Progress in Adhesion Adhesives 9781119625292, 1119625297, 9781119625254

Table of contents :
Content: Intro
Title Page
Copyright
Preface
Chapter 1: Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects in Microelectronic Packaging: A Critical Review
1.1 Introduction
1.2 Polymer Thermal Interface Material -Metal Interface Adhesion Phenomena
1.3 Ball Grid Array Solder Attach Adhesion Phenomena
1.4 Summary
Nomenclature
References
Chapter 2: Influence of Silicon-Containing Compounds on Adhesives for and Adhesion to Wood and Lignocellulosic Materials: A Critical Review
2.1 Introduction 2.2 An Overview of Compounds and Natural Minerals Containing the Element Si, which are the Most Relevant in the Science and Technology of Lignocellulosics2.3. Si-containing Compounds in Adhesives and in Lignocellulosic Substrates and their Influence on the Performance of Adhesive Bonds
2.4 Interactions of the Si Compounds with Lignocellulosics
2.5 Wood- and Lignocellulose-based Composites Containing Si Compounds
2.6 Summary and General Remarks
2.7 Acknowledgments
References Chapter 3: Recent Advances in Adhesively Bonded Lap Joints Having Bi-Adhesive and Modulus-Graded Bondlines: A Critical Review3.1 Introduction
3.2 Bi-adhesive Joints
3.3 Modulus-Graded Bondline
3.4 Summary
Acknowledgement
Nomenclature
References
Chapter 4: Adhesion between Compounded Elastomers: A Critical Review
4.1 Introduction
4.2 Co-crosslinking
4.3 Adhesion Between Vulcanized Rubber and Unvulcanized Rubber or Partially Vulcanized Rubber
4.4 Adhesion Between Vulcanized Rubber and Vulcanized Rubber
4.5 Summary
Acknowledgements
List of Symbols
List of Abbreviations
References Chapter 5: Contact Angle Measurements and Applications in Pharmaceuticals and Foods: A Critical Review5.1 Introduction
5.2 Contact Angle Measurements in Pharmaceutical Field
5.3 Contact Angle Measurements in Foods
5.4 Summary
Acknowledgement
References
Chapter 6: The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure to Oxygen or Ammonia Plasma: A Critical Review
6.1 Introduction
6.2 Oxygen Plasma Treatment
6.3 Ammonia Plasma for Introduction of NH2 Groups onto Polyolefin Surfaces
6.4 Discussion
6.5 Summary
Acknowledgement
References Chapter 7: Surface Free Energy Determination of Powders and Particles with Pharmaceutical Applications: A Critical Review7.1 Introduction
7.2 Surface Thermodynamic Quantities of Pure Materials
7.3 Contact Angle Methods
7.4 Determination of Surface Free Energy using IGC and AFM
7.5 Characterizing Surface Properties by Inverse Gas Chromatography
7.6 Pharmaceutical Applications
7.7 Summary
References
Chapter 8: Understanding Wood Bonds-Going Beyond What Meets the Eye: A Critical Review
8.1 Introduction: Macroscopic Knowledge for Successful Adhesive Bonding of Wood

Citation preview

Progress in Adhesion and Adhesives

Scrivener Publishing 100 Cummings Center, Suite 541J Beverly, MA 01915-6106 Publishers at Scrivener Martin Scrivener ([email protected]) Phillip Carmical ([email protected])

Progress in Adhesion and Adhesives Volume 4

Edited by

K.L. Mittal

This edition first published 2019 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA and Scrivener Publishing LLC, 100 Cummings Center, Suite 541J, Beverly, MA 01915, USA © 2019 Scrivener Publishing LLC For more information about Scrivener publications please visit www.scrivenerpublishing.com. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.

Wiley Global Headquarters 111 River Street, Hoboken, NJ 07030, USA For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com. Limit of Liability/Disclaimer of Warranty While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials, or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Library of Congress Cataloging-in-Publication Data ISBN 978-1-119-62525-4

Cover images: K.L. Mittal Cover design by Russell Richardson Set in size of 10pt and Minion Pro by Exeter Premedia Services Private Ltd., Chennai, India

Printed in the USA 10 9 8 7 6 5 4 3 2 1

Contents Preface 1

2

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects in Microelectronic Packaging: A Critical Review Dinesh P R Thanu, Aravindha Antoniswamy, Roozbeh Danaei and Manish Keswani 1.1 Introduction 1.2 Polymer Thermal Interface Material -Metal Interface Adhesion Phenomena 1.2.1 Basics of Thermal Interface Material Adhesion 1.2.2 Current Status of Thermal Interface Materials and their Bonding Mechanisms 1.2.3 Chemical Bonding 1.2.4 Mechanical Interlocking 1.2.5 Weak Boundary Layer 1.3 Ball Grid Array Solder Attach Adhesion Phenomena 1.3.1 Solder Alloy Selection 1.3.2 Flux Selection 1.4 Summary Nomenclature References Influence of Silicon-Containing Compounds on Adhesives for and Adhesion to Wood and Lignocellulosic Materials: A Critical Review Marko Petricˇ 2.1 Introduction 2.2 An Overview of Compounds and Natural Minerals Containing the Element Si, which are the Most Relevant in the Science and Technology of Lignocellulosics 2.2.1 Silica – SiO2 2.2.2 Silicates and Clay 2.2.3 Silicones 2.2.4 Silanes and Silsesquioxanes 2.3 Si-containing Compounds in Adhesives and in Lignocellulosic Substrates and their Influence on the Performance of Adhesive Bonds

xiii 1

2 3 3 5 6 12 14 14 15 18 19 20 21 25 26

29 29 30 32 33

35 v

vi

Contents

2.3.1 Compounds of Silicon in Adhesives 2.3.1.1 Inorganic Compounds of Si (Silica, Silicates, Clay, and Other Inorganic Compounds) 2.3.1.2 Organosilicon Compounds in Adhesives 2.3.2 Silicon-containing Compounds in Lignocellulosics with Regard to the Properties of Adhesive Bonds 2.3.3 Influence of Si in Coatings or in Lignocellulosic Substrates with Regard to Coatings Adhesion to the Substrates 2.4 Interactions of the Si Compounds with Lignocellulosics 2.4.1 Interactions with Silica 2.4.2 Interactions with Silicates 2.4.3 Interactions with Silicones 2.4.4 Interactions with Organosilicon Compounds and Coupling Agents 2.4.4.1 Interactions with Organosilicon Compounds 2.4.4.2 Coupling Agents 2.5 Wood- and Lignocellulose-based Composites Containing Si Compounds 2.5.1 Composites Containing Silica 2.5.2 Composites Containing Silicates and Clay 2.5.3 Composites Containing Silicones 2.5.4 Composites with Organosilicon Compounds 2.6 Summary and General Remarks 2.7 Acknowledgments References 3

4

35 35 40 42 44 46 46 48 49 50 50 52 57 57 59 60 61 64 65 65

Recent Advances in Adhesively Bonded Lap Joints Having Bi-Adhesive and Modulus-Graded Bondlines: A Critical Review Özkan Öz and Halil Özer 3.1 Introduction 3.2 Bi-adhesive Joints 3.2.1 Numerical and Analytical Studies 3.2.2 Experimental Studies 3.3 Modulus-Graded Bondline 3.3.1 Numerical and Analytical Studies 3.3.2 Experimental Studies 3.4 Summary Acknowledgement Nomenclature References Adhesion between Compounded Elastomers: A Critical Review K. Dinesh Kumar, M.S. Satyanarayana, Ganesh C. Basak and Anil K. Bhowmick 4.1 Introduction 4.2 Co-crosslinking

77 77 80 80 84 88 88 91 94 94 94 94 99

100 101

Contents vii

4.2.1

Adhesion Between Unvulcanized Rubber (Filled with Crosslinking Agents) and Unvulcanized Rubber (Filled with Crosslinking Agents) by Co-crosslinking 4.2.2 Adhesion Between Partially Vulcanized Rubber (Filled with Crosslinking Agents) and Partially Vulcanized Rubber (Filled with Crosslinking Agents) by Co-crosslinking 4.3 Adhesion Between Vulcanized Rubber and Unvulcanized Rubber or Partially Vulcanized Rubber 4.3.1 Adhesion between Vulcanized Rubber and Unvulcanized Rubber (Filled with Crosslinking Agents) 4.3.2 Adhesion between Vulcanized Rubber and Partially Vulcanized Rubber (Filled with Crosslinking Agents) 4.4 Adhesion Between Vulcanized Rubber and Vulcanized Rubber 4.5 Summary Acknowledgements List of Symbols List of Abbreviations References 5

6

Contact Angle Measurements and Applications in Pharmaceuticals and Foods: A Critical Review Davide Rossi, Paola Pittia and Nicola Realdon 5.1 Introduction 5.1.1 Prospects 5.2 Contact Angle Measurements in Pharmaceutical Field 5.2.1 Pharmaceutical Powders 5.2.2 Solvents for Pharmaceutical Use 5.2.3 Injectable Solutions for Parenteral Use 5.3 Contact Angle Measurements in Foods 5.3.1 Solid Foods 5.3.2 Liquid Foods and Beverages 5.3.3 Food Packaging 5.4 Summary Acknowledgement References The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure to Oxygen or Ammonia Plasma: A Critical Review Jörg Friedrich 6.1 Introduction 6.1.1 Reasons for Polyolefin Surface Functionalization 6.1.2 Energetic Considerations, Thermodynamics and Probability of Reactions 6.1.3 Processes on Molecular Level at Polyolefin Surface 6.2 Oxygen Plasma Treatment

104

118 138 140 164 166 184 186 186 187 189 193 194 199 200 200 211 218 222 222 231 234 236 236 237 241 242 242 245 249 254

viii

Contents

6.2.1 Formation of O Functional Groups at Polyolefin Surfaces on Exposure to Oxygen Plasma 6.2.2 Kinetics of Polyolefin Oxidation – Dependence on Parameters 6.2.3 Influence of Type of Plasma Gas 6.2.4 Influence of Polymer Composition 6.2.5 Auto-Oxidation 6.2.6 Oxidation by Exposure to Noble Gas Plasmas 6.2.7 Generation of OH Groups on the Surface of Polyolefins by Deposition of a Thin Layer of Poly(allyl alcohol) Plasma Polymer 6.3 Ammonia Plasma for Introduction of NH2 Groups onto Polyolefin Surfaces 6.3.1 Production of Primary Amino Groups on Exposure to Plasma 6.3.2 Thermodynamic Aspects 6.3.3 Ammonia Plasma 6.3.4 Formation of Functional Groups on Exposure to NH3 Plasma 6.3.5 Kinetics of N and NH2 Introduction on Exposure to Ammonia or Nitrogen-Hydrogen Plasmas 6.3.6 Side-Reactions at Polyolefin Surfaces on Exposure to NH3 Plasma 6.3.6.1 Hydrogenation and Dehydrogenation 6.3.6.2 Post-Plasma Oxidation 6.3.6.3 Nitrile Formation 6.3.7 NH2 Groups via Plasma Polymerization of Allylamine and Other N-Precursors 6.3.8 Attempts to Increase the Concentration of NH2 Groups by Addition of Ammonia to Allylamine Plasma Polymerization 6.3.9 Significant Side-Reactions During and After Plasma Polymerization of Allylamine 6.4 Discussion 6.5 Summary Acknowledgement References 7

Surface Free Energy Determination of Powders and Particles with Pharmaceutical Applications: A Critical Review Frank M. Etzler and Douglas Gardner 7.1 Introduction 7.2 Surface Thermodynamic Quantities of Pure Materials 7.3 Contact Angle Methods 7.3.1 The Zisman Method

254 260 262 263 265 267

269 272 274 275 277 278 280 283 284 286 286 290

294 294 297 303 304 304 315 315 316 320 320

Contents ix

7.3.2

The van Oss, Chaudhury and Good Method 7.3.2.1 Methods for Calculating the van Oss, Chaudhury and Good Parameters 7.3.3 The Chang – Chen Method 7.4 Determination of Surface Free Energy using IGC and AFM 7.4.1 Application of the Fowkes Method to IGC Data 7.4.2 Application of the van Oss, Chaudhury and Good Method to IGC Data 7.4.3 Application of the Chang-Chen Model to IGC Data 7.4.4 AFM Methods 7.5 Characterizing Surface Properties by Inverse Gas Chromatography 7.5.1 IGC Measurements - Experimental Considerations 7.5.2 Finite Dilution IGC 7.6 Pharmaceutical Applications 7.6.1 Surface Free Energy and Crystal Planes 7.6.2 Compaction of Tablets 7.6.3 Effects of Processing on Surface Free Energy 7.6.4 Performance of Dry Powder Inhalers 7.6.5 Powder Flow 7.7 Summary References 8

Understanding Wood Bonds–Going Beyond What Meets the Eye: A Critical Review Christopher G. Hunt, Charles R. Frihart, Manfred Dunky and Anti Rohumaa 8.1 Introduction: Macroscopic Knowledge for Successful Adhesive Bonding of Wood 8.2 Bond Formation (Developing Adhesion) 8.2.1 Influence of Wood Structure on Bonding 8.2.2 Influence of Wood Surface Quality on Bonding 8.2.2.1 Mechanical Damage at the Wood Surface 8.2.2.2 Surface Chemistry Barriers to Bonding 8.2.3 Adhesive Penetration 8.2.3.1 Void Penetration (Bulk Flow) 8.2.3.2 Cell Wall Penetration (Infiltration) 8.2.4 Adhesive Properties that Influence Void and Cell Wall Penetration 8.3 Properties of Adhesive-Wood Assemblies 8.3.1 Zones in a Wood Bond 8.3.2 How Adhesives Accommodate Wood Swelling 8.3.3 Two Classes of Adhesives 8.3.4 Methods for Determining Void and Cell Wall Penetration

320 324 325 326 326 328 329 329 331 332 339 340 340 341 342 344 345 346 347 353

353 356 356 360 361 365 367 368 370 373 375 375 376 377 379

x

Contents

8.2.4.1 Quantifying Depth of Void Penetration 8.3.5 Shortcomings of Standardized Tests 8.4 A More Detailed Approach than Standard Wood Failure Analysis 8.4.1 Going Beyond What Meets the Eye to Understand Epoxy Failure 8.4.2 Using SEM to Detect Brittle Failure in UF 8.4.3 Alternative Mechanical Methods of Testing for More Information 8.5 Unresolved Questions in Wood Bonding Research 8.5.1 How Do We Make Wood Surfaces Better for Bonding? 8.5.2 Does the Adhesive Have Good Penetration Into the Wood Structure? 8.5.3 How Does the Adhesive Interact with the Wood at the Nanoscale and Molecular Level? 8.5.4 Can We Improve the Resistance of Bonds to the Dimensional Changes inWood with Variation in Moisture? 8.5.5 How do Primers Work? 8.5.6 Where Does the Bond Failure Initiate and How Does it Propagate? 8.5.7 How Do We Optimize the Benefits of Cell Wall Penetration? 8.5.8 How Does the Adhesive Form a Suitable Polymer Matrix to Bridge Between the Two Wood Surfaces? 8.5.9 Will Adhesives Based on Renewable Resources be the Future in Wood Bonding? 8.5.10 How Much the Experience with Solid Wood Bonding can be Used to Understand Wood Based Particulate Bonding? 8.5.11 How Do We Compare Results Obtained in Different Laboratories with Different Wood Species with Different Adhesives? 8.6 Summary List of Abbreviations References 9

Dispersion Adhesion Forces between Macroscopic Objects–Basic Concepts and Modelling Techniques: A Critical Review Youcef Djafri and Djamel Turki 9.1 Introduction 9.2 Basic Concepts 9.3 Modeling Techniques 9.3.1 The Microscopic Theory (Hamaker’s Approach) 9.3.2 The Proximity-Force Approximation

386 387 388 389 391 391 393 393 394 394

395 395 396 396 397 397

398

398 399 399 400 421 421 422 424 424 426

Contents xi

9.3.3

The Retardation Effect 9.3.3.1 The Retarded vdW Forces 9.3.3.2 Retardation in Macroscopic Bodies 9.3.4 The Casimir Effect 9.3.5 Worldline Calculations of the Casimir Effect 9.3.6 The Macroscopic Theory of Van der Waals Forces (DLP Method) 9.3.7 The Coupled Dipole Method 9.4 Discussion and Prospects 9.5 Summary References

427 427 428 429 431 431 434 437 438 439

Preface The current book constitutes Volume 4 in the series “Progress in Adhesion and Adhesives” initiated in 2015. Volume 3 (published in 2018) in this vein was based on 12 review articles published initially in 2017 in the journal Reviews of Adhesion and Adhesives (RAA) by recognized experts and world-class active researchers in the wide domain of adhesion science and adhesive technology. The sole purpose of RAA is to publish concise, critical, illuminating and thought-provoking review articles on any topic within the broad purview of Adhesion Science and Adhesive Technology. Volume 2 (published in 2017) documented 14 review articles published originally in 2016 in RAA addressing a number of topics of current interest. The premier Volume 1 (although we did not designate it as Volume 1 as we had no crystal ball to prognosticate the future of this series) was the result of 13 review articles originally published in RAA in the year 2014. The first three volumes were warmly received and had served the intended purpose and this motivated us to bring out the current Volume 4. We have received excellent comments from many members of the adhesion community and this provided ample justification for making these books available. These hardbound books provide an easily accessible resource for critical information on a number of topics of current interest and relevance in the broad domain of Adhesion Science and Adhesive Technology. It should be stressed here that the authors of the review articles endorsed cheerfully the idea of bringing out these hardbound books. With the information being published at an ever-increasing rate, critical review articles serve an excellent purpose for anyone wishing to stay informed about the latest developments on a topic of his/her interest. As a matter of fact, anyone embarking on research in an area should start with a critical review article. The rationale for bringing out Volume 4 was the same as was applicable to its predecessors, i.e., the RAA has limited circulation so this set of books should provide broad exposure and wide dissemination of valuable information published on many and varied aspects of Adhesion & Adhesives in RAA. The chapters in this Volume 4 follow the same order as the review articles published initially in RAA in the year 2018. The subjects of these 9 review articles fall into the following areas.

xiii

xiv

Preface

O Adhesion to wood and wood bonds. O Adhesive joints. O Adhesion in microelectronic packaging. O Surface modification. O Contact angle, wettability and surface free energy. It should be pointed out that in 2018 the reason why only 9 review articles were published was that some review articles were very long and the total number of pages printed had to be maintained. The topics covered include: Adhesion phenomena in microelectronic packaging; adhesives for wood and lignocellulosic materials; adhesion to wood and lignocellulosic materials; adhesively bonded lap joints having bi-adhesive and modulus-graded bondlines; adhesion between compounded elastomers; applications of contact angle measurements in pharmaceuticals and foods; oxygen or ammonia plasma treatment of polyolefin surfaces; surface free energy determination of powders and particles; wood bonds; and dispersion adhesion forces between macroscopic objects. Now comes the pleasant task of acknowledging all those who were instrumental in making this book available. First and foremost, I would like to profusely thank the authors of review articles for their whole-hearted support and enthusiastically endorsing the idea of bringing out Volume 4 as they strongly felt that this series of books was an alternative, but very useful, medium to make the information contained in these review articles to a much wider audience. Also, thanks to Martin Scrivener (publisher) for conceiving the idea of publishing this series of books and for his steadfast support in this book project. Kash Mittal P.O. Box 1280 Hopewell Jct., NY 12533 E-mail: [email protected] May 2019

1 Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects in Microelectronic Packaging: A Critical Review Dinesh P R Thanu1*, Aravindha Antoniswamy2, Roozbeh Danaei3 and Manish Keswani4 1

Department of Materials Science and Engineering, University of Arizona, Tucson, AZ 85721, USA 2 Department of Materials Science and Engineering, University of Texas at Austin, TX 78712, USA 3 School of Mechanical and Materials Engineering, Washington State University, Pullman, WA 99163, USA 4 Department of Materials Science and Engineering, University of Arizona, Tucson, AZ 85721, USA

Abstract High performance and diverse power computing needs in desktop, server, communication, automotive, and artificial intelligence microelectronic sectors demand microprocessors with different form factors and intricate package designs. For such complex package architectures, semiconductor chips in addition to microprocessor, such as in-package memory, transceivers or a combination of both are essential to attain maximum performance. Integrated Heat Spreader (IHS) assembly with thermal interface material (TIM) layers plays a vital role in providing heat dissipation for these integrated circuit chips and aids them to perform with maximum efficiency for a long duration. Additionally, for Ball Grid Array (BGA) semiconductor packages, interfacial adhesion quality of solder attach material is very critical in determining package quality and reliability. There are numerous challenges associated with developing and optimizing such an IHS assembly process and solder attach material for high volume manufacturing with good throughput, quality and yield. Here we provide a comprehensive review of the adhesion mechanisms and challenges of polymer TIMs to IHS-metal interface and various techniques proposed in the literature to enhance their adhesion. Complexities involved in solder attach adhesion including material selection and chip assembly interaction are reviewed in detail as well.

*Corresponding author: [email protected]

K.L. Mittal (ed.) Progress in Adhesion and Adhesives, Volume 4 (1–24) © 2019 Scrivener Publishing LLC

1

2 Progress in Adhesion and Adhesives, Volume 4

Keywords: Adhesion, ball grid array, integrated heat spreader, interconnect, microelectronic packaging, solder flux, thermal interface material

1.1

Introduction

Integrated Heat Spreader (IHS) assembly is an integral part of a microelectronic package to accomplish efficient thermal dissipation from one or more semiconductor chips and to attain overall device thermal performance [1–5]. State-of-the-art IHS assembly process is used in various application sectors including desktop, server, communication, automotive, and artificial intelligence. In an assembly process, semiconductor chips or dies are first placed on an organic or ceramic substrate. Front layer interconnect and an underfill layer are further used to hold the dies mechanically. Furthermore, an IHS is glued on the substrate to protect the die and aid with its heat dissipation [6–8]. During IHS assembly, a thermal interface material (TIM) is placed or dispensed on the surface of the die and an adhesive or sealant material is further applied and cured on the organic substrate layer (motherboard) to mechanically hold the IHS in place. A TIM layer is placed between the die and the IHS referred to as TIM1 and between IHS and heat sink referred to as TIM2 as shown in Figure 1. A Ball Grid Array or BGA package is a form of surface mount technology (SMT) that is being used increasingly for integrated circuits. It has become one of the most popular packaging alternatives for high input/output devices in the industry. Apart from the improvement in connectivity they offer, BGAs have other advantages. They offer a much lower thermal resistance between the silicon chips than the quad flat pack devices. This allows heat generated by the integrated circuit inside the package to be conducted out of the device on the PCB faster and more effectively. The whole bottom surface of the device can be used on a BGA as compared to just the perimeter on a Land Grid Array (LGA) Primary heat transfer path

I II III IV V VI VII

(a)

VIII

(b)

Figure 1.1 Schematics of multiple thermal architectures. (a) Designs typically used in laptops without an IHS (b) Architectures used in desktop, server and high-end gaming applications with an IHS. I- Heat Sink; II-TIM2; III- IHS; IV- TIM1; V- Silicon Die; VI- Underfill; VII- Substrate; VIII- Solder Ball [Adapted from [8]].

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects 3

package. BGA packages have solder balls pre-attached to the bottom of the substrate, hence resulting in a higher stand-off height compared to LGA packages. Thermal efficiency and reliability of TIM1, TIM2 and solder balls define the performance of a microelectronic package and its long-lasting characteristics depend a lot on the interfacial adhesion between the layers [9]. In this review, focus will be mainly on the adhesion characteristics of these materials due to their quintessential role in the microelectronic packaging. This review will be divided into two sections excluding the Introduction. Section 2 provides an overview of the thermal interface material and its interfacial adhesion phenomena. Section 3 on the other hand is devoted to the solder joint adhesion characteristic and its dependence on the material quality and assembly process.

1.2 1.2.1

Polymer Thermal Interface Material -Metal Interface Adhesion Phenomena Basics of Thermal Interface Material Adhesion

Shown in Figure 2 is a simple graphical representation of a microelectronic package resistance network. The key temperatures of interest include Tj or Tjmax which refers to the maximum junction temperature in the device, Tc which refers to the IHS temperature, Tsink which refers to the heat sink temperature, and Ta which refers to the local ambient air temperature. Total thermal resistance which depends on the interfacial adhesion can be described as two components: junction to IHS surface resistance or package thermal resistance and heat sink surface to ambient resistance or system thermal resistance. The maximum allowable junction temperature is one of the key factors that limits the power dissipation capability of a device. The maximum power that a device can dissipate, also referred to as thermal design power (TDP), can be written as

TDP = (Tjmax–Ta)/ TDP = (Tjmax–Ta)/(

jc

(1)

ja

+

)

ca

T Ψ

TIM + Heatsink T

Ψ Ψ

TIM + IHS T

Ψ =Ψ + Ψ

Figure 1.2 Classical example of a one-dimensional resistance network [Adapted from [10]].

(2)

4 Progress in Adhesion and Adhesives, Volume 4

jc

+

ca

= (Tjmax–Ta)/TDP

(3)

In other words, TDP is the maximum amount of heat that a processor can generate for a thermally significant period while running real applications. ja is the overall thermal resistance, jc is the package thermal resistance and ca is the system thermal resistance. If we consider only the package level thermal resistance, equation (3) can be re-written as, jc =

(Tjmax – Tc)/TDP

(4)

Thermal impedance in a package is represented universally in industry as Rjc and is measured as package thermal resistance ( jc) under uniform power times the area of the silicon die (Adie) as shown in equation (5).

Rjc = (

)

jc uniform power

× Adie

(5)

Poor adhesion of the TIM layer has a huge impact on the junction temperature, increasing Rjc and, therefore, impacts TDP leading to poor power capability of a microelectronic device. Real TIMs in a semiconductor package look like as shown in Figure 3. Real TIMs have a finite bondline thickness (BLT) and at the interface have voids/delamination because of their inability to completely wet the surface. From Figure 3 it can be inferred that the total thermal impedance (RTIM) of a real TIM can be written as

TIM

Rc2

Rbulk = BLT kTIM

Distance

Material 2

BLT

Rc1 Material 1

Rbulk=Bulk thermal resistance BLT=Bond line thickness kTIM=Thermal conductivity of TIM

Temperature

Figure 1.3 Cross-sectional illustration of a TIM layer [Adapted from [8]].

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects 5

RTIM = (BLT/ kTIM) + Rc1 + Rc2

(6)

where Rc1 and Rc2 represent the contact resistances of the TIM at the silicon-TIM and TIM-metal interfaces, respectively. There is a great drive in the semiconductor industry to decrease the RTIM as much as possible to provide efficient thermal dissipation on high-end packages. One way to do this is by increasing the thermal conductivity of the TIM (kTIM) using novel materials and process optimization; however, it will be of no use if the TIM fails to adhere to the surface initially or the adhesion bond weakens over a period of time. Both these failures lead to quality issues either during manufacturing impacting yield and productivity or in the field during use condition affecting product users. Customer returns because of these types of quality issues are becoming a norm these days and hence understanding the fundamentals of adhesion bonds and the root cause of adhesion failures is very critical.

1.2.2

Current Status of Thermal Interface Materials and their Bonding Mechanisms

With the recent developments in TIMs, a Polymer Thermal Interface Material (PTIM) is preferred over a Solder Thermal Interface Material (STIM) due to its inherent cost benefits and comparable thermal performance to STIM [11, 12]. IHS attachment process, in particular, can be performed in two stages: room temperature assembly and high temperature curing. Room temperature assembly is, however, not sufficient enough to bond the PTIM/STIM and sealant with IHS as these materials are not cured. Both sealant and PTIM/STIM need to be cured at high temperature (~150 °C to 180 °C) with pressure on top of these to achieve their targeted material properties and BLT. Polymer to IHS bonding is caused by a combination of different adhesion mechanisms such as physical adsorption, mechanical interlocking, and chemical bonding [13–17]. Physical adsorption forces are a resultant of interatomic and intermolecular interactions due to the van der Waals bonds. On the other hand, mechanical interlocking is brought about by surface roughness of the substrate; however it is debatable whether enhancement in adhesion strength is due to mechanical interlocking or increase in the effective area through secondary bonding [18]. Primary chemical bonds are one order of magnitude stronger than secondary chemical forces such as van der Waals force (Table 1). The primary bond type also includes acid-base theory of adhesion which is very popular and has been studied extensively on various surfaces including polymers [19–21]. By increasing the number of bonds across the interface, polymer-IHS metal bonding could be enhanced considerably. Chemists tend to associate adhesion with the energy liberated when two surfaces meet to form an intimate contact characterized as an interface as shown in Figure 4. In other words, adhesion may be defined as the energy required to dismantle the interface between two materials. Physicists and engineers describe adhesion in terms of forces, with the force of adhesion being the maximum force exerted when two adhered materials are separated. Many theories on the mechanisms of adhesion are usually attributed to adsorption, chemical bonding and mechanical interlocking all of which play significant roles in interfacial

6 Progress in Adhesion and Adhesives, Volume 4

Table 1.1 Bonding energy range for different types of bonds [22]. Category

Bond type

Primary Bond

Ionic

Bond energy [kJ/mol] 600–1100

Covalent

60–700

Metallic

110–350

Brönsted acid-base interaction Secondary Bond

Up to 1000

Hydrogen bonds

Up to 40

Van der Waals bonds

Up to 40

A1

1 1 Coating 2

2 Substrate

A3

A2

A1 = Coating surface to be adhered A2 = Real surface to substrate A3 = Interface between coating and substrate

Figure 1.4 Pattern of surface effects determining the overall adhesion [Adapted from [23]].

bonding. The energy required to separate the polymer from a metal surface is a function of the adhesion level i.e. interactions at the interface, but it also depends on the mechanical and viscoelastic properties of the polymeric material. Modern multi-chip packages have inherent manufacturing variability in chip stacks including IHS flatness, die stack-up gaps and die dynamic warpage affecting TIM1 and TIM2 adhesion and BLTs. This, in turn, affects the Rjc (equation 5) and the package reliability. Listed in Table 2 are some of the commercially available TIM1/TIM2 materials which exhibit unique adhesion property.

1.2.3

Chemical Bonding

The adhesion results from molecular contact between polymer and substrate due to surface forces. For these forces to develop, the polymer must make an intimate contact with the substrate surface. The process of establishing a continuous contact is termed as wetting which can be measured by contact angle. The prerequisite for wetting is a contact or wetting angle of less than 90 degrees. A complete spontaneous wetting occurs when the contact angle is zero degrees. Schematic illustration of good and poor wetting is presented in

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects 7

Table 1.2 TIM materials widely used in industry and are commercially viable. TIM materials

Advantages

Disadvantages

Polymers

1. Conform to surfaces 2. Fail cohesively, depending on polymer chemistry

Physical movement and delamination failure mode

Adhesives

1. Conform to surfaces 2. No physical movement or migration

1. 2. 3. 4.

Aligned Carbon Fibers

1. Very high thermal conductivity 2. Easy to handle 3. No physical movement or migration

1. High pressure required to compress 2. Not recoverable

Phase-Change Materials

1. High thermal conductivity 2. Compressible 3. Conform to surfaces

1. Difficult to handle 2. Not recoverable

Thermal Greases

1. High thermal conductivity 2. Compressible 3. Conform to surfaces

1. Difficult to handle 2. Not recoverable

Gap Fillers/Gels

1. Conform to surfaces 2. Compressible 3. No physical movement or migration

1. Low thermal conductivity 2. Cure process required 3. Difficult to handle 4. Delamination failure mode 5. Not recoverable

Gap Pads/ Elastomers

1. Easy to handle 2. Fill larger BLT gaps 3. May be recoverable

1. High contact resistance 2. Low thermal conductivity in general

Low thermal conductivity Cure process required Delamination failure mode Not recoverable

Figure 5. Wetting is favoured when the substrate surface tension (in mN/m), also known as the surface free energy (in mJ/m2), is higher than that of the adhering polymer. Low surface energy polymers, therefore, easily wet the high surface energy substrates such as metals and glass. On the other hand, substrates with low surface energy such as polyethylene, fluorocarbons, etc. will not be wetted. Surface tension is an important factor that determines the ability of a polymer coating to wet and adhere to a substrate. The ability of a coating to wet a substrate has been shown to be improved by using solvents with lower surface tensions. Wetting may be quantitatively defined with reference to a liquid drop resting in equilibrium on a solid surface (Figure 5). The smaller the contact angle, the better the wetting. When q is zero, the liquid wets the

8 Progress in Adhesion and Adhesives, Volume 4 Good

Better

Poor

Vapor

lv

Liquid sv

Solid

sl

Figure 1.5 Cartoon illustration of good and poor wetting [Adapted from [24]].

solid surface completely at a rate depending on the liquid viscosity and the solid surface roughness. The equilibrium contact angle for a liquid drop sitting on ideally smooth, flat, and nondeformable surface is related to various interfacial tensions by Young’s equation: lv

cos =

sv

–

sl

(7)

Where lv is the surface tension of the liquid in equilibrium with its own saturated vapor, sv is the surface free energy of the solid in equilibrium with the saturated vapor of the liquid, and sl is the interfacial tension between the solid and the liquid. For spontaneous wetting to occur, the surface tension of the liquid must be lower than the surface free energy of the solid. It is also possible for a liquid to spread and wet a solid surface when is greater than zero, but this requires the application of a force to the liquid. The ranges of van der Waals forces and the hydrogen bonds are extremely short for this purpose. For optimum adhesion it is, therefore, absolutely essential to ensure good wetting by the coating material applied, thus creating ideal conditions for causing the film forming agent molecules to approach the substrate. The condition for good wetting is always fulfilled whenever the surface free energy of the substrate is higher than that of the liquid coating material. Such a requirement can easily be fulfilled when coating metals because of their high surface free energy. With various nonpolar plastics such as polyethylene or polypropylene, with surface free energy values less than 30 mJ/m2, it is not possible to achieve good adhering coatings because of the inferior wettability without appropriate surface treatments. The surface tension values of the involved materials, the liquid coating and the solid substrate, are most important for substrate wetting. Figure 6 and Table 3 highlight the surface tension in mN/m or surface free energy in mJ/m2 of some of the commercially available metals and polymer materials. It is interesting to note that the polarity of the substrate and possible surface structures (porosity, roughness) will also influence the adhesion mechanism. For example, for surfaces such as wood, the surface tension will not be the same across the whole surface and will vary. Additional surface irregularities can be due to contamination of the surface, which will then cause wetting problems in the form of craters in some areas. Finally, there is also a time aspect. The surface free energy of the substrate is constant, but the surface tension of the liquid phase changes due to solvent evaporation and crosslinking reactions. If, in this process, the surface tension of the liquid exceeds that of the substrate, dewetting can occur, if the film viscosity is still low enough. When a polymeric

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects 9

Re

2500

Os w Ru

Ir

Surface tension of molten metals (mNm–1)

Mo

2000

Co

Rh v

Fe Ni Ti

Cu

Al

1000

Ag

Nb

Hf

U Pd Zr

1500

Au

Pt Cr

Ta

Mn Be Th Sc Lu Si Y

Ce La Pr Nd Tb Hb Cd In Mg Ge Gd Dy Er Sn Pu Pb 500 Hg Tl Sm Sb Ca Li Bi Yb Te Na Ba Eu Cs Rb K Se Zn

Ga

500

1000

1500 2000

2500

3000 3500

Melting point (K)

Figure 1.6 Surface tension of molten metals at their melting points [Adapted from [25]].

Table 1.3 Surface tension of commercially available polymers. Polymer

Surface tension (mN/m)

Polyperfluoropropylene

16

Polytetrafluoroethylene (Teflon)

18.5

Poly(dimethylsiloxane)

24

Polyethylene

31

Polystyrene

34

Poly(methyl methacrylate) (acrylic)

39

Poly(vinyl chloride) (PVC)

40

Poly(ethylene terephthalate) (polyester)

43

Poly(hexamethylene adipate) (nylon)

46

10 Progress in Adhesion and Adhesives, Volume 4

coating is applied on a substrate, a chemical reaction takes place between the two materials i.e. the substrate as well as the coating. It is often desirable to modify the substrate to ensure reactivity at the interface by removing contamination and/or introducing functional groups. This simplified view of the interfacial or interphase bonding neglects physical forces between the two materials, which are influenced by surface roughness. For a comprehensive characterisation of coatings, surface analysis of the substrate (chemical as well as topographic) and thermal analysis of cured polymeric coating materials are of great importance. Due to the higher bonding energy of the primary bonds in comparison to secondary bonds, working on different types of primary bonds has been more attractive for researchers. Various types of primary bonds, such as ionic and covalent at different interfaces have been reported in the literature. For instance, bonding of brass and rubber occurs by curing with the existence of sulfur due to the creation of polysulfide bonds [26]. Using coupling agents, such as adhesion promoter molecules, is one of the most interesting approaches for interfacial chemical bonding. These molecules are able to make a bond with both polymer and metal [27, 28]. The most common adhesion promoters are silane-based molecules. Examples of silane-based adhesion promoters are listed in Table 4. Schuberth et al. [29] used 2,2’-Spirobi[4H-1,3,2-benzodioxasilin] and 2-(3-aminon-propyl)-2-methyl-4H-1,3,2-benzodioxasilin as twin monomers in order to improve the polymer-metal bonding by introducing chemical bonding between the two. In their work, they considered the interaction of chemical adhesion promoter versus surface treatments on steel-fiber reinforced polymer (FRP) and aluminum-FRP as shown in Figure 7. Interestingly, it was concluded that using these two techniques together was not generally more effective than the individual ones, and surface treatment should be adjusted for the purpose and application. It is definite that all the mechanisms mentioned above can affect adhesion and thus bond strength. It is also undisputed that the individual mechanisms of adhesion only make significant contributions if the prerequisites have been met. If one endeavors to establish

Table 1.4 Examples of silane-based adhesion promoters reported in the literature. Adhesion promoter

Organic structure

Trimethoxyvinylsilane

(CH2)=CHSi(OCH3)3

(3-Chloropropyl)trimethoxysilane

Cl(CH2)3Si(OCH3)3

(3-Mercaptopropyl)trimethoxysilane

HS(CH2)3Si(OCH3)3

(3-Aminopropyl) diethoxymethylsilane

H2N(CH2)3Si(CH3)(OC2H5)2

(3-Aminopropyl) triethoxysilane

H2N(CH2)3Si(OC2H5)3

(3-Aminopropyl) trimethoxysilane

H2N(CH2)3Si(OCH3)3

Triethoxy –(3-ureidopropyl)silane

H2NCONH(CH2)3Si(OC2H5)3

Triethoxy –(3-methacryloxypropyl)silane

CH2=C(CH3)COO(CH2)3Si(OCH3)3

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects

11

the causes of interactions, numerous theories are available in the literature [30–33]. The explanations range from mechanical attachment of the coating on the cavities and fissures in the substrate (mechanical anchoring) and attachments of film forming agent molecules by diffusion or contact charges and the creation of mirror forces through interactions of polar functional groups, hydrogen bonds or chemical links between the coating and the substrate as illustrated in Figures 8 and 9. Figure 9 shows curves and ranges of the potential energies of van der Waals forces as well as hydrogen bonds as the causes for adhesion. Van der Waals forces are orientation forces (dipole-dipole), induction forces (dipole/induced dipoles) and dispersion forces. Assuming a suitable chemical structure and an appropriate substrate there are also effective hydrogen bonds.

Steel Aluminum

12

10.2 9.5

10 8.4 Ra (μm)

8

7.5

7.3

6 3.4

4 2 0 Grit-blasted

(a)

(b)

30

35

Without bonding agent With bonding agent

Bonding strength [MPa]

Bonding strength [MPa]

35

Thermally sprayed NiAl 95/5

25 20 15 10 5 0 A

B

C

D

(c)

30

Laser-structured

Without bonding agent With bonding agent

25 20 15 10 5 0 A

B

C

D

A=untreated, B=grit-blasted, C=thermally-sprayed NiAl 95/5, D=laser-structured

Figure 1.7 (a) Surface roughness of steel and aluminum according to different surface treatments. Bonding strength of (b) Steel-FRP and (c) aluminum-FRP determined by shear tension test [Adapted from [29]].

12 Progress in Adhesion and Adhesives, Volume 4 Mechanical anchoring

Contact charging + + – –

Mechanical attachment of the coating in the cavities and fissures in the substrate

+ + + + – – – –

Dipole-dipole-interaction

– + – +

Diffusion

+ – + –

Hydrogen bridging bonds OH

Attachment of film forming agent molecules by diffusion

Mirror forces in metals

Chemical crosslinking O C H

N

O O

C H

N

O O

C H

OH OH

Oxide layers

N

O

–

–

–

+

+

+

–

–

–

Figure 1.8 Physical and chemical bonding of polymer coatings to the metal surface [Adapted from [23]].

1.2.4

Mechanical Interlocking

Lee and Qu studied the effect of surface roughness by presenting the outcome of different types of oxidation on interfacial fracture toughness [34]. In their study, it was shown that copper as a function of oxidation time from exposure to the environment creates different oxidized forms. Within the first 30 seconds of copper exposure, pebble-like cuprous oxide (Cu2O) was formed on the surface with an approximate thickness of 0.2 μm. After 2 minutes, needle-shape cupric oxide (CuO) was formed. Formation of cupric oxide causes interfacial (adhesion) failure mechanism in the polymer coating (Figure 10). Schaubroeck, et al. exposed a polymer resin surface to KMnO4/NaOH solution for different time periods to roughen the surface by means of etching [35]. By controlling the time of exposure, surface roughness was controlled and after that surfaces were treated by polydopamine according to the method described by Lee et al. [36]. Afterward, copper deposition was carried out on polydopamine (DOPA)-modified and non-modified etched surfaces. Peel strength of the deposited copper on non-modified Epoxy Cresol Novolac (ECN) resin substrates increased by an increment in surface roughness. Moreover, it shows that lower electroplating bath temperature leads to higher peel strength in comparison to higher temperatures as indicated in Figure 11.

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects

13

kJ/mol Epot

Repulsion

40 30

Van der Waals forces (dipole-dipole, dipole-inducted dipole, and dipersion)

20 10 0

1

2

3

4

5

6

Distance r (Å)

Attraction

–10 –20 –30

Hydrogen bonds Chemical bonds

–40 –50

Type of van der Waals forces Orientation forces Dipole-dipole (Keesom forces) Induction forces Dipole-induced dipole (Debye forces) Dispersion forces (London forces)

–60

Hydrogen bridging bonds

Average energy

Range

25 kJ/mol

E=

15–20 kJ/mol

E=

5–10 kJ/mol

E=

40–50 kJ/mol

2μ4 3kTr6 2α.μ2 r6 3hv0.α 4r6

E= e–kr

Figure 1.9 Potential energies of van der Waals forces as well as hydrogen bonds [Adapted from [23]].

120

GIC (J/m2)

100 80 60 40 20 0

0

5

10 15 Oxidation time (min)

20

Figure 1.10 Fracture toughness (GIC) as a function of oxidation time and representative SEM images at different times [Adapted from [34]].

On both polydopamine modified substrates as well as non-modified surfaces, peel strength does not have a linear correlation with surface roughness. This result has a striking similarity with Schuberth et al. work which showed that employing both adhesion promoter and surface roughness does not necessarily lead to higher bond strength of metal-polymer bonds in comparison to their individual effect [29].

Non-modified Epoxy cresol Novolac (ECN) 1.2 High Low

1 0.8 0.6 0.4 0.2 0

Polydopamine-modified ECN Peel strength (N/mm)

Peel strength (N/mm)

14 Progress in Adhesion and Adhesives, Volume 4

1.2 High-DOPA Low-DOPA

1 0.8 0.6 0.4 0.2 0

0

200 400 600 800 1000 1200 Rrms (nm)

0

200 400 600 800 1000 1200 Rrms (nm)

Figure 1.11 Peel strength of the non-modified (left) and polydopamine-(DOPA) modified ECN surface (right) after copper plating on top. Two series of measurements in each graph are representative of low (35oC) and high (47oC) electroless plating temperature [Adapted from [35]].

1.2.5

Weak Boundary Layer

According to a theory proposed by Bikerman, adhesion failures occur due to the presence of a weak layer at the interface between the adhesive and target surface as illustrated in Figure 12 [37]. This theory suggests that the root cause of adhesion failure is the cohesive failure within the weak boundary layer. One of the common weak layers is the hydrocarbon contamination on the target surface [38]. Plasma surface treatment is one of the wellknown ways to remove hydrocarbon layer from the surface [39]. Jang et al. investigated the nonconductive film (NCF)-SiO2 adhesion improvement by oxygen plasma cleaning [40]. Applying the oxygen plasma to the oxidized silicon wafer causes removal of the hydrocarbons and increase in silanol groups and hydrophilicity of SiO2 surface. However, strikingly, surface roughness decreased after plasma treatment in comparison with non-treated surfaces and due to this plasma treatment could not enhance the adhesion despite increment in surface energy. However, de-ionized water (DIW) rinse was applied after surface plasma cleaning. This step could change the surface roughness during plasma treatment due to the presence of hydroxyl groups on the oxidized silicon surface [41, 42]. With DIW rinse step, NCF-SiO2 bond strength was improved as shown in Figure 13. In another study, Coulon et al. carried out plasma surface treatment on the polymer resin and then covered the surface with 1 μm thick evaporated aluminum film [43]. It was demonstrated regardless of the roughness of the surface, plasma treatment brought about surface energy and adhesion strength increments (Figure 14). X-ray photoelectron spectroscopy (XPS) analysis of treated and non-treated polymer resin surfaces revealed that atmospheric plasma created reactive carbonyl groups and metal-carbonyl bonds led to higher adhesion strength [44, 45].

1.3

Ball Grid Array Solder Attach Adhesion Phenomena

Ball grid array packages have solder balls pre-attached to the bottom of the substrate, hence resulting in a higher stand-off height compared to land grid array packages. This higher

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects

15

Adhesive

Weak boundary layer

Target surface

RMS : 0.248 nm STDEV : 0.01 nm

RMS : 0.162 nm STDEV : 0.007 nm

RMS : 0.331 nm

(a)

(b)

(c)

Shear strength (MPa)

Figure 1.12 Schematic presence of a weak boundary layer, such as environmental contamination, as a failure mechanism in adhesion [Adapted from [37]].

(d)

9 8 7 6 5 4 3 2 1 0 DIW rinse

Plasma

Plasma + DIW rinse

Figure 1.13 Atomic force microscopy (AFM) measurements of surface roughness of (a) un-treated (b) atmospheric plasma treated (c) and surface subjected to both plasma treatment and DIW rinse. (d) Shear strength of NCF-SiO2 bonding with different surface treatments [Adapted from [40]].

stand-off height and improved planarity of the package have been shown to positively affect the temperature cycle performance of the electronic devices [46]. The adhesion/joint quality and reliability of these solder joints and the yields during the surface mount process heavily depend on the material and process selections. The materials involved in the ball attach process are solder sphere, flux, or solder paste. The solder sphere selection determines the strength and reliability of the joint and hence affects its performance under stress and temperature that the devices go through in the field. The flux or solder paste serves to clean the oxide on the solder sphere and the pad on the substrate and attaches the solder sphere through a metallurgical joint.

1.3.1

Solder Alloy Selection

Tin-silver-copper alloys are widely used as a lead-free alternative in the semiconductor packaging industry. This is because they possess an attractive combination of wettability to the pad surface, good mechanical properties, microstructural stability and availability [47]. However, the melting range of tin-silver-copper (SnAgCu, also known as SAC) alloys is

16 Progress in Adhesion and Adhesives, Volume 4

Polar component Dispersion component

Surface energy (mJ/m2)

80 60 40 20 0 Composite A

(a)

Composite Composite Composite A-treated B B-treated

Resin

Resintreated

Adhesion strength (MPa)

8 7

No treatment

6 Plasma treatment

5 4 3 2 1 0

(b)

Composite A

Composite B

Resin

Figure 1.14 (a) Surface energy evolutions on different surfaces before and after atmospheric plasma treatment, (b) adhesion strength of physical vapor deposited Al on top of the un-treated and plasma treated composites and resin [Adapted from [43]].

high and requires a typical reflow peak temperature of 245 °C. This leads to higher warpage of the package during assembly and may lead to poor joint formation or adhesion [48]. Since Sn is a reactive species, the same intermetallic compounds are formed with both Cu and Ni surfaces during solder/substrate interaction between all common Sn-based solders such as Sn-Ag-Cu, Sn-Cu, and Sn-Bi alloys [49]. With a Cu surface, these alloys form Cu3Sn and Cu6Sn5 as can be seen in the phase diagram that these two are the stable intermetallic compositions in the temperature range of interest [50]. With a Ni surface, these alloys primarily form Ni3Sn4 and it can be seen that this is one of the stable compositions in the phase diagram and forms at the lowest temperature upon cooling among all intermetallics [51]. When Cu is present in the solder, even in much lower amount than Sn, (Cu, Ni)6 Sn5 intermetallic is observed as well.

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects

17

Once the joint is formed, it is important that the adhesion holds or the joint survives through mechanical shock and severe thermal changes due to operation of the device as well as external changes in temperature [52]. Under mechanical shock, the primary mode of failure is cracking along the intermetallic layer as seen in Figure 15.This is primarily because at high strain rates, the yield strength of the bulk solder increases drastically making the intermetallic layer the weaker portion of the joint [53]. This is referred to as strainrate sensitivity. It has been found that a decrease in elastic modulus, which is modulated primarily by the composition of the solder, and a decrease in yield strength can help in increasing the overall toughness of the joint during shock, for example SAC105 showed better shock resistance than SAC405 [54]. A lower silver content was found to be generally associated with higher ductility and lower strength of SAC solders in the 0-5% range [55]. The other method to improve shock resistance would be to change the nature and properties of the intermetallic layer itself. This has been achieved by addition of Ni to SAC alloys for example [52]. The effect of Ni addition, in the right proportion, was found to decrease intermetallic thickness or growth through void reduction and also through increasing the liquidus temperature of the solder [56, 57]. Under thermal cycling, however, the crack originates and propagates through the body of the solder [52]. The crack nucleation sites include the interfaces between the intermetallics and the bulk solder and the grain boundaries of the bulk solder created by localized recrystallization reactions due to differential plastic strain stored within the solder [59]. The crack growth has been proposed to be an intergranular propagation along the recrystallized network of high angle grain boundaries as illustrated in Figure 16 [59, 60]. Due to the dependence of recrystallization on the critical strain energy, higher strength alloys were shown to be beneficial, however, this could also result in poor drop shock performance as discussed earlier. Sn-Bi solders are lower melting alternatives to SAC solders. It was found that Sn-Bi solders wetted Au/Ni/Cu surface better than bare Cu or bare Ni [61]. In addition, the

Interfacial crack (package side)

characterized by a lack of bulk solder deformation and an absence of bulk solder cracking

Interfacial crack (board side)

Figure 1.15 An example of a typical failure mode of solder joint (SAC 405) under mechanical shock or high strain rate deformation [Adapted from [58]].

18 Progress in Adhesion and Adhesives, Volume 4

interfacial tension between the liquid solder and substrate decreases with an increase in Bi content leading to better wetting at higher Bi concentrations [61]. The relationship between wettability, surface energy and adhesion/joint strength has been discussed [62]. One study found that a Sn-30Bi-0.5Cu alloy had a much lower contact angle or superior wettability to an SAC305 solder [63]. However, a composition beyond the eutectic concentration of Bi (58%) leads to embrittlement due to the presence of Bi-rich phase [64]. In addition, once the Sn is depleted for intermetallic compound formation, a Bi-rich layer forms adjacent to the interface leading to potential weakness in the solder joint [65].

1.3.2

Flux Selection

Intergranular propagation along recrystallized network of high angle grain boundaries

Polarized light

Bright light

Flux type, composition and quantity play a crucial role in controlling the adhesion of the solder to the pad. In addition, they also might modulate the adhesion of other materials such as the underfill (though not directly relevant to second-level interface), when a noclean flux is used [66]. Fluxes used for soldering in the microelectronics industry including the ball attach process fall into two major categories: no-clean and water washable. No-clean fluxes do leave a residue while any water washable flux can be removed by washing with water, which is typically heated and sprayed. However, no-clean flux residues are hydrophobic in nature and cannot be water washed but these are tolerated since they are nonconductive and noncorrosive [67, 68]. The flux reactions with the substrate pad and the solder sphere are much more than just oxidation-reduction and involve acid-base, co-ordination and adsorption type reactions as well. Most fluxes form salt and water with metallic oxides and the salt further helps in promoting solder wetting [69]. In general, for proper adhesion, fluxing of lead-free solders such as SAC alloys is more challenging than lead containing solders. This has been attributed to 3 different factors: 1.

Figure 1.16 Intergranular crack propagation during thermal cycling of an SAC alloy [Adapted from [60]].

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects

19

Tin salt that forms at high temperatures is harder to clean than lead salt, 2. Higher flux activity is needed to boost the wetting of an SAC solder leading to more side reactions, and 3. Higher reflow profile is needed for SAC compared to an eutectic Sn-Pb solder. There are other challenges caused by lead-free solder alloys other than SAC alloys, prevalent in the industry today. Low temperature solders such as Sn-Bi need fluxes with a lower activation temperature and highly oxidizable alloys such as the ones containing Zn need a flux with a high oxygen barrier capability or needs to be reflowed under an inert atmosphere [67]. Hence modern fluxes need to have a superior compatibility to lead-free solders based on factors such as superior thermal stability, etching ability, and a temperature range of activity that is compatible with the melting range of the solder. This often necessitates custom development based on solder alloy package/board pad surface finish selection.

1.4

Summary

In this review paper, a detailed overview of interfacial adhesion phenomena of the polymer layers to a metal surface is given. Various bonding mechanisms which play a key role in polymer to metal bonding are discussed in detail. For optimum adhesion, it is absolutely essential to ensure good wetting. Ideally, for good substrate wetting the surface tension of the polymer material should be lower than the surface tension of the substrate. The quality and durability of polymer to metal bonding are directly related to the nature of adhesion. Many theories influencing the mechanism of interfacial adhesion including adsorption, chemical bonding and mechanical interlocking are discussed. The energy required to separate the polymer from a metal surface also depends on the properties of the polymer material. It is definite that all the mechanisms mentioned could modulate the adhesion and thus the bond strength. Novel ideas and procedures to enhance polymer to metal bonding and to mitigate adhesion failures are provided which include chemical surface modification and plasma treatment. According to Schuberth et al. [29] and Schaubroeck et al. [35] surface chemical modification and adhesion promoters used in combination on the metallic surface do not necessarily enhance the bond strength significantly compared to employing them individually on the surface. In other words, efficacy of adhesion promoters depends on the surface and it is more effective on smoother surfaces. In another section, the effect of plasma treatments of polymer substrates is reviewed. Jang et al. [40] studied the adhesion phenomena of SiO2 substrates using oxygen plasma treatment with and without de-ionized water treatment. Coulon et al. [43] measured the adhesion strengths on carbon/epoxy composite surfaces treated with atmospheric plasma. Remarkably, from these two works, it becomes clear that there is a difference in applying plasma treatment on oxidized silicon surfaces as opposed to polymer surfaces. In Jang et al. work [40], it was shown that plasma treatment on SiO2 showed no enhancement in its surface roughness and did not add any value to bond strength. Use of de-ionized water after plasma treatment helped to improve bond strength drastically with the formation of Si(OH)x bonds. But on the other hand, Coulon et al. [43] demonstrated that plasma treatment of the polymer surface, regardless of its roughness, would help PVD coated aluminum bond strength.

20 Progress in Adhesion and Adhesives, Volume 4

To assemble latest technology BGA microelectronic packages with improved reliability and performance has been one of the challenging tasks for semiconductor industries. It involves mounting solder balls onto the land side of the substrate. The adhesion and survivability of solders depend heavily on selection of solder spheres and flux materials. The melting range of the solders has a significant effect on package warpage and its bonding phenomena. Sn-Ag-Cu or SAC solders are the common choice for lead-free solder spheres and it has been found that an increase in the silver content leads to superior thermal cycling survivability but poor shock survivability and vice versa. Bi tends to increase the strength of the solder and Bi containing low melting solder is available, however, it can lead to embrittlement of the joint. Fluxes fall into no-clean and water washable categories and selection needs to be made here based on application and tolerability. Fluxing SAC solders is much more challenging when compared to lead based solders for several reasons and the flux choice needs to be made suitably as well with regards to activation range, thermal stability and oxide-removal capability.

Nomenclature Adie BGA BLT Cu2O CuO DIW ECN FRP IHS kTIM LGA NCF PCB PTIM Rc1 Rc2 Rjc SMT STIM TIM TIM1 TIM2 Ta TDP Tc

Area of a Semiconductor Chip Ball Grid Array Bond Line Thickness Cuprous Oxide Cupric Oxide De-ionized Water Epoxy Cresol Novolac Fiber Reinforced Polymer Integrated Heat Spreader Thermal Conductivity of the Thermal Interface Material Land Grid Array Non-Conductive Film Printed Circuit Board Polymer Thermal Interface Material Contact Resistance of the TIM at the Silicon to TIM Interface Contact Resistance of the TIM at the TIM to Metal Interface Thermal Impedance at the Junction to IHS Surface Surface Mount Technology Solder Thermal Interface Material Thermal Interface Material Die to IHS Thermal Interface Material Layer IHS to Heat Sink Thermal Interface Material Layer Ambient Air Temperature Thermal Design Power Integrated Heat Spreader Temperature

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects

Tj Tsink XPS ja jc ca

21

Maximum Junction Temperature Heat Sink Temperature X-ray Photoelectron Spectroscopy Overall Thermal Resistance Package Thermal Resistance System Thermal Resistance

References 1. S. J. Dent, L. J. Larson, R. T. Nelson, and D. C. Rash, Semiconductor package and method of preparing the same, US Patent 6940177 (2005). 2. S. F. De Cecco and G. W. Cheshire, Cooling assembly using Heat Spreader, US Patent 9318410 (2016). 3. A. B. Chong, Multi-chip packaging (MCP) or not MCP, in Proc. of the International MultiConference of Engineers and Computer Scientists (IMECS), Vol. II, pp. 1–4 (2012). 4. R. Mahajan, C. P. Chiu, and G. Chrysler, Cooling a microprocessor chip. Proc. IEEE 94, 1476– 1486 (2006). 5. L. A. Polka, H. Kalyanam, G. Hu, and S. Krishnamoorthy, Package technology to address the memory bandwidth challenge for Tera-scale computing. Intel Technol. J. 11, 197–205 (2007). 6. I. Sauciuc, R. Prasher, J. Y. Chang, H. Erturk, G. Chrysler, C. P. Chiu, and R. Mahajan, Thermal performance and key challenges for future CPU cooling technologies, in Proceedings of IPACK2005, ASME InterPACK’05, San Francisco, California, USA (2005). 7. R. Mahajan, C.P. Chiu, and R. Prasher, Thermal interface materials: A brief review of design characteristics and materials, Electronics Cooling 10, 1–12 (2004). 8. R. Prasher, Thermal interface materials: Historical perspective, status, and future directions. Proc. IEEE 94, 1571–1586 (2006). 9. K. L. Mittal and T. Ahsan (Eds.), Adhesion in Microelectronics, Wiley-Scrivener, Beverly, MA (2014). 10. H. K. Dhavaleswarapu, C. M. Jha, S. F. Smith, S. Kothari, B. Bicen, S. K. Saha, and A.  Gupta, Challenges and opportunities in thermal management of multi-chip packages, in Proc. InterPACK, Vol. 1, pp. 1–7 (2015). 11. R. D. Lowe Jr, S. Jain, and J. C. Matayabas, Polymer thermal interface material having enhanced thermal conductivity, US Patent Application 20140264818 (2014). 12. B. Liu, S. Jain, J. C. Viskota, N. R. Raravikar, and J. C. Matayabas, Adhesive polymer thermal interface material with sintered fillers for thermal conductivity in micro-electronic packaging, US Patent Application 2017111945 (2017). 13. A. N. Gent and G. R. Hamed, Fundamentals of adhesion, in Handbook of Adhesives, 3rd ed., I. Skeist (Ed.), pp. 39–73, Springer US, New York (1990). 14. J. Schultz and M. Nardin, Theories and mechanisms of adhesion, in Handbook of Adhesive Technology, A. Pizzi and K. L. Mittal (Eds.), pp. 19–33, Marcel Dekker, New York (1994). 15. D. E. Packham, The mechanical theory of adhesion, in Handbook of Adhesive Technology, 2nd ed., A. Pizzi and K. L. Mittal (Eds.), pp. 69–93, CRC Press, Boca Raton, FL (2003). 16. D. J. Gardner, Theories or mechanisms of adhesion, in Handbook of Adhesive Technology, 3rd ed., A. Pizzi and K. L. Mittal (Eds.), pp. 3–18, CRC Press, Boca Raton, FL (2018).

22 Progress in Adhesion and Adhesives, Volume 4

17. J. W. McBain and D. G. Hopkins, On adhesives and adhesive action. J. Phys. Chem. 29, 188–204 (1925). 18. J. D. Venables, Adhesion and durability of metal-polymer bonds. J. Mater. Sci. 19, 2431–2453 (1984). 19. M. M. Chehimi, A. Azioune, and E. Cabet-Deliry, Acid-base interactions: Relevance to adhesion and adhesive bonding, in Handbook of Adhesive Technology, 2nd ed., A. Pizzi and K. L. Mittal (Eds.), pp. 95–144, CRC Press, Boca Raton, FL (2003). 20. F. M. Fowkes, Acid-base interactions in polymer adhesion, in Physicochemical Aspects of Polymer Surfaces, Vol. 2, K. L. Mittal (Ed.), pp. 583–603, Plenum Press, New York (1983). 21. K. L. Mittal (Ed.), Acid-Base Interactions: Relevance to Adhesion Science and Technology, Vol. 2, CRC Press, Boca Raton, FL (2000). 22. L. Pauling, The Nature of the Chemical Bond and the Structure of Molecules and Crystals: An Introduction to Modern Structural Chemistry, 3rd ed., Cornell University Press, Ithaca, NY (1960). 23. A. Goldschmidt and H. J. Streitberger, Coatings, in BASF Handbook on Basics of Coating Technology, W. Andrew (Ed.), pp. 323–433 (2003). 24. A. C. Santos, A. P. Luz, and S. Ribeiro, Melting temperature and wetting angle of AlN/Dy2O3 and AlN/Yb2O3 mixtures on SiC substrates. J. Mater. Res. 18, 957–962 (2015). 25. B. J. Keene, Review of data for the surface tension of pure metals. Intl. Mater. Reviews 38, 157–192 (1993). 26. H. Neusser and W. Schall, Studies on some promising possibilities for improvement of plywood. 1. Gluing with aminoplasts, Holzforschung Holzverwertung 24, 108–116 (1972). 27. M. Huang and E. R. Pohl, Organofunctional silanes for sealants, in Handbook of Sealant Technology, K. L. Mittal and A. Pizzi (Eds.), pp. 27–49, CRC Press, Boca Raton, FL (2009). 28. K. L. Mittal (Ed.), Silanes and Other Coupling Agents, Vol. 4, CRC Press, Boca Raton, FL (2007). 29. A. Schuberth, M. Goring, T. Lindner, G. Toberling, M. Puschmann, F. Riedel, I. Scharf, K. Schreiter, S. Spange, and T. Lampke, Effect of new adhesion promoter and mechanical interlocking on bonding strength in metal-polymer composites, IOP Conf. Ser. Mater. Sci. Eng. 118, 1–6 (2016). 30. C. Bischof and W. Possart, Adhäsion: Theoretische and Experimentelle Grundlagen, Akademie-Verlag (1983). 31. J. Comyn, Theories of adhesion, in Handbook of Adhesives and Sealants, P. Cognard (Ed.), Vol. 2, pp. 1–50, Elsevier (2006). 32. W. C. Wake, Theories of adhesion and uses of adhesives: A review, Polymer 19, 291–308 (1978). 33. D. E. Packham, The mechanical theory of adhesion- changing perceptions 1925–1991. J. Adhesion 39, 137–144 (1992). 34. H. Y. Lee and J. Qu, Microstructure, adhesion strength and failure path at a polymer/roughened metal interface. J. Adhesion Sci. Technol. 17, 195–215 (2003). 35. D. Schaubroeck, E. Van Den Eeckhout, J. De Baets, P. Dubruel, L. Van Vaeck, and A. Van Calster, Surface modification of a photo-definable epoxy resin with polydopamine to improve adhesion with electroless deposited copper. J. Adhesion Sci. Technol. 26, 2301–2314 (2012). 36. H. Lee, S. M. Dellatore, W. M. Miller, and P. B. Messersmith, Mussel-inspired surface chemistry for multifunctional coatings. Science 318, 426–430 (2007). 37. J. J. Bikerman, Causes of poor adhesion: Weak boundary layers. Ind. Eng. Chem. 59, 41–44 (1967). 38. W. C. Luo, Surface property of passivation layer on integrated circuit chip and solder mask layer on printed circuit board, IEEE Trans. Electron. Packag. Manuf. 26, 345–351 (2003). 39. D. F. O’Kane and K. L. Mittal, Plasma cleaning of metal surfaces. J. Vac, Sci. Technol. 11, 567–569 (1974).

Adhesion Phenomena Pertaining to Thermal Interface Materials and Solder Interconnects

23

40. M. S. Jang, S. W. Ma, J. Song, M. Sung, and Y. H. Kim, Adhesion of NCF to oxidized Si wafers after oxygen plasma treatment, Microelectron. Reliab. 78, 220–226 (2017). 41. A. U. Alam, M. M. R. Howlader, and M. J. Deen, Oxygen plasma and humidity dependent surface analysis of silicon, silicon dioxide and glass for direct wafer bonding. ECS J. Solid State Sci. Technol. 2, 515–523 (2013). 42. S. Bengtsson and P. Amirfeiz, Room temperature wafer bonding of silicon, oxidized silicon, and crystalline quartz. J. Electronic Mater. 29, 909–915 (2000). 43. J. F. Coulon, N. Tournerie, and H. Maillard, Adhesion enhancement of Al coatings on carbon/ epoxy composite surfaces by atmospheric plasma. Appl. Surf. Sci. 283, 843–850 (2013). 44. M. Thomas and K. L. Mittal (Eds.), Atmospheric Pressure Plasma Treatment of Polymers, WileyScrivener, Beverly, MA (2013). 45. M. Strobel, C. S. Lyons, and K. L. Mittal (Eds.), Plasma Surface Modification of Polymers: Relevance to Adhesion, CRC Press, Boca Raton, FL (1994). 46. S. Joshi, B. Arfaei, A. Singh, M. Gharaibeh, M. Obaidat, A. Alazzam, M. Meilunas, L.  Yin, M. Anselm, and P. Borgesen, LGAs vs. BGAs-lower profile and better reliability, in Proc. Surface Mount Technology Association International Conference, pp. 1–4 (2012). 47. Y.T. Chin, P.K. Lam, H.K. Yow, and T.Y. You, Investigation of mechanical shock testing of leadfree SAC solder joints in fine pitch BGA package. Microelectron. Reliab. 48, 1079–1086 (2008). 48. L. Kondrachova, S. Aravamudhan, R. Sidhu, and D. Amir, Fundamentals of the non-wet open BGA solder joint defect formation, in Proc. International Conference on Soldering and Reliability (ICSR), pp. 1–4 (2012). 49. J. Glazer, Microstructure and mechanical properties of Pb-free solder alloys for low-cost electronic assembly: A review. J. Electronic Mater. 23, 693–700 (1994). 50. J.H. Shim, C.S. Oh, B.J. Lee, and D.N. Lee, Thermodynamic assesment of the Cu-Sn system. Zeitschrift Metallkunde. 87, 205–212 (1996). 51. P. Møller, J.B. Rasmussen, S. Kohler, and L.P. Nielsen, Electroplated Tin-Nickel coatings as a replacement for Nickel to eliminate Nickel dermatitis. National Association for Surface Finishing (NASF) Surface Technology White Papers 78, 15–24 (2013). 52. T. T. Mattila, J. Hokka, and M. Paulasto-Krockel, The reliability of microalloyed Sn-Ag-Cu solder interconnections under cyclic thermal and mechanical shock loading. J. Electronic Mater. 43, 4090–4102 (2014). 53. J. H. L. Pang and F. X. Che, Drop impact analysis of Sn–Ag–Cu solder joints using dynamic highstrain rate plastic strain as the impact damage driving force, in Proc 56th IEEE-ECTC Conf, pp. 49–54 (2006). 54. D. Suh, Dong W. Kim, P. Liu, H. Kim, K. A. Weninger, C. M. Kumar, A. Prasad, B. W. Grimsley, and H. B. Tejada, Effects of Ag content on fracture resistance of Sn-Ag-Cu lead-free solders under high-strain rate conditions. Mater. Sci. Eng: A., 460, 595–603 (2007). 55. J. Keller, D. Baither, U. Wilke, and G. Schmitz, Mechanical properties of Pb-free SnAg solder joints. Acta Materialia 59, 2731–2741 (2011). 56. L. Garner, S. Sane, D. Suh, T. Byrne, A. Dani, T. Martin, M. Mello, M. Patel, and R. Williams, Finding solutions to the challenges on package interconnect reliability, Intel Technol. J. 9, 297–308 (2005). 57. V. Vuorinen, T. Laurila, T. Mattila, E. Heikinheimo, and J.K. Kivilahti, Solid state reactions between Cu(Ni) alloys and Sn. J. Electronic Mater. 36, 1355–1362 (2007). 58. D. A. Shnawah, M. F. M. Sabri, and I. A. Badruddin, A review on thermal cycling and drop impact reliability of SAC solder joint in portable electronic products. Microelectron. Reliab. 52, 90–99 (2012).

24 Progress in Adhesion and Adhesives, Volume 4

59. D. Schmitz, S. Shirazi, L. Wentlent, S. Hamasha, L. Yin, A. Qasaimeh, and P. Borgesen, Towards a quantitative mechanistic understanding of the thermal cycling of SnAgCu solder joints, in Proc. Electronic Components and Technology Conference, pp. 371–378 (2014). 60. H. Chen, M. Mueller, T.T. Mattila, J. Li, X. Liu, K-J Wolter, and M Paulasto-Krockel, Localized recrystallization and cracking of lead-free solder interconnections under thermal cycling. J. Mater. Res. 26, 2103–2116 (2011). 61. C. Lee, S. Jung, Y. Shin, and C. Shur, The effect of Bi concentration on wettability of Cu substrate by Sn-Bi solders. Mater. Trans. 42, 751–755 (2001). 62. K. L. Mittal, The role of the interface in adhesion phenomena. Polym. Eng. Sci. 17, 467–473 (1977). 63. X. Zhang, H. Matsuura, F. Tsukihashi, and Z. Yuan, Wettability of Sn-Zn, Sn-Ag-Cu and Sn-Bi-Cu alloys on copper substrates. Mater. Trans. 53, 926–931 (2012). 64. M. McCormack, H.S. Chen, G.W. Kammlott, and S. Jin, Significantly improved mechanical properties of Bi-Sn solder alloy by Ag-doping. J. Electronic Mater. 26, 954–958 (1997). 65. L. E. Felton, C. H. Raeder, and D. B. Knorr, The properties of Tin-Bismuth alloy solders. J. Metals 45, 28–32 (1993). 66. S. K. Tran, D. L. Questad, and B. G. Sammakia, Adhesion issues in flip-chip on organic modules, IEEE Trans. Components Packaging Technol. 22, 519–524 (1999). 67. N. C. Lee and M. Bixenman, Flux technology for lead-free alloys and its impact on cleaning, in Proc. Electronic Components and Technology Conference, pp. 316–322 (2002). 68. C. L Chung, K. S Moon, and C. P. Wong, Influence of flux on wetting behavior of lead-free solder balls during the infrared-reflow process. J. Electronic Mater. 34, 994–1001 (2005). 69. G. J. Sprokel, The use of radioisotopes to determine the chemistry of solder flux. IBM J. Res. Development 5, 218–225 (1961).

2 Influence of Silicon-Containing Compounds on Adhesives for and Adhesion to Wood and Lignocellulosic Materials: A Critical Review Marko Petricˇ University of Ljubljana, Biotechnical Faculty, Department of Wood Science and Technology, Jamnikarjeva ulica 101, 1000 Ljubljana, Slovenia

Abstract Silicon (Si) is the second most abundant element in the earth’s crust and it is no surprise that various contents of this element may be found in plants, and also in different lignocellulosic materials. This may have some undesired effects when products from wood or other lignocellulosic materials are produced and utilised, typically with regards to adhesion of adhesives or surface coatings. However, there are plenty of positive effects of silicon-containing compounds in wood- and lignocellulosebased products, so very often silicon-containing compounds are added into the products on purpose. Sometimes, they might decrease adhesion properties of adhesives and coatings, but their advantages are a major priority and care has to be taken to make sure the drawbacks do not become predominant. Moreover, sometimes the silicon-containing compounds in lignocellulosic products are a necessity. A typical example is silicon-containing compounds to function as coupling agents in wood-polymer composites. The number of publications on Si in lignocellulosics, including wood, is enormous. Even if we narrow our search on this topic only to reports related to adhesion, the number of papers is still astonishing. Therefore, this review does not discuss all recent publications about this topic and it must be rather regarded as a critical review, exhibiting the most important issues that must be taken into account when considering the influence of Si in lignocellulosic materials with regards to adhesion. On the basis of selected papers, it is hoped that the reader will gain a satisfactory insight into the topic. In the first part, after Introduction, there is a brief overview of the most important compounds of Si that are relevant for the field of wood science and technology. In the next section, the influence of the presence of silicon in adhesives or in coatings or substrates on adhesion is discussed. In order to understand these effects, it is essential to know about the interactions/reactions of Si— compounds with cellulose, lignin and wood, which is the topic of the subsequent section. Reactions with the components of lignocellulosics are essential for proper functioning of coupling agents in wood-polymer composites, so these agents are presented in this review as well. Finally, an insight into

Corresponding author: [email protected]

K.L. Mittal (ed.) Progress in Adhesion and Adhesives, Volume 4 (25–76) © 2019 Scrivener Publishing LLC

25

26 Progress in Adhesion and Adhesives, Volume 4

the vast field of lignocellulosic composites is given, with a focus on a variety of different compounds of silicon and their function in the composites. Keywords: Wood, lignocellulose, silicon, silica, silicate, silicone, silane, adhesive, coating, coupling agent, wood-polymer composite

2.1

Introduction

Silicon (Si) is the second most abundant element in the earth’s crust [1-2], making up over a quarter of its mass [3]. Being an element of almost all parent materials from which soils develop, Si is one of the basic components in most soils [4]. Commonly, Si is considered as a ‘Metalloid’ [5]. ‘Metalloids’ is a term that is frequently used to group elements that possess physical and chemical characteristics that are intermediate between those of metals and non-metals. For instance, they conduct heat and electricity better than non-metals but not as well as metals (i.e., they are semiconductors). Also, they generally form amphoteric oxides, being an acid and a base. They are grouped along a diagonal line through the p block of the periodic table from boron to astatine. In addition to silicon, commonly the following elements are also considered metalloids: boron (B), germanium (Ge), arsenic (As), antimony (Sb), tellurium (Te), polonium (Po) and astatine (At) [5]. Si rarely occurs as a pure free element in nature but it rather occurs in more than 370 rock-forming minerals [4], because of the tendency to form strong bonds with oxygen. This is why Si generally exists as silica or silicate compounds [6]. The majority of metalloids are toxic when present in large concentrations, with the exception of Si [5]. Because Si is such a frequent element, the use of its compounds has a long history. For instance, silica for the production of glass dates back to the early stages of human civilisation [5]. Si is a crucial element for various plant species [7]. In the element mobility classification, Si can be regarded either as stationary or mobile element. Mobile Si substances play a primary role in the so-called global biogeochemical cycle, and so also in the so-called global biological cycle [8]. So, Si enters the cycle in the various soil-plant systems which can be accessed via the determination of mobile Si forms in the soil and of the total content of Si in the plant associations [8]. For instance, there was a study on the concentrations of monosilicic acid, polysilicic acids and acid-extractable Si in unmowed meadow, mowed meadow, birch-aspen forest, spruce wood and agricultural land soil-plant systems. Concentrations were measured at soil depths of 0-10, 20-30 and 50-60 cm in a Russian region, south of Moscow [8]. The investigation showed that the biological cycle of Si was characterized by 40 to 80 kg Si ha-1 annually removed from a specific soil type, according to the Russion Classification called the Grey Forest Soil, or Luvisol, with respect to the classification of Food and Agriculture Organization (FAO) of the United Nations [8]. In another study in the field of agriculture it was shown that Si plays a number of important roles in the mineral nutrition of plants [1]. The principal crops in the US collectively can annually take up 9.55 million tons of Si, as estimated from Si content in very young plants on agricultural plantations (shoots), or remove as much as 11.1 million tons from the planted soils [1].

Influence of Silicon-Containing Compounds on Adhesives for and Adhesion 27

Phytogenic Si was defined as Si precipitated in roots, stems, branches, leaves, or needles of plants. Silicon is taken up by plants from soil solution either passively with the mass flow of water or actively, e.g., by rice plants [5]. Although silicon (Si) is generally considered nonessential for plant growth and development, Si uptake by plants can alleviate both biotic and abiotic stresses [2]. Therefore, Si can be found also in wood where it is usually categorised into the group of ash-forming elements [9]. Typically, the ash content (0.2-0.7 %) in four wood species – spruce, pine, birch and aspen - harvested in Scandinavia was low compared to the ash content in the bark tissues (1.9–6.4%) and the foliage (2.4–7.7%). Si in the wood ash was determined to be in the range between 50–190  ppm [9]. On the other hand, even 5000–11,300 ppm of Si was determined in the ash of spruce needles [9]. In another study, the variation of wood elements in wood of six species with anatomically distinct to rather indistinct tree rings from a Thai monsoon forest was investigated [10]. In all species, the X-ray images showed crystals. It was shown that the crystals consist of calcium or silicon in the case of Chukrasia tabularis species, as major elemental components [10]. If we look at some other lignocellulosic tissues, we can see that, for example, the amounts of holocellulose (which is the total polysaccharide fraction of a lignocellulosic material, made up of cellulose and all of the hemicelluloses and is obtained by removing the extractives and the lignin), lignin and cellulose in the date palm rachis and in Posidonia oceanica were similar to those found in softwoods and hardwoods. But, extractives in different solvents and ash contents were relatively high and it was shown that silicon was the major component of ash (17.7%) of P. oceanica. Some other lignocellulosic materials contain even higher contents of silicon [11]. A typical and well-known example is rice husks. The average composition of rice husk according to Ugheoke and Mamat [12] exhibits that there is 17 % of ash in rice husk, consisting of 94 % of Si. Genieva and co-authors report that rice husks contain nearly 20 mass % silica, which is present in hydrated amorphous form [13]. The issue of Si content in wood and lignocellulosic materials becomes highly important with regards to utilisation of biomass for the production of energy and in biorefineries. Biorefineries aim to convert low value biomasses into high value products. The feedstock biomasses are often high-silica agricultural waste products such as rice straw, wheat straw, corn stover, sugar cane bagasse, or empty fruit bunches. But, this causes challenges, since the presence of silica may cause problems in industrial processes, where it forms waterinsoluble precipitates that are hard to remove, block filtration systems, and cause instrumental defects [14]. There are also many studies on undesired role of Si in woody and other lignocellulosic tissues in the field of utilisation of biomass for burning. Just as an example, pointed out by Werkelin and co-authors, there is no doubt that the high Si content in the spruce needles has a large impact on the ash chemistry, and may cause problems in combustion devices. For instance, ash with a high Si content may adhere to the heat transfer surfaces, causing fouling and corrosion [9]. Also, silicon dioxide was identified in the exhaust gases from engines powered using sewage and landfill gas. The production of energy from biogas is severely compromised by its volatile organic silicon compound content. In this case, the Si present is of an anthropogenic origin. The cycle starts with the production and use of poly(dimethylsiloxane)polymer in a wide range of industrial

28 Progress in Adhesion and Adhesives, Volume 4

and domestic applications and its further dispersion into the environment. Silicon dioxide from biogas in power plants deposits on valves, cylinder walls, and liners, causing abrasion and blockage of pistons, cylinder heads, and valves. In gas turbines, siloxane deposits usually form on the nozzles and blades, causing erosion of the turbine blades and subsequently decreasing the operating efficiency [15]. While the presence of Si in biomass can be undesired, as described in the previous paragraph, it can also be regarded as a useful one, finding various useful ways of application, some of them are mentioned in the subsequent text. A perfect example is high content of Si in rice husks. Rice husks have now become a great source of raw biomass material for manufacturing value-added silicon composite products, including silicon carbide, silicon nitride, silicon tetrachloride, magnesium silicide, pure silicon, zeolites, fillers of rubber and plastic composites, cement, adsorbents, and supports of heterogeneous catalysts [13], as also shown in the review by Ugheoke and Mamat [12]. With the aim to improve selected properties of wood and lignocellulose-based products, silicon compounds have found a number of applications, so it is a common practice to add silicon compounds into wood, wood based composites and similar products and not just exploit the benefits of naturally present compounds of silicon. Probably, the most well-known example is the application of silicon compounds to function as coupling agents in various composite materials [16, 17]. By the way in [16], there is an extensive review about the formation of an effective interphase and the characterization of silane-treated surfaces and interphases in composites. Also, in this paper there is also a review about coupling agents in composites (section 4.3.2), however this overview focuses on a variety of composites (e.g. particleboards, medium dense fiberboards (MDFs), wood-flour composites, fiber-reinforced composites, etc.) where the coupling agents have found application. Otherwise, the coupling agents have been extensively studied, e.g. see [17], [18] and [19]. In addition, due to their expected function, silicon-containing compounds are not only added into or on wood or lignocellulosic substrates but also into various other products that are used in the field of wood science and technology, typically into adhesives and surface coatings. For instance, quite recently geopolymer binders have been identified as an emerging class of mineral polymers that can be manufactured from natural raw materials and industrial by-products containing high amounts of silica and alumina in mineral compositions. Geopolymers (mineral polymers resulting from geochemistry or geosynthesis) are materials produced by the activation of aluminosilicate powder components with an alkaline solution and an activator [20, 21]. Another example is modification of soy-based adhesives made from soybean seed with sodium montmorillonite clay [22]. Similarly, silicone resins have recently attracted a great deal of interest, especially in the construction industry, where products such as silicone resin emulsion paints and renders (the first, underlying coats) are gaining in popularity [23]. The aim of this review paper is to provide an overview of the applications of siliconcontaining compounds in the field of wood science and technology and also in relation to some other lignocellulosic compounds, with regards to adhesion between a substrate and an adhesive or surface coating. The other equally important topic in this review is an overview of silicon-containing coupling agents and of silicon compounds in various wood

Influence of Silicon-Containing Compounds on Adhesives for and Adhesion 29

based- and other lignocellulosic composites. Si compounds in lignocellulose substrates and in adhesives and coatings have found numerous other applications. Most probably, the predominant ones are related to hydrophobic properties of wood, wood-based and some other lignocellulosic materials, and fire retardant properties are also a very interesting area. However, the number of publications about these topics is much too large for all to be included in this review.

2.2

2.2.1

An Overview of Compounds and Natural Minerals Containing the Element Si, which are the Most Relevant in the Science and Technology of Lignocellulosics Silica – SiO2

The English word silica has a very broad connotation: it includes silicon dioxide in all its crystalline, amorphous, soluble, or chemically combined forms in which the silicon atom is surrounded by four or six oxygen atoms. This definitely excludes all the organosilicon compounds made by man in which carbon atoms have been linked directly to silicon atoms— commonly referred to as “silicones”, which do not occur in nature [24]. Combined with the oxides of magnesium, aluminum, calcium, and iron, silica forms silicate minerals in rocks and soil [24]. The building block of silica and the silicate structures is the SiO4 tetrahedron, with four O2- at the corners of a regular tetrahedron with a Si4+ at the center cavity or centroid. The O2- is so much larger than the Si4+ that the four oxygens of a SiO4 unit are in mutual contact and the Si4+ is said to be in a tetrahedral hole (Figure 1). Natural silicas can be crystalline, as in quartz, cristobalite, tridymite, coesite, and stishovite, or amorphous, as in opal. Crystalline silica polymorphs are divided according to their network density (SiO2 groups per 1000 Å3) into pyknosils and porosils, and the latter are further divided into clathrasils and zeosils depending on whether the pores are closed or open, i.e., accessible to adsorption [24]. Although in most silicas and silicates, the silicon atom is surrounded by four oxygen atoms, forming the tetrahedral unit [SiO4]4-, a sixfold octahedral coordination of the silicon atom has also been observed in stishovite and coesite. The arrangements of [SiO4]4- and [SiO6]8- and the tendency of these units to form a three-dimensional network structure are fundamental to silica crystal chemistry [24]. The polymorphism of silicas is based on different linkages of the tetrahedral [SiO4]4- units. Quartz has the densest structure, and tridymite and cristobalite have a much more open structure. All three forms exist in - and -forms, which correspond to low- and high temperature modifications, respectively [24]. In the Introduction in the paper of Pabst and Gregorová [25], it is cited from various sources that 12 silica phases exist, and that some other authors have identified 14 and even more than 20 phases of silicon dioxide. Apart from crystalline silicas, the amorphous ones have attracted considerable attention as well. In amorphous silica the bulk structure is determined, as opposed to the crystalline silicas, by a random packing of [SiO4]4- units, which results in a nonperiodic structure (Figure 2) [24].

30 Progress in Adhesion and Adhesives, Volume 4

(a)

(b)

(c)

(d)

Figure 2.1 Representation of tetrahedral coordination of oxygen ions with silicon: (a) ball and stick model, (b) solid tetrahedron, (c) skeletal tetrahedron, and (d) model with packed spheres. (Adapted from Figure 1 in [24]).

One reason for such an importance lies in the fact that amorphous silica and silica based materials have been widely developed for optoelectronics and optical telecommunication technology [26]. So silicon dioxide in its crystalline form, also in the glassy one, is of immense technological importance, with applications ranging from cladding skyscraper facades to manufacturing microchips [27]. Since nanotechnology is rapidly sweeping through all vital fields of science and technology such as electronics, aerospace, defense, medical, and dental, it is normal that considerable attention has been dedicated also to the preparation, characterisation and utilisation of nanosized silica, SiO2 [28]. Nanosilica, also known as the nanoform (700 m2 g−1) and narrow pore size distributions [42]. It was shown that mesoporous silicas free from any metal ions can act as an efficient oxidation catalyst in the process of oxidation of lignin model compounds under microwave activation [40]. Also, silicates (see the next section) are being used in a mesoporous form, as demonstrated by Chen et al. in the case of preparation of Ultralow Density Wood Fiber Composite [43].

2.2.2

Silicates and Clay

According to Belton et al. [44] the term silicate means a chemically specific ion having negative charge (e.g. SiO32-), and the term is also used to describe salts (e.g. sodium silicate,

Influence of Silicon-Containing Compounds on Adhesives for and Adhesion 31

Figure 2.2 Schematic representation of amorphous SiO2. (Adapted from a part of Figure 2 in [24]).

Na2SiO3). The silicates are built up by sharing of oxygen atoms. The various ways in which the SiO4, tetrahedra may be linked form the basis for a structural classification of silicas and silicates [45]. Two different SiO4 groups may share only one oxygen atom, but any or all of the four oxygen atoms on a SiO4 group may be shared with adjacent groups. Sharing of two oxygen atoms per unit yields a chain, three oxygen atoms a sheet, and four oxygen atoms a three-dimensional network. The crystalline silicas quartz, tridymite, and cristobalite are truely network silicates, each silicon being bound to four oxygens and each oxygen being bound to two silicons [24]. Silicates are classified, based on their structure, into six silicate groups: nesosilicates (isolated (SiO4)4- tetrahedra connected only by interstitial cations); sorosilicates (isolated (Si2O7)6- double tetrahedra); cyclosilicates (linked tetrahedra with (SixO3x)2x- rings, which exist as 3-member (Si3O9)6-, 4-member (Si4O12)8- and 6-member (Si6O18)12- rings); inosilicates (interlocked silicate tetrahedra leading to either two single chains of SiO3 or (Si4O11)6- double chains); phyllosilicates (parallel tetrahedral sheets of silica and alumina (Si2O5)2-, (AlSi3O10)5- or (Al2Si2O10)6-) and tectosilicates (three-dimensional networks of silicate tetrahedra with formulas of SiO2, (AlSi3O8)1- or (Al2Si2O8)2-) [43].

32 Progress in Adhesion and Adhesives, Volume 4

Clays are silicates belonging to the phyllosilicate group [46]. Clay is a term for naturally occurring mineral aggregates with various clay minerals contents and degrees of purity. Clays were formed at the site of the parent rocks and were not transported by any of the various agencies such as wind and water. Primary clays (e.g. china clays) are usually found in irregular pockets with unaltered rocks remaining. These deposits are coarse-grained and non-plastic. The formation of clay is a chemical process that is assisted by mechanical breakdown and the separation of fine particles from coarse grains [46]. Bentonite is a term for rock whose dominant clay minerals are smectites, formed through the weathering of volcanic glass, and smectite is a family name, which includes sodium and calcium montmorillonites [46]. Similarly to silicon-containing substances, clays have also found applications in wood technology. For instance, poly(vinyl acetate) (PVAc), a thermoplastic polymer, has poor performance at elevated temperatures and humid conditions as a wood adhesive. Kaboorani and Riedl showed that shear strength of wood joints increased, as measured both in dry and wet states, by adding nano-clay to PVAc. Inclusion of nanoclay improved the thermal stability of PVAc to different degrees depending on nano-clay loading and type [47]. Another example was presented by Cai et al. who prepared woodpolymer nanocomposites from solid aspen wood, water-soluble melamine-urea-formaldehyde (MUF) resin, and silicate nanoclays [48]. Clay in various nanoclay forms can also be used as nanofillers in wood coatings, as presented in an extensive review of Nikolic and coauthors [49].

2.2.3

Silicones

Organosilicates are silicates in which Si-O is bonded to an organic group or even to a more complex organic structure. Such structures are considered to be inorganic because the silicon atom is bonded to carbon through an oxygen atom unlike organosilicon structures where the silicon atom bonds directly to carbon. The organic group is typically a methyl [CH3] or an ethyl [CH2CH3] [45]. Silicones are entirely synthetic polymers containing a repeating Si-O backbone and organic groups attached directly to the silicon atom via silicon-carbon bonds (Figure 3). The most common example is poly(dimethylsiloxane) (PDMS). This is a synthetic polymer with a repeat unit of (CH3)2SiO [45]. The term silicone is actually a misnomer. It was incorrectly thought that the early silicone polymers were silicon-based ketones, hence the contraction silicone. Despite this error, the term is still widely used and accepted [50]. Oxygen is usually bound to two silicon atoms to form the so-called “siloxane linkage”. A series of linked siloxane units form a polysiloxane compound. Silicon-oxygen (Si-O) bonds are remarkably stable, and are inorganic in character [23]. The silicon-carbon (Si-C) bond is also exceptionally stable. The carbon atom bonded to the silicon atom, together with any other chemical groups to which it is attached, is referred to as the organic group R. Thus, the Si-C or Si-R bond may also be referred to as an “organosilicon linkage” [50]. The combination of organosilicon and siloxane linkages has given rise to the terms organosilicon compound, silico-organic compound and organopolysiloxane in connection with silicones [50]. We encounter silicones every day, though we hardly ever notice them. Under the

Influence of Silicon-Containing Compounds on Adhesives for and Adhesion 33

hood, silicone rubber protects the car electronics against moisture and dirt; in car lacquers silicone additives provide gloss; in washing machines, silicone are used as antifoam agents in detergents; in shampoo they give hair its sheen; they provide woolen garments with a typical soft hand, and as silicone resin emulsion paints, they give masonry water repellency, while allowing water vapor and carbon dioxide to diffuse out of its interior [50]. The three basic products, silicone fluids, silicone rubbers and silicone resins, form the basis of more than a thousand silicone products, such as the ones mentioned in previous sentences, and also as greases, release agents, paper-coating agents, to mention but a few [23]. Silicones, like silanes, are the frequently used reagents in decreasing the surface energy for preparing superhydrophobic and superoleophobic surfaces [51]. For instance, superhydrophobic wood was fabricated by spray-coating silicone nanoparticles with a necklace structure, which were prepared by the hydrolytic condensation of methyltrichlorosilane (in toluene with a water concentration of 230 ppm) [51]. Flax fibers coated with two different types and various amounts of silicones were used to make flax–polyurethane composites. Flax– polyurethane composites with silicone interphases demonstrated improvement in impact toughness up to 100% [52]. Another example is an application of an amino-silicone (AS; amino-poly(dimethylsiloxane) micro-emulsion for preservation of wood against basidiomycetes [53]. Scots pine (Pinus sylvestris L.) samples were vacuum impregnated with quaternary ammonium (quat)-silicone micro-emulsion (P1

Average

Bi-adhesive joint P3 P3 P3>P2>P1

r

y

Average of the brittle adhesive

Average of the bi-adhesive

Figure 3.10 Schematic shear stress distributions at failure in mono- and bi-adhesive joints (adapted from [33]).

86 Progress in Adhesion and Adhesives, Volume 4

Figure 3.11 Mono and bi-adhesive joints fabrication method (adapted from [34]).

The use of a bi-adhesive bondline may provide significant advantages in applications where large temperature differences exist. Reduction of the joint strength may be prevented over large temperature ranges encountered in flight at high altitudes. da Silva and Adams [35] studied adhesive joints with a bi-adhesive over a wide temperature range (–55 to 200 °C). Their joint strength predictions showed that for identical adherends (titanium/ titanium), the bi-adhesive technique was of only little benefit. Experiments were then presented for titanium/titanium and titanium/composite double lap joints. It was shown that for a joint with dissimilar adherends, the combination of two adhesives gives a better performance over the temperature range than a high temperature adhesive alone. Marques et al. [36] studied a ceramic–metal joint, representative of the thermal protection systems of some aerospace vehicles. A comparison was performed between the experimentally tested bi-adhesive joint configuration and two alternative configurations: a symmetrical square configuration with silicone fully surrounding the epoxy central section, and a ramped configuration with a tapered ceramic substrate and a tapered adhesive layer. It was concluded that the alternative geometries allowed the introduction of additional flexibility on the joint but at the cost of lower failure load. Figure 12 shows the evolution of a bi-adhesive joint strength in a wide range of temperatures [36]. It is known that adhesively bonded patches have problem of stress concentration at the edges where crack initiation is prone to occur. Marques and da Silva [37] tried to overcome this problem by the use of a taper and a spew fillet at the end of the patch and using a biadhesive technique where a ductile adhesive is placed at the edges of the patch (Figure 13).

Recent Advances in Adhesively Bonded Lap Joints Having Bi-Adhesive and Modulus-Graded 87

Joint strength

High temperature adhesive (LTA)

Bi-adhesive Low temperature adhesive (LTA) –65˚C

Temperature

100˚C

Bi-adhesive shear stress distribution

LTA HTA LTA

Figure 3.12 Bi-adhesive joint strength in a wide range of temperatures (adapted from [36]).

Bi-adhesive without taper 5.7mm

Stiff adhesive Flexible adhesive and 35 Bi-adhesive with taper 5.7mm

Figure 3.13 Bi-adhesive joint with and without taper (adapted from [37]).

They concluded that a taper angle is beneficial only for the brittle adhesive, and the use of two adhesives is advantageous for the taperless configuration. Akpinar et al. [38] investigated the normal and shear stress distributions in a bi-adhesive bonded T-joint using a non-linear 3D FE analysis. In addition, experimental studies were also performed on two different types of T-joint samples (mono-adhesive T-joint and biadhesive T-joint). For the bi-adhesive T-joint, a flexible adhesive was applied to the ends of the overlap, while the central region was bonded with a stiff adhesive (Figure 14). It was observed that a higher degree of shear stress was transferred into the inner regions of the bi-adhesive bondline, and a decrease was observed in the shear stress values.

88 Progress in Adhesion and Adhesives, Volume 4

Stiff adhesive

Flexible adhesive

Flexible adhesive

Figure 3.14 T-joint with bi-adhesive (adapted from [38]).

The influence of stress-reduction methods on the strength of adhesively bonded joints composed of brittle adherends was studied by Valle´e et al. [39]. Experimental and numerical investigations were carried out on two types of adherends: fibre reinforced polymers and timber, considering three different stress-reduction methods: adhesive rounding, chamfering, and adhesive grading. A probabilistic strength prediction method was also applied. They showed that stress-reduction and strength increase of adhesively bonded joints are greatly affected by the brittleness of the adherends. Fabrication of bi-adhesive joints is more complex than mono-adhesive joints. New dispensing process must be developed to apply the adhesives with different characteristics along the bondline. Chiminelli et al. [40] analyzed the potential of the mixing adhesives approach for an SLJ and developed a computational fluid-dynamic (CFD) model. They used a special device developed for assembling bi-adhesive joints (Figure 15). This technique was experimentally validated through joint tests. Experimental results showed that the ultimate load per unit depth for both mono-adhesive configurations is similar, but for the bi-adhesive one there is about a 70% improvement with respect to the flexible one.

3.3

Modulus-Graded Bondline

3.3.1 Numerical and Analytical Studies A simple analytical model by making use of power series expansions to study the performance of the modulus-graded joints was developed by Carbas et al. [41]. They observed that the joints having modulus-graded bondline show a higher strength when compared with the joints with homogeneous adhesive properties along the overlap and this behaviour becomes more pronounced with an increase in the overlap length. Figure 16 shows the boundary conditions and loadings in their model. An efficient model for the stress analysis of modulus-graded adhesive joints was proposed by Stein et al. [42]. Two differential equations for the displacements were derived for the case of an axially loaded SLJ. The differential equations were solved using a power series approach. Stein et al. [43] compared the existing analytical models for planar modulus-graded adhesive SLJs. In addition, a novel approach and possible extensions of the presented approaches are

Recent Advances in Adhesively Bonded Lap Joints Having Bi-Adhesive and Modulus-Graded 89

ff sti of re le xtu xib Mi d fle es v an hesi ff ve ad Sti hesi ad

le xib ve Fle hesi ad

Lower adhere n displa cemen d t

Figure 3.15 Dispenser head for fabrication of five-band joints (adapted from [40]).

Figure 3.16 Model used in FE analysis (adapted from [41]).

outlined and discussed in their study. They concluded that regarding the reduction of the peak adhesive stresses, the joint configuration and the adhesive Young’s modulus profile are of minor importance, whereas the ratio between the maximum and minimum values of the adhesive elastic modulus has a significant effect on the peak adhesive stresses. Composite and metallic tubular structures are used to transport petroleum, oil, fluids, liquid chemicals and gases. Tubular structures are mostly assembled using welding, fastening and bolting. It is well known that fastening and bolting increase the weight of assembly and welding damages micro-structure. A bi-adhesive joint with structural adhesives can be another alternative to these joining methods for the tubular structures. Kumar [44] presented a theoretical framework for the stress analysis of the modulusgraded tubular lap joints based on a variational principle. A multi-step variation of the adhesive modulus was considered so as to reduce the shear and peel stress concentrations at the ends of the overlap. His model including two tubes of different materials and dimensions is shown in Figure 17. In this study, the varying modulus of the adhesive along the bond length was expressed by suitable functions.

90 Progress in Adhesion and Adhesives, Volume 4 Modulus-graded bondline

P t1

t2

P

Inner tube

Outer tube ta 2l

Figure 3.17 Modulus-graded bondline tubular joint (adapted from [44]).

Spaggiari and Dragoni [45] performed an analytical stress analysis of a tubular SLJ under torsion with a modulus-graded adhesive. The feasibility limits were discussed considering the maximum change in the shear modulus as well as in the adhesive thickness, while the influence of the overlap length was analyzed showing that having short overlap helps in achieving optimum joint configuration. Nimje and Panigrahi [46] studied the modulus-graded adhesively bonded tubular lap joints of laminated fiber reinforced plastic composites under varying loadings using 3D geometrically nonlinear FE analysis. Strain energy release rate (SERR) is utilized as the characterizing and governing parameter for assessing damages originating from the critical location. The damage propagation behavior of the tubular joints with pre-embedded damages at the critical location was compared between conventional mono-modulus adhesives and modulus-graded adhesives with appropriate material gradation profile. Stein  et al. [47] presented a new model for the stress analysis of modulus-graded adhesive lap joints with composite adherends. Their model is applicable to various joint configurations such as SLJs, L-joints, T-joints, reinforcement patches, corner joints or balanced double lap joints. Nimje and Panigrahi [48] performed a 3D non-linear FE analysis to assess the structural behavior of adhesively bonded double supported T-joints of laminated FRP composites having embedded interfacial failures. Depending on the SERR magnitude, it was observed that the interfacial failure propagates under mixed mode condition. Nimje and Panigrahi [49] carried out the 3D stress analysis of modulus-graded adhesively bonded T-joints made of laminated fiber reinforced polymeric composite using geometrically non-linear FE analysis. Linear and exponential material gradation function profiles were used to grade the adhesive layer in the T-joint. An improved analytical model for the stress analysis of interface stiffness graded axisymmetric adhesive joints was presented by Kumar and Scanlan [50]. The joint is composed of similar or dissimilar anisotropic and/or isotropic adherends and a modulus-graded adhesive bondline. They observed that the shear and normal stress concentrations at the overlap ends are much less than those of mono-modulus bondline adhesive joints under the same axial load. Stein et al. [51] proposed an analytical model for modulus-graded adhesive joints with composite adherends under mechanical and thermal loadings. Asymmetric laminated adherends were studied using the first-order shear deformation theory and taking into account

Recent Advances in Adhesively Bonded Lap Joints Having Bi-Adhesive and Modulus-Graded 91

bending-extension coupling. It was shown that the predicted stress distributions agreed well with numerical results. Khan and Kumar [52] employed the principle of minimum complementary energy with a variational method to theoretically determine the stresses in each of the constituents of 3D axisymmetric single-side composite patch repair assembly. In their study, a specific gradation scheme was demonstrated to minimize peak stresses in the adhesive. Kumar and Khan [53] presented a shear-lag model for stress transfer through an adhesive layer of variable stiffness. They investigated the effect of such inhomogeneous bondline on the interfacial shear stress distribution in comparison to that of a homogeneous bondline anchor subjected to monotonic axial tension. Specific closed-form expressions for interfacial stresses were given for constant and linear distribution of shear modulus.

3.3.2

Experimental Studies

Breto et al. [54] focused on the modulus-graded adhesive joints to improve the strength of aluminum/composite joints under a shear load. They showed that the way in which the continuous grading is approximated by means of bands strongly affects the level of improvement finally attained. Carbas et al. [55] used induction heating and cooling in order to obtain a modulus-graded adhesive joint. They compared the performances of the joints cured isothermally and gradually. They found that the modulus-graded joints have the highest failure load and similar displacement as the joints cured isothermally at high temperature. Carbas et al. [56] investigated the fracture behavior of wood beams repaired with an adhesively bonded carbon fiber reinforced polymer. In their study, the repair of wood structures with carbon fiber reinforced polymer was made using a homogeneous cure and a graded cure. The graded cure was performed by induction heating. Cohesive zone model was used to simulate the crack initiation and propagation. The results of the bending tests on the wood beams showed that graded joints can be used to improve the strength and reliability of repaired beams. The use of different amounts of additives in the adhesive bondline is one of the techniques applied to grade the bondline properties. An appropriate additive amount in the bondline length is crucial to increase the strength of joint. Carbas et al. [57] used different amounts of carbon black along the bondline length in order to obtain a modulus-graded joint. Figure 18 shows a schematic variation of the carbon black amount along the overlap length. They also used an analytical model to predict and assess the possible effectiveness of the graded joint concept. They concluded that the graded joints with carbon black have the highest failure load when compared with the joints with a homogeneous bondline. Stapleton et al. [58] constructed an analytical model to compare the stresses in a buttend joint configuration with four different functions of modulus-graded adhesive. These functions are constant (mono-adhesive), discrete (bi-adhesive), linear, and exponential. The analytical findings are complemented with an experimental ‘‘proof-of-concept’’ testing to illustrate the benefits of modulus-graded adhesives. The adhesive used was AF 163-2k (3M Scotch-Weld™, United States). The adhesive was graded by adding different volume percentages of glass beads (Figure 19). It was shown that there are significant stress reductions possible by using modulus-graded adhesives. Additionally, it was concluded that grading did not result in an increase in strength when the failure without grading occurred in the adherend.

92 Progress in Adhesion and Adhesives, Volume 4

Carbon Black amount (Vol %)

20

15

10

5

0 0

0.25

0.5

0.75

1

Normalized overlap length

Figure 3.18 Carbon black distribution along the bondline (adapted from [57]).

Doubler

Adherend

Adhesive Glass Beads

Figure 3.19 Diagram of single strap joint with glass beads (adapted from [58]).

Kawasaki et al. [59] proposed a novel method to fabricate modulus-graded adhesive joints. Two types of second-generation acrylic adhesives (SGAAs), one that is flexible and one that is rigid and brittle, were selected. It was concluded that the mechanical properties of SGAAs could be controlled by varying the mixing ratio of the SGAAs. Stapleton et al. [60] considered the geometry of the gradation and the material properties of the adhesive at different gradation levels. They created a new adhesive gradation system by using a polyurethane-based adhesive with varying amounts of acrylate, and a numerical analysis was performed to determine the potential advantages of the adhesive gradation. Figure 20a shows the modulus varying along the bondline with two linear symmetric gradations. The other grading type performed with tanh function is shown in Figure 20b. Graded adhesives with tanh function showed higher potential for improvement in maximum load over the outermost adhesive based system. They concluded that for modulus-graded adhesive joints, the design and application should be tightly coupled with predictive models because the gains can be either negligible or nonexistent without the right design. 3D printing can be an alternative method to other methods using nano-particles and induction heating for gradation of bondline properties. Kumar et al. [61] studied the mechanics of

Recent Advances in Adhesively Bonded Lap Joints Having Bi-Adhesive and Modulus-Graded 93 4500

4500

x

x

P,D 2l 4000 Ea (MPa)

Ea (MPa)

4000

3500

3500

3000

3000

2500

P,D 2l

Lin + Min 0

0.2

0.4 0.6 x/2l (mm/mm) (a)

0.8

Lin – Max 1

2500 0

Tanh – Max

Tanh + Min 0.2

0.4 0.6 x/2l (mm/mm)

0.8

1

(b)

Figure 3.20 Functionally-graded adhesive bondline: (a) linear function and (b) tanh function (adapted from [60]).

Stiff adhesive

Flexible adhesive

Constant modulus adhesive

Stiff adhesive contributes to center stiffness

Flexible adhesive contributes to edge compilance

Compilance-tailored adhesive

Figure 3.21 Spatially tailored adhesive compliance in an SLJ (adapted from [61]).

deformation and failure of tensile loaded SLJs with a compliance-tailored adhesive. Utilizing 3D printing, the modulus of the adhesive was spatially varied along the bond length. The different configurations tested are shown in Figure 21. It was seen that stiff adhesive contributes to stiffness in the middle and flexible adhesive contributes to compliance at the ends.

94 Progress in Adhesion and Adhesives, Volume 4

3.4

Summary

Bi-adhesive and modulus-graded adhesive joints are alternative stress-reduction techniques for the classical bonding applications. Here, some of the most recent scientific literature regarding the use of these joints was summarized. Results reported can be easily summarized that the adhesive type and the bond-length ratio (ratio of length of flexible adhesive to length of stiff adhesive) play an important role in the bi-adhesive bondline. The joints with bi-adhesive and modulus-graded bondlines showed a high strength and a low stress concentration when compared with the joints with mono-adhesives along the overlap.

Acknowledgement The authors are grateful to reviewers for their valuable comments on this article.

Nomenclature A,B a D Ea, Hp Hr Wp Wr L l n P s ta t1 t2 , y r

: : : : : : : : : : : : : : : : : : : : : :

Constants of integration Half-length of crack Displacement Adhesive Young’s modulus Plate length Patch height Plate width Patch width Un-bonded adherend area Half-length of bondline Finite number of discrete adhesives Load Half-length of stiff adhesive Adhesive thickness Inner tube thickness Outer tube thickness Spew fillet angle Peel stress Constants Shear stress Shear yield strength Shear strength

References 1. R.D. Adams and N.A. Peppiatt, Effect of Poisson’s ratio strains in adherends on stresses of an idealized lap joint. J. Strain Anal. Eng. Design 8, 134–139 (1973). 2. M.Y.Tsai, D.W. Oplinger, and J. Morton, Improved theoretical solutions for adhesive lap joints. Int. J. Solids Struct. 35, 1163–1185 (1998).

Recent Advances in Adhesively Bonded Lap Joints Having Bi-Adhesive and Modulus-Graded 95

3. R.D. Adams and N.A. Peppiatt, Stress analysis of adhesive bonded lap joints. J. Strain Anal. Eng. Design 9, 185–196 (1974). 4. A. Kimiaeifar, E. Lund, O.T. Thomsen, and J.D. Sørensen, Asymptotic sampling for reliability analysis of adhesive bonded stepped lap composite joints. Eng. Struct. 49, 655–663 (2013). 5. J.H. Kim, B.J. Park, and Y.W. Han, Evaluation of fatigue characteristics for adhesively bonded composite stepped lap joint. Composite Struct. 66, 69–75 (2004). 6. H.S. Kim, S.J. Lee, and D.G. Lee, Development of a strength model for the cocured stepped lap joints under tensile loading. Composite Struct. 32, 593–600 (1995). 7. L.J.H. Smith, Adhesive-bonded scarf and stepped lap joints. Technical report NASA CR 112237 (1973). 8. S.B. Kumar, S. Sivashanker, A. Bag, and I. Sridhar, Failure of aerospace composite scarf-joints subjected to uniaxial compression. Mater. Sci. Eng. A 412, 117–122 (2005). 9. Y. Hua, L. Gu, and M. Trogdon, Three-dimensional modeling of carbon/epoxy to titanium singlelap joints with variable adhesive recess length. Int. J. Adhesion Adhesives 38, 25–30 (2012). 10. G. Belingardi, L. Goglio, and A. Tarditi, Investigating the effect of spew and chamfer size on the stresses in metal/plastics adhesive joints. Int. J. Adhesion Adhesives 22, 273–282 (2002). 11. S.V. Nimje and S.K. Panigrahi, Stress and failure analyses of functionally graded adhesively bonded joints of laminated FRP composite plates and tubes: A critical review. Rev. Adhesion Adhesives 5, 162–194 (2017). 12. S.K. Panigrahi and R.N. Das, Study and analysis of damages in functionally graded adhesively bonded joints of laminated FRP composites: A critical review. Rev. Adhesion Adhesives 4, 152–165 (2016). 13. M.K. Apalak, Functionally graded adhesively bonded joints. Rev. Adhesion Adhesives 2, 56–84 (2014). 14. C. Raphael, Variable-adhesive bonded joints. Appl. Polym. Symp. 3, 99–108 (1965). 15. O. Volkersen, Rivet strength distribution in tensile-stressed rivet joints with constant cross-section. Luftfahrorschung 15, 41–47 (1938). 16. P.J.C. das Neves, L.F.M. da Silva, and R.D. Adams, Analysis of mixed adhesive bonded joints Part I: Theoretical formulation. J. Adhesion Sci. Technol. 23, 1–34 (2009). 17. P.J.C. das Neves, L.F.M. da Silva, and R.D. Adams, Analysis of mixed adhesive bonded joints Part II: Parametric study. J. Adhesion Sci. Technol. 23, 35–61 (2009). 18. S.A. Yousefsani and M. Tahani, Relief of edge effects in bi-adhesive composite joints. Composites B 108, 153–163 (2017). 19. H. Özer and Ö. Öz, A comparative evaluation of numerical and analytical solutions  to the biadhesive single-lap joint. Math. Probl. Eng. Article ID 852872, 16 pages (2014). 20. O. Bavi, N. Bavi, and M. Shishesaz, Geometrical optimization of the overlap in mixed adhesive lap joints. J. Adhesion 89, 948–972 (2013). 21. R. Breto, A. Chiminelli, M. Lizaranzu, and R. Rodríguez, Study of the singular term in mixed adhesive joints. Int. J. Adhesion Adhesives 76, 11–16 (2017). 22. S. Temiz, Application of bi-adhesive in double-strap joints subjected to bending moment. J. Adhesion Sci. Technol. 20, 1547–1560 (2006). 23. F.-R. Kong, M. You, and X.-L. Zheng, Three-dimensional finite element analysis of the stress distribution in bi-adhesive bonded joints. J. Adhesion 84, 105–124 (2008). 24. S. Kumar and P.C. Pandey, Behaviour of bi-adhesive joints. J. Adhesion Sci. Technol. 24, 1251–1281 (2010). 25. G. Dean, L. Crocker, B. Read, and L. Wright, Prediction of deformation and failure of rubbertoughened adhesive joints. Int. J. Adhesion Adhesives 24, 295–306 (2004). 26. I. Pires, L. Quintino, and R.M. Miranda, Numerical simulation of mono- and bi-adhesive aluminium lap joints. J. Adhesion Sci. Technol. 20, 19–36 (2006).

96 Progress in Adhesion and Adhesives, Volume 4

27. H. Fekirini, B. B. Bouiadjra, M. Belhouari, B. Boutabout, and B. Serier, Numerical analysis of the performances of bonded composite repair with two adhesive bands in aircraft structures. Composite Struct. 82, 84–89 (2008). 28. A.S. Bouchikhi, A. Megueni, S. Gouasmi, and F.B. Boukoulda, Effect of mixed adhesive joints and tapered plate on stresses in retrofitted beams bonded with a fiber-reinforced polymer plate. Mater. Design 50, 893–904 (2013). 29. A. Afkar and M.N. Camari, Finite element analysis of mono- and bi-adhesively bonded functionally graded adherend. J. Failure Anal. Preven. 14, 253–258 (2014). 30. A. Çalık and S. Yıldırım, Effect of adherend recessing on bi-adhesively bonded single-lap joints with spew fillet. Sadhana 42, 317–325 (2017). 31. I. Pires, L. Quintino, J.F. Durodola, and A. Beevers, Performance of bi-adhesive bonded aluminium lap joints. Int. J. Adhesion Adhesives 23, 215–223 (2003). 32. M.D. Fitton and J.G. Broughton, Variable modulus adhesives: An approach to optimised joint performance. Int. J. Adhesion Adhesives 25, 329–336 (2005). 33. L.F.M. da Silva and M.J.C.Q. Lopes, Joint strength optimization by the mixed-adhesive technique. Int. J. Adhesion Adhesives 29, 509–514 (2009). 34. Ö. Öz and H. Özer, An experimental investigation on the failure loads of the mono and bi-adhesive joints. J. Adhesion Sci. Technol. 31, 2251–2270 (2017). 35. L.F.M. da Silva and R.D. Adams, Adhesive joints at high and low temperatures using similar and dissimilar adherends and dual adhesives. Int. J. Adhesion Adhesives 27, 216–226 (2007). 36. E.A.S. Marques, R.D.S.G. Campilho, and L.F.M. da Silva, Geometrical study of mixed adhesive joints for high-temperature applications. J. Adhesion Sci. Technol. 30, 691–707 (2016). 37. E.A.S. Marques and L.F.M. da Silva, Joint strength optimization of adhesively bonded patches. J. Adhesion 84, 915–934 (2008). 38. S. Akpinar, M.D. Aydin, and A. Özel. A study on 3-D stress distributions in the bi-adhesively bonded T-joints. Appl. Math. Model 37, 10220–10230 (2013). 39. T. Valle´e, T. Tannert, J. Murcia-Delso, and D.J. Quinn, Influence of stress-reduction methods on the strength of adhesively bonded joints composed of orthotropic brittle adherends. Int. J. Adhesion Adhesives 30, 583–594 (2010). 40. A. Chiminelli, R. Breto, S. Izquierdo, L. Bergamasco, E. Duvivier, and M. Lizaranzu, Analysis of mixed adhesive joints considering the compaction process. Int. J. Adhesion Adhesives 76, 3–10 (2017). 41. R.J.C. Carbas, L.F.M. da Silva, M.L. Madureira, and G. W. Critchlow, Modelling of functionally graded adhesive joints. J. Adhesion 90, 698–716 (2014). 42. N. Stein, H. Mardani, and W. Becker, An efficient analysis model for functionally graded adhesive single lap joints. Int. J. Adhesion Adhesives 70, 117–125 (2016). 43. N. Stein, J. Felger, and W. Becker, Analytical models for functionally graded adhesive single lap joints: A comparative study. Int. J. Adhesion Adhesives 76, 70–82 (2017). 44. S. Kumar, Analysis of tubular adhesive joints with a functionally modulus graded bondline subjected to axial loads. Int. J. Adhesion Adhesives 29, 785–795 (2009). 45. A. Spaggiari and E. Dragoni, Regularization of torsional stresses in tubular lap bonded joints by means of functionally graded adhesives. Int. J. Adhesion Adhesives 53, 23–28 (2014). 46. S.V. Nimje and S.K. Panigrahi, Strain energy release rate based damage analysis of functionally graded adhesively bonded tubular lap joint of laminated FRP composites. J.  Adhesion 93, 389–411(2017). 47. N. Stein, P. Weißgraeber, and W. Becker, Stress solution for functionally graded adhesive joints. Int. J. Solids Struct. 97–98, 300–311 (2016).

Recent Advances in Adhesively Bonded Lap Joints Having Bi-Adhesive and Modulus-Graded 97

48. S.V. Nimje and S.K. Panigrahi, Interfacial failure analysis of functionally graded adhesively bonded double supported tee joint of laminated FRP composite plates. Int. J. Adhesion Adhesives 58, 70–79 (2015). 49. S.V. Nimje and S.K. Panigrahi, Numerical simulation for stress and failure of functionally graded adhesively bonded tee joint of laminated FRP composite plates. Int. J. Adhesion Adhesives 48, 139–149 (2014). 50. S. Kumar and J.P. Scanlan, On axisymmetric adhesive joints with graded interface stiffness. Int. J. Adhesion Adhesives 41, 57–72 (2013). 51. N. Stein, P.L. Rosendahl, and W. Becker, Homogenization of mechanical and thermal stresses in functionally graded adhesive joints. Composites B 111, 279–293 (2017). 52. M.A. Khan and S. Kumar, Interfacial stresses in single-side composite patch repairs with material tailored bondline. Mech. Adv. Mater. Struct. 25, 304–318 (2018). 53. S. Kumar and M.A. Khan, A shear-lag model for functionally graded adhesive anchors. Int. J. Adhesion Adhesives 68, 317–325 (2016). 54. R. Breto, A. Chiminelli, E. Duvivier, M. Lizaranzu, and M.A. Jiménez, Finite element analysis of functionally graded bond-lines for metal/composite joints. J. Adhesion 91, 920–936 (2015). 55. R.J.C. Carbas, L.F.M. da Silva, and G.W.Critchlow, Adhesively bonded functionally graded joints by induction heating. Int. J. Adhesion Adhesives 48, 110–118 (2014). 56. R.J.C. Carbas, G.M.S.O. Viana, L.F.M. da Silva, and G.W. Critchlow, Functionally graded adhesive patch repairs of wood beams in civil applications. J. Composite Constr. 19, 1–11 (2014). 57. R.J.C. Carbas, L.F.M. da Silva, and L.F.S. Andrés, Functionally graded adhesive joints by graded mixing of nanoparticles. Int. J. Adhesion Adhesives 76, 30–37 (2017). 58. S.E. Stapleton, A.M. Waas, and S.M. Arnold, Functionally graded adhesives for composite joints. Int. J. Adhesion Adhesives 35, 36–49 (2012). 59. S. Kawasaki, G. Nakajima, K. Haraga, and C. Sato, Functionally graded adhesive joints bonded by honeymoon adhesion using two types of second generation acrylic adhesives of two components. J. Adhesion 92, 517–534 (2016). 60. S.E. Stapleton, J. Weimer, and J. Spengler, Design of functionally graded joints using a polyurethane-based adhesive with varying amounts of acrylate. Int. J. Adhesion Adhesives 76, 38–46 (2017). 61. S. Kumar, B.L. Wardle, and M.F. Arif, Strength and performance enhancement of bonded joints by spatial tailoring of adhesive compliance via 3D printing. ACS Appl. Mater. Interfaces 9, 884−891 (2017).

4 Adhesion between Compounded Elastomers: A Critical Review K. Dinesh Kumar1, M.S. Satyanarayana1, Ganesh C. Basak2 and Anil K. Bhowmick2* 1

Department of Materials Science and Engineering, Indian Institute of Technology Patna, Patna 801106, India 2 Rubber Technology Centre, Indian Institute of Technology Kharagpur, Kharagpur 721302, India

Abstract Rubber to rubber adhesion is very unique in a way that both joining substrates are flexible and of relatively low strength. Bonding of two different rubber compounds either through vulcanization or by using an adhesive compound bears significant importance from both scientific and technological points of view. This problem has significant implication especially in four areas of the rubber industry. The first is the adhesion between an unvulcanized rubber compound and an unvulcanized rubber compound which has direct relevance in the manufacturing process of tires and conveyor belts where different layers of unvulcanized rubbers are joined together and co-crosslinked. The second is the co-vulcanization between similar or dissimilar rubbers. The third is the adhesion between an unvulcanized rubber compound and a vulcanized rubber compound. This is generally practiced for repairing the damaged vulcanized rubber portions in a conveyor belt by pressing the unvulcanized rubber compounds against the damaged vulcanized rubber portions using heat and pressure. The fourth is the adhesion between a vulcanized rubber compound and a vulcanized rubber compound using a special bonding agent for retreading of tires.  There are wide varieties of distinct variables affecting rubber to rubber bonding. The variables affecting the bonding between compounded rubber and compounded rubber have been identified and discussed in detail in this review. In addition, the precise nature of the interactions at interfaces of various rubber to rubber joints has been elaborated. The recent developments in the area of rubber to rubber bonding are collated and compared with old literature to present the current understanding of the subject. Keywords: Adhesion, interface, rubber, elastomer, vulcanization, adhesive

*Corresponding author: [email protected]

K.L. Mittal (ed.) Progress in Adhesion and Adhesives, Volume 4 (99–192) © 2019 Scrivener Publishing LLC

99

100 Progress in Adhesion and Adhesives, Volume 4

4.1

Introduction

In the previous review, we have discussed in detail the theoretical and practical aspects of adhesion between two unvulcanized elastomers [1]. The adhesion was primarily dependent on the mobility of elastomer chains across the interface. The unrestricted mobility of the elastomer chains across the interface of two unvulcanized elastomers joints is due to the absence of any crosslinking agents. However, in practical scenario, elastomers can only be used after crosslinking the molecular chains by crosslinking agents and compounding, which in turn imposes severe restrictions on the molecular mobility across the interface. Therefore, it is obviously necessary to investigate the nature of adhesion between two elastomers where the molecular mobility across the interface from one side or either side is partially or fully restricted by manipulating the crosslinking reactions. This problem has such an important bearing upon three major areas in the rubber industry such as: 1. In the case of rubbers used in tires and conveyer belts, different layers of unvulcanized rubbers filled with standard compounding ingredients and crosslinking agents are joined together and co-crosslinked under pressure and temperature in a molding press. In this case mobility of elastomer chains across the interface takes place from both sides initially. The nature and level of crosslinking agents and other ingredients play a dominant role in deciding the adhesion strength between the two co-crosslinked elastomer specimens in the end. The lower the crosslinking density at the interface, the poorer will be the bond strength and there will be a high risk of failure in such laminates. Also, mobility of the elastomer chains may decrease due to the presence of highly interacting ingredients. 2. Often vulcanized rubber joints need repairing / joining during applications. For example, a conveyor belt when it runs several miles, it fails at certain parts of the belt or at the joints. This requires repairing of the damaged vulcanized conveyer belt layer using a rubber compound which is unvulcanized (filled with crosslinking agents). The unvulcanized rubber compound (filled with crosslinking agents) is generally pressed over the damaged vulcanized rubber layer and crosslinked by application of heat and pressure. The crosslinking system in the unvulcanized rubber compound will help in stitching the damaged vulcanized rubber layer. In this case, the mobility of elastomer chains from one side is restricted. It is a challenge to bond rubber layers in the absence of such molecular mobility. 3. In the last but not the least case, often rubber articles are vulcanized and need joining together. In this case, the mobility of elastomer chains from both sides is restricted. The bonding between these two layers takes place with the help of special bonding agents that stitch both surfaces and improve the bond strength. A classical example is retreading of tires.

Adhesion between Compounded Elastomers: A Critical Review 101

The current review will discuss the following cases in detail: a. Adhesion between unvulcanized rubber (filled with crosslinking agents) and unvulcanized rubber (filled with crosslinking agents) by co-crosslinking b. Adhesion between partially vulcanized rubber (filled with crosslinking agents) and partially vulcanized rubber (filled with crosslinking agents) by co-crosslinking c. Adhesion between vulcanized rubber and partially vulcanized rubber (filled with crosslinking agents) d. Adhesion between vulcanized rubber and unvulcanized rubber (filled with crosslinking agents) e. Adhesion between vulcanized rubber and vulcanized rubber Both the practical and theoretical aspects of the above cases will be covered using the state-of-the-art literature. The review will also critically analyze the work done by various authors in this field and point out the areas for future research.

4.2

Co-crosslinking

Vulcanization is a process generally applied to rubbery or elastomeric materials. Vulcanization can be defined as a process which increases the retractile force and reduces the amount of permanent deformation remaining after removal of the deforming force [2]. Thus, vulcanization increases elasticity while it decreases plasticity [2]. It is generally accomplished by the formation of a crosslinked molecular network (Figure 1). Vulcanised network

Unvulcanised

Sulphur

Sulphur

Rubber molecules

Figure 4.1 Network formation through vulcanization.

Sx

Crosslinks

102 Progress in Adhesion and Adhesives, Volume 4

Vulcanization, thus, is a process of chemically producing network junctures by the insertion of crosslinks between rubber chains [2]. A crosslink may be a group of sulfur atoms in a short chain, a single sulfur atom, a carbon to carbon bond, a polyvalent organic radical, an ionic cluster, or a polyvalent metal ion [2]. The process is usually carried out by heating the rubber, mixed with vulcanizing agents, in a mold under pressure [2]. Unvulcanized rubber is generally not very strong, does not maintain its shape after a large deformation, and can be very sticky. Therefore, most useful rubber articles, such as tires and mechanical goods, cannot be made without vulcanization [2]. In the tire and conveyer belt industry, several layers of elastomers mixed with respective crosslinking agents and other compounding ingredients are assembled one over another and finally co-crosslinked by application of heat and pressure in a molding press. This process is generally called co-crosslinking. The formation of an elastomer-to-elastomer interface by co-crosslinking can be separated into three different steps [3]: wetting of surfaces leading to an intimate contact; diffusion across the interface of free or dangling chains; and covalent bonding of the diffused chains across the interface (co-crosslinking) when the crosslinking agents start to react with elastomer chains. Crosslinking of the joined polymers leads to the formation of a three-dimensional network in the bulk material. When chains have crossed the interface, they can be covalently bonded to the opposite network on both sides. In order to distinguish the chains crosslinked at the interface and in the bulk, the term co-crosslinking is generally used although the same chemical bonds are created at the interface and in the bulk. Schematically outlined in Figure 2 is the mechanism involved in the co-crosslinking process between two unvulcanized elastomers filled with crosslinking agents. At the initial contact time (t 0) of the two elastomer surfaces, wetting of the surfaces takes place which leads to an intimate contact (Figure  2 (a)). At some intermediate contact time (t 0), diffusion of the elastomer chains across the interface takes place (Figure 2 (b)). This is the stage at which the increased penetration depth of the diffusing chains occurs resulting in increased entanglements. Ingredients added in elastomer may also move along with the elastomer chains, depending on the characteristics of the ingredients. At longer contact time (t t ), covalent bonding takes place across the interface (shown by stars in Figure 2 (c)) by the process called co-crosslinking and the covalent bonding / ionic bonding takes place in the bulk (shown by dots in Figure 2 (c)) by the process called crosslinking. It is commonly accepted that the strength of the interface is related to the number of interlinks that are created at the interface. This number generally is often considered as equal to the number of crosslinks in the bulk [3], which may not be the case in many instances, as a result of which the joints fail at the interface. Bhowmick and Gent [4] and Ahagon and Gent [5] have shown that when two elastomer sheets are co-crosslinked in contact, the mechanical strength of the joint under

Adhesion between Compounded Elastomers: A Critical Review 103

(a)

(b)

(c)

Figure 4.2 Formation of the interface between two elastomer sheets. (a) wetting stage (b) interdiffusion stage (c) co-crosslinking stage (dots: crosslinks, stars: co-crosslinks). Figure reproduced with permission from Ref. [3]. Copyright 2000 John Wiley & Sons Inc.

near-equilibrium conditions (or threshold strength) is directly proportional to the amount of chemical interlinking, as evaluated from measurements of the bulk density of crosslinking. They have suggested that the exact contributions of the interdiffusion and interlinking mechanisms are quite difficult to determine because, in general, crosslinking is performed at a high temperature, above 120 °C. At these temperatures, the diffusion of the chains will be very rapid and a complete healing of the interface might occur before the chains are immobilized by the crosslinks [4, 5].

104 Progress in Adhesion and Adhesives, Volume 4

Two classical cases in the co-crosslinking process of elastomers will be discussed first [3, 6–14]. These cases are schematically explained below for better clarity (Figure 3). a. Adhesion between unvulcanized rubber (filled with crosslinking agents) and unvulcanized rubber (filled with crosslinking agents) by co-crosslinking (Figure 3 (a)) b. Adhesion between partially crosslinked rubber (filled with crosslinking agents) and partially crosslinked rubber (filled with crosslinking agents) by co-crosslinking (Figure 3 (b))

4.2.1

Adhesion Between Unvulcanized Rubber (Filled with Crosslinking Agents) and Unvulcanized Rubber (Filled with Crosslinking Agents) by Co-crosslinking

Bhowmick and Chakraborty [6] studied the bond strength between natural rubber (NR) and ethylene propylene diene monomer rubber (EPDM), NR and styrene-butadiene rubber (SBR), acrylonitrile-butadiene rubber (NBR) and SBR, and NR and NBR on cocrosslinking [6]. The measurement of bond strength was carried out with reference to the variation in the vulcanizing system. The test samples for 180 ° peel test were prepared as described below. Compounded rubber sheets were sheeted out with a thickness of approximately 3.5 mm. The surfaces of the synthetic rubbers were made smooth by

Interface

(a)

Co-crosslinking

Unvulcanized rubber (filled with crosslinking agents)

Unvulcanized rubber (filled with crosslinking agents)

Fully crosslinked rubber-rubber joints

Interface

(b) Co-crosslinking

Partially crosslinked rubber

Crosslinking agents

Partially crosslinked rubber

Partial crosslink

Fully crosslinked rubber-rubber joints

Bulk crosslink

Co-crosslink

Figure  4.3 Schematic for co-crosslinking of unvulcanized rubber to unvulcanized rubber (a) and partially crosslinked rubber to partially crosslinked rubber (b).

Adhesion between Compounded Elastomers: A Critical Review 105

pressing them at 100 °C and 3 MPa for 2 min between platens. Fabric backing on one side was also introduced during this operation. Pieces of fabric backed rubber were cut having dimensions of 10x10 cm2. Two layers of rubber each having fabric backing were brought into intimate contact and vulcanized for a specified time. The bond strength of different co-crosslinked NR-EPDM joints, NR-SBR joints, SBR-NBR joints, and NR-NBR joints was measured by 180° peel test and the values are reported in Table 1. Among the

Table 4.1 Bond strengths of different co-crosslinked rubber-to-rubber joints.

#

Compounds#

Bond strength ( 102 J/m2)

Natural rubber (10.5) – Ethylene propylene diene monomer rubber (21)

25.3

Natural rubber (10.5) – Ethylene propylene diene monomer rubber (17)

33.8

Natural rubber (10) – Ethylene propylene diene monomer rubber (17)

28.2

Natural rubber (10) – Ethylene propylene diene monomer rubber (23)

26.1

Natural rubber (9.5) – Ethylene propylene diene monomer rubber (25)

24.5

Natural rubber (9.5) – Ethylene propylene diene monomer rubber (23)

30.4

Natural rubber (9.5) – Ethylene propylene diene monomer rubber (21)

25.3

Natural rubber (13) – Ethylene propylene diene monomer rubber (17)

36.2

Natural rubber (13) – Ethylene propylene diene monomer rubber (23)

35.1

Natural rubber (13) – Ethylene propylene diene monomer rubber (21)

40.6

Natural rubber (32) – Ethylene propylene diene monomer rubber (25)

39.4

Natural rubber (32) – Ethylene propylene diene monomer rubber (21)

33.4

Natural rubber (9.5) – Styrene-butadiene rubber (22)

45.1

Natural rubber (10) – Styrene-butadiene rubber (17)

56.2

Natural rubber (10.5) – Styrene-butadiene rubber (14.5)

66.6

Natural rubber (13) – Styrene-butadiene rubber (13.5)

39 (sample broke near the grip)

Styrene butadiene rubber (22) – Acrylonitrile-butadiene rubber (9.5)

3.7

Styrene butadiene rubber (17) – Acrylonitrile-butadiene rubber (10)

8.3

Styrene butadiene rubber (14.5) – Acrylonitrile-butadiene rubber (11.5)

8.9

Styrene butadiene rubber (13.5) – Acrylonitrile-butadiene rubber (13.5)

9.2

Natural rubber (9.5) – Acrylonitrile-butadiene rubber (24)

2.1

Natural rubber (10) – Acrylonitrile-butadiene rubber (20)

2.2

Natural rubber (10.5) – Acrylonitrile-butadiene rubber (10)

2.6

Natural rubber (13) – Acrylonitrile-butadiene rubber (13.5)

3.4

Values in parentheses are the respective optimum cure times (min)

106 Progress in Adhesion and Adhesives, Volume 4

various pairs studied, NR-SBR joints displayed the highest strength and NR-NBR and SBR-NBR joints displayed the poorest joint strength. Though in NR-EPDM some pairs had close similarity in curing characteristics, development of adhesion was not good as that of NR-SBR (Table 1). The reason for this was that the combination of ingredients chosen for crosslinking a particular rubber had only a little tendency to co-crosslink the other rubber. It was pointed out that diffusion of compounding ingredients across the interface during crosslinking plays an important role in determining the interfacial adhesion [6]. Interestingly, the same authors investigated joint strength between the rubbers having compounding ingredients (before vulcanization). The compounded stocks were less tacky than the uncompounded ones due to migration of ingredients to the surface thus preventing the hundred percent contact between the rubber chains and interdiffusion. NR-SBR joints exhibit higher bond strength than the rest of the joints. The individual rubbers having different optimum cure times were prepared using different vulcanizing systems. The co-crosslinking time and temperature were based on vulcanizing characteristics of the slowest vulcanizing rubber. In order to understand the variation of bond strength with different vulcanizing systems, cure curves were analyzed [6]. It was observed that when the cure time of both substrates was closely matched, the joint was stronger. Hence, a new concept, called the mismatch factor, was introduced for understanding the bond strength of such dissimilar joints. The mismatch factor was defined as [6]:

MF

(cure rate A cure rate B ) (cure time A cure time B )

(1)

where, A and B indicate the two substrates A and B. Equation (1) can be written as

MF

1

1

(t 90 t 2 )A

(t 90 t 2 )B

[t 90A

t 90B ]

(2)

where, t90 indicates time for 90% cure and t2 indicates time for two units rise above the minimum torque. The parameters in the above equation were extracted from the standard cure curves from Monsanto rheometer [6]. The variations of bond strength with mismatch factor for NR / EPDM and NR / SBR joints are shown in Figure 4 (a) and those for NR / NBR and SBR / NBR joints are shown in Figure 4 (b). From Figure 4 (a–b), it is observed that a high mismatch factor gives rise to poor bond strength. The relationship follows a straight line up to a certain mismatch factor. The intercept and the initial slope were different for various rubber pairs studied. However, when reinforcing fillers were added into rubber and similar experiments were carried out, the fracture started at the interface and then veered into one of the substrates. It was difficult to obtain a measure of the adhesion strength. Hence, Loha et al. devised a test method using a separator or a perforated sheet [7].

Adhesion between Compounded Elastomers: A Critical Review 107 11

80

10 9 Bond strength (x102J/m2)

Bond strength (x102J/m2)

70 60 50 40 30 20

7 6 5 4 3 2 1

10 0

(a)

8

5

10

15

20

25

30

35

40

0

(b)

Mismatch factor

5

10

15

20

25

30

35

40

Mismatch factor

Figure  4.4 (a). Plot of bond strength against mismatch factor. , NR-EPDM joints; , NR-SBR joints and (b) Plot of bond strength against mismatch factor. , NR-NBR joints; , SBR-NBR joints.

Loha et al. [7] described a modified 180° peel test for studying the adhesion between co-crosslinked rubber to rubber joints. Filled and unfilled NR to NR, polybutadiene rubber (BR) to BR, and NR to BR joints were tested by conventional 180° peel test and modified 180° peel test. The sample preparation method of modified 180° peel test is explained below. Two layers of uncured rubbers (containing crosslinking agents) each having a fabric backing on one side were brought into intimate contact and finally co-crosslinked for a specified time at the curing temperature. When required, a perforated plastic sheet was inserted at the interface and the joints were finally co-crosslinked for a specified time at the curing temperature. Figure  5 (a–c) shows the components of the laminates and the perforated sheet used in the modified 180° peel test. The formulations of the various mixes are given in Table 2. In the case of conventional 180° peel test, the adhesion strength was calculated in terms of Ga using the equation given below [7].

Ga

2F w

(3)

where Ga is the work of adhesion, F is the applied force and w is the width of the joint. On the other hand, although the perforated plastic sheet was used in the modified 180° peel test, it was mentioned that the energy criterion will not change but the width w will be reduced to w1. Hence

Ga

2F (w w1 )

(4)

108 Progress in Adhesion and Adhesives, Volume 4 Perforated sheet Rubber layer 1

Perforated sheet

Rubber layer 2

Perforated circles

(a)

(b)

(c)

Perforated sheet at the interface

Figure 4.5 (a)–(b) Modified peel test using a perforated sheet inside and (c) the perforated sheet. Figure reproduced with permission from Ref. [7]. Copyright 1987 Elsevier.

Table 4.2 Formulations (in phr) of Mixes [7]. (1) Gum

(2) Filled

a) Natural rubber (i)

(ii)

(iii)

NR

100

100

100

–

BR

–

–

–

100

ZnO

6

5

5

6

5

Stearic acid

(ii)

d) Polybutadiene rubber

(iii)

(i)

(ii)

(i)

(ii)

–

–

100

100

–

–

100

100

–

–

100

100

5

6

6

6

6

0.5

2

2

0.5

2

2

0.5

0.5

0.5

0.5

–

–

–

–

–

–

30

–

30

–

b

N-660

–

–

–

–

–

–

–

30

–

30

Naphthalene oil

–

–

–

–

–

–

3

3

3

3

0.65

0.5

–

0.6

0.55

–

0.65

0.65

0.6

0.6

MBTSd

–

–

0.6

–

–

0.5

–

–

–

–

TMTDe

–

–

0.05

–

0.3

0.3

–

–

–

–

DPGf Sulphur

–

–

–

0.3

–

–

–

–

0.3

0.3

2.5

2.2

2.2

1.8

1.7

1.7

2.5

2.5

1.8

1.8

N-220, ISAF Carbon black N-660, GPF Carbon black c N-cyclohexylbenzothiazyl sulphenamide d Mercaptobenzothiazyl disulphide e Tetramethyl thiuram disulphide f Diphenyl guanidine b

(i)

c) Natural rubber

N-220a

CBSc

a

b) Polybutadiene rubber

Adhesion between Compounded Elastomers: A Critical Review 109

The value of (w – w1) can be calculated from the diameters of the perforation circles (as shown in Figure 5(c)). The tests were made on rubber-rubber joints with and without perforated sheets. The results are shown in Table 3. It was shown that there was no significant difference in the values of bond strength for various pairs of perforated/non-perforated joints with a range of materials. The traces obtained for NR-NR joints with and without perforated sheets are shown in Figure 6. It was shown that the variations in the values of crack arrest and crack initiation were lower in the case of samples prepared using perforated sheets. The authors mentioned that as soon as the new surfaces were generated at the interface, they were deformed and torn. The tear path could not veer, as in the other test without the perforated sheet [7]. In addition, the authors reported that there was a great difficulty in measurement of bond strength for filled rubber-rubber joints or joints where the substrates can diffuse

Table 4.3 Values of bond strength of gum rubber-rubber joints prepared with and without perforated sheet [7].

Nature of compound and formulation* NR-BR 1a(i)-1(b)(i)

Type of joint Perforated

Cure time at contact (min) 20

Bond strength of rubber-rubber (gum-gum) joints (J/m2)

Nature of failure

990

Interfacial

Non-perforated

20

900

Interfacial

Perforated

15

820

Interfacial

Non-perforated

15

810

Interfacial

NR-BR 1a(ii)-1(b)(ii)

Perforated

14

1300

Interfacial

Non-perforated

14

1350

Interfacial

NR-NR 1a(i)-1(a)(i)

Perforated

15

13560

Interfacial

Non-perforated

15

Slipped from the grip

Cohesivea

Perforated

20

15250

Interfacial

Non-perforated

20

Slipped from the grip

Cohesivea

Perforated

15

1350

Interfacial

Non-perforated

15

1250

Interfacial

Perforated

20

1525

Interfacial

Non-perforated

20

3450

Cohesivea

BR –BR 1b(i)-1(b)(i)

*See Table 2 a Cohesive failure in rubber

110 Progress in Adhesion and Adhesives, Volume 4

700

600 NR-NR joints without peforated sheet

Peel force (N)

500

400

300 NR-NR joints with peforated sheet 200

100

0 Distance

Figure 4.6 Force versus distance traces for NR-NR joints with and without perforated sheet at the interface. Figure adapted with permission from Ref. [7]. Copyright 1987 Elsevier.

rapidly at the interface and have very high cohesive strength [7]. Therefore, it was not possible to control the tear path for carbon black filled NR-BR joints without having a perforated sheet at the interface (Figure  7). The failure took place between the rubber and the backing fabric (Figure  7). On the other hand, Figure  7 shows the trace for the same joint using a perforated sheet. It was observed that the propagation of fracture was forced to be interfacial and meaningful results were obtained. The results for several different joints prepared using perforated sheets at the interface are reported in Table 4. The NR-NR joints established highest peel strength amongst all the joints due to the reasons given earlier. Bhowmick et al. [8] then studied the adhesion between co-crosslinked NR and polybutadiene rubber (BR), unfilled and filled with carbon black (CB) and silica. The formulations of the various mixes comprising of NR and BR are given in Table 5. Two layers of unvulcanized rubber having fabric backing on each side were brought into contact with a perforated sheet (a plastic material, Tm 212 °C and Tg 87 °C, diameter 0.28 cm and thickness 0.03 cm) in between and co-crosslinked for a specified time at the crosslinking temperature. The cure time of various rubbers at a particular temperature was found using a rheometer. Peel testing (180°) was carried out in order to determine the

Adhesion between Compounded Elastomers: A Critical Review 111

300

NR-BR joints without peforated sheet

Peel force (N)

250

200 150

NR-BR joints with peforated sheet

100 50 0 Distance

Figure 4.7 Test results for carbon black filled NR-BR joints with and without a perforated sheet at the interface. Figure adapted with permission from Ref. [7]. Copyright 1987 Elsevier. Table 4.4 Values of bond strength of filled rubber-rubber joints prepared with a perforated sheet inside [7] Cure time at contact (min)

Nature of compound

Formulation*

NR-BR

2a(i)-2(b)(i)

15

2a(ii)-2(b)(ii) NR-NR

2a(i)-2(a)(ii)

BR-BR

Bond strength of rubber-rubber (gumgum) joints (J/m2)

Nature of failure

9620

Interfacial

15

4660

Interfacial

6

16720

Interfacial

2a(i)-2(a)(i)

6

51350

Interfacial

2b(i)-2(b)(i)

9

1310

Interfacial

2b(ii)-2(b)(ii)

9

1080

Interfacial

*See Table 2

bond strength of the rubber-rubber joints. There was no problem of deviation of fracture path. It was also established that the presence of perforated sheet did not change the peel strength. Since the curing system used for NR was different from that of BR, the effect of the nature of the rubber and crosslinking system on the bond strength of various rubber

ZnO

(III)

(II)

BR 100, ZnO 5, stearic acid 2, sulphur 1.7, MBTS 0.5, TMTD 0.3

NR 100, ZnO 5, stearic acid 2, sulphur 2.2, MBTS 0.5, TMTD 0.5

(iii)

Tetramethylthiuram disulphide

BR 100, ZnO 5, stearic acid 3, precipitated silica 40, naphthenic oil 5, sulphur 1.0, CBS 1.5, silane coupling agent 3

NR 100, ZnO 5, stearic acid 3, precipitated silica 40, naphthenic oil 5, sulphur 1.3, CBS 1.5, silane coupling agent 3

BR 100, ZnO 6, stearic acid 0.5, N660 30 naphthenic oil 3, sulphur 1.8, CBS 0.6, DPG 0.3

NR 100, ZnO 6, stearic acid 0.5, N660 30 naphthenic oil 3, sulphur 2.5, CBS 0.65

BR 100, ZnO 5, stearic acid 2, sulphur 1.7, CBS 0.55, TMTD 0.3

NR 100, ZnO 5, stearic acid 2, sulphur 2.2, CBS 0.5

(ii)

Mercaptobenzothiazyl disulphide, TMTD

BR 100, ZnO 5, stearic acid 3, precipitated silica 40, naphthenic oil 5, sulphur 1.8, CBS 0.8, TMTD 0.3, triethanol amine 2

NR 100, ZnO 5, stearic acid 3, precipitated silica 40, naphthenic oil 5, sulphur 2.5, CBS 0.8, TMTD 0.2, triethanol amine 2

BR 100, ZnO 6, stearic acid 0.5, N220 30 naphthenic oil 3, sulphur 1.8, CBS 0.6, DPG 0.3

NR 100, ZnO 6, stearic acid 0.5, N220 30 naphthenic oil 3, sulphur 2.5, CBS 0.65

BR 100, ZnO 6, stearic acid 0.5, sulphur 1.8, CBS 0.6, DPG 0.3

cyclohexylbenzothiazyl sulphenamide, MBTS

Polybutadiene rubber (silica filled)

(b)

Zinc oxide, CBS

Natural rubber (silica filled)

Polybutadiene rubber (carbon black filled)

(b)

(a)

Natural rubber (carbon black filled)

(a)

Polybutadiene rubber (gum)

(b)

NR 100, ZnO 6, stearic acid 0.5, sulphur 2.5, CBS 0.65

Natural rubber (gum)

(I)

(a)

(i)

Compound

Mix

Table 4.5 Formulations (in phr) of mixes [8].

112 Progress in Adhesion and Adhesives, Volume 4

Adhesion between Compounded Elastomers: A Critical Review 113

joints was studied. Three systems were studied and the cure times of the joints are given in Table 6. It was found that NR-NR joints gave maximum adhesion in each case, as expected, due to the mutual solubility and strain induced crystallization of NR compared to that of BR (Table 6). The bond strength of BR-BR joints was greater than that of NR-BR joints due to the mutual solubility of BR in BR (Table 6). Compounds la (iii) gave the highest bond strength of the three systems (Table 6). The crosslinking systems using S, MBTS and TMTD in NR and BR were found to be more effective among the three systems studied, which produced higher interlinking densities. The influence of various fillers on the strength of NR-BR interface is shown in Table 7. For all joints, NR-NR, NR-BR and BR-BR, the joint strength decreases with increased cure time. It has been shown that the influence of adsorption, diffusion and cross-linking affects the strength of the interface. The strength of black filled NR-BR joints was considerably improved compared to the gum compound, because of the reinforcing effect of black in BR (Tables 6 and 7). NR-BR joints were intermediate between NR-NR and BR-BR joints in adhesion strength. It was mentioned that as soon as carbon black was added to the weaker phase, namely BR, the strength of the interface at NR-BR joints was enhanced due to increase of cohesive strength of BR with filler reinforcement at probably similar levels of diffusion of NR molecules. However, the filler restricts the chain mobility, reducing interfacial interactions. As a result, the filled BR-BR joints showed weaker bond strength than NR-BR joints (Table  7). Unlike gum compounds which showed higher bond strength of the BR-BR joints than the NR-BR joints, it was found that the bond strength of NR-BR joint was greater than that of the BR-BR after addition of fillers (Table 6 and 7). Between the two carbon black systems studied, N220 systems with higher reinforcing capacity showed higher bond strength mainly because of improved strength of the substrates (Table 7). On the other hand, silica showed different behaviour. Table 4.6 Effect of crosslinking systems and rubbers on different rubber joints of gum natural and polybutadiene rubber compounds [8].

Compound*

Cure time and contact time (min)

Ia(i)-Ia(i) Ia(i)-Ib(i)

Bond strength of rubber-rubber joints (Jm 2) 15250

20

990

Ib(i)-Ib(i)

1525

Ia(ii)-Ia(ii)

41240

Ia(ii)-Ib(ii)

14

1100

Ib(ii)-Ib(ii)

2430

Ia(iii)-Ia(iii)

57640

Ia(iii)-Ib(iii) Ib(iii)-Ib(iii) *See Table 5

15

1155 3940

114 Progress in Adhesion and Adhesives, Volume 4

Table 4.7 Effect of fillers on different joints of natural and polybutadiene rubber compounds [8]. Adhesion strength of rubber-rubber joints (Jm 2)

Fillers N-220

Cure time at contact (min)

IIa(i)-IIa(i)+ NR-NR

IIa(i)-IIb(i)+ NR-BR

IIb(i)-IIb(i)+ BR-BR

6

51350

23060

1520

9

16700

9910

1310

15

–

9620 +

N-660

IIa(ii)-IIa(ii) NR-NR

IIa(ii)-IIb(ii) NR-BR

IIb(ii)-IIb(ii)+ BR-BR

6

33280

5645

2745

9

32270

4950

1080

15

18450

4660 +

Silica with TEA#

IIIb(i)-IIIb(i)+ BR-BR

IIIa(i)-IIIa(i) NR-NR

IIIa(i)-IIIb(i) NR-BR

4

––-

18940

–-

7

12103

17650

–-

10

–

17295

–-

IIIa(ii)-IIIa(ii) NR-NR

IIIa(ii)-IIIb(ii) NR-BR

IIIb(ii)-IIIb(ii)+ BR-BR

15

27670

16425

–

20

–

18160

–

25

–

+

16140 +

IIIa(ii)-IIIa(ii) NR-NR Silica with Si-69*

785 +

+

Silica with Si-69#

1120 +

– +

IIIa(ii)-IIIb(ii) NR-BR

IIIb(ii)-IIIb(ii)+ BR-BR

12

–

96040

24010

16

277555

34745

–

20

–

31070

–

#indicates that the mixing was done in open mill *indicates that the mixing was done in internal mixer + See Table 5

The strength of the interface of most silica filled compounds was very high, 4–5 times higher than the carbon black filled NR-BR joints and about 15 times greater than the corresponding unfilled vulcanizates (Table 7) (The adhesive properties, however, were

Adhesion between Compounded Elastomers: A Critical Review 115

dependent on adhesion between silica and rubber). It was also shown that the bond strength between two substrates depends to a large extent on the individual properties of the substrates (tensile strength, modulus at 100% elongation and elongation at break, and hardness). In the case of peel joints with substrates having fabric backing, the energy is concentrated mostly at the interface. What happens if the joints are made differently and the failure does not take place under peeling mode? This becomes more complicated when there is dynamic loading. Sarkar and coworkers [9–10] studied the failure induced by stress concentration at co-crosslinked rubber-rubber interfaces and also studied the stress distribution using photoelastic method in co-crosslinked NR to NR joint subjected to uniaxial tension. The formulations of the mixes are given in Table  8. The test specimens were prepared in two stages. In the first stage, a uniform unfilled uncured sheet (filled with crosslinking agents) was prepared between two thin aluminium sheets under a light pressure (2 MPa) in a mold for a very short period of time at 100 °C. In the second stage, a uniform CB filled uncured sheet (filled with crosslinking agents) was prepared between two thin aluminium sheets under a light pressure (2 MPa) in the mold for a very short period of time at 100 °C. The aluminium sheets were then removed from both sides of the filled and the unfilled sheets. Finally, two halves of dissimilar materials (one half from the unfilled sheet and one half from the filled sheet) of uniform thickness were joined together under controlled pressure as shown in Figure 8. The optimum cure time (12.5 min), temperature (150 °C) and pressure (5 MPa) of the uncured rubber compound were employed for co-crosslinking. Table 4.8 Formulation (in phr) of the birefringent and the opaque materials [10]. I

II

NR

100

100

ZnO

5

–

Stearic acid

2.5

–

HAFa

50

–

CBS

0.6

–

S

2.5

–

DCPb

–

1.0

1.53

0.50

–

1.92

Small strain modulus, MPa (100% elongation) Stress optical coefficient, Pa a

1

HAF – High abrasion furnace black b DCP – Dicumyl peroxide

10

9

116 Progress in Adhesion and Adhesives, Volume 4 Softer gum NR

(a)

Opaque and stiffer CB filled NR

(b)

Figure  4.8 A model rubber-rubber two-component specimen. A. Sarkar, A. K. Bhowmick and S. Majumdar, 1991, Photoelastic studies on rubber-to-rubber joints. Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [10].

Photoelastic studies were made to evaluate whole-field information about the stress distributions in the rubber-rubber joints [10]. The birefringence was produced when the rubber-rubber two-component specimen (as shown in Figure 8) was stretched (in a specially prepared grip) in the field of a plane polariscope in between the polariser and the analyser [10]. To understand the stress distribution and the mechanism of fracture, an investigation was conducted by preparing simple rubber-rubber joints [10]. Two types of two-component specimens were prepared and studied as shown in Figure 9 [10]. The only difference between the two-component type I and type II specimens was that the materials were interchanged. The isochromatic fringe patterns, as obtained at different load levels in a monochromatic light field, are shown in Figures 10 (a) and 10 (b) for two-component specimen type I, while Figures 11(a) and 11(b) show the same for specimen type II. Symmetric fringe patterns were observed in both types of model composite specimens which demonstrate the actual stress gradient at different levels of loading. From Figures 10 and 11, it was noted that the fringe patterns were generated in different zones depending on the geometry of the two-component joint and the stress. For the type I specimen, the fringes were generated at the angle tip and travelled towards the edges of the joint (Figures  10 (a–b)). This indicated the maximum stress concentration site at the angle tip. For the type II two-component specimen, however, the fringes appeared near the edges of the joint and moved towards the angle tip of the joint (Figures 11 (a–b)). The fringes were also numbered in order of their generation for type I and type II specimens as shown in the individual photographs (Figures 10 and 11). Moreover, it was mentioned that there was a stress gradient at the angle tip for the

Adhesion between Compounded Elastomers: A Critical Review 117

Softer material

Stiffer material Type I specimen

Type II specimen

(a)

(b)

Figure  4.9 Two types of two-component specimens, (a) Specimen type I ; (b) Specimen type II. A. Sarkar, A. K. Bhowmick and S. Majumdar, 1991, Photoelastic studies on rubber-to-rubber joints. Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [10].

(a)

(b)

Figure  4.10 Isochromatic fringe patterns of two-component type I specimen at different loading conditions: (a) 6.3N; (b) 18.25N. A. Sarkar, A. K. Bhowmick and S. Majumdar, 1991, Photoelastic studies on rubber-to-rubber joints. Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [10].

type I specimen at all load levels and it was found to be more severe at the higher load levels (~18N) [Figure 10 (b)]. These studies were also theoretically confirmed from the Finite Element Analysis by the same authors [10].

118 Progress in Adhesion and Adhesives, Volume 4

(a)

(b)

Figure 4.11 Isochromatic fringe patterns of two-component type II specimen at different loading conditions: (a) 6.5N; (b) 15.6N. A. Sarkar, A. K. Bhowmick and S. Majumdar, 1991, Photoelastic studies on rubber-to-rubber joints. Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [10].

4.2.2

Adhesion Between Partially Vulcanized Rubber (Filled with Crosslinking Agents) and Partially Vulcanized Rubber (Filled with Crosslinking Agents) by Co-crosslinking

Chang and Gent [11] studied the effect of interfacial bonding (degree of chemical crosslinking) on the adhesion strength of various co-crosslinked elastomer joints. Two identical layers of an elastomer were prepared, partially crosslinked to the same extent. These layers were then pressed into intimate contact and the co-crosslinking reaction was carried out to completion (Figure 12). By varying the extent of crosslinking before the layers were brought into contact, the degree of chemical interlinking between the layers was varied over the entire range. Flat sheets about 0.5 mm thick were prepared from each mix by pre-molding them for a few minutes at 80 °C between Mylar films. After this pre-molding step, the elastomer sheets were crosslinked to a certain degree by placing them for a time t1 in a heated press at 150 °C (as shown in Figure 12 (a)). After 24 h at room temperature, the Mylar films were removed and the two elastomer sheets, crosslinked to the same degree, were immediately pressed into intimate contact and subjected to a further crosslinking for an additional period t2 by heating at 150 °C (as shown in Figure 12 (b)). The total time of heating tl t2 was chosen so that the crosslinking process was substantially complete after this time (as shown in Figure 12 (c)). The degree of crosslinking at various reaction times was estimated by studying the equilibrium degree of swelling of the samples in a suitable solvent. Values of the volume

Adhesion between Compounded Elastomers: A Critical Review 119 Step 1 Partially crosslinked for time t1

Step 2 Crosslinked in contact for a further time t2

V1

Vf ΔV

Vf

V1

(a)

Vf Crosslink density

(b)

L V1

t1

(c)

Time

t1+t2

Figure  4.12 Sketch of co-crosslinking reaction. Figure adapted with permission from Ref. [11]. Copyright 1981 John Wiley & Sons Inc.

fraction r of the rubber in the swollen gel were calculated. In addition, the number of molecular network strands per unit volume was calculated from the Flory-Huggins relationship:

A[In(1

r)

c

2 1/3 1 r ]/ V1 ( r

1

2 r)

(5)

where A is Avogadro’s number, V1 is the molar volume of the swelling liquid, 1 is the rubber-liquid interaction parameter and r is the volume fraction of rubber in the swollen gel. A series of samples (Recipe I, Recipe II, Recipe III and Recipe IV as shown in Table 9) were made with varying amounts of interfacial interlinking by varying the time t1 and hence the amount of crosslinking of the elastomer sheets before they were brought into contact and the crosslinking process taken to completion. The final degree of crosslinking of all the samples in the series was identical, because the total cure time tl t2 was held constant. It corresponded to the fully cured state (as shown in Figure 12 (c)). The r and values for a series of samples crosslinked by varying the time t1 are reported in Table 9. 180 ° peel test was performed at threshold condition (at higher temperatures) to determine the work Ga of detachment per unit area of interface. Figures  13 (a–c) are plots where the threshold energy of detachment G0 for the two interlinked sheets (corresponding to Recipes I, II, III and IV) is plotted against the increase in density of network, inferred from swelling measurements on the sheets before and after they were joined together. From Figure 13 (a) it was shown that there exists a direct proportionality between the adhesion strength obtained under threshold conditions and the density of chemical interlinking between the two BR elastomer layers. This proportionality extended over the entire range of interlinking from zero, when only dispersion forces were assumed to be operative at the interface, up to the fully crosslinked state, when cohesive rupture took place by tearing. From Figure 13 (b) it was shown that there exists a direct proportionality between the threshold adhesion strength and the inferred density of chemical interlinking between the two EPR elastomer layers. On the other hand, from Figure 13 (c) it was observed that there was a markedly nonlinear dependence of adhesion strength obtained under threshold conditions on the density of chemical interlinking. The reason for this was attributed to the feature of the particular crosslinking system employed and rough failure surface.

120 Progress in Adhesion and Adhesives, Volume 4

Table 4.9 Kinetics of crosslinking for Recipes I, II, III and IV [11]. Cure time (Min at 150 °C) BR, 0.08% DCP (Recipe I) 15 20 25 30 35 50 90 140 BR, 0.2% DCP (Recipe II) 11 13 20 25 40 60 80 EPR, 2.7% DCP 0.32% S (Recipe III) 10 20 30 40 60 90 EPDM, sulphur cured (Recipe IV) 15 17.5 20 25 30 35 70 100 a

a r

b

( 10–26 m 3)

0.090 – 0.201 0.220 0.289 0.259 0.308 0.335

0.10 – 0.36 0.42 0.49 0.58 0.81 0.92

– 0.262 0.316 0.336 0.392 0.427 0.438

– 0.59 0.82 0.03 1.34 1.67 1.79

0.193 0.271 0.324 0.338 0.362 0.360

0.13 0.59 1.04 1.20 1.47 1.48

0.081 0.094 0.166 0.246 0.280 0.292 0.312 0.324

0.06 0.08 0.30 0.80 1.12 1.25 1.51 1.67

BR and EPDM swollen in n-heptane, ethylene propylene rubber (EPR) swollen in benzene Calculated from r by means of Equation (5) using the following values for x1: x1 0.37 0.32 r for BR, n-heptane system : x1 0.495 0.256 r for EPR, benzene system: x1 0.44 for EPDM-n-heptane system

b

Adhesion between Compounded Elastomers: A Critical Review 121 80 70

60 50 50

50 40 30

0.2% DCP

20 10 0 0.0

(a)

G0 (J/m2)

40

0.08% DCP

G0 (J/m2)

G0 (J/m2)

60

30 20 10

1.0

1.5

2.0

–26 –3 Δ x 10 (m )

2.5

0.0

(b)

30 20 10 0 0.0

0 0.5

40

0.2

0.4

0.6

0.8

(c)

Δ x 10–26 (m–3)

0.5

1.0

1.5

2.0

Δ x 10–26 (m–3)

Figure 4.13 Threshold detachment energy (Go) versus increase in density of network chains ( ) while two (a) BR sheets were crosslinked in contact, Recipes I and II; (b) EPR sheets were crosslinked in contact, Recipe III; and (c) EPDM sheets were crosslinked in contact using sulfur, Recipe IV. Figure adapted with permission from Ref. [11]. Copyright 1981 John Wiley & Sons Inc.

Bhowmick and Gent [4] studied the effect of interfacial bonding (degree of chemical crosslinking) on the adhesion strength between various co-crosslinked SBR-SBR joints and chloroprene rubber (CR)-CR joints. Two identical layers of an elastomer were prepared, partially crosslinked to the same extent. These layers were then pressed into intimate contact and the co-crosslinking reaction was taken to completion (similarly to Figure 12). By varying the extent of crosslinking before the layers were brought into contact, the degree of chemical interlinking between the layers was varied over the entire range. Flat sheets, about 2 mm thick, were prepared in all cases by a molding process, between films of Mylar for SBR and Teflon for CR. The elastomer sheets were partially crosslinked by heating for a time t1 at the vulcanization temperature. The protective films were then removed and the two sheets, crosslinked to the same degree, were pressed into intimate contact and subjected to further crosslinking for a time t2 at the vulcanization temperature. The total time (t1 t2) was chosen from rheometer curves to give essentially complete crosslinking. The times t1 and t2 were chosen to give a wide range of additional crosslinking while the sheets were in contact. It is assumed that the degree of interlinking of the sheets is represented by the additional crosslinking while in contact. The mix formulations and the molding conditions are given in Table 10. Typical diagrams of the peel test and tear test are shown in Figure 14. The peel experiments were employed to measure the work of detachment over a range of rates and temperatures and also for swollen samples. The work of detachment for the swollen samples was calculated using the relationship.

Ga

2 s2 F w

where s linear swelling ratio for (unswollen samples the width of the bonded interface (see Figure 14).

(6) s

1), F is the peel force and

is

122 Progress in Adhesion and Adhesives, Volume 4

Table 4.10 Mix formulations, in parts by weight, and vulcanization conditions employed for preparing test specimens [4] Chloroprene rubber (CR) (A)

Oxide crosslinks : CR, 100; magnesium oxide, 2; zinc oxide, 5 cured at 150 °C for a total time (t1 90 min

t2) of

(B)

Sulfur crosslinks : CR, 100; sulphur, 2; diorthotolyl guanidine (DOTG), 0.5; tetramethylthiuram disulfide, 1.5, vulcanized at 140 °C for a total time (t1 t2) of 80 min

(C)

Oxide and sulfur crosslinks : CR, 100; sulphur,1; tetramethylthiuram monosulfide (TMTM), 1; DOTG, 0.5; magnesium oxide; 4; zinc oxide, 5; stearic acid 1, vulcanized at 150 °C for a total time (t1 t2) of 120 min

Styrene-butadiene rubber (SBR) SBR-1502, 100; dicumyl peroxide, 0.5, vulcanized at 150 °C for a total time (t1 100 min

t2) of

F F

w

F

F

(a)

(b)

F

F

Figure  4.14 (a) Peel test and (b) tear test. Figure reproduced from Ref. [4], with permission from Rubber Chemistry and Technology. Copyright (1984), Rubber Division, American Chemical Society Inc.

Estimates of the degree of crosslinking at various stages of vulcanization were obtained from stress-strain relations in tension. The tensile measurements were plotted in the Mooney-Rivlin form:

f / 2 A0 (

2

) C1 C2

1

(7)

Adhesion between Compounded Elastomers: A Critical Review 123

where f is the tensile force, A0 is the cross-sectional area of the specimen in the unstrained state, is the extension ratio and C1 and C2 are elastic constants, obtained from the intercept and slope of a plot of the quantity on the left-hand side of Equation (7) against 1. The values of C1 and C2 as vulcanization proceeds are given in Table 11. The constant C1 was taken as directly proportional to the degree of crosslinking as

C1

RT 2

(8)

Increase in C1, denoted C1, as crosslinking takes place in two contacting sheets was taken as a measure of the corresponding degree of interlinking. 180 ° peel test was performed at threshold condition (at higher temperatures) to determine the threshold energy of detachment Ga per unit area of interface. Figures  15 (a-c) are plots where the threshold energy Ga,0 of detachment for the two interlinked sheets is plotted against the increase in the elastic constant C1. Again, there exists a direct proportionality between the threshold work of detachment, Ga,0 and increase C1 in the elastic constant of SBR layers over the entire range. From Figure  15 (b-c), it is also shown that there exists a direct proportionality between the threshold energy of detachment, Ga,0 and increase C1 in the elastic constant of CR compounds (CR compound A, CR compound B and CR compound C). There was no apparent difference observed between -Sx- and -O- crosslinking. At a similar degree of crosslinking and of interlinking, both -Sx- and -O- crosslinking systems showed the same value for the work of detachment (between CR compound A and CR compound B). Interestingly, it was noted that the slope of each line in Figure 15 is different. At a given degree of interlinking, the strength of adhesion is lower when the final degree of crosslinking is higher and the molecular strands are shorter. It was earlier proposed that the work of fracture under threshold conditions Gc,0, increases with molecular weight, Mc of the network strands as follows.

Gc ,0

kM 1c /2

(9)

If the same relationship is assumed to hold at the interface, the work of detachment will be given by

Ga,0

k( C1 / C1max )M 1c /2

(10)

where, C1max denotes the maximum value of the elastic constant C1 after complete vulcanization. 1 Since C1max Mc , it can be easily written

Ga,0

/2 k C1 / C13max

(11)

124 Progress in Adhesion and Adhesives, Volume 4

Table 4.11 Progress of crosslinking for CR and SBR compounds [4]. Vulcanization time, min

C1, kPa

C2, kPa

CR Compounds A

B

C

SBR Compounds

10

2

20

17

28

24

40

80

20

60

105

20

80

110

22

90

120

24

20

10

20

25

34

24

35

66

20

40

80

20

50

100

20

60

110

20

80

120

20

15

8

60

19

44

90

24

100

95

30

160

110

50

210

110

120

230

110

5

10

20

10

20

20

20

35

80

40

80

120

70

150

120

90

235

100

100

275

100

Thus, the slope Ga,0/ C1 of the experimental relationship between Ga,0 and C1 as shown above should be proportional to C1max. As shown in Figure 16, the data points of Figure 15 conform reasonably well to a linear relationship, as predicted. The principal structural factors governing both cohesive and adhesion strengths under threshold

Adhesion between Compounded Elastomers: A Critical Review 125

Ga,0 (J/m2)

Ga,0 (J/m2)

60 40

60

Ga,0 (J/m2)

100

80

50

40

20

20 0

0

(a)

200 ΔC1 (kPa)

50

400

100

0

150

ΔC1 (kPa)

(b)

100

200

ΔC1 (kPa)

(c)

Figure 4.15 Threshold energy of detachment, Ga,0 versus C1 the elastic constant of network chains for two (a) SBR layers crosslinked in contact, (b) CR compound A layers crosslinked in contact, o; CR compound B layers crosslinked in contact, ; and (c) CR compound C layers crosslinked in contact. Figure adapted from Ref. [4], with permission from Rubber Chemistry and Technology. Copyright (1984), Rubber Division, American Chemical Society Inc.

0.7 0.6 0.5 ∂Ga,0 ∂(ΔC1)

mm

0.4 0.3 0.2 0.1 0.0 0.0

0.5

1.0 C

1.5 3/2

1max

2.5

2.0 8

3.0

3.5

(× 10 Pa

–3/2

)

/2 Figure 4.16 Ga,0/ C1 vs. C13max values of CR, ; BR, o; and SBR, . Figure reproduced from Ref. [4], with permission from Rubber Chemistry and Technology. Copyright (1984), Rubber Division, American Chemical Society Inc.

conditions appear to be the number and length of molecular strands comprising the interface. Using the above swelling method, Chun and Gent [12] carried out detailed experiments to understand the contribution of length and number of interlinking molecules to the strength of adhesion. They used standard SBR and dicumyl peroxide as the crosslinking agent. The joints were prepared exactly the same way as done by others in the Gent’s laboratory [4,11].

126 Progress in Adhesion and Adhesives, Volume 4

The degree of crosslinking after a time t of reaction was characterized by the corresponding number N of network molecular strands per unit volume. Values of N were determined by three measurements: Young’s modulus E, from equilibrium stress-strain relations at small tensile strains up to about 10%, Mooney-Rivlin coefficient C1 from tensile stress-strain relations up to about 200% elongation; and volume swelling ratio Q in toluene. These values at different cure times are given in Table 12. As usual, the values of E and C1 increased, whereas Q decreased with increasing cure time. Average values of N from the three methods were employed to characterize the degree of crosslinking. The relation between the average value of N and cure time is given in Figure 17. With increasing cure time, the number of network molecular strands increased and then reached a plateau. The degree of interlinking for the two sheets crosslinked for a further time, t while in contact was represented by the corresponding increase, N in density of network strands while in contact. The length L of interlinking strands was assumed to be the same as that of strands in the two interlinked sheets, crosslinked to a final degree: N2 N1 N, where, N1 initial crosslink density and N2 final crosslink density. By a judicious choice of times tl and t2 (= tl t), (see Figure 12) interlinked sheets were prepared with the same density N of interlinks but with different strand lengths L. Alternatively, by bringing the sheets together at different times tl and then interlinking them by carrying out the crosslinking reaction to completion, the density N of interlinks was varied but the strand length L was held constant. The length L of the interlinking molecules was represented by their molecular weight between the crosslinks Mc, which is calculated using the equation shown below:

Mc

A/N

(12)

Molecular characteristics of the two systems are listed in Tables 13 and 14, respectively.

Table 4.12 Elastic properties E and C1, and swelling ratio Q, for various cure times [12]. Cure Time (min)

E (MPa)

C1 (MPa)

Q

5

0.468

0.017

11.61

10

0.754

0.076

8.66

15

0.899

0.132

6.59

20

1.279

0.162

5.94

40

1.486

0.182

5.42

60

1.533

–

4.99

80

1.503

0.204

4.94

105

1.692

0.206

4.90

Adhesion between Compounded Elastomers: A Critical Review 127

N (10–25 /m3)

15

10

5

0

50 100 t (min)

Figure 4.17 Density N of network strands vs. cure time t. Average values of N from the three relations. Figure reproduced with permission from Ref. [12]. Copyright 1996 John Wiley & Sons Inc.

Table 4.13 Density N of network strands, density N of interlinks and molecular weight Mc of interlinks* [12]. t1/t2 (min at 150 °C) 10/5

N ( 10–25/m3)

N ( 10–25/m3)

Mc (g/mol)

6.44

2.03

8980

10/10

7.81

3.40

7300

10/15

8.66

4.24

6590

10/20

9.63

4.85

6150

10/40

10.29

5.88

5535

10/90

11.45

7.04

4960

*Strand and interlink lengths are represented by molecular weight Mc

The near-equilibrium strength of adhesion was determined at 80 °C at peel rates of 10 and 0.5 mm/min. In the first case (samples shown in Table 13), the strength of adhesion was found to be more or less independent of the extent of additional curing, i.e., of the density of interlinking (Figure 18). This surprising observation was contrary to previous studies of the effect of density of interlinking bonds, where a direct proportionality was noted [4,11]. The authors justified this behavior by the change in the length of interlinking molecules as cure proceeds, and the anomalous result was attributed to combined effects of increasing density and decreasing length of interfacial bonds. In order to study the relationship between strength of adhesion and length L of interlinking molecules, represented by their molecular weight Mc, a second set of specimens

128 Progress in Adhesion and Adhesives, Volume 4 80 0.5 mm/min 10 mm/min

70 60

G0 (J/m2)

50 40 30 20 10 0 0

20

40

60

80

100

Δt (min)

Figure  4.18 Effect of degree of interlinking N on peel strength (for a further time t while in contact). Strength of adhesion determined at 0.5 mm/min, ; and strength of adhesion determined at 10 mm/min, . Figure adapted with permission from Ref. [12]. Copyright 1996 John Wiley & Sons Inc.

Table 4.14 Samples prepared with similar densities N of interlinks but different interlink lengths L, represented by standard molecular weight Mc [12]. t1/t2 (min at 150 °C) 10/10

N ( 10–25/m3)

N ( 10–25/m3)

Mc (g/mol)

7.81

3.40

7300

11.5/12

8.45

3.33

6760

12/15

8.94

3.57

6390

15/21

9.68

3.24

5890

19/37

10.50

2.92

5420

24/76

11.45

2.89

4960

(samples shown in Table  14) was prepared with approximately constant density N of interlinking bonds, about 3.2 0.3 1025/m3, and with different values of N, and hence Mc (Table 14). Strand densities N ranged from 7.80 1025/m3 to 11.45 1025/m3 (Table 14). The quasi-equilibrium strengths of adhesion are given in Figure 19. They are seen to increase strongly with Mc. Using logarithmic scales for both axes, a linear relation was obtained:

G0

M 1c .62

(13)

Adhesion between Compounded Elastomers: A Critical Review 129

G0 (J/m2)

1.6

1.2

0.8 3.7

3.8

3.9

log Mc (g/mol)

Figure  4.19 Variation of threshold peel strength G0 with molecular weight Mc of interlinking molecules. N is held constant. Figure adapted with permission from Ref. [12]. Copyright 1996 John Wiley & Sons Inc.

The relation between the length of interlinking molecular strands and strength of adhesion was also investigated using the first set of samples (samples shown in Table 13) in which both N and Mc were varied. Since G0 is generally thought to increase in direct proportion to N, the quantity (G0/ N) was plotted against Mc in Figure 20 in order to correct for variations in density of interlinks. Using logarithmic scales for both axes, a linear relation is again obtained:

G0 / N

M 1c .51

(14)

Finally, the authors showed that the variation of threshold strength of adhesion with length of interlinking molecules is roughly in accord with the predictions of the LakeThomas theory [12].

G0

1.0 C (U / a1/2 )L3/2 N

(15)

where G0 is the strength of adhesion, C is a numerical factor (generally between 2 and 10), that characterizes the intrinsic flexibility of the molecule), U is the dissociation energy for each C-C bond, a is the length of a C-C bond, L is the length of a fully stretched network strand and N is the degree of interlinking. All the above measurements were done under near-equilibrium conditions. However, Sarkar and Bhowmick [13] studied the adhesion between co-crosslinked filled and unfilled rubber-to-rubber similar and dissimilar joints made from NR and EPDM by 180° peel test under non-equilibrium conditions where the dissipation

130 Progress in Adhesion and Adhesives, Volume 4

log (G0/ΔN)

1.5

1.0

0.5

3.6

3.8 log Mc (g/mol)

4.0

Figure  4.20 Variation of reduced peel strength G0/ N ( 1025 J.m 2) with molecular weight Mc of interlinking molecules. Note that both N and Mc were varied. Figure reproduced with permission from Ref. [12]. Copyright 1996 John Wiley & Sons Inc.

of energy was not minimized. The compositions of various mixes based on NR and EPDM rubbers are given in Table 15. The authors investigated the effect of degree of interfacial linking on the peel strength among the various elastomer joints using these compositions. Filled and unfilled rubber to rubber joints (similar and dissimilar joints) from NR and EPDM were prepared, partially crosslinked to the same extent. These layers were then pressed into intimate contact (by placing a perforated sheet at the interface) and co-crosslinking reaction was taken to completion. By varying the extent of crosslinking before the layers were brought into contact, the degree of chemical interlinking between the layers was varied over the entire range. The crosslinking density, in terms of volume fraction of rubber for each sheet, was calculated by measuring the equilibrium degree of swelling in benzene [11]. The interlinking density ( Vr) was calculated from the difference in the volume fraction of rubber [11]. The interfacial linking density between the rubber phases at various cure times is given in Table 16. From the 180° peel test, it is seen that the peel strength of the gum samples increases as the interlinking density increases (Figure  21). The gum EPDM/EPDM joints showed higher peel strength than gum NR/EPDM joints because of the difference in cure rates of the latter two rubbers and also because of the difference in interdiffusion in these dissimilar rubbers as described earlier (Figure 21). Interestingly, there was many-fold increase in peel strength in the carbon black loaded joints as compared with the gum rubber-to-gum rubber joint. The results are

Adhesion between Compounded Elastomers: A Critical Review 131

Table 4.15 Formulations and characterization of the mixes [13]. Ingredients (in phr) NR(RMA-IX) Mitsui EPT3045 ZnO Stearic Acid N-330(HAF) CBSa TMTDb MBTc ZDECd Sulphur RESULTS Scorch Time, min Optimum cure time, min Modulus at 300% elongation, MPa Elongation at break, % Tensile strength, MPa Hardness, Shore A

I

II

III

IV

100 5 2.5 0.6 2.5

100 5 2.5 40 0.6 2.5

100 5 1 1.5 1.0 0.6 1.5

100 5 1 40 1.5 1.0 0.6 1.5

4.5 14 2.4 1050 21.3 33

3.0 12.7 9.6 660 26.7 58

3.7 14.7 1.01 320 1.4 47

3.0 12.5 4.6 670 18.4 64

a

CBS : Cyclohexyl Benzothiazyl Sulphenamide TMTD : Tetramethyl Thiauram Disulfide c MBT : Mercaptobenzothiazole d ZDEC : Zinc Diethyldithiocarbamate b

Table 4.16 Interlinking density, Vr between the rubber phases at various partial cure times [13]. NRg/NRg

NRf/NRf

NRg/EPDMg

EPDMg / EPDMg

Partial cure time (min)

Partial cure time (min)

Partial cure time (min)

Partial cure time (min)

Vr

Vr

Vr

Vr

2

0.209

2

0.219

2

0.175

2

0.206

4

0.141

4

0.149

4

0.107

4

0.115

6

0.066

6

0.059

6

0.032

6

0.056

8

0.021

8

0.027

8

-

8

0.029

NRg

NR gum, NRf

NR filled, and EPDMg

EPDM gum

shown in Figure 22. The higher peel strength of the filled rubber matrix was achieved due to higher rubber-filler interaction, reinforcement and adequate diffusion of rubber chains [13].

132 Progress in Adhesion and Adhesives, Volume 4 140 NRg/NRg EPDMg/EPDMg NRg/EPDMg

120

Peel strength, Ga (J/m2)

100

80

60

40

20

0 0

0.05

0.10

0.15

0.20

0.25

Interlinking density, ΔVr

Figure 4.21 Plot of peel strength against the interlinking density ( Vr). A. Sarkar and A. K. Bhowmick, 1992, Fatigue failure of rubber-to-rubber joints, Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [13].

Peel strength (Ga) kJ/m2

15

10

5

0 Gum

HAF (40 phr)

Cured adhesion

Figure 4.22 Plot of peel strength against the cured adhesion (two rubber substrates strongly adhering to each other during curing process by application of heat and pressure in a molding press) between gum NR to gum NR joints and filled NR to filled NR joints. A. Sarkar and A. K. Bhowmick, 1992, Fatigue failure of rubber-to-rubber joints, Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [13].

Adhesion between Compounded Elastomers: A Critical Review 133

Gent and Lai [14] have studied the effect of surface roughness on the adhesion between filled and unfilled co-crosslinked rubber-rubber joints. Three elastomers were used: SBR; 50/50 blend of BR/SBR filled with CB; and NR filled with CB. Crosslinking of SBR was carried out using dicumyl peroxide, crosslinking of BR/SBR blend filled with CB was carried out using sulfur, and crosslinking of NR filled with CB was carried out using either peroxide or sulfur. A layer of rubber compound was sandwiched between two Mylar films and pressed at 45 °C for 30 min to form a sheet about 1 mm thick. The sheet was then crosslinked to a given degree by heating at a suitable temperature, say 150 °C, for a time t1, in a heated press in between various mold surfaces (Mylar film, smooth mild steel, and roughened mild steel) to impart different surface roughnesses to the partially cured samples. The peel strength Ga required to separate a unit area of bonded interface was determined by peeling apart long strips cut from the bonded sheets at test temperatures ranging from 40 °C to +130 °C and at a constant peel rate of 40 μm/s. Values of peel strength Ga are plotted in Figure  23 against test temperature for SBR samples (crosslinked with DCP) that were partially crosslinked for 20 min at 150 °C before being brought into contact and interlinked for a further 80 min to complete the crosslinking. More than one-half of the crosslinking density of the fully crosslinked sheet was generated in this way before the sheets were interlinked. Therefore, there was very little diffusion of elastomer chains across the interface after the sheets were brought together. From Figure 23 it is observed that the strength of adhesion decreases to a larger extent, and appears to reach a lower limit of about 5 J/m2 at 100 °C or higher. The results for

5

Log Ga (J/m2)

4 3 2 1 0 –50

0

50

100

150

Temperature (ºC)

Figure 4.23 Strength of autohesion Ga vs. test temperature for SBR sheets interlinked by heating in contact for 80 min at 150 °C after being crosslinked for 20 min at 150 °C against various mold surfaces: both surfaces rough steel, ; one surface rough steel, one surface Mylar, ; both surfaces smooth steel, ; both surfaces Mylar, . Figure reproduced from Ref. [14], with permission from Rubber Chemistry and Technology. Copyright (1995), Rubber Division, American Chemical Society Inc.

134 Progress in Adhesion and Adhesives, Volume 4

the samples prepared with different surface conditions were all the same over the entire temperature range. Thus, the degree of roughness or type of mold surface used to prepare sheets for interlinking had little, if any, effect. This was attributed to the cancellation of two opposing effects: some elastic repulsion by asperities and some benefit from enhanced bonded area. Values of peel strength Ga are plotted in Figure 24 against test temperature for NR samples (crosslinked with DCP) that were partially crosslinked for 20 min at 150 °C before being brought into contact and interlinked for a further 100 min to complete the crosslinking. While the bond strength was rather low at all temperatures with smooth (Mylar film, smooth mild steel) surfaces, the adhesion was extremely high at low temperatures against the rough surfaces (roughened mild steel). At high test temperatures it decreased considerably but remained about twice as high as for samples prepared with smooth surfaces (Mylar film, smooth mild steel) because of increased area of bonding. The authors showed that the effect of surface roughness is to enhance the bond strength by only a small factor, about 2 , when the adhesion strength is low but to cause a much larger increase when the adhesion strength is high. The relatively high adhesion strength shown by NR compounds was also attributed to the strain-induced crystallization at the peel front as the sheets were peeled apart. Values of peel strength Ga are plotted in Figure  25 against test temperature for NR samples (filled with CB and crosslinked with sulfur) that were partially crosslinked for 17 min at 140 °C before being brought into contact and interlinked for a further 13 min to complete the crosslinking. The degree of crosslinking was found to be rather small and the

4

Log Ga (J/m2)

3

2

1

0 –50

0

50

100

150

Temperature (ºC)

Figure  4.24 Autohesion Ga vs. test temperature for NR samples (peroxide cured) interlinked for 100 min at 150 °C after being crosslinked for 20 min. Symbols: both surfaces rough steel, ; both surfaces smooth, ; both surfaces Mylar, . Figure reproduced from Ref. [14], with permission from Rubber Chemistry and Technology. Copyright (1995), Rubber Division, American Chemical Society Inc.

Adhesion between Compounded Elastomers: A Critical Review 135 5

Log Ga (J/m2)

4

3

2

1

0 –50

0

50

100

150

Temperature (ºC)

Figure 4.25 Autohesion Ga vs. temperature for CB-filled NR samples interlinked for 13 min at 140 °C after being crosslinked for 17 min. Symbols: both surfaces rough steel, ; both surfaces smooth, ; both surfaces Mylar, . Figure adapted from Ref. [14], with permission from Rubber Chemistry and Technology. Copyright (1995), Rubber Division, American Chemical Society Inc.

values of Young’s modulus, E, were about the same (6 MPa) before and after interlinking. Strength of adhesion was correspondingly low. Only a slight increase in relative adhesion was shown by samples with a sheet premolded against a rough surface. Ruch et al. [3] studied the effect of interdiffusion mechanism on the interfacial cocrosslinking and adhesion between EPDM rubber joints. Two different crosslinking procedures leading to C-C bonds were used. The classic procedure of crosslinking by peroxide, where the diffusion of the chains at the interface and the network formation occur simultaneously, was performed at a high temperature. In addition, the authors also used electron beam at room temperature to separate the effect of the covalent bonds between the chains from the effect of the diffusion of the chains across the interface. In the first case, the authors tried to understand the role of pre-cure time and cocrosslinking on the adhesion strength between two EPDM rubber joints. Two elastomer sheets containing 2.5 phr of DCP were cured separately at 150 °C for various cure times (t1). The joint formation is described in Figure 26 (a). For times shorter than the decomposition time of the peroxide (about 100 min at 150 °C), only part of the peroxide reacted and the sheets were partially crosslinked to a degree, v1.Then the sheets were placed into contact and further co-crosslinked for t2 up to 150 min (t1 t2 150), leading to a bulk crosslinking degree of v. In this last step, high diffusion and co-crosslinking effects are expected because of the high temperature. The results plotted as adhesion strength vs. pre-curing time in Figure 27 were obtained at a peel rate of 5 mm/min. When the pre-cure time became shorter, the peel strength was higher. When two fully cured sheets were joined (t1 150 min), the peel strength was low

136 Progress in Adhesion and Adhesives, Volume 4 Pre-curing peroxide at 150 ºC t1 = varied

Contact

v1

Co-crosslinking peroxide at 150 ºC Total crosslink time (t1+t2) = 150 min v

Simultaneous interdiffusion and crosslinking

(a) Contact time = varied temperature = 90 ºC

Pre-curing peroxide at 150 ºC t1 = 20 min

(b)

Co-crosslinking peroxide at 150 ºC Time = 130 min

Contact time = 1 h to 624 h temperature = room temperature

Pre-curing peroxide at 160 ºC t1 = 1 h

Electron beam irradiation temperature = room temperature Dose = 130 kGy

(c) Contact time = 1 h temperature = 25 ºC to 170 ºC

Pre-curing peroxide at 160 ºC t1 = 1 h

Electron beam irradiation temperature = room temperature Dose = 130 kGy

(d)

Figure  4.26 Procedures used for joint formation and expected influences. Figure adapted with permission from Ref. [3]. Copyright 2000 John Wiley & Sons Inc.

G (J/m2)

1000

100

0

50

100

150

t1 (min)

Figure  4.27 Variation of peel strength with pre-curing time at a peel rate of 5  mm/min. Figure adapted with permission from Ref. [3]. Copyright 2000 John Wiley & Sons Inc.

(20 J.m 2), which was, however, higher than the reversible work of adhesion (0.05 J.m 2), indicating energy dissipation at the interface. In the second case, the authors tried to understand the role of chain diffusion and cocrosslinking on the adhesion strength between two EPDM rubber joints. The EPDM sheets containing 2.5 phr of DCP were partially crosslinked for 20 min at 150 °C (v 2–4 mol/m3 and sol fraction (fs) 40–50%). Residual peroxide was still active in the elastomer and

Adhesion between Compounded Elastomers: A Critical Review 137

free chains were present in large quantity. The sheets were then joined for various lengths of time at 95 °C before the crosslinking step at 150 °C for 130 min. The joint formation is described in Figure 26 (b). At 95 °C, only chain diffusion occurs whereas at 150 °C both crosslinking and chain diffusion are possible. The results shown in Figure 28 indicate that the contact step has no effect on the interfacial strength of peroxide post-crosslinked samples. The chains diffuse enough before the crosslinking immobilizes them and the co-crosslinking efficiency is the same in the interfacial domain whatever the degree of penetration reached after the contact step. In the third case, sheets of EPDM (100 phr of rubber 1 phr of DCP 1 phr of isopropylN’-phenyl-p-phenylenediamine (IPPD) were crosslinked for 1  h at 160 °C to generate lightly crosslinked networks (v 10 mol/m3) . There was a significant amount of polymer (25%) in the sol fraction (fs). There was no peroxide left after this heat treatment. The joints were made by contact of two sheets at room temperature for lengths of time ranging from 1 h up to 720 h (1 month). After this contact, the samples were irradiated with a constant dose of 130 kGy to produce networks with v 105 mol/m3 and fs 8%. The joint formation is summarized in Figure 26 (c). It was shown that up to approximately 500 h, there was no effect of the contact time on the interfacial strength. For longer times, there were some trends for increased strength. In the fourth case, sheets of EPDM (100 phr of rubber 1 phr of DCP 1 phr of IPPD) were crosslinked for 1 h at 160 °C to obtain lightly crosslinked networks (v 10 mol/m3). There was a significant amount of polymer (25%) in the sol fraction (fs). There was no peroxide left after this heat treatment. Two series of experiments were carried out (1) the contact was made at room temperature at various lengths of time. (2) the contact was made at various temperatures (25 °C to 170 °C) at a constant contact time (1  h). After this contact, the samples were irradiated with a constant dose of 130 kGy to

10000

tc=0min Tc= 95 ºC tc=5min Tc= 95 ºC tc=30min Tc= 95 ºC tc=60min Tc= 95 ºC tc=165min Tc= 80 ºC

G (J/m2)

1000

0.01

0.1

1

10

100

R (mm/min)

Figure 4.28 Peel strength vs. peel rate, R for peroxide post-crosslinked samples. Degree of pre-cure, v1 2–4 mol/m3, where tc represents time of contact and Tc represents temperature of contact. Figure adapted with permission from Ref. [3]. Copyright 2000 John Wiley & Sons Inc.

138 Progress in Adhesion and Adhesives, Volume 4

produce networks with v 105 mol/m3 and fs 8%. The joint formation is summarized in Figure 26 (d). In the first case, there was no effect of contact time on interfacial strength. On the other hand, interfacial strength was directly related to contact temperature for the second experiment. It has been shown that the efficiency of the crosslinking reaction in the interfacial zone is related to the interpenetration depth. The results showed that interfacial strength is directly related to the temperature of the contact step. The peel strength increased gradually with the temperature of contact. In order to check the effect of time of contact, selected samples were tested by varying the time of contact (30 min up to 6 h) at various temperatures. It was found that the time did not seem to play an important role, as shown by the average peel strengths obtained for times ranging from 30 min up to 6 h at various temperatures.

4.3

Adhesion Between Vulcanized Rubber and Unvulcanized Rubber or Partially Vulcanized Rubber

The previous section of this review has discussed the mechanisms and factors involved in the adhesion between unvulcanized rubber to unvulcanized rubber containing compounding ingredients and partially vulcanized rubber to partially vulcanized rubber through cocrosslinking. This section will discuss the mechanisms and factors involved in the adhesion between unvulcanized rubber to vulcanized rubber and partially vulcanized rubber to vulcanized rubber. In the first case, the adhesion between the two symmetric or asymmetric rubber layers is dictated by the magnitude of the diffusing rubber chains across the interface from the rubber layers. However, in the latter case, there are restrictions on the diffusing rubber chains across the interface, because of the immobilization of chains by crosslinking. Therefore, in this scenario, it is difficult to establish good bond strength between the joining rubber layers through diffusion. However, in many practical applications, rubber technologists are forced to face such challenging scenarios. One such classical scenario is repairing of the damaged conveyor belt. Generally, conveyor belts are designed in such a way as to run continuously for several miles which can lead to failure at certain parts of the belt or at the joints. This demands immediate repairing of the damaged portions using an unvulcanized rubber filled with crosslinking agents and other compounding ingredients. While repairing, the unvulcanized rubber compound is pressed over the damaged vulcanized portion of the conveyor belt and subsequently crosslinked by application of heat and pressure. The diffusion of the elastomer chains from the unvulcanized portion into the vulcanized portion and subsequent crosslinking will help to stitch and repair the damaged vulcanized rubber layer. The adhesion at the unvulcanized rubber - vulcanized rubber and partially vulcanized rubber - vulcanized rubber interfaces by crosslinking can be divided into various steps: Initially, there should be good wetting between the two joining rubber layers, which can lead to intimate contact. Once the contact is established, there will be diffusion of rubber chains across the interface from the unvulcanized rubber layer or partially vulcanized rubber layer into the vulcanized rubber layer. The diffused rubber chains get entrapped

Adhesion between Compounded Elastomers: A Critical Review 139

Vulcanized rubber layer

Unvulcanized rubber layer (filled with crosslinking agents) (a)

(b)

(c)

Figure 4.29 Formation of the interface between two elastomer sheets. (a) wetting stage (the open circles in the unvulcanized rubber layer correspond to the unreacted crosslinking agent in the unvulcanized rubber layer) (b) interdiffusion stage - diffusion of elastomer chains from the unvulcanized rubber layer to vulcanized rubber layer (the open circles in the unvulcanized rubber layer correspond to the unreacted crosslinking agent in the unvulcanized rubber layer) (c) crosslinking stage (dots: crosslinks in the bulk, stars: crosslinks between the diffused rubber chains from unvulcanized rubber layer and the rubber chains in the vulcanized rubber layer during crosslinking reactions).

inside the vulcanized rubber layer during crosslinking process. Schematically outlined in Figure 29 is the mechanism involved in the formation of interface between an unvulcanized rubber layer and a vulcanized rubber layer by crosslinking. At the initial contact time (t 0) of the two elastomer surfaces wetting of the surfaces takes place which leads to an intimate contact (Figure  29 (a)). The open circles in the unvulcanized rubber layer correspond to the unreacted crosslinking agent in the unvulcanized rubber layer (right side image of Figure 29 (a)). The crosslinks that form in the bulk of vulcanized rubber layer are shown by dots (left side image of Figure 29 (a)). At some intermediate contact time (t 0), diffusion of the elastomer chains across the interface takes place from the unvulcanized rubber layer into the vulcanized rubber layer (Figure 29 (b)). The open circles in the unvulcanized rubber layer correspond to the unreacted crosslinking agent in the unvulcanized rubber layer (right side image of Figure 29 (b)). The crosslinking that takes place in the bulk of vulcanized rubber layer is shown by dots (left side image of Figure 29 (b)). At longer contact time (t t ), the diffused rubber chains get entangled/entrapped inside the vulcanized

140 Progress in Adhesion and Adhesives, Volume 4

rubber layer during crosslinking (shown by stars in Figure 29 (c)). These are shown by dots in the images of Figure 29 (c). In the literature, there are research papers which describe (a) adhesion between vulcanized rubber and partially vulcanized rubber (filled with crosslinking agents) (Figure 30 (a)) and (b) and adhesion between vulcanized rubber and unvulcanized rubber (filled with crosslinking agents) (Figure 30 (b)) [14–21].

4.3.1

Adhesion between Vulcanized Rubber and Unvulcanized Rubber (Filled with Crosslinking Agents)

Bhowmick and co-workers [15] investigated the peel adhesion between vulcanized EPDM rubber (unmodified and modified with electron beam (EB) irradiation) and gum natural rubber (NR) by a 180° peel test. The unmodified gum NR was compression molded at a temperature of 100 °C and under 5 MPa pressure. Two minutes were given for preshaping the unmodified gum NR sheet (1 mm thick). Compounding of EPDM rubber with standard compounding ingredients was done in a two-roll open mixing mill. The compounded EPDM rubber mixture was compression molded between Teflon sheets at a temperature of 150 °C and a pressure of 5 MPa for the optimum cure time in an electrically heated press. Modification of the EPDM vulcanizate surface was carried out by EB irradiation with or

Interface

(a)

Vulcanized rubber

Partially vulcanized rubber

Fully crosslinked rubber-rubber joints Interface

(b)

Vulcanized rubber

Curatives

Unvulcanized rubber (filled with crosslinking agents)

Partial crosslink

Fully crosslinked rubber-rubber joints

Crosslink

Co-crosslink

Figure 4.30 Schematic for interlinking of (a) vulcanized rubber to partially vulcanized rubber, and (b) vulcanized rubber to unvulcanized rubber.

Adhesion between Compounded Elastomers: A Critical Review 141

without EB sensitizers such as trimethylol propane triacrylate (TMPTA) and tripropylene glycol diacrylate (TPGDA). The doses of both EB and TMPTA/TPGDA were varied in separate experiments keeping one of them constant. The surface modified vulcanized EPDM rubber sample (2  mm thick) was placed into a 3  mm mold cavity along with an aluminium foil of 2 cm width on its upper portion. The preshaped unmodified NR sheet was kept over the vulcanized modified or unmodified EPDM rubber samples. The whole assembly was placed in the hot press until curing of the partially vulcanized NR was completed. During this process, the fully vulcanized EPDM rubber in the assembly remained in the hot press well beyond its optimum cure time. The gel content was measured, but there was no change. Hence, overvulcanization did not change the state of cure. The surface features were analyzed by using several analytical techniques such as attenuated total reflection infrared spectroscopy (ATR-IR), X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDX), surface energy measurements, and free sulphur analysis. ATR-IR spectra of EB irradiated unsensitized EPDM vulcanizate showed C–O–C polar groups (C–O–C symmetric stretching at 1032 and 1088 cm−1) after irradiation. The band intensities of both peaks portrayed significantly higher values than that obtained from the unmodified state, especially at 20 kGy irradiation dose; at still higher doses, these decreased to lower values (Figure 31 (a)). It was seen that insignificant amounts of >C=O groups (due to oxidation) were found on the EPDM surface. This was due to the higher ‘bond energy’ of C=O as compared to C–O–C. On irradiation, free radicals were likely to be generated which subsequently reacted with atmospheric oxygen to form oxygenated groups such as ethers. The absorbance values of C–O–C groups at 1088 cm−1 and 1032 cm−1and for different irradiation doses at constant (10 wt%) TMPTA concentration are plotted in Figure 31 (b). The band intensities increased up to 50 kGy irradiation dose and then dropped. The presence of small >C=O stretches at 1730 cm−1 in the EPDM surface was observed in this case, but no significant variation in its concentration was noticed on further increasing the dose (Figure  31 (b)). The >C=O stretching was mainly due to the presence of TMPTA and partly due to oxidation of the elastomer in presence of the sensitizer. Increasing band intensity of C–O–C implied more surface oxidation of the sensitized rubber vulcanizates. The decreasing trend (of C–O–C) was possibly due to escaping tendency of the monomers (light weight molecules) at higher irradiation doses as a consequence of high heat generation. With increasing irradiation dose at constant TMPTA concentration (10 wt%), the band intensity at 698 cm−1 corresponding to C–S–C stretching vibration (cross-links), decreased (Figure 31 (c)). This indicated a breakdown of cross-links and subsequent blooming of the free sulfur to the surfaces under higher irradiation doses. Variation of TMPTA concentration also showed maximum absorbance values for A1088 and A1032 at 10 wt% concentration (Figure  31 (d)). A1730 increased with TMPTA concentration. TPGDA showed similar results (Figure 31 (d)).

0.14

A1032

0.12

A1088

0.10 0.08 0.06 0.04 0.02 0.00

0

50

100

150

Absorbance of polar groups

Absorbance of polar groups

142 Progress in Adhesion and Adhesives, Volume 4 0.14

A1088 A1032 A1730

0.12 0.10 0.08 0.06 0.04 0.02 0.00 0

200

Absorbance of polar groups

A698

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 50

100

150

100

150

200

Irradiation dose (kGy)

Irradiation dose (kGy)

0

50

200

Irradiation dose (kGy)

A1088 A1032 A1730

0.12 0.10 0.08 0.06 0.04 0.02 0.00 0

5

10

15

20

TMPTA concentration (wt%)

Figure 4.31 Plots of absorbance of polar groups for (a) EPxkGy/0, (b) EPxkGy/10 TM samples as a function of the irradiation dose, (c) plot of absorbance of C–S–C stretching vibration for EPx kGy/10 TM samples as a function of the irradiation dose and (d) plots of absorbance of polar groups for EP100 kGy/y TM samples as a function of the TMPTA concentration, where EPx kGy/0 represents EB irradiated EPDM rubber (with various irradiation doses) without sensitizer, EPx kGy/10 TM represents EB irradiated EPDM rubber (with various irradiation doses) with 10 wt % TMPTA, and EP100 kGy/y TM represents EB irradiated EPDM rubber (100 kGy irradiation dose) with various TMPTA concentrations. G. C. Basak, A. Bandyopadhyay, Y. K. Bharadwaj, S. Sabharwal and A. K. Bhowmick, 2009, Adhesion of vulcanized rubber surfaces: Characterization of unmodified and electron beam modified EPDM surfaces and their co-vulcanization with natural rubber, Journal of Adhesion Science and Technology, reprinted by permission of Taylor & Francis Ltd., Ref. [15].

It was further affirmed from the XPS analysis (Figure 32) that the peaks corresponding to C-O and C=O significantly shifted towards higher binding energies which indicates oxidative modification of the carbon at the EPDM surface in the form of C-O-C and C=O. It was also reported that the concentration of oxygen increased more with increasing irradiation dose in the presence of TMPTA than the samples devoid of TMPTA. This implies that presence of TMPTA as a sensitizer favors formation of more oxidized groups on the surface. For instance, the oxygen concentration increased from 28.1 to 30.8 % when EPDM

Adhesion between Compounded Elastomers: A Critical Review 143 EP 0/0

EP 20kGy/0 C1s

20000

40000

20000

0

Intensity (Counts/sec)

40000

C1s

60000

Intensity (counts/sec)

60000

0 282

(a)

284

286

288

282

290

(b)

Binding energy (eV)

284

286

288

Binding energy (eV)

60000

EP 10kGy/10TM

O1s

60000

0

20000

282

284

286

528

288

(d)

Binding energy (eV)

160

S2p

20 0

12000 Intensity (Counts/sec)

40

10000 8000 6000 4000 2000

530

532

534

536

S2p

0

159 160 161 162 163 164 165 166

(e)

0/0

EP 20kGy/0

14000

120

60

EP

Binding energy (eV)

16000

EP 0/0

140

80

EP 20kGy/0

0

(c)

100

Intensity (Counts/sec)

20000

40000

EP 100kGy/10TM

Intensity (Counts/sec)

40000

Intensity (Counts/sec)

C1s

159 160 161 162 163 164 165 166

(f)

Binding energy (eV) 16000

EP 100kGy/10TM

14000

Binding energy (eV)

S2p

10000 8000 6000 4000 2000

Intensity (Counts/sec)

12000

0

(g)

159 160 161 162 163 164 165 166 Binding energy (eV)

Figure  4.32 XPS spectra for different systems: (a–c) C1  s (d) O1  s and (e–g) S2p spectra for EP0/0, EP20kGy/0 and EP100kGy/10TM samples, where EP0/0 represents EPDM rubber without irradiation, EP20kGy/0 represents EB irradiated EPDM rubber (20 kGy irradiation dose) without sensitizer and EP100kGy/10TM represents EB irradiated EPDM rubber (100 kGy irradiation dose) with 10 wt% TMPTA concentration. G. C. Basak, A. Bandyopadhyay, Y. K. Bharadwaj, S. Sabharwal and A. K. Bhowmick, 2009, Adhesion of vulcanized rubber surfaces: Characterization of unmodified and electron beam modified EPDM surfaces and their co-vulcanization with natural rubber, Journal of Adhesion Science and Technology, reprinted by permission of Taylor & Francis Ltd., Ref. [15].

144 Progress in Adhesion and Adhesives, Volume 4

surface was modified at 20 kGy irradiation dose and from 28.1 to 36.5% for the 100 kGy irradiation dose in presence of TMPTA ( 10 wt%) as a sensitizer (Figure 32). The increased polarity was further determined from surface energy parameters calculated from contact angle measurements. The surface energy of the modified surfaces increased up to certain extent and then decreased in presence of TMPTA or TPGDA system in comparison to only EB modified surfaces [e.g. 64.2 mJ/m2 to 84.0 mJ/m2 maximum for unmodified to EB modified systems, while 64.2 mJ/m2 to 88.0 mJ/m2 maximum for TMPTA modified systems]. These results correlated well with the analysis done by using ATR-IR. On the contrary, the free sulfur increased on the modified surfaces after irradiation which is probably due to breakage of some polysulfidic crosslinks under high energy, offering free sulfur to bloom to the surface. This was confirmed from EDX as well as free sulfur determination through chemical analysis. A morphological analysis of EDX study of a few representative samples is shown in Figure 33. SEM micrographs of the representative unmodified and modified EPDM surfaces are displayed in Figure 34. The white deposits over the surface of unmodified EPDM vulcanizate in Figure 34 (a) indicated the migration of different ingredients which causes microroughening of the surface. This increased on irradiating the sample with 20 kGy (Figure 34 (b)) as a consequence of surface oxidation and modification (grafting, breakdown of crosslinks, and sulfur blooming). Figures 34 (c–d) show SEM micrographs of 10 wt% TMPTA soaked and irradiated at 100 and 200 kGy EPDM vulcanizates, respectively. The deposits in the latter sample were significantly less than in the former. The deposits were mainly composed of some grafted TMPTA along with other ingredients. At a higher irradiation dose (20 kGy), loss of especially ungrafted and free sensitizer molecules probably decreased the amount of deposits and the irregularity. This, on the other hand, decreased the surface polarity and surface energy value. The adhesion strength between surface-modified vulcanized EPDM rubber and unmodified natural rubber (NR) was studied by the 180o peel test. For the unmodified EPDM vulcanizate, peel strength increased up to 20 kGy (39% increment from the original value) and then decreased. When EPDM was subjected to different irradiation doses at a constant TMPTA or TPGDA concentration (Figure 35 (a)), the peel strength increased up to

Ska, 3

Ska, 4 (b)

(a)

2 μm

Ska, 4 (c)

2 μm

2 μm

Figure 4.33 EDX mapping of sulfur for (a) EP0/0, (b) EP20kGy/0 and (c) EP100kGy/10TM samples. G. C. Basak, A. Bandyopadhyay, Y. K. Bharadwaj, S. Sabharwal and A. K. Bhowmick, 2009, Adhesion of vulcanized rubber surfaces: Characterization of unmodified and electron beam modified EPDM surfaces and their co-vulcanization with natural rubber, Journal of Adhesion Science and Technology, reprinted by permission of Taylor & Francis Ltd., Ref. [15].

Adhesion between Compounded Elastomers: A Critical Review 145

(a)

(b)

(c)

(d)

Figure 4.34 SEM micrographs of (a) EP0/0, (b) EP20kGy/0, (c) EP100kGy/10TM, and (d) EP200kGy/10TM samples, where EP200kGy/10TM represents EB irradiated EPDM rubber (200 kGy irradiation dose) with 10 wt% TMPTA concentration. G. C. Basak, A. Bandyopadhyay, Y. K. Bharadwaj, S. Sabharwal and A. K. Bhowmick, 2009, Adhesion of vulcanized rubber surfaces: Characterization of unmodified and electron beam modified EPDM surfaces and their co-vulcanization with natural rubber, Journal of Adhesion Science and Technology, reprinted by permission of Taylor & Francis Ltd., Ref. [15].

100 kGy and then decreased. Notably, the TMPTA modified sample showed the maximum improvement (71% for TMPTA vs. 43% for TPGDA) in adhesion strength. Figure 35 (b) demonstrates the peel behavior between the modified EPDM vulcanizates with gum NR rubbers where the vulcanizate systems were modified with different TMPTA and TPGDA concentrations (0–20 wt%) and irradiated at a fixed dose of 100 kGy. These results showed a similar trend to the earlier system, depicted in Figure 35 (a). The adhesion strength gradually increased, reached a maximum and then finally decreased; the TMPTA soaked EPDM vulcanizates showed better results than TPGDA soaked samples. Basically, the peel strength results between the modified EPDM vulcanizate and NR (cured by joining a fully vulcanized EPDM rubber sheet and a gum NR sheet containing vulcanizing agent) are found to be dependent on various factors like oxidation, grafting, breakdown of crosslinks and blooming of sensitizer to the surface. In addition, surface polarity improved the interaction between the polar materials at the interface, while surface irregularity contributed

146 Progress in Adhesion and Adhesives, Volume 4 120

EP xkGy/0(c) -NR(uc)

Percent change in peel strength

EP xkGy/10TM(c) -NR(uc) 90

EP xkGy/10TP(c) -NR(uc)

60

30

0 0 (a)

50

100

150

200

Irradiation dose (kGy)

120

0

5

10

15

20

EP xkGy/0(c) -NR(uc) EP xkGy/10TM(c) -NR(uc)

Percent change in peel strength

100

EP xkGy/10TP(c) -NR(uc)

80 60 40 20 0 0

(b)

5

10

15

20

Monomer concentration (wt.%)

Figure 4.35 Plots of percent change in peel strength between modified EPDM vulcanizate and uncured NR; EPDM surface modified (a) with varying irradiation dose with and without using a sensitizer (10 wt%) (b) with varying sensitizer concentration (wt%) at constant 100 kGy irradiation dose and varying irradiation dose without using sensitizer (suffix c and pc imply fully crosslinked rubber layer and partially crosslinked rubber layer, respectively). G. C. Basak, A. Bandyopadhyay, Y. K. Bharadwaj, S. Sabharwal and A. K. Bhowmick, 2009, Adhesion of vulcanized rubber surfaces: Characterization of unmodified and electron beam modified EPDM surfaces and their co-vulcanization with natural rubber, Journal of Adhesion Science and Technology, reprinted by permission of Taylor & Francis Ltd., Ref. [15].

to mechanical anchorage. Therefore, higher polarity and surface micro-roughness result in higher peel strength [16]. On the other hand, peel strength decreased at higher irradiation doses due to the migration of sulfur and other ingredients to the rubber surface which reduced the interfacial contact and consequently retarded the adhesion. Basak et al. [17] investigated the adhesion strength between vulcanized EPDM rubber (unmodified and modified with gamma irradiation) and unmodified NR by the 180o peel

Adhesion between Compounded Elastomers: A Critical Review 147

test. The procedure for preparing the samples was the same as before. Surface modification of EPDM vulcanizate was carried out by: (a) irradiation of the surface by gamma radiation in the presence and absence of trimethylol propane triacrylate (TMPTA) as a sensitizer. The samples were then characterized extensively by various sophisticated techniques as in the case of electron beam modification. The -modified samples in both the presence and absence of TMPTA showed higher peel strengths at a particular irradiation than the unmodified sample. However, it was noted that the sample in the presence of TMPTA at 1 kGy -irradiation dose showed maximum peel strength in comparison to the other modified samples (Figure 36). From the SEM micrographs, it was observed that the surface texture increased only marginally upon exposure to irradiation both in the presence or absence of TMPTA and the increment was not prominent after a certain dose. On the contrary, the concentration of free sulfur on the modified surface also increased with increasing irradiation dose. Therefore, it was suggested that a 1 kGy dose both in the presence and absence of 10 wt% TMPTA was optimum in this study. A maximum of ~76% improvement in peel strength was obtained for the -modified system in the presence of TMPTA at 1 kGy -irradiation dose (Figure  36). Similar experiments were carried out with TCI modified samples [17]. It was observed that the peel strength increased with increasing concentration of TCI up to 0.5 wt%, beyond which it decreased. Plasma modification of EPDM vulcanizate was also undertaken in this elaborate study [18]. The adhesion strength between vulcanized EPDM rubber (unmodified and modified with argon/oxygen irradiation) and unmodified NR was measured by the 180o peel test. The Irradiation dose (kGy) at constant 10 wt%TMPTA 0.0

0.5

1.5

2.0

2.5

3.0

EP xkGy/10TM (c) -NR(uc)

0.12 Peel strength (N/mm)

1.0

EP xkGy/ 0 (c) -NR(uc)

0.10

0.08

0.06 0.0

0.5

1.0 1.5 2.0 2.5 Irradiation dose (kGy)

3.0

Figure 4.36 Peel strength of EPxkGy/0 and EPxkGy/10TM (suffixes c and uc denote fully crosslinked rubber layer and uncrosslinked rubber layer respectively). G. C. Basak, A. Bandyopadhyay, Y. K. Bharadwaj, S. Sabharwal and A. K. Bhowmick, 2010, Characterization of EPDM vulcanizates modified with gamma irradiation and trichloroisocyanuric acid and their adhesion behavior with natural rubber, Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [17].

148 Progress in Adhesion and Adhesives, Volume 4

pre-shaped NR sample was prepared by pressing a layer of NR at 100 °C for 2 min between smooth aluminium foils at 5 MPa pressure in an electrically heated press. Compounding of EPDM rubber with standard compounding ingredients was done in a two-roll open mixing mill. The compounded EPDM rubber mixture was compression molded at a temperature of 150 °C and a pressure of 5 MPa for 15 min in an electrically heated press. Surface modification of EPDM vulcanizate was carried out by treating the vulcanized EPDM rubber surface in a radiofrequency capacitively coupled low pressure argon/oxygen plasma reactor. The surface modified vulcanized EPDM rubber sample of 1.5 mm thickness was kept in the 3 mm mold cavity with an aluminium foil of 20 mm width on its upper portion. The Al foil was used to produce a jaw for peel strength measurement. The pre-shaped NR was placed over the modified or unmodified vulcanized EPDM rubber sample. The whole assembly was placed into the hot molded press at a temperature of 150 °C and a pressure of 5 MPa until curing of the partially vulcanized NR was completed. Figure  37 shows the ATR-FTIR spectrum in the region of 3000–750 cm−1 for the untreated vulcanized EPDM rubber surface (EP0/0/0). The untreated sample was characterized by three major bands corresponding to hydrocarbon moieties: (1) the two bands located in the 2926–2856 cm−1 region, which resulted from saturated hydrocarbon backbone; (2) the bands located in the region of 1460 cm−1 and 1376 cm−1, which were caused by –CH2 scissoring and –CH3 stretching vibration of ethylene–propylene group, and (3) the band around 722 cm−1 which was due to –CH2 rocking for long ethylene sequence. Similarly, ATR-FTIR spectra of EP0/0/0 and the vulcanized samples after treatment with various rf powers at constant 10 min. exposure and 90% Ar concentration are presented in 0.5 Absorbance (a.u.)

Absorbance (a.u.)

0.4

0.20

0.15

0.10 722

617 698

0.05

623 632

609

0.3 1154

0.00 720 680 640 Wavenumber (cm–1)

0.2 2926

600

1376 1460

1088

1248 1032 1014 760 722

1538 2856

0.1

972

698

0.0 3000

2500

2000

1500

1000

–1

Wavenumber (cm )

Figure 4.37 ATR-FTIR spectrum of untreated EPDM vulcanizate surface. Figure reproduced with permission from Ref. [18]. Copyright 2011 Elsevier.

Adhesion between Compounded Elastomers: A Critical Review 149

Figure 38 (a). New functional group (C=O) at 1730 cm−1 was observed as a result of surface oxidation. The absorbances at 1154 cm−1 and 1730 cm−1 due to asymmetric stretching vibration of C–O–C and carbonyl group (-C=O) against plasma power are plotted in Figure 38 (b). The peak intensity at 1154 cm−1 registered a slight upward trend, while that at 1730 cm−1 exhibited substantial increase with plasma power. The peak intensity (marginal for ether group and significant for carbonyl group) was enhanced up to about 100 W of plasma power, beyond which there was no significant change. This was attributed to various reactions, e.g., chain scission, self-crosslinking, and disproportionation reactions at higher power. The topographic features of the untreated and plasma treated surfaces were studied using SEM. Figure 39 (a)-(c) displays the SEM micrographs for plasma treated samples at two different concentrations of Ar and compares these with the surface of EP0/0/0. The treatment

1730 EP

EP

0.15 0.10 0.05

EP

EP

3000

(a)

0.5

698

0.4 EP

0.15 0.10 0.05

0.15 0.10 0.05

125W/10/90Ar

1376 1088 9721 032 1460 1248 760 1154 722 101

2500

–1

1730 cm

100W/10/90Ar

Absorbance (a.u.)

Absorbance (a.u.)

1538

0.15 2926 2856 0.10 0.05 0.00 0.20 0.15 0.10 0.05

50W/10/90Ar

25W/10/90Ar

–1

1154 cm

0.3 0.2 0.1 0.0

0/0/0

2000

1500

1000

–1

Wavenumber (cm )

0

(b)

20

40

60

80

100

120

140

Plasma power (W)

Figure 4.38 (a) ATR-FTIR spectra of different plasma treated samples and (b) C–O–C (1730 cm 1) and C=O (1154 cm 1) peak intensities against various plasma powers. Figure reproduced with permission from Ref. [18]. Copyright 2011 Elsevier.

(a)

(b)

(c)

Figure  4.39 SEM micrographs of (a) EP0/0/0, (b) EP100  W/10/90Ar, (c) EP100  W/10/95Ar, where EP0/0/0 represents untreated EPDM rubber sample, EP100 W/10/90Ar represents EPDM rubber treated with 90% concentration of Ar at 100 W power for 10 min, and EP100 W/10/95Ar represents EPDM rubber treated with 95% concentration of Ar at 100 W power for 10 min. Figure reproduced with permission from Ref. [18]. Copyright 2011 Elsevier.

150 Progress in Adhesion and Adhesives, Volume 4

with different Ar percentages showed that the unevenness of the surface increased with increasing Ar concentration. Active species generated in argon plasma increased the surface roughness of the vulcanized EPDM rubber. Figure 40 (a–f) displays the AFM micrographs of the untreated and the treated surfaces of vulcanized EPDM. By increasing the plasma power, the unevenness on the surface was enhanced in comparison to that of the untreated one. Figure 41 illustrates that the peel strength was a function of plasma power, exposure time and composition of the gas mixture used. Significant decrement of peel strength was observed with increasing Ar concentration (Figure 41 (a)). It was observed (from the SEM and the AFM analysis) that surface roughness was enhanced with increasing concentration of Ar, while the polarity of the surface (as determined from the ATR-FTIR analysis, surface energy values and XPS analysis) decreased with increasing Ar percentage. Hence, the values of peel strength indicated that surface polarity governs the peel strength [16]. On the other hand, it was observed that the sample treated at 100 W plasma power, exposure time of 10 min. and 90/10 Ar/O2 gas mixture showed maximum peel strength in comparison to the other modified samples (Figure 41 (b)). The adhesion strength of the joints increased with increasing exposure time of plasma up to 10 min beyond which it showed almost constant value (Figure 41 (c)).

10.0

(a)

0

2.5

5.0

7.5

(d)

0

2.5

5.0

7.5

10.0

(b)

10.0

(c)

7.5

7.5

7.5

5.0

5.0

5.0

2.5

2.5

2.5

0 10.0 μm 0 10.0

2.5

5.0

7.5

(e)

0 10.0 10.0

0

2.5

5.0

7.5

(f)

0 10.0 10.0

7.5

7.5

7.5

5.0

5.0

5.0

2.5

2.5

2.5

0 10.0 μm 0

2.5

5.0

7.5

0 10.0 μm 0

2.5

5.0

7.5

0 10.0

Figure 4.40 AFM micrographs of (a) EP0/0/0, (b) EP50 W/10/90Ar, (c) EP100 W/10/90Ar, (d) EP125 W/10/90Ar, (e) EP100  W/5/90Ar, and (f) EP100  W/10/95Ar, where EP50  W/10/90Ar represents EPDM rubber treated with 90% concentration of Ar at 50 W power for 10 min, EP100 W/10/90Ar represents EPDM rubber treated with 90% concentration of Ar at 100 W power for 10 min, EP125 W/10/90Ar represents EPDM rubber treated with 90% concentration of Ar at 125  W power for 10 min, EP100  W/5/90Ar represents EPDM rubber treated with 90% concentration of Ar at 100 W power for 5 min, and EP100 W/10/95Ar represents EPDM rubber treated with 95% concentration of Ar at 100 W power for 10 min. Figure reproduced with permission from Ref. [18]. Copyright 2011 Elsevier.

Adhesion between Compounded Elastomers: A Critical Review 151 0.150

0.14 10min-90% Ar

100W-2 min 0.12 0.10 Peel strength (N/mm)

Peel strength (N/mm)

0.125

0.100

0.075

0.08 0.06 0.04 0.02

0.050

0.00 75

(a)

80

85 90 Concentration of Ar (%)

95

100

0

20

40

(b)

60

80

100

120

140

RF power (W)

0.15 100W-90% Ar

0.14 0.13

Peel strength (N/mm)

0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0

(c)

4

8

12

16

20

24

Exposure time (min)

Figure 4.41 Peel strength as a function of (a) concentration of Ar gas used, (b) plasma power, and (c) plasma treatment time. Figure reproduced with permission from Ref. [18]. Copyright 2011 Elsevier.

The authors concluded that the degree of surface modification was dependent on the plasma composition and operating parameters including rf power and treatment time. The surface treatment of vulcanized EPDM rubber by plasma was found to significantly improve the adhesion between vulcanized EPDM rubber and uncured NR with vulcanizing agents. The authors suggested that the plasma treatment of the vulcanized EPDM rubber surface is more efficient for improving the adhesion strength between vulcanized EPDM rubber and NR when compared to other surface treatment techniques such as electron beam irradiation, gamma irradiation and chemical treatment [16–18]. Commonly, tackifiers are used for enhancing adhesion between two rubber sheets. Basak et al. [19] studied the effect of tackifier compatibility and blend viscoelasticity on the peel

152 Progress in Adhesion and Adhesives, Volume 4

behavior of vulcanized EPDM rubber joined with EPDM rubber containing vulcanizing agents. The samples for adhesion studies were prepared in two stages. The fully vulcanized EPDM rubber was prepared in stage one and the details are given below. The EPDM rubber was compounded with standard compounding ingredients (without tackifier) in a two-roll open mixing mill. The compounded EPDM rubber (without tackifier) was completely vulcanized to a given degree at 150 °C under 5 MPa in an electrically heated press. Another substrate of EPDM rubber was compounded with standard compounding ingredients (along with tackifier) in a two-roll open mixing mill. The compounded EPDM rubber (along with tackifier) was pre-shaped in an electrically heated press at 100 °C under 5 MPa pressure for 2 min. The authors confirmed that there was no crosslinking during this procedure. Two different resins namely hydrocarbon (HC) and coumarone indene (CI) were used as tackifiers in this study. A wide range of concentrations of the tackifiers from 2 to 32 wt% with respect to EPDM was investigated. The vulcanized EPDM sheet of 1.5 mm thickness was placed in the 3 mm mold cavity along with an aluminium foil of 20 mm width on its upper portion. The pre-shaped EPDM rubber was kept over the fully vulcanized EPDM rubber sample. The whole assembly was placed in the hot press until curing of the partially vulcanized EPDM rubber portion was completed. The adhesion strength was measured by the 180° peel test. The tan vs. temperature plots for pure tackifiers are shown in Figure  42. The glass transition temperature (Tg) of the neat HC resin tackifier was 58 °C and that of neat CI resin tackifier was 84 °C. Tan vs. temperature plots for HC-EPDM blends are illustrated in Figure 43. For all the HC–EPDM blends, only one well-defined Tg lying in between the Tg of neat EPDM 0.6 58 ºC

84 ºC

0.5 HC resin CI resin

Tan δ

0.4

0.3

0.2

0.1

0.0 0

20

40 60 Temperature (ºC)

80

100

Figure 4.42 Tan against temperature plots of pure tackifiers. Figure reproduced with permission from Ref. [19]. Copyright 2010 Elsevier.

Adhesion between Compounded Elastomers: A Critical Review 153 1.6 1.4

–38 ºC –37 ºC–35 ºC –34 ºC –32 ºC –26 ºC –23 ºC

1.6 1.4

1.2

1.2 1.0 Tanδ

Tanδ

1.0 0.8

0 phr 2 phr 4 phr 8 phr 16 phr 24 phr

0.8 0.6

0.6

0.4

0.4

0.0

0.2 –60

0.2

–40 –20 0 Temperature (ºC)

20

40

80

32 phr

0.0 –60

–40

–20

0

20

60

Temperature (ºC)

Figure  4.43 Tan vs. temperature plots of HC tackifier modified EPDM vulcanizates. Figure reproduced with permission from Ref. [19]. Copyright 2010 Elsevier.

( 38 °C) and neat HC (58 °C) was observed. The Tg value of EPDM rubber gradually shifts to higher temperature with increasing concentration of HC resin tackifier. Simultaneously, tan max was reduced on increasing HC concentration in the blend. Existence of a single Tg was a clear evidence of good compatibility between EPDM and HC throughout the concentrations studied. Increase of Tg with the addition of tackifier indicated a reduction in free volume available for local segmental motions. Simultaneous broadening of tan curves was possibly due to increase in the extent of molecular relaxations and plasticization in HC–EPDM blends. The Tg values of HC resin tackifier filled EPDM rubber vulcanizates are given in Table 17. Figure 44 shows the dynamic storage modulus curves of HC-EPDM blends. In the rubbery plateau region, the tackifier acts as a diluent and causes a decrease in the storage modulus value. The entanglement density in the rubbery plateau zone was accurately determined from the parameters such as entanglement molecular weight (Me) and network density ( ) (moles of network strands per cubic centimeter). The aforementioned parameters were calculated to understand the effect of the HC resin tackifier on the rubbery plateau modulus. The entangle0 ment molecular weight (Me) was estimated from the plateau modulus ( Gn ) as follows:

Me

RT / Gn0

(16)

where is the density of the polymer or blend, R is 8.31 107 dyne-cm/mol K, T is the absolute temperature where Gn0 is located, and Gn0 is determined from the storage modulus

154 Progress in Adhesion and Adhesives, Volume 4

Table 4.17 Effect of tackifiers on the viscoelastic properties of EPDM vulcanizates [19].

Sample designation

Plateau modulus Tan max (Tg) (°C)

Gn (MPa)

Entanglement molecular weight (oe) (g/mol)

Network density v 103 (mol/cm3)

EP

38

2.72

736

1.17

EPHC2

37

2.48

807

1.06

EPHC4

36

2.39

838

1.03

EPHC8

34

2.12

944

0.91

EPHC16

32

2.02

991

0.87

EPHC24

26

1.83

1094

0.78

EPHC32

22

1.72

1164

0.74

EPCI2

39

2.83

703

1.21

EPCI4

39,36

2.93

666

1.34

EPCI8

39,60

3.01

611

1.39

EPCI16

38,67

3.23

566

1.44

EPCI24

38,80

3.56

520

1.51

EPCI32

38,84

3.67

475

1.56

where EP represents neat EPDM, EPHC2 represents 2 wt% of HC-EPDM, EPHC4 represents 4 wt% of HC-EPDM, EPHC8 represents 8 wt% of HC-EPDM, EPHC16 represents 16 wt% of HC-EPDM, EPHC24 represents 24 wt% of HC-EPDM, EPHC32 represents 32 wt% of HC-EPDM, EPCI2 represents 2 wt% of CI-EPDM, EPCI4 represents 4 wt% of CI-EPDM, EPCI8 represents 8 wt% of CI-EPDM, EPCI16 represents 16 wt% of CI-EPDM, EPCI24 represents 24 wt% of CI-EPDM, and EPCI32 represents 32 wt% of CI-EPDM

(G ) at the onset of the rubbery region (usually where tan reaches minimum following the prominent maximum). Furthermore, the plateau modulus was related to the network density ( ). The relationship between the plateau modulus and network density ( ) is given by the equation:

Gn0

g n RT

(17)

On rearranging the above equation, the network density was calculated from the following equation:

Gn0 / g n RT

(18)

In the above equation, ( ) is the moles of network strands per cubic centimeter, gn is 0 a numerical factor and is generally taken as unity. The Gn , Me and values of the neat

Adhesion between Compounded Elastomers: A Critical Review 155 3.5 3.0 Log storage modulus (MPa)

0 phr 2 phr

2.5

4 phr 8 phr

2.0

16 phr 24 phr

1.5

32 phr 1.0

0.5 0.0 –80

–60

–40

40 –20 0 20 Temperature (ºC)

60

80

100

Figure 4.44 Log storage modulus vs. temperature plots of HC blended EPDM vulcanizates. Figure reproduced with permission from Ref. [19]. Copyright 2010 Elsevier.

EPDM rubber and EPDM /HC resin blends are reported in Table 17. It was observed that the Me value increased from 736 to 1164 for the compatible EPDM/HC resin blends. It was suggested that the HC resin tackifier significantly reduced the entanglement of the EPDM polymer, and the HC resin tackifier essentially acted as a diluent in the plateau region. The network density ( ) value decreased with the increase in the concentration of the HC resin tackifier. This further confirms the diluent effect shown by the HC resin tackifier in the plateau zone. It was pointed out that the observed depression of storage plateau modulus in EPDM rubber– HC resin blends enhances the interfacial contact and compliance between the joining rubber layers and raises the contact area for better adhesion. Figure 45 shows tan vs. temperature plots of neat EPDM and representative CI-EPDM blends. Tan peak did not change appreciably with incorporation of CI resin in the EPDM rubber matrix. Furthermore, CI modified EPDM system showed a single Tg up to 2 wt% concentration, while the other compositions had two Tg’s close to individual rubber and tackifier indicating a complete phase separation between the two; in fact the Tg of the rubber phase stayed at 39 °C in all blend compositions. The tan max corresponding to CI resin phase in the blend was shifted towards higher temperature with an increase in tackifier concentration and at 32 wt% CI concentration, two Tg’s were exactly pointed at 38 and 84 °C. This further indicated coexistence of two separate phases in EPDM–CI blends. The Tg values of CI tackifier filled EPDM rubber vulcanizates are given in Table 17. Figure 46 shows the dynamic storage modulus curves of CI-EPDM blends. The Gn0 , Me and values of the neat EPDM rubber and EPDM /CI resin blends are reported in

156 Progress in Adhesion and Adhesives, Volume 4

1.6 1.4

Tan δ

Tan δ (arbitrary unit)

1.2 1.0 0.8 0.6 0.4 0.2

0.20

0.0

32 phr

–40 Temperature (ºc)

–20 0.15 Tan δ

–60

16 phr

0.10

84 ºC

67 ºC 36 ºC

8 phr 0.05

2 phr

0.00 20

0 phr –100

–50

50

0

100

40

60

80

100

Temperature (ºC)

150

Temperature (ºC)

Figure  4.45 Tan vs. temperature plots of CI tackifier modified EPDM vulcanizates. Figure reproduced with permission from Ref. [19]. Copyright 2010 Elsevier.

3.5

Log storage modulus (MPa)

3.0

0 phr 2 phr 4 phr

2.5

8 phr 16 phr 24 phr 32 phr

2.0 1.5 1.0 0.5 0.0 –50

0 Temperature (ºC)

50

100

Figure 4.46 Log storage modulus vs. temperature plots of CI blended EPDM vulcanizates. Figure reproduced with permission from Ref. [19]. Copyright 2010 Elsevier.

Adhesion between Compounded Elastomers: A Critical Review 157

Table 17. It was observed that the Me value decreased from 736 to 475 for the EPDM/CI resin blends. This means that the CI resin tackifier significantly increases the entanglement of the base polymer, and the CI resin tackifier essentially acts as a filler (no dilution) in the plateau region. The network density ( ) value increases with the increase in the concentration of the CI resin tackifier. This further confirms the reinforcing effect shown by the CI resin tackifier in the plateau zone. It was pointed out that the observed increase of storage plateau modulus in EPDM rubber- CI resin blends reduced the interfacial contact compliance and hence the adhesion strength between the two joining rubber layers. From viscoelastic property measurements, it was found that the HC resin tackifier reduces the plateau modulus, increases the entanglement molecular weight and reduces the network density of the EPDM rubber. On the other hand, it was found that the CI resin tackifier increases the plateau modulus, reduces the entanglement molecular weight and increases the network density of the EPDM rubber. This clearly confirms a better dilution effect of HC resin tackifier in EPDM rubber in comparison to the CI resin tackifier. Therefore, the authors suggested that the degree of diffusion of elastomer chains from the HC resin tackifier filled EPDM rubber into the fully vulcanized EPDM rubber will be significantly higher in comparison to the degree of diffusion of elastomer chains from the CI resin tackifier filled EPDM rubber into the fully vulcanized EPDM rubber. Therefore, it was anticipated that the HC resin filled EPDM system should show higher adhesion strength in comparison to the CI resin filled EPDM system. There will be very negligible or no diffusion

(a)

(b)

(c)

(d)

Figure 4.47 SEM micrographs of (a) EPHC8, (b) EPHC24, (c) EPCI8 and (d) EPCI24. Figure reproduced with permission from Ref. [19]. Copyright 2010 Elsevier.

158 Progress in Adhesion and Adhesives, Volume 4

of elastomer chains from the fully vulcanized EPDM rubber into the tackifier filled EPDM rubber due to the absence of mobile elastomer chains in the fully vulcanized EPDM rubber. The SEM images of representative HC filled and CI filled EPDM samples are compared at the same magnification in Figure 47. The concentration of resin was same in these samples. The white phases were most likely the dispersed resins in the EPDM matrix. It was observed that HC was more homogeneously dispersed (Figure 47 (a) and (b)) than CI (Figure 47 (c) and (d)). In the latter, these white domains were comparatively of much larger size in the range of 4.27–22.58 μm. This clearly indicated coexistence of a second resin phase in these blends, as predicted by dynamic mechanical analysis (DMA). Further confirmation was obtained from the AFM phase images of these samples show in Figure 48 (a–d). The white phases were confirmed as resins, since these are the harder domains at room temperature. The HC containing samples were almost co-continuous (Figure  48 (a) and (b)) showing excellent correlation with their compatibilities (single Tg). CI, on the other hand, clearly exhibited incompatible, phase-separated morphologies, confirming existence of two different Tg’s (Figure 48 (c–d)). Figure 49 exhibits the results of peel strength vs. different tackifier contents in EPDM assembly where the uncured part was loaded with HC or CI resins at various concentrations. The peel strength increased slowly up to 8 wt% HC content, beyond which significant increase was noticed up to 24 wt%. Finally, it reached a constant value at higher

0

2.5

5.0

7.5

10.0

10.0

7.5

7.5

5.0

5.0

2.5

2.5

0 10.0 μm 0

(a)

0 (c)

2.5

5.0

7.5

(b)

2.5

5.0

7.5

0 10.0 μm

10.0

10.0

7.5

7.5

5.0

5.0

2.5

2.5

0 10.0 μm 0

2.5

5.0

7.5

0 10.0 μm

(d)

Figure 4.48 AFM micrographs of (a) EPHC8, (b) EPHC24, (c) EPCI8 and (d) EPCI24. Figure reproduced with permission from Ref. [19]. Copyright 2010 Elsevier.

Adhesion between Compounded Elastomers: A Critical Review 159 1.0 0.9

Peel strength (N/mm)

0.8 0.7

HC resin CI resin

0.6 0.5 0.4 0.3 0.2 0.1 0

5

10

15

20

25

30

35

40

Concentration of resin (wt%)

Figure 4.49 Plots of peel strength against concentration of the resin tackifier. Figure reproduced with permission from Ref. [19]. Copyright 2010 Elsevier.

concentration. On the other hand, CI modified systems also showed an increase in peel strength up to 8 wt% of resin concentration, but thereafter the strength reduces and levels off at 32 wt% resin content. A maximum of 138% improvement in peel strength was observed for the HC containing system at 24 wt% concentration of tackifier, while only 27% improvement was found for the CI filled system. Here, single side migration from the partially vulcanized phase was anticipated to improve the peel strength which was certainly more effective in the case of HC containing EPDM system than in the CI filled EPDM system, as predicted from the viscoelastic properties. Basak et al. [20] investigated the effect of unmodified nanoclay, Cloisite Na+ on the peel strength behavior between vulcanized EPDM rubber and uncured EPDM rubber joints. The samples for adhesion studies were prepared exactly in the same way except that nanoclay was used in place of the resins. The nanoclay filled unvulcanized EPDM rubber was pre-shaped (1.5  mm thick) for 2 min at 100 °C under 5 MPa pressure in a hydraulic press. The vulcanized EPDM sheet of 1.5 mm thickness was placed into the 3 mm mold cavity along with an aluminium foil of 20 mm width on its upper portion. The pre-shaped EPDM rubber was kept over the fully vulcanized EPDM rubber sample. The same procedure was followed for vulcanization and adhesion testing. AFM phase images of EP and EPNA4 are displayed in Figure 50 (a)-(b). The existence of clay particles on the outermost surface of EPDM vulcanizate was clearly seen in Figure 50 (b), whereas no such behavior was noticed for EPDM vulcanizate devoid of nanoclay (Figure  50 (a)). Representative phase images in Figure  50 (a–b) showed that surface

160 Progress in Adhesion and Adhesives, Volume 4

Cloisite Na+

2 (a)

4

6

2 8

μm

(b)

4

6

8

μm

Figure 4.50 Phase surface morphologies of (a) EP and (b) EPNA4, where EP represents neat EPDM rubber and EPNA4 represents EPDM rubber with 4 phr of Cloisite NA nanoclay. Figure reproduced with permission from Ref. [20]. Copyright 2011 Elsevier.

Figure  4.51 Transmission electron micrograph of EPgNA4. Figure adapted with permission from Ref. [20]. Copyright 2011 Elsevier.

morphology was changed in presence of nanoclay. The roughness value, Ra was found to be 11.70 and 13.05 nm for EP and EPNA4, respectively. A representative transmission electron microscopy (TEM) micrograph at 4 parts of clay loading is shown in Figure 51. The dark lines are the intersectional view of the dispersed clay layers, whereas the off- white phases are the EPDM rubber matrix. An intercalatedexfoliated morphology was observed.

Adhesion between Compounded Elastomers: A Critical Review 161 1.0

12 EP/EPNA

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 –2

EPp/EP EPp/EPNA4 EPp/EPNA8

10

0.8 Peel force (kNx10–3)

Peel strength (kN/m)

0.9

8 6 4 2 0

0

2 4 6 8 10 Concentration of clay (wt %)

12

0

20

40

60

80

100

Distance (mx 10–3)

Figure 4.52 (a) Plots of peel strength against nanoclay concentration and (b) peel force vs. distance curves of unfilled and filled EPDM vulcanizates, where EPp/EP represents EPDM precured rubber vulcanizate/ gum EPDM rubber, EPp/EPNA4 represents EPDM precured rubber vulcanizate/ gum EPDM rubber with 4 phr of Cloisite NA nanoclay, and EPp/EPNA8 represents EPDM precured rubber vulcanizate/ gum EPDM rubber with 8 phr of Cloisite NA nanoclay. Figure reproduced with permission from Ref. [20]. Copyright 2011 Elsevier.

The peel strength value increased with increasing nanoclay amount up to 4phr loading beyond which it decreased (Figure  52). The peel strength of the unmodified EPDM vulcanizate was 0.37 kN/m. A maximum of 51% improvement in peel strength was found after incorporation of 4 parts of Cloisite Na+. On further increase of clay content, the peel strength decreased due to agglomeration. From all these results, it was inferred that the addition of clay up to a certain extent can improve adhesion strength of the rubber joints. Choi et al. [21] carried out a similar study much later. The adhesion between thermally aged and unaged vulcanized and unvulcanized rubber joints was investigated. The BR and NR compounds were made of rubber (100.0 phr), carbon black (CB) (50.0 phr), zinc oxide (4.0 phr), stearic acid (2.0 phr), polymerized 2,2,4-trimethyl-1,2-dihydroquinoline (TMDQ, 2.0 phr), wax (2.0 phr), N-tert-butyl-2-benzothiazole sulfenamide (TBBS, 1.5 phr), and sulfur (1.5 phr). The compounds had the same formulation except the kind of rubber (BR or NR). Mixing was performed in a two-roll mixing mill at a roll speed of 18 rpm. The master batch (MB) compounds were prepared as follows: (1) The rubber was loaded into a two-roll mill and preheated for 4.0 min. (2) CB was compounded into the rubber for 10.0 min. (3) ZnO, stearic acid, N-(1,3-dimethylbutyl)-N’-phenyl-p-phenylenediamine (HPPD), and wax were mixed for 6.0 min. The final mix (FM) compounds were prepared by mixing the TBBS and sulfur with the MB compounds for 5.0 min. Dimensions of the compounded sheet were 140 mm 140 mm with 2 mm thickness. Four adhesion specimens of fully vulcanized BR/unvulcanized BR, fully vulcanized NR/ unvulcanized NR, fully vulcanized BR/ unvulcanized NR, and fully vulcanized NR/ unvulcanized BR joints were prepared. The fully vulcanized sheet was prepared by curing at

162 Progress in Adhesion and Adhesives, Volume 4

160 °C for 8.0 and 6.0 min for the BR and NR compounds, respectively, in a press mold (140 mm 140 mm with 2 mm thickness). The peel test specimens were prepared as follows: (1) Papers were inserted between the two rubber sheets which were then attached. (2) The combined rubber sheet was vulcanized at 160 °C for 6.0 min for the fully vulcanized NR/unvulcanized NR specimen and for 8.0 min. at 160 °C for the other specimens in a press mold (4 mm 140 mm 140 mm). (3) The inserted papers were removed. The peel test was performed at 50 mm/min. Thermal aging was performed at 90 °C for 3 days in a convection oven. Organic materials in each sheet of the adhesion specimen before and after thermal aging were extracted with tetrahydrofuran (THF) (about 60.0 mg sample/1.0 mL THF) at 70 °C for 5 h. Analysis of the extracted organic materials was performed with gas chromatography / mass spectrometry (GC/MS). From the GC/MS total ion chromatogram (TIC), chromatograms of the extracted organic materials from the individual rubber layers of the assembled specimen after thermal aging showed that the relative amounts of the organic materials were notably reduced by the thermal aging at 90 °C for 3 days. Decrements of the principal organic materials such as MBT, sulfur, stearic acid, and TMDQ dimer are summarized in Table 18. Decrements of the antidegradant TMDQ dimer in the BR/BR specimen are much larger than those in the NR/NR specimen. This was attributed to the interactions of TMDQ with NR in comparison with BR. For the fully vulcanized BR/unvulcanized NR and fully vulcanized NR/ unvulcanized BR specimens, the TMDQ decrements in the BR part are lower than those in the NR (Table 18). Especially for the fully vulcanized BR/ unvulcanized NR specimens the TMDQ decrement in the BR sheet is much lower than that in the NR over three times. This implies that some TMDQ dimer molecules migrate from the NR sheet to the BR through the interface during the thermal aging. Decrements of stearic acid in the NR/NR specimen are larger than those in the BR/BR one and those in the unvulcanized sheets are larger than those in the fully vulcanized ones, as shown in Table 18. This was explained by participation in the chemical reactions of stearic acid as well as evaporation because the cure rate of NR is faster than of BR and stearic acid molecules are consumed by participating in the crosslinking reactions. For the fully vulcanized BR/unvulcanized NR and fully vulcanized NR/ unvulcanized BR specimens, the stearic acid decrements in the uncured sheets are larger than those in the fully vulcanized ones. Adhesion strength of the specimens before and after thermal aging was measured and the changes in the adhesion strength by thermal aging are summarized in Table 19. The adhesion strength except for the fully vulcanized BR/unvulcanized NR specimen was reduced by the thermal aging. Decrement of the adhesion strength of the fully vulcanized NR/unvulcanized NR specimen by the thermal aging is much smaller than that of the fully vulcanized BR/unvulcanized BR and the decreased value of the adhesion strength of the fully vulcanized NR/unvulcanized BR specimen lies between those of the fully vulcanized NR/unvulcanized NR and the fully vulcanized BR/unvulcanized BR specimens. Enhancement of the adhesion strength of the fully vulcanized BR/unvulcanized NR specimen by the thermal aging was explained by the crosslink density changes. For the fully vulcanized BR/unvulcanized NR

59.0

36.5

25.4

Stearic acid

TMDQ dimer

21.2

91.0

92.5

44.8

36.1

Unvulcanized BR sheet

Sulphur

Fully vulcanized BR sheet

MBT

Organic chemical

15.9

53.1

–

71.1

Fully vulcanized NR sheet

12.6

69.3

–

39.1

Unvulcanized NR sheet

10.2

8.0

–

28.4

Fully vulcanized BR sheet

31.8

50.6

–

86.9

Unvulcanized NR sheet

25.5

35.5

–

77.3

Fully vulcanized NR sheet

21.5

38.4

83.1

60.2

Unvulcanized BR sheet

Table 4.18 Decrements (%) of major organic materials in the specimens consisting of the fully vulcanized BR/ unvulcanized BR sheets, fully vulcanized NR/unvulcanized NR sheets, fully vulcanized BR/unvulcanized NR sheets, and fully vulcanized NR/unvulcanized BR sheets after thermal aging at 90 °C for 3 days [21].

Adhesion between Compounded Elastomers: A Critical Review 163

164 Progress in Adhesion and Adhesives, Volume 4

Table 4.19 Variation in adhesion strength (%) after thermal aging at 90 °C for 3 days [21].

Types of adhesion samples

Change in adhesion strength

Change in crosslink density

Fully vulcanized BR/Unvulcanized BR

58.2

+12.7/+17.2

Fully vulcanized NR/ Unvulcanized NR

18.6

+11.8/+33.1

Fully vulcanized BR/ Unvulcanized NR

+36.1

+1.9/+21.4

Fully vulcanized NR/ Unvulcanized BR

38.8

+17.9/+22.3

specimen, the crosslink density of the fully vulcanized BR sheet is only slightly increased by the thermal aging, whereas that of the unvulcanized NR is notably enhanced as listed in Table 19. This implies that the differences in the hardness and modulus of the two sheets are reduced by the thermal aging and crosslink density in the interfacial region will be enhanced by migration of organic materials. For the fully vulcanized BR/unvulcanized NR specimen, decrements of the principal organic materials such as MBT, stearic acid, and TMDQs in the NR sheet are much larger than those in the BR one. This implies that lots of the organic materials will migrate from the NR sheet to BR one through the adhesion interface and some of them will participate in crosslinking reactions in the interfacial region. It was shown that the crosslink densities of the unvulcanized sheets increased more than those of the fully vulcanized ones as shown in Table 19 since the curatives and their residues are more retained in the unvulcanized sheets than in the fully vulcanized ones. Crosslink density changes of the NR sheets are also relatively larger than those of the BR ones due to the faster cure characteristics of NR. Ratios of the increased crosslink densities of the unvulcanized and fully vulcanized sheets are 1.35, 2.81, 11.26, and 1.25 for the fully vulcanized BR/ unvulcanized BR, fully vulcanized NR/ unvulcanized NR, fully vulcanized BR/unvulcanized NR, and fully vulcanized NR/unvulcanized BR, respectively. This implies that the increased crosslink density ratios are related to the adhesion strength changes after the thermal aging.

4.3.2

Adhesion between Vulcanized Rubber and Partially Vulcanized Rubber (Filled with Crosslinking Agents)

Gent and Lai [14] investigated the adhesion for vulcanized rubber to partially vulcanized rubber joints having different roughnesses. Three elastomers were used: an SBR; a 50/50 blend of PB/SBR filled with CB; and NR filled with CB. Three examples are given below. An unvulcanized SBR sheet (containing DCP) was partially crosslinked to a given degree by heating for 20 min at 150 °C. A partially vulcanized SBR sheet was brought into contact with a fully-vulcanized SBR sheet and interlinked for further 100 min at 150 °C to complete the crosslinking of partially vulcanized sheet. Values of peel strength Ga are plotted in Figure 53 against test temperature for an SBR sheet interlinked to a fully vulcanized SBR sheet by heating in contact for 100 min at 150 °C. A higher bond strength was obtained when the fully-crosslinked sheet had a rough surface, in accord with an increase in bonded area.

Adhesion between Compounded Elastomers: A Critical Review 165 5

Log Ga (J/m2)

4

3

2

1

0 –50

0

50

100

150

Temperature (ºC)

Figure 4.53 Strength of autohesion vs. temperature of an SBR sheet interlinked to a fully cured SBR sheet by heating in contact for 100 min at 150 °C premolded against Mylar, ; against rough steel, ; SBR sheet crosslinked for 20 min before being interlinked to a fully-crosslinked SBR sheet, both molded against Mylar, . Figure adapted from Ref. [14], with permission from Rubber Chemistry and Technology. Copyright (1995), Rubber Division, American Chemical Society Inc.

However, when one sheet was lightly crosslinked (for 20 min) before being crosslinked in contact with a fully-crosslinked sheet, the samples prepared with rough mold surfaces spontaneously debonded, whereas those prepared with smooth mold surfaces still showed appreciable adhesion strength, Figure  53. The authors suggested that the elastic stresses responsible for quashing the surface asperities were apparently large enough in this case to break the interlinking bonds. A layer of unvulcanized NR sheet (containing sulfur) was partially crosslinked to a given degree by heating in a press. A partially vulcanized NR sheet was brought into contact with a fully vulcanized NR sheet and interlinked for further 30 min at 140 °C to complete the crosslinking of partially vulcanized NR. Values of peel strength Ga are plotted in Figure 54 against test temperature for NR sheet interlinked to a fully vulcanized NR sheet by heating in contact for 30 min at 140 °C. From Figure 54, it is observed that for a sheet premolded against a rough surface, adhesion at moderate temperatures was greatly enhanced in comparison to samples premolded against smooth surfaces. The much higher adhesion strength obtained with a rough surface is tentatively attributed to a greater degree of strain-induced crystallization during peeling. At high temperatures, when crystallization was reduced if not prevented altogether, the adhesion strength for a rough surface fell to the expected level, about 3 . As a third example, a layer of unvulcanized PB/SBR sheet (filled with CB and containing sulfur) was partially crosslinked to a given degree by heating for 27 min at 150 °C. A filled partially vulcanized BR/SBR sheet was brought into contact with a CB filled fully-vulcanized

166 Progress in Adhesion and Adhesives, Volume 4 5

Log Ga (J/m2)

4

3

2

1

0 –50

0

50

100

150

Temperature (ºC)

Figure 4.54 Autohesion vs. temperature of NR sheet interlinked to a fully cured sheet by heating for 30 min at 140 °C (sulfur cure). Symbols: both surfaces rough steel, ; both surfaces smooth steel, ; both surfaces Mylar, . Figure adapted from Ref. [14], with permission from Rubber Chemistry and Technology. Copyright (1995), Rubber Division, American Chemical Society Inc.

PB/SBR sheet and interlinked for further 50 min at 150 °C to complete the crosslinking of partially vulcanized CB filled PB/SBR sheet. Values of peel strength Ga are plotted in  Figure  55 against test temperature for PB/SBR sheet interlinked to a fully vulcanized BR/SBR sheet by heating in contact for 50 min at 150 °C. The samples premolded against a smooth Mylar surface gave the highest adhesion and the samples premolded against a rough steel surface gave the lowest values. This was ascribed to the stresses generated in forcing the rough surfaces of relatively stiff materials into contact.

4.4

Adhesion Between Vulcanized Rubber and Vulcanized Rubber

This section will discuss the mechanisms and factors involved in the adhesion between vulcanized rubber and vulcanized rubber [22–40]. In this case, the mobility of rubber chains from both sides is restricted. Consequently, achieving good adhesion strength between vulcanized rubber layers is very challenging. The bonding between two vulcanized rubber layers is achieved with the help of special bonding agents that stitch both vulcanized rubber surfaces and improve the bond strength. There are many practical situations which demand adhesion between two vulcanized rubber layers. The bonding of vulcanized rubber to vulcanized rubber is important in the fabrication of large structures such as hovercraft loop and finger assemblies (skirt components) and inflatable objects. It is also important in the retreading process of tires, where a vulcanized tread is bonded to the buffed casing of a tire. The bond must possess adequate strength and must be capable of withstanding harsh

Adhesion between Compounded Elastomers: A Critical Review 167 5

Log Ga (J/m2)

4

3

2

1

0 –50

0

50

100

150

Temperature (ºC)

Figure  4.55 Strength of autohesion vs. temperature for BR/SBR samples precured for times t1 at 150 °C and fully cured in contact; t1 27 min for one sheet and 50 min for the other, premold surfaces: Mylar film, ; rough steel, . Figure adapted from Ref. [14], with permission from Rubber Chemistry and Technology. Copyright (1995), Rubber Division, American Chemical Society Inc.

Bonding agent/compound solution at the interface

Vulcanized rubber layer Vulcanized rubber

Vulcanized rubber

Curatives

Vulcanized rubber layer Very negligible amount of inter-diffusions at the interface due to bonding agent/compound solution

Crosslink

Figure 4.56 Schematic for interlinking of vulcanized rubber to vulcanized rubber.

environmental conditions such as extended periods of water immersion, severe tire running conditions, etc. The mechanism of adhesion between two vulcanized rubber layers is different from that of unvulcanized or partially vulcanized rubbers. The contribution from interdiffusion of rubber chains across the interface will be less significant. However, the application of the compound solution and the compound strip may provide some interdiffusion. The contribution to adhesion may arise principally from factors of surface energy and rubber hysteresis (the percentage energy loss for cycle of deformation). Bonding arises from several kinds of interactions that may be physical/or chemical in nature. Schematically outlined in

168 Progress in Adhesion and Adhesives, Volume 4

Figure 56 is the mechanism involved in the adhesion between two vulcanized rubber layers in the presence of a bonding agent at the interface. In the literature, various authors have studied the adhesion between two symmetric or asymmetric vulcanized rubber layers. This section will report on the observations from various research papers related to adhesion between two vulcanized rubber layers. In addition, the mechanisms and factors involved in the adhesion between two vulcanized rubber layers have been explained in detail. Oldfield and Symes [22] studied the adhesion between various vulcanized rubber layers such as NR, polybutadiene rubber (BR), SBR, isobutylene-isoprene rubber (IIR) and NBR (unmodified and TCI/ethyl acetate (EA)-modified) using the T-peel test. A polyamidecured epoxy adhesive was used in this study. The authors have shown that the use of an organic chlorinating agent, TCI dissolved in EA, to treat vulcanized elastomer surfaces produced strong and durable bonds with a polyamide cured epoxy adhesive. NR, BR, SBR, IIR and bromo-butyl rubber (BBR) rubbers were used. Compounding was performed in a small internal mixer and curing agents were added on a laboratory two-roll mixing mill. The curing characteristics of each formulation were determined on a Rheometer and the test sheets were moulded in a single cavity mould in an electrically heated press at 160 °C for the time taken to reach 90 percent of the maximum torque in the Rheometer. Surface treatment of the vulcanized elastomers was performed using the following methods. (i) Solvent wipe: The vulcanized elastomer surface was wiped with a tissue paper moistened in toluene and then allowed to dry for at least 24 h before bonding. (ii) Abrasion: The elastomer surface was mechanically abraded and the debris was subsequently removed from the elastomer surface by wiping with a toluene moistened tissue paper. (iii) Aqueous chlorination: The elastomer was immersed for 5 min in a solution consisting of concentrated hydrochloric acid (2 ml), sodium hypochlorite solution (4 ml, 15% w/v) and water (100 ml). The elastomer was then rinsed well in distilled water and finally allowed to dry. (iv) Cyclisation: The vulcanizate was immersed in concentrated sulphuric acid for one to two minutes, then rinsed in distilled water and allowed to dry. The surface was flexed before bonding to induce fine cracks in the hardened surface layer. (v) Organic chlorination: The vulcanizate surface was wiped successively with clean tissue paper moistened in EA, TCI in EA and finally acetone. The surface was allowed to dry for at least 4 h before bonding. Adhesive joints for peel test were prepared as follows. A bead of adhesive, mixed from epoxy resin and polyamide hardener in the ratio from 1 : 0.5 to 1 : 2 parts by weight, was spread on either side from the centre of the treated surface of an elastomer strip (100 mm 25 mm 6.35 mm). A second similar strip was placed on top with a masking tape at one end between the surfaces to leave a non-bonded area (25 mm 25 mm). A rectangular steel bar weighing about 500 g was placed on each adhesive joint which was then allowed to stand at 20 °C for at least 96 h before testing. Bond

Adhesion between Compounded Elastomers: A Critical Review 169

thickness was typically about 0.125 mm and was assumed to be the same for all specimens. Specimens immersed in seawater were stored in screw capped glass jars in an oven at 60 °C. The bonded specimens were trimmed to 19 mm in width with a parallel bladed cutter and the mean force required to peel the two adherends at an angle of 180° was measured on a tensile testing machine using a crosshead speed of 50 mm/min.

Table 4.20 Peel strength values obtained from various vulcanized rubber to vulcanized rubber joints [22]. Peel strength, kN/me

Elastomera type and formulation no.

Natural rubber (1)

Bromobutyl rubber (2)

Nitrile rubber (3)

Styrenebutadiene rubber (4) a

Surfaceb treatment

Original

A B C D E A B C D E A B C D E A B C D

0.1c 1c 10d 1c 18d 1c 1c 3c 0.1c 20d 8.0d 5c 8d 18d 21d 0.2c 1c 12d 12d

E

11d

After immersion in seawater at 60° C for 1 month

After immersion in seawater at 60° C for 6 month

After immersion in seawater at 60° C for 12 month

10d

14d

12d

12d

16d

16d

12d 5c

15d 3c

11d 4c

10d 6d 14d

14d 3c 16d

15d 5c 11d

8d 10d

8d 8d

6d 6d

9d

6d

6d

Formulation number. A solvent-wipe, B abrasion, C aqueous chlorination, D cyclisation, E organic chlorination, 3.0% TCI. c Rubber/adhesive failure. d Cohesive failure in rubber. e Adhesive used was epoxy resin: polyamide hardener 1:1 w/w. Average of 3 specimens in each case. b

170 Progress in Adhesion and Adhesives, Volume 4

The peel strengths of the specimens from four different elastomers with five different surface treatments that were bonded with an adhesive consisting of epoxy resin and polyamide hardener in the ratio of 1 : 1 by weight are presented in Table 20. The initial peel strengths measured after allowing the adhesive to cure for at least 96  h at 20 °C show that the performance of a surface treatment depends on the elastomer chosen. For NR, only the chlorination treatment procedures (aqueous and organic) resulted in good bonding. These results were maintained during immersion of the specimens in seawater at 60 °C for up to 12 months. Throughout the trial, the organic chlorine donor (TCI) gave higher peel strengths than the aqueous chlorination procedure. Bromobutyl rubber was considered to be difficult to bond in the vulcanized state. Bromobutyl rubber showed low peel strengths for all surface treatment procedures except TCI. In addition, the TCI treatment resulted in bonds between the two vulcanized bromobutyl rubber layers which produced failure in the elastomer layer (both initially and after up to 12 months in seawater at 60 °C). SBR proved to be easier to bond than NR with both chlorination and cyclisation methods giving bonds of similar strength. The relative performance of the three systems studied did not show much change after12-month immersion in sea water. Nitrile rubber proved to be the easiest rubber to bond in the vulcanized state. In the nitrile rubber joints, the highest initial peel strength was shown by the TCI treatment and this was maintained for up to 6-month immersion in sea water. On the other hand, after 12 months the aqueous chlorination treatment showed stronger effect. Martin-Martinez et al. [23] studied the adhesion between two sulfur vulcanized SBR rubber layers (unmodified, TCI/EA-modified, ultrasonic cleaning-modified, and surface roughening-modified) using the T-peel test. Three different surface modifications were performed on the SBR rubber surface. 1. Ultrasonic cleaning: The samples were immersed in a bidistilled water ultrasonic bath for 15 min to remove their surface contaminants. 2. Roughening: Mechanical roughening of the SBR rubber surface was carried out using a sandpaper, until 0.5 mm of surface was removed. 3. Chlorination: Different amounts (1–9 wt%) of TCI of 99% minimum purity and 98% active chlorine were dissolved in MEK. These solutions were applied over previously ultrasonic-cleaned SBR samples. In some specific tests, acetone and EA were also used. Polyurethane (PU) based adhesives of different viscosities were used to join the vulcanized SBR rubber layers. The adhesive solutions were prepared by mixing 15 wt% of solid PU polymer with MEK; the mixture was stirred for 2 h at room temperature in a laboratory mixer. 180 ° peel tests were carried out on PU adhesive joints between two SBR strip test pieces. A layer of PU adhesive was applied to each of the ultrasonic cleaned test strip pieces. Peel strength values were low and increased with the adhesive viscosity above 2 Pa.s. This enhancement was explained by an increase in molecular weight of the PU adhesive due to larger polymer chains. When the amount of adhesive was increased by applying two layers to each SBR

Adhesion between Compounded Elastomers: A Critical Review 171

strip test piece, there was some increase in the adhesion strength. Therefore, the increased amount of adhesive slightly increased the adhesion strengths with SBR. On the other hand, this enhancement was not significantly seen with the increase in viscosity of the adhesive. The peel strength values obtained with roughened SBR surfaces were greater than the peel strength values obtained with ultrasonic-cleaned ones. In addition, this difference was more noticeable for the lowest viscosity. It was observed that the peel strength did not change much with the adhesive viscosity, nor with the amount of adhesive applied to the roughened SBR surface. Hence, the authors concluded that the original SBR surface contained some products that decreased the SBR adhesion to PU adhesive. Sanchez-Adsuar et al. [24] studied the adhesion between two vulcanized SBR rubber layers (unmodified and TCI/2-butanone modified) using the T-peel test. Five onecomponent PU (polycaprolactone polyurethane) adhesives (C1, C2, C3, C4 and C5) based on -polycaprolactone were used. The PU adhesives (C1, C2, C3, C4 and C5) prepared have different configurations i.e., the relative fraction and distribution of hard to soft segments: (i) PU with a relatively low hard/ soft segment ratio (C1, C2, and C4) and (ii) PU with a relatively high hard/ soft segment ratio (C3 and C5). T-peel tests were carried out between identically roughened rubber surfaces. After roughening, two sets of samples were prepared. In the first set of samples, the PU adhesive was directly applied to the roughened rubber surface. In the second set of samples, chlorination was performed on the roughened rubber surfaces and finally PU adhesive was applied. A linear relationship was observed between the T-peel strength (of unchlorinated and chlorinated rubber/PU adhesive/ rubber joints) and the crystallization temperature, the softening temperature and the relative intensity of the (110) reflection in the wide angle X-ray diffraction (WAXD) pattern. For unchlorinated rubber, a higher adhesion corresponds to PUs with lower crystallization temperature, smaller softening temperature and a lower degree of crystallinity, i.e., the polymer with a greater hard/soft segment ratio. On the contrary, for chlorinated rubber the variations in the properties are more marked and follow an opposite trend with respect to the unchlorinated sample; thus the greater T-peel strength corresponds to PUs with the greatest crystallization and higher softening temperature and the highest crystallinity, i.e., the polymer with a lower hard/soft segment ratio. The authors concluded that chlorination creates polar groups on the rubber surface which interact with the polar groups of PU, and thus the different trends were found for unchlorinated and chlorinated rubber. Pastor-Blas et al. [25] attempted to prevent the weak boundary layer (WBL) between two sulfur vulcanized surface modified SBR rubber layers (unmodified and TCI/EA-modified and solvent wiping-modified). PU based adhesive was used to join the vulcanized SBR rubber layers. The sulfur-vulcanized SBR rubber was surface modified by solvent wiping using organic solvents and chlorination using TCI. A vulcanized SBR, about 5 mm thick, was prepared using a moulding process (carried out at 150°C for 50 minutes) after open-mill mixing. To prevent and/or eliminate WBL, several surface treatments were applied to this rubber. T-peel strength of chlorinated rubber/polyurethane adhesive joints was always around 6 kN/m for the heat-treated chlorinated rubber, independent of the percentage of chlorination

172 Progress in Adhesion and Adhesives, Volume 4

agent used. Therefore, with respect to the room temperature-chlorinated rubber, smaller T-peel strengths were obtained for heat-treated chlorinated rubber for the amount of TCI lower than 3 wt%, probably because a certain amount of chlorination agent was used to remove the anti-adhesion compounds, giving a reduced global degree of chlorination in the rubber. For TCI percentages higher than 5 wt%, there were, however, higher T-peel strengths for the heat-treated rubber in comparison to the room temperature chlorinated rubber (Figure 57). In addition, the use of higher amounts of chlorinating agent produced a WBL, as a consequence of the excess of TCI as well as of cyanuric acid on the rubber surface, giving a noticeable decrease in T-peel strength. WBL was produced by the presence of anti-adhesion compounds in the rubber formulation (zinc stearate, microcrystalline paraffin wax). The WBL was not effectively removed by solvent wiping, whether followed by washing with an ethanol/water mix or not. Although this treatment allowed a significant removal of zinc stearate, the paraffin wax concentration on the surface was not greatly reduced; thus, poor adhesion of the rubber was obtained. Chlorination with small amounts of EA solutions of trichloro isocyanuric acid (0.5–5 wt% TCI/EA) and/or an extended halogenation treatment increased the adhesion strength and effectively eliminated the zinc stearate from the rubber surface. On the other hand, the WBL was also created by an excess of chlorination agent applied to the rubber surface. The excess of chlorination agent produced lack of adhesion in the rubber because there was significant damage to the rubber surface and a non-rubber surface layer was formed (mainly

10

T-Peel strength (kN/m)

8 6

4

2 0 0

2

4

6

8

10

TCI %

Figure  4.57 T-peel strength obtained for EA wiped rubber: (o) chlorinated, ( ) chlorinated+ 25 wt% ethanol/water mixture and ( ) chlorinated 25 wt% ethanol/water mixture dried in vacuum, where A and R signify adhesional failure and cohesive failure in the rubber respectively, ti and tr mean immersion time and reaction time in air, respectively. M. M. Pastor-Blas, M. S. SanchezAdsuar and J. M. Martin-Martinez, 1995, Weak boundary layers in styrene-butadiene rubber, Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [25].

Adhesion between Compounded Elastomers: A Critical Review 173

due to oxidized chlorinating agent residues and cyanuric acid), which contributed to the formation of WBL. Figure 57 shows T-peel strength of chlorinated rubber/PU adhesive joints as a function of the concentration of chlorinating agent in EA. The halogenation treatment was applied by solvent wiping with ethyl acetate followed by chlorination with TCI/EA solutions. There was a sudden increase of adhesion for a small amount of TCI, which does not vary by increasing the percentage of chlorinating agent, until a sudden decrease in T-peel strength is produced for TCI percentages larger than 7 wt%. This loss of adhesion produced with a high amount of chlorination agent has been attributed to the formation of a WBL on the surface. Pastor-Blas et al. [26] studied the adhesion between two vulcanized SBR rubber layers (unmodified and trichloroisocyanuric acid (TCI)/ ethyl acetate (EA)-modified) using the T-peel test. A polyurethane (PU) based adhesive was used to join the vulcanized SBR rubber layers. The sulfur-vulcanized SBR was surface modified with EA solutions containing different amounts of TCI. The surface treatment to the vulcanized SBR rubber surface was performed as follows. The SBR was immersed in ethyl acetate for 20 seconds, dried in room temperature (RT) air for 30 min and then immersed in the chlorination solution (1–7 wt% TCI solution in EA). The chlorination process was carried out for one hour. The reaction of TCI with SBR rubber surface produced solid TCI crystallites and other residues. Therefore, a post-chlorination treatment was also done. The post-chlorination treatment was carried out by immersing the SBR in an aqueous solution which contained 25 wt% ethanol. The SBR rubber was removed from the aqueous solution and left to dry in an open air for one hour. A thermoplastic, one-component polyester-urethane (PU) adhesive was used to join the two vulcanized SBR rubber layers. The PU adhesive was prepared by dissolving 15 wt% PU in 2-butanone in a laboratory mixer having a Brookfield viscosity of 1.3 Pa.s (23 °C). The adhesive joints were prepared by applying approximately 150 mg of adhesive to each of the identically-treated vulcanized SBR rubber surfaces. After allowing the solvent to evaporate for 30 minutes, the dry adhesive film was melted at 80 °C, placing them into contact immediately under a pressure of 0.8 MPa. The adhesive joints were conditioned for 72 hours at 23°C and 50% relative humidity. The strength of the adhesive joints was determined using a T-peel test with a peel rate of 100 mm/min. Figure 58 presents the T-peel strength of chlorinated SBR/polyurethane adhesive joints. Chlorination of SBR with small amounts of TCI produced a noticeable increase in practical adhesion, producing cohesive failure in the rubber. The increase in the amount of chlorinating agent produced a sudden decrease in T-peel strength and the failure mode changed to interfacial (adhesional). The treatment of SBR with 7 wt% TCI did not give any adhesion to the polyurethane adhesive, which was attributed to the formation of a weak boundary layer. The chlorination of SBR with small amounts of TCI (lower than 2 wt%) produced an enhancement of surface free energy and the formation of C-Cl and C-O groups on the rubber which were responsible for the increment of the T-peel strength. The increase in the amount of chlorinating agent did not modify the surface free energy. There was an increase

174 Progress in Adhesion and Adhesives, Volume 4

T-Peel strength (kN/m)

15

100% C

10

10%A–90%C

5

100% A

100% A 0 0

5

10

TCI (wt%)

Figure 4.58 T-peel strength of chlorinated SBR/polyurethane joints as a function of TCI percentage. Locus of failure: (A and C imply adhesional failure and cohesive failure respectively). M. M. Pastor-Blas, J. M. Martin-Martinez and J. G. Dillard, 1997, Surface characterization of chlorinated synthetic vulcanized styrene-butadiene rubber using contact angle measurements, infra-red spectroscopy and XPS, Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [26].

of COO- groups on the surface of treated SBR, which resulted in poor adhesion to PU adhesive due to the existence of a weak boundary layer. Romero-Sánchez et al. [27] further studied the adhesion between two vulcanized SBR rubber layers (unmodified and TCI/EA-modified, TCI/propyl acetate (PA)-modified, and TCI/butyl acetate (BA)-modified) using the T-peel test. PU based adhesive was used to join the vulcanized SBR rubber layers. The sulfur-vulcanized SBR was surface modified with ethyl acetate (EA), propyl acetate (PA) and butyl acetate (BA) solutions containing different amounts of trichloroisocyanuric acid (TCI). Two synthetic sulphur-vulcanized SBRs (designated as R1 and R2) were selected in this study. R2 rubber contained larger amounts of polybutadiene, paraffin wax and zinc stearate than R1 rubber. R1 and R2 rubber samples were chlorinated with solutions of 0.5 and 2 wt% TCI, using a brush. The effect of the solvent used for chlorination using TCI on the adhesion property of surface chlorinated R2 rubber was evaluated from T-peel tests by making R2 rubber/PU adhesive/R2 rubber joints. The adhesion strength of the joint produced with the as-received R2 rubber exhibited poor adhesion (0.5 kN/m) due to the presence of wax on its surface. Chlorination treatment of R2 rubber consistently increased the T-peel strength above 10 kN/m, irrespective of the solvent used for chlorination. However, the peel strength value of R2/ PU adhesive joint obtained using 0.5 wt% TCI/EA was slightly higher than those obtained with the other solvents. This was attributed to the absence of wax migration when EA is used in the chlorinating solution. Adhesion failure was observed in the treated R2 rubber/ PU adhesive/R2 rubber joints.

Adhesion between Compounded Elastomers: A Critical Review 175

T-peel strength of treated R1 rubber/PU adhesive /R1 rubber joints also showed excellent and durable adhesive joints, irrespective of the nature of the solvent used in the chlorinating solution. Cohesive failure in the adhesive was, however, observed in the treated R1 rubber/PU adhesive /R1 rubber joints (clearly seen by visual inspection). There was a slight increase in T-peel strength value as the evaporation rate of the solvent increased. From the above results, it was concluded that paraffin wax in the rubber formulation was mainly responsible for the reduction of the adhesion strength between vulcanized SBR layers. On the other hand, it was mentioned that zinc stearate in the rubber formulation did not produce detrimental effect on the adhesion between vulcanized SBR layers. In addition, the authors showed that ageing (ageing in an environmental chamber for 72 h at 70 °C and 50% relative humidity) had negative effects on the T-peel strengths of both chlorinated-R2 rubber adhesive joints and chlorinated-R1 rubber adhesive joints which was attributed to the degradation of the PU adhesive during ageing. Romero-Sánchez et al. [28] studied the combined effect of surface roughness and surface modification (unmodified and TCI/MEK-modified) on the adhesion strength between two vulcanized SBR rubber layers by the T-peel test. PU based adhesive was used to join the vulcanized SBR rubber layers. The sulfur vulcanized SBR was surface modified with MEK solution containing different amounts of TCI. To determine the adhesion properties of vulcanized SBR rubber, roughened or unroughened plus chlorinated SBR rubber/PU adhesive joints were prepared. The PU adhesive solution was prepared by dissolving 18 wt% PU pellets in MEK in a laboratory mixer. The Brookfield viscosity of the PU solution was 2700 mPa.s (23 °C). The influence of roughening of SBR rubber prior to chlorination was considered in this study. The roughening of SBR rubber surface was performed using an aluminum oxide abrasive cloth, and about 0.5 mm of rubber was removed. The effects of the chlorination produced on the unroughened SBR rubber surface were compared with those produced by chlorination on the freshly roughened SBR rubber surface. Two concentrations of chlorinating agent were used (0.5 and 2 wt% TCI/MEK). The chlorination solutions were applied using a brush and, in general, the PU adhesive was applied on the treated SBR rubber (roughened and unroughened) 15 h after the chlorination treatment was performed. The polyurethane adhesive solution was always applied 30 min before the joints were produced. Unroughened treated SBR rubber/PU adhesive/unroughened treated SBR rubber joints and roughened treated SBR rubber/PU adhesive/roughened treated SBR rubber joints were prepared by applying the PU adhesive solution (about 0.8 ml was applied on each rubber piece). The solvent of the adhesive was allowed to evaporate for 45 min. The dried adhesive films were melted (re-activation process) at 100 °C under IR irradiation and immediately placed into contact under a pressure of 0.8 MPa for 10 seconds. T-peel test (72 h after joint formation) was carried out at a peel rate of 0.1 m/min. The chlorination with 0.5 wt% TCI/MEK was sufficient to produce a high adhesion. At all concentrations of TCI, the joints produced with roughened chlorinated SBR rubber provided higher peel strength values than those for unroughened chlorinated SBR

176 Progress in Adhesion and Adhesives, Volume 4 20 18

Peel strength (kN/m)

16

Roughened R2

sive joint

rubber/PU adhe

12

int

ive jo

14

/P

bber

R2 ru

hes U ad

10 8 6 4 2 0 0.00

0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 wt% TCI/MEK

Figure 4.59 T-peel strength values of unroughened and roughened SBR rubber chlorinated with 0.5 and 2 wt% TCI/PU adhesive joints. Values obtained 72 h after joint formation. M. D. Romero-Sanchez, M. M. Pastor-Blas and J. M. Martin-Martinez, 2002, Improved peel strength in vulcanized SBR rubber roughened before chlorination with trichloroisocyanuric acid, Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [28].

rubber (Figure 59). The improvement in the adhesion strength of the joints produced with roughened chlorinated SBR rubber was attributed to the higher roughness and higher degree of chlorination of the roughened SBR rubber. Pastor-Blas and Martin-Martinez [29] studied the adhesion between two vulcanized SBR rubber layers (unmodified and solvent modified TCI/EA-modified, and oxygen plasma-modified) using the T-peel test. PU based adhesive was used to join the vulcanized SBR rubber layers. The sulfur-vulcanized SBR rubber was surface modified using either chemical treatment (using TCI) or treatment with oxygen plasma. The authors compared the effect of these two surface modification methods on the adhesion strength. The chlorination was carried out by immersing the vulcanized SBR rubber in EA for 30 s and then in solutions of 1–7 wt% TCI in EA, followed by a post-chlorination treatment with aqueous solutions containing 25 wt% ethanol. Oxygen plasma treatment was carried out in a plasma reactor operating at 13.56 MHz and 50 W for 1 min and 40 min. The PU adhesive solution was prepared by dissolving 15 wt% of poly -caprolactone urethane pellets in 2-butanone in a laboratory mixer (500 rpm for 2.5  h). The adhesive solution obtained had a Brookfield viscosity of 1690 mPa.s (23 °C). Adhesive joints were made using two rubber-strip test pieces which were similarly treated. The PU adhesive solution was applied using a brush (about 0.9 ml of solution was placed on each rubber piece). After allowing the solvent to evaporate for 1 h, the dry adhesive film was melted at 80 °C under IR radiation and the PU-coated rubber pieces were

Adhesion between Compounded Elastomers: A Critical Review 177

placed into contact immediately under a pressure of 0.8 MPa for 10  s. The thickness of the adhesive layer was about 0.5 mm. The adhesive joints were kept at 23 °C and 50% relative humidity for 72 h before undergoing the T-peel test. The strength of the adhesive joints was determined using a T-peel test at a peel rate of 0.1 m/min. The percentage of oxygen on the SBR rubber surface increased with the concentration of TCI. Also, the percentage of oxygen on the SBR rubber surface increased with the extended plasma treatment. Surprisingly, the T-peel strength values were much higher for the 2 wt% TCI chlorinated SBR rubber/PU adhesive joint than for the oxygen plasmatreated SBR rubber/PU adhesive joint (Figure 60). On the other hand, solvent wiping of SBR rubber with EA or MEK did not increase the peel strength values of SBR rubber/PU adhesive joints. Furthermore, when a high percentage of TCI was applied on the SBR rubber (7 wt%), the T-peel strength decreased to 0.2 kN/m and an adhesional failure (visually assessed) was produced. This was a consequence of the formation of a weak boundary layer consisting mainly of unreacted TCI and/or isocyanuric acid, and of mechanical degradation of the rubber. On the contrary, the increase in the length of the oxygen plasma treatment up to 40 min enhanced the adhesion due to surface ablation, which helped to remove surface contaminants and anti-adhesion moieties from the rubber surface. Romero-Sánchez et al. [30] studied the adhesion between two sulfur-vulcanized SBR rubber layers (unmodified, TCI/EA-modified, and TCI/methyl ethyl ketone (MEK)modified) using the T-peel test. PU based adhesive was used to join the vulcanized SBR rubber layers. The sulfur-vulcanized SBR was surface modified with MEK and EA solutions containing different amounts of TCI. The chlorination was produced by brushing the SBR rubber surface with solutions of 2 wt% TCI in MEK (2 wt% TCI/MEK) or EA (2 wt% TCI/

8

7 wt% TCI/EA

2 wt% TCI/EA

O2 (50 W, 10 min)

O2 (50 W, 1 min)

2

EA

4

MEK

6 As-received

T-peel strength (kN/m)

10

0 Surface treatment

Figure 4.60 T-peel strength of solvent-wiped, oxygen plasma-treated, and chlorinated SBR rubber/ PU adhesive joints where O2 represents oxygen plasma treated samples. M. M. Pastor-Blas and J. M. Martin-Martinez, 2002, Different surface modifications produced by oxygen plasma and halogenation treatments on a vulcanized rubber, Journal of Adhesion Science and Technology, reprinted by permission of Taylor & Francis Ltd., Ref. [29].

178 Progress in Adhesion and Adhesives, Volume 4

EA). The concentration of chlorine in these TCI solutions was obtained by adding an acidified potassium iodide solution. The adhesion strengths were obtained from T-peel test on surface-chlorinated SBR rubber/PU adhesive joints. The joint produced with the as-received SBR rubber showed low peel strength due to the poor wettability and to the existence of the hydrocarbon-rich layer on the SBR rubber surface. The chlorinated and the oxidized moieties (produced on the SBR rubber surface by treatment with 2 wt% TCI/MEK) and removal of paraffin wax from the SBR rubber surface led to a noticeable increase in the T-peel strength. Similar peel values were obtained when chlorination was produced with solutions that were used before two weeks of preparation. However, the chlorination with 2 wt% TCI/MEK solution that was used after six weeks of preparation showed a marked decrease in peel strength value. This was ascribed to the less effectiveness of the chlorination of SBR rubber surface. A relationship between the concentration of chlorine in the 2 wt% TCI/MEK solution and the adhesion between two surface treated vulcanized SBR rubber layers was established. Tyczkowski et al. [31] studied the adhesion between two sulfur-vulcanized surface modified SBR rubber layers (unmodified, modified by wet chemical method, and modified by plasma technique) using the T-peel test. PU based adhesive was used to join the vulcanized SBR rubber layers. Two types of synthetic SBR rubbers (SBR-1 and SBR-2) were used in this study. Butadiene-to-styrene ratios were 69–31 wt% and 27–73 wt% for SBR 1 and SBR 2, respectively. Here, the authors have attempted to replace the wet chemical method by a clean plasma technique. The plasma treatment was carried out in two parallel plate reactors, the first system with radio-frequency (rf, 13.56 MHz) and the second one with audio-frequency (af, 20 kHz) glow discharges. As chlorine precursors, chlorine (Cl2), trichloromethane (CHCl3) and tetra-chloromethane (CCl4) were used. To compare the efficacy of plasma treatment and the wet chemical modification procedures, vulcanized SBR rubber surfaces were chemically chlorinated. The chlorination was performed by immersion of the samples for 30 s in the chlorination solution (2 wt% of TCI in propanone). To determine the adhesion strength between two vulcanized SBR rubber layers, T-peel tests were carried out. Adhesive joints were made using the rubber samples which were surface mechanically roughened before the modifying treatments. The one-component solventborne PU adhesive was spread on each adherend and dried for 15 min. T-peel strength measurements were performed at a peel rate of 1.67 10–3 m/s. The samples were plasma treated in an rf reactor using the CHCl3 vapour as a chlorine source. The plasma treatment evidently improved the peel strength compared to the nontreated surface in the case of the SBR-1 elastomer. When the pure CHCl3 vapour was used, the peel strength was higher (by approx. 56%) than that of the samples treated by TCI. The use of oxygen in the reactive mixture deteriorated the adhesion of the treated surfaces. However, the low power supplied to the plasma reactor produced peel strength of a similar value to that measured for the TCI treated samples (plasma CHCl3/O2/10 W vs. TCI treated). It was striking to note that when the discharge power was low, the peel strength of the adhesive joints was better. The authors observed such behaviour for both the plasma treatment by pure CHCl3 and also when the CHCl3/O2 mixture was used.

Adhesion between Compounded Elastomers: A Critical Review 179

On the other hand, in the case of the SBR-2 elastomer, the plasma treatment did not increase the peel strength. In fact, the peel strength of the plasma treated samples was either the same or only a bit higher in comparison with the non-treated samples. However, the peel strength of the plasma treated samples was evidently lower than that obtained after the wet chemical process. The evidently higher peel strength for samples of SBR-2 after the chemical treatment in comparison to the plasma treatment was attributed to the increase in surface energy (formation of higher number of C–Cl groups) which consequently led to an increase in the thermodynamic adhesion [16]. It was also found that the peel strength between two SBR-2 surfaces (TCI treated) was lower than the peel strength between two SBR-1 surfaces (TCI treated). This was attributed to the different chemical constitutions of the copolymers present in SBR-1 and SBR-2. SBR-1 contains 69 wt% of butadiene in the structure, whereas the amount of butadiene in SBR-2 is only 27 wt%. The authors explained that the efficiency of the chlorination process depends on the positions where chlorine atoms are substituted in the copolymer chains. Based on the analogy with polychloroprene rubber it was predicted that the carbon atoms in the 1, 2-addition configuration of the monomer units in the polybutadiene blocks, substituted with chlorine atoms, act as active cross-linking sites. Accordingly, after surface chlorination, SBR-1 having higher concentration of polybutadiene blocks shows a greater peel strength of adhesive joints in comparison to SBR-2 having lower concentration of polybutadiene blocks. The chlorination process results in the anchorage of chlorine atoms to the substrate surface. The anchored chlorine atoms participate in the crosslinking between the substrate surface and the adhesive. Pastor-Sempere et al. [32] studied the adhesion between two vulcanized SBR rubber layers (unmodified and fumaric acid (FA)-modified) using the T-peel test. PU based adhesive was used to join the vulcanized SBR rubber layers. The PU used in this study had a medium thermoplasticity, very high crystallization rate and short open-time. PU adhesive solutions containing 18 wt% of polymer in 2-butanone were prepared in a laboratory mixer, simultaneously adding amounts of FA between 0.5 and 3 wt% (with respect to the PU adhesive). The mixture was stirred for 2 hours at room temperature. Adhesion strength was measured from T-peel tests on roughened vulcanized SBR/PU adhesive/roughened vulcanized SBR joints. The SBR surface was roughened by using a sandpaper (0.5  mm of the external surface was removed). Subsequently, 100 mg of PU adhesive was applied to each rubber surface to be joined, and left to dry for 30 minutes. The dried PU film was quickly heated to 70 °C using infrared radiation in order to facilitate the contact between the PU adhesive applied to the two SBR test pieces. The melted PU adhesive films were placed in contact and a pressure of 3 atm was applied immediately for 10 seconds to achieve a suitable joint. Test samples were kept at 23 2 °C and 50% relative humidity for 72 hours before the T-peel strength was measured in a tensile machine (peel rate: 0.l m/min). T-peel strengths of roughened SBR/PU adhesive/roughened SBR joints are given in Figure 61. The addition of amounts higher than 1 wt% FA to the PU adhesive produces a noticeable increase in T-peel strength. Figure 61 shows the T-peel strengths of roughened SBR/PU adhesive/roughened SBR joints. A slight decrease in T-peel strength is produced due to chain cleavage and disruption of crystallinity in the PU structure. However, the

180 Progress in Adhesion and Adhesives, Volume 4 Roughened SBR 5

T-Peel strength (kN/m)

4

3

2

1

0 0

1

2

3

Wt% Fumaric acid

Figure  4.61 T-peel strength of roughened SBR/PU adhesives with FA/roughened SBR joints. N. Pastor-Sempere, J. C. Fernández-García, A. C. Orgilés-Barceló, M. S. Sánchez-Adsuar and J. M. MartínMartínez, 1996, Improved adhesion properties of polyurethane adhesive containing fumaric acid, Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [32].

improved adhesion between SBR and PU adhesive containing FA is maintained even after 30 days of adhesive preparation. The authors could not clearly explain this improvement because the addition of FA causes a modification of the segmented structure of PU, its mechanical, physical and thermal properties are reduced and there is no increase in the surface free energy of the PU films. Therefore, the authors attributed this effect to the diffusion process being much faster than the hydrolysis reaction of the polyurethane chains. It was assumed that the higher concentration of FA on the surface would help to maintain the surface energy and the improved T-peel strengths even after one month of addition of FA to the PU adhesive. Cepeda-Jiménez et al. [33] studied the adhesion between two sulfur-vulcanized surface modified SBR rubber layers (unmodified and sulfuric acid-modified) using the T-peel test. PU based adhesive was used to join the vulcanized SBR rubber layers. The treatment of SBR rubber surface with sulfuric acid was carried out by the following consecutive steps: (i) The rubber was immersed in concentrated sulfuric acid for a given time (immersion time); (ii) It was then removed from the acid and allowed to continue to react in air for a given time (reaction time in air); (iii) The acid was neutralized using hot distilled water ammonium hydroxide (15 wt% ammonia) exhaustive washing under distilled water, the neutralization was carried out until a pH 7 was obtained in the washed water; and (iv) The treated rubber (a continuous film of liquid remained on

Adhesion between Compounded Elastomers: A Critical Review 181

the rubber surface) was dried under infrared radiation at a moderate temperature (lower than 60 °C) for 30 minutes. To determine the adhesion strength of surface-treated SBR rubber, two identicallytreated specimens were joined using an adhesive solution, based on one-component thermoplastic PU. The adhesive solution was prepared by mixing 18 wt% PU in 2-butanone. A PU adhesive solution with a Brookfield viscosity of 1.7 Pa.s (20 °C) was obtained. The strength of the joints was evaluated using T-peel tests. T-peel strength measurements were carried out for the adhesive joints produced with the strip test pieces which were treated in the same way. Before applying the adhesive, the treated rubber pieces were flexed to develop cracks and facilitate mechanical interlocking with the adhesive. The PU adhesive solution was applied on the treated rubber surface with a brush and the solvent was allowed to evaporate for one hour. The dried PU adhesive film was heated to 100 °C under infrared radiation in order to facilitate interlocking of the chains of the two PU films applied to the two identically surface-treated rubber strips. The strips were then placed in contact and a pressure of 0.8 MPa was immediately applied for 10 s to achieve a suitable joint. The T-peel strength was measured at a peel rate of 0.1 m/min. The adhesive joints were conditioned at 25 °C and 50% relative humidity before undergoing the T-peel test. Once the joints were formed, the T-peel tests were performed after 15 min and 72 h. From Figure  62 (a) it is seen that although the adhesion was enhanced by treatment with H2SO4, an excessive modification of the outermost surface layer was not appropriate since the failure was directed within the damaged surface. Therefore, an immersion time in sulfuric acid shorter than 2 minutes was optimal to produce an adequate performance in the adhesive joints and with negligible mechanical degradation of the rubber. From Figure 62 (b) it is noticed that there was a noticeable increase in immediate T-peel strength up to 1 minute reaction time in air. Reaction times in air higher than 1 minute did not improve adhesion. Reaction time of 1 minute was sufficient to produce an adequate performance. Therefore, the reaction time with sulfuric acid in air influences to a lesser extent the properties and the adhesion of SBR rubber when compared to the effect of other immersion times. Pastor-Blas et al. [34] studied the adhesion between two sulfur-vulcanized surface modified cold SBR rubber layers (unmodified and oxygen plasma-modified) using the T-peel test. T-Peel strength values of the adhesive joints vs. length of plasma treatment are given in Figure  63. All adhesive joints showed adhesional type of failure. The joints between untreated rubber materials produced very poor adhesion because of the existence of a waxrich layer on the rubber surface (constituted by hydrocarbon species). The oxygen plasma treatment produced an increase in joint strength which was maximum for 10 min of treatment; the increase in the length of treatment produced a decrease in adhesion strength which was ascribed to the existence of oxidized sulfur particles on the treated rubber surface and to the excessive surface ablation produced by an extended length of the treatment. Hirahara et al. [35] studied the adhesion of vulcanized fluorinated rubber (FR) and vulcanized nitrile rubber joints using the T-peel test. A new curing agent, the tetrabutylammonium salt of 1,3,5-triazine-2,4,6-trithio (TATT) for FR was developed, for making

182 Progress in Adhesion and Adhesives, Volume 4

10 72 h 15 min

T-Peel strength (kN/m)

8

6

4

2

0 0

2

4

6

(a)

12

8 10 t1 (min)

14

9

16

72 h 15 min

8

T-Peel strength (kN/m)

7 6 5 4 3 2 1 0 0 (b)

1

2

3 t1 (min)

4

5

6

Figure 4.62 T-peel strength values of H2SO4-treated SBR rubber/PU adhesive joints as a function of (a) immersion time in H2SO4 95% solution, ti 1 min.; NH4OH 15% solution and (b) reaction time in air. H2SO4 95% solution, tr 0.5min.; NH4OH 15% solution (where ti represents the immersion time and tr represents the reaction time). , T-peel tests were performed after 72 h once the joint was formed; , T-peel tests were performed after 15 min once the joint was formed. C. M. Cepeda-Jimenez, M. M. Pastor-Blas, J. M. Martin-Martinez and T.P. Ferrandiz-Gomez, 2000, Surface characterization of vulcanized rubber treated with sulfuric acid and its adhesion to polyurethane adhesive, Journal of Adhesion, reprinted by permission of Taylor & Francis Ltd., Ref. [33].

Adhesion between Compounded Elastomers: A Critical Review 183

T-Peel strength (kN/m)

5 4 3 2 1 0 0

10

20

30

40

50

Length of treatment (min)

Figure 4.63 T-Peel strength of oxygen plasma-treated SBR rubber/polyurethane adhesive joints as a function of the length of treatment. Figure reproduced with permission from Ref. [34]. Copyright 1998 John Wiley & Sons Inc.

composite materials with other rubbers. In this study, TATT was found to be an effective curing agent for the direct adhesion between FR and NBR without any adhesive. A mixture of triazine thiolate (TT) (17.7 g) and NaOH (4.1 g) was stirred at 40 °C in water (300 ml). Into this aqueous solution, tetrabutylammonium bromide (TBAB, 31.0 g) in water (100 ml) was poured with stirring at 60 °C over 3 h. The mixture was then cooled to 10 °C and the precipitate was filtered from the mixture. The crude TATT was recrystallized from ethanol solution, with a yield of 92%, melting point of 183 °C. Master batches containing FR or NBR, CB, stearic acid, and metal activators were mixed for 20 min in a Banbury mixer and then blended for 10 min on a two-roll mill to obtain the base compounds. All the other additives, such as curing agents and accelerators, were added to the master batch on a laboratory mill for 10 min at 60 °C to prepare rubber compounds, which were hot-pressed in a 2 mm thick cavity of a metal mold at 160 °C for 30 min to produce the vulcanizates. A sheet of FR compound and a sheet of an NBR compound, each 1.6 mm thick, were placed in a metal mold 3 mm thick with air vents for 30 min at 160°C under a pressure of 15 MPa to obtain the FR-NBR joints. The peel strength of the FR-NBR joints was determined by a T-peel test. In the first case, polyol curing agent was used to generate direct adhesion of FR to NBR. The conventional polyol curing system acts only on FR or NBR compounds, to give respectively, FR and NBR vulcanizates but no interpolymer at the FR-NBR interface. It was shown that the conventional polyol curing system consisting of a bisphenol-type curing agent and an organic phosphate-type accelerator was not capable of reacting with NBR. The joints obtained in this study thus showed no peel strength at all, due to the absence of interfacial bonds. In the second case when TATT is used as the curing agent, the TATT reacts with FR in the presence of Ca(OH)2 and MgO, and also reacts with NBR in the presence of CBS and ZnO. TATT serves as a curing agent for FR, and thiol groups in TT react with NBR at the interface in the presence of CBS, DTDM, and ZnO which ultimately gives an FR-NBR

184 Progress in Adhesion and Adhesives, Volume 4

interpolymer. The high peel strength was derived from the formation of an interpolymer due to TATT curing agent. Job and Joseph [36] studied the adhesion strengths of vulcanized NR-vulcanized NR and vulcanized NR/BR (70/30) and vulcanized NR/BR (70/30) joints using the peel test. NR and reclaimed latex waste based adhesives were used in the study. The bonding characteristic was evaluated by determining the peel strength with a Universal Testing Machine (UTM) at a test speed of 500 mm/min. The thickness of the rubber compound strip was found to have a profound effect on the vulcanized rubber - vulcanized rubber bonding. In the case of NR/NR and NR-BR/ NR-BR blends, the peel strength increased with the increase in the thickness of the strip, reached a maximum and thereafter decreased when the bonding time and bonding temperature were kept constant. The authors concluded that there is an optimum thickness of the compound strip below which it cannot keep the two thick substrates intact. In addition, it was also mentioned that at higher thickness the strip remained undercured as the time of bonding is kept constant resulting in a reduction in bond strength. It was also shown that the temperature of bonding was an important variable in controlling the bond strength. NR based sheets had maximum peel strength at a bonding temperature of 130 °C, whereas NR-BR blend based sheets had a maximum peel strength at a bonding temperature of 140 °C. Petrova et al. [37] studied the adhesion between two sulfur-vulcanized EPDM rubber layers using the T-peel test. Adhesives based on the oligomeric nitrile–butadiene rubber (NBR having ~5 wt % carboxylic groups) and epoxy resin (5 to 15 wt %) were used in this study. The joint strength was determined by the 180 ° peel test at room temperature at a crosshead speed of 10 mm/min. From the test results, it was seen that at low degrees of cure, the peel resistance was insignificant and the failure (along the adhesive composition) was cohesive in character. As the degree of cure of the NBR epoxy resin (adhesive) blend increased, the strength increased and the character of failure became mixed. At the complete conversion of carboxylic groups, i.e., after the formation of a three-dimensional network of crosslinks in the adhesive, the value of peel strength depended on the lamination conditions. If curing is run under the conditions of permanent contact with the substrate, the adhesion strength attains a maximal value showing a mixed mode of failure. Substrates (rubbers) and adhesives with different moisture contents were prepared to understand the effect of moisture on the adhesion strength. It was shown that the water content of rubbers during their scheduled conditioning at a humidity of 65–70% led to the spontaneous redistribution of water between the substrate and the adhesive. This process was accompanied by retardation of the formation of the adhesive network structure and, as a consequence, by a decline in the strength of adhesive joints.

4.5

Summary

Rubber to rubber bonding bears significant importance from scientific and industrial points of view. This fascinating problem has prompted many scientists and technologists to

Adhesion between Compounded Elastomers: A Critical Review 185

explore this subject and come up with a clear understanding and new findings. This paper reviews the literature on adhesion between unvulcanized elastomer / partially vulcanized elastomer to unvulcanized elastomer / partially vulcanized elastomer, unvulcanized elastomer to vulcanized elastomer, and vulcanized elastomer to vulcanized elastomer. It should be pointed out here that the adhesion strength between two rubber substrates under the above different conditions is found to be dependent on the combination of a variety of factors / mechanisms. The following paragraphs summarize the significance and factors governing the rubber to rubber bonding under various conditions. 1. The adhesion between two unvulcanized rubber layers through co-crosslinking is very important in the tire and conveyor belt manufacturing industries. The manufacturing process involves assembling of several layers of unvulcanized rubber (similar or dissimilar) one on top of another and finally co-crosslinking them using heat and pressure in a mold. The contribution from interdiffusion of elastomer chains (interlinking molecules) across the interface from either side is significant in controlling the final adhesion strength. It has been found that the adhesion strength correlates well with the length of the interlinking chains and the density of the interlinking chains. The length of the interlinking chains and the density of the interlinking chains are governed by the type of crosslinking system and crosslinking conditions such as crosslinking temperature and crosslinking time. The adhesion between two partially vulcanized rubber layers through co-crosslinking may not have great importance from industrial point of view. However, it demands attention from scientific point of view. The contribution from interdiffusion of elastomer chains (interlinking molecules) across the interface from either side towards improving the adhesion strength is slightly reduced because of the restriction on the mobility of elastomer chains due to partial crosslinking. However, the adhesion strength between the two partially vulcanized rubber layers is still found to be dependent on the length of the interlinking chains / density of the interlinking chains that are governed by the type of crosslinking system and crosslinking conditions. In some instances, it has also been found that the surface roughness of the partially vulcanized substrates has some influence in improving or deteriorating the bond strength by enhancing or reducing the interfacial contact at the interface. 2. The adhesion between an unvulcanized elastomer and vulcanized elastomer is important in the scenario where any damaged vulcanized rubber layer needs to be repaired. For example, the damaged vulcanized portion of the conveyor belt is repaired by pressing an unvulcanized rubber compound against the damaged vulcanized portion by application of heat and pressure. The crosslinking system in the unvulcanized rubber compound will help in stitching the damaged vulcanized rubber layer. In this case, the mobility of elastomer chains from one side is restricted. The contribution from interdiffusion of elastomer chains (interlinking molecules) across the interface from either side towards improving the adhesion

186 Progress in Adhesion and Adhesives, Volume 4

strength is very low. The adhesion strength between an unvulcanized elastomer and vulcanized elastomer is found to be dependent on various factors such as surface free energy, extent of diffusion of elastomer chains from one side, and surface roughness / texture of the substrates. The manipulation of surface free energy for enhanced adhesion can be achieved by surface modification procedures (modification by chlorinating agents and high energy) applied on the surface of the fully vulcanized rubber substrate. Enhancing the diffusion of elastomer chains from one side (unvulcanized portion) by addition of tackifiers is also very effective in improving the bond strength. It has also been shown that adding special additives such as nanoclay particles in the unvulcanized rubber portion significantly increases the adhesion strength due to the reinforcement action of the nanoclay particles at the interface. 3. The adhesion between two vulcanized elastomer layers is extensively practiced in retreading tire industry. The mechanisms involved in adhesion between two fully vulcanized elastomer layers are completely different from the earlier two cases. In this case, the interdiffusion of elastomer chains across the interface from both sides is completely restricted. However, applying adhesive solution at the interface can provide some interdiffusion. The adhesion strength between two fully vulcanized elastomers layers is found to be dependent on various factors such as free energy between the substrate and the adhesive solution and rubber hysteresis. It has been found that the bonding between the substrates and adhesive arises from several kinds of interactions that can be physical and/or chemical in nature. The manipulation of surface free energy for enhanced adhesion can be achieved by surface modification procedures (modification by chlorinating agents, acids, and high energy) applied to the surfaces of the substrates and by a judicious selection of the type of adhesive (polarity, functional groups, viscosity, adhesive layer thickness and molecular weight) applied on the surfaces of the modified or unmodified substrates. Among the various mechanisms involved in rubber to rubber bonding, the relative importance and the appropriate way in which they should be combined will vary from case to case. It is not advisable to exclude any mechanism without careful consideration, understanding and exploration. Therefore, it is very clear that the adhesion bond in the rubber-to-rubber joints achieves its strength from a combination of a variety of factors and mechanisms.

Acknowledgements One of the authors (AKB) gratefully acknowledges the support received from INAE, New Delhi in the form of Chair Professor and the funding received from MHRD, New Delhi for development of high performance rubber composites using new generation materials for application in tire (UAY-HPN project).

Adhesion between Compounded Elastomers: A Critical Review 187

List of Symbols Symbol T F A k t A0 tan v C1 and C2 N2 t2 R C1 N C1max N1 t1 Vr s

MF V1 Mc N gn

Gn0 1

L fs Q f G0 Gc,0 c w Ga E

Explanation Absolute temperature Applied force Avogadro’s number Boltzmann’s constant Contact time Cross-sectional area of the specimen in the unstrained state Damping factor Degree of crosslinking Density of the rubber Elastic constants Extension ratio Final crosslink density Final crosslink time Gas constant Increase in C1 Increase in density of network Increase in density of network strands Increase in the maximum value of elastic constant, C1 Initial crosslink density Initial crosslink time Interlinking density Linear swelling ratio Mismatch factor Molar volume of the swelling liquid Molecular weight of the network strands Number of molecular network strands per unit volume / density of network Number of network molecular strands per unit volume Numerical factor Plateau modulus Rubber-liquid interaction parameter Rubber strand length Sol fraction Swelling ratio Tensile force Threshold peel strength Threshold work of detachment Volume fraction of rubber in the swollen gel Width of the joint Work of adhesion Young’s modulus

List of Abbreviations AFM

Atomic force microscopy

188 Progress in Adhesion and Adhesives, Volume 4

ATR-FTIR af BIIR BR BA CB CR CI CBS DCP DOTG DMA EB EDX EA EPDM EPR FM FR FTIR FA GC/MS Tg HAF HC IIR IPPD MB Tm MBTS MEK HPPD NR NBR PET PU PA rf RT SEM Si SBR TBBS TBAB TATT THF

Attenuated total reflection Fourier transform infrared spectroscopy Audio frequency Brominated isobutylene-isoprene rubber Butadiene rubber Butyl acetate Carbon black Chloroprene rubber Coumarone indene Cyclohexylbenzothiazyl sulphenamide Dicumyl peroxide Diorthotolyl guanidine Dynamic mechanical analysis Electron beam Energy dispersive X-ray spectroscopy Ethyl acetate Ethylene propylene diene monomer rubber Ethylene-propylene rubber Final mix Fluorinated rubber Fourier transform infrared spectroscopy Fumaric acid Gas chromatograph/ mass spectrometry Glass transition temperature High abrasion furnace black Hydrocarbon Isobutylene-isoprene rubber Isopropyl-N’-phenyl-p-phenylenediamine Master batch Melting temperature Mercaptobenzothiazyl disulphide Methyl ethyl ketone N-(1,3-dimethylbutyl)-N’-phenyl-p-phenylenediamine Natural rubber Nitrile-butadiene rubber Poly(ethylene terephthalate) Polyurethane Propyl acetate Radio frequency Room temperature Scanning electron microscopy Silica Styrene-butadiene rubber Tert-butyl-2-benzothiazole sulfenamide Tetrabutylammonium bromide Tetrabutylammonium salt of 1,3,5-triazine-2,4,6-trithiol Tetrahydrofuran

Adhesion between Compounded Elastomers: A Critical Review 189

TMTD TMTM TIC TEM TT TCI TEA TMPTA TPGDA UTM WBL WAXD XPS ZDEC ZnO TMDQ

Tetramethylthiuram disulphide Tetramethylthiuram monosulfide Total ion chromatogram Transmission electron microscopy Triazine thiolate Trichloroisocyanuric acid Triethanolamine Trimethylol propane triacrylate Tripropylene glycol diacrylate Universal Testing Machine Weak boundary layer Wide angle X-ray diffraction X-ray photoelectron spectroscopy Zinc diethyldithiocarbamate Zinc oxide 2,2,4-trimethyl-1,2-dihydroquinoline

References 1. K. D. Kumar, G. C. Basak and A. K. Bhowmick, Adhesion between unvulcanized elastomers: A critical review. Rev. Adhesion Adhesives 5, 195–267 (2017). 2. A. Y. Coran, Vulcanization, in: Science and Technology of Rubber, 3rd edition., J. E. Mark, B. Erman and F. R. Eirich (Eds.), pp.337–346, Elsevier Academic Press, New York (1994). 3. F. Ruch, M. O. David and M. F. Vallat, Adhesion in EPDM joints: Role of the interdiffusion mechanism on interfacial co-crosslinking. J. Polym. Sci. Part B. Polym. Phys. 38, 3189–3199 (2000). 4. A. K. Bhowmick and A. N. Gent, Effect of interfacial bonding on the self-adhesion of SBR and neoprene. Rubber Chem. Technol. 57, 216–226 (1984). 5. A. Ahagon and A. N. Gent, Effect of interfacial bonding on the strength of adhesion. J. Polym. Sci., Polym. Phys. Edn. 13, 1285–1300 (1975). 6. A. K. Bhowmick and B. Chakraborty, Bond strength in various rubber-rubber joints. Plast. Rubber: Process. Appl. 11, 99–106 (1989). 7. P. Loha, A. K. Bhowmick and S. N. Chakravarty, Modification of the peel test for testing of rubber to rubber joints. Polym. Test. 7, 153–163 (1987). 8. A. K. Bhowmick, P.Loha and S. N. Chakravarty, Studies on adhesion between natural rubber and polybutadiene rubber, Int. J. Adhesion Adhesives 9, 95–102 (1989). 9. A. Sarkar, D. Dutta and A. K. Bhowmick, Failure induced by stress concentration at rubberrubber interfaces. Plast. Rubber: Process. Appl. 14, 49–55 (1990). 10. A. Sarkar, A. K. Bhowmick and S. Majumdar, Photoelastic studies on rubber-to-rubber joints. J. Adhesion 36, 161–175 (1991). 11. R. J Chang and A. N. Gent, Effect of interfacial bonding on the strength of adhesion of elastomers. I. Self-adhesion. J. Polym. Sci., Polym. Phys. Edn. 19, 1619–1633 (1981). 12. H. Chun and A. N. Gent, Effect of length and number of interlinking molecules on the strength of adhesion. J. Polym. Sci. Part B. Polym. Phys. 34, 2223–2229 (1996). 13. A. Sarkar and A. K. Bhowmick, Fatigue failure of rubber-to-rubber joints. J. Adhesion 37, 225–237 (1992).

190 Progress in Adhesion and Adhesives, Volume 4

14. A. N. Gent and S. M. Lai, Adhesion and autohesion of rubber compounds: Effect of surface roughness. Rubber Chem. Technol. 68, 13–25 (1995). 15. G. C. Basak, A. Bandyopadhyay, Y. K. Bharadwaj, S. Sabharwal and A. K. Bhowmick, Adhesion of vulcanized rubber surfaces: Characterization of unmodified and electron beam modified EPDM surfaces and their co-vulcanization with natural rubber. J. Adhesion Sci. Technol. 23, 1763–1786 (2009). 16. K. L. Mittal, The role of the interface in adhesion phenomena. Polym. Eng. Sci., 17, 467–473 (1977). 17. G. C. Basak, A. Bandyopadhyay, Y. K. Bharadwaj, S. Sabharwal and A. K. Bhowmick, Characterization of EPDM vulcanizates modified with gamma irradiation and trichloroisocyanuric acid and their adhesion behavior with natural rubber. J. Adhesion 86, 306–334 (2010). 18. G. C. Basak, A. Bandyopadhyay, S. Neogi and A. K. Bhowmick, Surface modification of argon/oxygen plasma treated vulcanized ethylene propylene diene polymethylene surfaces for improved adhesion with natural rubber. Appl. Surf. Sci. 257, 2891–2904 (2011). 19. G. C. Basak, A. Bandyopadhyay and A. K. Bhowmick, Effect of tackifier compatibility and blend viscoelasticity on peel strength behavior of vulcanized EPDM rubber co-cured with unvulcanized rubber. Int. J. Adhesion Adhesives 30, 489–499 (2010). 20. G. C. Basak, A. Bandyopadhyay and A. K. Bhowmick, Influence of nanoclay on adhesion of EPDM vulcanizate, Int. J. Adhesion Adhesives 31, 209–219 (2011). 21. S. Choi, J. Kim and H. Lee, Analysis of thermally aged adhesion specimen between precured and uncured rubber sheets. J. Ind. Eng. Chem. 15, 624–627 (2009). 22. D. Oldfield and T. E. F. Symes, Surface modification of elastomers for bonding. J. Adhesion 16, 77–96 (1983). 23. J. M. Martin-Martinez, J. C. Fernandez-Garcia, F. Huetra and A. C. Orgiles- Barcelo, Effect of different surface modifications on the adhesion of vulcanized styrene-butadiene rubber. Rubber Chem. Technol. 64, 510–521 (1991). 24. M. S. Sanchez-Adsuar, M. M. Pastor-Blas, R. Torregrosa-Macia and J. M. Martin-Martinez, Relevance of polyurethane configuration on adhesion properties. Int. J. Adhesion Adhesives 14, 193–200 (1994). 25. M. M. Pastor-Blas, M. S. Sanchez-Adsuar and J. M. Martin-Martinez, Weak boundary layers in styrene-butadiene rubber. J. Adhesion 50, 191–210 (1995). 26. M. M. Pastor-Blas, J. M. Martin-Martinez and J. G. Dillard, Surface characterization of chlorinated synthetic vulcanized styrene-butadiene rubber using contact angle measurements, infrared spectroscopy and XPS. J. Adhesion 63, 121–140 (1997). 27. M. D. Romero-Sanchez, M. M. Pastor-Blas and J. M. Martin-Martinez, Adhesion improvement of SBR rubber by treatment with trichloroisocyanuric acid solutions in different esters. Int. J. Adhesion Adhesives 21, 325–337 (2001). 28. M. D. Romero-Sanchez, M. M. Pastor-Blas and J. M. Martin-Martinez, Improved peel strength in vulcanized SBR rubber roughened before chlorination with trichloroisocyanuric acid. J. Adhesion 78, 15–38 (2002). 29. M. M. Pastor-Blas and J. M. Martin-Martinez, Different surface modifications produced by oxygen plasma and halogenation treatments on a vulcanized rubber. J. Adhesion Sci. Technol. 16, 409–428 (2002). 30. M. D. Romero-Sanchez, M. M. Pastor-Blas and J. M. Martin-Martinez, Improved adhesion between polyurethane and SBR rubber treated with trichloroisocyanuric acid solutions containing different concentrations of chlorine. Composite Interfaces 10, 77–94 (2003).

Adhesion between Compounded Elastomers: A Critical Review 191

31. J. Tyczkowski, I. Krawczyk and B. Wozniak, Modification of styrene-butadiene rubber surfaces by plasma chlorination. Surf. Coat. Technol. 174–175, 849–853 (2003). 32. N. Pastor-Sempere, J. C. Fernández-García, A. C. Orgilés-Barceló, M. S. Sánchez-Adsuar and J. M. Martín-Martínez, Improved adhesion properties of polyurethane adhesive containing fumaric acid. J. Adhesion 59, 225–239 (1996). 33. C. M. Cepeda-Jimenez, M. M. Pastor-Blas, J. M. Martin-Martinez and T.P. Ferrandiz-Gomez, Surface characterization of vulcanized rubber treated with sulfuric acid and its adhesion to polyurethane adhesive. J. Adhesion 73, 135–160 (2000). 34. M. M. Pastor-Blas, J. M. Martin-Martinez and J. G. Dillard, Surface characterization of synthetic vulcanized rubber treated with oxygen plasma. Surf. Interf. Anal. 26, 385–399 (1998). 35. H. Hirahara, K. Mori and Y. Oishi, Direct adhesion of fluorinated rubbers to nickel plated steel and nitrile rubber during curing using the tetrabutylammonium salt of 1,3,5-triazine-2,4,6trithiol. J. Adhesion Sci. Technol. 11, 1459–1474 (1997). 36. L. Job and R. Joseph, Studies on the adhesives for rubber to rubber bonding. J. Adhesion Sci. Technol. 9, 1427–1434 (1995). 37. T. F. Petrova, A. A. Shcherbina and A. E. Chalykh, Effect of water on the strength of rubber bonding with adhesives based on reactive oligomers. Polymer Science Series C 49, 89–94 (2006). 38. M. M. Pastor-Blas, M. S. Sanchez-Adsuar and J. M. Martin-Martinez, Surface modification of synthetic vulcanized rubber. J. Adhesion Sci. Technol. 8, 1093–1114 (1994). 39. M. M. Pastor-Blas, J. M. Martin-Martinez and J. G. Dillard, Influence of the nature and formulation of styrene-butadiene rubber on the effects produced by surface treatment with trichloroisocyanuric acid. J. Adhesion Sci. Technol. 11, 447–470 (1997). 40. M. M. Pastor-Blas, T. P. Ferrandiz-Gomez and J. M. Martin-Martinez, Chlorination of vulcanized styrene-butadiene rubber using solutions of trichloroisocyanuric acid in different solvents. J. Adhesion Sci. Technol. 14, 561–581 (2000).

5 Contact Angle Measurements and Applications in Pharmaceuticals and Foods: A Critical Review Davide Rossi1,2*, Paola Pittia2 and Nicola Realdon1 1

Department of Pharmaceutical and Pharmacological Sciences, University of Padova, Via F. Marzolo, 5, 35121 Padova, Italy 2 Faculty of Biosciences and Technologies for Agriculture Food and Environment, University of Teramo, Via Balzarini, 1, 64100 Teramo, Italy

Abstract Contact angle (CA) analysis is a noninvasive and rapid method to determine the wettability of various materials including biomaterials. Contact angle measurements and interpretations have been performed since the beginning of the 19th century. This review provides an overview of contact angle analysis and measurements performed on surfaces of pharmaceuticals and foods in relation to other analytical approaches. The contact angle method represents an analytical approach widely applied in order to determine the surface free energy of complex materials in these fields. Our review is focused on contact angle analysis and surface free energy characterization of pharmaceutical powders, injectable drugs, and foods. Correlations between chemistry, rheology and surface properties of these materials were considered in order to demonstrate the importance of the integrated analytical approach (IAA) for a complete characterization of these kinds of materials and the critical issues encountered in CA measurements. This review is organized in two parts: the first one is an overview of the history of research in CA method since its beginning, and the second one is focused on an overview of particular applications of CA method in pharmaceuticals and foods. In this work, it is shown that the topic of the surface properties of materials in the new “pharma-food” field is currently a very important area of research for the interpretation of the phenomena and the control of the quality and processing of these materials. Keywords: Contact angle, skin hydration, pharmaceutics, foods, cosmetics, surface free energy

*Corresponding author: [email protected]

K.L. Mittal (ed.) Progress in Adhesion and Adhesives, Volume 4 (193–240) © 2019 Scrivener Publishing LLC

193

194 Progress in Adhesion and Adhesives, Volume 4

5.1

Introduction

The contact angle measurements have been carried out since the 19th century during which for the first time the contact angle (CA; deg) was described as the result of the contact between a solid (s) and a liquid (l). The curvature of l depends on the equilibrium of three interfacial tensions (ST) at liquid- vapor ( l/v), solid-liquid ( s/l), and solid-vapor ( s/v) interfaces [1]. These first studies provided the basis for the concept of wettability of materials which is measurable through the determination of the CA at the interface between s and l and which depends on the balance between the gravity force and the cohesion force within the liquid depending on its chemical composition and the force of adhesion at the interface between the two materials [2]. During this period the first technique for measuring the surface pressure by surface tension (ST; mN/m) measurement was developed [3], opening new perspectives in the interpretation of the CA formation. All the studies performed in these years were developed from the investigations of the capillary phenomenon in which the measurement of the height reached by the liquid that rises in a capillary represented the main method for testing the theories of capillary action [4]. In the 19th century, important studies using the thermodynamic approach for the explanation of CA were also performed [5] and the influence of vapour (v) on the work of adhesion (WA; mJ/m2) was studied [6], opening a new aspect of CA investigations. Successively, in the first part of the 20th century the attention of researchers was focused on the observation of the behaviour of drops deposited on a substrate with the aim to characterize the orientation of the molecules present in the surface of liquids taking in account the total energy, the chemical composition and its influence on the ST [7]. These investigations helped to demonstrate the correlation between CA and polarity of solid surfaces [8] and underlined the importance of flat surfaces in the determination of CA at s/l interface in relation to the liquid drop size [9]. Other studies demonstrated the influence of the roughness on the CA analysis also [10] (Figure 1). The upper left-hand sketch of Figure 1 presents a plan view of the solid surface and the line M-N represents a segment of the periphery of the wetted area advancing from left to right. The studies on the capillary flow in a bundle of parallel cylindrical tubes previously conducted in the 19th century were extended to the imbibition phenomenon of a liquid drop into porous materials [11]. Successively, it was possible to develop a method for determining CA on a finely divided substrate [12]. All these new researches performed during the first years of the 20th century on different kinds of solid surfaces opened the application of the CA method in different research fields for ST and surface free energy (SFE; mJ/m2) characterization of liquids and solid materials. Other studies started to take into account the surface area occupied by a drop of liquid onto a solid in relation to the measured CA, the volume of the drop, the density of the liquid and its ST [13]. At the beginning of 1940s, the need for determining the film pressure of a liquid for the determination of the pressure at the interface between a solid and aqueous solutions of organic substances led to the development of a new apparatus for CA measurements using a tilting plate coupled with a Wilhelmy type film balance [14]. In the 1940s the attention was focused on the surfaces of solids and the first studies were performed by Washburn [11] and Bartell and Whitney [12] in order to confirm

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 195

N

S2

B

C S12 12

A

D

S1 A

M

S2

N

D

‘

B

C rS12

G

F

H

E

S2 cos

rS1 A M

rA

S2 cos ‘

Figure 5.1 Vector relations of surface forces:; (above) smooth solid surface, (below) rough solid surface where S1 and S2 are force vectors acting on a unit length of the periphery, A-B is unit length of the periphery, A is the adhesion tension, S2 is the surface tension of the liquid, r is the roughness factor, EFGH is the area on which the specific interfacial energy is concentrated. rA is the product measured by the term S2 cos calculated for real solids, and the contact angle (Adapted from [10]).

the suitability of the CA method for the determination of micropore-size distribution in porous solids. The micropore-size distribution can be obtained by calculating the pressure required to force a liquid into a pore depending on the radius of the pore, the ST of the liquid and the CA (deg) measured at the interface [15]. The studies performed on porous solids were accompanied in the same period by the works of Cassie and CassieBaxter [16] whose equation was used to describe the phenomenon of superhydrophobicity shown by natural and artificial surfaces characterized by extremely high (> 150°) CA at the interface with water [17].This approach opened up the investigation on low surface energy substrates [16]. The Cassie-Baxter approach was based on the concept of advancing ( a) and receding ( r) CA measurements. This approach focuses on the interfacial force between a drop and an inclined substrate. The difference between a and r is called Contact Angle Hysteresis (CAH) and it influences the determination of the surface tension (ST) of a liquid [18, 19]. In order to determine in an easy way the CA of a liquid on a solid surface, in the 1940s a novel goniometer was built and from which the concept was developed for the modern tensiometers [20]. Due to the employment of the new goniometer and on the basis of previous studies, new concepts of CA ( ) were studied in which it was demonstrated that its value was linked directly with the interfacial tensions at solid-liquid ( s/l),solid-vapour ( s/v) and liquid-vapour ( l/v) interfaces [21]. In Figure 2 is

196 Progress in Adhesion and Adhesives, Volume 4

reported an example of CA measured at the interface between a heated solid surface and a drop of liquid. In the 1950s the attention of many researchers was focused on investigation of the wettability property of solids with low SFE e.g., for fluorinated compounds [22]. In Figure 3 is reported the cosine of CA for n-alkanes on surfaces composed of -CF3, -CF2H, -CF2- and -CH3 groups in correlation with ST (dynes/cm) by a linear equation. These works compared the wettability of fluorinated compounds (CF2H and CF3 groups) after reactions with esters, amines, alcohols, organic acids and other polar components such as glycerol based on the previous studies performed by Fox and Zisman which demonstrated that the wettability of low energy solids depends on the atomic composition of their surfaces [23–25]. All these studies demonstrated that in the case of solid surfaces with low SFE, the CA variations and the ST of a large variety of liquids are linked and the

Liquid Vapor lv

sl

sv

Heated surface

Figure 5.2 Scheme of a drop of liquid on a heated solid surface where is CA, lv is interfacial tension at liquid-vapour interface, sv is interfacial tension at solid-vapour interface, and sl is the interfacial tension at solid-liquid interface (Adapted from [21]).

10

Cosine

8

6

4

2

18

22 26 Surface tension (dynes/cm at 20˚C)

30

Figure 5.3 Linear correlations between CA (deg) and ST (dynes/cm) of n-alkanes on low energy surfaces of -CF3 (O), -CF2 ( ) and -CH3 (Δ) groups, (Adapted from [22]). 2

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 197

wetting of the substrate depends on the kind of surface. New thermodynamic approaches at solid-liquid-vapour (s/l/v) systems by using the techniques of Gibbs gave the possibility to derive the Young’s equation for a sessile drop on a solid in which adsorption and gravity were also considered [26]. This approach demonstrated for the first time that Young’s equation was valid in all situations studied. From these observations in the early 1960s a relation between the critical ST of spreading for low energy solid surfaces and the constitution of their surfaces was found, formulating for the first time the constitutive law of wettability [27] (Figure 4). Zisman studies on low SFE solids performed in the 1950s and 1960s proved to be fundamental for the successive developments of mathematical models for the evaluation of the interfacial tension (IF; mN/m) between liquids having different ST values [28]. In light of these considerations, an equation relating the free energies of cohesion of separate phases to the free energy of adhesion, and the ST to the IF was developed using CA as the main parameter [29]. In the same period, the CAH, previously studied by Cassie&Baxter [16] and McDougall and Ockrent [18], was described using a digital model able to determine a and r on ideal heterogeneous surfaces. By this digital model it was possible to define the CAH as a balance between the vibrational energy of the drop and the heights of energy barriers between different metastable states [30]. In the 1960s, new investigations on the interfacial force (IF) between a solid and a liquid by the CA method opened a new frontier of science evoking new fields of applications [30]. In this context, further new studies were conducted for the evaluation of liquid flow in a dynamic wetting situation [31] demonstrating its influence on the phenomena that occur at l/s interface and opening new perspectives in the analysis of the drop motion time by the CA method (Figures 5, 6). The increased need for CA research in a number of scientific fields led to an improvement in instruments for CA measurements with the objective to enhance the performance of the goniometer apparatus by modifying the instrument developed by Zisman in the 1940s (goniometer) and Hayman in the 1950s and the tangentometer scheme is reported

Critical surface tension (dynes/cm at 20˚C)

60 50 By chlorine 40 30

By fluorine

20 10 0 0

75 25 50 100 Atom percent of hydrogen replaced

Figure 5.4 Effect of progressive halogen substitution on the critical surface tension of polyethylenetype surfaces (Adapted from [27]).

198 Progress in Adhesion and Adhesives, Volume 4

C ontact angle, degrees

109

105

101

97

93

89 0

4

8 12 16 T ime, thous andth s ec.

24

20

Amplitude of drop top (10–2 mm)

Figure 5.5 Experimental characteristic of drop oscillation relative to time (Adapted from [32]).

57.93 wt. % glycerol in water drop volume = 3.69 mm3

12 8 4

1/e amp. 4.53 × 10–2 mm = 38.65 × 10–3 sec

0 0

10

20 30 Time, 10–3 sec

40

50

Figure 5.6 Correlation between the decrease of the amplitude of drop top (mm) and time (sec) where represents the experimental decay time, 1/e is the decrease value of the amplitude of drop’s top motion (AMP.) from the original value where e is the base of the natural logarithms (Adapted from [32]).

by Fenrick [33] while a modification of the Langmuir & Shaeffer apparatus was performed with an increase in the simplicity of manual operation, ruggedness, high precision and, consequently, the accuracy of the drop profile method [34]. All the researches performed by the CA method in more than a century and reported above led to the development of mathematical conversion models capable to determine in a precise and accurate way the surface free energy (SFE; mJ/m2) and dispersion (DC; mJ/m2) and polar components (PC; mJ/m2) of liquids and solids of different nature in the 1970s, 1980s and 1990s. The aim of

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 199

this review is to present a condensed overview of the recent progresses and perspectives of the CA method for the characterization of systems in “pharma-food” field with the main focus on some recent methods applied to some particular situations in which the tensiometric technique has been used for the characterization of the wettability.

5.1.1

Prospects

Various theoretical approaches have been developed for the determination of the ST of liquids and SFE of solids by the CA method in the fields of food and pharmaceutics. On the basis of the fact that the SFE of a solid dictates its surface properties and interfacial phenomena such as adsorption, wetting or adhesion with liquid phases, the knowledge of ST and SFE is of great interest in many fields which consider the interfacial phenomena useful for numerous industrial applications. For example, in the cases of packaging, food science, biomedical applications and pharmaceutical products, cleaning processes, adhesive technology, painting, coating and many other applications in many fields in relation to wettability of their systems, the CA analysis can be applied widely [35]. The CA method has been known since the 19th century, during which the theoretical basis of CA phenomena and surface free energy of materials was developed; actually the measurements and interpretations of CAs are still debated in the scientific literature because of the need to better understand the fundamental mechanisms at s/l interface. The novel applications of the CA method in different research fields linked to the human health within the Integrated Analytical Approach (IAA) can give more information about the links between the structure and surface of the systems investigated. In particular, the ST and SFE characterization in the field of food science is deeply investigated, especially for packaging and coating applications and suggests different strategies in order to improve the processing and quality of these materials [35]. These experimental results demonstrated that the wettability is one of the fundamental parameters in food emulsions and cosmetics industries as the water-oil (W/O) emulsions and oil-water (O/W) emulsions are stabilized by fine solid particles and the stability of these formulations depends directly on the particle wettability. In the pharmaceutical industry the wettability of powders is important for the production of formulations and the manufacturing and optimization of the production process of drugs [36]. We focus on some recent studies on particular and innovative aspects and applications of CA approach in the study of pharmaceutics and foods of actual industrial and commercial interest within the novel “pharma-food” field. In this context in the last years the pharma-food concept has had impact on pharmacological and food research because several food components are employed as medicines and pro-drugs [37]. Visioli [37] recently reported that in the context of “pharma-food” there are areas in which the difference between food and pharma is not well-defined because nutrition could involve bioactive compounds (drugs), fibers, plant molecules and other substances (nutrients). The chronic diseases can be treated by controlling the macro- and micronutrients in pharmacological doses (supplements, nutraceuticals, functional foods). Visioli highlighted that the traditional pharmacotherapy is often accompanied by supplementary treatments with nutrition-derived remedies to decrease the doses of medicines and reduce their side effects [37]. In the context of pharma-food,

200 Progress in Adhesion and Adhesives, Volume 4

the CA approach for a non-invasive investigation of different pharmaceutical forms can be useful for the determination of the quality and processing in food systems also and can be considered as a common analytical approach capable to link these two fields.

5.2 5.2.1

Contact Angle Measurements in Pharmaceutical Field Pharmaceutical Powders

Generally, the CA analysis of pharmaceutical products has considered mainly the powder form in which the wettability of small particles influences their ability to float on a liquid. One of the most common methods for the CA determination of particles is the h-epsilon (h- ) method [38] which involves e.g. the measurement of the maximum height of a droplet containing solid particles (Figure 7). In general the data obtained from the h-epsilon method are independent of the size of the particles and the porosity of the material, making this method applicable for investigation of any powder system. The most important result obtained from the application of the h-epsilon method is that hydrophobic powder materials are characterized by large particle size [39]. According to Lerk et al. [39], the CA method was found to be suitable for investigations on hydrophobic powders also and the wettability becomes an important parameter for a non-invasive characterization of many kinds of pharmaceutical powders together with the calculation of their dispersion (DC: mJ/m2) and polar (PC: mJ/m2) components. In this context, Marston et al. investigated the dynamics of the drop impact onto powder surfaces, opening the possibility to study the temporal evolution of the spreading phase, drop deformation and the resulting crater morphology [40]. For water drop impacting on dry and partially saturated powders, a low impact speed 0.15 m/s and a high impact speed (2.2 m/s) were investigated. In particular, when a drop wets a pharmaceutical powder bed, the corresponding times available for the penetration should be considered as a function of moisture content as demonstrated by the analysis of apparent contact diameter with time as a function of the speed of drainage (ui). After the impact of the liquid drop on the powder, the imbibition is characterized by two limited cases: the constant drawing area (CDA) (Equation 1) and decreasing drawing area (DDA) (Equation 2) both depending on CA measurements.

t CDA

1.35

t DDA

V02/3 2

R pore

(1) cos

p

(2)

9t CDA

where Vo is the drop volume, is the liquid dynamic viscosity, the porosity of powder bed, is the ST of the liquid drop and p is the apparent CA. The time of penetration depends on the size of the pore (Rpore) of the powder bed (Equation 3).

R pore

2 /((1- )S0

s

(3)

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 201 1 0.5 c

60

0

cos

cos

c

c

120

c

(degree)

180

–0.5 –1

0 0.010 0.015 0.020 0.025 0.030 0.035 0.040

(eV) w

w h

= 0.015 eV

C

= 0.025 eV

= 0.035 eV

Figure 5.7 CA measurements method for molten Ag droplets on graphene where c can be determined from height (h), is the energy supplied (eV) and w changes clearly with between 0.01 eV and 0.04 eV (Adapted from [38]).

where is the porosity of the bed, S0 is the particle specific surface area and s is the packing density of the pores. Marston et al. [40] showed that in the case of pharmaceutical powders, the CA values can be determined by the capillary rise method using a force tensiometer and Equation (4). 2

m2

c

cos

p

t

(4)

where m is the mass of liquid in the capillary at time t, is the density of the liquid that rises and c is a constant depending on the material used. As reported by Marston et al. [40], many experimental and application studies have been performed in order to investigate the impact of a water drop on pre-wetted moisturized pharmaceutical powders demonstrating that the moisture content has only a little influence on CA measurements and drop spreading appears similar for different moisture contents. However, the maximum drop spread and subsequent drainage into the powder are significantly influenced by the liquid and the impact speed. Marston et al. [40] showed that in the case of a low impact speed, the imbibition moment during the initial spreading phase is small. In the case of a dry powder, after the impact on the surface the drop can drain at both low and high impact speeds. The high impact speed can cause the formation of granules, while for wet powders with various moisture contents the drop drainage into the powder occurs without the formation of granules and splashing in pre-wetted powders. As evidenced previously, the spreading time is a strong function of the moisture content of the powder bed and the drop impact speed determines the maximum spread diameter from impact to complete drainage, and the penetration time in the case of a pre-wetted powder is shorter than for dry powders whose

202 Progress in Adhesion and Adhesives, Volume 4

saturation influences the speed of penetration. The analyses of these parameters are very important for the study of the granulation process of pharmaceutical powders for which the effective pore radius, pore size, CA and permeability represent fundamental parameters. Anyway, considering the dynamics of the penetration process of a liquid into a powder bed, Han et al. proposed the droplet penetration method (DPM) as a new tool capable of characterizing the wettability of pharmaceutical powders [41]. The DPM considers that the capillary pressure inside the porous powder is the only dominant driving force for the imbibition of drops, and the contact area between the penetrating drop and the powder bed is constant. The DPM was developed on the basis of the penetration process observed after the deposition of droplets of two different liquids on a slightly compressed powder bed and recording their penetration [41]. Han et al. demonstrated that DPM can evaluate in a new way the influence of the particle size distribution on the wetting behavior of pharmaceutical powders [41]. Han et al. considered, as an example, the lactose monohydrate excipient powder with particles in 38–45, 45–53, 53–63, 63–75, 75–90, 90–106 size ranges m), and anhydrous caffeine as active pharmaceutical ingredient (API). The test liquid used was deionized water. In this case, DPM demonstrated that the lactose powder with particle size smaller than 75  m has approximately the same CA and a significant decrease occurs when its particles are larger than 75 m . In the case of caffeine powder Han et al. found that larger particles resulted in smaller CA values. In light of these experimental evidences, the DPM represents a useful tool for controlling the granulation process of pharmaceutical powders [41]. However, Marston et al. [40] demonstrated that a force tensiometer was efficient for CA measurements on powders, and the wettability of dispersed materials of pharmaceutical powders could be determined through the capillary penetration method using the traditional Washburn apparatus (Figure 8) [42]. Teipel and Mikonsaari [42] reported that the Washburn equation combines the Hagen-Poiseuille Equation (5) with the equation developed for the evaluation of the capillary pressure of liquids or Laplace equation (Equation 6) to obtain an expression that describes the liquid through a cylindrical capillary (Equation 7).

pr 2 8 h

dV dt

p 2 h2

tr

lv

cos cos 2

lv

(5) 1 r

(6) (7)

where h is the length wetted by the liquid with volume dV in time t, the term p represents the Laplace pressure, r is the capillary radius, represents the viscosity of the liquid, t is the flow time of the liquid, lv is the ST of the liquid and is the advancing CA. The Washburn equation is a common method to analyze the increase in mass measured as a function of time. Teipel and Mikonsaari [42] reported that in the Washburn method the

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 203

Balance

Bulk material

Frit

Liquid

Figure 5.8 Apparatus for the Washburn method (Adapted from [42]).

pharmaceutical powder is placed in a closed tube with a frit at the bottom end. The liquid rises through the glass frit into the powder and the increase in weight with time is measured with a balance. Teipel and Mikonsaari showed that the application of the Washburn equation allows to characterize the wetting behavior of powders, and in particular the critical SFE of pharmaceutical powders such as amylobartbitone, acetaminophen and adipic acid could be correlated with other methods such as sedimentation. The capillary rise method is comparable to the sessile drop measurements performed on pharmaceutical powders previously compacted or on large single crystals and the agreement between these two methods for the analysis of pharmaceutical powders led to the estimation of the value of SFE of solid ( s) using an equation of state. As reported by Teipel and Mikonsaari, and more recently Thakker et al. developed a modified Washburn apparatus for the measurement of powder wettability (Figure 9) [43]. As reported by Thakker et al. [43], the traditional Washburn apparatus was coupled to a high resolution electromagnetic microbalance system which can give a resolution R of 1 mg (equation 8) depending on the resistivity of the metal wire ( ), the length of the wire (l) and the change in the area of the wire (A) which influences the resistance and, consequently, a voltage signal proportionally to the force is produced.

R

l A

(8)

The CA measurements obtained with this novel Washburn apparatus are precise and accurate and demonstrate that the changes in the powder wettability are in accord with the hydrophilic and hydrophobic nature of the materials used [43, 44] (Table 1). The results obtained by Thakker et al. [43] are in accord with those obtained with a standard instrument. As an example, the new instrument can be used to evaluate the uncoated or nano-coated powders and treated polytetrafluoroethylene (PTFE) powders. The PTFE

204 Progress in Adhesion and Adhesives, Volume 4

RS 232 interface

Microbalance

Powder material Filter paper Liquid

Labview based computer data acquisition system Mechanical platform

Figure 5.9 Modified Washburn apparatus (Adapted from [43]). Table 5.1 Wettability results of uncoated and nano-coated starch [43]. Time (s)

Mass2 (g2)

Material constant C × 10–15

Contact angle  (deg)

Water

3

0.34

7.97

80.3

0.5% R972 coated (hydrophobic)

Water

3

0.29

11.0

83.7

1.0% R972 coated (hydrophobic)

Water

3

0.13

12.80

87.1

1.0% 200P coated (hydrophilic)

Water

20

4.04

13.5

78.8

1.5% 200P coated (hydrophilic)

Water

20

5.11

15.1

77.8

2.0% 200P coated (hydrophilic)

Water

20

6.82

15.9

74.5

Powder sample

Test liquid

Starch uncoated

R972: Aerosil nano-silica (hydrophobic) coating. 200P: Aerosil nano-silica (hydrophilic) coating,

increases the water sorption of the powder providing a wide range of applications in food and pharmaceuticals (Figure 10). The analysis of wettability of powders is of fundamental importance for the control of the granulation process of pharmaceutical powders and also the sample preparation method has a strong influence on the results after the process of penetration of the liquid into the powder [42]. The first step in the granulation process involves wetting and nucleation phases and is crucial for the preparation of the final pharmaceutical form. In

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 205 0.6

Mass2 (g2)

Untreated PTFE 1% AOT treated PTFE

0.4

0.2

0 0

5

10

15

Time (sec)

Figure 5.10 Mass versus time graphs of untreated and 1% AOT treated PTFE in water (Adapted from [43]).

the case of non-wettable powders, the lifetime of the drop on the surface increases significantly enabling the use of hydrophobic powders for the production of granules. The analysis of hydrophobic and hydrophilic pharmaceutical powders performed by the CA method led to the calculation of their sd (dispersion component) and the selection of an appropriate solvent for the granulation process during which the liquid is added to the powder with the aim to form granules and product tablets [40]. In the case of the prewetted powders, the moisture content is generally not considered as a control parameter and the penetration time is commonly longer than that for a dry powder. This is because during the granulation process the moisture changes systematically and the drop impacts dry, partially wet and saturated powder. In this context, Zhang et al. [45] demonstrated that the prediction of the performance of granulating solvent is based on the solvent–drug spreading coefficient which is linked to the density, porosity and friability of the granules obtained. Ahfat et al. [46] were the first to demonstrate that the dispersion component determined from the mathematical conversion of CA (deg) to surface free energy SFE (mJ/m2) was consistent with that determined by traditional analytical approach such as inverse gas chromatography (IGC); however, the correlation between CA and IGC techniques showed that the CA approach was difficult to apply to hydrophilic pharmaceutical powders due to the penetration of the drop in the core of the pharmaceutical form [46] (Tables 2 and 3). The retention time, i.e., the time taken by the probe to elute from the column, was used to calculate the dispersion component and polar acid/base parameters (KA and KD) of the powders using a series of thermodynamic equations. The WA can be obtained from the free energy of adsorption (Equation 9).

G°ads = N a WA

(9)

where WA is the work of adhesion, N is the Avogadro number, and a is the surface area of the liquid probe molecule. The WA can be expressed as the geometric mean of the dispersion components of the solid ( ds) and liquid ( dl) (Equation 10).

206 Progress in Adhesion and Adhesives, Volume 4

Table 5.2 The dispersion components and acid / base parameters of powders determined by IGC with the standard deviations shown in parentheses [46]. KA (10–2)

KD (10–2)

KD/KA

46.8 (0.8)

0.3 (0.1)

27.3 (1.4)

94.3

Spray dried lactose

41.4 (0.7)

4.8 (0.2)

24.9 (0.6)

5.2

Anhydrous lactose

41.3 (1.3)

4.3 (0.6)

28.8 (11.6)

6.7

Sample

d

Zamifenacin

s

(mJ m2)

Starch

39.8 (0.7)

5.1 (0.2)

7.7 (0.3)

1.5

Lactose monohydrate

39.3 (1.1)

1.9 (0.1)

39.8 (2.5)

20.5

Table 5.3 Surface free energy components of the test powders using the contact angle data for the three polar liquids (mJ/m2) [47]. TOT s

LW s

AB s

S

S

Theophylline

44.5

43.8

0.7

0.0

6.7

Caffeine

47.9

44.5

3.4

0.5

5.9

TOT + S

+

LW s

–

AB

= total surface energy of the solid, = Lifshitz-van der Waals contribution, s = acid-base contribution, s = electron-acceptor acid-base contribution, S- = electron-donor acid-base contribution

WA =2 (

d s

d 1/2 l

)

(10)

By combining equations (9) and (10) it is possible to correlate surface free energies with IGC data (Equation 11).

RTlnVN = 2N ( ds)1/2 a ( dl)1/2 + C

(11)

where VN is the net retention volume, a ( dl) is the product of the surface area and dispersion component of an apolar liquid such as alkane, ( ds) is the dispersion component of the unknown solid and C is a constant which takes into account the weight, surface area and vapour pressure of the probe in the gaseous state. The ds of the solid can be obtained from the slope of the line (Figure 11). The comparison of the values of dispersion component of the surface free energy of powders obtained by IGC and dynamic angle tester (DAT) showed that both techniques ranked the powders in the same order. In Table 2 are reported the acid and base parameters determined from the slope and intercept of the line. As an example, the CA/IGC combined approach was applied to study the wetting behaviour of caffeine and theophylline powders using the infinite dilution IGC [47]. According to Dove et al. [47], the pharmaceutical powders were adhered to a glass slide (glass slide method) and analyzed in tablet form using a classic Wilhelmy plate wetting method by which it is possible to determine the CA on a tablet surface (Equation 12).

cos = mg/p

LV

(12)

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 207

Polar probe

RT ln VN (kJ mol–1)

Alkane line

GABads

GDads

a.( ld)1/2 (Å2.J1/2.m–1)

RTInVN (J.mol–1)

Figure 5.11 The free energy of adsorption of a polar probe in relation to the alkane line (Adapted from [46]).

6000 5000 4000 3000 2000 1000 0 –1000 –2000 –3000 –4000 –5000 –6000

Alkane line Polar probes

5

5.5

6

6.5 a(

7 D L

7.5

8

8.5

9

9.5

)0.5 (Å2.J0.5.m–1)

Figure 5.12 Passage of gaseous polar probes through a theophylline powder column at 35 °C where on the x axis is reported the dispersion contribution to the surface tension of the liquid ( DL ) and the molecular area of adsorbed molecule (a) (Adapted from [47]).

where is the contact angle on the powder, m is the mass (mg), g the acceleration of gravity, p the perimeter of the compact solid and LV is the surface tension of the liquid (mN/m). Using the CA data it is possible to obtain the surface free energy components of powders such as theophylline and caffeine by using the models such as the van Oss, Chaudhury and Good [48] (Table 3). According to Ahfat et al. [46], plotting a graph of RT In VN against a ( dl )1/2 it is possible to determine the DC of the powders ( ds )1/2 from the gradient of the straight line (Figure 12).

208 Progress in Adhesion and Adhesives, Volume 4

On the other hand, to determine the polar character of the powders, a selection of polar probes with known acid-base characteristics is needed. Plotting RT In VN with Hdvap (dispersion component of the heat of vaporization of the probe) it is possible to determinate the free energy of adsorption ( Ga–sp) (Figure 13). As reported in Figure 13, the distance separating each polar probe from the point on the apolar straight line vertically below corresponds to the specific Ga–sp. The Ga–sp can be used to calculate the acid/base parameters of the powders (KA, KD ). Dove et al. [47] showed that the critical issue in the analysis of a powder tablet by the Wilhelmy plate method is that the data obtained are inconsistent with very high values of  s because the complexity due to topography and porosity of the powder plate. In general, the application of the Wilhelmy plate method is suitable for s values of about 44.5 and 47.9 mJ/m2 as shown in the case of theophylline and caffeine surfaces (Table 3). Another issue that can affect the values of CA in the Wilhelmy plate method is the possibility that data can partially reflect the CA of the underlying adhesive used to fix the powder on the support. For these reasons, the  s values obtained by IGC cannot be correlated with those from CA measurements. This situation demonstrated that the Wilhelmy plate method is not suitable for the analysis of the wettability of most of the pharmaceutical powders. The wettability of pharmaceutical powders is a fundamental parameter for the preparation of dispersed systems. The analysis of the wettability of pharmaceutical powders leads to the determination of the quantitative relationships between wettability and chemical structure parameters of pharmaceutical powders. However, still potential problems with interpretation of capillary wetting data for swellable solids using combined capillary wetting and IGC approaches (CA/IGC) persist. These two methods are suitable for the analysis of SFE of a wide range of industrial microcrystalline cellulose powders commonly used in the pharmaceutical sector, such as morphine sulfate powder, the wettability of which can be determined by the capillary rise 5000

Alkane line Polar probes

4000

RTlnVN (J.mol–1)

3000 2000 1000 0 –1000 –2000 –3000 –4000 22

24

26

28

30

32

34

36

38

40

42

Hdvap (kJ.mol–1)

Figure 5.13 Passage of gaseous polar probes through a theophylline powder column at 35 °C where Hdvap reported on the x axis is the dispersion component of the heat of vaporization of the probe (Adapted from [47]).

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 209

approach. However, the wetting rate data are reproducible for the liquids, and the sessile drop CA method on pharmaceutical compacted powder is not possible because the wetting tension data depend on its different crystal faces. As reported by Williams [49] in his recent review, this particular issue can be solved only if CAs are measured onto crystal faces with sufficiently large surfaces and data are correlated with other analytical techniques such as AFM and ToF-SIMS analytical techniques that provide information on the composition of the crystal facets of pharmaceutical compounds [49]. In Figures 14a and 14b are reported the surface free energy distributions as a function of size fraction, the lines show the surface free energies for specific facets of d-mannitol determined by contact angles on large d-mannitol crystals [49]. The work confirmed that the surface free energy distribution is sensitive to the shape and thus to the population of crystal facets found in a crystalline powder system [49]. Table 4 shows the surface free energy of untreated and silanised d-mannitol as determined with the Owens-Wendt [50] approach [49]. Using this combined approach, the CA method can effectively discriminate between different crystal facets in terms of their specific surface chemistries. As reported by Williams

50.0 (010) (mJ/m2)

(011)

46.0

42.0

s

(120)

d

48.0

Decreasing aspect ratio

d

Facet (010)

s

Facet (120)

s

Facet (011)

s

44.0

40.0 5 mm

hmax

Figure 5.22 Mechanism representing formulation of Liquisolid system (LS) (Adapted from [57]).

5.2.3

Injectable Solutions for Parenteral Use

The injectable suspensions contain hydrophilic powders generally suspended in aqueous systems. In this context, Patel demonstrated in his overview that the wettability of the ingredients assumes a crucial importance [58]. In this case, Patel considered nonionic surfactants and non-aqueous solvents such as glycerol and propylene glycol as wetting agents commonly used in the preparation of injectable suspensions [58]. This is because glycerol and propylene glycol reduce the CA at the interface between the particles and the liquid, increasing the wetting efficiency and stability of the final product. Patel highlighted that the surfactants with hydrophilic/lipophilic balance (HLB) values in the range of 7 to 9 are considered suitable for the preparation of this kind of drugs [58]. Proper concentrations of the surfactants (range 0.05%-0.5%) were found to be important because foaming or caking processes gave undesiderable issues in the case of their excessive concentration. Patel [58] indicated that the most common wetting agents used for the preparation of non-aqueous suspensions of drugs such as Cefazolin sodium in peanut oil are lecithin, Polysorbate 20, Polysorbate 80, Pluronic F68 and Sorbitan trioleate (Span 85). As reported by Patel, the solvent systems normally used in parenteral suspensions can be aqueous or non-aqueous and their choice is made on the basis of their solubility, stability and desired release characteristics of the final product. Non-aqueous solvents include both water miscible and water immiscible vehicles

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 219

and the agents used as cosolvent with water lead to the formation of soluble and stable parenteral preparations (ethanol, glycerin, propylene glycol, n–lactamide). As reported by Patel, the most common undesiderable effects of injectable drugs are myotoxicity and hemolysis (propylene glycol). These negative effects depend mainly on modifications of factors such as dielectric constant, pH, viscosity, and surface tension (ST). In this context, Patel provided evidence that these issues occur when the cosolvent is present in inadequate concentration which causes interactions with tissues such as skeletal muscle fibers and red blood cells. For these reasons, the CA approach assumes an important role in the evaluation of the quality of the injectable drugs. On the basis of these, Trivino and Chauhan [59] demonstrated in their review that the CA approach showed its usefulness in the characterization of solid lipid nanoparticles (SLNPs) carrier systems (50–1000 nanometers) that also use solid lipids as a matrix for drug delivery [59]. The lipids with a high CA may affect the stability and production of nanoparticles and the lipid chain length influences drug supersaturation levels and precipitation inhibition in self-emulsifying drug delivery systems. Thus, the appropriate selection of lipids for the preparation of Solid Lipid Nanoparticles (SLNPs) is important for a successful formulation (Figure 23). Considering that in the case of SLNPs the drugs are dispersed in the lipid matrix and are stabilized with an emulsifier, Trivino and Chauhan demonstrated that the fine size of SLNPs

GUT WALL Mucus Crystalline drug

hes

Ad

Solid lipid nanoparticles

EEnnzzyy maa ddeegg m rraaddaa ttiicc ttiioonn

Absorption ABSORPTION

ion

Drug release

Micelle formation

Lipid

Figure 5.23 The effects of lipids in prepared SLNPs on drug absorption. An immobilized crystalline drug is molecularly dispersed in the crystalline solid lipids forming the SLNPs whose stability depends on the bulk properties and the length of the lipids influencing the release and the absorption of the drug due to the adhesion and enzymatic degradation of SLNPs (Adapted from [59]).

220 Progress in Adhesion and Adhesives, Volume 4

leads to a real drug release system because of the high surface area with the advantage to reduce the potential issues that can affect the final formulation. As reported by Trivino and Chauhan, the SLNPs can be considered very important for the pharmaceutical industry due to their capability to incorporate both lipophilic (poorly water soluble drugs) and hydrophilic drugs (highly water soluble drugs). The SLNPs have great advantages because these are constituted by biocompatible lipids that substitute the organic solvents traditionally used. The enzymatic degradation of lipids results in the formation of micelles that improves the solubility of the drug, increasing its release. Trivino and Chauhan showed that the CA analysis is an important tool for selecting the different lipids excluding those that show high CA. This is important because lipids with high values of CA influence negatively the stability and production of nanoparticles. An appropriate selection of lipids for the preparation of SLNPs on the basis of their wettability is important for a successful formulation. In the field of nanotechnology, more recently De Solorzano et al. [60] reported that the poly(d,l lactic-co-glycolic acid) (PLGA) is particularly important as a biodegradable polymer widely used in drug delivery field in many areas such as vaccines, cancer treatment and inflammatory diseases [60]. De Solorzano et al. highlighted that the wide use of PLGA is due to its biodegradability and biocompatibility, usability in parenteral administration, capability to encapsulate hydrophilic/hydrophobic active pharmaceutical ingredients (APIs), and capability to protect the pharmaceutical form from biochemical degradation. Other characteristics of PLGA are capability to carry the nanoparticles (NPs) to specific tissues or cells, capability to sustain the release process over time, and possibility to easily modify its surface properties. In this context, in 2017 Nakashima et al. [61] developed a particularly highly dispersible PLGA in an aqueous fluid to solubilize a model drug. Aripiprazole (model drug) was dissolved in an aqueous fluid (containing PLGA) and successively spraydried to produce microparticles (MRPs) and further co-processed with water-soluble additives (ADDs) and surfactants (S) to improve their dispersion behavior [61]. The preparation of MRPs and granulated microparticles (G-MRPs) is performed by a Spray Drying Process step (step 1) and a Drop Freeze-Drying Process second phase (step 2) [61]. The physicochemical properties of the complex system constituted by MRPs and MRPs with ADDs(GMRPs) were evaluated by X-ray powder diffraction (XRPD)/DSC/CA combined approach taking into account the possibility to consider them as injectable pharmaceutical forms having long-acting release properties. In this case, Nakashima et al. [61] showed that the CA measurement of water gave important information about the wettability of the materials investigated and measurements were performed onto the surfaces of compressed G-MRPs considering the height of the drop (H) and the diametrical length of the water lens formed on the surface of the tablet (L) as the main parameters. The height (H) and diametrical length (L) of the water lens were measured immediately after formation of the droplet. The CA ( ) was calculated by equation 13:

= 2tan–1 (2H /L)

(13)

The crystalline state of the G-MPRs was evaluated by XRPD. The XRPD analysis was performed on intact Aripripazole (ARP) crystals, and various G-MRPs such as MRP,

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 221

G-MRP 20%HPC, G-MRP 20%PVA (poly (vinyl alcohol)), G-MRP50%MNT (granulated microparticles with 50%D-Mannitol) in relation to bulk mannitol ( -form) and bulk mannitol ( -form) [61]. As reported by Nakashima et al. [61], the CA values of the MRPs and G-MRPs underlined that G-MRPs are more wettable than the association between drug and MRP only and, consequently, are characterized by an excellent dispersibility in the aqueous phase (Figure 24). However, in the studies of Nakashima et al. [61] the CA method was applied as an analytical support for traditional analytical techniques such as X-ray powder diffraction (XRPD) and differential scanning calorimetry (DSC) that characterized the above-mentioned crystals and particles as ARP and its mixture, MRP, G-MRP, and G-MRP 50%MNT [61]. Asiri et al. [62 ] recently applied SEM/AFM/CA combined approach in which CA measurement was performed for the evaluation of the wettability of complex systems constituted by carbon nanofibers (CNFs) embedded in PLGA [62]. The main objective of Asiri et al. [62] was the demonstration that the cardiomyocyte functions increase with CNFs aligned in PLGA. The CNFs aligned in PLGA (poly(lactic-co-glycolic acid) showed greater cardiomyocyte density than both CNFs randomly oriented in PLGA and pure PLGA (CNFs in the PLGA matrix at 50:50 wt%). The alignment mimics the natural anisotropy of the cardiac tissue. The AFM (atomic force microscopy) analysis of the 50:50 CNF:PLGA aligned composite demonstrated an alignment of the CNFs on the PLGA surface [62]. This increase is because CNFs are characterized by high hydrophobicity and thus enhance the adsorption of key proteins such as fibronectin, laminin, and vironectin that are considered promoters of the cardiomyocyte adhesion. The increase of the cardiomyocyte adhesion is considered an expression of the cardiomyocyte functions. Asiri et al. [62] demonstrated that the CA method could help to improve the design of CNF/PLGA composites for numerous cardiovascular applications, opening new perspectives in the application of the CA method in the field of medicine also (Table 8). The cardiomyocyte cultured in vitro in the designed CNFs/PLGA by the CA method represented a greater expression of important cardiomyocyte biomarkers (cTNT, Cx43, -SCA) [62].

Contact angle (degree)

100

92.3 78.0

80 59.4

60

59.6 45.5

40

30.6

20 0 MRP

PVA bulk

G-MRP G-MRP MNT 50% PVA 20% PVA bulk

G-MRP 50% MNT

Figure 5.24 CA of Purified Water Droplets on Compacts Prepared by Compressing Various Formulations of Granulated Microparticles (G-MRPs) and D-mannitol (MNT) (Adapted from [61]).

222 Progress in Adhesion and Adhesives, Volume 4

Table 5.8 Comparison between the wettability poly(d,l lactic-co-glycolic acid ( PLGA) [62]. Sample

of

carbon

nanofibers

(CNFs)

and

Contact angle (degrees)

Pure PLGA

82 ± 3

50:50 wt% randomly oriented CNFs in PLGA

135 ± 2

50:50 wt% aligned CNFs in PLGA

145 ± 3

V dd D 00

d d LV

Ud

h

Ra

e

SV

FOOD

θ FOOD

Tf

Figure 5.25 Drop impact on a food surface where Vd is drop volume, d is drop viscosity, d is drop density, D0 is drop diameter, Ud is drop velocity, LV is drop surface tension, Tf is temperature, h is food surface distance from drop, Ra is food surface roughness, SV is surface free energy and is the CA at s/l interface (Adapted from [63]).

5.3 5.3.1

Contact Angle Measurements in Foods Solid Foods

As in the case of pharmaceutical powders, the tensiometric parameters SFE, DC and PC of solid foods need investigations on the impact of a drop of liquid on their surfaces as reported in the studies of Andrade et al. [63] (Figure 25). The main parameters involved in the dynamics of drop impact at controlled temperature and pressure are drop volume (Vd), drop viscosity ( d), drop density ( d), drop diameter (D0), drop velocity (Ud), drop surface tension LV and temperature (Tf), food surface distance from drop (h), food surface roughness (Ra), food surface free energy ( SV) and CA( ) for food surfaces [63]. Wetting behavior of impinging drops such as deposition, splash and rebound was investigated, respectively, onto banana, eggplant, and purple cabbage surfaces. The drop deposited into a thin disk (banana); disintegrated into secondary

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 223

droplets, splashed (eggplant); or rebounded (cabbage) demonstrating that splashing was due to high drop kinetic energy whereas rebound was the result of low food surface energy [63]. Andrade et al. [63] report that the Reynolds-Re (14), Weber-We (15) and Ohnesorge-Oh (16) numbers describe the behaviour of the impact of a drop on the surface of a solid food.

Re

We

D0U 0

(14)

D0U 02

(15)

LV

Oh

We Re

(16)

where U0 is the impact velocity of the drop on the surface. An optimized edible coating for food is defined as a uniform coating with a constant thickness of controlled microstructure [63]. For a good coating for the food, the drop impact behavior must be controlled, without splash, without rebound and with a controlled thickness. For a single drop impact, the thickness e is a function of the drop volume V and the maximum spreading diameter Dmax. Equation (17) shows that at constant drop volume the Dmax must be large to obtain a small thickness e.

Vd

6

D03

2 Dmax e 4

(17)

Andrade et al. [63] also highlighted that the effect of surface temperature on drop impact length behavior depends on the surface free energy. In the case of high surface free energy (hydrophilic substrate), the effect of substrate temperature appears to be more pronounced than for hydrophobic substrate. Another parameter that controls the wettability characteristics of the surface (hydrophobic or hydrophilic) of foods is the surface free energy [63]. Also in the case of foodstuffs, the value of the equilibrium CA depends on the anisotropy of the surface. The surface roughness is another important physical property of solid foods that influences the visual, sensorial attributes and their behavior during processing and storage. In particular, in the coating process, the surface roughness has a great effect on the adhesion between the coating solution and the food surface [63]. An increase in the roughness of the surface leads to prompt splashing. Andrade et al. [63] reported also that the ST plays a vital role in the splashing and rebounding rates. For example, in the case of cabbage, the rebounding phenomenon is weaker when ST of the liquid decreases [63]. The drop viscosity is also an important parameter affecting the impact process during which in the first stage of spreading the inertial force of impact is much larger than the viscous force [63]. Andrade et al. [63] highlighted that all the parameters that influence the drop impact behaviour can be stored in a database. In this way, from the physical properties of

224 Progress in Adhesion and Adhesives, Volume 4

the coating solution, it is possible to find the optimal drop impact conditions to obtain the desired deposition on a food surface [63] (Figure 26). As in many other fields, the main issues with the tensiometric analysis of solid foods are represented by the roughness and the irregularity of their surfaces which influence the CA measurements [64]. To address this issue, Parreidt et al. [64] applied a new CA measurement method for food materials and irregular surfaces in general. The new CA method called “DropSnake” and performed using ImageJ software was able to characterize foods such as strawberries (SFE; 21.26 mJ/m2, DC: 19.96 mJ/m2, PC: 0.99 mJ/m2) and endive salad (DC: 19.72 mJ/m2; PC: 23.39 mJ/m2) that are characterized by rough and irregular surfaces. For these kinds of surfaces the “DropSnake” approach was demonstrated to be a useful tool for CA measurements on irregular and rough solid food surfaces in general (Figure 27). Input database Output to prediction by neural network

Density Edible coating composition

Viscosity Surface Tension tension Splash Drop diameter

Drop impact conditions

Deposition Deposition

Impact velocity

Rebound Rebound

Roughness Food surface

Thickness

Surface Free free energy Energy Surface Tension tension

Figure 5.26 Influential parameters (input) for predicting the drop impact behavior (output) (Adapted from [63]).

Seeds

Figure 5.27 The “DropSnake” optical method to measure CA on a stacked grain (Adapted from [64]).

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 225

The “DropSnake” plug-in reported by Parreidt et al. [64] and Kurkuri et al. [65] is ideal for the analysis of asymmetric drops because no shape assumptions are used. As reported by Reinke et al. [66] also, the CA approach performed with ImageJ demonstrated the possibility to evaluate the alteration of some kinds of food surfaces such as chocolate blooming. These alterations are represented by the formation of white spots on the surface of chocolate products due to the presence of sugar or fat crystals on the surface. These white spots cause issues such as aesthetic changes and deterioration of the taste that are often associated with polymorphic changes of the crystal structure on the surface [66]. The analysis of the CA variation with time on the cocoa butter surface was investigated by the sessile drop directly after droplet deposition at 0s (92 deg), 5s (68 deg), 10s (64 deg) and 30s (61 deg). The variation of CAs of water was studied on untempered cocoa butter stored for 1 week at 20 °C at 0s (95 deg), 5s (94 deg), 10s (94 deg) and 30s (91 deg). The CAs of ethylene glycol were measured on tempered cocoa butter at 0s (91 deg), 5s (87 deg), 10s (86 deg) and 30s (83 deg), and on the same kind of cocoa butter the CAs of diiodomethane were measured at 0s (59 deg), 5s (45 deg), 10s (43 deg) and 30s (40 deg) [66]. Reinke et al. [66] demonstrated that the CA measured by imageJ on inclined surfaces can distinguish tempered and untempered foods such as chocolate and cocoa butter in accord with SEM results (Figure 28). Microscopic investigations using SEM were also performed on cocoa butter, dark chocolate,milk chocolate, freshly untempered cocoa butter and dark chocolate and milk chocolate [66]. Reinke et al. [66] demonstrated that the influence of the roughness of the surface on CA measurements depends mainly on the wetting ability of the liquid (Figure 29) [66]. Contact angles of water, ethylene glycol and diiodomethane on tempered cocoa butter (CB) and directly after production and after storage at 30 °C for 1 week [66] were also investigated. A particular field in the food industry is the sector of the dairy powders frequently used for their convenience in transportation, handling, processing, and for product formulations. As reported by Sharma et al. [67] in their review, the powdered foods in the nutrition field have particular physical and functional properties that control their structure, particle size distribution, powder density, bulk density, particle density, wettability, sinkability, dispersibility, solubility, hygroscopicity, heat stability, emulsifying properties, glass transition temperature, water activity, stickiness, and caking [67]. In particular, the wettability of a food powder depends on particle size, density, porosity, surface charge, surface area, the presence of amphipathic substances, and the surface activity of the particles. Surface coverage with hygroscopic components such as lactose yields good wetting properties because of the small CA. The wettability of food powder such as milk depends also on temperature. In particular, when milk powder is instantly prepared its wettability and CA values change rapidly over 15s. A rapid wetting of milk powder is also favored by large particles of high porosity, and natural surfactants such as soy lecithin are commonly used to increse the wettability of the powder [68]. The characterization of food powders is similar to that of pharmaceutical powders and Górnicki et al. [68] have emphasized that in the nutritional field the rehydration process of dried fruits has a particular importance and is based on the capillary movement of water in the powder samples as

226 Progress in Adhesion and Adhesives, Volume 4 Ethylene glycol Advancing Approx. max

100 75 50 25

Contact angle (°)

Contact angle (°)

Ethylene glycol 125

125

0

75 50 25 0 CB Temp CB Untemp DC Temp DC Untemp

CB Temp CB Untemp DC Temp DC Untemp Water Advancing Approx. max

100 75 50 25

Water

125 Contact angle (°)

Contact angle (°)

125

0

Receding Approx min.

100 75 50 25 0

CB Temp

CB Untemp

DC Temp

DC Untemp

CB Temp

Diiodomethane Advancing Approx. max

100 75 50 25

125 Contact angle (°)

125 Contact angle (°)

Receding Approx min.

100

100

CB Untemp

DC Temp

DC Untemp

Diiodomethane Receding Approx min.

75 50 25 0

0 CB Temp

CB Untemp

DC Temp DC Untemp

CB Temp CB Untemp DC Temp DC Untemp

Figure 5.28 Advancing and receding contact angles of ethylene glycol, water and diiodomethane measured on inclined tempered and untempered cocoa butter (CB), and tempered and untempered dark chocolate (DC). Approximated maximum and minimum contact angles are from approximated experimental data (Adapted from [66]).

shown for pharmaceuticals [66]. The flux of water through a dried material follows the capillary motion as described by the Washburn equation. In the case of foods, the Washburn equation can also be applied. As reported by Górnicki et al. [68], the milk powder is one of the most consumed dried foods and its wettability can be evaluated easily and quickly by CA measurements using the Washburn equation and apparatus. The application of the Washburn equation is possible because of the ability of milk powder to adsorb water on the surface. Górnicki et al. [68] highlighted that the wettability of dry milk is linked to its solubility and depends on the particle size, density, porosity, surface charge, surface area, presence of amphipathic substances, and surface activity of the particles. Consequently, dried milk with fast wetting property is characterized by large particles which represents an essential condition for their solubilization in water. The control of the insolubility index (II) of dry milk after the addition of surfactants such as soy lecithin is a method commonly used to enhance and control the wettability of the food powder in relation to the sediment formed as II is considered a typical correlation parameter to CA measurements.

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 227 120 Contact angle (°)

Water 110 100 90 Contact angle water 1s Contact angle water 10s Contact angle water 20s

80 70 0

20

40

60

80

100

120

140

160

180

Roughness Sp ( m) Ethylene glycol

Contact angle (°)

120

Contact angle ethylene glycol 1s Contact angle ethylene glycol 10s Contact angle ethylene glycol 20s

110 100 90 80 70 60 0

20

40

140

160

180

Diiodomethane

60 Contact angle (°)

60 80 100 120 Roughness Sp ( m)

Contact angle diiodomethane 1s Contact angle diiodomethane 10s Contact angle diiodomethane 20s

50 40 30 20 10 0

20

40

60 80 100 120 Roughness Sp ( m)

140

160

180

Figure 5.29 Contact angles of water, ethylene glycol and diiodomethane on dark chocolate with variation in roughness where Sp values are the maximum peak heights for roughness determination at different times (Adapted from [66]).

In particular, the movement of water through the dried powder material follows the capillary motion as described by the Lucas-Washburn equation, assuming that the food structure consists of multi-individual pores; the flow is one dimensional, steady state and fully developed; and the liquid is Newtonian with negligible inertia effect. On these bases, equation 18 was obtained [68]. dh(t ) dt

k1 k h(t ) 2

(18)

228 Progress in Adhesion and Adhesives, Volume 4

where k1 and k2 are parameters linked to the wettability of powder and are expressed in terms of CA (equations 19 and 20) [68].

k1 k2

r cos 4

(19)

r2 g 8

(20)

where is liquid density (kg m−3), the liquid viscosity (Pa.s), the surface tension (N m−1), r the the mean pore radius, the advancing liquid contact angle (rad), and g gravitational constant (m s−2). In light of these considerations, recently Banach [69] in his PhD thesis showed that the proteins pattern represents one of the more important ingredients of milk and in the food industry the dry milk is one of the main sources of milk protein concentrate (MPC) from which it is possible to obtain high-protein nutrition (HPN) food forms [69]. It is well known that MPCs produce HPN bars commonly commercialized, however one of the main issues is represented by the changes in their texture that could occur during their storage [69]. Banach [69] recorded also confocal micrographs of high-protein nutrition (HPN) bars formulated with unmodified spray dried milk protein concentrate with 80% protein (MPC80), 4h toasted MPC80 at 75 °C and 110 °C, and extruded MPC80 at die temperature of 65 °C and 120 °C. Confocal micrographs showed that changes in MPC80 occurred during the storage [69]. The changes in HPN bars are also visible in the images of the model high-protein nutrition (HPN) bars after storage for 6 and 29 weeks (Week 0 represents t0) at 22 °C or 32 °C. MPC80 represents spray dried milk protein concentrate with 80% protein, while E105 and E116 are two extrudates of HPN formulated with unmodified spray dried milk protein concentrate with 80% protein (MPC80) with die-end melt temperatures of 105 and 116°C respectively [67]. The protein ingredient, storage temperature, and storage time all had significant effects (P  0.05). However, some observations show that MPCs are wettable, but it appears that these high-protein powders are not wettable. On the basis of these experiments, Banach [69] demonstrated that by ADCA method it is possible to evaluate the water absorption speed of the extruded MPC and its modifications in the 0–60s

230 Progress in Adhesion and Adhesives, Volume 4

range and the instant at which the volume of water droplet changes. This demonstrated that the ST decreases when the protein is dissolved in the water droplet. For this reason, the CA method can be considered as an important tool for the control of MPC quality. MPC with high protein content (hydrophobic) is less wettable with respect to the same containing high concentration of lactose (hydrophilic). This explains why the extruded MPC powder particles easily hydrate during the process of manufacture of HPN bars. Consequently, a fast decrease of CA occurred. In this context, Banach [69] demonstrated that the CA approach for the study of MPCs is particularly important because it is able to evaluate the integrity of the structure of HPN bars which could be modified when water molecules separate from the protein during storage. By CA analysis it is possible to show also that the extrusion process of MPCs causes a decrease in protein solubility and surface hydrophobicity, improving the ability to interact with water and, consequently, a rapid powder hydration occurs during the production of HPN bars. This allows a good and constant hydration of the protein during storage. More recently Nushtaeva focused on the role of the emulsifiers in medicine and gastronomy fields [70]. It is well known that some natural insoluble powders, such as mustard and ground cinnamon, can be used as solid emulsifiers for O/W emulsions that are formed by aqueous dispersions of food powders and olive oil. In the work of Nushtaeva [70] the mustard and cinnamon powders demonstrated their emulsifying property, and in particular the ground cinnamon ensures high stability of food emulsions in general with antioxidant and antimicrobial properties also. As reported by Nushtaeva, the evaluation of these properties can be performed by investigating the hydrophile/lipophile ratio (HLR) of the powders. Nushtaeva demonstrated that HLR can be expressed by using the CA values (CA/HLR combined approach). The hydrophile-lipophile ratio for solid particles is the analog of hydrophilicity/hydrophobicity coefficient and the hydrophile-lipophile ratio (HLR) can be expressed via the heats of wetting (Equations 21 and 22).

HLRwt

HLR

Hw Ho

1 cos 1 cos

(21) 2

ow

(22)

ow

where ΔH is the heat of wetting of 1g of solid phase, ΔHw is the heat of water wetting of the powder, ΔHo is the heat of oil wetting, ow is the contact angle measured on the surface of O/W emulsions stabilized by powder particles such as natural milk powder, ground ginger, mustard powder, ground cinnamon, potato starch and ground nutmeg. The HLR of a powder is strongly linked to the hydrophilicity/hydrophobicity coefficient ( ) that is based on the heat of wetting of solid phase and characterizes both water and oil wetted powders. By employing CA measurements for HLR determination, Nushtaeva demonstrated that mustard and cinnamon powders have a high capacity to protect the O/W food emulsions from coalescence and sedimentation. Another interest in the food industry is the quality of the reconstitution process of a beverage from a dehydrated powder. As reported by Dupas et al.

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 231

[71], in this process many physical mechanisms occur which make it very difficult to produce a homogeneous drink even under the best conditions. In this case the particle size distribution and the role of CA measurements for the prediction of the wettability behaviour of food powders assume great importance. Considering CA and porosity as the main parameters for the evaluation of the particle size distribution of a dehydrated powder, it is possible to calculate the fraction of particles that float to improve the wettability of the powder. The CA measurements can be done at the interface between water and tablets obtained from high compaction of powders of milk, cocoa, chocolate and coffee for beverages production. Following this method, the wettability of the food powders for beverages can be determined by measuring the percentage of the powder that remains at l/v interface after 45s.

5.3.2

Liquid Foods and Beverages

Liquid foods are represented generally as food suspensions containing lipid coated particles. In this case Wang et al. [72] demonstrated that the CA measurements can evaluate the wettability of lipid surfaces as well as the influence of pH, salt concentration, protein concentration, surfactant type and concentration [72] (Figure 32a–32d). The glass slides were immersed in surfactant solutions for 16 h and the measurement were made after 15min [72].

Tween 20 100 80 60 40 20

100 80 60 40 20

0 0.1

0.5 1 Concentration (g/L)

5

0

(b)

120

80 60 40 20

0.1

0.5 1 Concentration (g/L)

5

10

0.5 g/L whey protein solution

120

100

100 80 60 40 20 0

0 Control Tween 20 Tween 40 Tween 60 Tween 80

(c)

10

Contact angle (deg)

Minimum contact angle (deg)

0 0

(a)

Whey protein

120 Contact angle (deg)

Contact angle (deg)

120

Surfactants

1.8

whey

(d)

2.5

3.5

4.5

5

5.5

65

pH

Figure 5.32 (a) Contact angle (degrees) of water on lipid coated glass slides treated with different concentrations of Tween 20 solutions (Adapted from [72]). (b) Contact angles (degrees) of water on lipid coated glass slides treated with whey protein (Adapted from [72]). (c) Minimum contact angles (degrees) of water on lipid coated glass slides treated with different surfactant solutions (Adapted from [72]). (d) Contact angles (degrees) of water on lipid coated slides treated with 0.5g/L whey protein solution (50g/L citrate buffer) (Adapted from [72]).

232 Progress in Adhesion and Adhesives, Volume 4

Generally, the increase in the concentration of the emulsifier causes a decrease in CA measured for a drop of water on lipid surface, while low values of pH are linked to high values of CA and vice versa. Wang et al. [72] showed that the very low CA obtained in the presence of high concentration of emulsifier with high pH is due to the effect of the ionic strength (e.g., high concentration of Na citrate) independent of the concentration of protein. As a consequence, high concentration of protein and high ionic strength increase the wettability of hydrophobic surfaces (e.g., for protein 5 g/L and NaCl 0.5mol/L) with CA about 10.7 ± 2.7. The best emulsifiers present in nature are the proteins of whey that have shown to be very efficient at low concentration in water (0.5g/L). In the study of food suspensions, the CA value evaluates the surface pressure (SP) c at the air–water interface also by the Langmuir equation for planar surfaces obtained by using the tensiometer (goniometer) (equation 23) [70].

cos )2

LA (1 c

2 3

(23)

where c is the critical SP, LA is the ST of liquid (water), and is the CA. These studies of Wang et al. [72] on food suspensions were of fundamental importance because many micronutrients are added to liquid foods including beverages and the role of the liquid foods as suspensions, makes the intake of micronutrients fundamental for the iron fortification in food suspensions depending on its encapsulation. Wang et al. [72] demonstrated that the adjustment of parameters such as pH, ionic strength and emulsifier concentration based on the CA method leads to obtain more stable and also with good flavor food suspensions with micronutrients. In the field of foods, the tensiometric characterization of liquid foods can be used for the determination of the quality of food oils also. One of the most interesting results obtained from CA analysis of food oils was the characterization of their polar component [73]. As reported by Michalski et al. [73], the food oils tested and the CA values converted into SFE by the Owens-Wendt model showed some degree of PC; however, in general it is considered negligible because it is only 5% of the dispersion component. The measurement of the interfacial tension between apolar and polar liquids was performed by the DuNoüy ring method. The dL component of polar liquids was calculated using Equation (24): d L

(

a

aL )

L

4

(24)

a

where is the the surface tension of apolar liquid (a) and L is the surface tension polar liquid (L). The device for CA measurements used by Michalski et al. [73] is shown in Figure 33. The measurements of CAs were useful for the determination of DC ( doil) and PC ( poil) of food oils (Equations 25 and 26). The dispersion component of oils was determined from the measurements of CAs on the surface of Parafilm®. The polar component was obtained from the total surface free energy.

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 233

oil (cos oil , Parafilm

d oil

4 p oil

1)

2

(25)

Parafilm

oil

d oil

(26)

This residual PC ( poil) is due mainly to the presence of fatty acids that are commonly present in similar concentrations in food oils from olive, sunflower, and soybean. The effect of surface polarity, calculated by polar, acid-base, or hydrogen bonding ST components, on oil experimental adhesion was determined through van Oss, Chaudhury and Good; and Owens and Wendt equations that were also used to investigate the adhesion of inks to polymer surfaces (Equation 27) [73].

WA

2

d d L S

2

p p L S

2

h h L S

(27)

where h is the surface free energy component due to hydrogen bonding. These studies led to the evaluation of the effects of surface hydrophilicity on oil experimental adhesion [73]. There are small differences between the surface tensions of the different oils, and the differences between them can be evaluated by considering parameters such as viscosity, nature of the components, and physico-chemical properties of the test solids used for CA measurements. On the other hand, the studies of the sensorial perception of liquid foods in relation to the oral perception due to the mouth friction have recently found a great interest in the food industry. For this reason, some studies were recently performed on the adhesion and spreading of food emulsion droplets on solid surfaces in relation to their capability to decrease the friction on the oral surface [74]. The main interactions occurring

Microsyringe Camera

Closed chamber

Screen viewing

Light source

Light filter

Rotating sampler holder

Computer analysis

Figure 5.33 Device for contact angle measurement by image analysis (Adapted from [73]).

234 Progress in Adhesion and Adhesives, Volume 4

between the droplets and solid surfaces are electrostatic, steric and hydrophobic because they determine the adhesion and spreading of the food emulsion droplets. In particular, the hydrophobic interactions are crucial for developing stable emulsions capable of spreading on oral surfaces due to a low friction phenomenon at the interface. As reported by Dresselhuis [74], in the case of lipids, the increase in drop spreading is related to an increase in the perception of fats at tongue level. Generally, the increase in adhesion is due to the presence of protein-poor emulsion droplets that adhere better to hydrophobic surfaces than protein-rich droplets. Dresselhuis [74] demonstrated that it was due to the presence of various colloidal forces causing a decrease in electrostatic and steric repulsions. In particular, the distinction between adhesion and spreading of emulsion droplets is commonly used for investigations on the correlation between the degree of aggregation and the droplet size. Normally, the CA analysis applied in this kind of studies is performed with light microscopy (LM) and flow cell (FC) analysis by LM/FC/CA combined approach [74]. In this case, the CA method analyzes the spreading of emulsion droplets and the change in their diameter with time. The LM/FC/CA combined approach can distinguish between non-adhered, adhered and spread emulsion droplets. Dresselhuis [74] first considered the pig tongue as an ideal tissue for the investigation of fat perception. The CAs were measured on slightly moist pig tongue surface with water droplet after 2 sec, after 5 sec, dry tongue surface, dry tongue after 1.5 min and dry tongue after 3 min [74]. Dresselhuis [74] explained that the tongue can be considered as a surface in need of lubrication and requires protection against external agents, e.g., acids. A hydrophobic surface in combination with coverage by mucins makes possible the lubrification and protection function [74]. Dresselhuis [74] demonstrated the friction forces between the tongue and glass with the emulsion acting as a lubricant. This can be measured in a reproducible manner despite the biological material originating from different pigs tongues (measurement performed 5 times with different pieces of tongue) [74]. The friction force (N) was investigated on an emulsion such as 1 wt% of Whey Protein Isolate (WPI) and 40 wt% of sunflower oil sheared between tongue and glass [74]. Dresselhuis [74] measured also the friction force (N) when shearing the same kind of emulsion between glass and poly(dimethylsiloxane) (PDMS) rubber [74]. On the basis of these observations, Dresselhuis [74] demonstrated that a knowledge of the characteristics of pig’s tongue enables modification of PDMS surfaces in order to make them more closely mimic a tongue surface. PDMS is an ideal surface for this since it is easily moulded in various shapes, allowing control of the surface roughness. The decrease in CA means an increase in spreading of protein-poor emulsion droplet which is very efficient as a lubricant [74]. As a consequence of these properties, the protein-poor emulsions show high sensorial potential in fats perception [74].

5.3.3

Food Packaging

The increase in environmental, economic and safety concerns caused by the traditional food packaging materials has prompted in recent years to partially substitute the petrochemical-based polymers such as polystyrene with biodegradable ones [75]. An example of a biodegradable thermoplastic polymer for food packaging is the poly(lactic acid) (PLA)

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 235

that can be obtained from renewable sources. The properties of PLA and their possible modifications by using various modifiers for blending, copolymerizing are fundamental for the production of PLA films, wrappings, laminates and containers such as bottles and cups. As reported by Jamshidian et al. [75], the PLA possesses antimicrobial and antioxidant characteristics that are fundamental requisites for the conservation of foods. The nanotechnology has helped the development of new PLA nanocomposites combined with nanomaterials which has improved the quality and functions of PLA, thus decreasing concentrations and prices. Jamshidian et al. [75] demonstrated that the PLA with added antioxidants can be evaluated by XPS/AFM/RAMAN/FTIR/TEM combined approach in order to verify the influence of antioxidants on the surface of PLA [75]. In particular Jamshidian et al. [75] recorded also FTIR spectra of PLA films, PLA-antioxidant films and pure PLA. TEM analyses were performed on PLA films containing antioxidants. The surface energy of a polymer influences its adhesion property [76] and it is important for processes such as printing and heat sealing in the packaging field. PLA is biocompatible and is widely used in the medical field. Due to its low wettability and low surface energy, PLA influences the proliferation of cells. Considering that the SFE of a polymer is an important factor that can influence its adhesion property, the CA measurements and the determination of SFE, DC, and PC of PLA films with and without antioxidants becomes a useful tool for the characterization of biodegradable packaging materials [75]. Synthetic phenolic antioxidants (SPAs) including butylated hydroxyanisole (BHA), butylated hydroxytoluene (BHT), propyl gallate (PG), and tert-butylhydroquinone (TBHQ) were added (1%, w/w) to a poly (lactic acid) (PLA) film [75]. The contact angle of the PLA film with 1% SPA decreases from approximately 75.3° to the lowest value 44.5° for PLA-AP (recto), 69.4° for PLA-AP (verso), 70.5° for TBHQ (recto) where recto is the surface in contact with air, and verso is the surface in contact with Teflon. As regards other antioxidants, the modification is negligible. In the case of PLA, the CA level increases after the addition of antioxidants because of their hydrophobic nature and the variation depends on the concentration of antioxidant in the PLA film, and its distribution on the surface and throughout the polymer network. Jamshidian et al. [75] found a slight increase in CA values measured on PLA surface with added antioxidants. Jamshidian et al. [75] also showed that the increase in the number of hydrophobic groups present on the PLA surface improves the printing processes. One of the first parameters that characterizes the quality of foods and the packaging materials is the absence of various microbial classes (Escherichia coli, Salmonella typhimurium, Staphylococcus aureus, Listeria monocytogenes, Enterococcus faecalis) which improves the efficiency with respect to the conservation of these materials. The novel Non-Thermal Plasma (NTP) technology has shown high potential for decontamination in the food industry and it can be used for the surface treatment of raw products, dry food powders such as dried milk, herbs, spices and packaging materials for the dry disinfection of food surfaces such as meat, poultry, fish, fresh vegetables and sprouted seeds [77]. The NTP technology involves an intense bombardment of micro-organisms with radicals causing deep biological surface lesions. As reported by Tiwari [77] the damage to the surfaces of microorganisms is due to the action of the electrostatic forces occurring from the accumulation of charges on the surfaces of cell membranes. In this context, the CA measurements on food surfaces is a useful tool for

236 Progress in Adhesion and Adhesives, Volume 4

the evaluation of NTP performance [77]. As an example, the surface of lettuce shows an increase in its wettability after plasma treatment and the decontamination can be confirmed by FTIR and SEM investigations [77].

5.4

Summary

This review has focused on the applications of CA in the study of pharmaceuticals and foods. In this context, we evidence new perspectives in the applications of the CA method in the novel “pharma-food” combined field. In our opinion, the application of the tensiometric technique in this novel field is in its beginning stage because the tensiometric investigations actually are performed separately in pharmaceutical and foods fields. Actually, the CA investigations are still sparse and do not take into account yet the concept of an Integrated Analytical Approach (IAA) by which it is possible to obtain a broad view of a material. The IAA approach can give more information about the elements-strucrure-surface correlations that govern all natural and artificial materials. The studies reported here demonstrated that the CA method is still considered as an analytical support in the pharmaceutical and food sectors in which other traditional techniques are more widely used. The other critical point is that the development of new tensiometric methods based on mathematical models such as Owens, Wendt, Rabel&Kaelble; Wu; van Oss, Chaudhury and Good; etc. for the characterization of surface free energy (SFE; mJ/m2) and surface tension (ST; mN/m) of pharmaceutical and food products is quite absent. This could be caused by the fact that although there are several methods for determination of surface free energy of a liquid (surface tension of liquid), the experimental determination of surface free energy of a solid is still not a fully solved problem [78–79]. In particular, Della Volpe and Siboni [80] have focused on many misleading concepts presented in the scientific literature such as confusion between sliding angle and contact angle hysteresis, inaccurate definition of superhydrophobicity, uncriticized occurrence of meaningless negative contact angles, the so-called “petal effect,” or the lotus leaf described as a chemically hydrophilic surface with a superhydrophobic behaviour attributable to only its geometrical structure. On the basis of the results of this first review on recent pharma-food sector, we can conclude that the CA method can be a non-invasive analytical approach capable to lead the investigations towards the correlation between the pharmaceuticals and foods. This could help in developing an IAA based common investigation protocol for the non-invasive determination of the surface free energy and surface tension of products. The same concept can be extended to a non-invasive and rapid evaluation of the quality of the industrial processes in order to improve these, with the aim to increase the quality of the final product. For these reasons, we think that we are at the beginning for the application of IAA approach in pharma-food sector in which the tensiometric approach could play a crucial role for solving the several issues in pharmaceutical and food sectors.

Acknowledgement The authors wish to express special thanks to Dr. Kash Mittal for his many valuable comments/suggestions.

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 237

References 1. T. Young, An essay on the cohesion of fluids. Phil. Trans. R. Soc. Lond. 95, 65–87 (1805) 2. T. Tate, On the magnitude of a drop of liquid formed under different circumstances. Phil. Mag. 27, 176–180 (1864). 3. L.A. Wilhelmy, Ueber die abhängigkeit der capillaritäts-constanten des alkohols von substanz und gestalt des benetzten fasten körpers. Annals Phys. Chem. 119, 177–217 (1863). 4. F. Bashforth and J. C. Adams, An Attempt to Test the Theories of Capillary Action by Comparing the Theoretical and Measured Forms of Drops of Fluid, Cambridge University Press, Cambridge, UK, 1883, With an Explanation of the Method of Integration Employed in Constructing the Tables which Give the Theoretical Form of Such Drops, by J. C. Adams. 5. A. Dupre, Theorie Mechanique de la Chaleur. Gauthier-Villars, Paris, p 369 (1869). 6. D. Berthelot, Sur le mélange des gaz. Comptes Rendus Hebdomadaires Séances Académie Sciences. 126, 1803–1855 (1898). 7. I. Langmuir, The constitution and fundamental properties of solids and liquids. II. Liquids. J. Am. Chem. Soc. 39, 1848–1906 (1917). 8. N. K. Adam and G. Jessop, Contact angles and polarity of solid surfaces J. Chem. Soc. Trans. 127, 1863–1868 (1925). 9. L. Mack, The determination of contact angles from measurements of dimensions of small bubbles and drops I: The spheroidal segment method for acute angles and II: The sessile drop method for obtuse angles.  J. Phys. Chem. 40, 159–167 (1936). 10. R.N. Wenzel, Resistance of solid surfaces to wetting by water. Ind. Eng. Chem. 28,988–994 (1936). 11. E. W. Washburn. The dynamics of capillary flow. Phys. Rev. 18, 273–283 (1921). 12. F. E. Bartell and C.E. Whitney, Adhesion tension III. J. Phys. Chem. 36, 3115–3126 (1932). 13. J.J. Bikermann, A method of measuring contact angles. Ind. Eng. Chem. Anal. Ed. 13, 443–444 (1941). 14. F. M. Fowkes and W. D. Harkins, The state of monolayers adsorbed at the interface solid—aqueous solution. J. Am. Chem. Soc. 62, 3377–3386 (1940). 15. H.L. Ritter and L.C. Drake. Pore size distribution in porous materials. Ind. Eng. Chem. Anal. Ed.  17, 782–786 (1945).   16. S. Baxter and A.B.D. Cassie, The water repellency of fabrics and a new water repellency test. J. Textile Inst. 36, 67–90 (1945). 17. A. Carré and K.L. Mittal (Eds.), Superhydrophobic Surfaces, CRC Press, Boca Raton, FL (2009). 18. G. McDougall and C. Ockrent, Surface energy relations in liquid/solid systems I. The adhesion of liquids to solids and a new method of determining the surface tension of liquids. Proc. Royal Soc. A. 180, 151–173 (1942). 19. V.R. Gray, Surface aspects of wetting and adhesion. Chem. Ind. 23, 969–977 (1965). 20. W.C. Bigelow, D.L. Pickett, and W.A. Zisman, Oleophobic monolayers: I. Films adsorbed from solution in non-polar liquids. J. Colloid Sci. 1, 513–538 (1946). 21. W.M. Rohsenow, A method of correlating heat transfer data for surface boiling of liquids. Trans. ASME. 74, 969–975 (1952). 22. A. H. Ellison, H. W. Fox, and W. A. Zisman, Wetting of fluorinated solids by hydrogen bonding liquids. J. Phys. Chem. 57, 622–627 (1953). 23. H. W. Fox and W. A. Zisman, The spreading of liquids on low energy surfaces. I. Polytetrafluoroethylene. J. Colloid Sci. 5, 514–531 (1950). 24. H. W. Fox and W. A. Zisman, The spreading of liquids on low-energy surfaces. II. Modified tetrafluoroethylene polymers. J. Colloid Sci. 7, 109–121 (1952).

238 Progress in Adhesion and Adhesives, Volume 4

25. H. W. Fox and W. A. Zisman, The spreading of liquids on low-energy surfaces. III. Hydrocarbon surfaces. J. Colloid Sci. 7, 428–442 (1952). 26. R. E. Johnson Jr., Conflicts between Gibbsian thermodynamics and recent treatments of interfacial energies in solid-liquid-vapor systems. J. Phys. Chem. 63, 1655–1658 (1959). 27. E. G. Shafrin and W. A. Zisman, Constitutive relations in the wetting of low energy surfaces and the theory of the retraction method of preparing monolayers. J. Phys. Chem. 64, 519–524 (1960). 28. F. M Fowkes, Determination of interfacial tensions, contact angles, and dispersion forces in surfaces by assuming additivity of intermolecular interactions in surfaces. J.  Phys. Chem. 66, 382–390 (1962). 29. L. A. Girifalco and R. J. Good, A theory for the estimation of surface and interfacial energies. I. Derivation and application to interfacial tension, J. Phys. Chem. 61, 904–909 (1957). 30. R. E. Johnson Jr and R. H. Dettre, Contact angle hysteresis. III. Study of an idealized heterogeneous surface. J. Phys. Chem. 68, 1744–1750 (1964). 31. W. A. Zisman, The solid/liquid interface—An essential and active frontier of science, Adv. Chem. Sci. 87, 1–9 (1968). 32. D. E. McIntyre, A study of dynamic wettability on a hydrophobic surface. Doctor Dissertation, The Institute of Paper Chemistry, Appleton, Wisconsin, USA (1969). 33. W. J. Fenrick, Simple tangentometer. Rev. Sci. Instrum. 35, 1616–1617 (1964). 34. T. Fort Jr and H.T. Patterson, A simple method for measuring solid-liquid contact angles. J. Colloid Sci. 18, 217–222 (1963). 35. T. Karbowiak, F. Debeaufort, and A. Voilley, Importance of surface tension characterization for food, pharmaceutical and packaging products: A review. Critical Rev. Food Sci. Nutrition. 46, 391–407 (2006). 36. E. Nowak, G. Combes, E. H. Stitt, A. and A. W. Pacek, A comparison of contact angle measurement techniques applied to highly porous catalyst supports. Powder Technol. 233, 52–64 (2013). 37. F. Visioli, Pharma and nutrition: Crossing the Rubicon. Pharma Nutrition 1, 9 (2013). 38. D. Schebarchov and S. C. Hendy, Uptake and withdrawal of droplets from carbon nanotubes. Nanoscale 3, 134–141 (2011). 39. C. F. Lerk, A.J.  Schoonen, and T. Fell, Contact angles and wetting of pharmaceutical powders. J. Pharm. Sci. 65, 843-847 (1976). 40. J. O. Marston, J. E. Sprittles, Y. Zhu, E. Q. Li, I. U. Vakarelski, and S.T. Thoroddsen, Drop spreading and penetration into pre-wetted powders. Powder Technol. 239, 128–136 (2013). 41. Y. Han, G. Drazer, G. Callegari, and Z. S. Shojaei,  Droplet penetration method as a wettability test for pharmaceutical powders. Ph.D. dissertation, Rutgers University, New Brunswick, New Jersey, USA (2017). 42. U. Teipel and I. Mikonsaari, Determining contact angles of powders by liquid penetration. Particle Particle Syst. Charact. 21, 255–260 (2004). 43. M. Thakker, V. Karde, D. O Shah, P. Shukla, and C. Ghoroi, Wettability measurement apparatus for porous material using the modified Washburn method. Measurement. Sci. Technol. 24, 1–8 (2013). 44. E. Chibowski, L. Holysz and A. Szczes, Wettability of powders, in: Adhesion in Pharmaceutical, Biomedical, and Dental Fields, K.L. Mittal and F.M. Etzler (Eds.) pp.  23–49, Wiley-Scrivener, Beverly, MA (2017). 45. D. Zhang, J. H. Flory, S. Panmai, U.Batra, and J. Kaufman, Wettability of pharmaceutical solids: Its measurement and influence on wet granulation. Colloids Surfaces A. 206, 547–554 (2002).

Contact Angle Measurements and Applications in Pharmaceuticals and Foods 239

46. N. M. Ahfat, G. Buckton, R. Burrows, and M. D. Ticehurst, An exploration of inter-relationships between contact angle, inverse phase gas chromatography and triboelectric charging data. Eur. J. Pharm. Sci. 9, 271–276 (2000). 47. J. W. Dove, G. Buckton, and C. Doherty, A comparison of two contact angle measurement methods and inverse gas chromatography to assess the surface energies of theophylline and caffeine. Int. J. Pharm. 138, 199–206 (1996). 48. C. J. van Oss, M.K. Chaudhury, and R.J. Good, Interfacial Lifshitz-van der Waals and polar interactions in macroscopic systems. Chem. Rev. 88, 927–941 (1988). 49. D. R. Williams, Particle engineering in pharmaceutical solids processing: Surface energy considerations. Current Pharmaceutical Design 21, 2677–2694 (2015). 50. D. K. Owens, and R. C., Wendt, Estimation of the surface energy of polymers. J. Appl. Polym. Sci. 13, 1741–1747 (1969). 51. T. Ghafourian and P. A. Khan, Quantitative relationships between contact angles of pharmaceutical powders and computable molecular descriptors. Pharm. Sci. 3, 69–77 (2000). 52. T. Phaechamud and C. Savedkairop, Contact angle and surface tension of some solvents used in pharmaceuticals. Res. J. Pharm. Biol. Chem. Sci. 3, 513–529 (2012). 53. C. J. Yarce, D. Pineda, C. E. Correa and C. H. Salamanca, Relationship between surface properties and in vitro drug release from a compressed matrix containing an amphiphilic polymer material. Pharmaceuticals 9, 34–54 (2016). 54. J.R. Chokshi, Z. Hossein, K. S. Harpreet, H. S. Navnit, and A. M. Waseem, Improving the dissolution rate of poorly water soluble drug by solid dispersion and solid solution—pros and cons. Drug Delivery 14, 33–45 ( 2007). 55. S. Verma, and S. V. Rudraraju, Wetting kinetics: An alternative approach towards understanding the enhanced dissolution rate for amorphous solid dispersion of a poorly soluble drug. AAPS Pharm. Sci. Technol.16, 1079–1090 (2015). 56. K. Min-Soo, K. Jeong-Soo, and H. Sung-Joo, Enhancement of wettability and dissolution properties of Cilostazol using the supercritical antisolvent process: Effect of various additives. Chem. Pharm. Bull. 58 230—233 (2010). 57. I. Khan, M.I. Khan and U. Khan; Liquisolid technology: An emerging and advanced technique for enhancing solubilization. PharmaTutor 2, 31–41 (2014). 58. R. M. Patel, Parental suspension : An overview. Int. J. Current Pharm. Res. 2, 5–13 (2010). 59. A. Trivino and H. Chauhan, Drug-excipient compatibility for the formulation development of solid lipid nanoparticles. Amer. Pharmaceutical Review 18, 2–6 (2015). 60. I. O. De Solorzano, L. Uson, A. Larrea, M. Miana, V. Sebastian and M. Arruebo, Continuous synthesis of drug-loaded nanoparticles using microchannel emulsification and numerical modeling: Effect of passive mixing. Int. J. Nanomedicine 11, 3397–3416 (2016). 61. A. Nakashima, T. Izumi, K. Ohya, K. Kondo and T. Niwa, Design of highly dispersible PLGA microparticles in aqueous fluid for the development of long-acting release injectables Chem. Pharm. Bull. 65, 157–165 (2017). 62. A. M. Asiri, H. M. Marwani, S. B. Khan, and T. J. Webster, Understanding greater cardiomyocyte functions on aligned compared to random carbon nanofbers in PLGA. Int. J. Nanomedicine 10, 89–96 (2015). 63. R. Andrade, O. Skurtys, and F. Osorio, Drop impact behavior on food using spray coating: Fundamentals and application. Food Res. Int. 54, 397–405 (2013). 64. T. S. Parreidt, M. Schmid and C. Hauser, Validation of a novel technique and evaluation of the surface free energy of food. Foods 6, 1–14 (2017).

240 Progress in Adhesion and Adhesives, Volume 4

65. M. D. Kurkuri, C. Randall, and D. Losic, New method of measuring the angle of repose of hard wheat grain. Intl. Proc. Chemeca, Wellington, New Zealand (2012). 66. S.  K. Reinke, K. Hauf, J.  Vieira, and S.  Palzer, Changes in contact angle providing evidence for surface alteration in multi-component solid foods. J. Phys. D: Appl. Phys. 48, 1–15 (2015) 67. A. Sharma, A. H. Jana, and R. S. Chavan, Functionality of milk powders and milk-based powders for end use applications—A review. Comprehensive Rev. Food Sci. Food Safety. 11, 518–529 (2012). 68. K. Górnicki, A. Kaleta, R. Winiczenko, A. Chojnacka and M. Janaszek, Some remarks on modelling of mass transfer kinetics during rehydration of dried fruits and vegetables, Mass Transfer. 16, 431–457 (2013). 69. J. C. Banach, Modifed milk protein concentrates in high-protein nutrition bars. PhD Dissertation, Iowa State University, Ames, Iowa, USA (2016). 70. A.V. Nushtaeva, Natural food-grade solid particles for emulsion stabilization. Colloids Surfaces A 504, 449–457 (2016). 71. J. Dupas, L. Forny, and M. Ramaioli, Powder wettability at a static air–water interface. J. Colloid Interface Sci. 448, 51–56 (2015). 72. Z. Wang, G. Narsimhan, and D. Kim, Characterization of the effect of food emulsifier on contact angle and dispersibility of lipid coated neutrally buoyant particles. Lebensmittel-Wissenschaft Technologie, 41, 1232–1238 (2008). 73. M. C. Michalski, S. Desobry, M. N. Pons and J. Hardy, Adhesion of edible oils to food contact surfaces. J. Amer. Oil Chem. Soc 75, 447–454 (1998). 74. D. M. Dresselhuis, The fate of fat: Tribology, adhesion and fat perception of food emulsions. Wageningen Universiteit, Nederland (2008). 75. M. Jamshidian, E. A. Tehrany, M. Imran, M. Jacquot and S. Desobry, Poly-Lactic Acid: Production, applications, nanocomposites, and release studies. Comprehensive Rev. Food Sci. Food Safety 9, 552–571 (2010). 76. K.L. Mittal, The role of the interface in adhesion phenomena. Polym. Eng. Sci. 17, 467–473 (1977). 77. B. K. Tiwari, Nonthermal plasma inactivation of food-borne pathogens. Food Eng. Rev. 3, 159–170 (2011). 78. F. M. Etzler, Characterization of surface free energies and surface chemistry of solids, in: Contact Angle, Wettability and Adhesion, Volume 3, K.L. Mittal (Ed.) pp. 219–264, CRC Press, Boca Raton, FL (2003) 79. F.M. Etzler, Determination of the surface free energy of solids: A critical review. Rev. Adhesion Adhesives 1, 3–45 (2013). 80. C. Della Volpe and S. Siboni, Use, abuse, misuse and proper use of contact angles: A critical review. Rev. Adhesion Adhesives 3, 365–385 (2015).

6 The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure to Oxygen or Ammonia Plasma: A Critical Review Jörg Friedrich Technical University Berlin, Fasanenstrasse 90, 10623 Berlin, Germany

Abstract Hydroxy (-OH) and primary amino groups (-NH2) are important as anchoring points at polymer surfaces for covalent coupling of molecules used in bio-medicine or components in polymer composites. O2 and NH3 plasmas or the coating of polyolefin surfaces with thin films of plasma polymers from allyl alcohol and allylamine are most often used for generation of OH and NH2 groups. The analysis of polyolefin surfaces exposed to oxygen and noble gas plasma treatments has shown very similar spectra of O-functional groups when the plasma treated samples were post-plasma exposed to air. Therefore, it was not surprising that the variety of O-functional groups formed by the two types of plasmas corresponds well to that of chemical auto-oxidation. This oxidation proceeds via formation of radicals, attachment of molecular oxygen, generation of peroxy radicals, and decomposition of hydroperoxide groups to many types of O-functional groups, among them only a few OH groups are formed. Coating of polyolefins with allyl alcohol plasma polymer leads to a surface with higher concentration of OH groups. It is assumed that many OH groups of the monomer survive the plasma polymerization process and additional OH groups are formed by post-plasma auto-oxidation on exposure to air. Ammonia plasma treatment produces also a low yield in desired NH2 groups at a polyolefin surface, due to unfavorable thermodynamics. Moreover, several side reactions are observed, such as hydrogenation at the topmost surface as well as in subjacent layers, dehydrogenation by UV irradiation, polyene formation, crosslinking and post-plasma extensive auto-oxidation on exposure to air in deeper layers. Allylamine plasma polymerization is characterized by a strong loss in NH2 groups of allylamine and extensive side-reactions, such as dehydrogenation to imine and nitrile groups ( CH2 NH2 CH NH C N). The auto-oxidation of plasma polymer introduces many O-functional groups.

Corresponding author: [email protected]

K.L. Mittal (ed.) Progress in Adhesion and Adhesives, Volume 4 (241–314) © 2019 Scrivener Publishing LLC

241

242 Progress in Adhesion and Adhesives, Volume 4

Yield, selectivity, advantages, limitations and drawbacks of the above-mentioned processes for polyolefin surface modification with OH or NH2 groups are discussed in some detail with respect to the mechanism, kinetics and thermodynamics of reactions. In conclusion, introduction of functional groups into the polymer surface is always a compromise between attachment of different types of functional groups and simultaneous degradation and decomposition of the surface region of polymer substrates with weakening of its mechanical strength. Keywords: Auto-oxidation, OH groups, NH2 groups, thermodynamics, NH3 plasma, O2 plasma, plasma polymer layers

6.1

Introduction

6.1.1 Reasons for Polyolefin Surface Functionalization Polyolefins are the most often used polymers because of their excellent chemical, physical and processing properties. They represent linear or slightly branched long-chain alkanes consisting of carbon and hydrogen. This absence of polar groups, the partial crystallinity, and the protection of the C-C backbone from the direct attack of plasma particle bombardment of the covalently bonded hydrogen atoms, forming a protecting shield or cloud of hydrogen around the polyethylene backbone of methylene groups, produce the extraordinary chemical inertness. Such protective behavior of hydrogen atoms of the backbone is also known from chemical chlorination of poly(vinyl chloride) to polymer products with increased Cl concentration without degradation of poly(vinyl chloride) [1]. On the other hand, the inertness of polyolefins, also caused by the absence of any polar groups, is often a disadvantage. Thus, any interaction or bonding with other materials is problematic, which is an important drawback for using polyolefins in composite materials. Therefore, chemical bonding of coating molecules, adhesion-promoters, metal atoms, sensor or biomolecules is not possible. This chemical inertness must be overcome by introduction of functional groups realized by variation of structure using metallocene catalysts, copolymerization, coating, oxidation or mechanical pretreatment. Oxidation of polyolefins and introduction of O-functional groups is the only way which is thermodynamically favored as seen from polyolefin or paraffin combustion with considerable heat release. However, a broad variety of oxygen-containing groups are formed as found by Langenbeck and Pritzkow, and others [2–4]. Oxidations of polyolefins by oxygen or fluorine are strong exothermic reactions and proceed as chain-reaction with release of high amount of reaction enthalpy. The exothermic reaction is a significant reason for the rapid self-acceleration of oxidation. Therefore, such thermally initiated oxidation has a low selectivity regarding the oxidative attack to CH, CH2 or CH3 groups. Depending on oxidation conditions, fatty acids, alcohols, aldehydes, ketones, acids, esters, lactones, peroxides, hydroperoxides, CO2 and H2O are the products, i.e. they are formed during the (auto-) oxidation process. However, low-pressure plasma oxidation may allow tuning the oxidation process at polyolefin surfaces.

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

243

However, the “Holy Grail” [5] of interface design in polymer composites is the formation of covalent bonds between the polyolefin surface and the coating as schematically presented in Figure 1. In this scheme, the true diversity of functional groups was not considered [6]. As is evident from Figure 1 the precondition of equipping the surface with monosort OH or NH2 groups on exposure to O2/allyl alcohol plasma or NH3/allylamine plasma is not fulfilled in practice. Plasma-chemical followed by wet-chemical processing allows producing monosort functional groups as was discussed in a foregoing review [7]. In this review, basics of formation of OH and NH2 groups at polyolefin surfaces on exposure to O2 or NH3 plasma are analyzed as well as the survival of OH and NH2 groups by deposition of thin plasma polymer layers using allyl alcohol or allylamine. Moreover, a few attempts shall be discussed to increase the yield in OH or NH2 groups using additives in the process [8, 9]. The additives can shift the thermodynamic equilibrium in the sense of the Le Chatelier-Braun principle (equilibrium law, “any change in status quo prompts an opposing reaction in the responding system”). The validity of this idea was checked to see if it was also working under plasma conditions [10]. Plasma applied to polymers offers some advantages, such as well controllable surface activation also of inert polyolefin surfaces, but it shows also a multitude of disadvantages, such as the nearly total decomposition of polyolefins in the near-surface layer (Table 1).

OH

OH OH

OH

O2 plasma

O

Covalent bond

O O

O

Polyolefin

NH2 NH2

NH2 NH2

NH3 plasma

NH

Covalent bond

NH NH

NH

Polyolefin

Chemically inert

Wettable Adhesion promotion Anchoring points

Covalent bonding of biomolecules, sensor molecules, metal deposits, coatings, etc.

Figure 6.1 Schematics of plasma pretreatment of polyolefin surfaces and the covalent bonding of an epoxy resin (grey) to hydroxy or primary amino groups.

244 Progress in Adhesion and Adhesives, Volume 4

Table 6.1 Some advantages and disadvantages of plasma processing applied to polymers. Advantages

Disadvantages

All types of chemical bonds in polymers can be broken

Energy in the plasma is much higher than the binding energies in polymer

Generated radicals render the surface reactive

High energy tail of electron energy distribution function exceeds many times the C-C and C-H binding energies

Plasma gas atoms or molecules can be incorporated into the surface accompanied by formation of functional groups

Plasma emits high-energy ultra-violet radiation which acts not only at the topmost surface but also in some micrometer depth

Functional groups can post-plasma react following rules of classic chemistry

Polymers are degraded and lose mechanical strength

By choice of plasma gas the type of introduced functional group can be roughly predetermined

Polymers are crosslinked by vacuum-UV radiation in near-surface layers

Temperature in the plasma as well as at polymer surface is low

Radicals are formed

Plasma treatment can be easily tuned

Radicals react post-plasma with oxygen from the air

 

Many different functional groups are generated

 

Monotype functional groups, as needed for the following chemical graft reactions, are very seldom generated

 

Original polymer structure is destroyed

 

Original orientation is eliminated and amorphous structure is generated

 

Polymers are etched physically as well as chemically in case of oxygen plasma

 

Etching of polymers to gaseous products depends on composition and structure

 

Hydrogen is abstracted and polyenes or trienes are generated

 

Sometimes aromatic structures are produced

 

Nitrile and hydrocyanic acid are produced in presence of nitrogen in the plasma or in the polymer

 

Hydrogen is absent in O2 plasma; it is needed for the formation of OH, CHO, COOH groups

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

6.1.2

245

Energetic Considerations, Thermodynamics and Probability of Reactions

Thermal load, oxidation, degradation, ageing, etc. can strongly deteriorate the advantageous properties of polymers. If a polymer is in contact with plasma, high energy doses are transferred to the polymer and are consumed in its topmost layer of a few or even tens of nanometer thickness, producing a high energy density capable of initiating all types of chemical reactions. Also organic polymers are not thermally conducting, i.e., the transferred energy from the plasma cannot be broadly distributed over a crystal lattice as with inorganics or metals. Therefore, this energy is focused on one or two layers of atoms at the topmost surface of polymers (Table 2). Such a local energy concentration promotes endothermic chemical reactions, such as preferred cracking of aromatic rings in benzene plasma polymerization. The electron energy distribution, depicted in Figure 2, representative of ordinary plasma produced by a low-pressure glow discharge, shows an average kinetic energy just in the range of chemical bonds (in this example Ekin 3 eV), i.e. most electrons have the needed kinetic energy to dissociate C-C and C-H bonds (ca. Ediss=3.5 eV). For such processes, the activation barrier must be exceeded (Figure 2). In case of vinyl or acrylic monomers, the plasma-initiated chain-growth polymerization needs activation energy of about 1 eV for initiation. Such plasma-initiated chain-growth polymerization is exothermic and produces additional reaction heat. However, a few properties of such plasma are not compatible with the sensitivity of polymers towards high energy exposure (Figure 2), such as:. The high energy tail of the electron energy distribution function with energies much higher than Ekin >10 eV exceeds significantly the binding energies in polyolefin polymers (3–4 eV) and is able to fragment monomers and polymers to atoms or small fragments (“Atomic Polymerization” [11, 12]). The energetically dominating emission of energy-rich vacuum ultra-violet (VUV) radiation is capable to produce some radiation defects in the near-surface layer (Figure 3). Energy is continuously supplied to the plasma from electrical network. Table 6.2 Rough comparison of approximate energies occurring as binding energies in polymers and as measured electron or ionization energies in low-pressure glow discharges. Polymer

Plasma

Binding energy

kJ/mol

eV

kJ/mol

eV

C-H, C-C

350–400

3.5–4

0–5000 300–6000

0–50 (3–6 average)

Electron energy Gas, Ion energy

Hydrogen bonds

20–50

0.2–0.5

5

0.05

Van der Waals

5–10

0.05–0.1

1000–2500

10–25

Energy

Ionization

246 Progress in Adhesion and Adhesives, Volume 4

Enthalpy

Hdiss-plasma=2648 kJ/mol Hformation=3458 kJ/mol Energy of bonds in the substrate

Hoxidation= –810 kJ/mol energy of bonds in CO2 and H2O

F(v)=dn / dv (a.u.)

Reaction coordinate

3 0

1

2

3

4

5

6

7

8

9

10 11 12 13 14

Kinetic energy (eV)

Figure 6.2 Energy distribution of electrons, gas atoms or molecules and ions in the low-pressure glow discharge in relation to dissociation energies of C-C and C-H bonds in polyolefins, inset: exothermic chain-growth polymerization and energy needed from plasma to overcome the activation barrier.

Plasma processes (15%)

Wall processes (20%) Vacuum-UV (VUV) 55%

UV-vis (10%)

Figure 6.3 Percentages of energy flow to elementary processes in the low-pressure plasma.

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

247

The following three plasma processes generate functional groups at polymer surfaces: oxygen, ammonia or other molecular gas plasmas with attachment of plasma gas species as functional groups onto the polymer surface noble gas plasma with post-plasma introduction of O-functional groups, and plasmas in gas, vapor or liquid phase with deposition of plasma polymers carrying functional groups in their structure. The energy consumed per molecule and the ionization and dissociation energies are the most important for reaction in plasma as well as at polymer surface. The following energetic conditions have to be considered for these processes: 1. To sustain the plasma state, oxygen must be continuously ionized. Kinetic energies needed for it are in the range of about 12–15 eV (1100–1400 kJ/mol), or ≥10.4 eV (≈1000 kJ/mol) for NH3 which are far from dissociation energies found in polymers. 2. Formation of trapped radicals in polymers on exposure to plasma. 3. For plasma polymerization, the higher the energy dose per molecule, the more the fragmentation of monomers or precursors occurs [13, 14]. At lower energy dose, the probability for the (partial) survival of monomer or precursor structures in the plasma polymer is increased because of lower density of high-energy species. 4. Classic chemical oxidation of polyolefins is strongly exothermic and, therefore, produces additional enthalpy. 5. Classic chemical chain-growth polymerization also produces additional enthalpy because of the exothermic character of chain-growth polymerization but needs some additional activation energy in the range of 1 eV (100 kJ/mol) (styrene). Thus, the situation is as follows: the chemical (radical, ionic) polymerization (chaingrowth polymerization) of classic vinyl monomers is exothermic, and additionally they consume about 10 to 100 eV or even more per molecule during their residence time in the low-pressure radio-frequency plasma before conversion into a plasma polymer layer [15, 16]. Thus, the chain-growth polymerization to high-molecular weight linear macromolecules is accompanied by monomer fragmentation and random recombination of fragments and atoms. Low molecular weight alkanes (aliphatic non-polymerizable compounds), such as n-hexane, are not polymerizable. In the plasma, hexane needs about 1000 eV for conversion of only 60% alkane precursors into plasma polymer as shown by calorimetric measurements [17] because the plasma polymerization of an aliphatic precursor requires the prior elimination of hydrogen. The residual 40% precursor has passed the plasma with and without fragmentation as gas products. Such high energy input of 1000 eV into a single n-hexane molecule, as is needed for its plasma polymerization, results in its almost complete dissociation/fragmentation into atoms and small fragments. A simple calculation shows that the dissociation of

248 Progress in Adhesion and Adhesives, Volume 4

all the bonds of an n-hexane molecule (5xC-C and 14xC-H) requires about 70 eV, but in reality 1000 eV kinetic energy is transferred from the plasma to the molecule. Such high energy excess indicates the absence of any selectivity. The end products of such processes are highly crosslinked plasma polymers (or even carbon or “diamondlike” carbon layers) and hydrogen. As mentioned before, Yasuda denoted such process as “Atomic Polymerization” and proposed the “Yasuda-factor” [14]. This factor (Y=W/FM, W=power input, F=flow rate, M=molecular weight of precursor) influences the preservation of precursor/monomer structure in the plasma polymer. Low Y produces more  structural retention but tends to form more low-molecular weight oligomers; high Y generates crosslinked network with insignificant structural retention. Electrons are accelerated much more strongly in the electrical field because of their very low mass in comparison to that of ions. The electron energy in the low-pressure glow discharge Ee is Ee=3/2kTe, where k is the Boltzmann constant and Te the electron temperature, and the electron velocities follow a Maxwell-Boltzmann distribution. Each eV is equivalent to 11,605 K. Thus, electrons have much higher kinetic energy than ions and gas molecules (Figure 2) [18]. Thus, collisions of electrons with atoms or molecules lead to excitation and emission of fluorescence radiation, ionization and dissociation (fragmentation), which is applicable for noble gas plasmas. Additionally, chemical processes can accompany plasma effects, as in case of oxygen plasma etching of polyolefins, which can exceed physical etching of argon plasma by about 5–10 times (Figure 4) [19]. Hydrogen, carbon dioxide and water as well as carbon monoxide in smaller percentages are the most found gaseous oxidation products of polyolefins. The fast electrons impact the polyolefin surface first, charging it negatively resulting in stronger attraction of positive ions. Thus, the plasma forms a steady-state transition

10–7

O2

O2 plasma

Ion current (A)

O2 flow Plasma ignition

O2 flow Plasma-off

10–8

H2O H H2 CO/(N2)

10–9 –20

CO2

–10

0

10 20 Reaction time (s)

30

40

50

Figure 6.4 Time dependence of gas evolution on exposure of polypropylene to the low-pressure oxygen plasma as determined by mass spectrometry.

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

249

layer (Debye layer) at the electrically insulating polymer surface with a characteristic bias potential (Figure 5). Excited neutrals are not affected and etch the polyolefin or form functional groups at polymer surface. Oxygen species are transported to the polymer surface through the Debye layer and in the opposite direction gaseous products of the etch process diffuse to the plasma phase. Plasma particle bombardment produces radicals at the topmost surface of a few nanometers [20]; ultra-violet radiation (resonance-fluorescence) generates also radicals below the topmost surface up to the micrometer or even millimeter range [19]. Such transport processes dilute the oxygen concentration above the polymer surface and reduce the etching. Thus, the released gaseous etch products (water, carbon dioxide or carbon monoxide) are transported through the Debye layer and dilute the plasma gas as well as decrease the electron temperature in the plasma because of their high cross section of electron capture. On the other hand, this effect can be compensated by high hydrogen release (crosslinking, formation of unsaturations) which leads to higher electron temperature because of the small size of hydrogen species and, therefore, the reduced probability of collisions which leads to higher acceleration of electrons [17, 20]. Etching on exposure to oxygen plasma proceeds linearly in the absence of structural inhomogeneities and is dependent on wattage, pressure, gas flow and composition of polymer. O-containing polymers tend to etch more rapidly with increasing O/C elemental ratio in their structure. Aliphatics and N-containing, crosslinked and aromatic polymers are more stable than O-rich polymers [21–25]. A steady-state is adjusted between continuous ingress of etching front into the polymer accompanied by the permanent renewal of O-functionalization at the etching front as well as the release of gaseous etching products as shown in Figures 4 and 6.

6.1.3

Processes on Molecular Level at Polyolefin Surface

Plasma exposure in general and oxygen plasma in particular with its strong chemical action can attack the two elemental bonds in polyolefins: C-C and C-H bonds. The following two processes should be considered on exposure to plasma: plasma-initiated reaction with excess of plasma energy in comparison to needed dissociation energy (plasma particle bombardment and irradiation by plasma UV), thus eliminating selectivity of reactions by dissociation of all kinds of bonds followed by random radical recombination frequently to unusual chemical products including products of endothermic reactions initiation of chemical reactions, for example, oxidation initiated by plasma species (oxygen, halogens) and auto-oxidation initiated by plasma-produced trapped radicals in the polymer (reaction sequence in oxygen plasma or oxygen atmosphere) or initiation of chemical chain-growth polymerization to linear polymer molecules. Chemical reactions are determined by thermodynamics and kinetics, i.e. by reaction enthalpy, entropy and activation barrier. General reactions of monomers or polymers are

250 Progress in Adhesion and Adhesives, Volume 4 Species density

Electrostatic potential V ni

V=0 Wall

Wall

Vwall

Gas concentrations O 2

n0=ne=ni

Plasma sheath

Wall ne

Plasma bulk

Plasma sheath

Plasma bulk

CO2, H2O, CO, H2

Plasma sheath

Plasma bulk

Figure 6.5 Debye sheath of oxygen plasma above an electrically insulating polyolefin surface and generation of (floating) electrostatic potential (V) (left), electron (ne) and ion (ni) density (middle) as well as gaseous exchange of released etching gases to plasma and diffusion of oxygen to the polymer surface as shown by concentrations schematically (right).

Oxygen plasma

Etched layer Surface

O

OH

O

OH

Modified layer

3-5 nm sampling depth of XPS

Polyene Crosslinked

Polyene Crosslinked

O

OH

Polyene Crosslinked

ca. 1500 nm sampling depth of ATR

Polymer bulk

0s

Polymer bulk

2s

Polymer bulk

4s

Polymer bulk

8s

Exposure time

Figure 6.6 Steady-state between etching and permanent renewal of O-functionalization of polymer surface on exposure to oxygen plasma.

determined by the Gibbs-Helmholtz equation: G= H–T S. The enthalpy can be approximately calculated by the Hess rule [26]. In Figure 7, two substitution reactions onto polyethylene chains are schematically shown. Using the example of UV/plasma-initiated bromination, the same number of reactants and products are formed, i.e., the entropy does not play an essential role. Then, the dissociation enthalpies of C-C and C-H bonds were calculated using the Hess law. As a

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

251

result, the C-C bonds are broken more easily (356 kJ/mol) than C-H bonds (395 kJ/mol) (Figure 7). As a result, each plasma exposure of polyethylene leads rather to polymer degradation than to substitution of H atoms bonded to the polyethylene chain by atoms or fragments from plasma. Therefore, any type of successful polymer surface modification (plasma, radiative, chemical, mechanical) is a compromise between surface functionalization and decomposition of polymer. This preferred degradation is not only reflected in the lower dissociation enthalpy of C-C bonds but also by high heats of formation of CO2 and H2O formed as the main degradation and etching products of all polyolefins leading to self-acceleration of oxidation as present in combustion. Dehydrogenation and formation of polyenes as well as crosslinking, however, are also known from reactions initiated by high-energy irradiation characterized by low G-values (here, G-value means number of chain scissions per 100 eV) [27]. However, the actual goal of polymer surface modification is the introduction and the strong and permanent anchoring of oxygen-containing or other plasma-specific groups onto the polymer surface. C-H substitution by bromine

Br

Plasma

Br

H

H

C

C

H

H

395 kJ/mol Plasma

H +

Br

C

C

H

H

H

H

Recombination

2 Br H C

C

H

H

H

Br

C

C

H

H

H

Br

+

HBr

+

HBr

Nucleophilic substitution +

2 Br

C

C

C

C

H

H

H

H

C-C substitution by bromine

356 kJ/mol H

2 Br

+

C H

Plasma

C

C

C

H

H

H

H

H

H +

H

H

H

recombination C

C

H

H

+

C H H

Br

+

Br

C H

Figure 6.7 H substitution or C-C dissociation in polyolefins are schematically displayed using the example of the UV-stimulated (or plasma-initiated) bromination.

252 Progress in Adhesion and Adhesives, Volume 4

Unfortunately, because of the situation summed up in Figure 7, there is no known chemical way to substitute selectively H atoms from polyethylene chains by atoms or groups, generally, and specially nothing under plasma conditions with high energy excess. However, it is well known from chemical chlorination of poly(vinyl chloride) that the hydrogen atoms of the polymer backbone shield the C-C bond scissions fairly efficiently as also mentioned before (Figure 8). Thus, the desired C-H substitution is partially possible although it is still associated with considerable C-C chain scissions and generation of radiation-induced defects. At the beginning of plasma exposure, the polymer chains are not broken immediately by the attack of plasma. Then, the shield of hydrogen atoms is broken and the C-C bond scissions begin. With introduction of a few defects, the shielding effect of the hydrogen jacket/hose around the C-C backbone is eliminated and the reduction of molar mass begins, associated with the strong decrease in mechanical strength of the polymer. On the other hand, it was shown by Hansen and Schonhorn as well as by Hudis, about 50 years ago, that non-oxygen containing plasmas (noble gases, hydrogen) can produce crosslinking (Crosslinking by Activated Species of INert Gas=CASING process) associated with mechanical strengthening of the polymer surface region [28, 29] below the topmost surface caused by vacuum-UV irradiation from the plasma [30]. However, such crosslinking at polyolefin surfaces alters the fundamental structure and properties of the polymer. Summarizing the thermodynamic aspects, due to the compromise between strong mechanical weakening of the topmost polymer surface layer on exposure to any plasma and energy needed to break C-H (and C-C) bonds for anchoring of functional (polar) groups onto the polyolefin surface, plasma surface functionalization is associated with significant drawbacks.

H C H

H C H

H H

C H

C H

Figure 6.8 Principal scheme of (partial) shielding of polyethylene backbone by the surrounding hydrogen “hose” or “jacket” from direct attack from plasma particles (Backbone=thick bond line).

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

253

Nevertheless, plasma modification is often used and applied with some success. Here, introduction of OH and NH2 groups onto the polymer surface is investigated. OH-groups can be plasma-chemically generated at polyolefin surfaces in a few ways (Figure 9). Using the O2 plasma, the problem is the absence of hydrogen, thus OH (and COOH) groups cannot be produced in a direct way. Hydrogen must be first released from the polymer and transferred to the plasma for formation of OH features. As will be shown, noble gas plasmas also produce OH groups at polymer surfaces, however, only on post-plasma exposure to air by the so-called auto-oxidation process. A further possibility is the coating of polyolefins with a (well-adhering) plasma polymer bearing OH groups. This can be performed with the hope of dominant chemical chain-growth polymerization with monomers, such as allyl alcohol (AAl) [31, 32] or hydroxyethylmethacrylate (HEMA) [33] or by simple precursors reacting only under plasma conditions, such as alcohols [34].

Variant-1: gas plasmas

Variant-2: plasmas polymers Chain-growth polymerization of monomers

Plasma atmospheres suited for introduction of OH groups into the surface of polyolefins

O2

CO2

H2O

H2O2

Plasma polymers with OH groups deposited as thin topcoat onto the surface of polyolefins made from vinyl, allyl or acrylic monomers

H2+O2

Allyl alcohol

Hydroxyethylmethacrylate (HEMA)

Vinyacetate (hydrolysis is needed)

Non-polymerizable precursors Plasma polymers with OH groups deposited as thin topcoat onto the surface of polyolefins made from vinyl, allyl or acrylic monomers

Plasma atmospheres suited to produce radicals suited for post-plasma introduction of OH groups into the surface of polyolefins

Ethylene glycol He

Ne

Ar

Kr

H2

Butanediol

Ethanol

Methanol

Phenol

i-propanol

Addition of donors

The O2 plasma does not provide H for formation of OH, CHO or COOH

Plasma polymers with OH groups from vinyl, allyl or acrylic monomers with addition of OH donors

Allyl alcohol + O2

Allyl alcohol + H2O

Allyl alcohol + H2O2

Figure 6.9 Principal plasma-chemical processes for introduction of OH groups onto polyolefin surfaces, or covering polyolefin surfaces with well-adhering plasma polymers bearing OH groups.

254 Progress in Adhesion and Adhesives, Volume 4

6.2 6.2.1

Oxygen Plasma Treatment Formation of O Functional Groups at Polyolefin Surfaces on Exposure to Oxygen Plasma

Oxygen plasma can introduce O atoms onto polyolefin surfaces by endothermic reactions: O2 + electron or hv~2.0. Atomic oxygen can abstract hydrogen atom from polyolefin: >CH-H + .0. ~ >CH. + .OH. Highly branched polypropylene is most readily attacked by atomic oxygen. Lowpressure oxygen plasma consists of 10–20% atomic oxygen O(3P) and 10–20% molecular singlet oxygen 1O2(1Ag) as reactive species [35]. Bonding of atomic oxygen occurs as follows: >CH. + 0 ~ >CH-O . Such alkoxy radical has a short lifetime. Secondary reactions with hydrogen released from polyolefin (>CH-H + electron or hv ~ >CH. + .H and >CH-O. + .H ~ >CH-OH or >CH-O. + RH ~ >CH-OH + R.) or rearrangements are needed to form a complete functional group, such as OH. The same is true for the exothermic reactions at C radical sites with molecular oxygen in the triplet ground state 3~ g- from the air or present in the oxygen plasma: >CH. + .0-0. ~ >CH -0-0. (peroxy radical), which reacts secondarily with H from polyolefin starting a chain reaction (auto-oxidation) forming aliphatic hydroperoxides: >CH-O-O. + RH ~ >CH-O-OH + R . The R. radical formed can react with oxygen and re-initiate the process (auto-oxidation) or recombine with other radicals or atoms. Shard and Badyal have compared plasma and vacuum-UV exposure of polystyrene, which can also be considered as polyolefin in a wider sense [36]. They report similar percentages of C-Ox products for both oxygen plasma and VUV irradiation; however, photooxidative attack arises from excitation of the phenyl ring to form either a charge-transfer complex with molecular oxygen or a peroxy intermediate. The oxidation products were identified using X-ray Photoelectron Spectroscopy (XPS) following Clark´s fitting strategy to assign the C-O species to one, two, three or four C-O bonds [37, 38], shake-up satellite and specific aromatic bonds (phenol and benzoic acid) (Figure 10). The interpretation of oxidation products on exposure of polystyrene to oxygen plasma or UV by Shard and Badyal raises a question. The proposed formation of the carbonate (-O-CO-O-) structure, indicated by the appearance of a signal at a binding energy higher than 290 eV, is unlikely. Immediate decarboxylation (CO2 t ) of carbonate groups on plasma exposure was found [39–41]. Here, the interpretation as benzoic acid structure (aryl-COOH) or even as (not very stable) peroxyacid (R-CO-O-OH) is also plausible for a fitted band at a binding energy of about 290 eV [42]. However, it is assumed that alkoxy radicals can react with CO2 to carbonates, in analogy to the proposed formation of carboxylic groups in the CO2 plasma [43]. The formation of epoxy groups (C –OC) with atomic oxygen onto aliphatic chains is problematic because 2 hydrogen atoms must be removed from 2 adjacent carbon atoms from

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

255

Intensity (cps)

Untreated polystyrene standard Polystyrene, 32 s oxygen plasma

CHarom 285.0 eV

C-OH, C-O-C 286,6 eV >C=O, CHO 287.3 eV COOH, COOR 289.0 eV CO-O-OH, aryl-COOH O-CO-O 289.8 eV 290.5 eV

CH, CH2 285.3 eV

Shake-up satellite 291.0 eV

292

290

288

286

284

282

280

Binding energy (eV)

Figure 6.10 C1s peak of untreated polystyrene (in grey) and after exposure to the oxygen plasma (in black, cw-rf plasma, 50 W, 6 Pa, 32 s) [19].

the polyethylene backbone simultaneously. In analogy to ethylene oxide synthesis, the attachment of oxygen atoms to double bonds is more probable as found for graphene or cyclization of ether to epoxy [44]. Gerenser also confirms the presence of epoxy groups and ascribes them to the peak at a binding energy of 286.1 eV [45], however, the identification (derivatization) without doubt is missing. Ketone formation (RR´C=O) at polyethylene backbone needs also the substitution of 2 H atoms from a carbon atom of the methylene group:> CH2 + O >C=O + H2 otherwise an intermediate alkoxy radical from attachment of atomic oxygen is formed (>CH-O ) which must release atomic hydrogen, which is disadvantageous energetically: >CH-O >C=O + H . Another way would be the reaction of a carbene C: with atomic oxygen: >C: + O >C=O. Recombination of alkoxy and alkyl radicals to ether bonds (-C-O-C-) is highly conceivable. As demonstrated with the formation of OH groups, hydrogen is not present originally in the oxygen plasma and must be released by substitution reaction or dehydrogenation of olefins or by irradiation or plasma bombardment. In a following reaction step OH, CHO and COOH groups can be formed. A further problem is that CHO and COOH groups indicate chain scission, i.e. the polymer backbone is broken. High concentrations of these groups show reduced molecular weight or even formation of small fragments. This means that the cohesive strength of such fragments with functional groups in the polymer bulk is weak, known as weak boundary layer [47].

256 Progress in Adhesion and Adhesives, Volume 4

A better explanation for the formation of the diversity of O-functional groups is the concept of auto-oxidation occurring in oxygen plasma, in oxygen atmosphere and in the air as can be illustrated by the mechanism of paraffin oxidation outlined by Langenbeck, Pritzkow and Asinger (Figure 11) [2–4, 48]. As mentioned before, polyolefin oxidation is strongly exothermic and needs only an activation energy, here delivered from the oxygen plasma, for initiating the auto-oxidation process [26, 49]:

CH4 + 2O2

H0reaction = –810 kJ/mol

CO2 + 2 H2O

as calculated by Hess´rule with the following dissociation enthalpies [50]:

C-H

Hdiss = +413 kJ/mol

O-O

Hdiss = +498 kJ/mol

C=O

Hform = –803 kJ/mol

O-H

Hform = –463 kJ/mol

Calculated Hmethane = 4.413 + 2.498 - 2.803 - 4.463 = -810 kJ/mal. Hydrocarbon OH O Secondary hydroperoxide Side-reaction Main reaction OH

O Secondary alcohol

ketone

OH O

O

-ketohydroperoxide O Diketone

O

!

O

Carboxylic acid

O Aldehyde

OH

1

H O

Peroxy acid O

+ O OH

O

O OH O O ester

+

O OH

Figure 6.11 Mechanism of thermal paraffin oxidation as found by Langenbeck, Pritzkow and Asinger [2, 48].

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

257

For polyethylene, the complete oxidation of each CH2 group produces about HCH ≈ 2 –653 kJ/mol [51]. Such auto-oxidation is a pure chemical process initiated by carbon radicals, and these react with molecular oxygen in the ground state to produce peroxy and hydroperoxy groups which decompose to many O-functional groups as schematically depicted in Figures 11 and 12. As shown, atomic oxygen and excited molecular oxygen attack the polyolefin surface readily (Figure 12) [52]. Nevertheless, all C radical sites generated by plasma exposure react easily with molecular oxygen in the ground state to peroxy radicals on exposure to oxygen plasma. Additionally, radicals also react post-plasma on exposure to pure oxygen atmosphere in the reactor and, finally, on exposure of the plasma treated sample to ambient air. Thus, the auto-oxidation dominates ultimately the types of O-functional groups formed and their percent distribution among all O-functional groups. The mechanism of auto-oxidation, as exemplified for paraffins in Figure  11, can explain the existence and percent distribution of all O-functional groups after exposure to oxygen plasma. Fitting of the XPS-C1s peak of oxidized polyolefins was performed using a routine procedure introduced by Clark and Dilks [37] and further improved by Briggs [53, 54] as well as Pleuel and Simon [55]. The XPS-C1s peak position depends on the number of C-O bonds in the respective O-functional group (Figure 13). The subpeak at a binding energy (BE) of 286.1–286.6 eV in polyethylene is the most intense new peak after exposure to oxygen plasma and is assigned to OH, C-O-C, epoxy (C –OC) and C-O-O(H) (peroxy/hydroperoxy) groups. Those at BE≈287.5 and that at BE=288.8 to 289.3 eV are assigned to ketones, diketones and aldehydes as well as carboxylic

Oxygen plasma irradiation

Oxygen plasma

Air

Etching

diffusion of oxygen

Vacuum-UV irradiation (h )

Auto -oxidation

Particle bombardment

Linear chains, unbranched structure

! H-abstraction and formation of radical sites appropriate for attachment of atoms and fragments (functional groups)

O-

1

O O

O-O Polymer degradation, chain scissions of backbone, lowering of molecular weight

O

O-O

O O-O

O-O

Crosslinked structure by covalent peroxy linking of macro molecules in presence of oxygen dissolved in polymer or as contamination in plasma

Chain scissions of backbone, introduction of functional groups

O

CO2, CO, H2O

O-O Crosslinked structure by formation of inter- or intramolecular C-C bonds, increase of molecular weight

Etching, emission of etching gases, introduction of functional groups

O

O OH

further introduction of O-functional groups by auto-oxidation always leading to the same spectrum of O-functional groups

Figure 6.12 Degraded and crosslinked polymer structures caused by plasma and post-plasma processes.

258 Progress in Adhesion and Adhesives, Volume 4

Low-pressure oxygen plasma (32 s) Untreated polyethylene

Intensity (a.u.)

C-O 286.6 eV

C=O, O-C-O 287.9 eV

O=CO-O 290.5 eV

292

CH, CH2, CH3 285.0 eV

O-C=O 289.1 eV

290

288

286

284

282

Binding energy (eV)

Figure 6.13 XPS-C1s peak of polyethylene after exposure to the low-pressure oxygen plasma fitted after peak fitting routine introduced by D. T. Clark [37].

acids, esters, peroxy acids (R-CO-O-OH) and peroxy esters (R1-CO-O-OR2), as depicted in Figure 10. A further peak at BE=290.5 eV was discussed before and it can be assigned to carbonate groups or to benzoic acid (aryl-COOH). In the immediate vicinity, the shake-up signal at BE≈291.5 eV of aromatic rings is located. The occurrence of this peak was proposed by Tibbitt et al. for the plasma polymerization of ethylene [56]. Moreover, it is known that gaseous products of the etch process, such as CO2, H2O or H2, form a plasma polymer at another place by re-deposition. Thus, also traces of aromatic rings can be formed via re-deposition as phenolic or benzoic rings [57]. A further verification of all these O-functional groups is possible by surface-enhanced FTIR spectroscopic techniques, such as surface-enhanced IR absorption (SEIRA), infrared reflection-absorption spectroscopy (IRRAS) or attenuated total reflectance (ATR). In particular, OH and NHx groups, CH3 groups, carbonyl, olefinic, acetylenic, aromatic features, etc. can be identified by such IR techniques in addition to XPS [58]. In Figure 14, the various theoretically occurring O-functional groups are listed. The oxygen plasma is not able to generate a complete functional group in a single step as known in wet-chemistry, for example by SN2 reactions. Hydrogen abstraction or C-C chain scission must occur. A mechanism of H-abstraction and simultaneous attachment of a plasma gas fragment or atom in a single step is not probable as shown before. As already mentioned, formation of carbonyl units needs also replacing of 2 hydrogen atoms from the backbone (ketone and aldehyde) or C-C bond scission (carboxyl and ester groups). Thus, most plausible interpretation of detected groups at polyolefin surfaces is their generation by chemical auto-oxidation even in the oxygen plasma also in pure oxygen atmosphere after switching off the plasma as well as on exposure to oxygen from ambient air as presented in Figure 11.

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

259

Polymer substrate

Functional groups O-OH OH

Peroxyacid

Hydroxyl

Radicals OH Carboxyl O

O

OR

O-O-R

Ester

Peroxyester

OH Phenol (?)

O

O

O C O O

O

Epoxy (?)

CH2 O

Ether

O-O-R

Ketone

O-R*

Carbonate (?)

Peroxide

O

O-R

Allyl

Peroxy

O-O

O

CH2 Trapped CH2 (dangling) CH2

O

Primary ozonide

O O CH O

Secondary ozonide

HC

Ketene

CH2-O

2

CH

Acetal

CH CH CH

Alkyl

O-OH Hydroperoxide

CH2 O

CH2

Carbene

H Aldehyde O

O-R*

Hemiacetal

C=C=O

Alkoxy

OH

Figure 6.14 Possible O-functional groups at polyolefin surface after exposure to the oxygen plasma.

Table 3 presents an overview of possible O-functional groups after exposure of polyethylene to oxygen plasma bonded with or without bond scission of polymer backbone. It should be mentioned again that C-C bond scission acts very negatively on the mechanical strength of polymers, especially in the topmost surface layer. The situation can be summarized as follows: polyolefins (….–CH2-CH2-CH2-CH2-….) do not have any functional (polar) group, therefore, they are chemically inert and unreactive one exception, polyolefins are easily oxidatively attacked followed by chain scissions and loss in mechanical strength and in extreme case, vaporization by combustion occurs oxidations proceed spontaneously (exothermic) with oxygen as well as with fluorine or chlorine and all reactions are also possible without plasma the plasma oxidation of polyolefins has two enthalpy sources, chemical reaction enthalpy and plasma-electrical energy input in contrast to uncontrolled combustion, the plasma oxidation is better tunable because of low temperature and low oxygen concentration in the oxygen low-pressure plasma

260 Progress in Adhesion and Adhesives, Volume 4

plasma oxidation is limited to surface and near-surface layers plasma oxidation does not produce such high temperature as combustion all consumed electrical power input and the released reaction enthalpy of oxidation is focused in the topmost layer of a few nanometers of polymer surface variation of plasma gas enables attachment of different types of functional groups onto polyolefin surface plasma oxidation produces a broad variety of O-functional groups important side-reactions in the plasma are chain scissions, polymer degradation, dehydrogenation with the formation of polyene structures and crosslinking in layer below the surface most important post-plasma side-reaction is the oxidation via attachment of molecular oxygen onto radical sites such as peroxy radicals, which cannot be avoided post-plasma oxidation is known in polymer chemistry as auto-oxidation, which forms the final O-functionalization of polyolefin surfaces.

6.2.2

Kinetics of Polyolefin Oxidation – Dependence on Parameters

As often reported, etch rates of polymers exposed to oxygen plasma are constant with time [23, 59–63] as shown by linear mass loss of polypropylene exposed to the low-pressure oxygen plasma (Figure 15). Table 6.3 Assignments of functional groups and structures associated with C-C chain scission or preferably with substitution onto the intact backbone. Additionally, XPS binding energies are presented for these features for polyethylene exposed to the oxygen plasma. XPS-C1sBE

Structure

Terminal group and C-C scission (CS) or H-substitution with intact chain (IC)

285.0 eV

-CH2-CH2-CH2=CH2=CH2+CH3>CH-CH
CH-OH >CH-OH -CH2-O-CH2 >CH-O-OH

Primary hydroxy group (intact) Secondary hydroxy group (intact) Tertiary hydroxy group (scission) Ether (scission) Hydroperoxide (intact)

287.5 eV

>C=O -CHO -O-CH2-O-

Ketone (intact) Aldehyde (scission) Methylene glycol (unlikely)

289.0 eV

-COOH -COOR

Carboxyl (scission) Ester (scission)

290.5 eV

-O-CO-O-

Carbonate (scission)

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

261

60

Loss in sample mass (a.u.)

50 40 30 20 10 0 0

10

20

30

40

50

60

Time (s)

Figure 6.15 Mass loss of polypropylene exposed to the low-pressure oxygen rf plasma.

Supermolecular structures (spherulites, domains, quenched (amorphous) surface layers, etc.) of polypropylene seem not to have a significant influence on etch rate (Figure 15). The percentages of C1s components remain constant in dependence on exposure time (Figure 16). The two diagrams in Figure 16 show that most of the O-functional groups are introduced onto the polyolefin surface within 2 s. The number (x) of fitted (C-O)x bonds (see Figure 13) show (often) a ratio of 5 (C-O)1 :3 (C-O)2 : 2 (C-O)3. After 18 s, no further increase in O/C takes place and the final steadystate is reached. Using angle-dependent Near-Edge X-ray Absorption Fine Structure (NEXAFS) analysis, the etching progress of aliphatic layers (octadecyltrichlorosilane self-assembled monolayers, OTS-SAMs) as model for polyethylene, its functionalization and disorientation of its anisotropic structured aliphatic molecules, which were covalently linked to the Si-wafer by C-O-Si bonds, could be clarified (see also Figure 17) [26, 64]. It was found that the anisotropic orientation of OTS molecules and also the macromolecular orientation in biaxially stretched polypropylene were nearly stable for about 2 s exposures to oxygen plasma [41]. More than 2 s plasma exposure ends up in a complete disorientation of the OTS monolayer, which has a thickness of about 3 nm. The same was observed for polypropylene within the NEXAFS (Near-Edge X-ray Absorption Fine Structure) sampling depth. The thickness of the modified polymer layer was verified by comparison of results obtained from CK and OK edges assuming about 3 and 5 nm sampling depths, respectively [65]. In Table 4 the identification of plasma-caused processes by important analytical methods is listed.

262 Progress in Adhesion and Adhesives, Volume 4 32 Functionalization

Penetration/saturation

Steady-state functionalization-etching

Measurement [19] Measurement [42]

20

Functionalization of deeper layers Functionalization of the topmost layer

Ototal

20 C1s [%]

O-introduction [% O/C]

25

24

Steady-state (etching)

30

28

16 12

15 C-O 10

8

C=O

5

4

O-C=O

0 0 0

10

20 30 40 50 Time of exposure to O2 plasma (s)

0

60

2

4 6 8 10 12 14 16 Time of exposure to O2 plasma (s)

18

20

Figure 6.16 Exponential increase of oxygen introduced onto polyethylene surfaces on exposure to the O2 plasma (cw-rf, 6 Pa, 100 W) and fitted C1s components on polyethylene surfaces in dependence on exposure time [19, 31, 41].

Table 6.4 Sampling depths of methods for identification of essential processes on exposure of polyolefins to low-pressure oxygen plasma from the topmost surface to the bulk (SEC-Size Exclusion Chromatography, EDX-Energy-dispersive X-ray spectroscopy) [49, 50]. Structural elements

XPS/NEXAFS (3–5 nm)

IRRAS (1–3 nm)

FTIR-ATR (≈2000 nm)

Range of zone from topmost surface [nm]

Functionalization

x

x

0

3

(Post-plasma) oxidation

x

x

(x)

>>3

Disorientation

x

x

0

>3

Polyene

x

x

x

>>2000 (EDX)

Crosslinking

-

-

-

>>>2000 (SEC)

In Figure 17, a general summary of processes is shown schematically.

6.2.3

Influence of Type of Plasma Gas

Chemical etching in oxygen plasma dominates over physical etching. Noble gas plasmas exhibit no chemical, only physical etching (sputtering) (Table 5) [59]. Table 5 confirms the dominance of chemical etching of polyethylene on exposure to oxygen plasma compared to its exposure to physical etching alone in non-oxidative plasmas. Therefore, oxygen and water vapor plasmas etch most strongly. Noble gas plasmas can only etch by plasma particle bombardment (“sputtering”).

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

CO2

Plasma phase

H2O

H2

Oxygen plasma

H2

Surface

Idealized (oriented) aliphatic model

OH

CO2

H2O

CO2 H2O

COOH

Topmost-surface (level of functionalization)

H2

COOH

O O

263

O

O

O

O

O

O O

3 nm

OH CO-OOH

O-OC

Near-surface (level of radical formation. crosslinking and polyene formation)

HO

O-OH

O

O-OH

O-O

HO-O

O O-O

3 - 30 000 nm

O O O-O O

Unaffected bulk

o s exposure to O2 plasma

2 s exposure to plasma

4 s exposure to plasma and subsequent exposure to air

Figure 6.17 Schematic course of oxygen plasma induced processes at anisotropic structured aliphatics. Table 6.5 HDPE etched/exposed to different kinds of low-pressure plasmas (HDPE-O2 plasma= 100). Plasma

Relative etching rate

O2

6.2.4

100

H2O

59

Ar

36

H2

12

NH3

13

N2

11

Influence of Polymer Composition

Conclusions on the mechanism of oxygen plasma etching can be drawn from the etching rates of different polymers. It was concluded that the O/C ratio plays the most important role for etching and predominates over structural features (Figure 18) [23, 59–63]. Figure 18 confirms that the etching rate depends nearly linearly on O/C ratio of the polymer. The O/C determines the etch rate and, therefore, the rates of CO2 and H2O formation.

264 Progress in Adhesion and Adhesives, Volume 4 20 OH HO O

O

O

Oxygen containing

OH

Etching rate (mg/cm2s)

15 83% O/C

C H3 C OO -C H 3

C H2

C C O O -C H 3

Aliphatic

50% O/C

10

50% O/C

O

O O C

C O CH2CH2 O

O C

O C O (CH2)4-

C H3 C H C H2

CH2 CH2

CH2 CH2

Stabilized by aromatic rings Olefins-tendency to crosslinking

5

40% O/C

33% O/C

0% O/C

0% O/C

0% O/C CH CH2

0% O/C

0% O/C

PS

Natural rubber

0

Cellulose

PVAc

PMMA

PET

PBT

PP

LDPE

HDPE

16

Etching rate [mg/cm2s]

14 12

atic

ph

10

Ali

8 6

tic

ma

Aro

4

?

ic fin

2 0

Ole 0

20

40

60

80

O/C (%)

Figure 6.18 Nearly linear dependence of etching rate on O/C ratio of various polymers (cellulose, poly(vinyl acetate-PVAc, poly(methyl methacrylate-PMMA, poly(ethylene terephthalate)-PET, poly (butylene terephthalate)-PBT, polypropylene-PP, low-density polyethylene-LDPE, high density polyethylene-HDPE, polystyrene -PS, natural rubber).

However, it should be added that independently of O/C ratio, the remaining polymer in the near-surface layer undergoes characteristic changes in molecular structure. Four general types of polymers could be identified which show a characteristic response to molecular structure on exposure to oxygen plasma with the formation of specific types of functional groups [19]: 1. photo-oxidative degradation, for example poly(ethylene terephthalate), poly(bisphenol-A-carbonate) 2. crosslinking as well as polyene formation, for example polyethylene 3. depolymerization (unzipping), for example PMMA 4. random degradation, for example poly(ethylene oxide).

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

6.2.5

265

Auto-Oxidation

When the polyolefin is exposed to the oxygen plasma, its surface reacts with ions, excited species, energy-rich ozone and atoms, is etched and modified, is irradiated with far-UV radiation from the plasma and radical sites are generated. Additionally, most of oxygen in the plasma gas phase is molecular oxygen in the ground state. This stable 3O2 (3~ g) state possesses some biradical character ~ .0-0. or . Thus, such pure chemical auto-oxidative (C.+.O-O. ~ C-O-O.) reaction of oxygen molecules with C radical sites predominates and continues the plasma-initiated oxidation reactions. This plasma-less auto-oxidation occurs preferably at the polymer surface because the diffusion of oxygen into the near-surface layers is hindered because of its low-pressure in the range of 10 Pa. As shown in the next Section in detail, after switching-off the plasma the O2 gas atmosphere remains (for a short time) and the pure chemical auto-oxidation proceeds. The susceptibility to auto-oxidation increases from linear polyethylene to branched polyethylene and to polypropylene [50]. On exposure of the O2 plasma-treated polyolefin to ambient air, the remaining trapped radicals in deeper layers can then also react with slowly diffusing oxygen at atmospheric pressure and complete the auto-oxidation over a period of some weeks or months [66–69] following the process as shown in Figure 11. On plasma exposure, a great diversity of C radicals is produced (Figure 19) [70]. Initial reaction is the formation of a C radical site by radiation or plasma particle impact: R-H + hv ~ R. + .H or R-R’+ hv ~ R. + R' •. This is followed by peroxy radical formation by reaction with molecular oxygen: R. + .0-0. ~ R-O-O. (rapid) and chain growth: R-O-O. + R’H ~ R-O-O. + R’. or crosslinking reactions: R-O-O. + R’-0-0 H C C C C Polyethylene

------.-Ar plasma

:

-H

.

H C

.. C

H H C C

Polyethylene Mid-chain radical

H C

Allyl radical

Polyenyl radical

C C nC Polyethylene

- ____________________ :n-f,e: __

Vinylene bond

H H H H C C C C Polyethylene

H C C C C Polyethylene

-H

H-migration (radical “walking” mechanism)

H, C

Hj C

•

C

HI C

Continuation by auto-oxidation on exposure to air

Polyethylene

-CH2-CH2-CH2-CH2-CH2-CH2-

-CH2-CH2-CH2-CH2-CH2-CH2-

-CH2-CH2-CH2-CH2-CH2-CH2-

-CH2-CH2-CH2-CH2-CH2-CH2-

cross-linking

Immobilized free radical (dangling bond)

Figure 6.19 Different ways of radical formation in polyolefins as proposed in [70].

266 Progress in Adhesion and Adhesives, Volume 4

R-R´ + 2 O2 or R. + R’. ~ R-R’ or R-O-O. + R’. ~ R-O-O-R´ as already shown in Figure 11. Double bond formation proceeds by H-release from the aliphatic chain followed by formation of a mid-chain radical and completed finally by generation of the midchain vinylene double bond: -CH2-CH2-CH2-CH2-CH2-CH=CH-CH2- or vinyl end group: -CH2-CH2-CH2-CH3 -CH2-CH2-CH=CH2 or adjacent to conjugated polyene structures. Sometimes, migration of radicals along the macromolecular chain is observed (“walking”). Other O-functional groups can be formed by the following reactions: alcohol: R-O-OH ~ RO. + .OH, RO. + R’H ~ ROH + R’.; ether: RO. + R’ ~ R-O-R’; ketone: RO. + R’. ~ R-CO-R’; aldehyde: R-O-OH + R’. ~ OHC-R + HO-R´; carboxyl: -CH2-CH(O-OH)CH(O-OH)-CH2- ~ -CH2-COOH + OHC-CH2-. The dominating products of oxygen plasma exposure and weathering on exposure to Florida climate are found to be nearly the same as found in the beginning of 1980s (“plasma ageing”, Figure 20) [71–73]. Oxygen plasma aging

Weathering

Polyolefin

Polyolefin

!•

O2 plasma O

OH

~

OH

I II

O O O

~

Reaction with oxygen

OH OH OH O

~

j

O OH

Sunlight

Reaction with oxygen

OH

OH OH

O

O

I I

Modified ("aged") zone

J 1’--1. 1"('"

Modified (aged) zone

j

Steady-state of progressive ageing and etching

:t·.~. t·nij·t-1-:

O OH OH O OH O COOH

Etched zone

modified ("aged") zone f•••••••:••:••••••]

O

O

Progressive penetration of ageing front

COOH OH

lmn"LI.-:-::J’

Modified (aged) zone

Figure 6.20 Schematic comparison of polyolefin ageing on exposure to oxygen plasma and on weathering.

The Formation Processes of Functional Groups at Polyolefin Surfaces on Exposure

267

The dimension of “plasma aged” zone in polyolefin is much smaller compared to natural weathering. Plasma vacuum UV radiation forms polyenes in deeper layers. Such polyenes in a moderate distance from the surface (several hundreds nm) absorb the vacuum UV and limit its penetration to deeper polymer layers (see Figures 6 and 12) [71–73]. It must be added that peroxy acid groups can also be formed by exposure to energyrich radiation as detected in the IR spectrum [74, 75]. The mechanism of their formation in the plasma is mostly unknown, however, it can be assumed that the attack by excited molecular oxygen on aldehydes produces peroxy acids: R-CHO + O2* R-CO-O-OH or .OH radicals react with carboxy groups: R-COOH + 2 .OH ~ R-CO-O-OH + H2O. In the XPS spectra, the corresponding C1s peak is assumed to be at about 290 eV as described by Clark, who presents the following order in binding energies [42] (see also [76]): alkanes 0) the walls of a capillary Eqn. (13) becomes:

P

2 cos( ) r

(14)

Washburn [13] has described the rate of penetration, , of liquid into cylindrical capillaries. Washburn’s equation is written as follows:

dl dt

r

L

cos 4 l

(15)

where is the liquid viscosity and l the depth of penetration. For < 90°, is positive indicating that penetration is spontaneous, while for > 90° is negative indicating liquid withdrawal from capillaries is spontaneous. The above equations represent the principal surface thermodynamic relations that are likely to form the scientific basis for a large number of development problems associated with product performance or product manufacture. The ability to design, control and determine surface free energies would allow for easier development of products which require knowledge of wetting, adhesion, spreading and liquid penetration for their use or manufacture. Knowledge of the relevant interfacial and surface free energies is critical to the understanding of each of the above discussed phenomena.

320 Progress in Adhesion and Adhesives, Volume 4

7.3

Contact Angle Methods

Various methods employing the measurement of contact angles have been used over the past few decades. As measurements of contact angles on small particles is usually impractical or impossible the methods are only highlighted here. Etzler [14] has published a review that has discussed these methods in detail. Many of the models described here have also been used in conjunction with IGC (Inverse Gas Chromatography).

7.3.1

The Zisman Method

Zisman and co-workers (see for instance, [15]) also noted that plots of cos( ) vs. LV were nearly linear, particularly when homologous series of liquids were used. It thus follows that:

cos

1

c

(16)

LV

where c is referred to as the Zisman critical surface tension. A plot of cos ( ) vs. LV will intersect cos ( ) = 1 when LV = c. Zisman’s relation is empirical. Despite the fact that c is not the solid surface free energy, the critical surface tension has been shown to correlate with the known surface chemistry of several solids. Determination of c is an adequate measure of solid surface free energy for many practical problems. Mittal [16] has discussed the relation between c and sv.

7.3.2

The van Oss, Chaudhury and Good Method

Several investigators including Fowkes [17–21], Good [22, 23], van Oss [24] and Chang [25] have constructed statistical thermodynamic models for wetting and adhesion. These models may be considered as statistical thermodynamic in the sense that they offer molecular interpretations for origins of wetting and adhesion. In this section, the basic principles common to these statistical thermodynamic theories are explored and the van Oss, Chaudhury and Good [22, 24] model is, in particular, discussed. The van Oss, Chaudhury and Good model is frequently used for the determination of surface free energy of solids when contact angles are employed. Intermolecular forces between molecules result from interactions between their corresponding electron orbitals. The principal non-bonding interactions result from induced dipole-induced dipole (London), dipole-dipole (Keesom) and dipole-induced dipole (Debye) interactions. The intermolecular potential energy function for each of these three types of interactions is of the same form.

U

r

12 6

(17)

If London dispersion forces are considered Eqn. (17) can be expressed as follows. d 12

2 I1 I 2 I1 I 2

d d 11 22

1

2

2 d 11

a2 a1

d d 11 22 d 22

a1 a2

(18)

Surface Free Energy Determination of Powders and Particles 321

Here the subscripts 11, 22, 12 refer to interactions between like molecules (11, 22) and dissimilar molecules (12). is the coefficient in Eqn. (17). Also, I is the ionization potential and the polarizability. If I1 ≈ I2 then d 12

12

d d 11 22

(19)

Eqn. (19) forms the basis of the Berthelot principle [26] (also see for instance, Chang [27]) which states that dispersion interactions between dissimilar molecules can be estimated as the geometric mean of the interactions between like molecules. Alternatively, if 1 ≈ 2 then d 12

d d 11 22 d d 11 22

2

(20)

The harmonic mean estimation expressed in Eqn. (20) has been used less often and is frequently numerically similar to the geometric mean approximation. The interaction potentials between molecules have been used to determine the interactions between macroscopic bodies. For a column of material 1 interacting with material 2, d , to move the plate of material 2 from distance d to infinity is: the free energy, G12 d G12

d 12 N 2 3

W12d d12

6x

A12

N1dx

2 12 d12

(21)

Here A12 is the Hamaker constant (see for instance [5, 27]) and

A12

2

(22)

12 N1 N 2

N1 and N2 are the numbers of molecules of types 1 and 2, respectively. Recalling Berthelot’s principle [26]

A12

1

A11 A22

(23)

2

and further assuming that the intermolecular distance, d12, can be approximated as the geometric mean of the intermolecular distances found in the pure components (d11 and d12):

d12

12

(24)

(d11d22 )

Thus the work of adhesion, W12d , resulting from London dispersion forces is

W12d

A11 A22

12

12 d11d22

W11dW22d

12

2

d d 1 2

12

(25)

322 Progress in Adhesion and Adhesives, Volume 4

Fowkes [20, 21] suggested that the surface free energy of materials could be considered to be a sum of components resulting from each class of intermolecular interaction; thus,

(26)

i i

Eqn. (26) might be expanded as d

p

AB

...

(27)

Here the superscript d refers to dispersion forces, p refers to dipole-dipole (Keesom) and dipole-induced dipole (Debye) interactions, and AB refers to Lewis acid - base interactions. London, Kessom and Debye interactions are non-bonding orbital interactions and Lewis acid-base interactions, by definition, involve electron acceptance and donation. van Oss, Chaudhury and Good choose to express surface free energy in terms of two principal components - Lifshitz-van der Waals (LW) and Lewis acid-base (AB) components. The Lifshitz-van der Waals term is composed of the interactions covered by Eqn. (17) (London, Keesom and Debye). The work of adhesion attributable to Lifshitz-van der Waals interactions is estimated using the geometric mean rule discussed above. Thus

WALW

2(

LW LW 1 2 1 2 )

(28)

The use of the geometric mean approximation with regard to Lifshitz-van der Waals interactions is not unique to the van Oss, Chaudhury and Good approach and is also used in the models to be discussed later by Chang-Chen and by Fowkes. The harmonic mean approximation has also sometimes been used although the results of the two calculations are often nearly identical. The use of the geometric mean approximation is not a subject of current controversy. The relative merits of the geometric and harmonic mean approximations have been discussed in the literature [19, 28–32]. According to the van Oss, Chaudhury and Good model [22, 24] the Lewis acid-base component is modeled as follows AB i

12

2

where + is the Lewis acid parameter and and Good further choose

–

i

i

i

(29)

the Lewis base parameter. van Oss, Chaudhury

i

0

(30)

for alkanes, methylene iodide and -bromonaphthalene which presumably interact only through Lifshitz-van der Waals interactions. For water H2 O

H2 O

25.5 mJ/m 2

(31)

Surface Free Energy Determination of Powders and Particles 323

Based on these above numerical choices + and – have been experimentally determined using contact angles for a variety of liquids. van Oss [24] has compiled and reviewed the determination of these values (Also see [14]). Della Volpe and co-workers [33–35] have argued that the choice of + and – for water is inappropriate and van Oss [36] has argued that this numerical choice is not scientifically significant. Kwok [37], for instance, has criticized the use of surface free energy components altogether. Earlier Owens and Wendt [29] had described surface free energy in terms of two components which were called dispersion d and polar p. Thus d

p

(32)

While it is generally recognized that d ≈ LW the meaning of p is perhaps hopelessly confused in the literature. According to Fowkes[20, 21], p should refer to dipole-dipole (Keesom) interactions. In the van Oss, Chaudhury and Good model such interactions are incorporated into LW. Good [38] no longer recommends the use of p. Good’s argument follows in the next paragraph. Eqn. (29) reminds us that for monopolar materials ( + or – = 0) AB = 0. On the other hand, for two dipolar ( +, – ≠ 0 ) interacting materials p > 0; thus AB ≠ p. For example, the surface tensions of carbon tetrachloride and chloroform are nearly identical yet their interfacial tensions with water are 45.0 mN/m and 31.6 mN/m, respectively. Because chloAB roform is a monopolar acid ( + ≠ 0, – = 0 = 0) a descriptor such as p is inadequate to describe the difference in the observed interfacial tensions as AB = 0 for these two substances. Instead Eqn. (29) is a better descriptor. From a practical point of view, reported values of d and p can be regarded as LW and AB, respectively. The values, however, should be interpreted in terms of the van Oss, Chaudhury and Good model. The symbols d and p should no longer be used; LW and AB should be used in their stead. The Owens and Wendt method for the above reasons should be considered obsolete. Recalling Eqns. (10), (12) and (29) together with the relation, LW

AB

(33)

,

it follows that

WA

l

2

1 cos LW LW s l

12

2

12 l

s

2

12 l

(34)

s

If the van Oss, Chaudhury and Good parameters are known for at least three liquids and the contact angles of these liquids on a solid are measured, then Eqn. (34) can be used to determine the van Oss, Chaudhury and Good parameters for the surface free energy of the solid. van Oss [24] has reviewed the numerous publications which have reported the determination of the van Oss, Chaudhury and Good parameters for various liquids. A table of van Oss, Chaudhury and Good parameters for various test liquids can be found in Etzler’s review [14].

324 Progress in Adhesion and Adhesives, Volume 4

7.3.2.1 Methods for Calculating the van Oss, Chaudhury and Good Parameters The simultaneous solution of Eqn. (34) for a set of several liquids would seem straightforward, at first glance. Gardner et al. [39], however, have pointed out that different calculation approaches may give numerically different results and that a particular author’s choice may not be readily apparent to the reader. The choice of liquids used for the determination of surface free energy should not be completely arbitrary and careful selection is required. Dalal [40] discussed the choice of liquid sets used to determine surface free energy parameters. While Dalal’s discussion addresses the older Owens - Wendt [29] model much of the discussion applies directly to the van Oss, Chaudhury and Good model as well. Because the Owens-Wendt model has only two parameters, it is only necessary to measure contact angles for two liquids. Dalal noted that the calculated values for the surface free energy components depended on the choice of liquids. The use of dissimilar liquid pairs (e.g. water, methylene iodide) minimized the dependence of the calculated results upon the precise choice of probe liquids. Dalal [40] recommended that many liquids be used and that the contact angle results from this overdetermined set of liquids be used to find the best fit surface free energy components. The present author [41] concurs with Dalal’s conclusions but realizes that many authors have not heeded this advice. Recent work by Etzler [42] has shown that the number of liquids used to determine the surface free energy should be the number of model parameters plus three. If the requisite number of liquids are not available then replicate measurements of each available liquid may be adequate. In this work by Etzler [42] a method for determining the best (statistically) model to use is discussed in detail. For the solution of Eqn. (34), most authors choose to measure the contact angles with at least one non-polar liquid ( +, – = 0) such as methylene iodide and two polar liquids one of which is usually water as water is most dissimilar to other liquids. If a non-polar liquid is chosen, then Eqn. (34) reduces to

WA

l

1 cos

2

LW LW l s

12

(35)

The Lifshitz-van der Waals component may thus be calculated from a single contact angle measurement. (Subject, of course, to Dalal’s warnings. Also see Etzler [42]) The parameters sj , (j = LW, +, –) in Eqn. (34) have been interpreted in two ways for fitting purposes. The first method involves determination of sj directly from Eqn. (34). In this j . The first case one will find sj 0 andWA 0. In the second case investigators let c sj s second choice allows c sj to become negative during the fitting process and thus WA may also be negative. Again, Gardner et al. [39] have explored the consequences of these two choices. The first choice is the correct van Oss, Chaudhury and Good [22, 24] model. The fitting to Eqn. (34) of contact angle data can be accomplished in at least three ways. The first way involves direct fitting to Eqn. (34) using data from the entire pool of chosen liquids. This method works best when an overdetermined set (more than three liquids) of data are available. Care should be taken to select well - conditioned set of liquids as Dalal[40] suggested earlier. If this first method is chosen when data for only a few liquids

Surface Free Energy Determination of Powders and Particles 325

are available, it is possible to obtain an unrealistic result. The second method involves the use of Eqn. (35) to fit the data from measurements made with non-polar liquids (alkanes, methylene iodide, -bromonaphthalene, etc.). to determine sLW . Data from the polar liquids and Eqn. (34) are then used to determine s and s . When data from many liquids are available, both the first and second methods will yield comparable results. When only a few liquids have been chosen the second method is preferable. A third method would be to borrow from Owens and Wendt the expression

WA

l

1 cos

2

LW LW l s

12

2

AB AB l s

12

(36)

Eqn. (36) and data from contact angle measurements using polar and nonpolar liquids can be used to determine sLW and sAB . Equation (34) can subsequently be used to calculate s and s . Eqn. (36) must also be simultaneously true. It is important to the reader of the literature to realize that various authors have calculated so-called van Oss, Chaudhury and Good parameters using various numerical methods which may influence the calculated result significantly. The van Oss, Chaudhury and Good model expressed in (Eqn. 34) requires WA > 0 and sj lk 0. Furthermore, lk 0 for all investigated liquids (j,k = +,–). Again, tabulated values for the van Oss, Chaudhury and Good parameters have been compiled by van Oss, for instance [14, 24]. The best estimates of the van Oss, Chaudhury and Good parameters require the use of an overdetermined set of probe liquids and carefully and properly measured contact angles. The set of chosen liquids should contain liquids that are non-polar as well as liquids that are polar. As the van Oss, Chaudhury and Good parameters can be calculated using different algorithms it is important for readers of the literature to know which algorithm has been used. The model, as van Oss had intended, requires the parameter values to be positive. Care must be taken that the selected probe liquids do not exhibit stick-slip advancing and that contact angle does not change over the measurement time. See Etzler [42] for the appropriate statistical considerations.

7.3.3

The Chang – Chen Method

The Chang-Chen model [25, 27] for interfacial free energy is largely based on the same principles which govern the van Oss, Chaudhury and Good model. Both models treat Lifshitz-van der Waals interactions in the same way. Calculation of the surface free energy components requires the knowledge of the same experimental data. The two models, however, differ in the way that Lewis acid-base interactions are modeled. Recall that

WA

WALW WAAB

(37)

and LW

AB

.

(38)

326 Progress in Adhesion and Adhesives, Volume 4

The Chang-Chen model uses the same geometric mean approximation for WALW as does the van Oss, Chaudhury and Good model. Thus

WALW

WAL

2

12

LW LW 1 2

P1L P2L

(39)

where

Pi L

2

L i

12

(40)

PiL is the dispersion parameter. The superscript L is equivalent to LW. Like the van Oss, Chaudhury and Good model, the acid-base interaction is modeled using two parameters. These parameters, Pia and Pib, are referred to as principal values. The acid-base work of adhesion can be represented using the following relation:

WAAB

GAAB

(P1a P2b

P1b P2a )

(41)

PaPb

(42)

The surface free energy of the material is thus LW

AB

1 L P 2

2

Tabulated Pia and Pib values [14] are substituted into Eqn. (41) such that the work of adhesion is maximized and the free energy of adhesion is minimized. The acid-base character of a material is characterized by the sign of Pia and Pib. If Pia = Pib = 0 then the material is neutral (or non-polar). If Pia and Pib are both positive then the material is monopolar acidic and if both are negative then the material is monopolar basic. If Pia and Pib are of opposite sign then the material is amphoteric. A table of Chang-Chen parameters for test liquids is given by Etzler [14]. Despite some similarities to the van Oss, Chaudhury and Good model, the ChangChen model differs from the former model in a number of ways. The Chang-Chen model only applies the geometric mean rule to Lifshitz-van der Waals interactions. In determining values for Pia and Pib, interactions involving only n-alkanes are assumed to exclusively result from Lifshitz-van der Waals interactions. The van Oss, Chaudhury and Good model, for instance, assumes that both methylene and -bromonaphthalene also interact exclusively by Lifshitz-van der Waals interactions. A major difference is that the Chang-Chen model allows for both attractive and repulsive interactions. In other words, whereas in the van Oss, Chaudhury and Good model W AB 0 . WAAB A The Lewis acid-base concept is general enough to include traditional ion-ion and dipoledipole repulsions and thus it may not be unreasonable to suggest the existence of repulsive interactions [27]. Furthermore, entropic effects may contribute to the overall repulsion.

7.4 7.4.1

Determination of Surface Free Energy using IGC and AFM Application of the Fowkes Method to IGC Data

As discussed above, Fowkes [20, 21] first suggested that surface free energy could be considered as a sum of components resulting from different classes of intermolecular interactions.

Surface Free Energy Determination of Powders and Particles 327

The van Oss, Chaudhury and Good, and Chang-Chen models both draw upon the idea of Fowkes and as such all use the geometric mean approximation to model Lifshitz-van der Waals interactions. Fowkes [17, 20, 32] however, suggested a different approach to evaluating the acid-base character of surfaces (also see [43]). Fowkes criticized the use of contact angles for determination of interfacial properties [19]. His approach is, for experimental reasons, more applicable to powdered samples (e.g. pharmaceutical powders). As stated previously,

WALW WAAB

WA

(43)

WAAB is then, according to Fowkes, expressed by the following relation

WAAB

f N

H AAB

(44)

where N is the number of sites per unit area and

f

1

lnWAAB ln T

(45)

and

f ≈ 0.2 ... 1.0

(46)

When using the Fowkes approach some authors have taken f as unity although this does not seem to be a good approximation [43]. Because f and N are generally not known, direct calculations of the work of adhesion are often not made. Determination of HAB for multiple probe liquids on a given solid together with models by Drago [44] or Gutmann [45, 46] can be used to assess the acid-base nature of the surface (also see [47]). Lewis acid-base interactions encompass hydrogen bonding, electron donor-acceptor and organic nucleophile-electrophile interactions. Mulliken theory [48] is based on a valence bond model for the ground state energy of the donor-acceptor complex. This model considers the complex in terms of two resonance forms – non-bonded and ionic (electrons fully transferred). This notion of covalent and Coulombic contributions forms the basis for the Drago parameters (also see, for instance [49]). According to Drago

H AB

E A EB C A C B

(47)

where Ei’s are the electrostatic susceptibility parameters and Ci’s are the covalent susceptibility parameters. Drago parameters allow one to easily consider the hardness of the acidbase interaction. Hardness is related to polarizability. Substances with low polarizability are referred to as hard and those with high polarizability are said to be soft. In Drago’s model, E/C is a measure of hardness. Drago’s equation shows the preference for interaction with molecules of like hardness. While useful, Drago’s model does not allow one to consider easily the amphoteric nature of many materials (For a discussion of Hard and Soft Acids and Bases see [50, 51]).

328 Progress in Adhesion and Adhesives, Volume 4

Many adsorbates are amphoteric; therefore, it becomes necessary to recognize the dual nature of many compounds. Gutmann [45, 46] introduced the notion of electron-donor numbers (DNs) and electron-acceptor numbers (ANs) [47]. These parameters are similar to the van Oss, Chaudhury and Good surface free energy parameters as they both describe the same molecular parameters but from different points of view. In 1966, Gutmann [45] introduced the donor number based on the interaction with SbCl5. DN has units corresponding to enthalpy (e.g. kJ/mol). In 1975, Mayer et al. [52] introduced the acceptor number based on the relative 31P shift induced by triethylphosphine oxide. AN has arbitrary units. In 1990, Riddle and Fowkes [47] removed the dispersion component from AN. The corrected AN* values have the usual units of enthalpy. According to Gutmann, water is somewhat more nucleophilic. Chang [25, 27] also finds that PHb O PHa O which suggests that water 2 2 is somewhat more nucleophilic than it is electrophilic. These conclusions contrast with van Oss’ assumption that water is equally electrophilic and nucleophilic [24]. (Again, the question of the electrophile/nucleophile balance for water remains an open issue as does the exact relation between each of the various acidity scales proposed in the literature [33–36].) See Etzler [14] for Tables of AN, AN* and DN. According to Gutmann’s theory

H AB

K a DN K d AN *

(48)

Where Ka and Kd reflect, respectively, the acceptor and donor characteristics of a solid. Gutmann’s model works best with hard (low polarizability) atoms.

7.4.2

Application of the van Oss, Chaudhury and Good Method to IGC Data

Das and co-workers [53] have recently illustrated the use of van Oss, Chaudhury and Good model to IGC data. They [53] have determined the surface free energy components of lactose monohydrate (Pharmatose 450 M). The discussion by Theilmann and co-workers [54] is also useful. The first task was to measure the sorption isotherms of alkanes. A fit to the BET sorption isotherm was used to determine the monolayer coverage amount (nm). VN, the specific retention volume,was then calculated as below (Eqn. (50)) except that VN ( Eqn. (56) in section 5.1) is calculated at different surface coverages. This latter step allows for the calculation of surface free energy as a function of surface coverage. (See section 5.1 for further experimental details) As discussed above, the free energy of adsorption is split into two components:

Gads

LW Gads

AB Gads

(49)

The LW contribution to adsorption can be calculated using Eqns. (57) and (58) as described in section 5.1. The acid - base components are calculated using the following relation: AB Gads

I sp

2N Aa

1/ 2 L S

1/ 2 S L

(50)

Surface Free Energy Determination of Powders and Particles 329

Where NA is Avogadro’s number, a is the surface area of the molecule. In the present instance [53] a monopolar acid (dichloromethane, + = 5.20 mJ/m2) and monopolar base (ethylacetate, – = 19.20 mJ/m2, + = 0.01 mJ/m2) were used to determine +S and S– . For lactose monohydrate the surface free energy values were LW = 250 mJ/m2 and AB = 50 mJ/m2 at infinite dilution. These values appear to be rather high for an organic material and it is not clear why the surface free energy values are so high. The mixing of acid and base parameters determined by contact angle (those determined by van Oss and co-workers) and IGC data deserves further consideration. The surface free energy components determined in the study by Das et al. [53] are dependent on surface coverage. The values decline with surface coverage. In contrast to Fowkes’ method, Das’ method only requires data to be collected at one temperature.

7.4.3

Application of the Chang-Chen Model to IGC Data

The present author is not aware of the application of the Chang –Chen model to IGC data. It  would appear possible to use an approach similar to that of Das et al. [53] using the Chang-Chen model to calculate surface free energy of solids. As in the above section, mixing probe liquid properties obtained by contact angle measurements [14] with IGC data may deserve further consideration.

7.4.4

AFM Methods

The Atomic Force Microscope (AFM) has been shown, in recent years, to be a valuable tool for studying surfaces in a number of ways. A volume edited by Drelich and Mittal discusses many aspects regarding the design and use of AFM [55]. AFM can be used to measure the adhesion force between the cantilever or colloidal particle and a solid surface. As discussed below the adhesion force can be used, in principle, to also determine the surface free energy. The AFM measures the adhesion force over a small area of the sample and thus offers an advantage over other methods discussed in this work as it can be used to study surface heterogeneity. Relative determinations of surface free energy may be adequate for the study of surface heterogeneity. Here we explore briefly the use of AFM for the determination of surface free energy. Mazzola et al. [56] as well as Beach and co-workers [57] have discussed the use of AFM in the present context. The DMT (Derjaguin, Muller, Toporov) model has been used to describe the relation between the force of adhesion, Fad, and the work of adhesion between a spherical particle and a flat surface. The particles interact primarily through Lifshitz-van der Waals attractions. According to the DMT model the relation is

Fad

2 RWA

(51)

where R is the particle radius. The DMT model has been successfully used when particles are small and the surface is rigid. This model, furthermore, assumes that there is no adhesion hysteresis, which is physically implausible.

330 Progress in Adhesion and Adhesives, Volume 4

The JKR (Johnson, Kendall, Roberts) model is in popular use. The JKR model allows for adhesion hysteresis unlike the DMT model above. The model also considers long range interactions outside the area of contact. According to the JKR model, the force of adhesion is given by the following equation.

Fad

3 RWA 2

(52)

Work by Maugis [58] suggests that the JKR and DMT models are the limiting cases. It appears that the “best” model has not been, as yet, agreed upon. The JKR model does, however, appear to be a better choice for many organic surfaces. Beach and co-workers [57] have determined the surface free energy of methyl terminated monolayers using AFM (See Table 1.). These results have been compared with results obtained using contact angle measurements (van Oss, Chaudhury and Good model). The observed variability appears to be consistent with expected errors in the tip radius and cantilever spring constant. Mazzola and co-workers [56] have recently proposed a model for determining surface free energy that uses a nanoindenter equipped with a spherical tip. In this method, the interaction of the tip with the surface as the tip approaches is recorded. The AFM method described above, in contrast to the present case, uses pull-off forces. Here the work of adhesion is calculated from pressure, P, changes as tip moves from some distance hmax to contact the surface (h = 0). The work of adhesion thus is:

WA

1 P h 2 max max

(53)

For a complete discussion and derivation of the above equation see Mazzola et  al. [56]. Table 2 compares the surface free energies of several polymers determined using the method of Mazzola to those calculated from contact angle data using the Owens and Wendt model. The polymers studied are polycarbonate (PC), poly(methyl methacrylate) (PMMA), polypropylene (PP), poly(tetrafluoroethylene) (PTFE) and acrylonitrile-butadiene-styrene (ABS). Mazzola and co-workers note that when surface features are smaller

Table 7.1 Surface Free Energy of CH3- Terminated Self-Assembled Monolayers Determined by AFM and Contact Angle Measurements by the van Oss, Chaudhury and Good (vOCG) Method [57]. A comparison of DMT and JKR models is shown. SV

Cantilever

(mJ/m2) DMT

SV

(mJ/m2) JKR

#1

24.3 ± 9.3

32.4 ± 21.2

#2

26.9 ± 9.4

35.9 ± 22.4

(mJ/m2) Contact Angle (vOCG) SV

24.6 ± 0.9

Surface Free Energy Determination of Powders and Particles 331

Table 7.2 Surface Free Energy (mJ/m2) of Polymers Determined by Mazzola’s Method and by Contact Angle (Owens and Wendt) [56].

Mazzola’s Method

PC

PMMA

PP

PTFE

ABS

80.1 ± 21.7

74.8 ± 17.9

39.7 ± 36.1

20.4 ± 4.9

44.6 ± 13.7

50.3

45.3

38.4

20.1

44.9

Owens & Wendt

It appears that AFM methods offer less precision in the determination of surface free energy of solids when compared to contact angle methods. AFM methods do, however, offer advantage when measuring surface heterogeneity at small scale. Presumably, future advances in technology will increase the usefulness of AFM for the determination of surface free energy. Contact angle measurements are a poorer choice for heterogeneous surfaces.

than 10 nm, large differences in surface free energies determined by the two methods are observed.

7.5

Characterizing Surface Properties by Inverse Gas Chromatography

As discussed above, IGC provides information on the Lifshitz - van der Waals component of surface free energy ( LW) as well as the Lewis acid and Lewis base parameters for material surfaces (KA and KB respectively if Fowkes’ model combined with Gutmann’s acid-base model is chosen). It has been recognized that materials having a monofunctional acidic (or basic) surface do not interact strongly with each other [59]. Knowledge of the surface free energy of individual formulation components allows for an understanding of components interactions in multi-component systems. Below we discuss how surface free energy and surface free energy variations may affect processes and product performance. Acid-base interactions are an important consideration in developing pharmaceutical formulations made from various materials that include both excipients and active drug ingredients. If the Fowkes model and the Gutmann acidbase model are used, for instance, acid-base (or electron acceptor-donor) interaction parameters (PAD) can be calculated by matching KA of one component with KB of another component [60, 61]: Whether PAD should be maximized or minimized will depend on the application. Strong adhesion is desired in some applications but minimal adhesion is required in others. Here

PAD

K A , e K B ,i

1/ 2

K B , e K A ,i

1/ 2

,

(54)

where subscripts e and i are for excipient and active ingredient, respectively. Larger values of PAD indicate greater adhesion between materials. The body of surface science literature reveals differences in adopting KA and KB values for calculating the acid-base interaction parameters expressed in Eqn. (54). The KA or KB values of a solid sample are investigated in IGC experiments using a range of acid-base (polar) liquids. The acid-base probes are usually characterized by acceptor and donor numbers representing the Lewis acidic and basic character of the probe liquids. The donor number, DN [62], has units of energy/mol (Table 3). The acceptor numbers have two types of scaling: AN [52] which is unitless or dimensionless, and AN* [45] which is in energy/mol. Acceptor and donor

332 Progress in Adhesion and Adhesives, Volume 4

numbers are used to calculate KA and KB values based on the specific (or acid-base) enthalpies (ΔHadssp) of adsorption of a range of polar probes on a specific solid sample [63]: sp H ads

K A donor no.

K B acceptor no. .

(55)

Referring to Eqn.( 55), the units, and hence values, of the calculated KA or KB are expected to differ according to whether AN* or AN is used as the acceptor number. Table 4 summarizes the units of KA and KB using different calculation approaches. Calculations using AN* are assigned as Approach 1, and those using AN as Approach 2. Approach 1 is used for determination of PAD [60]., and Approach 2 is employed for calculation of Ia–b, thus the resulting value, though assigned an arbitrary unit, is in energy/mol (see Table 4). Ia-b is equivalent to PAD except AN has been used instead of AN* . When the calculated interaction parameter is in units of energy/mol, it can be designated as the enthalpy of the acid-base interaction between an excipient and active ingredient (ΔHAB) ( For a more detailed discussion, see [64]). The Fowkes approach (Section 4.1) does not allow for the calculation of the overall surface free energy of the material as the acid-base scale is based on enthalpy not free energy.

7.5.1

IGC Measurements - Experimental Considerations

For IGC measurements, materials in powder form are packed into a column. Generally, the column is both straight and short (a few centimeters). Vapors of analytical-grade Table 7.3 Properties of the Probes Commonly Used in IGC Experiments. Polarizability index (1049 C3/2m2V-1/2)

DN (kcal/ mol)

AN

AN* (kcal/ mol)

Specific characteristic

n-Hexane

9.2

−

−

−

Non-polar

n-Heptane

10.3

−

−

−

Non-polar

n-Octane

11.4

−

−

−

Non-polar

n-Nonane

12.5

−

−

−

Non-polar

a

n-Decane

13.6

-

-

-

Non-polar

n-Undecanea

14.7

-

-

-

Probe

b

Non-polar b

Chloroform

7.8

−

23.1

4.8

Acidic

Acetone

5.8

17.0

12.5

2.5

Amphoteric

Ethyl acetate

7.9

17.1

9.3

1.5

Amphoteric

Diethyl ether

7.3

19.2

3.9

1.4

Basic

Tetrahydrofuran

6.8

20.0

8.0

0.5

Basic

Note: Polarizability index from [67], DN from [71], AN from [72], and AN* from [45] a From Mills et al. [73] b The AN value for chloroform is in accord with the original work of Mayer et al. [52] as opposed to the 25.1 stated in [47]. The AN* for chloroform was calculated using 23.1 as AN, based on the formula established by Riddle and Fowkes [47].

Surface Free Energy Determination of Powders and Particles 333

Table 7.4 Units of Quantities Involved in the Calculations of KA and KB, and Acid-Base Interaction Parameters. Acceptor number

Donor number, DN

KA

KB

PAD

Ia-b

1

Energy/mol (AN*)

Energy/mol

Unitless

Unitless

Unitless

–a

2

Unitless (AN) Energy/mol

Unitless

Energy/mol

–a

Energy/ mol

Approach

Note: a Not indicated as it is irrelevant to the approach and interaction parameters adopted by references [60, 67].

alkanes and acid-base probes are individually sampled from vials containing the probe liquids. A small amount of the probe is injected into the IGC column through the inlet port at a specified temperature. The amount of probe in column should be such that the sorption is in the Henry’s Law portion of the isotherm (unless finite dilution IGC is desired). This is referred to as the infinite dilution region. The probe material is carried through the IGC column by purified helium gas. The eluted probe is detected by a flame ionization or thermal conductivity detector near the outlet port which is maintained at a specified temperature. The resulting chromatogram is integrated using software to determine the retention time of the probe. The IGC experiments are typically conducted at a minimum of three different temperatures (Fowkes’ model) to allow for calculation of ΔHads. If Fowkes’ acid-base model is not used it is possible to use a single temperature. The retention time of the probe is used to calculate the net specific retention volume as [65]:

VN

J

273.15 Q(t r t m ) TW

(56)

where VN is the net specific retention volume per gram of sample (ml/g), J is the James-Martin correction factor, which corrects the retention time for the pressure drop in the column, T is the column temperature (K), W is the weight (g) of sample packed in the column, Q is the flow rate (ml/min) of carrier gas, tr is the retention time (min) of the probe and tm is the retention time (min) of a reference gas (typically methane) which is used to determine the dead volume of the column. The net specific retention volume for a homologous series of n-alkane probes is used to determine the free energy of adsorption of a methylene group on sample surfaces [66]:

Gads (

CH 2 )

RT VN (Cn 1H2n 4 ) ln , VN (Cn Hn 2 ) 1000

(57)

where [ Gads( CH2 )] is the free energy of adsorption (in kJ/mol) of a methylene group, R is the gas constant (8.3145 J K-1 mol-1), n is the number of carbon atoms of the alkane probes

334 Progress in Adhesion and Adhesives, Volume 4

which typically ranges from hexane (n = 6) to undecane (n = 11). The factor 1000 in the equation converts the energy unit from J to kJ. Eqn. (57) expresses the difference in (RT ln VN) for every one increment of carbon atom of an alkane probe. Therefore, [ Gads( CH2 ) ] is normally obtained from the slope of plots (RT ln VN) versus the number of carbon atoms of alkane probes. (see example given in Figure 1) The [ Gads( CH2 )] value calculated is used to determine the Lifshitz - van der Waals component of the surface free energy of the solid samples [66]:

1 4

LW s

2 Gads ( (CH 2 )

CH 2 )

N 2 a2

1012

(58)

where sLW is the Lifshitz-van der Waals component of the solid surface free energy (in mJ/m2), [ (CH )] is the surface free energy of pure methylene groups (35.6 mJ/m2, assum2 ing a close-packed structure as in polyethylene), N is Avogadro’s number (6.0221 × 1023 mol-1), (equals 6 × 10–20 m2 or 6 Å2) is the area of an adsorbed methylene group, and the factor 1012 converts [–ΔGads(–CH –)]2 from kJ2/mol2 to mJ2/mol2. Note that the literature often 2 refers to the Lifshitz- van der Waals component as the London dispersion component. More aggrievedly, this component is sometimes referred to as the dispersive component of the surface energy. It is neither dispersive nor an energy. There are three approaches used to extract the specific free energy (ΔGsp) and subsequently (ΔHsp) if necessary. The approaches are the 1. probe polarizability approach 2. probe vapor pressure approach 3. molecular area approach These approaches are discussed and referenced below. The free energy of specific (acid-base) adsorption can be determined using the probe polarizability approach which is shown to be adequate for describing interactions between an isolated (gas) molecule and a solid surface. Probe polarizability as a molecular descriptor has the advantage of not showing, for adsorption of certain acid-base probes, a total free energy of adsorption that appears to be lower than the corresponding Lifshitz-van der Waals component. Such a problematic observation can be experienced, especially for solid surfaces having a large Lifshitz-van der Waals component of the surface free energy [67], when using other descriptors involving either probe vapor pressure [68] or molecular area [69]. The use of the probe polarizability approach is an attempt to avoid measurement difficulties, so that the free energy of specific (acid-base) adsorption can be obtained by subtracting the non-polar component from the total free energy of adsorption. (See Figure 1 for an illustration) In brief, the specific, or Lewis acid-base, interaction of an acid-base probe with the solid sample is calculated from the following equation: sp Gads

VN RT ln ref , 1000 VN

(59)

Surface Free Energy Determination of Powders and Particles 335

where Gspads is the free energy of specific (acid-base) adsorption (in kJ/mol), VN is the net specific retention volumes (in ml/g) of a polar probe observed from experiment and calculated from equation (57), and VNref is the net specific retention volumes (in ml/g) predicted from the plot of RTlnVN versus polarizability index (tabulated in Table 4) for a series of n-alkanes ( also see Figure 1). If Fowkes’ acid-base model is used, the specific interaction for an acid-base probe is plotted versus temperature. The enthalpy ( Hspads; in kJ/mol) and entropy ( Sspads; in kJ mol-1 K-1) of adsorption of the particular polar probe with test samples are determined, respectively, from the intercept and the slope based on the thermodynamic function: sp Gads

sp sp H ads T Sads

(60)

Alternatively, a plot of Gspads/T vs 1/T can be made. The slope will be Hspads The Hspads for a series of acid-base probes subsequently allows determination of the Gutmann acid and base parameters (KA and KB) of sample surfaces using the Gutmann acceptor and donor numbers of the probes [63] as shown in equation (61). Equation (61) can be rewritten as a linear function so that KA (dimensionless) and KB (dimensionless) for Approach 1 (Table 4) can be obtained respectively from the slope and the intercept of the plots: sp H ads

4.184 AN

DN KA AN *

*

(61)

KB ,

12

C10

RT ln VN (kJ/mol)

8 THF C9 4 CHCl3 Ether 0

C6H6 Isp

C8

C7 CCl4

C6 –4 160

200

240 a(

280

LW 1/2 L

)

320

360

400

[Å2(mJ)1/2m]

Figure 7.1 RTln(VN) versus ( LLW)1/2. Squares: alkane liquids. Triangles: other liquids. Slope of linear fit to alkane liquids points gives sLW and vertical displacement of triangular points from linear fit gives Isp. Isp (ΔGab) is the free energy resulting from acid-base interactions, x- axis reflects the use of the molecular area approach. For the other approaches, the x-axis differs as described in the text.

336 Progress in Adhesion and Adhesives, Volume 4

where the factor 4.184 converts the unit of Hspads (kJ/mol) to kcal/mol to be consistent with the unit of the denominator AN*. When AN is used (Approach 2), the KA (dimensionless) and KB (kJ/mol) values can be obtained respectively from the slope and the intercept of the plots: sp H ads

DN K AN A

4.184 AN

(62)

KB ,

where the denominator 4.184 converts the unit of DN (kcal/mol) to kJ/mol to be consistent with the unit of Hspads and KB. In the probe polarizability approach, the polarizability index is given by [(h L)1/2 O,L], where h is the Planck’s constant (6.626 × 10–34 J s), is the characteristic electronic frequency (in s-1) of the probe, 0 is the deformation polarizability (in C m2 V-1) of molecules, and the subscript L refers to the probe liquid. The energy term, h , is calculated from:

1 2

h

or h

1 2

e2 , 0me

(63)

(1.602)2 1.986 10 0 8.187

37

(64)

where e is the elementary charge (1.602 x 10–19 Coulomb), me is the mass of electron (9.109 × 10–31 kg or the mass -energy equivalent (E = mc2) of 8.187 × 10–14 J), O is in C m2 V-1, and the factor 1.986 is for converting the calculated h value to the unit of J. Equation (64) can be simplified to:

h

5.56 10

38

1

(65)

0

Based on Equation (65), the polarizability index, (h )1/2 0, (in C3/2 m2 V-1/2) can be expressed as: 1/ 2

(h )1/2

5.56 10

38

1

0

0 0

3/ 4

or (h )1/2

0

2.358 10

19

0

(66)

The values of the polarizibility index [(h )1/2 0] for IGC probes in Table 3 (Column 2) had been calculated in the same way by Donnet et al. [67] .

Surface Free Energy Determination of Powders and Particles 337

Regardless of the molecular descriptors (vapor pressure, polarizability, etc.), the total free energy of adsorption (ΔGads; in kJ/mol) is generally expressed as [66]:

VN Ps , g RT ln , 1000 106 S s

Gads

(67)

where R is the gas constant (8.3145 J K-1 mol-1), T is the column temperature (in K), VN is the net specific retention volume per gram of sample (in ml/g), S is the specific surface area of the sample (in m2/g), Ps,g is the adsorbate vapor pressure in the gaseous standard state (1.013 × 105 Pa or N/m2), and πs is the surface or spreading pressure of the gas in the standard adsorption state (3.38 × 10-4 N/m according to De Boer’s definition of the standard state (twodimensional ideal gas at 1 atm), and 6.08 × 10-4 N/m according to Schreiber’s definition (see [70] for further details)). The factor 1000 in Eqn. (67) converts the energy unit from J to kJ, and the factor 106 converts the retention volume from ml to m3. Using a constant, K1, equal 10 6 Ps , g , Equation (67) can be rewritten as: to S s

Gads Gads

RT 1000

lnVN

RT lnVN 1000 Gads

ln K1 , or RT ln K1 , or 1000

RT lnVN 1000

C

(68)

where C represents [10-3 RT ln K1], which is a constant value for an IGC column at a given temperature. In the probe polarizability approach, the London dispersion interaction between an adsorbate (probe) and an adsorbent (solid sample) is equal to the potential energy of interaction between two non-identical molecules, expressed as [67]: d Gads

1 3 N 1000 4 4 0 d Gads

6

1 2

h

rS , L

1 K h 1000

1/ 2

1/ 2 S

0,S

S

0,S

h

1/ 2 L

h

1/ 2 L

0, L ,

0, L ,

or

(69)

(70)

where ΔGdads is the London dispersion component of the free energy (in kJ/mol) of adsorption, is the permittivity in vacuum (8.8542 × 10-12 C2 J-1 m-1), rS,L is the distance between adsoro bent (solid; subscript S) and adsorbate (liquid; subscript L) molecules (assuming constant as

338 Progress in Adhesion and Adhesives, Volume 4

0.3 × 10-9 m), K in Equation (70) is the collective constant (in J2 C-4 m-4 mol-1) involving N, , , and rS,L of Equation (69). The factor 1000 converts the energy unit from J to kJ. o For adsorption of non-polar probes, the London interaction (ΔGdads) is also the total free energy of adsorption of the probes (ΔGads), and thus combining Eqns. (68) and (70), the following equation is obtained:

Gads RT lnVN 1000

d Gads ,

C

or

(71)

K (h S )1/2 1000

0,S

(h L )1/2

0, L .

Based on Eqn. (71) a linear regression can be performed on the plots of [RTlnVN] versus [(h L)1/2 0,L] (the polarizability index tabulated in Table 3) for a series of n-alkanes ( also see Figure 1). The intercept obtained includes the constant, C. The slope obtained is the representation of [K(h S)1/2 0,S] which is related to the London dispersion component ( dS) of the solid surface, and is characteristic of a given solid sample. For adsorption of polar probes, the same regression constants established in the preceding paragraph can be used. The polarizability index [(h L)1/2 O,L] of the polar probe liquid (from Table 3) is inserted into the regression formula obtained using alkane probes to predict the value of [RTlnVN], and equation (71) can be more specifically expressed as:

RT lnVNref 1000

C

K (h )1/2 1000 S

0,S

(h

1/ 2 PL )

0 , PL ,

(72)

where the superscript ref means value predicted from the reference alkane line, and the subscript PL refers to polar probe liquid. The corresponding VNref is thus the VN for the polar probe if it behaved as an alkane. The term on the right-hand side of Eqn. (72), based on Eqn. (70), can be replaced by –ΔGadsd (in kJ/mol):

RT lnVNref 1000

d , Gads

C

(73)

The specific, or Lewis acid-base, interaction of a polar probe [ΔGspads; in kJ/mol] with the solid sample is calculated by subtracting the London dispersion component from the total free energy of adsorption: sp Gads sp Gads

Gads

d Gads , or

Gads

(74)

d Gads

By inserting Eqns. (68) and (73), Eqn. (74) becomes: sp Gads

RT RT lnVN C lnVNref 1000 1000 RT RT lnV lnVNref VN 1000 1000

C (75)

Surface Free Energy Determination of Powders and Particles 339

Equation (75) is graphically depicted in Figure 1, where the –ΔGspads (Isp) value is determined from the vertical difference between a point of [RTlnVN] (see Figure 1) and the corresponding alkane line. Eqn. (75) can also be simplified to Eqn. (59).

7.5.2

Finite Dilution IGC

Surface free energies determined by IGC often differ from those determined using contact angles. Infinite dilution IGC measurements are thought to reflect surface sites with the highest energies while surface free energies reflect the average energy of the saturated surface. To partially compensate for the nature of the occupied surface sites, Das and co-workers [74] have described the use of IGC at finite dilutions. The Lifshitz-van der Waals component of surface free energy is determined as described in section 5.1 above. In this instance, the value of VN is determined at finite and specified surface coverage. As in previously discussed examples

G

G LW

G AB

ΔGAB is determined using the following relation.

G AB

I sp

2N Aa

1/ 2 L S

1/ 2 S L

where +L and –L are the van Oss’ values for the acid and base prameters of surface free energy obtained from earlier contact angle measurements. The application of the van Oss parameters to the present problem is an appropriate subject for scientific controversy. While Das and co-workers [74] did not apply the Chang-Chen model in their analysis, there is no reason that it could not be employed in the analysis. If IGC measurements were made at several temperatures thus allowing for the calculation of ΔHAB then the Fowkes’ model could also be employed. Figure 2 shows the surface free energy components for Pharmatose 450 (a grade of lactose monohydrate). The data in the figure are those of Das et al. [74] The figure shows that the surface free energy components generally decrease in an asymptotic manner. Yao and co-workers[75] have determined the surface free energy of powdered pearl oyster shell using contact angles determined using the Washburn method to determine contact angles (see [14, 76]) and IGC at infinite dilution and finite dilution. Table 5 summarizes their results. The data in the table suggest that surface free energies determined using contact angles and IGC may differ. This difference may result from the use of van Oss parameters determined using contact angles or the fact that even finite dilution IGC most often refers to surface coverages far less than a single monolayer. The Washburn method relies upon the wicking of liquid into a powder bed to determine the contact angle between the liquid and solid substrate. This method is generally unsuitable for pharmaceuticals as most pharmaceuticals are soluble in one or more of the necessary probe liquids.

340 Progress in Adhesion and Adhesives, Volume 4 T

250

AB LW

(mJ/m2)

200

150

100

50

0 0

0.02

0.04 n/nm

0.06

0.08

Figure 7.2 The surface free energy components of Pharmatose 450 as a function of surface coverage. The figure is redrawn from data by Das et al. [74]. The asymptotic decline of the surface free energy components with surface coverage is typical. Table 7.5 Surface Free Components of Powdered Pearl Oyster Shell. Data are by Yao et  al. [75]. Note that the surface free energy components determined by IGC differ from those obtained using contact angles. T

LW

AB

+

0.005

-

vOGC

41.4

40.6

0.8

IGC ID

82.4

59.3

23.1

12.1

30.1 12.9

IGC FD

65.3

48.0

17.3

5.8

9.3

ID - infinite dilution FD - finite dilution vOCG - van Oss, Chaudhury and Good

7.6

Pharmaceutical Applications

As indicated in the beginning of this paper, surface free energy and variations in surface free energy can affect both industrial process and product quality. Here we will consider a few subjects relevant to the pharmaceutical industry. Sun [77] has also recently discussed the role of surface free energy in pharmaceutical processes.

7.6.1

Surface Free Energy and Crystal Planes

The surface free energies of various crystal planes of form I paracetamol were determined by Heng and co-workers.[78]. (Also see Williams [79]). In this work it was found that the

Surface Free Energy Determination of Powders and Particles 341

various crystal planes have unique values of surface free energy. Crystallization processes that result in different fractions of the various crystal planes would produce a material which may behave differently when subsequently processed. Crystallization from different solvents, for instance, may produce material with different fractions of the various crystal faces. Figure 3 shows the surface free energy of the faces of form 1 paracetamol vs. the surface OH group density. Ho and co-workers [80] similarly investigated the surface free energy of various crystal planes of D-mannitol. These authors also observed distinct surface free energies for each of the investigated crystal planes.

7.6.2

Compaction of Tablets

Etzler and Pisano [81] had previously proposed a model for understanding the tensile strength of tablets based on tablet porosity and material surface free energy. This model is based on the Ryshkewitch-Duckworth equation which describes the relation between tablet tensile strength and tablet porosity. The model shows the importance of the material surface free energy. Both tablet porosity (contact area) and surface free energy are import factors contributing to tablet tensile strength. Trasi and co-workers [82] have also investigated the relationship between tablet hardness and the Lifshitz-van der Waals component of surface free energy. These investigators studied glucose monohydrate that was subsequently dehydrated at various temperatures. Figure 4 shows the relation to tablet tensile strength.

75

001

70 011

(mJ/m2)

65

201

60 110 55

010

50 45 0

0.004

0.008 0.012 OH density (Å–2)

0.016

0.02

Figure 7.3 The surface free energy of form 1 paracetamol vs the surface OH density of various crystal planes. Data by Heng and co-workers [78].

342 Progress in Adhesion and Adhesives, Volume 4

8 140

Tablet hardness (kP)

7 6 5 4 120

100 80

3 2 32

34 LW

36 (mJ/m2)

38

40

Figure 7.4 Tablet hardness (tensile strength) in kiloponds (kP) versus LW for glucose monohydrate dehydrated at various temperatures. Numbers refer to the dehydration temperature in Celsius degrees. Data by Trasi et al. [82]

Fichtner and co-workers[83] have also commented on the relation between tablet tensile strength and material surface free energy.

7.6.3

Effects of Processing on Surface Free Energy

The crystallization of pharmaceuticals and subsequent processing of pharmaceutical powders may affect the surface free energy of pharmaceuticals. Etzler and co-workers [84] investigated the role of surface contamination in the adhesion between lactose particles. It appears that lactose is contaminated with protein materials contained in milk. The contamination is a source of variation in surface chemistry, particle adhesion and surface free energy. Cares-Pacheco and co-workers[85] have studied polymorphs of d-mannitol by finite coverage IGC. Cryomilled and spray dried mannitol were also investigated. Table 6 shows the Lifshitz-van der Waals component of surface free energy for the polymorphs and treated materials at a surface coverage of 0.1. Otte and Carvajal [86] studied the effect of cryomilling on the surface free energy of two pharmaceutical compounds, Griseolfulvin and Ketoconazole . Their results also indicate an increase in LW with milling. The major increase in surface free energy occurs after approximately 1 min of milling. Milling for longer times has only a small effect on surface free energy. Plots of ln (VN/T) vs. 1/T become non-linear with increasing milling time. This behavior suggests the creation of amorphous material at the surface. This amorphous material appears to undergo a glass transition at a particular temperature.

Surface Free Energy Determination of Powders and Particles 343

Luner and co-workers [87] demonstrated that milling affects the surface free energy of succinic acid (see Figure 5). Dry milling and wet milling using different solvents were explored. This suggests that wet milling produces material with the highest ( LW). Zhang and co-workers [88] have similarly shown surface free energy differences between crystalline and amorphous lactose as measured by AFM pull-off force. The amorphous form has the greater surface free energy. Patera and co-workers [89] have noted that lactose samples manufactured by DMV (Amersfoort, The Netherlands) and Meggle (Wasserburg, Germany) have different values for LW. Both spray dried lactose and milled and seived (200 mesh) lactose were investigated. These authors also noted that milled lactose exposed to 40 °C and 90% RH Table 7.6

LW

for Various Forms of D-Mannitol. Form

LW

(mJ/m2) 72.1 70.7

-cryomilled

39.4 -spray dried

51.7

-cryomilled

52.4

Data from Cares-Pacheco and co-workers [85].

45

35

LW

(mJ/m2)

40

30

TB M d

m ille

W

et

et W

E

A IP d m ille

m Dr y

Un m

ille

ille d

d

25

Figure 7.5 LW of succinic acid subjected to various types of milling. IPA - isopropanol. MTBE methyl t-butyl ether. Data by Luner et al. [87]

344 Progress in Adhesion and Adhesives, Volume 4

exhibited a decreasing value of LW over time. LW decreased from approximately 37 mJ/m2 to 28 mJ/m2 over 15 hours. It also appears that the acidity of the surface is reduced over time. Environmental factors thus may alter the surface free energy of pharmaceutical materials. Planinesk et al. [90] have described the creation of surface amorphous material on indomethacin particles during milling. In general, the surface amorphous content increases with milling time. As surface amorphous content increases so does LW. Figure 6 shows the dependence of LW with surface amorphous content.

7.6.4

Performance of Dry Powder Inhalers

Etzler and co-workers [84] have commented extensively on the relation between lactose surface contamination and the performance of dry powder inhalers. Lot-to-lot variations in surface contamination appear to contribute to lot-to-lot variations in the performance of dry powder inhalers. Traini and co-workers [91] investigated the performance of dry powder formulations containing salbutamol sulfate and various crystal forms of lactose. An NGI (Next Generation Impactor) was used to determine the percentage of particles in the respirable size range (the fine particle fraction, FPF). Particles whose diameter was less than 4.46 μm were considered to be respirable. Figure 7 summarizes the results of this study. Stank and Steckel [92] investigated milling salbutamol sulfate with magnesium stearate or glycerol monostearate. Milling with a lubricant allowed for the production of smaller particles and for the reduction of particle surface free energy. 48

40

LW

(mJ/m2)

44

36

32 0

20

40 60 80 % Amorphous by area

100

Figure 7.6 LW of indomethacin vs surface amorphous content. Data by Planinsek et al. [90]. Amorphous content increases with milling time.

Surface Free Energy Determination of Powders and Particles 345

Alpha-monohydrate

20

FPF (%)

16

Beta-anhydrous

12 Beta-treated

8 Alpha-anhydrous

4 120

140

160 180 (mJ/m2)

200

220

Figure 7.7 Fine particle fraction (FPF) vs surface free energy of various lactose forms . Figure redrawn from data by Traini et al. [91] The treated beta-lactose has been purified to remove traces of alpha-lactose. Beta-anhydrous lactose is as received and contains some alpha-lactose. The decreasing FPF with increasing surface free energy is consistent with increased interparticle adhesion at high surface free energy.

7.6.5

Powder Flow

Shah and co-workers [93, 94] have studied the flow properties of mefenamic acid powders. The flow properties were studied on as received and milled samples. Furthermore, samples that were silanized with alkyl groups were also investigated. Silanization is the process of reacting surfaces with siloxane compounds (e.g. dimethyldichlorosilane) in order to alter the surface chemistry (see references in Shah for details [93, 94]) In addition to the flow properties, the surface free energies of the various samples were also determined. It was found that surface free energy increased with the number of milling cycles, while silanization reduced the surface free energy of the material. Silanization was found to reduce the angle of friction and cohesion in the powder. The cohesion and the angle of friction are determined by plotting yield stress versus normal stress of a powder in a shear cell. The cohesion is the intercept and the angle of friction is the angle of the linear plot of yield stress versus normal stress with respect to the normal stress axis (see [93, 94] for experimental details.) A higher angle of friction corresponds to higher friction between powder particles. The results indicate that both particle size and surface free energy are important properties contributing to powder flowability. Shah and co-workers [93, 94] results are partially summarized in Figure 8.

346 Progress in Adhesion and Adhesives, Volume 4

4

50

Cohesion (kPa)

3 30 2 20 1

10

Angle of friction (deg)

40

n/nm = 0.007 0

0 48

52

56

60 LW (mJ/m2)

64

68

Figure 7.8 Cohesion (circles) and angle of friction (triangles) vs LW of mefenamic acid measured at n/nm = 0.007(infinite dilution). Increasing surface free energy corresponds to an increased number of milling cycles. Data by Shah and co-workers [93, 94].

7.7

Summary

The surface free energy of solids is an important factor contributing to product and process performance. The surface free energy of solids can be determined using contact angles, Atomic Force Microscopy (AFM) and Inverse Gas Chromatography (IGC). The latter two methods are useful on powdered samples. Inverse gas chromatography is experimentally the least labor intensive when used on powdered samples and thus is more convenient for the study of pharmaceutical materials. For pharmaceutical materials, solubility of the material in probe liquids is a particular concern when employing contact angle methods. It is nearly impossible to select an appropriate set of liquids to determine surface free energy without encountering a major problem with solubility. The use of IGC avoids this issue and is thus an important method for the characterization of pharmaceutical materials. In this paper, the basic models used for the determination of surface free energy of solids are reviewed. Particular emphasis is given to the determination of surface free energy by inverse gas chromatography although other methods are discussed as well. It is observed that surface free energies of a material by IGC, AFM and contact angles may differ. In the case of AFM, the differences are likely attributable to experimental difficulties and the ability to adequately sample the surface. In the case of infinite dilution IGC, differences in surface free energies determined by contact angles and IGC likely result as only the surface sites with the highest energy are occupied by the probe molecules. Differences between the two methods are also observed when finite dilution IGC is used. In the case of finite dilution IGC, it must be recognized that the surface coverage is still much less than 1 monolayer. Furthermore, probe liquid properties derived from contact angles are frequently used in

Surface Free Energy Determination of Powders and Particles 347

the calculations. It is not clear that the use of probe liquid properties derived from contact angle experiments are appropriate when using IGC measurements. Despite the inadequate level of current understanding, it is clear from the literature that the surface free energy of pharmaceuticals is an important quantity to be considered for product performance. This literature is clear that various processes, such as milling, can influence the surface free energy of pharmaceuticals. Thus, the surface free energy of a particular solid may not be constant but dependent on the previous history of the material. The literature shows that surface free energy is important to the performance of dry powder inhalers. Such dosage forms are particularly sensitive to changes in interparticle and particle -surface adhesion. A series of papers, discussed above, show that surface free energy is a contributing factor in determination of tablet tensile strength. Tablets must have sufficient tensile strength to be marketable. In the future, careful consideration to the role of surface free energy and its possible variation in product and process performance should be given. At present, this quantity is often ignored.

References 1. K. L. Mittal and F. M. Etzler, Is the world basic?: Lessons from surface and adhesion science, in: Contact Angle, Wettability and Adhesion, Vol. 6, K. L. Mittal (Ed.), pp. 111–123, CRC Press, Boca Raton, FL (2009). 2. F. M. Etzler, Characterization of surface free energies and surface chemistry of solids, in: Contact Angle, Wettability and Adhesion, Vol.3, K. L. Mittal (Ed.), pp. 219–264, CRC Press, Boca Raton, FL (2003). 3. M. Zenkiewicz, Methods for the calculation of surface free energy. J. Achievements Mater. Manuf. Eng. 24, 137–145 (2007). 4. R. R. Deshmukh and A. R. Shetty, Comparison of surface energies using various approaches and their suitability. J. Appl. Polym. Sci. 107, 3707–3717 (2008). 5. A. W. Adamson, The Physical Chemistry of Surfaces, 4th edition, John Wiley, New York (1990). 6. G. Beilby, Aggregation and Flow of Solids, MacMillan and Co., London, UK (1921). 7. R. Eötvös, Uber den zausammenhang der oberflachenspannung der flussigkeiten mit molecularvolume. Weid. Ann. 27, 448–459 (1886). 8. E. A. Guggenheim, The principle of corresponding states. J. Chem. Phys. 13, 253–261 (1945). 9. I. Langmuir, The distribution and orientation of molecules, in: Colloid Symposium Monograph, pp. 48–75, The Chemical Catalog Company, New York (1925). 10. T. Young, On the cohesion of fluids, in: Miscellaneous Works, Vol. 1, G. Peacock (Ed.), p. 418, Murray, London (1855). 11. A. Dupre, Theorie Mechanique de la Chaleur, Gauthier-Villars, Paris (1869). 12. P. S. Laplace, Mechanique Celeste, Supplement to Book 10, J.B.M. Dupart, Paris (1806). 13. E. W. Washburn, Dynamics of capillary flow. Phys. Rev. Ser. 2 17, 374–375 (1921). 14. F. M. Etzler, Determination of the surface free energy of solids: A critical review. Rev. Adhesion Adhesives 1, 3–45 (2013). 15. W. A. Zisman, Relation of equlibrium contact angle to liquid and solid constitution, in: Contact Angle, Wettability and Adhesion, Adv. Chem. Ser. No. 43, pp. 1–51, American Chemical Society, Washington, DC (1964).

348 Progress in Adhesion and Adhesives, Volume 4

16. K. L. Mittal, The role of the interface in adhesion phenomena. Polym. Eng. Sci. 17, 467–473 (1977). 17. F. M. Fowkes and M. A. Mostafa, Acid-base interactions in polymer adsorption. Ind. Eng. Chem. Prod. Res. Dev. 17, 3–7 (1978). 18. F. M. Fowkes, K. L. Jones, G. Li, and T. B. Lloyd, Surface chemistry of coal by flow microcalorimetry. Energy Fuels 3, 97–105 (1989). 19. F. M. Fowkes, Role of acid-base interfacial bonding in adhesion. J. Adhesion Sci. Technol. 1, 7–27 (1987). 20. F. M. Fowkes, Calculation of the work of adhesion by pair potential summation. J. Colloid Interface Sci. 28, 493–505 (1968). 21. F. M. Fowkes, Determination of interfacial tensions, contact angles, and dispersions forces in surfaces by assuming additivity of intermolecular interactions in surfaces. J. Phys. Chem. 66, 382 (1962). 22. R. J. Good, Contact angle, wetting and adhesion: A critical review, in: Contact Angle, Wettability and Adhesion, K. L. Mittal (Ed.), pp. 3–36, CRC Press, Boca Raton, FL (1993). 23. L. A. Girifalco and R. J. Good, Theory for the estimation of surface and interfacial energies. I. Derivation and application to interfacial tension. J. Phys. Chem. 61, 904–909 (1957). 24. C. J. van Oss, Interfacial Forces in Aqueous Media, Marcel Dekker, New York (1994). 25. F. Chen and W. V. Chang, Applicability study of a new acid- base interaction model in polypeptides and polyamides. Langmuir 7, 2401–2404 (1991). 26. D. Berthelot, Sur le melange des gaz. Compt. Rend. 126, 1857 (1898). 27. W. V. Chang and X. Qin, Repulsive acid - base interactions: Fantasy or reality?, in: Acid-Base Interactions: Relevance to Adhesion Science and Technology, Vol. 2, K. L. Mittal (Ed.), pp. 3–54, CRC Press, Boca Raton, FL (2000). 28. J. Panzer, Components of solid surface free energy from wetting measurements J. Colloid Interface Sci. 44, 142–161 (1973). 29. D. K. Owens and R. C. Wendt, Estimation of the surface free energy of polymers J. Appl. Polym. Sci. 13, 1741–1747 (1969). 30. S. Wu, Compression mechanics of polymer monolayers. J. Polym. Sci., Part C 34, 265–271 (1971). 31. S. Wu, Polymer Interface and Adhesion, Marcel Dekker, New York (1982). 32. F. M. Fowkes, Donor-acceptor interactions at interfaces. J. Adhesion 4, 155–159 (1972). 33. C. Della-Volpe and S. Siboni, Troubleshooting of surface free energy acid-base theory applied to solid surfaces: The case of the Good, van Oss and Chaudhury theory, in: Acid-Base Interactions: Relevance to Adhesion Science and Technology, Vol. 2, K. L. Mittal (Ed.), pp. 55–90, CRC Press, Boca Raton, FL (2000). 34. C. Della-Volpe, Some reflections on acid–base solid surface free energy theories. J.  Colloid Interface Sci. 195, 121–136 (1997). 35. C. Della-Volpe, A. Deimichei, and T. Ricco, A multiliquid approach to the surface free energy determination of flame-treated surfaces of rubber-toughened polypropylene. J.  Adhesion Sci. Technol. 12, 1141–1180 (1998). 36. C. J. van Oss, Irrelevance of the ratio of the electron-accepticity to the electron donicity of water with respect to the determination of polar surface tension components, interfacial tensions and free energies of interaction of liquids and/or solids, in: Acid-Base Interactions: Relevance to Adhesion Science and Technology, Vol. 2, K. L. Mittal (Ed.), pp. 173–180, CRC Press, Boca Raton, FL (2000). 37. D. Y. Kwok, The usefulness of Lifshitz-van der Waals/acid-base approach for surface tension components and interfacial tensions. Colloids Surfaces A 156, 191–200 (1999).

Surface Free Energy Determination of Powders and Particles 349

38. R. J. Good, On the acid/base theory of contact angles, in: Acid-Base Interactions: Relevance to Adhesion Science and Technology, Vol. 2, K. L. Mittal (Ed.), pp. 167–171, CRC Press, Boca Raton, FL (2000). 39. D. J. Gardner, S. Q. Shi, and W. T. Tze, Comparison of acid-base characterization techniques on lignocellulosic surfaces, in: Acid-Base Interactions:Relevance to Adhesion Science and Technology, Vol. 2, K. L. Mittal (Ed.), pp. 363–384, CRC Press, Boca Raton, FL (2000). 40. E. N. Dalal, Calculation of surface tensions. Langmuir 3, 1009–1015 (1987). 41. F. M. Etzler, J. Simmons, N. Ladyzhynsky, V. Thomas, and S. Maru, Assessment of acid - base character of polymer surfaces from contact angle and other surface chemical data, in: Acid - base interactions: Relevance to Adhesion Science and Technology, Vol. 2, K. L. Mittal (Ed.), pp. 385–397, CRC Press, Boca Raton, FL (2000). 42. F. M. Etzler, Determination of the surface free energy of solid surfaces: Statistical considerations, in: Advances in Contact Angle, Wettability and Adhesion, Vol. 3, K. L. Mittal (Ed.), pp. 301–330, Wiley-Scrivener, Beverly, MA (2018). 43. M. D. Vrbanac and J. C. Berg, The use of wetting measurements in assessment of acid-base interactions at solid-liquid interfaces, in: Acid-Base Interactions: Relevance to Adhesion Science and Technology, K. L. Mittal and H. R. Anderson Jr. (Eds.), pp. 67–78, CRC Press, Boca Raton, FL (1991). 44. R. S. Drago, G. C. Vogel, and T. E. Needham, Four-parameter equation for predicting enthalpies of adduct formation. J. Amer. Chem. Soc. 93, 6014–6026 (1971). 45. V. Gutmann, A. Steininger and E. Wychera, Donorstarken in 1,2 dichlorathan. Montash. Chem. 97, 460–467 (1966). 46. V. Gutmann, The Donor-Acceptor Approach to Molecular Interaction, Plenum, New York (1978). 47. F. L. Riddle and F. M. Fowkes, Spectral shifts in acid-base chemistry. 1. Van der Waals contributions to acceptor numbers. J. Amer. Chem. Soc. 112, 3259–2364 (1990). 48. R. S. Mulliken, Lewis acids and bases and molecular complexes J. Chem. Phys. 19, 514–515 (1951). 49. J. C. Berg, Role of acid-base interactions in wetting related phenomena, in: Wettability, J. C. Berg (Ed.), pp. 75–144, Marcel Dekker, New York (1993). 50. L. H. Lee, Relevance of the density-functional theory to acid-base interactions and adhesion in solids. J. Adhesion Sci. Technol. 5, 71–92 (1991). 51. R. J. Pearson, Hard and soft acids and bases. J. Amer. Chem. Soc. 85, 3533–3539 (1963). 52. U. Mayer, V. Gutmann, and W. Gerger, The acceptor number - a quantitative empirical parameter for the electrophilic properties of solvents. Montash. Chem. 106, 1235–1257 (1975). 53. S. C. Das, I. Larson, D. A. V. Morton, and P. J. Stewart, Determination of polar and total surface energy distributions of particulates by inverse gas chromatography. Langmuir 27, 521–523 (2011). 54. F. Theilmann, M. Naderi, D. Burnett, and H. Jervis, Investigation of the acid-base properties of an mcm-supported ruthenium oxide catalyst by inverse gas chromatography and dynamic vapour sorption, in: Catalysis in Application, S. D. Jackson, J. S. J. Hargraves and D. Lennon (Eds.), pp. 233–239, Royal Society of Chemistry, London, UK (2003). 55. J. Drelich and K. L. Mittal (Eds.), Atomic Force Microscopy in Adhesion Studies, CRC Press, Boca Raton, FL (2005). 56. L. Mazzola, M. Sebastiani, E. Bemporad, and F.Carassiti, An innovative non-contact method to determine surface free energy on micro areas. J. Adhesion Sci. Technol. 26, 131–150 (2012). 57. E. R. Beach, G. W. Tormoen, and J. Drelich, Pull-off forces measured between hexadecanethiol self-assembled monolayers in air using an atomic force microscope. J. Adhesion Sci. Technol. 16, 845–868 (2002).

350 Progress in Adhesion and Adhesives, Volume 4

58. D. Maguis, Adhesion of spheres: The JKR-DMT transition using a dugdale model. J.  Colloid Interface Sci. 150, 243–269 (1991). 59. J. C. Berg, The importance of acid-base interactions in wetting, coating, adhesion and related phenomena. Nordic Pulp Paper Res. J. 8, 75–85 (1993). 60. J. M. Felix, P. Gatenholm, and H. P. Schreiber, Plasma modification of cellulose fibers: Effects on some polymer composite properties. J. Appl. Polym. Sci. 51, 285–295 (1994). 61. S. J. Park and J. B. Donnet, Anodic surface treatment on carbon fibers: Determination of acidbase interaction parameter between two unidentical solid surfaces in a composite system. J. Colloid Interface Sci. 206, 29–32 (1998). 62. V. Gutmann and E. Wychera, Coordination reactions in nonaqueous solutions - the role of the donor strength. Inorganic Nuclear Chem. Letters 2, 257–260 (1966). 63. C. Saint-Flour and E. Papirer, Gas-solid chromatography: Method of measuring surface free energy characteristics of short fibers. 2. Through retention volumes measured near zero surface coverage. Ind Eng. Chem.: Product Res. Development 21, 666–669 (1982). 64. E. Pisanova and E. Mäder, Acid–base interactions and covalent bonding at a fiber–matrix interface: Contribution to the work of adhesion and measured adhesion strength, J. Adhesion Sci. Technol. 14, 415–436 (2000). 65. J. R. Conder and C. L. Young, Physicochemical measurements by gas chromatography, WileyInterscience, New York (1979). 66. G. M. Dorris and D. G. Gray, Adsorption of n-alkanes at zero surface coverage on cellulose paper and wood fibers. J. Colloid Interface Sci. 77, 353–362 (1980). 67. J. B. Donnet, S. J. Park, and H. Balard, Evaluation of specific interactions of solid surfaces by inverse gas chromatography. Chromatographia 31, 434–440 (1991). 68. C. Saint-Flour and E. Papirer, Gas-solid chromatography: A quick method of estimating surface free energy variations induced by the treatment of short glass fibers. J. Colloid Interface Sci. 91, 69–75 (1983). 69. J. Schultz, L. Lavielle, and C. Martin, The role of the interface in carbon fibre-epoxy composites. J. Adhesion 23, 45–60 (1987). 70. P. Mukhopadhyay and H. P. Schreiber, Aspects of polymer surface characterization by inverse gas chromatography. J. Polym. Sci.: Part B: Polym. Phys. 32, 1653–1656 (1994). 71. V. Gutmann, Coordination Chemistry in Non-Aqueous Solutions., Springer-Verlag, New  York (1968). 72. U. Mayer, A semiempirical model for the description of solvent effects on chemical reactions,. Pure Appl. Chem. 51, 1697–1712 (1979). 73. R. H. Mills, D. J. Gardner, and R. Wimmer, Inverse gas chromatography for determining the dispersive surface free energy and acid–base interactions of sheet molding compound—part II: 14 ligno cellulosic fiber types for possible composite reinforcement J. Appl. Polym. Sci. 110, 3880–3888 (2008). 74. S. C. Das, I. Larson, D. A. V. Morton, and P. J. Stewart, Determination of the polar and total surface energy distributions of particulates by inverse gas chromatography. Langmuir 27, 521–523 (2011). 75. Z. Yao, D. Wu, J. Y. Y. Heng, S. Lanceros-Méndez, E. Hadjittofis, W. Su, J. Tang, H. Zhao, and W. Wu, Comparative study of surface properties determination of colored pearl-oyster-shellderived filler using inverse gas chromatography method and contact angle measurements. Intl. J. Adhesion Adhesives 78, 55–59 (2017).

Surface Free Energy Determination of Powders and Particles 351

76. E. Chibowski, L. Holysz, and A. Szczes, Wettability of powders, in: Adhesion in Pharmaceutical, Biomedical and Dental Fields, K. L. Mittal and F. M. Etzler (Eds.), pp. 23–52, Wiley-Scrivener, Beverly, MA (2017). 77. C. C. Sun, The role of surface free energy in powder behavior and tablet strength, in: Adhesion in Pharmaceutical, Biomedical and Dental Fields, K. L. Mittal and F. M. Etzler (Eds.), pp. 75–88, Wiley-Scrivener, Beverly, MA (2017). 78. J. Y. Y. Heng, A. Bismark, A. F. Lee, K. Wilson, and D. R. Williams, Anisotropic surface energies and wettability of macroscopic form 1 paracetamol crystals. Langmuir 22, 2760–2769 (2006). 79. D. R. Williams, Particle engineering in pharmaceutical solids processing: Surface energy considerations. Current Pharmaceutical Design 21, 2677–2694 (2015). 80. R. Ho, S. J. Hinder, J. Watts, S. E. Dilworth, D. R. Williams, and J. Y. Y. Heng, Determination of the surface heterogeneity of d-mannitol by sessile drop contact angle and finite concentration inverse gas chromatography. Intl. J. Pharm. 387, 79–86 (2010). 81. F. M. Etzler and S. Pisano, Tablet tensile strength: Role of surface free energy, in: Advances in Contact Angle, Wettability and Adhesion, Vol. 2, K. L. Mittal (Ed.), pp. 397–418, Wiley-Scrivener Beverly, MA (2015). 82. N. S. Trasi, S. X. M. Boerrigter, S. R. Byrn, and T. M. Carvajal, Investigating the effect of dehydration conditions on the compactability of glucose. Intl. J. Pharm. 406, 55–61 (2011). 83. F. Fichtner, D. Mahlin, K. Welch, S. Gaisford, and G. Alderborn, Effect of surface energy on powder compactibility. Pharm. Res. 25, 2750–2759 (2008). 84. F. M. Etzler, R. Deanne, T. R. Burk, T. H. Ibrahim, G. A. Willing, and R. D. Neuman, The effect of the acid-base chemistry of lactose on its adhesion to gelatin capsules: Conclusions from contact angles and other surface chemical techniques., in: Contact Angle, Wettability and Adhesion, Vol. 2, K. L. Mittal (Ed.), pp. 13–43, CRC Press, Boca Raton, FL (2002). 85. M. G. Cares-Pacheco, R. Calvert, G. Vaca-Medina, A. Rouilly, and F.Espitalier, Inverse gas chromatography a tool to follow physico-chemical modifications of pharmaceutical solids: Crystal habit and particle size effects. Int. J. Pharm. 494, 113–126 (2015). 86. A. Otte and M. T. Carvajal, Assessment of milling disorder of two pharmaceutical compounds. J. Pharm. Sci. 100, 1793–1804 (2011). 87. P. E. Luner, Y. Zhang, Y. A. Abramov, and M. T. Carvajal, Evaluation of milling method on the surface energetics of molecular crystals using inverse gas chromatography. Cryst. Growth Design 12, 5271–5282 (2012). 88. J. Zhang, S. Ebbens, X. Chen, Z. Jin, S. Luk, C. Madden, N. Patel, and C. J. Roberts, Determination of the surface free energy of crystalline and amorphous lactose by atomic force microscopy adhesion measurement. Pharm. Res. 23, 401–407 (2006). 89. J. Patera, P. Zamostny, D. Litva, Z. Berova, and Z. Belohav, Evaluation of functional characteristics of lactose by inverse gas chromatography. Procedia Engineering 42, 644–650 (2012). 90. O. Planinsek, J. Zadnik, M. Kunaver, S. Sriic, and A. Godec, Structural evolution of indomethacin particles upon milling: Time-resolved quantification and localization of disordered structure studied by IGC and DSC. J. Pharm. Sci. 99, 1968–1981 (2010). 91. D. Traini, P. M. Young, F. Thielmann, and M. Acharya, The influence of lactose pseudopolymorphic form on salbutamol sulfate - lactose interactions in dpi formulations. Drug Develop. Ind. Pharmacy 34, 992–1001 (2008). 92. K. Stank and H. Steckel, Physico-chemical characterization of surface modified particles for inhalation. Intl. J. Pharm. 448, 9–18 (2013).

352 Progress in Adhesion and Adhesives, Volume 4

93. U. V. Shah, D. Olusanmi, A. S. Narang, M. A. Hussain, M. J. Tobyn, and J. Y. Y. Heng, Decoupling the contribution of dispersive and acid-base components of surface energy on the cohesion of pharmaceutical powders. Intl. J. Pharm. 475, 592–596 (2014). 94. U. V. Shah, D. Olusanmi, A. S. Narang, M. A. Hussain, M. J. Tobyn, S. J. Hinder, and J. Y. Y. Heng, Decoupling the contribution of surface energy and surface area on the cohesion of pharmaceutical powders. Pharm. Res. 32, 248–259 (2015).

8 Understanding Wood Bonds–Going Beyond What Meets the Eye: A Critical Review Christopher G. Hunt1*, Charles R. Frihart1, Manfred Dunky2 and Anti Rohumaa3 1

Forest Products Laboratory, Madison, Wisconsin, USA Kronospan GmbH Lampertswalde, Lampertswalde, Germany 3 FibreLaboratory, South-Eastern Finland University of Applied Sciences (XAMK), Savonlinna, Finland 2

Abstract Understanding why wood bonds fail is an excellent route toward understanding how to make them better. Certifying a bonded product usually requires achieving a specific load, percent wood failure, and an ability to withstand some form of moisture exposure without excessive delamination. While these tests protect the public from catastrophic failures, they are not very helpful in understanding why bonds fail. Understanding failure often requires going beyond what meets the naked eye, conducting additional tests, probing the wood surface, the fracture surface, adhesive properties, and the interaction of wood and adhesive during bond formation and service. This review of wood bond analysis methods reviews fundamentals of wood bonding and highlights recent developments in the analyses and understanding of wood bonds. It concludes with a series of challenges facing the wood bonding community. Keywords: Wood adhesive, wood bond, microscopy, penetration, failure

8.1

Introduction: Macroscopic Knowledge for Successful Adhesive Bonding of Wood

Although adhesives have been used for bonding wood products since before recorded history [1], making bonded wood products is a continuous challenge because of the everchanging wood sources and products, greater performance needs, and desire for lower costs. In recent decades, engineered wood products have mainly replaced solid wood for structural and other wood applications. All these changes and their interacting effects have driven a continuous demand for bonding a wider variety of wood substrates to achieve a

*Corresponding author: [email protected]

K.L. Mittal (ed.) Progress in Adhesion and Adhesives, Volume 4 (353–420) © 2019 Scrivener Publishing LLC

353

354 Progress in Adhesion and Adhesives, Volume 4

higher performance than in the past [2]. These products must perform for decades or even centuries without failing. Good adhesives and bonding systems have been developed through intelligent empirical approaches and application of the understanding of wood bonds to date. As a result, over 65% of wood products are now bonded [3]. However, this does not mean that adhesive manufacturers and users are done innovating. The continuous change of the market and the unresolved questions addressed in Section 5 of this review present significant challenges [4]. To meet these challenges in a timely and cost-effective way, wood bonding professionals need to understand how bonds work. More importantly, they also need to understand why bonds fail, and this often requires using methods that go beyond standard tests and visual evaluations. They must go beyond what meets the eye. Studying the mechanisms responsible for wood adhesive bonding has been an important aspect of wood science research over the past 50 years [2, 5]. It has been proposed that a better understanding of wood adhesion mechanisms at the micro and nanoscale has the potential to accelerate development of better adhesive systems using more efficient and effective processing methods for the wide array of wood products [4, 6, 7]. Creating an effective wood bond requires good adhesive wetting, efficient solidification of the adhesive to provide strength, and sufficient deformability of the cured adhesive to reduce the stresses that occur during joint formation and to avoid stress concentrations [8]. This review will first discuss wood properties, the preparation of surfaces, and the application of adhesives. Then we discuss bond performance, followed by examples of investigations that go beyond what meets the naked eye, and a survey of techniques. Finally, we discuss unresolved questions in the field. The first step in bonding is the preparation of the substrate (wood) surface so that it has minimal surface contamination and, for most products, is smooth. A weak surface layer can lead to bond failure [9–11], but it can be hard to prove the presence of a weak layer because that normally requires techniques beyond simple visualization. Several researchers have emphasized that in wood, such weak boundary layers can be caused by chemical or physical deficiencies, or both [5, 12]. Consequently, a closer examination of the wood surface is needed beyond what can be felt by hand or visualized by eye. The general theories of adhesion have been intensely studied for a wide variety of substrates [7, 13–17]. For any bonding to take place the adhesive must spread across and wet the wood surface to develop molecular interactions between the adhesive and wood. The maximal surface interactions occur when the surface energy of the adhesive and the wood are nearly equal [18]. Wood is an unusual substrate because of its abundant micrometer scale, interconnected voids. Adhesive flows into these voids and provides better mechanical interlocking with a larger surface area compared to most other substrates [19]. However, just mechanical interlock is not able to provide sufficient strength without intermolecular interactions [20]. These specific interactions are enhanced in those adhesives capable of penetrating wood cell walls at the molecular scale [21]. Some adhesives have components that diffuse into the wood cell wall (cell wall penetration, or infiltration) and polymerize

Understanding Wood Bonds–Going Beyond What Meets the Eye 355

there, potentially providing a continuous chain of covalent bonds from the bulk adhesive into the cell wall. Because wood adhesives penetrate below the surface, it is necessary to differentiate the zone of pure adhesive, which we call the glueline, from the entire zone impacted by the presence of adhesive, which we call the bondline. In other words, the bondline contains both the glueline and the interphase. Another unusual property of wood as a substrate is its great variability. Because of the cellular and multilayered composite nature of wood, the bonding surface can vary from the voids of open lumens, the various intercellular pitting patterns, and ends of ray cells to the relatively smooth middle lamella surfaces [22]. The variety of wood cell types and structures serves to increase the types of wood surfaces for bonding [5]. This problem becomes more complex when bonding chips, particles and fibers using a binder adhesive, which can be thought of as spot welding the wood together, as opposed to a continuous bondline in a laminated product. Because of the micrometer and molecular level penetration and wood variability, adhesion is highly dependent on a complex interaction of the adhesive formulation, production process, and the wood. After obtaining adhesion by wetting and penetration, the adhesive must solidify by moisture loss, chemical cure, or both, to develop good cohesive strength. Good internal strength (cohesion) is needed to ensure permanency of the bond and transfer load across the glueline, and is controlled by different factors than adhesion to the wood. Most wood adhesives develop cohesive strength through some type of chemical reaction (often condensation) that forms large intertwined polymers and gelled structures [23]. However, strong networks are often difficult to form near an interface with a substrate because the network formation is disrupted at the boundary [11]. Thus, cohesive failure in the adhesive may occur in the weak interphase layer while the bulk of the adhesive appears normal. Other causes of weak adhesive interphase in wood can be due to interference of cure by extractives from the wood, loss of reactive components to the wood, or alteration of the pH or moisture content by contact with the wood. Adhesives make possible an efficient use of wood resources to make a wide variety of products, most of which cannot be made out of a solid piece of wood. The variety of wood products, their performance expectations, and the anatomical variety within wood means that demands on the adhesive vary widely. Comparing the performance of a poly(vinyl acetate) (PVA) adhesive holding a chair together to a phenol-formaldehyde (PF) holding together the layers of structural laminated veneer lumber can seem ridiculous, but both depend on the wood surface, the interaction of the adhesive with the wood, and the demands placed on the bond. Both cases depend on developing molecular level contact between the adhesive and wood substrates and maintaining that contact while the joint is loaded. Obviously, we cannot easily see all molecular level contacts and failures in such a complex environment, but we can certainly determine what makes a good or bad bond at a finer level of examination than what is normally visible to the eye. This review is meant to assist the reader understand the fundamental process of wood bonding and choose from the many methods that have been developed, especially over the past few decades, for understanding wood adhesives at a microscopic level.

356 Progress in Adhesion and Adhesives, Volume 4

8.2

Bond Formation (Developing Adhesion)

Bond formation is the process of adhesion and cohesion; this means strong adhesion, or attraction between the substrate and the adhesive, as well as sufficient internal (cohesive) strength of the adhesive. Various interactions between substrate and adhesive determine adhesion strength; these interactions can be both physical and chemical in nature. For good adhesion, there must be molecular level contact between the adhesive and the substrate, which includes flow of adhesive over the substrate surface and into the irregularities of the surface to increase the contact area (interface) [14]. For most materials, this involves preparing the surface by solvent wipe, chemical treatment, or ablation to develop good contact between adhesive and substrate, while for wood, sawing, planing, chipping, fibrillation, and sanding are the most common methods. The problem faced daily by those making wood bonds is not how to make the ideal bond, but rather how to make satisfactory bonds with minimal cost, and with the available equipment and material. This involves balancing many contributing factors. For example, a less expensive wood source, or wood preparation method, might require more adhesive or a different adhesive to bond adequately. The influence of wood anatomy, surface preparation, and other production variables on bond quality have been generally described [5, 22–27]. Therefore, our discussion will briefly survey general principles and focus on recent discoveries. Because optimizing this balance of economy and quality requires understanding the role of each material in bond formation, the following sections will discuss the role of each component of bond formation in the order: wood properties, surface quality, adhesive penetration, and adhesive properties that influence penetration. Section 3 will then discuss the properties of bonded assemblies and tools for measuring penetration, section 4 discusses some examples of detailed approaches and methods, and section 5 points out some unresolved questions in the field of wood bonding.

8.2.1

Influence of Wood Structure on Bonding

How an adhesive interacts with the wood surface depends as much on the wood as it does on the adhesive’s properties. Wood is a complex material due its multi-layered composite structure and great variability. Wood properties can have important effects on wood bond performance, and these properties vary not only between but also within species and within the tree. Wood properties can influence bond formation primarily through a) adhesive flow (impacted by density, porosity, anisotropy, and grain angle), b) wetting (impacted by wood surface chemistry and extractives), c) rate and extent of adhesive cure (wood moisture content, pH, buffering capacity, and extractives), d) wood mechanical properties, damage and distortions, including those that occur during the bonding process, and e) swelling characteristics, as swelling can create internal stresses. In practice, the ideal bonding conditions are wood and process dependent. For instance, it is not unusual to find that in production of laminated beams, different pressures or adhesive formulations work better for different species. Therefore, we suggest using caution in

Understanding Wood Bonds–Going Beyond What Meets the Eye 357

drawing conclusions about the relative bond performance across species in a study without detailed evaluation of all the variations. Data within a species are usually more reliable, but still can be confounded by sapwood versus heartwood, the presence of juvenile or reaction wood, or even the age of the tree [5, 26, 28, 29]. For example, the elastic and shear moduli within individual annual rings show considerable variations around the trunk and at different heights [30, 31]. Because of the interaction between wood and adhesive properties, studies often use a single adhesive with different species of wood, or several adhesives with one species of wood. There are fewer studies that involve a matrix effect of different adhesives and wood species. Konnerth et al. showed that the adhesive performance was dependent on the adhesive, wood species, and test method, and that trying to extrapolate the adhesive performance from one wood species to another is risky [32]. Information used to classify wood species into bondability categories has been collected over several decades [33, 34], and has been summarized in Table 1. Though current adhesives or current commercially available wood quality might behave somewhat differently, the basic trends are still helpful. When encountering a new wood species for wood bonding, it is likely you will look up its properties. Many published wood properties values originated in the Wood Handbook, published by the Forest Products Laboratory (FPL) and is free to the public on their website [35]. Many of the values in those tables were already present in the 1940 edition, meaning that the properties were largely determined using old growth (slow growing, mature) wood. The modern wood supply generally contains more knots, smaller diameters, and significant quantities of juvenile wood, which has generally inferior mechanical properties relative to mature wood [28, 29]. Reflecting the change in the wood supply over time, the standard mechanical properties values of the most important species group for structural applications in the US, the southern pines, were recently modified to lower values [36]. The user must exercise caution, however, because the southern pines are the only species group where we are aware that the values have been changed, and even if the average value in the Wood Handbook is accurate, there is no indication of the variability within species, which in wood can be quite large [26, 31, 37]. Many people assume that wood from trees that grow fast is inferior to old growth or slow growth wood. Softwood species do tend to have a higher percentage of earlywood with fast growth, resulting in lower density and consequently lower mechanical properties in general [26, 29, 37], as shown in Figure 1. This contributed to the lowering of mechanical properties values for engineering calculations just discussed. In ring porous hardwood species, however, faster growth rates lead to higher density and better mechanical properties [26]. This is because these woods produce a very low density layer of mostly vessels early in the spring, and dense fibers the rest of the year, as shown in Figure 1. Besides density, the larger or smaller number of vessels carrying adhesive away from the bondline might require changes in bonding conditions or adhesive used. While traditional bondability information is useful in general, it does not provide guidance on why specific wood bonds fail and does not help individuals solve their specific bonding problems. In Table 1, the easier to bond species are generally lower in density

Magnolia

Willow, black

Aspen

Basswood

Walnut, black

Yellow-poplar

Hackberry

Maple, soft

Tupelo

Rock

Sycamore

Elm

American

Sweetgum

Butternut

Bond well

b

Cottonwood

Chestnut, U.S.

Alder

Bond easilya

U.S. hardwoods

Ponderosa

Sugar

Pine

Larch, western

Douglas-fir f

Redwood

Pacific

Redcedar, Eastern

Spruce, Sitka

Redcedar, Western

Western white

Eastern white

Pine

Noble

Grand

White

Fir

U.S. softwoods

Table 8.1 Categories of selected wood species according to ease of bonding.

Wallaba

Sucupira

Sapele

Mahogany American

Peroba rosa

Limba

Spanish cedar

Opepe

Jarrah

African

Okoume

Obeche

Yellow

White

Light red

Meranti (lauan)

Roble

Purpleheart

Hura

Iroko

Banak

Avodire

Angelique

Andiroba

Afromosia

Determa

e

Courbaril

Cativo

Balsa

Non-U.S. Woods

358 Progress in Adhesion and Adhesives, Volume 4

e

e

Lapacho Lignumvitae Rosewood Teak

Balau Greenheart Kaneelhart Kapur

Persimmon

Keruing

Ramin

Radiata

Caribbean

Pine

Parana pine

Pau marfim

Meranti (lauan), dark red

Balata

Karri

Bubinga

Benge

Azobe

Angelin

Osage-orange

White

Red

Oak

Pines, southern

Port Orford cedar

Yellow cedar

b

Bond very easily with adhesives of a wide range of properties and under a wide range of bonding conditions. Bond well with a fairly wide range of adhesives under a moderately wide range of bonding conditions. c Bond satisfactorily with good-quality adhesives under well-controlled bonding conditions. d Satisfactory results require careful selection of adhesives and very close control of bonding conditions; may require special surface treatment. e Difficult to bond with some phenol-formaldehyde adhesive formulations. f Wood from butt logs with high extractive content is difficult to bond.

a

d

Bond with difficulty

Cherry

Maple, hard

Yellow

True

Birch

Madrone

Pecan

Beech, American

Sweet

Hickory

Ash, white

Bond satisfactorilyc

Understanding Wood Bonds–Going Beyond What Meets the Eye 359

360 Progress in Adhesion and Adhesives, Volume 4

Figure 8.1 Effect of growth rate on (Top row) softwood, southern pine group and (Lower row) ring porous hardwood, red oak group. Left to right: slow to fast growth. Images courtesy of Alex Wiedenhoeft, FPL Center for Wood Anatomy, using surface sanding method described in Section 3.4. Image height: top row 15.6 mm, bottom row 5.2 mm. Color figures online.

than those that are harder to bond. Lower density woods have more void space, allowing better adhesive flow into the wood, and also generally swell and shrink less with moisture, resulting in less swelling stress. It should also be noted that low density woods are generally weaker, so there is less stress on the bondline in lower density woods at failure. A major problem encountered with low density species is that it can be difficult to avoid overpenetration, where too much adhesive flows away from the bondline. For example, discussions with particleboard producers reveal that they believe overpenetration causes many problems when low density poplar furnish is used. In many cases, the reasons for poor bond performance have not been examined in much detail. Was it a wood preparation problem, a bonding problem leading to poor penetration, or a curing issue? An example where the authors tried to understand why failure took place with some adhesives was the use of 18 adhesives for bonding acetylated wood. Their evaluation of why the bonds failed was useful in solving the associated bonding problems [38]. As a followup to this study, additional insights into the problems in bonding acetylated wood has been published [39]. However, the literature is still limited in most part to bond strength and percent wood failure with a few analyses of the failure surfaces by microscopic or other techniques.

8.2.2

Influence of Wood Surface Quality on Bonding

It has long been recognized that the quality of the wood surface has a strong influence on the bond performance properties. Wood surface property issues include chemical

Understanding Wood Bonds–Going Beyond What Meets the Eye 361

heterogeneity, surface inactivation, weak boundary layers, and processing impacts, such as machining, drying, and aging [7]. The influence of wood surface preparation, as well as other production steps, on bond quality has been ably and extensively described elsewhere [5, 22, 24–26]. These references describe how the best bonds are made with freshly prepared, cleanly cut wood surfaces that mate to form uniformly thin bondlines. Here we review recent developments in the understanding of how surface preparation influences bond performance. 8.2.2.1 Mechanical Damage at the Wood Surface It did not take sophisticated modern techniques to realize that if the wood surface was damaged, the bond made at that surface was also compromised. The mechanical weak boundary layer (MWBL) is the damaged wood adjacent to the bondline [9, 10, 12], which must be supported or repaired by the adhesive. Even when making “perfect” surfaces, cells can be crushed or torn by planers and veneer peelers. Crushing of surface cells can occur if the surface is mechanically sanded, which is a major reason why sanding must be used with caution on surfaces to be glued. Sanding tends to crush or gouge out the earlywood on wood surfaces where there is a large density difference between earlywood and latewood. In some cases, this freshly prepared surface is an improvement, as in the case of particleboard and fiberboard where mechanical sanding is used to remove the surface layer that is overdried/ overcured and can contain a high wax content. Besides being intuitively straightforward, the MWBL and means to avoid it have been well described [5, 9, 22, 26]. Good wood adhesion is thought to rely on the adhesive penetrating to the depth of solidly attached fibers, i.e., below the MWBL, before cure [40]. Discussion of the MWBL here will focus on developments since those texts were written. In the production of veneer-based products such as plywood and laminated veneer lumber, lathe checks are invariably formed, but their depth and frequency can be influenced by processing conditions [22]. Lathe checks have long been suspected to lower bond strength and product quality but the literature is conflicting regarding the important parameters and mechanisms [41–45]. Pulling plywood specimens with lathe checks open was known to be detrimental to bond strength but the literature values of checks pulled open are quite varied: lower by 14% to 94% relative to checks pulled closed [42, 43, 45, 46]. Figure 2 shows a veneer in the process of peeling, with evident lathe checks, and the difference between pulling open and closed. The effect of lathe checks was greatly clarified by the use of a dye to clearly show (and thus quantify) lathe checks, and by sanding both sides of a veneer to produce veneers that were identical except for lathe check depth. With these innovations, a strong correlation was shown between the depth of lathe checks, as a percent of veneer thickness, and shear strength. The effect on strength was large when the bonds were pulled open, and much smaller when bonds were pulled closed as shown in Figure 3 [47]. The mechanism behind the lathe check effect was further explained by looking beyond the naked eye and closely observing (with magnification and a camera) samples during testing. Figure 4a clearly shows that lathe checks, when pulled open, induced a rolling

362 Progress in Adhesion and Adhesives, Volume 4

F

(b) (a)

F

Checks pulled open (c)

F

Checks pulled closed

F

Figure 8.2 (a) Lathe checks forming during the peeling process are visible on upper side of veneer, top right. The grid was painted on the end grain of the log before peeling (Forest Products Laboratory (FPL) photo). (b) Plywood tested with checks pulled open. (c) Plywood tested with checks pulled closed. Failure zones and applied loads are highlighted in b and c. b and c are adapted from [47]. 7

Dry: Open checks Dry: Closed checks Soaked: Closed checks Soaked: Open checks

Shear strength (MPa)

6 5 4 3 2 1 0 30

40

50

60 70 Depth of lathe checks (%)

80

90

Figure 8.3 Impact of lathe check depth and direction of testing (pulled open or closed) on observed shear strength of plywood, both soaked and dry samples. % on x axis indicates % of veneer thickness [47].

mode I failure through the veneer instead of the typical mode II failure observed when lathe checks were pulled closed and when checks were shallow (Figure 4b). Another interesting result is that mode I failures propagate very close to the bondline, while mode II

Understanding Wood Bonds–Going Beyond What Meets the Eye 363

(a)

(b)

(c)

(d)

Figure 8.4 (a) Localized mode I failure of plywood with deep (80% of veneer thickness) lathe checks when pulled open, (b) Mode II failure of plywood with deep lathe checks when pulled closed, (c, d) Typical failure surfaces of samples with deep checks pulled open and closed, respectively [47, 48]. 4a and 4b correspond to Figure 2b and 2c, respectively.

failures typically propagate deep in the wood, often from the tip of one check to the tip of a neighboring check [47]. Shallow wood failure is often considered a sign of poor adhesive performance, though in this case the depth of failure appears to be driven by the wood surface preparation far more than by the adhesive. This suggests that the depth of lathe checks and pull direction (open or closed) should be given more attention during bonding studies than they typically had in the past. The influence of lathe checks on LVL (laminated veneer lumber) properties has also been clarified. Finite element analysis has shown that lathe checks that do not get filled with adhesive are quite detrimental to edgewise shear modulus [49], though not to other properties. Other works suggest that lathe checks impact edgewise shear and other LVL properties as well [50, 51]. Surface mechanical preparation, such as sanding, planing or cutting, will produce different surface qualities and will also affect roughness. Generally, the roughening of a surface will improve bonding quality [52] and adhesives often form much stronger bonds to porous than to smooth surfaces [53], though extremely rough wood surfaces contribute to poor adhesive transfer through lack of contact and uneven bondline pressure [5]. Even though the role of topography is unclear [54], roughness parameters have been used to explain adhesive bond formation and quality [55]. The critical roughness values needed for optimal wood bonding are not fully understood and so far no consensus has been found [44, 56–58]. Sellers [59], quoting Walser [60], claimed the maximum roughness depth for acceptable veneer bonding is about 500 μm, while some authors claim that roughness differences of only tens of micrometers are important [55]. Some of these differences in the “optimal” roughness of veneer can be related to adhesive formulation. For instance, fillers can serve to prevent overpenetration (improve holdout) of the adhesives so that they can bridge the gap between surfaces. Different fillers can

364 Progress in Adhesion and Adhesives, Volume 4

have very different properties [61]. Also, it has been shown that adhesives with low elastic modulus tolerate thick bondlines much better than adhesives with modulus much higher than the wood being glued [62]. The differences in observed impact of roughness on bond quality might also be explained by the use of different wood species and the diverse measurement techniques used. The conflicting results might also be, in part, due to the fact that different peeling parameters to obtain different roughnesses will also change other surface properties such as the extent of damage to cells at or below the surface, or the frequency and depth of lathe checks. It is remarkable that there is no easy, widely used technique to assess the presence or severity of a MWBL in wood. To address this deficiency, a new method of quantifying the MWBL, referred to as the surface integrity in the paper, was recently developed [63]. The test consists of attaching a metal tab to a veneer surface with a double sided tape, and tearing the tab off. The tape is then imaged to determine the number and size of wood particles removed from the wood surface. It was shown that veneers with deep lathe checks lost many large fiber bundles, which could easily initiate failure in the product, while shallow lathe checks produced fewer and smaller bundles. On the tight side of veneers, which lack lathe checks, only single fibers or fiber fragments were removed. While this test was developed for veneer, we expect that with an appropriately strong adhesive (tape or otherwise), damaged cells from other types of surface preparations could also be detected. Another new development was to discover an interaction between the temperature of log soaking and peeling, and the felling season of birch logs. When soaked and peeled at 70 °C, contact angles and bond strengths were only slightly different between logs felled in different seasons. When peeled at 20 °C, however, the logs felled in winter had significantly higher contact angle and lower bond strength than spring or autumn felled logs [64]. While seasonal variation in the density of logs, presumably from extractives stored in the wood, had been documented, it was rather small and the authors did not expect it to impact bonding [65]. Many assumed that the differences in contact angles and bond strengths of veneers peeled at different temperatures originated in the different cutting behaviours of the wood at different temperatures. It was a surprise, then, to learn that birch logs heated to 70 °C and then cooled to 20 °C before peeling had contact angle and bond strength properties closer to veneers peeled at 70 °C than to those soaked and peeled at 20 °C [66]. This suggests that some of the changes in wood that occur at 70 °C soaking temperature are relevant to bonding and are not reversible by cooling. Whether the relevant changes are physical (increased cellulose crystallinity for instance) or chemical is not known. Moisture weakens wood, and temperature enhances this effect. It has been estimated that with a typical waterborne adhesive loading of 150 g/m2, the cells nearest the glueline can reach the fiber saturation point [67]. Therefore, these cells, especially earlywood, are more prone to buckle during pressing, creating a new MWBL, as shown in Figure 5. This is commonly seen in micrographs of solid wood bondlines, as in Figure 5a, and also occurs in composites. In particleboard, the wood structure and orientation on either side of a bond are extremely varied [68, 69], suggesting that conditions favorable for buckling will exist throughout the board. Cell wall buckling appears to be quite common in particleboard but

Understanding Wood Bonds–Going Beyond What Meets the Eye 365

(a)

(b)

Figure 8.5 Buckling of cell walls and subsequent MWBL from a combination of moisture, pressure, and heat. (a) bonded poplar radial faces, UF fluorescence emission is yellow, wood is blue. Glueline and collapsed vessels (CV) at top of image, distorted wood structure throughout is evidenced by kinks in ray cells, shown by white arrows. Unpublished image from study reported in [70]. (b) particleboard with light blue UF (urea-formaldehyde) adhesive and purple wood. Method reported in [71]. Color images online. Image from Hogger Elfriede, Wood K plus - Competence Centre for Wood Composites and Wood Chemistry, Linz, Austria. Scale bars: 100μm.

is very seldom described. Thus, the normal macroscopic view of wood surfaces, and lack of attention to this issue, may lead to an underestimation of this source of failures. 8.2.2.2 Surface Chemistry Barriers to Bonding In addition to a mechanical weak boundary layer (MWBL) on wood from surface damage, a chemical weak boundary layer (CWBL) can also exist, impeding bond formation [5, 7, 12]. A macroscopic inspection of the wood surface cannot reveal a CWBL: other methods that are sensitive to chemical properties are needed [72]. Wood species with large concentrations of nonpolar extractives are generally more difficult to bond [73]. Wood surfaces tend to become less hydrophilic over time, which impedes the spread of water based adhesives across the wood surface. Freshly planed surfaces had lower contact angles and better wetting, or contact between wood and waterborne adhesives, compared to surfaces that had been stored in the laboratory for only a few hours [74] or longer [75–77]. This is one reason some adhesive standards require planing of the wood surface “just prior to gluing” [78] or “within 24 hours of bonding” [79]. Good manufacturing practice is to apply the adhesive as soon as possible to the freshly cut surface, especially if the wood contains high levels of extractives. Contact angle with water, polar liquids, or the adhesive of interest is the standard analytical method of assessing surface wettability [80–85]. A low contact angle indicates that surface tension of the solid is high relative to the liquid, as described by Young’s equation. Low contact angles not only indicate adhesive spreading across the wood surface, but also that the adhesive will make molecular scale contact with the wood. The ideal interfacial strength occurs when the surface energies of the adhesive and wood are equal [18].

366 Progress in Adhesion and Adhesives, Volume 4

Contact angle also influences the rate that adhesive advances through a capillary such as a lumen [86]. Hse supported the utility of contact angle by showing correlation between low contact angle and a thinner bondline (increased void penetration) with 36 different PF formulations [87]. While the phenomenon of lower bond quality with aged surfaces is not disputed, the correlation between contact angle and bond strength is much weaker [80, 81]. Instead of a “lower is always better” relationship, contact angle can be viewed as having a threshold value for a given system. Too high a contact angle can prevent sufficient wetting and spreading and molecular contact between wood and adhesive, but once sufficient wetting is achieved, there is little further improvement by reducing the contact angle. Once an adhesive is able to fully wet the wood, further reduction in contact angle might promote overpenetration and/or distribution of the adhesive over too large an area [87]. Further caution regarding contact angle measurements was advised when Petricˇ concluded that in most cases it is useless to compare contact angle data on wood obtained in different laboratories with different wood specimens of the same species [88]. There are many factors that are likely to contribute to the gradual loss of surface energy at the wood surface [89–91]. The most likely source is migration of extractives like resin and fatty acids and their esters, waxes, sterols and terpenes to the wood surface. Waterborne adhesives with high pH are often used when high levels of extractives on the surface are suspected to be a problem, whereas neutral or acidic adhesives are typically less effective. There could also be chemical changes to the molecules on the surface, not only those already mentioned but also oligosaccharides, phenols and tannins migrating to the surface. In addition to lowering surface energy, these molecules that dissolve in the adhesive could potentially interfere with the cure of the adhesive [73, 92–94], or influence cure through altering adhesive pH [95–103]. This is especially the case with very acidic woods like oak, which have been known to interfere with cure if the adhesive is not strongly buffered [104]. The surface acidity of many wood species has been documented [105, 106]. In addition, fresh, high energy surfaces attract contaminants from the air which also lowers surface energy. Wiping the wood surface with solvent has also been shown to improve bonding, and is a common practice for small scale woodworkers and very oily woods [107]. An extreme version of surface inactivation comes from overdrying of wood after it has been cut, such as veneer for plywood LVL, or OSB (oriented strandboard) flakes. Surface inactivation by overdrying is likely a combination of many factors [24, 25]. Physical effects include degradation of the mechanical properties of the surface layer, and closing of surface cracks, thus reducing surface area. Chemical changes include migration of hydrophobic wood ingredients to the surface and chemical modification of surface bonding sites, which includes oxidation, molecular rearrangement of various functional groups on the surface, and elimination of hydroxyl sites [24, 25]. It has also been shown that even the time of year that a tree is harvested can impact wettability and bond performance, presumably because of different levels of stored sugars in the stem [64, 108]. Hanetho discussed the influence of seasonal variations on wood quality in particleboard production [109]. Freshly cut wood harvested in winter and used immediately caused problems. The cause was identified as high contact angles from high levels of extractives. Harvesting in different seasons or letting winter felled logs

Understanding Wood Bonds–Going Beyond What Meets the Eye 367

age both had the effect of decreasing contact angle, improving resin contact with the wood, and improving board properties. The lower contact angle achieved by letting logs age must not be confused with the higher contact angle typically experienced when cut surfaces age.

8.2.3

Adhesive Penetration

Having discussed wood structure and wood surfaces, we now turn to how adhesives penetrate into the wood by considering the following factors [5]: i. Wood-related parameters, such as wood species, diameter of the lumens and their exposure on the wood surface due to slope of grain, pit frequency and aspirations, cutting orientation, wood density, and moisture content. Also important is the presence of knots, bark, decay, stain, heartwood, juvenile wood, and occlusions from extractives or tyloses. Wood strength to withstand clamping pressure, wettability of the wood surface and surface energy also play a role. ii. The adhesive properties that control penetration such as chemical structure, molecular weight distribution, additives, solids content, viscosity [110] and surface energy, buffering capacity, hardening time and rate of resin curing or solidification [111]. iii. Processing parameters of bonding such as spread rate, open and closed assembly time, temperature, pressure [70, 112, 113], and moisture profile over time through the material. It should be noted that the temperature and moisture level of the wood surface and of the bondline continuously change the viscosity of the resin (which also depends on the degree of curing). Resin viscosity is also changed by selective loss of some components through cell wall penetration. Penetration is the ability of an adhesive to enter into the wood structure [5, 19, 110]. Unfortunately, the word penetration has long been used to mean not only wood void penetration (flow into cracks, voids, and lumens in the wood), but also cell wall penetration (molecular scale mixing of adhesive molecules between cell wall polymers) [110]. These are two different phenomena that are controlled by different factors and result in different physical and mechanical properties for the bonded product [6]. Void penetration into wood occurs on the millimeter to sub-micrometer level as a result of the hydrodynamic flow and wetting of the liquid adhesive. The adhesive flows from the glueline (adhesive material between the two wood surfaces) into the porous and capillary structure of wood, mostly filling cell lumens, as well as encapsulating fractures and surface debris caused by wood surface preparation. Also known as adhesive bulk flow, lumen filling, or tissue penetration, void penetration is governed by pore size (wood anatomy), bulk adhesive viscosity, external compression force from pressure applied to the wood, and wetting behavior [6, 110, 114]. Flow through adhesive-filled capillaries is mathematically described by the Washburn equation [86]. The resulting large area of close contact between the adhesive and the internal surface of the substrate plays an important role in developing

368 Progress in Adhesion and Adhesives, Volume 4

good adhesion. This is at times discussed strictly in terms of mechanical interlock, but it also includes any type of surface adhesion through dispersion, polar, hydrogen, or ionic bonds [7, 17]. In contrast, cell wall penetration (infiltration, cell wall modification) involves diffusion of low molecular weight adhesive components into the polymer matrix of wood cell walls. Cell wall penetration generally fortifies, increasing strength and lowering cell wall swelling. Whether a given molecule penetrates the cell wall is controlled by its molecular size, shape, and its solubility parameter, as well as the moisture content and chemistry of the wood. In general, polar molecules and especially hydrogen bond donors are good at entering wood cell walls [115], but molecules above MW 1000 are largely excluded from normal wood [116]. Many people have asked “What is the ideal level of penetration for an adhesive bond?” To answer this, consider that an acceptable bond for a given product is determined by the use application. Not every bond has to provide maximal strength, water resistance, loading or moisture cycling, etc. In addition, almost every bonding operation is unique: different species, wood surface properties, adhesives, and bonding processes. Therefore, the question of ideal penetration is only answerable for a particular process and product. Industry experts have confirmed that the amount of flow and infiltration at a particular mill under a particular set of conditions and wood supply do correlate to bond performance, but a general answer to the question does not exist. 8.2.3.1 Void Penetration (Bulk Flow) An early discussion of wood bond strength relied on void filling to explain bond strength [19]. However, this strictly mechanical interlock concept was deemed insufficient because there can also be specific adhesion through chemical interactions between the adhesive and the wood surface [20]. Some have indicated that for a strong bond the adhesive must penetrate deep enough into the wood substrate to reinforce weakened cells and to obtain a large contact surface [77]. An emphasis on penetration has continued because of the recognition that mechanical interlock helps to distribute stress, and bond performance is presumed to be significantly influenced by the degree of adhesive penetration [110]. Improvements have been made in the microscopy techniques and data analyses used to measure the degree of void penetration. However, despite the ease of conducting these tests, there are still many questions around the issue of how much penetration is needed, especially when filling lumens far from the surface. Wood void penetration is the first step in the formation of the interphase, a threedimensional zone containing both wood and adhesive on both sides of the glueline. The void filling is clearly visualized in Figure 6a, showing an epoxy adhesive filling the wood voids both in the cross-sectional and transverse sections [110]. This reference is a good review of the techniques to analyze void filling, with detailed discussion. While crosssectional microscopy has been useful over the years to examine void penetration, it provides only a two-dimensional view that leaves questions on how the adhesive gets to lumens far from the glueline. Images b and c in Figure 6 demonstrate how tomography can help visualize the three-dimensional flow of adhesive into wood structures.

Understanding Wood Bonds–Going Beyond What Meets the Eye 369

(a)

(b)

(c)

Figure 8.6 (a) Assembled epi-fluorescence images of a 2-part epoxy bondline in red oak stained with 0.5% Safranin O. Transverse view of horizontal bondline on bottom, flatsawn view in upper left, and radial cut on right side. Bright areas are resin [110]. (b) Tomographic image of bromo-PF bondline in loblolly pine. (c) Same as (b) but wood removed to show void penetration of adhesive [117].

When inspecting cross sections, it is common to observe penetration into a lumen three or more cells deep from the surface where there is no adhesive in the first or second cell. The logical explanation was that deep penetration was due to capillary action of the adhesive up a lumen of a cell cut above or below the plane of the particular cross section. While this explanation was well accepted, there was no way to adequately address the issue with the existing technology. X-ray computed tomography, however, now allows visualization of wood, adhesive, and voids in 3-D using labeled adhesives [118–124]. With this tool it is now possible to investigate the 3-D wood anatomy, bondline, and adhesive distribution on the micrometer scale and to address speculation on adhesive-wood void interactions. The adhesives were shown to flow up open lumens that deviate from the wood surface providing a deeper than expected penetration. Adhesive has also been observed to flow through pits into longitudinal tracheids away from the bondline [123, 125]. Adequate penetration of the adhesive into the wood surface needs to occur before adhesive curing and solidification to provide a sufficiently large bonding interface. Low bond strengths can result from underpenetration, where the adhesive is not able to move into the wood enough to create a large bonding interface within the wood interphase, or does not extend beyond a weak boundary layer. Weak bonds can also result from so-called starved joints, caused by overpenetration, when an insufficient amount of adhesive remains in the glueline that bridges the wood surfaces [5, 126]. Often the adhesive is applied to only one surface, and thus, transfer to the adjoining piece of wood is a critical process. In open assembly time, the adhesive only penetrates into the applied surface and the solvent evaporates, while closed assembly time allows transfer and penetration into both substrates [59]. Because the adhesive represents a significant cost to the manufacturer, there is a strong incentive to learn how to apply just enough adhesive, and no extra, in every situation. Adhesive penetrates relatively easily into fiber cells or vessel elements which are physically ruptured, such as during the veneer manufacturing process. SEM images show that adhesives flow predominantly through cut tracheids and rays, simple pits are a minor obstacle, and bordered pits (under these conditions) block the flow of resin from one cell to an adjacent cell [127]. Within a wood species, the penetration can differ greatly between

370 Progress in Adhesion and Adhesives, Volume 4

earlywood and latewood and be influenced by the presence of ray cells, grain deviations, and many other factors. Thus, as illustrated in Figure 6, void penetration is not a uniform process, but varies along the bondline. 8.2.3.2 Cell Wall Penetration (Infiltration) Although adhesive penetration into the wood voids has long been shown using microscopy of cross sections, penetration into the cell wall (infiltration) was often not assessed in part because it was technically more difficult. Over the years, however, a wide variety of methods capable of demonstrating cell wall penetration have been developed. The migration of phenol-formaldehyde (PF) resins into cell walls has been shown using fluorescence microscopy [128], autoradiography [129], transmission electron microscopy (TEM) [130], scanning electron microscopy with energy dispersive X-ray spectroscopy (EDXS) [131], dynamic mechanical analysis (DMA) [132], and anti-shrink efficiency [133]. For polymeric methylene diphenyl diisocyanate (pMDI), the presence of adhesive in cell walls has been more complicated in that it has been shown to occur by nuclear magnetic resonance (NMR), DMA, atomic force microscopy (AFM), and nanoindentation [134, 135], while other studies using X-ray adsorption spectroscopy, DMA, and solid state NMR spectroscopy found none [136, 137]. These and other techniques such as UV microscopy [138] and nanoindentation [139] have been used to show the presence of urea-formaldehyde (UF) and melamine-formaldehyde (MF). UF resin penetration in cell walls in particleboards has been visualized with EDX [140], confocal microscopy [141], and others. We refer the reader to Table 2 in section 3.4 for a more extensive list of references and techniques. The fact that all the adhesives mentioned in the prior paragraph are in-situ polymerized and none are prepolymerized supports the concept of an important functional difference between the two adhesive groups in the way they interact with the wood [142]. Prepolymerized adhesives are highly limited by the size of the adhesive molecules as only low molecular weight (MW) portions of an adhesive can infiltrate the cell wall. In the past, determining the size of molecules that will infiltrate was based on the cell wall penetration of poly(ethylene glycol) or dextrans [116, 143]. Today we are able to evaluate cell wall micro- and nano-domains with microscopy, spectroscopy and/or mechanical tests using adhesives as penetrants. While it is clear that small molecules should infiltrate cell walls better, it is valuable to quantify the relationships between penetrant size, penetration ability, and influence on cell wall properties. Interpretation of dynamic mechanical data pointed out that while a very low MW PF caused stiffening of the cell wall by enhanced intermolecular coupling near the main glass transition of wood lignin, a high MW PF resin did not [132]. Qin et al. noted that the first cell row adjacent to the glueline was more reinforced by UF penetration than cells further from the glueline [144]. Often, images of cell wall penetration show diminished cell wall penetration with distance from the glueline, which is likely caused by the depletion of components able to swell cell walls as the adhesive flows away from the glueline. Examination of brominated PF bondlines with nanoindentation and X-ray fluorescence microscopy (XFM) (Figure 7) showed that a low MW bromo-PF was more

Understanding Wood Bonds–Going Beyond What Meets the Eye 371

0.2 0 0.2

q longitudinal (Å-1)

Unmodified

(c)

0.2

0

0.2

q tangential (Å-1)

(a)

300 40 5 0.7 0.1

2 μm

0.2

3000

0

(b)

Intensity 4000

2000 1000

0.2

Br signal

q longitudinal (Å-1)

d-PF

(d)

500

0.2

0

0.2

q tangential (Å-1)

Figure 8.7 (a) Br signal in X-ray fluorescence of bromophenol-formaldehyde adhesive. The glueline is near the bottom of figure. (b) AFM image after nanoindentation of cell walls with resin content calculated from a [145]. (c) No neutron scattering diffraction pattern is seen from unmodified wood after soaking in 35% H2O /65% D2O solution, but (d) shows diffraction pattern indicating ~4 nm spacing of deuterated PF domains between cellulose “crystals” in the microfibril [147, 149], color images online.

effective at maintaining cell wall stiffness under high humidity conditions than an equal quantity of higher MW adhesive [145]. This discovery was only possible through the ability to quantitatively probe essentially the same approximately one-square-micrometer sized area of an individual cell wall with both chemical and mechanical analyses [145, 146]. An additional technique developed to look at the interaction of adhesive components within the cellulose microfibril involved small angle neutron scattering (SANS) [147, 148]. This work showed that deuterated PF entered the cellulose microfibrils and swelled the spaces between cellulose elementary fibrils [117, 147, 149]. Some adhesives have intrinsic differences which differentiate them from wood, and so the amount of cell wall penetration can be directly measured. An example is the distinct UV absorption spectra of wood and UF resin. Quantitative UV absorbance of thin sections in a microscope allowed MUF (melamine-urea-formaldehyde) and MF to be conclusively identified inside the cell wall, supporting the idea that with these resins, cell wall penetration could create an interpenetrating network of adhesive inside the cell wall covalently bonded to the adhesive in the lumen. The small difference in PF and wood UV absorbance has also allowed some to use the same method to conclude that low molecular weight PF had penetrated the cell walls [151]. Using scanning thermal microscopy (SThM), a PRF was shown to infiltrate while a PUR adhesive did not [152]. Electron energy loss spectroscopy (EELS) has been applied to the detection of partly methylated hydroxymethyl melamine into wood

372 Progress in Adhesion and Adhesives, Volume 4

cell walls [153]. After covalently binding a fluorescent dye to the adhesive, confocal laser scanning microscopy was used to determine the amount of UF resin penetration into fibers used for the production of medium density fiberboard (MDF) [141, 154–156]. These examples show that many analytical tools can be used with certain adhesives to ascertain cell wall infiltration, sometimes quantitatively. The presence of covalent bonds between adhesive components and cell wall polymers has often been theorized but could not be clearly proven until Yelle and coworkers [157, 158] used solution-state two-dimensional NMR. They showed that covalent bonds were present between PF and wood polymers [157] but not between pMDI and wood under typical bonding conditions [158]. After determining that adhesive components have entered the cell wall, the next issue is to determine how they influence the cell wall properties. There is considerable literature in the area of wood modification on the effect of different levels of infiltration by various chemicals on the changes in wood properties. This has traditionally been done by measuring bulk properties, such as the anti-shrink efficiency and biological decay resistance [159–161]. Nanoindentation, first applied to wood in 1997 [162], allows measurement of mechanical properties of individual cell walls and middle lamella corners. Cell wall hardness, measured by nanoindentation, has been extensively used to observe the effect of cell wall penetration [150, 162, 163]. Gindl et al. coupled nanoindentation with UV absorbance spectroscopy to investigate PRF (phenol-resorcinol-formaldehyde) and PUR (polyurethane) bondlines [164, 165]. Significant amounts of PRF resin infiltrated into the cell wall, whereas no PUR could be detected. In both adhesive assemblies examined, cell walls at the immediate surface were damaged during machine planing as shown by a significantly reduced hardness and indentation modulus. The infiltration of the PRF adhesive into cell walls clearly increased the hardness of PRF-impregnated cell walls, but did not significantly change the elastic indentation modulus of sound cells, probably because elastic moduli of sound dry wood cells and PRF are similar. The standard analysis of nanoindentation data assumes that the substrate being indented is a uniform half-sphere. Because this assumption is never true for wood bondlines, the values obtained using this method contain systematic errors and are only relative numbers. Therefore, methods that allow consistent, absolute measures of hardness and modulus even for a cell wall adjacent to an empty lumen were developed for working with wood bondlines [166, 167]. Methods for obtaining elastic and plastic moduli (broadband nanoindentation spectroscopy) were also developed and used to investigate pMDI-infiltrated cell walls [168, 169]. This series of papers have described nanoindentation on unembedded wood samples obtained from wood bonding, and obtained absolute hardness and modulus instead of just relative values of adhesives penetrating the cell wall. In addition, they have shown pMDI adhesive changed the elastic modulus and hardness of cell walls, and swelled the S2 cell layer, clearly visible in atomic force microprobe (AFM) images [135]. This was interpreted as support for the existence of an interpenetrating network of pMDI and wood polymers originally proposed by Frazier and Ni [137]. Nanoindentation has also been coupled with AFM-IR to study the impact of pMDI penetration [170]. AFM-IR is a promising

Understanding Wood Bonds–Going Beyond What Meets the Eye 373

technique in that it allows IR absorption spectra to be obtained on sub-micrometer sized areas, such as an individual cell wall or compound middle lamella. Further information on the interaction of pMDI with wood and wood cell walls using several other advanced analytical methods can be found elsewhere [171]. The ability of wood cell walls to absorb adhesive components not only involves the reactive components, but can also involve low molecular weight adhesive components or additives. Thus, wood adhesives can contain materials that would normally reduce bond strength, but significant adverse effects are avoided because these materials move from the adhesive into the wood [172]. While this might have no impact on most adhesives [173], this phenomenon seems relevant to protein-based adhesives and also potentially to those adhesives that contain large quantities of unreacted urea [174]. When Browne and Brouse proposed specific adhesion as an alternative to just mechanical interlocking [20], methodology did not exist to test their concept. Over the years, specific adhesion has been defined by dispersion forces, polar bonds, hydrogen bonds, and covalent bonds [17], all of which, except covalent bonds, are likely relevant to most adhesives.

8.2.4

Adhesive Properties that Influence Void and Cell Wall Penetration

Although many aspects of the wood influence its bondability, there are also many properties of the adhesive that play a role in controlling bond strength. Molecular weight/ molar mass distribution and branching pattern, viscosity, solids content, reactivity, pH, solubility parameter, and surface tension of the liquid phase of the adhesive will all influence penetration. In addition, additives are used to alter some of these characteristics. Consequently, the adhesive formulation has a tremendous influence over the penetration behavior of adhesives into wood. General statements comparing the penetration characteristics of various adhesive types must be offered with the knowledge that specific formulations can have significantly different properties. For example, a PF resin for composite panel faces is usually lower in molecular weight and/or lower in reactivity compared to core resin, because panel faces experience significantly higher temperature, for longer, than cores. Adhesive specifications generally involve such factors as viscosity, percent solids, pH, and gel time. Although these factors are important for ensuring consistency in production, they do not provide enough data to understand the type and degree of penetration into the wood. While a suitable viscosity is needed for spreading over the surface and into voids, it is influenced by the solids content, temperature, molecular weight distribution, pH, and the presence of additives or fillers in the adhesive. Optimizing an adhesive requires consideration not only of all the adhesive properties but also the details of its formulation as well as the wood and process being used. It should also be remembered that as soon as the adhesive touches the wood, all these properties change because of void and cell wall penetration, moisture movement, wood buffering capacity, temperature changes, etc. Clearly it is not only the initial viscosity, pH, etc. of the adhesive, but also the changes in these properties over the time as they interact with the wood that actually determine adhesive performance. When measuring rheology of adhesives over time, some researchers have

374 Progress in Adhesion and Adhesives, Volume 4

replaced the standard aluminum plates with wooden ones, so as to better understand the changes over time [175–177]. Factors influencing void penetration were extensively studied in a recent series of papers with different wood species and different orientations using UF adhesive at three different DOCs (degree of condensation), which resulted in different viscosities for the three resins [70, 178–181]. In their work, low DOC (lower viscosity due to lower MW, higher polarity due to higher number of remaining methylol groups) adhesive always flowed more deeply (Figure 8). They also showed the expected relationships of higher pressure creating deeper void penetration, cellular collapse at the bondline with higher pressures, and deeper penetration in the radial than the tangential direction, because of the availability of rays to carry adhesive deeply into the wood [182]. It should be noted that wetting properties probably also changed with viscosity [85]. Qin et al. [144] selected emulsion polymer isocyanate (EPI) and UF adhesives as typical systems to investigate the microstructure of wood-adhesive interphases by fluorescence microscopy and confocal laser scanning microscopy (CLSM). Further, a quantitative micromechanical analysis of the interphases was conducted using nanoindentation. The results showed that the UF resin penetrated voids to a greater extent than EPI, and that the average penetration depth for these two adhesive systems was higher in the case of earlywood, as is typical. CLSM allowed visualization of the resin distribution with contrasting colors, showing that EPI did not infiltrate the cell wall, whereas UF resin did. The micromechanical properties of the cell walls were almost unaffected by EPI, but were significantly affected by UF infiltration, especially in the first cell wall from the bondline. This further confirmed that resin infiltration can improve the mechanical properties of cells in the interphase regions.

(a)

(b)

(c)

(d)

(e)

(f)

Figure 8.8 Fluorescence micrographs for three UF resins (mixed with Safranin O as dye) on poplar, using 25 μm thick sections. Left to Right, low, medium and high degree of condensation (viscosity). Upper (a,b,c): bonding of radial faces; Lower (d,e,f): bonding of tangential faces [178].

Understanding Wood Bonds–Going Beyond What Meets the Eye 375

We present only a few of the studies on the effect of adhesive composition on penetration as examples of using multiple newer analytical methods to better understand the phenomenon. Many more studies exist in the literature that can provide insight into the formation of strong bonds. A few summaries and textbooks are provided here [23, 27, 67], and many primary references are provided in section 3.4. Of course, there are many more analytical methods available that are primarily used to study adhesives on their own, which are occasionally applied to bondlines. Nuclear magnetic resonance (NMR) spectroscopy has been highly useful in determining how formaldehyde reacts with urea, melamine, phenol, and resorcinol to form linear and branched structures depending on reaction conditions with solution state being used for the monomers and oligomers, while solid state can be used after cure. Infrared (IR or Fourier transform FTIR) is useful for functional group determination at all stages of reaction. Thermal properties have been determined by differential scanning calorimetry (DSC) and dynamic mechanical analysis (DMA) for the polymeric products as well as for following curing reactions. Molecular weight distribution is measured using size exclusion chromatography. Curing rate can be determined by gel time, dynamic mechanical analysis [183], and small scale mechanical analysis of bonded specimens via the so-called Automated Bonding Evaluation System method [184, 185].

8.3

Properties of Adhesive-Wood Assemblies

Adhesives are used to hold wood pieces together, whether it is in the structurally strong laminated veneer lumber or the relatively weak particleboard. Without the new methods that let us examine the adhesive-wood interaction zone and the adhesive region in great detail as separate entities, understanding bond performance was mainly limited to speculation. In order to understand and improve adhesive performance, we need tools that let us dissect the bonded assembly to understand what is leading to bond failure. In this section we will discuss models of wood bonds, followed by discussions of methods for analyzing penetration and bond mechanical properties.

8.3.1

Zones in a Wood Bond

Given that an adhesive needs to hold two substrates together under a variety of conditions, most information about adhesive performance comes from analyzing bond failure. As stated by Marra, a bond fails at its weakest link [5], with the zones (links) illustrated in Figure 9. However, rather than just thinking about a bond with equal stress in all the zones as in the Marra model, it is useful to keep in mind that the failure often occurs where the stress is concentrated the most [186, 187], and that forces are not equally distributed among the different zones. For example, internal strain usually occurs in the adhesive-wood interphase when the adhesive shrinks due to curing reactions and solvent loss, while the substrate dimensions remain unchanged or swell from the absorbed water [22, 187]. Although this simple link system is easy to visualize, a real bond has more of a continuum of different states or overlapping zones with dissimilar forces, in both magnitude and direction, in different zones.

376 Progress in Adhesion and Adhesives, Volume 4

Figure 8.9 Marra’s description of bondline failure zones (links) using the terminology described in this review rather than Marra’s original terms [5]. In this review links 1–7 are referred to as the bondline, while links 1–3 are referred to as the glueline.

8.3.2

How Adhesives Accommodate Wood Swelling

Besides the requirement that the dry strength of the adhesive bond be greater than the wood strength, another key requirement of the bond is to resist fracture or delamination when exposed to wet conditions. Surprisingly, even though wood loses strength due to plasticization when wet, this exposure often leads to failure in, or close to, the glueline. Therefore, it is often wet testing that differentiates acceptable from unacceptable wood adhesives. Since the failure can even occur in some tests with no applied external load on the bondline [79, 188–191], the swelling and shrinking of the wood is a main source of stress that breaks the adhesive bond [22]. Additional tests involve a load applied to the wet sample [192–194]. Because wood typically swells much more than the adhesive, the constraint imposed by the adhesive leads to complex and sometimes very large stresses on the bondline [22]. The response of a bonded assembly to water exposure is illustrated in Figure 10. Figure 10a shows the dry assembly as formed. Figure 10b illustrates a possible swelling and cupping of the unbonded wood when wetted. For the bonded case, the adhesive resists the dimensional change in wood, resulting in large normal stresses at the interface both perpendicular (Figure 10c) and parallel (Figure 10d) to the bondline. If these stresses are not well distributed, bond rupture can result, even without any external load. Thus, adhesive interactions with the wood that minimize stress concentrations in the bondline are important components of bond durability [21, 142]. Recognizing the value of avoiding stress concentrations, many suppliers offer “flexibility modified” condensation resins where the glueline strength, brittleness, toughness and flexibility have been modified, usually using proprietary techniques.

Understanding Wood Bonds–Going Beyond What Meets the Eye 377 Initial, Dry

Wet wood warping + swelling Un-bonded

Bonded

Stress perpendicular to bondline: Tension compression

Adhesive

Adhesive

(c) (a)

Stress parallel to bondline:

(b)

Compression tension compression (d)

Figure 8.10 Illustrative swelling stress in two dimensions. (a) dry wood, bonded; (b) wet unbonded, with adhesive in place illustrating swelling and cupping of the wood substrates; (c) image of specimen overlaid with plot of stress perpendicular to bondline (green), expected along the dashed line. In this specimen, cupping pulls edges away from bondline; (d) image of specimen overlaid with plot of stress parallel to bondline (red), expected along the dashed line. When wet wood attempts to expand sideways, adhesive is under tension as it restrains the wood while wood adjacent the bondline is under compression, color images online.

Structural wood-adhesive bonds have to withstand considerable stresses caused by mechanical loads and moisture content changes. Moisture-related durability of such bonds depends on the ability to withstand stresses generated by moisture–induced dimensional changes in the wood.

8.3.3

Two Classes of Adhesives

While adhesives have been classified using a wide variety of criteria, distinction between insitu polymerized and prepolymerized adhesives seems to be especially useful for wood adhesives. These two adhesive classes generally differ in their chemical curing, morphological, physical, and mechanical properties [16, 142, 195]. The largest group by far in volume sold is the in-situ polymerized group, including the aminoplastic and phenolic adhesives that use formaldehyde as a co-monomer, epoxies, and pMDI, see Figure 11. These adhesives typically produce rigid polymers on curing [196] and often contain reactive components that are low enough in MW that they can penetrate cell walls. These polymers are mainly thermoset. The second group of adhesives involves pre-formed polymers that are too large to infiltrate cell walls, and develop adhesive strength by losing water and/or by cross-linking the flexible polymers. These adhesives include PVAc, PUR, and proteins. After physical solidification (coalescence), these are generally more flexible than the in-situ polymerized adhesives.

378 Progress in Adhesion and Adhesives, Volume 4 In-situ polymerized—phenolic, aminoplastic, epoxy, pMDI

Prepolymerized—polyurethane, poly(vinyl acetate), EPI, protein

Figure 8.11 Schematics for the in-situ polymerized and prepolymerized classes of adhesives.

The in-situ/prepolymerized categorization is useful for wood adhesives because these two groups appear to have different mechanisms for minimizing water-induced stress [142]. An abrupt change in mechanical properties within a material can result in significant stress concentrations and failure. In-situ polymerized adhesives that infiltrate cell walls have been shown repeatedly to modify cell wall properties in the wood interphase region [145, 163, 170, 197], which in addition to void penetration generate a gradual transition of mechanical properties from the stiff glueline to the bulk wood. Infiltrating in-situ polymerized adhesives have also been shown to reinforce weak, mechanically damaged cells near the glueline [163]. In addition, cell wall penetration provides the opportunity to have cured adhesive molecules that both interpenetrate cell wall polymers and extend into the bulk adhesive, strengthening the wood-adhesive interphase [21]. Prepolymerized adhesives, on the other hand, are typically not as rigid and so can stretch to accommodate wood swelling, thereby avoiding stress concentrations. In-situ polymerized adhesives that are typically durable on nonswelling substrates have failed water soak tests on wood bonds when the adhesive does not have a component that can penetrate cell walls. Examples include PF lacking a low molecular weight component, and epoxy, where bisphenol A is not expected to penetrate cell walls based on its low polarity [21, 142, 198]. Although the two adhesives classes were hypothesized based on many prior studies in the literature, new techniques allow measuring physical and mechanical properties at the cell wall level and smaller. Instead of relying on a single strain measurement from a traditional strain gage, digital image correlation to measure full-field strains within and near the bondline is now available. Analysis of MUF, PRF, PUR, and EPI adhesives supported the proposal that in-situ polymerized adhesives stabilize the wood and the prepolymerized adhesives distribute strain in the adhesives [199]. Minimal creep of the bond is important for structural wood products as evidenced by a wide use of creep tests [200, 201] and minimum performance criteria [79, 202, 203]. The in-situ polymerized adhesives typically have very good creep performance because they are

Understanding Wood Bonds–Going Beyond What Meets the Eye 379

thermoset polymers, often with branching and nearly complete polymerization, forming a rigid three-dimensional network. Imparting creep, moisture, and heat resistance to prepolymerized adhesives has been more challenging but the more durable ones are generally lightly crosslinked to maintain some elastomeric properties.

8.3.4

Methods for Determining Void and Cell Wall Penetration

The essential challenge in measuring penetration is to differentiate between wood and adhesive, on a size scale relevant to the question being addressed. When measuring void penetration only, resolution of up to a few micrometers is often sufficient, while sub-micrometer resolution is needed to quantify cell wall penetration with confidence. Ideally, a natural feature of the adhesive such as color or natural fluorescence allows it to be unambiguously identified. Often the darkness of a phenolic adhesive stands out against the light color or autofluorescence of wood, while isocyanates, proteins, and sometimes other adhesives will naturally fluoresce differently from the wood. Enhancing contrast between the adhesive and wood is often necessary, however. In many cases, dyes are applied to the sample just before visualization, such as Lugol’s iodine (not Povidone iodine) to color PVAc and starch [204]; brilliant sulphaflavine to make amine bearing adhesive fluorescent [205, 206], toluidine blue to suppress the autofluorescence of wood [156], or osmium tetroxide and uranyl acetate to darken oxidizable structures in TEM. Another approach is to add the contrast agent to the adhesive before applying it to the wood. Adding small molecular tracers is attractive because they spread evenly through the adhesive, making every part of it visible. The danger with molecular tracers is that they sometimes migrate independently of the actual adhesive. An example of this was the use of RbOH as the alkali to make a PF: while the Rb did successfully enhance X-ray absorption, it did not stay with the PF [207]. Insoluble tracers like pigment particles are attractive because they do not leave the adhesive, and show up as unmistakable bright specks not only in fluorescence but in element-sensitive mapping such as backscatter SEM, EDX, EELS, TEM, etc. While the concentrated signal from pigment particles are typically more obvious than from a dye, not all the adhesive is visible, just the pigment particles. Because of their size, pigment particles can be retained by anatomical features like pits or sieve plates when the adhesive passes, defeating their purpose [208]. Also, pigments are too large to penetrate cell walls. Another approach is to chemically modify the adhesive backbone to integrate the contrast enhancement. Because of the chemical similarity of PF and lignin, it is often difficult to use chemical techniques to distinguish PF and cell walls. To overcome this, Kamke and coworkers made their adhesive from iodophenol and formaldehyde, ensuring one iodine atom stayed attached to each phenolic unit, making the adhesive highly absorbing for X-ray tomography [118, 119, 123, 209–211]. Others similarly used bromophenol with various analytical techniques including SEM-EDX [129, 131], neutron activation analysis [212], and X-ray fluorescence [145]. Grigsby and coworkers [213–218] have made a series of PF and UF adhesives where fluorescent molecules are incorporated into the polymer backbone, typically at ~20% substitution. Grigsby claimed to have done size exclusion

380 Progress in Adhesion and Adhesives, Volume 4

chromatography to show that the fluorophore was uniformly present in all the different molecular weights of the applied adhesive [219]. While this approach is sound analytically, care in interpreting results is needed because incorporation of different monomer units (acriflavine, bromophenol) in the polymer means that the adhesive has different molecular size, solubility parameter, branching pattern, etc. than the unlabeled adhesive, which can change the penetration and mechanical properties. Prepolymerized adhesives are in many ways the easiest to label because the backbone polymers are typically so large that covalently attaching a few heavy atoms or fluorescent dyes is unlikely to significantly change their properties. Also there are typically many functional groups available for covalent reaction with the tracer. While this is a promising technical route, it has been little exploited to date [220]. Visible light is of course not the only technique that can differentiate wood and adhesive: absorption or interactions with radiation of other wavelengths can be exploited by ultraviolet or X-ray absorbance, Raman, IR, NMR, or X-ray fluorescence. Interactions with electrons provide not only simple observation of topography with SEM but a whole collection of analytical techniques from EDXS to EELS. Nanoindentation allows probing the mechanical properties of the compound middle lamella, S2 cell wall, and (resin filled) lumen independently. Often a more complete understanding is only possible by using a variety of methods to examine different properties of the same specimen. Table 2 is provided as a brief overview of methods for measuring void and cell wall penetration, and their relative merits. Typically, increasing resolution results in a smaller volume or area examined, and this is especially relevant to tomography. It is not unusual to acquire thousands of 2D images for a single 3D reconstruction. Therefore, high intensity radiation (which might cause sample degradation or drying, accompanied by shrinkage), long acquisition times, and small samples are much more the norm for tomography than 2D imaging. Because only small volumes are scanned, the volumes being examined need to be carefully selected to ensure that they are representative of the norm. Traditionally, light microscopy is often used in transmission with thin sections. This provides an excellent view of the wood and bondline, but thin sections can be very time consuming to prepare, and preparation becomes much more difficult as the area becomes larger. Reflection or fluorescence microscopy from the surface of a block of wood can image a large area more quickly, but it has traditionally been challenging to create a surface smooth enough for such images to be useful. Researchers have started polishing large areas with standard sanding equipment and airflow or vacuum to draw debris out of the sanding zone so it does not clog the wood pores [247, 250]. Removal of sanding dust is facilitated by using extremely porous sanding pads rather than traditional sandpaper. Tape can also be used to remove residual dust. This allows much larger areas to be prepared quickly, which can mean multiple samples prepared at once, or much larger areas in a given sample for better statistical reliability. The samples for Figure 1 were prepared by this method. This method is fast and simple enough to be practical for industrial quality control. If near-nanometer smoothness is needed, such as preparation of samples for nanoindentation, an ultramicrotome is typically used [145, 166], but successful sample

The traditional thin section method Visible light transmission or fluorescence emission: fluorescence often has higher contrast Thin sections provide best samples as light reflects off wood below the image plane making features somewhat fuzzy Sample preparation can be time consuming and difficult, and one needs to be careful of smearing while cutting Staining of sections (wood or adhesive) or labelling of adhesive are common Especially useful if there is natural contrast between wood and adhesive Fluorescence is especially useful for light colored adhesives that are typically hard to see, especially when failure occurs in the bondline Fluorescence often uses UV illumination. Wood naturally fluoresces with UV.

Visible light reflection or fluorescence emission from the surface of a block. The size of surface damage on the sample should be smaller than objects of interest. Small surfaces can easily be made with a sharp razor blade or microtome; large surfaces are very difficult to produce with a knife. Sanding can generate large sample areas for observation.

Microscopy of the smooth surface of a thick specimen

Characteristics

Transmission microscopy; fluorescence microscopy (traditional thin sections)

Method

Table 8.2 Methods of determining void or cell wall penetration in wood.

(Continued)

Efficient method for making smooth surfaces on large areas: [247] Observing penetration on blocks: [247–250] Bond and glueline thickness on blocks: [251] Failure analysis on bulk specimens: [22, 204]

Overviews: [110, 221–225] Specific studies: [70, 71, 77, 110, 114, 122, 144, 178–181, 198, 199, 205–207, 226–246]

References

Understanding Wood Bonds–Going Beyond What Meets the Eye 381

[123, 129, 131, 140, 156, 172, 207, 227, 235, 253, 254, 259]

[127, 153]

SEM or TEM with ability to map location and concentration of nearsurface elements based on energy of emitted X-rays

EELS detectors are added on to an SEM, TEM, or ion mill Sensitive to surface elements (including C, O and N) and their bonding states. Nanometer resolution possible

Sensitive to index of refraction, hence whether cell walls are infiltrated

[198, 264] Radioactive atoms (14C, 3H) are placed in the adhesive, and location of emitted particle from atomic decay is recorded. Slowly decaying isotopes take long time to image (sometimes months for a single image)

SEM or TEM with analyzers for emitted X- rays (EDXS, EDS, EDX, EDXA, EDAX, XEDS); WDS

Electron energy loss spectroscopy (EELS)

Interference microscopy

Autoradiography

[263]

[129, 140, 198, 227, 235, 236, 258–262]

Highest resolution (nm scale) Requires extremely thin sections (~100 nm) and staining with electrondense materials. Very small fields of view

Transmission electron microscopy (TEM)

[77, 127–129, 131, 207, 226, 227, 236, 246, 252–257]

References

Images formed from electrons scattered off or ejected from the surface: low contrast between wood and adhesive Typically imaged under vacuum 10 cm field of view and nm resolution are possible Easy to change scan area from cm to μm With proper detectors, it can be sensitive to topography (secondary ion), or changes in density/elemental composition (backscatter mode), and several other modes possible Greater depth of field, resolution, and variable magnification than light microscopy

Characteristics

Scanning electron microscopy (SEM)

Method

Table 8.2 Cont.

382 Progress in Adhesion and Adhesives, Volume 4

[256, 274]

[152, 278]

3D image of sample assembled from a series of x-ray absorption images at different angles Sample size and instrument time limited by desired resolution Requires different absorption between resin and wood, often achieved by covalently coupling a heavy atom into the adhesive Would theoretically work for nitrogen-containing adhesives without additional labelling

X-rays passing through a sample excite fluorescence from the (heavy) atoms within the sample. Resolution from nanometer to millimeter, depending on instrument Synchrotrons commonly used because of their high X-ray intensity

Uses absorption of neutrons by sample to create an image or, by imaging at multiple angles, a tomogram Highly sensitive to hydrogen, therefore avoids need for labeling

Observes rate of heat flow from AFM tip to sample

Observes thermal expansion of ~200 nm thick specimen when IR light is absorbed Traditional IR data with nanometer spatial resolution

X-ray computed (micro)tomography (XCT, micro-CT); synchrotron x-ray (micro)tomography

X-ray fluorescence (XRF) microscopy; XRF-tomography

Neutron imaging, tomography

Scanning thermal microscopy (SThM)

IR-AFM

[170, 279, 280]

[123, 125, 145]

(Continued)

[117–123, 125, 207, 209, 210, 272–277]

[127, 138, 151, 164, 165, 266, 271]

Classical method for wood and lignin studies: measures UV absorbance through thin sections

UV absorbance microscopy

[128, 141, 144, 154–156, 163, 205, 215, 216, 220, 226, 227, 238, 265–270]

Out of focus light is blocked by the optical system, giving much clearer images than standard fluorescence microscopes Possible to “optically section”, or only collect light from the focal plane, not above or below Serial optical sections can be used to create 3D images Multiple fluorescence signals as well as reflected light (for topography information) or transmitted light (if samples are thin) can be simultaneously collected

Confocal microscopy or Confocal laser scanning microscopy (CLSM); Spinning disk confocal microscopy

Techniques not widely used for penetration before 2000

Understanding Wood Bonds–Going Beyond What Meets the Eye 383

[136]

[156, 281, 288–290]

Creep and viscous properties observed with nanoindentation apparatus

Analogous to transmission light microscopy, using x-rays Extremely high spatial resolution possible Sensitive to nitrogen in the adhesive

Highly sensitive to top few nm but large area illuminated so only mm spatial resolution Sensitive to all elements except H and He, indicating both their presence and oxidation states

Micrometer-scale tip rasters over surface to create map of surface. Topography is most common but many different modes available to enhance sensitivity to particular specimen properties

Observes molecular vibrations, difficult to interpret. Has potential for quality control where little variability is expected

Nanoindentation mechanical spectroscopy

X-ray microscopy

X-ray photoelectron spectroscopy (XPS) or (ESCA)

AFM

Near-infrared (NIR)

[206, 243]

[243, 257]

Methods: [168, 169] Application to wood bondlines: [170]

Method for eliminating specimen compliance: [167] Effect of microfibril angle: [282] Bondline studies: [144, 145, 166, 172, 196, 197, 217, 238, 243, 257, 271, 283–287]

Measures the mechanical properties of solids, shows impact of cell wall penetration Sample volumes of less than 1μm3 are possible Typical values obtained are elastic modulus and hardness.

Nanoindentation stiffness, modulus

[243, 257, 266, 281]

References

Sample is imaged by scanning, IR or Raman spectrum acquired at each spot. Spot size limited by frequency of excitation, theoretical limit of ~0.5 to 2 μm

Characteristics

IR/Raman imaging

Method

Table 8.2 Cont.

384 Progress in Adhesion and Adhesives, Volume 4

Chemical bond information at resolution beyond the diffraction limit

Superresolution infrared (IR) or Raman spectroscopy

[134, 137, 237, 298, 301–304]

Observes elastic and plastic moduli, which change when adhesive infiltrates wood Typically used for wood-adhesive interaction studies, but cannot provide spatial information

Observes heat flow into/out of sample with temperature change: sensitive to chemical and physical changes

Change in mean spacing between cellulose nanocrystals is observed when deuterated adhesive or water enters cellulose microfibrils, showing how adhesive impacts swelling inside cell wall nanostructure

Measures mobility of hydrogen atoms on picosecond time scale. Provides information on water and polymer mobility inside the cell wall, as affected by resin penetration

Dynamic mechanical analysis (DMA); Thermal mechanical analysis (TMA)

Differential scanning calorimetry (DSC)

Small angle neutron scattering (SANS)

Quasi-elastic neutron scattering (QENS)

[147]

[117, 147–149]

[305]

[134, 157, 158, 281, 291–300]

Sensitive to the chemical bonds and environment of atomic isotopes with unpaired spins (13C, 1H, 15N, etc.) Sample can be solid or liquid. Liquid gives better resolution.

[170]

[126]

Nuclear magnetic resonance (NMR)

Methods to investigate interaction of wood and adhesive

Quantitative measure of the volume of pores of each diameter Wood samples measured before/after bonding to determine which pores were filled

Mercury intrusion porosimetry (MIP)

Understanding Wood Bonds–Going Beyond What Meets the Eye 385

386 Progress in Adhesion and Adhesives, Volume 4

preparation by polishing with a series of sandpapers and then with a diamond abrasive has also been claimed [306]. It is probably not surprising that the textbooks describing how to cut, mount, and stain thin sections were initially written many decades ago [225, 255, 307]. Because it can be difficult to find resources, we provide some references here. Basic wood anatomy, making wood sections by hand, and using a hand lens is described in English and Spanish in these three book chapters available for free online [308–310]. Hand lens use and making sections by hand are also described by Hoadley [311]. Microscopy, wood specimen preparation, and staining are often described together [224, 312–314]. Collections of wood-related micrographs are also available, and are not only beautiful but also useful for demonstrating how to make extremely informative images [221, 223]. Inspecting bondlines after product failure is discussed, with photos, for an extensive list of potential adhesive problems by Marra [5]. Lukowsky does not address as many adhesive issues but his color photos are easier to interpret and he includes a schematic depicting different types of penetration issues [204], as well as many practical examples of failure analysis of wood products. 8.3.4.1 Quantifying Depth of Void Penetration Any measure of void penetration must consider the strong impact wood structure can have on penetration. Differences of earlywood vs latewood, radial vs tangential face, cells parallel to the cut surface vs sloping grain, lathe checks, damaged surface cells, high or low spots on the wood surface, and the porosity of the wood on the other side of the glueline could all significantly influence penetration depth. Careful attention to these variables, as well as a sufficient number of measurements in appropriate regions of the bondline are necessary to obtain meaningful penetration data. Perhaps the simplest, and an early-used method of quantifying void penetration was to apply a uniform amount of adhesive and measure the thickness of the cured glueline. A thicker glueline means less adhesive has moved into the wood [87, 241]. Reporting the “Mean Penetrated Adhesive”, or % of an area filled with adhesive, at several different distance increments from the bondline, [114] is simple and extremely informative. With automated image analysis, it is possible to make many segments, resulting in informative histograms of the % of each region filled with adhesive as a function of distance from the glueline; given the variability of adhesive penetration it is important to sample many areas to obtain a representative value. Probably the most common approach is to quantify the extent of adhesive penetration at each increment of distance along the length of the bondline. The average of these values is termed average penetration (AP) depth [144, 179, 238], or specific penetration. Effective Penetration (EP) is a commonly used measure of void penetration [119, 122, 144, 205, 238, 244]. Originally proposed by Sernek et al., [205], EP is the sum of area outside the glueline containing adhesive divided by the length of bondline measured. In theory, EP + glueline thickness should be highly correlated to spread rate. One potential uncertainty in this measurement is whether to classify cell walls infiltrated with adhesive as cell wall or

Understanding Wood Bonds–Going Beyond What Meets the Eye 387

adhesive. Paris and Kamke [119] found that two adhesives of very different penetration patterns had indistinguishable EP’s, leading them to recommend “Weighted Penetration”, or EP weighted by the distance the adhesive is from the glueline, which did distinguish between their adhesives. Maximum Penetration (MP), also proposed by Sernek et al. [205], needs modification, or at least specific clarification, to be comparable between laboratories and species, and even with that clarification, it has limited utility. MP is the average distance from the glueline to the five most distant resin spots. First, any such measure should be normalized to the length of bondline observed, because as more bondline is measured, more spots will be observed far from the bondline. It has limited utility because it only reflects the most extreme values, which are statistically quite variable. MP has been shown to be highly influenced by a few very efficient flow paths, such as rays or cracks [180]. MP would be valuable, however, in cases where the extreme penetration values are the important information, such as when trying to ensure there is no adhesive bleed-through. Hass and coworkers improved the utiltiy of MP by displaying their tomography results in quartile plots, which are much more informative than single value expressions [272]. At the end of a series of papers on the factors influencing penetration depth, Gavrilovic´Grmuša and coworkers settled on six different measurements: AP, MP, MP/AP, total interphase region, I, which is the width of the interphase multiplied by the length of bondline inspected; A or area of filled lumens and rays in the region I; and A/I, the percentage of the interphase filled with adhesive [70]. The percentage of interphase filled with adhesive is essentially the same as Johnson and Kamke’s measurement [114], simplified to a single average value. As an improvement on EP, Hass et al. were able to show a good correlation between void penetration and bond strength with PUR, PVAc, and UF resins in beech when penetration was quantified as the degree of saturation of the pore space, specifically excluding cell wall material from the calculation [272]. This is a promising development, and identifying three phases (lumen, wood, and adhesive) instead of just two should be relatively straightforward with data from many techniques. Edalat and coworkers tallied resin filled areas using the number of cells as their unit rather than micrometers, which has the advantage of capturing the influence of wood anatomy on void penetration [250]. This approach was able to identify different penetration depths for boards made with wood at different moisture contents. If this technique were standardized to the length of bondline inspected (presumably measured in number of cells) there is a chance that numbers could be compared among laboratories. However, it seems more cumbersome than summing the areas using automated analysis methods.

8.3.5

Shortcomings of Standardized Tests

Standardized tests for wood bonding were developed to allow comparisons of adhesives from different manufacturers, using exactly the same methodology, often with a requirement for testing after exposure to heat or moisture, or even boiling water and/or climatic cycling. However, standard methods seldom cover how the bonded sample is made.

388 Progress in Adhesion and Adhesives, Volume 4

Exceptions are ASTM D 2559, which is used for qualifying adhesives for structural products sold in the United States, and is fairly specific on the wood and the bonding conditions used, as is ASTM D7998 [185]. Often it is hard to discern why the test method uses the specified conditions, although the commentaries in versions of D2559 after 2011 are an exception. The problem for industry and academics is that the results from one national standard cannot necessarily be compared to another national standard. A prime example is the many different test methods used by regulators around the world to monitor formaldehyde emissions from wood panels where not only the procedures but also the strategy and aim of the tests vary between jurisdictions [67, 315]. How does one compare the results learned in Europe using an ISO or an EN standard with European wood species to the results from an ASTM test using a North American wood species? An additional complication is that there are no standard adhesive or standard wood specimen, such as the standard epoxy, 316 stainless steel, and pressure-sensitive tape for peel or shear tests. With wood, it is difficult even to obtain consistent samples of a given wood species. In some cases, standards that appear very similar have subtle differences that might have significant impact, such as the requirement in ASTM D906 that half of plywood bonds are pulled open and half pulled closed, while the parallel standard EN314 does not specify how the bonds are pulled [194, 316]. Although the chances of having harmonized international standards are not very likely, the more we understand about wood bond performance beyond the readily visible failure load and percent wood failure, the better we can start to compare studies using different adhesives, wood species and bonding conditions. The increasing level of sophistication of analysis of intact and failed bond samples allows us to better understand bonds from a fundamental perspective.

8.4

A More Detailed Approach than Standard Wood Failure Analysis

It can be very hard to determine what properties of an adhesive lead to success. The adhesive is considered successful when the bond does not fail and meets other requirements such as cost, fitting the production process, meeting customer’s specifications, etc. Improving performance by paying attention to the impact of changes in the wood preparation, adhesive formulation, and bonding conditions has been going on for a long time. Accelerated tests that typically involve changing moisture or temperature, as well as different static loads, or combination of these are helpful and have been used to estimate long term durability [317]. When a product fails in the bondline, however, the researcher can begin to understand what is critically missing from the adhesive’s performance. As Marra indicated, all the links between the two wood pieces need to hold together to prevent bond failure [5]. Solving the weak link problem leads to commercial success, but understanding the weak link allows for more effective problem solving in the long term. Thus, a careful examination of the failure surfaces is useful to understand in what region and why the bond failed. Many traditional and new techniques have been developed to understand the where and why of bond failures. Unfortunately, it is easy to misinterpret the analysis results because wood is such a non-homogeneous material and bonding is

Understanding Wood Bonds–Going Beyond What Meets the Eye 389

complex. At a minimum, the researcher needs to be sure that the area examined is representative of the main failure zone. Below we will look at two examples of studies that used failure analysis to improve understanding of why bonds failed, and then briefly discuss how nonstandard mechanical tests can provide insight.

8.4.1

Going Beyond What Meets the Eye to Understand Epoxy Failure

A study of epoxy failure is a good example of how looking deeper into failure modes using a variety of methods has greatly improved the understanding of a common problem, i.e. water durability of epoxy bonds to wood. Epoxy bonds often fail water soak tests when bonding wood, but provide strong, durable adhesion to many plastics, metals and concrete [318]. The failure of wood-epoxy bonds has been known for a long time and significant research has been done to find formulations that work well [318–320]. A study was done to explore why water soaking induced failure of epoxy bonds using a commercial epoxy resin [318]. In a typical fashion the bonded products exhibited poor strength during wet testing. Normal naked eye and microscope images showed what looked like a “wood” side and a shiny adhesive side, with a wood-like texture was evident on both sides (Figure 12). This apparent adhesion failure was surprising since cross-sectional analysis of the bonded specimen showed good void penetration of epoxy on both sides of the bond. This reinforces the idea that wood penetration may be a necessary, but it is not a sufficient condition for durable bonds, especially when exposed to water. Normally at this stage, researchers go to the SEM to obtain more detailed images. The SEM images in some spots appear to show adhesion failure, as illustrated in Figure 13. However, a closer examination showed that most of the lumens on the “wood” side of the bond appeared to be filled with epoxy. Because this analysis can be very subjective, the SEM analysis did not offer any definitive conclusion on failure location. Attempts to stain the adhesive and the wood showed that there seemed to be an epoxy coating on the wood side, but the results were inconclusive. Fortunately, one of the epoxy

(a)

(b)

Figure 8.12 Optical microscopy images of the failure surfaces from ASTM D 905 sample of a commercial epoxy on wood to illustrate the orientation or surface features parallel to the wood longitudinal direction on (a) the side that appears to be epoxy and (b) the side that appears to be bare wood. Approximate image width; 1.2mm.

390 Progress in Adhesion and Adhesives, Volume 4

(a)

(b)

(c)

Figure 8.13 SEM images of the failure surfaces from ASTM D 905 testing of a commercial epoxy on wood to illustrate the difficulty of distinguishing failure (a) unbonded wood, (b) the “wood” side of the failed bond that contrasts with (c) the “epoxy” side of the bond. Approximate image width; a, b 1.2mm; c, 280μm.

Epoxy film

Epoxy residue

Figure 8.14 Fluorescent image of the wood side failure surface of epoxy bonded wood. Fractures primarily oriented along the wood grain direction. Scale bar: 100μm.

adhesives tested happened to provide good fluorescence on curing. The image in Figure 14 shows that the adhesive covers the wood surface even on the wood side, suggesting a combination of failure modes: limited adhesion failure and mainly cohesive failure of the epoxy film very close to the wood surface. Another telling feature was that the fracture planes were oriented parallel to the wood grain and applied load. Fracture planes typically form perpendicular to the main loading direction. This suggests that that the primary load was perpendicular to the applied force, as would be expected if failure were induced by wood swelling. If the “wood” side was generated due to adhesion failure, x-ray photoelectron spectroscopy (XPS, also called ESCA) should show a surface that looks wood-like, not epoxy-like. The XPS data, however, showed epoxy on both failure surfaces [290] as did the FTIR. These different analytical methods showed that what looked like an adhesion issue was really a cohesive failure very close to the wood surface, i.e., in the adhesive interphase. This result

Understanding Wood Bonds–Going Beyond What Meets the Eye 391

led to the idea that internal strain from the swelling of the wood limited the amount of exterior load the bond could withstand. Supporting this result was the observation that the use of certain primers improved the durability of epoxy bonds to wood [321]. The expected inability of bisphenol A in epoxy to penetrate cell walls provided a counter example to most in-situ polymerized adhesives, and was critical to Frihart’s development of the idea that stiff, in-situ polymerized adhesives must penetrate and modify cell walls in the bondline to be durable [142]. Primers such as hydroxymethylated resorcinol (HMR) might help epoxy durability by penetrating and modifying the wood cell walls. For more discussion of primers see section 5.

8.4.2

Using SEM to Detect Brittle Failure in UF

Stress concentration points can initiate failure, which can lead to global failure during mechanical testing, as discussed at length by River [62, 187]. Therefore, the adhesive ductility, or ability to avoid stress concentration in the bondline, can be an important adhesive property [322–325]. UF is a brittle adhesive and fractures on curing due to volume shrinkage from water loss and condensation reactions [326]. Replacing ammonium chloride used for curing the UF with organic ammonium chlorides which are less volatile resulted in higher strength values. Inspecting the failure surfaces with SEM revealed that along with the increased strength came increased signs of ductility such as necking of the adhesive rather than sharp edges [324–326]. The increased ductility appears to explain the improvements in bond strength. An SEM is commonly used to examine fracture surfaces to identify the difference between adhesion and cohesive failures, and the sharpness of fracture planes on brittle failure surfaces compared to the more distorted surfaces in plastic failure. Because these observations are typically highly magnified, and therefore cover only a small area, it is important that the area selected is typical of the bonded surface area. In addition, the conductive coating traditionally applied has obscured observations. Recent improvements in SEM technology providing higher resolution, better sensitivity, and lower specimen damage are likely to yield better images of fracture surfaces in the future. In addition, advances in techniques that can be added to an SEM, such as EDXS or WDS, allow researchers to simultaneously map the physical structure and the chemical composition of the surface.

8.4.3

Alternative Mechanical Methods of Testing for More Information

The key requirement of a wood adhesive is that it holds the wood together under normal use conditions. Standard test methods for wood bond strength are numerous depending on the application. In the US, the ASTM methods D905, D906, D1037, D2559, D7247, and D7519 are most commonly used [78, 79, 191, 194, 327, 328]. Europe typically uses the CEN methods EN 205, EN 302, EN 314, EN 391, EN 12765 (formerly EN 204), EN 14080, and EN 15425 [190, 192, 202, 203, 316, 329–331]. These are just the most commonly used standards for bonded wood products. They do not begin to cover specialty applications or other countries.

392 Progress in Adhesion and Adhesives, Volume 4

While standard visual methods are very useful commercially, they rarely provide clear guidance as to why the bond failed. If crack propagation or elongation properties of the cured adhesive between wood substrates are needed, the dual cantilever beam (DCB) method is often used [332, 333]. DCB has been useful in developing understanding of wood bonding relationships such as the influence of adhesive thickness and plasticity on bond strength [62, 334–339]. In this test, two pieces of wood are glued together, a notch is introduced in the bondline at one end, and the pieces of wood are pulled apart perpendicular to the bondline at the notched end. This procedure enables testing the cohesive strength of an adhesive in a joint. This value is typically used to calculate energy release rate, or energy required for a crack to grow. It has been shown that the energy required to grow a crack in PUR is typically much larger than in PRF or other in-situ polymerized wood adhesives [187]. Recently, efforts have been made to use DCB to understand the performance of composite panels, with some success in differentiating adhesives better than the standard internal bond tests [340–342]. Stress concentrations can initiate failure, which can lead to global failure during mechanical testing [187]. Prepolymerized adhesives appear to rely on adhesive ductility to avoid stress concentrations in the bondline, while in-situ polymerized adhesives appear to avoid stress concentrations by creating a gradient of properties through the interphase. Digital Image Correlation (DIC) and Electronic Speckle Pattern Interferometry (ESPI) are useful methods for measuring the displacement and strain fields of a bondline during loading. ESPI uses interference between a reference beam and the light reflected off a sample [343, 344]. DIC tracks the motion of high contrast objects, such as specks of spray paint, on the sample surface in a series of images taken as the sample deforms. The work of Kläusler et al. [345] shows the striking difference in ductility between two structural adhesives, a PRF and a PUR. PRF strain-at-break ranged from 1% dry to 2.5% wet, while PUR strain-at-break was consistently 25–30%. Figure 15 shows how this difference

(a)

(b)

Figure 8.15 Strain distribution at nominal 10MPa shear stress. Red = high strain. (a) Strain is concentrated at the ends of the joint with brittle PRF adhesive, while (b) ductile PUR distributes strain (and therefore stress or load) more uniformly [346], color images online.

Understanding Wood Bonds–Going Beyond What Meets the Eye 393

in ductility manifests in bondline deformation, as measured with DIC (red = high strain) [346]. The movement and load in brittle PRF (A) are concentrated at the ends of the joint, while the ductile PUR (B) allows movement across the entire joint, which results in more uniform load distribution. A complete discussion of the details of DIC methodology can be found elsewhere [347]. Tracking the displacements of individual elements of a bondline is also possible using tomography, with the advantage of observing all three dimensions of the specimen. Tomography data have led to a micromechanics model of equivalent strain and stress of each volume element of the adhesive bond under load [211]. The experimental lap-shear test results from the same specimens validated the model. In most tests, it is not possible to extract the mechanical properties of an adhesive independent of the wood due to adhesive-wood interactions or the thin, non-uniform nature of the bondline. Nanoindentation, by contrast, is able to reliably extract mechanical properties of an adhesive in the glueline or in a single lumen (for references see Table 2). Dynamic mechanical analysis (DMA), sometimes called dynamic thermal mechanical analysis (DTMA), has been used to probe the changes in mechanical properties of wood pieces with adhesive infiltration. The mechanical response with temperature, moisture, and time scale has provided new insights into the interaction between adhesives and wood polymers [132, 134, 348]. Mechanical properties of cubic micrometer scale volumes, particularly the impact of cell wall penetration, are now commonly probed using nanoindentation (see Table 2). Many interesting discoveries are expected as more labs acquire humidity and temperature control of the specimen chamber, and the practice of dynamic analysis using nanoindentation becomes more common.

8.5

Unresolved Questions in Wood Bonding Research

Ideally, scientists and engineers should be able to predict bond strength and durability with an adhesive of known composition used for bonding wood of known characteristics under a prescribed set of bonding conditions. While we are still far from this goal, progress in measuring and understanding properties from the macro to the molecular scale is bringing us closer to this goal. Much of this progress in understanding comes from using tools beyond the naked eye for examination of bond properties and bond failure. Even more progress is possible when multiple techniques, measuring complementary properties, are brought together on the same sample set. This approach not only leads to a more thorough understanding of a particular case, but also uncovers inconsistencies and contradictions between the conclusions from different approaches. These contradictions, or conversely, the ability to quantitatively predict properties, are fertile areas for developing our understanding, but often requires a single sample to be studied by multiple, diverse, quantitative methods.

8.5.1

How Do We Make Wood Surfaces Better for Bonding?

Adhesives need to wet the wood surface. Water drop tests are a quick way of assessing wettability with waterborne adhesives, and a large decrease in contact angle after a solvent

394 Progress in Adhesion and Adhesives, Volume 4

wipe suggests that the wipe removed an oily surface that could have caused a chemical weak boundary layer. However, more sophisticated methods are needed beyond these tests. Mechanical weak boundary layers can be identified with cross-sectional microscopy and a tape test can be used to measure loose wood surface pieces [63]. There is no proven test for evaluating the frequency of crushed and buckled cells on the surface or their impact on bond performance. The tests we do analyze small areas and do not give a view of the entire surface. Can large surfaces be quickly and efficiently analyzed before bonding? Macroscopically rough surfaces can easily be judged by their look and feel, finer measurements have not led to consistent comparisons with bonding performance. What techniques can be developed to relate surface properties to bond performance? Could an online monitoring system be developed that would allow dynamic process modifications to account for the quality of the wood surface? Penetration of the adhesive into the wood is important for developing a strong bond, but how do we assess wood porosity in relation to adhesive penetration? Porosity of the wood is considered important for forming good bonds, but too much porosity can lead to excess costs and starved bonds. Is there a way to judge wood porosity prior to its use so that the adhesive spread rate and/or its formulation can be adjusted prior to production? The swelling of wood in bonded products is a major problem. Are there ways to measure and reduce stress this swelling causes in the bonded assemblies? How important is the high pH in phenolic resins for promoting good contact between bulk adhesive and the cell wall, as well as cell wall penetration? If the cut-open cells on the surface are important for mechanical interlocking and bonding, does the chemical structure of the surface of wood matter? Does the change in composition of the cell lumen walls between species affect bonding?

8.5.2

Does the Adhesive Have Good Penetration Into the Wood Structure?

The question is, what does “good penetration” mean? Too little or too much penetration will cause low bond strength and therefore bond failures, especially when the bond is exposed to water. Starved gluelines typically show cohesive failure in the bulk adhesive or in the adhesive interphase. Thick gluelines from low penetration or excessive spread rate are also weak [349] and waste adhesive. Several of the EN standards for testing and classification of bonds require the preparation and testing of thick joints [190, 350]. How do we determine the most important factors for proper adhesive penetration, viscosity of the adhesive, molecular weight of the adhesive components, loss of water to the wood, adhesive cure rate, pressure, temperature of bonding, etc.? How do these factors interact with wood properties?

8.5.3

How Does the Adhesive Interact with the Wood at the Nanoscale and Molecular Level?

Although wood bond formation and fracture seem like macroscopic properties, they are the sum of a multitude of interactions at different length scales. How do we determine the

Understanding Wood Bonds–Going Beyond What Meets the Eye 395

relative contribution of chemical cross-linking vs. secondary chemical bonding (van der Waals, hydrogen bond, etc.) vs. mechanical interlock? Once we begin to understand them, how do we use that information to improve bond performance?

8.5.4

Can We Improve the Resistance of Bonds to the Dimensional Changes in Wood with Variation in Moisture?

The level of moisture-induced stresses in bondlines depends on various factors, such as the geometrical properties of the lamellas and the glued element as well as the wood properties. With increasing density, shrinkage and swelling coefficients as well as stiffness and strength properties generally increase, so moisture changes generate higher stress levels [5]. Durable bonds have been difficult to make with ash (Fraxinus excelsior L.) and beech (Fagus sylvatica L.), because these wood species have significantly higher densities than other wood species typically used for glulam in Europe, such as spruce (Picea abies L. Karst) [241, 245, 351]. An adhesive that expands with the wood as it wets may partially explain the moisture durability of wood bonded with such resin systems. Some success in studying the hygro-mechanical behavior of adhesive films has been realized [173, 287, 343, 352–354]. It has been demonstrated that the mechanical properties of cured adhesives depend on their MC, so this must be considered in modeling and predicting bond performance [345, 355]. PUR and EPI adhesives are more elastic and therefore allow for smoother strain transition, showing less distinct strain peaks than PRF and MUF [199]. How does the bond withstand not only dynamic loading, but also duration of load and fatigue? Which types of loadings are worst for different types of adhesives? With regard to moisture durability, was Frihart really correct about how the rigid and brittle adhesive layers of in-situ polymerized adhesives and the flexible bondlines of prepolymerized adhesives distribute stress [142]? Are there details still to be worked out, such as a gradual reduction in brittleness/increased ductility with in-situ polymerized adhesives? Can this theory be turned into predictions of how much void and cell wall penetration are necessary for a given pair of wood surfaces?

8.5.5

How do Primers Work?

Kläusler et al. reported that the tensile shear strength and wood failure percentage of wet bonds were substantially lower than the samples that were dried before testing [356]. The microscopic images suggest that the wet samples lose adhesion. Improvement of tensile shear strength and wood failure percentage of 1C PUR bonded wood joints tested wet was achieved with HMR and dimethylformamide primers. In some cases, the adhesives provide good strength, but a low percentage of wood failure. Primers, in particular HMR, have been found to be effective in improving both wood failure and bond strength for a variety of adhesives and wood species, but the mechanisms of primers is unclear [304, 321, 357–361]. Is it possible that HMR allows the joint to benefit from the dimensional stabilization and reinforcement of the interphase typically afforded by in-situ polymerized adhesives, as well as from the ductility of a prepolymerized adhesive? In what situations do primers help

396 Progress in Adhesion and Adhesives, Volume 4

the most? Why? What do primers do, what bond properties are influenced, and when will primers improve bond quality?

8.5.6

Where Does the Bond Failure Initiate and How Does it Propagate?

One characteristic of failure propagation is the stress concentration at the edge of a defect exceeding the adhesion or cohesive strength at that point. The growth of micro-cracks at stress concentrations results in a gradual loss of mechanical integrity and, hence, weakening of the joint. While bond strength is determined by the weakest link in the adhesive bonding chain, it is currently difficult or impossible to predict whether a crack will remain where it started or move to the glueline, the interphase, or to the wood. In an ideal bond, or one that passes most quality standards, most of the failure occurs in wood, far from the bondline, so the only information obtained is that the bond strength exceeds the wood strength. Given the complexity of crack propagation, how can we understand the source and propagation of failures? Different failures need to be studied by different, often complementary methods to arrive at firm conclusions. Do the different methods agree? How do we take the understanding of a few observed failures and apply that knowledge to all the possible variations that occur in commercial wood bonds? Even with good information about a particular failure, wood is so varied that a statistical sample of different possible configurations is warranted. When stresses exceed bond strength, delaminations can develop and affect the remaining service life [362]. How do we consistently determine the effect of an accumulation of microdelaminations that over time, can lead to macroscopic failure?

8.5.7

How Do We Optimize the Benefits of Cell Wall Penetration?

It is well known that pMDI is an effective adhesive, but it has been shown not to react with the wood polymers [158]. It is unclear how much the pMDI enters the wood cell walls because sometimes it is observed to penetrate the cell wall [134, 135, 170] and sometimes not [136]. Grigsby and Thumm claimed that either MDI or pMDI (the paper is ambiguous) penetration into MDF fiber was present but very shallow [218]. The question is, how important is pMDI cell wall penetration to bond performance? Do the details of how a bond is made determine whether or how much cell wall penetration occurs? If pMDI does not enter the cell wall and does not bind to cell wall polymers, how does it develop good bonds? The literature shows that some adhesives enter the cell wall and others do not usually because of molecular weight limitations for cell wall infiltration. Less well understood is the impact of solubility parameter of the adhesive and moisture content of the wood on wood bonding. Adhesive solubility parameter could determine in what domain of the cell wall the adhesive components will reside, and how the adhesive modifies the properties of that domain. Does cell wall infiltration contribute to adhesive interphase failure in epoxy by allowing the amine monomers to enter the cell wall while excluding bisphenol A, resulting in a depletion of amine? Does this occur in other adhesive systems as well?

Understanding Wood Bonds–Going Beyond What Meets the Eye 397

8.5.8

How Does the Adhesive Form a Suitable Polymer Matrix to Bridge Between the Two Wood Surfaces?

The chemical structure of the solid adhesive influences the adhesive’s cohesive strength (rigid and brittle vs. ductile and elastic) and the temperature behavior of the glueline, with duroplastic behavior relatively independent of temperature and thermoplastic behavior losing bond strength at the glass transition temperature (Tg) of the solid adhesive. The formation of the solid bondline typically involves shrinkage due to the loss of the water and polymerization or cross-linking of the adhesive molecules. How does the bond deal with this volume reduction without creating too much internal strain? How does the adhesive respond to the strain from external mechanical forces as well as from wood swelling and shrinking? Significant differences have been observed between the properties of adhesives intended for solid wood products compared to adhesives intended for use in composites [196]. Does multiscale modeling and simulation of wood adhesion provide useful insight into bonded wood performance? Does this allow data from standard laminated samples to be used for understanding cross-laminated structures and new types of wood products such as mass timber or plywood?

8.5.9

Will Adhesives Based on Renewable Resources be the Future in Wood Bonding?

The very first glues were made of natural, renewable raw materials. Today, the great majority of adhesives are derived from fossil fuels because of performance and cost. Nonetheless, there has been great interest in renewable adhesives in recent years driven by concerns about human health and sustainability. The main safety and health concerns have focused on urea-formaldehyde and isocyanates. The biomaterials under investigation include lignin, proteins, bio-oils, hemicelluloses, and others. Several overview and review papers have been published [3, 23, 27, 363–366]. Expressing the renewable content will avoid some confusion about what is a biobased adhesive, and may promote the use of lignin and proteins both as base adhesives and as extenders, as well as provide a mechanism to give credit to existing bio-based fillers and extenders. There appears to be demand for bio-based wood adhesives in the marketplace, especially in Europe. For example, one of the world’s largest furniture retailers, Ikea, has committed to 40% natural raw materials for all their adhesives by 2025 and 80% by 2030 [367]. The broader use of bio-based content in adhesives is hindered by a lack of information and understanding on these bio-based systems. Synthetic adhesives are often more consistent than natural products. In synthetic adhesives, the manufacturer starts with known, well characterized, relatively pure raw materials of consistent quality and properties. Natural materials are often much more complex in composition, structure, and seasonal variability. This leads to the following questions:

398 Progress in Adhesion and Adhesives, Volume 4

What is the structure of the cured adhesive and how do variations in the cure conditions influence mechanical properties and performance? How do these bio-based resins interact with the wood? What components can be added and under what conditions to improve this interaction? How does an adhesive manufacturer manage the differences in the growth and extraction of natural materials? How do they deal with variation from season to season, and as the natural material ages after harvest, and geographic differences? How do they ensure a consistent and continuous supply of the raw materials? How do they characterize these extremely complex raw materials for their adhesive properties, when we understand so little about what controls the final bond strength with these materials?

8.5.10

How Much the Experience with Solid Wood Bonding can be Used to Understand Wood Based Particulate Bonding?

Oriented strandboard (OSB) and particleboard (PB) are assembled from wood particles, while the structure of fiberboard fibers is substantially different from the original wood. As long as solid wood is present (independent of the size of the strands or particles), the basic principles of bonding two wood surfaces remain applicable. In principle, the only difference between solid wood bonding and panel production is the size of an individual bond: several m2 for plywood and cross-laminated timber, possibly as much as 2000 mm2 for two OSB strands, and finally as small as a few mm2 when bonding PB particles. However, the relative importance of various effects will change: some adverse effects that are minor in solid wood bonding become much more common, such as the buckling of cell walls during pressing and the combination of different grain directions. This is especially important in PB, where no consistent grain direction of the particles is present. The main test methods of mechanical properties for PB, OSB, and MDF are internal bond, bending strength, and bending stiffness; however no evaluation is done concerning proportion of wood failure simply because of the difficulty of the procedure. How do we relate what we find with laminated wood to glued wood products that have spot welds instead of a continuous adhesive layer? Do the test results reflect more the wood particle properties and density profile or the adhesive properties?

8.5.11

How Do We Compare Results Obtained in Different Laboratories with Different Wood Species with Different Adhesives?

Through the extraordinary work of researchers around the world, we are learning many things about wood bonding. However, it can be hard to know how broadly the results can be extrapolated. How do we know the repeatability of the results, with different researchers and different wood samples, even when they are of the same species? How do we compare work done in Europe on beech or ash with that done in the US using Douglas fir and loblolly pine? How do we relate the work on penetration and bond performance from one set of wood/adhesive/surface preparation/bonding conditions to another?

Understanding Wood Bonds–Going Beyond What Meets the Eye 399

8.6

Summary

Wood bonding is enormously complex given the multitude of variables and unknowns. The intelligent empirical work up to now has led us to bonded wood products that generally meet the customer’s need for performance, safety and cost. However, there are always drivers for improvement, whether they are product cost reduction, wood type and quality, new types of bonded wood products, or societal changes. There has been substantial research into why wood bonds fail and what is needed so that they do not. Much of the research discussed in this review has focused on factors that cannot be observed by examination with the naked eye. The factors that contribute to bond performance span several microscopic levels, including the cellular, nano, and even molecular level. This review has tried to summarize the breadth of this work and provides challenges for future research.

List of Abbreviations AFM AP CLSM CV CWBL DCB DIC DMA DOC DSC EDXS EELS EP EPI ESPI FPL FT-IR HMR LVL MC MDF MF MIP MOE MOR MP MUF MW

Atomic Force Microscopy Average Penetration Confocal Laser Scanning Microscopy Collapsed Vessel Chemical Weak Boundary Layer Dual Cantilever Beam Digital Image Correlation Dynamic Mechanical Analysis (also known as DMTA) Degree of Condensation Differential Scanning Calorimetry Energy Dispersive X-Ray Spectroscopy, also known by EDX, EDXA, and XEDS Electron Energy Loss Spectroscopy Effective Penetration Emulsion Polymer Isocyanate Electronic Speckle Pattern Interferometry Forest Products Laboratory Fourier Transform-Infrared Hydroxymethylated Resorcinol Laminated Veneer Lumber Moisture Content Medium Density Fiberboard Melamine-Formaldehyde Mercury Intrusion Porosimetry Modulus of Elasticity Modulus of Rupture Maximum Penetration Melamine-Urea-Formaldehyde Molecular Weight

400 Progress in Adhesion and Adhesives, Volume 4

MWBL NIR NMR OSB PB PF pMDI PRF PUR PVAc QENS SANS SEM SThM TEM Tg TMA UF WDS XCT XFM XPS

Mechanical Weak Boundary Layer Near-Infrared Nuclear Magnetic Resonance Oriented Strand Board Particleboard Phenol-Formaldehyde polymeric Methylene Diphenyl Diisocyanate Phenol-Resorcinol-Formaldehyde Polyurethane Poly(vinyl acetate) Quasi-Elastic Neutron Scattering Small Angle Neutron Scattering Scanning Electron Microscopy Scanning Thermal Microscopy Transmission Electron Microscopy Glass Transition Temperature Thermal Mechanical Analysis Urea-Formaldehyde Wavelength Dispersive Spectroscopy X-Ray Computed (Micro)Tomography X-Ray Fluorescence Microscopy X-ray Photoelectron Spectroscopy, also known as ESCA

References All recent publications authored by USDA Forest Service, Forest Products Laboratory (FPL) employees are freely available to the public and can be found through internet search or downloaded from the FPL website, https://www.fpl.fs.fed.us/search. FPL employees referenced include: A.W. Christiansen, C.R. Frihart, C.G. Hunt, J.J. Jakes, N. Plaza, B.H. River, H. Tarkow, C. Vick, A.C. Wiedenhoeft, D.J. Yelle, and J. Youngquist. 1. F.A. Keimel, Historical development of adhesives and adhesive bonding, in Handbook of Adhesive Technology, second edition, A. Pizzi and K.L. Mittal (Eds), pp. 1–12, CRC Presss, Boca Raton, FL (2003). 2. J.W. Koning, Forest Products Laboratory, 1910–2010, University of Wisconsin Press, Madison, WI (2010). 3. A. Pizzi, Wood products and green chemistry. Annals Forest Sci. 73, 185–203 (2016). 4. C.R. Frihart, Wood adhesives: Vital for producing most wood products. Forest Prod. J. 61, 4–12 (2011). 5. A.A. Marra, Technology of Wood Bonding: Principles in Practice, Van Nostrand Reinhold, New York (1992). 6. C.R. Frihart, Wood structure and adhesive bond strength, in Characterization of the Cellulosic Cell Wall, D. Stokke and L. Groom (Eds), pp. 241–253, Blackwell Publishing, Ames, IA (2006).

Understanding Wood Bonds–Going Beyond What Meets the Eye 401

7. D.J. Gardner, Adhesion mechanisms of durable wood adhesive bonds, in Characterization of the Cellulosic Cell Wall, D. Stokke and L. Groom (Eds), pp. 254–265, Blackwell Publishing, Ames, IA (2006). 8. R.E. Baier, E.G. Shafrin, and W.A. Zisman, Adhesion: Mechanisms that assist or impede it. Science 162, 1360–1368 (1968). 9. J.J. Bikerman, The Science of Adhesive Joints, second edition, Academic Press, New York (1968). 10. R.J. Good, Theory of “cohesive” vs “adhesive” separation in an adhering system. J. Adhesion 4, 133–154 (1972). 11. L.H. Sharpe, Interfaces, interphases and “adhesion”: A perspective, in The Interfacial Interactions in Polymeric Composites, Vol 230, G. Akovali (Ed), pp. 1–20, NATO ASI Series (Series E: Applied Sciences) Springer, Dordrecht (1993). 12. M. Stehr and I. Johansson, Weak boundary layers on wood surfaces. J. Adhesion Sci. Technol 14, 1211–1224 (2000). 13. F.H. Chung, Unified theory and guidelines on adhesion. J. Appl. Polym. Sci. 42, 1319–1331 (1991). 14. A.V. Pocius, Adhesion and Adhesives Technology: An Introduction, third edition, Carl Hanser Verlag (2012). 15. J. Schultz and M. Nardin, Theories and mechanisms of adhesion, in Handbook of Adhesive Technology, second edition, A. Pizzi and K.L. Mittal (Eds), pp. 53–67, CRC Press, Boca Raton, FL (2003). 16. C.R. Frihart, Wood adhesion and adhesives, in Handbook of Wood Chemistry and Wood Composites, second edition, R.M. Rowell (Ed) pp. 255–313, CRC Press, Boca Raton, FL (2013). 17. D.J. Gardner, M. Blumentritt, L. Wang, and N. Yildirim, Adhesion theories in wood adhesive bonding. Rev. Adhesion Adhesives 2, 127–172 (2014). 18. K.L. Mittal, The role of the interface in adhesion phenomena. Polym. Eng. Sci. 17, 467–473 (1977). 19. J.W. McBain and D.G. Hopkins, On adhesives and adhesive action. J. Phys. Chem. 29, 188–204 (1925). 20. F. Browne and D. Brouse, Nature of adhesion between glue and wood 1: A criticism of the hypothesis that the strength of glued wood joints is due chiefly to mechanical adhesion. Ind. Eng. Chem. 21, 80–84 (1929). 21. C.R. Frihart, Adhesive bonding and performance testing of bonded wood products. ASTM International 2, 1–12, February (2005). 22. B.H. River, C. Vick, and R.H. Gillespie, Wood as an adherend, in Treatise on Adhesion and Adhesives, J.D. Minford (Ed), pp. 1–238, Marcel Dekker, New York (1991). 23. M. Dunky, Adhesives in the wood industry, in Handbook of Adhesive Technology, second edition, A. Pizzi and K.L. Mittal (Eds), pp. 887–956, CRC Press, Boca Raton, FL (2003). 24. A.W. Christiansen, How overdrying wood reduces its bonding to phenol-formaldehyde adhesives: A critical review of the literature. Part II. Chemical reactions. Wood Fiber Sci. 23, 69–84 (1991). 25. A.W. Christiansen, How overdrying wood reduces its bonding to phenol-formaldehyde adhesives - A critical review of the literature. 1. Physical responses. Wood Fiber Sci. 22, 441–459 (1990). 26. A. Panshin and C. de Zeeuw, Textbook of Wood Technology: Structure, Identification, Properties, and Uses of the Commercial Woods of the United States and Canada, fourth edition, McGrawHill College, Blacklick, OH (1980).

402 Progress in Adhesion and Adhesives, Volume 4

27. M. Dunky, Adhesives in the wood industry, in Handbook of Adhesive Technology, third edition, A. Pizzi and K.L. Mittal (Eds), pp. 511–574, CRC Press, Boca Raton, FL (2017). 28. P.R. Larson, D.E. Kretschmann, A.I. Clark, and J.G. Isebrands, Formation and properties of juvenile wood in southern pines: A synopsis, General Technical Report FPL-GTR-129. 2001, 42 pp. U.S. Department of Agriculture, Forest Service, Forest Products Laboratory, Madison, WI, USA. 29. J.G. Haygreen and J. Bowyer, Forest Products and Wood Science, Iowa State University Press, Ames, IA (1996). 30. S. Cramer, D. Kretschmann, R. Lakes, and T. Schmidt, Earlywood and latewood elastic properties in loblolly pine. Holzforschung 59, 531–538 (2005). 31. D.E. Kretschmann, S.M. Cramer, R. Lakes, and T. Schmidt, Selected mesostructure properties in loblolly pine from Arkansas plantations, in Characterization of the Cellulosic Cell Wall, D. Stokke and L. Groom (Eds), pp. 149–170, Blackwell Publishing, Ames, IA (2006). 32. J. Konnerth, M. Kluge, G. Schweizer, M. Miljkovic´, and W. Gindl-Altmutter, Survey of selected adhesive bonding properties of nine European softwood and hardwood species. European J. Wood Wood Products 74, 809–819 (2016). 33. C.R. Frihart and C.G. Hunt, Adhesives with wood materials- Bond formation and performance, in Wood Handbook: Wood as an Engineering Material, R. Ross (Ed), pp. 10.1–10.24, General Technical Report FPL-GTR-190. U.S. Department of Agriculture, Forest Service, Forest Products Laboratory, Madison, WI (2010). 34. FPL, Gluing of wood, in Wood Handbook: Wood as an Engineering Material 1974, US Government Printing Office, Washington, DC. 35. FPL, Wood Handbook: Wood as an Engineering Material, Vol 72, USDA Forest Products Laboratory, Madison WI (2010). 36. Supplement No. 13 to the Southern Pine Inspection Bureau Grading Rules 2002 Edition, SPIB (2013). 37. B.J. Zobel and J.P. Van Buijtenen (Eds), Wood Variation: Its Causes and Control, Springer Science & Business Media, Berlin (1989). 38. C.B. Vick and R.M. Rowell, Adhesive bonding of acetylated wood. Intl. J. Adhesion Adhesives 10, 263–272 (1990). 39. C.R. Frihart, R. Brandon, J.F. Beecher, and R.E. Ibach, Adhesives for achieving durable bonds with acetylated wood. Polymers 9, 731–742 (2017). 40. B. Bryant, Wood adhesion, in Handbook of Adhesives, second edition, I. Skeist (Ed), pp. 669–678, Van Nostrand Reinhold Company, New York (1977). 41. T. Ebihara, Shear properties of laminated-veneer lumber (LVL) (In Japanese). J. Japan Wood Res. Soc. 27, 788–791 (1981). 42. P. Koch, Effects of seven variables on properties of southern pine plywood. Part II, Maximizing wet shear strength. Forest Prod. J. 15, 463–465 (1965). 43. S. Chow, Lathe-check influence on plywood shear strength, Report VP-X-122:24. 1974, Canadian Forest Service, Vancouver, BC. 44. J. Neese, J. Reeb, and J. Funck, Relating traditional surface roughness measures to gluebond quality in plywood. Forest Prod. J. 54, 67–73 (2004). 45. D. DeVallance, J. Funck, and J. Reeb, Influence of several preparation conditions on Douglas-fir plywood gluebond quality test results. Forest Prod. J. 56, 47–50 (2006). 46. J.S. Bethel and J.B. Huffman, Influence of lathe checks orientation on plywood shear test results, Technical Report #1. 1950, Agricultrual Experiment Station, North Carolina State College, Raleigh, NC.

Understanding Wood Bonds–Going Beyond What Meets the Eye 403

47. A. Rohumaa, C.G. Hunt, M. Hughes, C.R. Frihart, and J. Logren, The influence of lathe checking depth and orientation on the bond quality of phenol-formaldehyde-bonded birch plywood. Holzforschung 67, 779–786 (2013). 48. A. Rohumaa, The impact of log preheating on birch veneer surface quality, bond formation and plywood performance. PhD thesis, Aalto University, Helsinki, Finland (2016). 49. G. Pot, L.-E. Denaud, and R. Collet, Numerical study of the influence of veneer lathe checks on the elastic mechanical properties of laminated veneer lumber (LVL) made of beech. Holzforschung 69, 337–345 (2015). 50. W. Darmawan, D. Nandika, Y. Massijaya, A. Kabe, I. Rahayu, L. Denaud, and B. Ozarska, Lathe check characteristics of fast growing sengon veneers and their effect on LVL glue-bond and bending strength. J. Mater. Process. Technol. 215, 181–188 (2015). 51. I. Rahayu, W. Darmawan, N. Nugroho, and R. Marchal, The effect of jabon veneer quality on laminated veneer lumber glue bond and bending strength. Jurnal Ilmu dan Teknologi Kayu Tropis 13, 98–110 (2015). 52. A. Gent and S. Lai, Adhesion and autohesion of rubber compounds: Effect of surface roughness. Rubber Chem. Technol. 68, 13–25 (1995). 53. D.E. Packham, Surface energy, surface topography and adhesion. Intl. J. Adhesion Adhesives 23, 437–448 (2003). 54. A. Baldan, Adhesion phenomena in bonded joints. Intl. J. Adhesion Adhesives 38, 95–116 (2012). 55. I. Aydin, G. Colakoglu, and S. Hiziroglu, Surface characteristics of spruce veneers and shear strength of plywood as a function of log temperature in peeling process. Intl. J. Solids Structures 43, 6140–6147 (2006). 56. T. Faust and J. Rice, Characterizing the roughness of southern pine veneer surfaces. Forest Prod. J. 36, 75–81 (1986). 57. D. DeVallance, J. Funck, and J. Reeb, Douglas-fir plywood gluebond quality as influenced by veneer roughness, lathe checks, and annual ring characteristics. Forest Prod. J. 57, 21–28 (2007). 58. I. Aydin, Activation of wood surfaces for glue bonds by mechanical pre-treatment and its effects on some properties of veneer surfaces and plywood panels. Appl. Surface Sci. 233, 268–274 (2004). 59. T. Sellers, Plywood and Adhesive Technology, Marcel Dekker, New York (1985). 60. D.C. Walser and T.A. McLauchlan, A new development in veneer peeling, in: Proceedings of Forest Products Research Society Annual Meeting (1977). 61. C.E. Frazier and X. Wang, Tack measurements for resole formuations, in: Proceedings of International Conference on Wood Adhesives, pp. 340–356 (2017). 62. B.H. River, Fracture of adhesively-bonded wood joints, in Handbook of Adhesive Technology, A. Pizzi and K.L. Mittal (Eds), pp. 151–177, Marcel Dekker, New York (1994). 63. A. Rohumaa, T. Antikainen, C.G. Hunt, C.R. Frihart, and M. Hughes, The influence of log soaking temperature on surface quality and integrity performance of birch (Betula pendula Roth) veneer. Wood Sci. Technol 50, 463–474 (2016). 64. A. Rohumaa, C.G. Hunt, C.R. Frihart, P. Saranpää, M. Ohlmeyer, and M. Hughes, The influence of felling season and log-soaking temperature on the wetting and phenol-formaldehyde adhesive bonding characteristics of birch veneer. Holzforschung 68, 965–970 (2014). 65. V. Möttönen and K. Luostarinen, Variation in density and shrinkage of birch (Betula pendula Roth) timber from plantations and naturally regenerated forests. Forest Prod. J. 56, 34–39 (2006). 66. A. Rohumaa, A. Yamamoto, C.G. Hunt, C.R. Frihart, M. Hughes, and J. Kers, Effect of log soaking and the temperature of peeling on the properties of rotary-cut birch (Betula pendula Roth) veneer bonded with phenol-formaldehyde adhesive. Bioresources 11, 5829–5838 (2016).

404 Progress in Adhesion and Adhesives, Volume 4

67. M. Dunky and P. Niemz, Wood Based Panels and Resins: Technology and Influence Parameters (in German), Springer, Heidelberg (2002). 68. M.P. Wolcott, F.A. Kamke, and D.A. Dillard, Fundamental aspects of wood deformation pertaining to manufacture of wood-based composites. Wood Fiber Sci. 26, 496–511 (1994). 69. M.A. Irle, Cell wall collapse and the pressing of particleboard, in: Proceedings of 4th European Panel Products Symposium, pp. 70–80 (2000). 70. I. Gavrilovic´-Grmuša, M. Dunky, M. Djiporovic´-Momcˇilovic´, M. Popovic´, and J. Popovic´, Influence of pressure on the radial and tangential penetration of adhesive resin into poplar wood and on the shear strength of adhesive joints. Bioresources 11, 2238–2255 (2016). 71. E. Mahrdt, F. Stöckel, H.W.G. van Herwijnen, U. Müller, W. Kantner, J. Moser, and W. GindlAltmutter, Light microscopic detection of UF adhesive in industrial particle board. Wood Sci. Technol. 49, 517–526 (2015). 72. M.A. Tshabalala, J. Jakes, M.R. VanLandingham, S. Wang, and J. Peltonen, Surface characterization, in Handbook of Wood Chemistry and Wood Composites, second edition, R. Rowell (Ed), pp.217–252, CRC Press, Boca Raton, FL (2013). 73. C.H. Hse and M. Kuo, Influence of extractives on wood gluing and finishing—A review. Forest Prod. J 38, 52–56 (1988). 74. V. Gray, The wettability of wood. Forest Prod. J 12, 452–461 (1962). 75. T. Nguyen and W.E. Johns, The effects of aging and extraction on the surface free energy of Douglas fir and redwood. Wood Sci. Technol. 13, 29–40 (1979). 76. M. Scheikl, M. Dunky, and H. Resch, Wettability of European soft- and hardwoods by ureaformaldehyde adhesives, in: Proceedings of International Conference on Wood Adhesives, pp. 43–46 (1995). 77. D. Hare and N. Kutscha, Microscopy of eastern spruce plywood gluelines. Wood Sci. 6, 294–304 (1974). 78. Standard test method for strength properties of adhesive bonds in shear by compression loading, ASTM D905-08 (2013) 79. Standard specification for adhesives for structural laminated wood products for use under exterior exposure conditions, ASTM D2559-12ae1 (2011). 80. H.J. Zhang, D.J. Gardner, J.Z. Wang, and Q. Shi, Surface tension, adhesive wettability, and bondability of artificially weathered CCA-treated southern pine. Forest Prod. J. 47, 69–72 (1997). 81. R.M. Nussbaum and M. Sterley, The effect of wood extractive content on glue adhesion and surface wettability of wood. Wood Fiber Sci. 34, 57–71 (2007). 82. M. De Meijer, S. Haemers, W. Cobben, and H. Militz, Surface energy determinations of wood: Comparison of methods and wood species. Langmuir 16, 9352–9359 (2000). 83. G. Sinn, J. Sandak, and T. Ramananantoandro, Properties of wood surfaces - characterisation and measurement. A review COST Action E35 2004 -2008: Wood machining - micromechanics and fracture. Holzforschung 63, 196–203 (2009). 84. J. Rathke and G. Sinn, Evaluating the wettability of MUF resins and pMDI on two different OSB raw materials. European J. Wood Wood Products 71, 335–342 (2013). 85. M. Scheikl and M. Dunky, Measurement of dynamic and static contact angles on wood for the determination of its surface tension and the penetration of liquids into the wood surface. Holzforschung 52, 89–94 (1998). 86. E.W. Washburn, The dynamics of capillary flow. Phys. Rev. 17, 273–283 (1921). 87. C.-Y. Hse, Properties of phenolic adhesives as related to bond quality in southern pine plywood. Forest Prod. J. 21, 44–52 (1971).

Understanding Wood Bonds–Going Beyond What Meets the Eye 405

88. M. Petricˇ and P. Oven, Determination of wettability of wood and its significance in wood science and technology: A critical review. Rev. Adhesion Adhesives 3, 121–187 (2015). 89. D.A. Stumbo, Influence of surface aging prior to gluing on bond strength of Douglas-fir and redwood. Forest Prod. J. 14, 582–589 (1964). 90. E. Back, Oxidative activation of wood surfaces for glue bonding. Forest Prod. J. 41, 30–36 (1991). 91. K. Plomley, W. Hillis, and K. Hirst, The influence of wood extractives on the glue-wood bond. I. The effect of kind and amount of commercial tannins and crude wood extracts on phenolic bonding. Holzforschung 30, 14–19 (1976). 92. S. Bockel, I. Mayer, J. Konnerth, P. Niemz, C. Swaboda, M. Beyer, S. Harling, G. Weiland, N. Bieri, and F. Pichelin, Influence of wood extractives on two-component polyurethane adhesive for structural hardwood bonding. J. Adhesion 94, 1–17 (2018). 93. E. Roffael, Significance of wood extractives for wood bonding. Appl. Microbiol. Biotechnol. 100, 1589–1596 (2016). 94. J. Slay, P. Short, and D. Wright, Catalytic effects of extractives from pressure-refined fiber on gel time of urea-formaldehyde resin. Forest Prod. J. 30, 22–23 (1980). 95. Y. Akaike, The inhibitory effect of kapur wood extracts on the gelation of the urea resin adhesive. Mokuzai Gakkaishi 20, 224–229 (1974). 96. Z. Gao, X.-M. Wang, H. Wan, and Y. Liu, Curing characteristics of urea–formaldehyde resin in the presence of various amounts of wood extracts and catalysts. J. Appl. Polym. Sci. 107, 1555–1562 (2008). 97. C. Xing, S. Zhang, J. Deng, B. Riedl, and A. Cloutier, Medium-density fiberboard performance as affected by wood fiber acidity, bulk density, and size distribution. Wood Sci. Technol. 40, 637–646 (2006). 98. C. Xing, S. Zhang, and J. Deng, Effect of wood acidity and catalyst on UF resin gel time. Holzforschung 58, 408–412 (2004). 99. G. Han, K. Umemura, S. Kawai, and H. Kajita, Improvement mechanism of bondability in UF-bonded reed and wheat straw boards by silane coupling agent and extraction treatments. J. Wood Sci. 45, 299–305 (1999). 100. S. Medved and J. Resnik, Influence of the acidity and size of beech particles on the hardening of the urea-formaldehyde adhesive. Acta Chimica Slovenica 51, 353–360 (2004). 101. N. André, T. Young, and T. Rials, On-line monitoring of the buffer capacity of particleboard furnish by near-infrared spectroscopy. Appl. Spectroscopy 60, 1204–1209 (2006). 102. R. Pedieu, B. Riedl, and A. Pichette, Measurement of wood and bark particles acidity and their impact on the curing of urea formaldehyde resin during the hot pressing of mixed panels. Holz Roh Werkstoff 66, 113–117 (2008). 103. B. Stefke and M. Dunky, Catalytic influence of wood on the hardening behavior of formaldehyde-based resin adhesives used for wood-based panels. J. Adhesion Sci. Technol. 20, 761–785 (2006). 104. E. Roffael and W. Rauch, Extractives of oak and their influence on gluing with alkaline phenolicformaldehyde resins. Holz Roh Werkstoff 32, 182–187 (1974). 105. W.E. Johns and K.A. Niazi, Effect of pH and buffering capacity of wood on the gelation time of urea-formaldehyde resin. Wood Fiber Sci. 12, 255–263 (1981). 106. V. Gray, The acidity of wood. J. Inst. Wood Sci. 1, 58–64 (1958). 107. D.J. Gardner, M.P. Wolcott, L. Wilson, Y. Huang, and M. Carpenter, Our understanding of wood surface chemistry in 1995, in: Proceedings of International Conference on Wood Adhesives, pp. 29–36 (1995).

406 Progress in Adhesion and Adhesives, Volume 4

108. A. Yamamoto, A. Rohumaa, E. Kontturi, M. Hughes, P. Saranpää, M. Andberg, and T. Vuorinen, Colorimetric behavior and seasonal characteristic of xylem sap obtained by mechanical compression from silver birch (Betula pendula). ACS Sustainable Chem. Eng. 1, 1075–1082 (2013). 109. P. Hanetho, Seasonal induced quality problems with particleboards by using fresh wood (in German), in: Proceedings of FESYP, pp. 129–136 (1987). 110. F.A. Kamke and J.N. Lee, Adhesive penetration in wood - A review. Wood Fiber Sci. 39, 205–220 (2007). 111. J. Resnik, M. Šernek, and F.A. Kamke, High-frequency heating of wood with moisture content gradient. Wood Fiber Sci. 29, 264–271 (2007). 112. H. Baumann and J. Marian, Gluing pressure with wood as a function of physical factors. Holz Roh Werkst 11, 441–446 (1961). 113. R. Kurt and M. Cil, Effects of press pressures on glue line thickness and properties of laminated veneer lumber glued with phenol formaldehyde adhesive. Bioresources 7, 5346–5354 (2012). 114. S.E. Johnson and F.A. Kamke, Quantitative analysis of gross adhesive penetration in wood using fluorescence microscopy. J. Adhesion 40, 47–61 (1992). 115. G.I. Mantanis, R.A. Young, and R.M. Rowell, Swelling of wood, Part II Swelling in organic liquids. Holzforschung 48, 480–490 (1994). 116. J.E. Stone and A.M. Scallan, A structural model for the cell wall of water swollen wood pulp fibres based on their accessibility to macromolecules. Cellulose Chem. Tehnol 2, 343–358 (1968). 117. J.E. Jakes, N.Z. Plaza, X. Arzola-Villegas, and C.R. Frihart, Improved Understanding of Moisture Effects on Outdoor Wood-Adhesive Bondlines, General Technical Report 246. 2017, US Department of Agriculture, Forest Service, Forest Products Laboratory, Madison, WI. 118. J.L. Paris, F.A. Kamke, and X. Xiao, X-ray computed tomography of wood-adhesive bondlines: Attenuation and phase-contrast effects. Wood Sci. Technol. 49, 1185–1208 (2015). 119. J.L. Paris and F.A. Kamke, Quantitative wood–adhesive penetration with X-ray computed tomography. Intl. J. Adhesion Adhesives 61, 71–80 (2015). 120. T. Walther and H. Thoemen, Synchrotron X-ray microtomography and 3D image analysis of medium density fiberboard (MDF). Holzforschung 63, 581–587 (2009). 121. P.D. Evans, O. Morrison, T.J. Senden, S. Vollmer, R.J. Roberts, A. Limaye, C.H. Arns, H. Averdunk, A. Lowe, and M.A. Knackstedt, Visualization and numerical analysis of adhesive distribution in particleboard using X-ray micro-computed tomography. Intl. J. Adhesion Adhesives 30, 754–762 (2010). 122. A. Bastani, S. Adamopoulos, T. Koddenberg, and H. Militz, Study of adhesive bondlines in modified wood with fluorescence microscopy and X-ray micro-computed tomography. Intl. J. Adhesion Adhesives 68, 351–358 (2016). 123. P. McKinley, F.A. Kamke, A. Sinha, V. De Andrade, and J.E. Jakes, Analysis of adhesive penetration into wood using nano-X-ray computed tomography. Wood Fiber Sci. 50, 66–76 (2018). 124. J.L. Paris, F.A. Kamke, J.A. Nairn, M. Schwarzkopf, and L. Muszynski, Wood-adhesive penetration: Non-destructive, 3D visualization and quantification, in: Proceedings of International Conference on Wood Adhesives, (2013). 125. J.E. Jakes, C.R. Frihart, C.G. Hunt, D.J. Yelle, N.Z. Plaza, L. Lorenz, W. Grigsby, D.J.  Ching, F. Kamke, and S.-C. Gleber, X-ray methods to observe and quantify adhesive penetration into wood. J. Mater. Sci. 54, 705–718 (2018). 126. W. Wang and N. Yan, Characterizing liquid resin penetration in wood using a mercury intrusion porosimeter. Wood Fiber Sci. 37, 505–513 (2005). 127. W. Gindl, SEM and UV-microscopic investigation of glue lines in Parallam PSL. Holz Roh Werkstoff 59, 211–214 (2001).

Understanding Wood Bonds–Going Beyond What Meets the Eye 407

128. H. Saiki, The effect of the penetration of adhesives into cell walls on the failure of wood bonding (in Japanese). J. Japan Wood Res. Soc. 30, 88–92 (1984). 129. L.A. Smith, Resin penetration in wood cell walls- Implications for adhesion of polymers to wood. PhD thesis, SUNY, Syracuse, NY (1971). 130. W.T. Nearn, Wood-adhesive interface relations. Official Digest of the Federated Socioty of Paint Technology 37, 720–733 (1965). 131. L.A. Smith and W.A. Cote, Studies of the penetration of phenol-formaldehyde resin into wood cell walls with the SEM and energy dispersive X-ray analyzer. Wood Fiber Sci. 4, 56–57 (1971). 132. M.P.G. Laborie, L. Salmen, and C.E. Frazier, A morphological study of the wood/phenol-formaldehyde adhesive interphase. J. Adhesion Sci. Technol. 20, 729–741 (2006). 133. A.J. Stamm and R.M. Seborg, Minimizing wood shrinkage and swelling - Treating with synthetic resin-forming materials. Ind. Eng. Chem. 28, 1164–1169 (1936). 134. J.J. Marcinko, S. Devathala, P.L. Rinaldi, and S. Bao, Investigating the molecular and bulk dynamics of PMDI/wood and UF/wood composites. Forest Prod. J. 48, 81–84 (1998). 135. J. Jakes, D. Yelle, J. Beecher, C. Frihart, and D. Stone, Characterizing polymeric methylene diphenyl diisocyanate reactions with wood: 2. Nano-indentation, in: Proceedings of International Conference on Wood Adhesives, pp. 367–373 (2009). 136. C.J. Buckley, C. Phanopoulos, N. Khaleque, A. Engelen, M.E.J. Holwill, and A.G.  Michette, Examination of the penetration of polymeric methylene di-phenyl-di-isocyanate (pMDI) into wood structure using chemical-state x-ray microscopy. Holzforschung 56, 215–222 (2002). 137. C.E. Frazier and J. Ni, On the occurrence of network interpenetration in the wood-isocyanate adhesive interphase. Intl. J. Adhesion Adhesives 18, 81–87 (1998). 138. W. Gindl, E. Dessipri, and R. Wimmer, Using UV-microscopy to study diffusion of melamineurea-formaldehyde resin in cell walls of spruce wood. Holzforschung 56, 103–107 (2002). 139. W. Gindl and H.S. Gupta, Cell-wall hardness and Young’s modulus of melamine-modified spruce wood by nano-indentation. Composites Part A 33, 1141–1145 (2002). 140. A.J. Bolton, J.M. Dinwoodie, and D.A. Davies, The validity of the use of SEM/EDAX as a tool for the detection of UF resin penetration into wood cell walls in particleboards. Wood Sci. Technol. 22, 345–356 (1988). 141. C. Loxton, A. Thumm, W.J. Grigsby, T.A. Adams, and R.M. Ede, Resin distribution in medium density fiberboard. Quantification of UF resin distribution on blowline- and dry-blended MDF fiber and panels. Wood Fiber Sci. 35, 370–380 (2003). 142. C. Frihart, Adhesive groups and how they relate to the durability of bonded wood. J. Adhesion Sci. Technol. 23, 601–617 (2009). 143. H. Tarkow, W.C. Feist, and C.F. Southerland, Penetration versus molecular size. Forest Prod. J. 16, 61–66 (1966). 144. L. Qin, L. Lin, F. Fu, and M. Fan, Microstructure and quantitative micromechanical analysis of wood cell–emulsion polymer isocyanate and urea–formaldehyde interphases. Microscopy Microanalysis 23, 687–695 (2017). 145. J.E. Jakes, C.G. Hunt, D.J. Yelle, L. Lorenz, K. Hirth, S.-C. Gleber, S. Vogt, W. Grigsby, and C.R. Frihart, Synchrotron-based X-ray fluorescence microscopy in conjunction with nanoindentation to study molecular-scale interactions of phenol−formaldehyde in wood cell walls. ACS Appl. Mater. Interfaces 7, 6584–6589 (2015). 146. J.E. Jakes, S.-C. Gleber, S. Vogt, C.G. Hunt, D.J. Yelle, W. Grigsby, and C.R. Frihart, New synchrotron-based technique to map adhesive infiltration in wood cell walls, in: Proceedings of The 2013 Annual Meeting of the Adhesion Society, pp. 1–3 (2013).

408 Progress in Adhesion and Adhesives, Volume 4

147. N.Z. Plaza, Neutron scattering studies of nano-scale wood-water interactions. PhD thesis, University of Wisconsin-Madison, Madison, WI (2017). 148. N.Z. Plaza, S.V. Pingali, S. Qian, W.T. Heller, and J.E. Jakes, Informing the improvement of forest products durability using small angle neutron scattering. Cellulose 23, 1593–1607 (2016). 149. N. Plaza, J. Jakes, C. Frihart, C.G. Hunt, D. Yelle, L. Lorenz, W. Heller, S.V. Pingali, and D. Stone, Small angle neutron scattering as a new tool to evaluate moisture-induced swelling in the nanostructure of chemically modified wood cell walls. Forest Prod. J. (In Press). 150. W. Gindl, F. Zargar-Yaghubi, and R. Wimmer, Impregnation of softwood cell walls with melamine-formaldehyde resin. Bioresource Technol. 87, 325–330 (2003). 151. Y. Huang, B. Fei, and R. Zhao, Investigation of low-molecular weight phenol formaldehyde distribution in tracheid cell walls of Chinese fir wood. Bioresources 9, 4150–4158 (2014). 152. J. Konnerth, D. Harper, S.H. Lee, T.G. Rials, and W. Gindl, Adhesive penetration of wood cell walls investigated by scanning thermal microscopy (SThM). Holzforschung 62, 91–98 (2008). 153. A.O. Rapp, H. Bestgen, W. Adam, and R.D. Peck, Electron energy loss spectroscopy (EELS) for quantification of cell-wall penetration of a melamine resin. Holzforschung 53, 111–117 (1999). 154. C. Xing, B. Riedl, A. Cloutier, and S.M. Shaler, Characterization of urea-formaldehyde resin penetration into medium density fiberboard fibers. Wood Sci. Technol. 39, 374–384 (2005). 155. C. Xing, B. Riedl, and A. Cloutier, Measurement of urea-formaldehyde resin distribution as a function of MDF fiber size by laser scanning microscopy. Wood Sci. Technol. 37, 495–507 (2004). 156. H. Pakdel, P.L. Cyr, B. Riedl, and J. Deng, Quantification of urea formaldehyde resin in wood fibers using X-ray photoelectron spectroscopy and confocal laser scanning microscopy. Wood Sci. Technol. 42, 133–148 (2008). 157. D.J. Yelle and J. Ralph, Characterizing phenol–formaldehyde adhesive cure chemistry within the wood cell wall. Intl. J. Adhesion Adhesives 70, 26–36 (2016). 158. D.J. Yelle, J. Ralph, and C.R. Frihart, Delineating pMDI model reactions with loblolly pine via solution-state NMR spectroscopy. Part 2. Non-catalyzed reactions with the wood cell wall. Holzforschung 65, 145–154 (2011). 159. R. Rowell, Chemical modification of wood, in Handbook of Wood Chemistry and Wood Composites, R. Rowell (Ed) pp.381–420, Taylor & Francis, Boca Raton, FL (2005). 160. C.A.S. Hill, Wood Modification: Chemical, Thermal, and Other Processes, John Wiley and Sons, New York (2006). 161. E.E. Thybring, The decay resistance of modified wood influenced by moisture exclusion and swelling reduction. Intl. Biodeterioration Biodegradation 82, 87–95 (2013). 162. R. Wimmer and B.N. Lucas, Comparing mechanical properties of secondary wall and cell corner middle lamella in spruce wood. IAWA J. 18, 77–88 (1997). 163. J. Konnerth and W. Gindl, Mechanical characterisation of wood-adhesive interphase cell walls by nanoindentation. Holzforschung 60, 429–433 (2006). 164. W. Gindl, T. Schoberl, and G. Jeronimidis, The interphase in phenol-formaldehyde and polymeric methylene diphenyl-di-isocyanate glue lines in wood. Intl. J. Adhesion Adhesives 24, 279–286 (2004). 165. W. Gindl, T. Schoberl, and G. Jeronimidis, Corrigendum to “The interphase in phenolformaldehyde (PF) and polymeric methylene di-phenyl-di-isocyanate (pMDI) glue lines in wood”. Int. J. Adhesion Adhesives 24, 535–535 (2004). 166. J.E. Jakes, C.R. Frihart, J.F. Beecher, R.J. Moon, and D.S. Stone, Experimental method to account for structural compliance in nanoindentation measurements. J. Mater. Res. 23, 1113–1127 (2008).

Understanding Wood Bonds–Going Beyond What Meets the Eye 409

167. J.E. Jakes, C.R. Frihart, J.F. Beecher, R.J. Moon, P.J. Resto, Z.H. Melgarejo, O.M. Suárez, H. Baumgart, A.A. Elmustafa, and D.S. Stone, Nanoindentation near the edge. J. Mater. Res. 24, 1016–1031 (2009). 168. J.E. Jakes, R.S. Lakes, and D.S. Stone, Broadband nanoindentation of glassy polymers: Part II. Viscoplasticity. J. Mater. Res. 27, 475–484 (2012). 169. J.E. Jakes, R.S. Lakes, and D.S. Stone, Broadband nanoindentation of glassy polymers: Part I. Viscoelasticity. J. Mater. Res. 27, 463–474 (2012). 170. X. Wang, L. Zhao, Y. Deng, Y. Li, and S. Wang, Effect of the penetration of isocyanates (pMDI) on the nanomechanics of wood cell wall evaluated by AFM-IR and nanoindentation (NI). Holzforschung 72, 301–309 (2018). 171. C.E. Frazier, Isocyanate wood binders, in Handbook of Adhesive Technology, second edition, A. Pizzi and K.L. Mittal (Eds), pp.681–694, CRC Press, Boca Raton, FL (2003). 172. C.G. Hunt, J. O’Dell, J.E. Jakes, W.J. Grigsby, and C.R. Frihart, Wood as polar size exclusion chromatography media: Implications to adhesive performance. Forest Prod. J. 65, 9–14 (2015). 173. J. Konnerth, A. Jaeger, J. Eberhardsteiner, U. Mueller, and W. Gindl, Elastic properties of adhesive polymers. II. Polymer films and bond lines by means of nanoindentation. J. Appl. Polym. Sci. 102, 1234–1239 (2006). 174. G.I. Mantanis, E.T. Athanassiadou, M.C. Barbu, and K. Wijnendaele, Adhesive systems used in the European particleboard, MDF and OSB industries. Wood Mater. Sci. Eng. 13, 104–116 (2018). 175. G. Stapf, N. Zisi, and S. Aicher, Curing behaviour of structural wood adhesives - parallel plate rheometer results, in: Proceedings of International Conference on Wood Sci. and Engineering, pp. 109–117 (2013). 176. M. Kariz, M. Kitek Kuzman, and M. Sernek, The effect of the heat treatment of spruce wood on the curing of melamine–urea–formaldehyde and polyurethane adhesives. J. Adhesion Sci. Technol. 27, 1911–1920 (2013). 177. M. Witt, Novel plate rheometer configuration allows monitoring real-time wood adhesive curing behavior. J. Adhesion Sci. Technol. 18, 893–904 (2004). 178. I. Gavrilovic´-Grmuša, M. Dunky, J. Miljkovic´ and M. Điporovic´-Momcˇilovic´, Influence of the viscosity of UF resins on the radial and tangential penetration into poplar wood and on the shear strength of adhesive joints. Holzforschung 66, 849–856 (2012). 179. I. Gavrilovic´-Grmuša, M. Dunky, J. Miljkovic´ and M. Djiporovic´-Momcˇilovic´, Radial penetration of urea-formaldehyde adhesive resins into beech (Fagus Moesiaca). J.  Adhesion Sci. Technol. 24, 1753–1768 (2010). 180. I. Gavrilovic´-Grmuša, J. Miljkovic´ and M. Ðiporovic´-Momcˇilovic´, Influence of the degree of condensation on the radial penetration of urea-formaldehyde adhesives into silver fir (Abies alba, Mill.) wood tissue. J. Adhesion Sci. Technol. 24, 1437–1453 (2010). 181. I. Gavrilovic´-Grmuša, M. Dunky, J. Miljkovic´ and M. Djiporovic´-Momcˇilovic´, Influence of the degree of condensation of urea-formaldehyde adhesives on the tangential penetration into beech and fir and on the shear strength of the adhesive joints. European J. Wood Wood Products 70, 655–665 (2012). 182. J.F. Siau, Transport Processes in Wood, Springer Verlag, Berlin (1984). 183. J. Wang, M.P.G. Laborie, and M.P. Wolcott, Kinetic analysis of phenol-formaldehyde bonded wood joints with dynamical mechanical analysis. Thermochimica Acta 491, 58–62 (2009). 184. P.E. Humphrey, Thermal and chemical injection effects on PF, PMDI and UF adhesion kinetics, in: Proceedings of International Conference on Wood Adhesives 2009, pp. 213–223 (2010).

410 Progress in Adhesion and Adhesives, Volume 4

185. Standard test method for measuring the effect of temperature on the cohesive strength development of adhesives using lap shear bonds under tensile loading, ASTM D7998–15 (2015). 186. D.A. Dillard, Fundamentals of stress transfer in bonded systems, in Adhesion Science and Engineering, Volume 1: The Mechanics of Adhesion, D.A. Dillard and A.V. Pocius (Eds), pp. 1–44, Elsevier, Amsterdam, The Netherlands (2002). 187. B.H. River, Fracture of adhesively-bonded wood joints, in Handbook of Adhesive Technology, second edition, A. Pizzi and K.L. Mittal (Eds), pp. 325–350, CRC Press, Boca Raton, FL (2003). 188. American national standard for engineered wood flooring, ANSI/HPVA EF 2012 (2012). 189. Standard test method for multiple-cycle accelerated aging test (automatic boil test) for exterior wet use wood adhesives, ASTM D3434-00 (2013). 190. Adhesives for load-bearing timber structures - Test methods - Part 1: Determination of longitudinal tensile shear strength, CEN EN 302-1 (2013). 191. Standard test method for internal bond strength and thickness swell of cellulosic-based fiber and particle panels after repeated wetting, ASTM D7519-11 (2011). 192. Plywood bonding quality. Part 2: Requirements, CEN EN 314-2 (1997). 193. Standard practices for resistance of adhesives to cyclic laboratory aging conditions (metal bonding committee), ASTM D1183-03 (2011). 194. Standard test method for strength properties of adhesives in plywood type construction in shear by tension loading, ASTM D906-98 (2017). 195. C.R. Frihart and J.M. Wescott, Why do some wood-adhesive bonds respond poorly to accelerated moisture-resistant tests, in: Proceedings of The 9th Pacific Rim Bio-Based Composites Symposium. , pp. 51–58 (2008). 196. F. Stoeckel, J. Konnerth, and W. Gindl-Altmutter, Mechanical properties of adhesives for bonding wood--A review. Intl. J. Adhesion Adhesives 45, 32–41 (2013). 197. F. Stockel, J. Konnerth, W. Kantner, J. Moser, and W. Gindl, Tensile shear strength of UF- and MUF-bonded veneer related to data of adhesives and cell walls measured by nanoindentation. Holzforschung 64, 337–342 (2010). 198. W.T. Nearn, Application of the ultrastructure concept in industrial wood products research. Wood Sci. 6, 285–293 (1974). 199. M. Knorz, P. Niemz, and J.-W. van de Kuilen, Measurement of moisture-related strain in bonded ash depending on adhesive type and glueline thickness. Holzforschung 70, 145–155 (2016). 200. Standard test method for creep and time to failure of adhesives in static shear by compression loading (wood-to-wood), ASTM D4680-98 (2017) 201. Standard test method for resistance to creep of adhesives in static shear by compression loading (wood-to-wood), ASTM D7966/D7966M - 16 (2016) 202. Adhesives - One component polyurethane (PUR) for load bearing timber structures. Classification and performance requirements, CEN EN 15425 (2008) 203. Timber structures - Glued laminated timber and glued solid timber - Requirements, CEN EN 14080 (2013) 204. D. Lukowsky, Failure Analysis of Wood and Wood-Based Products, McGraw Hill Educational, New York (2015). 205. M. Sernek, J. Resnik, and F.A. Kamke, Penetration of liquid urea-formaldehyde adhesive into beech wood. Wood Fiber Sci. 31, 41–48 (1999). 206. M. Riegler, W. Gindl-Altmutter, M. Hauptmann, and U. Müller, Detection of UF resin on wood particles and in particleboards: Potential of selected methods for practice-oriented offline detection. European J. Wood Wood Products 70, 829–837 (2012).

Understanding Wood Bonds–Going Beyond What Meets the Eye 411

207. G. Modzel, F. Kamke, and F. De Carlo, Comparative analysis of a wood: adhesive bondline. Wood Sci. Technol. 45, 147–158 (2011). 208. H. Matsunaga, M. Kiguchi, and P.D. Evans, Microdistribution of copper-carbonate and iron oxide nanoparticles in treated wood. J. Nanoparticle Res. 11, 1087–1098 (2009). 209. F.A. Kamke, P.E. McKinley, D.J. Ching, M. Zauner, and X. Xiao, Micro X-ray computed tomography of adhesive bonds in wood. Wood Fiber Sci. 48, 2–16 (2016). 210. J.L. Paris, F.A. Kamke, R. Mbachu, and S.K. Gibson, Phenol formaldehyde adhesives formulated for advanced X-ray imaging in wood-composite bondlines. J. Mater. Sci. 49, 580–591 (2014). 211. F.A. Kamke, J.A. Nairn, L. Muszynski, J.L. Paris, M. Schwarzkopf, and X. Xiao, Methodology for micromechanical analysis of wood adhesive bonds using X-ray computed tomography and numerical modeling. Wood Fiber Sci. 46, 15–28 (2014). 212. M.S. White, G. Ifju, and J.A. Johnson, Method for measuring resin penetration into wood. Forest Prod. J. 27, 52–54 (1977). 213. W. Grigsby, A. Thumm, and P. Burrell, Towards an understanding of fiber-adhesive interactions in MDF manufacture, in: Proceedings of International Conference on Wood Adhesives, pp. 303–310 (2005). 214. W. Grigsby, K. Murton, and A. Thumm, Effect of process conditions on resin mobility during medium density fiberboard (MDF) production, in: Proceedings of International Wood Composites Symposium, pp. 5–8 (2004). 215. W. Grigsby and A. Thumm, The interactions between wax and UF resin in medium density fibreboard. European J. Wood Wood Products 70, 507–517 (2012). 216. W.J. Grigsby and A. Thumm, Resin and wax distribution and mobility during medium density fibreboard manufacture. European J. Wood Wood Products 70, 337–348 (2012). 217. C.G. Hunt, J.E. Jakes, and W. Grigsby, Evaluation of adhesive penetration of wood fibre by nanoindentation and microscopy, in: Proceedings of Pacific Rim Bio-Based Composites Conference, pp. 216–226 (2010). 218. W. Grigsby and A. Thumm, Fundamental fiber-adhesive interactions in MDF manufacture, in: Proceedings of International Conference on Wood Adhesives 2009, (2010). 219. W. Grigsby, Personal Communication (2010). 220. W. Grigsby and A. Thumm, Visualization of biomaterials on wood fiber, in: Proceedings of International Conference on Wood Adhesives 2009, (2010). 221. L. Donaldson and J. Bond, Fluorescence Microscopy of Wood (CD-ROM), SCION, Rotorua, NZ (2005). 222. S. Ruzin, Plant Microtechnique and Microscopy, Oxford University Press, New York (1999). 223. N. Kutscha, A Compilation of Micrographs on Wood and Wood Products, Publication Number 7225. 2007, Forest Products Society, Madison, WI. 224. E. Ives, A Guide to Wood Microtomy - Making Quality Microslides of Wood Sections, Self Published, Sproughton, England (2001). 225. G.P. Berlyn and J.P. Miksche, Botanical Microtechnique and Cytochemistry, Iowa State University Press, Ames, Iowa (1976). 226. A. Nuryawan, B.-D. Park, and A.P. Singh, Penetration of urea–formaldehyde resins with different formaldehyde/urea mole ratios into softwood tissues. Wood Sci. Technol. 48, 889–902 (2014). 227. H. Wan and M.G. Kim, Distribution of phenol-formaldehyde resin in impregnated southern pine and effects on stabilization. Wood Fiber Sci. 40, 181–189 (2008). 228. E. Mahrdt, H.W.G. van Herwijnen, W. Kantner, J. Moser, J. Giesswein, R. Mitter, U. Müller, and W. Gindl-Altmutter, Adhesive distribution related to mechanical performance of high density wood fibre board. Intl. J. Adhesion Adhesives 78, 23–27 (2017).

412 Progress in Adhesion and Adhesives, Volume 4

229. D.E. Brady and F.A. Kamke, Effects of hot-pressing parameters on resin penetration. Forest Prod. J. 38, 63–68 (1988). 230. L. Murmanis, B.H. River, and H. Stewart, Microscopy of abrasive-planed and knife-planed surfaces in wood-adhesive bonds. Wood Fiber Sci. 15, 102–115 (1983). 231. L. Murmanis, G.C. Myers, and J.A. Youngquist, Fluorescence microscopy of hardboards. Wood Fiber Sci. 18, 212–219 (1986). 232. J. Youngquist, G.C. Myers, and L. Murmanis, Resin distribution in hardboard: Evaluated by internal bond strength and fluorescence microscopy. Wood Fiber Sci. 19, 215–224 (1987). 233. W. Ginzel and G. Stegmann, Subsequent colouring of urea-formaldehyde resins on glued wood particles for visual estimation of glue distribution. Holz Roh Werkstoff 28, 289–292 (1970). 234. T.M. Gruver and N.R. Brown, Penetration and performance of isocyanate wood binders on selected wood species. Bioresources 1, 233–247 (2007). 235. T. Furuno and T. Goto, Structure of the interface between wood and synthetic polymer (III): The penetration of MMA monomer into woody cell wall. Mokuzai Gakkaishi 19, 271–274 (1973). 236. A.P. Singh, C.R. Anderson, J.M. Warnes, and J. Matsumura, The effect of planing on the microscopic structure of Pinus radiata wood cells in relation to penetration of PVA glue. Holz Roh Werkstoff 60, 333–341 (2002). 237. J. Zheng, S.C. Fox, and C.E. Frazier, Rheological, wood penetration, and fracture performance studies of PF/pMDI hybrid resins. Forest Prod. J. 54, 74–81 (2004). 238. L. Qin, L. Lin, and F. Fu, Microstructural and micromechanical characterization of modified urea-formaldehyde resin penetration into wood. Bioresources 11, 182–194 (2016). 239. N. Kutscha, Factors affecting the bond quality of hem-fir finger-joints. Forest Prod. J 37, 43–48 (1987). 240. J.E. Marian and K. Suchsland, Experimental investigation of gluing and finishing problems through application of fluorescence microscopy and incident light. Forest Prod. J 7, 74–77 (1957). 241. M. Knorz, M. Schmidt, S. Torno, and J.-W. van de Kuilen, Structural bonding of ash (Fraxinus excelsior L.): Resistance to delamination and performance in shearing tests. European J. Wood Wood Products 72, 297–309 (2014). 242. M. Guan, C. Yong, and L. Wang, Microscopic characterization of modified phenol-formaldehyde resin penetration of bamboo surfaces and its effect on some properties of two-ply bamboo bonding interface. Bioresources 9, 1953–1963 (2014). 243. O. Kläusler, W. Bergmeier, A. Karbach, W. Meckel, E. Mayer, S. Clauß, and P. Niemz, Influence of N, N-dimethylformamide on one-component moisture-curing polyurethane wood adhesives. Intl. J. Adhesion Adhesives 55, 69–76 (2014). 244. A. Bastani, S. Adamopoulos, and H. Militz, Gross adhesive penetration in furfurylated, N-methylol melamine-modified and heat-treated wood examined by fluorescence microscopy. European J. Wood Wood Products 73, 635–642 (2015). 245. M. Schmidt, P. Glos, and G. Wegener, Gluing of European beech wood for load bearing timber structures (in German). European J. Wood Wood Products 68, 43–57 (2010). 246. M. Knorz, E. Neuhaeuser, S. Torno, and J.-W. van de Kuilen, Influence of surface preparation methods on moisture-related performance of structural hardwood–adhesive bonds. Intl. J. Adhesion Adhesives 57, 40–48 (2015). 247. J.-C. Cerre, Macrophotographs of cross sections of woods: Part One, https://www.youtube.com/ watch?v=CuFCfzm5BdU (2012).

Understanding Wood Bonds–Going Beyond What Meets the Eye 413

248. L. Gollob, R.L. Krahmer, J.D. Wellons, and A.W. Christiansen, Relationship between chemical characteristics of phenol-formaldehyde resins and adhesive performance. Forest Prod. J. 35, 42–48 (1985). 249. S. Ellis, Effect of resin particle size on waferboard adhesive efficiency. Wood Fiber Sci. 25, 214–219 (1993). 250. H. Edalat, M. Faezipour, V. Thole, and F.A. Kamke, A new quantitative method for evaluation of adhesive penetration pattern in particulate wood-based composite. Wood Sci. Technol 48, 703–712 (2014). 251. J. Luedtke, C. Amen, A. van Ofen, and C. Lehringer, 1C-PUR-bonded hardwoods for engineered wood products: Influence of selected processing parameters. European J. Wood Wood Products 73, 167–178 (2015). 252. C.B. Vick and T.A. Kuster, Mechanical interlocking of adhesive bonds to CCA-treated southern pine -A scanning electron-microscopic study. Wood Fiber Sci. 24, 36–46 (1992). 253. T. Furuno, Y. Imamura, and H. Kajita, The modification of wood by treatment with low molecular weight phenol-formaldehyde resin: A properties enhancement with neutralized phenolicresin and resin penetration into wood cell walls. Wood Sci. Technol. 37, 349–361 (2004). 254. B. Collett, Scanning electron microscopy: A review and report of research in wood science. Wood Fiber Sci. 3, 113–133 (1970). 255. L.P. Futo, Direct electron microscopic presentation of glue lines and surface coatings of wood base materials (in German). Holz Roh Werkstoff 31, 52–61 (1973). 256. P. Niemz, D. Mannes, E. Lehmann, P. Vontobel, and S. Haase, Investigations of the distribution of adhesive in the area of the bond line using neutron radiography and microscopy (in German). Holz Roh. Werkstoff 62, 424–432 (2004). 257. S. Clauß, J. Gabriel, A. Karbach, M. Matner, and P. Niemz, Influence of the adhesive formulation on the mechanical properties and bonding performance of polyurethane prepolymers. Holzforschung 65, 835–844 (2011). 258. Z. Koran and R.C. Vasishth, Scanning electron microscopy of plywood glue lines 1. Wood Fiber Sci. 3, 202–209 (1972). 259. A.P. Singh, A. Nuryawan, B.-D. Park, and K.H. Lee, Urea-formaldehyde resin penetration into Pinus radiata tracheid walls assessed by TEM-EDXS. Holzforschung 69, 303–306 (2015). 260. O. Suchsland, Über das Eindringen des Leimes bei der Holzverleimung und die Bedeutung der Eindringtiefe für die Fugenfestigkeit (in German). Holz Roh Werkstoff 16, 101–108 (1958). 261. A.P. Singh, E.A. Dunningham, and D.V. Plackett, Assessing the performance of a commercial wood stain by transmission electron microscopy. Holzforschung 49, 255–258 (1995). 262. A.P. Singh and B.S. Dawson, The mechanism of failure of clear coated wooden boards as revealed by microscopy. IAWA J 24, 1–11 (2003). 263. L. Donaldson and T. Lomax, Adhesive/fibre interaction in medium density fibreboard. Wood Sci. Technol. 23, 371–380 (1989). 264. R.M. Nussbaum, E.J. Sutcliffe, and A.-C. Hellgren, Microautoradiographic studies of the penetration of alkyd, alkyd emulsion and linseed oil coatings into wood. J. Coatings Technol. 70, 878, 49–57 (1998). 265. S. Lee, T.F. Shupe, L.H. Groom, and C.Y. Hse, Wetting behaviors of phenol-and urea-formaldehyde resins as compatibilizers. Wood Fiber Sci. 39, 482–492 (2007). 266. N. Gierlinger, C. Hansmann, T. Röder, H. Sixta, W. Gindl, and R. Wimmer, Comparison of UV and confocal Raman microscopy to measure the melamine–formaldehyde resin content within cell walls of impregnated spruce wood. Holzforschung 59, 210–213 (2005).

414 Progress in Adhesion and Adhesives, Volume 4

267. F.A. Kamke, C.A. Lenth, and H.G. Saunders, Measurement of resin and wax distribution on wood flakes. Forest Prod. J. 46, 63–68 (1996). 268. W.J. Grigsby, A. Thumm, and F.A. Kamke, Determination of resin distribution and coverage in MDF by fiber staining. Wood Fiber Sci. 37, 258–269 (2005). 269. F. Van De Velde, F. Weinbreck, M.W. Edelman, E. Van Der Linden, and R.H. Tromp, Visualisation of biopolymer mixtures using confocal scanning laser microscopy (CSLM) and covalent labelling techniques. Colloids Surfaces B 31, 159–168 (2003). 270. P.-L. Cyr, B. Riedl, and X.-M. Wang, Investigation of urea-melamine-formaldehyde (UMF) resin penetration in medium-density fiberboard (MDF) by high resolution confocal laser scanning microscopy. Holz Roh Werkstoff 66, 129–134 (2008). 271. F. Stöckel, J. Konnerth, J. Moser, W. Kantner, and W. Gindl-Altmutter, Micromechanical properties of the interphase in pMDI and UF bond lines. Wood Sci. Technol. 46, 611–620 (2012). 272. P. Hass, F.K. Wittel, M. Mendoza, H.J. Herrmann, and P. Niemz, Adhesive penetration in beech wood: Experiments. Wood Sci. Technol. 46, 243–256 (2012). 273. D. Mannes, F. Marone, E. Lehmann, M. Stampanoni, and P. Niemz, Application areas of synchrotron radiation tomographic microscopy for wood research. Wood Sci. Technol. 44, 67–84 (2010). 274. D. Mannes, P. Niemz, and E. Lehmann, Tomographic investigations of wood from macroscopic to microscopic scale. Wood Res. 54, 33–44 (2009). 275. S.J. Sanabria, P. Wyss, J. Neuenschwander, P. Niemz, and U. Sennhauser, Assessment of glued timber integrity by limited-angle microfocus X-ray computed tomography. European J. Wood Wood Products 69, 605–617 (2011). 276. J. Sobczyk, P. Fra˛czek, M. Obarzanowski, J.M. del Hoyo-Meléndez and Ł. Bratasz, Digital radiography (DR) and imaging analysis for evaluating the penetration and distribution of organic substances used in wood conservation. Wood Sci. Technol. 48, 981–994 (2014). 277. S.M. Shaler, H. Wang, E. Landis, D.T. Keane, L. Mott, and L. Holzman, Microtomography of cellulosic structures, in: Proceedings of Process and Product Quality Conference and Trade Fair (1998). 278. D. Xu, Y. Zhang, H. Zhou, Y. Meng, and S. Wang, Characterization of adhesive penetration in wood bond by means of scanning thermal microscopy (SThM). Holzforschung 70, 323–330 (2016). 279. X. Wang, Y. Deng, Y. Li, K. Kjoller, A. Roy, and S. Wang, In situ identification of the molecularscale interactions of phenol-formaldehyde resin and wood cell walls using infrared nanospectroscopy. RSC Advances 6, 76318–76324 (2016). 280. C. Marcott, M. Lo, K. Kjoller, C. Prater, R. Shetty, J.E. Jakes, and I. Noda, Nanoscale infrared spectroscopy of biopolymeric materials, in: Proceedings of Society for the Advancement of Material and Process Engineering Technical Conference, p. 9 (2012). 281. Z.-M. Liu, F.-H. Wang, and X.-M. Wang, Spectroscopic analysis of the interface for wheat straw specimen glued with PMDI. Wood Fiber Sci. 39, 109–119 (2007). 282. J. Konnerth, N. Gierlinger, J. Keckes, and W. Gindl, Actual versus apparent within cell wall variability of nanoindentation results from wood cell walls related to cellulose microfibril angle. J. Mater. Sci. 44, 4399–4406 (2009). 283. K. Liang, G.B. Du, O. Hosseinaei, S.Q. Wang, and H. Wang, Mechanical properties of secondary wall and compound corner middle lamella near the phenol-formaldehyde (PF) adhesive bond line measured by nanoindentation. Advanced Materials Research 236–238, 1746–1751 (2011). 284. J. Konnerth, A. Valla, and W. Gindl, Nanoindentation mapping of a wood-adhesive bond. Appl. Phys. A 88, 371–375 (2007).

Understanding Wood Bonds–Going Beyond What Meets the Eye 415

285. E.G. Herbert, P.S. Phani, and K.E. Johanns, Nanoindentation of viscoelastic solids: A critical assessment of experimental methods. Current Opinion Solid State Mater. Sci. 19, 334–339 (2015). 286. X. Wang, Y. Li, S. Wang, W. Yu, and Y. Deng, Temperature-dependent mechanical properties of wood-adhesive bondline evaluated by nanoindentation. J. Adhesion 93, 640–656 (2017). 287. J. Follrich, F. Stöckel, and J. Konnerth, Macro-and micromechanical characterization of woodadhesive bonds exposed to alternating climate conditions. Holzforschung 64, 705–711 (2010). 288. W. Grigsby, A.G. McDonald, A. Thumm, and C. Loxton, X-ray photoelectron spectroscopy determination of urea formaldehyde resin coverage on MDF fibre. Holz Roh Werkstoff 62, 358–364 (2004). 289. W. Grigsby and A. Thumm, Visualisation of UF resin on MDF fibre by XPS imaging. Holz Roh Werkstoff 62, 365–369 (2004). 290. J.F. Beecher and C.R. Frihart, X-ray photoelectron spectroscopy for characterization of wood surfaces in adhesion studies, in: Proceedings of Internationl Conference on Wood adhesives, pp. 83–89 (2006). 291. M.P.G. Laborie and C.E. Frazier, 13C CP/MAS NMR study of a wood/phenol-formaldehyde resin bondline. J. Mater. Sci. 41, 6001–6005 (2006). 292. C. Lapierre, B. Pollet, and C. Rolando, New insights into the molecular architecture of hardwood lignins by chemical degradative methods. Res. Chemical Intermediates 21, 397–412 (1995). 293. S. Bao, W.A. Daunch, Y. Sun, and P.L. Rinaldi, Solid state two-dimensional NMR studies of polymeric diphenylmethane diisocyanate (PMDI) reaction in wood. Forest Prod. J. 53, 63–71 (2003). 294. S. Das, M.J. Malmberg, and C.E. Frazier, Cure chemistry of wood/polymeric isocyanate (PMDI) bonds: Effect of wood species. Intl. J. Adhesion Adhesives 27, 250–257 (2007). 295. S.L. Wendler, and C.E. Frazier, Effect of moisture content on the isocyanate/wood adhesive bondline by 15N CP/MAS NMR. J. Appl. Polym. Sci. 61, 775–782 (1996). 296. S.L. Wendler, and C.E. Frazier, The effects of cure temperature and time on the isocyanate-wood adhesive bondline by 15N CP/MAS NMR. Intl. J. Adhesion Adhesives 16, 179–186 (1996). 297. M. Nishida, T. Tanaka, T. Miki, Y. Hayakawa, and K. Kanayama, Integrated analysis of solidstate NMR spectra and nuclear magnetic relaxation times for the phenol formaldehyde (PF) resin impregnation process into soft wood. RSC Advances 7, 54532–54541 (2017). 298. C. Simon, B. George, and A. Pizzi, UF/pMDI wood adhesives: Networks blend versus copolymerization. Holzforschung 56, 327–334 (2002). 299. A.J. Allen, J. Marcinko, T. Wagler, and A.J. Sosnowick, Investigations of the molecular interactions of soy-based adhesives. Forest Prod J. 60, 534–540 (2011). 300. J.J. Marcinko, A.A. Parker, P.L. Rinaldi, and W.M. Ritchey, Application of NMR cross-polarization/modulus correlations to a series of polyurethane elastomers. J. Appl. Polym. Sci. 51, 1777–1780 (1994). 301. M.P.G. Laborie, L. Salmen, and C.E. Frazier, Cooperativity analysis of the in situ lignin glass transition. Holzforschung 58, 129–133 (2004). 302. D. Ren and C.E. Frazier, Wood/adhesive interactions and the phase morphology of moisturecure polyurethane wood adhesives. Intl. J. Adhesion Adhesives 34, 55–61 (2012). 303. D.A. Riedlinger, N.J. Sun, and C.E. Frazier, T-g as an index of conversion in PMDI-impregnated wood. Bioresources 2, 605–615 (2007). 304. N.J. Sun and C.E. Frazier, Probing the hydroxymethylated resorcinol coupling mechanism with stress relaxation. Wood Fiber Sci. 37, 673–681 (2005). 305. G. He and N. Yan, Effect of moisture content on curing kinetics of pMDI resin and wood mixtures. Intl. J. Adhesion Adhesives 25, 450–455 (2005).

416 Progress in Adhesion and Adhesives, Volume 4

306. T. Zimmermann, V. Thommen, P. Reimann, and H.J. Hug, Ultrastructural appearance of embedded and polished wood cell walls as revealed by atomic force microscopy. J. Structural Biology 156, 363–369 (2006). 307. N. Kutscha and J.R. Gray, The suitability of certain stains for studying lignification in balsam fir, Abies balsamea (L.) Mill, TB53. 1972, University of Maine Life Sciences and Agriculture Experiment Station, Orono, ME. 308. A. Wiedenhoeft, Basic wood biology—Anatomy for identification, in Identification of Central American Woods, Publication# 7215-11, pp.11–20, Forest Products Society, Madison, WI (2011). 309. A. Wiedenhoeft, Correct use of a hand lens, in Identification of Central American Woods, Publication# 7215-11, pp.21–22, Forest Products Society, Madison, WI. (2011). 310. A. Wiedenhoeft, Bloodless wood specimen preparation for hand lens observation, in Identification of Central American Woods, Publication# 7215-11, pp.23–30, Forest Products Society, Madison, WI (2011). 311. R.B. Hoadley, Identifying Wood: Accurate Results with Simple Tools, Taunton Press, Newtown, CT (1990). 312. W. Schoch, I. Heller-Kellenberger, F. Schweingruber, F. Kienast, and D. Schmatz, Wood anatomy of central European species, www.woodanatomy.ch/preparation.html (2004). 313. G. von Arx, A. Crivellaro, A.L. Prendin, K. Cˇufar, and M. Carrer, Quantitative wood anatomy—practical guidelines. Frontiers Plant Sci. 7, 781–793 (2016). 314. J.C. Tardif, and F. Conciatori, Microscopic examination of wood: Sample preparation and techniques for light microscopy, in Plant Microtechniques and Protocols, E.C.T. Yeung, C. Stasolla, M.J. Sumner, and B.Q. Huang (Eds), pp.373–415, Springer, Cham, Switzerland (2015). 315. M. Risholm-Sundman, A. Larsen, E. Vestin, and A. Weibull, Formaldehyde emission—comparison of different standard methods. Atmospheric Environment 41, 3193–3202 (2007). 316. Plywood - bonding quality - Part 1: Test methods, CEN EN 314-1 (2005) 317. D. Caster, Correlation between exterior exposure and automatic boil test results, in: Proceedings of Wood Adhesives - Research, Application and Needs Symposium, pp. 199–208 (1980). 318. C.R. Frihart, Are epoxy-wood bonds durable enough?, in: Proceedings of International Conference on Wood Adhesives pp. 241–246 (2005). 319. W. Olson and R.F. Blomquist, Epoxy-resin adhesives for gluing wood. Forest Prod. J 12, 74–80 (1962). 320. P. Lavisci, S. Berti, B. Pizzo, P. Triboulot, and R. Zanuttini, A shear test for structural adhesives used in the consolidation of old timber. European J. Wood Wood Products 59, 145–152 (2001). 321. C.B. Vick, K. Richter, B.H. River, and A.R. Fried, Hydroxymethylated resorcinol coupling agent for enhanced durability of bisphenol-A epoxy bonds to Sitka spruce. Wood Fiber Sci. 27, 2–12 (1995). 322. R.O. Ebewele, B.H. River, and G.E. Myers, Behavior of amine-modified urea–formaldehydebonded wood joints at low formaldehyde/urea molar ratios. J. Appl. Polym. Sci. 52, 689–700 (1994). 323. R.O. Ebewele, B.H. River, and G.E. Myers, Polyamine-modified urea-formaldehyde-bonded wood joints. 3. Fracture-toughness and cyclic stress and hydrolysis resistance. J. Appl. Polym. Sci. 49, 229–245 (1993). 324. R.O. Ebewele, B.H. River, G.E. Myers, and J.A. Koutsky, Polyamine-modified urea-formaldehyde resins 2. Resistance to stress-induced by moisture cycling of solid wood joints and particleboard. J. Appl. Polym. Sci. 43, 1483–1490 (1991).

Understanding Wood Bonds–Going Beyond What Meets the Eye 417

325. R.O. Ebewele, G.E. Myers, B.H. River, and J.A. Koutsky, Polyamine-modified urea-formaldehyde resins. 1. Synthesis, structure, and properties. J. Appl. Polym. Sci. 42, 2997–3012 (1991). 326. B.H. River, R.O. Ebewele, and G.E. Myers, Failure mechanisms in wood joints bonded with urea-formaldehyde adhesives. Holz Roh Werkstoff 52, 179–184 (1994). 327. Standard test methods for evaluating properties of wood-base fiber and particle panel materials, ASTM D1037-12 (2012). 328. Standard test method for evaluating the shear strength of adhesive bonds in laminated wood products at elevated temperatures, ASTM D7247-17 (2017). 329. Adhesives - Wood adhesives for non-structural applications - Determination of tensile shear strength of lap joints, CEN EN 205 (2016). 330. Glued laminated timber - Delamination test of glue lines (standard withdrawn), CEN EN 391 (2002). 331. Classification of thermosetting wood adhesives for non-structural applications, CEN EN 12765 (formerly EN 204) (2016). 332. Standard test method for mode I interlaminar fracture toughness of unidirectional fiber-reinforced polymer matrix composites, ASTM D5528-13 (2013). 333. J. Whitney, C. Browning, and W. Hoogsteden, A double cantilever beam test for characterizing mode I delamination of composite materials. J. Reinforced Plastics Composites 1, 297–313 (1982). 334. J.A. Nairn, Energy release rate analysis for adhesive and laminate double cantilever beam specimens emphasizing the effect of residual stresses. Intl. J. Adhesion Adhesives 20, 59–70 (2000). 335. J.M. Gagliano, and C.E. Frazier, Improvements in the fracture cleavage testing of adhesivelybonded wood. Wood Fiber Sci. 33, 377–385 (2001). 336. S. Veigel, J. Follrich, W. Gindl-Altmutter, and U. Müller, Comparison of fracture energy testing by means of double cantilever beam-(DCB)-specimens and lap joint testing method for the characterization of adhesively bonded wood. European J. Wood Wood Products 70, 3–10 (2012). 337. R.O. Ebewele, B.H. River, and J.A. Koutsky, Tapered double cantilever beam fracture tests of phenolic-wood adhesive joints Part II. Effects of surface roughness and surface ageing on joint fracture. Wood Fibre Sci. 12, 40–65 (1980). 338. R.O. Ebewele, B.H. River, and J.A. Koutsky, Tapered double cantilever beam fracture tests of phenolic-wood adhesive joints Part I: Development of specimen geometry: Effects of bondline thickness, wood anisotropy and cure time on fracture energy. Wood Fibre Sci. 11, 197–213 (1979). 339. B.H. River, and E.A. Okkonen, Contoured wood double cantilever beam specimen for adhesive joint fracture tests. J Test Eval 21, 21–28 (1993). 340. J. Rathke, G. Sinn, M. Harm, A. Teischinger, M. Weigl, and U. Müller, Fracture energy vs. internal bond strength–mechanical characterization of wood-based panels. Wood Mater. Sci. Eng. 7, 176–185 (2012). 341. J. Rathke, G. Sinn, M. Harm, A. Teischinger, M. Weigl, and U. Müller, Effects of alternative raw materials and varying resin content on mechanical and fracture properties of particle board. Bioresources 7, 2970–2985 (2012). 342. J. Rathke, G. Sinn, M. Weigl, and U. Müller, Analysing orthotropy in the core layer of wood based panels by means of fracture mechanics. European J. Wood Wood Products 70, 851–856 (2012). 343. J. Konnerth, W. Gindl, and U. Müller, Elastic properties of adhesive polymers. I. Polymer films by means of electronic speckle pattern interferometry. J. Appl. Polym. Sci. 103, 3936–3939 (2007).

418 Progress in Adhesion and Adhesives, Volume 4

344. J. Konnerth, A. Valla, W. Gindl, and U. Mueller, Measurement of strain distribution in timber finger joints. Wood Sci. Technol. 40, 631–636 (2006). 345. O. Kläusler, S. Clauss, L. Lübke, J. Trachsel, and P. Niemz, Influence of moisture on stress– strain behaviour of adhesives used for structural bonding of wood. Intl. J. Adhesion Adhesives 44, 57–65 (2013). 346. E. Serrano and B. Enquist, Contact-free measurement and non-linear finite element analyses of strain distribution along wood adhesive bonds. Holzforschung 59, 641–646 (2005). 347. M.A. Sutton, J.J. Orteu, and H. Schreier, Image Correlation for Shape, Motion and Deformation Measurements: Basic Concepts, Theory and Applications, Springer Science & Business Media, New York (2009). 348. F. Lopez-Suevos and C.E. Frazier, Fracture cleavage analysis of PVAc latex adhesives: Influence of phenolic additives. Holzforschung 60, 313–317 (2006). 349. J. Tomblin, P. Harter, W. Seneviratne, and C. Yang, Characterization of bondline thickness effects in adhesive joints. J. Composites Technol. Res. 24, 332–344 (2002). 350. Classification of thermoplastic wood adhesives for non-structural applications, CEN EN 204 (2016) 351. D. Ohnesorge, K. Richter, and G. Becker, Influence of wood properties and bonding parameters on bond durability of European beech (Fagus sylvatica L.) glulams. Annals Forest Sci. 67, 601–601 (2010). 352. A. Bolton and M. Irle, Physical aspects of wood adhesive bond formation with formaldehyde based adhesives Part I. The effect of curing conditions on the physical properties of urea formaldehyde films. Holzforschung 41, 155–158 (1987). 353. L. Muszyn´ski, F. Wang, and S.M. Shaler, Short-term creep tests on phenol-resorcinol-formaldehyde (PRF) resin undergoing moisture content changes. Wood Fiber Sci. 34, 612–624 (2002). 354. R. Wimmer, O. Kläusler, and P. Niemz, Water sorption mechanisms of commercial wood adhesive films. Wood Sci. Technol. 47, 763–775 (2013). 355. J. Konnerth, F. Stöckel, U. Müller, and W. Gindl, Elastic properties of adhesive polymers. III. Adhesive polymer films under dry and wet conditions characterized by means of nanoindentation. J. Appl. Polym. Sci. 118, 1331–1334 (2010). 356. O. Kläusler, P. Hass, C. Amen, S. Schlegel, and P. Niemz, Improvement of tensile shear strength and wood failure percentage of 1C PUR bonded wooden joints at wet stage by means of DMF priming. European J. Wood Wood Products 72, 343–354 (2014). 357. F. López-Suevos and K. Richter, Hydroxymethylated resorcinol (HMR) and novolak-based HMR (n-HMR) primers to enhance bond durability of eucalyptus globulus glulams. J. Adhesion Sci. Technol. 23, 1925–1937 (2009). 358. A.W. Christiansen, Chemical and mechanical aspects of HMR primer in relationship to wood bonding. Forest Prod. J. 55, 73–78 (2005). 359. J. Son and D.J. Gardner, Dimensional stability measurements of thin wood veneers using the Wilhelmy plate technique. Wood Fiber Sci. 36, 98–106 (2004). 360. C.B. Vick and E.A. Okkonen, Durability of one-part polyurethane bonds to wood improved by HMR coupling agent. Forest Prod. J. 50, 69–75 (2000). 361. D.J. Yelle, Solution state NMR analysis of hydroxymethylated resorcinol cured in the presence of crude milled wood lignin from Acer saccharum. J. Appl. Polym. Sci. 134, 1–9 (2017). 362. P. Niemz, H. Bärtschi, and M. Howald, Investigation of moisture distribution and stress formation in timber construction materials under changing climatic conditions. Schweiz. Zeitsch. Forstw. 156, 92–99 (2005).

Understanding Wood Bonds–Going Beyond What Meets the Eye 419

363. M.A. Hubbe, A. Pizzi, H. Zhang, and R. Halis, Critical links governing performance of selfbinding and natural binders for hot-pressed reconstituted lignocellulosic board without added formaldehyde: A review. Bioresources 13, 1–67 (2017). 364. A. Pizzi, Recent developments in eco-efficient bio-based adhesives for wood bonding: Opportunities and issues. J. Adhesion Sci. Technol. 20, 829–846 (2006). 365. V. Hemmilä, S. Adamopoulos, O. Karlsson, and A. Kumar, Development of sustainable bioadhesives for engineered wood panels–A Review. RSC Advances 7, 38604–38630 (2017). 366. C.R. Frihart and L. Lorenz, Protein adhesives, in Handbook of Adhesive Technology, third edition, A. Pizzi and K.L. Mittal (Eds), pp.145–175, CRC Press, Boca Raton, FL (2017). 367. J. Bruck, Current and future needs of the wood based furniture industry, in: Proceedings of International Conference on Wood Adhesives, (2017).

9 Dispersion Adhesion Forces between Macroscopic Objects–Basic Concepts and Modelling Techniques: A Critical Review Youcef Djafri1* and Djamel Turki2 1

University Ibn Khaldoun, Univ-Tiaret, Laboratoire de Synthèse et Catalyse, Tiaret, Algeria 2 University Ibn Khaldoun, Laboratoire de Génie électrique et des plasmas, BP 78, 14000 Tiaret, Algeria

Abstract Dispersion adhesion forces are fundamental forces on which the behaviour of nano-systems is crucially dependent, thus calculating these forces accurately is a major problem to study adhesion in such systems. We have reviewed in this paper the basic concepts related to the process of modelling dispersion adhesion forces between macroscopic objects, a brief theoretical survey of the different modelling techniques used is presented, and each model’s limitations have been emphasised. The aim of this review was to present the reader with a sufficient knowledge which he/she can use to understand and study cohesion and adhesion between macroscopic objects. Keywords: Dispersion forces, Casimir effect, adhesion forces, retardation, van der Waals interactions, Coupled Dipole Method

9.1

Introduction

It has been more than 80 years since London had presented the first explanation and physical description of the origin of the van der Waals forces which were previously predicted to exist between atoms and molecules, and since then this phenomenon has been the focus of researchers from many diverse fields ranging from pure theoretical physics to biology and even to cosmology [1]. From a practical viewpoint, which is what we are interested in, understanding these forces is a fundamental necessity for a better comprehension of the adhesion and cohesion in many practical systems in dry environment.

*Corresponding author: [email protected]

K.L. Mittal (ed.) Progress in Adhesion and Adhesives, Volume 4 (421–442) © 2019 Scrivener Publishing LLC

421

422 Progress in Adhesion and Adhesives, Volume 4

The prediction and control of the behavior of microsystems in different real circumstances while under usage cannot be accomplished with success and accuracy if the processes underlying this behavior are not properly studied and modeled. Many forces are known to cause adhesion and cohesion between macroscopic objects, one of them is the capillary force which manifests due to the presence of a liquid medium between the interacting objects [2]; also the presence of a charge gives rise to the electrostatic force [2–4], and at last dispersion forces that exist in all cases regardless of the type of matter or medium [5, 6], The effect of these forces can lead to sticking of the mechanical elements in nano-/ micro-electromechanical systems (NEMS/MEMS)[7–10]. Due to their critical effect, the study of dispersion forces is very significant and crucial for predicting and controlling the behavior of such small systems [8, 9]. In this review we will focus only on the van der Waals (vdW) forces. The aim of this review is to present an adequate summary of the basic concepts and the different modeling approaches used to study this phenomenon. This review will also focus on updating the knowledge of the reader on the latest works presented by researchers to predict and understand dispersion adhesion forces.

9.2

Basic Concepts

London [11] was the first to present a quantum mechanical description of these forces, finding the dispersion energy to be proportional to the distance by r–6.

EA

3 4

1 0

r

2 6

(1)

Where 1,2 are the polarizabilities of the atoms, 0 is the characteristic frequency, h is given by h/2 with h is Planck’s constant and r is the distance between the center points of the interacting atoms. However, this description was incomplete, since London had disregarded the finite speed of photons transferred between interacting atoms, and he thus considered an instantaneous interaction (Figure 1). To overcome this limitation, Casimir and Polder [12] introduced a model which accounted for the finite speed of electromagnetic waves (Figure 1), and they found that when the distance is sufficiently large, the energy starts to decay faster and becomes proportional to r–7. Arising from quantum fluctuations of charges on atoms and having a macroscopic manifestation gives these forces the uniqueness of being a quantum macroscopic phenomenon. since the discovery of the van der Waals forces, the process of modelling these interactions between macroscopic objects has presented a challenge for researchers [13]. The interaction between macroscopic objects presents a difficult problem to model theoretically. A simple way to solve this problem is to consider the force to be additive and sum it over all possible two-body interactions [14]. This approach presents important information about the general behavior of dispersion forces, but when it comes to predicting the exact value of these forces, the pairwise summation method gives results far from exact, especially for nano-objects. The problem with this model lies in the assumption of

Dispersion Adhesion Forces between Macroscopic Objects–Basic Concepts 423

additivity on which it was built. Moreover, the only mode of interaction considered is the two-body interaction which means that the many-body effects are excluded. Casimir later studied the interaction between two perfectly conducting parallel plates in the vacuum (Figure 1), and he demonstrated that due to zero-point energy a pressure is generated on the plates leading to an attractive force between them. This effect is now known as the Casimir effect [15]. Lifshitz [16] generalized Casimir findings for all dielectric materials, where the interaction between macroscopic objects was calculated from the measured optical (dielectric) properties of the materials, which were considered as a continuum. The great success of this approach lies in the fact that it accounts for all many-body interactions, as well as the retardation effect [5]. Dzyaloshinskii, Lifshitz, and Pitaevskii (DLP) generalized Lifshitz’s model to include the existence of a third medium between the interacting objects [17, 18]. Through the DLP framework, many surprising and critical findings were found, such as the existence of a repulsive as well as an attractive force between objects, depending on the nature of the medium [18]. DLP provides results for only a certain limited number of shapes such as spheres, cylinders and half-spaces, yet for the case of arbitrarily shaped systems, modelling the van der Waals interaction forces using this method can present a difficult problem to tackle.

P1 P2

(a)

Van der Waals non-retarded interactions

p Casimir interaction (c)

Retarded vdW interaction (b)

Figure 9.1 Key manifestations of the dispersion forces, with the first approximation given by London describing an instantaneous dipole P1 generating a field which induces a dipole P2 simultaneously without any retardation (a). The effect of retardation is considered by introducing the finite speed of light into the interaction (b). When macroscopic objects are involved, dispersion forces rise in the form of a Casimir pressure (c).

424 Progress in Adhesion and Adhesives, Volume 4

9.3

Modeling Techniques

Before we embark on our discussion of the different techniques, we should recognize that the convention of the terms used in this paper is that which considers the term van der Waals forces to be reserved for interactions between individual atoms. The Casimir effect, however, is defined to be the interaction between macroscopic objects, and the general term for these forces would be dispersion forces.

9.3.1

The Microscopic Theory (Hamaker’s Approach)

Considering that the dispersion forces exist between two symmetrical and electrically neutral atoms, the potential energy is given by [19]:

EA

(2)

r6

Where is a constant related to the nature of atoms, and r is the distance between the center points of the two atoms. Hamaker assumed that the interaction force between two macroscopic objects can be calculated by summing all the interactions between individual pairs of molecules or atoms [7, 14]:

E

1dr

3

v1

2 dr

3

v2

(3)

r6

Where 1, 2 are the numbers of molecules per unit volume in the two bodies, v1 and v2 are their volumes. When considering a system of two interacting infinite half-spaces as shown in Figure 2, Hamaker demonstrated that the free dispersion energy E||(d) for a unit surface is given by [14]:

AH 1 12 d 2

E d

(4)

Similarly, through a full integration he obtained the interaction energy for dissimilar spherical particles [14]:

Es

s

r

2R1 R2 AH 2 6 r R1 R2

2R1R2 r2

R1 R2

2

ln

r2

R1 R2

r2

R1 R2

2

(5)

Where the Hamaker constant is defined primarily as AH = 2 2 , R1 and R2 are the radii of the two interacting particles and r is the distance between their centers. When the particles

Dispersion Adhesion Forces between Macroscopic Objects–Basic Concepts 425

are at close proximity ([r/(R1 + R2)] 1). Equation (5) is simplified mathematically to the well-known formula for the interaction energy [5, 14]:

Es s (d R)

AH R1R2 1 6 R1 R2 d

(6)

This model has been and is still used to calculate dispersion forces contribution to adhesion and cohesion between particulate materials, especially in powder studies and simulations [19–24]. Vold [25] later applied Hamaker’s approach to study the interactions between anisometric particles and proposed models to calculate the dispersion interaction energies for rectangular rods, cylinders and spherical particles. She showed that for a constant volume and increasing distances, the dispersion interaction energy between rod-shaped anisometric particles becomes remarkably larger than that of spherical particles [25]. In addition, the interaction energy calculated for the geometries studied (rods, cylinders and spheres) is demonstrated to converge with increasing distances, leading to energies of the same order of magnitude [25]. Since its first proposal by Hamaker, the pairwise summation approach was popular for decades due its simplicity, yet only basic geometries had been studied due to the difficult mathematical problems faced when attempting to calculate these forces for non-symmetrical arbitrary shapes. The different models for the basic shapes studied are reviewed sufficiently in the literature [26–28]. There are though some papers that have attempted to calculate these forces between non-traditional geometrical configurations [28]. The most interesting one is the model developed for the interaction between two torus-shaped particles [29], which is a significant geometry for applications in hemodynamics, as it can be used to calculate the contribution of dispersion forces to the adhesion and cohesion of blood cells [29].

R1 r

d

d

R2

Figure 9.2 Van der Waals interactions between two spheres (left), and between parallel half-spaces (right)

426 Progress in Adhesion and Adhesives, Volume 4

For small distance, which is the case related to the process of adhesion and cohesion, the interaction energy between similar and highly symmetrical basic geometries (i.e. spherical, cylindrical, cubic ) is given by a simple elegant formula [30, 31]:

Em (d )

AH 2

Reff

1 m/2

1 m / 2 Lm d

1 m/2

(7)

Where m = 0, 1, 2 for spheres, cylinders and cubes respectively, is the gamma function, L is the length of the particle, and Reff is the effective radius, given as R1 R2 R1 R2 for perfectly spherical objects of radii R1 and R2, and as R R 1 2

9.3.2

R1 R2 for cylinders.

The Proximity-Force Approximation

The Derjaguin approximation, also known as the Proximity-Force Approximation (PFA), is a method for calculating the interaction force between curved objects at very close distances using the interaction energy between parallel surfaces [32]. Figure (3) shows a demonstration of how this method is applied to calculate the force between identical spherical particles. The dispersion interaction energy EPFA between two objects in vicinity is given by the general formula of the PFA theorem which states [33–35]:

EPFA d

2

Reff E r dr

(8)

d

Where E|| is the interaction free energy between two semi-infinite plates, and the effective radius Reff is chosen with respect to the geometry of the interacting objects. For

dr r R1

R2

d

Figure 9.3 Scheme of the Derjaguin approximation for spherical particles, where the interaction force between the two spheres is calculated by summing (integrating) the forces between small circular sections of the two spheres.

Dispersion Adhesion Forces between Macroscopic Objects–Basic Concepts 427

spherical particles, the result of this method coincides with Hamaker’s simple model (Equation 6), which proves how powerful this method is. Nevertheless, it should be emphasized that although the condition on which this approximation is established confines its applicability, yet this method is suitable for studying cohesion and adhesion for large systems. There have been some attempts to generalize the PFA method with the aim to overcome the geometrical restrictions of this approach. The surface element integration [36] and the Surface Integration Approximation [37] thus have been proposed. These modeling techniques have been introduced for calculating  dispersion forces between colloidal particles. Even though these two approaches are based on different assumptions, they still produce the same result [38]. In the Surface Element Integration (SEI) method, the interaction energy between two objects of arbitrary geometries is calculated by taking the interaction energy per unit area between two infinite flat plates E(r), and then double-integrating it over the planes on which the surfaces of these objects are projected [36, 39]. The general formula for this technique is given as follows [40]:

E(r )

S

n2 k2

n1 k1 E (r )dS n1 k1

(9)

Where n1 and n2 are the unit normal vectors pointing outward the surfaces of objects under consideration, k1 and k2 are the unit vectors pointing in the direction of the z-axes of both objects’ coordinate systems. The SEI method has been used primarily to calculate the interaction between an infinite half-space and a particle of arbitrary shape [36], as well as a particle of a randomly generated rough surface [39]. Another relatively recent work has used the same method to calculate adhesion force between colloidal cylindrical and spherical particles [40].

9.3.3

The Retardation Effect

9.3.3.1 The Retarded vdW Forces Models derived based on London’s formula were all developed under the false assumption that the interaction between atoms was instantaneous. It is known that this interaction occurs through a transfer of virtual photons between atoms [30], thus if the finite velocity of photons is considered, this assumption of instantaneity proves to be imprecise [27]. The retardation effect represents the delay in the transfer of virtual photons, which ultimately decreases the intensity of the interaction forces between particles [27]. Casimir and Polder [12] demonstrated that the interaction between two neutral atoms at short distances is proportional to r–6, which coincides with London’s non-retarded model. However, when the distance between the two atoms increases, the retardation becomes significant and the interaction energy becomes proportional to r–7 [12]. These results indicate that the effect of retardation leads to a rapid decay of the van der Waals forces for large

428 Progress in Adhesion and Adhesives, Volume 4

distances, which means that this effect increases with increasing distances. From the results of Casimir and Polder the interaction energy for large distances becomes [12]:

E r

c

23 4

1

r

2

(10)

7

Where c is the speed of light. The retarded van der Waals interaction energy between two atoms E(r) is approximated classically by introducing a correction factor f(p) to the non-retarded energy expression (Equation 1) proposed by London [41, 42]:

E r

r6

f p

(11)

where f(p) depends on p (the dimensionless reduced distance) which is given by p = 2 r/ , with as the characteristic wavelength generally agreed to be 100  nm [41, 42]. Overbeek [43] proposed a correction function f(p) made of two parts (Table 1). Schenkel and Kitchener [44] gave a single relation for this function, which gave a discrepancy less than 15% when it was compared to Overbeek’s model [42, 44]. A better expression for the correction function f(p) for retardation between single atoms has been proposed by Anandarajah and Chen [45]. When their relation was compared with Casimir’s result, this correction function showed an error less than 10%, which makes it the most precise approximation found in the literature. 9.3.3.2 Retardation in Macroscopic Bodies A simple approach to calculate the retarded dispersion forces between macroscopic bodies would be by introducing into the integrand (Equation 3) the correction functions previously stated (Table 1). Hence, within the framework of Hamaker’s pairwise summation method, several models have been proposed in the literature for calculating the retarded dispersion energy between macroscopic bodies. Accordingly, and by using his expressions for the retarded interaction energy between two atoms, Overbeek [43] developed a two-part model Table 9.1 Correction functions proposed in the literature. f(p)

Condition

Ref

f1(p) = 1.10 – 0.14p

0