Spherical nucleic acids. Volume 3 9781003056706, 9789814800358, 9780429200151, 9789814877237, 9781000092486, 1000092488, 9781000092530, 1000092534, 9781000092585, 1000092585, 1003056709

Spherical Nucleic Acids (SNAs) are typically comprised of a nanoparticle core and a densely packed and highly oriented n

159 31 43MB

English Pages xiv, 494 [509] Year 2020

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Spherical nucleic acids. Volume 3
 9781003056706, 9789814800358, 9780429200151, 9789814877237, 9781000092486, 1000092488, 9781000092530, 1000092534, 9781000092585, 1000092585, 1003056709

Table of contents :
Cover......Page 1
Half Title......Page 3
Title Page......Page 5
Copyright Page......Page 6
Table of Contents......Page 7
Part 6: Colloidal Crystallization Processes and Routes to Hierarchical Assembly......Page 17
48: Assembly and Organization Processes in DNA-Directed Colloidal Crystallization......Page 19
48.1 Introduction......Page 20
48.2.1 Formation of Aggregates via Slow Cooling through the Melting Temperature......Page 21
48.2.2 In situ Measurements of Formation......Page 27
48.3 Conclusions......Page 34
48.4 Materials and Methods......Page 35
49: Critical Undercooling in DNA-Mediated Nanoparticle Crystallization......Page 39
49.1 Introduction......Page 40
49.2 Results and Discussion......Page 42
49.3 Conclusion......Page 47
49.4.2 DNA Synthesis and Characterization......Page 48
49.4.5 Temperature-Dependent UV-Visible Spectroscopy Measurements (Freezing and Melting Experiments)......Page 49
50: DNA-Mediated Nanoparticle Crystallization into Wulff Polyhedra......Page 55
50.1 Introduction......Page 56
50.2 Methods Summary......Page 64
51: Growth Dynamics for DNA-Guided Nanoparticle Crystallization......Page 69
51.1 Introduction......Page 70
51.2.1 Interaction Potential......Page 72
51.2.2 Isolated SNA Crystal Growth......Page 75
51.2.3 Coalescence Dynamics......Page 80
51.3 Conclusion and Outlook......Page 87
51.4.1 Model Interaction......Page 88
51.4.2 Grain Boundary Classification......Page 91
52: Nonequilibrium Anisotropic Colloidal Single-Crystal Growth with DNA......Page 97
52.1 Introduction......Page 98
52.2.1 DNA-Programmable Assembly of Anisotropic Colloidal Crystals......Page 102
52.3 Discussion......Page 104
53: Density-Gradient Control over Nanoparticle Supercrystal Formation......Page 113
53.1 Introduction......Page 114
53.2 Materials......Page 126
53.5 Silica Encapsulation of Microcrystals......Page 127
53.6 Characterization Methods......Page 128
Part 7: Dynamic Nanoparticle Superlattices......Page 133
54: Reconstitutable Nanoparticle Superlattices......Page 135
54.1 Introduction......Page 136
55: Contraction and Expansion of Stimuli-Responsive DNA Bonds in Flexible Colloidal Crystals......Page 149
55.1 Introduction......Page 150
56: Topotactic Interconversion of Nanoparticle Superlattices......Page 161
56.1 Introduction......Page 162
57: Dynamically Interchangeable Nanoparticle Superlattices through the Use of Nucleic Acid-Based Allosteric Effectors......Page 173
57.1 Introduction......Page 174
58: Transmutable Nanoparticles with Reconfigurable Surface Ligands......Page 185
58.1 Introduction......Page 186
59: pH-Responsive Nanoparticle Superlattices with Tunable DNA Bonds......Page 197
59.1 Introduction......Page 198
Part 8: Surface-Based and Template-Confined Colloidal Crystallization......Page 207
60: Stepwise Evolution of DNA-Programmable Nanoparticle Superlattices......Page 209
60.1 Introduction......Page 210
60.2 Methods and Results......Page 211
60.3 Conclusion......Page 218
60.4 Experimental Section......Page 219
61: Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates......Page 223
61.1 Introduction......Page 224
61.2 Methods and Results......Page 227
61.3 Conclusion......Page 235
62: Epitaxy: Programmable Atom Equivalents versus Atoms......Page 239
62.1 Introduction......Page 240
62.2 Results and Discussion......Page 242
62.4.1 DNA Functionalization of Gold Nanoparticles......Page 247
62.4.2.1 Patterned template synthesis......Page 248
62.4.2.4 Substrate DNA functionalization......Page 249
62.4.3.2 DNA-NP superlattice thin-film assembly......Page 250
62.4.4 Silica Embedding......Page 251
62.4.5.2 Grazing-incidence SAXS......Page 252
62.4.6.1 RMS roughness and mean thickness calculation......Page 253
63: Lattice Mismatch in Crystalline Nanoparticle Thin Films......Page 257
63.1 Introduction......Page 258
63.2 Methods......Page 260
63.3 Results and Discussion......Page 263
64: Building Superlattices from Individual Nanoparticles via Template-Confined DNA-Mediated Assembly......Page 275
64.1 Introduction......Page 276
64.2 Methods......Page 277
64.3 Results and Discussion......Page 282
65: Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly......Page 289
65.1 Introduction......Page 290
65.2 Results and Discussion......Page 292
65.4 Experimental Section......Page 300
66: DNA-Mediated Size-Selective Nanoparticle Assembly for Multiplexed Surface Encoding......Page 307
66.1 Introduction......Page 308
66.2 Methods......Page 310
66.3 Discussion......Page 316
67.1 Introduction......Page 321
67.2.3 Coupled Dipole Simulations......Page 330
Part 9: Optics and Plasmonics with Nanoparticle Superlattices......Page 335
68. Plasmonic Photonic Crystals Realized through DNA-Programmable Assembly......Page 337
68.1 Introduction......Page 338
68.2.1 FDTD Calculations......Page 346
68.2.3 Optical Experiments......Page 347
69: Nanoscale Form Dictates Mesoscale Function in Plasmonic DNA-Nanoparticle Superlattices......Page 351
69.1 Introduction......Page 352
69.2 Optical Properties of Plasmonic Nanoparticle Assemblies......Page 354
69.3 DNA-Programmable Assembly of Mesoscale Superlattices......Page 356
69.4 Optical Characterization of Plasmonic Superlattices......Page 358
69.5 Crystal Habit as a Design Parameter for Optical Response......Page 360
69.6 Conclusions......Page 363
70: Directional Emission from Dye-Functionalized Plasmonic DNA Superlattice Microcavities......Page 367
70.1 Introduction......Page 369
70.2.4 TCSPC Lifetime Detection Limit......Page 379
71: Defect Tolerance and the Effect of Structural Inhomogeneity in Plasmonic DNA-Nanoparticle Superlattices......Page 383
71.1 Introduction......Page 384
71.2 Results......Page 386
71.3 Discussion......Page 396
71.4.2 Optical Measurements......Page 397
72: Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties......Page 401
73: Plasmonic Metallurgy Enabled by DNA......Page 419
74: Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices......Page 431
74.1 Introduction......Page 432
74.2 Building PCs with Plasmonic NP Superlattices......Page 434
74.3 PPCs Realized through DNA-Programmable Assembly......Page 441
74.4 Building PCs with Materials Other Than Au......Page 443
74.5 Conclusions......Page 444
74.6.3 Superlattice Assembly......Page 445
74.6.4 Optical Experiments......Page 446
75: Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled by DNA-Programmable Assembly......Page 451
76: Polarization-Dependent Optical Response in Anisotropic Nanoparticle-DNA Superlattices......Page 465
Part 10: Postsynthetic Modification and Catalysis with Nanoparticle Superlattices......Page 479
77.1 Introduction......Page 481
77.2.2 Silica Embedding......Page 491
77.2.4 Transmission Electron Microscopy......Page 492
78: Controlling Structure and Porosity in Catalytic Nanoparticle Superlattices with DNA......Page 495
78.1 Introduction......Page 496
78.2.2 Calcination......Page 498
78.2.3 Catalytic Oxidation of 4-Hydroxybenzyl Alcohol......Page 500
78.2.4 Evaluation of Structure Following Catalytic Experiments......Page 501
78.3 Conclusions......Page 502
78.4.1 Materials and General Procedures......Page 503
78.4.2 Nanoparticle Assembly and Crystallization......Page 504
78.4.4 Preparation of Supported, Nonassembled Gold Nanoparticles......Page 505
78.4.9 Small-Angle X-ray Scattering (SAXS)......Page 506
78.4.11 Catalysis Experiments......Page 507

Citation preview

Spherical Nucleic Acids

Spherical Nucleic Acids Volume 3

edited by

Chad A. Mirkin

Published by Jenny Stanford Publishing Pte. Ltd. Level 34, Centennial Tower 3 Temasek Avenue Singapore 039190

Email: [email protected] Web: www.jennystanford.com British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

Spherical Nucleic Acids, Volume 3 Copyright © 2020 by Jenny Stanford Publishing Pte. Ltd.

All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher. ISBN ISBN ISBN ISBN

978-981-4800-35-8 (Set) (Hardcover) 978-0-429-20015-1 (Set) (eBook) 978-981-4877-23-7 (Volume 3) (Hardcover) 978-1-003-05670-6 (Volume 3) (eBook)

Contents

Part 6

Colloidal Crystallization Processes and Routes to Hierarchical Assembly

48. Assembly and Organization Processes in DNA-Directed Colloidal Crystallization Robert J. Macfarlane, Byeongdu Lee, Haley D. Hill, Andrew J. Senesi, Soenke Seifert, and Chad A. Mirkin 48.1 Introduction 48.2 Results and Discussion 48.2.1 Formation of Aggregates via Slow Cooling through the Melting Temperature 48.2.2 In situ Measurements of Formation 48.3 Conclusions 48.4 Materials and Methods

49. Critical Undercooling in DNA-Mediated Nanoparticle Crystallization Matthew N. O’Brien, Keith A. Brown, and Chad A. Mirkin 49.1 Introduction 49.2 Results and Discussion 49.3 Conclusion 49.4 Methods 49.4.1 Nanoparticle Synthesis and Characterization 49.4.2 DNA Synthesis and Characterization 49.4.3 Nanoparticle Functionalization with DNA 49.4.4 Nanoparticle Assembly 49.4.5 Temperature-Dependent UV-Visible Spectroscopy Measurements (Freezing and Melting Experiments)

939

940 941 941 947 954 955 959

960 962 967 968

968 968 969 969

969

vi

Contents

50. DNA-Mediated Nanoparticle Crystallization into Wulff Polyhedra Evelyn Auyeung, Ting I. N. G. Li, Andrew J. Senesi, Abrin L. Schmucker, Bridget C. Pals, Monica Olvera de la Cruz, and Chad A. Mirkin 50.1 Introduction 50.2 Methods Summary

51. Growth Dynamics for DNA-Guided Nanoparticle Crystallization Subas Dhakal, Kevin L. Kohlstedt, George C. Schatz, Chad A. Mirkin, and Monica Olvera de la Cruz 51.1 Introduction 51.2 Results and Discussion 51.2.1 Interaction Potential 51.2.2 Isolated SNA Crystal Growth 51.2.3 Coalescence Dynamics 51.3 Conclusion and Outlook 51.4 Materials and Methods 51.4.1 Model Interaction 51.4.2 Grain Boundary Classification

52. Nonequilibrium Anisotropic Colloidal Single-Crystal Growth with DNA Soyoung E. Seo, Martin Girard, Monica Olvera de la Cruz, and Chad A. Mirkin 52.1 Introduction 52.2 Results 52.2.1 DNA-Programmable Assembly of Anisotropic Colloidal Crystals 52.2.2 Tunable Structural Parameters at the Nanoscale 52.3 Discussion 53. Density-Gradient Control over Nanoparticle Supercrystal Formation Taegon Oh, Jessie C. Ku, Jae-Hyeok Lee, Mark C. Hersam, and Chad A. Mirkin 53.1 Introduction

975

976 984 989

990 992 992 995 1000 1007 1008 1008 1011 1017

1018 1022 1022 1024 1024 1033

1034

Contents

53.2 53.3 53.4 53.5 53.6

Part 7

Materials Preparation of PAEs and Their Hybridization Preparation of PAE Microcrystals Silica Encapsulation of Microcrystals Characterization Methods

Dynamic Nanoparticle Superlattices

54. Reconstitutable Nanoparticle Superlattices Boya Radha, Andrew J. Senesi, Matthew N. O’Brien, Mary X. Wang, Evelyn Auyeung, Byeongdu Lee, and Chad A. Mirkin 54.1 Introduction

55. Contraction and Expansion of Stimuli-Responsive DNA Bonds in Flexible Colloidal Crystals Jarad A. Mason, Christine R. Laramy, Cheng-Tsung Lai, Matthew N. O’Brien, Qing-Yuan Lin, Vinayak P. Dravid, George C. Schatz, and Chad A. Mirkin 55.1 Introduction 56. Topotactic Interconversion of Nanoparticle Superlattices Robert J. Macfarlane, Matthew R. Jones, Byeongdu Lee, Evelyn Auyeung, and Chad A. Mirkin 56.1 Introduction 57. Dynamically Interchangeable Nanoparticle Superlattices through the Use of Nucleic Acid-Based Allosteric Effectors Youngeun Kim, Robert J. Macfarlane, and Chad A. Mirkin 57.1 Introduction

58. Transmutable Nanoparticles with Reconfigurable Surface Ligands Youngeun Kim, Robert J. Macfarlane, Matthew R. Jones, and Chad A. Mirkin 58.1 Introduction

1046 1047 1047 1047 1048 1055

1056 1069

1070 1081

1082 1093

1094 1105

1106

vii

viii

Contents

59. pH-Responsive Nanoparticle Superlattices with Tunable DNA Bonds Jinghan Zhu, Youngeun Kim, Haixin Lin, Shunzhi Wang, and Chad A. Mirkin 59.1 Introduction

Part 8

Surface-Based and Template-Confined Colloidal Crystallization

60. Stepwise Evolution of DNA-Programmable Nanoparticle Superlattices Andrew J. Senesi, Daniel J. Eichelsdoerfer, Robert J. Macfarlane, Matthew R. Jones, Evelyn Auyeung, Byeongdu Lee, and Chad A. Mirkin 60.1 Introduction 60.2 Methods and Results 60.3 Conclusion 60.4 Experimental Section

61. Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates Sondra L. Hellstrom, Youngeun Kim, James S. Fakonas, Andrew J. Senesi, Robert J. Macfarlane, Chad A. Mirkin, and Harry A. Atwater 61.1 Introduction 61.2 Methods and Results 61.3 Conclusion 62. Epitaxy: Programmable Atom Equivalents versus Atoms Mary X. Wang, Soyoung E. Seo, Paul A. Gabrys, Dagny Fleischman, Byeongdu Lee, Youngeun Kim, Harry A. Atwater, Robert J. Macfarlane, and Chad A. Mirkin 62.1 Introduction 62.2 Results and Discussion 62.3 Conclusion 62.4 Methods 62.4.1 DNA Functionalization of Gold Nanoparticles

1117

1118

1129

1130 1131 1138 1139 1143

1144 1147 1155 1159

1160 1162 1167 1167 1167

Contents

62.4.2 Substrate Preparation and Functionalization 62.4.2.1 Patterned template synthesis 62.4.2.2 Silanization of patterned templates 62.4.2.3 Unpatterned substrate preparation 62.4.2.4 Substrate DNA functionalization 62.4.3 Layer-by-Layer DNA-Nanoparticle Superlattice Thin-Film Assembly 62.4.3.1 Determining the thin-film annealing temperature 62.4.3.2 DNA-NP superlattice thinfilm assembly 62.4.4 Silica Embedding 62.4.5 Small-Angle X-ray Scattering 62.4.5.1 SAXS experimental conditions 62.4.5.2 Grazing-incidence SAXS 62.4.6 Focused Ion Beam–Scanning Electron Microscopy 62.4.6.1 RMS roughness and mean thickness calculation

63. Lattice Mismatch in Crystalline Nanoparticle Thin Films Paul A. Gabrys, Soyoung E. Seo, Mary X. Wang, EunBi Oh, Robert J. Macfarlane, and Chad A. Mirkin 63.1 Introduction 63.2 Methods 63.3 Results and Discussion 64. Building Superlattices from Individual Nanoparticles via Template-Confined DNA-Mediated Assembly Qing-Yuan Lin, Jarad A. Mason, Zhongyang Li, Wenjie Zhou, Matthew N. O’Brien, Keith A. Brown, Matthew R. Jones, Serkan Butun, Byeongdu Lee, Vinayak P. Dravid, Koray Aydin, and Chad A. Mirkin 64.1 Introduction 64.2 Methods

1168

1168 1169

1169

1169

1170

1170

1170 1171 1172

1172 1172 1173 1173

1177

1178 1180 1183 1195

1196 1197

ix

x

Contents

64.3

Results and Discussion

65. Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly Wenjie Zhou, Qing-Yuan Lin, Jarad A. Mason, Vinayak P. Dravid, and Chad A. Mirkin 65.1 Introduction 65.2 Results and Discussion 65.3 Conclusion 65.4 Experimental Section 66. DNA-Mediated Size-Selective Nanoparticle Assembly for Multiplexed Surface Encoding Qing-Yuan Lin, Edgar Palacios, Wenjie Zhou, Zhongyang Li, Jarad A. Mason, Zizhuo Liu, Haixin Lin, Peng-Cheng Chen, Vinayak P. Dravid, Koray Aydin, and Chad A. Mirkin 66.1 Introduction 66.2 Methods 66.3 Discussion

67. Conformal, Macroscopic Crystalline Nanoparticle Sheets Assembled with DNA Jessie C. Ku, Michael B. Ross, George C. Schatz, and Chad A. Mirkin 67.1 Introduction 67.2 Experimental Section 67.2.1 Substrate Preparation 67.2.2 Silica Embedding and Film Delamination 67.2.3 Coupled Dipole Simulations

Part 9

Optics and Plasmonics with Nanoparticle Superlattices

68. Plasmonic Photonic Crystals Realized through DNAProgrammable Assembly Daniel J. Park, Chuan Zhang, Jessie C. Ku, Yu Zhou, George C. Schatz, and Chad A. Mirkin 68.1 Introduction

1202 1209

1210 1212 1220 1220 1227

1228 1230 1236 1241

1241 1250 1250 1250 1250

1257

1258

Contents

68.2

Methods 68.2.1 FDTD Calculations 68.2.2 EMT Approximation 68.2.3 Optical Experiments

69. Nanoscale Form Dictates Mesoscale Function in Plasmonic DNA-Nanoparticle Superlattices Michael B. Ross, Jessie C. Ku, Victoria M. Vaccarezza, George C. Schatz, and Chad A. Mirkin 69.1 Introduction 69.2 Optical Properties of Plasmonic Nanoparticle Assemblies 69.3 DNA-Programmable Assembly of Mesoscale Superlattices 69.4 Optical Characterization of Plasmonic Superlattices 69.5 Crystal Habit as a Design Parameter for Optical Response 69.6 Conclusions

70. Directional Emission from Dye-Functionalized Plasmonic DNA Superlattice Microcavities Daniel J. Park, Jessie C. Ku, Lin Sun, Clotilde M. Lethiec, Nathaniel P. Stern, George C. Schatz, and Chad A. Mirkin 70.1 Introduction 70.2 Methods 70.2.1 FDTD Calculations 70.2.2 Effective Medium Theory Approximation 70.2.3 Optical Experiments 70.2.4 TCSPC Lifetime Detection Limit 71. Defect Tolerance and the Effect of Structural Inhomogeneity in Plasmonic DNA-Nanoparticle Superlattices Michael B. Ross, Jessie C. Ku, Martin G. Blaber, Chad A. Mirkin, and George C. Schatz 71.1 Introduction 71.2 Results

1266 1266 1267 1267 1271

1272

1274

1276 1278

1280 1283 1287

1289 1299 1299

1299 1299 1299 1303

1304 1306

xi

xii

Contents

71.3 71.4

Discussion Materials and Methods 71.4.1 Optical Simulations 71.4.2 Optical Measurements

72. Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties Kaylie L. Young, Michael B. Ross, Martin G. Blaber, Matthew Rycenga, Matthew R. Jones, Chuan Zhang, Andrew J. Sensei, Byeongdu Lee, George C. Schatz, and Chad A. Mirkin 73. Plasmonic Metallurgy Enabled by DNA Michael B. Ross, Jessie C. Ku, Byeongdu Lee, Chad A. Mirkin, and George C. Schatz 74. Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices Lin Sun, Haixin Lin, Kevin L. Kohlstedt, George C. Schatz, and Chad A. Mirkin 74.1 Introduction 74.2 Building PCs with Plasmonic NP Superlattices 74.3 PPCs Realized through DNA-Programmable Assembly 74.4 Building PCs with Materials Other Than Au 74.5 Conclusions 74.6 Methods 74.6.1 FDTD Calculations 74.6.2 EMT Approximation and TMM 74.6.3 Superlattice Assembly 74.6.4 Optical Experiments 75. Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled by DNA-Programmable Assembly Matthew R. Jones, Kevin L. Kohlstedt, Matthew N. O’Brien, Jinsong Wu, George C. Schatz, and Chad A. Mirkin

1316 1317 1317 1317 1321

1339

1351

1352 1354

1361 1363 1364 1365 1365 1365 1365 1366 1371

Contents

76. Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices Lin Sun, Haixin Lin, Daniel J. Park, Marc R. Bourgeois, Michael B. Ross, Jessie C. Ku, George C. Schatz, and Chad A. Mirkin,

Part 10

1385

Postsynthetic Modification and Catalysis with Nanoparticle Superlattices

77. Transitioning DNA–Engineered Nanoparticle Superlattices from Solution to the Solid State Evelyn Auyeung, Robert J. Macfarlane, Chung Hang J. Choi, Joshua I. Cutler, and Chad A. Mirkin 77.1 Introduction 77.2 Experimental Section 77.2.1 Nanoparticle Crystallization 77.2.2 Silica Embedding 77.2.3 Small-Angle X-ray Scattering 77.2.4 Transmission Electron Microscopy

78. Controlling Structure and Porosity in Catalytic Nanoparticle Superlattices with DNA Evelyn Auyeung, William Morris, Joseph E. Mondloch, Joseph T. Hupp, Omar K. Farha, and Chad A. Mirkin 78.1 Introduction 78.2 Results and Discussion 78.2.1 Superlattice Synthesis 78.2.2 Calcination 78.2.3 Catalytic Oxidation of 4-Hydroxybenzyl Alcohol 78.2.4 Evaluation of Structure Following Catalytic Experiments 78.3 Conclusions 78.4 Experimental Section 78.4.1 Materials and General Procedures 78.4.2 Nanoparticle Assembly and Crystallization 78.4.3 Silica Embedding

1401

1401 1411 1411 1411 1412 1412 1415

1416 1418 1418 1418 1420

1421 1422 1423 1423 1424 1425

xiii

xiv

Contents

78.4.4 Preparation of Supported, Nonassembled Gold Nanoparticles 78.4.5 Calcination of Superlattice Assemblies 78.4.6 Thermogravimetric Analysis (TGA) 78.4.7 Fourier Transform Infrared Spectroscopy (FTIR) 78.4.8 N2 Isotherm Measurements 78.4.9 Small-Angle X-ray Scattering (SAXS) 78.4.10 Inductively Coupled Plasma Mass Spectrometry (ICP-MS) 78.4.11 Catalysis Experiments

1425

1426 1426

1426 1426 1426 1427 1427

Part 6

Colloidal Crystallization Processes and Routes to Hierarchical Assembly

Chapter 48

Assembly and Organization Processes in DNA-Directed Colloidal Crystallization*

Robert J. Macfarlane,a,b Byeongdu Lee,c Haley D. Hill,a,b Andrew J. Senesi,a,b Soenke Seifert,c and Chad A. Mirkina,b aDepartment

of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA bInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cX-ray Science Division, Advanced Photon Source, Argonne National Laboratory, 9700 S. Cass Ave., Argonne, IL 60439, USA [email protected]

We present an analysis of the key steps involved in the DNA-directed assembly of nanoparticles into crystallites and polycrystalline aggregates. Additionally, the rate of crystal growth as a function of increased DNA linker length, solution temperature, and selfcomplementary versus non-self-complementary DNA linker strands (1- versus 2-component systems) has been studied. The data show that the crystals grow via a 3-step process: an initial ‘‘random binding’’ phase resulting in disordered DNA-AuNP aggregates, *Reprinted with permission from Macfarlane, R. J., Lee, B., Hill, H. D., Senesi, A. J., Seifert, S. and Mirkin, C. A. (2009). Assembly and organization processes in DNAdirected colloidal crystallization, PNAS 106(26), 10493–10498.

Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

940

Assembly and Organization Processes in DNA-Directed Colloidal Crystallization

followed by localized reorganization and subsequent growth of crystalline domain size, where the resulting crystals are wellordered at all subsequent stages of growth.

48.1

Introduction

The chemical and physical properties of most materials are determined by the placement of individual atoms relative to one another [1–4]. These atom–atom interactions include forces such as covalent and ionic bonds, van der Waals forces and London dispersion forces. However, in the field of nanomaterials, the length scales of particle assembly are significantly larger and the assembly process is governed by a more complicated set of interactions [5–7]. The tailorable arrangement of nanoscale materials through directed-mechanisms has proven to be the most practical method to create ordered arrangements of nanoparticles in solution [8–15]; crystalline nanoparticle aggregates have potential implications for the development of materials with unique plasmonic [16, 17], photonic [18, 19], electrical [20], and magnetic properties [21]. Previous strategies developed to create well-ordered nanoscale assemblies have used electrostatic interactions [11, 12], hydrogenbonding networks [10, 13], and peptide recognition properties [9]. Over a decade ago, we introduced the concept of synthetically programmable particle assembly through the use of DNA and polyvalent oligonucleotide nanoparticle conjugates [22]. Recently, we [23, 24] and the Gang group [25] independently used these principles to construct highly-ordered nanoparticle crystallites via DNA hybridization, where crystal type and lattice parameters can be programmed through design of DNA linker. The formation of these crystals involves multiple types of molecular interactions that are highly dependent and predictable based on the DNA base sequence [22, 26–28], where the hybridization of the DNA linkers drives the crystallization process. Moreover, the ultimate structure formed is typically the one that maximizes the number of hybridization events, and therefore, if enough thermal energy is provided, the system will typically equilibrate to this structure [23, 25, 29, 30]. Indeed, only highly-ordered crystalline aggregates present the thermodynamically most stable arrangement

Results and Discussion

of nanoparticles, because they allow for a maximum of nearestneighbor complementary DNA interactions. Previous work has shown that the complexity of the nanoparticle systems studied thus far necessitates significant thermal annealing to create the thermodynamically favorable crystal structures [23, 24]. Combining particles at 22°C often yields disordered structures or semicrystalline materials that lack the long-range order observed after the annealing process. Fundamentally, the mechanism by which individual DNA-functionalized nanoparticles (DNA-AuNPs) assemble into crystalline materials is of interest because the forces governing crystal formation are significantly different from their atomic analogues. Herein, we present a study of the formation process for these DNA-linked nanoparticle crystal systems, using synchrotron small angle X-ray scattering (SAXS) to monitor the initial formation and growth of highly-ordered 3D assemblies in 3 distinct crystal systems, showing variations in assembly parameters as a function of solution temperature, DNA length and 1- versus 2-component DNA-AuNP systems.

48.2

Results and Discussion

48.2.1 Formation of Aggregates via Slow Cooling through the Melting Temperature DNA-functionalized gold nanoparticles (DNA-AuNPs) were prepared via the addition of excess synthetic oligo- nucleotides containing a 5’ hexyl-thiol moiety to a solution of either 5- or 10 nm AuNPs (9.3 ± 0.9 nm, 5.3 ± 0.7 nm). This was followed by a slow salting process (see Materials and Methods), which yielded densely functionalized oligonucleotide nanoparticle conjugates (58 ± 5 DNA strands/NP for 10 nm AuNPs, 12 ± 3 DNA strands/NP for 5 nm AuNPs). The salt screens the electrostatic interactions between negatively charged oligonucleotides, allowing for more DNA strands to enter the coordination sphere of the particle [31]. All of the DNA sequences used in these experiments have been demonstrated to direct the formation of well-ordered nanoparticle ‘‘crystal structures’’ (Scheme 48.1) [23, 24]. DNA-AuNPs were crystallized via the addition of linker strands containing a region complementary to the

941

942

Assembly and Organization Processes in DNA-Directed Colloidal Crystallization

AuNP-bound DNA Sequence 1, an unpaired flexor base and a 4- or 7-base linker-linker overlap region. The long FCC linker (Scheme 48.1, containing a poly dT sequence to increase DNA length) was hybridized with a complementary 39-mer poly dA filler DNA before hybridizing the linker to the DNA-AuNPs. Note that DNA-linkers containing a self-complementary 5’ CGCG end form face-centered cubic (FCC) crystals, whereas DNA-linkers that contain non-selfcomplementary 5’ ends form body-centered cubic (BCC) crystals and that, individually, these short overlap sequences result in weak binding interactions. Aggregation of the particles is induced because of the polyvalent nature of the DNA-AuNPs, which allows for multiple linker-linker hybridization events per particle. Using the method devised by Marky et al. [32], we can estimate the ΔG of hybridization for these overlaps to be 10.3 and 12.0 kcal/mol, respectively. Given these similarly low estimated values, we can qualitatively state that the slight difference between the 2 calculated free energies of hybridization is negligible for these individually-weak DNA binding interactions. Therefore, the energetics of DNA hybridization can be treated as approximately equivalent for the self- and non-selfcomplementary systems, meaning that differences in hybridization energies should not play a major factor in these studies.

Scheme 48.1  DNA sequences used. AuNPs were functionalized with ≈58 ± 5  (10 nm AuNPs) or 12 ± 3 (5 nm AuNPs) strands of DNA Sequence 1, linked via a  5’ hexyl-thiol moiety. Crystals were formed via the addition of linker sequences with a 3’ end complementary to the Au-bound DNA, a single flexor base used to  increase DNA strand flexibility, and a 5’ end that is either self-complementary  (FCC crystals) or non-self-complementary (BCC crystals).

All structures were characterized by small angle X-ray scattering (SAXS), using Argonne National Laboratory’s Advanced Photon

Results and Discussion

Source. In brief, particles in solution are treated as inelastic Rayleigh scatterers—photons that are scattered by the nanoparticles in solution create distinct 2D interference patterns that are dependent on the position of the scatterers relative to one another [33]. Azimuthal averaging of these 2D patterns yields 1D scans that allow for an analysis of DNA-AuNP crystal structures obtained. We first studied the effects of increased DNA interconnect length on the rate of crystal formation, using 10 nm DNA-AuNPs and the long FCC and short FCC linkers presented in Scheme 48.1. Initial aggregates were formed via the addition of linker DNA strands (1.25 μM) to DNA-AuNPs (25 nM) in a microcentrifuge tube. This solution was then transferred to a 2.0-mm quartz capillary tube for annealing and X-ray scattering experiments. The samples were heated to 60°C (several degrees above the temperature at which the DNAAuNP aggregates dissociate) and slowly cooled at a rate of 1°C/min, with images taken at ~1-min intervals. The scattering experiments show that there is a progression from dispersed (nonaggregated) DNA-AuNPs to an ordered crystal structure that is observable over the monitored time scale (Fig. 48.1). However, the long FCC linker system reaches the highest-ordered state more slowly (Fig. 48.1A) than the short FCC linker system (Fig. 48.1B). As shown in Fig. 48.1B, the short FCC linker shows a distinct shift between unbound DNA-AuNPs and FCC crystals, with no readily observable intermediates. Maximum intensity peaks appear within 3 minutes of the first observed FCC scattering peak. (The subsequent decrease in intensity can be attributed to both DNA degradation because of X-ray beam exposure and aggregates settling out of the path of the X-ray beam.) However, the long FCC linker system shows an initial weak, diffuse peak, which persists for several minutes before the formation of peaks corresponding to an FCC crystal. The peaks associated with the FCC structure continue to increase in intensity for several minutes after their initial formation. A complete analysis of this transition will be discussed at the end of this section. Analogous to the growth of molecular crystals [34], the size of the crystalline domains only increases if the incoming building materials adopt positions in the growing crystal that are an extension of the ordered structure. It is important to make a distinction between growth in the size of the crystalline domain and growth in the size of the DNA-AuNP aggregate. Crystal growth refers to the movement

943

944

Assembly and Organization Processes in DNA-Directed Colloidal Crystallization

of a DNA-AuNP from a position outside of the crystal lattice to a position in line with the existing crystallographic arrangement. This can occur via binding of a free DNA-AuNP, or via repositioning of an already bound DNA-AuNP from a disordered to an ordered position. Conversely, aggregate growth refers to the binding of a DNA-AuNP to an existing aggregate (regardless of whether or not the AuNP is ordered relative to the crystal lattice). Therefore, crystal growth can occur at a rate different from the rate of aggregate growth.

Figure 48.1 1D SAXS profiles of the long FCC (A) and short FCC (B) linker systems. As the DNA-AuNPs are cooled from above their melting temperature (1°C/min), initial DNA linkages create random aggregates with no long-range order. Given time to anneal, these aggregates then reform into well-ordered crystalline domains. The vertical lines notated as 2π/ΔAu and √6π/ΔAu indicate the position of predicted q0 peaks corresponding to disordered aggregates and perfect FCC crystals, respectively.

The rate of crystal growth as a function of DNA length is largely affected by 2 principle factors: an increased flexibility of the DNA interconnects and an increase in the distance between the outer ‘‘recognition’’ end of the DNA strands with increasing DNA linker length. In the long FCC linker system, the increased flexibility of the DNA strands provides a larger number of degrees of freedom for both the DNA and the DNA-AuNP, thereby slowing the rate of crystal

Results and Discussion

growth (although not necessarily aggregate growth). Additionally, an increase in DNA linker length increases the hydrodynamic radius of the DNA-AuNP, resulting in a subsequent increase in the distance between the free ends of the DNA strands (~6.6 nm for short FCC, ~11.9 nm for long FCC) on 1 particle, due to surface curvature. In order for a DNA-AuNP to move into a position in line with the existing crystal lattice, the DNA strands linking the AuNP must undergo dehybridization and rehybridization events to allow the AuNP to move to a position where the largest number of DNA duplexes can occur (the most stable crystal formation). If the DNA strands do not undergo this process, the DNA-AuNPs are locked into the initial, disordered structure. Increasing the distance between the free ends of the linking DNA strands decreases the rate at which these subsequent DNA hybridization events can occur during the reorganization process. This in turn results in the long FCC DNAAuNP aggregates restructuring more slowly than their short FCC linker counterparts. As we state above, a clear delineation of the transition from unbound DNA-AuNPs to crystals can be seen with the long FCC linker system, which allows us to probe the method of crystal formation. In this system, the first appearance of the first-order DNA-AuNP scattering peak (q0) occurs at 54°C; however, this peak is broad and centered at a q-value noticeably lower than the position of the q0 peak of the highest ordered crystal (T 39°C). The breadth and weak intensity of this peak indicates that the aggregates are both small and highly disordered, whereas the position of the q0 peak indicates that the crystallographic arrangement of nanoparticles is different from the aggregates observed at later stages of growth. By plotting the q0 peak positions relative to the q0 peak position of the most ordered crystal q0/q0max (Fig. 48.2A), we can see that there is a shift in peak position between 50 and 47°C. We attribute this to a reorganization of the nanoparticles within the aggregate—for FCC crystals and amorphous aggregates, the q0 peak position should occur at values of √6π/dAu and 2π/dAu, respectively, where dAu is the distance between DNA-AuNP nearest neighbors. Note that, for a system with the same interparticle spacing, the peak positions for FCC and disordered crystals occur at relative values of 1.0 and ~0.82, further indicating that the structural shift seen in Fig. 48.2A is due to a transition from amorphous aggregates to a crystalline material.

945

946

Assembly and Organization Processes in DNA-Directed Colloidal Crystallization

Other possible crystalline arrangements such as hexagonal closepacked (HCP) or random hexagonal close-packed (rHCP) would exhibit a q0 peak position very close to that of FCC. However, we can distinguish between FCC crystals and these other structures by comparing the relative positions of higher order scattering peaks— the appearance of peaks at (√8/3)q0 and (√11/3)q0 (along with the positions of other, higher order scattering peaks) confirms that the crystals are FCC.

Figure 48.2 Plots demonstrating the growth of crystalline aggregates in the long FCC linker system. (A) Plot of q0/q0max, showing the appearance of NP scattering peaks at 54°C and the transition to an FCC formation between 50 and 49°C. (B) Plot of 2π/Δq0, demonstrating the growth in nanocrystal size at temperatures 150 nm). This low particle density is attributed to the two-fold decrease in substratebound DNA linkers, which makes particle-substrate interactions less favorable. The decrease in attractive interaction is also observed by monitoring the thermal desorption transition, which broadens and shifts to lower temperatures compared to the desorption of SNA-AuNPs from mono-functionalized DNA substrates. In the

Figure 60.3  Stepwise GISAXS characterization of (100)- and (110)-textured bcc superlattices. (a, e) Idealized schematic and 2D GISAXS scattering patterns after  1, 2 and 3 half deposition-cycles on mono-functionalized (100)-directing and bi-functionalized (110)-directing DNA substrates, respectively. (b, f) Normalized GISAXS horizontal linecuts at af = ai after successive half-cycles on mono-functionalized and bi-functionalized DNA substrates, respectively. The curves are offset  for clarity and plotted on a log–log scale. Note the systematic absence (b) or presence (f) of the (211) peak, which can be used as a diagnostic for determining crystallographic orientation. The peaks at q = 0.02 ((b), 5 and 7 half-cycles) and q=0.023 ((f), 7 and 10 half-cycles) are from diffuse scattering and are not attributed to the particle arrangement. The crystal grain sizes increase with deposition cycles, both in the plane of the substrate (c, g) and out of the plane (d, h), for (100)- and (110)-textured superlattices, respectively. The gray bars in (c, g) show average domain size, while the colored markers show various structures deconvoluted from the GISAXS line cuts (“ML” denotes the presence of a monolayer). Error bars are determined from the standard deviation of at least three separate measurements, and in many cases are smaller than the marker size. The dashed lines in (d, h) show the expected film thickness if a full layer were deposited per half-cycle.

Methods and Results 1137

1138

Stepwise Evolution of DNA-Programmable Nanoparticle Superlattices

context of superlattice growth, this sub-monolayer acts as a seed layer, such that after two half-cycles, only 2D aggregation occurs with no increase in film thickness. The structure is consistent with the (110) plane of a bcc crystal, and the characteristic first-order scattering peak at q = 0.024 Å−1 was observed through the first four half-cycles (Fig. 60.3e,f). The presence of this monolayer after several deposition cycles indicates island formation, which is often observed when substrate–particle interactions are weaker than particle–particle interactions. 3D ordering with bcc symmetry (q0 at 0.028 Å−1) was first observed with as little as two crystal layers, which here occurred after three half-cycles. As before, subsequent growth resulted in a linear increase in film thickness with deposition cycles and crystallite in-plane growth (Fig. 60.3g,h).

60.3

Conclusion

We have demonstrated that DNA-mediated crystallization can be used to provide a simplified model for understanding crystal growth in which adlayers display direct bonding interactions with a substrate but lack the periodic potential inherent to atomic systems. The interfacial energy between a thin-film superlattice and the substrate, and consequently the orientation, can be controlled by appropriate choice of DNA interconnects. This work creates a new design rule for these structures: the orientations of such programmable crystalline thin-films will be dictated by the crystal planes that maximize complementary interactions with the substrate. This strategy could easily be applied to other binary crystal symmetries, or to surfaces patterned with DNA for lithographically templating nanoparticle superlattices. Furthermore, the ideas set forth in this work suggest a route for growing single-crystal nanoparticle superlattices by controlling epitaxial processes. We have also shown that the number of layers in SNA-NP superlattice thin films can be controlled, which is useful for determining thickness-dependent material properties. The additional level of control extended to the system through direct substrate–adlayer bonding interactions will be important for the

Experimental Section

development of materials that take advantage of the periodicity of the inorganic core material, such as optical metamaterials, photonic bandgap materials, and magnetic storage media.

60.4

Experimental Section

Nanoparticles (10-nm-diameter AuNPs) were functionalized with DNA according to literature procedures [10]. Au-coated silicon wafers (8 nm Au, 2 nm Cr adhesion layer) were diced into 7.5 × 15 mm chips and functionalized overnight at various molar ratios of B:A DNA from a 2 mM total concentration of thiolated DNA in 1M phosphate buffer saline (PBS, 10 mM phosphate, 1M NaCl). After washing 3 × in 0.5M PBS, linker DNA (0.5 μM total concentration) was hybridized to the DNA substrates at the same molar ratio as the thiolated sequences in 0.5M PBS by slowly cooling from 75°C to room temperature over 2 h. SNA-AuNP films were grown by sequential immersion in B- and A-type SNA-AuNPs in 0.5 M PBS for 1–1.5 h per half-cycle at various temperatures. After each half-cycle, unbound SNA-AuNPs were carefully removed in 0.5M PBS (5×). X-ray scattering. All GISAXS experiments were conducted at the 12ID-B station at the Advanced Photon Source (APS) at Argonne National Laboratory using collimated 12 keV (1.033 Å) X-rays.

Acknowledgments

C.A.M. acknowledges support from AFOSR Awards FA9550-1110275 and FA9550-09-1-0294, AOARD Award FA2386-10-1-4065, Department of the Navy/ONR Award N00014-11-1-0729, DoD/ NSSEFF/NPS Awards N00244-09-1-0012 and N00244-09-1-0071 and the non-Equilibrium Energy Research Center (NERC), an Energy Frontier Research Center funded by DoE/Office of Science/Office of Basic Energy Sciences Award DE-SC0000989. D.J.E. and E.A. acknowledge support from the DoD, Air Force Office of Scientific Research, National Defense Science and Engineering Graduate (NDSEG) Fellowship, 32 CFR 168a. M.R.J. acknowledges a Graduate

1139

1140

Stepwise Evolution of DNA-Programmable Nanoparticle Superlattices

Research Fellowship from the NSF. Portions of this work were carried out at beamline 12-ID-B at the Advance Photon Source (APS). Use of the APS, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science by Argonne National Laboratory, was supported by the U.S. DOE under contract no. DEAC02-06CH11357. Electron microscopy was performed at the EPIC facility of the NUANCE Center at Northwestern University.

References

1. Cheng, Z., Russel, W. and Chaikin, P. (1999). Nature, 401, 893–895.

2. Shevchenko, E., Talapin, D., Kotov, N., O’Brien, S. and Murray, C. (2006). Nature, 439, 55–59.

3. Wong, S., Kitaev, V. and Ozin, G. A. (2003). J. Am. Chem. Soc., 125, 15589–15598. 4. Ganapathy, R., Buckley, M. R., Gerbode, S. J. and Cohen, I. (2010). Science, 327, 445–448.

5. Henzie, J., Grünwald, M., Widmer-Cooper, A., Geissler, P. and Yang, P. (2012). Nat. Mater., 11, 131–137.

6. Leunissen, M., Christova, C., Hynninen, A.-P., Royall, C., Campbell, A., Imhof, A., Dijkstra, M., van Roij, R. and van Blaaderen, A. (2005). Nature, 437, 235–240.

7. Kalsin, A. M., Fialkowski, M., Paszewski, M., Smoukov, S. K., Bishop, K. J. M. and Grzybowski, B. A. (2006). Science, 312, 420–424.

8. Tang, Z., Zhang, Z., Wang, Y., Glotzer, S. C. and Kotov, N. A. (2006). Science, 314, 274–278.

9. Mirkin, C. A., Letsinger, R., Mucic, R. and Storhoff, J. (1996). Nature, 382, 607–609.

10. Park, S. Y., Lytton-Jean, A., Lee, B., Weigand, S., Schatz, G. C. and Mirkin, C. A. (2008). Nature, 451, 553–556.

11. Nykypanchuk, D., Maye, M., van der Lelie, D. and Gang, O. (2008). Nature, 451, 549–552.

12. Chen, C.-L. and Rosi, N. L. (2010). Angew. Chem., 122, 1968–1986; Angew. Chem. Int. Ed., 49, 1924–1942.

13. Collier, C., Saykally, R., Shiang, J., Henrichs, S. and Heath, J. (1997). Science, 277, 1978–1981.

14. Lazarides, A. and Schatz, G. C. (2000). J. Phys. Chem. B, 104, 460–467.

References

15. Vlasov, Y. A., Bo, X.-Z., Sturm, J. C. and Norris, D. J. (2001). Nature, 414, 289–293. 16. Alaeian, H. and Dionne, J. A. (2012). Opt. Express, 20, 15781–15796.

17. Zhu, Z., Meng, H., Liu, W., Liu, X., Gong, J., Qiu, X., Jiang, L., Wang, D. and Tang, Z. (2011). Angew. Chem., 123, 1631–1634; Angew. Chem. Int. Ed., 50, 1593–1596.

18. Urban, J. J., Talapin, D. V., Shevchenko, E. V., Kagan, C. R. and Murray, C. B. (2007). Nat. Mater., 6, 115–121.

19. Sun, S., Murray, C., Weller, D., Folks, L. and Moser, A. (2000). Science, 287, 1989–1992. 20. Khizroev, S., Hijazi, Y., Amos, N., Chomko, R. and Litvinov, D. (2006). J. Appl. Phys., 100, 063907.

21. (a) Cutler, J. I., Auyeung, E. and Mirkin, C. A. (2012). J. Am. Chem. Soc., 134, 1376–1391; (b) Macfarlane, R. J., O’Brien, M. N., Petrosko, S. H. and Mirkin, C. A. (2013). Angew. Chem., 125, doi:10.1002/ange.201209336; Angew. Chem. Int. Ed., 51, doi:10.1002/anie.201209336.

22. Macfarlane, R. J. Jones, M. R., Senesi, A. J., Young, K. L., Lee, B., Wu, J. and Mirkin, C. A. (2010). Angew. Chem., 122, 4693–4696; Angew. Chem. Int. Ed., 49, 4589–4592.

23. Jones, M. R., Macfarlane, R. J., Lee, B., Zhang, J., Young, K., Senesi, A. J. and Mirkin, C. A. (2010). Nat. Mater., 9, 913–917.

24. Leunissen, M. E., Dreyfus, R., Sha, R., Seeman, N. C. and Chaikin, P. M. (2010). J. Am. Chem. Soc., 132, 1903–1913. 25. Macfarlane, R. J., Lee, B., Jones, M. R., Harris, N., Schatz, G. C. and Mirkin, C. A. (2011). Science, 334, 204–208.

26. Noh, H., Hung, A. and Cha, J. (2011). Small, 7, 3021–3025.

27. Auyeung, E., Cutler, J. I., Macfarlane, R. J., Jones, M. R., Wu, J., Liu, G., Zhang, K., Osberg, K. D. and Mirkin, C. A. (2012). Nat. Nanotechnol., 7, 24–28.

28. Taton, T. A., Mucic, R., Mirkin, C. A. and Letsinger, R. (2000). J. Am. Chem. Soc., 122, 6305–6306.

29. Niemeyer, C. M., Ceyhan, B., Noyong, M. and Simon, U. (2003). Biochem. Biophys. Res. Commun., 311, 995–999.

30. Zou, B., Ceyhan, B., Simon, U. and Niemeyer, C. M. (2005). Adv. Mater., 17, 1643–1647. 31. Lee, B., Park, I., Yoon, J., Park, S., Kim, J., Kim, K.-W., Chang, T. Ree, M. (2005). Macromolecules, 38, 4311–4323.

1141

1142

Stepwise Evolution of DNA-Programmable Nanoparticle Superlattices

32. Auyeung, E., Macfarlane, R. J., Choi, C. H. J., Cutler, J. I. and Mirkin, C. A. (2012). Adv. Mater., 24, 5181–5186.

33. Binks, B. P. and Horozov, T. S. (2006). Colloidal Particles at Liquid Interfaces (Cambridge University Press, New York). 34. Thompson, C. V. (1990). Annu. Rev. Mater. Sci., 20, 245–268.

Chapter 61

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates*

Sondra L. Hellstrom,a Youngeun Kim,b James S. Fakonas,a Andrew J. Senesi,c Robert J. Macfarlane,c Chad A. Mirkin,b,c and Harry A. Atwatera aKavli Nanoscience Institute, California Institute of Technology, Pasadena, CA 91125, USA bDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA cDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA dX-ray Science Division, Advanced Photon Source, Argonne National Laboratory, 9700 S. Cass Ave., Argonne, IL 60439, USA [email protected]

*Reprinted with permission from Hellstrom, S. L., Kim, Y., Fakonas, J. S., Senesi, A. J., Macfarlane, R. J., Mirkin, C. A. and Atwater, H. A. (2013). Epitaxial growth of DNAassembled nanoparticle superlattices on patterned substrates, Nano Lett. 13, 6084−6090. Copyright (2013) American Chemical Society. Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1144

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates

DNA-functionalized nanoparticles, including plasmonic nanoparticles, can be assembled into a wide range of crystalline arrays via synthetically programmable DNA hybridization interactions. Here we demonstrate that such assemblies can be grown epitaxially on lithographically patterned templates, eliminating grain boundaries and enabling fine control over orientation and size of assemblies up to thousands of square micrometers. We also demonstrate that this epitaxial growth allows for orientational control, systematic introduction of strain, and designed defects, which extend the range of structures that can be made using superlattice assembly. Ultimately, this will open the door to integrating self-assembled plasmonic nanoparticle materials into on-chip optical or optoelectronic platforms.

61.1

Introduction

The assembly of well-ordered, 3D, isotropic, optically active materials has been an important goal in the photonic crystal and metamaterials communities for many years. Many such materials are predicted to have interesting optical properties, including tunable permittivities [1], complex resonances [2], and negative refractive indices [3]; these properties are determined mainly by the physical structures of the assemblies, which must be designed and controlled on the nanometer scale. A wide range of techniques exist for constructing these materials [4, 5], including colloidal sedimentation [6, 7], Langmuir−Blodgett trough assembly [8], drying-based [9], depletion-based [10], and matrix-filling [11] strategies. Unfortunately, such techniques lack independent control over the shape, size, and composition of the plasmonic components and the lattice structure. Moreover, the densities of defects are generally high, and the crystal orientations or lattice parameters cannot be easily or widely controlled. Plasmonic nanoparticles have been assembled a variety of different ways using DNA-based approaches [12]. Indeed, DNA and

Introduction

anisotropically functionalized particles have been used to create clusters that have been described as “small plasmonic molecules” [13−18]. DNA origami has also been used to form scaffolds upon which nanoparticles can organize [19−23]. Solutions of these materials have little long-range order, and individual structures are difficult to measure [24]. Nonetheless these materials have been found to exhibit interesting optical properties, including Fano-like resonances [25] and giant circular dichroism [23]. Alternatively, DNA and densely loaded, isotropically functionalized particles have been used to make superlattices with exquisite control over crystal symmetry and lattice parameters [26−29]. Spherical nanoparticles functionalized in this way are referred to as spherical nucleic acids (SNAs) [30] and act as “programmable atom equivalents (PAEs)” [31, 32], in which the nanoparticles take the place of atoms, and oligonucleotides take the place of chemical bonds. By adjusting different components of PAEs, hundreds of different crystal structures with both static [29] and dynamic [33, 34] tunability have already been made, and a series of design rules has been established to explain and predict solution-phase crystal stability [29]. Typical PAE superlattices have grain sizes of a few hundred nanometers up to 1−2 μm and form polycrystalline aggregates with uncontrollable edges, sizes, and orientations. To construct and probe a useful optical material, a single crystal on the order of tens of micrometers or larger must be fabricated, with a controllable orientation relative to a substrate and to other optical components. Polycrystalline body-centered cubic (bcc) (100) and (110) thin films have already been grown on unpatterned DNA-modified substrates [35], opening the door to the formation of crystals with a fixed location and orientation. Although thin-film assemblies grown in this manner with appropriate annealing display a uniaxial texture (preferential orientation normal to the substrate), they remain polycrystalline with domain sizes similar to those of crystals in solution.

1145

1146

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates

Figure 61.1 (a) Template fabrication process. (b) Schematic illustrating the DNA binding scheme between template and nanoparticles or between nanoparticles of opposite types (not to scale). (c) Schematic of an ideal bcc (100) crystal epitaxially grown on a surface. Gold surfaces functionalized with DNA containing A-type single-stranded ends are shown in red, and gold surfaces with complementary B-type single-stranded ends are shown in blue. Full color illustrates the portion discussed in this work. For clarity, the DNA itself is not included. (d) Submonolayer growth on a bcc (100) template, imaged via SEM, with nanoparticles in false color. Submonolayer growth is shown to better illustrate three types of binding modes: center-bound nanoparticles are colored green, edge-bound nanoparticles are colored yellow, and probable cornerbound nanoparticles are colored orange. The associated schematic illustrates these corner, edge, and center-bound particles. In the schematic, template sites presenting A-type sticky ends are shown in red, and deposited particles presenting B-type sticky ends are shown in blue.

Epitaxial growth, that is, the use of a crystalline substrate to control order and orientation in a growing crystal, is typically a very effective method of transferring large-scale order to materials that would otherwise display amorphous or polycrystalline material

Methods and Results

growth mechanisms [36]. Nonetheless, prepatterned surfaces have only been used in a limited capacity to direct assembly of DNAmodified nanoparticles [37, 38]. In this work, in analogy to methods in conventional thin-film growth, we have developed a technique to epitaxially grow thin-film PAE single-crystal superlattices. Electronbeam lithography was used to pattern gold features on silicon; after DNA functionalization, the resulting substrate was roughly equivalent to the first layer of a PAE crystal. We subsequently used this as a template for solution-phase homoepitaxy. Adlayer growth on single-domain 100 μm2 templates was examined with regards to defect formation, lattice mismatch, and the role of the templated crystal plane on superlattice orientation. We demonstrated control over crystal size, orientation, and location on a substrate, and we also exerted control over defect density and type. This technique will ultimately enable sophisticated optical measurements, assist elucidation of optical structure−property relationships, and potentially extend the range of properties which these materials can exhibit.

61.2

Methods and Results

Epitaxial growth of bcc superlattices was effected in a manner similar to that used for preparing polycrystalline thin films of PAEs [35], but instead of using an amorphous gold substrate, a nanofabricated one resembling a superlattice crystal plane was employed. Specifically, a template was designed and fabricated using conventional electronbeam lithography in poly(methylmethacrylate) (PMMA; Fig. 61.1a). After exposure and development, 3 nm Ti or Cr and 12 nm Au were sequentially deposited, and the PMMA was lifted off. The fabricated template was then functionalized with 3¢-propylthiol-modified DNA, and oligonucleotide linkers with seven base long “sticky ends” were introduced to enable hybridization to PAEs. In this work, we used PAEs with a 30 nm diameter Au core designed to assemble into a bcc crystal with a 62 nm lattice parameter, and we focused on the formation and structure of the first superlattice monolayer. We left our templates exposed to PAE in solution for a full 24 h, to promote the formation of a thermodynamically favorable crystal state [39]. Lattices with a bcc crystal structure are ideal for use in surfacesupported stepwise crystal growth, because they are composed of

1147

1148

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates

two complementary distinct types (A and B) of PAE (Fig. 61.1b,c). The bcc crystal is grown by exposing the surface alternately to A-type and B-type nanoparticles, suppressing crystal nucleation in solution. Since the nanoparticles exposed to the surface interact with one another primarily by steric hindrance and electrostatic repulsion, the attachment of the first adlayer of nanoparticles to the epitaxial substrate is best understood as a process of site-specific adsorption with no driving force for clustering or island growth. We first studied the adsorption of a monolayer of PAEs on templates designed to be equivalent to a bcc (100) plane. For a submonolayer film, we observed three largely distinct adsorption patterns, corresponding to PAE attachment to either one (corner-bound), two (edge-bound), or four (center-bound) DNAfunctionalized lithographic features (Figure 61.1d). This behavior is most easily observed for submonolayer films but also holds true for full monolayer structures. Since the patterned features on the surface are the equivalents of A-type particles in a bcc lattice, center-bound attachment of PAEs corresponds to the continuation of the lattice. Therefore, in this context edge-bound or corner-bound PAEs constitute defects in the growing crystal. Notably, in contrast to other reports [40], for substrates composed of Cr films or of Si with native oxide we only rarely found evidence of nonspecific binding of nanoparticles to the substrate surface, even without additional surface functionalization. For well-ordered lattice-matched monolayer (100) films, vacancies remained the most common kind of defect (Fig. 61.2a). This could have a number of causes, but we believe that two in particular dominate. First, thiol modification of gold surfaces is known to be very dependent on experimental conditions [41], and we found that the history of the substrate surface prior to modification (type of cleaning and storage) had a strong impact on particle coverage during monolayer growth. This implies that high-density functionalization of the gold template with DNA is important for growing wellordered monolayers and that variations in oligonucleotide density on the template can lead to vacancies. Second, there is a window of crystallization, temperatures below which nanoparticles cannot diffuse on the substrate surface to promote order, and above which nanoparticle desorption substantially decreases particle coverage.

Methods and Results (a)

(b)

(c)

(d)

Figure 61.2  (a) Percent of the specified type of each site occupied by particles,  as a function of lattice strain. Note that a particle bound to a given site typically blocks binding to adjacent sites of different types. (b) Percent of the total number  of  occupied  sites  that  are  occupied  in  the  specified  manner  as  a  function  of  lattice strain, illustrating the relative prevalence of each type of binding in a monolayer of particles. (c) x- and z-plane slices of a simulation used to calculate the degree of DNA hybridization between particles in different configurations  (an edge-bound particle is demonstrated in these images). Dashed lines in each scheme illustrate the location of the plane represented in the orthogonal image. Blue space represents buffer, green space represents DNA linker, yellow space  represents gold, and red space represents that occupied by a single-stranded DNA sticky end. The amount of DNA hybridization is taken to be proportional to the volume occupied by intersecting red shells. (d) Estimated probabilities of site occupancy fit to (1), with fit parameter ε = −2.8 meV (0.064 kcal/mol) per  hybridized sticky end. Qualitative agreement—and the prediction of which site is thermodynamically favorable—is very good.

DNA hybridization occurring at the template surface places a PAE in a lower-energy state compared to a free nanoparticle in solution. The energy benefit to attachment in a given configuration can be calculated using a variant of the “complementary contact model” established in previous work [29]. This model states that the stability of a given arrangement of particles is proportional to the

1149

1150

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates

amount of contact between complementary sticky ends on adjacent objects, that is, that the degree of hybridization can be calculated geometrically. In this work we assume that the number of DNA sticky ends that hybridize in a multiparticle system is proportional to the overlapping volume of space between those that contain interpenetrating complementary sticky ends (for simplicity, the density s of hybridized sticky ends is assumed to be constant in this volume). Every gold object (yellow) in a functionalized system is modeled as having a halo of sticky ends (red) surrounding it at a distance corresponding to the approximate length of the DNA linker (green) (Fig. 61.2c). We assume that volumes containing DNA are penetrable and volumes containing metal or silicon are not. Then, for a particle in a given configuration, the reduction in energy compared to a free particle is ΔE = εsV, where ε is the energy gained by forming a single DNA duplex between two sticky ends and V is the overlapping sticky-end volume. Arguably the simplest method of comparing this estimate of the energies of different sites with our experimental observations in Fig. 61.2a,b is with a Boltzmann distribution function. Here, this means that the relative probability of finding a center-bound particle compared to an edge- or corner-bound particle is: pcenter =

Â

e - e sVcenter /kBT

V= corner, edge, center

e - e sV /kBT

(61.1)

We calculated overlap volumes for different configurations numerically based on geometry and fit the results to experimental probabilities using Eq. 61.1 with ε as a fit parameter (Fig. 61.2d). T and s can be estimated from experimental conditions (here, we use 7.6 × 10−3 sticky ends·nm−3, or approximately 300 sticky ends per 30 nm diameter Au particle, a 4 nm halo [42], and a temperature of 296 K). Qualitatively, the results match well; in particular, the dominant particle attachment site for a given geometry is always readily predictable. Quantitatively, our fit ε = −2.8 meV per hybridized sticky end is quite small compared with measurements of free DNA hybridization in solution [43−45]. This may be an indication that steric hindrance and electrostatic repulsion substantially counterbalance DNA hybridization in this system or that there are significant sources of disorder other than temperature.

Methods and Results

Experimentally, with crystallization at 23°C, 98% of the particles on our best single lattice-matched 100 μm2 template were centerbound, clearly demonstrating the viability of largescale crystal growth using this technique. However, after counting over 5000 attached particles across different substrates, we found that on average 86% of the particles attached to a lattice-matched bcc (100) surface were center-bound (Fig. 61.2b), 12% were edge-bound, and about 2% were corner-bound. We hypothesize that the best method to improve this fraction of center-bound states is to increase the overlap volume of a particle in that state relative to an edge-bound state, by adjusting the geometry of the template. Alternatively, given that for the lattice-matched case the center-bound state is already the one with the greatest overlap volume, a greater percentage of center-bound states could be achieved by increasing ε, or decreasing the crystallization temperature, to the extent possible without inducing kinetic crystallization. In this regard, a process of slow cooling may be helpful in allowing the crystal to most easily settle into its thermodynamically favored state.

Figure 61.3 SEM images of monolayer bcc (100) crystals grown on epitaxial templates  with  −10%,  −5%,  +5%,  +10%,  and  +20%  strain,  respectively,  as  indicated, and 2D fast Fourier transforms of the images, center point removed and cropped to feature lower frequencies. The amorphous matter in the SEM images is left over silica from a sol-gel embedding process required to stabilize  the superlattices for drying and imaging with SEM.

1151

1152

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates

To test the flexibility of PAE thin films assembled on lithographically patterned substrates, we investigated growth of the originally designed lattice with a lattice parameter of 62 nm on lithographic features of variable distances (57 nm, 59 nm, 65 nm, 68 nm, and 76 nm) (Fig. 61.3). The amorphous matter in the SEM images is leftover silica from a sol-gel embedding process required to stabilize the superlattices for drying and imaging with SEM [46]. The most notable defects in the compressed lattices were vacancies. In contrast, strained lattices showed a tendency toward increasing numbers of edge-bound nanoparticles. Up to 10% strain, the first crystal layer grew pseudomorphically with the lattice. However, in the 20% strain case, so many nanoparticles became edgebound that chains of edge-bound particles became the dominant growth mode. This transition is predictable using calculations of overlap volume (Fig. 61.2d). Around 20% strain, the adlayer began to resemble, were the crystal continued further into three dimensions, a layer consistent with a uniaxially strained bcc (110) lattice—or a bcc (100) lattice with an a/√2 lattice parameter and every other center atom missing—rather than a biaxially strained bcc (100). While this result has promise for enabling growth of crystals that cannot be formed in solution or on unpatterned surfaces, the real bulk crystal structure obtained by growth of additional layers on these substrates has yet to be determined. Using lithographically patterned features allows predictable and programmable control of PAE attachment onto substrates. To test this control, we studied PAE attachment onto templates representing different crystal orientations. Figure 61.4 shows results of nanoparticle monolayers grown on templates designed to reproduce lattice-matched bcc (100), bcc (110), and bcc (111) crystal planes, compared with a monolayer grown on an unpatterned surface shown as reference. Note that, while the real bcc (110) crystal plane contains both A-type and B-type particles, the template must be fabricated missing the B-type particle, since the surface has only been functionalized with one type of DNA. In all orientations, a certain defect density notwithstanding, the monolayer adopted a single-domain crystalline structure unique to the patterned surface. This is visible in the 2D fast Fourier transforms of the SEM images, which illustrate that the real crystals maintain the predesigned lattice parameters to a high degree.

Methods and Results

Figure 61.4 Nanoparticle monolayers attached to bcc (100), (110), and (111) epitaxial templates. From left to right, the images show the templates, a 2D FFT  of the template image, the adlayer grown on top of the template, and a 2D FFT of the final image. All FFTs, which were provided to illustrate the degree of order  in the films, have their center points removed and are cropped to emphasize  lower frequencies. Schematics illustrate the dominant observed growth pattern of PAEs on the templates, with the template in red and the adlayer particles in blue. The amorphous matter in the SEM images is leftover silica from a solgel embedding process required to stabilize the superlattices for drying and imaging with SEM.

The nanoparticles attached to all of the templates were physically located in the x and y directions (parallel to the substrate surface), in the locations corresponding to the next layer of particles in bulk 3D bcc (100), (110), and (111) crystals, which makes further growth in these orientations promising. It is important to note that, while technically speaking the atoms in a bcc lattice are indistinguishable, in the case of the bcc (110) template, half of the nanoparticles in the grown adlayer had the opposite DNA functionalization relative to those in the predicted bulk structure. While the bulk bcc (100) crystal is naturally organized as a layer of A-type particles followed by a layer of B-type particles, this

1153

1154

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates

is not the case for bulk bcc (110). Each layer of particles in a bulk bcc (110) crystal contains a mixed composition of A- and B-type particles, one of which is unavailable when only one monolayer is grown. The atoms omitted from the templated crystal plane were not filled in by the grown monolayer, presumably because they were blocked by the silicon substrate. In the case of an unpatterned substrate surface, bulk bcc (110) structures were obtained after multilayer growth and annealing enabled them to relax into their thermodynamically most favorable state. Further experiments with multilayer growth and annealing will tell whether or not such reorganization occurs on epitaxial substrates as well. It is very interesting, though, that even with 50% incorrect functionalization, and without annealing, the particles nevertheless adopted in x and y the exact symmetry suggested by the (110) template. A related effect was observed in the bcc (111) system. The bulk bcc (111) crystal is, like the bcc (100), organized as a layer of A-type particles followed by a layer of B-type particles. However, each layer has a comparatively low particle density, which means that substantial potential binding area remains available to particles in solution after a layer is complete. We observed binding of B-type particles to template sites as was consistent with (111) bulk growth, but in addition binding of B-type particles to template sites where A- or B-type particles would normally bind at the same x and y but at somewhat different z. This occurred because, at the time of growth, there were no other particles present to block the attachment of the B-type particles to the template. Again, multilayer growth and annealing may reorganize the structure into a true 3D bcc (111) bulk crystal, or alternatively a new 3D strained crystal might be formed. In either case, as with the bcc (110), it is interesting that even with this additional binding the particles adopt in x and y the exact symmetry suggested by the bcc (111) template. While (100) and (110) orientations can be grown on planar DNA-functionalized substrates through judicious choice of DNA interconnects, higher energy facets such as the (111) orientation are not thermodynamically accessible. Patterning the substrate to appear like (111) seems to force growth consistent with this orientation, albeit with the caveats mentioned above and higher defect density than the other two orientations. This observation suggests that superlattices with novel symmetries not observed in

Conclusion

bulk superlattice formation, for example chiral or quasicrystals, may be possible as thin-film superlattices on templated substrates.

61.3

Conclusion

In conclusion, we studied epitaxial growth of PAE superlattices and have vastly extended grain sizes in PAE thin films to thousands of square micrometers, limited only by the size of the lithographic template. Further, on this scale we maintained orientation and crystal size control over PAE thin films, which was not possible when crystals were assembled in solution or on unpatterned surfaces. For example, we were able to fabricate strained lattices and higher surface energy orientations by carefully controlling the lattice mismatch and type of lithographically defined template. In addition, we even preprogrammed defects at specific points within the superlattices. We anticipate that this process could be extended to other particle sizes, lattice parameters, and crystal symmetries. Importantly, this marriage of top-down and bottom-up assembly techniques utilizes the programmability afforded by DNA and the precision and alignment control afforded via lithography to compensate for each techniques’ respective drawbacks. We envision that the combination of the two methodologies will have a major impact in both assembling crystal structures not achievable via conventional DNA-programmed assembly or lithography alone, such as lattices with tetragonal or more complex unit cells, with lattice constants larger or smaller than can be currently achieved, and via heteroepitaxial growth involving, for example, different particle shapes or compositions. As a result, we envision that this technique will allow for previously unattainable levels of control over the plasmonic, optical, magnetic, catalytic, or other emergent properties that arise as a result of assembling nanoparticle superlattices.

Acknowledgments

S.L.H. thanks the Caltech KNI staff for training and equipment expertise; Tiffany Kimoto and Jennifer Blankenship for excellent support; and Dr. Matt Sheldon for technical expertise. H.A.A. and C.A.M. are grateful for funding through the Bioprogrammable

1155

1156

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates

One-, Two-, and Three-Dimensional Materials no. FA9550-11-1-0275 MURI. C.A.M. also acknowledges support from the AFOSR (FA955012-1-0280) and the DOE (611-8289300-60024682-01) through the NU Non-Equilibrium Research Center. Use of the Advanced Photon Source at Argonne National Laboratory was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.

References

1. Ciattoni, A., Marinelli, A., Rizza, C. and Palange, E. (2013). Appl. Phys. B: Laser Opt., 110, 23−36.

2. Fan, J. A., Wu, C., Bao, J., Bardhan, R., Halas, N. J., Manoharan, V., Nordlander, P., Shvets, G. and Capasso, F. (2010). Science, 328, 1135−1138. 3. Rockstuhl, C., Lederer, F., Etrich, C., Pertsch, T. and Scharf, T. (2007). Phys. Rev. Lett., 99, 017401.

4. Yao, J., Le, A.-P., Gray, S. K., Moore, J. S., Rogers, J. A. and Nuzzo, R. G. (2010). Adv. Mater., 22, 1102−1110. 5. Hur, K., Francescanto, Y., Giannini, V., Maier, S. A., Hennig, R. G. and Wiesner, H. (2011). Angew. Chem. Int. Ed., 50, 11985−11989.

6. Tao, A. R., Ceperley, D. P., Sinsermsuksakul, P., Neureuther, A. R. and Yang, P. (2008). Nano Lett., 8, 4033−4038. 7. Lu, Y., Yin, Y., Li, Z.-Y. and Xia, Y. (2002). Nano Lett., 2, 785−788.

8. Tao, A., Sinsermsuksakul, P. and Yang, P. (2007). Nat. Nanotechnol., 2, 435−440.

9. Talapin, D. V., Shevchenko, E. V., Bodnarchuk, M. I., Ye, X., Chen, J. and Murray, C. B. (2009). Nature, 461, 964−967.

10. Henzie, J., Grünwald, M., Widmer-Cooper, A., Geissler, P. and Yang, P. (2012). Nat. Mater., 11, 131−137. 11. Vignolini, S., Yufa, N. A., Cunha, P. S., Guldin, S., Rushkin, I., Stefik, M., Hur, K., Wiesner, U., Baumberg, J. J. and Steiner, U. (2012). Adv. Mater., 24, OP23−OP27.

12. Tan, S. J., Campolongo, M. J., Luo, D. and Cheng, W. (2011). Nat. Nanotechnol., 6, 268−276. 13. Alivisatos, A. P., Johnsson, K. P., Peng, X., Wilson, T. E., Loweth, C. J., Bruchez, M. P., Jr. and Schultz, P. G. (1996). Nature, 382, 609−611.

References

14. Loweth, C. J., Caldwell, W. B., Peng, X., Alivisatos, A. P. and Schultz, P. G. (1999). Angew. Chem. Int. Ed., 38, 1808−1812.

15. Mastroianni, A. J., Claridge, S. A. and Alivisatos, A. P. (2009). J. Am. Chem. Soc., 131, 8455−8459.

16. Maye, M. M., Nykypanchuk, D., Cuisinier, M., van der Lelie, D. and Gang, O. (2009). Nat. Mater., 8, 388−391.

17. Xu, X., Rosi, N. L., Wang, Y., Huo, F. and Mirkin, C. A. (2006). J. Am. Chem. Soc., 128, 9286−9287.

18. Pal, S., Sharma, J., Yan, H. and Liu, Y. (2009). Chem. Commun., 6059−6061. 19. Pal, S., Deng, Z., Ding, B., Yan, H. and Liu, Y. (2010). Angew. Chem. Int. Ed., 49, 2700−2704. 20. Ding, B., Deng, Z., Yan, H., Cabrini, S., Zuckermann, R. N. and Bokor, J. (2010). J. Am. Chem. Soc., 132, 3248−3249.

21. Aldaye, F. A. and Sleiman, H. F. (2007). J. Am. Chem. Soc., 129, 4130−4131. 22. Aldaye, F. A. and Sleiman, H. F. (2006). Angew. Chem. Int. Ed., 45, 2204−2209.

23. Kuzyk, A., Schreiber, R., Fan, Z., Pardatscher, G., Roller, E.-M., Högele, A., Simmel, F. C., Govorov, A. O. and Liedl, T. (2012). Nature, 483, 311−314. 24. Sebba, D. S., Mock, J. J., Smith, D. R., LaBean, T. H. and Lazarides, A. A. (2008). Nano Lett., 8, 1803−1808.

25. Fan, J. A., He, Y., Bao, K., Wu, C., Bao, J., Schade, N. B., Manoharan, V. N., Schvets, G., Nordlander, P., Liu, D. R. and Capasso, F. (2011). Nano Lett., 11, 4859−4864. 26. Mirkin, C. A., Letsinger, R. L., Mucic, R. C. and Storhoff, J. J. (1996). Nature, 382, 607−609.

27. Park, S. Y., Lytton-Jean, A. K. R., Lee, B., Weigand, S., Schatz, G. C. and Mirkin, C. A. (2008). Nature, 451, 553−556.

28. Nykypanchuk, D., Maye, M. M., van der Lelie, D. and Gang, O. (2008). Nature, 451, 549−552.

29. Macfarlane, R. J., Lee, B., Jones, M. R., Harris, N., Schatz, G. C. and Mirkin, C. A. (2011). Science, 334, 204−208.

30. Cutler, J. I., Auyeung, E. and Mirkin, C. A. (2012). J. Am. Chem. Soc., 134, 1376−1391.

31. Macfarlane, R. J., O’Brien, M. N., Petrosko, S. H. and Mirkin, C. A. (2013). Angew. Chem. Int. Ed., 52, 5688−5698.

32. Zhang, C., Macfarlane, R. J., Young, K. L., Choi, C. H. J., Hao, L., Auyeung, E., Liu, G., Zhou, X. and Mirkin, C. A. (2013). Nat. Mater., 12, 741−746.

1157

1158

Epitaxial Growth of DNA-Assembled Nanoparticle Superlattices on Patterned Substrates

33. Maye, M. M., Kumara, M. T., Nykypanchuk, D., Sherman, W. B. and Gang, O. (2010). Nat. Nanotechnol., 5, 116−120.

34. Kim, Y., Macfarlane, R. J. and Mirkin, C. A. (2013). J. Am. Chem. Soc., 135, 10342−10345. 35. Senesi, A. J., Eichelsdoerfer, D. J., Macfarlane, R. J., Jones, M. R., Auyeung, E., Lee, B. and Mirkin, C. A. (2013). Angew. Chem. Int. Ed., 52, 6624.

36. Venables, J. A. (2000). Introduction to Surface and Thin Film Processes (Cambridge University Press, New York). 37. Kershner, R. J., Bozano, L. D., Micheel, C. M., Hung, A. M., Fornof, A. R., Cha, J. N., Rettner, C. T., Bersani, M., Frommer, J., Rothemund, P. W. K. and Wallraff, G. M. (2009). Nat. Nanotechnol., 4, 557−561.

38. Lalander, C. H., Zheng, Y., Dhuey, S., Cabrini, S. and Bach, U. (2010). ACS Nano, 4, 6153−6161. 39. Macfarlane, R. J., Lee, B., Hill, H. D., Senesi, A. J., Seifert, S. and Mirkin, C. A. (2009). Proc. Natl. Acad. Sci. U.S.A., 106, 10493−10498.

40. Kannan, B., Kulkarni, R. P. and Majumdar, A. (2004). Nano Lett., 4, 1521−1524.

41. Herne, T. M. and Tarlov, M. J. (1997). J. Am. Chem. Soc., 119, 8916−8920.

42. Chi, Q., Wang, G. and Jiang, J. (2013). Physica A, 392, 1072−1079.

43. Holbrook, J. A., Capp, M. W., Saecker, R. M. and Record, M. T., Jr. (1999). Biochemistry, 38, 8409−8422. 44. Mikulecky, P. and Feig, A. L. (2006). Biochemistry, 45, 604−616.

45. Biancaniello, P. L., Kim, A. J. and Crocker, J. C. (2005). Phys. Rev. Lett., 94, 058302. 46. Auyeung, E., Macfarlane, R. J., Choi, C. H. J., Cutler, J. I. and Mirkin, C. A. (2012). Adv. Mater., 24, 5181−5186.

Chapter 62

Epitaxy: Programmable Atom Equivalents versus Atoms*

Mary X. Wang,a,b Soyoung E. Seo,b,c Paul A. Gabrys,e Dagny Fleischman,f Byeongdu Lee,g Youngeun Kim,b,d Harry A. Atwater,f Robert J. Macfarlane,e and Chad A. Mirkina,b,c,d aDepartment of Chemical and Biological Engineering, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA bInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA dDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA eDepartment of Materials Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA fThomas J. Watson Laboratories of Applied Physics, California Institute of Technology, 1200 East California Boulevard, Pasadena, CA 91125, USA gX-ray Science Division, Advanced Photon Source, Argonne National Laboratory, 9700 S. Cass Ave., Argonne, IL 60439, USA [email protected]; [email protected]

Mary X. Wang and Soyoung E. Seo contributed equally to this work.

*Reprinted with permission from Wang, M. X., Seo, S. E., Gabrys, P. A., Fleischman, D., Lee, B., Kim, Y., Atwater, H. A., Macfarlane, R. J. and Mirkin, C. A. (2017). Epitaxy: programmable atom equivalents versus atoms, ACS Nano 11, 180−185. . Further permissions related to the material excerpted should be directed to the ACS. Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1160

Epitaxy: Programmable Atom Equivalents versus Atoms

The programmability of DNA makes it an attractive structuredirecting ligand for the assembly of nanoparticle (NP) superlattices in a manner that mimics many aspects of atomic crystallization. However, the synthesis of multilayer single crystals of defined size remains a challenge. Though previous studies considered lattice mismatch as the major limiting factor for multilayer assembly, thin-film growth depends on many interlinked variables. Here, a more comprehensive approach is taken to study fundamental elements, such as the growth temperature and the thermodynamics of interfacial energetics, to achieve epitaxial growth of NP thin films. Both surface morphology and internal thin-film structure are examined to provide an understanding of particle attachment and reorganization during growth. Under equilibrium conditions, single-crystalline, multilayer thin films can be synthesized over 500 × 500 μm2 areas on lithographically patterned templates, whereas deposition under kinetic conditions leads to the rapid growth of glassy films. Importantly, these superlattices follow the same patterns of crystal growth demonstrated in atomic thinfilm deposition, allowing these processes to be understood in the context of well-studied atomic epitaxy and enabling a nanoscale model to study fundamental crystallization processes. Through understanding the role of epitaxy as a driving force for NP assembly, we are able to realize 3D architectures of arbitrary domain geometry and size.

62.1

Introduction

The epitaxial deposition of thin films has been key to the semiconductor industry in its efforts to control material properties as a function of crystal structure. The transfer of order and orientation from a substrate to a deposited crystal is dependent upon many interlinked variables (e.g., interfacial chemical potential, crystal lattice parameters, defect stability) that vary as a function of both atomic composition and deposition protocol [1−3]. As a result, significant effort has been expended to fully understand atomic epitaxy, leading to a wealth of information about thin-film

Introduction

crystallization behavior. For nanoscale systems, many strategies have also been developed to assemble nanomaterials into thin films [4−8]; however, these methods often lack the ability to precisely control the overall 3D structure of the resulting crystals (e.g., size, shape, and orientation). Recent developments in nanoparticle (NP) assembly have shown that NPs functionalized with a dense monolayer of oligonucleotides can form ordered superlattice structures with programmable lattice parameters and crystallographic symmetries, and these building blocks exhibit many crystallization behaviors similar to those observed in atomic systems [9−11]. Thus, these “programmable atom equivalents” (PAEs) hold promise for tailoring material structure at the nanoscale in a precise and controllable manner. In the context of PAE thin films, assembly on unpatterned surfaces has been demonstrated to produce rough, polycrystalline films with a lack of long-range order or alignment [12]. Assembly of PAEs on patterned substrates has also been attempted but was limited to monolayers of single-crystalline thin films, as the combination of both NP−substrate and NP−NP binding events significantly increases the complexity of multilayer epitaxial crystal formation [4, 13, 14]. In order to fully control thin-film morphology in PAE superlattices, these complexities must be better understood via investigations into the thermodynamics of lattice growth as a function of different variables. The fundamental information gained from these comprehensive studies not only provides the opportunity to develop superlattice morphologies with complex 3D structures but also has the potential to provide insight into the process of atomic thin-film epitaxy. Like atomic systems, there are many design parameters that can affect multilayer epitaxy, such as factors inherent to the deposition protocol (e.g., thermal annealing temperature) and factors dictated by PAE design (e.g., DNA hybridization strength) [10, 12, 15]. However, unlike atomic epitaxy where only the deposition protocol can be modulated, parameters related to the individual PAE building blocks can be precisely controlled as a function of DNA [9, 10, 12], particle [16−19], or substrate pattern design [13]. Fully understanding how

1161

1162

Epitaxy: Programmable Atom Equivalents versus Atoms

PAE epitaxy can be manipulated as a function of these variables will potentially yield single-crystal superlattices with controlled 3D geometries, allowing for the realization of materials possessing the desired optical [20−22], electronic [23], and magnetic responses [24]. Here, we report a stepwise method for synthesizing large epitaxial thin films of PAEs up to 10 layers thick and 500 μm wide and show that the corresponding growth process mimics atomic thin-film epitaxy, allowing us to study epitaxy as a driving force for building a nanomaterial as it evolves from a 2D monolayer to a 3D crystal lattice.

62.2

Results and Discussion

Epitaxial growth of NP superlattices was realized by depositing PAEs layer-by-layer onto lithographically defined substrates designed to resemble a continuous (100) plane of a body-centered cubic (bcc) lattice (Scheme 62.1). Using standard electron-beam lithography (EBL) techniques, 500 μm × 500 μm arrays of gold posts were synthesized on a silicon wafer, such that post diameters and postto-post distances were comparable to the PAE NP core diameters and the lattice parameter of the superlattice. These posts were functionalized with DNA and hybridized with complementary DNA linkers that presented a single-stranded recognition region to which PAEs could bind. Stepwise thin-film growth was done via successive immersion of the template into suspensions of PAEs displaying a single-stranded recognition region complementary to that of the previous layer. They were then embedded in silica and characterized by synchrotron-based small-angle X-ray scattering (SAXS, Fig. 62.1 middle) and grazing-incidence small-angle X-ray scattering (GISAXS) to determine the overall degree of epitaxy [25]. Lattices were also examined by scanning electron microscopy (SEM, Fig. 62.1 top), which allowed for real-space imaging of superlattice surface morphology; focused ion beam milling was used to etch selected sections of the silica-embedded lattices, allowing for SEM characterization of the internal structure (FIB-SEM, Fig. 62.1 bottom).

Results and Discussion

Scheme 62.1 Layer-by-layer assembly of PAE superlattice thin films on a DNAfunctionalized template.

Figure 62.1  SEM, SAXS, and FIB-SEM characterization of DNA-NP thin films.  (a)  2-,  5-,  and  10-layer  DNA-NP  thin  films  assembled  at  25°C  exhibit  kinetic  roughening and nonepitaxial growth beyond four layers of deposited PAEs. (b)  5- and 10-layer DNA-NP thin films assembled at 25°C and thermally annealed  after the full deposition process demonstrate enhanced ordering, but only the 5-layer sample is fully epitaxial since only PAEs that are close to the initial four  epitaxial layers experience sufficient driving force to align with the patterned  template.  (c)  10-layer  DNA-NP  thin  film  where  each  layer  is  assembled  at  an  elevated temperature; this process produces smooth, crystalline thin films fully  epitaxial with the patterned substrate. Scale bars for SEM and FIB-SEM are 500  and 200 nm, respectively.

We first investigated the deposition of particles onto templated substrates via far-from-equilibrium conditions (i.e., low growth temperature) to understand the effectiveness of the EBL-patterned template itself as a driving force for multilayer epitaxy. When compared to atomic systems, this low temperature deposition is analogous to chemical bath deposition, where atoms rapidly precipitate from solution, resulting in disordered materials [26]. Here, when conducting templated PAE deposition at 25°C, initial layers conform epitaxially to the substrate, but subsequent layers transition to a kinetically roughened, glasslike state (Fig. 62.1a). This can be observed in the SAXS data, where a sample with two

1163

1164

Epitaxy: Programmable Atom Equivalents versus Atoms

deposited layers shows diffraction spots (corresponding to aligned, single-crystal bcc lattices), while 5- and 10-layer films exhibit diffuse scattering, corresponding to disordered PAE aggregates. The relative degree of epitaxy (XA) at each layer was determined by comparing the SAXS intensity of the spots from the (110) peak of the epitaxial PAEs and the intensity of the diffuse ring, where an XA value of 1 indicates complete epitaxy of the PAEs; the value of XA decays from 0.99 to 0.88 after 5 layers and to 0.65 at 10 layers (Fig. 62.2a). This was corroborated by the FIB-SEM cross-sectional images, which showed that the first few layers of all samples are indeed epitaxial, up to a critical layer number of ~4, with PAEs above the critical layer adopting a kinetic, glassy state (Fig. 62.1a, bottom). This is a morphological transition commonly observed in atomic thin films, which exhibit temperature-dependent surface roughening when the adsorption rate is faster than the reorganization rate [2]. Similarly, PAEs adsorbed at low temperature are stuck in kinetic traps, leading to an accumulation of defects in the film, which increases the surface area available for subsequent NPs to bind. This results in amorphous, rough films; the root-mean-squared roughness (RRMS) increases by 140% from 2- to 10-layer deposited films. The rapid, nonlinear increase in height, as measured by the mean z-distance of topmost NPs from the substrate, is further indication of nonequilibrium growth (Fig. 62.2b). These data clearly show that the EBL template does serve as a strong driving force for epitaxy, but this driving force rapidly decays with increasing layer number when lattices are assembled at nonequilibrium conditions. It is possible to reorganize thin films into more thermodynamically preferred configurations by adding thermal energy. This is done frequently in atomic systems to turn disordered or polycrystalline thin films (e.g., from sputter coating) into a film with a singlecrystal orientation. This annealing process was mirrored in the PAE system by first depositing 5 and 10 layers at 25°C and then heating the sample slightly below the film’s melting temperature (Tm, the temperature at which the superlattice dissociates). Interestingly, in the process of determining the films’ Tm, it was observed that they exhibit thickness-dependent melting point depression, analogous to atomic thin-film systems. The thermal stability of the film, measured by monitoring lattice decomposition using SAXS, showed a concomitant increase with thickness due to the decreasing surface-

Results and Discussion

to-volume ratio, as described by Lindemann’s criterion and the Gibbs−Thomson relationship. Upon annealing the samples at (Tm − 2)°C, the 5-layer film became crystalline and epitaxial while retaining the same height and RRMS. On the other hand, the 10-layer sample became crystalline, but not epitaxial (Fig. 62.1b). This is most likely due to the fact that the previously observed critical layer thickness for epitaxy is ~4, indicating that, in the 5-layer sample, only the topmost layers of NPs were disordered. The differences in epitaxy for these two samples mirror previous findings for nonepitaxial PAE systems, in which post-assembly annealing is capable of inducing crystallization but grain boundaries are difficult to remove once formed [15].

Figure 62.2  Quantitative  characterization  of  films  shows  that  depositing  PAEs  at  near-equilibrium  conditions  induces  (a)  higher  degree  of  epitaxy  (as  determined  by  SAXS)  and  (b)  more  controlled  growth  and  a  smoother  film  morphology (as determined from FIB-SEM).

The greatest degree of ordering in thin films can be achieved when the entire deposition process occurs under near-equilibrium conditions. To achieve this in atomic systems, molecular beam epitaxy is performed at high temperatures where deposition and desorption occur at equivalent rates, allowing each adatom to find its thermodynamic position in the monolayer before the next layer is introduced. To achieve this effect in the PAE system, each layer was deposited at an optimized growth temperature (Tm − 4)°C. With this method, nearly perfect Frank−van der Merwe (layer-by-layer) growth was observed, with films remaining epitaxial (XA = 0.99)

1165

1166

Epitaxy: Programmable Atom Equivalents versus Atoms

far beyond the critical layer thickness of room-temperature growth (Figs. 62.1c and 62.2a). Film cross sections examined with FIBSEM show smooth surfaces with an absence of kinetic roughening; the RRMS of the 10-layer film is 51% less than that of the equivalent film assembled at 25°C. Film height increases linearly with deposition layer, indicating that deposition occurred under equilibrium conditions (Fig. 62.2b). Both GISAXS and SAXS confirmed that the film is well-ordered over a large area and nearly completely epitaxial with the patterned template (Fig. 62.1c, middle). Notably, a loss of radial broadening in the SAXS pattern, corroborated by FIB-SEM, indicates that as the film grows from 2 to 5 layers the stability of the PAE network increases to such a point that the particles become locked within a single domain. The appearance of thin lines of diffuse scattering between the peaks originates from the vibrational motion of the particles. This equilibrium growth condition was therefore able to create a crystalline film of well-defined and arbitrary crystal habit that is epitaxial over a domain of 500 μm (Fig. 62.3).

Figure 62.3  Optical image of DNA-functionalized NP thin film grown from a  template exhibiting an arbitrary geometry. SEM images show that the thin film  possesses the same crystallographic orientation across the entire structure.

Finally, an interesting aspect of this PAE system that does not have an atomic analogue is the ability to tune interfacial potential. In atomic crystals, chemical potentials between adatoms and the substrate are influenced by atomic identity and crystallographic symmetry. However, in PAEs, the chemical potential between NPs and the substrate can be tuned by adjusting DNA bond strength; we have recently demonstrated that this can be accomplished after assembly using DNA intercalators [27, 28]. Here, these intercalators can be used to “staple” each PAE layer after deposition and

Methods

annealing, thereby increasing their binding strength and preventing reorganization during subsequent cycles of growth. When this stapling method was employed, 10-layer thin films exhibited similarly high epitaxy (XA = 0.99) but 66% higher RRMS than the nonintercalated counterpart, which can be attributed to defect immobilization by the intercalators. These data and the observed thickness-dependent melting point depression indicate that thermal annealing after each round of deposition induces reorganization not only in the topmost layer of PAEs but also in subsurface layers. This reorganization is critical for achieving perfect epitaxy, indicating the importance of being able to precisely modulate PAE binding strength during deposition.

62.3

Conclusion

In this work, we have determined that DNA-mediated NP crystallization follows similar thin-film growth processes that are observed in atomic thin films, but importantly, PAEs offer a set of parameters distinct from atomic systems that can be independently tuned to control crystallization outcome. Unlike atomic systems, epitaxy can be controlled lithographically and through the choice of oligonucleotide bonding elements. These observations allow one to grow precisely defined crystalline NP architectures of arbitrary shape and size over thousands of μm2. Future studies will be able to take advantage of the tunable nature of the bonding interactions between the PAEs and the substrate to investigate how different parameters (DNA sequence, grafting density of DNA on PAEs) affect the epitaxial deposition process, laying the groundwork for making functional device architectures from crystalline NP networks.

62.4

62.4.1

Methods

DNA Functionalization of Gold Nanoparticles

DNA functionalization on gold nanoparticles (AuNPs, 20 nm diameter from Ted Pella) was done using previously described methods [10, 27]. The 3¢ propylmercaptan protecting group of the

1167

1168

Epitaxy: Programmable Atom Equivalents versus Atoms

thiolated DNA was cleaved with 100 mM dithiolthreitol (SigmaAldrich) for 1 h, followed by desalting on a NAP5 size-exclusion column (GE Healthcare). The deprotected DNA was combined with colloidal AuNPs in a ratio of 6 nmol of DNA per 1 mL of gold colloid. After a 30 min incubation, 1 wt % sodium dodecyl sulfate was added to bring the solution concentration to 0.01%. NaCl (5 M) was added stepwise, followed by 10 s of sonication after each salt addition, until the final concentration of 0.5 M NaCl was achieved. The solution was then allowed to incubate in a shaker overnight to maximize the DNA loading (140 rpm, 37°C). Unbound DNA and excess salt were removed by four successive rounds of centrifugation and resuspension in nanopure water using a 100 kDa filter centrifuge tube (Millipore) on a swinging bucket centrifuge (2500 rpm, 5 min). After the last round of centrifugation, the DNA-NPs were concentrated down to the total volume of 500 μL. The concentrations of resulting DNA-NPs were determined using a Cary 5000 UV-Vis-NIR spectrophotometer (Agilent) and known extinction coefficients from Ted Pella.

62.4.2

Substrate Preparation and Functionalization

62.4.2.1 Patterned template synthesis Si wafers with native oxide (·100Ò, B doped, 10 Ω·cm (Silicon Quest International)) were cleaned (2 min acetone rinse, 2 min methanol rinse, drying under nitrogen) and baked for 2 min at 180°C. PMMA resist (495-A4) was spun onto the wafers (3500 rpm for 60 s) and postbaked (5 min at 180°C). Once they were cooled, 950-A2 PMMA resist was spun coat onto the coated wafers at 3500 rpm for 1 min to create a bilayer and postbaked (5 min at 180°C). EBL was used to write the desired pattern with an optimized beam current (500−700 pA) and dose range (640−1500 μC/cm2). The substrates were developed in cold MIBK/IPA in a 1:3 ratio for 60 s, briefly rinsed in IPA, and dried under nitrogen. The posts were then deposited using an electron-beam evaporator: 3 nm of Cr at a rate of 0.5 Å/s followed by 30 nm of Au at a rate of 1 Å/s. Wafers were then diced into pieces with one pattern per chip. Liftoff was done in heated (100°C−150°C) PG remover (Microchem); the chips were then rinsed (acetone followed by IPA) and finally dried under nitrogen.

Methods

62.4.2.2 Silanization of patterned templates A hydrophobic hexamethyldisilazane (HMDS, Sigma-Aldrich) coating was employed to prevent nonspecific adsorption of DNA-NPs to the silicon chip, so that DNA-NP assembly occurred only on the DNAfunctionalized Au posts of the template. This is a widely used vapor coating technique where the Si of HMDS reacts to form a strong bond with the oxidized silicon, creating a hydrophobic surface. Substrates were prebaked in a Vulcan 3-550 burnout furnace for 1 h at 150°C−200°C to remove adsorbed water molecules. Silanization was performed by incubating the prebaked substrates in a sealed, dry chamber with a small open beaker containing 5 mL of HMDS and 5 mL of anhydrous hexane (Sigma-Aldrich) for 24 h. After 24 h, the substrates were rinsed and sonicated in water or ethanol for a few seconds.

62.4.2.3 Unpatterned substrate preparation

Unpatterned substrates were prepared by depositing a 2 nm Cr adhesion layer followed by 8 nm of Au on Si wafers using a PVD 75 E-beam evaporator (Kurt J. Lesker) at a base pressure of 5 × 10−8 Torr. Cr and Au were evaporated at the rates of 0.3 and 0.5 Å/s, respectively. These conditions yielded a smooth Au film, which is crucial for the crystalline DNA-NP thin-film growth.

62.4.2.4 Substrate DNA functionalization

DNA functionalization of substrates was performed by incubating each substrate in 2 mL Eppendorf tubes (Fisher Scientific) containing 5 μM HS-A DNA solution diluted in buffer A (0.5 M NaCl, 10 mM phosphate buffered saline (PBS)) overnight. The propylmercaptan protecting group on the thiolated DNA was cleaved prior to the functionalization, as described above. The substrates were then washed three times in buffer A with vigorous agitation to remove unbound DNA and then hybridized with “linker” sequences. “Linkers” consist of one complementary section that hybridizes to the thiolated DNA sequence, two double-stranded “duplexed” regions, and a short single-stranded sticky end. To prepare the duplexed linker stocks (100 μM), “duplexer” strands were added to linkers A and B in a 2:1 molar ratio in 0.5 M NaCl. The linkers were heated up to 70°C for 5 min and cooled to room temperature over 4 h

1169

1170

Epitaxy: Programmable Atom Equivalents versus Atoms

to achieve full hybridization. Duplexed linker stock solutions of 100 μM concentration were made fresh every few weeks. Complementary linker (linker A) was hybridized to the DNA-functionalized substrate by incubating substrates in 0.5 μM duplexed linker A solution at 0.5 M NaCl at 35°C for 4 h. Prior to layer-by-layer growth of DNA-NPs, the substrates were rinsed five times in buffer A.

62.4.3 Layer-by-Layer DNA-Nanoparticle Superlattice Thin-Film Assembly 62.4.3.1 Determining the thin-film annealing temperature

To find the appropriate annealing temperature for the DNA-NP thin films, it is first necessary to determine the Tm of DNA-NP aggregates. Although the thermal melting and desorption behaviors are slightly different for solution-phase aggregates versus thin films, the Tm DNA− NP aggregates can be used as a quick way to inform thin-film assembly and annealing temperature. NP superlattice aggregates were prepared by mixing 0.5 pmol of DNA-functionalized 20 nm AuNPs (HS-A and HS-B) and 400 equiv per particle of each duplexed linker in a final concentration of 0.5 M NaCl at room temperature. Assembly was mediated by the complementary pendant “sticky ends” displayed on the linkers. After the sample was allowed to aggregate over 5−10 min, the thermal melting behavior was monitored using a Cary 5000 UV-Vis-NIR spectrophotometer. The extinction of the solution was monitored at 520 and 260 nm while the solution was heated from 25°C to 60°C at a ramp rate of 0.25°C/min, and Tm was calculated from the point of inflection of the melting curve. Typically, reorganization and crystallization can be achieved by annealing a thin-film superlattice at 2°C below Tm,aggregate for 15 min. This is experimentally determined by confirming the crystallinity of annealed DNA-NP thin films grown on unpatterned substrates using SEM.

62.4.3.2 DNA-NP superlattice thin-film assembly

A- and B-type DNA-NP assembly solutions were made by hybridizing each type of NP with its corresponding linker DNA (duplexed as described above) at 400 linkers per NP and incubated at 35°C for

Methods

5 min. The NPs were subsequently diluted to 1 nM concentration (0.5 M NaCl, 10 mM PBS) and used for five layers of PAE assembly. DNA-NP superlattices were grown from the patterned substrates in a layer-by-layer fashion using four different growth conditions: (1) low-temperature growth, (2) low-temperature growth followed by annealing step, (3) elevated temperature growth, and (4) same conditions as 3 with an intercalation step after each annealing step. The intercalator solution was prepared by diluting [Ru(dipyrido[3,2a:2¢,3¢-c]phenazine)(4,4¢-dimethyl-2,2¢-bipyridine)2]Cl2 in 10 mM PBS buffer. This compound was synthesized according to previously reported methods [29]. Layer-by-layer assembly was accomplished in the following way. Substrates functionalized with A-type DNA were incubated in a suspension of B-type DNA-NPs for 4 h. Then the substrates were washed five times in buffer A and immersed in A-type DNA-functionalized AuNP for 4 h. This constituted two layers of DNA-NPs. For the annealing step, the substrate was incubated in buffer A at an elevated temperature for 15 min. This process was repeated until the desired number of layers was achieved. After the desired number of layers was reached, the samples were stored in buffer A at 25°C.

62.4.4

Silica Embedding

In order to transfer liquid-phase thin-film superlattices to the solid state for characterization by SEM and GISAXS while preserving the structure, samples were embedded in silica using a sol-gel process [30]. First, 3 μL of N-(trimethoxysilyl)-propyl-N,N,Ntrimethylammonium chloride (TMSPA, Gelest, 50% in methanol) was added to the thin-film superlattices in 1 mL of buffer A and left to fully associate with the DNA bonds within the superlattices for 30 min on an Eppendorf Thermomixer R (1400 rpm, 25°C). Then 5 μL of triethoxysilane (Sigma-Aldrich) was added, and the sample was shaken for another 30 min before being removed. The samples were rinsed with running water and blown dry with N2. In the case of unsuccessful silica embedding, NPs will dissociate during the rinsing step. In order to prevent such failure, it is crucial to use dry, relatively fresh silane solutions (stored in a desiccator).

1171

1172

Epitaxy: Programmable Atom Equivalents versus Atoms

62.4.5

Small-Angle X-ray Scattering

All SAXS and GISAXS experiments were conducted at the 12ID-B station at the Advanced Photon Source (APS) at Argonne National Laboratory. The samples were probed using 14 keV (0.8856 Å) X-rays, and the sample-to-detector distance was calibrated with a silver behenate standard. The beam was collimated using two sets of slits, and a pinhole was used. The beam size was ~200 μm × 50 μm. Scattered radiation was detected using a Pilatus 2 M detector.

62.4.5.1 SAXS experimental conditions

Unembedded samples were probed using a vertical sample holder made from two coverslips that allowed a buffered environment to be maintained around the sample to preserve DNA hybridization. Embedded samples were mounted on a horizontal sample holder allowing for movement in the in-plane direction (normal to the beam). Sector averaging of diffraction patterns was used to determine the degree of epitaxy.

62.4.5.2 Grazing-incidence SAXS

Embedded samples were aligned to the beam on a sample positioning stage in the x (parallel to the beam), y, z (normal to the substrate), θ (rotation around the y axis), and φ (rotation around the z axis) directions. The sample was centered in the X-ray beam by aligning the pattern to the center of the goniometer’s θ and φ rotations and by adjusting the sample height to be normal to the φ rotation axis. Data were collected at incident angles of 0.1°. After alignment of the sample, scans were taken at several. The sample was centered in the X-ray beam by aligning the pattern to the center of the goniometer’s θ and φ rotations and by adjusting the sample height to be normal to the φ rotation axis angles (rotation around the z axis). In-depth information about GISAXS analysis is available from Senesi et al. [12] and Li et al. [25]. GISAXS scattering patterns of 5-layer and 10-layer films were indexed to bcc crystals with (100) orientation corresponding to space group I4/mmm (#139).

Methods

62.4.6

Focused Ion Beam–Scanning Electron Microscopy

After embedding samples in silica, a representative cross-section SEM image of each sample was obtained on a Helios Nanolab 600 dual-beam focused ion beam milling system with a 52° relative difference between the ion and electron beam. After a layer of titanium was deposited over the area of interest, a 15 μm × 1 μm area, aligned lengthwise with the (100) followed by the (110) plane of the superlattice, was milled with a 93 pA (30 kV) ion beam. Each cross section was imaged with an 86 pA (5 kV) electron beam using the in-lens detector on the SEM without using the software’s tilt correction. Postimage collection, SEM images of the cross sections were lengthened in the y direction by the appropriate factor to account for the tilt. Data analysis on cross sections was done for entire 15 μm cross section; the larger image was subsequently cropped to a representative section and included in the figures for qualitative reference. The degree of epitaxy was determined using Photoshop and Matlab to track the positions of internal PAEs relative to the positions of the templated posts.

62.4.6.1 RMS roughness and mean thickness calculation

To calculate the thin-film thickness and surface root-mean-square (RMS) roughness, the 15 μm cross section of (100) plane images was cropped into one image, where the posts met the substrate and the topmost PAEs were marked using Photoshop. Mean thickness was measured from the substrate surface to the center of the topmost PAE core and averaged over the entire cross section. Since FIBSEM images were taken at an angle, the images were adjusted for the tilt using Matlab prior to data processing. RMS roughness was calculated in its standard fashion: RRMS = √(∑(yi2)/N), where N is the number of PAEs on the thin-film surface and yi = height-mean height.

Acknowledgments

This work was supported by the following awards: AFOSR FA955011-1-0275 and FA9550-12-1-0280; the Department of Defense National Security Science and Engineering Faculty Fellowship

1173

1174

Epitaxy: Programmable Atom Equivalents versus Atoms

N00014-15-1-0043; and the Center for Bio-Inspired Energy Science (CBES), an Energy Frontier Research Center funded by the U.S. Department of Energy (DOE), Office of Science, Basic Energy Sciences under award DE-SC00009890002. This work was also supported by the National Science Foundation’s (NSF) MRSEC program (DMR-1121262) and made use of its Shared Facilities at the Materials Research Center of Northwestern University, specifically the EPIC facility of the NUANCE Center, which also receives support from the Soft and Hybrid Nanotechnology Experimental (SHyNE) Resource (NSF NNCI-1542205). X-ray experiments were carried out at beamline 12-ID-B at the Advanced Photon Source (APS), a U.S. DOE Office of Science User Facility operated by Argonne National Laboratory under Contract No. DE-AC02-06CH11357. EBL was performed at the Kavli Nanoscience Institute’s shared instrumentation center. FIBSEM was performed at the Shared Experimental Facilities supported in part by the MRSEC Program of the NSF (DMR1419807). M.X.W. acknowledges support from the National Science Foundation Graduate Research Fellowship, a Ryan Fellowship, and the Northwestern University International Institute for Nanotechnology. S.E.S. acknowledges support from the Center for Bio-Inspired Energy Sciences Fellowship and the Northwestern University International Institute for Nanotechnology. Y.K. acknowledges support from a Ryan Fellowship and the Northwestern University International Institute for Nanotechnology.

References

1. Zhang, Z. and Lagally, M. G. (1997). Science, 276, 377−383.

2. Bartelt, M. and Evans, J. W. (1995). Phys. Rev. Lett., 75, 4250−4253.

3. Venables, J., Spiller, G. and Hanbucken, M. (1984). Rep. Prog. Phys., 47, 399−459.

4. Shevchenko, E. V., Talapin, D. V., Kotov, N. A., O’Brien, S. and Murray, C. B. (2006). Nature, 439, 55−59.

5. Hua, F., Shi, J., Lvov, Y. and Cui, T. (2002). Nano Lett., 2, 1219−1222.

6. Rupich, S. M., Castro, F. C., Irvine, W. T. and Talapin, D. V. (2014). Nat. Commun., 5, 5045. 7. Tao, A., Kim, F., Hess, C., Goldberger, J., He, R., Sun, Y., Xia, Y. and Yang, P. (2003). Nano Lett., 3, 1229−1233.

References

8. Kotov, N. A., Dekany, I. and Fendler, J. H. (1995). J. Phys. Chem., 99, 13065−13069. 9. Park, S. Y., Lytton-Jean, A. K., Lee, B., Weigand, S., Schatz, G. C. and Mirkin, C. A. (2008). Nature, 451, 553−556.

10. Macfarlane, R. J., Lee, B., Jones, M. R., Harris, N., Schatz, G. C. and Mirkin, C. A. (2011). Science, 334, 204−208.

11. O’Brien, M. N., Jones, M. R., Lee, B. and Mirkin, C. A. (2015). Nat. Mater., 14, 833−839.

12. Senesi, A. J., Eichelsdoerfer, D. J., Macfarlane, R. J., Jones, M. R., Auyeung, E., Lee, B. and Mirkin, C. A. (2013). Angew. Chem. Int. Ed., 52, 6624−6628. 13. Hellstrom, S. L., Kim, Y., Fakonas, J. S., Senesi, A. J., Macfarlane, R. J., Mirkin, C. A. and Atwater, H. A. (2013). Nano Lett., 13, 6084−6090.

14. Lalander, C. H., Zheng, Y., Dhuey, S., Cabrini, S. and Bach, U. (2010). ACS Nano, 4, 6153−6161. 15. Macfarlane, R. J., Thaner, R. V., Brown, K. A., Zhang, J., Lee, B., Nguyen, S. T. and Mirkin, C. A. (2014). Proc. Natl. Acad. Sci. U.S.A., 111, 14995−15000. 16. Zhang, C., Macfarlane, R. J., Young, K. L., Choi, C. H. J., Hao, L., Auyeung, E., Liu, G., Zhou, X. and Mirkin, C. A. (2013). Nat. Mater., 12, 741−746.

17. Zhang, Y., Lu, F., Yager, K. G., van der Lelie, D. and Gang, O. (2013). Nat. Nanotechnol., 8, 865−872. 18. Lee, J.-S., Lytton-Jean, A. K., Hurst, S. J. and Mirkin, C. A. (2007). Nano Lett., 7, 2112−2115. 19. Brodin, J. D., Auyeung, E. and Mirkin, C. A. (2015). Proc. Natl. Acad. Sci. U.S.A., 112, 4564−4569.

20. Evanoff, D. D. and Chumanov, G. (2005). ChemPhysChem, 6, 1221−1231.

21. Kelly, K. L., Coronado, E., Zhao, L. L. and Schatz, G. C. (2003). J. Phys. Chem. B, 107, 668−677.

22. Storhoff, J. J., Lazarides, A. A., Mucic, R. C., Mirkin, C. A., Letsinger, R. L. and Schatz, G. C. (2000). J. Am. Chem. Soc., 122, 4640−4650.

23. Choi, J.-H., Wang, H., Oh, S. J., Paik, T., Sung, P., Sung, J., Ye, X., Zhao, T., Diroll, B. T. and Murray, C. B. (2016). Science, 352, 205−208.

24. Jonsson, P. E. (2003). Adv. Chem. Phys., 128, 191−248.

25. Li, T., Senesi, A. J. and Lee, B. (2016). Chem. Rev., 116, 11128.

26. Meyers, S. T., Anderson, J. T., Hung, C. M., Thompson, J., Wager, J. F. and Keszler, D. A. (2008). J. Am. Chem. Soc., 130, 17603−17609.

1175

1176

Epitaxy: Programmable Atom Equivalents versus Atoms

27. Seo, S. E., Wang, M. X., Shade, C. M., Rouge, J. L., Brown, K. A. and Mirkin, C. A. (2016). ACS Nano, 10, 1771−1779. 28. Pal, S., Zhang, Y., Kumar, S. K. and Gang, O. (2015). J. Am. Chem. Soc., 137, 4030−4033.

29. Shade, C. M., Kennedy, R. D., Rouge, J. L., Rosen, M. S., Wang, M. X., Seo, S. E., Clingerman, D. J. and Mirkin, C. A. (2015). Chem. Eur. J., 21, 10983−10987.

30. Auyeung, E., Macfarlane, R. J., Choi, C. H. J., Cutler, J. I. and Mirkin, C. A. (2012). Adv. Mater., 24, 5181−5186.

Chapter 63

Lattice Mismatch in Crystalline Nanoparticle Thin Films*

Paul A. Gabrys,a Soyoung E. Seo,b,c Mary X. Wang,c,d EunBi Oh,b,c Robert J. Macfarlane,a and Chad A. Mirkinb,c,d,e aDepartment of Materials Science and Engineering, Massachusetts Institute of Technology (MIT), 77 Massachusetts Avenue, Cambridge, MA 02139, USA bDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA dDepartment of Chemical and Biological Engineering, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA eDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA [email protected]; [email protected]

Paul A. Gabrys, Soyoung E. Seo, and Mary X. Wang contributed equally to this work.

*Reprinted with permission from Gabrys, P. A., Seo, S. E., Wang, M. X., Oh, E., Macfarlane, R. J. and Mirkin, C. A. (2018). Lattice mismatch in crystalline nanoparticle thin films, Nano Lett. 18, 579−585. Copyright (2018) American Chemical Society. Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1178

Lattice Mismatch in Crystalline Nanoparticle Thin Films

For atomic thin films, lattice mismatch during heteroepitaxy leads to an accumulation of strain energy, generally causing the films to irreversibly deform and generate defects. In contrast, more elastically malleable building blocks should be better able to accommodate this mismatch and the resulting strain. Herein, that hypothesis is tested by utilizing DNA-modified nanoparticles as “soft,” programmable atom equivalents to grow a heteroepitaxial colloidal thin film. Calculations of interaction potentials, small-angle X-ray scattering data, and electron microscopy images show that the oligomer corona surrounding a particle core can deform and rearrange to store elastic strain up to ±7.7% lattice mismatch, substantially exceeding the ±1% mismatch tolerated by atomic thin films. Importantly, these DNAcoated particles dissipate strain both elastically through a gradual and coherent relaxation/broadening of the mismatched lattice parameter and plastically (irreversibly) through the formation of dislocations or vacancies. These data also suggest that the DNA cannot be extended as readily as compressed, and thus the thin films exhibit distinctly different relaxation behavior in the positive and negative lattice mismatch regimes. These observations provide a more general understanding of how utilizing rigid building blocks coated with soft compressible polymeric materials can be used to control nano- and microstructure.

63.1

Introduction

Heteroepitaxy is the process of depositing a thin film of one material atop a dissimilar material while maintaining the crystallinity of each composition. However, when the two lattice parameters are misaligned, significant strain is generated in the deposited crystals. In cases where there is minimal strain, the lattices form a coherent interface with a continuous crystallographic alignment. Unfortunately, in atomic thin films only heteroepitaxial processes with a low lattice mismatch (max ±1%) tend to be coherent, and strain is instead generally alleviated by the formation of defects within the deposited crystal [1−4]. Since atoms behave like and are often modeled as hard spheres, engineering this crystal

Introduction

interface requires significant modulation of lattice composition (e.g., introducing dopant atoms to adjust lattice parameter). As a result, material composition must often be compromised in order to prevent the formation of undesirable defects [1]. However, if a more elastically malleable building block was used, the equilibrium thinfilm crystal structures would be more accommodating of strain from lattice mismatch as a function of film thickness [5, 6]. Although it is not possible to adjust an individual atom’s “softness,” programmable atom equivalents (PAEs) generated from rigid nanoparticles (NPs) and soft, highly tunable DNA ligand shells can be chemically adjusted in a deliberate and rational manner [7−11]. Indeed, such PAEs can be assembled into ordered, crystalline structures in a manner that is analogous to atomic crystallization [12−14]. We have previously shown that PAEs and lithographically defined templates can be used to grow single-crystalline thin films in an epitaxial manner [15, 16]. In principle, this enables the study of epitaxy using building blocks that are more tunable than atoms. In this work, we extend this motif to the concept of heteroepitaxy by inducing controlled lattice mismatch between the substrate and PAEs. While significant research effort has been devoted to investigating atomic heteroepitaxy [2], PAEs have not been fully explored in this context yet. Understanding how strain is alleviated in these thin films is key to improving control over nano- and microscale structure in NP-based crystalline materials. This understanding could enable the fabrication of more complex nanodevices, particularly those requiring variations in NP composition with different lattice constants within the lattice. In heteroepitaxial systems with lattice mismatch, elastic strain energy accumulates within the deposited thin film with the addition of each layer [2, 4]. In atomic systems, all strain energy must be stored within the atomic bonds themselves, raising the overall potential energy of each individual building block. An oversimplified model can describe crystalline lattices as masses (atoms) connected by springs in each of the lattice directions. Each spring has an equilibrium length corresponding to an ideal interatomic spacing based on the energetics of binding and a spring constant that

1179

1180

Lattice Mismatch in Crystalline Nanoparticle Thin Films

opposes deviation [17−19]. Similarly, PAEs exhibit a well-defined, equilibrium interparticle distance that balances favorable DNA hybridization between neighbors with unfavorable steric repulsion between DNA coronae [10, 12, 20, 21]. To an extent, PAEs can store elastic energy in different modes such as compression, extension, bending, and rearrangement of the DNA bonds between each set of neighboring PAEs, suggesting that they are significantly “softer” building blocks than atoms [11, 22, 23]. Given the numerous modes of elastic energy storage available to soft matter, we hypothesize that the PAE thin films will more readily accommodate any deviation due to lattice mismatch.

63.2

Methods

To explore the energetic stability of a heteroepitaxial PAE thin film, a mathematical model based on mean field approximation [21] was used to calculate theoretical PAE interaction potential energies (Fig. 63.1). These calculations used a PAE design consisting of 20 nm spherical gold NPs functionalized with one of two oligonucleotide sequences bound to linker strands with complementary “sticky ends.” This binary system was assembled into body-centered cubic (bcc) crystals with an equilibrium lattice parameter of 65 nm. The foundation for this model is based on the assumptions that (i) all DNA sticky ends within the region of overlap between complementary PAEs are hybridized, and (ii) the PAEs are assembled into a rigid lattice with well-defined x,y-spacing (i.e., dissipation of strain energy was not considered). Previous studies show that the main thermodynamic contributor to lattice stability is the balance between attraction from DNA hybridization and interparticle repulsion due to excluded volume interactions between the DNA brushes [24−27]. A “bulk” energy was first calculated by modeling a free-standing PAE film of a given thickness at the equilibrium lattice parameter. This is the energy associated with an unstrained lattice without the additional stability arising from being bound to a substrate. In the presence of a heterogeneous interface for the thin-film cases,

Methods

an interfacial energy term was included in the model by adding the attractive potential arising from the hybridization between a DNAfunctionalized substrate and the superlattice film.

Figure 63.1  Modeled  PAE  thin  films  are  energetically  stable  up  to  roughly  ±9% lattice mismatch at 10 layers. Calculated PAE thin-film potential energies  relative to bulk (a) as a function of film thickness and (b) as a function of induced  strain. In all figures, blue data correspond to negative lattice mismatch, red data  to positive lattice mismatch; squares correspond to 5 layer samples, “X”s to 10 layers. Black lines correspond to “ideal”/“bulk” values.

The impact of lattice mismatch on PAE thin-film energy was investigated by calculating the potential energy of coherently assembled PAE thin films subjected to various amounts of lattice mismatch strain (Fig. 63.1). The lattice energy calculations, compared to their bulk analogues, demonstrate that the soft DNA shell allows

1181

1182

Lattice Mismatch in Crystalline Nanoparticle Thin Films

the assembled thin films to withstand a significant amount of strain. However, strain energy accumulates with each layer of the film and eventually destabilizes the overall structure to the point that the film is energetically less stable than the bulk state (Fig. 63.1a). This also implies that at greater lattice mismatch the energy barrier for coherent, heteroepitaxial growth will be prohibitively high, resulting in incoherent growth. The model reveals that 10 layer films beyond roughly ±9% strain are less stable than their bulk analogue (Fig. 63.1b). Thus, for the PAE design studied here, 10 layer thin films are predicted to only exhibit epitaxial alignment when the heterogeneous interface induces less than ±9% strain. Beyond this regime, the analytical model predicts that the accumulated strain energy will make thin-film growth energetically unstable, favoring the formation of defects or dewetted amorphous (bulk) structures. To investigate these hypotheses experimentally, PAEs consistent with the modeled design (DNA functionalized 20 nm gold NPs that can arrange in a bcc lattice with a 65 nm lattice parameter upon annealing) were synthesized following literature protocols [12, 28]. Arrays of gold dots commensurate in size with the PAEs were deposited on a silicon wafer in a pattern mimicking the (001) plane of the targeted bcc superlattice using electron beam lithography. Once the dots were functionalized with one of the DNA strands and linkers, PAEs functionalized with the complementary sequence could bind and form a monolayer on the patterned array. Previous studies have shown that the PAEs bind to the dot array in the position that maximizes the number of DNA linkages [15, 16]. In this case, the particle was driven to sit in the center of four dots, epitaxially continuing the patterned crystal plane (i.e., the formation of the (002) plane). This process was continued in a layer-by-layer fashion to assemble multiple layers on each template. To study the phenomenon of heteroepitaxy and the impact of lattice mismatch, arrays were fabricated with different lattice parameters than the ideal bulk PAE superlattice (up to ±10.8% lattice mismatch). The thin films were embedded in silica following established protocols [29] and the structure of each was determined by synchrotron-based transmission small-angle X-ray scattering (SAXS), as well as focused ion beam (FIB) cross-sectioning followed by scanning electron microscopy (SEM).

Results and Discussion

63.3

Results and Discussion

First, SAXS was used to determine the degree of crystallinity and epitaxy. Because the incident X-ray beam was normal to the substrate for these measurements, the resulting SAXS data shown in Fig. 63.2 are 2D projections of the x,y-plane positions in reciprocal space. Each SAXS pattern shows a high degree of single-crystalline ordering consistent with the patterned bcc (001) plane (Fig. 63.2a). In these data, epitaxially aligned PAEs contribute to the intensity of the reciprocal space lattice positions. Any deviation in alignment results in a broadening of each diffraction spot. Regions of nonepitaxial PAEs within the film add to the intensity of a diffuse ring encircling the beam center. Therefore, we can define an order parameter by comparing the integrated SAXS intensity of a diffraction spot to the relative integrated intensity of the diffuse ring over the same q-range (Fig. 63.2b). Consistent with both the hypotheses and the calculations of interaction potentials, all samples exhibit a reasonably high degree of order (intensity at the diffraction spot 2.5 times greater than the integrated intensity of the corresponding diffuse ring), even the cases of ±10.8% lattice mismatch. Within ±7.7%, the thin films exhibit near perfect ordering, matching the analytical model. This is in stark comparison to atomic thin-film heteroepitaxy, which rarely remains epitaxial and coherent above ±1% lattice mismatch [2]. SAXS also confirms that the interparticle distance between PAEs in the thin film (calculated from the q(110) position) conform to the induced mismatch (Fig. 63.2c). Finally, additional confirmation of the epitaxial alignment was achieved with SEM (Fig. 63.2d) and atomic force microscopy (AFM) of the thin films, which both show a transition from an aligned and ordered surface into an amorphous layer beyond ±7.7% mismatch. The strong parallels between the modeled and experimental results indicate that a majority of the accumulated strain energy in this “soft” heteroepitaxy system is likely stored within the DNA shell. However, broadening of the diffraction spots (which increases with larger values of lattice mismatch) is indicative of deviation of the PAEs from ideal lattice positions and suggests that some of the strain energy is being dissipated throughout the thin-film structure. In

1183

1184

Lattice Mismatch in Crystalline Nanoparticle Thin Films

Figure 63.2  PAE  thin  films  maintain  coherency  with  the  patterned  crystallography up to ±7.7% lattice mismatch. (a) The 2D transmission SAXS data centered on the (110) reciprocal spot, (b) order parameter, calculated by comparing the integrated intensity of the (110) spot to the intensity of the amorphous ring, as a function of lattice mismatch, (c) the maximum q(110) value from the measured SAXS data compared to the templated q(110) position, and (d) representative SEM micrographs of the thin-film surface.

general, a material can alleviate strain energy in two ways: elastically or plastically. Elastic alleviation is strain removal that is done without breaking any bonds between individual building blocks, while plastic

Results and Discussion

deformation requires the breaking of individual bonds and generally manifests in the form of defects [3]. Heteroepitaxial thin films (atomic or PAE-based) that can dissipate strain elastically largely do so through a gradual change of the interparticle distances in lattice planes further from the heterogeneous interface (Scheme 63.1). In other words, lattice planes further from the mismatched interface will have interparticle distances closer to the ideal spacing. Atoms in a crystalline lattice have relatively rigid and directional atomic orbitals resulting in a sharp and deep energy well as a function of position. Therefore, atoms incur significant energetic penalties for deviating too readily from an ideal lattice position. At high lattice mismatch, the strain energy present in the bond between two atoms outweighs the energy cost of breaking that nearest-neighbor bond, driving the atoms to rearrange their equilibrium structure and form defects such as misfit dislocations or lattice vacancies [3]. The mechanisms of alleviating the strain from lattice mismatch available to heteroepitaxial PAE thin films can be deduced from their resulting equilibrium structure. In contrast with atoms, spherical PAEs are isotropic and have omni-directional binding. Additionally, the flexibility of the DNA corona allows for a wider range of achievable binding distances at a lower energy cost [11]. Therefore, these “soft” building blocks are hypothesized to more readily allow for elastic alleviation of strain.

Scheme 63.1 PAE thin films elastically alleviate strain from lattice mismatch. Representation of the layer-by-layer PAE epitaxy platform and heteroepitaxial PAE  thin  films  under  negative  (left)  and  positive  (right)  lattice  mismatch  (compressive and tensile strain, respectively) with coherent interfaces elastically relieving strain. Cross section of the (010) plane is shown.

Indeed, the SAXS data clearly reveal an elastic mode of strain relief (Fig. 63.3). Intensities at larger scattering angles (larger q

1185

1186

Lattice Mismatch in Crystalline Nanoparticle Thin Films

values) correspond to smaller real space distances, and vice versa. By averaging radially along only the 45° angle, trends in the average interparticle distance can be observed (Fig. 63.3a). In all mismatch cases, the (110) peak intensity is strongest where matched with the template position and exhibits a tail that leans toward the bulk, ideal interparticle spacing. Plotting the (110) peak maxima and their respective rightward and leftward peak widths at half max, the data demonstrate that when the thin films are compressed under negative lattice mismatch, the PAEs relax toward larger spacing and vice versa (Fig. 63.3b). While this conclusion may be intuitive for soft matter in general, it highlights the behavior unique to a “soft” building block in the construction of crystalline materials. While strain energy in these PAE systems is alleviated predominantly in an elastic manner by stretching or compressing soft DNA shells, plastic deformation would be expected to occur when the accumulated strain energy in the collective DNA “bond” between two particles exceeds the energy penalty associated with breaking that bond (i.e., the DNA ligand shell is too stretched or compressed for the DNA sticky-end duplexes to remain in the hybridized state). Given the high grafting density of oligomers on the surface of the PAEs (which results in a crowded environment for the DNA strands) [30, 31] and the fact that the DNA bonds consist almost entirely of duplex DNA, we hypothesize that the DNA corona surrounding an individual particle will display different physical characteristics under compressive and tensile lattice mismatch. Specifically, under negative mismatch the volume available to a PAE in the lattice is reduced, generally confining it to its local lattice position. However, under positive mismatch the free accessible volume of each lattice site is increased, providing a larger space for each PAE to occupy. Thus, PAEs under positive lattice mismatch have higher degrees of translational freedom (variation around ideal lattice sites) than films of a corresponding compressive mismatch. According to the analytical model in this work, an increased interparticle distance reduces the overlap region between complementary particles, leading to fewer DNA sticky-end hybridization events and a higher overall energy. Therefore, the maximum number of DNA hybridization events may occur when the particle is shifted away from the lattice site, a behavior that is not considered in the model. By releasing entirely from one neighbor (breaking the “bond” between the two PAEs) and

Results and Discussion

shifting toward its other complementary nearest neighbors, each PAE in a larger-than-ideal unit cell could increase its overall DNA hybridization, resulting in a lower equilibrium, lattice energy.

Figure 63.3  PAE thin  films  alleviate strain  energy elastically in the  x,y-plane through gradual retraction/expansion of interparticle distance toward the bulk value. (a) The 1D radial line averages of the SAXS data along the close-packed direction with the “ideal” bulk phase (110) peak position noted for reference. Dotted lines are 5 layer films and solid lines are 10 layers. Inset: representative  schematic of direction and width of radial line cuts. (b) Plot of max (110) peak positions relative to the template and the peak width at half max of each line cut in the high-q (small interparticle distance) and low-q (large interparticle distance) directions displayed as error bars.

This type of random, plastic deviation results in azimuthal, instead of radial, broadening of SAXS diffraction spots as the PAEs within the film transition from precise locations to more random, amorphous ones. After taking broad azimuthal cuts (from q-values

1187

1188

Lattice Mismatch in Crystalline Nanoparticle Thin Films

of 0.0093 to 0.0183 Å−1) that encompass the entirety of the (110) diffraction spot in each sample, the 1D averaged intensities were plotted as a function of angle; the relative breadth of these peaks was then plotted to determine the degree of translational freedom for each sample (Fig. 63.4a). As hypothesized, the tensile samples have a much higher degree of translational freedom. To corroborate this conclusion from the SAXS data, the 10 layer thin films were cross-sectioned via FIB milling and imaged with SEM. Each cross section was aligned with the (010) plane of the thin film to readily observe any deviation in the PAE positions. The significant difference between translational freedom in the negative versus positive mismatch cases is visually apparent by observing the [001] vectors moving in the z direction in the representative FIB cross sections comparing the +7.7% lattice mismatch sample to the −7.7% sample (Fig. 63.4b). Similar to atomic materials under tensile strain [3], the PAE films under positive lattice mismatch show signs of “micro-tears” (gaps) where void space within the structure is a result of PAEs locally breaking bonds in the same direction. The fact that these gaps are not observed in the negative mismatch samples indicates that the PAEs’ “soft” coronae can be compressed with the addition of strain but not as readily stretched. This is consistent with prior experimental results on the mechanical properties of single DNA strands [32−36]. This nonreciprocal difference gives rise to the different physical characteristics and amounts of plastic versus elastic deformation of the PAE lattices under positive and negative lattice mismatch. The microscopy images corroborate the broad conclusions of the SAXS data; all thin films appear, in general, epitaxial with increasing frequency of defects as mismatch increases. These defects culminate in the formation of large “glassy” regions under extreme mismatch. Under positive lattice mismatch, the defects are commonly random deviations in the spacing between lattice planes (Fig. 63.4b). Interestingly, the FIB cross sections, particularly for the negative lattice mismatch films, also show the presence of other plastic deformation mechanisms to alleviate strain that are commonly observed in atomic systems [1, 2]. These include both dislocations and vacancies, which alleviate strain by breaking bonds between particles and changing the local crystal structure of the lattice.

Results and Discussion

Figure 63.4  Plastic  alleviation  of  strain  in  PAE  thin  films  in  the  presence  of  high strain. (a) Plot of relative breadth of the (110) SAXS peak in the azimuthal direction (corresponding to the degree of PAE translational freedom within the thin film) versus lattice mismatch. Inset: representative schematic of direction  and width of azimuthal cuts. (b) The higher frequency of random, lateral deviations in x,y-planes under large positive mismatch (SEM in red) versus large negative mismatch (SEM in blue) was verified visually from the FIB cross  sections. Scale bars are 250 nm.

The conclusions presented here reveal the complexity of using PAEs as thin-film building blocks. They exhibit remarkable strain

1189

1190

Lattice Mismatch in Crystalline Nanoparticle Thin Films

tolerance based on their design, accommodating a lattice mismatch up to ±7.7% while maintaining coherency. Unlike atoms, PAEs are readily able to store the accumulated strain energy within the DNA bonds between particles and alleviate some of the strain elastically through a change in lattice parameter as a function of layer number. While the release of any strain in these PAE thin films is primarily through elastic relaxation, these building blocks still undergo plastic defect formation. The observed misfit dislocations and vacancies are comparable to atomic thin films. In principle, these PAEs should be tunable to behave like soft matter or hard sphere atoms depending on synthetic variation of the building block design and deposition parameters. “Soft” heteroepitaxy, while maintaining similarities to atomic heteroepitaxy, contains unique mechanistic differences that can greatly affect material synthesis. This effect might be even more prominent for crystals in solution, where macroscopic curvature could provide additional mechanisms for strain relief. In principle, the induced strain at the interface could be used to control the thickness of overgrown material atop a dissimilar material with different lattice parameter, that is, in a core–shell structure [37]. Ultimately, this study yields unparalleled control over nano- and microstructure in NP-based systems. The platform investigated in this work can provide a toolkit for novel NP devices requiring improved control over interparticle distances, as well as provide a proxy system for the study of interface/thin-film science at the boundaries of heteroepitaxial interfaces.

Acknowledgments

This work was supported by the following awards: Air Force Office of Scientific Research FA9550-16-1-0150 (oligonucleotide syntheses and purification), FA9950-17-1-0348 (DNA-functionalization of gold nanoparticles), and FA9550-17-1-0288 Young Investigator Research Program (substrate fabrication, particle assembly and electron microscopy characterization); the Vannevar Bush Faculty Fellowship program sponsored by the Basic Research Office of the Assistant Secretary of Defense for Research and Engineering and funded by the Office of Naval Research through Grant N00014-151-0043 (substrate functionalization); the Center for Bio-Inspired Energy Science, an Energy Frontier Research Center funded by the

References

U.S. Department of Energy, Office of Science, Basic Energy Sciences award DE-SC0000989 (nanoparticle superlattice thin-film assembly and characterization). Fabrication and SEM characterization were performed at the Materials Technology Laboratory at MIT. SAXS experiments were carried out at beamline 12-IDB at the Advanced Photon Source, a U.S. DOE Office of Science User Facility operated by Argonne National Laboratory under Contract DE-AC02-06CH11357, and the authors particularly acknowledge the help of Byeongdu Lee. FIB was performed at the Shared Experimental Facilities in the Center for Materials Science and Engineering at MIT, supported in part by the MRSEC Program under National Science Foundation award DMR-1419807. AFM was performed at the Materials Research Center of Northwestern University supported by National Science Foundation award DMR1121262. P.A.G. acknowledges support from the NSF Graduate Research Fellowship Program under Grant NSF 1122374. S.E.S. acknowledges partial support from the Center for Computation and Theory of Soft Materials Fellowship. M.X.W. was supported by the NSF Graduate Research Fellowship.

References

1. Palmstrom, C. J. (1995). Annu. Rev. Mater. Sci., 25, 389−415.

2. Ohring, M. (2002). Materials Science of Thin Films: Deposition and Structure, 2nd ed. (Academic Press, San Diego, CA).

3. Callister, W. D. and Rethwisch, D. G. (2010). Materials Science and Engineering: An Introduction, 8th ed. (John Wiley & Sons, Hoboken, NJ).

4. Ayers, J. E., Kujofsa, T., Rago, P. and Raphael, J. (2016). Heteroepitaxy of Semiconductors: Theory, Growth, and Characterization, 2nd ed. (CRC Press, New York).

5. Rupich, S. M., Castro, F. C., Irvine, W. T. M. and Talapin, D. V. (2014). Nat. Commun., 5, 5045.

6. Van Blaaderen, A., Ruel, R. and Wiltzius, P. (1997). Nature, 385, 321−324. 7. Mirkin, C. A., Letsinger, R. L., Mucic, R. C. and Storhoff, J. J. (1996). Nature, 382, 607−609.

8. Hill, H. D., Macfarlane, R. J., Senesi, A. J., Lee, B., Park, S. Y. and Mirkin, C. A. (2008). Nano Lett., 8, 2341−2344.

1191

1192

Lattice Mismatch in Crystalline Nanoparticle Thin Films

9. Park, S. Y., Lytton-Jean, A. K. R., Lee, B., Weigand, S., Schatz, G. C. and Mirkin, C. A. (2008). Nature, 451, 553−556.

10. Nykypanchuk, D., Maye, M. M., van der Lelie, D. and Gang, O. (2008). Nature, 451, 549−552.

11. Senesi, A. J., Eichelsdoerfer, D. J., Brown, K. A., Lee, B., Auyeung, E., Choi, C. H. J., Macfarlane, R. J., Young, K. L. and Mirkin, C. A. (2014). Adv. Mater., 26, 7235−7240.

12. Macfarlane, R. J., Lee, B., Jones, M. R., Harris, N., Schatz, G. C. and Mirkin, C. A. (2011). Science, 334, 204−208.

13. Auyeung, E., Li, T. I. N. G., Senesi, A. J., Schmucker, A. L., Pals, B. C., de la Cruz, M. O. and Mirkin, C. A. (2013). Nature, 505, 73−77.

14. Macfarlane, R. J., Lee, B., Hill, H. D., Senesi, A. J., Seifert, S. and Mirkin, C. A. (2009). Proc. Natl. Acad. Sci. U.S.A., 106, 10493−10498.

15. Wang, M. X., Seo, S. E., Gabrys, P. A., Fleischman, D., Lee, B., Kim, Y., Atwater, H. A., Macfarlane, R. J. and Mirkin, C. A. (2017). ACS Nano, 11, 180−185. 16. Hellstrom, S. L., Kim, Y., Fakonas, J. S., Senesi, A. J., Macfarlane, R. J., Mirkin, C. A. and Atwater, H. A. (2013). Nano Lett., 13, 6084−6090.

17. Fetter, A. L. and Walecka, J. D. (2003). Theoretical Mechanics of Particles and Continua (Courier Corporation, Waltham, MA). 18. Kittel, C. (1967). Am. J. Phys., 35, 547−548.

19. Born, M. and Huang, K. (1955). Am. J. Phys., 23, 474−474.

20. Wang, M. X., Brodin, J. D., Millan, J. A., Seo, S. E., Girard, M., Olvera de la Cruz, M., Lee, B. and Mirkin, C. A. (2017). Nano Lett., 17, 5126−5132.

21. Seo, S. E., Li, T., Senesi, A. J., Mirkin, C. A. and Lee, B. (2017). J. Am. Chem. Soc., 139, 16528−16535.

22. O’Brien, M. N., Girard, M., Lin, H.-X., Millan, J. A., Cruz, M. O. de la; Lee, B. and Mirkin, C. A. (2016). Proc. Natl. Acad. Sci. U.S.A., 113, 10485−10490.

23. Lee, O.-S., Cho, V. Y. and Schatz, G. C. (2012). J. Phys. Chem. B, 116, 7000−7005. 24. Maye, M. M., Nykypanchuk, D., van der Lelie, D. and Gang, O. (2007). Small, 3, 1678−1682. 25. Tkachenko, A. V. (2002). Phys. Rev. Lett., 89, 148303.

26. Nykypanchuk, D., Maye, M. M., van der Lelie, D. and Gang, O. (2007). Langmuir, 23, 6305−6314.

27. Jin, R., Wu, G., Li, Z., Mirkin, C. A. and Schatz, G. C. (2003). J. Am. Chem. Soc., 125, 1643−1654.

References

28. Hill, H. D. and Mirkin, C. A. (2006). Nat. Protoc., 1, 324−336.

29. Auyeung, E., Macfarlane, R. J., Choi, C. H. J., Cutler, J. I. and Mirkin, C. A. (2012). Adv. Mater., 24, 5181−5186.

30. Hurst, S. J., Lytton-Jean, A. K. R. and Mirkin, C. A. (2006). Anal. Chem., 78, 8313−8318.

31. Hill, H. D., Millstone, J. E., Banholzer, M. J. and Mirkin, C. A. (2009). ACS Nano, 3, 418−424. 32. Bustamante, C., Bryant, Z. and Smith, S. B. (2003). Nature, 421, 423−427.

33. Rief, M., Clausen-Schaumann, H. and Gaub, H. E. (1999). Nat. Struct. Biol., 6, 346−349. 34. Marko, J. F. and Cocco, S. (2003). Phys. World, 16, 37.

35. Bustamante, C., Smith, S. B., Liphardt, J. and Smith, D. (2000). Curr. Opin. Struct. Biol., 10, 279−285.

36. Kosikov, K. M., Gorin, A. A., Zhurkin, V. B. and Olson, W. K. (1999). J. Mol. Biol., 289, 1301−1326.

37. Seo, S. E., Wang, M. X., Shade, C. M., Rouge, J. L., Brown, K. A. and Mirkin, C. A. (2016). ACS Nano, 10, 1771−1779.

1193

Chapter 64

Building Superlattices from Individual Nanoparticles via Template-Confined DNA-Mediated Assembly*

Qing-Yuan Lin,a,c Jarad A. Mason,a,b Zhongyang Li,d Wenjie Zhou,a,b Matthew N. O’Brien,a,b Keith A. Brown,a,b Matthew R. Jones,a,c Serkan Butun,d Byeongdu Lee,e Vinayak P. Dravid,a,c Koray Aydin,d and Chad A. Mirkina,b,c aInternational

Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA bDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA dDepartment of Electrical Engineering and Computer Science, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA eX-Ray Science Division, Advanced Photon Source, Argonne National Laboratory, 9700 S. Cass Ave., Argonne, IL 60439, USA [email protected]; [email protected]; [email protected] Qing-Yuan Lin and Jarad A. Mason contributed equally to this work.

*From Lin, Q.-Y., Mason, J. A., Li, Z., Zhou, W., O’Brien, M. N., Brown, K. A., Jones, M. R., Butun, S., Lee, B., Dravid, V. P., Aydin, K. and Mirkin, C. A. (2018). Building superlattices from individual nanoparticles via template-confined DNA-mediated assembly, Science 359(6376), 669–672. Reprinted with permission from AAAS.

Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1196

Building Superlattices from Individual Nanoparticles

DNA programmable assembly has been combined with top-down lithography to construct superlattices of discrete, reconfigurable nanoparticle architectures on a gold surface over large areas. Specifically, the assembly of individual colloidal plasmonic nanoparticles with different shapes and sizes is controlled by oligonucleotides containing “locked” nucleic acids and confined environments provided by polymer pores to yield oriented architectures that feature tunable arrangements and independently controllable distances at both nanometer and micrometer length scales. These structures, which would be difficult to construct via other common assembly methods, provide a platform to systematically study and control light–matter interactions in nanoparticle-based optical materials. The generality and potential of this approach are explored by identifying a broadband absorber with a solvent polarity response that allows dynamic tuning of visible light absorption.

64.1

Introduction

DNA has become a powerful tool for constructing highly ordered materials from nanoparticle (NP) building blocks [1–3]. Indeed, the tunability afforded by the sequence-specific interactions inherent to oligonucleotides has been leveraged to assemble NPs into many exotic structures, including colloidal crystals that feature over 30 different lattice symmetries, tunable interparticle distances ranging from below 3 nm to above 130 nm, and multiple well-defined crystal habits [4–7]. In contrast to the diversity of structures that have been synthesized in solution, DNA has only been used to generate a relatively limited set of NP structures on surfaces [8–10]. Moreover, the use of DNA—as well as any other bottom-up or combination of bottom-up and top-down assembly technique—to transfer colloidal NPs from solution to a surface has led to either NP monolayers or three-dimensional (3D) extended lattices [11–14], while the synthesis of isolated nanostructures that incorporate multiple NP sizes, shapes, and/or composition has remained elusive. The ability to predictably, rapidly, and precisely place individual NPs into desired arrangements—regardless of size, shape, or composition—over

Methods

large areas on a surface in both two and three dimensions would represent a significant advance in structural control, dramatically expanding the range of nanomaterials that can be synthesized and enabling new properties, many of which have likely never even been contemplated because of a lack of access to such structures. With most assembly techniques, it is extremely challenging to control the thermodynamics of interactions both between NPs and between NPs and a surface, as would be required to build discrete, surface-bound architectures with a level of structural control. Here, by using DNA programmable interactions to direct the layer-by-layer assembly of colloidal NPs within a polymer template, we realized oriented superlattices of multicomponent NP architectures. Confined environments provided by the pores of the polymer template enable the construction of architectures perpendicular to the substrate, while precisely engineered NP-NP interactions mediated by DNA allow architectures to be assembled a single NP at a time with controlled interparticle distances. These NP superlattices can be specified and independently controlled by the two-dimensional (2D) template and the one-dimensional (1D) arrangement of the oriented, NP architecture. Because of the oligonucleotide bonding elements that hold them in place, these architectures undergo reversible structural changes in response to chemical stimuli, allowing interactions with visible light to be dynamically tuned.

64.2

Methods

To arrange colloidal NPs into desired architecture on a surface (Fig. 64.1), we used electron-beam lithography (EBL) to pattern a uniform array of pores into a 300-nm-thick layer of poly(methyl methacrylate) (PMMA) affixed to a gold-coated silicon substrate [9]. The selectively exposed gold surfaces at the bottom of each pore were then densely functionalized with oligonucleotides bearing a terminal propylthiol. Complementary oligonucleotides were then hybridized to the single-stranded DNA at the base of each pore to yield a monolayer of rigid, double-stranded DNA with a short singlestranded region, or “sticky end,” at the solution-facing terminus.

1197

1198

Building Superlattices from Individual Nanoparticles

Figure 64.1 Programmable assembly of reconfigurable nanoparticle (NP) architectures. To assemble NP architectures within a confined environment, 1D pores are fabricated in a PMMA-coated gold substrate using top-down lithography, and the gold surface at the bottom of each pore is densely functionalized with DNA. DNA-functionalized colloidal NPs of controlled size and shape are then assembled in a layer-by-layer fashion by designing each layer of NPs to have a terminal DNA sequence complementary to that of the previous layer. The porous PMMA template is removed to generate NP superlattices with 2D periodicity that are composed of oriented NP architectures. Bottom images depict cross-sectional views of a single pore.

Colloidal gold NPs of different shapes and sizes were similarly modified with DNA. Architectures of DNA functionalized NPs were then assembled within each pore of the PMMA template in a layerby-layer fashion by designing the sticky-end DNA sequence present on a selected NP to be complementary to that of the previous layer (Fig. 64.1). To achieve a predictable, thermodynamically favored arrangement of NPs in each layer, the assembly process must be governed by the maximization of canonical Watson–Crick DNA hybridization events between complementary DNA sticky ends-a principle known as the complementary contact model [3, 15]. For this interaction to dominate, we needed to mitigate the effects of other nonspecific interactions that compete at low temperatures, such as interactions between noncomplementary nucleic acids and between DNA and the PMMA template. Anisotropic NPs with flat facets and adenine- or guaninecontaining sticky ends are particularly prone to a range of noncanonical interactions [8, 16]. The impact of these interactions

Methods

can be reduced by performing the assembly at higher temperatures, but high temperatures may also lead to the de hybridization and desorption of NP layers that have already been assembled [8]. Indeed, finding a suitable assembly temperature for the synthesis of architectures in both high yield and high purity proved to be challenging when sticky ends were composed of conventional nucleic acids. To increase the strength of Watson–Crick DNA hybridization interactions between sticky ends relative to noncanonical ones, we replaced three adenine nucleotides in sticky-end sequences with “locked” versions containing the same base. Locked nucleic acids (LNA) are modified RNA nucleotides in which the ribose group is rigidified by connecting the 2’ oxygen to the 4’ carbon with a methylene bridge [17, 18]. This modification reduces conformational flexibility and increases the strength of canonical base-pairing interactions. Incorporating just three LNA bases into the sticky-end sequences of nanocubes increased the melting temperature associated with canonical hybridization interactions by 9°C, while also decreasing the melting temperature for noncanonical interactions by 1°C. This 10°C greater window for NP assembly enabled the predictable synthesis of highly uniform superlattices composed of one-, two-, and three-layer NP architectures. In addition to DNA, the size, shape, depth, and arrangement of PMMA pores provide a critical element of structural control during NP assembly. For example, the depth of each pore, which was determined by the thickness of PMMA, provided a confined environment for the assembly of each NP layer. We could then build ID architectures a single NP at a time exclusively in a direction perpendicular to the substrate. Additionally, NPs would not assemble in a pore smaller than the NP, whereas multiple NPs would assemble in larger pores. Choosing a pore size slightly larger than the total size of NP and DNA thus allowed us to assemble uniform monolayers of different size and shape NPs into arrays (Fig. 64.2) [9]. The pore shape further offers the ability to align the average orientation of anisotropic NPs, as observed for cubes assembled in square pores and triangular prisms assembled in triangular pores (Fig. 64.2B,C). After assembly, the porous PMMA template could be dis solved and the NP superlattices transferred intact to the solid state for imaging by scanning electron microscopy (SEM) (Fig. 64.1).

1199

1200

Building Superlattices from Individual Nanoparticles

Figure 64.2 Monolayer of DNA-functionalized gold NPs assembled in PMMA templates with different shaped pores. (A) Circular pores (left) are used to assemble a monolayer of gold spheres 60 and 100 nm in diameter (right). (B) Square pores (left) are used to assemble a monolayer of gold cubes 55 and 80 nm in edge length (right). (C) Triangular pores (left) are used to assemble a monolayer of gold triangular prisms 90 and 165 nm in edge length (right). Note that all SEM images of assembled nanoparticles are shown after the removal of the PMMA template. Scale bars: 200 nm.

In addition to monolayers of isolated NPs, we used templateconfined, DNA-mediated assembly to synthesize superlattices with 2D periodicity featuring ten different unit cells that each consist of a ID architecture of either two or three NPs oriented normal to the surface (Fig. 64.3A). These ID architectures include building blocks of the same size and shape, building blocks of decreasing size, and building blocks of different sizes and shapes. Superlattices of these low-symmetry architectures were synthesized over areas of at least 600 µm by 600 µm with high uniformity (Fig. 64.3B,D). Grazing-incidence small-angle X-ray scattering (GISAXS) confirmed that the superlattices exhibited the expected diffraction patterns—lines in reciprocal space—of a material with 2D periodicity

Methods

Figure 64.3 Synthesis of oriented superlattices of two- and three-layer NP architecture. (A) Scanning electron microscopy (SEM) images of oriented superlattices (periodicity = 500 nm) attached to a gold surface after removal of the PMMA template. The superlattices are composed of 1D architectures of gold NPs in the order of disk-cube, cube-cube, prism-cube, disk-cube-sphere, cube-cube-sphere, prism-cube-sphere, disk-cube-cube, cube-cube-cube (same size), and cube-cube-cube (decreasing sizes). Scale bar: 300 nm. (B) Large-area SEM image of the disk-cube-sphere superlattice. Inset: fast Fourier transform (FFT) pattern of the SEM image. Scale bar: 4 μm. (C) Grazing-incidence smallangle X-ray scattering (GISAXS) pattern of the disk-cube-sphere superlattice. (D) Defects in the disk-cube-sphere superlattice were quantified through analysis of SEM images that contained 773 individual unit cells within the superlattice.

(Fig. 64.3C). This bottom-up, layer-by-layer assembly process overcame several challenges in the fabrication of multilayer architectures with conventional top-down techniques, such as focused

1201

1202

Building Superlattices from Individual Nanoparticles

ion beam (FIB) milling [19] or EBL stacking [20], which require time-consuming, small-scale cutting of materials or complex serial alignment, exposure, development, and metal evaporation to form each layer. Here, biomolecular interactions operating at molecular length-scales allow for the rapid, automatic, and precise alignment of each NP layer with tunable interlayer distances.

64.3

Results and Discussion

Not surprisingly, these NP superlattices contained several types of defects analogous to thermodynamically inevitable defects in atomic crystals, including “vacancies,” where a NP was missing from a superlattice position, and “interstitials,” where an extra NP was present within the superlattice [21]. Each individual architecture within the superlattice could be resolved by electron microscopy, so these NP defects can be rigorously quantified and used to establish structure-property relations. For example, a representative diskcube-sphere three-layer superlattice of 773 architectures contained 73% defect-free structures, 1% vacancies, 9% interstitials, and 17% other defects (Fig. 64.3D). In general, the number of defects tended to increase as the number of potential competing interactions increased in moving from one to two to three layer structures. Note that the SEM images depict snapshots of the structures in the perturbed solid-state, a consequence of drying, and the solutionbased architectures likely exhibit greater uniformity. In addition to directing NP assembly, oligonucleotide bonds between NPs are dynamic and undergo reversible contractions and expansions in response to changes in solvent polarity that allow the distance between NPs to be precisely tuned [22]. Although it is a challenge to accurately measure the distance between NPs of different sizes and shapes in isolated architectures, cross-sectional SEM images of disk-cube-sphere architectures, which were encased in silica to attempt to preserve their solution-phase arrangements [22], showed that the average distance between NPs decreased from >12 nm to 690 nm), affording both very large wavelength tunability and amplitude modulation (Fig. 64.4B,C). To experimentally confirm the predicted optical response, the computationally identified NP superlattice—composed of discrete disk-cube-sphere architectures—was synthesized. Absorption spectra were measured with an inverted optical microscope for the superlattice coated with a thin layer of solvent with increasing ratios of EtOH to H2O (Fig. 64.4E). As predicted, changing the average coupling distance be tween NPs led to dramatic changes in the absorption spectra, which are in excellent agreement with simulations—indicating that any defects or inhomogeneities present

Results and Discussion

in the assembled superlattice do not significantly affect its optical properties—and confirmed the high tunability of this reconfigurable absorber (Fig. 64.4E,F). Specifically, the superlattice had a 75% increase in the average absorption of light from 550 to 800 nm when decreasing solvent polarity—and decreasing average gap lengths— from 0 to 80% EtOH in H2O, with a maximum increase of 443% at 732 nm (increased ab sorption from 14 to 73%). As expected, the changes in optical response could also be observed visually as the color of the surface changed from maroon to dark green to brown as gap lengths decreased (Fig. 64.4G). In addition to the amplitude, the absorption band edge λedge, could be tuned from 650 nm (1.9 eV) to 775 nm (1.6 eV) (Figs. 64.4E,F). The 125 nm shift in wavelength represents a wavelength tuning figure of merit (wavelength shift divided by the initial band edge wavelength prior to tuning) of 19%. The tunability reported here exceeds that of stretchable substrates, which have exhibited wavelength tuning of up to 11% in the longer wavelength infrared region [29], and is comparable to that achieved recently using temperature responsive polymers [30]. Importantly, all structural changes, and the changes they induce in optical properties, were reversible for at least five cycles between large and small gaps, with no appreciable changes to absorption spectra observed (Fig. 64.4H). The ability to control the arrangement, spacing, and sequence of NPs within each architecture is critical to the realization of tunable broadband absorption. Indeed, superlattices with other sequences of disk, cube, and sphere AuNP architectures are predicted to exhibit very different optical responses with considerably reduced tunability. Beyond tunable absorption, the ability to make responsive plasmonic nanoarchitectures not yet achievable via other techniques should dramatically increase the diversity of structures and compositions that can now be explored by theorists and experimentalists to access new and useful optical properties. It should be possible to synthesize even more sophisticated architectures through the use of more intricate pore designs, and new DNA sequence designs should enable responsiveness to be extended to light and biological signals, in addition to chemical ones. Additionally, although we have only synthesized three-layer architectures here, the number of NP layers could in principle be increased by using deeper PMMA pores.

1205

1206

Building Superlattices from Individual Nanoparticles

Acknowledgments This material is based upon work supported by the Center for Bio-Inspired Energy Science, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences under Award DE-SC0000989 and the Air Force Office of Scientific Research under Award Numbers FA9550- 12-l0280, FA9550-14-l-0274, and FA9550-17-l-0348. Use of the Center for Nanoscale Materials, an Office of Science user facility at Argonne National Laboratory, and GISAXS experiments at beamline 121D-B at the Advanced Photon Source (APS) at Argonne National Laboratory, were supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DEACO2-06CH11357. This work made use of the EPIC facility of the NUANCE Center at Northwestern University, which has received support from the Soft and Hybrid Nanotechnology Experimental (SHyNE) Resource (NSF NNCI-1542205); the MRSEC group (NSF DMR-1121262) at the Materials Research Center; the International Institute for Nanotechnology (IIN); the Keck Foundation; and the State of Illinois, through the IIN, Q.-Y.L., Z.L., and M.R.J. gratefully acknowledge support from the Ryan Fellowship at Northwestern University, and M.N.O gratefully acknowledges the National Science Foundation for a Graduate Research Fellowship. We thank C. Laramy and H. Lin for assistance with some nanoparticle syntheses. The authors declare no competing financial interests.

References

1. Mirkin, C. A., Letsinger, R. L., Mucic, R. C. and Storhoff, J. J. (1996). Nature, 382, 607–609. 2. Nykypanchuk, D., Maye, M. M., van der Lelie, D. and Gang, O. (2008). Nature, 451, 549–552.

3. Jones, M. R., Seeman, N. C. and Mirkin, C. A. (2015). Science, 347, 1260901. 4. Wang, Y., Wang, Y., Breed, D. R., Manoharan, V. N., Feng, L., Hollingsworth, A. D., Weck, M. and Pine, D. J. (2012). Nature, 491, 51–55.

5. Liu, W., Tagawa, M., Xin, H. L., Wang, T., Emamy, H., Li, H., Yager, K. G., Starr, F. W., Tkachenko, A. V. and Gang, O. (2016). Science, 351, 582– 586.

References

6. Rogers, W. B., Shih, W. M. and Manoharan, V. N. (2016). Nat. Rev. Mater., 1, 16008. 7. O’Brien, M. N., Lin, H.-X., Girard, M., Olvera de la Cruz, M. and Mirkin, C. A. (2016). J. Am. Chem. Soc., 138, 14562–14565.

8. O’Brien, M. N., Radha, B., Brown, K. A., Jones, M. R. and Mirkin, C. A. (2014). Angew. Chem. Int. Ed., 53, 9532–9538.

9. Lin, Q.-Y., Li, Z., Brown, K. A., O’Brien, M. N., Ross, M. B., Zhou, Y., Butun, S., Chen, P.-C., Schatz, G. C., Dravid, V. P., Aydin, K. and Mirkin, C. A. (2015). Nano Lett., 15, 4699–4703.

10. Wang, M. X., Seo, S. E., Gabrys, P. A., Fleischman, D., Lee, B., Kim, Y., Atwater, H. A., Macfarlane, R. J. and Mirkin, C. A. (2017). ACS Nano, 11, 180–185.

11. Henzie, J., Andrews, S. C., Ling, X. Y., Li, Z. and Yang, P. (2013). Proc. Natl. Acad. Sci. U.S.A., 110, 6640–6645.

12. Liu, X., Biswas, S., Jarrett, J. W., Poutrina, E., Urbas, A., Knappenberger Jr., K. L., Vaia, R. A. and Nealey, P. F. (2015). Adv. Mater., 27, 7314–7319.

13. Flauraud, V., Mastrangeli, M., Bernasconi, G. D., Butet, J., Alexander, D. T. L., Shahrabi, E., Martin, O. J. F. and Brugger, J. (2017). Nat. Nanotechnol., 12, 73–80. 14. Hughes, R. A., Menumerov, E. and Neretina, S. (2017). Nanotechnology, 28, 282002.

15. Macfarlane, R. J., Lee, B., Jones, M. R., Harris, N., Schatz, G. C. and Mirkin, C. A. (2011). Science, 334, 204–208.

16. Bloomfield, V. A., Crothers, D. M. and Tinoco, I. (2000). Nucleic Acids: Structures, Properties, and Functions (University Science Books, Sausalito, CA). 17. Koshkin, A. A., Singh, S. K., Nielsen, P., Rajwanshi, V. K., Kumar, R., Meldgaard, M., Olsen, C. E. and Wengel, J. (1998). Tetrahedron, 54, 3607–3630. 18. Owczarzy, R., You, Y., Groth, C. L. and Tataurov, A. V. (2011). Biochemistry, 50, 9352–9367. 19. Valentine, J., Zhang, S., Zentgraf, T., Ulin-Avila, E., Genov, D. A., Bartal, G. and Zhang, X. (2008). Nature, 455, 376–379.

20. Liu, N., Guo, H., Fu, L., Kaiser, S., Schweizer, H. and Giessen, H. (2008). Nat. Mater., 7, 31–37. 21. Boles, M. A., Engel, M. and Talapin, D. V. (2016). Chem. Rev., 116, 11220–11289.

1207

1208

Building Superlattices from Individual Nanoparticles

22. Mason, J. A., Laramy, C. R., Lai, C.-T., O’Brien, M. N., Lin, Q.-Y., Dravid, V. P., Schatz, G. C. and Mirkin, C. A. (2016). J. Am. Chem. Soc., 138, 8722– 8725.

23. Halas, N. J., Lal, S., Chang, W.-S., Link, S. and Nordlander, P. (2011). Chem. Rev., 111, 3913–3961.

24. Urbas, A. M., Jacob, Z., Negro, L. D., Engheta, N., Boardman, A. D., Egan, P., Khanikaev, A. B., Menon, V., Ferrera, M., Kinsey, N., DeVault, C., Kim, J., Shalaev, V., Boltasseva, A., Valentine, J., Pfeiffer, C., Grbic, A., Narimanov, E., Zhu, L., Fan, S., Alù, A., Poutrina, E., Litchinitser, N. M., Noginov, M. A., MacDonald, K. F., Plum, E., Liu, X., Nealey, P. F., Kagan, C. R., Murray, C. B., Pawlak, D. A., Smolyaninov, I. I., Smolyaninova, V. N. and Chanda, D. (2016). J. Opt., 18, 093005. 25. Zheludev, N. I. and Plum, E. (2016). Nat. Nanotechnol., 11, 16–22.

26. Qian, Z. and Ginger, D. S. (2017). J. Am. Chem. Soc., 139, 5266–5276.

27. Moreau, A., Ciracì, C., Mock, J. J., Hill, R. T., Wang, Q., Wiley, B. J., Chilkoti, A. and Smith, D. R. (2012). Nature, 492, 86–89.

28. Akselrod, G. M., Huang, J., Hoang, T. B., Bowen, P. T., Su, L., Smith, D. R. and Mikkelsen, M. H. (2015). Adv. Mater., 27, 8028–8034.

29. Pryce, I. M., Aydin, K., Kelaita, Y. A., Briggs, R. M. and Atwater, H. A. (2010). Nano Lett., 10, 4222– 4227.

30. Ding, T., Valev, V. K., Salmon, A. R., Forman, C. J., Smoukov, S. K., Scherman, O. A., Frenkel, D. and Baumberg, J. J. (2016). Proc. Natl. Acad. Sci. U.S.A., 113, 5503–5507.

Chapter 65

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly*

Wenjie Zhou,a,b Qing-Yuan Lin,c Jarad A. Mason,a,b Vinayak P. Dravid,c and Chad A. Mirkina,b,c aDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA bInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA [email protected]

Template-based strategies are becoming increasingly important for controlling the position of nanoparticle-based (NP-based) structures on surfaces for a wide variety of encoding and device fabrication strategies. Thus, there is an increasing need to understand the behavior of NPs in confined spaces. Herein, a systematic investigation of the diffusion and adsorption properties of DNA-modified NPs is presented in lithographically defined, high-aspect-ratio pores using a template-confined, DNA-mediated assembly. Leveraging

*Reprinted with permission from Zhou, W., Lin, Q.-Y., Mason, J. A., Dravid, V. P. and Mirkin, C. A. (2018). Design rules for template-confined DNA-mediated nanoparticle assembly, Small 14, 1802742. Copyright © 2018, John Wiley and Sons. Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1210

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly

the sequence-specific binding affinity of DNA, it is discovered that although NP adsorption in deep polymer pores follows a traditional Langmuir adsorption model when under thermodynamic control, such NPs kinetically follow Fick’s classical law of diffusion. Importantly, these observations allow one to establish design rules for template-confined, DNA-mediated NP assembly on substrates based on pore dimensions, NP size and shape, NP concentration, temperature, and time. As a proof-of-concept example, these design rules are used to engineer a vertical, four-layer assembly consisting of individual octahedral NPs stacked on top of one another, with inplane positioning defined by pores generated by e-beam lithography.

65.1

Introduction

Nanoparticles (NPs), due to their small size, well-defined shapes [1–4], and unusual properties [5–16], are promising ideal building blocks for constructing higher-ordered materials [10, 17–19], DNA, due to its sequence programmability and adjustable length, has become a versatile tool for making highly ordered materials from NP building blocks [20–22], both in solution and on surfaces [17, 19, 20, 22–29]. Indeed, we and others have used programmable assembly to map out the design space for NP superlattices of over 500 different crystal types spanning over 40 different lattice symmetries and 4 different crystal habits [20–22, 24–27]. Moreover, we recently showed that a wide variety of structures not attainable through conventional top-down lithographic methods could be made using a combination of top-down lithographic and bottomup DNA-programmable assembly [17, 19, 29], thereby establishing a platform for the metamaterial community to consider a more diverse set of architectures to discover, design, and realize unusual NP-based materials with desirable properties. The design and synthesis of these new colloidal NP surface architectures will benefit from an improved understanding of the fundamental diffusion and adsorption behaviors of NPs on surfaces and in confined environments. Although the behavior of solution-dispersed NPs has been extensively studied using wellestablished theoretical [30] and analytical techniques [31–33], there are comparatively few studies [34, 35] of NP interactions with surfaces due to the complexity of interactions involved and a lack of reasonable experimental models. Recently, we found that despite

Introduction

the complex interactions between NPs and surfaces, NP adsorption on flat surfaces can be described by a simple Langmuir adsorption isotherm [36], which is normally used to model molecular surface adsorption. This work established an initial understanding of the collective interactions between ligand-functionalized NPs and a flat surface with complementary binding sites. In less ideal situations, flat-surface-bound binding sites cannot be guaranteed, and a substantial number of NP-surface interactions involve degrees of physical confinement on the NP. For instance, in directed NP assembly systems [17–19, 29], NP adsorption—and the quality of the final structure—is not only dependent on the NP interactions with lithographically defined pores, but also with the surface at the base of the pores. Such interactions are challenging, if not impossible, to directly observe in situ with existing characterization techniques. Therefore, in order to understand the diffusion and adsorption of NPs in porous environments, a suitable model or platform to investigate such behavior is required. The behavior of NPs is strongly governed by Brownian motion; hence tracking individual NP interactions with a surface covered by a porous template is experimentally challenging. Given the difficulty of studying such interactions in situ, we adopted a recently developed template-confined, DNA-mediated assembly technique to investigate NP adsorption thermodynamics and kinetics [19]. This technique involves using electron beam lithography (EBL) to fabricate 1D pore channels at fixed positions in a poly(methyl methacrylate) (PMMA) polymer thin film on a gold-coated silicon substrate with a specific DNA sequence covalently attached to the exposed gold at the base of each pore. This approach offers a highly programmable system where critical diffusive and adsorptive parameters such as pore diameter, pore depth, NP shape, NP concentration, temperature, and time can be precisely and independently defined. Herein, a statistical approach was designed and utilized to analyze and describe NP diffusion and adsorption in porous templates. Briefly, we investigate the relationships between adsorption yield and NP size, NP concentration, pore size, pore depth, temperature, and time. The diffusion and adsorption behaviors of NPs were studied under both equilibrium and nonequilibrium conditions. From this analysis, we show how pore diameter and depth are directly related to particle adsorption yield as well as the time required to reach equilibrium. Remarkably, despite the complexity of the system and the large size of the NP building blocks (compared with the size of

1211

1212

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly

molecules), particle diffusion and adsorption in this system are well described and modeled using Fick’s diffusion law and the Langmuir adsorption model, thereby making it possible to predict both the kinetic and thermodynamic behavior of NPs in pores.

65.2

Results and Discussion

In a typical experiment, 1D pore channels were designed and fabricated using EBL to pattern nanoscale circular pores ranging from 40 to 220 nm in diameter and 100 to 600 nm in depth in PMMA thin films deposited on gold-coated silicon substrates (Fig. 65.1) [17, 19]. Subsequently, DNA-mediated NP assembly (diffusion and adsorption) within these pores was studied over a temperature range of 25 to 45°C and time periods of 0.5 to 96 h. After assembly of the DNA-modified NPs, the PMMA template was removed to allow the NP arrays to be transferred intact from the solution phase to the Noccupied solid state. The adsorption yield, θ, where θ = (Noccupied Ntotal is the number of NP-occupied binding sites, while Ntotal is the total number of potential binding sites), was calculated based upon NP position data taken from scanning electron microscopy (SEM) measurements (Fig. 65.1). In this system, pore size plays a key role in controlling the degree of physical confinement. Qualitatively, NPs cannot diffuse into pores whose diameters are too narrow, while pores that are too wide tend to be filled with multiple NPs. However, a quantitative relationship between pore size and the adsorption yield at the single-particle level remains unknown. To systematically address this problem, we performed a series of experiments under equilibrium conditions, where only the pore size was varied. To ensure the system was in an equilibrium state, each experimental condition was carried out for multiple lengths of time, and the adsorption yields were calculated. When an increase in time no longer led to a measurable increase in yield, the system was deemed to have reached equilibrium. Adsorption yield is plotted as a function of pore size for various sizes of NPs and exhibits a sigmoidal shape with a sharp transition (Fig. 65.2). It should be noted that the DNA sequences [17, 19] used in this experiment add ~20 nm to the diameter of the gold core for each

Results and Discussion

NP. Interestingly, the minimum pore diameter required to achieve 100% adsorption yield is consistently ~1.4 times the diameter of the DNA-functionalized spherical NP. This is a general guideline to maximize adsorption yield while maintaining positional control. In addition, the transition from 0% to 100% yield occurs across a consistent pore size to NP diameter (hydrodynamic diameter, which includes the ligand shell) ratio of 1.1–1.4 when the pore depth is at least 200 nm. Importantly, this indicates that in order for NPs to diffuse through polymer nanopores, a pore size of at least 1.1 times the total size of a NP (including ligand shell) is required. However, to reach 100% occupancy, pore sizes of at least 1.4 times the NP size are necessary.

Figure 65.1 Scheme and SEM characterization of template-confined, DNAmediated NP assembly. (a) Arrays of nanoscale pores are fabricated using EBL, after which both NPs and the bottom of each pore are densely functionalized with thiolated DNA (red and blue indicates complementary DNA sequences). By designing the NPs to have a terminal DNA sequence complementary to that at the bottom of pores, NPs are adsorbed onto surfaces at precisely defined locations. The porous polymer templates are removed to generate solidstate NP arrays after assembly. (b) SEM images of polymer templates (left), NPs assembled in the templates (middle), and NP arrays with the templates removed (right). Note that three images were captured from three separated samples prepared with the same experimental conditions. Scale bars: 500 nm.

1213

1214

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly

Figure 65.2 Effect of pore diameter on assembly yield. (a) Scheme and SEM images of 80 nm diameter spherical NPs assembled into nanopores with different sizes (pore diameters = 100, 130, and 200 nm, from left to right) under equilibrium conditions. Scale bar: 500 nm. (b) Assembly yield as a function of pore diameter in which 60, 80, and 100 nm diameter spherical NPs were assembled at equilibrium. All pore depths are 200 nm, assembly temperature is 25°C, NP concentrations are 150 pM, and assembly times are longer than 48 h. Each data point represents an average of two independent experiments, 900 sites were counted for each experiment, with an error bar as the standard deviation.

In addition to pore diameter, pore depth also affects the physical confinement of the NPs, with deeper pores expected to provide stronger confinement. However, it is unknown whether pore depth affects the thermodynamics of NP adsorption. It was previously shown that DNA-mediated NP adsorption on flat surface follows the Langmuir adsorption model in the equilibrium state [36], but it is unclear if the same model applies under strong physical confinement in pores. In the Langmuir adsorption model, the relationship of surface coverage and adsorption parameters is described as c(1 − θ) ka = θ kd, where c is the concentration of NPs; ka and kd are the adsorption and desorption rate constants, respectively. By further k K Lc . If defining KL = a , the equation can be simplified to θ = 1 + K Lc kd the NP concentration in solution is held constant, ka should describe both adsorption to the DNA-mediated binding sites and diffusion

Results and Discussion

through the pore, because adsorption events occur at the bottom of pores. Likewise, desorption processes combine both desorption from binding sites and diffusion out of the pore. It is unclear whether changing the diffusion length will affect ka, or kb, and thus KL, as well as the relationship between pore size and adsorption yield.

Figure 65.3 Thermodynamic analysis of NP assembly in pores. a) Assembly yield as a function of pore size. Spherical NPs (diameter = 80 nm, concentration = 150 pM) were adsorbed in nanopores (depth = 200, 300, and 400 nm) at a temperature of 25°C for at least 96 h. b) Assembly yield as a function of NP concentration. Spherical NPs (diameter = 80 nm) were adsorbed in nanopores (diameter = 140 nm, depth = 200 and 400 nm) at a temperature of 25 and 45°C for at least 96 h. Each data point represents an average of two independent experiments, 900 sites were counted for each experiment, with an error bar as the standard deviation.

To elucidate the effect of pore depth on the thermodynamics of NP adsorption, we carried out experiments with pores of various depths (H = 100, 200, 300, 400 nm) and diameters (110 and 140 nm). For 80 nm diameter NPs, we found consistent adsorption yields for both small (110 nm) and large diameter pores (140 nm), regardless of pore depth. These experiments show that, for the conditions explored here, θ is independent of pore depth (Fig. 65.3a). This result indicates that under equilibrium conditions, pore depth does not alter the thermodynamics of NP adsorption. To further test whether KL varies with pore depth, we investigated θ as a function of NP concentration. The results show that NP adsorption in high-aspect-ratio pores still fits the Langmuir adsorption model (Fig. 65.3b). In fact, KL values for different pore depths at temperatures below the DNA melting temperature are nearly constant. This further indicates that pore depth does not influence the equilibrium result

1215

1216

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly

of NP surface adsorption. Given sufficient time to reach equilibrium, NPs diffuse through deep pores with high aspect ratios, indicating that deep pores can be used to provide a high degree of physical confinement without compromising adsorption yield. While pore depth does not affect the thermodynamics of NP adsorption, it does significantly affect the kinetics of NP adsorption, where deeper pores increase the time required to reach equilibrium. To quantify how the kinetics of NP diffusion are affected, we designed a series of ex situ experiments to investigate the relationship between θ and assembly time for pore depths of 100, 200, 300, 400, and 600 nm (Fig. 65.4a).

Figure 65.4 Kinetic analysis of NP diffusion in pores. (a) Assembly yield as a function of assembly time. Spherical NPs (diameter = 80 nm, concentration = 150 pM) were adsorbed in nanopores (diameter = 140 nm, depth = 100, 200, 300, 400, and 600 nm) at 25°C for varying time periods. Each data point represents an average of two independent experiments, with an error bar as the standard deviation. (b) The time required to reach 20%, 40%, 60%, 80%, and 90% yields is plotted as a function of the square of pore depth.

Importantly, we found that the diffusion of NPs during adsorption in nanopores can be rationalized using Fick’s laws of diffusion, which describe the diffusion flux of molecules from high-concentration (high chemical potential) to low-concentration regions (low chemical potential). In our template-confined DNA-mediated assembly system, each pore effectively represents a 1D channel for NP diffusion. As soon as the nanopores are immersed in a colloidal NP solution, the NPs begin to diffuse from a high-concentration region (solution) to a low—effectively zero—concentration region (the bottom of each pore). It should be noted that the interactions between the DNA shell and PMMA walls are assumed to be collision

Results and Discussion

only. Such an assumption is applicable in DNA-mediated surface assembly, where DNA:NP size ratios are designed to be small in order to maintain the shape of NPs. Using a simplified 1D diffusion model, the NP concentration at the bottom of each pore can be calculated using Fick’s law [37]: Ê H ˆ c ( H , t ) = co erfc Á Ë 2 Dt ˜¯

(65.1)

where c represents the diffusive concentration at a distance H from the environment (top of pore) after a dissolution time, t, co is the environmental concentration, D is the diffusion coefficient, and erfc is the complementary error function. As previously discussed, template-confined, DNA-mediated assembly satisfies the Langmuir adsorption model, which indicates that the same surface coverage is determined by the NP concentration, strength of interaction between the NP and surface, ratio of pore diameter to NP size, and temperature, but not by the pore depth. As a result, the adsorption H yield for various pore depths, must be constant (H > 0, θ ≤ 1), 2 Dt indicating that the assembly time must be directly proportional to the square of the pore depth (i.e., t µ H2). We calculated the assembly time required to reach 20%, 40%, 60%, 80%, and 90% yield for pores of 100, 200, 300, 400, and 600 nm in depth, respectively (Fig. 65.4a). As predicted by Fick’s law, the results show that the assembly time is indeed proportional to the square of pore depth (Fig. 65.4b). This confirmed that, despite the complexity of the system, classical molecular diffusion models can be utilized to describe the diffusive behavior of NPs, improving our fundamental understanding of NP behaviors in confined environments [17, 19, 38]. While we focused on spherical NPs, the adsorption behavior can be extended to anisotropic NPs. Specifically, we found that the effective diameters of sphere-like anisotropic NPs (as compared with their spherical counterparts) are described by the diameter of the smallest sphere required to circumscribe the anisotropic NP. For cubic NPs, the effective diameter is 3 times the cube edge length; while, for octahedral NPs, the effective diameter is 2 times the edge length. Equilibrium adsorption yields were plotted as functions of pore sizes for spherical NPs with diameters of 80 nm, cubic NPs

1217

1218

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly

with edge lengths of 80 nm, and octahedral NPs with edge lengths of 80 nm (Fig. 65.5). Cubes of this size should effectively behave as a sphere with a diameter of 140 nm, and therefore the minimal pore size required for 100% yield should be 215 nm (including a 20 nm DNA shell); for octahedra, the corresponding pore size should be 180 nm. The experimental results agree closely with the expected pore size (Fig. 65.5). Such behavior is primarily under the control of rapid Brownian motion; the anisotropy of NPs plays a negligible role in the interaction between NPs and pores.

Figure 65.5 Effect of NP shape on assembly yield. Assembly yield as a function of pore diameter with different shapes of NPs assembled at equilibrium. Assembled NPs are spherical NPs with 80 nm in diameter, octahedron NPs with 80 nm in edge length, cubic NPs with 80 nm in edge length. The effective sizes of anisotropic NPs are essentially the diameter of their circumscribed spheres, which is the effective shape under strong Brownian motion in solution. All pore depths are 200 nm, assembly temperature is 25°C, assembly times are longer than 48 h. Each data point represents an average of three independent experiments, 400 sites were counted for each experiment, with the error bar as the standard deviation.

As a proof-of-concept, we show how the fundamental relationships elucidated in this work can be leveraged to synthesize sophisticated nanostructures with a high degree of structural control. Specifically, we targeted the assembly of four-layer NP architectures (Fig. 65.6). Gold octahedra with an average edge length of 80 nm were synthesized and functionalized with DNA. To direct

Results and Discussion

NP assembly, PMMA nanopores (180 nm diameter, 400 nm depth) were patterned on a gold-coated Si substrate. DNA-mediated layerby-layer assembly was then used to grow the 1st (60 h), 2nd (48 h), 3rd (24 h), and 4th (12 h) layers in successive fashion. Less time was required for each successive layer since the pore depth decreases with the addition of each layer. SEM characterization confirmed that the NPs assembled in high yield (>90%), with near perfect facet registry among the two-, three-, and four-layer structures. As compared with the previous three-layer structures assembled using this technique [19], the assembly yield is significantly higher than the reported results (~70%). Therefore, this fundamental study of NP adsorption thermodynamics and kinetics allows one to significantly improve the yield of multilayer NP architectures and should allow researchers to program and realize even more sophisticated structures in the future.

Figure 65.6 Template-confined vertical assembly of one-, two-, three-, and four-layer architectures of octahedral NPs. Guided by PMMA nanopores (diameter = 165 nm, depth = 400 nm), octahedral NPs (edge length = 80 nm) are assembled layer-by-layer through DNA-mediated assembly. NP architectures show good facet registry, which leads to high stability of assembled structures. Different colors represent octahedral NPs capped with different DNA sequences, where blue is complementary to red. Scale bars: 50 nm, tilting angle of SEM images: 70°.

1219

1220

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly

65.3

Conclusion

In conclusion, this work provides valuable insight into the thermodynamic and kinetic considerations of colloidal NP assembly, which can be generalized for other systems where the interactions between NPs and pores play a crucial role [39–46]. The design rules elucidated by this work will significantly advance the level of structural control that is possible when using NPs as building blocks for generating structurally and compositionally sophisticated and complex particle-based architectures [19]. With the diversity of NP shapes, sizes, and compositions that can serve as building blocks [47], the lessons learned from this work will be extremely valuable for the design, synthesis, and integration of NP-based materials relevant to applications in biology [39, 43], optics [17, 19], and catalysis [44, 45].

Rule 1: Minimal NP concentration for a desired adsorption yield can be predicted for specific NPs, since the relationship between adsorption yield in pores and NP concentration at equilibrium can be K Lc . described by the Langmuir adsorption equation, where θ = 1 + K Lc Rule 2: Minimal required assembly time can be predicted for pores of specific depths, since the relationship between assembly time and pore depth at fixed adsorption yield follows Fick’s law, where t µ H2. Rule 3: To maximize NP adsorption, the pore radius should be ≥1.4 times the hydrodynamic radius of the adsorbing NPs.

Rule 4: For pseudo-spherical anisotropic NPs (e.g., cubes and octahedra), the size can be approximated based upon the diameter of the smallest sphere required to circumscribe such NPs. Rule 5: Equilibrium adsorption yield is independent of pore depth.

65.4

Experimental Section

Spherical gold NPs capped with citrate ligands with nominal diameters of 60, 80, and 100 nm were purchased from Ted Pella, Inc. and used as received. Single-crystalline gold nanocubes and octahedrons were synthesized via the seed-mediated method

Experimental Section

described in detail by O’Brien et al. [4]. Briefly, uniform, singlecrystalline spherical gold NPs were synthesized via an iterative chemical refinement process. These NPs were subsequently used as “seeds” to template the growth of cubes. Importantly, this synthetic procedure results in structurally uniform anisotropic NPs produced with high yield. PMMA pores were fabricated on Au-coated Si substrates with EBL. First, a Si wafer was cleaned via an O2 plasma (~0.2 mbar) at 50 W for 5 min. Following plasma cleaning, a 5 nm Cr adhesion layer, followed by a 100 nm Au layer were deposited onto the Si substrate with electron beam evaporation, at a rate of 0.25 Å s−1 (Kurt J. Lesker Company). The wafer was then cut into smaller pieces (1.5 × 1.5 cm2) and stored in vacuum desiccator. Substrates were then spin-coated with positive e-beam resist PMMA. The PMMA was baked at 200°C for 60 s, followed by EBL (FEI Quanta 650 ESEM) to define the size and position of the pores. An accelerating voltage of 30 kV with a dosage of 200–1500 µC cm−2 was used to pattern pores with 90–260 nm diameters. The substrates were developed in a methyl isobutyl ketone/ isopropyl alcohol (IPA) 1:3 solution for 60 s, rinsed with IPA, and blown dry with N2. After development, each substrate was cleaned via O2 plasma (~0.2 mbar) at 50 W for 1 min to remove potential PMMA residue at the bottom of pores and cut into four smaller pieces to fit in 2 mL Eppendorf tubes. In brief, we functionalized substrates and NPs with orthogonal 3¢ thiolated DNA sequences, denoted as X and Y, respectively. Then, complementary linker DNA strands were hybridized to both the NPs and the substrate to provide rigidity to the DNA shell. Each linker strand possessed a short 5-base terminus designed to link the NPs to the substrates through complementary DNA hybridization events. NPs were functionalized as described by Jones et al. [25] and O’Brien et al. [36]. In brief, 3¢ alkylthiol-modified oligonucleotides were first treated with a 100 × 10−3 m solution of dithiothreitol (DTT) in 170 × 10−3 m sodium phosphate buffer (pH = 7.4), followed by purification on a Nap-5 size exclusion column (GE Life Sciences) to remove DTT. During this time, 1 mL aliquots of the cube solutions were centrifuged for 8 min at 6000–10,000 rpm depending on the shapes of the NPs, the supernatant was removed, and the NPs were resuspended in water. The NPs were then centrifuged a second time, the supernatant was removed, and then the purified DNA and

1221

1222

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly

water were immediately added to the pellet. Specifically, 5 µmol of thiolated DNA was added per 1 mL of original NP solution. The NP solution was then brought to 0.01 m sodium phosphate buffer (pH = 7.4) and 0.01 wt% sodium dodecyl sulfate (SDS) in water. Stepwise addition of 2 m NaCl was carried out every half hour, such that the NaCl concentration was stepped through 0.05, 0.1, 0.2, 0.3, 0.4 m, and finally arrived at 0.5 m. Following this process, the NPs were placed on a shaker at 1000 rpm and left overnight to ensure a dense loading of oligonucleotides. After functionalization, the NP solutions were centrifuged three times to remove excess DNA. After each of the first two rounds of centrifugation, NPs were resuspended in 0.01 wt% SDS, and after the last centrifugation step, the NPs were resuspended in 0.5 m NaCl, 0.01 m sodium phosphate buffer (pH = 7.4), and 0.01 wt% SDS solution. For the functionalization of the substrate with DNA, the procedure was similar as described above. Specifically, 1 µmol of DTT-cleaved DNA in water was added to each substrate. However, instead of stepwise addition of NaCl, the substrates were brought to 1 m NaCl in one addition, and then shaken for 1 h at 1000 rpm. Substrates were then rinsed three times with water and placed in a 0.5 m NaCl, 0.01 m sodium phosphate buffer (pH = 7.4), and 0.01 wt% SDS solution. After functionalization of thiolated DNA strands, linker strands were hybridized to both the substrates and NPs. To determine the appropriate number of linkers, the concentration of the NP solution was measured using UV-vis [48]. Subsequently, 10,000 strands of linkers were added per NP. Substrates were incubated in a solution containing 0.5 × 10−6 m linker. Both the substrates and the NP solutions were then heated to 55°C for 30 min, and then allowed to slowly cool to room temperature to ensure full hybridization between anchor and linker DNA sequences. Following linker hybridization, the substrates were rinsed in 0.5 m NaCl, 0.01 m sodium phosphate buffer, and 0.01 wt% SDS solution three times, while NPs were used without further processing. For the assembly of NPs, substrates were first placed in the NP solution and shaken at 1000 rpm for varying time (0.5– 96 h) and at variant temperature (25 to 45°C). After assembly, the substrates were rigorously rinsed three times in 0.5 m NaCl, 0.01 m sodium phosphate buffer, and 0.01 wt% SDS solutions to remove

References

unbounded NPs. Then the substrates were immersed in 80% IPA in water (by volume) with 0.2 m ammonium acetate (AA) at 45°C for 30 min to fully remove the PMMA. After PMMA removal, the substrates were rinsed three times with a solution of 80% IPA and 0.2 m AA, then blown dried with N2.

Acknowledgments

This material is based upon work supported by the Air Force Office of Scientific Research Awards FA9550-12-1-0280, FA9550-14-10274, and FA9550-17-1-0348. This work made use of the EPIC facility of the NUANCE Center at Northwestern University, which has received support from the Soft and Hybrid Nanotechnology Experimental (SHyNE) Resource (NSF NNCI-1542205); the MRSEC program (NSF DMR1720139) at the Materials Research Center; the Keck Foundation, and the State of Illinois, through the IIN.

References

1. Jin, R. C., Cao, Y. W., Mirkin, C. A., Kelly, K. L., Schatz, G. C. and Zheng, J. G. (2001). Science, 294, 1901. 2. Liu, M. and Guyot-Sionnest, P. (2005). J. Phys. Chem. B, 109, 22192. 3. Eustis, S. and El-Sayed, M. A. (2006). J. Appl. Phys., 100, 044324.

4. O’Brien, M. N., Jones, M. R., Brown, K. A. and Mirkin, C. A. (2014). J. Am. Chem. Soc., 136, 7603.

5. Nehl, C. L., Liao, H. and Hafner, J. H. (2006). Nano Lett., 6, 683.

6. Kelly, K. L., Coronado, E., Zhao, L. L. and Schatz, G. C. (2003). J. Phys. Chem. B, 107, 668.

7. Narayanan, R. and El-Sayed, M. A. (2004). Nano Lett., 4, 1343.

8. Yan, N., Zhang, J., Yuan, Y., Chen, G.-T., Dyson, P. J., Li, Z.-C. and Kou, Y. (2010). Chem. Commun., 46, 1631.

9. Narayanan, R. and El-Sayed, M. A. (2005). J. Phys. Chem. B, 109, 12663.

10. Cutler, J. I., Auyeung, E. and Mirkin, C. A. (2012). J. Am. Chem. Soc., 134, 1376.

11. Calabrese, C. M., Merkel, T. J., Briley, W. E., Randeria, P. S., Narayan, S. P., Rouge, J. L., Walker, D. A., Scott, A. W. and Mirkin, C. A. (2015). Angew. Chem. Int. Ed., 54, 476.

1223

1224

Design Rules for Template-Confined DNA-Mediated Nanoparticle Assembly

12. Zhang, C., Hao, L. L., Calabrese, C. M., Zhou, Y., Choi, C. H. J., Xing, H. and Mirkin, C. A. (2015). Small, 11, 5360.

13. Radovic-Moreno, A. F., Chernyak, N., Mader, C. C., Nallagatla, S., Kang, R. S., Hao, L. L., Walker, D. A., Halo, T. L., Merkel, T. J., Rische, C. H., Anantatmula, S., Burkhart, M., Mirkin, C. A. and Gryaznov, S. M. (2015). Proc. Natl. Acad. Sci., 112, 3892.

14. Goya, G., Berquo, T., Fonseca, F. and Morales, M. (2003). J. Appl. Phys., 94, 3520.

15. Seo, W. S., Jo, H. H., Lee, K., Kim, B., Oh, S. J. and Park, J. T. (2004). Angew. Chem. Int. Ed., 43, 1115.

16. Park, T.-J., Papaefthymiou, G. C., Viescas, A. J., Moodenbaugh, A. R. and Wong, S. S. (2007). Nano Lett., 7, 766.

17. Lin, Q. Y., Li, Z. Y., Brown, K. A., O’Brien, M. N., Ross, M. B., Butun, Y. Z. S., Butun, S., Chen, P. C., Schatz, G. C., Dravid, V. P., Aydin, K. and Mirkin, C. A. (2015). Nano Lett., 15, 4699. 18. Flauraud, V., Mastrangeli, M., Bernasconi, G. D., Butet, J., Alexander, D. T., Shahrabi, E., Martin, O. J. and Brugger, J. (2017). Nat. Nanotechnol., 12, 73.

19. Lin, Q.-Y., Mason, J. A., Li, Z., Zhou, W., O’Brien, M. N., Brown, K. A., Jones, M. R., Butun, S., Lee, B., Dravid, V. P., Aydin, K. and Mirkin, C. A. (2018). Science, 359, 669.

20. Macfarlane, R. J., Lee, B., Jones, M. R., Harris, N., Schatz, G. C. and Mirkin, C. A. (2011). Science, 334, 204.

21. Macfarlane, R. J., Jones, M. R., Lee, B., Auyeung, E. and Mirkin, C. A. (2013). Science, 341, 1222.

22. Mirkin, C. A., Letsinger, R. L., Mucic, R. C. and Storhoff, J. J. (1996). Nature, 382, 607. 23. Oh, M. and Mirkin, C. A. (2005). Nature, 438, 651.

24. Park, S. Y., Lytton-Jean, A. K. R., Lee, B., Weigand, S., Schatz, G. C. and Mirkin, C. A. (2008). Nature, 451, 553.

25. Jones, M. R., Seeman, N. C. and Mirkin, C. A. (2015). Science, 347, 1260901. 26. Kim, Y., Macfarlane, R. J., Jones, M. R. and Mirkin, C. A. (2016). Science, 351, 579. 27. Auyeung, E., Li, T., Senesi, A. J., Schmucker, A. L., Pals, B. C., de la Cruz, M. O. and Mirkin, C. A. (2014). Nature, 505, 73.

28. Nykypanchuk, D., Maye, M. M., van Der Lelie, D. and Gang, O. (2008). Nature, 451, 549.

References

29. Lin, Q.-Y., Palacios, E., Zhou, W., Li, Z., Mason, J. A., Liu, Z., Lin, H., Chen, P.-C., Dravid, V. P. and Aydin, K. (2018). Nano Lett., 18, 2645. 30. Damasceno, P. F., Engel, M. and Glotzer, S. C. (2012). Science, 337, 453.

31. Navarro, E., Baun, A., Behra, R., Hartmann, N. B., Filser, J., Miao, A.-J., Quigg, A., Santschi, P. H. and Sigg, L. (2008). Ecotoxicology, 17, 372.

32. Stender, A. S. and Fang, N. (2013). MRS Online Proc. Library Archive, 1546. 33. Maurya, N. K. and Mandal, A. (2016). Pet. Sci. Technol., 34, 429.

34. Malaquin, L., Kraus, T., Schmid, H., Delamarche, E. and Wolf, H. (2007). Langmuir, 23, 11513.

35. Mertens, K., Putkaradze, V., Xia, D. and Brueck, S. (2005). J. Appl. Phys., 98, 034309.

36. O’Brien, M. N., Radha, B., Brown, K. A., Jones, M. R. and Mirkin, C. A. (2014). Angew. Chem. Int. Ed., 53, 9532.

37. Fick, A. (1995). J. Membr. Sci., 100, 33.

38. Kuemin, C., Nowack, L., Bozano, L., Spencer, N. D. and Wolf, H. (2012). Adv. Funct. Mater., 22, 702.

39. Ragusa, A., GarcÍa, I. and Penadés, S. (2007). IEEE Trans. NanoBiosci., 6, 319.

40. Cai, D., Mataraza, J. M., Qin, Z.-H., Huang, Z., Huang, J., Chiles, T. C., Carnahan, D., Kempa, K. and Ren, Z. (2005). Nat. Methods, 2, 449. 41. Hu, X. and Dong, S. (2008). J. Mater. Chem., 18, 1279.

42. Wildgoose, G. G., Banks, C. E. and Compton, R. G. (2006). Small, 2, 182.

43. Haes, A. J. and Van Duyne, R. P. (2002). J. Am. Chem. Soc., 124, 10596.

44. Astruc, D., Lu, F. and Aranzaes, J. R. (2005). Angew. Chem. Int. Ed. 44, 7852. 45. Haruta, M. (2003). Chem. Rec., 3, 75.

46. Hutchings, G. J. (2008). Chem. Commun., 1148.

47. Zhang, C., Macfarlane, R. J., Young, K. L., Choi, C. H. J., Hao, L. L., Auyeung, E., Liu, G. L., Zhou, X. Z. and Mirkin, C. A. (2013). Nat. Mater., 12, 741.

48. O’Brien, M. N., Jones, M. R., Brown, K. A. and Mirkin, C. A. (2014). J. Am. Chem. Soc., 136, 7603.

1225

Chapter 66

DNA-Mediated Size-Selective Nanoparticle Assembly for Multiplexed Surface Encoding* Qing-Yuan Lin,a,b Edgar Palacios,b,c Wenjie Zhou,b,d Zhongyang Li,b,c Jarad A. Mason,b,d Zizhuo Liu,b,c Haixin Lin,b,d Peng-Cheng Chen,a,b Vinayak P. Dravid,a,b Koray Aydin,b,c and Chad A. Mirkina,b,d aDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA bInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cDepartment of Electrical Engineering and Computer Science, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA dDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA [email protected]; [email protected]; [email protected]

Qing-Yuan Lin and Edgar Palacios contributed equally to this work.

*Reprinted with permission from Lin, Q.-Y., Palacios, E., Zhou, W., Li, Z., Mason, J. A., Liu, Z., Lin, H., Chen, P.-C., Dravid, V. P., Aydin, K. and Mirkin, C. A. (2018). DNA-mediated size-selective nanoparticle assembly for multiplexed surface encoding, Nano Lett. 18, 2645−2649. Copyright (2018) American Chemical Society. Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1228

DNA-Mediated Size-Selective Nanoparticle Assembly for Multiplexed Surface Encoding

Multiplexed surface encoding is achieved by positioning two different sizes of gold nanocubes on gold surfaces with precisely defined locations for each particle via template-confined, DNA-mediated nanoparticle assembly. As a proof-of-concept demonstration, cubes with 86 and 63 nm edge lengths are assembled into arrangements that physically and spectrally encrypt two sets of patterns in the same location. These patterns can be decrypted by mapping the absorption intensity of the substrate at λ = 773 and 687 nm, respectively. This multiplexed encoding platform dramatically increases the sophistication and density of codes that can be written using colloidal nanoparticles, which may enable high-security, highresolution encoding applications.

66.1

Introduction

Nanostructures have been utilized as encoding materials for a variety of applications, including biomolecular detection [1] and analysis [2], clinical diagnostics [3] and tracking [4], anticounterfeiting [5], food quality and safety assessment [6], and forensic marking [7]. Because of their small size, nanomaterials enable encoding with high density [8], high crypticity [9], high sensitivity [10], and minimal influence on the host target [11]. Indeed, nanomaterials provide a pathway to design high-resolution, high-density multiplexed platforms for high-security encoding [5]. Before this can be realized, however, new synthetic techniques are needed that allow multiple sizes, shapes, and/or compositions of colloidal nanoparticles to be assembled on surfaces with individual particle control [12−15]. Herein, we leverage template-confined, DNA-mediated nanoparticle assembly to position two different sizes of gold nanocubes on gold surfaces with precisely defined locations for each particle. Importantly, because different cube sizes give rise to different optical absorption spectra when positioned a fixed distance above a gold film, this sizeselective assembly approach can be used for multiplexed surface encoding applications. As a proof-of-concept demonstration, cubes with 86 and 63 nm edge lengths are assembled into patterns that physically and spectrally encrypt two sets of patterns in the same location. Through reflectivity spectroscopy and finite-difference time-domain (FDTD) simulations, we find that these patterns can

Introduction

be decrypted by mapping the absorption intensity of the substrate at different wavelengths. This multiplexed encoding platform dramatically increases the sophistication and density of codes that can be written using colloidal nanoparticles, which should enable high-security, high-resolution encoding applications. Over the past few decades, a diverse range of nanoscale architectures, such as nanowires [16], nanodisks [17], nanorods [18], and nanopillars [19], have been designed and synthesized to rationally encode information through optical signals [18, 20] such as luminescence [16, 21] and fluorescence [22]. Colloidal nanoparticles represent a particularly appealing class of materials for encoding applications, because they can be synthesized in many different sizes, shapes, and compositions at large scale [23−25] and can be manipulated through colloidal chemistry to achieve desired properties [26, 27]. For example, when colloidal metal nanocubes are positioned a fixed distance above a metal film, a plasmonic gap mode emerges that allows for tunable absorption and emission [12, 28, 29]. The resonance of the cavity mode is dictated by the size of the particle [30] as well as the size of the gap between the nanoparticle and surface [31], making it an ideal system for surface encoding at tunable wavelengths [14]. Despite these advantages, using colloidal particles for high-resolution multiplexed encoding, an attractive approach to increasing the complexity and information content of codes and, therefore, the difficulty of counterfeiting [11, 16, 19−21] through the incorporation of multiple codes within a single platform, has proven challenging due to a lack of synthetic approaches for positioning of multiple nanoparticles that differ in size on a surface with single-particle resolution. Recent approaches that combine top-down patterning and bottom-up colloidal assembly point toward the possibility of assembling multiple particles, which differ in size, at precise locations on surfaces [14, 32, 33]. For example, DNA-mediated assembly allows one to precisely control the interactions both between nanoparticles and between nanoparticles and a surface in a highly specific manner. Moreover, through the choice of oligonucleotide sequence and length one can systematically tune interaction strength and interparticle distance [34−36]. Indeed, DNA has been used as a surface ligand to program the assembly of vertical stacks of nanoparticles on surfaces [12, 13, 37−40], enabling the creation

1229

1230

DNA-Mediated Size-Selective Nanoparticle Assembly for Multiplexed Surface Encoding

of entirely new classes of optical materials. Importantly, in the case of DNA-driven processes, this control has been limited to the surface normal (generating one-dimensional stacks). In addition, capillarydriven nanoparticle assembly has been used to position individual particles into lithographically defined patterns [15, 41−43], but this approach is limited to the assembly of only one type of nanoparticle on a given substrate. To date, the ability to robustly immobilize different-sized sub-100 nm nanoparticles with independent control over the location of every individual particle has not been realized [43−45]. If this limitation could be overcome, one could literally optically program a substrate with information that is encoded in the patterns of the nanoparticles. Herein, we report the realization of a novel multiplexed surface encoding technique based upon the size-selective, DNA-mediated assembly of nanoparticles. By using a polymer template to provide size selectivity, two types of gold nanocubes, which differ in size, were assembled on a gold substrate at precisely defined locations. Since the resonance wavelengths of the plasmonic gap modes depend on the size of the cubes, we can encode the surface by patterning cubes that have a specific size and decode the information by measuring the intensity of absorption at the corresponding gap mode wavelength, allowing one to encode multiple sets of patterns on the same area. As a proof-of-concept example, we encoded two superimposed patterns consisting of either the letter “N” or the letter “U” and show that the encoded information can be decoded by measuring the absorption intensity map of the surface at the representative resonance wavelengths of the two different gap modes. Broadly, this method provides a high level of structural control for multicomponent nanoparticle assembly on surfaces with single-particle resolution, and it creates a tunable platform for highresolution multiplexed surface encoding.

66.2

Methods

In a typical experiment to construct encoded surfaces, two different sizes of gold nanocubes were assembled onto substrates in a stepwise manner via template-confined, DNA-mediated assembly (Fig. 66.1a). Specifically, electron beam lithography (EBL) was used to pattern

Methods

100 nm thick poly(methyl methacrylate) (PMMA) pores of two different diameters at defined locations on gold-coated Si substrates, such that the bottom of each pore consisted of exposed gold. As has been previously reported [12], the nanoparticles and the exposed gold in the pores were densely functionalized with 3¢propylthiolated DNA and hybridized with complementary linker strands. Then, the nanoparticles were assembled within the pores by designing the DNA on the gold surface to be complementary to the DNA on the nanoparticles. The sizes of the pores were deliberately designed such that larger cubes can only fit inside the larger set of pores, while the smaller cubes can fit into both sizes of pores. Therefore, the larger set of pores was first filled by immersing the substrate in a solution containing only the larger cubes. After assembly, cubes not attached to the surface were removed by rinsing with buffer solution. Subsequently, the smaller set of pores was filled by performing a second assembly with the smaller cubes. After assembly, the PMMA template was dissolved without significantly disturbing the location of the assembled nanoparticles [46]. Finally, the nanoparticle arrays were either transferred back to buffer solution for optical characterization, or dried for characterization by scanning electron microscopy (SEM). Note that although EBL combined with metal deposition and lift-off may be used to create similar patterns, the structures demonstrated here are fundamentally different from those fabricated only through top-down lithography. One advantage of our approach is the ability to position singlecrystalline metallic nanoparticles of different sizes on surfaces, which is extremely difficult, if not impossible to do with EBL. In addition to generating polycrystalline structures with inferior optical properties, EBL is unable to make structures with different heights on this scale. Additionally, our approach provides chemical differentiation of particle building blocks based upon DNA sequence. Importantly, others will be able to extend what we report here to a myriad of colloidal materials, as long as they can be modified with DNA. Finally, our method can be scaled with other higher throughput techniques that allow the direct patterning of DNA such as polymer pen lithography, which is also not possible using EBL.

1231

1232

DNA-Mediated Size-Selective Nanoparticle Assembly for Multiplexed Surface Encoding

Figure 66.1 Assembly of two sizes of gold nanocubes on gold substrates at precisely defined locations. (a) Scheme showing the stepwise DNA-mediated size-selective nanoparticle assembly. Two different sizes of polymer pores were used to assemble large (purple) and small (green) cubes into four types of encoding pixels on the surface. (b) The assembly yield as a function of pore diameter for cubes with 86 and 63 nm edge lengths. The yield curves clearly show the size-selective behavior of the template-confined assembly with a range of pore sizes (orange shade) that result in high yield of small cube assembly and low yield of large cube assembly. (c−f) SEM images of an array of (c) 63 nm edge length cubes assembled in 130 nm diameter pores, (d) 86 nm edge length cubes assembled in 200 nm diameter pores, (e) polymer pores with 130 and 200 nm diameter pores, and (f) cubes with 63 and 86 nm edge lengths assembled in the polymer template shown in panel e. All arrays have a periodicity of 300 nm. The scale bars are 1 μm for the large images and 100 nm for the insets.

To experimentally evaluate this size-selective assembly of nanocubes, single-crystalline gold cubes with an average edge length of 86 ± 3 and 63 ± 2 nm were synthesized [25] and functionalized with propylthiolated DNA [12]. To determine the optimal pore sizes for stepwise assembly, we explored how the assembly yield was affected by pore diameter ranging from 90 to 280 nm for both

Methods

sizes of cubes. As shown in Fig. 66.1b, 63 nm cubes required pore diameters larger than 120 nm for assembly to occur at high yield, while 86 nm cubes required pore diameters of at least 170 nm. The appropriate selection of pore diameters allowed gold cubes with both sizes to assemble into ordered arrays with >99% yield over an area of 100 by 100 μm (Fig. 66.1c,d). The relationship between yield, pore size, and nanoparticle size allowed us to determine that pores of 125−135 nm in diameter resulted in both a high yield of assembled 63 nm cubes (>99%) and a low yield of assembled 86 nm cubes ( 0. At about 365 nm (Re(εeff) ≈ 0, Fig. 72.1e), ENZ behavior is seen in the superlattice [13, 21]. Here, attenuation of the incident field is seen at the boundary of the superlattice. Additionally, some fraction of the incident field is immediately reflected due to the significant boundary mismatch of the refractive index between the superlattice (near zero) and the host media (1.4). At 415 nm, where Re(εeff) < 0, we see a slightly altered effect from the ENZ fields. A material with Re(εeff) < 0 acts “optically metallic,” showing a finite skin depth, after which the field is attenuated (Fig. 72.1f). Thus, a superlattice consisting of ~20% Ag, and thus ~80% water or dielectric, would appear optically glossy like a metal, yet it would be nonconductive, like a dielectric. For comparison, the electric fields at wavelengths associated with the absorption and extinction maxima (Fig. 72.1d, lower panel) for this superlattice are shown in Figs. 72.1g and 72.1h. Notably, the strength of the electric field on the interior of the superlattice is more intense for the absorption and extinction maxima compared to the fields in Figs. 72.1e and 72.1f. Forward scattering contributes significantly to the extinction maximum (550 nm), and thus the incident field is focused in the direction of propagation (Fig. 72.1h). Having demonstrated that Ag superlattices with volume fractions between 15% and 25% should demonstrate either ENZ or optically metallic behavior that is tunable with the lattice constant, we utilized DNA-programmable methods to experimentally synthesize such structures. AgNPs are more challenging to incorporate experimentally than gold nanoparticles (AuNPs) due to the chemical degradation of the AgNPs under DNA functionalization conditions

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

and their susceptibility to surface oxidation [27]. However, our group recently developed a method for functionalizing AgNPs with a dense layer of oligonucleotides [43], which are analogous to the AuNP-based spherical nucleic acids (SNAs) [44]. In this work, AgNPs with diameters of ~20 and ~30 nm were functionalized with DNA oligonucleotides containing three terminal cyclic disulfide (DSP) groups to serve as a robust anchor. DNA linkers with single-stranded “sticky ends” that facilitate hybridization between NPs were added in a predetermined ratio, and the samples were annealed slightly below their melting temperature to form crystalline aggregates (Fig. 72.2a). Importantly, these AgNP aggregates displayed sharp melting transitions indicative of co-operative binding, a requirement for the synthesis of well-ordered nanoparticle superlattices [33, 45].

Figure 72.2 Characterization of FCC and BCC superlattices assembled from AgNPs. (a) Schematic illustration of DNA-programmable nanoparticle crystallization where the length of the linker strands can be tuned by increasing the value of n. DSP refers to cyclic dithiol. (b) SAXS pattern and corresponding unit cell and (c, d) STEM images of an FCC lattice synthesized from 30 nm AgNPs. The zoomed-out image (d) shows multiple crystalline domains within a superlattice. (e) SAXS pattern and corresponding unit cell and (f) STEM image of a BCC lattice synthesized from 20 nm AgNPs. Experimental data are shown in blue and predicted scattering patterns are shown in black. All scale bars are 100 nm. (g) SAXS patterns of 20 nm AgNP superlattices with BCC symmetry with tunable interparticle spacings (red: 31.7 nm (n = 0), green: 52.8 nm (n = 1), blue: 71.8 nm (n = 2)).

1327

1328

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

The AgNPs functionalized with a dense layer of DNA oligonucleotides were then used to synthesize one-component face-centered cubic (FCC) and two-component body-centered cubic (BCC) crystal structures using linkers that were self-complementary (5¢-GCGC-3¢) and non-self-complementary (5¢TTCCTT-3¢ and 5¢-AAGGAA-3¢), respectively (Fig. 72.2) [15, 18]. Superlattices with more complex unit cells were also synthesized. As we have previously demonstrated, the most stable crystal structure is that which maximizes the number of DNA hybridization events between neighboring nanoparticles [18]. The 3D crystalline structures of the superlattices were characterized by in-situ synchrotron smallangle X-ray scattering (SAXS), and the data were compared to the scattering patterns of the analogous atomic lattices for position and intensity of the allowed reflections. The large number of sharp diffraction peaks that are observed for the FCC and BCC AgNP superlattices (Fig. 72.2b,e) are indicative of highly ordered crystals with domain sizes on the micrometer-scale. In a complementary method, the AgNP superlattices were embedded in a polymer resin, sectioned, and characterized by scanning transmission electron microscopy (STEM) to obtain a “snapshot” of the crystals that exist in solution (Fig. 72.2c,d,f) [46]. It is important to note that the embedding process results in a reduction of the interparticle spacing, but maintains the crystallographic symmetry calculated from the SAXS pattern. In order to measure the optical properties of individual AgNP superlattices without modifying their symmetry or lattice spacing, the AgNP superlattices were stabilized in the solid state using a silica encapsulation method [47]. The nearest-neighbor spacing, and therefore lattice parameters, of the FCC and BCC crystals can be tuned with nanometer precision by changing the number of DNA bases in the linkers [48]. For the AgNP superlattices, each nucleobase adds 0.244 nm. In this work, the length of the linkers was tuned using modular blocks of 40 bases, denoted by parentheses in Fig. 72.2a, allowing us to probe the effects of interparticle plasmonic coupling in an extended periodic 3D structure. Each integer increment of n corresponds to an increase of ~9.7 nm on each NP, or ~19.5 nm total between neighboring NPs. Using 20 nm AgNPs, BCC superlattices were assembled and characterized with interparticle distances of 31.7 ± 0.6 nm (n = 0), 52.8 ± 0.8 nm (n = 1), and 71.8 ± 1.2 nm (n = 2) (standard

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

deviation determined from four separate batches of superlattices), corresponding to surface-to-surface distances between neighboring AgNPs of approximately 12, 32, and 52 nm, respectively. The SAXS pattern for each of the three systems confirms high-quality crystals with BCC symmetry (Fig. 72.2g).

Figure 72.3 Optical characterization of AgNP superlattices (experimental data and simulations). (a) UV-vis spectra taken in water of as-synthesized AgNP BCC superlattices with varying interparticle spacings, and thus, metal fill fractions (solid traces; red: n = 0, 17.1% Ag; green: n = 1, 3.7% Ag; blue: n = 2, 1.5% Ag). Simulated spectra of randomly generated orientationally averaged superlattices in a refractive index of 1.4 are underlaid with corresponding colors, albeit a lighter shade. (b) Single superlattice measurements of absorption spectra taken with a micro-spectrophotometer. Spectra were chosen that are representative of the statistical data. Note the difference in the scales of the x axes.

The collective optical properties of the superlattices were then analyzed experimentally, using both UV-vis spectroscopy for ensemble measurements and optical microspectrophotometry for individual superlattice measurements. Using UV-vis spectroscopy (Fig. 72.3a), red-shifting of the superlattices’ bulk LSPR with decreased nanoparticle spacing is observed. This can be explained

1329

1330

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

primarily by the increase in Ag fill fraction (i.e., an increase in dipole– dipole interactions, provided that they are weak) with decreased interparticle spacing [49]. Additionally, electrodynamic simulations were performed using the CDM. For correlation with the UV-vis ensemble measurements, superlattices with varying shape, but with well-defined interparticle spacing, were randomly generated to best mimic the superlattices observed with STEM. CDM simulations on these structures are in good agreement with the experimental UV-vis extinction spectra (Fig. 72.3a). It is difficult, however, to determine rigorous structure-property relationships with ensemble measurements. The inherent averaging and solution-based nature of UV-vis spectroscopy makes it difficult to separate the effects of nanoparticle spacing and superlattice size [50]. Thus, individual superlattices were analyzed using a micro-spectrophotometer [51, 52]. After depositing silica-embedded Ag nanoparticle superlattices on a Formvar grid, STEM was used to select superlattices with sizes between 700 and 800 nm. Because the Formvar and silica encapsulation affect scattering, and therefore extinction measurements, absorption spectra were determined from transmission and reflection measurements. Thus, the plasmonic response of the AgNP superlattices with varying interparticle spacings was determined with bright-field measurements using the micro-spectrophotometer. Figure 72.3b shows that “n = 0” superlattices demonstrate the most red-shifted absorption, with an average maximum absorption, λ max, of 594 nm. With increased spacing, the absorption is blue-shifted (Fig. 72.3b), with average λmax values of 530 nm (n = 1) and 448 nm (n = 2). We note that these measurements are consistent both between superlattices from a single sample and between different batches of superlattices, with standard deviations in λmax of less than 4% for all spacings. The redshift seen between the micro-spectrophotometer measurements (Fig. 72.3b) and the UV-vis ensemble measurements (Fig. 72.3a) is attributed to the silica-embedding process, which changes the local refractive index and the interaction between the superlattice and the Cu Formvar TEM grid. Additionally, nonuniformities in the superlattice aggregates can lead to the formation of additional lines in the individual superlattice measurements. For example, in the green trace of Fig. 72.3b, an extra peak can be observed at ~600 nm.

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

This is likely due to asymmetry in the superlattice shape; the complex structure of the aggregates can result in multiple plasmon resonances from different parts of the aggregate (see, for example, Fig. 72.1). In addition, it is possible that the distance between the superlattice and the Formvar is not uniform across the entire structure, resulting in changes to the structure of the LSPR spectrum. The low variation in absorption maxima from different superlattices suggests that the absorption of similarly sized superlattices is consistent. From this, we conclude that the primary reason that the LSPR redshifts with decreased interparticle spacing is due to the increase in Ag fill fraction [35]. Once the emergent optical properties of DNA-mediated AgNP superlattices were investigated, we introduced DNA-functionalized AuNP building blocks to allow for the assembly of binary (AgNP)– (AuNP) superlattices. Although there have been some examples of DNA-programmable heterogeneous nanoparticle assemblies in 3D space to date [53–56] the functional properties of these materials have yet to be explored. To synthesize the binary Ag–Au superlattices, a two-component system with non-self-complementary linkers, similar to those used to create the AgNP BCC crystals, was utilized. The two components consisted of AgNPs (20 or 30 nm diameter) functionalized with DSP-terminated DNA oligonucleotides and AuNPs (15, 20, or 30 nm diameter) functionalized with monothiolterminated DNA oligonucleotides (Fig. 72.4a). The simplest binary system, in which the Ag and AuNPs are identical in size, and added in a 1:1 ratio, results in an AB-type crystal (isostructural with CsCl), characterized by SAXS and STEM (Fig. 72.4b,c). As shown in Fig. 72.4b, the experimental SAXS pattern is in strong agreement with the calculated diffraction pattern, confirming the long-range order of the system. Crystal structures with increasingly complex unit cells can be synthesized with a two-component system by tuning the hydrodynamic radii and linker ratios of each component according to the design rules that have been established for DNAmediated AuNP crystallization [18]. Binary (AgNP)–(AuNP) superlattices of the measurements AB2-type (isostructural with AlB2) with a hexagonal unit cell (cell parameter c/a ratio = 0.73) were synthesized using 30 nm AuNPs with the “n = 2” linker at the corners of the unit cell and 20 nm AgNPs with the “n = 1” linker in the middle (Fig. 72.4e,f). Importantly, an AB2-type crystal can also

1331

1332

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

be achieved by swapping the components, i.e., using 30 nm AgNPs with the “n = 2” linker and 20 nm AuNP building blocks with the “n = 1” linker; thus demonstrating the versatility of the DNA-mediated approach to assembly. By varying the radii and linker ratios of the Ag and Au components, binary AB3-type (isostructural with Cr3Si) and AB6-type (isostructural with Cs6C60) superlattices were also synthesized (Fig. 72.4h–k). All binary superlattices were confirmed with SAXS and imaged by electron microscopy (Fig. 72.4).

Figure 72.4 Characterization of binary superlattices assembled from Ag and AuNPs. (a) Schematic illustration of the assembly of binary Ag–AuNP superlattices where the linker lengths can be tuned by increasing n. (b, c, e, f, h–k) SAXS patterns (and the corresponding unit cell) and STEM images of binary Ag–AuNP superlattices that are isostructural with CsCl (b, c), AlB2 (e, f), Cr3Si (h, i), and Cs6C60 (j, k). In the SAXS patterns, experimental data are shown in red and predicted scattering patterns are shown in black. The insets of the STEM images show the unit cell viewed along the appropriate projection axis. (d, g) EDX elemental maps of AB- (d) and AB2-type (g) binary lattices with the zone axis shown in the insets. Gold signal is shown in red and silver signal is shown in blue. All scale bars are 100 nm.

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

Although the AgNPs have lower contrast than AuNPs in the STEM images of the binary Ag–AuNP crystals, the identity of the NPs can be determined using energy-dispersive X-ray (EDX) elemental mapping. When EDX maps of Au and Ag are overlaid on the corresponding STEM image (Fig. 72.4d,g), the zone axis becomes immediately obvious. Thus, in addition to potential emergent collective properties, the incorporation of Ag into DNA-mediated superlattices provides an elemental handle to unequivocally confirm the symmetry of the unit cell that is determined using SAXS. Interestingly, when solution-phase ensemble UV-vis measurements were collected for the 3D binary CsCl-type (AgNP)– (AuNP) superlattices, the peak at 410 nm (corresponding to the LSPR of the AgNPs) was significantly dampened when compared to a solution of dispersed Ag and AuNPs in an equal ratio (Fig. 72.5a). Furthermore, the dampening is more pronounced as the interparticle distance was decreased through the use of shorter DNA linkers. This observation was further investigated by simulating the system using generalized multiparticle Mie (GMM) theory (Fig. 72.5b) [57–59]. At the closest nearest-neighbor surface-to-surface distance of ~10 nm (n = 0), the LSPR peaks of Ag and Au are equivalent in intensity and significantly broadened. This is surprising because AgNPs have a much higher extinction coefficient than AuNPs of the same diameter [27]. To understand the interaction between the Ag and AuNPs in the extended lattice structure, the per-particle optical absorption and scattering cross sections were plotted for each material separately (Fig. 72.5c,d). As is shown in Fig. 72.5c, there is a Fano-like [26] dip in the absorption efficiency of Au at 370 nm when the surfaceto-surface distance is 10 nm. Additionally, a greater decrease is seen in the Ag absorption per-particle relative to the Ag scattering per particle (Fig. 72.5d). Recently, plasmonic coupling has been investigated between Au and AgNP dimers [60, 61] and in 2D arrays of plasmonic nanoparticles made lithographically [62] or chemically and separated by nonplasmonic NPs [32]. Theoretical studies by Bachelier et al. have predicted the presence of a Fano profile in dimers of AgNPs and AuNPs due to coupling of the LSPR of the AgNP with the Au interband transitions [60]. The experimental observation of this Fano profile is challenging due to the orientational averaging that occurs in solution-phase measurements [63] and the need to isolate the absorption cross sections of the Ag and Au separately

1333

1334

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

[61]. In the binary Ag–Au colloidal crystals, the LSPR of the AgNPs serves as the discrete level and the gold interband transitions serve as the continuum. Thus the dampening of the Ag LSPR in the binary (AgNP)–(AuNP) superlattices, observed experimentally and in GMM simulations, is due to a Fano-like interference between the resonant silver mode and the interband transitions of gold.

Figure 72.5 Optical characterization and simulations of AB-type Ag–Au binary superlattices. (a) Experimental data: UV-vis spectra of unassembled, DNA-functionalized Ag and Au nanoparticles (blue), Ag–Au superlattices with n = 1 linker (green), and Ag–Au superlattices with n = 0 linker (orange). (b) Simulated data: extinction of Ag–Au superlattices at various surface-to-surface distances. (c) Simulated absorption efficiency per nanoparticle of Ag–Au binary superlattices at various surface-to-surface distances. (d) Simulated scattering efficiency per nanoparticle of Ag–Au binary superlattices at various surface-tosurface distances. All simulations performed with generalized multiparticle Mie theory.

In conclusion, electrodynamic simulations were used to identify that superlattices of AgNPs have the potential to exhibit unusual metamaterial behavior, including ENZ behavior and an “optically metallic” region that exhibits high reflectivity despite the fact that the superlattices are over 80% water. DNA was used as a programmable linker to assemble spherical silver

References

nanoparticles into 3D periodic assemblies, and the optical response of the AgNP superlattices was shown to be heavily dependent on the interparticle spacing (and hence the metal fill fraction) within the superlattices. DNA-programmable assembly allows for the precise control of NP spacing, making it attractive for the study of structure–property relationships and the construction of designer materials. The experiments detailed herein represent one of the first studies of the functional properties of colloidal superlattices engineered with DNA. The use of DNA as a programmable assembly tool in conjunction with electrodynamic simulations can be used for the effective design of plasmonic metamaterials with potential for use in the creation of optical circuitry and interconnects [10, 23, 24], optical cloaking materials [9, 21], and data exchange [13, 20].

Acknowledgments

C.A.M. and G.C.S. acknowledge support from AFOSR MURI Award FA9550–11–1–0275 and the Department of Energy Office (DOE Award DE-SC0000989) through the Northwestern University Nonequilibrium Energy Research Center. C.A.M. also acknowledges support from AFOSR Awards FA9550–09–1–0294 and FA9550– 12–1–0280 and NSF/MRSEC award DMR-1121262. G.C.S. also acknowledges support from NSF/MRSEC award DMR-0520513. K.L.Y. and M.B.R. gratefully acknowledge support through NDSEG graduate fellowships. K.L.Y. and M.R.J. gratefully acknowledge support from the NSF through the Graduate Research Fellowship Program (GRFP). Use of the Advanced Photon Source was supported by the Office of Basic Energy Sciences, US DOE under Contract DEAC02–06CH11357. Electron microscopy was carried out in the Electron Probe Instrumentation Center facility of the Northwestern University Atomic and Nanoscale Characterization Experimental Center.

References

1. Mirkin, C. A., Letsinger, R. L., Mucic, R. C. and Storhoff, J. J. (1996). Nature, 382, 607.

2. Macfarlane, R. J., O’Brien, M. N., Petrosko, S. H. and Mirkin, C. A. (2013). Angew. Chem. Int. Ed., 52, 5688.

1335

1336

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

3. Nie, Z., Petukhova, A. and Kumacheva, E. (2010). Nat. Nanotechnol., 5, 15. 4. Travesset, A. (2011). Science, 334, 183.

5. Gwo, S., Lin, M.-H., He, C.-L., Chen, H.-Y. and Teranishi, T. (2012). Langmuir, 28, 8902. 6. Stebe, K. J., Lewandowski, E. and Ghosh, M. (2009). Science, 325, 159.

7. Alaeian, H. and Dionne, J. A. (2012). Opt. Express, 20, 15781.

8. Tao, A. R., Ceperley, D. P., Sinsermsuksakul, P., Neureuther, A. R. and Yang, P. (2008). Nano Lett., 8, 4033. 9. Smith, D. R., Pendry, J. B. and Wiltshire, M. C. K. (2004). Science, 305, 788.

10. Alu, A. and Engheta, N. (2007). Opt. Express, 15, 7578.

11. Chen, P. Y., Soric, J. and Alu, A. (2012). Adv. Mater., 24, Op281.

12. Silveirinha, M. and Engheta, N. (2006). Phys. Rev. Lett., 97, 157403.

13. Engheta, N. (2013). Science, 340, 286.

14. Solomon, M. J. (2011). Curr. Opin. Colloid Interface Sci., 16, 158.

15. Park, S. Y., Lytton-Jean, A. K. R., Lee, B., Weigand, S., Schatz, G. C. and Mirkin, C. A. (2008). Nature, 451, 553.

16. Nykypanchuk, D., Maye, M. M., van der Lelie, D. and Gang, O. (2008). Nature, 451, 549.

17. Tan, S. J., Campolongo, M. J., Luo, D. and Cheng, W. (2011). Nat. Nanotechnol., 6, 268. 18. Macfarlane, R. J., Lee, B., Jones, M. R., Harris, N., Schatz, G. C. and Mirkin, C. A. (2011). Science, 334, 204.

19. Vesseur, E. J. R., Coenen, T., Caglayan, H., Engheta, N. and Polman, A. (2013). Phys. Rev. Lett., 110, 013902.

20. Alu, A., Silveirinha, M. G., Salandrino, A. and Engheta, N. (2007). Phys. Rev. B, 75.

21. Alu, A. and Engheta, N. (2007). Opt. Express, 15, 3318.

22. Alu, A. and Engheta, N. (2006). J. Appl. Phys., 99. 23. Atwater, H. A. (2007). Sci. Am., 296, 56.

24. Engheta, N. (2007). Science, 317, 1698.

25. Kelly, K. L., Coronado, E., Zhao, L. L. and Schatz, G. C. (2003). J. Phys. Chem. B, 107, 668.

26. Halas, N. J., Lal, S., Chang, W. S., Link, S. and Nordlander, P. (2011). Chem. Rev., 111, 3913.

References

27. Rycenga, M., Cobley, C. M., Zeng, J., Li, W., Moran, C. H., Zhang, Q., Qin, D. and Xia, Y. (2011). Chem. Rev., 111, 3669.

28. Blaber, M. G., Arnold, M. D. and Ford, M. J. (2010). J. Phys.: Condens. Matter, 22, 143201. 29. Kuzyk, A., Schreiber, R., Fan, Z., Pardatscher, G., Roller, E.-M., Hogele, A., Simmel, F. C., Govorov, A. O. and Liedl, T. (2012). Nature, 483, 311.

30. Liu, Q., Cui, Y., Gardner, D., Li, X., He, S. and Smalyukh, I. I. (2010). Nano Lett., 10, 1347.

31. Liu, Q., Senyuk, B., Tang, J., Lee, T., Qian, J., He, S. and Smalyukh, I. I. (2012). Phys. Rev. Lett., 109, 088301.

32. Ye, X., Chen, J., Diroll, B. T. Murray, C. B. (2013). Nano Lett., 13, 1291.

33. Macfarlane, R. J., Jones, M. R., Senesi, A. J., Young, K. L., Lee, B., Wu, J. and Mirkin, C. A. (2010). Angew. Chem., 122, 4693.

34. Macfarlane, R. J., Jones, M. R., Lee, B., Auyeung, E. and Mirkin, C. A. (2013). Science, 341, 1222.

35. Lazarides, A. A., Lance Kelly, K., Jensen, T. R. and Schatz, G. C. (2000). J. Mol. Struct. Theochem., 529, 59.

36. Collinge, M. J. and Draine, B. T. (2004). J. Opt. Soc. Am. A, 21, 2023. 37. Draine, B. T. and Flatau, P. J. (1994). J. Opt. Soc. Am. A, 11, 1491.

38. Draine, B. T. Flatau, P. J. (2012). arXiv:1202.3424 [physics.comp-ph]. 39. Draine, B. T. and Goodman, J. (1993). Astrophys. J., 405, 685. 40. Flatau, P. J. and Draine, B. T. (2012). Opt. Express, 20, 1247.

41. Johnson, P. B. and Christy, R. W. (1972). Phys. Rev. B, 6, 4370.

42. Kreibig, U. and von Fragstein, C. (1969). Z. Phys., 224, 307.

43. Lee, J.-S., Lytton-Jean, A. K. R., Hurst, S. J. and Mirkin, C. A. (2007). Nano Lett., 7, 2112. 44. Cutler, J. I., Auyeung, E. and Mirkin, C. A. (2012). J. Am. Chem. Soc., 134, 1376.

45. Hill, H. D., Millstone, J. E., Banholzer, M. J. and Mirkin, C. A. (2009). ACS Nano, 3, 418. 46. Jones, M. R., Macfarlane, R. J., Lee, B., Zhang, J., Young, K. L., Senesi, A. J. and Mirkin, C. A. (2010). Nat. Mater., 9, 913. 47. Auyeung, E., Macfarlane, R. J., Choi, C. H. J., Cutler, J. I. and Mirkin, C. A. (2012). Adv. Mater., 24, 5181.

48. Hill, H. D., Macfarlane, R. J., Senesi, A. J., Lee, B., Park, S. Y. and Mirkin, C. A. (2008). Nano Lett., 8, 2341.

1337

1338

Using DNA to Design Plasmonic Metamaterials with Tunable Optical Properties

49. Atay, T., Song, J.-H. and Nurmikko, A. V. (2004). Nano Lett., 4, 1627.

50. Storhoff, J. J., Lazarides, A. A., Mucic, R. C., Mirkin, C. A., Letsinger, R. L. and Schatz, G. C. (2000). J. Am. Chem. Soc., 122, 4640.

51. Mock, J. J., Barbic, M., Smith, D. R., Schultz, D. A. and Schultz, S. (2002). J. Chem. Phys., 116, 6755.

52. Sherry, L. J., Chang, S. H., Schatz, G. C., Van Duyne, R. P., Wiley, B. J. and Xia, Y. N. (2005). Nano Lett., 5, 2034.

53. Zhang, C., Macfarlane, R. J., Young, K. L., Choi, C. H., Hao, L., Auyeung, E., Liu, G., Zhou, X. and Mirkin, C. A. (2013). Nat. Mater., 12, 741.

54. Auyeung, E., Cutler, J. I., Macfarlane, R. J., Jones, M. R., Wu, J., Liu, G., Zhang, K., Osberg, K. D. and Mirkin, C. A. (2012). Nat. Nanotechnol., 7, 24.

55. Kostiainen, M. A., Hiekkataipale, P., Laiho, A., Lemieux, V., Seitsonen, J., Ruokolainen, J. and Ceci, P. (2013). Nat. Nanotechnol., 8, 52. 56. Sun, D. and Gang, O. (2011). J. Am. Chem. Soc., 133, 5252.

57. Xu, Y.-l. (1995). Appl. Opt., 34, 4573. 58. Xu, Y.-l. (1997). Appl. Opt., 36, 9496.

59. Xu, Y.-l. and Wang, R. T. (1998). Phys. Rev. E, 58, 3931.

60. Bachelier, G., Russier-Antoine, I., Benichou, E., Jonin, C., Del Fatti, N., Vallée, F. and Brevet, P. F. (2008). Phys. Rev. Lett., 101, 197401.

61. Sheikholeslami, S., Jun, Y.-W., Jain, P. K. and Alivisatos, A. P. (2010). Nano Lett., 10, 2655. 62. Haynes, C. L., McFarland, A. D., Zhao, L., Van Duyne, R. P., Schatz, G. C., Gunnarsson, L., Prikulis, J., Kasemo, B. and Käll, M. (2003). J. Phys. Chem. B, 107, 7337.

63. Yoon, J. H., Zhou, Y., Blaber, M. G., Schatz, G. C. and Yoon, S. (2013). J. Phys. Chem. Lett., 4, 1371.

Chapter 73

Plasmonic Metallurgy Enabled by DNA*

Michael B. Ross,a,b Jessie C. Ku,b,c Byeongdu Lee,d Chad A. Mirkin,a,b,c and George C. Schatza,b aDepartment

of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA bInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA dX-ray Science Division, Advanced Photon Source, Argonne National Laboratory, 9700 S. Cass Ave., Argonne, IL 60439, USA [email protected]; [email protected]

Noble-metal nanoparticles are ideal as nanoscale building blocks for new materials because their optical responses are dominated by localized surface plasmon resonance (LSPR) excitation, providing efficient and tunable absorption that is controlled by their size, shape, and composition [1–7]. The precise placement of such nanoparticle *Reprinted with permission from Ross, M. B., Ku, J. C., Lee, B., Mirkin, C. A. and Schatz, G. C. (2016). Plasmonic metallurgy enabled by DNA, Adv. Mater. 28, 2790–2794. Copyright © 2016, John Wiley and Sons.

Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1340

Plasmonic Metallurgy Enabled by DNA

building blocks enables one to dictate the structural properties of a material including 3D crystal habit [5, 6], metal-semiconductor “nanoparticle doping” [8], nanoparticle packing density [9, 10], and programmability over multiple length scales [2, 11]. Indeed, such architectural control provides greater opportunities for designing optical properties than is available for most other materials. Here, we use DNA-programmable assembly to study metallurgical control over arrangement, metal composition, and hierarchical order in assemblies of plasmonic nanoparticles, leading to “alloy” and “bimetallic” assemblies that have diverse optical properties in which homogenized dielectric properties are not always applicable even though metal volume fractions are just a few percent. In “alloy” nanoparticle assemblies (Fig. 73.1, left), where the distribution of metallic composition is random, absorption and color can be linearly and precisely controlled by changing the Ag:Au ratio. This applies to both assemblies comprised of atomic alloy (AgAu) nanoparticles and also to nanoscopic alloy assemblies comprised of homogeneous nanoparticles (Ag or Au) that are randomly distributed within the superstructure. A third possibility involves “bimetallic” assemblies, here defined as structures where the metallic composition is homogeneous within an individual (subwavelength) layer (Fig. 73.1, right), but where interlayer interactions lead to asymmetric reflectivity in the collective optical response. In these systems, the optical properties are asymmetric in that they differ depending on the material orientation to incident light. NA uniquely enables the precise placement and manipulation of nanoparticles across multiple dimensions [7, 12–19]. Nanoparticles that are densely functionalized with DNA can be used as programmable atom equivalents (PAEs), which enables the separation of the nanoparticle identity from arrangement [18, 19]. Recent advances in DNA-programmed assembly have enabled the synthesis of layer-by-layer structures, where the DNA on particle A in a layer is complementary with the DNA on particle B in the subsequent layer, enabling precise control over layer composition and film thickness (Fig. 73.2a) [20].

Plasmonic Metallurgy Enabled by DNA

Figure 73.1 Controlling optical properties with plasmonic metallurgy. (Left) Alloying (random placement) of Ag or Au building blocks (on the atomic or nanoscopic scale) can be used to linearly change absorption (and thus color). (Right) Bimetallic structures, where different metal building blocks are ordered (here by homogeneous layering), can be used to precisely control reflectivity.

Plasmonic building blocks that are 20–30 nm diameter spheres and comprised of Ag, Au, or AgAu alloys were synthesized and densely functionalized with DNA by adapting existing protocols [7, 21]. These building blocks exhibit LSPRs that vary in location and extinction coefficient; increasing the Au content linearly red shifts and dampens the LSPR (Fig. 73.2b,c) [21]. The optical properties of the atomic alloy particles cannot be described by a combination of the Ag and Au dielectric functions, since each alloy has a unique electronic structure with distinct metallic dielectric properties that reflect electron transfer between the two components [21, 22]. The dense DNA shell assembles the building blocks into ~175 nm thick isostructural bcc films (eight nearest neighbors) independent of the composition of the underlying metallic nanoparticles. The assemblies studied in this work thus exhibit local order (bcc), which previous work has been shown is sufficient for a homogeneous optical response at these nanoparticle spacings [5, 6, 23, 24]. Two types of alloy assemblies are accessed with this strategy: one is homogeneous on the nanoscale (each building block is identical in composition), but “alloyed” on the atomic scale, i.e., each nanoparticle is a random mixture of Ag and Au atoms (Fig. 73.3a), and the other is inhomogeneous on the nanoscale (a spatially random mixture of two different particles), but where each nanoparticle is a “pure”

1341

1342

Plasmonic Metallurgy Enabled by DNA

building block, i.e., Ag or Au (Fig. 73.3f). The former assemblies are synthesized by densely functionalizing atomic alloy nanoparticles with DNA and the latter by mixing different ratios of Ag and Au nanoparticles with identical DNA (either A or B). Importantly, the optical properties of both classes of alloy architectures can be precisely tuned by varying the Ag:Au ratio. As such, fine control over absorption provides a means for precisely controlling color, an important use of plasmonic materials due to their high extinction coefficients and nondiffraction limited scattering [25–27].

Figure 73.2 Synthesis of DNA-nanoparticle assemblies. (a) Scheme depicting two-component (A–B type) DNA-programmable assembly on an Au surface. (b) Optical image of aqueous suspensions of metal nanoparticle building blocks (left to right, Au, Au80Ag20, Au60Ag40, Au40Ag60, Au20Ag80, Ag). (c) Experimental (solid) and simulated (dashed) extinction spectra of building blocks shown in (b).

Plasmonic Metallurgy Enabled by DNA

Figure 73.3 Systematic control over absorption and color by atomic and nanoscopic alloying. (a) Scheme depicting atomically alloyed nanoparticles (left) and their arrangement into a thin-film assembly. (b, c) Optical images (b) and simulated color (c) of atomically alloyed nanoparticle assemblies (left 100% Ag, right 100% Au). (d, e) Measured (d) and simulated (e) extinction of plasmonic nanoparticle assemblies with varying atomic alloy compositions. (f) Scheme depicting Ag and Au nanoparticle building blocks (left) and their arrangement into alloy thin-film assemblies. (g, h) Optical images (g) and simulated color (h) of nanoscopically alloyed nanoparticle assemblies (left 100% Ag, right 100% Au). (i, j) Measured (i) and simulated (j) extinction of plasmonic nanoparticle assemblies with varying nanoscopic alloy compositions.

1343

1344

Plasmonic Metallurgy Enabled by DNA

Similar to the free building blocks (Fig. 73.2c), assemblies comprised of AgAu alloy nanoparticles (Fig. 73.3a) exhibit one collective LSPR that red shifts and broadens with increasing Au content (Fig. 73.3d), exhibiting very good agreement with coupled dipole simulations of bcc structures with equivalent nanoparticle spacings (Fig. 73.3e). Notably, an average of the dielectric functions of Ag and Au does not reproduce the metallic properties of the atomic alloy. In nanoscopic alloyed assemblies (Fig. 73.3f), two distinct LSPRs are observed that vary linearly with the ratio of Ag (~400 nm) and Au (~520 nm) nanoparticles [note the extinction coefficient of Ag nanoparticles is larger than that of Au [21]] (Fig. 73.3i), demonstrating excellent agreement with coupled dipole simulations (Fig. 73.3j). Optical images of both the atomic alloy assemblies (Fig. 73.3b) and nanoscopic alloy assemblies (Fig. 73.3g) demonstrate systematic variation in color from white/ yellow (Ag assembly, left) through yellow and orange to red (Au assembly, right). Using the Commission Internationale de l’Elcairage (CIE) 1931 color matching functions, the simulated spectra (Fig. 73.3e,j) are converted to red-green-blue (RGB) color values (Fig. 73.3c,h); these color swatches approximate how the optical properties of the fi lm would appear to the human eye. In general, the trend in color change compares favorably with the experimental optical images, for example Fig. 73.3b,g both exhibit a clear transition from yellow to burgundy with increasing gold content. Additionally, the atomic alloys (Fig. 73.3b,c) are yellow for 20% and 40% Au in both measurements and simulations, while nanoscopic alloys of the same composition are comparably more orange (Fig. 73.3g,h). Variation between the simulated color and optical images could be due to the optical collection geometry, imperfect calibration to “white” light, and more complex scattering phenomena not captured in the simulations. These data show that deliberate mixing of Ag and Au on two distinct length scales (atomic and nanoscopic) provide smoothly tunable absorption and color that are well described by coupled dipole simulations. In a variety of nanoscale systems, order leads to robust and tunable optical properties [3–6, 8, 9]. Here we consider a third (bimetallic) structural motif for Ag/Au mixtures wherein the order of homogeneous Ag and Au layers is varied. Figure 73.4a shows that

Plasmonic Metallurgy Enabled by DNA

the reflective properties of DNA-programmed plasmonic films vary drastically in two model systems. The most significant differences are observed in assemblies that are grouped by material, e.g., three layers of Ag nanoparticles on top of three layers of Au nanoparticles (and its inverse). Notably, the extinction properties of the plasmonic films do not change appreciably depending on the film ordering; the characteristic LSPRs for both Ag (~400 nm) and Au (~520 nm) are observed (Fig. 73.4b, solid). However, when light is first incident on the three Ag layers, the reflectivity of the film is significantly increased near 450 nm (Fig. 73.4c, solid). For comparison, when light is first incident on the three Au layers, the reflectance is greatly attenuated near 400–450 nm, and increases toward the red (~600 nm) (Fig. 73.4c, solid). Fresnel thin-film simulations reproduce both the equivalence of the extinction spectra and differences in reflectivity (Fig. 73.4b,c dashed). Importantly, Fresnel thin fi m and coupled dipole simulations demonstrate asymmetric reflectivity both with and without the inclusions of an 8 nm Au layer (which serves as the substrate for growth of the assembly), though inclusion of the Au layer leads to the strongest agreement with experiment. To probe the sensitivity of reflectivity to layering identity, two structures were synthesized that alternate three layers each of Ag and Au nanoparticles, where the only difference is the identity of the starting layer. Again, the extinction of both structures is similar (Fig. 73.4d) but the reflectivity is again asymmetric, here between 475 and 525 nm (Fig. 73.4e). This clearly demonstrates that the fine structure of the plasmonic assembly and layering can be used to control the reflectivity with noticeable differences for just one layer. We attribute the sensitivity of reflectivity to layering identity to two factors: (i) the strong extinction coefficients and attenuation properties of plasmonic assemblies [13] and (ii) interlayer interference to which the reflectivity is sensitive [28]. The plasmonic assemblies in this work (~5% metal by volume) significantly attenuate light, with skin depths of approximately ~100 nm for Ag at 400 nm and ~250 nm for Au layers at 520 nm [3]. Comparison of how efficiently different wavelengths absorb as a function of film thickness reveals that the wavelengths which favor reflection depend on layering order.

1345

1346

Plasmonic Metallurgy Enabled by DNA

Figure 73.4 Asymmetric reflectivity in bimetallic plasmonic assemblies. (a) Scheme depicting asymmetric reflectivity (and symmetric extinction) in bimetallic assemblies which depends on the orientation of the assembly to incident light. (b, c) Extinction (b) and reflection (c) of bimetallic assemblies where the layers are segregated by composition (i.e., three layers of Ag nanoparticles on three layers of Au nanoparticles, and its inverse). (d, e) Extinction (d) and reflection (e) of bimetallic assemblies where Ag nanoparticle layers and Au nanoparticles layers alternate.

Interference effects also influence the reflectivity; the assemblies are optically thick enough (t ≈ λ/n to t ≈ λ /3n) to exhibit interference

Plasmonic Metallurgy Enabled by DNA

effects that typically onset when t > λ /4n (assuming n = 1.6 at its maximum from 300 to 700 nm). Tunable reflection has been demonstrated in strongly absorbing systems as thin as ~λ/13n [29]. Indeed, asymmetry in bimetallic assemblies is predicted to occur with as few as two homogeneous layers (t ≈ λ /4n to t ≈ λ /9n). Together, Fresnel–EMT theory [6, 23, 24] (which only includes optical properties that depend on a locally averaged optical response) and coupled dipole calculations (which explicitly describe the nanoparticles) clarify which factors cause asymmetric reflectivity in multi-metallic plasmonic assemblies. Recall that the properties of such structures are not accurately described by a homogeneous material approximation, i.e., as an average of Au and Ag dielectric properties, so it is the fine structure of the assembly that dictates the optical response. We have demonstrated that the rational and deliberate mixing of silver and gold across multiple length scales enables tunable optical properties in DNA-programmed plasmonic assemblies. Color (and absorption) can be systematically tuned by controlling the “alloying” ratio of silver and gold on both the atomic scale (i.e., with alloy nanoparticles) and on the nanoscale (i.e., with alloy assemblies). In bimetallic assemblies with homogeneous ordered layers, reflectivity can be precisely tailored by controlling the inter-layer interactions between Ag and Au nanoparticles. Together, these data demonstrate that deliberate control over placement and ordering on the nanoscale can result in strikingly different optical properties for layered structures even though the layer structuring involves length scales much shorter than the wavelength of light. It is likely that the exploration of more complex mixed metal nanoscale systems, including those that contain other metals and those that have more complex crystalline symmetries, will further unveil diverse optical properties and functions relevant to applications in metamaterials, catalysis, and solar energy harvesting.

Acknowledgments

This research was supported by AFOSR MURI grant FA9550-11-10275 and by the Northwestern Materials Research Center under NSF grant DMR-1121262. Theory research was supported by the

1347

1348

Plasmonic Metallurgy Enabled by DNA

Department of Energy, Office of Basic Energy Science, under grant DE-FG02-09ER16109. M.B.R. and J.C.K. gratefully acknowledge support through the NDSEG graduate fellowship program. Computational time was provided by the Quest High-Performance Computing facility at Northwestern University, which was jointly supported by the Office of the Provost, the Office for Research, and Northwestern University Information Technology. GISAXS experiments were performed at beamline 12-ID-B at the Advanced Photon Source (APS), Argonne National Laboratory, and use of the APS was supported by the DOE (DE-AC02-06CH11357).

References

1. Kelly, K. L., Coronado, E., Zhao, L. L. and Schatz, G. C. (2003). J. Phys. Chem. B, 107, 668.

2. Jones, M. R., Osberg, K. D., Macfarlane, R. J., Langille, M. R. and Mirkin, C. A. (2011). Chem. Rev., 111, 3736.

3. Ross, M. B., Blaber, M. G. and Schatz, G. C. (2014). Nat. Commun., 5, 4090. 4. Soukoulis, C. M. and Wegener, M. (2011). Nat. Photonics, 5, 523.

5. Park, D. J., Zhang, C., Ku, J. C., Zhou, Y., Schatz, G. C. and Mirkin, C. A. (2015). Proc. Natl. Acad. Sci. U.S.A., 112, 977.

6. Ross, M. B., Ku, J. C., Vaccarezza, V. M., Schatz, G. C. andMirkin, C. A. (2015). Nat. Nanotechnol., 10, 453.

7. Young, K. L., Ross, M. B., Blaber, M. G., Rycenga, M., Jones, M. R., Zhang, C., Senesi, A. J., Lee, B., Schatz, G. C. and Mirkin, C. A. (2014). Adv. Mater., 26, 653. 8. Cargnello, M., Johnston-Peck, A. C., Diroll, B. T., Wong, E., Datta, B., Damodhar, D., Doan-Nguyen, V. V. T., Herzing, A. A., Kagan, C. R. and Murray, C. B. (2015). Nature, 524, 450. 9. Tao, A., Sinsermsuksakul, P. and Yang, P. (2007). Nano Lett., 2, 435.

10. Lin, M.-H., Chen, H.-Y. and Gwo, S. (2010).J. Am. Chem. Soc., 132, 11259.

11. Ariga, K., Li, J., Fei, J., Ji, Q. and Hill, J. P. (2015). Adv. Mater., doi:10.1002/ adma.201502545. 12. Mirkin, C. A., Letsinger, R. L., Mucic, R. C. and Storhoff, J. J. (1996). Nature, 382, 607.

13. Alivisatos, A. P., Johnsson, K. P., Peng, X., Wilson, T. E., Loweth, C. J., Bruchez, M. P. and Schultz, P. G. (1996). Nature, 382, 609.

References

14. Park, S. Y., Lytton-Jean, A. K. R., Lee, B., Weigand, S., Schatz, G. C. and Mirkin, C. A. (2008). Nature, 451, 553.

15. Nykypanchuk, D., Maye, M. M., van der Lelie, D. and Gang, O. (2008). Nature, 451, 549. 16. Tan, S. J., Campolongo, M. J., Luo, D. and Cheng, W. (2011). Nat. Nanotechnol., 6, 268.

17. Schreiber, R., Do, J., Roller, E.-M., Zhang, T., Schüller, V. J., Nickels, P. C., Feldmann, J. and Liedl, T. (2013). Nat. Nanotechnol., 9, 74.

18. Macfarlane, R. J., O’Brien, M. N., Petrosko, S. H. and Mirkin, C. A. (2013). Angew. Chem. Int. Ed., 52, 5688.

19. Jones, M. R., Seeman, N. C. and Mirkin, C. A. (2015).Science, 347, 1260901. 20. Senesi, A. J., Eichelsdoerfer, D. J., Macfarlane, R. J., Jones, M. R., Auyeung, E., Lee, B. and Mirkin, C. A. (2013).Angew. Chem. Int. Ed., 52, 6624.

21. Link, S., Wang, Z. L. and El-Sayed, M. A. (1999).J. Phys. Chem. B, 103, 3529. 22. Peña-Rodríguez, O., Caro, M., Rivera, A., Olivares, J., Perlado, J. M. and Caro, A. (2014).Opt. Mater. Express, 4, 403.

23. Ross, M. B., Ku, J. C., Blaber, M. G., Mirkin, C. A. and Schatz, G. C. (2015). Proc. Natl. Acad. Sci. U.S.A., 112, 10292. 24. Lazarides, A. A. and Schatz, G. C. (2000).J. Phys. Chem. B, 104, 460.

25. Koay, N., Burgess, I. B., Kay, T. M., Nerger, B. A., Miles-Rossouw, M., Shirman, T., Vu, T. L., England, G., Phillips, K. R., Utech, S., Vogel, N., Kolle, M. and Aizenberg, J. (2014).Opt. Express, 22, 27750.

26. Olson, J., Manjavacas, A., Liu, L., Chang, W.-S., Foerster, B., King, N. S., Knight, M. W., Nordlander, P., Halas, N. J. and Link, S. (2014).Proc. Natl. Acad.Sci. U.S.A., 111, 14348.

27. Yokogawa, S., Burgos, S. P. and Atwater, H. A. (2012). Nano Lett., 12, 4349. 28. Altewischer, E., van Exter, M. P. and Woerdman, J. P. (2003). Opt. Lett., 28, 1906.

29. Kats, M. A., Blanchard, R., Genevet, P. and Capasso, F. (2012). Nat. Mater., 12, 20.

1349

Chapter 74

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices*

Lin Sun,a,b Haixin Lin,b,c Kevin L. Kohlstedt,c George C. Schatz,b,c and Chad A. Mirkina,b,c aDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA bInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA [email protected]; [email protected]

Photonic crystals have been widely studied due to their broad technological applications in lasers, sensors, optical telecommunications, and display devices. Typically, photonic crystals are periodic structures of touching dielectric materials with

*Reprinted with permission from Sun, L., Lin, H., Kohlstedt, K. L., Schatz, G. C. and Mirkin, C. A. (2018). Design Principles for photonic crystals based on plasmonic nanoparticle superlattices, PNAS 115(28), 7242–7247.

Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1352

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

alternating high and low refractive indices, and to date, the variables of interest have focused primarily on crystal symmetry and the refractive indices of the constituent materials, primarily polymers and semiconductors. In contrast, finite-difference time-domain (FDTD) simulations suggest that plasmonic nanoparticle superlattices with spacer groups offer an alternative route to photonic crystals due to the controllable spacing of the nanoparticles and the high refractive index of the lattices, even far away from the plasmon frequency where losses are low. Herein, the stopband features of 13 Bravais lattices are characterized and compared, resulting in paradigmshifting design principles for photonic crystals. Based on these design rules, a simple cubic structure with an ~130 nm lattice parameter is predicted to have a broad photonic stopband, a property confirmed by synthesizing the structure via DNA programmable assembly and characterizing it by reflectance measurements. We show through simulation that a maximum reflectance of more than 0.99 can be achieved in these plasmonic photonic crystals by optimizing the nanoparticle composition and structural parameters.

Significance

In this article, we derive a set of design principles for making photonic crystals with desired photonic stopband properties by taking advantage of spacer group, a design parameter enabled by recent advances in bottom-up assembly processes. The concept of spacer groups is experimentally realized through DNAprogrammable assembly of Au nanoparticles, showing that highly reflective structures can be generated with cubic lattices and flexible spacer groups that can enable lighter and more compact 3D photonic crystals with precisely designed and even reconfigurable photonic properties. It also allows one to explore the combined effects of both photonic bandgap and the plasmonic properties of NPs, which may prove useful in fields spanning plasmonic cavity structures, optical nanocircuits, subwavelength imaging, and low-loss metamaterials.

74.1

Introduction

Photonic crystals (PCs) are materials with periodically varied refractive indices, in which optical control is achieved by refractive

Introduction

index contrast and diffraction. When the effective wavelength of light satisfies the Bragg criterion (when the wavelength is twice the periodicity), light propagation in certain directions inside the material is “forbidden” [1]. This gives rise to a photonic band gap (PBG) conceptually analogous to the electronic band gap in semiconductors. PCs have been intensively studied for use in a wide range of technologies, such as semiconductor lasers, optical integrated circuits, optical switches, and solar cells [2–5]. In addition, they are commercially used for light-emitting diodes, sensors, and optical fibers. Conventionally, PCs are structures made of dielectric materials (e.g., polymers and semiconductors) and prepared via top-down [6, 7] or bottom-up fabrication processes [8–10]. In general, bottomup techniques are attractive, because they are often simpler, less expensive, and more scalable [9, 11]. PCs made with bottom-up processes are typically close packed or touching, and the techniques used to make them provide little control over crystal symmetry and lattice parameter [12]. Thus far, the primary considerations in designing PCs have been crystal symmetry and the choice of dielectric materials to increase the index contrast between the high- and low-index materials, which is crucial for achieving good photonic properties [13–16]. However, there remain several intrinsic challenges, including overcoming the strain that results from interfacing materials with large lattice mismatches [17], poor crystal quality [13], and low index contrast [16, 18]. Moreover, since the high-index materials are closely spaced and sometimes touching [2, 15], planes along the light propagation direction are always composed of a mixture of both high- and lowindex materials—there is no well-defined separation between the high- and low-index layers, which reduces the index contrast. Hence, alternative fabrication methods that incorporate high-index materials and allow spatial separation of the high- and low-index materials may help solve challenges in PC fabrication and property tailoring. With the advent of methods for chemically programming the formation of colloidal crystals [19, 20], additional variables can be used to tune PC properties. For example, interparticle distance, which can be finely tuned with some of the emerging techniques that incorporate spacer groups for particle assembly [21, 22], represents

1353

1354

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

a yet-to-be explored parameter that could prove useful in PC design. Herein, we use finite-difference time-domain (FDTD) simulations and take into account the Bragg criterion to study the origin of the PBG along certain directions for 13 of 14 Bravais lattices composed of Au nanoparticles (NPs). From this analysis, general design principles are established, and based on these design rules, a simple cubic (cP) structure with an ~130 nm lattice parameter is predicted to exhibit a broad photonic stopband, a property confirmed by synthesizing it via DNA programmable assembly [19, 22, 23] and characterizing it by reflectance measurements. Most of the data pertain to AuNPs, but the lessons learned extend to other plasmonically active materials. Importantly, we show through simulation that a maximum reflectance of >0.99 can be achieved by optimizing NP composition and crystal structural parameters, despite the concern that metallic plasmonic NPs are lossy.

74.2

Building PCs with Plasmonic NP Superlattices

Because the NP size is smaller than the free electron mean free path, the free electrons in AuNPs are confined and interact strongly with light, giving rise to a localized surface plasmon resonance (LSPR) that can be used to focus light beyond the diffraction limit [24]. The frequency of the LSPR is highly dependent on the NP’s size and shape, the dielectric environment, and the presence of neighboring plasmonic NPs [25, 26]. Consequently, plasmonic NPs have been exploited for uses in many fields spanning chemical and biological sensing [27], Raman spectroscopy [28], nanoantennas [29], and therapeutics [30]. Other than having local control over detailed properties at the nanoscale, the macroscopic properties of the NP ensemble, such as its effective refractive index (neff), can also be tuned by various structural parameters [31]. Particularly, within a range of volume fraction where plasmonic coupling is not too strong, the neff of spherical NP ensembles can be calculated using Maxwell–Garnett effective medium theory (EMT) [32], allowing qualitative understanding of the structure dependence of their optical properties. Although metal NPs are absorptive close to the plasmon resonance frequency, EMT studies have shown that the

Building PCs with Plasmonic NP Superlattices

real part of the permittivity can be enhanced well away from the plasmon frequency, where absorption is relatively small. Also, the optical response associated with plasmonic NPs is so strong that it is possible to use lattices with relatively low volume fraction (i.e., the NPs are highly separated) in constructing functional photonic lattices. Therefore, ensembles of plasmonic NPs serve as a promising candidate for the high-index material in PCs [plasmonic photonic crystals (PPCs)] (Fig. 74.1). Although a high effective index can be realized in plasmonic NP ensembles and there has been extensive work on plasmonic NP assemblies [20, 33, 34], large stopbands are not typically observed in such structures due to the dense-packed arrangements of NPs [35, 36]. This analysis is consistent with the conclusion that plasmonic NPs must be well-separated to design effective PPCs.

Figure 74.1 Schematic representation describing the design of PPCs with AuNPs. The stopband features that are generated by light incident normal to the x-y plane are investigated. Along the z direction, the superlattice can be viewed as alternating NP and dielectric layers with high and low indices, respectively; 13 of 14 Bravais lattices are studied. In the layered structure scheme, the NPs are embedded in a homogeneous matrix. mC, base-centered monoclinic; oC, base-centered orthorhombic; oF, face-centered orthorhombic; oI, bodycentered orthorhombic.

1355

1356

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

Conceptually, a periodic structure can be achieved by building a crystal where the NPs are located at each lattice site and embedded in a homogeneous dielectric medium. For a chosen direction relative to the crystal lattice, each crystalline plane contains a layer of NPs that serves as the high-index layer, and the medium between each lattice plane serves as the low-index layer. The crystal can have different lattice symmetries and lattice constants, which in turn, will affect the effective refractive index (neff) of the NP layer and the periodicity. To systematically investigate the effect of structural parameters on the photonic properties and provide guidelines for the construction of PPCs with large PBGs along selected directions, we have studied the optical properties of 13 of the 14 Bravais lattices to identify the roles that each lattice parameter plays (Fig. 74.1). In particular, FDTD simulations are carried out to investigate how the lattice parameters of PPCs—the lattice constants in all three dimensions and the angle—affect the photonic stopband features. We set up lattices with spherical AuNPs (diameter of 108 nm), all embedded in a silica matrix (as such structures can be made experimentally). To minimize computational time and keep the simulations tractable, we found that seven layers of NPs sufficiently represented the thin PPC films and were used in all simulations unless otherwise specified. Normal incidence of light (z direction) onto the (001) plane (x-y plane) of the lattice is investigated (Fig. 74.1). Rule 1: Layer Periodicity Compared with the Lattice Constant along the Light Propagation Direction Dictates the Location of the Photonic Stopband

Figure 74.2 summarizes the maximum reflectance (Rmax) of the stopband and the peak wavelength, at which Rmax is obtained, for cP, body-centered cubic (cI), and face-centered cubic (cF) lattices as a function of the periodicity of each layer along the z direction (Fig. 74.2A). In other words, a single cI or cF unit cell has three layers (top, middle, bottom), and the periodicity of each layer in the z direction is one-half of the lattice constant (Fig. 74.2A). A clear trend between the peak wavelengths and layer spacing is observed between the three lattice types (Figs. 74.2B and 74.2C). It is remarkable how the peak wavelengths and maximum reflectance share similar

Building PCs with Plasmonic NP Superlattices

values and trends, especially above 200 nm layer spacing. As the lattice constant increases, the peak wavelength increases, and Rmax increases until it reaches a maximum value and then decreases. This shows the importance of controlling the lattice constant, which is not easily achievable with conventional fabrication techniques. When the properties are plotted as a function of lattice constant, the correspondence between the three lattice types becomes weak. This suggests that, instead of the lattice constant in the z direction (c), the spacing between each NP layer in the light propagation direction (layer periodicity) should be considered as the periodicity of the PC (Fig. 74.2A).

Figure 74.2 Layer periodicity dictates the location of the photonic stopband. (A) If the superlattice is viewed as alternating layers that do and do not contain the NPs, then the layer periodicity is defined as the spacing between two adjacent layers that contain NPs. (B and C) Dependence of (B) maximum reflectance (Rmax) and (C) its corresponding wavelength (λpeak) on the layer periodicity of cP, cI, and cF superlattices. (D and E) Dependence of (D) Rmax and (E) center wavelength (λ0) on the layer periodicity for tP and tI lattices. The lattice constant within each layer is kept constant at 200 nm. Only the λ values of a reflectance larger than 0.9 are considered as the stopband, and its λ0 is plotted as a function of layer periodicity in panel (E).

1357

1358

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

The above point is directly shown in lattices with larger layer periodicity (>200 nm) (Fig. 74.2C), where the peak wavelength increases almost linearly, compared with the slow change at smaller periodicities. This complication arises, because plasmonic NPs are used as building blocks. The stopband overlaps with the plasmonic modes of the NP lattice at smaller lattice constants (140–200 nm). In this situation, the exact wavelength of the stopband becomes hard to predict. We simplify the discussion further by focusing on lattices where the plasmonic modes and stopband are spectrally separated. This also allows us to later test the feasibility of using EMT to estimate the neff and reproduce the reflectance spectra, since the EMT prediction is less accurate on the blue side of the LSPR where quadrupole modes are important in the NP response and there is more significant absorption. In addition, as the lattice constant further increases and the first-order Bragg peak red shifts, a second-order peak arises. For simplicity, only first-order peaks are considered. Although the cubic lattice system can provide insight into the structure-function relationship of PPCs, the situation is limited due to the high crystal symmetry in the cubic lattice system, prompting the investigation of other lattice symmetries. Next, we move to investigate tetragonal crystal systems. Compared with the cP lattice, c is different from the lattice constant in the x-y plane (a), which allows more freedom in teasing out the structure-function relationship. Figure 74.2D shows the dependence of the Rmax on the layer periodicity of both tetragonal lattices (tP) while keeping a constant (200 nm). The largest Rmax is reached with c ≈ 240 nm. Additional inspection shows that the periodicity at which the largest Rmax is obtained involves an optimization of both layer periodicity and layer number. To further determine the characteristics of the stopband, we study only those with their Rmax larger than 0.9 and define the bandwidth (Δλ) to be the width of the band with reflectance ≥0.9. Fig. 74.2E shows again that the wavelength of the center of the band (λ0) depends linearly on the layer periodicity. Although Δλ increases, the normalized bandwidth Δλ/λ0 decreases due to the faster increase in λ0. Interestingly, the properties of bodycentered tetragonal (tI) lattices are almost identical to a tetragonal counterpart with the same layer periodicity, which we emphasize again is not c but c/2 in tI lattices. This indicates that the properties

Building PCs with Plasmonic NP Superlattices

of the stopband may be independent of the relative position of the NPs between different layers as discussed below.

Volume fraction

Volume fraction

Figure 74.3 The volume fraction of each NP layer dictates the Rmax, λ0, and Δλ/λ0 of the stopband. (A) The volume fraction within each NP layer is defined in the equation, where N is the number of NPs in each unit cell, Vsphere is the volume of the NP, and A × (2r) is the volume of a unit cell: A is the area of the facet in the x-y plane of the unit cell, r is the radius of the NP. Thus, the diameter of the NP defines the thickness of the NP layer. (B) Rmax, (C) λ0, and (D) normalized bandwidth (Δλ/λ0) of tP, oP, and hP lattices as a function of volume fraction within each NP layer. One lattice constant within the NP layer is chosen for body-centered orthorhombic (oI), face-centered orthorhombic (oF), and base-centered orthorhombic (oC), and the stopband features show good agreement with the oP lattices as predicted. The layer periodicity of all data points is 240 nm.

Rule 2: The Volume Fraction Rather Than the Exact NP Arrangement of Each NP Layer Dictates the Bandwidth Now that layer periodicity has been identified as a key structural parameter, we investigate the relationship between a (with fixed layer periodicity) and stopband features. As a increases, Rmax, λ0, and Δλ/λ0 all decrease due to the reduced index contrast between the NP and the silica layer as the amount of Au in the NP layer is diluted by increasing a. Next, we explore the stopband features of orthorhombic lattices (oP) and hexagonal lattices (hP) and compare them with those of the tetragonal structures (Fig. 74.3). The layer

1359

1360

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

periodicity is kept constant at 240 nm for all lattice types. From the perspective of the NP arrangement in each layer, the NPs are arranged in squares, in rectangles with the length of one side fixed (b = 250 nm), and in diamonds with 60° angle, while a is varied for tP, oP, and hP lattices, respectively (Fig. 74.3A). To enable comparison between different lattice symmetries, the properties are plotted against the NP volume fraction within each NP layer (Fig. 74.3A). A striking fact is that all of the plots for the three different lattice structures overlap. The stopband is dictated by the volume fraction rather than the exact arrangement of the NPs within each layer for fixed NP size and layer periodicity. We see in Fig. 74.3B that, as volume fraction increases, the Rmax increases monotonically until it saturates at ~0.97, and the 0.03 loss is due to absorption by the material. Interestingly, λ0 and Δλ/λ0 increase with increasing volume fraction (decreasing lattice constant); this is a consequence of the low-energy band edge experiencing a blue shift, while the highenergy edge remains relatively unchanged.

Rule 3: The Nanoparticle Registry between Layers Does Not Affect Stopband Features Features of the stopband of trigonal lattices (hR) are summarized in Fig. 74.4B–D. Here, the lattice constant is fixed at 240 nm, while the angle is changed from 40° to 80°, and both Rmax and λ0 increase (Fig. 74.4B–D). So far, we have shown the important roles that layer periodicity and volume fraction in each layer play, while the exact arrangement of NPs in each layer has negligible influence. However, the effect of registry between different layers has not been considered. While our observations in all of the body-centered and face-centered structures (cubic and tetragonal) indicate that registry plays a trivial role, its effect can be studied more clearly in monoclinic lattices. A monoclinic lattice (mP) allows even more degrees of freedom, where all three lattice constants (a, b, and c) and the angle (α) between layers can be independently varied. Two sets of simulations are performed where the lattice constants in each layer are fixed at a = 250 nm and b = 160 nm, while the lattice constant in the [001] direction is set, such that either the layer periodicity (fixed z) or c (fixed c) is kept constant. Rmax, λ0, and Δλ/λ0 of the lattice with fixed c show a similar trend to that of the hR; in comparison, those of the fixed z lattices are constant, and the reflectance spectra overlap

PPCs Realized through DNA-Programmable Assembly

at all α. This proves that the registry between different layers indeed does not affect the stopband features of the lattice. This observation coincides with our expectation that the relative arrangement of NPs in each layer and between layers does not matter (when the NPs are spaced such that no strong plasmonic interaction occurs).

A

B

fixed c

fixed z

C

D

Figure 74.4 The registry between NP layers has a negligible effect on the stopband features. (A) Two sets of structural parameters in mP and basecentered monoclinic (mC) lattices with either fixed layer periodicity (fixed z) or fixed lattice constant c (fixed c) while changing the angle. (B) Rmax, (C) λ0, and (D) normalized bandwidth (Δλ/λ0) of hR, mP, and mC lattices with either fixed c or z as a function of the angle α. In the fixed c (z) case, c (z) is 240 nm.

To further explore this concept, we discuss the feasibility of using EMT combined with the transfer matrix method [37] (TMM; EMT + TMM) to reproduce the spectra calculated by FDTD. The good qualitative agreement between the EMT + TMM and FDTD results indicates that the superlattice properties in the z direction can be treated as alternating layers of high (NP)- and low (dielectric)-index materials.

74.3

PPCs Realized through DNA-Programmable Assembly

We have derived guidelines to design PPCs from analysis of the stopband features along one dimension; however, the results can be applied to any direction for a 3D PPC; 3D PPCs offer more compact

1361

1362

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

Wavelength (nm)

Figure 74.5 Experimental measurement and FDTD simulations of cubic NP superlattices. (A) Schematic representation (lower left) and scanning electron microscope image (lower right) of a superlattice made through DNA-programmable assembly of nanocubes after encasing in silica. The Au nanocube building blocks have an 88 nm edge length and a 5 nm corner rounding. The lattice constant of the superlattice is 134 nm and defined by the duplex DNA interconnects (sequences used: anchor strand: TCA ACT ATT CCT ACC TAC AAA AAA AAA A SH; linker strand: GTA GGT AGG AAT AGT TGA A TTTTTTTTTTTT ACT GAG CAG CAC TGA TTTTTTTTTTTTT A GCGC; and duplexer strand: AAAAAAAAAAAAA TCA GTG CTG CTC AGT AAAAAAAAAAAA; all strands are listed from 5¢ to 3¢). An enlarged view of one hybridized DNA pair between nearest-neighbor nanocubes in a lattice is shown in upper. (B, upper) Simulation result of a cP superlattice with spherical NPs that has the same lattice constant and volume fraction as the superlattice shown in A. (B, lower) Simulation and experimental results for the superlattice shown in (A).

design and are sought after for applications involving all-optical integrated circuits [9]. Here, we explore experimental methods for realizing such 3D PPCs. DNA-programmable assembly is a promising emerging method for making PPCs, since it provides fine control and even subnanometer tunability over particle spacing [21–23]. Moreover, interparticle spacing can be dynamically tuned after PC formation [21, 38], enabling dynamic tuning of the stopband location. Here, as proof of concept, we use micrometer-sized cP superlattices with well-faceted cubic crystal habits [39] made from DNA-functionalized cubic NPs to experimentally explore this concept. The cubic crystal habit facilitates the alignment of the microcrystal with its (001) facet facing up. Specifically, cubic NPs with 88 ± 4 nm edge length and 5 ± 1 nm corner rounding were used, and the

Building PCs with Materials Other Than Au

lattice constant as measured by small angle X-ray scattering was 134 nm. The simulation results for a superlattice made of spherical NPs with the same volume and lattice constants show the existence of a broad stopband (Fig. 74.5B, upper), a property that is observed both experimentally and through simulation in the cubic NP superlattice (Fig. 74.5B, lower). Moreover, the experimentally observed stopband matches remarkably well with the simulation prediction, emphasizing that the design rules articulated above are not limited to spherical particles. Indeed, NPs with different shapes can be used as the building blocks for superlattices that show similar stopband properties. Although only a cP lattice is studied experimentally, other lattice structures (over 500 different crystals spanning over 30 different crystal symmetries) have been made through DNA programmable assembly [19, 22]. For lattices with symmetries that do not belong to the cubic lattice system, one can obtain different stopband properties using different crystal orientations of the same superlattice.

74.4

Building PCs with Materials Other Than Au

Although the aforementioned design principles have been studied for spherical AuNPs, they also extend to other plasmonic NPs and lattice structures. Indeed, by performing a set of simulations on seven-layer cubic superlattices (134 nm lattice constant) composed of 88 nm edge-length cube-shaped NPs consisting of different plasmonic materials (Ag, Au, Al, or Cu), a broad stopband can be observed for all structures. The superlattice made with Ag particles exhibits the highest reflectance (Rmax = 0.986) and lowest absorption, a consequence of the lower losses in Ag. Indeed, Rmax can be further increased on optimization of lattice constant and particle size. One simulation on a structure consisting of the same cube-shaped NPs in a tP with a = 140 and c = 300 nm shows that an Rmax greater than 0.996 can be achieved. This large value is comparable with that of high-quality dielectric PCs and indicates the suitability of AgNP superlattices for a wide range of applications. It is also important to note that, since the plasmonic resonance of Al NPs can be tuned into the UV region, distinct stopbands can be realized to cover the entire visible region with Al NP superlattices.

1363

1364

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

Since the advantageous properties of PPCs are attributed to the large index contrast between the NP and dielectric layers, even far away from the plasmon frequency, we also have looked into the role of plasmonic NPs by comparing results for plasmonic NP superlattices with those for superlattices composed of other metallic NPs with poor plasmonic properties (Cr, Fe, Ti). While a significant stopband is present for the plasmonic NP superlattices, it is absent for lattices made from NPs with poor plasmonic properties. By varying the number of layers and the lattice constants, we show that the stopband features for superlattices made of TiNPs are always much weaker than those for superlattices made of AuNPs with similar structures over a wide range of structural parameters. Additionally, to benchmark our technique with conventional PC fabrication methods (where dielectric materials are in general used), we investigated the stopband features of lattices made with various dielectric NPs through simulation. These lattices obey the same design principles derived above. However, the stopbands of lattices made with dielectric NPs saturate much more slowly, and the band widths are much narrower compared with the plasmonic NP ones with the exact same lattice parameters. Finally, we explore the effect of dielectric medium. This shows that, with the same superlattice, the spectral location of the stopband can be tuned, in a similar fashion as the LSPR, by immersing or embedding it in a different dielectric medium [12].

74.5

Conclusions

Through a systematic study of the stopband features of 13 Bravais lattice structures along the z direction, we have shown that nontouching PPC superlattices can be treated as periodically alternating layers of high- and low-index materials along the light propagation direction. We have identified two key parameters that dictate the stopband features: the layer periodicity and the volume fraction of each NP layer. Interestingly, when the NPs are spaced sufficiently far away such that strong plasmonic coupling is minimized, the exact arrangement of NPs in each layer and the registry between different layers have negligible influence on the stopband properties. From a fabrication standpoint, this work

Methods

conclusively shows that DNA-programmable assembly is especially and perhaps uniquely useful for making 3D PPCs, since it provides control over the spacing between NPs and lattice symmetry. The high tunability of the stopband features realized through this technique (by changing lattice parameters, NP size and composition, and the dielectric matrix) should lead to PPCs with interesting applications as cavities and filters. Compared with PCs made with purely dielectric materials, PPCs are smaller in size and lighter in weight [40]. For example, with cP lattices, a saturated stopband is realized for a superlattice with a lattice constant of only 140 nm and a seven-layer thickness (the total thickness is ~1 μm). Importantly, the volume fraction of the metal for a lattice with 200 nm particle spacing is less than 0.10. Moreover, the technique, which allows one to explore the combined effects of both the PBG and the plasmonic properties of NPs, hold promise for making and exploring PPC materials that may prove useful in plasmonic cavity structures [41, 42], optical nanocircuits [43], subwavelength imaging [44], and lowloss metamaterials [45].

74.6

74.6.1

Methods

FDTD Calculations

The FDTD simulations were performed with Lumerical FDTD solutions. The structures were designed as NP lattices embedded in a homogeneous dielectric background.

74.6.2

EMT Approximation and TMM

Both are performed in Matlab. Maxwell–Garnett EMT approximation is used to approximate the refractive indices of superlattices.

74.6.3

Superlattice Assembly

DNA strands used in this work were designed according to the literature and synthesized on a solid support with an MM48 synthesizer (BioAutomation). Nanocubes were synthesized

1365

1366

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

according to the literature. DNA functionalization and assembly were done according to the literature.

74.6.4

Optical Experiments

Reflectance was performed using a Zeiss microscope (Axio Observer Z1) coupled with a spectrometer (50 g/mm grating; Princeton Instrument and charge coupled device; PyLoN). A Xenon lamp (XBO 75) with a broadband spectrum (300–1000 nm) was used as the light source.

Acknowledgments

This material is based on work supported by the following awards: Air Force Office of Scientific Research Grant FA9550-171-0348 (FDTD simulation); Asian Office of Aerospace Research and Development (AOARD) Grant FA2386-13-1-4124 (optical measurement); the Center for Bio-Inspired Energy Science, an Energy Frontier Research Center funded by US Department of Energy, Office of Science, Basic Energy Sciences Award DE-SC0000989 (DNA-programmable assembly); Department of Energy Grant DE-SC0004752 (theory methods); and National Science Foundation Grant CHE-1414466 (transfer matrix analysis). This research was supported in part through the computational resources and staff contributions provided for the Quest High-Performance Computing facility at Northwestern University, which is jointly supported by the Office of the Provost, the Office for Research, and Northwestern University Information Technology. This work made use of the Electron Probe Instrumentation Center (EPIC) facility of Northwestern University’s Northwestern University Atomic and Nanoscale Characterization Experimental Center (NUANCE) Center, which has received support from the Materials Research Science and Engineering Center (MRSEC) Program (National Science Foundation Grant DMR-1121262) at the Materials Research Center. L.S. acknowledges the International Institute for Nanotechnology (IIN) for the Ryan Fellowship. H.L. acknowledges the IIN for the IIN Postdoctoral Fellowship.

References

References 1. Joannopoulos, J. D., Johnson, S. G., Winn, J. N. and Meade, R. D. (2008). Photonic Crystals: Molding the Flow of Light (Princeton Univ. Press, Princeton). 2. Norris, D. J. (2007). Nat. Mater., 6, 177–178.

3. Maldovan, M. and Thomas, E. L. (2004). Nat. Mater., 3, 593–600.

4. Huang, M. C. Y., Zhou, Y. and Chang-Hasnain, C. J. (2007). Nat. Photonics, 1, 119–122. 5. Arsenault, A. C., et al. (2006). Nat. Mater., 5, 179–184.

6. Yablonovitch, E., Gmitter, T. J. and Leung, K. M. (1991). Phys. Rev. Lett., 67, 2295–2298.

7. Miyake, M., Chen, Y.-C., Braun, P. V. and Wiltzius, P. (2009). Adv. Mater., 21, 3012–3015. 8. Hermatschweiler, M., Ledermann, A., Ozin, G. A., Wegener, M. and von Freymann, G. (2007). Adv. Funct. Mater., 17, 2273–2277.

9. Vlasov, Y. A., Bo, X.-Z., Sturm, J. C. and Norris, D. J. (2001). Nature, 414, 289–293.

10. Nelson, E. C., et al. (2011). Nat. Mater., 10, 676–681.

11. von Freymann, G., Kitaev, V., Lotsch, B. V. and Ozin, G. A. (2013). Chem. Soc. Rev., 42, 2528–2554.

12. García-Santamaría, F., Salgueiriño-Maceira, V., López, C. and LizMarzán, L. M. (2002). Langmuir, 18, 4519–4522.

13. Jang, S. J., Song, Y. M., Yeo, C. I., Park, C. Y. and Lee, Y. T. (2011). Opt. Mater. Express, 1, 451–457.

14. Ya-Chih, T., Kenneth, W. K. S. and John, B. P. (1998). J. Phys. Condens. Matter, 10, 753. 15. Bulletin, M. R. S., Armstrong, E. and O’Dwyer, C. (2015). J. Mater. Chem. C, 3, 6109–6143.

16. Waterhouse, G. I. N. and Waterland, M. R. (2007). Polyhedron, 26, 356– 368. 17. Liu, Y.-S., et al. (2016). Appl. Phys. Lett., 109, 081103.

18. Oh, J. R., Moon, J. H., Yoon, S., Park, C. R. and Do, Y. R. (2011). J. Mater. Chem., 21, 14167–14172.

19. Jones, M. R., Seeman, N. C. and Mirkin, C. A. (2015). Science, 347, 1260901.

1367

1368

Design Principles for Photonic Crystals Based on Plasmonic Nanoparticle Superlattices

20. Boles, M. A., Engel, M. and Talapin, D. V. (2016). Chem. Rev., 116, 11220–11289. 21. Kim, Y., Macfarlane, R. J., Jones, M. R. and Mirkin, C. A. (2016). Science, 351, 579–582. 22. Macfarlane, R. J., et al. (2011). Science, 334, 204–208.

23. Mirkin, C. A., Letsinger, R. L., Mucic, R. C. and Storhoff, J. J. (1996). Nature, 382, 607–609.

24. Liz-Marzán, L. M. (2006). Langmuir, 22, 32–41.

25. Kelly, K. L., Coronado, E., Zhao, L. L. and Schatz, G. C. (2003). J. Phys. Chem. B, 107, 668–677.

26. Halas, N. J., Lal, S., Chang, W.-S., Link, S. and Nordlander, P. (2011). Chem. Rev., 111, 3913–3961.

27. Kabashin, A. V., et al. (2009). Nat. Mater., 8, 867–871.

28. Laor, U. and Schatz, G. C. (1981). Chem. Phys. Lett., 82, 566–570. 29. Akselrod, G. M., et al. (2014). Nat. Photonics, 8, 835–840.

30. Giljohann, D. A., et al. (2010). Angew. Chem. Int. Ed. Engl., 49, 3280– 3294. 31. Kubo, S., et al. (2007). Nano Lett., 7, 3418–3423.

32. Ross, M. B., Mirkin, C. A. and Schatz, G. C. (2016). J. Phys. Chem. C, 120, 816–830.

33. Henzie, J., Grünwald, M., Widmer-Cooper, A., Geissler, P. L. and Yang, P. (2011). Nat. Mater., 11, 131–137.

34. Shevchenko, E. V., Talapin, D. V., Kotov, N. A., O’Brien, S. and Murray, C. B. (2006). Nature, 439, 55–59.

35. Tao, A. R., Ceperley, D. P., Sinsermsuksakul, P., Neureuther, A. R. and Yang, P. (2008). Nano Lett., 8, 4033–4038. 36. Karg, M., Hellweg, T. and Mulvaney, P. (2011). Adv. Funct. Mater., J. Mater. Chem., 4668–4676. 37. Smith, D. R., Vier, D. C., Koschny, T. and Soukoulis, C. M. (2005). Phys. Rev. E: Stat. Nonlinear Soft Matter Phys., 71, 036617.

38. Mason, J. A., et al. (2016). J. Am. Chem. Soc., 138, 8722–8725.

39. O’Brien, M. N., Lin, H.-X., Girard, M., Olvera de la Cruz, M. and Mirkin, C. A. (2016). J. Am. Chem. Soc., 138, 14562–14565.

40. Sigalas, M. M., Chan, C. T., Ho, K. M. and Soukoulis, C. M. (1995). Phys. Rev. B, 52, 11744–11751.

41. Ruben, E., Javier, A. and Garnett, W. B. (2014). New J. Phys., 16, 013052.

References

42. Park, D. J., et al. (2017). Proc. Natl. Acad. Sci. U.S.A., 114, 457–461. 43. Engheta, N. (2007). Science, 317, 1698–1702.

44. Luo, C., Johnson, S., Joannopoulos, J. and Pendry, J. (2003). Opt. Express, 11, 746–754.

45. Xiao, S., et al. (2010). Nature, 466, 735–738.

1369

Chapter 75

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled by DNA-Programmable Assembly*

Matthew R. Jones,a,b Kevin L. Kohlstedt,b,c Matthew N. O’Brien,b,c Jinsong Wu,a George C. Schatz,b,c and Chad A. Mirkina,b,c aDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA bInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA [email protected]

The physical properties of matter rely fundamentally on the symmetry of constituent building blocks. This is particularly true for structures that interact with light via the collective motion of

*Reprinted with permission from Jones, M. R., Kohlstedt, K. L., O’Brien, M. N., Wu, J., Schatz, G. C. and Mirkin, C. A. (2017). Deterministic symmetry breaking of plasmonic nanostructures enabled by DNA-programmable assembly, Nano Lett. 17(9), 5830– 5835. Copyright (2017) American Chemical Society.

Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1372

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

their conduction electrons (i.e., plasmonic materials), where the observation of exotic optical effects, such as negative refraction and electromagnetically induced transparency, require the coupling of modes that are only present in systems with nontrivial broken symmetries. Lithography has been the predominant fabrication technique for constructing plasmonic metamaterials, as it can be used to form patterns of arbitrary complexity, including those with broken symmetry. Here, we show that low-symmetry, 1D plasmonic structures that would be challenging to make using traditional lithographic techniques can be assembled using DNA as a programmable surface ligand. We investigate the optical properties that arise as a result of systematic symmetry breaking and demonstrate the appearance of π-type coupled modes formed from both dipole and quadrupole nanoparticle sources. These results demonstrate the power of DNA assembly for generating unusual structures that exhibit both fundamentally insightful and technologically important optical properties. Self-assembly approaches that utilize colloidal nanoparticles as building blocks have a number of advantages over lithographic methods for the construction of plasmonic metamaterials: (1) solution-phase synthesized nanoparticles are often single crystalline and possess superior optical properties due to their low defect density [1], (2) particles can be generated in large quantities and arranged simultaneously, resulting in increased yield and scale [2], (3) the length scales that are routinely accessed via selfassembly methods are considerably smaller than those available to traditional lithographic techniques, as they are defined by molecular interactions [3, 4] (4) solution-dispersible materials exist in a 3D environment and can exhibit optical properties that are less sensitive to orientation than those confined to a 2D substrate [5]. However, the majority of self-assembly methodologies rely on spherical building blocks that interact via nearly isotropic potentials that can result in densely packed, complex structures but with invariably high-symmetry [6−9]. Consequently, these approaches offer limited access to nanoparticle arrangements that exhibit metamaterial-like properties [10−14].

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

Directed assembly based on sequence-programmable DNA interactions represents a particularly powerful approach for synthesizing complex particle-based architectures [15−19]. Nucleic acids can either be folded into discrete molecular templates with location-specific nanoparticle binding sites [18, 20, 21], or anchored to rigid nanoparticle cores that act to bundle and orient strands in a surface-normal direction [6, 18, 22], constructs known as programmable atom equivalents (PAEs) [19]. In this work, we exploit three important features of the assembly of PAEs to construct symmetry-broken, plasmonically coupled nanoparticle superlattices: (1) The use of anisotropic building blocks whose shape imposes directional interactions between particles that lead to reduced-dimensionality superlattices [23−28] (2); a DNA design in which two sets of differently sized or shaped particles are modified with oligonucleotides that have complementary sequences, forcing the co-crystallization of structurally dissimilar particles in a manner that breaks symmetry [29] and (3) the use of nonspherical plasmonic nanoparticles that support strongly confined dipolar and quadrupolar resonances [30, 31], necessary for the observation of extended π-type coupled modes that are more amenable to metamaterial behavior than the more common σ-type modes [32]. This set of design elements allows for the generation of extremely sophisticated assembled superlattices that would be impractical to fabricate with traditional lithographic methods. The particle building blocks that form the basis of this work are 2D plate-like nanostructures composed of gold [31]. These materials are sufficiently thin (~7 nm) and anisotropic (lateral dimensions tunable between 50 and 200 nm) so as to present a dense array of highly oriented oligonucleotides orthogonal to their broad, atomically flat facets, and considerably fewer, poorly oriented oligonucleotides orthogonal to their edges. As a consequence, these structures have been shown to greatly favor face-to-face hybridization interactions both kinetically and thermodynamically [24, 25] resulting in the formation of highly ordered 1D superlattices (Fig. 75.1a) [23]. These particles can be synthesized with a circular cross-sectional shape (circular disks), which exhibit a single in-plane dipole resonance, and a triangular cross-sectional shape (triangular prisms), which exhibit both in-plane dipole and in-plane quadrupole resonances [31, 33]. In addition, the small thickness of the particles results in

1373

1374

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

strong in-plane (longitudinal) modes with no measurable out-ofplane (transverse) modes [31], a property that will be important to understand the plasmonic response of the assembled superlattices.

Figure 75.1 DNA-programmable assembly of low-symmetry nanoparticle superlattices. (a) Schematic illustration of the numerous DNA sticky-end interactions and preferential face-to-face alignment used to drive the formation of 1D superlattices. (b) DNA design in which a nanoparticle-bound strand is hybridized to a linker oligonucleotide presenting sequence-programmable sticky ends. Small-angle X-ray scattering (c, e, g) and transmission electron microscopy (d, f, h) characterization of circular disk superlattices showing control over array width (c, d), interparticle spacing (e, f), and periodicity (g, h) by controlling the particle size, DNA length, and sticky end sequence, respectively. Simulated SAXS data for perfect crystals are shown as dotted lines directly beneath experimental data.

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

The particles are assembled using a previously reported DNA design [23, 29]: terminal hexyl-thiol-modified oligonucleotides are used to anchor and orient strands relative to the gold nanoparticle surface and programmable linker oligonucleotides are added to dictate particle interactions via the nucleobase sequence of the terminal “sticky ends” (Fig. 75.1b). Superlattices are characterized using synchrotron-based small-angle X-ray scattering (SAXS) to probe a large number of structures simultaneously and assign symmetries on the basis of the structure factor, S(q) [23, 29]. All superlattices are encapsulated in silica to preserve their lattice symmetry and spacing for imaging via transmission electron microscopy. In order to rigorously describe the assembled superlattices and quantify how symmetries are broken through rational structural modification, we have classified each system using the nomenclature of 3D group theory. Superlattices assembled from self-complementary circular disk particles (sticky end: 5¢ GCGC 3¢) with different diameters (Fig. 75.1c,d) and DNA lengths (Fig. 75.1e,f) result in monoperiodic arrays with tunable widths and interparticle spacings, respectively. If particles of two different sizes are modified with complementary sticky-end sequences (5¢ AGAGA 3¢/5¢ TCTCT 3¢), biperiodic superlattices are formed (Fig. 75.1g,h); relative to the monoperiodic array, this results in a decrease in translational symmetry. It is worth noting that the broken symmetries of these nanoparticle cocrystals would be challenging to achieve using traditional self-assembly approaches, wherein particle sizes and shapes must be precisely matched to prevent phase separation [34]. In order to understand the plasmonic consequences of programmable symmetry breaking, we utilize the conceptual framework of plasmon hybridization theory [32]. When the discrete modes of two nanostructures couple, one expects a hybrid lowenergy bonding mode, in which the dipoles on each particle interact favorably, and a hybrid high-energy antibonding mode, in which the dipoles on each particle interact unfavorably. In most cases (e.g., spherical particles) [13, 14, 22, 32], these modes arise from dipole alignment in a head-to-tail (bonding) or head-to-head (antibonding) fashion, a geometry known as σ-type hybridization. Because the antibonding configuration possesses no net dipole, it is unable to interact with light (a dipole excitation source) and is described as

1375

1376

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

optically dark. Only the low-energy (redshifted) bonding modes possess a net dipole, are optically bright, and therefore are most frequently observed in cases of nanoparticle assembly [13, 14, 22]. In contrast, the dipolar mode for circular disks is strongly confined to the plane of the particle [31] and therefore is oriented orthogonal to the unit lattice vector. This leads to an alternative side-by-side orientation of dipoles that align antiparallel to form a low-energy bonding mode and align parallel to form a high-energy antibonding mode (Fig. 75.2a). As a consequence, the low-energy bonding mode is dark and the high-energy antibonding mode is bright. We observe this effect by collecting solution-phase extinction spectra before and after the assembly of 1D monoperiodic superlattices (Fig. 75.2a,b). Indeed, the plasmon resonance shows a pronounced blueshift as a result of the antibonding π-type (i.e., side-by-side) dipole coupling in the array. Because the extent to which the antibonding mode is blueshifted should depend on the particle coupling strength and dipole magnitude, the proposed dipole alignment can be tested experimentally by decreasing the particle spacing and increasing the particle size, respectively (Fig. 75.2e,f). As predicted, this results in greater mode splitting. These results are confirmed in silico using the discrete dipole approximation (DDA) method to simulate the extinction spectra of perfectly aligned gold disk arrays. One of the most powerful features of DNA-mediated assembly is the ability to program particle interactions via sequencespecific hybridization and form cocrystalline biperiodic arrays (Fig. 75.1g,h) [18, 29]. In this case, the unequal magnitude of the two dipoles suggests that the bonding mode should no longer be dark, as the broken symmetry allows for a net dipole on the structure (Fig. 75.2c). To test this prediction, we compared the extinction spectra for isolated particles and coupled superlattices composed of two differently sized circular disks bearing complementary oligonucleotides (Fig. 75.2c,d). Indeed, biperiodic superlattices support two plasmon modes, one that is redshifted and one that is blueshifted with respect to the isolated particle plasmons. In addition, because the magnitude of the net dipole determines the extent to which light is able to excite a mode, we observe the bonding (redshifted) peak having a lower extinction than the antibonding (blueshifted) peak. Superlattices cocrystallized from combinations of disks with a larger difference in their diameters and thus more

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

disparate discrete plasmon energies show less efficient coupling, a smaller energy shift, and weaker bonding/antibonding modes (Fig. 75.2g). Simulations of the electric field profile and dipole orientations at the energies of the blueshifted and redshifted peaks clearly indicate the presence of parallel and alternating net dipole orientations, confirming the assignment of antibonding and bonding mode character, respectively (Fig. 75.2h,i).

Figure 75.2 Optical characterization of 1D nanoparticle superlattices. Plasmon mode hybridization diagrams (a, c), and extinction spectra (b, d) for monoperiodic (a, b) and biperiodic (c, d) arrays showing the absence of a low-energy π-type bonding mode (red X, a, b) except when symmetry is rationally broken (c, d). Discrete modes are shown in blue or red while coupled modes are shown in purple. Simulated data (dotted lines) is shown offset adjacent to experimental data (solid lines). Monoperiodic arrays assembled with shorter DNA lengths (e) and larger circular disks (f) show antibonding modes with greater energy shifts with respect to the discrete modes (ΔEnergy = Energyassembled − Energydiscrete). (g) Biperiodic arrays assembled from pairs of circular disks with increasing size ratio show less-efficient bonding and antibonding mode splitting. In (e−g), experimental data are shown in solid symbols while simulated data are shown in open symbols. Electric field distributions and dipole orientations calculated for biperiodic arrays at the antibonding (h) and bonding (i) mode energies confirm the parallel and antiparallel net dipole arrangements, respectively.

While π-type hybridized modes are well-known in the literature [32], their observation is almost exclusively limited to structures fabricated or deposited on substrates with optical sources intentionally polarized to excite them [10, 35, 36]. The DNAassembled superlattices described here have been probed optically

1377

1378

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

in a solution-phase, orientation-averaged measurement that negates the role of polarization-dependent effects. In addition, because the circular disks do not possess an observable transverse mode [31], the structure cannot be excited along the long dimension (lattice vector) of the superlattice and thus σ-type coupling is forbidden. As a result, the only bright modes available to the system are those with a π-type antibonding character. This geometry of dipoles is known to possess a magnetic resonance and is considered one of the fundamental building blocks of all metamaterials [32]. Indeed, our simulations reveal that the circular disk superlattices presented here possess magnetic and Fano modes at visible frequencies. Manipulation of these coupled resonances is particularly promising for the development of advanced solution-phase metamaterials. Although examples of blueshifted modes have been reported for solution-phase gold nanorods [37], and copper sulfide disks [38], the optical consequences of symmetry breaking in these systems have been challenging to elucidate experimentally because of a lack of programmability in the forces governing particle assembly. Further symmetry breaking is possible via the assembly of nanoparticle building blocks with greater shape anisotropy. Triangular prism nanoparticles have a similar thickness to circular disks (~7 nm) but have lower in-plane rotational symmetry (3fold versus continuous) and, as a result, possess a partially bright quadrupolar resonance [33]. In addition to the dipolar modes observed for circular disks (see above), monoperiodic arrays of triangular prisms show a coupled π-type antibonding quadrupolar resonance. A tomographic TEM tilt series illustrates the 3D structure of the superlattice and reveals that the tips of the triangular prims are in registry. This is consistent with the hypothesis that the system will assemble in a manner that maximizes the number of interparticle DNA linkages [6, 18]. When triangular prisms are cocrystallized with circular disks, the additional shape-derived symmetry breaking results in the appearance of new hybrid π-type modes. As expected, tomographic TEM tilt data demonstrate that the alternating arrangement of particles results in a lack of triangular prism tip registry (Fig. 75.3a) [29]. The energy alignment of plasmon modes results in dipolar coupling to generate bonding (996 nm, 1πp/d) and antibonding (760 nm, 1π*p/d) modes but also results in a new quadrupole-like

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

mode (618 nm, 2π*p/d, Fig. 75.3b−d). This is surprising because quadrupolar modes on circular disks are only weakly excited with far-field light, as they are symmetry-forbidden. Simulations reveal that this resonance arises because the partially bright quadrupole mode on the triangular prisms can excite a quadrupole resonance on the neighboring circular disks as a result of near-field coupling between the two nanostructures (Fig. 75.3d).

Figure 75.3 Characterization of shape-biperiodic symmetry-broken nanoparticle superlattices. Arrays consist of alternating arrangements of circular disks and triangular prisms (a−d) or circular disks and rod dimers (d−h). (a, e) Images captured from a tomographic series and 3D modeled structures, each tilted by 15° increments out-of-plane illustrate the 3D features of the materials. (b, e) Extinction spectra show discrete particles in red, blue, or green, denoting triangular prisms, circular disks, and rods, respectively, while superlattice modes are shown in purple. Simulated data (dotted lines) are shown below experimental data (solid lines). Relevant modes are denoted by labels in (b, f) with corresponding diagrams (c, g) and simulations for the normalized local electric field distribution and dipole orientations (d, h). Labels denote the mode order (left superscript, dipole l = 1, quadrupole l = 2), bonding or antibonding character (right superscript), and participating particles (right subscript, p = triangular prism, d = circular disk, r = rod dimer). Note: field intensity color scale for the 1π*p/d mode has been decreased by a factor of 2 for clarity.

Triangular prisms are known to have another partially bright multipole, the l = 3 hexapolar mode [33, 39]. When triangular prisms

1379

1380

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

are cocrystallized with larger circular disks (94 nm), a similar set of coupled dipolar and near-field-induced quadrupolar resonances are observed, but with an additional hexapolar-like mode. In order for a circular disk to support multipolar modes, the circular symmetry of the disk has to be broken by the near-field of the neighboring triangular prism. Although it is not surprising that larger disks can more easily accommodate multipolar modes [40], it is important to understand how the symmetry of the coupled modes supported by the shape-biperiodic arrays can be tuned by the size of the components. The lowest-symmetry superlattice we constructed formed via the co-crystallization of circular disks and rods (Fig. 75.3e). Interestingly, although there is no explicit attractive interaction between rods, they adopt a parallel dimer motif that separates each circular disk in a 1D array; tomographic TEM tilt data shows that rod dimers lack orientational correlation along the unit lattice vector (Fig. 75.3e). The modes that result from this structure can be best thought of as arising from coupling between circular disks and rod dimers (Figs. 75.3f−75.3h). A bright antibonding mode on the rod dimer can π-type hybridize with the circular disk dipole in a bonding fashion to generate a redshifted mode (900 nm, 1πd/r), and in an antibonding fashion to generate a blueshifted mode (700 nm, 1π*d/r, Fig. 75.3f−h). Alternatively, when the rod dimer is dark (bonding), it does not participate in plasmon coupling, resulting in a mode in which the circular disks couple in an antibonding fashion, albeit with an increased effective lattice spacing (800 nm, 1π*d/d, Fig. 75.3f−h). In conclusion, we have demonstrated that fundamental metamaterial properties arise as a direct consequence of deterministic symmetry breaking enabled by DNA-programmable nanoparticle assembly. The ability to engineer colloidal particle interactions in a manner that does not dictate high-symmetry arrangements has been a longstanding goal in chemistry, physics, and materials science [2, 15, 41]. This accomplishment has particular relevance for the scalable synthesis of photonic and plasmonic metamaterials, as low-symmetry arrangements are often compulsory and traditional lithographic methods are low-yielding. Natural extensions of this work to program compositional or dynamically switchable asymmetry are expected to establish these structures as building blocks for future solution-phase metamaterials.

References

Acknowledgments This material is based upon work supported by the AFOSR GRANT12204315 and the National Science Foundation’s MRSEC program under award DMR-1121262. Portions of this work were carried out at the DuPont-Northwestern-Dow Collaborative Access Team (DND-CAT) beamline located at Sector 5 of the Advanced Photon Source (APS). DND-CAT is supported by E. I. DuPont de Nemours & Co., Dow Chemical Company, and the state of Illinois. The transmission electron microscopy work was carried out in the EPIC facility of NUANCE Center at Northwestern University, which has received support from the Soft and Hybrid Nanotechnology Experimental (SHyNE) Resource (NSF NNCI-1542205); the MRSEC program (NSF DMR-1121262) at the Materials Research Center; the International Institute for Nanotechnology (IIN); the Keck Foundation; and the State of Illinois, through the IIN. M.R.J. acknowledges the NSF for a graduate research fellowship and Northwestern University for a Ryan Fellowship. M.N.O. also acknowledges the NSF for a graduate research fellowship.

References

1. Huang, J.-S., Callegari, V., Geisler, P., Brüning, C., Kern, J., Prangsma, J. C., Wu, X., Feichtner, T., Ziegler, J., Weinmann, P., Kamp, M., Forchel, A., Biagioni, P., Sennhauser, U. and Hecht, B. (2010). Nat. Commun., 1, 150.

2. Gao, B., Arya, G. and Tao, A. R. (2012). Nat. Nanotechnol., 7, 433.

3. Moreau, A., Ciraci, C., Mock, J. J., Hill, R. T., Wang, Q., Wiley, B. J., Chilkoti, A. and Smith, D. R. (2012). Nature, 492, 86.

4. Fan, J. A., Wu, C., Bao, K., Bao, J., Bardhan, R., Halas, N. J., Manoharan, V. N., Nordlander, P., Shvets, G. and Capasso, F. (2010). Science, 328, 1135. 5. Sheikholeslami, S. N., Alaeian, H., Koh, A. L. and Dionne, J. A. (2013). Nano Lett., 13, 4137. 6. Macfarlane, R. J., Lee, B., Jones, M. R., Harris, N., Schatz, G. C. and Mirkin, C. A. (2011). Science, 334, 204.

7. Shevchenko, E. V., Talapin, D. V., Kotov, N. A., O’Brien, S. and Murray, C. B. (2006). Nature, 439, 55.

8. Boles, M. A. and Talapin, D. V. (2015). J. Am. Chem. Soc., 137, 4494.

1381

1382

Deterministic Symmetry Breaking of Plasmonic Nanostructures Enabled

9. Kalsin, A. M., Fialkowski, M., Paszewski, M., Smoukov, S. K., Bishop, K. J. M. and Grzybowski, B. A. (2006). Science, 312, 420.

10. Zhang, S., Genov, D. A., Wang, Y., Liu, M. and Zhang, X. (2008). Phys. Rev. Lett., 101, 047401. 11. Kanté, B., Park, Y.-S., O’Brien, K., Shuldman, D., Lanzillotti-Kimura, N. D., Jing Wong, Z., Yin, X. and Zhang, X. (2012). Nat. Commun., 3, 1180. 12. Soukoulis, C. M. and Wegener, M. (2011). Nat. Photonics, 5, 523.

13. Ross, M. B., Ku, J. C., Vaccarezza, V. M., Schatz, G. C. and Mirkin, C. A. (2015). Nat. Nanotechnol., 10, 453.

14. Ye, X., Chen, J., Diroll, B. T. and Murray, C. B. (2013). Nano Lett., 13, 1291.

15. Wang, Y., Wang, Y., Breed, D. R., Manoharan, V. N., Feng, L., Hollingsworth, A. D., Weck, M. and Pine, D. J. (2012). Nature, 491, 51.

16. Rogers, W. B. and Manoharan, V. N. (2015). Science, 347, 639.

17. Yan, W., Xu, L., Xu, C., Ma, W., Kuang, H., Wang, L. and Kotov, N. A. (2012). J. Am. Chem. Soc., 134, 15114. 18. Jones, M. R., Seeman, N. C. and Mirkin, C. A. (2015). Science, 347, 1260901.

19. Macfarlane, R. J., O’Brien, M. N., Petrosko, S. H. and Mirkin, C. A. (2013). Angew. Chem. Int. Ed., 52, 5688.

20. Kuzyk, A., Schreiber, R., Fan, Z., Pardatscher, G., Roller, E.-M., Hogele, A., Simmel, F. C., Govorov, A. O. and Liedl, T. (2012). Nature, 483, 311.

21. Kuzyk, A., Schreiber, R., Zhang, H., Govorov, A. O., Liedl, T. and Liu, N. (2014). Nat. Mater., 13, 862.

22. Mirkin, C. A., Letsinger, R. L., Mucic, R. C. and Storhoff, J. J. (1996). Nature, 382, 607.

23. Jones, M. R., Macfarlane, R. J., Lee, B., Zhang, J., Young, K. L., Senesi, A. J. and Mirkin, C. A. (2010). Nat. Mater., 9, 913. 24. Jones, M. R., Macfarlane, R. J., Prigodich, A. E., Patel, P. C. and Mirkin, C. A. (2011). J. Am. Chem. Soc., 133, 18865. 25. Damasceno, P. F., Engel, M. and Glotzer, S. C. (2012). Science, 337, 453.

26. Henzie, J., Grünwald, M., Widmer-Cooper, A., Geissler, P. L. and Yang, P. (2011). Nat. Mater., 11, 131.

27. Boneschanscher, M. P., Evers, W. H., Geuchies, J. J., Altantzis, T., Goris, B., Rabouw, F. T., van Rossum, S. A. P., van der Zant, H. S. J., Siebbeles, L. D. A., Van Tendeloo, G., Swart, I., Hilhorst, J., Petukhov, A. V., Bals, S. and Vanmaekelbergh, D. (2014). Science, 344, 1377.

References

28. O’Brien, M. N., Girard, M., Lin, H.-X., Millan, J. A., Olvera de la Cruz, M., Lee, B. and Mirkin, C. A. (2016). Proc. Natl. Acad. Sci. U.S.A., 113, 10485.

29. O’Brien, M. N., Jones, M. R., Lee, B. and Mirkin, C. A. (2015). Nat. Mater., 14, 833.

30. Kelly, K. L., Coronado, E., Zhao, L. L. and Schatz, G. C. (2003). J. Phys. Chem. B, 107, 668.

31. O’Brien, M. N., Jones, M. R., Kohlstedt, K. L., Schatz, G. C. and Mirkin, C. A. (2015). Nano Lett., 15, 1012.

32. Halas, N. J., Lal, S., Chang, W.-S., Link, S. and Nordlander, P. (2011). Chem. Rev., 111, 3913.

33. Shuford, K. L., Ratner, M. A. and Schatz, G. C. (2005). J. Chem. Phys., 123, 114713.

34. Paik, T. and Murray, C. B. (2013). Nano Lett., 13, 2952.

35. Funston, A. M., Novo, C., Davis, T. J. and Mulvaney, P. (2009). Nano Lett., 9, 1651.

36. Sheikholeslami, S., Jun, Y.-w., Jain, P. K. and Alivisatos, A. P. (2010). Nano Lett., 10, 2655.

37. Jain, P. K., Eustis, S. and El-Sayed, M. A. (2006). J. Phys. Chem. B, 110, 18243.

38. Hsu, S.-W., Ngo, C. and Tao, A. R. (2014). Nano Lett., 14, 2372.

39. Losquin, A., Zagonel, L. F., Myroshnychenko, V., Rodríguez-González, B., Tencé, M., Scarabelli, L., Förstner, J., Liz-Marzán, L. M., García de Abajo, F. J., Stéphan, O. and Kociak, M. (2015). Nano Lett., 15, 1229. 40. Schmidt, F. P., Ditlbacher, H., Hofer, F., Krenn, J. R. and Hohenester, U. (2014). Nano Lett., 14, 4810. 41. Chen, Q., Bae, S. C. and Granick, S. (2011). Nature, 469, 381.

1383

Chapter 76

Polarization-Dependent Optical Response in Anisotropic Nanoparticle− DNA Superlattices*

Lin Sun,a,b Haixin Lin,b,c Daniel J. Park,b,c Marc R. Bourgeois,c Michael B. Ross,b,c Jessie C. Ku,a,b George C. Schatz,b,c and Chad A. Mirkin,a,b,c aDepartment

of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, IL 60208, USA bInternational Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA cDepartment of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208, USA [email protected]; [email protected]

DNA-programmable assembly has been used to prepare superlattices composed of octahedral and spherical nanoparticles, respectively.

* Reprinted with permission from Sun, L., Lin, H., Park, D. J., Bourgeois, M. R., Ross, M. B., Ku, J. C., Schatz, G. C. and Mirkin, C. A. (2017). Polarization-dependent optical response in anisotropic nanoparticle−DNA superlattices, Nano Lett. 17, 2313−2318. . Further permissions related to the material excerpted should be directed to the ACS.

Spherical Nucleic Acids, Volume 3 Edited by Chad A. Mirkin Copyright © 2020 Jenny Stanford Publishing Pte. Ltd. ISBN 978-981-4877-23-7 (Hardcover), 978-1-003-05670-6 (eBook) www.jennystanford.com

1386

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

These superlattices have the same body-centered cubic lattice symmetry and macroscopic rhombic dodecahedron crystal habit but tunable lattice parameters by virtue of the DNA length, allowing one to study and determine the effect of nanoscale structure and lattice parameter on the light–matter interactions in the superlattices. Backscattering measurements and finite-difference time-domain simulations have been used to characterize these two classes of superlattices. Superlattices composed of octahedral nanoparticles exhibit polarization-dependent backscattering but via a trend that is opposite to that observed in the polarization dependence for analogous superlattices composed of spherical nanoparticles. Electrodynamic simulations show that this polarization dependence is mainly due to the anisotropy of the nanoparticles and is observed only if the octahedral nanoparticles are well-aligned within the superlattices. Both plasmonic and photonic modes are identified in such structures, both of which can be tuned by controlling the size and shape of the nanoparticle building blocks, the lattice parameters, and the overall size of the 3D superlattices (without changing habit). Gold nanoparticles (NPs) are widely studied optical building blocks due to their strong interactions with visible light, which is confined into small volumes close to the NP surface due to localized surface plasmon resonance (LSPR) excitation [1−6]. The LSPR is sensitive to NP size, shape, composition, dielectric environment, and proximity to other plasmonic NPs [7−10]. Ordered arrays and other structures of plasmonic NPs exhibit a variety of interesting properties, such as the ability to guide light around sharp corners [11], a broadband optical response [12], Fano resonances [13], and a negative index, which is important for the development of metamaterials [14]. Moreover, the optical properties of such ordered plasmonic structures can be tuned by changing the distance between individual NPs [7−9, 15, 16]. Therefore, great effort has been devoted to research on making 2D and 3D periodic structures of plasmonic NPs using top down [11, 12, 14] and bottom up [10, 17−19] techniques. DNA-programmable assembly has emerged as a robust and flexible tool for synthesizing superlattices with control over NP size and shape, lattice structure, and crystal habit. In these

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

structures, nanoparticles with different shapes [20, 21], sizes [22], and compositions [21, 23, 24] can be assembled and also lattice symmetry and nanoparticle spacing can be tuned [22, 25], giving rise to robust [26] and compositionally tunable thin-film optical modes [24]. In addition, the micron length scales associated with well-formed superlattices lead to optical cavity modes, such as FP resonances [27] that arise due to interference of light traveling between the parallel top and bottom facets of the superlattice [27, 28], and shape-dependent scattering [29] that is dictated by the crystal habit (i.e., the size and shape of the micro- or macroscopic superlattice). In principle, one can uniquely use DNA-programmable methods to assemble different NP building blocks into macroscopic superlattices where the superlattices have the same crystal symmetry and macroscopic crystal habit but that comprise different NP shapes. This type of comparison enables separation of the effect of NP shape from the effects of lattice symmetry and microscale faceting. Herein, we show that octahedral and spherical NPs can be assembled into body-centered cubic (bcc) superlattices with identical rhombic dodecahedra crystal habits and similar lattice parameters but different optical properties (Fig. 76.1). These superlattices have been characterized by backscattering measurements and finite-difference time-domain (FDTD) simulations. In addition to focusing on the importance of the shape of the NPs, we can also use simulations and, in certain cases, experiment to independently assess the importance of lattice parameter and size of the microscale superlattices on optical response. Therefore, this DNAprogrammable technique allows one to separate contributions of the nanoscale building blocks from the microscale architecture to the optical properties of the superlattice. Importantly, superlattices composed of octahedral NPs exhibit polarization-dependent backscattering spectra at all volume fractions studied (5.1−20.3%), while the opposite polarization dependence behavior is seen in superlattices composed of spherical NPs with volume fractions larger than 10%. Finally, it was found that the orientation and alignment of the octahedral NPs inside each superlattice is crucial for observing such polarization-dependent behavior.

1387

1388

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

Figure 76.1 Schematic depiction of the structure of superlattices made from either spherical (orange) or octahedral (green) NPs. The shape of the NPs gives rise to different polarization dependence of their LSPRs. Both superlattices have bcc lattice structures and the exposed facets are the closest packed (110) planes for the rhombic dodecahedral crystal habit. Single nanoparticle extinction spectra were obtained from FDTD calculations with the refractive index of the dielectric environment set as 1.45. Note that different wavelength ranges are used in the spectra.

We begin by using optical simulations to probe different types of resonances in the superlattices. Previous work done by our group has shown that both plasmonic modes and FP type photonic modes exist and interact in superlattices composed of spherical NPs (Fig. 76.2A) [27]. These two types of modes couple strongly to each other, leading to band gap behavior in some cases [27, 30]. Here, FDTD simulations with an infinite slab model were used to approximate the optical properties of the superlattices. This model matches well with the experimental setup. Two polarizations, 0° and 90° (perpendicular and horizontal to the long axis of the rhombic dodecahedron, respectively), are used to characterize the light polarization (Fig. 76.2A).

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

Figure 76.2 Optical modes of the superlattice consist of both plasmonic and FP modes. The roles of each mode are shown in the backscattering spectra. (A) Schematic depiction of the superlattice where both plasmonic and FP modes exist in comparison to a thin slab with a thickness of only 1 u.c., which is too thin to support FP modes (center). Structures of a single unit cell of superlattices made from spherical (left) and octahedral (right) NPs along the (110) direction are shown with the 0° and 90° indicating the corresponding polarizations of the incident light. (B) FDTD simulation of a spherical superlattice with the polarization along 0° (top, solid line) and 90° (bottom, solid line) as defined in (A). The gray dashed line indicates the maximum wavelength of the main peaks, which is the same for the two polarizations. The blue and red dashed lines are the spectra of the 1 u.c. thick slab, which shows that the location of the main peak is determined by the plasmonic modes. The dips at longer wavelength, as indicated by the yellow arrows, are FP modes. (C) The same simulations were performed for superlattices made from octahedral NPs, where there is a difference (Δλ ≈ 10 nm) between the main peaks of the two polarizations.

Importantly, the LSPRs of the octahedral NPs along 0° and 90° are different due to anisotropy of the NP shape (shape anisotropy, Fig. 76.1). In addition, the top and bottom facets of the superlattice adopt the (110) plane of the bcc lattice structure [31], resulting in a smaller interparticle spacing between NPs in 90° compared to 0°. The anisotropy in the lattice structure (structure anisotropy)

1389

1390

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

would result in different interparticle interactions that depend on polarization, with sufficient volume fractions, even for superlattices composed of spherical NPs, where the NP shape anisotropy is absent. The solid lines in Fig. 76.2B,C show the simulated backscattering spectra of superlattices made from spherical and octahedral NPs, respectively. The average diameter of the spherical NPs is 40 ± 3.2 nm, while the edge length of octahedral NPs is 51 ± 2 nm. Tips of the octahedral NPs are rounded with an average radius of curvature of 4.8 ± 0.5 nm, a value determined from scanning electron microscopy (SEM) images of the as synthesized NPs used in the experiments. These NP structural parameters are kept constant throughout this work. The interparticle spacing, defined as the NP surface-to-surface distance along the 90° direction, is 40 nm, and the thickness of the slab is set to be 1.5 μm. These parameters are set such that they fall within a range that can be realized experimentally [22, 32−35]. The simulated spectra exhibit a convolution of plasmonic and FP modes, however, the influence of nano- and microscale structure can be disentangled (Fig. 76.2). FP modes are purely photonic modes that are determined by superlattice thickness and effective refractive index within the cavity. Therefore, we eliminate FP modes by simulating a 1 unit cell (u.c.) thick superlattice, where the superlattice is optically too thin to support any FP modes; such a structure should only exhibit the optical response of the plasmonic nanoparticles [29]. Indeed, only one peak is observed for the 1 u.c film (dashed lines, Fig. 76.2B,C). Moreover, this peak is spectrally close to the main peak found in the thicker samples (1.5 μm). Therefore, the main peak in these figures is primarily determined by the plasmonic properties of a single layer of NPs, while the dips at longer wavelengths are from FP modes (orange arrows, Fig. 76.2B,C). From these simulations, we find that for the superlattices with spherical NPs there is negligible difference between the main peaks, namely the backscattering maximum around 560 nm, along the 0° and 90° polarizations. However, for the superlattices with octahedral NPs the main peak at around 610 nm red shifts by Δλ ≈ 10 nm when the polarization is aligned along 90° compared to 0°. This difference suggests the crucial role that shape anisotropy plays. Notably, Maxwell-Garnett effective medium theory (EMT), which has been used to quantitatively explain a variety of optical properties in

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

spherical nanoparticle based superlattices [24, 27, 36, 37] can no longer be applied to the superlattice made from octahedral NPs. To experimentally demonstrate the different responses to the two polarizations in the two types of superlattices, spherical and octahedral NPs were assembled into bcc lattices with a rhombic dodecahedral crystal habit and ~2 μm parallel face-to-face dimensions (Fig. 76.1). In particular, the directional DNA interaction will align octahedral NPs in a face-to-face manner [20], as can be seen in the schematic drawing of a single unit cell superlattice in the (100) orientation. After the assembly, the superlattices undergo a silica embedding process in order to preserve their structure in the solid-state [38], enabling optical measurements and electron microscopy imaging. Small angle X-ray scattering (SAXS) was used to extract information on the crystal structure and lattice constant (in the solution- and in the solid-state), as well as to ensure their high-quality crystallinity. Backscattering measurements were performed on these superlattices with a microscope-coupled spectrometer. Figure 76.3A,C shows SEM images of typical superlattices made from spherical and octahedral NPs, respectively, where the lattice constants, as measured by SAXS, are 69 and 102 nm, respectively. As such, the interparticle spacing is 29 nm for the spherical NPs and 37 nm for the octahedral NPs. The good agreement between simulation (top) and experiment (bottom) for both superlattices, Fig. 76.3B,D, validates both the simulation model and the high quality and fidelity of the DNA-programmed NP superlattices. Polarization dependence is primarily observed for the main peak as dictated by nanoscale effects, such as NP shape, which controls the LSPR, and structure anisotropy of the lattice, which affects particle−particle coupling. At wavelengths far from the main peak, both simulation and experiment show a much smaller difference between the FP modes at the two polarizations. To explore the effect of both shape and structure anisotropy, a set of superlattices where the interparticle spacing is varied from 20 to 70 nm and the superlattice thickness is held constant at 1.5 μm were simulated. Figure 76.4A summarizes the difference between the 90 0 main peak along the 90° and 0° directions, namely, Δλ = lmax - lmax . A spectral representation can be seen in Fig. 76.4B, where the

1391

1392

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

backscattering spectra of superlattices made from spherical (top) and octahedral (bottom) NPs with 20 nm interparticle spacing are plotted. In contrast with the polarization-dependent red shift as seen in the superlattices with octahedral NPs, the main peak blue shifts when the polarization is changed from 0° to 90° in the superlattices with spherical NPs, resulting in negative values of Δλ. In fact, as the interparticle spacing decreases, λmax along both 0° and 90° are red-shifted, though the shift is more significant along 0° for superlattices made of spherical NPs. This red shift with decreasing NP separation is routinely observed and attributed to enhanced interparticle coupling [8, 39, 40]. However, the opposite trend (blueshifting) is observed for the polarization dependence, by which we mean that although the NPs are more closely spaced (smaller 90 interparticle spacing) along the 90° direction, lmax is to the blue of 0 lmax for spherical NP superlattices.

Figure 76.3 Experimental confirmation of the absence and presence of polarization-dependence in superlattices composed of spherical and octahedral NPs, respectively. (A, B) SEM image of a superlattice consisting of spherical NPs and its backscattering spectra obtained from simulation (top) and experiment (bottom). (C, D) The same set of data for a superlattice consisting of octahedral NPs. Inset in (C) is a magnified area of the SEM image. The interparticle spacings are 29 and 37 nm for superlattices made of spherical and octahedral NPs, respectively. Scale bars are 1 μm for (A,C), and 200 nm for inset in (C).

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

Figure 76.4 Effect of structural and NP shape anisotropy and NP alignment on polarization dependence. While keeping the NP size and shape the same, simulations show the effect of changing the interparticle spacing from 20 to 70 nm for spherical and aligned octahedral NPs. Differences between maxima of the main peaks at 90° and 0° polarization (Δλ) for all superlattices are summarized in (A). (B) Simulated spectra of spherical and aligned octahedral NPs with 20 nm interparticle spacing. Interestingly, for the superlattices consisting of spherical NPs, the peaks blue shift at 90° compared to 0°, resulting in negative Δλ values as marked by the −Δλ. The opposite happens for the superlattices consisting of octahedral NPs. (C) In order to show the importance of alignment of the NPs, a set of simulations was set up with the same set of parameters but for randomly oriented (top) and well-aligned (bottom) NPs. Interparticle spacing here is 70 nm. (D) The backscattering spectra from the setup in (C). The polarization dependence is lost in the former case.

Evidently, in this case long-range coupling between many NPs plays an important role, necessitating explicit consideration of the dipole lattice sum. Under the assumption that each NP can be considered as a point dipole, this sum describes the near- and farfield coupling between NPs and is solely dependent on the geometric parameters of the lattice (i.e., the arrangement of the NPs) [41, 42]. Applying this method to calculate the extinction spectra for a (110) bcc lattice plane along 0° and 90° shows that the blue shift seen in superlattices made of spherical NPs is due to the difference in the lattice sums for the two polarization directions [8, 9, 43]. As the

1393

1394

Polarization-Dependent Optical Response in Anisotropic Nanoparticle−DNA Superlattices

interparticle spacing increases (>60 nm), negligible polarization dependence is observed (Fig. 76.4A), and the dipole lattice sums and extinction spectra show that this is due to the smaller difference in the lattice sums along the different polarization directions for this 0 90 spacing. Likewise, the larger red shifts of lmax compared to lmax as the interparticle spacing decreases can be attributed to a larger difference in the lattice sums along the two polarizations. 90 For superlattices made of octahedral NPs, lmax is to the red 0 of lmax . Although the dipole lattice sum is the same for the same NP arrangement, compared to the case of superlattices made of spherical NPs, octahedral NPs have different effective polarizabilities (α) along the different polarization axes. Larger values of α along the 90° (tip-to-tip) will result in a reduced (1/α), which would then intersect the dipole lattice sum at larger wavelength, which red shifts the resonance position (λmax). In other words, a stronger tipto-tip coupling along 90° contributes to a more significant red shift compared to edge-to-edge coupling along 0° as the interparticle spacing become small. As the interparticle spacing decreases, large plasmonic coupling plays an especially important role in the large 90 red shift observed in lmax . As a result, the effect of structure anisotropy (i.e., different dipole lattice sums) on the polarization dependence becomes significant for both superlattices made of spherical and anisotropic NPs, respectively, at close distances (Fig. 76.4A) with Δλ becoming nonzero even for spherical NPs. The same trends are also demonstrated experimentally (see details in SI). From these data, we conclude that structural anisotropy is important only in closepacked superlattices (interparticle spacing