Advances in Solar Energy: Volume 17 : An Annual Review of Research and Development in Renewable Energy Technologies 9781315793221, 9781844073146

Citation preview

Advances in Solar Energy An Annual Review of Research and Development Volume 17

Editor-in-Chief D. Yogi Goswami Clean Energy Research Center, University of South Florida, Tampa, FL

Associate Editors Viacheslav M. Andreev Ioff Physico-Technical Institute, St Petersburg, Russia

J. Douglas Balcomb National Renewable Energy Laboratory, Golden, CO

Adolf Goetzberger Fraunhofer Institut Solare Energiesysteme, Freiburg, Germany

Yoshihiro Hamakawa Ritsumeikan University, Kusatsu Shiga, Japan

Lu Weide China Rural Energy Industry Association, Beijing, China

Antonio Luque Universidad Politécnica de Madrid, Madrid, Spain

Gerald R. Nix National Renewable Energy Laboratory, Golden, CO

Morton Prince Melrose Park, PA

Aldo Steinfeld Swiss Federal Institute of Technology, Zurich, Switzerland

Gunnar Svedberg Mid Sweden University, Sundsvall, Sweden

Steven Szokolay University of Queensland, Kenmore, Australia

Lorin Vant-Hull University of Houston, Houston, TX

Nejat Veziroglu University of Miami, Coral Gables, FL

Carl Weinberg Walnut Creek, CA

Roland Winston University of Chicago, Chicago, IL

Emeritus Editor-in-Chief Karl W. Böer University of Delaware, Newark, DE

Advances in Solar Energy An Annual Review of Research and Development Volume 17

Edited by D. Yogi Goswami

American Solar Energy Society, Inc., Boulder, CO

First published by Earthscan in the UK and USA in 2007

For a full list of publications please contact: Earthscan 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN 711 Third Avenue, New York, NY 10017 Earthscan is an imprint of the Taylor & Francis Group, an informa business Copyright © American Solar Energy Society, 2007 Published by Taylor & Francis.



American Solar Energy Society, Inc. 2400 Central Avenue, Suite A Boulder, CO USA All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Notices Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.

ISBN

978-1-84407-314-6 (hbk)

Typeset by Domex e-Data, India Cover design by Yvonne Booth A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data has been applied for

FOREWORD Global climate change continues to be a hot topic on the world stage. However, we have now moved away from the debate of whether global warming exists to what strategies are needed to combat the challenge. Another ongoing debate concerns the peaking and eventual decline of world oil production and how this will affect other conventional energy resources including coal, natural gas and uranium. It is becoming increasingly clear that whether we are concerned about global climate change or the peaking and eventual decline of conventional resources, we must ramp up the use of renewable energy resources worldwide. Indeed, Schindler and Zittel argue in their chapter in this volume of Advances in Solar Energy that by 2050 as much as 50 per cent of our energy will have to come from renewable energy. In our continuing series of country- and region-specific articles we focus on the Middle East and North Africa (MENA) region – Alnaser, Treib and Knies show that by 2050 as much as 50 per cent of the electrical power for the MENA countries may come from concentrating solar power (CSP). At present, commercial deployment of renewable energy technologies is increasing at rates of 20–40 per cent per year. However, in order to increase the deployment of renewable energy technologies to the levels foreseen by Schindler and Zittel, we need continued research and development to raise the efficiency and reduce the costs of these technologies. The chapters in this volume inform us on progress in this direction, for example advances in quantum dot solar cells and organic solar cells. Other topics in this volume include such applications of solar energy as solar-hydrogen production, solar photocatalytic detoxification of water, solar drying and solar industrial process heat. In the first chapter, Schindler and Zittel question the assumptions made by the International Energy Agency (IEA) in forecasting the future availability of energy resources, and present an alternative world energy outlook and a possible path towards a sustainable future. According to the authors the world will experience the peaking of oil first, followed by natural gas. They show that the resulting energy supply gap cannot be filled by a rising share of nuclear energy. This gap could be filled partially by coal; however, without CO2 sequestration this would lead to emissions at unacceptable levels. Their analysis shows that a transition in our energy resource use to renewable energy will occur. However, they question whether there is enough time left for the transition to occur smoothly. They recommend that the earlier we start with the transition, the better we will be able to manage it. Chapters 2 and 3 deal with the new developments in photovoltaic cells. Chapter 2 deals with quantum well solar cells and the progress made in this area over the last 15 years. The author, Ekins-Daukes, writes, ‘The ability to adjust the absorption profile of a bulk semiconductor through the inclusion of quantum wells gives rise to a number of practical, near-term applications for quantum solar cells.’ Some of these applications include multi-junction solar cells and thermophotovoltaics. The author concludes that even though to date only the band gap has been engineered with the application of quantum wells, it is likely that in the future it will be possible to control other properties, such as optical transitions and phonon modes in order to produce highly efficient devices. Chapter 3 describes the recent progress in organic photovoltaic materials and devices. Organic and polymeric materials have the potential to reduce the cost of solar cells

vi

ADVANCES IN SOLAR ENERGY

making them cost competitive. At present the conversion efficiencies of organic solar cells are relatively small (< 6 per cent), which can be attributed to a number of reasons, such as, photon, exciton and carrier losses, poor material morphologies, and unoptimized structures. However, there is plenty of room for improvement. The paper outlines the approaches being followed in improving the performances. Chapter 4 describes the progress in thermal and material characterization of immersed heat exchangers used in solar domestic water heating systems. The authors have made recommendations based on published literature and their own work over the last 20 years. As manufacturers consider the use of polymers for heat exchangers, their durability and reliability must also be considered in addition to thermal performance. The authors also point out areas for additional research. Chapter 5 deals with solar photocatalytic detoxification, in which there has been tremendous research and development over the past ten years, although it is a relatively recent environmental application of solar radiation. In this paper, the authors review the scientific progress and applications of solar photocatalysis. Even though the process is very attractive in remediating water pollution, the quantum efficiency is very low. The authors describe the research and development (R&D) approaches to increase the quantum yield, such as modifying the catalyst structure and composition and finding new catalysts with band gaps that better overlap the solar spectrum. The authors also describe the progress in using photo-Fenton treatment, and combined photocatalytic-biological treatment. At present there is considerable interest in the potential use of hydrogen for transportation, to replace the eventually depleting supplies of oil. However, for hydrogen to provide a viable solution, it must be produced economically and from a clean source, such as solar energy. Solar photo-electrochemical generation of hydrogen is one such method that has a viable potential. However, this method needs some fundamental R&D. In Chapter 6, the authors systemically describe the solid-state chemistry-related issues that must be considered in order to improve the viability of this method. The authors explain the parameters such as band gap and Fermi level and the concepts of modifying them. Chapter 7 reviews the recent progress in using solar heat for industrial processes, with specific examples from Europe. The authors review the technology and its potential for use in low (< 60ϒC), medium (60–150ϒC) and medium-high (150–250ϒC) temperature industrial applications. They present case studies for many industries in Spain and Portugal, arguing that as much as 7 per cent of total final energy demand in southern European countries may be fulfilled by providing solar process heat at temperatures below 250ϒC. The authors state that the present costs of solar thermal systems range from ⇔250 to ⇔1000/kW of thermal power, leading to average thermal energy costs in southern Europe of 2–5¢/kWh for low temperature applications and 5–15¢/kWh for medium temperature applications. However, according to the authors, these costs can be reduced by 50 per cent by 2010 by mass production, reducing operation and maintenance expenses and improving collector efficiency and collector design. The final paper (Chapter 9) of this volume is focused on the Middle East and North Africa (MENA) as a part of our series of country-specific articles. This article, however,

FOREWORD

vii

covers many countries with a common potential in using concentrating sunlight to produce electrical power. The MENA region has the potential to provide not only for its own electrical needs from solar energy but also for those of parts of Europe. The authors show that with a growth rate of 25–35 per cent concentrating solar thermal power (CSP) can be used to fulfil as much as 14 per cent of the electricity demand of MENA by 2025 and 57 per cent by 2050. According to the authors, CSP may also be able to provide a majority of the acute water needs of the region via desalination. The study also shows that the total carbon emissions of electricity generation of all MENA countries can be reduced from the present 770 million tons per year to 475 million tons per year by 2050, instead of increasing to 2000 million tons. I would like to thank all the authors for sharing their knowledge and expertise with the rest of the world via their papers in this volume of Advances in Solar Energy. I would also like to thank my editorial assistant Ms Barbara Graham for her tireless work in carefully editing many articles. Finally, I would like to thank the American Solar Energy Society, its President Ronald Larson and Executive Director Brad Collins, and the Board of Directors for their commitment in continuing to publish Advances in Solar Energy for the benefit of researchers and practitioners of solar energy around the world.

D. Yogi Goswami Editor-in-Chief

This page intentionally left blank

Contents Foreword List of figures and tables About the authors 1

Alternative World Energy Outlook 2006: A Possible Path towards a Sustainable Future

v xii xxi 1

Jörg Schindler and Werner Zittel 1.1 Winds of Change: The Transition Period 1.2. Future Availability of Fossil and Nuclear Energy Sources 1.3 Alternative World Energy Scenarios Notes and References Annex I: Renewable Energy Potentials Annex II: Simulation Parameters for Renewable Energy Scenarios Annex III: Growing Share of Renewable Energy since 1990

1 7 28 38 41 42 43

2

45

Quantum Well Solar Cells

N. J. Ekins-Daukes 2.1 History 2.2 Quantum Well Electronic Structure 2.3 Basic Operation of the P-I-N Quantum Well Solar Cell 2.4 Near-term Applications for Quantum Well Solar Cells 2.5 Efficiency Limits 2.6 Conclusion Acknowledgements References

45 46 49 61 64 66 66 67

3

74

Recent Progress of Organic Photovoltaics

Sam-Shajing Sun 3.1 Introduction 3.2 Organic versus Inorganic Semiconductors 3.3 Organic/Polymeric Solar Cell Developments 3.4 Organic Solar Cell Fabrications 3.5 Organic Solar Cell Optimizations 3.6 Conclusions and Future Perspectives Acknowledgements References

4

Thermal and Material Characterization of Immersed Heat Exchangers for Solar Domestic Hot Water

74 75 77 84 87 94 95 95

99

Jane H. Davidson, Susan C. Mantell and Lorraine F. Francis 4.1 Introduction 4.2 Thermal Characterization and Design

99 102

x

ADVANCES IN SOLAR ENERGY

4.3 Mechanical Characterization of Polymers 4.4 Scaling of Candidate Polymers 4.5 Conclusion Acknowledgements References

5

Photocatalytic Detoxification of Water with Solar Energy

112 118 124 125 125

130

Sixto Malato, Julián Blanco, Diego C. Alarcón, Manuel I. Maldonado, Pilar Fernández-Ibáñez and Wolfgang Gernjak 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8

Introduction Solar Collectors for Photochemistry Fundamental Parameters in Solar Photocatalysis Factors Affecting Solar Photocatalysis Solar UV Photocatalytic Degradation of Contaminants Evaluation of Solar UV Radiation Installed Solar Photocatalytic Treatment Plants Photocatalytic Detoxification of Water with Solar Energy: Outlook for the Future Acknowledgements References

6

Solar-Hydrogen: A Solid-State Chemistry Perspective

130 132 138 147 149 151 154 160 163 163

169

J. Nowotny, T. Bak, L. R. Sheppard and C. C. Sorrell 6.1 Introduction 6.2 Solar-Hydrogen 6.3 The Concept of Solar-Hydrogen Generation 6.4 Materials Property Requirements for Photo-Electrodes 6.5 Electronic Structure 6.6 Why Titania? 6.7 Reduced-Band-Gap Titania 6.8 Impact of Defect Chemistry on the Properties of Titania 6.9 Collective and Local Factor 6.10 Spin-off Applications of Titania 6.11 Multiphase Photo-Sensitive Systems 6.12 Solar Cell Equipped with Space-Based Solar Energy Collector 6.13 Solar-Oxygen 6.14 Economic and Environmental Considerations of Solar-Hydrogen 6.16 Solar-Methanol 6.17 Conclusions Notes Acknowledgements References

170 174 175 178 180 190 193 194 203 204 204 205 207 207 209 210 211 211 211

CONTENTS

7

Solar Heat for Industrial Processes

xi

216

H. Schweiger, J. Farinha Mendes, Ma. J. Carvalho, K. Hennecke, D. Krüger 7.1 Introduction 7.2 Application Potential 7.3 Available Solar Collector Technology 7.4 Guidelines for Evaluation and System Design 7.5 Case Studies 7.6 Conclusions References

8

Solar Energy Technology in the Middle East and North Africa (MENA) for Sustainable Energy, Water and Environment

216 217 220 238 247 257 258

261

W. E. Alnaser, F. Trieb and G. Knies 8.1 Introduction 8.2 Determination of the Electricity Demand in MENA 8.3 Renewable Energy Resources in EU-MENA 8.4 CSP Demand-Side Potential in MENA 8.5 The Critical Issue of Water and Energy in the GCC Countries 8.6 Water Resources and Water Demand in MENA 8.7 The Potential for Desalination by Concentrating Solar Power 8.8 Environmental Impact of Using Solar Technology in MENA Acknowledgements References

262 264 265 274 282 284 291 295 302 302

List of Figures and Tables FIGURES 1.1

World energy demand 1971–2030 as given in the IEA’s World Energy Outlook 2004 and 2005 1.2 The basic dilemma – Business as usual or climate policy 1.3 Oil production of countries outside OPEC and the FSU 1.4 Canadian oil production 1960–2030 1.5 Cumulative worldwide gas discoveries and production 1.6 Annual gas production 1920–2004 and extrapolation based on a bell-shaped profile and an estimated ultimate recovery of 12,000Tcf 1.7 Gas supply in the US 1.8 UK gas production 1970–2030 1.9 Natural gas supply of Europe – Probable development until 2020 according to scenario calculations by LBST 1.10 Gas production forecast for Russia 1.11 Worldwide gas production according to LBST scenario calculations 1.12 Installed capacity of nuclear power plants and various forecasts 1.13 World uranium reserves and cumulative uranium consumption 1.14 Possible world coal production profiles based on 180 years and 370 years of remaining reserves 1.15 High fossil scenarios of future production of fossil and nuclear fuels 1.16 Low fossil scenario of future production of fossil and nuclear fuels 1.17 Carbon dioxide emissions according to the production profiles as outlined in the high fossil scenarios in Figure 1.15 1.18 Principle of the logistic growth function and the meaning of its parameters, P, T0 and b 1.19 Technical potentials of electricity production from renewable energies 1.20 Possible market penetration of renewable energy sources 1.21 High fossil alternative world energy outlook until 2100 1.22 Low fossil alternative world energy outlook until 2100 1.23 Comparison between fossil fuel costs and renewable installation costs 1.A.1 Cumulative installed capacity of wind energy converters, geothermal power plants, photovoltaic modules and solar collectors

5 7 10 11 15

16 17 19 20 21 22 23 24 25 26 27

28 29 30 32 34 35 36

43

LIST OF FIGURES AND TABLES

1.A.2 Energy production from renewable sources 1.A.3 Primary energy from renewable sources 2.1 a) Layer structure showing three QWs surrounded by barrier material; b) Band diagram for three QWs, showing two confined states for electrons and holes; c) Energy vs in-plane momentum dispersion relation for a QW 2.2 Density of states for a 3D bulk semiconductor g3D and for a 2D quantum well structure g2D 2.3 a) QW p-i-n layer structure; b) Band diagram for the QW p-i-n solar cell showing the photogeneration and recombination processes, together with the carrier capture and escape routes 2.4 Typical quantum efficiency for a GaAsP0.06/In0.1GaAs 75Å, 35 MQW structure at various biases 2.5 Quantum efficiency for an Al0.04GaAsP0.06/In0.1GaAs 75Å, 35 MQW showing how background impurities lead to a collapse of the QE at forward bias 2.6 Light IV curve for a GaAsP0.06/In0.1GaAs 75Å, 35 MQW structure under approximately 1 sun AM0 equivalent 3000K tungsten halogen illumination 2.7 Plot of J0 vs absorption edge for a variety of lattice matched, mismatched and MQW devices 2.8 Strained GaAs/InGaAs 3period MQW device: a) Schematic diagram showing the layer configuration; the relaxed lattice parameter is indicated laterally; b) Electronic band-structure for the device 2.9 a) External quantum efficiency for a strained GaAs/InGaAs MQW sample and GaAs control; b) Dark IV curves for GaAs/InGaAs strain-balance MQW cell, a relaxed GaAs/InGaAs MQW cell, a strained GaAs/InGaAs MQW cell and a GaAs control 2.10 Strain-balance GaAsP/InGaAs 3period MQW device: a) Schematic diagram showing the layer configuration; b) Electronic band structure for the device 2.11 a) External quantum efficiency for the strain-balanced GaAsP/InGaAs MQW sample and GaAs control; b) Dark IV curves for GaAsP/InGaAs strain-balance MQW cell, a relaxed GaAs/InGaAs MQW cell, a strained GaAs/InGaAs MQW cell and a GaAs control 2.12 AM0 iso-efficiency contour plot for a monolithic tandem structure 3.1 Lightweight and flexible thin film ‘plastic solar cells’ are very attractive for camping tents or any mobile units 3.2 Scheme of exciton Coulombic binding energy (in log scale) versus exciton radius

xiii

43 44

47 48

50 51

52

53 54

56

57

58

59 62 75 76

xiv

3.3

3.4

3.5 3.6 3.7 3.8

3.9

3.10 3.11 3.12 3.13 3.14 3.15

3.16

4.1

4.2 4.3 4.4

ADVANCES IN SOLAR ENERGY

First generation single layer type organic photovoltaic cell or ‘Schottky cell’ with: a) device structure; and b) energy diagram Second generation donor/acceptor bilayer type organic photovoltaic cell or ‘Tang cell’ with: a) device structure; and b) energy diagram Representative organic/polymeric electron donors (p-type semiconductors) Representative organic/polymeric electron acceptors (n-type semiconductors) Frontier orbital levels of representative polymeric electron donors and acceptors Third generation donor/acceptor blend type organic photovoltaic cells or ‘bulk heterojunction cells’ with: a) device structure; and b) energy diagram Fourth generation donor/acceptor bicontinuous phase separated and nano-structured organic photovoltaic cell with: a) device structure; and b) energy diagram General scheme of an organic/polymer solar cell with ‘buffering’ layers Chemical structure of PSS-PEDOT Scheme of a –DBAB- type of block copolymer ‘primary structure’ Scheme of a potential –DBAB- type of block copolymer ‘secondary structure’ Scheme of a potential –DBAB- type of block copolymer ‘tertiary structure’ Scheme of molecular frontier orbitals and photo-induced electron transfer as well as Dexter energy transfer processes in a donor/acceptor binary light harvesting system Donor RO-PPV exciton quenching parameter, charge recombination rate constant, and charge recombination quenching parameter versus LUMO offset of RO-PPV/SF-PPV-I pair Conceptual drawing of an indirect integral solar collector storage (ICS) system with a load-side immersed heat exchanger for solar domestic water heating Conceptual sketch of an immersed heat exchanger in a vertical thermal storage tank Three dimensional streamlines (top) and isotherms (bottom) for a rectangular storage with a single immersed tube Sketch of a divided indirect thermal storage vessel with two storage compartments

78

79 79 80 80

82

83 85 85 89 89 90

92

93

101 102 106 109

LIST OF FIGURES AND TABLES

4.5

4.6

4.7 4.8 4.9 4.10

4.11

5.1 5.2 5.3 5.4 5.5 5.6 5.7

5.8

5.9 5.10 5.11

5.12

Predicted ratio of delivered energy of a divided storage and an undivided storage as a function of the dimensionless output volume and NTU Measured ratio of delivered energy of a divided and undivided storage as a function of ratio of the volume of delivered hot water to the volume of storage fluid for nominal heat exchanger NTU of 2.5 and 7 Creep compliance of PA66 in air and exposed to hot chlorinated water Tensile strength (top) and strain (bottom) of PA66 before and after exposure to hot chlorinated water SEM images of PA66: a) native unexposed surface; and b) after 292 hours exposure at 830 ± 40mV, 59 ± 2ϒC SEM micrographs of scale formed on: a) polyamide 6,6; b) polypropylene; c) polybutylene; and d) copper tubes in a tube-in-shell heat exchanger SEM micrographs of scale formed on PEX tubes after: a) 5 hours; and b) 7.5 hours of exposure to flowing hard water at room temperature The various loss factors (d) affecting the photon flux (I) inside a PTC photo-reactor Structural schematic view of a DSSR reactor showing water flowing in the transparent box Schematic drawing of CPCs Graphics related to the adjustment of data to an L-H type kinetic model Pesticide decomposition at different initial concentrations Application of the proposed kinetic model for mineralization of a pesticide mixture Average UV direct and UV global (direct + diffuse) irradiance for each month of the year at Plataforma Solar de Almería, Spain Overall dichloroacetic acid degradation rate (as TOC disappearance reaction rate) comparison for a concentrating (PTC) and a non-concentrating (CPC) collector system Plots of pesticide concentration as a function of experiment time (top) and accumulated energy (bottom) Schematic of the pilot plant batch operation Schematic diagram of solar detoxification demonstration plant constructed in SOLARDETOX project at HIDROCEN, Madrid Schematic diagram of the container washing and photocatalytic water treatment plant concept

xv

110

111 115 116 117

121

122 133 135 138 139 140 142

143

143 153 154

157 159

xvi

5.13 5.14 6.1

6.2

6.3

6.4

6.5 6.6 6.7

6.8 6.9

6.10 6.11

6.12

6.13 6.14 6.15 6.16

ADVANCES IN SOLAR ENERGY

Schematic diagram of solar detoxification demonstration plant constructed by ALBAIDA at La Mojonera, Almería, Spain Schematic of a solar treatment coupled with an aerobic biological treatment Emission of carbon dioxide over the period 1700–2000 according to direct atmospheric measurements and ice probing Emission of carbon dioxide in Australia over the period 1990–2000 and the emission level according to the Kyoto target Electrochemical cell showing the principle of electrochemical hydrogen generation through water photolysis using solar energy and related reactions Electrical circuit representing the photo-electrochemical device formed by a semiconducting photo-anode and metallic cathode The electrochemical chain of TiO2-x-based photo-electrochemical device and related charge transfer Band model of n-type semiconductor without surface charge and involving surface charge Effect of light in splitting of the Fermi energy level into two quasi-levels corresponding to electrons and electron holes for an n-type semiconductor The schematic Jonker plot of thermoelectric power vs log m showing the critical parameters Solar energy spectrum in terms of the number of photons vs their energy, showing the flux density for both commercial titania and reduced-band-gap titania Solar energy spectrum in terms of the radiation energy vs wavelength Defect diagram showing the concentration of defects vs oxygen partial pressure for undoped TiO2 at 1273K in absence of foreign elements forming donors or/and acceptors Effect of Fermi energy on the charge transfer between the surface of semiconducting solid and adsorbed species forming either acceptors or donors Effect of cluster and grain size on electronic structure and related band gap width Circuit of the hybrid cell of Morisaki et al (1976) and related electrochemical chain Schematic illustration of the solar-hydrogen generation system using space solar energy collector Schematic illustration of the use of both solar-hydrogen and solar-oxygen as input gases for the production of electricity using a fuel cell

160 161

170

172

176

176 177 182

184 188

192 192

199

201 203 205 206

207

LIST OF FIGURES AND TABLES

6.17 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 7.10 7.11 7.12 7.13 7.14 7.15 7.16 7.17 7.18

7.19 7.20 7.21 7.22 7.23 7.24 7.25 7.26 7.27

The concept of the Australian initiative of the solar-methanol technology Distribution of the heat demand by temperatures Technical and economical potential of solar industrial process heat Distribution of the heat demand in the industry according to different industrial sectors Instantaneous efficiency for different solar collector types Evacuated tube collector Layout of a CPC type collector with tubular absorbers Parabolic trough collector Possibilities for the coupling of the solar system with the conventional heat supply Solar system without storage Solar system with heat storage Yearly energy yield delivered to a process for three sites, depending on the process temperature Heat costs for the three systems at different process temperatures Solar industrial process heat plants in operation Short-term mismatch between heat demand and available solar heat Global solar radiation on a horizontal surface (H) on the Iberian Peninsula Costs of useful heat for the different systems as a function of the mean annual working temperature Combination of solar thermal system and waste heat recovery Primary energy saving with respect to a conventional steam boiler, for 1000MWh of industrial process heat produced either by solar thermal systems or conventional cogeneration system Geographical location of the proposed demonstration plants Heat demand by processes in the malting factory in seville Scheme of the proposed solar system for the malting factory in Seville Malting factory at Poceirão, Portugal: Ground layout Layout of the proposed solar system for the malting factory at Poceirão, Portugal Solar radiation in Portugal and location of the Beiralã plant Scheme of the proposed solar system for the Beiralã plant Heat, cold and domestic hot water demand of the installations of Bodegas Mas Martinet Total heat demand of Bodegas Mas Martinet and solar contribution throughout the year

xvii

210 219 219 220 221 222 223 224 225 226 226 228 229 238 240 241 242 244

244 247 248 249 250 250 252 252 253 254

xviii

7.28 7.29 7.30 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8

8.9

8.10

8.11

8.12

8.13 8.14 8.15 8.16 8.17 8.18 8.19 8.20

ADVANCES IN SOLAR ENERGY

Scheme of the proposed solar system for Bodegas Mas Martinet Autoclave process (simplified) Solar field integration via Ruth-type (saturated steam) storage Countries of the EU-MENA region analysed within this study Annual direct normal irradiance for 2002 Gross hydropower potentials in EU-MENA Temperature at 5000m depth for hot dry rock geothermal power technology Map of biomass productivity Annual average wind speed at 80m above ground level Annual global irradiation on surfaces tilted south with latitude angle Domestic electricity demand in MENA and electricity supplied by existing power stations, by new power plants and by CSP for domestic consumption, export and seawater desalination for the scenario ‘Closing the Gap’ Growth rate of CSP production in MENA in the scenario ‘Closing the Gap’ including domestic power supply, export electricity and seawater desalination Domestic electricity demand in North Africa and electricity supplied by old power stations, by new power plants and by CSP for domestic consumption, export and seawater desalination for the scenario ‘Closing the Gap’ Electricity demand in western Asia and electricity supplied by old power stations, by new power plants and by CSP for domestic consumption, export and seawater desalination for the scenario ‘Closing the Gap’ Electricity demand of the Arabian Peninsula and electricity supplied by old power stations, by new power plants and by CSP for domestic consumption, export and seawater desalination for the scenario ‘Closing the Gap’ Water demand structure in MENA and its evolution until 2050; scenario ‘Closing the Gap’ Water supply from sustainable sources and deficits in MENA a) Water consumption growth rates in MENA; b) Water consumption per capita in MENA Water demand structure in North Africa and its evolution until 2050 Water demand structure in western Asia and its evolution until 2050 Water demand structure for Arabian Peninsula and its evolution until 2050 Groundwater withdrawals as percentage of safe yield for selected countries Water demand, sustainable freshwater resources, non-sustainable supply and potential future supply by CSP via cogeneration by MED plus direct generation via solar electricity in RO plants

255 256 256 262 268 269 270 271 272 273

274

275

278

279

280 286 287 288 289 290 291 292

293

LIST OF FIGURES AND TABLES

8.21

8.22 8.23

8.24 8.25 8.26 8.27 8.28

8.29 8.30

Covering the future freshwater deficits in MENA by transitory non-sustainable sources and by CSP plants using MED in cogeneration plus solar electricity for RO Capacity potential for CSP desalination plants with MED and RO in the three main MENA regions CO2emissions of electricity generation in million tons per year for all countries for the scenario CG/HE and emissions that would occur in a business as usual case Annual per capita CO2 emissions of power generation (Scenario CG/HE) Annual electricity demand and generation within the analysed countries in the MED-CSP scenario Installed power capacity and peak load within the analysed countries in the scenario CG/HE Share of different technologies for electricity generation in 2000 Total electricity consumption and share of different technologies for electricity generation in the analysed countries in 2050 according to the MED-CSP scenario Example of electricity costs and learning in the MED-CSP scenario Projection of a future trans-Mediterranean grid interconnecting the best sites for renewable energy use in EU-MENA

xix

294 294

299 299 300 300 301

301 302

303

TABLES 1.1 1.2 2.1

4.1

6.1

7.1 7.2 7.3 7.4 7.5

Energy consumed for the construction of new power plants Projected cost reductions of renewable energy technologies AM0 Junction parameters for a GaInP top cell together with parameters for lower junction cells composed of GaAs p/n, strain-balanced MQW, DBR enhanced strain-balanced MQW, strained MQW and relaxed MQW Scale accumulation and scaling rate for tubes in a tube-in-shell heat exchanger as determined by chemical analysis Effect of Cr on the energy conversion efficiency (ECE) of a photo-electrochemical cell involving photo-anode made of Cr-doped TiO2 Overview of the industries studied in the POSHIP project Potential for solar industrial process heat in the Spanish industry Total investment costs for the solar collector field related to the gross collector area List of solar process heat plants Selection of solar collector types according to working temperatures

32 35

63

120

202 218 220 227 231 243

xx

7.6 7.7 7.8 7.9 7.10 7.11 7.12 8.1

8.2 8.3

8.4

8.5

8.6

8.7 8.8 8.9 8.10 8.11 8.12

ADVANCES IN SOLAR ENERGY

Evaluation criteria for solar industrial process heat systems for the Iberian Peninsula List of the most relevant parameters of the demonstration projects proposals Characteristic data of the proposed system for solar air heating in a malting factory in Seville, Spain Characteristic data of the proposed system for air preheating for malt drying in Portugal Characteristic data of the proposed system for a textile drying process Characteristic data of the proposed system for space heating and cooling of a wine cellar Annual system performance for different sites Renewable electricity performance indicators representing the average renewable electricity yield of a typical facility in each country Economic potentials of renewable energy sources in the southern EU and MENA region Electricity demand in MENA and generation by conventional old and new plants and by CSP technologies in MENA until 2050 Domestic electricity demand in North Africa and generation of domestic power, export electricity and power for desalination by CSP technologies until 2050 Domestic electricity demand in western Asia and generation of domestic power, export electricity and power for desalination by CSP technologies until 2050 Domestic electricity demand of the Arabian Peninsula and generation of domestic power, export electricity and power for desalination by CSP technologies until 2050 Annual per capita water consumption in 2000 and projected demand in 2010 for the GCC countries Solar resources in the GCC countries Comparison of energy consumption for six types of water desalination The available desalination capacity in the GCC countries, in 2000, using different desalination technologies Global solar radiation, monthly and annual means Areas required for renewable electricity generation in 2050 for the scenario CG/HE

245 247 249 251 253 255 257

267 267

276

278

279

281 282 283 283 284 296 297

About the Authors Diego C. Alarcón Padilla is a project researcher at the Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas (CIEMAT) at Plataforma Solar de Almería (PSA) and has been working on different solar thermal research topics since 1994. Since 2000 he has been working within the PSA environmental applications of solar energy area, with a special involvement in solar thermal seawater desalination projects. His first research and development work was on computer simulation of optical devices for solar thermal energy applications and software development for sun-tracking devices. Dr. Alarcón is also currently a lecturer in computer science at the University of Almería, Spain. W. E. Alnaser holds a Ph.D. in physics and has been Professor of Applied Physics at the University of Bahrain since 1997, where he served as the Dean of the College of Science and also the Dean of Scientific Research. His research work is diversified but concentrated in solar radiation modelling, measurement and application. Prof. Alnasar devised a small mobile solar and wind energy electricity generator (2.5kW) as well as participated in constructing a mobile solar reverse osmosis desalination unit (250 gallons per day). He is the author of more than 100 refereed articles on material science and superconductivity, renewable energy, astronomy and astrophysics, and environmental physics and 30 books on physics, renewable energy and astronomy. He is currently serving as Chairman for the Arab Section at the International Solar Energy Society (Germany), as well as President for the International Energy Foundation for Middle East. He is a Fellow of the Institute of Physics in London. Professor Alnaser has received many awards at local and professional levels, including the Shuman Prize (Jordon) for Best Youth Physicist, the Islamic Educational, Scientific and Cultural Organization (ISESCO) Prize for Best Research, and the State Prize (Bahrain) for Community Science Services. He is the past editor-in-chief of the Journal of Arab Association Universities for Basic and Applied Sciences and is now its managing director. He has been a referee for more than 12 leading international journals and currently serves as associate editor for two international journals. Tadeusz Bak is a senior research fellow at the Centre for Materials Research in Energy Conversion, University of New South Wales, Sydney, Australia. He received his Ph.D. from the Academy of Mining and Metallurgy, Cracow, Poland in 1982. In 1996 he joined the School of Materials Science and Engineering, University of New South Wales as a research fellow. His expertise includes electrochemistry and materials science. His current research interests include diffusion in ionic solids and photo-electrochemistry of metal oxides. He has published over 100 refereed papers and 17 book chapters. Julián Blanco Gálvez earned his Ph.D. at the University of Almería in 2002 following specialization in industrial engineering at Seville University. Dr. Blanco is a senior researcher at the Spanish Ministry of Education and Science. Since 1995 he has served

xxii

ADVANCES IN SOLAR ENERGY

as both the head of the Solar Chemistry Department of Plataforma Solar de Almería (PSA), and the Spanish National Representative in the Task II group of the International Energy Agency – SolarPACES (Solar Power and Chemical Energy Systems). Dr. Blanco has been involved in many national and contractual research and development projects, as well as work for the European Union related to the development of solar technologies as applied to waste water treatment processes and techniques. Dr. Blanco is author of 1 book and co-author of 3 books and 8 chapters in different books, some 40 publications in indexed international journals and more than 90 contributions to international congress and symposiums. He has been honoured with the Jury’s Grand Prix at the European Grand Prix for Innovation Awards, 2004, in Monaco (www.european-grandprix.com/index_en.htm). Maria João Carvalho has worked in solar thermal energy since the 1980s and is currently the senior researcher in the Renewable Energy Department of the National Institute for Engineering, Technology and Innovation (INETI), a state laboratory in Portugal. She is the director of the Solar Collector Testing Laboratory of INETI. Dr. Carvalho is the author of 19 papers published in international technical journals and 38 papers published in conference proceedings. She has given invited lectures on renewable energy for universities and also seminars on the subject of solar thermal energy with special emphasis on collector and system testing. She is a member of the Portuguese Section of ISES. Jane H. Davidson has been a professor in the Department of Mechanical Engineering at the University of Minnesota since 1993. Before coming to Minnesota, she was a faculty member at the University of Delaware and Colorado State University. Dr. Davidson is past editor-in-chief of the Journal of Solar Energy Engineering. She was the 2004 recipient of the ASME John I. Yellott Award for outstanding research in solar energy and is a fellow of the American Society of Mechanical Engineering and the American Solar Energy Society (ASES). Her publications include more than 150 papers in journals and conferences and 3 book chapters. She was the 2005 recipient of the University of Minnesota’s Distinguished Woman Scholar Award in Science and Engineering. Her research interests include solar thermal systems and energy efficient buildings. N. J. Ekins-Daukes is a lecturer at the University of Sydney’s School of Physics, having earned his Ph.D. from Imperial College, London in 2000 for his work on strain-balanced quantum well solar cells. He later held a fellowship from the Japan Society for the Promotion of Science (JSPS) at the Toyota Technological Institute in Nagoya, Japan. Dr. Ekins-Daukes, present research involves the application of nano-structures to high efficiency photovoltaic devices as well as contributing towards projects on the radiation resistance of III-V solar cells and the application of III-V solar cells in concentrator systems and the application of nano-structures to photovoltaic devices. Dr. Ekins-Daukes is a member of the Institute of Electrical and Electronics Engineers (IEEE). He has published papers on a range of topics covering quantum well solar cells, luminescent up-conversion, radiation resistance and photovoltaic concentrator systems. João A. Farinha Mendes is a mechanical engineer (thermodynamics), currently working as a senior researcher and is the Head of the Solar Energy Unit in the Renewable Energy

ABOUT THE AUTHORS

xxiii

Department (DER) at the National Institute for Engineering, Technology and Innovation (INETI), a state laboratory in Portugal. His main research activity has been devoted to solar thermal applications, serving as national representative for several European projects, leading to a number of papers on non-imaging optics, phase change heat transfer and solar system design published in international technical journals and conference proceedings. He is the Portuguese representative for the International Energy Agency’s Executive Committee on Solar Heating and Cooling Implementing Agreement and has participated in the standardization work of solar collectors at the national, European and international levels. He has given invited lectures on solar thermal energy for universities and participated in numerous seminars in Portugal and other countries in Europe. Pilar Fernández Ibáñez earned her Ph.D. in applied physics from the University of Granada in 2004, following specialization in environmental sciences and applied physics at the University of Almería, Spain. Dr. Fernández has ten years of work experience in various research sectors, participating in ten European Union and two national research and development projects related to disinfection and wastewater treatment with the help of solar technologies development. She is co-author of three books and six chapters in other collaborative books. She holds two patents, and is co-author of 30 international technical journal articles and 50 contributions to different international congresses and symposiums. She has tutored the scientific work of students on solar disinfection and detoxification of wastewater. Dr. Fernández has also been Invited Guest Associate Editor for special issues of the indexed journals Catalysis Today and Solar Energy. Lorraine F. Francis is a professor of chemical engineering and materials science at the University of Minnesota. She joined the University in 1990 and currently serves as the Director of Undergraduate Studies in Materials Science and Engineering. Dr. Francis received the National Science Foundation (NSF) Young Investigator Award (1993–1998) and the DuPont Young Professor Grant (1994–1997). Her research interests include polymer and ceramic coatings, processing-microstructure-property relationships in coatings and composites, interfacial studies and biomaterials. Dr. Francis is a member of the American Ceramic Society (ACerS), the Materials Research Society (MRS) and ASM International. Wolfgang Gernjak earned his Ph.D. in land and water management from the University of Natural Resources and Applied Life Sciences (BOKU) in Vienna, following specialization in analytical and physical chemistry at the Vienna University of Technology. He has been involved in five European Union research and development projects connected with the application of solar photocatalysis and solar photo-Fenton for the detoxification of industrial wastewater and disinfection of drinking water. He is author of 20 articles in indexed international journals and 33 contributions to international congress and symposiums. Klaus Hennecke earned his Doctor of Engineering degree in aerospace technologies in 1978 at the University of the Federal German Forces, Munich, Germany. In 1989 he became a manager of a series of research projects in the Solar Research Division of the Institute of Technical Thermodynamics (ITT) within the German Aerospace Center (DLR) in

xxiv

ADVANCES IN SOLAR ENERGY

Cologne, Germany. His research is particularly involved with the field of direct steam generation in parabolic trough collectors and solar process heat applications. As head of the applications subdivision Dr. Hennecke is currently responsible for the transfer of research and development results into the market. Engaged in international cooperations under the umbrella of the International Energy Agency (IEA), implementing the Solar Power and Chemical Energy Systems (SolarPACES) agreement, Dr. Hennecke leads the subtask ‘System Integration and Demonstration’ of the new IEA SHC Task 33/SolarPACES Task 4 on Solar Heat for Industrial Processes (SHIP). Gerhard Knies holds a Ph.D. in physics and has worked in the fields of high energy physics and elementary particle research at Deutsches Elektronen Synchrotron (DESY) Hamburg, Germany; the European Centre for Nuclear Research (CERN) in Geneva, Switzerland; and the Stanford Linear Accelerator Center (SLAC) in both Stanford and Berkeley, California, US. Dr. Knies developed a road map towards energy, water and climate security by use of solar energy. In 1981 he co-founded the Hamburg Scientists Peace Initiative, a disarmament campaign against the threatening East–West arms race, and made studies on the vulnerability of industrial civilization. In 1995 he founded the Hamburg Climate Protection Foundation, which in 2003 allied with other groups to form the Trans-Mediterranean Renewable Energy Cooperation movement for development, climate stabilization and good neighbourhoods. Dr. Knies has been retired since 2001. Dirk Krueger earned his B.Sc. in energy and environmental engineering at the University of Applied Sciences in Aachen–Juelich, Germany. Since 1997 he has specialized in investigating parabolic trough collector systems for process heat applications. Dr. Krueger’s major research fields are collector testing, simulations for output forecasts and project assistance for companies engaged in parabolic trough technology. Sixto Malato Rodríguez is a senior researcher of the Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas (CIEMAT) at Plataforma Solar de Almería (PSA), Spain, and has 18 years of experience in various research sectors. Since 1990 his work at the PSA-CIEMAT has involved all the European Union research and development projects linked to the solar detoxification of water and solar wastewater treatment technologies. He is author of 1 book and co-author of 5 books as well as 21 chapters in others. Dr. Malato has also co-authored more than 80 publications in indexed international journals, 21 articles in technical journals and more than 140 contributions to 68 different international congresses and symposiums and holds 4 patents. In 2004 he received the Jury’s Grand Prix at the European Grand Prix for Innovation Awards in Monaco (www.european-grandprix.com/index_en.htm). Manuel I. Maldonado Rubio earned his Ph.D. from the University of Almería in 2001, following specialization in environmental sciences and chemistry at the University of Granada and the Instituto de Investigaciones Ecológicas in Málaga, Spain. Dr. Maldonado has worked for the the Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas (CIEMAT) at Plataforma Solar de Almería, Spain, since 2002, and has been involved in seven European Union research projects, three national

ABOUT THE AUTHORS

xxv

research projects and three research and development contracts related to the development of solar technologies applied to wastewater treatment. He has authored 29 peer-reviewed articles for international publications, co-authored 1 book and 5 chapters in others, and given presentations at 52 international and national congresses. Susan C. Mantell is a professor of mechanical engineering at the University of Minnesota. Dr. Mantell has received several awards for excellence in teaching and research including a National Science Foundation (NSF) Young Investigator Award (1994–1999) and a McKnight Land-Grant Professorship (1995–1997). She is a leading expert in the field of composite materials and manufacturing and most recently has been working on development of polymer heat exchangers for solar energy applications and for use in the transportation industry. Dr. Mantell has served as an associate editor of the Journal of Composite Materials since 2000 and the Journal of Materials Processing and Manufacturing Science from 1998 to 2003. She has published over 75 papers and 2 book chapters on composite materials and polymer processing. Janusz Nowotny received his Ph.D. in solid-state chemistry from the Institute of Physical Chemistry, Polish Academy of Science, in 1967, and following specialized research in diffusion in nonstoichiometric compounds earned a D.Sc. in materials science from the Academy of Mining and Metallurgy, Cracow, Poland in 1974. Dr. Nowotny is Director of the Centre for Materials Research in Energy Conversion, University of New South Wales, Sydney, Australia. He has been a visiting professor at several universities in France, Germany and Japan. Dr. Nowotny s expertise includes photo-electrochemistry, solid-state electrochemistry, defect chemistry and the science of materials interfaces. His research includes defect chemistry of oxide semiconductors, diffusion in ionic solids, segregation in nonstoichiometric compounds and photo-electrochemistry. He has published over 400 refereed papers, and edited 17 books. Jörg Schindler is a business economist and since 1984 has worked for the LudwigBölkow-Systemtechnik GmbH, becoming its managing director in 1992. His major activities include the technical and economic analyses of market introduction of photovoltaics and other renewable energy sources, clean drive systems for road transport, clean fuels produced from renewable energy sources (like hydrogen), and the future availability of fossil fuels. Mr. Schindler is currently a member of the board of ‘e5 – The European Business Council for a Sustainable Energy Future’ and from 1999 to 2003 was a member of the Enquete Commission of the Bavarian House of Representatives entitled ‘New Energy for the New Millennium’. Enquete Commissions serve the national government by collecting scientific state-of-the-art information and then transmitting it through publicly discussed booklets. He is also a member of the board of the Global Challenges Network. Hans Schweiger is director of the thermo-energetical systems and renewable energies company energyXperts.BCN, Barcelona, Spain. Since 1991, Dr. Schweiger has been active in the field of solar thermal energy. He has worked in collaboration with several national and international research projects in the fields of solar thermal energy (solar collector development, transparent insulation, building simulation and absorption cooling) and

xxvi

ADVANCES IN SOLAR ENERGY

numerical methods for heat transfer and fluid dynamics (software development for CFD and radiation heat transfer). He served as project coordinator of the European Union research project on solar industrial process heat, POSHIP (NNE5-1999-0308). Dr. Schweiger also teaches master’s and post-graduate courses on solar thermal energy at the Catedra UNESCO (Polytechnical University of Catalonia). He was co-founder of the engineering cooperative Aiguasol Engineering (Barcelona, Spain), where he was active from 1999 to 2005. He earned his degrees from the Politechnical University of Catalonia (UPC, Terrassa, Barcelona, Spain), specializing in the optimization of solar thermal absorber elements with transparent insulation. Leigh Sheppard is a postdoctoral fellow at the Centre for Materials Research in Energy Conversion, School of Materials Science and Engineering, University of New South Wales, Sydney, Australia. In 2006 he received the Deutsche Akademische Austausch Dients award of the Hahn-Meitner Institute to work with Professor H. Tributsch. Dr. Sheppard’s research interests include photo-electrochemistry of TiO2. He has published 16 refereed papers and 2 book chapters. Charles C. Sorrell is Professor of Ceramic Engineering, School of Materials Science Engineering, University of New South Wales, having earned his Ph.D. in ceramic engineering at the University of New South Wales in 1987. His main research interest areas include ceramic processing, phase equilibria, microstructure and performance. Professor Sorrell’s other research activities are focused on titania-based materials for environmentally friendly applications, bioceramics and microwave heating, glass-ceramics, refractories, high temperature superconductors, graphite, silicon nitride, zirconia, and gemstone heat treatment. Professor Sorrell is a fellow of the Australian Institute of Energy (AIE) and the Institution of Engineers/Australia (IEAUST). He has been editor of the Journal of the Australasian Ceramic Society, Advances in Applied Ceramics, Metals and Materials International, and Ceramics International. Professor Sorrell has published over 350 papers and supervised 75 B.Sc. projects, as well as 35 postgraduate students. Sam-Shajing Sun obtained his Ph.D. degree in polymer/materials chemistry from the University of Southern California in 1996. After a postdoctoral appointment at the Loker Hydrocarbon Research Institute, Dr. Sun joined the Norfolk State University in early 1998. He is currently leading an organic and polymeric materials research programme at the Center for Materials Research (CMR) at Norfolk State University. Dr. Sun is also directing a Center for Research and Education in Advanced Materials sponsored by NASA. Dr. Sun’s research interests include the moulding, design, synthesis, processing and characterization of novel organic and polymeric electronic and photonic materials. Current research projects of his lab focus on the development of novel organic/polymeric photovoltaic materials for potential ‘plastic’ solar cells and photodetector applications. Dr. Sun has published more then 50 refereed papers and presented more then 40 papers and lectures relevant to organic/polymeric optoelectronic materials at international scientific communities and academic/governmental organizations. He has recently edited a new CRC Press book titled Organic Photovoltaics: Mechanisms, Materials and Devices, a comprehensive book covering the state of the art of the field. Dr. Sun is a member and technical referee of a number of major scientific organizations.

ABOUT THE AUTHORS

xxvii

Franz Trieb has worked in the field of renewable energies since 1983. After the implementation of hydrogen storage for an autonomous renewable energy system at the University of Oldenburg, Germany, he obtained specialized training in renewable energy at the National University of Tacna, Peru. Since 1994, Dr. Trieb has worked as project manager at the Institute of Technical Thermodynamics of the German Aerospace Center (DLR), working on solar energy resource assessment by satellite remote sensing, market strategies for concentrating solar power and renewable energy scenarios. Two of Dr. Trieb’s recent studies look at the future demand and the sustainable supply of electricity using the potential of renewable energy sources in the countries of the Mediterranean Region and Europe (www.dlr.de/tt/med-csp and www.dlr.de/tt/trans-csp). Werner Zittel obtained his Ph.D. in physics in 1987 from the Max-Planck Institute for Quantum Optics, Garching, and at the Technical University of Darmstadt, both in Germany. He has worked as a scientific collaborator for the German Aerospace Center (DLR) and at the Institute for Technical Physics of the German Research Center for Aircraft and Space Technology (DFVLR, today DLR) in Stuttgart and at the Fraunhofer Institute for Solid State Technology, in Munich, Germany. Since 1989, he has worked for Ludwig-BölkowSystemtechnik GmbH, Ottobrunn, Germany. Dr. Zittel actively researches climate change effects on the trace gas methane (especially in Russia); the external and social costs of energy use in the public transport sector; the energetic amortization of renewable energy generation technologies; advanced hydrogen storage technologies; the use of hydrogen in existing natural gas grids; hydrogen production from biomass; analysis of the existing hydrogen markets; introduction scenarios for hydrogen into the energy and transport economy; fundamental questions of the power industry; analysis of fossil resource availability; and advanced hydrogen storage systems for automotive use. He is a member of the Association for the Study of Peak Oil, Germanwatch e.V., and Global Challenges Network e.V.

This page intentionally left blank

1

Alternative World Energy Outlook 2006: A Possible Path towards a Sustainable Future Jörg Schindler and Werner Zittel

Abstract This paper sketches a possible path towards the future energy supply in the light of foreseeable challenges and opportunities. First, the shortcomings of the World Energy Outlook, which is biennially published by the International Energy Agency, are revealed. The deficiencies of these reports are the reason why a more progressive approach and alternative scenarios are needed. An alternative approach must address both the challenges as well as promising pathways towards a sustainable energy future. Second, we briefly address the challenges and discuss those non-sustainable approaches which represent today’s conventional wisdom in the energy community. Finally, and most important, we want to describe the real potential of renewable energies in a scenario called ‘Alternative World Energy Outlook 2006’. It is not the intention of this paper to draw a complete and exact picture of what our energy future will look like, as every attempt to do so is certain to fail. However, the aim is to describe the limits of what is possible and what is impossible simply due to geological and physical restrictions. Most relevant in this respect is to take into account the peaking of oil production in the near future and the peaking of natural gas production in the mid-term future.

■ Keywords – World Energy Outlook; renewable energy; climate change; economic principles; logistic growth

1.1 WINDS OF CHANGE: THE TRANSITION PERIOD The main thesis of this paper is that we are at the beginning of a structural change of our economic system. This change will be triggered by declining fossil fuel supplies and will influence almost all aspects of our daily life. Climate change will also force mankind to change the energy consumption pattern away from fossil fuel combustion. This is a very serious problem. However, the focus of this contribution is on resource depletion aspects, as these are much less transparent to

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

2

J. SCHINDLER AND W. ZITTEL

GLOSSARY OF TERMS USED IN THIS PAPER API ASPO AWEO bbl boe BGR BP BTU CO2 DoE DTI EIA EUR EWEA FSU Gb GoM Gtoe GW IAEA IEA IHS IPCC LBST LNG M Mb Mtoe NEB NPD OECD OPEC

R/P SO2 SOT t toe Tcf UK USA WEO yr

American Petroleum Institute Association for the Study of Peak Oil Alternative World Energy Outlook barrel (1 barrel = 159 litres) barrel oil equivalent Bundesanstalt für Geowissenschaften und Rohstoffe (German federal agency for resources) company name (originally standing for British Petroleum) British thermal unit carbon dioxide Department of Energy Department of Trade and Industry, UK Energy Information Administration estimated ultimate recovery European Wind Energy Association former Soviet Union gigabarrel (=109 barrels) Gulf of Mexico gigatons of oil equivalent (= 109 tons) gigawatt (= 109 watts) International Atomic Energy Agency International Energy Agency IHS-Energy, company name Intergovernmental Panel on Climate Change Ludwig-Bölkow-Systemtechnik GmbH, company name liquefied natural gas million million barrels megatons of oil equivalent (= 106 tons) National Energy Board, Canada National Petroleum Directorate, Norway Organisation for Economic Co-operation and Development Organization of Petrol Exporting Countries (OPEC member states are: Algeria, Indonesia, Iran, Iraq, Kuwait, Libya, Nigeria, Qatar, Saudi Arabia, the United Arab Emirates and Venezuela) reserve to production (ratio) sulphur dioxide solar thermal power plants ton ton oil equivalent (1 toe = 7.1 boe) teracubic feet (= 1012 cubic feet) United Kingdom United States of America World Energy Outlook year

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

Alternative World Energy Outlook 2006

3

the public. The following scenarios try to figure out if the present resource depletion problem eventually could be mastered just by increasing the extraction rate of other conventional energy sources. This transition period probably has its own rules which are valid only during this phase. Things might happen which we have never experienced before and which we may never experience again once this transition period is over. Specifically our way of dealing with energy topics may change completely, which will have consequences for our economic systems. The International Energy Agency (IEA) denies that such a fundamental change of our energy supply is likely to happen in the near or medium-term future and therefore does not give a warning that our economic system is in danger. Therefore the views of the IEA are briefly analysed. It will be shown that the ‘business as usual’ approach of the biennially published World Energy Outlook does not describe our energy future appropriately. After that the main drivers determining the transition to a changing energy future are discussed. It is important to understand that we are entering such a transition period, leading to fundamental changes. The acknowledgement of this development and the resulting mindset are necessary preconditions for coping with this situation in an appropriate way. This will lead to completely different behaviour on the part of individuals, companies and governments. The imminent transition is not a voluntary act in which people might or might not engage because changing boundary conditions will force us to adapt our energy system and also our way of life.

1.1.1 CRITIQUE OF THE WORLD ENERGY OUTLOOK 2004/2005 The IEA biennially publishes a World Energy Outlook (WEO) describing the most probable development of energy supply and demand over the next 20 to 30 years.1 These outlooks differ only in minor details from year to year and follow this general pattern: ● Demand is projected by extrapolating past experience. According to these demand

projections, energy consumption will rise tremendously until 2030. There will be no sign of rising supply problems or an imminent peak of oil production. ● Fossil fuel use will dominate the supply. Renewable energy sources (apart from hydropower or traditional biomass use) will contribute only a minor share and will only rise marginally over the next 25 years. Even by 2030 the contribution will still be only about 2 per cent. A more detailed analysis reveals that the secured supply of this rising demand over the next 30 years relies on a few basic assumptions which are not questioned by the IEA.2, 3 The main assumptions are: ● On the one hand it is claimed that oil resources are sufficient to supply the projected

rise of consumption. At the same time, the IEA recognizes that proved reserves are not sufficient and new reserves must be discovered. In addition, it is stated that the uncertainty and poor quality of data is of growing concern to all involved in the oil industry. Even with this background, the IEA concludes that growth of oil

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

4

J. SCHINDLER AND W. ZITTEL

consumption is possible and likely disregards the uncertainties. More detailed analyses by critical observers (for example by the Association for the Study of Peak Oil) show that it is much more probable that the existing production basis is shrinking and the peak of oil production is very close. The evidence provided by independent oil geologists and analysts is not discussed seriously by the IEA and is more or less ignored. ● The IEA states that many more financial resources than in the past will have to be invested in the upstream sector in order to compensate for the rising decline of the existing production base. Thus the future availability of sufficient resources is reduced to a purely financial problem rather than lack of resource base. At the same time the IEA observes that it is very uncertain whether oil companies and oilproducing states will increase their upstream investments sufficiently. ● As a consequence of the assumed rise of fossil fuel supply, the future growth of renewable energies is projected based on unchanged conditions and thus assuming no need to raise their share. Even 25 years from now the IEA foresees only a marginal contribution from renewable energy sources. However, it must be noted that all previous WEO reports have considerably underestimated the actual growth of wind energy. The now projected future growth rates still fall far behind the experienced rise of all renewables in the past 10–20 years. The IEA approach shows the fundamental difference between observers influenced by economic reasoning and those influenced by geological reasoning: the economic observers completely ignore the influence of geological restrictions, instead seeing possible future supply problems simply as a consequence of insufficient investments in the upstream sector. In the latest WEO report, the IEA argues that oil prices will decline as soon as oil companies invest more money in the upstream sector. Independent of whether oil resources are sufficient or not, it is hard to imagine that oil companies will invest more money to reduce prices. Regarding future oil prices, many observers have changed their views over the last year. Hardly anybody still believes that future oil prices will return to low levels of below US$40–50/bbl. In the following paragraphs some results of the latest two WEOs are discussed. Figure 1.1 shows the projections of the WEO as published by the IEA in 2004 and 2005. It shows the development of the primary energy supply since 1965 and the forecast for the next 25 years (until 2030). The fossil energy sources (oil, coal and natural gas) together provide almost 80 per cent of present world energy needs. Oil is the most important fuel with a share of almost 40 per cent. Biomass, mostly used in an unsustainable manner for cooking and heating in developing countries, covers between 10 and 15 per cent of the world energy demand. Nuclear and hydropower produce roughly the same amount of electricity at world level, each with a share of about 18 per cent of world electricity production. In the graph the share of nuclear energy is exaggerated due to the primary energy supply calculation method, which assumes an efficiency of 33 per cent for primary nuclear fuel to electric power provision and almost 100 per cent for hydropower to electricity. Finally, the smallest

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

Alternative World Energy Outlook 2006

5

Mtoe 6000 oil 5000

2005

WEO 2004 WEO 2005 WEO 2005-altern. WEO 2005-low invest gas

4000

coal 3000 2000 biomass 1000 Nuclear (x3) 0 1965

Other-WEO2005-alt Other-WEO2005 Other-WEO2004

hydro

1985

2005

2025

Year Source: Historical data – BP Statistical Review of World Energy Outlook – International Energy Agency 2004/2005

FIGURE 1.1 World energy demand 1971–2030 as given in the IEA’s World Energy Outlook 2004 and 2005

contribution, with about 1 per cent is derived from all other renewables such as wind, direct solar energy, geothermal energy and wave power. The contribution from these sources might rise to a share of 2 per cent in 2030, according to the WEO projection. The graph tells us that oil will remain the most important fuel. Its consumption and production levels will increase over the next 25 years by the same amount as they did during the last 40 years. The message to consumers is that in the next two decades we will see no major changes. Not shown in the graph, the price of oil is assumed to rise to US$27/bbl by 2030 in WEO 2004 and to US$39/bbl in WEO 2005. The latest WEO 2005 report also discusses a ‘low investment’ scenario and an ‘alternative policy’ scenario. The projected oil consumption for these two alternative scenarios is also shown in Figure 1.1. According to the report, the low investment scenario would result in an oil price rise of up to US$52/bbl by 2030. In contrast to these scenarios, however, oil was traded at US$60–70/bbl for several months in 2005 and early in 2006 and has risen above US$75/bbl. Another IEA message is that natural gas and coal supplies will compete for the second most important position behind oil, with natural gas supplies probably developing more rapidly than coal: the supply of natural gas is expected to more than double within the next 25 years. At present, North America and Europe are the biggest natural gas markets, consuming more than half of the world’s natural gas production. In both regions natural gas production has already peaked. North America is experiencing a decline of its domestic supply. Only very small amounts of natural gas are imported from foreign sources due to limited import capacities. There is no doubt that large quantities of natural gas are still in the ground. However, it is doubtful that future global extraction rates will exceed past

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

6

J. SCHINDLER AND W. ZITTEL

levels when production in North America and Europe is already in decline. Any new natural gas supplies must first close this gap before there can be a net increase. This may turn out to be a big challenge! A few comments can also be made on the other energy sources. The growth of biomass use is supposed to continue as in the past. On nuclear energy the report spends just 1 out of more than 400 pages, claiming without any further justification that total production will remain almost at present levels. New renewable energies like wind or solar will almost triple their contribution within the next 25 years. This is based on an expected average growth rate of about 4.5 per cent per year. However, this rate is far below the actual experience: during the last 15 years the average growth rate of wind energy was about 30–35 per cent per year, for solar thermal energy about 10–15 per cent and for solar electricity production 15–20 per cent and strongly accelerating in the last few years. It is obvious that the IEA systematically overestimated the future role of fossil energy sources and underestimated that of renewable energy sources.

1.1.2 DRIVING FORCES OF FUTURE ENERGY DEMAND/SUPPLY What are the driving forces that influence our present energy supply and may even grow in importance in the years to come? Three main drivers can be identified. Most prominently, in the last 20 years mankind has learned a lot about the relationship between fossil fuel combustion and its role in heating the Earth’s atmosphere. Basic physics teaches us that the more we pollute the atmosphere with carbon dioxide, the more we change the radiation balance. This results in heating the Earth’s surface temperature with the consequence of increased evaporation and rainfall. An increase of water vapour in the air is almost identical to more energy in the air and consequently must result in stronger rainfalls and storms. These effects today are beyond any doubt. The remaining uncertainties are: To what amount can competing effects dampen the rise in temperature and how large is the range of the natural variability of our climate? How much does the temperature still need to rise until we are convinced that the observed rise is related to anthropogenic influence? Figure 1.2 sketches the dilemma: economic principles imply that gross domestic product needs to grow each year in order to maintain the living standard of the population. All efforts are directed towards this growth. The main advice to developing countries is to accelerate their economic growth. But economic growth is almost completely linked to the growth of material products based on rising consumption of raw materials and fossil energy. All this happens in spite of the indisputable but ignored fact that unrestricted growth of limited resources is simply not sustainable. The best way to illustrate this basic dilemma is to consider climate change. Figure 1.2 shows the history of world energy consumption, which is directly related to the level of greenhouse gas emissions. The projected alternative trends exhibit schizophrenia. On the one hand it is believed that a growing consumption of fossil fuels is needed to keep the economies running and to increase welfare; on the other it is known that only a strong reduction of fossil fuel use will save our planet. Another potent driver influencing the energy future is the natural endowment of the Earth with fossil energy resources. Oil is the first resource whose availability is beginning

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

Alternative World Energy Outlook 2006

7

Gtoe/yr (Gigatons oil equivalent/year) 20 18 16

Business as usual: economic growth

14 12 other

10 8 6

coal

4

gas

2

oil

1920

1940

1960

1980

Emission reduction: ecological welfare 2000

2020

2040

Year

Source: BP Statistical Review of World Energy

FIGURE 1.2 The basic dilemma – Business as usual or climate policy; growing fossil energy consumption is tantamount to growing air pollution and reinforced greenhouse effect

to decline. Within a few years it will be followed by the decline of natural gas production and after that by the geological restrictions of uranium and coal supply. The question is not whether we like this or not – depletion will occur even if one tries to ignore it. The limited resource base will be detailed in the following section. The third major driver is more positive. During recent years we have seen lots of improvements and technical innovations in promoting the use of renewable energy sources. This positive driver will make it easier to manage the transition from the unsustainable present to a more sustainable future. However, this is not an automatism. How fast the major economies of the world especially will move in such a direction depends on many players – these are the political leaders, the industry leaders, and of course the consumers and voters. The extent and speed of action will be strongly influenced by the assessment of the present situation.

1.2 FUTURE AVAILABILITY OF FOSSIL AND NUCLEAR ENERGY SOURCES 1.2.1 OIL The conditions for the world’s oil supply have entered into a new phase, with increasing demand pressures, worries about the security of supply in important oil-producing countries, speculative factors, and clear indications that limitations on the supply side have caused unexpected and high price increases. In view of the fact that increasing oil production is obviously becoming more and more difficult, it is now almost irrelevant whether the peak of oil production has already been reached or whether growth of production ‘just’ can’t keep pace with rising demand anymore. With accumulating evidence we will soon be able to decide between the conflicting views held by the optimists and pessimists. According to the doctrine of the optimists

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

8

J. SCHINDLER AND W. ZITTEL

(mostly economists), rising prices will induce a fast increase of oil exploration and production which in turn will lead to a relaxation in the oil market in the near future. In contrast, the pessimists (mostly influenced by geological considerations) expect that it will become increasingly difficult to balance the increase in demand by a sufficient rise in supply. As a consequence production will not be able to keep up with demand and, after a short phase of stagnation, will inevitably decline. Many events during the last five years vindicate the theses of the pessimists. But it would be much more appropriate to leave these misleading categories behind and to speak only about ‘realistic’ and ‘unrealistic’ views. Facts and not beliefs will decide the issue. In the following sections the present state of the world’s oil supply will be outlined. (For a more detailed analysis see Notes 4, 5 and 6.)

1.2.1.1 General pattern of oil production The different phases of oil production can be described schematically as follows. In the early phase of the search for oil, the easily accessible oil fields are found and developed. With increasing experience the locations of new oil fields are detected in a more systematic way. This leads to a boom in which more and more new fields are developed, initially in the primary regions, later all over the world. Those regions which are more difficult to access are explored and developed only when sufficient new oil can no longer be found in the easily accessible regions. As nobody will look for oil without also wanting to produce it, in general shortly after the discovery of new fields their development will begin. With increasing production the pressure of an oil field diminishes and the water levels rise, and after some time the production rate begins to decline. This trend can be controlled to a certain extent so that the decline of the production rate is delayed or reduced: by injecting gas or water into the reservoir in order to increase the pressure, by heating the oil or by injecting chemicals in order to reduce the viscosity of the oil. In every oil province the big fields will be developed first and the smaller ones later. As soon as the first big fields of a region have passed their production peak, an increasing number of new and generally smaller fields have to be developed in order to compensate for the decline of the production base. From there on, it gets increasingly difficult to sustain the rate of production growth. A race begins which can be described as follows. More and more large oil fields show declining production rates. The resulting gap has to be filled by bringing into production a larger number of smaller fields. However, these smaller fields reach their peak much faster and then contribute to the overall production decline. As a consequence, the region’s production profile, which results from the aggregation of the production profiles of the individual fields, becomes more and more ‘skewed’, the aggregate decline of the producing fields becomes steeper and steeper. This decline has to be compensated for by the ever faster connection of more and more ever smaller fields. This pattern can be observed very well in many oil provinces. However, sometimes this general pattern has not been followed, either because the timely development of a ‘favourable’ region was not possible for political reasons or because of the existence of huge surplus capacities so production was held back for a longer period of time. However,

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

Alternative World Energy Outlook 2006

9

the more existing spare capacities are reduced, the closer the production profile will follow the described pattern. In the history of oil production, which is now extending over more than 150 years, we can identify some fundamental trends: ● ● ● ●

the world’s largest oil fields were all discovered more than 50 years ago; since the 1960s annual oil discoveries have decreased tangentially; since 1980 annual consumption has exceeded annual new discoveries; till now more than 42,000 oil fields have been found, but the 400 largest oil fields (1 per cent) contain more than 75 per cent of all oil ever discovered; and ● the historical maximum of oil discoveries must be followed by a maximum of oil production (the ‘peak’). How close to the peak have we already got? How steep will the decline be after the peak? These are the crucial questions remaining.

1.2.1.2 Countries outside OPEC and the former Soviet Union At the global level, the development of different oil regions took place at different times and at varying speeds. Therefore today we are able to identify production regions being in different development stages and with this empirical evidence we can validate with many examples the simple considerations described above. Looking at the countries outside the former Soviet Union (FSU) and OPEC, it can be noticed that their total production increased until about 2000, but since then total production has been declining. A detailed analysis of the individual countries within this group shows that most of them have already reached their production peaks and that only a very limited number of countries will still be able to expand production, particularly Brazil and Angola. Responsible for the stagnation of the oil production in this group of countries was the peaking of the oil production in the North Sea which occurred around 2000 (1999 in Great Britain, 2001 in Norway). Worldwide onshore oil production had reached a plateau much earlier and has been declining since the mid-1990s. This decline could be balanced by the fast development of offshore fields, which now account for almost 50 per cent of the production of all countries in this group. The North Sea alone has a share of almost 40 per cent of the total offshore production within this group. The peaking of the North Sea was decisive because the production decline could not be compensated any more by a timely connection of new fields in the remaining regions – it was only possible to hold the plateau for a few years. Crucial for the further development was the production peak of Cantarell in Mexico, the world’s biggest offshore field. This field, discovered in 1978, even today contributes one half of Mexican oil production. It reached a plateau for some years and started to decline in 2005. Furthermore, the quality of the oil produced in Mexico has degraded steadily; the share of light oil has halved since 1997. This steady degradation of the quality of the oil produced can be observed in almost all regions which have passed their peak and poses an additional challenge for the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

10

J. SCHINDLER AND W. ZITTEL

Mb/day history

40 35 30

Ecuador 99 Colombia 99 Argentina 98

25 20 15 10 5

Mexico 04

Norway 01 Oman 01 Australia 2000 United Kingdom 99

Malaysia 97 Gabon 97 Syria 95 India 95 Egypt 93 Indonesia 77 NGL Romania 76 Alaska Canada (conv.) 74 Rest-USA 7 1 Germany 67 Austria 55 Texas 71

1900 10

20

30

40

50

60

70

80

Yemen Neutral Zone Brazil Angola China

89 GOM

90 2000 10

Source: IHS 2003, BP Stat Rev 2005: Denmark, UK, Canada, Norway, Alaska, USA, Mexico, Brazil, Argentina: 2005 estimate based on Jan-Nov data from government statistics, Analyses and Forecast LBST

FIGURE 1.3 Oil production of countries outside OPEC and the FSU

existing downstream infrastructures: refineries have to operate with oil of decreasing quality. The share of lower oil qualities is steadily increasing – this will additionally drive upwards the prices for the remaining good oil grades. Particularly interesting is the example of Indonesia, the only OPEC member state which is included in this group of countries as it will probably soon leave OPEC, because in March 2004 for the first time more oil was imported in this country than exported. Oil production in regions having passed their peak can be forecasted with some certainty for the next 10 years. Even if it is assumed that the remaining regions with growth potential (Angola, Brazil and the Gulf of Mexico) will considerably expand their production by 2015 (in accordance with the forecasts of the companies operating in these regions), total oil production for this group of countries will decline by 10–15Mb/day by 2015. As the production of conventional oil is declining, this group of countries will be able to supply additional amounts only from non-conventional sources. Non-conventional oil sands in Canada and Venezuela will contribute 3–4Mb/day in 2015, provided that the already announced expansion plans will be realized without any further delay.

1.2.1.3 The former Soviet Union Oil production in the FSU peaked reaching a production rate of more than 12Mb/day at the end of the 1980s. Production then collapsed by almost 50 per cent within five years. The production peak at the end of the 1980s had been forecasted by Western geologists based on the depletion patterns of the largest oil fields. However, the following production

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

Alternative World Energy Outlook 2006

11

Mb/day 5

history

4 Bitumen (4Gb)

3 Synthetic crude oil (6.7Gb)

2

Sasketchwan Alberta (5.5Gb)

1 0

Total conv New Foundland oil production (18Gb) (1.4Gb)

1960

1970

1980

1990

2000

2010

2020

2030

Tarsands (1 0.9 >8

~0.3 0.3 >0.3

3.6 2.7 3.6

1.9 2.1 3

6.8 2.5 6.8

0.1 0.24 1

0.6 1.1 15.3

>1 1.2 >10

~0.1 0.1 >0.1

3.8 4.0 4.2

1.2 1.3 1.9

2.6 1.5 2.8

0.1 0.4 0.3

2.3 2.2 5.2

>1 1 >3

~0.2 0.2 >0.2

10.6 6.0 10.6

3.5 3.5 5.6

17

~0.1 0.3 >0.1

4.6 4.2 4.6

2.2 2.4 3.6

17 1 0.4 1 >4 6 >27

~0.5 1 >0.5 ~0.8 0.5 >0.8 ~0.6 0.3 >0.6 ~0.3 1 >0.3

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

42

J. SCHINDLER AND W. ZITTEL

Annex II: Simulation Parameters for Renewable Energy Scenarios ELECTRICITY Hydro Wind PV OECD – North America P (GW) 400 2010 T0 (year) b (years) 30 OECD – Pacific P (GW) 90 T0 (year) 2020 b (years) 30 OECD – Europe P (GW) 300 T0 (year) 2020 b (years) 65 Transition economies P (GW) 700 T0 (year) 2040 b (years) 20 China P (GW) 500 T0 (year) 2030 b (years) 17 East Asia P (GW) 360 T0 (year) 2039 b (years) 16 South Asia P (GW) 250 T0 (year) 2031 b (years) 15 Latin America P (GW) 700 T0 (year) 2040 b (years) 24 Middle East P (GW) 80 T0 (year) 2052 b (years) 20 Africa P (GW) 500 T0 (year) 2063 b (years) 20

SOT

Geo

Biomass1

HEAT Geo2

Solar3

2500 2040 6.5

2900 2055 5.7

500 2050 6.2

50 2030 11

5000 2030 20

750 2050 10

300 2045 20

900 2040 5.5

1500 2036 4.5

1000 2060 10

30 2035 10

100 2045 12

600 2046 6.5

110 2026 12

1500 2023 5

1100 2035 4.4

600 2050 5.5

50 10 2040

200 2025 22

500 2040 10

250 2027 10

2000 2060 5.7

3500 2070 5

0

100 2050 6.5

300 2037 11

1000 2060 8.4

300 2050 3.5

1500 2040 4.8

1600 2045 4

0

3 2039 7

200 1950 40

300 2040 8

1000 2020 5.5

1000 2060 5.4

770 2050 4

0

100 2047 12

100 1970 11

2000 2060 5

500 2040 6

1000 2050 8

900 2040 3.8

100 2035 3.5

1 2030 8

150 1990 7,5

2 2035 15

400 2030 6

1500 2045 4.5

2000 2060 4.5

50 2040 4

100 2070 12

600 2020 10

3000 2065 7

1000 2030 6

30 2050 6

680 2050 3.8

3000 2060 5.5

6 2040 10

10 2020 13

30 2050 10

100 2035 7

3000 2060 6

4000 15,000 2090 2070 6.8 5.5

60 2060 9

500 2005 16

200 2060 8

1000 2045 7

1 Potential in Mtoe 2 Potential in GW thermal energy 3 Potential in million m2

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

Alternative World Energy Outlook 2006

43

Annex III: Growing Share of Renewable Energy Since 1990

Year For reasons of comparison the installed collector area is converted into power by the factor 0.7kW per m2. The data have been collected over the years by the LBST from various different statistics, including data from Notes 43–49. Though the absolute share of photovoltaics is very small, its growth rate is accelerating, exceeding 40 per cent over recent years. Its absolute contribution is lagging about ten years behind wind energy.

FIGURE 1.A.1 Cumulative installed capacity of wind energy converters, geothermal power plants, photovoltaic modules and solar collectors

~9–10% p.yr.

Year Photovoltaics, geothermal, wind and biomass are restricted to electricity generation. For reasons of comparison the heat generation from solar collectors is also included. Not included are contributions from hydropower or heat from geothermal sources and biomass. Fuels production from biomass is also not included. The data have been collected over the years by the LBST from various statistics including data from Notes 43–49.

FIGURE 1.A.2 Energy production from renewable sources

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

44

J. SCHINDLER AND W. ZITTEL

Wind power Biomass (incl. traditonal use)

Year Year Biomass includes so-called traditional use in developing countries. Electricity from hydropower and other sources is directly converted into Mtoe. The share of other renewables is relatively small, but exhibits by far the largest growth rates. The data have been collected over the years by the LBST from various statistics including data from Notes 43–48. The conversion from capacity to energy production was performed by the LBST.

FIGURE 1.A.3 Primary energy from renewable sources

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 1–44

2

Quantum Well Solar Cells N. J. Ekins-Daukes

Abstract The quantum well solar cell represents the highest efficiency nano-structured solar cell demonstrated to date. The concept and practical constraints for this cell are discussed together with its potential application in high efficiency multi-junction solar cells and thermophotovoltaic devices. Finally the ultimate efficiency limit of the quantum well solar cell is discussed.

■ Keywords – quantum well; uncertainty principle; density of states; recombination; voltage enhancement; thermophotovoltaics; electroluminescence; entropy

2.1 HISTORY Since its inception, the quantum well (QW) has been applied widely in semiconductor devices. Its roots stem from the resonant tunnel diode and it has found applications in a large number of electronic devices, such as photodetectors, light-emitting diodes, lasers, optical modulators and high mobility transistors. The almost universal utility of QWs is due to the ability to engineer the optical and electronic properties of a QW structure to suit a particular application. It is then reasonable to suppose that some advantages may exist when using QWs in photovoltaic structures. Early work (Chaffin et al, 1984) suggested that a superlattice could provide a means of achieving highly efficient multi-junction solar cells, allowing the constituent band gaps to be matched to the solar spectrum, without recourse to lattice mismatched techniques or complex quaternary materials. However, the most decisive step in bringing QWs to photovoltaics was made in the early 1990s with the proposal of the multiple quantum well (MQW) solar cell (Barnham and Duggan, 1990). Here it was suggested that the QWs should be located in the intrinsic region of a pin diode and the proposal was quickly followed by a successful experimental demonstration in the AlGaAs/GaAs material system (Barnham et al, 1991) showing a considerable increase in short-circuit current (Jsc). The ability to adjust the absorption profile of a bulk semiconductor through the inclusion of QWs gives rise to a number of practical, near-term applications for QW solar cells, such as a component of a multi-junction solar cell, or in thermophotovoltaics where low band-gap materials are required.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

46

N. J. EKINS-DAUKES

QWs and other low dimensional structures also offer some possibilities for testing unconventional concepts in photovoltaics. For example, the proposal by Barnham and Duggan also predicted that some advantage in the open-circuit voltage (Voc) could be attained over bulk semiconductor material. This aspect has stimulated an interesting debate over what electronic properties are necessary to achieve a photovoltaic voltage enhancement in low dimensional semiconductors. Many excellent review papers have already been written on lattice matched QW solar cells (Barnham et al, 2002; Nelson, 2001 and 1995) and are accompanied by a large volume of journal papers and conference proceedings. To distinguish this review paper from earlier ones, the role of strain-balance is emphasized and is presented as a means of broadening the range of materials that can be usefully applied in QW photovoltaic devices.

2.2 QUANTUM WELL ELECTRONIC STRUCTURE Over the past 20 years an extremely large volume of literature has been published on the properties of QWs. Most pertain to devices other than solar cells, but the general properties of the QW remain the same, irrespective of the application. A number of books have now been published on the physics of QW devices (Weisbuch and Vinter, 1991; Bastard, 1988) and one book has a particular emphasis on QW solar cells (Barnham and Vvedensky, 2001). What follows is a brief description of the properties of QWs and the reader is referred to the detailed texts above for a full discussion. In essence, a QW is a thin layer of material possessing a lower band gap than the surrounding barrier or host material. The QW must be sufficiently thin to bring about quantum effects through the confinement of electrons and holes, so its dimensions are of the order of the electron and hole de-Broglie wavelength, typically a few nanometres wide. The structural properties of the QW should be similar to the barrier, meaning that there is no discontinuity in crystal symmetry and ideally the interfaces should be abrupt. Under these conditions, the QW potential can be treated as a perturbation of the crystal potential, allowing delocalized states to extend across the two materials. The properties of these states closely resemble those of a quantum 1D potential well, and they can be described in an analogous fashion using the envelope function approximation (Barnham and Vvedensky, 2001). Figure 2.1a shows how the QW and barrier layers are physically stacked. The resulting electronic structure when the QWs are made sufficiently thin is shown in Figure 2.1b. The envelope functions for each confined energy level are shown for two bound electron and hole levels in each QW. The lowest bound level in the QW does not lie at the bottom of the QW, but exists at an elevated energy. This is a direct result of confining the electron and hole and can be simply understood through Heisenberg’s uncertainty principle 6 x 6 p * h/2. Decreasing the well width, 6 x increases the uncertainty in the momentum 6p and hence increases the energy of the lowest bound level. The absorption edge of the QW can therefore be controlled simply by varying the well width. The parity of the confined levels is well defined and therefore restricts the allowed optical transitions, so for example the e1h1 will be a strong transition, while e1h2 is weak. As the electron and hole are only confined in one dimension, the carriers are free to move in a 2D plane, giving continuous in-plane momentum dispersion curves for each bound level. This is illustrated in Figure 2.1c

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

47

k||

FIGURE 2.1 a) Layer structure showing three QWs surrounded by barrier material; b) Band diagram for three QWs, showing two confined states for electrons and holes; c) Energy vs in-plane momentum dispersion relation for a QW, showing two confined electron levels (e1 and e2) and the first heavy hole (hh1) and the first light hole (lh1) levels

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

48

N. J. EKINS-DAUKES

for two bound electron levels (e1 and e2) and the first heavy and light hole levels. The degeneracy in the hole levels at k=0 is lifted when the QW is strained (Barnham and Vvedensky, 2001). The quantum confinement of carriers also serves to restrict the density of states. Figure 2.2 compares the step-like behaviour of the QW 2D density of states to the continuous 3D density of states; the expressions for the 2D and 3D densities of states are expressed below:

m∗ π¯ h2

(1)

1 2m∗ 3 √ ( 2 )2 E 2π 2 ¯ h

(2)

g2D (E) =

g3D (E) =

where m* is the carrier effective mass. It is clear that in the 2D case, the density of states is independent of energy for each bound level and depends only on the effective mass. It is this simple dependence of the QW density of states that gives rise to the many applications for QWs in electronic devices. For example, the strength of optical transitions is governed by the density of states, so materials can be chosen to optimize the oscillator strength of a particular transition, and the well width can be adjusted to achieve the desired transition energies. Finally, the number of bound states in a QW is of particular importance for the QW solar cell, as each bound level adds further states from which absorption can

FIGURE 2.2 Density of states for a 3D bulk semiconductor g3D and for a 2D quantum well structure g2D

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

49

take place. The number of bound states in a QW is given by the following expression (Bastard, 1988):



n(L) = 1 + Int[

2m∗ Vb L2 ] π 2 ¯h2

(3)

where Vb is the potential of the confining barrier, L is the well width and Int returns the lowest integer value of the argument. The role of the effective mass is again important and can lead to situations, such as in an Al0.3GaAs/GaAs QW, where the band offset gives a deeper well for the electrons than for holes, yet the larger hole mass more than compensates for the band offset, giving more heavy hole bound states than bound electron states. When a series of QWs is grown, if the barriers are thick, such that only a negligible fraction of each QW wave function penetrates into the neighbours, then each QW can be considered isolated and the structure is termed a multiple QW (MQW). The properties of the MQW can be considered simply as a linear multiple of single QWs. However, when the barriers separating each well are sufficiently thin to allow sufficient overlap of the wave functions from adjacent wells, then the QWs become coupled and a superlattice is formed (Dingle et al, 1975). Under these circumstances, the energy levels characteristic of an isolated QW broaden to form mini-bands that extend throughout the superlattice. There is no clearly defined point at which an MQW becomes a superlattice, as narrowing of the barriers leads to an ever increasing broadening of the confined levels. Strictly it is only by defining an arbitrary bandwidth that the transition from MQW to superlattice can be determined. However, for most structures discussed in this review, the thermal broadening of the confined levels at room temperature is greater than any superlattice effects, so the structures can be safely called MQW devices. Some consideration is given to superlattices in section 2.3.4. Before concluding this section, it is important to note that the QW band alignment considered above is one where both the electron and hole are confined and is called a type-I alignment. On occasions, the band alignment of the QW and barrier material are such that only the electron or hole is confined, but not both, and the QW is then said to be type-II. To date only type-I QW arrangements have been pursued for photovoltaic applications as the optical transitions are generally stronger. Nevertheless, type-II band alignments have some desirable properties, such as the suppression of Auger recombination (Tang et al, 1995), and have been exploited for optoelectronic applications (Gevaux et al, 2001).

2.3 BASIC OPERATION OF THE P-I-N QUANTUM WELL SOLAR CELL The QW solar cell proposed by Barnham and Duggan (1990) is shown schematically in Figure 2.3a. The device consists of a p-i-n structure with the QWs grown in the i-region. Epitaxial techniques such as molecular beam epitaxy (MBE) and metal organic vapour phase epitaxy (MOVPE) can be used to fabricate QW structures with monolayer accuracy (Barnham and Vvedensky, 2001; Stradling and Klipstein, 1989). Figure 2.3b shows the band diagram for the QW solar cell and shows the main carrier photogeneration and recombination paths in the barrier and QW layers.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

50

N. J. EKINS-DAUKES

FIGURE 2.3 a) QW p-i-n layer structure; b) Band diagram for the QW p-i-n solar cell showing the photogeneration and recombination processes, together with the carrier capture and escape routes

2.3.1 PHOTOGENERATION AND CARRIER ESCAPE For the QWs to contribute to the photocurrent, the carriers must escape from the QW through a combination of thermal and tunnelling processes (Nelson et al, 1993); this is shown schematically in Figure 2.3b together with the corresponding carrier capture process. The carrier escape is efficient providing that there is sufficient thermal energy (Barnes et al, 1996; McFarlane et al, 1999) and the transverse electric field is sufficiently strong (Nelson et al, 1993; Serdiukova et al, 1999). In most cases studied to date, there is sufficient thermal energy at room temperature for the carriers to escape the QW and in the presence of a strong electric field the probability for escape is unity under short-circuit conditions. An example of efficient photogenerated carrier collection is shown in the quantum efficiency (QE) profile in Figure 2.4 for the GaAsP/InGaAs MQW p-i-n device whose structure is depicted in Figure 2.3. The GaAsP barriers are responsible for the majority of the QE from 400 to 860nm and the InGaAs QW extends the solar cell response from 860 to 920nm. In this example only one electron level is bound and the first transition in the step-like density of states is obscured by the presence of exciton levels giving rise to sharp peaks in the absorption. These exciton levels persist at room temperature, due to the increased exciton binding energy in QW layers (Barnham and Vvedensky, 2001). It is clear that, apart from a modest Stark-shift in the QW QE (Monier et al, 1999), the collection efficiency remains high from short-circuit through to forward bias. Nevertheless, if the background doping is too high, then the field across the QWs becomes insufficient to remove carriers efficiently and the carrier collection collapses, both in the bulk and QW. A spectacular example of this is demonstrated in Figure 2.5, showing bias-dependent QE for a strain-balance structure composed of Al0.04GaAsP0.06/

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

51

FIGURE 2.4 Typical quantum efficiency for a GaAsP0.06/In0.1GaAs 75Å, 35 MQW structure at various biases

InGaAs. The inclusion of just 4 per cent Al into the GaAsP barriers introduces sufficient background impurities1 to give an almost negligible spectral response in forward bias. Similar results are observed if the QWs are placed in the doped regions of the device, at which point the diffusion length is reduced to the order of the barrier width (Ekins-Daukes, 2000; Barnham et al, 2002). There are also instances where there is insufficient thermal energy to allow efficient carrier escape from the QWs, for example in InP/InGaAs QWs (Zachariou et al, 1998a) and in particularly deep strain-balanced InGaAs/InGaAs QWs (Rohr et al, 2002a). However, if the electric field is maintained across the QWs and the temperature is sufficiently high, then efficient carrier escape is maintained beyond the operating point for most MQW solar cell devices. This is illustrated in the case of the GaAsP/InGaAs device by the light IV curve shown in Figure 2.6. The temperature coefficient for MQW solar cells has been found to be marginally better than for equivalent p-i-n solar cells made from the well and barrier

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

52

N. J. EKINS-DAUKES

FIGURE 2.5 Quantum efficiency for an Al0.04GaAsP0.06/In0.1GaAs 75Å, 35 MQW showing how background impurities lead to a collapse of the QE at forward bias

material (Ballard et al, 1998 and 2001). Historically the AlGaAs/GaAs material system was the first material system in which the p-i-n QW solar cell was demonstrated (Barnham et al, 1991) and was followed by other lattice matched material systems, such as InGaP/GaAs (Osbourne, 1994; Zachariou et al, 1998b), InP/InGaAs (Freundlich et al, 1994; Zachariou et al, 1996; Zachariou, 1996) and InGaAsP/InP (Rohr et al, 1998 and 1999; Raisky et al, 1998; Rohr et al, 2000a; Raisky et al, 2001). A number of strained material systems have also been investigated, for example GaAs/InGaAs (Barnes et al, 1994; Ragay et al, 1994a; Barnes et al, 1996; Barnes, 1994), InP/InAsP (Freundlich et al, 1994), InP/InGaAs (Serdiukova et al, 1997; Freundlich, 2000) and also some strain-balance structures based on GaAsP/InGaAs (Ekins-Daukes et al, 1999), GaAs/InGaAs (Ekins-Daukes, 2000; EkinsDaukes et al, 1998), InGaAs/InGaAs (Rohr, 2000; Rohr et al, 2000b and 2002b). The strained and strain-balance materials are discussed further in section 2.3.3.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Current / mA·cm–2

Quantum Well Solar Cells

J

SC=32.1

53

mA·cm–2

Voc = 09.55V FF = 82.9% Eff=18.7%

Bias/V Source: Bushnell (2002)

FIGURE 2.6 Light IV curve for a GaAsP0.06/In0.1GaAs 75Å, 35 MQW structure under approximately 1 sun AM0 equivalent 3000K tungsten halogen illumination

Despite the complexity of the absorption profiles of the MQW solar cell, it has been possible to model the spectral response accurately using the envelope function approximation outlined earlier (Paxman et al, 1993; Renaud et al, 1994; Nelson, 1995). Computer modelling has enabled advanced cell structures to be designed, featuring compositionally graded layers and back surface mirroring, to improve minority carrier transport and optical absorption (Connolly, 1997; Connolly et al, 1998).

2.3.2 CARRIER RECOMBINATION AND DARK IV Recombination takes place across both the barrier and QW material and can be understood using standard recombination models for radiative and non-radiative SRH recombination (Nelson, 2003). Experimentally, it is generally found that the Voc of a lattice matched QW solar cell is lower than that of a control cell made of the barrier material, but higher than that of a control cell made of the well material. If a comparison is made between the Voc of barrier and well p-i-n control cells grown in the same MOVPE or MBE machine, one observes that the QW cell has a marginally higher Voc than for a bulk cell of equivalent absorption edge (Barnham et al, 1996 and 2002), estimated from a linear interpolation between well and barrier p-i-n control cells. This effect is commonly called voltage enhancement, but the term sometimes leads to confusion; it is the voltage that is enhanced with respect to a bulk p-i-n cell composed of the well material, not the barrier material.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

54

N. J. EKINS-DAUKES

To give a broad indication of the recombination behaviour that can be expected from a QW solar cell, Figure 2.7 shows values of J0 plotted against the cell absorption edge for a large number of solar cells, with band gaps spanning the solar spectrum. A theoretical estimate of the radiative limit is also plotted, showing the lower limit for J0, assuming unity absorptivity (Shockley and Queisser, 1961).

Ln(J0)

FIGURE 2.7 Plot of J0 vs absorption edge for a variety of lattice matched, mismatched and MQW devices

The values of J0 have been extracted by a simple process, taking experimental values for Jsc and Voc and assuming that the solar cell behaves as an ideal diode. Under these conditions J0 is estimated as follows:

Jsc = J0 e

qVoc kT

−1

(4)

where q is the electronic charge, k is Boltzmann’s constant and T is the device temperature. Extracting J0 in this way only gives an approximate means of comparing PV devices. In particular, it assumes that the fill-factor is recombination limited and that radiative recombination (unity diode ideality) dominates. Under 1 sun illumination, most QW and p-i-n solar cells have a diode ideality of two, whereas good quality p/n cells usually have an ideality of one. Such differences in ideality have been neglected here, as the comparison in J0 is made across many orders of magnitude. Light IV data has been taken from a large number of sources: AlGaAs/GaAs MQW (Paxman, 1992), GaAsP/InGaAs MQW (Bushnell, 2002), InGaAs/InGaAs MQW (Connolly and Rohr, 2003),

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

55

InGaAsP/InGaAsP MQW (Connolly and Rohr, 2003), bulk GaInP (Yang et al, 1997), bulk GaAs (Tobin et al, 1990), lattice mismatched InGaAs (Takamoto et al, 2000; Hoffman et al, 1998), bulk GaInNAs (Friedman et al, 1998; Li et al, 2000), bulk In0.53GaAs (Takamoto et al, 1997), bulk InGaAsP (Takamoto et al, 1997) and bulk Ge (Nagashima et al, 2003). It is clear from Figure 2.7 that similar performance is observed from most of the MQW structures as bulk cells. Of particular interest are the low band-gap MQWs (Connolly and Rohr, 2003) that show exceptionally low recombination, in comparison with the neighbouring bulk p/n cells, making the cells attractive for thermophotovoltaics, as discussed in more detail in section 2.4.2. It is also noteworthy that the strain-balanced GaAsP/InGaAs cells (Ekins-Daukes et al, 1999) have a similar J0 to the lattice mismatched InGaAs cells, leading to possible applications in multi-junction solar cells, discussed in section 2.4.1. When attempting to model the Voc or dark IV of an MQW cell, it is usually assumed that the MQW system is sufficiently close to quasi-thermal equilibrium to allow electron and hole quasi-Fermi levels to be used to describe the carrier populations in the QW. The extent to which quasi-thermal equilibrium is established in an MQW is discussed in section 2.5, but nevertheless, the dark currents in a number of QW systems have been modelled accurately (Nelson et al, 1995; Nelson, 1995) including the effects of asymmetric QW location in the i-region (Nelson et al, 1999). The success of computer modelling of MQW structures has led to more general MQW device optimization studies (Connolly et al, 2000 and 2001). While actual device performance clearly depends on a large number of material parameters, these models now have a certain degree of predictive power that make them useful for assessing suitable MQW device designs.

2.3.3 STRAINED AND STRAIN-BALANCED QW SOLAR CELLS Changing the QW width allows the absorption threshold to be adjusted to higher energies, but there are situations where it is desirable to lower the absorption threshold. When using GaAs or InP barriers, these lower band-gap materials are usually strained, so for example InGaAs is a particularly good choice for devices based on GaAs (Barnes et al, 1994; Ragay et al, 1994a; Barnes et al, 1996; Barnes, 1994) and InP (Serdiukova et al, 1997; Freundlich, 2000; Freundlich et al, 1994; Freundlich, 1998). As an example, a device composed of compressively strained InGaAs QWs and GaAs barriers grown on a GaAs substrate is considered; the associated layer structure and band diagram is shown in Figure 2.8. Note that the relaxed lattice parameter is indicated laterally, illustrating the compressive strain required to grow the InGaAs QWs on the GaAs lattice. In Figure 2.9, the quantum efficiency and dark IV characteristics for two strained QW GaAs/InGaAs solar cells and a GaAs p-i-n control cell are shown. The increased photocurrent due to the QWs is evident from 900nm to 1000nm; however, this is accompanied by a very large rise in dark current due to defects introduced through strain relaxation taking place at the top and bottom of the MQW stack (Griffin et al, 1996; Mazzer et al, 1996). In the case of the 23 QW cell, the strain relaxation is so great as to degrade the short wavelength QE. Unless measures are taken to passivate the defects associated with the strain-induced misfit dislocations (Okada et al, 2002; Suzuki et al, 1999), the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

56

N. J. EKINS-DAUKES

FIGURE 2.8 Strained GaAs/InGaAs 3period MQW device: a) Schematic diagram showing the layer configuration; the relaxed lattice parameter is indicated laterally; b) Electronic band-structure for the device

effects of strain relaxation are so severe that strained GaAs/InGaAs structures can never offer a substantial increase in current over a GaAs cell without suffering a catastrophic loss in voltage (Ekins-Daukes et al, 2001). In the strained GaAs/InGaAs structure, strain builds up with each InGaAs layer. However, it is possible to minimize the build-up of strain by incorporating alternate tensile and compressively strained layers. Through an appropriate choice of alloy composition and layer thickness, and accounting for differences in elastic constant, it is possible to achieve structures which are locally strained but exert no net force on the substrate or neighbouring repeat units. The layer thickness required to achieve a strain-balanced structure can be estimated using the following expression (Ekins-Daukes et al, 2002):

a0 =

A1 t1 a1 a22 + A2 t2 a2 a21 A1 t1 a22 + A2 t2 a21

(5)

where a0 is the substrate lattice parameter, a1 and a2 are the respective QW and barrier lattice parameters, t1 and t2 the respective QW and barrier layer thicknesses, and A1 and A2 constants for the QW and barrier and defined through the elastic stiffness coefficients Cxx as follows:

A = C11 + C12 −

2 2C12 C11

(6)

In principle, the critical thickness of a strain-balanced structure is infinite. In practice, however, there will always be some shortcomings, both in the growth and design of a

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

57

FIGURE 2.9 a) External quantum efficiency for a strained GaAs/InGaAs MQW sample and GaAs control; b) dark IV curves for GaAs/InGaAs strain-balance MQW cell, a relaxed GaAs/InGaAs MQW cell, a strained GaAs/InGaAs MQW cell and a GaAs control

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

58

N. J. EKINS-DAUKES

strain-balance stack. The layer thicknesses determined from Equation 5 are based on linear elastic theory and Vegard’s law (Vegard, 1921) is usually used to relate the material composition to lattice parameter. These are good approximations, but will become increasingly inaccurate at high strain. Finally, the material parameters used to determine the strain-balance condition are usually those measured at room temperature, yet the device is grown at an elevated temperature (up to 650°C), and differences in thermal expansion between the component layers ensure that no structure can be exactly strainbalanced at both the growth temperature and room temperature. A typical strain-balanced structure is shown in Figure 2.10. Here GaAsP is used as the tensile material to strain compensate the compressive InGaAs layer. Such strain-balanced structures have demonstrated unprecedented photovoltaic performance for QW solar cells (Ekins-Daukes et al, 1999). The quantum efficiency and dark IV are shown in Figure 2.11. The quantum efficiency plot indicates the additional absorption due to the QWs over a GaAs p-i-n control sample. The dark current profiles show excessively high rates of recombination for the strained and relaxed GaAs/InGaAs cells, but the strain-balanced cell has a dark current profile comparable to the GaAs p-i-n control cell. Therefore, by using the strain-balanced QW approach, strained, lower band-gap semiconductors can be incorporated into a solar cell with no additional misfit dislocations and can therefore maintain a high Voc. A remaining challenge is to increase the absorption of the MQW stack, as the 20 MQW sample only has a QE of approximately 20 per cent due to the overall low optical thickness of QW absorbing material. This could be solved by simply growing more QWs into the structure, and the current state-of-the-art device is a 50 MQW GaAsP/InGaAs device (Bushnell et al, 2003b). However, one cannot increase

FIGURE 2.10 Strain-balance GaAsP/InGaAs 3period MQW device: a) Schematic diagram showing the layer configuration; b) Electronic band structure for the device

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

59

FIGURE 2.11 a) External quantum efficiency for the strain-balanced GaAsP/InGaAs MQW sample and GaAs control; b) Dark IV curves for GaAsP/InGaAs strain-balance MQW cell, a relaxed GaAs/InGaAs MQW cell, a strained GaAs/InGaAs MQW cell and a GaAs control

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

60

N. J. EKINS-DAUKES

the number of QWs indefinitely. First, increasing the number of QWs places an ever tighter tolerance on the growth process (Ekins-Daukes et al, 2002), so a growth technique capable of producing a 20 QW structure may not be suitable for a 100 QW sample. Second, the necessity to maintain a field across the QW for efficient carrier escape places a stringent demand on the background doping level in the i-region. The devices discussed above typically have i-regions approximately 1μm wide, which through a simple electrostatic calculation (Ekins-Daukes, 2000) requires a background doping level below 1 × 1015cm-3. If the background doping level rises to a point that the MQW is no longer fully depleted, then the catastrophic loss in photocurrent observed in Figure 2.5 is experienced. Third, the strained QW and barriers must not exceed their individual critical thickness (Matthews and Blakeslee, 1974), thereby placing some limits on the well width and number of bound levels in the QW. Indeed, even when the well and barrier remain below their critical thickness, strain-induced composition modulation can degrade the MQW under certain growth conditions (Bushnell et al, 2003b; Bushnell, 2002). Such problems lead one to conclude that MQW devices could benefit from light-trapping schemes (Yablonovitch and Cody, 1982; Steinke, 1996; Hepp, 1998). One effective technique has been to incorporate a distributed Bragg reflector (DBR) into the structure (Bushnell et al, 2003a), thereby doubling the optical path through the QW layers. To approach unity absorption, however, the optical path length needs to be extended further. Textured coatings and grating structures have been proposed to increase the optical path through the QWs and some practical means for their formation investigated (Bushnell et al, 2002; Bushnell, 2002).

2.3.4 SUPERLATTICE STRUCTURES In comparison to the MQW, relatively little experimental work has been performed on applying superlattices for photovoltaics, but nevertheless there are some interesting opportunities in this area. While the superlattice maintains the advantages offered by MQWs for engineering the band structure, it also allows transverse carrier transport within the mini-bands, alleviating the need for carriers to escape the QWs. The early work by Chaffin (Chaffin et al, 1984; Chaffin, 1987) proposed using superlattices in multijunction solar cells to obtain an optimal match to the solar spectrum. In this proposal the superlattice was doped to form a p/n junction and this concept is currently being pursued in an attempt to create an all silicon Si/SiO2 multi-junction solar cell (Green, 2000). Superlattices have also been proposed to improve transport when working with difficult materials (Varonides and Berger, 1997; Kojima et al, 2000) and in particular to provide a 1eV junction for a multi-junction solar cell using InGaAsN (Bedair et al, 2000).

2.3.5 RADIATION RESISTANCE OF QW SOLAR CELLS The radiation resistance of InP/InAsP p-i-n QW solar cells has been investigated, comparing the MQW against an InP p-i-n control cell (Walters et al, 2000a). The Jsc of the QW is more robust against 3MeV protons than the InP pin control cell. However, this advantage appears to be offset by a greater loss in Voc in the QW cell over the control, leading to approximately the same power degradation in both cells. A detailed study

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

61

showed that the proton irradiation introduced a mid-gap trap that is likely to be associated with the QW or QW Barrier interfaces (Walters et al, 2000b and 2001). Further work on the radiation resistance of MQW solar cells is required, especially on GaAs-based QW devices, which could usefully enhance the GaInP/GaAs/Ge multi-junction space solar cell, discussed in section 2.4.1.

2.4 NEAR-TERM APPLICATIONS FOR QW SOLAR CELLS Due to the high cost of launching a spacecraft, typically US$70,000/kg, the space photovoltaic industry is able to use expensive and specialized photovoltaic technology, with the primary aim of achieving a high specific power (power per unit mass). In general, the rigid panel structures on which the PV cells are mounted dominate the specific power, so the panel specific power scales with the efficiency of the solar cell. This makes high efficiency cells such as the GaInP/GaAs/Ge triple junction particularly attractive for space use (Fatemi et al, 2000). QW solar cells could therefore find practical applications as a component of a multi-junction space solar cell (Freundlich and Serdiukova, 1998; Freundlich, 2000 and 2002; Ekins-Daukes et al, 2000) and this application is discussed in more detail below. The terrestrial photovoltaic industry is continually looking for lower cost materials in an attempt to bring down the cost of photovoltaic power generation. Nevertheless, there are opportunities for highly specialized devices such as QW solar cells in high concentration photovoltaics, where at sufficiently high concentration the cell cost becomes a negligible component of the overall concentrator system cost (Yamaguchi and Luque, 1999). QWs have been considered for concentrator applications (Barnham, 1992) and, as for the multi-junction space solar cell, QWs hold some advantages for a concentrator multi-junction cell (Tibbits et al, 2003). In general, any photovoltaic application where precise tuning of the absorption edge is required will lend itself to the use of QW structures. Examples include thermophotovoltaics (Connolly and Rohr, 2003) (discussed below) or wavelength splitting schemes such as the luminescent concentrator (Barnham et al, 2000).

2.4.1 MQW MULTI-JUNCTION SOLAR CELLS There have been plenty of proposals for using MQWs in multi-junction solar cells, but it is only recently that experiment has shown results that could be promising for multi-junction applications. A possible near-term application is in the InGaP/GaAs solar cell, where it is well known that the cell could be improved if the lower cell junction had a band gap somewhat lower than that of GaAs. The problem can be overcome to some extent if the InGaP layer is thinned, in order to match the photocurrents (Kurtz et al, 1990), but the cell still operates with a compromised voltage. There is then an opportunity to engineer a material using the strain-balance approach that could provide an ideal low band gap bottom junction for a tandem solar cell. Figure 2.12 shows how the efficiency of a two terminal tandem solar cell varies as a function of the top and bottom cell band gaps. The extent to which the existing InGaP/GaAs is not optimally matched is clear, as is the efficiency enhancement if the lower cell possessed an absorption edge around 1.3eV. This approach can be considered an alternative to the popular bulk lattice mismatched layers

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Upper Cell Band Gap/eV

62

N. J. EKINS-DAUKES

Monolithic tandem lso-Efficiency Contours One Sun; AM 0 Illumination Temperature = 300K

Lower Cell Band Gap/eV FIGURE 2.12 AM0 iso-efficiency contour plot for a monolithic tandem structure; the calculation is described in Ekins-Daukes et al (2000)

(Dimroth et al, 2000), the advantage being that strain relaxation and the associated degradation in material quality due to dislocations are avoided altogether. To give an indication of the performance of these strain-balanced structures in a tandem solar cell, Table 2.1 shows the projected Voc for both the lower cell and the tandem structure, together with the Jsc and the tandem efficiency values. The calculation is based on published data taken from a GaInP top cell (Yang et al, 1997) and a variety of MQW candidates for the lower junction (Bushnell et al, 2003a; Ekins-Daukes, 2000). Values for J0 are estimated from Jsc and Voc as described earlier, using equation 4. The GaInP junction is modelled with J0 =5 × 10–23mA˙cm–2 and can deliver 20mA˙cm-2, making the 2J cell current limited by the lower junction in all cases. The GaInP/GaAs efficiency provides the baseline for comparison at 26 per cent and is in reasonable agreement with experimental values (Takamoto et al, 2001). An improved performance is achieved with both the GaInP/SB MQW and the GaInP/DBR SB-MQW, the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

63

TABLE 2.1 AM0 Junction parameters for a GaInP top cell together with parameters for lower junction cells composed of GaAs p/n, strain-balanced MQW, DBR enhanced strain-balanced MQW, strained MQW and relaxed MQW; the thick GaInP cell is modelled with J0 = 5 × 10–23 mA•cm–2 and can deliver 20mA•cm–2, making the 2J cell current limited by the lower junction in all cases.

GaAs p/n SB-MQW DBR SB-MQW Strn. MQW Rlxd. MQW

LOWER CELL ONLY J0mA•cm–2 1.7 × 10–16 6.0 × 10–15 6.0 × 10–15 1.9 × 10–13 9.6 × 10–10

V0C/V 1.006 0.920 0.923 0.831 0.611

2J GaInP/ LOWER CELL JSC/mA•cm–2 VOC/V 15.8 2.411 17.1 2.326 18.3 2.328 17.1 2.236 17.1 2.016

EFF. 27.1% 26.1% 28.8% 25.9% 23.2%

latter giving an efficiency in excess of 28 per cent AM0. These figures illustrate the relative importance of an increase in Jsc over a loss in lower junction Voc. The GaInP/strained and GaInP/relaxed MQW cell experiences an efficiency loss, due to the excessive dark current evident from Figure 2.11b. The figures projected in Table 2.1 look encouraging for the MQW cell as a component of a multi-junction solar cell. However, the efficiency estimates are somewhat generous to the MQW in terms of the fill factor. While J0 has been chosen to fit the Jsc and Voc, it has been assumed that the diode ideality is unity, which is reasonable for p/n devices and yields a fill factor around 90 per cent for the structures described in Table 2.1. However, in practice the diode ideality is closer to two for MQW and p-i-n devices under 1 sun illumination. This serves to reduce the fill factor to around 82 per cent and reduces the efficiency advantage for the DBR cell to a negligible margin. Before the introduction of the strain-balance technique, the MQW Voc was the limiting factor in a multi-junction device. Now the strain-balance MQW devices give good Voc, so work should proceed to raise the fill factor and identify the processes responsible for the high MQW diode ideality. In addition, the absorption in the strain-balanced structure is still quite weak, even with the DBR structure, and the QW current could be raised to match the GaInP cell by employing the more sophisticated light-trapping techniques discussed in section 2.3.3.

2.4.2 THERMOPHOTOVOLTAICS In a thermophotovoltaic system the radiative emission from a heat source is collected using a photovoltaic device (Coutts and Fitzgerald, 1998). The heat source is usually followed by a selective emitter, which radiates strongly only over a narrow energy band. It is therefore desirable to match the solar cell to the selective emitter, and for this purpose the band-gap engineering offered by QW structures can be useful (Connolly and Rohr, 2003). The first practical demonstration of a QW thermophotovoltaic cell was based on lattice matched InGaAs QWs on InP (Griffin et al, 1997 and 1998) and showed an absorption edge at 1.55μm. The InGaAsP material system is extremely versatile as it offers a wide range of band gaps that are lattice matched to InP. A particular virtue of this material is that both the QWs and barriers can be grown from InGaAsP. This facilitates the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

64

N. J. EKINS-DAUKES

MOVPE growth process as neither the group III nor group V sources are shut off completely, as is necessary when growing InP/InGaAs structures. InGaAsP MQW devices can be designed to possess an absorption edge between 0.9 and 1.6μm (Rohr et al, 1998, 1999 and 2000a). Such devices are particularly suitable for erbia selective emitters operating at temperatures around 2000K. There are some environmental advantages to lowering the selective emitter temperature to 1500K, and for this purpose an emitter operating at longer wavelengths is required (Abbott et al, 2002). Suitable emitters for this temperature include thulia and holmia, requiring photovoltaic cells that can absorb emission at 1.75μm and at 1.95μm respectively. Strained InGaAs QWs on InP are suitable for these long wavelengths (Serdiukova et al, 1997; Freundlich, 2000) and the same strain-balance technique can be used to ensure that the device is free of dislocations (Rohr, 2000; Rohr et al, 2000b and 2002b). A strain-balanced InGaAs/InGaAs MQW device grown on an InP substrate has recently been demonstrated, with an absorption edge at 2μm (Rohr et al, 2002a).

2.5 EFFICIENCY LIMITS It has already been demonstrated that the QW solar cell has practical advantages in terms of band-gap engineering. However, the issue of whether a QW solar cell has some inherent efficiency advantage over traditional materials, such as GaAs, has been the subject of some debate. In the original proposal, Barnham and Duggan (1990) predicted that the Jsc of a QW solar cell would be enhanced over a solar cell made from the barrier material. As discussed above, this result is well established and marks one of the highly useful aspects offered by QW solar cells. However, they also predicted that the upper limit for the Voc of the QW solar cell would be dominated by the high band-gap barrier material. This prediction was controversial as it implied that the thermodynamically rigorous limit for photovoltaic power conversion established by Shockley and Queisser (1961) could be broken. A vigorous debate ensued which is still ongoing in certain areas. What follows is a brief summary of the various theories that have been proposed and the objections that have been raised. The terminology used here follows a recent review of the arguments for efficiency enhancements in QW solar cells (Anderson, 2002). Briefly, a ‘global efficiency enhancement’ is defined as a theoretical limit for ideal solar cells, where the QW cell outperforms a conventional bulk cell, when both are operating under idealized conditions. A ‘relative efficiency enhancement’ is a practical demonstration of a superior efficiency for a real QW solar cell when compared with a conventional bulk solar cell. The applications for the QW solar cell described in section 2.4 fall into a category defined as ‘ancillary advantages’ and are not of concern to us here. In support of the original paper by Barnham and Duggan, a number of device level predictions have been made indicating a global efficiency advantage (Corkish and Green, 1993; Anderson, 1995; Rimada and Hernandez, 2001a and b) from QW solar cells. In general they all assume that the relationship between absorption and recombination can be broken. However, these device models are found to be thermodynamically inconsistent when carriers are in quasithermal equilibrium in only the conduction and valance band (Araújo et al, 1994; Ragay et al, 1994b; Araújo and Marti, 1995; Luque and Marti, 1997a).

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

65

However, an important feature of the early thermodynamic consistency arguments was that they assumed that the carrier mobility is infinite throughout the device and that the carrier population in the QW is described by the same quasi-Fermi level as the barrier. However, if carriers in the QW are subjected to an additional driving force, that is not in equilibrium with the lattice, then the QWs can possess their own quasi-Fermi levels and the possibility of a global efficiency enhancement emerges once more (Kettemann and Guillemoles, 1995). This concept has been shown to be thermodynamically consistent in a general form, where a third ‘intermediate’ band is postulated from which carriers are excited optically via solar photons into the conduction and valance band (Luque and Marti, 1997b). The operation of an intermediate band QW device (Bremner et al, 1999) is similar to that of a QW infrared photodetector (Levine, 1993). To date a number of mechanisms have been proposed through which a three-band, QW device could show an efficiency enhancement (Green, 2000; Honsberg et al, 2002); this has been extended to cover multiple bands (Brown and Green, 2003; Peng et al, 2003). The subject of thermodynamic consistency (Luque et al, 2001) has again been discussed, as has the detrimental effect of transitions between intermediate bands in > 3-band models (Würfel, 2003), but there is broad agreement that, at least in theory, a global efficiency advantage is possible if the MQWs can establish a third band. Alternatively, if a temperature difference can be maintained between the QW and barrier material, a global efficiency enhancement is also possible (Kettemann and Guillemoles, 2002). The question then arises as to whether in practice a third band, or something approximating it, exists in an MQW solar cell. The experimental evidence that shows an enhanced Voc (Barnham et al, 1996) is a strong indication of a relative efficiency enhancement – that the MQW p-i-n can outperform a bulk p-i-n cell – but it is important to note that it does not confirm the existence of a global efficiency enhancement. Nevertheless, a more detailed picture emerges when modelling the dark current (Nelson et al, 1995) and electroluminescence (Nelson et al, 1997) from a QW solar cell. In both cases the quasi-Fermi level in the QW has to be suppressed by approximately 20meV with respect to the barrier material to obtain a good fit to experimental results. It is important to note that the suppressed quasi-Fermi level is observed in the dark, where electrically injected carriers are falling into the QW. So the quasi-Fermi level suppression may be due to some transport-limiting process leading to a quasi-Fermi level gradient. In this case, if light is absorbed in the QW and generates a net photocurrent, then the gradient in the quasi-Fermi levels must be reversed to describe the flow of current and should lead to an inflated quasi-Fermi level in the QW and therefore higher recombination (Luque et al, 2001). This is then an example of how a gradient in a quasi-Fermi level will always generate entropy (Kondepudi and Prigogine, 1998) and therefore a lower photovoltaic efficiency. However, recent experiments show that there is no change in luminescence when moving from electrical injection, through Voc to photovoltaic behaviour (EkinsDaukes et al, 2003a and 2003b). This implies that there is no change in quasi-Fermi level under the reversal of current and is inconsistent with the prediction that a suppressed QW quasi-Fermi level in the dark will lead to an inflated quasi-Fermi level under illumination. Such behaviour is essentially in conflict with the second law of thermodynamics, so

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

66

N. J. EKINS-DAUKES

requires an alternative explanation. We can speculate that it may be inappropriate to define quasi-Fermi levels over distances as short as a QW (Settler and Lundstrom, 1994; Corkish and Honsberg, 1997). Also much of the QW luminescence stems from the recombination of excitons, whose population has a strongly non-thermal momentum distribution (Koch et al, 2002) and leads to a breakdown of Kirchoff‘s law relating absorptivity and emissivity, on which efficiency limits derived through the principle of detailed balance rely (Araújo and Marti, 1994). Nevertheless, one does not abandon quasiFermi levels lightly, as the analysis of photovoltaic efficiency becomes difficult without the convenience of quasi-thermal equilibrium (Würfel, 1995). Such speculation, however, is only hinting at possible mechanisms for the 20meV quasi-Fermi level suppression, and a much larger effect is required for a practical intermediate band device. It has also been pointed out that the MQW is not expected to be the best material to demonstrate an intermediate band (Luque et al, 2001), on account of the freedom of movement for carriers in the plane of the well, which promotes rapid phonon-mediated relaxation of carriers. A quantum dot solar cell with discrete levels may be better in this respect (Marti et al, 2000), but nevertheless the same concern applies over whether, in practice, an intermediate band can be established. The conclusion is therefore that plenty of mechanisms exist on paper that predict a global efficiency enhancement in an MQW device, but that experimental work should proceed to establish what properties can realistically be exploited in QW and other low dimensional materials. For example, a detailed investigation of electroluminescence from AlGaAs/GaAs double QW samples composed of a wide and narrow QW (Kluftinger et al, 2000 and 2001; Kluftinger, 2000) showed a quasi-Fermi level suppression in the wide well, but not in the narrow well. The reason for this remains elusive and may be the key to understanding the origin of the 20meV quasi-Fermi level suppression.

2.6 CONCLUSION Over the last 15 years a considerable quantity of work has been performed on the QW solar cell and this marks one of the first attempts to apply low dimensional structures to photovoltaics. The ability to engineer the device band gap has led to practical applications in multi-junction solar cells and thermophotovoltaics, while the strain-balance technique broadens the range of materials whose properties can be exploited in MQW or superlattice structures. To date only the band gap has been deliberately engineered in QW solar cell structures, but it is likely that other properties, such as the optical transitions and phonon modes, will also be controlled in order to produce highly efficient devices in the future.

ACKNOWLEDGEMENTS Extensive discussions with Keith Barnham, Massimo Mazzer and Jenny Nelson are gratefully acknowledged. David Bushnell kindly provided recent experimental data and Martin Green and Peter Würfel are thanked for useful discussions concerning efficiency limits. The Japan Society for the Promotion of Science (JSPS) and Masafumi Yamaguchi are thanked for their respective financial and scientific support.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

67

AUTHOR CONTACT DETAILS N. J. Ekins-Daukes, School of Physics, University of Sydney, NSW 2006, Australia. Tel: +61 2 9036 9259; Fax: +61 2 9351 7726; e-mail: [email protected]

NOTE 1 The contamination is most likely to be due to C which is inevitably incorporated into AlGaAs during MOVPE growth on account of the strong AlC bond.

REFERENCES Abbott, P., Rohr, C., Connolly, J. P., Ballard, I., Barnham, K., Ginige, R., Corbett, B., Clarke, G., Bland, S. W. and Mazzer, M. (2002) ‘A comparative study of bulk InGaAs and InGaAs/InGaAs straincompensated quantum well cells for thermophotovoltaic applications’, proceedings of 29th IEEE PV Specialists Conference, IEEE, New Orleans, US, May, pp1058–1061 Anderson, N. G. (1995) ‘Ideal theory of quantumwell solar cells’, J. Appl. Phys., vol 78, no 3, pp1850–1861 Anderson, N. (2002) ‘On quantum well solar cell efficiencies’, Physica, vol E14, no 12, pp126–131 Araújo, G. and Marti, A. (1994) ‘Absolute limiting efficiencies for photovoltaic energy conversion’, Solar Energy Materials and Solar Cells, no 33, pp213–240 Araújo, G. L. and Marti, A. (1995) ‘Electroluminescence coupling in 24 multiple quantum well diodes and solar cells’, Appl. Phys. Lett, vol 66, no 7, pp894–895 Araújo, G., Marti, A., Ragay, F. and Wolter, J. (1994) ‘Efficiency of multiple quantum well solar cells’, proceedings of 12th European Photovoltaic Solar Energy Conference, H. S. Stephens & Associates, Bedford, UK, pp1481–1484 Ballard, I., Barnham, K. W. J., Connolly, J. P., Nelson, J., Rohr, C., Roberts, C., Roberts, J. and Button, C. (2001) ‘The effect of temperature on the efficiency of multi quantum well cells for solar and thermophotovoltaic applications’, proceedings of 17th European Conference and Exhibition on PV Solar Energy, WIP and ETA, Munich, Germany and Florence, Italy, pp41–44 Ballard, I., Barnham, K. W. J., Nelson, J., Connolly, J. P., Roberts, C., Roberts, J. S. and Pate, M. A. (1998) ‘The effect of temperature on the efficiency of multiquantum well solar cells’, proceedings of 2nd World PV Energy Conversion Conference European Commission, Ispra, Italy and Vienna, Austria, July, pp3624–3626 Barnes, J. (1994) ‘An experimental and theoretical study of GaAs/InGaAs quantum well solar cells and carrier escape from QWs’, Ph.D. thesis, Imperial College, University of London, available from www.sc.ic.ac.uk/~q_pv/ Barnes, J., Ali, T., Barnham, K. W. J., Nelson, J. and Tsui, E. S. M. (1994) ‘Gallium arsenide/indium gallium arsenide multiquantum well solar cells’, proceedings of 12th European PV Solar Energy Conference, H. S. Stephens & Associates, Bedford, UK, pp1374–1377 Barnes, J., Nelson, J., Barnham, K., Roberts, J., Pate, M., Grey, R., Dosanjh, S., Mazzer, M. and Ghiraldo, F. (1996) ‘Characterization of GaAs/InGaAs QWs using photocurrent spectroscopy’, J. Appl. Phys., no 79, pp7775–7779 Barnham, K. (1992) Solar Cell, UK Patent Application PCT/GB92/01913, EC Patent Application 92921571.3, US Patent Application 211850, Australian Patent Application 27905/92 Barnham, K. W. J. and Duggan, G. (1990) ‘A new approach to highefficiency multiband-gap solar cells’, J. Appl. Phys., no 67, pp3490–3493 Barnham, K., and Vvedensky, D. (2001) Low Dimensional Semiconductor Structures: Fundamentals and Device Applications, Cambridge University Press, Cambridge, UK Barnham, K. W. J., Braun, B., Nelson, J., Paxman, M., Button, C., Roberts, J. S. and Foxon, C. T. (1991), ‘Shortcircuit current and energy efficiency enhancement in a low dimensional structure photovoltaic device’, Appl. Phys. Lett., no 59, pp135–137 Barnham, K. W. J., Connolly, J., Griffin, P., Haarpaintner, G., Nelson, J., Tsui, E., Zachariou, A., Osborne, J., Button, C., Hill, G., Hopkinson, M., Pate, M., Roberts, J. S. and Foxon, T. (1996) ‘Voltage enhancement in quantum well solar cells’, J. Appl. Phys., vol 80, no 2, pp1201–1206

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

68

N. J. EKINS-DAUKES

Barnham, K., Marques, J. L., Hassard, J. and O’Brien, P. (2000) ‘Quantum dot concentrator and thermodynamic model for the global redshift’, Appl. Phys. Lett., vol 76 no 9, pp1197–1199 Barnham, K., Ballard, I., Connolly, J., Ekins-Daukes, N., Kluftinger, B., Nelson, J. and Rohr, C. (2002) ‘Quantum well solar cells’, Physica, vol E14, no 12, pp27–36 Bastard, G. (1988) Wave Mechanics Applied to Semiconductor Heterostructures, Les Editions de Physique, les Ulis Bedair, S., Roberts, J., Jung, D., Moody, B., El-Masry, N. and Katsuyama, T. (2000) ‘GaAsP/InGaAsN strained layer superlattices for solar cell applications’, proceedings of 28th IEEE PV Specialists Conference, IEEE, US, pp1269–1272 Bremner, S., Corkish, R. and Honsberg, C. (1999) ‘Detailed balance efficiency limits with quasi-Fermi level variations’, IEEE Trans. Elect. Dev., vol 46, no 10, pp1932–1939 Brown, A. and Green, M. (2003) ‘Intermediate band solar cell with many bands: Ideal performance’, J. Appl. Phys., vol 94, no 9, pp6150–6158 Bushnell, D. (2002) ‘Optimisation of strain compensated multi-quantum-well solar cells’, Ph.D. thesis, Imperial College, University of London, available from www.sc.ic.ac.uk/~q_pv/ Bushnell, D. B., Barnham, K. W. J., Connolly, J. P., Ekins-Daukes, N. J., Airey, R., Hill, G. and Roberts, J. S. (2002) ‘Lighttrapping structures for multiquantum well solar cells’, proceedings of 29th IEEE PV Specialists Conference, IEEE, New Orleans, US, May, p1035 Bushnell, D., Ekins-Daukes, N., Barnham, K., Connolly, J., Roberts, J., Hill, G., Airey, R. and Mazzer, M. (2003a) ‘Shortcircuit current enhancement in Bragg stack multi-quantum-well solar cells for multi-junction space applications’, Solar Energy Materials and Solar Cells, vol 75, no 12, pp299–305 Bushnell, D. B., Barnham, K. W. J., Connolly, J. P., Mazzer, M., Ekins-Daukes, N. J., Roberts, J. S., Hill, G., Airey, R. and Nasi, L. (2003b) ‘Effect of barrier composition and well number on the dark current of quantum well solar cells’, proceedings of 3rd World Conference on Photovoltaic Energy Conversion, Osaka, Japan, May Chaffin, R. (1987) Quantum Well Multi-Junction Photovoltaic Cell, US Patent 4,688,068 Chaffin, R., Osbourn, G., Dawson and L., Biefeld, R. (1984) ‘Strained superlattice, quantum well, multi-junction photovoltaic cells’, proceedings of 17th IEEE PV Specialists Conference, IEEE, US, pp743–746 Connolly, J. (1997) ‘Modelling and optimising GaAs/AlGaAs mqw solar cells’, Ph.D. thesis, Imperial College, University of London, available from www.sc.ic.ac.uk/~q_pv/ Connolly, J. P. and Rohr, C. (2003) ‘Quantum well cells for thermophotovoltaics’, Semiconductor Science and Technology, vol 18, no 5, p216 Connolly, J. P., Barnham, K. W. J., Nelson, J., Roberts, C., Pate, M. and Roberts, J. S. (1998) ‘Short circuit current enhancement in GaAs/AlGaAs MQW solar cells’, proceedings of 2nd World PV Energy Conversion Conf. European Commission, Ispra, Italy, pp3631–3634 Connolly, J. P., Nelson, J., Barnham, K. and Ballard, I. (2000) ‘Multiple quantum well solar simulator’, proceedings of 28th IEEE PV Specialists Conference, Anchorage, IEEE, US, pp1304–1307 Connolly, J. P., Nelson, J., Ballard, I., Barnham, K. W. J., Rohr, C., Button, C., Roberts, J. and Foxon, T. (2001) ‘Modelling multi quantum well solar cell efficiency’, proceedings of 17th European Conference and Exhibition on PV Solar Energy, WIP, Munich, Germany Corkish, R. and Green, M. (1993) ‘Recombination of carriers in quantum well solar cells’, proceedings of 23rd PV Specialists Conference, IEEE, New York, pp675–689 Corkish, R. and Honsberg, C. (1997) ‘Dark currents in double-heterostructure and quantumwell solar cells’, proceedings of 26th IEEE PV Specialists Conference, pp955–960 Coutts, T. J. and Fitzgerald, M. C. (1998) ‘Thermophotovoltaics’, Scientific American, vol 279, no 3, pp90–95 Dimroth, F., Schubert, U. and Bett, A. (2000) ‘25.5% efficient GaInP/GaInAs tandem solar cells grown on GaAs substrates’, IEEE Elect. Dev. Lett., vol 21, no 5, pp209–211

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

69

Dingle, R., Gossard, A. and Wiegmann, W. (1975) ‘Direct observation of superlattice formation in a semiconductor heterostructure’, Phys. Rev. Lett., vol 34, no 21, pp1327–1330 Ekins-Daukes, N. (2000) ‘An investigation into the efficiency of strained and strain-balanced quantum well solar cells’, Ph.D. thesis, Imperial College, University of London, available from www.sc.ic.ac.uk/~q_pv/ Ekins-Daukes, N., Mazzer, M., Barnes, J., Ballard, I., Barnham, K., Holmes, S., Grey, R., Roberts, J., Pate M. and Hill, G. (1998) ‘New approaches in GaAs/InGaAs quantum well solar cells’, proceedings of 2nd World Conf. and Exhibition on Photovoltaic Solar Energy Conversion, European Commission, Ispra, Italy, p3627 Ekins-Daukes, N., Barnham, K., Connolly, J., Roberts, J., Clark, J., Hill, G. and Mazzer, M. (1999) ‘Strain-balanced GaAsP/InGaAs quantum well solar cells’, Appl. Phys. Lett., vol 75, no 26, pp4195–4197 Ekins-Daukes, N., Bushnell, D., Zhang, J., Barnham, K. and Mazzer, M. (2000) ‘Strain-balanced materials for highefficiency solar cells’, proceedings of 28th IEEE PV Specialists Conference, IEEE, US, pp1273–1276 Ekins-Daukes, N., Barnes, J., Barnham, K., Connolly, J., Mazzer, M., Clark, J., Grey, R., Hill, G., Pate and M., Roberts, J. (2001) ‘Strained and strain-balanced quantum well devices for high efficiency tandem solar cells’, Solar Energy Materials and Solar Cells, no 68, pp71–87 Ekins-Daukes, N., Kawaguchi, K. and Zhang, J. (2002) ‘Strain-balanced criteria for multiple quantum well structures and its signature in X-ray rocking curves’, Crystal Growth & Design, vol 2, no 4, pp287–292 Ekins-Daukes, N., Ballard, I., Calder, C., Barnham, K., Hill, G. and Roberts, J., (2003a) ‘Photovoltaic efficiency enhancement through thermal up-conversion’, Appl. Phys. Lett., vol 82, no 12, pp1974–1976 Ekins-Daukes, N., Calder, C., Ballard, I., Barnham, K., Nelson, J., Hill, G. and Roberts, J. (2003b) ‘Quantum well and bulk quasiFermi level behaviour under illuminated conditions’, proceedings of 3rd World Conference on Photovoltaic Energy Conversion, Osaka, Japan, May Fatemi, N. S., Pollard, H. E., Hou, H. Q. and Sharps, P. R. (2000) ‘Solar array trades between very high-efficency multi-junction and Si space solar cells’, proceedings of 28th IEEE PV Specialists Conference, IEEE, US, pp1083–804 Freundlich, A. (1998) Strained Quantum Well Photovoltaic Energy Converter, US Patent 5,851,310 Freundlich, A. (2000) Quantum Well Thermophotovoltaic Convertor, US Patent 6,150,604 Freundlich, A. (2002) Multi-Quantum Well Tandem Solar Cell, cont. US Patent 6,372,980 Freundlich, A. and Serdiukova, I. (1998) ‘Multi-quantum well tandem solar cells with efficiencies exceeding 30% AM0’, proceedings of 2nd World Conf. and Exhibition on Photovoltaic Solar Energy Conversion, European Commission, Ispra, Italy, Vienna, Austria, July, p3707 Freundlich, A., Rossignol, V., Vilela, M., Renaud, P., Bensaoula, A. and Medelci, N. (1994) ‘InPbased quantum well solar cells grown by chemical beam epitaxy’, proceedings of 1st World Conference and Exhibition on Photovoltaic Solar Energy Conversion, IEEE, US, pp1886–1889 Friedman, D., Geisz, J., Kurtz S. and Olson, J. (1998) ‘1ev solar cells with GaInNAs active layer’, J. Crystal Growth, no 195, pp409–415 Gevaux, D. G., Green, A. M., Phillips, C. C., Vurgaftman, I., Bewley, W. W., Felix, C. L., Meyer, J. R., Lee, H. and Martinelli, R. U. (2001) ‘GaInSb/InAs/AlGaAsSb “W” quantum-well light emitting diodes’, Appl. Phys. Lett., vol 79, no 25, pp4073–4075 Green, M. (2000) ‘Prospects for photovoltaic efficiency enhancement using low-dimensional structures’, Nanotechnology, vol 11, no 4, pp401–405 Griffin, P. R., Barnes, J., Barnham, K. W. J., Haarpainter, G., Mazzer, M. Zanotti-Fregonara, C., Grunbaum, E., Olson, C., Rohr, C., David, J. P. R., Roberts, J. S., Grey, R. and Pate, M. A. (1996) ‘Effect of strain relaxation on forward bias dark currents in GaAs/InGaAs multiquantum well pin diodes’, J. Appl. Phys. no 80, pp5815–5820 Griffin, P., Ballard, I., Barnham, K., Nelson, J., Zachariou, A., Epler, J., Hill, G., Button, C. and Pate, M. (1997) ‘Advantages of quantum well solar cells for TPV’, in T. J. Coutts, C. S. Allman and J. P. Benner (eds) Thermophotovoltaic Generation of

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

70

N. J. EKINS-DAUKES

Electricity: Third NREL Conference, vol 401 of AIP Conference Proceedings, American Institute of Physics, Woodbury, New York, Colorado Springs, Colorado, US, May, pp411–422 Griffin, P., Ballard, I., Barnham, K., Nelson, J., Zachariou, A., Epler, J., Hill, G., Button, C. and Pate, M. (1998) ‘The application quantum well solar cells to thermophotovoltaics’, Solar Energy Materials and Solar Cells, no 50, pp213–219 Hepp, M. (1998) ‘Light trapping in quantum well solar cells grown on patterned substrate’, Master’s thesis, Imperial College of Science, Technology and Medicine, University of London., available from www.sc.ic.ac.uk/~q_pv/ Hoffman, R., Fatemi, N., Stan, M., Jenkins P. and Weizer, V. (1998) ‘High efficiency InGaAs on GaAs devices for monolithic multi-junction solar cell applications’, proceedings of 2nd World Conf. and Exhibition on Photovoltaic Solar Energy Conversion, European Commission, Ispra, Italy, pp3604–3608 Honsberg, C., Bremner, S. and Corkish, R. (2002) ‘Design tradeoffs and rules for multiple energy level solar cells’, Physica, vol E14, no 12, pp136–141 Kettemann, S. and Guillemoles, J. (1995) ‘Limiting efficiency of LDS solar cells’, proceedings of 13th European Photovoltaic Solar Energy Conference, Stephens, Bedford, UK, pp119–121 Kettemann, S. and Guillemoles, J. (2002) ‘Thermoelectric field effects in lowdimensional structure solar cells’, Physica, vol E14, no 12, pp101–106 Kluftinger, B. (2000) ‘An experimental study of the quasiFermi level separations in quantum well solar cells’, Ph.D. thesis, Imperial College, University of London, available from www.sc.ic.ac.uk/~q_pv/ Kluftinger, B., Barnham, K., Nelson, J., Foxon, T. and Cheng, T. (2000) ‘Temperature dependent study of the quasi-Fermi level separation in double quantum well pin structures’, Microelectronic Engineering, no 5152, pp265–274 Kluftinger, B., Barnham, K., Nelson, J., Foxon, T. and Cheng, T. (2001) ‘Temperature dependent study of radiative losses in double quantum well solar cells’, Sol. Energy Mater. Sol. Cells, no 66, pp510–509 Koch, S., Meier, T., Hoyer, W. and Kira, M. (2002) ‘Theory of the optical properties of semiconductor nanostructures’, Physica, vol E14, no 12, pp45–52 Kojima, N., Goetzberger, O., Ohshita, Y. and Yamaguchi, M. (2000) ‘Approach to the carbonbased superlattice structure solar cell’, proceedings of 28th IEEE PV Specialists Conference, IEEE, US, pp873–875 Kondepudi, D. and Prigogine, I. (1998) Modern Thermodynamics, John Wiley & Sons Kurtz, S., Faine, P. and Olson, J. (1990) ‘Modeling of 2junction, series-connected tandem solar cells using top cell thickness as an adjustable parameter’, J. Appl. Phys., vol 68, no 4, pp1890–1895 Levine, B. (1993) ‘Quantum well infrared photodetectors’, J. Appl. Phys., vol 74, no 8, R1–R81 Li, N. Y., Sharps, P. R., Stan, M., Newman, F., Hills, J. S., Hou, H. Q., Gee, J. M. and Aiken, D. J. (2000) ‘Development of 1.25 eV InGaAsN for triple junction solar cells’, proceedings of 28th IEEE PV Specialists Conference, IEEE, US, pp986–989 Luque, A. and Marti, A. (1997a) ‘Entropy production in photovoltaic conversion’, Phys. Rev. B., vol 55, no 11, pp6994–6999 Luque, A. and Marti, A. (1997b) ‘Increasing the efficiency of ideal solar cells by photon induced transitions at intermediate levels’, Phys. Rev. Lett., vol 78, no 26, pp5014–5017 Luque, A., Marti, A. and Cuadra, L. (2001), ‘Thermodynamic consistency of subband-gap absorbing solar cell proposals’, IEEE Trans. Electron Devices, vol 48, no 9, pp2118–2124 Marti, A., Cuadra, L. and Luque, A. (2000) ‘Quantum dot intermediate band solar cell’, proceedings of 28th IEEE PV Specialists Conference, IEEE, US, pp940–943 Matthews, J. W. and Blakeslee, A. E. (1974) ‘Defects in epitaxial multilayers (I)’, J. Crystal Growth, no 27, pp118–125 Mazzer, M., Grünbaum, E., Barnham, K., Barnes, J., Griffin, P., Holt, D., Hutchinson, J., Norman, A., David, J., Roberts, J. and Grey, R. (1996) ‘Study of misfit dislocations by EBIC, CL and HRTEM in GaAs/InGaAs lattice strained multi-quantum well pin solar cells’, Materials Science and Engineering, no B42, pp43–51 McFarlane, S. C., Barnes, J., Barnham, K. W. J., Tsui, E. S. M., Button, C. and Roberts, J. S. (1999) ‘Space charge effects in carrier escape from single quantum well structures’, vol 86, no 9, pp5109–5115

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

71

Monier, C., Freundlich, A. and Vilela, M. (1999) ‘Oscillator strength of excitons in InGaAs/GaAs QWs in the presence of a large electric field’, vol 85, no 5, pp2713–2718 Nagashima, T., Okumura, K., Murata, K. and Yamaguchi, M. (2003) ‘A germanium backcontact type cell for thermophotovoltaic applications’, proceedings of 3rd World Conference on Photovoltaic Energy Conversion, Osaka; Japan, May Nelson, J. (1995) ‘Quantum well structures for photovoltaic energy conversion’, in M. Francome and J. Vossen (eds) Physics of Thin Films, Academic Press, New York, vol 21, pp311–368 Nelson, J. (2001) ‘Quantum well solar cells’, in M. D. Archer and R. Hill (eds) Clean Electricity from Photovoltaics, Imperial College Press, London, pp71–87 Nelson, J. (2003) The Physics of Solar Cells, Imperial College Press, London Nelson, J., Paxman, M., Barnham, K. W. J., Roberts, J. S. and Button, C. (1993) ‘Steady state carrier escape from single QWs’, IEEE Journal of Quantum Electronics, no 29, pp1460–1467 Nelson, J., Kluftinger, B., Tsui, E., Barnham, K., Roberts J. and Button, C. (1995) ‘QuasiFermi level separation in quantum well solar cells’, proceedings of 13th European Photovoltaic Solar Energy Conference, H. S. Stephens & Associates, Bedford, UK, pp150–153 Nelson, J., Barnes, J., Ekins-Daukes, N., Kluftinger, B., Tsui, E., Barnham, K. W. J., Foxon, T., Cheng, T. and Roberts, J. S. (1997) ‘Observation of suppressed radiative recombination in single quantum well pin photodiodes’, no 82, pp6240–6246 Nelson, J., Ballard, J., Barnham, K. W. J., Connolly, J., Roberts, J. S. and Pate, M. (1999) ‘Effect of QW location on quantum well solar cell dark currents’, vol 86, no 10, pp5898–5905 Okada, Y., Seki, S., Takeda, T. and Kawabe, M. (2002) ‘Control of dark currents in multi-quantum well solar cells fabricated by atomic H assisted molecular beam epitaxy’, J. Cryst. Growth, no 237, pp1515–1518 Osbourne, J. (1994) ‘InGaP/GaAs quantum well solar cells’, Master’s thesis, Imperial College, University of London Paxman, M. (1992) ‘An experimental study of GaAs/AlGaAs multiple quantum well solar cells’, Ph.D. thesis, Imperial College, University of London Paxman, M., Nelson, J., Braun, B., Connolly, J., Barnham, K. W. J., Foxon, C. T. and Roberts, J. S. (1993) ‘Modeling the spectral response of the quantum well solar cell’, no 74, pp614–621 Peng, R., Mazzer, M. and Barnham, K. (2003) ‘Efficiency enhancement of ideal photovoltaic solar cells by photonic excitations in multiintermediate band structures’, Appl. Phys. Lett., vol 83, no 4, pp770–772 Ragay, F., Wolter, J., Marti, A. and Araújo, G. (1994a) ‘Experimental analysis of GaAsInGaAs MQW solar cells’, proceedings of 1st World Conference and Exhibition on Photovoltaic Solar Energy Conversion, IEEE, US, p1754 Ragay, F., Wolter, J., Marti, A. and Araujo, G. (1994b) ‘Experimental analysis of the efficiency of MQW solar cells’, proceedings of 12th European Photovoltaic Solar Energy Conference, H. S. Stephens & Associates, Bedford, UK, pp1429–1433 Raisky, O. Y., Wang, W. B., Alfano, R. R., Reynolds, Jr., C.L. Stampone, D. V. and Focht, M. W. (1998) ‘In1-xGaxAs1-yPY /InP multiple quantum well solar cell structures’, J. Appl. Phys., vol 84, no 10, pp5790–5794 Raisky, O. Y., Wang, W. B., Alfano, R. R. and Reynolds, Jr., C.L. (2001) ‘Carrier screening effects in photoluminescence spectra of InGaAsP/InP multiple quantum well photovoltaic structures’, Appl. Phys. Lett., vol 79, no 3, pp430–432 Renaud, P., Vilela, M., Freundlich, A., Bensaoula, A. and Medelci, N. (1994) ‘Modeling p-i-n multi quantum well n solar cells: A contribution for a near optimum design’, proceedings of 1st World Conference and Exhibition on Photovoltaic Solar Energy Conversion, IEEE, US, pp1787–1790 Rimada, J.C. and Hernandez, L. (2001a) ‘Modelling of ideal AlGaAs quantum well solar cells’, Microelectronics, vol 32, no 9, pp719–723 Rimada, J.C. and Hernandez, L. (2001b) ‘A new approach to ideal AlGaAs MQW solar cells’, Mod. Phys. Lett., vol B15, no 1719, pp778–781

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

72

N. J. EKINS-DAUKES

Rohr, C. (2000) ‘InGaAsP quantum well cells for thermophotovoltaic applications’, Ph.D. thesis, Imperial College, University of London Rohr, C., Connolly, J. P., Barnham, K. W. J., Griffin, P. R., Nelson, J., Ballard, I., Zachariou, A., Button, C. and Clark, J. (1998) ‘InGaAsP quantum well cells for thermophotovoltaics’, proceedings of 2nd World Conf. and Exhibition on Photovoltaic Solar Energy Conversion, Ispra, Italy and Vienna, Austria, July 1998, European Commission, pp230–233 Rohr, C., Connolly, J. P., Barnham, K. W. J., Griffin, P. R., Nelson, J., Ballard, I., Button, C. and Clark, J. (1999) ‘Optimisation of InGaAsP quantum well cells for hybrid solarthermophotovoltaic applications’, in T. J. Coutts, J. P. Benner and C. S. Allman (eds) Thermophotovoltaic Generation of Electricity: Fourth NREL Conf., vol 460 of AIP Conference Proceedings, Denver, Colorado, US, October 1998, American Institute of Physics, Woodbury, New York, pp83–92 Rohr, C., Barnham, K. W. J., Connolly, J. P., Nelson, J., Button, C. and Clark, J. (2000a) ‘Potential of InGaAsP quantum well cells for thermophotovoltaics’, proceedings of 26th International Symposium on Compound Semiconductors, no 166 in Institute of Physics Conference Series, Berlin, Germany, August 1999, Institute of Physics Publishing, Bristoland, Philadelphia, US, pp423–426 Rohr, C., Connolly, J., Barnham, K. W., Mazzer, M., Button, C. and Clark, J. (2000b) ‘Strain-balanced In0.62Ga0.38As/In0.47Ga0.53As (InP) quantum well cell for thermophotovoltaics’, proceedings of 28th IEEE Photovoltaic Specialists Conference, IEEE, US, pp1234–1237 Rohr, C., Abbott, P., Ballard, I., Connolly, J. P., Barnham, K. W., Nasi, L., Ferrari, C., Lazzarini, L., Mazzer, M. and Roberts, J. (2002a) ‘Straincompensated InGaAs/InGaAs quantum well cell with 2μm band-edge’, in T. J. Coutts, G. Guazzoni and J. Luther (eds) The Fifth Conference on Thermophotovoltaic Generation of Electricity. Vol. 653 of AIP Conference Proceedings, American Institute of Physics, Woodbury, New York, pp344–353 Rohr, C., Connolly, J. P., Ekins-Daukes, N., Abbott, P., Ballard, I., Barnham, K. W., Mazzer, M. and Button, C. (2002b) ‘InGaAs/InGaAs strain-compensated quantum well cells for thermophotovoltaic applications’, Physica E: Low-dimensional Systems and Nanostructures, vol 14, no 12, pp158–161 Serdiukova, I., Newman, F., Aguillar, L., Vilela, M. F., Monier, C. and Ferundlich, A. (1997) ‘Strained In0.53Ga0.47As/InxGa1-xAs (x>0.6) multiquantum well thermophotovoltaic converters’, proceedings of 26th IEEE PV Specialists Conference, IEEE, US, pp963–966 Serdiukova, I., Monier, C., Vilela, M. and Freundlich, A. (1999) ‘Critical built-in electric field for an optimum carrier collection in multi-quantum well pin diodes’, Appl. Phys. Lett., vol 74, no 19, pp2812–2814 Settler, M. and Lundstrom, M. (1994) ‘A detailed investigation of hetero junction transport using a rigorous solution to the Boltzmann equation’, IEEE Trans. Elec. Dev., vol 41, no 4, pp592–600 Shockley, W. and Queisser, H. (1961) ‘Detailed balance limit of efficiency of pn junction solar cells’, J. Appl. Phys., vol 32, no 3, pp510–519 Steinke, L. (1996) ‘Light trapping in MQW solar cells’, Master’s thesis, Imperial college, University of London, available at www.sc.ic.ac.uk/~q_pv/ Stradling, R. and Klipstein, P. (1989) Growth and Characterization of Semiconductors, Adam Hilger, UK Suzuki, Y., Kikuchi, T., Kawabe, M. and Okada, Y. (1999) ‘Atomic hydrogen assisted molecular beam epitaxy for the fabrication of multi-quantum well solar cells’, J. Appl. Phys., vol 86, no 10, pp5858–5861 Takamoto, T., Ikeda, E., Kurita, H., Tanabe, T., Tanaka, S., Matsubara, H., Mine, Y., Takagishi, S. and Yamaguchi, M. (1997) ‘InGaP/GaAs and InGaAs mechanically stacked triplejunction solar cells’, proceedings of 26th PV Specialists Conference, IEEE, pp1031–1034 Takamoto, T., Agui, T., Ikeda, E. and Kurita, H. (2000) ‘High efficiency InGaP/InGaAs tandem solar cells on Ge substrates’, proceedings of 28th IEEE PV Specialists Conference, IEEE, US, pp976–981 Takamoto, T., Agui, T., Ikeda, E. and Kurita, H. (2001) ‘High efficiency InGaP/In0.01Ga0.99As tandem solar cells lattice matched to Ge substrates’, Solar Energy Materials and Solar Cells, no 66, pp511–516

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

Quantum Well Solar Cells

73

Tang, P., Pullin, M., Phillips, C. and Stradling, R. (1995) ‘Photoluminescence studies of NIPI superlattices in InSb and InAs suppression of Auger recombination due to type-II potentials’, Semicond. Sci. Technol., vol 10, no 4, pp476–482 Tibbits, T., Ballard, I., Barnham, K., Bushnell, D., Ekins-Daukes, N., Airy, R., Hill, G. and Roberts, J. (2003) ‘The potential for strain-balanced quantum well solar cells in terrestrial concentrator applications’, proceedings of 3rd World Conference on Photovoltaic Energy Conversion, Osaka, Japan, May Tobin, S., Vernon, S., Bajgar, C., Woktczuk, S., Melloch, M., Keshavarzi, A., Stellwag, T., Venkatensan, S., Lundstrom, M. and Emery, K. (1990) ‘Assessment of MOCVD and MBE grown GaAs for high efficiency solar cell applications’, IEEE Trans. Electron Devices, vol 37, no 2, pp469–477 Varonides, A. and Berger, A. (1997) Proceedings of 15th European Photovoltaic Solar Energy Conference, H. S. Stephens & Associates, Bedford, UK, pp1712–1715 Vegard, L. (1921) ‘Die Konstitution der Mischkristalle und die Raumfüllung der Atome’, Z. Phys., no 5, pp17–26 Walters, R., Summers, G., Messenger, S., Freundlich, A., Monier, C. and Newman, F. (2000a) ‘Radiation hard multiquantum well InP/InAsP solar cells for space applications’, Prog. Photovolt. Res., vol 8, no 3, pp349–354 Walters, R., Summers, G., Messenger, S., Romero, M., Araújo, D., Garcia, R., Freundlich, A., Newman, F. and Vilela, M. (2000b) ‘Electron beam induced current and cathodoluminescence study of proton irradiated InAsP/InP quantum well solar cells’, proceedings of 28th IEEE PV Specialists Conference, IEEE, US, pp1312–1316 Walters, R., Summers, G., Messenger, S., Romero, M., AlJassim, M., Garcia, R., Araújo, D., Freundlich, A., Newman, F. and Vilela, M. (2001) ‘Electron beam induced current and cathodoluminescence study of proton irradiated InAsP/InP quantumwell solar cells’, J. Appl. Phys., vol 90, no 6, pp2840–2846 Weisbuch, C. and Vinter, B. (1991) Quantum Semiconductor Structures: Fundamentals and Applications, Academic Press Würfel, P. (1995) ‘Is an illuminated semiconductor far from thermodynamicequilibrium’, Sol. Energy Mater. Sol. Cells, vol 38, no 14, pp23–28 Würfel, P. (2003) ‘Improvement of solar cell efficiencies by impurity transitions’, Sol. Energy Mater. Sol. Cells, vol 79, no 1, pp153–161 Yablonovitch, E. and Cody, G. (1982) ‘Intensity enhancement in textured optical sheets for solar cells’, IEEE Trans. Electron Devices, ED29, pp300–305 Yamaguchi, M. and Luque, A. (1999) ‘High efficiency and high concentration in photovoltaics’, IEEE Trans. Elec. Dev., vol 46, no 10, pp2139–2144 Yang, M. J., Yamaguchi, M., Takamoto, T., Ikeda, E., Kurita, H. and Ohmori, M. (1997) ‘Photoluminescence analysis of InGaP top cells for high-efficiency multi-junction solar cells’, Solar Energy Materials and Solar Cells, no 45, pp331–339 Zachariou, A. (1996) ‘An experimental study of InP/InGaAs quantum well solar cells’, Ph.D. thesis, Imperial College of Science, Technology and Medicine, University of London, available from www.sc.ic.ac.uk/~q_pv/ Zachariou, A., Barnham, K. W. J., Griffin, P., Nelson, J., Button, C. C., Hopkinson, M., Pate, M. and Epler, J. (1996) ‘A new approach to pdoping and the observation of efficiency enhancement in InP/InGaAs quantum well solar cells’, proceedings of 25th IEEE PV Specialists Conference, IEEE, US, pp113–116 Zachariou, A., Barnes, J., Barnham, K. W. J., Nelson, J., Tsui, E. S. M., Epler, J. and Pate, M. (1998a) ‘A carrier escape study from InP/InGaAs single quantum well solar cells’, J. Appl. Phys., vol 83, no 2, pp877–881 Zachariou, A., Barnham, K. W. J., Griffin, P., Nelson, J., Osborne, J., Hopkinson, M. and Pate, M. (1998b) ‘GaInP/GaAs quantum well solar cells’, proceedings of 2nd World Conf. and Exhibition on Photovoltaic Solar Energy Conversion, European Commission, Ispra, Italy, pp223–226

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 45–73

3

Recent Progress of Organic Photovoltaics Sam-Shajing Sun

Abstract This article briefly reviews and summarizes recent key developments in organic and polymeric photovoltaic materials and devices. Specifically, the article briefly reviews the four major stages and/or types of the organic and polymeric photovoltaic developments from single layer ‘Schottky junction cell’, to donor/acceptor bilayer ‘Tang cell’, to donor/acceptor blend ‘heterojunction cell’ to donor/acceptor ‘bicontinuous phase separated nano-structured cell’. The current relatively low photoelectric power conversion efficiencies (less than 6 per cent) of purely organic and polymeric photovoltaic materials and devices are attributed mainly to the heavy ‘photon loss’, the ‘exciton loss’ and ‘carrier loss’. However, high efficiency can be achieved as these ‘losses’ can be addressed via optimizations in both space and energy domains as discussed. Organic versus inorganic photovoltaics is also briefly compared.

■ Keywords – organic and polymeric photovoltaics; plastic solar cells; organic and inorganic semiconductors; band theory; excitons; light harvesting; portable power

3.1 INTRODUCTION While inorganic crystalline-based solar cell technology has become relatively mature – over 30 per cent photoelectric power conversion efficiencies have been demonstrated,1-4 and much higher theoretical efficiencies have been predicted3 – this technology suffers from a relatively high manufacturing cost and a current severe shortage of the feedstock materials. As the need for renewable and solar energy technologies is rapidly growing, alternative materials or technologies that could reduce the solar cell manufacturing costs and have sufficient feedstock supplies become very attractive and critical. Organic or polymeric photovoltaic materials and technology fall right into this category.5–7 In comparison to the traditional inorganic solar cell, recently developed organic and polymeric conjugated semiconducting materials appear very promising for photovoltaic applications for several reasons:

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

75

● their lightweight, flexible shape, versatile materials synthesis and device fabrications,

and low cost on large-scale industrial production; ● almost continuous tunability of the materials’ energy levels and gaps via molecular

design, synthesis and processing; ● ultra-fast optoelectronic response and charge carrier generation at organic

donor/acceptor interfaces (this makes organic photovoltaic materials also attractive for developing potential fast photodetectors); and ● integrability into other products such as textiles that can be used for fabricating solar cell tents (see Figure 3.1) and clothing, packaging systems, consumption goods and future ‘all-plastic’ optoelectronic devices.5–7

The Sun Future Camping/Hiking

Plastic Solar Cell Tent

FIGURE 3.1 Lightweight and flexible thin film ‘plastic solar cells’ are very attractive for camping tents or any mobile units.

3.2 ORGANIC VERSUS INORGANIC SEMICONDUCTORS In organic conjugated semiconductor materials, typically the valence shell / electrons are responsible for the optoelectronic properties.5–7 When the material is at its lowest ground state, the highest occupied molecular orbital (HOMO) typically refers to / bonding orbital, and the lowest unoccupied molecular orbital (LUMO) typically refers to /* anti-bonding orbital. In typical organic semiconductors, including most organic crystalline semiconductors, the spatial and therefore the electronic overlap or coupling of / orbitals are much poorer compared to their inorganic semiconductor counterparts. Therefore traditional long-range delocalized conduction band (CB) and valence band (VB) can hardly be formed in organics due to this poor frontier orbital overlap, particularly the poor inter-molecular overlap.7 Specifically, poor inter-molecular orbital overlap results in relatively small charge delocalization length or small frontier orbital (i.e., HOMOs and LUMOs) band size (BS)8 (l, the extent of charge delocalization) that would limit or suppress the exciton radius (r). Since the exciton Coulombic binding energy E is inversely proportional to the exciton radius (E=K/¡r, as shown in Figure 3.2), for most organic or polymeric materials where small radius and strongly correlated ‘Frenkel’ type of exciton are typically generated9 and the binding energy is mostly larger than

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

76

S-S. SUN

the room temperature thermal energy (the top diamond curve in Figure 3.2), in other words a secondary force is needed to dissociate the photogenerated exciton. This is called the secondary photo carrier generation process (or excitonic mechanism) and will be discussed in the following section. In fact, the initial sunlight harvesting steps of photosynthesis in natural plants usually follow the ‘secondary’ process more closely.10 However, as shown in Figure 3.2, VB to CB band-to-band-like photo carrier generation (also called the primary photo carrier generation process or SSH model8) is also possible if the orbital overlap is really good, such that the exciton radius is very large, or the dielectric constant ¡ of the material is very large so the exciton Coulombic binding energy is below the thermal energy level. A ‘Wannier’ type of exciton curve affiliated with large dielectric constant materials such as silicon crystal is also shown (as triangles) in Figure 3.2. Due to the typical small band size (BS), the excitation energy gap Eg in organic or polymeric materials only represents the energy difference between the relatively localized LUMO and HOMO, not the traditional delocalized VB and CB. When a photon with energy equal to or over Eg excite the organic or polymeric molecule, an electron first transfers from the HOMO to the LUMO and then quickly relaxes with the hole to form a ‘Frenkel’ type exciton, also called polaron-exciton in organics.8 Depending on the dielectric constant of the materials, the ‘Frenkel’ type exciton typically has a size of less than 2nm with binding energies larger than 0.1eV.

'Frenkel' Exciton Curve

kT

'Wannier' Exciton Curve

100

Coulombic Energy E (eV)

10

1

0.1

0.01

0.001 0

10

20

30

40

Exciton Radius r (nm) The diamond curve (top) is obtained from an organic exciton. The triangle curve (bottom) is obtained from an inorganic exciton.

FIGURE 3.2 Scheme of exciton Coulombic binding energy (in log scale) versus exciton radius.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

50

Recent Progress of Organic Photovoltaics

77

As discussed above, one key difference between organic and inorganic semiconductors appears to be the extent of the spatial (and therefore the electronic) overlap of the frontier orbitals, or the band size (BS), as BS is directly affected by the extent of orbital spatial overlap or coupling. Recent experimental data on the charge mobility of an organic crystal, rubrene, has confirmed this argument. Rubrene is an organic molecule that can self-assemble to form a highly ordered crystalline structure such that its frontier orbitals are spatially well overlapped between adjacent molecules, so the CB and VB band-like states might be formed, and charge mobility up to 20cm2/Vs has been observed in a rubrene.11 On the mechanisms of photo carrier generation, it is possible the exciton is generated with its binding energy below the thermal energy line, so that a photon plus room temperature may instantaneously generate a free electron (also called negative polaron) at CB and a free hole (also called positive polaron) at VB, in other words a ‘band-to-band’ or ‘primary photo carrier generation’ mechanism applies.

3.3 ORGANIC/POLYMERIC SOLAR CELL DEVELOPMENTS 3.3.1 SINGLE LAYER ‘SCHOTTKY’ TYPE SOLAR CELLS The earliest photoconductivity observation in organic materials was reported in Anthracene in 1907.12 Later, some organic dyes and biological materials, such as methylene blue, carotenes, chlorophylls and related porphyrins or phthalocyanines were also found to exhibit photoconductive or very weak photovoltaic effects,13–14 with best photoelectric power conversion efficiency of 0.05 per cent obtained on a chromium/chlorophyll-a/mercury cell.15 Conjugated polymers such as polyacetylene,16 polythiophenes,17 and poly-p-phenylenevinylenes (PPVs)18 were also investigated for photovoltaic effects, with best power conversion efficiency of 0.1 per cent achieved under white light illumination,18 and a best open-circuit voltage (Voc) of up to 1.7 volt was reported using calcium as the small work function electrode.19 The low photoelectric conversion efficiency can be attributed mainly to the ‘Frenkel exciton’ model. As shown in Figure 3.3, a single layer organic cell is composed of an organic semiconductor sandwiched between two different metal electrodes with different work functions. The small work function electrode (SWFE) typically acts as the negative electrode to collect photogenerated electrons, and the large work function electrode (LWFE) acts as the positive electrode to collect photogenerated holes. At least one electrode is very thin and semi transparent, so the light can pass through. As Figure 3.3 shows, when an energymatched photon strikes most part of the organic layer (assumed defect free), only a strongly bound ‘Frenkel’ exciton is generated, which typically decays radiatively or nonradiatively back to the ground state without contributing to the carrier generation (Figure 3.3b). This explains the very low photoelectric efficiency of single layer organic PV cells. However, for those few excitons that were generated at or diffused into the ‘Schottky junction’ area near the organic/metal interface as shown in Figure 3.3, the exciton can be dissociated into an electron and a hole by a strong field formed from the orbital bending or ‘depletion region’ in the Schottky junction area.15, 20 Additionally, frontier energy level offsets due to the presence of molecular oxygen21 and certain impurity or structure defect sites (such as carbonyl groups)22 also contribute to the charged carrier generations. However, within the typical exciton lifetime of pico- to nano-seconds, most ‘Frenkel’

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

78

S-S. SUN

excitons can only travel 5–50nm (also called average exciton diffusion length, AEDL), much less than the thickness of most PV materials films, which are typically over 100nm; most excitons are therefore wasted. This ‘exciton loss’ is a major loss mechanism. Those early cells are also called ‘Schottky cells’ or ‘defect cells’ due to the carriers mainly being generated in the Schottky junction or defect sites. These single layer solar cells may be called first generation cells.

Schottky Junction

Vacuum Level

SWFE

LWFE

Schottky Junction

hv LUMO

hv

ex

em ex

δE SWFE

LWFE HOMO

(a)

(b)

FIGURE 3.3 First generation single layer type organic photovoltaic cell or ‘Schottky cell’ with (a) device structure and (b) energy diagram

3.3.2 DONOR/ACCEPTOR BILAYER HETEROJUNCTION CELL OR ‘TANG CELL’ From the spatial (geometry) structure point of view, the second generation solar cells were of the donor/acceptor bilayer structure type as shown in Figure 3.4. This bilayer heterojunction type cell was first demonstrated in 1986 with a surprising 1 per cent power conversion efficiency under AM 2.23 Since then, many other donor/acceptor (D-A) bilayer systems have been investigated extensively,5–6 including, for instance, molecule-fullerene,24 polymer-fullerene25–27 and polymer-polymer D-A bilayer systems,28–29 with a best 1 per cent power conversion efficiency achieved in a PPV/C60 bilayer cell.27 It was also found that the average exciton diffusion length (AEDL) in PPV is about 9nm.27 As shown in Figure 3.4a, once a photogenerated ‘Frenkel’ exciton in either donor or acceptor layer diffuses to the middle donor/acceptor interface, charge separation occurs whereby the electrons transfer to or remain in the acceptor layer and holes transfer to or remain in the donor layer. The charge carrier is now generated at such an artificially created donor/acceptor interface and the frontier orbital energy offset between the donor and acceptor has become the key driving force for exciton dissociation. Due to both the electrode-induced internal field and chemical potential driving forces,30–31 the electrons and holes than diffuse or ‘hop’ to their respective electrodes much quicker than in single layered cells. The likelihood of carrier

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

79

recombination is also relatively small due to electrons and holes now moving in two separate layers or domains. However, a major limiting factor for the double layer cell is still the relatively thick donor or acceptor layer (typically over 100nm) versus the relatively short average exciton diffusion length (5–50nm), in other words many photogenerated excitons decay before they reach the D-A interface (Figures 3.4a and b). On the other hand, if the

D/A Interface

Vacuum Level

SWFE

LWFE

D/A Interface

hv

D-LUMO

hv

δE

ex

Donor

Acceptor

em

A-LUMO

ex

em

ex

D-HOMO

A-HOMO

(a)

(b)

FIGURE 3.4 Second generation donor/acceptor bilayer type organic photovoltaic cell or ‘Tang cell’ with (a) device structure and (b) energy diagram

N

N

N

N

N

O

O

N M N n

N

PPV n

O

Phthalocyanines

n

O MDMO-PPV

MEH-PPV

O S O

n

RO-PPV-10

n

R PTh

N

S

S

N

n

PBZT (D)

FIGURE 3.5 Representative organic/polymeric electron donors (p-type semiconductors)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

80

S-S. SUN

O

O

R N

N

O R

O

O

O

C60

Perylenes R SO2

OR

NC

RO

PCBM

n

n

O

O

O

N

N

N

N

n

R

CN-PPV

BBL

SF-PPV

FIGURE 3.6 Representative organic/polymeric electron acceptors (n-type semiconductors) vacuum –2

vacuum Donor (hole acceptor)

PPV 2.7 eV -

MDMO MEH -PPV -PPV

Acceptor (electron acceptor)

C10O -PPV

-2.83 eV -2.80 eV

–3

-2 –2.6 eV

LUMOs

Ca

-2.91 eV

P3HT P3OT P3DDT –3.53 eV –3.53 eV –3.55 eV

PCBM C CN-PPV 60

–3.75 eV

–3.83 eV

–3

SF-PPV

–3.43 eV

–3.60 eV

–4

–4 Al –4.3 eV

ITO –4.7 eV

–5

PSS-PEDOT –5.2 eV

–6

-5

–4.90 eV –5.20 eV

–5.31 eV

–5.17 eV –5.20 eV

Au –5.1 eV

–5.22eV

–5.29 eV

HOMOs

–5.80 eV –5.89 eV

-6

–6.10 eV –6.10 eV

FIGURE 3.7 Frontier orbital levels of representative polymeric electron donors and acceptors; the work functions of several representative electrodes are also shown on the sides

film thickness is too thin, this results in severe ‘photon losses’. Figure 3.5 shows mostly used or studied organic donors. Figure 3.6 shows mostly used organic acceptors, and Figure 3.7 shows the frontier HOMO/LUMO levels of some commonly used conjugated

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

81

semiconducting polymers. The donor/acceptor (or p/n) double layer solar cells may be called second generation organic cells.

3.3.3 DONOR/ACCEPTOR BLEND ‘BULK HETEROJUNCTION’ SOLAR CELLS Since the donor/acceptor interface is critical for organic exciton dissociation, and the average ‘Frenkel’ exciton diffusion length (AEDL) in most organic semiconductors is between 5 and 50nm, it is therefore desirable that D-A interfaces are within the AEDL dimension everywhere in the bulk of the materials. The donor/acceptor blended/composite or ‘bulk heterojunction’ organic and polymeric solar cells (shown in Figure 3.5) are thus a logical approach. These may be called third generation organic cells.32–40 These cells were fabricated by intimately blending a donor material with an acceptor material mechanically. In this way, the donor/acceptor interface (and so the exciton dissociation sites) would be located everywhere in the bulk, so that it would be more convenient for an exciton generated anywhere in the bulk to reach a nearby donor/acceptor interface and be dissociated into carriers. For instance, it was found that a cell with a D-i-A tri-layer structure, where D is a donor layer, A is an acceptor layer and i represents a donor/acceptor blend layer, had nearly doubled photoelectric power conversion efficiency over that of a corresponding D-A bilayer cell under similar conditions.32 So far, numerous ‘bulk heterojunction’ cells using PPV or polythiophene derivatives (MEH-PPV, MDMO-PPV, P3HT and P3OT, for example) as the donor and CN–PPV or fullerene derivatives (mostly fullerenes such as PCBM) as the acceptor have been intensely studied,33–40 and near unity photo-induced charge separation (also called internal quantum efficiency) and between 1 and 6 per cent photoelectric power conversion efficiencies have been reported under different conditions.33–40 For instance, a 4.4 per cent power conversion efficiency (under NREL calibrated AM 1.5 and 1 sun condition) was recently described on a PCBM/P3HT binary cell with carefully controlled thin film processing and device fabrications.40 The higher efficiencies of D/A blend cells over the previous D-A double layered cells can be attributed mainly to the reduction of ‘exciton loss’ due to a larger donor/acceptor interface, and also to the ‘photon loss’ as the films now can be made thicker to harvest more light. However, even if the charge carrier generation may be efficient, the carrier transport to the electrodes still appears not so easy or might even be worse compared to the bilayer cell – in other words a major problem is ‘carrier loss’. This is due to the fact that the two phases may not be really ‘bicontinuous’ between the two electrodes (Figure 3.5a), in other words a carrier transport pathway may be interrupted easily by the other phase. This would cause carriers to recombine more frequently. Additionally, if the donor and acceptor are in direct contact with both electrodes, carrier recombination at the organic/electrodes interface would be severe, and carrier collection efficiency at electrodes would be poor. One interesting approach was to fabricate a D/A bilayer first and then enable donor and acceptor to partially diffuse into each other to form a D-D/a-d/A-A gradient type cell structure:41 this would increase the D/A interface and at the same time still keep the D/A spatial asymmetry between the two electrodes.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

82

S-S. SUN

LWFE

D

A

SWFE

Vacuum

D-LUMO A-LUMO SWFE LWFE

(a)

D-HOMO A-HOMO

(b)

FIGURE 3.8 Third generation donor/acceptor blend type organic photovoltaic cells or ‘bulk heterojunction cells’ with (a) device structure and (b) energy diagram

3.3.4 DONOR/ACCEPTOR BICONTINUOUS PHASE SEPARATED AND NANO-STRUCTURE ORDERED SOLAR CELL If the carrier transport in a donor/acceptor blend system is poor due to transport pathways disorder, then a donor/acceptor bicontinuous phase separated and nano-structure ordered solar cell as illustrated in Figure 3.6 become desirable.42–45 Since this structure was proposed in a conjugated block copolymer system,42, 43 a number of other approaches such as ‘honeycomb’- style patterned inorganic n-type semiconductors with p-type polymers,46 acceptor-type carbon nano-tubes with donor-type polymers,47 and n-type semiconducting nano-rods with p-type polymers48 have been pursued in order to realize it. In this solar cell type, a donor/acceptor bicontinuous phase separated nano-structure can be in the form of ordered column or cylinder nano-structure or ‘honeycomb’ pattern sandwiched and perpendicularly oriented between the two electrodes, each column or cylinder cross-section diameter controlled to be within the average exciton diffusion length of most organic semiconductors (5–50nm, Figure 3.6a).42, 43 In comparison to the third generation donor/acceptor blend system, the excitons now still easily reach the donor/acceptor interfaces, particularly in the direction perpendicular to the column (Figure 3.6a), yet the carriers now have a continuous or uninterrupted transportation pathway toward their respective electrodes along the direction of the column. In this way the active layer thickness (or lengths of the columns) can be much greater than the average exciton diffusion length, so more photons can be captured. Discotic-shaped small molecules stacked in p/n bicontinuous column style crystalline structures are also desirable. While one molecular phase stacked crystalline structure has already been reported,49, 50 both donor and acceptor bicontinuous stacked columnar structures have yet to be

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

83

demonstrated. Most importantly, these kind of bicontinuous ordered nano-structures may be applied to all organic D/A, inorganic p/n, or any hybrid inorganic/organic p/n binary type solar cells, as the two common features are that the nano-domain junctions dissociate excitons while the continued/uninterrupted channels provide smooth transport pathways for the carriers. These may be called fourth generation organic cells.

D

A SWFE

LWFE

Vacuum D/A Interface

D-LUMO

A-LUMO

D-HOMO

A-HOMO

LWFE-Femi

SWFE-Femi

(b)

(a)

FIGURE 3.9 Fourth generation donor/acceptor bicontinuous phase separated and nano-structured organic photovoltaic cell with (a) device structure and (b) energy diagram

3.3.5 TANDEM STRUCTURED CELLS In inorganic solar cells, one well-known approach to minimize the solar ‘photon loss’ has been to fabricate a cell with multiple serially connected and parallel stacked sub-cells having different energy gaps (also called energy gap graded ‘tandem’ structure cells).1–4, 51, 52 The advantages of the tandem structure include: 1) increased solar photon capture due to both increased overall cell thickness and energy gap grading, and 2) increased open-circuit voltage as a result of Voc serial summation. It is obvious that the optical excitation of the cell should match the energies of the photons to be harvested, and the frontier orbital energy offsets between the donor and acceptor must also be ‘just right’ in order to dissociate the exciton most efficiently and at the same time minimize the charge recombination.44, 45, 53, 54 For this reason, optimizations at energy domain are also critical. Since sunlight radiation spans a wide range from UV all the way to IR, and in each cell either donor or acceptor has only one energy gap and can only capture a very narrow energy range of photons, a tandem-style stacked and serially connected cell structure is therefore desirable, with the energy gaps graded among the stacked cells with an energy range spanning the whole solar spectrum, so that most of the sunlight can be captured.51, 52 Since the open-circuit voltage of each cell is correlated closely to the donor HOMO and acceptor LUMO,55 it is critical to connect the cells in a serial manner so that photovoltage

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

84

S-S. SUN

can be added up. In this way, large solar cell voltages together with a high power conversion efficiency can be achieved.8 For instance, over 30 per cent power conversion efficiencies have been reported in inorganic triple-junction tandem-type solar cell.1–4 For organics, a 5.7 per cent power conversion efficiency (under 1 sun AM 1.5 condition) and an open-circuit voltage of 1.2 volts was reported in a tandem-type PV device containing two serially connected and stacked sub-cell units.56 In this device, though both sub-cells are composed of copper phthalocyanine (CuPc) as the donor and C60 as the acceptor, the front sub-cell was tuned (via optical engineering) to absorb mainly the 650nm wavelength low energy light, while the back sub-cell mainly captures the 450nm wavelength high energy photons due to the optical match issue.56 In an ideal tandem cell, however, it is desirable that the front sub-cells absorb high energy photons while the rear sub-cells capture low energy light.2, 3, 51, 52 In this way, the largest-gapped cell at the front would capture highest energy photons first, but allow lower energy photons to pass through to the lower-gapped cells behind, where the lower energy photos can be captured by lower energy gap sub-cells, and so on. Even if some excitons in the front large-gapped cells do not dissociate and relax to emit a smaller energy photon, that photon can be captured by the next lower-gapped cells. However, current density needs to be balanced between subcells.1–3, 51, 52 As a whole, as the cell could capture most sunlight, it should appear as a dark color.7

3.4 ORGANIC SOLAR CELL FABRICATIONS In a typical organic or polymeric solar cell device, as shown in Figure 3.10, the organic or polymeric semiconductor layer (called the active layer) is typically sandwiched between a transparent conducting electrode (TCE) (for example Indian-Tin-Oxide or ITO coated glass as the LWFE, bottom) and a metal electrode (for example aluminum as the SWFE, top). It has recently been found that a poly(ethylene dioxythiophene):polystyrene sulfonic acid layer (PSS-PEDOT, chemical structure shown in Figure 3.11) greatly facilitates the hole transfer between the active layer and the ITO electrode, and a thin layer LiF greatly facilitates the electron transfer between the active layer and the metal electrode as will be further elaborated below. For small organic molecules, typically high vacuum (at least 10-6 Torr pressure) vapor deposition and occasionally solution crystallization protocols are used to grow thin films on the TCE substrate, followed by the vacuum deposition of the metal electrode on top of the photovoltaic active layer. For polymers, typically solution spin coating (small devices), inkjet printing, or roll-to-roll printing (large sized sheets) protocols can be used. Solution processing generally offers advantages of low cost and convenience on large-scale industrial productions.5, 6 The PSS-PEDOT layer is coated between the ITO glass electrode and the active layer in order to optimize the hole transfer between the active layer and the ITO electrode.5, 6 The conductivity of PSS-PEDOT can reach 80S/cm if produced by electro-polymerization, and can be as low as 0.03S/cm if produced by chemical polymerization. This is believed to be due to different composition.57 Sheet resistance of PSS-PEDOT as low as 350–500 ohm/square was reported.58 It became an extremely attractive material when it was found that the performance and stability of polymer light emitting diodes (LEDs) could be improved by inserting this material between the polymer active layer and ITO

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

85

SWFE, e.g., Al, Ca

LiF

Photovoltaic Active Materials PSS-PEDOT

Transparent LWFE, e.g., ITO Transparent substrate, e.g., glass

Light FIGURE 3.10 General scheme of an organic/polymer solar cell with ‘buffering’ layers

O

O

O S +

S O

O

O

_

O

+ S

S

S O

O

O

S O

O

_

O OH O O S OO S OO S O

FIGURE 3.11 Chemical structure of PSS-PEDOT

glass as a buffer layer.59, 60 PSS-PEDOT is available in an aqueous dispersion, which can form uniform, transparent, conductive film by spin coating. The most important physical properties for its application in devices are the high work function (-5.2eV) and the smooth surface. The improvement of the device performance after PSS-PEDOT layer application may be attributed to a number of factors in either spatial or energy regimes. In the spatial regime, for instance, commercial ITO glass surfaces have been found to be very rough and the conductivities were area sensitive.61 The PSS-PEDOT layer was believed to help smooth both the surface roughness and the conductivities at different spots. In the energy regime, it is believed that the work function of PSS-PEDOT lies between the work function of the ITO (-4.7eV) and the HOMO levels of most organic donor materials. This intermediate level would facilitate the hole transfer between the polymer and the ITO. It is

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

86

S-S. SUN

also possible that PSS-PEDOT simply prevents the acceptor from being in direct contact with the ITO, so charge recombination at the ITO interface is minimized. However, PSS-PEDOT layer thickness should be optimized since too thin a layer may yield many pin holes, while thicker layers cause larger increase in the series resistance.61 Insertion of a thin layer of lithium fluoride (LiF) between the active layer and the SWFE metal electrode has been shown to improve electron injection in some organic LEDs.62 The LiF layer is typically very thin (< 1nm), since thicker layers are found to be detrimental to electron injection. For instance, 3Å thickness of LiF was found to be optimal for both the fill factor and the current density in one system.62 While the exact causes were not very clear, the improved device optoelectronic performance of incorporating LiF in some cases may be attributed to, for instance, the LUMO of the LiF lying between the acceptor LUMO and work function of the metal, or the LiF-modified work function of the metal electrode due to the dipolar nature of LiF layer. Both would facilitate the electron transfer. Another cause might be that the LiF layer prevents the donor from being in direct contact with the metal, or prevents the chemical reactions between the organic active layer and the metal, or reduces the serial resistance of the interface between the metal and the active layer.61 The equivalent circuit of an organic/polymeric solar cell J-V curves can be approximated by:63 (1) where J0 is the saturation dark current density, q is the elementary charge, n is the diode ideality factor, V is the applied voltage of the cell, RS and RP are the serial and parallel resistivity, Jsc is the short-circuit photocurrent density (A/cm2), k is the Boltzman constant, and T is the absolute temperature. For a pin-hole free organic active layer, the parallel (shunt) resistivity was approximately the same for many organic photodiodes (typically over 1k1), but the total serial resistivity (RS) may vary depending on materials and device fabrications. In one device, for instance, Rs decreased from 101 with no LiF layer to 41 for LiF thickness between 3 and 9Å thick and increased to 51 for LiF 12–15Å thick for one cell.63 In comparison to inorganic solar cells, organics typically have very large RS due to the typically very poor orbital overlaps. When characterizing and comparing organic solar cells, like in inorganic solar cells, the open-circuit voltage (Voc), short-circuit current density (Jsc), the fill factor (FF), and the overall photoelectric power conversion efficiencies (d) are critical parameters. Like in inorganic solar cells, the open-circuit voltage (Voc) can be experimentally obtained from the current–voltage curves of an illuminated device when the current is zero. Because of the special charge generation and separation mechanisms in organic devices, the efficiencylimiting factors are therefore distinct from those in conventional inorganic solar cells (for example silicon p-n junction solar cells). For instance, while the maximum photovoltages (Voc) achievable in silicon cells are generally limited to the magnitude of the built-in potential, it is common to observe experimentally Voc greater than the built-in potential in organic-based photovoltaic devices.55 It is believed the Voc of an organic donor/acceptor binary cell is closely correlated to the HOMO of the donor and the LUMO of the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

87

acceptor.55, 64 The short-circuit current density (Jsc) is the photocurrent density value under zero applied bias. It is a result of the photo-induced charge separation (or exciton dissociation) and the charge transport driven by the internal field and the chemical potentials of the material. The photoelectric power conversion efficiency (PCE or dPCE) is the ratio of cell maximum output electrical power (Pout) over the total radiated input optical power (Pin):

(2) where a fill factor (FF) is related to the percentage of maximum electrical power that can be extracted from the solar cell. The fill factor for devices is described as

(3)

where Jm and Vm are the values for the current density and voltage at maximum power Pout/max = (JV)max. To achieve a high FF factor in organic solar cells, it is desirable that the parallel resistivity is large to prevent short or leakage current, while the serial resistivity is small to enable a large forward current. The series resistivity simply adds up from all series resistive contributions in the device, including both carrier transport in materials and interface charge transfer. One key problem in organics is the very high series resistance due to poor molecular frontier orbital overlaps, so charges have to ‘hop’ between molecules instead of diffuse in ‘bands’. However, molecular self-assembly may solve this problem in the future.

3.5 ORGANIC SOLAR CELL OPTIMIZATIONS The overall photoelectric power conversion efficiency of an organic/polymeric solar cell is determined by at least the following five critical steps: 1 2 3 4 5

photon capture and exciton generation; exciton diffusion to the donor/acceptor interface; exciton dissociation and/or carrier generation at donor/acceptor interface; carrier diffusion to the respective electrodes; and carrier collection by the respected electrodes.

For all currently reported organic/polymeric photovoltaic materials and devices, none of the above mentioned five steps have been optimized. For instance, in the first step, the ‘photon loss’ is heavy due to most organic semiconductors having too high energy gaps (typically larger than 2eV), meaning that only a very small fraction of the sunlight could be captured. In the second step, the ‘exciton loss’ is severe unless a ‘phase separated and ordered nano-structure’ can be materialized. In the third step, both the ‘exciton loss’ and ‘carrier loss’ are severe unless the frontier orbital energy offsets are at their optimal value

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

88

S-S. SUN

so that the exciton dissociation is maximized and at the same time the charge recombination is minimized. In the fourth step, the ‘carrier loss’ is notoriously severe and well known in most organic/polymeric semiconductors due to the typically poor orbital overlaps, so that sizable ‘bands’ are difficult to form, and carrier transport is mainly through inter- and intra-molecular ‘hopping’ mechanisms. In the last step, the ‘carrier loss’ may be severe if the energy levels or chemical states between the organic semiconductors and the electrodes are not optimized. It is thus not surprising that the power conversion efficiencies of most currently reported organic or polymeric solar cells are relatively small in comparison to their inorganic counterparts. However, there is plenty of room to improve, and the optimizations should be done in both energy and spatial domains.

3.5.1 OPTIMIZATIONS IN THE SPATIAL DOMAIN As mentioned earlier, in the spatial domain, a donor/acceptor ‘bicontinous phase separated and nano-structure ordered’ morphology appears ideal in order to minimize both the ‘exciton loss’ and the ‘carrier loss’.42–45 As mentioned in section 3.4, there are several approaches being pursued that could potentially realize this structure. In the cases of carbon nano-tubes and semiconducting nano-rods, the main challenges are to fabricate cells with uniformly distributed and well-aligned tubes/rods perpendicular to the horizontal conducting substrates. In the cases of any techniques using block copolymers, the main challenge lies in both synthetic chemistry and block copolymer processing. Block copolymer solid melts are well known to exhibit behavior similar to conventional amphiphilic systems such as lipid–water mixtures, soap, and surfactant solutions.65, 66 The covalent bond connection between different blocks imposes severe constraints on possible equilibrium states, this results in unique supra-molecular nano-domain structures such as lamellae (LAM), hexagonally (HEX) packed cylinders or columns, spheres packed on a body-centered cubic lattice (BCC), hexagonally perforated layers (HPL) and at least two bicontinuous phases: the ordered bicontinuous double diamond phase (OBDD) and the gyroid phase.65–67 The morphology of block copolymers is affected by chemical composition, block size, temperature, processing and other factors. Clearly, the block copolymer approach to photovoltaic function offers some intrinsic advantages and has attracted a number of research efforts.42–45 For instance, block copolymers containing a conjugated PPV donor block with a non-conjugated polystyrene acceptor block derivatized with fullerenes have been investigated, and phase separations at nano-scale were indeed observed in some cases.68, 69 However, the ‘carrier loss’ problem in the non-conjugated polystyrene phase would be a problem compared to the / conjugated main chain systems. On the other hand, when a conjugated donor block was connected directly to a conjugated acceptor block to form a direct p-n type conjugated diblock copolymer, while energy transfer from higher gap block to lower gap block was observed, no charge separated states (which are critical for photovoltaic functions) were detected.70 A –DBABtype of block copolymer and its potential ‘tertiary’ supra-molecular nano-structure was therefore designed (Figures 3.12–3.14),42–44 where D is a / electron conjugated donor block, A is a / electron conjugated acceptor block, and B is a non-conjugated and flexible bridge unit. Additionally, the flexibility of the bridge unit enables the rigid donor and acceptor conjugated blocks more easily to self-assemble, phase separate and become

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

89

less susceptible to distortion of the main chain conjugation (Figures 3.12 and 3.13). Since both donor and acceptor blocks are / electron conjugated chains, if they are selfassembled in planes perpendicular to the molecular plane like a /–/ stacking morphology well known in a typical / conjugated system (Figure 3.13),71–73 good carrier transport in both donor and acceptor phases would become feasible.

B

D

A

B

Paradigm Paradigm Paradigm Paradigm

Paradigm Paradigm Paradigm Paradigm Paradigm

B U

Paradigm Paradigm Paradigm Paradigm Paradigm - Paradigm

Source: S. Sun, Photovoltaic Devices Based on a Novel Block Copolymer, US patent publication # 20040099307

FIGURE 3.12 Scheme of a –DBAB- type block copolymer ‘primary structure’

B

D A

B

B

Source: S. Sun, Photovoltaic Devices Based on a Novel Block Copolymer, US patent publication # 20040099307

FIGURE 3.13 Scheme of a potential –DBAB- type block copolymer ‘secondary structure’

While the –DBAB- block copolymer backbone structure may be called ‘primary structure’ (Figure 3.12), the conjugated chain / orbital closely stacked and ordered morphology may be called ‘secondary structure’ (Figure 3.13). This ‘secondary structure’ has been known to possess dramatically enhanced charge carrier mobility as demonstrated in, for instance,

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

90

S-S. SUN

crystalline / orbital stacked molecular rubrene,10 self-assembled regioregular polythiophenes,72 or template aligned poly-p-phenylenevinylenes.73 Finally, through the adjustment of block size, block derivatization, and thin film processing protocols, a ‘tertiary structure’ (Figure 3.14) could be obtained, where a ‘bicontinuous’ columnar (or ‘HEX’) type of morphology of the donor and acceptor blocks vertically aligned on top of the substrate and sandwiched between two electrodes may be formed.6 Even better, a thin donor layer could be inserted between the ITO and the active ‘HEX’ layer (preventing the acceptor from being in direct contact with the ITO), and a thin acceptor layer could be inserted between the metal electrode and the active layer (preventing the donor from being in direct contact with metal) (Figure 3.9a).42 The terminal donor and acceptor layers not only prevent charge easy recombination at the electrodes, they also enable a desired asymmetry and favorable chemical potential gradient for asymmetric (selective) carrier diffusion and collection at respective electrodes.30, 31 Since the diameter of each donor or acceptor block column can be conveniently controlled via design, synthesis and processing to be within the average exciton diffusion length (AEDL) of 5–50nm, every photo-induced exciton will be in convenient reach of a donor/acceptor interface along the direction perpendicular to the column. At the same time, photogenerated charge carriers can diffuse more smoothly to their respective electrodes via a truly ‘bicontinuous’ block copolymer columnar morphology. Additionally, with appropriate adjustment of donor and acceptor block sizes and their substituents, energy levels, gaps, and tandem device structures, it is expected that the ‘photon loss’, the ‘exciton loss’, and the ‘carrier loss’ (including charge recombination) issues can all be addressed and optimized simultaneously in one such block copolymer photovoltaic device. To examine the feasibility of this design,42, 43 a series of –DBAB- type block copolymers has been developed.44, 45, 74–80 Preliminary experimental results have indeed shown the photovoltaic properties of the –DBAB- block copolymer are much better than the corresponding D/A blend cells.44, 45, 79, 80

D A D

D A D A

A

D

A

'HEX' Columnar Morphology Source: S. Sun, Solar Energy Materials and Solar Cells, 79, pp257–264 (2003)

FIGURE 3.14 Scheme of a potential –DBAB- type block copolymer ‘tertiary structure’

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

91

3.5.2 OPTIMIZATIONS IN THE ENERGY DOMAIN ‘Photon loss’, ‘exciton loss’ and ‘carrier loss’ are all critically related to the frontier orbital levels of the donor, the acceptor and the work functions of the two electrodes. In sunlight harvesting applications, the solar radiation spectrum is very broad, from UV all the way to IR, with maximum photon flux ranging between 1.3 and 2.0eV on the surface of the Earth (air mass 1.5) and between 1.8 and 3.0eV in outer space (air mass zero).5–6 For optical telecommunications and signal processing, an optical excitation energy gap of 0.8eV (for 1.55 +m IR signal) is needed. Energy excitation gaps (energy difference between the HOMO and LUMO) in both the donor and the acceptor should be fine tuned to match the photon energy, as both can absorb photons and incur charge separation at the donor/acceptor interface. The questions are: 1) how to capture as much solar photons as possible while each material (donor or acceptor) has typically one energy gap that can only capture a very narrow energy range of photons; and 2) what would be the optimal frontier orbital level offsets between the donor and the acceptor, which is the key driving force for exciton dissociation. For the first question, the best answer would be a tandem-type cell structure with energy gap grading as has been discussed extensively in the literature.1–4, 49, 50 For instance, over 30 per cent power conversion efficiencies have been reported in inorganic tandemtype solar cells,1–4 and a 5.7 per cent power conversion efficiency was reported in a two sub-cell stacked tandem-type organic device.56 However, it is obvious the more cells with more energy gaps, the smaller the ‘photon losses’. For the second question, the current widely cited view is that the frontier orbital energy offset between the donor and the acceptor should be no less than the exciton binding energy EB (in other words the minimum energy needed to overcome the electric Coulomb forces and dissociate the ‘Frenkel’ type exciton into a separate or ‘free’ electron at acceptor LUMO and a ‘free’ hole at donor HOMO).81 Indeed when the LUMO energy offset is too small, photo-induced charge separation appears to become inefficient.82 On the other hand, if the energy offset is too large, the Marcus ‘inverted’ region would slow down charge separation,83–89 and thermal ‘ground state’ charge separation without photo excitation may also occur.7 These are not desirable for light harvesting applications. A large energy offset also reduces open-circuit voltages.55 Therefore an analysis of optimal donor/acceptor energy offset is very critical and necessary44, 45, 51, 52 and is briefly summarized here. For solar cell purposes, the photo-induced charge separated state (Figure 3.15, after steps 3 or 7) is the desired starting point. However, the exciton charge separation is also competing with exciton decay (Figure 3.15, steps 2 and 6). The ratio of charge separation rate constant (ks) versus exciton decay rate constant (kd) can be defined as exciton quenching parameter (EQP, mathematically represented as Yeq) as:

YeqX = ksX / kdX

(4)

Here X=D (donor) or A (acceptor). The parameter Yeq reflects to a certain degree the efficiency of exciton–— > charge conversion. It was experimentally observed that the charge separation could be orders of magnitude faster then the exciton decay in MEHPPV/fullerene binary systems.33, 34 Secondly, the charge separations (steps 3 and 7) are

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

92

S-S. SUN

also competing with charge recombination (step 4). The ratio of charge separation rate constant (ks) over charge recombination rate constant (kr) can therefore be defined as recombination quenching parameter (RQP, mathematically represented as Yrq):

YrqX = ksX / kr

(5)

For any light harvesting applications, such as solar cell applications, it is desirable that both Yeq and Yrq parameters are as large as possible. From semi-classical electron transfer theory, when LUMO orbital energy offset bE is set as variable (Figure 3.15),83–89 and based on measured, calculated and estimated parameters of RO-PPV/SF-PPV-I donor/acceptor pairs developed recently,42, 45 a plot of normalized YeqD, kr, and YrqD versus bE is shown in Figure 3.16.51, 52 As Figure 3.16 shows, when bE varies, kr, YeqD and YrqD all exhibit their own maximum values. For the RO-PPV/SFPPV-I pair, the fastest photo-induced charge separation occurs when the RO-PPV/SF-PPVI LUMO offset (bE, driving force) equals the sum of charge separation reorganization energy and the donor exciton Coulombic potential energy.51, 52 Also, the fastest charge recombination occurs at a LUMO offset far away from the optimum charge separation offset as well as the actual RO-PPV/SF-PPV-I offset. Therefore, the charge recombination in the RO-PPV/SF-PPV-I pair does not seem to be of a major concern as long as the LUMO offset is nearby the bEeqD. Figure 3.16 also shows the recombination quenching parameter YrqD (ksD/kr) does not reach its maximum until a positive energy offset. At this positive energy offset, the photo-induced charge separation might be too slow to be attractive for efficient photovoltaic function; therefore, the positive bErqD value appears not critical in this particular case. It is desirable that through careful molecular design, the bErqD is

Vacuum 3

D-LUMO

EA

E

A-LUMO IP

2

1 4 6

D-HOMO

5

7

A-HOMO D/A Interface FIGURE 3.15 Scheme of molecular frontier orbitals and photo-induced electron transfer as well as Dexter energy transfer processes in a donor/acceptor binary light harvesting system

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

ET Rates/Rate Ratios (Normalized)

Kr

–3.5

–2.5

–1.5

Yeq(D)

–0.5

93

Yrq(D)

0.5

1.5

Donor/Acceptor Frontier Orbital Energy Offset (δE/eV) Source: S. Sun, Sol. Energy Mat. Sol. Cells, 85, 261-267 (2005)

FIGURE 3.16 Donor RO-PPV exciton quenching parameter (Yeq(D)=ks(D)/kd(D), middle solid curve), charge recombination rate constant (Kr, left long dashed curve), and charge recombination quenching parameter (Yrq(D)=ks(D)/kr(D), right short dashed curve) versus LUMO offset of RO-PPV/SF-PPV-I pair

coincident with or close to bEeqD, and that bErD is far away from bEeqD. Similar results can also be deducted for the acceptor and for the donor/acceptor pair.44, 52 This donor/acceptor charge separation model can readily be used in the impurity or defect ‘photo doping’ cases (including the single layer ‘Schottky junction’ cell), as the impurities or structural defects can be either donor or acceptor type in the energy domain. The model can also be used in ground state ‘chemical doping’ cases.7 ‘Chemical doping’ typically refers to a mobile or free charge carrier generation phenomenon when one material is mixed or in direct contact with a second material at room temperature without the intentional use of strong external excitation forces such as light or high heat. This is sometimes also called ‘ground state charge separation’. However, there are two different situations involved. In the first situation, the acceptor LUMO is lower then the donor HOMO. The energy offset bE between the D-HOMO and A-LUMO therefore acts as a key driving force (in addition to other potential driving forces such as thermal force kT) for the electron transfer from the donor HOMO directly to the acceptor LUMO. In this case, the most efficient or optimal charge separation would occur when the energy offset bE plus the thermal force kT equals the charge separation reorganization energy and the attractive Coulomb force. Even at absolute zero temperature when kT=0, the electron transfer still occurs due to the presence of the driving bE. However, in the second scenario, the acceptor LUMO might be the same or a little higher then the donor HOMO and the thermal driving force kT becomes critical. In this case, the most efficient electron transfer would occur when the driving kT is equal to the charge separation reorganization energy plus the Coulomb attractive force plus the D-HOMO/A-LUMO energy offset increase. Because of the kT driving force involved, this process actually is not really ‘ground state’ electron transfer, though it may occur at room temperature. Thermally induced chemical doping mechanism may also be used to explain static electrical charge generation when two different materials are rubbed against each other, as the friction at the interface might

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

94

S-S. SUN

generate enough thermal energy to incur electron transfer. In chemical doping cases, since the dopant is typically a minority component and it traps one charge carrier, the opposite charged carrier at non-dopant majority phase then becomes the mobile or free charge carrier and can transport at the majority component phase.

3.6 CONCLUSIONS AND FUTURE PERSPECTIVES The current relatively small (less than 6 per cent) photoelectric power conversion efficiencies of organic and polymeric photovoltaic devices can be attributed mainly to the severe ‘photon loss’, ‘exciton loss’ and ‘carrier loss’ due to materials’ improper frontier energy levels and gaps, energy offsets between the donors and the acceptors, poor material morphologies, and cell structures and fabrications that are not optimized. However, there are plenty of areas where improvements might be made. Optimizations in both spatial and energy/time domains should be pursued simultaneously in order to achieve high efficiency organic and polymeric photovoltaic materials and devices. In the spatial domain, the ideal structure or morphology of a single cell appears to be a donor/acceptor nano-phase separated and bicontinuous ‘columnar’ type morphology, where the diameter of each column phase should be within the average exciton diffusion length (AEDL) of 5–50nm depending on the materials involved, and with the columns aligned perpendicular to the electrode planes. Among the several approaches mentioned, a –DBAB- type of block copolymer and its potential self-assembled columnar ‘tertiary’ supra-molecular nano-structure has been proposed and preliminarily examined experimentally with promising results. In this system, along the carrier transport direction which is perpendicular to the two electrode surfaces, it is ‘bicontinuous’ of the two phases, so both positive and negative charges have smooth transport pathways. Yet in the direction parallel to the electrode planes, it has donor/acceptor phase separated morphology, and each phase diameter is within the AEDL. In the energy/time domain, first the optical excitation energy gaps in each donor and acceptor phase should match the intended photon energy. For sunlight harvesting, an energy gap graded tandem-type cell structure is desirable to minimize ‘photon losses’. Second, the optimal donor/acceptor energy offset that is the critical driving force for photoinduced charge separation should be identified and materialized. Specifically, in an electron transfer dynamic regime, there exists an optimal donor/acceptor LUMO (or HOMO) level offset where exciton-charge conversion is most efficient (or exciton quenching parameter EQP reaches its maximum), and another optimal LUMO (or HOMO) level offset can be identified where charge recombination is relatively slow compared to charge separation (or recombination quenching parameter RQP become largest). Though the cell efficiency is mainly affected by the EQP, the molecules should be designed and developed such that the maximum RQP is close to or coincides with maximum EQP. There also exists a third energy offset where the charge recombination becomes most severe. The molecules should be designed and developed such that this worst charge recombination is far away from maximum EQP. These orbital offset values are critically important in molecular structure and energy level fine tuning. Finally, since sunlight is a very broad radiation with photon energy ranging from UV all the way to IR, a tandem-style serially connected and parallel stacked cell structure with energy gaps gradually descending from UV to IR along

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

95

the light radiation direction appears ideal. This would enable a broad capture of most solar photons and the summation of induced photovoltages. With optimizations in both space and energy/time domains, it is expected that high efficiency photoelectric power conversion efficiency organic light harvesting systems, including organic photovoltaic cells, photo detectors or any artificial photo-charge synthesizers/converters, can be realized. The dream of renewable, clean, inexpensive, portable and low cost energy supply can be a reality.

ACKNOWLEDGEMENTS The author would like to thank all those involved in the research cited in the literature and for the research/educational grant supports from a number of funding agencies including NASA, the Air Force Office of Scientific Research, the National Science Foundation, the Department of Education (Title III award) and the Dozoretz foundation.

AUTHOR CONTACT DETAILS Sam-Shajing Sun, Center for Research & Education in Advanced Materials and Chemistry Department, Norfolk State University, 700 Park Avenue, Norfolk, VA 23504, US Tel: 757-823-2993; Fax: 757-823-9054; e-mail: [email protected]

REFERENCES 1 M. D. Archer and R. Hill (eds) (2001) Clean Electricity From Photovoltaics, Imperial College Press, London 2 A. Luque and S. Hegedus (eds) (2003) Handbook of Photovoltaic Science and Engineering, John Wiley, Chichester, UK 3 M. A. Green (2003) Third Generation Photovoltaics: Advanced Solar Energy Conversion, Springer, Berlin 4 A. Hepp, S. Bailey and R. Raffaelle (2005) ‘Inorganic photovoltaic materials and devices: Past, present, and future’ in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p19 5 C. Brabec, V. Dyakonov, J. Parisi, and N. Sariciftci (2003) Organic Photovoltaics: Concepts and Realization, Springer, Berlin 6 S. Sun and N. S. Sariciftci (eds) (2005) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US 7 S. Sun (2006) ‘Organic and polymeric solar cells’, in H. S. Nalwa (ed) Handbook of Organic Electronics and Photonics, American Scientific Publishers, Los Angeles, California, US, Vol 2, Chapter 23 8 A. J. Heeger (2000) ‘Semiconducting and metallic polymers: the fourth generation of polymeric materials’, Nobel Lecture in Chemistry. 9 Y. I. Frenkel (1931) Phys. Rev., no 37, p1276 10 R. Blankenship (2005) ‘Natural organic photovoltaic solar energy transduction’, in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p37 11 V. Podzorov, E. Menard, A. Borissov, V. Kiryukhin, J. A. Rogers and M. E. Gershenson (2004), Phys. Rev. Lett., 93, 086602. 12 Pochettino (1906) Acad. Lincei Rendiconti, no 15, pp355–363 13 R. Bube (1960) Photoconductivity of Solids, Wiley, New York 14 S. Anthoe (2002) Rom. Rep. Phys., no 53, p427 15 C. W. Tang and A. C. Albrecht (1975) J. Chem. Phys., no 62, p2139

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

96

S-S. SUN

16 B. R. Weinberger, M. Akhtar and S. Gau (1982) Synth. Metals, no 4, p187 17 S. Glenis, G. Tourillon and F. Garnier (1986) Thin Solid Films, no 139, p221 18 S. Karg, W. Riess, V. Dyakonov and M. Schwoerer (1993) Synth. Metals, no 54, p427 19 R. Marks, J. Halls, D. Bradley, R. Friend and A. Holmes (1994) J. Phys.: Cond. Matter, no 6, p1379 20 H. Antoniadis, B. Hsieh, M. Abkowitz, S. Jenekhe and M. Stolka (1994) Synth. Metals, no 62, p265 21 E. Frankevich et al (1992) Phys. Rev. B, no 46, p9320 22 L. Rothberg et al (1996) Synth. Metals, no 80, p41 23 C. W. Tang (1986) Appl. Phys. Lett. no 48, pp183–185 24 Y. Yamashita, W. Takashima and K. Kaneto (1993) Jpn. J. Appl. Phys., no 32, p1017 25 N. Sariciftci et al (1993) Appl. Phys. Lett., no 62, p585 26 S. Morita, A. Zakhidov and K. Yoshino (1993) Jpn. J. Appl. Phys., no 32, p873 27 J. Halls et al (1996) Appl. Phys. Lett., no 68, p3120 28 K. Tada et al (1999) Synth. Metals, no 102, p982 29 M. Granstrom, K. Petritsch, A. Arias, A. Lux, M. Andersson and R. Friend (1998) Nature, no 395, p257 30 B. Gregg (2005) ‘Coulomb forces in excitonic solar cells’, in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p139 31 B. A. Gregg (2003) J. Phys. Chem. B, no 107, pp4688–4698 32 M. Hiramoto, H. Fujiwara and M. Yokoyama (1991) Appl. Phys. Lett., no 58, pp1062–1064 33 N. S. Sariciftci, L. Smilowitz, A. J. Heeger and F. Wudl (1992) Science, no 258, p1474 34 B. Kraabel, J. Hummelen, D. Vacar, D. Moses, N. Sariciftci, A. Heeger and F. Wudl (1996) J. Chem. Phys., no 104, pp4267–4273 35 G. Yu, J. Gao, J. Hummelen, F. Wudl and A. Heeger (1995) Science, no 270, pp1789–1791 36 J. Halls et al (1995) Nature, no 376, p498 37 L. S. Roman, M. Anderson, T. Yohannes and O. Inganas (1997) Adv. Mater., no 9, p1164 38 H. Hoppe and N. S. Sariciftci (2004) J. Mater. Res., no 19, pp1924–1945 39

H. Hoppe and N. Sariciftci (2005) ‘Bulk heterojunction solar cells’, in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p217

40 G. Li, V. Shrotriya, J. Huang, Y. Yao, T. Moriarty, K. Emery and Y. Yang (2005) Nature Materials, no 4, pp864–868 41 M. Drees, R. Davis and R. Heflin (2005) ‘Polymer-fullerene concentration gradient photovoltaic devices by thermally controlled interdiffusion’, in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, (2005) p559 42 S. Sun, Photovoltaic Devices Based on a Novel Block Copolymer, US patent publication no 20040099307 (application dated 14 November 2002) 43 S. Sun (2003) Sol. Energy Mat. Sol. Cells, no 79, pp257–264 44 S. Sun, Z. Fan, Y. Wang and J. Haliburton (2005) J. Mater. Sci., no 40, pp1429–1443 45 C. Zhang, S. Choi, J. Haliburton, R. Li, T. Cleveland, S. Sun, A. Ledbetter, C. Bonner (2006) Macromolecules, no 39, pp4317-4326. 46 K. Coakley and M. McGehee (2004) Chem. Mater., no 16, pp4533–4542 47 M. Jin and L. Dai (2005) ‘Vertically aligned carbon nanotubes for organic photovoltaic devices’, in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p579 48 Y. Kang, N. Park and D. Kim (2005) Appl. Phys. Lett., no 86, 113101. 49 L. Schmidt-Mende A. Fechtenkötter, K. Müllen, E. Moons, R. H. Friend and J. D. MacKenzie (2001) Science, no 293, p1119

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

Recent Progress of Organic Photovoltaics

97

50 B. Kippelen, S. Yoo, J. Haddock, B. Domercq, S. Barlow, B. Minch, W. Xia, S. Marder and N. Armstrong (2005) ‘Liquid crystal approaches to organic photovoltaics’, in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p271 51 G. V. Tsarenkov (1975) Sov. Phys. Semicond., vol 9, no 2, pp166–171 52 G. Sassi (1989) J. Appl. Phys., no 65, pp322–328 53 S. Sun (2005) Sol. Energy Mat. Sol. Cells, no 85, pp261–267 54 S. Sun (2005) Mater. Sci. Eng. B., vol 116, no 3, pp251–256 55 C. J. Brabec, A. Cravino, D. Meissner, N. S. Sariciftci, T. Fromherz, M. Minse, L. Sanchez and J. C. Hummelen (2001) Adv. Funct. Mater., no 11, pp374–380 56 J. Xue, S. Uchida, B. Rand and S. R. Forrest (2005) Appl. Phys. Lett., no 86, p5757 57 X. Crispin, S. Marciniak, W. Osikowicz, G. Zotti, A. W. D. van der Gon, F. Louwet, M. Fahlman, L. Groenendaal, F. De Schryver and W. R. Salaneck (2003) Journal of Polymer Science Part B – Polymer Physics, no 41, pp2561–2583 58 A. J. Mäkinen, I. G. Hill, R. Shashidhar, N. Nikolov and Z. H. Kafafi (2001) Applied Physics Letters, no 79, pp557–559 59 Y. Cao, G. Yu, C. Zhang, R. Menon and A. J. Heeger (1997) Synthetic Metals, no 87, pp171–174 60 S. A. Carter, M. Angelopoulos, S. Karg, P.J. Brock and J.C. Scott (1997) Applied Physics Letters, no 70, pp2067–2069 61 A. Djuristic and C-Y. Kwong (2005) in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p453 62 W. J. H. van Gennip, J. K. J. van Duren, P. C. Thüne, R. A. J. Janssen and J. W. Niemantsverdriet (2002) J. Chem. Phys., no 117, pp5031–5035 63 P. Lane and Z. Kafafi (2005) in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p49 64 Y. Gao (2005) in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p421 65 N. Hadjichristidis, S. Pispas and G. Floudas (eds) (2003) Block Copolymers: Synthetic Strategies, Physical Properties, and Applications, John Wiley & Sons, Inc., New York 66 M Lazzari, and M. Lopez-Quintela (2003) Adv. Mater., no 15, pp1584–1594 67 F. S. Bates and G. H. Fredrickson (1990) Ann. Rev. Phys. Chem., no 41, pp525–557 68 B. de Boer, U. Stalmach, P. F. van Hutten, C. Melzer, V. V. Krasnikov and G. Hadziioannou (2001) Polymer, vol 42, no 21, pp9097–9109 69 R. Segalman, C. Brochon and G. Hadziioannou (2005) ‘Solar cells based on diblock copolymers: a PPV donor block and a fullerene derivatized acceptor block’, in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p403 70 X. L. Chen and S. A. Jenekhe (1996) Macromolecules, no 29, p6189 71 T. A. Skotheim, R. L. Elsenbaumer and J. R. Reynolds (eds) (1998) Handbook of Conducting Polymers, 2nd ed, Marcel Dekker, New York 72 Z. Bao, A. Dodabalapur and A. J. Lovinger (1996) Appl. Phys. Lett., no 69, p4108 73 T. Nguyen, J. Wu, V. Doan, B. Schwartz and S. H. Tolbert (2000) Science, no 288, pp652–656 74 S. Sun, Z. Fan, Y. Wang, C. Taft, J. Haliburton and S. Maaref (2002) in D. Fichou and K. Kafafi (eds) Organic Photovoltaics II, SPIE, Bellingham, WA, US, vol 4465, pp121-128 75 S. Sun, Z. Fan, Y. Wang, J. Haliburton, C. Taft, K. Seo and C. Bonner (2003) Syn. Met., no 137, pp883–884 76 S. Sun (2004) in P. Lane and Z. Kafafi (eds) Organic Photovoltaics IV, SPIE, Bellingham, WA, US, vol 5215, pp195–205 77 S. Sun (2004) in P. Lane and Z. Kafafi (eds) Organic Photovoltaics V, SPIE, Bellingham, WA, US, vol 5520, pp126–135 78 S. Thomas, C. Zhang and S. Sun (2005) J. Poly. Sci. (A) Polym. Chem., vol 43, pp4280–4287 79 S. Sun, Z. Fan, Y. Wang, K. Winston and C. E. Bonner (2005) Mater. Sci. Eng. B., vol 116, no 3, pp279–282

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

98

S-S. SUN

80 S. Sun, C. Zhang, J. Haliburton, A. Ledbetter, C. Bonner, M. Drees and N. Sariciftci (2005) in P. Lane and Z. Kafafi (eds) Organic Photovoltaics VI, SPIE, Bellingham, WA, US, vol 5938, pp59380D1-8 81 M. Knupfer (2003) Appl. Phys. A, no 77, pp623–626 82 S. Sensfuss and M. Al-Ibrahim (2005) in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, p529 83 R. A. Marcus (1993) Rev. Mod. Phys., no 65, pp599–610 84 V. Balzani (ed) (2000) Electron Transfer in Chemistry, Wiley-VCH, New York 85 E. Peeters, P. Hal, J. Knol, C. Brabec, N. Sariciftci, J. Hummelen and R. Janssen (2000) J. Phy. Chem. B, no 104, pp10174–10190 86 E. Neuteboom, S. J. Meskers, P. van Hal, J. van Duren, E. Meijer, R. Janssen, H. Dupin, G. Pourtois, J. Cornil, R. Lazzaroni, J. Brédas and D. Beljonne (2003) J. Am. Chem. Soc., no 125, pp8625–8638 87 J. Cornil, V. Lemaur, M.C. Steel H. Dupin, A. Burquel, D. Beljonne and J. L. Brédas (2005) ‘Electronic structure of organic photovoltaic materials: modeling of exciton-dissociation and charge-recombination processes’, in S. Sun and N. S. Sariciftci (eds) Organic Photovoltaics: Mechanisms, Materials and Devices, CRC Press, Boca Raton, Florida, US, pp161–182 88 J. Miller, L. Calcaterra and G. Closs (1984) J. Am. Chem. Soc., no 106, pp3047–3049 89 A. Weller (1982) ‘Photoinduced electron transfer in solution’, Phys. Chem., no 133, pp93–98

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 74–98

4

Thermal and Material Characterization of Immersed Heat Exchangers for Solar Domestic Hot Water Jane H. Davidson, Susan C. Mantell and Lorraine F. Francis

Abstract This chapter provides an overview of methods to characterize the thermal performance of heat exchangers immersed in thermal storage vessels intended for solar domestic water heating systems and to assess the mechanical durability and scaling potential of polymers for this application. Recent research at the University of Minnesota is summarized and recommendations are made for future research to further the development of polymer heat exchangers and piping.

■ Keywords – heat exchangers; domestic hot water; integral collector storage; polymers; natural convection; nylon; heat transfer

4.1 INTRODUCTION Immersed heat exchangers are common in European combistore water storage tanks for combined domestic hot water and space heating, but in the US, low temperature solar thermal systems are used almost exclusively for domestic hot water and immersed heat exchangers are atypical. This situation is expected to change as lower cost and simplified solar domestic hot water (SDHW) systems are developed for the US market. The recent focus of US efforts to reduce system cost is an indirect integral collector storage (ICS) system (Figure 4.1). An immersed heat exchanger is required to discharge the unpressurized solar storage. During discharge, the domestic pressurized water flows through the heat exchanger and is delivered to the conventional water heater. Because the load-side flow is driven by the domestic water pressure, no mechanical pump is required. Recent research and development for this system, as well as for systems intended for cold climates, addresses the barriers and opportunities to shift from copper and glass components to integrated systems manufactured using mass production techniques, such as those associated with polymeric materials. Significant cost savings

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

100

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

GLOSSARY OF TERMS USED IN THIS PAPER a A C cp D E Er f g h i k Ksp L mj n NuD Nu*D NTU o P Pr . Q r’ Ri Ro Rw RaD RaD s S t t0 Tf,in Tf,out Ts Ts,0 Tw U UA Vr Vs . V w _

ionic activity, units depend on constituent heat transfer surface area of each heat exchanger tube, m2 empirical constant specific heat of water, J/kg•K outer diameter of heat exchanger tube delivered energy, J ratio of energy delivered by divided storage to energy delivered by undivided storage subscript used to indicate heat exchange fluid gravitational constant, m/s2 convective heat transfer coefficient, W/m2·K subscript indicating inside of tube or inner diameter thermal conductivity, W/m·K solubility product for crystalline CaCO3, moles/l length of heat exchanger tube (per compartment), m heat exchanger mass flow rate, kg/s number of heat exchanger tubes Nusselt number, hD/kf average Nusselt number with initial storage temperature as reference temperature number of transfer units subscript indicating outside of tube tube pitch, or center to center spacing, m Prandtl number, p/_ heat transfer, W radial distance from the tube center, m convective resistance of forced flow through the heat exchanger, K/W convective resistance of natural convection on the outer or storage side of the heat exchanger, K/W conductive resistance across the heat exchanger wall, K/W Rayleigh number, g`D3(6T)/(_i) initial Rayleigh number, g`D3(Tw –Ts,0)/(_i) subscript indicating storage side medium supersaturation of water with respect to calcium carbonate, time, s time scale, s water temperature at the inlet of the heat exchanger, K water temperature at the outlet of heat exchanger, K average storage temperature, K initial storage temperature, K temperature of the outer tube wall, K velocity scale, m/s overall heat transfer coefficient-area product, W/K ratio of hot water volumetric output to storage volume volume of storage fluid, m3 tube-side volumetric flow rate, m3/s subscript used to indicate tube surface thermal diffusivity, m2/s

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

` 6 ¡ a hn i O – O o q

coefficient of volumetric thermal expansion, K–1 indicates temperature difference heat exchanger effectiveness dimensionless variable, ¡(lcp)f/(lcp)s. positive roots of the Bessel function kinematic viscosity, m2/s dimensionless temperature dimensionless tank averaged temperature dimensionless time geometric parameter

Yo

Bessel function of the second kind

101

FIGURE 4.1 Conceptual drawing of an indirect integral solar collector storage (ICS) system with a load-side immersed heat exchanger for solar domestic water heating

are anticipated using multi-component injection molding or extrusion, and integration of components and fittings. Additional savings are expected due to weight reduction, which translates to reductions in shipping, and installation costs. The status of material selection for the glazing, collector enclosure and heat exchanger were first described in volume 15 of Advances in Solar Energy (Davidson et al, 2002). The present paper summarizes the results of our continuing study of heat transfer and long-term durability and reliability of polymers for immersed heat exchangers. In section 4.2, a summary of the operation of immersed heat exchangers is presented with an emphasis on study and characterization of natural convection heat transfer. In section 4.3, the durability of candidate polymers in hot chlorinated water is discussed and recent data are presented. Section 4.4 summarizes the state of knowledge regarding the formation of calcium carbonate, usually called scale, on polymeric surfaces.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

102

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

4.2 THERMAL CHARACTERIZATION AND DESIGN 4.2.1 THERMAL PERFORMANCE Immersed heat exchangers may be used to charge or discharge water storage tanks. Heat transfer between the heat exchange surface and the storage fluid is via natural convection. Thus the heat exchanger should be located to take advantage of the greatest difference between the temperature of the storage fluid and the temperature of the fluid flowing through the heat exchanger (Figure 4.2). To discharge the tank, the heat exchanger should be located near the top of the storage tank where the storage temperature is presumably the warmest. The cold plumes formed in the boundary layer at the heat exchange surface will move downward in the tank. Conversely, to charge the storage tank, the heat exchanger should be placed at the bottom of the tank. These arrangements ensure that the storage vessel can be completely charged or discharged.

(a)

(b)

FIGURE 4.2 Conceptual sketch of an immersed heat exchanger in a vertical thermal storage tank; the heat exchanger is shown here in two positions depending on the intended mode of operation: a) top mounted heat exchanger intended to discharge the tank, and b) bottom mounted heat exchanger intended to charge the tank

The thermal performance of the heat exchanger is described by the overall heat transfer coefficient-area product (UA), which is the inverse of the sum of the convective and conductive thermal resistances: (1) In equation (1), Ri is the thermal resistance of the forced convection flow through the heat exchanger, Rw is the conductive resistance across the heat exchanger wall and Ro is the natural convection thermal resistance on the outer storage-side of the heat exchanger. The effectiveness of an immersed heat exchanger is related to the UA by: (2)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

103

The subscript s denotes the storage fluid and the subscript f denotes the heat exchange fluid. The bracketed term on the right hand side of equation (2) is referred to as the number of transfer units (NTU) of the heat exchanger. For metal heat exchangers, the conductive resistance across the tube wall (Rw) may be neglected (unless there is substantial scaling due to use in hard water or if double wall tubes are used). If the tube is plastic, thin-walled tubes with a relatively high ratio of outer diameter-to-wall thickness, termed the standard diameter ratio (SDR), are desirable to minimize Rw (Liu et al, 2000; Davidson et al, 2002). The relative magnitude of the thermal resistances depends on many factors including the flow rate and flow regime (laminar or turbulent) through the heat exchanger, the geometry of the combined heat exchanger and storage tank, and the temperature difference between the outer heat exchange surface and the storage fluid (the state of charge). The natural convection heat transfer may limit the rate of heat transfer, especially if the internal flow is turbulent. In this case, it is very important to accurately characterize Ro: U (3) UU h o Ao

where ho is the natural convection heat transfer coefficient. Determination of the ho of heat exchangers immersed in solar storage tanks has been the subject of numerous experimental and numerical studies (Feiereison et al, 1982; Farrington and Bingham, 1986 and 1987; Reindl, 1992; Reindl et al, 1992a and 1992b; Liu et al, 2003, 2004 and 2005; Su and Davidson, 2005). The problem poses unique challenges compared to the extensively studied problem of natural convection to bodies in an unbounded fluid. First, the process is transient. During both charge and discharge, the temperature difference that drives natural convection to the heat exchanger decreases in magnitude as charge or discharge procedure progresses. Thus in the expected situation where the natural convection poses a significant thermal resistance compared to wall conduction and forced convection heat transfer, the overall heat transfer coefficient U, and, thus, effectiveness ¡ and NTU decrease with time. Second, the spatial distribution of the storage temperature may change during the discharge process and depends on the specific heat exchanger/tank geometry. Thus a generalized choice for the appropriate temperature difference to describe natural convection has been elusive. The starting point for most efforts to quantify the natural convection heat transfer coefficient has been the conventional empirical correlation of Nusselt and Rayleigh numbers: (4) where: (5)

(6)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

104

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

The characteristic length scale in expressions (5) and (6) is the outer diameter D of an individual tube in a tube bundle or coil. The empirical constant C and exponent m should depend only on the geometry of the heat exchanger/tank and the range of Rayleigh number. The real challenge in developing correlations for this application is measurement and identification of an appropriate characteristic temperature difference between the heat exchange surface and the storage fluid, 6T = Tw – Ts. Early studies of smooth and finned coiled and bayonet heat exchangers (Feiereison et al, 1982; Farrington, 1986; Farrington and Bingham, 1986 and 1987) did not measure the heat exchanger surface temperature but rather used various definitions of the log-mean temperature difference to correlate data. The storage fluid was assumed to be fully mixed. The correlations were generally unsuccessful in the sense that the ‘constant’ C was temperature dependent. Natural convection to a single tube (Liu et al, 2003) and several tube bundle heat exchangers (Liu et al, 2003 and 2004) immersed in a thin rectangular storage tank with a height to width aspect ratio of 9.3:1 and inclined at 30 degrees to the horizontal was measured at the University of Minnesota. The tube wall temperature was measured directly. When the storage fluid was initially isothermal, it was possible to correlate the data using the average storage temperature. When the storage fluid was initially stratified, it was necessary to use the local storage fluid temperature surrounding the heat exchanger tubes to define Ts and to obtain statistically significant correlations of the measured data. For a single tube, the data were best correlated by: NuD = (0.675 ± 0.001)RaD0.25, for 106 ) RaD ) 108.

(7)

For a bundle of eight tubes with pitch-to-diameter (P/D) = 2.43: NuD = (0.728 ± 0.002)RaD0.25, 105 ) RaD ) 107.

(8)

For bundles of 240 tubes with 1.5 ) P/D ) 3.3: NuD = (2.45 ± 0.03)RaD0.188, 102 ) RaD ) 104.

(9)

Three important findings emerged from these studies. First, comparison of equations (7)–(9) to the correlations developed for a horizontal tube in an unbounded fluid (Morgan, 1975) proved that the presence of the storage tank enhanced heat transfer. The heat transfer enhancement is attributed to the buoyancy driven flow in the tank. For a single tube, the correlation predicts average heat transfer coefficients ~20 per cent larger than those for a heated tube in an unbounded fluid. Second, fluid motion in the storage fluid was sufficient to maintain a fully mixed tank for an initially isothermal storage tank. If the storage fluid was initially stratified, operation of the heat exchanger destroyed stratification relatively quickly. Third, for tube bundles, the pitch to diameter ratio (P/D) had no statistically significant impact on the overall heat transfer for 1.5 ) P/D ) 3.3. This result implies that it is possible to pack the heat exchanger tubes in close proximity to each other without substantially decreasing overall heat transfer. For the same heat exchange area, multiple tubes reduce the pressure drop across the heat exchanger.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

105

Use of equations (7)–(9) for design of heat exchangers is difficult because it requires knowledge of the temporal behavior of the storage fluid temperature during the discharge process. An alternate approach that relies solely on the initial storage temperature was first suggested by Reindl et al (1992b). From a scale analysis of the problem, a semiempirical expression for the time-dependent Nusselt number based on the initial Rayleigh number of an initially isothermal tank was proposed. Su and Davidson (2005) extended the scale analysis to consider discharge of a storage tank with tube bundle heat exchangers and presented the results in a more generalized dimensionless form. Su and Davidson (2005) divide the process into four temporal periods referred to as conduction, quasi-steady, fluctuating and decay. General formulations for the transient Nusselt number and the volume averaged water temperature in the enclosure were developed for each period. Three dimensional CFD results illustrate the flow and temperature fields within the storage tank (Figure 4.3). At the initiation of the discharge, heat transfer is dominated by conduction. The convective flow is only apparent near the heat exchanger and the buoyancy-induced velocity is very low. The dimensionless temperature in the tank during the conduction period may be determined analytically and is given by: (10) where hn are the positive roots of the Bessel function of the second kind,Y0 (hn) = 0, and r⬘ is the dimensionless radial distance measured from the heat exchanger. The volume averaged storage temperature is defined as: (11) – where Ts is the average storage temperature at a specified time, and Ts,0 is the fully-mixed storage temperature at the start of the charge or discharge process (at t = 0). The average Nusselt number is expressed as: (12) The asterisk on the Nusselt number denotes that it is defined in terms of the initial, and presumed known, difference between the initial surface temperature of the heat exchanger and the initial temperature of the storage fluid, i.e. 6T = Tw – Ts,0. The conduction period — . The dimensionless time, o, is given by: ends relatively quickly at o = 3Pr s (13) where t0 is the ratio of the convective length scale and a velocity scale for natural convection. The natural convection velocity scale U is given by: (14) Thus the time scale is: (15)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

106

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

FIGURE 4.3 Three dimensional streamlines (top) and isotherms (bottom) for a rectangular storage with a single immersed tube; 6T = |Tw – To| = 20K (T0 = 353K and Tw = 333K) corresponding to Ra*D = 2.345 × 107 during (a) conduction (o = 0.029, t = 0.014s); (b) conduction (o = 1.953, t = 0.955s); (c) quasi-steady (o = 11.778, t = 5.76s); (d) fluctuating (o = 111.78, t = 54.66s); (e) decay (o = 2441, t = 1194s); and (f) decay (o = 65,710, t = 32,315s)

Once the conduction period ends, heat transfer becomes dominated increasingly by natural convection. Initially, the presence of the storage container does not impact the flow field or heat transfer to the heat exchanger. Consequently, the convective heat transfer is described by the average Nusselt number for a body immersed in an unbounded fluid, given in the form of equation (4). This quasi-steady period endures until

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

107

the flow in the enclosure begins to disrupt the boundary layer of the heat exchanger. The duration depends on the geometry of the tank, the position of the heat exchanger within the tank and the magnitude of the convective velocity given in equation (14). At the end of the quasi-steady period, the flow field and heat transfer are influenced by the presence of the tank. CFD results show there is a short period of intense mixing of the storage fluid and intermittent disruption and growth of the boundary layer at the heat exchanger surface. The Nusselt number fluctuates around the quasi-steady value. The duration of the fluctuating period is not well defined except by observation of numerical or measured data. The last period and most important one in terms of overall system performance is the decay period during which over 90 per cent of the energy is removed (discharge) or added (charge) to the storage tank. It is during this period that all measured data have been obtained. The driving temperature difference between the heat exchanger surface and the storage fluid decreases exponentially. The transient Nusselt number is expressed as: (16) where the dimensionless transient temperature difference is given by:

(17) The geometric parameter q is specified by the heat exchanger/storage configuration. It equals the volume of the storage fluid divided by the product of the heat transfer surface area of the heat exchanger and the characteristic length scale for natural convection heat transfer. For N tubes, each of diameter D and outer heat transfer surface area Ao: (18) A generalized expression for the transient Nusselt number is:

(19) The strength of equation (19) is that it requires only the initial storage temperature and the heat exchanger geometry and operating conditions, from which Ra*D and q can be determined, to determine heat transfer during the entire decay period. The values of C and m must be determined empirically and depend only on the geometry. Su and Davidson (2005) used the data of Liu et al (2003, 2004 and 2005) for single- and multiple-tube heat exchangers to test the usefulness of equation (19). The data for the single-tube heat exchanger in an initially isothermal storage were successfully correlated with m = 0.25 and C = 0.50. For the 240-tube heat exchanger with P/D = 3.3, the figures were m = 0.188 and C = 2.6.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

108

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

In situations where storage fluid is stratified at the initiation of the charge or discharge and stratification is maintained by the use of manifolds or baffles, equation (19) is not applicable. If the fluid is stratified, the average tank temperature does not represent the thermal environment of the heat exchanger. In this case, use of the initial Rayleigh number in equation (19) will underestimate the natural convection film coefficient at the beginning of the charge or discharge process and overestimate it as the storage tank reaches a full or depleted state of charge, respectively. In this case, correlations of the form of equation (4) may be used but it will be necessary to characterize the temperature of the storage fluid surrounding the heat exchanger during use. Once the UA of the heat exchanger has been estimated, the transient behavior of the heat exchanger/storage tank can be predicted. The transient storage temperature Ts(t) and delivered outlet water temperature Tf,out(t) can be estimated from energy balances on the heat exchanger and storage. Assuming the inlet fluid temperature and mass flow rate · ) are fixed and the storage fluid remains fully mixed through the heat exchanger (m f throughout the charge or discharge process, the transient storage temperature and heat exchanger outlet temperature are given by: (20)

(21) The parameter a = ¡(lcp)f/(lcp) is approximately equal to the heat exchanger effectiveness ¡ when the heat exchange and storage fluids are the same fluid. The volume ratio Vr equals the volume of fluid passed through the heat exchanger divided by the ˙ ft/lf /Vs. storage volume: Vr=m

4.2.2 DESIGNS FOR IMPROVED PERFORMANCE Efforts to improve the thermal performance of thermal stores with immersed heat exchangers have focused on three concepts: shrouds or baffles to direct the buoyant flow that develops at the heat exchanger, stratification manifolds, and partitioned storage vessels. Several European combistores include a shroud around the immersed heat exchanger connected to a stratification pipe to prevent the natural convective plumes that develop in the boundary layer of the coil from mixing with the surrounding fluid. Work to characterize these systems is reported by the International Energy Agency (IEA) Task 26 (Drück, 2002; Weiss, 2003). This review focuses on the use of a divided or partitioned storage vessel to improve performance. The divided storage concept is an adaptation of the use of multiple storage tanks connected in series first suggested by Bejan (1982). Analyses of the concept are presented by Sekulic and Krane (1992a and 1992b) for two tanks in series each with an immersed heat exchanger and by Boies and Homan (2004 and 2005) and Boies et al (2005) for multiple storage tanks. Mather et al (2002) proposed using multiple tanks for large solar thermal storage systems with the view that a series of small storage tanks should be more economical and more practical than the single large (greater than 2000 liter) storage tank used in a combistore. For smaller SDHW systems, dividing a single tank

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

109

into two or more storage compartments is practical and has the advantage of reduced thermal losses to the ambient. At the University of Minnesota, we have conducted experiments with a single storage tank divided into two compartments (Davidson et al, 2005). The dual compartment storage tank, illustrated in Figure 4.4 for discharge of the store, provides a staged heating process. The inlet of the heat exchanger is in the first compartment. The heat exchange fluid entering the second compartment is warmer than that entering the first compartment and thus the storage fluid in compartment two remains warmer for longer periods of time. Consequently, for much of the discharge process, the heat exchanger outlet temperature is higher than it would be using the same heat exchanger in an undivided storage. The advantage of the approach is that the enhanced heat transfer attributed to convective motion in the storage (Liu et al, 2003, 2004 and 2005) is maintained in each compartment. Boies and Homan (2004) point out that the benefit of this concept can be realized in either of two ways. For a fixed total storage volume, a higher energy delivery rate is maintained for a longer period, which means more hot water is produced. Alternately, the same energy output can be provided by a smaller storage volume.

FIGURE 4.4 Sketch of a divided indirect thermal storage vessel with two storage compartments; the heat exchanger is positioned for discharge

The advantage of the divided storage depends on the effectiveness, or equivalently the NTU, of the immersed heat exchanger, and the size of the thermal store relative to the hot water draw (Vr). Too low an NTU will limit the energy transfer rate and diminish the advantage of the divided storage. On the other hand, increasing the NTU is advantageous only up to a certain value. At high NTU, heat transfer is limited by the effective temperature difference between the heat exchange and storage fluids. The effects of NTU and Vr on cumulative delivered energy of a storage tank with two compartments are illustrated in Figure 4.5 for an idealized situation in which the heat exchanger NTU is

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

110

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

1.18 1.16

NTU = 12

Energy Ratio (Er)

1.14 NTU = 6

1.12 1.10

NTU = 4

1.08 Equal outlet temperatures

1.06 1.04

NTU = 2

1.02 NTU = 1

1.00 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Volume Ratio (Vr)

FIGURE 4.5 Predicted ratio of delivered energy of a divided storage and an undivided storage as a function of the dimensionless output volume and NTU. The divided storage has two compartments. The dashed line indicates the volume ratio at which the heat exchanger outlet temperatures of the two storage configurations are equal

constant during discharge and thermal losses are neglected. Here the ratio of energy delivered by a divided storage to that delivered by an undivided storage, Er, is plotted versus Vr for 1 ) NTU ) 12. The NTU values shown are those of the undivided storage. For the divided storage, the NTU is assumed to be divided equally between the two storage compartments. The energy ratio increases as overall NTU is increased from 1 to 10, with little change from NTU = 10 to 12. For a given NTU, the energy ratio depends on the volume ratio, or extent of depletion of the storage. The maximum cumulative delivered energy is at Vr ⬇ 0.85. For example, at NTU = 10, Er is 1.15. The dashed line on the plot indicates the value of Vr for which the outlet temperatures of the undivided and divided storage are equal. The region to the left of the dashed line, where the divided storage delivers higher temperature water than the undivided storage, represents the desirable operating range. The only case for which there is no advantage to the divided storage system is if the storage is too small to meet the hot water load and the stored energy is completely discharged during a hot water draw. When the storage unit is completely discharged, Er = 1. The measured performance of a divided storage vessel is shown in Figure 4.6. These data were obtained for a 126 liter rectangular storage vessel with a copper heat exchanger with a total surface area of 2.38m2 immersed in the storage fluid. Experiments were conducted with one and two compartments. In the divided storage vessel, the compartments were separated with a 2.54cm thick polystyrene sheet (R-value = 0.881K-m2/W). Figure 4.6 is a plot of Er versus Vr for nominal NTUs of 2.5 and 7. With NTU = 7, the divided storage delivers 11 per cent more energy than the undivided storage at Vr = 0.8 (i.e. when 100 liters of hot water or 55 per cent of the stored energy has been delivered). Higher outlet temperatures are sustained by the divided storage until Vr = 1.2. At this point, the divided

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

111

1.14

Energy Ratio (Er)

1.12 NTU = 7

1.10 1.08 1.06 1.04

NTU = 2.5

1.02 1.00 0.0

0.5

1.0

1.5

2.0

2.5

3.0

Volume Ratio (Vr) FIGURE 4.6 Measured ratio of delivered energy of a divided and undivided storage as a function of ratio of the volume of delivered hot water to the volume of storage fluid for nominal heat exchanger NTU of 2.5 and 7

storage provides 8 per cent more energy than the undivided storage and the storage is 74 per cent depleted. With NTU = 2.5, the divided storage delivers 5 per cent more energy at Vr = 0.8. At Vr = 1.3, when the outlet temperatures of the divided and undivided storage are equal, the divided storage provides 4.6 per cent more energy and the stored energy is 65 per cent depleted. The measured advantage of the divided storage is slightly lower than that predicted because the NTU of the heat exchanger decreases as the tank is discharged. However, even with a simple partition, the divided storage vessel provides the same benefit as that predicted for multiple storage tanks. The concept is expected to be easy to implement in solar systems at minimal or no extra cost.

4.2.3 SUMMARY Substantial progress has been made in recent years to understand and characterize the thermal/fluid processes in storage vessels with immersed heat exchangers. Natural convection heat transfer correlations have been developed for tube bundles and the generalized approach to express measured data in the form of equation (19) holds promise for fully mixed storage vessels. The empirical constant in this expression will depend on the combined heat exchanger/tank design and must be determined from experiment or CFD. The computational problem requires substantial computational resources. It is anticipated that most developers and designers will choose the experimental approach. Once the constant is determined, the transient charge or discharge process can be predicted using only the initial conditions. If baffles or other stratification devices are shown to be effective, equation (4) may be used. In this case, the key to predicting the natural convection heat transfer is to determine the thermal environment to which the heat exchanger is exposed.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

112

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS .

4.3 MECHANICAL CHARACTERIZATION OF POLYMERS Traditionally, immersed heat exchangers are copper. The switch from copper to polymer materials requires careful consideration of the compatibility of the material with the working fluids. Polymer materials must maintain mechanical properties in the working fluid over their target lifetimes. Although thin walled tubes with a high SDR are desirable for maximizing heat transfer (see section 4.2), tube wall thickness must be sufficient to ensure structural integrity throughout the lifetime of the heat exchanger. The traditional approach to evaluating long-term performance of polymers has been to consider burst strength. However, the creep compliance is equally important, because material deformation may exceed acceptable limits long before the polymer component ruptures (Wu et al, 2004). In the past, we have focused on a number of potential materials. Of these, we selected polysulfone (PSU), polybutylene (PB) and polyamide 6,6 (PA66) for additional study. PSU was selected because it is currently used in aqueous environments for pipe fittings and water heater dip tubes. PB was selected because it is used outside the US for hot water distribution and in the US as an internal liner tank in a commercially available plastic electric water heater. In addition, PB represents polyolefins, which use additives to protect them from degradation in water. Polyamide 6,6 was selected because it is a high strength material. In addition it is known to absorb water and this property is of interest in terms of understanding the mechanisms for degradation of mechanical properties and scaling (discussed in section 4). Cross-linked polyethylene (PEX) was not selected for further study because PEX tubes are designed for a 40-year lifetime of exposure to hot chlorinated water (following ASTM F2023, 2000). In potable water, chlorine and pH combine to create an oxidative environment, characterized by the oxidative reduction potential (ORP), that can chemically attack a polymer, resulting in permanent loss of mechanical strength and increase in creep compliance. Water absorption and hydrolysis can also diminish polymer mechanical properties. The mechanism of polymer degradation will depend on the polymer molecular structure. Thus both polymer morphology data and mechanical properties are required to understand the relationship between degradation and loss of mechanical performance.

4.3.1 BACKGROUND ON POLYMER DEGRADATION AND PRIOR STUDIES The molecular structures of PSU, PB and PA66 differ and predispose each polymer to different degradation mechanisms in hot potable water (Scott, 1999). Of the three polymers, PSU is the most resistant to molecular degradation because the polymer chain is comprised of aromatic rings and strong carbon, sulfur and oxygen bonds within the polymer backbone. Polybutylene, a polyolefin comprised of carbon and hydrogen atoms, will degrade by oxidation: a hydrogen atom is easily abstracted from a tertiary carbon within the polymer chain. The lifetime of PB in an oxidative environment is extended by including an antioxidant. Polybutylene performance will not degrade significantly until the antioxidant has been depleted. Polyamide 6,6 is a semi-crystalline polyamide that has good mechanical properties in the air but is prone to water absorption and hydrolysis. The absorbed water acts as a plasticizing agent in PA66 and increases polymer chain mobility resulting in a reduction in tensile strength (Aharoni, 1997). The properties may be restored if the water is removed. However, if PA66 remains in water for longer periods, hydrolysis

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

113

is likely. The PA66 polymer chain includes an amide group which can by hydrolyzed in water, leading to chain scission and permanent damage. The effect of hot water on the molecular structure of polyolefins was analyzed using infrared spectroscopy, size exclusion chromatography, thermal analysis and visual observations (Gedde et al, 1994). Cross-linked polyethylene (PEX), polybutene-1, polypropylene and medium-density polyethylene were studied. These researchers found that there is an incubation period during which there is a decrease in the antioxidant concentration in a polyolefin. The incubation period ends when the antioxidants are completely depleted, leaving the material with an extremely limited lifetime. Oxidation was found to be the main mechanism responsible for polymer degradation. The study, although limited to polyolefins, established a method for predicting the lifetime of a polymer based on antioxidant depletion. Prior work addressing the effect of a hot potable water environment on mechanical properties of polymer tubing has focused on polymer strength. The National Sanitary Foundation (NSF) and the American Society for Testing and Materials (ASTM) have both developed testing protocols, NSF standard P171 (1999) and ASTM standard F2023 (2000), to evaluate the burst strength of polymer tubing in hot chlorinated water. In these tests, the tube rupture must be a result of the oxidative environment as indicated by a failure in which there are many cracks emanating from the interior tube wall and extending to the tube surface (stage III failure). The water chlorine level ranges between 3 to 5ppm and pH level ranges from 6 to 8 such that an ORP of 825mV is maintained throughout the test. (Note that the ORP of typical tap water ranges from 400 to 550mV (ASTM, 2000).) Tests are performed over several combinations of pressure and temperature so as to extrapolate the burst strength of the tubing after 10 to 40 years’ exposure to hot chlorinated water. Test results are limited to the particular tube geometries (diameter to thickness ratio) tested. These standards provide a methodology to predict polymer lifetime based on a material strength limit, the hydrostatic burst strength. Material deformation (in other words creep compliance) is not evaluated during testing and is not considered in predicting material lifetime. Bradley et al (2000) used a similar test to determine the effects of temperature and chlorine levels on the tensile strength and weight of polymer tubes. Polymers were tested in conditions of 28ϒC, 60ϒC and 90ϒC with 5ppm chlorine and also at 90ϒC with 0ppm chlorine. The strength of the materials when immersed in a hot chlorinated environment for 1500 hours was evaluated under (1) no load and (2) continuous loading conditions. The harshest condition, 90ϒC and 5ppm chlorine, was very detrimental to PA66, causing it to lose 70 per cent of its weight and 95 per cent of its tensile strength. Reducing the temperature to 60ϒC from 90ϒC caused the degradation of PA66 to slow dramatically. PSU, polyphenylsulfone, polyphenylsulfide and polyvinylidene fluoride exhibited good resistance to chlorine exposure at high temperature with little weight or tensile strength loss. The creep compliance was not evaluated. Limited creep data in oxidative environments for PB and PA66 have been published by researchers at the University of Minnesota (Walter et al, 2003; Wu et al, 2004). Because the test data were obtained for only one environmental condition (825mV ORP and 82ϒC), the effects of changes in ORP and temperature could not be evaluated. No tests of chemical or physical degradation (such as molecular weight or SEM) were performed.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

114

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

4.3.2 EVALUATION OF POLYMER MATERIALS: PSU, PB, AND PA66 Recent work at the University of Minnesota has focused on creep of PSU, PB and PA66 in an oxidative aqueous environment (Freeman, 2005; Freeman et al, 2005). Other indicators of polymer degradation, tensile strength, molecular weight and surface morphology (through SEM images) were also evaluated. A review of this work is provided in this section. Creep and tensile strength were measured in a series of experiments in which the polymers were submerged in water at controlled temperature and ORP levels. Specimens were either injection molded (PSU and PB) or cut from extruded sheets (PA66). The test environments consisted of combinations of two temperature settings, 60ϒC and 80ϒC, and two ORP settings, 550mV and 825mV, for a total of four test environments. Data were also obtained in air at room temperature. Creep compliance was measured with specimens that were continuously loaded at stress levels of approximately 11MPa (PSU), 10.5MPa (PB), and 7MPa (PA66). The stress levels were chosen to give measurable creep over the test duration and to ensure creep data were in the linear range for each material. Creep was measured for specimens exposed for 300 hours. Measurements of tensile strength and molecular weight were obtained for specimens that had been exposed to the full range of test conditions for 300 hours and the most aggressive condition (80ϒC and 825mV) for up to 1100 hours. Measurement of molecular weight of the PSU, PB and PA66 required different techniques. Gel permeation chromatography (GPC) was used to analyze the molecular weight of PSU. As recommended by the material supplier, the molecular degradation of PB was determined using melt flow index tests according to ASTM D1238 (ASTM, 2001). A higher melt flow index indicates a lower molecular weight. Changes in molecular weight of PA66 were based on the measurement of relative viscosity (adaptation of ASTM D789 (ASTM, 2004)). A lower viscosity indicates material degradation. Scanning electron microscopy was used to observe the morphology of the unexposed and exposed polymer surfaces after 300 hours’ exposure. For the relatively short duration of these experiments, both PSU and PB were stable in hot chlorinated water. After 300 hours the measured creep compliance of PSU in air at room temperature ranged from 0.2 to 0.7GPa–1, comparable to the manufacturer’s reported compliance of approximately 0.4GPa–1. The creep compliance in all water environments ranged from 0.3 to 1.2GPa–1. The data for air and water are indistinguishable based on the experimental uncertainty (± 0.4GPa–1). The average tensile strength of all specimens tested (air and hot water environments) is 77 ± 2MPa, comparable to the manufacturer’s reported value of 70MPa (Solvay Advanced Polymers, 2002). Strain at failure is relatively low (5–7 per cent) for exposed and unexposed specimens. The GPC analysis and SEM images support the stable nature of this polymer. The Mw of unexposed specimens ranged from 57,000 g/mole to 67,000 g/mole. The Mw of the exposed specimens ranged from 44,000 g/mole to 63,000 g/mole. SEM images taken of exposed PSU specimens are featureless and provide no evidence of surface degradation. The creep compliance of PB in air ranged from 0.3 to 0.6GPa–1, while the creep compliance in water ranged from 0.5 to 1.7GPa–1 (after 300 hours). Even though specimens exposed to water show larger creep compliance than in air, the difference is

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

115

within the experimental uncertainty (± 0.3GPa–1). The strength of the unexposed specimens and those exposed to the potable water environments is 50 ± 20MPa. The strength data are comparable to the manufacturer’s reported value of 36.5MPa (Basell Polyolefins, 2004). Additional tests indicated that the large variation in the tensile strength can be reduced significantly if the injection flow direction is consistent among specimens (Freeman, 2004). Strain at failure data were also comparable between unexposed and exposed specimens, ranging from 2 to 6 per cent for unexposed specimens and from 1.9 to 6.9 per cent for specimens exposed to 830mV, 80ϒC. The melt flow index of the unexposed specimens and those exposed to chlorinated water for 250 hours were within the instrument uncertainty: the average melt flow index of the unexposed material is 0.60g/10 min and that of the exposed specimens is 0.63 ± 0.05g/10 min. SEM images of the unexposed and exposed PB specimens are similar and do not indicate degradation of the material. The effect of exposure to hot, chlorinated water on PA66 is dramatic. The loss in mechanical properties of PA66 is evident in both the creep and tensile strength/strain at failure data. Creep compliance is plotted in Figure 4.7. The creep compliance of specimens exposed to water is greater than that in air for all conditions. The creep compliance in air ranged from 0.2 to 1.0GPa-1 after 160 hours. The creep compliance in water ranged from 6.5 to 14.5GPa–1 (after 230–340 hours). The average tensile strength of the unexposed material is 94 ± 1MPa, comparable to the manufacturer reported strength of 85MPa (DuPont, 2004). After 24 hours of exposure to any of the hot potable water conditions, the strength is reduced more than half, to approximately 45MPa, regardless of temperature or ORP (Figure 4.8). This decrease in tensile strength is attributed primarily to the absorption of water. Approximately 15 per cent of the strength is not recovered by drying, indicating some molecular degradation had occurred. Strain at failure for unexposed specimens ranges from 26 to 30 per cent. For specimens exposed to 830mV,

Creep compliance (1/GPa)

100

Air, 24˚C nominal 570 ± 50 mV, 27 ± 1˚C 550 ± 20 mV, 60 ± 1˚C 830 ± 20 mV, 60 ± 1˚C 540 ± 50 mV, 80 ± 2˚C 830 ± 20 mV, 80 ± 1˚C

10

1

0.1 0.1

1

10

100

1000

Time (hours) FIGURE 4.7 Creep compliance of PA66 in air and exposed to hot chlorinated water

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

116

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

Tensile strength (MPa) G

100

No exposure 550 ± 20 mV, 60 ± 1˚C 830 ± 20 mV, 60 ± 1˚C 830 ± 20 mV, 80 ± 8˚C

80

uncertainty, stress

60

40

20

0

uncertainty, strain

Strain at failure (%)

250 200 150 100 50 0 0

50

100

150

200

250

300

Time (hours) FIGURE 4.8 Tensile strength (top) and strain (bottom) of PA66 before and after exposure to hot chlorinated water

80ϒC conditions, strain at failure varies depending on the exposure time (Figure 4.8). Considerable elongation occurs for samples aged from 2 to 5 hours (strain at failure during this period was as high as 200 per cent). After 12 hours’ exposure the strain at failure ranges from 35 to 45 per cent. This variation in elongation over time may be associated with initial water absorption followed by hydrolysis. Although strain at failure is similar for unexposed and exposed specimens (exposure time greater than 10 hours), the exposed material has consistently lower strength and reduced stiffness.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

117

There is no difference in the relative viscosity of unexposed and exposed PA66 specimens: the average relative viscosities of the unexposed PA66 specimens and those specimens exposed to 80ϒC water with ORP levels of 540 and 800mV for 240 hours are 50.4 and 50.3 ± 0.6, respectively. Degradation of the surface may not be detected because of the large volume of bulk material required for relative viscosity tests. SEM images of the PA66, however, show significant surface degradation. The surface of the unexposed PA66 is smooth with no visible defects (Figure 4.9a). Surface cracks, ranging from 1+m to 10+m in length, are visible on the surface of all specimens exposed to water (Figure 4.9b). Although the physical breakdown of polymer chains is not indicated by the relative viscosity, these images, along with the creep and tensile test data, provide convincing evidence of material degradation.

(a)

(b)

FIGURE 4.9 SEM images of PA66: (a) native unexposed surface and (b) after 292 hours’ exposure at 830 ± 40mV, 59 ± 2ϒC. The surface of the unexposed PA66 is smooth with no visible defects. Surface cracks, ranging from 1+m to 10+m in length, are visible on the surface of all specimens exposed to water

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

118

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

4.3.3 SUMMARY These results highlight the importance of designing heat exchanger components based on data from specimens that have been exposed to an aqueous, oxidative environment. The challenge is to elicit degradation within the test duration. It is difficult to accelerate the test sufficiently using a traditional ORP control approach. The data presented here show that for up to 1100 hours exposure in hot potable chlorinated water, PSU and PB (with antioxidant additives) maintain their mechanical properties, while PA66 degrades significantly. These test results are consistent with the expected degradation mechanisms for each polymer. PSU has a rigid backbone with strong chemical bonds and was expected to be resistant to oxidation. PSU tubing is commercially available in limited sizes. Manufacturing process development is required to cost-effectively produce small diameter, thin-walled PSU tubes required for heat exchanger applications. The PB included an antioxidant package, such that degradation of the polymer is delayed until the antioxidant is depleted. The combination of test environments and durations studied were not sufficient to deplete the antioxidant. Further investigation of the relationship between depletion of the antioxidant additives in polyolefins is recommended to determine the expected life of these materials. Polyamide 6,6 is not recommended for this application as test results indicate substantial degradation of the material in relatively short times. The polymer readily absorbs water and the polymer chain includes an amide group which can be hydrolyzed.

4.4 SCALING OF CANDIDATE POLYMERS 4.4.1 BACKGROUND Scaling is the formation of a hard mineral deposit on a surface that is in contact with untreated water containing dissolved minerals (in other words hard water). Scale deposits are composed mostly of calcium carbonate, and this type of scale is the focus here. Scaling is pervasive on heated surfaces because the solubility of calcium carbonate decreases with increasing temperature and the rates of the deposition processes are enhanced at elevated temperature. In solar collectors (Vliet and Baker, 1998; Baker, 2000; Baker and Vliet, 2001; Baker and Vliet, 2003) and heat exchangers (Epstein, 1986; Marner and Suitor, 1987) scaling can be a serious problem. Scale may reduce heat transfer rates due to the additional conductive resistance across the heat exchanger wall. In addition, the scale build-up increases pressure drop across the heat exchanger. In extreme cases, the deposit may completely block the flow of water. Until recently, literature reports have focused on scaling of metal tubes (see, for example, Branch and Muller-Steinhagen, 1991; Khan et al, 1996; Budair et al, 1998). With the increased use of polymer tubes, more attention has been paid to scaling of polymers. The focus of this section is scaling of candidate polymer tubes and more broadly the formation of calcium carbonate on polymer surfaces. Scaling of a heated surface involves several steps, as described more completely elsewhere (Cowan and Weintritt, 1976; Hasson, 1981; Knudsen, 1981; Ferguson, 1984; Bott, 1988; Karabelas, 2002). First, the surface (in this case a polymer) comes into contact

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

119

with water that is supersaturated with respect to calcium carbonate. The supersaturation (S) is commonly given by:

(22) where aCa2+ and aCO32– are the activities of calcium and carbonate ions, respectively and Ksp is the solubility product for crystalline CaCO3 (e.g. calcite) (Snoeyink and Jenkins, 1980; Stumm and Morgan, 1996). When S is greater than 1, the solution is supersaturated and the formation of solid calcium carbonate is possible. A higher value of S indicates a greater thermodynamic driving force for scaling. To calculate S, several chemical reactions that control species concentrations must be taken into account along with the temperature, which influences the reaction rates as well as the solubility product. Next, solid calcium carbonate forms on the polymer surface by nucleation and growth (Nancollas, 1979; Mullin, 1993). During the initial nucleation stage, the chemical and physical properties of the polymer surface, as well as the supersaturation and the temperature, are expected to play a role. During growth, ions from the water are transported to the surface and the calcium carbonate scale layer grows. Growth can be controlled by transport to the surface or reaction at the surface. When transport controls, the flow rate (e.g., Reynolds number) enters in as another important variable in scaling, as flow impacts the mass transfer to the surface. Lastly, the scale can be removed, depending on the flow conditions and the adhesion of the scale to the tube. Hence at least four variables impact the scaling process: the composition of the water, the temperature, the surface composition and structure, and the flow condition.

4.4.2 SCALING OF POLYMERS Various configurations have been used to study scaling of polymers. Experiments have been performed with 1) flow through heated tubes, 2) isothermal flow through tubes, and 3) in stagnant water under isothermal conditions. Some researchers have focused solely on the behavior of polymers while others have included polymers as one of several materials under study. Very few studies have tackled scaling of polymer tubes using conditions that simulate the application of a polymer in a heat exchanger or solar absorber. An early exploration of heat exchangers with Teflon tubing explored scaling using a boiler test and data on the thermal performance of the heat exchanger (Githens et al, 1965). Results of the boiler test suggested that the smooth surface of the Teflon reduced adherence of the scale under the agitation from boiling. The scale also had less of an impact on the heat transfer performance of the Teflon-based heat exchanger compared with that of the one containing metal tubes; this result is due in part to the similarity in the thermal conductivity between the polymer and the scale. More recently, Wang and co-workers at the University of Minnesota (Wang et al, 2005) studied scaling of a variety of candidate polymer tubes in a tube-in-shell heat exchanger. In this research commercial tubes of candidate polymers were studied using conditions

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

120

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

representative of solar water heating applications. Tubes were oriented vertically in a tube-in-shell geometry with 60ϒC shell side temperature. Tap water with its composition adjusted by the addition of calcium nitrate, sodium bicarbonate and potassium hydroxide was used as the inlet water. The inlet water was heated as it traveled down the length of the 0.89m tubes. While the diameters of the tubes varied (see Table 4.1), the average flow rate was kept constant and all flow was in the laminar regime. Since the experiment was carried out for a fixed time, different quantities of water passed through each tube. There were two phases of the study: (i) a pretreatment phase with a lower supersaturation inlet water (S = 2, [Ca2+]t = 2 x 10–3 M, pH = 8) and (ii) an accelerated phase with a higher supersaturation inlet water (S = 8, [Ca2+]t = 4 x 10–3 M, pH = 9). Scale deposition was monitored periodically by removing sections of the tubes and characterizing the deposit by chemical analysis, scanning electron microscopy (SEM) and X-ray diffraction (XRD). All materials in the study formed a calcium-rich scale on their surfaces. Figure 4.10 shows representative SEM micrographs of the scale layer deposited on several of the materials investigated in the study. A layer of new material of a distinctly different morphology from the native polymer is apparent. Small particles (~100nm) appear in the layer. Chemical analysis data showed that the scale contained calcium with minor quantities of iron and phosphorus. However, the only crystalline phases found by XRD were CaCO3 polymorphs (calcite and aragonite), indicating that there may be some amorphous iron- and phosphorus-containing phases present. In addition, the scale formed on Teflon and a high temperature nylon (HTN) was amorphous or nano-crystalline. Table 4.1 provides a summary of the final chemical analysis results of the study. Scale accumulated on all polymer tubes in quantities and at rates that were comparable to those on copper, with two exceptions – more scale was found on polyamide 6,6 and less on HTN. The greater amount on PA66 was linked to the moisture absorption by the polymer, which is an order of magnitude more than the other polymers in the study, and to hydrolysis, which creates anionic groups capable of binding Ca cations

TABLE 4.1 Scale accumulation and scaling rate for tubes in a tube-in-shell heat exchanger1 as determined by chemical analysis MATERIAL

Polyamide 6,6, PA66 High Temperature Nylon, HTN Polybutylene, PB Polypropylene, PP Teflon Copper

EXTERNAL DIAMETER (mm) 3.25 5.89

WALL THICKNESS (mm) 0.41 2.19

2.54 2.18 1.52 4.57

0.76 1.07 1.48 1.78

VOLUME OF WATER2 (L) 1158 2535 478 303 327 143

SCALE ACCUMULATED CaCO3 (g /m2) 1.275 0.082

6.6 x 10–7 8.3 x 10–9

0.172 0.202 0.173 0.395

8.9 x 10–8 1.0 x 10–7 8.9 x 10–8 2.0 x 10–7

1

Scale accumulation and rate calculated assuming that all calcium was present as CaCO3.

2

Volume of water passed through each tube in 540 hours during accelerated scaling phase.

Source: Wang et al (2005)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

SCALING RATE (g /m2s)

Thermal and Material Characterization of Immersed Heat Exchangers

121

Source: Wang et al (2005)

FIGURE 4.10 SEM micrographs of scale formed on (a) polyamide 6,6, (b) polypropylene, (c) polybutylene and (d) copper tubes in a tube-in-shell heat exchanger. SEM micrographs were taken at the end of 540 hours of accelerated scale testing. A layer of new material of a distinctly different morphology from the native polymer is apparent

(Aizenberg et al, 1999). The lesser scaling rate for the HTN was attributed to the tube’s larger wall thickness, which prevented the interior surface temperature from reaching as high a temperature as the other materials. For this water chemistry, the temperature had little effect on the supersaturation, but the lower temperature in the case of HTN resulted in slower kinetics of deposition. These conclusions were consistent with a surfacereaction limited model for scaling that took into account differences in temperatures due to the differing wall thicknesses and thermal conductivities of the materials. From these experimental and modeling results, several of the candidate polymers for solar thermal applications (polypropylene (PP), PB, HTN and Teflon) had very similar scaling tendencies. Scaling studies have also been carried out using flow-through systems under isothermal conditions. Andritsos and co-workers (Andritsos et al, 1996 and 1997) studied scale formation at room temperature under turbulent flow conditions. Their experimental system used a syringe pump to boost the pH, and hence the supersaturation, in-line. This design allowed them to achieve high supersaturation without the risk of precipitation of calcium carbonate particles in the bulk liquid. Their results showed a marked increase in scale deposition rate with temperature and flow velocity, consistent with work by Hasson

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

122

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

et al (1968). Interestingly, the scaling rate was found to undergo a dramatic increase at a critical value of supersaturation (S = 8). The authors speculate that below the critical supersaturation the formation of scale is by a surface-controlled mechanism and above the critical value a diffusion-controlled mechanism dominates. Likewise, the scale accumulation with time depended on the supersaturation; at low S, the scale accumulated slowly in the early stage, with an apparent induction period, but at high S, the scale quantity increased linearly with time. The researchers compared the morphologies of the scale deposits on steel, copper and Teflon and found that the tube material had little effect. The tenacity of the scale was less on Teflon, however. Using a similar isothermal flow-through system, Sanft et al (2006) carried out accelerated scaling experiments at room temperature on tubes of cross-linked polyethylene (PEX), a polypropylene copolymer (PP-r) and copper. Two tests were run under identical conditions using distilled water with its composition adjusted (S = 7.8, [Ca2+]t = 3 x 10–3 M, pH = 9). The accumulation of scale was slow at early times and then increased, similar to the results of Andritsos et al (1996). There was large variability in the data, however, which was attributed to the randomness of the nucleation process. Within this limitation, scaling tendencies of the materials in the study were similar. SEM micrographs revealed isolated calcite and vaterite particles, as shown in Figure 4.11. Coverage is incomplete, indicative of the early nucleation-dominated stage of scaling. Another approach is to expose polymers to supersaturated water without flow. In one such study, Roques and Girou (1974) noted a connection between the type of material used for the walls of their reactor, which contained supersaturated water, and the formation of calcium carbonate on the walls. The induction time for this heterogeneous nucleation on the walls was decreased in the following order: poly(vinyl chloride), poly(methyl methacylate), glass and stainless steel. In another study, nucleation times were compared

Source: Sanft et al (2006)

FIGURE 4.11 SEM micrographs of scale formed on PEX tubes after (a) 5 hours and (b) 7.5 hours of exposure to flowing hard water at room temperature. After exposure, isolated calcite and vaterite particles are visible. Coverage is incomplete, indicative of the early nucleation-dominated stage of scaling

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

123

for various substrate materials that were immersed in hard water kept at constant temperature (Ben Amor et al, 2004). The nucleation time was determined by monitoring the pH and the calcium ion content of the water. The researchers observed that under low hardness conditions the nucleation times were much different, decreasing from polyamide to polyvinyl chloride to chrome to steel, consistent with Roques and Girou. However, when harder water was used, the nucleation times were all shorter and roughly the same. This result concurs with evidence from flow-through experiments that the nature of the material has the greatest impact when scaling conditions are not severe and nucleation dominates. A series of reports document the formation of calcium carbonate on the surfaces of fine polymer particles circulating in supersaturated water under isothermal conditions (Dalas, et al, 1988, 1999 and 2000; Kanakis and Dalas, 2000; Dousi et al, 2003). In these studies, polymer particles are placed in a constant concentration reactor in which a drop in pH associated with calcium carbonate formation triggers the addition of reactant solutions, allowing for quantitative analysis of deposition. The results of studies of calcium carbonate formation on sulfonated polystyrene and polystyrene divinylbenzene (Dalas et al, 1988), carboxylated poly(vinyl chloride) copolymer (Dalas et al, 1999) and cellulose (Dalas et al, 2000) polymers parallel earlier work on calcium carbonate formation on calcite seeds (Nancollas and Reddy, 1971; Reddy and Nancollas, 1971; Kazmierczak and Tomson, 1982), supporting the finding that the induction time decreases and the calcium carbonate formation (nucleation and growth) rate increases as the supersaturation of the water is increased. These studies generally point to a surfacecontrolled or surface diffusion-controlled mechanism and reveal some of the effects of polymer chemistry. An important chemical factor is the ability of the polymer to bind calcium ions. For example, calcium ions can bind to the carboxylate (C=O) functional group (Dousi et al, 2003). With this binding ability, the local supersaturation at the surface is high and so the induction time drops. This research on polymer particles gives good guidance on the selection of candidate polymers for solar water heating systems: polymers lacking active functional groups, such as PEX, PP-r, and PP, have the greatest promise in terms of minimizing scaling. In recent work at the University of Minnesota, Wang (2004) and McGill (2005) studied the formation of calcium carbonate on polymers in isothermal cells containing hard water. They immersed thin sheets of the candidate polymers, including those studied by Wang et al (2005), as well as polysulfone and polyphenylsulfone, in hard water. Calcium carbonate deposition was quantified by chemical analysis of the deposit. Conditions were chosen to study nucleation, and hence the randomness of the process led to some variability in the data. These researchers found the scaling tendencies of the candidate polymers to be the same within the experimental error. Experiments were conducted at room temperature and approximately 50ϒC using water that was supersaturated to approximately the same degree at the two temperatures. A significantly greater amount of scale formed during the elevated temperature test, showing the importance of kinetics in enhancing the rate of deposition.

4.4.3 SUMMARY While progress has been made in recent years, more research is needed to understand and control the scaling of polymers. Results to date indicate all surfaces are prone to

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

124

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

scale, but the scaling rate may be different if the conditions ensure that nucleation or surface reaction control the deposition. Scale formation on polymers is linked in part to the chemical structure of the material. For example, PA66 appears to form scale more readily due to its susceptibility to water absorption and hydrolysis (Wang et al, 2005). Yet unknown is the importance of the physical structure, including surface roughness, which has been shown to influence the scaling of metals (Keysar et al, 1994), and stress concentration, which is documented as trouble causing in calcification of biomedical polymers in the body (Banas and Baier, 2000). In general, scaling is enhanced for water with higher supersaturation. Scaling rate also increases with temperature, an effect due, to some extent, to an increase in supersaturation and perhaps to a greater extent to the acceleration of the kinetics of deposition process. The flow conditions become critical when the scale grows by ion transport to the surface – scaling rate increases with flow rate. This phenomenon is well documented for turbulent flow (Hasson et al, 1968), but less so for laminar flow, which is characteristic of flow through narrow polymer tubes such as might be used in heat exchangers. The shear removal of scale from polymer tubes has not been investigated extensively. Here, the difference in scale adherence, which will likely be related to the chemical and physical structure of the polymer, may be important. Lastly, the field of scaling research on polymers as well as metals is in need of innovation in characterization tools. Some progress has been made on this front with several reports of new in situ methods: microscopic imaging (Kim et al, 2002), rotating disc electrode monitoring (for metals) (Chen et al, 2005), attenuated total reflectance infrared sensing (Smith et al, 2004), quartz crystal microbalance monitoring (Kohler et al, 2001), optical fiber sensing (Lyons et al, 2001), and monitoring of laser reflectance and scattering (Euvrard et al, 2004; Sanft, 2005).

4.5 CONCLUSION Thermal characterization of heat exchangers immersed in solar water storage vessels for domestic hot water systems has been the subject of study for nearly 20 years. Yet predicting thermal performance remains a challenge because the transient natural convection heat transfer coefficients for such heat exchangers depend on the geometry of the heat exchanger/tank combination, the state of charge and the extent of thermal stratification. Based on recent work summarized in this paper, we recommend equation (19) to predict the transient natural convection Nusselt number during either charge or discharge. The advantage of this transient formulation is that only the initial Rayleigh number and an empirical constant C are required. The empirical constant must be determined for each heat exchanger/tank combination. The correlation is valid for fully mixed storage tanks. In most cases, charge or discharge of the tank with an immersed heat exchanger maintains a fully mixed tank and is likely to destroy existing stratification. In situations where stratification of the storage fluid is maintained by the use of manifolds or other devices, equation (19) is not applicable. In this case, correlations of the form of equation (4) may be used and it will be necessary to characterize the temperature of the storage fluid surrounding the heat exchanger during use. Either a validated transient zonal model of the heat exchanger/tank such as that used in TRNSYS or computational fluid dynamics models will be needed to predict the temperature distribution in the tank.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

125

As manufacturers and system integrators consider the use of polymer materials for immersed heat exchangers, durability and reliability of the heat exchanger must be considered. A critical aspect is the long-term mechanical stability of materials in water at elevated temperature. In addition, chlorinated water may create an oxidative environment in which certain polymers are chemically attacked, resulting in permanent loss of mechanical strength and increase in creep compliance. Water absorption and hydrolysis can also diminish polymer mechanical properties. The mechanism of polymer degradation will depend on the polymer molecular structure. Based on testing of polysulfone, (PSU), polybutylene (PB) and polyamide 6,6 (PA66), we do not recommend the use of polyamide 6,6 for this application. Data for polybutylene is inconclusive. Although the material did not degrade after 1100 hours of exposure to hot, chlorinated water, PB is expected to fail once the antioxidant additives have been depleted. Further investigation of the relationship between depletion of the antioxidant additives in polyolefins is recommended to determine the expected life of these materials. Polysulfone is a promising material as it is stable in the intended environment and does not require the use of antioxidants. All heat exchangers operated in an open loop with water supersaturated with respect to calcium carbonate or other inverse soluble salts are prone to scale. Measurement of scaling in a number of candidate polymer tubes indicates the scaling rate for polymers is of the same order of magnitude as that for copper. Additional research is needed to assess scale adhesion on polymers and to develop polymers that are resistant to scale.

ACKNOWLEDGEMENTS The authors appreciate the collaboration of Dr. Jay Burch at the US National Renewable Energy Laboratory, Professor F. A. Kulacki and graduate students at the University of Minnesota, and Professor Kelly Homan at the University of Missouri-Rolla. We gratefully acknowledge the financial support of the National Renewable Energy Laboratory, the US Department of Energy and the University of Minnesota Initiative for Renewable Energy and the Environment.

AUTHOR CONTACT DETAILS Dr. Jane H. Davidson (corresponding author), Department of Mechanical Engineering, University of Minnesota, 111 Church St., S.E., Minneapolis, MN 55455 e-mail: [email protected]; [email protected] Dr. Susan C. Mantell, Department of Mechanical Engineering, University of Minnesota, 111 Church St., S.E., Minneapolis, MN 55455 e-mail: [email protected] Dr. Lorraine F. Francis, Department of Chemical Engineering and Materials Science, University of Minnesota, 421 Washington Ave., S.E., Minneapolis, MN 55455

REFERENCES Aharoni, S. M. (1997) n-Nylons: Their Synthesis, Structure and Properties, John Wiley & Sons, New York Aizenberg, J., Black, A. J. and Whitesides, G. M. (1999) ‘Oriented growth of calcite controlled by self-assembled monolayers on functionalized alkanethiols supported on gold and silver’, J. Am. Chem. Soc., vol 121, pp4500–4509 Andritsos, N., Kontopoulou, M., Karabelas, A. J. and Koutsoukos, P. G. (1996) ‘Calcium carbonate deposit formation under isothermal conditions’, Can. J. Chem. Eng., vol 74, p911

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

126

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

Andritsos, N., Karabelas, A. J. and Koutsoukos, P. G. (1997) ‘Morphology and structure of CaCO3 scale layers formed under isothermal flow conditions’, Langmuir, vol 13, pp2873–2879 ASTM (2000) ‘F 2023-00: Standard test method for evaluating oxidative resistance of crosslinked polyethylene (PEX) tubing and systems to hot chlorinated water’, in Annual Book of ASTM Standards, American Society of Testing and Materials, West Conshohocken, PA, US ASTM (2001) ‘D 1238: Standard test method for melt flow rates of thermoplastics by extrusion plastometer’, in Annual Book of ASTM Standards, American Society of Testing and Materials, West Conshohocken, PA, US ASTM (2004) ‘D789-04: Standard test method for determination of relative viscosity of polyamide (PA)’, in Annual Book of ASTM Standards, American Society of Testing and Materials, West Conshohocken, PA, US Baker, D. K. (2000) ‘An investigation of calcium carbonate scaling rates based on experiments and modeling’, Ph.D. Thesis in Mechanical Engineering, University of Texas, Austin, US Baker, D. K. and Vliet, G. C. (2001) ‘Designing solar hot water systems for scaling environments’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 123, no 1, pp43–47 Baker, D. and Vliet, G. (2003) ‘Identifying and reducing scaling problems in solar hot water systems’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 125, pp61–66 Banas, M. D. and Baier, R. E. (2000) ‘Accelerated mineralization of prosthetic heart valves’, Mol. Cryst. and Liq. Crystl., vol 354, pp249–267 Basell Polyolefins (2004) Product Data Sheet: Polybutylene-1, PB 4137 Grey Bejan, A. (1982) Entropy Generation Through Heat and Fluid Flow, John Wiley & Sons, New York, p165 Ben Amor, M., Zgolli, D., Tlili, M. M. and Manzola, A. S. (2004) ‘Influence of water hardness, substrate nature and temperature on heterogeneous calcium carbonate nucleation’, Desalination, vol 166, pp79–84 Bradley, S. W., El-Hibri, J., Bersted, B. H. and Bradley, W. L. (2000) ‘A study of the effect of chlorinated water on engineering thermoplastics at elevated temperatures’, presented at ANTEC, Orlando, US Branch, C. A. and Muller-Steinhagen, H. M. (1991) ‘Influence of scaling on the performance of shell-and-tube heat exchangers’, Heat Transfer Engineering, vol 12, pp37–45 Boies, A. M. and Homan, K. O. (2004) ‘Improving discharge characteristics of indirect integral collector storage systems with multi-element storage’, ASME International Solar Energy Conference, Portland, OR, US, June 2004, CDROM Boies, A. M. and Homan, K. O. (2005) ‘System performance of an indirect collector storage preheater in series with a conventional electric resistance water heater’, 2005 ISES Solar World Congress, Orlando, FL, US, August 2005, ASME, paper ISEC2005-76200, CDROM Boies, A. M., Homan, K. O., Davidson, J. H. and Liu, W. (2005) ‘A variable effectiveness model for indirect thermal storage devices’, 2005 ASME Summer Heat Transfer Conference, San Francisco, CA, US, July 2005, ASME, paper HT2005-72711, CDROM Bott, T. R. (1988) ‘Crystallisation fouling – Basic science and models’, Fouling Science and Technology, vol 145, pp251–260 Budair, M. O., Khan, M. S., Zubair, S. M., Sheikh, A. K. and Quddus, A. (1998) ‘CaCO3 scaling in AISI 316 stainless steel tubes – Effect of thermal and hydraulic parameters on the induction time and growth rate’, Heat & Mass Transfer, vol 34, pp163–170 Chen, T., Neville, A., and Yuan, M. (2005) ‘Calcium carbonate scale formation – Assessing the initial stages of precipitation and deposition’, J. Petroleum Science and Engineering, vol 46, pp185–194 Cowan, J. C. and Weintritt, D. J. (1976) Water-Formed Scale Deposits, Gulf Publishing Co., Houston, TX, US Dalas, E., Kallitsis, J. and Koutsoukos, P. G. (1988) ‘The crystallization of calcium carbonate on polymeric substrates’, J. Crystal Growth, vol 89, pp287–294 Dalas, E., Klepetsanis, P. and Koutsoukos, P. G. (1999) ‘The overgrowth of calcium carbonate on poly(vinyl chloride)-co-vinyl acetate-co-maleic acid’, Langmuir, vol 15, pp8322–327

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

127

Dalas, E., Klepetsanis, P. G. and Koutsoukos, P. G. (2000) ‘Calcium carbonate deposition on cellulose’, J. Colloid. Inter. Sci., vol 224, pp56–62 Davidson, J. H., Mantell, S. C. and Jorgensen, G. (2002) ‘Status of the development of polymeric solar water heating systems’, in D.Y. Goswami (ed) Advances in Solar Energy, American Solar Energy Society, vol 15, pp149–186 Davidson, J. H., Ragoonanan, V. and Homan, K. (2005) ‘Divided storage in an indirect integral collector storage with immersed heat exchanger’, Proceedings of the 2005 Solar World Congress, Orlando, FL, US Dousi, E., Kallitsis, J., Chrissanthopoulos, A., Mangood, A. H. and Dalas, E. (2003) ‘Calcite overgrowth on carboxylated polymers’, Journal of Crystal Growth, vol 253, pp496–503 Drück, H. (2002) ‘Influence of different combistore concepts on the overall system performance’, IEA-SHC Task 26, Industry Workshop, Oslo, Norway, pp39–46 DuPont (2004) Product Data Sheet: Zytel 103 HSL NC010 Epstein, N. (1986), ‘Fouling of heat exchangers’, in J. W. Palen (ed.) Heat Exchanger Source Book, Hemisphere Publishing Corporation, New York, pp677–697 Euvrard, M., Membrey, F., Filiatre, C. and Foissy, A. (2004) ‘Crystallization of calcium carbonate at a solid/liquid interface examined by reflectance of a laser beam’, Journal of Crystal Growth, vol 265, pp322–330 Farrington, R. B. and Bingham, C. E. (1986) ‘Testing and analysis of immersed heat exchangers’, SERI/TR-253-2866, National Renewable Energy Laboratory, Golden, Colorado, US Farrington, R. B. and Bingham, C. E. (1987) ‘Testing and analysis of load-side immersed heat exchangers for solar domestic hot water systems’, SERI/TR-254-3094, National Renewable Energy Laboratory, Golden, Colorado, US Feiereison, T. J., Klein, S. A., Duffie, J. A. and Beckman, W. A. (1982) ‘Heat transfer from immersed coils’, Paper No 82 WA/SOL-18, American Society of Mechanical Engineers, New York Ferguson, R. J. (1984) ‘A kinetic model for calcium carbonate deposition’, Materials Performance, vol 23, pp25–34 Freeman, A. (2004) ‘Mechanical performance of polysulfone, polybutylene, and nylon 6,6 in hot, chlorinated water’, MS Thesis in Mechanical Engineering, University of Minnesota, Minneapolis, US Freeman, A., Mantell, S. C. and Davidson, J. H. (2005) ‘Mechanical performance of polysulfone, polybutylene, and polyamide 6,6 in hot chlorinated water’, Solar Energy, vol 79, pp624–637. Gedde, U. W., Viebke, J., Leijstrom, H. and Ifwarson, M. (1994) ‘Long-term properties of hot-water polyolefin pipes – A review’, Polymer Engineering and Science, vol 34, no 24, pp1773–1787 Githens, R. E., Minor, W. R. and Tomsic, V. J. (1965) ‘Flexible tube heat exchangers’, Chem. Eng. Prog., vol 61, no 7, pp55–62 Hasson, D. (1981); ‘Precipitation fouling: A review’, in E. F. C. Somerscals and J. G. Knudsen (eds) Fouling of Heat Transfer Equipment, Hemisphere Publishing Corporation, New York, pp527–568 Hasson, D., Avriel, M., Resnick, W., Rozenman T. and Windreich, S. (1968) ‘Mechanism of calcium carbonate scale deposition on heat-transfer surfaces’, I&EC Fundamentals, vol 7, no 1, pp59–65 Kanakis, J. and Dalas, E. (2000) ‘The crystallization of vaterite on fiber’, Journal of Crystal Growth, vol 219, pp277–282 Karabelas, A. J. (2002) ‘Scale formation in tubular heat exchangers – Research priorities’, International Journal of Thermal Science, vol 41, pp682–692 Kazmierczak, T. F. and Tomson, M. B. (1982) ‘Crystal growth of calcium carbonate: A controlled composition kinetic study’, J. Phys. Chem., vol 86, pp103–107 Keysar, S., Semiat, R., Hasson, D. M. and Yahalom, J. (1994) ‘Effect of surface roughness on the morphology of calcite crystallizing on mild steel’, J. Colloid and Interface Sci., vol 162, p311 Khan, M. S., Zubair, S., Budair, M., Sheikh A. and Quddus, A. (1996) ‘Fouling resistance model for prediction of CaCO3 scaling in AISI 316 tubes’, Heat and Mass Transfer, vol 32, pp1–2 Kim, W. T., Bai, C. and Cho, Y. I. (2002) ‘A study of CaCO3 fouling with a microscopic imaging technique’, International Journal of Heat and Mass Transfer, vol 45, pp597–607

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

128

J. H. DAVIDSON, S. C. MANTELL AND L. F. FRANCIS

Knudsen, J. G. (1981) ‘Cooling water fouling – A brief review’, American Society of Mechanical Engineers, Heat Transfer Division, vol 17, pp29–38 Kohler, N., Courbin, G. and Ropital, F. (2001) ‘Static and dynamic evaluation of calcium carbonate scale formation and inhibition’, European Formation Damage Control Conference, Expanding Horizons, Society of Petroleum Engineers, The Hague, The Netherlands Liu, W., Davidson, J. H., and Mantell, S. C. (2000) ‘Analysis of polymer heat exchangers for solar water heating: A case study’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 122, no 2, pp84–91 Liu, W., Davidson, J. H., Mantell, S. C. and Kulacki, F. A. (2003) ‘Natural convection from a horizontal tube heat exchanger immersed in a tilted enclosure’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 125, no 1, pp67–75 Liu, W., Davidson, J. H. and Kulacki, F. A. (2004) ‘Natural convection from a tube bundle in a thin inclined enclosure’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 126, no 2, pp702–709 Liu, W., Davidson, J. H., and Kulacki, F. A. (2005) ‘Thermal characterization of prototypical integral collector storage systems with immersed heat exchangers’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 127, no 1, pp21–28 Lyons, W. B., Ewald, H., Flanagan, C., Lochmann, S. and Lewis, E. (2001) ‘A neural networks based approach for determining fouling of multipoint optical fibre sensors in water systems’, Meas. Sci. Technol., vol 12, pp958–965 Mather, D., Hollands, K. G. T. and Wright, J. L. (2002) ‘Single- and multi-tank energy storage for solar heating systems: fundamentals’, Solar Energy, vol 73, no 1, pp3–13 Marner, W. J. and Suitor, J. W. (1987) Fouling with Convective Heat Transfer, John Wiley & Sons, New York McGill, K. N. (2005) ‘Calcium carbonate scaling of polymers in isothermal, stagnant water’, Mechanical Engineering, University of Minnesota, Minneapolis, MN, US Morgan, V. T. (1975) ‘The overall convective heat transfer from smooth circular cylinders’, Advances in Heat Transfer, vol 11, pp199–264 Mullin, J. W. (1993) Crystallization, Butterworth-Heinemann, Oxford, UK Nancollas, G. H. (1979) ‘The growth of crystals in solution’, Advances in Colloid and Interface Science, vol 10, pp215–252 Nancollas, G. H. and Reddy, M. M. (1971) ‘The crystallization of calcium carbonate: II Calcite growth mechanism’, J. Colloid and Interface Sci., vol 36, pp824 NSF (1999) ‘NSF Protocol P171: Chlorine resistance of plastic piping materials’, National Sanitary Foundation International, Ann Arbor, MI, US Reddy, M. M. and Nancollas, G. H. (1971) ‘The crystallization of calcium carbonate: I. Isotopic exchanger and kinetics’, J. Colloid and Interface Science, vol 36, p166 Reindl, D. T. (1992) ‘Source driven transient natural convection in enclosures’, Ph.D. Thesis in Mechanical Engineering, University of Wisconsin, Madison, US Reindl, D. T., Beckman, W. A. and Mitchell, J. W. (1992a) ‘Transient natural convection from a vertical flat plate in a rectangular enclosure’, Proceedings, ASME 28th National Heat Transfer Conference and Exhibition, San Diego, CA, US, pp91–98 Reindl, D. T., Beckman, W. A. and Mitchell, J. W. (1992b) ‘Transient natural convection in enclosures with application to solar thermal storage tanks’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 114, pp175–181 Roques, H. and Girou, A. (1974) ‘Kinetics of the formation conditions of carbonate tartars’, Water Research, vol 8, p907 Sanft, P. M. (2005) ‘Development of an apparatus for in situ analysis of calcium carbonate formation on polymer surfaces’, MS Thesis in Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, US Sanft, P., Francis, L. F. and Davidson, J. H. (2006) ‘Calcium carbonate formation on cross-linked polyethylene (PEX) and polypropylene random copolymer (PP-r)’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 128, no 2, pp251–254 Scott, G. (1999) Polymers and the Environment, The Royal Society of Chemistry, Cambridge, UK

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

Thermal and Material Characterization of Immersed Heat Exchangers

129

Sekulic, D. P. and Krane, R. J. (1992a) ‘Use of multiple storage elements to improve the second law efficiency of a thermal storage system. Part I: Analysis of the storage process’, in A. Valero and G. Tsatsaronis (eds) ECOS ‘92 International Symposium on Efficiency, Cost, Optimization, ASME, New York, pp61–66 Sekulic, D. P. and Krane, R. J. (1992b) ‘Use of multiple storage elements to improve the second law efficiency of a thermal storage system. Part II: Completion of the analysis and presentation of results’, in A. Valero and G. Tsatsaronis (eds) ECOS ‘92 International Symposium on Efficiency, Cost, Optimization, ASME, New York, pp67–72 Smith, J. K., Yuan, M., Lopez, T. H., Means, M. and Przybylinski, J. L. (2004) ‘Real time and in-situ detection of calcium carbonate scale in a west Texas oil field’, Society of Petroleum Engineers Production and Facilities, vol 19, pp94–99 Snoeyink, V. L. and Jenkins, D. (1980) Water Chemistry, John Wiley & Sons, New York Solvay Advanced Polymers (2002) Product Data Sheet: P-1700 NT11, NT06 & CL2611, Alpharetta, GA, US Stumm, W. and Morgan, J. J. (1996) Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters, John Wiley & Sons, New York Su, Y. and Davidson, J. H. (2005) ‘Natural convective flow and heat transfer in a collector storage with an immersed heat exchanger: Numerical study’, Journal of Solar Energy Engineering, Transactions of ASME, vol 127, no 3, pp324–332 Vliet, G. C. and Baker, D. K. (1998) ‘Designing solar hot water systems for scaling environments’, Proceedings of the 1998 American Solar Energy Society Annual Conference, Albuquerque, NM, US, pp307–317 Walter, D., Mantell, S. C. and Davidson, J. (2003) ‘Performance of polymer tubes in hot chlorinated water’, SOLAR 2003: Conference Proceedings, American Solar Energy Society, pp77–82 Wang, Y. (2004) ‘Experimental study of calcium carbonate scaling on polymer tubes and films’, MS Thesis in Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, US Wang, Y., Davidson, J. H. and Francis, L. F. (2005) ‘Scaling in polymer tubes and interpretation for use in solar water heating systems’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 127, pp3–14 Weiss, W. (2003) Solar Heating Systems for Houses: A Design Handbook for Solar Combisystems, International Energy Agency, London and James and James, London Wu, C., Mantell, S. C. and Davidson, J. H. (2004) ‘Polymers for domestic solar hot water: Long-term performance of PB and nylon 6,6 in hot water’, Journal of Solar Energy Engineering, Transactions of the ASME, vol 126, no 1, pp581–586

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 99–129

5

Photocatalytic Detoxification of Water with Solar Energy Sixto Malato, Julián Blanco, Diego C. Alarcón, Manuel I. Maldonado, Pilar Fernández-Ibáñez and Wolfgang Gernjak

Abstract During the past ten years there has been tremendous research and development in the area of photocatalysis. One of the major applications of this technology is the degradation of organic pollutants in water, by what are called ‘advanced oxidation processes’ (AOPs). This paper reviews the use of sunlight to produce •OH radicals for AOPs. The pilot plant-scale systems necessary for performing solar photocatalytic experiments, including the basic pilot plant components and the fundamental parameters related to solar photocatalytic reactions, are described. The paper also summarizes most of the recent research related to solar photocatalytic degradation of water contaminants, and how it could significantly contribute to the treatment of very persistent toxic compounds. It further describes the possibility of using the solarilluminated photo-Fenton reaction to extend the horizon of solar photocatalysis applications. Various solar reactors for photocatalytic water treatment based mainly on non-concentrating collectors erected during the last few years are also described in detail in the last part of this review.

■ Keywords – photocatalysis; non-concentrating collectors; reaction rate; mineralization; electrons; degradation; collector efficiency

5.1 INTRODUCTION The main causes of surface and groundwater contamination are industrial discharges (even in small amounts), excessive use of pesticides, fertilizers (agro-chemicals) and domestic waste landfills. Wastewater treatment is based on various mechanical, biological, physical and chemical processes. In fact, it is a combination of many operations, such as filtration, flocculation, sterilization or chemical oxidation of organic pollutants. After filtration and elimination of particles in suspension, the ideal process is biological treatment (natural decontamination). Unfortunately, some organic pollutants, classified as bio-recalcitrant, are not biodegradable. Advanced oxidation processes (AOPs) may be used for

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

131

decontamination of water containing these compounds (Andreozzi et al, 1999; Gogate and Pandit, 2004a; Legrini et al, 1993; Pera-Titus et al, 2004). These methods rely on the formation of highly reactive chemical species which degrade even the most recalcitrant molecules into biodegradable compounds. Although there are different reacting systems, all of them are characterized by the same chemical feature: production of OH radicals (•OH), which are able to oxidize and mineralize almost any organic molecule, yielding CO2 and inorganic ions. Rate constants (kOH, r = kOH [•OH] C) for most reactions involving hydroxyl radicals in aqueous solution are usually in the order of 106 to 109M–1s–1. They are also characterized by their not very selective attack, which is a useful attribute for wastewater treatment and solution of pollution problems. The versatility of AOPs is also enhanced by the fact that there are different ways of producing hydroxyl radicals, facilitating compliance with the specific treatment requirements. Methods based on UV, H2O2/UV, O3/UV and H2O2/O3/UV combinations use photolysis of H2O2 and ozone to produce the hydroxyl radicals. Other methods, like heterogeneous photocatalysis and homogeneous photo-Fenton, are based on the use of a wide band-gap semiconductor and the addition of H2O2 to Fe2+ salts, respectively, and irradiation with UV-VIS light. Since 1976, photocatalytic detoxification has been discussed in the literature as an alternative method for cleaning up polluted water (Carey et al, 1976). Today’s renewed interest (Pignatello, 1992) in the photo-assisted Fenton process, discovered by Fenton in the 19th century, is underlined by a significant number of studies devoted to wastewater treatment (Gogate and Pandit, 2004b). Both these processes are of special interest since sunlight can be used for them. The main disadvantage of AOPs is their high cost (expensive reactants such as H2O2 and UV generation). Future applications of these processes could therefore be improved through the use of catalysis and solar energy. The heterogeneous solar photocatalytic detoxification process consists of making use of the near-ultraviolet (UV) band of the solar spectrum (wavelength shorter than 380nm) to photo-excite a semiconductor catalyst in the presence of oxygen (Goswami and Blake, 1996). Under these circumstances, oxidizing species, either bound hydroxyl radicals (•OH) or free holes, which attack oxidizable contaminants, are generated producing a progressive break-up of molecules yielding CO2, H2O and diluted inorganic acids. The most commonly used catalyst is the semiconductor TiO2, which is cheap, non-toxic and abundant. The homogeneous solar photocatalytic detoxification process (photo-Fenton) is based on the production of •OH radicals by Fenton reagent (H2O2 added to Fe2+ salts). The rate of degradation of organic pollutants with Fenton-like reagents is strongly accelerated by irradiation with UV-VIS light. This is an extension of the Fenton process which can make use of UV-VIS light irradiation at wavelengths over 300nm. Under these conditions, the photolysis of Fe3+ complexes enables Fe2+ regeneration and Fenton reactions due to the presence of H2O2 (Bauer et al, 1999; Sagawe et al, 2001). Although these processes have been studied for at least two decades, industrial/commercial applications, engineering systems and engineering design methodologies have only been developed recently (Blanco and Malato, 2003; Goswami et al, 1997 and 2003). This paper summarizes such work done during the last decade and attempts to continue previous work published in Advances in Solar Energy on this subject (Goswami, 1995).

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

132 S. MALATO et al

5.2 SOLAR COLLECTORS FOR PHOTOCHEMISTRY 5.2.1 CONCENTRATING COLLECTORS Contrary to solar thermal processes, which collect large amounts of photons at any wavelength to reach a specific temperature range, solar photochemical processes use only high-energy short wavelength photons to promote photochemical reactions. Most solar photochemical processes use UV or near-UV sunlight (300 to 400nm), but in some photochemical synthesis processes, sunlight up to 500nm can be absorbed and photoFenton heterogeneous photocatalysis uses sunlight up to 580nm. Sunlight at wavelengths over 600nm is normally not useful in any photochemical process. Nevertheless, the specific hardware needed for solar photochemical applications has much in common with those used for thermal applications. As a result, both photochemical systems and reactors have followed conventional solar thermal collector designs, such as parabolic troughs and non-concentrating collectors (Goswami, 1997). At this point, their design begins to diverge, since in photochemical processes: (i) the fluid must be directly exposed to the solar radiation and, therefore, the absorber must be transparent to the photons, and (ii) temperature usually does not play a significant role, so no insulation is required. The original solar photo-reactor designs (Goswami, 1995) for photochemical applications were based on line-focus parabolic trough concentrators (PTCs). In part, this was a logical extension of the historical emphasis on trough units for solar thermal applications. Furthermore, PTC technology was relatively mature and existing hardware could be easily modified for photochemical processes. The parabolic trough collector consists of a structure that supports a reflective concentrating parabolic surface. This structure has one or two motors controlled by a solar tracking system on one or two axes respectively which keep the collector aperture plane perpendicular to the solar rays. In this situation, all the solar radiation available on the aperture plane is reflected and concentrated on the absorber tube that is located at the geometric focal line of the parabolic trough. The first outdoor engineering-scale reactor developed (in the US) was a converted solar thermal parabolic trough collector in which the absorber/glazing tube combination had been replaced by a simple Pyrex glass tube through which contaminated water could flow. It was designed and built jointly by the Solar Energy Research Institute (SERI), now known as the National Renewable Energy Laboratory (NREL), and Sandia National Laboratories at the end of the 1980s. This system was installed at the National Solar Thermal Test Facility at the Sandia National Laboratories in Albuquerque, New Mexico (US) in 1989 (Pacheco et al, 1990; Blake, 1994). The facility was made up of a total of six aligned parabolic trough collectors with single-axis solar tracking, with an aperture of 2.13m and a length of 36.4m, for a total of 465m2 aperture area. The collector concentrated the sunlight about 50 times on the photo-reactor (Anderson et al, 1991; Pacheco and Tyner, 1990; Prairie et al, 1992). Immediately afterwards, in the 1990s, a similar facility was designed and built at the Plataforma Solar de Almería. The first engineering-scale solar photochemical facility for water detoxification in Europe was developed by Centro de Investigaciones Medioambientales, Energéticas y Tecnológicas (CIEMAT) (Minero et al, 1993), using 12 two-axis solar tracking PTCs, each having a total of 32 mirrors in 4 parallel parabolas with a collecting area of 32m2. Typical overall optical efficiencies obtained in PTCs were around

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

133

50 per cent (Curcó et al, 1996), mainly due to: ● the solar tracking error, attributable to the control system (ds); ● the module construction (dC); ● the reflectivity (which is a function of the wavelength) of the aluminized surface of

the parabolic mirrors (dR,h). As this surface is outdoors it gets dirty and can be damaged, requiring that the reflectivity should be measured periodically; and ● the spectral transmissivity of the absorber tube (dT,h). Figure 5.1 shows the path of direct radiation (ID) until it arrives inside the absorber tube. It must reach the surface and be reflected (part is lost due to dR,h) in the right direction (here affected by ds) by the real mirror surface (dC), before penetrating (dT,h) in the tube. Furthermore, the parabolic trough concentration factor must also be considered (ratio of surface area of the parabola capturing the radiation and surface area of the tube, Sp/ST). Therefore the effective photon flux corresponding to the direct UV inside the absorber (ID,E) is:

I D ,E = f I D ,

Sp Sr

,ηc ,ηR ,ηTl

(1)

Global Radiation is also collected by the PTCs, but only that small fraction which falls directy on the transparent absorber tube without intervention of the collector and is only affected by the transmissivity of the glass, dT,h:

I G E = f I G ,ηT λ

(2)

where IG,E is the effective photon flux corresponding to global UV inside the absorber and therefore the total radiation reaching a PTC is the sum of both components.

U U U U

UU U

U U Paradigm Paradigm

U

UU Paradigm Paradigm

U ID U IG U Paradigm ID Paradigm ID Paradigm ID,E Paradigm U Paradigm Paradigm Paradigm UParadigm Paradigm IG Paradigm IG,E Paradigm Paradigm Paradigm

FIGURE 5.1 The various loss factors (d) affecting the photon flux (I) inside a PTC photo-reactor

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

134 S. MALATO et al

Both these facilities, which were followed by others, were based on parabolic trough collectors with hundreds of square meters of collector surface and can be considered the starting point of solar photocatalytic technology development (Goswami, 1995). Parabolic trough collectors make efficient use of direct solar radiation and, as an additional advantage, the thermal energy collected from the concentrated radiation could simultaneously be used for other applications. The reactor is small, while receiving a large amount of energy per volume unit. The flow is turbulent and volatile compounds do not evaporate, making handling and control of the liquid to be treated simple and inexpensive. The main disadvantages are that the collectors (i) use only direct radiation, (ii) are expensive and (iii) have low optical and quantum efficiencies (at least for TiO2 applications) (Alfano et al, 2000).

5.2.2 NON-CONCENTRATING COLLECTORS One-sun (non-concentrating) collectors are, in principle, cheaper than PTCs, as they have no moving parts or solar tracking devices (Wyness et al, 1994a and 1994b). They do not concentrate radiation, so efficiency is not reduced by factors associated with concentration and solar tracking. Manufacturing costs are cheaper because their components are simpler, which also means easy, low-cost maintenance. Nonconcentrating collector support structures are also easier and cheaper to install and the surface required for their installation is smaller because, since they are static, they do not project shadows on the others. They are able to utilize the diffuse as well as the direct portion of the solar UV-A. It should be noted that with AM (air mass) at 1.5, the diffuse and direct portion of the solar radiation reaching the surface of the earth are almost the same. This means that a light concentrating system, for example a PTC, can in principle only employ half of the solar radiation available in this particular spectral region. An extensive effort in the design of small non-tracking collectors has resulted in the testing of several different non-concentrating solar reactors: ● Free-falling film. The process fluid falls slowly over a tilted plate which faces the

sun and is open to the atmosphere (Goslich et al, 1997a), with a catalyst supported in a permanent matrix on the reactor surface (Guillard et al, 2003) or suspended or dissolved in the fluid (photo-Fenton applications) (Gernjak et al, 2003). Two large pilotscale fixed-bed photocatalytic reactors were tested in Australia (Feitz et al, 2000). One (6.5 x 0.5m, total volume 200L) was based on a TiO2-coated (5g m–2) woven glass fiber mesh supported on an inclined corrugated support. The other was based on a packed bed configuration (1 x 2m2) containing TiO2-coated raschig rings (6.5cm deep). The packed bed configuration processed a similar concentration of phenol seven times faster under similar solar light conditions. ● Pressurized flat plate. This consists of two plates between which the fluid circulates using a separating wall (Dillert et al, 1999a). A kind of this nonconcentrating reactor is the double skin sheet reactor (DSSR), which consists of a flat and transparent structured box made of Plexiglas♦ (Well et al, 1997), with channels and walls for water circulation. The structure of the reactor is schematically drawn in Figure 5.2. The suspension containing the model pollutant and the photocatalyst is

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

135

pumped through these channels. The Plexiglas♦ used to manufacture the double skin sheets has high transmittance in the UV-A portion of the solar spectrum below 400nm. A similar system (based on a double skin acrylic panel) was developed by the Solar Energy and Energy Conversion Laboratory (University of Florida) for the treatment of BTEX-contaminated groundwater (Goswami, 1995). A detailed description and discussion of the experiments can be found in Srinivasan et al (1997 and 1998). ● Solar ponds. These are small, shallow on-site pond reactors (Bedford et al, 1994; Giménez et al, 1999). The flat reactor resides in a cylindrical tank in which several perforated tubes lie at the bottom, through which air circulates, bubbling the solution to be treated. The system operates in a discontinuous mode and the catalyst is maintained in suspension by using the air stream. Air also supplies the oxygen necessary for the pollutants’ oxidation. Based on all of the above, the main advantages and disadvantages of the different technologies for solar photocatalytic applications can be summarized.

FIGURE 5.2 Structural schematic view of a DSSR reactor showing water flowing in the transparent box. The channels are shown connected in series. The channels may also be connected in parallel to reduce pressure drop

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

136 S. MALATO et al

Advantages of non-concentrating systems: 1 they can make use of both direct and diffuse solar radiation; 2 they are simpler systems, with lower investment cost and fewer maintenance requirements; 3 water may not heat up significantly if no special measures are taken to that effect; and 4 they have both high optical efficiency and high quantum efficiency since there is a lower recombination of e–/h+, given that the photonic density is lower than in a concentrating system. Disadvantages of non-concentrating systems: 1 construction problems associated with weather resistance, chemical inertness and ultraviolet transmission; 2 they usually work with laminar flow, which can cause mass transfer problems in photocatalysis; 3 reactants vaporization; and 4 reactants contamination. Advantages of concentrating systems: 1 they have a noticeably smaller reactor tube area (a shorter loop in which to confine, control and handle the water); 2 more practical use of a supported catalyst; 3 turbulent flow (favors mass transfer and avoids possible catalyst sedimentation problems); and 4 there is no evaporation of volatile compounds. Disadvantages of concentrating systems: 1 they can only make use of direct solar radiation; 2 usually expensive; 3 lower efficiency, both optical and quantum, the second arising from a higher recombination of e–/h+ than in non-concentrating systems; and 4 there can be problems with water overheating.

5.2.3 COMPOUND PARABOLIC CONCENTRATORS (CPCs) The non-concentrating devices referred to above are of the flat plate type, even when tubes are directly used, since they are just laid side by side. But as mentioned, there is a category of low concentration collectors called Compound Parabolic Concentrators (CPCs), which are used in thermal applications both with fins in non-evacuated concentrators and with tubes in vacuum collectors. These are an interesting option for thermal applications since they combine characteristics of parabolic concentrators and static flat systems. They concentrate solar radiation but they conserve the properties of the flat plate collectors,

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

137

being static and collecting diffuse radiation. Thus they also constitute a good option for solar photochemical applications (May et al, 1991). CPCs are static collectors with a reflective surface designed to be ideal in the sense of non-imaging optics and can be designed for any given reactor shape. When the absorber has another shape a more precise way of describing them would be CPC-type collectors. CPCs were invented in the 1960s (Welford and Winston, 1978) to achieve solar concentration with static devices (Collares-Pereira, 1995), since they were able to concentrate on the receiver all the radiation that arrives within the collector’s ‘angle of acceptance’. They do this by illuminating the complete perimeter of the receiver, rather than just the ‘front’ of it, as in conventional flat plates or in the case of tubes laid side by side. These concentrating devices have ideal optics, thus maintaining both the advantages of the PTCs and static systems. The concentration factor (CCPC) of a two dimensional CPC collector is given by Equation 3:

CCPC =

1 sinθ a

(3)

The normal values for the semi-angle of acceptance (ea) for photocatalytic applications are going to be between 60 and 90 degrees. This wide angle of acceptance allows the receiver to collect both direct and a large part of the diffuse light (1/C of it), with the additional advantage of decreasing errors of both the reflective surface and receiver tube alignment, which become important for achieving a low cost photo-reactor. A special case is that in which ea = 90°, whereby CCPC = 1 (non-concentrating solar system) and each CPC curve is an ordinary involute (Figure 5.3a). When this occurs, all the UV radiation that reaches the aperture area of the CPC (direct and diffuse) can be collected and redirected to the reactor. If the CPC is designed for an acceptance angle of +90° to –90°, all incident solar diffuse radiation can be collected. The light reflected by the CPC is distributed all around the tubular receiver (Figure 5.3b) so that almost the entire circumference of the receiver tube is illuminated and the light incident on the photo-reactor is the same as would impinge on a flat plate. The advantages of CPCs can be summarized as follows. Just as in the parabolic trough collector, in CPC-type collectors, the photo-reactor is tubular so that the water can be pumped easily. CPC devices for photocatalytic applications are generally fabricated with aluminum reflectors and the structure is usually made of a simple frame which, in turn, forms the support for connecting the glass tubes that make up the photo-reactor (Romero et al, 1999a; Blanco et al, 2000). CPCs have the advantage of both technologies (PTCs and non-concentrating collectors) and none of the disadvantages, apparently making them the best option for photocatalytic processes based on the use of solar radiation (Ajona and Vidal, 2000; Blanco et al, 2000; Jubran et al, 1999; Malato et al, 2002a and 2003a; Robert et al, 1999). Advantages include: 1 they can make highly efficient use of both direct and diffuse solar radiation without the need for solar tracking (Blanco et al, 1999); 2 there is no evaporation of possible volatile compounds and water does not heat up; 3 they have high optical efficiency, since they make use of almost all the available radiation, and high quantum efficiency, as they do not receive a concentrated flow of photons; and 4 flow is turbulent inside the tube reactor.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

138 S. MALATO et al

U

‘A’ Paradigm Paradigm Paradigm Paradigm Paradigm Paradigm Paradigm

U U

U

Paradigm U

First First Paradigm Paradigm

‘A’ First First

FIGURE 5.3 Schematic drawing of CPCs

5.3 FUNDAMENTAL PARAMETERS IN SOLAR PHOTOCATALYSIS 5.3.1 INITIAL CONTAMINANT CONCENTRATION INFLUENCE As oxidation proceeds, less and less of the surface of the TiO2 particles is covered as the contaminant is decomposed. Evidently, at total decomposition, the rate of degradation is zero and a decreased photocatalytic rate is to be expected with increasing illumination time. Most authors agree that, with minor variations, the expression for the rate of photomineralization of organic substrates with irradiated TiO2 follows the LangmuirHinshelwood (L-H) law for the same saturation-type kinetic behavior in any of the four possible situations: 1 the reaction takes place between two adsorbed substances; 2 the reaction occurs between a radical in the solution and the adsorbed substrate; 3 the reaction takes place between the radical linked to the surface and the substrate in the solution; or 4 the reaction occurs with both species in solution. In all cases, the expression is similar to the L-H model. From kinetic studies only, it is therefore not possible to find out whether the process takes place on surfaces or in solution. Although the L-H isotherm has been useful in modeling the process, it is generally agreed that both rate constants and orders are only ‘apparent’ (Cunningham and

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

139

Sedlak, 1996; Minero, 1999). They serve to describe the rate of degradation, and may be used for reactor optimization, but they have no physical meaning and may not be used to identify surface processes. Thus while not a useful tool for describing the active species involved in oxidation, engineers and solar designers seem to have a common understanding of the usefulness of the unmodified L-H model. For L-H standard data treatment it is therefore assumed that the reaction occurs on the surface, which is also the assumption most widely accepted as possible. Under these conditions, two extreme situations are defined to illustrate the adsorption on the catalyst surface: (1) substrate and water compete for the active catalyst sites, and (2) the reactant and the solvent are adsorbed on the surface without competing for the same active catalyst sites. According to the L-H model, the reaction rate (r) is proportional to the fraction of surface covered by the substrate (ex). In each case the following expression can be obtained:

r=-

kr KC dC = krex = dt 1 + KC + K sCs

(4)

k KC dC = krθ x = r dt 1 + KC

(5)

r=-

where kr is the reaction rate constant, K is the reactant adsorption constant, C is the concentration at any time, KS is the solvent adsorption constant and CS is its concentration (in water CS 5 55.5 M). As CS >> C and CS remains practically constant, the part of the catalyst covered by water is unalterable over the whole range of C and the previous equations can be integrated. Using an L-H model, graphics similar to those depicted in Figure 5.4 may be obtained from the experimental data and from the linearization of the previous equations. The effect of the initial concentration on the degradation rate is shown in Figure 5.4; due to the saturation produced on the semiconductor surface as the concentration of the reactant increases, it reaches a point at which the rate becomes steady. It should be emphasized that photo-decomposition gives rise to intermediates, which could also be adsorbed competitively on the surface of the catalyst. The concentration of these intermediates varies throughout the reaction up to their mineralization and thus Equation 6 may also take the following form:

r0

r0–1

t1/2

U

U U

UU

U

U

U

FIGURE 5.4 Graphics related to the adjustment of data to an L-H type kinetic model

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

140 S. MALATO et al

r=

kr KC (6)

n

1 + KC + Â K iCi (i = 1, n ) i =1

UU

U

UU

UU UU

rpublished

published published L–1

where i is the number of intermediates formed during degradation (the solvent is also included in the summation). An understanding of the reaction rates and how the reaction rate is influenced by different parameters is important for the design and optimization of an industrial system. The L-H reaction rate constants are useful for comparing the reaction rate under different experimental conditions. Once the reaction constants kr and K have been evaluated, the disappearance of reactant can be estimated if all other factors are held constant. Due to this, a series of tests at different initial substrate concentrations must be performed to demonstrate whether the experimental results could be adjusted with this model. The concentration range must be wide enough to allow correct fit of the L-H linearization. This means from the lowest concentration at which the initial rate could be determined until the limit where the relationship between initial reaction and initial concentration remains constant (see Figure 5.4). Since hydroxyl radicals react non-selectively, numerous intermediates are formed en route to complete mineralization at different concentrations. Because of this, all tests should be carried out using TOC (total organic carbon) as a crucial parameter (instead of concentration of parent compound, C), because the photocatalytic treatment must destroy not only the initial contaminant, but any other organic compound as well. The results shown in Figure 5.5 are examples of the experiments carried out with mixtures of different commercial pesticides. It is possible to see that mineralization, once begun, maintains the same slope until at least 60–70 per cent of the initial TOC has been degraded.

U

UU

U

UU UU U

UU

UU –1 LU UU UUU UU U

UU

U U

U

U

UU

UU

Q, kJ L–1

UU

UU

UU

FIGURE 5.5 Pesticide decomposition at different initial concentrations. ‘Maximum rate’ as a function of TOCmax is shown in the inserted graphic

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

141

As the reaction is not expected to follow simple models like first- or zero-order kinetics, overall reaction rate constants cannot be calculated. The complexity of the results, of course, is caused by the fact that the TOC is a sum parameter often including many products which undergo manifold reactions. One parameter is proposed in order to obtain a practical point of comparison for various experiments: the maximum gradient of the degradation curve, which is the slope of the tangent at the inflection point (rQ,0). This has the unit of a zero-order rate constant (mg/kJ instead of mg/min) and therefore appears to be easy to handle. Furthermore this gradient can be roughly considered as the initial rate of the mineralization reaction, because it is preceded by a period of nearly constant TOC level. The parameter rQ,0 is referred to as ‘maximum rate’. In the graphic insert in Figure 5.5, it may be observed that the initial rate is steady from 20–30mg of TOC per liter. At this concentration, saturation occurs and the reaction rate becomes constant. Once the optimum initial concentration is known, a model for predicting plant behavior is necessary. This model must allow calculation of the area of solar collectors required for treating water contaminated with different amounts of pesticides. As commented before, the L-H model is not a perfect explanation of the mechanism of the photocatalytic process, but the behavior of the reaction rate versus reactant concentration can very often be adjusted to a mathematical expression with it. In the present case, instead of using the LH model (r = kKC/(1+KC)) directly, the use of an alternative model is preferred for fitting experimental data in large solar photocatalytic plants, using an approximate kinetic solution of the general photocatalytic kinetic system, which has the analytical form of an L-H equation. With these considerations, the rate of TOC disappearance is given by Equation 7 (analogous to the L-H model but without its original significance).

rQ ,0 =

β 1 [TOC ]max β 2 + β 3 [TOC ]max

(7)

The experimental results shown in Figure 5.5 have been used to calculate the constants (`i). By inversion of Equation 7 these constants can be calculated from the intercept and the slope of the line of fit (Equation 8), which is shown in the inset in Figure 5.5.

β β 1 β3 β2 1 = + ; 3 = 1.67mg -1 ◊ kJ; 2 = 5 .07kJ ◊ L-1 β1 rQ ,0 β1 β1 [TOC ]max β1

(8)

Using these values, experimental results and the corresponding lines of fit are shown in Figure 5.6. The lines of fit were drawn with Equation 9, using the constants reported previously.

Ê [TOC ]max ˆ 1 ÏÔ + β3 [TOC ]max - [TOC ] Ì β ln β1 ÓÔ 2 ÁË [TOC ] ˜¯

(

¸

)Ô˝ = Q ˛Ô

UV

(9)

The experimental results agree reasonably well with the model proposed and the constants calculated. This equation allows TOC degradation to be predicted as a function of initial TOC

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

142 S. MALATO et al

U

U

rpublished published

published published L–1

UU

UU

UU U

U

U

U

U

published published

UU

UU

U

UU

UU

Q, kJ L–1 FIGURE 5.6 Application of the proposed kinetic model for mineralization of a pesticide mixture. The inset shows the fit of Equation 9

and available radiation, and the reverse, incident energy on the reactor necessary to reach a specific degree of mineralization. Therefore, useful design equations may be obtained with an L-H type model, in spite of not fitting the heterogeneous photocatalytic reaction mechanism. For now, these equations must be obtained at pilot plant size; however, they will be useful for larger plants if the same type of collector is used.

5.3.2 RADIANT FLUX Since 1990 there has been a clarification of the kind of solar technology which should be involved in detoxification (Dillert et al, 1999a; Alfano et al, 2000; Mukherje and Ray, 1999). The question posed was whether it is necessary to concentrate radiation for photocatalysis and if a non-concentrating collector can be as efficient as concentrating ones. Initially, it was thought that non-concentrating collectors were the ideal alternative and, in fact, the first large pilot plants operated with them. However, their high cost and the fact that they can only operate with direct solar radiation (which implies their location in highly insolated areas) led researchers to consider the alternative of static non-concentrating collectors. The reason for using one-sun systems for water treatment is firmly based on two factors: (1) the high percentage of UV photons in the diffuse component of solar radiation (see Figure 5.7), and (2) the low order dependence of reaction rates on light intensity. As commented before, the first non-concentrating collectors for detoxification were tested a few years after the first tests with parabolic troughs. Early engineering-scale tests with solar systems were based on parabolic troughs with concentration factors in the range of 10X, but immediately a second generation of non-concentrating collectors started being tested on thin films, flat plates and CPCs. Figure 5.8 presents degradation tests comparing a CPC and a two-axis parabolic trough. Both systems were operated in parallel at Plataforma Solar de Almería (Spain), at the same catalyst concentration (200mg L–1 Degussa P-25), at the same dichloroacetic acid initial concentration (5mM) and

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

143

FIGURE 5.7 Average UV direct and UV global (direct + diffuse) irradiance for each month of the year at Plataforma Solar de Almería, Spain (5 years average)

FIGURE 5.8 Overall dichloroacetic acid degradation rate (as TOC disappearance reaction rate) comparison for a concentrating (PTC) and a non-concentrating (CPC) collector system. The performance ratio (CPC/PTC) and the mean value of global UV-radiation during the experimental periods are also shown

performing one test near the 15th of each month during one year. The tests were performed on perfect, sunny days to permit the parabolic trough to operate all day from sunrise to sunset. The ratio encountered between both systems working in analogous conditions for degradation is higher than 5 in favor of the CPC, and this ratio would be even higher taking cloudy periods into account.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

144 S. MALATO et al

It has been experimentally shown that above a certain flux of UV photons, the reaction rate changes from one to a half-order dependence on intensity. Most authors believe that the transition from r = f (I1.0) to r = f (I0.5) occurs due to the excess of photogenerated species (e–, h+ and •OH). A very simple explanation could be the following (based on the first stages of the process). The first stages considered are: (i) formation of electron/hole pairs (Equation 10), (ii) recombination of the pairs (Equation 11), and (iii) oxidation of a reactant R (Equation 12). (10) (11)

(12)

From these reactions, the concentration of holes is:

(13)

where I is the intensity of incident radiation. If it is considered that [e] 5 [h], then in stationary state: (14)

When I is very high, a large number of holes and electrons are generated and therefore kR[h]2 ›› ko[h]R: (15) As the reaction rate depends on the amount of hydroxyl radicals present, and these are generated in the holes, then r_0.5 when I is high. Under these conditions, the quantum yield diminishes because of the high rate of recombination of e–/h+ pairs formed. In the same manner, when I is small, the inverse is true, kR[h]2 ‹‹ ko[h]R: (16) At higher radiation intensities, another transition from r = f (I0.5) to r = f (I0) is produced. At this moment, mass transfer limits the photocatalytic reaction, therefore the rate is constant although the radiation increases. There may be several factors that limit the mass transfer, such as the lack of electron scavengers (O2), organic molecules in the proximity of the TiO2 surface and/or excess of products occupying active centers of the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

145

catalyst. These phenomena appear more frequently when working with a supported catalyst (Pozzo et al, 1997) and/or at a low agitation level. This implies a low catalyst surface in contact with the liquid and smaller turbulence, which diminishes the contact of reactants with the catalyst and the diffusion of products from the proximity of the catalyst to the liquid. These effects may be appreciably attenuated if some reagent that reduces the importance of the electron/hole recombination is added. When the electrons are trapped, recombination of e–/h+ is impeded. Either way, addition of oxidants can improve the efficiency of the process at high illumination intensities. Moreover, this type of compound can increase the quantum yield even at low irradiation levels due to their strong oxidizing character. The use of inorganic peroxides has been demonstrated to remarkably enhance the rate of degradation of different organic contaminants because they trap the photogenerated electrons more efficiently than O2 (Chen et al, 1998; Doong and Chang, 1997; Malato et al, 1998 and 2000a; Poulios et al, 1998). Hydrogen peroxide is the obvious candidate and it has been tested with a large number of compounds. Also it is a very commonly used chemical and so is inexpensive. Being an electron acceptor, hydrogen peroxide can be a beneficial oxidizing agent because it can react with conduction band electrons to generate hydroxyl radicals which are required for the photo-mineralization of organic pollutants (Equation 17): (17) The effect depends on H2O2 concentration, generally showing in the optimum range, but that strength usually depends on the H2O2/contaminant molar ratio. An optimal molar ratio of between 10 and 100 has been found by different authors (Akmehmet and Ynel., 1996; Bellobono et al, 1994; Harada et al, 1990; Hofstadler et al, 1994; Poulios et al, 1998). Inhibition could be explained in terms of TiO2 surface modification by H2O2 adsorption, scavenging of photo-produced holes and reaction with hydroxyl radicals (Equations 18 and 19). This produces additional problems when working with complex mixtures, because the molar concentration is usually different for each compound in the mixture. Moreover, for the treatment of real wastewater, the concentration of each product is unpredictable. (18) (19) Peroxydisulfate can be a beneficial oxidizing agent in photocatalytical detoxification because SO4•– is formed from the oxidant compound by reaction with the semiconductor – ). The peroxydisulfate ion accepts an electron and photogenerated electrons (eCB dissociates (Equation 20). The sulfate radical proceeds through the reactions shown in Equations 21 and 22. In addition, the strongly oxidizing SO4•– (Eo = 2.6V) can directly participate in the degradation processes. The improvement in the mineralization reaction rate, the relatively low consumption of this reactant and the apparent non-relation

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

146 S. MALATO et al

between molar ratio S2O82–/contaminant and reaction rate shows peroxydisulfate as a good choice. It must be mentioned here that in many highly toxic wastewaters where degradation of organic pollutants is the major concern, the increase of salt content by addition of an inorganic anion to enhance the organic degradation rate may be justified. (20) (21) (22) Other oxidants have been used in photocatalysis for reducing solar (or artificial) UV exposure time: ClO3–, BrO3–, IO4– and HSO5–. Nevertheless, these additives are very expensive compared to hydrogen peroxide and peroxydisulfate, and their application would dramatically increase treatment cost. Even more importantly, they do not dissociate into harmless products (Br– and I–), and hundreds of milligrams per liter of these anions are undesirable in water.

5.3.3 QUANTUM YIELD In photochemistry, quantum yield is used to evaluate the results obtained and compare different experimental conditions. The quantum yield of a photochemical reaction is defined with regard to the number of reacting molecules and the number of photons absorbed (\ = 6n Na–1). A photocatalytic heterogeneous system is made up of a suspended solid (TiO2), a gas (O2) in bubbles and/or dissolved, and an aqueous solution of a multitude of compounds (initial substrate, intermediates, H+, anions). It is very difficult to determine the amount of photons absorbed by the catalyst from the behavior of the radiation incident on a suspension such as this. In order to calculate this, if so desired, one would have to: 1) evaluate the light absorption of a very complex mixture, which, moreover, changes its composition throughout the reaction, 2) from this basis, determine the photon flux that arrives at each particle of the catalyst to photo-activate it, and 3) estimate the photons absorbed and dispersed. And it seems that this, for the moment, is a difficult undertaking (Serpone et al, 1996). Remember that in heterogeneous catalysis the reaction rate is usually expressed as a function of the grams of catalyst. In photocatalysis, it should include the number of active centers as well as the surface area of the catalyst. But as a consequence of the above comments, the number of active centers is unknown and the surface of the catalyst exposed to light is undetermined. This is therefore simplified by considering only the radiation of a certain wavelength incident inside the reactor for the calculation of Na. The value obtained from this is called the estimated quantum yield: \E. No distinction is made between the photons corresponding to each wavelength, assuming that all of them have the same effect on the surface of the catalyst. In all cases, this simplification is accepted as valid by a multitude of authors and widely used in the literature. Consequently, the reported ‘quantum yields’ have sometimes been reported as lower limits not allowing for scattered light. A simple means of assessing process efficiencies for equal absorption of photons is therefore

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

147

desirable in heterogeneous photocatalysis. The initial photo-conversion of phenol has been proposed as the standard process and Degussa P-25 titania as the standard photocatalyst (Serpone and Emeline, 2002). The choice of phenol was dictated by the recognition that the molecular structure of phenol is present in many organic pollutants and, like many of them, is essentially degraded by oxidation rather than reduction.

(23)

In Equation 23, cr is called relative photonic efficiency (Serpone and Emeline, 2002). When the reaction rate for the test substances and phenol (secondary actinometer) are obtained under identical experimental conditions, there is no need to measure the photon flux. The use of relative photonic efficiency renders comparison of process efficiencies between studies carried out in different laboratories or pilot plants possible because cr is basically independent of the fundamental photocatalysis parameters (light intensity, reactor geometry and TiO2 concentration for a given catalyst). However, it depends on the initial concentration of substrate and on temperature. In any case, based on initial rates of degradation, cr illustrates only one aspect of photo-degradation and is also useful for comparing different photocatalyst materials for water treatment purposes. The mineralization rate (measured by TOC analysis) is also included because efficiencies based on the disappearance of organic carbon (cr ,TOC, Equation 24) provide more practical information:

r, TOC

=

rateof disapearanceof substrateTOC rateof disappearance of TOC from phenol

(24)

5.4 FACTORS AFFECTING SOLAR PHOTOCATALYSIS 5.4.1 INFLUENCE OF OXYGEN In semiconductor photocatalysis for water purification, the pollutants are usually organic and, therefore, the overall process can be summarized by Equation 25. Given the reaction stoichiometry of this equation, there is no photo-mineralization unless O2 is present. The literature provides a consensus regarding the influence of oxygen (Herrmann, 1999). It is necessary for complete mineralization and does not seem to be competitive with other reactives during the adsorption on TiO2 since the places where oxidation takes place are different from those of reduction. The O2 avoids the recombination of e–/h+ and photoactivated oxygen (O2•–) also reacts directly: (25)

The concentration of oxygen also affects the reaction rate, which is faster when the partial pressure of oxygen (pO2) in the atmosphere in contact with the water increases. In any case, it seems that the difference between using air (pO2 = 0.21atm) or pure oxygen

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

148 S. MALATO et al

(pO2 = 1atm) is not drastic. In an industrial plant it would be purely a matter of economy of design. Injection of pure O2 becomes necessary in once-through plants at low flow rates. At high flow rates or with recirculation, the addition of oxygen is not always necessary since the illumination time per pass is short. The water again recovers the oxygen consumed when it reaches the tank (open to the atmosphere and stirred).

5.4.2 INFLUENCE OF CATALYST CONCENTRATION Whether in static, slurry or dynamic flow photo-reactors, the initial reaction rates were found to be directly proportional to catalyst mass. This indicates a truly heterogeneous catalytic regime. However, above a certain value, the reaction rate levels off and becomes independent of catalyst mass. This limit depends on the geometry and working conditions of the photo-reactor and is for a specific amount of TiO2 in which all of the particles (in other words all the surface) exposed are totally illuminated. When catalyst concentration is very high, after traveling a certain distance on an optical path, turbidity impedes further penetration of light in the reactor. In any given application, this optimum catalyst mass must be found in order to avoid excess catalyst and ensure total absorption of efficient photons. There are a number of studies in the literature discussing the influence of catalyst concentration on process efficiency (Cassano and Alfano, 2000). The results are very different, but from all of them it may be deduced that radiation incident on the reactor and path length inside the reactor are fundamental in determining the optimum catalyst concentration. If the lamp is outside, but the path length is several centimeters long (similar to a reactor illuminated by solar radiation), the appropriate catalyst concentration is several hundreds of milligrams per liter. In this case, it is very clear that the optimum rate is attained at lower catalyst concentrations when the photo-reactor diameter is increased. As one important factor related to the photo-reactor design is its diameter, it seems obvious that a uniform flow must be maintained at all times in the reactor, since irregular flows cause non-uniform residence times that can lower efficiency compared to the ideal (Blake et al, 1997). In the case of the heterogeneous process with TiO2 in suspension, sedimentation and deposition of the catalyst along the hydraulic circuit must also be avoided, so turbulent flow in the reactor is guaranteed (Malato et al, 2003b). Turbulent flow makes pressure loss an important parameter that can condition design, especially in the case of an industrial plant with long reactor tube lengths. For these reasons diameters of less than 20–25mm are not feasible. On the other hand, diameters over 50–60mm are considered impractical. Furthermore, every photo-reactor design must guarantee that all the useful incoming photons are used and do not escape without having intercepted a particle in the reactor. Although TiO2 suspensions absorb solar photons with less than 390–400nm wavelengths, there is also a strong light scattering effect due to particles. Both effects must be considered in determining the optimum catalyst load as a function of light path length in the photo-reactor. After many experiments with different photoreactors, the optimum TiO2 concentration obtained with sunlight has been found to be several hundreds of milligrams per liter (Dillert et al, 1999a; Giménez et al, 1999; Goslich et al, 1997a; Guillard et al, 1999; Minero et al, 1996a) and the ideal diameter of the photoreactor for use with sunlight must be in the range of 25 to 50mm.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

149

Fe3+ (related species and organic complexes) absorbs solar photons as a function of its absorptivity (Mazellier et al, 1997). This effect must be considered when determining the optimum load as a function of light path length in the photo-reactor. These calculations must be experimentally demonstrated because they are strongly affected by different complexes formed by Iron (III) during the photo-Fenton process. The optimum concentrations of 0.2–0.5mM were obtained after many experiments with different photoreactors under sunlight (Fallmann et al, 1999; Gernjak et al, 2003; Malato et al, 2002b).

5.4.3 TEMPERATURE INFLUENCE Because of photonic activation, photocatalytic systems do not require heating and operate at room temperature. The true activation energy is nil, while the apparent activation energy is often very low (a few kJ/mol) in the medium temperature range (20–80ϒC). However, at very low temperatures (–40–0ϒC), activity decreases and activation energy becomes positive. The decrease in temperature favors reactants’ adsorption, which is a spontaneous exothermic phenomenon, but also favors adsorption of the final reaction products, desorption of which tends to be the rate-limiting step. By contrast, at ‘high’ temperatures (>70–80ϒC) for various types of photocatalytic reactions, the activity decreases and the apparent activation energy becomes negative. When temperature increases above 80ϒC, nearing the boiling point of water, the exothermic adsorption of reactants is disfavored and this tends to become the rate-limiting step. In addition to these mechanical effects, other consequences of plant engineering must be considered. At high temperatures the oxygen concentration in water decreases. Also materials used at high temperatures must be stable at those temperatures. Consequently, the optimum temperature is generally between 20 and 80ϒC. The absence of a need for heating is attractive for photocatalytic reactions carried out in aqueous media and in particular for environmental purposes (photocatalytic water purification). In the case of photo-Fenton treatment, reaction rates increase with temperature (Lee and Yoon, 2004; Sagawe et al, 2001). As a consequence this process seems especially interesting for wastewater streams leaving an industrial process at elevated temperatures. Additionally, the application of heat exchangers or insulation can be an attractive plant design adaptation (Sagawe et al, 2001).

5.5 SOLAR UV PHOTOCATALYTIC DEGRADATION OF CONTAMINANTS Research activities devoted to environmental protection have developed relatively quickly as a consequence of the special attention paid to the environment by international social, political and legislative authorities, leading in some cases to very severe regulations (Angelakis et al, 1999). The fulfillment of strict quality standards is especially called for in the case of those toxic substances which affect the biological sphere and prevent activation of biological degradation processes. These non-biodegradable contaminants are an accumulative problem with unpredictable consequences for the mid-term future (Hayo, 1996). The destruction of toxic pollutants such as biologically recalcitrant compounds must be handled by non-biological technologies. Photocatalysis aims at the mineralization of the contaminants into carbon dioxide, water and inorganics. Up to now, practical applications of solar technologies have been studied and developed most

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

150 S. MALATO et al

intensively for heterogeneous TiO2 photocatalysis and homogeneous photo-Fenton. In this context, treatment of industrial wastewater seems to be one of the most promising fields of application of solar detoxification, despite inefficient production of hydroxyl radicals and slow kinetics which may limit economic feasibility. There is no general rule here at all – each case is completely different (Malato et al, 2003a). Consequently, preliminary research is always required to assess potential pollutant treatments and optimize the best option for any specific problem on nearly a case-by-case basis. Solar photocatalytic degradation technology might be feasible for the treatment of wastewater containing hazardous contaminants at medium or low pollutant concentrations when treatment plants are impossible. The technology is dependent on the energy flux, as is the associated investment contingent on the collector surface. Reasonable orders of magnitude for inflow into typical treatment plants would be in the range of from several dozen up to a few hundred m3 per day (Bekbölet et al, 1996; Freudenhammer et al, 1997; Goswami et al, 1997; Malato et al, 1996, 2000b; Funken et al, 2001). In general, the types of compounds which have been degraded include alkanes, haloalkanes, aliphatic alcohols, carboxylic acids, alkenes, aromatics, haloaromatics, polymers, surfactants, herbicides, pesticides and dyes. Equation 26 generally holds true for an organic compound of general formula CnHmOp:

(26) In the case of organic compounds containing halogens, Equation 27 shows how the corresponding halide is formed:

(27)

Under photocatalytic oxidative conditions, sulfur is recovered as sulfate in sulfurcontaining compounds according to Equation 28:

(28)

In photo-degradation, transformation of the parent organic compound is desirable in order to eliminate its toxicity and persistence, but the principal objective is to mineralize all pollutants. The effectiveness of degradation is not demonstrated only because the entire initial compound is decomposed. Furthermore, the stoichiometry proposed for the general reactions (Equations 26–28) has to be demonstrated in each case by a correct mass balance. Reactants and products might be lost (evaporation, adsorption on reactor components), which introduces uncertainty in the results. The mineralization rate is determined by monitoring inorganic compounds, such as CO2, Cl–, SO42–, NO3– and PO43–. When organics decompose, a stoichiometric increase in the concentration of inorganic anions is produced in the water treated and there is very often an increase in the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

151

concentration of hydrogen ions (decrease in pH). For this reason, the analysis of these two products of the reaction is of interest for the final mass balance. However, the decrease in pH is not a very reliable parameter of this balance, except in some cases, because it is influenced by other processes which take place in the medium: the effect of the TiO2 suspension, the formation of CO2 and intermediates. The oxidation of carbon atoms into CO2 is relatively easy. It is, however, in general, markedly slower than the de-aromatization of the molecule. Until now, the absence of total mineralization has been observed only in the case of s-triazine herbicides, for which the final product obtained was essentially 1,2,5-triazine-2,4,6, trihydroxy (cyanuric acid), which is, fortunately, not toxic (Minero et al, 1996b). This is due to the strong stability of the triazine nucleus, which resists most methods of oxidation. For chlorinated molecules, Cl– ions are easily released into the solution. Nitrogen-containing molecules are mineralized into NH4+ and mostly NO3–. Ammonium ions are relatively stable and the proportion depends mainly on the initial oxidation degree of nitrogen and on the irradiation time. Organo-phosphorous pesticides produce phosphate ions. However, phosphate ions in the pH range used remain adsorbed on TiO2. This strong adsorption partially inhibits the reaction rate, though it remains acceptable. Until now, the analyses of aliphatic fragments resulting from the degradation of the aromatic ring have only revealed formate and acetate ions. Other aliphatics (presumably acids, diacids and hydroxylated compounds) are very difficult to separate from water and to analyze. Formate and acetate ions are relatively stable, which in part explains why total mineralization takes much longer than de-aromatization (Franch et al, 2002). Information concerning degradation of contaminants at pilot plant-scale using solar collectors is available in several recently published reviews (Alfano et al, 2000; Bahnemann et al, 2004; Dillert et al, 1999a; Gogate and Pandit, 2004a; Konstantinou and Albanis, 2003; Malato et al, 2002a; Malato and Agüera, 2003; Pera-Titus et al, 2004).

5.6 EVALUATION OF SOLAR UV RADIATION Solar ultraviolet radiation is an essential parameter for the correct evaluation of data obtained during photocatalytic experiments in a solar water decontamination pilot plant. The kinetic constants of photocatalytic processes can be obtained by plotting substrate concentration as a function of three different variables: time, incident radiation inside the reactor and photonic flux absorbed by the catalyst. The complexity of obtaining these constants, as well as their applicability, varies depending on the procedure. When the photonic flux absorbed by the catalyst is used as an independent variable, extrapolation of the results is better. However, many parameters must be known for determining photonic flux absorbed by the catalyst (including incident photons passing through the reactor without interacting with the catalyst, directions of light scattering and size distribution of the TiO2 particles suspended in the liquid), making it almost impracticable in large size solar photo-reactors (Cassano and Alfano, 2000; Curcó et al, 1996; Sagawe et al, 2003a and 2003b). Use of the experimental time as the calculation unit could give rise to misinterpretation of results, because the reactor consists of illuminated and nonilluminated elements. Large experimental reactors must be as versatile as possible and

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

152 S. MALATO et al

require a great deal of instrumentation, thus substantially increasing the non-illuminated volume. With use of residence time, in other words the time the water was irradiated, the conclusions would be erroneous too. This is because when time is the independent variable, the differences in the incident radiation in the reactor during an experiment are not taken into account. Goslich et al (1997b) have proposed a very useful mathematical approach for the treatment of such data obtained in real solar experiments using a relationship between experimental time (t), plant volume (Vt), collector surface (Ar) and the radiant power density (UVG = WUVm–2) measured by a specific UV radiometer. They suggested using a so-called ‘mass area ratio’ Si of the organic pollutant i. This is the overall mass of pollutant per surface element of the reactor: Si = CiVtAr–1. As radiation data are collected continuously, ——– it is very easy to calculate the average incident radiation on the collector surface (UVG,n) for each period of t, and therefore calculate the accumulated energy during this period, Euv,ges. It is reasonable to divide the ‘mass area ratio’ of the pollutant by the accumulated energy, yielding a value for an entity called ‘efficiency’: Eff = Si (Euv,ges)–1. Its dimension is mass per unit energy. Instead of using Euv,ges and Eff, we have proposed using QUV and rQ (Malato et al, 2000c). QUV fits better with the term radiant energy (Q) recommended by IUPAC (Braslavsky and Houk, 1988) and makes it possible to use C (reactant concentration) in the plots, which is easier to understand than Si. Besides, rQ is better than Eff to define a parameter which is not dimensionless. At the same time, we think that symbol rQ (-dC/ dQUV) is more consistent with r (dC/dt), which is commonly used in the scientific community. Consequently, the amount of energy collected by the reactor (per unit of volume) from the beginning of the treatment until each sample is collected may be found by:

QUV,n = QUV,n -1 + Dt n UV G,n

Ar Vt

(29)

Dt n = tn - t n -1 where tn is the experimental time of each sample and QUV,n is the accumulated energy (per unit of volume, kJ L–1) incident on the reactor for each sample taken during the experiment. This equation comprises all those proposed by Goslich et al (1997a), substituting the parameter used by these authors (Euv,ges) by QUV,n. Figure 5.9 shows the improvement obtained using this equation to calculate the reaction rate in a two-day photocatalytic degradation experiment with a model compound. Obviously, UV power changes during the day, and clouds during the first day make this variation still more noticeable, but with Equation 29, the data for both days can still be combined and compared with other photocatalytic experiments. Consequently, with QUV, the initial reaction rate (rQ,0 = -dC/dQUV; C1(QUV) = C0 - kapQUV) is expressed in terms of pesticide degraded per kJ of UV incident on the collector surface. A similar treatment could be done for first order kinetics. As Goslich et al have remarked, if rQ (in their case Eff) is known, collector efficiency is already included through the use of incident surface radiation, since different rQ, with the same substance and different solar collectors, means collector efficiency is different.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

U 2014

2014 2014

U

U

U

U

U published UU

published published

U U

UU

C, mg.L–1

U

UVG, W.m–2

153

U

UUU

U 2014 2014

2014

U

U

U

U

U

U

C, mg.L–1

U

U

QUV, kJ.L–1 FIGURE 5.9 Plots of pesticide concentration as a function of experiment time (top) and accumulated energy (bottom). Solar UV power throughout the experiment is also shown

As commented by Wolfrum and Turchi (1992), this procedure for deducing rQ is exact only if the concentration of reactives is constant along the entire reactor (in other words if the system is at steady state). Solar detoxification pilot plants are frequently operated in a recirculating batch mode as depicted in Figure 5.10. In this scheme, the fluid is continuously pumped between a reactor zone and a tank in which no reaction occurs until the desired degradation is achieved (Bahnemann et al, 2004). Since these are dynamic systems, the photo-reactor (see Figure 5.10) outlet concentration (C1) is not exactly the same as the mixed tank outlet concentration (C2). In a large field system, the amount of conversion each time the mixture passes through the reactor is noticeable (Goswami, 1995). As the relatively clean water in the reactor is mixed with the ‘dirty’ water in the batch tank, the water sent from it to the reactor has an increasingly lower concentration. Because the rate usually decreases with concentration, the overall rate in the reactor responds likewise. Thus unless properly accounted for, the presence of the tank will alter the perceived performance of the photo-reactor. Two solutions are available to solve this problem. The first solution is to use a very high flow rate to achieve low conversion each time through. This high flow rate must allow more than 1 per cent conversion per pass (C1(t) 5 C2(t)) to be avoided. We have selected this 1 per cent because it is lower than the error associated with any analytical chemical method applied for Ci analysis. The second solution is that the concentration in the reactor outlet at time t, C1(t) is determined from the inlet concentration (which does not change with time) and the reaction kinetics. However, because the system is a transient process, the normal steady-state plug flow reactor

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

154 S. MALATO et al

equation cannot be used to model the photo-reactor. Because the inlet concentration changes with time, C2(t) is defined by what went into the reactor one residence time prior to t, C2(t-o), and the kinetics. The batch tank could still be modeled as a well-mixed tank. Solving these equations is more difficult than the low-conversion-per-pass case and a numerical routine is required to fit the data from the batch test. This routine also takes into account the volume of the piping between the reactor outlet and the batch tank. This methodology can be used to numerically solve for the concentration profile in any batch process, even though this method is only necessary when the recirculation flow is not high enough. As also pointed out by Mehos et al (1992), one must account for the mixing in the dark tank only if substantial conversion per pass is obtained (C2 > C1).

FIGURE 5.10 Schematic of the pilot plant batch operation

5.7 INSTALLED SOLAR PHOTOCATALYTIC TREATMENT PLANTS Despite its obvious potential for the detoxification of polluted water, there has been very little commercial or industrial use of photocatalysis as a technology to date. Several years ago, and according to a review by Goswami (1997), the published literature showed only two engineering-scale demonstrations for groundwater treatment in the US and one industrial wastewater treatment in Spain (at Plataforma Solar de Almería). But recently more installations have been erected, mainly based on non-concentrating collectors. In laboratory and bench-scale experiments Dillert et al have treated biologically pretreated industrial wastewaters from the Volkswagen AG factories in Wolfsburg (Germany) and Taubaté (Brazil). The results of the experiments, which were performed using the DSSR mentioned in section 5.2.2, were so promising that a pilot plant was installed in the Wolfsburg factory during the summer of 1998 (Dillert et al, 1999b). The flowsheet of a more recent version of this pilot plant, which was installed in 2000, has been recently published by Bockelmann et al (2004). This pilot plant, which is operated in a recycle batch mode, consists of 12 double-skin sheet photo-reactors (manufactured by Solacryl) with a total irradiated area of 27.6m2. These reactors are connected by a recirculation pipe with a tank (approx. 1.1m3 volume). The water (500L) coming from the biological treatment plant is pumped into the tank and mixed with a catalyst slurry. The solar photocatalytic treatment starts after the tank has been filled. The suspension (total volume approx. 1m3) recycles between the tank and the DSSRs for 8 to 11 hours during the daytime. The flow rate can be varied between 830 and 1300L per hour in the reactor sheets. After the desired treatment time the suspension is pumped out of the reactors into the tank. The photocatalyst is allowed to settle during the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

155

night. After this sedimentation period the supernatant liquid (500L) is pumped out of the tank and the treatment cycle can be started again by filling the tank with a new batch of wastewater. The initial values of the TOC were determined to vary between 26.9 and 7.3mg L–1. During the solar-catalytic treatment a decrease in the concentration of the organic pollutants, determined as COD (chemical oxygen demand) and TOC, was observed in most cases. More than 50 per cent of the organic pollutants initially present in the mixed water inside the pilot plant could thus be degraded within 8 to 11 hours of illumination. Naturally, the total mass of the degraded contaminants was found to depend on the initial pollutant concentration, the time of illumination and, in particular, on the solar UV energy flux density. In 1997 Freudenhammer et al reported their results from a pilot study using thin film fixed bed reactors (TFFBRs), which was performed in various Mediterranean countries and showed that biologically pretreated textile wastewater can be cleaned by solar photocatalysis with a maximum degradation rate of 3g COD m–2 per hour (Freudenhammer et al, 1997). It was concluded that photocatalysis should be a suitable technology as the final stage of purification of biologically or physically pretreated wastewater in particularly sunrich areas. However, it was also pointed out that there is a strong demand for cheap photocatalysts of higher activity for industrial applications of solar wastewater treatment on a larger scale to be competitive with treatment methods already established on the market. Based on these results, a pilot plant, financed by the European Commission, has been built at the site of a textile factory in Tunisia (Menzel Temime). The TFFBR was chosen because previous studies showed sufficient degradation rates with the selected textile wastewater in combination with its simple, low cost construction and the low energy consumption. However the possibility of using suspended catalysts has also been considered, to integrate results obtained with suspended catalysts showing, in some cases, a higher efficiency than the fixed system,. The pilot plant and the flow chart have been published recently (Bahnemann et al, 2004). Two TFFBRs of width 2.5m and length 10m, corresponding to a total illuminated area of 50m2, were built in concrete and are oriented to the south with an inclination angle of 20ϒ. The reactors can be operated in parallel or as a cascade flow and in a continuous or a recycling mode, depending on the reaction kinetics. The pumps are designed for a maximum flow of 3m3 per hour. The plant is connected to the sewerage system of the textile factory for continuous and long-term experiments. Two sequencing batch reactors (SBRs) and a membrane aeration system are connected to the TFFBR for pre- and post-treatment, each with a total volume of 15m3. An automatic injection system of chemical compounds can be used for pH regulation or other necessities. The plant is designed to operate the two reactors independently. This allows a comparison of different catalysts, fixation techniques, hydrodynamics and effluents/model compounds under identical conditions. The plant can be operated with suspended and fixed catalysts. A sedimentation tank is connected to the reactors (via a storage tank) to separate and recycle the catalyst. Instrumentation and control instruments are installed and connected to a computer system to allow automatic operation: pumps and valves control flow rates, tank levels, UV-A intensity, dissolved oxygen, pH and conductivity. Experiments are currently being conducted to test the operation conditions of this newly installed facility (Bousselmi et al, 2004).

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

156 S. MALATO et al

Under the Solar Detoxification Technology for the Treatment of Industrial NonBiodegradable Persistent Chlorinated Water Contaminants (SOLARDETOX) project, a consortium coordinated by Plataforma Solar de Almería, Spain, has been formed in Europe for the development and marketing of solar detoxification treatments for recalcitrant water contaminants. The main project objective was the technical and economic optimization of real solar detoxification applications. The main innovations were in the engineering, there being no specific developments in the solar detoxification technology itself, since it was derived from the already existing solar thermal technology with only minor modifications. This general objective was divided into several parts: ● assessment of the performance and best working conditions for newly developed

● ● ● ●

catalyst powders compared with the efficiencies of commercial products (Degussa P25) in the treatment of chlorinated solvent compounds; assessment of degradation mechanisms, photonic efficiencies, formation of sideproducts and kinetic modeling of the process from degradation data; development of a reactor with highly UV-transmissive glass in the solar UV range; improvement of the solar collector with a highly efficient UV-reflective surface; and demonstration of the technical and economic feasibility of achieving a well-defined system under real conditions; optimization of the design and standardization of the components (collectors, tubes, catalyst, control and operating systems) in order to minimize costs of production; erection and operation through the formation of a consolidated industrial and institutional consortium.

The main goal (financed by the EC-DGXII (European Commission Directorate General XII) through the Brite Euram III Program, 1997–2000) was to develop a commercial nonconcentrating solar detoxification system using compound parabolic collector (CPC) technology, with a concentration ratio of one. The beauty of the solar CPC system (as commented in section 5.2.3) is its intrinsic simplicity while remaining cost-effective and easy to use – it requires low capital investment. Field demonstration was intended to identify any pre- or post-processing requirements, potential operating problems, and capital and operating costs. Based on accumulated experience in pilot plant design, construction and testing (Blanco et al, 2000), a full-size demonstration plant was erected at the facilities of HIDROCEN (Madrid, Spain). This plant was designed to treat 1m3 of water contaminated with 100m2 of collector aperture area (see Figure 5.11). The main plant characteristics are: i) 2 rows of 21 collectors each; ii) total collector (tilted 40°, local latitude) aperture area = 98m2; iii) total loop volume = 675L; iv) total plant volume = 975L; v) 200mg/L TiO2 slurry; and vi) 1.5 x 1.5m collectors with sixteen 29.2mm internal diameter (ID) tubes. The CPC reflector is made of a highly reflective anodized aluminum sheet held by a galvanized frame supporting 16 parallel 1.5m-long tubes, each with an appropriate connector for the adjacent tube. A complete module is formed by a series of collectors connected in a row. The final prototype plant consists of E–W oriented parallel rows of 21 collectors each. The structure was slightly tilted (1 per cent) in the same direction to drain rain water and avoid its accumulation in the CPC troughs. Final system design is completely modular, with the collectors attached in series using HDPE (high density

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

Photocatalytic Detoxification of Water with Solar Energy

157

polyethylene) quick connections between glass reactor tube absorbers. Water flows simultaneously through all parallel tubes and there is no limit to the number of collector components modules may have. Water enters and exits rows through two manifolds at opposite ends. As this plant is to be a demonstration of what a commercial plant would be like, operation is fully automatic and maintenance requirements are minimal. Among the electronic control devices the most innovative system is a solar UV-A sensor that integrates solar UV with time during the treatment. This sensor is connected to a programmable logic controller (PLC) and once sufficient energy (see Equation 29) for finishing the treatment has been achieved (based on preliminary testing for plant design according to the specific contaminated water to be treated), the PLC stops the main pump and advises the operator that the treatment has been completed. The PLC also receives other data signals (flow rate, tank level, temperature, etc) for controlling system pumps and valves.

20142014 20142014 20142014 2014

0.1m3 2014

2014 2014 0.2m3

0.3m

3

2014 2014 2014

2014 20142014 2014 2014 2014 2014 2m3 2014

2m3 2014

AND AND AND AND

2014 2014 2014

AND AND AND AND

2014 2014 2014

20142014 2014 2014 2014 2014

2014 2014 2014 2014 100m2 800L 2014

~5

0m

FIGURE 5.11 Schematic diagram of solar detoxification demonstration plant constructed in SOLARDETOX project at HIDROCEN, Madrid

Since late 1999, this plant has been fully functional and its operation is easily described. Wastewater enters the system from a 2m3 wastewater storage tank while detoxified water from a previous run returns to the catalyst sedimentation tank. When the system is completely full (0.8m3), a pump recirculates through the collectors and a small tank (0.1m3). The concentrated TiO2 slurry and the air necessary for the reaction are injected into the circuit. Once the water is detoxified, the entire volume passes to the

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 130–168

158 S. MALATO et al

sedimentation tank and the system is filled with more wastewater for another batch. Meanwhile, the detoxified water and the TiO2 in the sedimentation tank undergo pH adjustment to provoke fast sedimentation of the catalyst (Fernández–Ibáñez et al, 2003). The concentrated catalyst slurry is transferred from the bottom of the tank to another smaller tank from which the catalyst enters the photo-reactor. The supernatant is removed through an outlet in the side almost at the bottom of the sedimentation tank and enters another tank (0.3m3) where the small fraction of the initial catalyst contained in the supernatant ( 20 per cent, possess limited corrosion and photo-corrosion resistance, resulting in short lifespans. Reports on the development of PEs (Onishi et al, 1975; Mavroides et al, 1975; Memming, 1980; Soliman and Seguin, 1981) have concentrated on the effects of selected individual issues on the ECE. Progress in this area requires the identification and optimization of all of the materials properties that have an impact on the ECE, including defect chemistry and related electronic structure, semiconducting properties, composition, nonstoichiometry, and microstructure (Bak et al, 2002). Most photo-electrodes that exhibit sustainable performance are fabricated from oxide materials (Bak et al, 2002). However, another group of materials used for photo-electrodes are valence semiconductors, such as GaAs (El Zayat et al, 1998; Licht et al, 2000). Extensive analysis of the property requirements of photo-electrodes (Bak et al, 2002; Nowotny et al, 2005) indicates that the following properties are essential (the physical meanings of these properties have been outlined elsewhere): ● Band gap: The band gap (the forbidden energy gap) is one of the most critical

properties of PEs. In the case of water photolysis, the band gap should remain between 1.23eV (theoretical lower limit) and ~2eV (practical upper limit). The electronic structure and related band gap of semiconductors based on nonstoichiometric compounds, such as TiO2-x, are very sensitive to defect disorder

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

Solar-Hydrogen: A Solid-State Chemistry Perspective

●

●

●

●

●

●

179

(Khan et al, 2002). On the other hand, in the case of single-crystal Si, used for photovoltaic (PV) cells, the band gap is fixed at 1.1eV and it is a material property. Therefore, the electronic structure of nonstoichiometric compounds may be modified through the control of the defect chemistry. Electrical resistance: Efficient charge transfer within the circuit of a photo-electrochemical cell requires the electrical resistance to be at its minimal level. Consequently, this requires the resistance of all elements of the circuit, including the photo-electrode, to be minimized. The resistivity of the photo-electrode can be minimized through the imposition of suitable defect disorder. Concurrently, the materials engineering must be undertaken so as to retain or achieve the optimal band gap. Schottky barrier: The Schottky-type barrier is needed for effective charge separation (prevention of recombination of the electron-hole pair). The consequence of this separation is that electrons are transported into the n-type solid while electron holes are transported toward the photo-electrode/electrolyte interface, resulting in the formation of a p-type surface layer (Bak et al, 2002). In the case of nonstoichiometric compounds used as photo-electrodes, the potential barrier may be formed either by the imposition of a defect concentration gradient during processing or through the effect of segregation leading to enrichment of the surface layer. These have the effect of imposing a surface Fermi energy at a level different from that of the bulk phase. Microstructure: The properties of polycrystalline solids are well known to be very sensitive to microstructure. The effects of the microstructure on properties may be considered in terms of, inter alia, grain size and the concentration of grain boundaries. The effect of the surface microstructure may be considered in terms of the local properties of the linear defects formed at the intersections between the grain boundaries and the external surface. So far, little is known of the effects of these defects on the ECE. Further, the effect of the grain size (bulk/surface) may be considered in terms of the interfacial energy and its impact on the local defect chemistry and related charge transfer across the photo-electrode/electrolyte interface. Flat band potential: The flat band potential should be higher than the redox potential of the H+/H2 couple. This potential may be engineered by the imposition of specific surface-to-bulk composition gradient and related potential gradient. The flat band potential was described by Bak et al (2002). Helmholtz potential barrier: The Helmholtz potential barrier, which is formed at the interface between the photo-electrode and the electrolyte, plays an important role in retarding/accelerating the charge transfer within the electrochemical chain shown in Figure 6.5 (Bak et al, 2002). Therefore its value is determined by the surface compositions of the photo-electrode and the composition of the electrolyte. Nonstoichiometry and related defect disorder: Although it is well known that nonstoichiometry and point defects have a substantial impact on material properties, so far little is known on the effect of defect disorder on electroactivity and photo-sensitivity. Awareness is growing that properties of photo-electrodes based on

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

180

J. NOWOTNY, T. BAK, L. R. SHEPPARD AND C. C. SORRELL

nonstoichiometric compounds, such as TiO2-x, can be closely related to defect disorder and the related electronic structure. Consequently, the desired properties of the photo-electrodes may be achieved through the modification of the defect disorder. The recent report by Khan (Khan et al, 2002) confirms that the defect disorder has a significant impact on the electronic structure and related photo-sensitivity of TiO2. ● Corrosion resistance: Most photo-active materials are not resistant to corrosion in aqueous environments, which results in the degradation of their properties. However, little is known of effective corrosion inhibition in these materials. Consequently, naturally occurring and stable oxides are the most promising materials for photo-electrodes owing to their ability to meet this criterion. Since TiO2-x is among the most stable of all oxides in aqueous environments at ~pH = 7 (Fujishima et al, 1999), it is the leading candidate. ● Segregation-induced surface barrier: Segregation results in the formation of concentration gradients in the surface layer of photo-electrodes that may affect the charge transfer across the electrode/electrolyte interface. So far, knowledge about segregation and its impact on the function of the photo-electrode is still limited. However, the segregation-induced potential barrier in oxide materials may be of the order of 1V (Nowotny, 1991), so its effect on charge transfer may be crucial. The list of the materials properties that are essential to achieve high-performance photoelectrodes is longer than that shown above. Some properties, such as band gap, electrical resistance, flat band potential and microstructure are probably the most frequently used in studies of photo-electrodes. The most extensive outline of these properties are reported elsewhere (Nowotny et al, 2005). However, because there is no simple relationship between these properties and their impact on the ECE, a multi-variant approach is required in the development of photo-electrode materials rather than monovariant investigations. In summary, there is a considerable experimental and theoretical body of evidence that indicates that TiO2-x is the best candidate for highefficiency photo-electrode PEs. However, there is a need to optimize the electronic structure in order to reduce the effective band gap required for ionization. At the same time there is a need to address other properties that have an impact on energy losses, such as the energy losses related to recombination and the charge transfer. Subsequent sections outline the issues related to the electronic structure of TiO2-x, including the definition of basic quantities describing electronic structure and the experimental approaches of their determination.

6.5 ELECTRONIC STRUCTURE 6.5.1 DEFINITION OF BASIC TERMS The electronic structure of semiconducting solids may be described in terms of diagrams of the density of states and represented in the form of plots of the number of electron energy levels as a function of energy (Kofstad, 1972; Green, 1986).

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

Solar-Hydrogen: A Solid-State Chemistry Perspective

181

The key energy quantities describing the electronic structure of solids and having an impact on the photo-sensitivity include: ● ● ● ●

energy of the bottom of the conduction band, EC; energy of the top of the valence band, EV; energy levels of the donors and acceptors within the band gap, ED and EA; and Fermi energy, EF.

It is well known that the width of the forbidden gap (Eg = EV – EC) is the key property controlling ECE (Bak et al, 2002; Chandra, 1985; Gerischer, 1977 and 1997; Seraphin, 1979). Another quantity that has a crucial impact on the reactivity of photo-electrodes is the Fermi energy. There is a close relationship between the Fermi energy and the lattice defect disorder of semiconducting oxides. This involves point defects that form either donors, such as oxygen vacancies and cation interstitials, or acceptors, such as cation vacancies. The Fermi energy is determined by the location of the associated energy levels within the band gap and the related density of states. Doping of metal oxides with aliovalent ions results in the formation of either donors or acceptors located within the band gap. The positions of these levels are determined by the incorporation mechanisms of the specific ions into the lattice of the host semiconducting oxides. The densities of states are controlled by the concentration of dopants within their solubility ranges. The mobility of electronic charge carriers within the bands or mid-gap levels is determined by the density of allowed states forming the bands and the donor and acceptor states. Specifically, at high and low density of states, the charge carriers are quasi-free or localized respectively. At low density of states, when these levels are isolated, the charge transport occurs by the hopping mechanism. Increase of the density of states to very high values may, in extreme cases, lead to metallic conduction (Sheppard, 2004). Detailed analysis of the electronic structure of semiconducting solids is beyond the scope of the present work, which is limited to a brief analysis of the following two quantities: 1 band gap; and 2 Fermi energy. These two quantities are selected for consideration for the following reasons: ● They are the most important in the assessment of photo-sensitivity and reactivity at

the water/photo-electrode interface. ● They may be assessed relatively easily by measurement of electrical properties

commonly available in materials science laboratories.

6.5.1.1 Band gap As shown in Figure 6.6, the band gap Eg is the energy difference between the bottom of the conduction band, EC, and the top of the valence band, EV. The band gap width is one

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

182

J. NOWOTNY, T. BAK, L. R. SHEPPARD AND C. C. SORRELL

of the most critical factors that control the properties of intrinsic semiconductivity in general and photo-sensitivity in particular. The band gap is the energy difference through which heat- and photon-induced ionization occur. This ionization results in the formation of quasi-free charge carriers (electrons and electron holes) that take part in the charge transfer within the electrode and its electroactivity (certain density of states is required for the charge carriers to be quasi-free, alternatively they are isolated). In other words, only the photons of the energy equal to or larger than that of the band gap may be absorbed and be available for conversion of the electromagnetic energy into another form of energy, such as electrical and chemical. The critical impact of Eg on photo-sensitivity and ECE has been discussed previously (Bak et al, 2002).

Centre for Materials Research in Energy Conversion, UNSW

E0=0 O¯ Φ1

Φ2

EC

ΔΦS

EF Eg

ΔΦS=-ΔEF

O¯

EC

O¯

EF

O¯ O¯

EV EV

DISTANCE FROM THE SURFACE FIGURE 6.6 Band model of n-type semiconductor without surface charge (left) and involving surface charge (right)

6.5.1.2 Fermi energy Fermi energy (EF) is a specific parameter of the Fermi-Dirac statistics of electrons at which 50 per cent of states are ionized. These statistics describe the distribution of electrons with respect to their energy levels (Kofstad, 1972):

fn ( E ) = ÈÎ1 + exp ( Ei - E F ) / kT ˘˚

-1

(5)

where fn(E) is the probability of occupation of the energy level Ei. The similar expression for fp(E) describes the distribution of electron holes: (6)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

Solar-Hydrogen: A Solid-State Chemistry Perspective

183

The EF may be referenced quantitatively by comparison to the energy of either the bottom of the conduction band or the top of the valence band. The EF level may also be referenced to the energy level where electrons are free of electrostatic interactions, which is outside the solid surface. The energy difference between these two levels is termed the ‘work function’. A schematic representation of the flat band model (free of surface charge) for an n-type semiconductor, showing the EF and the work function  , is shown on the left side of Figure 6.6. In solid-state physics, the EF is used to characterize semiconducting properties of solids with respect to their ability to donate or accept electrons. The EF is also a key quantity in the evaluation of the charge transfer accompanying chemical reactions, such as catalysis. However, in chemistry, chemical equilibria are typically described in terms of the chemical potentials of reactants and reaction products. In this case, the EF is equivalent to the chemical potential of electrons, which describes the electroactivity of the solids and so their ability either to accept or donate electrons during chemical reaction. Knowledge of the value of the chemical potential is essential for quantitative evaluation of the reactivity of solids whenever charge transfer takes place. The reactivity (viz photo-reactivity) of solids involved in photo-induced processes is no exception. Therefore, it is clear that the EF appropriate for photo-reaction of the photoelectrode with water must be optimized in order to achieve the desired photo-reactivity. The semiconducting properties of the photo-electrode and thus the resultant chemical potential are established during processing. Therefore it is useful to monitor the chemical potential of electrons of the solid during processing. The desired chemical potential of electrons can be established by conducting the processing in a controlled manner. The Fermi-Dirac statistics of electrons and electron holes, represented by Equations 5 and 6, can be approximated by the Maxwell-Boltzmann statistics when (Kofstad, 1972): (7) and (8) where EC and EV represent the energies of the bottom of the conduction band and the top of the valence bands respectively, k is the Boltzmann constant, and T is the absolute temperature. Then for a non-degenerated semiconductor there is a simple relationship between the EF and the concentration of electronic charge carriers:

(9)

(10) where n and p denote the concentrations of electrons and electron holes respectively, and Nn and Np denote the densities of states of electrons and electron holes respectively.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

184

J. NOWOTNY, T. BAK, L. R. SHEPPARD AND C. C. SORRELL

Equations 9 and 10 may be used for the determination of the basic electronic structure when the concentrations of the electronic carriers are known (or vice versa). In the case of a flat-band structure, corresponding to semiconductors that are free of surface charge, the value of EF is independent of the distance from the surface. Alternatively, the presence of a surface charge, caused, for example, by the adsorption of charged species, results in a difference in the Fermi energies of the surface and the bulk by the energy component 6 s, as shown on the right side of Figure 6.6. There is a growing body of empirical data that indicate that the EF in the bulk is quite different from that at the surface (Nowotny, 1991). This difference is due to segregationinduced concentration gradients at interfaces (surfaces and grain boundaries) and the resultant electrical potential gradients (Adamczyk and Nowotny, 1986). These gradients have a significant impact on the functional properties of solids, including photo-sensitivity. Therefore quantitative analysis of surface versus bulk properties is a critical issue in the interpretation of photo-induced effects. Accordingly, the imposition of specific surface versus bulk properties during the processing of photo-sensitive materials is required for the achievement of desired photo-sensitivity.

ENERGY

Centre for Materials Research in Energy Conversion, UNSW

PHOTON PENETRATION THICKNESS

*

(EF)n

EC EF

hν

*

(EF)p EV

DISTANCE FROM THE SURFACE Source: Gerischer (1977)

FIGURE 6.7 Effect of light in splitting of the Fermi energy level into two quasi-levels corresponding to electrons and electron holes for an n-type semiconductor

The preceding discussion applies to solids in thermal electronic equilibrium. However, when light is applied, then the system is in a dynamic state. Then EF may then be considered in terms of EFs splitting into two quasi-Fermi levels, (EF)n and (EF)p, shown in Figure 6.7 (Gerischer, 1977).

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

Solar-Hydrogen: A Solid-State Chemistry Perspective

185

6.5.2 EFFECT OF LIGHT ON FERMI ENERGY The light-induced split of the Fermi energy is the driving force for the charge transfer within the photo-electrochemical cell. The split results from the change in the concentration of electronic charge carriers within the valence band and the conduction band as a result on the light-induced ionization. The quasi-Fermi energy levels will be determined by increased concentrations of electrons and electron holes, due to ionization, by the respective components 6n and 6p:

(11)

(12) Therefore: (13)

(14) where 6EF is the change of the Fermi energy induced by light and no and po denote the concentration of the electronic charge carriers before the light is applied. The effect of light on (EF)n in an n-type semiconductor such as reduced TiO2 is relatively small because the component no is relatively large compared to 6n. However, the effect of light on (EF)p in the n-type material is substantial because the component 6p is large compared to po, which, in the dark, corresponds to the minority charge carriers. Consequently, application of the light results in a substantial increase of the oxidation potential represented by (EF)p leading to the shift of the equilibria represented by Equations 2–4 to the right. Then the rate of these reactions is determined by the light intensity and the resulting (EF)p level.

6.5.3 BASIC RELATIONSHIPS There have been an enormous number of reports on the application of electrical methods to investigate the properties of materials, in particular their functional properties (Kofstad, 1972). The electrical properties of semiconducting solids are very sensitive to the structure and chemical composition (including nonstoichiometry). The direct assessment of the structure and composition of solids can be time-consuming, complicated and expensive. The determination of these factors during processing at elevated temperatures is much more difficult if at all possible. However, measurements of the electrical properties at both room and elevated temperatures are more straightforward. The disadvantage of the use of electrical properties for the assessment of structure and composition is that these properties can be assessed only indirectly. However, there are three substantial advantages to materials assessment through the measurement of electrical properties:

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

186

J. NOWOTNY, T. BAK, L. R. SHEPPARD AND C. C. SORRELL

1 they are extremely sensitive; 2 they require relatively uncomplicated analytical infrastructure; and 3 they can be used not only for assessment of structure and composition but also for in situ monitoring of changes to them during processing. The electrical properties used most frequently for the assessment and monitoring of semiconducting solids include (Kofstad, 1972; Nowotny, 1997): ● ● ● ●

electrical conductivity; thermoelectric power (Seebeck coefficient); electrical conductivity and thermoelectric power measured simultaneously; and work function.

The following sections discuss the relationships between these three electrical properties, which can be determined experimentally, and the quantities describing the electronic structure, Eg and EF.

6.5.3.1 Electrical conductivity The electrical conductivity, m, includes the conductivity components of all charge carriers taking part in conduction (electrons, electron holes and ions): (15) where: (16) (17)

(18) where e is the electron charge, + is the mobility, n, p and [i] are the concentrations of electrons, electron holes and ions respectively, z is the charge of ionic species, and l is their number. Therefore: (19) The temperature dependence of the electrical conductivity can be used to determine the electronic structure of a solid (Kofstad, 1972). Becker and Frederikse (1962) showed that there is an explicit relationship between the Eg of a semiconductor that exhibits an n–p transition, such as titania, and the temperature dependence of the minimal value of the m at the transition: (20)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

Solar-Hydrogen: A Solid-State Chemistry Perspective

187

where mmin is m at the n–p transition point, which then assumes a minimum. Solids, particularly nonstoichiometric compounds, can be well defined when they are in a state of gas/solid equilibrium. Consequently, well-defined data for the m should be determined at elevated temperatures and under controlled gas phase compositions. In the case of oxide materials, oxygen is the most important component of the gas phase under which the material is annealed or equilibrated since oxygen is a component of the lattice. Thus its activity in the gas phase results in the imposition of its activity in the lattice and the related defect disorder. The determination of the electrical conductivity of oxide materials has been described elsewhere (Nowotny, 1997). As seen in Equations 16–19, the determination of the concentration of charge carriers from the electrical conductivity data is possible when the mobility terms are known (or vice versa). The evaluation of the mobility terms corresponding to electronic charge carriers requires the application of more sophisticated experimental and theoretical approaches (Kofstad, 1972). The electrical conductivity measurements may also be applied for monitoring the gas/solid equilibration kinetics for a metal oxide/oxygen system (Kofstad, 1972). While the electrical conductivity is the most commonly measured electrical property, its physical meaning is complicated because it incorporates both concentration and mobility terms (Kofstad, 1972). The commonly assumed interpretation of the electrical properties of materials is based on the assumption that changes in the electrical conductivity are determined by the concentration term alone, while the mobility term remains independent of concentration. Unfortunately, this assumption frequently is not valid, which is the reason why studies based solely on electrical conductivity may lead to a misleading picture of semiconducting properties.

6.5.3.2 Thermoelectric power Although the experimental determination of thermoelectric power, S, is more complicated than that of electrical conductivity, thermoelectric power is independent of the mobility term and thus is determined solely by the concentration term (Nowotny, 1997). The thermoelectric power is the temperature coefficient related to the electrical potential difference 6^ (termed the ‘Seebeck voltage’ or ‘thermovoltage’) generated along a temperature gradient 6T imposed on a solid. Knowledge of both 6^ and 6T are required for the determination of the thermoelectric power or thermopower (Nowotny, 1997): (21) Since S may vary with temperature, correct determination of the S requires measurement of the thermovoltage at temperature gradients as small as possible. Assuming the Maxwell-Boltzmann statistics, which may be applied at the limitations imposed by Equations 7 and 8, the S may be related directly to the EF (Nowotny, 1997): (22)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

188

J. NOWOTNY, T. BAK, L. R. SHEPPARD AND C. C. SORRELL

(23) where Sn and Sp denote the Seebeck coefficient components corresponding to electrons and electron holes respectively, and An and Ap are kinetics constants related to the scattering of electrons and electron holes respectively. Experimental determination of Sn and Sp is straightforward when one type of charge carrier is predominant. The separation of S into the two components Sn and Sp within the n–p transition regime is more complicated. Methods for the determination of the components of S in terms of different charge carriers are described elsewhere (Wagner, 1972). Combination of Equations 9 and 10 with Equations 22 and 23 may be used to establish a direct interrelationship between S, which is a measurable quantity, and the concentration of electronic charge carriers, which are essential to the assessment of the charge transfer. Experimental determination of thermoelectric power at elevated temperatures and under controlled gas phase compositions has been described previously (Nowotny, 1997).

6.5.3.3 Thermopower versus electrical conductivity The simultaneous determination of both m and S is a very useful experimental approach in assessing the charge transport. Both m and S may be used for the determination of Centre for Materials Research in Energy Conversion, UNSW

) σmin

σi

0

eB ⎯⎯⎯ k ln10

D

lg 2

(

k ln ⎯⎯ 4e B - 1 2B - ⎯ e k

2B

THERMOPOWER, S

Sp

Sn

log σ Source: Jonker, 1968; Nowotny, 1997

FIGURE 6.8 The schematic Jonker plot of thermoelectric power vs log m showing the critical parameters (meanings of the symbols are explained in text)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

Solar-Hydrogen: A Solid-State Chemistry Perspective

189

semiconducting properties of metal oxides that exhibit measurable n–p transitions according to the analysis proposed by Jonker (1968). The concept of this analysis is based on determination of the key points of the Jonker plot, which consists of m versus log S. This plot, which exhibits a pear-like shape, is shown in Figure 6.8 and reveals that the coordinate system used is described by the characteristic parameters mmin, B and D (Jonker, 1968; Nowotny, 1991). Quantitatively, these parameters are defined in Equations 24–26:

(24) where: (25) and (26) where +n and +p are the mobilities of the electrons and electron holes respectively. The parameters B and D, which can be determined from the Jonker’s formalism, are directly related to the band gap and the ratio of the mobility terms respectively. Since the Jonker analysis allows determination of the mobilities, then Equation 19 may be used for the determination of the concentration of charge carriers from the electrical conductivity data. Also Equation 25 allows direct determination of the Eg when the kinetics terms An and Ap are known. A specific feature of the analysis described by Equations 24–26 is that knowledge of the p(O2) under which the oxide specimen is equilibrated is not required. However, it is critical that both m and S are determined in equilibrium under identical p(O2). The simultaneous determination of both the m and S of oxide semiconductors is possible using a hightemperature Seebeck probe, which has been described previously (Nowotny, 1991). A prototype of this instrument is available in the Centre for Materials Research in Energy Conversion, University of New South Wales (UNSW).

6.5.3.4 Work function According to the band model of semiconducting materials, as shown in Figure 6.6, the work function q is the energy difference between the reference energy level (E = 0) and the Fermi energy level EF: (27) Therefore: (28) Consequently, the   is a direct measure of the EF at the surface. Therefore   measurements may be used for monitoring changes in EF during processing. The most

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

190

J. NOWOTNY, T. BAK, L. R. SHEPPARD AND C. C. SORRELL

widely applied method for the determination of work function changes of compounds at elevated temperatures and under controlled gas phase environment is based on the determination of the contact potential difference (CPD) (Nowotny, 1997): (29) where  R is the work function of the reference material. Pt has been used for the reference material in measurements of work function changes as a function of p(O2). Therefore knowledge of the reference energy level allows determination of the EF at the surface: (30) The q changes in oxide materials at elevated temperatures (up to 1300K) in gas phases of controlled oxygen activities using a high-temperature Kelvin probe have been described previously (Nowotny, 1991). A prototype of this instrument is available in the Centre for Materials Research in Energy Conversion, UNSW. This unique surface-sensitive tool may be used for: ● determination of the chemical potential of electrons; and ● in situ monitoring of the changes in the chemical potential of electrons during the

processing of oxides at elevated temperatures enabling the establishment of desired nonstoichiometry and related defect disorder through either imposition of specific oxygen activities or incorporation of aliovalent ions (forming donors and acceptors).

6.6 WHY TITANIA? There are many candidates for photo-electrode materials for solar-hydrogen. While some materials are known to demonstrate outstanding photo-sensitivity (Licht et al, 2000), these are not suitable for commercial applications because they have limited resistance to corrosion and photo-corrosion in aqueous environments. Consequently, it is essential to consider all of the requirements that the photo-electrode must meet. TiO2-x has a unique set of properties that make it the most promising candidate for photo-electrodes. The relevant considerations are as follows: ● TiO2-x is reactive with both light and water, which may be attributed to the ease

with which the Ti ions (Ti3+ and Ti4+) alter their valence (Kofstad, 1972; Matzke, 1981). ● TiO2-x has excellent resistance to chemical and photochemical corrosion in aggressive aqueous environments (Bak et al, 2002). ● The properties of TiO2-x can be altered by varying the defect chemistry and associated electronic structure through the introduction of aliovalent ions and the alteration of the concentration of intrinsic defects. ● TiO2-x is substantially less expensive than other photo-sensitive materials.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 169–215

Solar-Hydrogen: A Solid-State Chemistry Perspective

191

● The different processing technologies for TiO2-x production are relatively

straightforward and uncomplicated compared to those required for valence semiconductors. The economic feasibility of photo-electrochemical generation of solar-hydrogen requires the energy conversion efficiency (ECE) to be increased from current levels of 1000m2), with a corresponding increase in heat costs.

7.4.2 GUIDELINES FOR SYSTEM DESIGN 7.4.2.1 Solar collector field The selection of an appropriate solar collector basically depends on the operating temperatures (Table 7.5), though other aspects, such as the possibility of roof integration or system size, also have to be considered. The necessary collector field size can be estimated from the results presented in section 7.3. The maximum annual energy gain that can be obtained from a solar thermal system varies from 350 to 1100kWh/m2, depending on the site and the working temperature. For an accurate design and sizing of the systems, dynamical system simulations should be carried out taking into account the specific demand profile. The peak thermal power of solar collectors is about 500W/m2 for medium temperature and as high as 800W/m2 for low temperature systems.

160

6

9

140

4 11

Energy cost [€/MWh]

120

5

100

17 3

20

80

22

23

24 60

18 19

15

10

12 1

8

16 40

13 20

21 2 14

0 0

20

40

60

80

100

120

140

Working temperature

Notes: Colours: very high (black), high (grey), medium (white) solar radiation. Q: systems with continuous demand;

䉬: Systems with continuous demond during the week and weekend breaks; J: systems with

continuous demand during the week and weekend breaks, but with long-term (weeked) storage; 䉱: systems with only seasonal operation. Source: POSHIP (2001)

FIGURE 7.16 Costs of useful heat for the different systems as a function of the mean annual working temperature

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

160

Solar Heat for Industrial Processes

243

TABLE 7.5 Selection of solar collector types according to working temperatures TEMPERATURE RANGE 40ϒC 40–70ϒC 70–100ϒC > 100ϒC

PROCESS Unglazed selective collectors or low cost standard flat plate collectors Highly selective flat plate collectors or CPC collectors CPC collectors, evac. tubes or other high efficient stationary collectors; concentrating collectors for medium and large systems Concentrating collectors; evac. tubes with CPC

Source: POSHIP (2001)

Optimum annual heat gains per unit collector area in the Iberian Peninsula (latitude from 37ϒto 42ϒ) are obtained for non-tracking systems at collector inclinations of 30ϒand a row separation of at least 1.5 times the gross collector (row-) height. The fraction of gross collector area to roof area in this case is less than two thirds. If a limited roof area is available and maximum total solar heat gains are desired, the collector inclination can be set down to 10ϒapproximately. Nevertheless, for glazed collectors a lower limit of 20ϒfor the collector’s inclination is recommended in order to avoid rain water infiltration and enhance self-cleaning of the glass sheet by rain water. A deviation from south orientation of up to 45ϒ can be tolerated, leading to about a 10 per cent reduction in the system’s output. Parabolic trough collectors should be tracked around an axis oriented N–S, to maximize the annual heat production. The row spacing should be about three times the aperture width to avoid excessive shadowing in the morning and evening. By an appropriate design of flow rates, pipe diameters and pipe insulation, the electricity consumption for fluid circulation can be below 1 per cent of the overall heat gains. Thermal losses in the piping and storage should not be above 5 per cent of the overall heat gains for medium and large size systems. In earlier studies, depending on the load characteristics and profile, different solutions have been studied, like single-pass versus multi-pass strategies (Gordon and Rabl, 1982; Collares Pereira et al, 1984) or constant versus variable flow rate (Gordon and Zarmi, 1985) in order to assure good temperature stratification in the storage. Solar gains are highest if the load and solar supply flow rates are matched with the help of flow regulators. A bypass for the preheating of the primary circuit during the morning hours and the extraction of the residual heat after sundown by appropriate regulation can improve solar gains by up to 20 per cent.

7.4.2.2 Storage An introduction to solar systems with heat storage has been given in section 7.3.3. Short-term heat storage is recommended whenever a mismatch between available solar radiation and heat demand occurs. For short-term storage (several hours), volumes of about 25 litres/m2 are recommended. Short-term storage may even be recommended for continuously operating processes, in order to lower the mean working temperature of the solar system and thereby improving its efficiency, especially if low cost solar collectors with high thermal loss coefficients are used. The larger the system’s size, the more effective the heat storage over longer periods (for example weekends). Weekend storage becomes economic from about 500m2 and above. Weekend storage volumes should be about 250 litres/m2. In this case, storage

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

244

H. SCHWEIGER et al

costs can range from 10 to 20 per cent of the total system costs. Storage for longer periods (seasonal storage) can only be considered for very large systems (> 5000m2).

7.4.2.3 Coupling to the existing heat supply and control Coupling the solar system to the existing heat supply should always be attempted at the lowest possible temperature. Nevertheless, for liquid and air preheating, solar heat should Solar irradiation

Hot water to process

Solar collectors

Waste water Heat exchanger Heat recovery

Source: POSHIP (2001)

FIGURE 7.17 Combination of solar thermal system and waste heat recovery

3000 2500

MWh

2000 1500 1000 500 0 solar heat

solar heat &

cogeneration

cogeneration

electricity

high eff.

low eff.

Note: Conversion efficiency (electrical) for: (a) highly efficient (combined cycle) cogeneration plants (50%); (b) low efficiency cogeneration plants (25%) Source: POSHIP (2001)

FIGURE 7.18 Primary energy saving with respect to a conventional steam boiler, for 1000MWh of industrial process heat produced either by solar thermal systems or conventional cogeneration system

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

Solar Heat for Industrial Processes

245

TABLE 7.6 Evaluation criteria for solar industrial process heat systems for the Iberian Penninsula CRITERION Working temperature

INFLUENCE ON SYSTEM PERFORMANCE Working temperatures not higher than 150ϒC, best performance below 100ϒC Very good conditions in the southern and central regions of the Iberian Peninsula. In regions with medium or low levels of solar radiation (H < 1600kWh/m2) systems should only be considered if all other conditions are favourable (low temperature, continuous demand).

Climate

Continuity of the demand: Annual variation Daily variation

System size

Annual solar gains Solar fraction Available roof or ground area

Static load aspects of the roof

Waste heat recovery and cogeneration

Breaks in summer reduce system performance. Losses in solar gains are more than proportional to the time interval of the break. Continuous demand or demand with peaks at daytime are favourable. Short interruptions (several hours) can be buffered by low volume storage with little increase in system cost. The economic performance of solar thermal systems depends strongly on the system size. Resulting solar energy costs are up to 50% lower for large systems (> 1000m2, thermal peak power > 0.5MW) than for small systems (100m2, thermal peak power 50kW). Annual solar gains of a solar system should be at least 500 kWh/m2 for economic profitability. Systems should be designed for solar fractions not higher than about 60% (for continuous demand). Sufficient roof or ground area should be available in order to obtain solar fractions from 5 to 60%. Orientation to the south with inclination of about 30ϒis the optimum. Small deviations from these values are tolerable (±45ϒfrom south orientation, ±15ϒfrom optimum inclination). Long piping should be avoided. The need for reinforcement of roof structures increases system cost and therefore reduces economic performance. The additional static load of solar collectors is 25–30kg/m2 for standard collectors. First all means of improving energy efficiency by waste heat recovery and by co-generation should be exploited. Solar systems should be designed to cover (part of) the remaining heat demand.

Source: POSHIP(2001)

be introduced only after first preheating by waste heat recovery systems (Figure 7.17), and not as an alternative to these systems. Even if the waste heat recovery raises the working temperature in the solar system, the combination of both systems yields better results than a solar system at lower temperature without heat recovery. If solar heat is being fed to several processes, a control strategy should be chosen in order to reach an optimum overall energy saving. In most cases feeding solar energy

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

246

H. SCHWEIGER et al

to the process with the lowest temperature might be the best choice, but in some cases giving preference to heat production at higher temperatures during the hours around noon with high insolation results in a better overall performance. The control strategy should be optimized for the specific cases using dynamic system simulation techniques.

7.4.2.4 Solar thermal systems and cogeneration As a general rule, it is possible to state that solar thermal systems should be designed as a complementary system to cogeneration plants covering the remaining heat demand. Only in some cases can solar thermal energy be considered as an alternative heat supply system. Shutting down an existing cogeneration plant during periods when solar heat is available in excess would imply a higher importation of electricity from the grid during this period. In Figure 7.18 the primary energy saved by producing 1000MWh of process heat by solar thermal systems or by cogeneration, instead of a conventional steam boiler, is compared. The primary energy savings for the produced electricity are taken into account, using as a reference the mean efficiency of the present electricity supply (about 35 per cent). In addition, the possibility of simultaneous heat and electricity production from solar thermal systems is considered, assuming a conversion efficiency of 30 per cent from heat to electricity. It can be seen that solar heat saves primary energy compared to cogeneration plants only if either the electrical efficiency of the installed cogeneration plant is lower than the mean efficiency of the grid, or if the solar system itself produces both heat and electricity.

7.4.2.5 Summary The most important evaluation criteria for solar process heat plants are summarized in Table 7.6.

7.5 CASE STUDIES Within the POSHIP project, case studies have been carried out for more than 20 companies in Spain and Portugal that showed some interest in applying solar thermal energy in their factories. Based on a detailed analysis of the heat (and cold) demand of the industries, appropriate solar systems have been proposed. A prediction of the annual energy gains of these systems based on dynamic system simulations (TRNSYS) and economic analyses have been performed. A description of some selected systems is given in the following sections. For an overview see Table 7.7. The geographical locations of the proposed demonstration plants are shown in Figure 7.19.

7.5.1 MALTING FACTORY, ANDALUCÍA (SPAIN) The malting factory is located in the province of Seville, in southern Spain. The malt produced is used for beer production in the same factory. About 80 per cent of the heat and cold consumption in the malting factory is used for air heating for drying the malt (hot

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

Solar Heat for Industrial Processes

247

TABLE 7.7 List of the most relevant parameters of the demonstration projects proposals COMPANY SECTOR LOCATION

PROCESS

WORKING TEMPERATURE RANGE

COLLECTOR TYPE**

PROJECTED COLLECTOR AREA (m2)

ESTIMATED COST (⇔)

Malte Ibérica Malting factory Poceirão, Portugal

Air for malt drying preheating

Low temperature (20–60ϒC)

FPC or CPC

3500

624,000

Heineken España S.A. Malting factory Sevilla, Spain

Air preheating for malt drying

Low temperature (10–80ϒC)

SFPC

3046

930,000

Beirala Lanificios Textile factory Seia, Portugal

Water preheating Low temperature in dyeing process (20–80ϒC)

FPC or CPC

1750

340,000

CPC

60

55,000*

Bodegas Mas Martinet Cooling of wine Medium Wine cellar stores and space temperature Tarragona, Spain heating (50–85ϒC) Notes: * cost for solar system and absorption cooling system.

**FPC: flat plate collector; CPC: compound parabolic concentrator; SFPC: selective flat plate collector. Source: POSHIP (2001)

Source: POSHIP (2001)

FIGURE 7.19 Geographical location of the proposed demonstration plants

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

248

H. SCHWEIGER et al

Cold generation 19%

Malt drying 81% Source: POSHIP (2001)

FIGURE 7.20 Heat demand by processes in the malting factory in Seville

air at about 60ϒC). Most of the remaining 20 per cent is required for air cooling in the germination process (Figure 7.20). The profile of the heat demand in this factory is ideal for the implementation of a solar system: ● The heat demand is high and continuous during 7 days a week and 24 hours a day,

so nearly all the collected solar energy can be used directly in the process without need of large storage volumes. ● The required temperatures are low. Air is preheated by the solar system only a few degrees above ambient temperature. ● The factory is located in a region with a high solar radiation (1770kWh/m2a). ● Energy costs are a significant cost factor in the malt production process, so a reduction of energy consumption can help improve competitiveness. A solar system is proposed for the preheating of air for the malt drying process. An additional hot water to air heat exchanger will be installed in series before the existing heat exchanger using steam from the conventional heat supply system. The solar system consists of a solar collector field of a total surface of 5000m2 and a storage tank of 1000m3 (Figure 7.21 and Table 7.8). The solar collectors will be integrated in the saw-tooth shaped roof structure forming a rain-tight ‘solar roof’.

7.5.2 MALTIBÉRICA – SOCIEDADE PRODUTORA DE MALTE S.A. (PORTUGAL) Maltibérica is a Portuguese–Spanish factory producing malt for breweries pertaining to the UNICER Group in Portugal and also exporting it to Spain. Although it is located not far from a very industrialized zone of Portugal, the Setúbal Peninsula, around 50km from Lisbon, no other factories are located nearby, increasing the difficulties to extend the natural gas network to that area. As a result, it is very expensive to change from the actual usage of heavy fuel oil in great quantities to a cleaner conventional energy source.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

Solar Heat for Industrial Processes

249

Source: POSHIP (2001)

FIGURE 7.21 Scheme of the proposed solar system for the malting factory in Seville

TABLE 7.8 Characteristic data of the proposed system for solar air heating in a maltting factory in Seville, Spain Company Location Ind. sector Process Working temperature Solar collector area Solar collector type

Heineken España S.A. Seville, Spain Malting factory Air preheating for malt drying 10–80ϒC 5000m2 Selective collectors without glass cover integrated into a saw-tooth roof structure 1000m3 4770MWh 7% ⇔1,320,000 ⇔410,000 ⇔105,000 4 years

Solar storage Annual solar gains (useful heat) Solar fraction Total investment Total own contribution Yearly net savings Payback Source: POSHIP (2001)

The factory produces malt, which undergoes three steps during the process: 1 wetting of barley; 2 germination of green malt (cold); and 3 drying of malt (heat). The largest part of the energy consumption is for the production of superheated water (180ϒC, 14 bars), which is used with the water–air heat exchangers for drying the malt

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

250

H. SCHWEIGER et al

Source: POSHIP (2001)

FIGURE 7.22 Malting factory at Poceirão, Portugal: Ground layout

(60–80ϒC). A small part is consumed for keeping thick fuel at its appropriate storage temperature. The factory is quite new (less than 10 years old) and its design incorporates heat recovery measures that improve the potential use of solar energy when considering new energy-saving actions.

preheated air 25–50°C

Collector field

Source: POSHIP (2001)

FIGURE 7.23 Layout of the proposed solar system for the malting factory at Poceirão, Portugal

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

Solar Heat for Industrial Processes

251

There are several favourable factors for the implementation of a solar system in the factory: ● ● ● ●

the factory is in operation all day throughout the year round; energy can be used at the time of its collection, which avoids the need solar storage; solar energy is used at a low temperature to preheat the outside air for drying; the factory is located in a site with very good annual sunny conditions: more than 2800hours of sunshine and 1720kWh/m2a; ● the solar energy system would remain totally independent from the existing conventional system. The ground layout of the factory is given in Figure 7.22 and the layout of the proposed solar system in Figure 7.23. The most relevant data of the proposed plant are summarized in Table 7.9. TABLE 7.9 Characteristic data of the proposed system for air preheating for malt drying in Portugal Company Location Ind. sector Process Working temperature Solar collector area Collector type Storage volume Annual solar gains (useful heat) Solar Fraction Total investment Total own contribution Yearly net savings Pay back

Grupo Unicer Poceirão (Setúbal) Malting factory Air preheating for malt drying 20–80ϒC 3500m2 Selective flat plate or CPC collectors —3762MWh/year 20% ⇔624,121 ⇔405,679 ⇔118,320 4 years

Source: POSHIP (2001)

7.5.3 BEIRALÃ – LANIFÍCIOS S.A. (PORTUGAL) Beiralã is a Portuguese textile factory that produces woollen garments and cloth. It is located in the north-west of central Portugal, at the foot of the highest Portuguese mountain, Serra da Estrela, with average annual solar radiation conditions: 1520kWh/m2a and 2400 hours of sunshine (Figure 7.24). In the dyeing process, only a small fraction of the process heat consumption (20 per cent) is at 120ϒC and most does not exceed 80ϒC. A great quantity of hot water at 60–70ϒC is consumed for washing the dyeing vessels. There is a variety of solar collector technologies available on the market for this level of temperature appropriate for this application. Favourable conditions for the implementation of a solar system apply due to the following:

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

252

H. SCHWEIGER et al

Source: POSHIP (2001)

FIGURE 7.24 Solar radiation in Portugal and location of the Beiralã plant ● access to the national natural gas network is not expected in the short term; ● due to the high transportation costs, it is very expensive to change from heavy fuel

to a cleaner fossil fuel like propane gas; ● solar energy can complement the natural gas, therefore lowering the final energy

costs and is ideal to meet the strategic ISO 14000 certification target; ● a large area, with good orientation, is available on the roof of the factory for placing

solar collectors; ● the solar system can be designed in a very common way, storing the collected

energy in an already existing concrete storage tank; and ● the dyeing process and/or the washing process will be supplied by water from the

solar tank after increasing its temperature by the back-up conventional energy system as necessary.

Back-up

Source: POSHIP (2001)

FIGURE 7.25 Scheme of the proposed solar system for the Beiralã plant

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

Solar Heat for Industrial Processes

253

The scheme of the proposed solar system is given in Figure 7.25. The most relevant data for the proposed plant are summarized in Table 7.10. TABLE 7.10 Characteristic data of the proposed system for a textile drying process Company Location Ind. sector Process Working temperature Solar collector area Collector type Storage volume Annual solar gains (useful heat) Solar fraction Total investment Total own contribution Yearly net savings Payback

Beiralã Seia Textile Preheating of water for dyeing process 20–80ϒC 1750m2 Stationary collectors, selective flat plate or CPC type 75m3 1331MWh/year 33% ⇔340,430 ⇔221,280 ⇔84,496 3 years

Source: POSHIP (2001)

7.5.4 BODEGAS MAS MARTINET, TARRAGONA (SPAIN) The Bodegas Mas Martinet wine cellar is located in the Priorat region, in the south of Catalonia. The company produces a high quality wine for national consumption and for export. The cellar has no connection to the electricity grid, so it tries to cover most of the energy demand by solar energy (thermal and photovoltaic). Most of the heat and cold consumption in the wine cellar arises for the cooling of the cellar during the summer months and for the heating of an office building and a residential building during the winter season. A small fraction of the heat demand is used for domestic hot water production. The yearly distribution of the heating and cooling demands are shown in Figure 7.26. Electricity generation using the installed diesel generator is expensive, so Cooling

Heating

DHW

10000

8000

[kWh]

6000

4000

2000

be r

be r De ce m

No ve m

be r

t

O ct ob er

Se pt em

Au gu s

Ju ly

Ju ne

ay M

Ap ril

ar ch M

ry Ja nu a

Fe br ua ry

0

Source: POSHIP (2001)

FIGURE 7.26 Heat, cold and domestic hot water demand of the installations of Bodegas Mas Martinet

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

254

H. SCHWEIGER et al

Solar contribution

Auxiliary Heat Demand

10000

8000

6000

4000

2000

be r De ce m

be r No ve m

O ct ob er

t

be r Se pt em

Au gu s

Ju ly

Ju ne

ay M

Ap ril

ar ch M

Fe br ua ry

Ja nu a

ry

0

Source: POSHIP (2001)

FIGURE 7.27 Total heat demand of Bodegas Mas Martinet and solar contribution throughout the year

favourable conditions are encountered for an alternative thermal cooling system that substitutes the electricity consumption of the compression chillers. A solar system is proposed for the air conditioning of the wine cellars and the office building using a lithium bromide absorption cooling machine (Yazaki Company). The excess heat from the solar collectors during the winter months is used for space heating and domestic hot water production. In the proposed system, fan-coils are used for the distribution of heat and cold to the wine cellar and the connected office building, and a floor heating system for a nearby residential building. The solar system consists of a solar collector field of 60m2 of CPC collectors producing hot water at 80–90ϒC and 1500L of hot water storage. The absorption chiller has a nominal cooling power of 35kW and is buffered by a 3000L cold storage. The total heat demand of the factory (including the heat demand of the absorption chiller) and the solar contribution is given in Figure 7.27. Figure 7.28 shows the hydraulic scheme of the plant and Table 7.11 gives the characteristics of the solar system.

7.5.5 PRE-FEASIBILITY STUDY ON SOLAR PROCESS STEAM GENERATION FOR THE PRODUCTION OF POROUS CONCRETE A pre-feasibility study on solar process steam generation for the production of porous concrete was carried out in cooperation with the YTONG Company (Hennecke et al, 2002). Porous concrete is a building material with low heat conductivity, supporting energy conservation in the building sector. The manufacture of 1m3 of this material consumes some 155kg steam at 10 bars, equivalent to about 87kWhth. The motivation for this investigation is to partly substitute this energy by steam produced in solar parabolic trough collectors, reducing the fossil energy consumption and improving further the overall energy balance of this material.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

Solar Heat for Industrial Processes

Zona casa

floor heating system

cooling from pool

Parte ya instalada ACS casa

Bacs-casa Suelo radiante Ttemp-SR

Vcasa Vtemp-SR

255

Retorno a balsa

Agua de la balsa

T BSR

T VSR Caldera calefacción

Bcasa

Tbalsa

Filtro

TSR-ret

Bbalsa Campo de colectores

Acumulador de calor

Thcol-o

Vdep1500

Ftorre

Máquina de absorción

Caldera auxiliar

Ths-top

Thx-hi

Tcaldera

Thot-ma Vhot-ma

Bhot-ma

Vbalsa

1 500 l

B1ario Thx-ho

B2ario

Ths-bot

Tret-2ario

Benfriamiento

Agua de renovación

Tevap-o ACS oficina

Vinter-dep

Tacs-of-top

solar thermal system Sala de calderas

Bacs-of

Vcal-ida T cal-ida

Vcr-of-ret

Tcs-top

TFC-ret

Vfrío-aux2

absorption cooling system

VD-evap2

Vcr-of

3 000 l

Bof

VD-evap1 Tcs-bot Acumulador de frío / calor

Vfrío-ida

Vfrío-aux1

Balm

Tfrío-ida Bbod F.C.- B2

F.C.- B1

F.C.- B3

F.C.- B4

F.C.- Alm.

F.C-Of.1 F.C-Of.2 F.C-Of.3

Source: POSHIP (2001)

FIGURE 7.28 Scheme of the proposed solar system for Bodegas Mas Martinet

TABLE 7.11 Characteristic data of the proposed system for space heating and cooling of a wine cellar Company Location Ind. sector Process Working temperature Solar collector area Solar collector type Solar storage Annual solar gains (useful heat) Solar fraction Total investment Total own contribution Yearly net savings Payback

Bodegas Mas Martinet Tarragona, Spain Wine cellar Space cooling and heating 50–85ϒC 60m2 CPC collectors 1500L (hot) + 3000L (cold/hot) 21MWh 58% (cooling) 35% (heating) ⇔55,000 ⇔35,000 ⇔3000 11 years

Source: POSHIP (2001)

Reference data for this study were obtained from a typical factory in Hamm, Germany. The main steam-consuming process is the curing of the material under precisely controlled conditions in large autoclaves (Figure 7.29). Several autoclaves are operated in a staggered mode, and Ruth-type (saturated steam) storages are commonly used to

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

256

H. SCHWEIGER et al

pressure [bar] steam flow [t/h] 15 10 5 0 -5

0

2

4

6

8

time [hours] 10

–10 –15 Source: Hennecke et al (2002)

FIGURE 7.29 Autoclave process (simplified)

manage the steam flows efficiently during charging and discharging. It should be pointed out that the flexibility of the system, designed to cope with the severe load changes caused by the batch processes, also provides distinct advantages for the integration of solar process steam. A solar field was designed to fit the roof of the main factory hall, consisting of small collectors like the parabolic trough collector from Industrial Solar Technology (IST), with a total aperture area of about 1400m2 (Figure 7.30). The nominal power output (600kWth) represents about 5 per cent of the total fossil-fired boiler capacity installed. The area of the surrounding storage yards is large enough to provide up to 100

Source: Hennecke et al (2002)

FIGURE 7.30 Solar field integration via Ruth-type (saturated steam) storage

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

Solar Heat for Industrial Processes

257

per cent of the factory’s total steam demand with collector systems mounted on elevated structures. Care must be taken in the long, horizontal and asymmetrically heated boiler tubes to avoid potentially dangerous flow patterns of the two-phase flow. Theoretical investigations indicate that the problem of flow stratification will be reduced at lower pressures. However, below about 5 bars the low steam density leads to high velocities and pressure drops, which might prevent the achievement of high steam qualities at the outlet of the solar steam generator. The recirculation process provides a convenient solution, particularly if the existing ruth’s steam storages can be used as the steam drum to separate steam and water. During times when no steam is extracted for the production, the solar energy input will lead to rising pressure in the system, increasing the exergetic value of the stored energy and consequently reducing live steam consumption requirements by the same amount. At the same time, any direct interference between the solar steam generator and the delicate production process can be avoided. The annual system performance has been simulated for different sites as a basis for cost estimations, indicating that in the Mediterranean area the solar steam may be competitive with conventional boilers (Table 7.12). At favourable sites, 10 per cent of the annual steam demand can be satisfied from the solar field sized to provide 5 per cent of the fossil boiler capacity. TABLE 7.12 Annual system performance for different sites LOCATION Würzburg, Germany Portorozˇ, Slovenia Mugla, Spain

DNI kWh/m2a 894 1330 1900

ANNUAL HEAT PRODUCTION kWh/m2a 294 472 761

LEVELIZED HEAT COST ⇔/kWh 0.11 0.07 0.045

Note: DNI: direct normal irradiance Source: Hennecke et al (2002)

7.6 CONCLUSIONS Industrial solar thermal systems as a renewable energy source can cover a significant fraction of the industrial heat and electricity demand. This industrial heat demand constitutes about a third of the total final energy demand in southern European countries. About 7 per cent of the total final energy is consumed in the form of industrial process heat at temperatures below 250ϒC. Fulfilling the goal of installing 2,000,000m2 of solar thermal collectors for industrial process heat and solar cooling, as outlined in the ‘Campaign for Take-Off’ of the European Commission would represent a primary energy saving of about 2,000,000MWh/year. Therefore, solar thermal energy in industry can be an important contribution to a reliable, clean, safe and cost-effective energy supply based on renewable energy sources. Solar collectors for hot water production at low temperatures are a well-known and well-extended standard technology. With recently developed high performance solar collectors, heat at temperatures up to 250oC can be produced with excellent efficiency. Heat at these temperatures is required in many industrial processes, such as steam generation, washing, drying, distillation and pasteurization.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

258

H. SCHWEIGER et al

At present, industrial solar thermal systems are operating in Europe with a total installed collector surface of about 10,000m2. Solar collector arrays can be integrated either into industrial roofs or installed on an available ground area. Large-scale industrial installations may lead to very low cost systems, meaning solar systems for industrial process heat production can become economically competitive with fossil fuels on a very short term. Present investment costs for solar thermal systems range from 250 to 500⇔/m2 (corresponding to 250–1000⇔/kW of thermal power), leading to average energy costs in southern Europe from 2 to 5 cents/kWh for low temperature applications and from 5 to 15 cents/kWh for medium temperature systems. The present cost per kWh of saved primary energy in industrial applications is less than that of small- to medium - scale solar domestic hot water applications and can be further reduced. Cost reductions can be obtained by large-scale production, reducing operation and maintenance costs and improving collector efficiency and system design, especially for medium temperature solar collectors. Cost reductions of up to 50 per cent on a medium term are expected by 2010.

AUTHOR CONTACT DETAILS H. Schweiger (corresponding author),Creu dels Molers, 15, 2-1, 08004 Barcelona, Spain Tel: (+34) 93 441 53 93; Fax: (+34) 93 441 53 95; e-mail: [email protected] Dr. Schweiger is director of the engineering company energyXperts.BCN, working in the fields of energy efficiency and renewable energies (www.energyxperts.net). J. Farinha Mendes, INETI (Instituto Nacional de Engenharía e Tecnología Industrial), Lisbon, Portugal, www.ineti.pt Ma. J. Carvalho, INETI, Lisbon, Portugal, www.ineti.pt K. Hennecke, DLR (Deutsches Zentrum für Luft- und Raumfahrt e.V.), www.dlr.de D. Krüger, DLR, www.dlr.de

REFERENCES Carvalho, M. J., Collares Pereira, M., Oliveira, J. C., Mendes, J. F., Haberle, A. and Wittwer, V. (1995) ‘Optical and thermal testing of a new 1.12X CPC solar collector’, Solar Energy Materials and Solar Cells, no 37, pp175–190 Collares Pereira, M., Gordon, J. M., Rabl, A. and Zarmi, Y. (1984) ‘Design and optimisation of solar industrial hot water systems with storage’, Solar Energy, no 32, p121 Garg, H. P. (1987) Advances in Solar Energy Technology, vol 2, Reidel Publishing Company, Dordrecht, The Netherlands Gee, R., Cohen, G., Winston, R., Greenwood, K. and McGuffey, B. (2003) ‘Design, installation and early operation of a roofintegrated solar cooling and heating system’, proceedings of the ASES Annual Conference, Dordrecht, The Netherlands, June 21–26 Gordon, J. M. and Rabl, A. (1982) ‘Design, analysis and optimisation of solar industrial process heat plants without storage’, Solar Energy, vol 28, no 6, pp519–530 Gordon, J. M. and Zarmi, Y. (1985) ‘Single-pass open-loop solar thermal energy systems: On maximum energy delivery by variable flow rate’, Journal of Solar Energy Engineering, no 107, p273 Hennecke, K., Meinecke, W. and Krüger, D. (2000) ‘Integration of solar energy into industrial process heat and cogeneration systems’, proceedings of 9th International Symposium on Solar Thermal Concentrating Technologies, Font-Romeu, France, June

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216 –260

Solar Heat for Industrial Processes

259

Hennecke, K., Kötter, J., Michel, O. and Peric, D. (2002) ‘Solar process steam generation for the production of porous concrete’, in A. Steinfeld (ed) 11th SolarPACES International Symposium on Concentrated Solar Power and Chemical Energy Technologies, Zurich, Switzerland, 2–4 September, Paul Scherrer Institut, Villigen, Switzerland Huerdes, J. V. and Lachal, B. (1986) Calor Solar Industrial, Spanish edition, Delta Energy Ltd, Rafz, Switzerland IDAE – Instituto para la Diversificación y el Ahorro de la Energía (2001) Private communication: official data from the Spanish Ministry for Industry and Energy (MINER) for 1999 Kreider, J. F. (1979) Medium and High Temperature Solar Processes, Academic Press, New York Krüger, D., Hoffschmidt, B., Hennecke, K., Pitz-Paal, R., Rietbrock, P. and Fend, T. (2000) ‘Results from parabolic trough collectors for process heat at the DLR Cologne’, in H. Kreetz, K. Lovegrove and W. Meike (eds) Proceedings of the 10th International Symposium on Solar Thermal Concentrating Technologies, Sydney, 8–10 March, Australia National University, Canberra, Australia Krüger, D., Pitz-Paal, R., Lokurlu, A. and Richarts, F. (2002) ‘Solar cooling and heating with parabolic trough collectors in the Mediterranean’, in A. Steinfeld (ed) Proceedings of the 11th International Symposium on Concentrated Solar Power and Chemical Energy Technologies, Zurich, Switzerland, 2–4 September, Paul Scherrer Institut, Villigen, Switzerland Laue, H. J. and Reichert, J. (1994) ‘Potential for medium and large sized industrial heat pumps in Europe’, Directorate General for Energy (DGXII), European Commission, contract no XVII/7001/90-8, final report Lewis, C. W. (1980) ‘The prospects for solar energy use in industry within the United Kingdom’, Solar Energy, no 24, pp47–53 Nitsch, J. and Luther, J. (1990) Energieversorgung der Zukunft, Springer-Verlag, Berlin POSHIP (2001) ‘The potential of solar heat for industrial processes’, Project NNE5-1999-0308, final report, funded by the European Commission Rabl, A. (1985) Active Solar Collectors and Their Applications, Oxford University Press, Oxford, UK Rabl, A. (1976) ‘Comparison of solar concentrators’, Solar Energy, no 18, p93 Schreitmüller, K. (1986) Erfahrungen mit solaren Proze`wärmeanlagen Solar Thermische Proze`wärme, Klaus Scharmer, Gesellschaft für Entwicklungstechnologie mbH, Aldenhoven Schreitmüller, K. (1987) ‘Stand und Perspektiven der solaren Proze`wärmebereitung unter besondrere Berücksichtigung der Aspekte von Entwicklungsländern’, DFVLR, Institut für technische Thermodynamik TNO Bouw (1995) ‘Potentieelstudie Hoogrendement-zonnecollectoren en Nederland’, Report 94-BBI-R 1368, TNO, Delft, The Netherlands TRNSYS (2000) Software for Dynamical System Simulation, Solar Energy Laboratory, Madison, Wisconsin, US, Version 14.2 (1997) and 15.0 Wackelgard, E., Niklasson, G. A. and Granqvist, C. G. (2001) ‘Selectively solar-absorbing coatings’, J. Gordon (ed) Solar Energy, the State of the Art, James & James, London Welford, W. T. and Winston, R. (1978) The Optics of Non-Imaging Concentrators – Light and Solar Energy, Academic Press Winter, C.-J. (1997) Sonnenenergie nutzen. Technik, Wirtschaft, Umwelt, Klima, VDE-Verlag GmbH, Berlin

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 216–260

This page intentionally left blank

8

Solar Energy Technology in the Middle East and North Africa (MENA) for Sustainable Energy, Water and Environment W. E. Alnaser, F. Trieb and G. Knies

Abstract Our study reveals that concentrating solar power (CSP) can be used to fulfil the total electricity demand in Middle East (ME) and North African (NA) countries of 1700TWh/y in 2025, rising to 3600TWh/y in 2050, especially as an expanding technology with a growth of 25 to 35 per cent a year. A strong competition of several CSP technologies could lead to higher overall CSP growth rates than those assumed in the present study (around 30 per cent per year). In 2025 14 per cent of the electricity demand of MENA may be covered by CSP, which will become the dominating power source with a share of 57 per cent by 2050. According to our scenarios the mix of renewable energies will cost around 5 cents/kWh in 2050. The cost of CSP will be reduced from the current 8 cents/kWh to 6 cents/kWh in 2010 to 5 cents/kWh in 2020 (other scenarios quote 15 cents/kWh). From 2020 until 2050, the MENA region will save US$250 billion compared to the business as usual policy scenario, where it was assumed that the fuel prices start at US$25/bbl for oil and US$49/ton for coal and escalate by only 1 per cent per year (currently the prices are at a level of US$55/bbl and US$65/ton respectively and the escalation rates have amounted to 40 per cent per year since 2003). The study shows that solar thermal power plants are suitable for seawater desalination. A concentrating solar thermal collector array required for desalinating 1 billion m3/y would cover a total land area of approximately 10 x 10km, corresponding to about 10m3 of desalinated water per m2 of collector area. About 10 per cent of the desalinated water would suffice to irrigate the desert land beneath the collectors with a water column of 1m/y. The study shows that the total carbon emissions of electricity generation of all MENA countries can be reduced from 770 million tons per year to 475 million tons in 2050, instead of increasing to 2000 million tons. Furthermore, the greenhouse emissions from constructing the plants for most renewable energy technologies are between 10 and 25g/kWh of useful energy. In the case of photovoltaic systems, reductions are possible in the medium term to

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

262

W. E. ALNASER, F. TRIEB AND G. KNIES

about 50g/kWh. The scenario reaches a per capita emission of 0.58 tons/cap/y in the power sector in 2050, which is acceptable in terms of the recommended total emission of 1–1.5 tons/cap.

■ Keywords – installed power capacity; global solar radiation; water deficiency; MENA; wave and tidal power

8.1 INTRODUCTION The MED-CSP study ‘Concentrating solar power for the Mediterranean region’ (DLR, 2005), which aims to catalyse the cooperation between Europe and Africa in solar energy application, focused on the electricity and water supply of the regions and countries illustrated in Figure 8.1, including southern Europe (Portugal, Spain, Italy, Greece, Cyprus and Malta), North Africa (Morocco, Algeria, Tunisia, Libya and Egypt), western Asia (Turkey, Iran, Iraq, Jordan, Palestine & Israel, Lebanon and Syria) and the Arabian Peninsula (Saudi Arabia, Yemen, Oman, United Arab Emirates, Kuwait, Qatar and Bahrain). The Arab countries in North Africa (NA) and in West Asia (WA), in general, lack energy and water. The average annual rainfall in Arab countries is from 5 to 45mm/y, while in European countries (mainly those in the Mediterranean EU) it ranges from 200 to 400mm/y. The total internal water reserve is only 100km3 for Arab countries, while in Europe it may exceed 400km3 (bear in mind that the world internal water reserve is 43,764km3). Nearly 50 per cent of the Arab countries have water availability per capita less than the absolute water scarcity level (200m3/capita/year) while the rest, except Iraq, are in water scarcity threshold level (1000m3/capita/year) (ESCWA, 2003). In Europe, the per

Southern Europe Western Asia Arabian Peninsula North Africa

FIGURE 8.1 Countries of the EU-MENA region analysed in this study

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

263

capita water resource ranges as high as 85,478m3/year (Norway). In fact among the countries with least water resource (UNDP, 2002) we have 14 Arab countries (Kuwait, UAE, Qatar, Libya, Saudi Arabia, Jordan, Bahrain, Yemen, Oman, Algeria, Tunisia, Egypt, Morocco and Palestine) where the per capita water supply ranges from 10m3/year (Kuwait) to 971m3/year (Egypt). In 1999, electricity consumption per capita in Western Europe was never less than 4000kWh, while it was as low as 46kWh in Sudan. In the majority of the Arab countries, except Arabian Gulf countries (GCC), the average person consumes electricity at the rate of 1200kWh annually. Probably more than 1 million Arab citizens (of some 300 million) have no access to electricity (UNDP, 2002). Furthermore, the European Organisation for Economic Co-operation and Development (OECD) countries emitted 3800 million tons of CO2, which is 15.2 per cent of the global emission (25,000 million tons), in 2001. The CO2 emissions from the Middle East are only 4.8 per cent compared to North America (27.7 per cent), Eastern Europe (12.6 per cent), Western Europe (15.6 per cent), Africa (8.8 per cent), Central and South America (4.1 per cent), and the Far East and Oceania (31.5 per cent) (World Almanac, 2004). According to the latest reports, Germany had managed to reduce emission of CO2 by only 20 per cent (compared to the base year 1990), while others had actually increased their emissions or stayed the same – for example the UK 12.5 per cent, Italy 6.5 per cent, France 0 per cent, Russia 6 per cent, the US 7 per cent and Japan 6 per cent (Global Emission Newsletter, 2004). These industrialized nations are required to reduce their CO2 emissions – according to the Kyoto Protocol, which took effect on 16 February 2006 – by 5.2 per cent of their emissions in 1990. The increased temperature of the globe due to global warming – due to extensive use of fossil fuels – has led to more extreme weather and hence catastrophes. According to Swiss Re-insurance Company (World Almanac, 2004) in its report on natural catastrophes and man-made disasters in 2003, nearly 36 natural accidents had occurred in Europe with 424 victims – wasting US$2173 million (11.8 per cent of the total wasting cost budget) – while in the Middle East and Asia 178 accidents had occurred with 51,894 victims – wasting US$1447 million (7.8 per cent). Therefore, mutual cooperation between Arab and European countries in the field of energy production using solar radiation is strongly advisable. The Arab countries are characterized and blessed with abundant direct solar radiation, ranging from 4.1kWh/m2/day in Mosul, Iraq to 6.7kWh/m2/day in Nouakchott, Mauritania (Alnaser et al, 2004; ESCWA, 2001). Of course, the total solar radiation will be larger than these figures by nearly 30 per cent. Moreover, the maximum recorded annual mean sunshine duration ranges from 7.5 hrs in Tunis to 10.7 hrs in Egypt. These figures are much larger, at least three times, compared to European countries (Alnaser et al, 2004). The electricity demand and temporal behaviour among MENA and European countries is found to complement each other. Therefore cooperation will enhance renewable energy utilization worldwide and it will increase to more than the current level of 13.8 per cent of the total primary energy supply, where 2.3 per cent is from hydro, 11.0 per cent from combustible renewable and waste and only 0.5 per cent covers solar (0.039 per cent), wind (0.026 per cent), tidal (0.004 per cent) and geothermal (0.440 per cent) power. The solar power market is already growing – it was 1000MW in 2000 and is expected to be

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

264

W. E. ALNASER, F. TRIEB AND G. KNIES

14,000MW by 2010 and 70,000MW in 2020 (UNDP, 2002). This, of course, will minimize the cost per watt for each renewable energy source, especially solar thermal and photovoltaic. The present study is the result of an analysis of the Middle East and North Africa (MENA) countries with respect to their electricity demand and the possibility to satisfy this demand by solar thermal power plants in the coming decades. Most of the findings of this paper are derived from the final report ‘Concentrating solar power for the Mediterranean Region’ (DLR, 2005). This paper describes the importance of clean and sustainable energy exports among MENA countries and Europe in the form of high voltage potential produced in Arab countries, using solar thermal (the most suitable technology for the region) or other renewable energy resources; this is, of course, after using part of this energy for water desalination in Arab countries, where 65 per cent of the water resources are politically in debate with non-Arab countries, which may ignite conflicts. It will also show how the environment will be protected against CO2 emissions if solar thermal technology is utilized.

8.2 DETERMINATION OF THE ELECTRICITY DEMAND IN MENA 8.2.1 DRIVING FORCES The scenarios for electricity demand – without electricity used for desalination – were calculated using GDP per capita and population as driving forces. The GDP per capita and population scenarios are the same as in the scenarios for the demand for desalination.

8.2.1.1 Population All scenarios share the same population prospect – the World Population Prospect by the UN, middle path, and revision 2002, including the age structure of the population. The last is not taken into account explicitly but will be used to discuss the feasibility of economic growth and what might happen if the work force and not the population would be considered as a driving force. In addition, it should be noted that according to the scenario in the next 50 years most of the states under investigation are in a demographic transformation due to decreasing birth rates, which will raise the share of the workforce in the population to a high level. At the end of the period the share of old people will increase significantly, however. Even with given wage levels, such a demographic transformation offers the opportunity to reap a socalled demographic dividend within the next 30 years when the share of working people will increase. Therefore rising income per capita and rising savings can be expected.

8.2.1.1 GDP per capita When talking about economic development, you must decide which indicator should be used. In the following Gross National Income (GNI) in Purchasing Power Parity (PPP), US$-2001 per capita will be employed. While GDP is a better measure for economic activities within a country the GNI is a better measure of the population’s income. The PPP used is calculated on the basis of a US-basket of goods and services. A regression suggests that a 1.45 per cent increase in GNI in Atlas-$ is necessary to reach a 1 per cent increase in GNI (PPP). Different functions suggest that at the lower end of the income scale a lower increase in GNI in Atlas-$ is necessary while at the upper end a higher increase is required.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

265

8.2.2 THE LINK BETWEEN GDP PER CAPITA AND ELECTRICITY CONSUMPTION PER CAPITA The essential part for deriving the electricity demand is a link between GDP per capita and electricity consumption per capita. With the population growth, the overall electricity demand can obviously be calculated. The link between the per capita values used the following steps: 1. For every year in the period 1960–2001, regressions were calculated between GDP per capita and total final consumption (TFC) for electricity per capita (IEA, 2002; Heston et al, 2002). 2 For the two parameters of the regression equations, time trends were estimated (using power and alternatively linear functions). A power function gave a significantly better fit for the absolute term. For the power it was difficult to distinguish a linear trend from a power trend. So both were used, resulting in two alternative links between GDP per capita and electricity consumption per capita. The linear trend gives a scenario with high efficiency increase while the power trend results generally in increasing electricity intensities. It should be noted that with these parameter trends, the link between GDP per capita and electricity per capita no longer depends solely on growth rates. Also the absolute value shows an influence. 3 From the TFC for electricity, the domestic demand was derived using International Energy Agency (IEA) data for 2001 (IEA, 2002) on distribution losses, consumption in the energy sector and the so-called ‘own use’. These consumptions were split into a proportional and a fixed term. The fixed term is meant to accommodate the use for oil production. The proportional term was linearly reduced to a level which is now common in industrial countries (8 per cent). It should be noted that the data on these terms are not of a high quality and are sometimes missing. Fortunately, the impact of these terms is generally small. 4 The resulting general functions were calibrated to individual countries assuming a linear mix of the current values and the estimated value. The weight of the estimated value is assumed to increase linearly from a current 0 to 1 in 2050. 5 The two scenarios are obtained by combining high economic growth with high efficiency growth and low economic growth with low efficiency growth, as the increase of efficiency is coupled to investment and the higher growth rates result in higher investment rates and a higher share of new machineries.

8.3 RENEWABLE ENERGY RESOURCES IN EU-MENA The renewable energy resources in the Euro-Mediterranean region were assessed on the basis of spatial information available from different sources described later in this chapter. The direct normal irradiance (DNI) used by concentrating solar power systems was assessed by the German Aerospace Center’s (DLR’s) high resolution satellite remote sensing system (SOLEMI, 2004), while the data for the other renewable energies were taken from materials provided by the renewable energy scientific community.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

266

W. E. ALNASER, F. TRIEB AND G. KNIES

Both the technical and economic potentials were defined for each renewable energy resource and for each country. For each resource and for each country, a performance indicator was defined that represents the average renewable energy yield with which the national potential could be exploited (Table 8.1). The economic potentials are those with a sufficiently high performance indicator that will allow new plants in the medium and long term to become competitive with other renewable and conventional power sources, considering their potential technical development and economies of scale (Table 8.2). The renewable energy potentials for power generation differ widely in the countries analysed within this study. Altogether they can cope with the growing demand of the developing economies in MENA. The economic wind, biomass, geothermal and hydropower resources combined amount to about 400TWh/y. Those resources are more or less locally concentrated and not available everywhere, but can be distributed through the electricity grid, which will be reinforced in the future in line with the growing electricity demand of this region. The greatest resource in MENA is solar irradiance, with a potential that is by several orders of magnitude larger than the total world electricity demand. The solar energy irradiated on the ground equals one to two barrels of fuel oil per square metre per year of primary energy. This magnificent resource can be used both in distributed photovoltaic systems and in large central solar thermal power stations. Thus, both distributed rural and centralized urban demand can be covered by renewable energy technologies. The accuracy of a global resource assessment of this kind cannot be better than ± 30 per cent for individual sites as it depends on many assumptions and simplifications. However, it gives a first estimate of the order of magnitude of the renewable energy treasures available in Europe and MENA.

8.3.1 RESOURCES FOR CONCENTRATING SOLAR POWER Weather satellites such as the Meteosat-7 of the European Organization for the Exploitation of Meteorological Satellites (EUMETSAT) are geo-stationary satellites at a distance of 36,000km at a fixed point over the globe that send half-hourly images for weather forecasting and other purposes. From those images, the optical thickness of clouds can be derived obtaining half-hourly cloud values for every site. Of all atmospheric components, clouds have the strongest impact on the direct irradiation intensity on the ground. Therefore the very high spatial (5 x 5km) and temporal (0.5 hour) resolution provided by Meteosat is required for this atmospheric component. Aerosols, water vapour, ozone and so forth have less impact on solar irradiation. Their atmospheric content can be derived from several orbiting satellite missions like NOAA and from re-analysis projects like GACP or NCEP/NCAR and transformed into corresponding maps/layers of their optical thickness. The spatial and temporal resolution of these data sets can be lower than that of clouds. The elevation above sea level also plays an important role as it defines the thickness of the atmosphere. It is considered by a digital elevation model with 1 x 1km spatial resolution. All layers are combined to yield the overall optical transparency of the atmosphere for every hour of the year. Knowing the extraterrestrial solar radiation intensity and the varying angle of incidence, the direct normal irradiation can be calculated for every site and for every hour of the year. Electronic maps and GIS data of the annual sum of

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

267

direct normal irradiation can now be generated, as can an hourly time series for every single site. The mean bias error of the annual sum of direct normal irradiation, which is decisive for economic assessment, is usually in the order of ± 5 per cent. More information TABLE 8.1 Renewable electricity performance indicators representing the average renewable electricity yield of a typical facility in each country

TABLE 8.2 Economic potentials of renewable energy sources in the southern EU and MENA region in TWh/y

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

268

W. E. ALNASER, F. TRIEB AND G. KNIES

FIGURE 8.2 Annual direct normal irradiance for 2002

can be found at the websites www.dlr.de/steps, www.solemi.com and http:// swera.unep.net. The analysis was performed for the countries is shown in Figure 8.2 for 2002. A oneyear basis is not sufficient for the development of large CSP projects, as the annual climatic fluctuations can be in the range of ± 15 per cent. For project development purposes, at least 5–15 years of data should be processed. However, for the assessment of national solar electricity potentials and their geographic distribution, this basis is sufficient, especially because in most MENA countries the total solar energy potential is some orders of magnitude higher than the demand.

8.3.2 OTHER RENEWABLE ENERGY RESOURCES 8.3.2.1 Hydropower The national technical and economic hydropower potentials were taken from the literature (WEC, 2004; Horlacher, 2003). The annual full load hours are used as the performance indicator. They were calculated from the installed capacity and the annual electricity generation of the plants installed at present in each country (ENERDATA, 2004). The map of gross hydropower potentials (Figure 8.3) illustrates the geographic distribution of the hydropower potentials. The total economic hydropower potential of all countries analysed within the study is 432TWh/y. In 2000 approximately 70GW of hydropower were installed, producing 155TWh/y of electricity. There is certain evidence that climate change is possibly having an increasing impact on hydropower generation with the possibility of reductions of up to 25 per cent in the long term in the southern Mediterranean countries (Bennouna, 2004; Lehner et al, 2005). Although we have not quantified such impacts in the study we believe that this is a

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

269

000 000 0000 0000 00000 GWh/y on 30 x 50 km pixel

Source: Lehner et al (2005)

FIGURE 8.3 Gross hydropower potentials in EU-MENA

serious concern that should be taken into account in energy planning. Efficiency of hydropower use should be enhanced systematically in order to, at least, partially counteract such effects.

8.3.2.2 Geothermal power Considerable conventional geothermal resources are available in Italy (already used to a great extent), Turkey and Yemen. Conventional geothermal resources were taken from the literature (Gawell et al, 2004). For Europe, medium-term geothermal power potentials from the literature were taken for cross-checking (EU, 2004). A map of subsoil temperatures at 5000m depth was taken to assess the total areas with temperatures higher than 180ϒC as the economic potential for hot dry rock (HDR) technology. It was assumed that a layer with 1km thickness at 5000m depth was used as a heat reservoir (Bruchmann, 2004; Hurter et al, 2000). The technical HDR potential for temperatures below 180ϒC was not assessed. The temperature at 5000m depth was used as the performance indicator (Figure 8.4). With that information, the efficiency d and the specific investment cost (Inv) of a HDR plant was compared with that of a reference plant and calculated according to the following equation, using a scaling exponent of 0.7: Inv = InvReference•(dReference/d)0.7

(1)

The data of the reference plant were taken from Chopra (2003).

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

270

W. E. ALNASER, F. TRIEB AND G. KNIES

Source: Wittig (2004)

FIGURE 8.4 Temperature at 5000m depth for hot dry rock geothermal power technology

The annual amount of electricity that can be generated from hot dry rock depends on the heat in place and the time of extraction. That time was assumed to be 1000 years in order to ensure that the geothermal potentials can be renewed within this time span. At such a slow rate, the geothermal power potentials can be considered as renewable energies that could be used continuously without limitations in time like other renewable energy sources. In 2000 approximately 600MW of conventional geothermal power capacity was installed in the analysed countries producing 4.6TWh/y of electricity. The total economic potential was estimated to be around 400TWh/y; this is, however, a quite rough and conservative estimate.

8.3.2.3 Electricity from biomass The electricity potential of municipal waste, solid biomass (wood) and agricultural residues was calculated. From the literature, agricultural residues such as biogases, which at present are mainly unused for power purposes, were taken as reference (WEC, 2004). An electricity conversion factor of 0.5MWh/ton of biomass was assumed for the calculation of the potential electricity yield from agricultural waste biomass. It was assumed that 80 per cent of this potential will be used in 2050. The amount of potentially available municipal waste was calculated in proportion to the growing urban population in each country. The growth of population was taken from the UN medium growth model. Due to growing urban population, the biomass potential from municipal waste grows

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

271

steadily with each year. We have assumed a constant municipal waste productivity of 0.35tons/cap/year and a waste-to-electricity conversion factor of 0.5MWh/ton. It was estimated that 80 per cent of this potential will be used until 2050. Solid biomass (mainly wood) potentials were assessed from a global map of biomass productivity in tons/ha/year and from the existing forest areas of each country (Figure 8.5). A possible change in productivity or forest areas in the future has been neglected. Results were cross-checked for plausibility with historical data from European countries (WEC, 2004). There will also be competition with traditional fuel wood use in most MENA countries which must be taken into consideration. Therefore the rate of use of the fuel wood potential was assumed to be 40 per cent until 2050. Annual full load hours are used as performance indicator.

Source: Bazilevich (1994)

FIGURE 8.5 Map of biomass productivity

The total installed capacity of biomass power plants in the analysed countries in 2000 amounted to 1.8GW; plants were generating a total of 6.4TWh/year of electricity. For the total region, a biomass electricity potential of 400TWh/y was identified, of which about 50 per cent might be used until 2050. Potential from residues dominate in MENA, while power from solid and other biomass sources is also very important in Europe.

8.3.2.4 Wind energy Wind power resources are given in the literature for European countries, including Malta and Cyprus, and for Morocco, Tunisia, Egypt and Turkey (EU, 2004; Gardner et al, 2002; OME, 2002). There is additional information on wind power potentials for Morocco, Jordan, Egypt and Turkey in GTZ (2002, 2004).

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

272

W. E. ALNASER, F. TRIEB AND G. KNIES

Source: Gra`l (2003a)

FIGURE 8.6 Annual average wind speed at 80m above ground level in m/s

For the other countries, electricity potentials were estimated from the wind map in Figure 8.6 taking into account wind speed and area restrictions and site exclusion similar to that used for CSP but adapted to wind power. The original wind speed was taken from ECMWF (2002) for 33 and 144m height and was interpolated to 80m height. This map gives a very rough estimate of the distribution of wind speed as an average for an area of 50 x 50km. The original data has a geographic resolution of 1.12 degrees. Wind electricity potentials were calculated as functions of the average wind speed according to well-known equations. We have assumed a maximum installed capacity of 10MW per square kilometre of land area. Areas with annual full load hours over 1400h/y equivalent to a capacity factor of 16 per cent were considered as long-term economic potential. Results were cross-checked and eventually corrected for those countries that have made a national resource assessment (WEC, 2004; OME, 2002; GTZ, 2004; Alnaser and Al-Karaghouli, 2000; Gra`l et al, 2003a). Annual full load hours (capacity factor) define the performance indicator. These have been derived from the literature, the World Wind Atlas (WWA, 2004) for a selection of sites in each country, and from the wind speed map. Potentials include both onshore and offshore. In 2000 a total of 3.3GW of wind capacity was installed in the analysed region producing 7.2TWh/y of wind electricity (ENERDATA, 2004). The total economic wind power potential in the region amounts to 440TWh/y, of which 285TWh/y could be exploited by 2050.

8.3.2.5 Photovoltaic energy Photovoltaic (PV) applications are in principle unlimited. There are no criteria for site exclusion for PV systems, as they can be installed almost anywhere. However, their

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

273

expansion is still limited by their high investment cost. Using present growth rates and scenarios for very large PV systems and distributed applications, PV potentials were assessed in a relatively intuitive way. For EU states, the literature gives mid-term potentials for PV (EU, 2004). The global irradiance on a surface tilted according to the latitude was used as a performance indicator (Figure 8.7) (Meteonorm, 2004). Although we have not introduced any economic threshold, the learning curves of PV suggest that this technology will become competitive by the middle of this century under the irradiance conditions of the MENA region.

Source: ECMWF (2002), Gra`l (2003a)

FIGURE 8.7 Annual global irradiation on surfaces tilted south with latitude angle in kWh/m) /year

PV systems are especially suited to decentralized small scale applications in remote regions where they often are already competitive with conventional diesel motor generator power supply schemes. This special market segment has been assessed by several studies (OME, 2002; GTZ, 2002, 2004). In our study we have only included global PV potentials without quantifying grid-connected and remote systems separately. In 2000 about 0.02TWh/y of solar electricity was produced by PV, mainly in Italy and Spain. Including very large scale PV systems (VLS-PV) of up to 1.5GW of capacity in the desert regions, by 2050, as suggested by IEA (2003), we estimate the PV potential in the region to be about 218TWh/y with a total installed capacity of 125GW.

8.3.2.6 Wave and tidal power Wave and tidal power potentials were taken from the literature (EU, 2004). Performance indicators are the annual full load hours which have been set for all locations at 4000h/y.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

274

W. E. ALNASER, F. TRIEB AND G. KNIES

More information on the equations used to estimate the potential of each type of renewable energy resource can be obtained from Alnaser (1995).

8.4 CSP DEMAND-SIDE POTENTIAL IN MENA

86 86 86 86 86 86 86 86 86 86 86 86 86 86 86 1

Acknowledgments Acknowledgments What's

According to the estimates of the reference scenario ‘Closing the Gap’ for each MENA country, assuming high economic growth rates and at the same time high efficiency gains in the power sector, the total MENA electricity demand will increase from 800TWh/y today to 1710TWh/y in 2025 and to 3605 TWh/y in 2050 (Figure 8.8). This reflects an increase by a factor of 4.5. In the coming decades, the necessity of re-investing in the power sector in MENA countries by replacing older conventional power plants with solar energy plants is the most suitable and economic decision to take. The concentrating solar power (CSP) plant is the candidate. These new plants will coincide with the increasing growth rates of the power demand – by that time the CSP capacities will still be very limited by their possible maximum speed of expansion. Although CSP will expand rapidly with growth rates of 25 to 35 per cent per year, major CSP shares of the domestic power supply will not be seen before 2025 (Figure 8.9). Very high growth rates will be achieved in the first five years. There will probably be competition between various collector designs, which will accelerate the production capacity

8686

What's What's 86 86

Export 8686 What's What's

Desalination What's What's

What's MENA What's What's

FIGURE 8.8 Domestic electricity demand in MENA and electricity supplied by existing power stations, by new power plants and by CSP for domestic consumption, export and seawater desalination for the scenario ‘Closing the Gap’

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

275

100

CSP Growth Rate [%/y]

90 80 70 60 50 40 30 20 10

20 50

20 45

20 40

20 35

20 30

20 25

20 20

20 15

20 10

20 05

20 00

0

Year FIGURE 8.9 Growth rate of CSP production in MENA in the scenario ‘Closing the Gap’ including domestic power supply, export electricity and seawater desalination

expansion. A long-term growth rate of 30 per cent, as achieved in the scenario between 2015 and 2025, is quite realistic and has already been observed in other technologies. Nevertheless, until 2025 considerable production of conventional and other renewable electricity of 1440TWh/year will have to be built in the form of new conventional power plants or other renewable power technologies such as wind, hydro or geothermal power. Once installed, this new power plant capacity will be there for a time span of about 30–40 years, until it will again be replaced. The scenario assumes that once CSP has reached appropriate production volumes, the expansion of new fossil-fuel-fired power plants will stagnate. After 2025 most new plants will be hybrid, using CSP and fossil fuels as input energies (Table 8.3). Part of the existing fossil-fuel-powered capacity could then be replaced subsequently with concentrating solar power, if the land area required for the collector fields is considered in the planning of those power plants. This additional CSP potential has been neglected except for the islands of Cyprus and Malta. In our scenario the CSP capacity of each country is proportional to each country’s share of the total MENA power deficit. Countries showing larger deficits will also require larger CSP capacities. The total supply of electricity by CSP in MENA will reach 235TWh/y by 2025 and 2045TWh/y by 2050. It is assumed that the installed CSP capacities would be distributed evenly among the MENA countries. In the real world, there might be competition among the MENA countries to achieve higher growth rates within each individual country. A strong competition among several CSP technologies could lead to higher overall CSP growth rates than those assumed here (around 30 per cent per year). In 2025 14 per cent of the domestic electricity demand of

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

276

W. E. ALNASER, F. TRIEB AND G. KNIES

TABLE 8.3 Electricity demand in MENA and generation by conventional old and new plants and by CSP technologies in MENA until 2050 (all numbers rounded to 5) YEAR Old Plants New Plants CSP Supply MENA Electricity Demand CSP Desalination CSP Export Total CSP in MENA

TWh/y TWh/y TWh/y TWh/y TWh/y TWh/y TWh/y

2000 785

– – 785

– – –

2025 40 1440 235 1710 115 45 395

2050 0 1560 2045 3605 525 645 3215

MENA will be covered by CSP, which will become the dominating power source with a share of 57 per cent by 2050. The necessary solar capacities for seawater desalination are quantified in section 8.7. A CSP production of 115TWh/y in 2025 and 525TWh/y in 2050 would be used for desalination purposes. After 2030, the CSP desalination capacities would be large enough to cope with the freshwater demand and desalination will grow much more slowly. While in 2025 about 29 per cent of the total CSP production would be used for desalination, in 2050 only 16 per cent would be used for that purpose. Export of solar electricity will be limited to minor volumes before the scheduled finalization of the Mediterranean Ring Interconnection in 2015. Also after that, additional transmission capacities will be required to export considerable quantities of electricity from MENA to Europe, as the Mediterranean Ring transmission capacity will be limited to about 500 to 1000MW. The scenario assumes that export through additional high voltage direct current (HVDC) interconnection of North Africa and Europe will start in 2015 with 2TWh/y. An extended HVDC grid including other MENA regions will transmit 45TWh/y by 2025 and 645TWh/y by 2050, which would be equivalent to about 15 per cent of the European electricity demand of that time. In the scenario, export capacities are distributed to all MENA countries in proportion to their share of the total MENA domestic power demand. The idea behind this is that if a country installs high capacities for domestic supply, it will also have a good industrial base for export. However, there are different priorities for electricity export and different starting times for interconnections, depending on the distance to Europe, the solar energy resources and the priority of the domestic demand for power and water. By 2025, 11 per cent of the total CSP production of MENA will be exported, increasing to about 20 per cent by 2050. The islands are not considered as potential CSP exporters in this scenario. On the other hand, the electricity interconnection between the Gulf Cooperation Council countries (southwest Asia – Saudi Arabia (SA), Kuwait (KU), Qatar (QR), United Arab Emirates (UAE), Bahrain (BN) and Oman (ON)) has already been agreed on, with a budget of US$1.2 billion; it started in early 2005 and will be complete in 2008. The scenario is a rough estimate of the CSP potential in MENA. There will be three types of plants distributed among all three CSP categories (domestic, export and desalination) used in different combinations:

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

277

1 CSP plants for cogeneration with coupled seawater desalination must be at the coast, as the cogenerated heat cannot be transferred over long distances. Their electricity can be used for additional reverse osmosis (RO) desalination, domestic use or export. As the coastal regions in MENA are strongly used by other human activities, this plant type will be limited to regions with appropriate site conditions and available land area. 2 CSP plants used exclusively for power generation can be anywhere on the grid. Their electricity can be transmitted to any other place and used for domestic supply, export or RO desalination. This type of plant will be placed where good irradiation coincides with good infrastructure conditions. 3 CSP plants for industrial cogeneration will be limited to appropriate industrial sites. While their heat will be used on site, their electricity might be used on site too or be sold to the grid for domestic use, export or RO desalination. Cogeneration plants are considered as part of the domestic CSP production potential. In the real world, there will be a mix of these three plant types, which will vary according to the regional demand of each country and the local supply-side conditions. The scenario can only give a rough estimate of the overall potentials of the region, showing the amounts of energy potentially used for domestic supply, export or desalination. However, it cannot distinguish and quantify the different plant types that will be erected in each country, which will be subject to national strategic power expansion planning.

8.4.1 CSP DEMAND-SIDE POTENTIAL IN NORTH AFRICA According to the reference scenario ‘Closing the Gap’, the domestic North African electricity demand will increase from 160TWh/y today to 490TWh/y in 2025 and to 1230TWh/y in 2050 (Figure 8.10). This reflects an increase by a factor of 7.7, which is almost twice the MENA average. The export share of the Northern African countries is assumed to become relatively large, due to their proximity to Europe. Export will start in 2015, achieve 20TWh/y, equivalent to 14 per cent of the total CSP production, by 2025 and 385TWh/y, equivalent to 25 per cent of the total CSP production, by 2050. Due to the considerable growth of demand, domestic electricity production will be the dominant potential for CSP in North Africa, with equally important short-term potentials for desalination and considerable long-term potentials for export electricity. In 2025 15 per cent of the domestic electricity demand will be covered by CSP, which will grow to become the dominating electricity source with a share of 65 per cent in 2050 (Table 8.4). On the basis of country statistics from FAO (2004), a considerable demand for solar powered desalination can only be detected for Libya and Egypt. However, the desalination potential of those two countries alone accounts for 38 per cent of the total CSP production in North Africa in 2025 and 20 per cent in 2050. CSP export of the western Asian countries is of relative low priority because the distance to Europe is quite large for most countries. However, land and solar radiation resources are available as in Northern Africa. Export may start in 2025 with approximately 10TWh/y, equivalent to 8 per cent of the total CSP production, and achieve 120TWh/y, equivalent to 19 per cent of the total CSP production in 2050.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

278

W. E. ALNASER, F. TRIEB AND G. KNIES

TABLE 8.4 Domestic electricity demand in North Africa and generation of domestic power, export electricity and power for desalination by CSP technologies until 2050 YEAR Old Plants New Plants CSP Supply North Africa Demand CSP Desalination CSP Export Total CSP in North Africa

TWh/y TWh/y TWh/y TWh/y TWh/y TWh/y TWh/y

2000 160

– – 160

– – –-

2025 10 415 70 490 55 20 145

2050 0 430 800 1230 285 385 1470

FIGURE 8.10 Domestic electricity demand in North Africa and electricity supplied by old power stations, by new power plants and by CSP for domestic consumption, export and seawater desalination for the scenario ‘Closing the Gap’

The interconnection of electricity between European and MENA countries will resemble that interconnecting the seven Mediterranean states (Spain, Italy, Morocco, Algeria, Libya and Egypt) through the UCTE (Union of Coordination for Transmission of Electricity) which is supervising projects in more than 20 European countries providing more than 40 million consumers with 2100TWh. There is also an interconnection of electricity between African states which link Egypt to Sudan and to the Congo. This link will be illustrated later.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

279

8.4.2 CSP DEMAND-SIDE POTENTIAL IN WESTERN ASIA The domestic electricity demand in western Asia will increase from 375TWh/y today to 810TWh/y in 2025 and 1590TWh/y in 2050 (Figure 8.11 and Table 8.5). This reflects an increase by a factor of 4.2, which is about the MENA average. CSP export of the western Asian countries is of relative high priority. However, land and solar radiation resources are more limited than in Northern Africa and the distance to Europe is greater. Export may start in 2020, achieve 10TWh/y, which is equivalent to 8 per cent of total CSP production, in 2025 and 145TWh/y, equivalent to 13 per cent of the total CSP production, in 2050. Export volumes of CSP from western Asia are smaller than those of North Africa, as

FIGURE 8.11 Electricity demand in western Asia and electricity supplied by old power stations, by new power plants and by CSP for domestic consumption, export and seawater desalination for the scenario ‘Closing the Gap’

TABLE 8.5 Domestic electricity demand in western Asia and generation of domestic power, export electricity and power for desalination by CSP technologies until 2050 YEAR Old Plants New Plants CSP Supply Western Asia Demand CSP Desalination CSP Export Total CSP in Western Asia

TWh/y TWh/y TWh/y TWh/y TWh/y TWh/y TWh/y

2000 375

– – 375

– – –

2025 20 680 110 810 10 10 130

2050 0 685 905 1590 60 145 1110

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

280

W. E. ALNASER, F. TRIEB AND G. KNIES

the site conditions in western Asia are not as conducive as North Africa and the domestic demand in western Asia is very large and will probably be binding the capacities. On the basis of country statistics from the FAO (2004), a considerable demand for solar powered desalination can only be detected for Jordan, Israel and Syria. The potential for desalination will achieve 8 per cent of the total CSP production in 2025 and 5 per cent in 2050. Due to the considerable growth of demand, domestic electricity production will be the dominant potential for CSP in western Asia, with comparatively small potentials for desalination and for export electricity. In 2025 14 per cent of the domestic electricity demand will be covered by CSP, which will grow to become the dominate electricity source with a share of 57 per cent in 2050.

8.4.3 CSP DEMAND-SIDE POTENTIAL OF THE ARABIAN PENINSULA The domestic electricity demand of the Arabian Peninsula will increase from 235TWh/y today to 410TWh/y in 2025 and 790TWh/y in 2050 (Figure 8.12 and Table 8.6). This reflects an increase by a factor of 3.4, which is well below the MENA average. The low value is due to an already rather high consumption of electricity per capita in most countries of the peninsula except Yemen and Oman. On the basis of country statistics from the FAO (2004), all countries of the Arabian Peninsula have a large freshwater deficit and hence a large demand for solar powered desalination. The potential for desalination will achieve 43 per cent of the total CSP production in 2025 and will still make up 28 per cent in 2050.

FIGURE 8.12 Electricity demand of the Arabian Peninsula and electricity supplied by old power stations, by new power plants and by CSP for domestic consumption, export and seawater desalination for the scenario ‘Closing the Gap’

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

281

TABLE 8.6 Domestic electricity demand of the Arabian Peninsula and generation of domestic power, export electricity and power for desalination by CSP technologies until 2050 (rounded to 5) YEAR Old Plants New Plants CSP Supply Demand Arabian Peninsula CSP Desalination CSP Export Total CSP Arabian Peninsula

TWh/y TWh/y TWh/y TWh/y TWh/y TWh/y TWh/y

2000 235

– – 235

– – –

2025 10 345 55 410 50 10 115

2050 0 445 345 790 180 120 645

Due to the considerable growth of demand, especially in Saudi Arabia and Yemen, domestic electricity production will be the dominant potential for CSP on the Arabian Peninsula, followed by a large potential for desalination and comparatively small potentials for export. In 2025 14 per cent of the domestic electricity demand will be covered by CSP, which will grow to become a very important electricity source with a share of 44 per cent in 2050.

8.4.4 SUMMARY From this analysis we can conclude that there are several consequences related to the scenario ‘Closing the Gap’: ● Until 2025, the CSP market in MENA will be limited neither by the available solar

energy resources nor by the expected energy demand, but only by the intrinsic achievable speed of CSP capacity expansion. ● Under the assumptions of the scenario ‘Closing the Gap’, CSP would cover about 60 per cent of the expected electricity demand for power and water in MENA in 2050, the rest would have to be covered by conventional power technologies using fossil fuels. In spite of the considerable expansion of CSP and other renewable energies, CO2 emissions in MENA would still double by 2050. However, without CSP CO2 emissions in 2050 would become roughly five times those of 2000. ● Economic growth and a reasonable growth of the population do not necessarily lead to an excessive consumption of natural resources if sustainable pathways are taken immediately. This includes the intensification of energy and water efficiency and the use of all available renewable energy technologies. ● Due to the abandoning of CSP in the beginning of the 1990s in California, about 15 years of development, learning and expansion have been lost. These missing 15 years will lead to a considerable gap in the energy and water supply in MENA during the 2020s, which must be bridged by non-sustainable resources like fossil fuels and groundwater reserves, hoping that those resources will be able to carry on long enough. An expansion of CSP in time could have avoided those gaps. Now, every year of delay of the introduction and expansion of CSP will lead to more critical

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

282

W. E. ALNASER, F. TRIEB AND G. KNIES

situations in the MENA countries. The cost of CSP varies widely. It depends on the producer and consumer ability. Fully exporting this technology, relying fully on expatriates and foreign manufacturing and services of CSP will make the cost twice that found in our scenario ● Export of solar electricity and of CSP technology can become an important economic factor in those MENA countries that will be able and willing to use this new, clean and everlasting natural resource to increase their national income, while their income from fossil reserves will inevitably decrease.

8.5 THE CRITICAL ISSUE OF WATER AND ENERGY IN THE GCC COUNTRIES The majority of the population in the Arab world is concentrated in urban areas. In the Arabian Gulf countries, which are the poorest in water resources, approximately 15.7 per cent of its people inhabit rural areas with no access, or very limited access, to suitable water and energy resources (figures range from 22.3 per cent in Oman to 16.5 per cent in Saudi Arabia, 8.4 per cent in Qatar, 2.5 per cent in Kuwait and probably 1.5 per cent in Bahrain (ESCWA, 2001). These countries are the poorest in natural water resources. In 2000 Oman had only 320m3/capita/year, Bahrain 170m3, Saudi Arabia 100m3, Kuwait 20m3 and Qatar 10m3. Given that the poverty level is 500m3/capita/year and the critical level is 100m3, one can see how critical the situation is. Freshwater scarcity has been a serious challenge and freshwater supplies from conventional sources are unable to meet basic demands. The desalination option continues to be favoured. With desalination, which consumes power derived from oil and natural gas, the annual per capita water consumption in the GCC in 2000 had become reasonable (Table 8.7, including the projected demands for 2010). The GCC countries are, generally, the wealthiest in thermal heat, which can be used for desalination, either directly using trough solar concentrating or solar tower or solar dish technology, or possibly photovoltaic technology, to run the pumping of the reverse osmosis system. Table 8.8 shows the solar potential in these countries.

TABLE 8.7 Annual per capita water consumption in 2000 and projected demand in 2010 for the GCC countries – The poorest in natural water resources COUNTRY

Bahrain Kuwait Oman Qatar Saudi Arabia United Arab Emirates Total

DOMESTIC DEMAND (m3) 2000 2010 237 279 308 302 54 76 258 296 140 137 340 352 1337 1442

AGRICULTURAL DEMAND (m3) 2000 2010 197 215 64 60 467 417 341 349 978 880 573 638 2622 2559

Source: Dabbagh (1995)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

TOTAL DEMAND (m3) 2000 2010 436 494 372 362 521 493 599 645 1118 1017 913 990 3959 4001

Solar Energy Technology in the Middle East and North Africa

283

TABLE 8.8 Solar resources in the GCC countries COUNTRY Bahrain Kuwait Oman Qatar Saudi Arabia United Arab Emirates

GLOBAL SOLAR RADIATION (kWh/m3/day) 6.4 6.2 5.1 5.5 7.0 6.5

DIRECT NORMAL SOLAR RADIATION (kWh/m3/day) 6.5 6.5 6.2 5.6 6.5 6.0

Source: Global Emission Newsletter (2004)

With relatively large land area available in these countries, it is possible to utilize the solar power to produce enough water in the GCC (and other Arab states), while the electricity can be transmitted to Europe, which suffers from heavy air pollution due to burning conventional fuel, especially in those countries with heavy industry and advanced technology. Cooperation between MENA and Europe is indeed needed. The investment in solar water desalination may be very successful. According to certain newspaper reports, the GCC is planning to spend more than US$40 billion. Table 8.9 shows the useful electrical equivalent per m3 of distillate (UEPCMD) in kWh, and the specific electrical energy input per m3 of distillate (SEEPCMD) in kWh/m3 for typical multi-stage flash (MSF), multiple effect distillation (MED), thermal vapour compression (TVC), mechanical vapour compression (MVC), reverse osmosis (RO) and electrolysis (ED). TABLE 8.9 Comparison of energy consumption for six types of water desalination DESALINATION TYPE UEPCMD(kWh) SEEPCMD(kWh/m3)

MSF 16 4

MED 11.5 2.0

TVC 14.56 1.7

MVC – 11.5

RO – 1.5

ED – 4

Source: Jawad (2001); ESCWA (2003)

The future of the electrical power transmission between different countries is affordable. If this happens, then there will be no problem at all in producing electricity and water using solar energy systems, such as the concentrating solar power CSP for desalination or by photovoltaic. We have to bear in mind that, today, the cost of water desalination using fossil fuel ranges from US$0.5/m3 (reverse osmosis) to US$1/m3 (thermal desalination methods). According to the ESCWA (2003) report, MSF desalination plants and RO systems are the most predominant installed desalination plants in the GCC countries. These systems achieve 7.859 and 1.811 million m3 per day respectively. VC systems achieve 19,469m3 per day, ED 64,424m3 per day. The minimum installed capacity for ME is only 21,204m3 per day (Table 8.10). In conclusion: ● the GCC countries have the largest installed MSF desalination capacities in the world,

accounting for 80 per cent of the world total of 9.8 million m3 per day;

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

284

W. E. ALNASER, F. TRIEB AND G. KNIES

TABLE 8.10 The available desalination capacity in the GCC countries, in 2000, using different desalination technologies, in cubic metres per day DESALINATION TYPE FEED WATER RO

MSF

VC

MED

ED

Brackish Sea Waste and others Total Brackish Sea Total Brackish Sea Waste Total Brackish Brine Sea Total Brackish Total

Total

BAHRAIN KUWAIT

GCC COUNTRIES OMAN QATAR

SAUDI ARABIA

65,504 48,217

44,103 3000

5151 13,840

4154 5000

974,201 444,363

–

–

–

1130

70,133

113,721 47,103 – – 297,202 1,468,036 297,202 1,468,036 – – 1635 – – – 1635 – 1135 – – 1272 – 1200 1135 2470 3350 2418 3350 2418 416,861 1,520,029

UNITED ARAB EMIRATES 73,969 58,344

TOTAL BY TECHNOLOGY (m3/day) 1,167,082 572,764

–

71,263

18,991 10,284 1,488,697 132,313 – – 1300 – 161,015 499,954 3,491,385 1,940,596 161,015 499,954 3,492,685 1,940,596 1515 – 6362 – 1600 19,590 59,012 103,735 – – 1242 – 3115 19,590 66,616 103,735 – – – 500 – – – – 3000 2542 6900 4655 3000 2542 6900 5155 – – 56,156 2500 – – 56,156 2500 186,121 532,370 5,111,054 2,184,299

1,811,109 1300 7,858,006 7,859,306 7877 185,572 1242 194,691 1635 1262 18,297 21,204 64,424 64,424

Source: ESCWA (2001)

● the GCC countries have 9.35 million m3 per day or some 19.4 per cent of the total RO

world capacity; and ● the world total of ED, ME and VC technologies in 2000 was 2.95 million m3 per day

(280,319m3 per day for GCC countries, representing only 9.5 per cent of the world total) (ESCWA, 2003).

8.6 WATER RESOURCES AND WATER DEMAND IN MENA The analysis leads to a scenario for the prediction of the demand and the resources of freshwater at the country level. Inside a country, there might be regions with deficits that cannot be identified on the basis of countrywide statistical data. The analysis of Spain or Italy at that level would not yield any deficits. Excessive withdrawal of groundwater is also a common problem in many regions. This study concentrates on those cases that can be identified as problematic on the basis of national statistics. Sub-national demand for nonconventional freshwater resources is not considered. Most of the actual data on water resources and use has been obtained from the AQUASTAT database of the Food and Agriculture Organization of the United Nations (FAO) (FAO, 2004; AQUASTAT, 2004). Extrapolations have been made on the basis of population and GDP growth rate expectations as described in this report. ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

285

The extrapolation of future water demand at country level is based on the assumptions that: ● Agricultural production and its water demand per capita will be maintained as today.

This means that the demand of the agricultural sector will be growing proportionally to population. ● The demand of the domestic and industrial sector will grow proportionally to the GDP, which is calculated for every country adding the population growth rate to the per capita GDP growth rate. ● The efficiency of water use in the agricultural and municipal sector will be increased from today’s country-specific values to a maximum value which depends on the selected scenario, the water demand growth rate thus becoming lower than the population or GDP growth rates. Enhanced technologies will additionally de-couple water demand from economic growth as experienced in Australia in recent decades (AQUASTAT, 2004; Gleick, 1998). In the analysis of the power sector, two different economic scenarios have been used to determine baselines for water demand predictions: 1 The scenario ‘Following Up’ assumes an average per capita GDP growth rate of only 1.2 per cent for every country from today until 2050. This implies that the relative distance between the actual GDP/capita (US$-PPP) of the respective country and the US will remain constant because the GDP of the US at the same time will also be growing by 1.2 per cent. Efficiencies of the agricultural and the municipal water supply system and the reuse of wastewater increase gradually from the present national performance values to a future better value of an enhanced system. However, the efficiency enhancements are limited by the slow economic development. Population growth and the agricultural sector dominate the water demand growth rates in this case. Decoupling of the water demand from the economic growth by using enhanced water supply technologies is also limited in this scenario (Gleick, 1998; Heston et al, 2002). 2 The scenario ‘Closing the Gap’ assumes that the relative distance between the actual GDP/capita (US$-PPP) in the US and the respective country is reduced to 50 per cent by 2050 while the GDP of the US at the same time is growing by 1.2 per cent. This scenario assumes that the MENA countries will by 2050 achieve GDP per capita values close to that of the European countries. In this case, the industrial and domestic sectors will dominate the water demand growth. However, efficiencies will also be increased and a significant de-coupling of water demand and economic growth as experienced in Australia in the past decades will take place. The water demand in the MENA region today is made up of 85 per cent agricultural use, 9 per cent domestic use and 6 per cent industrial use. The future demand is calculated individually for each country and aggregated to the regions of North Africa, western Asia and the Arabian Peninsula. Under the assumptions of the scenario ‘Following Up’, the share of agricultural water use will fall to about 80 per cent and the domestic and industrial share will increase to 12 ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

286

W. E. ALNASER, F. TRIEB AND G. KNIES

per cent and 8 per cent respectively. The total water demand will increase from 300 billion m3/y today to about 510 billion m3/y in 2050. The scenario reflects the influence of enhanced water management, policies and efficiencies that are of highest priority for a sustainable water future in MENA, but that are limited by the slow economic growth within this scenario. Under the assumptions of the scenario ‘Closing the Gap’, the share of agricultural water use will fall to about 66 per cent and the domestic and industrial share will increase to 18 per cent and 16 per cent respectively, more and more dominating the water demand. The total water demand will increase from 300 billion m3/y today to about 540 billion m3/y in 2050 (Figure 8.13). The scenario also reflects the pronounced influence of enhanced water management, policies and efficiencies, giving them highest priority for a sustainable water future in MENA, especially since this scenario is oriented towards a high economic growth. In terms of water demand, both scenarios are rather optimistic compared to other scenarios that predict a doubling of demand for 2025. This is determined by extrapolating

600

Freshwater Demand [billion m³/y]

500 Industrial Domestic

400

Sus tainable Water

300 200

Agricultural

100 0

00 20

05 20

10 20

15 20

20 20

30 25 20 20 Year

35 20

40 20

45 20

50 20

Note: The white line indicates the sustainable renewable freshwater resources.

FIGURE 8.13 Water demand structure in MENA and its evolution until 2050; scenario ‘Closing the Gap’

the water demand growth rates as experienced in the last decades. However, we believe that a reduction of the agricultural sector and the successive de-coupling of economic growth and water demand are more realistic approaches. At first glance, it is surprising that both scenarios culminate in a rather similar water demand of 510 to 540 billion m3/y by 2050, which obviously will be achieved with or without economic growth. It reflects the positive impact of economic stability and development on the water supply. In the scenario ‘Following Up’, consumption is limited by availability, while in the scenario ‘Closing the Gap’, it is rather limited by the enhanced efficiency of the supply system.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

287

As the future deficits and the additional demand for non-conventional resources will not change considerably assuming one scenario or the other, the scenario ‘Closing the Gap’ (which is more desirable from the point of view of the MENA countries) will be used as reference in further analyses. Western Asia still has large sustainable water resources that will be increasingly exploited in the future. However, even in this region, non-sustainable use as from fossil fuelled desalination and from unsustainable groundwater withdrawal is already experienced on a local level and shows an increasing trend in the future. Unsustainable water supply from fossil-fuelled desalination and from excessive groundwater withdrawal is considered as potential future deficits (Figure 8.14). The sustainable freshwater resources of North Africa are being used today almost to their limits, and therefore no considerable increase of their exploitation can be expected for the future. Unsustainable use from fossil desalination and from excessive groundwater withdrawal is already taking place to a considerable extent, with a dramatic increase of this situation ahead. On the Arabian Peninsula, the relation of sustainable and unsustainable use of water is even more dramatic. The total annual water deficits in MENA will increase from 35 billion m3/y today, at present supplied by excessive groundwater withdrawals and fossil-fuelled desalination, to about 155 billion m3/y by 2050. There is no sustainable resource in sight to supply such deficits except renewable energies. The cost of fossil fuels is already today too high (US$50 in August 2004) for intensive seawater desalination and its volatility and the fact that fossil fuels are limited in time eliminate fossil fuels as a resource for sustainable water security in MENA. Nuclear power is a very limited and costly resource, and in addition faces unsolved problems such as nuclear waste disposal, proliferation and other serious security issues.

FIGURE 8.14 Water supply from sustainable sources and deficits in MENA (Closing the Gap)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

288

W. E. ALNASER, F. TRIEB AND G. KNIES

(a)

Consumption Growth Rate [%/y]

The water demand growth rates will decline in all three MENA regions from about 1.5 to less than 1 per cent per year. The per capita water demand and its future trend is different in the three regions (Figures 8.15a and 8.15b). The MENA average per capita demand is expected to stay almost constant at about 800m3/capita/y. Western Asia will reduce its per capita demand from 1000 to about 900m3/capita/y, while the demand in North Africa will grow from 700 to about 800m3/capita/y, due to a relatively moderate growth of the population and an increasing importance of the domestic and industrial sector, mainly in Egypt. The specific consumption on the Arabian Peninsula will fall from 600m3/capita/y (currently) to about 400m3/capita/y in 2050 due to a strong growth of the

3.0 2.5 2.0

North Africa Western Asia Arabian Peninsula Total MENA

1.5 1.0 0.5 0.0

00 05 10 15 20 25 30 35 40 45 50 20 20 20 20 20 20 20 20 20 20 20 Year

1400

Consumption [m³/capita/y]

(b)

1200 1000 North Africa Western Asia Arabian Peninsula Total MENA

800 600 400 200 0 00 05 10 15 20 25 30 35 40 45 50 20 20 20 20 20 20 20 20 20 20 20

Year FIGURE 8.15 a) Water consumption growth rates in MENA (Closing the Gap); b) Water consumption per capita in MENA (Closing the Gap)

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

289

population and the persisting importance of the agricultural sector, coupled with very limited natural water resources.

8.6.1 NORTH AFRICA

Consumption [billion m³/y]

The scenario assumptions lead to a linear growth of the water demand in North Africa from 100 billion m3/y today to 200 billion m3/y in 2050 (Figure 8.16). The reduction in the agricultural sector is compensated by the growth of the domestic and industrial sectors. Sustainable sources in North Africa cannot be exploited to a greater extent than today. All countries will experience growing deficits, with Egypt being by far the dominating case, due a very strong agricultural sector and large population, followed by Libya and the Maghreb countries. The deficit of Egypt expected for 2050 might rise to the present water capacity of the Nile River of about 70 billion m3/y. An official expectation of a deficit of 35 billion m3/y until 2025 was recently published. All countries will experience a reduction of their water demand growth rates of about 0.5 per cent by 2050. The per capita demand is highest in Egypt and Libya (about 250 200 Industrial Domestic Agricultural

150 100 50 0

00

20

05

20

10

20

15

20

20

20

25

20

30

20

35

20

40

20

45

20

50

20

Year

FIGURE 8.16 Water demand structure in North Africa and its evolution until 2050

1000m3/cap/y), and lowest in Algeria and Malta (200m3/cap/y), with a slightly increasing trend in all countries. The strong economic growth of the scenario ‘Closing the Gap’ reveals the challenge of this path, as the water demand of the industrial and domestic sector will grow very quickly and overcompensate for possible reductions in the agricultural sector.

8.6.2 WESTERN ASIA The water demand in western Asia will increase from 175 billion m3/y today to about 275 billion m3/y in 2050, showing a slight stabilization trend by that time (Figure 8.17). There are vast sustainable water resources in that region which will be increasingly exploited in the future. However, local deficits will occur in Syria, Jordan and Israel and later also in Iraq.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

W. E. ALNASER, F. TRIEB AND G. KNIES

Consumption [billion m³/y]

290

300 250 200

Industrial Domestic Agricultural

150 100 50 0 20

00

20

05

20

10

20

15

20

20

20

25

20

30

20

35

20

40

20

45

20

50

Year FIGURE 8.17 Water demand structure in western Asia and its evolution until 2050

The demand growth rates are high in Jordan but at a very low level of per capita demand, as can be appreciated from Figure 8.20. Strong consumers are Iraq, Turkey and Syria, with only Syria facing a short-term deficit. The average per capita demand of the western Asian region will be slightly reduced from 950 to 850m3/cap/y, while in all countries the consumption growth rates will be reduced.

8.6.3 THE ARABIAN PENINSULA The Arabian Peninsula is characterized by a strongly growing population and a dominating agricultural sector. The demand will increase from 30 to 65 billion m3/y (Figure 8.18). The region’s water demand is dominated by Saudi Arabia and Yemen, both relying to a great extent on non-sustainable sources, including fossil-fuelled desalination and excessive groundwater withdrawal. Due to the combination of high population and high dependency on agriculture, both countries will be facing considerable deficits if their water supply is persistently based on the limited resources of fossil fuels and non-renewable groundwater. The sustainable natural resources of this region are very limited. The average per capita consumption on the Arabian Peninsula will be reduced from 600 to 450 m3/cap/y. Saudi Arabia and UAE will have the highest consumption per capita (about 700–800m3/cap/y). The strongest decrease of per capita consumption will be experienced in Yemen. In terms of population growth and share of the agricultural sector, Yemen is a very specific case among the MENA countries. The per capita consumption will decrease from 400 to 250m3/cap/year, but the consumption growth rates will not decrease until after 2030. The scenario ‘Closing the Gap’ would require a continuous GDP growth rate of Yemen of 11 per cent until 2050 (a necessary 7.8 per cent per capita growth rate to close the GDP per capita gap with the US plus a 3.2 per cent population growth rate), which is rather unrealistic.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Consumption [billion m³/y]

Solar Energy Technology in the Middle East and North Africa

291

70 60 50 Industrial Domestic Agricultural

40 30 20 10 0

20

00

20

05

20

10

20

15

20

20

20

25

20

30

20

35

20

40

20

45

20

50

Year FIGURE 8.18 Water demand structure for Arabian Peninsula and its evolution until 2050

8.7 THE POTENTIAL FOR DESALINATION BY CONCENTRATING SOLAR POWER The previous analysis of water deficits in MENA showed that there is a pressing need for new, non-conventional, sustainable water sources in many countries of this region. The hot spots can be found in North Africa (mainly Egypt and Libya) and the Arabian Peninsula (mainly Yemen and Saudi Arabia), while the situation is by far less critical in the countries of western Asia. However, Syria, Jordan and Israel also face considerable deficits. Although the demand of the agricultural sector, which in MENA makes up 85 per cent of the total, will not grow as fast as in the past decades. This will be compensated for by a quickly growing demand of the urban centres and industry (FAO, 2003). The use of water is today heavily subsidized in many MENA countries (Al-Zubari, 2002). This reflects the fact that the cost of supplying water is already too high today considering the per capita income level, especially in the agricultural sector. Today, the cost of desalinating water using fossil fuels ranges between US$0.5/m3 (reverse osmosis) and US$1/m3 (thermal desalination methods), which would be higher than the prices paid for water in most MENA countries. Economies building their water supply to a great extent on desalination with fossil fuels would suffer from additional subsidy loads, from the volatility of fossil fuel costs and from the gradual depletion and cost escalation of fossil energy resources. A severe stagnation of investments in the water sector is the consequence of this situation, the total water sector becoming more and more dependent on national and international subsidization. Today, many countries try to avoid an increasing dependency on desalination and fossil fuels by exploiting their groundwater resources. However, in many countries the exploitation rate is much higher than the renewable groundwater resources (Figure 8.19), making this solution not more sustainable than the dependency on fossil fuels. Therefore,

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

292

W. E. ALNASER, F. TRIEB AND G. KNIES

Saudi Arabia

1456%

Kuwait

1275%

Qatar Gaza Bahrain Jordan Yemen Syria UAE Oman Algeria Tunisia Morocco Cyprus 0%

50%

100%

150%

200%

250%

300%

350%

400%

Source: Saghir (2000)

FIGURE 8.19 Groundwater withdrawals as percentage of safe yield for selected countries

a renewable, sustainable freshwater source with much lower and more stable costs than fossil fuels is required. In the present study, we have assumed that the unsustainable water supplied by groundwater depletion and fossil-fuelled desalination represents a potential future deficit that could be covered by concentrating solar power (CSP) plants in cogeneration with multi-effect desalination (MED) and additionally using the remaining electricity for reverse osmosis (RO) desalination. To estimate the minimum CSP capacity potential in the water supply sector, we have assumed that all plants would be coupled to MED desalination plants, while the electricity generated is completely used for RO desalination in order to produce larger amounts of desalinated water. In view of the quick increase of future deficits of water in MENA, this could become necessary in order to substitute for the unsustainable water supply more quickly. This approach leads to the minimum installed capacity of CSP that is necessary to cover the future water deficits in MENA. The capacity potential for CSP would, in reality, be higher, as part of the plants would be only used for cogeneration of city power and MED desalination, but without RO desalination. The installation of such plants would be limited to the coast. Another part would only be used for power generation for RO, but without making use of cogeneration

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

293

with MED plants. Those CSP plants could be anywhere on the grid, while only the RO desalination plant must be located at the coast. To this capacity, the potential of CSP electricity generating and cogeneration plants outside the water sector may be added. This is described in other chapters of this study.

8.7.1 POTENTIAL FOR CSP DESALINATION IN MENA Today, 35 billion m3/y of the water consumption in MENA are covered by non-sustainable water sources. According to the reference scenario ‘Closing the Gap’, this deficit will increase to about 155 billion m3/y by 2050 (Figures 8. 20 and 8.21). The figures show how these deficits could be subsequently covered by CSP desalination plants, reducing the non-sustainable water supply and providing most of the non-conventional water by 2030 and afterwards using solar energy. Deficits will have to be bridged by fossil-fuelled desalination and groundwater withdrawals, hoping that those resources will remain available and affordable until then. This may seem optimistic, but there are no sustainable and affordable alternatives. It is a reassuring fact that the potential of CSP is neither limited by the solar energy resource nor by its cost, but only by the possible speed of CSP capacity expansion, and that there is a solution for the freshwater deficits in MENA that can be realized by 2030. However, a lot of time has been unnecessarily lost, leading to a considerable increase of non-sustainable water until 2030. This calls for the intensive additional use of other renewable sources like geothermal and wind power for non-conventional water production, and also calls for an intensive freshwater management programme with urban and rural applications. Only a decided employment and efficient combination of all possible measures will lead to satisfactory and sustainable water supply security in MENA. The capacity of CSP plants until 2050, if installed exclusively for seawater

Water Demand and Supply billion m³/y

600 500 400 CSP-Desalination Non-Sustainable Supply Sustainable Supply

300 200 100

50

45

20

40

20

30

35

20

20

25

20

20

20

15

20

10

20

05

20

20

20

00

0

Year FIGURE 8.20 Water demand, sustainable freshwater resources, non-sustainable supply and potential future supply by CSP via cogeneration by MED plus direct generation via solar electricity in RO plants

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

294

W. E. ALNASER, F. TRIEB AND G. KNIES

FIGURE 8.21 Covering the future freshwater deficits in MENA by transitory non-sustainable sources and by CSP plants using MED in cogeneration plus solar electricity for RO

desalination, could amount to a total of 67GW. North Africa (35GW) has the largest potential for CSP desalination plants, followed by the Arabian Peninsula (24GW) and the western Asian countries (8GW) (Figure 8.22).

FIGURE 8.22 Capacity potential for CSP-Desalination plants with MED and RO in the three main MENA regions

8.7.1.1 North Africa The deficit in North Africa will grow from 15 billion m3/y in 2000 to 80 billion m3/y in 2050 with a major share taken by Egypt. The CSP capacity potential for desalination amounts to 30GW for Egypt and 4GW for Libya, while the other countries have minor shares. On

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

295

the basis of country statistics, no potential can be detected for Morocco, Malta and Tunisia, although there may be deficits at the local level.

8.7.1.2 Western Asia The deficit in western Asia will grow from 0.5 billion m3/y in 2000 to 20 billion m3/y in 2050 with a major share taken by Syria. After 2040, Iraq will be part of the deficit mix. The CSP capacity potential for desalination amounts to 10GW for Syria and 1GW each for Israel and Jordan. The other countries have minor shares. On the basis of country statistics, no potential can be detected for Cyprus, Lebanon, Turkey and Iran, although there may be deficits at the local level.

8.7.1.3 The Arabian Peninsula The deficit on the Arabian Peninsula will grow from 20 billion m3/y in 2000 to 55 billion m3/y in 2050 with the major share taken by Saudi Arabia and Yemen. The CSP capacity potential for desalination amounts to 32GW for Saudi Arabia and 17GW for Yemen, while the other countries have minor shares. It is evident that it is very feasible to use CSP technology in order to produce electricity using solar energy in the Arab countries, which have ample solar radiation (Table 8.11). In fact, if only 1 per cent of the Saudi Arabian desert was used for CSP, the electricity gained would be enough to provide Europe and the Arab countries with electrical power and water. European countries would provide the technology and training and, in return, could get enough electrical power from the Arab states through the interconnection shown below. By this, European countries could minimize their air pollution while Arab countries would be receiving 100 billion m3 of water annually.

8.8 ENVIRONMENTAL IMPACT OF USING SOLAR TECHNOLOGY IN MENA The main environmental impact of our scenario is the reduction of carbon emissions by about 40 per cent from electricity generation in spite of growing demand. In the following we highlight two important issues in this regard.

8.8.1 LAND USE FOR CSP The specific land requirement of hydropower ranges between 10km2/TWh/y for microhydropower to over 400km2/TWh/y for very large schemes like the Aswan dam (Table 8.12). The average value results in 165km2/TWh/y for the total analysed region. Geothermal power requires little land (1 to 10km2/TWh/y), and the areas affected are in the subsoil at thousands of metres depth. In our scenario, biomass is produced mainly by agricultural and municipal residues (no extra land use) and from wood, resulting in an average land use of only 2km2/TWh/y. Energy crops – with a very high land use – were not considered in the MENA countries, as they would compete with food and water supply. For wind power, the average land use was 46km2/TWh/y. The specific values differ considerably according to the different performance indicators in each country. Table 8.12 the two columns on the right show the total area of each country and the percentage of this area used for power generation by renewable energy sources in 2050.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

296

W. E. ALNASER, F. TRIEB AND G. KNIES

TABLE 8.11 Global solar radiation, monthly and annual means (in kWh/m2/day) CITY Amman Aquaba Abu Dhabi Mohraaq Alain Tunis Cabes Alger Oran Tamanrset Riyadh El-Medina Taif Khartoum Juba Damas Almsleamea Bagdad Mosul Mascat Oman Jerusalem Doha Kuwait Tripoli Sirte Cairo Aswan Casablanca Marrakech Nouak Chatt Sanaa Mocha

JAN 2.7 3.0 4.3 3.6 3.9 2.4 2.9 2.2 2.8 5.2 3.5 4.5 4.4 5.5 5.5 3.1 2.0 3.0 2.0 4.0 4.6 2.7 3.7 3.1 2.9 3.4 5.7 4.6 2.7 3.4 5.7 4.0 4.9

FEB 3.7 4.6 5.0 4.8 4.3 3.1 3.7 3.0 3.7 6.1 4.6 5.4 5.2 6.4 5.5 3.5 3.0 3.8 2.8 4.7 4.7 3.4 4.4 4.1 4.0 4.2 4.0 5.6 3.3 4.2 6.2 4.4 5.4

MAR 5.0 5.9 5.7 4.6 5.3 4.4 4.9 4.1 4.9 6.9 5.1 6.2 5.6 6.8 5.4 4.6 4.1 4.8 3.6 5.5 5.5 5.0 4.9 5.5 5.0 5.0 5.2 6.5 4.5 5.2 7.2 4.8 6.1

APR 6.8 6.7 6.7 6.1 6.6 5.7 5.9 4.9 6.2 7.7 5.5 6.6 5.8 7.2 5.6 5.8 5.3 5.7 4.5 6.3 6.3 6.0 5.7 6.2 6.0 5.6 6.3 7.3 5.4 6.0 7.8 4.4 6.3

MAY 7.8 7.6 7.6 6.9 7.5 6.5 6.8 6.0 6.9 7.8 5.6 7.4 5.8 6.6 5.6 7.4 6.2 6.5 5.5 6.9 6.9 6.9 6.2 7.1 6.5 6.2 7.1 7.4 6.3 6.7 7.7 5.6 6.2

JUN 8.4 8.3 7.6 7.4 7.5 7.3 7.2 6.2 7.4 7.7 6.1 8.2 6.4 6.3 5.2 8.0 7.9 7.3 6.3 6.7 6.7 7.6 6.5 7.9 7.0 6.9 7.6 8.0 6.5 7.3 7.7 5.3 5.8

JULY 8.2 8.0 7.0 7.0 6.8 7.4 7.3 7.0 7.5 7.6 6.1 8.3 6.3 6.1 5.0 7.9 7.7 7.2 6.2 6.1 6.1 7.6 6.0 7.5 7.1 6.9 7.4 7.8 6.4 7.6 7.2 4.7 4.3

AUG 7.5 7.5 6.7 6.3 6.6 6.6 6.5 6.4 6.8 7.4 5.9 8.0 5.9 6.0 5.4 7.2 7.1 6.6 5.6 6.0 6.0 7.2 5.8 7.1 6.5 6.3 6.8 7.4 5.9 7.0 7.1 4.6 4.8

SEP 6.4 6.5 6.5 5.8 6.4 5.3 5.4 5.1 5.6 6.7 5.7 6.8 5.5 6.0 5.9 6.1 6.1 5.7 4.8 5.8 5.8 6.1 5.5 6.2 5.5 5.6 5.9 6.6 5.0 5.9 6.7 5.2 5.3

OCT 4.8 5.1 5.7 4.9 5.7 4.0 4.2 3.3 4.0 6.0 5.3 5.8 5.2 5.9 5.8 4.5 4.1 4.4 3.5 5.2 5.2 4.8 4.8 4.8 4.0 4.1 4.7 5.7 3.8 4.6 6.3 4.0 6.1

NOV 3.6 4.0 4.8 3.9 4.7 2.9 3.2 2.7 2.9 5.2 4.5 4.8 4.6 5.7 5.5 3.0 2.8 3.3 2.3 4.4 4.4 3.5 4.1 3.4 3.1 3.2 3.5 4.8 2.7 3.6 5.5 4.0 5.4

DEC 2.7 3.4 4.0 3.0 4.6 2.3 2.7 2.0 2.4 4.8 3.6 4.3 4.4 5.5 5.3 2.3 2.0 2.7 1.8 3.8 3.8 2.6 3.5 2.9 2.3 3.0 5.9 4.3 2.4 3.2 5.1 3.9 5.1

ANNUAL 5.6 5.9 6.0 5.3 5.8 4.8 5.1 4.4 5.1 6.6 5.1 6.4 5.4 6.2 5.5 5.2 4.9 5.1 4.1 5.4 5.4 5.3 5.1 5.5 5.0 5.0 5.4 6.3 4.6 5.4 6.7 4.6 5.5

Source: Global Emission Newsletter (2004)

Hydropower surface demand varies strongly between countries. Photovoltaic surface demand considers only 50 per cent of the total because many plants will be installed on roofs. Wind power and CSP surface demand is calculated as if exclusively used for power generation. Biomass surface demand is only considered for fuel wood energy. Concentrating solar thermal power schemes have a specific land use of 6–10km2/TWh/y. However, land could be gained from waste land if multi-purpose CSP

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

297

TABLE 8.12 Areas required for renewable electricity generation in 2050 for the scenario CG/HE

plants are applied. This would mean winning additional land rather than land ‘consumption’. Photovoltaic energy has no additional land use if installed on roofs, and a slightly higher land use than CSP if installed in large installations. An average land use of 7km2/TWh/y was assumed. It may seem paradoxical that solar and geothermal power generation has the best land use efficiency among all power technologies, even when not considering the potential land gain effect. The total mix of renewable energies in 2050 within the scenario ‘Closing the Gap/High Efficiency’ (CG/HE) has an average land use of 22.5km2/TWh/y, which is in the same order as the average value of natural-gas-fired combined cycle power stations, which represent the best available fossil-fuelled power technology. Disposal of sequestrated CO2 is not considered within this figure. The land use of oil- or coal-fired steam cycles is between 50 and 100km2/TWh/y. Considering the long time during which areas are affected by nuclear waste disposal and uranium mining, nuclear plants also have a high land consumption in the order of 100km2/TWh/y. This figure does not account for nuclear accidents like the one at Chernobyl. The change to renewable energies will therefore lead to more efficient land use for power generation. Solar thermal power plants will also be used for seawater desalination. A concentrating solar thermal collector array required for desalinating 1 billion m3/y would cover a total land area of approximately 10 x 10km, corresponding to about 10m3 desalinated water per m2 of collector area. In case of linear Fresnel or multi-tower technology, the collectors could act like blinds, blocking the intense direct solar radiation and creating a cool space underneath with sufficient light for horticulture or other purposes.

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

298

W. E. ALNASER, F. TRIEB AND G. KNIES

About 10 per cent of the desalinated water would be sufficient for irrigating the desert land beneath the collectors with a water column of 1m/y. In 2050, our scenario arrives at 2900TWh/y of electricity (including solar power generation and desalination) and 160 billion m3/y of desalted water. For this a collector field of 120 x 120km2 would be necessary, which is equivalent to not more than 0.15 per cent of the Sahara Desert.

8.8.2 CSP AND THE REDUCTION OF EMISSION OF GREENHOUSE GASES AND OTHER POLLUTANTS The emissions of renewable energy technologies mainly occur during the production of the plant’s components, because most plants are produced within today’s industrial production schemes that use mostly fossil energies. Thus the emission occurs from fossil power plants that are at present used to provide energy for the production of plant components. The life cycle emissions are valid for a power peak with average CO2 emissions of 700g/kWh. During operation, only biomass and geothermal plants produce emissions. The emission of greenhouse gases (CO2 equivalent) of renewable energy technologies are by orders of magnitude lower than those of fossil-fuelled technologies. Coal plants usually have emissions of 900–1100kgCO2/MWh, oil plants around 600–700kgCO2/MWh. Even coal plants with CO2 sequestration would still emit more CO2 than solar or wind power plants, as about 20 per cent of their emissions would still reach the atmosphere. Moreover, it is not yet clear for how long CO2 reservoirs of sequestration would remain isolated from the atmosphere. Other emissions that mainly occur during combustion, such as nitrates (NOx) and sulphates (SOx), as well as phosphoric acids are also avoided. These can lead to acidification and over-nutrition (eutrophication) of soils and water bodies. Emissions of CSP plants in hybrid operation will gradually be reduced with time and applying increased solar thermal storage capacities. At present, the total carbon emissions of electricity generation of all countries analysed in the study amount to approximately 770 million tons per year. Instead of growing to 2000 million tons per year, which would be expected for 2050 in a business as usual case, our scenario achieves a reduction of emissions to 475 million tons within that same time span (Figure 8.23). The scenario avoids a total of 28 billion tons of carbon dioxide until 2050, which is equivalent to the present total annual CO2 emissions worldwide. The scenario reaches a per capita emission of 0.58tons/cap/y in the power sector in 2050 (Figure 8.24). This is acceptable in terms of the recommended total emission of 1–1.5tons/cap (EU Commission, 1997). Comparing Figure 8.25 (‘electricity generation’) with Figure 8.26 (‘installed capacity’) reveals that the installed concentrating solar power capacity by 2050 is as large as that of wind, PV, biomass and geothermal plants together, but due to their built-in solar thermal storage capability, CSP plants deliver twice as much electricity per year as those resources. The use of CSP or renewable energy technologies for electricity and water production will contribute positively in reducing air pollution, haze, smog, ozone deterioration and global warming. In fact the most visible impact of air pollution is the haze, a brownish layer of pollutants and particles from biomass burning and industrial emissions that pervades many regions in Asia and is transported far beyond the source region, particularly during December to April (EU Commission, 1997).

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

299

2

Solar Energy Technology in the Middle East and North Africa

CO2 CO2

CO2

CO2

FIGURE 8.23 CO2 emissions of electricity generation in million tons per year for all countries for the scenario CG/HE and emissions that would occur in a business as usual case (BAU)

FIGURE 8.24 Annual per capita CO2 emissions of power generation (Scenario CG/HE)

For the security of power generation and our planet, the diversity of energy production, using different sources of energy, should transform from the existing pattern (Figure 8.27) to that proposed by our scenarios (Figure 8.28).

8.8.3 THE SOCIO-ECONOMIC IMPACTS OF THE SCENARIO The most important benefit will be in stabilization of electricity costs at a relatively reasonable price level. Another benefit will be the reduction of subsidy requirements in

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

300

W. E. ALNASER, F. TRIEB AND G. KNIES

FIGURE 8.25 Annual electricity demand and generation within the analysed countries in the MED-CSP scenario

FIGURE 8.26 Installed power capacity and peak load within the analysed countries in the scenario CG/HE

the energy sector (Figure 8.29). Renewable energies in the future will be the least cost option and may be the only option to obtain sustainable energy and water security in the MENA region. In most countries, the dependency on energy imports is reduced, opening new business opportunities for industrial development. In the total EU-MENA region there may be two million direct and indirect jobs in the renewable energy sector by 2050. Renewable energies are characterized by their diversity of resources and technologies and their enormous capacity range from a few watts to hundreds of MW. They can be adapted to any kind of energy service and closely interlocked with conventional modern energy technologies in order to provide full power availability and security of supply at any time and place. Renewable energy technologies fit very well into modern supply systems that are increasingly relying on distributed generation and network integration, like in ‘virtual power plants’. Furthermore, intercontinental grid connections can effectively combine the different regional resources to yield the necessary redundancy of supply and address the sustainability goal of international cooperation (Figure 8.30). Large centres of supply will evolve at sites with

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

301

FIGURE 8.27 Share of different technologies for electricity generation in 2000

FIGURE 8.28 Total electricity consumption and share of different technologies for electricity generation in the analysed countries in 2050 according to the MED-CSP scenario

very abundant, and thus cost-effective, renewable energy resources, providing electricity and renewable hydrogen to the regions of demand, in other words large urban areas in industrialized and developing countries, by means of high voltage direct current (HVDC) transmission and by pipelines respectively (ABB, 2004). At the same time, such centres will become a regional nucleus of economic development and wealth and will help to stabilize socio-economic structures. Many of those centres will be established in developing countries, contributing considerably to the positive progress of our developing world (Knies, 2004). Using solar energy means manufacturing machines that use renewable energies. It means replacing minerals from the subsoil by capital goods. Renewable energies require a large amount of labour on all industrial levels from base materials like steel, glass and concrete to civil engineering and advanced technology applications. Increased industrial

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

302

W. E. ALNASER, F. TRIEB AND G. KNIES

FIGURE 8.29 Example of electricity costs and learning in the MED-CSP scenario

activities will create job opportunities and reduce the brain-drain from MENA to the industrial countries. Considerable shares of the equipment and construction materials of the solar field and the power block can be produced domestically in many countries with potential CSP deployment. For parabolic trough systems, an evaluation of the supply capability of selected countries like Morocco, Spain and Brazil indicates domestic shares ranging between 40 and 60 per cent for the first plants. Local supply shares can be increased for subsequent projects if domestic industries adopt increased production of solar field and power block components.

ACKNOWLEDGEMENTS The authors wish to thank Prof. Yogi Goswami, University of South Florida, Co-director of the Clean Energy Research Center, Dr. Russell L. Shoemaker, University of Bahrain, and Barbara J. Graham, Clean Energy Research Center at the University of South Florida, for their assistance in editing this paper.

AUTHOR CONTACT DETAILS W. E. Alnaser (corresponding author), Physics Department, University of Bahrain, P.O. Box 32038, Kingdom of Bahrain e-mail: [email protected] F. Trieb, German Aerospace Center (DLR), Institute of Technical Thermodynamics, Stuttgart, Germany e-mail: [email protected] G. Knies, Trans-Mediterranean Renewable Energy Cooperation TREC, Hamburg and Amman e-mail: [email protected]

REFERENCES ABB (2004) HVDC references – ABB HVDC classic and light projects, ABB power transmission, www.abb.com/global/ abbzh/abbzh251.nsf!OpenDatabase&db=/global/ gad/gad02181.nsf&v=17EA&e=us&c =C1256D71001E0037C1256C330030750D Alnaser, W. E. (1995) ‘Renewable energy resources in the State of Bahrain’, Applied Energy, vol 5, pp23–30 Alnaser, W. E. et al (2004) ‘First solar radiation atlas for the Arab world’, Renewable Energy, vol 29, pp1085–1107

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Solar Energy Technology in the Middle East and North Africa

303

FIGURE 8.30 Projection of a future trans-Mediterranean grid interconnecting the best sites for renewable energy use in EU-MENA Alnaser, W. E. and Al-Karaghouli, A. (2000) ‘Wind availability and its power utility for electricity production in Bahrain’, Renewable Energy, vol 21, pp247–254 Al-Zubari, W. K. (2002) ‘Alternative water resource management policies in west Asia’, Industry and Environmental Journal, vol 25, no 1, pp–43–47 AQUASTAT (2004) Aquastat, FAO’s information system on water and agriculture, online database, www.fao.org/ag/agl/aglw/aquastat/dbase/index.stm Bazilevich, N. (1994) Global Primary Productivity Database, unpublished work, Clark University, Worcester, MA, US Bennouna, A. (2004) Perspektiven der Netzgebundenen Umwandlung Regenerativer Energien in Marokko, Rabat Bruchmann, U. (2004) Umweltpolitik – Geothermie – Energie der Zukunft, Bundesministerium für Umwelt, Naturschutz und Reaktorsicherheit, Berlin Chopra, P. (2003) ‘The economics (hot rock energy power plant)’, Australian National University, Canberra, 30 July 2003, http://hotrock.anu.edu.au/economics.htm Dabbagh, T. A. (1995) ‘The role of desalination and water management in sustaining economic growth in the Gulf’, International Desalination Association CIDA, World Congress in Desalination and Water Sciences, Abu Dhabi, 18–24 November DLR (2005) ‘Concentrating solar power for the Mediterranean region’, German Aerospace Center, Institute of Technical Thermodynamics, Federal Ministry for the Environment, Nature Conservation and Nuclear Safety, Germany, www.dlr.de/tt/med-csp ECMWF (2002) Re-analysis generated for 15 years (ERA-15) from December 1987 to February 1994, European Center for Medium-Range Weather Forecasts, London, http://wms.ecmwf.int/research/era/Era-15.html ENERDATA (2004) GasElecStat 2004 databases, ENERDATA S.A. ESCWA (2001) Energy Options for Water Desalination in Selected ESCWA Member Countries, United Nations, New York ESCWA (2003) Review of Sustainable Development and Productivities, no 1, United Nations, New York EU Commission (1997) ‘Energy for the future: Renewable sources of energy’, white paper, COM (97) 599 final EU (2004) The Share of Renewable Energy in the EU, Commission of the European Communities, Brussels FAO (2003) Review of World Water Resources by Country, FAO, Rome FAO (2004) FAO database collection ‘FAOSTAT 2004’, http://faostat.fao.org/faostat/collections?version=ext&hasbulk=0 Gardner, P. et al (2002) ‘Wind energy – The facts’, European Wind Energy Association, Berlin, www.ewea.org/06projects_events/proj_WEfacts.htm Gawell, K., Reed, M. and Wright, M. (2004) ‘Geothermal energy: The potential for clean power from the Earth’, Geothermal Energy Association, Washington, www.geo-energy.org/PotentialReport.htm

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

304

W. E. ALNASER, F. TRIEB AND G. KNIES

Gleick, W. (1998) Water Futures: A Review of Global Water Resources Projections Global Emission Newsletter (2004) vol 3, no 4, April Gra`l, H. et al (2003a) Welt im Wandel – Energiewende zur Nachaltigkeit, Wissenschaftlicher Beirat der Bundesregierung Globale Umweltveräderungen, Berlin März Gra`l, H., et al (2003b) Über Kioto hinaus denken – Klimaschutzstrategien füdras 21. Jahrhundert, Wissenschaftlicher Beirat der Bundesregierung Globale Umweltveränderungen, Berlin GTZ (2002) Stromproduktion aus erneuerbaren Energien: Energiewirtschaftliche Rahmenbedingungen in 15 Entwicklungs- und Schwellenländern, Deutsche Gesellschaft für Technische Zusammenarbeit, Eschborn, Germany GTZ (2004) Energiepolitische Rahmenbedingungen für Strommärkte und erneuerbare Energien – 21 Länderanalysen, Deutsche Gesellschaft für Technische Zusammenarbeit, Eschborn, Germany Heston, A., Summers, R. and Aten, B. (2002) Penn World Tables Version 6.1, Center of International Comparisons at the University of Pennsylvania (CICUP) Horlacher, H. (2003) Globale Potenziale der Wasserkraft, Wissenschaftlicher Beirat der Bundesregierung Globale Umweltveränderungen, Berlin, Heidelberg Hurter, S. et al (2000) Atlas of Geothermal Resources in Europe: Planning Exploration and Investments, Geowissenschaftliche Gemeinschaftsaufgaben, Hannover, Germany International Energy Agency (IEA) (2002) Annual Report IEA (2003) Energy Statistics of Non-OECD Countries 1971–2001, International Energy Agency Jawad, A. (2001) ‘Energy sources for coupling with desalination plants in GCC countries’ in consulting report for ESCWA Knies, G. (2004) ‘Solar energy & water for sustainable prosperity in MENA’, Trans-Mediterranean Renewable Energy Collaboration TREC, Tripoli Lehner, B., Czisch, G. and Vassolo, S. (2005) ‘The impact of global change on the hydropower potential of Europe: A modelbased analysis’, Energy Policy, vol 33, no 7, pp839–855 Meteonorm (2004) Global Meteorological Database for Applied Climatology Meteornorm Version 5.0, Meteotest, Bern, www.meteotest.com Mortensen, N. G. et al (1996) Wind Atlas for the Gulf of Suez 1991–95, New and Renewable Energy Authority, Meteorology and Wind Energy Department, Cairo/Roskilde Mortensen, N. G. et al (2003) Wind Atlas for the Gulf of Suez, New and Renewable Energy Authority, Wind Energy Department, Cairo/Roskilde NGDC, www.ngdc.noaa.gov/seg/cdroms/ged_iib/datasets/b02/baz.htm#top OME (2002) ‘Large scale integration of solar and wind power in Mediterranean countries’, MED2010, Observatoire Méditerranéen de ‘Éergie, Paris, www.ome.org Saghir, J. (2000) ‘Urban water and sanitation in the Middle East and North Africa: The way forward, building knowledge and expertise in infrastructure finance’, 24 January–4 February, World Bank, Amman SOLEMI (2004) ‘Solar energy mining’, www.solemi.de UNDP (2002) Human Development Report, New York UNEP (2003) The Asian Brown Cloud: Climate and Other Environmental Impacts, Center for Clouds, Chemistry and Climate WEC (2004) ‘Energie für Deutschland – Fakten, Perspektiven und Positionen im globalen Kontext’, Dt. Nat. Komitee DNK des Weltenergierates, Dusseldorf, Germany Wittig, G. (2004) Bestec GmbH, Stand des europäischen Hot-Dry-Rock Forschungsprojektes in Soultz-sous-forets, EWIV Wärmebergbau, Strasbourg World Almanac, the (2004) World Almanac Books, New York WWA (2004) World Wind Atlas 2004, Meteotest, Bern, www.meteotest.com

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 261–304

Index / electrons, 75 1.5 kW Society, 37 Abdul-Baqi, 13, 38 absorber tube, 132–3 accumulated energy, 152–3 active wells, 17 adsorption, 139, 145, 147, 149–51, 164 advanced oxidation processes, 130, 163 Alaska, 10, 14 Al-GaAs/GaAs, 45, 49, 52, 54, 66, 71 alternative fuels, 37 aluminum reflector, 137 ammonia, 16, 18 Angola, 9–10 annual discoveries, 14 annual per capita CO2 emissions of power generation, 299 anti-bonding orbital, 75 application potential, 217–18, 220 Arabian Gulf countries, 282 areas required for renewable electricity generation, 297 artificial photo-charge synthesizers/converters, 95 Association for the Study of Peak Oil, 2, 4, 14, 22, 38 Auger recombination, 49, 73 Azerbaijan, 11–12 Azeri-Chirag-Guneshli, 11 background doping, 50, 60 Bakhtiari, 13, 38 Barent Sea, 21 batch mode, 153–4 bi-continuous double diamond phase (OBDD), 88 biodegradability, 162 biomass, 3–6, 30–5, 37, 39, 41–4 body-centered cubic lattice (BCC), 88 bonding orbital, 75 bound states, 48–9 BP, 2, 5, 7, 10, 12, 19–20, 22, 25, 39 Bragg reflector, 60 Brazil, 9–10 bulk heterojunction, 81–2, 96 calcium carbonate, 100–1,118–19, 121–3, 125–9 Canada, 2, 10–11, 17–18, 39 Cantarell, 9 capacity of CSP plants, 293 capacity potential for CSP, 292, 294 carbon dioxide emissions, 25, 27–8, 36 carbon emissions, 261, 295, 298 carrier escape, 50–1, 60, 70–1, 73

carrier loss, 74, 81, 87–8, 90–1, 94 Caspian Sea, 12 catalyst, 131, 134–6, 139, 142, 145–8, 151, 154–8, 160–1, 163, 165–6 catalyst mass, 148 charge mobility, 77 charge separation reorganization energy, 92–3 charge transfer, 170, 177, 179–80, 182–3, 185, 188, 193–5, 200–1, 203, 205, 212 chemical doping, 93–4 chemical potential, 78, 87, 90 China, 10, 21–2, 30, 41–2 chlorinated, 151, 156, 167 chlorine, 112,–13, 128 climate change, 1–2, 6, 28, 33 closing the gap, 274–5, 277–81, 285–90, 293, 297 CN-PPV, 80–1 CO2 emissions, 263–4, 281, 298–9 coal, 4–5, 7, 24–7, 32–3, 35–7 collector efficiency, 130, 152, 160 collector surface, 134, 150, 152 compound parabolic concentrators, 136 compressive strain, 55 concentration factor, 133, 137, 142 concentrator, 61, 68, 73 conventional wisdom, 1, 12 cooperation between Arab and European, 263 cooperation between MENA and Europe, 283 cost reduction, 34–5 CPC, 136–8, 142–3, 156, 158–9, 163, 165, 167 critical thickness, 56, 60 CSP market in MENA, 281 CSP plants, 277, 292–4, 298 dark current, 55, 58, 63, 65, 68–9, 71 defect chemistry, 169–70, 178–9, 190–1, 194–5, 200, 202, 211–14 defect disorder, 169–70, 178–81, 187, 190, 193–7, 203, 211, 215 degradation, 130–1, 138–43, 145–7, 149–52, 155–6, 162–7 degradation of mechanical properties, 112 degradation process, 145, 149 Degussa P-25, 142, 147 demand pressure, 7 density of states, 45, 48, 50 desalination by concentrating solar power, 291 dichloroacetic acid, 142–3 diffuse radiation, 137 direct normal irradiation, 266–7 domestic electricity demand in MENA, 274 domestic electricity demand in Western Asia, 279

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 305–308

306

ADVANCES IN SOLAR ENERGY

domestic electricity demand of the Arabian Peninsula, 280–1 domestic hot water, 99, 124, 127 domestic Supply, 5 double-skin sheet photo-reactors, 154 driving forces, 6 East Asian countries, 21 economic potentials of renewable energy sources, 267 economic principles, 1, 6 effective mass, 48–9 effective photon flux, 133 efficiency, 4, 26, 32, 37 electricity, 4, 6, 17, 25–6, 29–33, 35, 37, 39, 41–4 electricity consumption per capita, 263, 265 electricity demand, 261, 263–6, 274–81, 300 electricity from biomass, 270 electroluminescence, 45, 65–7 electron acceptor, 145, 161 electron scavengers, 144 electron/hole pairs, 144 electron/hole recombination, 145 electrons, 130, 144–5 energy consumption, 1, 3, 6–7, 31–2, 37, 39 energy demand, 4–6, 29 energy gap grading, 83, 91 energy losses, 26–7 energy offset, 78, 83, 87, 91–4 entropy, 45, 65, 70 ereep, 112–15, 117, 125 Europe, 2, 5–6, 14, 16, 18–22, 31, 40–2 exciton, 50, 66, 71 exciton dissociation, 78, 81, 87–8, 91, 98 exciton loss, 74, 78, 81, 87–8, 90–1, 94 exciton quenching parameter (EQP), 91, 93–4 exciton radius, 75–6 excitonic mechanism, 76 experimental time, 151–2

hard water, 103, 118, 122–3 heat, 4, 6, 8, 17, 31–3, 35, 37, 40–3 heat exchangers, 99, 101–5, 107–9, 111–13, 115, 117–19, 121, 123–9 heat transfer, 99–107 heterogeneous photocatalysis, 131–2, 147, 161, 165, 167 hexagonally (HEX), 88 hexagonally perforated layers (HPL), 88 high fossil scenario, 14, 26–8, 34 history of oil production, 9 holes, 131, 144–5 HOMO, 75–6, 78–80, 82–3, 85–6, 91–4 homogeneous, 161 homogenous photo-Fenton, 150, 161 honeycomb, 82 hydrogen, 16, 25, 37–8 hydrogen peroxide, 145–6, 164, 167 hydropower, 3–4, 30, 32, 39, 43–4 hydrostatic burst, 113 hydroxyl radicals, 131, 140, 144–5, 150 industrial wastewaters, 154 industry, 216, 218, 220, 236 infrastructure, 10, 12, 15–16 InGaAs/InGaAs, 51–2, 54, 64, 67, 72 InGaAs/Ingaes, 51–2, 54, 64, 67, 72 InGaAsP/InP, 52, 71 InGaP/GaAs, 52, 61, 71–2 InP/InAsP, 52, 60, 73 InP/InGaAs, 51–2, 64, 73 installed capacity, 23, 43 installed power capacity, 262, 300 integral collector storage, 99, 126–8 interconnection, 276, 278, 295 Intergovernmental Panel on Climate Change, 2, 28 intermediate band, 65–6, 68, 70–1 intermediates, 139–40, 146, 151, 162, 165–6 International Atomic Energy Agency, 2, 22–3, 39 International Energy Agency, 1–3, 5, 38, 40

Fenton reagent 131 fill factor (FF), 86–7 first order kinetics, 152 Fischer-Tropsch, 25 flexible shape, 75 Former Soviet Union, 2, 9–10 fossil fuels, 6, 25, 33–6 Frenkel type of exciton, 76 future Energy Supply, 1

Japan, 21, 39

GaAs/InGaAs, 51–2, 54–9, 64, 67, 69–70, 72 GaAs/Ingaes, 51–2, 54–9, 64, 67, 69–70, 72 GaAsP/InGaAs, 50–2, 54–5, 58–9, 68–9 geothermal energy, 5, 30, 39 geothermal power, 269–70, 275, 295, 297 Ghawar, 12–13 global solar radiation, 262, 283, 296 Great Britain, 9 ground state charge separation, 91, 93 groundwater treatment, 154 growth rate of CSP production in MENA, 275 Gulf of Mexico, 2, 10

lamellae (LAM), 88 largest oil fields, 9–10 lattice mismatched, 45, 55, 61 L-H isotherm, 138 light intensity, 142, 147, 164 light scattering, 148, 151 light-trapping, 60, 63 lightweight, 75 limited resources, 6 LNG, 2, 18–20 LNG terminals, 18–19 logistic function, 29 logistic growth, 1, 29

Kamchagarak, 12 Kara Sea, 21 Kashagan, 12 Kazakhstan, 11–12 kinetic constants, 151 kinetic studies, 138, 164 Korea, 21

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 305–308

INDEX

low fossil scenario, 14, 22, 27–8, 35 LUMO, 75–6, 78–80, 82–3, 86, 91–4 LWFE, 77–9, 82–5 market penetration, 28, 32, 36 mature, 14, 17, 26 MDMO-PPV, 79, 81 Mediterranean ring interconnection, 276 medium temperature, 216–21, 227, 242–3, 248 Medvezhye, 21 MEH-PPV, 79, 81 membrane, 155, 158, 163 metal organic vapor phase epitaxy (MOVPE), 49 methanol, 16, 18, 39 Mexico, 2, 9–10 mineralization, 130, 138–42, 145, 147, 149–51, 162, 166 mineralization rate, 147, 150 molar ratio, 145–6 molecular beam epitaxy (MBE), 49 multi-junction, 45, 55, 60–1, 63, 66, 68–70, 73 natural convection, 99–108, 111, 124, 128 natural gas, 1–2, 4–7, 14, 16–20, 33, 36–7 net energy balance, 25, 31 non-biodegradable contaminants, 149 non-concentrating collector, 130, 132, 134, 137, 142, 154 non-conventional oil, 10–11 non-imaging optics, 137, 164 North Africa, 20, 31 North America, 5–6, 14, 16–18, 21–2, 41–2 North Sea, 9 Norway, 2, 9–10, 20, 39 NREL, 81 NTU, 100, 103, 109–11 nuclear energy, 4, 6–7, 14, 22, 25–6, 33–4, 36 Nusselt number, 100, 105–7, 124 nylon, 120, 125, 127, 129 O2, 147–8 oil, 1–19, 22, 25–7, 32–39 oil price, 4–5, 11, 35 oil Provinces, 8 oil sand, 10, 39 OPEC, 2, 9–10, 12 open circuit voltage, 77, 83–4, 86, 91 optical efficiency, 136–7 optical path, 148 optimum catalyst concentration, 148 organic contaminants, 145, 162–3 Ormen-Lange, 19, 39 oscillator strength, 48, 71 oxidative reduction potential, 112 oxygen, 131, 135, 147–9, 155, 160 P3HT, 80–1 P3OT, 80–1 parabolic trough collector, 132, 134, 137 parallel (shunt) resistivity (Rp), 86 parallel resistivity, 86–7 parity, 46 PCBM, 80–1

307

PCE or dPCE, 87 peak of oil production, 3–4, 7 peroxydisulfate, 145–6 pesticides, 130, 140–1, 150–1, 158, 160, 164–6 phenol, 134, 147, 165–6 photo doping, 93 photocatalysis, 136, 138, 142, 146–7, 149–50, 154–5, 158, 161, 163–7 photochemical, 132, 137, 146, 166 photo-efficiency, 160 photo-electrochemistry, 169–70, 211 photo-electrode, 169–70, 174–5, 177–81, 183, 190, 194–5, 198, 201–4, 206, 211 photo-Fenton, 130–2, 134, 149–50, 159, 161, 163–6 photon flux, 133, 146–7 photon loss, 74, 80–1, 83, 87, 90–1, 94 photo-reactor, 148–9 photosynthesis, 76 photovoltaic systems, 261, 266 pilot plant, 130, 142, 147, 151, 153–6, 163–4, 166 p-i-n, 49–55, 58, 60, 63, 65, 71 pipeline, 16–17, 19–20 plant size, 142 pollutant, 130–1, 134–5, 145–7, 149–50, 152, 155, 162, 165–7 polyacetylene, 77 polyamide, 6, 112, 118, 120–1, 127 polybutylene (PB), 112, 125 polymer, 99, 101, 112–14, 117–29 polyolefin, 112–13, 115, 118, 125–27 poly-p-phenylenevinylenes (PPVs), 77, 90 polysulfone (PSU), 112, 125 polythiophenes, 77, 90 primary structure, 89 process heat, 216–21, 226–7, 231–2, 234, 236, 238–9, 241, 246–7, 258–60 production profile, 8–9, 14–15, 19, 25, 28, 39 PSS-PEDOT, 80, 84–6 quantum dot, 66, 68, 70 quantum efficiency, 136–7 quantum well, 45–6, 48–9, 67–9, 70–3 quantum yield, 144–6, 161 quasi-Fermi level, 55, 65–6, 68, 70 R/P-Ratio, 20, 25–7 radiation resistance, 60–1 Rayleigh number, 100, 103–5, 108, 124 reaction rate, 130, 139–49, 151–2, 161 reaction rate constant, 139–41 reactors, 22–4, 26 recombination, 45, 49–50, 53–5, 58, 64–6, 68, 71, 73 recombination quenching parameter (RQP), 92–4 reduction of emission of greenhouse gases, 298 reflectivity, 133 regio-regular, 90 relative photonic efficiency, 147 renewable electricity performance indicators, 267

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 305–308

308

ADVANCES IN SOLAR ENERGY

renewable energy, 1, 3–4, 6–7, 28–33, 35–7, 39–43 renewable energy scenario, 28, 31, 33, 42 reserves, 3, 12–16, 19, 23–5, 38–9 residence time, 148, 152, 154 resources for concentrating solar power, 266 Rubrene, 77, 90 Saleri, 13, 38 Saudi Arabia, 2, 11–13, 38 Saudi Aramco, 13, 38 scaling, 99, 103, 112, 118–29 Schottky junction, 74, 77–8, 93 seawater desalination, 261, 274–80, 287, 297 secondary structure, 89 sedimentation, 136, 148, 155, 157–8 semiconductor, 131, 139, 145, 147, 163, 165–7 semiconductor surface, 139 serial resistivity (RS), 86–7 Shockley and Queisser, 54, 64 Si/SiO2, 60 Simmons, 13, 38 size distribution, 151 socio-economic impacts, 300 solar detoxification, 150, 153, 156–7, 159–60, 162–7 solar energy, 5 solar energy conversion, 169–70, 214–15 solar photochemical processes, 132 solar photo-reactor, 132, 151 solar power for Mediterranean region, 262, 264, 304 solar powered desalination, 277, 280 solar spectrum, 131, 135, 160–1 solar thermal, 216–18, 221, 226, 230, 243, 245–7, 256, 258–60 solar thermal power plants, 2, 31 solar-hydrogen, 170–1, 173–5, 177, 179, 181, 183, 185, 187, 189–91, 193–95, 197, 199, 201, 203, 205–15 spare capacity, 13 spectral response, 51, 53, 71 SSH model, 76 Statoil, 12 storage, 99–111, 124, 126–9 storage tank, 99, 102–5, 107–9, 111, 124, 128 strain balance, 46, 50–2, 55–9, 61–4, 66, 69, 72, 73 strain relaxation, 55–6, 62, 69 stratification manifolds, 108 stratified, 104, 108 sulphur, 2, 12 superlattice, 45, 49, 60, 66, 68–70, 73 supersaturation, 100, 119–24 supported catalyst, 136, 145 suspended catalyst, 155 sustainability, 36 SWFE, 77–9, 82–6

synthetic crude, 11, 16, 25 tandem structure cell, 83 technical potential, 30–1 Tengiz, 12 tensile strength, 112–16 tertiary structure, 90 textile wastewater, 155, 164 The Netherlands, 20, 39 thermophotovoltaics, 45, 55, 61, 63, 66, 68, 70, 72 Thorium, 24 TiO2, 131, 134, 151, 164 TiO2 particles, 138, 151, 164 titanium dioxide, 161, 163–5, 170, 194, 212–15 TOC, 140–1, 143, 147, 155, 159, 162 top-down analysis, 14–15 total annual water deficits in MENA, 287 total organic carbon, 140 toxic substances, 149 toxicity, 150, 160, 162, 164, 166, 168 transition period, 1, 3, 29, 33, 36, 38 transparent conducting electrode (TCE), 84 tube bundle, 104–5, 111, 128 turbulent flow, 136, 148 two axis parabolic trough, 142 type-1, 49, 73 type-11, 49 ultra-fast, 75 Uncertainty principle, 45–6 Uranium, 7, 23–4 Urengoy, 21 US Energy Information Agency, 12 UV radiometer, 152 Venezuela, 2, 10 voltage enhancement, 45–6, 53, 67 Wannier type of exciton, 76 water deficiency, 262 water demand in MENA, 284 water demand in North Africa, 289 water demand in Western Asia, 289 water demand predictions, 285 water detoxification, 132, 162, 165, 168 water photolysis, 169–70, 174–6, 178, 204, 208, 210–11 water resources and water demand, 284 wave and tidal power, 262, 273 wave power, 5 well width, 45–6, 53, 67 wind energy, 2, 4, 6, 29, 31–2, 39–40 wind power, 271–2, 293, 295, 296, 298, 305 World Energy Outlook, 1–3, 5, 33–4, 36, 38 Yamburg, 21 zero order kinetics, 141

ADVANCES IN SOLAR ENERGY ■ 2007 ■ VOLUME 17 ■ PAGES 305–308