Zone-Melting Recrystallization for Crystalline Silicon Thin-Film Solar Cells Dissertation zur Erlangung des akademischen Grades des Doktors der Naturwissenschaften (Dr. rer. nat.) an der Universität Konstanz Fachbereich Physik vorgelegt von THOMAS KIELIBA Fraunhofer Institut für Solare Energiesysteme Freiburg 2006 Kieliba, Thomas: Zone-Melting Recrystallization for Crystalline Silicon Thin-Film Solar Cells [Elektronische Ressource] / Thomas Kieliba. – Konstanz, Univ., Diss., 2006 Zugl. – Berlin : dissertation.de – Verlag im Internet GmbH, 2006 ISBN 3-86624-196-8 Bibliografische Information der Deutschen Bibliothek Die Deutsche Bibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über <http://dnb.ddb.de> abrufbar. Dieses Dokument ist urheberrechtlich geschützt. Einzelne Vervielfältigungen, z.B. Kopien und Ausdrucke, dürfen nur zum privaten und sonstigen eigenen Gebrauch angefertigt werden (Paragraph 53 Urheberrecht). Die Herstellung und Verbreitung von weiteren Reproduktionen ist nur mit ausdrücklicher Genehmigung des Urhebers gestattet. Dissertation der Universität Konstanz Tag der mündlichen Prüfung: 22.09.2006 Referenten: Priv. Doz. Dr. Gerhard Willeke Prof. Dr. Ulrich Rüdiger ii v3.4b (doc: 2006-11-06) Acknowledgements I would like to thank my advisor PD Dr. Gerhard Willeke for the faith in my abilities and the continuous support of my work. He allowed me great freedom in this research. I am very grateful to Prof. Dr. Ulrich Rüdiger for acting as a second reviewer and an examiner. The Fraunhofer ISE provided an excellent environment for creative work. Quite a lot of people contributed to this very comfortable working atmosphere, and there are too many to mention all individually. I am especially grateful to Dr. Stefan Reber for many stimulating discussions, including the consideration of industrial applications, and his helpful comments on the final draft. I very much appreciate the valuable input of Dr. Achim Eyer on crystal growing techniques and equipment design. I am extremely grateful to Dr. Wilhelm Warta for many helpful discussions on defects, their characteri- zation, and on the appropriate communication of scientific results. The chapters on dislocation modeling and characterization profited very much from his con- structive input. I would like to thank Dr. Albert Hurrle for many inspiring dis- cussions that often helped me get back on track. I am very appreciative of the help I received from the department staff in solar cell processing and characterization. Norbert Schillinger and Fridolin Haas are thanked for construction of the ZMR furnace that withstood many torturous experiments. Mira Kwiatkowska, Harald Lautenschlager, Toni Leimenstoll, and Christian Schetter kindly prepared solar cells on numerous, sometimes exotic, substrates. Many thanks to Elisabeth Schäffer for uncounted measurements, her helping hand whenever needed and the good humor she spread. Daniel M. Spinner is thanked for his perfect support regarding all computer software and hardware issues. I would like to thank all my colleagues who have contributed to this work in different ways. The following people deserve special mentioning: I would like to thank Stephan Riepe for many fruitful discussions on defects and their appro- priate modeling, and also for MFCA measurements. I am very appreciative of iii iv ACKNOWLEDGEMENTS Dr. Stefan Peters from the Fraunhofer ISE »outpost« in Gelsenkirchen for a wealth of discussions on photovoltaics and everything under the sun. He was always a very welcome guest in the Rennweg apartment. I would like to thank Dominik Huljić for many contributions, inspiring discussions, and the great time we spent together. I am grateful to Dr. Sandra Bau for the good teamwork in silicon film preparation and many useful discussions. Further, I am indebted to Johannes Pohl whose studies contributed a lot to this work, and to Stefan Janz who took over the operation of the ZMR lab. Transferring the responsibilities over to him was a pleasure, and I know they are in good hands, which has helped me to concentrate on this thesis. Much of the research was conducted in conjunction with national and inter- national projects. I appreciate the valuable contribution and helpful discussions of all partners involved. Along side the »official« projects some fruitful collabo- rations developed. I would like to thank Christian Schmiga from ISFH, Hameln/Emmerthal for hydrogen passivation of numerous samples. I am indebted to Dr. Melanie Nerding from the University of Erlangen-Nürnberg for TEM and EBSD characterization, and for helpful discussions. Thanks are also due to Dr. Gaute Stokkan from NTNU, Trondheim for the work on dislocation density measurements. I appreciate the funding of this work by the scholarship program of the German Federal Environmental Foundation (Deutsche Bundesstiftung Umwelt) and I would especially like to thank Dr. Maximilian Hempel for his support. I am very grateful to Nicole Kuepper for proofreading the English text and her incredibly fast response with helpful corrections from »Down Under«. Finally, I would like to express my deepest gratitude to my parents for their continuous support, and to my sister and my brothers for their support at all times. Abstract Thin-film solar cells from crystalline silicon combine advantages from silicon wafer based technology with thin-film features. On one hand, the use of a sup- porting substrate minimizes the consumption of highly pure silicon and enables integrated interconnection of the individual solar cells within a module. On the other hand, these solar cells profit from the established silicon technology and the abundance of quartz sand as a raw material. The thin-film solar cell technology investigated in this work belongs to the so-called »high temperature approaches«. On a low-cost substrate, an intermedi- ate barrier layer is deposited. Then a thin silicon film is applied on top by chemical vapor deposition (CVD) and afterwards transferred into a large grained structure by zone-melting recrystallization (ZMR). The recrystallized film acts as a »seed« for subsequent epitaxial growth, again by silicon CVD. The final silicon thin film has a thickness of 20–30 µm. For silicon thin-film formation, ZMR is a key technology. This process largely determines film quality since defects, such as dislocations or grain boundaries, are replicated by the subsequent epitaxial growth. The ZMR method yields large grains with sizes comparable to those in multicrystalline silicon wafers grown by directional solidification. However, an inevitable feature of ZMR is the creation of low angle grain boundaries within the grains, so-called subgrain boundaries (SGBs). These subgrain boundaries induce stripes with high dislocation density in the epitaxial silicon layer. Therefore, they are especially harmful for the solar cell device. The analyzed films showed that dislocation density and therefore electronic material quality in subgrain boundary regions is related to the run of the subgrain boundaries relative to the ZMR scan direction. For equally spaced subgrain boundaries running parallel to the scan direction, their effect was found to be smallest. Investigations by optical and electron microscopy support an earlier theory, which explained subgrain boundary for- mation by tilting of subgrains and polygonization of dislocations. For the ZMR process, the effect of different material and process parameters on film quality has been investigated. These include the substrate type, the sili- v vi ABSTRACT con film thickness, the scan speed and the type of capping used to prevent agglomeration of molten silicon. Regarding the capping type, a ∼0.15 µm thick rapid thermal oxide (RTO), which was grown inside the ZMR reactor, has been compared to a standard 2 µm thick SiO2 layer deposited by plasma enhanced CVD. For 8 µm thick sili- con films, the thin thermal oxide was found to yield better film quality and solar cell performance. Scan speed is the most crucial parameter for costs of the ZMR process. An automated process control based on image analysis of the molten zone has been developed, which allows high-speed ZMR. On model substrates, the dependence of dislocation density and solar cell performance on scan speed has been studied for values between 10 mm min−1 and 100 mm min−1. The minimum value is similar to the pull speed for common silicon ribbon materials, such as Edge- defined Film-fed Growth (EFG) or String-Ribbon, and yielded comparable crystal quality. For a higher scan speed compromises regarding crystal quality and solar cell performance have to be made. For solar cells prepared without bulk hydrogenation, the tenfold increase of scan speed from 10 mm min−1 to 100 mm min−1 resulted in a relative decrease in solar cell conversion efficiency of approximately 35%. The choice of substrate material and its preparation have a major effect on quality of the ZMR films. For the investigated materials (Si3N4, SiSiC and ZrSiO4 ceramics, SSP ribbons) three key issues have been identified, which cur- rently lead to a drawback in solar cell performance compared to devices fabri- cated on »model« substrates. (i) Thermal expansion has to match very precisely with silicon. The maximum tolerable difference in length is estimated to be below 1 ‰. (ii) The stability of the intermediate layer is crucial for successful ZMR processing. Irregularities observed in situ during ZMR processing could often be traced back to a damaged