Available online at www.sciencedirect.

com

Energy Procedia
www.elsevier.com/locate/procedia

Energy Procedia 00 (2011) 000000 Energy Procedia 10 (2011) 61 65

European Materials Research Society Conference Symp. Advanced Inorganic Materials and Concepts for Photovoltaics

Polycrystalline silicon thin films by high-rate electronbeam evaporation for photovoltaic applications Influence of substrate texture and temperature
Christiane Becker*, Tobias Sontheimer, Simon Steffens, Simone Scherf, and Bernd Rech
Helmholtz-Zentrum Berlin fr Materialien und Energie, Institut Silizium Photovoltaik, Kekulstr. 5, 12489 Berlin, Germany

Abstract Polycrystalline silicon (poly-Si) thin-films have the potential to overcome the limits of todays silicon thin-film solar cell technology because of the chance to grow high-quality material by low-cost deposition techniques. As poly-Si thin-film fabrication is independent of rather slow and costly PECVD processes, it is possible to switch to physical high-rate deposition methods such as electron-beam (e-beam) evaporation exhibiting a strong cost-saving potential. In this contribution, the challenges and opportunities of two different poly-Si solar cell preparation techniques using e-beam evaporated silicon are investigated: Thermal solid phase crystallization (SPC) of initially amorphous silicon layers, and direct growth of poly-Si films at elevated temperatures >500C. For both approaches, attention is given to the interplay between substrate texture, structure of the grown silicon and the final solar cell performance. Poly-Si thin film solar cells with 7.8% conversion efficiency were prepared on a smooth substrate in cooperation with CSG solar, demonstrating the equality of e-beam evaporation as deposition technique compared to PECVD. However since e-beam evaporation leads to a non-conformal deposition of rough surfaces, the implementation of a textured substrate for light trapping is highly desired but still challenging. A different behavior is observed for poly-Si thin films directly grown at higher temperatures. For the best photovoltaic performance, a certain substrate microstructure is even necessary and can be optimized by the use of glass coated by differently textured ZnO films as substrate.

2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Organizers of European 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of the EMRS 2011 Materials Research Society (EMRS) Conference: Symposium on Advanced Inorganic Materials and Concepts symposium organizers: G. Conibeer; Yongxiang Li; J. Poortmans; M. Kondo; A. Slaoui; M. Tao; M. for Photovoltaics. Topic
Polycrystalline silicon; thin film solar cells; electron-beam evaporation; crystallization kinetics * Corresponding author. Tel./Fax: +49 - 30 - 8062 41375/41333. E-mail address: christiane.becker@helmholtz-berlin.de

1876-6102 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Organizers of European Materials Research Society (EMRS) Conference: Symposium on Advanced Inorganic Materials and Concepts for Photovoltaics. doi:10.1016/j.egypro.2011.10.153

62

Christiane Becker et al. / Energy Procedia 10 (2011) 61 65 Author name / Energy Procedia 00 (2011) 000000

1. Introduction Electron-beam (e-beam) evaporation is an appealing deposition method for polycrystalline silicon thinfilm solar cells. With deposition rates of about 1 micrometer per minute e-beam evaporation yields a huge cost saving potential for photovoltaic applications compared to widely used deposition techniques based on plasma enhanced chemical vapor deposition (PECVD) where typically only 20-30 times smaller rates are achieved. Furthermore, for e-beam evaporation of silicon no toxic, explosive or greenhouse gases have to be used, such that the corresponding technological safety requirements are much lower than in a PECVD laboratory. A common method to fabricate polycrystalline silicon thin-film solar cells is the solid phase crystallization (SPC) of amorphous silicon [1,2]. Recently, it could be shown that e-beam evaporated poly-Si thin-film mini modules by SPC yield equivalent efficiencies as respective devices grown by PECVD: On planar substrates an efficiency of 6.7% could be achieved that could even be improved to 7.8% by implementation of a rear side texture for light trapping [3]. However, a further increase of conversion efficiency by implementation of a textured substrate for light trapping failed so far. The reason might be that e-beam evaporation is a directed, non-conformal deposition method favoring shadowing effects [4] and material growth with minor electronic quality at steep edges of the texture [5]. In the present study, the influence of the substrate texture on poly-Si thin film growth will be analyzed in detail for two cases: The kinetics of the solid phase crystallization process of amorphous silicon on the one hand (section 3.1), and the direct growths of poly-Si at elevated substrate temperatures >500C on the other hand (section 3.2). Furthermore, it will be shown that the substrate temperature during deposition plays an important role as well and is immediately connected with the poly-Si growth process. It will be shown that the poly-Si growth can be controlled by smart setting of texture and temperature. 2. Experimental Substrates with different surface topographies were prepared by covering Schott Borofloat 33 glass with SiNx and/or aluminium doped zinc oxide (ZnO:Al) layers. The smoothest substrate with a root mean square roughness rms= 0.3 nm was achieved by coating the glass with a 100 nm thick SiNx layer in an industrial PECVD reactor. By non-reactive RF- sputtering of 850 nm thick ZnO:Al films on top of the SiNx at temperatures between 200 and 300C from ceramic targets with 1wt% or 2wt% Al2O3 content, layer surface textures of rms=5 nm, 10 nm and 29 nm were prepared. The surface topography of these substrates was measured by an atomic force microscope (Anfatec Instruments) equipped with a 10 nm large Si tip. The AFM was operating in tapping-mode at a resonance frequency of 330 kHz and investigated a 2x2 m2 surface area. To analyze the surface texture the root mean square roughness rms was determined. 1 m thick silicon thin films with about 2x1016 cm-3 boron doping were deposited by e-beam evaporation of silicon and co-evaporation of boron from an effusion cell. The substrate temperature was varied between 150 and 300C for preparation of amorphous silicon films that should be further solid phase crystallized by thermal annealing at 600C (section 3.1). The directly polycrystalline grown silicon films were deposited in the temperature range from 500 to 700C (section 3.2) since the transition from amorphous to nanocrystalline silicon growth occurs in the range between 350 C and 450C [6]. The SPC process was monitored by analysis of optical microscopic images from the partly crystalline Si films taken in transmission mode [7]. High resolution images were taken with a Hitachi S4100 cold field emitter scanning electron microscope (SEM). In order to visualize single crystal grains some poly-Si films were defect-etched by a Secco etch solution with 2 parts hydrofluoric acid (50%) and 1 part 0.15 M K2Cr2O7 [8]. Open circuit voltages of corner etched n+/p-/p+ poly-Si stacks were determined by Sinton SunsVoc measurements reflecting the absorber material quality. A white paper was used as back reflector.

Christiane Becker et al. / Energy Procedia 10 (2011) 61 65 Author name / Energy Procedia 00 (2011) 000000

63

3. Results and discussion In this section the influence of substrate texture and temperature on the growth process of e-beam evaporated poly-Si thin films are investigated. Two approaches are distinguished: The solid phase crystallization of amorphous silicon by thermal annealing (3.1) and the direct growth of poly-Si (3.2). 3.1. Polycrystalline silicon by solid phase crystallization In order to achieve deeper insights into nucleation and growth during the SPC process by thermal annealing at 600C, the dependency of the crystalline fraction of the silicon layers from annealing time was measured by evaluation of optical micrographs in transmission mode. Crystalline parts of the silicon films can be easily distinguished from amorphous parts due to the much lower optical absorption.

Fig. 1. Temporal evolution of the crystalline fraction of initially amorphous silicon thin films during thermal annealing at 600C. As substrate either ZnO:Al-coated glass with a surface roughness of rms= 5nm (left side) or SiNx-coated glass with a surface roughness of rms= 0.3nm (right side) is used. The silicon was deposited at substrate temperatures of 200C (blue triangles) and 300C (black circles).

In Fig.1 the crystalline fraction versus annealing time is depicted for two different substrates, SiNx- and ZnO:Al-coated glass with surface roughness 0.3nm and 5nm, respectively, as well as for two different deposition temperatures, 200 and 300C. The time lag - i.e. the annealing time when first crystals occur is found to be strongly dependent on the deposition temperature. The higher the substrate temperature during deposition, the shorter the time lags for both kinds of substrates. There is also a remarkable difference between the two substrate types. While time lag and full crystallization time are very similar for both substrates at a deposition temperature of 200C, the temporal evolution of the crystalline fraction differs significantly between both substrates when the silicon was deposited at 300C. The different time lags during crystallization of silicon films deposited at different temperatures might be explained by a different structural disorder of the as-deposited amorphous silicon. Raman spectra indicate a larger intermediate-range disorder in amorphous silicon deposited at lower temperatures leading to a longer time lag [9]. By variation of the substrate temperature during silicon deposition selective nucleation layers can be implemented as nucleation and growth mainly starts in the high-T regions. The influence of the substrate texture on the grain size of fully crystallized poly-Si films can be seen in Fig.2. The silicon films were deposited at 300C and subsequently annealed at 600C for 20 hours. An increase of substrate surface roughness from 0.3nm to 29nm (a,c,e,g) decreases the poly-Si grain size from several micrometers to about 200nm (b,d,f,h). This behavior can be explained by a nucleation rate strongly increasing with substrate texture.

64

Author nameet / Energy Procedia 00 (2011) 000000 Christiane Becker al. / Energy Procedia 10 (2011) 61 65

Fig. 2. SEM images of various substrates (a,c,e,g, 30 view) and corresponding poly-Si thin-films grown on top by ebeam evaporation at Tdep=300C and subsequent thermal annealing at 600C (b,d,f,h, top view). A Secco etching step has been applied to the poly-Si in order to clarify single grains (dashed ovals). SiNx-coated glass with a surface texture of rms= 0.3nm (a, b) and ZnO:Al-coated glass partly flattened by a thin SiNx layer having a surface roughness of rms= 5nm (c,d), rms= 10nm (e,f) and rms= 29nm (g,h) were used as substrate. The figures b, d and f are taken from reference [9].

3.1. Direct growth of polycrystalline silicon Silicon by e-beam evaporation grows directly in crystalline phase at substrate temperatures above 400500C. The grains have a columnar shape and the grain size increases with deposition temperature [6]. However, also the substrate texture seems to influence crystal growth as shown in Fig.3. On a comparable rough ZnO:Al-substrates larger poly-Si crystals are growing than on smooth glass substrates (left side). The crystal size can be directly translated into photovoltaic performance (right side).

Fig. 3. Left side: SEM images of poly-Si thin films grown directly at temperatures between 500 and 700C on glass (rms< 0.3nm, upper row) and ZnO:Al (rms= 5nm, lower row, from [6]). Right side: Open circuit voltages of poly-Si thin-film solar cells by SPC (red symbols) and direct growth (purple symbols) treated by the post-deposition treatments hydrogen-passivation (HP) and rapid thermal annealing (RTA). The small symbols mark SiNx-, the big stars mark ZnO:Al-coated glass substrates.

Christiane Becker et al. / Energy Procedia 10 (2011) 61 65 Author name / Energy Procedia 00 (2011) 000000

65

For only hydrogen-passivated (HP) samples made by SPC, a rough ZnO:Al-substrate leads to a lower open circuit voltage than a smooth SiNx-substrate which could be explained by smaller grains as described in section 3.1. For directly grown poly-Si layers, however, there is a huge reverse difference of the Voc between a rough ZnO:Al-substrate (>400mV) and a smooth SiNx-substrate (250mV). 4. Conclusion In conclusion, the growth of e-beam evaporated poly-Si can be controlled by setting the substrate texture and temperature during deposition. By varying the deposition temperature Tdep between 150 and 300C the time lag, i.e. the start of nucleation, in a subsequent SPC step can be adjusted in a controlled way. The grain size of silicon deposited at 300C and subsequent SPC decreases from several micrometers to about 200nm by increase of the substrate surface roughness from rms= 0.3nm to 29nm. The grain size of directly grown poly-Si increases with increasing deposition temperature. Contrary to poly-Si by SPC the directly grown films dont exhibit the largest crystals and highest Voc values for the smoothest substrates but for a certain micro roughness. The finding of an optimum substrate texture for directly grown poly-Si, enabling high quality crystal growth in combination with light trapping, is therefore of big interest for photovoltaic applications. Acknowledgements The authors would like to thank J. Hpkes from Forschungszentrum Jlich, Germany, for providing the ZnO:Al layers and CSG Solar, Australia, for providing the glass substrates and the SiN coatings. A. Schpke from Helmholtz Zentrum Berlin is gratefully acknowledged for carrying out atomic force microscopy analyses. The work has been partially supported by the European Commission by the FP7 research project PolySiMode (Contract No. 240826). T. Sontheimer acknowledges SCHOTT AG for financial support. References
[1] Matsuyama T, Terada N, Baba T, Sawada T, Tsuge S, Wakisaka K, Tsuda S. High-quality polycrystalline silicon thin film prepared by a solid phase crystallization method . J. Non-Cryst. Solids 1996;940:198200 [2] Green MA, Basore PA, Chang N, Clugston D, Egan R, Evans R, Hogg D, Jarnason S, Keevers M, Lasswell P, OSullivan J, Schubert U, Turner A, Wenham SR, Young T. Crystalline silicon on glass (CSG) thin-film solar cell modules. Sol. Energy 2004;77:857. [3] Sontheimer T, Becker C, Ruske F, Klimm C, Bloeck U, Gall S, Kunz O, Young T, Egan R, Hupkes J, Rech B. Challenges and opportunities of electron beam evaporation in the preparation of poly-Si thin film solar cells. Proceedings of the 35th IEEE Photovoltaic Specialists Conference, Hawaii, USA, 2010, p. 614619. [4] Mattox DM. Handbook of physical vapor deposition (PVD) processing. Park Ridge, New Jersey: Noyes Publications; 1998 [5] Werner M, Schubert U, Hagendorf C, Schneider J, Keevers M, Egan R. Thin film morphology, growth and defect structure of ebeam deposited silicon on glass. Proceedings of the 24th European Photovoltaic Solar Energy Conference, Hamburg, Germany, 2009, p. 24822485. [6] Becker C, Ruske F, Sontheimer T, Gorka B, Bloeck U, Gall S, Rech B. Microstructure and photovoltaic performance of polycrystalline silicon thin films on temperature-stable ZnO:Al layers. J. Appl.Phys. 2009;106:084506. [7] Sontheimer T, Becker C, Bloeck U, Gall S, Rech B. Crystallization kinetics in electron-beam evaporated amorphous silicon on ZnO:Al-coated glass for thin film solar cells. Appl.Phys.Lett. 2009;95:101902. [8] Secco dAragona F. Dislocation Etch for (100) Planes in Silicon. J. Electrochem. Soc. 1972;119:948. [9] Sontheimer T, Scherf S, Klimm C, Becker C, Rech B. Characterization and control of the solid phase crystallization kinetics of electron beam evaporated silicon for thin film solar cells. Accepted for publication in J.Appl.Phys. 2011.