Laser-induced crystallization of Si films continues to present researchers with opportunities to investigate phase-transformation-related topics or to generate device-optimized Si films for various electronic applications. A historical overview of the field, as well as its current status with regards to fundamental understanding and technological progress, will be provided in this presentation. In addition to the excimer-laser-based methods that are presently being employed to manufacture advanced LCDs and AMOLED displays (and which are poised to further evolve and grow with the application of the techniques for making large-AMOLED TVs), we will also discuss a cw-laser/radiant-beam-based crystallization technique, referred to as mixed-phase solidification (MPS), which may eventually permit a seed-layer-based fabrication route for manufacturing crystalline-Si-film-based high-efficiency solar cells on low-cost large-area substrates. Regarding pulsed-laser irradiation of Si films, we will focus on presenting the recent findings that pertain to the melting and solidification dynamics (within the partial-melting regime) of PECVD-deposited amorphous films and "pre-crystallized" polycrystalline Si films. These findings, in particular, have profound implications on the potential ways through which the multiple-pulsed-based excimer-laser-annealing (ELA) technique may be optimized and adopted for making large displays (e.g., via location-controlled firing of a short-axis spatial-profile-optimized line beam). Regarding the MPS method, we will focus on describing the basic mechanisms and fundamental factors that enable the approach to generate large-grained and highly (100)-surface-textured Si films on SiO₂.

Solid phase crystallised thin-film silicon solar cells on glass substrates are limited in their efficiency due to inherently high defect density associated with processing temperature constraints of the glass. Large-area diode laser crystallisation provides a pathway to achieving low defect densities and higher cell efficiencies than solid phase crystallised silicon while maintaining the cost advantage of thin-film crystalline silicon on cheap glass substrates. This work demonstrates the feasibility of large-area diode laser crystallised silicon films for solar cell applications. An 808 nm cw diode laser with 11 mm line width, an optical power density up to 0.15 MW/m2 and a scan speed up to 20 mm/s is used to crystallise a few micron thick electron beam evaporated a-Si films on borosilicate glass substrates. The process parameters for diode laser crystallisation are investigated and a regime of continuous lateral growth is found, whereby the crystal growth front is seeded by the preceding crystallised region forming long parallel crystals in the direction of laser scanning. Another mode is found which results in much smaller crystals grown perpendicular to the scan direction. The grain orientation distributions are analysed by electron backscatter diffraction (EBSD) and the samples which showed lateral growth are found to contain a large number of ?3 twin boundaries and very few other boundaries. In contrast the samples with perpendicular growth show random grain orientation and random boundary angles similar to those created by solid phase crystallisation. The intra-grain defect density is analysed by transmission electron microscopy (TEM) and found to be at most 2E9/cm2 for both growth modes of the diode laser crystallised silicon which is an order of magnitude lower than solid-phase crystallised silicon. The grain size is investigated with scanning electron microscopy (SEM) and the laser crystallised samples with lateral growth are shown to have very large grains ranging from 100 nm - 200 µm in width and up to millimetres in length, the laser crystallised samples with perpendicular growth have grains 1 - 10 µm in size, while solid-phase crystallised silicon samples have grains 0.5 - 5 µm in size.

Amongst th¬e thin-film based approaches for photovoltaics, which aim to combine high conversion efficiencies (> 10%) and low cost manufacturing (< 1$/Wp), poly-crystalline silicon (poly-Si) based solar cells (< 10 ?m thick) are nowadays considered a promising candidate. One of the most followed approaches is the development of poly-Si on inexpensive (e.g. glass) substrates upon solid-phase crystallization (SPC), i.e. an annealing procedure under a controlled temperature ramp up to 600-650 °C, of plasma- deposited amorphous silicon (a-Si:H) films. Here the technological challenges are the development of large area, high growth rate a-Si:H and development of large (in the range of few ?m) grain poly-Si. Among the research questions, the impact of the a-Si:H microstructure on the crystallization kinetics (incubation-nucleation-grain development) and on the grain size development represent the key towards large grain poly-Si and process up-scaling. We report here selected insights into the crystallization of a-Si:H by addressing the above mentioned questions. Experiments have been performed by means of the expanding thermal plasma (ETP), a remote plasma deposition technique that allows for an independent parameter variation with a large freedom in operating conditions.1,2 State-of-the art poly-Si layers characterized by grain sizes developing up to a 1 ?m in diameter were obtained upon SPC of ETP deposited a-Si:H layers.1 In detail, larger grains are found to be promoted by an increase in the a-Si:H microstructure parameter R*.2 Next to the role of R* in controlling the grain size, the morphology of a-Si:H films also affects the grain size development in poly-Si films, as concluded by studies related to the crystallization of a-Si:H films deposited at ultra high growth rates (11- 58 nm/s). Furthermore, poly-Si layers characterized by large grains (~1.5 µm) were obtained from disordered a-Si:H films deposited at 11- 25 nm/s.3 Finally, the incubation time towards nuclei formation appears to be not only controlled by the medium range order in the a-Si:H films, as previously reported in literature,4 but also by the density of nano-sized voids which undergo a faster hydrogen out-diffusion and chemical bond rearrangement towards a higher medium range order and more ordered microstructure.5 Reference list: [1] A. Illiberi, K. Sharma, M. Creatore, M.C.M. van de Sanden, Matt. Lett., 63, 1817-1819 (2009). [2] K. Sharma, A. Branca, A. Illiberi, F. D. Tichelaar, M. Creatore, and M. C. M. van de Sanden, Adv. Energy Mater, 1, 401-406 (2011). [3] K. Sharma, M. V. Ponomarev, M. A. Verheijen, O. Kunz, F. D. Tichelaar, M. Creatore, and M. C. M. van de Sanden, J. Appl. Phys., Accepted (2011). [4] A. H. Mahan, T. N. Su, D. L. Williamson, L. M. Gedvilas, S. P. Ahrenkiel, P. A. Parilla, Y. Q. Xu, and D. A. Ginley, Adv. Funct. Mater. 19, 2338 (2009). [5] K. Sharma, M. A. Verheijen, M. C. M. van de Sanden, M. Creatore, J. Appl. Phys., Accepted (2011)

Optical microscopy has been used to examine crystallite nucleation and growth in thermally annealed a-Si thin films (1). When PECVD films containing 10-12 at. % H films were periodically laser ablated in an evenly spaced grid pattern prior to film annealing, it was found that the crystallite nucleation rate (rn) was significantly retarded near these grids. µ-Raman spectroscopy measurements on partially crystallized films have shown, from the position of the c-Si TO phonon peak, that the film tensile stress is high in film areas far away from these grid patterns, but is relieved near these grid patterns, thus enabling for the first time an examination of how film stress affects crystallite nucleation and growth in the same film (2). It has been suggested that these laser ablated grid patterns interrupt the film connectivity, thus enabling the film to structurally relax. In this work, we stepwise anneal our PECVD films at anneal temperatures between 540-600C, and count the crystallites that are observed to appear versus anneal time to enable a determination of rn at each anneal temperature. Such temperature dependent measurements enable a determination of the nucleation rate activation energies (EA), and how they are affected by film stress. We find that while the EA in our stressed film areas is similar to that of Lee et al. (3) for their PEVCD films, our EA is significantly increased when the film stress is relieved. This suggests that either the energy barrier height at the amorphous-crystalline interface and/or the size of the critical nucleus increases with stress relief, both of which could contribute to the lower rn. We also present measurements on higher H content PECVD films and on HWCVD films containing similar (10-12 at. %) H contents to illustrate how variations in initial film properties affect stress relief, and its consequent crystallite nucleation and growth. Information on the relative crystallite growth rates will also be presented. Finally, we present a tentative model for how stress relief in our 10-12 at. % H PECVD films extends laterally over such large areas (100-200 µm in a 0.1 µm thick film). (1) A.H. Mahan, M.S. Dabney, R.C. Reedy Jr., D. Molina, and D.S. Ginley, J. Appl. Phys. 2011 (in press). (2) See M.S, Dabney et al., this conference. (3) J.N. Lee, B.J. Lee, D.G. Moon, and B.T. Ahn, Jpn. J. Appl. Phys. 36 (1997) 6862.

The common approach to improve the surface texture in hydrogenated microcrystalline silicon (?c-Si:H) thin-film solar cells is to vary the process parameters and to compare solar cell results. The characterization of the texture itself leads to simplified quantities like the root mean square roughness, haze, etc. For the modeling of light scattering properties, scalar approaches are frequently used by taking into account statistical values instead of local structures of the texture. By applying statistical approaches, a detailed knowledge about the light scattering process cannot be achieved and, therefore, the success to improve the solar cell efficiency will be limited. Therefore, we present optical simulations of µc-Si:H solar cells based on the real three-dimensional structure of the whole layer stack. The simulations are done by Finite-Difference Time-Domain method. Different aspects are in the focus of interests. The extraction of the local light trapping efficiency of individual scatterers makes it possible to get rid of statistical analysis with the objective of identifying the best surface features. The modification of the surface morphology due to the growth of ?c-Si:H on top of the front contact layer (ZnO:Al) by plasma-enhanced chemical vapor deposition is an important issue. In the past, the front texture is assumed to be conformally transferred to the back-side in optical simulations. But it is found that the total absorption is significantly improved due to the non-conformality. We compare the situation with conformal interfaces of both, front texture and back texture, to non-conformal interfaces, where we take into account the real surface modifications due to the silicon deposition at the same location. From this study, we will address the following questions: Which texture has the higher light trapping potential, front texture or back texture? At which interface is light scattering more effective, front side or back side? Is it reasonable anyway to separate the two interfaces for the interpretation? Which surface features will provide the best light trapping?

In thin film silicon, voids, i.e. volume parts which are empty or have a strongly reduced density are an important defect and may cause electronic defect states. One may distinguish between isolated and interconnected voids. Recently attention was drawn to the presence of isolated voids as it was proposed that most Si-bonded hydrogen in device grade hydrogenated amorphous silicon (a-Si:H) is incorporated in crystalline silicon type of divacancies (Ref.1), i.e. in well defined isolated voids. Here we study the presence of isolated voids in various a-Si:H type materials by effusion of implanted helium. Isolated voids are detected by an effusion peak of helium occurring at high temperatures (> 600°C) (Ref.2), i.e. at temperatures at which He incorporated in bulk material should have diffused out. A-Si:H films prepared by various techniques like plasma deposition, hot wire deposition, sputtering and e-beam evaporation at substrate temperatures between 200 and 300°C all show the presence of isolated voids. We do not find the correlation between void concentration and hydrogen content predicted by Ref. 1. The presence of isolated voids in alloys of a-Si:H with C, N, O and Ge was also studied. Isolated voids generally disappear when interconnected voids show up. Similar as for silicon (Ref.3), we find for crystalline and amorphous germanium which was hydrogenated (at the same level) by hydrogen implantation a significant difference in the concentration of isolated voids: in a-Ge:H the concentration of isolated voids is found to be more than an order of magnitude smaller than in (hydrogenated) crystalline germanium. These results suggest that for both a-Si:H and a-Ge:H, hydrogen in dense material resides predominantly in cavities which do not trap helium, i.e. which are smaller in size than divacancies. The nature of isolated voids in amorphous silicon materials will be discussed. 1. A.H.M. Smets, C.R. Wronski, M. Zeman and M.C.M. van de Sanden, MRS Symp. Proc. 1245 (2010) 303-308 2. W. Beyer, Physica Status Solidi (c) 1 (2004) 1144-1153 3. W. Beyer, W. Hilgers, P. Prunici, D. Lennartz, Proceedings ICANS 24, August 21-26, 2011, Nara, Japan, J. Non Cryst. Solids, to be published.

In this contribution we present additional proof that divacancies play an important role in the Staebler-Wronski effect. The network and nature of hydrogenated amorphous silicon (a-Si:H) has been extensively interpreted in terms of a continuous random network (CRN) in which the isolated dangling bond (a coordination defect) is considered as the dominant defect type. Wronski et al. identified at least three defects states in the sub gap, referred to as A, B, and C-states having different creation kinetics under light soaking (fast degradation: A and B; slow degradation: B and C) [1]. In the last years, we have made the case that NOT coordination defects but ones related to volume deficiencies, like not fully H passivated divacancies, are plausible candidates to represent the dominant A, B and C defect states in a-Si:H [2]. Using carrier-lifetime measurements de Wolf et al. [3] recently showed that under light soaking the generation of fast metastable defects states can be monitored in great detail at the interface of a-Si:H and n-type c-Si(100). Here we report results in which the characterization of the fast metastable defect states at a-Si:H/c-Si(100) interface, using carrier-lifetime measurements was carried out on both n- and p-type wafers. In these experiments the light soaking times covered 10 orders of magnitude timescale (10 ?s up to several hours) while the temperature dependence was studied up to 400 C. The observed evolution of fast metastable defects with light soaking has the exact same scaling as observed for the A and B states in the a-Si:H bulk [1]. Considering the fact that plasma processing on c-Si(100) surfaces creates a large density of divacancies at the interface, the similarity between a-Si:H bulk and a-Si:H/c-Si(100) interfaces is additional proof that divacancies play a crucial role in the formation of fast metastable defect states. Furthermore, under light soaking the recombination velocity at the a-Si:H/n-doped-c-Si(100) interface is enhanced, whereas it is reduced at the a-Si:H/p-doped-c-Si(100) interface, while the absolute change in recombination velocities is independent of the doping of the wafer. This clearly demonstrates that the fast metastable defects are negatively charged and supports our claim that negatively charged divacancies correspond to the fast metastable A-state. In addition, the significant reduction in the light induced metastable defect generation when the wafers are annealed above 350 C further supports the divacancy model, since at these elevated temperatures divacancies in silicon become unstable and agglomerate into larger volume deficiencies. [1] C. Wronski, J. Deng, X. Niu, and A.H.M. Smets, Proceedings of the 35th IEEE PVSC conference, pp. 146 (2010). [2] A.H.M. Smets, C.R. Wronski, M. Zeman and M.C.M. van de Sanden, Mat. Res. Symp. Proc. Vol. 1245-A14-02 (2010). [3] S. De Wolf et al., Phys. Rev. B 83, 233301 (2011).

While prevalent in low-power/low-cost applications, the limited efficiency of hydrogenated amorphous silicon (a-Si:H) photovoltaic (PV) devices has precluded the adoption of the material as a bulk absorber layer in large scale PV installations. A major factor contributing to the low efficiency of a-Si:H based cells is the poor hole mobility, arising from the substantial hole trapping in the inherently disordered material. Through statistical analysis of an ensemble containing over two thousand computational a-Si:H structures, we have acquired insight into the geometric signals of hole traps, including a correlation between trap strength and atomic rearrangement under the presence of a hole. Through the application of stress, we demonstrate the feasibility of attenuating these traps, observing a trend of improving mitigation of strong trap states under increasing stress application, as well as a dependence of the improvement on the dimensionality of the stress.

In Electrically Detected Magnetic Resonance (EDMR), paramagnetic states are detected via their participation in electronic transport processes such as recombination. This detection method is significantly more sensitive than conventional electron spin resonance. Nevertheless the reported sensitivity strongly depends on the type of sample, the contact geometry and the specific transport mechanisms. In this contribution, we systematically study the dependence of the EDMR signal amplitude on the contact geometry and the bias voltage for a-Si:H and ?c-Si:H thin film samples at different temperatures. We find that the EDMR signal measured in sandwich geometry is up to a factor of 50 smaller than in coplanar test structures of the same material. Using the Kaplan-Solomon-Mott and the Lepine model for EDMR we will demonstrate that the charge transport properties of the a-Si:H films are the origin of the observed differences. In particular, the ratio between the spin flip time induced by resonant microwaves and the electron transit time between contacts has a strong influence on the EDMR signal amplitude. This work is funded by BMBF (Contract number 03SF0328B, EPR-Solar).

United Solar has used roll-to-roll manufacturing technology to make a-Si:H/a-SiGe:H/a-SiGe:H triple-junction solar laminates on flexible thin stainless steel substrates. A typical laminate has a power rating of 144W, corresponding to an aperture area stable efficiency of 8.2% [1]. In order to increase the efficiency further, we have extensively carried out research and development of a-Si:H and nc-Si:H multi-junction solar cell technology since 2001 [2-5], and made significant progress in improving cell and module efficiencies. We have achieved (i) a 16.3% initial active-area (~0.25 cm2) solar cell efficiency using an a-Si:H/a-SiGe:H/nc-Si:H triple-junction structure [3], (ii) an NREL measured stable total-area (~0.25cm2) efficiency of 12.5% [4], and (iii) NREL measured initial and stable aperture-area (~400 cm2) module efficiencies of 12.0% and 11.4%, respectively [5]. These represent new record efficiencies for thin-film silicon solar cells. Based on these achievements, we have started implementing roll-to-roll manufacturing technology using a-Si:H and nc-Si:H multi-junction structures on flexible substrates. In this presentation, we will review nc-Si:H material optimization and characterization, multi-junction solar cell design and fabrication, light-trapping and light management in a-Si:H and nc-Si:H multi-junction configurations, and discuss challenges and opportunities for thin film silicon photovoltaic technology. [1] J. Yang, A. Banerjee, and S. Guha, Sol. Energy Mater. Sol. Cell, 78, 597 (2003). [2] B. Yan, K. Lord, J. Yang, S. Guha, J. Smeets, and J-M. Jacquet, Mater. Res. Soc. Symp. Proc. 715, 629 (2002). [3] B. Yan, G. Yue, L. Sivec, J. Yang, S. Guha, C.-S. Jiang, Appl. Phys. Lett. 99, 113512 (2011). [4] G. Yue, L. Sivec, J. M. Owens, B. Yan, J. Yang, and S. Guha, Appl. Phys. Lett. 95, 263501 (2009). [5] A. Banerjee, T. Su, D. Beglau, G. Pietka, F. Liu, B. Yan, J. Yang, and S. Guha, Mater. Res. Soc. Symp. Proc. 1321, 3 (2011).

Sanyo has been conducting research and development on thin-film Si solar cells for over 30 years. The world's first a-Si solar cell, called AMORTONTM, and the hetero-junction with intrinsic thin-layer solar cell, called HITTM, which has achieved the world's top level of conversion efficiency were produced by Sanyo. For the next-generation solar cell, Sanyo has targeted the high-efficiency a-Si/?c-Si tandem solar cell, and developed a fabrication technology for high-quality, device-grade ?c-Si thin-film with a higher deposition rate. A stabilized module efficiency of 10.0% was reported with a very high deposition rate of mc-Si thin-film (2.4 nm/s) on a large-area glass substrate (1,100×1,400 mm2).[1] We have achieved the world's highest position for tandem solar module performance with a combination of conversion efficiency, deposition rate of ?c-Si thin-film, and substrate size. In order to promote this position, we have been studying the very high-rate deposition of high-quality, device-grade ?c-Si thin-film by Localized Plasma Confinement (LPC) CVD to eliminate the production bottleneck. By applying the essence of this technique to full-size CVD equipment, we succeeded in achieving high-quality, uniform ?c-Si thin-film with a very high deposition rate. To develop higher quality ?c-Si thin-film, we are going to study the reaction process of CVD combining an optical emission study and plasma simulation technique. Due to the progress of our research, we achieved the world's highest stabilized conversion efficiency of 12.2% for small-size cells (1 cm2) and the world's top level of stabilized module conversion efficiency of 10.7% for large-area glass substrates (Gen. 5 class), respectively.[2] These conversion efficiency values were improved by advanced techniques, such as material design, optical confinement and laser patterning. These modules already comply with the IEC 61646 standard and, to investigate the possibility of further long-term reliability, more severe reliability tests are being conducted. In the Damp Heat test, the module performance was not degraded at a level of over 7 times higher than that of the standard. References [1] M. Matsumoto, A. Kuroda, H. Katayama, T. Kunii, K. Murata, M. Hishida, W. Shinohara, I. Yoshida, A. Kitahara, H. Yoneda, Y. Aya, A. Terakawa, M. Iseki and M. Tanaka: the Journal of Photovolatics (to be published). [2] A. Terakawa, M. Hishida, S. Yata, W. Shinohara, A. Kitahara, H. Yoneda, Y. Aya, I. Yoshida, M. Iseki and M. Tanaka: 26th EUPVSEC (2011) 3BO.4.2 (to be published). Acknowledgments This work was partially supported by NEDO as a part of the New Sunshine Program under METI.

Over the past decade, the PV industry has witnessed tremendous growth in manufacturing scale and technology advancement, with PV generated electricity cost ever approaching grid parity. As the market continues to expand geographically as well as in size, technology innovation in all aspects of the industry including materials, processes and equipment will remain as one of the key factors driving the industry growth. Si based thin film technology has made substantial progress in demonstrating its inherent advantages in lower material cost, ease of manufacturing, higher energy yield, etc. More recently, reduced product prices and competing technologies from crystalline silicon and other thin film technologies have made amorphous and microcrystalline silicon based thin film technology very challenging, and require further increase in module efficiency and decrease in manufacturing cost. As one of the few companies in the world with significant manufacturing capacity for thin film Si PV products, Astronergy has been at the forefront of technology development for the mass production of large-scale (Gen. 5, 1.43m2) Si thin film solar modules in the last 5 years. We will review major technology advancements which have led to the mass produced tandem silicon thin film module "Micromorph" with 10.0% plus stabilized efficiency and high performance in term of temperature coefficient and low light performance. We will also discuss major trends in technology development and near term criteria for thin film Si to become a preferred technology over crystalline Si in the market place.

Manufacturers of PV modules have been aiming for large scale production, efficiency improvements and a lower consumption of raw materials to reduce production costs. In case of hydrogenated amorphous and microcrystalline silicon thin-film tandem solar cells (a-Si:H/µc-Si:H) most of the improvement has been obtained by increasing the efficiency through improved light-trapping and reducing the intrinsic layer thickness. Microcrystalline silicon makes up for about 80 % of a tandem cell thickness. Hence, its deposition rate has a very large impact on the overall deposition time. Results published within the last years indicate that the very high frequency plasma enhanced chemical vapour deposition (VHF-PECVD) process can obtain deposition rates of 3 nm/s without compromising on solar cell performance. The time needed to grow the corresponding µc-Si:H absorber layer can therefore be reduced to less than 10 minutes. However, for an increasing frequency the effect of standing waves leads to a non-uniform deposition on large areas. Several solutions have been suggested in order to deal with this problem (e.g. lens shaped electrode, ladder-shaped electrode, array antenna). In the present work we use linear plasma sources and a dynamic VHF-PECVD deposition process. The linear plasma sources are scaled up in only one dimension with the number of power feeds being adapted to the length of the electrode and the operating frequency. It is thus possible to ensure a good homogeneity along the electrode for high frequencies. In order to deposit homogeneously on a large area the substrate moves past the linear plasma sources at a constant speed (similar to a roll-to-roll process). For intrinsic µc-Si:H we obtained deposition rates from 0.6 nm/s up to 1.4 nm/s. The corresponding deposition time for the solar cell absorber layer was reduced to 12.5 minutes without compromising on solar cell performance. The solar cell efficiencies of the optimized cells reached 7.5-8.0% and were independent of the deposition rate. For intrinsic a-Si:H we investigated deposition rates up to 1.1 nm/s. The homogeneity of the absorber layer thickness was typically better than ± 5 %. Stabilized solar cell efficiencies of 7.6 % to 6.1 % at deposition rates ranging from 0.34 nm/s up to 1.0 nm/s have been obtained. The influence of the deposition rate on the intrinsic layer properties and solar cells performance has been studied for RF- and VHF-PECVD in order to investigate the influence of the excitation frequency. We found that at a given deposition rate the a-Si:H solar cells incorporating intrinsic layers deposited with VHF-PECVD outperformed solar cells incorporating intrinsic layers deposited with RF-PECVD by about 0.2 %.

Best performing thin film silicon solar cells currently use tandem or triple junction configurations, based on amorphous (a-Si:H) and microcrystalline (µc-Si:H) silicon and their alloys. To get the most out of those structures, efficient light trapping is mandatory. However defective growth of µc-Si:H material, as can typically be observed at high deposition rates on highly textured surfaces, leads to severe electrical performances losses, hence discarding potentially better morphologies for light trapping. In this contribution we address the specific issue of obtaining high quality and resilient µc-Si:H material through PECVD on challenging morphologies. Up to now FTPS and FTIR were favored opto-electrical characterization tools to determine the bulk quality of the deposited intrinsic µc-Si:H material. However we will show that they are not sufficient as they do not account for extrinsic defects, defined here as the defective porous regions (also called cracks), which develop where the growth fronts encounter each other during the film deposition on textured surfaces. We find that apparently similar high-quality materials in terms of bulk electronic defects, as assessed by these characterization methods, can lead to large discrepancies in the solar cells performances. This is attributed to the presence of those porous regions which are found to be significantly more sensitive to the plasma process than the bulk phase. A two-phase material model is presented, taking into account the bulk material quality and its intrinsic defects along with the presence of those extrinsic defects which can both drive the cell performances. The model is validated by experiments showing how substrate roughness and plasma conditions affect these two different phases, and how the unencapsulated cells react to damp heat exposure. We show that it is possible to find plasma processes where the bulk phase is worse, but where the crack quality is improved, leading to better performing devices. Eventually, we'll show that, considering growth on rough substrate and the sensitivity of plasma processes to the porous phases, a precise control of interfaces is of tremendous importance for achieving high performances and reproducibility. In particular, special attention is usually given to the p/i interface as it is known to be a determinant factor for achieving high Voc devices. We will show here that the typical trade-off can be lifted and that a highly crystalline p/i interface while retaining a high Voc is achievable. Thanks to these new understandings, we have developed very high efficiency microcrystalline silicon p-i-n solar cells in an industrial type reactor with efficiencies well above 10%, while sustaining good performances on very rough as-grown LPCVD ZnO as well. We'll report on these devices and on the opportunities created for the realisation of higher performance micromorph devices as well as on the use of more aggressive light trapping structures.

Hydrogenated nanocrystalline silicon (nc-Si:H) has shown a significant advantage over amorphous silicon (a-Si:H) and amorphous silicon germanium (a-SiGe:H) alloys as the absorber layer of the bottom cell in multi-junction solar cells because of high photo-current density and low light-induced degradation. Normally, nc-Si:H solar cells not only have much higher photocurrent density than a-Si:H and a-SiGe:H solar cells, but also higher photocurrent density than c-Si solar cells with the same thickness. Besides the great contribution from light trapping using proper back reflectors (BR), the bi-phase structure consisting of amorphous tissue and nanometer sized crystallites is also an important factor for effective light scattering. With the extensive study of nc-Si:H material quality and light trapping schemes, state-of-the-art nc-Si:H solar cells show a short circuit current density (Jsc) in the range of 24-27 mA/cm2 [1-3]. We have reported on optimizing the Ag/ZnO BRs for high efficiency nc-Si:H solar cells and achieved Jsc=29.2 mA/cm2 [4]. In this paper, we report our recent improvement of nc-Si:H solar cells. We use modified very high frequency glow discharge to deposit nc-Si:H solar cells. We optimized Ag/ZnO BRs for effective light trapping. We find that a textured Ag layer and a thin ZnO layer is the best structure at this moment for nc-Si:H solar cells. Low texture on the Ag surface can produce sufficient light scattering. The low surface texture also reduces the impact of substrate texture on the quality of the subsequently deposited nc-Si:H. A thin ZnO layer is used as a buffer layer to shift the plasmonic resonance frequency to reduce plasmonic absorption. In addition, we have improved the quality of the nc-Si:H material so that we can use a relatively thick intrinsic layer. Combining the optimized Ag/ZnO BR and the improved nc-Si:H material, we made nc-Si:H single-junction solar cells with Jsc=28.0 mA/cm2 with an intrinsic layer thickness of 1.9 ?m and Jsc=30.5 mA/cm2 with an intrinsic layer thickness of 3.3 ?m. Comparing solar cells on specular stainless steel, the gain in Jsc by the Ag/ZnO BR is around 50-60%, which is much higher than gains reported in the literature. We compared the quantum efficiency (QE) to the estimations using the classical limits of 4n2 by consideration of the losses from ITO reflection and absorption in the doped layers and BRs. We find the QE curves are close to the theoretical estimation, indicating that our Ag/ZnO BR is very effective for light trapping. We will present the details of Ag/ZnO optimization and nc-Si:H improvements. Furthermore, we will report the recent progress of our work with a-Si:H and nc-Si:H based multi-junction solar cells. [1] H. Sai, et al., Appl. Phys. Lett. 93, 143501 (2008). [2] P. Cuony, et al., Appl. Phys. Lett. 97, 213502 (2010). [3] M. Berginski, et al., Sol. Energy Mater. Sol. Cells, 92, 1037 (2008) [4] G. Yue, et al., Appl. Phys. Lett. 95, 263501 (2009).

Good electronic transport in solar cells based on thin amorphous (a-Si) or microcrystalline (?c-Si) silicon films requires thin active layers, resulting in poor light absorption of long wavelengths and consequently lowered short circuit current density (Jsc). Light scattering at interface textures is the most successfully used approach to enhance the absorption and thus solar cell efficiency. However, the textured interfaces can lead to the growth of defective material which limits the cell efficiency by decreasing the open circuit voltage (Voc) and the fill factor (FF) compared to flat reference cells. In this contribution we propose new approaches for the design of back reflector used as substrate in the growth of thin film silicon solar cells. These substrates allow the decoupling of electrical and optical performances. In the first part, we present silver based back reflectors with low roughness ?rms that allow growing good material quality while being still sufficiently rough for light scattering and absorption enhancement in the absorber. This type of substrate is thus "electrically flat but optically rough" and yields high Voc and FF as well as increased Jsc. In the spectral range between 530 and 810 nm, co-deposited single junction a-Si cells on substrates with ?rms=9, 13 and 17 nm show an increased current of 12, 18 and 21% respectively when compared to the flat reference (?rms=3 nm). Due to the low roughness of these substrates, Voc and FF are comparable to the flat cell. Finally, an initial cell efficiency of 10.3% (Voc=955 mV, FF=72.4%, Jsc=14.87 mA/cm2) for 220 nm of intrinsic layer is obtained, one of the best value obtained for an n-i-p single junction cell of that thickness in our lab. Tandem a-Si/a-Si cells with initial (stabilized) efficiency of 10.5% (8.5%) on glass and 10% (8.2%) on plastic substrates will also be presented. In a second approach, better suited for ?c-Si cells, a variation of the FLiSS (flattened light scattering substrate) concept recently discussed by Sai et al. (Appl. Phys. Lett. 98, 113502 (2011)) is presented; we use a flat Ag layer covered with the pyramidal texture of LP-CVD ZnO which is then coated with a dummy a-Si layer. The a-Si film is then polished until the tips of the ZnO are exposed. Different to the FLiSS concept, we use an undoped a-Si layer to limit the parasitic absorption in this layer and the current is collected by point contacts through the exposed tips of the ZnO film. The resulting ?c-Si cells are flat and keep high Voc and FF even for thick absorber. We obtain an efficiency of 9.5% (Voc=520 mV, FF=67%, Jsc=27.3 mA/cm2) for a thick cell of 3.9 ?m. This promising result proves the feasibility of this approach, yielding again substrates that are electronically flat but optically rough. Such substrates are good candidates for the realization of record efficiency triple junction cells a-Si/?c-Si/?c-Si or a-Si/a-SiGe/?c-Si.

Thin film silicon solar cells can be a low cost alternative to bulk silicon based silicon solar cells. Especially multiple junction cells are a very promising approach to low cost, high efficiency photovoltaics. Normally, glass is used as a substrate for these type of cells. As an alternative to these glass substrates we aim to use cheap plastics, which demands cell processing below the deformation temperatures of the plastics. One of the issues with the processing on plastics is the longer deposition times due to low deposition rates associated with deposition with high hydrogen dilution of the silane precursor gas that is necessary for obtaining device quality material at low temperatures [1]. This time factor becomes even more severe when we consider multijunction cells. We propose to solve this problem by reducing the total thickness of the cell to around 1000 nm. In this study we fabricated very thin silicon tandem cells in a p-i-n configuration on glass at a substrate temperature of 130°C, which is low enough for polycarbonate substrates. As a front TCO we used texture etched aluminium doped zinc oxide (ZnO:Al), which was sputter deposited on glass. As a back contact we used evaporated silver, after sputter depositing a ZnO:Al back reflector. A double p-layer (nc-Si/a-Si:H) was used to make proper contact with ZnO:Al front TCO. We used very thin layers as the active layers: 275 nm of amorphous silicon for the top cell combined with a nanocrystalline silicon bottom cell with an i-layer thickness of only 900 nm, where normally 1500 to 2000 nm or even more is used. Spectral response measurements show that there is good spectral splitting due to the combination of a relatively high band gap (1.9eV) top cell and 1.1 eV bottom cell. The high bandgap top cell is a consequence of the low deposition temperature. This configuration allows us to achieve a very high Voc while using a broad spectral range. Moreover, we can greatly reduce the thickness of the bottom cell, which has a number of advantages. First of all it results in a decrease in material usage. Secondly, the thinner layers mitigate the deleterious effect of relatively higher defect density (resulting from deposition at lower than optimum temperature) on FF and Voc. Thirdly it decreases the deposition time. In this concept, the deposition time for all i-layers together is less than 1 hour. Fourthly, thinner layers induce less stress on the substrate, which is a very important property when using plastics as a substrate The thicknesses are confirmed through XTEM studies. Using this configuration we were able to fabricate a tandem cell with an initial conversion efficiency of 9.5%. This bottom cell limited tandem cell had a Jsc of 10.5 mA/cm2, a high Voc of 1.40 V and a FF of 65%. We are conducting further studies to transfer this concept onto plastic substrates. [1] J.K. Rath, M.M. de Jong, A. Verkerk, M. Brinza, R.E.I. Schropp, Mater. Res. Soc. Symp. Proc. Vol. 1153 (2009) 1153-A22-04.