Manufacturing Technology of Amorphous and Nanocrystalline Silicon Solar Cells

Subhendu Guha, United Solar Ovonic, 3800 Lapeer Road, Auburn Hills, Michigan 48326, USA

Abstract-- The last two decades have witnessed tremendous progress in the science and technology of amorphous and nanocrystalline silicon-based photovoltaic. Advances in the understanding of materials and devices have led manufacturers to expand their production capacity; the production of solar panels based on this technology exceeded 85 MW in 2006. In this paper we shall review the properties of the optimum material for device application, the cell design to obtain high efficiency, the manufacturing technology and the production status.

I. INTRODUCTION

Hydrogenated amorphous silicon alloy (a-Si:H) has received a great deal of attention as a candidate for low-cost solar cells. Amorphous semiconductors absorb sunlight very efficiently because of the inherent disorder, and only a very thin film (<500 nm) is needed to complete the solar cell structure. The material cost for the solar cell is thus low. Innovative manufacturing processes have also been developed to reduce production cost, and a-Si:H technology is now extensively used for the production of solar cells and panels for a variety of applications [1]. The industrial focus has been mainly on single- and multi-junction cells using a-Si:H and amorphous silicon germanium (a-SiGe:H) alloys. Recently, hydrogenated nanocrystalline silicon (nc-Si:H) has been receiving a great deal of attention as a low-cost replacement for a-SiGe:H [2]. The superior long wavelength response of this material offers the possibility of reaching higher efficiencies, and several manufacturers have announced the construction of manufacturing plants incorporating this technology.

a-Si:H alloy is usually prepared by glow-discharge decomposition of silane. It was first shown by Guha et al. [3] that superior film properties could be obtained by using a dilute mixture of silane in hydrogen. As the hydrogen dilution is increased, the transition from amorphous to nanocrystalline phase takes place. Recent work has shown that the best amorphous silicon is grown at a dilution just below the edge of amorphous to nanocrystalline transition [4]. On the other hand, the highest efficiency nanocrystalline solar cells are made when the hydrogen dilution pushes the material just above the edge [5]. The material near the edge thus has received a great deal of attention. In this paper, after briefly discussing the material properties of thin-film silicon in this deposition regime, we shall describe the cell design and the manufacturing technology of solar cells using this material. We shall also discuss the products and future directions.

II. MATERIAL AND CELL RESEARCH IN THE AMORPHOUS REGIME

a-Si:H alloy is a very intriguing material. The lack of long-range order and the presence of dangling bonds and other defects give rise to states in the mobility gap that hinder carrier transport. In addition, exposure to light creates metastable states in the gap that degrade material and cell performance. An enormous effort has gone into obtaining a better understanding of these phenomena. Since the first observation [3] of improvement of stability of film properties after light exposure in films grown with hydrogen dilution, the technique has been used extensively for growing both a-Si:H and a-SiGe:H. It is generally believed that increasing the hydrogen dilution improves the hydrogen coverage of the growing surface. The impinging SiH3 species, therefore, diffuse over a larger distance before being incorporated in the film, resulting in an improved order. In fact, hydrogen dilution beyond a certain limit can lead to even better order, resulting in the formation of nanocrystalline material. The best quality material is grown at a hydrogen dilution just below the threshold of the amorphous-to-nanocrystalline transition [4]. The material has an improved order, as confirmed by Raman, transmission electron microscopy, and x-ray diffraction studies. The improved order of this proto-crystalline or “on the edge” (OTE) material is considered to be responsible for the observed superior stability against light exposure.

Fig. 1. Schematic of triple-junction cell

In order to obtain high efficiency with good stability, a spectrum-splitting triple-junction structure (Fig. 1) has been developed. The top cell, which uses a-Si:H

978-1-4244-1728-5/07/$25.00 ©2007 IEEE

material, has a bandgap of about 1.8 eV and absorbs the blue photons. The middle and bottom cells, using a-SiGe:H alloys of different Si to Ge ratios, have bandgaps of around 1.6 and 1.4 eV and absorb green and red photons, respectively. The nine-layer, triple-junction stack is deposited onto a thin, flexible stainless steel (SS) substrate coated with textured silver/zinc oxide (Ag/ZnO) back reflector (BR) to facilitate light trapping. Transparent-conductive indium-tin oxide (ITO) is deposited on top of the top cell, serving as the top contact as well as an antireflection coating. Thin and highly conductive p and n layers are used in individual cells; they also form tunnel junctions between adjacent cells. Key factors leading to high cell efficiency include using hydrogen dilution during intrinsic (i) layer growth, bandgap profiling in a-SiGe:H i layers, microcrystalline players, highly conductive tunnel junctions, and highly reflective textured back reflectors. Details of these have been reported previously [1]. The best performance achieved in our laboratory using a triple-junction configuration is 14.6% initial and 13% stable AM1.5 efficiencies over a 0.25 cm2 active-area cell.

III. MANUFACTURING

Several manufacturers have been in production of a-Si:H solar panels using either a superstrate or a substrate structure. In the superstrate structure, the substrate on which the various layers of thin films are deposited to form the solar cell serves as a window to the cells. Although flexible substrates can also be used for this structure, the most common superstrate is glass, and several companies in Europe and Japan have been in production of single- and multi-junction solar panels using a monolithic laser-integrated approach [6]. In the substrate configuration [1], the cell is deposited on an opaque substrate and the light enters through the top anti-reflection coating layer as shown in Fig. 1. The advantage of the substrate configuration is the use of a flexible substrate that can be used in a roll-to-roll operation. We have developed a roll-to-roll automated process for manufacturing solar cells on stainless steel. The first prototype roll-to-roll machine was built in 1982-83 timeframe, and since then several production machines have been built incorporating more advanced features. In the current production plant running since 2003, rolls of stainless steel, a mile and a half (2500 m) long, 14 in (36 cm) wide, and 5 mil (125 μm) thick, move in a continuous manner in four machines to complete the solar cell fabrication.

Fig. 2. Schematic of the four production roll-to-roll deposition machines.

The machines are (Fig. 2): 1) The wash machine that washes the web one roll at a time; 2) the back reflector machine that deposits the back reflector by sputtering Al

and ZnO on the three rolls of washed webs at a time; 3) the triple junction amorphous silicon alloy processor that deposit the nine layers of a-Si and a-SiGe alloy layers on six rolls of back reflector coated stainless steels at a time; and 4) the anti-reflection coating machine that sputters indium tin oxide (ITO) on top of the three rolls of the triple-junction solar cell at a time.

The most sophisticated machine in the plant is the triple-junction processor (Fig. 3) where six rolls of stainless steel coated with the back reflector are processed in 62 hours to produce nine miles of solar cells. The processor incorporates many unique features developed over the last two decades:

• Gas gates to isolate the chambers depositing intrinsic layers from those using dopant gases;

• Magnetic rollers to support web transport that prevent front surface scratching;

• Improved cathode design, gas distribution, pumping and heating system that ensure uniform deposition;

• In-situ diagnostic to monitor thickness of layers and component cell performance while the run is progressing.

Fig. 3 Amorphous silicon triple-junction cell processor

The module assembly operation consists of the following steps (Fig. 4). The finished roll of the coated web is first cut into 23.9 cm x 36 cm slabs using a semi-automated press; coupons are also cut during the same operation at preset intervals along the length of the web. These coupons are processed off-line for QA/QC evaluation. The slabs are then processed to define cell size, passivated to remove shunts and shorts, and tested to ascertain quality. Grid wires and contact pads are next applied, and the slabs are cut into predetermined cell sizes for the various product requirements. The cells are next interconnected and the cell-block laminated to provide protection against outside atmosphere. Depending on the application, frames and junction boxes are added, and the finished modules undergo a hi-pot test (by applying 2000 V between the shorted terminals and the frame/ground while the surface of the module is kept wet) and performance measurement under global AM1.5 illumination before they are shipped out.

Amorphous Silicon Alloy Deposition

Machine

Wash Machine

Back Reflector Deposition Machine

Anti-Reflection Coating Deposition

Machine

Fig. 4. Semiautomatic module assembly operation.

The plant has an annual capacity of 25-30 MW and is producing a variety of products for many different applications. The largest focus is on building-integrated photovoltaic. We have developed several solutions so that our flexible and lightweight PV laminates can be directly bonded to various roofing materials ranging from metals to polymeric membranes (Fig. 5). This reduces the installation cost significantly.

Fig. 5. Building-Integrated PV

The plant has been running since 1993; we practice “lean manufacturing” in the plant to improve productivity, quality and safety and to reduce cost. The enthusiastic response of the market to our products has resulted in significant sales growth, and we have embarked on a very aggressive expansion plan as shown in Fig. 6. As of now with the two new plants in Auburn Hills, MI we have an annual production capacity of 58 MW that will increase to 180 MW by the end of 2008. To facilitate fast build up and quick ramp up, we are using essentially the existing design principles for the two new plants of total annual capacity of 120 MW. We are evaluating the introduction of new technology, such as nanocrystalline silicon, for the next generation machines.

Fig. 6. Expansion plan of United Solar Ovonic

IV. NANOCRYSTALLINE SOLAR CELL

The main motivation for using a-Si:H for large-scale manufacturing of solar cells is its low material cost. The inherent disorder and dangling bonds present in this material limits the highest efficiency achievable; there have been efforts to develop other thin film technologies that would still use inexpensive materials, but would attain higher efficiency. nc-Si:H silicon is a natural choice. It can be grown the same way as a-Si:H by glow

discharge decomposition of silane by increasing the hydrogen dilution beyond the edge. nc-Si:H offers enhanced long wavelength response and thus can increase the short circuit current density. Germane is more expensive than silane; substituting a-SiGe:H with nc-Si:H is thus economically attractive. However, due to the nature of the indirect optical transition in nc-Si:H, a relatively thick i layer is required for obtaining high current in the solar cell. How to achieve high efficiency nc-Si:H cells with high deposition rates so that the total deposition time is not too long remains a challenging issue for researchers. Over the last decade, significant advances in our understanding of this material have emerged leading to achieving efficiencies comparable to those of multi-junction a-Si alloy solar cells [7]. Many different deposition methods have been used including radio frequency, very high frequency, and microwave for decomposition of silane. We have been actively involved in this research activity, and have already achieved initial efficiency of 13.5% using a double-junction structure with nc-Si:H as the bottom cell, and 15.3% using an a-Si:H/a-SiGe:H/nc-Si:H structure [8,9]. The highest stable efficiencies using the two structures are 11.8% and 13%, respectively. We have reached a stable efficiency of 13.3% using a-Si:H/nc-Si:H/nc-Si:H structure. We should mention that the triple-junction cell structure with nc-Si as the middle or bottom cell is at a preliminary stage of optimization and there are great opportunities for further improvement. A great deal of efforts has also been made to increase the deposition rate of nc-Si:H. Just cranking up the rf power to increase the deposition rate is not beneficial since it results in high energy ion bombardment and also creation of polymers in the plasma. The most commonly used methods to obtain high quality material at high deposition rates are use of very high frequency (vhf) or deposition at high pressure [10]. Use of very high frequency results in low energy ion bombardment that is believed to improve the structure. At high pressures, there is further reduction of ion bombardment. Using a modified vhf deposition method, we have achieved an initial active-area efficiency of 13.8% using an a-Si:H/a-SiGe:H/nc-Si:H structure where the nc-Si:H layer was deposited at about 0.5 nm/sec. The cell efficiency stabilized at 11.4% after prolonged light soaking. The success in the laboratories has led several manufacturers to introduce multi-junction modules incorporating nc-Si:H in the market. Prototype modules have been supplied by Kaneka for evaluation purposes; Sharp is also building a plant of annual capacity 15 MW. Applied Materials has announced supply of equipment to several manufacturers using a-Si:H and nc-Si:H technology. This is good news for the thin-film silicon community.

V. CONCLUSION

Thin-film silicon made using hydrogen dilution near the edge of amorphous to nanocrystalline transition has emerged as a viable candidate for large-scale manufacture of solar cells. The material below the edge has received more attention, and manufacturing plants are in operation producing a variety of products using a-Si:H

Cutting of Slabsand Coupons

Short Passivation andCell Definition

Bonding ofElectrodes

Cutting ofCells

CellInterconnection

QA/QC QA/QC QA/QC

LaminationFraming andFinishing

Hi-Pot TestTest and Ship

Cumulative Production Capacity (MW)

58

118

180

300

0

50

100

150

200

250

300

350

2005 2006 2007 E 2008 E 2010 E

28

Auburn Hills IAuburn Hills II

Greenville, MI IExpansion (60MW)

Underway

Greenville, MI II AdditionalExpansion (60MW)

Underway

technology. nc-Si:H grown with hydrogen dilution above the edge is emerging as a viable candidate for the bottom cell of the multi-junction structure. Impressive progress has been made in this field as well, and thin-film silicon solar cells are expected to play a major role in the photovoltaic market.

ACKNOWLEDGEMENTS

The authors are grateful to J. Yang, A. Banerjee, K. Lord, B. Yan, G. Yue, K. Hoffman, and J. Call for discussion and collaboration, and K. Welty for preparation of the paper.

REFERENCES

[1] J. Yang, A. Banerjee, and S. Guha, Solar Energy Mat. & Solar Cells 78 (2003) 597.

[2] A.V. Shah, J. Meier, E. Vallat-Sauvain, J. Wyrsch, U. Kroll, C. Droz, and U. Graf, Solar Energy Mat. & Solar Cells 78 (2003) 469.

[3] S. Guha, K. L. Narasimhan, and S. M. Pietruszko, J. Appl. Phys. 52 (1981) 859.

[4] S. Guha, J. Yang, A. Banerjee, B. Yan, and K. Lord, Solar Energy Mat. & Solar Cells 78 (2003) 329.

[5] O. Vetterl, F. Finger, R. Carius, P. Hapke, L. Houben, O. Kluth, A. Lambertz, A. Muck, B. Rech, and H. Wagner, Solar Energy Mat. & Solar Cells 62 (2000) 97.

[6] Y. Tawada, H. Yamagishi, and K. Yamamoto, Solar Energy Mat. & Solar Cells 78 (2003) 647.

[7] K. Saito, M. Sano, H. Otoshi, A. Sakai, S. Okabe, and K. Ogawa, Proc. of 3rd World Conf. on Photovoltaic Energy Conversion (2003) 2793.

[8] B. Yan, G. Yue, J. Yang, S. Guha, D. L. Williamson, D. Han, and C. Jiang, Appl. Phys. Lett. 85 (2004) 1955.

[9] G. Yue, B. Yan and S. Guha, ICANS 2007 (to be published in J. Non-Cryst. Solids)

[10] M. Kondo, Solar Energy Mat. & Solar Cells 78 (2003) 543.