Recent development in monocrystalline Si (c-Si) solar cell technology is significant both for reducing the cell cost andfor increasing the photoelectric conversion efficiency. The recent improved efficiency and cost-cutting of c-Si solar cell is mainly due to technology that (a) improves the carrier lifetime via surface passivation, (b) uses higher quality crystalline Si, and (c) reduces wafer thickness. A challenge in achieving further cost reduction, however, is reducing the thickness of c-Si while maintaining its crystallinity. Fabrication of c-Si is commonly attained by wire-saw slicing of a crystalline ingot or by a layer transfer process (peeled off). The wire-saw slicing can keep high quality of c-Si but, cause kerf loss and the limitation of reducing thickness. In contrast, technical challenges for layer transfer processes are 1) the formation of a high-quality thin film Si, 2) achieving a structure that can easily be peeled off from Si substrate, 3) improving the growth rate and Si raw material yield (necessary equipment costs are determined by the growth rate), and 4) being able to use the substrate after layer transfer without any waste. In the paper, I will introduce the zone melting / heating (re-) crystallization (ZMC/ZHR), shown in Fig.1, to fabricate monocrystalline Si thin film as a key technology in two different kinds of layer transfer process for cost-cutting solar cells with high efficiency. In both layer transfer processes, the control of nano-surface roughness was crucial to improve the crystal quality of Si thin film. Fabrication of quasi-monocrystalline Si thin film without seed layer by high-speed ZMC[1~3] Silicon-on-insulator (SOI) films for solar cells were fabricated by high-speed ZMC. For ZMC, amorphous Si films were deposited on SiO2layers by rf sputtering. After SOI (Si/SiO2) was covered with another SiO2layer (forming a sandwich structure of SiO2/Si/SiO2), the SOI films were crystallized by ZMC. In ZMC, monocrystalline Si thin film was prepared by melting and crystallizing in a line (using a lamp line heater) and then scanning in one direction. Due to the stability at the SiO2/melted Si interface, the result is quasi-monocrystalline Si without requiring any seed for crystal growth. The surface roughness (several ten nanometer) of deposited Si film determined the crystal defect density of the ZMC film. Monocrystalline Si thin film with 10 times faster growth rate on nano smoothing porous Si by ZHR for cost-cutting solar cells with high efficiency[4~6] A high-quality thin film monocrystallinesilicon was developed for solar cells. The crystal defect density of the film could be reduced to the silicon wafer level at a growth rate that is more than 10 times higher than typical CVD. First, double-layer nano-order porous Si was formed from Siwafer using an electrochemical oxidation. Subsequently, the surface was smoothed to a roughness of 0.2 to 0.3 nm by ZHR. This substrate was used for high-speed growth of monocrystalline Si by Rapid vapor deposition (RVD), developed by S. Noda et al.[7]. The grown film can easily be peeled off using the double-layer porous Si layer, and the substrate can be reused or used as an evaporation source for RVD, which significantly reduces material loss. When the surface roughness of the porous Si substrate is reduced by changing the ZHR conditions, the defect density of the Si thin film, measured by ESR, could be reduced to the Si wafer level of about 1/10th. The results showed that a surface roughness in the range of only 0.1-0.2 nm (level of atoms to several tens of layers) has an important impact on the formation of crystal defects, which is also of interest as a crystal growth mechanism. (Figure 1 showed the outline of these results.) Acknowledgement I summarized the results of reference papers [1~6] focusing on zone melting / heating (re-) crystallization and surface roughness control as an invited talk in the 234thMeeting of The Electrochemical society (ECS). I appreciate for Prof. S. Noda, Dr. K. Hasegawa, Dr. A. Lukianov, Ms. C. Takazawa and all authors of these papers. Figure 1
Read full abstract