Silicon sheet materials grown by recrystallisation from linear and closed molten zones R. Gamboa, J. C. Henriques, J. M. Serra, J. Maia Alves, A. M. Vallera Universidade de Lisboa, Departamento de Fisica/ Citecmat, P- 1749-016 Lisboa, Portugal Tel: +351 21 7500000, Fax +351 21 7573619, rgamboa@fc.ul.pt G. Hahn, P. Geiger University of Konstanz, Faculty of Physics, P.O.Box X916, 78457 Konstanz, Germany Tel.: +49-7531-88-3644, Fax: +49-7531-88-3895 ABSTRACT: As a step in the development of new silicon sheet growth techniques, we have demonstrated that it is possible to form a stable molten zone in a prismatic silicon tube, and use such a closed molten zone (CMZ) in a new recrystallisation process to obtain crystalline silicon sheet material. The main advantage of such a process, as compared to linear molten zones with open side meniscus, is that liquid edge instabilities are avoided. Sheet silicon materials obtained in different conditions, grown from linear or closed molten zones, were characterised by lifetime measurements and by spectral response of test solar cells. Once the main sources of contamination were removed, CMZ materials demonstrated average lifetimes of ~2μs and diffusion lengths of ~100μm in test cells with no BSF nor passivation procedures. Keywords: Silicon - 1: Ribbons - 2: Recrystallisation – 3 1. INTRODUCTION Silicon sheet growth techniques may prove vital in the mid and long term development of photovoltaic solar energy applications where low cost and high quality materials are needed. This was realised very early in the modern history of photovoltaics, and much effort has been devoted to the development of an extremely varied array of techniques. That the problem is far from trivial is demonstrated, for instance, by the fact that it is still far from decided today, after several decades of research and development, which techniques will survive the test of industrial production in a competing market: R&D is proceeding, for instance, on the two main classes of processes of growth from the liquid, parallel and perpendicular [1]. Parallel processes, where crystal growth proceeds (almost) parallel to the sheet normal, should permit high area generation rates, but generally require a solid substrate. This substrate can be detached from the silicon sheet and re-used, as in the RGS process, being investigated by Bayer [2], or become part of a final composite sheet, as in the SILICON FILM under development in Astropower [3]. Perpendicular processes, where crystal growth proceeds along a direction in the sheet plane, are generally slower area generators than parallel ones. Interestingly, this class includes the only process that reached well established industrial production, the EFG process [4]. In this process, the problem of meniscus stability is solved with the help of a shaper dye, made of graphite, partly immersed in the liquid silicon bath; and the problem of the free meniscus edges is solved by avoiding them altogether: the dye has a closed polygonal shape, and the final product is a prismatic silicon tube. Among other perpendicular processes, one could mention the String Ribbon [5], where the problem of stability of the meniscus edges is solved by a pair of “strings”, which effectively stabilise the lateral sheet shape; and the Dendritic Web process [6], where lateral collapse of the liquid meniscus is avoided by an acicular silicon dendrite on either side. A different approach of perpendicular growth from the melt is exemplified by the SSP process [7], where, rather than a liquid silicon bath, only a molten zone is kept liquid. 2. MOLTEN ZONE PROCESSES Sheet silicon growth from a floating molten zone has interesting advantages when compared to conventional multicrystalline silicon ingot production or other sheet growth techniques: there is no need for a crucible, contamination by foreign materials can be avoided, and, in principle, it can be a low energy process, since only a small volume of silicon has to be kept liquid. The main technical difficulties concern molten zone stability and the need for a pre-ribbon. Previous work on such molten zone techniques in our group resulted in the development of the STRETCH process, in which a multicrystalline silicon sheet is produced, with area multiplication and thickness reduction, by zone melting recrystallisation of a flat pre-ribbon [8]. Good surface finish, square centimetre grain size and reduction of metallic impurity content were achieved. However, it was shown that the floating zone requires an oxidising atmosphere, necessary for stabilising the free edges by oxide encapsulation of the whole molten zone. This is the main reason why it was not possible to demonstrate a consistent acceptable quality in the samples produced in the presence of oxygen: although the diffusion length of the record sample was quite high (~100 μm), it generally fell below 40 μm. This was not unexpected, since the quality of oxygen containing silicon material is well known to be strongly dependent on thermal history (which includes the thermal profile the sheet is subjected to during the growth process itself). The high oxygen content therefore strongly limits the interest of this growth process.