DISTRIBUTION OF GRAIN BOUNDARY TYPES IN MULTICRYSTALLINE SILICON Stokkan, Gaute 1 , Stoss, Aleksander 2 , Kivambe, Maulid 2 , Ervik, Torunn 2 , Ryningen, Birgit 1 and Lohne, Otto 2 1) SINTEF Materials and Chemistry, Department of Solar Cell Silicon, P.O.Box 4760 Sluppen, NO-7465 Trondheim, NORWAY. Tel: (+47) 41 55 98 51, Fax: (+47) 73 59 27 86, Email: gaute.stokkan@material.ntnu.no, birgit.ryningen@sintef.no 2) NTNU, Department of Materials Science and Engineering, Alfred Getz vei 2, 7491 Trondheim, NORWAY. Tel: (+47) 73 55 12 00, Fax: (+47) 73 55 02 03, Email: stoss@stud.ntnu.no, kivambe@MIT.EDU, torunn.ervik@material.ntnu.no, otto.lohne@material.ntnu.no ABSTRACT: Grain boundary properties and dislocation density were investigated for an ingot produced in a pilot scale directional solidification furnace. The material is characteristic of traditional multicrystalline silicon, different from the current trend of producing small grained material, which has significantly lower dislocation density. Large, twinned grains were present from the bottom part of the ingot, and grain size increased as growth proceeded. While the total density of grain boundaries decreased, the fraction of all types of grain boundaries remained stable. The ingot developed extensive dislocation networks towards the top, covering large areas in the range of several cm 2 . Together with these regions also large regions with very low dislocation density exists. The characteristics of grain boundary types and their evolution during growth in this, former generation of multicrystalline silicon is a useful tool for evaluating product and process parameters of high performance multicrystalline silicon, although no direct comparison to such material has been performed in this work. Keywords: Grain Boundaries, dislocations, multicrystalline silicon 1 INTRODUCTION Grain boundaries and dislocations are the most important defects in multicrystalline silicon because of their density and their impact on the solar cell performance. Whereas grain boundaries are often found to be of less significance, dislocation clusters and sub grain boundaries account for important efficiency reductions [1-3]. Furthermore grain boundaries are found to be important sources of dislocations [4-6], and by employing computer simulation based on constitutive models, such as the Alexander-Haasen model [7], it can be shown that dislocation multiplication increases in the vicinity of grain boundaries due to higher stresses [8]. Recent development in production techniqes for multicrystalline silicon has resulted in a preference to production of multicrystalline silicon with significantly smaller grain size than has been traditionally preferred [9], since this method enables the control of dislocation density to a higher extent. While dislocations are generated locally, the clusters are not allowed to proliferate uncontrolled as growth proceeds, as is often seen for traditional multicrystalline material [5], and it is thus important also to focus on the ability of grain boundaries to release thermal stress and to function as "sinks" for dislocation clusters, not only sources. It should be noted that High performance multicrystalline silicon is a generic term which refers to material containing the properties described above, and it does not specify a particular nucleation or growth method. Different producers may have different approaches to achieving these properties. In this work grain boundary properties of an ingot produced in a pilot scale furnace (12 kg charge, boron doped to 1.2 Ωcm resistivity, cylindrical, electronic grade crucible) is systematically monitored by the use of Electron Backscatter Diffraction (EBSD), as function of height. The aim of the grain boundary investigations was to provide more detailed information about distribution of grain boundary properties (such as the presence of Coincidence Site Lattice (CSL) than has previously been reported [10], e.g. as function of height, which may be used as a comparison to properties of high performance multicrystalline silicon. The investigation of dislocation density is added onto the grain boundary analysis, to show if any correlation between the grain boundary characteristics and dislocation density could be inferred. 2 EXPERIMENTAL 2.1 Sample preparation Wafers representing the entire height of the ingot was selected. The wafer numbers are given in Table 1. Table 1 Sample selection Wafer number Number of samples Top 200 4 Middle high 150 4 Middle low 121 3 Bottom 10 4 From each wafer, four 50x50 mm 2 samples were laser cut. These wafers were subsequently grinded and polished, including chemical mechanical polishing, in order to gain a high signal to noise ratio during EBSD measurements. After the EBSD measurements the samples were etched for 25 seconds in Sopori etchant [11] to reveal dislocations and grain boundaries. 2.2 Measurements The EBSD measurements were performed using a JEOL JSM 840A Scanning Electron Microscope equipped with TSL OIM software for data acquisition and analysis and a NORDIF Phosphor detector. In order to obtain large scale measurements for statistical analysis, the Comboscan mode was used, in which several areas are imaged by scanning the electron beam, each scan separated by a transfer of the sample stage. The acquisition software stitches these images together to form a large area measurement. Very low magnification of 30x is used, and the step size was 60 µm. An area of 25x50 mm was scanned for each sample, and the measurements formed adjacent regions to each 28th European Photovoltaic Solar Energy Conference and Exhibition 1418