Growth Texture of Polycrystalline Silicon Prepared by Chemical Vapor Deposition

Abstract

Fig. 5. Scanning electron micrograph of precipitation front. Backscattered electron image of alloy is darker than that of Ni matrix Iomellae.logous to an isothermal eutectoid transformation in a homogeneous solid and v is related to the lameUar spacing, 8, bywhere D A is the alloy diffusion coefficient and f is the relative supersaturation (1), which will be here estimated as 0.1. If segregation is achieved by boundary diffusion along the a/d interface, then an approximate growth equation analogous to [3] is obtained by substituting D B b/5 for D A, where D B is the diffusion constant within the boundary of width b. To a first approximation then v = 0.1 DA/8 volume diffusion [4] v = 0.1 DBb/~ ~ boundary diffusion [5] Substitution of the values of v and 5 reported in Fig. 2 into [4] and [5] assuming b = 10A yields D A ~ 10 -9 cm 2 sec -1 or D B ,~ 10 -6 cm 2 sec -1. Extrapolation of high temperature results (17) for volume interdiffusion in the Ni-Mo solid solution yields, at 700~ D A 10 -i s cm ~ sec -1, indicating that segregation must be achieved by the second mechanism of diffusion along the advancing interface. This is almost universally the case with discontinuous precipitation. It is concluded from the simultaneous observations of linear kinetics, uniform lamellar morphology, and a discontinuous change in concentration at the reaction front that the mechanism of internal sulfidation of Ni-20Mo is that of discontinuous precipitation. It is not possible, however, to conclude that the reaction rate is controlled entirely at the precipitation front. Since the kinetics of the over-all reaction are approximately linear, so too are those of external scale growth. Such behavior in an alloy exposed to HeS/H2 atmospheres has frequently been ascribed to rate control by a process occurring at the solid-gas interface. If this is the case in the present instance, then the penetration velocity and interlamellar spacing of the internal precipitation zone may be partially determined by this external constraint.We are undertaking a study of the high temperature deformation of polycrystalline silicon as part of a program to find low cost methods of manufacturing silicon solar cells. The material used as samples in these deformation experiments was commercially available, Permanent address: The Technion, ttaifa, Israel.

Recrystallization of Polycrystalline CVD Grown Silicon

“…The observed spread in reflection angle is m a i n l y due to the radial symmetry of the bars from which the samples are sawn. This is in agreement with earlier results for this kind of material (5). The same kind of analysis on only one of the striated clusters (area 10 mm~) show it to be very strongly ~l l O ~ fiber textured.…”

“…This result was in reasonable agreement with another recently published analysis (2). However, there are three reports in the literature claiming that cobalt microprecipitates are found, either in the as-deposited material (3), or after heat-treatment (4,5). Since this issue is important in defining the relation between the microstructure and possible hardening mechanisms, we have carried out further experiments to study this point.…”

Microstructural Evolution of Mechanically Deformed Polycrystalline Silicon for Kerfless Photovoltaics

“…Kulkarni and Graham et al carried out plastic compression on CVD polysilicon at 1380 C with various strain rates both along and normal to the original <110> fiber orientation. [10,11,31] Their deformed polysilicon completed the recrystallization process after annealing at 1380 C for 10 min when the sample was compressed perpendicular to the original <110> direction by 30% at 1380 C. This finding is, to a certain extent, of equivalent timescale to the secondary recrystallization behavior observed by us. They claimed their recrystallized material is textured with a <110> axis parallel to the compression axis and a <111> axis along the original <110> axis.…”

“…Our characterization results for as-received polycrystalline silicon are broadly consistent with those from other studies. [12,30,31] In previous reports, the <110> preferred growth orientation was not only observed in bulk CVD polycrystalline silicon, [30] but also in polysilicon films produced by the CVD method on different silicon substrates. [32] It was suggested that the high surface energy of the (110) planes causes the crystallites to grow in the <110> direction, that is, nucleation of {110} islands is kinetically favored.…”

“…[32] It was suggested that the high surface energy of the (110) planes causes the crystallites to grow in the <110> direction, that is, nucleation of {110} islands is kinetically favored. [32] However, the surface microstructure of polycrystalline silicon in previous studies [12,[30][31][32] was only examined qualitatively by optical metallography and the existence of growth twins was only inferred from the appearance of striated needle clusters. Our study provides a more direct characterization of the defect morphology and crystallography.…”