Float-zone silicon

Abstract: A baffle plate of a showerhead gas distribution system and method of using the baffle plate wherein the baffle plate is effective for reducing particle and/or metal contamination during processing of semiconductor substrates such as silicon wafers. The showerhead can be a showerhead electrode of a plasma processing chamber such as an etch reactor. The baffle plate comprises silicon on at least one surface thereof and is adapted to fit in a baffle chamber of the gas distribution system such that the silicon containing surface is adjacent to and faces the showerhead. The silicon containing baffle plate can consist entirely of silicon or silicon carbide of at least 99.999% purity. The silicon can be single crystal silicon or polycrystalline and the silicon carbide can be CVD silicon carbide, sintered silicon carbide, non-sintered silicon carbide or combination thereof. The non-sintered silicon carbide can be silicon carbide formed by reaction synthesis of silicon vapor with a carbon material such as graphite. Openings in the silicon containing baffle plate can be offset from openings in the showerhead to avoid a line-of-sight between plasma in the chamber and the openings in the silicon containing baffle plate.

Abstract: Outdiffusion profiles of nitrogen in silicon were measured by secondary ion mass spectrometry to determine its diffusion coefficient in a temperature range of 800–1200 °C. The total amount of the nitrogen outdiffusion agrees with the change in infrared absorption by heat treatment. The experimental results give the diffusion coefficient of nitrogen as D=2.7×103 exp(−2.8eV/kT)cm2/s. This value is five orders of magnitude larger compared with the reported expression of 0.87 exp(−3.29eV/kT)cm2/s. Nitrogen–nitrogen‐pair‐like molecule in crystals corresponds to the former value and substitutional nitrogen atom to the latter one. These two types of nitrogen in silicon may allow us to clarify the various effects of nitrogen such as strengthening of crystals and suppression of swirls and D‐defect generation.

Abstract: For optimum performance of solar cells featuring a locally contacted rear surface, the metallization fraction as well as the size and distribution of the local contacts are crucial, since Ohmic and recombination losses have to be balanced. In this work we present a set of equations which enable to calculate this trade off without the need of numerical simulations. Our model combines established analytical and empirical equations to predict the energy conversion efficiency of a locally contacted device. For experimental verification, we fabricate devices from float zone silicon wafers of different resistivity using the laser fired contact technology for forming the local rear contacts. The detailed characterization of test structures enables the determination of important physical parameters, such as the surface recombination velocity at the contacted area and the spreading resistance of the contacts. Our analytical model reproduces the experimental results very well and correctly predicts the optimum contact spacing without the use of free fitting parameters. We use our model to estimate the optimum bulk resistivity for locally contacted devices fabricated from conventional Czochralski-grown silicon material. These calculations use literature values for the stable minority carrier lifetime to account for the bulk recombination caused by the formation of boron-oxygen complexes under carrier injection.

Abstract: Accurate measurements of the minority carrier‐ and lattice scattering‐diffusion constant and mobility in float zone silicon have been determined using photoconductance decay. For the more lightly doped specimens our results indicate slightly higher mobility than published majority carrier values. This is attributed to purer samples which allow a more accurate measurement of the lattice scattering mobility. In the dopant range 1015–1017 cm−3 the results for both electrons and holes are, within experimental error, equal to the majority carrier values. Unlike other methods this technique is a very direct measurement of the diffusion constant as only the thickness and decay time of the wafer need to be determined. The method is estimated to have a one standard deviation uncertainty of 2%–4% which is comparable to the best accuracy previously obtained for majority carrier measurements.