Crystallographic Evaluation of Ultrathin SOI Substrates and Impurity Diffusion Modeling for Nanometer LSIs

点击翻译按钮获取中文摘要

1.Crystalline Quality Evaluation of SOI Substrates(1)Changes of Carrier Density Distribution in SOI Substrates with AnnealingCarrier distribution changes in SOI substrates with annealing were measured. The changes depend on both Si layer thickness of SOI substrates and annealing temperature. For Si layer thicker than 1 μm, the inversion layer is generated with annealing near 1000℃ at the buried oxide interface of p-type SOI substrates, while the accumulation layer is generated at the interface of n-type SOI substrates.. For Si layer thinner than 1 μm, n-type donor is generated with annealing of 700-800℃. The donor level is below 0.03eV and the origin of the donor is considered to be due to the oxygen precipitation, i.e., so-called new donor.(2)Electrical Characteristics Evaluation of Buried Oxide (BOX) Interfacethe carrier recombination velocity of BOX interface was measured using μ-PCD method. The velocity is 500-1800 cm/s which is about 100 times larger than that for the conventional … More interface of Si and thermal silicon oxide.2.Impurity Diffusion Modeling for SOI Substrates(1)Comparison of B Diffusion Profiles in 3 kinds of Ultrathin SOI SubstratesB diffusion profiles in 3 kinds of ultra-thin SOI substrates (SIMOX, UNIBOND and ELTRAN) were compared. The diffusion depths in the SOI substrates were shallower than those in bulk Si and the depth in SIMOX is the shallowest among 3kinds of SOI substrates. These results show the crystalline quality of BOX interface and the recombination velocity of point defects at the interface is the fastest for SIMOX substrates.(2)Accuracy of Process Simulator on the Market and Diffusion ModelingFrom the B and P predeposition profiles in bulk Si and SOI substrates, diffusion parameters for simulation were extracted. Using the parameters, the drive-in diffusion profiles were simulated using the process simulator on the market and compared with experimental profiles. The experimental profiles are deeper than those simulated, which suggests that the diffusion enhancement mechanism in drive-in process should be taken in the process simulator. Less