Search

We propose a simplified reaction model for SiGe epitaxial growth by chemical vapor deposition (CVD). The SiH4 and GeH4 adsorption at vacant sites, the exchange reaction between SiH and GeH at the surface, and the decomposition of SiH and GeH are considered in the model. The decomposition of SiH4 or GeH4 into SiH2 or GeH2 is representative of complicated reactions in the gas phase. SiGe growth is carried out on a 300 mm Si substrate by reduced-pressure CVD from the SiH4–GeH4–H2 system in the original epitaxial furnace. The dependences of the growth rate and Ge content in the SiGe layer on the temperature and GeH4 flow rate are determined by the characterization of specimens cut from certain positions in 300 mm wafers. By using the chemical reaction analysis software package chemkin, the rate constants in the proposed model are determined by comparison with experimental results. In this report, we validate that the model can explain the behavior of SiGe CVD.

Nanoaggregates of three butadiyne derivatives, i.e., butadiyne derivatives with one tetrathiafulvalene (TTF) moiety at one end (BUT1), two TTF moieties at both ends (BUT2), and one TTF moiety directly bound to the butadiyne group (TBU), and their charge-transfer (CT) complexes with 2,5-difluoro-7,7,8,8-tetracyanoquinodimethane (F2TCNQ) were prepared by the reprecipitation method. The nanoaggregate sizes of BUT1, BUT2, and TBU were about 80–200 nm and those of the CT complexes were reduced to about 100 nm. Upon UV irradiation, nanoaggregates of BUT1 showed regular 1,4-addition polymerization in the solid state while those of BUT2 and TBU polymerized in an irregular manner. However, solid-state polymerization of their CT complexes was not confirmed and decomposition of the complexes was observed. From the X-ray study, the crystallinity of the nanoaggregates was found to degrade in general, compared with that of the corresponding bulk crystals. Interestingly, nanoaggregates of BUT1-F2TCNQ and BUT2-F2TCNQ complexes were found to have conductivities of more than 10−6 S/cm, although their bulk crystals did not show measurable conductivities.

In order to develop the low-temperature silicon epitaxial growth process, the change in the surface morphology of the silicon dioxide film and the silicon surface is studied in a transient state, for the first time, at 1223 K at a pressure of 40–101 kPa in a hydrogen ambient. A very smooth and clear silicon substrate surface is achieved using a uniform silicon dioxide film formed using ozonated ultrapure water and a very low flow rate of hydrogen gas. The surface morphology of the silicon substrate finally becomes smooth and the pit formation is suppressed, although the surface shows a nonuniform island appearance during the removal of the silicon dioxide film. The microroughness of the silicon substrate surface is improved by heating in a hydrogen ambient and by the silicon epitaxial film growth after complete removal of the ozone oxide film.

Oxygen concentration in the top silicon layer of silicon-on-insulator materials formed by low-dose implantation of oxygen has been measured by means of secondary ion mass spectroscopy. A value of less than 1×1017 atoms/cm3 has been obtained. We propose that the oxygen concentration in the top silicon layer is controlled by oxygen out-diffusion during the wafer cooling process.

The enhancement effect of Fe impurities on the generation of surface and bulk microdefects, such as oxidation-induced stacking faults, oxide precipitates, precipitate-dislocation complexes and bulk stacking faults, has been observed in annealed silicon wafers prepared from Czochralski grown crystals intentionally contaminated with iron. The effect remains significant even for an Fe concentration as low as 1012 atoms/cm3. It has been found that Fe facilitates the nucleation of oxide precipitates in silicon. The mechanism of Fe-assisted nucleation of oxide precipitates is discussed. The effect of Fe on the generation of oxidation-induced stacking faults is explained, assuming that both oxide precipitates and Fe:Si precipitates formed near the wafer surface serve as active nucleation sites of these microdefects.

X-ray plane-wave topography in a Laue-transmission case has been widely used to observe microdefects in otherwise perfect silicon crystals. To show the experimental effect of specimen thickness, plane-wave topographic observations were conducted before and after the stepwise chemical etching of the specimen crystal. The observations showed a heightening of the strain sensitivity when reducing the specimen thickness to less than several times the extinction distance of the relevant diffraction. The thickness effect was interpreted through X-ray dynamical diffraction theory for distorted crystals (Takagi-Taupin theory). The computer simulations of the diffraction theory proved that correlations exist amongst the divergence of the incident X-ray beams, the crystal thickness, and the quantity of the minute strain.

In order to reveal more minute microdefects in as-grown CZ silicon crystals, thinned crystals as thin as the order of the extinction distance have been successfully observed by ultra plane-wave topography using synchrotron X-radiation. Various kinds of microdefects reveal themselves only in the thinned region. Using thinned crystals for plane-wave topography is essential for the present experiment. Observed morphologies of microdefects are classified into five categories. One of the observed microdefects has an image extinction rule suggesting a specified strain field around it. The morphology of microdefects has been described in detail.

In order to observe microdefects in silicon crystals, plane-wave topography using synchrotron X-radiation has been successfully applied to thin silicon crystals. We have aimed increasing a ratio of a diffraction signal from microdefects to diffracted intensity from the matrix crystal by taking advantage of an extinction effect of X-ray dynamical intensity. Our plane-wave topography revealed that microdefects for such thin crystals always show a distinct contrast, though we have not obtained zero intensity at the background because of a long-range elastic strain. It is proposed that when silicon wafers are thinned, stress relief at the specimen surface allows a strain field to apread around microdefects to an extent which is detectable using plane-wave topography.