Nobukatsu Fujino

Theoretical Calculation of Fluorescent X-Ray Intensities in Fluorescent X-Ray Spectrochemical Analysis.

The theoretical formulas of the fluorescent x-ray intensity (the primary, the secondary and the tertiary fluorescent x-rays) are obtained and actual calculations are made for nickel-iron, iron-chromium, and nickel-iron-chromium alloys. In the calculation, the continuous intensity distribution of the irradiating x-rays for the wavelength is taken into account. The theoretical results agree with the experiments very well, the deviations being 1% at most, and it is concluded that the theoretical estimation of the fluorescent x-ray intensity emitted from samples of various composition is possible. The matrix effect, absorption effect, and enhancement effect are evaluated from the present theory. In the enhancement effect, secondary fluorescent x-rays are quite significant (for instance, the chromium secondary fluorescent x-rays from Cr 2% and Fe 98% alloy reach the 55% of the chromium primary fluorescent x-rays and they can not be neglected), but the tertiary fluorescent x-rays are only a few per cent. The usual approximation methods, where the irradiating x-rays are assumed as monochromatic rays and the secondary fluorescent rays are not taken into consideration are also discussed and their meaning is made clear.

A Method of Quantitative Contamination with Metallic Impurities of the Surface of a Silicon Wafer

A method of quantitative and uniform contamination of the surface of a silicon wafer with a metallic impurity in low-level quantities using spin coating with a contaminated aqueous solution was developed. Fe, Ni, Cu and Al were examined and the quantities of these metallic impurities on the surface were controlled by the concentration of metal ions in the contaminated solution. Using this method, the influence on the recombination carrier lifetime was quantitatively investigated. A significant degradation of the recombination lifetime was observed for more than 1011 atoms/cm2 of impurities Cu, Ni and Fe on the surface.

Behavior of Defects Induced by Metallic Impurities on Si(100) Surfaces

The formation of defects induced by Ni, Cu and Fe and their retardation by intrinsic gettering (IG) were studied under several heat treatments. The behavior of Fe in Si was different from that of Ni and Cu. After annealing at 1150°C and subsequent cooling, shallow pits (SP) due to Ni and Cu precipitates were observed only at the surface. Fe precipitation was obtained by holding at temperatures below 850°C, where Fe was supersaturated in the matrix. We believe that metal precipitation is dominated by the diffusion rate. During additional oxidation, oxidation-induced stacking faults (OSF) were transformed from each precipitate. It seems that metal precipitates act as nuclei of OSF. Effects of IG for the three elements were compared. The retardation of defects induced by Fe was less than that of Ni and Cu. This difference is closely related to the growth rate of Fe precipitates.

Degradation of Gate Oxide Integrity by Metal Impurities

The degradation of gate oxide integrity (GOI) by metal impurities on Si wafers was studied. Ni and Cu tended to precipitate at the Si surface after high-temperature annealing. When these precipitates existed before gate oxidation, they penetrated into the gate oxide film and degraded GOI. Fe tended to remain in the oxide film after oxidation and degraded GOI. The degradation was observed for annealed samples when the surface metal concentration exceeded 1.0×1012 atoms/cm2 of Ni or 5.0×1012 atoms/cm2 of Cu. It was also observed with contamination of 1.0×1013 atoms/cm2 of Fe without annealing.

Influence of Metal Impurities on Leakage Current of Si N + P Diode

Dependence of the leakage current of Si N+P diodes on the surface metal (Cu, Ni or Fe) concentration after quantitative contamination was investigated, and the causes of the leakage current were studied by SIMS, TEM and optical microscopy. Cu was gettered in the N+ area, forming many large dislocations in the N+ area and inducing some dislocations in the substrate near the junction; thus, the leakage current increased remarkably. Ni was gettered in the N+ area also, but did not form large dislocations, so the leakage current did not increase. Fe was not gettered easily, and hence the leakage current increased corresponding to the Fe concentration.

Transmission Electron Microscopy Observation of Defects Induced by Fe Contamination on Si(100) Surface

The behavior of Si(100) surface defects induced by Fe contamination was studied with transmission electron microscopy. After annealing at 1150°C for 1 hour and a subsequent heat treatment at 850°C for 2 hours, Fe-containing precipitates were observed in Si substrate in close vicinity to the interface of Si and SiO2 formed during the annealing. One of these precipitates is identified as FeSi-type silicide. In addition, the inclusions which were confirmed to be Fe3O4 or γ-Fe2SiO4 were observed in the surface thermal SiO2 layer. These results demonstrate that Fe atoms diffuse into the Si substrate during annealing at 1150°C and precipitate at a Si/SiO2 interface, while Fe atoms left on the surface form inclusions in the surface SiO2 layer. Under an additional thermal oxidation at 1000°C, oxidation-induced stacking faults were formed. They were not decorated at all in contrast with those induced by Cu or Ni contamination.

TEM Observation of Defects Induced by Cu Contamination on Si(100) Surface

The behavior of a Si(100) surface defect induced by intentional Cu contamination was studied with transmission electron microscopy. After annealing at 1150°C for 1 hour in N2 atmosphere, colony precipitates lying along (110) planes were observed only on the surface of the wafer. These precipitates were estimated to be Cu6Si-type silicides by the selected area diffraction pattern. During additional annealing in an oxidation atmosphere, stacking faults were formed from each of colony precipitates. This indicated that colony precipitates were the nucleus of oxidation-induced stacking faults.

Dependence of Gettering Efficiency on Metal Impurities

The dependence of gettering efficiency on metal impurities was investigated by means of an intentional contaminating method using Ni or Fe and the MOS C-t method with an eye to using gettering techniques appropriately. The results showed that Ni is easily gettered but Fe is not. Therefore, as a second step, the condition improving gettering efficiency for Fe was investigated. It was consequently found that adding heat treatment at low temperature to the final heat cycle is effective for gettering of Fe.

Comparison of Gettering Techniques by Means of Intentional Quantitative Cu Contamination

Gettering is a very important technique for the LSI process. The effects of gettering techniques such as intrinsic gettering (IG), backside damage (BD), and polysilicon gettering (PG) were investigated by the MOS c-t method after intentional quantitative Cu contamination. These three techniques showed different gettering behaviors in the heat treatments simulating a device heat process. IG and BD were not effective in the beginning of the heat process, but they began to show stronger gettering effects with heat treatments and gained sufficient abilities. The IG kept its effect until the end of the heat process, but BDs effect decreased with subsequent heat treatments. PG had a sufficient gettering effect from the beginning of the heat process, but PG also showed a decrease of gettering effect at the end of the heat process.

A Model of Thermal Transfer in Czochralski Silicon Molten

Simultaneous multipoint measurement has been performed to determine the temperature of silicon molten of a Czochralski (CZ) system during the growth of a monocrystalline ingot. A high magnitude of temperature fluctuation was found to be totally random without correlation among temperatures of various points in the molten, a phenomenon which could not be explained by models based on a lamina flow. The magnituge of the fluctuation, however, was observed to be related with effective thermal transfer of the molten. A model which took account of thermal transfer induced by local dynamical random motion, a kind of turbulence, of the molten was proposed. The time-averaged temperature distribution taken from the solution of a static diffusion equation with the effective thermal diffusivity agreed very well with the experimental results.