Kimio Hashimoto

Diffusion Coefficient of Interstitial Iron in Silicon

The in-depth profiles of iron in silicon diffused with iron at 800, 900, 1000 and 1070°C were investigated by DLTS measurements. The diffusion coefficient of interstitial iron in silicon was represented by the expression DFe=9.5×10-4 exp (-0.65/kT) cm2s-1 in the temperature range of 800–1070°C.

Diffusion Coefficient of Iron in Silicon at Room Temperature

The pairing reaction of interstitial iron and substitutional boron for samples diffused with iron into boron-doped p-type silicon was studied by measuring the concentration of interstitial iron by DLTS as a function of the storage time at the temperatures of 0, 27, 42, 57 and 72°C. The diffusion coefficient of interstitial iron in silicon was determined at the temperature range between 0 and 72°C. The diffusion coefficient was represented by the expression DFe=3.3×10-1 exp(-0.81/kT)cm2s-1.

Diffusion Mechanism of Nickel and Point Defects in Silicon

The in-diffusion of nickel into silicon and the annealing of nickel in silicon have been studied experimentally at 900°C. The variations in concentration of substitutional nickel with time are described well by an exponential function predicted from a dissociative mechanism, in both the in-diffusion and annealing processes. The time constant in the annealing process is substantially independent of the initial concentration of substitutional nickel. These results support the theory that nickel in silicon diffuses by a dissociative mechanism and that the dominant point defects in silicon are vacancies.

Deep impurity levels and diffusion coefficient of manganese in silicon

Manganese‐related deep levels in n‐ and p‐type silicon have been investigated by deep level transient spectroscopy and Hall effect. Two electron traps of Ec−(0.12±0.01) eV and Ec−(0.41±0.01) eV, and a hole trap of Ev+(0.32±0.01) eV are found in manganese‐doped silicon. The energy levels of these traps correspond to the transitions between four charge states (Mn−, Mn0, Mn+, Mn++ ) of interstitial manganese. An additional donor‐type electron trap of Ec−(0.51±0.02) eV is observed in the n‐type samples, and the trap can be tentatively assigned to substitutional manganese. Furthermore, an electron trap of Ec−(0.50±0.02) eV is observed for n+p junction samples diffused with manganese in boron‐doped p‐type silicon. The trap is attributed to the manganese‐boron complex, which is formed owing to the pairing reaction of interstitial manganese and substitutional boron. From the investigation of the pairing reaction, the diffusion coefficient DMn of interstitial manganese is determined in the temperature range 14–90 ...

Note on the Analysis of DLTS and C2-DLTS

The difference in the peak temperatures of DLTS and C2-DLTS is discussed with reference to the measurement of deep levels in semiconductors. The emission rate of a trap level can be accurately determined when the peak temperature observed in DLTS agrees with that in C2-DLTS. When the peak temperatures disagree, the trap depth and trap density cannot be determined correctly, and the junction profile and trap density have a great effect on the peak temperatures of DLTS and C2-DLTS. These phenomena appear in DLTS and C2-DLTS measurements on Au-doped Si p+n diodes. It is necessary to use C2-DLTS together with DLTS in the study of deep impurities in semiconductors.

Method of Analysis of a Single-Peak DLTS spectrum with Two Overlapping Deep-Trap Responses

In the usual DLTS technique, the trap activation energy is obtained from Arrhenius plots of the inverse of time constants set up experimentally vs the inverse peak temperatures of the DLTS signals. Considerable error, however, may appear in the activation energy when DLTS singals with a hump or shoulder are observed. In a special case, a single-peak DLTS is obtained when the signals due to two or more individual traps overlap. However, it is possible in this case to separate each deep-trap response from the overlapping DLTS signal by observing the injection pulse width dependence of the transient capacitance, and to determine the trap parameters such as the activation energy correctly. This paper presents a method for determining the trap parameters from the injection pulse width dependence of the DLTS spectrum.

In-Diffusion and Annealing of Copper in Germanium

The in-diffusion and anealing of copper in germanium are examined experimentally over the temperature range 600 to 805°C, and using the experimental results, discussions are made as to whether the dissociative mechanism or the kick-out mechanism is at work in these processes. The experimental results support the theory that copper in germanium diffuses by the dissociative mechanism and the dominant point defects in germanium are vacancies.

Deep impurity levels of cobalt in silicon

It has been believed until now that cobalt in silicon forms an acceptor level at around Ec–0.53 eV and a donor level at around Ev+0.35 eV. However, it is found that an acceptor level at Ec–(0.40±0.02)eV and a donor level at Ev+(0.23±0.01) eV are attributed to the amphoteric cobalt levels from the DLTS measurements for the samples diffused with cobalt deposited from the evaporation of a pure cobalt wire. The former levels (0.53 eV and 0.35 eV) are observed only for the samples prepared by the cobalt deposition from a tungsten filament wrapped with cobalt wire, and the tungsten contamination is regarded to be the cause of these levels.

Point defects in silicon studied by nickel diffusion

Abstract The in-diffusion and the annealing processes of nickel in silicon are studied experimentally at 900°C to distinguish between vacancies and self-interstitials as the dominant point defects in silicon. The experimental results show that, in both the in-diffusion and the annealing processes, the concentration of substitutional nickel atoms varies with time t, following the linear dependence on t during a short time after the onset of both processes, and that the time constant of the annealing process is indpendent of the initial concentration of nickel. These results support the theory that nickel in silicon diffuses by the dissociative mechanism and the dominant point defects in silicon are vacancies.

Method for Estimating Accurate Deep-Trap Densities from DLTS of Junctions Containing Several Kinds of Deep-Traps

The density of a deep trap is estimated from the peak height of the DLTS signal and the steady-state capacitance on a reverse-biased semiconductor junction. When several kinds of traps exist in the depletion region and the trap densities are not much smaller than the net shallow-level density, a considerable error appears in the trap-density estimations. For such a case, it is necessary to solve exactly Poissons equation for the depletion region in transient sequence to deduce the trap densities. A method for accurate trap-density evaluation is introduced, and the results of the analysis for the DLTS spectrum of a gold-doped n-type silicon Schottky diode are discussed.