Kikuji Sato

Stress and Thermal‐Expansion Coefficient of Chemical‐Vapor‐Deposited Glass Films

The stress in thin films of chemical‐vapor‐deposited (CVD) glass on Si substrates was measured by the Newton‐ring method. The CVD films studied were Silane‐oxidized SiO2, phosphosilicate glass and borosilicate glass deposited at 400°–450°C. Stress reduction due to moisture absorption was observed in the CVD films, but it was not observed in the sputtered SiO2 and the thermally grown SiO2 films. The initial tensile stress in the CVD films changed into compressive stress after heat treatments at 600°–900°C. By measuring the stress at elevated temperatures, the intrinsic stress was reduced from the total stress. From the thermal stress measurement, the thermal‐expansion coefficient and the elastic constant of the CVD glass films were estimated.

Hillock-free aluminum thin films for electronic devices

The addition of an alloying element was proposed in order to suppress hillocks which grow on the surface of deposited aluminum conductors after they have been subjected to thermal cycling (200°C to room temperature) or high temperature heat treatment (400°C). The alloying element, tin, which has a small diffusion coefficient, a large binding energy with a lattice vacancy and a large atomic diameter, and manganese, which has a relatively small solid solubility into aluminum, were evaluated since they also have proper values of vapor pressure for ease of evaporation with aluminum. Alloy composition was determined to be just above the solid solubility of the element in aluminum at the deposition temperature of 350°C. It was proved experimentally that Al-0.06 wt pct Sn alloy and Al-0.1 wt pct Mn alloy, which had been selected for the abovementioned reason, had a marked effect in suppressing the growth of hillocks.

Transverse 1/f noise in InSb thin films and the signal‐to‐noise ratio of related Hall elements

Transverse 1/f noise in approximately 2‐μm‐thick InSb thin films is investigated experimentally at room temperature. Linear dependence of noise voltage on dc bias current is shown quantitatively. The noise intensity is inversely proportional to the number of conduction electrons in the bulk. The temperature rise of specimens due to Joule heating does not affect the noise intensity coefficient. The coefficient differs from sample to sample, which is reduced by the heat treatment of specimens, but is independent of the doped impurity concentration. As a result, the signal‐to‐noise ratio of related Hall elements is formulated for the first time for audio magnetic heads applications. The signal‐to‐noise ratio is nearly 80 dB for a 10‐G magnetic field in the audio frequency range.

Evaporated InSb thin films and their application to bubble domain devices

Indium antimonide thin films were deposited by the three-temperature method onto various kinds of amorphous substrates such as Corning 7059 glass using source materials of 99.9999 % purity. The optimum ratio of vapor pressures of indium and antimony, and the optimum temperature of the substrate, Tsub, were investigated for each substrate in order that the deposited film had the highest carrier mobility, μH, and Hall coefficient, RH, as well as being perfectly stoichiometric when tested by the x-ray diffraction method. The microstructure of the films was also investigated as a function of Tsub. The thickness of the film varied between 1.0 – 3.2 µm. Among the films thus obtained the best one gave the characteristics of μH = 23,000 cm2/V·sec and RH = 480 cm3/C at room temperature. The latter value is entirely the same as that of the purest bulk material.This film was photo—etched to a cross—shaped Hall detector; the width of both the current lead and the output lead was 4µm. The device was applied to the detection of magnetic bubble domains of 10 µm diameter. The output voltage of 1.4 mV is larger than those ever reported for a bubble of the same order of diameter.

Reply to ’’Comment on ’Transverse 1/f noise in InSb thin films and the signal‐to‐noise ratio of related Hall elements’ ’’

Noise formula describing the 1/f noise in InSb films is pointed out. The noise‐intensity coefficient K is introduced as a variable depending on film‐preparation methods. Experimentally, K2 changes by two orders of magnitude. However, existing noise calculations show that K2 should remain constant except for a dimension factor. Therefore, the experimental results cannot be explained by the existing theory.