High electric field effects in n-indium antimonide

Abstract

IN a recent report(l) the authors described experimental evidence for the creation of electron-hole pairs in n-InSb by means of hot electrons. These observations included current-voltage and Hall effect data taken in a 7000 oersteds magnetic field. The previous experiments have been extended to study the magnetic field dependence of the phenomena. Studies of the effect of longitudinal magnetic fields have also been undertaken, indicating a markedly different behavior from that with the magnetic field transverse. The pulse technique for obtaining these data has already been described.(l) All the measurements were made at 77°K. The single crystal n-InSb specimen was the one discussed in the earlier report.(r) Fig. 1 shows the current density-electric field characteristics for no magnetic field and for a transwrse magnetic field, H, of 3500 oersteds. The Hall coefficient, En, is also shown as a function of the electric field, E, in the presence of the 3500 oersteds. For E x 150-200 V/cm RH is seen to decrease rapidly for small further increases in E. This decrease is interpreted as evidence for the creation of electron-hole pairs by impact ionization across the forbidden gap. Other mechanisms that might have explained these results have been considered(l) and abandoned on the basis of additional critical experiments and the known properties of n-InSb. In particular, the possibility of minority carrier injection can be ruled out because the transit time (Z 10-s set) was 102 times longer than the lifetime (_ 10-s set) of such carriers.(s) For H = 0, the current density increases as Ed.0 in the region above 240 V/cm. Although this is a rapid increase in current it might have been expected to be even steeper.t3) It is believed that the shape of the current densityelectric field curve is determined principally by the electron-hole recombination processes. In particular, surface recombination may be very important. Several other second order effects can also contribute to producing the observed dependence. Both the increase@) in the effective mass of the electron and the increase(s) in Eg, the band gap, as the conduction band is filled, can dictate the need for an increased electric field to get higher current densities. Several attempts have been made to detect the band-gap recombination radiation which should be present once the threshold for creation is exceeded. A sensitive bridge circuit was used to detect possible photoconductive changes in either a gold-doped germanium or an InSb detector placed very close to the crystal. To date, these attempts have not yielded any observable signals that could be attributed to the recombination radiation. These negative results are ascribed to the strong self-absorption in the crystal and to the optical inefficiency of the detecting system. Fig. 2 shows the results when a longitudinal magnetic field is applied to the crystal. In the sub-threshold region the magnetoresistance is now much less(s) than for the corresponding transverse case, as is expected for the electrons with their isotropic effective mass. There are two other marked differences between the longitudinal and transverse magnetic field data. In Fig. 2 the curve for H = 3500 oersteds exhibits a vertical rise in current density when E = 268 V/cm. This vertical rise is also observed for values of H = 1050 oersteds and 7000 oersteds, the onset electric field being a monotonically increasing function of the magnetic field. For H = 350 oersteds the curve (a portion of which is shown dotted) undergoes a peculiar loop in the interval 150-260 V/cm.