No abstract

The energy and niomentiiin relaxation of conduction electrons by optical phonons in lnSb are studied using the energy band structure calculated by Kane. For simplicity, the drifted Maxwellian approach is used and its form for a n arbitrary band structure is presented. KO intrinsic breakdown is obtained. On the contrary, the model gives a voltagecontrolled differential negative mobility for high fields. I n calculations of electron temperature and mobility in 111-V semiconductors the drifted Maxwellian approach has been used by several authors with various degrees of approximation [I t o 41. Optical phonon scattering has been considered t o be the dominant scattering mechanism and the band structure used has been simple parabolic bands. However, as has been shown by Kane [5], non-parabolic effects are very pronounced inInSb. The purpose of this paper is to present analogous calculations of mDbility and electron temperature using Kane's band structure.I t is not a t once obvious how the drifted Maxwellian should be written in a semiconductor with non-parabolic energy bands. One form was presented in [el, where a drift-wave number was introduced. This, however, seems to be more of an a d hoc assumption. The definition of the drifted Maxwellian should be that the distribution is Maxwellian in a coordinibte system moving with the drift velocity of the charge carriers. The relation between the energy as a function of the wave number in a coordinate system a t re&, E U ( k ) , and one moving with the drift, velocity of the charge carriers, E ( k ) , can be written This is just the velocity addition law for two coordinate systems moving with relative velocity vd. I€ this relation is integrated the energy as a function of the wave number in the moving coordinate system is obtained :

^)/f 3.0 W 1.0 o d-8mm a d-12mm • d-16mm o d-20rnm A d-32mm • a • • o A AO o O ° MAIN RES. SECOND RES. THIRD Rg$.' 10 10 kl m d w' ft: FIG. 2. The diagram of (ojp 2 )/oo 2 versus kT e /md 2 uvariable for the second and the third resonance is made plausible by the fact that a smooth curve can be fitted to the experimental points in Fig. 2. The main resonance is seen to be of different character. This resonance, however, has been explained in the theory by Herlofson, and, as a particular solution, in the theory by Gould.Following Herlofson and taking into account the finite plasma radius and the surrounding glass walls, we can compute (u)p 2 )/io 2 for the main resonance in a homogeneous plasma column. The results of this computation are compared with the measured values of (u>£ 2 )/a> 2 in Table I, and the agreement is found to be good.

No abstract

No abstract