The interaction between hydrogen and platinum is studied in n-and p-type silicon using deep-level transient spectroscopy. Hydrogen is introduced by wet-chemical etching or during crystal growth. In both cases we find that hydrogen forms only electrically active complexes with platinum. Four platinum-hydrogen related deep levels are identified: E͑90͒ at E C Ϫ0.18 eV, E͑250͒ at E C Ϫ0.50 eV, H͑150͒ at E V ϩ0.30 eV, and H͑210͒ at E V ϩ0.40 eV. These levels belong to at least three different platinum-hydrogen complexes. Level E͑250͒ is identical to the so-called midgap level in Pt-doped Si, which is believed to control the minority-carrier lifetime in Pt-doped silicon. Level H͑150͒ is an acceptor and is present both in n-and p-type samples after hydrogenation. It belongs to a platinum-hydrogen complex which contains more hydrogen atoms than the complexes responsible for the other hydrogen-related levels. Annealing at temperatures above 600 K results in a complete dissociation of all the platinum-hydrogen related defects and the substitutional platinum concentration is fully restored.

In the present paper a mechanism of diffusion and electrolytic conduction in solids is discussed, which is based on a formal treatment given by Frenkel. It is assumed that in a crystal in thermal equilibrium some of the atoms or ions are removed from their normal positions in the lattice to irregular ones in the interlattice space. Then diffusion and electrolytic conduction is possible by two processes; first by migration of the ions in the interlattice space, second by migration of the vacant places. The number of ions in the interlattice space can be calculated. If one considers the influence of polarization the result agrees in the order of magnitude with the observed data. It is also shown that the activation energy connected with the processes of movement is of the correct order of magnitude. Thus a satisfactory explanation of the exponential factor occuring in the empirical conductivity (or diffusion) formula is obtained. The constant factor multiplying the exponential can be explained at once for one group of cases; for the other group, where the constants apparently are too high, a tentative explanation is proposed.

No abstract

GaAs lightly doped with the heaviest group-V atom, bismuth (Bi), has been studied by conventional electron-spin resonance (ESR) and by ESR detected via the magnetic-circular-dichroism (MCD) absorption. A new Bi-related sharp-line MCD band has been observed on which two MCD-ESR lines have been discovered. They are shown to arise from the singly ionized Biz, double donor. Most remarkably, a substantial fraction, about 10%, of the total Bi content is found to occupy the Ga site. The BiG, MCD absorption band is tentatively assigned to an exciton deeply bound to the singly ionized double donor + Bi~,