CHEMICAL INTERACTIONS AMONG DEFECTS IN Ge AND Si 577 



lowered. In ion pairs it is in fact to be expected that the donor level will 

 be moved up into the conduction band and the aceptor level down into 

 the valence band.* This change in energy level structure should be ap- 

 parent in Hall coefficient measurements at low temperature. Experiments 

 of this sort have been conducted and are reported in this paper. Under 

 certain conditions this phenomenon may be useful for the elimination of 

 trapping"*^ levels from the forbidden gap. 



(C) Change of Carrier Mohilitij 



Ion pairs possess dipolar fields, and consequently, scattering cross-sec- 

 tions very much smaller than those of point charges. The addition of 

 hthium to a sample under such conditions that more than half the added 

 lithium becomes paired should therefore increase rather than decrease 

 the mobility of holes. The latter effect is the one to be expected in the 

 absence of pairing. In other words not only carriers but also the scat- 

 terers are removed by compensating the acceptor with donor. Experi- 

 ments of this sort have been performed. They are described later in this 

 paper. Since they allow us to measure the degree of pairing with good 

 accuracy they have been very valuable in validating the theory, and also 

 in exploring the nature of the potential function in the neighborhood of 

 an isolated acceptor. 



(D) Relaxation Times 



A semiconductor containing unpaired donors and acceptors at one 

 temperature can be cooled to a lower temperature, and the impurities 

 should then pair. If the temperature is lowered sufficiently, the pairing 

 process will be slow enough to be followed, kinetically, by observing any 

 parameter (such as carrier mobility) sensitive to pairing. Experiments of 

 this sort have been performed and will be described later. 



The process of pairing can be characterized by a calculable relaxation 

 time, which depends on the acceptor concentration, the diffusivity of 

 the mobile donor, the dielectric constant, and the charges on the ions 

 among other things. The measured time can therefore be used as a means 

 of determining any one of these parameters. 



(E) Diffusion 



It is evident that pairing should reduce the diffusivity of a mobile 

 donor. Studies of diffusion in the presence of an immobile acceptor should 



* A rough calculation indicates that about 0.5 e.v. would be required to place 

 an additional electron on an ion pair. 



