1024 THE BELL SYSTEM TECHNICAL JOURNAL, SEPTEMBER 1956 



conduction or the valence band. This in turn depends both on the surface 

 potential and on the ratio of cross-sections for transitions to the two 

 .bands. For F « — 1, one expects (52^/(95) y/(62s/ay)5 to have the value 

 — X~'; for F » +1, the vakie +X. These limiting values may be deduced 

 by a somewhat general argument. 



At some intermediate value of surface potential, the above ratio 

 must change sign. If the distribution of surface states in energy is known 

 from the field effect measurements, then the value of F at which the 

 above ratio changes sign determines the ratio of cross-sections for those 

 traps which are close to the Fermi level for that value of F. By repeating 

 the experiment for samples of differing bulk resistivity, it is then possible 

 to determine whether the same ratio holds for the states at some dif- 

 ferent position in the energy gap. 



III. EXPERIMENTAL DETAILS 



Fig. 1 shows the experimental arrangements. The sample of ger- 

 manium, of dimensions shown, was prepared by cutting, sandblasting, 

 etching in CP4 and washing in distilled water. The exposed faces were 

 approximately (100). The end contacts were made by sandblasting and 

 soldering. The slots A, A' in the ceramic were incorporated in order to 



GERMANIUM 



GOLD 



GOLD 



--GERMANIUM 



BINDING POSTS- 



Fig. 1 — Experimental arrangement. 



