802 THE BELL SYSTEM TECHNICAL JOURNAL, JULY 1956 



if) 



o 

 > 



2 

 > 



50 100 150 200 250 300 350 400 450 500 

 TIME IN MICROSECONDS 



Fig. 15 — Forming voltage pulse for HF treated surface. 



undergoes a succession of breakdown and recovery intervals. In Fig. 

 l-l(b) is the accompanying plot of current against time. Comparison of 

 these two plots shows that following the application of the voltage, the 

 resistance of the contact decreases until a rather sudden more rapid de- 

 crease in resistance occurs, taking place at time h . In view^ of this time 

 scale, the first decrease can be attributed to a heating of the contact, 

 a form of thermal breakdoAvn at the metal-semiconductor surface.^^ 

 Any reason invoked to account for the second more rapid decrease in 

 resistance must account for the short time (a few /xs) in which this change 

 occurs. In any event, shortly after the second "breakdown," a quenching 

 results, with the collector resistance returning to a \'alue nearer to its 

 original value. This sequence of events is roughly repeated until the 

 condenser is discharged. 



The properties of the contact at nominal reverse voltage and currents 

 are usually changed as soon as one such condenser discharge pulse has 

 occurred, and often one such pulse is sufficient to reach the forming ob- 

 jective. A typical forming pulse obtained under similar conditions to 

 those for Fig. 14 is shown in Fig. 15, with the exception that the surface 

 has been treated in HF for a few minutes. On this case it is apparent that 

 the second, rapid breakdown is entirely absent. The well-defined form- 

 ing pulse of Fig. 14 is usually obtained on surfaces with good pre-forming 

 diode characteristics, and results in production of a usable transistor. 



From results of the previous sections it is well established that etched 

 surfaces treated with reagents in which germanium dioxide is soluble 

 provide point contact diode characteristics unsuited to electrical pulse 

 forming. 



It is often assumed, on the basis of the results of Valdes,^ that forming 

 effects result from the diffusion of impuiities from the point into the 

 semiconductor during the forming pulse. Since the high temperature 

 required for such diffusion results from the power dissipated at the metal 



