ELECTKU'AL XOISK l.\ SKMIt'OMDUCTORS 91)3 



applies to pr , the coiu'eutratioii of holes emitted from ihc li^ht-haiid 

 side. 



UndcM- the iiilliuMice of a magnetic field pusluii<i holes toward th(> rifi;ht, 

 the eoneeiit rations will ('hanj«;<' to those shown in part (b) of the fignre. 

 The magnetic field will {)nll holes through the filament and tend to pre- 

 voni diffusion from right to left. For some moderate value of field, the 

 value of ./■> is not increased enough to change ./i appreciably, so the 

 value of pi is nearly the same as with no field. At tlu^ same time the 

 efTects of diffusion are suppressed by the field so that the concentration 

 pi extends nearly to the right side of the figure. By the same action, (he 

 concent i-ation of holes emitted from the right surface drops to zero very 

 quickly. 



From the curves of Fig. 2 and relation (2), we see that the ai'ea under 

 the density curve, and hence the lifetime of holes injected at the left is 

 at most doubled by the magnetic field, while the lifetime of those in- 

 jected at the right is reduced nearly to zero. Recalling that the noise 

 is proportional to a summation of the scjuare of the lifetimes, we see 

 that the noise power is at most doubled at a suitable value of magnetic 

 field. 



Higher \'alues of field will sweep so many holes to the right-hand 

 surface as to substantially reduce pi , so at very high fields the noise 

 decrea.ses monotonically to zero. 



Thus it is seen that the noise behavior is the result of competing ten- 

 dencies. On the one hand, the magnetic field helps holes escape from 

 the surface at which they are emitted, but on the other hand it tends to 

 push these holes against the opposite surface and thereby reduce their 

 lifetime. The relative importance of those two tendencies depends on 

 the surface recombination properties and the strength of the magnetic 

 field. 



Calculation of the lifetime along the lines just discussed involves solu- 

 tion of the continuity eciuation 



ax- ax 



with suitable boundary conditions. The results of such a calculation 

 carried out by Shockley and Suhl in tlie work already referred to are 

 plotted in Fig. 10. , 



In order to make the results independent of sample dimensions, the 

 following parameters are used. The first i)arameter is proportional to 

 the applied magnetic field, and is defined as the effecti\e transverse 

 potential in units of kT/q: 



