ELECTRICAL XOISE IN SE.MICO.NDUCTOHS 



971 



0.9 



O 0.7 



0.6 



Z 0.5 

 O 



0.1 0.2 0.3 0.4 0.6 0.8 1.0 2 3 4 6 8 10 20 



RATIO OF LIFE PATH TO SEGMENT LENGTH 



Fig. 15 — Noise correlation. The dotted curve includes allowance for losses at 

 the side arms. 



by a factor of two. In Fig. 16 the form of the experimental curve seems 

 different from that calculated. In particular, the slope is steeper, and 

 the curve tends to level off at a correlation of about 0.8. It seems pos- 

 sible to explain the discrepancies between the experimental data and 

 the calculations on the basis of two considerations which were not in- 

 cluded in the model, (a) The pair of side arms separating the two seg- 

 ments of the filament serve to drain off some holes which would other- 

 wise contribute to the correlation. The dashed curve in Fig. 15 shows 

 the calculated effect, on the assumption that the absorption in the side 

 arms is equivalent to an extra section of filament equal in length to half 

 a segment. The actual distance across the side arms is only 20 per cent 

 of a segment, but it is not hard to believe that the decay rate in this 

 region might increase by a factor of two or three due to the reduced elec- 

 tric field and loss of holes down the side arms, (b) The model assumed a 

 unifoim distribution of noise sources along the filament. There is ex- 

 perimental evidence that the distribution may be quite spotty. This 

 can have a substantial effect on the form of the correlation curve. For 

 example, the dashed curve in Fig. 16 shows the curve calculated for 

 noise sources lumped at the mid-point of each segment. Other assumed 

 positions might shift the curve considerably along the horizontal axis. 



