I THEORY OF THE SWEPT INTRINSIC STRUCTURE 1275 



i| fent generation. It will be convenient to choose a; = so that .t„ = 

 —Xp/h. Then the origin is nearer to the NI junction for 6 > 1. Now 

 from this and the boundary conditions (6.4) and I = 2qgL we have the 

 positions of the junctions: 



L 1 + 6' L 1 + 6 ^^'^^ 



A.S before, the junctions are at .r = ± L for 6=1. 



We can now find the fictitious carrier flow Jp + J„/6 and the fictitious 

 current q{Jp — Jn/i>) as functions of x. 



■fp+T= (M^) !'^ (6.6) 



where the dimensionless parameter j8 = (6^ — l)/46. Thus the fictitious 

 current q(Jp — J„/b) is equal to the actual current times a linear func- 

 tion of X. This function is always positive and varies from a minimum of 

 1/6 to a maximum of 1. 

 Combining (6.6) with (6.1) and integrating gives the equation 



that we had before. Now, however, E is not a minimum at the same point 

 where s is a maximum. As before, when recombination is negligible 

 throughout all of the intrinsic region, A determines the voltage; and, 

 when recombination is important over part of the region, A determines 

 both the voltage and the length of the intrinsic region Xp — Xn > 2L = 

 \/o 



' ibining (6.7) with (6.2) gives 





(6.9) 



which is similar to the previous (3.6) except that / is nuiltiplied by the 

 factor 1 + j3.r/L, which ^'aries from 1 + 1/6 to 1 + 6. The same argu- 

 ments used in Section V apply here and show that the second term in 

 brackets (the diffusion term) can be neglected except near the junctions. 

 In other words, although / is always part drift and pai't diflusion, 

 7(1 + ^x/L) is approximately pure drift except at the junctions. 



Eliminating s between (6.9) and (6.8) and neglecting the diffusion 



