<title>Sudden breakdown of bulk semiconductor switches</title>

Abstract

Order of magnitude estimates suggest that optically controlled bulk semiconductor switches should be able to withstand voltages up to the product of their thickness and the dielectric strength of their material. In reality, however, the devices fail -i.e., exhibit a behavior that resembles dielectric breakdown -already at voltages which are much lower. This deficiency threatens to limit the prospects of the device concept quite seriously and has so far not completely been understood. In our paper, we discuss several mechanisms which may underlie the observed phenomenon, and focus in particular on the dynamical aspects of it, namely on the sudden transition ("sudden breakdown" ) which takes the switch within a few ns from from the resistive off-state to a highly conductive on-state. We investigate a scenario that relates this transition to a second effect also seen during breakdown, namely to the spontaneous onset of current filamentation, and speculate that the magnetic self-contraction of the current (known as the "pinch effect") may play an essential role in the process. On the basis of a mathematical device model which incorporates the effects of particle transport and magnetic interaction, we obtain quantitative results for the speed and the threshold of magnetically driven filamentation, and find those numbers to lie in the A and the ps region, respectively. We conclude that the magnetic pinch may play a essential role in the dynamics of current filamentation and fast breakdown, but cannot explain the fast observed current rise in the ns-range by itself.

<title>High-power switching with electron-beam-controlled semiconductors</title>

“…As the transition to the lock-on mode is related to the filling of the deep carrier traps, we can roughly estimate the closure time by r = N/S, where N 1016 is a typical trap density. To obtain a more quantitative formulation, we have to employ a complete transient simulation of equations (1) - (4). For the sake of conciseness, we confine ourselves to the discussion of one case which is typical for our experimental results.…”

“…(4) We have solved the equations numerically under the assumption of steady state, closely following a procedure which is more extensively discussed in reference [2], and in the transient case where we have implemented a hybrid code which combines a Lagrangian treatment of the free carriers with an Eulerian description of the trapped charges. (This approach, which virtually eliminates the spurious numerical diffusion inherent to finite differencing methodes, ensures a satisfactory accuracy even for a relatively small number of discretization points.)…”

“…Recently, much attention has been focused on the non-linear behavior of optically and electron beam controlled semiconductor switches, concentrating on features like the non-linear dark current characteristics, the break-down voltage, and on the so-called lock-on effect [3,4,5]. The experimental results of different groups vary considerably.…”

“…The strong dependence of the switch characteristics on impurities and defects is on one hand an obstacle for standardization ofsemiconductor switches, but can also be seen as an opportunity: "Tailoring" of the switch material becomes possible. Deep traps in the band gap are responsible for the high resistance of semi-insulating GaAs and determine largely the I-V-characteristics of the material [4]. Some of them, like the generally dominant electron trap EL2, are related to defects resulting from the composite nature ofGaAs and are hence intrinsic to the material, others, however, can be introduced by doping the crystal with suitable acceptors or donors.…”

Filamentation in Optically Controlled Semiconductor Switches

“…It is interesting to compare this result with that of our previous investigation [1], where we had neglected the dynamical influence of the charge carrier generation, but evaluated the rest of the dynamics with a somew'hat more sophisticated model. In the correct limit (i.e., neglecting impact ionization here and the diffusion pressure in the other model), the two results agree up to a numerical factor of order unity, reassuring us of the order-of-magnitude validity of our results.…”