What is the SOS effect?

It is widely known that the forward/reverse switching of power semiconductor diodes in inductively loaded circuits is accompanied by overvoltage arising at these diodes, because in certain conditions the decay time of the reverse current may prove to be shorter than the duration of the high reverse conductivity phase during recovery of the diodes. For the purpose of traditional applications of the diodes as a.c. rectifiers this effect is unwanted, since it impairs reliability of the diodes and other circuitry elements. A great many methods for suppression of this effect have been developed. These methods rely either on selection of a certain distribution of dopants in the semiconductor structure (soft recovery diodes) or on the use of special circuit designs intended to suppress overvoltages at the diodes (by-passing varistors or combinations of R - C elements).

We tried to attack this problem from a different viewpoint and asked ourselves a question: Is it possible, oppositely, to enhance this effect and use high-voltage semiconductor diodes as opening switches in high-power pulsed systems with inductive energy storages? The experiments performed by S.K. Lyubutin, S.N. Rukin and S.P. Timoshenkov in 1991 [1] showed that, given a certain combination of the density of the forward and reverse currents and the pass time of the current through a semiconductor structure, the reverse current decay time decreases to tens and units of a nanosecond. Simultaneously, characteristic values of the current density equal tens of kA/cm2, while the current pass time is hundreds of nanoseconds. This effect of nanosecond interruption of superdense currents has been termed the SOS effect (Semiconductor Opening Switch) [5].

Difference of the SOS-effect from Other Current Switching Methods

The theoretical investigations performed by S.A. Darznek and S.N. Tsyranov [5, 14, 18 and 25] showed that the SOS effect represents a qualitatively new principle of current switching in semiconductor devices. The main distinction is that the current interruption process occurs not in the low-doped base of the structure, as is the case in other devices, but in its narrow high-doped regions. The base and the p-n junction of the structure remain to be filled with a dense excess plasma whose concentration is nearly two orders of magnitude higher than the initial level of doping. These two factors provide the combination of a high density of the interrupted current and the nanosecond cutoff time.

One more significant distinction is that at the stage of current interruption the SOS effect is characterized by an automatic uniform distribution of the voltage over series-connected structures. Therefore it is possible to construct megavolt opening switches [30] by simply connecting the structures in series without the use of external voltage dividers. The mechanism of the uniform voltage distribution is connected with an intensive avalanche multiplication of carriers in a narrow region of a high electric field in the structure at the stage of current interruption. Additional plasma is rapidly (tenth fractions of a nanosecond) produced at the expense of impact ionization in structures where an electric field arises for certain reasons. The increase in the plasma concentration is followed by the reduction of the electric field and equalization of voltage in the structures.

The Limiting Current Interruption Time with the SOS effect

Since the process of current interruption with the SOS effect is connected with the dynamics of the excess electron-hole plasma in the structure, the structure pumping conditions, which determine the distribution profile of the excess plasma concentration, also affect the current interruption time. We performed experiments [17, 19] taking short reverse pumping times and a high input rate of reverse current to the structure. Those experiments showed that the semiconductor opening switch, which is essentially a plasma-filled diode, possesses a property inherent in other plasma current interrupters: the current cut-off parameter is improved with increasing rate of the current input to the switch. When the reverse current input time was decreased from 80-100 ns to 10-15 ns, the current cut-off time was reduced from 5-10 ns to 500-700 ps. The registered current cut-off times were limited by the bandwidth of the measuring equipment.

The achieved current cut-off time confirmed the earlier statement that the process of current interruption develops in a narrow region of the structure, which is a few tens of micrometers wide, and does not require complete removal of the excess plasma from the diode base. Given the current cut-off time of 500 ps and assuming that at the stage of current interruption the region of the electric field expands at a maximum possible speed equal to the saturation velocity of carriers in silicon, this region will be not more than 50 чm wide. Further investigations are planned in order to clarify the picture of the phenomenon of subnanosecond interruption of current during the SOS effect.

Potentials of the SOS effect

Thanks to the aforementioned qualities of the SOS effect, powerful nanosecond generators boasting of record-breaking parameters among semiconductor switches were designed already in 2-3 years after the phenomenon had been detected. Using standard rectifier high-voltage columns as a semiconductor opening switch, we developed nanosecond generators having the output voltage up to 1 MV, average power of tens of kW, pulsed current of units and tens of kA, and the pulse power of the gigawatt level [3, 4, 6, 9, 10].