This was/is a central activity of the laboratory (overlapping also with other researches) in which many laboratory members – former and present – were involved in different periods of time. Below, the milestones are presented
In the Soviet Union, the studies targeting establishment of the power semiconductor electronics were started in the mid-1950s at the Ioffe Physical-Technical Institute under a leadership of Prof. Vladimir Maximovich Tuchkevich. By the early 1960s, in collaboration with industry, the new technology was emerged enabling large-series production of power diodes and transistors, whose characteristics met the world standards in those years.
Active development of the conversion technique and, especially, of radio-electronics demanded for an enhancement of the electric power commutable by the semiconductor devices, and for a reduction of the switching time. In the mid-70s, investigations into switching dynamics of high-voltage power thyristors resulted in a creation of the new-class thyristors (named “modulation thyristors”) whose switching times, for devices operating at the voltages in excess of KiloVolt, were reduced from the usual level of several microseconds to less than 20 nanoseconds. In term of switching time, such devices were much superior to their foreign counterparts in those years, and already at the end of 1970s the Svetlana factory in Leningrad has launched mass production of high-speed modulation thyristors. However further reduction of the switching time within the double-injection mechanism, even if advanced to double-field-injection or to avalanche injection option, encountered limitations. Therefore, new approaches to the problem of a rapid current commutation were needed.
Impact ionization devices
In 1979, a phenomenon was discovered which enabled enhancement of the maximum power commutable by the semiconductor devices by almost three orders of magnitude, for the current rise time within tens of picoseconds. It was, namely, revealed that, if a reverse bias is very rapidly, faster than with a rate of 1012 V/s, applied to the high-voltage diode, the maximal value of such a bias may exceed the quasi-stationary breakdown voltage by 1.5-2 times. The subsequent abrupt reduction (relocation into a load) of the voltage was shown to occur within a time shorter by an order of value than the time of a carrier transit through the base of the device with the saturated velocity. In other words, there first occurs a retard of initiation of the impact ionization process. Further ultrafast switching is owing to formation of the impact ionization front near the p+-n-junction, which is propagating from the junction toward the opposite contact and is leaving the dense electron-hole plasma behind. The propagation speed can substantially exceeds the carrier saturation velocity. Due to filling of the base area with a plasma, the diode is switched from the blocking into the conducting state; this process seems to be the fastest ever known for a high-density plasma generation in semiconductors. For example, it enables to switch the silicon diode with a 2 kV blocking voltage into the on-state within just 40-50 picosecond.
It has also been proven that more complex two-pn-junctions- (i.e. transistor) or even three-pn-junctions- (four-layer dynistor) structures might be turned-on using an effect of the impact ionization front – and so the power current pulses with a large duration and a nanosecond rise time can be generated in the output.
Formation of the voltage pulses with a rise rate about 1012 V/s, launching the ionization front, became possible due to the effect of a superfast recovery of the blocking properties of the high-voltage diode after switching it from the conductive into the turn-off state. For this purpose, the widths and dopant concentrations of the emitter and base layers, the magnitude and duration of the forward-current pulse as well as the reverse-current rise rate should be properly adjusted. The goal of an adjustment is to warrant the conditions for meeting the plasma fronts – that arise under the reverse current flow – exactly in the plane of the p-n junction. At the next stage, the current would flow just due to the majority carriers moving with saturation velocity from the junction plane; there appears the space charge region and the current will be turned-off within nanoseconds. Such a mechanism of current turn-off can also come into effect in the structures with multiple pn-junctions.
The design of these devices, called the diode or dynistor pulse sharpeners and the fast turn-off diode, is rather simple – and one can easily connect them in series so that to build the high-voltage modules. Using such modules, the high-voltage pulse generators can be created, with a peak power in the range of tens of MWatt, which are widely used in all developed countries worldwide.
As demonstrated in the recent experiment with the Silicon and ZnSe n+nn+ structures without pn-junctions, they exhibit ultrafast (<200 ps) switching into the conducting turn-on state with a base n-layer uniformly filled with an electron-hole plasma, when the voltage pulse having a rise rate in excess of 1012 V/s is applied. This feature may be promising for creation of the wide-bandgap light-emitting devices.
Reverse Switching Dynistor
Usually, for commutation of large currents in case of a millisecond or longer pulse duration, the thyristor-like devices are used, whose four-layer n+-p/-n0-n+ Silicon structure is switched from the blocking to the conducting state by feeding a short current pulse via the control circuit between the n+ and p layers. The process of switching, first initiated in the small spot around the control electrode, is slowly expanding over the entire device area. However, in the sub-millisecond range, a substantial part of the area cannot get involved. For that range, a structure was elaborated, wherein, instead of a control electrode, the uniformly distributed n+-channels are incorporated into the p+-emitter layer. Such a device can be turned-on by means of a short-time reversing of the terminal voltage polarity; during that time, a current pulse passes in n+-channels creating a quasi-homogeneous thin electron-hole plasma layer. A saturable-core choke, connected in series with the thyristor, disconnects the control and power circuits for a short time. After the choke is saturated, the initial polarity of a voltage across the device is restored – and the external electric field extracts holes from the plasma into the p-layer provoking thereby an opposite electron injection from the n+-emitter, almost uniformly over the whole area. The switching therefore occurs simultaneously and uniformly, no matter how large the area is, that enables commutation of very high currents. Such a device was named a Reverse Switching Dynistor (RSD). The pulse generators with RSDs are employed in many countries.
High voltage opening switches
Further development in the field of pulse switching technique will be, to great extent, related with the silicon carbide (4HSiC) devices. In 2012 we elaborated the first device of this kind – a high-voltage opening switch.
The p+-p/-n0-n+ structure of this diode is fabricated by sequentially growing the layers onto the n+-wafer. After a plasma is created during the pumping-current pulse with a density of ~200 A/cm2, the reverse-current pulse with a density of 3.5 kA/cm2 is interrupted within a time shorter than 300 ps – and the terminal device voltage will increase up to ~2 kV.