DEVELOPMENT AND MODELING OF A NEW FAMILY OF INJECTION-GATED DOUBLE-INJECTION SWITCHING DEVICES.

摘要

The double injection (DI) devices consist of an anode (a p('+) contact) and a cathode (a n('+) contact) for hole and electron injection, respectively, into a high resistivity semiconductor substrate containing deep traps. With the appropriate concentration of the deep traps, thermal free carriers, and the separation between the anode and the cathode, the DI diodes exhibit S-type switching similar t the SCR's. To control the switching behavior of the DI devices, a p('+) gate has been diffused in the "channel" region between the anode and the cathode. Some extremely interesting characteristics have been observed in these injection-gated DI devices.;These injection-gated DI devices are fabricated in the planar configuration using the conventional microfabrication technology. The starting material is 5 to 10 (OMEGA)-cm. silicon wafers of about 10 mils thickness; the orientation of the surface, seemingly, is not of major significance in these devices. The p('+) and n('+) contacts are made by thermal diffusion using BN and POCl(,3) as sources, respectively, while keeping the oxide at the back of the wafer intact. Prior to the gold diffusion, this oxide is removed from the back and an indirect-source diffusion of gold is performed. Next, the contact windows are opened, aluminum is vacuum evaporated, the metal pattern is etched, and finally the devices are sintered in the nitrogen ambient.;Some extremely interesting results obtained with these injection-gated DI devices have already been reported in the M.S. thesis by the author. They include the SCR-like behavior of DI devices, with the gate exercising very sensitive control over the threshold voltage V(,Th), and the S-type differential negative resistance (DNR) being converted into a variable N-type DNR as a positive gate-cathode bias V(,GC) was applied to the appropriately placed gate electrode. Some of the recently fabricated injection-gated DI devices have shown that their holding voltage V(,H) can be reduced to zero, or even to a negative voltage, by applying sufficient negative V(,GC). Also, an optimum location for the gate placement in the channel has been sought. A number of experiments have been designed and performed to investigate especially into the post-threshold behavior of these DI devices.;The DI diodes, with properly adjusted parameters, show pulse width modulation (PWM) effect, which is a unique quality of these devices. The PWM effect has been characterized in terms of the dwell time or t(,d), which is the interval between the leading edge of the input pulse and the point at which leading edge of the output pulse reaches 50% of the peak value. It was observed that the inverse of the dwell time (1/t(,d)) showed excellent linear correlation with the input pulse peak voltage Vp. The dwell time was found to be affected by external magnetic field also. Transient behavior of the three terminal device showed that for a certain range of voltages, the anode and gate pulses had opposing effects on the output voltage, i.e. increasing the anode pulse peak voltage decreased the output pusle width, or increased the dwell time, while the gate pulse had the oppposite effect on the output. This effect was observed even when the anode was pulsed a few tens of (mu)s before the gate.;This dissertation also contains; a theoretical investigation into the deviations from the low-injection square law behavior due to a finite electric field at the anode, some refinements in the theoretical model for the N-type DNR, a phenomenological model for the gate-controlled holding voltage, in depth modeling of the dwell time variation with the pulse peak voltage and the effects of the magnetic field upon the dwell time, and finally, a semi-quantitative model of the transient behavior of the three-terminal injection-gated DI devices. The chapter on the analysis of the data and modeling, forms the major portion of the dissertation.