No abstract

No abstract

attention in the recent years because of their use in the high In this paper, we report the effects of high-energy electron irradiation on the DC characteristics of polyimide passivated InPnnGaAs single heterojunction bipolar transistors. In contrast with the results of electron irradiation of unpassivated devices, the polyimide-passivated devices show much less degradation of current gain and no change in the collector output conductance. The decrease of collector current in the active regime is found to be typically -9 percent for a cumulative equivalent I-MeV dose of 2 . 7~1 0 '~ e/cm2 (-620 Mrad (InGaAs)). For low base currents, the devices show an .i -performance lightweight electronic systems in military and commercial space satellites [2-61. We have reported results on electron irradiation effects in unpassivated InPAnGaAs single heterojnnction bipolar transistors (SHBTs) and double heterojunction bipolar transistors (DHBTs) in our earlier studies [5,6]. Our results showed a significant degradation of the transistor performance for irradiation doses greater than 1x10" e/cm2. The degradation effects include a decrease of current gain, an increase in the output conductance and an increase of V C~,~.~, the collector -emitter volvage at which the transistor comes out of saturation. In this oaner, we investigate I I increase in the cnrrent gain for smaller doses (<2.5X1O8' e/Cm') the electron irradiation effects in polyimide passivated followed by a decrease at the higher doses. The increase in the InPAnGaAs SHBTs. Our present studies show that the electron current gain at low doses is attributed to the trapped charge in irradiation induced degradation of polyinlide passivated the polyimide layer near the periphery Of the B-E Junction. The devices is much less than that of the unpassivated devices. Our most significant effect of electron irradiation on the passivated show that the electron energy loss in the polyimide devices is a decrease in the slope of the IC-Vc~ characteristics layer is very low and hence the polyimide layer does not of some devices in the saturation regime. We believe this provide much shield'. we believe that the increased decrease in slope is caused by an increase in the collector series radiation tolerance of the po~yimide-passivated devices is due resistance after irradiation. Finally, devices with smaller to the smaller radiation damage at the peripheries of the baseemitter size are shown to have less radiation degradation than emitter and base-collector junctions as to the the larger emitter devices. This is explained by the smaller unpassivated junctions, radiation damage at the junction peripheries of the passivated devices. EXPERIMENTAL DETAILS

High-electron mobility transistors (HEMTs) have become an important device for high frequency and low noise applications. The maximum cut-off frequency, fT, that has been achieved thus far in a real device is 610 GHz [1]. Through simulation, we have been investigating how these frequencies may be pushed even higher. For this, we have used a full-band, cellular Monte Carlo transport program, coupled to a full Poisson solver to study a variety of InAs-rich, InGaAs pseudomorphic HEMTs and their response at high frequency, concentrating on devices with a structure (from the substrate) InP/InAlAs/InGaAs/InAlAs/InGaAs, with the quantum well composed of Ino.75Gao.25As [2,3,4]. By studying the frequency response behavior as the gate length was shortened in a scaled structure, we were able to establish in previous work a theoretical maximum forfT given a device that was 300 nm in total source-drain length. This upper limit was found to 2.9 THz for a device with an 18 nm channel thickness [3], and 3.1 THz for a 10 nm channel thickness [4]. These limits were achieved by, first relating the physical gate length to an effective length which included depletion extensions, and then extrapolating to zero physical gate length, while at the same time reducing the gate to channel distance. A key element of these calculations was the establishment of an accurate definition for the effective gate length. Importantly, we found that this quantity was considerably shorter than the depletion length when one used the results of full RF simulations in conjunction with an integration over the velocity in the gate region.In the present work, we consider other physical factors beyond the physical gate length, which may enhance or inhibit fT, to see if the limits discussed above are indeed achievable in real devices. For example, it has been traditional microwave transistor design to use an Lsg that is about a third the total device length (100 nm in this case), which is what was used in the previous simulations to extract the upper limits [3,4]. However, we have found that when Lsg is reduced below this value, fT is increased, by more than 10%. We believe this to be a result of a reduction in the effective gate length of the device, a quantity that our previous limit studies showed to be crucial.The lower the effective gate length is for a given physical gate length, the higher fT can be. We shall also discuss other means by which this may be achieved, and we will present a thorough investigation of device operation as a function of gate and drain voltage to examine where a "sweet spot" for fT may be found. An important physical limitation for device operation is gate leakage. Of course, this can be reduced by increasing the gate to channel separation, and we study how far this separation can be made without sacrificing too much in terms of a highfT. As we shall show, even increasing the gate to channel separation by a more than a factor of three, one can still achieve an fT greater than 2 THz in a device with a short physical gate length, an...

No abstract