The use of epitaxial junctions, especially heterojunctions, like InGaAs/InAlAs, Al- GaN/GaN, has yielded transistors switching at high frequency or high voltage. It is however generally found that a high frequency transistor remains deficient in breakdown voltage and vice versa, a limitation of heterojunctions between similar materials by epi- taxial methods. Wafer bonding, on the contrary, allows junctions between mismatched materials, and so may be a promising technique to overcome the trade-off between high frequency operation of and the available power from a transistor. This work focuses on the experimental demonstration of a well-behaved wafer-bonded junction and transistor, namely, an InGaAs/InGaN junction and an InGaAs/III-Nitride transistor, respectively. The latter is referred to as a wafer-bonded current aperture vertical transistor, BAVET, comprising a channel in InGaAs and a drift region in InGaN/GaN. In this work, key fea- tures of a BAVET are identified, designed, and fabricated. Fundamental to the operation of the transistor is the design of a conductive aperture and an insulating current-blocking layer (CBL). This property in a BAVET results in an on current as high as 600 mA.mm−1 and a transconductance of 132 mS.mm−1. Despite the fact that these results mark the first demonstration of transistor operation in a BAVET, other aspects of the on- and off- state performances remain weak. For instance, saturation voltage, on-resistance, turn-on voltage, and output conductance are anomalously high and off-state pinch-off is poor.
Mechanisms pertaining to virtual gate, drain resistance, weak field plating and trap xi
ionization are found to be the cause of these anomalies. Experiments prove that all anomalies are local to the wafer-bonded interface (WBI). This work determines that passivation of traps at WBI is a solution to these drawbacks in device performance. An in-situ trap passivation process is devised, which uses hydrogen species and the layer structure electrostatics to effect trap activity. This improves all aspects of the BAVET performance, with the saturation voltage, drain resistance, turn-on voltage and output conductance significantly lowered and critical field to trap-ionization made higher.
In addition to designing the BAVET, the dissertation makes progress in understanding the physics of WBI and its relation to functioning of a BAVET. In passivating traps at WBI, a means to improve the wafer-bonded junction and transistor is conceived. The work succeeds in demonstrating a trap-free wafer-bonded InGaAs/InGaN junction, an anomaly-free III-Arsenide/III-Nitride vertical transistor and so reveals a new space of possibilities in the field of wafer-bonded electronics.