Analysis of High-Temperature Effects on InAs/In0.3Al0.7As/InSb/In0.3Al0.7 As pHEMTs on Accessing RF/Analog performance: A Machine Learning Predictive Modeling

In this paper, we delve into the intriguing realm of Pseudo-morphic High Electron Mobility Transistors (pHEMTs) composed of InAs/In0.3Al0.7As/InSb/In0.3Al0.7 layers, utilizing Silvaco-TCAD for simulation. Our focus centers on the assessment of RF and analog electrical characteristics, with a keen eye on the high-temperature effects. The influence of temperature on device performance is meticulously evaluated in comparison to a reference device operating at room temperature. Traditionally, the critical parameters such as threshold voltage ( V-th ), transconductance ( g(m) ), and I-on/I-off ratio have been calculated within the temperature range spanning from 300 K to 700 K. The primary pHEMT device in our study exhibits impressive attributes, featuring a drain current of 950 mA, a threshold voltage of -1.75 V, a high transconductance ( g(m) ) value of 650 mS/mm, an I-on/I-off ratio of 1x10(6) , a transition frequency ( ft ) soaring to 790 GHz, and a maximum frequency ( f(max) ) reaching a staggering 1.4 THz. However, as we traverse the temperature spectrum, our findings unveil a compelling narrative. The impact of rising temperature is unequivocal, triggering a cascade of transformations within the device. Notably, as the temperature escalates, we observe a noticeable decrease in current, a reduction in transconductance (g(m) ), and a diminishing I-on/I-off ratio. To unravel the intricacies of these temperature-induced effects, we introduce the infusion of Machine Learning (ML) into our analysis.