Design and Analysis of the performance of strained Ge-based pin switch through Machine Learning framework for application in mm-wave frequency

We discuss the design and analysys of the performance of a strain modulated Ge/Ge_0.98Sn_0.02 vertical channel pin-based switch for application in mm-wave frequency. The device's performance in the mm-wave region is assessed using a Nano-mixed Quantum Corrected Strain Modified Drift–Diffusion Nonlinear mathematical (NQCSM-DD) model along with Machine Learning Framework. The study investigates the switching characteristics of the device, considering V-I characteristics, reverse recovery time, power dissipation, Insertion Loss (IL), and Isolation (ISOL).The inherent material attributes of the DUT (Device Under Test) are improved considerably by the addition of 2% of Sn into the intrinsic Ge material. The NQCSM-DD model is calibrated by analyzing the experimental and simulated performance of a flat structure-based Si pin device under similar circumstances. The detailed investigation and analysis proves that the switching performance of the proposed DUT is significantly enhanced. The results, compared with the super-lattice structure-based GaN/AlGaN pin device, show that Ge/Ge_0.98Sn_0.02 outperforms its GaN/AlGaN counterpart in terms of reverse recovery tim, power dissipation, and, IL and ISOL. The proposed DUT offer low IL (0.121 dB and 0.03671 dB for series-shunt & shunt SPST switches, respectively) and high ISOL (69.72 dB and 80.23 dB for series-shunt & shunt SPST switches, respectively) at 120 GHz . Furthermore, the Random-Forest-Regression (R-F-R) model within a Machine Learning Framework (MLF) is applied to determine the device’s efficiency. The proposed model’s reliability study is reported in this paper in details. Ge/Ge_0.98Sn_0.02 vertical channel pin-based device for the application in mm-wave frequency.