Journal of Engineering Science and Technology Vol. 18, No. 6 (2023) 3077 - 3096 © School of Engineering, Taylor’s University 3077 PREDICTIVE ANALYTICS OF JUNCTIONLESS DOUBLE GATE STRAINED MOSFET USING GENETIC ALGORITHM WITH DOE-BASED GREY RELATIONAL ANALYSIS K. E. KAHARUDIN 1,2 , F. SALEHUDDIN 1,* , N. A. JALALUDIN 1 , F. ARITH 1 , A. S. M. ZAIN 1 , I. AHMAD 3 , S. A. M. JUNOS 1 1 Micro & Nano Electronics (MiNE), CeTRI, Faculty of Electronics and Computer Technology and Engineering, Universiti Teknikal Malaysia Melaka, Hang Tuah Jaya, Durian Tunggal, 76100 Melaka, Malaysia 2 Faculty of Engineering, Lincoln University College (Main Campus), Wisma Lincoln, 47301 Petaling Jaya, Selangor, Malaysia 3 College of Engineering (CoE), Universiti Tenaga Nasional (UNITEN), 43009 Kajang, Selangor, Malaysia *Corresponding Author: fauziyah@utem.edu.my Abstract This paper explores the application of Genetic Algorithm (GA) incorporated with design of experiment (DoE) based on Grey Relational Analysis (GRA) in predicting the optimal design parameters of n-type Junctionless Double Gate Strained MOSFET (JLDGSM). The GRA is applied to predict the optimum level of multiple design parameters in attaining the best multiple device characteristics. The GA approach is applied to further optimize the design parameters for much improved device characteristics. The initial step is to select the best possible level of four design parameters (Ge mole fraction, high-k material thickness, source/drain doping concentration and metal work-function) within specific upper and lower boundary limits. The predictive analytics are initiated with the employment of GRA in finding the grey relational grade (GRG) which represents the multiple electrical characteristics (ION, IOFF, on-off ratio, gm, fT and fmax) for 18 sets of experiment. The computed GRGs are then processed using multiple regression analysis to derive the objective function that summarizes the relationship between the design parameters and the GRG. Finally, the genetic algorithm is utilized to predict the optimum level of design parameters based on the derived objective function. The final result reveals that the proposed predictive analytics have successfully optimized ION, IOFF, on-off ratio, gm, fT and fmax of the device by ~34%, ~40%, ~50%, ~18%, ~10% and ~4% respectively. The best combinational magnitudes of Ge mole fraction, Thigh-k, Nsd and WF for the most optimum device characteristics are predicted to be 0.1 (10%), 3 nm, 3×10 13 cm -3 and 4.6 eV respectively. The results exhibits significant potential for junctionless transistor revealing the alternative way and configuration in developing future highly efficient nano-scaled devices and ion-sensitive sensors. Keywords: Maximum oscillation frequency, Off-current, On-current, On-off ratio transconductance, Unity-gain frequency.