TY - GEN
T1 - Machine Learning Approach to Characteristic Fluctuation of Bulk FinFETs Induced by Random Interface Traps
AU - Butola, Rajat
AU - Li, Yiming
AU - Kola, Sekhar Reddy
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Interface traps are of particular concern for highly scaled-down semiconductor devices. They cause trapping and de-trapping of charge carriers and have an adverse effect on device characteristics and variability. Therefore, in this work, the influence of randomly generated interface traps (RITs) on device characteristics of 16-nm-gate high-κ/metal gate bulk fin field-effect transistors (FinFETs) is investigated for experimentally validated simulated data. A machine learning (ML) model is proposed here to imitate the device simulation results. The impact of variation of these multi-point defects is analyzed by generating RITs at the interface of gate-oxide and silicon channel of the explored bulk FinFETs. The statistical fluctuations induced by RITs are analyzed by predicting the variations in threshold voltage (VTH), subthreshold slope (SS), drain-induced barrier lowering (DIBL), off-state current (IOFF), and transconductance (gm) using the proposed ML model with high accuracy and small error, in much less computational cost. This work shows the possibility of accelerating the random defects analysis using the technique of machine learning.
AB - Interface traps are of particular concern for highly scaled-down semiconductor devices. They cause trapping and de-trapping of charge carriers and have an adverse effect on device characteristics and variability. Therefore, in this work, the influence of randomly generated interface traps (RITs) on device characteristics of 16-nm-gate high-κ/metal gate bulk fin field-effect transistors (FinFETs) is investigated for experimentally validated simulated data. A machine learning (ML) model is proposed here to imitate the device simulation results. The impact of variation of these multi-point defects is analyzed by generating RITs at the interface of gate-oxide and silicon channel of the explored bulk FinFETs. The statistical fluctuations induced by RITs are analyzed by predicting the variations in threshold voltage (VTH), subthreshold slope (SS), drain-induced barrier lowering (DIBL), off-state current (IOFF), and transconductance (gm) using the proposed ML model with high accuracy and small error, in much less computational cost. This work shows the possibility of accelerating the random defects analysis using the technique of machine learning.
KW - Bulk FinFET
KW - characteristic fluctuation
KW - interface trap
KW - machine learning
KW - random forest regressor
UR - http://www.scopus.com/inward/record.url?scp=85133757083&partnerID=8YFLogxK
U2 - 10.1109/ISQED54688.2022.9806233
DO - 10.1109/ISQED54688.2022.9806233
M3 - Conference contribution
AN - SCOPUS:85133757083
T3 - Proceedings - International Symposium on Quality Electronic Design, ISQED
BT - Proceedings of the 23rd International Symposium on Quality Electronic Design, ISQED 2022
PB - IEEE Computer Society
T2 - 23rd International Symposium on Quality Electronic Design, ISQED 2022
Y2 - 6 April 2022 through 7 April 2022
ER -