TY - GEN
T1 - Total Ionizing Dose Effect in Tri-gate Silicon Ferroelectric Transistor Memory
AU - Aabrar, Khandker Akif
AU - Read, James
AU - Kirtania, Sharadindu Gopal
AU - Stepanoff, Sergei
AU - Wolfe, Douglas E.
AU - Yu, Shimeng
AU - Datta, S.
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - For the first time, we demonstrate the survivability of programmed states in a ferroelectric field effect transistor (FeFET) memory under gamma-ray irradiation. The un-irradiated Tri-gate Si FeFET (control) shows a large memory window (MW) of 1.6V, a high read current window of 2×105, a long retention of 2.3 times 10 {6} mathrm{s} (27 days) and high endurance (>108 cycles). The Tri-gate FeFETs, when exposed to radiation, retain ferroelectric hysteresis with a current window >102 up to a high radiation dose of 10Mrad. This confirms the survivability of the FeFETs and makes FeFET a potential candidate for data storage and compute-in-memory (CIM) in harsh ionizing environment. Analysis of array-level performance of FeFET-based CIM accelerator trained on CIFAR-10 dataset using VGG-8 neural network model under 5Mrad and 10Mrad radiation shows inference accuracy of 90% and 80%, respectively, versus 92% for control.
AB - For the first time, we demonstrate the survivability of programmed states in a ferroelectric field effect transistor (FeFET) memory under gamma-ray irradiation. The un-irradiated Tri-gate Si FeFET (control) shows a large memory window (MW) of 1.6V, a high read current window of 2×105, a long retention of 2.3 times 10 {6} mathrm{s} (27 days) and high endurance (>108 cycles). The Tri-gate FeFETs, when exposed to radiation, retain ferroelectric hysteresis with a current window >102 up to a high radiation dose of 10Mrad. This confirms the survivability of the FeFETs and makes FeFET a potential candidate for data storage and compute-in-memory (CIM) in harsh ionizing environment. Analysis of array-level performance of FeFET-based CIM accelerator trained on CIFAR-10 dataset using VGG-8 neural network model under 5Mrad and 10Mrad radiation shows inference accuracy of 90% and 80%, respectively, versus 92% for control.
UR - http://www.scopus.com/inward/record.url?scp=85147530152&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85147530152&partnerID=8YFLogxK
U2 - 10.1109/IEDM45625.2022.10019566
DO - 10.1109/IEDM45625.2022.10019566
M3 - Conference contribution
AN - SCOPUS:85147530152
T3 - Technical Digest - International Electron Devices Meeting, IEDM
SP - 3271
EP - 3274
BT - 2022 International Electron Devices Meeting, IEDM 2022
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2022 International Electron Devices Meeting, IEDM 2022
Y2 - 3 December 2022 through 7 December 2022
ER -