Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
Accurate and Fast STT-MRAM Endurance Evaluation Using a Novel Metric for Asymmetric Bipolar Stress and Deep Learning
Ist Teil von
2022 IEEE Symposium on VLSI Technology and Circuits (VLSI Technology and Circuits), 2022, p.373-374
Ort / Verlag
IEEE
Erscheinungsjahr
2022
Quelle
IEEE Xplore
Beschreibungen/Notizen
A framework for STT-MRAM endurance test and prediction has been successfully developed. To improve prediction accuracy, we have constructed a Unified Voltage Polarities Power-law model (UVPP) using a novel stress metric by unifying unipolar, symmetric bipolar and asymmetric bipolar stresses. To keep TDDB test within a realistic time frame, we have applied a time series deep learning model, Transformer, for forecasting MTJ resistance series data of 10 6 -10 15 cycles and breakdown points by using tested resistance series data of 1-10 6 cycles, which reduces test time by up to 10 9 x. The proposed framework enables fast STT-MRAM reliability verification for high-volume mass production.