Saxena, NishantNishantSaxenaWang, ChengchengChengchengWangHu, ZeyuZeyuHuWang, GuoshengGuoshengWangNaqi, MuhammadMuhammadNaqiZhang, LinxiLinxiZhangZhang, WeidongWeidongZhangChai, ZhengZhengChaiGarbin, DanieleDanieleGarbinDegraeve, RobinRobinDegraeveClima, SergiuSergiuClimaRavsher, TarasTarasRavsherBelmonte, AttilioAttilioBelmonte2026-09-232026-09-2320260018-9383https://imec-publications.be/handle/20.500.12860/60464Ovonic threshold switch (OTS) devices are used as selectors for emerging nonvolatile memory and neuromorphic computing applications, where their performance and reliability are strongly governed by defect states and carrier localization dynamics. In this work, we present a comparative investigation of GeAsSe and GeAsTe OTS devices to elucidate the impact of chalcogen chemistry on trap energy barriers, defect density, localization kinetics, and overall switching behavior. By combining characterizations with DC, ultrafast pulse, and double-pulse techniques across nanoseconds-to-seconds timescales, we show that material composition affects the localized states, which in turn set the characteristic energy barriers for both field-assisted and thermal-assisted delocalization, and control the leakage current, threshold voltage (V th) , and recovery dynamics. GeAsSe devices exhibit lower leakage, improved on/off selectivity, and lower recovery-induced Vth shift due to larger delocalization energy, while GeAsTe devices show lower switching field (F th) , and higher on-state current, leading to more field-driven and thermal-assisted delocalized defects. These insights establish a direct link between defect energetics and device performance, offering material-level guidelines for OTS optimization.engJournal article10.1109/ted.2026.3716191WOS:0018378510000011557-9646