| PhD Seminar

Name of the Speaker: Mr. Sakthikumaran (EE18D040)
Guide: Dr. Anbarasu Manivannan
Venue: ESB-210B (Conference Hall)
Date/Time: 30th October 2025 (Thursday), 3.00pm
Title: Unravelling the Role of Tellurium Network Connectivity in Sub-nanosecond Threshold Switching Dynamics of Amorphous GexTe100-x Devices

Abstract :

Speaker Bio:

Contemporary Artificial Intelligence systems are rapidly advancing towards Artificial General Intelligence (AGI) and the eventual emergence of Artificial Superintelligence (ASI). This trajectory places unprecedented demands on memory technologies, requiring faster speeds, higher densities, and improved energy efficiency. Conventional memory hierarchies, however, are fundamentally constrained by latency and power bottlenecks, especially under the data-intensive workloads of modern AI architectures. Emerging paradigms such as memristor-based in-memory computing offer promising alternatives. Yet, their performance hinges critically on the behaviour of selector devices, particularly the threshold switching (TS) dynamics. Our research addresses this challenge by investigating how amorphous network connectivity modulates the TS behaviour in the GexTe100-x alloy system.

Using time-resolved electrical switching dynamics of GeTe, GeTe₄, and GeTe₆ devices, we demonstrate that network rigidity tuned by tellurium concentration directly influences switching delay, threshold voltage, and energy dissipation. GeTe, with its rigid amorphous network, exhibits elevated threshold fields and longer delays. In contrast, GeTe₆, with a floppy network rich in Te-Te bonds, achieves sub-nanosecond switching with minimal overvoltage. GeTe₄, occupying an isostatic regime, shows a hybrid response, making it a strong candidate for both memristive and Selector-Only Memory (SOM) applications.

Our findings establish atomic connectivity as a tunable design parameter, not merely a passive structural feature. By correlating steady-state vibrational mode distributions (via Raman spectroscopy) with transient I–V characteristics, we reveal the pivotal role of local phonon dynamics and network topology in threshold switching. This work paves the way for faster, yet low-power chalcogenide-based memory and selector technologies.