Dec 3, 2024
9:00am - 9:30am
Sheraton, Second Floor, Republic A
Asir Intisar Khan1,2,Xiangjin Wu2,Heungdong Kwon2,Kenneth Goodson2,H.S. Philip Wong2,Eric Pop2
University of California, Berkeley1,Stanford University2
Asir Intisar Khan1,2,Xiangjin Wu2,Heungdong Kwon2,Kenneth Goodson2,H.S. Philip Wong2,Eric Pop2
University of California, Berkeley1,Stanford University2
Today’s nanoelectronics systems are reaching the limits of energy and latency for numerous data-intensive applications. Among the existing non-volatile memory technologies, phase-change memory (PCM) holds promise for high-density storage [1]. However, this technology must achieve low-power and stable operation at nanoscale dimensions to be useful in heterogeneously integrated logic and memory. This talk will delve into our recent efforts on the atomic-scale interface and electro-thermal engineering of new phase-change materials for low-power and brain-inspired computing.<br/>Using a combination of phase-change heterostructures and nanocomposites, we have demonstrated sub-1 picojoule switching energy and sub-1 V switching voltage in nanoscale PCM devices [2,3], which are promising for on-chip logic and memory integration. These devices also show low resistance drift, good endurance, fast switching (40 ns), and a large on-off ratio (>100) [1, 2]. The energy-efficient switching is enabled by strong heat confinement within the superlattice material interfaces [3,4]. The heat confinement is further amplified in low-thermal conductivity flexible polyimide substrates, a new paradigm in low-power memory for flexible electronics [5]. We also demonstrate that the material and thermal properties of the heterostructures are controlled by the interface density, which ultimately plays a crucial role in regulating the device performance [6,7]. Additionally, phase-change nanocomposites facilitate the fast-switching speed and gradual bi-directional conductance change in PCM devices, favorable for low-power spiking neural networks [8,9]. These results demonstrate the promise of novel heterostructures and their interface-driven transport modulation for low-power and high-density neuro-inspired memory.<br/><b>References:</b><br/>[1] S. Raoux, E. Pop <i>et al</i>., MRS Bulletin 39, 703-710 (2014). [2] X. Wu, A.I. Khan, E Pop <i>et al</i>., Nature Commun. 15, 13 (2024). [3] A.I. Khan, E. Pop <i>et al</i>., IEEE Electron Dev. Lett. 43, 204-207 (2022). [4] H. Kwon, A.I. Khan, E. Pop, K.E. Goodson <i>et al</i>., Nano Letters 21, 5984–5990 (2021) [5] A.I. Khan, E. Pop <i>et al</i>., Science 373, 1243-1247 (2021). [6] A.I. Khan, E. Pop <i>et al</i>., Nano Letters 22, 6285–6291 (2022). [7] J. Zhao, A.I. Khan, E. Pop, L. Allen <i>et al</i>., Nano Letters 23, 4587-4594 (2023). [8] A.I. Khan, E. Pop <i>et al.,</i> Adv. Materials 35, 2300107 (2023). [9] S.B. Hamid, A.I. Khan, E. Pop <i>et al</i>., IEEE DRC (2024).