10:30 AM - EN15.07.01
Harnessing Anharmonicity and Electron Correlations for Energy Effecient Computing
Oak Ridge National Laboratory1
Metal oxide-based Resistive Random-Access Memory (RRAM) exhibits multiple resistance states, arising from the activation/deactivation of a conductive filament (CF) formed by oxygen vacancies inside a switching layer – due to an underlying metal-insulator transition (MIT). In a crossbar device architecture, you have a highly dense information-storage&computing system, that you want to be reliable, fast switching and utilizing low-power. This can enable emulating ‘brain like’ neuromorphic computing with co-located memory and compute. Similarly, ferroelectric materials – such as Hafnia – are promising candidates for synaptic weight elements in neural network hardware because of their nonvolatile multilevel memory effects. But conventional RRAM materials require high forming potentials and show high variability (device-to-device or cycle-to-cycle) and are plagued by less reliability as well as the voltage-time dilemma, similar to ferroelectric synapses. To address these challenges, we are working to answer the following open questions: what material characteristics we need when choosing a memristor material? What factors triggers a state-change (e.g. MIT, or ferroelectric-switching) in these materials? What determines the dynamics of the switching mechanism?
Using a combination of high-throughput phase-field and machine-learning methods  we discovered that harnessing electron-electron correlations in binary oxides can be advantageous for improved performance of RRAM devices. Using a combination of various correlated electronic structure methods, we further uncovered the underlying factors that control the MIT in non-stoichiometric correlated binary oxides – such as VO2 [2,3]. We subsequently demonstrated how many of the correlated perovskite metals that undergo MIT are negative charge-transfer metals, with the magnitude of ligand-hole being the key to controlling MIT. These works provides a fundamental understanding of the role of electron correlations in resistive switching materials. For ferroelectric-based synapses, we explored the recently discovered layered-thiophosphate family of materials . We discovered  presence of strong anharmonic coupling in these materials between the ferroelectric polar-mode and a Raman active symmetric-mode, even down to the single-layer limit, that could allow optical or mechanical control of ferroelectric switching, potentially alleviating the voltage-time dilemma in conventional ferroelectrics.
” High-throughput phase-field simulations and machine learning of resistive switching in resistive random-access memory”, npj Computational Materials volume 6, Article number: 198 (2020), Kena Zhang, Jianjun Wang, Yuhui Huang, Long-Qing Chen, P. Ganesh* & Ye Cao.
” Doping a bad metal: Origin of suppression of the metal-insulator transition in nonstoichiometric VO2”, Phys. Rev. B 101, 155129, (2020), P. Ganesh*, Frank Lechermann, Ilkka Kylänpää, Jaron T. Krogel, Paul R. C. Kent, and Olle Heinonen
"Metal-insulator transition tuned by oxygen vacancy migration across TiO2/VO2 interface", Scientific Reports, 10, 1854 (2020), Qiyang Lu, Changhee Sohn, Guoxiang Hu, XiangGao, Matthew F. Chisholm, Ilkka Kylänpää, JaronT. Krogel, Paul R. C. Kent, Olle Heinonen, P.Ganesh* and Ho Nyung Lee
”Origin of Metal-Insulator Transitions in Correlated Perovskite Metals”, arXiv:2103.09809, M. Chandler Bennett, Guoxiang Hu, Guangming Wang, Olle Heinonen, Paul R. C. Kent, Jaron T. Krogel, P. Ganesh*.
”Tunable quadruple-well ferroelectric van der Waals crystals”, Nature Materials, 19, 43 (2020), John A. Brehm, Sabine M. Neumayer, Lei Tao, Andrew O’Hara, Marius Chyasnavichus, Michael A. Susner, Michael A. McGuire, Sergei V. Kalinin, Stephen Jesse, P. Ganesh*, Sokrates T. Pantelides, Petro Maksymovych and Nina Balke
”Origin and stabilization of ferrielectricity in CuInP2Se6”, arXiv:2106.08783, Nikhil Sivadas, Peter Doak, P. Ganesh*