Research motivation and thrust

Human development and improved standards of living have historically coincided with increases in energy consumption. As billions of humans enter the global middle class, the environmental consequences of fossil energy use place immense strains of the earth’s ecosystem. It is necessary to decouple development with carbon emissions through energy efficiency and renewable energy. Electrochemical systems play crucial roles towards achieving these goals.

The Li+ group aims to advance fundamental understanding and develop technologies in electrochemical materials and devices. We study the underlying physical principles behind such materials, devices, and systems, and exploit them into developing new technologies towards energy solutions. We specifically investigate transition metal oxides with point defects. Point defects, such as lithium interstitials and oxygen vacancies, act as electronic dopants and enable these materials to conduct both electrons and ions, or “mixed conduction.” Mixed conduction combined with electrochemistry enables the simultaneous addition or removal of electron and ions using current and voltage, allowing us to add or remove ~1022 cm-3 dopants dynamically under device operating conditions. This process is known as electrochemical ion insertion, and is crucial for a number of technologies including batteries, electrochromics, solid fuel cells, electro-catalysts, etc. The two main thrusts of this group is to develop materials and devices for batteries and for in-memory, low-energy computing. 

Reference: Y. Li and W. C. Chueh, “Electrochemical and Chemical Insertion for Energy Transformation and Switching.” Annual Reviews for Materials Research. 48:137-165 (2018)

Electrochemical redox memory

Machine learning and artificial intelligence require large data centers, consume extraordinary amounts of energy, will soon be major contributors to climate change. How can we make AI more energy efficient?

The energy intensity of of data-heavy processes like AI result from the need to move information from logic to processor elements. Neural networks can be 1000x more energy efficient by computing on a memory crossbar using physics (Ohm’s & Kirchoff’s Laws) to perform matrix multiplications, the most ubiquitous process.

Neuromorphic comptuting containing crossbar arrays of non-volatile memory elements enable orders of magnitude reduction in the energy consumption of data-intensive processes like machine learning.

For the past several decades, most research in the nonvolatile memory element have focused onto filament-based information storage elements like memristors and phase-change memory. In such devices, information is stored in discrete numbers of point defects within the filament. Due to kinetic theory, individual atoms follow random, stochastic motion, leading to memory devices with irreproducible and stochastic switching.

Our approach is to use the average behavior of all point defects in the bulk. With over 106 lattice sites in a small (30-nm)3 volume, the average behavior of all point defects is statistical. Harnessing this statistical behavior results in predictable and deterministic analogue switching.


Filamentary and bulk resistive memory. (a) Conventional filamentary memory devices store analogue information states in atomic-sized filaments. (b) Such devices switch stochastically due to the discrete nature of atoms in the filament and kinetic theory. (c) Bulk memory instead stores information using the point defect concentration, a continuous variable. (d) Bulk memory yields deterministic switching.

We seek to both understand the fundamental ion insertion and migration processes within these redox memory cells, as well as design improved devices. We envision an analogue “neural-core” hardware accelerator than can be placed in all chips to accelerate machine learning in an energy-efficient manner.


Past work:

Y. Li, E. J. Fuller, J. D. Sugar, S. Yoo, D. S. Ashby, C. H. Bennett, R. D. Horton, M. S. Bartsch, M. J. Marinella, W. D. Lu, A. A. Talin. “Filament-free bulk resistive memory enables deterministic analogue switching.” Advanced Materials, Accepted (2020)

E. J. Fuller, S. T. Keene, A. Melianas, Z. Wang, S. Agarwal, Y. Li, Y. Tuchman, C. D. James, M. J. Marinella, J. J. Yang, A.Salleo, A. A. Talin. ”Parallel programming of an ionic floating-gate memory array for scalable neuromorphic computing.” Science. 364, 570-574 (2019)

Heterogeneity in Li-ion battery particles

Batteries typically comprise of micron-sized particles in porous electrodes with additives, electrolyte, and binder. While every particle has the same functionality to store energy, they do not necessarily have the same properties. An unknown question is whether or not there are certain particles that have much better electrochemical performance than other particles. If we can increase the population of “good” particles while decreasing the population of “bad” particles, this can lead to an overall improvement in the performance of the battery electrode. 

Our goal is to measure the electrochemical performance of individual battery particles, and understand how they are different from one another. We’ve been able to conduct charge-discharge profiles of individual battery particles.

Past work:

Y. Li, H. Chen, K. Lim, H. D. Deng, J. Lim, D. Fraggedakis, P. M. Attia, S. C. Lee, N. Jin, J. Moskon, Z. Guan, W. E. Gent, J. Hong, Y. S. Yu, M. Gaberscek, M. S. Islam, M. Z. Bazant, W. C. Chueh. ”Fluid-enhanced Surface Diffusion Controls Intra-Particle Phase Transformations.” Nature Materials. 17, 915-922 (2018)

J. Lim,* Y. Li,* D. H. Alsem, H. So, S. C. Lee, P. Bai, D. A. Cogswell, X. Liu, N. Jin, Y. Yu, N. J. Salmon, D. A. Shapiro, M. Z. Bazant, T. Tyliszczak, W. C. Chueh. “Origin and hysteresis of lithium compositional spatiodynamics in battery primary particles.” Science, 353, 566-71 (2016)