Publications

Building artificial neurons and synapses is key to achieving the promise of energy efficiency and acceleration envisioned for brain-inspired information processing. Emulating the spiking behavior of biological neurons in physical materials requires precise programming of conductance nonlinearities. Strong correlated solid-state compounds exhibit pronounced nonlinearities such as metal–insulator transitions arising from dynamic electron–electron and electron–lattice interactions. However, a detailed understanding of atomic rearrangements and their implications for electronic structure remains obscure. In this work, we unveil discontinuous conductance switching from an antiferromagnetic insulator to a paramagnetic metal in ε-Cu_0.9V₂O₅. Distinctively, fashioning nonlinear dynamical oscillators from entire millimeter-sized crystals allows us to map the structural transformations underpinning conductance switching at an atomistic scale using single-crystal X-ray diffraction. We observe superlattice ordering of Cu ions between [V₄O₁₀] layers at low temperatures, a direct result of interchain Cu-ion migration and intrachain reorganization. The resulting charge and spin ordering along the vanadium oxide framework stabilizes an insulating state. Using X-ray absorption and emission spectroscopies, assigned with the aid of electronic structure calculations and measurements of partially and completely decuprated samples, we find that Cu 3d and V 3d orbitals are closely overlapped near the Fermi level. The filling and overlap of these states, specifically the narrowing/broadening of V 3d_xy states near the Fermi level, mediate conductance switching upon Cu-ion rearrangement. Understanding the mechanisms of conductance nonlinearities in terms of ion motion along specific trajectories can enable the atomistic design of neuromorphic active elements through strategies such as cointercalation and site-selective modification.

Vanadium oxides are widely tunable materials, with many thermodynamically stable phases suitable for applications spanning catalysis to neuromorphic computing. The stability of vanadium in a range of oxidation states enables mixed-valence polymorphs of kinetically accessible metastable materials. Low-temperature synthetic routes to, and the properties of, these metastable materials are poorly understood and may unlock new optoelectronic and magnetic functionalities for expanded applications. In this work, we demonstrate topochemical reduction of α-V₂O₅ to produce metastable vanadium oxide phases with tunable oxygen vacancies (>6%) and simultaneous substitutional tin incorporation (>3.5%). The chemistry is carried out at low temperature (65 °C) with solution-phase SnCl₂, where Sn²⁺ is oxidized to Sn⁴⁺ as V⁵⁺ sites are reduced to V⁴⁺ during oxygen vacancy formation. Despite high oxygen vacancy and tin concentrations, the transformations are topochemical in that the symmetry of the parent crystal remains intact, although the unit cell expands. Band structure calculations show that these vacancies contribute electrons to the lattice, whereas substitutional tin contributes holes, yielding a compensation doping effect and control over the electronic properties. The SnCl₂ redox chemistry is effective on both solution-processed V₂O₅ nanoparticle inks and mesoporous films cast from untreated inks, enabling versatile routes toward functional films with tunable optical and electronic properties. The electrical conductance rises concomitantly with the SnCl₂ concentration and treatment time, indicating a net increase in density of free electrons in the host lattice. This work provides a valuable demonstration of kinetic tailoring of electronic properties of vanadium–oxygen systems through top-down chemical manipulation from known thermodynamic phases.

We show that the deposition of the solid-state electrolyte LiPON onto films of V₂O₅ leads to their uniform lithiation of up to 2.2 Li per V₂O₅, without affecting the Li concentration in the LiPON and its ionic conductivity. Our results indicate that Li incorporation occurs during LiPON deposition, in contrast to earlier mechanisms proposed to explain postdeposition Li transfer between LiPON and LiCoO₂. We use our discovery to demonstrate symmetric thin film batteries with a capacity of >270 mAh/g, at a rate of 20C, and 1600 cycles with only 8.4% loss in capacity. We also show how autolithiation can simplify fabrication of Li iontronic transistors attractive for emerging neuromorphic computing applications. Our discovery that LiPON deposition results in autolithiation of the underlying insertion oxide has the potential to substantially simplify and enhance the fabrication process for thin film solid state Li ion batteries and emerging lithium iontronic neuromorphic computing devices.

Bimetallic transition metal complexes with site-specific redox properties offer a versatile platform for understanding electron polarization, intramolecular electron transfer processes, and customizing electronic and magnetic properties that might impact reactivity and catalyst design. Inspired by the dissymmetric nickel sites in the Acetyl CoA Synthase (ACS) Active Site, three new bimetallic Ni(N₂S₂)–Ni(S₂C₂R₂) complexes based on Ni(N₂S₂) metalloligand donor synthons, Ni_d, in mimicry of the nickel site distal to the redox-active iron sulfur cluster of ACS, and nickel dithiolene receiver units, designated as Ni_p, the nickel proximal to the 4Fe4S cluster, were combined to explore the influence of ligand environment on electronic structure and redox properties of each unit. The combination of synthons gave a matrix of three S-bridged dinickel complexes, characterized by X-ray crystallography, and appropriate spectroscopies. Computational modeling is connected to the electronic characteristics of the nickel donor and receiver units. This study demonstrated the intricacies of identifying sites of electrochemical redox processes, within multi-metallic systems containing non-innocent ligands.

Crystal structures are characterized by atomic bases within a primitive unit cell that repeats along a regular lattice throughout 3D space. The periodic and infinite nature of crystals poses unique challenges for geometric graph representation learning. Specifically, constructing graphs that effectively capture the complete geometric information of crystals and handle chiral crystals remains an unsolved and challenging problem. In this paper, we introduce a novel approach that utilizes the periodic patterns of unit cells to establish the lattice-based representation for each atom, enabling efficient and expressive graph representations of crystals. Furthermore, we propose ComFormer, a SE(3) transformer designed specifically for crystalline materials. ComFormer includes two variants; namely, iComFormer that employs invariant geometric descriptors of Euclidean distances and angles, and eComFormer that utilizes equivariant vector representations. Experimental results demonstrate the state-of-the-art predictive accuracy of ComFormer variants on various tasks across three widely-used crystal benchmarks. Our code is publicly available as part of the AIRS library (https://github.com/divelab/AIRS).

In this work, we developed three methods to map crystallographic variants of samples at the nanoscale by analyzing precession electron diffraction data using a high-temperature shape memory alloy and a VO₂ thin film on sapphire as the model systems. The three methods are (I) a user-selecting-reference pattern approach, (II) an algorithm-selecting-reference-pattern approach, and (III) a k-means approach. In the first two approaches, Euclidean distance, Cosine, and Structural Similarity (SSIM) algorithms were assessed for the diffraction pattern similarity quantification. We demonstrated that the Euclidean distance and SSIM methods outperform the Cosine algorithm. We further revealed that the random noise in the diffraction data can dramatically affect similarity quantification. Denoising processes could improve the crystallographic mapping quality. With the three methods mentioned above, we were able to map the crystallographic variants in different materials systems, thus enabling fast variant number quantification and clear variant distribution visualization. The advantages and disadvantages of each approach are also discussed. We expect these methods to benefit researchers who work on martensitic materials, in which the variant information is critical to understand their properties and functionalities.

In this work, we developed a method using precession electron diffraction data to map the residual elastic strain at the nano-scale. The diffraction pattern of each pixel was first collected and denoised. Template matching was then applied using the center spot as the mask to identify the positions of the diffraction disks. Statistics of distances between the selected diffracted disks enable the user to make an informed decision on the reference and to generate strain maps. Strain mapping on an unstrained single crystal sapphire shows the standard deviation of strain measurement is 0.5%. With this method, we were able to successfully measure and map the residual elastic strain in VO₂ on sapphire and martensite in a Ni_50.3Ti_29.7Hf₂₀ shape memory alloy. This approach does not require the user to select a “strain-free area” as a reference and can work on datasets even with the crystals oriented away from zone axes. This method is expected to provide a robust and more accessible alternative means of studying the residual strain of various material systems that complements the existing algorithms for strain mapping.