Neuromorphic computing inspired by mammalian intelligence aims to emulate the nonlinear dynamics of biological neurons and synapses to achieve fast, low-energy, and highly efficient information processing. Brain-inspired computing relies on the design and discovery of materials exhibiting nonlinear current–voltage profiles, frequently underpinned by electronic state transitions, to achieve spiking neurons and dynamically tunable synapses. A signature challenge in the design of artificial neurons is controlling the steepness of first-order transitions in active elements, as abrupt transitions are at risk of driving unstable voltage and temperature oscillations, which result in catastrophic device failure. A critical knowledge gap is the lack of structure–function correlations mapping the composition and atomistic structure of crystalline solids to nonlinear dynamical response characteristics. Here, we address the key question of how modification of atomistic structure correlates with alteration of neuron-like functionality. Constructing oscillator circuits from millimeter-scale single crystals enables high-resolution atomic structure solutions, which we use to demonstrate that the selective positioning of Pb cations modifies charge ordering along a one-dimensional CuxV2O5 framework even at low insertion stoichiometries, thereby providing an atom-precise design parameter for damping first-order transitions. We use temperature-variant X-ray diffraction and X-ray spectroscopy to elucidate the suppression of Cu-ion shuttling based on the precise positioning of Pb ions in seven-coordinated tunnel interstitial sites as the mechanistic basis for transition broadening, thus bridging a critical gap between statistical mechanics and quantum chemical descriptions of phase transitions. Such mechanistic understanding thus paves the way to site-selective modification strategies for modulating the sharpness of first-order transitions, with an exemplary demonstration here in tuning neuronal signal processing.
Publications
Knowledge gaps for neuromorphic ionic computing
Neuromorphic ionic computing, which uses principles similar to a human brain, represents a groundbreaking direction in computational technology, promising substantially improved energy efficiency compared with traditional silicon-based platforms. In a Review, Aluru et al. highlight essential gaps in knowledge spanning multiple domains such as materials science, device design, system integration, chemical compatibility, and biocompatibility that must be addressed. The authors emphasize the critical role of interdisciplinary collaboration in realizing the full promise of this emerging field. By advancing these areas, neuromorphic ionic systems could provide new possibilities for energy-efficient computing, with applications ranging from artificial intelligence to robotics and beyond. —Yury Suleymanov
Intrinsic Nonlinearity Modulation in Two-Dimensional (Cu,Ag)InP2S6 for Selectorless Nonvolatile Memory Array
March 12, 2026
Selectorless resistive random-access memory is essential for scaling high-density crossbar arrays, yet suppressing sneak path currents (SPCs) without external selector components remains a major challenge. In this work, we investigated a two-dimensional (2D) van der Waals (vdW) mixed cation crystal Cu0.5Ag0.5InP2S6 (CAIPS) as a switching layer and systematically compared its resistive switching with CuInP2S6 (CIPS) and AgInP2S6 (AIPS). The coexistence of Cu+ and Ag+ ions produces asymmetric out-of-plane diffusion barriers, as confirmed by first-principles density functional theory (DFT) calculations, leading to self-rectifying transport and the intrinsic suppression of leakage in arrays. CAIPS-based devices exhibit stable bipolar resistive switching, a high intrinsic nonlinearity factor (>10 under a V/3 read scheme), a large memory window (>9× at Vread = 0.1 V), and low variability (coefficient of variation down to 5.1%), surpassing the performance of both CuInP2S6 (CIPS) and AgInP2S6 (AIPS). These features, combined with low operational switching voltages, robust endurance, and built-in nonlinearity highlight CAIPS as a promising material for scalable selectorless memory arrays, with direct relevance to energy-efficient neuromorphic and edge-computing architectures.
Augmenting Molecular Graphs with Geometries via Machine Learning Interatomic Potentials
Augmenting Molecular Graphs with Geometries via Machine Learning Interatomic Potentials
February 22, 2026
Accurate molecular property predictions require 3D geometries, which are typically obtained using expensive methods such as density functional theory (DFT). Here, we attempt to obtain molecular geometries by relying solely on machine learning interatomic potential (MLIP) models. To this end, we first curate a large-scale molecular relaxation dataset comprising 3.5 million molecules and 300 million snapshots. Then MLIP pre-trained models are trained with supervised learning to predict energy and forces given 3D molecular structures. Once trained, we show that the pre-trained models can be used in different ways to obtain geometries either explicitly or implicitly. First, it can be used to obtain approximate low-energy 3D geometries via geometry optimization. While these geometries do not consistently reach DFT-level chemical accuracy or convergence, they can still improve downstream performance compared to non-relaxed structures. To mitigate potential biases and enhance downstream predictions, we introduce geometry fine-tuning based on the relaxed 3D geometries. Second, the pre-trained models can be directly fine-tuned for property prediction when ground truth 3D geometries are available. Our results demonstrate that MLIP pre-trained models trained on relaxation data can learn transferable molecular representations to improve downstream molecular property prediction and can provide practically valuable but approximate molecular geometries that benefit property predictions. Our code is publicly available at: https://github.com/divelab/AIRS/.
Modulating charge transport via 2 MeV He+ irradiation in VO2
Modulating charge transport via 2 MeV He+ irradiation in VO2
February 20, 2026
Alkali-Metal Interlocking of 2D V4O10 Sheets Defines Discretized Interlayer Shear Relationships
Alkali-Metal Interlocking of 2D V4O10 Sheets Defines Discretized Interlayer Shear Relationships
February 19, 2026
Low-dimensional materials manifest structural anisotropy, quantum confinement, and tightly bound excitonic states, which make them attractive building blocks that can be assembled within three-dimensional laterally stitched heterostructures, stacked van der Waals solids, and complex moiré superlattices. Ion intercalation in the galleries between layered materials provides a means of modifying interlayer separation and coupling, but it is also known to drive the shearing of the layers. In this article, we explore the distinct ligand coordination environments afforded by vanadyl oxygens of singular [V4O10] sheets and examine how the size, polarizability, and stoichiometry of Group I cations sandwiched between such layers determine the interlocking of the sheets in stacked structures. Based on the topochemical insertion of alkali-metal ions into the layered λ-V2O5, we identify seven types of guest ion coordination sites discretized into four distinct regimes of interlayer shear in units of half octahedral widths. The coordination preferences of intercalated cations govern how they interlock 2D [V4O10] sheets and engender specific shear conformations. We present evidence that static and dynamic disorder in guest ion arrangement modulate the magnetic structure of the intercalated compounds based on electrostatic polarization, localization of charge and spin density, and lattice distortion. The results illustrate the use of topochemical ion insertion to modulate stacking relationships and magnetic transition characteristics.
Low-cost calculation and analysis of 2D IR spectra of model diiron trinitrosyl complexes in the NO stretch region with vibrational perturbation theory
Two-dimensional infrared spectroscopy offers unique capabilities for probing vibrational coupling in complex metal–ligand systems. In this paper, we combine two-dimensional infrared spectroscopy with vibrational perturbation theory to investigate vibrational coupling in a diiron trinitrosyl complex across three stable redox states. Although these systems are challenging for electronic structure methods, we demonstrate that key features of experimental 2D IR spectra can be accurately reproduced using reduced-dimensional anharmonic calculations with a small harmonic frequency scaling. Analysis reveals that N–O stretching modes maintain high locality across all redox states, with coupling patterns that directly reflect variations in Fe–N bond strength. Using curvilinear coordinate analysis, we demonstrate these differences result from systematic changes in cubic anharmonic force constants rather than mode delocalization. Our results establish N–O stretches as sensitive probes of metal–ligand bonding strength, expanding the toolkit for studying biologically relevant nitrosyl complexes.
Magnetic and EPR Spectroscopic Studies of Thiolate Bridged Divalent Ni, Pd, and Pt Ions Capped with VO(N2S2) Metalloligands
January 21, 2026
Reactions of the metallodithiolate complex VO(bme-dach) (hereafter abbreviated as V, where bme-dach = N,N′-bis(2-mercaptoethyl)-1,4-diazacycloheptane) with [PdII(CH3CN)4](BF4)2 and [PtII(CH3CN)4](BF4)2 yield the V–M–V trimetallic compounds [VPdV](BF4)2 (2) and [VPtV](BF4)2 (3). Reaction of a similar metalloligand, VO(bme-daco) (hereafter abbreviated as V′ where bme-daco = N,N′-bis(2-mercaptoethyl)-1,5-diazacyclooctane) with [NiII(CH3CN)6](BF4)2 afforded the related salt [V′NiV′](BF4)2 (1). X-ray structural analyses revealed that cations in 1, 2, and 3 adopt a stairstep C2h structure consisting of two terminal VO(N2S2) moieties bridged via thiolate sulfur to the group 10 metal ions. Weak ferromagnetic superexchange coupling (J = 0.282 cm–1 for 1, 0.954 cm–1 for 2, and 1.372 cm–1 for 3) was observed between the two S = 1/2 VIV centers separated by distances in the range of 5.9–6.3 Å, with J values varying following the order Ni < Pd < Pt. Frozen-solution EPR spectra measured on the more soluble [BArF24]− (BArF24– = tetrakis((3,5-trifluoromethyl)phenyl)borate) analogues revealed that the [VPtV]2+ cation exhibits a 15-line hyperfine splitting of 225 MHz at g = 4 in parallel mode, confirming exchange coupling between the two 51V, I = 7/2 nuclei. Density-functional theory (DFT) calculations indicate an S = 1 ground state for 1–3. These results demonstrate that the choice of paramagnetic metallodithiolate ligand and diamagnetic bridge in such trimetallic species influences the sign and magnitude of magnetic interactions.
Correlation control of the Mott transition in LaTiO3/SrTiO3 heterostructures
The Mott metal-insulator transition arises from electron-electron interactions determined by the ratio of Coulomb to kinetic energy scales (U/t). While temperature, pressure, and doping can induce Mott transitions, direct control of U in solid-state systems remains largely unexplored, particularly in the inhomogeneous environments of emerging neuromorphic devices, where conductive filaments create complex gradients of local properties. Here we show that interface-induced screening can continuously tune the electron-electron interaction strength U to drive an isothermal Mott transition. Using thickness-graded LaTiO3/SrTiO3 heterostructures, we used diffraction and photoemission spectroscopy to reveal a continuous, isothermal quantum phase transition from a Fermi liquid quasiparticle with incoherent excitations to a Mott insulator with Hubbard bands. The primary determinant of the transition is the enhanced local screening environment, which directly influences the interaction strength U, driving the system metallic. This demonstrates that interface engineering of the local screening environment provides a promising approach to manipulate Mott physics through correlation control, beyond traditional bandwidth or filling approaches.
Interlayer Exciton Polarons in Mesoscopic V2O5 for Broadband Optoelectronic Synapses
Interlayer Exciton Polarons in Mesoscopic V2O5 for Broadband Optoelectronic Synapses
November 20, 2025
Persistent photoconductivity and optoelectronic synaptic behavior are demonstrated in solution-processed mesoscopic α-phase vanadium pentoxide (V2O5) thin films. First-principles simulations coupled with the two-site Holstein polaron hopping model show that vacancies at the terminal oxygen position lead to long recombination times because photoexcited electrons and holes reside on different layers separated by the van der Waals gap, forming a weakly coupled interlayer exciton polaron. Mid-gap polaronic states also significantly broaden the photoresponse of the films to span across visible and infrared wavelengths. By controlling the amplitude/intensity, duration, and/or number of optical pulses, tunable optoelectronic memory functions, such as short-term and long-term plasticity, are experimentally established in V2O5-based optoelectronic synapses. Device fabrication was extended to mechanically flexible ultrathin glass substrates. Flexible optoelectronic synapses maintained high performance after 150 bending cycles.