Alexander Groetsch, KTH Royal Institute of Technology

Tuning ZnO-Coated Laser-Printed Nano-Architected Mechanical Metamaterials

Alexander Groetsch presents a parametric study of bending-dominated nanoarchitected metamaterials whose mechanical response is tuned through both relative density and conformal zinc oxide (ZnO) coating thickness. Polymer Kelvin foam scaffolds with 6×6×12 unit cells were fabricated by two-photon laser printing on a Nanoscribe Photonic Professional GT, achieving feature sizes of approximately 200 nm in the lateral plane and 500 nm vertically. Three distinct architectures were generated by tuning contour and slicing distance during fabrication while holding laser power constant, and atomic layer deposition (ALD) using diethylzinc and water at 50°C produced conformal ZnO coatings of 20 nm-,40 nm-, 80 nm-, and 160-nm thickness without thermal damage to the polymeric scaffolds. Approximately 270 ex situ micro-compression tests across coating thicknesses, architectures, and three strain rates (10^-3, 10^-2, 10^-2 s^-1) revealed that thicker ZnO coatings increase yield strength and stiffness but produce more localized failure, with denser architectures showing a smaller relative gain from coating because of their lower surface-to-volume ratio. Layer-by-layer failure in some configurations contrasted with stable plastic plateaus in others, and nano-computed x-ray tomography with Paganin phase retrieval related the 3D printed geometry back to the measured mechanical response.

Jeffrey Lipton, Northeastern University

Fabrication-Directed Entanglement for Additive Manufacturing of Multifunctional Foams

Written by Corrisa Heyes

Jeffrey Lipton presents fabrication-directed entanglement (FDE) via viscous thread printing (VTP), an additive manufacturing strategy that exploits fluid-mechanical coiling instabilities during extrusion to produce continuous stochastic entangled networks rather than deterministic lattices. The viscous thread instability behaves dimensionally: a one-dimensional thread produces straight extrusion, two-dimensional coiling produces alternating loops, and three-dimensional self-stabilization produces non-woven textiles and foams whose path height self-regulates. Real-time modulation of deposition parameters allows local control of density, connectivity, and anisotropy on standard desktop fabrication platforms. In polymer foams, integrating topology optimization with VTP produces mechanical metamaterials exhibiting directional stiffness, tunable Poisson’s ratios including auxetic response, and chiral deformation modes that translate compression into shear force, all from process-encoded mesoscale structure without multi-material interfaces. Lipton’s group also extended FDE to metallic foams using copper, bronze, and stainless steel filled fused filament feedstocks followed by debinding and sintering, producing both discrete and continuous porosity gradients that can be machined into final parts. Mechanical, thermal, and electrical characterization across these systems revealed enhanced crack resistance relative to deterministic architectures, a multifunctional property envelope arising from stochastic entanglement rather than explicit geometric design.

Achieving Ultra-Fast Locomotion in Insect-Scale Soft Robots via iCVD-Based Nanoadhesive Technology

Written by Corrisa Heyes

Minki Kim presents a nanoadhesive technology based on initiated chemical vapor deposition (iCVD) that resolves an energy-transmission bottleneck in piezoelectric insect-scale soft robots. In multilayer piezoelectric actuators, the adhesive between active and passive layers must simultaneously be thin and stiff for efficient strain transfer (Crawley shear-lag model) and thick and compliant for reliable wet-out (Dahlquist criterion); commercial pressure-sensitive adhesives compromise both, with viscoelastic dissipation at the high resonant frequencies of miniaturized robots compounding the loss. Kim’s group designed a copolymer of glycidyl methacrylate and 2-hydroxyethyl acrylate (pGH) deposited via solvent-free, room-temperature iCVD with nanometer-level thickness control. A hydrogen bonding network between epoxy and hydroxyl groups produces low loss modulus above 100 Hz, where commercial adhesives dissipate significantly. Bending actuators with the pGH layer reached 3-5 times the tip velocity of equivalents using commercial ultra-thin tapes. An insect-scale robot oscillating at 383 Hz with 164 - 180 degree body deformation achieved 1.1 m/s (45 body-lengths/second), matching Periplaneta americana. The robot demonstrated high functionality: pushed loads exceeding 40x body weight, recovered from impacts via mid-air flipping, jumped 105 mm gaps, and transitioned from ground to water-surface sliding.

Additive Manufacturing of Responsive Phononic Subsurfaces for Drag Reduction

Written by Corrisa Heyes

Abigail Juhl presents architected phononic subsurfaces (PSubs) fabricated by additive manufacturing that passively delay laminar-to-turbulent flow transition on aircraft surfaces, reducing drag without active actuation hardware. The design exploits fluid-structure coupling: a Tollmien-Schlichting wave approaching transition forces a phononic crystal embedded beneath the aircraft skin, and the crystal’s engineered dynamic response feeds back to attenuate the instability. A naïve phononic crystal sized to address the relevant frequency range would be too large to fit inside a wing, so the team coils the unit cell by locking rotational degrees of freedom at primitive-cell edges and applying full beam-axis rotations between them, producing a compact subsurface architecture compatible with AM fabrication. Coupled CFD using AFRL’s high-order implicit Navier-Stokes solver with large-eddy simulation, exchanging loads and displacements between fluid and solid domains, predicted attenuation depending on the phase relationship between the TS wave and the subsurface displacement. Wind tunnel experiments at Florida State University FCAAP confirmed CFD trends, though rigid control surfaces unexpectedly acted as resonators. A multi-input multi-output PSub design superposes responses from multiple coupled units to broaden the attenuation envelope across the flow conditions of interest.

Our congratulations go to the 2026 MRS Spring Meeting Chairs Meeting Chairs Vicky Doan-Nguyen Trigg (The Aerospace Corporation), Taeghwan Hyeon (Seoul National University), Naoya Shibata (The University of Tokyo), and Benjamin Tee (National University of Singapore) for putting together an excellent technical program along with various special events. MRS would also like to thank all the Symposium Organizers and Session Chairs for their part in the success of this Meeting. A thank you goes to Symposium Support, and to the sponsors of the Meeting and of the special events and activities, and to the Exhibitors whose commitment and enthusiasm made the Materials Science Exhibit a success.

Contributors to news on the 2026 MRS Spring Meeting & Exhibit include Meeting Scene reporters Sophia Chen, Corrisa Heyes, Soumyajyoti Mondal, Matthew Nakamura, Yuying Ning, Kendra Redmond, Abhinanda Sengupta, and Melody Yiyuan Zhang; reporters for special events Anthony Salazar and Solbin Yang; and graphic artist Stephanie Gabborin; with newsletter production by Jason Zimmerman.

Thank you for subscribing to the MRS Meeting Scene newsletters. We hope you enjoyed reading them and continue your subscription as we launch into the 2026 MRS Fall Meeting & Exhibit. The conversation already started at #F26MRS! We welcome your comments and feedback.

L. Cate Brinson, FMRS Duke University
From Data to Discovery: The AI-Enabled Future of Materials Science

Written by Sophia Chen

In the last few years, artificial intelligence tools have rapidly changed materials research workflows to accelerate simulations and discovery. Companies continue to roll out new tools quickly, such as agentic AI. During her Symposium X presentation, L. Cate Brinson of Duke University discussed ways her laboratory has incorporated these tools in a talk called “From Data to Discovery: The AI-Enabled Future of Materials Science.”

“We have to learn to move as these models move and to work with them,” she said.

Brinson studies AI in the context of her overall research goals of merging experiments, computation, and data science to discover and design complex hierarchical materials. One example her research group recently worked on was a metamaterial that has a gradient stiffness, but a constant acoustic impedance. Such a metamaterial has biomedical applications, where it can be used for ultrasound imaging while interfacing with soft and sensitive human tissue.

A key step in developing AI tools for materials research is to make existing data comprehensible to them. Brinson’s group has worked on so-called “data curation,” even before the current generative AI models took off. In the past, the group found that researchers were unwilling to take the time to submit their data. “People are happy to give you their files, but they don’t want to do all the annotation and put things in the right boxes,” she said.

From her group’s research, agentic AI appears to be helpful for doing this type of annotation and curation, Brinson said. “It could take up to six hours for a human expert to fully curate 10 samples from a paper,” she said. An automated pipeline that includes agentic AI “can do thousands of samples from hundreds of papers in 10 minutes.”

However, one issue among researchers is whether to trust the output, as one attendee pointed out in the questions and answer session. “It’s never going to be 100%,” she said. “You have to be willing to accept that there are going to be some errors in the data that’s there, but presumably they will be found over time.”

Currently, many AI models operate as “black boxes,” where it is difficult or impossible for a human to understand the process by which it arrives at an answer. An alternative to this black box paradigm is “interpretable” AI, where the model formulates its response in terms of physical concepts. Brinson thinks that interpretability is important for materials research, although not all experts agree, she said. Sometimes it’s impossible to create an interpretable model. For now, Brinson calls for a “blended approach,” she said. “I think it’s important to find interpretability when you can.”

Attendees also asked Brinson about how AI is affecting the culture of materials research and education. Brinson said that a famous coder had expressed that his coding skills had degraded because he relied on agents to do the work. “I think that’s what we need to watch out for, that we don’t get lazy and just accept everything from the AI that we use,” said Brinson.

AI tools in materials research offer a “huge opportunity” right now, she said, via the US Department of Energy’s Genesis Mission. “My hope is that we can have the agentic structures built in such a way that they can help us generate hypotheses, but that there’s always a human in the loop […] to redirect and to find that interpretability and to put constraints on the system and send it back to the starting line if it’s going off in the wrong direction,” said Brinson.

Million-Atom Simulation of the Set Process in Phase Change Memories at the Real Device Scale

Written by Melody Yiyuan Zhang

Omar Abou El Kheir presented a major advance in atomistic simulation of phase-change memory materials by extending machine-learned interatomic potentials to the million-atom scale. Using a neural network potential with van der Waals corrections for the prototypical compound Ge₂Sb₂Te₅ (GST), the research group constructed a realistic three-million-atom model that mimics crystallization in an operating memory device. The simulation captured heat diffusion from an upper electrode thermostat into a semi-cylindrical amorphous hot spot and followed the subsequent recrystallization dynamics over nanosecond timescales. Results showed that nearly 80% of atoms recrystallized within 1.55 ns, with the overwhelming majority arising from crystal growth at the crystalline–amorphous interface rather than from homogeneous nucleation within the amorphous dome. While some coarsening and isolated nucleation events were observed, interface-driven growth clearly dominated the active region. Overall, the work demonstrates how machine learning potentials now enable direct visualization of crystallization kinetics, bridging the long-standing gap between density functional theory accuracy and realistic device length and time scales.

Xia Cao, Pacific Northwest National Laboratory

Floatable Protective Layer and Stable Electrolyte Design for Enhanced Lithium Utilization in Anode-Free Lithium Batteries

Anode-free lithium batteries offer a pathway toward higher energy density, but their practical deployment is limited by poor lithium utilization arising from continuous solid-electrolyte interphase (SEI) breakdown and dead lithium formation. Xia Cao presented an integrated strategy combining electrolyte design with a floatable protective layer (FPL) to stabilize lithium deposition. The approach introduces a surfactant-assisted, weakly adhered protective layer on the copper current collector. Unlike conventional coatings, the FPL dynamically “floats” during cycling: it lifts during lithium plating and settles during stripping. This behavior enables lithium to deposit beneath the layer, effectively isolating freshly formed lithium from direct electrolyte exposure. As shown in cross-sectional microscopy, lithium growth occurs uniformly underneath the layer, suppressing parasitic reactions. Complementary electrolyte optimization, particularly using ether-based solvents and localized high-concentration formulations, further enhances stability, achieving ~71% capacity retention after 100 cycles. The combined system reduces SEI accumulation and improves coulombic efficiency to ~99.6%. This work demonstrates that decoupling lithium from electrolyte exposure through dynamic interfacial design provides a mechanistic route to improving lithium utilization in anode-free systems.

Robert Kuphal, Michigan State University

Enabling High-Energy Hybrid Lithium-Ion/Lithium-Metal Batteries

Improving both fast-charging capability and energy density in lithium-ion batteries requires rethinking how graphite anodes operate under aggressive cycling conditions. In this presentation, Robert Kuphal discussed a hybrid lithium-ion/lithium-metal strategy in which reversible lithium plating on graphite enables operation at substantially lower negative to positive capacity (N/P) ratios while maintaining stable cycling behavior. Rather than treating lithium plating solely as a failure mechanism, the work reframed controlled lithium-metal deposition as a pathway toward higher practical energy density. The study focused heavily on electrolyte engineering to overcome kinetic limitations and stabilize the graphite interface. Conventional ethylene carbonate and dimethyl carbonate (EC/DMC) electrolytes exhibited poor lithium reversibility, whereas high-volume propylene carbonate and fluoroethylene carbonate (PC/FEC)-based electrolytes, particularly with lithium bis(fluorosulfonyl)imide (LiFSI) salt chemistry, enabled significantly improved lithium plating and stripping behavior. Electrochemical analysis using incremental capacity analysis plot showed reduced graphite phase-transition signatures and minimized exfoliation behavior in the PC/FEC system. Mechanistic characterization revealed that LiFSI decomposition drives the formation of a lithium oxide (Li₂O)-rich solid-electrolyte interphase (SEI), which appears critical for improving interfacial stability and lithium reversibility. The work isolated lithium plating behavior through dedicated reversibility protocols and demonstrated that stable SEI formation governs long-term cycling performance. Importantly, the electrolyte maintained stable cycling even at an aggressive N/P ratio of 0.7, achieving more than 1000 cycles while supporting fast-charging operation and suppressing over-lithiation effects. The study further demonstrated compatibility with practical cell configurations, including single-layer pouch cells and high-voltage cycling conditions. Overall, the presentation highlighted how electrolyte-controlled SEI chemistry can fundamentally reshape graphite behavior, enabling hybrid lithium-ion/lithium-metal operation without sacrificing manufacturability or cycling stability.

Taylor D. Sparks, The University of Utah

KnowMat2—An Open-Source Agentic Framework for Knowledge-Driven Materials Discovery

Taylor D. Sparks introduced KnowMat2, an open-source agentic framework designed to accelerate materials discovery by transforming unstructured scientific literature into datasets using AI tools. The system targets four major categories of materials information, including entities, process conditions, structural characteristics, and materials properties, and is capable of extracting data not only from text, but also from figures and tables. While baseline large language models can capture general trends, they often suffer from hallucination and poor reliability in detailed value extraction. To address this, KnowMat2 employs a modular multi-agent workflow consisting of reasoning, action, observation, validation, and managerial oversight. A schema-based extraction strategy, combined with document parsing and dedicated validation, manager, and flagging agents, allows the framework to detect inconsistencies and reduce overcorrection during iterative refinement. Benchmarking with ROUGE and BERT-based metrics showed clear performance improvements over vanilla LLM extraction baselines. At an estimated cost of only $0.70 per paper, KnowMat2 demonstrates a scalable path toward autonomous synthesis–structure–property knowledge mining and self-driving materials design.

Wenhao Sun, University of Michigan

The Synthesis Bounty Board - A Community Challenge to Experimentally Realize Computationally-Predicted Materials

Wenhao Sun addressed one of the central bottlenecks in modern computational materials discovery: although high-throughput simulations and generative AI can predict vast numbers of promising compounds, experimental synthesis remains difficult and poorly standardized. To bridge this gap, Sun proposed the “Synthesis Bounty Board,” an annual community-wide challenge aimed at experimentally realizing computationally predicted materials with exceptional target properties. Inspired by scientific benchmark competitions, the framework establishes objective synthesis metrics, independent adjudication, and public reporting of both successful and failed attempts. Several examples highlighted the complexity of predicting synthesizability, including solid-state synthesis pathways, competing phases in solution synthesis, and the role of thermodynamic versus kinetic control in suppressing unwanted by-products. Robotic ceramic synthesis and closed-loop make-and-predict workflows were presented as emerging tools for systematically exploring reaction combinations and optimal process conditions. By creating a shared benchmark that connects computational prediction, synthesis science, and public datasets, the Synthesis Bounty Board offers a promising route toward more reliable validation of AI-predicted materials and accelerated autonomous materials discovery.

Amir Gildor, Technion – Israel Institute of Technology

Harnessing VO₂ Phase Transition for Automatic Gain Control in Transimpedance Amplifiers

Amir Gildor of the Technion – Israel Institute of Technology presented a study on using the phase transition in VO₂ to enable automatic gain control in transimpedance amplifiers. The research group showed that conventional amplifiers with fixed gain can saturate under high input currents, leading to slow recovery. To address this, the group introduced a VO₂-based switching element that exploits its reversible insulator–metal transition. The device was first characterized independently, where its switching behavior and short-term memory effects were examined under electrical pulses. The results indicated that the switching dynamics depend strongly on operating conditions, particularly the applied voltage and temperature relative to the transition point. When integrated into the amplifier circuit, the VO₂ device enabled adaptive gain behavior, allowing the system to respond dynamically to varying input signals. In addition, under constant bias, the device exhibited self-sustained oscillations linked to its thermal switching dynamics. The study shows how phase-change behavior in VO₂ can be used to introduce compact and energy-efficient functionality in electronic circuits.

Chihwan An, Seoul National University

Ultrahigh Remnant Polarization in Ferroelectric Hf_xZr_1-xO₂ Thin Films Grown by Atomic Layer Epitaxy

Chihwan An of Seoul National University presented a study on achieving enhanced remnant polarization in epitaxial Hf_xZr_1-xO₂ thin films grown by atomic layer epitaxy on yttria-stabilized zirconia substrates. Hafnia-based ferroelectrics typically show a limited polarization range, which constrains device operation. In this work, epitaxial films with controlled composition were grown and structurally characterized, confirming epitaxial growth across the studied compositions. Electrical measurements revealed a substantially higher remnant polarization compared to commonly reported values. The analysis suggests that the enhancement is linked to changes in oxygen-sublattice displacement enabled by epitaxial constraints, pointing to a polarization mechanism that differs from the conventional picture in polycrystalline films. The results indicate that combining epitaxial growth with compositional tuning provides a pathway to modify and extend the ferroelectric response in hafnia-based systems.

John F. Hardy, Northern Arizona University

Architecture Enabled Low-Power Switching in WO_x-Based RRAM Devices

John F. Hardy of Northern Arizona University presented a study on reducing switching power in WO_x-based resistive memory devices by modifying the active-layer architecture. He compared conventional thin-film WO_x layers with porous helical structures fabricated using a glancing angle deposition technique. At higher operating currents, both types of devices showed similar switching characteristics. However, at lower operating currents, the difference became clear. The porous helical structures continued to show stable and reproducible switching, while the thin-film devices lost reliability. The helical architecture enabled lower switching voltages and currents, along with improved separation between high and low resistance states due to reduced leakage. It was also noted that these improvements did not depend strongly on the film thickness, indicating that the geometry of the active layer plays the dominant role. The talk highlighted how structural design can be used to achieve low-power operation in resistive memory devices.

In a trio of short talks, three MRS award recipients share highlights and opportunities from their materials research and outreach endeavors.

Miaofang Chi, a Corporate Fellow at Oak Ridge National Laboratory, received the Innovation in Materials Characterization Award. With modern scanning transmission electron microscopy (STEM), we can image materials at the atomic scale and probe local chemistry and electronic structures, Chi said. New techniques such as monochromated electron energy loss spectroscopy are providing information on bandgaps, phonon vibrations, and other quasiparticles, revealing key insights for energy and quantum materials, she told attendees.

Still, using advanced STEM techniques to study batteries at extremely cold temperatures or access key phenomena in quantum materials has been challenging. At low temperatures, mechanical drift and thermal fluxes create significant stage instabilities, Chi said. But this is changing.

Recent stability improvements mean we can now perform atomic resolution imaging and spectroscopy at low temperatures, she said. In particular, developments in liquid helium cooling stages are opening “an entire new experimental region” for quantum materials, Chi told attendees.

Combining the new techniques with stable cryo stages “will allow us to directly probe charge lattice, quasiparticles, and spin-orbit coupling at the temperature where they actually emerge,” she said, noting that it’s “a fundamental new way of accessing those hidden states in quantum materials.”

Sihong Wang, a molecular engineering professor at the University of Chicago and recipient of the Outstanding Early-Career Investigator Award, is working on high-performance bioelectronics. Today’s wearable devices are limited in scope, he said, while implantable devices have short functional lifetimes. His approach to long-lasting bioelectronic interfaces is based on biomimetic semiconducting polymers.

Brains are highly curvilinear and hearts constantly deform, Wang told attendees. To be truly accepted by the body, bioelectronics need stretchability, adhesion, and immune system compatibility. His group designs stretchable polymer semiconductors and devices with these properties.

The body’s immune system response is “arguably the most widely existing challenge for any type of implantable materials and devices,” Wang said. Within a year or two, scar tissue usually hinders signals from reaching implanted devices. This is driven by chemical and physical properties, motivating Wang’s research team to optimize the polymers for immune-compatible properties. They’ve made progress by, for example, developing an aqueous processing method to create semiconducting polymer-based hydrogels.

Wang’s group is also taking a broader view, working to integrate soft neuromorphic computing with its biosensing devices and utilizing stretchable optoelectronics for displays and biostimulation.

Jerrold A. Floro, a materials science and engineering professor at the University of Virginia, received the MRS Impact Award for contributions to the Society’s public outreach efforts. Floro highlighted three projects from the first decade of the MRS Public Outreach Committee, for which he was the inaugural chair in 2005.

Strange Matter, a touring museum exhibit, established MRS’s credibility in public outreach, Floro said. Designed in collaboration with the Ontario Science Centre, the exhibit toured for 15 years and attracted more than 6 million visitors.

The four-part NOVA series Making Stuff was produced in cooperation with MRS and aired on PBS. More than 14.5 million viewers tuned in during its premiere and another 16 million people during rebroadcasts.

During that time, MRS was also a primary partner in NISENet, the Nanoscale Informal Science Education Network. The network shared information and resources, such as NanoDays kits, with more than 550 museums, universities, and other organizations, reaching a combined 30 million people.

In major outreach undertakings like these, securing funding is usually the hardest part, Floro told attendees. The National Science Foundation supported all three of these efforts.

Design of Thermally Robust HfAlO_x/HfZrO₂ Hetero-Hafnia Stacks for High-Temperature Microelectronics Integration

Written by Soumyajyoti Mondal

Donghoon Kim of the Samsung Advanced Institute of Technology presented a study on improving the thermal stability of hafnia-based ferroelectric devices for high-temperature processing conditions. The research group showed that conventional HfZrO₂-based devices degrade significantly after thermal treatment, with a reduction in memory window and loss of functionality under standard operating conditions. To address this, a hetero-hafnia stack combining HfAlO_x and HfZrO₂ was introduced. The devices maintained stable operation even after high-temperature exposure, with no degradation in switching behavior and, in some cases, improved performance. The stability was attributed to the suppression of the non-ferroelectric monoclinic phase that typically forms during thermal processing. The importance of stack design was also highlighted, where placing the HfAlO_x layer at the channel interface improves thermal robustness and limits defect-related degradation. The results show that careful engineering of the ferroelectric stack can improve device stability under realistic processing conditions.

Property-Constrained Materials Design via a Hybrid Diffusion-Genetic Algorithm Framework

Written by Melody Yiyuan Zhang

Shivang Agarwal presented a novel hybrid diffusion–genetic algorithm (Diff-GA) framework for inverse materials design under user-defined compositional and property constraints. The work addresses a major limitation of existing crystal generative approaches: conventional genetic algorithms largely search within existing structural spaces, screening pipelines are restricted to database entries, and diffusion-based generative models often require expensive retraining or large property-labeled datasets for each new target. Diff-GA overcomes these issues by coupling a pre-trained diffusion model with genetic optimization during intermediate denoising stages. Partially denoised crystal candidates are evaluated using black-box scoring functions, and their latent representations are iteratively refined through selection, crossover, and mutation to guide subsequent generations toward desired properties. This closed-loop strategy enables controllable property optimization without modifying the underlying generative model. Demonstrations in Ti–Al–N, Li-halide, and Cs–Sn–I systems showed substantial improvements over diffusion-only and GA-only baselines, while maintaining structural novelty and diversity, highlighting the promise of Diff-GA for automated discovery of battery and functional materials.

Hrushikesh Pravin Sahasrabuddhe (Lawrence Berkeley National Laboratory, University of California, Berkeley), Kangan Wang (Lawrence Berkeley National Laboratory, University of California, Berkeley), Sanat Ghosh (Columbia University), Elena Cortese (Justus-Liebig-Universität Giessen), Aki Hiraoka (The University of Tokyo), and Mohamed Elekhtiar (University of Nebraska–Lincoln)

Designing Wide-Temperature Electrolytes for High-Energy Batteries

Written by Abhinanda Sengupta

Achieving reliable battery performance across wide temperature ranges remains a central challenge for next-generation energy storage, particularly for electric vehicles operating under sub-zero conditions. Jijian Xu presented a solvation-structure-driven framework for electrolyte design that moves beyond conventional formulations toward mechanistically tuned interfacial chemistry. The talk outlined a three-stage evolution in electrolyte engineering. The first stage focuses on cation solvation, where “soft-solvating” solvents are selected to reduce lithium desolvation energy, thereby enhancing ion transport and facilitating interphase formation. The second stage expands the design space to anion solvation, emphasizing solvent–anion interactions that influence both bulk transport and interfacial chemistry. Finally, a synergistic cation–anion solvation strategy is introduced, deliberately balancing coordination environments to simultaneously optimize transport, stability, and temperature tolerance. Experimental insights demonstrated by Xu highlight why conventional carbonate-based electrolytes fail at low temperatures due to unfavorable freezing points and sluggish kinetics. In contrast, weakly coordinating solvents with optimized donor numbers enable improved solvation structures, supporting ion mobility even under extreme conditions. This work reframes electrolyte design as a solvation-controlled problem, where interfacial stability, transport kinetics, and temperature resilience are co-optimized through molecular-level engineering.

Gwanghyeon Jang, Kangwon National University

Built-In Field Engineering in Ferroelectric Hf_0.5Zr_0.5O₂ Capacitors via Electrode Replacement

Gwanghyeon Jang of Kangwon National University presented a study on controlling internal electric fields in ferroelectric Hf_0.5Zr_0.5O₂ capacitors through electrode engineering. The work focuses on how the choice of electrode material influences polarization switching behavior. The devices were first fabricated with symmetric tungsten electrodes to stabilize the ferroelectric phase, after which the top electrode was replaced with metals of different work functions. This created an asymmetry between the top and bottom electrodes, leading to a built-in electric field across the ferroelectric layer. The effect of this internal field was observed as a shift in the polarization–electric field hysteresis loops, with the direction of the shift determined by the relative work function of the electrode. Since the ferroelectric phase remained unchanged during electrode replacement, the observed changes can be attributed to electrical effects rather than structural variations. The study shows that electrode work-function differences can be used to tune switching behavior and control the operating voltage in ferroelectric capacitors.

Anudeep Tullibilli, Indian Institute of Science

Investigating the Role of Oxygen Vacancy Redistribution in Ferroelectric Behavior of Yttria-Doped Hafnia Thin Films Using Scanning Probe Microscopy

Anudeep Tullibilli of the Indian Institute of Science presented a study on the role of oxygen vacancy redistribution in yttria-doped hafnia thin films and its influence on ferroelectric behavior. The work focuses on how the movement of oxygen vacancies affects both electrical properties and surface morphology. Using scanning probe microscopy, the evolution of the films during electrical cycling was examined. After the wake-up process, the films showed a uniform height contrast between biased and pristine regions, indicating a reversible and controlled motion of oxygen vacancies. In contrast, samples exhibiting leakage displayed significant surface roughening, associated with dendritic patterns of vacancy accumulation. It was further discussed that, in leaky samples, changes in the oxidation state of Mn in the LSMO layer contribute to this behavior. These observations suggest that different modes of oxygen vacancy migration, along with interfacial redox processes, govern the transition between stable ferroelectric behavior and degraded states. The study provides additional insight into the origin of leakage in ultra-thin hafnia-based ferroelectric films, a key challenge for device applications.

Vincent Consonni, Université Grenoble Alpes

Atomic Layer Deposition of Epitaxial Ga₂O₃ Thin Films on c-Plane Sapphire

Vincent Consonni of Université Grenoble Alpes presented a study on the growth of Ga₂O₃ thin films using atomic layer deposition on c-plane sapphire substrates. The work explores the use of triethyl gallium and ozone as precursors over a range of deposition temperatures. A stable growth window was identified, within which uniform films with controlled thickness and smooth surface morphology were obtained. Structural analysis showed that films grown at lower temperatures remain amorphous, while increasing the deposition temperature leads to the onset of crystallization and eventually to well-defined epitaxial growth. At higher temperatures, clear diffraction features corresponding to κ-phase Ga₂O₃ were observed, indicating improved crystalline quality. Additional characterization using Raman spectroscopy, transmission electron microscopy, and x-ray photoelectron spectroscopy confirmed the structural and chemical properties of the films. The study demonstrates that atomic layer deposition can be used to achieve controlled and relatively low-temperature growth of epitaxial Ga₂O₃ thin films.

Nadire Nayir, Paul Drude Institute, Germany

Rational Computational Design of Metal-Organic Precursors for Complex Oxide Thin Film Growth via Multi-Physics Modeling—Titanium(IV)-Isopropoxide Thermolysis as a Case Study

Understanding how metalorganic precursors decompose during thin film growth is essential for controlling deposition processes, yet the underlying mechanisms are often simplified. In this talk, Nadire Nayir from Paul Drude Institute presented a multiscale modeling framework to investigate the decomposition of Titanium(IV) isopropoxide (TTIP), a widely used precursor in chemical vapor deposition.

Conventional models assume that TTIP primarily decomposes through β-hydride elimination, involving C–O bond dissociation. However, by combining quantum chemistry, reactive molecular dynamics, and metadynamics simulations, Nayir showed that this pathway represents only part of the process. Additional reactions, including β-X elimination mechanisms, were identified as significant contributors to precursor breakdown.

This expanded understanding provides a more complete picture of how metalorganic molecules evolve during deposition. Beyond this specific system, the modeling framework offers a general approach for predicting and tailoring precursor behavior.

These results highlight the potential for computationally guided design of next-generation precursors with optimized decomposition pathways, enabling improved control in thin film growth technologies.

Franziska Elisabeth Muenzer, University Duisburg-Essen

Lateral Heterostructures of Electrochemically Exfoliated MoS₂ and Graphene Flakes via Scalable Langmuir Assembly

Scalable assembly of two-dimensional materials into functional heterostructures remains a key challenge, particularly for lateral architectures that can be integrated over large areas. In this talk, Franziska Elisabeth Muenzer from University Duisbur-Essen presented a solution-based Langmuir deposition approach for creating mixed flake films composed of graphene and MoS₂.

Graphene and MoS₂ were first processed into stable inks and then assembled at the air–water interface. The surface pressure during deposition was used to control how the flakes pack together. At low pressure, the flakes remain well ordered but separated. Increasing the pressure promotes edge to edge contact, while further compression leads to local buckling, indicating the limit of maintaining a flat film.

Spectroscopic and microscopic characterization revealed how these structural changes influence materials behavior. Raman mapping confirmed spatial separation between graphene and MoS₂, while photoluminescence measurements showed a redshift in MoS₂ emission as graphene contact increased, indicating stronger electronic interaction. Electrical measurements further showed a percolation threshold as the graphene concentration increased, marking the transition to conductive pathways.

These results demonstrate a versatile and scalable method for engineering large area lateral heterostructures with tunable morphology and electronic properties, supporting the development of flexible electronic devices based on two-dimensional materials.

Max Christian Lemme, AMO GmbH / RWTH Aachen University

Engineering 2D MoS₂ Films for Ion-Based Memristive Switching

Memristors are key components in neuromorphic computing, where electronic devices mimic the behavior of biological synapses. In this talk, Max Lemme from AMO GmbH and RWTH Aachen University presented device designs and switching mechanisms in memristors based on two-dimensional materials, along with their integration into silicon CMOS circuits.

Lemme showed that the weak van der Waals bonding in layered materials such as MoS₂ creates nanoscale gaps that can guide the formation of conductive filaments. In vertical devices fabricated from MoS₂ films, silver ions migrate through these gaps under an applied electric field, enabling volatile resistive switching at low voltages. These devices exhibit consistent switching below 1 V, fast response times on the order of hundreds of nanoseconds, and stable operation over thousands of cycles.

Similar ion migration behavior was observed in lateral devices with submicrometer channel lengths, which operate at even lower voltages and do not require an initial forming step. Operando transmission electron microscopy provided direct visualization of filament formation during device operation.

These results demonstrate how two-dimensional materials enable controlled switching mechanisms for low-power memristors, supporting their application in neuromorphic circuits and next-generation computing architectures.

Strain-Induced Mechanical Instabilities in Freestanding Oxide Membranes

Written by Soumyajyoti Mondal

Nini Pryds of the Technical University of Denmark presented a study on freestanding oxide membranes based on SrTiO₃ and BaTiO₃. Advances in thin-film growth and membrane transfer enable these materials to relax their built-in strain when suspended over lithographically defined cavities, leading to out-of-plane deformation through buckling instabilities. The deformation is governed by the interplay between residual strain, membrane thickness, bending stiffness, and cavity geometry, resulting in well-defined stable and metastable configurations. Experimental observations, supported by finite-element modelling, show strong agreement between predicted and measured buckling profiles. These mechanically locked states correspond to distinct strain distributions and can undergo reversible snap-through transitions under external force. In ferroelectric BaTiO₃ membranes, the deformation produces spatial variations in electromechanical response, reflecting the coupling between strain and polarization. The results describe how mechanical instabilities in freestanding oxide membranes can be understood in terms of geometry, strain, and material properties.

Low Temperature Atomic Layer Deposition Processes for Large-Area Synthesis of 2D Transition Metal Dichalcogenides

Written by Yuying Ning

Scalable and controllable synthesis of two-dimensional materials remains essential for their integration into nanoelectronic devices. In this talk, Ageeth A. Bol from University of Michigan presented recent progress in the large-area growth of MoS₂ thin films using plasma-enhanced atomic layer deposition (ALD) with advanced cycle schemes.

By tuning plasma conditions such as gas composition and power, the research team demonstrated improved control over film stoichiometry and crystallinity in wafer-scale polycrystalline MoS₂. These process parameters play a critical role during the plasma step, influencing sulfur incorporation and overall film quality.

Bol also addressed limitations of conventional ALD processes that rely on hydrogen sulfide (H₂S), a hazardous and difficult-to-handle gas. As an alternative, a modified process using hydrogen plasma and a liquid sulfur precursor, di-tert-butyl disulfide (TBDS), was introduced. This approach reduces safety concerns while maintaining compatibility with existing ALD systems.

Comparison of the two methods shows that the alternative precursor route offers a promising pathway for safer and more practical large-scale deposition of MoS₂ films. These results support the development of scalable and industry-compatible processes for 2D semiconductor integration.

Electric-Field Control of Topological Phases in Germanene

Written by Yuying Ning

Controlling topological phases across different length scales offers new opportunities for designing low-power electronic and quantum devices. In this talk, Pantelis Bampoulis from the University of Twente presented a systematic exploration of topological behavior in epitaxial germanene as its dimensionality is reduced from two dimensions to one and zero dimensions.

Bampoulis first demonstrated that monolayer germanene exhibits quantum spin Hall behavior, characterized by a sizable bulk gap and robust edge states. By applying a perpendicular electric field, the system can be tuned through a sequence of phase transitions, from a topological insulator to a Dirac semimetal and eventually to a trivial insulating state.

When confined into zigzag-terminated nanoribbons, the material shows a dimensional crossover. Below a critical width of approximately 2 nm, the two-dimensional topological phase breaks down, giving rise to one-dimensional states with symmetry-protected edge modes. In ultranarrow ribbons, scanning tunneling microscopy measurements combined with theoretical modeling revealed reversible, all-electric control of localized zero-dimensional states.

These results demonstrate germanene as a versatile platform for engineering topological phases across dimensions, with potential applications in low-energy electronics and quantum information technologies.

Matthew J. McDermott, Newfound Materials, Inc.

Can We Actually Make Any of This Stuff? Bridging Prediction and Experiment in the AI Era of Materials Discovery

Recent advances in AI-driven crystal structure generation have enabled the prediction of millions of inorganic compounds, yet many of these materials remain experimentally inaccessible. Matthew J. McDermott addressed the critical gap between computational discovery and laboratory realization by introducing a high-throughput framework for assessing synthetic accessibility. Rather than treating thermodynamic stability as the sole indicator of synthesizability, the presented approach explicitly evaluates reaction pathways through network analysis, pathway selectivity, and practical processing constraints. McDermott emphasized that stability alone is often insufficient: some theoretically favorable materials may still fail experimentally due to kinetic barriers or limited precursor routes. By expanding the searchable reaction and process space, however, previously hidden synthetic pathways can emerge. A cloud-based platform integrating reaction simulation, pathway scoring, synthesis planning, and experimental feedback was developed to benchmark and predict what materials can realistically be made. Overall, the work provides a more practical and data-driven strategy for prioritizing experimentally realizable compounds, helping accelerate the translation of predicted materials into scalable synthesis.

Luis Barroso-luque, Meta

Advances in Machine Learning for Atomic-Scale Modeling—Progress and Perspectives from FAIR Chemistry

This presentation reviewed the FAIR Chemistry team’s six-year effort in advancing open-science machine learning models for atomistic simulations. Luis Barroso-luque emphasized that recent progress in large-scale datasets and machine-learned interatomic potentials (MLIPs) has enabled substantial gains in predictive accuracy across catalysis, chemistry, and materials science. A central theme was that model performance improves stepwise with increased high-quality training data, but conventional energy mean absolute error alone is insufficient for evaluating downstream reliability. Instead, physically meaningful metrics such as energy drift over time in NVE molecular dynamics, and the stability of predicted vibrational modes provide a more rigorous assessment of model robustness. Several FAIR Chemistry architectures were discussed, including OMat24, which reduces systematic softening bias, and UMA, a universal ML potential trained on a combinatorial dataset spanning broad chemical space. The talk highlighted Meta’s broader strategy of building unified, scalable, and generalizable atomistic ML frameworks capable of supporting diverse real-world scientific applications.