The scale of the problem is easy to underestimate. When neutrons slam into the lattice of a metal such as iron or tungsten, they knock atoms out of position, leaving behind vacancies and clusters of vacancies. As irradiation continues, these vacancy clusters grow into nano-voids, tiny empty cavities just nanometers across. Individually they are minuscule, but collectively they cause the material to swell, harden, and creep, degrading mechanical properties in ways that can shorten component lifetimes and compromise reactor safety. Fusion devices are particularly demanding because, unlike fission reactors, they subject structural materials to an especially harsh neutron spectrum that simultaneously creates both displacement damage and new chemical species inside the metal.

Among those species, helium is the most insidious. It is not present in the material to begin with; instead it is born inside the lattice through nuclear reactions between incoming neutrons and the atoms of the alloy itself. Helium atoms are essentially insoluble in most metals, so they drift toward defects, vacancies, dislocations, and especially grain boundaries, the interfaces between the microscopic crystallites that make up a polycrystalline metal. There they coalesce into bubbles. Grain boundaries act as traps, soaking up the mobile helium, but trapping is a double-edged solution: over time the accumulated helium can build pressure and concentration to the point where the boundary itself weakens, a process that embrittles the material and undermines the very interfaces that hold it together.

To understand the first half of this damage cascade, the HFIPS team needed to know what nano-voids actually look like when they are stable, a surprisingly difficult question because the possible atomic arrangements of a cavity grow combinatorially with its size. Their answer came from a new computational framework built on the differential evolution algorithm, an evolutionary optimization method that searches vast configuration spaces for the arrangement that minimizes energy. Applied to body-centered cubic metals, the framework mapped out the ground-state structures, energetic properties, and a linear discrete model describing how nano-void properties change with size. The simulations revealed a clear pattern: small nano-voids tend to adopt compact, polyhedral shapes, while larger voids develop faceted geometries bounded by low-energy crystallographic surfaces, notably the (110) and (100) planes. In other words, as voids grow, the metal effectively carves out cavities whose walls align with the crystallographic directions that cost the least energy.

That structural insight had a kinetic payoff. The researchers found that vacancies situated around a nano-void do not all behave identically; their energy levels depend on the local atomic environment, meaning that a vacancy adjacent to a smooth facet sits in a different energetic landscape than one tucked into a corner or edge of the cavity. By linking the energy levels of nano-voids directly to their atomic structures, the team built a bridge between what a void looks like and how it moves. This connection helps explain why some voids are remarkably stable while others migrate through the lattice, a distinction that ultimately governs whether defects disperse harmlessly or accumulate into the clusters that cause swelling, hardening, and creep. A linear discrete model relating void energetics to size gives researchers a quantitative tool for predicting this behavior without simulating every atom of every possible configuration.

With the defect physics established, the team turned to the second, and arguably more dangerous, half of the problem: what to do about the helium. Grain boundaries had long been viewed mainly as sinks, passive traps where helium collects and bubbles nucleate. Li’s group asked a different question: could some grain boundaries, rather than merely storing helium, actually help move it out of the material entirely? To answer it, the researchers combined high-throughput calculations with object kinetic Monte Carlo simulations, a modeling approach that tracks defect populations over experimentally relevant length and time scales far beyond what atom-by-atom dynamics alone can reach.

The simulations delivered a result with real engineering promise. Grain boundaries are not interchangeable; their atomic structure varies widely depending on how the adjacent crystals are misoriented. The team found that some boundaries, those with suitable atomic structures, act as rapid diffusion channels for helium. Along these boundaries, helium atoms move orders of magnitude faster than they can through the perfect lattice. Crucially, when such a fast-diffusion boundary is connected to a free surface of the material, it becomes a pipeline: helium that would otherwise stagnate in the interior can instead be carried along the boundary and released at the surface. The researchers observed this same behavior in both iron and tungsten, the two signature bcc metals of fusion structural materials research, suggesting the mechanism is general rather than an accident of one particular element.

From these findings the researchers distilled a design philosophy they describe as a capture-transport-expulsion strategy. Instead of relying on grain boundaries purely as passive sinks, a properly engineered microstructure would use them as a coordinated system: boundaries capture the mobile helium, transport it along fast diffusion pathways, and, when those pathways intersect free surfaces, expel it from the material altogether. This reframing moves helium management from passive capture toward active removal, offering a theoretical basis for grain-boundary engineering as a route to materials that resist helium-induced damage rather than merely tolerating it. In practice, such engineering would mean tailoring the types, densities, and connectivity of boundaries in an alloy so that the microstructure functions as a built-in helium plumbing network.

The significance for fusion energy development is considerable. One of the central obstacles to a viable fusion power plant has always been the lifetime of first-wall and blanket materials, which must endure decades of neutron exposure in a commercial-scale device. Every mechanism that mitigates swelling, embrittlement, and helium bubble formation extends component lifetimes, reduces maintenance, and improves the economics of fusion power. The HFIPS work contributes on two fronts simultaneously: a predictive, size-resolved model of how the vacancy defects that drive swelling form and evolve, and a microstructural design principle aimed specifically at the helium that embrittles boundaries. Because the atomistic picture was developed for body-centered cubic metals, it speaks directly to tungsten, the leading candidate for plasma-facing components, and to the iron-based ferritic and reduced-activation steels favored for structural applications.

The team frames the achievement in terms of fundamental understanding feeding directly into design. By understanding how these defects form and evolve, Li Xiangyan said, the researchers hope to provide useful guidance for developing fusion materials with better radiation resistance. The combination of evolutionary algorithms for ground-state searching, discrete models of defect energetics, and kinetic Monte Carlo simulation of defect populations illustrates how modern computational materials science can close the gap between atomic-scale mechanisms and engineering-scale strategies. What emerges is a picture of radiation damage that is not an unavoidable fate but a set of processes that can, in principle, be routed and redirected, with the right atomic-scale architecture turning a material’s own grain boundaries from collection points for helium into escape routes for it.

Subject of Research: Irradiation damage mechanisms and helium mitigation in fusion reactor structural materials

Article Title: Researchers reveal irradiation damage mechanisms and mitigation strategies for fusion reactor structural materials

Article References: Researchers reveal irradiation damage mechanisms and mitigation strategies for fusion reactor structural materials. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: fusion reactors, nano-voids, irradiation damage, helium bubbles, grain boundaries, tungsten, iron, differential evolution algorithm, kinetic Monte Carlo, radiation resistance, structural materials, HFIPS

The stakes could hardly be higher. Lithium metal is the dream anode, packing a theoretical specific capacity of 3860 mAh per gram with a remarkably low mass density of 0.534 g per cubic centimeter and an aggressive reducing potential of −3.04 volts against the standard hydrogen electrode. Paired with nonflammable inorganic solid electrolytes such as garnet-type LLZO, it underpins not only next-generation solid-state cells but also exotic lithium–sulfur and lithium–air chemistries. The problem is that a solid–solid contact is never truly conformal. Unlike the liquid electrolytes that wet every crevice of a conventional electrode, a rigid ceramic pressed against soft lithium leaves microscopic gaps, and those imperfections amplify into spatially heterogeneous deposition and dissolution. Years of experiments, from pressure-dependent stripping tests on Li–LLZO cells to operando synchrotron X-ray tomography, have converged on a single conclusion: contact loss during stripping, not interphase growth, is the primary driver of current constriction and cell failure.

What has been missing is a mechanistic account of how an individual void actually behaves once it forms. Does it simply grow monotonically under stripping, as many models assumed? Or can it shrink, migrate, or even detach from the interface entirely? The research team behind the new work, using a three-dimensional kinetic Monte Carlo framework, simulated a 50-by-50-by-50 cubic lattice representing a 17.5-nanometer cube of lithium against LLZO, and tracked a single hemispherical void through the competing atomic events that shape it. Four mechanisms were woven into the model: bulk migration of lithium vacancies away from the interface, surface diffusion of lithium atoms across the evolving void boundary, electro-dissolution at active contact sites governed by Butler–Volmer kinetics, and transport of lithium ions into the electrolyte bulk. Each event carried an Arrhenius rate constant calibrated to literature values, with activation barriers of 0.3 eV for terrace diffusion, 0.15 eV for step diffusion, and 0.5 eV for interlayer diffusion.

The simulations revealed a strikingly choreographed two-act drama. In the first act, current focusing concentrates the electric field at the sharp circular edge of the hemispherical void, where the electrolyte’s potential distribution distorts around the non-conducting cavity. The enhancement decays with a 1/r-cubed dependence, so dissolution attacks the void rim preferentially, and the void spreads laterally along the interface while its depth barely changes. In the second act, surface diffusion takes over: vacancies redistribute from the interface down along the void surface, the void’s footprint retracts slightly, and its depth increases substantially. The team quantified this morphological ballet with geometric descriptors, including a circularity index that dropped as the footprint elongated and roughened, then recovered as surface diffusion smoothed the shape back toward symmetry, and a vertical shape factor that stayed flat during lateral growth before climbing sharply during the deepening phase.

The most consequential discovery concerns the three distinct modes of surface diffusion, classified through a terrace-step-kink framework that distinguishes atomic hops by their coordination environment. Terrace diffusion, in which adatoms glide across flat crystalline planes, and step diffusion, in which they travel along step edges, both act to keep voids compact and anchored at the interface. Interlayer diffusion, by contrast, shuttles atoms between adjacent atomic layers and directs vacancies into the depth of the electrode. When interlayer diffusion dominates, voids detach from the interface and drift into the lithium bulk, which paradoxically allows the interface to heal as fresh contact is restored. Phase maps sweeping the activation barriers from 0.05 to 0.5 eV showed that suppressing interlayer diffusion while keeping terrace diffusion active produces flat, pancake-shaped voids, whereas strong step diffusion inflates voids in both lateral and depth directions, and combined dominance of interlayer and bulk diffusion can dissolve a void entirely into the electrode.

To unify these observations, the researchers introduced an electrochemical Damköhler number, the ratio of the reaction rate to the bulk vacancy diffusion rate, which compares how fast vacancies are generated at the interface with how fast they can be swept away into the anode interior. When this number is far below unity, vacancy removal outpaces generation, vacancies disperse, and voids shrink or vanish altogether, defining a no-void regime. When it is far above unity, vacancies pile up around the void perimeter, and the resulting morphology is dictated by the surface diffusion ratios. Strong terrace diffusion yields compact but irregular voids; strong step diffusion carves the deepest attached voids; strong interlayer diffusion promotes detachment at moderate Damköhler numbers and complete dissolution when bulk diffusion is also strong. Notably, the analysis showed that bulk diffusion and interlayer diffusion are two independent routes by which a void can effectively disappear into the electrode, a finding with direct implications for how operating conditions might be tuned to favor healing over degradation.

Temperature emerged as a double-edged lever. Sweeping the operating temperature from −10 to 80 degrees Celsius, a range that spans subzero winter starts to sustained elevated-temperature operation in real cells, the model showed that higher temperatures accelerate both surface diffusion and bulk vacancy transport through Arrhenius kinetics, effectively enabling a transport-driven healing mechanism that suppresses void growth. But the authors caution that this benefit is chemistry-dependent. For garnet oxides like LLZO, whose interface with lithium is kinetically passivated, the morphological advantage of warmth is captured directly by the model. Sulfide electrolytes, however, can undergo interfacial reduction and form interphases that grow faster, and sometimes more dangerously, at elevated temperature, so chemical degradation may offset the healing benefit. The optimal operating temperature therefore depends on the balance between morphological recovery and chemical stability, and it will vary with electrolyte chemistry in ways the current framework deliberately leaves for future work.

Perhaps the most dramatic finding is the percolation transition. As stripping proceeds, thousands of single-site vacancy clusters nucleate randomly at the interface and along void edges, where current focusing makes dissolution most aggressive. These point defects act as nuclei that grow through aggregation and coalescence. The team tracked cluster populations across four stages of stripping and watched the original 6-cubic-nanometer void swell to 9 cubic nanometers as neighbors merged into it. Two coupled mechanisms drive the merging: current focusing is strongest in the narrow lithium ligaments separating adjacent voids, so those bridges dissolve preferentially, and curvature-mediated surface diffusion then transports atoms away from the high-curvature necks, lowering surface energy and consolidating each merged pair into a single compact void. At a critical void fraction of 0.0256, the void network suddenly connects across the simulation domain, snapping from an isolated-defect state to a percolated structure composed entirely of one giant cluster.

The electrochemical consequences of that geometric transition are severe. Using a flux-tube formulation, the researchers estimated the interfacial constriction resistance directly from the simulated contact morphology and found that it climbs by roughly two orders of magnitude across the percolation window, peaking at the minimum contact fraction before partially recovering as void detachment restores contact. The percolation threshold itself is tunable: strong terrace diffusion keeps voids as isolated compact clusters and prevents percolation entirely, while strong interlayer diffusion detaches voids before they can link laterally. A connected, contact-destroying network emerges only when both modes are sufficiently weak, and within that regime, step diffusion acts as the controlling mechanism, with strong step diffusion delaying the onset of percolation by carving deeper, more isolated voids instead of shallow lateral ones.

The model’s predictions align qualitatively with a decade of experimental evidence: the lateral-then-deep growth sequence matches operando dilatometry observations, the Damköhler threshold echoes the critical stripping current above which voids accumulate, and the partial healing under enhanced surface diffusion mirrors the recovery of dewetted interfaces after thermal treatment. The framework deliberately excludes mechanical effects such as stack pressure and lithium creep, which would supply material toward the interface and slow void growth, meaning the present results represent a conservative, pressure-free bound on degradation. Even so, the study delivers something solid-state battery engineers have lacked: a theoretical baseline identifying which surface diffusion modes dictate void morphology, and therefore a design compass for selecting electrolyte chemistries, operating temperatures, and current densities that keep lithium’s molecular potholes from connecting into a highway to failure.

Subject of Research: Void formation and percolation dynamics at the lithium metal–solid electrolyte interface during stripping in solid-state batteries

Article Title: Mechanistic Insights Into Instability From Single Void Dynamics at the Lithium Metal‐Solid Electrolyte Interface

Article References: Banerjee, S., Vishnugopi, B. S., Lin, G., & Mukherjee, P. P. (2026). Mechanistic Insights Into Instability From Single Void Dynamics at the Lithium Metal‐Solid Electrolyte Interface. Advanced Science, Article e78028. https://doi.org/10.1002/advs.78028

Image Credits: AI Generated

DOI: 10.1002/advs.78028

Keywords: solid-state batteries, lithium metal anode, solid electrolyte, void formation, kinetic Monte Carlo, surface diffusion, vacancy transport, percolation, LLZO, electrochemical stripping, interfacial stability, dendrite formation

At the heart of the work is a fundamental challenge in additive manufacturing: the properties of a 3D-printed metal part depend intimately on its grain structure, the spatial arrangement and orientation of crystalline regions separated by grain boundaries. Fine grains typically enhance strength and toughness through grain-boundary strengthening, while larger or anisotropic grains can improve ductility and creep resistance. Controlling these attributes during printing requires understanding how processing conditions such as laser scanning velocity, melt pool geometry, and heat-affected zone dimensions translate into grain evolution, a relationship complicated by vast multidimensional parameter spaces and non-linear thermal dynamics.

The research team, led by Dipayan Sanpui and Subramanian K.R.S. Sankaranarayanan, built their framework on a repository of 1,524 successfully retrieved three-dimensional polycrystalline microstructures generated with the open-source SPPARKS kinetic Monte Carlo code. These simulations modeled hundreds of laser passes over a substrate using a modified Potts Monte Carlo approach, capturing curvature-driven grain growth, melting, remelting, and nucleation in the heat-affected zone. Rather than relying on abstract statistical descriptors such as two-point statistics or chord length distributions used in earlier studies, the team extracted physically interpretable features directly from two-dimensional slices of the simulated structures.

Four descriptors formed the quantitative vocabulary of the study: average grain size, surface-to-volume ratio, sphericity, and roundness. Grain size serves as a first-order measure linked directly to mechanical performance through the Hall-Petch relationship, while surface-to-volume ratio distinguishes microstructures with similar grain dimensions but different boundary complexity. Sphericity captures overall grain compactness and elongation, whereas roundness reflects the sharpness of grain edges and corners, quantities that differ when two grains share similar elongation but diverge in boundary smoothness due to solidification dynamics. The team computed these measures directly from grain identifiers in the SPPARKS output files, streamlining integration into automated materials design pipelines.

With the features extracted, the researchers framed the problem as supervised regression, using seven processing conditions, including scanning velocity, scanning pattern, molten zone width, depth, and tail length, and heat-affected zone width and tail length, as inputs. They trained and compared multiple algorithms: Random Forest, Gradient Boosting Machine, Extreme Gradient Boosting (XGBoost), a feed-forward neural network, and a one-dimensional convolutional neural network, with hyperparameters optimized through six-fold grid search cross-validation and Keras-Tuner’s Hyperband search for the neural architectures.

Tree-based ensemble methods consistently outperformed the neural networks. XGBoost emerged as the best overall performer, achieving a coefficient of determination of 0.977 for grain size prediction, 0.946 for surface-to-volume ratio, 0.954 for sphericity, and 0.957 for roundness on test data, with a mean absolute error of just 6.62 pixels for grain size. The authors attribute grain size’s high predictability to its status as a first-order reflection of the temperature distribution across the simulation domain, while shape descriptors such as sphericity and roundness depend on secondary morphological effects, including grain boundary curvature and irregular grain interactions, that are harder to infer from processing conditions alone.

A particularly striking finding emerged from the parity plots: grain size predictions split into two distinct clusters corresponding to different microstructural regimes. The first cluster, with average grain sizes between 0 and 450 pixels, represented anisotropic structures of alternating equiaxed and elongated grains formed at higher scanning velocities and broader heat-affected zones. The second cluster, spanning 600 to 1200 pixels, arose from low-velocity, narrow-heat-affected-zone conditions that promote localized remelting and coarser, irregular grains. When trained and evaluated within the well-behaved first regime, the model reached an R² of 0.982 with an error of only 2.305 pixels; in the morphologically heterogeneous second regime, accuracy dropped to R² of 0.561, revealing where the surrogate’s reliability limits lie.

To open the black box, the team applied Shapley Additive exPlanations (SHAP), a game-theory-based framework that quantifies each input feature’s marginal contribution to predictions. The analysis showed that scanning velocity and heat-affected zone characteristics dominate grain size outcomes globally, consistent with established solidification theory in which thermal gradients, cooling rates, and grain-boundary mobility govern grain evolution. At high velocities, scanning speed and HAZ tail length controlled the outcome, with rapid cooling producing fine, anisotropic grains. At minimum velocity, scanning direction and HAZ tail length became the primary drivers: cross-hatch patterns induced multidirectional heat flux and rapid cooling that favored clusters of fine equiaxed grains, while unidirectional scanning promoted elongated, directionally aligned grains.

Beyond interpretation, the framework demonstrated practical utility for process optimization. By filtering candidate parameter combinations according to the relative error between predicted and desired grain sizes, the researchers identified viable alternative processing conditions under fixed maximum or minimum laser velocities. Under high velocity, neighboring points in parameter space with variations in melt-zone geometry yielded nearly identical grain structures, suggesting flexibility in achieving consistent microstructures. Under low velocity, HAZ tail length and scanning direction offered the most meaningful degrees of freedom. The computational payoff is substantial: simulating a single 100 × 100 × 100 site microstructure took roughly 25 minutes across four compute nodes, while the trained models deliver predictions in minutes once trained.

The authors envision their approach as a foundation for digital twins and closed-loop process control, in which sensor-derived process variables feed continuously into trained models for online quality monitoring and adaptive parameter optimization. They acknowledge limitations, notably that featurization currently operates on two-dimensional slices rather than full three-dimensional volumes, and they point toward extending the framework to experimental microscopy images, inverse design, and active-learning-based calibration as next steps. By coupling high-fidelity physics simulation with interpretable machine learning, the study illustrates how explainable AI can transform additive manufacturing from a process of empirical trial and error into a predictive, quantitatively grounded engineering discipline.

Subject of Research: Interpretable machine learning prediction of process–structure relationships in metal additive manufacturing using kinetic Monte Carlo microstructure simulations

Article Title: Data-driven discovery of process–structure relationships in additive manufacturing via featurization from kinetic Monte Carlo simulations and interpretable machine learning

Article References: Sanpui, D., Chan, H., Koneru, A., Banik, S., Manna, S., & Sankaranarayanan, S. K. (2026). Data-driven discovery of process–structure relationships in additive manufacturing via featurization from kinetic Monte Carlo simulations and interpretable machine learning. Machine Learning with Applications, 25, Article 100985. https://doi.org/10.1016/j.mlwa.2026.100985

Image Credits: AI Generated

DOI: 10.1016/j.mlwa.2026.100985

Keywords: additive manufacturing, kinetic Monte Carlo, machine learning, microstructure, XGBoost, SHAP, grain size, process-structure linkages, featurization, surrogate modeling, explainable AI, metal 3D printing