Energy materials are now designed by how ions move

The bottleneck is labels, not models

A preprint roadmap on lithium battery electrolyte additives synthesises three decades of research into a structured computational framework, running from quantum-chemical screening to self-driving laboratories. Its central judgement is sobering for a field fond of new models: simple frontier orbital rules are insufficient for modern high-voltage applications. m:…4313

The authors argue that the field must now prioritise generating high-quality, function-specific experimental labels, because without them, computational screening has little reliable ground truth to learn from. That reframes additive discovery as a data problem first and an algorithm problem second, and it sets the context for the other preprints in this set, which each attack a piece of the ion-movement problem. m:…4313

Mapping where ions can go

A preprint on a new open-source toolkit, EnGOAT, analyses the free-energy surfaces of mobile species within crystalline hosts. It automates partitioning those surfaces into basins and transition states, allowing directional self-diffusion coefficients to be calculated via kinetic Monte Carlo simulation. m:…3476

Its distinctive choice is to define accessibility through energy isosurfaces rather than static van der Waals radii, giving what the authors describe as a more accurate, probe-specific representation of the energy landscape. The toolkit is aimed at high-performance battery electrolytes and porous materials, where knowing which paths an ion can actually take matters more than knowing the static crystal structure. m:…3476

A preprint on hydrogen molybdenum bronzes supplies the experimental counterpart. Analysing the structural response of these materials to high pressure with in situ neutron diffraction, researchers found that proton incorporation levels significantly influence framework stress response and compressibility, and identified specific local hydrogen environment distortions in D0.36MoO3 under compression up to 7.8 GPa. The findings are presented as fundamental insight into proton-conducting materials relevant to energy storage and catalysis. m:…1135

From mapping to making

A preprint on nanofiltration membranes shows the loop closing from prediction to synthesis. Selective Mg2+/Na+ separation matters for resource recovery from desalination brine, and the authors curated a dataset of 1,253 entries, used extremely randomised tree regressors to predict membrane performance with high accuracy, and then validated the workflow by experimentally synthesising and testing two optimised membrane recipes. m:…9559

That validation step is what separates this preprint from a purely computational claim. Two recipes were made and tested, not just ranked, which is the standard the battery roadmap implicitly demands when it calls for function-specific experimental labels. m:…9559 m:…4313

The roadmap and the toolkit point in the same direction, from opposite ends. The roadmap says screening needs better labels before it can be trusted for high-voltage additives, and the toolkit gives researchers a way to turn a crystal structure into directional diffusion numbers that such labels can be checked against. The membrane work then shows what happens when the loop is completed: a target selectivity is set, a model predicts which recipes meet it, and two of those recipes are made and tested. m:…4313 m:…3476 m:…9559

All four studies are preprints, and none reports a commercial device. What they jointly show is a field converging on ion movement as the design variable: better labels for additives, better maps of diffusion paths, better measurements of proton environments under pressure, and membranes designed by inverting the selectivity target. The next test for each is whether lab-scale insight survives scale-up and real operating conditions. m:…4313 m:…3476 m:…1135 m:…9559