Machine Learning for High-Throughput Configuration Sampling of Li−La−Ti−O Disordered Solid-State Electrolyte
Bibliographic record
Abstract
Most solid-state lithium electrolytes are disordered ionic crystalline materials that possess crystallographic sites that can be vacant or occupied by different ions. The presence of these partially occupied sites enables lithium diffusion through their lattice and makes such materials promising for developing all-solid batteries. High-throughput computational screening of such materials must bypass costly DFT sampling of disordered configurations and, therefore, commonly relies on the computationally efficient Coulomb approximation to find just a few representative low-energy ionic configurations, for which DFT is then used to quickly predict a number of important materials’ properties, such as the electrochemical stability window. This work demonstrates, using the Li−La−Ti−O solid electrolyte (LLTO) as an example, that the Coulomb approximation fails to correctly detect the most stable arrangement of Li and La ions in the LLTO, which has a noticeable impact on the accuracy of subsequent computational prediction of the electrochemical stability window of the material. The analysis herein shows that the sampling problem arises from the relatively modest geometry relaxation of the LLTO lattice. A kernel ridge regression machine learning (ML) method employing the smooth overlap of atomic positions as a structure descriptor (SOAP-KRR) leads to significant improvements in detecting the most stable configurations of the LLTO. The universal ML potential based on the multiple atomic cluster expansion is also found to be reliable but to a lesser extent than SOAP-KRR. Remarkably, accurate energies can be obtained with SOAP-RKK trained on as few as 40 LLTO structures, making this method promising for designing force matching ML potentials that can serve as a computationally inexpensive alternative to the costly DFT structure relaxation in high-throughput screening of large data sets of ionic materials.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".