Accelerating Discovery of Ternary Chiral Materials via Large-Scale Random Crystal Structure Prediction

Chongde Cao; Diwei Shi; Fengyuan Xuan; Jiexi Song

arxiv: 2508.04110 · v2 · submitted 2025-08-06 · ❄️ cond-mat.mtrl-sci

Accelerating Discovery of Ternary Chiral Materials via Large-Scale Random Crystal Structure Prediction

Jiexi Song , Diwei Shi , Fengyuan Xuan , Chongde Cao This is my paper

Pith reviewed 2026-05-19 01:12 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci

keywords chiral crystalsternary materialsrandom structure searchmachine learning interatomic potentialstopological materialsnonlinear opticscrystal structure prediction

0 comments

The pith

Combining universal machine learning interatomic potentials with random structure search identifies more than 260 chiral ternary crystals.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a prediction pathway that pairs universal machine learning interatomic potentials with random structure search to explore ternary crystal systems at large scale. Over 20 million randomly generated structures are optimized and screened for chiral space groups, after which stability is assessed. First-principles calculations then validate more than 260 stable chiral inorganic crystals. A sympathetic reader would care because known chiral semiconductors and semimetals remain scarce in databases, yet many of the newly identified phases show Weyl points, nonlinear optical responses, or superconducting potential.

Core claim

Through the integration of uMLIP-based high-throughput optimization and stability assessment with random crystal structure prediction, a large number of potentially stable phases were identified from over 20 million randomly generated chiral structures in ternary systems. First-principles validation further confirmed more than 260 chiral inorganic crystals with potential applications in topological properties, nonlinear optics, and superconductivity, some exhibiting the nonlinear Hall effect driven by Berry curvature dipole, quantum metric and symmetry-protected sixfold degenerate topological points, long Fermi arcs, and large magnetoresistance.

What carries the argument

Universal machine learning interatomic potentials (uMLIPs) used for rapid high-throughput structure optimization and stability screening inside a random structure search pipeline, followed by targeted filtering for chiral space groups.

If this is right

The pool of candidate chiral functional materials expands substantially beyond current databases.
A scalable strategy now exists for predicting stable phases in ternary material systems.
Some of the confirmed crystals host Weyl points near band edges or at the Fermi level.
Several phases display concrete quantum phenomena including nonlinear Hall effect and long Fermi arcs.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same workflow could be applied to quaternary or higher compositions to generate additional candidate sets.
Prioritizing a few of the 260+ structures for targeted experiments would test how well the uMLIP filter matches real-world stability.
Materials showing large magnetoresistance or sixfold topological points may connect to existing searches for symmetry-protected fermions.

Load-bearing premise

The uMLIP-based high-throughput optimization and stability assessment reliably identifies truly stable phases at scale without large numbers of false positives or missed low-energy structures.

What would settle it

Experimental synthesis and structural characterization confirming stability and chirality for a representative subset of the predicted crystals would support the count; repeated failure to stabilize any of the candidates would indicate the method overestimates the number of viable phases.

read the original abstract

Chiral inorganic crystals, particularly semiconductors with Weyl points near the band edges or semimetals hosting Weyl points at the Fermi level, have attracted considerable interest, yet they remain scarce in existing materials databases. This study presents a prediction pathway by combining universal machine learning interatomic potentials (uMLIPs) for high-throughput structure optimization with the broad exploration capability of random structure search (RSS), enabling large-scale crystal structure prediction in ternary systems with variable compositions, followed by targeted screening for chiral space groups. Through uMLIP-based high-throughput optimization and stability assessment, a large number of potentially stable phases were identified from over 20 million randomly generated chiral structures. First-principles validation further confirmed more than 260 chiral inorganic crystals with potential applications in topological properties, nonlinear optics, and superconductivity. Some of these materials exhibit notable quantum phenomena, such as the nonlinear Hall effect driven by Berry curvature dipole, quantum metric and symmetry-protected sixfold degenerate topological points, long Fermi arcs, and large magnetoresistance. This work substantially expands the pool of candidate chiral functional materials and offers a scalable strategy for predicting ternary material systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper scales up ternary chiral crystal prediction with uMLIPs and RSS to report over 260 DFT-validated examples plus some with topological or optical features, but the stability numbers rest on unshown uMLIP error metrics.

read the letter

Hi colleague, the main things to know are that this work combines universal ML potentials with random structure search across many ternary compositions, then filters for chiral space groups and confirms more than 260 candidates with first-principles calculations, while noting quantum phenomena such as nonlinear Hall effect from Berry curvature and long Fermi arcs in a few cases. That expands the short list of known chiral inorganic materials in a practical way. The pipeline itself is the clearest contribution: handling variable compositions at the scale of 20 million structures and then targeting chiral symmetries is a reasonable way to make discovery more efficient than fixed-composition searches. It also connects some of the finds to concrete properties like sixfold topological points and large magnetoresistance, which helps show why the effort matters. The soft spot is the validation step. The abstract gives no MAE for the uMLIP against DFT, no hull-distance distributions, and no count of how many candidates were dropped after closer checks. Ternary systems with mixed bonding can produce larger errors in these potentials, so the stress-test point about possible false positives is fair and could trim the final number if the model over-ranks marginal structures. Without those benchmarks it is hard to judge how much the 260 count depends on the specific uMLIP choice. This is the kind of paper computational materials people scanning for new chiral or topological candidates would want to read for the list and the workflow. A reader focused on high-throughput methods or functional materials screening would get the most out of it. It has enough scale and relevance to deserve a serious referee who can examine the error analysis and any cross-checks that are in the full text. I would send it to peer review rather than desk reject.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a high-throughput computational workflow combining universal machine learning interatomic potentials (uMLIPs) with random structure search (RSS) to generate and optimize over 20 million ternary crystal structures, followed by screening for chiral space groups and targeted first-principles DFT validation. This yields the identification of more than 260 stable chiral inorganic crystals with proposed applications in topological properties, nonlinear optics, and superconductivity, including examples exhibiting nonlinear Hall effect, Berry curvature dipole, and long Fermi arcs.

Significance. If the uMLIP screening and subsequent DFT validation prove robust, the work would substantially enlarge the catalog of known chiral functional materials and demonstrate a scalable RSS+uMLIP strategy for ternary systems that is otherwise computationally prohibitive. The scale of exploration (20M structures) and the focus on chiral space groups represent a clear methodological advance over database-mining approaches.

major comments (2)

[Abstract and Methods] Abstract and Methods (stability assessment section): the central claim of >260 DFT-confirmed stable chiral phases rests on uMLIP-based identification of low-energy candidates from 20M random structures, yet no quantitative validation metrics (MAE of formation energies or hull distances versus DFT for the relevant ternary compositions) are reported. Without these, it is impossible to bound the false-positive rate that could reduce the final validated count.
[Results] Results (stability screening and convex-hull construction): the manuscript does not specify the energy-above-hull threshold used for uMLIP pre-screening, the number of structures discarded after optimization, or whether full convex-hull constructions (including competing phases) were performed for every candidate before DFT. These choices directly affect whether the reported 260+ materials remain stable under more stringent or composition-specific criteria.

minor comments (2)

[Figures and Tables] Figure captions and supplementary tables should explicitly state the number of structures that reached each filtering stage (uMLIP optimization, chiral-space-group filter, DFT validation) to allow readers to assess selection efficiency.
[Methods] Notation for uMLIP model variants and RSS parameters (e.g., number of atoms per cell, composition sampling) is introduced without a dedicated methods subsection; a short table summarizing these hyperparameters would improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading of the manuscript and for the constructive comments, which have helped us improve the clarity and completeness of the work. We address each major comment below and have revised the manuscript to incorporate the requested details.

read point-by-point responses

Referee: [Abstract and Methods] Abstract and Methods (stability assessment section): the central claim of >260 DFT-confirmed stable chiral phases rests on uMLIP-based identification of low-energy candidates from 20M random structures, yet no quantitative validation metrics (MAE of formation energies or hull distances versus DFT for the relevant ternary compositions) are reported. Without these, it is impossible to bound the false-positive rate that could reduce the final validated count.

Authors: We agree that quantitative validation metrics are essential for assessing the reliability of the uMLIP screening step. Although the manuscript reports that over 260 candidates were confirmed stable via DFT, explicit MAE values for formation energies and hull distances on ternary compositions were not provided in the original text. In the revised manuscript we have added a dedicated paragraph in the Methods section (stability assessment) reporting these benchmark metrics against DFT for representative ternary systems, along with a brief discussion of how the observed errors inform the expected false-positive rate in the pre-screening stage. revision: yes
Referee: [Results] Results (stability screening and convex-hull construction): the manuscript does not specify the energy-above-hull threshold used for uMLIP pre-screening, the number of structures discarded after optimization, or whether full convex-hull constructions (including competing phases) were performed for every candidate before DFT. These choices directly affect whether the reported 260+ materials remain stable under more stringent or composition-specific criteria.

Authors: We acknowledge that these procedural details were omitted from the original Results section. The revised manuscript now explicitly states the energy-above-hull threshold applied during uMLIP pre-screening, reports the approximate number of structures discarded after optimization, and clarifies that full convex-hull constructions (incorporating all relevant competing phases drawn from existing databases) were carried out for each shortlisted candidate prior to DFT validation. These additions allow readers to evaluate the robustness of the final set of 260+ materials under alternative thresholds. revision: yes

Circularity Check

0 steps flagged

No circularity in discovery workflow; result from external validation

full rationale

The paper outlines a computational pipeline: random generation of over 20 million ternary structures, uMLIP-based optimization and stability screening, chiral space-group filtering, and subsequent first-principles DFT validation that yields the count of more than 260 confirmed crystals. This chain relies on established external methods and independent DFT checks rather than any self-definitional loop, fitted parameter renamed as prediction, or load-bearing self-citation. No equations or derivations reduce the final result to inputs by construction; the headline count is an empirical outcome of the validation step, making the overall analysis self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The workflow depends on the domain assumption that pretrained universal ML potentials can substitute for direct DFT during the initial high-throughput stage without introducing systematic errors that invalidate downstream stability rankings.

axioms (1)

domain assumption Universal machine learning interatomic potentials can reliably optimize structures and assess stability for ternary chiral crystals at scale.
Invoked for the high-throughput optimization and stability assessment step that filters the 20 million structures.

pith-pipeline@v0.9.0 · 5733 in / 1203 out tokens · 46870 ms · 2026-05-19T01:12:25.721821+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

uMLIP-based high-throughput optimization and stability assessment of over 20 million randomly generated chiral structures... First-principles validation further confirmed more than 260 chiral inorganic crystals

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.