Kinetic Monte Carlo-Ising Machine Optimization for Atomistic Inverse Design of Solid Electrolytes

Ai Koizumi; Ryo Tamura; Tomofumi Tada

arxiv: 2606.31261 · v1 · pith:MR6YCJ3Unew · submitted 2026-06-30 · ❄️ cond-mat.mtrl-sci

Kinetic Monte Carlo-Ising Machine Optimization for Atomistic Inverse Design of Solid Electrolytes

Ai Koizumi , Tomofumi Tada , Ryo Tamura This is my paper

Pith reviewed 2026-07-01 05:03 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci

keywords solid electrolytesionic conductivitykinetic Monte Carlofactorization machineIsing machine optimizationdopant configurationyttria-stabilized zirconiainverse design

0 comments

The pith

A framework pairing kinetic Monte Carlo with Ising-machine optimization locates dopant arrangements in solid electrolytes that raise conductivity by an order of magnitude over random and experimental values.

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

The paper presents a computational method for the atomistic inverse design of solid electrolytes that maximizes ionic conductivity. Kinetic Monte Carlo simulations compute conductivity for any chosen dopant placement while factorization machine quadratic optimization annealing iteratively builds a surrogate model from limited data and proposes new placements expected to perform better. Partitioning of the configuration space allows the optimizer to handle the combinatorial scale of realistic simulation cells. Applied to 8 mol % yttria-stabilized zirconia, the procedure returns a single dopant configuration whose conductivity exceeds both random arrangements and the experimentally reported value by roughly a factor of ten. The authors note that the same loop, when paired with laboratory measurements, could be used to infer microscopic structures from observed conductivity.

Core claim

The KMC-FMQA framework evaluates ionic conductivity via kinetic Monte Carlo for proposed dopant configurations and uses factorization machine with quadratic-optimization annealing to iteratively learn a surrogate model and propose new configurations that are expected to maximize conductivity, with space partitioning to handle large simulation cells; application to bulk 8 mol % yttria-stabilized zirconia yields a dopant configuration with an order-of-magnitude higher conductivity than random configurations and experimental values.

What carries the argument

The KMC-FMQA framework, in which kinetic Monte Carlo supplies conductivity evaluations and factorization machine quadratic optimization annealing supplies the black-box search over dopant configurations, with explicit partitioning of the configuration space to enable parallel optimization.

If this is right

The same partitioned optimization loop can be applied to dopant design in other solid-electrolyte chemistries.
When experimental conductivity data are available, the framework can be run in reverse to recover candidate microscopic structures consistent with the measurements.
The approach directly addresses the combinatorial explosion that otherwise prevents exhaustive search of dopant placements in large KMC cells.
Iterative surrogate learning reduces the number of full KMC evaluations required to reach high-conductivity configurations.

Where Pith is reading between the lines

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

The method supplies a practical route to test whether particular local dopant orderings control long-range ionic transport, a question left open by purely random or average-structure models.
If the identified configuration can be realized experimentally, it offers a concrete benchmark for validating or refining the underlying KMC hopping barriers.
The surrogate model built during optimization may later serve as a fast predictor for conductivity in related compositions without repeated full simulations.

Load-bearing premise

The kinetic Monte Carlo model used to evaluate conductivity for each dopant configuration accurately captures the real ionic transport mechanisms in the material.

What would settle it

Direct experimental measurement of ionic conductivity on a sample prepared with the specific dopant configuration identified by the method, checking whether the measured value is approximately ten times higher than both random configurations and the previously reported experimental value.

read the original abstract

Maximizing ionic conductivity remains a fundamental challenge in the atomistic design of solid electrolytes. To this end, we present a framework that combines kinetic Monte Carlo (KMC) and factorization machine with quadratic-optimization annealing (FMQA), an Ising-machine-based black-box optimization algorithm for large-scale combinatorial optimization. KMC evaluates the ionic conductivity for a given dopant configuration, whereas FMQA iteratively learns a surrogate model from a small configuration-conductivity dataset and proposes configurations expected to maximize conductivity. To address the severe combinatorial explosion in large KMC simulation cells, we partition the configuration space for parallel optimization. As a proof of concept, we apply this KMC-FMQA framework to bulk 8 mol % yttria-stabilized zirconia, identifying a dopant configuration with an order-of-magnitude higher conductivity than that of random configurations and the experimentally reported conductivity. Combined with experiments, this framework will enable the determination of microscopic structures from measured conductivity, providing insight into the underlying transport mechanisms.

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 pairs KMC conductivity scoring with FMQA Ising optimization plus partitioning to search dopant configs in 8YSZ and reports a 10x gain, but the result hinges on an unbenchmarked KMC model.

read the letter

The core contribution is a practical loop that uses kinetic Monte Carlo to score ionic conductivity for each dopant arrangement, then feeds those scores into FMQA to propose better arrangements, with a spatial partition step to keep the search tractable in large cells. They demonstrate it on 8 mol % YSZ and claim one configuration beats both random doping and the experimental benchmark by roughly an order of magnitude.

What works is the engineering of the search: partitioning the configuration space lets them run parallel FMQA instances without exploding the combinatorial cost, and the abstract makes clear that the optimizer learns from actual KMC evaluations rather than a purely fitted surrogate. That combination is new enough to be worth noting for people already using Ising machines or KMC in materials design.

The main soft spot is the missing validation step. The stress-test note is right: the abstract gives no evidence that the same KMC setup reproduces the known experimental conductivity of random 8YSZ within a factor of two or so. If the absolute scale of the hop rates or vacancy interactions is off, the reported 10x improvement over experiment is just a model artifact. Without that check, or at least a clear statement of how the energy barriers were obtained and tested, the central claim stays provisional.

This is for computational materials groups already running KMC or looking for black-box optimizers for electrolyte design. A reader who wants a ready-to-use inverse-design script will get the most out of it once the validation details are filled in. The work shows clear thinking on the optimization side and honest engagement with the combinatorial problem, so it deserves a serious referee to check the KMC implementation and any internal benchmarks that may exist in the full text.

I would send it to review rather than desk-reject; the method is concrete and the partitioning trick is useful even if the YSZ result needs more grounding.

Referee Report

1 major / 0 minor

Summary. The manuscript introduces a KMC-FMQA framework that couples kinetic Monte Carlo conductivity evaluations with factorization-machine quadratic annealing (an Ising-machine optimizer) to perform inverse design of dopant configurations in 8 mol% YSZ. The central claim is that the method identifies a dopant arrangement whose KMC-evaluated ionic conductivity is an order of magnitude higher than both random configurations and the experimentally reported value for 8YSZ.

Significance. If the KMC model is shown to be quantitatively accurate, the approach would provide a practical route to combinatorial optimization over large dopant-configuration spaces in solid electrolytes, potentially enabling data-driven discovery of high-conductivity microstructures. The partitioning strategy for parallel optimization and the use of a learned surrogate are technically relevant to scaling KMC-based design.

major comments (1)

[Abstract] Abstract (and the KMC evaluation step described therein): the claim of an order-of-magnitude conductivity improvement over experiment rests on the absolute accuracy of the KMC hop-rate model. No benchmark is supplied showing that the same KMC reproduces the known experimental conductivity of random 8 mol% YSZ within a factor of ~2 (or any stated tolerance). Without this calibration, the reported gain relative to experiment cannot be distinguished from a possible offset in the surrogate energy landscape.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback and positive view of the KMC-FMQA approach. We address the single major comment below and will revise the manuscript accordingly.

read point-by-point responses

Referee: [Abstract] Abstract (and the KMC evaluation step described therein): the claim of an order-of-magnitude conductivity improvement over experiment rests on the absolute accuracy of the KMC hop-rate model. No benchmark is supplied showing that the same KMC reproduces the known experimental conductivity of random 8 mol% YSZ within a factor of ~2 (or any stated tolerance). Without this calibration, the reported gain relative to experiment cannot be distinguished from a possible offset in the surrogate energy landscape.

Authors: We agree that the manuscript would benefit from an explicit benchmark of the KMC hop-rate model against the experimental conductivity of random 8 mol% YSZ. The KMC implementation follows standard literature parameters for yttria-stabilized zirconia, but no direct comparison to the experimental value is provided in the current version. In the revised manuscript, we will add a section or paragraph in the methods or results detailing the KMC-predicted conductivity for random configurations and compare it to experiment, including any scaling factors or notes on model limitations. This will allow readers to assess the absolute scale of the reported improvements. The relative 10x gain over random configurations remains robust as it is evaluated consistently within the same model. revision: yes

Circularity Check

0 steps flagged

No circularity: KMC evaluations are independent inputs to the FMQA optimizer

full rationale

The paper's core loop uses KMC to compute conductivity for each dopant configuration as an external black-box evaluator, then feeds those values into FMQA to learn a surrogate and propose new configurations. This is a standard surrogate-assisted search; the final reported conductivity for the optimized configuration is the direct KMC output on a searched point, not a redefinition or statistical artifact of the inputs. No equations, self-citations, or ansatzes in the provided abstract reduce the conductivity gain to a fitted parameter or prior result by construction. The framework is therefore self-contained against external benchmarks (KMC runs), consistent with a score of 0.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only; no explicit free parameters, axioms, or invented entities are stated. The framework implicitly assumes KMC conductivity calculations are reliable and that FMQA surrogate learning generalizes in the partitioned space.

pith-pipeline@v0.9.1-grok · 5706 in / 1048 out tokens · 33931 ms · 2026-07-01T05:03:11.556009+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

1 extracted references · 1 canonical work pages

[1]

Linking Microscopic Structures to Ionic Currents in Solid Electrolytes via Kinetic Monte Carlo and Regression Analysis

(1) Koizumi, A.; Ivonina, M.; Tamura, R.; Tada, T. Linking Microscopic Structures to Ionic Currents in Solid Electrolytes via Kinetic Monte Carlo and Regression Analysis. J. Phys. Chem. C 2026, 130 (10), 3897–3908. https://doi.org/10.1021/acs.jpcc.5c06852. (2) Yamahara, K.; Jacobson, C. P.; Visco, S. J.; De Jonghe, L. C. Influence of Powders on Ionic Cond...

work page doi:10.1021/acs.jpcc.5c06852 2026

[1] [1]

Linking Microscopic Structures to Ionic Currents in Solid Electrolytes via Kinetic Monte Carlo and Regression Analysis

(1) Koizumi, A.; Ivonina, M.; Tamura, R.; Tada, T. Linking Microscopic Structures to Ionic Currents in Solid Electrolytes via Kinetic Monte Carlo and Regression Analysis. J. Phys. Chem. C 2026, 130 (10), 3897–3908. https://doi.org/10.1021/acs.jpcc.5c06852. (2) Yamahara, K.; Jacobson, C. P.; Visco, S. J.; De Jonghe, L. C. Influence of Powders on Ionic Cond...

work page doi:10.1021/acs.jpcc.5c06852 2026