Local Structure Dictates Ionic Transport and Mechanical Properties in Glassy Solid Electrolytes for Lithium Batteries

Kasper Tolborg; Morten M. Smedskjaer; Qi Zhang; Rasmus Christensen; Tao Du; Timoth\'ee Jamin; Xiaoyi Xu; Yong Li; Yuanzheng Yue; Zhencai Li

REVIEW 2 major objections 7 minor 2 references

Local Structure Dictates Ionic Transport and Mechanical Properties in Glassy Solid Electrolytes for Lithium Batteries

T0 review · 2 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A moderate dose of P2S5 simultaneously enhances lithium-ion conduction and nano-ductility in a glassy sulfide electrolyte, while excess P2S5 re-polymerizes the network and blocks transport.

desk verdict Solid computational study with a plausible coupled transport-mechanics story; the mechanics half needs DFT spot-checks at high strain before the central claim is fully established. read the letter →

arxiv 2608.06895 v1 pith:TYMIK2U6 submitted 2026-08-07 cond-mat.mtrl-sci

Yong Li , Tao Du , Rasmus Christensen , Timothée Jamin , Zhencai Li , Qi Zhang , Xiaoyi Xu , Kasper Tolborg

show 2 more authors

Yuanzheng Yue Morten M. Smedskjaer

This is my paper · ORCID

classification cond-mat.mtrl-sci

keywords glassysolidelectrolyteslithium-iontransportbrittle-to-ductiletransitionmachine-learnedinteratomicpotentialthioborate-thiophosphatenetworksall-solid-statebatteriesmoleculardynamicsnetworktopology

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper uses machine-learned molecular dynamics to explain why small amounts of P2S5 improve lithium-ion conduction in glassy sulfide electrolytes while larger amounts degrade it, and why the same additions make the glass less brittle. It argues that moderately added P2S5 forms isolated PS4 units that fragment the rigid boron-sulfur network, opening fast diffusion paths for Li$^+$ and introducing flexible P-S-P linkages that dissipate strain instead of fracturing. Excessive P2S5 forms condensed P2S6 and P2S7 polyphosphates that reconnect the network, raising the activation energy for lithium hopping. The payoff is a design rule: intermediate compositions balance high conductivity against mechanical robustness for all-solid-state batteries.

What carries the argument

The load-bearing object is the compositional speciation of the glass network, quantified by the $Q^n$ distribution of PS4 tetrahedra, where n is the number of bridging sulfur atoms around each tetrahedron, and by the fractions of isolated PS4, meta P2S6, hypo P2S6, and P2S7 units. The paper uses this speciation as a single structural order parameter: low P2S5 content maximizes isolated PS4 and fragments the B-S framework, while high P2S5 content shifts the balance toward P2S6/P2S7 chain and dimer species that re-polymerize it. This same order parameter is claimed to control both the percolation of lithium diffusion pathways and the switch from brittle crack localization to nano-ductile energy dissipation.

What would settle it

Take the most strained snapshots from the tensile simulations, at 20 to 50 percent strain, and recompute their atomic forces with direct density functional theory; if the machine-learned force errors are large there, the predicted brittle-to-ductile ranking is not supported. A complementary experiment would be to measure crack or indentation response in real LSPBI glasses from a5 to a100 and look for the predicted monotonic shift from localized fracture to distributed ductile flow.

Watch

Extended reading notes

Core claim

The central claim is that in glassy Li-S-P-B-I electrolytes, the addition of P2S5 induces a critical structural transformation that couples ionic transport and mechanical response. At moderate P2S5 content, incorporated PS4 tetrahedra depolymerize the rigid boron-sulfur framework, creating percolative diffusion pathways for Li$^+$; at the same time, flexible P-S-P configurations allow strain energy to be dissipated through bond bending and torsion, producing a brittle-to-ductile transition. At excessive P2S5 content, condensed polyphosphate species such as P2S6 and P2S7 re-polymerize the network and raise the activation barrier for lithium hopping. The authors conclude that a single axis, network fragmentation versus re-polymerization, governs both properties, and that intermediate compositions balance peak conductivity against mechanical robustness.

Load-bearing premise

The entire mechanical conclusion rests on the assumption that the machine-learned potential, trained on high-temperature melt configurations and near-equilibrium structures, remains accurate when bonds are stretched to 50% strain during tensile fracture.

Editorial extensions

If this is right

At low P2S5 content, around a3 to a5 in the studied series, isolated PS4 units fragment the B-S network, lowering the Li$^+$ activation barrier and increasing diffusivity.
At higher content, a10 and above, condensed P2S6 and P2S7 species re-polymerize the network, raising the activation energy and reducing long-range lithium motion.
Tensile strength decreases monotonically with P2S5 content because P-S bonds are weaker than B-S bonds, while failure shifts from localized brittle fracture to distributed nano-ductile flow.
Intermediate compositions, a25 to a50, should deliver usable conductivity together with improved ductility, providing a practical compromise for all-solid-state batteries.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The authors do not simulate Li$^+$ transport beyond a10, so their recommendation that a25 to a50 balances conductivity and ductility is an extrapolation; extending the diffusion analysis to those compositions would directly test the claimed sweet spot.
The energy-dissipation mechanism is argued by analogy to related sodium thioborate glasses rather than measured here; tracking P-S-P bond-angle distributions during tensile loading would provide a direct test.
If the fragmentation mechanism is general, other network-former substitutions that yield isolated tetrahedral units without forming condensed dimers should also raise conductivity while preserving stiffness, a design rule that could be screened computationally before synthesis.
The simulations predict a monotonic trend in fracture behavior across compositions that could be checked experimentally by indentation or fracture-toughness tests on melt-quenched LSPBI glasses.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The short version: this is a competent, clearly written computational study that gives the first atomistic mechanical picture of the Li-S-P-B-I glass family, and it is honest about its own discrepancies. The caveat that matters: the brittle-to-ductile conclusion rests on neural-network forces at 50% strain, far outside the training data, and the proposed bond-bending dissipation mechanism is argued by analogy rather than measured in the trajectories.

What is actually new: the composition-series structural quantification (Qn evolution, P2S6/P2S7 speciation, BS3/BS4 ratios) and the first mechanical characterization of LSPBI across a0-a100, including the brittle-to-ductile transition and the stress-strain ranking. The non-monotonic Li conductivity trend was already reported experimentally (ref 16) and a5 diffusion was simulated before (ref 17), but the authors use those as validation anchors rather than overselling them. The MLIP is trained with a sensible protocol, checked against independent AIMD at 600 and 1000 K, experimental density (2.28 vs 2.3 g/cm3) and a literature S(Q). Size convergence (beyond 20,000 atoms) and strain-rate robustness are checked, and code and potential are public.

The soft spots, in proportion. First and most important: the tensile simulations run to 50% strain, where bonds break and reform. The validation set covers near-equilibrium configurations, with no DFT reference reported for any deformed snapshot from the tensile runs, and the ZBL repulsion is removed after training stage one. The brittle-to-ductile transition and the composition ranking at large strain therefore depend on unverified extrapolation. Two or three DFT single-point checks on representative strained configurations would settle this. Second, the energy-dissipation mechanism via P-S-P bond bending is inferred from analogy to Na2S-P2S5-B2S3 glasses; the paper does not report S-P-S or P-S-P bond-angle distributions under load, so it is a plausible interpretation, not a measured one. Third, the diffusion analysis uses single-trajectory MSD slopes without error bars. Given the disclosed a3-versus-a5 offset, that is a minor issue, but error bars would help separate trend from noise. And 'percolative pathways' is asserted without a percolation analysis; a simple ring or percolation statistic would make it concrete.

None of this kills the paper. The central mechanism—network fragmentation by isolated PS4 units helps Li+ mobility and flexibility, while P2S6/P2S7 re-polymerization impedes both—is consistent with the structural data and the experimental anchor. But the mechanical half is not yet established at the level the abstract implies. The paper deserves a serious referee; the review should ask for the strained-configuration DFT checks and the bond-angle statistics. With those additions, the claims would be much stronger. I would send it to peer review rather than desk reject, and I would cite the structural series and the MLIP validation if I worked in this area.

Referee Report

2 major / 7 minor

Summary. The paper uses DeePMD machine-learning interatomic potentials to simulate a series of Li-S-P-B-I (LSPBI) glassy electrolytes of composition 30Li2S-25B2S3-45LiI-aP2S5. It reports that moderate P2S5 addition (a3/a5) fragments the B-S network through isolated PS4 units, enhancing Li+ mobility, while higher P2S5 content forms condensed P2S6/P2S7 species that re-polymerize the network and suppress transport. Parallel tensile simulations across a0-a100 are interpreted as evidence of a composition-driven brittle-to-ductile transition, attributed to flexible P-S-P configurations that dissipate energy by bond bending. The authors conclude that network fragmentation versus re-polymerization is the single structural order parameter linking ionic conductivity and mechanical behavior, and that intermediate compositions (a25-a50) provide a practical balance.

Significance. If the central claim survives scrutiny, the paper would establish a unified structural mechanism governing both ionic transport and mechanical response in a practically relevant glassy electrolyte family, with immediate design implications for all-solid-state batteries. The transport half is credibly supported: the MLIP is validated against independent AIMD trajectories, the simulated density (2.28 g/cm3) matches experiment (2.3 g/cm3) for a5, the neutron-weighted structure factor agrees with literature data for 75Li2S-25P2S5, and no target quantity is fitted (conductivities come from MSD slopes and Arrhenius fits). The mechanical half, however, rests on MLIP predictions in a strongly deformed regime that is not covered by the validation data, and the proposed bond-bending dissipation mechanism is asserted by analogy rather than measured. The paper is otherwise careful in reporting size and strain-rate checks, and the trained potential and scripts are made publicly available, which strengthen its reproducibility.

major comments (2)

[§4.3/Table S1 and §4.8/Figure 4] The brittle-to-ductile transition is not validated in the regime where it is claimed. Section 4.3 trains the DeePMD potential on AIMD data at 3000 K and on 300-1000 K equilibrium-like configurations (Table S1), and the ZBL short-range repulsion is removed after the first training stage. Section 4.8 then stretches the glasses to 50% strain, where bonds are broken and reformed and local environments contain stretched pairs and large voids. Supporting Figure S2 validates only 600 K and 1000 K equilibrium states. I request DFT single-point energy and force checks on representative configurations drawn from the tensile trajectories at strains of roughly 0.3-0.5 for at least a5 and a100, or retraining with such configurations; without this, the mechanical half of the central claim rests on extrapolation of the MLIP outside its validated domain.
[§2.4] The energy-dissipation mechanism is not directly evidenced. The text states that 'flexible P-S-P configurations enable energy dissipation through bond bending,' but this is argued by analogy to Na2S-P2S5-B2S3 glasses (refs 40, 43), and no S-P-S or P-S-P bond-angle distributions are reported from the present tensile trajectories. Please compute bond-angle statistics as a function of strain and composition, and ideally quantify the relative contributions of bond stretching versus bond-angle bending to the dissipated work. This would directly test the proposed structural origin of the brittle-to-ductile transition.

minor comments (7)

[§2.3/Figure 3c] The simulated activation-energy minimum occurs at a3 rather than a5, while the experimental conductivity peak is at a5. The paper acknowledges this but does not quantify the uncertainty on the fitted Ea values or explain the offset structurally. Please add error bars or a statistical comparison, and comment on whether a3 and a5 are distinguishable within the simulation.
[Figure 1] The caption lists panels (d-f), but the text refers to 'Figure 1e' when discussing the structure factor S(Q), which is panel (f). Please correct the cross-reference.
[Table S1 vs §4.3] Table S1 lists LSPBI amorphous structures at 300 K in the training dataset, whereas Section 4.3 states that LSPBI AIMD data at 600, 1000, and 3000 K were added to the training set and that 300 K data appear in the validation set. Please resolve this inconsistency.
[§4.8] The lateral dimensions are held fixed during loading, which is uniaxial strain, not uniaxial stress. The phrase 'uniaxial tensile deformation' is imprecise; please state the strain-controlled condition explicitly in the main text.
[Figure 4c] The stress-strain curves are stated to be averages over three MD simulations, but no standard deviations or error bands are shown. Please report the variability or state that it is smaller than the line width.
[§2.2] The terms 'P2S6 meta' and 'P2S6 hypo' are used in Figure 2c but not defined at first use. Please define these species and explain how they differ structurally.
[General] There are several typographical errors: 'xhibits' in Section 2.3, 'the the' in Section 2.1, and 'Current Wrok' in Figure S1 of the Supporting Information.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: transport and mechanical predictions emerge from MD/MSD analysis, not from fitted inputs.

full rationale

The paper's central derivation chain is not circular. Lithium conductivities are obtained from mean-square-displacement slopes and Arrhenius fits rather than from any parameter fitted to the experimental conductivity maximum, and the disclosed a3-versus-a5 mismatch shows the simulation did not tune itself to the reported peak composition. The structural descriptors (Qn distributions, P2S6/P2S7 fractions) are counted from simulated configurations and then correlated with transport and stress-strain behavior; they are not used to define the transport outcome. The DeePMD potential is a machine-learned surrogate for DFT, and while the training set includes AIMD trajectories for the same LSPBI compositions, the predicted properties are emergent outputs of MD, not regression targets. Validation is provided by independent AIMD trajectories (Supporting Figure S2), external experimental density and S(Q) benchmarks, and the experimental conductivity trend from ref 16. The self-citation (ref 31) is used only to support the generic non-Arrhenius behavior and as an S(Q) comparison reference; it is not load-bearing for the coupled fragmentation/ductility mechanism. The brittle-to-ductile conclusion relies on DeePMD forces at 50% strain, a regime lacking explicit DFT spot-checks, but that is an extrapolation/correctness risk rather than circularity, since no quantity was fit to force that outcome. No step in the derivation reduces by construction to its own inputs.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities; its contributions are structural descriptors and a trained MLIP. The ledger captures the simulation-protocol choices and the two extrapolation bets, MLIP accuracy at high strain and melt-quench representativeness, on which the conclusions rest. The Qn counting cutoffs are a further unquantified analysis choice.

free parameters (4)

Melt-quench cooling rate = 2.5 K/ps
Protocol choice in Section 4.4 for producing all glasses; faster than experimental cooling and can bias network topology, although density and S(Q) match experiment for representative compositions.
Low-temperature Arrhenius fitting window and point set = 325-375 K; 3 points for some compositions, 5 for others
Section 4.7 selects temperatures per composition to stay below the ~400 K dynamic crossover; the choice influences the reported Ea values and the position of the Ea minimum (a3 versus experimental a5).
Tensile strain rate = 5e9 /s
Adopted in Section 4.8 after a rate survey; stress magnitudes depend on rate, though composition ranking is reported stable, so the brittle-to-ductile ordering is the robust output.
Coordination cutoffs for Qn and bridging-sulfur counting = not reported numerically
The structural statistics in Section 2.2 (Qn distributions, P2S6/P2S7 fractions) depend on the P-S and B-S distance cutoffs, which are described but not given as values in the paper.

assumptions (4)

domain assumption PBE-D3 DFT provides adequate reference energies and forces for Li-S-P-B-I glasses
All MLIP training data come from CP2K PBE-D3 AIMD (Section 4.2); no benchmark against higher-level methods or experimental thermodynamics is given.
domain assumption The DeePMD potential remains accurate for 300 K dynamics and for tensile strains up to 0.5
Sections 4.3 and 4.8; validation covers equilibrium structure and forces, while deformed states are outside the training set.
domain assumption Melt-quench MD at 2.5 K/ps produces glass structures representative of experimental LSPBI glasses
Section 4.4; validated by density and S(Q) agreement for a5 and 75Li2S-25P2S5, respectively; transferability to all compositions is assumed.
domain assumption Uniaxial strain loading (fixed lateral dimensions) is a suitable comparative measure of fracture behavior
Section 4.8; the imposed boundary condition prevents Poisson contraction; accepted for ranking compositions, not for absolute fracture toughness.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Local Structure Dictates Ionic Transport and Mechanical Properties in Glassy Solid Electrolytes for Lithium Batteries." pith.science (2026). https://pith.science/paper/TYMIK2U6

@misc{pith2026260806895,
  author       = {Pith},
  title        = {Pith review of: Local Structure Dictates Ionic Transport and Mechanical Properties in Glassy Solid Electrolytes for Lithium Batteries},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TYMIK2U6}},
  note         = {Machine review of arXiv:2608.06895}
}

read the original abstract

Electrolytes composed of sulfide and halide glasses are promising candidates for all-solid-state lithium batteries owing to their processability, lack of grain boundaries, and relatively high ionic conductivity. Nevertheless, their ionic conductivity and mechanical properties are still not satisfying for the real-world applications. Significant advances in solid electrolytes require a thorough understanding of their microstructures. Here, we reveal the connections among structure, ionic transport properties, and mechanical stability in a series of glassy solid electrolytes by employing molecular dynamics simulations based on a machine learning interatomic potential. Specifically, we explore how the interplay between B-S and P-S networks in glassy Li-S-P-B-I (LSPBI) governs ionic conductivity and deformation behavior. The introduction of P2S5 into a B2S3-based glass induces a critical structural transformation, through which both ionic conductivity and mechanical nano-ductility can be enhanced. For a moderate P2S5 content, incorporated PS4 units depolymerize the rigid boron framework, creating percolative diffusion pathways for fast ionic transport. Concurrently, the flexible P-S-P configurations enable energy dissipation through bond bending, leading to the brittle-to-ductile transition. However, excessive P2S5 increases the fraction of polyphosphates (e.g., P2S6 and P2S7), thereby polymerizing the structural network and ultimately impeding Li+ mobility. Our work thus provides atomistic principles for engineering glass electrolytes with balanced ionic conductivity and mechanical robustness.

Figures

Figures reproduced from arXiv: 2608.06895 by the authors.

**Figure 1.** Performance evaluation of trained MLIP for LSPBI‑a5 glassy solid electrolyte. (a) Atomic configuration of generated glass structure from AIMD. (b-c) Parity plots of atomic forces and energies predicted by DeePMD‑kit versus DFT reference. (d-f) Structural properties at 300 K derived from MD simulations using present DeePMD potential, as well as reference MACE‑Omat-0 and MACE‑MP-0a potentials, benchmarked against AIMD… view at source ↗

**Figure 2.** Composition-dependent structural evolution in LSPBI glassy electrolyte. (a-b) Qn distribution of (a) PS4 and (b) BS4 units in glassy LSPBI electrolytes with compositions a0, a5, a10, a25, a50, a75, and a100. (c) Fractions of P2S7, P2S6 meta, P2S6 hypo, and PS4 units in glassy LSPBI electrolytes with compositions a0, a5, a10, a25, a50, a75, and a100. Next, we analyze the compositional evolution of BS4 tetrahedra. In … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages

[1]

(1) Chen, Z.; Du, T.; Krishnan, N. M. A.; Yue, Y .; Smedskjaer, M. M. Disorder -induced enhancement of lithium-ion transport in solid-state electrolytes. Nature Communications 2025, 16 (1),

work page 2025
[1057]

Structural and electronic features of binary Li(2)S- P(2)S(5) glasses

(2) Ohara, K.; Mitsui, A.; Mori, M.; Onodera, Y .; Shiotani, S.; Koyama, Y .; Orikasa, Y .; Murakami, M.; Shimoda, K.; Mori, K.; et al. Structural and electronic features of binary Li(2)S- P(2)S(5) glasses. Scientific Reports 2016, 6, 21302

work page 2016

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

(1) Chen, Z.; Du, T.; Krishnan, N. M. A.; Yue, Y .; Smedskjaer, M. M. Disorder -induced enhancement of lithium-ion transport in solid-state electrolytes. Nature Communications 2025, 16 (1),

work page 2025

[2] [1057]

Structural and electronic features of binary Li(2)S- P(2)S(5) glasses

(2) Ohara, K.; Mitsui, A.; Mori, M.; Onodera, Y .; Shiotani, S.; Koyama, Y .; Orikasa, Y .; Murakami, M.; Shimoda, K.; Mori, K.; et al. Structural and electronic features of binary Li(2)S- P(2)S(5) glasses. Scientific Reports 2016, 6, 21302

work page 2016