{"id":"cabf1305-b1ab-45f9-b3d2-0e0274793c3f","arxiv_id":"2608.06902","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Vitrifying ZIF-4 and ZIF-62 drops the simulated Li-ion migration barrier from about 0.35 eV to 0.16 eV and switches anisotropic cage-hopping transport into faster, isotropic Fickian diffusion.","lead":"Simulations show that turning crystalline ZIF frameworks into glasses lowers the energy barrier for lithium ions to move and makes their motion equally easy in all directions. This suggests MOF glasses could be designed as solid battery electrolytes that combine fast ion transport with mechanical robustness.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The glassy-state Li+ diffusion results, especially Ea = 0.16 eV and the order-of-magnitude room-temperature enhancement, rely on a DeePMD potential whose Li validation is limited to crystalline ZIF-62; the central claim is therefore conditional on MLIP transferability to amorphous Li environments.","rationale":"The reader's weakest-assumption analysis identified MLIP transferability to amorphous Li environments as the pivotal issue, and my stress-test concurs. The paper is otherwise well-executed: the glass densities match experiment, the structural descriptors are internally consistent, the replica-averaged MSD protocol addresses statistical sampling, and the GitHub release of workflows and the trained potential enables independent checks. Those are real strengths and support the qualitative picture of disorder-broadened barriers and isotropization. However, the quantitative headline numbers—0.16 eV Ea and the specific room-temperature D values—are only as secure as the MLIP's ability to describe Li in the glass, and the only direct Li-in-ZIF validation is crystalline. I also note an internal-consistency warning the authors themselves provide: at 700 K the crystalline diffusivity exceeds the glassy one, so the low-temperature 'disorder unlocks fast diffusion' claim depends on the precise Arrhenius slope crossing. If the MLIP underestimates the glass barriers by even a few tens of meV, or if the 300 K crystalline MSD is subdiffusive and mis-extracted, the ordering could reverse or shrink. These are testable, and the proposed DFT NEB check on glass hops would settle the matter. I do not see grounds to reject the paper; the appropriate verdict is conditional, matching the reader's conclusion.","tokens_in":17598,"tokens_out":3262,"duration_ms":39215,"concrete_test":"Take 10 representative Li+ hop events from the ZIF-62 glass MLIP trajectories at 300-400 K (selected from the broad Gs(r,t) population in Figure 2d), and recompute their minimum-energy pathways with DFT NEB using the same neutral-framework, formal +1 Li treatment as in Section 4.1. If the mean absolute deviation between the MLIP and DFT migration barriers exceeds ~0.05 eV, the 0.16 eV glass activation energy and the room-temperature enhancement are not yet supported; if it is below that, the MLIP transferability concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that structural disorder lowers Ea from ~0.35 eV to 0.16 eV and increases room-temperature D by ~17x (ZIF-4) and ~7x (ZIF-62), with isotropization in the glass. The entire quantitative basis is the fine-tuned DeePMD potential of Section 4.1. Its validation is explicitly a 'direct Li+ insertion model' inside the ZIF frameworks, but the reported parity plot (SI Figure S13) and RMSEs (energy 5.51e-3 eV/atom, force 1.12e-1 eV/A) are for Li+ in crystalline ZIF-62 only. No DFT benchmark is provided for Li+ in the amorphous ZIF-4/ZIF-62 glasses whose 0.16 eV Ea and Fickian isotropic diffusion are the headline results. The force RMSE is not obviously negligible for a 0.16 eV barrier: even a 0.05 eV systematic error in local Li-framework interactions could change the glass Ea by ~30% and materially alter the room-temperature crossover between crystal and glass, which Figure 2a shows occurs between 300 and 700 K. The concern is sharpened by the 300 K crystalline NGP behavior: the authors state that NGP 'increases continuously at 300 K within the simulated time window' (Section 2.3), meaning the crystalline 300 K MSD may not have reached the linear diffusive regime used to extract D, so the Arrhenius fit that separates crystal and glass behavior is particularly sensitive to how non-diffusive short-time data are treated. Together, the missing amorphous-phase DFT check and the questionable 300 K crystalline D extraction make the central claim not fully established, though not contradicted.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses a DeePMD machine-learned interatomic potential, fine-tuned to include Li interactions, to simulate Li+ transport in crystalline and glassy ZIF-4 and ZIF-62. The central claim is that vitrification lowers the Li+ migration activation energy from about 0.35 eV in the crystals to 0.16 eV in the glasses, increases the extrapolated room-temperature diffusion coefficient by more than an order of magnitude for ZIF-4 and by nearly sevenfold for ZIF-62, and makes Li+ diffusion isotropic or near-isotropic by randomizing imidazolate/benzimidazolate ring orientations. The authors support the mechanistic picture with non-Gaussian parameter analysis, van Hove correlation functions, diffusion-tensor ellipsoids, and ring-orientation pole figures. The manuscript also reports a replica-based sampling strategy (20 crystalline and 40 glassy replicas) and makes the workflow and trained potential available on GitHub.","tokens_in":18021,"tokens_out":4799,"duration_ms":49523,"significance":"If the findings hold, the paper offers a concrete and falsifiable design principle for glassy MOF electrolytes: structural disorder lowers migration barriers and isotropizes Li+ transport. The two-system consistency (ZIF-4 and ZIF-62) and the combination of diffusion-tensor analysis with linker-orientation descriptors are strengths, as is the reproducible MLIP workflow with documented replica sampling. However, the quantitative central claim depends on the transferability of the fine-tuned Li potential to amorphous Li+ environments, which is not directly validated, and on room-temperature crystalline diffusion coefficients whose 300 K simulations are not clearly in the diffusive regime. The result is therefore plausible but conditional in its current form.","major_comments":[{"comment":"The fine-tuned DeePMD potential is validated against DFT on a direct Li+ insertion model, but the parity plot and RMSEs (energy 5.51×10^-3 eV/atom, force 1.12×10^-1 eV/Å) are for Li+ in crystalline ZIF-62 only. The headline results, Ea ≈ 0.16 eV in the glasses and the room-temperature D enhancement, are computed in glassy ZIF-4 and ZIF-62, where the local Li+ coordination environments are not represented in the validation set. Because the reported force RMSE is comparable to the energy scale of the 0.16 eV barrier, a systematic error of even about 0.05 eV in Li-framework interactions in the amorphous phase could materially change the glass Ea and the crystal/glass crossover in Figure 2a. Please add DFT benchmarks, at minimum energies and forces for Li+ in the melt-quenched glass structures, and ideally short AIMD trajectories in the glass, or otherwise demonstrate transferability of the Li-related terms to disordered environments.","section":"§4.1, SI Fig. S13"},{"comment":"The room-temperature diffusion coefficients for the crystalline phases are not established at the same level as the glassy values. Section 2.3 states that in the crystalline phase the NGP increases continuously at 300 K within the simulated time window, which indicates that the 300 K MSD has not reached the linear diffusive regime; nevertheless, the Arrhenius fits in Figure 2a appear to include the 300 K point, and the reported room-temperature D values (1.91×10^-8 and 1.12×10^-8 cm2/s) are then used to quantify the order-of-magnitude and sevenfold enhancements. Please either exclude the non-diffusive 300 K crystalline points from the Arrhenius fit, run longer trajectories to reach the diffusive regime at 300 K, or provide a sensitivity analysis showing that the enhancement factors are robust to the treatment of this point.","section":"§2.3, Fig. 2a"},{"comment":"No statistical uncertainties are reported for D, Ea, or D0. Given that the central claims are quantitative (barrier reduction from about 0.35 to 0.16 eV and an order-of-magnitude D increase), the absence of error bars leaves the significance of the crystal/glass differences unquantified. The replica-based sampling is in principle sufficient to report standard errors; please provide them for D(T), the fitted Ea, and the derived room-temperature values.","section":"§4.5, Eq. (7)"}],"minor_comments":[{"comment":"The phrase 'an grainboundary-free' should read 'a grain-boundary-free'.","section":"Abstract"},{"comment":"The reference to 'Supporting Figure S12' for the DeePMD parity plot should be 'Supporting Figure S13'.","section":"§4.1"},{"comment":"In the discussion of Eq. (17), the abbreviation 'NPG' should be 'NGP' to match the notation used throughout the paper.","section":"§4.5"},{"comment":"The caption lists 'MSDxy, MSDxy and MSDyz'; the second term should be 'MSDxz'.","section":"Supporting Information, Fig. S10 caption"},{"comment":"The force RMSE unit appears as 'ev/Å' and should be 'eV/Å'.","section":"§4.1"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this one if you care about MLIP predictions for amorphous electrolytes. The headline: the paper claims vitrification of ZIF-4/ZIF-62 cuts Li+ migration activation energy from ~0.35 eV to 0.16 eV, raises extrapolated room-temperature D by 7–17x, and isotropizes transport. That is genuinely new for ZIF glasses, as far as the cited record goes, and the mechanistic story is coherent: ordered ring orientations impose direction-dependent barriers and hopping dynamics, while glassy disorder broadens the barrier distribution and makes diffusion continuous and Fickian.\n\nWhat is done well: the study uses a fine-tuned DeePMD potential, ships the code and data, and validates the glass structures against experimental densities and PDFs. The replica-based MSD sampling (20–40 independent runs) is a reasonable answer to the problem of having only 10 Li ions per cell. The diffusion tensor and pole-figure analysis is a nice, direct way to tie ring orientation to anisotropic transport. The authors also deserve credit for labeling the room-temperature crystal D as extrapolated rather than simulation-direct.\n\nThe soft spots are real but not disqualifying. First, the MLIP validation set is Li+ in crystalline ZIF-62 only; the glass-phase results are the headline, and there is no direct DFT or experimental check of Li in the amorphous state. With a force RMSE of ~0.11 eV/Å, a 0.16 eV barrier could shift by tens of percent. Second, at 300 K the crystalline MSDs are still in the non-Gaussian, non-diffusive regime, so the extrapolated crystal D and hence the size of the glass enhancement carry more uncertainty than the glass numbers. There are no error bars on D or Ea anywhere. I would call these caveats rather than fatal flaws: the qualitative glass-vs-crystal trend appears in two systems and is supported by several independent descriptors.\n\nThe citation pattern is fine; building on the group's own prior DeePMD potential is normal, and the self-citations are load-bearing, not padding.\n\nWho gets value: anyone working on MOF-glass electrolytes, or on MLIP transferability to amorphous phases. It deserves a serious referee. I would send it to review, with the request that the authors validate the MLIP against a DFT-computed Li insertion or migration barrier in the glass structure and report uncertainties on the Arrhenius parameters.","headline":"A plausible MLIP-MD study claiming vitrification lowers Li+ barriers and isotropizes transport in ZIF glasses, but the key numbers rest on unvalidated amorphous-phase potential transferability and Arrhenius extrapolation.","tokens_in":18568,"tokens_out":2584,"would_cite":true,"duration_ms":26591,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Structural disorder from vitrification lowers the lithium migration barrier in ZIF glasses to 0.16 eV and makes diffusion isotropic.","keywords":["zeolitic imidazolate framework glass","solid-state battery electrolyte","lithium-ion diffusion","machine-learning interatomic potential","activation energy","isotropic ion transport","melt-quenched metal-organic framework"],"falsifier":"Measure the temperature-dependent Li+ conductivity of melt-quenched, Li-doped ZIF-4 glass by impedance spectroscopy or 7Li NMR; an activation energy near 0.35 eV instead of 0.16 eV would overturn the central claim. A complementary check is a DFT-based nudged-elastic-band scan of Li+ migration barriers in the glassy structure; if the barriers are not broadly distributed around 0.16 eV, the disorder mechanism is not supported.","tokens_in":17408,"feed_emoji":"🔋","tokens_out":10900,"duration_ms":105619,"temperature":0.7,"pith_summary":"This paper claims that turning a crystalline metal-organic framework into a glass is not a sacrifice but a gain for lithium-ion conduction. Simulating Li+ transport in crystalline and glassy ZIF-4 and ZIF-62 with a machine-learned interatomic potential, the authors find that vitrification lowers the migration activation energy from about 0.35 eV to 0.16 eV, raises the extrapolated room-temperature diffusion coefficient by more than an order of magnitude in ZIF-4 and nearly sevenfold in ZIF-62, and replaces rare anisotropic cage-to-cage hopping with continuous, Fickian-like isotropic diffusion. The structural cause is the randomization of imidazolate and benzimidazolate ring orientations, which removes direction-specific energy barriers. If correct, this makes ZIF glasses a concrete platform for solid-state battery electrolytes that need fast, direction-independent transport without grain boundaries.","feed_headline":"Disorder in ZIF glass electrolytes cuts Li-ion barrier to 0.16 eV","feed_subtitle":"Glass ZIF-4 conducts lithium about 17 times faster at room temperature, and isotropically.","key_machinery":"The load-bearing machinery is the structural disorder itself, produced by melt-quenching, together with a fine-tuned deep-learning interatomic potential that lets the authors run accurate nanosecond-scale molecular dynamics with only ten Li+ ions per supercell, sampled over 20 to 40 replicas for statistical convergence. The paper links disorder to transport through two quantitative tools: ring-orientation pole figures, which map the stereographic projections of imidazolate and benzimidazolate ring normals and show the loss of orientational texture on vitrification; and the diffusion tensor, whose diagonal components $D_{xx}$, $D_{yy}$, and $D_{zz}$ are drawn as projected ellipsoids and show the collapse of anisotropy in the glass. Van Hove correlation functions and the non-Gaussian parameter supply the mechanistic contrast between discrete hopping and continuous diffusion.","core_discovery":"Using a machine-learned interatomic potential fine-tuned on lithium-containing configurations, the authors simulate Li+ transport in crystalline and glassy ZIF-4 and ZIF-62. They find that the glassy frameworks show a Li+ migration activation energy of 0.16 eV, down from about 0.36 eV in ZIF-4 crystals and 0.34 eV in ZIF-62 crystals, and that the extrapolated room-temperature diffusion coefficient rises from 1.91e-8 to 3.21e-7 cm$^2$/s for ZIF-4 and from 1.12e-8 to 7.76e-8 cm$^2$/s for ZIF-62. The mechanistic picture is that crystalline ZIFs confine Li+ in cages and permit only rare, dynamically heterogeneous hops, whereas the glass presents a broad distribution of coordination geometries and barriers that supports continuous, Fickian-like movement. Directional analysis shows crystal diffusion is strongly anisotropic ($D_{yy} > D_{xx} > D_{zz}$, tied to ordered imidazolate and benzimidazolate ring orientations), while the glass is isotropic or near-isotropic because vitrification randomizes those ring orientations.","pith_inferences":["Editorial inference: a direct experimental measurement of Li+ conductivity in melt-quenched, Li-doped ZIF-4 and ZIF-62 by impedance spectroscopy or 7Li NMR would test whether the predicted 0.16 eV activation energy and the order-of-magnitude speedup appear in real glasses.","Editorial inference: a natural computational check is DFT-based migration-barrier calculations for representative glassy configurations, since the machine-learning potential used here was validated against DFT only for Li+ in crystalline ZIF-62.","Editorial inference: the ring-orientation mechanism suggests a search principle: other MOF frameworks with ordered linker apertures should show the same barrier reduction and isotropization upon vitrification, so ZIF-8, SALEM-2, or mixed-linker variants are natural next targets."],"forward_implications":["If the simulated values hold experimentally, a ZIF glass electrolyte would deliver room-temperature Li+ diffusivity roughly an order of magnitude above its crystalline parent without any grain-boundary engineering.","The 0.16 eV activation energy is close to the regime of practical solid electrolytes, meaning the glassy framework could sustain fast transport near ambient temperature.","Because diffusion in the glass is isotropic, battery cells would not need oriented or single-crystal electrolyte layers to avoid direction-dependent resistance.","The correlation between ring orientation and transport gives a design lever: linkers or processing routes that maximize orientational disorder should also lower migration barriers in other MOF glasses.","The hopping-to-continuous crossover implies that even at low lithium loadings, a substantial fraction of Li+ is mobile in the glass, which is favorable for dilute-solution electrolytes."],"supporting_citations":[{"why":"provides the deep-learning potential training code used to construct and fine-tune the interatomic potential.","marker":"[21]"},{"why":"provides the updated machine-learning potential framework used in the training workflow.","marker":"[22]"},{"why":"supplies the pre-existing ZIF deep-learning potential and the ring-orientation descriptor that the present study extends to ZIF-62.","marker":"[23]"},{"why":"establishes the two-stage melt-quenching protocol for making realistic ZIF glass structures.","marker":"[24]"},{"why":"gives experimental densities of melt-quenched ZIF glasses used to validate the simulated glass structures.","marker":"[27]"},{"why":"provides experimental ZIF-62 glass density and preparation context for structural comparison.","marker":"[28]"},{"why":"supplies the machine-learning analysis approach for lithium conduction and ion dynamics in glassy electrolytes.","marker":"[31]"},{"why":"provides the diffusion-tensor formalism for directional diffusivity used to quantify anisotropy.","marker":"[34]"}],"fun_headline_variants":["ZIF glass drops Li-ion barrier to 0.16 eV","Glass ZIF-4 conducts Li+ 17x faster at room temperature","ZIF glass makes Li-ion diffusion isotropic and fast","ZIF glass: 0.16 eV barrier, isotropic Li+ diffusion","Disorder in ZIF glass enables fast isotropic Li+ transport"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The fine-tuned machine-learning potential accurately describes how Li+ interacts with the disordered glassy ZIF framework, but it was benchmarked against DFT only for Li+ in crystalline ZIF-62, not for the glassy environments where the central claims live.","fun_headline_variants_meta":{"raw":{"variants":["ZIF glass drops Li-ion barrier to 0.16 eV","Glass ZIF-4 conducts Li+ 17x faster at room temperature","ZIF glass makes Li-ion diffusion isotropic and fast","ZIF glass: 0.16 eV barrier, isotropic Li+ diffusion","Disorder in ZIF glass enables fast isotropic Li+ transport"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001971,"raw_usage":{"total_tokens":7784,"prompt_tokens":1111,"completion_tokens":6673,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":727,"completion_tokens_details":{"reasoning_tokens":6582}},"tokens_in":727,"tokens_out":6673,"duration_ms":47581,"temperature":1.0,"reasoning_tokens":6582,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T18:40:02.567184+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the temperature-dependent Li+ conductivity of melt-quenched, Li-doped ZIF-4 glass by impedance spectroscopy or 7Li NMR; an activation energy near 0.35 eV instead of 0.16 eV would overturn the central claim. A complementary check is a DFT-based nudged-elastic-band scan of Li+ migration barriers in the glassy structure; if the barriers are not broadly distributed around 0.16 eV, the disorder mechanism is not supported.","supporting_citations":[],"review_version":1}