Pith. sign in

REVIEW 4 major objections 4 minor 29 references

A machine-learning funnel for generating metal-organic frameworks claims to identify two photocatalysts whose predicted multi-criterion fitness beats a benchmark material by 1.2- to 1.7-fold, while cutting screening cost by 4.13-fold.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 10:05 UTC pith:DGJ66CBL

load-bearing objection A transparent, code-released ML screening pipeline, but the headline claim of simultaneous improvements for the Cr candidate is contradicted by the paper's own data. the 4 major comments →

arxiv 2607.27295 v1 pith:DGJ66CBL submitted 2026-07-29 cond-mat.mtrl-sci cs.LG

MatCreatioNN: Machine learning-guided computational discovery of photocatalysts for environmental applications

Satya Kokonda This is my paper
classification cond-mat.mtrl-sci cs.LG
keywords photocatalysisCO2 reductionmetal-organic frameworksmachine learningcrystal graph neural networksreinforcement learningmaterials discoverymulti-objective optimization
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper seeks to establish that a fully computational pipeline can find new metal-organic framework (MOF) photocatalysts for CO2 reduction and environmental cleanup without brute-force simulation of every candidate. It couples a reinforcement-learning generator that creates 120,000 MOF structures with a multi-stage neural-network funnel that screens them on 13 descriptors spanning stability, catalytic promise, adsorption, cost, and sustainability. The headline result is that two generated materials, a zinc-based MOF and a chromium-based MOF, are predicted to score 1.20 ± 0.05 and 1.70 ± 0.25 times higher than the established photocatalyst PCN-224(Zr) on a composite fitness measure, and that the staged funnel needs only about a quarter of the inferences of exhaustive screening. The paper also argues that recurring structural motifs, a nitrogen-rich metal cluster and the bcg topology, mark reusable design rules. A sympathetic reader would care because, if the proxy descriptors hold up experimentally, the pipeline offers a fast, cheap route to synthesizable catalyst candidates.

Core claim

The central claim is that machine learning can replace exhaustive computation with a sequential funnel: generate a large set of candidate MOFs using reinforcement learning, then pass them through successive Crystal Graph Convolutional Neural Network (CGCNN) predictors—a graph-based neural net that estimates properties from atomic structure—so that each stage removes the worst 5% on a different criterion. The paper reports that this cut inference cost by 4.13-fold relative to evaluating all 13 properties on all 120,000 candidates. After screening, the two surviving candidates, a Zn-based MOF with predicted band gap 1.907 eV and a Cr-based MOF with predicted band gap 1.735 eV, are claimed to o

What carries the argument

The load-bearing pieces are the reinforcement-learning generator, the multi-stage CGCNN funnel, and the ensemble fitness evaluator. A CGCNN encodes each MOF crystal as a graph of atoms and bonds; thirteen such predictors estimate the descriptors used for screening. The funnel does the main work: instead of computing all 13 properties on all 120,000 candidates, it applies filters sequentially—stability, catalytic proxy, cost, sustainability, thermal stability, adsorption, synthesizability—so that later stages see fewer structures. Final ranking uses a generalized advanced-oxidation-process performance framework that scores each material under 13 different mathematical aggregations of five nor

Load-bearing premise

The ranking rests on the assumption that electrocatalysis-derived descriptors plus a band-gap weighting and five normalized proxy scores faithfully capture photocatalytic performance; the paper explicitly concedes that no quantum yield, turnover frequency, or band-edge calculations were performed.

What would settle it

Synthesize the Zn- or Cr-based MOF and measure its visible-light CO2 reduction quantum yield or turnover frequency against PCN-224(Zr) under identical conditions; if the 1.2–1.7x fitness advantage does not translate to activity, the proxy chain is broken. A cheaper computational check: compute DFT band-edge positions for the two structures; if the conduction and valence bands do not straddle the CO2-to-CO and H2O-to-O2 potentials, the candidates cannot be the photocatalysts claimed.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • If the predictions hold, the Zn- and Cr-based MOFs are concrete starting points for experimental synthesis, with simulated XRD fingerprints to guide verification.
  • The 4.13-fold inference reduction means the same funnel strategy can be applied to larger generated libraries or to other target reactions without a proportional increase in compute.
  • The recurring N262 metal cluster and bcg topology become explicit design rules: future generative searches can bias sampling toward these motifs instead of exploring blindly.
  • Because the Zn-based candidate beats the benchmark on all five descriptor categories, its ranking does not depend on how the fitness functions are weighted; the Cr-based candidate's ranking relies on its cost and sustainability advantages.

Where Pith is reading between the lines

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

  • Editorial inference: the paper's headline numbers stand or fall on transferability of electrocatalysis predictors (Faradaic efficiency, free energy, voltage) to photocatalysis; if shared *COOH/*CO intermediates do not dominate real light-driven MOF catalysis, the 1.2–1.7x gains could be proxy artifacts.
  • Editorial inference: no quantum yield, turnover frequency, carrier-lifetime, or band-edge calculation appears anywhere in the work, so the fitness improvements are best read as enrichment scores for synthesis triage, not as predicted reaction rates.
  • Editorial inference: a natural computational falsifier is a DFT band-edge calculation for the two structures; if the 1.7–1.9 eV gaps do not straddle the CO2-reduction and water-oxidation potentials, the photocatalytic rationale fails even though the fitness scores are high.
  • Editorial inference: the funnel's modular structure suggests it could be repurposed for pollutant-degradation photocatalysis by replacing the redox-related descriptors with reactive-oxygen-species or band-edge descriptors, an extension the author sketches but does not test.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes MatCreatioNN, a machine-learning pipeline that combines reinforcement-learning-based generation of 120,000 MOF candidates with a sequential CGCNN screening funnel over 13 descriptors (stability, catalytic proxies, cost, sustainability, adsorption, synthesizability). It reports that the funnel reduces inference cost by 4.13-fold, that two generated MOFs (Cr-based and Zn-based) have predicted fitness 1.70±0.25× and 1.20±0.05× higher than the PCN-224(Zr) benchmark, and that simulated XRD patterns indicate synthetic plausibility. Post-hoc analysis identifies enriched motifs (N262/N331/N536 clusters, bcg topology). The central quantitative claim—simultaneous improvements in light absorption, redox energetics, and framework durability—is not supported by the paper's own data: Section 5.2 and Table S7 show the Cr-MOF is below the benchmark in catalytic score and stability, and the 'all 13 fitness functions' statement in Section 4 is contradicted by Table S10.

Significance. If the central claims held, the work would demonstrate a scalable generative-screening framework for multi-objective MOF discovery. Strengths of the manuscript include public code and data (GitHub/Zenodo), explicit inference-count accounting for computational cost, an ensemble of 13 fitness functions, uncertainty propagation, and resampling-based confidence intervals. These are useful methodological components. However, the headline 'photocatalytic fitness' improvements are not established: the comparison is circular, the proxy transfer from electrocatalysis is unvalidated, and internal contradictions undermine the reported statistics. The paper could become a credible methods contribution after substantial revision, but as written the main claim is not defensible.

major comments (4)
  1. [Abstract; §4; §5.2; Table S7] The abstract's claim that both top candidates 'demonstrate simultaneous improvements in light absorption, redox energetics, and framework durability' is contradicted by Table S7 and Section 5.2. The Cr-based MOF has normalized catalytic score 0.515 vs PCN-224's 0.689 and stability 0.863 vs 0.885; it exceeds the benchmark only on cost, sustainability, and marginally on adsorption (0.760 vs 0.727). Section 5.2 explicitly states Cr 'sacrifices catalytic performance' and that 'Neither material simultaneously satisfies design criteria.' The 1.70±0.25× mean fold improvement is an average over 13 fitness functions, several of which are dominated by economic terms (e.g., S13 is the pure economic average). This is not a photocatalytic performance measure. The abstract and Section 4 must be corrected, and all 'simultaneous improvements' language removed unless supported by the descriptor-level dat
  2. [§4; §5.3; Table S10] Section 4 states 'All 13 fitness functions yielded statistically significant improvements for both Zn- and Cr-based MOFs after Bonferroni–Holm correction.' This is inconsistent with Table S10, which shows Z>2 in only 11/13 (Zn) and 12/13 (Cr) functions at r=0.11; at r=0.20, Zn drops to 0/13 and Cr to 10/13. Section 5.2 also reports Cr's Z=-2.90 (p=0.004) under the no-economic fitness function. The 'all 13' statement is therefore false. The statistical summary should be recomputed and reported per fitness function, with a clear statement of which functions support which candidate.
  3. [§3.2.1; §3.3; §3.4; Fig. 8] The benchmark comparison is a selection artifact. Generated candidates were pruned at every funnel stage using the same five descriptors (stability, catalytic score, cost, sustainability, adsorption) that define the ensemble fitness functions, while the control MOFs were not subjected to any equivalent funnel filtering before scoring. The 'complete dominance' and tiered separation in Fig. 8 therefore show that candidates surviving 5%-quantile filters score higher on the retained descriptors than unfiltered controls; they do not demonstrate that generated MOFs are superior photocatalysts. To support the discovery claim, controls should be passed through the same funnel or, preferably, the candidates should be evaluated on independent photocatalytic metrics (e.g., computed band edges, carrier dynamics, or experimental activity data).
  4. [§3.2.2; §5.4; §5.5] The proxy transfer from electrocatalysis models (Faradaic efficiency, free energy, voltage potential) to photocatalysis is asserted on mechanistic analogy but not validated. Section 5.5 concedes that no quantum yield or turnover frequency is predicted, and Section 5.4 concedes that no explicit DFT band-edge calculations were performed. The 'predicted photocatalytic fitness' is therefore a composite of proxy descriptors, not a predicted photocatalytic activity. The paper should either add validation (e.g., band-edge calculations, or comparison against experimental photocatalytic datasets) or explicitly restrict all conclusions to 'predicted multi-descriptor fitness' and remove 'photocatalytic fitness' and 'simultaneous improvements' from the abstract and conclusions.
minor comments (4)
  1. [S3.2] The supplementary has two sections numbered 'S3.2' ('Pareto-Front-Based Hyperparameter Selection' and 'Computational Modelling Proxy'). Renumber to avoid ambiguity.
  2. [§2] The statement '7 minutes as compared to 1.5 million hours' appears to conflate per-structure and total screening time. Clarify the basis of the comparison.
  3. [S6.3] The Cr-based fold improvement is reported as mean 1.70× but the bootstrap median is 1.53×. Report both mean and median in the main text, since the skewed distribution makes the mean misleading.
  4. [§5.5] The Mann–Whitney p-values are reported without multiple-testing correction across the three pairwise comparisons. Add a note on correction or justify the uncorrected values.

Circularity Check

2 steps flagged

The reported 1.2–1.7x fitness 'prediction' restates the selection objective: candidates were screened and ranked on the same five-descriptor composite used for the headline comparison, and Section 5.2 contradicts the abstract's 'simultaneous improvements' claim for Cr.

specific steps
  1. self definitional [Section 3.4 / Section 5.8 / Table S7]
    "Pareto dominance analysis: The Zn-based MOF outperformed PCN-224(Zr) across all five normalized descriptors (Fig. 5a), confirming that its superiority holds under any monotonic aggregation function."

    The 13 fitness functions (Eqs. S1–S13) are all monotone aggregations of the same five normalized descriptors (stability, catalytic activity, cost, sustainability, adsorption). Table S7 shows Zn ≥ PCN-224 on every descriptor, so f(Zn) > f(PCN) for every f in the ensemble is a mathematical consequence of those inputs, not an independent prediction. Because the funnel screened and ranked candidates on this same descriptor set (Section 3.2), the reported 1.20x (and, for Cr, the economics-dominated 1.70x) fold improvements restate the selection objective rather than test it.

  2. fitted input called prediction [Section 5.2 / Abstract / Table S7]
    "the Cr-based MOF sacrifices catalytic performance (0.52, falling 25% below the PCN-224 benchmark) to achieve exceptional economic metrics (cost: 1.00, sustainability: 1.00)... when economic factors were excluded ('No economic' function), Cr-MOF ranked below the PCN-224 benchmark (Z-score = -2.90, p = 0.004), demonstrating that its overall fitness superiority depends on cost and sustainability advantages rather than enhanced photocatalytic performance."

    The abstract's '1.70±0.25' and 'simultaneous improvements in light absorption, redox energetics, and framework durability' are drawn from a composite fitness whose five components include cost and sustainability. Section 5.2 shows Cr's catalytic score is below benchmark and it loses under the no-economics function (S10), yet Section 4 states 'All 13 fitness functions yielded statistically significant improvements.' The label 'photocatalytic fitness' converts an economic-weighted average into a claim about photocatalytic performance; the prediction is the fitted composite renamed.

full rationale

The central 'prediction' — that generated MOFs outperform PCN-224(Zr) by 1.20–1.70x — is largely forced by construction. Candidates were generated, funnel-filtered, and ranked using the same five normalized descriptors (stability, catalytic activity, cost, sustainability, adsorption) that define all 13 fitness functions used for the headline comparison. Zn-MOF dominates PCN-224 on every descriptor, so any monotone aggregation must rank it higher; the paper itself says this 'confirms' superiority, but that is a mathematical consequence, not an empirical validation. Cr-MOF's higher fold improvement is driven by cost and sustainability weights, and the paper's own Section 5.2 concedes it is below benchmark on the catalytic descriptor and under the no-economics fitness S10 — directly contradicting the abstract's 'simultaneous improvements' claim and Section 4's statement that all 13 functions gave significant improvements. I found no load-bearing self-citation pattern: the author's prior works [3,48] are not the basis of the main derivation, and external anchors (QMOF/MOFSimplify data, literature control MOFs, simulated-vs-experimental XRD) provide some independent support. The paper also honestly limits itself in Section 5.5 (no QY/TOF) and Section 5.4 (no explicit band-edge calculations), but those limitations reinforce rather than remove the central issue: the reported photocatalytic fitness is an unvalidated proxy composite, and comparing selected candidates to unselected controls on that same composite is an in-sample restatement of the selection rule.

Axiom & Free-Parameter Ledger

9 free parameters · 7 axioms · 1 invented entities

The central claim rests on unvalidated proxy transfer, hand-set scoring parameters, and a selection-evaluation loop. No new physical entity is required, but the N262/N331/N536 motif claims are unvalidated design conjectures.

free parameters (9)
  • Band-gap Gaussian center = 1.9 eV
    Optimal band-gap window 1.8-2.0 eV enforced via f(B) = 4.536 * exp(-(B-1.9)^2/(2*75)) in S3.2; hand-chosen, not learned from photocatalytic activity data.
  • Band-gap Gaussian width = 75
    Width in the same Gaussian weighting in S3.2; chosen to shape the band-gap preference, no independent justification.
  • Adsorption Gaussian center = -30 kJ/mol
    Truncated Gaussian weighting g(A) = exp(-(A+30)^2/(2*75)) in S3.2; selected from a literature heat-of-adsorption window, not fitted to photocatalytic outcomes.
  • Adsorption Gaussian width = 75
    Width in the adsorption weighting in S3.2; arbitrary smoothing scale.
  • Quantile cutoff per funnel stage = 5%
    Bottom-5% removal at each funnel stage (S3.1); the paper notes alternative thresholds were not exhaustively evaluated.
  • Uniform relative uncertainty r = 0.11 (also 0.05, 0.10, 0.15, 0.20)
    All five descriptors are assigned the same relative uncertainty in S5.1; this drives Z-scores, confidence intervals, and significance claims.
  • Pareto hyperparameter weighting = 0.76 MAE / 0.24 time
    Weighted Euclidean distance in S3.2; hand-picked to prioritize accuracy over speed.
  • Sustainability scaling factor = 1000
    Raw sustainability scores were multiplied by 1000 to avoid vanishing gradients (S3.3); rescaling affects training targets.
  • Expert-informed fitness function = Estab*Ecat*Eads*sqrt(Esust+1)*sqrt(Ecost+1)
    Post-hoc ranking function in S5.2; constructed by the authors and used to identify the top 1,000 MOFs and motifs, not part of the funnel.
axioms (7)
  • domain assumption Electrocatalysis descriptors transfer to photocatalysis
    Section 3.2.2 asserts shared *COOH/*CO intermediates and energetic landscapes justify using Faradaic efficiency, free energy, and voltage predictors as photocatalytic proxies; no photocatalysis dataset is used to validate transfer.
  • domain assumption CGCNN models generalize to generated hypothetical MOFs
    Section 3.3 restricts benchmarking to MOF domain but does not quantify extrapolation error for RL-generated structures outside the training distribution.
  • domain assumption The five-descriptor composite fitness captures photocatalytic performance
    Section 5.5 explicitly states the work does not predict quantum yield or turnover frequency, so the fitness score is an unvalidated proxy for actual photocatalytic activity.
  • domain assumption Literature windows for band gap and adsorption energy are correct
    Gaussian weightings in S3.2 encode band-gap 1.8-2.0 eV and adsorption -20 to -40 kJ/mol windows as optimal; these are literature heuristics, not measured optima for these candidates.
  • domain assumption Simulated XRD agreement implies synthesizability
    Section 5.6 uses matching simulated XRD profiles to argue physical plausibility, but no synthesis or experimental XRD was performed.
  • standard math First-order Gaussian error propagation is valid for the fitness functions
    S5.1 applies first-order propagation assuming independent descriptors and fixed relative uncertainty; standard but approximate.
  • standard math Bootstrap and jackknife resampling provide valid confidence intervals
    S6.3 uses 20,000 bootstrap resamples and leave-one-out jackknife across 13 fitness functions; standard statistical practice.
invented entities (1)
  • N262/N331/N536 metal-cluster motifs no independent evidence
    purpose: Postulated recurring structural design motifs for high photocatalytic fitness
    Identified post hoc in the top 1,000 scoring candidates; no experimental or external validation that these clusters cause photocatalytic activity. The labels are treated as potential design rules but have no independent falsifiable handle outside this paper.

pith-pipeline@v1.3.0-daily-deepseek · 31401 in / 14426 out tokens · 117442 ms · 2026-08-01T10:05:48.487258+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of MatCreatioNN: Machine learning-guided computational discovery of photocatalysts for environmental applications." pith.science (2026). https://pith.science/paper/DGJ66CBL

@misc{pith2026260727295,
  author       = {Pith},
  title        = {Pith review of: MatCreatioNN: Machine learning-guided computational discovery of photocatalysts for environmental applications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DGJ66CBL}},
  note         = {Machine review of arXiv:2607.27295}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The rational design of photocatalysts for environmental remediation and CO2 conversion remains limited by the high computational cost and sparse experimental data describing multi-parameter photocatalytic behavior. This work presents an integrated machine-learning framework that couples reinforcement learning-based metal-organic framework (MOF) generation with a multi-stage Crystal Graph Convolutional Neural Network (CGCNN) prediction funnel to identify photocatalysts optimized across multiple electronic and structural features. 120,000 MOF candidates were generated and screened using 13 key descriptors, including band-gap suitability, CO2/H2O selectivity, adsorption energy, and structural stability. The funnel approach reduced computational cost by 4.13-fold while maintaining predictive robustness. Two top candidates, a Cr-based and a Zn-based MOF, exhibited predicted photocatalytic fitness values of 1.70 +/- 0.25 and 1.20 +/- 0.05 fold higher respectively than benchmark materials such as PCN-224(Zr), demonstrating simultaneous improvements in light absorption, redox energetics, and framework durability. Simulated X-ray diffraction patterns confirmed strong structural agreement with experimentally synthesized MOFs, indicating high synthesizability. Post-hoc analysis revealed recurring structural motifs, such as the N262 metal cluster, that correlated strongly with high predicted photocatalytic activity. These results highlight the potential of data-driven methods to accelerate discovery of efficient and durable photocatalysts for environmental and energy-related transformations, providing a foundation for experimental realization and large-scale implementation of computationally designed MOFs.

Figures

Figures reproduced from arXiv: 2607.27295 by Satya Kokonda.

Figure 1
Figure 1. Figure 1: Overview of the MATCreatioNN machine-learning workflow for photocatalytic MOF discovery. Reinforcement-learning–based MOF generation is followed by sequential CGCNN screening for stability, catalytic proxies, adsorption, and economic/sustainability metrics. Candidates are evaluated using an ensemble of 13 multi-objective fitness aggregation functions to ensure robustness to fitness-function choice. Top-ran… view at source ↗
Figure 2
Figure 2. Figure 2: Workflow of the CGCNN funnel system. Thirteen sequential CGCNN modules predict physical, electronic, and catalytic descriptors of MOFs. At each stage, the lowest 5% of candidates are removed based on property-specific thresholds, maintaining dataset diversity while progressively enriching high-performing structures. This hierarchical filtration reduces overfitting and improves computational efficiency by 4… view at source ↗
Figure 3
Figure 3. Figure 3: Distribution of (a) topologies and (b) metal clusters among the 1,000 highest-ranked MOFs [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Multi-Fitness comparison of two identified MOFs compared to general top performing [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: (a) Normalized descriptor profiles (0–1 scaling) for the same MOFs, illustrating that [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Uncertainty propagation analysis showing robust separation of Zn- and Cr-based MOFs [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Simulated and experimental XRD comparison for top-performing generated MOFs. XRD analysis was conducted on the highest-ranked predicted structures to confirm structural feasibility and compare diffraction profiles against known synthesized analogs. Close alignment between simulated and experimental peaks validates the crystallographic plausibility of the generated MOFs. 14 [PITH_FULL_IMAGE:figures/full_fi… view at source ↗
Figure 8
Figure 8. Figure 8: Violin (ribbon) plots show the distribution of mean composite fitness scores (averaged over [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

29 extracted references · 17 canonical work pages · 1 internal anchor

  1. [1]

    Boukayouht, L

    K. Boukayouht, L. Bazzi, S. El Hankari, Sustainable synthesis of metal-organic frameworks and their derived materials from organic and inorganic wastes, Coord. Chem. Rev. 478 (2023) 214986.https://doi.org/10.1016/j.ccr.2022.214986

    arXiv 2023

  2. [2]

    H. Park, S. Majumdar, X. Zhang, J. Kim, B. Smit, Inverse design of metal–organic frameworks for direct air capture of CO2 viadeep reinforcement learning, Digit. Discov. 3 (2024) 728–741. https://doi.org/10.1039/D4DD00010B

    work page doi:10.1039/d4dd00010b 2024

  3. [3]

    Kim, K.-B

    M. Kim, K.-B. Hwang, An empirical evaluation of sampling methods for the classification of imbalanced data, PLOS ONE 17 (2022) e0271260.https://doi.org/10.1371/journal.pone.027 1260

    work page doi:10.1371/journal.pone.027 2022

  4. [4]

    Nandy, G

    A. Nandy, G. Terrones, N. Arunachalam, C. Duan, D.W. Kastner, H.J. Kulik, MOFSimplify, machine learning models with extracted stability data of three thousand metal–organic frameworks, Sci. Data 9 (2022) 74.https://doi.org/10.1038/s41597-022-01181-0

    doi:10.1038/s41597-022-01181-0 2022

  5. [5]

    Rosen, S.M

    A.S. Rosen, S.M. Iyer, D. Ray, Z. Yao, A. Aspuru-Guzik, L. Gagliardi, J.M. Notestein, R.Q. Snurr, Machine learning the quantum-chemical properties of metal–organic frameworks for accelerated materials discovery, Matter 4 (2021) 1578–1597.https://doi.org/10.1016/j.matt.2 021.02.015

    work page doi:10.1016/j.matt.2 2021

  6. [6]

    Gibaldi, A

    M. Gibaldi, A. Kapeliukha, A. White, J. Luo, R.A. Mayo, J. Burner, T.K. Woo, MOSAEC- DB: a comprehensive database of experimental metal–organic frameworks with verified chemical accuracy suitable for molecular simulations, Chem. Sci. 16 (2025) 4085–4100.https://doi.org/ 10.1039/D4SC07438F

    work page doi:10.1039/d4sc07438f 2025

  7. [7]

    Xie, J.C

    T. Xie, J.C. Grossman, Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties, Phys. Rev. Lett. 120 (2018) 145301. https: //doi.org/10.1103/PhysRevLett.120.145301. 45

    doi:10.1103/physrevlett.120.145301 2018

  8. [8]

    N. Redkar, CarbNN: A Novel Active Transfer Learning Neural Network To Build De Novo Metal Organic Frameworks (MOFs) for Carbon Capture, (2023).https://doi.org/10.48550/arX iv.2311.16158

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2311.16158 2023

  9. [9]

    Reddi, C

    V.J. Reddi, C. Cheng, D. Kanter, P. Mattson, G. Schmuelling, C.-J. Wu, B. Anderson, M. Breughe, M. Charlebois, W. Chou, R. Chukka, C. Coleman, S. Davis, P. Deng, G. Diamos, J. Duke, D. Fick, J.S. Gardner, I. Hubara, S. Idgunji, T.B. Jablin, J. Jiao, T.S. John, P. Kanwar, D. Lee, J. Liao, A. Lokhmotov, F. Massa, P. Meng, P. Micikevicius, C. Osborne, G. Pek...

    Pith/arXiv arXiv doi:10.48550/arxiv.1911.02549 2020

  10. [10]

    Zhang, X

    T. Zhang, X. Sun, S. Weng, S. Zhang, C. Xu, X. Gao, N. Zhu, Enhancing photocatalytic performance of rose-shaped Co/Ni bimetallic organic framework for reducing CO2 to CO under visible light, J. Mol. Struct. 1321 (2025) 140190.https://doi.org/10.1016/j.molstruc.2024.140190

    arXiv 2025

  11. [11]

    Rueda-Navarro, Z

    C.M. Rueda-Navarro, Z. Abou Khalil, A. Melillo, B. Ferrer, R. Montero, A. Longarte, M. Daturi, I. Vayá, M. El-Roz, V. Martínez-Martínez, H.G. Baldoví, S. Navalón, Solar Gas-Phase CO2 Hydrogenation by Multifunctional UiO-66 Photocatalysts, ACS Catal. 14 (2024) 6470–6487. https://doi.org/10.1021/acscatal.4c00266

    work page doi:10.1021/acscatal.4c00266 2024

  12. [12]

    C. Chen, Q. Mo, J. Fu, Q. Yang, L. Zhang, C.-Y. Su, PtCu@Ir-PCN-222: Synergistic Catalysis of Bimetallic PtCu Nanowires in Hydrosilane-Concentrated Interspaces of an Iridium(III)– Porphyrin-Based Metal–Organic Framework, ACS Catal. 12 (2022) 3604–3614.https://doi.org/ 10.1021/acscatal.1c05922

    work page doi:10.1021/acscatal.1c05922 2022

  13. [13]

    Stanley, K

    P.M. Stanley, K. Hemmer, M. Hegelmann, A. Schulz, M. Park, M. Elsner, M. Cokoja, J. Warnan, Topology- and wavelength-governed CO2 reduction photocatalysis in molecular catalyst- metal–organic framework assemblies, Chem. Sci. 13 (2022) 12164–12174.https://doi.org/10.1 039/D2SC03097G

    2022

  14. [14]

    F. Guo, S. Yang, Y. Liu, P. Wang, J. Huang, W.-Y. Sun, Size Engineering of Metal–Organic Framework MIL-101(Cr)–Ag Hybrids for Photocatalytic CO2 Reduction, ACS Catal. 9 (2019) 8464–8470.https://doi.org/10.1021/acscatal.9b02126

    work page doi:10.1021/acscatal.9b02126 2019

  15. [15]

    J. Yi, X. Wu, H. Wu, L. Zhang, K. Wu, J. Guo, Correction: Facile synthesis of novel NH2 -MIL-53(Fe)/AgSCN heterojunction composites as a highly efficient photocatalyst for ciprofloxacin degradation and H2 production under visible-light irradiation, React. Chem. Eng. 7 (2022) 201–201. https://doi.org/10.1039/D1RE90046C

    work page doi:10.1039/d1re90046c 2022

  16. [16]

    Karmakar, S

    S. Karmakar, S. Barman, F.A. Rahimi, D. Rambabu, S. Nath, T.K. Maji, Confining charge-transfer complex in a metal-organic framework for photocatalytic CO2 reduction in water, Nat. Commun. 14 (2023) 4508.https://doi.org/10.1038/s41467-023-40117-z

    work page doi:10.1038/s41467-023-40117-z 2023

  17. [17]

    Lingampalli, M.M

    S.R. Lingampalli, M.M. Ayyub, C.N.R. Rao, Recent Progress in the Photocatalytic Reduction of Carbon Dioxide, ACS Omega 2 (2017) 2740–2748.https://doi.org/10.1021/acsomega.7b0 0721

    work page doi:10.1021/acsomega.7b0 2017

  18. [18]

    Sroka, COMPARISON OF JACKKNIFE AND BOOTSTRAP METHODS IN ESTIMATING CONFIDENCE INTERVALS, Sci

    University of Economics, Katowice, Ł. Sroka, COMPARISON OF JACKKNIFE AND BOOTSTRAP METHODS IN ESTIMATING CONFIDENCE INTERVALS, Sci. Pap. Silesian Univ. Technol. Organ. Manag. Ser. 2021 (2021) 446–455.https://doi.org/10.29119/1641-346 6.2021.153.31

    work page doi:10.29119/1641-346 2021

  19. [19]

    Severiano, J.A

    A. Severiano, J.A. Carriço, D.A. Robinson, M. Ramirez, F.R. Pinto, Evaluation of jackknife and bootstrap for defining confidence intervals for pairwise agreement measures, PloS One 6 (2011) e19539.https://doi.org/10.1371/journal.pone.0019539

    work page doi:10.1371/journal.pone.0019539 2011

  20. [20]

    C. Wang, Y. Wan, S. Yang, Y. Xie, S. Chu, Y. Chen, X. Yan, Revealing the Untapped Potential of Photocatalytic Overall Water Splitting in Metal Organic Frameworks, Adv. Funct. Mater. 34 (2024) 2313596.https://doi.org/10.1002/adfm.202313596. 46

    work page doi:10.1002/adfm.202313596 2024

  21. [21]

    Khan, S.M

    S.T. Khan, S.M. Moosavi, Connecting metal-organic framework synthesis to applications using multimodal machine learning, Nat. Commun. 16 (2025) 5642.https://doi.org/10.1038/ s41467-025-60796-0

    2025

  22. [22]

    M. Khan, Z. Akmal, M. Tayyab, S. Mansoor, A. Zeb, Z. Ye, J. Zhang, S. Wu, L. Wang, MOFs materials as photocatalysts for CO2 reduction: Progress, challenges and perspectives, Carbon Capture Sci. Technol. 11 (2024) 100191.https://doi.org/10.1016/j.ccst.2024.100191

    arXiv 2024

  23. [23]

    Zhang, Y

    F. Zhang, Y. Jiang, J. Liu, A. Jiang, Y. Cao, S. Yu, K. Zheng, Y. Zhou, Exploration of ultrafast dynamic processes in photocatalysis: Advances and challenges, Fundam. Res. 5 (2025) 2838–2849.https://doi.org/10.1016/j.fmre.2024.04.003

    work page doi:10.1016/j.fmre.2024.04.003 2025

  24. [24]

    Q. Pan, M. Abdellah, Y. Cao, W. Lin, Y. Liu, J. Meng, Q. Zhou, Q. Zhao, X. Yan, Z. Li, H. Cui, H. Cao, W. Fang, D.A. Tanner, M. Abdel-Hafiez, Y. Zhou, T. Pullerits, S.E. Canton, H. Xu, K. Zheng, Ultrafast charge transfer dynamics in 2D covalent organic frameworks/Re-complex hybrid photocatalyst, Nat. Commun. 13 (2022) 845.https://doi.org/10.1038/s41467-02...

    work page doi:10.1038/s41467-022-28409-2 2022

  25. [25]

    Djurišić, Y

    A.B. Djurišić, Y. He, A.M.C. Ng, Visible-light photocatalysts: Prospects and challenges, APL Mater. 8 (2020) 030903.https://doi.org/10.1063/1.5140497

    work page doi:10.1063/1.5140497 2020

  26. [26]

    Swetha, T

    S. Swetha, T. Awad Alahmadi, M. Javed Ansari, S. Sudheer Khan, Strategically tailored double S-scheme heterojunction in h-MoO3 doped Bi7O9I3 decorated with Cr–CdS quantum dots for efficient photocatalytic degradation of phenolics, J. Clean. Prod. 449 (2024) 141656.https: //doi.org/10.1016/j.jclepro.2024.141656

    arXiv 2024

  27. [27]

    Ali, A.H

    Z.S.N. Ali, A.H. Bahkali, B. Janani, A.M. Elgorban, S.S. Khan, Constructing interface chemical coupling Schottky heterojunction NiMoO4/WC for enhancing photocatalytic oxidative rifampicin performance: Peroxymonosulfate activation and oxygen vacancy engineering, J. Water Process Eng. 79 (2025) 109008.https://doi.org/10.1016/j.jwpe.2025.109008

    arXiv 2025

  28. [28]

    Padha, S

    B. Padha, S. Verma, P. Mahajan, S. Arya, Electrochemical Impedance Spectroscopy (EIS) Performance Analysis and Challenges in Fuel Cell Applications, J. Electrochem. Sci. Technol. 13 (2022) 167–176.https://doi.org/10.33961/jecst.2021.01263

    arXiv 2022

  29. [29]

    Wang, J.-E

    Q. Wang, J.-E. Moser, M. Grätzel, Electrochemical Impedance Spectroscopic Analysis of Dye-Sensitized Solar Cells, J. Phys. Chem. B 109 (2005) 14945–14953.https://doi.org/10.102 1/jp052768h. 47

    2005