HCat-GNet: a Human-Interpretable GNN Tool for Ligand Optimization in Asymmetric Catalysis

iScience, Journal Year: 2025, Volume and Issue: 28(3), P. 111881 - 111881

Published: Jan. 24, 2025

Language: Английский

---

Experiment-Driven Atomistic Materials Modeling: A Case Study Combining X-Ray Photoelectron Spectroscopy and Machine Learning Potentials to Infer the Structure of Oxygen-Rich Amorphous Carbon

Journal of the American Chemical Society, Journal Year: 2024, Volume and Issue: 146(21), P. 14645 - 14659

Published: May 15, 2024

An important yet challenging aspect of atomistic materials modeling is reconciling experimental and computational results. Conventional approaches involve generating numerous configurations through molecular dynamics or Monte Carlo structure optimization selecting the one with closest match to experiment. However, this inefficient process not guaranteed succeed. We introduce a general method combine machine learning (ML) observables that produces structures compatible experiment by design. use approach in combination grand-canonical within modified Hamiltonian formalism, generate agree data are chemically sound (low energy). apply our understand oxygenated amorphous carbon (a-COx), an intriguing carbon-based material, answer question how much oxygen can be added before it fully decomposes into CO CO2. Utilizing ML-based X-ray photoelectron spectroscopy (XPS) model trained from GW density functional theory (DFT) data, conjunction ML interatomic potential, we identify a-COx compliant XPS predictions also energetically favorable respect DFT. Employing network analysis, accurately deconvolve spectrum motif contributions, both revealing inaccuracies inherent interpretation granting us insight a-COx. This generalizes multiple allows for elucidation directly thereby enabling experiment-driven degree realism previously out reach.

Language: Английский

---

AI in computational chemistry through the lens of a decade-long journey

Chemical Communications, Journal Year: 2024, Volume and Issue: 60(24), P. 3240 - 3258

Published: Jan. 1, 2024

This article gives a perspective on the progress of AI tools in computational chemistry through lens author's decade-long contributions put wider context trends this rapidly expanding field. over last decade is tremendous: while ago we had glimpse what was to come many proof-of-concept studies, now witness emergence AI-based that are mature enough make faster and more accurate simulations increasingly routine. Such turn allow us validate even revise experimental results, deepen our understanding physicochemical processes nature, design better materials, devices, drugs. The rapid introduction powerful rise unique challenges opportunities discussed too.

Language: Английский

---

A Graph Neural Network-Based Approach to XANES Data Analysis

The Journal of Physical Chemistry A, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 15, 2025

The determination of three-dimensional structures (3D structures) is crucial for understanding the correlation between structural attributes materials and their functional performance. X-ray absorption near edge structure (XANES) an indispensable tool to characterize atomic-scale local 3D system. Here, we present approach simulate XANES based on a customized graph neural network (3DGNN) model, XAS3Dabs, which takes directly system as input, inherent relation fine spectrum geometry considered during model construction. It turns out be faster than traditional fitting method when simulation optimization algorithm are combined fit given geometric features included in weighted message passing block XAS3Dabs importance investigated. demonstrates superior accuracy prediction compared most machine learning models. By extracting graphs constituted by edges related absorbing atom, our reduces redundant information, thereby not only enhancing model's performance but also improving its robustness across different hyperparameters. can generalized spectra systems with absorber having designed so meet expectations online data processing. expected key part analysis framework XAS-related beamlines high-energy photon source (HEPS) now under

Language: Английский

---

An Intelligent Prediction Model for the Synthesis Conditions of Metal–Organic Frameworks Utilizing Artificial Neural Networks Enhanced by Genetic Algorithm Optimization

Journal of Chemical Information and Modeling, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 19, 2025

In the field of emerging materials, metal–organic frameworks (MOFs) have gained prominence due to their unique porous structures, showing versatility in gas adsorption, storage, separation, and liquid processes. However, decomposition, collapse tendencies, complex synthesis make large-scale production costly challenging with no accurate method for predicting conditions. This work proposes an intelligent prediction model based on structural characteristics MOFs forecast A genetic algorithm-optimized back-propagation (BP) neural network was developed, starting feature selection via minimum redundancy maximum relevance algorithm rank importance. The optimal number inputs outputs determined basis performance, followed by optimization BP network. best initial population size hidden nodes were identified. study compared 10 models, including a simple results revealed that R coefficient optimized reached 96.2%, surpassing conventional methods all values approximately 85%. approach allows MOF conditions, aiding material manufacturing precise control over processes, improving quality, reducing raw waste.

Language: Английский

---

Active Learning‐Driven Discovery of Donor‐Acceptor Covalent Triazine Frameworks for High‐Performance Photocatalysts

Advanced Functional Materials, Journal Year: 2025, Volume and Issue: unknown

Published: May 2, 2025

Abstract Donor‐acceptor (D‐A) structure enables precise tuning of the electronic and optical properties materials, enabling widely applicable in organic semiconductors photocatalysts. However, vast diversity donor acceptor units their combinations pose considerable challenges to experimental development. Here, this study presents a screening strategy that integrates an active learning (AL)‐based multi‐model framework with synthesis validation discover high‐performance D‐A covalent triazine frameworks (CTFs) This combines AL model, trained on data reported D‐A‐CTFs, graph neural networks model establishes relationship between molecular properties. Meanwhile, expert chemical knowledge is incorporated into improve synthesizability stability, resulting 113 identified candidates from database 21807 structures. Experimental confirms 9 out 10 newly synthesized D‐A‐CTFs exhibit predicted photocatalytic performances. Notably, CTF‐[1,1′‐Biphenyl]‐4,4′‐dicarbaldehyde achieved record hydrogen evolution rate 33.29 mmol g −1 h for CTF‐based bulk Further feature engineering analysis reveals carbon nitrogen charges critically determine performance, offering optimization design. paves promising way accelerate discovery effective structured materials.

Language: Английский

---

Gaining More Insights from Synchrotron-Based X-ray Spectroscopy for Alkali Ion Rechargeable Batteries

Analytical Chemistry, Journal Year: 2024, Volume and Issue: 96(20), P. 8021 - 8035

Published: April 24, 2024

Alkali ion rechargeable batteries play a significant part in portable electronic devices and vehicles. The rapid development of renewable energy technology nowadays demands with even higher density for grid storage. To fulfill such demand, extensive research efforts have been devoted to optimizing electrochemical properties as well developing novel storage schemes designing new systems. In the investigation process, synchrotron-based X-ray spectroscopy plays vital role investigating detailed degradation mechanism schemes. Herein, we critically review applications battery recent years. This begins discussion different scientific issues alkali within various time space scales. Subsequently, principle is introduced, characteristics characterization techniques are summarized compared. Typical application cases then introduced into investigations. final presents perspectives direction both systems future.

Language: Английский

---

Machine-Learning Strategies for the Accurate and Efficient Analysis of X-ray Spectroscopy

Machine Learning Science and Technology, Journal Year: 2024, Volume and Issue: 5(2), P. 021001 - 021001

Published: May 24, 2024

Abstract Computational spectroscopy has emerged as a critical tool for researchers looking to achieve both qualitative and quantitative interpretations of experimental spectra. Over the past decade, increased interactions between experiment theory have created positive feedback loop that stimulated developments in domains. In particular, accuracy calculations led them becoming an indispensable analysis spectroscopies across electromagnetic spectrum. This progress is especially well demonstrated short-wavelength techniques, e.g. core-hole (x-ray) spectroscopies, whose prevalence following advent modern x-ray facilities including third-generation synchrotrons free-electron lasers. While based on well-established wavefunction or density-functional methods continue dominate greater part spectral analyses literature, emerging machine-learning algorithms are beginning open up new opportunities complement these traditional techniques with fast, accurate, affordable ‘ black-box ’ approaches. Topical Review recounts recent data-driven/machine-learning approaches computational spectroscopy. We discuss achievements limitations presently-available review potential expand scope reach spectroscopic studies.

Language: Английский

---

Machine Learning Confirms the Formation Mechanism of a Single-Atom Catalyst via Infrared Spectroscopic Analysis

The Journal of Physical Chemistry Letters, Journal Year: 2023, Volume and Issue: 14(49), P. 11058 - 11062

Published: Dec. 4, 2023

Single-atom catalysts (SACs) offer significant potential across various applications, yet our understanding of their formation mechanism remains limited. Notably, the pyrolysis zeolitic imidazolate frameworks (ZIFs) stands as a pivotal avenue for SAC synthesis, which can be assessed through infrared (IR) spectroscopy. However, prevailing analysis techniques still rely on manual interpretation. Here, we report machine learning (ML)-driven IR spectroscopy to unravel process Pt-doped ZIF-67 synthesize Pt–Co3O4 SAC. Demonstrating total Pearson correlation exceeding 0.7 with experimental data, algorithm provides coefficients selected structures, thereby confirming crucial structural changes time and temperature, including decomposition ZIF Pt–O bonds. These findings reveal confirm SACs. As demonstrated, integration ML algorithms, theoretical simulations, spectral introduces an approach deciphering characterization implying its broader adoption.

Language: Английский

---

A Connected Convolutional Neutral Network Protocol for Design of Two-Dimensional Materials Based on Modified Graphdiyne

The Journal of Physical Chemistry Letters, Journal Year: 2024, Volume and Issue: 15(31), P. 7840 - 7849

Published: July 25, 2024

In materials science, doping plays a crucial role in manipulating the electronic properties of materials. Conventional screening via trial-and-error strategy is challenging owing to enormous chemical space. We proposed connected convolutional neutral network (CCNN) for quick boron nitrogen (B–N) codoped graphdiyne terms band gap. A paired-atomic localized matrix (PALM) descriptor was designed describe local environment with form adapted network. An attribution analysis conducted, and quantitative relationship between structure gap proposed, which reveals more significant influence B–N at sp2 hybridized sites than sp on broadening GDY. The accuracy efficiency approach implicate its potential promoting design graphdiyne-based optoelectronic devices catalysts expected properties, opening new avenue rational novel

Language: Английский

---