Bayesian Optimization‐guided Discovery of High‐performance Methane Combustion Catalysts based on Multi‐component PtPd@CeZrOx Core–Shell Nanospheres DOI Open Access

Xilan Feng,

Xiangrui Gong,

Dapeng Liu

et al.

Angewandte Chemie, Journal Year: 2023, Volume and Issue: 135(47)

Published: Oct. 12, 2023

Abstract Formula regulation of multi‐component catalysts by manual search is undoubtedly a time‐consuming task, which has severely impeded the development efficiency high‐performance catalysts. In this work, PtPd@CeZrO x core–shell nanospheres, as successful case study, explicitly demonstrated how Bayesian optimization (BO) accelerates discovery methane combustion with optimal formula ratio (the Pt/Pd mole ranges from 1/2.33–1/9.09, and Ce/Zr 1/0.22–1/0.35), directly results in lower conversion temperature (T 50 approaching to 330 °C) than ones reported hitherto. Consequently, best sample obtained could be efficiently developed after two rounds iterations, containing only 18 experiments all that far less common human workload via traditional trial‐and‐error for compositions. Further, BO‐based machine learning strategy can straightforward extended serve autonomous material systems, other desired properties, showing promising opportunities practical applications future.

Language: Английский

Catalysts informatics: paradigm shift towards data-driven catalyst design DOI Creative Commons
Keisuke Takahashi, Junya Ohyama, Shun Nishimura

et al.

Chemical Communications, Journal Year: 2023, Volume and Issue: 59(16), P. 2222 - 2238

Published: Jan. 1, 2023

This work summarizes how catalysts informatics plays a role in catalyst design.

Language: Английский

Citations

39
Copper–oxygen adducts: new trends in characterization and properties towards C–H activation DOI Creative Commons
Jonathan De Tovar, Rébecca Leblay, Yongxing Wang

et al.

Chemical Science, Journal Year: 2024, Volume and Issue: 15(27), P. 10308 - 10349

Published: Jan. 1, 2024

Recent progresses in Cu–oxygen adducts towards recalcitrant C–H activation are reviewed with focus on Cu metalloenzymes and bioinspired synthetic models, mono- to polynuclear complexes, working under homogeneous heterogeneous catalytic conditions.

Language: Английский

Citations

5
Accelerated Development of Novel Biomass-Based Polyurethane Adhesives via Machine Learning DOI Creative Commons

Ye Cheng,

Takuma Araki,

Naofumi Kamimura

et al.

ACS Applied Materials & Interfaces, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 28, 2025

2-Pyrone-4,6-dicarboxylic acid (PDC) can be produced on a large scale from lignin by transformed bacteria, and its use as bifunctional monomer to synthesize biomass-based polymers has been reported. Recently, excellent adhesive properties of the resulting were also reported, but their performance not yet optimized. In this study, we focus improving PDC-based polyurethanes (PUs) combining experiments machine learning (ML). We synthesized an initial data set 25 samples different polyols isocyanates with isocyanate-to-hydroxyl ratios (r). Adhesive strengths measured after hot-pressing at varying temperatures (Theat, °C) durations (theat, h), following Taguchi L25 orthogonal design. Gaussian process-based Bayesian optimization (BO) was employed identify optimal PU function polyol type, isocyanate r ratio, heating temperature, time improved strength 10.04 ± 1.26 MPa only five iterations. This approach highlights effectiveness BO in guiding experimental conditions for enhanced performance. Random Forest regression used alternative ML supported conclusions. Overall, study demonstrates potential accelerating development novel materials.

Language: Английский

Citations

0
Leveraging machine learning engineering to uncover insights into heterogeneous catalyst design for oxidative coupling of methane DOI Creative Commons
Shun Nishimura, Xinyue Li, Junya Ohyama

et al.

Catalysis Science & Technology, Journal Year: 2023, Volume and Issue: 13(16), P. 4646 - 4655

Published: Jan. 1, 2023

Unveiling current issues in the investigation of highly-active heterogeneous catalysts using machine learning engineering techniques was discussed case oxidative coupling methane with support vector regression and Bayesian optimization.

Language: Английский

Citations

9
Bayesian Optimization‐guided Discovery of High‐performance Methane Combustion Catalysts based on Multi‐component PtPd@CeZrOx Core–Shell Nanospheres DOI

Xilan Feng,

Xiangrui Gong,

Dapeng Liu

et al.

Angewandte Chemie International Edition, Journal Year: 2023, Volume and Issue: 62(47)

Published: Oct. 12, 2023

Formula regulation of multi-component catalysts by manual search is undoubtedly a time-consuming task, which has severely impeded the development efficiency high-performance catalysts. In this work, PtPd@CeZrOx core-shell nanospheres, as successful case study, explicitly demonstrated how Bayesian optimization (BO) accelerates discovery methane combustion with optimal formula ratio (the Pt/Pd mole ranges from 1/2.33-1/9.09, and Ce/Zr 1/0.22-1/0.35), directly results in lower conversion temperature (T50 approaching to 330 °C) than ones reported hitherto. Consequently, best sample obtained could be efficiently developed after two rounds iterations, containing only 18 experiments all that far less common human workload via traditional trial-and-error for compositions. Further, BO-based machine learning strategy can straightforward extended serve autonomous material systems, other desired properties, showing promising opportunities practical applications future.

Language: Английский

Citations

8
Self-Optimizing Bayesian for Continuous Flow Synthesis Process DOI Creative Commons
Runzhe Liu, Zihao Wang, Wenbo Yang

et al.

Digital Discovery, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 1, 2024

A Bayesian algorithm with self-optimizing capabilities, tailored for process optimization in continuous flow synthesis small datasets enhancing efficiency.

Language: Английский

Citations

2
ABO4 as an Active Catalyst Structure for Direct Partial CH4 Oxidation as Identified through Screening of Supported Catalysts DOI
Junya Ohyama,

Yuriko Yoshioka,

Masato TSUKAMOTO

et al.

ACS Catalysis, Journal Year: 2024, Volume and Issue: 15(2), P. 697 - 705

Published: Dec. 24, 2024

In the present study, 76 different metal-oxide-supported-transition-metal catalysts were prepared using 11 metal oxides (MgO, Al2O3, SiO2, TiO2, V2O5, ZrO2, Nb2O5, MoO3, Ta2O5, WO3, and La2O3) seven 3d metals (V, Mn, Fe, Co, Ni, Cu, Zn). The supported catalysts, along with single oxides, screened to identify catalytically active lattice oxygen structures for partial oxidation of CH4 formaldehyde methanol. Fe/MoO3, Fe/V2O5, particularly Fe/Nb2O5 found be highly effective. Structural analysis Fe sites in was performed high-energy-resolution-fluorescence-detected K-edge X-ray absorption near-edge structure spectroscopy, revealing that FeNbO4, FeMoO4, FeVO4 species Fe/Nb2O5, respectively, are responsible their partial-oxidation activities. contrast, Fe2O3 formed Fe/Al2O3, Fe/SiO2, Fe/Ta2O5, Fe/WO3 complete CO2 than oxidation, as MgFe2O4, LaFeO3, TiFe2O5 Fe/MgO, Fe/La2O3, Fe/TiO2, interstitial solid solution Fe3+ ZrO2 generated Fe/ZrO2. Furthermore, while Fe/WO4 ineffective FeWO4 by a hydrothermal method exhibits high selectivity oxidation. Additionally, previous studies have shown CuWO4 CuMoO4 Accordingly, ABO4 (where A is B group 5 or 6 metal) indicated viable design basis development

Language: Английский

Citations

2
Development of graphical user interface for design of experiments via Gaussian process regression and its case study DOI Creative Commons
Yoshiki Hasukawa, Micke Kuwahara, Lauren Takahashi

et al.

Science and Technology of Advanced Materials Methods, Journal Year: 2024, Volume and Issue: 4(1)

Published: Jan. 4, 2024

Bayesian optimization, coupled with Gaussian process regression and acquisition functions, has proven to be a powerful tool in the field of experimental design. Nevertheless, it demands profound proficiency software programming, machine learning, statistical concepts. This steep learning curve presents substantial obstacle when implementing optimization for In order overcome this challenge, user-friendly graphical interface functions is proposed. accessible can readily accessed via web browsers, courtesy established CADS platform (available at https://cads.eng.hokudai.ac.jp/). Thus, offers perform without any programming or extensive prior knowledge about learning.

Language: Английский

Citations

1
Relationship between Redox Rate and Catalytic Activity of Cu Zeolite in the Partial Oxidation of Methane DOI Open Access
Junya Ohyama,

Airi Hirayama,

Hiroshi Yoshida

et al.

Journal of the Japan Petroleum Institute, Journal Year: 2023, Volume and Issue: 66(5), P. 180 - 184

Published: Aug. 31, 2023

CuゼオライトはCH4の部分酸化反応触媒として機能する。これまでに筆者らは様々なCuゼオライトのCH4部分酸化反応に対する触媒活性を評価し，Cu-CHAとCu-MORが比較的高い性能を示すことを明らかにしてきた。さらに，Cu-CHAとCu-MORの酸化還元挙動をin situ Cu K-edge X線吸収微細構造（XAFS）分光法を用いて評価してきた。本研究では，CO2選択性が高かったCu-MFIについて，in XAFS分光法を用いて解析し，その酸化還元速度を評価した。Cu-MFIのデータとこれまでのCu-CHAとCu-MORのデータを合わせて，Cu2＋/＋の酸化還元速度とCH4酸化活性および部分酸化物選択性の関係について調べた。その結果，Cu2＋からCu＋への還元速度はCH4酸化活性と強い相関があることが確認できた。これは，CH4のC–H活性化の際にCu2＋が還元されるためである。一方，Cu2＋/＋の酸化還元速度と部分酸化物選択性との間には相関は認められなかった。

Citations

1
Density Functional Theory-Assisted Active Learning-Driven Organic Ligand Design for CsPbBr3 Nanocrystals DOI
Zhaojie Wang, Yanwei Wen, Feifeng Wu

et al.

The Journal of Physical Chemistry C, Journal Year: 2024, Volume and Issue: 128(8), P. 3557 - 3566

Published: Feb. 20, 2024

Surface modification with organic ligands is pivotal in enhancing the stability and passivation of perovskite nanocrystals. Traditionally, design these has predominantly been dependent on expertise intuition researchers. We develop a density functional theory-assisted active learning framework to screen potential surface for CsPbBr3 nanocrystals large chemical space using dual-objective Bayesian optimization. Our approach successfully identified stable Pareto front after seven optimization iterations, resulting surrogate model demonstrating accurate predictions adsorption energies newly proposed molecules. Six promising candidates solutions without electronic traps are obtained through mere 80 calculations from 161,900 molecule spaces. Conspicuous enrichment featured fragments (halogen, ketone, imine, sulfide, benzene, their combinations) observed data set near front, which coincides features most reported excellent ligands. work demonstrates highly efficient accelerate ligand PNCs by rapidly screening large-scale data-driven workflow.

Language: Английский

Citations

0