Download Hi-Res ImageDownload to MS-PowerPointCite This:J. Phys. Chem. C 2020, 124, 41, 22471-22478

Get e-Alerts

C: Surfaces, Interfaces, Porous Materials, and CatalysisSeptember 16, 2020

A Machine Learning Model on Simple Features for CO₂ Reduction Electrocatalysts
基于 CO₂ 还原电催化剂简单特征的机器学习模型
Click to copy article link
点击复制文章链接Article link copied!

An Chen
An Chen
School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China
More by An Chen
Xu Zhang*
Xu Zhang
School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China
*Email: zhangxu@nankai.edu.cn
More by Xu Zhang
Letian Chen
Letian Chen
School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China
More by Letian Chen
Sai Yao
Sai Yao
School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China
More by Sai Yao
Zhen Zhou*
Zhen Zhou
School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China
*Email: zhouzhen@nankai.edu.cn
More by Zhen Zhou
http://orcid.org/0000-0003-3232-9903

Open PDF Supporting Information (1)

The Journal of Physical Chemistry C

Cite this: J. Phys. Chem. C 2020, 124, 41, 22471–22478

Click to copy citationCitation copied!

https://doi.org/10.1021/acs.jpcc.0c05964

Published September 16, 2020

Request reuse permissions

Article Views

Altmetric

Citations

164

Learn about these metrics

Article Views are the COUNTER-compliant sum of full text article downloads since November 2008 (both PDF and HTML) across all institutions and individuals. These metrics are regularly updated to reflect usage leading up to the last few days.

Citations are the number of other articles citing this article, calculated by Crossref and updated daily. Find more information about Crossref citation counts.

The Altmetric Attention Score is a quantitative measure of the attention that a research article has received online. Clicking on the donut icon will load a page at altmetric.com with additional details about the score and the social media presence for the given article. Find more information on the Altmetric Attention Score and how the score is calculated.

Abstract 抽象

Click to copy section link
单击以复制部分链接Section link copied!

Electroreduction of CO₂ is one of the most potential ways to realize CO₂ recycle and energy regeneration. The key to promoting this technology is the development of high-performance electrocatalysts. Generally, high-throughput computational screening contributes a lot to materials innovation, but still consumes much time and resource. To achieve efficient exploration of electrocatalysts for CO₂ reduction, we created a machine learning model based on an extreme gradient boosting regression (XGBR) algorithm and simple features. Our screening model successfully and rapidly predicted the Gibbs free energy change of CO adsorption (ΔG_CO) of 1060 atomically dispersed metal–nonmetal codoped graphene systems, and greatly reduced the research cost. The competitive reaction, the hydrogen evolution reaction (HER), is also discussed with respect to such a screening model. This work demonstrates the potential of machine learning methods and provides a convenient approach for the effective theoretical design of electrocatalysts for CO₂ reduction.
电还原 CO₂ 是实现 CO₂ 回收和能源再生的最有潜力的方法之一。推广这项技术的关键是开发高性能电催化剂。一般来说，高通量计算筛选对材料创新有很大贡献，但仍然会消耗大量的时间和资源。为了实现对用于 CO₂ 还原的电催化剂的有效探索，我们创建了一个基于极端梯度提升回归（XGBR）算法和简单特征的机器学习模型。我们的筛选模型成功快速地预测了 1060 个原子分散的金属-非金属共掺杂石墨烯体系的 CO 吸附（ΔG_CO）的吉布斯自由能变化，大大降低了研究成本。还讨论了竞争反应，即析氢反应（HER），与这种筛选模型有关。这项工作展示了机器学习方法的潜力，并为还原 CO₂ 的电催化剂的有效理论设计提供了一种便捷的方法。

This publication is licensed under the terms of your institutional subscription. Request reuse permissions.
本出版物根据机构认购。请求重用权限。

Subjects

Special Issue 所属专题

Published as part of The Journal of Physical Chemistry virtual special issue “Machine Learning in Physical Chemistry”.
作为 The Journal of Physical Chemistry 虚拟特刊“Machine Learning in Physical Chemistry”的一部分发布。

Introduction 介绍

Click to copy section link
单击以复制部分链接Section link copied!

Traditional energy conversion technologies are unable to meet the current demand. Thus, sustainable development and reuse of energy are of great significance nowadays. To achieve larger-scale energy reuse, new strategies are urgently needed. Electroreduction of carbon dioxide is an important process of sustainable production of fuels and value-added chemicals. (1) For this process, electrocatalysts are decisive. Bulk metals and alloys, (2,3) nonmetals, (4,5) and nanostructured materials (6,7) have been applied to this domain. Moreover, experiments and computations have demonstrated successful applications of carbon-based materials toward electroreduction of CO₂. (8−10) A kind of modified carbon-based catalysts with isolated single-atom sites, generally called single-atom catalysts (SACs), (11) shows the potential for the electroreduction process due to its distinctive structures and properties. Pan et al. (12) designed such catalysts for CO₂ electroreduction, with N-doped porous carbon spheres as the substrate to disperse Co atoms. These unique catalysts show both excellent selectivity and high stability for CO₂ electroreduction. Liu et al. (13) introduced Fe-doped graphdiyne (Fe/GDY) to CO₂ eletroreduction by means of density functional theory (DFT) computations. The results reveal the application potential of Fe/GDY, and the stability is also guaranteed by strong hybridization between Fe and GDY. Taking this kind of unique and controllable structures as the substrate brings many new ideas to materials design, which extends the content of novel catalysts.
传统的能源转换技术无法满足当前的需求。因此，能源的可持续发展和再利用在当今具有重要意义。为了实现更大规模的能源再利用，迫切需要新的策略。二氧化碳的电解还原是可持续生产燃料和增值化学品的重要过程。（1）对于此过程，电催化剂是决定性的。块状金属和合金、（2,3）非金属、（4,5）和纳米结构材料（6,7）已应用于该领域。此外，实验和计算已经证明了碳基材料在电还原 CO₂ 方面的成功应用。（8−10）一种具有孤立的单原子位点的改性碳基催化剂，通常称为单原子催化剂（SAC），（11）由于其独特的结构和性质，显示出电还原过程的潜力。Pan 等人（12）设计了这种用于 CO₂ 电还原的催化剂，以 N 掺杂多孔碳球作为分散 Co 原子的基材。这些独特的催化剂对 CO₂ 电还原表现出优异的选择性和高稳定性。Liu 等人（13）通过密度泛函理论（DFT）计算将 Fe 掺杂石墨炔（Fe/GDY）引入 CO₂ 电还原。结果揭示了 Fe/GDY 的应用潜力，Fe 和 GDY 的强杂化也保证了稳定性。以这种独特可控的结构为基材，为材料设计带来了许多新的思路，扩展了新型催化剂的内涵。

High-throughput experiments and computations have contributed much to the accurate preparation of new materials with sophisticated structures for CO₂ electroreduction. However, the cost of time and money for high-throughput investigations is huge, and the development of novel SACs is constricted. At this point, it is very urgent to optimize the efficient screening methods for SACs.
高通量实验和计算为准确制备具有复杂结构的 CO₂ 电还原新材料做出了重大贡献。然而，高通量研究的时间和金钱成本是巨大的，并且新型 SAC 的开发受到限制。在这一点上，优化 SACs 的高效筛选方法非常紧迫。

Machine learning (ML) has been applied to explore complicated materials for energy conversion, (14) such as MXenes, (15) perovskites, (16−18) and metal–organic frameworks (MOFs). (19−21) For SACs, machine learning has also been used as a powerful auxiliary tool. Zhu and co-workers (22) applied machine learning models based on DFT computations to explore the design principles of dual-metal-site catalysts (DMSCs) for the oxygen reduction reaction (ORR). By the best prediction model, several novel DMSCs were filtered with outstanding ORR activity compared with that of platinum. By using a deep neural network (DNN) algorithm, Zafari et al. (23) predicted three novel SACs for the nitrogen reduction reaction (NRR) from the results gained from a machine learning model. However, machine learning investigations on single-atom electrocatalysts for CO₂ reduction are lacking, and this technology is definitely helpful for this purpose. To be more comprehensive, ML-assisted materials design always selects features from both geometrical and electronic structures. In fact, the databases of complicated materials, especially for novel materials not prepared experimentally yet, are generally established by computations. If too many features are considered, the computational cost will be multiplied.
机器学习（ML）已被应用于探索用于能量转换的复杂材料，（14），例如 MXenes、（15）钙钛矿、（16-18）和金属有机框架（MOF）。（19−21）对于 SAC，机器学习也被用作强大的辅助工具。Zhu 及其同事（22）应用基于 DFT 计算的机器学习模型来探索用于氧还原反应（ORR）的双金属位催化剂（DMSC）的设计原理。通过最佳预测模型，过滤了几种新型 DMSC，与铂相比，具有出色的 ORR 活性。通过使用深度神经网络（DNN）算法，Zafari 等人（23）根据机器学习模型获得的结果预测了氮还原反应（NRR）的三种新型 SAC。然而，缺乏对用于 CO₂ 还原的单原子电催化剂的机器学习研究，这项技术绝对有助于实现这一目的。更全面地说，ML 辅助材料设计始终从几何和电子结构中选择特征。事实上，复杂材料的数据库，特别是尚未通过实验制备的新型材料的数据库，通常是通过计算建立的。如果考虑的特征过多，计算成本将成倍增加。

In this work, we use machine learning with simple and easily accessible features to train prediction models that can accurately describe the catalytic performance of single-atom catalysts with dispersed metal–nonmetal codoped graphene structures (for convenient expression, this kind of materials will be presented by SACs) (Figure 1). Combining with the excellent properties of this kind of materials, our work for mining novel SACs can accelerate the development of CO₂ reduction electrocatalysts. The first part is building a database for the target electrocatalysts by means of DFT computations. Then we focus on the establishment of machine learning models. Finally, we predict potential catalysts for CO₂ electroreduction through predicted ΔG_CO, which was widely used to explore CO₂ electroreduction mechanisms and catalytic performance in previous studies. (24−26) Our work provides guidance to explore CO₂ reduction electrocatalysts, and the selection of simple features is also helpful for other fast DFT-machine learning combining methods.
在这项工作中，我们使用具有简单易懂功能的机器学习来训练预测模型，这些模型可以准确描述具有分散金属-非金属共掺杂石墨烯结构的单原子催化剂的催化性能（为方便表达，这类材料将由 SAC 呈现）（图 1）。结合这种材料的优异性能，我们开采新型 SAC 的工作可以加速 CO₂ 还原电催化剂的开发。第一部分是通过 DFT 计算为目标电催化剂构建数据库。然后我们专注于建立机器学习模型。最后，我们通过预测的 ΔG_CO 预测 CO₂ 电还原的潜在催化剂，这在以前的研究中被广泛用于探索 CO₂ 电还原机制和催化性能。（24−26）我们的工作为探索 CO₂ 还原电催化剂提供了指导，简单特征的选择也有助于其他快速 DFT 机器学习组合方法。

Figure 1. Structure of SAC. Shallow blue, green, and gray balls represent transition metal (Ag, Au, Co, Cu, Fe, Ir, Mn, Mo, Ni, Os, Pd, Pt, Rh, Ru, Sc, Ti, V, Y, Zn or Zr), nonmetal (C, N, O, P, or S), and C atoms, respectively.
图 1.SAC 的结构。浅蓝色、绿色和灰色球分别代表过渡金属（Ag、Au、Co、Cu、Fe、Ir、Mn、Mo、Ni、Os、Pd、Pt、Rh、Ru、Sc、Ti、V、Y、Zn 或 Zr）、非金属（C、N、O、P 或 S）和 C 原子。

Computational Details 计算详细信息

Click to copy section link
单击以复制部分链接Section link copied!

Establishment of a Database
建立数据库

All the data were obtained from DFT computations that were performed with a plane-wave basis set as implemented in the Vienna ab initio simulation package (VASP). (27) The magnetic ions were initialized ferromagnetically. The projector augmented wave (PAW) was adopted to describe the ion-electron interactions. (28) Revised Perdew–Burke–Ernzerhof (RPBE) (29) was used to describe the adsorption. An energy cutoff of 450 eV was employed for the plane-wave basis set. A Monkhorst–Pack k point mesh of 4 × 5 × 1 was used to sample the first Brillouin zone. (30) Considering the strong electron correlation of some transition metal elements, the DFT+U with U-J parameters was used, and the corresponding U-J parameters were used in accordance with a previous report. (31) To prevent any artificial interactions between two neighboring structures, a vacuum space with 20 Å was inserted in the corresponding direction. To evaluate the stability of the catalysts, ab initio molecular dynamics (AIMD) simulations were performed in a NVT ensemble and lasted for 10 ps with a time step of 2 fs at 600 K. The temperature was controlled by the Nosé–Hoover method. (32)
所有数据均来自 DFT 计算，这些计算是使用 Vienna ab initio 仿真包（VASP）中实现的平面波基集执行的。（27）磁离子以铁磁方式初始化。采用投影仪增强波（PAW）来描述离子-电子相互作用。（28）修订的 Perdew-Burke-Ernzerhof （RPBE）（29）用于描述吸附。平面波基集采用 450 eV 的能量截止。使用 4 × 5 × 1 的 Monkhorst-Pack k 点网格对第一个布里渊区进行采样。（30）考虑到一些过渡金属元素的强电子相关性，使用了具有 U-J 参数的 DFT+U，并根据以前的报告使用了相应的 U-J 参数。（31）为了防止两个相邻结构之间发生任何人为的相互作用，在相应的方向上插入了一个 20 Å 的真空空间。为了评估催化剂的稳定性，在 NVT 集合中进行了从头计算分子动力学（AIMD）模拟，并在 600 K 下以 2 fs 的时间步长持续 10 ps。温度由 Nosé-Hoover 方法控制。（32）

The free energy change during adsorption of an adsorbate X can be calculated by (33)
吸附物 X 吸附过程中的自由能变化可由下式计算（33）

where μ_X* is the chemical potential of the adsorbate-catalyst complex, μ_X is the chemical potential of the adsorbate in the gas phase, and μ_* is the chemical potential of the catalyst. The chemical potential can be computed according to (33)
其中 μ_X* 是吸附物-催化剂复合物的化学势，μ_X 是吸附物在气相中的化学势，_μ* 是催化剂的化学势。化学势可以根据（33）计算

in which E is the energy calculated by DFT, E_ZPE is zero point energy which was computed based on the vibrational frequencies, C_p is the heat capacity, T represents temperature, S is entropy which could be obtained from NIST (National Institute of Standards and Technology) database for gas phase, Δμ_solv stands for the change in chemical potential from solvent stabilization, and Δμ_corr means the experimental correction to account for the difference between experimental chemical potentials and DFT-based chemical potentials.
其中 E 是由 DFT 计算的能量，E_ZPE 是根据振动频率计算的零点能量，CP_是热容，T 代表温度，S 是熵，可以从 NIST（美国国家标准与技术研究所）气相数据库获得，Δμ_solv 代表溶剂稳定引起的化学势变化， Δμ_corr 表示解释实验化学势和基于 DFT 的化学势之间差异的实验校正。

According to previous reports, (33,34) ∫C_p,_CO dT is 0.091 eV at 298 K and Δμ_corr,CO is 0.02 eV for CO gas. For adsorbed CO, ∫C_p,_CO* dT is 0.085 eV. For adsorbates, only vibrational entropy contributions were considered, which is consistent with a previous report. (35) The values of Δμ_solv,_CO* and Δμ_corr,_CO* are in agreement with those in a previous report. (33)
根据以前的报告，（33,34） ∫C_p，CO dT 在 298 K 时为 0.091 eV，Δμ_corr，CO 为 0.02 eV。对于吸附的 CO，∫C_p，CO_* dT 为 0.085 eV。对于吸附物，只考虑了振动熵贡献，这与之前的报告一致。（35） Δμ_solv，CO_* 和 Δμ_corr，CO_* 的值与之前报告中的值一致。（33）

SAC structures of the database are presented in Figure S1. Not only new SACs are included in this data set, but also some reported structures, such as Mn–N2C2, (36) Ni–N2O2, Ni–N4, (37) and Co–N4. (38)
数据库的 SAC 结构如图 S1 所示。该数据集中不仅包括新的 SAC，还包括一些已报道的结构，例如 Mn-N2C2、（36） Ni-N2O2、Ni-N4、（37）和 Co-N4。（38）

Machine Learning Methods 机器学习方法

In this work, we tried more than one ML algorithm, such as K-nearest neighbor regression (KNR), random forest regression (RFR), support vector regression (SVR), gradient boosting regression (GBR), extreme gradient boosting regression (39) (XGBR), and a kind of composited algorithms produced by TPOT (tree-based pipeline optimization tool). (40) For each algorithm, we tested various pairs of training and testing data with different separated ratios to gain ideal regression models.
在这项工作中，我们尝试了不止一种 ML 算法，例如 K 最近邻回归（KNR）、随机森林回归（RFR）、支持向量回归（SVR）、梯度提升回归（GBR）、极端梯度提升回归（39）（XGBR），以及 TPOT （tree-based pipeline optimization tool）产生的一种复合算法。（40）对于每种算法，我们测试了具有不同分离比率的各种训练和测试数据对，以获得理想的回归模型。

The performance of feature sets was evaluated by the Pearson correlation coefficient (p), which is determined by
特征集的性能由 Pearson 相关系数（p）评估，该系数由

in which f_i and F_i are the compared features. The value of p is in the range of −1.0 to 1.0, and a higher absolute value of p presents a stronger correlation.
其中 f 和 F 是比较的特征。p 的值在 −1.0 到 1.0 的范围内，p 的绝对值越高，相关性越强。

The coefficient of determination (r (2)) and root mean squared error (RMSE) were used to reflect the prediction accuracy of the models, and are defined as
决定系数（r（2））和均方根误差（RMSE）用于反映模型的预测准确性，定义为

where Y_i indicates the value of the DFT computations, y_i indicates the result predicted by the ML models, and Y̅ indicates the average value of the DFT data. Generally, an ideal model should have r² value close to 1, and a small RMSE close to 0.
其中 Y 表示 DFT 计算的值，y 表示 ML 模型预测的结果，Y̅ 表示 DFT 数据的平均值。通常，理想模型的 r² 值应接近 1，小 RMSE 接近 0。

We implemented the computations with the Scikit-Learn package. (41) Important parameters of machine learning models are given in Table S1, Supporting Information.
我们使用 Scikit-Learn 包实现了计算。（41）机器学习模型的重要参数在表 S1 支持信息中给出。

Results and Discussion 结果与讨论

Click to copy section link
单击以复制部分链接Section link copied!

Determination of Target Features for CO₂ Electroreduction
确定 CO₂ 电还原的目标特性

Before the establishment of prediction models, the evaluation parameter for the CO₂ reduction reaction (CO₂RR) and target features for machine learning models should be determined. Thermodynamic energies of microkinetic models or experimental characterizations have been used to represent catalytic performance. Moreover, it is a common and mature approach to use the adsorption energies or the Gibbs free energy change of intermediates to assess and predict electrocatalytic performances, such as hydrogen evolution reaction (HER) and NRR. In this work, toward CO₂RR, we chose the widely accepted computational hydrogen electrode model to present catalytic reactions and develop free energy profiles for the reactions. (42) As is known, CO is an important intermediate of multicarbon hydrocarbons and oxygenates in CO₂RR, (43) and previous relevant studies demonstrated that the CO adsorption energy is an ideal value to describe catalytic performance for CO₂RR. (44,45) Although choosing a single value as the evaluation criterion and target feature would undoubtedly lead to defective results, the original intention of our work is to preliminarily screen new catalysts efficiently with low research cost. Thus, it is reasonable to consider only vital features for the balance between accuracy and consumption.
在建立预测模型之前，应确定 CO₂ 还原反应（CO₂RR）的评估参数和机器学习模型的目标特征。微动力学模型或实验表征的热力学能量已被用于表示催化性能。此外，利用中间体的吸附能或吉布斯自由能变化来评估和预测析氢反应（HER）和 NRR 等电催化性能是一种常见且成熟的方法。在这项工作中，针对 CO₂RR，我们选择了被广泛接受的计算氢电极模型来呈现催化反应并为反应开发自由能曲线。（42）众所周知，CO 是 CO₂RR 中多碳烃和含氧化合物的重要中间体，（43）以前的相关研究表明，CO 吸附能是描述 CO₂RR 催化性能的理想值。（44,45）虽然选择单一值作为评价标准和目标特征无疑会导致结果有缺陷，但我们工作的初衷是以较低的研究成本初步有效地筛选新的催化剂。因此，为了在准确性和消耗之间取得平衡，只考虑重要特征是合理的。

Selection and Optimization of Features
特征的选择和优化

An ideal feature set is entrusted with the function of building good models and extracting physical rules from models. In our work, we considered three criteria for selecting suitable features: (1) Features should be easily accessible. (2) Since the catalytic activity is generally dependent on the electronic structure, features should relate to it. (3) Features can simply describe the geometry of the catalysts. We took no account of features that need DFT computations; thus, feature engineering of our study is more circumspect. This consideration for feature selection is acceptable due to a sharp reduction in research cost of structure optimization before prediction. We chose the number of d electrons of metal atoms (d), electron affinity of metal atoms (E_affi_M), Pauling electronegativity of metal atoms (E_M), first ionization energy of metal atoms (1E_M), atomic number of metal atoms (M_Z), the number of nonmetal atoms (n_NM), average electron affinity of nonmetal atoms (E_affi_NM), average Pauling electronegativity of nonmetal atoms (E_aNM), average covalent radius of nonmetal atoms (NM_acov), and average atomic number of nonmetal atoms (NM_aZ) as the initial features. Simple features such as atomic number were directly gained from materials. Electronic property features were obtained from the databases of NIST. However, some structural features which need DFT computations were excluded; thus, the final model can be used without geometric optimization for designed materials. For further describing the relationship between different atoms, the initial feature set should be extended. Features in the initial feature set were combined by simple mathematical operation. Composite features, such as E_aNM-E_M (value of the difference of electronegativity between metal and nonmetal atoms) and M_cov+NM_acov (value of the sum of covalent radius of metal atoms and average atomic number of nonmetal atoms), were added to a more complex feature set. Before the training process, the performance of the feature set is indeterminate. Excessive features will lead to redundant problems, and inadequate features cannot meet the basic demand of model establishment. In order to optimize the feature set, we analyzed the correlation between every two features by Pearson correlation coefficient heat map and the rank of feature importance. The color of each small square of the heat maps shows the correlation of feature pairs. Moreover, the rank of feature importance can clearly distinguish important features from the feature set. Thus, combining these two criteria, it is convenient to gain independent features and can promote the establishment of models.
一个理想的特征集被赋予了构建良好模型和从模型中提取物理规则的功能。在我们的工作中，我们考虑了选择合适的特征的三个标准：（1）特征应该易于访问。（2）由于催化活性通常取决于电子结构，因此特征应与其相关。（3）特征可以简单地描述催化剂的几何形状。我们没有考虑需要 DFT 计算的特征;因此，我们研究的特征工程更加谨慎。这种对特征选择的考虑是可以接受的，因为在预测之前结构优化的研究成本急剧降低。我们选择了金属原子的 d 电子数（d）、金属原子的电子亲和力（E_affi_M）、金属原子的鲍林电负性（E_M）、金属原子的第一电离能（1E_M）、金属原子的原子序数（M_Z）、非金属原子数（n_NM）、非金属原子的平均电子亲和力（E_affi_NM），非金属原子的平均鲍林电负性（E_aNM）、非金属原子的平均共价半径（NM_acov）和非金属原子的平均原子序数（NM_aZ）作为初始特征。原子序数等简单特征是直接从材料中获得的。电子属性特征是从 NIST 的数据库中获得的。然而，一些需要 DFT 计算的结构特征被排除在外;因此，最终模型可以在不对设计材料进行几何优化的情况下使用。为了进一步描述不同原子之间的关系，应扩展初始特征集。初始特征集中的特征通过简单的数学运算进行组合。复合特征，如 E_aNM-E_M （金属和非金属原子之间的电负性差值）和 M_cov+NM_acov （金属原子的共价半径之和与非金属原子的平均原子序数的值）被添加到更复杂的特征集中。在训练过程之前，特征集的性能是不确定的。特征过多会导致冗余问题，特征不足无法满足模型建立的基本需求。为了优化特征集，我们通过 Pearson 相关系数热图和特征重要性等级分析了每两个特征之间的相关性。热图的每个小方块的颜色显示特征对的相关性。此外，特征重要性的等级可以清楚地区分重要特征和特征集。因此，结合这两个标准，可以方便地获得独立的特征，并可以促进模型的建立。

Model Training and Performance
模型训练和性能

Different machine learning methods show different prediction performances even if the same feature set and training data are used. In this study, we mainly used the XGBR algorithm to build the model. KNR, RFR, SVR, and GBR were chosen as training algorithms to build prediction models for comparing and gaining the most suitable model. We also used a complicated machine learning method produced by TPOT, which can analyze thousands of possible combinations to find the best one for the models and parameters by genetic algorithm. Model selection and parameter adjustment of TPOT are totally automated. However, this complex and automation algorithm has two sides. One is that the algorithm has high prediction accuracy, and the other is that the multilayer structure of TPOT models makes the further adjustment of superparameters a challenge, namely, the improvement of TPOT models is out of control and restricted. Thus, we would not further improve this method in the following. Our data set (including DFT computational data of 171 designed SACs) was randomly separated to training set (80%) and testing set (20%), which is the best split ratio for most algorithms. The model performance figures of other ratios are shown in Figure S2. After the first training process, among all the trained models, the model trained by the XGBR algorithm showed the best performance. However, much credit of the successful establishment of this model must be attributed to the feature set optimization. The feature set might restrict the prediction accuracy when all the composite features were applied to train models, and cumbersome feature sets are wasteful. The model trained by the feature set including all the single features and mathematic-transformation composited features shows a prediction accuracy of r² = 0.902 and RMSE = 0.1652. Some composite features have good independence, such as d^∧2 and E_affi_M^∧2. However, in the heat map, there are also some composite features analogous to corresponding single features. Thus, we removed the lowest ranking and miscellaneous features, such as E_M^∧2, d^∧2, NM_acov^∧2, and 1E_M^∧2. To find the minimum of the number of features, we trained the model by the simplest feature set which only included independent and low-correlation coefficient features. The prediction performance of this model does not have a superiority performance score (r² = 0.797, RMSE = 0.2376), which may be due to the incomplete description of target values by inadequate features. Comparison Pearson correlation coefficient heat maps and feature importance rank of those two feature sets are shown in Figures S3 and S4. The prediction performances of these feature sets are shown in Figure S5. Then we eliminated the features of low importance and poor independence one by one, so as to reduce the redundant features and improve the prediction accuracy at the same time. Features were filtered step by step until the final feature set emerged with suitable size and provided assurance for model accuracy. By removing features without contribution from the biggest feature set, we gained the final feature set. Models built by the final feature set maintain the same r² (0.902) and RMSE (0.1652 eV) as the biggest feature set. The 5-fold cross validation results are presented in Table S2. The Pearson correlation coefficient heat map and the feature importance rank of the final feature set are shown in Figure 2a,b, and the final selected feature set demonstrates excellent adaptability to our data set. The comparison of DFT data and predicted results by the XGBR model is shown in Figure 2c, and comparison figures of other algorithm models are presented in Figure 3. After training by the XGBR algorithm, we successfully gained an acceptable prediction model for this kind of SACs. Other models show inferior prediction accuracy to the XGBR model, and the closest models are RFR and GBR. KNR and SVR models have unsatisfactory prediction performances because of low r² score and high RMSE value.
即使使用相同的特征集和训练数据，不同的机器学习方法也会表现出不同的预测性能。在本研究中，我们主要使用 XGBR 算法来构建模型。选择 KNR 、 RFR 、 SVR 和 GBR 作为训练算法来构建预测模型，以比较和获得最合适的模型。我们还使用了 TPOT 生成的复杂机器学习方法，该方法可以通过遗传算法分析数千种可能的组合，以找到最适合模型和参数的组合。TPOT 的模型选择和参数调整是完全自动化的。但是，这种复杂的自动化算法有两个方面。一是算法预测精度高，二是 TPOT 模型的多层结构使得超参数的进一步调整成为挑战，即 TPOT 模型的改进失控和受限。因此，我们在下面不会进一步改进这种方法。我们的数据集（包括 171 个设计 SAC 的 DFT 计算数据）被随机分为训练集（80%）和测试集（20%），这是大多数算法的最佳拆分比。其他比率的模型性能数据如图 S2 所示。经过第一次训练过程后，在所有训练的模型中，XGBR 算法训练的模型表现出最佳性能。然而，成功建立此模型的大部分功劳必须归功于功能集优化。当所有复合特征都应用于训练模型时，特征集可能会限制预测准确性，而繁琐的特征集会浪费。由特征集（包括所有单个特征和数学变换复合特征）训练的模型显示预测精度为 r² = 0.902 和 RMSE = 0.1652。一些复合特征具有良好的独立性，例如 d^∧2 和 E_affi_M^∧2。但是，在热图中，也有一些类似于相应的单个特征的复合特征。因此，我们删除了排名最低的杂项特征，例如 ^E_M∧2、d^∧2、NM_acov^∧2 和 1E_M^∧2。为了找到特征数量的最小值，我们用最简单的特征集训练了模型，该特征集只包括独立和低相关系数的特征。该模型的预测性能没有优越性性能得分（r² = 0.797， RMSE = 0.2376），这可能是由于特征不足对目标值的描述不完整。比较：这两个特征集的 Pearson 相关系数、热图和特征重要性排名如图 S3 和 S4 所示。这些特征集的预测性能如图 S5 所示。然后，我们逐一剔除低重要性和独立性差的特征，从而减少冗余特征，同时提高预测精度。逐步筛选特征，直到最终的特征集出现合适的大小，并为模型准确性提供保证。通过从最大的特征集中删除没有贡献的特征，我们获得了最终的特征集。由最终特征集构建的模型保持与最大特征集相同的 r² （0.902）和 RMSE （0.1652 eV）。表 S2 中显示了 5 倍交叉验证结果。皮尔逊相关系数热图和最终特征集的特征重要性排名如图 2a、b 所示，最终选择的特征集展示了对我们数据集的出色适应性。DFT 数据与 XGBR 模型预测结果的比较如图 2c 所示，其他算法模型的比较图如图 3 所示。经过 XGBR 算法的训练，我们成功地获得了这种 SAC 的可接受预测模型。其他模型的预测精度不如 XGBR 模型，最接近的模型是 RFR 和 GBR。由于 r² 分数低，RMSE 值高，KNR 和 SVR 模型的预测性能不令人满意。

Figure 2. (a) Pearson correlation coefficient matrix heat maps of the best feature set for ΔG_CO prediction. (b) Feature importance rank for the best feature set. (c) Prediction results of XGBR model vs DFT data.
图 2.（a） ΔG_CO 预测最佳特征集的 Pearson 相关系数矩阵热图。（b）最佳特征集的特征重要性排名。（c） XGBR 模型与 DFT 数据的预测结果。

Figure 3. Prediction performance of other models trained by different ML methods: (a) KNR, (b) RFR, (c) TPOT, (d) SVR, and (e) GBR.
图 3.由不同 ML 方法训练的其他模型的预测性能：（a） KNR、（b） RFR、（c） TPOT、（d） SVR 和（e） GBR。

Screening of Potential Candidates by the Prediction Model
通过预测模型筛选潜在候选人

After gaining the prediction model, we can use it to efficiently search potential SACs for CO₂ electroreduction by predicting ΔG_CO. Some designed structures are presented in Figure 4, and Figure 5 illustrates the heat map of ΔG_CO for 1060 designed SAC materials consisting of 20 transition metal atoms and 53 different nonmetal bonded environments. The results were predicted on the basis of the feature set that we determined above. From the heat map in Figure 5, potential materials can be quickly filtered by distinct colors. Each number in the x axis represents a nonmetal bonded environment, and the corresponding environments can be found in Table S3. The y axis represents the kind of metal atoms. SACs represented by brown color have strong CO adsorption, indicating that the catalytic activity of subsequent reactions will be restricted. Taking the ΔG_CO of available good electrocatalysts as a reference, (33,46) SACs presented by blue color have ideal ΔG_CO (−0.2 ± 0.1 eV). From this heat map, if we focus on doping metal atoms, structures with Co, Fe, Ir, Ni, Os, Sc, Ti, V, Y, and Zr are included in the candidate materials. For most nonmetallic coordination environments, such as M-C2S2, M-CNO2, M-CN2P, M-N2P2, M-CNS2, M-NS3, and M-OS3 structures, there is at least one suitable SAC structure. Moreover, some materials which are screened with our prediction model have been reported, such as M-C2N2 (36) and M-N2O2. (47) For a well-defined screening, we framed all the candidate structures in red lines.
获得预测模型后，我们可以使用它通过预测 ΔG_CO 来有效地搜索 CO₂ 电还原的潜在 SAC。图 4 列出了一些设计的结构，图 5 说明了由 20 个过渡金属原子和 53 种不同的非金属键合环境组成的 1060 种设计的 SAC 材料的 ΔG_CO 热图。结果是根据我们上面确定的特征集预测的。从图 5 的热图中，可以按不同的颜色快速筛选潜在材料。x 轴中的每个数字代表一个非金属粘合环境，相应的环境可以在表 S3 中找到。y 轴表示金属原子的种类。以棕色为代表的 SACs 具有很强的 CO 吸附性，表明后续反应的催化活性将受到限制。以可用良好电催化剂的 ΔG_CO 为参考，（33,46）蓝色表示的 SAC 具有理想的 ΔG_CO （-0.2 ± 0.1 eV）。从这张热图中可以看出，如果我们专注于掺杂金属原子，候选材料中包括了 Co、Fe、Ir、Ni、Os、Sc、Ti、V、Y 和 Zr 的结构。对于大多数非金属配位环境，例如 M-C2S2、M-CNO2、M-CN2P、M-N2P2、M-CNS2、M-NS3 和 M-OS3 结构，至少有一个合适的 SAC 结构。此外，已经报道了一些用我们的预测模型筛选的材料，例如 M-C2N2 （36）和 M-N2O2。（47）为了进行定义明确的筛选，我们将所有候选结构框在红线中。

Figure 4. Structures M-C2S2 and M-CNO2 as the designed SACs. Gray, navy blue, red, yellow, and shallow blue balls represent C, N, O, S, and transition metal atoms, respectively.
图 4.结构 M-C2S2 和 M-CNO2 作为设计的 SAC。灰色、海军蓝、红色、黄色和浅蓝色球分别代表 C、N、O、S 和过渡金属原子。

Figure 5. Prediction heat map of ΔG_CO for designed single-atom catalysts.
图 5.设计的单原子催化剂的 ΔG_CO 预测热图。

As is well known, HER is a possible concomitant reaction during CO₂ electroreduction. Thus, we also applied our machine learning method to evaluate the electrocatalytic activity of all designed materials for HER. By using the adsorption Gibbs free energy change of H (ΔG_H) as the prediction target, (48) we gained an acceptable prediction model after sorting out DFT data and selecting appropriate feature sets and machine learning algorithms. To gain an accurate and widely applicable model, we added more features, atomic number of metal atoms (M_Z), atomic radius of metal and nonmetal atoms (ar_M, ar_NM1, ar_NM2, ar_NM3, ar_NM4), d orbital occupancy (doo_M), and d band center of metal atoms (TM_d_band_center) which was calculated based on the data of MP (Materials Project) database, to our initial feature set. The prediction performances of different data split ratios and machine learning algorithms are presented in Figures S6 and S7. The best prediction model was trained by by XGBR algorithm with the 0.85:0.15 data split ratio. Figure S8 presents the Pearson correlation coefficient and feature importance of this final selected feature set. For HER, the absolute value of ΔG_H should be close to 0, which means that the HER process is prone to appear in catalytic reactions. (49) The prediction results of ΔG_H are also shown in Figure S9. The range of 0 ± 0.1 eV is presented by red color in the heat map of ΔG_H. From the results of HER catalytic activity, we can further filter our designed materials. Because the mixture of different colors will affect our resolution, we used an overlapping method to gain comprehensive analysis of the final potential materials. First, we overlapped the two prediction heat maps. Because the mixed color of blue and red is purple, if the candidate CO₂ electroreduction catalyst presented by blue squares turns into purple ones, it means that the catalyst also has good catalytic activity for HER. In Figure 6, if the structure squares in the red frame do not show purple color, the structures are the final choice. For example, Co-CS3, Fe–C2S2, Ni–C2NP, Sc-CN3, Ti–C2S2, V-NP3, and Zr-CN2S are considered to have good CO₂ electroreduction catalytic performance.
众所周知，HER 是 CO₂ 电还原过程中可能发生的伴随反应。因此，我们还应用了我们的机器学习方法来评估所有为 HER 设计的材料的电催化活性。以吸附吉布斯自由能变化 H （ΔG_H）作为预测目标，（48）在整理 DFT 数据并选择合适的特征集和机器学习算法后，我们得到了一个可接受的预测模型。为了获得准确且应用广泛的模型，我们增加了更多特征，金属原子的原子序数（M_Z）、金属和非金属原子的原子半径（ar_M、ar_NM1、ar_NM2、ar_NM3、ar_NM4）、d 轨道占用度（doo_M）和金属原子的 d 波段中心（TM_d_band_center）这些都是根据 MP （Materials Project）数据库的数据计算的，添加到我们的初始功能集。图 S6 和 S7 显示了不同数据拆分比率和机器学习算法的预测性能。最佳预测模型采用 XGBR 算法训练，数据分割比为 0.85：0.15。图 S8 显示了最终选择的特征集的 Pearson 相关系数和特征重要性。对于 HER，ΔG_H 的绝对值应接近 0，这意味着 HER 过程容易出现在催化反应中。（49） ΔG_H 的预测结果也显示在图 S9 中。0 ± 0.1 eV 的范围在 ΔG_H 的热图中用红色表示。根据 HER 催化活性的结果，我们可以进一步过滤我们设计的材料。因为不同颜色的混合会影响我们的分辨率，所以我们使用了重叠方法来全面分析最终的潜在材料。首先，我们将两个预测热图重叠。由于蓝色和红色的混合色是紫色，如果蓝色方块呈现的候选 CO₂ 电还原催化剂变成紫色，则说明该催化剂对 HER 也具有良好的催化活性。在图 6 中，如果红框中的结构方块没有显示紫色，则结构是最终选择。例如，Co-CS3、Fe-C2S2、Ni-C2NP、Sc-CN3、Ti-C2S2、V-NP3 和 Zr-CN2S 被认为具有良好的 CO₂ 电还原催化性能。

Figure 6. Overlapped prediction heat maps of ΔG_CO and ΔG_H of designed SACs.
图 6.设计 SAC 的 ΔG_CO 和 ΔG_H 的重叠预测热图。

To verify our predicted potential electrocatalysts, we randomly selected three designed materials, Sc-C2O2, Sc-CN2O, and Y-CN2O for further DFT computations, and checked the stability of these candidate materials through AIMD simulations. The AIMD results are shown in Figure S10, after a simulation at 600 K for 10 ps, the structures are intact and it demonstrates that these materials are stable. Moreover, we also compared the predicted ΔG_CO and the DFT computed ΔG_CO of these candidates. The prediction values of Sc-C2O2, Sc-CN2O, and Y-CN2O are −0.184 eV, −0.185 eV, and −0.145 eV, respectively, and the DFT computed ones are −0.359 eV, −0.380 eV, and −0.345 eV, respectively. The results demonstrate the catalytic activity of Sc-C2O2, Sc-CN2O, and Y-CN2O.
为了验证我们预测的电位电催化剂，我们随机选择了三种设计的材料 Sc-C2O2 、 Sc-CN2O 和 Y-CN2O 进行进一步的 DFT 计算，并通过 AIMD 模拟检查了这些候选材料的稳定性。AIMD 结果如图 S10 所示，在 600 K 下模拟 10 ps 后，结构完好无损，表明这些材料是稳定的。此外，我们还比较了这些候选物的预测 ΔG_CO 和 DFT 计算的 ΔG_CO。Sc-C2O2、Sc-CN2O 和 Y-CN2O 的预测值分别为 −0.184 eV、−0.185 eV 和 −0.145 eV，DFT 计算值分别为 −0.359 eV、−0.380 eV 和 −0.345 eV。结果表明 Sc-C2O2 、 Sc-CN2O 和 Y-CN2O 的催化活性。

In fact, after collecting the feature data from NIST, we could get prediction results within less than a minute, which greatly reduces the research cost compared with the general high-throughput methods, and the prediction results can indubitably be used as a reference for further experiments of novel CO₂ electroreduction catalysts.
事实上，从 NIST 收集特征数据后，我们可以在不到一分钟的时间内获得预测结果，与一般的高通量方法相比，这大大降低了研究成本，预测结果无疑可以作为新型 CO₂ 电还原催化剂进一步实验的参考。

Conclusions 结论

Click to copy section link
单击以复制部分链接Section link copied!

Overall, by using a DFT database and machine learning methods, we successfully established a prediction model that can fast explore potential catalysts for CO₂ electroreduction. After building the DFT database, feature engineering and model training, the final XGBR model shows the best prediction accuracy. By using this model, 1060 designed SACs including combinations of 20 transition metal atoms and 53 different nonmetal bonded environments were used as our prediction target materials. Moreover, the catalytic activity of HER of 1060 designed materials was also predicted by machine learning. By combining the prediction results of CO₂ electroreduction and HER, we rapidly screened suitable materials by overlapping the results, and then 94 potential SACs for CO₂ eletroreduction were proposed, which greatly improves the efficiency compared with high-throughput methods. The framework of establishment of prediction models is an excellent tool for preliminary investigations. Before further more accurate investigations, researchers can quickly find potential materials from the predicted heat maps. Considering the current huge demand for energy resources, the effective screening method with machine learning algorithms in this work can promote energy-related developments.
总体而言，通过使用 DFT 数据库和机器学习方法，我们成功建立了一个预测模型，可以快速探索 CO₂ 电还原的潜在催化剂。在构建 DFT 数据库、特征工程和模型训练后，最终的 XGBR 模型显示出最佳的预测精度。通过使用该模型，使用 1060 个设计的 SAC，包括 20 个过渡金属原子和 53 种不同的非金属键合环境的组合，作为我们的预测目标材料。此外，机器学习还预测了 1060 种设计材料的 HER 的催化活性。结合 CO₂ 电还原和 HER 的预测结果，通过重叠结果快速筛选出合适的材料，然后提出了 94 种潜在的 CO₂ 电还原 SAC，与高通量方法相比，效率大大提高。建立预测模型的框架是初步调查的极好工具。在进一步进行更准确的调查之前，研究人员可以从预测的热图中快速找到潜在的材料。考虑到当前对能源资源的巨大需求，本研究中采用机器学习算法的有效筛选方法可以促进能源相关发展。

Supporting Information 支持信息

Click to copy section link
单击以复制部分链接Section link copied!

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcc.0c05964.
支持信息可在 https://pubs.acs.org/doi/10.1021/acs.jpcc.0c05964 免费获取。

Figures of SAC structures of the training data set; CO/H adsorption energy prediction results of the comparison between the XGBR model and DFT computed data with different data split ratios and different feature sets; the Pearson correlation coefficient heat maps and feature importance ranks of different feature sets for CO/H adsorption energy prediction model; H adsorption prediction models trained by different algorithms; AIMD results of some predicted SACs; parameters of the final machine learning models; cross validation results of models; and the order number in the heat map of different kinds of structures (PDF)
训练数据集的 SAC 结构图;XGBR 模型与不同数据分割比、不同特征集的 DFT 计算数据对比的 CO/H 吸附能量预测结果;CO/H 吸附能预测模型的 Pearson 相关系数热图和不同特征集的特征重要性排名;不同算法训练的 H 吸附预测模型;一些预测的 SAC 的 AIMD 结果;最终机器学习模型的参数;模型的交叉验证结果;以及不同类型结构的热图（PDF）中的订单号

jp0c05964_si_001.pdf (1.04 MB)
jp0c05964_si_001.pdf （1.04 MB）

Terms & Conditions 条款和条件

Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.
大多数电子支持信息文件无需订阅 ACS Web 版本即可获得。此类文件可以按文章下载用于研究用途（如果有链接到相关文章的公共使用许可证，则该许可证可能允许其他用途）。可以通过 RightsLink 权限系统请求 ACS 以用于其他用途：http://pubs.acs.org/page/copyright/permissions.html。

Author Information 作者信息

Click to copy section link
单击以复制部分链接Section link copied!

Corresponding Authors 通讯作者
- Xu Zhang - School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China; Email: zhangxu@nankai.edu.cn
  张旭 - 南开大学材料科学与工程学院，新能源材料化学研究所，可再生能源转换与储存中心（ReCast），先进能源材料化学教育部重点实验室，天津，300350; 电子邮件： zhangxu@nankai.edu.cn
- Zhen Zhou - School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China; http://orcid.org/0000-0003-3232-9903; Email: zhouzhen@nankai.edu.cn
  周震 - 南开大学材料科学与工程学院，新能源材料化学研究所，可再生能源转换与储存中心（ReCast），先进能源材料化学教育部重点实验室，天津，300350; http://orcid.org/0000-0003-3232-9903; 电子邮件：zhouzhen@nankai.edu.cn
Authors 作者
- An Chen - School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China
  安晨 - 南开大学材料科学与工程学院，新能源材料化学研究所，可再生能源转换与储存中心（ReCast），先进能源材料化学教育部重点实验室，天津，300350
- Letian Chen - School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China
  陈乐天 - 南开大学材料科学与工程学院，新能源材料化学研究所，可再生能源转换与储存中心（ReCast），先进能源材料化学教育部重点实验室，天津，300350
- Sai Yao - School of Materials Science and Engineering, Institute of New Energy Material Chemistry, Renewable Energy Conversion and Storage Center (ReCast), Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, 300350, P. R. China
  尧赛 - 南开大学材料科学与工程学院，新能源材料化学研究所，可再生能源转化与储存中心（ReCast），先进能源材料化学教育部重点实验室，天津 300350
Notes 笔记
The authors declare no competing financial interest.
作者声明没有竞争性的经济利益。

Acknowledgments 确认

Click to copy section link
单击以复制部分链接Section link copied!

This work was supported by NSFC (21933006 and 91845112), China Postdoctoral Science Foundation (2019M660055), and the Fundamental Research Funds for the Central Universities, Nankai University (63201063).
这项工作得到了国家自然科学基金（21933006 和 91845112）、中国博士后科学基金（2019M660055）和中央高校基本科研业务费南开大学（63201063）的支持。

References 引用

Click to copy section link
单击以复制部分链接Section link copied!

This article references 49 other publications.
本文引用了其他 49 种出版物。

1
Ma, X.; Li, Z.; Achenie, L. E.; Xin, H. Machine-Learning-Augmented Chemisorption Model for CO₂ Electroreduction Catalyst Screening. J. Phys. Chem. Lett. 2015, 6, 3528– 33, DOI: 10.1021/acs.jpclett.5b01660
1 马，X.;李 Z.;Achenie， L. E.;Xin， H.用于 CO2 电还原催化剂筛选的机器学习增强化学吸附模型。J. Phys. Chem. Lett.2015， 6， 3528– 33， DOI. 10.1021/acs.jpclett.5b01660
Google Scholar 谷歌学术
1
Machine-Learning-Augmented Chemisorption Model for CO2 Electroreduction Catalyst Screening
Ma, Xianfeng; Li, Zheng; Achenie, Luke E. K.; Xin, Hongliang
Journal of Physical Chemistry Letters (2015), 6 (18), 3528-3533CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
We present a machine-learning-augmented chemisorption model that enables fast and accurate prediction of the surface reactivity of metal alloys within a broad chem. space. Specifically, we show that artificial neural networks, a family of biol. inspired learning algorithms, trained with a set of ab initio adsorption energies and electronic fingerprints of idealized bimetallic surfaces, can capture complex, nonlinear interactions of adsorbates (e.g., *CO) on multimetallics with ∼0.1 eV error, outperforming the two-level interaction model in prediction. By leveraging scaling relations between adsorption energies of similar adsorbates, we illustrate that this integrated approach greatly facilitates high-throughput catalyst screening and, as a specific case, suggests promising {100}-terminated multimetallic alloys with improved efficiency and selectivity for CO2 electrochem. redn. to C2 species. Statistical anal. of the network response to perturbations of input features underpins our fundamental understanding of chem. bonding on metal surfaces.
2
Zhao, Z.; Lu, G. Computational Screening of Near-Surface Alloys for CO₂ Electroreduction. ACS Catal. 2018, 8, 3885– 3894, DOI: 10.1021/acscatal.7b03705
阿拉伯数字赵 Z.;Lu， G.用于 CO2 电还原的近表面合金的计算筛选。ACS Catal.2018， 8， 3885– 3894， DOI： 10.1021/acscatal.7b03705
Google Scholar 谷歌学术
2
Computational Screening of Near-Surface Alloys for CO2 Electroreduction
Zhao, Zhonglong; Lu, Gang
ACS Catalysis (2018), 8 (5), 3885-3894CODEN: ACCACS; ISSN:2155-5435. (American Chemical Society)
Electrochem. conversion of carbon dioxide (CO2) into chem. feedstocks provides an attractive soln. to our pressing energy and environment problems. Here, we report that transition metal near-surface alloys (NSAs) are promising catalysts for CO2 electroredn. Based on first-principles calcns. on 190 candidates, we propose a no. of NSAs which show promise of highly active and selective catalysts for formic acid, carbon monoxide, methanol, and ethylene prodn., while simultaneously suppress competing hydrogen evolution reaction (HER). We predict that Pd/W, Au/Hf, and Au/Zr NSAs are more active than most known electrodes for formic acid formation with overpotentials significantly lower than that of HER. Ag/Hf and Ag/Zr are revealed as superior catalysts for the prodn. of carbon monoxide with overpotentials of 0.77 V lower than that on pure Ag electrode. We find that methanol and ethylene can be produced on Ag/Ta and Ag/Nb NSAs whose overpotentials are ∼15% lower than that on Cu (211) surface. On the other hand, their overpotentials for HER are six times more neg. than that on Cu (211). The work demonstrates the great potential of transition metal catalysts by modulating their near surface properties.
3
Gálvez-Vázquez, M. d. J.; Moreno-García, P.; Guo, H.; Hou, Y.; Dutta, A.; Waldvogel, S. R.; Broekmann, P. Leaded Bronze Alloy as a Catalyst for the Electroreduction of CO₂. ChemElectroChem 2019, 6, 2324– 2330, DOI: 10.1002/celc.201900537
3Gálvez-Vázquez，医学博士;莫雷诺-加西亚，P.;郭 H.;侯 Y.;杜塔，A.;Waldvogel， S. R.;Broekmann， P.铅青铜合金作为 CO2 电还原的催化剂。化学电子化学 2019， 6， 2324– 2330， DOI： 10.1002/celc.201900537
Google Scholar 谷歌学术
There is no corresponding record for this reference.
4
Zhang, Q.; Xu, W.; Xu, J.; Liu, Y.; Zhang, J. High performing and cost-effective metal/metal oxide/metal alloy catalysts/electrodes for low temperature CO₂ electroreduction. Catal. Today 2018, 318, 15– 22, DOI: 10.1016/j.cattod.2018.03.029
4 张 Q.;徐，W.;徐 J.;刘 Y.;Zhang， J.用于低温 CO2 电还原的高性能且经济高效的金属/金属氧化物/金属合金催化剂/电极。加泰罗尼亚。今日 2018， 318， 15– 22， DOI： 10.1016/j.cattod.2018.03.029
Google Scholar 谷歌学术
4
High performing and cost-effective metal/metal oxide/metal alloy catalysts/electrodes for low temperature CO2 electroreduction
Zhang, Qi; Xu, Wutao; Xu, Jie; Liu, Yuyu; Zhang, Jiujun
Catalysis Today (2018), 318 (), 15-22CODEN: CATTEA; ISSN:0920-5861. (Elsevier B.V.)
The electrochem. redn. of CO2 (ERC) has been proved to be both tech. and economic feasible. ERC using renewable energies and abandon nuclear/hydroelec. energies can convert CO2 to low-carbon fuels for energy storage and reducing CO2 emission as well as promoting waste (gas) recycling/utilization. One of the challenges is to find high electrochem. active, stable, product-selective and cost-effective catalysts for speeding up the electrode kinetics of ERC. Based on these specific requirements for catalysts, quite few of them are practically promising. In this mini-review paper, we try to give an overview of some important researches and the recent progress in ERC catalysts, focusing on the electrochem. characteristics of Cu-, Sn- and Zn-based catalysts for the efficient ERC. The catalysts reviewed in this paper include single pure metal, metal oxide, metal alloys and metal complexes. In recent years, more and more catalysts with novel nanostructures have been reported, such as those with shell-core structure, which exhibit desirable electrocatalytic performance. Moreover, carbon materials, such as carbon nanotube (CNT) and reduced graphene oxide (rGO), have also been explored as desirable catalyst supports in lab-scale studies. In the most recent years, more attention is moved on the fine crystal structures of catalysts, which are found to be quite crit. for achieving desirable product selectivity.
5
Fu, H. Q.; Zhang, L.; Zheng, L. R.; Liu, P. F.; Zhao, H.; Yang, H. G. Enhanced CO₂ electroreduction performance over Cl-modified metal catalysts. J. Mater. Chem. A 2019, 7, 12420– 12425, DOI: 10.1039/C9TA02223F
Google Scholar
5
Enhanced CO2 electroreduction performance over Cl-modified metal catalysts
Fu, Huai Qin; Zhang, Le; Zheng, Li Rong; Liu, Peng Fei; Zhao, Huijun; Yang, Hua Gui
Journal of Materials Chemistry A: Materials for Energy and Sustainability (2019), 7 (20), 12420-12425CODEN: JMCAET; ISSN:2050-7496. (Royal Society of Chemistry)
Ag nanoparticles with surface Cl- modification (Ag-Cl NPs) by in situ electroredn. of AgCl exhibit a high selectivity for CO2-to-CO conversion. The obsd. C-Cl bond suggests that electrons can be effectively transferred from the Cl- ions to the unoccupied orbital of CO2, and then activate nonpolar CO2 mols. on Cl- sites.
6
Plana, D.; Flórez-Montaño, J.; Celorrio, V.; Pastor, E.; Fermín, D. J. Tuning CO2 electroreduction efficiency at Pd shells on Au nanocores. Chem. Commun. 2013, 49, 10962– 10964, DOI: 10.1039/c3cc46543h
Google Scholar
6
Tuning CO2 electroreduction efficiency at Pd shells on Au nanocores
Plana, Daniela; Florez-Montano, Jonathan; Celorrio, Veronica; Pastor, Elena; Fermin, David J.
Chemical Communications (Cambridge, United Kingdom) (2013), 49 (93), 10962-10964CODEN: CHCOFS; ISSN:1359-7345. (Royal Society of Chemistry)
The faradaic efficiency of CO2 electroredn. is significantly affected by the thickness of Pd nanoshells on Au cores. The ratio of hydrogen evolution to CO2 redn. was detd. by differential electrochem. mass spectrometry. Decreasing the Pd shell thickness from 10 to 1 nm leads to a twofold increase in faradaic efficiency.
7
Cao, L.; Raciti, D.; Li, C.; Livi, K. J. T.; Rottmann, P. F.; Hemker, K. J.; Mueller, T.; Wang, C. Mechanistic Insights for Low-Overpotential Electroreduction of CO₂ to CO on Copper Nanowires. ACS Catal. 2017, 7, 8578– 8587, DOI: 10.1021/acscatal.7b03107
Google Scholar
7
Mechanistic Insights for Low-Overpotential Electroreduction of CO2 to CO on Copper Nanowires
Cao, Liang; Raciti, David; Li, Chenyang; Livi, Kenneth J. T.; Rottmann, Paul F.; Hemker, Kevin J.; Mueller, Tim; Wang, Chao
ACS Catalysis (2017), 7 (12), 8578-8587CODEN: ACCACS; ISSN:2155-5435. (American Chemical Society)
Recent developments of Cu-based nanomaterials have enabled the electroredn. of CO2 at low overpotentials. The mechanism of low-overpotential CO2 redn. on these nanocatalysts, however, largely remains elusive. The authors report here a systematic study of CO2 redn. on highly dense Cu nanowires, with the focus placed on understanding the surface structure effects on the formation of *CO (* denotes an adsorption site on the catalyst surface) and the evolution of gas-phase CO product (CO(g)) at low overpotentials (more pos. than -0.5 V). Cu nanowires of distinct nanocryst. and surface structures were studied comparatively to build up the structure-property relations, which are further interpreted by performing d. functional theory (DFT) calcns. of the reaction pathway on the various facets of Cu. A kinetic model reveals competition between CO(g) evolution and *CO poisoning depending on the electrode potential and surface structures. Open and metastable facets such as (110) and reconstructed (110) are likely the active sites for the electroredn. of CO2 to CO at the low overpotentials.
8
Baturina, O. A.; Lu, Q.; Padilla, M. A.; Xin, L.; Li, W.; Serov, A.; Artyushkova, K.; Atanassov, P.; Xu, F.; Epshteyn, A. CO₂ Electroreduction to Hydrocarbons on Carbon-Supported Cu Nanoparticles. ACS Catal. 2014, 4, 3682– 3695, DOI: 10.1021/cs500537y
Google Scholar
8
CO2 Electroreduction to Hydrocarbons on Carbon-Supported Cu Nanoparticles
Baturina, Olga A.; Lu, Qin; Padilla, Monica A.; Xin, Le; Li, Wenzhen; Serov, Alexey; Artyushkova, Kateryna; Atanassov, Plamen; Xu, Feng; Epshteyn, Albert; Brintlinger, Todd; Schuette, Mike; Collins, Greg E.
ACS Catalysis (2014), 4 (10), 3682-3695CODEN: ACCACS; ISSN:2155-5435. (American Chemical Society)
Activities of Cu nanoparticles supported on C black (VC), single-wall C nanotubes (SWNTs), and Ketjen Black (KB) toward CO2 electroredn. to hydrocarbons (CH4, C2H2, C2H4, and C2H6) are evaluated using a sealed rotating disk electrode (RDE) setup coupled to a gas chromatograph (GC). Thin films of supported Cu catalysts are deposited on RDE tips following a procedure well-established in the fuel cell community. Lead (Pb) underpotential deposition (UPD) was used to det. the electrochem. surface area (ECSA) of thin films of 40% Cu/VC, 20% Cu/SWNT, 50% Cu/KB, and com. 20% Cu/VC catalysts on glassy C electrodes. Faradaic efficiencies of four C-supported Cu catalysts toward CO2 electroredn. to hydrocarbons are compared to that of electrodeposited smooth Cu films. For all the catalysts studied, the only hydrocarbons detected by GC are CH4 and C2H4. The Cu nanoparticles are more active toward C2H4 generation vs. electrodeposited smooth Cu films. For the supported Cu nanocatalysts, the ratio of C2H4/CH4 faradaic efficiencies is believed to be a function of particle size, as higher ratios are obsd. for smaller Cu nanoparticles. This is likely due to an increase in the fraction of under-coordinated sites, such as corners, edges, and defects, as the nanoparticles become smaller.
9
Song, Y.; Chen, W.; Zhao, C.; Li, S.; Wei, W.; Sun, Y. Metal-Free Nitrogen-Doped Mesoporous Carbon for Electroreduction of CO₂ to Ethanol. Angew. Chem., Int. Ed. 2017, 56, 10840– 10844, DOI: 10.1002/anie.201706777
Google Scholar
9
Metal-Free Nitrogen-Doped Mesoporous Carbon for Electroreduction of CO2 to Ethanol
Song, Yanfang; Chen, Wei; Zhao, Chengcheng; Li, Shenggang; Wei, Wei; Sun, Yuhan
Angewandte Chemie, International Edition (2017), 56 (36), 10840-10844CODEN: ACIEF5; ISSN:1433-7851. (Wiley-VCH Verlag GmbH & Co. KGaA)
CO2 electroredn. is a promising technique for satisfying both renewable energy storage and a neg. C cycle. However, it remains a challenge to convert CO2 into C2 products with high efficiency and selectivity. Herein, the authors report a N-doped ordered cylindrical mesoporous C as a robust metal-free catalyst for CO2 electroredn., enabling the efficient prodn. of EtOH with nearly 100% selectivity and high faradaic efficiency of 77% at -0.56 V vs. the reversible H electrode. Expts. and d. functional theory calcns. demonstrate that the synergetic effect of the N hetero-atoms and the cylindrical channel configurations facilitate the dimerization of key CO* intermediates and the subsequent p-electron transfers, resulting in superior electrocatalytic performance for synthesizing EtOH from CO2.
10
Zhang, T.; Lin, L.; Li, Z.; He, X.; Xiao, S.; Shanov, V. N.; Wu, J. Nickel-Nitrogen-Carbon Molecular Catalysts for High Rate CO₂ Electro-reduction to CO: On the Role of Carbon Substrate and Reaction Chemistry. ACS Appl. Energy Mater. 2020, 3, 1617– 1626, DOI: 10.1021/acsaem.9b02112
Google Scholar
10
Nickel-Nitrogen-Carbon Molecular Catalysts for High Rate CO2 Electro-reduction to CO: On the Role of Carbon Substrate and Reaction Chemistry
Zhang, Tianyu; Lin, Lili; Li, Zhengyuan; He, Xingyu; Xiao, Shengdong; Shanov, Vesselin N.; Wu, Jingjie
ACS Applied Energy Materials (2020), 3 (2), 1617-1626CODEN: AAEMCQ; ISSN:2574-0962. (American Chemical Society)
Metal-N-C (M-N-C) mol. catalysts with NiN4 active structure were extensively studied as selective and active catalysts toward electrochem. redn. of CO2 to CO. The key challenge for a practical M-N-C catalyst is to increase the d. of at. metal active sites that achieves the partial c.d. of CO (jCO) relevant to the industrial level at lower overpotentials. Here, the authors revealed the effect of phys. and chem. properties of C substrates and synthetic processes on the tuning of the d. of at. metal active sites as well as the role of reaction chem. in enhancing the jCO and reducing the overpotential. The achievable loading of NiN4 active site in the Ni-N-C is detd. by the combined content of pyridinic and pyrrolic N functionalities and Ni-N coordination efficiency derived from the pyrolytic step rather than the uptake capability of Ni2+ in the adsorption step in the case of C black with high sp. surface area (>1000 m2/g). The N dopant content can be improved by modifying O functional groups on the surface of C black, optimizing the pyrolytic temp., and iterating the doping step. Through a combination of all optimum factors, the resultant Ni-N-C catalyst has a max. loading of ∼4.4% for at. Ni. This Ni-N-C catalyst exhibited faradaic efficiency (FE) of CO of 97% and jCO of -152 mA cm-2 at -0.93 V vs. RHE in a flow cell using 0.5M KHCO3 electrolyte while showing 93% FE of CO and jCO of -67 mA cm-2 at -0.61 V vs. RHE at 1 M KOH. Adding KI to the base electrolyte significantly magnified the jCO to larger than -200 mA cm-2 at a potential of -0.51 V vs. RHE while maintaining the almost unity FE of CO. The Ni-N-C is compatible with the membrane-electrode-assembly-based electrolyzer in which the jCO also achieved >200 mA cm-2 at a cell voltage of ∼2.7 V.
11
Li, Z.; Ji, S.; Liu, Y.; Cao, X.; Tian, S.; Chen, Y.; Niu, Z.; Li, Y. Well-Defined Materials for Heterogeneous Catalysis: From Nanoparticles to Isolated Single-Atom Sites. Chem. Rev. 2020, 120, 623– 682, DOI: 10.1021/acs.chemrev.9b00311
Google Scholar
11
Well-Defined Materials for Heterogeneous Catalysis: From Nanoparticles to Isolated Single-Atom Sites
Li, Zhi; Ji, Shufang; Liu, Yiwei; Cao, Xing; Tian, Shubo; Chen, Yuanjun; Niu, Zhiqiang; Li, Yadong
Chemical Reviews (Washington, DC, United States) (2020), 120 (2), 623-682CODEN: CHREAY; ISSN:0009-2665. (American Chemical Society)
A review. The use of well-defined materials in heterogeneous catalysis will open up numerous new opportunities for the development of advanced catalysts to address the global challenges in energy and the environment. This review surveys the roles of nanoparticles and isolated single atom sites in catalytic reactions. In the second section, the effects of size, shape, and metal-support interactions are discussed for nanostructured catalysts. Case studies are summarized to illustrate the dynamics of structure evolution of well-defined nanoparticles under certain reaction conditions. In the third section, we review the syntheses and catalytic applications of isolated single at. sites anchored on different types of supports. In the final part, we conclude by highlighting the challenges and opportunities of well-defined materials for catalyst development and gaining a fundamental understanding of their active sites.
12
Pan, Y.; Lin, R.; Chen, Y.; Liu, S.; Zhu, W.; Cao, X.; Chen, W.; Wu, K.; Cheong, W. C.; Wang, Y. Design of Single-Atom Co-N₅ Catalytic Site: A Robust Electrocatalyst for CO₂ Reduction with Nearly 100% CO Selectivity and Remarkable Stability. J. Am. Chem. Soc. 2018, 140, 4218– 4221, DOI: 10.1021/jacs.8b00814
Google Scholar
12
Design of Single-Atom Co-N5 Catalytic Site: A Robust Electrocatalyst for CO2 Reduction with Nearly 100% CO Selectivity and Remarkable Stability
Pan, Yuan; Lin, Rui; Chen, Yinjuan; Liu, Shoujie; Zhu, Wei; Cao, Xing; Chen, Wenxing; Wu, Konglin; Cheong, Weng-Chon; Wang, Yu; Zheng, Lirong; Luo, Jun; Lin, Yan; Liu, Yunqi; Liu, Chenguang; Li, Jun; Lu, Qi; Chen, Xin; Wang, Dingsheng; Peng, Qing; Chen, Chen; Li, Yadong
Journal of the American Chemical Society (2018), 140 (12), 4218-4221CODEN: JACSAT; ISSN:0002-7863. (American Chemical Society)
We develop an N-coordination strategy to design a robust CO2 redn. reaction (CO2RR) electrocatalyst with atomically dispersed Co-N5 site anchored on polymer-derived hollow N-doped porous carbon spheres. Our catalyst exhibits high selectivity for CO2RR with CO Faradaic efficiency (FECO) above 90% over a wide potential range from -0.57 to -0.88 V (the FECO exceeded 99% at -0.73 and -0.79 V). The CO c.d. and FECO remained nearly unchanged after electrolyzing 10 h, revealing remarkable stability. Expts. and d. functional theory calcns. demonstrate single-atom Co-N5 site is the dominating active center simultaneously for CO2 activation, the rapid formation of key intermediate COOH* as well as the desorption of CO.
13
Liu, X.; Wang, Z.; Tian, Y.; Zhao, J. Graphdiyne-Supported Single Iron Atom: A Promising Electrocatalyst for Carbon Dioxide Electroreduction into Methane and Ethanol. J. Phys. Chem. C 2020, 124, 3722– 3730, DOI: 10.1021/acs.jpcc.9b11649
Google Scholar
13
Graphdiyne-Supported Single Iron Atom: A Promising Electrocatalyst for Carbon Dioxide Electroreduction into Methane and Ethanol
Liu, Xin; Wang, Zhongxu; Tian, Yu; Zhao, Jingxiang
Journal of Physical Chemistry C (2020), 124 (6), 3722-3730CODEN: JPCCCK; ISSN:1932-7447. (American Chemical Society)
Electrochem. redn. of CO2 (CO2ER) to high-energy-d. multicarbon products is a quite promising technique for large-scale renewable energy storage, for which searching for stable, inexpensive, and efficient catalysts is a key scientific issue. The potential of an exptl. available single Fe atom supported on graphdiyne (Fe/GDY) as the CO2ER catalyst was explored by d. functional theory (DFT) computations. The authors' results revealed that Fe/GDY exhibits high stability due to the strong hybridization between the Fe 3d orbitals and the C 2p orbitals of GDY. Due to the small limiting potential of -0.43 V, the anchored Fe atom can effectively reduce CO2 to CH4 along the following pathway: CO2 → HCOO* → HCOOH* → HCO* → H2CO* → H3CO* → O* + CH4 → OH* → H2O, in which the hydrogenation of HCOOH* to HCO* is the potential-detg. step. Also, the unsatd. HCO* species on Fe/GDY can provide an active site for further coupling with CO to generate EtOH with a small activation energy for C-C coupling. The authors' theor. results not only propose a new approach to CO2ER to C2 products on a single-site catalyst but also further widen the potential applications of GDY.
14
Chen, A.; Zhang, X.; Zhou, Z. Machine learning: Accelerating materials development for energy storage and conversion. InfoMat 2020, 2, 553– 576, DOI: 10.1002/inf2.12094
Google Scholar
14
Machine learning: Accelerating materials development for energy storage and conversion
Chen, An; Zhang, Xu; Zhou, Zhen
InfoMat (2020), 2 (3), 553-576CODEN: INFOHH; ISSN:2567-3165. (John Wiley & Sons Australia, Ltd.)
With the development of modern society, the requirement for energy has become increasingly important on a global scale. Therefore, the exploration of novel materials for renewable energy technologies is urgently needed. Traditional methods are difficult to meet the requirements for materials science due to long exptl. period and high cost. Nowadays, machine learning (ML) is rising as a new research paradigm to revolutionize materials discovery. In this review, we briefly introduce the basic procedure of ML and common algorithms in materials science, and particularly focus on latest progress in applying ML to property prediction and materials development for energy-related fields, including catalysis, batteries, solar cells, and gas capture. Moreover, contributions of ML to expts. are involved as well. We highly expect that this review could lead the way forward in the future development of ML in materials science.
15
Frey, N. C.; Wang, J.; Vega Bellido, G. I.; Anasori, B.; Gogotsi, Y.; Shenoy, V. B. Prediction of Synthesis of 2D Metal Carbides and Nitrides (MXenes) and Their Precursors with Positive and Unlabeled Machine Learning. ACS Nano 2019, 13, 3031– 3041, DOI: 10.1021/acsnano.8b08014
Google Scholar
15
Prediction of Synthesis of 2D Metal Carbides and Nitrides (MXenes) and Their Precursors with Positive and Unlabeled Machine Learning
Frey, Nathan C.; Wang, Jin; Vega Bellido, Gabriel Ivan; Anasori, Babak; Gogotsi, Yury; Shenoy, Vivek B.
ACS Nano (2019), 13 (3), 3031-3041CODEN: ANCAC3; ISSN:1936-0851. (American Chemical Society)
Growing interest in the potential applications of two-dimensional (2D) materials has fueled advancement in the identification of 2D systems with exotic properties. Increasingly, the bottleneck in this field is the synthesis of these materials. Although theor. calcns. have predicted a myriad of promising 2D materials, only a few dozen were exptl. realized since the initial discovery of graphene. Here, we adapt the state-of-the-art pos. and unlabeled (PU) machine learning framework to predict which theor. proposed 2D materials have the highest likelihood of being successfully synthesized. Using elemental information and data from high-throughput d. functional theory calcns., we apply the PU learning method to the MXene family of 2D transition metal carbides, carbonitrides, and nitrides, and their layered precursor MAX phases, and identify 18 MXene compds. that are highly promising candidates for synthesis. By considering both the MXenes and their precursors, we further propose 20 synthesizable MAX phases that can be chem. exfoliated to produce MXenes.
16
Lu, S.; Zhou, Q.; Ouyang, Y.; Guo, Y.; Li, Q.; Wang, J. Accelerated discovery of stable lead-free hybrid organic-inorganic perovskites via machine learning. Nat. Commun. 2018, 9, 3405, DOI: 10.1038/s41467-018-05761-w
Google Scholar
16
Accelerated discovery of stable lead-free hybrid organic-inorganic perovskites via machine learning
Lu Shuaihua; Zhou Qionghua; Ouyang Yixin; Guo Yilv; Li Qiang; Wang Jinlan
Nature communications (2018), 9 (1), 3405 ISSN:.
Rapidly discovering functional materials remains an open challenge because the traditional trial-and-error methods are usually inefficient especially when thousands of candidates are treated. Here, we develop a target-driven method to predict undiscovered hybrid organic-inorganic perovskites (HOIPs) for photovoltaics. This strategy, combining machine learning techniques and density functional theory calculations, aims to quickly screen the HOIPs based on bandgap and solve the problems of toxicity and poor environmental stability in HOIPs. Successfully, six orthorhombic lead-free HOIPs with proper bandgap for solar cells and room temperature thermal stability are screened out from 5158 unexplored HOIPs and two of them stand out with direct bandgaps in the visible region and excellent environmental stability. Essentially, a close structure-property relationship mapping the HOIPs bandgap is established. Our method can achieve high accuracy in a flash and be applicable to a broad class of functional material design.
17
Im, J.; Lee, S.; Ko, T.-W.; Kim, H. W.; Hyon, Y.; Chang, H. Identifying Pb-free perovskites for solar cells by machine learning. npj Comput. Mater. 2019, 5, 37, DOI: 10.1038/s41524-019-0177-0
Google Scholar
There is no corresponding record for this reference.
18
Ali, A.; Park, H.; Mall, R.; Aïssa, B.; Sanvito, S.; Bensmail, H.; Belaidi, A.; El-Mellouhi, F. Machine Learning Accelerated Recovery of the Cubic Structure in Mixed-Cation Perovskite Thin Films. Chem. Mater. 2020, 32, 2998– 3006, DOI: 10.1021/acs.chemmater.9b05342
Google Scholar
18
Machine Learning Accelerated Recovery of the Cubic Structure in Mixed-Cation Perovskite Thin Films
Ali, Adnan; Park, Heesoo; Mall, Raghvendra; Aissa, Brahim; Sanvito, Stefano; Bensmail, Halima; Belaidi, Abdelhak; El-Mellouhi, Fedwa
Chemistry of Materials (2020), 32 (7), 2998-3006CODEN: CMATEX; ISSN:0897-4756. (American Chemical Society)
Data-driven approaches for materials design and selection have accelerated materials discovery along with the upsurge of machine learning applications. We report here a prediction-to-lab.-scale synthesis of cubic phase triple-cation lead halide perovskites guided by a machine learning perovskite stability predictor. The starting double-cation perovskite resulting from the incorporation of 15% dimethylammonium (DMA) in methylammonium lead triiodide suffers from significant deviation from the perovskite structure. By analyzing the X-ray diffraction and SEM, we confirmed that it is possible to recover the perovskite structure with the cubic phase at room temp. (RT) while minimizing the iterations of trial-and-error by adding <10 mol % of cesium cation additives, as guided by the machine learning predictor. Our conclusions highly support the cubic-phase stabilization at RT by controlling the stoichiometric ratio of various sized cations. This prediction-to-lab.-scale synthesis approach also enables us to identify room for improvements of the current machine learning predictor to take into consideration the cubic phase stability as well as phase segregation.
19
Fernandez, M.; Boyd, P. G.; Daff, T. D.; Aghaji, M. Z.; Woo, T. K. Rapid and Accurate Machine Learning Recognition of High Performing Metal Organic Frameworks for CO₂ Capture. J. Phys. Chem. Lett. 2014, 5, 3056– 3060, DOI: 10.1021/jz501331m
Google Scholar
19
Rapid and Accurate Machine Learning Recognition of High Performing Metal Organic Frameworks for CO2 Capture
Fernandez, Michael; Boyd, Peter G.; Daff, Thomas D.; Aghaji, Mohammad Zein; Woo, Tom K.
Journal of Physical Chemistry Letters (2014), 5 (17), 3056-3060CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
Quant. structure-property relationship (QSPR) models were developed using advanced machine learning algorithms that can rapidly and accurately recognize high-performing metal org. framework (MOF) materials for CO2 capture. QSPR classifiers were developed that can, in a fraction of a section, identify candidate MOFs with enhanced CO2 adsorption capacity ( > 1 mmol/g at 0.15 bar and >4 mmol/g at 1 bar). The models were tested on a large set of 292 050 MOFs that were not part of the training set. The QSPR classifier could recover 945 of the top 1000 MOFs in the test set while flagging only 10% of the whole library for compute intensive screening. Thus, using the machine learning classifiers as part of a high-throughput screening protocol would result in an order of magnitude redn. in compute time and allow intractably large structure libraries and search spaces to be screened.
20
He, Y.; Cubuk, E. D.; Allendorf, M. D.; Reed, E. J. Metallic Metal-Organic Frameworks Predicted by the Combination of Machine Learning Methods and Ab Initio Calculations. J. Phys. Chem. Lett. 2018, 9, 4562– 4569, DOI: 10.1021/acs.jpclett.8b01707
Google Scholar
20
Metallic Metal-Organic Frameworks Predicted by the Combination of Machine Learning Methods and Ab Initio Calculations
He, Yuping; Cubuk, Ekin D.; Allendorf, Mark D.; Reed, Evan J.
Journal of Physical Chemistry Letters (2018), 9 (16), 4562-4569CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
Emerging applications of metal-org. frameworks (MOFs) in electronic devices will benefit from the design and synthesis of intrinsically, highly electronically conductive MOFs. However, very few are known to exist. It is a challenging task to search for electronically conductive MOFs within the tens of thousands of reported MOF structures. Using a new strategy (i.e., transfer learning) of combining machine learning techniques, statistical multivoting, and ab initio calcns., we screened 2932 MOFs and identified 6 MOF crystal structures that are metallic at the level of semilocal DFT band theory: Mn2[Re6X8(CN)6]4 (X = S, Se,Te), Mn[Re3Te4(CN)3], Hg[SCN]4Co[NCS]4, and CdC4. Five of these structures have been synthesized and reported in the literature, but their elec. characterization has not been reported. Our work demonstrates the potential power of machine learning in materials science to aid in down-selecting from large nos. of potential candidates and provides the information and guidance to accelerate the discovery of novel advanced materials.
21
Wu, Y.; Duan, H.; Xi, H. Machine Learning-Driven Insights into Defects of Zirconium Metal-Organic Frameworks for Enhanced Ethane-Ethylene Separation. Chem. Mater. 2020, 32, 2986– 2997, DOI: 10.1021/acs.chemmater.9b05322
Google Scholar
21
Machine Learning-Driven Insights into Defects of Zirconium Metal-Organic Frameworks for Enhanced Ethane-Ethylene Separation
Wu, Ying; Duan, Haipeng; Xi, Hongxia
Chemistry of Materials (2020), 32 (7), 2986-2997CODEN: CMATEX; ISSN:0897-4756. (American Chemical Society)
Structural defects in metal-org. frameworks (MOFs) have the potential to yield desirable properties that could not be achieved by "defect-free" crystals, but previous works in this area have focused on limited versions of defects due to the difficulty of detecting defects in MOFs. In this work, a modeling library contg. 425 defective UiO-66 (UiO-66-Ds) with a comprehensive population (in terms of concn. and distribution) of missing-linker defects was created. Taking ethane-ethylene sepn. as a case study, we demonstrated that machine learning could provide data-driven insight into how the defects control the performance of UiO-66-Ds in adsorption, sepn., and mech. stability. We found that the missing-linker ratio in real materials could be predicted from the gravimetric surface area and pore vol., making it a useful complement for the challenges of directly measuring the defect concn. We further identified the "privileged" UiO-66-Ds that were optimal in overall properties and provided decision trees as guidance to access and design these top performers. This work offers a general strategy for fully exploring the defects in MOFs, providing long-term opportunities for the development of defect engineering in the adsorption community.
22
Zhu, X.; Yan, J.; Gu, M.; Liu, T.; Dai, Y.; Gu, Y.; Li, Y. Activity origin and design principles for oxygen reduction on dual-metal-site catalysts: A combined density functional theory and machine learning study. J. Phys. Chem. Lett. 2019, 10, 7760– 7766, DOI: 10.1021/acs.jpclett.9b03392
Google Scholar
22
Activity Origin and Design Principles for Oxygen Reduction on Dual-Metal-Site Catalysts: A Combined Density Functional Theory and Machine Learning Study
Zhu, Xiaorong; Yan, Jiaxian; Gu, Min; Liu, Tianyang; Dai, Yafei; Gu, Yanhui; Li, Yafei
Journal of Physical Chemistry Letters (2019), 10 (24), 7760-7766CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
Dual-metal-site catalysts (DMSCs) are emerging as a new frontier in the field of oxygen redn. reaction (ORR). However, there is a lack of design principles to provide a universal description of the relationship between intrinsic properties of DMSCs and the catalytic activity. Here, we identify the origin of ORR activity and unveil design principles for graphene-based DMSCs by means of d. functional theory computations and machine learning (ML). Our results indicate that several exptl. unexplored DMSCs can show outstanding ORR activity surpassing that of platinum. Remarkably, our ML study reveals that the ORR activity of DMSCs is intrinsically governed by some fundamental factors, such as electron affinity, electronegativity, and radii of the embedded metal atoms. More importantly, we propose predictor equations with acceptable accuracy to quant. describe the ORR activity of DMSCs. Our work will accelerate the search for highly active DMSCs for ORR and other electrochem. reactions.
23
Zafari, M.; Kumar, D.; Umer, M.; Kim, K. S. Machine learning-based high throughput screening for nitrogen fixation on boron-doped single atom catalysts. J. Mater. Chem. A 2020, 8, 5209– 5216, DOI: 10.1039/C9TA12608B
Google Scholar
23
Machine learning-based high throughput screening for nitrogen fixation on boron-doped single atom catalysts
Zafari, Mohammad; Kumar, Deepak; Umer, Muhammad; Kim, Kwang S.
Journal of Materials Chemistry A: Materials for Energy and Sustainability (2020), 8 (10), 5209-5216CODEN: JMCAET; ISSN:2050-7496. (Royal Society of Chemistry)
Prodn. of ammonia via electrochem. nitrogen redn. reaction (NRR) has recently attracted much attention due to its potential to play a vital role in producing fertilizers and other chems. High throughput screening of electrocatalysts for the NRR requires numerous calcns. in the search space, making the computational cost a bottleneck for predicting eligible electrocatalysts. Here we used a deep neural network (DNN) to predict efficient electrocatalysts for the NRR among boron(B)-doped graphene single atom catalysts (SACs). This model can noticeably reduce the time of computation by removing non-efficient catalysts from screening. Also, the adsorption energy and free energy can be predicted by the feature-based light gradient boosting machine (LGBM) model. These features represent the geometrical structure as well as bonding characteristics. Among the catalysts evaluated, three candidates were identified as very promising catalysts, offering excellent selectivity over the hydrogen evolution reaction (HER). CrB3C1 exhibited a minimal overpotential of 0.13 V for the NRR. This study provides a new pathway for the rational design of catalysts for nitrogen fixation by employing the most important features involved in the NRR by using machine learning methods.
24
Choukroun, D.; Daems, N.; Kenis, T.; Van Everbroeck, T.; Hereijgers, J.; Altantzis, T.; Bals, S.; Cool, P.; Breugelmans, T. Bifunctional Nickel-Nitrogen-Doped-Carbon-Supported Copper Electrocatalyst for CO₂ Reduction. J. Phys. Chem. C 2020, 124, 1369– 1381, DOI: 10.1021/acs.jpcc.9b08931
Google Scholar
24
Bifunctional Nickel-Nitrogen-Doped-Carbon-Supported Copper Electrocatalyst for CO2 Reduction
Choukroun, Daniel; Daems, Nick; Kenis, Thomas; Van Everbroeck, Tim; Hereijgers, Jonas; Altantzis, Thomas; Bals, Sara; Cool, Pegie; Breugelmans, Tom
Journal of Physical Chemistry C (2020), 124 (2), 1369-1381CODEN: JPCCCK; ISSN:1932-7447. (American Chemical Society)
Bifunctionality is a key feature of many industrial catalysts, supported metal clusters and particles in particular, and the development of such catalysts for the CO2 redn. reaction (CO2RR) to hydrocarbons and alcs. is gaining traction in light of recent advancements in the field. C-supported Cu nanoparticles are suitable candidates for integration in the state-of-the-art reaction interfaces, and here, the authors propose, synthesize, and evaluate a bifunctional Ni-N-doped-C-supported Cu electrocatalyst, in which the support possesses active sites for selective CO2 conversion to CO and Cu nanoparticles catalyze either the direct CO2 or CO redn. to hydrocarbons. The authors introduce the scientific rationale behind the concept, its applicability, and the challenges with regard to the catalyst. From the practical aspect, the deposition of Cu nanoparticles onto C black and Ni-N-C supports via an NH3-driven deposition pptn. method is reported and explored in more detail using x-ray diffraction, TGA, and H temp.-programmed redn. High-angle annular dark-field scanning TEM (HAADF-STEM) and energy-dispersive x-ray spectroscopy (EDXS) give further evidence of the presence of Cu-contg. nanoparticles on the Ni-N-C supports while revealing an addnl. relation between the nanoparticle's compn. and the electrode's electrocatalytic performance. Compared to the benchmark C black-supported Cu catalysts, Ni-N-C-supported Cu delivers up to a 2-fold increase in the partial C2H4 c.d. at -1.05 VRHE (C1/C2 = 0.67) and a concomitant 10-fold increase of the CO partial c.d. The enhanced ethylene prodn. metrics, obtained by virtue of the higher intrinsic activity of the Ni-N-C support, point out toward a synergistic action between the two catalytic functionalities.
25
Varley, J. B.; Hansen, H. A.; Ammitzbøll, N. L.; Grabow, L. C.; Peterson, A. A.; Rossmeisl, J.; Nørskov, J. K. Ni-Fe-S Cubanes in CO₂ Reduction Electrocatalysis: A DFT Study. ACS Catal. 2013, 3, 2640– 2643, DOI: 10.1021/cs4005419
Google Scholar
25
Ni-Fe-S Cubanes in CO2 Reduction Electrocatalysis: A DFT Study
Varley, J. B.; Hansen, H. A.; Ammitzboell, N. L.; Grabow, L. C.; Peterson, A. A.; Rossmeisl, J.; Noerskov, J. K.
ACS Catalysis (2013), 3 (11), 2640-2643CODEN: ACCACS; ISSN:2155-5435. (American Chemical Society)
The authors perform extensive mechanistic studies of CO2 (electro)-redn. by analogs to the active sites of CO dehydrogenase (CODH) enzymes. The authors explore structure-property relations for different cluster compns. and interpret the results with a model for CO2 electroredn. the authors recently developed and applied to transition metal catalysts. The authors' results validate the effectiveness of the CODH in catalyzing this important reaction and give insight into why specific cluster compns. were adopted by nature.
26
Back, S.; Lim, J.; Kim, N. Y.; Kim, Y. H.; Jung, Y. Single-atom catalysts for CO₂ electroreduction with significant activity and selectivity improvements. Chem. Sci. 2017, 8, 1090– 1096, DOI: 10.1039/C6SC03911A
Google Scholar
26
Single-atom catalysts for CO2 electroreduction with significant activity and selectivity improvements
Back, Seoin; Lim, Juhyung; Kim, Na-Young; Kim, Yong-Hyun; Jung, Yousung
Chemical Science (2017), 8 (2), 1090-1096CODEN: CSHCCN; ISSN:2041-6520. (Royal Society of Chemistry)
A single-atom catalyst (SAC) has an electronic structure that is very different from its bulk counterparts, and has shown an unexpectedly high specific activity with a significant redn. in noble metal usage for CO oxidn., fuel cell and hydrogen evolution applications, although phys. origins of such performance enhancements are still poorly understood. Herein, by means of d. functional theory (DFT) calcns., we for the first time investigate the great potential of single atom catalysts for CO2 electroredn. applications. In particular, we study a single transition metal atom anchored on defective graphene with single or double vacancies, denoted M@sv-Gr or M@dv-Gr, where M = Ag, Au, Co, Cu, Fe, Ir, Ni, Os, Pd, Pt, Rh or Ru, as a CO2 redn. catalyst. Many SACs are indeed shown to be highly selective for the CO2 redn. reaction over a competitive H2 evolution reaction due to favorable adsorption of carboxyl (*COOH) or formate (*OCHO) over hydrogen (*H) on the catalysts. On the basis of free energy profiles, we identified several promising candidate materials for different products; Ni@dv-Gr (limiting potential UL = -0.41 V) and Pt@dv-Gr (-0.27 V) for CH3OH prodn., and Os@dv-Gr (-0.52 V) and Ru@dv-Gr (-0.52 V) for CH4 prodn. In particular, the Pt@dv-Gr catalyst shows remarkable redn. in the limiting potential for CH3OH prodn. compared to any existing catalysts, synthesized or predicted. To understand the origin of the activity enhancement of SACs, we find that the lack of an at. ensemble for adsorbate binding and the unique electronic structure of the single atom catalysts as well as orbital interaction play an important role, contributing to binding energies of SACs that deviate considerably from the conventional scaling relation of bulk transition metals.
27
Kresse, G.; Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169– 11186, DOI: 10.1103/PhysRevB.54.11169
Google Scholar
27
Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set
Kresse, G.; Furthmueller, J.
Physical Review B: Condensed Matter (1996), 54 (16), 11169-11186CODEN: PRBMDO; ISSN:0163-1829. (American Physical Society)
The authors present an efficient scheme for calcg. the Kohn-Sham ground state of metallic systems using pseudopotentials and a plane-wave basis set. In the first part the application of Pulay's DIIS method (direct inversion in the iterative subspace) to the iterative diagonalization of large matrixes will be discussed. This approach is stable, reliable, and minimizes the no. of order Natoms3 operations. In the second part, we will discuss an efficient mixing scheme also based on Pulay's scheme. A special "metric" and a special "preconditioning" optimized for a plane-wave basis set will be introduced. Scaling of the method will be discussed in detail for non-self-consistent and self-consistent calcns. It will be shown that the no. of iterations required to obtain a specific precision is almost independent of the system size. Altogether an order Natoms2 scaling is found for systems contg. up to 1000 electrons. If we take into account that the no. of k points can be decreased linearly with the system size, the overall scaling can approach Natoms. They have implemented these algorithms within a powerful package called VASP (Vienna ab initio simulation package). The program and the techniques have been used successfully for a large no. of different systems (liq. and amorphous semiconductors, liq. simple and transition metals, metallic and semiconducting surfaces, phonons in simple metals, transition metals, and semiconductors) and turned out to be very reliable.
28
Kresse, G.; Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 1758– 1775, DOI: 10.1103/PhysRevB.59.1758
Google Scholar
28
From ultrasoft pseudopotentials to the projector augmented-wave method
Kresse, G.; Joubert, D.
Physical Review B: Condensed Matter and Materials Physics (1999), 59 (3), 1758-1775CODEN: PRBMDO; ISSN:0163-1829. (American Physical Society)
The formal relationship between ultrasoft (US) Vanderbilt-type pseudopotentials and Blochl's projector augmented wave (PAW) method is derived. The total energy functional for US pseudopotentials can be obtained by linearization of two terms in a slightly modified PAW total energy functional. The Hamilton operator, the forces, and the stress tensor are derived for this modified PAW functional. A simple way to implement the PAW method in existing plane-wave codes supporting US pseudopotentials is pointed out. In addn., crit. tests are presented to compare the accuracy and efficiency of the PAW and the US pseudopotential method with relaxed-core all-electron methods. These tests include small mols. (H2, H2O, Li2, N2, F2, BF3, SiF4) and several bulk systems (diamond, Si, V, Li, Ca, CaF2, Fe, Co, Ni). Particular attention is paid to the bulk properties and magnetic energies of Fe, Co, and Ni.
29
Hammer, B.; Hansen, L. B.; Nørskov, J. K. Improved adsorption energetics within density-functional theory using revised Perdew-Burke-Ernzerhof functionals. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 7413– 7421, DOI: 10.1103/PhysRevB.59.7413
Google Scholar
29
Improved adsorption energetics within density-functional theory using revised Perdew-Burke-Ernzerhof functionals
Hammer, B.; Hansen, L. B.; Norskov, J. K.
Physical Review B: Condensed Matter and Materials Physics (1999), 59 (11), 7413-7421CODEN: PRBMDO; ISSN:0163-1829. (American Physical Society)
A simple formulation of a generalized gradient approxn. for the exchange and correlation energy of electrons has been proposed by J. Perdew et al. (1996). Subsequently, Y. Zhang and W. Wang (1998) have shown that a slight revision of the Perdew-Burke-Ernzerhof (PBE) functional systematically improves the atomization energies for a large database of small mols. In the present work, we show that the Zhang and Yang functional (revPBE) also improves the chemisorption energetics of atoms and mols. on transition-metal surfaces. Our test systems comprise at. and mol. adsorption of oxygen, CO, and NO on Ni(100), Ni(111), Rh(100), Pd(100), and Pd(111) surfaces. As the revPBE functional may locally violate the Lieb-Oxford criterion, we further develop an alternative revision of the PBE functional, RPBE, which gives the same improvement of the chemisorption energies as the revPBE functional at the same time as it fulfills the Lieb-Oxford criterion locally.
30
Monkhorst, H. J.; Pack, J. D. Special points for Brillouin-zone integrations. Phys. Rev. B 1976, 13, 5188– 5192, DOI: 10.1103/PhysRevB.13.5188
Google Scholar
There is no corresponding record for this reference.
31
Xu, H.; Cheng, D.; Cao, D.; Zeng, X. C. A universal principle for a rational design of single-atom electrocatalysts. Nat. Catal. 2018, 1, 339– 348, DOI: 10.1038/s41929-018-0063-z
Google Scholar
31
A universal principle for a rational design of single-atom electrocatalysts
Xu, Haoxiang; Cheng, Daojian; Cao, Dapeng; Zeng, Xiao Cheng
Nature Catalysis (2018), 1 (5), 339-348CODEN: NCAACP; ISSN:2520-1158. (Nature Research)
Developing highly active single-atom catalysts for electrochem. reactions is a key to future renewable energy technol. Here we present a universal design principle to evaluate the activity of graphene-based single-atom catalysts towards the oxygen redn., oxygen evolution and hydrogen evolution reactions. Our results indicate that the catalytic activity of single-atom catalysts is highly correlated with the local environment of the metal center, namely its coordination no. and electronegativity and the electronegativity of the nearest neighbor atoms, validated by available exptl. data. More importantly, we reveal that this design principle can be extended to metal-macrocycle complexes. The principle not only offers a strategy to design highly active nonprecious metal single-atom catalysts with specific active centers, for example, Fe-pyridine/pyrrole-N4 for the oxygen redn. reaction; Co-pyrrole-N4 for the oxygen evolution reaction; and Mn-pyrrole-N4 for the hydrogen evolution reaction to replace precious Pt/Ir/Ru-based catalysts, but also suggests that macrocyclic metal complexes could be used as an alternative to graphene-based single-atom catalysts.
32
Martyna, G. J.; Klein, M. L.; Tuckerman, M. Nosé-Hoover chains: The canonical ensemble via continuous dynamics. J. Chem. Phys. 1992, 97, 2635– 2643, DOI: 10.1063/1.463940
Google Scholar
There is no corresponding record for this reference.
33
Tran, K.; Ulissi, Z. W. Active learning across intermetallics to guide discovery of electrocatalysts for CO₂ reduction and H₂ evolution. Nat. Catal. 2018, 1, 696– 703, DOI: 10.1038/s41929-018-0142-1
Google Scholar
33
Active learning across intermetallics to guide discovery of electrocatalysts for CO2 reduction and H2 evolution
Tran, Kevin; Ulissi, Zachary W.
Nature Catalysis (2018), 1 (9), 696-703CODEN: NCAACP; ISSN:2520-1158. (Nature Research)
The electrochem. redn. of CO2 and H2 evolution from water can be used to store renewable energy that is produced intermittently. Scale-up of these reactions requires the discovery of effective electrocatalysts, but the electrocatalyst search space is too large to explore exhaustively. Here we present a theor., fully automated screening method that uses a combination of machine learning and optimization to guide d. functional theory calcns., which are then used to predict electrocatalyst performance. We demonstrate the feasibility of this method by screening various alloys of 31 different elements, and thereby perform a screening that encompasses 50% of the d-block elements and 33% of the p-block elements. This method has thus far identified 131 candidate surfaces across 54 alloys for CO2 redn. and 258 surfaces across 102 alloys for H2 evolution. We use qual. analyses to prioritize the top candidates for exptl. validation.
34
Studt, F.; Abild-Pedersen, F.; Varley, J. B.; Nørskov, J. K. CO and CO₂ hydrogenation to methanol calculated Using the BEEF-vdW Functional. Catal. Lett. 2013, 143, 71– 73, DOI: 10.1007/s10562-012-0947-5
Google Scholar
34
CO and CO2 Hydrogenation to Methanol Calculated Using the BEEF-vdW Functional
Studt, Felix; Abild-Pedersen, Frank; Varley, Joel B.; Norskov, Jens K.
Catalysis Letters (2013), 143 (1), 71-73CODEN: CALEER; ISSN:1011-372X. (Springer)
The hydrogenation of CO and CO2 to methanol on a stepped copper surface was studied using the BEEF-vdW functional and is compared to values derived with RPBE. The inclusion of vdW forces in the BEEF-vdW functional yields a better description of CO2 hydrogenation as compared to RPBE. These differences are significant for a qual. description of the overall methanol synthesis kinetics and it is suggested that the selectivity with respect to CO and CO2 is only described correctly with BEEF-vdW.
35
Zhu, Y.-A.; Chen, D.; Zhou, X.-G.; Yuan, W.-K. DFT studies of dry reforming of methane on Ni catalyst. Catal. Today 2009, 148, 260– 267, DOI: 10.1016/j.cattod.2009.08.022
Google Scholar
35
DFT studies of dry reforming of methane on Ni catalyst
Zhu, Yi-An; Chen, De; Zhou, Xing-Gui; Yuan, Wei-Kang
Catalysis Today (2009), 148 (3-4), 260-267CODEN: CATTEA; ISSN:0920-5861. (Elsevier B.V.)
First-principles calcns. based on d. functional theory (DFT) were used to investigate the reaction mechanism of dry methane reforming on Ni(1 1 1). The most energetically favorable adsorption configurations of the species involved in this process are identified and the transition states for all the possible elementary steps are explored by the dimer method. Then, the related thermodn. properties at 973.15 K are calcd. by including the zero-point energy correction, thermal energy correction and entropic effect. CO2 dissocs. via a direct pathway to produce CO and O dominantly, and at. O is revealed to be the primary oxidant of CHx intermediates. Based on this information, two dominant reaction pathways are constructed as both the CH and C oxidn. are likely. The reaction network begins with the dissocn. of CO2 and CH4, and then the generated CH and C are oxidized by at. O to produce CHO and CO, followed by the CHO decompn. to finally generate CO and H2. As for these two reaction pathways, the oxidn. step is predicted to det. the overall reaction rate under the current investigated conditions, while the CH4 dissocn. is the rate-limiting step at lower temps.
36
Shang, H.; Sun, W.; Sui, R.; Pei, J.; Zheng, L.; Dong, J.; Jiang, Z.; Zhou, D.; Zhuang, Z.; Chen, W. Engineering Isolated Mn-N₂C₂ Atomic Interface Sites for Efficient Bifunctional Oxygen Reduction and Evolution Reaction. Nano Lett. 2020, 20, 5443– 5450, DOI: 10.1021/acs.nanolett.0c01925
Google Scholar
36
Engineering Isolated Mn-N2C2 Atomic Interface Sites for Efficient Bifunctional Oxygen Reduction and Evolution Reaction
Shang, Huishan; Sun, Wenming; Sui, Rui; Pei, Jiajing; Zheng, Lirong; Dong, Juncai; Jiang, Zhuoli; Zhou, Danni; Zhuang, Zhongbin; Chen, Wenxing; Zhang, Jiatao; Wang, Dingsheng; Li, Yadong
Nano Letters (2020), 20 (7), 5443-5450CODEN: NALEFD; ISSN:1530-6984. (American Chemical Society)
Oxygen-involved electrochem. reactions are crucial for plenty of energy conversion techniques. Herein, we rationally designed a carbon-based Mn-N2C2 bifunctional electrocatalyst. It exhibits a half-wave potential of 0.915 V vs. reversible hydrogen electrode for oxygen redn. reaction (ORR), and the overpotential is 350 mV at 10 mA cm-2 during oxygen evolution reaction (OER) in alk. condition. Furthermore, by means of operando X-ray absorption fine structure measurements, we reveal that the bond-length-extended Mn2+-N2C2 at. interface sites act as active centers during the ORR process, while the bond-length-shortened high-valence Mn4+-N2C2 moieties serve as the catalytic sites for OER, which is consistent with the d. functional theory results. The at. and electronic synergistic effects for the isolated Mn sites and the carbon support play a crit. role to promote the oxygen-involved catalytic performance, by regulating the reaction free energy of intermediate adsorption. Our results give an at. interface strategy for nonprecious bifunctional single-atom electrocatalysts.
37
Wang, Y.; Shi, R.; Shang, L.; Waterhouse, G. I. N.; Zhao, J.; Zhang, Q.; Gu, L.; Zhang, T. High-Efficiency Oxygen Reduction to Hydrogen Peroxide Catalyzed by Nickel Single-Atom Catalysts with Tetradentate N₂O₂ Coordination in a Three-Phase Flow Cell. Angew. Chem., Int. Ed. 2020, 59, 13057– 13062, DOI: 10.1002/anie.202004841
Google Scholar
37
High-Efficiency Oxygen Reduction to Hydrogen Peroxide Catalyzed by Nickel Single-Atom Catalysts with Tetradentate N2O2 Coordination in a Three-Phase Flow Cell
Wang, Yulin; Shi, Run; Shang, Lu; Waterhouse, Geoffrey I. N.; Zhao, Jiaqi; Zhang, Qinghua; Gu, Lin; Zhang, Tierui
Angewandte Chemie, International Edition (2020), 59 (31), 13057-13062CODEN: ACIEF5; ISSN:1433-7851. (Wiley-VCH Verlag GmbH & Co. KGaA)
C-supported NiII single-atom catalysts with a tetradentate Ni-N2O2 coordination formed by a Schiff base ligand-mediated pyrolysis strategy are presented. A NiII complex of the Schiff base ligand (R,R)-(-)-N,N'-bis(3,5-di-tert-butylsalicylidene)-1,2-cyclohexanediaminewasadsorbed onto a C black support, followed by pyrolysis of the modified C material at 300° in Ar. The Ni-N2O2/C catalyst showed excellent performance for the electrocatalytic redn. of O2 to H2O2 through a two-electron transfer process in alk. conditions, with a H2O2 selectivity of 96%. At a c.d. of 70 mA cm-2, a H2O2 prodn. rate of 5.9 mol gcat.-1 h-1 was achieved using a three-phase flow cell, with good catalyst stability maintained over 8 h of testing. The Ni-N2O2/C catalyst could electrocatalytically reduce O2 in air to H2O2 at a high c.d., still affording a high H2O2 selectivity (>90%). A precise Ni-N2O2 coordination was key to the performance.
38
Ren, C.; Wen, L.; Magagula, S.; Jiang, Q.; Lin, W.; Zhang, Y.; Chen, Z.; Ding, K. Relative efficacy of Co-X₄ embedded graphene (X = N, S, B, and P) electrocatalysts towards hydrogen evolution reaction: Is nitrogen really the best choice?. ChemCatChem 2020, 12, 536– 543, DOI: 10.1002/cctc.201901293
Google Scholar
38
Relative Efficacy of Co-X4 Embedded Graphene (X=N, S, B, and P) Electrocatalysts towards Hydrogen Evolution Reaction: Is Nitrogen Really the Best Choice?
Ren, Chunjin; Wen, Lu; Magagula, Saneliswa; Jiang, Qianyu; Lin, Wei; Zhang, Yongfan; Chen, Zhongfang; Ding, Kaining
ChemCatChem (2020), 12 (2), 536-543CODEN: CHEMK3; ISSN:1867-3880. (Wiley-VCH Verlag GmbH & Co. KGaA)
The authors perform 1st-principles calcns. to study whether or not N is the best dopant in system of Co-X4 embedded graphene (X = N, S, B, and P) electrocatalysts towards H evolution reaction(HER). The theor. results reveal that N, S, B, and P-doped graphene can enhance the catalytic activity toward HER compared with the pristine graphene, and S doped graphene exhibits more favorable performance than N doped graphene, consistent with the exptl. results. For the Co-X4 embedded graphene (X = N, S, B, and P), the authors predict that S may be a promising dopant in graphene supported single atom Co. The rather low H adsorption free energy (-0.07 eV) and activation energy barrier (0.78 eV) for the rate-detg. step, the downshift of the d band center, the enhanced charge d. of dz2 orbital as well as the reduced work function are responsible for the unexpected activity of Co-S4 embedded graphene for HER. Overall, Co-S4 embedded graphene catalyst could be a good candidate for H evolution reaction.
39
Chen, T.; He, T.; Benesty, M.; Khotilovich, V.; Tang, Y. Xgboost: extreme gradient boosting , R package version 0.4-2; 2015, 1. https://xgboost.readthedocs.io/en/latest/R-package/index.html
Google Scholar
There is no corresponding record for this reference.
40
Olson, R. S.; Bartley, N.; Urbanowicz, R. J.; Moore, J. H. Evaluation of a Tree-based Pipeline Optimization Tool for Automating Data Science. Proceedings of GECCO 2016, 485– 492, DOI: 10.1145/2908812.2908918
Google Scholar
There is no corresponding record for this reference.
41
Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Müller, A. Scikit-learn: Machine learning in Python. J. Mach. Learn. Res. 2011, 2825– 2830
Google Scholar
There is no corresponding record for this reference.
42
Peterson, A. A.; Abild-Pedersen, F.; Studt, F.; Rossmeisl, J.; Nørskov, J. K. How copper catalyzes the electroreduction of carbon dioxide into hydrocarbon fuels. Energy Environ. Sci. 2010, 3, 1311– 1315, DOI: 10.1039/c0ee00071j
Google Scholar
42
How copper catalyzes the electroreduction of carbon dioxide into hydrocarbon fuels
Peterson, Andrew A.; Abild-Pedersen, Frank; Studt, Felix; Rossmeisl, Jan; Norskov, Jens K.
Energy & Environmental Science (2010), 3 (9), 1311-1315CODEN: EESNBY; ISSN:1754-5706. (Royal Society of Chemistry)
D. functional theory calcns. explain copper's unique ability to convert CO2 into hydrocarbons, which may open up (photo-)electrochem. routes to fuels.
43
Kibria, M. G.; Edwards, J. P.; Gabardo, C. M.; Dinh, C. T.; Seifitokaldani, A.; Sinton, D.; Sargent, E. H. Electrochemical CO₂ Reduction into Chemical Feedstocks: From Mechanistic Electrocatalysis Models to System Design. Adv. Mater. 2019, 31, 1807166, DOI: 10.1002/adma.201807166
Google Scholar
There is no corresponding record for this reference.
44
Peterson, A. A.; Nørskov, J. K. Activity descriptors for CO₂ electroreduction to methane on transition-metal catalysts. J. Phys. Chem. Lett. 2012, 3, 251– 258, DOI: 10.1021/jz201461p
Google Scholar
44
Activity Descriptors for CO2 Electroreduction to Methane on Transition-Metal Catalysts
Peterson, Andrew A.; Noerskov, Jens K.
Journal of Physical Chemistry Letters (2012), 3 (2), 251-258CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
The electrochem. redn. of CO2 into hydrocarbons and alcs. would allow renewable energy sources to be converted into fuels and chems. However, no electrode catalysts were developed that can perform this transformation with a low overpotential at reasonable current densities. The authors compare trends in binding energies for the intermediates in CO2 electrochem. redn. and present an activity volcano based on this anal. This anal. describes the exptl. obsd. variations in transition-metal catalysts, including why Cu is the best-known metal electrocatalyst. The protonation of adsorbed CO is singled out as the most important step dictating the overpotential. New strategies are presented for the discovery of catalysts that can operate with a reduced overpotential.
45
Liu, X.; Xiao, J.; Peng, H.; Hong, X.; Chan, K.; Norskov, J. K. Understanding trends in electrochemical carbon dioxide reduction rates. Nat. Commun. 2017, 8, 15438, DOI: 10.1038/ncomms15438
Google Scholar
45
Understanding trends in electrochemical carbon dioxide reduction rates
Liu, Xinyan; Xiao, Jianping; Peng, Hongjie; Hong, Xin; Chan, Karen; Noerskov, Jens K.
Nature Communications (2017), 8 (), 15438CODEN: NCAOBW; ISSN:2041-1723. (Nature Publishing Group)
Electrochem. carbon dioxide redn. to fuels presents one of the great challenges in chem. Herein we present an understanding of trends in electrocatalytic activity for carbon dioxide redn. over different metal catalysts that rationalize a no. of exptl. observations including the selectivity with respect to the competing hydrogen evolution reaction. We also identify two design criteria for more active catalysts. The understanding is based on d. functional theory calcns. of activation energies for electrochem. carbon monoxide redn. as a basis for an electrochem. kinetic model of the process. We develop scaling relations relating transition state energies to the carbon monoxide adsorption energy and det. the optimal value of this descriptor to be very close to that of copper.
46
Zhong, M.; Tran, K.; Min, Y.; Wang, C.; Wang, Z.; Dinh, C.-T.; De Luna, P.; Yu, Z.; Rasouli, A. S.; Brodersen, P. Accelerated discovery of CO₂ electrocatalysts using active machine learning. Nature 2020, 581, 178– 183, DOI: 10.1038/s41586-020-2242-8
Google Scholar
46
Accelerated discovery of CO2 electrocatalysts using active machine learning
Zhong, Miao; Tran, Kevin; Min, Yimeng; Wang, Chuanhao; Wang, Ziyun; Dinh, Cao-Thang; De Luna, Phil; Yu, Zongqian; Rasouli, Armin Sedighian; Brodersen, Peter; Sun, Song; Voznyy, Oleksandr; Tan, Chih-Shan; Askerka, Mikhail; Che, Fanglin; Liu, Min; Seifitokaldani, Ali; Pang, Yuanjie; Lo, Shen-Chuan; Ip, Alexander; Ulissi, Zachary; Sargent, Edward H.
Nature (London, United Kingdom) (2020), 581 (7807), 178-183CODEN: NATUAS; ISSN:0028-0836. (Nature Research)
The rapid increase in global energy demand and the need to replace carbon dioxide (CO2)-emitting fossil fuels with renewable sources have driven interest in chem. storage of intermittent solar and wind energy1,2. Particularly attractive is the electrochem. redn. of CO2 to chem. feedstocks, which uses both CO2 and renewable energy3-8. Copper has been the predominant electrocatalyst for this reaction when aiming for more valuable multi-carbon products9-16, and process improvements have been particularly notable when targeting ethylene. However, the energy efficiency and productivity (c.d.) achieved so far still fall below the values required to produce ethylene at cost-competitive prices. Here we describe Cu-Al electrocatalysts, identified using d. functional theory calcns. in combination with active machine learning, that efficiently reduce CO2 to ethylene with the highest Faradaic efficiency reported so far. This Faradaic efficiency of over 80 per cent (compared to about 66 per cent for pure Cu) is achieved at a c.d. of 400 mA per square centimetre (at 1.5 V vs. a reversible hydrogen electrode) and a cathodic-side (half-cell) ethylene power conversion efficiency of 55 ± 2 per cent at 150 mA per square centimetre. We perform computational studies that suggest that the Cu-Al alloys provide multiple sites and surface orientations with near-optimal CO binding for both efficient and selective CO2 redn.17. Furthermore, in situ X-ray absorption measurements reveal that Cu and Al enable a favorable Cu coordination environment that enhances C-C dimerization. These findings illustrate the value of computation and machine learning in guiding the exptl. exploration of multi-metallic systems that go beyond the limitations of conventional single-metal electrocatalysts.
47
Wang, Y.; Shi, R.; Shang, L.; Waterhouse, G. I.; Zhao, J.; Zhang, Q.; Gu, L.; Zhang, T. High-Efficiency Oxygen Reduction to Hydrogen Peroxide Catalyzed by Nickel Single-Atom Catalysts with Tetradentate N₂O₂ Coordination in a Three-Phase Flow Cell. Angew. Chem., Int. Ed. 2020, 59, 13057– 13062, DOI: 10.1002/anie.202004841
Google Scholar
47
High-Efficiency Oxygen Reduction to Hydrogen Peroxide Catalyzed by Nickel Single-Atom Catalysts with Tetradentate N2O2 Coordination in a Three-Phase Flow Cell
Wang, Yulin; Shi, Run; Shang, Lu; Waterhouse, Geoffrey I. N.; Zhao, Jiaqi; Zhang, Qinghua; Gu, Lin; Zhang, Tierui
Angewandte Chemie, International Edition (2020), 59 (31), 13057-13062CODEN: ACIEF5; ISSN:1433-7851. (Wiley-VCH Verlag GmbH & Co. KGaA)
C-supported NiII single-atom catalysts with a tetradentate Ni-N2O2 coordination formed by a Schiff base ligand-mediated pyrolysis strategy are presented. A NiII complex of the Schiff base ligand (R,R)-(-)-N,N'-bis(3,5-di-tert-butylsalicylidene)-1,2-cyclohexanediaminewasadsorbed onto a C black support, followed by pyrolysis of the modified C material at 300° in Ar. The Ni-N2O2/C catalyst showed excellent performance for the electrocatalytic redn. of O2 to H2O2 through a two-electron transfer process in alk. conditions, with a H2O2 selectivity of 96%. At a c.d. of 70 mA cm-2, a H2O2 prodn. rate of 5.9 mol gcat.-1 h-1 was achieved using a three-phase flow cell, with good catalyst stability maintained over 8 h of testing. The Ni-N2O2/C catalyst could electrocatalytically reduce O2 in air to H2O2 at a high c.d., still affording a high H2O2 selectivity (>90%). A precise Ni-N2O2 coordination was key to the performance.
48
Feng, B.; Zhang, J.; Zhong, Q.; Li, W.; Li, S.; Li, H.; Cheng, P.; Meng, S.; Chen, L.; Wu, K. Experimental realization of two-dimensional boron sheets. Nat. Chem. 2016, 8, 563– 8, DOI: 10.1038/nchem.2491
Google Scholar
48
Experimental realization of two-dimensional boron sheets
Feng, Baojie; Zhang, Jin; Zhong, Qing; Li, Wenbin; Li, Shuai; Li, Hui; Cheng, Peng; Meng, Sheng; Chen, Lan; Wu, Kehui
Nature Chemistry (2016), 8 (6), 563-568CODEN: NCAHBB; ISSN:1755-4330. (Nature Publishing Group)
Two-dimensional boron sheets have been grown epitaxially on Ag(111) substrate. Two types of boron sheet, a β12 sheet and a χ3 sheet, both exhibiting a triangular lattice but with different arrangements of periodic holes, are obsd. by scanning tunnelling microscopy. D. functional theory simulations agree well with expts., and indicate that both sheets are planar without obvious vertical undulations. The boron sheets are quite inert to oxidization and interact only weakly with their substrate.
49
Ling, C.; Shi, L.; Ouyang, Y.; Zeng, X. C.; Wang, J. Nanosheet Supported Single-Metal Atom Bifunctional Catalyst for Overall Water Splitting. Nano Lett. 2017, 17, 5133– 5139, DOI: 10.1021/acs.nanolett.7b02518
Google Scholar
49
Nanosheet Supported Single-Metal Atom Bifunctional Catalyst for Overall Water Splitting
Ling, Chongyi; Shi, Li; Ouyang, Yixin; Zeng, Xiao Cheng; Wang, Jinlan
Nano Letters (2017), 17 (8), 5133-5139CODEN: NALEFD; ISSN:1530-6984. (American Chemical Society)
Nanosheet supported single-atom catalysts (SACs) can make full use of metal atoms and yet entail high selectivity and activity, and bifunctional catalysts can enable higher performance while lowering the cost than two sep. unifunctional catalysts. Supported single-atom bifunctional catalysts are therefore of great economic interest and scientific importance. Here, on the basis of first-principles computations, we report a design of the first single-atom bifunctional electrocatalyst, namely, isolated nickel atom supported on β12 boron monolayer (Ni1/β12-BM), to achieve overall water splitting. This nanosheet supported SAC exhibits remarkable electrocatalytic performance with the computed overpotential for oxygen/hydrogen evolution reaction being just 0.40/0.06 V. The ab initio mol. dynamics simulation shows that the SAC can survive up to 800 K elevated temp., while enacting a high energy barrier of 1.68 eV to prevent isolated Ni atoms from clustering. A viable exptl. route for the synthesis of Ni1/β12-BM SAC is demonstrated from computer simulation. The desired nanosheet supported single-atom bifunctional catalysts not only show great potential for achieving overall water splitting but also offer cost-effective opportunities for advancing clean energy technol.

Cited By

Click to copy section linkSection link copied!

Supporting
Supporting2
Mentioning
Mentioning159
Contrasting
Contrasting0

Explore this article's citation statements on scite.ai

This article is cited by 164 publications.

Jikai Sun, Jianzhong Wu. Unveiling Double Layer Effects in Electrocatalytic CO2 Reduction. The Journal of Physical Chemistry C 2025, 129 (19) , 8946-8954. https://doi.org/10.1021/acs.jpcc.5c02134
Zhongze Bai, Zhuo Zhi, Xi Zhuo Jiang, Kai H. Luo. Rational Design of Dual-Atom Catalysts for Electrochemical CO2 Reduction to C1 and C2 Products with High Activity and Selectivity: A Density Functional Theory Study. Industrial & Engineering Chemistry Research 2025, 64 (8) , 4378-4387. https://doi.org/10.1021/acs.iecr.4c04831
Yanyan Xia, Yihui Bao, Xinyi Lu, Zhencheng Ye, Yuhan Mei, Houyang Chen. Enhancing Electrocatalysis of CO2 to Ethanol via Intercalated Electron Boosters in an Atomically Dispersed Ca–N4-Doped Graphene Bilayer. Industrial & Engineering Chemistry Research 2025, 64 (4) , 2016-2024. https://doi.org/10.1021/acs.iecr.4c03238
Nana Zhou, Yaling Zhao, Qingzhang Lv, Yahong Chen. Using Machine Learning to Forecast the Conductive Substrate-Supported Heteroatom-Doped Metal Compound Electrocatalysts for Hydrogen Evolution Reaction. The Journal of Physical Chemistry C 2024, 128 (41) , 17274-17281. https://doi.org/10.1021/acs.jpcc.4c03846
Ruchi Agarwalla Riya Mudoi Unnati Bora Jyotirmoy Deb Madhulekha Gogoi Lakshi Saikia . Recent Advancements in Scalable Hydrogen Generation: An Integrated Approach of Experiments, Computation, and Machine Learning. , 25-45. https://doi.org/10.1021/bk-2024-1468.ch002
Xinyuan Cao, Jisi Huang, Kexin Du, Yawen Tian, Zhixin Hu, Zhu Luo, Jinlong Wang, Yanbing Guo. Machine-Learning-Assisted Descriptors Identification for Indoor Formaldehyde Oxidation Catalysts. Environmental Science & Technology 2024, 58 (19) , 8372-8379. https://doi.org/10.1021/acs.est.4c01691
Lourdes F. Vega, Daniel Bahamon, Ismail I. I. Alkhatib. Perspectives on Advancing Sustainable CO2 Conversion Processes: Trinomial Technology, Environment, and Economy. ACS Sustainable Chemistry & Engineering 2024, 12 (14) , 5357-5382. https://doi.org/10.1021/acssuschemeng.3c07133
Javad Shirani, Julio J. Valdes, Alain B. Tchagang, Kirk H. Bevan. Adsorbate-Dependent Electronic Structure Descriptors for Machine Learning-Driven Binding Energy Predictions in Diverse Single Atom Alloys: A Reductionist Approach. The Journal of Physical Chemistry C 2024, 128 (11) , 4483-4496. https://doi.org/10.1021/acs.jpcc.3c07398
Qin Zhu, Yating Gu, Xinzhu Wang, Yuming Gu, Jing Ma. The Synergistic Effect between Metal and Sulfur Vacancy to Boost CO2 Reduction Efficiency: A Study on Descriptor Transferability and Activity Prediction. JACS Au 2024, 4 (1) , 125-138. https://doi.org/10.1021/jacsau.3c00558
Haisong Feng, Meng Zhang, Zhen Ge, Yuan Deng, Pengxin Pu, Wenyu Zhou, Hao Yuan, Jing Yang, Feng Li, Xin Zhang, Yong-Wei Zhang. Designing Efficient Single-Atom Alloy Catalysts for Selective C═O Hydrogenation: A First-Principles, Active Learning and Microkinetic Study. ACS Applied Materials & Interfaces 2023, 15 (48) , 55903-55915. https://doi.org/10.1021/acsami.3c15108
Atiyeh Bashiri, Ali Sufali, Mahsa Golmohammadi, Ali Mohammadi, Reza Maleki, Abdollah Jamal Sisi, Alireza Khataee, Mohsen Asadnia, Amir Razmjou. Unveiling the Mechanisms of Catalytic CO2 Electroreduction through Machine Learning. Industrial & Engineering Chemistry Research 2023, 62 (47) , 20189-20201. https://doi.org/10.1021/acs.iecr.3c02698
Sergei Manzhos, Manabu Ihara. Neural Network with Optimal Neuron Activation Functions Based on Additive Gaussian Process Regression. The Journal of Physical Chemistry A 2023, 127 (37) , 7823-7835. https://doi.org/10.1021/acs.jpca.3c02949
Chen Chen, Bo Xiao, Zhengkun Qin, Jingxiang Zhao, Wenzuo Li, Qingzhong Li, Xuefang Yu. Metal-Doped C3B Monolayer as the Promising Electrocatalyst for Hydrogen/Oxygen Evolution Reaction: A Combined Density Functional Theory and Machine Learning Study. ACS Applied Materials & Interfaces 2023, 15 (34) , 40538-40548. https://doi.org/10.1021/acsami.3c07790
Linke Yu, Fengyu Li, Jingsong Huang, Bobby G. Sumpter, William E. Mustain, Zhongfang Chen. Double-Atom Catalysts Featuring Inverse Sandwich Structure for CO2 Reduction Reaction: A Synergetic First-Principles and Machine Learning Investigation. ACS Catalysis 2023, 13 (14) , 9616-9628. https://doi.org/10.1021/acscatal.3c01584
Yang Gao, Ludi Wang, Xueqing Chen, Yi Du, Bin Wang. Revisiting Electrocatalyst Design by a Knowledge Graph of Cu-Based Catalysts for CO2 Reduction. ACS Catalysis 2023, 13 (13) , 8525-8534. https://doi.org/10.1021/acscatal.3c00759
Haobo Li, Yunling Jiang, Xinyu Li, Kenneth Davey, Yao Zheng, Yan Jiao, Shi-Zhang Qiao. C2+ Selectivity for CO2 Electroreduction on Oxidized Cu-Based Catalysts. Journal of the American Chemical Society 2023, 145 (26) , 14335-14344. https://doi.org/10.1021/jacs.3c03022
Javad Shirani, Hanh D. M. Pham, Shuaishuai Yuan, Alain B. Tchagang, Julio J. Valdés, Kirk H. Bevan. Machine Learning Based Electronic Structure Predictors in Single-Atom Alloys: A Model Study of CO Kink-Site Adsorption across Transition Metal Substrates. The Journal of Physical Chemistry C 2023, 127 (25) , 12055-12067. https://doi.org/10.1021/acs.jpcc.3c02705
Honghao Chen, Yingzhe Zheng, Jiali Li, Lanyu Li, Xiaonan Wang. AI for Nanomaterials Development in Clean Energy and Carbon Capture, Utilization and Storage (CCUS). ACS Nano 2023, 17 (11) , 9763-9792. https://doi.org/10.1021/acsnano.3c01062
Lujun Li, Yiming Zhao, Haibin Yu, Zhuo Wang, Yongjia Zhao, Mingqi Jiang. An XGBoost Algorithm Based on Molecular Structure and Molecular Specificity Parameters for Predicting Gas Adsorption. Langmuir 2023, 39 (19) , 6756-6766. https://doi.org/10.1021/acs.langmuir.3c00255
Kajjana Boonpalit, Yutthana Wongnongwa, Chanatkran Prommin, Sarana Nutanong, Supawadee Namuangruk. Data-Driven Discovery of Graphene-Based Dual-Atom Catalysts for Hydrogen Evolution Reaction with Graph Neural Network and DFT Calculations. ACS Applied Materials & Interfaces 2023, 15 (10) , 12936-12945. https://doi.org/10.1021/acsami.2c19391
Hu Ding, Yawen Shi, Zeyang Li, Si Wang, Yujie Liang, Haisong Feng, Yuan Deng, Xin Song, Pengxin Pu, Xin Zhang. Active Learning Accelerating to Screen Dual-Metal-Site Catalysts for Electrochemical Carbon Dioxide Reduction Reaction. ACS Applied Materials & Interfaces 2023, 15 (10) , 12986-12997. https://doi.org/10.1021/acsami.2c21332
Diptendu Roy, Amitabha Das, Souvik Manna, Biswarup Pathak. A Route Map of Machine Learning Approaches in Heterogeneous CO2 Reduction Reaction. The Journal of Physical Chemistry C 2023, 127 (2) , 871-881. https://doi.org/10.1021/acs.jpcc.2c06924
Erhai Hu, Chuntai Liu, Wei Zhang, Qingyu Yan. Machine Learning Assisted Understanding and Discovery of CO2 Reduction Reaction Electrocatalyst. The Journal of Physical Chemistry C 2023, 127 (2) , 882-893. https://doi.org/10.1021/acs.jpcc.2c08343
Miaojuan Xing, Yunjiang Zhang, Shuyuan Li, Hong He, Shaorui Sun. Prediction of Carbon Dioxide Reduction Catalyst Using Machine Learning with a Few-Feature Model: WLEDZ. The Journal of Physical Chemistry C 2022, 126 (40) , 17025-17035. https://doi.org/10.1021/acs.jpcc.2c02161
Qin Zhu, Yuming Gu, Xinyi Liang, Xinzhu Wang, Jing Ma. A Machine Learning Model To Predict CO2 Reduction Reactivity and Products Transferred from Metal-Zeolites. ACS Catalysis 2022, 12 (19) , 12336-12348. https://doi.org/10.1021/acscatal.2c03250
Xu Zhang, Yun Tian, Letian Chen, Xu Hu, Zhen Zhou. Machine Learning: A New Paradigm in Computational Electrocatalysis. The Journal of Physical Chemistry Letters 2022, 13 (34) , 7920-7930. https://doi.org/10.1021/acs.jpclett.2c01710
Adeesh Kolluru, Muhammed Shuaibi, Aini Palizhati, Nima Shoghi, Abhishek Das, Brandon Wood, C. Lawrence Zitnick, John R. Kitchin, Zachary W. Ulissi. Open Challenges in Developing Generalizable Large-Scale Machine-Learning Models for Catalyst Discovery. ACS Catalysis 2022, 12 (14) , 8572-8581. https://doi.org/10.1021/acscatal.2c02291
Haisong Feng, Hu Ding, Si Wang, Yujie Liang, Yuan Deng, Yusen Yang, Min Wei, Xin Zhang. Machine-Learning-Assisted Catalytic Performance Predictions of Single-Atom Alloys for Acetylene Semihydrogenation. ACS Applied Materials & Interfaces 2022, 14 (22) , 25288-25296. https://doi.org/10.1021/acsami.2c02317
Xinyan Liu, Cheng Cai, Wanghui Zhao, Hong-Jie Peng, Tao Wang. Machine Learning-Assisted Screening of Stepped Alloy Surfaces for C1 Catalysis. ACS Catalysis 2022, 12 (8) , 4252-4260. https://doi.org/10.1021/acscatal.2c00648
Xu Zhang, Zhen Zhou. Perspective on Theoretical Models for CO2 Electrochemical Reduction. The Journal of Physical Chemistry C 2022, 126 (8) , 3820-3829. https://doi.org/10.1021/acs.jpcc.1c10870
Ziqing Shui, Yu Wang, Yafei Li. Activity Origin of Antimony Nanosheets toward Selective Electroreduction of CO2 to Formic Acid. The Journal of Physical Chemistry C 2022, 126 (8) , 4015-4023. https://doi.org/10.1021/acs.jpcc.1c10916
Sean Overa, Byung Hee Ko, Yaran Zhao, Feng Jiao. Electrochemical Approaches for CO2 Conversion to Chemicals: A Journey toward Practical Applications. Accounts of Chemical Research 2022, 55 (5) , 638-648. https://doi.org/10.1021/acs.accounts.1c00674
Fanping Sui, Ruiqi Guo, Zhizhou Zhang, Grace X. Gu, Liwei Lin. Deep Reinforcement Learning for Digital Materials Design. ACS Materials Letters 2021, 3 (10) , 1433-1439. https://doi.org/10.1021/acsmaterialslett.1c00390
Hao Yuan, Zhenyu Li. Intrinsic Descriptors for Coordination Environment and Synergistic Effects of Metal and Environment in Single-Atom-Catalyzed Carbon Dioxide Electroreduction. The Journal of Physical Chemistry C 2021, 125 (33) , 18180-18186. https://doi.org/10.1021/acs.jpcc.1c04637
Jaehyun Kim, Donghoon Kang, Sangbum Kim, Ho Won Jang. Catalyze Materials Science with Machine Learning. ACS Materials Letters 2021, 3 (8) , 1151-1171. https://doi.org/10.1021/acsmaterialslett.1c00204
Donghuan Wu, Jiayi Zhang, Mu-Jeng Cheng, Qi Lu, Haochen Zhang. Machine Learning Investigation of Supplementary Adsorbate Influence on Copper for Enhanced Electrochemical CO2 Reduction Performance. The Journal of Physical Chemistry C 2021, 125 (28) , 15363-15372. https://doi.org/10.1021/acs.jpcc.1c05004
Dunfeng Gao, Tianfu Liu, Guoxiong Wang, Xinhe Bao. Structure Sensitivity in Single-Atom Catalysis toward CO2 Electroreduction. ACS Energy Letters 2021, 6 (2) , 713-727. https://doi.org/10.1021/acsenergylett.0c02665
Andrew L. Ferguson, (Guest Editor)Johannes Hachmann, (Guest Editor)Thomas F. Miller, (Guest Editor)Jim Pfaendtner (Guest Editor). The Journal of Physical Chemistry A/B/C Virtual Special Issue on Machine Learning in Physical Chemistry. The Journal of Physical Chemistry A 2020, 124 (44) , 9113-9118. https://doi.org/10.1021/acs.jpca.0c09205
Andrew L. Ferguson, (Guest Editor)Johannes Hachmann, (Guest Editor)Thomas F. Miller, (Guest Editor)Jim Pfaendtner (Guest Editor). The Journal of Physical Chemistry A/B/C Virtual Special Issue on Machine Learning in Physical Chemistry. The Journal of Physical Chemistry B 2020, 124 (44) , 9767-9772. https://doi.org/10.1021/acs.jpcb.0c09206
Andrew L. Ferguson, (Guest Editor)Johannes Hachmann, (Guest Editor)Thomas F. Miller, (Guest Editor)Jim Pfaendtner (Guest Editor). The Journal of Physical Chemistry A/B/C Virtual Special Issue on Machine Learning in Physical Chemistry. The Journal of Physical Chemistry C 2020, 124 (44) , 24033-24038. https://doi.org/10.1021/acs.jpcc.0c09208
Huan Wang, Jikai Sun, Youyong Li, Weiqiao Deng. Machine learning and DFT database for C-H dissociation on single-atom alloy surfaces in methane decomposition. Scientific Data 2025, 12 (1) https://doi.org/10.1038/s41597-025-04885-1
Hwanyeol Park, Dae-Myeong Geum, Ho Jun Kim. Accelerating hydrogen evolution catalyst discovery via data-driven strategy for high-performance single-atom catalysts embedded in h-BN. Journal of Energy Chemistry 2025, 107 , 750-758. https://doi.org/10.1016/j.jechem.2025.04.002
Esraa Kotob, Mohammed Mosaad Awad, Mustapha Umar, Omer Ahmed Taialla, Ijaz Hussain, Shaima’ Ibrahim Alsabbahen, Khalid Alhooshani, Saheed A. Ganiyu. Unlocking CO2 conversion potential with single atom catalysts and machine learning in energy application. iScience 2025, 28 (6) , 112306. https://doi.org/10.1016/j.isci.2025.112306
Zhongze Bai, Xi Zhuo Jiang, Kai H. Luo. Enhanced CO2 electrochemical reduction on single-atom catalysts with optimized environmental, central and axial chemical ambient. Journal of Colloid and Interface Science 2025, 686 , 1188-1199. https://doi.org/10.1016/j.jcis.2025.02.015
Kexiang Guo, Xinyu Fan, Letian Chen, Xu Zhang, Zhen Zhou. Machine learning-based design of electrocatalysts and catalytic mechanism research. SCIENTIA SINICA Chimica 2025, 48 https://doi.org/10.1360/SSC-2025-0052
Rana Rashad Mahmood Khan, Ramsha Saleem, Syeda Satwat Batool, Shahzad Rasheed, Zohaib Saeed, Muhammad Pervaiz, Umer Younas, Muhammad Summer, Maira Liaqat. Electrochemical reduction of CO2 to liquid products: Factors influencing production and selectivity. International Journal of Hydrogen Energy 2025, 128 , 800-832. https://doi.org/10.1016/j.ijhydene.2025.04.077
Guo‐Jin Cao. Advancements in Machine Learning Predicting Activation and Gibbs Free Energies in Chemical Reactions. International Journal of Quantum Chemistry 2025, 125 (7) https://doi.org/10.1002/qua.70036
Qiaochu Shi, Boyu Zhang, Zhenhua Wu, Dong Yang, Hong Wu, Jiafu Shi, Zhongyi Jiang. Cascade Catalytic Systems for Converting CO 2 into C 2+ Products. ChemSusChem 2025, 18 (7) https://doi.org/10.1002/cssc.202401916
Yuhang Wang, Yaqin Zhang, Ninggui Ma, Jun Zhao, Yu Xiong, Shuang Luo, Jun Fan. Machine learning accelerated catalysts design for CO reduction: An interpretability and transferability analysis. Journal of Materials Science & Technology 2025, 213 , 14-23. https://doi.org/10.1016/j.jmst.2024.05.068
Wenyu Zhou, Haisong Feng, Shihong Zhou, Mengxin Wang, Yuping Chen, Chenyang Lu, Hao Yuan, Jing Yang, Qun Li, Luxi Tan, Lichun Dong, Yong‐Wei Zhang. Designing and screening single‐atom alloy catalysts for CO 2 reduction to CH 3 OH via DFT and machine learning. AIChE Journal 2025, 71 (3) https://doi.org/10.1002/aic.18678
Xueying Li, Woojong Kang, Xinyi Fan, Xinyi Tan, Justus Masa, Alex W. Robertson, Yousung Jung, Buxing Han, John Texter, Yuanfu Cheng, Bin Dai, Zhenyu Sun. Electrochemical CO2 reduction to liquid fuels: Mechanistic pathways and surface/interface engineering of catalysts and electrolytes. The Innovation 2025, 6 (3) , 100807. https://doi.org/10.1016/j.xinn.2025.100807
Tianyi Wang, Qilong Wu, Yun Han, Zhongyuan Guo, Jun Chen, Chuangwei Liu. Advanced theoretical modeling methodologies for electrocatalyst design in sustainable energy conversion. Applied Physics Reviews 2025, 12 (1) https://doi.org/10.1063/5.0235572
Qiumei Yu, Ninggui Ma, Chihon Leung, Han Liu, Yang Ren, Zhanhua Wei. AI in single-atom catalysts: a review of design and applications. Journal of Materials Informatics 2025, 5 (1) https://doi.org/10.20517/jmi.2024.78
Zhongjie Huang, Zilong Chen, Jiawen Cheng, Jiaqi Zhang, Shuyi Wang, Tingting Chen, Xiaodan Zhang, Huan Pang. Recent Progress in High‐Throughput On‐Chip Synthesis, Screening, and Data‐Driven Optimization: Toward an Electrocatalyst Chip for Catalysis Universe Exploration. Advanced Functional Materials 2025, 35 (9) https://doi.org/10.1002/adfm.202416117
Chen-Chen Er, Lutfi K. Putri, Yee Sin Ang, Siang-Piao Chai. Unveiling fundamental first-principles insights into single-atom transition metal photocatalysts for carbon dioxide reduction. Fuel 2025, 382 , 133746. https://doi.org/10.1016/j.fuel.2024.133746
Yiheng Huang, Jiarui Wang, Hui Hu, Zhengping Qiao, Yan Li, Chengxin Wang. Heteroatom-doped M-N4-C Single-atom catalysts towards electrochemical reactions of CO2: A machine learning-assisted DFT study. Molecular Catalysis 2025, 572 , 114793. https://doi.org/10.1016/j.mcat.2024.114793
Maksymilian Mądziel. Investigating Real-world Emissions from Liquefied Petroleum Gas-fueled Vehicles: A Modeling Approach that Utilizes Portable Emissions Measurement Systems. The Open Transportation Journal 2025, 19 (1) https://doi.org/10.2174/0126671212367963241223053201
Dongxu Jiao, Xinyi Li, Mingzi Sun, Lin Liu, Jinchang Fan, Jingxiang Zhao, Bolong Huang, Xiaoqiang Cui. Machine learning driven rational design of dual atom catalysts on graphene for carbon dioxide electroreduction. Nano Research 2025, 18 (1) , 94907044. https://doi.org/10.26599/NR.2025.94907044
Vuri Ayu Setyowati, Shiho Mukaida, Kaito Nagita, Takashi Harada, Shuji Nakanishi, Kazuyuki Iwase. Machine Learning Exploration of Experimental Conditions for Optimized Electrochemical CO 2 Reduction. ChemElectroChem 2024, 11 (24) https://doi.org/10.1002/celc.202400518
Jikai Sun, Rui Tu, Yuchun Xu, Hongyan Yang, Tie Yu, Dong Zhai, Xiuqin Ci, Weiqiao Deng. Machine learning aided design of single-atom alloy catalysts for methane cracking. Nature Communications 2024, 15 (1) https://doi.org/10.1038/s41467-024-50417-7
Jun Zhu, Mengdan Song, Qiling Qian, Yang Yue, Guangren Qian, Jia Zhang. Machine learning for deconstructing contributions of atomic characterizations to achieve hybridization-determined electron transfer in a perovskite catalyst. Journal of Materials Chemistry A 2024, 12 (44) , 30722-30728. https://doi.org/10.1039/D4TA05018E
Yang Liu, Zefei Wu, Chen Gu, Jianmei Chen, Yanwei Zhu, Longlu Wang. Curved Structure Regulated Single Metal Sites for Advanced Electrocatalytic Reactions. Small 2024, 20 (47) https://doi.org/10.1002/smll.202404758
Prince Joby, Yesaiyan Manojkumar, Antony Rajendran, Rajadurai Vijay Solomon. Computational catalysis on the conversion of CO2 to methane—an update. Frontiers of Chemical Science and Engineering 2024, 18 (11) https://doi.org/10.1007/s11705-024-2484-3
Nepal Sahu, Chandrashekhar Azad, Uday Kumar. Study and prediction of photocurrent density with external validation using machine learning models. International Journal of Hydrogen Energy 2024, 92 , 1335-1355. https://doi.org/10.1016/j.ijhydene.2024.10.339
Chen Liang, Bowen Wang, Shaogang Hao, Guangyong Chen, Pheng‐Ann Heng, Xiaolong Zou. Multi‐Task Mixture Density Graph Neural Networks for Predicting Catalyst Performance. Advanced Functional Materials 2024, 34 (45) https://doi.org/10.1002/adfm.202404392
Xiaoyun Lin, Shiyu Zhen, Xiaohui Wang, Lyudmila V. Moskaleva, Peng Zhang, Zhi-Jian Zhao, Jinlong Gong. Data-Driven Design of Single-Atom Electrocatalysts with Intrinsic Descriptors for Carbon Dioxide Reduction Reaction. Transactions of Tianjin University 2024, 30 (5) , 459-469. https://doi.org/10.1007/s12209-024-00413-1
M Karthikeyan, Durga Madhab Mahapatra, Abdul Syukor Abd Razak, Abdulaziz A.M. Abahussain, Baranitharan Ethiraj, Lakhveer Singh. Machine learning aided synthesis and screening of HER catalyst: Present developments and prospects. Catalysis Reviews 2024, 66 (4) , 997-1027. https://doi.org/10.1080/01614940.2022.2103980
Zijing Li, Yingchuan Zhang, Tao Zhou, Guangri Jia. Accelerating electrocatalyst design for CO2 conversion through machine learning: Interpretable models and data-driven innovations. Nexus 2024, 1 (3) , 100029. https://doi.org/10.1016/j.ynexs.2024.100029
Qiang Wang, Hehe Wei, Ping Liu, Zixiang Su, Xue-Qing Gong. Recent advances in copper-based catalysts for electrocatalytic CO 2 reduction toward multi-carbon products. Nano Research Energy 2024, 3 (3) , e9120112. https://doi.org/10.26599/NRE.2024.9120112
Xinyan Liu, Hong-Jie Peng. Toward Next-Generation Heterogeneous Catalysts: Empowering Surface Reactivity Prediction with Machine Learning. Engineering 2024, 39 , 25-44. https://doi.org/10.1016/j.eng.2023.07.021
Oliver Schilter, Philippe Schwaller, Teodoro Laino. Balancing computational chemistry's potential with its environmental impact. Green Chemistry 2024, 26 (15) , 8669-8679. https://doi.org/10.1039/D4GC01745E
Huang Qin, Hai Zhang, Kunmin Wu, Xingzi Wang, Weidong Fan. A systematic theoretical study of CO 2 hydrogenation towards methanol on Cu-based bimetallic catalysts: role of the CHO&CH 3 OH descriptor in thermodynamic analysis. Physical Chemistry Chemical Physics 2024, 26 (28) , 19088-19104. https://doi.org/10.1039/D4CP01009D
Sachidananda Nayak, Selvakumar Karuthapandi. Machine Learning Approaches to Catalysis. 2024, 101-125. https://doi.org/10.1002/9781394214167.ch9
Jihyeon Park, Jaeyoung Lee. Electrochemical energy conversion and storage processes with machine learning. Trends in Chemistry 2024, 6 (6) , 302-313. https://doi.org/10.1016/j.trechm.2024.04.007
H. Oliaei, N. R. Aluru. Study of the adsorption sites of high entropy alloys for CO2 reduction using graph convolutional network. APL Machine Learning 2024, 2 (2) https://doi.org/10.1063/5.0198043
Maria G. Minotaki, Julian Geiger, Andrea Ruiz-Ferrando, Albert Sabadell-Rendón, Núria López. A generalized model for estimating adsorption energies of single atoms on doped carbon materials. Journal of Materials Chemistry A 2024, 12 (18) , 11049-11061. https://doi.org/10.1039/D3TA05898K
Haoyan Zhang, Lin Cheng, Kai Li, Ying Wang, Zhijian Wu. CO 2 Electrochemical reduction on the two dimensional transition metal coordinated by 2,3,6,7,10,11-triphenylenehexathiol and 2,3,6,7,10,11-triphenylenehexamine, a computational survey. Molecular Physics 2024, 122 (9) https://doi.org/10.1080/00268976.2023.2274503
Lanjing Wang, Honghao Chen, Longqi Yang, Jiali Li, Yong Li, Xiaonan Wang. Single-atom catalysts property prediction via Supervised and Self-Supervised pre-training models. Chemical Engineering Journal 2024, 487 , 150626. https://doi.org/10.1016/j.cej.2024.150626
Sung Eun Jerng, Yang Jeong Park, Ju Li. Machine learning for CO 2 capture and conversion: A review. Energy and AI 2024, 16 , 100361. https://doi.org/10.1016/j.egyai.2024.100361
Qing-Meng Zhang, Zhao-Yu Wang, Hao Zhang, Xiao-Hong Liu, Wei Zhang, Liu-Bin Zhao. Micro-kinetic modelling of the CO reduction reaction on single atom catalysts accelerated by machine learning. Physical Chemistry Chemical Physics 2024, 26 (14) , 11037-11047. https://doi.org/10.1039/D4CP00325J
Yaxin Shi, Zhiqin Liang. Machine learning accelerates the screening of single-atom catalysts towards CO2 electroreduction. Applied Catalysis A: General 2024, 676 , 119674. https://doi.org/10.1016/j.apcata.2024.119674
Ilya V. Chepkasov, Aleksandra D. Radina, Alexander G. Kvashnin. Structure-driven tuning of catalytic properties of core–shell nanostructures. Nanoscale 2024, 16 (12) , 5870-5892. https://doi.org/10.1039/D3NR06194A
, Meilin Ren. Risk Identification and Countermeasures of Wetland Park Construction Project. Scientific Research Bulletin 2024, 1 (2) , 1-3. https://doi.org/10.71052/srb2024/JVNH8431
Shuaichong Wei, Yuhong Luo, Kai Zhang, Zisheng Zhang, Guihua Liu. Machine learning assisted prediction of copper-based catalysts towards carbon dioxide electroreduction into carbon monoxide. Chemical Physics 2024, 579 , 112197. https://doi.org/10.1016/j.chemphys.2024.112197
Eslam G. Al-Sakkari, Ahmed Ragab, Hanane Dagdougui, Daria C. Boffito, Mouloud Amazouz. Carbon capture, utilization and sequestration systems design and operation optimization: Assessment and perspectives of artificial intelligence opportunities. Science of The Total Environment 2024, 917 , 170085. https://doi.org/10.1016/j.scitotenv.2024.170085
Yipin Lv, Guozhu Chen, Rongwei Ma, Jin Yong Lee, Baotao Kang. Hybrid scheme of DFT and machine learning to accelerate the design of graphyne nanoribbons as electrocatalysts for the ORR and HER. Fuel 2024, 357 , 130017. https://doi.org/10.1016/j.fuel.2023.130017
Ali Ramazani, Brett A. Duell, Eric J. Popczun, Sittichai Natesakhawat, Tarak Nandi, Jonathan W. Lekse, Yuhua Duan. High-throughput ab initio calculations and machine learning to discover SrFeO3--based perovskites for chemical-looping applications. Cell Reports Physical Science 2024, 5 (2) , 101797. https://doi.org/10.1016/j.xcrp.2024.101797
Rafiuzzaman Pritom, Rahul Jayan, Md Mahbubul Islam. Unraveling the effect of single atom catalysts on the charging behavior of nonaqueous Mg–CO 2 batteries: a combined density functional theory and machine learning approach. Journal of Materials Chemistry A 2024, 12 (4) , 2335-2348. https://doi.org/10.1039/D3TA06742D
Srinivas Rangarajan. Artificial intelligence in catalysis. 2024, 167-204. https://doi.org/10.1016/B978-0-323-99135-3.00002-6
Nikolay O. Grebennikov, Daniil A. Boiko, Darya O. Prima, Malena Madiyeva, Mikhail E. Minyaev, Valentine P. Ananikov. Boosting the generality of catalytic systems by the synergetic ligand effect in Pd-catalyzed C-N cross-coupling. Journal of Catalysis 2024, 429 , 115240. https://doi.org/10.1016/j.jcat.2023.115240
Sergei Manzhos, Tucker Carrington, Manabu Ihara. Orders of coupling representations as a versatile framework for machine learning from sparse data in high-dimensional spaces. Artificial Intelligence Chemistry 2023, 1 (2) , 100008. https://doi.org/10.1016/j.aichem.2023.100008
Bin Chang, Hong Pang, Fazal Raziq, Sibo Wang, Kuo-Wei Huang, Jinhua Ye, Huabin Zhang. Electrochemical reduction of carbon dioxide to multicarbon (C 2+ ) products: challenges and perspectives. Energy & Environmental Science 2023, 16 (11) , 4714-4758. https://doi.org/10.1039/D3EE00964E
Lingyan Kong, Xiongyi Liang, Maohuai Wang, Chi-Man Lawrence Wu. Role of transition metal d-orbitals in single-atom catalysts for nitric oxide electroreduction to ammonia. Journal of Colloid and Interface Science 2023, 647 , 375-383. https://doi.org/10.1016/j.jcis.2023.05.158
Lianping Wu, Teng Li. Machine learning enabled rational design of atomic catalysts for electrochemical reactions. Materials Chemistry Frontiers 2023, 7 (19) , 4445-4459. https://doi.org/10.1039/D3QM00661A
Hong Liu, Jiejie Li, Jordi Arbiol, Bo Yang, Pengyi Tang. Catalytic reactivity descriptors of metal‐nitrogen‐doped carbon catalysts for electrocatalysis. EcoEnergy 2023, 1 (1) , 154-185. https://doi.org/10.1002/ece2.12
Mengbo Yan, Shizhi Dong, Yanshuai Li, Zhiyu Liu, Hewei Zhao, Zhenwei Ma, Fuyang Geng, Zhiyong Li, Chun Wu. Accelerating the design and optimization of catalysts for the hydrogen evolution reaction in transition metal phosphides using machine learning. Molecular Catalysis 2023, 548 , 113402. https://doi.org/10.1016/j.mcat.2023.113402
Chenyang Wei, Dingyi Shi, Zhaohui Yang, Zhimin Xue, Shuzi Liu, Ruiqi Li, Tiancheng Mu. Data-driven design of double-atom catalysts with high H 2 evolution activity/CO 2 reduction selectivity based on simple features. Journal of Materials Chemistry A 2023, 11 (34) , 18168-18178. https://doi.org/10.1039/D3TA02332J
Li‐Hui Mou, TianTian Han, Pieter E. S. Smith, Edward Sharman, Jun Jiang. Machine Learning Descriptors for Data‐Driven Catalysis Study. Advanced Science 2023, 10 (22) https://doi.org/10.1002/advs.202301020
Zhuo Wang, Zhehao Sun, Hang Yin, Honghe Wei, Zicong Peng, Yoong Xin Pang, Guohua Jia, Haitao Zhao, Cheng Heng Pang, Zongyou Yin. The role of machine learning in carbon neutrality: Catalyst property prediction, design, and synthesis for carbon dioxide reduction. eScience 2023, 3 (4) , 100136. https://doi.org/10.1016/j.esci.2023.100136
A.K. Priya, Balaji Devarajan, Avinash Alagumalai, Hua Song. Artificial intelligence enabled carbon capture: A review. Science of The Total Environment 2023, 886 , 163913. https://doi.org/10.1016/j.scitotenv.2023.163913

Load all citations

Download PDF

Get e-Alerts

The Journal of Physical Chemistry C

Cite this: J. Phys. Chem. C 2020, 124, 41, 22471–22478

Click to copy citationCitation copied!

https://doi.org/10.1021/acs.jpcc.0c05964

Published September 16, 2020

Request reuse permissions

Article Views

Altmetric

Citations

164

Learn about these metrics

Citations are the number of other articles citing this article, calculated by Crossref and updated daily. Find more information about Crossref citation counts.

Abstract
High Resolution Image
Download MS PowerPoint Slide
Figure 1
Figure 1. Structure of SAC. Shallow blue, green, and gray balls represent transition metal (Ag, Au, Co, Cu, Fe, Ir, Mn, Mo, Ni, Os, Pd, Pt, Rh, Ru, Sc, Ti, V, Y, Zn or Zr), nonmetal (C, N, O, P, or S), and C atoms, respectively.
High Resolution Image
Download MS PowerPoint Slide
Figure 2
Figure 2. (a) Pearson correlation coefficient matrix heat maps of the best feature set for ΔG_CO prediction. (b) Feature importance rank for the best feature set. (c) Prediction results of XGBR model vs DFT data.
High Resolution Image
Download MS PowerPoint Slide
Figure 3
Figure 3. Prediction performance of other models trained by different ML methods: (a) KNR, (b) RFR, (c) TPOT, (d) SVR, and (e) GBR.
High Resolution Image
Download MS PowerPoint Slide
Figure 4
Figure 4. Structures M-C2S2 and M-CNO2 as the designed SACs. Gray, navy blue, red, yellow, and shallow blue balls represent C, N, O, S, and transition metal atoms, respectively.
High Resolution Image
Download MS PowerPoint Slide
Figure 5
Figure 5. Prediction heat map of ΔG_CO for designed single-atom catalysts.
High Resolution Image
Download MS PowerPoint Slide
Figure 6
Figure 6. Overlapped prediction heat maps of ΔG_CO and ΔG_H of designed SACs.
High Resolution Image
Download MS PowerPoint Slide
References
This article references 49 other publications.
1. 1
  Ma, X.; Li, Z.; Achenie, L. E.; Xin, H. Machine-Learning-Augmented Chemisorption Model for CO₂ Electroreduction Catalyst Screening. J. Phys. Chem. Lett. 2015, 6, 3528– 33, DOI: 10.1021/acs.jpclett.5b01660
  1
  Machine-Learning-Augmented Chemisorption Model for CO2 Electroreduction Catalyst Screening
  Ma, Xianfeng; Li, Zheng; Achenie, Luke E. K.; Xin, Hongliang
  Journal of Physical Chemistry Letters (2015), 6 (18), 3528-3533CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
  We present a machine-learning-augmented chemisorption model that enables fast and accurate prediction of the surface reactivity of metal alloys within a broad chem. space. Specifically, we show that artificial neural networks, a family of biol. inspired learning algorithms, trained with a set of ab initio adsorption energies and electronic fingerprints of idealized bimetallic surfaces, can capture complex, nonlinear interactions of adsorbates (e.g., *CO) on multimetallics with ∼0.1 eV error, outperforming the two-level interaction model in prediction. By leveraging scaling relations between adsorption energies of similar adsorbates, we illustrate that this integrated approach greatly facilitates high-throughput catalyst screening and, as a specific case, suggests promising {100}-terminated multimetallic alloys with improved efficiency and selectivity for CO2 electrochem. redn. to C2 species. Statistical anal. of the network response to perturbations of input features underpins our fundamental understanding of chem. bonding on metal surfaces.
2. 2
  Zhao, Z.; Lu, G. Computational Screening of Near-Surface Alloys for CO₂ Electroreduction. ACS Catal. 2018, 8, 3885– 3894, DOI: 10.1021/acscatal.7b03705
  2
  Computational Screening of Near-Surface Alloys for CO2 Electroreduction
  Zhao, Zhonglong; Lu, Gang
  ACS Catalysis (2018), 8 (5), 3885-3894CODEN: ACCACS; ISSN:2155-5435. (American Chemical Society)
  Electrochem. conversion of carbon dioxide (CO2) into chem. feedstocks provides an attractive soln. to our pressing energy and environment problems. Here, we report that transition metal near-surface alloys (NSAs) are promising catalysts for CO2 electroredn. Based on first-principles calcns. on 190 candidates, we propose a no. of NSAs which show promise of highly active and selective catalysts for formic acid, carbon monoxide, methanol, and ethylene prodn., while simultaneously suppress competing hydrogen evolution reaction (HER). We predict that Pd/W, Au/Hf, and Au/Zr NSAs are more active than most known electrodes for formic acid formation with overpotentials significantly lower than that of HER. Ag/Hf and Ag/Zr are revealed as superior catalysts for the prodn. of carbon monoxide with overpotentials of 0.77 V lower than that on pure Ag electrode. We find that methanol and ethylene can be produced on Ag/Ta and Ag/Nb NSAs whose overpotentials are ∼15% lower than that on Cu (211) surface. On the other hand, their overpotentials for HER are six times more neg. than that on Cu (211). The work demonstrates the great potential of transition metal catalysts by modulating their near surface properties.
3. 3
  Gálvez-Vázquez, M. d. J.; Moreno-García, P.; Guo, H.; Hou, Y.; Dutta, A.; Waldvogel, S. R.; Broekmann, P. Leaded Bronze Alloy as a Catalyst for the Electroreduction of CO₂. ChemElectroChem 2019, 6, 2324– 2330, DOI: 10.1002/celc.201900537
  There is no corresponding record for this reference.
4. 4
  Zhang, Q.; Xu, W.; Xu, J.; Liu, Y.; Zhang, J. High performing and cost-effective metal/metal oxide/metal alloy catalysts/electrodes for low temperature CO₂ electroreduction. Catal. Today 2018, 318, 15– 22, DOI: 10.1016/j.cattod.2018.03.029
  4
  High performing and cost-effective metal/metal oxide/metal alloy catalysts/electrodes for low temperature CO2 electroreduction
  Zhang, Qi; Xu, Wutao; Xu, Jie; Liu, Yuyu; Zhang, Jiujun
  Catalysis Today (2018), 318 (), 15-22CODEN: CATTEA; ISSN:0920-5861. (Elsevier B.V.)
  The electrochem. redn. of CO2 (ERC) has been proved to be both tech. and economic feasible. ERC using renewable energies and abandon nuclear/hydroelec. energies can convert CO2 to low-carbon fuels for energy storage and reducing CO2 emission as well as promoting waste (gas) recycling/utilization. One of the challenges is to find high electrochem. active, stable, product-selective and cost-effective catalysts for speeding up the electrode kinetics of ERC. Based on these specific requirements for catalysts, quite few of them are practically promising. In this mini-review paper, we try to give an overview of some important researches and the recent progress in ERC catalysts, focusing on the electrochem. characteristics of Cu-, Sn- and Zn-based catalysts for the efficient ERC. The catalysts reviewed in this paper include single pure metal, metal oxide, metal alloys and metal complexes. In recent years, more and more catalysts with novel nanostructures have been reported, such as those with shell-core structure, which exhibit desirable electrocatalytic performance. Moreover, carbon materials, such as carbon nanotube (CNT) and reduced graphene oxide (rGO), have also been explored as desirable catalyst supports in lab-scale studies. In the most recent years, more attention is moved on the fine crystal structures of catalysts, which are found to be quite crit. for achieving desirable product selectivity.
5. 5
  Fu, H. Q.; Zhang, L.; Zheng, L. R.; Liu, P. F.; Zhao, H.; Yang, H. G. Enhanced CO₂ electroreduction performance over Cl-modified metal catalysts. J. Mater. Chem. A 2019, 7, 12420– 12425, DOI: 10.1039/C9TA02223F
  5
  Enhanced CO2 electroreduction performance over Cl-modified metal catalysts
  Fu, Huai Qin; Zhang, Le; Zheng, Li Rong; Liu, Peng Fei; Zhao, Huijun; Yang, Hua Gui
  Journal of Materials Chemistry A: Materials for Energy and Sustainability (2019), 7 (20), 12420-12425CODEN: JMCAET; ISSN:2050-7496. (Royal Society of Chemistry)
  Ag nanoparticles with surface Cl- modification (Ag-Cl NPs) by in situ electroredn. of AgCl exhibit a high selectivity for CO2-to-CO conversion. The obsd. C-Cl bond suggests that electrons can be effectively transferred from the Cl- ions to the unoccupied orbital of CO2, and then activate nonpolar CO2 mols. on Cl- sites.
6. 6
  Plana, D.; Flórez-Montaño, J.; Celorrio, V.; Pastor, E.; Fermín, D. J. Tuning CO2 electroreduction efficiency at Pd shells on Au nanocores. Chem. Commun. 2013, 49, 10962– 10964, DOI: 10.1039/c3cc46543h
  6
  Tuning CO2 electroreduction efficiency at Pd shells on Au nanocores
  Plana, Daniela; Florez-Montano, Jonathan; Celorrio, Veronica; Pastor, Elena; Fermin, David J.
  Chemical Communications (Cambridge, United Kingdom) (2013), 49 (93), 10962-10964CODEN: CHCOFS; ISSN:1359-7345. (Royal Society of Chemistry)
  The faradaic efficiency of CO2 electroredn. is significantly affected by the thickness of Pd nanoshells on Au cores. The ratio of hydrogen evolution to CO2 redn. was detd. by differential electrochem. mass spectrometry. Decreasing the Pd shell thickness from 10 to 1 nm leads to a twofold increase in faradaic efficiency.
7. 7
  Cao, L.; Raciti, D.; Li, C.; Livi, K. J. T.; Rottmann, P. F.; Hemker, K. J.; Mueller, T.; Wang, C. Mechanistic Insights for Low-Overpotential Electroreduction of CO₂ to CO on Copper Nanowires. ACS Catal. 2017, 7, 8578– 8587, DOI: 10.1021/acscatal.7b03107
  7
  Mechanistic Insights for Low-Overpotential Electroreduction of CO2 to CO on Copper Nanowires
  Cao, Liang; Raciti, David; Li, Chenyang; Livi, Kenneth J. T.; Rottmann, Paul F.; Hemker, Kevin J.; Mueller, Tim; Wang, Chao
  ACS Catalysis (2017), 7 (12), 8578-8587CODEN: ACCACS; ISSN:2155-5435. (American Chemical Society)
  Recent developments of Cu-based nanomaterials have enabled the electroredn. of CO2 at low overpotentials. The mechanism of low-overpotential CO2 redn. on these nanocatalysts, however, largely remains elusive. The authors report here a systematic study of CO2 redn. on highly dense Cu nanowires, with the focus placed on understanding the surface structure effects on the formation of *CO (* denotes an adsorption site on the catalyst surface) and the evolution of gas-phase CO product (CO(g)) at low overpotentials (more pos. than -0.5 V). Cu nanowires of distinct nanocryst. and surface structures were studied comparatively to build up the structure-property relations, which are further interpreted by performing d. functional theory (DFT) calcns. of the reaction pathway on the various facets of Cu. A kinetic model reveals competition between CO(g) evolution and *CO poisoning depending on the electrode potential and surface structures. Open and metastable facets such as (110) and reconstructed (110) are likely the active sites for the electroredn. of CO2 to CO at the low overpotentials.
8. 8
  Baturina, O. A.; Lu, Q.; Padilla, M. A.; Xin, L.; Li, W.; Serov, A.; Artyushkova, K.; Atanassov, P.; Xu, F.; Epshteyn, A. CO₂ Electroreduction to Hydrocarbons on Carbon-Supported Cu Nanoparticles. ACS Catal. 2014, 4, 3682– 3695, DOI: 10.1021/cs500537y
  8
  CO2 Electroreduction to Hydrocarbons on Carbon-Supported Cu Nanoparticles
  Baturina, Olga A.; Lu, Qin; Padilla, Monica A.; Xin, Le; Li, Wenzhen; Serov, Alexey; Artyushkova, Kateryna; Atanassov, Plamen; Xu, Feng; Epshteyn, Albert; Brintlinger, Todd; Schuette, Mike; Collins, Greg E.
  ACS Catalysis (2014), 4 (10), 3682-3695CODEN: ACCACS; ISSN:2155-5435. (American Chemical Society)
  Activities of Cu nanoparticles supported on C black (VC), single-wall C nanotubes (SWNTs), and Ketjen Black (KB) toward CO2 electroredn. to hydrocarbons (CH4, C2H2, C2H4, and C2H6) are evaluated using a sealed rotating disk electrode (RDE) setup coupled to a gas chromatograph (GC). Thin films of supported Cu catalysts are deposited on RDE tips following a procedure well-established in the fuel cell community. Lead (Pb) underpotential deposition (UPD) was used to det. the electrochem. surface area (ECSA) of thin films of 40% Cu/VC, 20% Cu/SWNT, 50% Cu/KB, and com. 20% Cu/VC catalysts on glassy C electrodes. Faradaic efficiencies of four C-supported Cu catalysts toward CO2 electroredn. to hydrocarbons are compared to that of electrodeposited smooth Cu films. For all the catalysts studied, the only hydrocarbons detected by GC are CH4 and C2H4. The Cu nanoparticles are more active toward C2H4 generation vs. electrodeposited smooth Cu films. For the supported Cu nanocatalysts, the ratio of C2H4/CH4 faradaic efficiencies is believed to be a function of particle size, as higher ratios are obsd. for smaller Cu nanoparticles. This is likely due to an increase in the fraction of under-coordinated sites, such as corners, edges, and defects, as the nanoparticles become smaller.
9. 9
  Song, Y.; Chen, W.; Zhao, C.; Li, S.; Wei, W.; Sun, Y. Metal-Free Nitrogen-Doped Mesoporous Carbon for Electroreduction of CO₂ to Ethanol. Angew. Chem., Int. Ed. 2017, 56, 10840– 10844, DOI: 10.1002/anie.201706777
  9
  Metal-Free Nitrogen-Doped Mesoporous Carbon for Electroreduction of CO2 to Ethanol
  Song, Yanfang; Chen, Wei; Zhao, Chengcheng; Li, Shenggang; Wei, Wei; Sun, Yuhan
  Angewandte Chemie, International Edition (2017), 56 (36), 10840-10844CODEN: ACIEF5; ISSN:1433-7851. (Wiley-VCH Verlag GmbH & Co. KGaA)
  CO2 electroredn. is a promising technique for satisfying both renewable energy storage and a neg. C cycle. However, it remains a challenge to convert CO2 into C2 products with high efficiency and selectivity. Herein, the authors report a N-doped ordered cylindrical mesoporous C as a robust metal-free catalyst for CO2 electroredn., enabling the efficient prodn. of EtOH with nearly 100% selectivity and high faradaic efficiency of 77% at -0.56 V vs. the reversible H electrode. Expts. and d. functional theory calcns. demonstrate that the synergetic effect of the N hetero-atoms and the cylindrical channel configurations facilitate the dimerization of key CO* intermediates and the subsequent p-electron transfers, resulting in superior electrocatalytic performance for synthesizing EtOH from CO2.
10. 10
  Zhang, T.; Lin, L.; Li, Z.; He, X.; Xiao, S.; Shanov, V. N.; Wu, J. Nickel-Nitrogen-Carbon Molecular Catalysts for High Rate CO₂ Electro-reduction to CO: On the Role of Carbon Substrate and Reaction Chemistry. ACS Appl. Energy Mater. 2020, 3, 1617– 1626, DOI: 10.1021/acsaem.9b02112
  10
  Nickel-Nitrogen-Carbon Molecular Catalysts for High Rate CO2 Electro-reduction to CO: On the Role of Carbon Substrate and Reaction Chemistry
  Zhang, Tianyu; Lin, Lili; Li, Zhengyuan; He, Xingyu; Xiao, Shengdong; Shanov, Vesselin N.; Wu, Jingjie
  ACS Applied Energy Materials (2020), 3 (2), 1617-1626CODEN: AAEMCQ; ISSN:2574-0962. (American Chemical Society)
  Metal-N-C (M-N-C) mol. catalysts with NiN4 active structure were extensively studied as selective and active catalysts toward electrochem. redn. of CO2 to CO. The key challenge for a practical M-N-C catalyst is to increase the d. of at. metal active sites that achieves the partial c.d. of CO (jCO) relevant to the industrial level at lower overpotentials. Here, the authors revealed the effect of phys. and chem. properties of C substrates and synthetic processes on the tuning of the d. of at. metal active sites as well as the role of reaction chem. in enhancing the jCO and reducing the overpotential. The achievable loading of NiN4 active site in the Ni-N-C is detd. by the combined content of pyridinic and pyrrolic N functionalities and Ni-N coordination efficiency derived from the pyrolytic step rather than the uptake capability of Ni2+ in the adsorption step in the case of C black with high sp. surface area (>1000 m2/g). The N dopant content can be improved by modifying O functional groups on the surface of C black, optimizing the pyrolytic temp., and iterating the doping step. Through a combination of all optimum factors, the resultant Ni-N-C catalyst has a max. loading of ∼4.4% for at. Ni. This Ni-N-C catalyst exhibited faradaic efficiency (FE) of CO of 97% and jCO of -152 mA cm-2 at -0.93 V vs. RHE in a flow cell using 0.5M KHCO3 electrolyte while showing 93% FE of CO and jCO of -67 mA cm-2 at -0.61 V vs. RHE at 1 M KOH. Adding KI to the base electrolyte significantly magnified the jCO to larger than -200 mA cm-2 at a potential of -0.51 V vs. RHE while maintaining the almost unity FE of CO. The Ni-N-C is compatible with the membrane-electrode-assembly-based electrolyzer in which the jCO also achieved >200 mA cm-2 at a cell voltage of ∼2.7 V.
11. 11
  Li, Z.; Ji, S.; Liu, Y.; Cao, X.; Tian, S.; Chen, Y.; Niu, Z.; Li, Y. Well-Defined Materials for Heterogeneous Catalysis: From Nanoparticles to Isolated Single-Atom Sites. Chem. Rev. 2020, 120, 623– 682, DOI: 10.1021/acs.chemrev.9b00311
  11
  Well-Defined Materials for Heterogeneous Catalysis: From Nanoparticles to Isolated Single-Atom Sites
  Li, Zhi; Ji, Shufang; Liu, Yiwei; Cao, Xing; Tian, Shubo; Chen, Yuanjun; Niu, Zhiqiang; Li, Yadong
  Chemical Reviews (Washington, DC, United States) (2020), 120 (2), 623-682CODEN: CHREAY; ISSN:0009-2665. (American Chemical Society)
  A review. The use of well-defined materials in heterogeneous catalysis will open up numerous new opportunities for the development of advanced catalysts to address the global challenges in energy and the environment. This review surveys the roles of nanoparticles and isolated single atom sites in catalytic reactions. In the second section, the effects of size, shape, and metal-support interactions are discussed for nanostructured catalysts. Case studies are summarized to illustrate the dynamics of structure evolution of well-defined nanoparticles under certain reaction conditions. In the third section, we review the syntheses and catalytic applications of isolated single at. sites anchored on different types of supports. In the final part, we conclude by highlighting the challenges and opportunities of well-defined materials for catalyst development and gaining a fundamental understanding of their active sites.
12. 12
  Pan, Y.; Lin, R.; Chen, Y.; Liu, S.; Zhu, W.; Cao, X.; Chen, W.; Wu, K.; Cheong, W. C.; Wang, Y. Design of Single-Atom Co-N₅ Catalytic Site: A Robust Electrocatalyst for CO₂ Reduction with Nearly 100% CO Selectivity and Remarkable Stability. J. Am. Chem. Soc. 2018, 140, 4218– 4221, DOI: 10.1021/jacs.8b00814
  12
  Design of Single-Atom Co-N5 Catalytic Site: A Robust Electrocatalyst for CO2 Reduction with Nearly 100% CO Selectivity and Remarkable Stability
  Pan, Yuan; Lin, Rui; Chen, Yinjuan; Liu, Shoujie; Zhu, Wei; Cao, Xing; Chen, Wenxing; Wu, Konglin; Cheong, Weng-Chon; Wang, Yu; Zheng, Lirong; Luo, Jun; Lin, Yan; Liu, Yunqi; Liu, Chenguang; Li, Jun; Lu, Qi; Chen, Xin; Wang, Dingsheng; Peng, Qing; Chen, Chen; Li, Yadong
  Journal of the American Chemical Society (2018), 140 (12), 4218-4221CODEN: JACSAT; ISSN:0002-7863. (American Chemical Society)
  We develop an N-coordination strategy to design a robust CO2 redn. reaction (CO2RR) electrocatalyst with atomically dispersed Co-N5 site anchored on polymer-derived hollow N-doped porous carbon spheres. Our catalyst exhibits high selectivity for CO2RR with CO Faradaic efficiency (FECO) above 90% over a wide potential range from -0.57 to -0.88 V (the FECO exceeded 99% at -0.73 and -0.79 V). The CO c.d. and FECO remained nearly unchanged after electrolyzing 10 h, revealing remarkable stability. Expts. and d. functional theory calcns. demonstrate single-atom Co-N5 site is the dominating active center simultaneously for CO2 activation, the rapid formation of key intermediate COOH* as well as the desorption of CO.
13. 13
  Liu, X.; Wang, Z.; Tian, Y.; Zhao, J. Graphdiyne-Supported Single Iron Atom: A Promising Electrocatalyst for Carbon Dioxide Electroreduction into Methane and Ethanol. J. Phys. Chem. C 2020, 124, 3722– 3730, DOI: 10.1021/acs.jpcc.9b11649
  13
  Graphdiyne-Supported Single Iron Atom: A Promising Electrocatalyst for Carbon Dioxide Electroreduction into Methane and Ethanol
  Liu, Xin; Wang, Zhongxu; Tian, Yu; Zhao, Jingxiang
  Journal of Physical Chemistry C (2020), 124 (6), 3722-3730CODEN: JPCCCK; ISSN:1932-7447. (American Chemical Society)
  Electrochem. redn. of CO2 (CO2ER) to high-energy-d. multicarbon products is a quite promising technique for large-scale renewable energy storage, for which searching for stable, inexpensive, and efficient catalysts is a key scientific issue. The potential of an exptl. available single Fe atom supported on graphdiyne (Fe/GDY) as the CO2ER catalyst was explored by d. functional theory (DFT) computations. The authors' results revealed that Fe/GDY exhibits high stability due to the strong hybridization between the Fe 3d orbitals and the C 2p orbitals of GDY. Due to the small limiting potential of -0.43 V, the anchored Fe atom can effectively reduce CO2 to CH4 along the following pathway: CO2 → HCOO* → HCOOH* → HCO* → H2CO* → H3CO* → O* + CH4 → OH* → H2O, in which the hydrogenation of HCOOH* to HCO* is the potential-detg. step. Also, the unsatd. HCO* species on Fe/GDY can provide an active site for further coupling with CO to generate EtOH with a small activation energy for C-C coupling. The authors' theor. results not only propose a new approach to CO2ER to C2 products on a single-site catalyst but also further widen the potential applications of GDY.
14. 14
  Chen, A.; Zhang, X.; Zhou, Z. Machine learning: Accelerating materials development for energy storage and conversion. InfoMat 2020, 2, 553– 576, DOI: 10.1002/inf2.12094
  14
  Machine learning: Accelerating materials development for energy storage and conversion
  Chen, An; Zhang, Xu; Zhou, Zhen
  InfoMat (2020), 2 (3), 553-576CODEN: INFOHH; ISSN:2567-3165. (John Wiley & Sons Australia, Ltd.)
  With the development of modern society, the requirement for energy has become increasingly important on a global scale. Therefore, the exploration of novel materials for renewable energy technologies is urgently needed. Traditional methods are difficult to meet the requirements for materials science due to long exptl. period and high cost. Nowadays, machine learning (ML) is rising as a new research paradigm to revolutionize materials discovery. In this review, we briefly introduce the basic procedure of ML and common algorithms in materials science, and particularly focus on latest progress in applying ML to property prediction and materials development for energy-related fields, including catalysis, batteries, solar cells, and gas capture. Moreover, contributions of ML to expts. are involved as well. We highly expect that this review could lead the way forward in the future development of ML in materials science.
15. 15
  Frey, N. C.; Wang, J.; Vega Bellido, G. I.; Anasori, B.; Gogotsi, Y.; Shenoy, V. B. Prediction of Synthesis of 2D Metal Carbides and Nitrides (MXenes) and Their Precursors with Positive and Unlabeled Machine Learning. ACS Nano 2019, 13, 3031– 3041, DOI: 10.1021/acsnano.8b08014
  15
  Prediction of Synthesis of 2D Metal Carbides and Nitrides (MXenes) and Their Precursors with Positive and Unlabeled Machine Learning
  Frey, Nathan C.; Wang, Jin; Vega Bellido, Gabriel Ivan; Anasori, Babak; Gogotsi, Yury; Shenoy, Vivek B.
  ACS Nano (2019), 13 (3), 3031-3041CODEN: ANCAC3; ISSN:1936-0851. (American Chemical Society)
  Growing interest in the potential applications of two-dimensional (2D) materials has fueled advancement in the identification of 2D systems with exotic properties. Increasingly, the bottleneck in this field is the synthesis of these materials. Although theor. calcns. have predicted a myriad of promising 2D materials, only a few dozen were exptl. realized since the initial discovery of graphene. Here, we adapt the state-of-the-art pos. and unlabeled (PU) machine learning framework to predict which theor. proposed 2D materials have the highest likelihood of being successfully synthesized. Using elemental information and data from high-throughput d. functional theory calcns., we apply the PU learning method to the MXene family of 2D transition metal carbides, carbonitrides, and nitrides, and their layered precursor MAX phases, and identify 18 MXene compds. that are highly promising candidates for synthesis. By considering both the MXenes and their precursors, we further propose 20 synthesizable MAX phases that can be chem. exfoliated to produce MXenes.
16. 16
  Lu, S.; Zhou, Q.; Ouyang, Y.; Guo, Y.; Li, Q.; Wang, J. Accelerated discovery of stable lead-free hybrid organic-inorganic perovskites via machine learning. Nat. Commun. 2018, 9, 3405, DOI: 10.1038/s41467-018-05761-w
  16
  Accelerated discovery of stable lead-free hybrid organic-inorganic perovskites via machine learning
  Lu Shuaihua; Zhou Qionghua; Ouyang Yixin; Guo Yilv; Li Qiang; Wang Jinlan
  Nature communications (2018), 9 (1), 3405 ISSN:.
  Rapidly discovering functional materials remains an open challenge because the traditional trial-and-error methods are usually inefficient especially when thousands of candidates are treated. Here, we develop a target-driven method to predict undiscovered hybrid organic-inorganic perovskites (HOIPs) for photovoltaics. This strategy, combining machine learning techniques and density functional theory calculations, aims to quickly screen the HOIPs based on bandgap and solve the problems of toxicity and poor environmental stability in HOIPs. Successfully, six orthorhombic lead-free HOIPs with proper bandgap for solar cells and room temperature thermal stability are screened out from 5158 unexplored HOIPs and two of them stand out with direct bandgaps in the visible region and excellent environmental stability. Essentially, a close structure-property relationship mapping the HOIPs bandgap is established. Our method can achieve high accuracy in a flash and be applicable to a broad class of functional material design.
17. 17
  Im, J.; Lee, S.; Ko, T.-W.; Kim, H. W.; Hyon, Y.; Chang, H. Identifying Pb-free perovskites for solar cells by machine learning. npj Comput. Mater. 2019, 5, 37, DOI: 10.1038/s41524-019-0177-0
  There is no corresponding record for this reference.
18. 18
  Ali, A.; Park, H.; Mall, R.; Aïssa, B.; Sanvito, S.; Bensmail, H.; Belaidi, A.; El-Mellouhi, F. Machine Learning Accelerated Recovery of the Cubic Structure in Mixed-Cation Perovskite Thin Films. Chem. Mater. 2020, 32, 2998– 3006, DOI: 10.1021/acs.chemmater.9b05342
  18
  Machine Learning Accelerated Recovery of the Cubic Structure in Mixed-Cation Perovskite Thin Films
  Ali, Adnan; Park, Heesoo; Mall, Raghvendra; Aissa, Brahim; Sanvito, Stefano; Bensmail, Halima; Belaidi, Abdelhak; El-Mellouhi, Fedwa
  Chemistry of Materials (2020), 32 (7), 2998-3006CODEN: CMATEX; ISSN:0897-4756. (American Chemical Society)
  Data-driven approaches for materials design and selection have accelerated materials discovery along with the upsurge of machine learning applications. We report here a prediction-to-lab.-scale synthesis of cubic phase triple-cation lead halide perovskites guided by a machine learning perovskite stability predictor. The starting double-cation perovskite resulting from the incorporation of 15% dimethylammonium (DMA) in methylammonium lead triiodide suffers from significant deviation from the perovskite structure. By analyzing the X-ray diffraction and SEM, we confirmed that it is possible to recover the perovskite structure with the cubic phase at room temp. (RT) while minimizing the iterations of trial-and-error by adding <10 mol % of cesium cation additives, as guided by the machine learning predictor. Our conclusions highly support the cubic-phase stabilization at RT by controlling the stoichiometric ratio of various sized cations. This prediction-to-lab.-scale synthesis approach also enables us to identify room for improvements of the current machine learning predictor to take into consideration the cubic phase stability as well as phase segregation.
19. 19
  Fernandez, M.; Boyd, P. G.; Daff, T. D.; Aghaji, M. Z.; Woo, T. K. Rapid and Accurate Machine Learning Recognition of High Performing Metal Organic Frameworks for CO₂ Capture. J. Phys. Chem. Lett. 2014, 5, 3056– 3060, DOI: 10.1021/jz501331m
  19
  Rapid and Accurate Machine Learning Recognition of High Performing Metal Organic Frameworks for CO2 Capture
  Fernandez, Michael; Boyd, Peter G.; Daff, Thomas D.; Aghaji, Mohammad Zein; Woo, Tom K.
  Journal of Physical Chemistry Letters (2014), 5 (17), 3056-3060CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
  Quant. structure-property relationship (QSPR) models were developed using advanced machine learning algorithms that can rapidly and accurately recognize high-performing metal org. framework (MOF) materials for CO2 capture. QSPR classifiers were developed that can, in a fraction of a section, identify candidate MOFs with enhanced CO2 adsorption capacity ( > 1 mmol/g at 0.15 bar and >4 mmol/g at 1 bar). The models were tested on a large set of 292 050 MOFs that were not part of the training set. The QSPR classifier could recover 945 of the top 1000 MOFs in the test set while flagging only 10% of the whole library for compute intensive screening. Thus, using the machine learning classifiers as part of a high-throughput screening protocol would result in an order of magnitude redn. in compute time and allow intractably large structure libraries and search spaces to be screened.
20. 20
  He, Y.; Cubuk, E. D.; Allendorf, M. D.; Reed, E. J. Metallic Metal-Organic Frameworks Predicted by the Combination of Machine Learning Methods and Ab Initio Calculations. J. Phys. Chem. Lett. 2018, 9, 4562– 4569, DOI: 10.1021/acs.jpclett.8b01707
  20
  Metallic Metal-Organic Frameworks Predicted by the Combination of Machine Learning Methods and Ab Initio Calculations
  He, Yuping; Cubuk, Ekin D.; Allendorf, Mark D.; Reed, Evan J.
  Journal of Physical Chemistry Letters (2018), 9 (16), 4562-4569CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
  Emerging applications of metal-org. frameworks (MOFs) in electronic devices will benefit from the design and synthesis of intrinsically, highly electronically conductive MOFs. However, very few are known to exist. It is a challenging task to search for electronically conductive MOFs within the tens of thousands of reported MOF structures. Using a new strategy (i.e., transfer learning) of combining machine learning techniques, statistical multivoting, and ab initio calcns., we screened 2932 MOFs and identified 6 MOF crystal structures that are metallic at the level of semilocal DFT band theory: Mn2[Re6X8(CN)6]4 (X = S, Se,Te), Mn[Re3Te4(CN)3], Hg[SCN]4Co[NCS]4, and CdC4. Five of these structures have been synthesized and reported in the literature, but their elec. characterization has not been reported. Our work demonstrates the potential power of machine learning in materials science to aid in down-selecting from large nos. of potential candidates and provides the information and guidance to accelerate the discovery of novel advanced materials.
21. 21
  Wu, Y.; Duan, H.; Xi, H. Machine Learning-Driven Insights into Defects of Zirconium Metal-Organic Frameworks for Enhanced Ethane-Ethylene Separation. Chem. Mater. 2020, 32, 2986– 2997, DOI: 10.1021/acs.chemmater.9b05322
  21
  Machine Learning-Driven Insights into Defects of Zirconium Metal-Organic Frameworks for Enhanced Ethane-Ethylene Separation
  Wu, Ying; Duan, Haipeng; Xi, Hongxia
  Chemistry of Materials (2020), 32 (7), 2986-2997CODEN: CMATEX; ISSN:0897-4756. (American Chemical Society)
  Structural defects in metal-org. frameworks (MOFs) have the potential to yield desirable properties that could not be achieved by "defect-free" crystals, but previous works in this area have focused on limited versions of defects due to the difficulty of detecting defects in MOFs. In this work, a modeling library contg. 425 defective UiO-66 (UiO-66-Ds) with a comprehensive population (in terms of concn. and distribution) of missing-linker defects was created. Taking ethane-ethylene sepn. as a case study, we demonstrated that machine learning could provide data-driven insight into how the defects control the performance of UiO-66-Ds in adsorption, sepn., and mech. stability. We found that the missing-linker ratio in real materials could be predicted from the gravimetric surface area and pore vol., making it a useful complement for the challenges of directly measuring the defect concn. We further identified the "privileged" UiO-66-Ds that were optimal in overall properties and provided decision trees as guidance to access and design these top performers. This work offers a general strategy for fully exploring the defects in MOFs, providing long-term opportunities for the development of defect engineering in the adsorption community.
22. 22
  Zhu, X.; Yan, J.; Gu, M.; Liu, T.; Dai, Y.; Gu, Y.; Li, Y. Activity origin and design principles for oxygen reduction on dual-metal-site catalysts: A combined density functional theory and machine learning study. J. Phys. Chem. Lett. 2019, 10, 7760– 7766, DOI: 10.1021/acs.jpclett.9b03392
  22
  Activity Origin and Design Principles for Oxygen Reduction on Dual-Metal-Site Catalysts: A Combined Density Functional Theory and Machine Learning Study
  Zhu, Xiaorong; Yan, Jiaxian; Gu, Min; Liu, Tianyang; Dai, Yafei; Gu, Yanhui; Li, Yafei
  Journal of Physical Chemistry Letters (2019), 10 (24), 7760-7766CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
  Dual-metal-site catalysts (DMSCs) are emerging as a new frontier in the field of oxygen redn. reaction (ORR). However, there is a lack of design principles to provide a universal description of the relationship between intrinsic properties of DMSCs and the catalytic activity. Here, we identify the origin of ORR activity and unveil design principles for graphene-based DMSCs by means of d. functional theory computations and machine learning (ML). Our results indicate that several exptl. unexplored DMSCs can show outstanding ORR activity surpassing that of platinum. Remarkably, our ML study reveals that the ORR activity of DMSCs is intrinsically governed by some fundamental factors, such as electron affinity, electronegativity, and radii of the embedded metal atoms. More importantly, we propose predictor equations with acceptable accuracy to quant. describe the ORR activity of DMSCs. Our work will accelerate the search for highly active DMSCs for ORR and other electrochem. reactions.
23. 23
  Zafari, M.; Kumar, D.; Umer, M.; Kim, K. S. Machine learning-based high throughput screening for nitrogen fixation on boron-doped single atom catalysts. J. Mater. Chem. A 2020, 8, 5209– 5216, DOI: 10.1039/C9TA12608B
  23
  Machine learning-based high throughput screening for nitrogen fixation on boron-doped single atom catalysts
  Zafari, Mohammad; Kumar, Deepak; Umer, Muhammad; Kim, Kwang S.
  Journal of Materials Chemistry A: Materials for Energy and Sustainability (2020), 8 (10), 5209-5216CODEN: JMCAET; ISSN:2050-7496. (Royal Society of Chemistry)
  Prodn. of ammonia via electrochem. nitrogen redn. reaction (NRR) has recently attracted much attention due to its potential to play a vital role in producing fertilizers and other chems. High throughput screening of electrocatalysts for the NRR requires numerous calcns. in the search space, making the computational cost a bottleneck for predicting eligible electrocatalysts. Here we used a deep neural network (DNN) to predict efficient electrocatalysts for the NRR among boron(B)-doped graphene single atom catalysts (SACs). This model can noticeably reduce the time of computation by removing non-efficient catalysts from screening. Also, the adsorption energy and free energy can be predicted by the feature-based light gradient boosting machine (LGBM) model. These features represent the geometrical structure as well as bonding characteristics. Among the catalysts evaluated, three candidates were identified as very promising catalysts, offering excellent selectivity over the hydrogen evolution reaction (HER). CrB3C1 exhibited a minimal overpotential of 0.13 V for the NRR. This study provides a new pathway for the rational design of catalysts for nitrogen fixation by employing the most important features involved in the NRR by using machine learning methods.
24. 24
  Choukroun, D.; Daems, N.; Kenis, T.; Van Everbroeck, T.; Hereijgers, J.; Altantzis, T.; Bals, S.; Cool, P.; Breugelmans, T. Bifunctional Nickel-Nitrogen-Doped-Carbon-Supported Copper Electrocatalyst for CO₂ Reduction. J. Phys. Chem. C 2020, 124, 1369– 1381, DOI: 10.1021/acs.jpcc.9b08931
  24
  Bifunctional Nickel-Nitrogen-Doped-Carbon-Supported Copper Electrocatalyst for CO2 Reduction
  Choukroun, Daniel; Daems, Nick; Kenis, Thomas; Van Everbroeck, Tim; Hereijgers, Jonas; Altantzis, Thomas; Bals, Sara; Cool, Pegie; Breugelmans, Tom
  Journal of Physical Chemistry C (2020), 124 (2), 1369-1381CODEN: JPCCCK; ISSN:1932-7447. (American Chemical Society)
  Bifunctionality is a key feature of many industrial catalysts, supported metal clusters and particles in particular, and the development of such catalysts for the CO2 redn. reaction (CO2RR) to hydrocarbons and alcs. is gaining traction in light of recent advancements in the field. C-supported Cu nanoparticles are suitable candidates for integration in the state-of-the-art reaction interfaces, and here, the authors propose, synthesize, and evaluate a bifunctional Ni-N-doped-C-supported Cu electrocatalyst, in which the support possesses active sites for selective CO2 conversion to CO and Cu nanoparticles catalyze either the direct CO2 or CO redn. to hydrocarbons. The authors introduce the scientific rationale behind the concept, its applicability, and the challenges with regard to the catalyst. From the practical aspect, the deposition of Cu nanoparticles onto C black and Ni-N-C supports via an NH3-driven deposition pptn. method is reported and explored in more detail using x-ray diffraction, TGA, and H temp.-programmed redn. High-angle annular dark-field scanning TEM (HAADF-STEM) and energy-dispersive x-ray spectroscopy (EDXS) give further evidence of the presence of Cu-contg. nanoparticles on the Ni-N-C supports while revealing an addnl. relation between the nanoparticle's compn. and the electrode's electrocatalytic performance. Compared to the benchmark C black-supported Cu catalysts, Ni-N-C-supported Cu delivers up to a 2-fold increase in the partial C2H4 c.d. at -1.05 VRHE (C1/C2 = 0.67) and a concomitant 10-fold increase of the CO partial c.d. The enhanced ethylene prodn. metrics, obtained by virtue of the higher intrinsic activity of the Ni-N-C support, point out toward a synergistic action between the two catalytic functionalities.
25. 25
  Varley, J. B.; Hansen, H. A.; Ammitzbøll, N. L.; Grabow, L. C.; Peterson, A. A.; Rossmeisl, J.; Nørskov, J. K. Ni-Fe-S Cubanes in CO₂ Reduction Electrocatalysis: A DFT Study. ACS Catal. 2013, 3, 2640– 2643, DOI: 10.1021/cs4005419
  25
  Ni-Fe-S Cubanes in CO2 Reduction Electrocatalysis: A DFT Study
  Varley, J. B.; Hansen, H. A.; Ammitzboell, N. L.; Grabow, L. C.; Peterson, A. A.; Rossmeisl, J.; Noerskov, J. K.
  ACS Catalysis (2013), 3 (11), 2640-2643CODEN: ACCACS; ISSN:2155-5435. (American Chemical Society)
  The authors perform extensive mechanistic studies of CO2 (electro)-redn. by analogs to the active sites of CO dehydrogenase (CODH) enzymes. The authors explore structure-property relations for different cluster compns. and interpret the results with a model for CO2 electroredn. the authors recently developed and applied to transition metal catalysts. The authors' results validate the effectiveness of the CODH in catalyzing this important reaction and give insight into why specific cluster compns. were adopted by nature.
26. 26
  Back, S.; Lim, J.; Kim, N. Y.; Kim, Y. H.; Jung, Y. Single-atom catalysts for CO₂ electroreduction with significant activity and selectivity improvements. Chem. Sci. 2017, 8, 1090– 1096, DOI: 10.1039/C6SC03911A
  26
  Single-atom catalysts for CO2 electroreduction with significant activity and selectivity improvements
  Back, Seoin; Lim, Juhyung; Kim, Na-Young; Kim, Yong-Hyun; Jung, Yousung
  Chemical Science (2017), 8 (2), 1090-1096CODEN: CSHCCN; ISSN:2041-6520. (Royal Society of Chemistry)
  A single-atom catalyst (SAC) has an electronic structure that is very different from its bulk counterparts, and has shown an unexpectedly high specific activity with a significant redn. in noble metal usage for CO oxidn., fuel cell and hydrogen evolution applications, although phys. origins of such performance enhancements are still poorly understood. Herein, by means of d. functional theory (DFT) calcns., we for the first time investigate the great potential of single atom catalysts for CO2 electroredn. applications. In particular, we study a single transition metal atom anchored on defective graphene with single or double vacancies, denoted M@sv-Gr or M@dv-Gr, where M = Ag, Au, Co, Cu, Fe, Ir, Ni, Os, Pd, Pt, Rh or Ru, as a CO2 redn. catalyst. Many SACs are indeed shown to be highly selective for the CO2 redn. reaction over a competitive H2 evolution reaction due to favorable adsorption of carboxyl (*COOH) or formate (*OCHO) over hydrogen (*H) on the catalysts. On the basis of free energy profiles, we identified several promising candidate materials for different products; Ni@dv-Gr (limiting potential UL = -0.41 V) and Pt@dv-Gr (-0.27 V) for CH3OH prodn., and Os@dv-Gr (-0.52 V) and Ru@dv-Gr (-0.52 V) for CH4 prodn. In particular, the Pt@dv-Gr catalyst shows remarkable redn. in the limiting potential for CH3OH prodn. compared to any existing catalysts, synthesized or predicted. To understand the origin of the activity enhancement of SACs, we find that the lack of an at. ensemble for adsorbate binding and the unique electronic structure of the single atom catalysts as well as orbital interaction play an important role, contributing to binding energies of SACs that deviate considerably from the conventional scaling relation of bulk transition metals.
27. 27
  Kresse, G.; Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169– 11186, DOI: 10.1103/PhysRevB.54.11169
  27
  Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set
  Kresse, G.; Furthmueller, J.
  Physical Review B: Condensed Matter (1996), 54 (16), 11169-11186CODEN: PRBMDO; ISSN:0163-1829. (American Physical Society)
  The authors present an efficient scheme for calcg. the Kohn-Sham ground state of metallic systems using pseudopotentials and a plane-wave basis set. In the first part the application of Pulay's DIIS method (direct inversion in the iterative subspace) to the iterative diagonalization of large matrixes will be discussed. This approach is stable, reliable, and minimizes the no. of order Natoms3 operations. In the second part, we will discuss an efficient mixing scheme also based on Pulay's scheme. A special "metric" and a special "preconditioning" optimized for a plane-wave basis set will be introduced. Scaling of the method will be discussed in detail for non-self-consistent and self-consistent calcns. It will be shown that the no. of iterations required to obtain a specific precision is almost independent of the system size. Altogether an order Natoms2 scaling is found for systems contg. up to 1000 electrons. If we take into account that the no. of k points can be decreased linearly with the system size, the overall scaling can approach Natoms. They have implemented these algorithms within a powerful package called VASP (Vienna ab initio simulation package). The program and the techniques have been used successfully for a large no. of different systems (liq. and amorphous semiconductors, liq. simple and transition metals, metallic and semiconducting surfaces, phonons in simple metals, transition metals, and semiconductors) and turned out to be very reliable.
28. 28
  Kresse, G.; Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 1758– 1775, DOI: 10.1103/PhysRevB.59.1758
  28
  From ultrasoft pseudopotentials to the projector augmented-wave method
  Kresse, G.; Joubert, D.
  Physical Review B: Condensed Matter and Materials Physics (1999), 59 (3), 1758-1775CODEN: PRBMDO; ISSN:0163-1829. (American Physical Society)
  The formal relationship between ultrasoft (US) Vanderbilt-type pseudopotentials and Blochl's projector augmented wave (PAW) method is derived. The total energy functional for US pseudopotentials can be obtained by linearization of two terms in a slightly modified PAW total energy functional. The Hamilton operator, the forces, and the stress tensor are derived for this modified PAW functional. A simple way to implement the PAW method in existing plane-wave codes supporting US pseudopotentials is pointed out. In addn., crit. tests are presented to compare the accuracy and efficiency of the PAW and the US pseudopotential method with relaxed-core all-electron methods. These tests include small mols. (H2, H2O, Li2, N2, F2, BF3, SiF4) and several bulk systems (diamond, Si, V, Li, Ca, CaF2, Fe, Co, Ni). Particular attention is paid to the bulk properties and magnetic energies of Fe, Co, and Ni.
29. 29
  Hammer, B.; Hansen, L. B.; Nørskov, J. K. Improved adsorption energetics within density-functional theory using revised Perdew-Burke-Ernzerhof functionals. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 7413– 7421, DOI: 10.1103/PhysRevB.59.7413
  29
  Improved adsorption energetics within density-functional theory using revised Perdew-Burke-Ernzerhof functionals
  Hammer, B.; Hansen, L. B.; Norskov, J. K.
  Physical Review B: Condensed Matter and Materials Physics (1999), 59 (11), 7413-7421CODEN: PRBMDO; ISSN:0163-1829. (American Physical Society)
  A simple formulation of a generalized gradient approxn. for the exchange and correlation energy of electrons has been proposed by J. Perdew et al. (1996). Subsequently, Y. Zhang and W. Wang (1998) have shown that a slight revision of the Perdew-Burke-Ernzerhof (PBE) functional systematically improves the atomization energies for a large database of small mols. In the present work, we show that the Zhang and Yang functional (revPBE) also improves the chemisorption energetics of atoms and mols. on transition-metal surfaces. Our test systems comprise at. and mol. adsorption of oxygen, CO, and NO on Ni(100), Ni(111), Rh(100), Pd(100), and Pd(111) surfaces. As the revPBE functional may locally violate the Lieb-Oxford criterion, we further develop an alternative revision of the PBE functional, RPBE, which gives the same improvement of the chemisorption energies as the revPBE functional at the same time as it fulfills the Lieb-Oxford criterion locally.
30. 30
  Monkhorst, H. J.; Pack, J. D. Special points for Brillouin-zone integrations. Phys. Rev. B 1976, 13, 5188– 5192, DOI: 10.1103/PhysRevB.13.5188
  There is no corresponding record for this reference.
31. 31
  Xu, H.; Cheng, D.; Cao, D.; Zeng, X. C. A universal principle for a rational design of single-atom electrocatalysts. Nat. Catal. 2018, 1, 339– 348, DOI: 10.1038/s41929-018-0063-z
  31
  A universal principle for a rational design of single-atom electrocatalysts
  Xu, Haoxiang; Cheng, Daojian; Cao, Dapeng; Zeng, Xiao Cheng
  Nature Catalysis (2018), 1 (5), 339-348CODEN: NCAACP; ISSN:2520-1158. (Nature Research)
  Developing highly active single-atom catalysts for electrochem. reactions is a key to future renewable energy technol. Here we present a universal design principle to evaluate the activity of graphene-based single-atom catalysts towards the oxygen redn., oxygen evolution and hydrogen evolution reactions. Our results indicate that the catalytic activity of single-atom catalysts is highly correlated with the local environment of the metal center, namely its coordination no. and electronegativity and the electronegativity of the nearest neighbor atoms, validated by available exptl. data. More importantly, we reveal that this design principle can be extended to metal-macrocycle complexes. The principle not only offers a strategy to design highly active nonprecious metal single-atom catalysts with specific active centers, for example, Fe-pyridine/pyrrole-N4 for the oxygen redn. reaction; Co-pyrrole-N4 for the oxygen evolution reaction; and Mn-pyrrole-N4 for the hydrogen evolution reaction to replace precious Pt/Ir/Ru-based catalysts, but also suggests that macrocyclic metal complexes could be used as an alternative to graphene-based single-atom catalysts.
32. 32
  Martyna, G. J.; Klein, M. L.; Tuckerman, M. Nosé-Hoover chains: The canonical ensemble via continuous dynamics. J. Chem. Phys. 1992, 97, 2635– 2643, DOI: 10.1063/1.463940
  There is no corresponding record for this reference.
33. 33
  Tran, K.; Ulissi, Z. W. Active learning across intermetallics to guide discovery of electrocatalysts for CO₂ reduction and H₂ evolution. Nat. Catal. 2018, 1, 696– 703, DOI: 10.1038/s41929-018-0142-1
  33
  Active learning across intermetallics to guide discovery of electrocatalysts for CO2 reduction and H2 evolution
  Tran, Kevin; Ulissi, Zachary W.
  Nature Catalysis (2018), 1 (9), 696-703CODEN: NCAACP; ISSN:2520-1158. (Nature Research)
  The electrochem. redn. of CO2 and H2 evolution from water can be used to store renewable energy that is produced intermittently. Scale-up of these reactions requires the discovery of effective electrocatalysts, but the electrocatalyst search space is too large to explore exhaustively. Here we present a theor., fully automated screening method that uses a combination of machine learning and optimization to guide d. functional theory calcns., which are then used to predict electrocatalyst performance. We demonstrate the feasibility of this method by screening various alloys of 31 different elements, and thereby perform a screening that encompasses 50% of the d-block elements and 33% of the p-block elements. This method has thus far identified 131 candidate surfaces across 54 alloys for CO2 redn. and 258 surfaces across 102 alloys for H2 evolution. We use qual. analyses to prioritize the top candidates for exptl. validation.
34. 34
  Studt, F.; Abild-Pedersen, F.; Varley, J. B.; Nørskov, J. K. CO and CO₂ hydrogenation to methanol calculated Using the BEEF-vdW Functional. Catal. Lett. 2013, 143, 71– 73, DOI: 10.1007/s10562-012-0947-5
  34
  CO and CO2 Hydrogenation to Methanol Calculated Using the BEEF-vdW Functional
  Studt, Felix; Abild-Pedersen, Frank; Varley, Joel B.; Norskov, Jens K.
  Catalysis Letters (2013), 143 (1), 71-73CODEN: CALEER; ISSN:1011-372X. (Springer)
  The hydrogenation of CO and CO2 to methanol on a stepped copper surface was studied using the BEEF-vdW functional and is compared to values derived with RPBE. The inclusion of vdW forces in the BEEF-vdW functional yields a better description of CO2 hydrogenation as compared to RPBE. These differences are significant for a qual. description of the overall methanol synthesis kinetics and it is suggested that the selectivity with respect to CO and CO2 is only described correctly with BEEF-vdW.
35. 35
  Zhu, Y.-A.; Chen, D.; Zhou, X.-G.; Yuan, W.-K. DFT studies of dry reforming of methane on Ni catalyst. Catal. Today 2009, 148, 260– 267, DOI: 10.1016/j.cattod.2009.08.022
  35
  DFT studies of dry reforming of methane on Ni catalyst
  Zhu, Yi-An; Chen, De; Zhou, Xing-Gui; Yuan, Wei-Kang
  Catalysis Today (2009), 148 (3-4), 260-267CODEN: CATTEA; ISSN:0920-5861. (Elsevier B.V.)
  First-principles calcns. based on d. functional theory (DFT) were used to investigate the reaction mechanism of dry methane reforming on Ni(1 1 1). The most energetically favorable adsorption configurations of the species involved in this process are identified and the transition states for all the possible elementary steps are explored by the dimer method. Then, the related thermodn. properties at 973.15 K are calcd. by including the zero-point energy correction, thermal energy correction and entropic effect. CO2 dissocs. via a direct pathway to produce CO and O dominantly, and at. O is revealed to be the primary oxidant of CHx intermediates. Based on this information, two dominant reaction pathways are constructed as both the CH and C oxidn. are likely. The reaction network begins with the dissocn. of CO2 and CH4, and then the generated CH and C are oxidized by at. O to produce CHO and CO, followed by the CHO decompn. to finally generate CO and H2. As for these two reaction pathways, the oxidn. step is predicted to det. the overall reaction rate under the current investigated conditions, while the CH4 dissocn. is the rate-limiting step at lower temps.
36. 36
  Shang, H.; Sun, W.; Sui, R.; Pei, J.; Zheng, L.; Dong, J.; Jiang, Z.; Zhou, D.; Zhuang, Z.; Chen, W. Engineering Isolated Mn-N₂C₂ Atomic Interface Sites for Efficient Bifunctional Oxygen Reduction and Evolution Reaction. Nano Lett. 2020, 20, 5443– 5450, DOI: 10.1021/acs.nanolett.0c01925
  36
  Engineering Isolated Mn-N2C2 Atomic Interface Sites for Efficient Bifunctional Oxygen Reduction and Evolution Reaction
  Shang, Huishan; Sun, Wenming; Sui, Rui; Pei, Jiajing; Zheng, Lirong; Dong, Juncai; Jiang, Zhuoli; Zhou, Danni; Zhuang, Zhongbin; Chen, Wenxing; Zhang, Jiatao; Wang, Dingsheng; Li, Yadong
  Nano Letters (2020), 20 (7), 5443-5450CODEN: NALEFD; ISSN:1530-6984. (American Chemical Society)
  Oxygen-involved electrochem. reactions are crucial for plenty of energy conversion techniques. Herein, we rationally designed a carbon-based Mn-N2C2 bifunctional electrocatalyst. It exhibits a half-wave potential of 0.915 V vs. reversible hydrogen electrode for oxygen redn. reaction (ORR), and the overpotential is 350 mV at 10 mA cm-2 during oxygen evolution reaction (OER) in alk. condition. Furthermore, by means of operando X-ray absorption fine structure measurements, we reveal that the bond-length-extended Mn2+-N2C2 at. interface sites act as active centers during the ORR process, while the bond-length-shortened high-valence Mn4+-N2C2 moieties serve as the catalytic sites for OER, which is consistent with the d. functional theory results. The at. and electronic synergistic effects for the isolated Mn sites and the carbon support play a crit. role to promote the oxygen-involved catalytic performance, by regulating the reaction free energy of intermediate adsorption. Our results give an at. interface strategy for nonprecious bifunctional single-atom electrocatalysts.
37. 37
  Wang, Y.; Shi, R.; Shang, L.; Waterhouse, G. I. N.; Zhao, J.; Zhang, Q.; Gu, L.; Zhang, T. High-Efficiency Oxygen Reduction to Hydrogen Peroxide Catalyzed by Nickel Single-Atom Catalysts with Tetradentate N₂O₂ Coordination in a Three-Phase Flow Cell. Angew. Chem., Int. Ed. 2020, 59, 13057– 13062, DOI: 10.1002/anie.202004841
  37
  High-Efficiency Oxygen Reduction to Hydrogen Peroxide Catalyzed by Nickel Single-Atom Catalysts with Tetradentate N2O2 Coordination in a Three-Phase Flow Cell
  Wang, Yulin; Shi, Run; Shang, Lu; Waterhouse, Geoffrey I. N.; Zhao, Jiaqi; Zhang, Qinghua; Gu, Lin; Zhang, Tierui
  Angewandte Chemie, International Edition (2020), 59 (31), 13057-13062CODEN: ACIEF5; ISSN:1433-7851. (Wiley-VCH Verlag GmbH & Co. KGaA)
  C-supported NiII single-atom catalysts with a tetradentate Ni-N2O2 coordination formed by a Schiff base ligand-mediated pyrolysis strategy are presented. A NiII complex of the Schiff base ligand (R,R)-(-)-N,N'-bis(3,5-di-tert-butylsalicylidene)-1,2-cyclohexanediaminewasadsorbed onto a C black support, followed by pyrolysis of the modified C material at 300° in Ar. The Ni-N2O2/C catalyst showed excellent performance for the electrocatalytic redn. of O2 to H2O2 through a two-electron transfer process in alk. conditions, with a H2O2 selectivity of 96%. At a c.d. of 70 mA cm-2, a H2O2 prodn. rate of 5.9 mol gcat.-1 h-1 was achieved using a three-phase flow cell, with good catalyst stability maintained over 8 h of testing. The Ni-N2O2/C catalyst could electrocatalytically reduce O2 in air to H2O2 at a high c.d., still affording a high H2O2 selectivity (>90%). A precise Ni-N2O2 coordination was key to the performance.
38. 38
  Ren, C.; Wen, L.; Magagula, S.; Jiang, Q.; Lin, W.; Zhang, Y.; Chen, Z.; Ding, K. Relative efficacy of Co-X₄ embedded graphene (X = N, S, B, and P) electrocatalysts towards hydrogen evolution reaction: Is nitrogen really the best choice?. ChemCatChem 2020, 12, 536– 543, DOI: 10.1002/cctc.201901293
  38
  Relative Efficacy of Co-X4 Embedded Graphene (X=N, S, B, and P) Electrocatalysts towards Hydrogen Evolution Reaction: Is Nitrogen Really the Best Choice?
  Ren, Chunjin; Wen, Lu; Magagula, Saneliswa; Jiang, Qianyu; Lin, Wei; Zhang, Yongfan; Chen, Zhongfang; Ding, Kaining
  ChemCatChem (2020), 12 (2), 536-543CODEN: CHEMK3; ISSN:1867-3880. (Wiley-VCH Verlag GmbH & Co. KGaA)
  The authors perform 1st-principles calcns. to study whether or not N is the best dopant in system of Co-X4 embedded graphene (X = N, S, B, and P) electrocatalysts towards H evolution reaction(HER). The theor. results reveal that N, S, B, and P-doped graphene can enhance the catalytic activity toward HER compared with the pristine graphene, and S doped graphene exhibits more favorable performance than N doped graphene, consistent with the exptl. results. For the Co-X4 embedded graphene (X = N, S, B, and P), the authors predict that S may be a promising dopant in graphene supported single atom Co. The rather low H adsorption free energy (-0.07 eV) and activation energy barrier (0.78 eV) for the rate-detg. step, the downshift of the d band center, the enhanced charge d. of dz2 orbital as well as the reduced work function are responsible for the unexpected activity of Co-S4 embedded graphene for HER. Overall, Co-S4 embedded graphene catalyst could be a good candidate for H evolution reaction.
39. 39
  Chen, T.; He, T.; Benesty, M.; Khotilovich, V.; Tang, Y. Xgboost: extreme gradient boosting , R package version 0.4-2; 2015, 1. https://xgboost.readthedocs.io/en/latest/R-package/index.html
  There is no corresponding record for this reference.
40. 40
  Olson, R. S.; Bartley, N.; Urbanowicz, R. J.; Moore, J. H. Evaluation of a Tree-based Pipeline Optimization Tool for Automating Data Science. Proceedings of GECCO 2016, 485– 492, DOI: 10.1145/2908812.2908918
  There is no corresponding record for this reference.
41. 41
  Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Müller, A. Scikit-learn: Machine learning in Python. J. Mach. Learn. Res. 2011, 2825– 2830
  There is no corresponding record for this reference.
42. 42
  Peterson, A. A.; Abild-Pedersen, F.; Studt, F.; Rossmeisl, J.; Nørskov, J. K. How copper catalyzes the electroreduction of carbon dioxide into hydrocarbon fuels. Energy Environ. Sci. 2010, 3, 1311– 1315, DOI: 10.1039/c0ee00071j
  42
  How copper catalyzes the electroreduction of carbon dioxide into hydrocarbon fuels
  Peterson, Andrew A.; Abild-Pedersen, Frank; Studt, Felix; Rossmeisl, Jan; Norskov, Jens K.
  Energy & Environmental Science (2010), 3 (9), 1311-1315CODEN: EESNBY; ISSN:1754-5706. (Royal Society of Chemistry)
  D. functional theory calcns. explain copper's unique ability to convert CO2 into hydrocarbons, which may open up (photo-)electrochem. routes to fuels.
43. 43
  Kibria, M. G.; Edwards, J. P.; Gabardo, C. M.; Dinh, C. T.; Seifitokaldani, A.; Sinton, D.; Sargent, E. H. Electrochemical CO₂ Reduction into Chemical Feedstocks: From Mechanistic Electrocatalysis Models to System Design. Adv. Mater. 2019, 31, 1807166, DOI: 10.1002/adma.201807166
  There is no corresponding record for this reference.
44. 44
  Peterson, A. A.; Nørskov, J. K. Activity descriptors for CO₂ electroreduction to methane on transition-metal catalysts. J. Phys. Chem. Lett. 2012, 3, 251– 258, DOI: 10.1021/jz201461p
  44
  Activity Descriptors for CO2 Electroreduction to Methane on Transition-Metal Catalysts
  Peterson, Andrew A.; Noerskov, Jens K.
  Journal of Physical Chemistry Letters (2012), 3 (2), 251-258CODEN: JPCLCD; ISSN:1948-7185. (American Chemical Society)
  The electrochem. redn. of CO2 into hydrocarbons and alcs. would allow renewable energy sources to be converted into fuels and chems. However, no electrode catalysts were developed that can perform this transformation with a low overpotential at reasonable current densities. The authors compare trends in binding energies for the intermediates in CO2 electrochem. redn. and present an activity volcano based on this anal. This anal. describes the exptl. obsd. variations in transition-metal catalysts, including why Cu is the best-known metal electrocatalyst. The protonation of adsorbed CO is singled out as the most important step dictating the overpotential. New strategies are presented for the discovery of catalysts that can operate with a reduced overpotential.
45. 45
  Liu, X.; Xiao, J.; Peng, H.; Hong, X.; Chan, K.; Norskov, J. K. Understanding trends in electrochemical carbon dioxide reduction rates. Nat. Commun. 2017, 8, 15438, DOI: 10.1038/ncomms15438
  45
  Understanding trends in electrochemical carbon dioxide reduction rates
  Liu, Xinyan; Xiao, Jianping; Peng, Hongjie; Hong, Xin; Chan, Karen; Noerskov, Jens K.
  Nature Communications (2017), 8 (), 15438CODEN: NCAOBW; ISSN:2041-1723. (Nature Publishing Group)
  Electrochem. carbon dioxide redn. to fuels presents one of the great challenges in chem. Herein we present an understanding of trends in electrocatalytic activity for carbon dioxide redn. over different metal catalysts that rationalize a no. of exptl. observations including the selectivity with respect to the competing hydrogen evolution reaction. We also identify two design criteria for more active catalysts. The understanding is based on d. functional theory calcns. of activation energies for electrochem. carbon monoxide redn. as a basis for an electrochem. kinetic model of the process. We develop scaling relations relating transition state energies to the carbon monoxide adsorption energy and det. the optimal value of this descriptor to be very close to that of copper.
46. 46
  Zhong, M.; Tran, K.; Min, Y.; Wang, C.; Wang, Z.; Dinh, C.-T.; De Luna, P.; Yu, Z.; Rasouli, A. S.; Brodersen, P. Accelerated discovery of CO₂ electrocatalysts using active machine learning. Nature 2020, 581, 178– 183, DOI: 10.1038/s41586-020-2242-8
  46
  Accelerated discovery of CO2 electrocatalysts using active machine learning
  Zhong, Miao; Tran, Kevin; Min, Yimeng; Wang, Chuanhao; Wang, Ziyun; Dinh, Cao-Thang; De Luna, Phil; Yu, Zongqian; Rasouli, Armin Sedighian; Brodersen, Peter; Sun, Song; Voznyy, Oleksandr; Tan, Chih-Shan; Askerka, Mikhail; Che, Fanglin; Liu, Min; Seifitokaldani, Ali; Pang, Yuanjie; Lo, Shen-Chuan; Ip, Alexander; Ulissi, Zachary; Sargent, Edward H.
  Nature (London, United Kingdom) (2020), 581 (7807), 178-183CODEN: NATUAS; ISSN:0028-0836. (Nature Research)
  The rapid increase in global energy demand and the need to replace carbon dioxide (CO2)-emitting fossil fuels with renewable sources have driven interest in chem. storage of intermittent solar and wind energy1,2. Particularly attractive is the electrochem. redn. of CO2 to chem. feedstocks, which uses both CO2 and renewable energy3-8. Copper has been the predominant electrocatalyst for this reaction when aiming for more valuable multi-carbon products9-16, and process improvements have been particularly notable when targeting ethylene. However, the energy efficiency and productivity (c.d.) achieved so far still fall below the values required to produce ethylene at cost-competitive prices. Here we describe Cu-Al electrocatalysts, identified using d. functional theory calcns. in combination with active machine learning, that efficiently reduce CO2 to ethylene with the highest Faradaic efficiency reported so far. This Faradaic efficiency of over 80 per cent (compared to about 66 per cent for pure Cu) is achieved at a c.d. of 400 mA per square centimetre (at 1.5 V vs. a reversible hydrogen electrode) and a cathodic-side (half-cell) ethylene power conversion efficiency of 55 ± 2 per cent at 150 mA per square centimetre. We perform computational studies that suggest that the Cu-Al alloys provide multiple sites and surface orientations with near-optimal CO binding for both efficient and selective CO2 redn.17. Furthermore, in situ X-ray absorption measurements reveal that Cu and Al enable a favorable Cu coordination environment that enhances C-C dimerization. These findings illustrate the value of computation and machine learning in guiding the exptl. exploration of multi-metallic systems that go beyond the limitations of conventional single-metal electrocatalysts.
47. 47
  Wang, Y.; Shi, R.; Shang, L.; Waterhouse, G. I.; Zhao, J.; Zhang, Q.; Gu, L.; Zhang, T. High-Efficiency Oxygen Reduction to Hydrogen Peroxide Catalyzed by Nickel Single-Atom Catalysts with Tetradentate N₂O₂ Coordination in a Three-Phase Flow Cell. Angew. Chem., Int. Ed. 2020, 59, 13057– 13062, DOI: 10.1002/anie.202004841
  47
  High-Efficiency Oxygen Reduction to Hydrogen Peroxide Catalyzed by Nickel Single-Atom Catalysts with Tetradentate N2O2 Coordination in a Three-Phase Flow Cell
  Wang, Yulin; Shi, Run; Shang, Lu; Waterhouse, Geoffrey I. N.; Zhao, Jiaqi; Zhang, Qinghua; Gu, Lin; Zhang, Tierui
  Angewandte Chemie, International Edition (2020), 59 (31), 13057-13062CODEN: ACIEF5; ISSN:1433-7851. (Wiley-VCH Verlag GmbH & Co. KGaA)
  C-supported NiII single-atom catalysts with a tetradentate Ni-N2O2 coordination formed by a Schiff base ligand-mediated pyrolysis strategy are presented. A NiII complex of the Schiff base ligand (R,R)-(-)-N,N'-bis(3,5-di-tert-butylsalicylidene)-1,2-cyclohexanediaminewasadsorbed onto a C black support, followed by pyrolysis of the modified C material at 300° in Ar. The Ni-N2O2/C catalyst showed excellent performance for the electrocatalytic redn. of O2 to H2O2 through a two-electron transfer process in alk. conditions, with a H2O2 selectivity of 96%. At a c.d. of 70 mA cm-2, a H2O2 prodn. rate of 5.9 mol gcat.-1 h-1 was achieved using a three-phase flow cell, with good catalyst stability maintained over 8 h of testing. The Ni-N2O2/C catalyst could electrocatalytically reduce O2 in air to H2O2 at a high c.d., still affording a high H2O2 selectivity (>90%). A precise Ni-N2O2 coordination was key to the performance.
48. 48
  Feng, B.; Zhang, J.; Zhong, Q.; Li, W.; Li, S.; Li, H.; Cheng, P.; Meng, S.; Chen, L.; Wu, K. Experimental realization of two-dimensional boron sheets. Nat. Chem. 2016, 8, 563– 8, DOI: 10.1038/nchem.2491
  48
  Experimental realization of two-dimensional boron sheets
  Feng, Baojie; Zhang, Jin; Zhong, Qing; Li, Wenbin; Li, Shuai; Li, Hui; Cheng, Peng; Meng, Sheng; Chen, Lan; Wu, Kehui
  Nature Chemistry (2016), 8 (6), 563-568CODEN: NCAHBB; ISSN:1755-4330. (Nature Publishing Group)
  Two-dimensional boron sheets have been grown epitaxially on Ag(111) substrate. Two types of boron sheet, a β12 sheet and a χ3 sheet, both exhibiting a triangular lattice but with different arrangements of periodic holes, are obsd. by scanning tunnelling microscopy. D. functional theory simulations agree well with expts., and indicate that both sheets are planar without obvious vertical undulations. The boron sheets are quite inert to oxidization and interact only weakly with their substrate.
49. 49
  Ling, C.; Shi, L.; Ouyang, Y.; Zeng, X. C.; Wang, J. Nanosheet Supported Single-Metal Atom Bifunctional Catalyst for Overall Water Splitting. Nano Lett. 2017, 17, 5133– 5139, DOI: 10.1021/acs.nanolett.7b02518
  49
  Nanosheet Supported Single-Metal Atom Bifunctional Catalyst for Overall Water Splitting
  Ling, Chongyi; Shi, Li; Ouyang, Yixin; Zeng, Xiao Cheng; Wang, Jinlan
  Nano Letters (2017), 17 (8), 5133-5139CODEN: NALEFD; ISSN:1530-6984. (American Chemical Society)
  Nanosheet supported single-atom catalysts (SACs) can make full use of metal atoms and yet entail high selectivity and activity, and bifunctional catalysts can enable higher performance while lowering the cost than two sep. unifunctional catalysts. Supported single-atom bifunctional catalysts are therefore of great economic interest and scientific importance. Here, on the basis of first-principles computations, we report a design of the first single-atom bifunctional electrocatalyst, namely, isolated nickel atom supported on β12 boron monolayer (Ni1/β12-BM), to achieve overall water splitting. This nanosheet supported SAC exhibits remarkable electrocatalytic performance with the computed overpotential for oxygen/hydrogen evolution reaction being just 0.40/0.06 V. The ab initio mol. dynamics simulation shows that the SAC can survive up to 800 K elevated temp., while enacting a high energy barrier of 1.68 eV to prevent isolated Ni atoms from clustering. A viable exptl. route for the synthesis of Ni1/β12-BM SAC is demonstrated from computer simulation. The desired nanosheet supported single-atom bifunctional catalysts not only show great potential for achieving overall water splitting but also offer cost-effective opportunities for advancing clean energy technol.
Supporting Information
Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcc.0c05964.
- Figures of SAC structures of the training data set; CO/H adsorption energy prediction results of the comparison between the XGBR model and DFT computed data with different data split ratios and different feature sets; the Pearson correlation coefficient heat maps and feature importance ranks of different feature sets for CO/H adsorption energy prediction model; H adsorption prediction models trained by different algorithms; AIMD results of some predicted SACs; parameters of the final machine learning models; cross validation results of models; and the order number in the heat map of different kinds of structures (PDF)
- jp0c05964_si_001.pdf (1.04 MB)
Terms & Conditions

Most electronic Supporting Information files are available without a subscription to ACS Web Editions. Such files may be downloaded by article for research use (if there is a public use license linked to the relevant article, that license may permit other uses). Permission may be obtained from ACS for other uses through requests via the RightsLink permission system: http://pubs.acs.org/page/copyright/permissions.html.

A Machine Learning Model on Simple Features for CO2 Reduction Electrocatalysts基于 CO2 还原电催化剂简单特征的机器学习模型Click to copy article link点击复制文章链接Article link copied!

The Journal of Physical Chemistry C

Article Views

Altmetric

Citations

Abstract 抽象

Special Issue 所属专题

Introduction 介绍

Figure 1 图 1

Computational Details 计算详细信息

Establishment of a Database建立数据库

Machine Learning Methods 机器学习方法

Results and Discussion 结果与讨论

Determination of Target Features for CO2 Electroreduction确定 CO2 电还原的目标特性

Selection and Optimization of Features特征的选择和优化

Model Training and Performance模型训练和性能

Figure 2 图 2

Figure 3 图 3

Screening of Potential Candidates by the Prediction Model通过预测模型筛选潜在候选人

Figure 4 图 4

Figure 5 图 5

Figure 6 图 6

Conclusions 结论

Supporting Information 支持信息

Terms & Conditions 条款和条件

Author Information 作者信息

Acknowledgments 确认

References 引用

Cited By

Recommended Articles

Machine Learning for Electrocatalyst and Photocatalyst Design and Discovery

Machine-Learning-Guided Discovery and Optimization of Additives in Preparing Cu Catalysts for CO2 Reduction

A Machine Learning Model To Predict CO2 Reduction Reactivity and Products Transferred from Metal-Zeolites

Machine Learning Assisted Understanding and Discovery of CO2 Reduction Reaction Electrocatalyst

Machine-Learning-Accelerated Catalytic Activity Predictions of Transition Metal Phthalocyanine Dual-Metal-Site Catalysts for CO2 Reduction

The Journal of Physical Chemistry C

Article Views

Altmetric

Citations

Recommended Articles

Machine Learning for Electrocatalyst and Photocatalyst Design and Discovery

Machine-Learning-Guided Discovery and Optimization of Additives in Preparing Cu Catalysts for CO2 Reduction

A Machine Learning Model To Predict CO2 Reduction Reactivity and Products Transferred from Metal-Zeolites

Machine Learning Assisted Understanding and Discovery of CO2 Reduction Reaction Electrocatalyst

Machine-Learning-Accelerated Catalytic Activity Predictions of Transition Metal Phthalocyanine Dual-Metal-Site Catalysts for CO2 Reduction

Abstract

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

References

Supporting Information

Supporting Information

Terms & Conditions

A Machine Learning Model on Simple Features for CO₂ Reduction Electrocatalysts
基于 CO₂ 还原电催化剂简单特征的机器学习模型
Click to copy article link
点击复制文章链接Article link copied!

Establishment of a Database
建立数据库

Determination of Target Features for CO₂ Electroreduction
确定 CO₂ 电还原的目标特性

Selection and Optimization of Features
特征的选择和优化

Model Training and Performance
模型训练和性能

Screening of Potential Candidates by the Prediction Model
通过预测模型筛选潜在候选人

Machine-Learning-Guided Discovery and Optimization of Additives in Preparing Cu Catalysts for CO₂ Reduction

A Machine Learning Model To Predict CO₂ Reduction Reactivity and Products Transferred from Metal-Zeolites

Machine Learning Assisted Understanding and Discovery of CO₂ Reduction Reaction Electrocatalyst

Machine-Learning-Accelerated Catalytic Activity Predictions of Transition Metal Phthalocyanine Dual-Metal-Site Catalysts for CO₂ Reduction

Machine-Learning-Guided Discovery and Optimization of Additives in Preparing Cu Catalysts for CO₂ Reduction

A Machine Learning Model To Predict CO₂ Reduction Reactivity and Products Transferred from Metal-Zeolites

Machine Learning Assisted Understanding and Discovery of CO₂ Reduction Reaction Electrocatalyst

Machine-Learning-Accelerated Catalytic Activity Predictions of Transition Metal Phthalocyanine Dual-Metal-Site Catalysts for CO₂ Reduction