Enhanced Hydrogen Evolution Performance at the Lateral Interface between Two Layered Materials Predicted with Machine Learning

Abstract

While economical and effective catalysts are required for sustainable hydrogen production, low-dimensional interfacial engineering techniques have been developed to improve the catalytic activity in the hydrogen evolution reaction (HER). In this study, we used density functional theory (DFT) calculations to measure the Gibbs free energy change (ΔG H) in hydrogen adsorption in two-dimensional lateral heterostructures (LHSs) MX2/M’X’2 (MoS2/WS2, MoS2/WSe2, MoSe2/WS2, MoSe2/WSe2, MoTe2/WSe2, MoTe2/WTe2, and WS2/WSe2) and MX2/M’X’ (NbS2/ZnO, NbSe2/ZnO, NbS2/GaN, MoS2/ZnO, MoSe2/ZnO, MoS2/AlN, MoS2/GaN, and MoSe2/GaN) at several different positions near the interface. Compared to the interfaces of LHS MX2/M’X’2 and the surfaces of the monolayer MX2 and MX, the interfaces of LHS MX2/M’X’ display greater hydrogen evolution reactivity due to their metallic behavior. The hydrogen absorption is stronger at the interfaces of LHS MX2/M’X’, and that facilitates proton accessibility and increases the usage of catalytically active sites. Here, we develop three types of descriptors that can be used universally in 2D materials and can explain changes in ΔG H for different adsorption sites in a single LHS using only the basic information of the LHSs (type and number of neighboring atoms to the adsorption points). Using the DFT results of the LHSs and the various experimental data of atomic information, we trained machine learning (ML) models with the chosen descriptors to predict promising combinations and adsorption sites for HER catalysts among the LHSs. Our ML model achieved an R 2 score of 0.951 (regression) and an F 1 score of 0.749 (classification). Furthermore, the developed surrogate model was implemented to predict the structures in the test set and was based on confirmation from the DFT calculations via ΔG H values. The LHS MoS2/ZnO is the best candidate for HER among 49 candidates considered using both DFT and ML models because it has a ΔG H of −0.02 eV on top of O at the interface position and requires only −171 mV of overpotential to obtain the standard current density (10 A/cm2).

When Machine Learning Meets 2D Materials: A Review

“…The introduction of ML to research on 2D materials has considerably increased the efficiency of discovering new 2D materials. As shown in Table 3 , most of these new materials are catalytic materials, [ 60 , 61 , 62 , 63 , 64 , 65 , 66 , 67 , 68 ] photoelectric materials, [ 69 , 70 , 71 , 72 , 73 , 74 , 75 , 76 , 77 ] and ferromagnetic materials. [ 78 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 90 , 91 , 92 ] In most cases, we search for 2D materials with specific desired properties in existing open‐source databases or in new material sets generated through methods such as element replacement within the original cell.…”

“…In studies related to material properties, ML has been combined with DFT and MD to explore the thermal properties, [ 17 , 33 , 34 , 35 , 36 , 37 , 38 , 46 ] bandgaps, [ 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 ] and mechanical properties [ 56 , 57 , 58 , 59 ] of various materials, thereby accelerating the pace of research in this field. Furthermore, ML has also been utilized for the discovery of novel 2D materials, including catalytic, [ 60 , 61 , 62 , 63 , 64 , 65 , 66 , 67 , 68 ] photoelectric, [ 69 , 70 , 71 , 72 , 73 , 74 , 75 , 76 , 77 ] and magnetic materials. [ 78 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 90 , 91 , 92 ] In terms of preparing 2D materials, ML methods have been ap...…”

Machine-Learning-Driven High-Throughput Screening of Transition-Metal Atom Intercalated g-C3N4/MX2 (M = Mo, W; X = S, Se, Te) Heterostructures for the Hydrogen Evolution Reaction

“…A generally simple yet powerful approach involving integration of artificial intelligence with data derived from density functional theory (DFT) calculations hold great promise in expediting the exploration and optimization of efficient catalyst screening. − Machine-learning (ML) algorithms thoroughly analyze extensive experimental/computational data by rapidly assessing an array of potential catalyst compositions and performances, thereby reducing the time and cost compared to traditional trial-and-error methodologies. , The selection of an appropriate ML model hinges on the careful definition of distinctive features that effectively encapsulate the catalysts in relation to the target variable within a diverse data set . Once ML algorithms are trained, they serve the purpose of predicting highly active catalysts, conducting feature importance analysis, integrating novel descriptors, and ultimately facilitating the utilization of optimized catalysts in the intended electrochemical reactions. , Nevertheless, the utilization of ML in the domain of hydrogen evolution, especially concerning intercalated heterostructures, remains relatively underexplored due to the scarcity of high-quality data concerning performance indicators of HER and the characteristics of these heterostructures . To this end, considerable attention has been directed toward developing such ML methodologies that enable direct catalyst design and screening, bypassing the need for extensive DFT computations.…”

“…35,36 Nevertheless, the utilization of ML in the domain of hydrogen evolution, especially concerning intercalated heterostructures, remains relatively underexplored due to the scarcity of high-quality data concerning performance indicators of HER and the characteristics of these heterostructures. 37 To this end, considerable attention has been directed toward developing such ML methodologies that enable direct catalyst design and screening, bypassing the need for extensive DFT computations.…”

Two-dimensional carbon/boron nitrides: modification, machine learning and beyond

“…44,45 This reduces the inefficiencies inherent in traditional trial-and-error approaches and expedites the design and development of novel materials. Significantly, ML expedites the process of 2D material discovery, 46–48 allowing for the rapid screening of thousands of potential structures to identify those with the most promising properties for specific applications. 49…”