Accurate Adsorption Energies Research Articles

Two-stage feature selection for machine learning-aided DFT-based surface reactivity study on single-atom alloys

This paper presents a feature-centric strategy for predicting adsorption energies of key CO2 reduction reaction (CO2RR) adsorbates, CO and H species, utilizing density functional theory-based calculations for eight adsorption sites and considering alloying effects of nine transition metals at single-atom concentrations. Here, we explore a class of materials consisting of a majority host metal where individual atoms of a different element are dispersed called single-atom alloys (SAA). A total of eight feature selection methods are assessed within Gradient Boosting Regression and Linear Regression models. This study proposes a practical and effective two-stage approach that narrows down the initial 86 features to subsets of 10 and 7 for CO and H adsorption energy predictions, respectively, with the arithmetic mean of valence electrons (VE-am) feature consistently emerging as highly influential, validated through permutation and Shapley additive explanations-based feature importance analyses. The models exhibit robust performance on unseen data, indicating their generalization capability. The findings emphasize VE-am as a potential key machine learning feature for CO2RR on SAA surfaces and underline the effectiveness of the feature-centric approach in understanding feature impacts in machine learning models for CO2RR on SAA systems. Additionally, while other features based on structural, electronic and elemental properties may not individually impact the model significantly, their collective contribution plays a vital role in achieving more accurate adsorption energy predictions.

Xsorb: A software for identifying the most stable adsorption configuration and energy of a molecule on a crystal surface

Molecular adsorption is the first important step of many surface-mediated chemical processes, from catalysis to lubrication. This phenomenon is controlled by physical/chemical interactions, which can be accurately described by first-principles calculations. Several computational tools have been developed to study molecular adsorption based on high throughput/automatized approaches in recent years. However, these tools can sometimes be over-sophisticated for non-expert users. Here we present Xsorb, a Python-based program for identifying the accurate adsorption energy and geometry of complex molecules on crystalline (reconstructed) surfaces. The program automatically samples the potential energy surface (PES) that describes the molecule-surface interaction by generating several adsorption configurations through symmetry operations. The set of the most representative ones is automatically identified through a fast pre-optimization scheme. Finally, the PES global minimum is identified through a full structural optimization process. We show the program capabilities through an example consisting of a hydrocarbon molecule, 1-hexene, adsorbed over the (110) surface of iron and the reconstructed (001) surface of diamond. This program, despite its conceptual simplicity, is very effective in reducing the computational workload usually associated with the creation and optimization of several adsorption configurations. Program summaryProgram title: XsorbCPC Library link to program files:https://doi.org/10.17632/kv97tgybx8.1Developer's repository link:https://gitlab.com/triboteam/xsorbed/Licensing provisions: CC by 4.0Programming language: Python (version 3.7 and above) and Quantum ESPRESSO (for the ab initio calculations).Nature of problem: Identifying the most stable adsorption configuration of a molecule over a substrate and compute its adsorption energy.Solution method: Creating a Python-based code that generates many adsorption configurations with different molecular orientations, performs a preliminary partial geometrical optimization with density functional theory calculations of all these configurations, identifies the most relevant ones for the full geometrical optimization and computes the adsorption energy.

Machine learned features from density of states for accurate adsorption energy prediction

Materials databases generated by high-throughput computational screening, typically using density functional theory (DFT), have become valuable resources for discovering new heterogeneous catalysts, though the computational cost associated with generating them presents a crucial roadblock. Hence there is a significant demand for developing descriptors or features, in lieu of DFT, to accurately predict catalytic properties, such as adsorption energies. Here, we demonstrate an approach to predict energies using a convolutional neural network-based machine learning model to automatically obtain key features from the electronic density of states (DOS). The model, DOSnet, is evaluated for a diverse set of adsorbates and surfaces, yielding a mean absolute error on the order of 0.1 eV. In addition, DOSnet can provide physically meaningful predictions and insights by predicting responses to external perturbations to the electronic structure without additional DFT calculations, paving the way for the accelerated discovery of materials and catalysts by exploration of the electronic space.

Large Impact of Approximate Exchange-Correlation Functionals on Modeling the Water Gas Shift Reaction on Copper

We investigate the water gas shift reaction (WGSR) on copper using three levels of exchange-correlation (XC) functionals of increasing complexity: the Perdew–Burke–Ernzerhof (PBE) functional, the Heyd–Scuseria–Ernzerhof (HSE) hybrid functional, and the exact exchange plus the random phase approximation (RPA) correlation functional. We show that DFT predictions for the kinetics of the WGSR strongly depend on the choice of XC functionals. It is important to have accurate CO adsorption energy. Due to PBE’s overestimation of CO’s adsorption energy, it predicts small free surface and a negative (incorrect) reaction order for CO. HSE and RPA largely avoid such overestimation and predict large free surface and positive reaction orders for CO. The key finding in this work is that the prediction for the WGSR mechanism also depends on the choice of XC functionals. PBE and HSE predict the carboxyl mechanism, while RPA predicts that both redox and carboxyl mechanisms are important. These results suggest that caution should be paid when using approximate XC functionals to model heterogeneous catalysis (such as the WGSR investigated here) in which several mechanisms compete. In addition, we also observed one problem for RPA: It underestimates the overall WGSR energy and predicts a negative reaction order for CO2. In addition, we examine formate’s role in the WGSR over copper. Both DFT and previous experiments suggest that formate does not participate much in the reaction and cannot cause a negative reaction for CO2 for the WGSR over copper.

Surface charging activated mechanism change: A computational study of O, CO, and CO2 interactions on Ag electrodes

Electrocatalytic and plasma-activated processes receive increasing attention in catalysis. Density functional theory (DFT) calculations are state-of-the-art tools for the fundamental study of reaction mechanisms and predicting the performance of catalytic materials. Proper application of DFT-based methods is crucial when investigating charge-doped electrode surfaces during electrocatalytic and plasma-activated reactions. Here, as a model electrode for plasma-activated CO2 splitting, we studied the interactions of O, CO, and CO2 with the neutral and progressively charged Ag(111) metal surfaces. We show that the application of correction procedures is necessary to obtain accurate adsorption energy profiles of O atoms, CO and CO2 molecules on Ag surfaces that are under the influence of additional electrons. Interestingly, the oxidation of CO is found to shift from a Langmuir–Hinshelwood mechanism on a neutral electrode to an Eley–Rideal mechanism on charged electrodes. Furthermore, we show that the surface charging of Ag(111) electrodes increase their CO2 reduction performance by enhancing the adsorption of O atoms and desorption of CO molecules. A further increase in the absolute charge-state of the electrode surface is expected to waive the thermodynamic barriers for the CO2 splitting reaction.

Efficient and accurate description of adsorption in zeolites.

Accurate theoretical methods are needed to correctly describe adsorption on solid surfaces or in porous materials. The random phase approximation (RPA) with singles corrections scheme and the second order Møller-Plesset perturbation theory (MP2) are two schemes, which offer high accuracy at affordable computational cost. However, there is little knowledge about their applicability and reliability for different adsorbates and surfaces. Here, we calculate adsorption energies of seven different molecules in zeolite chabazite to show that RPA with singles corrections is superior to MP2, not only in terms of accuracy but also in terms of computer time. Therefore, RPA with singles is a suitable scheme for obtaining highly accurate adsorption energies in porous materials and similar systems.

Machine Learning and Scaling Laws for Prediction of Accurate Adsorption Energy.

Finding an "ideal" catalyst is a matter of great interest in the communities of chemists and material scientists, partly because of its wide spectrum of industrial applications. Information regarding a physical parameter termed "adsorption energy", which dictates the degrees of adhesion of an adsorbate on a substrate, is a primary requirement in selecting the catalyst for catalytic reactions. Both experiments and in silico modeling are extensively being used in estimating the adsorption energies, both of which are an Edisonian approach, demand plenty of resources, and are time-consuming. In this paper, employing a data-mining approach, we predict the adsorption energies of monoatomic and diatomic gases on the surfaces of many transition metals (TMs) in no time. With less than a set of 10 simple atomic features, our predictions of the adsorption energies are within a root-mean-squared error (RMSE) of 0.4 eV with the quantum many-body perturbation theory estimates, a computationally expensive method with a good experimental agreement. Based on the important features obtained from machine learning models, we construct a set of mathematical equations using the compressed sensing technique to calculate adsorption energy. We also show that the RMSE can be further minimized up to 0.10 eV using the precomputed adsorption energies obtained with the conventional exchange and correlation (XC) functional by a new set of scaling relations.

Physisorption of Water on Graphene: Subchemical Accuracy from Many-Body Electronic Structure Methods.

Wet carbon interfaces are ubiquitous in the natural world and exhibit anomalous properties, which could be exploited by emerging technologies. However, progress is limited by lack of understanding at the molecular level. Remarkably, even for the most fundamental system (a single water molecule interacting with graphene), there is no consensus on the nature of the interaction. We tackle this by performing an extensive set of complementary state-of-the-art computer simulations on some of the world's largest supercomputers. From this effort a consensus on the water-graphene interaction strength has been obtained. Our results have significant impact for the physical understanding, as they indicate that the interaction is weaker than predicted previously. They also pave the way for more accurate and reliable studies of liquid water at carbon interfaces.

Chemically Accurate Adsorption Energies: CO and H2O on the MgO(001) Surface.

Hybrid MP2:DFT-D structure optimizations are performed at BSSE-free CBS-extrapolated potential energy surfaces for molecule-oxide surface interactions (BSSE, basis set superposition error; CBS, complete basis set limit). Subsequently single point MP2 calculations are performed to estimate the effects of increasing the basis set size in the CBS extrapolation and increasing the cluster model size. The resulting estimates of the periodic MP2 limit agree within 1 kJ/mol with Local MP2 calculations using periodic boundary conditions. Single point CCSD(T) calculations are performed to determine ΔCC = CCSD(T) - MP2 energy differences. The final hybrid MP2:DFT-D+ΔCC estimate for CO on the MgO(001) surface at low coverage, -21.2 ± 0.5 kJ/mol, is in close agreement with the reference energy derived from temperature-programmed desorption experiments, -20.6 ± 2.4 kJ/mol. For H2O on MgO(001), at limiting zero coverage, we predict an adsorption energy of -53.7 ± 4.2 kJ/mol which falls in the range of values, -55.8 ± 12.2 kJ/mol, derived from a high coverage low energy electron diffraction experiments and estimated lateral interactions.

A DFT study on the interaction of small molecules with alkali metal ion-exchanged ETS-10

Abstract In this paper, we present a systematic quantum-mechanical density functional theory (DFT) study of adsorption of small gas molecules in cation-exchanged Engelhard titanosilicate ETS-10 crystalline materials. Adsorbates with a range of polarities were considered, ranging from polar (H2O), quadrupolar (CO2 and N2), to apolar (CH4) atmospheric gases. Starting from the base-case of Na-ETS-10, other extra framework cations such as Li+, K+, Rb+ and Cs+ were considered. The DFT calculations were performed with the M06-L functional and were corrected for basis set superposition error with the counterpoise method in order to provide accurate and robust geometries and adsorption energies. For all adsorbates, the adsorption enthalpies decrease in the order Li+>Na+>K+>Rb+>Cs+, while adsorbate – cation interaction distances increase along the same order. For the two extreme cases, the enthalpies calculated at the M06-L/6-31++G** level of theory for CH4, N2, CO2, and H2O interaction with Li+(Cs+) exchanged materials are −21.8 (−1.7) kJ·mol−1, −19.0 (−10.7) kJ·mol−1, −34.4 (−21.3) kJ·mol−1, and −70.5 (−36.1) kJ·mol−1, respectively. Additionally, the calculated vibrational frequencies are found to be in quite good agreement with the characteristic vibrational modes of alkali metal cation-exchanged ETS-10 and also with the available experimental frequencies for CH4, N2, CO2, and H2O interactions with alkali metal cations in the 12-membered channel of ETS-10.

Hydrogen evolution reaction: The role of arsenene nanosheet and dopant

The state-of-the-art density functional theory (DFT) is employed to study the catalytic activity of arsenene for hydrogen evolution reaction (HER) and oxygen evolution reaction (OER). We have included dispersion correction to get accurate adsorption energy on the individual catalytic surface (top site). Using binding energy calculation, arsenene is shown to be a potential candidate for HER. Here we investigate the stability and electronic properties of the honeycomb structure of the arsenene system using first-principles calculation to find the effect of different dopants on the fundamental band gap, which is one of the primary parameters in the photocatalytic water splitting. Further, we sieved the dopant for better HER catalytic activity by substituting one of the arsenene (As) atoms by B, N, O, Ge, Ga and Se atoms to make arsenene a better candidate for HER. Our studies depict that HER activity is increased by 82% for O-doped arsenene and OER activity by 87% for B-doped arsenene as compared to pristine arsenene.

Chemically accurate adsorption energies for methane and ethane monolayers on the MgO(001) surface.

A hybrid QM:QM method that combines MP2 as high-level method on cluster models with density functional theory (PBE+D2) as low-level method on periodic models is applied to adsorption of methane and ethane on the MgO(001) surface for which reliable experimental desorption enthalpies are available. Two coverages are considered, monolayer (every second Mg2+ ion occupied) and one quarter coverage (one of eight Mg2+ ions occupied). Structure optimizations are performed at the hybrid MP2:(PBE+D2) level, with the MP2 energies and forces counterpoise corrected for basis set superposition error and extrapolated to the complete basis set limit. For the MP2 calculations on the adsorbate monolayer a two-body expansion of the lateral molecule-molecule interactions is applied. Higher order correlation effects are evaluated at the hybrid MP2:(PBE+D2) equilibrium structures as coupled cluster [CCSD(T)] - MP2 differences adopting smaller basis sets. The final adsorption energies obtained for monolayer coverage are -14.0 ± 1.0 and -23.3 ± 0.6 kJ mol-1 for CH4·MgO(001) and C2H6·MgO(001), respectively. They agree within 1 kJ mol-1 - well within chemical accuracy limits - with reference energies of -15.0 ± 0.6 and -24.4 ± 0.6 kJ mol-1, respectively. The latter have been derived from measured desorption enthalpy barriers, taking zero-point vibrational energy (ZPVE) and thermal enthalpy contributions into account.

Weak binding mode of CH4 on rutile crystallites from density functional theory calculations

In the recent years, the surfaces of oxides and fluorides have received considerable attention due to their suitability for emerging technologies such as fuel cells. Among these substances, well-known rutile TiO2, and rutile-like SnO2, MgF2 and ZnF2 are of particular interest owing to their utilization in a wide range of applications such as heterogeneous catalysis. In order to probe these materials for their catalytic activity, physisorption of small organic molecules, such as methane, on their respective dominant surfaces is of potential value. However, the lack of adequate experimental results on the adsorption energetics and geometries of binding particles, has stimulated a large number of computational studies. In this work, the van der Waals-driven sorption of CH4 on (110) face of all of the abovementioned structures has been examined computationally. A manifold of dispersion-corrected DFT methods was applied to evaluate the binding and surface energies as well as surface-induced atomic displacements. Only the most energetically favored configuration of the adsorption, has been examined. The calculated values were compared to the pertinent literature data, either theoretical or experimental, when available. Whilst optB86b-vdW produces the most consistent atomic displacements, despite its confinements, revPBE-D3 results in the most accurate adsorption energies for CH4-TiO2(110). Calculated surface energies were in agreement with the prior studies if available.

Corrosion Study of Mild Steel in Aqueous Sulfuric Acid Solution Using 4-Methyl-4H-1,2,4-Triazole-3-Thiol and 2-Mercaptonicotinic Acid-An Experimental and Theoretical Study.

The corrosion behavior of mild steel in 0.1 M aqueous sulfuric acid medium has been studied using weight loss, potentiodynamic polarization measurements, quantum chemical calculations, and molecular dynamic simulations in the presence and absence of 4-methyl-4H-1,2,4-triazole-3-thiol and 2-mercaptonicotinic acid. Potentiodynamic measurements indicate that these compounds mostly act as mixed inhibitors due to their adsorption on the mild steel surface. The goal of the study was to use theoretical calculations to better understand the inhibition. Monte Carlo simulation was used to calculate the adsorption behavior of the studied molecules onto Fe (1 1 1) and Fe2O3 (1 1 1) surface. The molecules were also studied with the density functional theory (DFT), using the B3LYP functional in order to determine the relationship between the molecular structure and the corrosion inhibition behavior. More accurate adsorption energies between the studied molecules and iron or iron oxide were calculated by using DFT with periodic boundary conditions. The calculated theoretical parameters gave important assistance into the understanding the corrosion inhibition mechanism expressed by the molecules and are in full agreement with the experimental results.

Properties of the water to boron nitride interaction: From zero to two dimensions with benchmark accuracy.

Molecular adsorption on surfaces plays an important part in catalysis, corrosion, desalination, and various other processes that are relevant to industry and in nature. As a complement to experiments, accurate adsorption energies can be obtained using various sophisticated electronic structure methods that can now be applied to periodic systems. The adsorption energy of water on boron nitride substrates, going from zero to 2-dimensional periodicity, is particularly interesting as it calls for an accurate treatment of polarizable electrostatics and dispersion interactions, as well as posing a practical challenge to experiments and electronic structure methods. Here, we present reference adsorption energies, static polarizabilities, and dynamic polarizabilities, for water on BN substrates of varying size and dimension. Adsorption energies are computed with coupled cluster theory, fixed-node quantum Monte Carlo (FNQMC), the random phase approximation, and second order Møller-Plesset theory. These wavefunction based correlated methods are found to agree in molecular as well as periodic systems. The best estimate of the water/h-BN adsorption energy is -107±7 meV from FNQMC. In addition, the water adsorption energy on the BN substrates could be expected to grow monotonically with the size of the substrate due to increased dispersion interactions, but interestingly, this is not the case here. This peculiar finding is explained using the static polarizabilities and molecular dispersion coefficients of the systems, as computed from time-dependent density functional theory (DFT). Dynamic as well as static polarizabilities are found to be highly anisotropic in these systems. In addition, the many-body dispersion method in DFT emerges as a particularly useful estimation of finite size effects for other expensive, many-body wavefunction based methods.

A density functional theory computational study of adsorption of Di-Meta-Cyano Azobenzene molecules on Si (111) surfaces

The adsorption of di-meta-cyano azobenzene (DMC) cis and trans isomers on non-passivated and passivated Si (111) (7×7) surfaces is studied using density functional theory (DFT) calculations. Our results reveal that on the non-passivated surface the 12 Si adatoms are accessible to form chemical bonds with DMC molecules. Interestingly, the trans isomer forms two chemical bonds near the corner hole atom in Si (111) (7×7) surface, which is not observed in the widely studied metallic surfaces. The DMC isomers show significant structural distortion in the chemisorption case. The strong chemical bonds (and high bonding energy) could be detrimental to conformation switching between these two isomers under external stimuli. The physisorption case is also examined. Monte Carlo (MC) simulations with empirical force fields were employed to search about 106 different adsorption positions and DMC molecule orientations to identify the stable adsorption sites (up to six). The DFT-PBE and DFT-D2 calculations were then carried out to obtain the relaxed atomistic structures and accurate adsorption energy. We find that it is imperative to take van der Waals (vdW) interaction into account in DFT calculations. Our results show that the adsorption sites generally are encompassed by either the Si adatoms or the passivated H atoms, which could enhance the long-range dispersion interaction between DMC molecules and Si surfaces. The molecular structures of both isomers remain unchanged compared with gas phase. The obtained adsorption energy results ΔEads are moderate (0.2–0.8eV). At some adsorption sites on the passivated surface, both isomers have similar moderate ΔEads (0.4–0.6eV), implying promises of molecular switching that should be examined in experiments.

DFT-Based Method for More Accurate Adsorption Energies: An Adaptive Sum of Energies from RPBE and vdW Density Functionals

In recent years, the popularity of density functional theory with periodic boundary conditions (DFT) has surged for the design and optimization of functional materials. However, no single DFT exchange–correlation functional currently available gives accurate adsorption energies on transition metals both when bonding to the surface is dominated by strong covalent or ionic bonding and when it has strong contributions from van der Waals interactions (i.e., dispersion forces). Here we present a new, simple method for accurately predicting adsorption energies on transition-metal surfaces based on DFT calculations, using an adaptively weighted sum of energies from RPBE and optB86b-vdW (or optB88-vdW) density functionals. This method has been benchmarked against a set of 39 reliable experimental energies for adsorption reactions. Our results show that this method has a mean absolute error and root mean squared error relative to experiments of 13.4 and 19.3 kJ/mol, respectively, compared to 20.4 and 26.4 kJ/mol f...

Molecular dynamics simulation studies of p-xylene on graphene surface: effect of partial charge calculation method on adsorption energies

We report the results of molecular dynamics (MD) simulations of the adsorption of p-xylene on graphene nanosheet in a canonical (NVT) ensemble at 293.15 K. The high adsorption energy is expected to result mainly from the π–π stacking interactions, because of the attraction between the π orbital on the graphene and the electronic density in the p-xylene aromatic rings. In this study, a developed intermolecular force field for the interaction between graphene and p-xylene is used. The improved performance of the developed potential function is shown to reproduce more accurate adsorption energies. Molecular dynamics simulations are also accomplished for investigating the wettability of the graphene surface by the organic liquid. Our findings denote a strong adhesion between the surface and the p-xylene liquid, and the low contact angle obtained θ = 49.16°, indicates strong liquid–surface intermolecular interactions and reflects the high wettability of the graphene surface by the liquid droplet, and thus increases the tendency of the liquid to spread over a large area of the surface.

Adsorption of glucose, cellobiose, and cellotetraose onto cellulose model surfaces.

Reliable simulation of molecular adsorption onto cellulose surfaces is essential for the design of new cellulose nanocomposite materials. However, the applicability of classical force field methods to such systems remains relatively unexplored. In this study, we present the adsorption of glucose, cellobiose, and cellotetraose on model surfaces of crystalline cellulose Iα and Iβ. The adsorption of the two large carbohydrates was simulated with the GLYCAM06 force field. To validate this approach, quantum theoretical calculations for the adsorption of glucose were performed: Equilibrium geometries were studied with density functional theory (DFT) and dispersion-corrected DFT, whereas the adsorption energies were calculated with two standard density functional approximations and five dispersion-containing DFT approaches. We find that GLYCAM06 gives a good account of geometries and, in most cases, accurate adsorption energies when compared to dispersion-corrected DFT energies. Adsorption onto the (100) surface of cellulose Iα is, in general, stronger than onto the (100) surface of cellulose Iβ. Contrary to intuition, the adsorption energy is not directly correlated with the number of hydrogen bonds; rather, it is dominated by dispersion interactions. Especially for bigger adsorbates, a neglect of these interactions leads to a dramatic underestimation of adsorption energies.

Band-Filling Correction Method for Accurate Adsorption Energy Calculations: A Cu/ZnO Case Study

We present a simple method, the "band-filling correction", to calculate accurate adsorption energies (Eads) in the low coverage limit from finite-size supercell slab calculations using DFT. We show that it is necessary to use such a correction if charge transfer takes place between the adsorbate and the substrate, resulting in the substrate bands either filling up or becoming depleted. With this correction scheme, we calculate Eads of an isolated Cu atom adsorbed on the ZnO(101̅0) surface. Without the correction, the calculated Eads is highly coverage-dependent, even for surface supercells that would typically be considered very large (in the range from 1 nm × 1 nm to 2.5 nm × 2.5 nm). The correction scheme works very well for semilocal functionals, where the corrected Eads is converged within 0.01 eV for all coverages. The correction scheme also works well for hybrid functionals if a large supercell is used and the exact exchange interaction is screened.