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

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Victor Fung ◽  
Guoxiang Hu ◽  
P. Ganesh ◽  
Bobby G. Sumpter
Keyword(s):  
Density Of States ◽  
Density Functional ◽  
Heterogeneous Catalysts ◽  
Computational Cost ◽  
Absolute Error ◽  
Electronic Density ◽  
Functional Theory ◽  
Energy Prediction ◽  
Computational Screening ◽  
Machine Learning Model

AbstractMaterials 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.

BAND NN: A Deep Learning Framework For Energy Prediction and Geometry Optimization of Organic Small Molecules

2019 ◽  
Author(s):  
Siddhartha Laghuvarapu ◽  
Yashaswi Pathak ◽  
U. Deva Priyakumar
Keyword(s):  
Machine Learning ◽  
Density Functional ◽  
Computational Cost ◽  
Geometry Optimization ◽  
Dft Methods ◽  
Energy Prediction ◽  
Machine Learning Model ◽  
Equilibrium Structures ◽  
High Level ◽  
Non Equilibrium

Recent advances in artificial intelligence along with development of large datasets of energies calculated using quantum mechanical (QM)/density functional theory (DFT) methods have enabled prediction of accurate molecular energies at reasonably low computational cost. However, machine learning models that have been reported so far requires the atomic positions obtained from geometry optimizations using high level QM/DFT methods as input in order to predict the energies, and do not allow for geometry optimization. In this paper, a transferable and molecule-size independent machine learning model (BAND NN) based on a chemically intuitive representation inspired by molecular mechanics force fields is presented. The model predicts the atomization energies of equilibrium and non-equilibrium structures as sum of energy contributions from bonds (B), angles (A), nonbonds (N) and dihedrals (D) at remarkable accuracy. The robustness of the proposed model is further validated by calculations that span over the conformational, configurational and reaction space. The transferability of this model on systems larger than the ones in the dataset is demonstrated by performing calculations on select large molecules. Importantly, employing the BAND NN model, it is possible to perform geometry optimizations starting from non-equilibrium structures along with predicting their energies.

scholarly journals Towards a density functional theory of molecular fragments. What is the shape of atoms in molecules?

2020 ◽  
Vol 44 (170) ◽  
pp. 269-279
Author(s):  
Victor H. Chávez ◽  
Adam Wasserman
Keyword(s):  
Density Functional Theory ◽  
Schrödinger Equation ◽  
Schrodinger Equation ◽  
Density Functional ◽  
Computational Cost ◽  
Atoms In Molecules ◽  
Electronic Density ◽  
Functional Theory ◽  
New Methods ◽  
The Cost

In some sense, quantum mechanics solves all the problems in chemistry: The only thing one has to do is solve the Schrödinger equation for the molecules of interest. Unfortunately, the computational cost of solving this equation grows exponentially with the number of electrons and for more than ~100 electrons, it is impossible to solve it with chemical accuracy (~ 2 kcal/mol). The Kohn-Sham (KS) equations of density functional theory (DFT) allow us to reformulate the Schrödinger equation using the electronic probability density as the central variable without having to calculate the Schrödinger wave functions. The cost of solving the Kohn-Sham equations grows only as N3, where N is the number of electrons, which has led to the immense popularity of DFT in chemistry. Despite this popularity, even the most sophisticated approximations in KS-DFT result in errors that limit the use of methods based exclusively on the electronic density. By using fragment densities (as opposed to total densities) as the main variables, we discuss here how new methods can be developed that scale linearly with N while providing an appealing answer to the subtitle of the article: What is the shape of atoms in molecules?

Quasicontinuum-Like Reduction of Density Functional Theory Calculations of Nanostructures

2008 ◽  
Vol 8 (7) ◽  
pp. 3729-3740
Author(s):  
Dan Negrut ◽  
Mihai Anitescu ◽  
Anter El-Azab ◽  
Peter Zapol
Keyword(s):  
Density Functional Theory ◽  
Density Functional ◽  
Computational Cost ◽  
Density Functional Theory Calculations ◽  
Optimization Techniques ◽  
Electronic Density ◽  
Functional Theory ◽  
Cross Domain ◽  
Explicit Consideration ◽  
Orbital Free

Density functional theory can accurately predict chemical and mechanical properties of nanostructures, although at a high computational cost. A quasicontinuum-like framework is proposed to substantially increase the size of the nanostructures accessible to simulation. It takes advantage of the near periodicity of the atomic positions in some regions of nanocrystalline materials to establish an interpolation scheme for the electronic density in the system. The electronic problem embeds interpolation and coupled cross-domain optimization techniques through a process called electronic reconstruction. For the optimization of nuclei positions, computational gains result from explicit consideration of a reduced number of representative nuclei and interpolating the positions of the rest of nuclei following the quasicontinuum paradigm. Numerical tests using the Thomas-Fermi-Dirac functional demonstrate the validity of the proposed framework within the orbital-free density functional theory.

scholarly journals Prediction of Homolytic Bond Dissociation Enthalpies for Organic Molecules at near Chemical Accuracy with Sub-Second Computational Cost

2019 ◽  
Author(s):  
Peter St. John ◽  
Yanfei Guan ◽  
Yeonjoon Kim ◽  
Seonah Kim ◽  
Robert Paton
Keyword(s):  
Density Functional ◽  
Organic Molecules ◽  
Soot Formation ◽  
Chemical Reactivity ◽  
Hydrogen Abstraction ◽  
Computational Cost ◽  
Absolute Error ◽  
Functional Theory ◽  
Small Organic Molecules ◽  
Bond Dissociation

Bond dissociation enthalpies (BDEs) of organic molecules play a fundamental role in determining chemical reactivity. However, BDE computations at sufficiently high levels of quantum mechanical (QM) theory require substantial computing resources. We have therefore developed A machine-Learning derived, Fast, Accurate Bond dissociation Enthalpy Tool (ALFABET), capable of accurately predicting BDEs for organic molecules in a fraction of a second. Automated density functional theory (DFT) calculations at the M06-2X/def2-TZVP level of theory were performed for 42,577 small organic molecules, resulting in a dataset of 290,664 BDEs. A graph neural network was trained on a subset of these results, achieving a mean absolute error of 0.58 kcal/mol for the BDE values of unseen molecules. An interface for the developed prediction tool is available online at https://ml.nrel.gov/bde. The model rapidly and accurately predicts major sites of hydrogen abstraction in metabolism of drug-like molecules and determines the dominant molecular fragmentation pathways during soot formation.

BAND NN: A Deep Learning Framework For Energy Prediction and Geometry Optimization of Organic Small Molecules

2019 ◽  
Author(s):  
Siddhartha Laghuvarapu ◽  
Yashaswi Pathak ◽  
U. Deva Priyakumar
Keyword(s):  
Machine Learning ◽  
Density Functional ◽  
Computational Cost ◽  
Geometry Optimization ◽  
Dft Methods ◽  
Energy Prediction ◽  
Machine Learning Model ◽  
Equilibrium Structures ◽  
High Level ◽  
Non Equilibrium

Recent advances in artificial intelligence along with development of large datasets of energies calculated using quantum mechanical (QM)/density functional theory (DFT) methods have enabled prediction of accurate molecular energies at reasonably low computational cost. However, machine learning models that have been reported so far requires the atomic positions obtained from geometry optimizations using high level QM/DFT methods as input in order to predict the energies, and do not allow for geometry optimization. In this paper, a transferable and molecule-size independent machine learning model (BAND NN) based on a chemically intuitive representation inspired by molecular mechanics force fields is presented. The model predicts the atomization energies of equilibrium and non-equilibrium structures as sum of energy contributions from bonds (B), angles (A), nonbonds (N) and dihedrals (D) at remarkable accuracy. The robustness of the proposed model is further validated by calculations that span over the conformational, configurational and reaction space. The transferability of this model on systems larger than the ones in the dataset is demonstrated by performing calculations on select large molecules. Importantly, employing the BAND NN model, it is possible to perform geometry optimizations starting from non-equilibrium structures along with predicting their energies.

scholarly journals The Effect of Point Defects on the Electronic Density of States of ScMN2-Type (M = V, Nb, Ta) Phases

2019 ◽  
Vol 4 (3) ◽  
pp. 70 ◽  
Author(s):  
Robert Pilemalm ◽  
Sergei Simak ◽  
Per Eklund
Keyword(s):  
Density Of States ◽  
Point Defects ◽  
First Principles ◽  
Density Functional ◽  
First Principles Calculations ◽  
Electronic Density ◽  
Functional Theory ◽  
Occupied State ◽  
Narrow Bandgap ◽  
Bandgap Semiconductors

ScMN2-type (M = V, Nb, Ta) phases are layered materials that have been experimentally reported for M = Ta and Nb. They are narrow-bandgap semiconductors with potentially interesting thermoelectric properties. Point defects such as dopants and vacancies largely affect these properties, motivating the need to investigate these effects. In particular, asymmetric peak features in the density of states (DOS) close to the highest occupied state is expected to increase the Seebeck coefficient. Here, we used first principles calculations to study the effects of one vacancy or one C, O, or F dopant on the DOS of the ScMN2 phases. We used density functional theory to calculate formation energy and the density of states when a point defect is introduced in the structures. In the DOS, asymmetric peak features close to the highest occupied state were found as a result of having a vacancy in all three phases. Furthermore, one C dopant in ScTaN2, ScNbN2, and ScVN2 implies a shift of the highest occupied state into the valence band, while one O or F dopant causes a shift of the highest occupied state into the conduction band.

scholarly journals Prediction of Homolytic Bond Dissociation Enthalpies for Organic Molecules at near Chemical Accuracy with Sub-Second Computational Cost

2019 ◽  
Author(s):  
Peter St. John ◽  
Yanfei Guan ◽  
Yeonjoon Kim ◽  
Seonah Kim ◽  
Robert Paton
Keyword(s):  
Density Functional ◽  
Organic Molecules ◽  
Soot Formation ◽  
Chemical Reactivity ◽  
Hydrogen Abstraction ◽  
Computational Cost ◽  
Absolute Error ◽  
Functional Theory ◽  
Small Organic Molecules ◽  
Bond Dissociation

Bond dissociation enthalpies (BDEs) of organic molecules play a fundamental role in determining chemical reactivity. However, BDE computations at sufficiently high levels of quantum mechanical (QM) theory require substantial computing resources. We have therefore developed A machine-Learning derived, Fast, Accurate Bond dissociation Enthalpy Tool (ALFABET), capable of accurately predicting BDEs for organic molecules in a fraction of a second. Automated density functional theory (DFT) calculations at the M06-2X/def2-TZVP level of theory were performed for 42,577 small organic molecules, resulting in a dataset of 290,664 BDEs. A graph neural network was trained on a subset of these results, achieving a mean absolute error of 0.58 kcal/mol for the BDE values of unseen molecules. An interface for the developed prediction tool is available online at https://ml.nrel.gov/bde. The model rapidly and accurately predicts major sites of hydrogen abstraction in metabolism of drug-like molecules and determines the dominant molecular fragmentation pathways during soot formation.

scholarly journals Prediction of Homolytic Bond Dissociation Enthalpies for Organic Molecules at near Chemical Accuracy with Sub-Second Computational Cost

2019 ◽  
Author(s):  
Peter St. John ◽  
Yanfei Guan ◽  
Yeonjoon Kim ◽  
Seonah Kim ◽  
Robert Paton
Keyword(s):  
Density Functional ◽  
Organic Molecules ◽  
Soot Formation ◽  
Chemical Reactivity ◽  
Hydrogen Abstraction ◽  
Computational Cost ◽  
Absolute Error ◽  
Functional Theory ◽  
Small Organic Molecules ◽  
Bond Dissociation

Bond dissociation enthalpies (BDEs) of organic molecules play a fundamental role in determining chemical reactivity. However, BDE computations at sufficiently high levels of quantum mechanical (QM) theory require substantial computing resources. We have therefore developed A machine-Learning derived, Fast, Accurate Bond dissociation Enthalpy Tool (ALFABET), capable of accurately predicting BDEs for organic molecules in a fraction of a second. Automated density functional theory (DFT) calculations at the M06-2X/def2-TZVP level of theory were performed for 42,577 small organic molecules, resulting in a dataset of 290,664 BDEs. A graph neural network was trained on a subset of these results, achieving a mean absolute error of 0.58 kcal/mol for the BDE values of unseen molecules. An interface for the developed prediction tool is available online at https://ml.nrel.gov/bde. The model rapidly and accurately predicts major sites of hydrogen abstraction in metabolism of drug-like molecules and determines the dominant molecular fragmentation pathways during soot formation.

A First-Principles Computational Comparison of the Aqueous Anatase TiO2 (001) Interface and the Disordered, Fluorinated TiO2 Interface

2018 ◽  
Author(s):  
Kyle Reeves ◽  
Damien Dambournet ◽  
Christel Laberty-Robert ◽  
Rodolphe Vuilleumier ◽  
Mathieu Salanne
Keyword(s):  
Density Of States ◽  
Density Functional ◽  
Water Dissociation ◽  
Water Molecules ◽  
Chemical Doping ◽  
Charge Balance ◽  
Functional Theory ◽  
Adsorption Of Water ◽  
Properties Of Materials ◽  
Computational Comparison

Chemical doping and other surface modifications have been used to engineer the bulk properties of materials, but their influence on the surface structure and consequently the surface chemistry are often unknown. Previous work has been successful in fluorinating anatase TiO<sub>2</sub> with charge balance achieved via the introduction of Ti vacancies rather than the reduction of Ti. Our work here investigates the interface between this fluorinated titanate with cationic vacancies and a<br>monolayer of water via density functional theory based molecular dynamics. We compute the projected density of states for only those atoms at the interface and for those states that fall within 1eV of the Fermi energy for various steps throughout the simulation, and we determine that the<br>variation in this representation of the density of states serves as a reasonable tool to anticipate where surfaces are most likely to be reactive. In particular, we conclude that water dissociation at the surface is the main mechanism that influences the anatase (001) surface whereas the change in<br>the density of states at the surface of the fluorinated structure is influenced primarily through the adsorption of water molecules at the surface.

Sign in / Sign up

Export Citation Format

Share Document