Automated exploration of the low-energy chemical space with fast quantum chemical methods

Physical Chemistry Chemical Physics, Journal Year: 2020, Volume and Issue: 22(14), P. 7169 - 7192

Published: Jan. 1, 2020

We propose and discuss an efficient scheme for thein silicosampling parts of the molecular low-energy chemical space by semiempirical tight-binding methods combined with a meta-dynamics driven search algorithm.

Language: Английский

---

Exploration of Chemical Compound, Conformer, and Reaction Space with Meta-Dynamics Simulations Based on Tight-Binding Quantum Chemical Calculations

Journal of Chemical Theory and Computation, Journal Year: 2019, Volume and Issue: 15(5), P. 2847 - 2862

Published: April 3, 2019

The semiempirical tight-binding based quantum chemistry method GFN2-xTB is used in the framework of meta-dynamics (MTD) to globally explore chemical compound, conformer, and reaction space. biasing potential given as a sum Gaussian functions expressed with root-mean-square-deviation (RMSD) Cartesian space metric for collective variables. This choice makes approach robust generally applicable three common problems (i.e., conformer search, exploration virtual nanoreactor, guessing paths). Because inherent locality atomic RMSD, functional group or fragment selective treatments are possible facilitating investigation catalytic processes where, example, only substrate thermally activated. Due approximate character method, resulting structure ensembles require further refinement more sophisticated, density wave function theory methods. However, extremely efficient running routinely on laptop computers minutes hours computation time even realistically sized molecules few hundred atoms. Furthermore, underlying energy surface containing almost all elements ( Z = 1-86) consistent including covalent dissociation process electronically complicated situations in, transition metal systems. As examples, thermal decomposition, ethyne oligomerization, oxidation hydrocarbons (by oxygen P450 enzyme model), Miller-Urey model system, forbidden dimerization, multistep intramolecular cyclization shown. For typical conformational search organic drug molecules, new MTD(RMSD) algorithm yields lower structures complete at reduced computational effort compared its already well performing predecessor.

Language: Английский

---

From DFT to machine learning: recent approaches to materials science–a review

Journal of Physics Materials, Journal Year: 2019, Volume and Issue: 2(3), P. 032001 - 032001

Published: Feb. 19, 2019

Abstract Recent advances in experimental and computational methods are increasing the quantity complexity of generated data. This massive amount raw data needs to be stored interpreted order advance materials science field. Identifying correlations patterns from large amounts complex is being performed by machine learning algorithms for decades. Recently, community started invest these methodologies extract knowledge insights accumulated review follows a logical sequence starting density functional theory as representative instance electronic structure methods, subsequent high-throughput approach, used generate Ultimately, data-driven strategies which include mining, screening, techniques, employ generated. We show how approaches modern uncover complexities design novel with enhanced properties. Finally, we point present research problems, challenges, potential future perspectives this new exciting

Language: Английский

---

Microkinetic Modeling: A Tool for Rational Catalyst Design

Chemical Reviews, Journal Year: 2020, Volume and Issue: 121(2), P. 1049 - 1076

Published: Nov. 18, 2020

The design of heterogeneous catalysts relies on understanding the fundamental surface kinetics that controls catalyst performance, and microkinetic modeling is a tool can help researcher in streamlining process design. Microkinetic used to identify critical reaction intermediates rate-determining elementary reactions, thereby providing vital information for designing an improved catalyst. In this review, we summarize general procedures developing models using parameters obtained from experimental data, theoretical correlations, quantum chemical calculations. We examine methods required ensure thermodynamic consistency model. describe parameter adjustments account heterogeneity inherent errors estimation. discuss analysis determine reactions degree rate control reversibility each reaction. introduce incorporation Brønsted–Evans–Polanyi relations scaling effects these catalytic performance formation volcano curves are discussed. review schemes terms maximum outline procedure kinetically significant transition states adsorbed intermediates. explore application generalized expressions prediction optimal binding energies important estimate extent potential improvement. also homogeneous catalysis, electro-catalysis, transient kinetics. conclude by highlighting challenges opportunities

Language: Английский

---

Progress in Accurate Chemical Kinetic Modeling, Simulations, and Parameter Estimation for Heterogeneous Catalysis

ACS Catalysis, Journal Year: 2019, Volume and Issue: 9(8), P. 6624 - 6647

Published: June 13, 2019

Chemical kinetic modeling in heterogeneous catalysis is advancing its ability to provide qualitatively or even quantitatively accurate prediction of real-world behavior because new advances the physical and chemical representations catalytic systems, estimation relevant parameters, capabilities modeling. This Perspective describes current trends future areas advancement modeling, simulation, parameter estimation: ranging from elementary step calculations multiscale role advanced statistical methods for incorporating uncertainties predictions. Multiple growing methodologies are covered, examples provided, forward-looking topics noted.

Language: Английский

---

Automated in Silico Design of Homogeneous Catalysts

ACS Catalysis, Journal Year: 2020, Volume and Issue: 10(3), P. 2354 - 2377

Published: Jan. 17, 2020

Catalyst discovery is increasingly relying on computational chemistry, and many of the tools are currently being automated. The state this automation degree to which it may contribute speeding up development catalysts subject Perspective. We also consider main challenges associated with automated catalyst design, in particular generation promising chemically realistic candidates, tradeoff between accuracy cost estimating catalytic performance, opportunities use large amounts data, even how define objectives design. Throughout Perspective, we take a cross-disciplinary approach evaluate potential methods experiences from fields other than homogeneous catalysis. Finally, provide an overview software packages available for silico design catalysts.

Language: Английский

---

Autonomous Discovery in the Chemical Sciences Part I: Progress

Angewandte Chemie International Edition, Journal Year: 2019, Volume and Issue: 59(51), P. 22858 - 22893

Published: Sept. 25, 2019

This two-part review examines how automation has contributed to different aspects of discovery in the chemical sciences. In this first part, we describe a classification for discoveries physical matter (molecules, materials, devices), processes, and models they are unified as search problems. We then introduce set questions considerations relevant assessing extent autonomy. Finally, many case studies accelerated by or resulting from computer assistance domains synthetic chemistry, drug discovery, inorganic materials science. These illustrate rapid advancements hardware machine learning continue transform nature experimentation modelling. Part two reflects on these identifies open challenges field.

Language: Английский

---

Machine learning in chemical reaction space

Nature Communications, Journal Year: 2020, Volume and Issue: 11(1)

Published: Oct. 30, 2020

Chemical compound space refers to the vast set of all possible chemical compounds, estimated contain 1060 molecules. While intractable as a whole, modern machine learning (ML) is increasingly capable accurately predicting molecular properties in important subsets. Here, we therefore engage ML-driven study even larger reaction space. Central chemistry science transformations, this contains reactions. As an basis for 'reactive' ML, establish first-principles database (Rad-6) containing closed and open-shell organic molecules, along with associated energies (Rad-6-RE). We show that special topology spaces, central hub molecules involved multiple reactions, requires modification existing ML-concepts. Showcased by application methane combustion, demonstrate learned offer non-empirical route rationally extract reduced networks detailed microkinetic analyses.

Language: Английский

---

Combustion chemistry in the twenty-first century: Developing theory-informed chemical kinetics models

Progress in Energy and Combustion Science, Journal Year: 2021, Volume and Issue: 83, P. 100886 - 100886

Published: Jan. 13, 2021

Language: Английский

---

Predictive chemistry: machine learning for reaction deployment, reaction development, and reaction discovery

Chemical Science, Journal Year: 2022, Volume and Issue: 14(2), P. 226 - 244

Published: Nov. 28, 2022

This review outlines several organic chemistry tasks for which predictive machine learning models have been and can be applied.

Language: Английский

---