Operando modeling of zeolite-catalyzed reactions using first-principles molecular dynamics simulations
- Author
- Veronique Van Speybroeck (UGent) , Massimo Bocus (UGent) , Pieter Cnudde (UGent) and Louis Vanduyfhuys (UGent)
- Organization
- Project
-
- Towards molecular control of electrophilic aromatic substitution reactions in homogeneous and heterogeneous environments through a combined ab initio molecular dynamics and conceptual density functional theory approach.
- Zeolite-induced shape-selectivity for waste-free, highly regioselective catalytic arylation or alkenylation of aromatic C-H bonds
- Bio based factory: Sustainable chemistry from wood
- Exploring strategies to tune the selectivity of the CO2-to-chemicals conversion on zeolite catalysts by molecular simulations
- DYNPOR (First principle molecular dynamics simulations for complex chemical transformations in nanoporous materials)
- Abstract
- Within this Perspective,we critically reflect on the role of first-principlesmolecular dynamics (MD) simulations in unraveling the catalytic functionwithin zeolites under operating conditions. First-principles MD simulationsrefer to methods where the dynamics of the nuclei is followed in timeby integrating the Newtonian equations of motion on a potential energysurface that is determined by solving the quantum-mechanical many-bodyproblem for the electrons. Catalytic solids used in industrial applicationsshow an intriguing high degree of complexity, with phenomena takingplace at a broad range of length and time scales. Additionally, thestate and function of a catalyst critically depend on the operatingconditions, such as temperature, moisture, presence of water, etc.Herein we show by means of a series of exemplary cases how first-principlesMD simulations are instrumental to unravel the catalyst complexityat the molecular scale. Examples show how the nature of reactive speciesat higher catalytic temperatures may drastically change compared tospecies at lower temperatures and how the nature of active sites maydynamically change upon exposure to water. To simulate rare events,first-principles MD simulations need to be used in combination withenhanced sampling techniques to efficiently sample low-probabilityregions of phase space. Using these techniques, it is shown how competitivepathways at operating conditions can be discovered and how broad transitionstate regions can be explored. Interestingly, such simulations canalso be used to study hindered diffusion under operating conditions.The cases shown clearly illustrate how first-principles MD simulationsreveal insights into the catalytic function at operating conditions,which could not be discovered using static or local approaches whereonly a few points are considered on the potential energy surface (PES).Despite these advantages, some major hurdles still exist to fullyintegrate first-principles MD methods in a standard computationalcatalytic workflow or to use the output of MD simulations as inputfor multiple length/time scale methods that aim to bridge to the reactorscale. First of all, methods are needed that allow us to evaluatethe interatomic forces with quantum-mechanical accuracy, albeit ata much lower computational cost compared to currently used densityfunctional theory (DFT) methods. The use of DFT limits the currentlyattainable length/time scales to hundreds of picoseconds and a fewnanometers, which are much smaller than realistic catalyst particledimensions and time scales encountered in the catalysis process. Onesolution could be to construct machine learning potentials (MLPs),where a numerical potential is derived from underlying quantum-mechanicaldata, which could be used in subsequent MD simulations. As such, muchlonger length and time scales could be reached; however, quite someresearch is still necessary to construct MLPs for the complex systemsencountered in industrially used catalysts. Second, most currentlyused enhanced sampling techniques in catalysis make use of collectivevariables (CVs), which are mostly determined based on chemical intuition.To explore complex reactive networks with MD simulations, methodsare needed that allow the automatic discovery of CVs or methods thatdo not rely on a priori definition of CVs. Recently, various data-drivenmethods have been proposed, which could be explored for complex catalyticsystems. Lastly, first-principles MD methods are currently mostlyused to investigate local reactive events. We hope that with the riseof data-driven methods and more efficient methods to describe thePES, first-principles MD methods will in the future also be able todescribe longer length/time scale processes in catalysis. This mightlead to a consistent dynamic description of all steps diffusion,adsorption, and reaction as they take place at the catalystparticle level.
- Keywords
- MACHINE LEARNING POTENTIALS, DENSITY-FUNCTIONAL THEORY, IN-SITU SPECTROSCOPY, AB-INITIO SIMULATION, MONTE-CARLO, FREE-ENERGY, METHYLATION REACTIONS, DIMETHYL ETHER, HETEROGENEOUS CATALYSIS, COLLECTIVE VARIABLES, molecular dynamics, kinetics, diffusion, operating conditions, enhanced sampling, densityfunctional theory, zeolites, catalysis
Downloads
-
acscatal.3c01945.pdf
- full text (Author's original)
- |
- open access
- |
- |
- 8.86 MB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-01HAVX1BRH2MP4EQJY7QEQ0Y70
- MLA
- Van Speybroeck, Veronique, et al. “Operando Modeling of Zeolite-Catalyzed Reactions Using First-Principles Molecular Dynamics Simulations.” ACS CATALYSIS, vol. 13, no. 17, 2023, pp. 11455–93, doi:10.1021/acscatal.3c01945.
- APA
- Van Speybroeck, V., Bocus, M., Cnudde, P., & Vanduyfhuys, L. (2023). Operando modeling of zeolite-catalyzed reactions using first-principles molecular dynamics simulations. ACS CATALYSIS, 13(17), 11455–11493. https://doi.org/10.1021/acscatal.3c01945
- Chicago author-date
- Van Speybroeck, Veronique, Massimo Bocus, Pieter Cnudde, and Louis Vanduyfhuys. 2023. “Operando Modeling of Zeolite-Catalyzed Reactions Using First-Principles Molecular Dynamics Simulations.” ACS CATALYSIS 13 (17): 11455–93. https://doi.org/10.1021/acscatal.3c01945.
- Chicago author-date (all authors)
- Van Speybroeck, Veronique, Massimo Bocus, Pieter Cnudde, and Louis Vanduyfhuys. 2023. “Operando Modeling of Zeolite-Catalyzed Reactions Using First-Principles Molecular Dynamics Simulations.” ACS CATALYSIS 13 (17): 11455–11493. doi:10.1021/acscatal.3c01945.
- Vancouver
- 1.Van Speybroeck V, Bocus M, Cnudde P, Vanduyfhuys L. Operando modeling of zeolite-catalyzed reactions using first-principles molecular dynamics simulations. ACS CATALYSIS. 2023;13(17):11455–93.
- IEEE
- [1]V. Van Speybroeck, M. Bocus, P. Cnudde, and L. Vanduyfhuys, “Operando modeling of zeolite-catalyzed reactions using first-principles molecular dynamics simulations,” ACS CATALYSIS, vol. 13, no. 17, pp. 11455–11493, 2023.
@article{01HAVX1BRH2MP4EQJY7QEQ0Y70, abstract = {{Within this Perspective,we critically reflect on the role of first-principlesmolecular dynamics (MD) simulations in unraveling the catalytic functionwithin zeolites under operating conditions. First-principles MD simulationsrefer to methods where the dynamics of the nuclei is followed in timeby integrating the Newtonian equations of motion on a potential energysurface that is determined by solving the quantum-mechanical many-bodyproblem for the electrons. Catalytic solids used in industrial applicationsshow an intriguing high degree of complexity, with phenomena takingplace at a broad range of length and time scales. Additionally, thestate and function of a catalyst critically depend on the operatingconditions, such as temperature, moisture, presence of water, etc.Herein we show by means of a series of exemplary cases how first-principlesMD simulations are instrumental to unravel the catalyst complexityat the molecular scale. Examples show how the nature of reactive speciesat higher catalytic temperatures may drastically change compared tospecies at lower temperatures and how the nature of active sites maydynamically change upon exposure to water. To simulate rare events,first-principles MD simulations need to be used in combination withenhanced sampling techniques to efficiently sample low-probabilityregions of phase space. Using these techniques, it is shown how competitivepathways at operating conditions can be discovered and how broad transitionstate regions can be explored. Interestingly, such simulations canalso be used to study hindered diffusion under operating conditions.The cases shown clearly illustrate how first-principles MD simulationsreveal insights into the catalytic function at operating conditions,which could not be discovered using static or local approaches whereonly a few points are considered on the potential energy surface (PES).Despite these advantages, some major hurdles still exist to fullyintegrate first-principles MD methods in a standard computationalcatalytic workflow or to use the output of MD simulations as inputfor multiple length/time scale methods that aim to bridge to the reactorscale. First of all, methods are needed that allow us to evaluatethe interatomic forces with quantum-mechanical accuracy, albeit ata much lower computational cost compared to currently used densityfunctional theory (DFT) methods. The use of DFT limits the currentlyattainable length/time scales to hundreds of picoseconds and a fewnanometers, which are much smaller than realistic catalyst particledimensions and time scales encountered in the catalysis process. Onesolution could be to construct machine learning potentials (MLPs),where a numerical potential is derived from underlying quantum-mechanicaldata, which could be used in subsequent MD simulations. As such, muchlonger length and time scales could be reached; however, quite someresearch is still necessary to construct MLPs for the complex systemsencountered in industrially used catalysts. Second, most currentlyused enhanced sampling techniques in catalysis make use of collectivevariables (CVs), which are mostly determined based on chemical intuition.To explore complex reactive networks with MD simulations, methodsare needed that allow the automatic discovery of CVs or methods thatdo not rely on a priori definition of CVs. Recently, various data-drivenmethods have been proposed, which could be explored for complex catalyticsystems. Lastly, first-principles MD methods are currently mostlyused to investigate local reactive events. We hope that with the riseof data-driven methods and more efficient methods to describe thePES, first-principles MD methods will in the future also be able todescribe longer length/time scale processes in catalysis. This mightlead to a consistent dynamic description of all steps diffusion,adsorption, and reaction as they take place at the catalystparticle level.}}, author = {{Van Speybroeck, Veronique and Bocus, Massimo and Cnudde, Pieter and Vanduyfhuys, Louis}}, issn = {{2155-5435}}, journal = {{ACS CATALYSIS}}, keywords = {{MACHINE LEARNING POTENTIALS,DENSITY-FUNCTIONAL THEORY,IN-SITU SPECTROSCOPY,AB-INITIO SIMULATION,MONTE-CARLO,FREE-ENERGY,METHYLATION REACTIONS,DIMETHYL ETHER,HETEROGENEOUS CATALYSIS,COLLECTIVE VARIABLES,molecular dynamics,kinetics,diffusion,operating conditions,enhanced sampling,densityfunctional theory,zeolites,catalysis}}, language = {{eng}}, number = {{17}}, pages = {{11455--11493}}, title = {{Operando modeling of zeolite-catalyzed reactions using first-principles molecular dynamics simulations}}, url = {{http://doi.org/10.1021/acscatal.3c01945}}, volume = {{13}}, year = {{2023}}, }
- Altmetric
- View in Altmetric
- Web of Science
- Times cited: