Development of the ChIMES Force Field for Reactive Molecular Systems: Carbon Monoxide at Extreme Conditions

Lindsey, Rebecca; Goldman, Nir; Fried, Laurence E.; Bastea, Sorin

doi:10.26434/chemrxiv.7940810

Cited by 4 publications

(13 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…MD simulations can supply an atomistically resolved view of this process but are challenging since relevant scales span over three orders of magnitude: from the molecular scale, where bond formation and breaking events occur, to the condensed domains scale, where formation and coarsening dynamics determine the system evolution. To overcome this, we have developed a reactive force field capable of first-principles accuracy for systems under extreme conditions 35,36 (including liquid carbon 35 ), which is computationally efficient and enables multi-million atom simulations that can reach nanosecond timescales (see "Methods" section for further details).…”

Section: 2mentioning

confidence: 99%

“…Model for Efficient Simulation (ChIMES) 35,36 was employed for simulations in the present work. ChIMES is a generalized many-body classical reactive force-field constructed from linear combinations of Chebyshev polynomials, machine learned to Density Functional Theory (DFT) 45 molecular dynamics simulations.…”

Section: Molecular Dynamics Simulations the Recently Developed Chebymentioning

confidence: 99%

“…ChIMES is a generalized many-body classical reactive force-field constructed from linear combinations of Chebyshev polynomials, machine learned to Density Functional Theory (DFT) 45 molecular dynamics simulations. The substantial flexibility afforded through the use of a polynomial basis enables ChIMES to serve as a high-fidelity proxy for DFT while providing a four-orders-ofmagnitude increase in simulation efficiency and linear system size scalability 35,36 . The ChIMES CO model used herein contains explicit two-and three-body interactions and has been shown to yield good agreement with the DFT-calculated structure, dynamics, and energetics at the conditions studied in this work 35 .…”

Section: Molecular Dynamics Simulations the Recently Developed Chebymentioning

confidence: 99%

See 2 more Smart Citations

Ultrafast shock synthesis of nanocarbon from a liquid precursor

et al. 2020

Self Cite

View full text Add to dashboard Cite

Carbon nanoallotropes are important nanomaterials with unusual properties and promising applications. High pressure synthesis has the potential to open new avenues for controlling and designing their physical and chemical characteristics for a broad range of uses but it remains little understood due to persistent conceptual and experimental challenges, in addition to fundamental physics and chemistry questions that are still unresolved after many decades. Here we demonstrate sub-nanosecond nanocarbon synthesis through the application of laser-induced shock-waves to a prototypical organic carbon-rich liquid precursorliquid carbon monoxide. Overlapping large-scale molecular dynamics simulations capture the atomistic details of the nanoparticles' formation and evolution in a reactive environment and identify classical evaporation-condensation as the mechanism governing their growth on these time scales.

show abstract

Section: 2mentioning

confidence: 99%

Section: Molecular Dynamics Simulations the Recently Developed Chebymentioning

confidence: 99%

Section: Molecular Dynamics Simulations the Recently Developed Chebymentioning

confidence: 99%

See 1 more Smart Citation

Ultrafast shock synthesis of nanocarbon from a liquid precursor

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…As will be described in greater detail below, ChIMES force fields are fit to forces (and optionally energies and stresses) arising from Kohn-Sham density functional theory (DFT) simulations of 2-20 ps and have typically leveraged 2+3-body interaction terms and an iterative refinement scheme, where frames from molecular dynamics (MD) simulations with the i th ChIMES force field are occasionally sent back to DFT for single point calculation and combined with the training repository, from which an i + 1 th ChIMES model is generated; the cycle is repeated until desired model performance is achieved. This iterative approach has worked well for molten carbon 22 , ambient water 25 , and dissociative carbon monoxide 26 , where in all cases species were small and chemistry was rapid when present. However, upon application to systems in which larger and more complex species form, shortcomings arising from use of a 2+3-body ChIMES many-body truncation have been identified 26 .…”

Section: A the Chimes Force Fieldmentioning

confidence: 99%

“…This iterative approach has worked well for molten carbon 22 , ambient water 25 , and dissociative carbon monoxide 26 , where in all cases species were small and chemistry was rapid when present. However, upon application to systems in which larger and more complex species form, shortcomings arising from use of a 2+3-body ChIMES many-body truncation have been identified 26 . Though the ChIMES equations can be easily extended to include higherbodied interactions, the resulting increase in model complexity necessitates intelligent and automated model development tools.…”

Section: A the Chimes Force Fieldmentioning

confidence: 99%

Active Learning for Robust, High-Complexity Reactive Atomistic Simulations

Lindsey

Fried²,

Goldman³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Machine learned reactive force fields based on polynomial expansions have been shown to be highly effective for describing simulations involving reactive materials. Nevertheless, the highly flexible nature of these models can give rise to a large number of candidate parameters for complicated systems. In these cases, reliable parameterization requires a well-formed training set, which can be difficult to achieve through standard iterative fitting methods. Here we present an active learning approach based on cluster analysis and Shannon information theory to enable semi-automated generation of informative training sets and robust machine learned force fields. Use of this tool is demonstrated for development of a model based on linear combinations of Chebyshev polynomials explicitly describing up to four-body interactions, for a chemically and structurally diverse system of C/O under extreme conditions. We show that this flexible training repository management approach enables development of models exhibiting excellent agreement with Kohn–Sham density functional theory (DFT) in terms of structure, dynamics, and speciation.

show abstract

Chemistry-Mediated Ostwald Ripening in Carbon-Rich C/O Systems at Extreme Conditions

Lindsey

Goldman

Fried

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

There is significant interest in establishing a capability for tailored synthesis of next-generation carbon-based nanomaterials due to their broad range of applications and high degree of tunability. High pressure (e.g. shockwave-driven) synthesis holds promise as an effective discovery method, but experimental challenges preclude elucidating the processes governing nanocarbon production from carbon-rich precursors that could otherwise guide efforts through the prohibitively expansive design space. Here we report findings from large scale atomistically-resolved simulations of carbon condensation from C/O mixtures subjected to extreme pressures and temperatures, made possible by machine-learned reactive interatomic potentials. We find that liquid nanocarbon formation follows classical growth kinetics driven by Ostwald ripening (i.e. growth of large clusters at the expense of shrinking small ones) and obeys dynamical scaling in a process mediated by carbon chemistry in the surrounding reactive fluid. The results provide direct insight into carbon condensation in a representative system and pave the way for its exploration in higher complexity organic materials. They also suggest that simulations using machine-learned interatomic potentials could eventually be employed as in-silico design tools for new nanomaterials.

show abstract

Development of the ChIMES Force Field for Reactive Molecular Systems: Carbon Monoxide at Extreme Conditions

Cited by 4 publications

References 32 publications

Ultrafast shock synthesis of nanocarbon from a liquid precursor

Ultrafast shock synthesis of nanocarbon from a liquid precursor

Active Learning for Robust, High-Complexity Reactive Atomistic Simulations

Chemistry-Mediated Ostwald Ripening in Carbon-Rich C/O Systems at Extreme Conditions

Contact Info

Product

Resources

About