The ReaxFF reactive force-field: development, applications and future directions

The ReaxFF reactive force-field: development, applications and future directions Thomas P. Senftle Sungwook Hong Md Mahbubul Islam Sudhir B. Kylasa Yuanxia Zheng Yun Kyung Shin Chad E. Junkermeier Roman Engel‐Herbert Michael J. Janik Hasan Metin Aktulga Toon Verstraelen Ananth Grama Adri C. T. van Duin

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

The ReaxFF reactive force-field: development, applications and future directions 2016 npj Computational Materials 10.1038/npjcompumats.2015.11 senftle2016-the-reaxff-reactive-force-field-de senftle2016-the-reaxff-reactive-force-field-de.txt, the project's staged copy cecdd4bb8ab7b8866c1ea32cee728cea34e88855 cc-by materials-science

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them the ReaxFF reactive force field - a classical interatomic potential that describes bond breaking and forming through a bond-order formalism, without explicit quantum mechanics Quotation withheld because this record is not cleared for redistribution. how a single functional form can cover reactive chemistry across phases, and what the resulting parameter sets can and cannot be combined with Quotation withheld because this record is not cleared for redistribution. a review of ReaxFF's development, applications and directions - including an explicit account of why its parameter sets fall into two mutually untransferable branches Quotation withheld because this record is not cleared for redistribution. method Quotation withheld because this record is not cleared for redistribution. a method whose defining choice is to give an element ONE description everywhere, and whose parameter sets have consequently forked into two populations that cannot be mixed Quotation withheld because this record is not cleared for redistribution. review review Quotation withheld because this record is not cleared for redistribution. ReaxFF's defining choice is that an element has ONE description regardless of environment, which is what lets an atom migrate between phases mid-simulation. The cost is that a parameter set cannot be specialised without changing the shared atom-level parameters - so the specialisation became a BRANCH instead of a TYPE. REFUSING TO SPLIT THE VOCABULARY SPLIT THE PARAMETER SETS one type, two branches Quotation withheld because this record is not cleared for redistribution. liquid water was not a development aim for the 2008 combustion field because every intended application was above the boiling point. That is a defensible scope decision at the time and it is why the field fails on liquid water in 2009, and why a second branch had to exist the scope decision that became a failure Quotation withheld because this record is not cleared for redistribution. parameter sets on the same branch combine by fitting only the bond and angular terms between the newly combined elements; across branches it takes extensive refitting, and combustion-to-aqueous is the easier direction. THE ASYMMETRY IS A FACT ABOUT THE FITTING HISTORY, not about the chemistry a fork that cannot be undone cheaply Quotation withheld because this record is not cleared for redistribution. several independent branches - B/N/H, Ni/C/H - began as non-transferable descriptions and were later merged into combustion or aqueous through extensive refitting. A vocabulary reconciled by redoing the work rather than by declaring the terms equivalent branches merge by refitting, not by agreement Quotation withheld because this record is not cleared for redistribution. QM structure and energy data train the empirical field, which then runs at scales orders of magnitude beyond what QM can reach. The potential is a compression of the quantum calculation, and every branch is a different compression of a different part of it accuracy traded for scale, on purpose Quotation withheld because this record is not cleared for redistribution. parameter sets cannot be used in any combination and expect satisfactory transferability - there are two branches and they are only intra-transferable Quotation withheld because this record is not cleared for redistribution. the absence of atom typing makes force-field development significantly more complex Quotation withheld because this record is not cleared for redistribution. the 2008 combustion force field fails to describe liquid water, and could not be reparameterised for it without changing general and atom-specific parameters Quotation withheld because this record is not cleared for redistribution. transferring parameters between branches requires extensive refitting, and the direction matters Quotation withheld because this record is not cleared for redistribution. empirical potentials trade accuracy for computational expense by construction Quotation withheld because this record is not cleared for redistribution. the bond-order concept from chemical bonding theory, used as the empirical function of interatomic distance that lets a classical potential break bonds Quotation withheld because this record is not cleared for redistribution. later parameterisations validated against the FULL training sets of the earlier ones they supersede - the 2008-C/H/O set against the entire 2001-C/H training set eliminates: a newer parameter set that is better on its own data and worse on the old - the earlier training set is re-run rather than retired · outcome: held Quotation withheld because this record is not cleared for redistribution. water tested in both phases under the same force field, rather than only in the phase the field was built for eliminates: a force field that looks general because it was only asked about the case it was fitted to - testing the other phase is what exposed the branch · outcome: failed Quotation withheld because this record is not cleared for redistribution. independent branches merged into combustion or aqueous through extensive refitting, with the high-energy-material description reported as merged without an obvious loss in accuracy eliminates: a merge that silently degrades the absorbed branch - the accuracy claim is made explicitly at the point of merging · outcome: held Quotation withheld because this record is not cleared for redistribution. the same element treated with one formalism across gas, liquid and solid, checked by letting atoms migrate between phases within a simulation eliminates: phase boundaries being imposed by the model rather than by the physics · outcome: held Quotation withheld because this record is not cleared for redistribution. other ReaxFF implementations developed outside the originating group, noted as existing eliminates: that a result depends on this particular ReaxFF implementation rather than on the force field itself · outcome (stated by the tagger, no quotation): absent-stated-about-other Quotation withheld because this record is not cleared for redistribution. ReaxFF ReaxFF Quotation withheld because this record is not cleared for redistribution. atom typing atom typing Quotation withheld because this record is not cleared for redistribution. CHARMM CHARMM Quotation withheld because this record is not cleared for redistribution. combustion branch combustion branch Quotation withheld because this record is not cleared for redistribution. aqueous branch aqueous branch Quotation withheld because this record is not cleared for redistribution. transferability transferability Quotation withheld because this record is not cleared for redistribution. gas-phase gas-phase Quotation withheld because this record is not cleared for redistribution. liquid liquid Quotation withheld because this record is not cleared for redistribution. training set training set Quotation withheld because this record is not cleared for redistribution. bond order bond order Quotation withheld because this record is not cleared for redistribution. an oxygen atom - IN ReaxFF's DESIGN: one type, treated with the same mathematical formalism whether it is O2 in gas, water in liquid, or a solid oxide Quotation withheld because this record is not cleared for redistribution. an oxygen atom - IN PRACTICE: described by combustion-branch parameters or by aqueous-branch parameters, which are not directly transferable Quotation withheld because this record is not cleared for redistribution. water - UNDER THE 2008 COMBUSTION FORCE FIELD: accurately described as a gas-phase molecule Quotation withheld because this record is not cleared for redistribution. water - UNDER THE SAME FORCE FIELD, AS A LIQUID: not described at all, because liquid water was not a development aim at temperatures above the boiling point Quotation withheld because this record is not cleared for redistribution. a parameter set - WITHIN ITS BRANCH: directly combinable, needing only bond and angular terms between the newly combined elements Quotation withheld because this record is not cleared for redistribution. a parameter set - ACROSS BRANCHES: requiring extensive refitting, with combustion to aqueous the easier direction Quotation withheld because this record is not cleared for redistribution. distinguishes-from → AMBER and CHARMM, non-reactive force fields that type atoms by chemical environment Quotation withheld because this record is not cleared for redistribution. qualifies → ReaxFF's own transferability claim, by stating that parameter sets fall into two branches that are only intra-transferable Quotation withheld because this record is not cleared for redistribution. builds-on → quantum-mechanical structure and energy data, used as the training data for every parameterisation Quotation withheld because this record is not cleared for redistribution. combustion branch —distinct-from→ aqueous branch combustion branch aqueous branch differs in: which O and H parameters, under an identical functional form · the relation itself is separately evidenced · strength of the relation: 0.95 · differs in (stated by the tagger, no quotation): which O and H parameters, under an identical functional form · the relation stated: the aqueous branch uses the same functional form as the combustion description with different O/H atom and bond parameters, and its parameter sets are not directly transferable with the pre-existing branch Quotation withheld because this record is not cleared for redistribution. ReaxFF —distinct-from→ CHARMM ReaxFF CHARMM differs in: whether an atom's parameters depend on the environment it is in · the relation itself is separately evidenced · strength of the relation: 0.95 · differs in (stated by the tagger, no quotation): whether an atom's parameters depend on the environment it is in · the relation stated: ReaxFF does not type atoms by chemical environment; CHARMM and AMBER do - which is why one oxygen in ReaxFF has to serve gas, liquid and solid at once Quotation withheld because this record is not cleared for redistribution. gas-phase —distinct-from→ liquid gas-phase liquid differs in: which phase the same molecule is in · the relation itself is separately evidenced · strength of the relation: 0.95 · differs in (stated by the tagger, no quotation): which phase the same molecule is in · the relation stated: the 2008 combustion force field describes water accurately as a gas-phase molecule and fails on liquid water, because liquid water was not a development aim at the temperatures intended Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span says the force-field describes water as a gas-phase molecule yet fails for liquid water; distinct-from is loose. Quotation withheld because this record is not cleared for redistribution. there is only one oxygen type in ReaxFF regardless of chemical environment, unlike AMBER and CHARMM there is only one oxygen type in ReaxFF regardless of chemical environment, unlike AMBER and CHARMM Quotation withheld because this record is not cleared for redistribution. parameter sets fall into two branches, combustion and aqueous, that are only intra-transferable parameter sets fall into two branches, combustion and aqueous, that are only intra-transferable Quotation withheld because this record is not cleared for redistribution. the 2008 combustion force field describes water as a gas accurately and fails to describe it as a liquid the 2008 combustion force field describes water as a gas accurately and fails to describe it as a liquid Quotation withheld because this record is not cleared for redistribution. The aqueous branch keeps the 2008-C/H/O functional form with different O/H atom and bond parameters. The aqueous branch keeps the 2008-C/H/O functional form with different O/H atom and bond parameters. Quotation withheld because this record is not cleared for redistribution. parameter sets on the same branch combine by fitting only bond and angular terms; across branches requires extensive refitting parameter sets on the same branch combine by fitting only bond and angular terms; across branches requires extensive refitting Quotation withheld because this record is not cleared for redistribution. independent branches such as B/N/H and Ni/C/H were later merged into combustion or aqueous through extensive refitting independent branches such as B/N/H and Ni/C/H were later merged into combustion or aqueous through extensive refitting Quotation withheld because this record is not cleared for redistribution. an element is treated with the same formalism in gas, liquid and solid, which is what allows atoms to migrate between phases in a simulation an element is treated with the same formalism in gas, liquid and solid, which is what allows atoms to migrate between phases in a simulation Quotation withheld because this record is not cleared for redistribution. an empirical interatomic potential cast within a bond-order formalism, describing chemical bonding implicitly and so allowing reactive events without expensive QM calculations ReaxFF Quotation withheld because this record is not cleared for redistribution. a parameter-set grouping employing the same functional form as the 2008 combustion description but with different O/H atom and bond parameters, not directly transferable with it aqueous branch Quotation withheld because this record is not cleared for redistribution. the C/H/O combustion force field that describes gas-phase water and fails on liquid water 2008 Quotation withheld because this record is not cleared for redistribution. transferability transferability Quotation withheld because this record is not cleared for redistribution. bond-order bond-order Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1038/npjcompumats.2015.11.