Accelerated molecular dynamics with the bond-boost method
Summary¶
Rare events such as surface diffusion hops, vacancy motion, and exchange mechanisms on metals often occur on time scales far longer than brute-force molecular dynamics can sample at experimentally relevant temperatures. This Journal of Chemical Physics article introduces bond-boost accelerated molecular dynamics (AMD), an empirical scheme that increases transition rates near potential-energy minima by applying a boost to the potential energy that depends on how far selected bond lengths deviate from their equilibrium values. The method does not require a predefined catalog of events; instead, it uses local bond distortions to bias dynamics toward escape from basins. The authors motivate the approach as a practical way to reach millisecond-scale phenomena while retaining an atomistic embedded-atom method (EAM) description of copper. The paper positions bond boost within the broader accelerated-dynamics family and focuses validation on copper surface processes where direct MD is prohibitively slow.
Methods¶
Method development (bond-boost accelerated MD)¶
The paper introduces bond-boost accelerated molecular dynamics (AMD): near local minima the potential energy surface is modified by adding a boost potential that grows with deviations of selected bond lengths from equilibrium, increasing transition rates while aiming to preserve relative rates among competing pathways when parameters are chosen appropriately (J. Chem. Phys. 119, 6210–6216 (2003); papers/Others/Miron_Fichthorn_Bond_Boost_2003.pdf). The approach needs no predefined event list and can pair with many-body classical potentials.
MD application (Cu(100) EAM benchmarks)¶
Engine / code: Molecular dynamics on Cu(100) using an embedded-atom method (EAM) potential (abstract). System & composition: Cu slab/supercell setups for adatom, vacancy, exchange, and dimer diffusion processes enumerated in the abstract (atom counts and supercell vectors in Methods). PBC: three-dimensional PBC for the surface cells as standard for slab MD (exact slab thickness / fixed layers in PDF). Ensemble / thermostat / timestep / duration: NVE/NVT assignments and timestep choices for the boosted MD stages are given in §§II–III of the article (N/A — numeric thermostat damping not transcribed from the indexed excerpt normalized/extracts/2018miron-the-journal-accelerated-molecular_p1-2.txt). Temperature: low-temperature surface diffusion regime emphasized in the Introduction/abstract framing (exact K in Methods). Barostat / pressure: N/A — not highlighted in the excerpted abstract (constant-volume surface MD typical; confirm in PDF). Electric field: N/A — not used. Enhanced sampling: bond-boost AMD is the paper’s enhanced-dynamics method (not umbrella/replica exchange).
Comparisons reported in the article¶
The authors compare boosted trajectories to brute-force MD and transition-state theory expectations for the same Cu(100) processes to assess whether relative rates remain faithful under boost.
Findings¶
For the Cu(100) test cases, bond-boost AMD reproduces the correct relative rates for adatom hopping, exchange, vacancy and dimer diffusion, and related processes summarized in the abstract, with speedups up to several orders of magnitude vs brute-force MD. The work argues local bond-length-based boosting can suffice for metallic surface diffusion without prelisting events, while warning that boost parameters must be tuned to avoid distorting relative kinetics among competing pathways.
Comparisons / validation. Validation is against direct MD and TST expectations for the same EAM Cu surface processes as described in the article.
Corpus honesty. This page corrects an earlier scaffold typo: the study uses EAM Cu MD, not ReaxFF; numerical boost settings live in the 2003 JCP PDF.
Limitations¶
Boost shape and strength are empirical knobs; transferability to different materials, reactive chemistry, or strongly anharmonic systems requires separate validation. The method inherits limitations of the underlying classical potential and of the assumption that relative transition rates remain meaningful under boost.
Relevance to group¶
Reference algorithm for rare-event atomistic simulation that can complement LAMMPS workflows and reactive studies where enhanced sampling is needed.