Approximate photochemical dynamics of azobenzene with reactive force fields

Evidence and attribution¶

Authority of statements

Sections below summarize the publication identified by doi, title, and pdf_path in the front matter.

Summary¶

Fits ReaxFF-type potentials separately to ground and first excited electronic states of azobenzene using genetic-algorithm global optimization. Coupled to a simple energy-gap transition probability model, the authors report force-field-only trajectories that reproduce cis↔trans photoisomerization with qualitatively reasonable quantum yields, aimed at enabling large-scale photochemical MD of molecular machines where both bond rearrangement and optical excitation matter. The approach sidesteps full surface-hopping on ab initio surfaces while retaining empirical reactive flexibility.

Methods¶

1 — MD application (atomistic dynamics). Photodynamics integrates S₁ dynamics using Langevin molecular dynamics with a 0.1 fs timestep, coupled to a simple energy-gap transition-rate model (κ, χ parameters defined in the article; normalized/extracts/2013hartke-gnuplot-plot-exp-eps_p1-2.txt is abstract-only—confirm details in pdf_path). System size & composition: gas-phase azobenzene (order 10¹ atoms). Boundaries / periodicity: N/A — isolated-molecule photodynamics as described in J. Chem. Phys. 139, 224303 (2013); confirm whether any PBC appears in pdf_path. Ensemble: Langevin implies stochastic thermal coupling (treat as NVT-like sampling of S₁; confirm target temperature in pdf_path). Timestep: 0.1 fs (abstract). Duration / stages: 2000 independent trajectories per cis→trans and trans→cis direction reported in the abstract (full segment lengths in pdf_path). Thermostat: Langevin thermostat coupling as stated. Barostat: N/A — not used. Temperature: N/A — explicit K thermostat target not in the indexed excerpt (read pdf_path). Pressure: N/A —. Electric field: N/A — photoexcitation handled via the energy-gap model rather than a static field bias. Replica / enhanced sampling: N/A — not reported.

2 — Force-field training. Parent FF / elements: ReaxFF-type reactive bond-order forms for azobenzene on ground (S₀) and first excited (S₁) surfaces. QM reference: FOCI-AM1 via development MOPAC for reference energies and gradients (Section 2 in the article). DFT is not the training QM engine here (N/A — plane-wave DFT benchmarks are outside the stated FOCI-AM1 protocol). Training set: S₀ and S₁ PES sampling targets for azobenzene as described in pdf_path. Optimization: genetic algorithm (GA) global search with hybrid global/local moves tailored to the underdetermined many-parameter ReaxFF-type form. Reference data / validation: quantum yields from the force-field + gap workflow are compared to reference Langevin simulations at the FOCI-AM1 level (abstract claim).

3 — Static QM / DFT. N/A — FOCI-AM1 semiempirical QM is the reference model rather than plane-wave DFT in this work.

Findings¶

1 — Outcomes & mechanisms. Coupling fitted S₀/S₁ ReaxFF-type surfaces to the energy-gap transition model enables cis↔trans photoisomerization trajectories with bond breaking/formation handled by the reactive force field while electronic hopping is approximated without full surface hopping.

2 — Comparisons. The abstract reports qualitatively acceptable quantum yields versus reference Langevin simulations at the FOCI-AM1 level for both cis→trans and trans→cis directions.

3 — Sensitivity & design levers. 2000-trajectory ensembles per direction stabilize yield estimates against rare branching in the gap model; κ/χ parameters control the transition rate model (see pdf_path).

4 — Limitations & outlook. The approach is framed as a proof of principle; accuracy is limited versus higher-level nonadiabatic methods, and reparameterization is needed for new substituents or condensed-phase environments.

5 — Corpus honesty. The repo filename gnuplot…eps is an ingest artifact; authoritative Methods/Results locators are the JCP article cited in front matter and pdf_path (papers/Others/Hartke_JCP_2013_azobenzene.pdf). The normalized/extracts file on record is abstract-only.

Limitations¶

Simple transition model; accuracy limited compared to ab initio nonadiabatic dynamics; transferability to substituted azobenzenes and condensed phases requires reparameterization.
Solvent friction and environmental screening of excited states are omitted in the gas-phase azobenzene benchmarks.

Relevance to group¶

Methodological bridge between ReaxFF culture and photochemistry, complementary to ground-state hydrocarbon/battery-focused parameter lines in the corpus.

Citations and evidence anchors¶

Abstract and methods: GA + gap model + isomerization benchmarks (J. Chem. Phys. 139, 224303 (2013); DOI above).

Corpus filename gnuplot…eps reflects an ingest artifact; the scholarly article is the JCP paper cited above.