Spin-Resolved Reaction-Network Exploration in AutoMeKin with Minimum-Energy-Crossing-Point Localization and an ⟨S2⟩ Single-Reference Check
- 1Department of Chemistry, KU Leuven, Celestijnenlaan 200F, Leuven, Belgium, 3001
Abstract
Automated exploration of reaction networks typically assumes a single potential-energy surface. However, 3d transition-metal catalysis often involves two-state reactivity, where the rate-determining step may occur on a different spin surface. This work presents an open-source extension of AutoMeKin that incorporates spin considerations throughout. The extension features a GFN2-xTB exploration engine with spin-handling capabilities, identification of closely lying spin states, localization of minimum-energy-crossing-points (MECP) at relevant crossings, integration of MECP nodes connected by nonadiabatic edges within the reaction graph, nonadiabatic transition-state-theory rate calculations, and a multireference diagnostic based on ⟨S2⟩ to disable the single-determinant approach where it is not applicable. The extension is tested on three different iron systems. For H2CO and Fe(CO)5, blind path sampling was utilized, whereas targeted transition-state optimization was applied to the more complex cases. Subsequently, MECPs were localized without manual intervention for Fe(CO)5 and Tp-Fe(II)-ethyl, but not for the Fe(I) alkene-isomerization catalyst.
Related
- Beyond the imbalance: Site-resolved dynamics probing resonances in many-body localization
- Time-resolved Coulomb explosion imaging of a photochemical ring opening reaction
- Enhancing the Efficiency and Flexibility of AutoMeKin: Integrating ORCA and Machine-Learning Potentials
- The Renormalization Group as a Stochastic Exploration Process
- Anchored Exploration in Product Evolution: A Phylogenetic Network Framework for Semiconductor Designs
BibTeX
@misc{sims2026spinresolved,
title = {{Spin-Resolved Reaction-Network Exploration in AutoMeKin with Minimum-Energy-Crossing-Point Localization and an ⟨S2⟩ Single-Reference Check}},
author = {Joshua M. Sims},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15010937/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010937/v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Spin-Resolved Reaction-Network Exploration in AutoMeKin with Minimum-Energy-Crossing-Point Localization and an ⟨S2⟩ Single-Reference Check
AU - Joshua M. Sims
PY - 2026
DA - 2026/10/23
PB - ChemRxiv
DO - 10.26434/chemrxiv.15010937/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010937/v1
AB - Automated exploration of reaction networks typically assumes a single potential-energy surface. However, 3d transition-metal catalysis often involves two-state reactivity, where the rate-determining step may occur on a different spin surface. This work presents an open-source extension of AutoMeKin that incorporates spin considerations throughout. The extension features a GFN2-xTB exploration engine with spin-handling capabilities, identification of closely lying spin states, localization of minimum-energy-crossing-points (MECP) at relevant crossings, integration of MECP nodes connected by nonadiabatic edges within the reaction graph, nonadiabatic transition-state-theory rate calculations, and a multireference diagnostic based on ⟨S2⟩ to disable the single-determinant approach where it is not applicable. The extension is tested on three different iron systems. For H2CO and Fe(CO)5, blind path sampling was utilized, whereas targeted transition-state optimization was applied to the more complex cases. Subsequently, MECPs were localized without manual intervention for Fe(CO)5 and Tp-Fe(II)-ethyl, but not for the Fe(I) alkene-isomerization catalyst.
ER -