Molecular Propensity as a Driver for Explorative Reactivity Studies

Alain C. Vaucher, Markus ReiherJournal of Chemical Information and Modeling 2016 · DOI

Introduction

In traditional quantum chemistry, researchers must define the "rules" of a simulation—such as the number of electrons (charge) and their spin alignment (multiplicity)—before the calculation begins. However, chemical systems are dynamic; a molecule might transition from a "singlet" state to a "triplet" state during a reaction if the latter becomes energetically favorable. If a researcher does not explicitly account for these alternative states, critical reaction pathways may be overlooked. This challenge is particularly acute in interactive quantum chemistry, where human-driven exploration is prone to bias, often focusing only on expected outcomes.

To address this, researchers have developed a "heads-up display" for quantum chemistry based on the concept of molecular propensity. This algorithm runs in the background of interactive simulations to automatically flag when a molecule is likely to change its spin, gain or lose an electron, or react to light, ensuring that "unseen" reactivity is captured.

Methodology: The Propensity Thread

The solution was implemented within the SAMSON real-time quantum chemistry framework. The core innovation is a dual-threaded architecture that decouples user-driven exploration from background propensity analysis.

  1. Main Exploration Thread: This thread calculates the energy and forces for the current molecular structure at the user-specified spin and charge. It provides immediate haptic or visual feedback, allowing researchers to "tug" on atoms in real-time.
  2. Propensity Thread: Operating in parallel, this thread periodically fetches the current geometry and performs rapid background calculations using semi-empirical methods (e.g., PM6 or DFTB).
The parallel architecture separates the main exploration thread from the background propensity monitoring thread to maintain real-time performance.

The propensity thread monitors three primary indicators:

  • Spin Propensity: Evaluates if alternative spin states offer lower energy configurations.
  • Redox Propensity: Calculates the energy required to add or remove an electron at the current geometry.
  • Photoexcitation Propensity: Identifies low-lying excited states reachable via photon absorption.

If an alternative state falls within a predefined energy threshold, the system triggers a user notification, enabling the researcher to "jump" to a different potential energy surface and continue their exploration.

Case Studies and Results

The algorithm was validated using four complex chemical systems known for non-trivial electronic behavior.

Two-State Reactivity in $FeO^+$

A classic example of spin-state crossing occurs during the oxidation of $H_2$ by $FeO^+$. The algorithm successfully identified points where the system prefers to switch from a sextet to a quartet state. As shown in the energy profiles, this transition significantly lowers the reaction barrier, a discovery that might be missed in a single-state simulation.

Energy profiles for the $FeO^+ + H_2$ reaction, illustrating the crossing between quartet (blue) and sextet (red) states along the exploration coordinate.

Redox and Photoexcitation Propensity

In studies of the Chatt–Schrock cycle for nitrogen fixation, the system flagged increased reduction propensity in the molybdenum complex immediately following proton transfer. Additionally, for [2+2] cycloadditions, the algorithm correctly predicted that the reaction would be "allowed" under light by identifying a low-lying excited state near the transition state, consistent with Woodward-Hoffmann rules.

Discussion and Key Insights

The authors propose that propensity should be viewed as an intrinsic property of a molecular geometry. By checking for these transitions at the current nuclear configuration (vertical transitions), the system aligns with Marcus theory, where structural fluctuations initiate electron transfer.

This approach significantly reduces human bias by acting as an unbiased observer. It automates the search for alternative electronic states that would otherwise be too tedious or counter-intuitive for a human expert to check manually.

Limitations and Future Work

While effective, the current implementation has limitations. The use of semi-empirical methods like PM6 prioritizes speed over absolute accuracy; thus, propensity serves as a "filter" for identifying structures that require subsequent rigorous ab initio refinement. Furthermore, setting energy thresholds for notifications remains a balancing act to avoid overwhelming the user with false positives. Future work aims to implement "protonation propensity" and develop more sophisticated, descriptor-based triggers to enhance the system's predictive power.

Summary updated

BibTeX
@article{vaucher2016molecular,
  author = {Alain C. Vaucher and Markus Reiher},
  title = {{Molecular Propensity as a Driver for Explorative Reactivity Studies}},
  journal = {Journal of Chemical Information and Modeling},
  year = {2016},
  doi = {10.1021/acs.jcim.6b00264},
}
Figure