[Paper Review] An Electronically Non-Adiabatic Generalization of Ring Polymer Molecular Dynamics

Timothy J. H. Hele|arXiv (Cornell University)|Aug 19, 2013

Advanced Physical and Chemical Molecular Interactions7 references17 citations

TL;DR

This paper introduces a non-adiabatic generalization of Ring Polymer Molecular Dynamics (RPMD) rate theory, enabling accurate modeling of quantum effects and recrossing dynamics in electronically non-adiabatic systems. The method extends classical RPMD by incorporating electronic state coupling through a modified potential energy surface, achieving agreement with exact quantum rates within a factor of two even in deep tunneling regimes, with computational cost scaling linearly with system size.

ABSTRACT

In this thesis I generalize Ring Polymer Molecular Dynamics (RPMD) rate theory to electronically non-adiabatic systems, followed by application to two one-dimensional curve crossing models and a multidimensional spin-boson model.

Motivation & Objective

• To extend Ring Polymer Molecular Dynamics (RPMD) rate theory to electronically non-adiabatic systems where nuclear motion occurs on multiple electronic potential energy surfaces.
• To develop a computationally efficient method that captures quantum effects such as tunneling and recrossing dynamics in non-adiabatic processes.
• To ensure the method reduces to conventional adiabatic RPMD in the single electronic state limit and maintains physical consistency in high-temperature limits.
• To validate the approach on one-dimensional curve-crossing models and a multidimensional spin-boson model with exact quantum benchmarks.
• To demonstrate that the method's computational cost scales linearly with system size while outperforming other approximate methods in accuracy.

Proposed method

• Generalizes the classical isomorphism by representing a quantum system as a ring polymer of classical beads connected by harmonic springs, with an extended phase space including electronic degrees of freedom.
• Derives a new effective potential energy surface for the ring polymer that incorporates electronic non-adiabatic coupling through a modified coupling term in the Hamiltonian.
• Applies the Bennett-Chandler factorization to decompose the RPMD rate into the product of the centroid probability density at the dividing surface and the flux of positive-momentum centroids in the product region.
• Uses a variational approach to determine the optimal dividing surface in asymmetric curve-crossing models.
• Implements the method using classical molecular dynamics on the extended phase space, with the reaction rate computed from time correlation functions of ring polymer centroids.
• Handles the challenge of divergent Quantum Transition State Theory (QTST) rates in the spin-boson model by using a convergent factorization into two physically meaningful, finite contributions.

Experimental results

Research questions

• RQ1 Can RPMD rate theory be generalized to electronically non-adiabatic systems while preserving its accuracy and computational efficiency?
• RQ2 How does the non-adiabatic RPMD rate compare to exact quantum rates in one-dimensional curve-crossing models with symmetric and asymmetric coupling?
• RQ3 Can the method accurately describe the spin-boson model with a Debye spectral density, especially in the limit of vanishing reaction coordinate mass?
• RQ4 Does the non-adiabatic RPMD method correctly recover the adiabatic RPMD limit when electronic states are decoupled?
• RQ5 Can the RPMD rate be reliably computed in systems where standard QTST fails due to divergent mass factors or undefined rates?

Key findings

• For the symmetric one-dimensional curve-crossing model, RPMD rates agree with exact quantum rates within a factor of two, even at 100 K where classical rates fail by orders of magnitude.
• In the asymmetric curve-crossing model, RPMD provides an even better approximation to the exact quantum rate than in the symmetric case, with deviations comparable to those in the adiabatic limit.
• For the spin-boson model with a Debye spectral density, the RPMD rate is calculable via a convergent factorization despite QTST being undefined due to divergent mass factors.
• The non-adiabatic RPMD rate is found to be closer to exact quantum rates than almost all other approximate methods tested, including popular mean-field and semiclassical approaches.
• The method correctly reduces to conventional adiabatic RPMD when electronic coupling is turned off, confirming consistency with established theory.
• The computational cost scales linearly with system size, making it suitable for large-scale simulations of electronically non-adiabatic dynamics.