A grid-free approach for simulating sweep and cyclic voltammetry.

Alec J Coffman, Jianfeng Lu, Joseph E SubotnikJournal of Chemical Physics 2021 · OpenReview

TL;DR

This paper introduces a novel "grid-free" computational method for simulating electrochemical sweep and cyclic voltammetry. By reformulating the problem using Green's functions and integral equations, the approach significantly reduces computational cost and complexity, particularly when dealing with adsorption effects, while achieving results comparable to traditional grid-based methods.

What Problem Does This Paper Address?

Electrochemical techniques like linear sweep voltammetry (LSV) and cyclic voltammetry (CV) are crucial for understanding electron transfer (ET) reactions and determining fundamental chemical and physical parameters. Simulating these experiments typically involves solving complex partial differential equations (PDEs) that describe diffusion and ET kinetics. A major challenge arises when trying to incorporate adsorption of electroactive species onto the electrode surface, as this introduces nonlinearities that make standard numerical methods (like implicit ODE solvers on a discretized grid) computationally expensive, prone to stiffening, and sometimes unstable. Existing methods struggle to efficiently handle the coupled dynamics of diffusion, ET, and adsorption, especially over the wide range of timescales involved in typical experiments.

Methods

The authors propose a "grid-free" approach that avoids spatial discretization altogether. Instead, they leverage Duhamel's principle to transform the PDEs governing species concentrations into integral equations. This allows them to focus solely on the concentrations at the electrode surface (x=0) and the surface concentrations of adsorbed species, rather than simulating the entire solution volume.

Key aspects of their method include:

  • Green's Function Approach: They utilize the Green's function for the diffusion equation to express the solution concentration at the electrode surface as an integral over time.
  • Integral Equation Formulation: The coupled differential equations for surface and bulk concentrations are recast into a system of integral equations.
  • Singularity Treatment: A crucial part of their method involves analytically treating the singularity that arises in the integral equations at the upper limit of integration (s=t), enabling accurate numerical evaluation.
  • Implicit ODE Solver for Surface Dynamics: While the bulk concentration dynamics are handled via integrals, the equations governing the adsorbed species concentrations (which are ODEs) are solved using an implicit ODE solver.
  • Iterative Coupling: The solution for bulk and surface concentrations are iteratively coupled until convergence is achieved at each time step.

This approach contrasts with traditional methods that rely on discretizing space into a grid and solving PDEs, which can become computationally prohibitive with increasing grid density or complexity of the problem (e.g., adsorption).

Datasets

Not discussed in the paper. The paper focuses on developing and validating a computational method, using simulated data for generating the presented cyclic voltammograms.

Key Insights and Ideas

  • Eliminating the Spatial Grid: The most significant insight is that by reformulating the problem in terms of integral equations and focusing on surface concentrations, the need for a spatial grid can be completely removed. This drastically simplifies the problem and reduces the number of variables to track.
  • Handling Adsorption Efficiently: The grid-free approach proves particularly effective for simulating adsorption effects, which are notoriously difficult for traditional grid-based methods due to nonlinearities and stiffening.
  • Computational Speedup: The method offers a substantial computational advantage, being significantly faster than traditional grid-based methods, especially as the complexity of the simulation increases.
  • Quantitative Agreement: The simulated cyclic voltammograms generated by the grid-free method show excellent agreement with results obtained from established grid-based PDE solvers, validating the accuracy of the new approach.

Key Results

  • The proposed grid-free algorithm successfully simulates cyclic voltammograms (CVs) for both systems with and without adsorption.
  • In the case of no adsorption, the method reproduces expected electrochemical behavior, such as peak separation that varies with scan rate.
  • With adsorption, the method captures the distinct behavior of surface-bound species, showing no peak separation as expected.
  • Computational performance is significantly improved:
  • Table I shows that the grid-free method's computation time is relatively invariant to scan rate, even with adsorption, unlike grid-based methods.
  • Table II demonstrates a substantial speedup, with the grid-free method completing simulations in 2.34 seconds, compared to 9.94 to 83.02 seconds for discretized methods with varying grid sizes.

Limitations and Future Work

  • Simplifying Assumptions: For simplicity in the derivation, the authors assumed equal diffusion coefficients for species A and B. While they note this can be modified for unequal diffusion constants, it's a simplification in the presented derivation.
  • Parameter Sensitivity: While not explicitly stated as a limitation, the accuracy of the simulations will depend on the accurate input of kinetic and thermodynamic parameters for ET and adsorption.
  • Future Work: The authors aim to apply this methodology to study various parameter regimes of electrochemical voltammetry experiments, particularly those involving more complex adsorption kinetics and beyond. They also suggest that the code implementing this algorithm is available for others to use and build upon.

Summary updated

BibTeX
@article{coffman2021grid,
  author = {Alec J Coffman and Jianfeng Lu and Joseph E Subotnik},
  title = {{A grid-free approach for simulating sweep and cyclic voltammetry.}},
  journal = {Journal of Chemical Physics},
  year = {2021},
  url = {https://openreview.net/forum?id=WgUNXoAqrA},
}
Figure