Benchmarking Proton Tunneling Splittings with a Wavefunction-Based Double-Well Model: Application to the Formic Acid Dimer

Technical Summary: Benchmarking Proton Tunneling Splittings with a Wavefunction-Based Double-Well Model

Problem Statement
Proton tunneling through hydrogen bonds is a fundamental quantum-mechanical phenomenon critical to molecular spectroscopy and hydrogen-bond dynamics. While multidimensional semiclassical methods (such as instanton theory) and path-integral techniques provide quantitatively reliable tunneling splittings for benchmark systems like malonaldehyde and the formic acid dimer, they are computationally intensive and often obscure simple scaling relations. Conversely, reduced-dimensional descriptions offer conceptual transparency but require rigorous validation to ensure they accurately capture the physics of tunneling observables. A systematic, internally consistent benchmarking framework that directly connects semiclassical estimates, wavefunction-based methods, and numerically exact solutions within a unified parametrized model remains underexplored.

Methodology
The study employs a reduced-dimensional, one-dimensional effective potential to model proton transfer. The methodology integrates two complementary approaches:

1. Effective Potential Construction: The author constructs a symmetric quartic double-well potential using a "Cornell-type" functional form ($V_C(x) = -\alpha/|x| + \beta|x|$) as a mathematical building block. This form is shifted and reflected to generate a tunable double-well profile with controllable barrier height ($V_b$), well separation ($d$), and local curvatures. The model is parameterized specifically for the formic acid dimer (FAD) to reproduce a literature-reported barrier height of $V_b \approx 2848 \text{ cm}^{-1}$.
2. Numerical Solution (Exact): The time-independent Schrödinger equation is solved using a finite-difference discretization of the kinetic-energy operator on a uniform real-space grid. A second-order central difference scheme is employed with Dirichlet boundary conditions. The resulting tridiagonal Hamiltonian matrix is diagonalized to obtain the two lowest eigenvalues ($E_0$ and $E_1$), from which the tunneling splitting is calculated as $\Delta E = E_1 - E_0$. Convergence is verified with respect to grid density ($N=3000$) and domain size.
3. Semiclassical Estimation (WKB): Tunneling splittings are independently estimated using the Wentzel–Kramers–Brillouin (WKB) approximation. The Euclidean action ($S$) is computed via numerical quadrature under the barrier between classical turning points defined by $V(x) = E$ (where $E$ is the numerically obtained ground-state energy). The splitting is calculated using the leading-order expression $\Delta E \approx (\hbar\omega/\pi) \exp(-S/\hbar)$.

Key Contributions and Results

• Scaling Trends: The study systematically examines the dependence of tunneling splittings on geometric parameters. Results confirm that $\Delta E$ decreases exponentially with increasing well separation ($d$), consistent with the semiclassical expectation $\Delta E \propto \exp(-S/\hbar)$. This validates the model's ability to reproduce the qualitative and quantitative scaling behavior of one-dimensional tunneling.
• Methodological Consistency: A direct comparison between the numerically exact finite-difference solutions and the WKB estimates reveals a systematic but modest deviation. Across a broad range of barrier heights and separations, the WKB approximation underestimates the numerical splittings by a few percent. This deviation is attributed to neglected higher-order corrections in the semiclassical prefactor, particularly for finite barrier heights. The agreement improves as the barrier height increases, approaching the asymptotic validity regime of the WKB method.
• Formic Acid Dimer Benchmark: Applying the model to the formic acid dimer with $V_b \approx 2848 \text{ cm}^{-1}$ yields a tunneling splitting of $\Delta E \approx 0.037 \text{ cm}^{-1}$. This result is in quantitative agreement with previously reported reduced-dimensional quantum calculations (e.g., Qu and Bowman). The WKB estimate for this specific case ($\approx 0.0356 \text{ cm}^{-1}$) deviates by approximately 4%, consistent with the finite-barrier regime.
• Sensitivity Analysis: The study demonstrates that the tunneling splitting is highly sensitive to physically relevant variations in the effective potential, such as barrier height, highlighting the importance of precise parameterization in reduced-dimensional models.

Significance and Claims
The paper positions itself not as a replacement for multidimensional instanton or path-integral approaches, but as a transparent, reproducible, and pedagogical reference for reduced-dimensional quantum tunneling models.

• Benchmarking Utility: The work establishes a unified framework for assessing the accuracy of approximate methods. By providing a controlled environment where semiclassical and wavefunction-based results can be directly compared, it clarifies the regimes of validity for the WKB approximation in effective one-dimensional potentials.
• Pedagogical Value: The study illustrates the capabilities and limitations of one-dimensional wavefunction-based approaches, making the underlying scaling behavior of tunneling explicit.
• Validation of Reduced-Dimensional Models: The successful reproduction of literature-quality tunneling splittings for the formic acid dimer validates the use of carefully parameterized one-dimensional models for prototypical hydrogen-bonded systems.
• Limitations: The author explicitly acknowledges that discrepancies relative to fully multidimensional treatments or experimental results arise primarily from the neglect of vibrational mode–mode coupling and environmental effects, rather than limitations of the numerical methodology itself.

In summary, the paper provides a rigorous numerical benchmark that bridges analytical semiclassical theory and exact numerical solutions, offering a reliable tool for understanding the fundamental physics of proton tunneling in reduced-dimensional frameworks.