Assessment of Approximate Methods for Anharmonic Free Energies

Venkat Kapil, Edgar A. Engel, Mariana Rossi, Michele Ceriotti

Abstract

Quantitative evaluation of the thermodynamic properties of materials—most notably their stability, as measured by the free energy—must take into account the role of thermal and zero-point energy fluctuations. While these effects can easily be estimated within a harmonic approximation, corrections arising from the anharmonic nature of the interatomic potential are often crucial and require computationally costly path integral simulations to obtain results that are essentially exact for a given potential. This study assesses the accuracy and efficiency of common approximate methods: the independent mode framework (IMF), the vibrational self-consistent field (VSCF), and self-consistent phonons (SCP). By benchmarking these against reference path integral calculations across diverse systems—from simple carbon allotropes to complex molecular crystals like paracetamol—we find that while these approximations excel for simple solids, they fail to provide reliable corrections for molecular systems with flexible, rotating units. We conclude that future efforts should prioritize reducing the computational expense of exact path integral methods rather than refining these specific approximate frameworks.

Introduction

In materials science, the free energy is the ultimate arbiter of stability, determining which crystal structure a substance will adopt under specific temperature and pressure conditions. Traditionally, researchers have relied on the harmonic approximation, which assumes atoms vibrate within perfectly parabolic potential wells. However, real atomic interactions are anharmonic; the effective "spring constants" between atoms change as they move further from their equilibrium positions.

To capture these effects accurately, researchers typically employ Path Integral (PI) simulations. While PI methods are physically exact for a given potential, they are notoriously expensive, often requiring orders of magnitude more computational power than harmonic models. To bridge this gap, several "shortcut" methods—IMF, VSCF, and SCP—have been developed. This paper provides a systematic benchmark to determine the reliability of these approximations across varying levels of material complexity.

Methodology

The study implemented three major approximate frameworks within the open-source i-PI engine to compare against Path Integral Thermodynamic Integration (QTI), the high-accuracy reference:

  • Independent Mode Framework (IMF): Treats each vibrational mode as independent but accounts for anharmonicity along that specific mode's coordinate.
  • Vibrational Self-Consistent Field (VSCF): Borrowing from quantum chemistry's Hartree-Fock method, it accounts for interactions between different vibrational modes through a "mean-field" potential.
  • Self-Consistent Phonons (SCP): Utilizes the Gibbs-Bogoliubov inequality to determine the optimal effective harmonic description that best approximates the true anharmonic system.

These methods were tested on a gradient of complexity: rigid carbon allotropes (diamond and lonsdaleite), polymorphs of ice (Ice XI), and paracetamol (Forms I and II), which contains flexible, rotating methyl groups.

Results and Discussion

Success in Simple Solids

For simple, rigid systems like diamond and lonsdaleite, the approximate methods performed exceptionally well. The anharmonic corrections to the Helmholtz free energy were in excellent agreement with the reference PI calculations, staying within approximately 2 meV/atom. In these cases, the approximations provide a high-fidelity result at a fraction of the computational cost.

Failure in Complex Molecular Systems

As the complexity of the system increased, the accuracy of the approximate methods deteriorated significantly:

  • Ice XI: Errors in the anharmonic correction jumped to 10–20 meV/molecule.
  • Paracetamol: The methods failed qualitatively. For paracetamol, the approximations overestimated the stability of one form by over 80 meV/molecule compared to the reference.

Crucially, the study found that even when these methods correctly predicted the relative stability between two phases, it was often due to fortuitous error cancellation. The methods were often equally incorrect for both phases, leading to a deceptive appearance of accuracy in relative terms that does not hold up under rigorous scrutiny.

The "Curvilinear Trap"

A central insight of this research is the identification of the "curvilinear trap." Approximate methods like IMF, VSCF, and SCP are fundamentally built upon Cartesian (straight-line) normal modes. In molecular crystals like paracetamol, atoms often move in arcs or rotations. Attempting to describe a circular or curved path using a linear coordinate system makes the potential appear much "stiffer" than it is in reality. This leads to an artificial "blue-shift" (stiffening) of vibrational modes, which is the primary source of error in organic and flexible systems.

Conclusion

The findings suggest that approximate methods based on linear normal modes are a "dead end" for systems characterized by large-amplitude, curved motions, such as flexible molecular crystals or biological systems. While they remain useful for simple, rigid solids, they do not offer a systematic improvement over the harmonic approximation for more complex materials.

Consequently, the authors argue that research efforts should shift away from refining these specific approximations. Instead, the community should focus on:

  1. Accelerating Path Integral Methods: Developing techniques like ring-polymer contraction or high-order estimators to make exact simulations more affordable.
  2. Machine Learning Potentials: Utilizing ML to create fast, high-fidelity potential energy surfaces that allow for the routine use of exact PI methods in complex chemical environments.

Summary updated

Abstract

Quantitative evaluation of the thermodynamic properties of materials-most notably their stability, as measured by the free energy-must take into account the role of thermal and zero-point energy fluctuations. While these effects can easily be estimated within a harmonic approximation, corrections arising from the anharmonic nature of the interatomic potential are often crucial and require computationally costly path integral simulations to obtain results that are essentially exact for a given potential. Consequently, different approximate frameworks for computing affordable estimates of the anharmonic free energies have been developed over the years. Understanding which of the approximations involved are justified for a given system, and therefore choosing the most suitable method, is complicated by the lack of comparative benchmarks. To facilitate this choice we assess the accuracy and efficiency of some of the most commonly used approximate methods: the independent mode framework, the vibrational self-consistent field, and self-consistent phonons. We compare the anharmonic correction to the Helmholtz free energy against reference path integral calculations. These benchmarks are performed for a diverse set of systems, ranging from simple weakly anharmonic solids to flexible molecular crystals with freely rotating units. The results suggest that, for simple solids such as allotropes of carbon, these methods yield results that are in excellent agreement with the reference calculations, at a considerably lower computational cost. For more complex molecular systems such as polymorphs of ice and paracetamol the methods do not consistently provide a reliable approximation of the anharmonic correction. Despite substantial cancellation of errors when comparing the stability of different phases, we do not observe a systematic improvement over the harmonic approximation even for relative free energies. We conclude that, at least for the classes of materials considered here, efforts toward obtaining computationally feasible anharmonic free energies should therefore be directed toward reducing the expense of path integral methods.
BibTeX
@inproceedings{kapil2019assessment,
  author = {Venkat Kapil and Edgar A. Engel and Mariana Rossi and Michele Ceriotti},
  title = {{Assessment of Approximate Methods for Anharmonic Free Energies}},
  booktitle = {Figshare},
  year = {2019},
  doi = {10.1021/acs.jctc.9b00596.s001},
}
Figure