Analytical nuclear gradients for the range-separated many-body dispersion model of noncovalent interactions

Martin A. Blood-Forsythe, Thomas Markovich, Robert A. DiStasio, Roberto Car, Alán Aspuru‐GuzikChemical 2015 · DOI

TL;DR

This paper introduces analytical gradients for the Many-Body Dispersion (MBD) model, a sophisticated method for capturing weak, long-range interactions crucial for molecular behavior. By developing these gradients, the authors enable efficient and accurate geometry optimizations of complex molecular systems, demonstrating MBD's superior performance over simpler dispersion models.

What Problem Does This Paper Address?

Accurately describing noncovalent interactions, particularly van der Waals (vdW) or dispersion forces, is fundamental for understanding molecular structure, dynamics, and function. While Density Functional Theory (DFT) is a workhorse in computational chemistry, standard DFT functionals struggle to capture these subtle, long-range electron correlation effects. The Many-Body Dispersion (MBD) model offers a more accurate approach by treating dispersion as a system of coupled quantum harmonic oscillators. However, performing geometry optimizations with MBD has been computationally expensive because it requires calculating forces numerically. This paper addresses this by deriving and implementing analytical gradients (forces) for the MBD model, making it practical for routine geometry optimizations and molecular dynamics simulations.

Methods

The core of this work is the derivation of analytical nuclear gradients for the range-separated MBD (MBD@rsSCS) energy. This involves:

  • Extending the MBD Model: The authors build upon the existing MBD framework, which models long-range correlation energy using coupled quantum harmonic oscillators.
  • Analytical Gradient Derivation: They meticulously derive the mathematical expressions for the forces acting on atomic nuclei. This is a complex task because the MBD energy depends on nuclear positions both explicitly (through interatomic distances) and implicitly (through the partitioning of electron density into effective atomic volumes, known as Hirshfeld partitioning).
  • Implementation: The derived gradients are implemented within the QUANTUM ESPRESSO (QE) code.
  • Geometry Optimizations: The authors use these new analytical gradients to perform geometry optimizations on various molecular systems.

Datasets

The paper uses established molecular systems for validation:

  • Benzene dimer: A classic system for studying π-π interactions and intermolecular dispersion.
  • Isolated small peptides: Used to investigate intramolecular interactions responsible for secondary protein structures (e.g., α-helices, β-pleated sheets).
  • C60@C60H28 buckyball catcher host-guest complex: A larger supramolecular system to test the method's scalability and accuracy for complex noncovalent assemblies.

The authors compare their results against high-level wavefunction theory (e.g., PBE/CCSD(T), MP2) and other dispersion correction methods (TS, D3(BJ)) as benchmarks.

Key Insights and Ideas

  • Analytical Gradients are Crucial: Neglecting the implicit dependence of MBD forces on nuclear positions (arising from the Hirshfeld partitioning) leads to significant errors (around 20% on average, sometimes exceeding 100%). This highlights the necessity of a rigorous analytical derivation.
  • MBD Outperforms Simpler Models: For the tested systems, MBD consistently provides more accurate geometries compared to pairwise dispersion corrections like TS and D3(BJ), especially for systems with significant intramolecular dispersion or complex supramolecular interactions.
  • Efficiency Gains: Analytical gradients dramatically reduce the computational cost of geometry optimizations compared to numerical differentiation, making MBD applicable to larger and more complex systems.
  • Implicit Dependencies Matter: The way electron density is partitioned into atomic volumes (Hirshfeld partitioning) and how these volumes change with nuclear positions is critical for accurate MBD forces.

Key Results

  • Accurate Geometries: Geometry optimizations using the analytical MBD gradients yielded excellent agreement with high-level wavefunction theory benchmarks for the benzene dimer and isolated peptides.
  • Benzene dimer: Mean RMSD of 0.01 Å compared to PBE/CCSD(T).
  • Peptides: Median RMSD of 0.07 ± 0.03 Å compared to MP2.
  • Improved Supramolecular Description: For the C60@C60H28 complex, MBD provided a more consistent description of the host-guest geometry compared to other methods.
  • Significant Force Errors Without Implicit Gradients: Ignoring the implicit coordinate dependence in MBD forces led to substantial errors, particularly in flexible molecular systems with intramolecular dispersion.

Limitations and Future Work

  • Computational Cost: While analytical gradients improve efficiency, MBD is still more computationally demanding than standard DFT functionals.
  • Self-Consistent Implementation: The current work treats MBD as an "a posteriori" correction. Future work will explore a self-consistent implementation of MBD, which is expected to further improve accuracy and be crucial for molecular dynamics simulations.
  • Periodic Systems: The current implementation and analysis focus on isolated molecules. Extending the analytical gradient implementation to condensed-phase (periodic) systems is a significant area for future research.
  • Implicit Coordinate Dependence: While the paper demonstrates the importance of these terms, further investigation into their impact across a wider range of systems is warranted.

Summary updated

Abstract

An accurate treatment of the long-range electron correlation energy, including van der Waals (vdW) or dispersion interactions, is essential for describing the structure, dynamics, and function of a wide variety of systems. Among the most accurate models for including dispersion into density functional theory (DFT) is the range-separated many-body dispersion (MBD) method [A. Ambrosetti <i>et al.</i>, <i>J. Chem. Phys.</i>, 2014, <b>140</b>, 18A508], in which the correlation energy is modeled at short-range by a semi-local density functional and at long-range by a model system of coupled quantum harmonic oscillators. In this work, we develop analytical gradients of the MBD energy with respect to nuclear coordinates, including all implicit coordinate dependencies arising from the partitioning of the charge density into Hirshfeld effective volumes. To demonstrate the efficiency and accuracy of these MBD gradients for geometry optimizations of systems with intermolecular and intramolecular interactions, we optimized conformers of the benzene dimer and isolated small peptides with aromatic side-chains. We find excellent agreement with the wavefunction theory reference geometries of these systems (at a fraction of the computational cost) and find that MBD consistently outperforms the popular TS and D3(BJ) dispersion corrections. To demonstrate the performance of the MBD model on a larger system with supramolecular interactions, we optimized the C<sub>60</sub>@C<sub>60</sub>H<sub>28</sub> buckyball catcher host-guest complex. In our analysis, we also find that neglecting the implicit nuclear coordinate dependence arising from the charge density partitioning, as has been done in prior numerical treatments, leads to an unacceptable error in the MBD forces, with relative errors of ∼20% (on average) that can extend well beyond 100%.
BibTeX
@inproceedings{bloodforsythe2015analytical,
  author = {Martin A. Blood-Forsythe and Thomas Markovich and Robert A. DiStasio and Roberto Car and Alán Aspuru‐Guzik},
  title = {{Analytical nuclear gradients for the range-separated many-body dispersion model of noncovalent interactions}},
  booktitle = {Chemical},
  year = {2015},
  doi = {10.1039/c5sc03234b},
}

Loading related content...

Figure