Orbital Entanglement in Bond-Formation Processes

Katharina Bogusławski, Paweł Tecmer, Gergely Barcza, Örs Legeza, Markus ReiherJournal of Chemical Theory and Computation 2013 · DOI · arXiv

Abstract

The accurate calculation of the (differential) correlation energy is central to the quantum chemical description of bond-formation and bond-dissociation processes. In order to estimate the quality of single- and multireference approaches for this purpose, various diagnostic tools have been developed. In this work, we elaborate on our previous observation that one- and two-orbital-based entanglement measures provide quantitative means for the assessment and classification of electron correlation effects among molecular orbitals. The dissociation behavior of some prototypical diatomic molecules features all types of correlation effects relevant for chemical bonding. We demonstrate that our entanglement analysis is convenient to dissect these electron correlation effects and to provide a conceptual understanding of bond-forming and bond-breaking processes from the point of view of quantum information theory.

Introduction

In quantum chemistry, the "holy grail" is the accurate calculation of correlation energy—the difference between the Hartree-Fock model, where electrons move independently, and reality, where their motions are intricately linked. This correlation is typically categorized into three distinct types: dynamic (instantaneous repulsion), static (near-degeneracy of electronic states), and nondynamic (required for correct fragment separation). Choosing the appropriate computational method depends heavily on which flavor of correlation dominates a given chemical process. This paper presents a quantitative diagnostic tool based on quantum information theory to visualize these effects and guide theoretical modeling.

Methodology: Quantifying Entanglement

The study utilizes the Density Matrix Renormalization Group (DMRG) algorithm. DMRG is uniquely suited for this task as it can handle large active spaces and capture all types of correlation simultaneously. From the DMRG wavefunction, two key quantum information measures are derived:

  1. Single-orbital entropy ($s(1)_i$): Measures the entanglement of a single orbital with all other orbitals.
  2. Mutual information ($I_{i,j}$): Measures the entanglement specifically between a pair of orbitals.

To calculate these, the system's density matrix must be constructed using a basis of local operators that describe transitions between the four possible states of an orbital: empty, spin-down, spin-up, and doubly occupied.

Table defining the 16 local operators based on transitions between orbital states.

As shown in the operator mapping above, sixteen operators $\mathcal{O}^{(m)}$ are defined. In the natural orbital basis, the single-orbital reduced density matrix $\rho_i$ becomes diagonal, with its non-zero elements representing the probabilities of the four possible orbital occupations.

Structure of the single-orbital reduced density matrix where diagonal elements correspond to occupation probabilities.

Case Studies: $N_2, F_2, \text{and } CsH$

The researchers applied this entanglement analysis to three paradigm diatomic molecules:

  • $N_2$ (Dinitrogen): A triple bond characterized by strong static and nondynamic correlation during dissociation.
  • $F_2$ (Difluorine): A weak single bond where dynamic correlation is dominant and notoriously difficult to capture.
  • $CsH$ (Cesium Hydride): Features an "avoided crossing," where the electronic state's nature changes abruptly as the bond stretches.

By plotting entropy and mutual information as a function of interatomic distance, "entanglement diagrams" are created. These diagrams act as a pictorial representation of the chemical bond, showing how entanglement is distributed and how it evolves as the bond breaks.

Key Insights and Results

The analysis reveals that orbital entropy serves as a precise monitor for bond breaking. A bond is effectively broken when the rate of change of orbital entropy reaches zero at large distances. In the case of $N_2$, entanglement analysis shows that the two $\pi$-bonds begin to break and gain entropy faster than the stronger $\sigma$-bond.

Furthermore, the study provides a quantitative classification for correlation types based on entropy ranges:

  • Nondynamic Correlation: $s(1)_i > 0.5$ (indicated by large mutual information values).
  • Static Correlation: $0.1 < s(1)_i < 0.5$.
  • Dynamic Correlation: $s(1)_i < 0.1$ (indicated by weak entanglement spread across many orbitals).

This classification explains why standard methods like CASSCF fail for $F_2$; the entanglement diagrams show a massive web of weak entanglement across many virtual orbitals that CASSCF typically ignores, accounting for the missing binding energy.

Conclusion

This work demonstrates that concepts from quantum information theory provide a powerful lens for understanding chemical bonding. By mapping the "flavor" of electron correlation, researchers can determine the suitability of specific simulation methods for complex chemical problems. Future work will focus on extending these measures to larger polyatomic systems and integrating them into standard quantum chemistry software packages to make these diagnostic tools accessible to the broader scientific community.

Summary updated

Abstract

The accurate calculation of the (differential) correlation energy is central to the quantum chemical description of bond-formation and bond-dissociation processes. In order to estimate the quality of single- and multireference approaches for this purpose, various diagnostic tools have been developed. In this work, we elaborate on our previous observation [J. Phys. Chem. Lett.2012, 3, 3129] that one- and two-orbital-based entanglement measures provide quantitative means for the assessment and classification of electron correlation effects among molecular orbitals. The dissociation behavior of some prototypical diatomic molecules features all types of correlation effects relevant for chemical bonding. We demonstrate that our entanglement analysis is convenient to dissect these electron correlation effects and to provide a conceptual understanding of bond-forming and bond-breaking processes from the point of view of quantum information theory.
BibTeX
@article{bogusawski2013orbital,
  author = {Katharina Bogusławski and Paweł Tecmer and Gergely Barcza and Örs Legeza and Markus Reiher},
  title = {{Orbital Entanglement in Bond-Formation Processes}},
  journal = {Journal of Chemical Theory and Computation},
  year = {2013},
  doi = {10.1021/ct400247p},
}
Figure