Chemical bonding without orbitals

https://doi.org/10.1016/j.comptc.2018.10.004Get rights and content

Abstract

This work presents an implementation of the original orbital-free Hohenberg-Kohn density functional theory in a form that is able to predict chemical bonding in molecules. The method is completely parameter-free and does not require analytical functional approximations. Instead, the proposed method is based on the idea that atoms are meaningful pieces of a molecule and thus, a promolecule, build from frozen spherical atomic entities, serves as a suitable model for the latter. This idea is imposed on the physical equations, originating from density functional theory converted into a bifunctional formalism.

The viewpoint proposed in this study offers a new strategic way of subsequent approximation levels in orbital-free density functional theory. In this work the zeroth order approximation is shown to predict chemical bonding in molecules, providing a concept of the chemical bond without involving orbitals.

Introduction

The chemical bond [1], [2], [3], [4] is a fundamental concept in chemistry and related sciences. Taking a very simplistic standpoint, the chemical bond somehow serves as glue, binding together the partaking atoms [5], [6]. One part of chemistry is to define and quantify chemical bonding in order to predict molecular stability and how the system will undergo possible changes induced by the environment. Unfortunately, the definition of that chemical bonding is ambiguous. This is due to the fact that atoms in molecules are no physical entities. From a puristic standpoint, the molecule is defined once the number of electrons and the nuclear positions are given. By solving the quantum mechanical equations [7], [5], the statistical distribution of the electrons in the field of the nuclei is determined. In that spirit, the molecule only consists as a whole, namely of all nuclei and all electrons, whereby the latter are usually (but not necessarily) described in terms of orbitals. The equations describe all orbitals at once and their solutions extend over the whole space. This form of wavefunction based quantum mechanics is said to be non-local [7], meaning that changes induced at one position in space might cause noticeable effects far away from the initial position. Such a non-local form of quantum mechanics is not suited for an atomic fragment approach, where those fragments are thought as being subjected to a physical interaction.

Indeed a local from of quantum theory is needed, further allowing for a meaningful separation into atomic fragments. This local form of quantum theory is called density-functional theory (DFT) [8], [9]. In the Hohenberg-Kohn variant[10], the system is described by the electron density itself, rather than by orbitals. This theory is guided by a local causality principle [11], [12], [13], [14], [15], [16], [17], [18] (effective interactions decrease with increasing distance) and, as will be shown in this work, allows for a meaningful separation into atomic fragments, while keeping the initial physical equations unaltered. Although the original ideas of Hohenberg and Kohn dates back more than fifty years, a widespread use of orbital-free DFT has been hampered by the lack of sufficiently accurate functional approximations for the kinetic energy [19]. Whereas considerable progress has been made for one-dimensional systems [20], the treatment of molecular systems or solids requires appropriate three-dimensional functional approximations. First work was based on conventional gradient expansions [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], whereas recent functional development is mainly focused on techniques based on the generalized-gradient-approximation motivated by conjoint arguments [33], [34] or the fulfillment of additional constraints [35], [36], [19]. Information-theory motivated functionals [37], [38], [39], functionals based on response theory [40], [41], and expansions in terms of moment densities [42] have been studied. Recently, a very promising approach based on the Liu-Parr power series expansion [43] has been suggested by Ludeña and coworkers [44], [45], including a detailed investigation concerning the ability of their ansatz to represent the atomic shell structure. Despite much efforts, it still remains difficult to design appropriate functional approximations, which yield properly structured electron densities from variational procedure [46]. This failure can be attributed to an insufficient incorporation of the Pauli exclusion principle [27], [8], [19] in the Pauli kinetic energy and the corresponding Pauli potential [47]. Both terms were intensively studied in the literature [48], [49], [50], [51], [52], [53], [54], [55], [56] as they represent the unknown part of the kinetic energy functional and the corresponding potential, respectively. As a consequence, there is up to now no reliably working orbital-free density functional implementation for the waste variety of chemical substances. There are a few noticeable exceptions treating variational orbital-free density functional approximations [57], [58], [59], [60], [61]. However, widespread applications covering all types of chemical substances without need of recursive adjustment of technical details are still out of reach.

This work offers a new strategy in orbital-free density functional theory based on the reformulation in terms of bifunctionals.

Section snippets

Formulation of the problem

In 1964 Hohenberg and Kohn (HK) rigorously founded density functional theory (DFT) [10]. In their seminal paper they proved the one-to-one correspondence between the energy E of the system and the external potential due to the nuclei vZ (first HK theorem) as well as the minimum principle of that energy functional E[ρ] (second HK theorem) for all electron densities ρ that are associated with some external potential. According to the first HK theorem, the electronic energy of a system E[ρ], given

Reformulation in terms of bifunctionals

The conventional strategy in the design of energy functionals is to start with an ansatz for the energy in the form of an analytical expression that is an integral over a function f, which in some way contains the electron density (and possibly some other ingredients):F[ρ]=f(ρ(r),)dr.From that functional, the potential, being the functional derivative, is obtained analytically, cf. Eq. (3). As mentioned, no ansatz for the kinetic energy has been found that is able to predict chemical

The chemical bond in the atomic fragment approximation

From a chemist’s viewpoint it seems natural to interpret [71], [72], [73] large aggregates as being composed from atomic fragments [74], [75], [76], [77] that somehow hold together. In the following it is shown how to link such a viewpoint with the quantum mechanical equations based on the bifunctional formalism and the atomic fragment approach.

A so-called promolecule [78], [76] is the sum of (usually spherical) atoms centered at the positions of the nuclei for the actual molecule of interest.

Discussion

This work presents an orbital-free implementation of the original Hohenberg-Kohn density functional theory, that is able to predict bonding in molecules. The proposed method is completely parameter-free and does not require analytical ansatzes for the energy functionals. Instead, it is based on the idea that atoms are meaningful pieces of a molecule and thus, a promolecule serves as a good model for the latter. This idea is imposed on the physical equations, originating from density functional

Acknowledgments

The author wishes to thank Dr. M. Kohout for fruitful discussions and substantial encouragement over years. Prof. Dr. M. Ruck is greatly acknowledged for academic support and valuable hints on the manuscript. The Technische Universität Dresden is acknowledged for funding in terms of a Maria-Reiche fellowship.

References (79)

  • A. Szabo et al.

    Modern Quantum Chemistry – Introduction to Advanced Electronic Structure Theory

    (1996)
  • R.M. Dreizler et al.

    Density Functional Theory

    (1990)
  • R.G. Parr et al.

    Density-Functional Theory of Atoms and Molecules

    (1989)
  • P. Hohenberg et al.

    Inhomogeous Electron Gas

    Phys. Rev. B

    (1964)
  • W. Kohn et al.

    Locality principle in wave mechanics

    Proc. Natl. Acad. Sci. USA

    (1978)
  • X.-P. Li et al.

    Density-matrix electronic structure method with linear system-size scaling

    Phys. Rev. B

    (1993)
  • W. Kohn

    Density functional and density matrix method scaling linearly with the number of atoms

    Phys. Rev. Lett.

    (1996)
  • R. Baer et al.

    Sparsity of the density matrix in Kohn-Sham density functional theory and an assessment of linear system-size scaling methods

    Phys. Rev. Lett.

    (1997)
  • E. Prodan et al.

    Nearsightedness of electronic matter

    Proc. Natl. Acad. Sci.

    (2005)
  • R.F.W. Bader

    Nearsightedness of electronic matter seen by a physicist and a chemist

    J. Phys. Chem. A

    (2008)
  • C. Cárdenas et al.

    Chemical reactivity descriptors for ambiphilic reagents: dual descriptor, local hypersoftness, and electrostatic potential

    J. Phys. Chem. A

    (2009)
  • S. Fias et al.

    Chemical transferability of functional groups follows from the nearsightedness of electronic matter

    Proc. Natl. Acad. Sci.

    (2017)
  • A. Cangi et al.

    Leading corrections to local approximations

    Phys. Rev. B

    (2010)
  • L.H. Thomas

    The calculation of atomic fields

    Proc. Cambridge Philos. Soc.

    (1927)
  • E. Fermi

    Eine statistische Methode zur Bestimmung einiger Eigenschaften des Atoms und ihre Anwendung auf die Theorie des periodischen Systems der Elemente

    Z. Phys.

    (1928)
  • C.F. von Weizsäcker

    Zur Theorie der Kernmassen

    Z. Phys.

    (1935)
  • D.A. Kirzhnits

    Quantum corrections to the Thomas-Fermi equation

    Sov. Phys. JETP

    (1957)
  • C.H. Hodges

    Quantum corrections to the Thomas-Fermi approximation – the Kirzhnits Method

    Can. J. Phys.

    (1973)
  • D.R. Murphy

    Sixth-order term of the gradient expansion of the kinetic-energy density functional

    Phys. Rev. A

    (1981)
  • W. Yang

    Gradient correction in Thomas-Fermi theory

    Phys. Rev. A

    (1986)
  • W. Yang et al.

    Various functionals for the kinetic energy density of an atom or molecule

    Phys. Rev. A

    (1986)
  • C.L. Lee et al.

    Density gradient expansion of the kinetic-energy functional for molecules

    Phys. Rev. A

    (1986)
  • P.M. Kozlowski et al.

    A graph approach to the gradient expansion of density functionals

    Int. J. Quant. Chem.

    (1986)
  • A.J. Thakkar

    Comparison of kinetic-energy density functionals

    Phys. Rev. A

    (1992)
  • J.-D. Chai et al.

    Modified statistical treatment of kinetic energy in the thomas-fermi model

    J. Phys. Chem. B

    (2004)
  • H. Lee et al.

    Conjoint gradient correction to the Hartree-Fock kinetic- and exchange-energy density functionals

    Phys. Rev. A

    (1991)
  • F. Tran et al.

    Link between the kinetic- and exchange-energy functionals in the generalized gradient approximation

    Int. J. Quant. Chem.

    (2002)
  • D. Lee et al.

    Condition on the Kohn-Sham kinetic energy and modern parametrization of the Thomas-Fermi density

    J. Chem. Phys.

    (2009)
  • V. Karasiev et al.

    Progress on new approaches to old ideas: orbital-free density functionals

  • Cited by (15)

    View all citing articles on Scopus
    View full text