The Hartree-Fock method

Result

The Hartree-Fock method

T-110Home CU-401Threads quantum

Statement

A self-consistent mean-field for many electrons.

Why it matters

hydrogen-atom-solution gave exact orbitals for a single electron in a central Coulomb field, but essentially every atom or molecule of chemical interest has more than one electron, and the Schrödinger equation for a genuinely interacting many-electron system has no exact analytical solution. The Hartree–Fock method is the foundational approximation that makes many-electron quantum chemistry computationally tractable at all, replacing the intractable problem of electrons instantaneously repelling one another with each electron instead moving in an averaged, self-consistent field created by all the others. Every more advanced electronic-structure method covered later in this unit — density-functional-theory included — is either built on the Hartree–Fock framework directly or is routinely benchmarked against it.

Hypotheses

The many-electron wavefunction is approximated as a single Slater determinant of one-electron spin-orbitals.A Slater determinant is the specific antisymmetrised product form required to satisfy the Pauli exclusion principle (indistinguishable fermions, antisymmetric under exchange of any two electrons) automatically; representing the true wavefunction by only one such determinant, rather than a sum of many, is the central approximation of the method, and is what ultimately limits its accuracy. Each electron experiences the other electrons only as an averaged, time-independent mean field, not their instantaneous, correlated positions.This mean-field approximation is precisely what "self-consistent field" refers to, and it is also precisely what is missing relative to the exact wavefunction: the true motion of electrons is correlated (each electron's instantaneous position depends on where the others actually are at that instant, tending to avoid one another beyond what the average field alone predicts), an effect called electron correlation, entirely absent from a single-determinant treatment by construction.

Proof

\Psi(1,2,\ldots,N) = \frac{1}{\sqrt{N!}}\det\big[\chi_i(j)\big]

Building the trial wavefunction as a single Slater determinant of \(N\) one-electron spin-orbitals \(\chi_i\) guarantees antisymmetry under any electron exchange automatically, via the determinant's own sign-flip-on-row-swap property, satisfying the Pauli exclusion principle (Hypotheses) by construction rather than as a separately imposed condition. A

E[\Psi] = \langle\Psi|\hat H|\Psi\rangle \ \text{minimised over all choices of orthonormal spin-orbitals } \{\chi_i\}

Applying the variational-principle (variational-principle, this unit) to the single-determinant trial wavefunction, the best possible set of spin-orbitals is the one that minimises the resulting energy expectation value, subject to the orbitals remaining orthonormal — this constrained minimisation is what generates the working equations of the method. B

\hat f\,\chi_i = \varepsilon_i\,\chi_i, \qquad \hat f = \hat h + \sum_j\left(\hat J_j - \hat K_j\right)

Carrying out the variational minimisation of Step 2 leads to a set of effective one-electron eigenvalue equations, the Hartree–Fock (or Fock) equations, in which each electron moves in the field of the bare nuclei (\(\hat h\)) plus an averaged Coulomb repulsion from the other electrons (\(\hat J\)) and a purely quantum-mechanical exchange term (\(\hat K\), arising directly from the antisymmetry requirement of Step 1, with no classical analogue). B

\text{Iterate: solve } \hat f\chi_i=\varepsilon_i\chi_i \text{ using a trial } \{\chi_i\}\text{, rebuild } \hat f \text{ from the new orbitals, repeat until } \{\chi_i\}\text{ no longer changes.}

Since the Fock operator \(\hat f\) itself depends on the very orbitals it is used to solve for (through the Coulomb and exchange terms), the equations are solved iteratively — the self-consistent field (SCF) procedure — starting from a guessed set of orbitals and repeating until the input and output orbitals (and the resulting energy) converge to within a chosen numerical tolerance. A

Result

\hat f\,\chi_i = \varepsilon_i\,\chi_i \ \text{(solved self-consistently)}, \qquad E_{\text{HF}} \geq E_{\text{exact}}

Reading. The many-electron problem is reduced to a set of coupled, effectively one-electron equations solved iteratively until self-consistency; because it uses a single Slater determinant (variationally optimised, so by the variational-principle its energy is a rigorous upper bound), the resulting Hartree–Fock energy is always at least as high as the true, exact ground-state energy.

Scope. Gives molecular orbitals, orbital energies, and total energies of generally good qualitative and often useful semi-quantitative accuracy, but systematically omits electron correlation (Hypotheses); the gap \(E_{\text{exact}}-E_{\text{HF}}\), the correlation energy, is small on the scale of total electronic energy but often chemically significant (comparable to or larger than typical bond energies), motivating the post-Hartree–Fock and density-functional methods that follow it.

Corollaries & converses

Koopmans' theorem is a direct corollary of the SCF orbital energies of Step 3: to a good approximation, the negative of an occupied orbital's energy \(-\varepsilon_i\) estimates the ionisation energy of removing an electron from that orbital, assuming the remaining orbitals do not relax upon ionisation.
basis-sets supplies the practical machinery (a finite set of basis functions used to represent each spin-orbital \(\chi_i\) numerically) that makes Step 4's iterative SCF procedure solvable on a computer at all; the Hartree–Fock equations themselves are basis-set independent in principle, but every real calculation requires choosing one.
Converse: any post-Hartree–Fock method aiming to recover electron correlation (going beyond a single Slater determinant) can be understood, and is conventionally benchmarked, relative to the Hartree–Fock reference energy and orbitals established here as its starting point.

Fails without

Drop the single-determinant restriction and hope the result still counts as "Hartree–Fock": using a sum of multiple determinants (a multi-configurational treatment) recovers some electron correlation but is then a genuinely different, more expensive family of method, not a Hartree–Fock calculation; conflating the two misrepresents both what has been computed and its expected accuracy.
Ignore that the Fock operator \(\hat f\) depends on the very orbitals being solved for, and attempt to solve Step 3 in a single non-iterative pass: without the self-consistent iteration of Step 4, the resulting orbitals and energy are not converged to a stationary solution of the underlying variational problem at all, and carry no meaningful accuracy guarantee.

Common errors

Assuming Hartree–Fock gives the exact energy of a many-electron system; by Step 2's variational bound it can only ever equal or exceed the true energy, never fall below it, and it generally does not exactly equal it due to the missing correlation energy (Result, Scope).
Confusing the Coulomb term \(\hat J\) (a classical-like average electrostatic repulsion) with the exchange term \(\hat K\) (a purely quantum-mechanical consequence of antisymmetry, with no classical analogue, that only acts between electrons of the same spin).
Treating Koopmans' theorem as an exact result rather than an approximation that neglects orbital relaxation upon ionisation, which can introduce meaningful error, particularly for larger or more polarisable systems.
Believing that using a larger basis set alone will eventually recover the exact energy from a Hartree–Fock calculation; an arbitrarily large basis set converges Hartree–Fock only to the Hartree–Fock limit itself, still missing electron correlation entirely, since that is a limitation of the single-determinant approximation (Step 1), not of basis-set incompleteness.

Discussion

Douglas Hartree developed the original self-consistent-field idea in the late 1920s using a simpler product wavefunction lacking proper antisymmetry; Vladimir Fock and John Slater independently showed, in the early 1930s, how to incorporate the Pauli exclusion principle correctly via a determinantal wavefunction, giving the method its modern Hartree–Fock form. Decades of subsequent computational development, alongside the growth of practical computing power, turned the method from a hand-calculation scheme feasible only for the simplest atoms into a routine, automated procedure applicable to large molecules.

The "exchange-correlation hole" concept, central to understanding both Hartree–Fock's exchange term and density-functional-theory's later, more complete treatment, describes the reduced probability of finding a second electron near a given one of the same spin (exchange, captured exactly by Hartree–Fock's \(\hat K\) term) or of opposite spin (correlation, entirely missing from Hartree–Fock) — a unifying picture across both major families of electronic-structure method.

Common misconception: that a "better" (larger) basis set can substitute for a fundamentally more accurate method. As the Result and Common errors note, basis-set size and level of theory (single-determinant Hartree–Fock versus a correlated post-Hartree–Fock or density-functional method) are two independent axes of computational accuracy; improving one does not substitute for deficiencies in the other.

Worked examples

\text{He atom: } E_{\text{HF}} \approx -2.8617\,\text{Hartree}; \quad E_{\text{exact (non-relativistic)}} \approx -2.9037\,\text{Hartree}

For the well-studied helium atom, the Hartree–Fock energy lies measurably above the essentially exact, numerically converged non-relativistic energy, exactly as Step 3's variational bound requires (\(E_{\text{HF}}\geq E_{\text{exact}}\)); the difference is the correlation energy. A

E_{\text{corr}} = E_{\text{exact}}-E_{\text{HF}} \approx -2.9037-(-2.8617) = -0.0420\,\text{Hartree}\ (\approx -1.14\,\text{eV})

Although this correlation energy is a small fraction of helium's total electronic energy, it is not negligible on a chemical scale, being comparable in magnitude to many chemically significant energy differences (bond energies, activation barriers), which is precisely why methods recovering correlation are often needed for quantitatively reliable chemical predictions. A

E_{\text{HF}} \geq E_{\text{exact}}\ \text{always}; \quad E_{\text{corr}}<0\ \text{always}

Reading. Even for the smallest genuinely many-electron atom, Hartree–Fock's single-determinant approximation leaves a measurable, chemically relevant energy gap relative to the true answer.

Scope. The same qualitative pattern — a variational upper bound with a chemically non-negligible correlation energy — holds across essentially every atom and molecule to which Hartree–Fock is applied.

Problems

Explain why the Hartree–Fock energy of any system can never be lower than the true, exact ground-state energy, referencing the variational-principle.
Solution
The variational-principle guarantees that the expectation value of the Hamiltonian for any normalised trial wavefunction, including a single Slater determinant, is greater than or equal to the true ground-state energy. Since Hartree–Fock's spin-orbitals are chosen specifically to minimise this expectation value over all possible single-determinant trial functions (Step 2), the resulting energy is the lowest possible value achievable by any single determinant, but that minimum can still only ever equal or exceed the true, exact minimum over all possible wavefunctions (which need not be restricted to a single determinant).
Distinguish, in one or two sentences, between the Coulomb operator \(\hat J\) and the exchange operator \(\hat K\) appearing in the Fock operator of Step 3.
Solution
\(\hat J\) represents the classical-like average electrostatic (Coulomb) repulsion felt by one electron due to the averaged charge density of all the others, acting between electrons of any spin. \(\hat K\), the exchange operator, has no classical analogue at all and arises purely from the antisymmetry requirement of the Slater determinant (Step 1); it acts only between electrons of the same spin, and its effect is to lower the energy relative to \(\hat J\) alone by keeping same-spin electrons systematically further apart than the Coulomb term alone would predict.
A student claims that running a Hartree–Fock calculation with an extremely large, near-complete basis set will eventually recover the exact energy of a many-electron molecule. Evaluate this claim using the Hypotheses and Result.
Solution
The claim is false. An arbitrarily large basis set only removes basis-set incompleteness error, converging the calculation to the "Hartree–Fock limit" — the best possible energy achievable by any single Slater determinant (Step 2). It does not, and cannot, recover electron correlation, since that deficiency stems from the single-determinant approximation itself (Hypotheses), not from basis-set size; a correlation-inclusive method (post-Hartree–Fock, or density-functional-theory) is required to close that remaining gap, regardless of how large the basis set is made.