The Hartree Product and Independent Electrons

The Hartree product itself is never used as a production-level method in modern computational chemistry — every serious electronic structure code, from Gaussian and ORCA to Psi4 and VASP, starts from an antisymmetrized Slater determinant (or a linear combination of them), not a bare product. But the conceptual skeleton of the Hartree product survives everywhere: the idea of a self-consistent mean field, iterated to convergence, is the operational core of Hartree-Fock theory, of Kohn-Sham density functional theory (the workhorse of computational materials science and drug design), and even of simplified semiempirical methods used for fast screening of candidate molecules in early-stage drug discovery pipelines. Understanding precisely why the Hartree product fails is also pedagogically indispensable in fields adjacent to chemistry: nuclear physics uses an analogous mean-field "shell model" for nucleons, which must likewise be antisymmetrized (nucleons are fermions too); condensed matter physics builds Bloch-orbital independent-electron pictures of metals and semiconductors that require the same fermionic antisymmetrization when computing exchange-correlation effects; and quantum computing algorithms for simulating molecular electronic structure (variational quantum eigensolver approaches, for instance) must explicitly encode fermionic antisymmetry into their qubit representations (via Jordan-Wigner or Bravyi-Kitaev transformations) precisely because a naive product-state ansatz on qubits would suffer the exact same defect discussed here. In every one of these settings, the lesson from the humble two-electron helium calculation generalizes directly: mean-field orbital thinking is an excellent zeroth approximation to real correlated physics, but the bookkeeping object that carries those orbitals must be antisymmetrized, or the resulting theory silently violates a fundamental symmetry of nature.

Ask the tutor: How exactly does swapping two rows of a Slater determinant generate the minus sign that a Hartree product can never produce?

Start from the electronic Schrödinger equation under the Born-Oppenheimer approximation, for $N$ electrons moving in the field of fixed nuclei:

The one-electron terms (kinetic energy plus nuclear attraction) are perfectly separable — sum them over electrons and the Hamiltonian would fall apart into $N$ independent one-electron problems, each solvable exactly (in principle) as a hydrogen-like problem. It is the electron-electron repulsion term $\sum_{i<j} 1/r_{ij}$ that ruins this: it depends on the coordinates of two electrons simultaneously, so the Schrödinger equation cannot be separated by simple variable separation, the way you separate $x$, $y$, $z$ in a particle-in-a-box. Faced with this, the Hartree strategy is: pretend the equation separates anyway. Assume each electron occupies its own spin-orbital $\chi_i(\mathbf{x}_i)$ — a function of one electron's spatial and spin coordinates — and write

This is the Hartree product. Each electron is assigned to a specific orbital, and the total wavefunction is just the product — mathematically identical in structure to how you'd write a wavefunction for $N$ distinguishable independent particles, like $N$ different colored balls each in its own box. That's exactly the conceptual crack that later widens into a fatal flaw: electrons are not colored balls, they are identical fermions, and quantum mechanics demands that swapping the labels of any two identical fermions must flip the sign of the total wavefunction. A bare product manifestly does not do this — swapping electrons 1 and 2 changes $\chi_1(\mathbf{x}_1)\chi_2(\mathbf{x}_2)$ into $\chi_1(\mathbf{x}_2)\chi_2(\mathbf{x}_1))$, which is neither symmetric nor antisymmetric under exchange in general, and certainly does not equal $-\Psi^{\text{HP}}$. Despite this flaw, the Hartree product is not useless — it is the seed of the entire mean-field idea. If you plug it into the variational principle and grind through the algebra (the derivation below), you find that each orbital $\chi_i$ obeys a one-electron Schrödinger-like equation in which the electron feels the nucleus plus the time-averaged charge cloud of every other electron. This is the origin of concepts like electron shielding, effective nuclear charge, and orbital energies in multi-electron atoms — ideas that survive, refined, into Hartree-Fock theory and even color the intuition behind density functional theory.

The mean-field logic can be visualized directly: each electron sees a smeared-out, time-averaged charge cloud from all the others rather than their instantaneous positions.

The central object is the trial wavefunction itself:

where:

• $\chi_i(i) \equiv \chi_i(\mathbf{x}_i)$ is a normalized one-electron spin-orbital (spatial part times spin function) assigned to electron $i$,
• the orbitals are taken mutually orthonormal, $\langle \chi_i | \chi_j \rangle = \delta_{ij}$, for later convenience,
• the shorthand "(i)" means "evaluated at the coordinates of electron $i$."

Reading: this says electron 1 lives entirely in $\chi_1$, electron 2 entirely in $\chi_2$, and so on, with zero correlation in their motion beyond what's baked into the average potential. It is the direct many-electron generalization of writing a separable wavefunction for independent, distinguishable particles. The energy expectation value of this trial function under the full electronic Hamiltonian, after the algebra below, separates cleanly:

where:

• $h_i = \displaystyle\int \chi_i^*(1)\left(-\tfrac{1}{2}\nabla_1^2 - \sum_A \frac{Z_A}{r_{1A}}\right)\chi_i(1)\, d\mathbf{x}_1$ is the one-electron core energy (kinetic + nuclear attraction) of orbital $\chi_i$,
• $J_{ij} = \displaystyle\iint |\chi_i(1)|^2 \frac{1}{r_{12}} |\chi_j(2)|^2\, d\mathbf{x}_1 d\mathbf{x}_2$ is the Coulomb integral, the classical electrostatic repulsion between the charge clouds of orbitals $i$ and $j$.

Reading: the total energy is just a sum of one-electron energies plus purely classical, Coulomb-only pairwise repulsions between charge clouds — there is no exchange term at all. This is the mathematical fingerprint of the missing antisymmetry: real electrons also feel an exchange stabilization (or destabilization) between same-spin pairs, and the Hartree product simply has no mechanism to produce it, because that term arises specifically from the interference between $\chi_i(1)\chi_j(2)$ and $\chi_i(2)\chi_j(1)$ that only appears once you antisymmetrize. Minimizing $E^{\text{HP}}$ subject to orbital orthonormality produces the Hartree equations, one per orbital:

where:

• the bracketed operator is the Hartree Fock-like operator (here, purely Hartree — no exchange piece) acting on electron 1,
• $\sum_{j\neq i}\int |\chi_j(2)|^2 (1/r_{12})\, d\mathbf{x}_2$ is the mean electrostatic potential felt by electron $i$ due to every other occupied orbital, obtained by smearing electron $j$'s charge density $|\chi_j|^2$ over space and computing its classical potential,
• $\varepsilon_i$ is the orbital energy — a Lagrange multiplier enforcing normalization, later reinterpreted physically.

Reading: each electron obeys its own one-electron Schrödinger equation, but the potential in that equation depends on the very orbitals you are trying to solve for. This circularity is resolved iteratively: guess orbitals, build the mean-field potential, solve for new orbitals, rebuild the potential, and repeat until the orbitals stop changing — the self-consistent field (SCF) procedure, the direct ancestor of every modern SCF-based method including Hartree-Fock and Kohn-Sham DFT.

Example 1 — Foundational: ground-state helium as a Hartree product. Model helium (nuclear charge $Z=2$, two electrons) with the crude but instructive assumption that both electrons occupy the same hydrogenic 1s orbital scaled by an effective charge $\zeta$,

with opposite spins so the spin-orbitals $\chi_1 = \phi_{1s}\alpha$, $\chi_2 = \phi_{1s}\beta$ are automatically orthogonal (via spin, not space). The Hartree product is $\Psi^{\text{HP}} = \phi_{1s}(1)\alpha(1)\,\phi_{1s}(2)\beta(2)$. The one-electron core energy for a hydrogenic 1s orbital with the true nuclear charge $Z=2$ but variational scale $\zeta$ is a standard result:

and the Coulomb repulsion integral for two electrons in the same 1s orbital of exponent $\zeta$ is the classic result

So the total Hartree product energy is

Minimize with respect to $\zeta$:

This is the famous "effective nuclear charge" result: each electron shields the other by $27/16 - 2 = -5/16 \approx 0.3125$ of a full unit of charge, consistent with elementary shielding rules. Plugging back in,

Compare to the experimental helium ground-state energy of $-2.9037$ hartree — the Hartree-product-with-optimized-orbital-exponent estimate is off by about $0.056$ hartree ($\sim 1.5$ eV), entirely due to missing electron correlation beyond this simple mean-field treatment (note: since both electrons here have opposite spin and share one spatial orbital, there is no exchange term to miss in this particular two-electron singlet case — the discrepancy here is dominated by dynamical correlation, not exchange, which is a useful reminder that the exchange defect specifically concerns same-spin electron pairs).

Limiting-case check: as $Z \to \infty$ (imagine a very highly charged nucleus), shielding becomes relatively negligible compared to nuclear attraction, and $\zeta^* \to Z$ exactly in the limit where the $5\zeta/8$ term is negligible next to $2Z\zeta$; more precisely here, minimizing $E = \zeta^2 - 2Z\zeta - (Z-2)...$ — carried through generally for charge $Z$, $\zeta^* = Z - 5/16$, so the shielding constant $5/16$ stays fixed while the fractional shielding $\to 0$ as $Z\to\infty$, exactly as expected physically.

Example 2 — Applied: Lithium's core-valence structure and why Hartree products get shell structure roughly right. Consider neutral lithium, $Z=3$, three electrons, in the Hartree product picture with configuration $1s^2 2s^1$:

Even without solving the full self-consistent equations numerically, the qualitative structure of the mean-field potential is illuminating. The 2s electron, in the Hartree mean-field picture, feels

At large $r_3$ (far from the nucleus), each of the two 1s-electron integrals approaches $1/r_3$ by Gauss's law (a spherically symmetric charge cloud looks like a point charge from outside), so

recovering the correct asymptotic form for a valence electron seeing an effective nuclear charge of $+1$ (three protons screened by two core electrons) — this is exactly the qualitative basis for the empirical idea of "effective nuclear charge" and Slater's rules taught in introductory inorganic chemistry, and it emerges naturally from the mean-field structure of the Hartree equations even before any antisymmetry is added. However, this Hartree product treatment still misses the exchange stabilization between the two same-spin core... — wait, here the two 1s electrons have opposite spin ($\alpha,\beta$), so no exchange integral connects them in the exact theory either. The place exchange would matter is between the $2s\,\alpha$ electron and the $1s\,\alpha$ electron (both spin-up): the true wavefunction has a $K_{1s,2s}$ exchange stabilization between these same-spin electrons that the Hartree product entirely omits, systematically pushing the Hartree-product energy for lithium too high relative to the true (still approximate, but antisymmetrized) Hartree-Fock energy.

Limiting-case check: setting the nuclear charge screening aside, as the number of core electrons $\to 0$ (a bare valence electron around a bare charge $+1$, i.e. hydrogen atom itself), the mean field vanishes identically and the Hartree equation reduces exactly to the hydrogen atom Schrödinger equation, confirming the one-electron limit is handled correctly — the Hartree machinery only introduces error once there are at least two electrons that can exchange.

Example 3 — Exam-level: quantifying the missing exchange energy for two same-spin electrons in a model system. Consider two electrons confined to a 1-dimensional box of length $L$ with hard walls (a toy model, but algebraically clean and pedagogically standard), both in the same spin state (say both $\alpha$), one in the ground orbital $\phi_1(x) = \sqrt{2/L}\sin(\pi x/L)$ and one in the first excited orbital $\phi_2(x) = \sqrt{2/L}\sin(2\pi x/L)$. Take the repulsion to be modeled by a simple contact (delta-function) interaction of strength $g$,

(a standard simplified stand-in used to make exchange-integral algebra tractable by hand, since real $1/r_{12}$ integrals in a box require more machinery). The Hartree product is $\Psi^{\text{HP}} = \phi_1(x_1)\phi_2(x_2)$, giving Coulomb-type energy

Using $\sin^2\theta = \tfrac{1}{2}(1-\cos 2\theta)$ for each factor and expanding the product of cosines, the standard result for this overlap integral over a full box length is

So

Since both electrons have the same spin ($\alpha\alpha$), the physically correct antisymmetrized (Slater-determinant) wavefunction must use the spatially antisymmetric combination

and its interaction energy works out to $J - K$, where the exchange integral for the contact interaction is

(a special feature of contact interactions: for a delta-function potential, the exchange integral between two real orbitals numerically equals the Coulomb integral, because both integrals collapse to the same coincidence-point overlap). So the correctly antisymmetrized same-spin pair has total repulsion energy

This is a clean, exact illustration of the Pauli exclusion / Fermi hole effect: two same-spin electrons in a contact-interaction model experience zero net repulsion energy in the correctly antisymmetrized treatment, because the antisymmetric spatial wavefunction vanishes whenever $x_1 = x_2$ — exactly where the delta-function interaction would otherwise act. The Hartree product, lacking this built-in node at coincidence, predicts a spurious repulsion energy of $J = 3g/4L \neq 0$, overestimating the energy by exactly this amount. For concrete numbers, take $g = 5.0$ (atomic-model units) and $L = 2.0$: the Hartree product wrongly predicts a repulsion energy of

while the true same-spin repulsion energy is exactly $0$ — a 100% relative error introduced purely by neglecting antisymmetry, with no approximation elsewhere in the model.

Limiting-case check: if instead the two electrons had opposite spin, the spin functions $\alpha(1)\beta(2)$ are already antisymmetric-compatible without requiring the spatial part to be antisymmetric (the total wavefunction antisymmetrizes via the spin part instead), so $K=0$ identically for opposite-spin pairs regardless of interaction type, and the Hartree product's Coulomb-only energy $J = 3g/4L$ is then the physically correct answer — confirming that the entire defect illustrated here is specifically a same-spin (exchange/Fermi-hole) phenomenon, not a generic flaw in treating repulsion.

Douglas Hartree developed the self-consistent field method in the late 1920s specifically to make atomic structure calculations tractable on the primitive calculating machines of the era — he was, by training and temperament, as much a numerical analyst and applied mathematician as a physicist, and he built (with his father) some of the earliest differential analyzer computing devices used in Britain. His method, published essentially as the Hartree product plus the self-consistent iterative procedure, was immediately recognized as physically insightful — mean-field orbitals with self-consistently determined potentials — but also immediately recognized (largely by Fock and Slater) as missing the antisymmetry required by Pauli's exclusion principle, which had only been properly formulated as a general quantum-mechanical statement a few years earlier.