Multi-Electron Atoms and the Orbital Approximation

Every electronic structure calculation used in modern chemistry — from Gaussian and ORCA input files run by synthetic chemists checking a reaction mechanism, to the density functional theory (DFT) codes used in materials science to design battery electrodes and catalysts — is built on some descendant of the orbital approximation. Hartree-Fock theory itself remains the starting point (the "reference wavefunction") for essentially all high-accuracy post-Hartree-Fock methods (MP2, coupled cluster, configuration interaction) that then add back the correlation energy the mean-field picture necessarily misses. In spectroscopy, the $\ell$-dependent orbital energy splitting derived here is exactly what explains why atomic emission and absorption spectra of multi-electron atoms (sodium's D-lines, for instance) look qualitatively different from hydrogen's simple Balmer/Lyman series, with term symbols and selection rules built on the same shell structure. In inorganic and organometallic chemistry, the same penetration logic explains transition-metal orbital ordering ($4s$ filling before $3d$, then emptying first on ionization) that governs the reactivity and magnetism of first-row transition metal complexes used industrially as catalysts (iron and cobalt in Fischer-Tropsch synthesis, palladium and nickel in cross-coupling). Even in drug design, semi-empirical and DFT calculations of ligand-protein binding energies inherit the orbital approximation's basic machinery, refined by decades of correlation-energy corrections, exchange-correlation functionals, and basis-set improvements — but the foundational move remains exactly what Hartree formalized: replace the impossible many-body problem with a self-consistent, iterated one-electron picture.

Ask the tutor: Can you walk me through why the variational helium calculation gives ζ* = 27/16 exactly, step by step from the energy expression?

Start with the honest Hamiltonian for a neutral atom of atomic number $Z$ with $N = Z$ electrons (in atomic units, distances in bohr, energies in hartree):

The first sum is the kinetic energy of each electron. The second is the attraction of each electron to the nucleus — this term alone is exactly what appears in the hydrogen atom problem, once per electron, and would let the equation separate into $N$ independent hydrogen-like equations if it were the only interaction. The catastrophe is the third sum: the repulsion between every pair of electrons, where $r_{ij} = |\mathbf{r}_i - \mathbf{r}_j|$ depends simultaneously on two electrons' coordinates. Because $r_{ij}$ mixes $\mathbf{r}_i$ and $\mathbf{r}_j$, you cannot write $\hat{H}$ as a sum of independent one-electron operators, and separation of variables — the trick that cracks the hydrogen atom into radial and angular pieces — fails outright. Even for helium ($N=2$), the equation

has never been solved in closed form; it is a genuine three-body problem (nucleus plus two electrons), mathematically kin to the three-body problem in celestial mechanics that has no general analytic solution either. Every value you have ever seen quoted for helium's ground-state energy (experimental $-79.0$ eV total electronic energy, or $-2.9037$ hartree) comes from numerical or perturbative methods, not an exact formula. The orbital approximation sidesteps this by a physically motivated act of averaging. Instead of tracking the instantaneous position of every other electron, we ask: what is the average electrostatic field that electron $i$ feels from the time-averaged charge distributions of all the other electrons? If we assume this averaged field is spherically symmetric — a good approximation because atoms are, on average, roughly spherical — then each electron obeys its own one-electron Schrödinger-like equation:

This has exactly the mathematical form of the hydrogen atom equation, so its solutions $\phi_i$ are hydrogen-like orbitals: the same $1s$, $2s$, $2p$, $3d$ angular and radial shapes, now computed in a modified effective potential rather than the bare $-Z/r$. The full atomic wavefunction is then built (to lowest approximation) as a product of these one-electron orbitals, later antisymmetrized into a Slater Determinants to respect the Pauli principle and electron indistinguishability. The physical picture that falls out is shielding: an electron far from the nucleus feels the full nuclear charge $Z$ partially cancelled by the electrons closer in, so it experiences an effective nuclear charge $Z_{\text{eff}} < Z$. Because s orbitals have larger amplitude near the nucleus than p or d orbitals of the same $n$ (better "penetration"), s electrons feel less shielding and sit lower in energy — the origin of the $ns < np < nd$ ordering that determines the entire structure of the periodic table, elaborated fully in Effective Nuclear Charge and Shielding.

The exact many-electron Hamiltonian.

where: $N$ = number of electrons, $Z$ = nuclear charge, $r_i$ = distance of electron $i$ from the nucleus, $r_{ij}$ = distance between electrons $i$ and $j$. Reading: the first two sums are separable (they are sums of one-electron operators); the third sum is a genuine two-electron operator for every pair, and it is the sole reason the equation cannot be split into independent one-electron problems.

The mean-field (orbital) approximation.

where: $V_{\text{eff}}(r_i)$ is a spherically symmetric potential representing the time-averaged repulsion electron $i$ feels from every other electron's charge cloud. Reading: this is the defining move of the orbital approximation — replacing instantaneous pairwise repulsions with an averaged, one-electron potential, at the cost of neglecting instantaneous correlation between electrons' motions.

The resulting effective one-electron equation.

where: $\phi_{n\ell m}$ is a one-electron orbital with the same angular shapes ($s,p,d,\ldots$) as hydrogen, and $\varepsilon_{n\ell}$ is its orbital energy. Reading: because $V_{\text{eff}}$ is spherically symmetric, the angular part of the Schrödinger equation is identical to hydrogen's — spherical harmonics $Y_\ell^m$ still solve it exactly — so orbital shapes survive unchanged; only the radial equation, and hence the energy and radial extent, is modified.

Effective nuclear charge (Slater-type bookkeeping).

where: $\sigma$ is a shielding constant summarizing how much nuclear charge is cancelled by the other electrons for a given orbital. Reading: this single number is a convenient shorthand for everything $V_{\text{eff}}(r)$ does; it is developed fully and quantitatively in Effective Nuclear Charge and Shielding, but the qualitative idea — that $s$ orbitals shield poorly and are shielded less — already falls directly out of the orbital approximation above.

Total energy is not simply the sum of orbital energies.

where: $\varepsilon_i$ are the one-electron orbital energies from the effective equation. Reading: summing orbital energies double-counts the electron-electron repulsion (each pairwise repulsion is implicitly included in both electrons' $V_{\text{eff}}$), so the true total electronic energy requires a correction term subtracting the double-counted repulsion — a detail that becomes essential once you build quantitative Hartree-Fock energies from orbital energies.

Example 1 — Foundational: perturbative estimate of helium's ground-state energy, checked against experiment. Using Step 4 of the first derivation, the zeroth-order (no-repulsion) energy for He ($Z=2$) with both electrons in hydrogenic $1s$ orbitals is

(using $1\ \text{hartree}=27.2$ eV, and doubling the single-electron energy $-Z^2/2$ for two electrons). The first-order repulsion correction is

so the first-order estimate of the total electronic energy is

Compare to the experimental total electronic energy of helium, $-79.0$ eV. The perturbative estimate is off by

This residual $-4.2$ eV is the electron correlation energy — the part of the true repulsion effect that a single averaged term (or a mean-field orbital treatment) cannot capture, because it reflects instantaneous, correlated avoidance of the two electrons rather than their time-averaged repulsion. Limiting-case check: as $Z\to\infty$ (a hypothetical very highly charged nucleus with only two electrons, i.e., a helium-like ion), $E_0 = -Z^2$ hartree grows as $Z^2$ while $E^{(1)}=5Z/8$ grows only linearly in $Z$; the fractional importance of repulsion, $E^{(1)}/|E_0| = (5Z/8)/Z^2 = 5/(8Z)$, shrinks to zero. This confirms that electron repulsion becomes relatively less important for very highly charged few-electron ions — a sanity check consistent with the orbital approximation becoming more accurate (not less) as $Z$ increases, since the nuclear attraction increasingly dominates over the fixed-form repulsion.

Example 2 — Applied: ranking orbital energies with penetration for a real multi-electron atom, sodium. Sodium has 11 electrons, configuration $1s^2 2s^2 2p^6 3s^1$. Consider the outermost region occupied by $3s$, $3p$, and (unoccupied but chemically relevant) $3d$ orbitals, and rank their energies using penetration. Using the standard Slater's-rules-style estimate for $Z_{\text{eff}}$ (a simple, non-round input set to keep the arithmetic asymmetric and checkable): suppose spectroscopic data give approximate one-electron energies (in eV) for a valence electron promoted into each of these orbitals in a sodium-like environment:

(these are close to the real spectroscopic term values for Na, used here as illustrative asymmetric numbers). Check the ordering predicted by the penetration argument (Step 4 of the second derivation): since $s$ penetrates more than $p$, more than $d$, we expect

Indeed $-5.14 < -3.04 < -1.52$, confirming the prediction. Convert to effective nuclear charges felt using $\varepsilon_{n\ell} = -\dfrac{(Z_{\text{eff}})^2}{2n^2}(27.2\ \text{eV})$ with $n=3$:

Notice $Z_{\text{eff}}(3d)\approx 1.00$: the $3d$ electron feels almost exactly the charge of a bare proton with all ten core electrons perfectly shielding it, because it barely penetrates the core at all — a striking confirmation that the orbital picture's "poor penetration → strong shielding → $Z_{\text{eff}}\to 1$" logic is not just qualitative but numerically self-consistent. Limiting-case check: for a hypothetical orbital with $\ell$ so large that it never penetrates the core electrons at all, the electron should feel exactly the charge of the ion core, $Z_{\text{eff}} \to (Z - N_{\text{core}}) = 11-10 = 1$ for sodium — exactly the trend the $3d$ number is approaching.

Example 3 — Competition-level: variational estimate of helium's ground state with a shielded orbital, showing the orbital approximation self-improving. Instead of the crude $Z=2$ hydrogenic $1s$ orbital (Example 1), let the effective nuclear charge $\zeta$ (zeta) felt by each electron be a variational parameter, reflecting that each electron partially shields the other — this is the orbital approximation applied variationally rather than perturbatively, and it is the calculation that historically convinced physicists the approximation was quantitatively serious. Using trial orbitals $\phi(\mathbf{r}) = \sqrt{\zeta^3/\pi}\,e^{-\zeta r}$ (normalized hydrogenic $1s$ with adjustable exponent $\zeta$) for both electrons, the expectation value of the full Hamiltonian works out (a standard but non-trivial integral) to

in hartree, where the $\zeta^2$ comes from kinetic energy scaling as $\zeta^2$, $-2Z\zeta$ from the two nuclear attractions (each scaling as $-Z\zeta$), and $\tfrac{5}{8}\zeta$ from the repulsion integral (which scales as $\zeta$ since $1/r_{12}$ has units of inverse length and $\zeta$ sets the length scale of both orbitals). For helium, $Z=2$:

Minimize with respect to $\zeta$:

This is a beautifully interpretable result: instead of feeling the bare $Z=2$, each electron effectively feels a shielded nuclear charge of $1.6875$ — the orbital approximation has variationally discovered its own shielding constant, $\sigma = Z - \zeta^* = 2 - 1.6875 = 0.3125$, purely from minimizing the energy, with no input assumption about shielding at all. Plug back in:

Convert to eV: $-2.8477 \times 27.2 = -77.5$ eV. Compare to the perturbative estimate ($-74.8$ eV, Example 1) and experiment ($-79.0$ eV): the variational shielded-orbital calculation is substantially closer to experiment than the crude unshielded estimate, even though it is still built entirely from the orbital-approximation ansatz (a simple product of hydrogenic-type orbitals). The remaining gap, $-79.0-(-77.5) = -1.5$ eV, is again correlation energy — smaller than before because letting $\zeta$ relax has already captured some of the repulsion effect (the average shielding), but a genuinely product-orbital wavefunction can never capture electrons dynamically dodging each other instant-to-instant. Limiting-case check: setting the repulsion term to zero artificially (as if $1/r_{12}\to 0$) would give $\zeta^* = Z = 2$ exactly (no shielding needed), recovering the trivial unshielded hydrogenic limit — confirming the $5/8$ repulsion coefficient is precisely what pulls $\zeta^*$ down from 2 to 1.6875.

The orbital approximation as a workable computational strategy traces most directly to Douglas Hartree, who in the late 1920s devised the "self-consistent field" method: guess a set of orbitals, compute the average field they produce, resolve each electron's one-electron equation in that field, get new orbitals, and iterate until the input and output fields agree — self-consistency. Hartree's original scheme used a simple product wavefunction without antisymmetry; it was Vladimir Fock and John Slater who, shortly after, showed how to enforce the Pauli principle rigorously using antisymmetrized determinants, giving the Hartree-Fock method its modern form. J.C. Slater additionally distilled the physics of shielding into simple empirical rules for estimating $Z_{\text{eff}}$ by hand — rules still taught because they capture, with pencil and paper, the qualitative penetration argument derived rigorously above.