Ch · Chemistry

When Do Kohn–Sham Orbitals Have No Physical Meaning?

Published By the 9to5Science editorial team

At a glance

  1. 1. Separate density from orbitals

    Kohn–Sham theory is designed to reproduce the electron density, not to make its auxiliary orbitals literal electron wavefunctions.

  2. 2. Check for nonuniqueness

    Occupied orbitals can be rotated without changing the density, so their individual shapes are not uniquely determined.

  3. 3. Consider stretched H₂

    A single doubly occupied orbital can reproduce the dissociation density while missing the physical electron–electron correlation.

  4. 4. Interpret energies cautiously

    Most Kohn–Sham eigenvalues, especially virtual ones, are not direct measurements of electron removal or addition energies.

Kohn–Sham (KS) orbitals are often drawn as if they were the paths or wavefunctions of electrons in a molecule. That picture can be useful, but it is not what density-functional theory guarantees. The central target of KS theory is the electron density; its orbitals belong to an auxiliary noninteracting system constructed to reproduce that density.

The clearest warnings come from two facts: different sets of orbitals can give exactly the same density, and a KS reference can have a very different many-electron wavefunction from the real system. These points do not make orbitals useless. They explain why orbital pictures should be treated as interpretations, not as uniquely observable objects.

What Kohn–Sham theory requires

In the KS construction, electrons in an auxiliary system occupy orbitals in an effective potential. For a closed-shell system, the density is

n(r)=2∑i occupied∣ϕi(r)∣2.n(\mathbf r)=2\sum_{i\,\mathrm{occupied}}|\phi_i(\mathbf r)|^2.

The factor of 2 accounts for the two spin states in each doubly occupied spatial orbital. The effective potential is chosen so that the auxiliary system’s density matches the interacting system’s density. With the exact exchange-correlation functional, this matching is exact in principle; practical calculations use approximations.

The equation constrains the sum of the occupied orbital densities. It does not say that each orbital is a physical electron wavefunction. Nor does it make an orbital plot an observable: measurements generally concern quantities such as total density, energies, or transition probabilities, rather than a uniquely defined individual KS orbital.

Example: rotate occupied orbitals without changing the density

Suppose two spatial orbitals, ϕ1\phi_1 and ϕ2\phi_2, are both doubly occupied. Define a rotated pair

ψ1=cos⁡θ ϕ1+sin⁡θ ϕ2,ψ2=−sin⁡θ ϕ1+cos⁡θ ϕ2.\begin{aligned} \psi_1 &= \cos\theta\,\phi_1+\sin\theta\,\phi_2,\\ \psi_2 &= -\sin\theta\,\phi_1+\cos\theta\,\phi_2. \end{aligned}

Because this transformation is unitary, the sum of their squared magnitudes is unchanged:

∣ψ1∣2+∣ψ2∣2=∣ϕ1∣2+∣ϕ2∣2.|\psi_1|^2+|\psi_2|^2=|\phi_1|^2+|\phi_2|^2.

So the electron density is identical for every value of θ\theta. In a molecular calculation, one representation may look delocalized across a molecule while another is localized near particular atoms or bonds. Both describe the same occupied space and density. The rotated orbitals may no longer be eigenfunctions of the canonical KS operator if the original eigenvalues differ, but that does not make the density or occupied subspace different.

This is a direct example of why the shape or label of an individual occupied orbital cannot always have a unique physical interpretation. What is robust is the occupied subspace and the density it produces—not every particular set of orbital pictures used to represent them.

Example: stretched H₂

Consider a hydrogen molecule whose nuclei are pulled far apart. Its ground state approaches a singlet with one electron on each hydrogen atom. The real system has strong electron correlation: the electrons avoid occupying the same atom, even though the total spin is zero.

A spin-restricted KS description can reproduce the separated-atom density using one spatial orbital occupied by two electrons. In the dissociation limit, that orbital is approximately a symmetric combination of the two hydrogen 1s functions. Its squared magnitude has weight on both atoms, and doubling it gives approximately one electron’s worth of density on each atom.

But interpreting that doubly occupied orbital literally would suggest that both electrons independently occupy the same delocalized spatial function. That is not the correlated electron arrangement of the physical singlet. The KS system gets the density right while its auxiliary determinant does not reproduce the real system’s detailed electron correlation. This makes stretched H₂ a striking demonstration that matching the density is not the same as giving the KS orbitals the interpretation of actual electron states.

Which orbital results are useful—and which are not

Orbital pictures can still help organize chemical reasoning. They can suggest where density is concentrated, provide a compact way to discuss bonding and symmetry, and help compare related calculations. Their usefulness is strongest when the interpretation is supported by other evidence and is not presented as a unique observable fact.

Orbital energies also need care. For the exact KS system, under the usual conditions, the highest occupied eigenvalue is related to the first ionization energy: εHOMO=−I\varepsilon_{\mathrm{HOMO}}=-I. This special result does not make every KS eigenvalue a measured energy. In particular, the lowest unoccupied KS eigenvalue is not generally equal to minus the electron affinity. Approximate functionals add further errors, and results can depend on the functional and basis set.

Virtual orbitals deserve especially cautious treatment. They help define the auxiliary one-electron problem, but their shapes and energies are not, in general, direct pictures or energies of states formed by adding an electron. Excited-state methods and electron-attachment calculations require their own interpretation.

Common mistakes

  • Calling every KS orbital a physical electron wavefunction. The density is the target; the orbitals are a means of constructing it.
  • Treating one orbital plot as uniquely determined. Occupied orbitals can be transformed while leaving the density unchanged.
  • Assuming a correct density proves a correct KS wavefunction. The auxiliary determinant can miss important correlation, as in stretched H₂.
  • Reading every eigenvalue as an experimental energy. The exact HOMO has a special ionization-energy relation, but that does not extend to all occupied and virtual orbitals.

A good practice is to state which orbitals were plotted and which convention or calculation produced them, then check important chemical conclusions against density-based quantities, energies, or appropriate excited-state and electron-attachment methods.

References

  1. W. Kohn and L. J. Sham, “Self-Consistent Equations Including Exchange and Correlation Effects”
  2. R. M. Martin, Electronic Structure: Basic Theory and Practical Methods
  3. R. G. Parr and W. Yang, Density-Functional Theory of Atoms and Molecules

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.