Molecular Orbitals and Density of States

Introduction

Plots of colorful molecular orbitals and tall density-of-states peaks make calculations feel visually direct. They are valuable maps of a model's electronic structure, but their shapes and eigenvalues are not automatically photographs of electrons or measured spectra. A computed orbital can help identify where a particular model places π character or lone-pair character. A density-of-states plot organizes many such levels. Careful interpretation asks what Hamiltonian, functional, basis, projection and energy reference produced the plot before drawing chemical conclusions.

Core explanation

In Hartree–Fock or Kohn–Sham density-functional theory, molecular orbitals solve an effective one-electron equation. Their squared amplitudes contribute to electron density when occupied, subject to the model's occupations. The orbitals are useful for discussing symmetry, nodal patterns and approximate chemical character. Yet in a many-electron molecule, an individual canonical orbital is not by itself a complete observable: several occupied orbitals can be mixed by allowed unitary transformations without changing the total occupied density in a closed-shell determinant. A visualization must therefore name its orbital definition and level of theory.

An orbital eigenvalue is also method-dependent. Hartree–Fock's frozen-orbital Koopmans approximation relates the negative HOMO energy to ionization energy under specific assumptions, but relaxation and correlation alter actual removal energies. For the exact Kohn–Sham density functional, the HOMO energy has a special connection to the first ionization energy; approximate functionals often fail to reproduce it quantitatively, and most other eigenvalues do not acquire a general simple spectroscopic interpretation. A LUMO–HOMO gap from a common approximate functional is not automatically the optical excitation energy or the fundamental gap (ionization energy minus electron affinity). Excited states involve electron–hole response, relaxation and sometimes spin selection. Primary analysis of Kohn–Sham orbital meaning compared orbital shapes and eigenvalues with other methods and ionization data; further primary work explains when orbital-energy differences can be useful approximations and why the exact-theory distinction matters.

For a finite molecule, the density of states (DOS) begins as a set of discrete eigenvalues. Formally one may write DOS(E) = Σi δ(E − εi), where each delta marks a level. A plotted DOS usually replaces delta spikes by Gaussian or Lorentzian broadening for readability. Width and peak height then depend on the arbitrary plotting kernel and should not be mistaken for intrinsic spectral linewidths. A large periodic crystal has many bands over sampled wave vectors, so its DOS approximates a continuous distribution more naturally. Comparing molecular and solid DOS therefore requires awareness of system size, normalization, smearing and k-point sampling.

A projected DOS (PDOS) weights each level by an atom, element, orbital type or fragment contribution. This can show, for instance, that levels near a band edge have metal d and ligand p character. But the numerical projection depends on the chosen basis or partition rule and may not sum exactly to a naive visual picture if basis functions overlap or interstitial density is important. A projected peak does not prove that an electron physically sits on one atom. Primary interface work using projected DOS illustrates how projections help discuss orbital localization while functional choice changes the computed gap.

An energy reference is indispensable. A diagram may set vacuum energy to zero, align to a Fermi level, or subtract the HOMO energy. A plot shifted by a constant can have the same level spacings but different displayed absolute values. Two molecules' HOMO energies cannot be compared from separate diagrams until the reference, method, solvent and geometry are matched. Near-degenerate orbitals may swap order with modest changes in basis or conformation, even though the broad electronic character persists. For reactions, total-energy differences and properly calculated free energies generally provide firmer quantitative evidence than a single frontier orbital picture.

Step-by-step reasoning

1. Identify the electronic-structure method, geometry, charge, spin and environment used to generate the orbitals. 2. Inspect orbital symmetry and localization without equating the image with an electron trajectory or a unique observable. 3. Establish the energy zero and confirm whether the DOS is discrete, broadened molecular levels or periodic bands. 4. Read any PDOS through its projection definition and verify that the assignments are robust to reasonable analysis choices. 5. Separate a frontier eigenvalue gap from a calculated excitation energy, ionization energy or measured optical edge. 6. Use total-energy or response calculations when the question requires a quantitative reaction or spectroscopic result.

Visual explanation

Draw an energy axis with three occupied horizontal levels below a marked HOMO and two empty levels above a marked LUMO. Place narrow spikes at the eigenvalues, then overlay a smooth curve obtained by broadening them. Next, color each spike according to its chosen fragment weight. The colored broadened plot is a PDOS. Label the common energy zero, the broadening width and the method. A separate arrow from HOMO to LUMO should be captioned “one-electron level difference,” not “measured absorption” without further calculation.

Real-world analogy

A city transit map organizes routes and transfer points but is not a street-level photograph or a record of every passenger's journey. Orbitals organize a model's one-electron states, while a DOS organizes their energies. The map can reveal where a connection may be possible; predicting how many passengers move requires further information. Likewise, an orbital gap alone does not provide an absorption intensity or a complete charge-transfer mechanism.

Real-world example

A materials researcher computes a semiconductor's PDOS and sees oxygen p character below the Fermi level and metal d character in states above it. This helps describe possible orbital contributions to electronic transitions. The researcher then calculates an optical response or compares with spectroscopy rather than reading the Kohn–Sham band gap as a measured absorption onset. If a second functional opens the computed gap while preserving broad orbital character, that difference is a method sensitivity, not evidence that the material has physically changed. In a molecular analogue, conformational changes can reorder near-degenerate frontier orbitals without a corresponding abrupt chemical transformation.

Why?

Why does broadening change a plotted DOS peak's height? The underlying finite set of levels has fixed locations, but smoothing spreads each level's weight over an energy interval. A narrower normalized kernel makes a taller, sharper peak; a broader kernel makes a shorter, wider one. Comparing peak heights from plots with different broadening parameters can therefore invent an apparent electronic difference. The integrated weight over a properly normalized level remains linked to the number or projection weight of states.

Common misconception

“An orbital is a photographed region where one electron lives” confuses a model function with a unique particle location. “The HOMO–LUMO gap is the color of the molecule” ignores response, selection rules and relaxation. “A high PDOS peak proves an atom owns that many electrons” overlooks projection conventions and occupied versus empty states. “Any two energy diagrams can be overlaid directly” fails if their zero references or computational methods differ.

Worked example

Suppose a hypothetical molecule has occupied levels at −7.2, −6.5 and −5.8 eV and an empty level at −2.0 eV under one approximate DFT calculation. Its reported HOMO–LUMO eigenvalue gap is (−2.0) − (−5.8) = 3.8 eV. If a linear-response excited-state calculation predicts a permitted transition at 3.1 eV, the two values answer different questions and need not match. Now broaden each level with a normalized Gaussian of width 0.20 eV; doubling the width to 0.40 eV lowers and merges peaks without changing the listed eigenvalues. If a PDOS assigns 60% of the LUMO weight to one chosen fragment, that 60% is a projection under the stated scheme, not a measured electron fraction. The numbers are illustrative, not a prediction for a named compound.

Quick check

1. Does a molecular DOS plot necessarily represent continuous energy bands? Answer: No. A finite molecule has discrete one-electron levels; plotting software commonly broadens their spikes. 2. Why is an approximate DFT HOMO–LUMO gap not automatically an optical gap? Answer: An optical excitation includes electronic response and other effects not supplied by simply subtracting two approximate ground-state eigenvalues.

Exam focus

Define a molecular orbital, DOS and PDOS at the level of a computational model. State the energy reference and smoothing convention before comparing plots. Compute a frontier gap by subtracting eigenvalues, but label it as such. Explain one limitation of assigning chemical properties from orbital energies alone and suggest an appropriate total-energy or excited-state calculation for a quantitative observable.

Advanced insight

In extended solids, band energies vary with wave vector, and the DOS integrates over sampled reciprocal space. A k-point mesh and smearing parameter can strongly affect a plotted narrow feature. The fundamental gap of an interacting system has a derivative-discontinuity contribution absent from a naive semilocal Kohn–Sham eigenvalue gap. Even when exact-theory results motivate interpretation of a frontier eigenvalue, approximate functional and environmental errors may be large. Projected orbital character remains useful for forming hypotheses, particularly when supported by independent density, spectroscopy and total-energy checks.

Summary

Orbitals and density-of-states plots are structured descriptions of an electronic-structure model. They reveal symmetry, energy ordering and approximate atomic or fragment character, provided the method, projection, smearing and energy zero are stated. Their eigenvalue differences should not be silently converted into measured ionization or optical energies. Use the visual model to frame a chemical question, then choose the corresponding total-energy or response calculation for a quantitative answer.

Practice questions

1. A HOMO lies at −6.0 eV and a LUMO at −2.4 eV. What is the computed eigenvalue gap? Answer: 3.6 eV; this is not by itself an optical excitation energy. 2. Why can a finite molecule display a smooth DOS curve? Answer: Discrete orbital levels were broadened with a chosen kernel for visualization. 3. What additional information is needed before comparing two PDOS plots from different studies? Answer: At least method, geometry, energy reference, projection definition, normalization and broadening; periodic plots also need k-point context. 4. If a level has 70% metal d PDOS weight, does that uniquely define a physical electron location? Answer: No. The percentage comes from the chosen orbital projection and does not represent a unique measurement of electron ownership.