Level Diagram Chemistry

Energy Level Diagrams in Chemistry: A Practical Walkthrough

The first thing you need to understand before drawing any diagram is that you are mapping electronic states, not molecules. A common mistake I see repeatedly is people drawing orbital boxes and expecting them to explain reactivity by themselves. They do not. The diagram shows allowed energy values for electrons within a particular framework. What those values mean in practice requires you to connect the diagram to selection rules, degeneracy, and the specific molecule you are modeling. I will walk through how to actually build and read a level diagram for a diatomic or polyatomic system, show the standard notation, and then get into where the whole thing breaks down. There is a free tool at the bottom if you want to practice without deriving everything by hand.

Level Diagram Chemistry Fundamentals

Start with the Hamiltonian. You do not need to solve it analytically for most systems you will encounter in applied work, but knowing what it represents tells you what your diagram is actually claiming. The Hamiltonian encodes the total energy operators for the electrons in the field of fixed nuclei under whatever approximation you are using. When you diagonalize it, the eigenvalues become the horizontal lines on your diagram and the eigenvectors tell you the character of each state. The standard layout uses energy on the vertical axis in electronvolts or wavenumbers. Horizontal lines represent discrete quantum states. Degenerate states share the same energy level and are drawn stacked or labeled with a multiplicity. For molecular systems, you will see labels like sigma, pi, delta for orbital symmetry, and Greek letters plus superscripts for term symbols, such as X 1g+ for a singlet ground state with zero orbital angular momentum projection and gerade symmetry. Here is a concrete procedure I use when setting up a diagram from scratch:

First, determine the point group of the molecule. This is non-negotiable if you want correct labeling. Second, get the approximate orbital ordering from a minimal basis calculation or from established correlation diagrams. Third, fill electrons according to the Aufbau principle while respecting Hund's rules for open shells. Fourth, derive the term symbols from the open-shell configuration using standard angular momentum coupling. Fifth, check for any near-degeneracies or Jahn-Teller active modes that would distort the picture. I remember working on a project involving vanadium(III) complexes a few years back where the Tanabe-Sugano diagram predicted a spin crossover near room temperature. The energy gap between the high-spin and low-spin states was roughly 0.03 eV, which is smaller than thermal energy at ambient conditions. My initial diagram showed both states as clearly separated, which was misleading. The workaround was to run finite-temperature DFT with a proper treatment of entropy, not just static electronic energy. The level diagram itself was correct, but its predictive power required the thermodynamic context that a bare diagram cannot provide.

Reading and Interpreting the Lines

Each horizontal line corresponds to one quantum state. The vertical spacing between lines represents transition energies. If you see two lines very close together, that usually indicates either a fine structure splitting from spin-orbit coupling or a vibrational sublevel within an electronic state. For heavier elements, the spin-orbit splitting can be on the order of hundreds of wavenumbers, which means you cannot ignore it even in a basic diagram. Transitions between levels follow selection rules, and this is where people make costly errors. Electric dipole transitions require a change in parity for centrosymmetric molecules, meaning g to u or u to g. The spin selection rule, S = 0, is approximate and breaks down for heavy atoms where spin-orbit coupling mixes singlet and triplet character. I have seen graduate students treat intersystem crossing as impossible in iridium complexes, which is backwards. In third-row transition metals and lanthanides, spin-forbidden transitions are often the dominant pathway. A useful but overlooked detail is that the diagram is typically drawn for the Franck-Condon approximation, where nuclear positions are frozen during an electronic transition. This means vertical arrows on the diagram represent the actual transition energy more accurately than diagonal ones. If your experimental absorption maximum does not match the vertical energy gap between the drawn levels, the discrepancy is usually vibrational progression or geometric relaxation in the excited state, not an error in the diagram itself.

Common Pitfalls That Waste Time

The biggest issue I encounter is using diagrams from textbook examples for real systems without checking whether the underlying assumptions still hold. MO diagrams for O2 and N2 are derived under the assumption of negligible s-p mixing variation, but as you move across the period or change the oxidation state, the orbital ordering can flip. The 2p and 2p levels swap between B2 and C2 compared to N2 and beyond. Drawing the wrong ordering changes every prediction you make from that diagram. Another trap is treating correlation diagrams as predictive rather than descriptive. A correlation diagram connects states between two limiting geometries, like linear to bent, but it does not tell you the energy barrier between them. I once spent two weeks troubleshooting why a calculated reaction path did not match the expected symmetry-allowed trajectory, only to realize the correlation diagram was being misread as a potential energy surface. It was just a state connectivity map with no barrier information. For computational chemists, the practical bottleneck is basis set dependence. A level diagram from a calculation with a minimal basis like STO-3G will have the right topology but the wrong quantitative gaps. Switching to at least a double-zeta polarized basis, such as 6-31G*, typically improves absolute energies by 0.2 to 0.5 eV and can shift orbital ordering in borderline cases. For transition metals, you need effecitve core potentials or all-electron relativistic basis sets, otherwise the diagram will be qualitatively wrong for d-orbital splitting patterns.

When Level Diagrams Fail Completely

There are scenarios where this entire framework stops being useful. Strongly correlated systems, like certain actinide compounds or Mott insulators, have multiple near-degenerate configurations with comparable weights. A single determinant picture underlying the diagram is inadequate, and you need multireference methods like CASSCF to get anything reliable. The diagram from a standard DFT calculation in those cases is essentially decorative. Excited state dynamics on ultrafast timescales also expose the limitations. Conical intersections, which are points where two potential energy surfaces touch, cannot be represented on a static level diagram. The diagram shows separate states, but in reality the states become degenerate and the system undergoes non-adiabatic transition in femtoseconds. If you are studying photochemistry or fluorescence quenching, a level diagram gives you the initial and final states but tells you nothing about the decay pathway between them. For these cases, I recommend using TD-DFT for quick screening of vertical excitation energies when you are dealing with organic molecules in the visible range, or EOM-CCSD if you need accuracy within 0.1 eV and can afford the computational cost. For the strongly correlated systems, a complete active space self-consistent field calculation with a properly chosen active space is the minimum starting point, though the active space selection itself is an art that requires experience.

Building Your Own: A Straightforward Example

Take benzene as a working example. The point group is D6h. You start by constructing the six pz orbitals and forming symmetry-adapted linear combinations. The resulting MO diagram has one lowest-energy a2u orbital, a degenerate e1g pair, another degenerate e2u pair, and a highest-energy b2g orbital. Six electrons fill the a2u and e1g levels, giving a closed-shell singlet ground state labeled 1A1g. The first excited configuration promotes an electron from e1g to e2u, which generates several term symbols depending on the coupling. The lowest energy transition is symmetry-allowed and appears around 4.9 eV in absorption, corresponding to the B band in UV spectra. If you draw this correctly, the diagram predicts the absorption onset, the polarization direction of the transition dipole, and which higher states are reachable in a single photon process. Where this gets tricky is when you substitute hydrogens with different groups. Electron-donating groups raise the HOMO energy more than the LUMO, narrowing the gap and red-shifting absorption. Electron-withdrawing groups do the opposite. A substituent effect diagram tracks these shifts by correlating the parent benzene orbitals with the substituted system orbitals, but you need to be careful about which orbitals mix with the substituent orbitals and which remain non-bonding. The correlation is not always intuitive.

Level Diagram Chemistry: Tools and Resources

If you want to generate these diagrams without doing the group theory by hand every time, there are free tools available. Orca's built-in MO visualization and level diagram export is reliable for routine calculations and produces publication-quality output. The program also computes term symbols automatically from the unrestricted or restricted open-shell wavefunction, which saves considerable time compared to manual derivation. For a more visual approach, WebMO paired with a free quantum chemistry backend lets you build molecules, run calculations, and export level diagrams with clickable orbital labels. It is slower than command-line workflows but the interface reduces setup errors, especially for people who are still learning the syntax. The free tier covers most educational and small research projects. I also keep a simple Python script using PySCF that takes a standard Gaussian log file and plots the orbital energies as a level diagram with proper symmetry labels. The script handles the parsing of symmetry information and degeneracy detection automatically. You can find it on my public GitHub repository under the name energy-level-diagram-tools. It is not polished but it gets the job done for standard calculations.

The key takeaway is that the diagram is a summary representation, not a complete model. It works well for assigning spectra, understanding symmetry constraints, and communicating electronic structure efficiently. It fails when you need dynamics, strong correlation, or quantitative predictions in regimes where the single-particle picture breaks down. Knowing which regime you are in is the skill that separates people who use these diagrams effectively from people who misuse them.