You run a TDDFT calculation on a donor-acceptor dye molecule using a GGA functional and find the predicted lowest excitation energy is 1.2 eV below the experimental absorption peak. Which explanation is most likely correct?
AThe GGA functional overestimates the exchange-correlation energy, artificially stabilizing the excited state
BThe calculation is using the wrong basis set, and a larger basis set would correct the error
CThe transition is a charge-transfer excitation, and GGA functionals severely underestimate CT excitation energies due to insufficient long-range exchange
DTDDFT is not applicable to dye molecules because they absorb only in the visible range
GGA and other semilocal functionals contain a self-interaction error that fails to properly penalize long-range separation of electron density. For charge-transfer excitations — where an electron moves from a donor fragment to an acceptor fragment far away — this error causes dramatic underestimation of excitation energies, often by 1–2 eV. Range-separated hybrid functionals (CAM-B3LYP, ωB97X) correct this by incorporating increasing Hartree-Fock exchange at long range. Basis set size (option B) is a second-order concern compared to functional choice for this class of excitation.
Question 2 Multiple Choice
What does a TDDFT linear response calculation directly compute, and what does it NOT compute?
AIt computes the absolute total energy of excited states; it does not compute oscillator strengths
BIt computes excitation energies (energy differences from the ground state) and oscillator strengths; it does not compute the absolute energy of excited states
CIt computes excited-state wavefunctions explicitly; it does not use the ground-state density
DIt computes excited-state geometries; it does not predict absorption wavelengths directly
TDDFT linear response calculates excitation energies — the energy required to promote the system from the ground state to each excited state — along with oscillator strengths that predict the intensity of each transition. These are the eigenvalues and related quantities from the Casida equations. TDDFT does NOT compute absolute excited-state energies (which would require placing the total energy on an absolute scale). A common misconception is treating the excitation energy as equivalent to an excited-state wavefunction or total energy; it is simply the gap between ground and excited state energetics.
Question 3 True / False
Standard GGA TDDFT functionals are equally reliable for valence excitations (local transitions) and charge-transfer excitations in large molecules.
TTrue
FFalse
Answer: False
This is the central practical limitation of TDDFT. For valence excitations — where the excited electron stays near its origin — GGA functionals perform reasonably well because the electron density doesn't move far. For charge-transfer excitations — where density shifts across the molecule — GGA functionals dramatically underestimate excitation energies because they lack the long-range exchange needed to correctly penalize charge separation. Range-separated hybrids (CAM-B3LYP, ωB97X) address this by blending in Hartree-Fock exchange at long range, but local functionals cannot be used reliably for CT states.
Question 4 True / False
TDDFT requires only a completed ground-state Kohn-Sham DFT calculation as input, making it far less computationally expensive than wavefunction-based excited-state methods.
TTrue
FFalse
Answer: True
This is the core practical appeal of TDDFT. The Casida equations — the eigenvalue problem that yields excitation energies — are constructed entirely from ground-state Kohn-Sham orbitals and orbital energies, plus the exchange-correlation kernel. No explicit excited-state wavefunctions are constructed. The main additional cost is one matrix diagonalization. By contrast, methods like EOM-CCSD or CASSCF/CASPT2 require substantially more computation because they explicitly construct multi-electron excited-state representations.
Question 5 Short Answer
Why do range-separated hybrid functionals perform significantly better than GGA functionals for charge-transfer excitations in TDDFT, and what physical feature of GGA functionals causes the failure?
Think about your answer, then reveal below.
Model answer: GGA functionals have a self-interaction error: an electron's Coulomb repulsion is not fully cancelled by the exchange-correlation term for long-range separations, making it artificially cheap to move electron density far across a molecule. For charge-transfer excitations, where an electron moves from donor to acceptor groups separated by distance, this error causes the excitation energy to be dramatically underestimated. Range-separated hybrid functionals correct this by including 100% Hartree-Fock exchange (which is self-interaction-free) at long interelectronic distances while retaining DFT exchange at short range. This asymptotic correction restores the correct long-range behavior of the exchange potential and gives accurate CT excitation energies.
The failure of GGA for CT states is not just numerical imprecision — it reflects a fundamental theoretical deficiency. The exact exchange-correlation potential must decay as -1/r at large distances (the image charge theorem), but approximate functionals violate this. Hartree-Fock exchange has the correct -1/r asymptotics, which is why range-separated hybrids that switch to 100% HF exchange at long range fix the problem.