Conceptual question regarding the applicability of SF-TDDFT method

Aarzoo · January 4, 2026, 1:30pm

Hi everyone,

I have a conceptual question regarding the applicability of SF-TDDFT method.

I am studying a benzene + O(¹D) system, where benzene is in its closed-shell singlet ground state, and the oxygen atom is in its excited singlet ¹D state. At large separation, the overall system is therefore singlet–singlet.

My question is:
Can this system be meaningfully treated using the SF-TDDFT framework?

As I understand it, SF-TDDFT requires a high-spin reference state (e.g., a triplet), from which lower-spin states are accessed by spin-flip excitations. However, in the benzene + O(¹D) case, both fragments are singlets.

Is it inappropriate to use SF-TDDFT for this system, at least in the non-interacting or weakly interacting regime? If so, would conventional TDDFT or a multireference approach be more appropriate for describing excited-state dynamics?

I would appreciate any clarification or guidance on how one should properly set up excited-state calculations for this type of system in Q-Chem.

Thank you very much.

I tried to put a job as well:
input:

$molecule
0 3
  H      1.954110     1.501681    -0.000017
  C      1.099407     0.844699    -0.000021
  C      1.281178    -0.530526    -0.000028
  H      2.277161    -0.943033    -0.000032
  C      0.181436    -1.373577    -0.000037
  H      0.322351    -2.442191    -0.000046
  C     -1.099388    -0.844754    -0.000030
  H     -1.954087    -1.501741    -0.000034
  C     -1.281160     0.530470    -0.000016
  H     -2.277141     0.942980    -0.000009
  C     -0.181409     1.373524    -0.000016
  H     -0.322324     2.442138    -0.000008
  O      0.000022    -0.000010    -3.000025
$end

$rem
   BASIS  =  6-31G*
   JOB_TYPE  =  sp
   METHOD  =  BHHLYP
   SYMMETRY_IGNORE = TRUE
   MAX_CIS_CYCLES = 500
   MAX_SCF_CYCLES = 500
   THRESH = 14
   SPIN_FLIP = true
   CIS_N_ROOTS = 6
   CIS_STATE_DERIVATIVE = 1
   sts_mom = true
   UNRESTRICTED = true
   SCF_CONVERGENCE  =  8
$end

and output is:

        SF-DFT Excitation Energies

(The first “excited” state might be the ground state)

Excited state 1: excitation energy (eV) = 1.6066
Total energy for state 1: -307.08509720 au
<S**2> : 2.0054
D( 16) → S( 1) amplitude = 0.6923
D( 17) → S( 2) amplitude = -0.6924

Excited state 2: excitation energy (eV) = 2.6689
Total energy for state 2: -307.04605844 au
<S**2> : 0.0190
D( 16) → S( 1) amplitude = -0.2698
D( 16) → S( 2) amplitude = 0.6312
D( 17) → S( 1) amplitude = -0.6326
D( 17) → S( 2) amplitude = -0.2700

Excited state 3: excitation energy (eV) = 2.6689
Total energy for state 3: -307.04605836 au
<S**2> : 0.0190
D( 16) → S( 1) amplitude = -0.6320
D( 16) → S( 2) amplitude = -0.2697
D( 17) → S( 1) amplitude = 0.2701
D( 17) → S( 2) amplitude = -0.6317

Excited state 4: excitation energy (eV) = 3.7159
Total energy for state 4: -307.00758397 au
<S**2> : 0.2174
D( 16) → S( 2) amplitude = -0.5899
D( 17) → S( 1) amplitude = -0.5868
S( 1) → S( 2) amplitude = 0.3663 alpha
S( 2) → S( 1) amplitude = 0.3726 alpha

Excited state 5: excitation energy (eV) = 4.1273
Total energy for state 5: -306.99246574 au
<S**2> : 1.0257
S( 1) → S( 1) amplitude = -0.6000 alpha
S( 1) → S( 2) amplitude = -0.1699 alpha
S( 2) → S( 1) amplitude = -0.2185 alpha
S( 2) → S( 2) amplitude = 0.7367 alpha

Excited state 6: excitation energy (eV) = 4.1631
Total energy for state 6: -306.99114905 au
<S**2> : 1.0124
S( 1) → S( 1) amplitude = 0.2073 alpha
S( 1) → S( 2) amplitude = -0.6416 alpha
S( 2) → S( 1) amplitude = 0.6714 alpha
S( 2) → S( 2) amplitude = 0.2201 alpha

AnnaKrylov · January 9, 2026, 2:02am

Yes, it is a good case for SF-TDDFT calculations. Your input os fine and the output makes sense: your lowest TDDFT state is triplet – probably corresponding to Bz(singlet)…O2(triplet) – and the first excited state is about 1.1 eV above is a singlet-- probably corresponding to Bz(singlet)…O2(singlet). I am not sure what your final goal is, but I would recommend the following:

To confidently assign states, use NO analysis (deployed with state_analysis true)
You can improve the description using MR-SFDFT (new feature in qchem 6.4)
For this size system, you can easily run EOM-SF-CCSD – at least you can use these data to validate SF-DFT.
Another method that might be useful but requires caution is EOM-DIP-CCSD (can only be reliably used with compact basis set, without diffuse functions).

You may find the following papers relevant:

D. Casanova and A. I. Krylov, Spin-flip methods in quantum chemistry, Phys. Chem. Chem. Phys. 22 , 4326 – 4342 (2020)
N. Orms, D. R. Rehn, A. Dreuw, and A. I. Krylov, Characterizing bonding patterns in diradicals and triradicals by density-based wave function analysis: A uniform approach, J. Chem. Theo. Comp. 14 , 638 – 648 (2018)
C.A. Taatjes, D.L. Osborn, T.M. Selby, G. Meloni, A.J. Trevitt, E. Epifanovsky, A.I. Krylov, B. Sirjean, E. Dames, and H. Wang, Products of the benzene + O(3P) reaction, J. Phys. Chem. A 114 , 3355 – 3370 (2010)