Problem oxygen potential energy scan using CCSD(T) and EA state PES using EA-EOM-CCSD

  1. I’m calculating PES of O2 using CCSD(T).
    I’m using the following input

$molecule
0 3
O
O 1 1.213185
$end

$rem
jobtype = pes_scan
frozen_scan = true
method = ccsd(t)
basis = aug-cc-pvtz
n_frozen_core = 0
scf_convergence = 8
max_scf_cycles = 500
cc_t_conv = 6

!ccman2 = true
$end

$scan
stre 1 2 0.80 8.0 0.10
$end

Problem: From the PES scan data when I’m trying to calculate Dissociation energy I’m getting DE as ~20 eV. Experimental data reported on NIST is ~5.11 eV.
What am I doing wrong?
Note: Same calculation I have done using ORCA 5.0. There I’m getting DE as ~5.4 eV which is close to experiment.

  1. I’m calculating EA states of O2 using EA-EOM-CCSD
    I’m using the following input

$molecule
0 3
O
O 1 1.213185
$end

$rem
jobtype = pes_scan
frozen_scan = true
method = eom-ccsd
basis = aug-cc-pvtz
n_frozen_core = 0
scf_convergence = 8
max_scf_cycles = 500
cc_t_conv = 6

ea_states = [4,2,3,3,2,3,3,3]
!ccman2 = true
$end

$scan
stre 1 2 0.85 3.1 0.11
$end

Problem: I’m unable to go beyond 3.1 Angstrom , I want the PES till 8.0 Angstrom.
What am I doing wrong?

UHF-CCSD(T) cannot describe the bond dissociations of triplet O2 or doublet [O2]-. These two problems are of significant multiconfigurational characters. It would be better to use multiconfigurational/multi-reference methods like CASSCF, CASPT2, NEVPT2, etc.

So, as per your suggestion I have done CASSCF followed by CASPT2 and My results are exact wrt experiment.

Is there any other method than EOM-EA-CCSD to calculate excited Electron attached state?
ORCA’s EOM-EA-CCSD doesn’t give good result, where as for QChem’s EOM-EA-CCSD, I’m unable to go beyond 3 angstrom.
Any suggestion?

Why can you not go beyond 3 A? Why not try doing the PES scan “by hand”, i.e., Run the calculation at 3 A (or whatever works successfully) to converge the MOs, then extend the bond distance a little bit in the subsequent job and use SCF_GUESS = READ. Note that the reproduce the effect of PES_SCAN, you would need to optimize with the bond length constrained.

I agree with what Prof. John Herbert said. For the PEC (potential energy curve) of diatomic molecules, it would be better to calculate one geometry by one, not via any PES scan keyword.

The electronic states of a diatomic molecule can be very tricky (especially for open-shell molecules). Even if one wants to read orbitals from a previous geometry, it is recommended to firstly calculate the geometry with the largest distance, and check the wave function stability of the ground state wave function, then read converged MOs from the previous geometry, say, 2.9 A reads MOs of 3.0 A.

For your asking electron attached state O2 → [O2]- problem, if you have to use single-reference methods, my suggestion is not to start from O2 or [O2]-, but start from [O2]^(2-). You can perform RHF for this dianion and check whether it is converged to (PI_u_2px)^2 (PI_u_2py)^2 (PI_g_2px)^2 (PI_g_2py)^2 configuration, a closed-shell state, and then use IP-EOM-CCSD or IP-EOM-CCSDT to make it lose one electron, so you obtain [O2]-. The basis set may be aug-cc-pVTZ. If you perform the calculation in this way, you start from a closed-shell RHF reference, thus the IP-EOM-CC results are expected to be very accurate. But anyway, this is just my suggestion. I believe Q-Chem developers and experiment users may have better ideas.

If you are still willing to use multi-reference methods, I think CAS(6o,8e), CAS(6o,9e) are expected to work perfectly for O2 and [O2]-, respectively. Electron affinity and excited electron attached state can also be calculated using multi-reference methods.

Best wishes,
Jingxiang

Thank you,
Prof. Herbert and Jingxiang.

I was able to perform the calculation as per your suggestion.

But, for few points I’m getting this error
Q-Chem fatal error occurred in module /scratch/qcdevops/jenkins/workspace/build_qchem_linux_distrib/tags/qc6113/qchem/ccman2/qchem/ccman2_main.C, line 26:

Key ‘f2_oo_float’ is not known.

Please submit a crash report at Q-Chem Crash Reporter

Can you tell me the reason behind it and how to debug it?

The input file given above runs to completion without errors using the most recent version of Q-Chem (v. 6.3.1). If you are experiencing a problem with an older version, please contact Q-Chem technical support to see if there is a workaround for your version.

thank you prof. for your valuable inputs.

I have few more queries.
I’m using CAP-EOM-EA-CCSD to find resonant states.
I want to know, is there any way to find out the ideal box dimension for CAP?

That is a question for the literature, I’m not an expert on those calculations.