RSPCTL WARNING-MAXIMUM NUMBER OF MICROITERATIONS

[ndhieu92]

WIsh you have a good day everyone.

I am trying to calculate for the spin-orbit coupling constant, but I have got a problem. Eventhough the calculation finished without any error, but the result of SOC value between 2 times of calculation is different between each other. Firstly, I checked the input file (mol. and dal. file) and I am sure that it is correct. So I checked the output file and then I always see the warning is: RSPCTL WARNING-MAXIMUM NUMBER OF MICROITERATIONS, 60. So now I am trying to add into dal. file: .MAX MICRO ITERATIONS and increase it into 200. Will this fix my problem? I attached 2 output file and also the new dal.file with my thinking. Please let me know your idea.

Thank you so much

[hjaaj]

Your spin-orbit coupling constants are garbage, because your calculation is triplet unstable (complex eigenvalues in the calculation of lowest triplet excitation). The differences between the two calculations are caused by numerical noise because of the non-convergence.

Either your DFT iterations do not converge to the ground state (do you get same total energy if you try with another program?), or B3LYP is a bad model for this case.

By the way, your ".MAX IT" keyword has no effect because the ".MAX IT" in that place specify the max. number of geometry iterations in a geometry optimization. But it does not help to increase the correct max iterations for response, because the convergence is stuck because of the above mentioned problems.

[ndhieu92]

Your spin-orbit coupling constants are garbage, because your calculation is triplet unstable (complex eigenvalues in the calculation of lowest triplet excitation). The differences between the two calculations are caused by numerical noise because of the non-convergence.

Either your DFT iterations do not converge to the ground state (do you get same total energy if you try with another program?), or B3LYP is a bad model for this case.

By the way, your ".MAX IT" keyword has no effect because the ".MAX IT" in that place specify the max. number of geometry iterations in a geometry optimization. But it does not help to increase the correct max iterations for response, because the convergence is stuck because of the above mentioned problems.

Thank you Dr. Hans for your reply. I am sorry but may I ask you a question: Do you mean the SCF energy which was shown in the output file is total energy? And I am thinking to try with another program like molpro to confirm it. I will try to change to MCSCF to see how it works.

[hjaaj]

I mean this final energy:

Code: Select all

@    Final DFT energy:          -2824.469146640524  

and these final HOMO-LUMO energies:

Code: Select all

    E(LUMO) :    -0.13437205 au (symmetry 1)
  - E(HOMO) :    -0.14494639 au (symmetry 1)
  ------------------------------------------
    gap     :     0.01057434 au

Note that the HOMO and LUMO orbital energies are very close, so no wonder there is triplet instability.

[ndhieu92]

I mean this final energy:

Code: Select all

@    Final DFT energy:          -2824.469146640524  

and these final HOMO-LUMO energies:

Code: Select all

    E(LUMO) :    -0.13437205 au (symmetry 1)
  - E(HOMO) :    -0.14494639 au (symmetry 1)
  ------------------------------------------
    gap     :     0.01057434 au

Note that the HOMO and LUMO orbital energies are very close, so no wonder there is triplet instability.

Maybe I will try to recalculate with gaussian 09 software with same basis set B3LYP/6-31G* in dal.file to confirm the DFT energy and HOMO-LUMO energies.
I will keep updating the result.
Thank you so much Dr. Hans

[xiongyan21]

Dear Prof. Jensen
Could you please give me an rough idea of how large the energy gap between LUMO and HOMO of the triplet state of a molecule would be, because in the uhf spin-flip TDDFT geometry search of thiabendazole with PCM, also can be very small. [/b] The advantage of the spin-flip triplet geometry, MECIs searches lies in it does not need an inital geometry guess.
I will let it supass Cs symmetry beginning and retry.

[xiongyan21]

After changing a functional and using C1 symmetry from the very beginning, the apparently successful UHF spin-flip TDDFT search of the triplet state of thiabendazole with PCM using GAMESS gives perhaps satisfactory HOMO and LUMO gaps, respectively. Of course, spin-flip TDDFT may perfer to some functionals. I use wxMacMolplt to view the orbital energies.

[xiongyan21]

I have done the virbational analysis of the triplet state using GAMESS and found there is no imaginary frequence, but it has a warning meaning it is not a true stationary point although the convergence criteria are met, thus HESSIAN analysis is not very rigorious. Perhaps with NWCHEM, there is no such warning.

[xiongyan21]

It seems this can be understood according to what Dr. Schmidt from Ames Lab told me before that HOMO an LUMO are not very suitable concepts.

[xiongyan21]

Dear Dr. Jensen
Yes, you are right.
According to the article of Prof. Anna Krylov in University of Southern California, the traditional single reference method for excited states will fail when the subspaces of the occupied and virtual become nearly degenerate, which is also true for ground states. According to an published article, SF-TDDFT can only deal with one electron excitation and doubly excited states, GAMESS can give energie gaps of the singlet to triplet, triplet to triplet, and triplet to mixed.

[xiongyan21]

The article Determination of thiabendazole in foods and waters by solid-phase transmitted room-temperature phosphorescence said the enhanced 472 nm is its naive phosphorescence.

[xiongyan21]

6-311++G(2d,2p) can make thiabendazole produce a small imaginary frequency with b3lyp geometry around 55 cm-1 using Cs symmetry and HF frequency analysis, perhaps caused by rotation, etc., while 6-31+(d,p) and b3lyp cannot when using GAMESS. Although it may really does not matter for the excitation energy calculation, it is perhaps a problem for the THz calculation for it.
Somebody recommended the decrease of the symmetry, e.g., using C1, to eliminate it. I have just tried thiabendazole, which also can be considered as to be planar even in C1, using gamess with 6-311++G(2d,2p) and b3lyp, and the approximate 55 cm-1 imaginary frequency has been removed during HF vibrational anaysis with the warning.
For the molecule thiol, if mp2 is used for its geometry optimization, 6-311++(3df, 2p ) is required to eliminate it with seminumerical mp2 gradients.

[xiongyan21]

Thiabendazole SF-TDDFT optimized geometry give the triplet to singlet emission energy having a large discrepency ( shorter about 76 nm )from the experimental enhanced data in Determination of thiabendazole in foods and waters by solid-phase transmitted room-temperature phosphorescence.
It is said in the paper the sample and enhancer solutions were deposited on a slip of paper (Whatman No. 4), previously humidified with pH 4.0 acetic/acetate buffer solution, after which the paper was dried and placed between 2 quartz plates. The diffuse transmitted phosphorescence intensity was measured directly at excitation of 303 nm and emission of 472 nm.

Monte Carlo search perhaps is needed for carbendazole first for the ground state due to its rotation，facilitating the geometry optimization at least.

[xiongyan21]

The article Highly sensitive and selective room-temperature phosphorescence determination of thiabendazole by the supramolecular interaction of thiabendazole/β-cyclodextrin/triton X-100 gives λex/λem = 298/481 nm resulting from a 1 ∶ 1 ∶ 1 β-cyclodextrin (β-CD)/thiabendazole (TBZ)/triton X-100 (TX-100) supramolecular ternary inclusion complex.

Flow injection analysis with on-line nylon powder extraction for room-temperature phosphorescence determination of thiabendazole gives maximum λem = 505 nm.

[hjaaj]

Dear Yan Xiong

Sorry, but your comments are not really related to the original posting in this thread by Hieu Nguyen. Furthermore, you discuss items (as UHF spin-flip and SF-TDDFT) which are not implemented in dalton, and therefore the dalton forum is not the right place for such a discussion.

[xiongyan21]

Dear Dr. Jensen
Sorry for this.

Thanks, anyway.

Very Best Regards!

[xiongyan21]

Since the triplet found in water is far from these published experimental data using BHHLYP, perhaps it is not useful for the phosphorescence of this molecule.

Please note GAMESS has incorrect analytic gradient for meta-GGA with SF-TDDFT for now.
GAMESS gives a deviance of 0.367 eV from 478nm using b3lyp and an sperical Dunning basis set in ethanol, and with sperical 6-31G+(d,p), it becomes 0.33 eV, questionable with the default grid.
Photophysical Properties of Benzimidazole and Thiabendazole and Their Homologues. Effect of Substituents and Solvent on the Nature of the Transition gives 453,478, and 500nm in ethanol at 77K with the lifetime of 157 ms.
The article Calculation of the Phosphorescence of Porphyrins by the Density Functional Method got 1.11 eV of the gap of T1 and S0 using 3-21G and B3LYP by DALTON。 The authors think 3-21G underestimated the energy of the T1 state in comparison with the experimental value( delta E=1.58 eV). Actually, what is calculated in the reference is porphin.
I have used UHF SF-TDDFT to calculate the T1 to So gap of porphyrin at nonplanar initial geometry with 3-21G and BHHLYP in toluene with GAMESS, got the transition energy of about 1.36 eV.
Theoretical DFT study of phosphorescence from porphyrins gives the gap to 3B2u of 1.46 eV using 6-31G(d,p) optimized ground state geometry with b3lyp.
Porphyrin Free Base Phosphorescence gives the heavy atom enhanced phosphorescence of 795 nm by time resolved spectra in 50% EPA and 50% ethyl iodide at 77K for proto-porphyrin.

[xiongyan21]

It is noted in the recently released GAMESS source code, including 2019.R1, the test, i.e., UHF SF-TDDFT geometry search of the triplet state of H2CO, not converged for GCC greater than 4.4.
Thus, perhaps, fix is wanted.

[magnus]

The Dalton forum is also not the right forum to discuss GAMESS calculations.

[xiongyan21]

UHF-SF-TDDFT using accd and b3lyp gives the deviance of 0.346eV from 478 nm in ethanol with the increased grid, still questionable and requiring further increases, above the default, for thiabendazole
Since it is particularly difficult for the quantum chemical caculation of the signlet-triplet gap with the geometry search of the triplet of thiabendazole, i.e., quantitative mismatich between the theoretical and the experimental results, if allowed, FCI, MCSCF, etc., or explicit solvation model might be carried out for the related search.