Page **2** of **3**

### Re: Can Dalton calculate oscillator strength

Posted: **27 Jul 2020, 07:02**

by **xiongyan21**

CAS 1185987-44-3 has been optimized with 6-311++G(2d,2p), and the excitation energies in hexane have been calculated with 6-31+G(d,p), all with GAMESS, finding three weak absoptions together can give the first experimetal peaks in an article published in China using Shimadzu LC一20AD where the middle one, with the largest oscilator strength, has an error of around 0.12 eVs, but the one of 204 nms caused by charge transfer fails.

CAMQTP01 in GAMESS, a functional modified for vertical ionization potential calculation, contributed last year by Dr. Gerasimov, in A.N.Nesmeyanov Institute of Organoelement Compounds of Russian Academy of Sciences, now is being tried.

The optimization using Dalton2018 with b3lyp and 6-31+G* has been finished, and 6-31++G(d,p) optimization with that also has been completed.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **28 Jul 2020, 09:09**

by **xiongyan21**

During the aug-cc-pvdz and b3lyp calculation of the above using 6-311++G(d,p) with b3lyp optimized geometry, at the beginnning of the DFT response employing Dalton2018, there is a warning

********* WARNING ORBDIA ***********

DIAGONAL ORBITAL HESSIAN HAS 1 NEGATIVE ELEMENTS

...

Perhaps .INA setting is not very suitable.

The result is completely wrong

...

@ Excited state no: 1 in symmetry 1 ( A ) - singlet excitation

-------------------------------------------------------------------

@ Excitation energy : 4.54909991E-02 au

@ 1.2378731 eV; 9984.1202 cm-1; 119.43660 kJ / mol

@ Total energy : -679.67677 au

@ Operator type: XDIPLEN

@ Oscillator strength (LENGTH) : 0.24231228 (Transition moment : 2.8266406 )

@ Operator type: YDIPLEN

@ Oscillator strength (LENGTH) : 2.23750730E-02 (Transition moment : 0.85894454 )

@ Operator type: ZDIPLEN

@ Oscillator strength (LENGTH) : 2.78751923E-03 (Transition moment : -0.30317391 )

...

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **29 Jul 2020, 06:01**

by **xiongyan21**

An article in China suggested that it has two maximum wavelengths at 204 and 265 nm in 70% hexane, and said the area of the first one is very large, using Shimadzu LC-20AD.

CAMQTP01 gives the first one. the fouth, 0.06 eVs deviating from the experimental data, using 6-31++G(d,p), but it gives the initial two of very small oscillator strengthes and small trasition dipoles, meaning " dark", and the third one divating 0.79 eVs from experimental data.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **01 Aug 2020, 00:27**

by **xiongyan21**

The 6-311++G(2d,2p) B3lyp frequency analysis with GAMESS has no imaginary frequency and no warning, meaning a true minimum obtained on the PES, which was first through Monte Carlo search. That is good because it seems the optimization is not easy.

I have used RHF and 6-31++G** to calculate ECD by Dalton2018 , but the third one, with oscillator strength around 0.45, is significant deviating form the experimental value, and velocity, length and London of rotational strength are all negative values ranging from around -47 to -50.

Has Dalton evolved to surpass the calculation of pi-pi* and sigma-pi* for ECD, or are the lengths and velocities correct even if the excitation energies have large discrepencies when compared with experimental ones?

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **03 Aug 2020, 10:53**

by **xiongyan21**

CAS 864237-81-0, a pesticide, is difficult to optimize, because it requires a lot of Monte Carlo search steps, and the final geometry is obtained using 6-311++G(2d,2p) and B3lyp, all with GAMESS. PCM in water, 6-31++G(d,p) and b3lyp excitation energies can be compared with the expeimental data in an article published in China. According to the calculation, the second absorption can give the data which can be compared with the experimental data because the first one is very weak. The experimental data shows one terrace and one slope nearby, and the calculation gives five absorptions around there where three of them are very weak, but it fails to predict the peak in 200 nms. The functional used for ionization potential fails for the excitation energy. If possible, I will also use Dalton2018 to study whether the two softwares can agree.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **07 Aug 2020, 10:37**

by **xiongyan21**

It should be noted that if the small basis set is used, the excited energies calculated near 200 nms may not be compared with experimental data in terms of eV., and perhaps even using aug-cc-pvdz and any functional, TDDFT results may also fail in the VUV and other regions in terms of eV.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **12 Aug 2020, 10:49**

by **xiongyan21**

Are there methods in Dalton 2018 to treat double excitations and above in the existence of solvation?

I don‘t know whether Prof. Truhlar's new method, the combination of MCSCF and DFT together, which may always overcome the difficulties of KS-DFT in multiconfigurational problems, transition metal chemistry, charge transfer, etc., or MCSCF in GAMESS can include solvation, because I haven't tried this.

Prof. Truhlar's group has demonstrated its abilities in several pubilished articles, e.g,, excitation spectra of retinal in vacuum, which has experimental data.

Very Best regards!

### Re: Can Dalton calculate oscillator strength

Posted: **13 Aug 2020, 12:11**

by **xiongyan21**

TD-CIS in GAMESS using 6-31++G(d,p) predicts the oscillator strength of the first and the fourth excitation energies are very small, and the transition dipole of the z direaction and the oscilator strength of the fourth is only the half of the first, and the excitation energy of the first is far from the experimental data, but the fifth, which can be compared with the experiment data(error: 0.27 eVs and aug-cc-pvdz: smaller) if the order is correct, and sixth, not shown completely in the experiment, are very strong. TD-CIS fails to give the relatively large oscilator strength of the first absorption, the terrance as well as the sople.

It seems the fifth one is of charge transfer character, and perhaps of multiconfiguration nature, thus whether TD-CIS is useful for this is questionable., but b3lyp can reproduce the peaks above 200 nms.

I wonder whether Dalton2018 can do TD-CISD.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **14 Aug 2020, 14:21**

by **hjaaj**

As long as you can specify the desired CISD with the RAS specifications, and it fits in memory and does not exceed some internal limits, then you can do a TD-CISD.

To do that you ask for .CI in the wavefunction part, and then under **RESPONSE the keywords for excitation calculation and oscillator strengths.

### Re: Can Dalton calculate oscillator strength

Posted: **16 Aug 2020, 07:03**

by **xiongyan21**

Dear Prof. Jensen

Thanks a lot for your explanation.

I first try TDDFT in TDA with Dalton2018, know TD-CIS should be done also with the specification of .CI, and am going to try TD-CISD.

By the way, spin-flip TDDFT geometry search with GAMESS using b3lyp and 6-31G(d,p) of phenyl methyl ketone can give T1-S0 vertical gap in ethanol, where the ground state targeted is not spin-contaminated at all, within 0.06 eVs deviating from the 0-0 band summarized by Prof. Wagner et al. previously in Communications to the Editor in JACS, and there are two imaginary frequencies during the following Hessian analysis below 380 cm^-1. The imaginary frequencies perhaps should be eliminated. Perhaps I should follow the test using spin-flip TDDFT instead of DFT to do Hessian, so the Hessian analysis here is not proper for ZPVE.

I will use Dalton2018 to calculate the phosphorescence lifetime.

I don't know whether Monte carlo minimum search, illustrating Monte Carlo AI, can be applied first to facilitate an triplet geometry optimization.

Very Best regards!

### Re: Can Dalton calculate oscillator strength

Posted: **18 Aug 2020, 07:42**

by **xiongyan21**

Using the triplet geometry found by Spin-Flip TDDFT in ethanol with GAMESS and with the following input for Dalton 2018 and 6-31G

**DALTON INPUT

.DIRECT

.PARALLEL

.RUN RESPONSE

**INTEGRALS

.SPIN-ORBIT

**WAVE FUNCTIONS

.DFT

B3LYP

*SCF INPUT

**RESPONS

*QUADRATIC

.ECPHOS

.PHOSPHORESCENCE

.PRINT

10

.ROOTS

3

**END OF INPUT,

the spin-orbit gap got in vacuum is within 0.04 eVs deviating from that got in ethanol with GAMESS SPIN-FLIP TDDFT, and the phosphorescence lifetime is around 0.3 ms(in terms of effective charge spin-orbit integrals) deviating from the experimental 8 ms in ethanol, i.e., the short-lived emission. It is estimated that in solvents, the long-lived phosphorescence is 75-76 kcal and the lifetime is from 0.35 to 2.25 seconds in solvents or gel; Dalton2018 gives one agreeing with the above energy and the other of about 1 second to 1.3 seconds in vacuum.

Prof. Wagner suggested the short-lived one was an mixture of the two triplet states, and the longer-lived one, which was calculated (not directly from experiments) by Prof. Lamola，who published the experimental data in another reference, in ethanol, might well involve the nonequilibrium loss an alpha proton for other ketones, but Prof. Lamola gave another explanantion first.

We can see spin-flip TDDFT method from the group headed by Prof.A.I.Krylov, once an M.Sc in Moscow State University, a PhD in The Hebrew University of Jerusalem, a post PhD in Prof. M. Head-Gordon's group, and now a US university distinguished professor having winned Dirac medal,Theoretical Chemistry Award, and Bessel Research Award, can facilitate the phosphorecence emission wavelength estimation through the vertical T1-S0 gaps directly provided after one run. They have developed and are developing other methods, eg. spin-flip coupled cluter methods(not in GAMESS), by spin-flip ansatz used for the singlet-triplet gaps of diradicals, triradicals, bond breaking situations, etc. as well as for the search of conical intersections, and are applying the tools in combustion, bioimaging, solar energy, etc., collaborating with experimental laboratories.

Thanks a lot for the suggestion by Prof. Ruud of using the triplet structure found by other software to calculate its phosphorescence lifetime.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **22 Aug 2020, 05:32**

by **xiongyan21**

Spin-flip TDDFT with GAMESS using 6-31G(d,p) of gaseous pyrazine give T1-S0 gap around 0.12 eVs deiviating from Prof. Nguyen's TDDFT theoretical work with 6-31+G(d,p) published in JCP before, and about 0.18 eVs deviating from the experimental data indexed there. This triplet geometry is very easy to be found.

The triplet structure perhaps has an imaginary frequncy during latter UHF DFT analysis which perhaps could be eliminated.

I will use Dalton2018 to calculate the phosphorescence.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **23 Aug 2020, 12:55**

by **xiongyan21**

According to Prof. Nguyen: "A mean absolute error ~MAE! of 0.08 eV is obtained by

B3LYP/6-311G(d,p) for 37 experimental energies of organic chromophores. Interestingly, the MAE increases to

0.14 eV after including zero-point energy corrections."

The increase of convergence tightness and the use of 6-31+G(d,p) makes the later seminumerical Hessian analysis of pyrazine triplet geometry found by spin-flip TDDFT have no imaginary frequency, in the existence of spin-flip TDDFT, and the T1-S0 vertical gap within 0.21 eVs deviating from the experimental value; the adiabatic theoretical/ experimental discrypency of T1-S0 gap with ZPVE within 0.06 eVs. ZPVE is very samll for this molecule, and the ground state geometry found with C1 symmetry is nonplanar.

Monte Carlo UHF minimum searches have been successful for the triplet phenyl methyl ketone in vacuum and in ethanol, further spin-flip triplet geometry optimization and its Hessian anaysis will be done.

I will use UHF DFT to find the triplet structure of pyrazine beginning with the one got by spin-flip TDDFT.

Very Best Regards！

### Re: Can Dalton calculate oscillator strength

Posted: **25 Aug 2020, 09:32**

by **xiongyan21**

The phosphorescence emission calculated with Dalton2018 wit 6-31++G** for pyrazine in vacuum using geometry found by spin-flip TDDFT with GAMESS is within 0.1 eVs deviating from the experimental value and the phosphorescence liftime less than 0.06-0.08 seconds. Neverthless, in the terminal, there is such information

1976: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 24

1991: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 30

1998: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 30

2061: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 6

2068: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 6

2083: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 12

2090: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 12

2105: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 18

2112: *** INFO, negative eigenvalues in reduced matrix. Symmetry 1, triplet F Dimension of reduced matrix = 18

Does it matter? Perhaps only spin-flip methods can allow negative roots, and of course here the negative eigenvalues have been ignored.

6-31+G* gives the discrepancy within 0.08 eVs.

Very Best Regards！

### Re: Can Dalton calculate oscillator strength

Posted: **27 Aug 2020, 10:06**

by **xiongyan21**

With GAMESS the UHF optimization of pyrazine in vacuum with the increased convergence tightness, 6-31+G(d,p) and pbe0 gives an imaginary frequency about 172 cm^-1, and when the basis set is changed into 6-311++G(2d,2p), that becomes about 260 cm^-1. The adiabatic T1-S0 gap after ZPVE(~0.21 eVs) is within 0.26 eVs smaller than the experimentl value.

Spin-flip has been expanded by Prof. M. S. Gordon, once a PhD student of Nobel Prize Winner John A Pople, now a distinguished US university professor and Ames Lab senior researcher having winned many metals, prizes, and prestigies for achievements, who are leading his group continue to increase famous GAMESS's quantum chemical capabilities rapidly, and his student to spin-flip ORMAS-CI, which still lacks dynamic correlation but gives results almost identical to those obtained using other multireference methods (for molecules in his recent article) and has just been implemented in the newest GAMESS. FCI is the strength of Prof. Gordon's Group and Ames Lab. GAMESS now can also do spin-flip MRMP2 using another kind of FCI for energy.

The phosphorescence of this structure with Dalton2018 gives a descrepancy of about 0.23 eVs using 6-31+G* when compare with experiments, and the lifetime around 0.06 to 0.08 seconds, accompanying the above warnings in the terminal.

If an imaginary frequency exists, the vibrational analysis will exclude it, making ZPVE not that reliable.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **28 Aug 2020, 00:02**

by **xiongyan21**

All the Dalton2018 calculations on Ubuntu 18.04 may need to be carefully examined, because although they are finished and the results can be compared with experimental data, it may contain serious errors.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **28 Aug 2020, 15:24**

by **hjaaj**

Negative eigenvalues in reduced matrices for triplet properties (as spin-orbit operators) are a sign that your reference state is triplet unstable, i.e. that an unrestricted DFT would give a lower energy (unrestricted DFT is not implemented in Dalton).

I think you should be able to get reasonable results by invoking the .TDA approximation, which is often better than the full model for triplet properties because of the triplet instability issue.

### Re: Can Dalton calculate oscillator strength

Posted: **28 Aug 2020, 15:59**

by **xiongyan21**

Dear Prof. Jessen

Thanks a lot for your elaboration, which makes me know one triplet structure obtained from one operator may not be stable when using others.

For the first excited state of thiabendazole searched in water, both Hessian analysis methods exhibit no imaginary frequency.

As for pyrazine, Prof. Nguyen said all the triplet states of all the molecules studied in his article except for planar stilbene , have no imaginary frequency, but spin-flip TDDFT nonplanar structure found by GAMESS does have when the frequency analysis is done with UHF DFT.

Prof. Gordon said in his recent article CIS, RPA and TDDFT notoriously fail to describe the correct topology of conical intersections. Perhaps because of electron correlation, it is very difficult to find the true minimum triplet geometry using UHF DFT and using spin-flip TDDFT in terms of later UHF Hessian analysis. I will carefully examine the bond lengths and angles of the triplet geometry found by spin-flip TDDFT , compare them with those in several previously published articles, and try the new methods. I have already repeated the test of search the quintet state of TMM in the article using ROHF with spin-flip ORMAS.

He also comments on if a PES （1A2 state of a molecule) is very flat, optimization including dynamic correlation to more quantitatively ascertain its shape is required, which I think is also beneficial to THz and low frequency RAMAN analyses.

Very Best Regard!

### Re: Can Dalton calculate oscillator strength

Posted: **04 Sep 2020, 12:20**

by **xiongyan21**

For the spin-flip found triplet geometry with 6-31+G(d,p) the lengths of NC, CC, CH are 0.056, 0.045, 0.01 Bohrs deviating from the empirically corrected values of 3B3u reported in J. Phys. Chem, and the bond angles are about 2, 1, 1 degrees different, the discrepencies from those of MCSCF STO-3G with GAMESS also reported in that reference are 0.117, 0.076, 0.019 Bohrs; 0.63, 0.32, 1.41 degrees.

UHF DFT triplet structure gives all the three bond angles are around 119-120 degrees, with the angle deviances of 1.24, 0.61, 0.56 degrees, and the bond length discrepencies are 0.037,0.045, 0.005 Bohrs when compared with the empirically corrected data.

No experimental data of these has been found.

It seems basis set does not make a significant difference, and the imaginary frequency should be eliminated to get a proper match, although the reference didn't do Hessian analysis.

According to an article in Theo. Chem. Acc. in 2011 for spin-flip TDDFT in another package, for the 1A3 of COH2, the angle of HCH is 9 degrees smaller then the experimental value when using b3lyp, and the deviances of bond length of C-H, C-O are 0.038, 0.038, respectively.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **06 Sep 2020, 03:09**

by **xiongyan21**

As for pyridine, spin-flip TDDFT, although using C1 symmetry a plananr geometry is found, gives vertical T1-S0 gap discrepencies of about 0.12, 0.05 and 0.06 eVs using PBE0, B3LYP and B3LYPV1R(matching electron gas formula in some packages), respectively, when compared with the experimental data of 3B2 indexed in the article of Prof. Bousquet et al in France and Japan. Spin-flip TDDFT perhaps improves the gap even using 6-31G(d.p). But their results are very close to the experimental data indexed by Prof. Nguyen of Air Force Research Laboratory in US, where the geometry is nonplanar and the experimental result matches adiabatic T1-S0 gap calculated by TDDFT. The spin-flip TDDFT vertical T1-S0 gap is 0.3 eVs different from that experimental data. Of course, there is an imaginary frequency of the planar triplet geometry found, making ZPVE unreliable.

For the S1-S0 gap of pyrazine, B3lyp with C1 symmetry can give adiabatic one deviating 0.2 eVs, respectively, from the experimental values of 1B1 and 2B1 indexed in the article of Prof. Klerier, in Los Alamos National Laboratory, et al. in US, agreeing very well with those from HF calculations using symmetry there. ZPVE makes the gap unacceptable. The descrepencies of angle CNC and lengths of CN, CC are about 6.9 degrees, 0.038 and 0.048 angstroms when compared with experimental data. Since the reference emphasizes localized excitation, delocalized orbitals are no longer used, and one localization method proposed by distinguished Professor Emerituz Ludenberg in Iowa State University and Ames Lab, etc., gives very similar results but changes the frequencies in THz region. I am not sure whether the localization is suitable here.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **06 Sep 2020, 14:14**

by **hjaaj**

Yes. If you e.g. search for "dft spin-orbit dalton" on google, you get some informative links.

### Re: Can Dalton calculate oscillator strength

Posted: **07 Sep 2020, 03:16**

by **xiongyan21**

Dear Prof. Jensen

I have understood the article, Influence of triplet instabilities in TDDFT of Prof.Tozer, an RSC metal winner, et al. in Durham University in United Kindom.

By the way, Prof. Tozer is one of the authors of the diagnostic of excited states natures.

GAMESS UHF or ROHF TDDFT can only handle with the identical multiplicities of the ground and excited states。

Can Dalton2018 do this?

Spin-flip TDDFT pbe0 T1-S0 vertical gap of furan, almost the same as that in the article of Prof. Bousquet, et al., using TDDFT and 6-311++G(2d,2p), is about 0.40 eVs deviating from the experimental value in the article of Prof.Flicker, et al. , in California Institue of Technology in US. Adiabatic pbe0 T1(spin-flip TDDFT)- S0(UHF DFT) gap is very small, thus failing. Spin-flip TDDFT, TDDFT and DFT with pbe0 fail to deal with this. Spin-flip CIS gives a larger descrepency of 0.53 eVs when compared with experimental data, and there is no imaginary frequency of the optimized structure during the subsequent Hessian analysis in the existence of spin-flip CIS, and the spin-squared operators are 0.2844 and 2.200 where perhaps a little spin contaminations makes the gap underestimted.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **07 Sep 2020, 07:16**

by **hjaaj**

The answer is yes, using quadratic response. II can see on this forum that you used this yourself in 2016 ...

### Re: Can Dalton calculate oscillator strength

Posted: **08 Sep 2020, 09:53**

by **xiongyan21**

I am trying Spin-flip ORMAS-CI, and arbitrarily send the first partition of active space 29 electrons, 36 active electrons in total among which there are 19 alpha and 17 beta electrons based on spin-flip TDDFT.

One of the original ORMARS programmer is Dr. M. Schmidt in Ames Lab and Prof. Gordon's group, already retired, whom has been deemed as a long-time tutor and friend by many PhD graduates.

I also will look into the forum and try that using Dalton2018.

Very Best Regards!

### Re: Can Dalton calculate oscillator strength

Posted: **19 Sep 2020, 11:24**

by **xiongyan21**

Spin-flip OMARS geometry search of the triplet structure of furan is OK, where a planar structure is found. The active orbitals are 115 and the active electrons are 36 in 3 partitions with nstate being equal to 3, where the final energy is significant different from the ROHF one. FCI should be applied to the ground state to obtain the discrepancy, but the spin-flip ORMAS search uses analytical gradient!

Actually, Prof. Flicker said in his article that the lowest inealstic tranition occurs in the region 3.3 eV to 4.9 eV, with a maxmum intensity at 3.99 eV. The other singlet-triplet gap is at 5.22 eV where the electron-impact energy-loss spectra peak is higher.

The adibatic T1(sf-ormas)-S0(ormas, nstate=4, mult=1, the other same as spin-flip) also is vanishly small. I will examine what is wrong here.

The vertical gaps T1(sf-ormas-CI)-S0(ormas-CI, the first of nstate=4-20, mult=1) is around 0.93 eVs deviating from 3.99 eVs. The inputs should be examined to verify whether the results can match the experiments.

Prof. Bousquet, et al using SAC-CI give the gap of 4.02 eV. and they believe the experimental value is close to 4.1 eV based on the experiments of Prof. Flicker, et al.

Very Best Regards!