search

gms-bbg / gamess-issues

GAMESS issue tracking
7 stars 1 forks source link

Does the warning in Test Raman-h2co matter #56

Open xiongyan21 opened 2 years ago

xiongyan21 commented 2 years ago

Test-h2co has a warning at the end MODE FREQ(CM**-1) SYMMETRY RED. MASS IR INTENS. RAMAN ACT. DEPOLARIZ. 1 86.452 B2 3.821881 0.672525 2.946837 0.750000 2 19.148 B1 2.879470 0.617671 0.877410 0.750000 3 0.071 B2 7.511561 0.000000 0.000001 0.750000 4 0.079 A1 7.502507 0.000000 0.000000 0.636416 5 0.151 B1 7.602040 0.000038 0.000055 0.750000 6 230.307 A2 1.007825 0.000000 3.876159 0.750000 7 1305.770 B2 1.345334 0.156786 0.047834 0.750000 8 1331.556 B1 1.318633 0.410328 0.972588 0.750000 9 1614.810 A1 1.131278 0.627055 10.953659 0.346079 10 1891.815 A1 6.215292 2.649156 14.930165 0.155271 11 3014.277 A1 1.050215 1.105944 142.451658 0.112713 12 3101.547 B1 1.119653 1.611123 63.308669 0.750000

 -------------------------------
 THERMOCHEMISTRY AT T=  298.15 K
 -------------------------------

USING IDEAL GAS, RIGID ROTOR, HARMONIC NORMAL MODE APPROXIMATIONS. P= 1.01325E+05 PASCAL. ALL FREQUENCIES ARE SCALED BY 1.00000 THE MOMENTS OF INERTIA ARE (IN AMU*BOHR2) 6.28646 46.18916 52.47562 THE ROTATIONAL SYMMETRY NUMBER IS 2.0 THE ROTATIONAL CONSTANTS ARE (IN GHZ) 286.82110 39.03704 34.36050 7 - 12 VIBRATIONAL MODES ARE USED IN THERMOCHEMISTRY. THE HARMONIC ZERO POINT ENERGY IS (SCALED BY 1.000) 0.027930 HARTREE/MOLECULE 6129.887380 CM-1/MOLECULE 17.526245 KCAL/MOL 73.329809 KJ/MOL

           Q               LN Q

ELEC. 1.00000E+00 0.000000 TRANS. 6.46197E+06 15.681445 ROT. 6.98656E+02 6.549158 VIB. 1.00399E+00 0.003978 TOT. 4.53269E+09 22.234581

          E         H         G         CV        CP        S
       KJ/MOL    KJ/MOL    KJ/MOL   J/MOL-K   J/MOL-K   J/MOL-K

ELEC. 0.000 0.000 0.000 0.000 0.000 0.000 TRANS. 3.718 6.197 -38.873 12.472 20.786 151.168 ROT. 3.718 3.718 -16.235 12.472 12.472 66.924 VIB. 73.395 73.395 73.320 1.451 1.451 0.251 TOTAL 80.832 83.311 18.212 26.394 34.708 218.343 VIB. THERMAL CORRECTION E(T)-E(0) = H(T)-H(0) = 64.998 J/MOL

          E         H         G         CV        CP        S
     KCAL/MOL  KCAL/MOL  KCAL/MOL CAL/MOL-K CAL/MOL-K CAL/MOL-K

ELEC. 0.000 0.000 0.000 0.000 0.000 0.000 TRANS. 0.889 1.481 -9.291 2.981 4.968 36.130 ROT. 0.889 0.889 -3.880 2.981 2.981 15.995 VIB. 17.542 17.542 17.524 0.347 0.347 0.060 TOTAL 19.319 19.912 4.353 6.308 8.295 52.185 VIB. THERMAL CORRECTION E(T)-E(0) = H(T)-H(0) = 15.535 CAL/MOL

 * THIS IS NOT A STATIONARY POINT ON THE MOLECULAR PES *
 *     THE VIBRATIONAL ANALYSIS IS NOT VALID !!!       *
 *********************************************************

This may be caused by Hessian calculated with another kind of method, e.g., DFT or another kind of basis set. Does it matter? If it does, how can we eliminate it?

I have used DFTB3 with conv=10d-12 and 3ob-3-1 parameters to calculate Raman for three pesticides(why no $Hess? ) , Nitenpyram, Thiamethoxam and Imidacloprid, and found no such a warning. I also have tried RHF Raman calculatation for one of them(optimized with B3lyp), and found such a warning appears either using RHF or DFT Hessian. TDHF calculation of Raman also gives the warning.

Very Best Regards!

gschoend commented 2 years ago

This also can be caused by numerical issues when using finite differencing.

On Thu, Jul 22, 2021 at 11:36 PM xiongyan21 @.***> wrote:

Test-h2co has a warning at the end MODE FREQ(CM**-1) SYMMETRY RED. MASS IR INTENS. RAMAN ACT. DEPOLARIZ. 1 86.452 B2 3.821881 0.672525 2.946837 0.750000 2 19.148 B1 2.879470 0.617671 0.877410 0.750000 3 0.071 B2 7.511561 0.000000 0.000001 0.750000 4 0.079 A1 7.502507 0.000000 0.000000 0.636416 5 0.151 B1 7.602040 0.000038 0.000055 0.750000 6 230.307 A2 1.007825 0.000000 3.876159 0.750000 7 1305.770 B2 1.345334 0.156786 0.047834 0.750000 8 1331.556 B1 1.318633 0.410328 0.972588 0.750000 9 1614.810 A1 1.131278 0.627055 10.953659 0.346079 10 1891.815 A1 6.215292 2.649156 14.930165 0.155271 11 3014.277 A1 1.050215 1.105944 142.451658 0.112713 12 3101.547 B1 1.119653 1.611123 63.308669 0.750000

THERMOCHEMISTRY AT T= 298.15 K

USING IDEAL GAS, RIGID ROTOR, HARMONIC NORMAL MODE APPROXIMATIONS. P= 1.01325E+05 PASCAL. ALL FREQUENCIES ARE SCALED BY 1.00000 THE MOMENTS OF INERTIA ARE (IN AMU*BOHR

2) 6.28646 46.18916 52.47562 THE ROTATIONAL SYMMETRY NUMBER IS 2.0 THE ROTATIONAL CONSTANTS ARE (IN GHZ) 286.82110 39.03704 34.36050 7 - 12 VIBRATIONAL MODES ARE USED IN THERMOCHEMISTRY. THE HARMONIC ZERO POINT ENERGY IS (SCALED BY 1.000) 0.027930 HARTREE/MOLECULE 6129.887380 CM -1/MOLECULE 17.526245 KCAL/MOL 73.329809 KJ/MOL

       Q               LN Q

ELEC. 1.00000E+00 0.000000 TRANS. 6.46197E+06 15.681445 ROT. 6.98656E+02 6.549158 VIB. 1.00399E+00 0.003978 TOT. 4.53269E+09 22.234581

      E         H         G         CV        CP        S
   KJ/MOL    KJ/MOL    KJ/MOL   J/MOL-K   J/MOL-K   J/MOL-K

ELEC. 0.000 0.000 0.000 0.000 0.000 0.000 TRANS. 3.718 6.197 -38.873 12.472 20.786 151.168 ROT. 3.718 3.718 -16.235 12.472 12.472 66.924 VIB. 73.395 73.395 73.320 1.451 1.451 0.251 TOTAL 80.832 83.311 18.212 26.394 34.708 218.343 VIB. THERMAL CORRECTION E(T)-E(0) = H(T)-H(0) = 64.998 J/MOL

      E         H         G         CV        CP        S
 KCAL/MOL  KCAL/MOL  KCAL/MOL CAL/MOL-K CAL/MOL-K CAL/MOL-K

ELEC. 0.000 0.000 0.000 0.000 0.000 0.000 TRANS. 0.889 1.481 -9.291 2.981 4.968 36.130 ROT. 0.889 0.889 -3.880 2.981 2.981 15.995 VIB. 17.542 17.542 17.524 0.347 0.347 0.060 TOTAL 19.319 19.912 4.353 6.308 8.295 52.185 VIB. THERMAL CORRECTION E(T)-E(0) = H(T)-H(0) = 15.535 CAL/MOL

  • THIS IS NOT A STATIONARY POINT ON THE MOLECULAR PES *
  • THE VIBRATIONAL ANALYSIS IS NOT VALID !!! *

This may caused by Hessian calculated with another kind of method, e.g., DFT or another kind of basis set. Does it matter? If it does, how can we eliminate it?

Very Best Regards!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/gms-bbg/gamess-issues/issues/56, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEJSYDMXPKX6FZDS2WDZCO3TZDPVNANCNFSM5A3HA46Q .

xiongyan21 commented 2 years ago

I have used 6-31G to try TDHF Raman calculation of imidacloprid, and will try larger basis sets. For comparison, I have chosen methane ( mp2 and 6-311++G(2d,2p) optimized) to calculate HF Raman(mp2 Hessian) and TDHF Raman, where there are above warnings, and got the difference within 45 cm-1 of the strongest peak. In HF Raman, there are three peaks extremely close after it, but there is only one more peak following in TDHF Raman. As for DFTB (conv=10d-12) geometry, DFTB3 Raman activity calculation(no warning) gives the strongest peak around 270 cm-1 lower, and two similar lines after it, whereas the lines before it having discrepencies about 200 cm-1. NWCHEM7.0.2 ao basis 6-311++G** and B3lyp Raman calculation with the same basis set and same functional optimized geometry gives a difference around 150 cm-1, and latter three peaks identical in position but slightly different in strength.

It seems NWCHEM gives questionable strengths if I haven't misunderstood the strength expressions.

of frequencies : 9

freq [ 1/cm] S [Ang**4/amu]

1 ... 4.6943E+06 2 ... 3.5818E+01 3 ... 2.3670E+01 4 ... 6.6062E+00 5 ... 2.6691E+00 6 ... 2.1850E+00 7 ... 8.0499E+00 8 ... 1.1637E+00 9 3127.45 1.6601E+00 Actually, there is something unusual in the NWChem output ... (start,last,rmmodes,nfreq,plot,width)=( 1, 15, 9, 1,normal , 20.00000000) rminfo( 1)=( 1341.19475598, , 2097.35357166, ) rminfo( 2)=( 1341.19525073, 35.81000501, 0.00753700, 35.81754200) rminfo( 3)=( 1341.19574540, 23.66503856, 0.00448652, 23.66952508) ... Perhaps, there are problems in the NWCHEM calculation.

Dalton2018 6-311++G numerical Hessian( gives the second highest peak at around 3200 cm-1(three lines) for 6-311++G mp2 geometry optimized with itself.

Mode Freq. Alpha2 Beta(a)2 Pol.Int. DepolInt. 180/0Int. DepRatio

1 ... 151.3303 ... 2 ...151.3311 ... 3 ...151.3390 ... 4 ... 391.7587 ... 5 ... 40.9031 ... 6 ... 40.8934 ... 7 ... 1.8654 ... 8 ... 1.8654 ... 9 ...1.8644 ...

Isotope effects are not considered in all of the above.

Very Best Regards!

xiongyan21 commented 2 years ago

According to the article High-Resolution Raman Spectroscopy of Gases Part XVI. The v3 Raman Band of Methane in JOURNAL OF MOLECULAR SPECTROSCOPY, 10, 448-483, (1963)

"" Methane (CH,) has four fundamental vibrations: v1 = 2916.7 cm-1 of species A, ; VP = 1533.6 cm-l of species E (doubly degenerate); v3 = 3018.9 cm -1; ard v4 = 1350.9 cm-’ both of species FZ (triply degenerate). All four vibrations are allowed in the Raman effect and the v1 , Q , and v3 bands have been studied at high resolution. " According to the calculations here, GAMESS DFTB3 can give results closest to these except for 1350.9 cm-1 and 1533.6 cm-1 which are cloest to the NWCHEM results, whereas Dalton and GAMESS TDHF gives lines having larger discrepencies among the results. Overally, DFTB's deviances are all within around 65cm-1. By the way, all the calculations except TDHF indicate the lines with the highest wavenumbers are also triply or nearly triply degenerate. Prof. Quinet gives in his article Vibrational Spectroscopies: Description of General Analytical TDHF Schemes for Their Simulation

␯ 1 (T 2 ) 1431 cm ⫺1 ␯ 2 (E) 1658 cm ⫺1 ␯ 3 (A 1 ) 3145 cm ⫺1 ␯ 4 (T 2 ) 3270 cm ⫺1 and values of a parameter A for degeneracy, and his simulated Stokes bands (only three)have lower wavenumbers than the above.

Can GAMESS give a criterion of degeneracy in TDHF package?

Very Best Regards!

xiongyan21 commented 2 years ago

Actually, DFT , HF and mp2 each have a well-known scaled factor for vibrational frequencies and zero-point energy, from Prof. Pople, respectively, but it seems if it is applied, it will underestimated the frequencies(different basis sets here).

I have found in an reference the scaled factor for SCC-DFTB is 0.9933.

Here, DFTB underestimates the frequencies except the one over 3000 cm-1, but the absolute relative errors are all smaller than about 4%.

It is a common practice to adopt adaptive scaled factors in order to make the calculated match the experimental.

Very Best Regards

xiongyan21 commented 2 years ago

By the way, I do not include LIBBCHEM (this needs another software to equip my computer), and CCSD3 and CCT3( several of these will fail on my computer), and NEO( which I am not interested in).

Very Best Regards!

xiongyan21 commented 2 years ago

There is another thing baffling, i.e., the Raman experimental spectra of pesticides are gereally obtained with SERS. For imidacloprid, the three detection lines calculated by DFTB (perhaps not the strongest nearby) are within 2.4% deviances when compared with reproted experimental SERS peaks, and for thiamethoxam(strongest nearby, four detection lines), those become within 3.6%, whereas for nitenpyram, those become within 4.6% if the strongest nearby are chosen of the three detection lines. I will try the complexes with DFTB, like Chemical enhancement in the SERS spectra of indigo: DFT calculation of the Raman spectra of indigo-Ag 14 complexes，The Raman and SERS spectra of indigo and indigo-Ag 2 complex: DFT calculation and comparison with experiment， etc.

It seems the deviances are large without scaled factors for pyruvic acid. I am searching experimental data.

There are no warnings now in the DFTB calcualtions, and previously for DFTB Raman calculation of thiameoxam not using DFTB geometry, not a stationary point warning appears.

Very Best Regards!

xiongyan21 commented 2 years ago

Unfortunately, there is no parametrization of Ag in 3ob-3-1and there are onl;y C-C|C-N|C-O in 3ob:freq, but there are no obvious changes of the chief static Raman peaks from the calculations of indigo and indigo-Ag2 complex(with small peaks might being eliminated), according to The Raman and SERS spectra of indigo and indigo-Ag 2 complex: DFT calculation and comparison with experiment.

Very Best Regards!

xiongyan21 commented 2 years ago

According to An Ab Initio Study of Pyruvic, experimental IR should be fitted to the mp2 6-311++G** calculated frequencies.

With DFTB and C1 symmetry, the first can be as low as 47 cm-1, over 70 cm-1 when compared with the experimental data indexed, and for some peaks, the frequencies calculated can be accepted. For imidacloprid THz deterction, DFTB results are not very good, but DFT frequency analysis of the 6-31+G** B3lyp geometry gives six peaks in the 0-2 THz domain, with the first two at 0.12 and 0.32 THz, and the relative strength pattern of the last three perhaps can be compared with the experimental one published in another paper. The one at 1.2 Thz is strong. For the fourth and sixth, the DFTB results(without the fifth one) are closer to the experimental results. There are no stationary point warning but floating-point exception warnings in the DFT calculations. RHF frequency analysis has no peaks ahead.

I cann't use MacMolplt to make a planar pyruvic acid drawn by Avogadra.

Very Best Regards!

xiongyan21 commented 2 years ago

It is OK if B3lyp and 6-31G** geometry and Raman calculations of methane run successively with one input like QA test h2o-raman_3.nw in terms of projected IR by NWCHEM7.0.2

 normal          Raman Scattering Plot

Excitation wavelength 514.50 nm Convolution lineshape is lorentzian
FWHM 20.00 1/cm

of frequencies : 9

freq [ 1/cm] S [Ang**4/amu]

1 1356.19 2.5245E+00 2 1356.19 2.5245E+00 3 1356.19 2.5246E+00 4 1578.56 3.2171E+01 5 1578.56 3.2171E+01 6 3046.80 1.3402E+02 7 3162.63 5.8496E+01 8 3162.64 5.8497E+01 9 3162.65 5.8496E+01

Very Best Regards!