CCL:G: Anharmonic Infrared Spectroscopy



Dear Jens SPANGET-LARSEN,

there can be few problems, maybe all of them together.

1. Generation of anharmonic force-field and dipole moment derivatives.

To get reliable results FF need to be generated for at least "tight" optimized structure,
and using at least int=ultrafine grid for DFT. If not, numerical errors may lead to completely un-reliable results,
like negative frequencies, huge corrections to energies etc.

2. Anharmonic resonances, in particular for intensities (1-1 resonances)

That is most likely the case of modes 2 and 3 which have very similar harmonic wavenumbers, but just off the default threshold (2cm-1) .
Solution is to change resonance criteria, for instance by using keyword
Resonances=(DFreq11=XX) xx=5 or 10 can be a good choice there also other options related to the magnitude of force constants (i.e Res11F3=0.0,Res11F4=0.0)

3. Coupling with low-freqency large amplitude motions (LAM) (i.e methyl rotation)

This can be overcome by excluding LAM from VPT2 treatment, that is well tested and reliable procedure
it can be done by using the keywords:

DataMod=SkipPT2=Modes RedDim=Inactive=1

mode number

You can note that several procedures, in particular for anharmonic resonances, including 1-1, have been modified to be more robust
in their default version in G16. But, most importantly there is a large number of keywords which allow to set up GVPT2 computations accordingly to the system,
i.e. not using VPT2 for floppy modes etc.

However, first "must" is reliable generation of force constants, so the warnings in this part need to be also checked.

Hope it helps,

Best wishes,

Malgorzata


2017-06-05 4:53 GMT+08:00 Robert Molt r.molt.chemical.physics-x-gmail.com <owner-chemistry|,|ccl.net>:

Is it plausible that there are degeneracies in the vibrational/rotational spectrum, leading to the divergences you observe? This would be the most obvious explanation for why the numbers blow up for any finite-order perturbation theory. Perhaps you have coupled methyl groups?


With smaller basis sets, oftentimes the divergences in finite order perturbation theory are "masked" (I've seen this with cc-pvdz vs. aug-cc-pVDZ), simply because you have not yet encountered the denominator problem. It's a "better" answer for wrong reasons.


On 6/4/17 4:38 AM, Jens Spanget-Larsen spanget**ruc.dk wrote:

Dear CCL!
When I perform Gaussian calculations with the option freq=anharm, I frequently get strange results (Gaussian 09, Revision D.01). Below is a portion of the results of a b3lyp/6-311++G(d,p) calculation on a planar, aromatic, C2v symmetrical hydrocarbon. The anharmonic results are obviously ridiculous. With smaller basis sets, the results tend to be more realistic. Any explanations?

Yours, Jens >--<

 

     ==================================================
              Anharmonic Infrared Spectroscopy
     ==================================================

 

 Fundamental Bands
 -----------------
    Mode(Quanta)       E(harm)   E(anharm)       I(harm)        I(anharm)
     1(1)              3476.017   3200.902     43.37436356    538.05352307
     2(1)              3202.964   3109.532      0.01728522    255.35137878
     3(1)              3200.221   3033.726      8.04709323    191.91610101
     4(1)              3175.408   3220.259      4.37202654   2730.70615470
     5(1)              2205.565   2153.406      1.97473796     11.71516009
     6(1)              1604.309   1584.143      6.91156654    634.57714516
     7(1)              1432.153   1374.467      3.54182113  12790.38891389
     8(1)              1263.675   1263.359      0.79105395    139.40836060
     9(1)              1112.643   1056.598      5.36740074   1394.13686552
    10(1)              1010.305   1031.530      0.00269831     47.10463873
    11(1)               717.365    793.822      1.28159522 103894.31384900
    12(1)               697.978   1794.196     67.17706473   1444.46852987
    13(1)               505.134   1553.018      6.40712425  16868.05994564
    14(1)               465.540    608.353      2.04849136  24943.02416835
    15(1)               115.463    594.923      0.77231918   6377.20573512
    16(1)               930.354  -1253.748      0.00000000      0.00000039
    17(1)               653.549  -2649.903      0.00000000      0.00000067
    18(1)               621.501 -20992.872      0.00000000      0.01789241
    19(1)               355.466  -1676.263      0.00000000      0.00001474
    20(1)               156.168  -1653.479      0.00000000      0.00002682
    21(1)               990.196   7085.310      0.11097137 176674.25100856
    22(1)               925.704  -2638.620     16.28434626 228593.72377779
    23(1)               811.788   1939.226     41.57436680 251187.21629870
    24(1)               701.424   1649.690     14.23131322 ***************
    25(1)               650.217   -656.428    106.61652268 433910.83299326
    26(1)               493.743  16521.965      0.05960240 ***************
    27(1)               378.821  -1503.535      5.22048529 779479.83775810
    28(1)               121.625  -1543.939      2.61491704  76301.87107925
    .......

    .......

 

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

  JENS SPANGET-LARSEN        Office:       +45 4674 2710

  Dept. of Science           Fax:          +45 4674 3011

  Roskilde University        Mobile:       +45 2320 6246

  P.O.Box 260                E-Mail:      spanget||ruc.dk

  DK-4000 Roskilde, Denmark  Web: thiele.ruc.dk/~spanget

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





--
Prof. Malgorzata Biczysko

International Centre for Quantum and Molecular Structures,
1/F Building E,
Shanghai University,
99 Shangda Road, Shanghai, 200444 China
e-mail: biczysko|,|shu.edu.cn

http://icqms.shu.edu.cn

上海大学量子与分子结构国际中心

上海市宝山区上大路99号E楼1楼