CCL: Atomic masses used in Orca for computing vibrational frequencies

 Sent to CCL by: =?utf-8?B?TWFyaXVzeiBSYWRvxYQ=?= [mariusz.radon]^[uj.edu.pl]
 Dear Grigoriy:
 > On 9 Dec 2024, at 08:27, Grigoriy Zhurko reg_zhurko[#]chemcraftprog.com
 <owner-chemistry|*|ccl.net> wrote:
 > Sent to CCL by: Grigoriy Zhurko [reg_zhurko===chemcraftprog.com]
 >> This is the standard atomic weight. An interval is given because the
 value depends on the origin of the material.
 >
 >> From the ORCA manual at https://www.faccts.de/docs/orca/6.0/manual/contents/typical/frequencies.html
 >> It is clear that ORCA uses atomic weights to calculate the hessian. But
 this can be changed afterwards using the orca_vib routine. Or during the
 calculation with the !Mass2016 keywords, that commands Orca to use the isotopic
 masses of the most abundant isotopes.
 >
 > I performed an Orca computation with !Mass2016 keyword. It indeed prints
 the masses of most abundant isotopes (12.0 for C, 1.00783 for H, etc). However
 the frequencies computed by Orca and my program do not match. I found that Orca
 in this case uses the mass 12.0004364861161 for C. For calculating this value, I
 computed a C4 molecule and divided a frequency produced by my program by the one
 printed by Orca.
 > I remember that Orca uses some QRRHO approach instead of RRHO; firstly I
 thought that this is applied only for computing the entropy, but maybe for the
 frequencies computations too?
 I doubt it can have anything to do with the QRRHO approach. As you wrote, this
 should only affect the computation of entropy (and maybe other thermodynamics
 functions).
 I wonder if the deviations you observe might be related to whether or not the
 translations and rotations are projected out before computing the frequencies.
 Different program may adopt slightly different algorithms at this stage,
 possibly explaining these discrepancies. In such a case, the discrepancies would
 be molecule-dependent and perhaps even frequency-dependent within a given
 molecule. Are they? Is the C-12 “effective mass” of 12.00044 amu
 that you obtained from Orca by “reverse-engineering" identical for
 (a) different normal modes of the same molecule? (b) different carbon-only
 molecules, like C2, C3, C4?
 Perhaps you are well aware of this, but let me add the following: It looks like
 that you are interested in tiny effects, which are often overlooked by most
 people, partly for the reason that these effects or related discrepancies would
 be irrelevant in most (if not all) practical applications due to other sources
 of larger errors in computed frequencies, such as: (a) inaccurate computation of
 the harmonic force constant, (b) anharmonicity effects being neglected or
 inaccurately modeled, (c) fundamentally inexact separation of rotations and
 vibrations. Also, when concerning isotopically averaged masses, their
 uncertainties are also relatively large, e.g., for C which you discussed before,
 the uncertainty of the standard atomic weight is 0.002 amu
 (12.0116−12.0096) due to uncertainties of the individual isotopes’
 abundances (atomic masses of the C-12 and C-13 isotopes are very accurately
 known). The uncertainty of the isotopically averaged mass is more important
 effect than including or neglecting the electrons’ masses (for C: 0.0003
 amu).
 > Also, I have one more question for all: would you find useful if Chemcraft
 computes the vibrational spectrum for each variant of isotopes distribution in
 the molecule, and averages them and shows at a singe graph (like in the
 computations of mass spectra)? I see that e.g. for Cl the isotope Cl37 produces
 different frequencies than Cl35, and its percentage is rather high - 25%.
 > Grigoriy.
 I do not use Chemcraft myself, but the strategy you described is definitely a
 good approach to tackle the problem of isotope mixtures (better than using
 isotopically averaged mass to compute a single set of frequencies). Such a
 feature would be useful for some people in specific applications, but I guess
 that in mainstream applications there exist other far more importance sources of
 errors in theory (inaccurate hessian and anharmonicity neglected, neglect of the
 environmental effects) and the experimental band are usually too broad to ever
 bother about such “subtleties”. The same holds true about the
 computation of statistical thermodynamics corrections based on the RRHO (or
 QRRHO) model, which is a crude approximation in some molecules.
 Best regards,
 Mariusz Radon
 --
 Mariusz Radon, Ph.D., D.Sc.
 Associate Professor
 Faculty of Chemistry, Jagiellonian University
 Address: Gronostajowa 2, 30-387 Krakow, Poland
 Room C1-06, Phone: 48-12-686-24-89
 E-mail: mradon|*|chemia.uj.edu.pl (mariusz.radon|*|uj.edu.pl)
 Web: https://tungsten.ch.uj.edu.pl/~mradon
 ORCID: https://orcid.org/0000-0002-1901-8521