Calculation of Vibrational Frequencies

In this chapter, all necessary information about the calculation of vibrational spectra and thermostatistical contributions are given.

Performing simple Vibrational Frequency calculations

Vibrational frequency calculations are available only through two-sided numerical differentiation of analytical gradients.

Consider a simple example like the following hydrogen abstraction reaction:

7

C          -0.12888312425142   -0.00640246259879   -0.00997057133406
H           1.44011699709596    0.12229812355524   -0.02854203428735
H          -0.41612454870604    1.02694842152161   -0.04938812535015
H          -0.26306601703832   -0.58286887757121   -0.90445094952445
H          -0.26440375738028   -0.51708010658031    0.92386857306799
O           2.45008586500521    0.26032015001761    0.01133571198248
H           2.61210033905108    0.98358191645276    0.62026402303033

By invoking the --hess command line argument, xtb executes a calculation of the Hessian matrix. The --ohess keyword may be used instead if a prior optimization of the structure is desired.

xtb min.xyz --hess --uhf 1

At the end of the frequency job you get an output like this:

        -------------------------------------------------
        |               Frequency Printout                |
        -------------------------------------------------
projected vibrational frequencies (cm-1)
eigval :       -0.00    -0.00    -0.00     0.00     0.00     0.00
eigval :       40.69   211.99   360.40   405.89   601.08   759.17
eigval :      829.07  1371.91  1375.70  1477.42  2297.37  3115.69
eigval :     3190.88  3197.05  3648.64
reduced masses (amu)
1:  4.63   2: 12.19   3: 12.31   4:  9.70   5:  9.30   6: 12.74   7:  1.74   8: 12.17
9:  1.45  10:  1.77  11:  1.92  12:  1.45  13:  2.39  14:  2.02  15:  2.03  16:  2.07
17:  2.17  18:  1.07  19:  2.09  20:  2.09  21:  1.86
IR intensities (amu)
1:  0.17   2:  0.46   3:  0.44   4:  0.41   5:  0.08   6:  0.35   7:  0.48   8:  0.26
9:  0.26  10:  0.27  11:  0.25  12:  0.41  13:  0.30  14:  0.04  15:  0.07  16:  0.52
17:  0.51  18:  0.06  19:  0.14  20:  0.12  21:  0.18
Raman intensities (amu)
1:  0.00   2:  0.00   3:  0.00   4:  0.00   5:  0.00   6:  0.00   7:  0.00   8:  0.00
9:  0.00  10:  0.00  11:  0.00  12:  0.00  13:  0.00  14:  0.00  15:  0.00  16:  0.00
17:  0.00  18:  0.00  19:  0.00  20:  0.00  21:  0.00
output can be read by thermo (or use thermo option).
writing <g98.out> molden fake output.
recommended (thermochemical) frequency scaling factor: 1.0

This output consists of the calculated vibrational frequencies and the vibrational modes. In the example above there are six frequencies which are identically zero. These frequencies correspond to the rotations and translations of the molecule. They have been projected out of the Hessian before the calculation of the frequencies and thus, the zero values do not tell you anything about the quality of the Hessian that has been diagonalized.

xtb writes an g98.out file in GAUSSIAN-format, which can be opened with the popular MOLDEN program to visualize the vibrational modes. Further, a hessian file is written, containing the projected Hessian matrix in turbomole format.

Calculation of thermochemical properties

Each frequency job provides the thermochemical properties at 298.15 K. (for other temperatures, see below). No further user-input is required to obtain all important thermostatistical contributions. The contributions are calculated following a coupled rigid-rotor-harmonic-oscillator approach. If a molecular symmetry is detected, the resulting rotational number is automatically accounted for. The symmetry detection can be adjusted in the $symmetry block of the xcontrol file if necessary.

        -------------------------------------------------
        |             Thermodynamic Functions             |
        -------------------------------------------------

        ...................................................
        :                      SETUP                      :
        :.................................................:
        :  # frequencies                          15      :
        :  # imaginary freq.                       0      :
        :  linear?                             false      :
        :  only rotor calc.                    false      :
        :  symmetry                               C1      :
        :  rotational number                       1      :
        :  scaling factor                  1.0000000      :
        :  rotor cutoff                   50.0000000 cm⁻¹ :
        :  imag. cutoff                  -20.0000000 cm⁻¹ :
        :.................................................:

    mode    ω/cm⁻¹     T·S(HO)/kcal·mol⁻¹    T·S(FR)/kcal·mol⁻¹   T·S(vib)
------------------------------------------------------------------------
    1     40.69    -1.55795 ( 30.48%)    -1.11458 ( 69.52%)    -1.24972
    2    211.99    -0.60419 ( 99.69%)    -0.62804 (  0.31%)    -0.60426
------------------------------------------------------------------------

temp. (K)  partition function   enthalpy   heat capacity  entropy
                                cal/mol     cal/K/mol   cal/K/mol   J/K/mol
298.15  VIB   13.7                 1501.827      9.485      9.210
        ROT  0.909E+04              888.752      2.981     21.094
        INT  0.125E+06             2390.579     12.466     30.304
        TR   0.184E+27             1481.254      4.968     36.401
        TOT                        3871.8331    17.4344    66.7050   279.0936

    T/K    H(0)-H(T)+PV         H(T)/Eh          T*S/Eh         G(T)/Eh
------------------------------------------------------------------------
    298.15    0.617016E-02    0.583013E-01    0.316937E-01    0.266076E-01
------------------------------------------------------------------------

        :::::::::::::::::::::::::::::::::::::::::::::::::::::
        ::                  THERMODYNAMIC                  ::
        :::::::::::::::::::::::::::::::::::::::::::::::::::::
        :: total free energy          -8.613409150740 Eh   ::
        ::.................................................::
        :: total energy               -8.640016786693 Eh   ::
        :: zero point energy           0.052131167146 Eh   ::
        :: G(RRHO) contrib.           -0.025523531193 Eh   ::
        :::::::::::::::::::::::::::::::::::::::::::::::::::::

Multiple temperatures can be calculated using the build in thermodynamic functions calculator by using a input file similar to this

$thermo
    temp=150.0,200.0,250.0,273.15,298.15

The final summary looks like

    T/K    H(0)-H(T)+PV         H(T)/Eh          T*S/Eh         G(T)/Eh
------------------------------------------------------------------------
 150.00    0.250495E-02    0.546739E-01    0.135034E-01    0.411705E-01
 200.00    0.361203E-02    0.557809E-01    0.192424E-01    0.365386E-01
 250.00    0.484240E-02    0.570113E-01    0.253913E-01    0.316200E-01
 273.15    0.545010E-02    0.576190E-01    0.283634E-01    0.292557E-01
 298.15    0.617016E-02    0.583013E-01    0.316937E-01    0.266076E-01 (used)
------------------------------------------------------------------------

xtb will always use the last entry from the temperature list for all further calculations and printouts.

Dealing with imaginary modes and non-minimum structures

If a frequency calculation is invoked using the --hess command line argument, xTB automatically checks the gradient norm for a non-zero value. For unoptimized structures with significant remaining grad. norm, a warning is printed. If you want xTB to exit with an error code instead of this warning, use the --strict command line argument.

########################################################################
# WARNING! Some non-fatal runtime exceptions were caught, please check:
#  - Hessian on incompletely optimized geometry!
########################################################################

A xtbhess.coord file is created in this case, containing the input structure distorted along the imaginary mode. In case of unwanted imaginary modes, this structure can be used as a starting point to perform further optimizations to get rid of the imaginary frequency and locate the true minimum.

Advanced options

Of course, the calculated frequencies depend on the masses used for each atom. Several options exist to modify/scale the default atomic masses in the $hess block of the xcontrol file.

$hess
    sccacc=real
        SCC accuracy level in Hessian runs

    step=real
        Cartesian displacement increment for numerical Hessian

    isotope: int,real
        set mass of atom number int to real (synonym to modify mass)

    modify mass: int,real
        set mass of atom number int to real (synonym to isotope)

    scale mass: int,real
        scale mass of atom number int by real

    element mass: int,real
        set mass of elements int to real

Changes regarding sccacc or step should be made with caution, as large displacements or loose SCC accuracy can lead to unreliable frequencies due to excessive numerical noise in the calculations.

The thermostatistical calculations can be influenced by the $thermo block of the xcontrol file.

$thermo
    temp=real
        temperature for thermostatistical calculation (default: 298.15 K)

    sthr=real
        rotor cut-off (cm-1) in thermo (default: 50.0)

Single Point Hessian (SPH) calculations

A prerequisite for accurate thermostatistics so far was to optimize the molecular input structures in order to avoid imaginary frequencies as discussed above. This inevitably leads to changes in the geometry if different theoretical levels are applied for geometry optimization and frequency calculations. Therefore, we propose a new method termed single-point Hessian (SPH) for the computation of HVF and thermodynamic contributions to the free energy within the modified RRHO approximation for general nonequilibrium molecular geometries. The main publication for SPH can be found at: JCTC. A SPH calculation is invoked using the --bhess command line argument. xTB automatically applies a biasing potential given as Gaussian functions expressed with the RMSD in Cartesian space in order to retain the initial geometry. In the xTB printout this is indicated by e.g.:

metadynamics with 1 initial structures loaded
           -------------------------------------------------
          |           Optimal kpush determination           |
           -------------------------------------------------
target rmsd / Å         0.100000
unbiased initial rmsd   0.415461

iter. min.        max.        rmsd       kpush
 1    0.000000    1.000000    0.023187   -0.500000
 2    0.000000    0.500000    0.052359   -0.250000
 3    0.000000    0.250000    0.090094   -0.125000
 4    0.000000    0.125000    0.135494   -0.062500
 5    0.062500    0.125000    0.108071   -0.093750
 6    0.093750    0.125000    0.098169   -0.109375
 7    0.093750    0.109375    0.102794   -0.101562
 8    0.101562    0.109375    0.100426   -0.105469
final kpush: -0.105469
           -------------------------------------------------
          |            Biased Numerical Hessian             |
           -------------------------------------------------
kpush                :  -0.10547
alpha                :   1.00000
step length          :   0.00500
SCC accuracy         :   0.30000
Hessian scale factor :   1.00000
frozen atoms in %    :   0.00000    0
RMS gradient         :   0.00028
estimated CPU  time      0.06 min
estimated wall time      0.02 min

writing file <hessian>.

The systematic shift of the HVF caused by the modification of the PES due to the biasing potential is subsequently removed approximately by individual frequency scaling, giving access to accurate thermostatistical contributions for general nonequilibrium geometries with low-level methods. The desired target RMSD between the input and constrained optimized structure can be influenced by the $metadyn block of the xcontrol file.

$metadyn
    rmsd=real
        target RMSD between input and optimized structure (default: 0.10 Å)