ESR/EPR¶

ESR/EPR g-tensor and A-tensor¶

A-tensor, no spin-orbit coupling

$ADFBIN/adf << eor
  ESR
  END
  CHARGE charge spinpolarization
  unrestricted
  {Relativistic scalar ZORA}
  {NUCLEARMODEL gaussian}
eor

If spin-orbit coupling is neglected, the spin in the effective spin Hamiltonian, which is commonly used for the interpretation of ESR experiments, is the real electronic spin of the paramagnetic molecule. In the spin-unrestricted DFT calculations one then uses eigenfunctions of S_z . The A-tensor can then simply be calculated as expectation value of the corresponding operator, see [96].

The A-tensor will be calculated for all nuclei. Terms due to the spin-polarization density at the nucleus are included in the evaluation of the A-tensor. For an accurate evaluation of the spin-polarization density at the nucleus it is important to use an all-electron basis set for the nuclei that one is interested in, avoiding the frozen core approximation. For heavy elements the incorporation of a Gaussian finite nucleus model can be important. However, one should have really large basis sets with tight basis functions to observe this effect in calculations. One possibility is to use the $ADFHOME/atomicdata/ZORA/QZ4P basis set, although even this large basis set is not large enough sometimes. The basis sets in the directory $ADFHOME/atomicdata/ZORA/jcpl (not available for all elements) are suitable for finite nucleus calculations.

In case one uses a finite nuclear model for the charge distribution, starting from ADF2013 ADF also uses a finite distribution of the nuclear magnetic dipole moment for the calculation of the A-tensor.

A-tensor, perturbative inclusion spin-orbit coupling

$ADFBIN/adf << eor
CHARGE charge spinpolarization
unrestricted
Relativistic scalar ZORA
Symmetry NOSYM
...
eor

$ADFBIN/cpl << eor

hyperfine
 atoms 1 2 :: calculates A-tensor for atom 1 and 2, input order
 SCF Converge=1e-7 {Iterations=25}
end
...
eor

The calculation of A-tensors is implemented in the CPL program as a second derivative property (spin-orbit coupling and nuclear magnetic field as perturbation) within the two-component relativistic zeroth-order regular approximation (ZORA), see Ref. [340]. This implementation allows for hybrid (only PBE0) DFT calculations, but not metaGGA’s and not metahybrids.

Note that the CPKS convergence in CPL has to be set tightly (1e-7 or 1e-8) to get converged PSOSO terms for the A-tensor. For hyperfine calculations the default value is 1e-7.

g-tensor, perturbative inclusion spin-orbit coupling

$ADFBIN/adf << eor
CHARGE charge spinpolarization
unrestricted
Relativistic scalar ZORA
Symmetry NOSYM
...
eor

$ADFBIN/nmr << eor
nmr
 gfactors
 u1k best
 calc all
 out iso tens
end
end input
eor

The calculation of g-tensors is implemented in the NMR program as a second derivative property (spin-orbit coupling and external magnetic field as perturbation) within the two-component relativistic zeroth-order regular approximation (ZORA), see Ref. [339]. This implementation allows for hybrid (B3LYP, PBE0, etc) DFT calculations, but not metaGGA’s and not metahybrids. This implementation requires the use of all electron basis sets.

For an older implementation of this method, see the EPR/NMR module documentation in the old ADF 2010 Properties document, and Refs. [120,121, 416,417].

g-tensor and A-tensor, self consistent spin-orbit coupling

$ADFBIN/adf << eor
ESR
END
CHARGE charge
unrestricted
Relativistic spinorbit ZORA
Collinear
Symmetry NOSYM
eor

In a spin-orbit coupled spin unrestricted relativistic ZORA calculation and the ESR block key, the g-tensor and the nuclear magnetic dipole hyperfine interaction (A-tensor) will be calculated, see also Refs. [95,96]. In such a calculation degenerate perturbation theory is used with the external magnetic field or nuclear magnetic field as perturbation. The calculation must use the collinear approximation, and symmetry must be NOSYM. This implementation does allow for metaGGA, and (meta-)hybrid DFT calculations, but then GIAO’s are not used. There may be more than one unpaired electron. Terms due to the spin-polarization density at the nucleus are included in the evaluation of the A-tensor. However, one can not set the number of unpaired electrons, the ‘spinpolarization’ argument of the key CHARGE will be ignored.

Note: in a spin-orbit coupled spin restricted relativistic ZORA calculation and the ESR block key, ADF will also calculate and print the nuclear magnetic dipole hyperfine interaction, but the terms due to the spin-polarization density at the nucleus are absent. Furthermore, if there is more than one unpaired electron, the computed results will simply be incorrect, without any warning from the program. On the other hand, in case of one unpaired electron, and very large effects of spin-orbit coupling, the spin-restricted calculation may be of interest, since it uses Kramer’s symmetry exact.

ESR/EPR Q-tensor¶

For the calculation of the ESR Q-tensor see the key QTENS.

ESR/EPR Zero-field splitting (D-tensor)¶

With the keyword ZFS the zero-field splitting (ZFS) of the ground state can be calculated.

ZFS
RELATIVISTIC SCALAR ZORA

Zero-field splitting is the breaking of degeneracies of the ground state that is not described by a standard non-relativistic Hamiltonian. ZFS as calculated by ADF is that exhibited by molecules whose ground state has spin S>1/2 and no spatial degeneracy. This type of ZFS has two contributions, second-order spin-orbit coupling and spin-spin coupling. In the present implementation only the spin-orbit coupling term is included. The calculation of ZFS with DFT is described in [312-315]

ZFS can be calculated in combination with LDA and GGAs but not hybrid or meta-GGA functionals. In order to calculate ZFS the RELATIVISTIC SCALAR ZORA option must be included.

Just the simple keyword ZFS is needed in order to calculate zero-field splitting. Several optional additional keywords can also be included. The complete list is:

ZFS {PEDERSON|NEESE} {ANALYSIS|FULLANALYSIS}

PEDERSON|NEESE

PEDERSON: The available approaches for calculating ZFS with DFT each differ subtly from the others. We believe that the method proposed by van Wüllen and coworkers [314,315] is the most theoretically complete but it may be that for certain systems the other approaches are more accurate. The van Wüllen formulation is the default but if the PEDERSON keyword in included then the equation proposed by Pederson and Khanna [312] is used. NEESE: If the NEESE keyword is included then the equation for ZFS proposed by Neese [313] is used.

ANALYSIS|FULLANALYSIS

ANALYSIS: Neese has presented some interesting analyses of ZFS [313]. If the ANALYSIS keyword is invoked then the contributions to the ZFS is divided into terms from alpha-beta, alpha-alpha, beta-beta and beta-alpha one-electron excitations. FULLANALYSIS: The output requested by the ANALYSIS keyword is further extended to analyze each of the alpha-beta, alpha-alpha, beta-beta and beta-alpha contributions in terms of the individual one-electron excitations.