Direct to:


Faculty of Chemistry


ECP Pseudopotentials


PES Database

Journal of Chemical Physics

Molecular Physics

Journal of Chemical Theory and Computation

Deutsche Forschungsgem. DFG

University Library


Excellence Initiative


Theoretical Spectroscopy:

Our research focusses mainly on the development of accurate quantum chemical methods as needed for a reliable prediction of spectroscopic properties of small molecules and molecular clusters (<25 atoms). It is the efficient and automated simulation of infrared, Raman, resonance Raman, photoelectron or photoionization spectra within a fully anharmonic framework and the account of high-order correlation effects, which dominates our work. All methods developed in our group are available in the Molpro program package.

Method development:

  • Automated construction of multi-dimensional potential energy surfaces:
    The accurate and efficient determination of multi-dimensional potential energy surfaces with one or several local minima based on high-level single and multi-reference ab initio methods is a non-trivial task, but is needed for the evaluation of vibrational wavefunctions or quantum-dynamical calculations. We develop techniques to reduce the computational effort while retaining very high accuracy [47]. A collection of highly accurate potential energy surfaces is provided in our PES database

  • Calculation of vibrational wavefunctions: We develop and use single and multi-reference self-consistent field methods (VSCF and VMCSCF [71,77]) to determine vibrational wavefunctions and vibrationally averaged molecular properties. This work is based on the full Watson-Hamiltonian [74].

  • Vibration correlation effects: We use methods in analogy to their well-known counterparts in electronic structure theory in order to account for correlation effects. These methods comprise single and multi-reference variational and perturbational approaches, as for example VCI [66], VMRCI [90], VMP2, VMRPT2 [90].

  • 2nd order vibrational perturbation theory: Besides variational methods, we have also an implementation of VPT2 theory, which allows us to to determine vibrational transition energies and many vibrational constants at this level.

  • Spectra relying on more than one potential energy surface: Photoelectron or photoionization spectra require the knowledge of more than one PES. Methods are developed, which allow for the efficient calculation of Franck-Condon factors or optical band shapes including Duschinsky effects [70]. All methods rely on highly correlated and fully anharmonic vibrational wavefunctions.



  • Calculation of accurate infrared and Raman spectra: In collaboration with experimentally working groups we compute and analyze complex vibrational spectra of molecules and molecular clusters. In the last years our focus has been on rather instable species, which often show strong Fermi resonances [86,93]. For small systems the remaining error bar in our calculations is typically in the range of 2-3 wavenumbers, which of course increases for larger systems or high lying vibrational states.

  • Calculation of tunneling splittings in double-well potentials: The accurate determination of tunneling splittings of excited vibrational states requires highly accurate potential energy surfaces and converged vibration correlation calculations [89]. Currently our approach is limited to rather small molecules and we are working on strategies to extend this to larger systems.

  • Calculation of photoelectron and photoionization spectra: The prediction of photoelectron spectra is tedious due the calculation of excited state Born-Oppenheimer surfaces. Besides that the Duschinsky rotation connecting the two surfaces leads to further complications. We compute such spectra beyond the harmonic approximation based on highly correlated wavefunction. Besides that we investigate alternative routes to determine optical bandshapes making use of time-independent Raman wavefunctions.

  • Calculation of molecular properties: In our collaborations we a frequently asked if we can provide vibrationally averaged molecular properties as for example geometrical parameters, rotational constants, spectroscopic constants in general or vibrationally averaged dipole moments. Yes, we can – at very high levels.