Can theory quantitatively model stratospheric photolysis? Ab initio estimate of absolute absorption cross sections of ClOOCl

We have calculated the absorption spectrum of dichlorine peroxide [ClOOCl] in the spectral range 250-400 nm. We have employed five different approaches to quantitatively model the absorption cross section (empirical broadening scheme, linearized harmonic reflection principle and full reflection principle with ground state density calculated using harmonic approximation, classical molecular dynamics, and path-integral molecular dynamics). We have also tested various single and multireference methods. We found that (i) the excitation characteristics of ClOOCl are sensitive on molecular geometries and therefore the ground state density has to be properly sampled and that (ii) single-reference methods tend to overestimate the absorption cross section and lead to a blueshift in the absorption maximum. The absorption spectrum calculated at the CASPT2 level with complete active space-configuration interaction (CAS-SCF) transition dipole moment with the ground state sampled via path-integral molecular dynamics is in a good agreement with the experiment. We, however, did not reach full agreement in the atmospherically relevant low-energy tail of the spectrum. The general message is that quantitative predictions are difficult even for a relatively small molecule as ClOOCl and the convergence, with respect to the electronic structure method, basis set used, and ground state sampling, needs to be carefully examined. (C) 2010 American Institute of Physics. [doi:10.1063/1.3499599]