Hi Ricardo! Thank you for the interest. Indeed, TDDFT always have an error in the absolute transition energies. Fortunately, this error is systematic and dependant on the chosen functional so we can always shift the spectra to compare with experimental data like you said. In this case it was not necessary because the observation of the peak's chemical shifts and the shape of the spectra was sufficient comparison. Therefore, all displayed TDDFT spectra is shown as it was computed with no shifts and the absolute values of the transition energies, which are underestimated if compared to the experiment. However, if we do an average of the peak's energies weighted by their intensities, we can see that the chemisorbed- to gas- phase chemical shift is quite constant, although a bit larger than experiment. We have all this data in the paper that it is already submitted, but let me know if you need anything else.

I have made calculations with both RPA and TDA and I have not seen any difference at all between the calculated spectra. At least not for the main 1s->pi* transition. 

Hi Breno! Thank you very much for the comments and I am often interested in giving (or trying to) this more insightful talks that could evolve into more technical aspects if necessary. Answering your question, the clusters are indeed frozen, but at their computationally optimized geometries calculated with periodic boundary conditions at RPBE level. Additionally, we have optimized the C-O bond length in the clusters at each calculated step of the desorption, as this is one of the most relevant parameters for the O1s-XAS, as it was shown in the presentation.