One-Photon Absorption Properties from a Hybrid Polarizable Density Embedding/Complex Polarization Propagator Approach for Polarizable Solutions
We present a formulation of the polarizable density embedding (PDE) method in combination with the complex polarization propagator (CPP) method for the calculation of absorption spectra of molecules in solutions. The method is particularly useful for the calculation of near-edge X-ray absorption fine structure (NEXAFS) spectra. We compare the performance of PDE-CPP with the previously formulated polarizable embedding (PE) CPP model for the calculation of the NEXAFS spectra of adenine,
... glycine and adenosine triphosphate (ATP) in water at the carbon and nitrogen K-edges, as well as of formamide and glycine at the oxygen K-edge. In general, we find only minor differences between the performance of PDE and PE for the targeted parts of the spectra, except in the case of transitions involving Rydberg states, of the number of configurations taken into account. The NRMSD applied here is defined in eq. S1 in the SI. The NRMSD values are significantly higher for the neutral ATP than for the protonated ATP in the first 2 -3 snapshots. However, the NRMSD of the neutral form converges more quickly so the NRMSD of both the neutral and protonated forms reaches similar values for the 10 configurations. Fig. S4 also indicates that a satisfying spectral convergence can be achieved even for a relatively small configurational sampling. At the nitrogen K-edge, the intense peak at 387.8 eV arises from the atoms N3, N7 and N1. For PDE this peak is more intense and at a lower frequency than for PE, as is the case for aqueous adenine but opposite from formamide and glycine because the N3 and N7 atoms in both adenine and ATP are hydrogen receptors when H-bonding with the nearest water molecules. In the experiment, the intense peak has been found at 400.3 eV. The following peak is a N10 transition. The position of this peak fluctuates significantly throughout the different snapshots, which results in broadening of the peak. The N10 peak corresponds to the weak experimental peak at 401.5 eV. The peak at 390.0 eV is a N9 excitation and it corresponds with the experimental peak at 402.5 eV. The second N10 peak can only be identified for the PE solvent model at 391.3 eV, while for PDE it is attenuated. The explanation for this is the same as for adenine, namely that it is a probably Rydberg transition, so PDE destabilizes it more strongly through the non-electrostatic repulsion. The PDE result is for this peak more in agreement with the experiment by Kelly et al., which also does not identify this peak for the aqueous ATP. In the protonated spectrum, the leftmost peak is a mixture of N7 and N3 excitations for PDE, and a more pure N7 excitation for PE. The third peak stems from the 1s-excitation at the N1 and N9 atoms for both embedding models, and not purely N9, as the neutral case. The differences between PDE and PE are more pronounced at the nitrogen K-edge, than at the carbon K-edge (except for the region above 278 eV of the carbon K-edge of ATP).