Comment on “State-Dependent Electron Delocalization Dynamics at the Solute-Solvent Interface: Soft-X-ray

In a recent letter, Bokarev et al. [1] create a general relationship between state-dependent fluorescence yield and electron delocalization in L-edge x-ray absorption spectra based on 2p core-electron excitations into eg and t2g orbitals for three aqueous ionic species. In this Comment, we show that Bokarev et al. do not account for previous findings and that their claims are unsubstantiated. The crucial role of state-dependent fluorescence yield was established previously [2–4]. Ab initio restricted active space self consistent field approaches for transition-metal L-edge spectroscopy have been reported before [5,6] and it was established that the state-dependence fluorescence yield for aqueous Cr3þ arises from local atomic effects [4]. In contrast to the assertions in Ref. [1] that the analysis in Ref. [4] is mostly based on semiempirical simulations of spectral line shapes, we note that the same kind of ab initio quantum-chemistry model was employed as in Ref. [1]. Furthermore, rigorous ab initio calculations of the fluorescence pathways are missing in Ref. [1] although they were shown to be essential for a correct interpretation of the fluorescence-yield x-ray spectra [4]. In spite of these findings and beyond the presented computational approach, Bokarev et al. advocate an interpretation in which the differences between total and partial fluorescence yield x-ray spectra arise, next to large x-ray optical effects, from a t2gand eg-dependent delocalization of the core-excited electron into the water solvation shell. X-ray optical effects, including, in general, more than the mentioned solvent-background variation and, in particular, the strong polarization dependence [7,8], are not quantitatively treated although this is necessary and straightforwardly possible [4]. The interpretation by Bokarev et al. is based on two arguments. The first is that the covalent overlap between the oxygen 2p derived lone-pair state of water is larger with the Fe eg than with the t2g orbitals. Second, Bokarev et al. determine the time-independent dipole transition-matrix elements for the 2p−1t12g core-excited states as smaller than for 2p−1e g. Neglecting the dominant (> 99%) nonradiative decay channels of the core-excited states, they postulate an unquantified longer lifetime of the 2p−1t 2g over the 2p−1e g states. With these assumptions they invoke a more “pronounced” or more “complete” electron delocalization in the 2p−1t 2g over the 2p−1e g resonances that should be visible as state depend-fluorescence yields depending on the orbital population of t2g and eg states. The claims by Bokarev et al. are not sufficiently supported. First, the correlation between 2p excitations into t2g orbitals and the appearance of spectral differences is not substantiated. Bokarev et al. display dominant t2g occupations for Fe3þ in Fig. 1(b) at 709.9, 710.4, and 710.7 eVand for 1⁄2FeClðH2OÞ5 2þ and 1⁄2FeCl2ðH2OÞ4 þ above 709 eV but the spectral distortions are minor or not present at these energies. In turn, the spectra deviate by 50% at 714.1 and 723.9 eV but no connection is made to orbital occupation. For Co2þ (Fig. S2) they report dominant t2g occupation for only one data point (775.4 eV). Instead, strong spectral differences at higher energies contrast with a dominant eg occupation, an inconsistency that appears to be selective interpretation. We also note that the experimental Co2þ spectra in Figs. S2 and 3 are shifted by 1.2–1.3 eV, impeding more detailed inspections. Second, Bokarev et al. estimate the lifetime of core-excited states from the minority channel of radiative decay although the lifetime is dominated by the nonradiative decays. Without calculating all decay channels and in particular the dominant Auger decays, this does not appear to be an adequate determination of the lifetime. It is important to note that the combined 2p3p3p, 2p3p3d, and 2p3d3d Auger decay channels are essentially energy independent, as has been exemplified, i.e., for the Auger decay of Ni2þ [3]. Third, the claim by Bokarev et al. that for core excitations with longer lifetimes the delocalization of the excited electron would be more pronounced or more complete appears to be mere speculation as no evidence is given to support this claim. In summary, we show that the claim by Bokarev et al. that the spectral differences between fluorescence-yield x-ray spectra and the true absorption cross section are largest for excitations of the 2p core electron into t2g orbitals and that, because the corresponding core-excited states have longer lifetimes, the core-excited electron delocalizes into the water solvation shell more “completely” is not substantiated; and that state-dependent fluorescence yield is an atomic effect without relevant electron delocalization. AComment by Green et al. [9] and the Reply by Bokarev et al. [10] have been published concurrently.