Article pubs.acs.org/JPCB
Ti3+ Aqueous Solution: Hybridization and Electronic Relaxation Probed by State-Dependent Electron Spectroscopy Robert Seidel,† Kaan Atak,† Stephan Thürmer,‡ Emad F. Aziz,†,§ and Bernd Winter*,† †
Institute of Methods for Material Development, Helmholtz-Zentrum Berlin, Albert-Einstein-Strasse 15, D-12489 Berlin, Germany Department of Chemistry, Graduate School of Science, Kyoto University, Kitashirakawa-Oiwakecho, Sakyo-Ku, Kyoto 606-8502, Japan § Department of Physics, Freie Universität Berlin, Arnimallee 14, D-14159 Berlin, Germany ‡
S Supporting Information *
ABSTRACT: The electronic structure of a Ti3+ aqueous solution is studied by liquid-jet soft X-ray photoelectron (PE) spectroscopy. Measured valence and Ti 2p core-level binding energies, together with the Ti 2p resonant photoelectron (RPE) spectra and the derived partial electron-yield L-edge X-ray absorption (PEY-XA) spectra, reveal mixing between metal 3d and water orbitals. Specifically, ligand states with metal character are identified through the enhancement of signal intensities in the RPE spectra. An observed satellite 3d peak structure is assigned to several different metal−ligand states. Experimental energies and the delocalized nature of the respective orbitals are supported by ground-state electronic structure calculations. We also show that by choice of the detected Auger-electron-decay channel, from which different PEY-XA spectra are obtained, the experimental sensitivity to the interactions of the metal 3d electrons with the solvent can be varied. The effect of such a state-dependent electronic relaxation on the shape of the PEY-XA spectra is discussed in terms of different degrees of electron delocalization.
■
INTRODUCTION Research into the electronic structure of aqueous solutions is a rapidly growing field in physical chemistry,1 propelled by the recent advances in making X-ray spectroscopy techniques applicable to highly volatile liquid solutions.2−4 Here, one key experimental development is the vacuum liquid microjet technique.5,6 In conjunction with tunable X-ray energies from modern synchrotron radiation facilities, it becomes possible to study in microscopic detail atomic and molecular species in a liquid environment, and there is a particular interest in the natural aqueous phase. Spectroscopies utilizing soft X-rays are ideal for accessing local solute−solvent interactions, both in the electronic ground state, e.g., giving accurate ionization energies, and in the electronically excited state, revealing the nature of orbital mixing, as well as short-lived reaction intermediates. Initially, the technique of total fluorescence-yield X-ray absorption (TFY-XA) was applied to study various organic and inorganic solutes in water.7−16 Recently, more powerful inelastic X-ray scattering (RIXS17) measurements were conducted,17−21 allowing state-dependent photon-emission analysis. Analogous electron-emission detection, in an energy-resolved fashion, has rarely been attempted, arguably because of the inherently greater difficulty in detecting electron energies from a highly volatile environment.22 To the best of our knowledge, fewer than a handful of resonant photoelectron (RPE) spectra and the derived partial electron-yield X-ray absorption (PEYXA) spectra have been reported to date from the aqueous phase.22−24 Those are all for transition metal ions. Total © XXXX American Chemical Society
electron-yield (TEY) measurements from aqueous solutions have been reported in refs 25−28. Neglecting electron-emission channels is certainly unsatisfying because electron and photon channels provide complementary electronic-structure information, probing different aspects of the electronic interactions. Also note that nonradiative decay by far dominates the radiative channels for sufficiently light elements.29 It should be pointed out that in many cases (also in this work) kinetic energies of the Auger electrons are large enough to probe sufficiently deep into solution. Hence, unlike often assumed, essentially bulklike regions can be probed, as has been shown in several recent experiments in aqueous solutions.30−32 This work reports RPE spectra from a TiCl3 aqueous solution. Our focus is on state-dependent electronic relaxation, through the detection of distinct Auger-decay channels, 2p3d3d (leading channel) and 2p-3p3d, following metal 2p−3d electron promotion (L-edge excitation). Both processes are illustrated in Figure 1. Ti3+ was chosen because of its 3d1 valence electronic configuration. Arguably,33−36 this aqueousphase transition metal ion is thus an ideal system for studying mixing between metal 3d states and the valence electronic states of water, which is relevant for metal reduction and oxidation in (solution) chemistry in the context of homogeneous and heterogeneous catalysis.28 Received: April 7, 2015 Revised: July 24, 2015
A
DOI: 10.1021/acs.jpcb.5b03337 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
partially based on the Ti2O3 oxide literature, provides a first hint at the electron delocalization and at the energies of involved electronic states (section D). This is inferred from the fact that the leading 2p-3d3d Auger channel is sensitive to (valence) orbital mixing between metal ion and water solvent but the 2p-3p3d channel is not. In the latter case, when bound (deeper) 3p electrons that do not participate in the electronic interaction between the solute and solvent are probed more strongly, electronic delocalization channels are not accessible at the given energy. To directly and unambiguously identify states of mixed metal−solvent character, we thoroughly analyze a series of RPE spectra, measured at certain characteristic positions (photon energies) of the XA spectra. Several mixed states as well as their energies are revealed from the signal enhancement in the resonant-emission process (section E). The enhancement is due to the interference of outgoing electron waves for two different electron-emission processes,37 Auger decay and direct ionization, which cannot be distinguished by the detected electron energy. At the same time, a photoelectron channel corresponding to the direct ionization of the same state exists. Both channels produce the same electronic one-hole final state (compare 2p-3d3d Auger and direct PE in Figure 1), which leads to interference (see Figure 1A, C). The resulting selective intensity enhancement of the respective emission lines37 thus allows effective detection even of subtle changes in the 3d state due to hybridization and charge transfer. In contrast, the photoionization cross section is typically too small to detect these 3d states in the off-resonant case; a similar observation has been made for Fe3+(aq).23 Although the RPE spectra provide an unambiguous fingerprint of mixed metal−water states, their actual nature cannot be inferred from experiment alone. To this end, we have performed electronic structure calculations that, even if performed for the ground state of the TiCl(H2O)52+ model complex, are yet useful for explaining several of our experimentbased qualitative conclusions (section F). For instance, calculated energies and orbital characters are consistent with the occurrence of the enhanced 3d-derived spectral features.
Figure 1. Illustration of the (A) 2p-3d3d and (B) 2p-3p3d Auger decays for 3d1 Ti3+(aq) in an octahedral ligand field following resonant excitation (2p−3d, shown by the green arrow), and of the normal photoemission process, for the same resonant photon energy (C). The final states for processes A and C are the same. Detection of process A leads to the 3d-XA spectrum and that of process B to the 3p-XA spectrum (see the text for details). Bent arrows illustrate the refill of the 2p core hole by a 3d or 3p electron. Straight black arrows in processes A and B represent the emission of an Auger electron with kinetic energy, KE, and in process C, the black arrow represents emission of a (direct) 3d photoelectron. Notice the same one-valence hole final state in processes A and C; process A is the case of participator decay.
The Results and Discussion is organized as follows. We first present valence (section A) and core-level (section B) photoelectron spectra from the TiCl3 aqueous solution from which electron binding energies of the solute and solvent can be determined for the ground-state solution. The valence spectrum also serves as a reference for interpreting the RPE valence spectra, which are superimposed by a large Augerelectron signal. Integration of the latter contribution as a function of excitation photon energy yields the electron-yield XA spectra; by choosing a particular Auger-decay channel, one generates different PEY-XA spectra. Auger-electron yields are an almost quantitative measure of the true X-ray absorption spectrum. The detection of such secondary-emission processes typically implies the possibility of partial quenching of certain Auger-decay channels as other radiative or nonradiative relaxation channels become available. As a result, a given PEY-XA spectrum differs in specific details from the true XA spectrum. The latter spectrum must be measured in a photon transmission experiment but has not been reported for Ti3+(aq). With regard to the PEY-XA spectra from a TiCl3 aqueous solution, our focus is on the small but important statedependent spectral differences. We start by presenting the experimental 2p or L-edge RPE spectra (section C), not yet discussing their spectral features in detail, and immediately turn to the signal-integrated RPE spectra as a function of excitation photon energy (section D). Although the RPE spectra are the primary experimental data, containing far more information than the EY-XA spectra, it is convenient to discuss the RPE spectra with reference to the characteristic absorption features. Also, the qualitative interpretation of the PEY-XA spectra,
■
METHODS Experimental Section. All photoelectron (PE) spectroscopy measurements were performed at the U41-PGM undulator beamline of the synchrotron radiation facility BESSY II in Berlin, Germany. The liquid jet with a diameter of 24 μm was injected into vacuum from a fused-silica nozzle; the jet velocity and temperature were approximately 40 ms−1 and 6 °C, respectively. Details of the technique and of the experimental setup have been described previously.3,6,22 Photoelectrons are detected normal to both the synchrotron light polarization vector and the flow of the liquid jet. Emitted photoelectrons pass from the main interaction chamber (operating at 10−4 mbar) through a 150 μm diameter orifice to the differentially pumped detector chamber (operating at 10−8 mbar) that houses a hemispherical electron energy analyzer equipped with a multichannel detector. The small distance of
