204
J. Phys. Chem. C 2011, 115, 204–209
Ultrafast Dynamics at the Na/D2O/Cu(111) Interface: Electron Solvation in Ice Layers and Na+-Mediated Surface Solvation Michael Meyer,*,†,| Mathieu Bertin,‡ Uwe Bovensiepen,†,§ Daniel Wegkamp,†,| Marcel Krenz,†,| and Martin Wolf†,| Department of Physics, Freie UniVersita¨t Berlin, Arnimallee 14, 14195 Berlin, Germany, Laboratoire de Physique Mole´culaire pour l’Atmosphe`re et l’Astrophysique, UniVersite´ Pierre et Marie Curie Paris 6, CNRS, 4 place Jussieu, 75252 Paris cedex 05, France, Fakulta¨t fu¨r Physik, UniVersita¨t Duisburg-Essen, Lotharstr. 1, 47048 Duisburg, Germany, and Department of Physical Chemistry, Fritz Haber Institute of the Max Planck Society, Faradayweg 4-6, 14195 Berlin, Germany ReceiVed: August 2, 2010; ReVised Manuscript ReceiVed: NoVember 24, 2010
We have studied the influence of sodium ions bound near the ice/vacuum interface on the electron solvation dynamics in amorphous D2O ice layers by means of femtosecond time-resolved two-photon photoelectron spectroscopy. Adsorption of submonolayer coverages of sodium on top of multilayers of amorphous ice leads to the formation of Na+ ions and to pronounced changes in the observed ultrafast dynamics compared to pure amorphous ice. We identify a Na+-induced species of excess electrons which exhibits a much longer lifetime compared to excess electrons in pure D2O ice and approximate the decay of the Na-induced contribution by two decay times τ2 ) 880 fs and τ3 ) 9.6 ps. In addition, a faster energetic stabilization of the excited electrons with a rate of Σ ) 0.73 eV/ps is observed. The population of these electrons depends nonlinearly on the sodium coverage. We attribute the Na-induced contribution to a transient electron/ion/water complex which is located at the ice/vacuum interface. This interpretation is corroborated by coverage-dependent measurements and by overlayer experiments. 1. Introduction Radiation-induced processes in molecular ices are of high relevance in numerous fields, from atmospheric chemistry and astrochemistry to (photo) catalysis.1,2 Besides effects induced directly by excitation with UV or VUV photons or ionizing particles, many indirect processes can be involved, such as reactions induced by secondary electrons or chemistry driven by radical formation. In water ice, such secondary electrons may be trapped at defects and stabilized by the molecular environment via local rearrangement of the surrounding polar molecules. This process of charge localization and stabilization is often referred to as electron solvation and has been extensively studied for photogenerated excess electrons in liquid water3,4 and more recently also for water ice condensed on metal substrates.5 In the latter experiments, photoinduced electron transfer from the metal into thin ice layers is followed by ultrafast localization and solvation of the photoinjected excess electrons. The dynamics of this process is governed by the competition between the electron solvation in the ice and the electron transfer back into the unoccupied states of the metal substrate.6 Depending on the phase of the water ice (i.e., amorphous ice clusters, wetting layers, or crystalline ice structures), it has been shown that photoinjected electrons can be localized in the bulk or at the surface of the ice, and their residence time in the ice can span from a few hundred femtoseconds to several minutes.7-9 Because of these characteristics, these long-lived (metastable) species are expected to actively participate in the reactivity of the water ice surfaces under ionizing irradiation (e.g. in the * Corresponding author. E- mail:
[email protected]. † Freie Universita¨t Berlin. | Fritz Haber Institute of the Max Planck Society. ‡ Universite´ Pierre et Marie Curie Paris 6. § Universita¨t Duisburg-Essen.
stratosphere), mediated by dissociative electron attachment to molecules which are coadsorbed on the ice surface. The highly reactive character of these species has been demonstrated in the case of chlorofluorocarbons (CFC) deposited on amorphous or crystalline water films, where the dissociation of surface CFC is triggered in an efficient way by excess electrons in the ice during electron10,11 or UV photon12,13 irradiation. The photoinjection of excess electrons from an electron donor into polar molecular ice and their subsequent dynamics of localization and lifetime are therefore key parameters for the understanding of the interactions between ionizing particles and ices of polar molecules. In addition, impurities are abundant in condensed water and will modify the energetics and dynamics of electron solvation. For instance, screening effects of excess charges in the ices, such as positively charged impurities (e.g., alkali ions Na+, K+, Cs+, and so forth), are expected to influence the population, localization, and lifetime of electronic states of excess electrons and therefore also the reactivity of such systems under ionizing irradiation. Here, we study the effects of coadsorption of alkali ions and water on the electron solvation dynamics at the ice/metal interface by means of femtosecond time-resolved two-photon photoelectron (2PPE) spectroscopy. Such a study is expected to bring valuable information on the effects of positive ions on transient electronic states of excess charges in ice films. Note that the metal substrate is on the one hand a template for the ice film but on the other hand serves as an electron source driven by UV light. This subject is important for several reasons. First, the influence of alkali ions on the water surfaces is of high interest in various fields, such as electrochemistry and stratospheric or atmospheric chemistry of the earth. Recently, Taylor et al. calculated the electrostatic surface potential of condensed water on a metal surface in the presence of Na ions by first-
10.1021/jp107253g 2011 American Chemical Society Published on Web 12/16/2010
Electron Solvation in Ice Layers and Na+-Mediated Surface Solvation principle reaction modeling.14 In the case of stratospheric or atmospheric chemistry of the earth, it is known that heterogeneous reactions on the surface of icy particles in stratosphere play an important role in the chemistry involving pollutants.15,16 Moreover, even in the dark, these ice particles are exposed to constant cosmic ionizing irradiation, leading to the generation of excess electrons which can lead to surface reactions.11 In the stratosphere, ice particles contain a sizable amount of positive ions, originating from NaCl brought by the ocean evaporation or from Na or K atoms brought by meteor sputtering into the atmosphere.17 How can such positively charged species change the stabilization and localization of the excess electrons in the ice? It is therefore important to study their role on secondary-electrons-induced reactivity under ionizing radiation. The second, more fundamental, motivation to study the alkali atoms at ice surfaces aims toward a better understanding of autoionization of alkali atoms in a polar molecular medium. Theoretical18,19 and experimental20,21 studies have already been performed, showing in this case that the autoionization proceeds via a delocalization of the outer s-electron into a solvated state which is stabilized by the ionic core and the surrounding polar molecules. Depending on the temperature and the alkali atom density, this solvated electron can subsequently induce the dissociation of surrounding water molecules or can decay to the metal substrate. Time-resolved studies of the dynamics of solvated electrons in the vicinity of the alkali cations are therefore highly relevant to gain insight into the dynamics of this autoionization process. Furthermore, alkali atoms directly adsorbed on a metal surface serve as a model system for adsorbate dynamics of chemisorbed species. The dynamics of the unoccupied alkali resonance, which is a consequence of a partial charge transfer of the alkali valence electron to the metal substrate, have been studied extensively theoretically22 and experimentally.23-25 In this article, the dynamics of excess electrons in wetting amorphous ice layers on a Cu(111) surface are investigated in the presence of coadsorbed Na ions. We show that after the adsorption of small amounts of Na atoms at coverages below 0.15 monolayers (ML), a charge transfer of the former 3s electron of the alkali atom to the metal substrate occurs and leaves a positive Na-ion at the water/vacuum interface. Furthermore, we show that photoinjected excess electrons produced by a UV pump laser pulse get trapped at the ice surface. The subsequent dynamics of these electrons are investigated by means of femtosecond time-resolved 2PPE spectroscopy and are compared to electron solvation in (pure) amorphous ice layers on Cu(111). 2. Experimental Details The experiments were performed in an ultrahigh vacuum (UHV) chamber with a base pressure below 10-10 mbar. The Cu(111) crystal is mounted on a liquid helium flow cryostat and can be cooled down to 35 K. The Cu(111) surface is prepared by repeated cycles of Ar+ sputtering and thermal annealing by using standard methods described elsewhere.26 The quality of the resulting surface is checked by two complementary criteria: low-energy electron diffraction (LEED) is used to verify the good crystalline ordering of the sample, and 2PPE spectroscopy allows us to check the purity of the surface by measuring its work function and the line width of the Cu(111) surface state.27 Deuterated water (Sigma-Aldrich, purity 99.9%) is purified by several freeze-pump-thaw cycles. Amorphous wetting D2O ice layers are grown by expansion of water vapor through a
J. Phys. Chem. C, Vol. 115, No. 1, 2011 205
Figure 1. Upper panel: schematic energy diagram of the D2O/Cu(111) system covered with Na adatoms and 2PPE. By absorption of photons hυUV, electrons are excited into unoccupied electronic states. With a second time-delayed photon hυVIS, the bound electron can be excited above the vacuum energy of the sample, and its kinetic energy is measured in a TOF spectrometer. Curved arrows indicate potential decay channels into the ice and the Cu substrate. Lower panel: work function of Na/D2O/Cu(111) as a function of the Na coverage in ML equivalent determined at a temperature of 60 K. The line is a guide for the eyes. The thickness of the ice film is 5 BL.
pinhole doser onto the Cu(111) surface kept at 60 K. The coverage of the prepared water layers, which is typically 5 bilayers (BL), is determined by means of thermal desorption spectroscopy (TDS) and work function measurements.7 Na is evaporated from a commercial getter source (SAES Getters) on top of the water ice layers while the surface is kept at 60 K. The evaporation rate is calibrated by evaporating Na atoms on the bare copper crystal and comparing the measured work function with the work function shift of the Cu(111) surface as a function of Na coverage obtained elsewhere.22,23 When following these previous studies, the Na coverage could be determined in a coverage range up to 0.15 ML with a precision of 0.02 ML. 2PPE spectroscopy has been performed by using a commercial amplified Ti:sapphire femtosecond laser system (Coherent RegA) which pumps an optical parametric amplifier (OPA) and a noncollinear OPA to generate visible (vis) femtosecond laser pulses employed in the probing step. By frequencydoubling the visible OPA output or the fundamental of the RegA at 800 nm, ultraviolet (UV) pulses are obtained for initial optical excitation. In the 2PPE process, which is schematically depicted in Figure 1, a first photon at pump energy hυUV excites an electron of the Cu substrate into an intermediate unoccupied bound state. By absorption of a second photon at the probe energy hυVIS, the electron is excited above the vacuum level of the sample, where it can propagate to the detector. The kinetic energy of the photoelectrons is measured by a time-of-flight (TOF) spectrometer. The energy with respect to the Fermi level of intermediate electronic states in the 2PPE process is calculated by: E - EF ) Ekin + Φ - hυvis, where Φ ) Evac- EF is the sample work function. For time-resolved measurements, 2PPE spectra are recorded as a function of the time delay between pump and probe pulse.
206
J. Phys. Chem. C, Vol. 115, No. 1, 2011
Meyer et al.
Figure 2. Left: false-color map of the 2PPE intensity detected in normal emission as a function of intermediate-state energy and time delay of the visible probe pulse with respect to the UV pump pulse (positive delay times). The peak at E - EF ) 2.9 eV is due to solvated electrons in the water film. Right: deposition of 0.08 ML of Na on top of the ice film results in pronounced changes in the electron spectra. Please note the different scaling of the time axes.
3. Results and Discussion 3.1. Characterization of Na on D2O Ice Layers. The interaction of Na atoms with amorphous water ice has already been studied before experimentally and theoretically.20,21,28-30 These previous investigations show that, after deposition of neutral Na atoms on the ice surface, the 3s electron of the Na is spontaneously liberated into the water environment, resulting in the formation of Na+ ions. For elevated temperatures (>110 K) and Na concentration above 0.2 ML, this electron transfer results in the exothermic dissociation of surroundings water molecules to form OH- ions and leads to the formation of NaOH. However, for sample temperatures below 110 K and for a coverage of Na below 0.2 ML, the dissociation of water is strongly suppressed.20,21 Under these conditions, Vondrak et al.28 showed by photoemission experiments, which were supported by theoretical calculations, that the electrons derived from the Na 3s orbital are instead trapped near the surface of the ice and stabilized by the water molecules, and the Na+ ions remain in the close vicinity of the surface. The studies of Vondrak et al. where performed at thick layers of ice (coverage of about 3000 ML). However, for very thin layers (few ML) adsorbed at a metal surface, the decay of the alkali valence electrons by charge transfer to unoccupied metal bulk states becomes very efficient. In this process, the Na 3s electron will transfer to the ice and decay to the manifold of unoccupied metal states by tunneling through the thin ice layer. Because the systems investigated in the present article have been prepared under very similar conditions, we expect that neutral Na atoms, which are deposited on the water ice, spontaneously autoionize after adsorption, without triggering dissociation of the water molecules. Furthermore, at temperatures below 100 K, diffusion of Na ions into the water film and to the metal surface is strongly suppressed. The result of this autoionization is a build-up of positively charged Na+ ions on top of the water-ice layer, and the resulting surface dipoles will lead to a lowering of the work function. In the lower panel of Figure 1, the work function of the Na/D2O/Cu(111) system is depicted as a function of Na coverage. With increasing Na coverage, the work function Φ decreases; for a Na coverage of 0.11 ML, we observe a work function change of ∆Φ ) 0.35 eV. This is in agreement with an accumulation of a positive charge at the water/vacuum interface, which experimentally confirms the autoionization of the Na atoms. 3.2. Time-Resolved 2PPE Spectroscopy of Na+/D2O/ Cu(111). In the following, we discuss the influence of Na+ ions on top of amorphous D2O on the electron dynamics. We compare the results of time-dependent 2PPE spectroscopy on
D2O/Cu(111) with and without Na ions. We start out with the results of a bare amorphous D2O layer on Cu(111). Figure 2 (left panel) displays a typical false-color map of the 2PPE intensity as a function of intermediate-state energy E - EF (left axis) and time delay (bottom axis) for 5 BL of D2O adsorbed on Cu(111). The 2PPE spectra are dominated by a peak located at an intermediate-state energy of E - EF ) 2.9 eV. With increasing time delay, the peak maximum shifts to lower energies, and its intensity decreases nonexponentially with an initial decay time of τ ) 140 fs which slows down to τ ) 320 fs for delays >300 fs. The work function of the Cu(111) surface covered with 5 BL D2O is Φ ) 3.95 eV, in accordance with earlier work.31 This feature and the electron dynamics in this system have been discussed extensively5 and originate from solvated electrons in the pure water-ice layer. The reorientation of the surrounding water molecules energetically stabilizes the electron, resulting in an increased binding energy and localization of the charge in the D2O film. For further details, see Sta¨hler et al.5 These species, which are photoinjected into the ice layer, present a dynamical change of the binding energy which is attributed to the formation of a solvated electron/water complex. For simplicity, we refer to this transient electronic state as solvated electrons. Because they decay to the metal substrate, they present a residence time of ∼1 ps. Note that an equilibrium state of solvated electrons is not reached because the electrons decay back to the metal substrate at time scales before an equilibrium state is formed. After deposition of 0.08 ML Na onto 5 BL D2O/Cu(111), clear changes can be seen in the time-resolved 2PPE spectra depicted in the right panel of Figure 2. The spectra are dominated by a broad peak centered around E - EF ) 2.6 eV for the spectrum at zero time delay, which shifts to lower energies with increasing time delay. The qualitative behavior of the new feature is similar to that of the solvated electrons in pure water, because both a peak shift to lower energies and a decay of its intensity can be observed. However, the quantitative behavior is remarkably different even though less than a tenth of ML Na was evaporated onto the water-ice layer. In the following, we will first concentrate on the discussion of the energetic shift of the peak which occurs at time zero at E - EF ) 2.6 eV. To show these changes more clearly, we present in Figure 3 (upper panel) 2PPE spectra for different delays as vertical cuts through the false-color map in Figure 2 for Na covered D2O. The resulting 2PPE spectra are ordered vertically with respect to the time delay between pump and probe pulses. After a time delay of ∆t ) 5 ps, the peak maximum has shifted by ∼400 meV to lower energies. This
Electron Solvation in Ice Layers and Na+-Mediated Surface Solvation
J. Phys. Chem. C, Vol. 115, No. 1, 2011 207
Figure 4. XC traces for pure water ice adsorbed on Cu(111) and for different coverages with Na coadsorbed on top of the amorphous ice layer. The XC traces are normalized to one for the maximum intensity. The solid lines are exponential decay fits to the data. For further details, see the text.
Figure 3. Upper panel: 2PPE spectra of 5 BL D2O/Cu(111) covered with 0.08 ML Na for different time delays. The dashed line shows the position of the peak maximum at time zero. Lower panel: energetic position of the peak maxima with respect to the Fermi level for 5 BL D2O (circles) and additional 0.08 ML Na (triangles). The inset shows the peak positions for time delays up to 400 fs.
increase in the binding energy, which is related to a stronger bound electron with respect to the vacuum level of the system, is attributed to the response of the polar water molecules which screen the excess charge by molecular rearrangement like for the bare ice layers discussed before. Hence, the question of how the Na adatoms influence the observed dynamics needs to be considered carefully. The lower panel of Figure 3 shows the position of the peak maximum as a function of time delay for an amorphous D2O film (circles) and the corrresponding feature for 0.08 ML Na adsorbed on top (triangles). At first, we compare the initial energy stabilization rates for bare and Na-covered ice layers. Up to ∆t ) 400 fs, the distribution of solvated electrons in bare D2O has an initial energy stabilization rate of Σ ) 0.32 eV/ps. After adding Na, the initial energetic stabilization has increased compared to that of pure D2O. A linear fit of the peak position gives an initial stabilization rate of Σ ) 0.73 eV/ps. At longer delays ∆t > 2 ps, this rate significantly slows down to Σ ) 15 meV/ps. The adsorption of Na not only affects the energetic stabilization of the solvated electrons but also leads to distinct changes in the population dynamics. In order to demonstrate this effect, cross-correlation (XC) traces are presented in Figure 4 for an amorphous D2O wetting layer (5 BL) and for various coverages of Na (0.02-0.08 ML). The correlation traces are obtained by integrating the 2PPE intensity over a selected energy window (as defined below) as a function of time delay. Because the XC curves are the integrated signal of the intensity of the spectroscopic feature of a particular state, the intensity of the XC traces is a direct probe of the population of the state. For the pure D2O layer, this energy window ranges from E - EF ) 2.3-3.1 eV. Because of the broad distribution of excited electrons, the integration was performed in the case of Na-covered layers from
E - EF ) 1.7-2.8 eV. All correlation traces exhibit a nonexponential decay. When comparing the XC traces of the Na-covered ice layers with the ones of the solvated electrons in pure water ice, a new component can be observed which exhibits significantly longer lifetimes. With increasing Na+ coverage, this component becomes more dominant. For the largest investigated Na coverage of 0.08 ML Na, the electron population can be observed up to a time delay of ∆t ) 20 ps. The decay of the electron population is clearly nonexponential. As a first approximation, we fit the XC traces with a triexponential decay within the first 5 ps after photo excitation. The first exponential decay takes the fast initial decay of the solvated electrons in pure water into account (τ1 ) 110(10) fs), in agreement with earlier findings.5 The latter two are assigned to the slower dynamics of the solvated electrons mediated by the presence of Na+ ions. The fit yields decay times of τ2 ) 880(50) fs and τ3 ) 9.6(5) ps for the second and third exponential decay, respectively. The slowing down of the back transfer with increasing time delay has also been observed for the electron solvation in pure water5 and ammonia.32 With ongoing solvation, the excess electrons gets more stabilized in energy, and the wave function of the solvated electron becomes more confined. This leads to a reduced wave-function overlap of the electron with unoccupied metal states and therefore to a smaller back-transfer probability. So far, we have demonstrated that the electron dynamics in amorphous ice layers are strongly changed by adsorption of submonolayer coverages of Na ions. We propose that the presence of the alkali species leads to a build-up of an electron/ alkali-ion complex which is solvated by the surrounding water dipoles. In order to verify this picture, we performed experiments to identify the binding site of the electron/alkali-ion/water complex on D2O ice. In a previous publication, we have shown by Xenon overlayer experiments8 that, for smooth amorphous water multilayers, the solvated electrons reside inside the bulk of the water layer. If the long -living electrons which we observe here are bound to the Na ions located at the ice/vacuumnterface, we can expect to influence their properties also by adding an additional ice layer on top of the Na ions. Figure 5 shows the results of such an experiment, where 3 BL of water were added on top of the Na-covered 5 BL D2O multilayer. The upper panel shows XC traces taken before and after the adsorption of the water overlayer. Clearly, the population decay of the solvated electrons is affected by the additional water ice. The lifetime of these excess electrons
208
J. Phys. Chem. C, Vol. 115, No. 1, 2011
Figure 5. Overlayer experiment to determine the binding sites of the solvated electrons in the case of Na-covered amorphous water adsorbed on Cu(111). Upper panel: XC traces of 5 BL D2O/Cu(111) (gray circles), 0.08 ML Na (gray triangles), and with an additional 3 BL thick D2O overlayer (black diamonds). Lower panel: time-dependent binding energy of the solvated electron.
determined by a single-exponential decay fit between 1 and 2 ps is reduced from τ ) 1.3(1) ps before water adsorption by 30% upon adlayer adsorption to a decay time of τ ) 0.9(1) ps. In addition to changes in the population dynamics, the energetic stabilization is modified as can be seen in the lower panel of Figure 5. The initial energetic stabilization rate is reduced from Σ ) 0.73 eV/ps before the adlayer to Σ ) 0.20 eV/ps by adding the top water layer. On the basis of these results, we conclude that the excess electrons that we probe for Na/D2O/Cu(111) are located at the Na-covered ice/vacuum interface and presumably form an electron/Na+/water complex, in contrast to the solvated electrons in amorphous ice multilayers where the solvated electrons reside in the bulk part of the layer. In the following, we discuss the contributions of the two solvated electron species, that is, the solvated electrons in pure water ice (species I) and the electron/Na+/water complex (species II), to the 2PPE spectra and the dependence of species II on the Na-ion coverage. The situation is depicted in a very schematic illustration in Figure 6. A separation of the two species through a spectral decomposition in the 2PPE spectra is not obvious. The spectroscopic signature from the solvated electrons is rather broad (∼0.5 eV) allowing no clear identification of two peaks which could be assigned to the two different species. Nevertheless, a separation can be made on the basis of the population dynamics. The initial characteristic decay time τ1 ) 110(10) fs of solvated electrons in amorphous ice can also be observed in the XC traces after Na deposition (see Figure 4), whereas the characteristic decay times of the Na-induced species are considerably longer (τ2 ) 880(50) fs and τ3 ) 9.6(5) ps). In order to investigate the dependence of the population dynamics on the Na+ coverage, we made the following assumption: the more Na is deposited on the ice layer, the more trapping sites are available to trap an electron at an alkali-ion/ water complex. If no interaction between neighboring sites occurs, the 2PPE intensity from species II in the XC curves should scale linearly with Na coverage; that is, doubling the
Meyer et al.
Figure 6. Upper panel: schematic representation of the solvated electron in a pure amorphous water layer referred to as species I which occupies a binding site in the bulk of the ice film. The solvated electrons at the Na+-ion/water complex are bound in the vicinity of the water/ vacuum- interface and are named species II. Lower panel: ratio of the amplitudes of the triexponential decay fit. A2/A1 and A3/A1 describe the proportion of the amplitude of the second and third decay, which describe the dynamics of the species II electrons, with respect to the first one, respectively. The behavior of the dashed lines would be achieved if the amplitude scaled linearly with the amount of Na ions, i.e., the Na coverage.
Na coverage should lead to a doubling of the amplitudes of the corresponding components in the fit of the decay dynamics. To verify this assumption, we fitted the XC trace of the 0.04 ML Na-covered water surface to determine the decay times and amplitudes. For the other Na coverages, fits were obtained for fixed decay times τ1, τ2, and τ3, only allowing the amplitudes A1, A2, and A3 to be adjusted. The results of this analysis are plotted in the lower panel of Figure 6. The amplitudes of the second and third decay (A2 and A3, respectively) are normalized with respect to the amplitude of the first decay, yielding the ratios A2/A1 and A3/A1 as a function of Na coverage. By taking into account that, in a 5 BL amorphous ice film, considerably more solvation sites for species I exist than sites for species II, A1 can be assumed to be independent of Na coverage, whereas species II is expected to scale according to increased number of available sites with Na coverage. We indeed find such an increase in A2/A1 and A3/A1 (see Figure 6), which depends, however, nonlinearly on Na coverage. For comparison, the dashed lines depict the expected linear dependence of the amplitude on the Na coverage for both ratios A2/A1 and A3/A1. We consider different possible explanations for the observed nonlinear increase, which will be discussed in the following. The dependence of the population dynamics on the Na+ coverage can be understood when the decrease in the work function for increasing Na coverage is taken into account (see lower panel of Figure 1). This change in the electronic structure of the system (from Φ ) 3.80 eV for 0.04 ML Na to Φ ) 3.65 eV for 0.08 ML Na) can lead to changes in the matrix elements involved in both the pump steps and the probe steps of the photoemission process, which can result in a higher probability to detect an excited electron. In addition, because the state of the electron/alkali-ion/water complex is pinned to the vacuum level of the system, it shifts to lower energies with respect to the Fermi level if the work function is decreased. This downshift can lead to an increased efficiency of the electron transfer in the pump step (e.g., because of a higher abundance of photo-
Electron Solvation in Ice Layers and Na+-Mediated Surface Solvation excited electron in the substrate) and thus to an enhanced population build-up of the Na-induced species II sites. Besides these global effects on the population and probing efficiency, also local effects at the solvation sites can have an influence on the relative amplitude of species II. By assuming a random distribution of Na ions on the ice surface, the mean distance of neighboring Na ions would be 17.9 or 14.6 Å for coverages of 0.04 or 0.06 ML Na+, respectively. On the basis of angle-resolved 2PPE measurements performed at solvated electrons in pure D2O ice, we estimated in an earlier work a spatial extent of the electronic wave function of the species I electrons with a diameter on the order of 10-20 Å.31 By assuming that the spatial extend of the electronic wave function of the species II electrons is similar, it seems likely that, above a critical Na coverage of ∼0.05 ML Na+, an interaction of the delocalized species II electron with two neighboring Na ions sets in. This interaction could result in energetically more favorable solvation sites in the proximity of neighboring Na ions. These trapping sites including two Na ions within the extend of the electron wave function can result in a higher population efficiency, because the better initial trapping of the excited electron reduces the back-transfer probability to the metal substrate. At even higher Na coverages exceeding the coverages investigated in the present work (>0.15 ML), the formation of Na dimers is possible. Mundy et al. showed in their theoretical work on Na(H2O)6 clusters that the autoionization process of the Na atom is more likely if they form Na2 dimers.18 Hence, the 3s electron of the Na is better stabilized, which means that, in the vicinity of the ionized Na2 dimer, an energetically favorable solvation site is formed. 4. Conclusions We have studied the influence of Na adsorption on the femtosecond solvation dynamics of photoinjected excess electrons in thin amorphous ice layers on Cu(111). We find that the energetic and population dynamics of excess electrons are strongly modified compared with the electron solvation in pure D2O layers. On the basis of our results, we propose the following scenario: adsorption of Na atoms at coverages
