Spectroscopy and dynamics of the hydrated electron at the water/air interface

The hydrated electron, e–(aq), has attracted much attention as a central species in radiation chemistry. However, much less is known about e–(aq) at the water/air surface, despite its fundamental role in electron transfer processes at interfaces. Using time-resolved electronic sum-frequency generation spectroscopy, the electronic spectrum of e–(aq) at the water/air interface and its dynamics are measured here, following photo-oxidation of the phenoxide anion. The spectral maximum agrees with that for bulk e–(aq) and shows that the orbital density resides predominantly within the aqueous phase, in agreement with supporting calculations. In contrast, the chemistry of the interfacial hydrated electron differs from that in bulk water, with e–(aq) diffusing into the bulk and leaving the phenoxyl radical at the surface. Our work resolves long-standing questions about e–(aq) at the water/air interface and highlights its potential role in chemistry at the ubiquitous aqueous interface.

Figure 3 assigns the transitions to the hydrated electron and the phenoxyl radical.In principle, the phenoxide anion could also lead to resonance enhancement of the SFG signals.
We consider this here.The S1 and/or S2 states are initially excited and could be visible in the measurements.The S1 state has been observed in transient absorption measurements taken following excitation at 266 nm, where it has a lifetime of ~20 ps (from fluorescence measurements). 6The S1 excited state absorption is around 515 nm whilst the emission is around 340 nm.Hence, neither are resonant with ω1, ω2 or ωSFG.For the S2 state, the excited state absorption spectrum has, to the best of our knowledge, not been observed. 7,8This is because the S2 state lifetime is very short leading to sub-picosecond photo-oxidation to form the phenoxyl radical and electron, 8 as seen in the current measurements.Hence, it seems unlikely that the S2 state will contribute to the SFG signal.Finally, from a dynamical perspective, we would anticipate that if the spectrum peaking around 720 in Figure 3 was arising from resonance enhancement from the S2 (or S1) state of phenolate, then its dynamics would be correlated with those of the phenoxyl radical, but they are not (Figure 2).We conclude that the most likely assignment is that the short-lived species arises from the interfacial hydrated electron and the long-lived component from the phenoxyl radical.

Consideration of electronic SFG spectrum's correspondence to the absorption spectrum
The electronic SFG spectra are expected to closely resemble the bulk absorption spectra on account of both measurements being proportional to the resonant transition dipole moment (TDM).For the case of a singly-resonant fundamental field, ω1, that induces the transition n ← g, the SFG signal contribution can be written as: 9,10 In the above expression, µng is the TDM for the transition n ← g, and state nʹ is the virtual state accessed by ω1 + ω2 or ωSFG from state g.Ωng is the resonant frequency of the n ← g transition, and Γng is its width.
In the case of e(aq) -, the primary resonance term is associated with ω1, where g is the 1s state and n the 1p state(s).The terms  ' and  ' are broadly nonresonant, in which case they are expected to be approximately constant in the wavelength regions probed in the experiment.Hence, we might anticipate that the electronic SFG spectrum will be approximately proportional to the absorption spectrum.
In the case of the phenoxyl radical, where ωSFG is resonant (as well as the sum ω1 + ω2), nʹ corresponds to an excited state of the molecule, the C 2 B1 excited state, and therefore n is a virtual state.In this scenario, the χ (2) becomes proportional to the two-photon TDM as well as that of the one-photon TDM.However, in general, a two-photon TDM is very small (especially for a small molecule with small hyperpolarizability) and the associated twophoton spectrum tends to be broader than the single-photon absorption.Hence, in the wavelength range considered here, the two-photon term is likely to be small and approximately constant so that the signal will predominantly follow the C 2 B1 ← X 2 B1 onephoton absorption spectrum.
To further support that the absorption spectra generally follow the linear absorption there have been a few studies comparing surface electronic SFG spectra with absorption spectroscopy (as well as many comparing vibrational SFG spectra with IR spectroscopy), which illustrate the often close agreement between the two.In general, these have relied on Malachite green as a test case.Sen et al. showed a close correspondence between the electronic SFG spectrum and the absorption spectrum for the S2 ← S0 transition using broadband electronic SFG. 11Qian et al. similarly showed the S1 ← S0 transition to be well captured by the electronic SFG spectrum. 12In both cases, the difference in spectral maximum between SFG and UV-vis was less than 10 nm.

Consideration of the probing depth of the SFG signal
Firstly, because the time-resolved SFG experiments are clearly different to bulk transient absorption data, 7 it seems unlikely that bulk dynamics are observed.Nevertheless, we consider here other possibilities.In principle, symmetry can be broken over an extended range because of bi-layer formation.Phenoxide has some surface affinity (although much weaker than phenol) and there is a significant (150 mM) concentration of sodium ions in the solution (due to the addition of NaOH to bring the solution to pH 13).The effect of concentration and pH was be probed by comparing previous measurements taken with ω1 at λ = 720 nm 13 to those presented here at the same ω1.Reducing the pH from 13 to 12 or reducing the concentration from 150 mM to 100 mM showed no noticeable effect on the dynamics.
We also briefly contrast the current work with our previous second-harmonic generation (SHG) work following charge-transfer-to-solvent (CTTS) excitation of iodide at the water/air interface. 14,15The hydrated electron was observed in these experiments with its kinetics appearing similar to those in the bulk and remaining visible for 100s ps, in contrast to current observations.However, the concentration was much higher (2 M NaI) and, according to molecular dynamics simulations, at such concentrations, the range over which symmetry is broken extends at least 1 nm. 16Indeed, this conclusion is consistent with experiments in which surfactants were added to the solution. 14The fact that hydrated electrons are observed for 100s ps in the SHG experiments for concentrated NaI solutions following CTTS excitation therefore suggests that the electrons remains within ~1 nm of the interface.Hence, the larger probe depth accessible in these experiments can account for the longer observation time.Alternatively, the very large concentration of counter ions (Na + ) might be driving the electrons to the interface.We intend to probe the effect of counter-ions in the current experiments in due course.

Consideration of symmetry
χ (2) , as probe in electronic SFG, is a macroscopic observable of the time-and orientationally averaged molecular hyperpolarisabilities of the surface, β.In the case of the hydrated electron, which in the bulk has a pseudo-spherically-symmetric structure, we expect the structure near the interface to be perturbed by the interfacial symmetry.More specifically, we expect the orbitals that protrude into/out of the interfacial layer, which lie along the vertical laboratory axis, to be the most affected, given they experience the largest changes in field gradients and solvent density along their length.Hence, for SFG fields in the PPP configuration that are incident at approximately 73° to the surface normal, the vertical surface component will be the largest contribution to the measured signal (χzzz (2) ).The components of χ (2) parallel to the plane of the surface are expected to have properties closer to the bulk hydrated electron, but these are probed more weakly.