A single electron in a polar environment like water is one of the simplest systems to figure the process of
solvation: The localized electron attracts the positively charged ends of the polar solvent molecules, thus
inducing a reorientation of the molecules to screen the electron, which results in an energy gain. This process
evolves on an ultrafast timescale of 10-1000 femtoseconds (1 fs=10-15s).
The dynamics of charge transfer, localization and solvation phenomena are of great importance for
electrochemistry and photocatalysis especially at interfaces since they determine the time intervals in which
the electron or ion actually can take part in a reaction. If one understands the basic mechanisms, one might
consider controlling these dynamics, for example to improve the efficiency of a certain reaction. Our study
shows a possible route for such an engineering by making use of the correlation of solvation dynamics and
structure of a solid solvent, in our case ice.
There are a number of related questions. For example, recently it became clear that solvated electrons enhance
the dissociation of chlorfluorocarbons adsorbed on ice, which is important to understand the depletion of the
ozone layer in the earth's atmosphere.
An ideal candidate to investigate the phenomenon of electron solvation is the photoinjection of an electron
into an ice layer adsorbed on a metal surface and to monitor the evolution of the electron's energy in real
time by fs time-resolved photoelectron spectroscopy.
Fig. 1: A fs-laser pulse transfers the electron from the copper into the ice layer, where the molecules reorient and form a solvation shell.
To realize this scheme we adsorb water molecules from the gas phase on a cold (100 K) Cu(111) metal surface.
The water, which covers the Cu surface, is supercooled and forms an amorphous ice layer of a few Angstroem
In the technique of two-photon photoemission spectroscopy (2PPE), an ultrashort laser pulse (duration 65 fs) in
the UV (320 nm wavelength or hv1=3.9 eV) excites electrons in the metal to energy levels above the Fermi energy, where
the electronic states in the metal overlap with the conduction band (CB) of the ice layer. Hence, a number of
excited electrons transfer from the metal into the ice as sketched in Fig. 1 (red arrow, see also Fig. 2, step 1).
In order to analyze the transferred electrons, we sent a second fs-laser pulse (hv2=1.95 eV) onto the sample to
photoemit the electrons. The electron kinetic energy and momentum is analyzed in a time-of-flight electron
spectrometer. The second pulse (the 'probe') is time-delayed with respect to the first 'pump'-pulse and thus
traces the solvation dynamics as the delay is varied.
Fig. 2: Scheme of the electron round trip: Energy diagram for ice on Cu(111) (left panel) and as a function of
the molecular configuration or solvent coordinate (right panel).
After injection into the CB of the ice layer the delocalized electron has several relaxation channels towards
lower energy: (a) It can instantanously be transfered back to metal and lose its energy within the metal
resulting at an energy level near the Fermi level EF. This means the electron will not contribute to
further processes in the ice layer. (b) It can also relax within the conduction band. A look at the
time-resolved 2PPE spectra in Fig. 3 reveals the following: At short times a rather small intensity is found
at the energy of the CB (3 - 4 eV) due to the presence of transferred electrons and their relaxation. This
feature is labeled eCB. The maximum intensity of the 2PPE spectra, however, is found at energies below the band
edge. As time evolves, this intense peak shifts to lower energies as marked by the red line. We call this
Fig. 3: 2PPE spectra as a function of pump-probe delay, the intensity is color-coded.
What is the origin of this behavior? The right panel of Fig. 2 shows the energy as a function of the
configuration of the ice, which is the sum of electronic energy and the energy caused by lattice distortion in
a harmonic approximation. Here, one find states at energies below the edge of CB at ice configurations different
from the equilibrium, which is marked by q1(0). If the lattice is stretched or pressed, the delocalized
electronic states shift to higher energies, as it is plotted for the conduction band by VCB(q1). Since the CB
has a certain band width, it is also possible to couple several states to form an electronic wave packet, which
represents a localized state. For certain configurations, this localized state shifts to energies below CB as
described by VS(q1). Therefore, the electron gains energy when it localizes, a process which occurs at favorable
sites in the ice layer.
Fig. 4: Angle dependent 2PPE spectra at time zero (left) and flattening of the dispersion at
100 fs delay proves localization of the initially delocalized electronic state (right).
It is the uniqueness of a time-resolved 2PPE experiment to be able to follow the transition
from a delocalized to a localized state by variation of the photoemission angle a, because one
has access to the electron momentum parallel to the surface since k|| is proportional to
sin a. The average momentum of a localized state in rest is zero and its energy does not depend on k: E(k)=const.
In contrast, a delocalized state is characterized by its dispersion, which is E(k) prop. k2
for a free electron. The left panel of Fig. 4 shows the angle dependent 2PPE spectra. One can readily
recognize the peak shift to higher energy with increasing a. On the right, the corresponding
dispersion is plotted for different delays. While we find a positive dispersion up to 50 fs,
E(k||) is constant for 100 fs time delay. This result unambiguously identifies eS as a localized
state, which emerges faster than 100 fs. In the scheme of Fig. 2 the localization is represented
by step (2): The transfer from the conduction band potential to the potential of the localized
So far, we considered the electron injected into the ice, where it localizes. At this point,
the strong Coulomb interaction between the electron and the water dipoles comes into play. It
leads to a rearrangement of the surrounding water molecules. Literally, the electron digs
itself a potential well. The polar environment shields the localized negative charge and the
binding energy of the electron increases, which we directly monitor in the 2PPE spectra as a
peak shift of eS to lower energy (Fig. 3, red line). The rate of the energy shift is 270 meV per picosecond (ps).
It is governed by the mobility of the water molecules that contribute to the solvation. One
might expect that it is difficult to reorient a molecule within the solid ice structure
compared to a liquid. But, if the energetic stabilization is larger than the hydrogen bond in ice,
the system gains energy in total. In fact, the energy release is dissipated due to energy
conservation into the ice and thereby increases the local mobility. In Fig. 2, the actual
electron solvation is sketched as step (3). The corresponding potential VS(q2) is plotted
shallower than the potential of the localization VS(q1), because the solvation process
proceeds on a much slower time scale. It even continues beyond 1 ps, while the localization
takes place within the first 100 fs.
ELECTRON BACK TRANSFER
Since a solvated electron is an excited electron, it will, sooner or later, return to its
thermal equilibrium, which is at in the metal substrate close to the Fermi level.
Thus, the solvated electron has to return to the substrate via a back transfer process
(step (4) in Fig. 2). The corresponding experimental observation is the decay of the 2PPE
intensity as a function of time delay. In Fig. 5, the time evolution of the 2PPE intensity
is plotted for peaks originating from electrons out of the CB and from solvated electrons,
respectively. The decay time of the latter is a measure for the back transfer rate back to
So far, all elementary steps of the electron 'round trip' at a water-metal interface initiated
by an ultrashort laser pulse could be separated. They comprise (1) electron transfer from the
metal to the ice layer, (2) localization at an energy level just below the band edge, (3)
electron solvation by rearrangement of the solvent molecules in proximity to the electron
and (4) electron transfer back to the metal.
We now consider the rates of the transfer processes. The shielding of the electron
wavefunction due to the solvation is expected to have consequences on the lifetime of the
electron in the ice layer, which is indirectly proportional to the overlap of the
wavefunctions of the electron in the ice and unoccupied metal states. Since the environment
surrounding the electron has changed from the initial configuration q1(0) and the electron is
screened, the overlap should become smaller and the lifetime longer. This is exactly what is
observed in the experimental data:
Fig. 5: Time dependence of the 2PPE intensity of the conduction band states (eCB) and the
solvated electrons (eS), the black line reflects the experimental time resolution.
SOLVATION DYNAMICS and SOLVENT STRUCTURE
By investigating adsorbed ice layers at surfaces the sensitivity to the microscopic structure
comes for free. The mobility of the solvent determines the solvation dynamics, due to the
higher mass of molecules compared to electrons. Thus, it is an intriguing question to study
the relation of structure and solvation dynamics, which is generally impossible for
experiments in the liquid phase.
The simplest change of the ice layer from an experimental point of view is the variation of
the coverage by changing the dosing time. The layer thickness is counted in bilayers
(1 BL=2.76 Å), a quantity that measures the distance of two layers in crystalline ice. An
individual layer is called a bilayer due to its puckered structure, which resembles two
separate sheets of absorbates. If we change the ice coverage from 1 to 5 bilayers, we
observe two distinct regimes in the solvation dynamics, as shown in Fig. 6.
Fig. 6: The solvation dynamics depend on the structure of the ice layer. Two separate regimes are
observed if the coverage is changed. For high coverages layers, the solvation proceeds
initially four times slower than for low coverages.
In this diagram, the peak shift of eS is plotted versus time for various coverages. A high
coverage regime is identified above 3 BL. Within this range, the shift is independent of the
coverage. Below 2 BL eS shifts at early delays about four times faster than for higher coverage,
hence the solvation process proceeds more rapidly. The origin of this behavior is most
likely related to different configurations for solvation at a specific coverage. For high
coverages, the peak shift is independent on the coverage, which make bulk states a good candidate for
possible sites of the solvated electron. At low coverages, the ice layer does not offer bulk
configurations, instead surface molecules will respond to accommodate the excess electron
(Fig. 7). Since ice forms a hydrogen bonded network, the number of hydrogen bonds is smaller
for a H2O at the surface compared to bulk. Therefore, an individual surface molecule is less
strongly bound. In this case, a configurational change requires less energy than for higher
coverages, which leads to a faster rearrangement. Since a surface is not ideally flat and the
coverage is usually not a perfect integer of a complete bilayer, edges will be present one the
surface. As a consequence, electron solvation might also occur on a hybrid position between
bulk and surface solvation sites, as shown in Fig. 7.
Fig. 7: Cartoons of various molecular configurations surrounding a solvated electron in ice
layers on a metal surface.
Such a correlation of the solvent structure and the solvation dynamics was not observed before
and is special for solid matter. Our work represents a first step to investigate the relation
of the solvent structure and the solvation dynamics. A possible next step is to change the
metal substrate, which acts as a template for the solvent. By this means, various strains
can be exerted on the solvent, eventually leading to a better understanding of the correlation
between structure and solvation. This knowledge might then open new pathways for electro-
/photochemistry, because a certain electron solvation and electron transfer time scale, which
is required for certain process or reaction, could be chosen by using the optimal structure.