Injecting Electrons into CeO2 via Photoexcitation of Embedded Au Nanoparticles

The electron injection efficiency and the steady state absorptance at different photon energies for a composite system made of Au NPs embedded in a cerium oxide matrix are reported. Cerium oxide can be coupled with plasmonic nanoparticles (NPs) to improve its catalytic properties by visible-light absorption. The present work is a study of the ultrafast dynamics of excited states induced by ultraviolet and visible-light excitation in Au NPs combined with cerium oxide, aimed at understanding the excitation pathways. The data, obtained by femtosecond transient absorption spectroscopy, show that the excitation of localized surface plasmon resonances (LSPRs) in the Au NPs leads to an ultrafast injection of electrons into the empty 4f states of the surrounding cerium oxide. Within the first few picoseconds, the injected electrons couple with the lattice distortion forming a polaronic excited state, with similar properties to that formed after direct band gap excitation of the oxide. At sub-picosecond delay times, we observed relevant differences in the energetics and the time dynamics as compared to the case of band gap excitation of the oxide. Using different pump energies across the LSPR-related absorption band, the efficiency of the electron injection from the NPs into the oxide was found to be rather high, with a maximum above 30%. The injection efficiency has a different trend in energy as compared to the LSPR-related static optical absorptance, showing a significant decrease in low energies. This behavior is explained considering different deexcitation pathways with variable weight across the LSPR band. The results are important for the design of materials with high overall solar catalytic efficiency.


XPS analysis
After the growth, all samples were characterized in situ by XPS, to obtain quantitative information on the deposited quantity of CeO2 and Au and on possible variations of the chemical state of cerium oxide and of the metal. Ce 3d spectra were used to estimate the Ce 3+ concentration, by fitting with Ce 3+ -and Ce 4+ -related components, following the procedure introduced by Skala et al. 4 The spectra and the corresponding fit of a 2 nm cerium oxide before and after the growth of Au nm are reported in Figure S1 a). In both cases the Ce 3+ concentration evaluated by the fit was below the detection limit, indicating that the films have a good CeO2 stoichiometry and that the Au NPs do not relevantly alter the Ce oxidation state. The Au 4f spectrum of the Au/CeO2 sample, shown in Figure S1b, is compatible with bulk Au. Figure S1: a) Ce 3d XPS spectra of a CeO2 film before and after the growth of 2 nm equivalent Au NPs and corresponding fit; b) Au 4f XPS spectrum after the growth of Au NPs.

Aspect ratio of the Au NPs
The aspect ratio of the Au NPs, defined as the ratio between the in-plane average size and the out-ofplane average size, was estimated from the SEM image shown in Figure 1a of the main text. The average out-of-plane size was calculated as ℎ = , where is the nominal deposited Au thickness and the is the fractional surface coverage. For the sample here investigated = 2 nm and = 0.5, so ℎ = 4 nm. Considering the lateral size distribution of the NPs shown in Figure 1 b), the average NP in-plane S3 size was calculated as 5.7 nm, which gives an average aspect ratio of 1.4. Given the width of the distribution of approximately 4 nm, and its asymmetric shape towards large in-plane NP sizes, the aspect ratio of the Au NPs in the sample ranges between approximately 1 and more than 2.

UV-Vis spectrophotometry
The static optical absorptance in the UV-Vis was measured using a Xenon lamp and a polarizer, which enables to select either p or s polarization, i.e. parallel or perpendicular to the optical plane.

Simulations of the polarizability of Au NPs in CeO2
The polarizability of Au NPs was simulated using the Maxwell-Garnett model, approximating the NP shape to an oblate spheroid, i.e. an ellipsoid with the three axis a, b and c following the relationship: a = b > c, where a and b are the in-plane dimensions of the nanoparticles and c is the out of plane axis.
The polarizability is given by: 5 where  () is the dielectric function of Au, 6  is the dielectric function of CeO2, 7 , are the depolarization factors in the out-of-plane and in-plane directions, given by: is the eccentricity of the ellipsoid: and is the aspect ratio AR of the nanoparticles.
The absorption cross section of the nanoparticles is proportional to the imaginary part of the polarizability, shown in Figure S3 for

S6
Full TA maps Figure S4 shows the complete false-color TA maps acquired on the Au@CeO2 sample for the two pump energies using the UV and Vis probes in the full delay time range investigated. The LSPR-related signals, between approximately 1.6 and 3 eV decay within the first few ps for both pump energies.

Evaluation of the injection efficiency
To evaluate the injection efficiency, the ratio of the intensity of the PIA feature in the delay time range 50-250 ps and the absorbed photon density -evaluated as explained in the main text -are calculated for each pump energy, and divided by the same ratio calculated for the pump at 4.5 eV. Figure S5 shows the