Controlling hot electron flux and catalytic selectivity with nanoscale metal-oxide interfaces

Interaction between metal and oxides is an important molecular-level factor that influences the selectivity of a desirable reaction. Therefore, designing a heterogeneous catalyst where metal-oxide interfaces are well-formed is important for understanding selectivity and surface electronic excitation at the interface. Here, we utilized a nanoscale catalytic Schottky diode from Pt nanowire arrays on TiO2 that forms a nanoscale Pt-TiO2 interface to determine the influence of the metal-oxide interface on catalytic selectivity, thereby affecting hot electron excitation; this demonstrated the real-time detection of hot electron flow generated under an exothermic methanol oxidation reaction. The selectivity to methyl formate and hot electron generation was obtained on nanoscale Pt nanowires/TiO2, which exhibited ~2 times higher partial oxidation selectivity and ~3 times higher chemicurrent yield compared to a diode based on Pt film. By utilizing various Pt/TiO2 nanostructures, we found that the ratio of interface to metal sites significantly affects the selectivity, thereby enhancing chemicurrent yield in methanol oxidation. Density function theory (DFT) calculations show that formation of the Pt-TiO2 interface showed that selectivity to methyl formate formation was much larger in Pt nanowire arrays than in Pt films because of the different reaction mechanism.

To characterize the conductive properties (e.g., Schottky barrier height, series resistance, and ideality factor of the nanodevice) of the fabricated Pt nanowires/TiO2 Schottky nanodiodes, current-voltage (I-V) curves were obtained using a Keithley Instrumentation 2400 sourcemeter. According to the thermionic emission equation, a forward-biased current on the Schottky diode can be described by where is the saturation electric current, is the elementary charge, is the series resistance, is the active area of the Schottky diode, * is the effective Richardson constant, is the temperature, is the Boltzmann constant, is the Schottky barrier height of the nanodiode, is the ideality factor of the device 1 . For separating the bias voltage as a function of the diode current, eq S1 can be written after taking the logarithm as By fitting the measured I-V curve to this equation, a Schottky barrier height of 0.85 eV, ideality factor of 2.26, and series resistance of 550 ohms were obtained.

Supplementary Note 2 | Maintained electrical stability of the catalytic nanodiodes under methanol oxidation reaction conditions.
Similar to the experiments on the Pt film/TiO2 catalytic nanodiode, we measured the electric currents as the partial pressure of methanol was changed for 1, 2, and 4 Torr at elevated reaction temperatures ( Supplementary Fig. 4). As shown in Supplementary Fig. 5 Fig. 6). Thus, owing to the excess oxygen and low temperature conditions in this reaction, the rectification characteristics of the Schottky device were well maintained. Also, Supplementary Fig. 7 shows the I-V curve data and Schottky barrier height varying with a partial pressure of methanol, and the change in barrier height was negligible. In addition, in situ I-V curves during the reaction were relatively unchanged, and it can be assumed that the state of the Pt nanowires was maintained as a metallic state during the reaction since the I-V curve represents the state at the junction between the Pt nanowires and TiO2.

Supplementary Note 3 | Negligible substrate effect on Pt film/TiO 2 catalyst.
As confirmed in the SEM image ( Supplementary Fig. 3d), this 5 nm-thick Pt film was deposited on TiO2 without any defects, so that bottom TiO2 was not exposed to the gas-phase.
That is, in the Pt film/TiO2 catalyst, only the Pt surface participated in the catalytic reaction. In addition, to further confirm the substrate effect discussed by the reviewer, we additionally compared the selectivity after depositing the Pt film with the same thickness of 5 nm on a SiO2 substrate instead of TiO2. It was confirmed that the selectivity of Pt film/TiO2 and Pt film/SiO2 were similar in all methanol conversions, and thus it was confirmed once again that the 5 nmthick Pt film had no substrate effect.

Supplementary Note 4 | Change of chemicurrent yield according to the product formation.
Since methanol oxidation is a multi-path reaction in which two products appear (i.e., CO2 and methyl formate), it is necessary to compare the hot electron excitation in catalytic nanodiodes of Pt nanowires/TiO2 and Pt film/TiO2 according to product formation. To this end, as the partial pressure of methanol was changed, we found a pressure range that produces only CO2 by full oxidation (methanol 1 and 2 Torr) and a minimum methanol pressure (methanol 4  Fig. 10a). As shown in Supplementary Fig. 10b, under methanol 1 and 2 Torr conditions, no significant difference in the chemicurrent yield was found for Pt nanowires/TiO2 and the Pt film/TiO2 nanodiode. As mentioned earlier, because of the excess oxygen conditions in the 1 and 2 Torr environments of methanol, methyl formate was not produced, and only CO2 was produced; thus, the chemicurrent yield was affected only by the production of CO2. Therefore, in these reaction environments (methanol 1-2 Torr and O2 to 760 Torr), the TOF values of the Pt nanowires/TiO2 and the Pt film/TiO2 were similar, resulting in similar chemicurrent yields. However, in the methanol 4 Torr environment, Pt nanowires supported on TiO2 showed higher chemicurrent yield than Pt film. In this increased methanol environment, a partial oxidation reaction took place, methyl formate formed, and the chemicurrent yield was determined by partial oxidation selectivity. Thus, the Pt nanowires showed higher selectivity due to the Pt-TiO2 interface, resulting in higher chemicurrent yield.
Similar trends of the partial pressure dependence of chemicurrent yield were also observed at different reaction temperatures ( Supplementary Fig. 11). Therefore, these findings suggest that the enhanced selectivity toward methyl formate formation on the Pt nanowires/TiO2 interface can significantly change the dynamics of hot electron excitation from chemical reactions.

S9
Furthermore, we conclude that the efficiency of hot electron generation was affected by the metal-oxide interface in a multi-path reaction (i.e., production of both methyl formate and CO2 under methanol 4 Torr and O2 to 760 Torr), not in a one-path reaction (i.e., only CO2 was generated under methanol 1-2 Torr and O2 to 760 Torr). Therefore, through this change of the chemicurrent yield according to partial pressure of methanol, it can be seen that chemicurrent yield, which represents the efficiency of hot electron excitation, was greatly influenced by selectivity to methyl formate, not reactivity.

Supplementary Note 5 | Comparison of activity and chemicurrent yield in Pt nanowires/TiO 2 and Pt film/TiO 2 under hydrogen oxidation reaction.
Similar to the reaction environment of methanol oxidation (i.e., methanol 4 Torr and O2 to 760 Torr), the catalytic reactions were conducted in hydrogen 4 Torr and oxygen to 760 Torr. As shown in Supplementary Fig. 12, the chemicurrents excited by hydrogen oxidation in Pt nanowires and Pt films were similar. The chemicurrent yield values were also similar for two catalytic nanodiodes, since the hydrogen oxidation reaction was the one-path reaction that differed from the methanol oxidation environment in which the selectivity existed ( Supplementary Fig. 12c). From these results, we confirmed that chemicurrent yield, the efficiency of hot electron generation, was attributed to selectivity, not reactivity. Furthermore, when TOF values were compared, the values were similar in both catalysts, and the activation energy values were almost the same for Pt nanowires/TiO2 and Pt films/TiO2; thus, we found that the effect of the Pt-TiO2 interface was negligible in the hydrogen oxidation reaction, the one-path reaction in which only water was formed ( Supplementary Fig. 13a-c). In the hydrogen oxidation reaction in which selectivity did not actually appear (i.e., only reactivity), the change in reactivity and chemicurrent yield by the Pt-TiO2 interface did not appear. Therefore, from these control experiments, it can be concluded that selectivity has a much greater effect than reactivity in determining chemicurrent yield.

Supplementary Note 6 | Charge transfer from the methanol adsorbate to oxide supports by theoretical calculations.
To reveal the Lewis acid/base properties of TiO2 and SiO2 supports under the methanol oxidation reaction environment, we investigated the charge transfer from the methanol adsorbate to the oxide support surface by using density functional theory (DFT) calculation when reactant methanol molecule was adsorbed on TiO2 and SiO2 support. In the chemistry of Lewis acid/base properties on oxide surfaces, it is a Lewis acid when electron charge is received and a Lewis base when electron charge is lost 3 . The charge transfer by the interaction between adsorbate molecule-oxide was investigated using the Bader charge analysis simulation method, which is commonly used to examine the Lewis acidity of oxide supports [3][4][5] . As in the model designed for DFT calculation on the Au-TiO2 interface in the Yates group 6,7 , three atomic layers of Pt nanorod bonded on top of the rutile TiO2(110) and SiO2 surface was used to model the Pt nanowires/TiO2 and Pt nanowires/SiO2. We used the interfacial structure of a Pt-oxide support, as shown in Supplementary Fig. 16, to calculate the charge transfer between methanol adsorbate and oxide surfaces.  Fig. 21b and 21c). In addition, the selectivity toward methyl formate was measured to confirm the metal-oxide interface effect at increased width of Pt nanowires ( Supplementary Fig. 21d).