Electrical tuning effect for Schottky barrier and hot-electron harvest in a plasmonic Au/TiO2 nanostructure

Schottky barrier controls the transfer of hot carriers between contacted metal and semiconductor, and decides the performance of plasmonic metal–semiconductor devices in many applications. It is immensely valuable to actively tune the Schottky barrier. In this work, electrical tuning of Schottky barrier in an Au-nanodisk/TiO2-film structure was demonstrated using a simple three-electrode electrochemical cell. Photocurrents excited at different wavelength significantly increase as the applied bias voltage increases. Analyzing and fitting of experimental results indicate that the photocurrent is mainly affected by the bias tuning position of Schottky barrier maximum, which shifts to metal–semiconductor interface as applied voltage increases, and enhances the collection efficiency of the barrier for plasmonic hot electrons. The conduction band curvature of 0.13 eV was simultaneously obtained from the fitting. This work provides a new strategy for facile tuning of Schottky barrier and hot-electron transfer across the barrier.


Scientific Reports
| (2021) 11:338 | https://doi.org/10.1038/s41598-020-79746-5 www.nature.com/scientificreports/ as width of depletion region, position of Schottky barrier maximum, height of barrier caused by band curvature, etc. 30 . These factors may seriously affect the transfer of hot carriers, but they were usually ignored. In this work, Schottky barrier of an Au-nanodisk/TiO 2 -film (AT) structure was electrically tuned in a simple three-electrode electrochemical cell. As the increase of reverse bias voltage, photocurrent originated from plasmonic hot-electron transfer from Au nanodisks to TiO 2 significantly increases. Based on fitting and analyzing of experimental results, it is found that the increase of photocurrent is mainly induced by electrical tuning of Schottky barrier maximum position, which improves collection efficiency of the barrier, rather than Schottky barrier height.

Results and discussion
To achieve the electrical tuning of Schottky barrier and further control of hot-electron injection from Au to TiO 2 , an external bias should be effectively applied on the Au-TiO 2 Schottky junction. In this sense, the existence of Schottky barrier on Au-TiO 2 interface of the investigated structure should be confirmed, while barrier on other interfaces in the structure should be avoided. In Fig. 1a, the current of an In-Ga/TiO 2 /In-Ga structure is proportional to voltage from − 2 to 2 V, showing an Ohmic characteristic. In-Ga alloy connects TiO 2 and electrode with an Ohmic contact, so the contact properties of TiO 2 with Au or ITO can be revealed by replacing one of the In-Ga alloy layers. In Au/TiO 2 /In-Ga structure, the Schottky contact of Au and TiO 2 can be corroborated by   Fig. 1c indicates that ITO and TiO 2 are also in Ohmic contact. Therefore, in AT structure fabricated on ITO glass, there is Schottky barrier only on the interface of Au and TiO 2 .
When AT structure was tested as working electrode in an electrochemical cell, voltage drop also occurs at metal-electrolyte interfaces, including those of nanoparticle-electrolyte and counter electrode-electrolyte, because of electric double layer on the interfaces. However, the existence of Au-TiO 2 Schottky barrier makes this fraction prone to be small. Besides, the different Fermi levels of TiO 2 and electrolyte give rise to a space charge layer at TiO 2 -electrolyte interface, and further induce another TiO 2 -electrolyte barrier on the interface. The barrier is parallel with Au-TiO 2 Schottky barrier, avoiding short circuiting the latter. Therefore, external bias can be effectively applied on the Schottky junction of Au and TiO 2 , while a fraction of voltage drop inevitably exists on other interfaces and components of the investigated system. Electrical tuning of the Schottky barrier can be revealed by experiments at different bias voltages.
I-V curve of AT sample in the electrochemical cell was also measured and shown in Fig. 2a. The nonlinear curve and calculated Schottky barrier height of approximately 0.8 eV is in accordance with the analysis above. Figure 2b shows the photocurrents of AT structure with nanodisk diameters of 80 nm (D80) and 100 nm (D100) at different bias voltages. Excited by incident light, hot electrons generated from localized surface plasmon (LSP) decay in Au nanodisks pass through the Schottky barrier to the TiO 2 film, and induce positive photocurrents. When 300 mV bias was applied, the photocurrents of both samples significantly increase several times. Considering hot electrons excited at a certain wavelength in nanodisks have almost the same energy distribution, it can be inferred that the enhancement of photocurrent comes from the electrical tuning effect for the Schottky barrier, which affects the hot-electron transfer efficiency from Au nanodisks to TiO 2 film.
Bias tunes the Schottky barrier in different aspects as illustrated in Fig. 3. When a reverse bias ( V a < 0 , positive on the semiconductor side, negative on the metal side) is applied to the Schottky junction, Fermi level of the semiconductor ( E f ) decreases at a degree of qV a to E ′ f . And semiconductor conduction band minimum also It can be seen that as the magnitude of applied reverse voltage increases, x m decreases and Schottky barrier shifts close to the metal-semiconductor interface.
Incident light with photon energy higher than Schottky barrier excites hot electrons in the metal and transports them from metal to semiconductor conduction band. This effect is also called internal photoemission and its efficiency ( η ) is influenced by multiple factors as follows: where A is optical absorbance of the metal part, F e is the fraction of photons generating photoelectrons and contributing to photocurrent, P E is the probability of photoexcited electrons overcoming Schottky barrier after scattering with cold electrons and boundary surface, η c is collection efficiency of the barrier. Among the above four parameters, A and P E are independent on Schottky barrier, while F e and η c are functions of Schottky barrier height and Schottky barrier maximum position, respectively 33 .
To reveal how the bias tuning Schottky barrier affects internal photoemission and hot electron harvest, relations between bias and F e , η c are analyzed. Incident photon-to-electron conversion efficiencies (IPCEs) of AT structure (D80) at different bias voltages were measured and calculated according to where n e , N, h, c, I, e, P and λ are collected photoelectron number, incident photon number, Plank constant, light velocity, photocurrent, electron charge, incident light power, and light wavelength, respectively. As Fig. 4a shows, IPCEs at all bias conditions exhibit a peak around 600 nm for LSPR of the structure. At this condition, the resonance results in an enhancement of plasmonic hot electron generation. Besides, the IPCEs significantly increase with the bias from − 100 to 700 mV, which is coherent with the photocurrents in Fig. 2. The peak IPCE with 700 mV bias is as high as about seven times of that without bias. Estimating from Eq. (2) with common TiO 2 parameters, bias (< 1 V) induced Schottky barrier height variation for an Au-TiO 2 interface is quite small (tens of milli-electron-volts) 29,34 . Compared with the total Schottky barrier height near 1 eV, the influence of applied voltage on F e , which is a function of Schottky barrier height, can be neglected, and it cannot support the significant increase of IPCEs with bias.
Then, it comes to the relation between bias and η c . The barrier collection efficiency η c indicates the probability of hot electrons migrating from metal interface to Schottky barrier maximum without scattering, and is given by   Fig. 4b, data points of IPCE with bias voltage at different excitation wavelength obtained from Fig. 4a were fitted using the relation of IPCE = ke −B(V i −V a) − 1 4 . Fitting parameters were given in Table 1. According to Eq. (3), k is proportional with AF e P E and varies with A at different wavelength. It is clear that the fitted dash curves in the figure match the experimental data well. Moreover, fitting at different wavelength is in good agreement, having the same fitting parameters of V i = 130 mV and B = 15 . The obtained conduction band curvature qV i of 0.13 eV here is quite close to the ones in references 35 , and it further supports the fitting results. Therefore, it can be deduced that the hot-electron harvest efficiency of  www.nature.com/scientificreports/ the plasmonic Au/TiO 2 structure is mainly determined by the bias tuning Schottky barrier maximum position x m . Taking N d = 2.5 × 10 25 m −3 , ε s = 8.85 × 10 −10 F · m −1 from a reference with similar samples into Eq. (2), x m can be estimated to be 0.3 nm 36 . For x m is proportional to (V i − V a ) − 1 4 , as the reverse bias increases, the Schottky barrier maximum position shifts to the metal-semiconductor interface and decreases to 63% of the primitive one at 700 mV reverse bias. In fact, x m can be further electrically tuned to 56% of the primitive value at a reverse bias voltage of 1.2 V, above which the electrolysis of water will happen. The deduced Schottky barrier maximum position significantly enhances the hot-electron collection efficiency of the Schottky barrier, and thus achieves an improved photocurrent.

Conclusion
In conclusion, Schottky barrier of AT structure was electrically tuned in a simple three-electrode electrochemical cell. Appling a reverse bias, photocurrents and IPCEs at different excitation wavelength significantly increase as the applied voltage increases. According to fitting results, the photoelectric response obeys a relation of It indicates the photocurrent is mainly controlled by bias tuning Schottky barrier maximum position, rather than Schottky barrier height. Schottky barrier maximum position increases photocurrent www.nature.com/scientificreports/ via shifting to the interface and enhancing collection efficiency of Schottky barrier for hot electrons. The conduction band curvature of Schottky barrier was also obtained to be 0.13 eV from the fitting. This work suggests a new strategy to facilely and reversely tune the Schottky barrier and hot-carrier transfer across the barrier. It is highly beneficial to improve the performance of plasmonic hot-carrier devices in photocatalysis and photovoltaic systems.

Methods
Sample fabrication. AT structure was fabricated on ITO glass substrate. First, a 2 nm thick Cr adhesion layer was deposited on the substrate, followed by sputtering a 100 nm thick TiO 2 film in an O 2 (3 sccm) and Ar (50 sccm) plasma at 5 mTorr and 0.8 kW DC. Then, Au nanodisks with average diameter 80 nm or 100 nm, and height 30 nm were fabricated on the TiO 2 film using hole-mask colloid lithography (HCL, Fig. 5a) followed by annealing at 350 °C. The structure and SEM image of AT sample are shown in Fig. 5b,c, respectively. Au nanodisks randomly distributed on TiO 2 film, and their diameter is relatively uniform. In-Ga/TiO 2 /In-Ga and Au/ TiO 2 /In-Ga structures were fabricated by coating In-Ga alloy layer, TiO 2 film, and In-Ga alloy or Au layer in sequence. In-Ga/TiO 2 /ITO structure was also fabricated by sequentially coating TiO 2 film and In-Ga alloy layer using ITO glass as substrate. Further details of sample fabrication are referred to the literature 37 .
Photocurrent measurement. Photocurrents of AT structure were measured in a three-electrode system using an Ag/AgCl wire reference electrode, a Pt counter electrode, and standard phosphate buffer saline (PBS) electrolyte (Fig. 5b). Photocurrents were collected at chronoamperometric mode with constant bias voltages against reference electrode. Incident light with different wavelength was filtered out by an acousto-optic tunable filter (AOTF) from a laser driven light source (LDLS, Energetiq). Further detailed description of procedures for photocurrent measurements is referred to the literature 37 .