Plasmon-enhanced light energy conversion using gold nanostructured oxide semiconductor photoelectrodes

Introduction

Global environmental and energy problems have motivated research and development of solar cells or artificial photosynthesis systems that can effectively use sunlight as a source of renewable energy [1–4]. We have advocated the concept of “effective utilization of a photon” using plasmonic nanostructures and investigated research on a highly efficient photochemical reaction using an optical antenna function related to the plasmonically enhanced optical near-field exhibited by metallic nanostructures [5–12]. Recently, we have promoted research on plasmon-enhanced light energy conversion, such as by solar cells or artificial photosynthesis systems. We demonstrated plasmon-enhanced photocurrent generation and water oxidation from visible to near-infrared light without deteriorating photoelectric conversion using photoelectrodes in which gold nanostructures were elaborately arrayed on the surface of a titanium dioxide (TiO₂) single crystal [13–17]. We recently successfully developed a plasmon-induced artificial photosynthesis system that can split a water molecule into hydrogen (H₂) and oxygen (O₂) and an ammonia synthesis system based on nitrogen fixation as an evolution of the plasmon-enhanced photoelectric conversion system [18, 19].

Photoelectric conversion systems and photocatalysts using an oxide semiconductor such as TiO₂ are mainly driven by ultraviolet irradiation [20–24]. However, ultraviolet radiation represents only approximately 4% of the sunlight energy that arrives at the surface of the earth, and therefore such systems do not use sunlight energy effectively. Accordingly, the development of photoelectrodes and photocatalysts that respond to visible light has been actively studied. Methodologies involving a response to visible light include the use of sensitizers such as organic dyes that absorb visible light [25–29] or the construction of a semiconductor that forms an impurity state by doping a transition metal [30–35]. These methods might be effective for a solar cell or a photocatalyst but are not optimal for water splitting. When the band gap of a semiconductor is narrowed to respond to visible wavelengths, it is difficult for the oxidation and reduction of water to proceed accompanied by a multi-electronic transition process. The thermodynamic reversible potential for water splitting is 1.23 V; however, an experimental oxygen evolution based on four electronic oxidations of water requires relatively high overpotential (leading to the requirements for additional a chemical bias) [36–38]. Thus, simultaneous oxidation and reduction of water using a single photocatalyst activated by visible light irradiation is difficult to achieve [39, 40]. Therefore, a Z-scheme process for water splitting using visible light irradiation is generally employed in which H₂ and O₂ evolution proceeds using photocatalysts with different energy levels in combination with an appropriate redox mediator [41–47]. However, in the Z-scheme, two different photocatalysts must be simultaneously excited, thus reducing the light irradiation area by half compared with a water splitting system driven by a single photocatalyst.

We previously determined that O₂ and hydrogen peroxide (H₂O₂) were stoichiometrically evolved upon irradiation with visible and near-infrared light based on four or two electronic oxidations of water molecules when photoelectrochemical measurements were performed in an aqueous electrolyte solution using gold nanostructure-loaded TiO₂ photoelectrodes. The gold nanoparticles appear to function as a co-catalyst of O₂ evolution because the oxidation reaction of water is accelerated in the plasmonically enhanced optical near-field. Therefore, it is also expected that water splitting can be realized with a small bias on a single photocatalyst by combining such a co-catalyst of O₂ and H₂ evolution. Here, we describe the latest results of our research on a plasmon-induced light energy conversion system based on photocurrent generation and artificial photosynthesis employing a plasmon-induced charge separation between the gold nanoparticles and the oxide semiconductor substrate.

Plasmon-enhanced photocurrent generation and water oxidation as a half reaction of water splitting

Gold nanoislands (Au-Nis) were prepared on a 0.05 wt% niobium-doped TiO₂ (Nb-TiO₂) single-crystal substrate by sputtering and annealing. A gold thin film with a thickness of 3 nm was deposited on the Nb-TiO₂ substrate, and the substrate was subsequently annealed at a temperature of 800 °C under a nitrogen (N₂) atmosphere. A scanning electron microscope image of the prepared Au-Nis/Nb-TiO₂ photoelectrode is shown in Fig. 1(a). The average estimated Au-Nis diameter was 20 nm, and the standard deviation of nanoisland size was 8 nm. Figure 1(b) shows the extinction spectrum of the Au-Nis/Nb-TiO₂ photoelectrode. The Au-Nis exhibit a localized surface plasmon resonance (LSPR) band peaking at 610 nm wavelength.

Fig. 1:

(a) SEM image of Au-Nis on Nb-TiO₂. (b) Extinction spectrum of the Au-Nis/TiO₂ photoelectrode. (c) Linear sweep voltammogram of the Au-Nis/TiO₂ electrode in the dark and under irradiation with visible light of 550 nm, 620 nm and 750 nm with light intensities of 1.31 mW/cm², 1.71 mW/cm² and 1.09 mW/cm², respectively. The scanning rate was 100 mV/s. (d) IPCE action spectrum measured using the Au-Nis/TiO₂ photoelectrode.

We performed a photocurrent measurement in an aqueous electrolyte solution (0.1 mol/dm³ KClO₄ aq.) using an Au-Nis/Nb-TiO₂ photoelectrode as the working electrode (WE) with a conventional three-electrode photoelectrochemical measurement system. Pt wire and a saturated calomel electrode (SCE) were used as the counter electrode (CE) and reference electrode (RE), respectively. Notably, the electrolyte solution in this system did not contain a sacrificial electron donor. The linear sweep voltammogram (current-potential curve) measured under conditions of visible light irradiation of 550 nm, 620 nm and 750 nm on the Au-Nis/TiO₂ photoelectrode is shown in Fig. 1(c). An anodic photocurrent was clearly observed above an applied potential of –0.2 V (vs. SCE). The IPCE action spectrum is shown in Fig. 1(d), and the IPCE value was calculated using the following eq. (1).

where J is the photocurrent value at 0.3 V vs. SCE measured by the quasi-steady-state current of the current-time curve, λ is the incident wavelength, and I is the intensity of the incident light. The IPCE action spectrum has a peak at a wavelength of approximately 600 nm, almost corresponding to the LSPR band, and thus plasmon-enhanced photocurrent generation was clearly observed.

Here, we discuss the possible mechanism of photocurrent generation. An energy diagram of electron excitation based on the plasmon-induced charge separation is shown in Fig. 2(a). The interband or intraband transition of gold was induced by the plasmonically enhanced optical near-field. The excited electrons in the Au-Nis are rapidly transferred to the conduction band of Nb-TiO₂, and the hole might oxidize water molecules as a result of the evolution of O₂ via four electronic transition processes because there is no electron donor in the aqueous electrolyte solution other than water molecules. To verify the evolution of O₂ from the Au-Nis/Nb-TiO₂ photoelectrode, O₂ content was determined quantitatively by a gas chromatography-mass spectrometry (GC-MS) apparatus using an aqueous electrolyte solution containing isotopic Water-18O (H₂¹⁸O). The chromatogram in Fig. 2(b) shows that O₂, which has a molecular weight of 34 (34-O₂), evolved over various irradiation times when the Au-Nis/Nb-TiO₂ photoelectrode was irradiated with visible light [from 450 to 750 nm (32.6 W/cm²)]. The quantity of O₂ evolution increased with increasing irradiation time. The number of moles of evolved O₂, as estimated from the irradiation dependence of the results of GC-MS analysis and the isotope abundance ratio, is plotted as a function of the observed photocurrent in Fig. 2(c), revealing a linear relationship. One oxygen molecule evolves per four electrons because water oxidation is a four electronic transition process, as shown in the following formula (2).

Fig. 2:

(a) Energy diagram of electron excitation based on plasmon-induced charge separation in the Au-Nis/TiO₂ system. (b) GC-MS spectrum of oxygen isotope 34-O₂ generated in the Au-Nis/TiO₂ system with different times of irradiation. (c) The number of moles of O₂ evolved at the observed photocurrent was estimated from the irradiation dependence of the GC-MS analysis and the isotope abundance ratio.

The yield of evolved O₂ at the observed photocurrent was estimated to be 91.6 %. This result demonstrates that water is functioning as an electron source in the photocurrent generation system and that O₂ evolves as a half-reaction of water splitting via the four-electron oxidation of water.

Plasmon-induced water splitting using both sides of a single- crystal strontium titanate substrate

To construct a water-splitting system using two electrodes acting as an anode and cathode, the electrochemical potential of H₂ evolution is important [48–50]. Therefore, we employed strontium titanate (SrTiO₃) as a photoelectrode because its conduction band potential is 0.2 V less than that of TiO₂. Au-Nis were formed on a 0.05 wt% niobium-doped SrTiO₃ (Nb-SrTiO₃) single crystal by sputtering and annealing in the same manner as for the preparation of the Au-Nis/Nb-TiO₂ photoelectrode. The LSPR band peaked at a wavelength of 600 nm, as shown in Fig. 4. For efficient H₂ evolution, platinum (Pt) was used as a co-catalyst. A Pt board was fixed on the backside of the Au-Nis/Nb-SrTiO₃ photoelectrode using an indium-gallium alloy to obtain ohmic contact between Nb-SrTiO₃ and Pt. The plasmon-induced water-splitting system using two sides of the same SrTiO₃ substrate was constructed using the prepared Au-Nis/Nb-SrTiO₃/Pt photoelectrode as shown in Fig. 3(a). The anode and cathode chambers were separated by the substrate; thus, the Au-Nis side of the photoelectrode functioned as the anode, whereas the Pt side functioned as the cathode. The gases were separated because H₂ and O₂ evolution were induced in the cathode-side and anode-side chambers, respectively. The chemical bias was induced by pH regulation, and the charge balance between the chambers was maintained using a salt bridge.

Fig. 3:

(a) Plasmon-induced water splitting system using two sides of same SrTiO₃ substrate and an Au-Nis/Nb-SrTiO₃/Pt photoelectrode. (b) The irradiation time dependence of H₂ (red triangle) and O₂ (blue inverted triangle) evolution at the pH combination of 1 and 13.

A plot of the quantity of H₂ and O₂ evolved as a function of light irradiation time is shown in Fig. 3(b). In this experiment, the pH of the cathode-side chamber was set at 1 and that of the anode-side chamber was set at 13 (that is, the chemical bias was 708 mV), and the irradiation wavelengths and intensity were 550 nm to 650 nm and 0.7 W/cm², respectively. The quantities of H₂ and O₂ evolved were linearly dependent on the irradiation time. Water splitting proceeded stoichiometrically; the quantity of evolved H₂ was twice that of O₂. Figure 4 shows the action spectrum of H₂ evolution using a histogram. The pH conditions were the same as in Fig. 3(b). In the 600 ± 50 nm, 700 ± 50 nm, and 800 ± 50 nm wavelength regions, the evolution efficiency (mol per hour) of H₂ corresponded to the LSPR band, which is indicated by the solid line in Fig. 4. Therefore, water splitting proceeded based on the plasmon-induced charge separation between Au-Nis and SrTiO₃.

Fig. 4:

The action spectrum of H₂ evolution at several wavelength regions displayed as a histogram. The solid line indicates the LSPR band.

Here, we discuss the possible mechanism of the plasmon-induced water splitting system. The plasmonically enhanced optical near-field promotes the interband or intraband transition of gold, and the excited electron can be transferred to the conduction band of SrTiO₃. The electrons injected into the Nb-SrTiO₃ reduce protons at the Pt board surface to evolve H₂, and the holes may oxidize hydroxyl ions and water molecules as a result of the evolution of O₂ at the gold nanostructured SrTiO₃ surface. The pH dependence clearly demonstrated that H₂ and O₂ evolve even at pH 3 and 6.8, respectively. The irradiation time dependence of H₂ and O₂ evolution with a pH combination of 3 and 6.8 is shown in Fig. 5. Linear H₂ and O₂ evolution were observed even at an irradiation time of 48 h, indicating the high stability of the system. Water splitting was induced by a chemical bias of approximately 230 mV, which corresponds to the difference between these pH values. In this study, water splitting was not confirmed at a pH combination of 4 and 13. This result indicates that the number of protons adsorbed to the surface of the platinum co-catalyst is critical for the plasmon-induced water splitting system, and thus, H₂ evolution is the rate-limiting process in this reaction.

Fig. 5:

The irradiation time dependence of H₂ (red triangle) and O₂ (blue inverted triangle) evolution at the pH combination of 3 and 6.8.

Plasmon-based ammonia synthesis system using visible light irradiation

As an analog of the plasmon-induced water splitting system, Au-Nis were formed on the Nb-SrTiO₃ single-crystal substrate for use in an ammonia (NH₃) synthesis system. Au-Nis were prepared on the Nb-SrTiO₃ substrate by sputtering and annealing. Ruthenium (Ru) was used as a co-catalyst for NH₃ evolution. Ru nanoparticles were deposited by an electron beam deposition system on the backside of the Nb-SrTiO₃ substrate. Figure 6(a) shows the plasmon-based NH₃ synthesis system using an Au-Nis/Nb-SrTiO₃/Ru photoelectrode. In this system, the Au-Nis side of the photoelectrode functioned as the anode (oxidation), whereas the Ru side functioned as the cathode (reduction). The chemical bias between the anode-side and cathode-side chambers was controlled by pH. Ethanol (10 vol %) was added to the anode-side chambers containing water as a sacrificial electron donor. An aqueous potassium hydroxide/ethanol solution with a pH of 13 was added to the anode-side chamber, and water vapor saturated nitrogen (N₂) gas was fed into an aqueous hydrochloric acid solution (pH 2) vessel in the cathode-side chamber to supply protons.

Fig. 6:

(a) Plasmon-based NH₃ synthesis system using an Au-Nis/Nb-SrTiO₃/Ru photoelectrode. (b) The irradiation time dependence of NH₃ formation on the cathode side of the chamber; Au-Nis-loaded Nb-SrTiO₃ with Ru co-catalyst with irradiation (pink triangle), Nb-SrTiO₃ without Au-Nis under visible light irradiation (green diamond), and Au-Nis-loaded Nb-SrTiO₃ with Ru co-catalyst without irradiation (black inverted triangle).

Figure 6(b) shows the irradiation time dependence of the quantity of evolved ammonia based on N₂ fixation in the cathode-side chamber. In this experiment, a spectrally filtered light with wavelengths from 550 nm to 800 nm (0.2 W/cm²) was used to irradiate the Au-Nis from the optical window of the anode-side chamber. The quantity of the evolved NH₃ increased linearly with light irradiation time, demonstrating that NH₃ evolution proceeds due to N₂ fixation under visible-light irradiation. By contrast, NH₃ evolution was not detected in the absence of irradiation or Au-Nis. Furthermore, in the anode-side chamber, both acetaldehyde from the oxidation of ethanol and O₂ from the oxidation of hydroxyl ions were detected and quantified. In the cathode-side chamber, H₂ was also simultaneously evolved because the redox potential of H₂ evolution (U⁰(H⁺/H₂)) is comparable to that of NH₃ evolution (U⁰(N₂/NH₃)).

Figure 7(a) shows a histogram of the action spectrum of the apparent quantum efficiency of NH₃ evolution for several wavelength regions. In the 550–800 nm region, the apparent quantum efficiency of NH₃ has a value approximately corresponding to the LSPR band, which is indicated by a solid line in Fig. 7(a), clearly indicating that plasmon-based NH₃ synthesis due to N₂ fixation occurred. By contrast, in the 410–550 nm region, the quantum efficiency was higher than the spectrum for Au-Nis. This indicates that NH₃ was evolved not only by LSPR excitation but also by direct excitation of the interband transition of gold. An energy diagram of the plasmon-based NH₃ synthesis system is shown in Fig. 7(b). The plasmonically enhanced optical near-field excites an electron in gold, which can be transferred to the conduction band of SrTiO₃ and electrochemically reduce N₂ at the surface of the Ru nanoparticles loaded in the SrTiO₃, resulting in the evolution of NH₃. The holes might oxidize ethanol or hydroxyl ions.

Fig. 7:

(a) Histogram of the action spectrum of the apparent quantum efficiency of NH₃ formation for several wavelength regions. The solid line indicates the LSPR band. (b) An energy diagram of the plasmon-induced ammonia photosynthesis system using an Au-Nis-loaded SrTiO₃ photoelectrode. The flat band potential of SrTiO₃ on the cathode side was estimated to be -0.2-0.059 × pH V vs. SHE because the SrTiO₃ conduction band potential is –0.2 V vs. SHE at pH 0. The pH for the cathodic surface of SrTiO₃ was less than 7 because a volatile HCl aqueous solution (0.01 mol/dm³) was injected into the cathode chamber.

Conclusion

In this study, we have demonstrated that a single-crystal substrate of TiO₂ or SrTiO₃ loaded with gold nanoparticles having an optical antenna function can be used to develop a photoelectric conversion system or an artificial photosynthesis system responding to visible and near-infrared light. To enhance energy conversion efficiency of the photovoltaic generation or artificial photosynthesis, practical use of the unused solar light energy of the longer wavelength region is indispensable. This study provides a means of accessing this region. The results of this study suggest that localized surface plasmon resonance can contribute not only to the physical process of light excitation of a chemical substance but also to the efficient propagation of chemical reactions, such as plasmon-induced charge separation, creating new opportunities for the field of “plasmonic chemistry” in photochemical research. Plasmonic chemistry will contribute to the construction of efficient solar light energy conversion systems.