Water splitting catalyzed by titanium dioxide decorated with plasmonic nanoparticles

Semiconductor-catalyzed photoelectrochemical water-splitting

The photocatalytic properties of TiO₂ arise from the photogeneration of electron/hole pairs, also called charge carriers, following light irradiation. When incident light with energy greater than the band gap interacts with a photocatalyst, electrons are injected to the conduction band of the material, while holes are left behind in the valence band (Fig. 2). Photogenerated holes and electrons then diffuse in the valence band (VB) and conduction band (CB) respectively to surface active sites where chemical reactions take place. The generation of hydrogen and oxygen gas by water splitting is driven by the concomitance of two events: (1) molecular oxygen and protons are produced by the oxidation of H₂O and (2) the excited electrons reduce the aforementioned protons to form hydrogen. Therefore, the four-electron process for splitting water into H₂ and O₂ requires a material with a band gap equal to at least 1.2 eV which corresponds to difference in the two redox potential of the couples involved in water splitting. The efficiency of a photocatalyst lies in its ability to absorb light in order to create charge carriers, by excitation of an electron form the VB to the CB [25]. In PEC cells, hydrogen and oxygen evolution take place on two electrodes, usually separate by a membrane as illustrated in Fig. 2 [3].

Fig. 2:

UV-driven water splitting with a semiconductor where oxygen and hydrogen evolution take place on the valence band and conduction band respectively (left) and diagram of the basic principles of water splitting for a photoelectrochemical cell with an n-type semiconductor photoanode where oxygen is evolved and a photocathode (Pt sheet) where hydrogen is evolved (right). Reproduced from Ref. [24] with permission of The Royal Society of Chemistry.

Additionally to the band gap width, the respective potential levels of the semiconductor VB and CB also have their importance when designing a photocatalyst for water splitting: they have to match the potential of the redox couples involved in the photo-driven reaction. Thus the top level of VB is required to be above the redox potential of O₂/H₂O (+1.2 eV VS SHE) and the bottom level of CB has to be more negative than the redox potential of H⁺/H₂ (0 eV). The width of the band gap as well as the potential of the CB and VB are characteristic of each semiconductor hence it has to be chosen adequately when designing photocatalysts. TiO₂ appears as a suitable photocatalyst since its band gap is large enough to split water (3.0 eV vs. 1.2 eV for the formation of H₂ and O₂) and its band structure matches redox potentials of water splitting as shown in Fig. 3 [4]. Metal oxides semiconductors have indeed been found to be highly active for the splitting of water, however they require UV irradiation because of their large band gap and are thus not suitable for visible-light-driven reactions [27]. One avenue in the field of TiO₂ photocatalysis has focused on the reduction of this material to access black TiO₂, a material able to absorb in the visible region [28].

Fig. 3:

Relationship between band structure of semiconductor and redox potentials of water splitting. Reprinted by permission from Macmillan Publishers Ltd: Nature (Ref. [26], copyright 2001.).

Another critical obstacle limiting the efficiency of photocatalytic reactions on semiconductors is the high rate of electron-hole recombination occurring before charge carriers can reach the semiconductor surface and perform the desired chemical transformations. Although efforts have been made to optimize photosensitivity while minimizing charge-charrier recombination, semiconductors usually exhibit limited efficiency. Their functionalization with PNPs as cocatalysts shows great promise for the design of more efficient sunlight driven water splitting catalysts, because it addresses the shortcomings of plain TiO₂.

Surface plasmon resonance in group XI metal nanoparticles

Cu, Ag and Au NPs feature characteristic strong absorption in the visible spectrum arising from SPR [21]. This phenomenon emerges from the collective resonance of the electrons when incident light matches their oscillation frequency against the restoring force of nuclei (Fig. 4) [30]. The SPR leads to a number of unique properties featured by PNPs: (1) As electrons oscillate in resonance, strong absorption and scattering of incoming light takes place. This leads to the intense color of PNPs suspensions [19]. (2) SPR causes an enhanced electromagnetic field at the vicinity of the NPs which is at the root of surface enhanced Raman spectroscopy (SERS) used for detection with molecular sensitivity. (3) Field enhancement also causes local heat generation and this has been used in the development of nanomedicines [31]. (4) Finally SPR allows for the generation of hot electrons at the surface of plasmonic NPs, a property exploited in catalysis [30], [32]. Owing to their morphology-dependant optical properties, PNPs offer a great deal of flexibility when engineering photocatalysts [33].

Fig. 4:

Schematic of plasmon oscillation for a sphere, showing the displacement of the conduction electron charge cloud relative to the nuclei. Reprinted with permission from [29]. Copyright 2003 American Chemical Society.

Upon excitation of PNPs, charge carriers relax within tens to femtoseconds either though scattering (radiative photon re-emission) or through absorption (non-radiative excitation of energetic charge carriers) [30]. Non-radiative pathways are necessary for the utilization of energetic charge carriers in chemical transformation but scattering by radiative photon re-emission is the dominant decay mode and prevent further use of e-h pairs. Solar-to-fuel catalyst efficiencies are currently limited by the kinetic discrepancy between the lifetimes of photogenerated charge carriers (picoseconds-microseconds) and the extended timescale necessary for photocatalytic reaction to occur at the semiconductor surface (milliseconds to seconds).

Plasmonic nanoparticle supported on titanium dioxide as photocatalyst for water splitting

In an effort to address the limitations of TiO₂ towards the photocatalysis of water splitting, researchers have explored the use of PNPs acting as a relay by trapping visible light in order activate TiO₂. PNPs offer a number of advantages as cocatalyst in the context of water splitting [23]. (1) The introduction of the PNPs at the surface of TiO₂ prevents recombination of the e-h pair by trapping them and promoting charge carrier separation. (2) The visible light absorption properties of PNPs allows more complete and efficient use of solar irradiation. (3) Because PNPs act as antennas for incoming light, less semiconductor materials is needed to trap the same amount of energy, thus increasing efficiency. (4) Metal NPs at the surface of TiO₂ constitute sites, which are potentially catalytically active [34]. (5) Finally the presence of metal NPs at the surface of TiO₂ plays a positive role in reducing the reflection losses taking place at the surface of the electrodes.

The PNP@TiO₂ hybrid system allows for the use of visible light to activate TiO₂. The mechanism of this reaction proceeds as follows. The PNP is able to absorb visible light and generate hot electrons [35]. These are subsequently transferred to the CB of TiO₂ (Fig. 5) [34]. Xiong and coworkers [36] determined the direction of charge transfers across the interface using charge kinetics models and evidenced the interplay relationships between the light harvesting material and the cocatalyst. Given the key contribution of cocatalysts to the photoactivity, synergetic engineering of its structure, its surface and the interface with the semiconductor is required to maximize performance. The PNP composition, exposed facets, phase and defects as well as the interface composition, location and facets play key roles in catalyst design. Another advantage of PNP is that, due to the interfacial upward energy band, called Schottky junction, electrons injected to the semiconductor cannot transfer back to the light harvesting metal [37]. Therefore, e-h recombinations and back reactions are prevented by spatial separation of the charge carriers.

Fig. 5:

Visible light-driven water splitting catalyzed by PNP@TiO₂ via plasmon-mediated electron transfer.

Photoelectrochemical studies performed by Wei’s group demonstrated that the addition of plasmonic cocatalysts generates long-lived charge carriers by plasmon-mediated electron transfer and could further enable visible-light-driven photocatalytic reactions [37]. Electrons transferred at the Au/TiO₂ heterojunction displayed an excited-state lifetime 1–2 orders of magnitude longer than the one displayed by electrons photogenerated by UV irradiation on bare TiO₂. The prolonged hot electron dynamic was attributed to the Schottky barrier at the interface which reduces electron-hole recombination [37].

Gold nanostructures on titanium dioxide

Because of the versatility of its synthesis and its chemical stability, Au NPs [38] have been by far the most used metal for this application [23] and have led to a large number of reports. We thus present below seminal works, followed by the most recent developments, over the last 5 years. Pioneer studies by Tian and Tatsuma [22] reported the first use of plasmonic Au NPs as sensitizers of TiO₂ films for plasmon-induced photoelectrochemistry. They later demonstrated that plasmonic Au NPs on TiO₂ can be used for the visible light induced photocatalytic oxidation of methanol and ethanol as well as the reduction of oxygen [39]. This paved the way for the design of a new class of visible light sensitive photocatalysts for chemical transformation. Tian and Tatsuma [40] suggested that upon excitation due to the SPR, charge separation is accomplished by transfer of photo-excited electrons from Au NPs to the TiO₂ conduction band and that such a charge separation was facilitated by the Schottky barrier formed by their interface. Following this early work, supported Au NPs were found to be applicable for various chemical transformations including of the oxidation of formaldehyde, alcohols, and amines to imines, C−C and amine–alkyne–aldehyde couplings, hydroamination of alkynes, oxidative degradation of phenol, oxidative aldehyde–amine condensation to amide, and hydroxylation of benzene, as well as the production of hydrogen or oxygen gas from organic molecules [41], [42], [43].

To effectively utilize the solar spectrum in catalytic water splitting, optical properties of the photocatalysts may be tuned by the careful design of both PNPs and their support. Photoanodes designed by Moskovits and coworkers harvest light through an array of aligned Au nanorods partially capped with TiO₂. In the plasmonic electrochemical cell reported by the Moskovits group in 2012, a cobalt-based oxygen evolution catalyst (Co-OEC) is deposited on exposed regions of PNPs [44]. Au nanorods can be excited by visible light to form hot electrons which are then collected at the Schottky junction and transferred to a platinum counter-electrode, followed by hydrogen evolution at its surface. The holes thus formed on the Au NPs are filled in by electrons formed by oxidation of water catalyzed on the surface of the cobalt oxygen evolution catalyst. Devices with Co-OEC exhibit excellent photocurrents as the oxidation catalyst act as a charge carrier mediator at the interface of the PNPs and the solution, increasing the charge transfers through the device. The role of each component of the photoanodes was subsequently confirmed by studying another Pt/TiO₂/Co-OEC@Au photocatalyst developed by Moskovits et al. [45] (Fig. 6). TiO₂ was found not to participate in the hydrogen production process but only acted as an electron channel; photoexcited electrons transferred from the Au nanorods to the platinum NPs through TiO₂ without transformation. Similarities between the visible region of the device extinction spectrum, dominated by the SPR band of Au PNPs, and the photoelectrochemical action spectrum were also observed, endorsing the mechanism initially proposed. Plasmon-mediated hot electrons were created by excitation of Au NPs and then injected in TiO₂ and Pt components for hydrogen evolution while photogenerated holes were extracted by the cobalt-based oxidation catalyst. TiO₂@Au nanodumbbells were reported in 2016 by the same group as an alternative to TiO₂ semi-coated Au nanorods with Co-OEC [47]. Moskovits’s group controlled the orientation and anisotropic growth of TiO₂ onto Au nanorods using surface-capping agents as soft template and obtained TiO₂-tipped PNPs exhibiting enhanced hydrogen production compared to counterpart core-shell structures. Addition of a co-catalyst on Au@TiO₂ for light-driven water splitting has also been investigated by Tanaka’s group [48]. The Ni-modified Au@TiO₂ catalyst proposed absorbed light via Au NPs, hot electrons were then injected in TiO₂ CB and reduced protons to form hydrogen at the surface of NiOx NPs.

Fig. 6:

Schematics of the Au-NP fabrication procedure via chemical reduction. Reproduced from Ref. [46] with permission from The Royal Society of Chemistry.

Engineering the morphology of TiO₂ supports can also enhance the efficiency of Au-decorated TiO₂ nanostructures. TiO₂ nanotubes arrays decorated with Au NPs of various shapes were reported extensively. Au nanospheres and nanorods deposited by Li and coworkers [49] on the surface of TiO₂ nanowire arrays improved the photocatalytic activity in the UV and visible regions respectively. Therefore by coating TiO₂ wires with a mixture of Au nanospheres and nanorods, increased water splitting photoactivity across the entire UV-visible region was observed. Misawa et al. [46] investigated three-dimensional TiO₂ structures loaded with Au PNPs. Deposited Au NPs on the walls of TiO₂ nanotunnels exhibited enhanced photocurrent generation compared to conventional two-dimensional devices due to the large specific area and high electron mobility of the hollowed TiO₂ phase (Fig. 6).

The Wang group has also designed visible light responsive Au NPs coated TiO₂ nanotubes for catalytic water splitting [50]. Attractive properties of tubular one-dimensional TiO₂ loaded with PNPs for photodriven reactions could be explained by optical confinement effects [51]. Diffusion of the hot electrons being constrained in TiO₂ walls, charge carrier separation efficiency increases and life time of charge carrier is extended. Guo et al. [52] proposed an Au/TiO₂/Au electrode formed by aligned TiO₂ nanosheets with Au NPs deposited on both sides. The device showed improved photocurrent in the visible region compared to core-shell structures due to the SPR coupling between Au NPs. The spatial arrangement and the interparticle spacing of the PNPs allows fine tuning of this coupling and optimize the photogeneration of e-h pairs. Interestingly, the Zheng group observed that Au NPs embedded in TiO₂ outperformed electrodes composed of Au NPs deposited on TiO₂ and designed PEC cells by incorporating layers of PNPs into the matrix of TiO₂ [53]. Simulations showed that the local electric field in the semiconductor was more intense near PNPs for Au-in-TiO₂ compared to Au-on-TiO₂ devices. Therefore, higher light absorption and enhancement of charge carrier generation were obtained with Au NPs embedded in semiconductor matrix, despite the absence of direct contact between PNPs and the liquid phase. Kang and coworkers [54] proposed to use photonic crystals (PC) and slow photon effect to enlarge absorption spectrum of photocatalyst in the visible region by tuning the SPR response. “Slow light” in photonic crystals and its attractiveness for photochemistry has been reported by several groups [55] and has been used in photocatalysis towards the generation of long-lived charge carriers [56]. Slow photons are observed when light interacts with the periodic dielectric structure of TiO₂ PCs and photons propagate at reduced velocity through ordered PCs. Therefore when the SPR band of Au NPs deposited on TiO₂ bi-layer structure, made of nanorod arrays and PCs, overlaps the slow photon effect region light can be harvested by the TiO₂ photonic crystals, and not only the PNPs. By engineering the size of the PCs to match the frequency of the SPR, the number of hot electrons photogenerated and injected in the TiO₂ nanorods CB was steeply increased.

Tuning of the semiconductor energy levels has also been investigated to enhance photocatalytic activity. The Wang group reported that the addition of Cu₂O, a semiconductor featuring a small band gap, enhanced light-harvesting and improved charge carrier separation of Au@TiO₂ electrodes [57]. Cu₂O is known for its appealing photocatalytic properties but suffers from decomposition by photocorrosion [58]. Wang et al. designed a sandwich-like Au@TiO₂/Al₂O₃/Cu₂O electrode (Fig. 7) where Cu₂O is protected by TiO₂ and promotes hydrogen evolution at its surface. The Al₂O₃ layer between Au@TiO₂ and Cu₂O increases the interfacial charge separation and the sensitivity of the device. Other methods such as nitrogen doping [59] have been used to lower the band gap of TiO₂ by creating localized state lower in energy and thus promote charge carrier separation [60].

Fig. 7:

(left) Energy level diagram superimposed on a schematic of an individual unit of the plasmonic solar water splitter, showing the proposed processes occurring in its various parts and in energy space. CB, conduction band; VB, valence band; EF, Fermi energy. Reprinted by permission from Macmillan Publishers Ltd: Nature Nanotechnology (Ref. [45]), copyright 2013. (right) Schematic diagram of excitation and separation of electrons and holes on TiO₂-1 wt% Au@TiO₂/Al₂O₃/Cu₂O heterostructure photoelectrodes in PEC system. Reproduced from Ref. [57] with permission of The Royal Society of Chemistry.

Silver nanostructures on titanium dioxide

AgNPs@TiO₂ were first reported to be active photocatalysts using near UV light by the Awazu group in 2008 [61]. They featured a seven-fold activity increase compared to bare TiO₂ for the degradation of methylene blue. Key to their design, was the protection of the Ag NP within an SiO₂ layer to prevent oxidation. Interestingly authors coined the term “plasmonic photocatalytic” in this report. Linic and coworkers [62] synthesized various shaped Ag nanostructures deposited onto TiO₂ and applied them towards the decomposition of methylene blue using visible light. They observed a sharp increase of activity compared to bare TiO₂ and showed that rational control of shape and size, and thus change in SPR, resulted in a modification of the wavelength at which photoactivation occurs. Ag NPs, both in supported and unsupported forms, have otherwise been used for organic transformations, including alkene epoxidation, and carbonyl hydrogenation [43], [63], [64]. The Linic group applied this principle to the PEC water splitting, using Ag nanocubes onto nitrogen-doped TiO₂. They demonstrated that Ag nanocubes where able to favor the light-induced formation of charge carriers at the nearby semiconductor surface and allowed a 10-fold increase in photocurrent [65]. Interestingly Au NPs in a similar setup showed only small activity enhancement. The mismatch between the Au NP SPR band and the semiconductor absorption was evoked to explain the small catalytic improvement while a large absorption spectrum overlap exits between Ag nanocubes and TiO₂, featuring a higher catalytic activity. Further to these discoveries, researchers explored the morphology of the semiconductor for this reaction and deposited Ag NPs onto TiO₂ nanotubes [66] and TiO₂ nanosheets [67] to afford catalysis enhancement of 3.3 and 8.5, respectively, compared to the corresponding bare TiO₂ materials. Finally the group of Lai designed TiO₂ nanotube arrays, organized in a parallel fashion and vertically on a substrate, using an electrochemical anodization technique [68]. They deposited Ag NPs on these structures using sonication and observed a 15-fold increase in the water splitting activity under visible light (Fig. 8). Using a similar synthetic procedure, Li and coworkers [69] developed photoelectrodes based on Ag quantum dots decorated TiO₂ nanotube arrays. The advantages of the nanotube array for PEC reside in a greatly enhanced specific surface and a faster charge transfer due to the coherent orientation of all the tubes.

Fig. 8:

Synthetic procedure for accessing TiO₂ nanotube vertical arrays, followed by the deposition of Ag NPs. Reproduced from Ref. [68] with permission of The Royal Society of Chemistry.

Copper nanoparticles on titanium dioxide

The development of Cu PNPs for photocatalytic applications has been slower than for the other coinage metals [70]. Indeed Cu NPs readily reacts with oxygen under atmospheric conditions to form an oxide layer preventing access to the active Cu(0) surface, and deeply affecting the optical properties [71]. The group of Van Duyne et al. [71] studied the effect of Cu oxide layer on Cu(0) NPs and demonstrated that it causes shifting and damping of the plasmonic properties. Cu(0) however is of great interest in the context of photocatalytic device manufacturing as it possesses excellent optical properties at the nanoscale, it is an earth-abundant element, and it is already heavily used in electronics. In 2013, the group of Linic et al. [70] reported the first application of plasmonically active Cu(0) NPs in catalysis, for the reaction of propylene epoxidation. Using in situ ultra violet extinction spectroscopy, they were able to prove that, under light irradiation at the NPs SPR wavelength, the oxide layer reduced and the particles were able to perform the selective formation of propylene oxide. Cu(0) NPs on TiO₂ was first reported to performed PEC water splitting in 2015. In this scheme, Cho and coworkers performed the reaction under inert conditions to ensure Cu would not oxidize. Cu(0) NPs of less than 5 nm, dispersed on TiO₂ were affording stable production of H₂ from water splitting, using glycerol as a hole scavenger [72].