Mechanisms Behind Plasmonic Enhancement of Photocurrent in Metal Oxides

Abstract

Surface plasmon resonance (SPR) of metal nanostructures has been proposed as a promising strategy to enhance the photoactivity of metal oxides for photovoltaic and photoelectrochemical applications. While several studies have suggested that metal nanostructures could act as photosensitizers to improve the photocurrent of metal oxides, the mechanism behind is still not well established. Here we briefly review some of the recent research progress and discuss several possible mechanisms underlying the observed plasmonic enhancement of photocurrent in metal-metal oxide systems.

Introduction

Photoelectrical conversion is a promising and environmentally friendly technology to solve faced energy crisis because sunlight delivers a power of 10⁶ TW/year continuously [1,2]. In photoelectrical conversion, semiconductors play an important role in capturing light to generate electron-hole pairs for subsequent electricity or chemical fuel production [3]. For instance, since Fujishima and Honda demonstrated the photocatalytic water splitting using anTiO₂ photoanode in a photoelectrochemical (PEC) cell [4], an enormous amount of work has been carried out in order to achieve photoassisted water splitting and develop sustainable methods for producing chemical fuel [5-10]. Metal oxide (MO) semiconductors afford advantages of low cost, high stability, and suitable oxidation potential for water oxidation. However, the limitation of most MOs is the intrinsically wide bandgap, resulting in poor ability to capture solar energy in the visible and near-infrared (NIR) regions. It is thus highly desired to develop MOs with small bandgaps or strategies for sensitizing MOs with large bandgaps so they can respond effectively to sunlight for enhancing solar energy conversion.

Surface plasmon resonance (SPR) is an unique optical property of noble metal nanoparticles (NPs) such as Au, Ag and Cu, and it arises from collective coherent oscillation of electrons in the NPs induced by light irradiation [11]. The SPR oscillation frequency is highly sensitive to the size and shape of the metal NPs as well as the dielectric constant of the surrounding environment, which often manifests as strong absorption in the visible and NIR regions [12,13]. One interesting question is: can metal NPs with strong SPR in the visible sensitize MO semiconductors for enhancing photoelectrical conversion applications?

To answer this question, a number of studies have been carried out with focus on interface charge transfer [14,15] as well as photoelectron conversion using plasmonic metal-MO nano hetero structures [16-19]. Several studies suggested that some of the hot electrons from SPR excitation can participate in chemical reactions. For example, Au-MO or Ag-MO nanocomposites have been proposed as visible light driven photocatalysts to degrade pollutant [20-23]. Likewise, plasmonic metal NPs have been demonstrated to sensitize MOs to generate photocurrent in photovoltaic (PV) cells [24-26].

Incident-photon-to-current efficiency (IPCE) of optimized Au-TiO₂ and electrolyte based PV as high as 26% under SPR excitation was reported [24]. In addition, a solid-state plasmonic solar cell fabricated using metal NPs, TiO₂ and hole conductor Spiro-OMeTAD showed photocurrent density of ~2.5 mA/cm² as compared to dye sensitized solid-state solar cell (0.5 mA/cm²) under high intensity visible LED illumination [26]. On the other hand, the same effect of photosensitization of MOs using metal NPs based on hot electron injection has also been proposed for application in PECs [27-29]. Lee et al. reported that the plasmonic photoanode made by a dense array of Au nanorods capped with TiO₂ could generate photocurrent and produce H2 under visible light illumination [29]. Fuel production efficiency is up to 20-fold higher in the visible region than in the UV region. In another study, plasmonic Au NPs were found to play dual roles in enhancing the efficiency of CdS-Au-TiO₂ PEC for hydrogen generation [30]. As the incident light shorter than 525 nm, Au NP served as an electron relay to facilitate the charge transfer between CdS and TiO₂. For wavelength from 525 nm - 725 nm, the Au NP acted as a photosensitizer to improve solar-to-hydrogen conversion. Likewise, Au-decorated TiO₂ nanowire (NW) electrodes for PEC water oxidation have been investigated recently [31]. Enhancement in photocurrent was observed in the entire UV-Vis region from 300 to 800 nm by controlling the cover density and shape of the decorating Au nanostructures including both Au NPs and nanorods (NRs). Although the most improved photoactivity of TiO₂ was found in the UV region, which is attributable to effective surface states passivation by Au NPs, the enhancement in visible region of TiO₂ photoelectrodes matched well the SPR absorption of Au NPs and Au NRs. The mixture of Au NPs and NRs on TiO₂ was able to enhance the photocurrent in the entire UV-Vis region from 300 to 800 nm, even though the efficiency of photoelectron conversion efficiency was very low, just ~0.01-0.02% in IPCE measurements.

The suggestion of metal NP-sensitized MOs with high IPCE in PV is at first somewhat surprising since the lifetime of hot electrons in Au and Ag NPs are usually very short, only 1~2 ps [32]. Thus, for photosensitization or hot electron transfer to be effective, the charge transfer rate has to be much faster than 1-2 ps. Indeed, plasmon-induced electron transfer from Au, to TiO₂ was reported to be 240 fs based on transient absorption spectroscopy with an IR probe [14]. Similarly, the half-lifetime of intra-band signal of hot electron from Au to CdS nanorod by SPR excitation was recently demonstrated to be 1.83 ± 0.22 ps [33]. While ultrafast charge transfer seems to be possible, its branching ratio or quantum yield with respect to other hot electron relaxation processes, such as photothermal conversion through electron-phonon coupling, is often not determined unambiguously. We have investigated the charge carrier dynamics in Au NP-decorated Zn OMWs by ultrafast transient pump-probe spectroscopy with the attempt to determine if there is direct evidence for hot electron injection from Au, to ZnO [34]. The observed hot electron lifetime of Au NP on ZnO NWs (3.2±0.5 ps) did not change noticeably from that of Au NPs alone. Moreover, the exciton lifetime in ZnO is essentially the same with and without Au decoration. The results indicate no charge transfer between ZnO NW and Au NPs or weak interaction between them. Based on the ultrafast study of hot electron lifetime in ZnO, we estimate that the quantum yield of hot electron injection is no more than 10%, based on the uncertainty of our measurement. This is consistent with the low photo current observed in our Au-TiO₂ work that also suggests a very low yield of hot electron injection, if injection is indeed responsible for the photocurrent observed in the visible [31]. Our results seem to be in interesting contrast to the much higher IPCE reported by others, as discussed earlier, which suggest that other possible enhancement mechanisms may be involved when high IPCE is observed or the IPCE may be very sensitive to the detailed experimental conditions that can easily differ for different systems studied.

In the hot electron generation mechanism, metal nanostructures serve as photosensitizers, similar to dyes or quantum dots (QDs), that generates high energy or hot electrons with enough energy to overcome the Schottky barrier upon SPR excitation and be injected into the conduction band of MO semiconductors (Figure 1a). Hot electron injection from metal to MOs will likely occur when the metal and MOs are strongly coupled. SPR excited Au nanostructures have been found to generate hot electrons in Au-ZnO photoanode under visible light irradiation based on in-situ X-ray absorption measurements [35]. As mentioned above, based on the very short lifetime of hot electrons, the probability of electron injection should be low if possible at all. Several studies have focused on improving the hot electron injection efficiency of plasmonic metal-MO systems by optimizing the detailed structures of MOs. For instance, three-dimensional (3D) hierarchical MO nanostructures have been designed and demonstrated to exhibit not only high interfaces between metal, MO and electrolyte, but also strong light harvesting in improving the performance of PEC performance [36,37]. The plasmonic metal combined with geometric 3D-MO nanostructures showed high photoactivity in the visible region, with photocurrent density ~0.125 mA/cm2 (0.5 V vs. Ag/AgCl) and ~0.4 mA/cm² (1 V vs. reversible hydrogen electrode or RHE, IPCE = 5.5 % at 550 nm) in dendritic Au-TiO₂nanorods and Au-ZnO branched nanowires, respectively [36,37]. Similarly, the combined 3D MO structures or photonic crystals with plasmonic metal NP-MO systems were claimed to generate hot electrons for enhancing PEC performance [38-40].

Figure 1:  Three possible mechanisms for enhancing photoelectric conversion in metal/MO systems: (a) hot electron generation, (b) interfacial electric field amplification, (c) light scattering effect.