WO₃ and W₂N nanowire arrays for photoelectrochemical hydrogen production

Abstract

Tungsten oxide (WO₃) nanowire array samples were nitrided in a NH₃ atmosphere to get complete conversion to tungsten nitride (W₂N) nanowires. UV–vis absorption spectroscopy shows that the band gap of WO₃ reduced from 2.9 eV to 2.2 eV after nitridation to W₂N. Photoelectrochemical properties of both WO₃ and W₂N nanowire array electrodes were investigated. WO₃ nanowire arrays show maximum incident photon-to-current conversion efficiency of 85% at 370 nm at 1.2 V vs. SCE. The high quantum efficiency is attributed to the nanowire architecture which ensures efficient light absorption and charge transport. The nanowire arrays were stable even up to 8 h of continuous gas evolution. W₂N nanowire arrays showed good photoactivity even at moderate bias. However, the pure W₂N electrodes were unstable with respect to photocorrosion. The mixed phase W₂N–WO₃ nanowires showed improvement in stability compared to pure W₂N nanowire arrays.

Introduction

Solar photolysis of water is one of the cleanest ways of producing hydrogen. The three main thermodynamic requirements of a semiconductor for spontaneous photolysis are (a) its band gap must be higher than 1.23 eV (b) its band edge positions must straddle the hydrogen and oxygen redox potential, and c) they must be stable against photocorrosion. See review by Bak et al. (2002) [1] for more extensive discussion. Although many metal phosphides and chalcogenides, such as GaP and CdS, have narrow band gaps that are suitable for water splitting, they are unstable against photo-anodic dissolution. Most of the currently used semiconductors that are stable are metal oxides such as TiO₂ and WO₃, but have wide band gaps (>3 eV), and thus only absorb ultraviolet photons. Band gap reduction of metal oxides (MO) with the addition of elements (Y) such C, N and S to form solid solutions (MO_1−xY_x) for enhanced visible light absorption have been reported previously for TiO₂, WO₃, In₂O₃ and ZnO [2], [3], [4], [5]. In TiO₂ alloys, the atomic orbital (typically 2p or 3p) of foreign elements are known to mix with the O 2p orbital to create states within the band gap that give rise to visible light activity. However in such materials, inability to incorporate high concentration of dopant and degradation of crystallinity [6] with incorporation has been reported. It has been theoretically [7] and experimentally [8] shown that the nitride counterparts of metal oxide have lower band gap than the corresponding oxides. For example, the band gap of Ta₂O₅, decreased from 4.0 eV to 2.1 eV after conversion to metal nitrides, Ta₃N₅ [8]. Many transition metal nitrides are thought to be stable against corrosion similar to their oxide counterparts, but not have been tested so far. Herein, we report the investigation of tungsten nitride, W₂N, as a potential candidate for a stable photocatalyst for water decomposition utilizing solar light.

In the recent years, transition metal nitrides such as W₂N have gained considerable interest for other applications because of their unique properties such as high melting point (>2000 K), exceptional hardness (>10 Gpa), chemical inertness in acidic and basic solutions, durability, good electrical conductivity and radiation resistance. They have long been known to be an effective heterogeneous catalysts for a number of reactions such as hydrodesulphurization [9], and NO dissociation [10]. It has been reported that in many of these reactions metal nitride has catalytic activities resembling that of noble metals such as platinum. Recent study by Zhonga et al. (2007) [11] showed that the W₂N is a good electrocatalyst for oxygen reduction reaction. They have also been used as a barrier coating in copper interconnects. Although metal nitrides such as WN_x finds many industrial application there is still a lack of even fundamental electronic and optical properties such as band gap. In this paper, we have studied the structural, optical and photoelectrochemical properties of W₂N and WO₃ nanowire arrays. This is so far the first report of W₂N as a material for photocatalyst.

Apart from the thermodynamic requirements, a suitable photoelectrode must also exhibit the following characteristics: (a). efficient charge transport through the semiconductor; (b). low overpotentials for the redox reactions; (c). low cost; and (d). high light harvesting efficiency. Nanoparticle electrodes have very high interfacial contact area which can make up for the slow kinetics of water oxidation reaction. However, numerous studies have shown that electron transport in these networks occurs by random hoping between particles. In fact, poor charge transport through nanocrystalline material is the one of the main factors in limiting the photocurrent of many semiconductors. Such poor transport combined with the inherently slow water oxidation reactions often results in the high degree of electron recombination either with the defect or trap states or within grain boundaries. To overcome this problem and to maximize both light harvesting and charge collection efficiency, one may use single crystalline 1-D nanowire array electrodes which offer several advantages. Since charge carriers are required to move axially along the length of the nanowire, they provide direct and faster electron transport to the back contact. This is shown schematically in Fig. 1. Array architectures provide longer dimension for light absorption [12]. Recombination at the grain boundary is avoided due to the single crystal nature of the wire. There have been several reports on the use of nanowires in other application such as dye sensitized solar cells [13], [14], [15]. However, there have been very few reports for water splitting reactions [16], [17]. For all the above reasons, we investigated the photoelectrochemical performance of WO₃ and W₂N nanowire array electrodes.