Intrinsic photoelectrocatalytic activity in oriented, photonic TiO₂ nanotubes

Abstract

In this work, we grow oriented photonic TiO₂ nanotubes (TNTs) with a well-established anodization method and their intrinsic catalytic and electronic properties are studied by means of (photo)electrochemical methods. The oriented photonic TNTs are assessed by the R_ctC_dl (ΩF) product, which is a measure of the intrinsic photoelectrocatalytic activity (IPA) of the material. The R_ctC_dl product decouples any surface area effects and it can be applied at any potential and accompanied reaction without knowing the actual electrochemically active surface area (EASA). Moreover, electrochemical impedance spectroscopy (EIS) is used in order to investigate the frequency dispersion in the single frequency Mott-Schottky (MS) measurement. The choice of frequency and the frequency dispersion in MS measurements in nanostructured and of high surface area semiconductors are rarely addressed and can result in erroneous values of two important parameters, i.e. flat band position and donor or acceptor densities. This work provides further insights into important electrochemical aspects and catalytic properties of high surface area nanomaterials for photoelectrochemical applications.

Introduction

Titanium dioxide has been studied extensively as a photocatalyst for the past half century [1]. Despite its low theoretical solar-to-hydrogen (STH) efficiency (< 1.2% for anatase) due to the high bandgap (3.0 eV for rutile and 3.2 eV for anatase), it is still placed among the best and most promising photocatalysts [2], [3]. The main reasons are its abundance, corrosion resistance, chemical and optical stability, low cost and non-toxic nature [4], [5]. Nanostructuring of TiO₂ can offer a series of improvements related to the increased surface area, improved light absorption and reduced recombination of the photogenerated energy carriers [6]. Nanostructures such as nanowires [7], [8], nanorods [9], [10], and nanotubes [11], [12], [13], are particularly attractive due to their 1D configuration, light confinement and short carrier diffusion lengths. The TiO₂ nanotubes (TNTs) are particularly interesting and a distinct category of their own, due to the simplicity in preparation and high level of control over their geometrical parameters [14], [15]. The latter permits the optimisation of the optical path length against the charge diffusion length, as well as the adjustment of surface-to-volume ratio, which can reach as high as 2000–3000 cm⁻¹ [16].

On the other hand, charge recombination centres and electron transfer processes are strongly dependent on the electron occupancy of shallow and deep level energy states, which usually scale with the surface area [17], [18]. Therefore, it is important that the comparison of performance between photoelectrocatalysts is not solely made on basis of the surface area, but also electronic and electrochemical properties must be assessed, calculated, and predicted, experimentally and theoretically. The estimation of the EASA of high surface area materials, such as TNTs, can provide a good indication of the intrinsic catalytic activity (after correcting per EASA) of a material for a given reaction. This is not always possible with voltammetric techniques (e.g. charge associated with surface electrochemistry, charge in non-faradic regions) for various reasons, as well as due to ohmic losses and accessibility of electrolyte to the interior of the nanostructured material [19]. These problems can be overcome by determining the capacitance of the electrode, i.e. the EASA, from EIS. Advantage of EIS is that is not limited by ohmic losses or slow kinetics, which in fact can be measured by the technique and in addition, it can distinguish between different capacitances.

Another application of EIS is in the Mott-Schottky (MS) method, which is widely used to obtain insights on the electronic properties of a photoelectrocatalyst. The MS analysis, which is also based on determining the capacitance of the system, can provide information about the Fermi level (E_F) of the electrons of the material. The E_F coincides with the flatband potential, which then determines the position of the conduction and valence bands, when the band gap energy is known. In addition, MS analysis can also estimate the donor densities of an n-type material and relates the capacitance to the flatband potential. In the derivation of the MS equation, a simple Randles equivalent circuit (EQC) is taken into account, and moreover, MS is valid when all donors are ionised and there are no surface states [20], [21]. In their paper, Taveira et al. [17], state that MS analysis is not applicable to systems with low donor densities, as the potential drop in the space charge layer (SCL) should be far greater than $\frac{\mathit{kT}}{e}$, i.e. 25.7 mV at 298 K. It is also assumed that the capacitance in the Helmholtz layer is stable and larger than the SCL capacitance, which is connected to the former in series. Therefore, the SCL capacitance dominates the capacitive response, and is the one probed by the MS measurement. Apart from uncommon cases [22], the majority of literature reports the MS measurements at different single frequencies, spanning over a wide range, failing or not considering to explain the actual choice of frequency [23], [24], [25], [26], [27], [28], [29], [30]. Severe frequency dispersion may be observed due to high porosity and nanostructuring, as well as due to the presence of surface and deep lying states. As a consequence, erroneous E_F and donor densities (for n-type semiconductors) may be reported, especially in the case of high surface nanomaterials [31].

In this work, we grow highly ordered, photonic TNTs and with the help of EIS, we decouple the SCL capacitance over the total capacitance of the interface between the electrode and electrolyte. On the one hand, the frequency dispersion in the single frequency MS is observed, presented and correlated to the SCL capacitance and on the other hand, the total capacitance is used to determine the IPA of the TNTs of different geometrical parameters grown in this study. We compare the IPA of the different TNTs films at the peak photoconversion efficiency potential, and for this purpose, the product of the overall charge transfer resistance with the double layer capacitance (with units of ΩF) is used [19], [32].