Effect of Ti foil size on the micro sizes of anodic TiO2 nanotube array and photoelectrochemical water splitting performance

Anodic


Introduction
As the firstly and mostly reported semiconductor in sunlight-driven water splitting (WS), TiO 2 has been intensively employed as the powder photocatalyst in photocatalytic WS and as the anode semiconductor in photoelectrochemical (PEC) cells. [1] However, the solar to hydrogen efficiency of TiO 2 is limited by its wide bandgap nature. To improve the TiO 2 catalytic efficiency in WS towards practical use of the process, great effort has been devoted to the approaches such as bandgap narrowing, surface step accelerating, and morphology controlling etc. Among them, TiO 2 morphology has been shown have strong influences on the performances both as powder and anode catalysts. Nanorods, [2] nanowires, [3] nanotubes, [4] and nanofibers [5] of TiO 2 have been successfully prepared and applied either as the only or component of the powder photocatalyst and photoanode in PECWS.
One-dimensional (1D) TiO 2 nanotube (NT) array prepared with anodic oxidation method have been an attractive catalyst for photoanode as it exhibits fast electron transfer, high specific surface area and low charge carrier recombination rate [6][7]. Moreover, TiO 2 NT array intrinsically takes the advantage of the metallic substrate, which is much more flexible and conformable in fabrication and compatible to the other components of the cell than the glass substrates [7]. It is also appealing towards a zero bias PECWS cell in academic research and future application due to an inherent advantage, an onset potential lower than Fe 2 O 3 and its composites as photoanode semiconductors (0.4-0.8 V RHE ) [8][9]. The first report of anodic grown TiO 2 NT application as anode in PECWS cell was illustrated with 1 M KOH electrolyte under UV light source as a function of anodization bath temperature by Mor et al. [10] in 2005, which was stimulated by the milestone works of Assefpour-Dezfuly et al. [11] and Zwilling et al. [12] for the anodic growth of TiO 2 NT arrays. Afterwards, many works [13][14][15][16][17] have examined the controlling of the dimensions of the anodic TiO 2 NT, e.g., the tube length, the inner diameter and the wall thickness etc. and their effect on the performance in PECWS. The reported photocurrent response of the pristine anodic TiO 2 NT, viz. without modifications with loading cocatalysts of metallic nature and forming junctions with other semiconductors, as anode in PECWS cell has been in the range of ca 50 μA/cm 2 to 1.0 mA/cm 2 . A vast difference, ca 50 times, exists due to the inconsistency of the conditions used in different labs for preparation and utilization! For instance, under AM 1.5 G simulated sunlight at 1.6 V RHE , a pristine TiO 2 NT array with an average tube length ~8 μm and diameter of 100 nm gave a 0.1 mA/cm 2 photo response in 1 M KOH solution, [18] while another, with a tube length of around 3.6 μm, showed a photocurrent of 0.52 mA/cm 2 in 1 M NaOH solution. [19] One of the highest photocurrent responses, ca. 0.71 mA/cm 2 in 1 M NaOH solution, reported with pristine TiO 2 NT was obtained with a tube length of 14.1 μm and a diameter of 115 nm on a Ti foil with 1.0 cm 2 active area Recent efforts on enhancing the anode performance of TiO 2 NT array are intensive. Improvement of the pristine TiO 2 NT is the base for achieving high efficient anode, thus the input of effort has been always active and important [20][21][22][23][24][25][26][27]. Nevertheless, the approaches on element doping and cocatalyst loading on the TiO 2 NT to widen the solar energy absorption spectrum and accelerate the surface steps for oxygen evolution reaction are also vital [28][29][30][31][32][33]. Very recently, Lucas et al. [34] coated a SrTiO 3 layer on the anodic TiO 2 NT surface and further doped the layer with La 3+ to form a La:SrTiO 3 /TiO 2 NT junction and obtained a photocurrent response of 0.1442 mA/cm 2 at 1.23 V RHE in a Na 2 SO 3 electrolyte buffered with phosphate at pH = 7.1 under AM 1.5G. A carbon bridged Bi 2 O 3 / TiO 2 NT gave 0.1915 mA/cm 2 output at 0.81 V RHE in a 0.05 M phosphate buffer solution (pH = 7) under visible light [26]. Xiao et al. [27] reported a 0D/1D g-C 3 N 4 and TiO 2 NT heterostructure with showing a 0.72 mA/cm 2 at 1.23 V RHE in a 0.1 M Na 2 SO 4 electrolyte under visible light.
Very recently in 2021, Wang et al. [35] specified the needs to standardize the testing and accreditation of efficiency, procedure and protocol of solar-to-chemical devices. To understand and optimize the photoelectrode preparation parameters and to correlate the effects and responses in different scale level would be important steps towards the commercial application of the PECWS convertors.
Herein, with the in-lab optimized electrolyte and cell configuration in anodic growth of the TiO 2 NT array, we illustrate the effects of the Ti foil size in the TiO 2 NT array growth on the micro-sizes of the nanotubes and on the photocurrent response of the TiO 2 NT anode in PECWS cell. Moreover, the relevance of the sizes in different length scale to the scale up of the anodic TiO 2 NT as anode of the PECWS cell is discussed.

Preparation of samples
Prior to the anodization, the Ti foil was cut into four sizes: 1.0 cm × 1.0 cm (a), 1.0 cm × 2.0 cm (b), 2.0 cm × 2.0 cm (c), and 2.0 cm × 3.0 cm (d). The Ti foils were ultrasonically cleaned with acetone, ethanol and DI water for 30 min sequentially and dried in an oven (UF 75, Memmert GmbH Co. KG) at 70 • C. The anodic oxidation was carried out in a two-electrode configuration with Ti foil as the anode and a Pt sheet (1.00 cm 2 ) as the counter electrode in the growth electrolyte under a constant voltage of 60 V powered with a DC power supply (E3647A, Keysight) for 3 h at room temperature. The electrolyte is a solution of a mixture of 0.3 wt% NH 4 F, DI H 2 O and EG (V DI : V EG = 3:47) [4]. The as formed TiO 2 NT samples were annealed at 450 • C for 30 min with a heating rate of 4 • C/min in a muffle oven (Nabertherm, Germany) in air. The samples are referred to as a, b, c and d corresponding to the four different Ti foil sizes.

Photoelectrochemical measurements
The PEC performance measurements were carried out with the TiO 2 NT sample as the work electrode, a Pt sheet as the counter electrode, and Ag/AgCl as the reference electrode in 1.0 M NaOH (pH = 13.6) as electrolyte under AM 1.5 G simulated sunlight from a Xenon lamp (94011A-ES, LCS-100, Newport). The active areas, i.e., the immersed area in the 1 M NaOH aqueous solution of the sample are listed in Table 1. The photocurrent was measured with an electrochemical workstation (ZENNIUM pro, Zahner-Elektrik, Germany). It should be noted that TiO 2 NT arrays always grow symmetrically on both sides of the Ti foil. However, in the condition range with interest for WS reaction, the photocurrent under dark is always negligible. Therefore, in the context of the photocurrent response description with respect to WS research, the irradiation is from one side and the response of the dark side is neglected.
The measured potential vs. Ag/AgCl electrode was converted to that versus reversible hydrogen electrode (RHE) with the Nernst equation [36]: The linear sweep voltammetry (LSV) curves were recorded with a scan rate of 10 mV/s in 1 M NaOH in both light and dark conditions.
The applied bias photon-to-current efficiency (ABPE) was calculated with the following equation.
where J is the photocurrent density (mA/cm 2 ) at the potential, and P is the incident light intensity of 100 mW/cm 2 .

Photoelectrochemical performance
In Fig. 1  and c achieved 50.6%, 41.3% and 21.3% enhancement, respectively. As compared with literatures, for example: Wang et al. [37] reported a TiO 2 NT with a sample size of 1*1 cm 2 , showing a photocurrent density of less than 0.4 mA/cm 2 at 1.23 V RHE . A nitrogen-doped carbon quantum dots anchored TiO 2 NT gave the highest performance of ca 1.1 mA/cm 2 [37]. Lucas et al. [34] reported a sample with the exposed area of 1.3 cm 2 . The pristine TiO 2 NT produced a photocurrent density of 0.0589 mA/cm 2 at the applied bias of 1.23 V RHE . A pristine TiO 2 NT fabricated on 2*1 cm 2 Ti foil showed a photocurrent density less than 0.5 mA/cm 2 at 1.23 V RHE . [38] Koiki et al. [39] reported a TiO 2 NT with a photocurrent density ca 0.05 mA/cm 2 at 1.23 V RHE , which was fabricated on 5*3 cm 2 Ti foil. The ABPE% of the four samples under light conditions are derived from the J-V curves and are illustrated in Fig. 1 (B).      (Table 1), respectively. With the increase of the size of Ti foil and active area of TiO 2 NT, the crystallite size of the TiO 2 NT is increased.  Fig. 4 (A), the recorded initial current increased from 37.0 mA with sample a to 49.0 mA with sample d, with the increase of the active growth areas. During the whole anodic oxidation TiO 2 NT growth process, the i-t curve for a specific sample gives a horizontal Sshape, but the sequence of the overall current of the four samples remains the same, i.e. d > c > b > a. In Fig. 4 (B) the current densities, i.e., the normalized current per cm 2 are plotted. The sequence of the initial values of the TiO 2 NT samples are reversed. The TiO 2 NT samples from a to d exhibit initial values as 54.8, 31.8, 19.6 and 14.2 mA/cm 2 , respectively, indicating a substantial decrease of the current density along with the increase of the size of the TiO 2 NT growth area of the Ti foil.

Size effects in micro scale
The photocurrent density response at bias 1.23 V RHE is the often discussed and one of the most important output of the photoanode in the PECWS cell. Here, we take the response values at bias 1.23 V RHE for comparison. Fig. 5 (A) plots the six size factors measured in this work as the Y-axis and the photocurrent density response values of the four samples at 1.23 V RHE as the X-axis. For visual aid, the sizes are the numbers read on the Y-axis multiplied by the scale in the parentheses after the term of the factor.
Both the crystallite size and the double wall thickness have a scale of 10 − 8 m level. With the increase of the two sizes in 10 − 8 m scale, the photocurrent density at 1.23 V RHE in PECWS cell is decreased. The double wall thickness has a stronger effect on the output of the anode than the crystallite size. The inner-wall diameter is of sub-micron scale (ca 0.1 μm, viz. 10 − 7 m) and tube length is set in micron-scale (ca 10 μm, viz. 10 − 5 m). With both the increase of inner diameter and tube length, the photocurrent density increased, while increase the tube length (10 − 5 m) seems a little bit more powerful than widening the wall thickness (10 − 7 m).
The effects of semiconductor component sizes in the micro-scale on the solar-to-energy efficiency in PECWS have been intensively investigated [29,40]. Tuning the size of quantum dots [41], the size of plasmonic metal nanodots [42], semiconductor particle size [43], semiconductor layer thickness [44] etc. have been shown effective in monitoring and enhancing PECWS efficiency. Here, the micro sizes of anodic grown TiO 2 NT including the tube length, inner wall diameter and wall thickness all affect the efficiency of the solar to energy device [7,29,[45][46].
Here, we also observed the periodically regular decoration of the nanoparticles and photonic crystals on the outer surface, illustrated as in Fig. 5 (B). These photonics structures in different scale, may contribute enhancement of the photon to electron transformation. These facts indicate that we are in need quantitative models to correlate such observations.

The growth of the nanotubes and the effect of macro size
The micro sizes of TiO 2 NT in anodic growth are known controllable with the reaction temperature, reaction time, the composition of organic solvent, the concentration of F − element and the applied voltage etc. [13,28] The morphology of the anodic grown TiO 2 NT has also been sensitive to the supplier of Ti metal with which the purity, thickness and other parameters are given as equivalent [47]. Here, we demonstrated that tailoring the macro size of Ti foil shows a remarkable effect on the mentioned micro sizes of TiO 2 NT. In turn, the micro-size factors measured all effectively influence the photocurrent response of the TiO 2 NT photoanode. The anodic growth of the TiO 2 NT array is complex and the mechanism and kinetics are still unclear [48]. However, some general outlines on these factors and the macro size effects can be provided. A dense oxide layer with high-resistance to oxidation always exists on the surface of a Ti foil. The growth of anodic TiO 2 NT has three steps. The primary stage: The fluoride (F − ) ions in the electrolyte etch the dense TiO 2 layer with the driving of the potential and pores form. The second stage: The continuous dissolution of Ti metal results in small cavities, which are gradually deepened and the continuous pores are developed to form an orderly independent nanotube structure. The competition between the formation and dissolution of TiO 2 is a crucial factor to determine the morphology of TiO 2 . The final stage: The length of the tube does not increase, when the rate of formation of the oxide and the dissolution rate are equal [49][50].
The macro size of the Ti foil in the TiO 2 NT growth reaction limits the space, and thus exerts a strong influence on the examined micro sizes: crystallite size, tube length, tube diameter and wall thickness. The transient i-t curves in Fig. 4 show clearly that the TiO 2 NT growth rate reflected by the current density becomes larger when the active area of the growth is reduced. The difference in the micro sizes between the samples can be explained with the space competition and the difference in growth rates.
As anodic grown TiO 2 NT arrays are gaining interest as a versatile photoanode catalyst in PECWS and scaling-up efforts of the process are underway [51], increasing the macro size of the photoelectrodes is of a lot of interest. This work clearly shows that the process parameters in anodic TiO 2 NT array growth strongly affect the TiO 2 NT anode efficiency in PECWS cell and a properly prepared sample gives an outstanding performance. The multi-scale modelling of the PEC water splitting electrodes will definitely facilitate the approach to practical use of the process [52].

Photon capture scheme of the anode
We propose the photon capture, charge carrier flow and water splitting mechanism (Fig. 6) to explain the outstanding performance of the anodic grown TiO 2 NT array. When light irradiated on the TiO 2 NT film, the open top of the tubes allows the photon directly running into the inside space of the tube and interacting with the inner wall. When photon excites the electrons (e − ) in the valence band (VB) of TiO 2 to the conduction band, stimulating an enrichment of VB holes (h + ) for O 2 evolution reaction on the tube tops. The excited e − drives to the tube bottom towards the Ti foil substrate and then flows in the circuit to the Pt electrode for the reduction of water to H 2 . The separated e − and h + transfer pathways in the tube wall but in opposite directions reduce the probability of recombination of the charge carries. The excellent conductivities of the TiO 2 NT semiconductor and the metallic Ti foil facilitate the charge flow in high rates.

Conclusion
The effects of the Ti foil macro size on both the anode performance of the TiO 2 NT array in PECWS and the micro sizes of the nanotubes in the array are examined in this work under exactly the same condition, i.e. same DC voltage and electrolyte during the growth. Four Ti foil sizes, with areas of 1, 2, 4 and 6 cm 2 , provided 4 active areas 0.65, 1.08, 2.30 and 3.45 cm 2 of the photoanode in the PECWS cell. With the increase of the Ti foil size, the photocurrent density of the TiO 2 NT anode decreased from 1.13 (sample a) to 0.75 mA/cm 2 (sample d). Compared to that of the sample d, sample a, b and c achieved 50.6%, 41.3% and 21.3% enhancement of the photocurrent just because of the decrease of the TiO 2 NT growth area. The highest ABPE value achieved with sample a was 0.957% at 0.324 V RHE . The characteristic micro sizes: crystallite size, double wall thickness, inner-wall diameter and tube length, measured with SEM and XRD techniques are found all influenced by the Ti foil size and in turn they decide the photocurrent output of the photoanode in PECWS reaction. The current density plotted in the transient i-t curve is a measure of the rate of anodic TiO 2 NT growth. The largest Ti foil (6 cm 2 ) with the largest active area (3.45 cm 2 ) gave the highest current 49.0 mA, but the lowest current density. The normalized transient i-t curves indicated that the sample grown on the smallest active area of 0.65 cm 2 on the 1 cm 2 Ti foil exhibited the highest current density of 54.8 mA/cm 2 , indicating the highest growth rate. Decrease the sizes set in the 10 − 8 m scale, the double tube wall thickness and crystallite size facilitate achieving the enhancement of the photocurrent density at 1.23 V RHE . Increase the sizes in the scale of 10 − 7 ~ 10 − 5 m, i.e. the tube length and inner-wall thickness will achieve PEC performance enhancement. In 10 − 2 m scale, decreasing the TiO 2 NT active area and the Ti foil size will improve the photocurrent density. The multi length scale effects and interactions can be explained with the chemical kinetics and physical forces during the TiO 2 phase nucleation and growth. The proposed mechanism shows that the separated e − and h + transfer pathway in the tube wall but in opposite direction reducing the probability of recombination of the charge carries. The results of this work will be useful to scaling-up of PECWS through helping construction multi length scale quantitative models to explain the multi scale effects both in the preparation and application of the photoelectrodes.

Author contributions
Xuelan Hou proposed the topic and contributed to experiment, analysis of the data and writing of the manuscript. Zheng Li contributed to the experiments. Lijun Fan contributed to SEM characterization, Jiashu Yuan contributed to XRD characterization, and both of them joined the discussion of experimental details. Peter D. Lund proof-read and commented the manuscript. Yongdan Li secured funding, lab facility, and discussed and improved the manuscript.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.