Effect of substrate pre-treatment on microstructure and enhanced electrochromic properties of WO₃ nanorod arrays

Abstract

WO₃ nanorod arrays (WNRAs) were successfully synthesized on an FTO substrate pre-coated with a layer of TiO₂ seeds. The morphology and micro-structure were characterized by scanning electron microscopy (SEM), transmission electron microscopy (TEM), high resolution transmission electron microscopy (HRTEM) and X-ray photoelectron spectroscopy (XPS). The effect of substrate pre-treatment on electrochromic (EC) properties of as-prepared WNRAs was investigated in detail. Compared with those synthesized on WO₃ seed layer, the WNRAs grown on TiO₂ seed layer showed a large optical modulation (77.5%, at 660 nm) and high coloration efficiency (142.7 cm² C⁻¹) under the optimal preparation conditions. In addition, the influence of annealing temperature of the TiO₂ seed layer on the EC performance was also discussed. More importantly, the WNRAs grown on the TiO₂ seed layer presented good stability and strong adhesion to the FTO substrate during the test process.

---

1 Introduction

As a versatile material for its polyvalency and diverse crystal forms, tungsten oxide (WO₃) has attracted considerable attention for the applications in electrochromic (EC) devices,^1,2 photoluminescent materials,³ photocatalysts,⁴ and gas sensors.^5,6 By alternately applying a small positive and negative voltage, amorphous and crystalline WO₃ can switch reversibly between colorless and blue color due to its electrochemical double injection/extraction of Li⁺ ions and electrons into/out of the host lattice of WO₃. In recent years, intensive researches have been carried out towards WO₃ thin film based on EC devices in order to improve its coloration efficiency and cyclic stability. It is found that ordered arrangement of WO₃ nanostructures such as well aligned nanowire arrays and nanorod arrays^7–9 vertical to the transparent conductive substrates (FTO/ITO substrates) compared with bulk materials, offer fast channels for cations (H⁺/Li⁺) and electrons inserting into or extracting out of the host materials due to their large surface area, short ion diffusion path length and low ions transport resistance.^8,9 Therefore, many researches have been focused on the controllable preparation of WO₃ nanorod arrays (WNRAs) on FTO/ITO substrates and related EC properties.^8–11 According to our previous works,^12–15 pre-coating the substrates with seeds of the same material as nanocrystals being grown is an effective way to control the morphology, growth density and orientation of the prepared nanorod/nanowire arrays. However, one drawback may limit their further EC properties: the poor contact between the nanostructured film and the substrate which made the ions and electrons transmission process difficult, resulted in a long time response and poor cyclic stability.¹⁶ Among various preparation methods,^17–21 hydrothermal method^22–24 is a suitable approach to fabricate crystal tungsten oxide films for its advantages of low reaction temperature, cost-effective and controllable size and morphology. Zhang et al.⁸ synthesized WO₃ nanowire arrays (WNWAs) film on FTO substrate pre-coated with a WO₃ seed layer by hydrothermal method. The as-prepared WNWAs showed a fast response time (7.6 s for coloring and 4.2 s for bleaching) and large coloration efficiency (CE) (102.8 cm² C⁻¹), but, it slowly dissolved in the electrolyte because of the not strong adhesion to the substrate. Ma et al.⁹ prepared WNRAs on FTO substrate pre-treated with a WO₃ seed layer and the WNRAs showed CE value of 98.3 cm² C⁻¹ and relatively lower transmittance modulation of 62%. Ng et al.¹⁶ synthesized WNRAs on W foil with a WO₃ seed layer by hydrothermal approach, which revealed a small reflectance modulation (less than 30%) and not so good stability probably due to the poor contact between the substrate and the WNRAs. So, improving the adhesion property between the nanostructured film and the substrate would ensure the stability and better performance of the WO₃ film. However, until now, no effort has been taken on this issue, and the effect of substrate pre-treatment on the WNRAs EC properties was rarely investigated.

In this paper, a TiO₂ seed layer was firstly adopted for the hydrothermal growth of WNRAs due to its structural stability and electron transmittal ability. The effect of substrate pre-treatment on EC properties of as-prepared WNRAs was systematically studied. It is demonstrated that pre-coating layer of TiO₂ nanoparticles played an important role in determining the cohesiveness between the thin film and the substrate, thus greatly affected the EC performance of the formed oriented WNRAs.

2 Experimental

2.1 Materials

All the regents used were of analytic grade without further purification. The aqueous solutions were prepared freshly with double distilled and de-ionized water. The FTO conductive glass (10–14 Ω cm⁻²) was used as the substrate and tailored into rectangular shape with dimensions of 1 × 5 cm².

2.2 FTO substrates pretreated with TiO₂ crystal seed layer

The FTO substrates were pre-coated with TiO₂ seed layer by spin-coating of TiO₂ colloid solution and then annealed in air. The TiO₂ colloid solution was prepared according to the previous literature¹⁵ as follows: the mixture of 1.7 mL deionized water, 34 mL ethanol and 0.13 mL hydrochloric acid (37% HCl) were added dropwise into the solution which was prepared by dissolving tetrabutyl and acetylacetone into 34 mL ethanol with the mole ratio of 1 : 1, then stirred for 30 min at room temperature. After that, the stable TiO₂ colloid solution was obtained. The FTO substrates were cleaned by acetone, ethyl alcohol and de-ionized water in sequence each for 10 min in an ultrasonic cleaner. Then, the obtained TiO₂ colloid solution was dropped onto the FTO substrates' conductive sides by spin-coating (KW-4A, made by the Chinese Academy of Science) at the speed of 3000 rpm for 30 s. After that, the as-coated FTO substrates were annealed at 550, 650, 700 °C for 30, 30, 15 min, respectively. The reduction of the annealing duration at 700 °C was to prevent the deformation of the FTO substrate. After these processes, the FTO substrates covered with TiO₂ seed layers were obtained.

As a contrast, the FTO substrate pretreated with WO₃ seed layer was also prepared by spin-coating H₂WO₄ colloid solution and then annealed in air. The preparation of H₂WO₄ colloid solution was described as follows: tungstic acid powder (H₂WO₄, 2.5 g, 0.01 mol) was dissolved into 20 mL hydrogen peroxide solution (30 wt%) to yield a homogeneous and stable colloid. Then, the colloid solution was dropped on the FTO substrate for spin-coating and the substrate was dried and annealed at 500 °C in air. The spin-coating and annealing processes were recycled three times, each for 10, 10, 30 min. After these processes, the FTO substrate covered with WO₃ seed layer was obtained.

2.3 Fabrication of WNRAs on pretreated FTO substrates

The precursor solutions for fabrication of WNRAs were prepared as follows: sodium tungstate dihydrate powder (Na₂WO₄·2H₂O, 8.25 g, 0.025 mol) was dissolved in distilled water (25 mL). Then, the final pH values of the precursor solutions were adjusted by the following two steps. Firstly, the pH value of the solution was acidified to 2.0 with HCl solution (2 mol L⁻¹) by magnetic stirring and a white precipitate was generated. Secondly, the solution was diluted to 250 mL and oxalic acid was added to the mixture to adjust the final pH value to 2.0. After that, a homogeneous precursor solution was formed. After stirring for 30 min, 20 mL of the precursor solutions were transferred into a 50 mL Teflon-lined autoclave with rubidium sulfate (Rb₂SO₄, 0.0025 mol) added. The pre-treated FTO substrates were put into the autoclaves with the TiO₂ or WO₃ seed layers down. Then the autoclaves were sealed and heated at 180 °C for 4 h. After the reaction finished, the autoclaves were cooled down to the room temperature naturally. Subsequently, the as-deposited substrates were rinsed repeatedly with de-ionized water and dried in air for further characterization. The as-prepared WNRAs samples grown on FTO substrates coated with TiO₂ seed layers and annealed at 550, 650, 700 °C were designed as A_Ti550, A_Ti650, A_Ti700, respectively. The WNRAs sample grown on FTO substrates coated with WO₃ seed layer was designed as A_W.

2.4 Characterization and analysis methods

The morphology and structure of the as-prepared WNRAs samples were observed on a field emission scanning electron microscope (FESEM, Zeiss, Supra-55, operated at 10 kV) and a high resolution transmission electron microscope (HRTEM, Philips, Tecnai F20). X-ray diffraction (XRD, Rigaku Dmax-2500 diffractometer using Cu-Kα radiation) and X-ray photoelectron spectroscopy (XPS, AXIS UTLTRADLD, equipped with a single Al Kα anode for photo excitation) were used to identify the crystal structure and composition. Cyclic voltammetry (CV) measurements were performed on an electrochemical workstation (CHI660C, Shanghai Chenhua Instrument, Inc.) using a three-electrode system in a beaker with 80 mL (1 mol L⁻¹ LiClO₄ + propylene carbonate (PC)) electrolyte. The WNRAs, a Pt wire and an Ag/AgCl electrode were used as working, counter and reference electrodes, respectively. The digital photographs of WNRAs were taken at different stages, which were applied at −3.0, +3.0 V (vs. Ag/AgCl) for 100 s.

Chronoamperometry (CA) of WNRAs by applying voltages of ±3.0 V for 100 s and in situ measurement of WNRAs with voltage step between −3.0 V (100 s) and +3.0 V (100 s) were carried out on an electrochemical workstation (CHI660C, Shanghai Chenhua Instrument, Inc.) in a colorimetric cell (1.0 cm² × 4.2 cm) with 3 mL (1 mol L⁻¹ LiClO₄ + PC) electrolyte. The transmittance spectra were measured using a UV-vis spectrophotometer (TU-1901, Beijing Purkinje General Instrument Co., Ltd) in the spectral region between 400 and 800 nm by applying voltages of ±3.0 V for 100 s, using a piece of bare FTO glass as the contrast. The in situ coloration/bleaching switching characteristics were recorded using a UV-vis spectrophotometer with a wavelength of 660 nm. The morphology feature parameters (diameter, length, number density and grain size of the seed layers) of all the WNRAs samples were calculated by artificial statistics from corresponding SEM images.

3 Results and discussion

3.1 Effect of substrate pretreatment by different seed layers on microstructure and EC properties of as-prepared WNRAs films

3.1.1 Effect of substrate pretreatment on microstructure and morphology of as-prepared WNRAs film. It is known that the structure and morphology of the WO₃ thin film played important roles in determining its EC properties. In our previous works, the effect of the WO₃ seed layer on the formation of the WNRAs was discussed in detail.²⁵ With the existence of the seed layer on FTO substrate, the uniform and well aligned WNRAs film was obtained compared with the film grown on bare FTO substrate. However, there was a problem that the film was easy to fall off due to its poor cohesiveness between the seed layer and FTO substrate, leading to not good EC properties. Therefore, in this paper, we substituted the WO₃ seed layer with the TiO₂ seed layer to improve the contact between the film and FTO substrate, and further to investigate the new seed layer's effect on the morphology and EC properties of synthesized WNRAs film. Fig. 1 showed the SEM images of the WNRAs in A_W and A_Ti700, respectively. Vertical growth of the WO₃ nanorods in A_W and A_Ti700 can both be seen from Fig. 1, and the shapes of the nanorods were almost the same. This suggests that a TiO₂ seed layer could also provide many active positions for the nucleation of the WO₃ particles and further lead to the formation of the WNRAs. The feature parameters of the WNRAs films in A_W and A_Ti700 were summarized in Table 1. Under the same hydrothermal conditions, the nanorods in A_Ti700 were longer (330 nm) in axial direction and smaller (55.2 nm) in radial direction compared to those in A_W samples. In addition, smaller grain size of TiO₂ seed (35.0 nm) meant relatively more active positions for the nucleation of WO₃ nanoparticles, which led to larger number density in A_Ti700 (1.0 × 10¹⁰ rods cm⁻²).

Fig. 1 SEM images of WNRAs: (a and b) A_W with low magnification and high magnification; (c and d) A_Ti700 with low magnification and high magnification. Insets in (b and d) cross-section view of SEM images of A_W and A_Ti700.

Table 1 Morphology feature parameters of WNRAs in A_W and A_Ti700

Sample	Average diameter (nm)	Average length (nm)	Density (rods cm^−2)	Grain size of seed layer (nm)
AW	57.9	230	6.9 × 10^9	37.5
ATi700	55.2	330	1.0 × 10^10	35.0

---

Fig. 2 gave the XPS spectra of the as-prepared A_Ti700. From the full spectrum, tungsten and oxide peaks could be identified obviously. The W4f core level spectrum of the A_Ti700 (Fig. 2(b)) exhibited well-resolved spin–orbit split doublet peaks corresponding to the W4f5/2 and W4f7/2, at 37.5 eV and 35.4 eV, respectively. This suggests that the film at bleaching state contained only W⁶⁺, which corresponded well to the result of the XRD in the ESI (Fig. S1†). Fig. 2(c) displayed the O1s core level spectrum which exhibited broad peak with high binding energy sides assigning several kinds of oxygen species. The O1s core level can be fit to yield three Gaussian curves, at binding energy 530.2, 531.8 and 533.1 eV, respectively. The component at binding energy 530.2 eV was assigned to oxygen atom (O²⁻) in the WO₃ lattice that formed a strong bond in the manner W=O. The second peak at 531.8 eV was assigned to a small mount hydroxyl group (OH⁻) on the WO₃ surface. The third components at binding energy 533.1 eV may be the oxygen from the atmospheric water molecule absorbed on the sample surface.²⁶

Fig. 2 XPS spectra of as-prepared A_Ti700 (a) full spectrum, (b) W4f and (c) O1s.

More detailed morphological and structural features of A_Ti700 and A_W were studied by TEM and HRTEM as shown in Fig. 3. Single-crystalline nanorods with varying diameters were illustrated in Fig. 3(a) and (c). From Fig. 3(b), it can be seen that the spacing of the lattice fringes was about 0.39 nm, which was well indexed to the (001) plane of the hexagonal structure WO₃, indicating that the WO₃ nanorod growing along [001] direction. This result agreed well with XRD pattern in Fig. S1.† In addition, it is also found that the lattice fringes of the nanorod were not distinct, suggesting that the crystallinity of the prepared nanorods on TiO₂ seed layer was not so high, which was similar to those in A_W (Fig. 3(d)).¹⁰ This meant that the substitution of the TiO₂ seed layer could not change the crystallinity of the synthesized WO₃ nanorods significantly.

Fig. 3 (a) TEM and (b) HRTEM images of A_Ti700, (c) TEM and (d) HRTEM images of A_W.

The formation mechanism of the as-prepared A_W was investigated in our previous work.²⁷ Like the WO₃ seed layer, the TiO₂ seed layer could provide more active positions than bare FTO substrate for further formation of nanorods, and the seed layers with relatively high crystallinity also helped the well orientation of the nanorods to form array structure. The method of adding addictives into the precursor to control the final morphology during the hydrothermal process is now common. With the different kinds of ions (cations, anions) added, the morphology of the WO₃ films was diverse. Zhang et al.⁸ used (NH₄)₂SO₄ as the addictive and thought that sulfate ions preferentially absorbed on the faces parallel to the c-axis of WO₃ nanocrystal, thus 1D single crystal nanowires were formed. In this paper, rubidium sulfate was used as the addictive and the as-prepared nanorod arrays showed a lower aspect ratio. So, we speculated that the cations could also affect the shaping of the 1D nanostructures. In this case, the Rb⁺ ions were supposed to might preferentially inhibit the growth of the rods on the axial direction, resulting in the low aspect ratio of the nanorods. When the amount of the rubidium sulfate was reduced, the nanorods with smaller diameters were obtained, which agreed well with our assumption.

More importantly, it should be noted that the as-prepared WNRAs film grown on the FTO substrates pre-coated with TiO₂ seed layer showed higher structure stability compared with that on WO₃ seed layer, suggesting that the TiO₂ seed layer has better adhesion ability to the FTO substrates. The main reasons might be ascribed to the follows: firstly, the fluoride doped SnO₂ (the conductive side of FTO substrate) has the same tetragonal rutile structure as rutile structure TiO₂ crystal seeds after annealing at 700 °C, so, the lattice mismatch between FTO substrate and TiO₂ was only 2%,²⁸ much lower than that between FTO substrate and WO₃ seeds layer (about 22%). This small lattice mismatch may promote the epitaxial nucleation of the TiO₂ nanoparticles of the seed layer, ensuring the seed layer had good crystallinity and good contact with the FTO substrate. Secondly, the extra H₂O₂ contained in preparing of WO₃ colloid solution would produce O₂ during the annealing process, so, the escape of O₂ between the FTO substrate and the colloid solution may further impact the contact of the substrate and the seed layer, resulting in the WO₃ seed layer easier falling off the substrate.

3.1.2 Effect of substrate pretreatment on EC properties of as-prepared WNRAs films. Generally, the characteristic parameters of WO₃ for EC properties are the cycling stability, optical modulation, responding times, and coloration efficiency. And the cycling stability is considered as one of the most important factors for EC properties. Generally, the cyclic voltammograms (CVs) is used to elucidate the cycling stability of WO₃. Fig. 4 showed the CV curves of the A_Ti700 between −1.0 and 1.0 V at a scan rate of 50 mV s⁻¹ for 1^st, 500^th and 1000^th cycles. As can be seen from the CVs, these WO₃ films gave a pair of redox peaks which could be ascribed to the intercalation/deintercalation of both Li⁺ ions and electrons into/out of the WO₃ crystal lattice, represented as the following equation:¹

WO₃ (transparent) + xe⁻ + xLi⁺ ↔ Li_xWO₃ (blue) (1)

Fig. 4 Cyclic voltammetry curves of A_Ti700 at 1^st, 500^th and 1000^th cycles in 1 mol L⁻¹ LiClO₄/PC at the scanning speed of 50 mV s⁻¹ between −1.0 V and +1.0 V.

It should be noted that no significant change in the shape of CVs appeared during the periods of 1000 cycles, suggesting good cycling stability of the as-prepared A_Ti700. In addition, the double ions intercalation/deintercalation capacities of the WNRAs film could be calculated by integrating the cathodic/anodic current density, which was equal to the amount of Li⁺ ions and electrons intercalated into/deintercalated out of the WNRAs film to form a blue Li_xWO₃ film/transparent WO₃ film. When the WNRAs film cycled for 1000^th, the total cathodic and anodic charge quantities became stable and were calculated to be about 17.2 and 14.9 mC cm⁻² respectively, implying there were still Li⁺ ions remaining in the A_Ti700 at the bleaching state. In addition, it could be found from the SEM images (Fig. S2†) that the nanorod arrays was unchanged after the CV test, indicating the good structure and cycling stability of A_Ti700.

Fig. 5 illustrated the visible transmittance spectra of A_W and A_Ti700 in colored and bleached states at −3.0 V and +3.0 V (vs. Ag/AgCl) for 100 s, respectively. Comparing with the sample A_W, the A_Ti700 presented larger transmittance at bleached state (81.0%) and lower transmittance at colored state (3.5%) when the wavelength was longer than 600 nm, resulting in a larger optical modulation of 77.5% at the wavelength of 660 nm, which was also larger than those in the former reports.^29,30 The obtained results were in accordance with that of Reyes-Gil et al.³¹ This phenomenon may be ascribed to the following reasons: the structures of the samples A_W and A_Ti700 were composed of two parts including the different seed layers of WO₃/TiO₂ and the similar WNRAs grown on the seed layers. It should be noticed that TiO₂ is also a kind of EC materials. The transmittance modulation of the as-annealed TiO₂ and WO₃ seed layers were also measured under the same experimental conditions, respectively as shown in Fig. S3.† It can be found that the transmittance modulation of the TiO₂ seed layer was very small (about 3%) at the wavelength of 660 nm while that of WO₃ seed layer was 24%. Considering the similar WNRAs grown on the two seed layers, the color's change of A_Ti700 was mostly attributed to the properties of the WNRAs rather than TiO₂ seed layer, while that of A_W was ascribed to the properties of the WNRAs along with WO₃ seed layer, which would lead to lower modulation of A_W than A_Ti700.

Fig. 5 The visible transmittance spectra of A_Ti700 (red line) and A_W (black line) at colored and bleached states measured at −3.0 V and +3.0 V for 100 s, respectively.

Fig. 6 gave the digital photographs of WNRAs film in colored and bleached states. It can be clearly seen that the WNRAs film showed a uniform deep blue color by applying a potential of −3.0 V, while the color of the film was quickly bleached to colorless by applying a potential of +3.0 V, revealing good color reversibility of the film.

Fig. 6 Digital pictures of A_Ti700 at (a) bleached state (measured at +3.0 V for 100 s) and (b) colored state (measured at −3.0 V for 100 s).

Fig. 7(a) presented the in situ coloring/bleaching transmittance response of A_W and A_Ti700 tested at 660 nm. It can be seen that A_Ti700 showed a good reversibility and a bit decrease of the modulation at bleached state after the first recycle. It is speculated that there were still Li⁺ ions stuck in the nanorods in A_Ti700, which was corresponding to the result of the CV test. The switching time was defined as the time required for 90% change in the full transmittance modulation at a certain wavelength (660 nm in this case). As indicated in Fig. 7(b), the switching time of A_Ti700 for coloring and bleaching were 10 s and 28 s, while A_W exhibited 26 s for coloring and 24 s for bleaching, respectively, which were not so fast when compared with some former reports,^9,32 due to the larger size of the WO₃ nanorods. The A_Ti700 showed a much shorter response time than A_W, which may be explained as follows: the larger diameters (57.9 nm) of the nanorods in A_W made the diffusion lengths for Li⁺ in each single nanorod longer than that (55.2 nm) in A_Ti700. With more ions intercalated into the larger nanorods, they would cause a stronger electronic field to resist the following ions' intercalation. In addition, the high crystallographic TiO₂ seed layer and its good contact with the substrate facilitated the fast transfer of electrons from the substrate to the nanorods, resulting in the faster coloring response of A_Ti700. Furthermore, there might be some Li⁺ ions to intercalate the WO₃ seed layer in A_W, delaying the coloring time.

Fig. 7 The switching response characteristics: (a) optical response of A_Ti700 (red solid line) and A_W (black solid line) through chronoamperometry method at +3.0 V and −3.0 V application for 100 s per step at 660 nm; (b) the switching time of A_Ti700 (red solid line) and A_W (black solid line) measured at ±3.0 V for 100 s, respectively.

In addition, it should be noticed that the bleaching time of A_Ti700 was longer than its coloring time, which may be explained by the EC mechanism of the WNRAs as illustrated in Fig. 8. When negative voltage was applied, Li⁺ ions diffused from the electrolyte to the surface of the nanorods. Due to the special lattice structure of h-WO₃, three possible channels³³ (trigonal cavity, hexagonal window and four-coordinated square window) facilitated the ions' transfer and accommodation. For the faster transfer in the substrate and the nanorods, the electrons combined the Li⁺ ions to form the Li_xWO₃ when the ions were just injected into the lattices. When the following Li⁺ ions intercalated into the nanorods, they combined the already existed electrons very quickly, leading to the rapid decrease of the transmittance. The Li⁺ ions would soon reach saturation with more Li⁺ ions in the WO₃ lattices and began to resist the subsequent ions diffusion, so the coloring process slowed down. On the contrary, when the positive voltage was applied, the electrons and Li⁺ ions were driven to deintercalate out of the lattices, back to the substrate and electrolyte. The Li⁺ ions intercalated near the surface were easy to transfer into the electrolyte, resulting in a fast rise in transmittance of the bleaching process. The electrostatic interaction between the electrons/Li⁺ ions and the Li⁺ ions in the electrolyte limited the deintercalation of the inner Li⁺ ions, leading to a longer bleaching time.

Fig. 8 The EC mechanism of the WNRAs: the interface reaction process of Li⁺ ions and electrons between WO₃ and the electrolyte as well as the diffusion process of Li⁺ ions and electrons in the WO₃ crystal lattice of (a) coloring process and (b) bleaching process.

Another important characteristic parameter of the EC properties is coloration efficiency (CE), which is defined as the change in optical density (OD) per unit of charge inserted into (or extracted from) the EC film. It can be calculated from the following equations:^34,35

CE (λ) = ΔOD/Q (2)

ΔOD (λ) = log(T_b/T_c) (3)

where T_c and T_b denote the transmittance of the film at its colored and bleached state, respectively, and the Q is the corresponding injected (or ejected) charge density per unit area. Fig. 9 illustrated the plots of the optical density (ΔOD) varying with the charge density (Q) of A_W and A_Ti700. The CE value was calculated as the slope of the line fitted to the linear region of the curve. A large CE value meant the EC film could exhibit a large optical density modulation with a small charge inserted (or extracted). The CE value of A_Ti700 could be up to 142.7 cm² C⁻¹, while that of A_W was 61.0 cm² C⁻¹, which could be compared with that in the former reports.^7,36 Combining the results shown in Fig. S4, S6† and 7, it can be concluded that, for A_Ti700, a major modulation of transmittance could be completed in a shorter time after voltage switching. However, as for A_W, it took longer time and more charge to accomplish the saturation of OD. So, when applied to eqn (3), the A_Ti700 showed a lager ΔOD value than A_W with smaller charge density. The relatively larger size of the nanorods in A_Ti700 was benefit for the electrolyte's immersing in the film, resulting in more Li⁺ ions having better contact with the nanorods to make the full use of the surface area. Meanwhile, the TiO₂ seed layer also contributed to the larger CE value of A_Ti700. The excellent stability of the TiO₂ itself and the good contact between the seed layer and FTO substrate/nanorods would make the fast and uniform diffusion of the electrons inserting into or extracting out of the nanorods. And the conduction band of WO₃ was higher than that of TiO₂,³⁷ leading to the electrons easier to transfer from TiO₂ to WO₃ for the driving force caused by the energy level difference. Thus, a large optical density modulation could be obtained with a small charge density.

Fig. 9 The variation of the in situ optical density (ΔOD) vs. the charge density for A_Ti700 (red square) and A_W (black dot) measured at 660 nm.

3.2 The effect of annealing temperature of the TiO₂ seed layers on microstructure and EC properties of as-prepared WNRAs films

According to our previous work,¹⁵ with increasing the annealing temperature of seed layer, the grain size and the crystallinity of the seeds increased, which would affect the morphology and size of the nanorod arrays grown on it. Fig. 10 showed the SEM images of the A_Ti550 and A_Ti650, and the corresponding morphology feature parameters were summarized in Table 2. It can be seen that both A_Ti650 and A_Ti550 had vertical grown orientation and similar shapes of rods. With increasing of the annealing temperature of the seed layers, the average diameter and length of as-prepared WNRAs increased from 48.2 to 55.2 nm, and 240 to 330 nm, respectively, which was mainly ascribed to the increasing grain size of the TiO₂ seeds from 15.0 to 35.0 nm (Fig. S5(a)–(c)†).

Fig. 10 SEM images of (a and b) A_Ti550 with low magnification and high magnification; (c and d) A_Ti650 with low magnification and high magnification. Insets in (b and d) cross-sectional view of SEM images of A_Ti550 and A_Ti650.

Table 2 Morphology feature parameters of WNRAs in A_Ti550, A_Ti650 and A_Ti700

Sample	Average diameter (nm)	Average length (nm)	Grain size of seed layer (nm)
ATi550	48.2	240	15.0
ATi650	53.6	275	25.0
ATi700	55.2	330	35.0

---

Fig. 11 illustrated the EC performances of A_Ti550, A_Ti650 and A_Ti700. Considering the same TiO₂ seed layer pre-coated on FTO substrates, the morphology variation of the WNRAs might be the main reason for the different EC properties of A_Ti550, A_Ti650 and A_Ti700. As shown in Fig. 11(a), A_Ti650 exhibited a similar optical modulation (77.4%) with A_Ti700 (77.5%), which was larger than A_Ti550 (72.7%) and A_W (62.8%) at the wavelength of 660 nm, further confirming the effect of the TiO₂ seed layer on improving the EC performance of WNRAs. Compared with the A_Ti700, both the WNRAs in A_Ti550 and A_Ti650 showed good reversibility, stability and faster response time as illustrated in Fig. 11(b) and (c). Specifically, with the annealing temperature increasing from 550 to 700 °C, the bleaching time of the WNRAs increased from 2.4 to 6.2 and 28 s while the coloring time changed a little (7.3, 12, 10 s), which may be mainly due to the increasing lengths and diameters of the WNRAs in A_Ti550 to A_Ti700. It can be also seen in Fig. 11(d) that the ΔOD values changed little in A_Ti550 and A_Ti650 with the charge density lower than 10 mC cm⁻², meanwhile, A_Ti650 had a larger optical modulation (77.4%) and a lower T_c (transmittance at colored state, 9.3%), so, when applied to eqn (3), with smaller values of T_c as the denominators, a series of higher ΔOD values could be gained. Combing the higher ΔOD values and nearly the same charge density values, a higher CE value was obtained in A_Ti650 (89.0 cm² C⁻¹) than A_Ti550 (74.5 cm² C⁻¹). Although A_Ti700 showed a little larger charge density values corresponding to the ordinal ΔOD values, the larger optical modulation and rather small T_c (2.0%) enabled it to obtain a higher CE value (142.7 cm² C⁻¹).

Fig. 11 EC performances of A_Ti550 (green), A_Ti650 (blue) and A_Ti700 (red): (a) the visible transmittance spectra at colored and bleached states measured at ±3.0 V for 100 s, respectively; (b and c) switching time characteristics between colored and bleached states measured at ±3.0 V with a wavelength of 660 nm; (d) coloration efficiencies measured at 660 nm.

4 Conclusions

In summary, WNRAs was synthesized on FTO substrate pre-coated with TiO₂ seed layer by using a hydrothermal method. The introduction of the TiO₂ seed layer did not change the morphology of the as-prepared WNRAs greatly when compared with those grown on WO₃ seed layer. The TiO₂ seed layer had great effect on the EC performance of the WNRAs. Due to the excellent adhesion and contact between the TiO₂ seed layer and the FTO substrate, the A_Ti700 showed larger modulation of transmittance (77.5%) and higher CE value (142.7 cm² C⁻¹). Moreover, with increasing the annealing temperature of the TiO₂ seed layers, all the WNRAs grown on them exhibited larger modulation of transmittance, faster response time and higher CE value. The small changes of the EC properties maybe mainly ascribed to their corresponding different thickness and diameter of the formed WNRAs. The high CE value, large modulation of transmittance and good stability make the WNRAs thin film promising to apply in practical EC devices.

Acknowledgements

The work was financially supported by the National Science Foundation of China (No. 51272025 and 51572020), the National Basic Research Program of China (No. 2014CB643401, 2013AA032003), and China Postdoctoral Science Foundation (No. 2015M581581).

References

1. B. W. Faughnan, R. S. Crandall and P. M. Heyman, RCA Rev., 1975, 36, 177–197.
2. C. Santato, M. Odziemkowski, M. Ulmann and J. Augustynski, J. Am. Chem. Soc., 2001, 123, 10639–10649.
3. M. Feng, A. L. Pan, H. R. Zhang, Z. A. Li, F. Liu, H. W. Liu, D. X. Shi, B. S. Zou and H. J. Gao, Appl. Phys. Lett., 2005, 86, 141901.
4. S. Das and W. M. A. Wan Daud, RSC Adv., 2014, 4, 20856–20893.
5. M. Penza, G. Cassano and F. Tortorella, Sens. Actuators, B, 2001, 81, 115–121.
6. K.-I. Choi, S.-J. Hwang, Z. Dai, Y. Chan Kang and J.-H. Lee, RSC Adv., 2014, 4, 53130–53136.
7. S. Gubbala, J. Thangala and M. K. Sunkara, Sol. Energy Mater. Sol. Cells, 2007, 91, 813–820.
8. J. Zhang, J. P. Tu, X. H. Xia, X. L. Wang and C. D. Gu, J. Mater. Chem., 2011, 21, 5492–5498.
9. D. Y. Ma, G. Y. Shi, H. Z. Wang, Q. H. Zhang and Y. G. Li, J. Mater. Chem. A, 2013, 1, 684–691.
10. F. Zheng, W. K. Man, M. Guo, M. Zhang and Q. Zhen, CrystEngComm, 2015, 17, 5440–5450.
11. G. F. Cai, J. P. Tu, D. Zhou, X. L. Wang and C. D. Gu, Sol. Energy Mater. Sol. Cells, 2014, 124, 103–110.
12. J. H. Qiu, M. Guo, Y. J. Feng and X. D. Wang, Electrochim. Acta, 2011, 56, 5776–5782.
13. T. Ma, M. Guo, M. Zhang and X. D. Wang, J. Nanosci. Nanotechnol., 2009, 9, 5920–5926.
14. Y. X. Li, M. Guo, M. Zhang and X. D. Wang, Mater. Res. Bull., 2009, 44, 1232–1237.
15. M. J. Gao, Y. X. Li, M. Guo, M. Zhang and X. D. Wang, J. Mater. Sci. Technol., 2012, 28, 577–586.
16. C. Y. Ng, K. A. Razak and Z. Lockman, J. Alloys Compd., 2014, 588, 585–591.
17. C. P. Li, F. Lin, R. M. Richards, C. Engtrakul, R. C. Tenent and C. A. Wolden, Sol. Energy Mater. Sol. Cells, 2014, 121, 163–170.
18. Y. Yamada, K. Tabata and T. Yashima, Sol. Energy Mater. Sol. Cells, 2007, 91, 29–37.
19. T. Pauporté, J. Electrochem. Soc., 2002, 149, C539–C545.
20. J. Thangala, S. Vaddiraju, S. Malhotra, V. Chakrapani and M. K. Sunkara, Thin Solid Films, 2009, 517, 3600–3605.
21. E. Khoo, P. S. Lee and J. Ma, J. Eur. Ceram. Soc., 2010, 30, 1139–1144.
22. J. M. Wang, E. Khoo, P. S. Lee and J. Ma, J. Phys. Chem. C, 2008, 112, 14306–14312.
23. D. Y. Ma, H. Z. Wang, Q. H. Zhang and Y. G. Li, J. Mater. Chem., 2012, 22, 16633–16639.
24. S. Adhikari and D. Sarkar, RSC Adv., 2014, 4, 20145–20153.
25. F. Zheng, H. Lu, M. Guo and M. Zhang, CrystEngComm, 2013, 15, 5828–5837.
26. V. V. Kondalkar, S. S. Mali, R. R. Kharade, R. M. Mane, P. S. Patil, C. K. Hong, J. H. Kim, S. Choudhury and P. N. Bhosale, RSC Adv., 2015, 5, 26923–26931.
27. F. Zheng, M. Guo and M. Zhang, CrystEngComm, 2013, 15, 277–284.
28. B. Liu and E. S. Aydil, J. Am. Chem. Soc., 2009, 131, 3985–3990.
29. G. Cai, X. Wang, D. Zhou, J. Zhang, Q. Xiong, C. Gu and J. Tu, RSC Adv., 2013, 3, 6896–6905.
30. G. Cai, J. Tu, D. Zhou, L. Li, J. Zhang, X. Wang and C. Gu, CrystEngComm, 2014, 16, 6866–6872.
31. K. R. Reyes-Gil, Z. D. Stephens, V. Stavila and D. B. Robinson, ACS Appl. Mater. Interfaces, 2015, 7, 2202–2213.
32. D. Zhou, D. Xie, F. Shi, D. H. Wang, X. Ge, X. H. Xia, X. L. Wang, C. D. Gu and J. P. Tu, J. Colloid Interface Sci., 2015, 460, 200–208.
33. S. Balaji, Y. Djaoued, A. S. Albert, R. Z. Ferguson and R. Bruning, Chem. Mater., 2009, 21, 1381–1389.
34. D. Ma, G. Shi, H. Wang, Q. Zhang and Y. Li, J. Mater. Chem. A, 2014, 2, 13541–13549.
35. Y. C. Her and C. C. Chang, CrystEngComm, 2014, 16, 5379–5386.
36. L. Liang, J. Zhang, Y. Zhou, J. Xie, X. Zhang, M. Guan, B. Pan and Y. Xie, Sci. Rep., 2013, 3, 1936.
37. M. Miyauchi, A. Nakajima, T. Watanabe and K. Hashimoto, Chem. Mater., 2002, 14, 4714–4720.

---

Footnote

† Electronic supplementary information (ESI) available: X-ray diffraction pattern, visible spectra of as-prepared seed layers, plots of current density vs. time of the samples, EIS and so on. See DOI: 10.1039/c5ra20970f

---