Efficient harvesting and storage of solar energy of an all-vanadium solar redox flow battery with a MoS₂@TiO₂ photoelectrode

Abstract

Solar redox flow batteries constitute an emerging technology that provides a smart alternative for the capture and storage of discontinuous solar energy through the photo-generation of the discharged redox species employed in traditional redox flow batteries. Here, we show that a MoS₂-decorated TiO₂ (MoS₂@TiO₂) photoelectrode can successfully harvest light to be stored in a solar redox flow battery using vanadium ions as redox active species in both the catholyte and anolyte, and without the use of any bias. The MoS₂@TiO₂ photoelectrode achieved an average photocurrent density of ∼0.4 mA cm⁻²versus 0.08 mA cm⁻² for bare TiO₂, when tested for the oxidation of V⁴⁺ to V⁵⁺, attributed to a more efficient light harvesting and charge separation for the MoS₂@TiO₂ relative to TiO₂. The designed solar redox flow cell exhibited an optimal overall solar-to-output energy conversion efficiency (SOEE) of ∼4.78%, which outperforms previously reported solar redox flow batteries. This work demonstrates the potential of the MoS₂@TiO₂ photoelectrode to efficiently convert solar energy into chemical energy in a solar redox flow battery, and it also validates the great potential of this technology to increase reliability in renewable energies.

---

Introduction

The direct storage of sunlight in the form of hydrogen via overall water splitting using photoelectrochemical cells (PECs) has been the subject of extensive research in the last few decades.^1–4 The main benefit of this technology is the clean production of hydrogen through the use of solar energy.^5,6 However, the sluggish kinetics of water splitting, particularly the oxygen evolution reaction, the low overall energy conversion efficiency, and the high cost of transportation and storage of hydrogen have somewhat slowed down the take-off of the hydrogen economy.^7,8 Because of this, researchers are increasingly shifting their efforts towards the development of alternative technologies that can offer faster chemical reaction kinetics to store solar energy. Among them, solar redox flow batteries (SRFBs) have attracted recent attention. SRFBs are hybrids between PECs and redox flow batteries (RFBs). They use renewable light energy to charge redox substances and solar energy in the form of chemical energy.^9–16 One example of recent developments in this area is the SRFB designed by Yangen et al. that used I³⁻/I⁻ and Br⁻/Br³⁻ as redox active pairs.¹⁷ The SRFB was driven by a WO₃-decorated BiVO₄ photoanode and provided 1.25% solar-to-output energy conversion efficiency. Yan et al. reported a SRFB consisting of a Li₂WO₄/LiI redox couple and a dye-sensitised TiO₂ photoelectrode, enabling a battery capacity of 0.0195 mA h mL⁻¹ at a discharge density of 0.075 mA cm⁻².¹ Recently, Amirreza et al. built a tandem with a bare hematite photoanode and two dye-sensitised solar cells connected in series;² the conversion efficiency from solar to chemical energy of the AQDS (anthraquinone-2,7-disulfonate)/iodide SRFB using only hematite as the photoanode was about 0.1%.

Due to their high reversibility and fast reaction kinetics, all-vanadium redox flow batteries, including vanadium-based SRFBs, have been widely investigated and developed around the world.^3–6 Hao et al. applied a nitrogen-doped TiO₂ photoanode to a microfluidic all-vanadium photoelectrochemical cell with an average photocurrent density of 0.1 mA cm⁻².⁷ Zi et al. demonstrated an AQDS/V⁴⁺ SRFB able to generate a relatively stable photocurrent of 0.14 mA cm⁻² using TiO₂ nanoparticles supported on fluorine-doped tin oxide (FTO) as a photoanode.⁸ Titanium dioxide (TiO₂) has been extensively investigated as a photocatalyst since the discovery of its unique properties in 1972, including remarkable stability against photocorrosion and wide band gap, large enough to provide sufficient energy to drive a variety of useful reactions.^9–12 Although the stability of TiO₂ is excellent, the lack of visible light absorption along with the rapid recombination of photogenerated electron–hole pairs produced usually results in low quantum efficiency and poor photocatalytic activity.¹³ Multiple strategies to improve the photocatalytic performance of TiO₂ have been explored, including doping with different elements,^7,14,15 decorating with noble metals,^16,17 and engineering heterojunctions with other semiconductors.^18,19 Coupling TiO₂ with other semiconductors to form heterojunctions is an effective way to boost the performance of TiO₂ (ref. 20 and 21) via bending of the photoelectrode band structure, which in turn provides an easier transfer path for the photogenerated electrons and holes and enables a more efficient process.^22,23 In addition, the design of a heterojunction with a semiconductor of smaller band gap provides a pathway to utilise the visible region of the solar spectrum, increasing the overall efficiency.^24,25 Some reported examples of such systems include Cu₂O@TiO₂,²⁶ g-C₃N₄@TiO₂ (ref. 27) and MoS₂@TiO₂.^28,29

Here we present for the first time a SRFB with exceptional solar-to-output energy conversion efficiency (SOEE), consisting of a MoS₂@TiO₂ photoelectrode (photoanode) and carbon felt as the counter electrode (cathode). Specifically, we have studied the redox pairs V⁵⁺/V⁴⁺ and V⁴⁺/V³⁺, whose redox potentials match the band gap of the photoelectrode heterojunction. We synthesised MoS₂@TiO₂ thin films supported on an FTO glass substrate via a hydrothermal route. The deposition of MoS₂ decorated TiO₂ ensured an increased specific surface area and effective light response, which translated into enhanced photon capturing and charge transfer. When sunlight reaches the photoelectrode, the photogenerated holes oxidize V⁴⁺ to V⁵⁺ at the interface between the photoelectrode and anolyte, while the photogenerated electrons reduce V⁴⁺ to V³⁺ at the interface between the carbon felt and catholyte. Thus, the oxidized and reduced forms of V⁴⁺, V⁵⁺ and V³⁺, respectively, retain the chemical energy that can be converted to electricity via the reverse reaction (Fig. 1). This SRFB can be written as TiO₂/MoS₂(s)|V⁴⁺, V⁵⁺‖V⁴⁺, V³⁺|carbon. The developed system provides a ∼0.4 mA cm⁻² photocurrent which is higher than those of similar TiO₂-based all-vanadium systems with acidic electrolytes.^7,8 This is the first time that the MoS₂@TiO₂ photoelectrode has shown good catalytic activity for vanadium redox couples, with an overall SOEE of 4.78%, remarkably higher than those of previously reported SRFB systems (Table S1, ESI†).^30,31

Fig. 1 Schematic representation of the SRFB system studied in this work.

Experimental

Material preparation

Preparation of TiO₂ nanorod arrays. TiO₂ nanorod arrays (NRs) on FTO were prepared using a hydrothermal method reported elsewhere.³² Briefly, 60 mL deionised water was mixed with 30 mL of concentrated hydrochloric acid (36.5% by weight), and the solution was sonicated for 15 min. Then, 0.1, 0.3, or 0.5 mL of titanium(IV) butoxide (purum, ≥97%, gravimetric, Sigma) was added to the solution and sonicated for a further 15 min. The FTO substrates were placed into a Teflon liner (200 mL). The hydrothermal synthesis was conducted at 180 °C for 12 h. The resultant sample was rinsed extensively with deionised water, then dried at 80 °C for 1 h and thermally treated at 500 °C in air for 1 h.

Fabrication of MoS₂ nanoflowers on TiO₂ NRs (MoS₂@TiO₂). MoS₂ nanoflowers were grown on the TiO₂ NRs (0.1/0.3/0.5 mL) by a facile hydrothermal reaction.²⁸ Firstly, 242 mg (0.001 mol) Na₂MoO₄·2H₂O powder (99%, Sigma) and 242 mg (0.002 mol) L-cysteine (99.99%, Sigma) were mixed in 90 mL deionised water under magnetic stirring for 10 min, then transferred to a Teflon-lined autoclave and stirred for an additional 5 min. Subsequently, the TiO₂ NRs previously prepared were placed into the autoclave containing the MoS₂ precursors. The hydrothermal synthesis and deposition of MoS₂ onto the TiO₂ NRs were conducted at 200 °C for 12 h, followed by rinsing with deionized water and drying at 80 °C for 1 h. The resultant samples were labelled MoS₂@TiO₂-0.1, MoS₂@TiO₂-0.3 and MoS₂@TiO₂-0.5.

Characterisation

The morphology and microstructure of the prepared photoelectrodes were characterised using a field-emission scanning electron microscope (FESEM, Philips FEI Quanta 200 FEG) equipped with an energy dispersive X-ray spectrometer (EDS, operating at an accelerating voltage of 30 kV). The crystal phases of samples were determined by X-ray diffraction (XRD) with a Panalytical Xpert Pro diffractometer using a Cu Kα radiation source (1.5418 Å). The optical absorption properties were determined using a UV-vis spectrophotometer (Lambda 950, PerkinElmer) equipped with an integrating sphere (PerkinElmer) within a wavelength range of 350 to 800 nm and a step of 1 nm. The surface chemical states of photocatalysts were analysed by X-ray photoelectron spectroscopy (XPS, Thermo Fisher Nexsa X-ray spectrometer, Al Kα monochromatic X-ray source). ICP-MS analysis was performed using a microwave plasma atomic emission spectrometer (4210 MP-AES).

Electrochemical and photoelectrochemical characterisation

A three-electrode electrochemical cell was employed to study the electrochemical behaviour of the carbon felt electrode against vanadium ion species. The carbon felts (3.18 mm thick, 99.0% Alfa Aesar) were pre-treated in an MTI 1200× tube furnace at 800 °C for 2 h, using a heating rate of 3 °C min⁻¹ and a nitrogen flow rate of 0.5 L min⁻¹.³³ The size of the carbon felt electrodes was 1.0 cm² (0.32 cm³). A KCl saturated calomel Hg₂Cl₂ electrode (SCE) and Pt foil were employed as reference and counter electrodes, respectively. Cyclic voltammetry (CV) experiments were conducted between −1.0 and 1.6 V versus SCE, at a scan rate of 10 mV s⁻¹, and in an electrolyte with a concentration of 0.1 M or 1.0 M V⁴⁺ (vanadium(IV) oxide sulfate hydrate, Sigma-Aldrich, 97%) and 1.0, 2.0 or 3.0 M H₂SO₄ (Sigma, 98%). Electrochemical impedance spectroscopy (EIS) experiments were conducted using a frequency range of 0.1 MHz to 0.1 Hz and a perturbation of 10 mV. All the electrochemical experiments were carried out using a potentiostat (Gamry IFC5000-05520).

Photoelectrochemical studies were conducted in a three-electrode electrochemical cell using TiO₂ or MoS₂@TiO₂ as a working electrode, and an SCE and Pt foil as reference and counter electrodes, respectively. 0.1 M and 1.0 M V⁴⁺ in 3.0 M H₂SO₄ were used as the electrolyte. Linear sweep voltammetry (LSV) measurements were conducted at a scan rate of 10 mV s⁻¹ in the voltage range of −0.5 V to 0.5 V vs. SCE. EIS photoelectrochemical experiments were also conducted at a frequency range of 0.1 MHz to 0.1 Hz and 10 mV perturbation. The illuminated area (back illumination) of the working electrode was 1.0 cm². The photoelectrode was irradiated by simulated solar light (400 W Xe lamp) using an AM 1.5G filter. The power density of the incident light was calibrated to 100 mW cm⁻² using an optical power meter with a silicon photodetector (Newport 818-SL) as a certified reference. The illuminated area of the working electrode was 1.0 cm².

Redox flow battery testing

Charge–discharge experiments were conducted in the voltage range 0 to 1.1 V using a current density of 1.0 mA cm⁻² with 1.0 M V⁴⁺. The electrolyte was pre-discharged to 0 V at a current density of 0.1 mA cm⁻². Carbon felts separated by a Nafion® membrane were used as positive and negative electrodes. The discharge current density after photocharge is 0.4 mA cm⁻². The carbon felts underwent the same thermal treatment described in the Electrochemical characterisation section. The Nafion membrane (115, 0.005 in. thick, DuPont de Nemours & Co) was subjected to the following treatment prior to its use: firstly, it was immersed in 50 mL 5 wt% H₂O₂ at 80 °C for 1 h, followed by treatment for an hour at 80 °C in distilled water, and another hour at 80 °C in 1.0 M H₂SO₄. The membrane was finally treated at 80 °C for 1 h in distilled water. A volume of 20 mL was employed as the initial catholyte and anolyte, respectively. The experiment was carried out using a peristaltic pump with a double head to control the flow of electrolyte on both sides of the redox flow cell to 2.5 mL min⁻¹. The size of carbon felt as the working electrode for redox flow battery testing is 5.0 cm² (1.59 cm³).

Solar flow battery testing

The SRFB consisted of a modified redox flow battery (Fig. S1a, ESI†), with a window in the anolyte side to enable the irradiation of the photoanode with the Xe lamp light source, and a redox flow battery (Fig. S1b, ESI†), coupled with two electrolyte tanks and a peristaltic pump. All electrochemical data were collected using a Gamry IFC5000-05520 potentiostat. The solar flow cell assembly included the MoS₂@TiO₂ photoelectrode and carbon felt, separated by a Nafion® membrane. Both carbon felts and Nafion® membrane were pretreated as explained in the previous section. Photocharging of the SRFB was conducted without any external bias through irradiation of the photoanode using simulated solar light (400 W Xe lamp) using an AM 1.5G filter. As the SRFB photocharges, the V⁴⁺ present oxidises to V⁵⁺ in the anolyte and reduces to V³⁺ in the catholyte. EIS studies of MoS₂@TiO₂-0.5 were conducted in a SRFB configuration, using 1.0 M V⁴⁺ in 3.0 M H₂SO₄, frequency range 0.1 MHz–0.1 Hz and perturbation of 10 mV. The illuminated area of the working electrode in the solar flow cell was 5.0 cm².

Results and discussion

Three TiO₂ thin films were synthesised using different amounts of titanium(IV) butoxide (0.1, 0.3 and 0.5 mL) and deposited onto FTO using a hydrothermal route. The TiO₂ films (Fig. 2) displayed a rod-like morphology typically observed in the literature³⁶ with a crystal structure corresponding to rutile, as determined by X-ray diffraction (Fig. S2, ESI†).

Fig. 2 SEM images of TiO₂ thin films obtained using (a) TiO₂-0.1, (b) TiO₂-0.3, (c) TiO₂-0.5, (d) MoS₂@TiO₂-0.1, (e) MoS₂@TiO₂-0.3 and (f) MoS₂@TiO₂-0.5; (g) EDS mapping of MoS₂@TiO₂-0.5.

The TiO₂ rods became larger with increasing the amount of titanium(IV) butoxide. The TiO₂ rods produced using 0.1 mL of precursor were relatively short (∼1.6 μm) and randomly oriented (Fig. 2a). As the amount of precursor is increased to 0.3 mL, generally the TiO₂ rods thickened and their length increased, by almost 1 μm, to ∼2.5 μm (Fig. 2b). However, in this case, the size, length, and orientation of the rods was still inhomogeneous. As the amount of titanium(IV) butoxide was further increased to 0.5 mL, the TiO₂ rods became much thicker (4.12 μm). Moreover, the orientation and size of the rods tended to be more uniform (Fig. 2c). The film itself also grew into a denser more compacted layer with additional titanium(IV) butoxide.

MoS₂ nanoflowers were grown on the three TiO₂ films prepared. In the case of the MoS₂@TiO₂-0.1 film, only a small amount of MoS₂ was deposited onto the TiO₂ rods, and also in the space between rods (Fig. 2d). The synthesis and growth of MoS₂ on TiO₂ with a tighter surface structure (MoS₂@TiO₂-0.3 and MoS₂@TiO₂-0.5) prevented the MoS₂ nanoflowers from depositing into the space between TiO₂ rods, leading to the formation of more uniform MoS₂ layers (Fig. 2e and f). A higher magnification of the MoS₂ nanoflowers on the upper right corner of Fig. 2f showed that each MoS₂ nanoflower had a diameter of ∼500 nm and exhibited >100 sheets. According to the literature, the higher the number of layers of MoS₂ nanosheets, the higher the photocatalytic performance, as more nanosheets will provide more active sites and area for the reaction.^32,34,35 EDS data (Fig. 2g and S3, ESI†) for the MoS₂@TiO₂-0.5 material supported the SEM data.

The UV-vis absorption spectra of TiO₂ and MoS₂@TiO₂ photoelectrodes in the front illumination mode (Fig. 3a) showed an increase in light absorption within the visible range for the samples containing MoS₂, as expected. The back illumination spectra (Fig. S4, ESI†) displayed similar behaviour with additional features due to interference in the case of MoS₂@TiO₂ photoelectrodes, although some reports have suggested that the additional bands observed may be due to MoS₂.^25–30 XPS was used to analyse the oxidation state and electronic environment of Ti, Mo, S and O. The Ti 2p line XPS spectrum for TiO₂-0.5 (Fig. 3b) confirmed that the surface of TiO₂ mainly consists of Ti⁴⁺ 2p_1/2, Ti⁴⁺ 2p_3/2, Ti–O bond and O–H bond as expected for TiO₂, with binding energies of 465.1 eV, 459.4 eV, 530.6 eV and 532.3 eV, respectively.³² In Fig. 3c and d, the Mo 3d and S 2p line XPS spectra for MoS₂@TiO₂-0.5 are shown, respectively. The Mo 3d line XPS spectrum indicated the existence of two different states of MoS₂: 1T-MoS₂ and 2H-MoS₂, both with Mo⁴⁺ as the dominant oxidation state. Previous research has indicated that the Mo 3d and S 2p XPS signals of the 1T-MoS₂ phase exhibit a higher binding energy (about 1 eV) than the corresponding signals for 2H-MoS₂.³⁶ The 2H and 1T phases exhibited binding energies of 229.4 and 232.2 eV for 3d_1/2 and 3d_3/2, respectively, for 2H-MoS₂ and 228.1 and 231.3 eV for 3d_5/2 and 3d_3/2, respectively, for 1T-MoS₂.³⁷ Two small features at 235.2 eV (Mo 3d_3/2) and 233.6 eV (Mo 3d_5/2) attributed to Mo⁶⁺ from some MoO₃ formed at the surface were also observed.^38,39 A clear peak corresponding to S 2s at 225.5 eV could also be detected. The S 2p line (Fig. 3d) exhibited two doublets corresponding to 2p_1/2 and 2p_3/2 at binding energies of 163.2 eV and 161.4 eV, respectively.⁴⁰ Additionally, peak contributions corresponding to bridging disulfides found in 1T-MoS₂ were observed at 162.3 eV for 2p_1/2 and at 160.9 eV for 2p_3/2.⁴¹ The XPS survey spectra corresponding to TiO₂-0.5 and MoS₂@TiO₂-0.5 and the XPS O 1s line for MoS₂@TiO₂-0.5 are shown in Fig. S5, ESI.†

Fig. 3 UV-vis spectra of TiO₂-0.5 and MoS₂@TiO₂-0.5 using (a) front-illumination mode, XPS spectra of (b) Ti 2p from TiO₂-0.5, and XPS spectra of (c) Mo 3d and (d) S 2p from MoS₂@TiO₂-0.5.

To study the influence of the electrolyte on the performance of the system, a CV experiment using carbon felt as a working electrode with 1.0 M/2.0 M/3.0 M sulfuric acid (without any V species) was conducted (Fig. S6, ESI†). The results showed that the ionic conductivity increased with increasing the concentration of sulfuric acid. CV and EIS of 0.1 M V⁴⁺ at different sulfuric acid concentrations were also conducted in a three-electrode electrochemical cell configuration (Fig. S7, ESI†). The results suggested that a higher concentration of H₂SO₄ facilitates the electrochemical reaction, resulting in more reversible redox processes at the highest concentration, 3.0 M H₂SO₄. The EIS result in Fig. S7c and d† confirmed that a higher concentration of sulfuric acid as an electrolyte can decrease the resistance through increasing the ionic conductivity of the electrolyte. The CV result indicated a redox potential difference between 0.1 M V³⁺/V⁴⁺ and V⁴⁺/V⁵⁺ of ∼0.49 V in 3.0 M H₂SO₄ using thermally treated carbon felt as the working electrode (Fig. 4a). There exists a reduction peak corresponding to V³⁺/V²⁺; however, no peak for oxidation of V²⁺ to V³⁺ was observed. This has been previously reported in the literature, due to the extremely easy oxidation of V²⁺ in air.^42,43 In order to match the photocurrent density, 1.0 mA cm⁻² was selected as the charge and discharge current density to test the reversibility and stability of the V³⁺/V⁴⁺ and V⁴⁺/V⁵⁺ redox pairs. The redox flow cell assembled with the same carbon felts and redox pairs exhibited a good reversibility upon charge and discharge cycling (Fig. 4b) with an average capacity of 1077 mA h L⁻¹ and coulombic efficiency of 99.2% at a current density of 1.0 mA cm⁻² within 25 cycles (Fig. 4c).

Fig. 4 (a) CV in 0.1 M V³⁺/V⁴⁺ and V⁴⁺/V⁵⁺, with a three-electrode electrochemical cell using, in 3.0 M H₂SO₄, carbon felt, an SCE and a Pt coil as the working, reference, and counter electrodes, respectively. (b) Charge/discharge experiments for a redox flow cell using 1.0 M V⁴⁺, 3.0 M H₂SO₄ electrolyte, and a current density of 1.0 mA cm⁻² from 0 V to 1.1 V. (c) Charge–discharge capacity and coulombic efficiency of the RFB.

The redox processes of V⁴⁺/V⁵⁺ and V⁴⁺/V³⁺ redox pairs in the SRFB are summarised in eqn (1)–(3) as follows:

Anolyte:

V⁴⁺ ⇌ V⁵⁺ + e⁻E₀ = 0.87 V vs. SCE (1)

Catholyte:

V³⁺ ⇌ V⁴⁺ + e⁻E₀ = 0.38 V vs. SCE (2)

Overall:

2V⁴⁺ ⇌ V⁵⁺ + V³⁺E₀ = 0.49 V vs. SCE (3)

After testing the performance of the RFB using V³⁺/V⁴⁺ and V⁴⁺/V⁵⁺, the electrochemical behaviour of the TiO₂ and MoS₂@TiO₂ photoanodes against V⁴⁺ was examined. Fig. 5a and b show the LSV measurements in 0.1 M and 1.0 M V⁴⁺, respectively, in 3.0 M H₂SO₄ for TiO₂-0.5 and MoS₂@TiO₂-0.5. The MoS₂@TiO₂ photoelectrode exhibited much higher performance than TiO₂-0.5, being able to achieve a photovoltage of ∼450 mV vs. SCE, about 100 mV higher than that of the TiO₂-0.5 photoelectrode for both V⁴⁺ concentrations. Upon applying a potential bias, there is an expected increase in photocurrent for both photoelectrodes, which is also more noticeable in the case of MoS₂@TiO₂-0.5. In the case of the MoS₂@TiO₂-0.5, the increase rate is faster for the more concentrated V⁴⁺ electrolyte (1.0 M), reaching 0.6 mA cm⁻² compared to ∼0.4 mA cm⁻² for 0.1 M V⁴⁺ at 0.4 V vs. SCE. Interestingly, there was no significant effect on the photocurrent response when increasing the V⁴⁺ concentration for the TiO₂-0.5 photoelectrode when applying a bias. We attribute this to the fact that the TiO₂ photoelectrode is already performing at its maximum capacity, in terms of the holes that can be produced at a given time, so a higher concentration of V⁴⁺ would not change the number of holes that can reach the electrolyte/TiO₂ interface.

Fig. 5 LSV in (a) 0.1 M V⁴⁺ and (b) 1.0 M V⁴⁺ with a three-electrode electrochemical cell using, in 3.0 M H₂SO₄, a SCE and a Pt coil as reference and counter electrodes, respectively. Chopped-chronoamperometry experiment for a SRFB using TiO₂-0.5 or MoS₂@TiO₂-0.5 as a photoanode and thermally treated carbon felt as a cathode in (c) 0.1 M V⁴⁺ and (d) 1.0 M V⁴⁺ with 3.0 M H₂SO₄ at a 2.5 mL min⁻¹ flow rate and with a light source of 100 mW cm⁻² (AM 1.5G).

Chronoamperometric measurements of all the photoelectrodes (TiO₂-0.1, TiO₂-0.3, and TiO₂-0.5 and their MoS₂@TiO₂ counterparts, Fig. S8, ESI†) conducted in a SRFB configuration showed that the performance of the samples was much more affected by the densification level and thickness of the TiO₂ layer than by the presence of MoS₂ in the TiO₂-0.1, TiO₂-0.3, MoS₂@TiO₂-0.1 and MoS₂@TiO₂-0.3 samples, all of which exhibited a similar photocurrent response. This is likely due to a better alignment of the TiO₂ rods for the more compact samples,³² TiO₂-0.5 and MoS₂@TiO₂-0.5. Therefore, a significant enhancement in the photoresponse when adding MoS₂ was only observed for these more compact TiO₂ layers, with the MoS₂@TiO₂-0.5 sample exhibiting a photocurrent density of 0.4 mA cm⁻²versus 0.08 mA cm⁻² for TiO₂-0.5 in 0.1 M V⁴⁺. Chopped-chronoamperometry experiments in 0.1 M (Fig. 5c) and 1.0 M V⁴⁺ (Fig. 5d), in 3.0 M H₂SO₄, also revealed that the MoS₂@TiO₂-0.5 photoanode presented a stable photocurrent density and quick photoresponse in V⁴⁺ flowing electrolyte.

A stability test was carried out for the best performing photoelectrode, MoS₂@TiO₂-0.5 in 1.0 M V⁴⁺, in 3.0 M H₂SO₄ electrolyte, using a SRFB configuration. Fig. 6a shows the photocharging process conducted for five cycles. As can be observed, a full photocharge was achieved in ∼60 min. As the number of charge/discharge cycles increased, the photocurrent density of MoS₂@TiO₂-0.5 decreased. After each full photocharge and discharge, the photocurrent density decreased from ∼0.4 mA cm⁻² to 0.26 mA cm⁻² for the 5th cycle. This is attributed to the low stability of MoS₂ under the harsh acidic working conditions and more studies to protect the MoS₂ are currently being explored to mitigate this effect. The fluctuations observed in the charge profiles are due to the slow photocorrosion process affecting the MoS₂ under the acidic working environment. When the photons inject into the MoS₂@TiO₂ photoanode, the photoanode tries to create a stable equipotential interface and dynamic equilibrium between the MoS₂@TiO₂ photoanode surface and electrolyte. Due to the existence of MoS₂ photocorrosion, the interface must re-establish itself after the photocorrosion of MoS₂, leading to photocharge fluctuations. This phenomenon was also observed for the 0.1 M V⁴⁺ concentration electrolyte (Fig. S9a, ESI†). Fig. 6b shows the discharge experiment after each full photocharge. Compared to the poor discharge performance (25.52 mA h L⁻¹ to 18.11 mA h L⁻¹) when using 0.1 M V⁴⁺ electrolyte (Fig. S9b, ESI†), the much higher capacity (98.93 mA h L⁻¹ to 74.26 mA h L⁻¹) of 1.0 M V⁴⁺ (Fig. 6c) can provide higher discharge potential to overcome the multiple resistances present in the system. The coulombic efficiency for 1.0 M and 0.1 M V⁴⁺ after five cycles is ∼90% (Fig. 7c) and ∼80% (Fig. S9c, ESI†), respectively. A SOC of 9.03% after 80 min of the first photocharging cycle was observed, decreasing to 6.78% after 80 min into the fifth photocharging cycle.

Fig. 6 (a) Cyclic photocharge and (b) discharge of MoS₂@TiO₂-0.5 and (c) the capacity and coulombic efficiency at a vanadium ion concentration of 1.0 M V⁴⁺. The discharge current density is 0.4 mA cm⁻² (100 mW cm⁻² (AM 1.5G) with 3.0 M H₂SO₄ and a 2.5 mL min⁻¹ flow rate).

Fig. 7 EIS of TiO₂-0.5 or MoS₂@TiO₂-0.5 as the working electrode under dark and light illumination (100 mW cm⁻², AM 1.5G) in a three-electrode electrochemical cell using (a and b) 0.1 M V⁴⁺ or (c and d) 1.0 M V⁴⁺ in 3.0 M H₂SO₄ as an electrolyte and a KCl saturated calomel Hg₂Cl₂ electrode (SCE) and a Pt coil as the reference and counter electrodes, respectively. Equivalent circuit model used to simulate the experimental data of (e) TiO₂-0.5 and (f) MoS₂@TiO₂-0.5 in 0.1 M V⁴⁺ and (g) TiO₂-0.5 and (h) MoS₂@TiO₂-0.5 in 1.0 M V⁴⁺.

During the photocharging process, the carriers from the semiconductor are excited to the conduction band and oxidise V⁴⁺ to V⁵⁺. The photoexcitation and electron transfer can be represented as follows:

TiO₂ + hν = e⁻ (TiO₂) + h⁺ (TiO₂) (formation) (step 1)

e⁻ (TiO₂) + h⁺ (TiO₂) = TiO₂ + hν (recombination) (step 2)

e⁻ (TiO₂) + h⁺ (TiO₂) = e⁻ (MoS₂) + h⁺ (MoS₂) (interface) (step 3)

e⁻ (MoS₂) + h⁺ (MoS₂) = MoS₂ + hν (recombination) (step 4)

V⁴⁺ + h⁺ (MoS₂) = V⁵⁺ (production) (step 5)

The average SOEE during photocharge in the SRFB was calculated according to eqn (4):⁴⁴

SOEE (%) = (∫I_dis × dt × ΔE₀)/(P_in × S × t) × 100 (4)

where I_dis is the discharge current density, ΔE₀ is the potential difference of the two redox couples in the reversible state, t is the illumination time (0.5 h or 1.33 h), P_in is the incident solar power (100 mW cm⁻²), and S is the area of the solar flow cell window (5.0 cm²).

As recorded in Table 1, the calculations showed that the SOEE of the SRFB in 1.0 M V⁴⁺ reaches a value of 4.78% in the first cycle and remains at 3.26% in the fifth cycle.

Table 1 Variation of SOEE from the cycling chronoamperometry plot for the MoS₂@TiO₂ under illumination

Cycle	Sample
	0.1 M			1.0 M
	Photocharge capacity (mA h L^−1)	Discharge capacity (mA h L^−1)	SOEE (%)	Photocharge capacity (mA h L^−1)	Discharge capacity (mA h L^−1)	SOEE (%)
1	32.68	25.52	0.17	114.44	98.93	4.78
2	26.23	22.50	0.14	109.09	94.47	4.31
3	23.92	21.00	0.12	96.99	87.79	3.99
4	21.38	18.69	0.11	79.72	76.74	3.49
5	21.28	18.11	0.09	78.29	74.26	3.26

---

EIS studies showed an increase in the impedance of MoS₂@TiO₂-0.5 with photocharging (Fig. S10, ESI†) reaching almost three times the initial value after 2 hours, revealing an increase in resistance as the photocharging process progresses, which was attributed to the low stability of the photoelectrode under strong acid electrolyte. This was confirmed via SEM images acquired after each cycle (Fig. S11, ESI†). The SEM images clearly show degradation of the MoS₂ layer upon cycling and suggest that the increase in resistance observed in the EIS may be linked to this degradation of MoS₂. The photoexcitation and electron transfer may be represented as shown in eqn (S1), ESI.† The amounts of Mo and Ti leached from the photoanode were determined by ICP-MS and the results are shown in Fig. S12, ESI.† The concentration of Mo in the electrolyte increases with the number of cycles, as expected, supporting the results obtained by SEM and EIS. The TiO₂ layer was less affected by the cycling, but slight leaching was also detected for the 1.0 M V⁴⁺ electrolyte.

To understand the influence of the photoanode structure and to use its light absorption ability to effectively photocharge our solar flow cell, EIS was measured under both dark and light conditions for 0.1 and 1.0 M V⁴⁺ (Fig. 7a–d). EIS data for TiO₂ and MoS₂@TiO₂ photoanodes both for low (0.1 M V⁴⁺) and high electrolyte concentration (1.0 M V⁴⁺) showed that the increase in thickness of the TiO₂ leads to a decrease in impedance, an effect that remains after adding the MoS₂ layer. In the low frequency range under dark conditions, the mass transfer dominates the reaction for TiO₂-0.5.

However, under illumination, the upward trend disappears, which means that charge transfer dominates in that case (Fig. 7a and c). For MoS₂@TiO₂-0.5, EIS results demonstrate that the addition of MoS₂ to the TiO₂ layer enhances the charge transfer properties of the photoelectrode (Fig. 7b and d). Under illumination, the process remains dominated by the mass transport effect for the MoS₂@TiO₂-0.5 photoanode at low frequencies, confirming its better performance compared to TiO₂-0.5. The equivalent circuits for TiO₂-0.5 and MoS₂@TiO₂-0.5 in 0.1 M V⁴⁺ (Fig. 7e and f) exhibit slightly different interface components due to the differences in morphology and roughness of both photoelectrodes.⁴⁵ The EIS data taken in a three-electrode electrochemical cell and equivalent circuit for TiO₂/FTO and MoS₂@TiO₂/FTO systems show that at lower V⁴⁺ concentration, the diffusion resistance is also lower, because when the charge transfer dynamics are not very fast, the charge transfer process and the mass transfer process (caused by diffusion) control the overall reaction: electrochemical polarisation and concentration polarisation coexist. When the concentration is low, the charge transfer process is the rate-determining process. At higher concentration of V⁴⁺, mass transport is introduced in the equivalent circuit as the Warburg impedance (W_o) in the equivalent circuit (Fig. 7g and h), in addition to the charge transfer resistance, as there is an increase in diffusion resistance with increasing concentration of V⁴⁺.

The fitted data for the TiO₂-0.5 and MoS₂@TiO₂-0.5 photoelectrodes in 0.1 M V⁴⁺ and 1.0 M V⁴⁺ are consistent with the results obtained from both dark and light experiments. R_s represents the series resistance (resistance of the electrolyte solution and cable of the potentiostat), R1 and C1 represent the charge transfer resistance and capacitance of the electrolyte/TiO₂ interface in TiO₂-0.5 while in MoS₂@TiO₂-0.5 they represent the MoS₂/TiO₂ interface, and R2 and CPE2 represent the charge transfer equivalent structure of the interface between SnO₂ (FTO) and TiO₂. In MoS₂@TiO₂-0.5, the R1 and C1 represent the impedance of MoS₂/TiO₂. Because the MoS₂ nanoflowers cannot form a dense layer to separate the TiO₂ and the electrolyte, R3 and C3 have a collective effect representing the charge transfer resistance and capacitance of the electrolyte/MoS₂ and electrolyte/TiO₂ interfaces. The C3 element in parallel with R1, C1 and R3 is also associated with the same phenomenon. The main values of the circuit elements in the dark and in light are listed in Table 2. The results from the fitting models of TiO₂-0.1/0.3 and MoS₂@TiO₂-0.1/0.3 in 0.1 M V⁴⁺ and 1.0 M V⁴⁺ fit well with the original data in Fig. S13 and S14, ESI.† The equivalent circuit is applicable for all the TiO₂ and MoS₂@TiO₂ combinations.

Table 2 Equivalent circuit main parameters of EIS

Sample		Element
		R3 (Ω)	C3 (F)	R1 (Ω)	C1 (F)	R2 (Ω)	CPE2-T (F)	CPE2-P
0.1 M V^4+	TiO2-0.5 dark	N/A	N/A	1.7	5.0 × 10^−6	70 441	1.9 × 10^−5	0.9
	TiO2-0.5 light	N/A	N/A	32.9	4.3 × 10^−5	297.3	7.5 × 10^−4	0.5
	MoS2@TiO2-0.5 dark	6.6	3.7 × 10^−6	13.4	2.0 × 10^−5	10 160	1.9 × 10^−4	0.8
	MoS2@TiO2-0.5 light	3.1	1.3 × 10^−5	23.9	1.1 × 10^−3	291.3	1.4 × 10^−3	0.9
1.0 M V^4+	TiO2-0.5 dark	N/A	N/A	1921	3.1 × 10^−6	2.6 × 10^5	1.4 × 10^−6	0.9
	TiO2-0.5 light	N/A	N/A	2943	1.4 × 10^−4	1776	4.3 × 10^−6	0.8
	MoS2@TiO2-0.5 dark	100.2	9.3 × 10^−5	53.8	3.4 × 10^−4	377.3	5.4 × 10^−4	0.8
	MoS2@TiO2-0.5 light	40.5	1.1 × 10^−4	19.3	6.2 × 10^−4	30.9	2.7 × 10^−3	1.0

---

Under dark conditions, due to the poor conductivity of the TiO₂, a large charge transfer resistance (R_ct) is generated at the interface between TiO₂ and SnO₂ (TiO₂/SnO₂), hindering electron transport. In 0.1 M V⁴⁺ and 1.0 M V⁴⁺, under illumination, by comparing the sum of the charge transfer resistance of R1 + R2 + R3, it is demonstrated that the introduction of the MoS₂ interface makes it easier for electrons to transfer in the photoanode, which makes the transmission of the photogenerated electrons in the photoelectrode more efficient. In addition, the injection of photons significantly reduces the R_ct of the interface, especially in the TiO₂/SnO₂ interface. Due to the photon injection, many carriers are generated and moved to the interface (MoS₂/TiO₂, TiO₂/SnO₂, electrolyte/TiO₂ and electrolyte/MoS₂). The enrichment of positive and negative charges on both sides of the interface increases the capacitance under illumination. By comparing the capacitance of TiO₂ and MoS₂@TiO₂, it can be concluded that MoS₂@TiO₂ exhibits more carrier accumulation at the interface, which also confirms that the heterojunction of MoS₂@TiO₂ can effectively increase the carrier density in the photoanode.

Conclusions

TiO₂ and MoS₂@TiO₂ thin films supported on FTO were synthesised and tested as photoelectrodes in a solar redox flow battery using vanadium redox active species. Different thicknesses of the TiO₂ layer were produced. The results showed that a larger amount of Ti precursor resulted in a more uniform, thicker, and denser TiO₂ thin film with more aligned TiO₂ rods. This, in turn, led to a more efficient attachment of the MoS₂ nanoflower overlayer which also translated into a higher photocurrent response. The designed MoS₂@TiO₂ photoelectrode exhibited around 5 times higher activity than bare TiO₂ for charging the all-V SRFB tested in this work. This was attributed to a better separation of charge carriers. The SRFB with 1.0 M V⁴⁺ electrolyte exhibited a remarkable performance, with an average capacity of 1077 mA h L⁻¹ and coulombic efficiency of 99.2% at a current density of 1.0 mA cm⁻² within 25 cycles, charged only by solar energy without the need for any bias. The SOEE reached 4.78%. The photoelectrode interfaces were analysed in detail through EIS, and successfully confirmed the enhancement of the photocurrent response with the addition of the MoS₂ layer to the TiO₂ photoelectrode. This indicates that the highly active MoS₂@TiO₂ photoanode is promising for the enhancement of photocurrent density in the system studied. The MoS₂ facilitated charge carrier separation and provided active sites for oxidation of redox species. However, the low stability of the photoanode under strong acidic conditions led to a significant degradation of the photoelectrode after only five cycles, resulting in the leaching of metal to the electrolyte. More research is currently under way into approaches to protect the photoelectrode against degradation.

Author contributions

GT: investigation, formal analysis, methodology, visualisation, writing – original draft and editing. RJ: formal analysis, writing and editing. JB: writing and editing. MT: writing and editing. AJS: conceptualisation, formal analysis, resources, funding acquisition, writing and editing.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors acknowledge Dr Ryan Wang, Jay Yan, Zhangyi Yao, Junrun Feng and Yue Wen at UCL Department of Chemical Engineering for their help with the ICP measurements. GT acknowledges his PhD scholarship funded by the Chinese Scholarship Council. AJS acknowledges her UKRI Future Leaders Fellowship (MR/T041412/1) for funding support.

Notes and references

1. N. Yan, G. Li and X. Gao, J. Mater. Chem. A, 2013, 1, 7012–7015.
2. A. Khataee, J. Azevedo, P. Dias, D. Ivanou, E. Dražević, A. Bentien and A. Mendes, Nano Energy, 2019, 62, 832–843.
3. M. Skyllas-Kazacos, M. Rychcik, R. G. Robins, A. Fane and M. Green, J. Electrochem. Soc., 1986, 133, 1057.
4. M. Skyllas-Kazacos, M. Chakrabarti, S. Hajimolana, F. Mjalli and M. Saleem, J. Electrochem. Soc., 2011, 158, R55.
5. J. Vázquez-Galván, C. Flox, J. Jervis, A. Jorge, P. Shearing and J. Morante, Carbon, 2019, 148, 91–104.
6. N. M. Delgado, R. Monteiro and A. Mendes, Nano Energy, 2021, 106372.
7. H. Feng, X. Jiao, R. Chen, X. Zhu, Q. Liao, D. Ye, B. Zhang and W. Zhang, J. Power Sources, 2019, 419, 162–170.
8. Z. Wei, H. Almakrami, G. Lin, E. Agar and F. Liu, Electrochim. Acta, 2018, 263, 570–575.
9. A. Fujishima, T. N. Rao and D. A. Tryk, J. Photochem. Photobiol., C, 2000, 1, 1–21.
10. H. Cui, W. Zhao, C. Yang, H. Yin, T. Lin, Y. Shan, Y. Xie, H. Gu and F. Huang, J. Mater. Chem. A, 2014, 2, 8612–8616.
11. T. Butburee, Y. Bai, H. Wang, H. Chen, Z. Wang, G. Liu, J. Zou, P. Khemthong, G. Q. M. Lu and L. Wang, Adv. Mater., 2018, 30, e1705666.
12. A. Fujishima and K. Honda, nature, 1972, 238, 37–38.
13. K. Nakata and A. Fujishima, J. Photochem. Photobiol., C, 2012, 13, 169–189.
14. U. Akpan and B. Hameed, Appl. Catal., A, 2010, 375, 1–11.
15. T. Ohno, T. Mitsui and M. Matsumura, Chem. Lett., 2003, 32, 364–365.
16. V. Subramanian, E. E. Wolf and P. V. Kamat, J. Am. Chem. Soc., 2004, 126, 4943–4950.
17. Z. Zheng, B. Huang, X. Qin, X. Zhang, Y. Dai and M.-H. Whangbo, J. Mater. Chem., 2011, 21, 9079–9087.
18. C. Gao, T. Wei, Y. Zhang, X. Song, Y. Huan, H. Liu, M. Zhao, J. Yu and X. Chen, Adv. Mater., 2019, 31, e1806596.
19. M. Humayun, F. Raziq, A. Khan and W. Luo, Green Chem. Lett. Rev., 2018, 11, 86–102.
20. Y. Bessekhouad, D. Robert and J.-V. Weber, Catal. Today, 2005, 101, 315–321.
21. J. Low, J. Yu, M. Jaroniec, S. Wageh and A. A. Al-Ghamdi, Adv. Mater., 2017, 29, 1601694.
22. C. Sotelo-Vazquez, R. Quesada-Cabrera, M. Ling, D. O. Scanlon, A. Kafizas, P. K. Thakur, T. L. Lee, A. Taylor, G. W. Watson and R. G. Palgrave, Adv. Funct. Mater., 2017, 27, 1605413.
23. H. Wang, L. Zhang, Z. Chen, J. Hu, S. Li, Z. Wang, J. Liu and X. Wang, Chem. Soc. Rev., 2014, 43, 5234–5244.
24. J. Li, M. Zhang, X. Li, Q. Li and J. Yang, Appl. Catal., B, 2017, 212, 106–114.
25. N. Yuan, J. Zhang, S. Zhang, G. Chen, S. Meng, Y. Fan, X. Zheng and S. Chen, J. Phys. Chem. C, 2020, 124, 8561–8575.
26. B. Y. Cheng, J. S. Yang, H. W. Cho and J. J. Wu, ACS Appl. Mater. Interfaces, 2016, 8, 20032–20039.
27. M. Alcudia-Ramos, M. Fuentez-Torres, F. Ortiz-Chi, C. Espinosa-González, N. Hernández-Como, D. García-Zaleta, M. Kesarla, J. Torres-Torres, V. Collins-Martínez and S. Godavarthi, Ceram. Int., 2020, 46, 38–45.
28. H. He, J. Lin, W. Fu, X. Wang, H. Wang, Q. Zeng, Q. Gu, Y. Li, C. Yan and B. K. Tay, Adv. Energy Mater., 2016, 6, 1600464.
29. L. Yang, K. Majumdar, H. Liu, Y. Du, H. Wu, M. Hatzistergos, P. Y. Hung, R. Tieckelmann, W. Tsai, C. Hobbs and P. D. Ye, Nano Lett., 2014, 14, 6275–6280.
30. W. Li, H. C. Fu, L. Li, M. Caban-Acevedo, J. H. He and S. Jin, Angew. Chem., Int. Ed. Engl., 2016, 55, 13104–13108.
31. Q. Cheng, W. Fan, Y. He, P. Ma, S. Vanka, S. Fan, Z. Mi and D. Wang, Adv. Mater., 2017, 29, 1700312.
32. Y. Liu, Y. Li, F. Peng, Y. Lin, S. Yang, S. Zhang, H. Wang, Y. Cao and H. Yu, Appl. Catal., B, 2019, 241, 236–245.
33. M. C. Ribadeneyra, L. Grogan, H. Au, P. Schlee, S. Herou, T. Neville, P. L. Cullen, M. D. Kok, O. Hosseinaei and S. Danielsson, Carbon, 2020, 157, 847–856.
34. D. Wang, Y. Xu, F. Sun, Q. Zhang, P. Wang and X. Wang, Appl. Surf. Sci., 2016, 377, 221–227.
35. C. Liu, L. Wang, Y. Tang, S. Luo, Y. Liu, S. Zhang, Y. Zeng and Y. Xu, Appl. Catal., B, 2015, 164, 1–9.
36. X. Geng, W. Sun, W. Wu, B. Chen, A. Al-Hilo, M. Benamara, H. Zhu, F. Watanabe, J. Cui and T. P. Chen, Nat. Commun., 2016, 7, 10672.
37. K. Chang, X. Hai, H. Pang, H. Zhang, L. Shi, G. Liu, H. Liu, G. Zhao, M. Li and J. Ye, Adv. Mater., 2016, 28, 10033–10041.
38. X. Zhou, M. Licklederer and P. Schmuki, Electrochem. Commun., 2016, 73, 33–37.
39. H. Liu, T. Lv, C. Zhu, X. Su and Z. Zhu, J. Mol. Catal. A: Chem., 2015, 396, 136–142.
40. H. Vrubel, D. Merki and X. Hu, Energy Environ. Sci., 2012, 5, 6136–6144.
41. M. Sun, Y. Wang, Y. Fang, S. Sun and Z. Yu, J. Alloys Compd., 2016, 684, 335–341.
42. W. Wang and X. Wang, Electrochim. Acta, 2007, 52, 6755–6762.
43. M. Ulaganathan, A. Jain, V. Aravindan, S. Jayaraman, W. C. Ling, T. M. Lim, M. P. Srinivasan, Q. Yan and S. Madhavi, J. Power Sources, 2015, 274, 846–850.
44. S. Liao, X. Zong, B. Seger, T. Pedersen, T. Yao, C. Ding, J. Shi, J. Chen and C. Li, Nat. Commun., 2016, 7, 11474.
45. H. Fakhr Nabavi and M. Aliofkhazraei, Surf. Coat. Technol., 2019, 375, 266–291.

---

Footnote

† Electronic supplementary information (ESI) available. See https://doi.org/10.1039/d2ta00739h

---