Effective Visible Light Exploitation by Copper Molybdo-tungstate Photoanodes.

The need for stable oxide-based semiconductors with a narrow band gap, able to maximize the exploitation of the visible light portion of the solar spectrum, is a challenging issue for photoelectrocatalytic (PEC) applications. In the present work, CuW1-x Mo x O4 (E g = 2.0 eV for x = 0.5), which exhibits a significantly reduced optical band gap E g compared with isostructural CuWO4 (E g = 2.3 eV), was investigated as a photoactive material for the preparation of photoanodes. CuW0.5Mo0.5O4 electrodes with different thicknesses (80-530 nm), prepared by a simple solution-based process in the form of multilayer films, effectively exhibit visible light photoactivity up to 650 nm (i.e., extended compared with CuWO4 photoanodes prepared by the same way). Furthermore, the systematic investigation on the effects on photoactivity of the CuW0.5Mo0.5O4 layer thickness evidenced that long-wavelength photons can better be exploited by thicker electrodes. PEC measurements in the presence of NaNO2, acting as a suitable hole scavenger ensuring enhanced photocurrent generation compared with that of water oxidation while minimizing dark currents, allowed us to elucidate the role that molybdenum incorporation plays on the charge separation efficiency in the bulk and on the charge injection efficiency at the photoanode surface. The adopted Mo for W substitution increases the visible light photoactivity of copper tungstate toward improved exploitation and storage of visible light into chemical energy via photoelectrocatalysis.

Introduction

Our need of storable forms of energy is continuously growing,[1,2] as well as the attention toward renewable energy sources, in order to prevent crucial fossil-fuel-related issues such as climate change and air pollution. The sun is the primary source of energy for our planet, and the evolution toward a solar energy society, already envisaged long time ago, has now become an inescapable choice.[3] Mimicking the action of bacteria and plants which are able to convert solar energy into nutrients via photosynthesis,[4] effective strategies were foreseen to convert solar light into energy-containing chemicals.[1,4−6] Among them, the photoelectrocatalytic (PEC) conversion of solar light into highly energetic chemical fuels was proposed, in thermodynamically uphill reactions such as hydrogen production from water splitting,[7,8] and intense efforts were made in identifying photoactive materials for the fabrication of efficient photoelectrodes, in particular photoanodes for water oxidation, as this reaction represents the kinetic bottleneck of the overall PEC water splitting process.[9]

Ternary oxides, such as BiVO4, MFe2O4 (M = Cu, Mg, Zn), InTaO4, and CuWO4, have emerged as promising candidate materials for this application[10−17] because of their high stability toward the harsh oxidative conditions required for oxygen evolution,[18] their valence band edge position being lower in energy with respect to the water oxidation potential of 1.23 V versus RHE.[19] Commonly employed BiVO4 and MFe2O4 oxides with a relatively narrow band gap may allow up to 8 and 10% harvesting of the solar radiation, respectively, which is still not suitable for industrial applications.[13,19] However, their efficient use as well as that of other ternary oxides such as InTaO4 and CuWO4 (having 2.6 eV[10] and 2.3 eV[20] band gap, respectively) requires their modification to narrow their band gap, for an effective solar energy conversion. Attempts to dope these materials with elements such as Ni,[21] N,[22] Cr,[22−24] and Mo[25,26] proved to increase their visible light absorption capability. In particular, partial substitution of W6+ with Mo6+ in the CuWO4 structure results in a band gap reduction of ca. 0.3 eV (from 2.3 to 2.0 eV),[25] corresponding to a significant red shift of the absorption onset of the ternary oxide material.[25−27]

Aiming at ascertaining the role that molybdenum for tungsten substitution has in increasing the PEC performance of intrinsically poorly performing CuWO4, in relation to the transport properties of the charge carriers photogenerated in differently thick photoactive layers and to the charge transfer efficiency at the oxide-solution interface, in the present work we investigate the CuW0.5Mo0.5O4 semiconductor oxide obtained by 50% Mo6+ for W6+ substitution, which is quite easy to achieve, the two ions having very similar radii.[28] This material was prepared by a facile solution-based synthetic route and is composed of a single phase. Its thorough PEC characterization allowed us to demonstrate that the attained reduction in band gap energy directly results in a more efficient exploitation of longer-wavelength photons compared with pure CuWO4 photoanodes. The role that molybdenum incorporation has on the efficiencies of photoproduced charge separation in the bulk and charge injection at the electrode/electrolyte interface has been clarified by investigating how the thickness of the photoactive layer in photoanodes affects their wavelength-dependent PEC performance, also in the presence of a suitable hole scavenger.

Experimental Section

Chemicals and Materials

The following chemicals, all purchased from Sigma-Aldrich, were employed as supplied: copper(II) nitrate trihydrate (99%, Cu(NO3)2·3H2O), ammonium metatungstate hydrate (99%, (NH4)6H2W12O40·xH2O), citric acid (99%), boric acid (99%) and ethanol (99%). Molybdenum(VI) oxide bis(2,4-pentanedionate) (99%, C10H14MoO6) was an Alfa Aesar product. Fluorine-doped 2 mm thick tin oxide (FTO) glass was purchased from Pilkington Glass (TEC-7).

Preparation of Photoelectrodes

A 0.5 M solution of CuW0.5Mo0.5O4 was prepared as follows. First, 0.270 g of citric acid, 0.122 g of copper nitrate, 0.062 g of ammonium metatungstate, and 0.082 g of molybdenum oxide bis-pentanedionate were added to 1.0 mL of an ethanol–water 2:1 solution. Complete dissolution of the metal precursors, corresponding to a 1:1 W:Mo molar ratio, was ensured in the chosen solvent composition, and any phase segregation was excluded in the resulting film. The precursors were dissolved by keeping the solution under constant stirring for 45 min at 80 °C. The as-obtained green paste is stable for several weeks. The photoelectrodes were prepared by spin coating it onto a 2.5 × 2.5 cm2 FTO glass at 4000 rpm for 30 s. Prior to deposition, the FTO glass was cleaned by 30 min-long sonication in a soap solution, followed by careful washing, sonication in ethanol for 30 min, and drying in air. The clean glass slices then underwent a 15 min-long UV-cleaner ozone treatment to remove any organic species deposited onto the FTO surface. They were finally soaked in isopropanol for a few seconds right before the spin coating deposition, to increase the FTO affinity for the metal oxide precursor solution and reduce the light scattering of the resulting films in the long-wavelength spectral region.

After deposition, the CuW0.5Mo0.5O4 films were dried at 250 °C for 10 min, followed by annealing at 550 °C for 1 h. Pure CuWO4 films were prepared by a similar procedure without adding the molybdenum precursor to the initial solution. Multilayer films (up to 5 layers), labeled as nL (n = 1–5) CuW0.5Mo0.5O4, were prepared by coating the as-deposited films with the same precursor solution, followed by annealing at 550 °C for 1 h after each deposition step.

Optical, Structural, Morphological, and Photoelectrochemical Tests

UV–visible absorption spectra were recorded in the transmission mode using a Jasco V-670 spectrophotometer. The crystalline phase of the materials was investigated through X-ray powder diffraction (XRPD) analysis using a Philips PW1820 diffractometer, equipped with a Cu-sealed tube that provided Kα radiation at 40 mA and 40 kV. A model LEO 1430 scanning electron microscope operating at a 10 kV accelerating voltage and at 8 mm working distance was used to acquire the top view and cross-sectional images of the films, up to three deposited layers. A Dektak XT Bruker profilometer was employed to measure the thickness of the thicker 4L and 5L CuW0.5Mo0.5O4 films.

Linear sweep voltammetry (LSV) measurements were carried out using a three-electrode cell equipped with two quartz windows. The FTO/CuW0.5Mo0.5O4 film was used as working electrode, an Ag/AgCl (3.0 M NaCl) as reference electrode, and a platinum gauze as counter electrode. The electrical bias was swept at 10 mV s–1 using an Autolab PGSTAT 12, controlled by the NOVA software. The light source was an Oriel, Model 81172 solar simulator equipped with an AM 1.5 G filter. The light intensity, measured by means of a Thorlabs PM200 power meter equipped with a S130VC power head with Si detector, was 100 mW cm–2. During a typical LSV test, 5 consecutive J–V scans were performed with each electrode, preceded by the Fermi level equilibrium under irradiation up to the open circuit potential (OCP). Identical OCP values were recorded prior to the beginning of each scan with each mono- or multilayer electrode and all of them provided stable and reproducible photocurrent from the first to the last scan.

The investigated films were tested as photoanodes under both back (through the FTO side) and front (through the deposited film side) irradiation configuration, in contact with a 0.1 M K3BO3 aqueous solution at pH 9.[29] The buffer borate solution was prepared by adding KOH to aqueous boric acid up to the desired pH. Furthermore, a series of different sacrificial agents, acting as electron donor species, was employed in LSV measurements. Aqueous solutions (0.1 M) of either H2O2, NH3, or NaNO2 were buffered at pH 9 in a K3BO3 solution, while the 0.1 M Na2SO3 aqueous solution did not contain K3BO3 as it is naturally at pH 9. The potential values versus Ag/AgCl were converted into the RHE scale using the following equation: ERHE = EAgCl + 0.059 pH + E°AgCl, with E°AgCl (3.0 M NaCl) = 0.210 V at 25 °C.

Incident photon to current efficiency (IPCE) measurements were carried out at 1.23, 1.5, and 1.7 V versus RHE under irradiation with a 300 W Lot-Oriel Xe lamp equipped with a Lot-Oriel Omni-λ 150 monochromator and a Thorlabs SC10 automatic shutter, in the above-described single-compartment three-electrode cell containing the K3BO3 buffered solution, to which the NaNO2 hole scavenger was eventually added. The IPCE values were calculated using the following equation:where J is the photocurrent density (mA cm–2) and Pλ (mW cm–2) is the power measured at each specific wavelength λ (nm).

Chopped chronoamperometric scans at different wavelengths were recorded in back side configuration at 1.23 V versus RHE within the 300–650 nm wavelength range, with a 10 nm step. A 420 nm filter was employed at λ > 500 nm, to avoid any contribution from the high-order harmonics originated from the monochromator.

Results and Discussion

Characterization of Photoelectrodes

As shown in Figure a, the multilayer CuW0.5Mo0.5O4 electrodes exhibit an absorption increasing with the increasing number of deposited layers, with an absorption tail due to light scattering observed with the thickest films. The gradual color increase can be appreciated also from their pictures shown in the inset of Figure a. The absorption onset of the material is at ca. 650 nm, corresponding to a band gap of ca. 1.9 eV, in agreement with the ca. 2.0 eV estimated band gap reported for a 50% degree of Mo for W substitution.[25] Thus, the absorption onset of pure copper tungstate, having a band gap energy of 2.3 eV,[30] is effectively extended toward the visible light region upon 50% Mo for W substitution.

Figure 1

(a) Absorption spectra of multilayer (1L–5L) CuW0.5Mo0.5O4 photoelectrodes. Inset: picture of the photoelectrodes. (b) X-ray powder diffraction (XRPD) patterns of 1L, 2L, and 3L CuW0.5Mo0.5O4 electrodes, in the 28–40 degrees 2θ range. The asterisks mark the reflections typical of wolframite.

The XRPD patterns relative to the 1L, 2L, and 3L CuW0.5Mo0.5O4 films, reported in Figure b, show the reflections characteristic of pure wolframite,[25,26] which become more intense with increasing material thickness. Thus, 50% Mo for W substitution has no effects on the crystalline phase formation, within the detection limits of the XRPD technique. The presence of Mo is confirmed by the change of lattice constants and the consequent peak shifts observed in the XRPD patterns of CuW0.5Mo0.5O4 in comparison with CuWO4 (see a detailed view of Figure b reported in Figure S1 of the Supporting Information), which are almost identical to those recently reported[26] for a CuW1–MoO4 sample with the same nominal Mo:W molar ratio.

The top view scanning electron microscopy (SEM) images of 1L, 2L, and 3L CuW0.5Mo0.5O4 electrodes (Figure a–c) reveal a structure composed of densely aggregated crystallites, which may facilitate the electrolyte diffusion across the film by increasing the contact area between the electrolyte solution and the photoactive material. In the case of the thickest film, a decrease in the size of the agglomerates can be observed in the top view SEM images acquired with the same electrode after PEC tests (see Figure e,f), resulting from an extended contact between the CuW0.5Mo0.5O4 film and the electrolyte.

Figure 2

Top-view SEM images of (a) 1L, (b) 2L, and (c) 3L CuW0.5Mo0.5O4 electrodes. (d) Absorption at 430 nm of multilayer photoanodes vs their thickness. SEM images recorded after PEC tests of (e) 2L and (f) 3L CuW0.5Mo0.5O4 electrodes. The scalebar is 200 nm.

The thickness values of the multilayer CuW0.5Mo0.5O4 electrodes, evaluated either by means of SEM cross-sectional images (Figure S2) or by profilometry, are reported in Table and indicate that the average increase in film thickness is ca. 110 nm upon each deposited layer. From these values, together with the absorption spectrum of the films, the absorption coefficients of CuW0.5Mo0.5O4 at different wavelengths were calculated according to the Lambert–Beer law, as the slopes of the lines obtained by plotting the absorbance of the film at a selected wavelength against the film thickness of multilayer CuW0.5Mo0.5O4 electrodes (see Figures d and S3). The best fitting was obtained for λ = 430 nm, the corresponding absorption coefficient being α430 nm = (3.41 ± 0.10) × 104 cm–1.

Table 1

Thickness of Multilayer CuW0.5Mo0.5O4 Electrodes Estimated from SEM Cross-Sectional Images and by Profilometry

sample	thickness/nm
1L	80 ± 20
2L	170 ± 10
3L	270 ± 10
4L	410 ± 20
5L	530 ± 30

PEC Performances

During a typical LSV test under simulated solar light irradiation, five consecutive photocurrent density versus applied potential (J–V) scans were performed with each electrode. The J–V curves reported in Figure a,b correspond to the fifth scan performed with each electrode. Each mono- or multilayer electrode provided stable and reproducible photocurrent from the first to the last scan, with no difference in photocurrent, as shown in Figure S4. Thus, high repeatability was obtained with all films, in line with the high photostability of CuWO4-based materials.

Figure 3

Linear sweep voltammetry (LSV) of (a) 1L–5L CuW0.5Mo0.5O4 electrodes and CuWO4 electrode in back configuration and (b) 1L and 3L CuW0.5Mo0.5O4 electrodes in back and front configuration; AM 1.5 G solar simulated irradiation, scan rate of 10 mV s–1. The current in the absence of irradiation is also shown in panel (a) (black dashed line). Incident photon to current efficiency (IPCE) of (c) CuWO4 monolayer and 1L–5L CuW0.5Mo0.5O4 electrodes at 1.23 V vs RHE and (d) the 3L electrode at different applied potentials.

Figure a shows the J–V plots recorded with the multilayer CuW0.5Mo0.5O4 electrodes and a monolayer CuWO4 electrode under back side solar simulated light (i.e., the electrodes were irradiated through the FTO glass), in contact with a 0.1 M K3BO3 solution at pH 9. First of all, the introduction of Mo into CuWO4 produces a more than 3-fold improvement in photocurrent density at 1.7 V versus RHE. Moreover, no significant difference in the LSV curves of CuW0.5Mo0.5O4 films could be noted up to an applied potential of ca. 1.3 V versus RHE, while progressively higher photocurrent density values were recorded at higher applied potentials with increasing number of deposited layers up to the 3L CuW0.5Mo0.5O4 electrode, which was found to be the most active one.

Consequently, the highest photocurrent density of ca. 1 mA cm–2 was recorded at 1.7 V versus RHE for the 3L film, while lower performances were obtained with thicker 4L and 5L electrodes, despite that they practically absorb 100% of the incident light, also at long wavelengths. This provides a first indication of electron mobility issues occurring in films composed of more than three layers (i.e., thicker than ca. 270 nm). Furthermore, the slightly higher activity resulting for the 5L film compared with the 4L film could be related to the contribution to photocurrent of the longest-wavelength photons, the exploitation of which increases with increasing film thickness, that is, with increasing absorption of long-wavelength radiation (see Figure S5, showing a magnification of the IPCE profile of Figure c).

The J–V curves of Figure a were compared with the results of LSV analyses performed under front-side illumination (Figure S6; i.e., by irradiating the material through the electrode/electrolyte interface). This allows us to get more insight into the charge-carrier mobility within the material and to assess if the internal charge transport is limited by either the minority or the majority charge carriers,[31] that is, by the holes or the electrons, respectively, in an n-type semiconductor material.[32] The LSV plots recorded with the 1L and 3L CuW0.5Mo0.5O4 electrodes (ca. 80 and 270 nm thick, respectively) under back and front side irradiation are compared in Figure b. Since for both selected films the photocurrent recorded under front side irradiation is lower than that recorded under back side irradiation, poor electron transport appears to be the limiting factor for this material. The 3L electrode exhibits the best PEC performance among the tested films either under front or back irradiation (compare Figure a and Figure S6). Thus, a ca. 270 nm film thickness seems to correspond to the best balance between photon absorption in the photoactive layer and electron transport efficiency within it.

The IPCE plots recorded under back side irradiation at 1.23 V versus RHE are reported in Figure c for the complete series of multilayer CuW0.5Mo0.5O4 electrodes and for a single-layer CuWO4 photoanode. By comparing the curves obtained with the two monolayer CuWO4 and CuW0.5Mo0.5O4 photoanodes, evidence is obtained of the higher efficiency of the composite molybdenum-containing material with respect to pure CuWO4, over the entire range of investigated wavelengths. This demonstrates that the increased visible light absorption capability (see Figure a) consequent to Mo for W substitution results in a higher number of photogenerated charge couples, which are effectively responsible for the higher visible light activity of CuW0.5Mo0.5O4 with respect to CuWO4.

Furthermore, by comparing the IPCE curves obtained with multilayer CuW0.5Mo0.5O4 electrodes (Figure c), the thinnest 1L and 2L CuW0.5Mo0.5O4 films are most efficient in the UV region up to 390 nm, whereas thicker films exhibit the best photoactivity in the visible light region. The high absorption coefficient value of CuW0.5Mo0.5O4 in the UV region (the extinction coefficient at 400 nm is 3 times greater than at 500 nm, see Figure S7) implies that UV radiation is absorbed almost quantitatively even by the thinner films, which results in high photocurrent values. Furthermore, in thicker electrodes (thicker than 2L) and under back side irradiation, most of the UV photogenerated holes need to travel across the whole film to reach their extraction sites at the film/electrolyte interface. The probability that they recombine with photopromoted electrons increases with increasing film thickness leading to lower IPCE values in the UV region with respect to thinner films. Concerning the relatively higher IPCE values of the thicker films in the visible region, owing to the lower absorption capability of the material at longer wavelengths (Figure S7), quantitative light harvesting can be attained only in thicker films, which are consequently able to better exploit visible light. The best compromise between efficient hole transport across the film and maximum visible light exploitation under back side irradiation is attained with the ca. 270 nm thick film (3L CuW0.5Mo0.5O4 electrode).

The recorded IPCE values are also in good agreement with the performances under full lamp irradiation recorded in LSV scans. In fact, as shown in Table S1, there is a good matching between the photocurrent density values at 1.23 V versus RHE in J–V curves and the photocurrent density calculated by integrating the product between the IPCE curves and the standard AM 1.5 G solar spectrum[33] over the entire range of investigated wavelengths (300–650 nm for the CuW0.5Mo0.5O4 multilayer films).

Besides at 1.23 V versus RHE, the IPCE curve for the best performing 3L CuW0.5Mo0.5O4 electrode was recorded in back side configuration also at 1.50 and 1.70 V versus RHE (Figure d), in order to evaluate the conversion efficiency of the material upon increasing charge separation. As shown in Figure d, the conversion efficiency gets higher with increasing applied potential over the whole investigated wavelengths, with the IPCE value recorded at 400 nm showing a 2.5- and 3.5-fold enhancement as the potential increases from 1.23 to 1.5 and to 1.7 V versus RHE, respectively. The observed behavior is in line with the photocurrent density enhancement recorded in LSV scans under simulated solar light irradiation. In particular, the IPCE almost linearly increases with increasing applied potential (see Figure d), because the holes generated by high-energy photons, which are mainly confined in the proximity of the FTO back contact, benefit from the enhanced charge carrier separation because of the progressively higher external bias and have larger probability to reach the film/electrolyte interface where O2 evolution occurs.

Finally, the chopped photocurrent measurements under monochromatic irradiation at different wavelengths performed with our photoanodes at 1.23 V versus RHE provide uncontroversial evidence that all CuW0.5Mo0.5O4-based photoanodes absorb light and exhibit photoactivity up to 650 nm, while CuWO4 is photoactive only below 550 nm. In fact, as shown in Figure , whereas all tested photoanodes, including the CuWO4-based one, originate a photocurrent signal under irradiation at 450 nm (Figure , bottom panel), the photocurrent signal is zero for CuWO4 under irradiation at 550 nm (Figure , middle panel), while all 1L–5L CuW0.5Mo0.5O4 photoanodes remain photoactive at this wavelength and maintain their photoactivity up to 650 nm (Figure , top panel). This demonstrates that Mo for W substitution in CuWO4 largely extends the photoactivity of this material into the visible region, at least up to 650 nm.

Figure 4

Chopped chronoamperometry at 1.23 V versus RHE of the 1L–5L CuW0.5Mo0.5O4 electrodes and of the CuWO4 electrode under irradiation at 650 nm (top panel), 550 nm (middle panel), and 450 nm (bottom panel).

Sacrificial Agents

Two are the main limiting factors for PEC water oxidation performed by semiconductor-based photoanodes, that is, (i) the transport of photopromoted electrons and photogenerated holes through the bulk material to the FTO back contact and to the film/electrolyte interface, respectively, and (ii) the interfacial hole injection kinetics, determining the oxygen evolution efficiency. Decoupling these two contributions is therefore required in order to shed light on the intrinsic properties of the investigated photoactive material. In particular, the use of a suitable hole scavenger-containing electrolyte, acting as electron donor, allows one to study the intrinsic properties of the bulk material, under the assumption that hole injection at the material/electrolyte interface is not rate limiting under such conditions.[34]

Previous studies pointed out that significant dark current is observed if CuWO4 electrodes are in contact with Na2SO3- and H2O2-containing solutions.[35] Therefore, the performance of CuW0.5Mo0.5O4 photoanodes was investigated in the presence of various hole scavengers (i.e., Na2SO3, H2O2, NaNO2, and NH3) to check the existence of dark currents and verify if and how the response of this material is affected by the redox potential of the donor species. The concentration of all sacrificial agents was kept constant at 0.1 M within 0.1 M K3BO3 buffer solutions at pH 9, except for the 0.1 M Na2SO3 solution, which was at pH 9 in the absence of buffer.

The dark current onsets in the presence of different sacrificial agents was measured first using both the 1L CuW0.5Mo0.5O4 film and a glassy carbon as working electrode, in order to attain an apparent redox scale for the employed species (see Figure S8). For the electrode immersed in 0.1 M K3BO3 in the absence of any hole scavenger, the dark current potential onset is the highest (i.e., around 1.8 V versus RHE). Furthermore, the most easily oxidizable species is H2O2, immediately followed by Na2SO3, with a dark current onset located at ca. 1.0 V versus RHE. More positive onset potentials (i.e., ca. 1.4 and 1.6 V versus RHE) were found for NaNO2 and NH3, respectively. Thus, moving from the most stable to the most oxidizable species, the following scale results: H2O (i.e., K3BO3 aqueous solution) > NH3 > NaNO2 ≫ Na2SO3 > H2O2.

The LSV curves obtained with a monolayer CuW0.5Mo0.5O4 electrode in contact with the hole scavenger-containing solutions under back side simulated solar irradiation are shown in Figure . The LSV curve obtained with the 0.1 M K3BO3 aqueous solution is also reported, as reference. In the absence of any electron donor species, the photocurrent onset potential was ca. 1.0 V versus RHE (i.e., 800 mV lower than under dark conditions). The photocurrent onset potentials in the presence of the hole scavengers, extrapolated from the curves shown in Figure , are reported in Table S2. Compared with the pure K3BO3 electrolyte solution, the use of Na2SO3 and H2O2 as hole scavengers led to the largest decrease of photocurrent onset potential, which was around 0.8 V versus RHE in both cases, in good agreement with the conduction band edge value of ca. 0.7 V versus RHE.[25] However, owing to the significant dark current generated in the presence of these sacrificial agents at a potential as low as 1.2 V versus RHE, Na2SO3 and H2O2 were discarded as hole scavengers in our PEC investigation.

Figure 5

LSV curves recorded with a monolayer CuW0.5Mo0.5O4 electrode in contact with a 0.1 M K3BO3 solution, 0.1 M K3BO3 solutions containing different hole scavengers at a 0.1 M concentration or 0.1 M Na2SO3, under dark (dashed lines) or back side AM 1.5 G irradiation conditions (continuous lines). Scan rate of 10 mV s–1. All solutions were at pH 9. The photocurrent onset potential was calculated by extrapolation of each LSV line (black dashed lines).

Additionally, in the presence of NH3 and NaNO2 hole scavengers, a lower negative shift of the photocurrent onset potential (located at ca. 0.9 V versus RHE) was observed and negligible dark current up to 1.75 and 1.6 V versus RHE, respectively. This allows the exploitation of a wider applied potential window in PEC measurements. Because of the high volatility of NH3, which may result in uncontrollable concentration variation of its aqueous solutions, we finally selected NaNO2 as the most suitable sacrificial agent for our PEC tests. In its presence, the photocurrent density at 1.23 V versus RHE is double compared with that measured in pure K3BO3 (0.30 versus 0.15 mA cm–2, see Figure ), with negligible dark current at such applied bias.

Anyway, regardless of the employed electron donor, the photocurrent just moderately increased with respect to water oxidation (in K3BO3). This is in contrast with the considerable increase typically observed with semiconductors the PEC performance of which is limited by surface hole accumulation and slow water oxidation kinetics. For instance, BiVO4 photoanodes generate up to 100 times higher current in contact with a hole-scavenger-containing solution,[36] whereas Fe2O3 electrodes in contact with H2O2 allow the complete consumption of surface holes.[37] This suggests that the transfer of surface holes to water or electron donor species has minor limiting effects on the PEC efficiency of CuW0.5Mo0.5O4.

The IPCE curves obtained at 1.23 V versus RHE with the 1L and 5L CuW0.5Mo0.5O4 electrodes in the presence of the NaNO2 hole scavenger (Figure ) evidence the extended visible light photoactivity of the CuW0.5Mo0.5O4 material, which is maximum for the thickest 5L photoanode.

Figure 6

Incident photon to current efficiency (IPCE) of 1L and 5L CuW0.5Mo0.5O4 electrodes at 1.23 V vs RHE in K3BO3 both in the absence (void symbols) and in the presence (full symbols) of NaNO2.

Charge Separation and Charge Injection Efficiencies

PEC measurements in the presence of NaNO2 as hole scavenger allowed us to evaluate the charge separation and charge injection efficiencies and thus assess if the performance of CuW0.5Mo0.5O4 photoanodes is limited by either the material bulk properties or the surface hole injection kinetics.

The photocurrent density J can be expressed as the combination of different contributions, according to eq :[37,38]where Jabs is the theoretical maximum photocurrent density obtained if all absorbed photons are converted into electricity, which was estimated by converting into current the integral of the product between the standard AM 1.5 G solar spectrum and the absorption spectrum of the photoelectrode over the 300–650 nm range (4.39 mA cm–2 for our 1L CuW0.5Mo0.5O4 material),[15] ηsep is the charge separation efficiency, and ηinj is the interfacial charge injection efficiency. From this equation the charge separation and charge injection efficiencies can be easily calculated by taking into account the photocurrent densities recorded in the presence (JNaNO) and absence (JK) of the hole scavenger, as follows:

The calculated ηinj and ηsep values for the 1L CuW0.5Mo0.5O4 electrode and for the CuWO4 electrode are shown in Figure as a function of the applied potential. At 1.23 V versus RHE ηinj and ηsep are 52% and 6%, respectively, for 1L CuW0.5Mo0.5O4, which clearly indicate that charge separation in the bulk material is the main limiting factor for the PEC performance of our investigated material, in line with the limited photoactivity improvement attained in the presence of electron donor species.

Figure 7

Charge injection efficiency ηinj (continuous lines) and charge separation efficiency ηsep (dashed lines) calculated for the monolayer CuW0.5Mo0.5O4 (fuchsia) and CuWO4 (light blue) electrodes vs the applied potential.

For our CuWO4 photoanode (with Jabs amounting to 3.60 mA cm–2 in the 300–550 nm range) ηinj = 83% and a ηsep = 3% at 1.23 V versus RHE. This implies that Mo incorporation into CuWO4 induces a 2-fold increase of the charge separation properties in the bulk, in line with recent reports,[27] while it negatively affects the hole injection efficiency at the CuW0.5Mo0.5O4 film/electrolyte interface, with a ηinj decrease.

The increased charge separation efficiency can result from the increased majority carrier (i.e., electron) concentration in Mo-doped CuWO4 with respect to CuWO4 evidenced through Mott–Schottky plots in previous studies on Mo-doped CuWO4.[25−27] This would improve the material conductivity and thus reduce charge recombination in CuW0.5Mo0.5O4 films. Additionally, the reduction in charge injection efficiency may be ascribed to a Mo-induced increase of the surface trap states of CuWO4,[12,29,35,39,40] which would decrease the efficiency of water oxidation especially at relatively low applied potentials, by inducing Fermi level pinning at the semiconductor–liquid junction. An indirect evidence of the increase of surface trap states in CuW0.5Mo0.5O4 films is the modest delay in photocurrent onset of the LSV curves recorded with CuW0.5Mo0.5O4-based photoanodes with respect to the CuWO4 photoanode appearing in Figure a and better evidenced in the enlargement of this figure at low potentials shown in Figure S9. As recently outlined,[27] such delay is compatible with an increase in surface trap states upon Mo incorporation, which hampers the photocurrent onset at low overpotentials, that is, when the driving force to separate photoproduced electron–hole couples is small. Taken together, the 2-fold improvement in charge separation and the decrease in charge injection efficiency result in the overall enhanced PEC activity of CuW0.5Mo0.5O4 with respect to CuWO4.

Conclusions

Substitution of Mo for W into the CuWO4 structure is responsible for the conduction band edge energy lowering and consequent band gap narrowing. Thus, the increased PEC performance of CuW0.5Mo0.5O4 with respect to CuWO4 mainly results from the extended visible light absorption ability and photoactivity of the CuW0.5Mo0.5O4 material up to 650 nm. This effect is maximum for the highest film thickness explored in this study ensuring the exploitation of long-wavelength photons in the visible region, which are efficiently harvested for longer optical paths within the photoactive films. However, Mo for W substitution also increases the efficiency of photoproduced charge separation, which is due to an increased conductivity of the material, while the lower charge injection efficiency of CuW0.5Mo0.5O4 with respect to pure CuWO4 may result from an increased amount of surface trap states. With its 2.0 eV band gap, copper molybdo-tungstate shows promise as photoanode material for practical water splitting. Further improvements are needed and might be attained through nanostructuring, the development of efficient oxygen evolution cocatalysts for this class of materials, and the fine modification of the electronic structure by doping with other elements.