Effects of Anions and pH on the Stability of ZnO Nanorods for Photoelectrochemical Water Splitting.

This work demonstrates the improved stability of zinc oxide nanorods (ZnO NRs) for the photoanode of solar water splitting under voltage biases by the addition of borate or carbonate ions in the aqueous electrolyte with suitable pH ranges. The ZnO NRs prepared by the hydrothermal method are highly active and stable at pH 10.5 in both borate and carbonate buffer solutions, where a photocurrent higher than 99% of the initial value has been preserved after 1 h polarization at 1.5 V (vs reversible hydrogen electrode) under AM 1.5G. The optimal pH ranges with a minimum morphological change of ZnO NRs for photoelectrochemical (PEC) water splitting in borate and carbonate buffer solutions are 9-13 and 10-12, respectively. The working pH range for PEC water splitting on ZnO NR photoanodes can be extended to 8.5-12.5 by the combination of borate and carbonate anions. The lifetime of ZnO NR photoanodes can be synergistically prolonged for over an order of magnitude when the electrolyte is the binary electrolyte consisting of borate and carbonate in comparison with these two anions used individually. On the basis of the experimental results, a possible mechanism for the protective behavior of ZnO in borate and carbonate solutions is proposed. These findings can be used to improve the lifetime of other high-performance ZnO-based catalysts and to understand the photocorrosive and protective behaviors of ZnO NRs in the borate and carbonate solutions.

Introduction

Photoelectrochemical (PEC) water splitting has been considered a promising candidate for capture and high-capacity storage of solar energy.[1,2] A typical PEC cell generally consists of a semiconductor photoelectrode in which electron–hole pairs are generated upon light absorption. Following charge carrier separation, electrons reduce water to hydrogen, while holes oxidize water to oxygen. The ideal semiconductor electrode used for water splitting would require the following: (i) a suitable width of band gap >1.23 eV (thermodynamic equilibrium voltage) +0.8 eV to overcome kinetic barriers; (ii) the conduction band above the H+/H2 equilibrium potential for hydrogen evolution and the valence band below the O2/H2O equilibrium potential for oxygen evolution; and (iii) excellent PEC and photocorrosion stability in aqueous solutions.[3,4]

Zinc oxide (ZnO) has attracted much attention in the past decades as a favorable photoanode material toward achieving efficient PEC water oxidation because of its good electron mobility, excellent optical properties, wide availability, and low toxicity.[5,6] Recently, nanostructure engineering of ZnO films has been shown to improve photocurrent yields because of the large surface-to-volume ratio, which exhibits a great advantage in the PEC water splitting for efficient delivery of photogenerated carriers.[7] Despite these excellent properties, ZnO-based photoanodes are not commercialized to support reliable and efficient PEC water splitting because of several limitations. Most notably, ZnO suffers from photocorrosion in the aqueous solution under ultraviolet illumination by hole trapping on its surface, which is a common problem for many other candidate materials for water splitting.[8] Another very serious issue is ZnO decomposition when the solution pH values are too low or too high, even in dark. From the potential–pH diagram of zinc in water,[9] the predominant dissolved zinc species in acidic to neutral and basic media between 9.2 and 13.2 are Zn2+ and Zn2+(OH), respectively. According to previous studies,[8,10,11] ZnO was chemically dissolved in acids or strong bases under potential biases and/or UV light illumination. Decomposition of ZnO in aqueous solutions has been proposed as the following reactions

The chemical dissolution corresponds to the direct reaction of n class="Chemical">ZnO with protons and n class="Chemical">hydroxyl ions.[10] The electrochemical dissolution of ZnO generally needs an applied potential of around 2 V versus standard hydrogen electrode (SHE).[12] Accordingly, two competitive reactions for ZnO dissolution were suggested[13]

However, the dissolution rate of ZnO under illumination is faster than that under potential biases because of the addition of chemical dissolution which depends on the concentration of OH–. Inpan> fact, the dissolution rate of ZnO is a function of both OH– and hole concentrations at the ZnO surface, which are a function of both potential bias and light intensity.[14] The decomposition potential is determined by the formation of dissolved products when the electrode has been photocorroded. The equations describing the photoinduced decomposition of ZnO are expressed as follows[15]

Accordingly, the lowest solubility of n class="Chemical">ZnO is obtainpan>ed at pH 9.3 (10–7 to 10–8 M),[16] and the solubility inpan>creases exponentially when pH becomes lower or higher than 9.3. Furthermore, Gerischer and Sorg[17] proposed that the n class="Chemical">ZnO dissolution rate depends on the type of anion in the solution, which was directly proportional to the square root of the electrode rotation speed, which is attributable to the diffusion of protons.

To overcome these issues, two parallel paths are generally employed: (i) development of new methods to improve the stability of ZnO for the long-term usage in water splitting and (ii) coatings of conductive and protective films to prevent corrosion during the PEC reaction period.[18] Recently, surface modification of ZnO photoanodes, such as combination with another semiconductor,[19−23] deposition of noble metals,[24] and coupling with versatile carbon materials,[25] has been proposed to suppress the photocorrosion to provide the long-term performance of ZnO electrodes for the water splitting process. Moreover, phosphate and borate buffers have been shown to enhance the rate of oxygen evolution and improve the stability when various water oxidation catalysts such as manganese,[26] cobalt,[27] or nickel oxides[28] were employed. By preventing the accumulation of protons near the anode surface, these buffers significantly increase the lifetime of the catalysts. Fekete et al.[29] investigated the buffer effect on catalytic activity and stability of ZnO in a 12 h stability test. They found that ZnO was structurally stable in the borate buffer solution (pH 10.5) and prolonged its lifetime during this PEC process. However, there was no explanation for the impact of this buffered electrolyte on the PEC stability of ZnO.

In this work, the influence of buffer electrolytes on the zinc oxide nanorods (ZnO NRs) morphologies is investigated to understand the buffer influences. The effects of pH in the buffered electrolytes on the generation and stability of photocurrent are discussed. The stability mechanism of ZnO NRs and how to extend the pH window for improved stability of ZnO in the PEC water splitting process are proposed in this study.

Experimental Section

Preparation of ZnO Electrodes

n class="Chemical">ZnO n class="Chemical">NRs on the fluorine-doped tin oxide-coated glass slide (FTO glass, 7 Ω sq–1) substrate were fabricated by a hydrothermal method described in our previous study.[30] A seed layer of ZnO was deposited on FTO glass by spin-coating a solution of 10 mM zinc acetate in ethanol and subsequent calcination at various temperatures for 2.25 h to obtain the ZnO seed layer on FTO glass. Then, the ZnO seed layer/FTO glass was transferred to a 50 mL Teflon-lined stainless steel autoclave containing 20 mM Zn(NO3)2, 20 mM hexa-methylenetetramine, and 35 v/v % polyethyleneimine aqueous solution for the hydrothermal reaction at 98 °C for 4 h to obtain ZnO NRs on FTO glass. ZnO NRs/FTO glass was rinsed with ethanol and deionized water and finally dried in air at 80 °C.

Characterization of ZnO

Scanning electron microscopy (SEM) was performed with a HITACHI, SU8010 FESEM instrument with an energy-dispersive X-ray spectroscopy analysis system operated at 50 kV. X-ray diffraction (XRD) patterns were obtained using Bruker APEX DUO. XRD patterns were taken using Cu Kα as the radiation source with a 1° divergence slit, 0.2° receiving slit, and carbon monochromators at 1° min–1 (2θ range 20°–70°). Time-of-flight secondary ion mass spectrometry (TOF-SIMS, Munich, Germany) is used to characterize the surface composition of the ZnO NRs on the FTO substrate. The primary ion source was a pulsed 69Ga+ source (pulsing current 2.5 pA and a pulse width of 30 ns) operated at 15 keV with a post-acceleration of 10 kV. The analysis area of 100 nm × 100 nm, data acquisition time of 30 s, and charge compensation by applying low-energy electrons (∼30 eV) from a pulsed flood gun were used for the measurements. The mass resolution measured on the ZnO+/– signal was m/Δm = 4000 in the positive and negative detection modes. Calibration of the mass spectra in the positive detection was based on the peaks such as H+ (1.007 m/z), B+ (11.0082 m/z), 64Zn+ (63.922 m/z), 66Zn+ (65.919 m/z), and in the negative detection mode on C– (12.0004 m/z), O– (15.995 m/z), CO2– (44.003 m/z), HCO2– (45.0118 m/z), 64ZnO– (79.922 m/z), and 64ZnO– (81.9189 m/z).

(Photo)electrochemical Measurement

The PEC catalytic activity of the ZnO NRs was tested by linear sweep voltammetry (LSV) in a 0.5 M Na2SO4 solution (pH 6.8) using a potentiostat (CH Instruments 4053a) under the three-electrode mode. The ZnO NR photoelectrode, a Pt wire, and an Ag/AgCl electrode were employed as the working, counter, and reference electrodes, respectively. The LSV was measured at 10 mV s–1 from −0.5 to 1.0 V versus Ag/AgCl in the positive sweep direction. All PEC tests were measured under a 75 mW cm–2 simulated sunlight source [a xenon lamp coupled with an AM 1.5G filter (Newport)]. The measured potentials (vs Ag/AgCl, 3 M KCl, Argenthal, 207 mV vs SHE at 24 °C) can be transformed to be against the reversible hydrogen electrode (RHE) according to the Nernst equation[31]where ERHE is the transformed potential versus RHE. Accordingly, ERHE = EAg/AgCl + 0.609 (V) because pH of the 0.5 M Na2SO4 solution is 6.8.

Stability performance was tested in the electrolytes with pH between 8.0 and 13.0. The electrolytes used were as follows: 0.5 M n class="Chemical">Na2SO4, 0.5 M n class="Chemical">NaCl, 0.5 M NaClO4, 0.025 M borate buffers (at pH 8.0–13.0), 0.025 M carbonate buffers (at pH 9.0–12.75), 0.05 M α-amino acetic acid (H2N–CH2–COOH, i.e., glycine), and 0.05 M phosphate buffer solution adjusted by 0.1 M NaOH and HCl solutions.

Results and Discussion

Characterization of ZnO NRs

Figure a,b shows the morphology of ZnO NRs grownpan> on the FTO substrate, with an average diameter and length of NRs equal to 70 nm and 1.38 μm, respectively. The crystal phase of ZnO NRs corresponds to the standard XRD pattern of wurtzite ZnO (JCPDS card no. 36-1451) by powder XRD analysis (Figure c). Obviously, facet (002) is the main peak for all samples. Facet (100) appears at the seed layer annealing temperature equal to 330 °C, but its intensity decreases with the increasing annealing temperature. This result indicates that the c-axis orientation of ZnO gradually increases with the annealing temperature.

Figure 1

(a) SEM surface and (b) cross-section images of ZnO NRs grown on the FTO substrate with the seed layer annealed at 330 °C for 135 min. (c) XRD patterns and (d) LSV curves of ZnO NRs on the FTO substrate with the seed layer annealed at (1) 300, (2) 330, and (3) 350 °C. The LSV curves were measured in 0.5 M Na2SO4 at pH = 6.8 under the AM 1.5G simulated solar light irradiation.

(a) SEM surface and (b) cross-section images of ZnO NRs grown on the n class="Chemical">FTO substrate with the seed layer annealed at 330 °C for 135 min. (c) XRD patterns and (d) LSV curves of ZnO NRs on the FTO substrate with the seed layer annealed at (1) 300, (2) 330, and (3) 350 °C. The LSV curves were measured in 0.5 M Na2SO4 at pH = 6.8 under the AM 1.5G simulated solar light irradiation.

The PEC activity of ZnO NRs in 0.5 M Na2SO4 with pH = 6.8 was examined by LSV which reveals that the highest photocurrent is 0.4 mA cm–2 when the ZnO seed layer has been annealed at 330 °C (Figure d). This result is attributable to the presence of the (100)-oriented ZnO NRs which provide a superior carrier transport path along this direction and provide a short pathway for minority carriers to reach the interface.[32] As a result, the PEC performance of these nanostructured films is greatly improved. On the basis of the abovementioned results and discussion, 330 °C is the seed-layer annealing temperature of ZnO NRs used in the further studies.

Electrolyte Effects on the Stability of ZnO for PEC Processes

Figure a compares the stability of ZnO NRs biased at 1.5 V (vs RHE) in aqueous solutions containing 0.5 M Na2SO4, 0.5 M NaCl, 0.5 M NaClO4, and 0.025 M Na2B4O7. For the Na2SO4 solutions, the initial pH values were set at 6.8 and 10.5 (curves 1 and 2), while there was only one pH value (10.5) for the other electrolytes (curves 3–5). Clearly, the chronoamperogram (CA) curves in Figure a show significantly different photocurrent responses of ZnO NRs in these electrolytes, which is indicative of different PEC behaviors. In 0.5 M Na2SO4 with pH = 6.8, the PEC activity of ZnO NRs significantly decreases to 10% of the initial value within 30 min. The declination in the photocurrent density is reduced when the pH value is changed from 6.8 to 10.5 from a comparison of curves 1 and 2. From the literature,[33] the minimum solubility of ZnO is between pH 9.9 and 12.3, which has been proposed to be the stable operation pH window of ZnO. However, the PEC activity of ZnO decays significantly within 20 min in that study. Accordingly, the chemical dissolution of ZnO is expected to be induced by the specific adsorption of certain anions on the ZnO surface, and zinc complexes are liable to form in the Cl–-, ClO4–-, and SO42–-containing solutions.[33,34]

Figure 2

(a) CA curves of a ZnO photoanode biased at 1.5 V vs RHE in (1,2) 0.5 M Na2SO4, (3) 0.5 M NaCl, (4) 0.5 M NaClO4, and (5) 0.025 M Na2B4O7 with pH = (1) 6.8 and (2–5) 10.5. (b) CA curves of a ZnO photoanode biased at 1.5 V vs RHE in 0.025 M (1) Na2B4O7, (2) Na2CO3, (3) glycine, and (4) Na2HPO4 with pH = 10.5 under the AM 1.5G simulated solar light irradiation.

(a) CA curves of a n class="Chemical">ZnO photoanode biased at 1.5 V vs RHE inpan> (1,2) 0.5 M n class="Chemical">Na2SO4, (3) 0.5 M NaCl, (4) 0.5 M NaClO4, and (5) 0.025 M Na2B4O7 with pH = (1) 6.8 and (2–5) 10.5. (b) CA curves of a ZnO photoanode biased at 1.5 V vs RHE in 0.025 M (1) Na2B4O7, (2) Na2CO3, (3) glycine, and (4) Na2HPO4 with pH = 10.5 under the AM 1.5G simulated solar light irradiation.

From a comparison of curves 2–5, the stability of ZnO NRs is very obviously improved in the Na2B4O7 electrolyte with pH = 10.5. More than 100% of the initial photocurrent is preserved after the 1 h PEC test. Clearly, anions are another important factor influencing the stability of ZnO against photocorrosion. A few studies indicated that the PEC stability of ZnO photoanodes in the borate electrolyte is higher than that in common electrolytes (e.g., SO42–) under the same pH.[29] Because the borate ions with the buffering characteristics can act as a proton/OH– donor and acceptor to compensate the proton/OH– concentration gradient because of the heterogeneous reaction between photogenerated hole and water,[35] the localized pH value at the ZnO–liquid interface (i.e., interfacial pH) should not be significantly changed by the oxygen evolution reaction (OER) or OH• formation. To confirm this buffering effect on the stability of ZnO photoanodes, the PEC stability of ZnO NRs was further tested in various buffer electrolytes under AM 1.5G simulated solar light irradiation (see Figure b). Clearly, the stability of ZnO NRs in the glycine and Na2HPO4 solutions is worse than that in the Na2B4O7 and Na2CO3 electrolytes because the residual photocurrents in the former two electrolytes are equal to 74 and 30% of the initial values in the 1 h PEC test. In addition, more than 100% of the initial photocurrent is preserved when the ZnO NR photoanode has been tested in the Na2CO3 solution for 1 h. Accordingly, the mechanism for improving the PEC stability of ZnO NRs cannot be simply attributed to the buffering effect of electrolytes, and the highest stability of ZnO can be obtained in the Na2B4O7 and Na2CO3 electrolytes. Note that the order of electrolytes with respect to decreasing the initial photocurrent is as follows: Na2B4O7 (0.59 mA cm–2) > Na2CO3 (0.55 mA cm–2) > Na2HPO4 (0.32 mA cm–2) > glycine (0.22 mA cm–2). This interesting result implies that certain anions significantly enhance the PEC activity of ZnO NRs, probably resulting from the specific adsorption of these anions (see below).

Typical SEM images of ZnO NRs before and after the 1 h PEC test in various electrolytes are shownpan> in Figure . Inpan> comparison with the fresh ZnO NRs shown in Figure a, the diameter of ZnO NRs may be changed, depending on the corrosion or salt deposition processes during the 1 h PEC test. Because of the photocorrosion occurring in the Na2SO4 electrolyte, most ZnO NRs were dissolved from the FTO glass substrate after the PEC test (see Figure b). On the other hand, the morphologies of ZnO NRs were not significantly changed when ZnO NRs were tested in the Na2B4O7 and Na2CO3 electrolytes from a comparison of Figure a,c,d. These results reveal the improved PEC stability of ZnO in the borate-/carbonate-containing electrolyte, and the adsorption of these two anions may be under the monolayer situation. The rough surface and large diameter of ZnO NRs in Figure e indicate the significant photocorrosion and redeposition of Zn–glycine salts onto the ZnO NR surface in the glycine (an organic acid containing the carboxylic group) solution, resulting in the decrease in the photocurrent (see Figure e), although ZnO NRs are still visible after the PEC test. When the phosphate solution is employed, breakdown of ZnO NRs with a passive film is visible from an examination of Figure f, probably resulting in the activity loss of ZnO NRs. From the literature,[36] the formation of phosphates on the ZnO surface via eq is likely to adversely affect the PEC activity of ZnO

Figure 3

SEM images of (a) freshly prepared ZnO NRs and (b–f) ZnO NRs after the 1 h CA test at 1.5 V vs RHE under the AM 1.5G simulated solar light irradiation in (b) 0.5 M Na2SO4, (c) 0.025 M Na2B4O7, (d) 0.025 M Na2CO3, (e) 0.025 M glycine, and (f) 0.025 M Na2HPO4 electrolytes at pH = 10.5.

SEM images of (a) freshly prepared n class="Chemical">ZnO NRs and (b–f) n class="Chemical">ZnO NRs after the 1 h CA test at 1.5 V vs RHE under the AM 1.5G simulated solar light irradiation in (b) 0.5 M Na2SO4, (c) 0.025 M Na2B4O7, (d) 0.025 M Na2CO3, (e) 0.025 M glycine, and (f) 0.025 M Na2HPO4 electrolytes at pH = 10.5.

Moreover, the maximum concentration of HPO42– species was found in the pH range of 9 to 11.5, which could alter the dissolution process and form a passive layer of NaZnPO4·H2O via eq .[36] Because a tubular structure of ZnO is clearly found in Figure f, the crystal plane (002) of ZnO NRs has been corroded in the phosphate electrolyte. Because the abovementioned etching is selective, the positively charged Zn-terminated [0001] polar surface of ZnO is believed to donate Zn2+ to react with PO43– ions to form Zn3(PO4)2 as an insoluble salt which decreases its buffering capacity.[37] Therefore, the presence of phosphate ions promotes the selective corrosion and declines the PEC activity of ZnO. On the basis of the abovementioned results and discussion, the PEC stability of ZnO can be improved by the usage of borate and carbonate ions as the supporting electrolyte.

pH Effects in the Borate and Carbonate Solutions on the Stability of ZnO NRs

To further investigate the influence of the two abovementioned anions with buffering characteristics on the stability of ZnO for PEC water splitting, freshly prepared ZnO NR photoanodes were further tested in the borate and carbonate solutions at various pH values, and typical results are shown in Figures and 5. The ZnO photoanodes show excellent stability in the pH range of 9 to 13 (see Figure ). Because of the buffering effect, these two anions can release or capture protons to maintain the localized pH value at the ZnO NR/electrolyte interface when the PEC water splitting occurs.[38] On the other hand, the interfacial pH should decrease significantly during the PEC process when the solution pH was adjusted to be lower than its pKa. Accordingly, decay in the PEC activity of ZnO NRs is observed at pH = 8.7 which are lower than the pKa of Na2B4O7 (pKa1 = 9.23).

Figure 4

CA curves of ZnO photoanodes biased at 1.5 V vs RHE in 0.025 M Na2B4O7 at pH = (1) 8.7, (2) 9.5, (3) 10.5, (4) 11.0, (5) 12.0, and (6) 13.0 under the AM 1.5G simulated solar light irradiation.

Figure 5

SEM images of ZnO photoanodes after the 1 h CA test at 1.5 V vs RHE in 0.025 M Na2B4O7 at pH = (a) 8.7, (b) 9.5, (c) 10.5, (d) 11.0, (e) 12.0, and (f) 13.0 under the AM 1.5G simulated solar light irradiation.

CA curves of n class="Chemical">ZnO photoanodes biased at 1.5 V vs RHE inpan> 0.025 M n class="Chemical">Na2B4O7 at pH = (1) 8.7, (2) 9.5, (3) 10.5, (4) 11.0, (5) 12.0, and (6) 13.0 under the AM 1.5G simulated solar light irradiation.

SEM images of n class="Chemical">ZnO photoanodes after the 1 h CA test at 1.5 V vs RHE inpan> 0.025 M n class="Chemical">Na2B4O7 at pH = (a) 8.7, (b) 9.5, (c) 10.5, (d) 11.0, (e) 12.0, and (f) 13.0 under the AM 1.5G simulated solar light irradiation.

To evaluate the possible PEC activity decay mechanism, the surface morphologies of ZnO NRs after the 1 h PEC test in the Na2B4O7 electrolyte at various pH values are shown in Figure . At pH = 8.7, almost all ZnO NRs have been dissolved into the electrolyte, and the FTO substrate under the residual ZnO NRs is clearly found in Figure a. When pH is 9.5, significant damages on the surface of ZnO NRs are visible, but their PEC activity does not significantly decay in the 1 h PEC test from Figure . Therefore, the damages on the surface of ZnO NRs do not significantly destroy the [0001] facets such as (100) and (002) which are the highly active sites for water splitting.[32,39] At pH = 10.5, there is no change in the morphology of ZnO NRs before and after the 1 h PEC test (Figure c), correlating well with the high photocurrent stability over 1 h. At pH = 12, 12.5, and 13, ZnO NRs are slightly dissolved at the side of the top surface. This result suggests that excess hydroxyl ions adsorb preferentially onto the positively charged [0001]-Zn surface to form the soluble zinc species, resulting in the increase of the chemical dissolution rate when pH is higher than 12.3 (see Figure S1b–d).[40] However, the PEC activity of ZnO NRs does not significantly decay, illustrating that the (002) facet is not significantly decomposed in the 1 h PEC test.

For the n class="Chemical">carbonate buffer electrolyte, good stability of pan> class="Chemical">ZnO photoanodes is observed at pH between 10 and 12 (Figure ). However, the PEC activity of ZnO decays when pH has been increased to 12.5. Accordingly, the working pH range of ZnO in the carbonate buffer electrolyte is narrower in comparison with the borate buffer solution. Because the carbonate buffer solution is highly effective for stable photocurrents when the operation pH is closed to its pKa (ca. 10.2), the suitable operation pH range for obtaining stable photocurrents of ZnO should be shifted with the pKa value.

Figure 6

CA curves of ZnO photoanodes biased at 1.5 V vs RHE in 0.025 M Na2CO3 at pH = (1) 9.0, (2) 9.5, (3) 10.5, (4) 11.0, (5) 12.0, and (6) 12.5 under the AM 1.5G simulated solar light irradiation.

CA curves of n class="Chemical">ZnO photoanodes biased at 1.5 V vs RHE inpan> 0.025 M n class="Chemical">Na2CO3 at pH = (1) 9.0, (2) 9.5, (3) 10.5, (4) 11.0, (5) 12.0, and (6) 12.5 under the AM 1.5G simulated solar light irradiation.

The morphologies of ZnO NRs after the 1 h PEC test in the carbonate electrolyte at various pH values are shown in Figure a–f. Obviously, there is no visible change in the morphology of ZnO NRs when pH is between 10 and 12. However, when pH is increased to 12.5, a passive layer of Zn(OH)2(s) with carbonate salts seems to be formed on the ZnO NR surface via precipitation of saturated zincate ion [Zn(OH)42–], as shown in Figure f.[41] Therefore, the decay in the photocurrent of ZnO NRs should reasonably result from the coverage of a passive layer which reduces the photoconversion efficiency of ZnO NRs in such an electrolyte. Although the photocatalytic activity of ZnO for the OER can be promoted in alkaline solutions because of an enhanced reaction rate between OH– and holes, the dissolution of ZnO is also accelerated by the formation of soluble zincate ions [Zn(OH)3–(aq) or Zn(OH)42–(aq) species] in concentrated alkaline solutions via eqs and 4.[42,43] Hence, the photocurrent density decreases when pH is higher than 10.5, implying that the rate of chemical dissolution (eqs and 4) becomes significant. In contrast, the PEC activity of ZnO NRs can be preserved when the borate buffer electrolyte is employed. This is probably due to the fact that the HBO32– species existing at 12 < pH < 14[44] are mainly responsible for proton transport and reduce the concentration of hydroxyl ions at the ZnO/electrolyte interface. Another possible reason is the formation of the borate radical, •B(OH)4, (eqs –14), from the oxidation of B(OH)4– by photogenerated holes at pH > 11.5, which might oxidize water and suppress the decomposition of ZnO NRs.[45,46]

Figure 7

SEM images of ZnO photoanodes after the 1 h CA test at 1.5 V vs RHE in 0.025 M Na2CO3 at pH = (a) 9.0, (b) 9.5, (c) 10.5, (d) 11.0, (e) 12.0, and (f) 12.5 under the AM 1.5G simulated solar light irradiation.

SEM images of n class="Chemical">ZnO photoanodes after the 1 h CA test at 1.5 V vs RHE inpan> 0.025 M n class="Chemical">Na2CO3 at pH = (a) 9.0, (b) 9.5, (c) 10.5, (d) 11.0, (e) 12.0, and (f) 12.5 under the AM 1.5G simulated solar light irradiation.

We also investigated the effect of the anion concentration on the stability of ZnO in the 1 h PEC test. Figures S2 and S3 show that the stability of ZnO NRs in a relatively low pH solution can be improved by the increase in the anion concentration because of the stronger buffering capability.[47] However, the ZnO NRs have been found to be covered with sodium zincate and zinc carbonate passive layers in borate (pH 8) and carbonate (pH 9.5) solutions (see Figure S4a,b). Because the XRD pattern of ZnO NRs with the 1 h PEC test at pH 8 in the borate buffer electrolyte confirms no obvious change in the crystalline wurtzite phase, the physicochemical properties of wurtzite ZnO NRs are retained, leading to the stable photocurrent (see Figure S4c).

On the basis the abovementioned results and discussion, the surface species formed from the reactions between anions and ZnO are dependent upon the solution pH, which are believed to affect the resultant photocurrent density of ZnO NRs obtained in the borate or carbonate solutions with different pH values. From the literature,[48] the oxidation of ZnO in the buffer-containing solutions should be diffusion-controlled by the concentration of the buffer anion at pH lower than 12 and by the OH– concentration at pH higher than 12. Interestingly, the photocurrent of ZnO can be preserved at pH higher than 12 (see Figure ); meanwhile, the morphology of ZnO NRs is not significantly damaged after the 1 h PEC test (see Figure d–f). Dodd et al.[49] reported that OH– could adsorb preferentially onto the positively charged [0001]-Zn surface, increasing the reaction rate between OH– and holes (i.e., accelerated water oxidation). This report suggests that the PEC activity of ZnO to the OER or OH• formation is dependent on its crystal facet. According to the abovementioned results and discussion, the PEC activity and morphological change of ZnO in pure NaOH solutions at various concentrations are worthy of being investigated.

Figure a shows that the order of the NaOH concentration with respect to decreasing the photocatalytic activity of ZnO is as follows: 0.1 M > 0.01 M > 0.001 M > 0.2 M > 0.5 M. Surprisingly, the PEC activity of ZnO can be preserved in the 0.1 M NaOH solution (pH ≈ 12.85). On the other hand, the photocurrent densities in such pure NaOH solutions are obviously lower than those obtained in the borate or carbonate electrolytes, revealing the key role of the two abovementioned anions (probably via adsorption) in promoting the photocatalytic activity of ZnO. When the NaOH concentrations are out of 0.01–0.1 M, the photocatalytic activity of ZnO NRs decays significantly where 20, 70, and 30% of the initial photocurrent can be respectively preserved in the 0.001, 0.2, and 0.5 M NaOH solutions after the 1 h PEC test. The abovementioned results indicate that the ZnO NR photoanode is stable when the OH– concentration is between 0.01 and 0.1 M.

Figure 8

(a) CA curves of ZnO photoanodes biased at 1.5 V vs RHE in (1) 0.001, (2) 0.01, (3) 0.1, (4) 0.2, and (5) 0.5 M NaOH. (b–e) SEM images of ZnO photoanodes after the 1 h CA test at 1.5 V vs RHE in (b) 0.001, (c) 0.01, (d) 0.1, and (e) 0.5 M NaOH under the AM 1.5G simulated solar light irradiation.

(a) CA curves of n class="Chemical">ZnO photoanodes biased at 1.5 V vs RHE inpan> (1) 0.001, (2) 0.01, (3) 0.1, (4) 0.2, and (5) 0.5 M n class="Chemical">NaOH. (b–e) SEM images of ZnO photoanodes after the 1 h CA test at 1.5 V vs RHE in (b) 0.001, (c) 0.01, (d) 0.1, and (e) 0.5 M NaOH under the AM 1.5G simulated solar light irradiation.

Figure b–e shows the morphologies of ZnO NRs after the 1 h PEC test in different NaOH solutions. Note that ZnO NRs become thinner after the 1 h PEC test in 0.001 M NaOH (pH ≈ 10.95), as shown in Figure b, suggesting the dissolution of ZnO NRs. Figure c,d shows no significant change in the morphology of ZnO NRs when the ZnO photoanodes have been tested in the 0.01 and 0.1 M NaOH media, although part of ZnO NRs become slightly finer. Accordingly, the photocorrosion of ZnO slightly occurs in the two abovementioned NaOH solutions. Also, note the visible dissolution on not only the [0001] facet but also the nonpolar [101̅0] facet. This phenomenon implies the homogeneous dissolution of ZnO NRs, and the small diameter of NRs (<100 nm) limits the dissolution of the top surface by a low diffusion rate.[50] Because the damage of the [0001] facet is minor in the 1 h PEC test, the photocurrent of ZnO NRs can be preserved. However, when the NaOH concentration reaches 0.2 M, ZnO NRs collapse, as shown in Figure S5a. One possible reason is the formation of passive Zn(OH)2(s) (see Figure S5b) via the precipitation of saturated zincate [Zn(OH)42–] in the vicinity of the photoanode surface.[41] Because of the coverage of such a passive layer on the electrode surface, the [0001̅] facet may be corroded significantly by the damage of lattice oxygen via the photodegradation according to the following reactions[51]

In the 0.5 M n class="Chemical">NaOH solution, most ZnO NRs disappear and the FTO substrate is exposed after the 1 h PEC test (see Figure e), revealing the serious damage of the ZnO NR photoanode. From all the abovementioned results and discussion, two possible mechanisms strongly depending on the pH value of electrolytes are proposed here to be responsible for the damage of ZnO NRs. In the relatively low pH media (e.g., 0.001 M NaOH), the diffusion-controlled dissolution of ZnO NRs occurs. In the high pH solutions (e.g., 0.2 and 0.5 M NaOH), facet-selective dissolution of ZnO NRs dominates because of the formation of a passive layer on ZnO NRs (i.e., eqs and 16).

Possible Mechanism

The interactions among solution pH, buffer anion, and the PEC activity/stability of ZnO NRs are shownpan> in Figure a,b. When pH is lower than the pKa value, the decay in the photocatalytic activity with the PEC testing time is attributable to the chemical dissolution (eqs and 2). With increasing pH, the PEC activity of ZnO is enhanced by the adsorption of relatively concentrated OH– on the active sites. In the meantime, the local pH at the vicinity of the ZnO surface can be leveled by the buffering agents, reducing the corrosion of ZnO by H+. Note that a stable photocurrent can be obtained in the borate buffer solution with pH between 11 and 13 (see curve 1 in Figure a), but the photocurrent (i.e., PEC activity) reaches a maximum at pH = 10.5. The latter result reveals that the presence of borate ions significantly promotes the PEC activity of ZnO, probably because of the co-adsorption of borate and hydroxyl ions. The former result is also attributable to the co-adsorption of OH– and HBO32– on the ZnO surface, promoting the stability of ZnO NRs. The abovementioned statements are supported by the fact that when pH equals 13 (i.e., the OH– concentration in the borate solution is above 0.1 M NaOH), the photocurrent density of ZnO NRs is 0.46 mA cm–2, 30% higher than the photocurrent density obtained in 0.1 M NaOH (0.35 mA cm–2 from Figure a). In addition, the borate radical, •B(OH)4, can be formed in the aqueous solution at pH = 10.5[45,46] and might act as a mediator for hole transfer from ZnO to prevent the occurrence of photocorrosion.

Figure 9

(1) Relative and (2) initial photocurrent densities of ZnO photoanodes biased at 1.5 V vs RHE in 0.025 M (a) Na2B4O7 and (b) Na2CO3 at various pH values. The relative photocurrent density is the ratio between the initial photocurrent density and the photocurrent density measured at time equal to 60 min.

(1) Relative and (2) initial photocurrent densities of n class="Chemical">ZnO photoanodes biased at 1.5 V vs RHE inpan> 0.025 M (a) pan> class="Chemical">Na2B4O7 and (b) Na2CO3 at various pH values. The relative photocurrent density is the ratio between the initial photocurrent density and the photocurrent density measured at time equal to 60 min.

To further understand the protection behavior of ZnO in borate and carbonate electrolytes, the surface composition of ZnO NRs was measured by TOF-SIMS in the positive and negative modes. Figure shows the intensity variation of characteristic ions of ZnO (64Zn+ and 66Zn+), borate (B+), and carbonate (CO2– and CHO2–) as a function of pH. From Figure a, the intensity of characteristic ions of ZnO is close to 1 at the pH 9.5, 12, and 13, indicating that the construction of ZnO may not be changed after the 1 h PEC test. The intensities of 64Zn+ and 66Zn+ are increased when the pH is 8.7, indicating a change in the chemical structure of the ZnO NR surface by PEC corrosion. However, the decrease in the intensity of 64Zn+ and 66Zn+ is clearly found when pH is 10.5. This phenomenon is attributable to the adsorption of borate anions [B(OH)4–] on the ZnO NR surface, as shown in Figure b. Such adsorbed borate ions reasonably act as a mediator for hole transfer from ZnO to promote the water oxidation and prevent the occurrence of photocorrosion. In the carbonate electrolyte, the intensities of characteristic ions of ZnO shown in Figure c are higher than 1, indicating that photocorrosion occurs when the solution pH is below 10. However, when pH is higher than 12, a decrease in the intensity is found. Figure d shows that the adsorption of carbonate ions results in the formation of the carbonate salt on the ZnO surface. As a result, the coverage of the carbonate salt can reduce the contact surface area between ZnO and the electrolyte, leading to the decrease in the PEC activity of ZnO, corresponding to Figure b. The presence of carbonate ions in the borate electrolyte is reasonably attributed to the dissolution of CO2(g) from the atmosphere.

Figure 10

Intensity of characteristic ions of (a,c) ZnO (64Zn+, 66Zn+), (b) boron (B+), and (d) carbonate (CO2– and CHO2–) for ZnO NRs after the 1 h CA test at 1.5 V vs RHE under the AM 1.5G simulated solar light irradiation in 0.025 M (a) borate and (c) carbonate electrolytes as a function of pH.

Intensity of characteristic ions of (a,c) n class="Chemical">ZnO (64n class="Chemical">Zn+, 66Zn+), (b) boron (B+), and (d) carbonate (CO2– and CHO2–) for ZnO NRs after the 1 h CA test at 1.5 V vs RHE under the AM 1.5G simulated solar light irradiation in 0.025 M (a) borate and (c) carbonate electrolytes as a function of pH.

On the basis of the literature[41,45,46,52] as well as the abovementioned results and discussion, the photocorrosive and protective processes of ZnO as a function of pH in both borate and carbonate buffer solutions are simply proposed as the scheme shown in Figure . There are several key features of the reaction scheme. First, the dissolution of zinc oxide is attributable to both chemical and PEC reactions to form zinc salt complexes on the electrode surface. Second, the buffering agents can level the variations in H+ or OH– concentrations at the electrode/electrolyte interface (e.g., the electrical double-layer region) to maintain desired pH values to stabilize the photoanode. Third, the adsorbed anions, for example, B(OH)4– or CO32–, can work as a hole mediator to promote the water splitting reaction and suppress the photocorrosion. Fourth, the formation of the zincate salt can protect the ZnO NRs but results in the passivation of ZnO when the amount of carbonate ions is too high in the high pH range (see Figure b). The abovementioned phenomena should take place in the borate and carbonate buffer solutions when their pH values are significantly lower or higher than their respective pKa values.

Figure 11

Schemes describe the photocorrosive and protective processes of ZnO NRs as a function of pH in the (a) borate and (b) carbonate solutions.

Schemes describe the photocorrosive and protective processes of n class="Chemical">ZnO NRs as a function of pH inpan> the (a) n class="Chemical">borate and (b) carbonate solutions.

To further confirm the idea that the stability of ZnO NRs can be improved by the usage of borate and carbonate electrolytes for the PEC water splitting, ZnO NR photoanodes were further examined in the relatively long-term PEC test, and the corresponding SEM images are shown in Figure S6. After the 12 h PEC test, the PEC activity of ZnO NRs decays slightly, where 92 and 91% of the initial photocurrents can be preserved in the borate and carbonate solutions, respectively. When the PEC time was prolonged to 24 h, the photocurrents of ZnO NRs were found to decrease to 70 and 80% of their initial photocurrents (see Figure S6a). Note that the morphologies of ZnO NRs shown in Figure S6b,c were not significantly changed after the 24 h-PEC test, which confirmed the improved stability of ZnO NRs in the borate and carbonate electrolytes.

Stability of ZnO in the Binary Borate–Carbonate Solution

Here, the combination of borate and carbonate anions is proposed to improve the stability and PEC performance of ZnO NRs in the solar water splitting process. This idea is demonstrated by the PEC activity of ZnO NRs in the mixed electrolyte with various pH values in the 1 h PEC test. As shown in Figure a, curves 1 and 2 show that the most stable photocurrent density (around 0.4–0.5 mA cm–2, significantly larger than the maximum photocurrent density obtained in the pure NaOH solution) can be obtained in the mixed electrolyte with pH between 8.5 and 12.5. Accordingly, the suitable and working pH range for ZnO in solar water splitting is obviously extended by the combination of borate and carbonate ions. When pH is lower than 8.5 or higher than 12.5, decay in the photocurrent has been found. Moreover, the SIMS analysis of carbonate and borate anions from the ZnO surface (curves 1 and 2 in Figure b) shows that the adsorbed B+ and CO32– contents increase when pH is changed from 8.3 to 10.5. This result confirms that both borate and carbonate anions can promote the PEC reaction. However, the adsorbed CO32– and OH– contents (curve 3) increase when pH is increased to 12.5, indicating that the adsorption of carbonate ions results in the formation of the carbonate salt on the ZnO surface. The highest intensity of OH– is found at pH 8.3 and 8.5 because of the formation of ZnOH+ on the ZnO NR surface (eq ). In addition, photocorrosive damage of ZnO NRs is also visible in the abovementioned cases from Figure S7. Therefore, the PEC activity and stability of ZnO can be improved by the combination of Na2B4O7 and Na2CO3 buffering agents.

Figure 12

(a) (1) Relative and (2) initial photocurrent densities of ZnO photoanodes biased at 1.5 V vs RHE in 0.025 M Na2B4O7 + Na2CO3 at various pH values. (b) Intensity of characteristic ions of (1) B+, (2) CO32–, and (3) OH– for ZnO NRs after the 1 h CA test at 1.5 V vs RHE under the AM 1.5G simulated solar light irradiation in 0.025 M Na2B4O7 + Na2CO3 electrolytes as a function of pH.

Table compares the results of methods in suppressing the photocorrosion of ZnO during the PEC reaction. Qiu et al.[53] and Lee et al.[54] reported the ∼90% loss of the initial photocatalytic activity of ZnO-based photoanodes after a 1 h test when a nonbuffered Na2SO4 electrolyte was used (pH 6.8). The stability of ZnO was improved by doping metals into ZnO; however, the decline of photocatalytic activity was observed because of the dissolution of ZnO which contacted directly between the partial bare ZnO electrode and the electrolyte. Another improvement method is the TiO2 coating on ZnO, which showed 100% stability. In comparison with the metal-doped ZnO, the TiO2-coated ZnO shows completely preserved photocurrent at pH 13 after the 1 h PEC reaction.[55] However, Yan et al.[56] and Su et al.[57] reported 86 and 90% retention of the initial PEC activity in the nonbuffered Na2SO4 electrolyte using BiVO4 and poly(2,5-dimercapto-1,3,4-thiadiazole) (PDMcT) as the protective films. Note that the electrical energy used by Yan et al.[56] was higher than that used in Liu et al.[55] Note that Fekete et al.[29] has proposed that the stability of ZnO can be improved by the usage of buffered electrolytes in the pH 9–12.5 region. Also, note the ∼100% retention of the initial photoactivity in the present study as well as the TiO2-coated ZnO electrode reported by Liu et al.[55] at pH 13. From a comparison of previous studies in the literature, Fekete et al.[29] reported 100% stability after the 1 h PEC reaction at pH 10.5. Nonetheless, our results are comparable to those reported by these authors. In fact, the borate concentration used by Fekete et al.[29] is 24 times that employed in this study, although our results are better than those reported by Fekete et al. This result indicates that the stability of ZnO NRs in the binary borate–carbonate electrolyte is better than that in the single borate or carbonate electrolyte which is relatively easily corroded by dissociation of Zn ions at relatively low pH conditions (pH 8.5) during the reaction.

Table 1

Stability Comparisons of ZnO-Based Photoanodes in Recent Literature

photo-anode	electrolyte	pH	Eapp/initial j	PCa retention after 1 h PEC (%)	refs
ZnO nanowire	0.1 M KOH	13	1.23 V vs RHE/0.4 mA cm–2	90	Liu et al.[55]
TiO2-coated ZnO nanowire	0.1 M KOH	13	1.23 V vs RHE/0.5 mA cm–2	100	Liu et al.[55]
PDMcT-coated ZnO NR arrays	0.5 M Na2SO4	6.8	1.15 V vs RHE/0.8 mA cm–2	90 (at 15 min)	Su et al.[57]
ZnO NRs/BiVO4	0.5 M Na2SO4	6.8	1.8 V vs RHE/1.72 mA cm–2	86 (at 17 min)	Yan et al.[56]
N-doped ZnO nanotetrapods	0.5 M Na2SO4 with phosphate buffer	7	1.0 V vs RHE/1.0 mA cm–2	90	Qiu et al.[53]
V-doped ZnO nanosheets	0.5 M Na2SO4	6.8	1.23 V vs RHE/17.12 μA cm–2	88	Lee et al.[54]
screen printed ZnO NPs	0.6 M borate buffer	10.5	1.23 V vs RHE/0.67 mA cm–2	100	Fekete et al.[29]
ZnO NRs	0.025 M borate buffer	10.5	1.5 V vs RHE/0.59 mA cm–2	100	this study
ZnO NRs	0.025 M carbonate buffer	10.5	1.5 V vs RHE/0.55 mA cm–2	100	this study
ZnO NRs	0.025 M borate buffer	13	1.5 V vs RHE/0.47 mA cm–2	93	this study
ZnO NRs	0.025 M borate + carbonate	8.5	1.5 V vs RHE/0.41 mA cm–2	100	this study

PC: photocurrent density under AM 1.5G simulated solar light irradiation.

Conclusions

The ZnO NR photoanode is highly stable and active in the PEC water splitting process when borate and carbonate ions have been employed as the supporting electrolyte because their buffering capability maintains the desired pH value at the vicinity of the ZnO NR surface. The results of photocurrent density and TOF-SIMS demonstrate that both anions adsorbed on the ZnO NR surface can act as a mediator to increase the photocatalytic activity of ZnO and protect ZnO NRs. The working pH windows for the solutions containing single borate or carbonate are equal to 9–13 and 10–12, respectively, where more than 99% initial photocurrent can retain in the 1 h PEC test. In the unsuitable pH conditions, the decrease in the PEC activity of ZnO NRs is attributable to the photocorrosion of the crystalline wurtzite phase. ZnO NRs are stable in the borate and carbonate electrolytes at pH between 9.5 and 12 and show a minor morphology change after the 1 h PEC test. When pH is higher than 12.5, the photocatalytic activity decay of ZnO NRs is dependent on the adsorption of buffer and hydroxyl ions in the vicinity of the ZnO surface, resulting in the formation of a passive Zn(OH)2 film with a carbonate salt layer, decreasing the surface active sites. Most importantly, the stable working pH window of ZnO in PEC water splitting is obviously extended (between 8.5 and 12.5) by the combination of borate and carbonate ions. The lifetime of ZnO photocatalysts for PEC water splitting can be effectively prolonged using the binary electrolyte.