Efficient BiVO4 Photoanodes by Postsynthetic Treatment: Remarkable Improvements in Photoelectrochemical Performance from Facile Borate Modification

Abstract Water‐splitting photoanodes based on semiconductor materials typically require a dopant in the structure and co‐catalysts on the surface to overcome the problems of charge recombination and high catalytic barrier. Unlike these conventional strategies, a simple treatment is reported that involves soaking a sample of pristine BiVO4 in a borate buffer solution. This modifies the catalytic local environment of BiVO4 by the introduction of a borate moiety at the molecular level. The self‐anchored borate plays the role of a passivator in reducing the surface charge recombination as well as that of a ligand in modifying the catalytic site to facilitate faster water oxidation. The modified BiVO4 photoanode, without typical doping or catalyst modification, achieved a photocurrent density of 3.5 mA cm−2 at 1.23 V and a cathodically shifted onset potential of 250 mV. This work provides an extremely simple method to improve the intrinsic photoelectrochemical performance of BiVO4 photoanodes.


Introduction
Water splitting by photoelectrochemical (PEC) cells is one of the most promising ways to obtain ar enewable H 2 fuel. [1] Since electrochemical photolysis of water at aT iO 2 photoanode was reported by Fujishima and Honda in 1972, [2] metal oxide based semiconductors have become attractive materials for photocatalysis and PEC cells. [3] An ideal semiconductor applicable for aPEC cell requires as uitable band gap to utilize as ignificant portion of the solar spectrum, an effective charge separation in the bulk, an efficient charge transfer at the semiconductor/electrolyte interface,a nd al ong-term stability in aqueous media. [4] Among the metal oxide based semiconductors,m onoclinic bismuth vanadate (BiVO 4 )i sc onsidered the most promising owing to its suitable band gap (ca. 2.4 eV) that enables it to absorb about 11 %ofthe visible light spectrum, its long carrier lifetime (ca. 40 ns), low cost, and good stability. [5] Under the standard AM 1.5 Gs unlight illumination, the theoretical photocurrent density of BiVO 4 is estimated to reach am aximum of 7.5 mA cm À2 ,r esulting in as olar-to-hydrogen conversion efficiency of close to 9.2 %. [4b,6] However,t he PEC performance of pure BiVO 4 photoanode is greatly limited by its low carrier mobility (ca. 4 10 À2 cm 2 V À1 s À1 ), short hole-diffusion length (ca. 100 nm), and slow water oxidation kinetics. [7] Plenty of approaches have been attempted to overcome these limitations,i ncluding element doping, [8] morphology engineering, [9] heterostructure formation, [10] oxygen evolution catalysts (OECs)-layer loading, [11] crystal facet engineering, [12] plasmonic enhancement, [13] and combinations thereof.However, the efficiency of BiVO 4 photoanodes is still far from an application level. [5a] Beside these well-studied techniques,aseries of postsynthetic treatments,aconcept proposed by Smith and Stefik, have recently emerged as as imple and effective strategy to enhance the intrinsic photocatalytic activity of BiVO 4 photoanodes. [14] Instead of requiring the use of additional materials,s uch posttreatments stand out as methods to change the defect chemistry,b oth at the surface and in the bulk of BiVO 4 .I t provides new mechanisms and opportunities to understand and enhance the intrinsic properties of BiVO 4 photoanodes for higher PEC performance.
To date,avariety of postsynthetic modifications have been reported, including annealing under H 2 or N 2 , [15] illumination (that is,photocharging), [16] UV curing, [17] electrochemical treatment, [18] acid vapor etching, [12b] Li/EDA(ethylenediamine) solution treatment, [19] and so on. Herein, we found an extremely facile postsynthetic treatment for the improvement of BiVO 4 photoanodes:m odifying the BiVO 4 electrodes with ab orate species at the molecular level. The treated BiVO 4 photoanodes (denoted as B-BiVO 4 )c onsistently exhibit excellent PEC performance for water oxidation under AM 1.5 Gi llumination, with an ear tenfold enhancement of photocurrent at 0.7 V RHE and ac athodic shift of the onset potential by 250 mV.Aseries of control experiments were performed;d etailed physical characterizations,e lectrochemical impedance spectroscopy (EIS), and kinetic isotope effect (KIE) studies were conducted to reveal the significant role played by the addition of the borate moiety.

Results and Discussion
Nanoporous BiVO 4 photoanodes were prepared according to an established method, with af ew minor modifications.
[5c] At ypical worm-like nanostructure of the resulting BiVO 4 with athickness of about 600 nm is shown in the SEM images (Supporting Information, Figure S1a,b). Them onoclinic phase and aband gap of 2.42 eV are indicated by X-ray diffraction and UV/Vis absorption spectra, respectively (Supporting Information, Figure S1c PEC performances of pristine BiVO 4 and B-BiVO 4 were monitored in at hree-electrode cell, with 0.5 m borate buffer (pH 9.3) as electrolyte,under simulated sunlight illumination (AM 1.5 G, 100 mW cm À2 ). Pristine BiVO 4 showed an onset potential of 0.57 V( defined at 0.1 mA cm À2 photocurrent density) and am aximum photocurrent density of only 1.6 mA cm À2 at 1.23 Vv s. ar eversible hydrogen electrode (RHE;F igure 1b). Surprisingly,ahighly improved photocurrent density was exhibited by B-BiVO 4 ,r eaching 3.5 mA cm À2 at 1.23 V. Theo nset potential cathodically shifted to 0.32 V. Photocurrent density of B-BiVO 4 at 0.7 V is approximately ten times higher than that of the pristine BiVO 4 .T he significantly enhanced PEC performance of B-BiVO 4 was further confirmed by the transient photocurrent (Figure 1c), applied bias photon-to-current efficiency (ABPE, Figure 1d), and incident photon-to-current conversion efficiency( IPCE) measurements ( Figure 1e). Am aximum ABPE of 1.1 %was obtained by B-BiVO 4 .IPCE of B-BiVO 4 at 0.7 Vs howed au niversal double increment compared to the pristine BiVO 4 and reached am aximum of 38 %a tawavelength of 460 nm.
Regarding the stability of B-BiVO 4 under an open-circuit condition, when B-BiVO 4 was stored under air for 24 h, the PEC performance showed only asmall decrease (Supporting Information, Figure S2);when B-BiVO 4 was stored in Milli-Q water overnight, the PEC performance kept approximately 85 %o fi ts incipient performance (Supporting Information, Figure S3). These observations distinguish B-BiVO 4 from the BiVO 4 after photocharging treatment, where the photocharged BiVO 4 totally lost its increment of PEC performance when stored in dark overnight in buffer solution, [16b] indicating that ad ifferent underlying mechanism is responsible for the improvement in the PEC performance of B-BiVO 4 . To investigate this underlying mechanism, we firstly established, by means of as eries of control experiments on the immersion treatment, that the remarkable effect is indeed caused by the involvement of the borate species.T he possibility that the improvement in PEC performance is due to the basic pH condition can be safely excluded as no obvious change in the photocurrent density is observed when the pristine BiVO 4 is soaked in aN aOH aqueous solution (pH 9.3) instead of the borate solution ( Figure 2a). Regarding the effect of salt ions,atreatment with neither NaOAc nor NaClO 4 solution brings in an improvement in the photocurrent of BiVO 4 .T he bare BiVO 4 ,treated with aphosphate buffer, showed some visible enhancement of PEC performance,b ut it was still far less than B-BiVO 4 .
Furthermore,t he borate treatment itself was studied in greater detail by changing the borate concentration, immersion time,t emperature,a nd pH value of the borate solution. PEC performances of the corresponding B-BiVO 4 photoanode markedly rose with the increase in the borate concentration under the same soaking duration (Figure 2b). PEC performances of the resulting B-BiVO 4 treated in the same borate solution improved with respect to the immersion time ( Figure 2c)during the first 12 h. Extension of the immersion time over 12 hled to negligible improvement, indicating that the full transformation of the pristine BiVO 4 to B-BiVO 4 was completed in the stipulated time.I nterestingly,i tw as found that the borate treatment can be considerably accelerated by increasing the reaction temperature (Supporting Information, Figure S4);B -BiVO 4 with the best PEC performance can be generated after only 25 min of treatment at 100 8 8C. Especially noteworthy is the fact that the effect of the modification is highly dependent on the pH of the borate solution. The highest improvement was achieved by the treatment with aborate solution in the pH range of 9to10, approaching boric acid pK a of 9. 24 (Figure 3c). These demonstrate that the nature of the bulk of B-BiVO 4 ,f or example,s tructure and absorbance,remain unchanged. Therefore,w ec an conclude that the alteration of the BiVO 4 film, induced by the borate modification, happens owing to the changes on the surface.R aman spectroscopy, high-resolution transmission electron microscopy (HRTEM), X-ray photoelectron spectroscopy (XPS), and near-edge Xray absorption fine structure (NEXAFS) spectroscopy,a ll of which are powerful techniques for surface characterization, were employed to explore structural details of the surface changes by the borate modification. Unfortunately,t he Raman spectra of the pristine BiVO 4 and B-BiVO 4 were found to be superimposable,s howing no identifiable structural changes (Supporting Information, Figure S6). No obvious interface or newly generated nanolayer was observed from the HRTEM images either (Figure 3d). TheX PS spectra of both species exhibited typical O1s, V2p, and Bi 4f peaks (Supporting Information, Figure S7). TheB i4f and V2pp eaks,a nd the O1sp eak at 529.4 eV,d isplayed negligible shifts before and after the borate treatment (Figures 3e-g). Theo nly obvious change is that the O1s peak at 531.6 eV,w hich is commonly attributed to chemisorbed ÀOH groups,s hifts to 532.1 eV,w ith an evidently higher density of such groups (Figure 3g). This change can be as ign of an increase in chemisorbed -OH groups due to the absorption of [B(OH) 4 ] À or À OH or both. [20] However,i t should be noted that surface contamination ( À CO and À CO 2 ) may also cause changes in the O1sp eak. [21] It has been clearly demonstrated earlier that ab orate moiety is involved in the modification of the surface of the BiVO 4 film to afford an efficient B-BiVO 4 species.U nfortunately,m ost of the above characterization methods failed to show the nature of the exact changes.E ven ab oron signal could not be identified in the elemental analysis by XPS (Supporting Information, Figure S7) or HRTEM-EDS (EDS corresponds to energy-dispersive X-ray spectroscopy;S upporting Information, Figure S8). However,this is not afactual contradiction, because boron is very light element. As it is in asystem with heavy metal, the detection limits of both these techniques are very high. [22] It is difficult to detect aBsignal when its content is not abundant in the sample.E ven for atypical boron-doped BiVO 4 with B-compositions of 3% and 10 %, [20,23] the observed Bsignals are very weak, indicating the level of ad etection limit. In comparison, the amount of surface absorbed borate in this case can be orders of magnitude lower. This should explain the failure in detecting aBsignal. Them issing Bs ignal in the regular physical characterization, in effect, indicates that the borate modification of the BiVO 4 surface is at am olecular level with an extremely low borate concentration.
To display the presence of trace amount of borate on BiVO 4 surface,w ec onducted am ore sensitive characterization, the NEXAFS measurements by using low-energy secondary electrons.T he NEXAFS B1se dge spectrum displayed that there may have been at race of Ba tt he surface of the B-treated BiVO 4 sample,asrevealed by asmall peak at approximately 194.1 eV owing to the boric acid/ borate species in Figure 3h. [24] NEXAFS measurements were accomplished using low energy secondary electrons of about 14 eV (more precisely over a13-15 eV range), noting that the inelastic mean free path (IMFP) of the detected secondary electrons,w hich is related to the escape depth and sampling depth of NEXAFS,i s3 .6 nm at this electron energy with inorganic materials. [25] Accordingly,t he NEXAFS data pertain to the sample surface indicating the presence of Ba t atrace level. It is unsurprising that NEXAFS located atrace of Bi nt he treated sample,a lthough XPS were unable to detect Bs ignal. Indeed, this is not ap recedent in the NEXAFS detection of trace elements owing to the enhanced sensitivity of NEXAFS.F or example,N EXAFS of the Fe Ledge yielded high-quality spectra with the detection of Fe II / Fe III states at an Fe depleted iron chalcogenide surface since the photoabsorption cross-section increased by several orders of magnitude,substantially boosting the analytical sensitivity of NEXAFS when the incident beam energy approached and resonated with the Fe L-edge. [26] To further reveal the underlying mechanism of the dramatic effect induced by the borate modification, we thoroughly investigated the photogenerated carrier transfer kinetics of BiVO 4 before and after the borate treatment. Mott-Schottky curves of both samples show positive slopes, as expected, for the n-type semiconductors (Figure 4a). Based on the slope of the Mott-Schottky curves,c arrier density increment of B-BiVO 4 ,c ompared to that of the pristine BiVO 4 ,i sn egligible.T he flat band potential (intercept on x axis) of the bare BiVO 4 anodically shifts by only as mall value of 25 mV.M oreover,a na nodic shift cannot contribute to the negative shift of the photocurrent onset potential of B-BiVO 4 for water oxidation. TheMott-Schottky analysis again demonstrates that the bulk properties of BiVO 4 are not affected by the borate modification. Electrochemical impedance spectroscopy (EIS) measurements show that the B-BiVO 4 photoanodes have the same series resistance R s but am uch smaller interfacial charge transfer resistance R ct as that of the pristine BiVO 4 (Figure 4b), indicating that the improvement in photocurrent density of B-BiVO 4 can be attributed to the enhanced surface charge transfer rather than to the bulk charge transport.
More precisely,t he contributions of the increased photocurrent density (J), which is determined by three fundamental components [given by Eq. (1)],n amely light absorption (represented as J abs ), charge transport efficiency in the bulk (h transport ), and charge transfer at the semiconductor/electrolyte interface for water oxidation (h transfer ), were studied to confirm the identification of the key factors for the high PEC performance observed in the case of B-BiVO 4 . Theborate treatment has trivial effect on J abs ,because the unmodified BiVO 4 and B-BiVO 4 have comparable light absorption properties,a ss hown by the similar UV/Vis absorption spectra for both. The h transport and h transfer were separately evaluated by employing ac onventional holescavenger method. Figure 4c shows the J-V curves for the pristine BiVO 4 and B-BiVO 4 photoanodes,determined in the electrolyte with and without ah ole-scavenger,N a 2 SO 3 .I n contrary to the differences in PEC performances for water oxidation, the pristine BiVO 4 and B-BiVO 4 exhibited comparable photocurrent density when sufficient Na 2 SO 3 was introduced in the electrolyte.Considering that J abs is the same for both samples,itisrational to deduce that B-BiVO 4 has the same h transport as the pristine BiVO 4 .I nc ontrast, h transfer of B-BiVO 4 ,asshown in Figure 4d,isatleast two-fold higher than that of the pristine BiVO 4 ,d epending on the applied potential.
Finally,wefound out that the immensely increased h transfer (that is,s urface catalytic efficiency) is the key factor in the observed improvement in PEC performances of B-BiVO 4 after the borate treatment. Three factors can be responsible for an increase in h transfer ,i ncluding al arger surface area, suppressed surface charge trapping,a nd an accelerated catalytic rate of water oxidation reaction. First, the surface areas of the BiVO 4 photoanode before and after the borate treatment were evaluated by electrochemical capacitance measurements (Supporting Information, Figure S9). Electrochemically active surface areas (EASA) of the pristine BiVO 4 and B-BiVO 4 were found to be similar, which rules out its contribution to the higher h transfer .T he case of surface charge trapping was investigated by measuring the open-circuit voltage (U oc ). [11d] When BiVO 4 is immersed in the electrolyte, the illumination induced increment in U oc depends on the photogenerated carrier density,which results in anew quasi-Fermi level. Thep hotovoltage for B-BiVO 4 was detected as 0.27 V, which was 50 mV higher than that detected for the bare BiVO 4 ,i ndicating the suppression of surface charge trapping on B-BiVO 4 (Figure 4e). The50mVofphotovoltage difference between BiVO 4 and B-BiVO 4 is much less than the 250 mV cathodic shift in the onset potential for water oxidation. Therefore,s uppression of surface charge trapping is one of the factors that must have played ar ole in the enhancement of h transfer of B-BiVO 4 .
Since the J-V curve for the B-BiVO 4 photoanode, determined with ahole-scavenger,did not show any cathodic shift, while its J-V curve for water oxidation cathodically shifted 250 mV (Figure 4c), it is obvious that the rate of water oxidation on the B-BiVO 4 photoanode was enhanced tremendously,w hich is the other important factor facilitating surface charge transfer in the case of B-BiVO 4 .T he faster water oxidation on the B-BiVO 4 surface can be further established by the study of photocurrent transients.Light onoff cycles in chopped light chronoamperometry is usually accompanied by photocurrent transient spikes,caused by the discrepancy between the fast carrier generation and slow surface reaction dynamics. [8b,11d,27] Thespikes for the B-BiVO 4 photoanodes are much smaller compared to that of the bare BiVO 4 ones;m oreover,n oc harge accumulation was found for B-BiVO 4 ,asshown by the damped-current during light-off (Figure 4f). These observations demonstrate that the borate modification accelerated the catalytic rate of water oxidation on the modified BiVO 4 surface.Indeed, the dramatic effect of surface modification on photocatalytic performance have been studied for other bismuth-based semicondutors. [28] Based on the control experiments,p hysical characterizations,and carrier transfer kinetics studies,wepropose that the immersion treatment in borate buffer solution is indeed as pontaneous process in which the tetrahedral [B(OH) 4 ] À gradually interacts with the active site (that is,defect) on the BiVO 4 surface ( Figure 5). Them ost likely sites for the tetrahedral [B(OH) 4 ] À are the defects formed as ar esult of vanadium loss. [23,29] Thea dsorbed [B(OH) 4 ] À may act as ap assivator to reduce charge recombination [22a] and to facilitate extraction of holes to the surface. [30] More importantly,t he anchoring of the borate moiety at the catalytic active site significantly accelerated the catalytic rate of water oxidation. Therole played by the self-anchored borate can be considered as aligand effect at the catalytic site on the BiVO 4 surface.I tc an modify the electronic configuration of the bismuth catalytic site and consequently,a ccelerate the OÀO bond formation rate.Atthe same time,the anchored borate, as an internal base,c an also assist the concerted protonelectron transfer,w hich has been shown to be essential for water oxidation by molecular catalysts, [31] metal oxides, [32] and semiconductor photoanodes. [33] KIE studies of the pristine BiVO 4 and B-BiVO 4 photoanodes indicated that proton transfer is involved in the rate determining step (RDS) because aK IE value of approximately 2.6 was observed for  Figure S13). The deactivation of B-BiVO 4 can be induced by photocorrosion [11b,34] or desorption of the borate from the photocharged surface of B-BiVO 4 or both. Deactivation of the bare BiVO 4 , without ac atalytic or passivating layer, has been widely observed under long-term PEC tests. [11b,35] Interestingly,when the process of borate treatment was repeated on B-BiVO 4 after 20 cycles of LSV scanning,similar PEC performances as that from afreshly-prepared B-BiVO 4 can be obtained again (Supporting Information, Figure S14). This self-recovery process can be repeated several times and projects borate treatment as apossible strategy to produce self-healing PEC cells,w hich can work during daytime and recover during the night (Supporting Information, Figure S15). Furthermore, modifying the B-BiVO 4 with co-catalyst can further increase its photocurrent density for water oxidation and dramatically improve the stability.These related studies are ongoing in our group.

Conclusion
In summary,w er eported the remarkable effect of modifying aBiVO 4 surface with borate by asimple immersion method, leading to as ignificant increase in photocurrent as well as ad ecrease in the onset potential for water oxidation, which is comparable to the effect of loading awater-oxidation co-catalyst. Detailed characterizations and carrier transfer kinetics investigations indicated that the adsorption of tetrahedral [B(OH) 4 ] À species near the active sites results in amolecular level modification. This acts as aregulating ligand and passivator,p laying an important role in accelerating water-oxidation rate and reducing charge trapping on the BiVO 4 surface.The post-synthetic borate treatment proposed in this work provides new opportunities to understand and improve the PEC performance of BiVO 4 photoanodes.T he method of small molecule modification can also be widely developed for improving the property of material-based catalysts and photocatalysts.