Electrode Preparation and Characterization

The Co@CB[5] supramolecular complex was assembled on the porous indium tin oxide (ITO) and porous BiVO₄ substrates to evaluate the electrochemical‐driven and PEC oxygen evolution activity of the proposed host–guest assembled devices, respectively. The porous ITO substrate was prepared by a doctor‐blade method (see the Supporting Information for details). The BiVO₄ thin film was prepared by modifying a method reported by Choi and co‐workers. ^[19] This porous BiVO₄ film has the advantages of improving electron‐hole separation and suppressing bulk carrier recombination, ^[12] while also providing a large specific surface area for loading the Co@CB[5] onto the electrode surface. The porous ITO substrate was composed of octahedron shape particles as indicated by scanning electron microscope (SEM) images (Figure S1a). A cross‐sectional view of the film showed that the porous ITO layer had an average thickness of approximately 1.3 μm (Figure S1b). The porous BiVO₄ film was composed of worm‐like particles as indicated by SEM images (Figure S2a). A cross‐sectional view of the film showed that the BiVO₄ layer had an average thickness of approximate 1 μm (Figure S2b). Energy dispersive spectroscopy (EDS) mapping analysis was performed on a selected cross‐sectional area to determine the elemental distribution (Figure S2c), which demonstrated a homogeneous distribution of Bi, V, and O elements in the film of the porous BiVO₄.

The preparation of Co@CB[5]/ITO and Co@CB[5]/BiVO₄ electrodes was briefly illustrated in Scheme 1. First, a solution of Co@CB[5] host–guest complex (1.0 mM) was prepared by mixing CB[5] and Co(NO₃)₂ in a molar ratio of 1:1. Mass‐to‐charge ratio (m/e) of 454.0879 corresponding to Co@CB[5]+H₃O (m/e 454.0979) could be found by electrospray mass spectrometry (Figure S3). Then, the porous ITO and BiVO₄ substrates were immersed in the solution of Co@CB[5] for 30 min, after washing with deionized water, resulting in the generation of Co@CB[5]/ITO and Co@CB[5]/BiVO₄ electrodes, respectively. The attenuated total reflection infrared (ATR‐IR) spectra of Co@CB[5]/ITO, CB[5]/ITO electrode and bare CB[5] powder were acquired using an infrared spectrometer with a resolution of 0.09 cm⁻¹ (Figure S4). Red‐shifts of ca. 9 and ca. 3 cm⁻¹ were observed for the carbonyl portal of the CB[5] component on Co@CB[5]/ITO and CB[5]/ITO electrode compared to that of CB[5] powder (1738 cm⁻¹), respectively. A similar red‐shift on the carbonyl portal of the CB[5] component on Co@CB[5]/BiVO₄ and CB[5]/BiVO₄ could also be observed (Figure S5). These ATR‐IR spectra suggested strong interaction between CB[5], Co cations and porous MO_x substrates, ^[20] and indicated that Co@CB[5] was immobilized on the surface of ITO and BiVO₄.

Scheme 1

The electrode preparation process by immobilizing Co@CB[5] on the surface of metal oxide substrates.

A time‐of‐flight secondary ion mass spectrometry (TOF‐SIMS) with Bi₃ ⁺ as the primary ion source was further employed to detect the presence of Co@CB[5] on the surface of ITO and BiVO₄. As shown in the SIMS images of Co@CB[5]/ITO and Co@CB[5]/BiVO₄ (Figures S6 and S7), some related secondary ions assigned to cobalt combined with the fragments of CB[5] could be observed uniformly distributing on the surface of electrodes, indicating that Co@CB[5] was successfully immobilized on the surface of ITO and BiVO₄.

The morphological features of Co@CB[5] functionalized electrodes were then investigated by SEM. As shown in Figures S8 and S9, no obvious morphological changes were observed after the immobilization of CB[5] and Co@CB[5] on the porous substrates, because the assembly process is a molecular level surface modification.

X‐ray photoelectron spectroscopy (XPS) measurements were performed to analyze the molecular level surface composition and electronic states of the Co@CB[5] complex functionalized MO_x substrates. The XPS survey spectra of Co@CB[5]/ITO, Co@CB[5]/BiVO₄ and reference electrodes are shown in Figures S10 and S11. In XPS high‐resolution spectra of CB[5]/ITO and Co@CB[5]/BiVO₄ (Figures 1 a,b, Figures 2 a,b), when CB[5] was immobilized on the surface of MO_x, the N 1s signal corresponding to the C‐N of CB[5] at a binding energy of 400.15 eV can be observed, which indicated that CB[5] can be successfully immobilized on the surface of porous MO_x substrates by the simple soaking method. When Co@CB[5] was immobilized on ITO or BiVO₄, clear signal of Co 2p were observed at the binding energy around 781 eV, which did not appear for the bare substrates and neither for the CB[5]/MO_x electrodes. Meanwhile, small offsets from the main signal of N 1s could be observed. The signal of Co 2p for Co@CB[5]/ITO was fitted by four peaks, namely Co 2p_3/2 at 781.0 eV, Co 2p_3/2 satellite at 786.3 eV, Co 2p_1/2 at 796.7 eV, and a Co 2p_1/2 satellite at 802.6 eV. The signal of Co 2p for Co@CB[5]/BiVO₄ was fitted into three peaks, Co 2p_3/2 at 780.7 eV, a Co 2p_3/2 satellite at 785.9 eV, and Co 2p_1/2 at 796.6 eV, (the Co 2p_1/2 satellite signal is not assigned here owing to its overlap with the BiVO₄ signal). Please note that the binding energies of Co 2p for Co@CB[5] complex were different from those of Co‐based oxides or hydroxides,[ ²¹ , ²² ] indicating a molecular nature of the Co@CB[5] on MO_x surfaces.

Figure 1

XPS spectra of ITO based electrodes in selected energy areas: a) N 1s, b) Co 2p. c) Energy‐dispersive X‐ray spectroscopy (EDS) maps of one Co@CB[5]/ITO particle. d) Atomic‐resolution HAADF image of one Co@CB[5]/ITO particle in selected small area.

Figure 2

High‐resolution XPS spectra of BiVO₄ based electrodes: a) N 1s and b) Co 2p. c) HRTEM image of the Co@CB[5]/BiVO₄ sample. d) HAADF‐STEM images of Co@CB[5]/BiVO₄ electrode and corresponding elemental mappings.

To further demonstrate that the Co@CB[5] is a molecular catalyst on MO_x, the surface and morphology properties of the Co@CB[5]/ITO and Co@CB[5]/BiVO₄ samples were studied at atomic level by spherical aberration corrected transmission electron microscope (ACTEM). As revealed by high‐angle annular dark‐field (HAADF) images in Figure S12, the Co@CB[5]/ITO nano‐particle exhibited octahedron shape with regular and clear edge. As shown in Figure 1 c, Energy‐dispersive X‐ray spectroscopy (EDS) mapping images of Co@CB[5]/ITO showed the presence of Co, N and C on the surface of ITO nano‐particle. Atomic‐scale HAADF images in the Figure 1 d revealed that there is no structural decomposition observed after the attachment of Co@CB[5] complex, meanwhile, no Co oxides or hydroxides could be observed on the terminations of ITO nano‐particle after the immobilization of Co@CB[5] complex. This is in consistence with the phenomena of molecular anchoring of Co@CB[5] complex on ITO surface. As the element atomic mass of tin and indium in bulk ITO nanoparticle is far larger than that of cobalt, therefore, it is hard to pinpoint the single cobalt molecular complex absorbed on the surface of ITO by comparing the difference in atomic resolution HAADF images. Figures S13 and S14 have shown atomic‐resolution TEM images and corresponding EDS mapping images of Co@CB[5]/ITO in different scales, which suggested that the cobalt components were uniformly distributed on the surface of ITO. In addition, according to the results of EDS analysis in selected area, the metal composition ratio of Co: In: Sn was estimated to be 1: 92: 7, this result was in accordance with the magnitude of surface mono‐molecular layer distribution for Co components. ^[23]

The interface morphology of Co@CB[5]/BiVO₄ was characterized by high‐resolution transmission electron microscope (HRTEM), as shown in Figure 2 c, there were no obvious metal oxide films or nanoparticles covered on the surface of Co@CB[5]/BiVO₄ nanoparticle. Unfortunately, when high energy electron beam was focused on BiVO₄, the surface of BiVO₄ was damaged, detailed structure information for the surface of Co@CB[5]/BiVO₄ could not be obtained by atomic resolution AC‐STEM. HAADF‐STEM and corresponding EDS mapping images of Co@CB[5]/BiVO₄ showed that elements of Co, N and C contents in the Co@CB[5] complex were uniformly distributed on the surface of the BiVO₄ nanoparticles (Figures 2 d and S15). All above characterizations suggested that the Co@CB[5] on the surface of MO_x is a molecular level immobilization.

Electrochemically Driven OER Catalysis by the Co@CB[5] Complex

The OER performance of Co@CB[5]/ITO was first evaluated by linear sweep voltammetry (LSV). LSV measurements were performed at 25 °C in a standard three‐electrode cell with 1.0 M borate buffer solution (pH 9.2) as the electrolyte. As shown in Figure 3 a, the ITO substrate and CB[5]/ITO electrode were silent for OER. In contrast, the anodic current density curve of the Co@CB[5]/ITO electrode rose sharply beyond a low onset potential of 1.05 V vs. normal hydrogen electrode (NHE), followed by a dramatic increase at more positive potentials. A current density of 1.0 mA cm⁻² was achieved at η=485 mV. For comparison, Co/ITO electrode was also prepared by immersing ITO in 1.0 mM Co²⁺ solution for 30 mins (the same immersion time as for Co@CB[5]/ITO). The Co/ITO electrode exhibited a much lower anodic current density than that of the Co@CB[5]/ITO electrode. Linear fitting of the Co@CB[5]/ITO electrode gives a Tafel slope of 59.5 mV dec⁻¹ (Figure 3 b). The steady‐state catalytic activities of Co@CB[5]/ITO were determined by chronopotentiometric measurements during electrolysis at constant current densities over 5 h (Figure 3 c). The Co@CB[5]/ITO electrode required only 1.23 V and 1.29 V vs. NHE to achieve the current density of 1.0 mA cm⁻² and 2.5 mA cm⁻², respectively. The overpotential requirement kept stable during the test, which indicated that the Co@CB[5]/ITO is stable for OER. The amount of electrochemically generated oxygen from the Co@CB[5]/ITO system was confirmed by gas chromatography (GC). When a total charge of 3.6 C passed through the electrode, 8.62 μmol of O₂ was detected by GC, leading to a Faraday efficiency of 92.6 % (Figure S16).

Figure 3

a) The LSV measurements of Co@CB[5]/ITO and comparison electrodes in 1.0 M borate buffer (pH 9.2) at a scan rate of 50 mV s⁻¹ with iR compensation. b) Tafel plots of Co@CB[5]/ITO. c) Chronopotentiometric measurement of OER at current density of 1.0 and 2.5 mA cm⁻² for 5 h without iR compensation. d) TOFs of Co@CB[5]/ITO based on the Co²⁺ loading (calculated according to electrochemical measurement and ICP‐OES) and current densities vs. various overpotentials.

The effective coverage of Co²⁺ on the surface of ITO was estimated as 4.54×10⁻⁹ mol cm⁻² according to the linear relationship between the peak current of Co^2+/3+ and the scan rate (Figure S17, Eq. S1 and Eq. S2), and the order of magnitude for such catalyst loading is consistent with that of molecules immobilized on the surface of porous metal oxide substrates. ^[24] This total loading amount of Co²⁺ was further confirmed by inductively coupled plasma optical emission spectrometer (ICP‐OES). The total Co²⁺ loading amount is 4.98×10⁻⁹ mol cm⁻², which was slightly larger than the redox active Co²⁺. The turn over frequencies (TOFs) of Co@CB[5]/ITO based on the Co²⁺ loading (redox active) current densities vs. various overpotentials were calculated based on LSV data (operated in a low scan rate of 10 mV s⁻¹) and Eq. S3. ^[25] The logarithm of the TOFs varied linearly on the applied overpotentials from 460 to 550 mV, as shown in Figure 3 d. A TOF of 0.3 s⁻¹ was achieved at η=460 mV, which reached 9.9 s⁻¹ at η=550 mV. Table S1 lists the performance of Co@CB[5]/ITO in the current work and previously reported anodes for electrocatalytic water oxidation, including molecular and inorganic material catalysts immobilized on conducting glass. This comparison shows that the Co@CB[5]/ITO electrode is one of the best OER anodes in terms of overpotential, current density, and TOF.

Photo‐electrochemical Water Oxidation by Co@CB[5]/BiVO₄

The PEC water oxidation performance of Co@CB[5]/BiVO₄ photoanode was measured by linear sweep voltammetry (LSV) with a scan rate of 10 mV s⁻¹. The absorption spectrum of BiVO₄ did not change after Co@CB[5] was immobilized on the surface of BiVO₄ (Figure S18), hence, the band gap of BiVO₄ and Co@CB[5]/BiVO₄ were the same.

As shown in Figure 4 a, the photocurrent density of the as‐prepared BiVO₄ at 1.23 V vs. RHE was approximate 1.8 mA cm⁻². In the presence of sulfite as hole scavenger, the photocurrent density of BiVO₄ increased to 5.4 mA cm⁻². These results are comparable with previously reported values, ^[26] which suggested that the quality of BiVO₄ film in our work is reliable. When Co@CB[5] was immobilized on the surface of BiVO₄, the photocurrent density of the Co@CB[5]/BiVO₄ electrode was measured to be 4.8 mA cm⁻² at 1.23 V vs. RHE without the hole scavenger sulfite, which is much greater than that of unmodified BiVO₄, CB[5]/BiVO₄ and Co/BiVO₄ electrodes. The CB[5]/BiVO₄ electrode exhibited an similar anodic photocurrent density curve than that of the pristine BiVO₄ electrode (Figure S19), indicating CB[5] itself can not affect the water oxidation of BiVO₄. After immersing BiVO₄ in Co²⁺ solution for 30 minutes, the photocurrent density of Co/BiVO₄ electrode at 1.23 V vs. RHE increased to 2.3 mA cm⁻², which is much lower than that of the Co@CB[5]/BiVO₄ electrode (4.8 mA cm⁻²). The applied bias photon to current efficiency (ABPE) was calculated from the corresponding LSV curve in Figure 4 a according to Eq. S4. The maximum ABPE of BiVO₄ was 0.32 % at 0.88 V vs. RHE. However, the maximum ABPE of Co@CB[5]/BiVO₄ was 1.79 % at 0.60 V vs. RHE (Figure 4 b), which is 5.5 times as high as that of unmodified BiVO₄. The incident photon to current efficiencies (IPCEs) of BiVO₄ and Co@CB[5]/BiVO₄ were measured and calculated according to Eq. S5. As shown in Figure 4 c, the IPCE values of BiVO₄ and Co@CB[5]/BiVO₄ were approximate 35 % and 88 % at 420 nm (1.23 V vs. RHE), respectively. The values of the ABPE and IPCE demonstrated the high photon‐to‐O₂ conversion efficiency of the host–guest complex assembled Co@CB[5]/BiVO₄ photoanode. On the basis of these measurements, the photocurrent densities were 4.85 mA cm⁻² at 1.23 V vs. RHE and 2.9 mA cm⁻² at 0.6 V vs. RHE for Co@CB[5]/BiVO₄, as estimated by integrating the IPCE curves over the AM 1.5G solar spectrum (Figure S20). These calculated values are in agreement with the measured data in Figure 4 a (4.8 mA cm⁻² at 1.23 V and 2.8 mA cm⁻² at 0.6 V).

Figure 4

a) LSV curves of Co@CB[5]/BiVO₄, Co/BiVO₄, BiVO₄, and BiVO₄ with the hole scavenger Na₂SO₃ in 1 M borate buffer (scan rate, 10 mV s⁻¹). b) ABPEs of Co@CB[5]/BiVO₄ and BiVO₄ calculated from LSV curves. c) IPCEs of Co@CB[5]/BiVO₄ at 1.23 V and 0.6 V vs. RHE, and BiVO₄ at 1.23 V vs. RHE. d) Light response of Co@CB[5]/BiVO₄ and BiVO₄ photoanodes under chopped irradiation at a constant bias of 0.6 V vs. RHE.

Under chopped‐light illumination, the bare BiVO₄ only generated a low photocurrent density less than 0.25 mA cm⁻² at a constant applied potential of 0.6 V vs. RHE; however, the Co@CB[5]/BiVO₄ photoanode generated a photocurrent density of ca. 2.4 mA cm⁻², underlining the indispensable role of Co@CB[5] in water splitting (Figure 4 d). During long‐term photolysis at 0.6 V vs. RHE under continued visible‐light irradiation, the magnitude of the photocurrent generated by Co@CB[5]/BiVO₄ slowly decreased from 2.5 to 2.3 mA cm⁻² over 30 min (Figure S21). The photogenerated O₂ in the headspace was quantified by GC. Assuming a 4 e⁻ process for O₂ evolution, the number of electrons passing through the electrode agreed well with the amount of O₂ detected, representing a Faradaic efficiency of 90.3 % for Co@CB[5]/BiVO₄ photoanode (Figure S22).

The photocurrent density at 1.23 V vs. RHE and the maximum ABPE obtained from Co@CB[5]/BiVO₄ are compared to those previously reported systems and the results are shown in Table S2. The performance of Co@CB[5]/BiVO₄ exceeded those of many catalyst‐modified undoped‐BiVO₄ photoanodes in both parameters. The host–guest complex assembled Co@CB[5]/BiVO₄ photoanode can avoid complicated organic synthesis, and the high activity of Co@CB[5]/BiVO₄ holds potential for supramolecular catalysts capable of replacing state‐of‐the‐art metal oxides and molecular catalysts.

Post‐characterization of the Electrode after the OER Test

Post‐characterization of the electrode after the OER test is necessary for further proof of reaction mechanism and the stability of the as‐fabricated hybrid electrodes. As shown in ATR‐IR spectra (Figure S23), Co@CB[5] maintained on the surface of electrodes after the OER test. As shown in Figures S24 and S25, the SEM images revealed that the Co@CB[5]/ITO and Co@CB[5]/BiVO₄ after OER test maintained a porous surface without significant changes. The composition and electronic states of the Co@CB[5]/ITO and Co@CB[5]/BiVO₄ samples after OER test were analyzed by XPS technique (Figures S26–S29). The binding energies of both N 1s and Co 2p for Co@CB[5]/ITO and Co@CB[5]/BiVO₄ were consistent with that of before OER test, indicating that Co@CB[5] was stable for OER. The EDS elemental mapping images with different magnitudes in Figures S30a, S31 and S32 confirmed the uniform distribution of Co, N and C in Co@CB[5]/ITO following the OER tests. Figure S30b showed the atomic‐resolution TEM images of tested Co@CB[5]/ITO particle. By further observing the edge of sample under greater magnification, a clear and well‐ordered surface could be observed of tested Co@CB[5]/ITO, indicating that there was no other amorphous layer or heterojunction film formed to covered ITO nano‐particle surface during the OER process (Figure S30c,d). In addition, according to the results of EDS analysis in selected area, the metal composition ratio of Co: In: Sn for tested Co@CB[5]/ITO particle was calculated to be 1: 91: 8, which was almost the same as the ratio of pristine Co@CB[5]/ITO. The EDS elemental mapping images with different magnitudes in Figure S33 and S34 confirmed the uniform distribution of Co, N and C in Co@CB[5]/BiVO₄ after the OER tests. Figure S35 showed the HRTEM images of the tested Co@CB[5]/BiVO₄ particles, there are no nanoparticles generated on the surface of Co@CB[5]/BiVO₄ during the OER process, indicating that Co@CB[5] was stable on the surface of BiVO₄ during the OER. By integrating the information obtained from the above characterizations, it can be convincingly concluded that Co@CB[5] maintains its supramolecular nature and is stable during the OER.

Kinetic Studies of Co@CB[5] for Water Oxidation

To investigate the transfer of carriers in the PEC system constructed by this novel Co@CB[5] supramolecular catalyst, kinetic studies were carried out. With sodium sulfite (Na₂SO₃) as a hole scavenger, extremely fast oxidation kinetics and negligible surface recombination were expected, as shown in Figure S36, the charge transfer efficiency (η_trans) at the electrode‐electrolyte interface was calculated according to Eq. S6 and the data from Figure 4 a. Using this method, a superior η_trans of Co@CB[5]/BiVO₄ was achieved (88 %), compared with that of bare BiVO₄ (33 %) at 1.23 V vs. RHE. This result suggests that the Co@CB[5]/BiVO₄ electrode has more rapid water oxidation kinetics than the bare BiVO₄. ^[27]

Electrochemical impedance spectroscopy (EIS) was performed to examine the charge transport phenomena in the bulk and at the surface of the electrodes. As shown in Figure 5 a, semicircles for Co@CB[5]/BiVO₄ and BiVO₄ electrodes fitted well with Randles equivalent circuit model. Charge transport at the photoanode‐electrolyte interface was estimated with the values for the charge‐transfer resistance (R_ct) obtained from the diameter of the semicircles. A smaller semicircle indicates better charge transfer of the corresponding photoanode at the interface. Notably, the Co@CB[5]/BiVO₄ electrode had a markedly smaller R_ct (96 Ω) value than that of bare BiVO₄ (303 Ω) under illumination at 0.6 V vs. RHE, illustrating a better ability to transfer holes at the electrode‐electrolyte interface of Co@CB[5]/BiVO₄, which was consistent with its high charge transfer efficiency (η_trans).

Figure 5

a) EIS of BiVO₄ and Co@CB[5]/BiVO₄ photoanodes measured under 0.6 V vs. RHE and AM 1.5G (100 mW cm⁻²) in 1.0 M borate buffer (pH 9.2). b) LSV curves of Co@CB[5]/ITO in anhydrous sodium borate H₂O and D₂O solutions, the inset exhibits the KIEs against potential. c),d) LSV curves of Co@CB[5]/BiVO₄ (c) and BiVO₄ (d) in anhydrous sodium borate H₂O and D₂O solutions under light irradiation, the inset exhibits the corresponding KIEs against potential.

The charge transfer efficiency depends on the rate of water oxidization catalytic processes on the surface of BiVO₄ and the rate of surface charge recombination. The deuterium kinetic isotope effects (KIEs) can reflect the proton transfer kinetic information on water oxidation reactions and help us to interpret the rate determine step (RDS) of the catalytic process. ^[28] The KIEs(H/D) can be defined as Eq. S7, the current densities in water and deuterated water electrolytes were contrasted at the same overpotential according to Eq. S8–Eq. S10 for KIEs(H/D) measurements in this work. ^[29]

As shown in Figure 5 b, the value of KIEs of Co@CB[5]/ITO electrode was larger than 2, indicating a primary KIEs for which O−H bond cleavage is involved in the RDS of electro‐driven water oxidation for Co@CB[5], corresponding to the water nucleophilic attack mechanisms. ^[28] When the Co@CB[5] was anchored on the surface of BiVO₄, the value of KIEs for Co@CB[5]/BiVO₄ photoanode was around 1.0 with the applied bias increasing, indicating the O−H bond cleavage is not involved in the RDS for light driven water oxidation by Co@CB[5]/BiVO₄ photoanode (Figure 5 c). In contrast, the value of KIEs for bare BiVO₄ photoanode was around 1.5 (Figure 5 d), suggesting that the proton transfer process was involved in the RDS of the light‐driven water oxidation process for bare BiVO₄. Thereby, the immobilization of Co@CB[5] on the surface of BiVO₄ could greatly accelerate the proton transfer processes (water oxidation) and shift the RDS of BiVO₄ photoanode to a non proton transfer involved process. Thence, Co@CB[5] indeed played an important role in the light‐driven catalytic water oxidation processes in this Co@CB[5]/BiVO₄ supramolecular assembly photoanode.

It has been known that a co‐catalyst can serve as passivation layer to improve charge‐separation and charge transfer processes across semiconductor‐liquid interfaces. ^[30] For example, the modification of CoPi on the surface of semiconductors can improve the performances of photoanodes, which primarily caused by reducing the surface charge recombination (passivation layer), rather than improving the catalytic ability (charge transfer kinetics).[ ³¹ , ³² ] Similar phenomena have also been observed in other systems, such as a thin layer of CoFeO_x modified hematite and BiFeO₃ passivated BiVO₄ photoanodes.[ ³³ , ³⁴ ] Converse systems have also been reported, for instance, thin layers of FeOOH can enhance the performance of BiVO₄ for water oxidation, for which the performance was improved primarily owing to the improvement of charge transfer, rather than suppressing the surface charge recombination.¹¹ Molecular‐based catalysts can also improve the light‐driven water oxidation ability of n‐type semiconductors, such as Co‐salophen complexes and Ni₄O₄ cubane have been reported as the co‐catalyst for BiVO₄,[ ³⁵ , ³⁶ ] in these cases, molecular catalysts can accelerate the water oxidation reaction as OER catalysts, meanwhile, reduce the surface charge recombination.

Our results confirmed that Co@CB[5] can effectively promote BiVO₄ for light‐driven water oxidation by serving as a good OER catalyst, however, it is necessary to further confirm whether Co@CB[5] on the semiconductor inhibited surface charge recombination by acting as a surface passivation layer. The charge transfer and surface recombination kinetics were quantified by intensity modulated photocurrent spectroscopy (IMPS) to understand the real role of Co@CB[5] on the BiVO₄. Figure 6 a and 6 b show typical IMPS responses of bare BiVO₄ and Co@CB[5]/BiVO₄ photoanodes in the complex plane. The IMPS spectrum consists of two semicircles in the 4th and 1st quadrants, which correspond to the resistance‐capacitance attenuation and the competition between charge transfer and recombination, respectively. ^[34] Accordingly, the frequency of the maximum imaginary corresponds to the sum of the charge transfer (k _trans) and charge recombination (k _rec) rate constants (k _trans+k _rec=2πf _max). The low frequency intercept in the normalized form, at which the imaginary part is equal to 0, corresponds to a charge transfer efficiency of k _trans/(k _trans+k _rec). ^[37] The key parameters k _rec and k _trans are therefore readily accessible.

Figure 6

a) IMPS spectra of BiVO₄ photoanode at various potentials. b) IMPS spectra of Co@CB[5]/BiVO₄ photoanode at various potentials. c) Rate constant for charge recombination, k _rec extracted from IMPS spectra. d) Rate constant for charge transfer, k _trans extracted from IMPS spectra.

Moreover, the upper semicircle of the BiVO₄ photoanode did not change notably as the applied potential was increased, the value of k _rec remained nearly constant with increasing potential for bare BiVO₄ (Figure 6 c), indicating that a high charge recombination exists for the BiVO₄ photoanode over a wide potential range. By contrast, the upper semicircle of the Co@CB[5]/BiVO₄ photoanode became smaller as the applied potential increased. Notably, k _rec was suppressed by a factor of 6.5–30 over the entire potential range after decoration of Co@CB[5]. For instance, at 0.6 V vs. RHE (the max point of ABPEs), the value of k _rec was 3.5 s⁻¹ for bare BiVO_4, which is 6.5 times as large as that of Co@CB[5]/BiVO₄ (0.54 s⁻¹). At higher potentials, this factor increased to 30 at 0.8 V vs. RHE. These results suggest that the charge recombination was markedly suppressed by the Co@CB[5]. Figure 6 d shows that the k _trans of Co@CB[5]/BiVO₄ surpassed that of BiVO₄ at all potentials. Particularly, at a potential of 0.6 V vs. RHE (the max point of ABPEs), the k _trans of Co@CB[5]/BiVO₄ was more than 4 times as great as that of BiVO₄. The charge transfer efficiency, defined as k _trans/(k _trans+k _rec), showed the same trend as the value determined by J _water/J _sulfite, indicating the reliability of our measurements (Figure S37). The ratio of k _rec /k _trans was positively proportional to the semicircle in 1st quadrant and a small value indicated faster charge transfer than charge recombination. ^[34] These results demonstrate that Co@CB[5] behaves as a molecular catalyst, which is in contrast to the mere passivation action of heterogeneous catalysts as reported,[ ³¹ , ³² , ³³ , ³⁴ ] Co@CB[5] not only accelerates the water oxidation reaction as a good OER catalyst for BiVO₄, but also reduces the surface charge recombination.

Based on the OER kinetics analyzed above, the function of Co@CB[5] on BiVO₄ was summarized and shown in Scheme 2. For the bare BiVO₄, the photo generated holes can be directly consumed for water oxidation (pathway 1), or recombined with electrons that trapped by surface state (pathway 2). Due to the low intrinsic OER catalytic activity of BiVO₄, the injection of holes to electrolyte was limited by the proton transfer involved water oxidation reaction, leading to a primary KIEs for bare BiVO₄. When Co@CB[5] was immobilized on the surface of BiVO₄, owing to the high OER catalytic activity of Co@CB[5], the photo‐generated holes could be effectively transferred and consumed for water oxidation (pathway 3); meanwhile, due to the faster consumption of holes, the photo‐generated electrons will have less chance to recombine with photo‐generated holes at the surface of BiVO₄, leading to the higher charge transfer efficiency of Co@CB[5]/BiVO₄ in comparison to that of the bare BiVO₄.

Scheme 2

Simplified elementary processes in a) bare BiVO₄ and b) Co@CB[5]/BiVO₄ photoanode. The movement of holes that were generated from photo‐induced charge separation have three pathways at the semiconductor‐electrolyte interface (solid lines: majority events; dotted lines: minority events): 1) direct water oxidation with low k _trans, 2) surface charge recombination, and 3) charge transfer via Co@CB[5] with high k _trans.