Introduction

Photocatalytic processes are one of the most attractive ways to utilize solar energy, which is of great significance to solve the current energy crisis and environmental pollution. Constructing composites with heterojunction structure has proved to be an effective means to increase the utilization of sunlight and prolong the carrier lifetime of photocatalysts, which ultimately significantly improves the photocatalyst efficiency1,2,3,4. Several high-performance heterojunction photocatalysts have been reported, exemplified by WO3/BiVO45, BiOBr/CuInS2/WO36 and Bi2MoO6/g-C3N4/Ag2MoO47, which demonstrate enhanced photocatalytic performance through optimized charge separation. Among various heterojunctions, TiO2/BiVO4 has emerged as a particularly promising system by synergistically addressing the inherent limitations of its components: while TiO2 suffers from poor visible-light absorption and rapid carrier recombination8, recent advances in defect engineering (e.g., oxygen vacancy-modulated TiO2) have enhanced its photoresponse9; Meanwhile, the short carrier lifetime and diffusion length of BiVO4 have been improved through nanostructuring and doping strategies (e.g., Al-doped BiVO4 achieves nearly 3× longer diffusion length and carrier lifetime)10. The TiO2/BiVO4 heterojunctions combine TiO2’s ideal conduction band edge, exceptional chemical stability, and biocompatibility with BiVO4’s visible-light harvesting (Eg ≈ 2.4 eV), overcoming shortcomings of both materials through interfacial charge engineering11,12,13,14.

Fig. 1
figure 1

The energy barrier at the interface is unfavorable to carrier transfer. Schematic diagram of carrier transfer in heterojunctions with (a) Type II and (b) Type I energy level structures under visible light irradiation.

Full size image

Interfacial charge transfer has been proved to critically determine the photocatalytic performance of TiO2/BiVO4 heterojunctions. In conventional Type-II configurations (Fig. 1a), the staggered band alignment drives electrons to migrate from BiVO4 to TiO2 while holes remain in BiVO4, resulting in spatial separation of photogenerated electrons and holes for efficient redox reactions13. For example, these separated carriers can directly participate in pollutant degradation: holes on the valence band (VB) of BiVO4 can oxidize organic pollutants (e.g., MO and MB) through •OH radical formation or directly decompose them via hole-transfer pathways, while electrons on the conduction band (CB) of TiO2 can reduce O2 to O2•− for further oxidative chain15– 16. Specifically, Feng and Fu et al. observed long-lived charge-separated states in Type-I TiO2/BiVO4, where photoexcited high-energy electrons from BiVO4 can possess sufficient energy to overcome the narrow potential barrier between TiO217. Despite significant advances in TiO2/BiVO4 heterojunction design, the charge dynamics under visible-light excitation, particularly how carriers overcome the large interfacial energy barriers in Type-I heterojunctions (Fig. 1b) and the influence of TiO2’s mid-gap states, remain poorly understood. Addressing this gap is fundamentally important, as mechanistic insights into TiO2/BiVO4’s charge transfer could directly guide interface engineering to enhance both visible-light absorption and carrier separation efficiency, which are the dual requirements of high-performance photocatalysis.

Fig. 2
figure 2

Graphical abstract: Enabling superior visible-light photocatalytic degradation of methyl orange in ‘inefficient’ Type-I TiO2/BiVO4 heterojunction via polaron engineering.

Full size image

In this work, we synthesized a TiO2/BiVO4 heterojunction with Type-I band structure and explored the mechanisms of carrier transfer, recombination and photocatalysis under visible light irradiation with phonon energy lower than the band gap of TiO2. Efficient charge migration pathway mediated by the localized polaronic mid-gap states of TiO2 was confirmed by structural and time-resolved spectroscopy results, which can effectively separate the photogenerated carriers of BiVO4 and TiO2, and eventually lead to the substantial improvement on the visible photocatalytic performance of TiO2/BiVO4 relative to pure BiVO4 and TiO2 (Fig. 2). Long-lived photogenerated holes in TiO2 and electrons in BiVO4 were both observed, and the degradation rate of methyl orange (MO) reached nearly 100% under weak visible light irradiation. This work broadens the feasible way to synthesize TiO2/BiVO4 photocatalysts for efficient use of solar energy and provides a new idea for improving the visible light utilization rate of TiO2.

Methods

All of the materials used in the experiments were purchased from Inno-chem and used without further purification.

Synthesis of BiVO4

The BiVO4 nanocatalysts were synthesized using a simple coupling method. In a typical procedure, 0.02 mol Bi(NO3)3·5H2O was dissolved in 30 mL HNO3 (4 mol/L). The mixture was denoted as solution A. 0.01 mol of NH4VO3 was dissolved in 30 mL of NaOH solution (1 mol/L) (solution B). Subsequently, solution A was then added dropwise to solution B with the addition of 1 g ethyl cellulose. Followed by the pH adjustment with an NaOH solution (2 mol/L), until the mixture reached a pH of 6. After stirring for 1 h, the mixture was transferred into a Teflon-lined stainless steel autoclave with a capacity of 100 mL. Then, the sealed reactor was heated to 180 ◦C for 8 h. The final products were filtered and washed with deionized water and ethanol several times, and finally dried at 100 °C for 8 h.

Synthesis of TiO2

A mixture of 2 mL HNO3 and 6 mL ethyl alcohol was added dropwise to the mixture of 20 mL butyl titanate and 30 mL ethyl alcohol with sustained stirring for 12 h to form a highly dispersed colloidal solution. Subsequently, the colloid was dried overnight in the oven and then was calcined at 500 °C for 2 h.

Synthesis of TiO2/BiVO4

The procedure for the synthesis of the nanocomposites was conducted according to the BiVO4 synthesis protocol. A certain amount of as-prepared TiO2 was added into the precursor solution of bismuth vanadate prior to the pH adjustment. The molar ratio of Ti and Bi was fixed to 1 : 0.06.

Femtosecond transient absorption measurement

The femtosecond transient absorption setup used for this study is based on a regenerative amplified Ti: sapphire laser system from Coherent (800 nm, 35 fs, 6 mJ pulse− 1, and 1 kHz repetition rate), nonlinear frequency mixing techniques and the Femto-TA100 spectrometer (Time-Tech Spectra). Briefly, the 800 nm output pulse from the regenerative amplifier was split into two parts with a 50% beam splitter. The transmitted part was used to pump a TOPAS Optical Parametric Amplifier (OPA) which generates a wavelength tunable laser pulse from 250 nm to 2.5 μm as a pump beam. The reflected 800 nm beam was split again into two parts. One part with less than 10% was attenuated with a neutral density filter and focused into a 2 mm thick sapphire window to generate a white light continuum (WLC) from 420 nm to 800 nm used for the probe beam. The probe beam was focused with an Al parabolic reflector onto the sample. After the sample, the probe beam was collimated and then focused into a fiber-coupled spectrometer with CMOS sensors and detected at a frequency of 1 kHz. The delay between the pump and probe pulses was controlled by a motorized delay stage. The pump pulses were chopped by a synchronized chopper at 500 Hz and the absorbance change was calculated with two adjacent probe pulses (pump-blocked and pump-unblocked). All experiments were performed at room temperature.

PL spectra and steady-state absorption measurements

PL spectra were obtained by using F-4700 Hitachi fluorescence spectrophotometer. Time-resolved PL spectra were recorded by FLS980 multifunction steady-state and transient-state fluorescence spectrometer. The UV-vis absorption spectra were recorded using a UV-Vis spectrometer (UV-3700, Shimadzu, Japan) with BaSO4 as a reference.

Structure characterization

The X-ray diffraction (XRD) patterns were obtained on the D8 Advance X-ray diffractometer and equipped with a Cu-Kα radiation. The data were expressed in the 2θ range from 20 ° to 80 °.

X-ray photoelectron spectra (XPS) were acquired on the Thermo Scientific escalab 250xi system with Al Kα (1486.6 eV) as the excitation source.

The morphology of the samples was explored using a Hitachi S4800 scanning electron microscope (SEM) (accelerating voltage 5 kV). The Energy Dispersive Spectra (EDS) were measured on HORIBA EX-250 instrument.

TEM images were obtained on a Talos F200X transmission electron microscope.

Electrochemical measurements

Electrochemical impedance and photocurrent measurements were conducted with a CHI660e electrochemical station in a typical three electrode cell, using a Pt sheet as the counter electrode and a saturated Ag/AgCl as the reference electrode. The working electrode was prepared by spreading the ethanolic slurry of photocatalyst over ITO plate with an area of 3.5 cm2. The suspension was prepared by dispersing the photocatalyst (20 mg) and ethyl cellulose (2 mg) into a mixed solution containing 1 mL ethanol. A 300 W Xe lamp with a λ > 420 nm optical filter was utilized as the light source and the electrolyte was an aqueous solution of Na2SO4 (0.5 M).

Photocatalytic test

The photocatalysis experiment was processed in multichannel photochemical reaction system (CEL-LB70, Ceaulight) with 300 mW/cm2 Xe lamp light source. The photocatalysis of visible light was carried out with a λ > 420 nm optical filter. The photocatalytic activity was studied by preparing 20 mg/L methyl orange (MO) as the experimental mode. 0.1 g of sample was added to 100 mL MO solution with continuous stirring. Before the photocatalytic experiment, the reaction system was treated in the dark for 30 min to achieve the adsorption-desorption equilibrium. After the light reaction for a certain time, 5 mL suspension was taken and separated by centrifugation. The concentration of MO was determined by spectrometer.

Results and discussion

Characterization

Herein, nanocomposites were synthesized using modified methods as previously reported (see Methods for the details)18,19. Powder X-ray diffraction (XRD) was employed to study the phase and crystallographic features of synthesized samples. The XRD diffraction patterns for TiO2/BiVO4, pure TiO2 and BiVO4 are shown and compared in Fig. 3a. The diffraction peaks of the BiVO4 powder are in great agreement with those of the monoclinic phase, while the diffraction peaks of the prepared TiO2 corresponding to an anatase are clearly observed. The TiO2/BiVO4 exhibits both the characteristic peaks of BiVO4 and TiO2, indicating the successful preparation of TiO2/BiVO4. The diffraction peaks at 2θ values of 25.56 o, 48.26 o, 54.08 o and 62.87 o of the TiO2/BiVO4 correspond to the diffraction related to the anatase structure, with the planes (101), (200), (105) and (213). Meanwhile, the 2θ values of 24.63 o, 30.74 o, 32.93 o, 34.54 o and 48.28 o can be attributed to diffraction related to the planes (200), (211), (112), (220) and (312) of BiVO4. In addition, the diffraction peaks with high intensity and narrow linewidth indicate that the prepared samples have good crystalline features.

The microstructures and microscopic morphologies of the as-prepared crystals were characterized by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). As confirmed by SEM images shown in Fig. 3b and Fig. S1, the irregular TiO2-spherical particles (Fig. 3b and S1a) are attached to the BiVO4-polyhedron (Fig. 3b and Fig. S1b) to form TiO2/BiVO4 compounds. The TEM image of the TiO2/BiVO4 composite shown in Fig. 3c provides the similar morphological feature as previous SEM images. Two different kinds of lattice fringes and the clear contact interface can be observed on the area selected high-resolution TEM (HRTEM) image exhibited in Fig. 3d, the facet distances of 0.148 nm and 0.238 nm correspond to the (213) crystallographic plane of TiO2 and the (220) lattice plane of BiVO4, respectively, which confirms the intimate interfacial contact of TiO2 and BiVO4 and the formation of heterojunction structure in TiO2/BiVO4. Furthermore, the elemental mapping of TiO2/BiVO4 heterostructures also suggests that all the elements (Bi, O, V and Ti) are uniformly distributed over the entire composite (Fig. 3e), confirming the successful synthesis of TiO2/BiVO4 again.

Fig. 3
figure 3

The TiO2/BiVO4 composite heterojunction was successfully synthesized. (a) XRD patterns of BiVO4, TiO2 and TiO2/BiVO4. (b) SEM image of TiO2/BiVO4, the spherical TiO2 particles are deposited on the BiVO4 polyhedron, indicating the successful synthesis of the TiO2/BiVO4 composite. (c) TEM and (d) High-resolution TEM (HRTEM) image of the selected area (the red square in Fig. 3c) of a typical TiO2/BiVO4. (e) The elemental mapping diagrams of TiO2/BiVO4.

Full size image

We then evaluated the photocatalytic performances of the BiVO4, TiO2 and TiO2/BiVO4 in terms of the photocatalytic degradation of methyl orange (MO) under visible light irradiation. Figure 4a shows the evolution of methyl orange (MO) concentration over time in the presence of different photocatalysts. Pure TiO2 exhibits almost no photocatalytic activity for methyl orange (MO) degradation under visible light irradiation, while the photocatalytic MO degradation rate of pure BiVO4 is ~ 64%. Besides, the photocatalytic activity of TiO2/BiVO4 composite is significantly improved and the photocatalytic degradation rate of methyl orange reaches nearly 100% under the same condition. The UV–vis absorption spectra of BiVO4, TiO2 and TiO2/BiVO4 under the same conditions are firstly compared in Fig. 4b to judge the influence of light absorption capacity on the improvement of photocatalytic performance. The TiO2/BiVO4 exhibits a decrease of photon absorption relative to pure BiVO4 in the visible light region, which is contrary to the results of photocatalysis, indicating that the improvement in photocatalysis performance is not due to the difference in light absorption capacity.

Meanwhile, electrochemical impedance spectroscopy (EIS) and photocurrent measurements were also performed to evaluate the charge transfer capabilities of the composites. As depicted in Fig. 4c, TiO2/BiVO4 exhibits the smallest semicircle capacitance compared to pure TiO2 and BiVO4, which implies a reduced charge-transfer resistance and the fastest interfacial charge transfer. The I-t curve of TiO2/BiVO4 also shows the largest photocurrent for four successive light-on and off cycles, while pure TiO2 has almost no response under the same condition, indicating the highest carrier separation efficiency of BiVO4 in TiO2/BiVO4 compared to the single component (Fig. 4d). The above results lead us to speculate that the more efficient charge separation and transfer should be the major mechanism for the enhanced photocatalytic performance of TiO2/BiVO4.

Fig. 4
figure 4

(a) Photocatalytic degradation of MO over pure TiO2, BiVO4 and TiO2/BiVO4. (b) Comparison of steady-state UV-vis absorption of TiO2, BiVO4 and TiO2/BiVO4. (c) Electrochemical impedance spectroscopy (EIS) Nyquist plots and (d) Transient photocurrent response under visible light irradiation of TiO2, BiVO4 and TiO2/BiVO4.

Full size image

The determination of localized polaronic mid-gap states

The band structure closely related to the charge transfer process is shown in Fig. 5a, the synthesized TiO2/BiVO4 nanocomposites exhibit a type I band alignment, where both the electron (ET) and hole (HT) transfer from BiVO4 to TiO2 are indeed unfavorable due to the energy barrier that exists at the interface. According to the Tauc plots (Figure S2a) and the Mott-Schottky tests (Figure S2b), the optical band gaps (Eg) were determined to be 3.34 eV for TiO2 and 2.47 eV for BiVO4, while the conduction bands (CB) are estimated to be -0.69 eV for TiO2 and − 0.61 eV for BiVO4 (vs. NHE) based on Eq. (1).

$$E_{{NHE}} = E_{{Ag/AgCl}} + {\text{ }}0.{\text{197}}$$
(1)

In this energy level arrangement, there should be almost no charge transfer between BiVO4 and TiO2 under our experimental conditions, since the white light source we used (λ ≥ 420 nm, Ephoton ≤ 2.95 eV) can only excite electrons in BiVO4 to the CB, but not TiO2, which is inconsistent with the phenomena we observed in photocatalysis, ESI and photocurrent experiments. Hence, it is reasonable to assume that there might be another carrier transport and recombination mechanism in TiO2/BiVO4.

Previous reports suggest that charge carriers produced by surface traps (O vacancies, etc.) can introduce localized polaronic mid-gap states in TiO2, i.e., excess electrons in the Ti sublattice near the O vacancy produce reduced Ti3+ centers, thus facilitating the formation of small polaron20,21,22,23,24. These localized small-polaronic (SP) states are estimated to be more than one eV above the VB edge of TiO2, which can not only enhance the visible light absorption of TiO2 but also induce different carrier traveling, trapping and recombination pathways. Moreover, the presence and increased density of O-vacancy in the synthesized TiO2/BiVO4, as well as the interaction between TiO2 and BiVO4, were affirmed by the XPS analysis. Thus, we suspected that it might be necessary to consider the influence of O-vacancy induced localized SP states.

Fig. 5
figure 5

The synthesized TiO2/BiVO4 nanocomposites exhibit a type I band alignment and increased O-vacancy density. (a) The energy diagram of TiO2/BiVO4. (b) The survey spectra of pure TiO2, BiVO4 and TiO2/BiVO4. The high-resolution XPS profiles of (c) Bi 4f, (d) V 2p, (e) O 1s and (f) Ti 2p, in TiO2, BiVO4 and TiO2/BiVO4.

Full size image

The survey spectra of pure TiO2 and BiVO4 exhibit Ti, O and Bi, V, O elements, respectively, while TiO2/BiVO4 contains the constituent elements of both TiO2 and BiVO4 (Fig. 5b). The high-resolution Bi 4f spectra exhibit two feature peaks around ~ 164.1 eV (Bi 4f5/2) and ~ 158.8 eV (Bi 4f7/2) in both BiVO4 and TiO2/BiVO4 (Fig. 5c), indicating that Bi3+ is the main bismuth species in both samples25. In Fig. 5d, two peaks centered at the binding energy of ~ 516.4 eV (V 2p3/2) and ~ 523.8 eV (V 2p1/2) show a slight blue shift in TiO2/BiVO4 compared with BiVO4, demonstrating the existence of the characteristic V5+ in both compounds and the interfacial interactions between TiO2 and BiVO4 in TiO2/BiVO4 composite4. Furthermore, the O 1s spectra of both pure BiVO4 and TiO2 can be deconvolved into two separate peaks (Fig. 5e), where peaks at low binding energy (529.5 eV, 529.8 eV) can be assigned to the O2− species in the lattice (Olattice) and the high binding energy peaks (530.6 eV, 532.4 eV) can be corresponded to hydroxyl groups bound to metal cations in the oxygen-deficient region (Ov)26. In the case of TiO2/BiVO4, the binding energy of Olattice is ~ 529.6 eV, which is located between BiVO4 and TiO2 and suggests the change of lattice environment. Meanwhile, the binding energy of Ov is close to pure TiO2, along with a slight increase in intensity, illustrating that the denser O-vacancy mainly exists in the TiO2 component for TiO2/BiVO4 composite.

The effect of O-vacancy can also be reflected by the XPS spectra of Ti 3d, as the creation of Ti3+ centers is closely related to the O-vacancies27. As shown in Fig. 5f, two peaks at 458.6 eV and 464.3 eV are observed for pure TiO2, which represent Ti 2p3/2 and 2p1/2 for Ti4+ in TiO2, respectively28. Along with these two peaks, a weak shoulder peak attributed to the Ti3+ center emerges at 456.8 eV29. All deconvoluted Ti 2p spectra show a shift of binding energy in TiO2/BiVO4, validating the formation of heterojunction structure. Notably, the shoulder peak assigned to Ti3+ 2p3/2 (~ 457.2 eV) becomes more intense and another new distinct peak belonging to Ti3+ 2p1/2 appears at 464.2 eV after combining with BiVO4 (Fig. 5f, lower panel), indicating the formation of Ti3+ is of greater preference in TiO2/BiVO4 and is consistent with the apparent Ov peak shown in the O 1s XPS spectra of TiO2/BiVO4, which further confirms the existence and higher density of O-vacancy in TiO2/BiVO4, and suggests the possible existence of small-polaronic mid-gap states.

Fig. 6
figure 6

The presence of mid-gap states was further proved by PL spectra. (a) PL spectra of TiO2, BiVO4 and TiO2/BiVO4 under 379 nm excitation. (b) Comparison of PL spectra of the TiO2, BiVO4 and TiO2/BiVO4 with a 430 nm laser illumination. Inset: An enlarged view of the spectrum in the virtual box region. (c) Speculated charge transfer in TiO2/BiVO4 under visible light irradiation.

Full size image

Excitation wavelength-dependent steady-state PL spectra were then measured to further prove the presence of mid-gap states and determine their approximate positions. As shown in Fig. 6a, TiO2/BiVO4 composites show a strong fluorescence emission at 467 nm and a shoulder peak centered at 526 nm under 379 nm laser excitation, which can be attributed to the carrier recombination of the VB and CB of TiO2 and BiVO4, respectively. However, there has been a noticeable change in the PL spectra of TiO2/BiVO4 when excited by photons with energies lower than the band gap of TiO2 (Fig. 6b), and only one peak centered at 522 nm appears. The origin of this fluorescence peak can be obtained by comparing it with the PL spectra of pure BiVO4 and TiO2 under the same conditions. As shown in Fig. 6b, the PL spectrum of pure BiVO4 is not significantly affected by changing the excitation light, and the center wavelength is still 532 nm. However, it’s worth noting that pure TiO2 exhibits a weak fluorescence centered at 517 nm, which can be well explained by the recombination of the excited SP states (inset of Fig. 6b). Therefore, it is reasonable to assume that this low-energy PL emission of TiO2/BiVO4 around 522 nm originates from the fluorescence superposition of BiVO4 and SP states, while the shift of the central wavelength and the attenuated intensity compared to BiVO4 reflect the interaction and charge transfer between TiO2/BiVO4. Moreover, the different dependence of excitation wavelength of TiO2 and BiVO4 indicates that mid-gap bands exist in TiO2 rather than BiVO4. Based on the spectral range of fluorescence emission of TiO2 (inset of Figs. 6b and 480 ~ 800 nm), it is estimated that the energy levels corresponding to the SP states are about 1.55 ~ 2.58 eV above the valence band of TiO2.

According to the above discussions, we refined the band structure of TiO2/BiVO4 and added the SP states in TiO2. As shown in Fig. 6c, the small-polaronic mid-gap states (represented by SP in Fig. 6c) and VB of TiO2, together with CB and VB of BiVO4, display a zigzagged band structure arrangement, which is more beneficial in promoting carrier transport and enhancing electron-hole separation at the interface of heterojunction. Upon visible light illumination, BiVO4 would be excited to generate electron-hole pairs, whereas electrons in TiO2 can be transformed from the VB into SP states. Since the potential of SP states of TiO2 (centered at ~ 0.5 eV vs. NHE) is more negative than the potential of VB of BiVO4 (1.86 eV vs. NHE), the accumulated photogenerated electrons at the SP states can smoothly recombine with the holes on the VB of BiVO4 through the S-scheme migration path driven by the internal electric field, band bending, and Coulombic attraction30,31,32. Consequently, the photogenerated electrons and holes are spatially separated and reserved on the CB of BiVO4 and VB of TiO2, respectively.

Mid-gap states induced charge transfer in TiO2/BiVO4

To gain deep insight into the detailed carrier dynamics and photophysical process in TiO2/BiVO4 heterostructures, time-resolved PL and femtosecond transient absorption (fs-TA) measurements were carried out to investigate the photoinduced carrier dynamics of different systems. Figure 7a and b show the comparative study of the PL kinetics of BiVO4 in pure BiVO4 and TiO2/BiVO4 with excited (400 nm Exc.) and unexcited SP states (500 nm Exc.). It can be observed that the decay becomes faster in TiO2/BiVO4 at 400 nm excitation (Fig. 7a), and the average lifetime of BiVO4 is reduced from 17.3 ns to 8.9 ns after loading TiO2 (Table S1). The shortened degree of PL lifetime of BiVO4 in TiO2/BiVO4 (~ 51%) is close to the quenched fluorescence shown in Fig. 6b, indicating that the two changes come from the same dynamic process. Extraction and more efficient recombination of carriers can both lead to the reduced fluorescence intensity and lifetime. However, BiVO4 exhibits similar fluorescence decay kinetics in pure BiVO4 and TiO2/BiVO4 when the SP states are not excited (Fig. 7b), which excludes the possible electron or hole transfer from BiVO4 to TiO2 and confirms the necessity of excited SP states. Hence, we believe that the interfacial recombination of photogenerated electrons in SP states and holes in BiVO4 is the main reason for the differential PL lifetime and intensity, supporting the model we proposed.

Carrier dynamics at faster time scales were observed by carrying out femtosecond transient absorption (fs-TA) measurements using a pump pulsed laser with photon energy (430 nm, 2.88 eV) much smaller than the band gap of TiO2 (3.34 eV) and a continuum probe pulse in the near infrared (NIR) region (800 ~ 1500 nm)33. As shown in Fig. 7c, the fs-TA spectra of pristine BiVO4 and TiO2 are both dominated by a broad photoinduced absorption (PIA) signal centered at 990 nm and 1200 nm, respectively, while TiO2 also exhibits a weak negative signal centered at 835 nm. For TiO2/BiVO4 composites, the PIA wavelength range is broadened to the entire spectrum range and the central wavelength is shifted to 1050 nm. These TA features contain obvious contributions from both BiVO4 and TiO2, suggesting the existence of BiVO4 and TiO2 and confirming the successful formation of heterostructure. According to the steady state absorption spectra (Fig. S3), the broad PIA signal can be assigned to the excited-state absorption (ESA), and the negative bleach peak can be assigned to ground state bleach (GSB) caused by the state filling of carriers after excitation34,35,36.

We then distinguish the contribution of electrons and holes in the PIA and GSB signals by measuring transient absorption spectra in the presence of electron or hole scavengers in order to track the electron and hole behaviors separately. Following literature methods28,37we used ethanol as the hole scavenger and benzoquinone (BQ) as the electron scavenger. In the presence of hole scavenger (in ethanol), the TiO2/BiVO4 and pure TiO2 exhibit similar broad ground state bleach (GSB) (Fig. S4a), and no TA signal is observed for pure BiVO4, which indicates that the contributions of photogenerated electrons to the PIA signals are negligible, but are closely related to the GSB signals of TiO2/BiVO4 and TiO2. Notably, unlike pure TiO2, the GSB signal of TiO2/BiVO4 is redshifted with increasing delay time, and the peak position shifted from ∼980 to ∼1035 nm (Fig. 7d and S4b). This shift may be attributed to the bandgap modification of TiO2 as a result of the strong electronic coupling between the BiVO4 and the TiO2, that is, the fast electron transfer process from TiO2 to BiVO4 leads to an unbalanced carrier distribution in TiO2/BiVO4 and therefore the altered structure of transient absorption spectra1,37,38.

The rate of electron transfer can be obtained by fitting the TA kinetics probed at 952 nm. As shown in Fig. 7e, the GSB signals of TiO2/BiVO4 and TiO2 are both fitted by three-exponential functions using the parameters shown in the table inset the figure. The TA kinetics of TiO2/BiVO4 and TiO2 show two similar time constants of ~ 0.64 ps and > 10 ns, which can be corresponded to the intrinsic electron trapping and recombination processes in the TiO2 component, respectively. In addition, TiO2/BiVO4 also yields a faster time constant of τ2 (~ 45 ps) than pure TiO2 (τ2 ~ 101 ps), indicating a new decay pathway for electrons in the TiO2/BiVO4 heterojunction, i.e., the interfacial electron transfer process20. The ET rate is approximately estimated to be ~ 0.01 ps− 1 in terms of Eq. (2).

$$k_{{ET}} = {\text{ 1}}/\tau _{{\text{2}}} ^{'} - {\text{1}}/\tau _{{\text{2}}}$$
(2)

where τ2’ and τ2 are the τ2 time constants of TiO2 in pure TiO2 and TiO2/BiVO4, respectively. In addition, we do not observe the electron transfer process from BiVO4 to TiO2 in TiO2/BiVO4, as there is no slower rising edge in TiO2/BiVO4 (Fig. S4c).

Fig. 7
figure 7

Detailed carrier dynamics in TiO2/BiVO4 heterostructures, and the presence of localized polaronic mid-gap states can enhance the interfacial carrier transfer and recombination in TiO2/BiVO4 heterojunctions. PL kinetics of BiVO4 in BiVO4 and TiO2/BiVO4 under (a) 400 nm and (b) 500 nm excitation. The solid curves represent the fitting results using the parameters listed in Table S1. (c) TA spectra of TiO2, BiVO4 and TiO2/BiVO4 collected at 3 ps. The TA spectra of TiO2/BiVO4 contain contributions from both TiO2 and BiVO4. (d) TA spectra of TiO2/BiVO4 in ethanol at indicated delay times under 430 nm excitation, exhibiting a red shift with the increase of delay time. (e) Normalized TA kinetics of TiO2 and TiO2/BiVO4 in ethanol probed at 952 nm, reflecting the decay of electrons. (f) Corresponding kinetics of holes probed at 1100 nm. Solid lines show the results of exponential fitting by using the parameters listed in the figure and Table S2.

Full size image

The dynamic processes associated with photogenerated holes are also stripped out by the TA results measured with the electron scavenger BQ. BiVO4, TiO2 and TiO2/BiVO4 crystals all exhibit broad positive spectra that decay with increasing delay time, with central wavelengths around 1080 nm, 1300 nm and 1240 nm, respectively (Fig. S5), corresponding to the PIA of holes. In Fig. 7f, the kinetics of TiO2/BiVO4, BiVO4 and TiO2 under the same irradiation condition are compared and fitted by triexponential functions with the time constants listed in Table S2. The lifetime of photogenerated holes in TiO2/BiVO4 crystals is measured to be ~ 196 ps, which is much shorter than the lifetime of ~ 735 ps in pure BiVO4 and slightly longer than the lifetime of ~ 156 ps in pure TiO2. More significantly, the long-lived component τ3 is extended to ~ 2187 ps after the formation of the heterojunction. These results rule out the possibility of hole transfer from TiO2 to TiO2/BiVO4 and indicate that, in TiO2/BiVO4, the decay of photogenerated holes in BiVO4 is accelerated, accompanied by the generation of long-lived photogenerated holes in TiO2, which is well combined with our expectations. That is, there is an ET from TiO2 to BiVO4 and photogenerated electrons on the SP states of TiO2 can recombine with the holes on the VB of BiVO4, resulting in the long-lived separated holes left in TiO2.

Photocatalytic mechanism over TiO2/BiVO4

The above experimental results unambiguously confirm the model presented in Fig. 6c and the role of small-polaronic mid-gap states in the enhancement of photocatalytic performance of TiO2/BiVO4. The possible mechanism of MO degradation can be concluded as shown in Fig. 8. Due to the existence of SP states, both TiO2 and BiVO4 can be photoexcited to generate electron–hole pairs for the fabricated TiO2/BiVO4 composites. The separation and transfer of the photoinduced electron-hole pairs follows a direct S-scheme path, in which the accumulated electrons on the SP states of TiO2 would thermodynamically transfer to the VB of BiVO4 and recombine with photoexcited holes13,17. Consequently, the photogenerated electrons on the CB of BiVO4 and holes on the VB of TiO2 are spatially separated and reserved. These preserved carriers with strong redox ability can be used to produce active species with strong oxidation ability or oxidize MO directly, which eventually leads to the significant improvement of the performance of TiO2/BiVO4 for photocatalytic degradation of MO.

Fig. 8
figure 8

Proposed photocatalytic reaction processes and charge transfer of TiO2/BiVO4. The separation and transfer of the photogenerated electron-hole pairs followed a direct S-scheme path induced by localized polaronic mid-gap states. The efficient elimination of electrons in TiO2 and holes in BiVO4 can prolong the lifetimes of the remaining carriers and lead to the enhancement of photocatalytic activity of TiO2/BiVO4.

Full size image

In the photocatalytic degradation reaction of methyl orange (MO), the electrons on the CB of BiVO4 (− 0.61 eV vs. NHE) have adequate potential to react with O2 to generate superoxide anion radicals (·O2–) (− 0.046 eV vs. NHE), and hydroxyl radicals (OH) can be further formed due to the reaction of water molecules (2.27 eV vs. NHE)39,40,41. Besides, holes left on the VB of TiO2 (2.65 eV vs. NHE) exhibit strong oxidation ability and would either oxidize MO molecules (1.64 eV vs. NHE)42 or react with water molecules to engender ·OH. Finally, the MO molecules are eventually decomposed into CO2 and H2O under the action of active substances, namely holes, superoxide anion radicals and hydroxyl radicals.

Conclusions

In summary, the TiO2/BiVO4 nanocomposites with large energy barrier developed in this work demonstrate exceptional visible-light photocatalytic performance, achieving ~ 100% methyl orange degradation rate within 1 h under visible light. Unique charge transfer in Type-I TiO2/BiVO4 heterojunction enabled by polarons was confirmed by comprehensive characterizations. Time-resolved spectra revealed the critical role of SP states in facilitating ultrafast electron transfer (kET ~ 0.01 ps− 1) from TiO2 to BiVO4 and promoting charge separation, which effectively prolonged the lifetimes of photogenerated carriers and eventually led to the outstanding visible photocatalytic activity of TiO2/BiVO4. These findings fundamentally advance our understanding of polaron-mediated photocatalysis by establishing how strategically introduced mid-gap states can simultaneously optimize visible-light harvesting and charge separation in heterojunction systems. The mechanistic insights and material design principles presented here provide a robust foundation for developing high-efficiency photocatalysts for environmental and energy applications.