Single-molecule quantification of photoredox activities and dynamics at the nanoscale on multi-faceted 2D materials

Abstract

Two-dimensional bismuth oxybromide (BiOBr) with multiple facets has been widely used in photocatalytic pollutant degradation and clean energy production. Herein, we used in situ single-molecule fluorescence microscopy to quantify structure-specific photoredox activities of facet-dependent BiOBr. The nanometric-mapping of photoredox reactions (resolution: 20 nm) clearly unveils the catalytic heterogeneity on {001} facet-dominant and {010} facet-dominant BiOBr, respectively (BiOBr-001 and BiOBr-010). The corners of BiOBr nanoplates exhibit the highest photoredox activities, followed by edges and basal planes, which are attributed to the unsaturated coordination sites in corners and edges. BiOBr-001 corners show the photoreduction and photo-oxidation activities of 108.0 ± 11.5 and 654.5 ± 83.2 s⁻¹ μm⁻², respectively, which are 1.2 and 3.4 times those of BiOBr-010 corners. Other structures of BiOBr-001 also exhibit superior reactivity to BiOBr-010. Such phenomena are ascribed to shorter charge transfer distance and the existence of high-index facets in BiOBr-001. Bulk activity evaluation further supports the single-molecule analysis. The investigation of temporal activity fluctuation reveals surface restructuring probably accounts for the activity enhancement at the nanoscale under practical reaction conditions. Hence, our study correlates the catalyst structure and reaction dynamics at nanometer resolution, which guides the performance improvement at both single-molecule and ensemble levels.

---

1. Introduction

Two-dimensional (2D) semiconductor materials play a pivotal role in photocatalytic renewable energy production and environmental remediation. Bismuth oxybromide (BiOBr) is a typical 2D oxyhalide material with numerous structural and photophysical merits, which has gained significant attention in photocatalysis.^1,2 BiOBr has a tetragonal crystalline structure in the space group P4/nmm. The bismuth ions (Bi³⁺) and oxygen ions (O²⁻) are coordinated to form a [Bi₂O₂]²⁺ slab via strong covalent and ionic bonds, which provides great intra-layer structural integrity. The bromide ions (Br⁻) are located between the [Bi₂O₂]²⁺ slabs and their bindings belong to van der Waals force. This weak interlayer interaction enables the easy cleavage of BiOBr crystals at certain planes, endowing it with a 2D sheet-like morphology. The alternating [Bi₂O₂]²⁺ slabs and Br⁻ layers generate a unique electrostatic driving force along the z-axis, facilitating the photoinduced charge separation.^3,4 The electronic bandgap energy of BiOBr is 2.5–2.9 eV,^5–8 enabling it to utilize visible light efficiently for solar-driven photocatalytic applications. Besides, its chemical stability, environmental friendliness, ease of synthesis and strong photo-oxidation potentials make it a promising candidate for various photocatalytic processes (e.g., dye degradation,⁹ air purification,¹⁰ antibacterial disinfection,¹¹ H₂ generation¹² and CO₂ reduction¹³). However, the understanding and evaluation of BiOBr photoactivities usually remain at the ensemble level, which falls short in providing intra-particle catalytic information at the nanoscale. The facet-dependent and structure-specific properties of BiOBr lead to significant heterogeneity in catalytic performance and dynamics within a single BiOBr nanoplate. The collective catalytic behaviors of individual particles contribute to the bulk activities of real-world applications. Therefore, there is a great need to fundamentally understand the photocatalytic reactivity and kinetics at nanometer resolution, aligning with structural analysis at the same scale.

Facet-dependent properties are commonly present in several single-crystalline catalysts such as TiO₂,¹⁴ ZnO,¹⁵ bismuth oxyhalides (BiOX, X = Cl, Br, I),^16,17 Cu₂O¹⁸ and CeO₂.¹⁹ Their catalytic activity, selectivity and kinetics largely depend on the crystallographic facets exposed due to their distinct atomic arrangements, surface energies and active sites. Several microscopic and spectroscopic techniques have been employed to investigate charge carrier dynamics, localization and the distribution of surface-active sites on different facets of single crystals. Kelvin probe force microscopy (KPFM) is a versatile technique for studying the surface electronic properties of catalysts at the nanoscale. It allows for detailed nanometric-mapping of surface potential and work function variations, which provides the information of charge transfer and spatial distribution on specific facets.^20–22 However, this measurement relies on the tip–sample interaction, which can be largely affected by probe calibration, the quality of the interaction and environmental factors (e.g., temperature, humidity and contamination).²³ Additionally, KPFM has a slow scanning speed and limited temporal resolution, making it unsuitable for analyzing time-resolved properties and fast chemical dynamics. Surface photovoltage spectroscopy (SPS) is effective for studying the movement of electrons (e⁻) and holes (h⁺) and their separation properties in multi-faceted nanocrystals.^24–26 This technique relies on the surface photovoltage effect of semiconductors and falls short in analyzing highly conductive materials like metals with minimal photovoltage effects. Without complementary tools, SPS provides only qualitative results of surface potentials and charge behaviors. It cannot quantify reaction activity and kinetics on individual facets or structures. BiOBr exhibits great heterogeneities in charge separation efficiency, adsorption ability, defect formation and redox potentials among different facets (i.e., {001}, {010} and {110} facets). Through facet-selective synthesis, BiOBr nanoplates with different dominant facets and anisotropic high-index surfaces can be prepared. In particular, high-index facets possess a higher density of active sites and greater surface energy, which enhance the photoactivity.²⁷ Hence, it is desirable to characterize these facet/structure-dependent properties using in situ quantitative techniques and reveal the “structure-dynamics” correlations with high spatiotemporal resolution.

Single-molecule fluorescence (SMF) microscopy is an effective tool to real-time monitor the catalytic process and localize the surface-active sites on catalyst materials with nanometer precision.^28,29 It can not only resolve reaction heterogeneity with high spatiotemporal resolution, but also quantify nanoscale activity and dynamics at individual structures/facets. Shen et al. used SMF imaging to spatially correlate photoreductive and photo-oxidative sites on {001} facet-dominant BiOBr (denoted as BiOBr-001).³⁰ Through coordinate-based colocalization calculations, it was concluded that e⁻ and h⁺ colocalize at the regions with fewer defects while selective charge carrier (either e⁻ or h⁺) accumulate at defective sites. However, this work did not provide the activity information of another typical type of BiOBr ({010} facet-dominant BiOBr, referred as BiOBr-010) and temporal fluctuation studies on both crystals. Li and co-workers used the photochemical probing method to detect the locations of e⁻ and h⁺ on BiOBr-001 and BiOBr-010 by photodepositing metals or metal oxides.²⁷ This photolabeling approach combined with electron microscopy examination cannot in situ quantify the reaction rates and kinetics with time-dependent properties. Besides, the initial deposits could alter subsequent charge migration routes, thus impeding the study of inherent charge behaviors and facet nature. Therefore, our work aims to use SMF imaging to address these challenges and investigate the structure-specific activities and dynamics of 2D multi-faceted crystals at the nanoscale.

In this work, in situ SMF microscopy was used to quantify photoredox activities on individual structural features of BiOBr nanoplates (i.e., corners, edges and basal planes (BPs)). Different facet-dominant BiOBr (i.e., BiOBr-001 and BiOBr-010) catalysts were prepared through a pH-controlled synthesis.^27,31 Both samples exhibit significant activity heterogeneities within a single particle. Generally, the corners of BiOBr nanoplates possess the highest photoredox turnover rates, followed by edges and BPs. Specifically, the BiOBr-001 corner exhibits a photo-oxidation activity up to 654.5 ± 83.2 s⁻¹ μm⁻², which is 36.2% higher than that of edges (480.6 ± 63.5 s⁻¹ μm⁻²) and 10.7 times of BPs (61.1 ± 12.0 s⁻¹ μm⁻²). When comparing the photoredox performance between the two samples, BiOBr-001 shows superior activity at each structure to BiOBr-010. This can be ascribed to the shorter charge diffusion distance, inhibiting charge recombination and the presence of high-index facets, providing more surface-active sites. Bulk activity measurements of photocatalytic H₂ generation and RhB degradation were also performed on BiOBr, which correspond to the single-molecule analysis. The study of temporal activity fluctuations under saturated substrate concentration ([S]) provides hints into the relationship between microscopic catalyst surface restructuring and macroscopic photocatalytic performance.

2. Materials and methods

2.1. Synthesis of the photocatalysts

The BiOBr-001 and BiOBr-010 samples were prepared through a pH-controlled synthesis based on previous reports.²⁷ In a typical preparation, 1.6 mmol of Bi(NO₃)₃·5H₂O and 1.6 mmol of KBr were separately dissolved in 24 mL of DI water under continuous stirring for 30 min. Subsequently, 24 mL of the KBr solution was added dropwise into the Bi(NO₃)₃ solution, followed by adjusting the pH value to 1 using 1 M NaOH solution. The mixture solution was transferred into a 70 mL autoclave, which was heated at 220 °C for 24 hours and then cooled to room temperature. The product solution was centrifuged at 6000 rpm for 20 min. The precipitate was washed thoroughly with ethanol and DI water, which was subsequently dried at 60 °C to obtain BiOBr-001. BiOBr-010 was obtained using the same synthetic procedure with the solution pH adjusted to 6.

2.2. Materials characterization

The crystal facet information of BiOBr samples was analyzed by X-ray diffraction (XRD) using a Bruker D2 Phaser diffractometer. Cu Kα irradiation (λ = 1.54184 Å) with a voltage of 30 kV and a current of 10 mA was employed as the incident beam. For morphology characterization, field emission scanning electron microscopy (FESEM) images were acquired using a JEOL JSM 6701F microscope. Transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) images were obtained on a JEOL JEM-2100Plus microscope. UV-vis diffuse reflectance spectroscopy (DRS) was collected on a Shimadzu UV-2450 spectrophotometer with reference to a BaSO₄ standard. Ultraviolet photoelectron spectroscopy (UPS) was performed using a Shimadzu Kratos Axis Supra spectrometer, which is equipped with a He–I photon source emitting at 21.21 eV. The spectra are calibrated with the Fermi levels of the instrument and samples located at 0 eV. Photoluminescence (PL) spectra were collected on a Cary Eclipse Fluorescence Spectrophotometer.

2.3. Photoelectrochemical measurement

The transient photocurrent was measured using a three-electrode electrochemical setup, comprising a working electrode, a Pt counter electrode and a Ag/AgCl reference electrode. The system was controlled by a CHI 660E electrochemical workstation. The catalyst suspension was prepared by dispersing 3 mg of BiOBr photocatalyst into a solution consisting of 20 μL of 5 wt% Nafion and 980 μL of ethanol. After sonicating the mixture for 20 min, 30 μL of the catalyst ink was deposited on the ITO glass (0.196 cm²). The ITO glass was then placed in a 100 °C oven to dry completely overnight. During the photocurrent measurement, the sample region was irradiated by a 300 W Xe lamp (Newport) with a cut-off filter (>400 nm) in 0.5 M Na₂SO₄ solution. With a bias voltage of 0.5 V, the photocurrent was measured under alternating on/off light cycles. Electrochemical impedance spectroscopy (EIS) was performed in the frequency range of 0.1 Hz to 10⁵ Hz with a bias voltage of 5 mV.

2.4. Photocatalytic activity evaluation

The photocatalytic activity of H₂ generation was assessed using a 300 mL glass reactor equipped with a gas circulation and evacuation system. In a standard experiment with Pt cocatalysts, 100 mg of BiOBr and 1 mL of H₂PtCl₆ solution (1 mg Pt per mL, equivalent to 1 wt% Pt loading) were mixed with 100 mL of 10 vol% triethanolamine (TEOA) solution. The reaction system was evacuated and filled with argon gas 4–5 times to eliminate air, then purged with argon at approximately 20 torr. The solution temperature was maintained at 18 °C using a water circulation cooling system. A 300 W Xe lamp (Newport) with a cut-off filter (>400 nm) served as the irradiation source. An online gas chromatograph (Agilent 6890N) was used to monitor the H₂ generation amount. The measurement without Pt cocatalysts was conducted under the same conditions, excluding the addition of H₂PtCl₆ to the suspension. In addition, the photodegradation of RhB by BiOBr was evaluated using the same setup under atmospheric pressure. In a typical procedure, 100 mg of BiOBr was uniformly dispersed in 100 mL of RhB solution (10 mg RhB per L). Prior to irradiation, the mixture was stirred in the dark for 1 h to reach adsorption/desorption equilibrium. During illumination, 2 mL of the suspension was collected every 20 min. The photocatalyst was removed by centrifuging the suspension at 6000 rpm for 5 min. The clear solution was examined by a Shimadzu UV-2450 spectrophotometer. The RhB concentration was determined by measuring its peak absorbance at 554 nm.

2.5. Single-molecule fluorescence microscopy

SMF microscopy was performed using a total internal reflection fluorescence (TIRF) microscope equipped with an oil-immersion objective lens (Nikon Plan Apo l 100×, NA 1.45). To prepare the samples, a suspension containing 1 mg per mL BiOBr in ethanol was drop-cast onto the coverslip and fully dried in a 120 °C oven. The coverslip was tightly sealed with a microflow cell (75.5 mm × 25.5 mm × 0.6 mm) and placed on the microscope stage. The O₂ in the substrate solution was removed by purging the solution with N₂ for 30 min. Next, for the structure-dependent activity measurement, a solution containing 500 nM resazurin or 50 nM amplex red was introduced into the cell at a flow rate of 30 μL min⁻¹. Before laser irradiation, bright-field images were collected. During the SMF imaging, a circularly polarized 532 nm laser (50 mW, L6CC Oxxius) and a 405 nm laser (1 mW) were employed to excite the resorufin molecules and BiOBr, respectively. To minimize the background PL signal from BiOBr, a band-pass emission filter (ET605/70m, Chroma) was inserted into the light path. The SMF images were captured using an electron-multiplying charge-coupled device (EMCCD) camera (Andor iXon3) at a frame rate of 100 Hz (exposure time: 10 ms). Each experiment typically involved a video composed of around 30 000 frames, which were subsequently processed by the software. The measurement of temporal activity fluctuations was performed on the same BiOBr nanoplate under the same conditions except for the substrate concentration of 1 μM resazurin or amplex red.

3. Results and discussion

The BiOBr-001 and BiOBr-010 nanoplates were prepared via a solvothermal route with tuning the pH of the solution. As revealed by the FESEM images in Fig. 1a and S1,† BiOBr-001 displays a quasi-square plate structure with a lateral size of 11–16 μm. The thickness of the plate is measured to be 424 nm (Fig. 1b). The atomic structure of BiOBr-001 shown in Fig. 1c consists of [Bi₂O₂] slabs and interlayered Br atoms. It can be observed that {001} surface is rich in electronegative oxygen atoms, which could bind with H⁺ in the surroundings and thus alter the crystal growth and morphology. In acidic conditions, the combination of H⁺ and surface terminated oxygen promotes the growth of {001} facets into a larger size.³¹ Previous studies demonstrated that the morphology of {001}-dominant BiOCl is controlled by the duration of solvothermal process.³² A shorter reaction time leads to the formation of square-shaped nanosheets with top/bottom and side surfaces identified as {001} and {010} facets, respectively. With prolonged solvothermal time, oblique facets such as {102} and {112} facets start to form at the corners and edges, transforming the plate morphology from a square to an octagonal shape. The crystal thickness also increases with the emergence of oblique lateral surfaces. Regarding our BiOBr-001 sample, the quasi-square shaped plates are acquired during the transition from square sheets to eighteen-faceted octagonal crystals. The TEM analysis further supports this by revealing the lateral crystal planes of BiOBr-001. In Fig. S2a and b,† the lattice fringes at the corner site of BiOBr-001 have a d-spacing of 0.228 nm, corresponding to the (112) atomic planes. Hence, oblique facets (i.e., {102} and {112}) are present at the side surfaces of BiOBr-001, instead of {010} facets. {001} facets are assigned to the top and bottom surfaces of BiOBr-001. The large spacing among the layers can generate polarization and dipoles along the [001] direction (Fig. 1c). The induced internal static electric fields drive the photogenerated e⁻ and h⁺ to move in opposite directions along the [001] axis. This configuration shortens the diffusion length of charges and minimizes their recombination in BiOBr-001, thereby enhancing charge separation efficiency. In Fig. 1d, S2c and d,† BiOBr-010 shows a distinct morphology compared to BiOBr-001 with a smaller lateral size and irregular shape. In the TEM image (Fig. S2e†), BiOBr-010 exhibits a sheet-like morphology with structural features such as corners, edges and BPs. The HRTEM image in Fig. S2f† reveals the crystal facets at the edge position of BiOBr-010. The lattice spacing of 0.404 nm is ascribed to the (002) crystal planes. Therefore, the top/bottom and side surfaces of BiOBr-010 are identified as {010} and {001}, respectively. As BiOBr-010 was synthesized in a neutral solution, less interaction between H⁺ and surface oxygen suppresses the growth of {001} facets, resulting in generation of more {010} facets. In addition, the lack of H⁺ induces a faster nucleation rate of BiOBr, leading to the formation of smaller nanoplates. The thickness of BiOBr-010 is measured to be 61 nm (Fig. 1e), much thinner than BiOBr-001. Since the dipole direction inside the BiOBr-010 crystal is perpendicular to the [010] axis (Fig. 1f), the charge migration distance in BiOBr-010 approximately corresponds to its lateral size, which is larger than the charge diffusion length in BiOBr-001 (i.e., the thickness of BiOBr-001). Such property increases the chance of charge coupling and negatively affects the photocatalytic performance of BiOBr-010.

Fig. 1 Morphology and structure of BiOBr-001 and BiOBr-010. FESEM images of (a and b) BiOBr-001 and (d and e) BiOBr-010. Atomic structure of (c) BiOBr-001 and (f) BiOBr-010.

The crystal facets of BiOBr nanoplates are further analyzed by XRD characterization. In Fig. 2a, the XRD peaks at 11.0°, 22.0°, 25.3°, 31.8°, 32.3°, 33.3°, 39.5°, 44.9°, 46.4°, 47.1°, 50.1°, 56.4° and 57.4° can be ascribed to the (001), (002), (101), (102), (110), (003), (112), (004), (200), (113), (104), (114) and (212) facets of BiOBr, respectively (JCPDS card no. 09-0393). The peak intensity ratio of (002) and (200) planes for BiOBr-001 is determined to be 3.3, much higher than that of BiOBr-010 (0.6). The intensity of other characteristic peaks of (001), (102), (003), (112) and (004) planes for BiOBr-010 also reduces relative to the (200) peak intensity. These results indicate a significant difference in the dominant facets between BiOBr-001 and BiOBr-010, which supports our findings in the morphology analysis. The optical properties of BiOBr nanoplates are evaluated by UV-vis DRS. As shown in Fig. 2b, both samples are responsive to visible light, suggesting their potential for visible-light-driven photocatalytic applications. The inset of Fig. 2b shows the Tauc plot, in which photo energy (hν) is plotted against (αhν)^1/2. BiOBr-001 and BiOBr-010 have similar bandgap energies (E_g) of 2.50 and 2.45 eV, respectively. Based on the UPS analysis, the valence band potentials (E_VB) of BiOBr-001 and BiOBr-010 are measured to be 2.38 and 2.32 eV, respectively versus the Fermi levels of the samples (Fig. S3a and b†). Accordingly, the conduction band potentials (E_CB) of BiOBr-001 and BiOBr-010 are calculated to be −0.12 and −0.13 eV, respectively, using the equation of E_g = E_VB − E_CB (Fig. S3c†). To study the photocatalytic properties of the samples at the nanoscale, SMF technology was applied to investigate photoredox activities at individual structural features of a single BiOBr nanoplate (Fig. 2c). Detailed descriptions of the experimental process and principles can be found in the ESI.† Briefly, resazurin and amplex red are used as probe molecules to detect e⁻ and h⁺via photoreduction and photo-oxidation reactions, respectively both of which form the same fluorescent product of resorufin. In Fig. 2d, each resorufin fluorescein molecule shown in the single frame represents a catalytic turnover event on BiOBr surface. The intensity trajectory of the fluorescent signal (Fig. 2e) is obtained from the cyan box in Fig. 2d. In Fig. 2e, τ_off manifests the time intervals between the two adjacent turnover events, reflecting the catalytic conversion kinetics. The dwelling duration (τ_on) of resorufin at the catalytic site before leaving indicates its dissociation properties. The excellent localization precision (i.e., spatial resolution) of 20 nm is achieved due to the merits of TIRF microscopy in our imaging setup (Fig. S4 and eqn (S1)–(S3)†). This configuration effectively enhances the photon collection via high numerical apertures and minimizes background noise.

Fig. 2 Photophysical properties of BiOBr and SMF imaging of photoredox reactions on BiOBr. (a) XRD patterns, (b) UV-vis DRS spectra (inset: Tauc plots) of BiOBr. (c) Schematic of resazurin photoreduction and amplex red photo-oxidation on BiOBr investigated by SMF imaging. (d) A typical image of fluorescent product molecules (in yellow circles) on BiOBr-001. (e) A segment of the fluorescence trajectory obtained from the spot in the cyan square in (d).

The bright-field image of a single BiOBr-001 plate is shown in Fig. 3a with structural features of corners, edges and BPs clearly identified. Nanometric-mapping of photoreduction reaction heterogeneities is first obtained on BiOBr-001. Fig. 3b shows the reconstructed SMF image of the distribution of fluorescent products after resazurin reduction. The corresponding activity density map derived from the reconstructed localization results reveals the density of photocatalytic events occurring at each specific structure (Fig. 3c). More single-molecule turnovers are found at corner sites, followed by the edge structure. The BP shows the darkest color in Fig. 3b and c, suggesting its inert nature in catalytic reactions. To quantify the photocatalytic activities of each structure, image segmentation is performed to determine the size of each unit subregion (Fig. S5 and S6†). The catalytic turnover rate (ν_T) is calculated by dividing the number of catalytic events in a unit subregion by time and area. In Fig. 3d, the average catalytic turnover rate 〈ν_T〉 is obtained by fitting ν_T to its frequency, signifying the photoactivity at specific positions. The 〈ν_T〉 for resazurin photoreduction at the corner and edge of BiOBr-001 is determined to be 108.0 ± 11.5 and 84.2 ± 10.0 s⁻¹ μm⁻², respectively which are 31.8 and 24.8 times that of the BP (3.4 ± 0.7 s⁻¹ μm⁻²) (Table S1†). The outstanding performance of the corner and edge structures can be attributed to the presence of surface unsaturated sites (e.g., oxygen vacancies, lattice disorders, highly curved surfaces and stepped surfaces). These defective structures could act as catalytically active sites with high surface energy, which makes them more likely to bind or interact with reactants, thus enhancing adsorption. Furthermore, the low-coordination sites could reduce the activation energy required for chemical conversion and provide reaction pathways with lower energy barriers, thereby accelerating the catalytic reaction rate.^33,34 These findings reveal that corners and edges exhibit a significantly higher photoreduction activity than inner BP regions, which establish a correlation between structure and performance at the nanoscale. The evaluation of photo-oxidation activities for each individual structure is performed on the same BiOBr-001 nanoplate. The catalyst surface is thoroughly cleaned without product molecules by flushing with DI water and photobleaching the surface for 30 min before introducing the amplex red solution (Fig. S7a†). The morphology of the BiOBr-001 plate after the photo-oxidation reaction is shown in Fig. 3e, proving its outstanding stability. As shown in the SMF image and density map in Fig. 3f and g, most of the resorufin molecules are concentrated at the corner and edge positions, similar to the resazurin photoreduction. However, the product density in amplex red oxidation is much higher than that in resazurin reduction (Fig. 3c and g), suggesting the stronger oxidation ability of BiOBr photocatalysts. In the histogram distribution (Fig. 3h), the photo-oxidation 〈ν_T〉 of corners, edges and BPs are found to be 654.5 ± 83.2, 480.6 ± 63.5 and 61.1 ± 12.0 s⁻¹ μm⁻², respectively (Table S1†). Corners exhibit the highest turnover rate, which is 36.2% higher than that of edges and 10.7 times of BPs. Although corners and edges display superior photoactivity, they account for a smaller portion of the BiOBr surface area, while BPs constitute a significant percentage of the total surface. Therefore, chemical activation and functionalization of BPs by introducing surface defects or functional groups is a promising strategy to enhance its overall photoactivity. Our study successfully achieves nanometric-mapping of photoactive species (i.e., e⁻ and h⁺) on a BiOBr-001 plate, which unveils great spatial heterogeneity. Such information is usually hidden in conventional ensemble-averaged measurements. The in situ SMF analysis provides further insights into the comprehension of intra-particle properties, which guides the morphology design and optimization.

Fig. 3 Nanometric-mapping and quantification of structure-dependent photoredox activities on BiOBr-001. The bright-field images of BiOBr-001 (a) before and (e) after photoredox reactions. The SMF images reconstructed from 30 000 frames by localizing all fluorescent bursts of (b) resazurin photoreduction and (f) amplex red photo-oxidation on BiOBr-001. Scalebar in (a and b) and (e and f): 2 μm. (c and g) The activity density maps processed from (b and f), respectively. Bin size: 50 nm × 50 nm. Histogram distribution of single-turnover rates of (d) photoreduction and (h) photo-oxidation at BPs, edges and corners obtained from (b and f), respectively (more than 30 positions investigated for each structure).

The photoredox behaviors of BiOBr-010 are also investigated using SMF microscopy at the single-molecule level. Fig. 4a shows the irregular shape of a BiOBr-010 nanoplate, which is much smaller than BiOBr-001. The structural features of corners, edges and BPs can be clearly distinguished. As shown in Fig. 4b and c, large quantities of product molecules are distributed at the corner and edge locations, which is similar to BiOBr-001. Less fluorescent signal is observed at the BP, indicating its poorer photoactivity. In Fig. 4d and Table S1,† the photoreduction 〈ν_T〉 of BiOBr-010 corners, edges and BPs are 93.3 ± 18.0, 17.3 ± 3.7 and 3.0 ± 0.4 s⁻¹ μm⁻², respectively. Fig. 5a compares the photoreduction 〈ν_T〉 at the same structures between BiOBr-001 and BiOBr-010. The right y-axis k_001/010 is calculated as the ratio of photoreduction 〈ν_T〉 of BiOBr-001 to that of BiOBr-010. BiOBr-001 exhibits superior photoreduction activity for all structures compared to BiOBr-010. In particular, the edge position of BiOBr-001 has a significantly higher 〈ν_T〉 than that of BiOBr-010. The k_001/010 for edges is found to be 4.9, much higher than that of corners (1.2). This notable difference in 〈ν_T〉 at the edges of BiOBr-001 and BiOBr-010 could be attributed to the presence of oblique facets (e.g., {102} and {112}) on the lateral sides of BiOBr-001. The photo-oxidation properties are analyzed on the same BiOBr-010 nanoplate after completely cleaning the catalyst surface (Fig. S7b†). In Fig. 4e, the nanoplate maintains excellent stability with no morphological changes after the photo-oxidation reaction. Similar to photoreduction, the majority of fluorescein molecules are located at the corner and edge segments (Fig. 4f and g), but their quantity is higher than that in photoreduction (Fig. 4b and c). In Fig. 4h and Table S1,† the photo-oxidation 〈ν_T〉 at the corner, edge and BP are 191.3 ± 35.3, 88.3 ± 19.4 and 4.7 ± 0.8 s⁻¹ μm⁻², respectively. These values are lower than those of BiOBr-001 for photo-oxidation, as reflected in Fig. 3h and 5b. The k_001/010 for photo-oxidation at each structure is much larger than one (Fig. 5b). Based on the above discussion, BiOBr-001 exhibits superior photocatalytic activity to BiOBr-010 for both reduction and oxidation reactions. This phenomenon could be explained by the different structures and physicochemical properties of the two samples. As mentioned in the morphology analysis (Fig. 1), an internal static electric field is generated in the layered BiOBr structure, which is parallel to the [001] direction and perpendicular to the [010] axis (Fig. 1c and f). Hence, the structural configuration of BiOBr-001 greatly reduces the charge migration distance before reaching the catalyst surface, enhancing the charge separation efficiency. In contrast, the charge transfer direction in BiOBr-010 is parallel to the {010} facets with a longer migration length, which increases the possibility of charge recombination. The results of transient photocurrent, EIS spectra and PL spectra support the view that BiOBr-001 exhibits superior charge separation compared to BiOBr-010 (Fig. S8†). In Fig. S8a,† both samples generate reproducible photocurrent signals during light on/off cycles. The photocurrent of BiOBr-001 is 69.3% higher than that of BiOBr-010, suggesting more efficient charge separation in BiOBr-001. The EIS study in Fig. S8b† reflects the interfacial charge transfer kinetics between the photoelectrode and the electrolyte. It is observed that the arc radius of BiOBr-001 is smaller than that of BiOBr-010, indicating the lower charge migration resistance of BiOBr-001. The PL spectra in Fig. S8c† reveal the charge separation efficiency of the two samples, as the recombination of e⁻ and h⁺ leads to PL emission. BiOBr-001 exhibits a reduced peak intensity compared to BiOBr-010, confirming its superior charge separation ability. In addition, the co-exposure of high-index facets at the lateral sides of BiOBr-001 further contributes to its superior activity. The high-index facets with higher surface energy contain numerous low-coordination sites and unsaturated bonds, which are more chemically reactive.^35,36 These sites have a strong affinity to reactant molecules, which promotes their activation and subsequent reaction. When the catalytic process involves multi-step reactions (e.g., amplex red photo-oxidation), the structural complexity of high-index facets could offer multiple types of reactive sites. This helps anchor different molecules and potential intermediates and provides suitable locations for each specific step in the catalytic pathways. Therefore, the nanoscale photoredox activity of BiOBr-001 is significantly higher than that of BiOBr-010 at each respective structure.

Fig. 4 Nanometric-mapping and quantification of structure-dependent photoredox activities on BiOBr-010. The bright-field images of BiOBr-010 (a) before and (e) after photoredox reactions. The SMF images reconstructed from 30 000 frames by localizing all fluorescent bursts of (b) resazurin photoreduction and (f) amplex red photo-oxidation on BiOBr-010. Scalebar in (a and b) and (e and f): 1 μm. (c and g) The activity density maps processed from (b and f), respectively. Bin size: 50 nm × 50 nm. Histogram distribution of single-turnover rates of (d) photoreduction and (h) photo-oxidation at BPs, edges and corners obtained from (b and f), respectively (more than 30 positions investigated for each structure).

Fig. 5 Structure-dependent and bulk photoredox activities of BiOBr. The comparison of structure-dependent (a) photoreduction and (b) photo-oxidation 〈ν_T〉 between BiOBr-001 and BiOBr-010. (c) Photocatalytic H₂ generation (100 mg of photocatalysts, 1 wt% Pt, 10 vol% TEOA, 300 W Xe lamp, >400 nm) on BiOBr. (d) Photocatalytic RhB degradation (100 mg photocatalyst, 10 mg per L RhB solution, 300 W Xe lamp, >400 nm) on BiOBr.

In addition to the quantification of single-turnover rates at the nanoscale, the photocatalytic performance at the ensemble level was also evaluated. Fig. 5c shows the photocatalytic H₂ generation rates of BiOBr samples under visible light irradiation (>400 nm), where 1 wt% Pt was added as the cocatalyst and 10 vol% aqueous TEOA solution was used as the hole scavenger. BiOBr-001 exhibits the H₂ production of 53.2 μmol g⁻¹ in four hours, which is 73.9% higher than BiOBr-010 (30.6 μmol g⁻¹). In the activity evaluation without Pt cocatalysts (Fig. S9†), BiOBr-001 also shows an enhanced H₂ generation rate (53.9% higher) compared to BiOBr-010. This demonstrates the superior photoreduction ability of BiOBr-001 to BiOBr-010 in agreement with the SMF results. To test the bulk performance of oxidation reactions on BiOBr, photodegradation of RhB was measured under visible irradiation. In Fig. 5d, BiOBr-001 fully degrades RhB in 120 min while BiOBr-010 decomposes 66% of RhB in the same duration. The evaluation of these two reactions reflects the photoreduction and photo-oxidation reactivity of BiOBr samples in real-world photocatalytic applications. Combined with SMF studies, this work provides a comprehensive understanding of photoredox properties of typical 2D materials at both single-molecule and ensemble-averaged levels. More importantly, the excellent nanometric-imaging ability of SMF microscopy successfully correlates catalyst structure with its performance at the nanoscale, which is often masked by conventional techniques.

To gain further insight into the structure-dependent catalytic dynamics, the temporal fluctuation of reaction rates was analyzed under saturated [S], which closely resembles the reaction conditions in practical photocatalytic applications. Fig. S10† shows the dependence of turnover rates on time, signifying changes in the number of turnovers per unit of time. These oscillatory properties result from the perturbation of chemical kinetics, particularly in catalytic conversion, dissociation or both under high [S].³⁷ The conversion and dissociation processes in this catalysis-induced activity perturbation can be represented by τ_off and τ_on, respectively. The contribution of τ_off and τ_on to the activity fluctuations can be evaluated using the autocorrelation function, C_τ(m) = 〈Δτ(0)Δτ(m)〉/〈Δτ(0)²〉,^38,39 where Δτ(m) = τ(m) − 〈τ〉, m is the event index number and τ represents either τ_off or τ_on. The plots of C_τoff–m and C_τon–m are shown in Fig. S11a and b,† respectively. Both C_τoff and C_τon(C_τoff, C_τon ≥ 0) exhibit an exponential decay with m, indicating the presence of fluctuations in the conversion (τ_off) and dissociation (τ_on). By calculating the fitting parameters of the decay curves, the obtained correlation times for τ_off and τ_on reveal the timescales of fluctuations in catalytic conversion and product dissociation, respectively. The fluctuation rates are calculated as the inverse of the correlation times. The histogram distributions of the correlation times for τ_off and τ_on are shown in Fig. S11c and d,† respectively. Its wide distribution indicates that the fluctuation rates of τ_off and τ_on on the same catalyst structure vary significantly. As the temporal perturbation arises from the catalytic reaction, the various discrete turnover rates lead to distinct τ_off and τ_on fluctuation rates. Therefore, the fluctuation rates of τ_off and τ_on are plotted with the catalytic turnover rates to study their relationships (Fig. S12–S14†). Based on previous reports, activity fluctuations are attributed to nanoscale surface restructuring on the catalyst.^40–42 Therefore, the fluctuation rate of each structure also reflects the extent of dynamic surface restructuring at this catalytic subregion. As shown in Fig. S12 and S13,† the fluctuation rates for the τ_off process are higher than those for τ_on at all structures for photoredox reactions, suggesting the larger contribution of catalytic conversion to activity fluctuations than dissociation. With increasing turnover rates, the fluctuation rates for τ_off increase more rapidly than those for τ_on at corner and edge structures (Fig. S12a, b, d, e, S13a, b, d and e†). This indicates that higher catalytic rates have a greater impact on activity fluctuations induced by the conversion process (τ_off) at these structures. Furthermore, the larger deviation of fluctuation rates at higher turnover rates unveils that the τ_off and τ_on lead to more distinct perturbation behaviors, both of which have no cross-correlation (Fig. S15†). It can be deduced that at higher turnover rates, conversion and dissociation occur at different sites, where the reactant-assisted dissociation (also referred as indirect dissociation) dominates. Under high [S] conditions, indirect dissociation is more favorable as abundant reactant molecules tend to replace product molecules at catalytic sites, which subsequently migrate to docking sites and eventually dissociate. In contrast, the τ_off and τ_on at BPs show almost the same slopes without such properties in Fig. S12c, f, S13c and f.† This is probably because of the steric effect of BPs. The flat surface of BPs facilitates the adsorption of planar aromatic resazurin molecules at docking sites via strong van der Waals forces, which hinders indirect dissociation. Fig. S14† shows a comparison of the fluctuation rates for corners, edges and BPs in the same catalytic process. It can be observed that BPs have the lowest turnover rates among all structures, which is consistent with the SMF results (Fig. 5a and b). Moreover, corners and edges generally exhibit higher activity fluctuation rates than BPs for the τ_off and τ_on processes in photoredox reactions. This demonstrates that the surface restructuring rates of corners and edges are much higher than that of BPs, which may account for the superior photoactivities of these structures to BPs. Based on above discussions, we discovered the relationship between temporal activity perturbations (i.e., nanoscale surface restructuring) and catalytic turnover rates for photoredox reactions on BiOBr nanoplates. The evaluation and analysis were conducted under saturated [S], which simulates the reaction conditions in real-world applications. This investigation helps establish “structure-dynamics” correlations at the single-molecule level, which facilitates the understanding of the practical catalytic performance from a nanoscale perspective.

This study has investigated the photoreduction and photo-oxidation activities of 2D crystalline materials at both single-molecule and ensemble levels. The facet-related and structure-dependent catalytic properties of BiOBr could provide guidance for the design and optimization of various 2D nanocatalysts. First, regarding bismuth oxyhalides (BiOX, X = Cl, Br or I) which share similar atomic structures, crystals with shorter charge diffusion lengths like {001} facet-dominant nanoplates can be synthesized, which have excellent charge separation ability and photoactivity. Facet engineering could be applied to fabricate BiOX nanocrystals with more high-index facets and anisotropic surfaces by controlling the synthetic conditions, thereby increasing the surface-active sites and regulating the charge flow. The conclusions can be extended to other analogous 2D materials such as layered oxyhalides (e.g., TiOX₂ and PbOX), transition metal dichalcogenides (TMDs) and black phosphorus. These materials possess internal static electric fields due to inherent polarization or structural asymmetry.^43–45 The induced dipoles are typically vertical to the plane of the 2D layers. Hence, reducing the material size along the dipole direction could greatly facilitate charge migration and separation. Second, our results reveal that corner and edge structures have significantly higher photoredox activities than BPs (generally more than 10 times). Edge engineering or creating material boundaries with higher curvature or more defects are promising routes to enhance the catalyst reactivity. Fabricating more edge structures or specific edge configurations can be realized by controlled edge-selective synthesis under precise growth conditions.^46,47 Exfoliation procedures could further reduce the lateral dimensions of 2D materials and generate more in-plane holes, increasing the density of edges and corners. Third, our results show that BPs have extremely low activities and low temporal fluctuation rates, signifying its inert nature in chemical reactions. However, it constitutes a major part of the 2D structure, which cannot be ignored. Surface modification and functionalization should be explored to chemically activate BPs. Introduction of functional groups (e.g., –OH and –COOH) to the BP would improve its interaction with reactant molecules. Incorporation of dopants or defects into BPs is an effective way to modify the electronic band structure of the material and enhance visible light absorption. Decorating BPs with metal nanoparticles or single atoms as co-catalysts could provide more reactive sites and improve the charge separation. Hence, our work offers valuable insights into the exploration of intra-particle catalytic heterogeneities of 2D materials. It provides a bottom-up strategy for optimizing the catalyst structure and activity at the nanoscale, leading to overall improvements in photocatalytic performance.

4. Conclusions

In conclusion, we used SMF imaging to investigate the structure-dependent photoredox activities and their temporal fluctuations in BiOBr nanoplates. The spatial resolution of activity density maps reaches 20 nm. Both BiOBr-001 and BiOBr-010 display notable heterogeneity in the distribution of photoreduction and photo-oxidation catalytic turnover events. Most single turnovers are concentrated at the corner and edge regions due to the low-coordination environment at these sites. The photo-oxidation activities at BiOBr-001 corners and edges are found to be 654.5 ± 83.2 and 480.6 ± 63.5 s⁻¹ μm⁻², respectively, which are 10.7 and 7.9 times of BPs (61.1 ± 12.0 s⁻¹ μm⁻²). In contrast, the corner, edge and BP of BiOBr-010 show photo-oxidation activities of 191.3 ± 35.3, 88.3 ± 19.4 and 4.7 ± 0.8 s⁻¹ μm⁻², respectively. Within the same structure, BiOBr-001 exhibits significantly higher activity than BiOBr-010 due to its superior charge separation and the existence of anisotropic high-index surfaces. The photoreduction performance of BiOBr structures follows the same trend as the photo-oxidation reaction. Bulk activity measurements further validate the single-molecule study, which correlates the photoactivity at both the nanoscale and ensemble levels. Such findings shed light on the quantitative study of photocatalytic dynamics at the nanoscale, which can be extended to other 2D-layered or multi-faceted materials. It would guide the catalyst design and structural optimization at the single-molecule level for efficient real-world photocatalytic applications.

Data availability

The data supporting this article have been included as part of the ESI.†

Author contributions

Shuyang Wu: conceptualization, data collection, analysis, methodology, writing of the original draft. Jinn-Kye Lee: SMF data collection and investigation. Mingyu Ma: visualization, review. Jia Xin Chan: editing, review. Zhengyang Zhang: supervision, funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We acknowledge the financial support from the Ministry of Education, Singapore, under its Academic Research Fund Tier 1 (No. RG60/21, RG1/23), and the Singapore Agency for Science, Technology and Research (A*STAR) AME YIRG grant (no. A2084c0065) and MTC IRG grant (no. M21K2c0110).

References

1. Z. Luo, X. Ye, S. Zhang, S. Xue, C. Yang, Y. Hou, W. Xing, R. Yu, J. Sun and Z. Yu, Nat. Commun., 2022, 13, 2230.
2. G.-Q. Zhao, J. Hu, J. Zou, X. Long and F.-P. Jiao, J. Environ. Chem. Eng., 2022, 10, 107226.
3. J. Sun, X. Li, Q. Zhao and B. Liu, Appl. Catal., B, 2021, 281, 119478.
4. M. Wang, R. Quesada-Cabrera, S. Sathasivam, M. O. Blunt, J. Borowiec and C. J. Carmalt, ACS Appl. Mater. Interfaces, 2023, 15, 49270–49280.
5. C. Gong, J. Chu, S. Qian, C. Yin, X. Hu, H. Wang, Y. Wang, X. Ding, S. Jiang and A. Li, Adv. Mater., 2020, 32, 1908242.
6. H. Feng, Z. Xu, L. Wang, Y. Yu, D. Mitchell, D. Cui, X. Xu, J. Shi, T. Sannomiya and Y. Du, ACS Appl. Mater. Interfaces, 2015, 7, 27592–27596.
7. H. Feng, J. Zhuang, A. D. Slattery, L. Wang, Z. Xu, X. Xu, D. Mitchell, T. Zheng, S. Li and M. Higgins, 2D Mater., 2017, 4, 025102.
8. J. Zhang, F. Shi, J. Lin, D. Chen, J. Gao, Z. Huang, X. Ding and C. Tang, Chem. Mater., 2008, 20, 2937–2941.
9. M.-L. Xu, X.-J. Jiang, J.-R. Li, F.-J. Wang, K. Li and X. Cheng, ACS Appl. Mater. Interfaces, 2021, 13, 56171–56180.
10. X. Li, Q. Dong, F. Li, Q. Zhu, Q. Tian, L. Tian, Y. Zhu, B. Pan, M. Padervand and C. Wang, Appl. Catal., B, 2024, 340, 123238.
11. K. Zhang, Y. Zhang, D. Zhang, C. Liu, X. Zhou, H. Yang, J. Qu and D. He, Chem. Eng. J., 2023, 468, 143581.
12. X. Zheng, L. Feng, Y. Dou, H. Guo, Y. Liang, G. Li, J. He, P. Liu and J. He, ACS Nano, 2021, 15, 13209–13219.
13. J. Meng, Y. Duan, S. Jing, J. Ma, K. Wang, K. Zhou, C. Ban, Y. Wang, B. Hu and D. Yu, Nano Energy, 2022, 92, 106671.
14. Y. Zhong, R. Wang, J. Chen, C. Duan, Z. Huang, S. Yu, H. Guo and Y. Zhou, ACS Appl. Mater. Interfaces, 2022, 14, 17601–17609.
15. K. Han, P. Ngene and P. de Jongh, ChemCatChem, 2021, 13, 1998–2004.
16. G. Zhou, B. Lei and F. Dong, ACS Catal., 2024, 14, 4791–4798.
17. Y. Peng, N. Zhao and J. Liu, ACS Appl. Nano Mater., 2022, 5, 2476–2482.
18. J. Cui, X. Zhang, H. Huang, M. Yang, B. Yang, Q. Yang, S. Liang and S. Sun, Adv. Funct. Mater., 2022, 32, 2111528.
19. J. Xu, Y. Wang, K. Wang, M. Zhao, R. Zhang, W. Cui, L. Liu, M. S. Bootharaju, J. H. Kim, T. Hyeon, H. Zhang, Y. Wang, S. Song and X. Wang, Angew. Chem., Int. Ed., 2023, 62, e202302877.
20. T. Hackl, G. Schitter and P. Mesquida, ACS Nano, 2022, 16, 17982–17990.
21. B. Mallada, A. Gallardo, M. Lamanec, B. De La Torre, V. Špirko, P. Hobza and P. Jelinek, Science, 2021, 374, 863–867.
22. M. Checa, A. S. Fuhr, C. Sun, R. Vasudevan, M. Ziatdinov, I. Ivanov, S. J. Yun, K. Xiao, A. Sehirlioglu and Y. Kim, Nat. Commun., 2023, 14, 7196.
23. T. Glatzel, U. Gysin and E. Meyer, Microscopy, 2022, 71, i165–i173.
24. R. Chen, F. Fan and C. Li, Angew. Chem., Int. Ed., 2022, 61, e202117567.
25. R. Chen, C. Ni, J. Zhu, F. Fan and C. Li, Nat. Protoc., 2024, 1–33.
26. J. Mao, Z. Gu, Y. Yu, H. Liu, J. Qu and X. An, ACS Appl. Energy Mater., 2021, 4, 2543–2551.
27. M. Shi, G. Li, J. Li, X. Jin, X. Tao, B. Zeng, E. A. Pidko, R. Li and C. Li, Angew. Chem., Int. Ed., 2020, 59, 6590.
28. M. Lelek, M. T. Gyparaki, G. Beliu, F. Schueder, J. Griffié, S. Manley, R. Jungmann, M. Sauer, M. Lakadamyali and C. Zimmer, Nat. Rev. Methods Primers, 2021, 1, 39.
29. T. Kondo, W. J. Chen and G. S. Schlau-Cohen, Chem. Rev., 2017, 117, 860–898.
30. M. Shen, T. Ding, W. H. Rackers, C. Tan, K. Mahmood, M. D. Lew and B. Sadtler, J. Am. Chem. Soc., 2021, 143, 11393–11403.
31. J. Jiang, K. Zhao, X. Xiao and L. Zhang, J. Am. Chem. Soc., 2012, 134, 4473–4476.
32. M. Li, S. Yu, H. Huang, X. Li, Y. Feng, C. Wang, Y. Wang, T. Ma, L. Guo and Y. Zhang, Angew. Chem., Int. Ed., 2019, 58, 9517–9521.
33. Y. Gao, M. Zhang, Y. Jin, M. Zhou, Y. Mao, J. Sun, W. Wang and Z. Song, Appl. Catal., B, 2024, 341, 123348.
34. Y. Jian, M. Tian, C. He, J. Xiong, Z. Jiang, H. Jin, L. Zheng, R. Albilali and J.-W. Shi, Appl. Catal., B, 2021, 283, 119657.
35. Q. Kuang, X. Wang, Z. Jiang, Z. Xie and L. Zheng, Acc. Chem. Res., 2014, 47, 308–318.
36. R. Huang, Y.-H. Wen, Z.-Z. Zhu and S.-G. Sun, J. Mater. Chem., 2011, 21, 11578–11584.
37. X. Zhou, W. Xu, G. Liu, D. Panda and P. Chen, J. Am. Chem. Soc., 2010, 132, 138–146.
38. K. S. Han, G. Liu, X. Zhou, R. E. Medina and P. Chen, Nano Lett., 2012, 12, 1253–1259.
39. H. P. Lu, L. Xun and X. S. Xie, Science, 1998, 282, 1877–1882.
40. W. Xu, J. S. Kong, Y.-T. E. Yeh and P. Chen, Nat. Mater., 2008, 7, 992–996.
41. P. Chen, X. Zhou, N. M. Andoy, K.-S. Han, E. Choudhary, N. Zou, G. Chen and H. Shen, Chem. Soc. Rev., 2014, 43, 1107–1117.
42. W. Xu, J. S. Kong and P. Chen, Phys. Chem. Chem. Phys., 2009, 11, 2767–2778.
43. H. Qu and Y. Li, Phys. Rev. B, 2023, 107, 235407.
44. F. Liu, J. Zhou, C. Zhu and Z. Liu, Adv. Funct. Mater., 2017, 27, 1602404.
45. L. Zhang, B. Wang, Y. Zhou, C. Wang, X. Chen and H. Zhang, Adv. Opt. Mater., 2020, 8, 2000045.
46. Y. Li, H. Lin, W. Zhou, L. Sun, D. Samanta and C. A. Mirkin, Sci. Adv., 2021, 7, eabf1410.
47. Y. Wen, J. Cai, J. Zhang, J. Yang, L. Shi, K. Cao, R. Chen and B. Shan, Chem. Mater., 2018, 31, 101–111.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ta07149b

---