Spatial separation of oxidation and reduction co-catalysts for efficient charge separation: Pt@TiO2@MnO x hollow spheres for photocatalytic reactions

The design and synthesis of fine Pt@TiO2@MnO x hollow spheres is described. Pt and MnO x are spatially separated by TiO2 shells. The catalyst exhibits high efficiency of charge-separation and surface-reaction.


Introduction
Efficient charge separation is a critical factor for solar energy conversion by semiconductor-based systems. It is crucial for increasing the performance of solar cells, photoelectrochemical and photocatalytic systems. 1,2 As an important application, photocatalytic oxidation requires exceptionally efficient charge separation. Within photooxidation reactions, the photooxidation of water into O 2 has long been considered as the bottleneck of the water-splitting process, so the invention of highly active water oxidation catalysts is a key step in the development of light-driven water splitting. 3 In addition, the photocatalytic oxidation system can also be used to produce highly valued chemicals such as aldehydes and ketones. Compared with traditional methods with high temperature, toxic oxidants and hazardous waste, 4 photocatalytic oxidation processes can be performed under mild and environmentally friendly conditions. Substances unstable at high temperatures may also be synthesized by such selective light-assisted processes. 4 Recently, the emergence of a new family of metal-free polymer photocatalysts provided ne methods to enhance charge separation. 5,6 For instance, heterojunctions between carbon nitride and sulfur-mediated carbon nitride have been constructed to promote photocatalytic reactions. 6 Besides, many other strategies were adopted to improve charge separation, while the results remain unsatisfactory because of diverse problems. 1,[7][8][9] Cocatalyst loading has been proved to be an effective approach to accelerate charge separation. 1 Cocatalysts can be designed for oxidation and reduction purposes. Oxidation cocatalysts tend to capture holes (PbO 2 , MnO x and PdS, 10 etc.), while reduction cocatalysts prefer electrons (Pt, Pd, 1 etc.). Meanwhile, they can promote the surface reaction of the redox process. 11 Yu et al. have reported the fabrication of TiO 2 nanosheets loaded with Pt. Compared with pure TiO 2 nanosheets, the photocatalytic activity of Pt decorated nanosheets was obviously improved. 2 Simultaneous loading of oxidation and reduction cocatalysts could further improve the photocatalytic activity. Colón et al. fabricated Pt-TiO 2 /g-C 3 N 4 -MnO x composites and successfully proved that Pt and MnO x trap electrons and holes, respectively, leading to a further separation of charges. 12 However, both reduction and oxidation cocatalysts are randomly distributed in this case, which results in a random ow direction of charge carriers when they migrate to the cocatalysts. In such a scenario, the possibility of electron-hole recombination will be increased. In addition, the short distance between the reduction and oxidation cocatalysts is unfavorable for redox reactions, for close redox sites may lead to severe back-reactions. Spatial separation of oxidation and reduction cocatalysts is a method to solve this problem, yet limited success has been reported. Carbon nitride nanosheets have been fabricated to separate different cocatalysts, promoting the locally-incompatible oxidation and reduction reactions on the two surfaces. 5 In addition, Pt particles (reduction cocatalyst) and MnO x particles (oxidation cocatalyst) have been nicely deposited on different facets of BiVO 4 , exhibiting increased performance when compared to the randomly deposited counterparts. 1 To achieve more complete spatial separation of oxidation and reduction cocatalysts, as well as to make full use of the photocatalyst, we design and synthesize Pt@TiO 2 @MnO x hollow spheres (PTM-HSs) with Pt particles and MnO x loaded onto the inner and outer surfaces of TiO 2 hollow spheres, respectively. Different cocatalysts can be completely separated by the TiO 2 shells without intermixing. The photocatalytic material can be fully used owing to the thin shell (about 40-60 nm) of TiO 2 hollow spheres. TiO 2 is chosen in this study since it serves as an excellent benchmark photocatalyst for comparison purposes. [13][14][15] Additionally, as a model structure, this concept (i.e., different cocatalysts are separated by TiO 2 shells completely) can be extended to other redox cocatalysts such as Au, Ag, PbO 2 , etc. 1 Upon generation from TiO 2 , electrons and holes will ow inward and outward of the spherical TiO 2 shells, respectively, accumulating on corresponding cocatalysts and then taking part in redox reactions (Fig. 1). Electrons can promote the photo-reduction, while holes participate in the photo-oxidation. 16 Thus, the oxidation of benzyl alcohol accompanied by the elimination of electrons by O 2 is described in Fig. 1a. The reduction and oxidation reactions take place at different sides of the TiO 2 shells, which is critical for effective charge separation. In addition, light will pass through the thin TiO 2 shell and keep reecting in the cavity, increasing the scattering length for enhanced light-absorption. Combined with the large surface area, the PTM-HSs exhibit high efficiency for the photocatalytic oxidation of benzyl alcohol, a representative alcohol containing phenyl groups. Generally, the appropriate catalysts for benzyl alcohol oxidation can also perform well in the oxidation of other similar chemicals. 17 Thus, the current photoredox catalysis system can be expanded for other relevant organic synthesis of ne chemicals. Additionally, we propose a two-step photooxidation mechanism for PTM-HSs, which may provide inspiration for the design of similar catalysts with high activity.

Results and discussion
The synthesis of PTM-HSs (Fig. 2) starts from SiO 2 nanospheres prepared by a modied Stöber method, with H 2 PtCl 6 adsorbed on the surface of SiO 2 . 18 Then they were calcined at 500 C for 2 hours (h) under H 2 atmosphere to form Pt particles anchored on SiO 2 ( Fig. 2a and f). Subsequent coating of an amorphous TiO 2 shell was performed by the hydrolysis of titanium tert-butoxide (TBOT) ( Fig. 2b and g). Then the samples were calcined to improve the crystallinity of TiO 2 . It should be noted that direct calcination would destroy the structure of the TiO 2 shells (Fig. S2 †). To preserve the morphology, another layer of SiO 2 was added as a coating to form the outermost protective layer ( Fig. 2c and h). Aer calcination, the outer and inner silica layers were removed by NaOH (1.67 M) etching at 70 C for 8 h to form Pt@TiO 2 hollow spheres (PT-HSs) ( Fig. 2d and i). Finally, MnO x was selectively deposited on the outer surface of PT-HSs to form PTM-HSs ( Fig. 2e and j and S3 †) by a photo-deposition method (Fig. S1 †).
The successful synthesis of the PT-HSs is further validated. Fig. 3a shows that most Pt particles with an average size of ca. 3.2 AE 1.3 nm (Fig. 3c) are uniformly loaded onto the inner surface of TiO 2 shells. A few Pt particles agglomerate to form larger ones. The high resolution transmission electron microscopy (HRTEM) image of the area near the inner surface of TiO 2 shells (Fig. 3b) shows that the lattice spacing of 0.2227 nm and 0.3520 nm match well with the (111) planes of Pt and the (101) planes of anatase. It can be seen that a hexagonal Pt particle is coated by a layer of TiO 2 (Fig. 3b), resulting from the strong interaction between Pt and TiO 2 . A STEM-energydispersive X-ray spectroscopy (EDS) line scan was performed through the center of an individual PT-HS particle (inset in Fig. 3d and e), which further conrms the position of Pt particles relative to the TiO 2 shell. Fig. 3d shows strong signals of Ti and O at the edge of the PT-HS, which reveals a typical structure of TiO 2 hollow spheres. The TiO 2 shell is marked in Fig. 3d by  two groups of vertical lines, indicating the wall of the TiO 2 shell based on Ti and O signals in the EDS line scan. Fig. 3e shows the magnied Pt EDS signal, demonstrating that Pt particles are loaded onto the inner surface of the TiO 2 shell. It is not possible to use an EDS area scan to perform such conrmation because of the low loading of Pt (1%, measured by ICP) (Fig. S4d †).
The structure of the PTM-HSs is conrmed by the same method ( Fig. 4), where MnO x is identied on the outer surface of TiO 2 shells. The universal existence of PTM-HSs can be demonstrated by the TEM image of a larger area (Fig. S3 †).
The pore size and Brunauer-Emmett-Teller (BET) surface area of PTM-HSs are measured by N 2 adsorption, and determined to be 5.0 nm (average pore size) and 298 m 2 g À1 , respectively (Fig. S5a †). The results indicate a mesoporous structure, which is the prerequisite for effective immigration of reactants. Additionally, X-ray diffraction (XRD, Fig. S5b †) and HRTEM of the spine-like layer ( Fig. S5c and d †) prove that the PTM-HSs consist of Pt, anatase and MnO x , where x is between 1.0 and 2.0. Highly crystallized anatase instead of a rutile shell is formed, which is capable of achieving higher photocatalytic activity. 19,20 The components of the catalysts can be further conrmed by X-ray photoelectron spectroscopy (XPS, Fig. S8 †).
For comparison, reference catalysts such as pure TiO 2 hollow spheres (T-HSs) ( Fig. S6a and     be greatly enhanced when the reduction/oxidation cocatalysts are spatially separated. The activity of T-HSs and PT-HSs is even lower. These results can be explained by the improvement of the charge separation, which can be demonstrated by the photoluminescence (PL) spectra (Fig. 5b). The emission intensity of T-HSs is the strongest, resulting from the severe recombination of photogenerated charges. In contrast, the intensity of PT-HSs, T/P/M-HSs and PTM-HSs are reduced by about 25%, 45% and 80%, respectively, indicating an enhanced suppression of the recombination process.
To illustrate that the enhancement of activity is indeed caused by the improvement of charge separation, the oxidation of benzyl alcohol was performed under UV light (l < 420 nm). UV light (1.46 mW cm À2 ) was adopted to ensure that light absorption remained the same (Fig. S7 †) when different catalysts were used. Thus, the inuence of light absorption can be eliminated. Benzyl alcohol was oxidized to benzaldehyde without any over-oxidized products such as benzoic acid (explained in Fig. S9 †). Besides, the structure of the catalysts remained unchanged aer the reaction, indicating excellent stability (Fig. S10 †). As shown in Fig. 5c, negligible amounts of benzaldehyde were detected without the catalyst or in the dark condition (traces 5 and 6, Fig. 5c To further conrm the enhancement of separation efficiency (h s ), the apparent quantum efficiency (AQE) of PTM-HSs and T/P/M-HSs at 254 nm were determined to be 63.14% and 45.94%, respectively, which is higher than that of conventional TiO 2 -based catalysts (oen in the range 4.3-39%). 23 Because of the similar structure of the two kinds of catalysts and the same loading amount of cocatalyst, the efficiency of adsorption (h a ) and surface-reaction (h c ) remain the same between the catalysts (Fig. S7 †). Considering the overall efficiency is determined by h a , h s and h c simultaneously, the enhancement of AQE can be attributed to the enhancement of h s . Thus, we can quantitatively demonstrate that the spatial separation of cocatalysts indeed leads to an efficient charge separation for photocatalytic oxidation. Moreover, even under visible light (l > 420 nm), the electrons (generated from catalysts containing Pt and TiO 2 ) still ow from TiO 2 to Pt (instead of the hot electrons injected into TiO 2 ), which can be proved by a probe experiment (Fig. S12 †).
In order to investigate the mechanism of photocatalytic benzyl alcohol oxidation, particularly the main active species, phenol, and 4-hydroxy-2,2,6,6-tetramethylpiperidine 1-oxyl free radical (4-OH-TEMPO) were used as the scavengers of holes (h + ) and superoxide radicals (O 2 À ), respectively. As shown in Fig. 6a, when phenol (h + scavenger) was added into the photocatalytic system, an extremely low oxidation activity was observed, indicating that h + plays an indispensable role in the oxidation process. However, when 4-OH-TEMPO (O 2 À scavenger) was added, the activity only partly decreased, indicating that O 2 À also takes part in the reaction, while it is not indispensable. These observations suggest that two photocatalytic steps might be involved. The rst step primarily depends on h + , while the second step is driven by h + and O 2 À simultaneously. An intermediate would be formed in step 1 and be oxidized in step 2, bridging the two steps.
To determine the intermediate species, a kinetic isotope effect (KIE) study was conducted for PTM-HSs with benzyl alcohol under aerobic conditions. The KIE value was 0.78, suggesting the breaking of a C-H bond instead of an O-H bond (see ESI †). Accordingly, the intermediate was determined to be Rh-cCH-OH (Rh stands for the benzene ring). The speculated mechanism of photocatalytic benzyl alcohol oxidation is shown in Fig. 6b and eqn (1) to (5): Upon the separation of photo-induced electrons and holes over the PTM-HSs, the holes react with benzyl alcohol molecules to form intermediate free radicals (Rh-cCH-OH). Subsequently, some of the free radicals react with O 2 À , while others are further oxidized by holes to form the ultimate products (Rh-CH]O). Meanwhile, oxy species nally exist as H 2 O 2 aer a series of changes, which are discussed in detail in the ESI. † 24

Conclusions
In summary, the PTM-HS catalyst with Pt and MnO x loaded separately onto the inner and outer surfaces of TiO 2 hollow spheres was successfully synthesized for the photooxidation of benzyl alcohol. Pt particles favor electron trapping, while MnO x tends to collect holes. The spatial separation of Pt and MnO x by the TiO 2 shell greatly enhances the separation of electrons and holes. With this structure, different cocatalysts can be completely separated without intermixing. In addition, the large surface area of the TiO 2 shell can provide plenary attachment sites for Pt and MnO x , accelerating the consumption of electrons, which is good for the oxidation of water and benzyl alcohol. Furthermore, considering the multiple improvements of light-absorption, charge-separation and surface catalytic effect, the structure of these PTM-HSs could provide inspiration for other photocatalytic systems such as water splitting and CO 2 reduction.

Characterization
TEM was performed on a JEOL JEM 2100F electron microscope operating at 200 kV. The photoluminescence (PL) spectrum was performed on a Hitachi F-4600 uorescence spectrophotometer. Crystalline structures were evaluated by XRD analysis using a Bruker D8 Focus operating at 40 kV and 40 mA equipped with nickel-ltered Cu Ka radiation (l ¼ 1.54056Å). The BET surface area and pore structure of catalysts were measured using a Micromeritics Tristar 3000 analyzer by nitrogen adsorption at 77 K. The specic surface areas were calculated from the isotherms using the BET method. The pore distribution and the cumulative volumes of pores were obtained by the BJH method from the desorption branch of the adsorption isotherms. XPS was performed under ultrahigh vacuum (<10 À6 Pa) on a Kratos XSAM 800 spectrometer with Mg Ka X-ray source (E ¼ 1253.6 eV).

Methods
Synthesis of Pt@TiO 2 hollow spheres (PT-HSs). TEOS (0.8 mL) was mixed with deionized water (5 mL), H 2 PtCl 6 $6H 2 O (0.2 g mL À1 , 1 mL), ethanol (20 mL) and an aqueous solution of ammonia (0.4 mL). Aer stirring for 6 h at room temperature, the precipitated silica particles were separated by centrifugation and washed twice with ethanol, then re-dispersed in 5 mL of ethanol under sonication. Subsequently, the samples were vacuum dried at 80 C overnight. Then they were calcined at 500 C for 2 h under H 2 atmosphere to form Pt particles anchored on the SiO 2 . Then such SiO 2 -Pt powders were dispersed in 5 mL ethanol under sonication overnight with shaking. Subsequently, the suspension was mixed with ethanol (20 mL), deionized water (0.1 mL) and HPC (0.1 g). Aer 30 min stirring, 5 mL of 2.84 M TBOT ethanol solution was injected into the mixture at a rate of 0.75 mL min À1 . Aer injection, the temperature was increased to 85 C with stirring under reuxing conditions for 100 min to give SiO 2 -Pt@TiO 2 core-shell structures. The precipitate was isolated using centrifugation and washed twice with ethanol and water. The above SiO 2 -Pt@TiO 2 composites were dispersed in 20 mL water into which 0.14 g of PVP was added. Aer 12 h, the precipitate was separated, redispersed in ethanol (20 mL), and then mixed with water (5 mL), TEOS (0.1 mL) and aqueous ammonia (0.4 mL) to form the SiO 2 outermost protective layer. Aer stirring for 6 h, the resulting SiO 2 -Pt@TiO 2 @SiO 2 composites were centrifuged, washed three times with ethanol and dried under 80 C for 12 h. Subsequently, the above powders were calcined in air for 2 h followed by H 2 for another 2 h at 500 C. Then the calcined samples were dispersed in 20 mL water under sonication and heated to 70 C. 1 mL of 1.67 M aqueous NaOH solution was added to the above suspension. Aer etching for 8 h, the PT-HSs were formed.
Synthesis of Pt@TiO 2 @MnO x hollow spheres (PTM-HSs). PT-HSs powders (0.05 g), MnSO 4 solution (0.06 M, 5 mL) and NaIO 3 solution (0.02 M, 5 mL) were mixed in 100 mL deionized water, and the suspension was then irradiated by a 300 W xenon lamp (l < 420 nm, 1.46 mW cm À2 ) under continuous stirring. Aer 5 h photo-deposition, the precipitate was isolated using centrifugation, washed with deionized water more than 3 times, and nally dried at 60 C overnight to give PTM-HSs structures.
Synthesis of TiO 2 hollow spheres (T-HSs) and TiO 2 /Pt/MnO x hollow spheres (T/P/M-HSs). By repeating steps of the synthesis of Pt@TiO 2 hollow spheres without adding H 2 PtCl 6 $6H 2 O, the T-HSs were synthesized. To synthesize the T/P/M-HSs, a SiO 2 -TiO 2 core-shell structure was synthesized at rst. Then the samples were mixed with H 2 PtCl 6 solution (0.01 g mL À1 , 20 mL). Aer 12 h stirring, the suspension was vacuum dried at 80 C overnight. Then SiO 2 was re-coated as the outermost layer. Then the samples were calcined in air and H 2 and etched in NaOH using methods mentioned in the step of the synthesis of Pt@TiO 2 hollow spheres. Finally, MnO x was loaded by the photo-deposition method mentioned above.
Photocatalytic oxidation of benzyl alcohol to benzaldehyde. Typically, 0.03 g of photocatalysts (PTM-HSs, T/P/M-HSs, PT-HSs or T-HSs) were added in a home-made reactor sealed with rubber stoppers. Then 5 mL of toluene (solvent), 50 mL of benzyl alcohol (reactant) and 40 mL of undecane (internal standard) were injected into the reactor. O 2 was bubbled through the mixture at 20 mL min À1 . The reactor was irradiated under magnetic stirring using a 300 W xenon lamp (PerfectLight Co.) to provide ultraviolet light (l < 420 nm) with an irradiation area of 6.25 cm 2 . The irradiation intensity was 1.46 mW cm À2 . Cooling water was used to eliminate the thermal effect in the reaction. The products were analyzed by a gas chromatograph system (GC 2060, Ramiin) with a ame ionization detector (FID). A standard solution (a mixture of 5 mL of toluene, 50 mL of benzyl alcohol, 40 mL of benzaldehyde and 40 mL of undecane) was used to calibrate the GC. The calibration factor (1.7) was obtained using the standard solution. Subsequently, the accurate amount of every sample was calculated using the area of every peak and the calibration factor. The correlation of peaks and products is determined by the retention time obtained from the prior analysis of pure chemicals, and further conrmed by the GC-MS analysis (Fig. S11 †). To investigate the mechanism, 0.06 g of phenol and 0.06 g of 4-OH-TEMPO were added to the mixture containing 0.03 g of PTM-HSs, 5 mL of toluene (solvent), 50 mL of benzyl alcohol (reactant) and 40 mL of undecane (internal standard). Subsequent reactions and detections were similar to those above.
Photocatalytic oxidation of water to O 2 . The photocatalytic O 2 evolution reactions were carried out in a closed gas circulation and evacuation system using a 300 W xenon lamp (Per-fectLight Co.) and an optical cutoff lter (PerfectLight Co. l > 420 nm). Normally, 0.03 g of photocatalyst was dispersed in 50 mL of 0.02 M NaIO 3 aqueous solution in a glass reaction cell. Before irradiation, the reaction system was thoroughly degassed by evacuation in order to drive off the air inside. The amount of evolved O 2 was determined by an online GC (TCD, Ar carrier). The rate of O 2 evolution in the initial 3 h was recorded for comparison.
Apparent quantum efficiency (AQE). The AQE was calculated with the following equation: AQE (%) ¼ N e /N p , where N e and N p are the amounts of electrons (taking part in the reaction) and incident photons, respectively. The amount of electrons is two times the amount of benzaldehyde, for every benzaldehyde molecule consumes two electrons during the reaction. The amount of benzaldehyde per unit mass and unit time was measured by a GC system aer reaction under irradiation with the wavelength at l ¼ 254 nm given by xenon lamp and a glass lter. The light intensity was xed at 0.15 mW cm À2 (measured by an irradiatometer, UV-A, Photoelectric Instrument Factory of Beijing Normal University), and the amount of incident photons per unit time was calculated by the intensity and the energy of a single photon at l ¼ 254 nm.
Turnover number (TON). TON ¼ (the number of transformed substrate molecules)/(the number of active sites). The number of transformed substrate molecules can be obtained by GC analysis, which has been discussed above (page 4 of ESI †). When calculating the denominator, it should be noticed that only cocatalysts (Pt and MnO x ) loaded onto TiO 2 act as active sites, which could be illustrated by the probe experiment (Fig. S12 †). The estimated TON is 4.6.