Copper on the inner surface of mesoporous TiO2 hollow spheres: a highly selective photocatalyst for partial oxidation of methanol to methyl formate

Changfu Li , Xuzhuang Yang *, Guanjun Gao , Yuanyuan Li , Weida Zhang , Xuetao Chen , Haiquan Su , Sijia Wang and Zhen Wang
School of Chemistry and Chemical Engineering, Inner Mongolia Key Lab of Rare Earth Materials Chemistry and Physics, Inner Mongolia University, Hohhot, Inner Mongolia 010021, P. R. China. E-mail: xzyang2007@yahoo.com

Received 9th August 2019 , Accepted 17th October 2019

First published on 17th October 2019


To raise the methyl formate (MF) selectivity at high methanol conversion is one of the most challenging topics for photocatalytic partial oxidation of methanol to MF. This work addresses the enhancement of the MF selectivity and methanol conversion by designing a new structure of Cu@TiO2 double layered hollow spheres with copper species dispersed on the inner surface of the mesoporous titania shell. We found that the new structure exhibited excellent MF selectivity at high methanol conversion for photocatalytic oxidation of methanol. The photocatalytic performance showed a close relation to its specific structure. The porous titania shell supplied sufficient surface areas for methanol chemisorption. The inner copper layer played important roles in accepting photoexcited electrons from the titania layer. Oxygen molecules chemisorbed on the negatively charged copper surfaces dissociated into active oxygen species and spilled to fill the oxygen vacancies on the titania layer. The pores on the titania shell were beneficial to oxygen inward diffusion, but methanol molecules were blocked from contact with the active oxygen species around the inner copper nanoparticles because of methanol molecules reacting with the surface hydroxyls at the opening of the pores as well as the refinement of the pore size, avoiding deep oxidation. The MF selectivity was proportional to the amount of surface hydroxyls. The light intensity was proportional to methanol conversion but of no relevance to the MF selectivity. The oxidation of methoxy to formaldehyde had a strong relation with light intensity and was the rate-determining step of the reaction, while the coupling reaction between methoxy and formaldehyde had a close relation to the reaction temperature. The catalyst had an induction period during which CuO was reduced to metallic Cu. The highest MF selectivity could be obtained at the stoichiometric ratio of methanol to oxygen, and the methanol conversion increased but the MF selectivity sharply deceased with the increase of oxygen partial pressure. The structure can be controlled by modulating the preparation parameters. This study may contribute to the design of new catalytic systems and provide an applicable green route to MF synthesis.


Introduction

High selectivity to the reactive partial oxidation intermediate is one of the foremost challenges in catalysis.1–4 Such reactions often occur in thermodynamically non-equilibrium systems,3–5 and the high selectivity to the partial oxidation intermediate is normally achieved through kinetically controlling the reaction: on the one hand, adjusting the pathway of the reaction by tailoring the microstructure of the catalyst; on the other hand, modulating the operation conditions such as the type of the reactor (batch or flow), the phase of the reaction, the concentration or partial pressure of reactants and the reaction temperature. A deeper understanding of the reaction as well as the catalyst properties will facilitate the design of the catalyst and enhance the partial oxidation selectivity.

Methyl formate (MF) is an important building block in synthesizing various chemicals such as formic acid, formamides, acetic acid and ethylene glycol.6–10 It is also more efficient in the transport and storage of CO and H2 since a molecule of methyl formate contains one more CO than methanol.10 The first MF production process was mentioned in the BASF patent in 1925 as a part of the formic acid production process.11 However, it has attracted great attention from researchers to study and develop new routes and technologies for MF synthesis in recent years because the current industrial production processes usually operate at relatively high pressure and temperature, and some processes use sulfuric acid or sodium methylate as the catalyst, resulting in serious corrosion of equipment and environmental problems.7,8,12–15 Partial oxidation of methanol is an important route for MF synthesis.9,10 But the reaction commonly occurs at elevated temperatures and/or pressures,6–8,13,14 leading to energy waste, multiple by-products and low MF selectivity. Wittstock et al. reported the selective oxidation of methanol to MF with high MF selectivity on a nanoporous gold catalyst at low temperatures,16 which shed new light on green MF production. Whiting et al. used titania supported Au–Pd nanoparticles as a catalyst for the reaction,17 and observed high MF selectivity at low temperatures and decreased MF selectivity at high temperatures. In recent years, photocatalytic MF synthesis from methanol has become a hot research topic due to its excellent catalytic performance and ambient reaction conditions.18–22 Liu et al. reported photocatalytic methanol oxidation on MoO3/TiO2 and TiO2 as early as 1985.23 Kominami et al. studied this reaction on an anatase-type TiO2 (ST-01) in 2010.22 However, the methanol conversion was very low at low temperatures and the MF selectivity decreased sharply at elevated temperatures.22,23 In order to improve the catalytic performance, we studied Au, Ag, Au–Ag alloy and Cu nanoparticles supported on TiO2 and ZnO as photocatalysts for MF synthesis from methanol at 15–45 °C under UV irradiation,18–20,24 and found that the methanol conversion increased remarkably at low temperatures but the MF selectivity, by comparison with metals on titania14,15 and pure titania,22 decreased by 5–20% although the MF formation rate increased greatly. Liu et al. and Colmenares et al. reported similar results on MoO3/TiO2, TiO2, Pd–Au/TiO2, and Pd–Cu/TiO2, respectively.5,21,23,25

According to the literature8,26 and our previous studies,18–20,24 methanol molecules could be chemisorbed and react with surface hydroxyls on the catalyst, giving rise to coordinated methoxy at room temperature without irradiation. The coordinated methoxy could be further oxidized to coordinated formaldehyde by the photo-generated hole under irradiation. And finally, a MF molecule was formed after the coupling of the coordinated formaldehyde with its neighbouring methoxy. All the above procedures occurred on the surfaces of the catalyst. The metal nanoparticles on the catalysts (TiO2 and ZnO) played a role in accepting the photo-excited electrons; on the other hand, the oxygen molecules adsorbed on the negatively charged surfaces of the metal nanoparticles could dissociate into oxygen atoms to participate in the reaction. Although metal nanoparticles were beneficial to enhancing the methanol conversion, the active oxygen species on their surfaces could fully oxidize methanol as well, resulting in low MF selectivity. In addition, the metal nanoparticles highly dispersed on the surface of the semiconductor can occupy the spaces of the surface hydroxyls, reducing the effective areas for the reaction. Thus, it is highly desirable to develop new structures to enhance the MF selectivity at high methanol conversion and low reaction temperatures.

In order to improve the MF selectivity at high methanol conversion, a new structure with copper species dispersed on the inner surface of mesoporous titania hollow spheres is designed. The methanol chemisorption, methoxy oxidation and coupling reaction are expected to occur on the outer surfaces of the mesoporous titania hollow spheres, and oxygen molecules are expected to diffuse through the porous tunnels in the titania shell and dissociatively adsorb on the inner copper species. Thus, the total outer surfaces of the mesoporous hollow spheres are the effective reaction areas. In addition, full oxidation of methanol can also be avoided or at least reduced because the newly formed methyl formate molecules will soon escape from the outer surface of the mesoporous titania shell before being deeply oxidized by the active oxygen species spilled from the inner copper species. Such a design has great potential to enhance the MF selectivity at high methanol conversion as well as to be used in other selective oxidation reactions. The new structure was prepared by a two-step wet chemical method using tetrabutyl titanate, cupric acetate and glucose as raw materials, and was used as a photocatalyst for selective oxidation of methanol to MF at 25 °C to 45 °C under UV irradiation. The objectives of this study were to investigate the influence of the new structure on the catalytic performance for selective oxidation of methanol, especially the MF selectivity. The influences of the methanol partial pressure, oxygen partial pressure, and light intensity as well as the reaction mechanisms of the catalysts were studied.

Results and discussion

Fig. 1 shows the morphologies of the copper-containing carbon spheres (1a), the spheres coated with titanium species after drying under supercritical conditions (1b) and the sample calcined at 500 °C (1c). The cupric ions in the mixture of glucose and cupric acetate could be reduced to metallic copper by the carbonyl groups of glucose under hydrothermal conditions, which could serve as the catalyst for further dehydration of glucose to the carbon structure. The metallic copper species were then wrapped up by the carbon wires from dehydrated glucose to form the copper-containing carbon spheres. The sizes of the spheres can be controlled by modulating the concentrations of glucose and cupric acetate, as well as the hydrothermal temperature. The sizes of the copper-containing carbon spheres used in this study were distributed in the range of 170–390 nm. There were full hydroxyls and carbonyls on the surface of the spheres (evidenced by the FTIR spectra in Fig. S2), which were connected to the hydrolysates of tetrabutyl titanate to give rise to a core–shell structure (1b). The thickness of the shell can be controlled by modulating the amount of tetrabutyl titanate and the hydrolysis duration. The mesoporous titania hollow spheres with copper species dispersed on the inner surface were formed after calcination (1c). The inner hollow space resulted from the decomposition of the carbon species in the core of the sphere, and the mesopores in the titania shell were formed by the aggregation of titania nanoparticles. By comparison of Fig. 1b and c, it can be seen that the average pore size increased and the average nanoparticle size of titania also increased, which is consistent with the results of N2 adsorption–desorption listed in Table 1. The increase of the pore size resulted from the growth of the titania nanoparticles. The release of CO2 resulting from the decomposition of the organic species during calcination might facilitate the formation of the pores as well. The TG-DSC result indicates that the carbon species completely decomposed at 450 °C (Fig. S3).
image file: c9cy01595g-f1.tif
Fig. 1 SEM (a) and TEM (b and c) images of the samples. (a) Uncoated copper-containing carbon spheres prepared by hydrothermal treatment, 3CuC; (b) spheres coated with titanium species after drying under supercritical conditions, 3CuC@5Ti-Dr-S; (c) mesoporous hollow spheres after calcination at 600 °C, 3CuC@5Ti-500.
Table 1 BET specific surface area, pore volume, pore size and crystal size of titania
Sample name S BET, m2 g−1 Pore volume, cm3 g−1 Pore size, nm Crystal size of titania, nm
3CuC@3Ti-Dr-S 45.0 0.071 5.7 19.7
3CuC@3Ti-500 54.8 0.242 10.9 12.5
3CuC@5Ti-Dr-S 95.0 0.266 8.7 10.1
3CuC@5Ti-500 47.0 0.156 8.9 10.5
3CuC@6Ti-500 81.5 0.264 9.1 10.4
3CuC@8Ti-500 78.6 0.295 10.3 10.8
3CuC@10Ti-500 83.2 0.404 10.5 10.9
3CuC@5Ti-300 106.6 0.343 10.3 8.6
3CuC@5Ti-400 119.1 0.325 8.9 10.1
3CuC@5Ti-600 45.5 0.174 8.4 15.2
3CuC@5Ti-700 18.3 0.064 8.6 29.8
1CuC@5Ti-600 46.4 0.131 7.4 15.8
5CuC@5Ti-600 43.5 0.134 9.1 12.3


The amount of tetrabutyl titanate is the most remarkable factor that influences the thickness of the titania shell. Fig. 2 shows the TEM images of the samples prepared with different amounts of tetrabutyl titanate. Images (a), (b), (c), (e) and (f) are from 3CuC@3Ti-500, 3CuC@5Ti-500, 3CuC@6Ti-500, 3CuC@8Ti-500 and 3CuC@10Ti-500, which were prepared with 3.4 ml, 5.1 ml, 6.8 ml, 8.5 ml and 10.2 ml of tetrabutyl titanate, respectively. Image (d) is the local enlargement of image (a). Sample 3CuC@3Ti-500 has a single layered wall composed of titania nanoparticles (a and d). The average size of the hollow sphere is about 100 nm, and is smaller than that of its precursor carbon-containing sphere (Fig. 1b), indicating remarkable size shrinking due to the decomposition of carbon species in the core during calcination. The thickness of the shell and the average size of the hollow spheres increase with the increase of the amount of tetrabutyl titanate. The titania shell cannot be formed when the amount of copper-containing carbon spheres is 0.2 g and that of tetrabutyl titanate is less than 3.4 ml. The tetrabutyl titanate hydrolysis duration has little effect on the morphology of the samples but has a remarkable influence on the crystal size of titania (Fig. S4 and S5).


image file: c9cy01595g-f2.tif
Fig. 2 TEM images of the samples prepared with different amounts of tetrabutyl titanate. (a) 3CuC@3Ti-500; (b) 3CuC@5Ti-500; (c) 3CuC@6Ti-500; (d) enlargement of 3CuC@3Ti-500; (e) 3CuC@8Ti-500; (f) 3CuC@10Ti-500.

The information of elemental dispersion on the catalyst can be obtained using HAADF-STEM (high angle angular dark field). Fig. 3 shows the dark field images of the samples, HAADF images and elemental maps of O, Ti and Cu in selected areas (one sphere in each sample) and the elemental line scan across selected areas. The bright regions in the angular dark field images are due to the electron diffraction that satisfies Bragg's conditions. The HAADF images in selected areas exhibit spheres with bright borders and dark central areas, indicating the shell structure of the spheres. The thickness of the bright border becomes wider and wider and the central dark area becomes smaller and brighter with the increase of the tetrabutyl titanate amount used in the experiment, suggesting that the shell of the sphere becomes thicker and thicker. The elemental dispersion maps of oxygen and titanium in each sample are similar to the corresponding HAADF image as oxygen and titanium are from titania from which the shell of the sphere is constructed. Since the content of Cu in each sample is constant and it is dispersed in the inner surface of the hollow sphere, the Cu dispersion map of each sample seems to be uniform compared with that of O or Ti, especially for the sample with a thicker border. The saddle-like shape of the line scan curves of O, Ti and Cu also implies the hollow sphere structure of the sample. Before calcination, the sample dried in supercritical ethanol such as 3CuC@3Ti-Dr-S or 3CuC@5Ti-Dr-S has a core–shell structure rather than a hollow sphere. During calcination, the carbon species and the organic residues were released from the core of the sphere, resulting in plenty of mesopores and thus a larger specific surface area such as that shown by 3CuC@5Ti-300 and 3CuC@5Ti-400 in Table 1. However, the specific surface area of the sample decreased sharply with the increase of the calcination temperature especially when it was higher than 400 °C. This is because all of the carbon and organic species had been decomposed completely when the calcination temperature was higher than 500 °C, and not only were no new pores formed but the titania crystals also grew larger at higher temperature. The copper content in the sample has little influence on the specific surface area, but the specific surface area increases with the increase of the titanium content. The total pore volume decreased with the increase of the calcination temperature but it has little influence on the pore size. This is because the sphere shrunk and the average size of the titania crystals grew larger at high calcination temperature. The adsorption–desorption isotherms in Fig. S6 exhibit type IV isotherms and type H3 hysteresis loops, indicating the mesoporous structure of these samples.27 The low nitrogen uptake and indistinct point B in the isotherms indicate the absence of micropores in the samples.


image file: c9cy01595g-f3.tif
Fig. 3 Elemental map and line scan of selected areas in the samples using HAADF-STEM. (a) 3CuC@3Ti-500; (b) 3CuC@5Ti-500; (c) 3CuC@6Ti-500; (d) 3CuC@8Ti-500.

The XRD profiles in Fig. 4 indicate that the titania shell is in the anatase phase. No diffractions concerning the copper species can be identified from the profiles because the content of copper species was very low and they were dispersed on the inner surface of the shell. The calcination temperature remarkably influences the crystallinity of titania, which was enhanced with the increase of the calcination temperature (Fig. 4a). The crystal size grew from 8.68 nm to 29.8 nm, calculated with Scherrer's equation using the (101) diffraction, when the calcination temperature was raised from 300 °C to 700 °C (see in Table 1). The titanium content has little effect on the crystallinity of anatase when the amount of tetrabutyl titanate varied from 3.4 ml to 10.8 ml (Fig. 4b), but it becomes worse and worse with the increase of the copper content in the sample (Fig. 4c). No diffractions can be observed from the sample dried in air, but the sample dried in supercritical ethanol exhibits weak diffractions attributed to anatase, suggesting that anatase could not be formed during hydrolysis but began to form during drying in supercritical ethanol (Fig. 4d). The hydrolysis of tetrabutyl titanate normally gives rise to the amphoteric compound titanium hydroxide. Apparently, titanium hydroxide can transform into titanium oxide through dehydration under supercritical conditions but it cannot occur in the case of drying in air.


image file: c9cy01595g-f4.tif
Fig. 4 XRD profiles of the samples. (a) Effect of calcination temperature; (b) effect of Ti content; (c) effect of Cu content; (d) effect of the drying method.

Anatase is an indirect band gap semiconductor.28 Direct recombination of photogenerated electrons from the conduction band minimum (CBM) with holes from the valence band maximum (VBM) is impossible, and thus anatase has a longer lifetime of the photoexcited electron–hole pair than other polymorphs such as rutile and brookite which are direct band gap semiconductors, resulting in better photocatalytic activity in various applications. The information of energy band gaps of the samples can be obtained from the UV-visible spectra in Fig. 5a and b. The samples calcined at temperatures lower than 400 °C (3CuC@5Ti-300 and 3CuC@5Ti-400) as well as the sample without calcination (3CuC@5Ti-Dr-S) exhibit broad absorption bands from 800 nm to 200 nm. The bands from 400 nm to 800 nm are attributed to the absorption from carbon species as well as the local surface plasma resonance (LSPR) of metallic copper nanoparticles (the humps near 500 nm) in these samples. The absorption bands from 200 nm to 400 nm resulted from the indirect electron interband transition of anatase in the samples. The TG analysis (Fig. S2) shows that the organic residues in the samples started to decompose at 300 °C and completely decomposed at 500 °C. The organic residues were carbonized through dehydration at temperatures lower than 500 °C and the carbon species were further oxidized to carbon dioxide at temperatures higher than 500 °C. Metallic copper was thus well preserved in a reducible environment at low calcination temperatures. The uncalcined sample 3CuC@5Ti-Dr-S exhibits weaker absorption in the visible region than sample 3CuC@5Ti-300 because it has fewer carbon species and most copper species were covered by organic species. Sample 3CuC@5Ti-400 also exhibits weaker absorption in the visible region than sample 3CuC@5Ti-300 because some of the carbon species in the sample are fully oxidized and released in the form of carbon dioxide. The small hump at 500 nm indicates metallic copper nanoparticles in sample 3CuC@5Ti-400. The samples calcined at temperatures higher than 500 °C exhibit two absorption bands from 410 to 510 nm and from 600 to 800 nm, which are ascribed to the absorption of the d–d transition of Cu2+ (ref. 29–31) and the charge transfer from the valence band to the conduction band of copper oxides,32,33 respectively. It indicates that metallic copper species were oxidized to CuO gradually with the increase of the calcination temperature. Apparently, there are more CuO species in sample 3CuC@5Ti-600 and sample CuC@5Ti-700. The band gap of sample 3CuC@5Ti-500, calculated by the Kubelka–Munk method, is 3.19 eV but those of 3CuC@5Ti-600 and 3CuC@5Ti-700 are 3.14 eV and 3.08 eV, respectively. Since the band gap of semiconductor nanoparticles increases with decreasing particle size,32,34 the decrease of the band gap with increasing calcination temperature for samples 3CuC@5Ti-500, 3CuC@5Ti-600 and CuC@5Ti-700 resulted from the growth of the anatase crystal size at high calcination temperatures. Fig. 5b shows the influence of the shell thickness, namely the titanium content, on the absorption of light. Sample 3CuC@3Ti-500 in Fig. 5b exhibits remarkable absorption in the range from 600 to 800 nm and from 410 to 510 nm. As mentioned above, the two absorption bands are attributed to the electron d–d transition of Cu2+ and the charge transfer from the valence band to the conduction band of copper oxides, respectively. With the increase of the titanium content in the sample, the intensity of the two absorption bands decreased. Sample 3CuC@5Ti-500 exhibits weak intensity of the two absorption bands but there is no absorption for samples 3CuC@5Ti-600, 3CuC@5Ti-700 and 3CuC@5Ti-800. It indicates that CuO is the main copper species in the as-prepared samples, and the thickness of sample 3CuC@5Ti-500 is suitable because light can just penetrate through the shell. The copper content as well as the hydrolysis time has little effect on light absorption (see Fig. S7).


image file: c9cy01595g-f5.tif
Fig. 5 UV-visible spectra of the samples. (a) Effect of calcination temperature; (b) effect of Ti content.

The state of copper species in the samples can be identified by H2-temperature programmed reduction (H2-TPR) analysis. Fig. 6a and b show the H2 consumption profiles of the samples with different Ti contents and those at different calcination temperatures, respectively. The UV-visible spectra evidenced that CuO existed in the as-prepared samples calcined at temperatures higher than 500 °C. However, these samples exhibit two reduction bands. The two bands are ascribed to the reduction of CuO to Cu2O and Cu2O to metallic copper. Normally, the reduction of CuO is easier than that of Cu2O, and it starts to be reduced at 280 °C while Cu2O starts to be reduced at 300 °C.35–38 The reduction temperatures as well as the reduction intermediates have a close relation to the partial pressure of hydrogen in the gas mixture, the heating rate and the flow rate of the gas mixture.35 Cu2+ or Cu+ will directly transform into Cu0 and no intermediate phases can be observed during the reduction for both CuO and Cu2O at a heating rate lower than 10 °C min−1 when the bulk sample is exposed to the hydrogen mixture. But when CuO is exposed to a very small amount of hydrogen, an intermediate Cu2O will be produced during the reduction.35 Since the copper species in the current samples are dispersed on the inner surfaces of the mesoporous titania shell, hydrogen in the gas mixture has to pass through the tunnels of the shell by diffusion to react with copper species. According to the results of the N2 adsorption–desorption analysis, most sizes of the tunnels are about 10 nm or less, which is far larger than the effective size of a single hydrogen molecule but two or three orders of magnitude smaller than the mean free path of hydrogen molecules in the gas mixture. As a result, the diffusion of hydrogen molecules will follow the Knudsen diffusion model instead of molecular diffusion. Once hydrogen in the gas mixture around the copper species starts to be consumed at a suitable temperature, a hydrogen-poor environment at the inner surface of the shell will be formed due to Knudsen diffusivity being far smaller than molecular diffusivity, and thus give rise to the intermediate Cu2O. Accordingly, the copper species in most of the samples underwent from CuO through Cu2O to metallic Cu during the reduction. For sample 3CuC@3Ti-500, the band from 200 °C to 300 °C is ascribed to the reduction of CuO to Cu2O and the band from 400 °C to 525 °C is ascribed to the reduction of Cu2O to metallic Cu. The two reduction peaks shift to the high temperature region with the increase of the titanium content in the sample (Fig. 6a), namely, the shell thickness, due to the increase of the diffusion path. The inset in Fig. 6a shows the relation of the reduction temperature corresponding to the reduction of CuO to Cu2O with the shell thickness, where in 267 °C, 415 °C, 485 °C, 617 °C and 634 °C correspond to the shell thickness of 12 nm, 23 nm, 45 nm, 173 nm and larger than 200 nm for 3CuC@3Ti-500, 3CuC@5Ti-500, 3CuC@6Ti-500, 3Cu@8Ti-500 and 3CuC@10Ti-500, respectively. In addition, the reduction peaks were broadened due to Knudsen diffusion as well. The results of the UV-visible investigation indicate that no CuO but Cu0 exists in the samples calcined at temperatures lower than 400 °C i.e. 3CuC@5Ti-300 and 3Cu@5Ti-400. The XPS results also evidenced the presence of Cu0 in samples 3CuC@5Ti-300 and 3CuC@5Ti-400. As a result, the peaks at about 600 °C for samples 3CuC@5Ti-300 and 3CuC@5Ti-400 are not from the reduction of copper species but the decomposition of the organic residues in the samples (TCD detector). The reduction peaks corresponding to the reduction of CuO shift from 415 °C to 253 °C with the increase of the calcination temperature (Fig. 6b). This is because the size of the sphere became smaller and smaller and the titania crystal size grew larger and larger with the increase of the calcination temperature, especially for sample 3CuC@5Ti-700, in which CuO was almost directly exposed to the reduction gas mixture. This results in the reduction of CuO directly to Cu0 but not via the intermediate Cu2O to Cu0.


image file: c9cy01595g-f6.tif
Fig. 6 H2-TPR profiles of the samples. (a) Effect of Ti content; (b) effect of calcination temperature.

The core level XPS spectra of O 1s from the samples calcined at different temperatures are shown in Fig. 7. The O 1s band of each sample is split into three peaks located at 533.2 eV (peak α), 531.9 eV (peak β) and 529.6 eV (peak γ), which are attributed to oxygen in the water molecule, oxygen in surface hydroxyl and lattice oxygen in TiO2,39–41 respectively. The peaks of α are intensified with the increase of the calcination temperature but the peaks of β and γ are weakened, indicating that the surface hydroxyls and water molecules adsorbed were gradually lost while the lattice of anatase was more and more structured with increasing calcination temperature. This is consistent with the results of the XRD investigation, namely the size of the anatase crystals growing larger and larger with increasing calcination temperature. The inset in Fig. 7 shows the deconvolution profiles of sample 3CuC@5Ti-500 as an example. The detailed deconvolution results of the samples calcined at different temperatures are listed in Table S1. The surface hydroxyls which have a significant influence on the MF selectivity of the reaction decrease with increasing calcination temperature. The titanium content has little influence on the surface hydroxyls, which is also consistent with the result of XRD.


image file: c9cy01595g-f7.tif
Fig. 7 Binding energy of O 1s XPS spectra for the samples calcined at different temperatures.

Fig. 8 shows the methanol conversion, MF selectivity, anatase crystal size and surface hydroxyls of the samples calcined at different temperatures. The methanol conversion increases with the increase of the calcination temperature before 500 °C and then decreases, but the MF selectivity deceases with the increase of the calcination temperature. The relation between the MF selectivity and the calcination temperature coincides with that between the surface hydroxyls and the calcination temperature, suggesting an inherent causality between surface hydroxyls and MF selectivity. This result is consistent with our previous studies.18,19,24 The crystal size of anatase increases with the increase of the calcination temperature, but it rises gently before 500 °C and steeply after 500 °C. The crystal size seems to have a close relation to the methanol conversion. Sample 3CuC@5Ti-500, with an anatase crystal size of ca. 10.5 nm, exhibits the maximum methanol conversion, while samples 3CuC@5Ti-300, 3CuC@5Ti-400, 3CuC@5Ti-600 and 3CuC@5Ti-700, with anatase crystal sizes of ca. 8.6 nm, 10.1 nm, 15.2 nm and 29.8 nm, respectively, exhibit lower methanol conversion. The anatase crystals with smaller sizes have broader band gaps, resulting in lower quantum efficiency and thus lower methanol conversion, while the anatase crystals with larger sizes have smaller surface areas, fewer active sites, fewer heterojunction interfaces and thus higher photo-excited electron–hole pair recombination rates, also resulting in lower methanol conversion. In addition, the organic residues in samples 3CuC@5Ti-300 and 3CuC@5Ti-400 can cover the active sites on anatase, which reduces the absorption of light and thus decreases the quantum efficiency and methanol conversion.


image file: c9cy01595g-f8.tif
Fig. 8 Methanol conversion, MF selectivity, anatase crystal size and surface hydroxyls for the samples calcined at different temperatures. Operation conditions: reaction temperature, 30 °C; flow rate, 20 ml min−1; methanol, 1 vol%; O2, 0.5 vol%; balanced with N2.

The titania shell thickness is also an important factor that influences the activity of the catalyst, as shown in Fig. 9. The methanol conversion remarkably decreases but the MF selectivity slightly increases with the increase of the shell thickness. The samples with shell thicknesses of 12 nm, 23 nm, 45 nm, 173 nm and >200 nm correspond to 3CuC@3Ti-500, 3CuC@5Ti-500, 3CuC@6Ti-500, 3CuC@8Ti-500 and 3CuC@10Ti-500, respectively. Sample 3CuC@3Ti-500 has a single layer of the TiO2 crystal wall. Most copper species are dispersed on the inner sides of the crystals, with the outer sides of the crystals being exposed to the feed gas. The photoexcited electrons can efficiently transfer to copper and thus reduce the electron–hole pair recombination rate, resulting in more oxidizing holes on the outer surface of the catalyst. The holes can not only partially oxidize methanol to MF but also further oxidize MF to CO2, resulting in high methanol conversion but low MF selectivity. With the growth of the shell thickness, the titania shell turns from a single layer to multiple layers, resulting in a low photoexcited electron transmission rate and thus a high electron–hole pair recombination rate. Accordingly, the methanol conversion rate decreased and the MF selectivity increased.


image file: c9cy01595g-f9.tif
Fig. 9 Methanol conversion and MF selectivity of the samples with different shell thicknesses. Operation conditions: reaction temperature, 30 °C; flow rate, 20 ml min−1; methanol, 1 vol%; O2, 0.5 vol%; balanced with N2.

The catalysts do not exhibit any activity without irradiation in the temperature range from 25 °C to 45 °C. However, the methanol conversion of the catalyst almost rises linearly with increasing light intensity under UV irradiation, while the MF selectivity does not show any relevance to light intensity (Fig. 10a). According to the literature and our previous study, the reaction occurs in three steps: firstly, dehydrogenation of methanol to methoxy by chemisorption, secondly, oxidation of methoxy to formaldehyde by the photoexcited holes, and finally, MF generation through methoxy coupling with formaldehyde. The first step can occur on the surface of the catalyst at room temperature without irradiation. Light is not involved in the third step as well. It just plays a role in the second step of the reaction. The relation between the methanol conversion and light intensity indicates that the second step is the rate-determining step of the reaction. Under UV irradiation, the methanol conversion increases with rising reaction temperature but the MF selectivity slightly decreases (Fig. 10b). The reaction turned from being temperature independent into temperature dependent. The temperature cannot play any role in the second step of the reaction. At constant light intensity and constant temperature, the reaction rate is determined by the second step of the reaction. So the temperature dependence of the reaction results from the third step of the reaction, namely, the temperature promotes the coupling reaction between methoxy and formaldehyde. In addition, the temperature can also expedite the side reaction, resulting in the decrease of the MF selectivity.


image file: c9cy01595g-f10.tif
Fig. 10 Influence of light intensity (a), temperature (b), time on stream (c) and the ratio of methanol to oxygen (d). Notes: 3CuC@5Ti-600 was used as a catalyst; (a)–(c) were operated at a flow rate of 20 ml min−1 with the composition of methanol, 1 vol% and O2, 0.5 vol% balanced with N2; (d) was operated under the same conditions except for the ratio of methanol to oxygen.

The copper species exist as copper oxide in the as-prepared samples calcined at temperatures higher than 500 °C. Copper oxide is also a semiconductor with a band gap of 1.7 eV,42,43 as well as appropriate conduction band edges,43–46 and thus has been used in many photocatalytic reactions45,47–55 in recent years. But it is unstable under irradiation because the redox potentials for reduction and oxidation of monovalent copper oxide lie within the band gap,56–58 which hinders its application. Fig. 10c shows the induction and activation period of the catalyst. The methanol conversion gradually rises with time on stream and the MF selectivity rises first and then remains at a constant level. Copper oxide is reduced into metallic copper by irradiation in this period, and thus it has strong oxidability, resulting in high CO2 content in the products and low MF selectivity at the beginning of the reaction.

The ratio of methanol to oxygen greatly influences the methanol conversion and MF selectivity, as seen in Fig. 10d. At a stoichiometric ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]0.5, the methanol conversion rises from ca. 72% to ca. 85%, while the MF selectivity rises from ca. 71% to ca. 77% with the increase of the reaction temperature. With the increase of the oxygen content, the methanol conversion increases remarkably but the MF selectivity decreased sharply. In particular, the methanol conversion rises with the increase of the reaction temperature but the MF selectivity decreases with the increase of the reaction temperature; the higher the oxygen content the more dramatically the MF selectivity decreases. Apparently, oxygen content is one of the most sensitive parameters that influence the reaction. The increase of oxygen content is beneficial to the full oxidization of methanol. Obtaining a higher MF selectivity has to be at the expense of the methanol conversion.

Compared with the samples prepared by loading copper species directly on titania (P25) in our previous study,59 the samples with copper species dispersed on the inner surface of the anatase shell in this study exhibit better MF selectivity at higher methanol conversion. For instance, the methanol conversion of the previous sample with the best MF selectivity of ca. 75% is less than 55%, but for the samples calcined at 500 °C in this study, the methanol conversion of the sample with the lowest MF selectivity of ca. 75% is higher than 85%. The superior photocatalytic performance of the catalyst results from its unique structure. Scheme 1 shows the structure and reaction mechanism of the catalyst for the photocatalytic oxidation of methanol to MF. As evidenced by the TEM observation and N2 adsorption–desorption experiment, the catalyst spheres are formed by the mesoporous anatase shell with copper or copper oxide dispersed on the inner surface of the shell. Copper oxide can be reduced to metallic copper during the reaction; thus the shell becomes a copper–anatase double layered shell. The thickness of the copper layer, namely the copper content in the catalyst, has little or negligible effect on the catalytic performance of the catalyst. This is because only the copper nanoparticles dispersed at the interface between copper and titania play the roles of accepting photogenerated electrons and dissociatively adsorbed oxygen during the reaction. The calcination temperature is an important factor that influences the catalytic performance of the reaction. The crystal size of anatase, the porosity of the titania shell and the surface hydroxyls can be controlled by modulating the calcination temperature. The crystal size of anatase has a close relation with light absorption; the porosity of the titania shell influences the diffusion of oxygen and the surface hydroxyls are the most important active sites for the reaction, on which methanol molecules dehydrate to methoxy groups. The titania shell thickness is another important factor that greatly influences the methanol conversion and MF selectivity, which can be controlled by modulating the amount of tetrabutyl titanate during the preparation. The titania shell of the sphere varies from a single layer to multiple layers. Taking the samples calcined at 500 °C as examples, the average size of anatase in these samples is about 10 nm and the shell thickness of the samples varies from ca. 10 nm to ca. 200 nm at different amounts of tetrabutyl titanate; thus the number of anatase layers is from 1 layer to about 20 layers. The sample with a single anatase layer exhibits the best methanol conversion but the lowest MF selectivity. The methanol conversion decreases with the increase of the number of anatase layers, but on the contrary the MF selectivity increases with the increase of the shell thickness. The reaction mechanism of the titania–copper double shell hollow sphere is shown in Scheme 1.


image file: c9cy01595g-s1.tif
Scheme 1 Structure of the catalyst and reaction mechanism for photocatalytic oxidation of methanol. (1) An electron–hole pair is generated under UV irradiation; (2) ohmic contact is formed at the interface between metallic copper and titania; (3) oxygen molecules that diffused through the mesopores dissociate into active oxygen species at the interface and spill to the surface of titania; (4) methanol molecules react with surface hydroxyls to give rise to methoxy which was further oxidized to formaldehyde by the photoexcited holes, and methyl formate is formed by crosslinking between formaldehyde and methoxy.

The porous titania shell supplies large surface areas which are teeming with surface hydroxyls. The methanol molecule can react with the surface hydroxyl to give rise to methoxy. This reaction can happen at room temperature without irradiation.7,8,26 Under UV irradiation, the electrons in the valence band of titania transfer to the conduction band and give rise to photoexcited electron–hole pairs. At the beginning of the reaction, the photoexcited electrons overcome the energy barrier at the interface between CuO and titania to reduce CuO into Cu. So an ohmic contact is formed at the interface between Cu and titania because the work function of Cu is less than that of titania.19 The photoexcited electrons are thus more feasible to transfer to the metallic copper layer due to the lower electrical resistance at the copper–titania interface. The methoxy on the surfaces of the titania shell can be oxidized to coordinated formaldehyde by the photoexcited holes on titania.60,61 The coordinated formaldehyde is coupled with its neighbouring methoxy to give rise to methyl formate. Both oxygen and methanol molecules enter the tunnels of the porous titania shell through Knudsen diffusion because the average size of the pores (ca. 10 nm) is far below the mean free path of the molecules.62–64 Oxygen molecules approach the negatively charged copper layer by diffusion and dissociate into oxygen atoms on the copper nanoparticles. The oxygen atoms spill over to fill the oxygen vacancies on the surfaces of titania,60 which result from the surface hydroxyls' involvement in the reaction. Compared with oxygen molecules, it is harder for methanol molecules to reach the deep inner surfaces of the tunnels because the size of the molecules is larger than that of oxygen. On the other hand, the methanol molecules easily chemisorb on the surfaces of titania by reacting with the surface hydroxyls. Once it happens near the openings of the tunnels, the methanol molecule's inward diffusion will be very difficult. As a result, it is hard for the active oxygen coming from the copper layer to be in direct contact with methanol, methoxy, formaldehyde and methyl formate, avoiding deep oxidation and thus leading to high methyl formate selectivity. However, with the increase of the shell thickness, it is hard for the photoexcited electrons to transfer to the copper layer, similar to the oxygen molecules. This leads to low methanol conversion.

Conclusions

A hollow sphere catalyst with a double layered shell was prepared by a wet chemical method. The outer layer of the shell was composed of mesoporous titania and the inner layer of the shell was composed of highly dispersed copper species. The outer titania shell provides sufficient areas with surface hydroxyls which are important active sites for the reaction. The tunnels on the porous shell are beneficial to the transfer of oxygen to the inner copper surfaces but block the diffusion of methanol molecules to the deeper part of the tunnels, avoiding direct contact with active oxygen species from the inner copper layer and thus deep oxidation, due to their reaction with the surface hydroxyls at the opening of the tunnels as well as the refinement of the pore size. The structure can be controlled by modulating the amount of tetrabutyl titanate used in the experiment, the hydrolysis time, the calcination temperature and the crystal size of anatase.

The new structure exhibits excellent MF selectivity at high methanol conversion during the photocatalytic oxidation of methanol. The MF selectivity is proportional to the amount of surface hydroxyls. The anatase crystal size influences the methanol conversion. The shell thickness is proportional to the MF selectivity but inversely proportional to the methanol conversion. The light intensity is proportional to methanol conversion but almost has nothing to do with the MF selectivity. The oxidation of methoxy to formaldehyde has a strong relation with light intensity and is the rate-determining step of the reaction, while the coupling between methoxy and formaldehyde has a close relation to the reaction temperature. There is an induction period of the catalyst during which CuO is reduced to metallic Cu. The highest MF selectivity can be observed at the stoichiometric ratio of methanol to oxygen, and the methanol conversion increases but the MF selectivity sharply deceases with the increase of the oxygen partial pressure.

Experimental

Materials

Tetrabutyl titanate, cupric acetate, absolute ethanol and methanol were purchased from J&K Scientific. Ammonium hydroxide and glucose were purchased from Sinopharm Chemical Reagent Co. Ltd. All chemicals were used as received.

Catalyst preparation

The structure was built by two steps of synthesizing carbon-containing copper nanospheres by hydrothermal treatment and coating titanium species by hydrolysis of tetrabutyl titanate in absolute ethanol, followed by drying under supercritical conditions and calcination to remove the organic species in the cores of the structure.
Synthesis of carbon-containing copper nanospheres. Typically, 13.52 g of glucose and a suitable amount (0 ml, 1 ml, 3 ml and 5 ml) of cupric acetate aqueous solution (0.02 M) were added into a 200 ml beaker, and a suitable amount of deionized water was added into the beaker to form a solution of 120 ml followed by vigorous stirring. The solution was then transferred to a 200 ml autoclave for hydrothermal treatment at 180 °C for 12 h. The solid obtained was washed twice with deionized water and twice with absolute ethanol. The samples of the carbon-containing copper nanospheres were labelled as 0CuC, 1CuC, 3CuC and 5CuC, and the number in the label represents the volume of the cupric acetate solution added into the sample.
Preparation of the catalyst. 0.2 g of carbon-containing copper spheres was ultrasonically dispersed in 200 ml of absolute ethanol for 1 h in a flask. About 1 ml of ammonium hydroxide (25%, mass fraction) was added into the suspension until pH 7.2 was reached. A suitable amount (3.4 ml, 5.1 ml, 6.8 ml, 8.5 ml and 10.2 ml) of tetrabutyl titanate was ultrasonically dispersed in 100 ml of absolute ethanol in a beaker. And then the suspension in the beaker was poured slowly into the flask under stirring, followed by dropping a suitable amount of ethanol solution (50 ml of absolute ethanol and 13.4 ml of water) into the suspension within 0.5 h and continuous stirring for certain hours (2 h, 4 h, 6 h and 8 h). The obtained solid was dried under supercritical conditions in ethanol (260 °C, 7 MPa) for 15 min after washing twice with absolute ethanol. The catalyst was then obtained after calcination at a certain temperature (300 °C, 400 °C, 500 °C, 600 °C and 700 °C), and labelled as xCuC@yTi-z where x represents the copper content, y represents the titanium content and z represents the temperature of calcination.

Catalyst characterization

X-ray diffraction (XRD) measurements were performed using a PANalytical B.V. Empyrean diffractometer with Cu Kα radiation operated at 40 kV, 40 mA. The scanning range (2θ) was 10–80°. The morphology of the samples was investigated using an FEI Tecnai S-Twin transmission electron microscope (TEM). Light absorbance was measured using a UVIKON/XL UV-vis diffuse reflectance spectrometer (UV-vis) with a scanning range of 200–800 nm. The metal content was measured by energy-dispersive spectrometry (EDS) using a Bruker-QUANTAX 200 microanalysis system attached to a Hitachi S-4800 field emission scanning electron microscope (SEM). Temperature programmed reduction with hydrogen (H2-TPR) was performed with a Micromeritics AutoChem 2910 analyzer. The gas mixture is 10% H2 balanced with N2 and the flow rate is 50 mL min−1. The heating rate is 10 °C min−1. 10 mg of sample was loaded in the quartz tube for each run. The X-ray photoelectron spectra (XPS) of the catalysts were recorded with a Kratos Amicus spectrometer using an Al Kα (1486.6 eV) radiation source. The binding energy (BE) was adjusted by the C1s transition at 284.6 eV. The light intensity was measured using a UV Integrator (UV-BIKESU). The typical process was as follows. Firstly, a 4 mm UV probe was installed under the quartz window used for the reaction, and the light source was fixed at a specific distance over the quartz window. Secondly, the light was turned on and timing was started. The light intensity was then obtained using the measured power accumulation divided by the area of the probe and the time of irradiation. The light intensity can be modulated by adjusting the distance between the light source and the quartz window.

Photocatalytic reaction

The photocatalytic activity of the catalyst was measured in a continuous flow magnesium–aluminium alloy reactor with a rectangle quartz window on the top and a dividing wall-type heat exchanger connected to the back of the reactor.18 The dimensions of the reactor can be seen in the ESI (Fig. S1). Three pieces of rectangle glass which were used as the catalyst holders were installed at the bottom of the reactor, with a thermocouple fixed in contact with the middle catalyst holder. 0.02 g of catalyst was loaded on the holders for each run. The dimensions of each piece of the glass holders were 26 mm × 80 mm × 1 mm. Cooling water was flowed through the heat exchanger to maintain a constant temperature. A 500 W high pressure mercury lamp (CEL-LAM500) with a peak wavelength of 365 nm without a filter was installed in a quartz cooling jacket which was positioned 2 cm over the quartz window of the reactor. A gas mixture containing 1–3 vol% methanol and 0.17–20 vol% O2 balanced with N2 was supplied at a flow rate of 50 ml min−1 into the reactor. Oxygen and nitrogen in the mixture were measured by mass flow meters. The reaction temperature was from 15 to 45 °C. The products were qualified using a GC-MS and an LC-MS in batches and quantified on line using a Shimadzu GC2014C equipped with an FID detector.

The methanol conversion was obtained using eqn (1), assuming that the volume flow rate was constant before and after the reaction due to the low reactant content in the feed gas.

 
image file: c9cy01595g-t1.tif(1)
where C is the methanol conversion, %; ρM0 is the initial methanol content, mg L−1; ρM1 is the methanol content in the off-gas after the reaction, mg L−1.
 
image file: c9cy01595g-t2.tif(2)
where S is the methyl formate selectivity, %; ρMF is the methyl formate content in the off-gas after the reaction, mg L−1.

The formation rate of methyl formate was calculated using eqn (3).

 
image file: c9cy01595g-t3.tif(3)
where Fr is the formation rate, mmol g−1 h−1; ρM0 is the initial methanol content in the feed gas, mg L−1; V is the flow rate of the feed gas, L h−1; C is the methanol conversion, %; S is the methyl formate selectivity, %; M is the molecular weight of methanol, g mol−1; m is the weight of the catalyst, g.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This project was funded by the National Natural Science Foundation of China (No. 21566026 and No. 21766020), the Natural Science Foundation of Inner Mongolia (No. 2019MS02011), the continuously supporting project of Grassland Talent of Inner Mongolia and the major basic research and open project of the Inner Mongolia Autonomous Region (30500-515330303).

Notes and references

  1. R. Blume, M. Hävecker, S. Zafeiratos, D. Teschner, A. Knop-Gericke, R. Schlögl, P. Dudin, A. Barinov and M. Kiskinova, Catal. Today, 2007, 124, 71–79 CrossRef CAS .
  2. J. A. Labinger and J. E. Bercaw, Nature, 2002, 417, 507–514 CrossRef CAS .
  3. Z. Guo, B. Liu, Q. Zhang, W. Deng, Y. Wang and Y. Yang, Chem. Soc. Rev., 2014, 43, 3480–3524 RSC .
  4. G. Palmisano, E. Garcia-Lopez, G. Marci, V. Loddo, S. Yurdakal, V. Augugliaro and L. Palmisano, Chem. Commun., 2010, 46, 7074–7089 RSC .
  5. K. Czelej, K. Cwieka, J. C. Colmenares, K. J. Kurzydlowski and Y. J. Xu, ACS Appl. Mater. Interfaces, 2017, 9, 31825–31833 CrossRef CAS PubMed .
  6. S. P. Tonner, D. L. Trimm, M. S. Wainwright and N. W. Cant, Ind. Eng. Chem. Prod. Res. Dev., 1984, 23, 384–388 CrossRef CAS .
  7. E. Tronconi, A. S. Elmi, N. Ferlazzo and P. Forzatti, Ind. Eng. Chem. Res., 1987, 26, 1269–1275 CrossRef CAS .
  8. A. S. Elmi, E. Tronconi, C. Cristiani, J. P. Gomez Martin and P. Forzatti, Ind. Eng. Chem. Res., 1989, 28, 387–393 CrossRef CAS .
  9. P. J. A. Tijm, F. J. Waller and D. M. Brown, Appl. Catal., A, 2001, 221, 275–282 CrossRef CAS .
  10. J. S. Lee, J. C. Kim and Y. G. Kim, Appl. Catal., 1990, 57, 1–30 CrossRef CAS .
  11. J. Y. Johnson, British Pat., 252848, 1925 Search PubMed .
  12. S. P. Tonner, D. L. Trlmm and M. S. Walnwrlght, Ind. Eng. Chem. Prod. Res. Dev., 1984, 23, 384–388 CrossRef CAS .
  13. M. Ai, J. Catal., 1982, 77, 279–288 CrossRef CAS .
  14. H. Liu and E. Iglesia, J. Phys. Chem. B, 2005, 109, 2155–2163 CrossRef CAS .
  15. G. A. Olah, Angew. Chem., Int. Ed., 2005, 44, 2636–2639 CrossRef CAS PubMed .
  16. A. Wittstock, V. Zielasek, J. Biener, C. M. Friend and M. Bäumer, Science, 2010, 327, 319–322 CrossRef CAS PubMed .
  17. G. T. Whiting, S. A. Kondrat, C. Hammond, N. Dimitratos, Q. He, D. J. Morgan, N. F. Dummer, J. K. Bartley, C. J. Kiely, S. H. Taylor and G. J. Hutchings, ACS Catal., 2014, 5, 637–644 CrossRef .
  18. C. Han, X. Yang, G. Gao, J. Wang, H. Lu, J. Liu, M. Tong and X. Liang, Green Chem., 2014, 16, 3603 RSC .
  19. J. Liu, C. Han, X. Yang, G. Gao, Q. Shi, M. Tong, X. Liang and C. Li, J. Catal., 2016, 333, 162–170 CrossRef CAS .
  20. X. Yang, A. Zhang, G. Gao, D. Han, C. Han, J. Wang, H. Lu, J. Liu and M. Tong, Catal. Commun., 2014, 43, 192–196 CrossRef CAS .
  21. J. C. Colmenares, P. Lisowski, D. Łomot, O. Chernyayeva and D. Lisovytskiy, ChemSusChem, 2015, 8, 1676–1685 CrossRef CAS PubMed .
  22. H. Kominami, H. Sugahara and K. Hashimoto, Catal. Commun., 2010, 11, 426–429 CrossRef CAS .
  23. Y. Liu, G. Griffin, S. Chan and I. Wachs, J. Catal., 1985, 94, 108–119 CrossRef CAS .
  24. X. Liang, X. Yang, G. Gao, C. Li, Y. Li, W. Zhang, X. Chen, Y. Zhang, B. Zhang, Y. Lei and Q. Shi, J. Catal., 2016, 339, 68–76 CrossRef CAS .
  25. P. Lisowski, J. C. Colmenares, D. Łomot, O. Chernyayeva and D. Lisovytskiy, J. Mol. Catal. A: Chem., 2016, 411, 247–256 CrossRef CAS .
  26. G. Ramis, G. Busca and V. Lorenzelli, J. Chem. Soc., Faraday Trans. 1, 1987, 83, 1591–1599 RSC .
  27. K. Sing, K. Sing, D. Everett, R. Haul, L. Moscou, R. Pierotti, J. Rouquerol and T. Siemieniewska, Pure Appl. Chem., 1982, 54, 2201 Search PubMed .
  28. J. Zhang, P. Zhou, J. Liu and J. Yu, Phys. Chem. Chem. Phys., 2014, 16, 20382–20386 RSC .
  29. H. Irie, S. Miura, K. Kamiya and K. Hashimoto, Chem. Phys. Lett., 2008, 457, 202–205 CrossRef CAS .
  30. X. Qiu, M. Miyauchi, K. Sunada, M. Minoshima, M. Liu, Y. Lu, D. Li, Y. Shimodaira, Y. Hosogi and Y. Kuroda, ACS Nano, 2012, 6, 1609–1618 CrossRef CAS PubMed .
  31. X. Sun and Y. Li, Chem, 2003, 9, 2229–2238 CrossRef CAS PubMed .
  32. S. Tsunekawa, T. Fukuda and A. Kasuya, J. Appl. Phycol., 2000, 87, 1318 CrossRef CAS .
  33. H. Wang, J.-Z. Xu, J.-J. Zhu and H.-Y. Chen, J. Cryst. Growth, 2002, 244, 88–94 CrossRef CAS .
  34. S. Banerjee and D. Chakravorty, Europhys. Lett., 2000, 52, 468 CrossRef CAS .
  35. J. Y. Kim, J. A. Rodriguez, J. C. Hanson, A. I. Frenkel and P. L. Lee, J. Am. Chem. Soc., 2003, 125, 10684–10692 CrossRef CAS PubMed .
  36. F. Boccuzzi, A. Chiorino, G. Martra, M. Gargano, N. Ravasio and B. Carrozzini, J. Catal., 1997, 165, 129–139 CrossRef CAS .
  37. J. C. Wu, I.-H. Tseng and W.-C. Chang, J. Nanopart. Res., 2001, 3, 113–118 CrossRef CAS .
  38. G. V. Sagar, P. V. R. Rao, C. S. Srikanth and K. V. Chary, J. Phys. Chem. B, 2006, 110, 13881–13888 CrossRef CAS PubMed .
  39. N. McIntyre and M. Cook, Anal. Chem., 1975, 47, 2208–2213 CrossRef CAS .
  40. F. Parmigiani, G. Pacchioni, F. Illas and P. Bagus, J. Electron Spectrosc. Relat. Phenom., 1992, 59, 255–269 CrossRef CAS .
  41. C. D. Wagner, D. A. Zatko and R. H. Raymond, Anal. Chem., 1980, 52, 1445–1451 CrossRef CAS .
  42. F. P. Koffyberg, J. Appl. Phys., 1982, 53, 1173 CrossRef CAS .
  43. Y. Xu and M. A. Schoonen, Am. Mineral., 2000, 85, 543–556 CrossRef CAS .
  44. V. Gombac, L. Sordelli, T. Montini, J. J. Delgado, A. Adamski, G. Adami, M. Cargnello, S. Bernal and P. Fornasiero, J. Phys. Chem. A, 2010, 114, 3916–3925 CrossRef CAS PubMed .
  45. Z. Wang, Y. Liu, D. J. Martin, W. Wang, J. Tang and W. Huang, Phys. Chem. Chem. Phys., 2013, 15, 14956–14960 RSC .
  46. S. N. Habisreutinger, L. Schmidt-Mende and J. K. Stolarczyk, Angew. Chem., Int. Ed., 2013, 52, 7372–7408 CrossRef CAS PubMed .
  47. D. Praveen Kumar, M. V. Shankar, M. M. Kumari, G. Sadanandam, B. Srinivas and V. Durgakumari, Chem. Commun., 2013, 49, 9443–9445 RSC .
  48. C. Ampelli, R. Passalacqua, C. Genovese, S. Perathoner, G. Centi, T. Montini, V. Gombac, J. J. Delgado Jaen and P. Fornasiero, RSC Adv., 2013, 3, 21776–21788 RSC .
  49. M. Hara, T. Kondo, M. Komoda, S. Ikeda, K. Shinohara, A. Tanaka, J. N. Kondoa and K. Domen, Chem. Commun., 1998, 357–358 RSC .
  50. X. An, K. Li and J. Tang, ChemSusChem, 2014, 7, 1086–1093 CrossRef CAS PubMed .
  51. P. D. Tran, L. H. Wong, J. Barber and J. S. C. Loo, Energy Environ. Sci., 2012, 5, 5902 RSC .
  52. H. Li, Y. Lei, Y. Huang, Y. Fang, Y. Xu, L. Zhu and X. Li, J. Nat. Gas Chem., 2011, 20, 145–150 CrossRef CAS .
  53. X. Hua, O. Shuxin, L. Lequan, W. Defa, K. Tetsuya and Y. Jinhua, Nanotechnology, 2014, 25, 165402 CrossRef PubMed .
  54. X. Li, J. Wen, J. Low, Y. Fang and J. Yu, Sci. China Mater., 2014, 57, 70–100 CrossRef .
  55. A. D. Handoko and J. Tang, Int. J. Hydrogen Energy, 2013, 38, 13017–13022 CrossRef CAS .
  56. P. E. de Jongh, D. Vanmaekelbergh and J. J. Kelly, Chem. Commun., 1999, 1069–1070 RSC .
  57. A. V. Walker and J. T. Yates, J. Phys. Chem. B, 2000, 104, 9038–9043 CrossRef CAS .
  58. A. Paracchino, V. Laporte, K. Sivula, M. Grätzel and E. Thimsen, Nat. Mater., 2011, 10, 456–461 CrossRef CAS PubMed .
  59. C. Li, X. Liang, Y. Li, W. Zhang, X. Chen, G. Gao and X. Yang, The 16th Inter. Congr. Catal., 2016, PE110 Search PubMed.
  60. K. R. Phillips, S. C. Jensen, M. Baron, S.-C. Li and C. M. Friend, J. Am. Chem. Soc., 2013, 135, 574–577 CrossRef CAS PubMed .
  61. Q. Guo, C. Xu, W. Yang, Z. Ren, Z. Ma, D. Dai, T. K. Minton and X. Yang, J. Phys. Chem. C, 2013, 117, 5293–5300 CrossRef CAS .
  62. R. B. Evans, G. M. Watson and E. A. Mason, J. Chem. Phys., 1961, 35, 2076–2083 CrossRef CAS .
  63. J. Gilron and A. Soffer, J. Membr. Sci., 2002, 209, 339–352 CrossRef CAS .
  64. M. F. L. Johnson and W. E. Stewart, J. Catal., 1965, 4, 248–252 CrossRef CAS .

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c9cy01595g

This journal is © The Royal Society of Chemistry 2019