Gas-generated thermal oxidation of a coordination cluster for an anion-doped mesoporous metal oxide

Central in material design of metal oxides is the increase of surface area and control of intrinsic electronic and optical properties, because of potential applications for energy storage, photocatalysis and photovoltaics. Here, we disclose a facile method, inspired by geochemical process, which gives rise to mesoporous anion-doped metal oxides. As a model system, we demonstrate that simple calcination of a multinuclear coordination cluster results in synchronic chemical reactions: thermal oxidation of Ti8O10(4-aminobenzoate)12 and generation of gases including amino-group fragments. The gas generation during the thermal oxidation of Ti8O10(4-aminobenzoate)12 creates mesoporosity in TiO2. Concurrently, nitrogen atoms contained in the gases are doped into TiO2, thus leading to the formation of mesoporous N-doped TiO2. The mesoporous N-doped TiO2 can be easily synthesized by calcination of the multinuclear coordination cluster, but shows better photocatalytic activity than the one prepared by a conventional sol-gel method. Owing to an intrinsic designability of coordination compounds, this facile synthetic will be applicable to a wide range of metal oxides and anion dopants.

Geochemical process coupled with gas generation is of great importance to the evolution of natural porous minerals. The porosity in the minerals is created by evaporation of gas bubbles. The gases comprised mostly of water steam, carbon dioxide but also contains a small amount of hydrogen sulphide, hydrogen fluoride and ammonia 1 . The anions in those gases react with minerals to be incorporated as anionic partners for metal ions 2 . Consequently, incorporation of anions and void formations in the minerals simultaneously occur, giving rise to natural porous minerals containing anions such as sulphur, fluorine and nitrogen.
Porous metal oxides represent promising materials for energy storage 3 , photocatalysis [4][5] , and photovoltaics 6-7 because of the large active surface area. By contrast, the control of chemical composition in metal oxides is also vital to these applications. In particular, incorporation of another anion into metal oxides, i.e. anion doping, provides excellent performance in ion-storage 8 and photocatalytic reaction [9][10] . However, synthesis of porous metal oxides and anion doping have been individually developed. In that context, a crucial challenge in this research field is to coherently integrate these two processes. These considerations inspire us to mimic the geochemical process to establish a facile synthetic method for anion-doped porous metal oxides.
Coordination compounds, wherein metal ions and organic ligands are rationally varied [11][12][13][14][15][16] , are candidates for precursor to apply the gas-generated thermal oxidation. Indeed, coordination compounds are thermally oxidized into metal oxides by calcination [17][18][19][20] . On the other hand, organic molecules are fragmented to generate gases by intense heating [21][22] . In particular, gases containing reactive anions are generated by the fragmentation of organic functional groups, which potentially act as dopant sources. In general, however, the organic ligands of coordination compounds are removed by heating before reaching temperatures where metal oxides are formed. Because of the temperature gap, a calcination of coordination compounds gives metal oxides even without anion doping.
Our strategy to overcome the problem is to improve thermal stability of organic ligands by robust coordination bonding 23 of carboxylates with a multinuclear metal cluster. As a model system, we design a multinuclear titanium coordination cluster comprised of a carboxylate ligand with a pendant amino-group. The carboxylate ligand is anchored by coordination bonding with the multinuclear titanium cluster until formation of metal oxides. Therefore, fragmentation of amino-group overlaps with thermal oxidation of the titanium coordination cluster. Consequently, TiO 2 is formed under evaporation of gases containing nitrogen atoms, giving rise to N-doped TiO 2 24-26 with permanent porosity. In other words, the porous N-doped TiO 2 can be obtained by a simple calcination of the coordination cluster.
Metal oxides doped with anion 27-29 has attracted much attention because of potential applications of visible-light photocatalyst for water splitting 30 , pollutant degradation 31-32 and solar energy conversion [33][34] . Porosity further improves the photocatalytic activity by increasing a surface area and improving the accessibility to catalytic active sites 35 . The mesoporous metal oxide has been fabricated by elaborate protocols, including templating method 36-37 , or particle assembly [38][39] . Sol-gel method is rather simple to synthesize mesoporous metal oxide, which can be easily combined with anion doping [40][41] . However, synthesis of mesoporous metal oxides via sol-gel method requires precise control of hydrolysis and condensation rates, which would conflict with anion doping approach. From simplicity of the protocol, calcination of coordination clusters will be an attractive strategy to fabricate anion-doped porous metal oxides (Fig. 1). Notably, coordination compounds can be rationally designed by a judicious choice of metal ions and organic ligands 42 . Therefore, the strategy presented here will be applicable to other types of metal oxides and anion dopants.

Results
We synthesized a titanium coordination cluster with 4-amino benzoic acid. A solvothermal reaction of titanium isopropoxide and 4-amino benzoic acid in acetonitrile gave cuboid crystals with a size of several hundred μ m. The resulting compound of Ti 8 O 10 (4-aminobenzoate) 12 (1) consists of Ti 8 O 10 cluster, where octanuclear titanium is linked by ten μ 2 -oxo bridges. The carboxyl groups of twelve 4-aminobenzoate further bridge each titanium to each of its neighbouring titanium in a bidentate fashion (Fig. 2a-c).
X-ray photon spectroscopy (XPS) was carried out to clarify the incorporation of nitrogen atoms in TiO 2 . A broad XPS peak of N 1s was observed in TiO 2 -(1) but not in TiO 2 -(2), suggesting that nitrogen in TiO 2 -(1) is originating from the amino group of 4-aminobenzoate ( Figure S2). The binding energy of N 1s (398 eV) corresponded to anionic N − in Ti-O-N which is in the range typically observed for substitutional nitrogen doping into TiO 2 [44][45][46] . Furthermore, the binding energies of Ti 2p1/2 (464 eV) and Ti 2p3/2 (459 eV) well matched with those of Ti in N-doped TiO 2 47 (Fig. 3b,c). As shown in Figure S4, Raman spectra of TiO 2 -(1) and TiO 2 -(2) showed the characteristic blue shift of E g(1) band by nitrogen doping (139.6 cm −1 for TiO 2 -(2) and 144.0 cm −1 for TiO 2 -(1)) 48 . These results suggested that nitrogen originating from amino group was incorporated into TiO 2 as a dopant, giving rise to N-doped TiO 2 . The nitrogen concentration in TiO 2 -(1) was estimated as 0.96%.
The resulting TiO 2 -(1) is yellow because of the nitrogen doping, whereas non-doped TiO 2 , including TiO 2 -(2), is white (Fig. 3d). As expected, TiO 2 -(1) showed absorption in the visible-light region (400-500 nm), but TiO 2 -(2) absorbs only light in ultraviolet (UV) region (Fig. 3e). This is because nitrogen doping into TiO 2 created a new energy level (N 2p level) above the valence band maximum. The new absorption band in 400-450 nm corresponds to the energy gap between conductance band and N 2p level (2.7 eV). These results suggest that TiO 2 -(1) is able to work as photocatalyst under visible light. Besides the nitrogen dope into TiO 2 , the porosity of TiO 2 -(1) and TiO 2 -(2) was evaluated by N 2 adsorption ( Figure S5a). The adsorption/desorption hysteresis was observed for TiO 2 -(1) and TiO 2 -(2) in the relative pressure (P/P 0 ) range of 0.4-0.9. This characteristic hysteresis is attributed to the mesopores of TiO 2 . The gradual adsorption in the hysteresis region, classified as H2 type adsorption, suggested mesopores with ununiform size and shape. The pore-size distribution, based on the desorption branch of the isotherm, was estimated by Barret, Joyner, and Halender (BJH) method, assuming a cylindrical pore model. The pore sizes of TiO 2 -(1) and TiO 2 -(2) were calculated to be around 4 nm ( Figure S5b). The mesopores of TiO 2 -(1) were also observed by TEM ( Figure S6). BET surfaces of TiO 2 -(1) and TiO 2 -(2) were estimated as 170.6 m 2 /g and 139.8 m 2 /g, which were relatively large compared to those of metal oxides prepared by calcination of coordination compounds 49,50 .
The series of measurements indicated that simple calcination of the coordination cluster allows the synthesis of mesoporous N-doped TiO 2 . To investigate the formation mechanism of mesoporous N-doped TiO 2 , variable-temperature XRD (VT-XRD) and thermogravimetry with differential thermal analysis (TG-DTA) were carried out. As seen in VT-XRD, 1 was decomposed and the formation of TiO 2 began over 400 °C (Fig. 4a). This result of VT-XRD was well matched with that of TG-DTA. TG-DTA showed the weight loss over 250 °C because of evaporation of acetonitrile. Note that the exothermic peak was observed in DTA over 350 °C (Fig. 4b). The exothermic peak is ascribed to the oxidation of titanium coordination clusters to form TiO 2 . The results of VT-XRD and TG-DTA suggested that TiO 2 began to be crystallized over 350-400 °C.
Quadrupol mass spectroscopy (Q-MS) of 1 under heating further gave the insight into gas generation and mechanism of nitrogen doping. As seen in Fig. 4c, the gases of benzene, aniline, HNO 3 and CO 2 were generated in the temperature region of 300-480 °C, suggesting the decomposition of 4-aminobenzoate. The organic ligand was decomposed to generate gases concurrently with the formation of TiO 2 . In other words, TiO 2 was crystalized during the generation of gases.
The decomposition of 4-aminobenzoate into benzene indicates that the covalent bond between the amino-group and phenyl ring was cleaved to generate the fragments containing nitrogen atoms (N-fragment). The generation of N-fragment was also confirmed by the detection of HNO 3 . HNO 3 was most likely formed by the oxidation of N-fragments. The rest of N-fragments reacted with TiO 2 and nitrogen atoms were incorporated into TiO 2 as a dopant.
This synchronic reaction was also observed in 2. 2 was decomposed to begin the formation of TiO 2 over 400 °C, which was characterized by VT-XRD and DT-XRD ( Figure S7 and Figure S8). Q-MS measurement of 2 showed that benzoate was decomposed into gases of benzene and CO 2 in 300-480 °C. Gas generation and formation of TiO 2 were overlapped in the temperature range of 350-480 °C ( Figure S9). Nitrogen was not doped into TiO 2 -(2) because of no nitrogen source (amino group) in the starting material of 2. However, gas generation during the formation of TiO 2 also resulted in the formation of mesoporous TiO 2 ( Figure S10).
Based on VT-XRD, TG-DTA, and Q-MS, we propose following the reaction mechanism of nitrogen doping. Ti 8 O 10 (4-aminobenzoate) 12 was decomposed to form TiO 2 over 350 °C. 4-aminobenzoate of 1 was decomposed into the gases of aniline, benzene, CO 2 and N-fragments. Nitrogen atoms in N-fragments reacted with TiO 2 to be incorporated into TiO 2 as a dopant, forming N-doped TiO 2 (Fig. 5(i) molecular scale). The gases, including CO 2 , benzene, were generated concurrently with the formation of TiO 2 . Thus, gas evaporation during the formation of TiO 2 created internal voids, leading to the formation of mesoporous N-doped TiO 2 (Fig. 5(ii) mesoscale). As mentioned above, the surface area of TiO 2 -(1) and TiO 2 -(2) are larger than the metal oxides synthesized by calcination of extended coordination frameworks 50 . This is because the gas generation synchronized with formation of TiO 2 created mesopores and significantly increased the surface area.
To evaluate the advantage of the new synthetic method, we synthesized mesoporous N-doped TiO 2 by a sol-gel method as a reference (TiO 2 -sg) 40,41 . Isopropanol solution of titanium isopropoxide was mixed with aqueous solution of urea and nitric acid to prepare precursor sol. The resulting sol was calcined to synthesize mesoporous N-doped TiO 2 . The nitrogen originating from urea was doped into TiO 2 . The mesoporosity and BET surface were evaluated by N 2 adsorption ( Figure S10). The mesoporosity is attributed to the interparticle voids as described in previous literatures 41 . As shown in Table S1, The BET surfaces of TiO 2 -(1) and TiO 2 -(2) were more than twice as large as that of  Figure S11 and Table S2). However, the concentration of nitrogen in TiO 2 -sg was slightly higher than TiO 2 -(1) (Figure S12-13).
We evaluated the visible-light photocatalytic activity of TiO 2 -(1), TiO 2 -(2) and TiO 2 -sg by degradation of methylene blue (MB) 51,52 . The crystals of TiO 2 -(1), TiO 2 -(2) or TiO 2 -sg were placed in a solution of MB and vigorously stirred under visible-light irradiation (> 410 nm). The absorption intensity of MB decreased over time, showing the photocatalytic activity of TiO 2 -(1) and TiO 2 -sg for the degradation of MB (Fig. 6 and Figure S14). The decrease rate of TiO 2 -(2) and no catalysts were nearly same, suggesting that the intensity decrease of MB was attributed to not the adsorption of MB on TiO 2 particles but the photocatalytic decomposition of MB. Although the nitrogen concentration of TiO 2 -sg was higher than TiO 2 -(1), TiO 2 -(1) decomposed MB much faster than TiO 2 -sg. MB was completely decomposed by TiO 2 -(1) in 150 min, while only a half amount of MB was decomposed by TiO 2 -sg. Since nitrogen concentration of TiO 2 -(1) is lower than TiO 2 -sg, the rapid degradation of MB is most likely attributed to the large surface area of TiO 2 -(1). The mesoporous N-doped TiO 2 can be easily prepared by calcination of the coordination cluster, but shows better photocatalytic activity than the one synthesized by a conventional sol-gel method.

Conclusion
In this contribution, we demonstrate a facile method for the synthesis of mesoporous anion-doped metal oxides. As a model system, we synthesized a multinuclear titanium coordination cluster with a pendant amino-group. A simple calcination of the coordination cluster resulted in synchronic reactions: thermal oxidation of the coordination cluster into TiO 2 and gas generation including N-fragments. The gas generation during the formation of TiO 2 allows the introduction of mesopores. Furthermore, nitrogen atoms in N-fragments reacted with TiO 2 to be incorporated as nitrogen dopant, thus leading to the formation of mesoporous N-doped TiO 2 . The resulting mesoporous N-doped TiO 2 showed photocatalytic activity under visible light better than TiO 2 prepared by a conventional sol-gel method, because of its larger surface area.
Notably, coordination clusters can be rationally designed by a choice of metal ions and organic ligands. Besides, doping amount can be potentially controlled by optimizing calcination conditions of coordination clusters ( Figure S15). The synthetic and calcination protocols of the coordination clusters do not require specialized instruments. Therefore, coordination clusters as precursors will be a promising method for anion-doped porous metal oxides, which will offer significant benefits for the fabrication of light emitting diodes, ion storage batteries and heterogeneous catalysts.

Synthesis of Ti 8 O 8 (benzoate)
A halogen lamp (SX-UI502M, USHIO SPAX INC.) was used as the light source. 400 nm cut-off filter was placed in front of the reactor.

X-ray Photon Spectroscopy (XPS).
Dried powders of TiO 2 -(1) and TiO 2 -(2) were placed on a carbon conductive tape to avoid the powders from swirling in the air. XPS data were collected by JEOL Ltd. JPS-9200. N 2 Gas Adsorption. N 2 adsorption measurements were carried out by Quantachrome Autosorb 6AG. The BET surface area was determined by the multipoint BET method using the adsorption branch in the relative pressure (P/P 0 ) range of 0.05-0.3. The pore-size distribution was estimated by applying Barret, Joyner, and Halender (BJH) method to the desorption branch of the isotherms. Powder X-ray Diffraction (XRD). PXRD data were collected by Bruker D8 Advance ECO. Scherrer equation is applied to 110 diffraction of anatase TiO 2 to estimate the average size of crystallite for TiO 2 -(1), TiO 2 -(2) and TiO 2 -sg. The instrumental broadening estimated by a standard sample (Al 2 O 3 ) is 0.042.