The oxidation of toluene to benzaldehyde by O₂ has been one of the reactions highly demanded in industry in the past decades [1-4] and it should be the best route to produce benzaldehyde, a highly useful chemical in many aspects. The reaction seems simple, just an insertion of one oxygen into the C–H bond of benzyl, but very difficult, for the heterogeneous catalytic processes for the reaction have been studied for several decades, of which performance, however, is still far below the demand [3, 5]. The present industrial processes have serious environmental and corrosion problems, as shown in Scheme 1, and thus produced benzaldehyde inevitably contained a little amount of halogens [6], which hindered high-end uses of the compound such as in perfumes or foods. Liquid oxidations of toluene by oxygen with noble metals based catalysts including Pd, Au, Pt, Ag, Ru and their alloys have become popular research topics in recent years, owing to their excellent catalytic activities [7-14], under solvents or solvent-free conditions [1, 3, 15-20]. Ag modified mixed-node MOF catalyst has been found highly selective for the oxidation of toluene to benzaldehyde, but just at very low conversions [8]. Bimetallic PdAu catalysts reported by Hutchings and co-workers for the oxidation of toluene toward benzyl benzoate do get outstanding activity at a relatively mild temperature (160 ℃, 1 MPa O₂) [4, 21, 22]. Generally, high selectivity to benzaldehyde at high toluene conversion has been still not available with these catalytic systems.

For the green oxidation of toluene by O₂ exclusively to benzaldehyde, a highly efficient catalytic system was explored in our previous work using a biphasic system with HDPA-FeO_x nanoparticles locating themselves at the interface of toluene and water to stabilize the O/W emulsion, i.e, Pickering emulsion [2]. Resasco et al. [23-25] have made the first use of Pickering emulsions for the biphasic hydrodeoxygenation and condensation in biomass refining as well as other reactions. We report here the further investigation on the doping effect of a series of metal oxides such as Mn, Co, Ni, Cu, Cr, Mo, V, and Ti on the catalyst HDPA-FeO_x, anticipating the enhancement of the mobility of lattice oxygen of the catalyst. In addition, we also use a special kind of nano alumina as support for the easy preparation of the modified catalysts and the high stability of the catalysts in application. With the well-defined particle structures, the catalysts show significantly improved and reliably repeated catalytic performance for the title reaction and the NiO co-loaded catalyst exhibits the best performance with ~83% conversion of toluene (a TOF of 0.027 nm^–2·s^–1) and ~100% selectivity to benzaldehyde under mild conditions. This is a very important progress toward the practical use of the catalytic process for production of high quality benzaldehyde.

Fig. 1 schematically shows the composition of the Pickering emulsion with the catalyst nanoparticles positioned at the interface of O/W. The long organic chains of HDPA molecules in a suitable density protect the nanoparticles from gathering and ensure the catalyst particles at the interface of the O/W emulsion, boosting the access of toluene to the nanoparticles. The phosphate head in HDPA adsorbed on the surface is important to tune the catalyst surface as well as to switch off unneeded sites unselective for the toluene oxidation. Considering the solubility of the reactants, the water has a significant effect on the transportation of O₂ from gas phase to the liquid phase as well as the products from the surface of the catalyst to the liquid phase (Figs. 1(a), 1(b), and 1(c)). Fig. 1(d) depicts the real laser confocal fluorescence photograph of the reaction mixture with small amount of rhodamine B as fluorescence dye adsorbed on the catalyst, confirming the O/W Pickering emulsion of the reactive system. Under such conditions, the surface-bonded HDPAs boost the access of the nonpolar toluene as well as the seceding of the relatively polar product, i.e., benzaldehyde, from the nonpolar surface to the polar environment, constituting a complete catalytic cycle of high performance.

The nano γ-Al₂O₃ used as support in current investigation was obtained according to our previous report [26]. Fe₂O₃/Al₂O₃ sample was prepared by incipient wet impregnation method with corresponding aqueous solution of Fe(NO₃)₃·9H₂O. The 25 wt% loading of Fe₂O₃ was set as the weight ratio of Fe₂O₃/(Fe₂O₃ + Al₂O₃). The impregnating mixture was sealed and aged at room temperature for 24 h. The resulting material was dried at 90 ℃ overnight and calcined in a tube furnace at 450 ℃ for 3 h in flowing air. The sample was labeled as 25Fe₂O₃/Al₂O₃. The figure before the metal oxides meant the content in weight percent of the transition metal oxide in the catalyst.

The (Fe₂O₃-MO_x)/Al₂O₃ (M = Mn, Co, Ni, Cu, Cr, Mo, V, Ti) samples were also prepared by incipient wet impregnation method. The precursors of the doped oxides were Mn(NO₃)₂, Co(NO₃)₂·6H₂O, Ni(NO₃)₂·6H₂O, Cu(NO₃)₂·3H₂O, Cr(NO₃)₃·9H₂O, (NH₄)₂MoO₄·4H₂O, NH₄VO₃, and tetrabutyl titanate, respectively. The mass ratio of Fe₂O₃ to MO_x was 4 : 1, the loadings of (Fe₂O₃-MO_x) in all samples were shown as the weight ratio of (Fe₂O₃-MO_x)/(Fe₂O₃-MO_x + Al₂O₃) and fixed at 25 wt%. The treatment of these mixtures was similar to Fe₂O₃/Al₂O₃ sample. These sample names were abbreviated as Fe₂₀Mn₅, Fe₂₀Co₅, Fe₂₀Ni₅, Fe₂₀Cu₅, Fe₂₀Cr₅, Fe₂₀Mo₅, Fe₂₀V₅ (oxalate was used to help the dissolution of NH₄VO₃.), and Fe₂₀Ti₅ (HCl was used to inhibit the hydrolysis of tetrabutyl titanate during preparation.), respectively. In addition, a series of (Fe₂O₃-xNiO)/Al₂O₃ or (Fe₂O₃-xCuO)/Al₂O₃ catalysts was prepared with different doping amount of Ni or Cu and the samples were correspondingly named as Fe₂₀Ni₅, Fe₁₅Ni₁₀, Fe₁₀Ni₁₅, Ni₂₅, Fe₂₀Cu₅, Fe₁₅Cu₁₀, Fe₁₀Cu₁₅, and Cu₂₅.

The decoration of HDPA was performed using a modified hydrothermal method. Taking the preparation of HDPA-25Fe₂O₃/Al₂O₃ as an example, certain amount of HDPA (Sigma-Aldrich) were added into a 100 mL beaker containing 40 mL of deionized water with vigorous stirring at 80 ℃ until HDPA completely dissolved and then corresponding amount of 25Fe₂O₃/Al₂O₃ powder was added into the above mixture. After stirring for 2 h, the solution was transferred into a Teflon-lined autoclave and heated at 120 ℃ for 24 h. After that, the mixture was washed with deionized water and absolute ethanol several times and filtered by suction. The achieved solid powders were dried at 80 ℃ overnight. For convenience, the final catalysts were hereafter denoted as H-Fe₂₅, H-Fe₂₀Mn₅, H-Fe₂₀Co₅, H-Fe₂₀Ni₅, H-Fe₂₀Cu₅, H-Fe₂₀Cr₅, H-Fe₂₀Mo₅, H-Fe₂₀V₅, and H-Fe₂₅Ti₅, respectively. For all of the samples, the HDPA loading in the catalysts was expressed as the weight ratio of HDPA/((Fe₂O₃-MO_x)/Al₂O₃) or HDPA/(25Fe₂O₃/Al₂O₃) and fixed at ~5 wt% (the density of surface HPDA was about ~1 HDPA/nm² of the sample). The process of catalyst preparation is schematically shown as Fig. 2. The nano alumina used as support brings about great convenience in catalyst preparation as well as the stability in performance of the catalyst in reaction. The names and compositions of all the catalysts investigated are elucidated in the Table 1 for clarity.

The X-ray diffraction (XRD) patterns were recorded using a XRD-6000 X-ray diffractometer (Shimadzu, Japan) with Cu Kα radiation (0.15418 nm) as X-ray source. The voltage and current were set at 40 kV and 30 mA, respectively. The surface area and textural characteristics of the catalysts were measured by N₂ sorption using a Micrometrics TriStar II 3020 analyzer. Transmission electron microscopy (TEM) images of catalysts were obtained by using a JEOL JEM-2100 at a voltage of 180 kV. Thermogravimetric analysis and derivative thermogravimetry (TG-DTA) analysis was conducted on a STA 449C apparatus (Netzsch) in air from room temperature to 800 ℃ with a heating rate of 10 ℃/min. Inductively coupled plasma-atomic emission spectrometry (ICP-AES) was carried out to analyze the leached metal from the catalysts in the solution after the reaction using an instrument of Optima 5300 DV (PerkinElmer).

In-situ Fourier transform infrared (FT-IR) spectra of toluene adsorption on the catalyst H-Fe₂₅ were recorded to analyze the interaction of toluene with the catalyst. The sample was pressed to a self-supported disk in the diameter of 0.5 inch and then was mounted in a quartz cell for evacuation, heat treatment and adsorption of toluene. The transmission FT-IR spectra were recorded through the CaF₂ windows of the quartz cell using a NEXUS870 FT-IR spectrometer.

The liquid oxidation of toluene was performed in a 40-mL of Teflon autoclave with 0.6 mL of toluene and 0.06 g of catalyst in 25 mL of deionized water, which formed biphasic system for catalyst dispersion and transport of reactants and products. The pH of the system was adjusted by H₂SO₄ solution. After purging with O₂ three times, 2.0 MPa of O₂ was fed into the autoclave. Then the autoclave was heated to reaction temperatures under stirring. Qualitative determination and quantitative analysis were conducted by GC-9560 (Shanghai Huaai) gas chromatography equipped with a SE-54 column and FID detector. After reaction, ethylbenzene was added as internal standard to calibrate the conversion of reactant and selectivity of products.

Fig. 3 depicts the TEM images for observation on the morphology and microstructure of the catalysts. The alumina support and the H-Fe₂₅ show almost the same morphologies of nanorods, implying that Fe₂O₃ mainly exists in a highly dispersed state on the γ-Al₂O₃ surface. According to the specific surface area of the alumina used, 132.7 m²/g, the thickness of iron oxide in amount of 25 wt% can be calculated as ~2–3 atomic layers, which is too thin to be clearly observed by the common TEM. The morphological features of H-FeM (not shown) are also very similar to that of H-Fe₂₅ and the results are consistent with the XRD measurements, which, as shown in Fig. 4, indicate the alumina is γ-Al₂O₃ (JCPDS no.29–0063). After loading with iron oxide, apart from the characteristic peaks of γ-Al₂O₃, some other weak peaks ascribed to Fe₂O₃ are detected. But the extra characteristic peaks ascribed to the other metal oxides are not observed for H-FeM, suggesting that these metal oxides are well dispersed on the corresponding catalysts. For the samples H-Fe₂₀Ni₅, H-Fe₁₅Ni₁₀ and H-Fe₁₀Ni₁₅, the diffraction peaks other than those ascribed to γ-Al₂O₃ cannot be observed, indicative of the highly dispersion of the mixture of oxides of iron and nickel in the catalysts (Fig. 4(b)). For the catalyst H-Fe₁₀Cu₁₅, however, the diffraction peaks ascribed to CuO are evident, implying the not so good compatibility or mixing of copper oxide with the iron oxide on the γ-Al₂O₃ support (Fig. 4(c)). For the catalysts H-FeCr, H-FeMo, and H-FeV, further characterizations on their structures have not been performed, because of their complete inertness for the reaction.

The N₂ sorption isotherms of typical samples (Fig. 5) exhibit all the IV-type isotherms with evident H2-type hysteresis loops, which is characteristic of irregular mesostructure and interstice mesoporous structure formed by nanoparticle assembly according to IUPAC [27, 28]. After loading metal oxides on the γ-Al₂O₃ surface, the porous textures of the samples seem unchanged much compared with the support alumina. The textural parameters of the samples are summarized in Table 2.

The results of FT-IR and TG measurements, as shown in Fig. 6, confirm the existence of HDPA on the surface of FeM nanorods. The two peaks located at 3000–2800 cm^–1 are attributed to the stretching vibration of –CH₂– and they show up with the samples decorated by HDPA adsorption [29]. By the results of TG measurements (Fig. 6(b)), the weight losses from the HDPA adsorbed samples are close to 5 wt%, similar to the amount of hexadecyl in the HDPA added in preparation of the catalyst. According to these results, the density of HDPA on the surface of the catalyst nanoparticles is ~1 HDPA/nm², which should be optimal to the catalytic performance of the catalyst for toluene oxidation and by tuning the active sites on the surface of FeM nanorods and the access of the reactants to the reactive sites, and the full coverage of HDPA on the surface of these samples is about 15 wt% according to our previous works [2].

As shown in Fig. 7(a), the conversion of toluene is ~71% and benzaldehyde is the only product with H-Fe₂₅ used as catalyst. A control experiment with Fe₂₅ as catalyst reports a conversion of toluene below 8% with a lower selectivity of benzaldehyde (~40.1%), other byproducts are benzyl alcohol (20.3%) and benzoic acid (39.6%). This means that the modification of HDPA greatly promotes the toluene conversion and ensures an exclusive selectivity in this catalytic system. With the modification of other metal oxides, the benzaldehyde is still the only product and the doping of NiO or CuO to H-Fe₂₅ gives promoted catalytic performance. Surprisingly, the catalysts are entirely inactive for the reaction when high valence metal oxides such as Cr, Mo, V, or Ti are used as dopants, even the dope amount is only 1~5 wt%, implying the limitation of the dopants on the mobility of the lattice oxygen of the catalysts.

In order to further investigate the dope effect of NiO or CuO on the property of the catalyst for the reaction, the H-FeM nanorods with different molar ratio of Ni (or Cu) to Fe were prepared and tested for the reaction, as shown in Fig. 7(b) and 7(c). It can be seen from the figures that the catalyst with iron oxide as major active constituent and minor dopant of nickel oxide gives the best catalytic performance for the reaction and, interestingly, the catalyst with nickel oxide or copper oxide as major active constituent shows fairly good catalytic performance for the reaction. But the stability of iron oxide catalyst is better than that of the nickel oxide and copper oxide catalysts, as discussed below.

After the use in reaction, the crystal structures (XRD) and morphology (TEM) of the catalysts, as shown in Figs. 7(d) and 7(e) are similar to those before reaction (Fig. 4(a) and Fig. 3(b)), indicating that the inherent structure of this kind of composite catalysts is rather stable. In addition, the concentration of leaching ions from the catalysts is always very low in the aqueous solution after reaction, though the leaching amount to the solution of Mn, Co, Ni, or Cu ions seems a little higher than that of iron (Table 3). As to the functions for the toluene conversion of the metal ions in the solution, control experiment with only Fe³⁺ and Ni²⁺ ions in the solution as catalysts, in similar concentrations to those with H-Fe₂₀Ni₅ as catalyst, was carried out and the conversion of toluene is only ~5% with a benzaldehyde selectivity of ~32% (Fig. 7(b)), other byproducts are benzyl alcohol (45.2%) and benzoic acid (22.8%). These facts indicate that the catalytic activity is mainly produced by the metal oxides composite nanoparticles and Fe₂O₃ is very important in this catalytic system. Therefore, it can be concluded that the loading of Fe₂O₃ has an essential influence on toluene conversion and the HDPA appears to ensure the exclusive selectivity of benzaldehyde as well as to promote the conversion of toluene. The moderately doped NiO is beneficial to increase the catalytic performance of the catalyst.

With the H-Fe₂₀Ni₅ is used as catalyst in the reaction, it is found that, similar to HDPA-FeO_x catalyst, the pH value has a prominent effect on toluene conversion, as shown in Fig. 8(a). The optimal pH value is ~2.5, neither lower nor higher pH value is detrimental to the catalytic activity. According to a previous results [30], the oxidation of toluene proceeded by hydride (H^–) transfer from toluene to the catalyst in water, nevertheless, hydride (H^–) was replaced by hydrogen atom (H·) in toluene solution. Moreover, the catalyst whose surface is positively charged will benefit to the transfer of hydride (H^–) to the catalyst and boost the oxidation of toluene. Therefore, relatively lower pH value in the biphasic catalytic system will lead to the surface of H-Fe₂₀Ni₅ positively charged. More protons in the solution, however, will lead to an unstable surface of the catalyst. Fig. 8(b) shows the influence of temperature on the reaction and the optimal value of temperature is 180 ℃ with exclusive selectivity. Higher temperatures used would result in the homogeneous oxidation initiated by O₂ diradicals and generate byproducts. The turnover frequencies (TOF, in unit of nm^–2 s^–1) based on the surface area of the catalysts are listed in Table 4.

In order to explore the interaction of toluene with the catalysts, in-situ FT-IR spectra of the catalyst H-Fe₂₅ were recorded with the surface adsorbed toluene. The sample was pressed into a self-supported disk of 0.5 inch in diameter and then was mounted in a quartz cell for evacuation, heat treatment and adsorption of toluene. The transmission FT-IR spectra were recorded through the CaF₂ windows of the quartz cell using a NEXUS870 FT-IR spectrometer. The results are shown in Fig. 9. The bands at ~2912 and ~2847 cm^–1 are attributed to the vibrations of C–H bonds in the methyl or methylene groups of the toluene (curve a) or the HDPA in H-Fe₂₅ (curves c, d and e) [29, 31]. The absorptions at ~3079 and ~3027 cm^–1 should be referred to the vibration of C–H bond connected to the aromatic ring of toluene. It appears that the toluene has hardly adsorbed at room temperature but strongly adsorbed on the catalyst at 180 ℃ (curve e), by the intensities of the peaks at ~3079 and ~3027 cm^–1. More importantly, the absorption at ~1219 cm^–1 shows up by the chemical adsorption of toluene at 180 ℃ which can be assigned to the formation of R–C–O–Fe by the insertion of the lattice oxygen of Fe₂O₃ into the C–H bond of benzyl [32, 33]. Coincidently, the shoulder peaks near 3079 and 3027 cm^–1 (i.e., 3091 and 3044 cm^–1) are also observed, which are reasonably attributed to the vibrations of aromatic C–H bonds in the adsorbed C₆H₅–C–O–Fe, which are also influenced by the narrow HDPA lanes. Summing up the results, we speculate that the lattice oxygen species are involved in the conversion of toluene to benzaldehyde with the assistance of HDPA at the catalyst surface.

With the results presented above, the mechanism of the catalytic system seems clear (Fig. 1). The HDPA-functionalized oxides nanorods locate themselves at the interface of toluene and water in the biphasic system as observed by laser confocal fluorescence microscope (Fig. 1(d)), and their hydrophilic/hydrophobic characteristics play an important role to stabilize the O/W emulsion (Fig. 1(a)). The surface HDPA molecules in a suitable density protect the nanorods from gathering and endow the catalyst amphipathic property. The HDPA in a surface density of ~1 molecule per nm² boosts the access of toluene toward the catalyst surface and, more important, it seems that HDPA molecules can align the toluene molecules touching the catalyst surface. The phosphate head in HDPA adsorbed on the surface may tune the catalyst surface to switch off unneeded sites unselective for the toluene oxidation. The hole dispersion system forms a Pickering emulsion under the stirring conditions, preventing the aggregation of droplets [34].

In addition, the water is helpful for the transfers of O₂ from gaseous as well as the products from the surface of the catalyst to the liquid phase (Figs. 1(b) and 1(c)), considering their relatively better solubility in water. In the process, the HDPA plays a role indispensable. As the consequence of synergistic interactions in the catalytic system, high-quality benzaldehyde is produced in exclusive selectivity at high conversions of toluene under mild conditions with gaseous O₂ as oxidant.

In summary, we have demonstrated a reliable strategy to prepare hexadecylphosphate acid (HDPA)-functionalized (Fe₂O₃-MO_x)/Al₂O₃ catalysts for the oxidation of toluene to benzaldehyde in toluene/water biphasic system with exclusive selectivity of benzaldehyde under mild conditions. The use of Al₂O₃ nanorods as support is greatly beneficial to the preparation and modification of the catalyst, with which the catalyst is more stable and easily tuned by deposition of HDPA and the dope of other metals. In the catalytic system, the HDPA-functionalized (Fe₂O₃-MO_x)/Al₂O₃ nanoparticles locate at the interface between oil and water phases to stabilize the Pickering emulsion. The loading of Fe₂O₃ and the HDPA are essential to ensure the exclusive selectivity of benzaldehyde and the catalytic activity for toluene conversion. Among the metal oxides of Mn, Co, Ni, Cu, Cr, Mo, V and Ti used as modifiers, the moderately doped nickel oxide enhances the catalytic performance of the system. Under optimal conditions, ~83% of toluene conversion and ~100% selectivity to benzaldehyde are obtained with molecular oxygen used as oxidant. This work displays an effective process, green and low cost, to produce high-quality (halogen-free) benzaldehyde from toluene oxidation using O₂ as oxidant and it should also be useful for other similar reactions to produce important chemicals in high quality.

Acknowledgments

The authors thank the financial supports from the National Natural Science Foundation NSF of China (91434101, 91745108) and the MOST of China (2017YFB0702900). The financial support from the Shanghai Research Institute of Petrochemical Technology of Sinopec is also appreciated.