Owing to the presence of the reactive C=N moiety, imines are an important nitrogen source in the fields of biology, medicine and agriculture [1-5]. A series of valuable nitrogen-containing derivatives can be synthesized through some simple chemical processes [6-8]. In industrial processes, most imine-containing compounds are obtained by condensing carbonyl compounds and primary amines under strongly acidic conditions [9-11]. Compared with this route, direct oxidative coupling of alcohols and amines to imines in the presence of air or oxygen is a more benign synthetic route, which is currently being intensively researched [12, 13].

Supported noble metal catalysts, such as Pd, Pt and Au, exhibit certain catalytic properties in the presence of alkali as additives or by using basic oxides as supports [14-20]. Considering the cost and environmental friendliness of catalysts, significant research has been devoted to developing transition metal oxide catalysts. Some reducible metal oxides, such as MnO₂, Fe₂O₃, V₂O₅ and CeO₂, have been reported to catalyze this reaction [21-25]. Our recent work showed that the reducibility of the transition metal oxides, which could directly influence the activation of molecular oxygen, was an important factor affecting the activity of the catalyst [26, 27]. The activity can be improved by optimizing the particle size or morphology of the metal oxides. However, the influence of other factors, such as the surface acidity and basicity of catalysts have been ignored in this reaction previously [28]. Some valuable solid acid-base catalysts composed of environmentally-friendly non-reducible metals have also been ignored in the previous work.

From the viewpoint of the reaction mechanism, a solid acid-base catalyst should also be suitable for this reaction. Generally, the title reaction undergoes two consecutive steps: (1) oxidative dehydrogenation of the alcohol to benzaldehyde and (2) condensation of benzaldehyde with aniline [23, 29, 30]. Current research based on transition metal oxides have shown that step (1) mainly occurs on the redox center, while step (2) mainly occurs on the acidic center [6, 31]. It is known that a solid base can also be used as an effective dehydrogenation active center in many catalytic processes [32]. Therefore, it is possible to construct a solid acid-base catalyst suitable for the above reaction by regulating the surface acid-base center of the catalyst.

Based on this hypothesis, Mg-Al mixed oxides were introduced as catalysts in the oxidative coupling of alcohols and amines to imine. The surface acid-base properties of Mg-Al oxides were modulated by changing the Mg/Al ratios, calcination temperature and treatment with probe molecules. Under the optimized conditions, Mg-Al mixed oxides with Mg/Al = 3 exhibited a relatively high activity in this reaction, which was comparable with that of supported FeO_x catalysts reported previously [26]. Various characterizations, including X-ray diffraction (XRD), ²⁷Al magic angle spinning nuclear magnetic resonance spectroscopy (MAS NMR), N₂-adsorption, NH₃-temperature-programmed desorption (TPD), CO₂-TPD and X-ray photoelectron spectroscopy (XPS) were performed to investigate the structure and surface properties of Mg-Al oxides. A series of detailed treatment conditions and catalytic tests were also carried out to clarify the influence of the surface acid-base properties on the catalytic performance for the oxidative coupling of alcohols and amines.

An Mg-Al mixed oxides catalyst was prepared by a co-precipitation method. In a typical process, a certain amount of Mg(NO₃)₂·6H₂O and Al(NO₃)₃·9H₂O was dissolved in 200 mL deionized water. After 0.5 h stirring, 12.5% NH₃·H₂O added to the mixture to adjust the pH value to 9.0. After continuous stirring for 2 h, the mixture was filtrated and washed with deionized water. The obtained solids were dried in an oven at 80 ℃ overnight, and then calcined at 500 ℃ for 4 h in a muffle furnace. The Mg/Al ratio of Mg-Al mixed oxides was adjusted from 0.33 to 5.0 by changing the amount of Mg(NO₃)₂·6H₂O and Al(NO₃)₃·9H₂O in the initial process. For comparison, MgO and Al₂O₃ were also prepared using the same method as the Mg-Al mixed oxides.

The X-ray diffraction (XRD) patterns were recorded on a Rigaku X-ray diffractometer using Cu K_α radiation (λ = 0.15418 nm) at 40 kV and 40 mA. N₂ adsorption-desorption isotherms were measured at –196 ℃, using a Micromeritics ASAP 2010N analyzer. Samples were degassed at 200 ℃ for 20 h before measurements. Specific surface areas were calculated using the Brunauer-Emmett-Teller (BET) model. Pore size distributions were evaluated from desorption branches of nitrogen isotherms using the Barret-Joyner-Halenda (BJH) model. Scanning electron microscopy (SEM) was performed using an HITACHI SU8020 instrument with 30 kV. XPS data were obtained on a Thermo ESCA LAB 250 system with a Mg K_α source (1254.6 eV). TPD was performed on a ChemBET Pulsar TPR/TPD instrument (Quantachrome Instruments). Before detection, the catalysts (50 mg) were first treated in a N₂ (99.99%) flow at 500 ℃ for 0.5 h. The ²⁷Al MAS spectra were recorded on a Bruker AVANCE Ⅲ 400 WB spectrometer equipped with a 4 mm standard bore CP MAS probe-head whose X channel was tuned to 104.27 MHz for ²⁷Al, where pulse sequence was one pulse (0.9 μs, π/12), using a magnetic field of 9.39 T at 24 ℃. The dried and finely powdered samples were packed in the ZrO₂ rotor closed with a Kel-F cap and were spun at 12 kHz rate. A total of 100 scans were recorded with a 2 s recycle delay for each sample. All ²⁷Al MAS chemical shifts were referenced to the resonances of [Al(H₂O)₆]³⁺ with standard (d = 0.00).

Catalytic screening for the oxidative coupling of alcohols and amines was carried out as follows: a 50-mL two-neck flask was charged with toluene (10 mL), catalyst (0.3 g), 1.0 mmol benzyl alcohol and 2.0 mmol aniline. The flask was placed in an 80 ℃ oil bath and connected with a balloon full of air. The products were extracted with a sampling pipe equipped with a filter, and the analyses were performed on a gas-chromatograph equipped with an HP-5 column and FID detector. The corresponding alcohols and amines and imines were used as standards.

The oxygen-free experiment was carried out in a Schlenk tube. First, the catalyst was put in the Schlenk tube and degassed for 1.5 h under vacuum. The tube was then charged with nitrogen (99.99%). The above procedure was repeated three times. Then the degassed reagents were introduced into the Schlenk tube and the tube was placed in an 80 ℃ oil bath. For the recycling experiment, the solid catalyst was separated by filtration after reaction without any washing treatment. Before another reaction, the catalyst was calcined at 500 ℃ for 1 h to remove the adsorbed reactants.

To understand the role of the acid-base sites in the reaction, pyridine and pyrrole poisoning experiments were carried out over the Mg-Al mixed oxide with Mg/Al = 3. The detailed process was as following: the catalyst (0.3 g) was added to a mixture of toluene (10 mL) and pyridine/pyrrole (100 μL). The mixture was stirred for 3 h at room temperature. The catalyst was then filtered out and washed with toluene several times.

Fig. 1 shows the XRD patterns of pure Al₂O₃, MgO and Mg-Al oxides with different Mg/Al ratios. All these samples were prepared with the same precipitation method, and ammonia solution was used as the pH adjusting agent. The samples were all calcined at 500 ℃ for 4 h before use as catalysts and characterization. As for Al₂O₃, the weak diffraction peaks centered at 2θ = 37.5°, 39.5°, 45.9° and 66.7° could be assigned to the (311), (222), (400), (440) planes of the γ-phase Al₂O₃, which was consistent with the JCPDS Card 10-0425 [33]. As for MgO, three diffraction peaks centered at 2θ = 37.0°, 43.0° and 62.5° could be observed, which were assigned to the (111), (200) and (220) planes of MgO (JCPDS Card 77-2364) [34]. As for Mg-Al oxides with different Mg/Al ratios, a mixed phase existed in all the samples. According to the JCPDS Card 33-0853, the main diffraction peaks of MgAl₂O₄ are located at 2θ = 36.8°, 44.8° and 65.6°, which are associated with the (311), (400) and (440) crystal planes of MgAl₂O₄. When Mg/Al≤1, both γ-Al₂O₃ and MgAl₂O₄ phases existed in the sample. The low intensity and broad peaks showed that the crystallinity of both γ-Al₂O₃ and MgAl₂O₄ phases was very low. When 2≤Mg/Al≤4, the diffraction peak at approximately 2θ = 36.8° and 44.8° broadened, and no obvious diffraction peaks assigned to the MgO phase could be observed. This result indicated that the Al³⁺ cations remained closely associated with the Mg-O and the local structure was amorphous. As for the sample with Mg/Al = 5, additional peaks assigned to the MgO crystal phase could be observed.

NMR is a powerful local-probe measurement that can detect all the structures regardless of crystalline or amorphous phase [35]. It can effectively complement Powder XRD by probing amorphous phases where direct information could only be obtained from samples with crystalline structure. Fig. 2 shows the ²⁷Al MAS NMR spectra of Mg-Al oxides with four typical Mg/Al ratios. There were two resonance peak in the areas around 8–10 ppm and 67–81 ppm, which could be assigned to Al³⁺ cations octahedrally (Al_o) and tetrahedrally (Al_t) coordinated to oxygen, respectively [36-38]. With the increase of the Mg/Al ratio, the peaks of Al_o and Al_t were slightly shifted downfield. This result suggested that electronegative Al³⁺ was being replaced by more electropositive Mg²⁺ as the nearest neighbor cation bonded to the Al-O polyhedral [39]. Different results were obtained for the Mg/Al = 5 sample, in which the peak assigned to Al_t split into two peaks and the peak assigned to Al_o moved to a lower chemical shift. This result mainly arose from the formation of the MgO and MgAl₂O₄ crystal phases at this Mg/Al ratio [40].

Combined with the results of XRD and ²⁷Al MAS NMR, it was shown that as the Mg content increased, more and more Mg²⁺ cations bonded to the Al-O polyhedral forming Mg-O-Al. When Mg/Al≤4, although some MgO and MgAl₂O₄ units were present in the sample, the samples still generally possessed the property of the amorphous state. The relatively wide XRD diffraction peaks supported this deduction. When Mg/Al was increased to 5, the formed MgO crystal phase broke the original bonding balance. Both MgO and MgAl₂O₄ crystalline phases could be clearly observed in the XRD patterns. Furthermore, the ²⁷Al MAS NMR spectra also showed that Al³⁺ cations in all the sample preferentially located in the octahedral positions. Comparing these samples, the Al_o/Al_t values decreased from 2.89 for Mg/Al = 0.33 to 2.04 for Mg/Al = 1.0, and then remained approximately constant when the Mg/Al ratio was increased further.

Fig. 3(a) shows the N₂ adsorption-desorption isotherms of pure Al₂O₃, MgO and Mg-Al oxides with different Mg/Al ratios. All these samples exhibited a type Ⅳ isotherm, which indicated a mesoporous character [41, 42]. Most of the hysteresis loops located at a relatively high pressure (p/p₀) zone, which indicated the presence of a relatively large pore size. Pore size distribution plots showed that the most integrable pore size of Mg-Al oxides with different Mg/Al ratios was centered at around 10–15 nm (Fig. 3(b)). The specific surface area showed a decreasing trend with the increase of the Mg/Al ratio (see Table 1). As for pure Al₂O₃, a relatively small pore size (around 5 nm) could be observed. As for pure MgO, the wide pore size distribution meant that the pores in the sample were mostly derived from the accumulated pores between particles. SEM images further confirmed the results from N₂-adsorption that a large amount of mesopores arose from the packing holes between the particles (Fig. 4). When the Mg/Al ratio was increased to 5, relatively large particles and pores could be clearly observed. This mainly arose from the remarkable crystallization occurring at such a high Mg/Al ratio.

The surface acidic and basic properties of the samples were investigated by TPD measurements using NH₃ and CO₂ as probe molecules, respectively. Fig. 5(a) shows the NH₃-TPD profiles of pure Al₂O₃, MgO and Mg-Al oxides with different Mg/Al ratios. All these samples exhibited a broad desorption peak in the detection region, which indicated the relatively complex surface acidic nature of these samples. The TPD profiles were deconvoluted into four desorption peaks (Fig. 5(a) and Table 2). According to the literature, the low-temperature peak could be assigned to a reversible H-bonded adsorption on Br nsted sites, and the high-temperature peak could be attributed to the irreversible coordinated adsorption on the Lewis sites [43, 44]. These Lewis sites included Al³⁺ cations in Al³⁺-O^2–-Mg²⁺ and Al³⁺-O^2– species, and Mg²⁺ cations in Mg²⁺-O^2- species [45-48]. From the perspective of total acidity, the largest surface acidity appeared in the Mg/Al = 0.33 samples. With the increase of the Mg/Al ratio, the total surface acidity decreased first and then increased. The sample of MgO exhibited the lowest surface acidity.

The CO₂-TPD profiles (Fig. 5(b)) showed that the surface basic properties of these samples were also complex. A series of basic centers with different strengths were present on the surface of all these samples. They separately came from OH⁻ groups, Mg-O pairs, Al-O pairs and O²⁻ anions [43-45]. As for the Mg-Al oxides, with an increase of the Mg content, the total number of surface basic sites increased at first then decreased. The largest number of surface basic sites was present in the Mg/Al = 3 samples (89.8 μmol·g^–1). A large number of weak basic centers existed on the surface of the sample (25.9 μmol·g^–1), which accounted for 28.8% of the surface basic centers (Table 2).

The above results showed that the change of surface acidity and basicity was closely related to the Mg/Al ratio of the samples. However, it should be noted that the composition of surface Mg/Al was not consistent with that of the body. Table 1 shows that the surface Mg/Al was much lower than that of the stoichiometric ratio, which indicated that more Mg atoms entered into the bulk of the samples. This part of the Mg atoms indirectly affected the acid-base properties of the catalyst surface through bonding with Al species (detected from NMR). With the increase of the Mg content, the O 1s XPS spectra (Fig. 6) shifted slightly toward a lower binding energy, which was consistent with the change of the surface basic properties. It should be noted that the total base amount of the Mg-Al oxides in our case was lower than that in some reports [41-43, 49, 50]. This was closely related to the distribution of surface Mg/Al elements. XPS results clearly showed that the sample prepared under our conditions possessed relatively low surface Mg/Al ratios compared with the samples reported previously.

Fig. 7(a) shows the catalytic activity of the above samples in the oxidative coupling of benzyl alcohol and aniline to an imine at 80 ℃. The mole ratio of benzyl alcohol to aniline in the reagent was 1:2. The yield of imine was calculated using benzyl alcohol as the standard. The selectivity calculated based on benzyl alcohol was larger than 99%. Therefore, the conversion of benzyl alcohol was almost the same as the yield (not shown here). Fig. 7(a) shows that both MgO and Al₂O₃ exhibited a low activity in this reaction. Only a 19.2% and 22.7% yield of imine could be observed over MgO and Al₂O₃ after reaction for 8 h. As for the Mg-Al oxides, the introduction of a small amount of Mg significantly improved the activity of the sample. When Mg/Al = 3, the yield of imine could reach to 93.5% (with benzyl alcohol and aniline as substrates) after reaction for 8 h. The activity could be further improved when the reaction temperature was increased from 80 ℃ to 100 ℃. This result was comparable with that of the supported FeO_x catalysts reported previously [26]. When Mg/Al was further increased, the activity of the samples decreased compared with that of the sample with Mg/Al = 3, but it was still higher than that of the MgO and Al₂O₃ samples. It should be pointed out that a trace amount of by-product, azoxybenzene, could be detected by GC. It was identified with Gas chromatography-mass spectrometry and a standard reagent. This by-product arose from the condensation of a small amount of aniline.

As for the sample with Mg/Al = 3, a hot filtration test was carried out to confirm the heterogeneous nature of the sample. The results showed that Mg-Al oxides behaved as an operationally heterogeneous catalyst, since the imine concentration in the filtrate did not increase (Fig. 7(b)). Almost no Mg and Al species in the filtration could be detected by ICP measurement. Furthermore, multiple reaction cycles were performed to investigate the recoverability of this sample. Two separate reactions were carried out at 80 and 100 ℃, respectively (Fig. 7(c)). After four cycles, the catalyst in the reaction at 100 ℃ still retained a high activity, whereas the activity of the catalyst in the reaction at 80 ℃ decreased significantly. It could be concluded that the adsorption and desorption of reactants and products on the catalyst surface have an important effect on the cyclic use of the catalyst. The adsorbed species could destroy the structure of Mg-Al oxides during the following thermal treatment. The surface area detected by N₂-adsorption decreased from 168 m²·g^–1 to 39 m²·g^–1. A relatively high reaction temperature was favorable for the desorption of products, which is beneficial for the recycling of catalysts.

It is known that Mg-Al oxides, as well as MgO and Al₂O₃, are all irreducible metal oxides. The catalytic process should be quite different from the reducible metal oxide catalysts reported previously [21-26]. Combined with the above characterization results, it could be proposed that the surface acid-base sites of Mg-Al oxides catalyst should be the main active sites for the oxidative coupling of benzyl alcohol and aniline. The influence of Mg/Al ratio on the catalytic performance was also mainly manifested by the change of the surface acidity and basicity of the catalyst. To understand the role of the acid-base properties of the catalyst in the title reaction, we attempted to use pyridine (basic molecule) and pyrrole (weak acidic molecule) to poison part of the surface acidic and basic active sites of the Mg-Al oxide catalyst (Mg/Al = 3). The detailed pre-treatment process has been described in the Experimental section. Fig. 8 shows the catalytic performance of the poisoned catalysts. Compared with the untreated sample, the catalytic activity of both poisoned samples decreased. The sample treated with pyrrole exhibited a much lower activity than that treated with pyridine. This result indicated that the surface basic sites should play a more important role in the reaction. NH₃-TPD and CO₂-TPD results showed that the sample with Mg/Al = 3 possessed a small number of surface acidic sites (34.8 μmol·g^–1, Table 2) and a large number of surface basic sites (89.8 μmol·g^–1, Table 2). Combined with the catalytic result, we could further confirm the important role of the surface basic sites in the oxidative coupling of benzyl alcohol and aniline.

To further understand the role of the surface acid-base properties in the catalytic process, we attempted to tune the surface properties of the sample (Mg/Al = 3) by changing the calcination temperature and then investigating their catalytic performance. XRD patterns (Fig. 9(a)) showed that the sample calcined at 400 ℃ (denoted as MgAl-400) exhibited a similar crystal structure to that calcined at 500 ℃ (the catalyst studied above, denoted as MgAl-500). The sample calcined at 800 ℃ (denoted as MgAl-800) contained more MgAl₂O₄ crystalline phases compared with the other two samples. N₂-adsorption results showed that MgAl-400 exhibited a similar specific surface area to MgAl-800, which was 97.1 and 102.1 m²·g^–1, respectively. These were a little lower than that of MgAl-500 (167.9 m²·g^–1, Table 3). The catalytic result showed that MgAl-500 exhibited a higher activity than MgAl-800. Whereas the MgAl-400 sample showed a very low activity under the same reaction conditions (Fig. 9(b)). NH₃-TPD results (Fig. 10(a)) showed that all these samples possessed a certain number of acidic sites on the catalyst surface. Among them, MgAl-400 possessed the largest amount of weak acidic sites. The CO₂-TPD (Fig. 10(b)) results showed that MgAl-400 was almost absent of weak basic sites, which was quite different to that for MgAl-500, MgAl-800 and the other samples studied above. The low activity of MgAl-400 in the title reaction should mainly arise from the absence of weak basic sites on the catalyst surface. Combined with above results, it can be concluded that weak basic sites play a very important role in the oxidative coupling of benzyl alcohol and aniline. These weak basic sites should mainly provide adsorption sites for benzyl alcohol, which is a weak acidic reagent. Aniline (pK_a = 4.63) was mainly adsorbed on the surface acidic sites of catalyst.

Our previous work has shown that oxidative dehydrogenation of benzyl alcohol is the rate-determining step in the reaction [26]. Over reducible oxide catalysts, the reaction proceeds in two consecutive steps, including oxidative dehydrogenation of alcohol to benzaldehyde and imine formation by reaction of benzaldehyde with aniline [23]. In this case, the step of oxidative dehydrogenation of alcohol to benzaldehyde hardly occurs on the irreducible Mg-Al oxides in the absence of aniline (Fig. 11(a)). This means that oxidative coupling of benzyl alcohol and aniline over Mg-Al oxides should experience a different reaction path. Molecular O₂ was demonstrated to be involved in this reaction since little imine can be observed in the absence of molecular O₂ (Fig. 11(b)). However, the activation of molecular oxygen and benzyl alcohol is a relatively complex process, which needs participation by aniline. This point is quite different from that over the reducible oxides catalysts, such as supported FeO_x and CeO₂, reported previously [23, 26]. The coupling of benzaldehyde and aniline is a fast step over Mg-Al oxide catalysts (Fig. 11(c)). Therefore, the activation of benzyl alcohol is still the rate-determining step for this oxidative coupling reaction although this process is not directly achieved by forming benzaldehyde. Further work is still needed to clarify the details of this process.

Furthermore, the general applicability of the Mg-Al oxides catalyst was also investigated. At least six kinds of imines could be obtained by oxidative coupling of amine with different alcohols. Table 5 shows that the Mg-Al mixed oxide catalyst was active in the reactions using benzyl alcohol derivatives with either electron-rich-or electron-poor-substituent groups, especially for 4-methoxy benzyl alcohol, 4-nitro benzyl alcohol and p-methyl benzyl alcohol (the yield of imine was 92.2 %, 85.6% and 80.1% after 12 h reaction, respectively).

Mg-Al mixed oxides with various Mg/Al ratios have been prepared by a co-precipitation method. They can effectively catalyze the oxidative coupling of alcohols and amines to imine. The sample with Mg/Al = 3, which possesses a large amount of weak basic sites and a relatively small amount of weak acidic sites, exhibits the highest activity in the reaction. These acidic and basic sites serve as adsorption and activation sites for amines and alcohols, respectively. Among them, the weak basic sites play a more important role than the acidic sites in the catalytic process. They are the reactive centers for benzyl alcohol activation, and the activation of benzyl alcohol is the rate-determining step for this oxidative coupling reaction. This work is very helpful to understand the role of the surface acid-base properties in the oxidative coupling reaction at low temperature, and will provide a new method for the development of irreducible acid-base catalysts for the imine synthesis.