An Efficient and Recyclable Sn-Based Phosphotungstic Acid with Tunable Brønsted/ Lewis Acidity for Selective Oxidation of Benzyl Alcohol

A series of metal ion (M = Sn, Fe, Ni, Co, Ag, Cu) exchanged tungstophosphoric acid (H3PW12O40; TPA) catalysts with tunable Brønsted/Lewis acidity were synthesized and exploited for the oxidation of benzyl alcohol (BzOH) to benzaldehyde (BzH) using hydrogen peroxide (H2O2) as oxidant. The structure of these M-TPA composite salts was also characterized by Fourier transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), thermogravimetric analyses (TGA) and solid-state P nuclear magnetic resonance (NMR) probe molecule method. Among these M-TPAs, the Sn1/2H2PW12O40 catalyst, which presented strong Brønsted acidity, the synergistic effect of Brønsted/Lewis and pseudo-liquid characteristic property, exhibited excellent catalytic activity and durability with 98.2% of BzH selectivity and 95.1% of BzH yield. The optimal conditions for the oxidation of BzOH optimized by response surface methodology (RSM) were as follows: n(BzOH)/n(H2O2) = 1:1.25, catalyst amount of 5.5 wt.% to BzOH, water amount of 17 mL, 3.3 h of reaction time, and temperature 393 K. Moreover, further kinetic study confirmed that the reaction order was 2.64 and the activation energy was 21.75 kJ mol.


Introduction
Carbonyl compounds are one of the most important intermediates in the fine chemicals such as perfumes, dyestuffs, and agro-processing chemicals industry. 1 The sustained growth of population, as well as the exponential growth of their quality of life, and the demand for more environmentally friendly synthesis methods in chemicals has increased. Traditional methods to obtain carbonyl compounds have involved oxidation of alcohols which is based on the use of oxidants under the stoichiometric amounts, like K 2 Cr 2 O 7 , KClO, KMnO 4 , MnO 2 , pyridinium chlorochromate (PCC) and so on. However, most of these reported systems suffer from economic and environmental problems for the phenomenon of large amounts of by-products, high reagent load and toxicity of the material. 2,3 Currently, catalyst-oriented liquid-phase selective oxidation of alcohols as a simple, economical route to the synthesis of carbonyl compounds has attracted the attention of researchers. Many transition metals which have been employed as catalysts in this reaction, such as Pd, Ru, Ag and Au, showed a nice selectivity and conversion. [4][5][6][7][8] Inevitably, there were still some defects in these metal catalysts due to toxicity of metal and easy desorption from the carrier.
Herein, a series of metal ion-exchanged phosphotungstic acid (M-TPA) have been synthesized, the adopted metals included Sn 2+ , Fe 3+ , Ni 2+ , Co 2+ , Ag + and Cu 2+ . The metal-modified TPA catalysts were characterized by Fourier transform infrared spectroscopy (FTIR), X-ray diffraction (XRD) and thermogravimetric analyses (TGA). Particularly, the acidic property was investigated by solidstate 31 P magic angle spinning nuclear magnetic resonance (MAS NMR) method. [49][50][51][52] On the basis of Box-Behnken design (BBD), the effects of different reaction parameters were optimized by response surface method (RSM) and the oxidation kinetic model was also established under optimal conditions.

Catalyst preparation
All materials were purchased from Aladdin Company (Shanghai, China) and used without purification. The metal ion-exchanged tungstophosphoric acid (TPA; H 3 PW 12 O 40 ) was prepared under the procedure described elsewhere. 40,41 Based on the typical synthesis procedure, a solution was obtained by dissolving 5.76 g H 3 PW 12 O 40 (0.002 mol) and appropriate amount of SnCl 2 into 40 mL deionized water, followed by continuous stirring for about 12 h at 368 K. After removal of water, the obtained product was washed with diether and followed by drying under vacuum at 348 K for 12 h. Thus, a series of Sn metal ion-exchanged TPA samples were synthesized and expressed as Sn x H 3−2x PW 12 O 40 (x = 1/2, 1, 3/2; x means the molar ratio of Sn/TPA). For comparison, the other metal ion-exchanged TPA samples were prepared with similar method by varying metal nitrate with M = Fe, Ag, Cu, Co and Ni.

Catalyst characterization
Spectral analysis using the conventional KBr pellet procedure was performed with aid of FTIR experiments on a Bruker IFS-28 spectrometer with resolution of 4 cm −1 over the range of 4000-400 cm −1 . Structure of sample was also studied by XRD (Bruker D8 ADVANCE X-ray diffractometer) using Cu Kα radiation (0.15418 nm) at 40 kV and 20 mA. The TGA were performed by heating the catalyst from 298 to 873 K at a rate of 10 K min −1 in flowing N 2 by TG209 (NETISCH). Solid-state 31 P MAS NMR sample characterization was carried out with Bruker-Biospin Avance-III 500 spectrometer at 202.46 MHz. The 31 P chemical shifts referred to the 85% H 3 PO 4 aqueous solution. The details of sample treatment were conducted according to the process of literature description. 53

Catalytic reaction
Under atmospheric conditions, the oxidation reaction was performed in a three-necked round bottom flask with a condenser. In a typical experiment, a mixture of substrate (0.05 mol benzyl alcohol and 0.1 mol 30% H 2 O 2 ) and catalyst (0.27 g, 5 wt.%) in deionized water (20 mL) was placed in reactor at room temperature. Then the mixture was placed into an oil bath by mechanical stirring at 393 K for 3.0 h. After reaction, the mixture was cooled to room temperature, extracted with acetic ether, and rinsed with brine solution and water. The catalyst was filtered from the mixture, and washed with diethyl ether for reuse. Productions for reaction were analyzed by gas chromatography (Agilent 7890B GC; HP-5; 30 m × 0.25 mm × 0.25 µm) with diphenyl as the internal standard. Catalytic properties of metal ion-exchanged TPAs catalysts were listed in Table 1 on the oxidation of BzOH with H 2 O 2 .

Response surface methodology
Based on RSM, a series of experimental designs for the experimental factors were applied to optimize the production process of BzH by oxidation of BzOH with H 2 O 2 over Sn 1/2 H 2 PW 12 O 40 . Four independent experimental variables, namely catalyst amount (x 1 ), BzOH/H 2 O 2 mole ratio (x 2 ), water amount (x 3 ), and reaction time (x 4 ) were utilized by BBD. All factors were established and shown in Table 2, and coded into three levels (−1, 0, +1). Additionally, a 3 4 full-factorial center composition designed with three coded levels was utilized, including 24 factorial and 5 central points, as listed in Table 3. The encoding values of these factors were derived from equation 1: where ∆X i represents the step-change value, while X 0 , x i and X i (i = 1-3) mean the central, coded, and real value of the various variables, respectively.
Utilizing the quadratic polynomial model equation given by RSM, the interactions between those variables were studied to make the course of reaction as effective as possible and forecast the product yield (i.e., BzH), which can be shown as: (2) where β 0 , β i , β ii , and β ij represent the regression coefficient of the variables, linear, quadratic, and interactive terms, respectively, while x i and x j (i, j = 1-4) denote the level of code for independent variables. The significance and effectiveness of the presented model were evaluated by statistic parameters on the basis of the analysis of variance (ANOVA) means.

Kinetic study
Oxidation kinetics under various conditions was studied by the initial reaction rate method, leading to the obtention of the kinetic expression. In this work, diverse experimental parameters were studied, such as BzOH/H 2 O 2 mole ratio, temperature, and so on, while other parameters remained unchangeable. The reaction rate (r) may be indicated as: where C A and C B indicate the instant concentration of BzOH and H 2 O 2 , respectively; k' is the rate constant, α and β the reaction order corresponding to BzOH and H 2 O 2 .
Taking the natural logarithm, the above equation can be expressed as: (4) where represents the modified parameter. Therefore, the rate constant was associated with the activation energy (E a ) by the Arrhenius equation: where R is the gas constant, T the reaction temperature, and k 0 the pre-exponential factor.

Results and Discussion
Catalysts characterization According to Figure S1 from the Supplementary Information section (hereinafter referred to as SI), FTIR spectra of original H 3 PW 12 O 40 and Sn x H 3−2x PW 12 O 40 (x = 1/2, 1, 3/2) samples were compared. Despite the reduction of   54 The phase purity of bulk samples was identified by powder X-ray analysis. The diffraction peaks of pure TPA and Sn x H 3−2x PW 12 O 40 series samples in Figure S2 (SI section) were consistent with available literature. 55 As exhibited in Figure S2b (SI section), the pure TPA displayed three main diffraction peaks of the Keggin polyanions structure at 2θ degree of 10.3, 25.3 and 34.6°. 53 Predictably, the characteristic peaks were also presented in Sn x H 3−2x PW 12 O 40 series samples, and only the intensity of peaks reduced with the bonding of Sn ( Figures S2c-S2e, SI section). On the other hand, the crystallinity of a series of catalysts was very similar to that of the parent TPA, which was consistent with previous reports. 55 However, the characteristic diffraction peak of SnCl 2 ( Figure S2a, SI section) could not be identified, indicating the existence of a new metal modified TPA salt with fine crystallinity. Moreover, additional peaks related to the incorporated Sn 2+ appeared in the 2θ region of ca. 50-60°, 31 which indicated the perfect anchoring of Sn 2+ onto the TPA. The conclusions were consistent with FTIR results.
The thermal properties of pure H 3 PW 12 O 40 and metal exchanged TPAs were also determined by sample was illustrated and discussed. As shown in Figure S3a (SI section), the pure SnCl 2 salt showed three major weight-loss peaks at 320, 420, and 720 K, which was ascribed to desorption of surface water, loss of crystal water, and partial decomposition of SnCl 2 , respectively. These results are distinct from the original TPA, which showed weight-loss peaks at 322, 470, and 663-788 K for the desorption of superficial water, loss of crystal water, and structural breakdown of Keggin frame, respectively. 56 By contrast, the Sn 1/2 H 2 PW 12 O 40 sample exhibited similar weight-loss peaks to that of the original TPA at 321, 445 and 673-780 K ( Figure S3c, SI section), in addition, there was a special existence of an extra weight-loss peak at 568 K, which may arise from the collapse of partial Keggin units formed in the existence of metal cations (Sn 2+ ). The thermal analysis results also indicated that metal cations (Sn 2+ ) were perfectly anchored with PW polyanions and kept steady at the temperature used for the catalytic studies.
In order to collect the acid properties information from samples, the 31 P-trimethylphosphine oxide (TMPO) MAS NMR could be employed to get the acid strength and the Brønsted or Lewis acid properties. 49,50 It was evident that the observed 31 P NMR chemical shift (d 31 P) of TMPO was facilitated by linear dependence with Brønsted acid strength. 52 Actually, the original TPA adsorbed by TMPO typically exhibited multiple 31 P signals in two chemical shift scopes (Figure 1a). The sharp 31 P signals situated at −10 and −15 ppm can clearly arise from polyanions (PW 12 O 40 3− ) of the TPA, while those in the range of 55-95 ppm owing to Brønsted acid sites of the M-TPAs adsorbed by TMPO. 40,41 It was remarkable that the original TPA had super acidic Brønsted acid sites (i.e., those with d 31 P ≥ 86 ppm), 49-52 as shown by the existence of 31 P resonance signals at 92, 88, and 83 ppm (Figure 1a) associated with three obtainable Brønsted H + sites adsorbed by TMPO (i.e., TMPOH + ). No resonance characteristics were observed for pure SnCl 2 adsorbed by TMPO (Figure 1e). Whereas the 31 P resonances was located within 55-80 ppm due to (TMPO) n H + (n ≥ 2) species. 51 Upon exchanging Sn 2+ with the acidic protons (H + ) of TPA, similar resonance signals to original TPA were captured, nevertheless, the peaks emerged to be broadened (Figures 1b-1d). This also indicated that Sn x H 3−2x PW 12

Oxidation of benzyl alcohol to benzaldehyde
The acidic property of catalyst is related to its catalytic activity in some reaction catalyzed by acid. Oxidation of alcohol to aldehyde (ketone) using H 2 O 2 as oxidant is a representative acid-catalyzed reaction. Strong acidic conditions can improve the oxidizability of H 2 O 2 . 58 The catalytic activities of all metal ion-exchanged TPA and the original TPA were investigated during the oxidation of BzOH and the results were listed at Table 1. As listed in Table 1, SnCl 2 exhibited ignorable BzH yield and BzOH conversion. All the metal ion-exchanged TPA salts displayed acceptable catalytic properties comparing with  samples showed inferior catalytic activities in contrast to the partially ion-exchanged TPA salts, which clearly indicated the synergistic influence of Brønsted-Lewis acid. More super-strong Brønsted acidity was not conducive to the oxidation of alcohols.
The synergy influence of Brønsted-Lewis acid can be attributed to the forceful ionic interactions among the Brønsted acidic proton (H + ), Lewis acidic metal ion (i.e., Sn 2+ ), and the PW 12 O 40 3− polyanion in ion-exchanged TPA salts. 29 A possible mechanism for BzOH oxidation to BzH over the M-TPAs was similar to the procedure described in literature. 59,60 The oxidation of BzOH to BzH was catalyzed by Lewis acid (path 1) and Brønsted acid (path 2) based on the structural characteristics and afore discussed acidity characterization of the Sn 1/2 H 2 PW 12 O 40 catalyst (Scheme 1). The catalytic performance of the catalyst was mainly related to the ratio of Brønsted acid to Lewis acid (B/L ratio) in the catalyst. In the cards, H 2 O 2 first underwent the elimination of the β-hydride and then combined with the carbonyl group to form the final products BzH and water. Based on previous results, the conclusion of synergy effect of Brønsted-Lewis acid was put forward. Apart from the synergy influence of Brønsted-Lewis acid, the pseudo-liquid character of TPA and strong acidity were also responsible for excellent catalytic properties on the oxidation of BzOH over Sn 1/2 H 2 PW 12 O 40 catalyst.

Process optimization
Since Sn 1/2 H 2 PW 12 O 40 catalyst exhibited optimal catalytic properties among different Brønsted-Lewis M-TPA salts, it was selected for further process optimization. Owing to the improved efficacy, the impact from parameters like catalyst amount, BzOH/H 2 O 2 mole ratio, time of the reaction and water amount was investigated. As is well-known, the catalyst amount always played an important role in this selectivity oxidation. 14,61 The effect of catalyst amount on the catalytic performance of BzOH oxidation with H 2 O 2 was listed in Figure 2a. The conversion of BzOH almost linearly improved from 72.14 to 98.32% with the catalyst amount increasing to 6 wt.%. It was related to the gradual increase of active acid sites for catalytic oxidation. At the catalyst amount of 5 wt.%, a maximum yield of BzH (93.24%) was obtained. However, the BzH yield reduced significantly with further increasing the amount of catalyst, which was due to the decomposition of H 2 O 2 resulted by the excessive amount of acid sites presented in the reaction system. 14,29 Excess catalyst amount easily led to unsatisfactory oxidation of BzH and other reactions.
Also, the role of BzOH/H 2 O 2 mole ratio was investigated while other experimental variables kept constant, and the results were displayed in Figure 2b. Clearly, the BzH yield improved with BzOH/H 2 O 2 mole ratio increasing. The highest yield of BzH obtained was 93.24% with BzOH/H 2 O 2 mole ratio of 1:2 in 4 h. This implied that appropriate amount of H 2 O 2 oxidant facilitated driving reaction equilibrium to form BzH. Nevertheless, further increase of BzOH/H 2 O 2 mole ratio led to a decrease of BzH yield probably due to the reactant (BzOH) and catalyst were too diluted, and the concentration of active center decreased with an excess of H 2 O 2 . Meanwhile, excessive amount of H 2 O 2 also provoked formation of undesirable oxidation of BzH, and thus, reducing BzH yield.
The influence of reaction time on BzH yield and BzOH conversion over the Sn 1/2 H 2 PW 12 O 40 catalyst was studied at the range of 1-5 h with the conditions of catalyst amount = 5 wt.%, BzOH/H 2 O 2 mole ration = 1:2, water amount = 20 mL, and reaction temperature = 393 K. As exhibited in Figure 2c, both BzH yield and BzOH conversion improved obviously at prolonged reaction time, reaching an excellent BzH yield (93.24%) at 4 h. Further increasing of reaction time caused a gradual decline in BzH yield, which may arise from the consumption of BzOH and the excess of BzH occupying the catalytic activity sites, hence, the oxidation of BzH of reaction improved with increasing reaction time and reduced BzH yield. Therefore, 4 h was chosen as reaction time for further study.  Figure 2d depicted the influence of water amount on the selective oxidation of BzOH to BzH. Obviously, both BzH yield and original BzOH conversion improved with the amount of water increasing, reaching a maximum at 20 mL water, then descending gradually afterward. During alcohol oxidation reaction, the presence of water can easily form special droplets and was conducive to the adsorption and/or activation of molecular oxygen. 8,62 Meanwhile, the existence of surface water can also enhance the oxygen mobility on catalyst to improve the number of oxygen vacant sites. 63,64 However, the reaction system was diluted by excess water, and the available acid moiety of catalyst decreased. Based on the above experimental results, the best catalytic oxidation activities can be obtained under the optimum experimental conditions: n(BzOH)/n(H 2 O 2 ) = 1:2, 5 wt.% catalyst amount relative to BzOH, 4 h, 20 mL H 2 O and 393 K.

Process optimization and model analysis
To optimize the reaction conditions of benzaldehyde (BzH) yield and assess interactions between experimental variable pairs, the factor-designed experiments and RSM were used. Table 3 listed the results of 29 experimental runs and the response values. By means of multiple regression analysis, using Design-Expert 6.0.5 software, 65 the response Y (i.e., BzH yield), which may be related to independent experimental parameters by a quadratic model based on equation 2, can be indicated as: Y = 93.23 − 0.42x 1 − 4.93x 2 + 1.15x 3 − 0.65x 4 − 5.10x 1 2 − 6.58x 2 2 − 3.18x 3 2 − 4.74x 4 2 − 2.91x 1 x 2 − 0.19x 1 x 3 − 4.72x 1 x 4 + 3.52x 2 x 3 + 1.72x 2 x 4 + 3.51x 3 x 4 (6) where the coded values x 1 , x 2 , x 3 and x 4 are catalyst amount, BzOH/H 2 O 2 mole ratio, water amount, and reaction time, respectively. From regression of equation 6, a value of 0.9891 was obtained for coefficient of determination (R 2 ), exactly confirming that the model was credible and well fitted for experimental and predicted values.
Moreover, the standard ANOVA was used to determine if the quadratic model was sufficient and fitted the data, and the conclusions were listed in Table 4. From Table 4, the F-value of model for BzH yield was much larger than the tabular F-value for the 5% significance level. The F-value of 90.64 implied that the model was significant, and less than 0.0001 at 5% confidence level for P-value meant that the model was very significant. In this case, the significant model terms, x 2 , x 3 and x 4 , showing a low P-value at 5% confidence level, demonstrated that these variables were significant. The value of Pred R-Squared (0.9424) was consistent with the Adj R-Squared of 0.9782, definitely implying that the model was of great significance. Besides, coefficient of variation (CV) was 1.04%, together with lack of fit F-value (3.11), implying that the model was more accurate. The experimental design was credible and there was less noise ratio and pure error.
The analysis chart was drawn according to the regression line equation, and the obtained two-dimensional (2D) contour plots together with three-dimensional (3D) response surface plots obtained were shown in Figures S4  and S5 (SI section), respectively. The contour plots visually reflected the effect of the interaction of various variables. The circle indicated that the interactive effect between two variables was not significant, and the ellipse indicates that the interaction was significant on these two factors. Therefore, from the Figures S4a and S5a (SI section), it can be observed that the interaction of x 1 x 2 was significant, and similarly, the interactions of x 1 x 4 , x 2 x 3 , x 2 x 4 , x 3 x 4 were significant.
Combined with the mathematical analysis of the regression model, the optimal process parameters for oxidation of BzOH over the Sn 1/2 H 2.0 PW 12 O 40 catalyst were obtained with amount of catalyst (x 1 ) = 5.49 wt.%, BzOH/H 2 O 2 mole ratio (x 2 ) = 1:1.26 mol mol −1 , water amount (x 3 ) = 16.95 mL and reaction time (x 4 ) = 3.33 h. Three parallel experimental runs were conducted to verify the validity of the model under the optimum condition, and the average BzH yield was 95.1% with x 1 = 5.5 wt.%, x 2 = 1:1.25 mol mol −1 , x 3 = 17 mL, and x 4 = 3.3 h. The results were in good agreement with predicted value (95.6%). Therefore, the regression model was deemed with effectiveness and accuracy in predicting BzH yield.

Catalyst recycling
For reducing the experiment cost, the stability and recyclability of Sn 1/2 H 2 PW 12 O 40 catalyst were investigated for the oxidation of alcohol. The tests were performed under above optimal conditions and the results were shown in Figure 3. The Sn 1/2 H 2 PW 12 O 40 catalyst was mainly deposited in the lowest layer of the component after the reaction, which made the catalyst better reused by simple filtration due to the low solubility of the catalyst in the reaction system. After the leach of catalyst from the system for each cycle, it was washed by ethyl ether and dried under vacuum at 353 K for 12 h. As displayed in Figure 3, the Sn 1/2 H 2 PW 12 O 40 catalyst exhibited good recyclability and stability after six consecutive experimental cycles. The BzOH conversion and BzH yield declined marginally from 99.1 and 95.1% of the first run to 94.9 and 92.8% after six consecutive running cycles, respectively. Additionally, the above optimum reaction conditions. The experiments were conducted at various reaction temperatures with different time. The reactant order of (BzOH; denoted as α) and oxidant (H 2 O 2 ; denoted as β) had already been defined in equations 3 and 4. To correlate reaction rate with reactant concentration, the changes in concentration versus time were indispensable. In order to obtain the reaction order of BzOH (α), the reaction was carried out by changing the concentration of BzOH while keeping its corresponding constant. A set of parallel experiments were conducted under the initial amount of benzyl alcohol varying from 5.4 (0.050 mol) to 3.6 g (0.033 mol) and the time changing from 5 to 45 min. By fitting the data of each point, four fitted lines were obtained, the slope was the reaction rate (r), and the intercept was the initial concentration. According to equation 4, the logarithmic curves of concentration and reaction rate were derived in Figure 4b. The reaction order of BzOH (α = 1.57) was obtained. Similarly, the linear decline of the H 2 O 2 concentrations over time was shown in Figure 5a. The reaction rate and concentration of H 2 O 2 were shown in Figure 5b and the reaction order of H 2 O 2 was 1.07.
Also, the pre-exponential factor (k 0 ) of the reaction and activation energy (E a ) were defined in equation 5. To obtain the pre-exponential factor k 0 and activation energy E a , the optimal reaction conditions were carried out under 5.5 wt.% of catalyst amount, 1:1.25 mol mol −1 of n(BzOH)/n(H 2 O 2 ), 17 mL of water, and 3.3 h for reaction with the temperature changing from 363 to 393 K (see Figure 6a). According to the temperature based on rate equation, the plots of ln k vs. 1/T were obtained by associating with equation 5 (Figures 6a and 6b). An activation energy E a = 21.75 kJ mol −1 was obtained. The value was slightly lower than that catalyzed by  [DMBPSH]H 2 PW 12 O 40 catalyst (36.18 kJ mol −1 ), 37 and much less than that obtained over [PheH]H 2 PW 12 O 40 catalyst (56.7 kJ mol −1 ) on the oxidation of BzOH to BzH with H 2 O 2 , 38 as well as oxidation of phenyl ethanol catalyzed by N-chlorinated p-toluenesulfonamide (p-TSA) salt (81.3 kJ mol −1 ). 66 The results indicated that Sn 1/2 H 2 PW 12 O 40 catalyst was a highly effective catalyst for the oxidation of BzOH with H 2 O 2 .
Over these synthesized catalysts, the Sn 1/2 H 2 PW 12 O 40 catalyst showed the best catalytic performances and was exploited in process optimization in order to acquire suitable conditions at catalyst amount of 5.5 wt.%, BzOH/H 2 O 2 mole ratio of 1:1.25, 17 mL water amount and 3.3 h of reaction time. An excellent yield of BzH (95.1%) was obtained. These optimum experimental parameters were consistent with the RSM prediction on the basis of BBD model. According to the developed kinetic model, the reaction order was 2.64 and the E a was 21.75 kJ mol −1 . Moreover, recyclability study had corroborated the effectiveness of Sn 1/2 H 2 PW 12 O 40 catalyst and reproducibility for oxidation of alcohols.

Supplementary Information
Supplementary data (FTIR, XRD, TGA-DTG spectra,  contour plots and 3D response surface plots) are available free of charge at http://jbcs.sbq.org.br as PDF file.