Selective Oxidation of Glycerol with Hydrogen Peroxide Using Silica-Encapsulated Heteropolyacid Catalyst

YUAN Mingming, LI Difan, ZHAO Xiuge, MA Wenbao, KONG Kang, NI Wenxiu, GU Qingwen,HOU Zhenshan

Key Laboratory for Advanced Materials, Research Institute of Industrial Catalysis, School of Chemistry & Molecular Engineering,East China University of Science and Technology, Shanghai 200237, P. R. China.

Abstract: The Keggin type heteropolyacids (HPAs) have attracted increasing attention due to their strong Bronsted acidity and excellent redox properties, which could play an important role in accelerating the conversion of bio-derived molecules. In this work, heteropolyacid (HPA, H4PMo11VO40) encapsulated by silica was synthesized by a sol-gel method and a sequential silylation technique(HPA@SiO2-N2-S). The as-synthesized material was characterized by Fourier transform infrared spectroscopy (FT-IR), X-ray diffraction (XRD),thermogravimetric analysis (TGA), scanning electron microscope (SEM) and transmission electron microscopy (TEM). The FT-IR spectra show that the HPA anions preserved their Keggin structure when incorporated into the catalyst. The XRD patterns show that HPA molecules are uniformly dispersed within the silica network. The SEM and TEM images confirm that the catalyst was composed of spherical nanometer-sized particles. The porous properties of the catalysts measured by the N2 adsorption-desorption isotherms indicate that the Brunauer, Emmett and Teller (BET) surface area of pure SiO2 was 287 m2·g?1, but upon encapsulation of HPA into the silica matrix, a lower surface area (245 m2·g?1) was measured for the resulting material. In addition, the pore diameter was reduced after silylation. Furthermore, the hydrophobicity of the catalysts was investigated by the measurement of contact angle (CA)with water. The SiO2 and SiO2/HPA catalysts were completely hydrophilic and the contact angle was close to 0°. However,the contact angle of the silylated catalyst was determined to be 137°, indicating that the silylation procedure significantly increased the hydrophobicity of the catalyst. The as-prepared catalysts were also used as heterogeneous catalysts for the selective oxidation of glycerol. The prepared material exhibited excellent catalytic activity towards glycerol oxidation, in which the glycerol can be selectively transformed into formic acid (ca. 70% selectivity) and glycolic acid (ca. 27% selectivity)using H2O2 as an oxidant under mild reaction conditions. The effect of the silylation procedure on the recyclability of catalyst wasalsoinvestigated in this work. The characterizations described above indicated that silylation procedure can significantly increase the hydrophobicity and limit the pore sizes, resulting in high leach-resistance towards HPA, thus improving the recyclability of the silica-encapsulated HPA catalyst, as compared to the SiO2/HPA catalyst prepared with the conventional impregnation method. Furthermore, the conversion in the second catalytic run is even higher than that of the initial run, which is likely because more active sites are exposed after the first run. The catalyst can be reused for at least five cycles without any leaching of HPA. The spent catalyst did not undergo structural changes, as revealed by FTIR, XRD, and SEM characterization. Moreover, it was found that the strong Bronsted acid additives played a crucial role in the catalytic oxidation of glycerol.

Key Words: Glycerol oxidation; Heteropolyacid; Encapsulation; Formic acid; Hydrogen peroxide

1 Introduction

Due to the depletion of fossil fuels and the improvement of awareness of environmental protection, biodiesel resources have been receiving extensive worldwide attention and are being rapidly flourished because of their excellent renewability,environmental performance, and innocuous advantages1–3.Biodiesel is the best known in biofuels, obtained through the transesterification between vegetable oils and short chain alcohols with a catalytic process4. The glycerol in an amount of ca. 10% (w, mass fraction) of the overall biodiesel production is generated as a by-product in this process, which has been identified as an important platform compounds for producing high value chemicals5. Representative reaction pathways for the glycerol valorization contain hydrodeoxygenation,hydrogenation, hydrogenolysis, oxidation and dehydration etc.Among these various utilization approaches, the glycerol oxidation is highly attractive, which can provide C1 products such as formic acid (FA), C2 products such as glycolic acids(GCA) and oxalic acid (OA), and many C3 products such as tartronic acid, glyceraldehyde, glyceric acid, and dihydroxyacetone. These products are all high-added chemical intermediates and are still being obtained through expensive processes or any other way6–8. Among the routes for the utilization of these chemicals, the increasing interest in efficient FA production aims at its application as hydrogen carrier9,10. As FA owns 4.4% hydrogen content and some well-known systems could decompose it under mild conditions to hydrogen and carbon dioxide11–13. Additionally, FA is also a convenient source of C1 raw material for the chemical synthesis. For example, the efficient hydrogenation/ disproportionation of formic acid (FA)to methanol by using iridium catalysts has been reported recently14.

Over the past decades, the selective oxidation of glycerol has been conducted to obtain valuable products with molecular oxygen. The seminal studies have founded that supported gold catalyst have high selectivity and conversion for glycerol oxidation15–17. In addition, the aerobic oxidation of glycerol has been widely studied using noble metal such as Pt18,19, Pd20etc.In these catalytic systems, the noble metals play a key role in glycerol oxidation and the catalytic performance depends on the basicity of the reaction medium. However, due to its basicity, it produces a lot of organic acid salts and thus the mixture after reaction needs further neutralization and acidification so as to obtain the target products. In addition to its basicity, the other disadvantage is that the supported noble metal catalysts are rather expensive for the further practical application and also tend to lose activity rapidly due to the oxidation of metal surface and thus result in the poisoning of the active sites21. Therefore,the design of other catalyst systems was becoming a hot point in the current research field.

On the other hand, hydrogen peroxide has been widely used as an oxidant in organic oxidation reactions owing to its environmentally friendly nature in that the degradation products of hydrogen peroxide are only oxygen and water22. Hydrogen peroxide is strong oxidant which can oxidize a broad variety of inorganic and organic substrates in liquid-phase reactions under very mild reaction conditions23. Hence, the catalytic oxidation of glycerol has been carried out by using the hydrogen peroxide as an oxidant recently24–26.

The catalyst acidity could play a very important role in accelerating the hydrolysis of glycerol molecules27,28. The Keggin type heteropolyacids (HPAs) have attracted increasing attention due to their strong Bronsted acidity and environmental friendliness29–31. Meanwhile, HPAs also showed excellent redox properties32. Their low surface area (＜ 10 m2·g?1) and high solubility in polar solvents limit its potential catalytic application. For example, HPAs have been employed as efficient catalysts in aqueous media for the oxidation of glycerol into FA,but the separation of homogenous HPA catalysts was tedious due to miscibility of HPAs and FA with water media33.

Generally, the heterogeneous catalysts have some particular advantages over homogeneous ones, for example, the solid catalysts can be removed easily from the reaction media and thus recycled. From this point of view, solid catalysts usually represent a better option in liquid phase reaction. On the basis of discuss above, in this work, we attempted to immobilize HPA(H4PMo11VO40) within silica matrix via a sol-gel method and the sequential silylation technique, which offered an appropriate hydrophobicity and also shielding of HPA, suppressing the leaching of HPA and but allowing penetration of reactants and products through porous silica during the aqueous reaction. It was found that the as-obtained catalyst exhibited excellent activity and reusability for glycerol oxidation using H2O2as an oxidant under very mild conditions.

2 Experimental

2.1 Materials

Disodium hydrogen phosphate, sodium metavanadate,sulfuric acid, sodium molybdenum oxide, ethyl ether, dimethyl sulfoxide, tetraethyl orthosilicate, acetonitrile, glycerol, toluene,dimethyldichlorosilane were purchased from Sinopharm Chemical Reagent Co., Ltd., Shanghai. Distilled water used in this work was produced by our own laboratory. High purity N2(99.999%) was supplied by ShangNong Gas Factory. All other chemicals (analytical grade) were from Sino pharm Chemical.Toluene (analytical grade) was dried by using the standard methods.

2.2 Catalyst preparation

2.2.1 Preparation of H4PMo11VO40(HPA)

In a typical synthesis route, 0.71 g of disodium hydrogen phosphate was dissolved in 10 mL of water, and then 0.61 g of sodium metavanadate that had been dissolved by boiling in 10 mL of water was mixed with the above solution under stirring.The cooled mixture was acidified with 0.5 mL of concentrated sulfuric acid to a red color. 11.26 g of sodium molybdenum was dissolved in 20 mL of water, and added to this mixture. Finally,8.5 mL of concentrated sulfuric acid was slowly dropped into the previously prepared aqueous solution with vigorous stirring. In the process, the dark red color changed to a lighter red and the pH of the solution was 1–2. The water solution was cooled and the heteropolyacid extracted with 40 mL of ethyl ether34. In the extraction process, it will be divided into three layers and the heteropoly etherate was present in the middle layer. After separation, a stream of air was passed through the heteropoly etherate layer to free it of ether, which offered the orange solid powder as HPA (3.70 g).

2.2.2 Preparation of SiO2-encapsulated HPA catalyst

Absolute ethanol (10 mL) was mixed with distilled water(0.12 g), then add tetraethyl orthosilicate (1.6 mL) into the mixture. Allowed the solution standing for thirty minutes and followed by adding dimethyl sulfoxide solution (4 mL)containing HPA (0.1 g); in the end, added 4 mL of NH3·H2O(28%). After 2 h, the as-obtained gel was dried. The sample was designated as HPA@SiO2sample. Additionally, silica gel (SiO2)was prepared by the same method besides HPA was not added.

2.2.3 The silylation of HPA@SiO2

Surface silylation of silica was carried out using dimethyldichlorosilane as the silylation agents. The procedures for surface modification of silica with the agents were similar to that in the literature35. 0.8 mmol of dimethyldichlorosilane was dissolved in 10.0 mL of dried toluene. 0.2 g of dried HPA@SiO2composite was dispersed in the above solution. After stirring for 20 min at 80 °C under a N2 atmosphere, the solid was isolated through filtrated, thoroughly washed with toluene for several times. The obtained product was dried for 12 h at 100 °C under vacuum. Then, the dried sample followed by calcination at 250 °C for 1 h in a nitrogen reduction furnace prior to reaction.This sample was named as HPA@SiO2-S-N2. The elemental analysis gave that the actual loading of HPA was about 13%. The as-synthesized silica gel (SiO2) support was impregnated conveniently with HPA to afford 13% HPA/SiO2catalyst.

2.3 Catalyst characterization

XRD analysis of types samples were performed in the 2θ range of 10°–80° on an D/MAX 2550 VB/PC instrument(Rigaku Corporation, Japan) using a graphite crystal a monochromator. The textural properties from N2adsorption isotherms were obtained on NOVA 2200e equipment(Quantachrome, USA). The surface area was obtained from the isotherms in the relative pressure range of 0.0–0.35. Pore volume was determined at p/p0 of 0.99. The inductively coupled plasmaatomic emission spectroscopy ICP-AES (Inductively Coupled Plasma-Atomic Emission Spectroscopy) analysis was carried out on a ICP-710ES instrument (Varian Company,Germany). The sample was putted in a nickel crucible mixed with some sodium hydroxide and then putted into muffle at 500 °C for 30 min to destroy the structure of sample, followed by dissolving with water. FT-IR spectra were recorded from pressed KBr pellets at room temperature on a Magna550(Nicolet, USA). A spectrum of dry KBr was also recorded as background. The thermal stability of catalysts was determined by TGA method (heating rate: 10 °C·min?1; N2 flow, 100 mL·min?1) using Perkin Elmer Pyris Diamond Analyser(Netzsch, Germany). The SEM images were performed on JSM electron microscopes (JEOL JSM-6360LV, Japan). TEM was performed in a JEM 2010 instrument (JEOL Corporation, Japan)operating at 200 kV with a nominal resolution of 0.25 nm. The samples for TEM were prepared by dropping the aqueous solutions containing the NPs onto the carbon-coated Cu grids.Water contact angle (CA) was carried out by the Contact Angle Meter (CA100C, Innuo Company, Shanghai, China) using the droplet profile as a method. The CA was determined using a tangent placed at the intersection of the liquid and solid. A water droplet with a volume of 2 μL was dispensed by a piezo doser onto each sample disk.

The amount of acid sites of the different samples was obtained with temperature programmed desorption of ammonia (NH3-TPD). About 0.15 g of catalyst was loaded in a quartz tube. Prior to each test, the sample was pre-treated in He at 300 °C for 1 h,cooled to 50 °C to remove surface water. Then, the sample was maintain at 50 °C for 1 h and saturated with a 10% NH3-in-N2mixture, and then flushed by He for 1 h to remove physically adsorbed ammonia. Then, the sample was heated to 400 °C at a heating rate of 10 °C·min?1in the same flow of He. The profiles of desorption were recorded using a thermal conductivity detector (TCD, Vodo, Zhejiang, China).

2.4 Catalytic reaction

The oxidation of glycerol was carried out as follows. A mixture of glycerol (1.25 mmol), acetonitrile (0.90 mL), 30%aqueous H2O2 (6.25 mmol), and the HPA@SiO2-S-N2 catalysts(0.06 g) were placed in a 25 mL Schlenk flask equipped with a magnetic stirrer and then stirred at 70 °C for 12 h. When the reaction was finished, the mixtures were filtrated by centrifugation. The resulting supernatant liquid was diluted 5 times with distilled water and analyzed by HPLC using LC-100 chromatograph equipped with a Aminex HPX-87H Column(300 × 7.8 mm, Bio-Rad Company, USA) used with a solution of 0.005 mol·L?1H2SO4 (0.5 mL·min?1) as the eluent at 55 °C.The recovered catalyst washed with ethanol and toluene for several times and dried in a vacuum at 60 °C for recycling tests.The conversion of glycerol and selectivity towards products were calculated as follows:

Hot filtration experiment for the HPA@SiO2-S-N2 catalyst was performed with two parallel experiments. The procedures were given by the following procedures: a mixture of glycerol(1.25 mmol), acetonitrile (0.88 mL), 30% aqueous H2O2 (6.25 mmol), and 0.06 g catalyst was placed in a 25 mL Schlenk flask equipped with a magnetic stirrer and condenser, followed by stirring at 70 °C. Two parallel experiments were performed without or with separating the catalyst from the hot reaction mixture after 8 h36. Hot filtration experiment for the HPA/SiO2catalyst was carried out in a similar method except that the catalyst was isolated after 50 min.

3 Results and discussion

3.1 Catalyst characterization

In this work, the HPA was introduced conveniently into the silica by a sol-gel procedure, followed by the silylation approach. The preparation route has been depicted briefly in Scheme 1.

Scheme 1 The preparation approach of the silica-encapsulated HPA catalyst.

Fig. 1 FT-IR spectra of (a) HPA; (b) SiO2; (c) HPA/SiO2 catalyst;(d) the fresh HPA@SiO2-S-N2 catalyst; (e) the HPA@SiO2-S-N2 catalyst after reused for 5 times.

The obtained samples were first characterized by FT-IR. A small band observed at 1635 cm?1and a broad band at 3420 cm?1was due to absorbed water in Fig. 1a–e. The pure HPA showed IR spectra with the specific bands of the Keggin structure as shown in Fig. 1a, which showed band at 1062 cm?1(vas P―Oa),961 cm?1(vasMo＝Od), 867 cm?1(vasMo―Ob―Mo), and 777 cm?1(vasMo―Oc―Mo), 591 cm?1(δ P―O), respectively.These absorptions are constant with the Keggin structure of the heteropolyanion37. As shown in Fig. 1b, SiO2showed the vibration band at 1095 cm?1, 950 cm?1, 800 cm?1and 584 cm?1,which can be assigned to vas (Si―O―Si), v (Si―O), vs(Si―O―Si) and δ (Si―O―Si) bonds, recpectively38. After the HPA was introduced into the silica matrix, the specific bands of the Keggin structure showed a slight change, as compared with that of the parent HPA. It was noted that bands of HPA/SiO2and HPA@SiO2-S-N2catalyst in the 1200–400 cm?1region were completely or partially overlapped by that of SiO2. For example,the band assigned to the P―O asymmetric stretching vibration at 1062 cm?1of HPA is completely overlapped by the strong band at 1095 cm?1of SiO2(Fig. 1c–e), while the two strong bands appeared in the spectra of HPA/SiO2and HPA@ SiO2-SN2appeared at 950 and 800 cm?1, as a result of the overlapping of the absorption bands of silica and HPA. However, the asymmetric stretching vibration band at 863 cm?1that assigned to Mo―Ob―Mo vibration, which was not found in SiO2, but it indeed appeared in SiO2-encapsulated/supported HPA catalysts(Fig. 1c–e). Additionally, a well defined band appeared at 2965 cm?1after silylation in Fig. 1d–e can be assigned to C―H stretching vibration in surface ―CH3groups39. All these results indicated clearly that HPA has been embedded within silica matrix and also the silylation was successful by dimethyldichlorosilane.

The XRD patterns of the obtained catalysts were depicted in Fig. 2. The most intensive reflections of bulk HPAs appear in the following range: 7°–10°, 17°–23° and 26°–32° (Fig. 2a), which are characteristic of the Keggin structure40, while HPA/SiO2 and HPA@SiO2-S-N2 catalysts did not give obvious diffraction peak and only displayed a large and flat diffraction peak appeared at 2θ = 15°–35°, which was assigned to amorphous silica (Fig. 2a–2e), which evidenced that bulk HPA did not form but HPA was highly dispersed on in the SiO2support.

The porous properties of the catalysts were measured by the N2adsorption-desorption isotherms method (Fig. S1). As shown in Table 1, the BET surface area of pure SiO2was 287 m2·g?1(Table 1, entry 1), the encapsulation of HPA into the silica matrix resulted in a lower surface area (Table 1, entries 3–5). But their surface areas are still higher than that of the catalyst prepared by impregnation (Table 1, entry 2). Interestingly, the encapsulation of HPA into silica matrix could result in the increase of pore diameter (Table S1, entries 1 vs. 2 and 3),possibly arising from the effect of bulky HPA molecules. In the next, the hydrophobicity of the catalysts has been investigated by the measurement of contact angle (CA). The SiO2and HPA/SiO2was completely hydrophilic and the contact angle was close to 0° (Table 1, entries 1 and 2). However, SiO2-encapsulated HPA catalyst (HPA@SiO2) showed slightly stronger hydrophibicity, as compared with that of SiO2 and HPA/SiO2 samples likely due to the difference in surface morphology. Especially, the silylated catalyst even gave a larger CA (Table 1, entry 4). As a result, the silylation procedure can increase obviously the hydrophobicity of the catalysts.Furthermore, the pore diameter reduced rationally after silylation, as compared with that of the patent catalyst (Table S1,entries 2 vs. 3).

Subsequently, the surface morphology of catalysts was examined by SEM. As shown in Fig. 3, the SEM images showed that the structure of SiO2is spherical (Fig. 3a and d), and HPA/SiO2catalyst maintained the morphology of the SiO2but we can see some particles on the surface (Figs. 3b and e). When the HPA was encapsulated into the silica by a sol-gel procedure,the SEM images showed that the structure of HPA@SiO2catalyst is spherical but showed rougher surface (Fig. 3c and f).After silylation, the images of HPA@SiO2-S-N2(Fig. 3g and i)catalyst were similar to that of HPA@SiO2catalyst.Interestingly, the HPA encapsulation into SiO2 through the solgel method can reduce the size of SiO2 particles significantly and the silylation did not influence the surface morphology. As discussed above, the encapsulation of HPA into SiO2(HPA@SiO2) can result in stronger hydrophobicity, as compared with that of HPA/SiO2, which could be attributed to the smaller particle size and the different surface morphology41,42.

Fig. 2 X-ray diffraction patterns of (a) HPA; (b) SiO2;(c) HPA/SiO2 catalyst; (d) the fresh HPA@SiO2-S-N2 catalyst;(e) the HPA@SiO2-S-N2 catalyst after reused for 5 times.

Table 1 Physicochemical properties and catalytic performances of the oxidation of glycerol using different catalysts a

In the next, the TEM images showed that the structure of HPA/SiO2is spherical with about 150 nm diameter (Fig. 4a),while the structure of HPA@SiO2-S-N2catalyst is spherical with about 12 nm diameter Fig. 5g, which was in well agreement with that of SEM images (Fig. 3b vs. 3g). Energy-dispersive X-ray spectrometry (EDS) elemental mapping analysis (Figs. 4 and 5)validated the immobilization of Mo species had a tendency to be distributed around silica particles on HPA/SiO2catalyst (Fig. 4e and f) but highly uniform dispersity within HPA@SiO2-S-N2catalyst (Fig. 5e and f).

The temperature-programmed desorption of ammonia (NH3-TPD) is a useful method to provide general information about the strength distribution of acid sites of solid acids. According to the desorption temperature of the absorbed ammonia, the strength of solid acid sites are formally assigned as weak and strong. The NH3-TPD curves of the samples were shown in Fig.6. The NH3-TPD patterns of both HPA@SiO2-S-N2and HPA@SiO2catalysts showed only a desorption peaks in similar temperature regions (50–200 °C). On the other hand, HPA/SiO2 samples exhibited an over-lapping peak including two peaks around 200 °C and 340 °C, respectively43. This indicated that HPA/SiO2showed a stronger surface acidity than that of HPA@SiO2-S-N2and HPA@SiO2, which was understandable because HPA was mainly loaded on surface silica for HPA/SiO2,leading to more accessible adsorption of NH3 molecules, as compared with that of silica-encapsulated HPA catalysts.

The thermal stability of the catalysts was investigated by TGA. As shown in Fig. S2, both SiO2and HPA@SiO2-S-N2showed a slight weight loss below 120 °C, which could be attributed to the removal of the adsorbed H2O on the surface of SiO2 and the HPA@SiO2-S-N2. Increasing temperature from 120 °C to 700 °C, it was seen that SiO2 only afforded a slight weight loss from 98% to 95%, which could be assigned to the increase of siloxane bridges (Si―O―Si) because of the condensation of isolated silanol groups44. The HPA@SiO2-S-N2catalyst gave a weight loss from 95% to 85.5% in the range of 300° to 700°, which was mainly attributed to the removal of gradual degradation of methyl on the surface of SiO2.

3.2 Catalytic performance

Fig. 3 SEM images of (a, d) SiO2; (b, e) HPA/SiO2;(c, f) HPA@SiO2; (g, i) the fresh HPA@SiO2-S-N2 catalyst;(h, j) HPA@SiO2-S-N2 catalyst after reused for 5 times.

Fig. 4 TEM images of HPA/SiO2 (a), and elemental mapping images of Si (b), O (c), Mo (d), Si and Mo (e), O and Mo (f).

Fig. 5 TEM image of HPA@SiO2-S-N2 (a, g), and elemental mapping images of Si (b), O (c), Mo (d), Si and Mo (e), O and Mo (f).

Fig. 6 NH3-TPD profiles of the (a) HPA@SiO2-S-N2 catalyst;(b) HPA@SiO2; (c) HPA/SiO2 catalyst.

The as-prepared catalysts have been tested for the glycerol oxidation reaction. As shown in Table 1, except SiO2other catalysts were all active for the oxidation of glycerol reaction(Table 1, entries 2–4). However, they exhibited different catalytic activity although the loading amounts of the HPA were same. The fresh HPA/SiO2catalyst showed high catalytic performance for glycerol oxidation, but HPA on catalyst surface is not stable and was visually leached into reaction media after reaction, resulting in yellow effluent and thus giving very poor activity in the second run (Table 1, entry 2). The HPA@SiO2 catalyst showed some improvement for catalyst recyclability,but the obvious decrease of activity was still observed in the second run (Table 1, entry 3). However, after the HPA@SiO2 sample was silylated and then calcined at 250 °C under the N2 flow, the as-obtained catalyst HPA@SiO2-S-N2can be recyclable and even the catalytic activity became better in the second run (Table 1, entry 4), which might be attributed obviously to the strong hydrophobicity and reducing pore sizes,resulting in shielding of HPA, suppressing the leaching of HPA and but allowing penetration of reactants and products through porous silica during the oxidation reaction. Obviously, the higher catalytic activity of fresh HPA/SiO2could be related with these easy accessible strong acidic sites, as presented at NH3-TPD spectra (Fig. 6).

The effects of reaction condition (temperature, reaction time,amounts of catalyst) were investigated over the HPA@SiO2-SN2catalyst. Firstly, the effect of the reaction temperature was shown in Fig. 7a, which indicated that the reaction temperature had a crucial impact on the reaction. It was found that the conversion of glycerol increased continuously and reached a maximum but that decreased above 70 °C, resulting from the decomposition of H2O2 under higher temperature. Additionally,the selectivity to FA kept unchanged but the selectivity to GCA decreases above 70 °C, indicating that GCA could be converted into FA under higher temperature. The parallel experiment with GCA as a substrate showed that GCA indeed can be oxidized into FA in 40% of conversion and 67% selectivity under the same conditions (Table 1, entry 5).

The effect of reaction time was obtained from Fig. 7b. Under the condition of 70 °C, the conversion of glycerol had a gradual increase within the first 5 h and then had a rapidly increment between 5 h and 10 h. However, longer reaction time could result in the decrease of FA and GCA selectivity. This can be explained that FA and GCA can be degraded further into CO2, which has been confirmed by GC-MS (7890A-5975C, Agilent, USA),showing that the gas samples indeed contained CO2, but no CO was detected from formic acid decomposition. Hence, the optimal reaction time is 12 h. Last, the influence of amounts of catalyst on the oxidation reaction of glycerol was also examined at 70 °C for 12 h. As shown in Fig. 7c, the conversion of glycerol was increased with the increase of catalyst until 0.06 g, in which can eliminate the influence of diffusion. The following reaction has been performed according to the conditions above.

Because acidity site has a crucial effect on catalytic activity,the impact of different acid additives on catalytic activity has been evaluated for the glycerol oxidation with H2O2. The results were listed in Table 2, which illustrated that Lewis acids used in this work almost did not influence the conversion of glycerol at all (Table 2, entries 2–4), but Bronsted acid additives can improve the conversion of glycerol significantly (Table 2, entries 5–9). Among of them, CF3SO3H is the most active acid from either conversion, or selectivity to FA (Table 2, entry 9). The control experiments indicated that when only acid additives were employed for glycerol oxidation, most of acids only offered low activity for glycerol conversion (Table S2, entries 1–7).CF3SO3H gave moderate activity (Table S2, entry 8), but the combination of CF3SO3H and HPA@SiO2-S-N2catalyst exhibited much higher catalytic performance (Table 2, entry 9).These results proved unambiguously that strong Bronsted acid played a critical role in enhancing the activity and selectivity of glycerol oxidation over HPA@SiO2-S-N2 catalyst.

Fig. 7 Dependence of glycerol conversion and selectivity on the reaction condition over HPA@SiO2-S-N2 catalyst.

On the base of the results above, HPA@SiO2-S-N2catalyst showed the highest catalytic performance. Hence, HPA@SiO2-S-N2catalyst, was employed for the oxidation of the other biomass platform molecules including ethylene glycol, glucose,sorbitol, fructose, xylitol and 1, 2-propanediol. As shown in Table 3, ethylene glycol, glucose and sorbitol had a low reactivity (Table 3, entries 1–3), while fructose, 1, 2-propanediol and xylitol had a relatively high conversion (Table 3, entries 4–6) in the absence of any acid additive. When CF3SO3H was added as an additive, all biomass-derived molecules afforded higher conversion, although the selectivity to FA was variable(Table 3, entries 1–6). For example, the conversion of fructose improved from 60% to 100% and selectivity to FA from 48% to 62% due to the addition of CF3SO3H. It was noting that although only CF3SO3H afforded the moderate conversion (Table S3) for the oxidation of the other biomass platform molecules in theabsence of the HPA@SiO2-S-N2catalyst, but the activity was much lower that of the combined CF3SO3H and HPA@SiO2-SN2 catalysts (Table 3, entries 1–6). These results indicated that Bronsted acid additives can enhance conversion of other biomass-derived molecules through acid hydrolysis for this reaction28.

Table 2 Different acid additives for the oxidation of glycerol reaction over HPA@SiO2-S-N2 catalyst a.

3.3 Catalyst stability

The reusability of the HPA@SiO2-S-N2catalysts was examined under the optimized conditions (Fig. 8). After reaction, the solid catalyst was recovered conveniently by simple centrifugation, followed by washing with ethanol and toluene for several times. And then it was dried at 60 °C under the vacuum.Then the recovered catalyst was reused for the next recycle. At first, the reusability of HPA@SiO2-S-N2catalyst was evaluated.It was found that the catalyst could be reused for at least five times with high conversion and selectivity to FA. It is worth noting that the conversion in the second run is even higher than that of the initial activity but remained unchanged as starting from the second recycle while the selectivity of the product remained the same (Fig. 8a). In contrast, after strong Bronsted acid CF3SO3H was added to the reaction system, the conversion of glycerol and selectivity to formic acid kept high and did not change obviously during the consecutive five recycle (Fig. 8b).These results strongly suggested that strong Bronsted acid, as well as redox sites is important for glycerol activation, and HPA@SiO2-S-N2catalyst could expose more active sites after the first run, and thereby increasing contact area of the catalyst and substrates, which can be proved by a loss of hydrophobicityarising from a partial removal of methyl group of silica surface by oxidation reaction (Fig. S3c vs. d).

Table 3 Different substrates for the oxidation reaction catalyzed by HPA@SiO2-S-N2 catalyst a.

Fig. 8 Recyclability of (a) HPA@SiO2-S-N2 catalyst without CF3SO3H as an additive; (b) HPA@SiO2-S-N2 catalyst with CF3SO3H (0.5 mmol) as an additive.

Fig. 9 Hot filtration experiments for the glycerol oxidation with H2O2 over (a) HPA@SiO2-S-N2 catalyst; (b) HPA/SiO2 catalyst.Solid square points (●): without isolating catalyst; Solid triangle points (▲): with isolating catalyst and then reaction in the filtrate.

The spent catalyst has also been examined for the possible structural changes. The characteristic vibration bands of heteropolyanions with the Keggin structure at 1095, 950, 863,800 and 584 cm?1were observed in FT-IR spectrum of the used HPA@SiO2-S-N2(Fig. 1d and e), indicating that the structure of the heteropolyanion remains intact. The XRD pattern of the reused catalyst was also in agreement with that of fresh one (Fig.2d and e). The surface morphology of the reused catalyst was shown in Fig. 3h and j, which indicated that the catalyst was still spherical and the morphology of the catalyst did not undergo obvious changes after reaction. The spent HPA@SiO2-S-N2catalyst still showed hydrophobicity (CA = 114°), although the water CA decreased slightly, as compared with that of the fresh one.

According to the above discussion, there came the idea that whether the present catalytic reaction went through a heterogeneous way or not. Thus, the fast hot filtration at the reaction temperature was employed to examine the filtrate for activity. For HPA@SiO2-S-N2 catalyst, as shown in Fig. 9a, no further glycerol was converted once the catalyst was removed,which evidenced unambiguously that the catalytically active species was not present in the homogeneous filtrate. Finally, the possible leaching of HPA into media after the catalytic reaction was determined by ICP-AES, which showed that no P, Mo and V species were found in effluents under our reaction conditions.This result proved that the catalytic reaction proceeded over HPA@SiO2-S-N2catalyst by a heterogeneous mechanism.However, for HPA/SiO2catalyst, as shown in Fig. 9b, because of the leaching of heteropolyacid active species into filtrate,there was still further glycerol was converted in the filtrate after the catalyst was removed.

4 Conclusions

In this work, we employed a sol-gel method and sequential silylation technique to encapsulate HPA into silica to afford an insoluble and easily separable heterogeneous catalyst. The asprepared HPA@SiO2-S-N2catalyst could be reused for at least five times with high conversion and showed no HPA leaching.The silylation procedure obviously increased the hydrophobicity but reduced the pore sizes, which led to high leach-resistance of HPAs during the consecutive reaction runs. Under the best reaction conditions, the conversion of the glycerol can be reached 50% and the selectivity to formic acid was 70% and to glycolic acid 27% by using H2O2as an oxidant, respectively. In addition, the addition of strong Bronsted acid into the reaction system can improve the catalytic activity significantly, which indicated that Bronsted acid sites have an important impact on activating glycerol molecule during the glycerol oxidation reaction.

Supporting Information:available free of charge via the internet at http://www.whxb.pku.edu.cn.