In order to address environmental issues, green chemistry-focused researchers typically concentrate on the application of various natural and renewable raw materials for the synthesis of polymeric materials. In particular, the use of vegetable oil-based materials have found numerous applications [1], such as stabilizers and plasticizers of thermoplasts [2, 3], lubricants [4], and polymer precursors for the preparation of various polymers (such as polyesters [5], polyamides [6], and polyurethanes [7]). Compared to polymers made from petroleum-based resources, bio-polymers are biodegradable, non-toxic, environmentally friendly, and inexpensive. In general, prior to being suitable polymer precursors, vegetable oils must be modified by converting their double bonds into more reactive functional groups such as, epoxide, acrylate, or hydroxyl groups [8]. Among those groups, epoxides are the most promising candidatesdue to the potential application.
Industrially, the epoxidation of vegetable oils is currently carried out via the Prileshajew reaction [9], in which unsaturated oils react with percarboxylic acid, which is prepared using formic or acetic acid, as well as hydrogen peroxide in the presence of strong mineral acids such as H2SO4, HCl, or HNO3 as catalysts. However, there are various drawbacks for this process, for example, the presence of the strong acid causes side reactions such as oxirane-ring opening, which produces diols, hydroxyesters, and other dimers [10], corrosions, in addition to issues associated with the separation of the water soluble mineral acid, etc. Hence, the exploration of heterogeneous catalysts is significant [11]. In this light, transition metal complexes such as MoⅣ [12], WⅥ [13], NbⅤ [14], and TiⅣ have emerged as catalysts for epoxidation reactions. Among them, Ti-containing materials, e.g., titanium-silica, Ti-MCM-41, and hierarchical TS-1 (HTS-1) [15-18] have been investigated due to their desirable activities and stabilities in epoxidations in the presence of hydrogen peroxide or organic hydroperoxides.
TS-1 is widely used as a highly selective and environmentally benign catalyst in the presence of H2O2 as an oxidant. However, the catalytic oxidation of bulkier substrates is limited due to steric constraints imposed by the 10 MR micropore apertures. Therefore, considerable efforts have been devoted to creating mesopores or macropores in crystalline TS-1 to increase the accessibility of bulky substrates to the internal surface, and consequently, enhance catalytic performance. In order to enhance the performance of HTS-1 in the epoxidation of bulky substrates, several methods such as dealumination, desilication, and other chemical treatments to form defect sites, hard or soft templating methods, have been introduced [19-25]. Recently, a new synthetic approach was developed by our group using low-cost and versatile polyquaternium-6 as a mesoporogen; the obtained materials contained remarkably active, easily accessible tetrahedral Ti sites and showed excellent oxidative desulfurization performance [26].
In general, olefin epoxidation is significantly influenced by the reaction conditions [27], such as the amount and concentration of the oxidant, reaction temperature and time, and amount of catalyst. Herein, in order to gain a deeper understanding of the influence of the aforementioned parameters on the epoxidation of vegetable oil-based unsaturated fatty acid methyl esters (FAMEs), epoxidation reactions were conducted using HTS-1 with polyquaternium-6 as a mesoporogen in the presence of hydrogen peroxide under different conditions. For simplicity, methyl oleate (MO) with only one C=C double bond was chosen as a model unsaturated FAME. Furthermore, three reaction variables, such as H2O2/C=C molar ratio, amount of catalyst, and temperature of the epoxidation process of MO were optimized. Response surface methodology (RSM)-based Box-Behnken Design in 17 experimental runs was used for the optimization using Design Expert 8.0.6 software.
HTS-1 was synthesized according to a literature procedure [26] using tetrapropylammonium hydroxide (TPAOH, 1.42 mol/L aqueous solution, Shanghai Kairui Chemical Co.) and polyquaternium-6 (39–43 wt.%, Haining Huangshang Chemical Co.) as micropore and mesopore structure-directing agents SDAs, respectively, as well as tetraethylorthosilicate (TEOS, Tianjin Kermel Chemical Co.) and tetrabutyltitanate (TBOT, Tianjin Kermel Chemical Co.) as silica and titanium sources, respectively. Briefly, 16.9 mL of an aqueous solution of TPAOH was added dropwise to 11.2 mL TEOS under vigorous stirring and the mixture was hydrolyzed for 1.5 h at room temperature. In a separate flask, 0.34 mL of TBOT was dissolved in 1.7 mL of isopropyl alcohol (IPA, Tianjin Kermel Chemical Co.), and then mixed with 10.1 mL of the TPAOH solution under stirring for 0.5 h to obtain a clear gel containing Ti. Next, both phases were mixed, stirred for 30 min and the alcohol was removed at 80 ℃, followed by the addition of 2.0 g of polyquaternium-6. The resulting mixture was stirred for another 24 h. The molar composition of the synthesized gel was 1SiO2:0.02TiO2:0.25TPAOH: 1IPA: 30H2O. The synthesized gel was transferred into a Teflon-lined stainless steel autoclave, in which the crystallization was continued at 170 ℃ for 96 h. Finally, the as-obtained solid was isolated by centrifugation, washed thoroughly with deionized water, dried at 100 ℃ overnight, and calcined at 550 ℃ for 8 h to remove the organic templates.
Powder X-ray diffraction (XRD) patterns were recorded on a Rigaku D/Max 2400 diffractometer using Cu Kα radiation run at 40 kV and 100 mA with a scan speed of 6°/min and a scan step size of 0.02°. The electronic state of Ti in the titanosilicate samples was analyzed using a JASCO UV550 spectrometer with a white broad as the standard. FT-IR spectra were recorded on a Bruker EQUINOX55 spectrometer, using the KBr pellet technique. The titanium content of the samples was determined by elemental analysis via optical emission spectrometry with inductively coupled plasma (ICP-OES, PerkinElmer/Nex ION 300D). The crystal size and morphology were determined by a NOVA NanoSEM 450 scanning electron microscope (SEM) from FEI. N2 physical adsorption-desorption measurements were carried out at –196 ℃ using a Quantachrome Autosorb-1MP after degassing the sample under vacuum at 350 ℃. The total specific surface area and pore-size distribution were calculated from adsorption data employing the Brunauer-Emmett-Teller (BET) method and Barrett-Joyner-Halenda (BJH) adsorption algorithm, respectively. Total pore volume was estimated from the amount of nitrogen adsorbed at a relative pressure of 0.99 and the micropore volume was determined by the t-plot method.
The catalytic reactions were performed batchwise in a reactor equipped with a reflux condenser using acetonitrile (Tianjin Kermel Chemical Co.) as the solvent, H2O2 (50 wt.%, Shanghai HABO Chemical Co.) as the oxidant, and chlorobenzene (99.8%, Tianjin Kermel Chemical Co.) as the internal standard. All catalysts were pretreated at 120 ℃ in dry air for 2 h prior to use. The typical procedure was as follows: 10 mg of catalyst, 20 μL (58 μmol) of substrate (methyl oleate, 99 wt.%, Aladdin) and 10.5 μL (169 μmol, oxidant/substrate molar ratio = 2.9) of H2O2 were added to 5 mL of acetonitrile, followed by the addition of 15 μL chlorobenzene. The reaction mixtures were stirred at 90 ℃ for 9 h, and the catalyst was recovered by centrifugation.
The reaction products were identified by GC-MS (HP6890/MS5973) with a HP-5 capillary column (0.25 mm × 30.0 m × 0.25 μm). Three types of analysis were performed to quantify the epoxidation reaction results. (ⅰ) GC analysis, using an HP6890 GC equipped with an HP-5 capillary column (0.25 mm × 30.0 m × 0.25 μm) and FID detector; conversion was defined as the mass ratio of converted MO to the initial mass of reactant and selectivity was defined as (moldesired product)/(molall products). (ⅱ) 1H NMR analysis, following a literature protocol [28], using a Bruker Avance Ⅱ 400 NMR spectrometer; conversion, selectivity, and yield were determined according to Eqs. (1)–(3), respectively. In Eqs. (1)–(3), A2, 0, A2, t, A5.35, 0, and A5.35, t are the integrations of the signals at δ 2.01 or 5.35 ppm in the substrate and in the product; similarly, AS, 0 and AS, t are the integrations of the resonance peaks of the internal standard in the substrate and the product spectra (the signal at δ= 0.88 ppm (–CH3, t) was used as internal standard); NS represents the number of protons in the internal standard, whereas N2, 01 represents the number of protons for the signal at 2.01 ppm. (ⅲ) The iodine value analysis, using the standard method of GB/T 5532-2008 for the evaluation of the conversion of the unsaturated bond (Eq. (4)), where (Iodine Number)0 and (Iodine Number)t represent the iodine number in the reactant and product, respectively.
The XRD profile of HTS-1 after calcination at 550 ℃ (Fig. 1(a)) clearly demonstrated that HTS-1 exhibited characteristic diffraction peaks at 2θ = 7.8°, 8.8°, 23.2°, 23.8°, and 24.3°, which were in agreement with MFI-structured materials [29]. In addition, the relatively high intensity of the diffraction peaks of HTS-1 indicated the high crystallinity of the material. The nature and coordination of titanium species in HTS-1 were characterized using a diffuse-reflectance (DR) UV-Vis spectrometer (Fig. 1(b)). The spectrum displayed a band with a maximum centered at 210–220 nm, which originated from the charge transfer of the oxygen 2p electron to the empty 3d orbit of framework Ti atoms, which was characteristic of isolated, tetrahedral coordination Ti species. The obtained HTS-1 zeolite also showed an absorption band centered at 260–300 nm, which was related to pentacoordinated Ti generated by the interaction of Ti species with moisture or hexacoordinated Ti species caused by the partial polymerization of non-framework Ti species through the formation of Ti–O–Ti bonds [30]. No band was observed near 330 nm, which indicated the absence of the anatase phase. The FT-IR spectrum of HTS-1 (Fig. 1(c)) suggested that there were nearly no silanol groups on the external surface of HTS-1, as indicated by the absence of peaks around 3500 and 3750 cm-1, which were characteristic stretching bands of hydrogen-bonded hydroxyl groups at defect sites and isolated silanol groups, respectively; thus, HTS-1 appeared to have a relatively high hydrophobicity. The bands centered at 1100, 800, and 450 cm-1 in the FT-IR spectrum were assigned to the asymmetrical stretch, symmetrical stretch, and bend stretch of Si–O in the SiO4 units [31], and the band at 960 cm-1 confirmed the presence of tetrahedral framework Ti species. The absorption band around 550 cm-1 was attributed to the vibration of double five-membered ring units, which are considered the fingerprint of the MFI structure.
The porosity of HTS-1 was examined by N2 adsorption and desorption measurements (Fig. 2). A combination of type Ⅰ isotherm with an uptake at p/p0 < 0.1 and type Ⅳ isotherm with hysteresis at p/p0 = 0.4-1.0 was observed. The hysteresis implied the presence of mesovoids, which were centered at 7 nm, calculated by the Barrett-Joyner-Halenda (BJH) method from the adsorption branch (Fig. 2, inset). The total pore volume of the sample was 0.43 cm3/g. The SEM images of the HTS-1 crystals (Fig. 3) exhibited well-defined crystal edges with a regular cubic morphology with a size of about 300 nm, which was similar to conventional TS-1 crystals.
With the purpose of exploring the reliability of the experimental data, three methods (GC analysis, 1H NMR analysis, and iodine value titration) were utilized to analyze the reaction mixtures (Table 1). Similar values for conversion were observed, which proved the reliability of the obtained data. The discrepancy in the selectivity values between GC and 1H NMR results might be due to the difference in the sensitivity factors [28].
The effect of H2O2/C=C molar ratio (0.8 to 5.7) on the epoxidation of MO was studied (Fig. 4). It was evident that by increasing the amount of oxidant, the conversion of MO increased almost linearly from 44% at 0.8 to 94% at 5.7, suggesting that higher H2O2/C=C molar ratios were favorable for the epoxidation of MO. Overall, the yield of epoxides increased in parallel with the increment of MO conversion. The selectivity for methyl 9, 10-epoxy stearate (ME) remained constant, whereas no oxidative cleavage by-products (methyl 9-oxononanate and pelargonic aldehyde or acid, Scheme 1) were detected by GC at low oxidant concentrations, which could be attributed to the weaker oxidation ability of H2O2 under the reaction conditions. The efficiency of H2O2 (100% (mol oxidized products)/(mol H2O2 consumed)) was also investigated (Fig. 4). Although, theoretically, equimolar amounts of hydrogen peroxide and the C=C of MO are required to complete the reaction, an excess H2O2 was necessary due to the decomposition of H2O2 into O2 and H2O, which was unavoidable in the catalytic system. The conversion of H2O2 leveled off around 90% with increased oxidant amounts, and only slightly decreased at the H2O2/C=C molar ratio of 5.7, indicating that the decomposition of H2O2 was more severe at high H2O2/C=C molar ratios, which resulted in the lower H2O2 efficiency.
The influence of oxidant concentration was studied at a fixed H2O2/C=C molar ratio (Table 2). No notable differences between the epoxidations using (30 and 50) wt.% H2O2 as the oxidant were observed. This observation contradicted the finding of Poli et al. [32]; however, the discrepancy could likely be attributed to the minor difference in amount of water accompanying with added H2O2 (Table 2).
The effect of catalyst mass was studied (Fig. 5) in the range of 5 to 70 mg. By increasing the amount of the catalyst, the conversion of MO increased from 62% at 5 mg to 96% at 50 mg, while further increase of the catalyst mass did not improve the conversion. Conversely, the selectivity of epoxy MO decreased continuously, which could be mainly attributed to side reactions that were likely to be more prevalent at higher concentrations of epoxides when large amounts of the catalyst were used. The turnover number (TON: calculated as the molar amount of MO converted after 9 h of reaction divided by the molar amount of Ti atoms in the catalyst), decreased with increasing amounts of catalyst, which indicated that the efficiency of Ti active sites of the catalyst decreased. Subsequently, the stereoselectivity issue in the product formation was studied. A cis/trans-epoxide molar ratio of nearly 1:1 was achieved in the absence of the catalyst, whereas the cis-epoxides were the major products when HTS-1 was used, thus confirming that the epoxidation of cis-MO followed a heterolytic mechanism in the presence of Ti active sites [15].
To confirm the thermodynamic impact on MO epoxidation, the reaction was studied under different temperatures (Fig. 6(a)). The conversion of MO increased while the epoxide selectivity slightly decreased upon increasing the reaction temperature. The cleavage by-products were not detected at 50 ℃, thus suggesting that the oxidative cleavage reaction did not take place at low temperatures. It was previously reported that the epoxidation of MO was limited by mass-transport processes [33], and thus high temperatures could improve the conversion. As a consequence, the temperature should be taken into account when considering the interplay of conversion and selectivity of this reaction. The conversion of MO increased significantly during the first 4 h (Fig. 6(b)), indicating that the epoxidation of MO was efficiently carried out using HTS-1. The epoxide selectivity did not appreciably change with reaction time and remained at about 95%, indicating that side reactions did not significantly affect the efficiency of this catalytic system.
A Box-Behnken experimental design was used to identify the relation between the response factor (yield of epoxy MO) and the three variables (H2O2/C=C molar ratio, catalyst mass, and temperature), which were determined to be significant, as well as to determine the optimal combination of variables (Table 3). A total of 17 experiments were carried out separately for the three independent variables at three levels to determine the experimental response of epoxy MO yield (Table 4) and the results were analyzed using the quadric equation obtained from response surface methodology (RSM) modeling via the Box-Behnken method (Eq. (5)). The quadratic regression model was expected to be very significant because its P value was below 0.01. The influence of each independent factor on the model was tested for its level of statistical significance by analyzing the variance. The tested terms included linear terms (A, B, and C), square terms (A2, B2, and C2), and interaction terms (AB, AC, and BC). The significance of each coefficient was determined by P values (Table 5). It was previously reported smaller P values indicated more significant corresponding coefficients [34]. The linear and quadratic effects of H2O2/C=C molar ratio and catalyst mass had a remarkable influence on the epoxy MO yield, as indicated by their significance level (P < 0.05). Therefore, these two factors were the limiting conditions and minimal variation in their values altered the yield of the epoxide. The response surface plots were constructed for the variations in target product yield according to H2O2/C=C molar ratio, catalyst amount, and reaction temperature (Fig. 7). In each plot, two factors were varied, while the other was kept constant at the zero level. The plots derived from the quadratic model were follows:
The yield of epoxy MO increased sharply as the amount of catalyst increased at low H2O2/C=C molar ratios (Fig. 7), while this tendency slowed down at high H2O2/C=C molar ratios. There was a moderate interaction between H2O2/C=C molar ratio and catalyst amount, as indicated by their elliptical contours and P value (P = 0.0869). Similarly, the mutual interaction effect between the catalyst amount and temperature was significant, whereas that of H2O2/C=C molar ratio and temperature was negligible. The optimal epoxy yield was achieved at H2O2/C=C molar ratio = 5.39, catalyst amount = 0.03 g, reaction temperature = 110 ℃; the experimental epoxy MO yield (94.90%) was consistent with the predicted one (95.30%). The H2O2 decomposition under different reaction conditions (Table 6) clearly indicated that the conversion of H2O2 was relatively low in the absence of the catalyst. However, when HTS-1 was introduced into the reaction system, the unproductive decomposition of H2O2 was significant, which confirmed that HTS-1 could also catalyze the decomposition of hydrogen peroxide under high temperatures and long reaction time. When MO was added into the reaction system, the conversion and efficiencies of H2O2 were found to be 96.63% and 19.83%, respectively. Hence, future studies will concentrate on the modification of the catalyst, so that a high conversion of MO and epoxide selectivity could be achieved under more moderate reaction conditions. Based on this, the decomposition of H2O2 can be suppressed and the utilization efficiency of H2O2 can be improved.
The recyclability of the HTS-1 catalyst in the MO epoxidation was evaluated (Fig. 8). The conversion of MO did not change significantly after the second run without any activation operation, whereas the conversion decreased to 79% in the third run, and only 71% yield of the epoxy product was obtained, mainly because the organic residues blocked the active Ti species. After being reactivated by calcination to remove the high-molecular-mass sticky substrates from the catalyst surface, the catalyst showed almost restored activity in the fourth run, demonstrating the excellent regenerability of the catalyst. On the other hand, the selectivity of the epoxy MO decreased slightly and was maintained at a high level during reuse.
The influence of various reaction parameters, i.e., H2O2/C=C molar ratio, oxidant concentration, catalyst amount, reaction temperature, and time on the epoxidation of unsaturated FAMEs was studied systematically. Specifically, the epoxidation of MO as a model substrate was carried out in acetonitrile, with H2O2 as the oxidant and HTS-1 employing polyquaternium-6 as a mesopore-forming agent as the catalyst. HTS-1 exhibited excellent catalytic performance in the MO epoxidation due to the highly active tetrahedral Ti sites, easy accessibility of active sites for bulky molecules, and excellent hydrophobicity. Process optimization was performed using response surface methodology. Based on the statistical analysis, the H2O2/C=C molar ratio and catalyst amount were the main factors affecting the MO epoxidation. The conversion of MO reached 100% with an epoxide selectivity of 94.9% by optimizing the process variables. As such, this work may pave the way for improving the conversion and selectivity in the industrial production of epoxidized vegetable oils via adjusting the reaction conditions.
2018,
Vol. 39 Issue (5): 964-972 
PDF
2018,
Vol. 39 Issue (5): 964-972 PDF