SiO2 Catalysts for

Dec 18, 2017 - Biobased Diesel Fuel Analysis and Formulation and Testing of Surrogate Fuel Mixtures. Industrial & Engineering Chemistry Research. Luni...
0 downloads 8 Views 2MB Size
Article pubs.acs.org/IECR

Cite This: Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Deactivation and Rejuvenation of Pellet MgO/SiO2 Catalysts for Transesterification of Soybean Oil with Methanol to Biodiesel: Roles of MgO Morphology Change in Catalysis Chien-Hsiu Hung,† Chin-Shuh Chen,† Hwo-Shuenn Sheu,*,‡ and Jen-Ray Chang*,§ †

Department of Food Science and Biotechnology, National Chung Hsing University, Taichung 402, Taiwan National Synchrotron Radiation Research Center, Hsinchu 300, Taiwan § Department of Chemical Engineering, National Chung Cheng University, Chia-Yi 621, Taiwan ‡

S Supporting Information *

ABSTRACT: Pellet MgO/SiO2 catalysts with pore diameters of 30 and 40 nm were prepared by incorporating Mg(NO3)2 on granular SiO2 followed with calcination at 500 °C. These pellet catalysts and MgO powder were tested and compared by using transesterification of soybean oil with methanol. Combined results of the test reaction, FT-IR, XRPD, and pore-size distribution suggested that the reaction is catalyzed by both Brønsted and Lewis base sites and is limited by mass transfer. The observation of Mg2+ in the reaction products and gum formed on the catalysts further indicated that the catalysts are deactivated by carbonaceous deposition and MgO leaching. Thus, the deactivated catalysts were rejuvenated by an ethanol wash followed with makeup MgO. However, inferred from synchrotron XRPD results, we suggested that the rejuvenation process cannot recover the morphological change of MgO that took place in the reaction and rejuvenation, leading to a decrease in product selectivity for FAME. alleviated.5,6,9−13 However, the reaction rate of transesterification catalyzed by solid catalysts appeared to be much slower;14 hence, higher reaction temperatures and pressure are necessary to deliver acceptable rates and conversion levels. The use of CaO14−20 and MgO14,15,20 as solid base catalysts for transesterification reactions has widely been reported in the literature. As opposed to homogeneous base catalysis, CaO catalysts in the process of vegetable-oil transesterifications are much easier to be removed from the reaction system, and thus, much less wastewater is produced in the purification of biodiesel. However, CaO leaching remains a problem.18,19 Ca2+ leaching leads to quick loss of catalyst activity and soap formation from its reaction with free fatty acids and may hamper its use in the subsequent batches or in continuous processes. Although basic sites of MgO are not as strong as that of CaO,14 it is easy to handle because of its rather stable properties under ambient conditions. Furthermore, being less prone to leach from the reaction system than CaO, the MgO catalytic

1. INTRODUCTION Transesterification of vegetable oils with methanol to produce fatty acid methyl esters (FAME) has been applied widely to produce biodiesel fuel because of its cost-effective technology as well as favorable physical properties of the fuel produced.1,2 The transesterification process includes three consecutive reversible reactions, namely conversion of triglyceride (TG) to diglyceride (DG), diglyceride (DG) to monoglyceride (MG), and monoglyceride (MG) to fatty acid esters and glycerol (FAME).3 Catalysts for the transesterification reaction are generally classified as homogeneous catalysts, biocatalysts, and heterogeneous catalysts.4−7 Among them, homogeneous catalysts and heterogeneous catalysts are used more widely than biocatalysts. In transesterification catalyzed by homogeneous basic catalysts, oil with high content of free fatty acid (FFA) would lead to soap and water formation, leading to product separation problem and oil losses. In addition, use of homogeneous catalysts, such as KOH and NaOH, would cause concerns in the areas of personal safety and environmental hazards.8,9 By use of heterogeneous catalysts, some drawbacks of homogeneous catalysts, such as separation of catalyst from the reaction mixture, formation of soap in the reaction, treatment of wastewater, and corrosion of process facilities, may be © XXXX American Chemical Society

Received: Revised: Accepted: Published: A

July 12, 2017 November 14, 2017 December 18, 2017 December 18, 2017 DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

by depositing MgO on the surface of Q50 and Q30 with average pore diameters of 40 and 30 nm, respectively. The granular silica were predried under vacuum at 200 °C for 2 h to remove water and then brought in contact with aqueous Mg(NO3)2 solution (4.8 N). A calculated amount of Mg(NO3)2·6H2O was added into the solution, so that the resulting solid would contain 5 wt % MgO. The water solvent was removed by evacuation at room temperature for 4 h, and the resulting solid was calcined in flowing air at 500 °C for 2 h. 2.3. Catalytic Performance Tests and Catalyst Rejuvenation. The catalytic performance was tested in a basket type reactor with an inside diameter of 6.3 and 10 cm height. The schematics of the experimental system used are shown in our previous paper.26 Feed per batch consists of 29 g of methanol and 125 g of soybean oil (i.e., methanol to oil molar ratio of 6:1). The tests were run with a MgO concentration of 0.2 wt % MgO/oil, which is equivalent to 0.25 g of MgO nanoparticles (noted as MgO) or 5.0 g of MgO/SiO2 in 125 g of soybean oil. The reactor was heated electrically and controlled by a PID temperature controller with a sensor at the outer wall of the reactor. The temperature difference between the outer reactor wall and the reaction temperature was about 10 °C. The reaction was carried out at 200 °C, 480 psig total pressure, and a stirring rate of 200 rpm for 5 h. Samples of 1 mL were taken at various time intervals during the reaction and analyzed with HPLC to determine soybean oil conversion and the formation of fatty acid methyl esters (FAME). After the first test run, reaction products were dumped from the reactor. Then, methanol and soybean mixtures were reloaded to the reactor, and the second test run was carried out with the same reaction conditions as those for the first test run. The same procedure was repeated four times for MgO/ SiO2(Q50) and two times for MgO/SiO2(Q30) in order to investigate catalyst deactivation. After 5-cycles of test runs, the used catalysts, MgO/SiO2(Q50)5, were first rejuvenated using an ethanol wash to remove the carbonaceous residue (gum) followed by 2 h drying in flowing dry air at 120 °C. The rejuvenated catalysts were referred to as MgO/Q50first‑re. After 3-cycles of test runs, the used catalysts, MgO/Q50first‑re-3, were rejuvenated again using an ethanol wash and air drying. In addition, to make up the leaching loss during test runs, 1 wt % MgO was loaded on the catalyst by impregnation. The solvent removal and calcination conditions for the rejuvenated catalysts were the same as those used in the preparation of the fresh catalysts. These rejuvenated catalysts with makeup MgO were noted as MgO/Q50second‑re. These catalysts were tested for three cycles to investigate the efficiency of gum removal and MgO reloading. The MgO morphology of the fresh MgO/SiO2(Q50) and used MgO/Q50second‑re-3 catalysts was characterized by synchrotron XRPD. 2.4. Analytical Methods. 2.4.1. HPLC Chromatographic Analysis. In this work, an analytical method using highperformance liquid chromatography (HPLC) developed by Carvalho et al.27 was used to determine total amounts of triglycerides (TG), diglycerides (DG), monoglycerides (MG), and fatty acid methyl esters (FAME) in the reaction system. The HPLC system was equipped with an LC10AT pump (Shimadzu, Tokyo, Japan), an SGE Enduro RP-C18 column (250 mm × 4.6 mm, particle size 5 μm), a column oven (SUPER CO-150, ENSHINE), and an ultraviolet detector (Shimadzu, Tokyo, Japan) set at 235 nm. Ten μL samples were injected into the HPLC system, and the flow-rate of 1 mL/min

activity was affected less in the presence of excess water in the vegetable oil, lending itself to be used in industrial biodiesel production;20 Di Serio et al., 2006 have reported that MgO can stand an excess of water up to 10,000 ppm in soybean oil transesterification.21 Initial studies in the use of pure MgO as a catalyst were not promising because of its low surface area, typically below 1 m2/ g. López et al.22 have reported only 18% conversion for the transesterification of triacetin with methanol at 60 °C after 8 h of reaction by using MgO with 0.45 m2/g surface area. To be commercially viable, preparation of high-surface-area MgO is necessary. Conventional high surface area MgO is normally prepared by calcination of precipitated Mg(OH)2. The total surface area of MgO powder prepared by these methods is higher than 100 m2/g. However, only the external surface of MgO granules is utilized in the catalytic reaction, because it is difficult for vegetable oils to diffuse into the interior of MgO pores. Moreover, being small in grain size, MgO is easy to be hydrated and leached out, leading to soap formation and loss of the catalysts. These drawbacks could be alleviated by depositing MgO on SiO2 to form pellet catalysts. In order to develop a quick and easy way of catalyst separation and rejuvenation after reaction, pellet SiO2supported MgO (MgO/SiO2 ) catalysts were made by impregnation. In order to investigate pore diffusion effects, catalysts composed of the same material with different pore diameters, namely MgO/SiO2(Q30) and MgO/SiO2(Q50) with pore diameters of 300 and 400 Å, respectively, were prepared and characterized by N2 adsorption−desorption isotherms. For characterizing active sites on MgO/SiO2, methanol adsorbed on the catalysts was characterized by in situ FT-IR spectroscopy. It has been reported that the structure of alkaline earth oxide catalysts and the corresponding catalytic activity are very sensitive to the method of catalyst preparation.15,18,23 Because of rather small MgO loading (5 wt %) on SiO2, it is difficult to characterize the detail crystalline phases contained in the catalysts by use of a laboratory diffrometer. X-rays from a synchrotron radiation source exhibit high brilliance, high collimation, tunable wavelengths over a wide range of energy, polarization, and short pulse lengths over a wide range of pulse periods.24,25 These advantages allow us to characterize the phase and morphology of MgO species on amorphous silica supports. The structure information, combined with catalytic performance tests, was used to explore the roles of reactionand rejuvenation-induced MgO morphology in catalysis.

2. EXPERIMENTAL SECTION 2.1. Materials. Commercial edible grade soybean oil (SBO, acid value of 0.04 mg KOH/g oil) was purchased from Unipresident Enterpriser Corp., Taiwan. Methanol (SK chemicals, HPLC grade) for transesterification reactions and MgO and Mg(NO3)2·6H2O (reagent grade) for catalyst preparation were purchased from Sigma-Aldrich Pty Ltd. Pellet catalyst supports SiO2 of 1.7 to 4.0 mm Q30 (100 m2/g) and Q50 (80 m2/g) were purchased from FUSI SILYSIA. Chemical grade isopropyl alcohol and n-hexanol (≥99%, liquid chromatography grade) for HPLC analyses were purchased from Macron Chemicals. Nitric acid and lanthanum oxide (La2O3) for Flame Atomic Absorption Spectrometry (AA) analyses were all purchased from Sigma-Aldrich Pty Ltd. 2.2. Catalyst Preparation. Pellet silica-supported MgO catalysts, MgO/SiO2(Q30) and MgO/SiO2(Q50), were prepared B

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

For comparison, methanol adsorbed on KBr was also measured and was noted as MeOH·KBr. 2.4.6. Synchrotron X-ray Powder Diffraction. X-ray powder diffraction (XRPD) was performed at the BL01C2 beamline of the National Synchrotron Radiation Research Center (NSRRC), with a wavelength of 0.7750 Å (16.0 keV) for the incidence X-ray beam. The storage ring of NSRRC was operated at an energy of 1.5 GeV with a typical current of 300 mA with a top-up injection mode. The synchrotron X-ray was produced from a 5.0 T superconducting wavelength shift magnet. Downstream from the collimating mirror, a double crystal Si (111) monochromator was employed for energy selection, which was followed by a refocusing toroidal mirror. The pellet samples were pulverized into powder and loaded into an XRPD cell for the measurement. During the X-ray exposure, the sample was kept fast spinning in order to make the orientations of powders more random. Two-dimensional diffraction patterns were recorded by a Mar345 imaging plate system, with a sample-to-detector distance of 300 mm. Diffraction angle 2θ was calibrated with silver behenate and Si powder (NBS640b) standards. XRPD (X-ray powder diffraction) profiles were integrated from selected fan-like areas of the symmetrical two-dimensional powder rings using the Fit2D program.28 2.4.3. Surface Area and Pore Size Distribution. Brunauer− Emmett−Teller (BET) surface area and pore size distribution of the catalyst samples were measured with the multipoint method from Micromeritics ASAP 2010. This was performed using nitrogen adsorption/desorption isotherms at liquid nitrogen temperature and relative pressures (P/P0) ranging from 0.04 to 0.4, where a linear relationship was maintained. 2.4.4. Flame Atomic Absorption Spectrometry (AA) Analysis. AA has been used for quantitative determination of MgO residue in biodiesel. The method of calculation was derived from a calibration curve drawn in accordance with absorbance and concentration of the standard solution. To have a concentration corresponding to the standard solution, stock solutions (1000 μg/mL) were diluted to 0.05−0.4 mg/L, while a lanthanum oxide solution was added to eliminate interference. A total of 250 mg of the biodiesel sample were weighed into a digestive tube (4.2 cm × 30 cm) accompanied by the addition of 25 mL of HNO3. The mixture was heated to 300 °C for 2 h. After cooling, 0.5 mL of lanthanum oxide solution was added and diluted to 25 mL. The wavelength of AA for Mg detection was 285.2 nm. Duplicate determinations were performed for each sample. All data presented are mean values and standard deviations of two independent experiments (n = 2).

with gradient elution of methanol (MeOH) and 2-propanol− hexane 5:4 (v/v) (PrHex) was used in sample analysis. The gradient program was 100% of MeOH at 0 min, 50% of MeOH and 50% of PrHex at 10 min, and then maintained isocratic elution for another 20 min. According to the equations below, the total conversion level (X) and mass percentages of FAME, MG, DG, or TG (Mi) were calculated, respectively Mi =

A i /Cf i A i /Cf ir × 100% = × 100% ∑ A i /Cf i ∑ A i /Cf ir

Conv. = 100% − M tri

where Ai (ATG, ADG, AMG, and AFAME) represents the respective areas of the peaks corresponding to i species (TG, DG, MG, and FAME), Cfi indicates their respective response factors, and Cfri indicates relative response factors to that of soybean oil (Cfi/CfTG). Standard solutions of FAME were prepared by diluting the desired weight of B100 (purified FAME, provided by Chinese Petroleum Corporation, Taiwan) in soybean oil. The response factors for FAME and TG were estimated based on the HPLC peak area in FAME concentration ranging from 0.1 to 0.9. Since there is no suitable standard for DG and MG, the response factors for these two were obtained from nonlinear regression of the following equation 0 Conv. × A TG =

ADG A A + MG + FAME r r r Cf DG Cf MG Cf FAME

where CfDG and CfMG, the regression parameters, were estimated from 10 chromatograms with conversion ranging from 0.2 to 0.95. The resulting Cfri values of FAME, MG, DG, and TG were 2.2, 2.1, 1.3, and 1.0, respectively. 2.4.2. FT-IR Spectroscopy. Diffuse reflectance infrared Fourier transform spectra (DRIFT) of MgO, MgO/SiO2(Q50), and methanol adsorbed on MgO/SiO2(Q50) were recorded with a Shimadzu FT-IR, IR Prestige-21 spectrophotometer, equipped with a DTGS detector having a spectral resolution of 2 cm−1. The pellet samples were pulverized into powder, diluted with KBr at a ratio of 1:5, and then loaded into an IR cell. The cell was purged with dry N2 at room temperature for 1 h, and IR spectra were recorded for MgO, silica support (Q50), and MgO/SiO2(Q50) samples. For the measurement of dehydrated samples, the cell was connected to a vacuum system and evacuated to obtain a vacuum better than 0.01 Torr. The temperature was increased at about 10 °C/min from room temperature to 120 °C and maintained for 1 h. The purpose of this in situ treatment is to remove water adsorbed on the catalyst surface, and this treatment process is noted as a dehydration process. After cooling to 40 °C, IR spectra were again recorded. The interactions between methanol and MgO/SiO2(Q50) were also characterized by FT-IR. Pulverized samples were loaded into a DRIFT cell, purged with dry N2 at room temperature for 1 h, and then heated under vacuum to 120 °C. After taking the background IR spectra, methanol containing about 7000 ppm water was introduced into the IR cell and maintained for about 20 min for equilibrium. Subsequently, the cell was evacuated to a pressure of approximately 0.1 Torr to remove free and physical adsorbed methanol, and IR spectra were, then, recorded and denoted as MeOH·MgO/SiO2(Q50).

3. RESULTS AND DISCUSSION 3.1. Catalytic Properties of MgO, MgO/SiO2(Q50), and MgO/SiO2(Q30). For solid catalysts tested in a batch system, the deactivation rate is usually estimated by repeating the reaction cycle several times and measuring the catalytic activity in each cycle. Since it is difficult to separate used powder nano-MgO from reaction products, reuse of the catalysts was normally not attempted; the purpose of the second cycle test for the nanoMgO in this study is for comparison. In contrast, for the reaction catalyzed by pellet MgO/SiO2, the reaction products can be dumped from the reactor and separated from catalysts in the basket easily, hence, a new reaction cycle can be started by simply charging reactants to the reactor. Figure 1 shows the soybean oil (TG) conversions for nanoMgO and successive cycles for MgO/SiO2(Q50) (noted as C

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

Figure 1. Triglyceride conversion catalyzed by MgO nanoparticles-1 (first cycle test of MgO) (purple ★), MgO/Q50-1 (first cycle test of MgO/SiO2(Q50)) (black ■), MgO/Q50-2 (blue ●), MgO/Q50-3 (olive ▲), MgO/Q50-4, (orange ◆), MgO/Q50-5 (green ▼), MgO/ Q30-1 (first cycle test of MgO/SiO2(Q30)) (orange □), MgO/Q30-2 (blue ο), and MgO/Q30-3 (wine △).

MgO/Q50). As shown in this figure, the TG conversion rate for the second cycle of MgO/SiO2(Q50) (noted as MgO/Q50-2) is higher than that of MgO. Moreover, MgO/SiO2(Q50) exhibited a rather stable activity, and the TG conversion level only dropped from 97.6% in the second cycle to 91.5% in the fifth cycle (Figure 1, Table S1 of the Supporting Information). The results indicated that SiO2-supported MgO catalysts are comparable to nano-MgO in activity and can be reused for a few reaction cycles. Notwithstanding catalyst deactivation caused by MgO leaching and other factors, such as carbonaceous deposition on the catalyst surface, the second cycle run for both MgO/ SiO2(Q50) and MgO/SiO2(Q30) presented a higher reaction rate and FAME yield than those of the first cycle run (Figure 1, Table S1). The lower TG conversion and FAME yield for the first cycle run appear to be due to the incomplete mixing of methanol with TG and the incomplete activation of MgO on SiO2 for a certain time period at the start of the run. This period is referred to as the induction period. 3.2. Active Sites of MgO/SiO2(Q50) Characterized by XRPD and FT-IR. XRPD patterns of MgO and MgO/SiO2(Q50) were shown in Figures 2a and 2b, respectively. For MgO/ SiO2(Q50), both broad backgrounds produced by amorphous SiO2 and small sharp Bragg peaks by crystalline phases of MgO and MgCO3 were present. For MgO, only Bragg peaks were observed. The crystalline phase of both samples was identified by matching the XRPD pattern with reference patterns of pure substances. The results indicated that the MgO sample is a mixture composed of brucite [Mg(OH)2] and a small amount of hydromagnesite [Mg5(CO3)4(OH)24H2O]. For MgO/ SiO2(Q50), the XRPD pattern indicated that periclase [MgO] and a small amount of nesquehonite [MgCO3(H2O)3] phases were formed on amorphous SiO2 support. Nesquehonite could be formed from the reaction of aqueous Mg(NO3)2 solution with CO2 dissolved in the solution. FT-IR spectra of hydrated and dehydrated MgO nanoparticles are shown in Figure 3. The sharp and intense peak at 3698 cm−1 is attributed to the hydroxyl groups in the crystal structure of brucite Mg(OH)2.29 Since different types of

Figure 2. X-ray powder diffraction pattern of (a) MgO nanoparticles and (b) MgO/SiO2(Q50).

hydroxyls on the MgO surface have been detected, a schematic diagram (Figure S1) inferred from the paper reported by Chizallet et al.30 was drawn to differentiate these hydroxyls of different IR absorption bands. The hydroxyls on brucite Mg(OH)2 could be classified as isolated types (1a, Figure S1; LC, low coordination with L = 3 for corners, 4 for edges, and 5 for terraces). The broad band at 3400 cm−1 shown in Figure 3a was attributed to the adsorbed water molecules. The shift of the peak from 3450 cm−1, the characteristic peak of water molecules, to 3400 cm−1 suggests a layer of water was adsorbed, and these water molecules are stabilized by mutual hydrogen bonding.31,32 A peak at 1635 cm−1 was attributed to the bending vibration of water molecules. After dehydration, the characteristic peaks (3400 and 1635 cm−1) characterizing the adsorbed water disappeared (Figures 3a and 3b). The peaks at 1420 and 1485 cm−1 are assigned as the characteristic peaks of the carbonate symmetric stretching vibration of [Mg5(CO3)4(OH)24H2O].33 FT-IR spectra of hydrated and dehydrated Q50 are shown in Figures 3c and 3d, respectively. For the dehydrated Q50 (Figure 3d), the peak at 3740 cm−1 was attributed to terminal hydroxyl groups bonded to Si (≡SiOH), while the broad peak appearing at 3600 cm−1 could be assigned to residue water molecules adsorbed on nonhydroxylic sites of amorphous silica.31 D

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

Figure 3. FT-IR spectra of (a) MgO, (b) dehydrated MgO, (c) Q50, (d) dehydrated Q50, (e) MgO/SiO2(Q50), and (f) dehydrated MgO/SiO2(Q50).

of the methanol molecule. The results are consistent with a blue shift of the CO stretching vibration peak (ν(C−O)) from 1010 to a higher wavenumber due to shortening the C−O bond of MgO−Hδ+--δ−OCH3. The results suggest that methanol can be chemisorbed on MgO (periclase)/SiO2(Q50) at room temperature (step 1, Figure S2).4,34 In order to explore the surface species formed on MgO/ SiO2(Q50) during the reaction, FT-IR was used to characterize the used catalysts of the first cycle test, and the results are presented in Figure 5(a). Since the product of the transesterification process, FAMEs, is chemically similar to soybean oil, the species deposited on the used catalysts were identified by IR characteristic peaks. The presence of a peak at 1360 cm−1 can be attributed to the glycerol group O−CH2 for MG, DG, and TG.35 Peaks 1070 and 1160 cm−1 are assigned to axial stretching −O−CH2−C for soybean oil. The peaks at 1446 and 1196 cm−1 correspond to the asymmetric stretching of (CO)− O−CH3 and C−O stretching, respectively, of FAMEs. The peak at 1745 cm−1 can be attributed to the stretching of CO for TG, DG, MG, and FAME.35 After the removal of TG, DG, MG, and gum from the surface by an ethanol wash followed by evacuation, the spectra were shown in Figure 5(b). A peak appearing at around 1085 cm−1 was assigned to the CO stretching vibration of methoxide anion (OCH3−), while another peak at around 1400 cm−1 was attributed to the asymmetric stretching of methoxide bound MgO (CH3−Oδ−--δ+MgO).36,37 The shoulder at 3550 cm−1 could be assigned as multicoordinated hydrogen bond donor hydroxyls in (3a, Figure S1). The bridging hydrogen could be originated from the hydrogen abstracted from methanol. The small peak appearing at 3775 cm−1 may be assigned as the characteristic peak of the multicoordinated

FT-IR spectra of hydrated and dehydrated MgO/SiO2(Q50) are shown in Figures 3e and 3f, respectively. The disappearance of 3740 cm−1 suggested that terminal silanol of SiO2 was covered by MgO. Consistent with XRPD results, there are no sharp OH stretching vibration (νOH) peaks at 3698 cm−1, suggesting no brucite [Mg(OH)2] crystals are formed on SiO2. Dehydration caused a disappearance of the water absorption band at 3400 and 1635 cm−1, whereas the appearance of a broad absorption band at about 3630 cm−1 for the dehydrated MgO/SiO2(Q50) (Figure 3f) suggests that trace water molecules could be dissociated on the defects of periclase MgO.29 The peaks at 1410 and 1452 cm−1 are assigned as stretching vibrations of carbonate in nesquehonite [MgCO3(H2O)3].33 FT-IR spectrum for physically adsorbed methanol on KBr (MeOH·KBr) is shown in Figure 4a. Without perturbation by adsorbent, OH stretching (νOH) and CO stretching vibration (ν(C−O)) frequency are observed at 3340 and 1010 cm−1, respectively. When a 7000 ppm water-containing methanol was introduced to MgO/SiO2(Q50), concomitant with the disappearance of a peak at 3340 cm−1, a new peak at 3680 cm−1 and a small shoulder at 3280 cm−1 were observed (Figure 4b). The results suggested the adsorption of methanol on MgO (step 1, Figure S2); note that a small amount of methanol could also interact with the hydroxyls on the defect of periclase MgO. The peak at 3680 cm−1 could be assigned as νOH of newly formed MgO−Hδ+ (2a, Figure S1), and the small shoulder at 3280 cm−1 could be assigned as νOH of the Hδ+-(OCH3)δ− moiety (2b, Figure S1) in MgO−Hδ+--(OCH3)δ−, and/or νOH of the methanol bond to MgO, CH3−OH--MgO (2c, Figure S1), and/or νOH of the methanol bond to MgOH, MgOH--H--OCH3 (2d, Figure S1). The red shift from 3340 to 3280 cm−1 could be associated with lengthening the OH bond E

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

methoxide anion with oxygen of MgO (C−H--O). The small peak at 1730 cm−1 (CO ester) indicates an ester linkage in the gum structure.38 This peak with −CH2 and −CH stretching vibrations at 2800, 2920, and 3030 cm−1 could be the characteristic peaks of the unremoved trace carbonaceous material (gum) on the MgO. The observation of the methoxide anion in FT-IR spectra suggests that methanol molecules were dissociatively adsorbed on periclase MgO (steps 1 and 2, Figure S2).4,34 Based on the reaction mechanism of base-catalyzed transesterification reported in the literature,34,39,40 the methoxide (CH3O−) ions were expected to attack a carbonyl carbon of the TG molecule and form a DG ion and FAME (step 3, Figure S2). The DG anion then extracts a proton from protonated MgO, MgO−H+, to produce a DG molecule and turns MgO− H+ into the initial form, MgO. In the next two sequential reactions, catalytic reaction cycles for MG and FAME formation are the same as that for the reaction of TG with methanol yielding DG. Besides heterogeneous catalysis, Mg2+ and OH− ion pairs formed from the dissolution of MgO into liquids also catalyze the transesterification reactions homogeneously. Normally, biodiesel production processes catalyzed homogeneously by the Brønsted base are relatively fast.22,39,41 Hence, MgO leaching enhances the reaction rate. However, it causes a loss of active sites of the solid catalysts and formation of soap, leading to a catalyst deactivation and difficulty in the separation product from catalysts. As shown in Table S2, the amount of Mg2+ leached from MgO nanoparticles is about 3 times that from MgO/SiO2(Q50). These two drawbacks limit the complete reutilization of MgO nanoparticles. 3.3. Factors Affecting Catalyst Activity and Reaction Kinetics. 3.3.1. Pore Structure and Pore Diameter. Particle size of MgO, adsorption isotherms, pore size distribution, and average pore diameter for the 3 catalyst samples are shown in Figure S3, Figure 6, Figure 7, and Table S4, respectively. According to an empirical classification of hysteresis loops given by IUPAC, IV type isotherm is assigned for MgO. Shown in Figure 6, the N2 isotherm for MgO/SiO2 is concave to the x (P/P0) axis in the range of P/P0 = 0.0 to 0.15; after that, the isotherm is convex to the x axis. Since the N2 isotherm is not

Figure 4. FT-IR spectra of (a) MeOH·KBr and (b) MeOHvMgO/ SiO2(Q50).

Figure 5. FT-IR spectra characterizing surface species on (a) used MgO/Q50-1 (first cycle test of MgO/SiO2(Q50)) and (b) used MgO/ Q50-1 treated by an ethanol wash followed by evacuation.

hydrogen bond acceptor OH group (3b, Figure S1). Peaks appearing at 2860 and 2935 cm−1 could be derived from −CH2 stretching vibrations of the methoxyl group.36 The peak appearing at 3130 cm−1 could be viewed as a weak hydrogen bond due to the interactions between hydrogen atoms of

Figure 6. Nitrogen adsorption isotherm at 77 K for MgO/SiO2(Q50) (red), MgO/SiO2(Q30) (black), and MgO (blue). F

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

conversion are less than those for MgO/SiO2(Q30), but the diffusion resistance is also lower to favor the reaction rate. The higher TG conversion for MgO/SiO2(Q50) (Figures 1 and 8 and Table S1) indicates that the transesterification catalyzed by MgO/SiO2 is a process with strong mass transfer resistance.

Figure 7. Pore size distributions for MgO/SiO2(Q50) (red), MgO/ SiO2(Q30) (black), and MgO (blue).

convex to the x axis over the entire range, IV type is assigned for MgO/SiO2.42,43 The shape of the hysteresis loop is related to the shape of the pores. With the loop of upward curvature at relative pressure (P/P0) between 0.8 and 0.98 (H1 hysteresis loop), the pore of MgO/SiO2 is of cylindrical type, while with a slow jump in the H4 hysteresis loop at the 0.45−0.95 P/P0 region, MgO is a mesoporous material with slit type porosity.43 In comparison with MgO/SiO2(Q50), MgO comprise much smaller granule size and pore diameter and much larger surface area (Figure 7, Table S4). Without considering the limitation of pore diffusion resistance, MgO is expected to be higher in catalytic activity. However, the experimental data show that the TG conversion rate for MgO was slightly lower than that for MgO/Q50-2 (Figure 1, Table S1), suggesting that the pore structure has a significant influence on the catalytic properties and activity in particular. As shown in Figure 6, MgO/SiO2 adsorption at P/P0 approaches zero, indicating no significant micropores present in these samples. In contrast, the microspores in MgO are about 0.025 cm3/g. These micropores are too small to catalyze TG transesterification reactions, because of pore diffusion limitation. For diffusion in a porous catalyst with mesopores, the effective diffusivity is altered by path length, pore cross section areas, and complexity of internal pore structures.44 In comparison with MgO/SiO2(Q50), the diffusion path for reactants to reach active sites on MgO nanoparticles is much shorter due to its small particle sizes, but the pore diameter (3 nm as opposed to 40 nm for MgO/SiO2(Q50), Figure 7) is much smaller, and the slit type pore structure greatly hinders diffusion of TG. Hence, despite more MgO being leached out to increase homogeneous reactions, and a larger surface area to enhance heterogeneous reactions, MgO nanoparticles are slightly lower in activity than MgO/SiO2(Q50) for TG conversion. The importance of the diffusion resistance and mass transfer limitation in this reaction becomes apparent when the performances of catalysts with supports of different pore sizes are compared. It can be seen in Figure 6 and Figure 7 that both MgO/SiO2(Q50) and MgO/SiO2(Q30) exhibit similar pore structure but different pore sizes. The larger pore diameter but smaller surface area for MgO/SiO2(Q50) (Table S4 and Figure 7) would suggest that active sites available for TG

Figure 8. Triglyceride conversion catalyzed by MgO nanoparticles-1 (first cycle test of MgO) (red •), MgO nanoparticles-2 (red ο), MgO/ Q50-2 (black □), and MgO/Q30-2 (blue Δ).

3.3.2. Reaction Kinetics and Rationalization of the Induction Period. The overall transesterification reaction can be written as TG + 3CH3OH ↔ 3FAME + G (Glycerin)

For the reaction system catalyzed by heterogeneous catalysts, the reaction kinetics could be described using a pseudo-firstorder rate expression as a function of the TG conversion.45−47 However, the rate expression could be further complicated by reactions involved in both homogeneous and heterogeneous, activation of active sites, and different levels of mass transfer resistance. Experimental results indicated that pseudo-firstorder kinetics may not be adequate for our experimental data, specifically, for the first cycle run. Test reaction and XRPD and FT-IR characterization results suggested that the mechanism for activation of MgO/SiO2 could involve adsorption followed with dissociation of methanol on periclase MgO leading to formation of a hydroxyl group and methoxide ions on MgO/SiO2. Thus, the actives sites consist of the periclase MgO and activated MgO, such as surface Mg(OH)2 and methoxide bound MgO (CH3Oδ−--δ+MgO). It appears that the rate processes for forming these active species lead to the induction period in the first reaction cycle of MgO/SiO2, which is rationalized as follows: FAMEs play a role like surfactant leading to an increase of solubility of oil in the methanol-FAME phase and that of methanol in the oil-FAME phase as well. The increase of solubility decreases both external and internal mass transfer resistance.48 In the first cycle run, upon adsorbing MeOH and trace water (7000 ppm), periclase MgO is converted gradually into protons donated MgO, methoxide ions bound MgO CH3Oδ−--δ+MgO, and Mg(OH)2, which are active catalytic species. In the presence of initially formed FAMEs, the mobility of reactants and surface catalytic species (OH− and CH3O−) increases, and G

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

period is not obvious (characteristic time is approaching zero). The experimental data can thus be fitted by a simple first-order model:

more vacant MgO sites become available for adsorption of MeOH and H2O to make more active catalytic species, leading to increased activities (intrinsic rate constant, ks). This chain of consecutive processes, including adsorption, active species formation, moving of active species for creating more active species, and reactions to convert TG, takes time, resulting in the induction period observed. 3.3.3. Reaction Model. To describe the reaction promoted by these active catalytic species and the presence of FAMEs, rate constant k was modified and formulated to k = ki + Δk*(1 − exp(−t/τ)), where ki is an initial rate constant contributed by periclase MgO, t is time elapsed, Δk is the difference between ki and kf (kf − ki), kf is the rate constant when ks (intrinsic rate constant) and kg (external mass transfer rate constant) will not be varied with time, and τ is the characteristic time for the increase rate constant from ki to kf due to mass-transfer enhancement and the active-sites activation. Hence, an empirical power-law model was reformulated as

dx /dt = k f × (1 − x) = ηG × ks × (1 − x)

The fitting results (line) and experimental data (symbol) are shown in Figures 1, 8, and 10, and the estimated kinetic parameters were summarized in Tables S3b, S5, and S6. It could be due to the fact that the MgO sites on SiO2 support have not been fully activated after the first cycle, ki + Δk of the first cycle is less than kf of the second cycle for both MgO/SiO2(Q50) and MgO/SiO2(Q30). It could also due to the fact that pores are full of active species, OH− and OCH3− after the first cycle of the run. These species promote the reaction rate, but that was not taken into account in the model. For MgO nanoparticles, the brucite [Mg(OH)2] phase has already been formed, and most of the MgO bound OCH3− occur in the external surface area, hence no significant induction period was observed. In order to take the activity loss due to MgO leaching into account, the rate constant is normalized by dividing overall rate constants by actual MgO in the reaction system, k n (normalized-rate-constant) = kf/w, where w is the actual weight of MgO in the reaction system. The calculated results were summarized in Table S3b. 3.3.4. Importance of the Reaction Catalyzed by Leached MgO. The leached out MgO is a good catalyst in the homogeneous phase and contributes to the overall reaction rate. As to the evaluation of the homogeneous contribution, a rigorous kinetic analysis was not attempted because of the complexity of the reaction system. Instead, a simple phenomenological equation

(rTG)obs = − 1 × (k i + Δk × (1 − exp( − t /τ ))) × C TG n (1)

By introducing CTG = CTG0 × (1−x) to the above equation, the equation was reformulated as dx

n−1 × {k i + Δk × [1 − exp( t τ )]} × (1 − x)n dt = (C TG0) (2)

In the equation, CTG0 is the initial concentration of TG, and x is conversion. The fitting results show that the estimated reaction order is in the range of 0.85 to 0.90, which is essentially 1, and the characteristic time approaches zero after the first cycle (Table S1a,b). For an nth-order reaction, the observed order will be (n+1)/ 2 due to diffusional falsification; therefore, the observed order is equal to the true order for a first-order reaction. Since our experimental data can be fitted well by eq 2 with first-order kinetics (R2 > 0.99), the intrinsic rate of eq 2 can thus be further simplified by first-order kinetics. For an intrinsic first-order reaction with the combination of the external and internal mass transfer resistance, the rate expression can be formulated as (rTG)obs = − ηG × ks × C TG ,

(4)

k f = klfs × wlfs + k fs × ws

(5)

was proposed to estimate the reaction rate contributed by MgO leached from MgO/SiO2(Q50) (or MgO nanoparticles), semiquantitatively, where wlfs and ws = time average weight of MgO leached from MgO/SiO2(Q50) (or MgO nanoparticles) and that of MgO containing solid MgO/SiO2(Q50), respectively

1/ηG = 1/η + (R /3) × (ks/kg)

wlfs + ws = w

(3)

klfs = the rate constant for the reaction catalyzed by MgO leached from MgO nanoparticles or MgO/SiO2(Q50), and kfs = the rate constant for the reaction catalyzed by solid MgO nanoparticles or solid MgO/SiO2(Q50). A hot filtration method49,50 was used to estimate the contribution by the leached MgO. After the run with the TG transesterification reaction in a slurry reactor for about 90 min, used MgO nanoparticles were filtrated out by centrifugation at 15,000 rpm followed with hot filtration. The filtrates were further reacted catalytically by the leached MgO for 300 min. The rate constant for the leached MgO species was determined from time-dependent TG conversion (Figure 9) based on firstorder reaction kinetics. By normalizing leached MgO species of 0.017 g, the estimated klfs for the leached MgO is 0.138 min−1 g−1. A similar procedure was used to estimate klfs for the MgO leached from MgO/SiO2(Q50) except that used pellet catalysts can be removed from the reaction system simply by removing the catalyst-containing basket. The testing results were shown in Figure 9. The leached MgO is 0.0068 g, and klfs for the MgO

where ηG and η are effectiveness factors for global and pore diffusion, respectively, ks and kg are intrinsic and external mass transfer rate constants, R is the radius of catalyst pellet, and (rTG)obs is the observed reaction rate for TG. External mass transfer resistance is much larger than internal resistance, and eq 3 was simplified as (rT)obs = (3/R) × kg × CTG; observed rate constants will not be varied with the activity change of active sites, whereas external mass transfer limitations will never exist unless internal diffusion limitations are also present.44 Internal mass transfer (pore diffusion) is much larger than external resistance, and eq 3 was simplified as (rT)obs = η × ks × CT ≈ 1/ϕ × ks × CT, where ϕ (Thiele modulus) = R/3 × ks/De , and De is effective diffusivity.44 In the first cycle, kinetic parameters in eq 3 are varied with time, and we are unable to estimate these time-varying parameters. For simplification, experimental data were fitted by 2 with n = 1, and the estimated ki, Δk, and τ were shown in Tables S3a and S6. After the first cycle test, the induction H

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

plugging could happen, leading to a significant decrease in effective diffusivity, De. Gum formed on MgO/SiO2(Q50) was extracted from the used catalyst by use of THF (tetrahydrofuran), and the molecule weight of the extracted gum was determined by GPC (Gel Permeation Chromatography equipped with HR 0.5, 3, 4, and 5 column sets, a Water 410 Differential Reframoter, and a Water 486 UV−vis detector). The results show that gum grows with no. of reaction cycles gradually from cycle 1 to 4 and then grows quickly at no. 5; the molecule weights of the extracted material were 1300, 2100, 5000 g/mol, for second, fourth, and fifth cycles, respectively. 3.4.2. Leaching Rate. The MgO leaching rate varies with the granular size, surface area, and pore structure of the catalysts. The data in Table S2 show the leaching rate decreases in the following order: MgO nanoparticles > MgO/SiO2(Q50) > MgO/ SiO2(Q30). For SiO2-supported MgO, MgO/SiO2(Q50) and MgO/SiO2(Q30), both samples have the same granular size, whereas MgO/SiO2(Q50) has larger pores but smaller surface area. The lower leaching rate for MgO/SiO2(Q30), as opposed to MgO/SiO2(Q50), suggests that instead of being associated with OH−, Mg2+ cations may be bound to large molecules such as TG, DG, and MG via the acceptance of electron pairs donated from the nucleophilic groups of these molecules, hence, the leaching rate is limited by the internal mass transfer resistance. The leaching rate of MgO nanoparticles is higher than the other two samples because of its smaller granular size and larger external surface. The rather high leaching rate for MgO nanoparticles results in a quick catalyst deactivation; as shown in Figure 8 and Table S3b, the catalytic activity of MgO nanoparticles lost about 45% after the first reaction cycle. For MgO/SiO2(Q50), upon washing to remove the carbonaceous residue, the porosity of the used second run became about 10% higher than that of the fresh catalyst (Figure S3). Apparently, the leaching and dissolution of MgO in the reaction system enlarged the average pore diameter and created some micro- and mesopores of about 15 nm in diameter. Hence, MgO leaching causes a loss of active sites and, on the other hand, could enlarge the catalyst pore, leading to an increase in effective diffusivity, De. Leaching and dissolution of MgO in the reaction system enlarged the average pore diameter, which was expected to increase Mg2+ diffusivity and could lead to an increase in Mg2+ leaching. However, as shown in Table S2, Mg2+ leaching for MgO/SiO2(Q50) smoothly declined from cycle 1 of 0.02 g of MgO leached to cycle 4 of 0.01 g and then sharply declined to 0.002 g in cycle 5. For cycle 5, the quick decrease of MgO leaching could be due to pore-mouth plugging. However, for cycles 1 to 4, the leaching results suggest other factors besides the pore-enlarging also influence Mg2+ leaching. One possible reason in influencing Mg2+ leaching is successive lay down of carbonaceous residues, which cover up the MgO leaching sites. The lay-down carbonaceous species were observed in FT-IR characterizing the washed used catalysts. The other likely cause is the change of MgO morphology from three-dimensional in shape to raft. The increase of SiO2−MgO interactions due to the morphology change might reduce the leaching rate. 3.4.3. Catalyst Deactivation. The contribution of the leached MgO to the rate constant, kf, is about 11, 8, 6, and 1%, respectively, for MgO/SiO2(Q50) at the second to fifth cycle runs. Based on these results, the normalized rate constant, kn, contributed from heterogeneous catalysis is calculated to be 47.3, 47.1, 48.4, and 40.1 min−1 g−1, respectively, for the second

Figure 9. Hot filtration tests: triglyceride conversion catalyzed by (a) MgO leached from MgO nanoparticles (black ■) and (b) MgO leached from MgO/SiO2(Q50) (red •).

leached from MgO/SiO2(Q50) is 0.152 min−1g−1, which is close to that estimated from MgO leached from MgO nanoparticles. The time average weight for the first run of MgO is 0.0315 g. For MgO nanoparticles, the rate constant, kf, contributed from the catalysis of MgO leaching is about 41%, which is about 3 times that of MgO/SiO2(Q50). The esterification of TG with ethanol is a three step consecutive reaction with three reaction rate constants. These rate constants can be very different for each catalyst. Besides the total conversion level, the final and intermediate products from each catalyst could be different due to the difference in these rates constant. However, homogeneous catalysts normally showed greater performance toward transesterification to obtain FAME.51 MG conversion could be more favorable in homogeneous reactions. Experimental results showed that a reaction catalyzed by the leached MgO from MgO nanoparticles is about 3 times that from MgO/SiO2(Q50). Hence, higher MG while lower FAMEs yield for MgO/SiO2, as opposed to MgO nanoparticles, may be mainly due to less MgO leached. The influence of MgO morphology on FAME yield cannot be ruled out. As shown in Figure 2(b), significant preferred orientations have been observed in the XRPD pattern of MgO/ SiO2(Q50). In heterogeneous reactions, the determining factor is the nature of the surface, and the conversion of MG is favored by low coordination step and corner MgO sites. In comparison to MgO nanoparticles, low coordination step and corner MgO sites associated with heterogeneous reactions are lower for MgO/SiO2(Q50), which could also be unfavorable for MG conversion. 3.4. Gum Formation, MgO Leaching, and Catalyst Deactivation. The deactivation of MgO/SiO2(Q50) for transesterification is mainly caused by deposition of carbonaceousresidue on active sites, plugging of pore mouth, and leaching loss of MgO. 3.4.1. Gum Formation. The high molecule weight carbonaceous-residue, gum, could be initiated from peroxy radicals that formed from inserting oxygen to unsaturated fatty acid moieties of glycerides during storage.3 Gum of low molecular weight could be strongly adsorbed on the active sites, resulting in a decrease in intrinsic rate constant, ks. As high molecular weight gum was formed, besides covering the active sites, pore mouth I

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

performance for the first rejuvenated catalysts deteriorated rapidly; the TG conversion and FAME selectivity are even lower than those of the deactivated one, MgO/Q50-5 (Table S5). The results indicated that the deactivated catalyst rejuvenated by ethanol washing alone is insufficient, consistent with that reported by Ngamcharussrivichai et al.52 3.5.2. Rejuvenation by Combined Ethanol Washing and MgO Reloading. The used MgO/Q50first‑re-3 was then rejuvenated by ethanol washing, adding 1 wt % of MgO by impregnation and calcining in air at 500 °C to obtain MgO/ Q50second‑re. The TG conversions in the second and third cycle runs for MgO/Q50second‑re were 97.5% and 96.2%, respectively, and the FAME yields were 40.3% and 38.2 (Figure 10, Table S6), respectively. Similar to the fresh catalysts prepared by air calcination at 500 °C, the conversion and FAME selectivity for the first cycle run of MgO/Q50second‑re are lower than those for the second run. The catalytic properties of the fresh and rejuvenated catalysts are strikingly different. In the rejuvenation process, the activity for TG conversion was recovered, but the selectivity for FAME was not. For the rejuvenated catalyst, MgO/Q50second‑re, the rate constant for TG conversion is slightly decreased, whereas FAME yield was significantly lower than the fresh catalyst MgO/SiO2(Q50) (Tables S1, S3, and S6). The difference in the catalytic properties for the fresh and the rejuvenated catalysts could be caused by the change in MgO morphology. 3.5.3. Effects of MgO Morphology on Catalysis. With an incidence synchrotron X-ray beam of 0.7750 Å wavelength, the characteristic peak at 18.3, 21.2, 30.1, 35.6, and 37.1° for fresh MgO/SiO2(Q50) (Figure 11a) was identified to be the peak of periclase MgO (111), (200), (220), (311), and (222) reflection planes, respectively. The orientation of crystalline grains can be manifested in the intensities of the diffraction pattern.53,54 In comparison with peak intensities of standard periclase MgO shown by the gray line (Figure 11a), except for the characteristic peak of (200), all peak intensities are lower than those of the fresh MgO/ SiO2(Q50). The results suggest that not all silica bound periclaseMgO grains on the cylindrical pore walls are randomly oriented; some crystal grains of terrace-like structure are also formed on the surface. After the test reactions of the rejuvenated catalysts, the intensities of the peak at 18.3, 30.1, 35.5, and 37.1° for MgO/Q50second‑re were further decreased and hard to observe. These results suggest that MgO crystal grains were destructed and recrystallized in the reaction and rejuvenation processes, leading to a change in the MgO morphology from three-dimensional in shape to a twodimensional terrace-like structure. It has been reported that preparation routes for alkalineearth-oxide catalysts significantly affect surface morphology of alkaline-earth-oxides and the corresponding catalytic activity of the catalysts.18,30,55,56 Computational modeling of MgO surfaces suggests that defect centers are energetically favored over (110) and (111) planes.57 A correlation between the activity of MgO nanocrystals and their basicity further reveals that (111) and (110) surface planes are more basic than that of the (100) plane and exhibit higher transesterification activities.15,56 Because of selection rules, no characteristic peak for (110) and (100) planes was observed for XRPD characterizing periclase MgO (Supporting Information), whereas the appearance of the characteristic peak of (200) and (220) planes indicates the presence of (100) and (110) surface planes

to fifth cycles of MgO/SiO2(Q50). In the normalized rate constant, kn, the active-sites loss due to leaching is taken into account by dividing overall rate constants by the actual weight of MgO. It is interesting to note that kn contributed from heterogeneous catalysis was not varied significantly from the second to fourth cycles, which could be because of the compensation effects, that is, decreasing ks due to strong adsorption of gum on active sites (Figure 5), increasing De due to pore enlarging and creation of new pores caused by MgO leaching (Figure S4), and possibly decreasing De due to pore shrinking caused by gum formation. In the fifth cycle run, the effect of MgO leaching is insignificant (Table S2), while the formation of long-chain gum could plug the pore mouth that hinders the diffusion of TG from reaching the active sites; as shown in Figure S4, there are only 15% mesopores left after the fifth reaction. The decrease of effective diffusivity, De, also leads to a quick drop in kn. 3.5. Catalyst Rejuvenation. Catalyst reusability, one of the most important features for commercialization, is the key value of supported catalysts. Apart from the cover of active sites by gum deposition, Mg 2+ leaching and change of MgO morphology could also result in a catalyst deactivation, in both activity and selectivity, and thus, impacting their reusability. As shown in Tables S3b and S1, for MgO/ SiO2(Q50), normalized rate constant and FAME yield decrease about 24 and 17%, respectively, from run 2 to run 5. Two different rejuvenation methods, namely, ethanol washing, MgO/Q50first‑re, and combined ethanol washing and MgO reloading, MgO/Q50second‑re, were used to demonstrate the efficacy of the rejuvenation process. 3.5.1. Rejuvenation by Ethanol Washing. The comparison of conversion and yield patterns for the fifth cycle of MgO/ SiO2(Q50) (MgO/Q50-5) and each cycle run for the rejuvenated catalysts are shown in Figure 10, Table S5, and Table S6. At first run, TG conversion and FAME yield for MgO/Q50first‑re were 95.3% and 41.6%, respectively, which is higher than those of the deactivated catalysts, MgO/Q50-5, but lower than those of MgO/Q50-2. After the first cycle run, the catalytic

Figure 10. Efficacy of catalyst rejuvenation by ethanol washing; triglyceride conversion catalyzed by MgO/Q50-5 (blue solid line ■), MgO/Q50first‑re-1 (red dot-dashed line ο), MgO/Q50first‑re-2 (red dotdashed line △), (ο) MgO/Q50first‑re-3 (red dot-dashed line △), MgO/ Q50second‑re-1 (black dashed line •), MgO/Q50second‑re-2 (black dashed line ▲), and MgO/Q50second‑re-3 (black dashed line ⧫). J

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research

loading until MgO is overloaded. In this study, MgO loading for both fresh and rejuvenated catalysts was kept at 5 wt %, because higher MgO loading may plug mesopores resulting in a decrease in surface area and catalyst porosity, leading to a decrease in both the intrinsic rate constant and effective diffusivity. Significant mesopore plugging was observed when MgO loading was increased from 5 to 7 wt % (Figure S4). Moreover, BET analysis results indicated that the surface area of SiO2(Q50) decreases from 80 m2/g to 78 m2/g upon loading 5 wt % MgO, but the surface area drops to 28 m2/g on loading 7 wt % MgO. Hence, higher loading of MgO over 5 wt % is not recommended according to this study. 3.6.2. Calcination Temperature. The MgO surface can be viewed as a source of electron density through O2− and OH− ligands. We might expect the electron donation by the MgO to correlate with surface basicity and the catalytic properties of MgO. The calcination temperature has the most influence on the basicity and basic strength of MgO. According to Tanabe, the basicity of MgO is 0.069 and 0.051 mequiv/g, for calcination at 500 and 700 °C, respectively.58 The Gates group has reported that electron density on Re bonded to the oxygen of partially dehydroxylated MgO (400 °C) is greater than that on Re atom bonded to almost fully dehydroxylated MgO (700 °C).59 Yacob et al. reported that nano MgO calcined at 600 °C gives the highest conversion of palm oil to biodiesel.60 Di Cosimo et al., however, reported that the increase in calcination temperature from 400 to 700 °C drastically decreased the density of strong base sites.61 It could be due to different calcination atmospheres, duration of calcination, and the nature of precursors from which the oxide was prepared; optimal calcination temperatures reported in the literature are different.60,61 However, in all cases, base strength increased with increasing calcination temperature until a maximum was reached at a temperature ranging from 400 to 600 °C. Hence, a partially dehydroxylated MgO (calcination temperature lower than 600 °C) was expected to have higher electron donation tendency and thus higher catalytic activity than a fully dehydroxylated one. For SiO2-supported alkali earth oxide catalysts used in TG transesterification, besides catalyst activity, reducing alkalimetal-oxide leaching from SiO2 support is crucial in minimizing the catalyst deactivation rate. To illustrate the roles of calcination in catalysis, catalytic performance for Mg(NO3)2/ SiO2(Q50) was tested and compared to the catalytic performance for MgO/SiO2(Q50). The results showed that TG conversion catalyzed by Mg(NO3)2/SiO2(Q50) in the first run is rather fast because the reaction is mostly catalyzed by leached Mg(NO3)2, whereas the conversion rate drops abruptly (Figure S5). Hence, the purpose of calcination was suggested as oxidative thermaldecomposition of Mg(NO3)2 into MgO, enhancing the interactions between MgO and SiO2 to minimize MgO leaching. Calcination at too low temperature leaves undecomposed nitrate ions, but, on the other hand, too high temperature calcination may destruct the SO2 pore structure and cause MgO aggregation and sintering. FT-IR characterizing oxidative thermal treatment of Mg(NO3)2/SiO2 indicates that nitrate ions will not be totally decomposed at temperatures below 500 °C; the presence of nitrate characteristic peaks, 1380 and 850 cm−1, for Mg(NO3)2/ SiO2 calcined at 400 °C (Figure S6) indicated that some amount of nitrate in precursors remains in SiO2 support.62 Hence, notwithstanding the Mg(NO3)2/SiO2 catalyst calcined

Figure 11. Powder X-ray pattern of (a) fresh MgO/SiO2(Q50) and (b) used MgO/Q50second‑re-3.

for the fresh MgO/SiO2(Q50), in addition to the (111) plane. Hence, the correlation between structure and catalytic performance results suggests that the reaction and rejuvenation process drives the destruction of coordination stepped (111) and (110) surface planes, which have more basicity and exhibit higher FAME selectivity in comparison with that of the (100) plane.56,57 Since MgO (100) is the most stable plane among (100), (110), and (111) surfaces because of its low surface energy, (110) and (111) planes are gradually eroded in the cycle tests. Besides the decrease of reaction contributed from homogeneous reactions catalyzed by the leached MgO, the erosion of the (110) and (111) planes could result in a decrease of FAME yield. As shown in Tables S1, S5, and S6, FAME yield decreases with increasing test cycles. After test reactions of the rejuvenated catalysts MgO/ Q50second‑re, an additional XRPD peak of 21.0° appears nearby the characteristic peak of periclase MgO (200) (Figure 11b). The peak could be caused by reloading 1 wt % MgO to the used MgO/Q50first‑re. The shift Bragg peaks of XRPD patterns to lower 2θ indicate the reloaded MgO having a slightly larger unit cell than that of the silica-bound MgO. The increase of the unit cell could be due to the lack of SiO2 restriction during the growth of MgO crystal grains. 3.6. Comments on the Preparation and Rejuvenation of Silica-Supported MgO Catalysts. 3.6.1. Metal Loading. Conversion was expected to be increased with increasing MgO K

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research at 400 °C having more strong base sites than that calcined at 500 °C, we calcined Mg(NO3)2/SiO2 at 500 °C. However, the change in MgO-SiO2 interactions with calcination temperature and the corresponding MgO leaching is a research topic of interest and will be studied in the future. 3.6.3. Merits of Pellet MgO/SiO2. The ultimate goal of this study is to develop commercially viable catalysts, which can be used in fixed bed reactors and can be regenerated and rejuvenated easily so as to reduce the efforts of separation catalysts from the reaction system and reduce the cost of catalyst preparation and spent catalyst disposal. In comparison with the catalyst reutilization methods reported in the literature, our method is different in two important aspects: namely, (1) to deposit the MgO on SiO2 support to make the catalyst size large and stable for fixed bed operation and (2) to add fresh MgO to make up the MgO leached out for long-term operation. By the use of the rejuvenation method proposed in this study, SiO2 and on average more than 80% of the catalyst precursor, Mg(NO3)2, can be reused. Hence, the cost of spent catalyst disposal and catalyst preparation cost are greatly reduced. Moreover, as to the separation of catalyst from the reaction products, pellet SiO2-supported metal-oxide catalysts loaded in a basket type reactor seem superior to nanoparticles loaded in a slurry reactor, because high-energy-consuming centrifugation and hot filtration processes will not be necessary. 3.6.4. Further Improvement in Catalyst Rejuvenation. Notwithstanding several merits of the pellet MgO/SiO2 catalyst having been presented, the procedures for catalyst rejuvenation can be improved and optimized, however. For example, in the regeneration step, instead of solvent washing using EtOH, the used catalyst can be washed with THF followed with oxidation in air to burn off the carbonaceous material to recover more activity and minimize the makeup MgO. In fact it was found that gum formed on the catalyst cannot be washed out completely, and two-layer MgO with raft morphology was formed on the rejuvenated catalysts leading to a decrease in activity and FAMEs yield. Yao and Huang have reported that residual oil on CaO/MgO catalysts can be removed completely by washing with ethanol at ultrasonic vibrations.63 It could be due to larger pellet size and more severe reaction conditions used in this study; with the same treatment method, trace gum on pellet MgO/SiO2 was still observed in FT-IR spectra (Figure 5). 3.6.5. Reasons for Using Pellet Catalysts and Choosing MgO as Active Sites. Experimental results indicated that SiO2supported MgO is comparable to MgO nanoparticles in activity, while the leaching rate is greatly reduced, so that MgO consumption and waste disposal can be minimized. Moreover, by the use of pellet catalysts, high-energy-consuming centrifugation and hot filtration processes will not be necessary in the separation of catalysts from the reaction products. Hence, instead of fine particles in the nano to micron range, 2−4 mm pellet catalysts were used in the performance tests. However, the catalytic activity of SiO2-supported MgO seems somewhat lower than that of the catalysts reported in the literature.63−69 Even this direct comparison might be inappropriate; because the test reactions in this study were under a lower MeOH/oil ratio and stirring rate, low activity for MgO could also be the reason for lower activity of MgO/SiO2. Actually, the catalytic activities for alkaline metal oxides decrease in the order of BaO > SrO > CaO > MgO.70 MgO were chosen in this study, because (1) the surface structure is robust and only moderately water sensitive, that is low leaching rate and (2) the relationship

between structure, surface polarizability of MgO nanocrystals, and catalytic reactivity has been characterized. A fundamental understanding of the relationship between catalyst preparation conditions, pore structure, alkaline metal-oxide morphology, and catalytic properties allows us to prepare tailor-made pellet catalysts for a continuous fixed-bed reaction system in commercial operations.

4. CONCLUSION The initial goal of this study is to develop a catalyst which is easy to separate from the products and rejuvenate after reaction. Pellet silica-supported MgO catalysts, MgO/SiO2(Q30) and MgO/SiO2(Q50), were prepared by the impregnation method. These two catalysts and MgO powder were tested for transesterification of soybean oil. Since pellet MgO/SiO2 was loaded in a basket type reactor, separation of the used catalysts for recycling is much more effective than MgO nanoparticles run in a slurry reactor, thus, a high-energyconsuming centrifugation and hot filtration step will not be necessary. Catalytic performance test results indicated that the activity of SiO2-supported MgO is comparable to that of MgO nanoparticles, while the leaching rate is greatly reduced. Used catalysts after an aging test were rejuvenated by ethanol washing followed by adding 1 wt % MgO to make up the leached out MgO. By this rejuvenation method, TG conversion was recovered to the level comparable to that of the fresh catalysts, whereas FAME yield is still lower. Correlation between the MgO structure characterized by synchrotron XRPD and performance test results suggests that, besides MgO leaching and strong pore diffusion limitation, the erosion of (110) and (111) planes in the reaction could result in a decrease of FAME yield. The lower FAME yield for the rejuvenated catalysts could also be caused by the change in MgO morphology in the catalyst rejuvenation. An incomplete washing of deposited carbonaceous material results in a layer of carbon material sandwiched in two MgO layers. Due to the lack of SiO2−MgO interactions during the growth of MgO crystal grains, terrace-like MgO may be formed on the carbon layer. Hence, the improved catalytic performance method could be a washing using THF, followed with air burning and reloading MgO to make up the leaching loss, as well as using mixedoxides to increase the catalyst activity.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.iecr.7b02859.



Figures and tables containing yield patterns, rate constants, and catalyst characterizations (PDF)

AUTHOR INFORMATION

Corresponding Authors

*J.-R.C. E-mail: [email protected]. *H.-S.S. E-mail: [email protected]. ORCID

Jen-Ray Chang: 0000-0002-5737-931X Notes

The authors declare no competing financial interest. L

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research



(18) Thitsartarn, W.; Kawi, S. An active and stable CaO−CeO2 catalyst for transesterification of oil to biodiesel. Green Chem. 2011, 13, 3423−3430. (19) López Granados, M.; Martín Alonso, D.; Alba-Rubio, A. C.; Mariscal, R.; Ojeda, M.; Brettes, P. Transesterification of triglycerides by CaO: increase of the reaction rate by biodiesel addition. Energy Fuels 2009, 23, 2259−2263. (20) Zabeti, M.; Daud, W. M. A. W.; Aroua, M. K. Activity of solid catalysts for biodiesel production: a review. Fuel Process. Technol. 2009, 90, 770−777. (21) Di Serio, M.; Ledda, M.; Cozzolino, M.; Minutillo, G.; Tesser, R.; Santacesaria, E. Transesterification of soybean oil to biodiesel by using heterogeneous basic catalysts. Ind. Eng. Chem. Res. 2006, 45, 3009−3014. (22) López, D. E.; Goodwin, J. G., Jr; Bruce, D. A.; Lotero, E. Transesterification of triacetin with methanol on solid acid and base catalysts. Appl. Catal., A 2005, 295, 97−105. (23) Islam, A.; Taufiq-Yap, Y. H.; Chu, C.-M.; Chan, E.-S.; Ravindra, P. Studies on design of heterogeneous catalysts for biodiesel production. Process Saf. Environ. Prot. 2013, 91, 131−144. (24) Margaritondo, G. Elements of synchrotron light: for biology, chemistry and medical research; Oxford University Press: USA, 2002; DOI: 10.1063/1.1712508. (25) Sheu, H.-S.; Liu, P.-H.; Cheng, H.-L.; Chao, K.-J.; Chang, Y.-P. Structural characterization of porous film materials and the supported metal catalysts by synchrotron powder X-ray diffraction. Catal. Today 2004, 97, 55−61. (26) Chang, J.-R.; Huang, S.-M. Pd/Al2O3 Catalysts for Selective Hydrogenation of polystyrene-block-polybutadiene-block-polystyrene thermoplastic elastomers. Ind. Eng. Chem. Res. 1998, 37, 1220−1227. (27) Carvalho, M. S.; Mendonça, M. A.; Pinho, D. M. M.; Resck, I. S.; Suarez, P. A. Z. Chromatographic analyses of fatty acid methyl esters by HPLC-UV and GC-FID. J. Braz. Chem. Soc. 2012, 23, 763− 769. (28) Hammersley, A. P. ESRF internal report; FIT2D V12.012 Reference Manual V6.0 ESRF98HA01T; 2004. (29) Foster, M.; Furse, M.; Passno, D. An FTIR study of water thin films on magnesium oxide. Surf. Sci. 2002, 502−503, 102−108. (30) Chizallet, C.; Costentin, G.; Che, M.; Delbecq, F.; Sautet, P. Infrared characterization of hydroxyl groups on MgO: a periodic and cluster density functional theory study. J. Am. Chem. Soc. 2007, 129, 6442−6452. (31) Little, L. H. Infrared spectra of adsorbed species; New York, London, 1966; pp 250−261. (32) Hair, M. L. Infrared spectroscopy in surface chemistry; Marcel Dekker: New York, 1967; pp 198−200. (33) Ramasamy, V.; Ponnusamy, V.; Sabari, S.; Anishia, S. R.; Gomathi, S. S. Effect of grinding on the crystal structure of recently excavated dolomite. Indian J. Pure Ap. Phys. 2009, 47, 586−591. (34) Di Serio, M.; Tesser, R.; Pengmei, L.; Santacesaria, E. Heterogeneous catalysts for biodiesel production. Energy Fuels 2008, 22, 207−217. (35) Rabelo, S. N.; Ferraz, V. P.; Oliveira, L. S.; Franca, A. S. FTIR analysis for quantification of fatty acid methyl esters in biodiesel produced by microwave-assisted transesterification. Int. J. Environ. Sci. Dev. 2015, 6, 964−969. (36) Masood, H.; Yunus, R.; Choong, T. S. Y.; Rashid, U.; Taufiq Yap, Y. H. Synthesis and characterization of calcium methoxide as heterogeneous catalyst for trimethylolpropane esters conversion reaction. Appl. Catal., A 2012, 425-426, 184−190. (37) Liu, X.; Piao, X.; Wang, Y.; Zhu, S.; He, H. Calcium methoxide as a solid base catalyst for the transesterification of soybean oil to biodiesel with methanol. Fuel 2008, 87, 1076−1082. (38) Shearer, G. L. An evaluation of Fourier transform infrared spectroscopy for the characterization of organic compounds in art and archaeology. Ph.D. Thesis, University College London, UK, 1989. (39) Lee, D.; Park, Y.; Lee, K. Heterogeneous base catalysts for transesterification in biodiesel synthesis. Catal. Surv. Asia 2009, 13, 63−77.

ACKNOWLEDGMENTS The support of the Ministry of Science & Technology, R.O.C. (Contract No. MOST-105-2218-E-194-007), National Synchrotron Radiation Research Center (NSRRC), and Refining & Manufacturing Research Institute, CPC Corporation, Taiwan is acknowledged.



REFERENCES

(1) Freedman, B.; Pryde, E. H.; Mounts, T. L. Variables affecting the yields of fatty esters from transesterified vegetable oils. J. Am. Oil Chem. Soc. 1984, 61, 1638−1643. (2) Meher, L. C.; Vidya Sagar, D.; Naik, S. N. Technical aspects of biodiesel production by transesterificationa review. Renewable Sustainable Energy Rev. 2006, 10, 248−268. (3) Ma, F.; Hanna, M. A. Biodiesel production: a review. Bioresour. Technol. 1999, 70, 1−15. (4) Ejikeme, P. M.; Anyaogu, I. D.; Ejikeme, C. L.; Nwafor, N. P.; Egbuonu, C. A. C.; Ukogu, K.; Ibemesi, J. A. Catalysis in biodiesel production by transesterification processes-an insight. E-J. Chem. 2010, 7, 1120−1132. (5) Thanh, L. T.; Okitsu, K.; Boi, L. V.; Maeda, Y. Catalytic technologies for biodiesel fuel production and utilization of glycerol: a review. Catalysts 2012, 2, 191−222. (6) Albuquerque, M. C. G.; Santamaría-González, J.; Mérida-Robles, J. M.; Moreno-Tost, R.; Rodríguez-Castellón, E.; Jiménez-López, A.; Azevedo Diana, C. S.; Cavalcante, C. L., Jr.; Maireles-Torres, P. MgM (M = Al and Ca) oxides as basic catalysts in transesterification processes. Appl. Catal., A 2008, 347, 162−168. (7) Talha, N. S.; Sulaiman, S. Overview of catalysis in biodiesel production. J. Eng. Appl. Sci. 2016, 11, 439−448. (8) Nakagaki, S.; Bail, A.; dos Santos, V. C.; de Souza, V. H. R.; Vrubel, H.; Nunes, F. S.; Ramos, L. P. Use of anhydrous sodium molybdate as an efficient heterogeneous catalyst for soybean oil methanolysis. Appl. Catal., A 2008, 351, 267−274. (9) Bobade, V. V.; Kulkarni, K. S.; Kulkarni, A. D. Application of heterogeneous catalyst for the production of biodiesel. IJAET 2011, 2, 184−185. (10) Lengyel, J.; Cvengrošová, Z.; Cvengroš, J. Transesterification of triacylglycerols over calcium oxide as heterogeneous catalyst. Pet. Coal 2009, 51, 216−224. (11) Watanabe, Y.; Shimada, Y.; Sugihara, A. Continuous production of biodiesel fuel from vegetable oil using immobilized Candida antarctica lipase. J. Am. Oil Chem. Soc. 2000, 77, 355−360. (12) Sakai, T.; Kawashima, A.; Koshikawa, T. Economic assessment of batch biodiesel production processes using homogeneous and heterogeneous alkali catalysts. Bioresour. Technol. 2009, 100 (13), 3268−3276. (13) Atadashi, I. M.; Aroua, M. K.; Abdul Aziz, A. R.; Sulaiman, N. M. N. The effects of catalysts in biodiesel production: a review. J. Ind. Eng. Chem. 2013, 19, 14−26. (14) Patil, P. D.; Deng, S. Transesterification of camelina sativa oil using heterogeneous metal oxide catalysts. Energy Fuels 2009, 23, 4619−4624. (15) Lee, A. F.; Bennett, J. A.; Manayil, J. C.; Wilson, K. Heterogeneous catalysis for sustainable biodiesel production via esterification and transesterification. Chem. Soc. Rev. 2014, 43, 7887−7916. (16) Albuquerque, M. C. G.; Jiménez-Urbistondo, I.; SantamaríaGonzález, J.; Mérida-Robles, J. M.; Moreno-Tost, R.; RodríguezCastellón, E.; Jiménez-López, A.; Azevedo, D. C. S.; Cavalcante, C. L., Jr.; Maireles-Torres, P. CaO supported on mesoporous silicas as basic catalysts for transesterification reactions. Appl. Catal., A 2008, 334, 35−43. (17) Kesić, Ž .; Lukić, I.; Zdujić, M.; Mojović, L.; Skala, D. Calcium oxide based catalysts for biodiesel production: a review. Chem. Ind. Chem. Eng. Q. 2016, 22, 391−408. M

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX

Article

Industrial & Engineering Chemistry Research (40) Boey, P. L.; Maniam, G. P.; Hamid, S. A. Performance of calcium oxide as a heterogeneous catalyst in biodiesel production: A review. Chem. Eng. J. 2011, 168, 15−22. (41) Kim, H.; Kang, B.; Kim, M.; Park, Y. M.; Kim, D.; Lee, J.; Lee, K. Transesterification of vegetable oil to biodiesel using heterogeneous base catalyst. Catal. Today 2004, 93−95, 315−320. (42) Thommes, M. Physical adsorption characterization of nanoporous materials. Chem. Ing. Tech. 2010, 82, 1059−1073. (43) Condon, J. B. Surface area and porosity determinations by physisorption measurements and theory; Elsevier: UK, 2006. (44) Froment, G. F.; Bischoff, K. B. Chemical reactor analysis and design; John Wiley & Sons: New York, 1990; pp 171−173, 145−157. (45) Di Serio, M.; Cozzolino, M.; Giordano, M.; Tesser, R.; Patrono, P.; Santacesaria, E. From homogeneous to heterogeneous catalysts in biodiesel production. Ind. Eng. Chem. Res. 2007, 46, 6379−6384. (46) Jahangir Alam, M.; Ali Zaker Shawon, S. M.; Sultana, M.; Waslkur Rahman, M.; Maksudur Rahman Khan, M. Kinetic study of biodiesel production from soybean oil; 2014 Power and Energy Systems: Towards Sustainable Energy (PESTSE), 14 March 2014, 432-4336, India, DOI: 10.1109/PESTSE.2014.6805328. (47) Nasreen, S.; Liu, H.; Lukic, I.; Qurashi, L. A.; Skala, D. Heterogeneous kinetics of vegetable oil transesterification at high temperature. Chem. Ind. Chem. Eng. Q. 2016, 22, 419−429. (48) López Granados, M.; Martín Alonso, D.; Al ba-Rubio, A. C.; Mariscal, R.; Ojeda, M.; Brettes, P. Transesterification of triglycerides by CaO: increase of the reaction rate by biodiesel addition. Energy Fuels 2009, 23, 2259−2263. (49) Xu, S.; Zeng, H.-Y.; Cheng, C.-R.; Duan, H.-Z.; Han, J.; Dinga, P.-X.; Xiao, G.-F. Mg−Fe mixed oxides as solid base catalysts for the transesterification of microalgae oil. RSC Adv. 2015, 5, 71278−71286. (50) Nakatake, D.; Yazaki, R.; Matsushima, Y.; Ohshima, T. Transesterification reactions catalyzed by a recyclable heterogeneous Zinc/Imidazole catalyst. Adv. Synth. Catal. 2016, 358, 2569−2574. (51) Hymavathi, D.; Prabhakar, G.; Sarath Babu, B. Biodiesel production from vegetable oils: an optimization process. IJCPT 2014, 4, 21−30. (52) Ngamcharussrivichai, C.; Totarat, P.; Bunyakiat, K. Ca and Zn mixed oxide as a heterogeneous base catalyst for transesterification of palm kernel oil. Appl. Catal., A 2008, 341, 77−85. (53) Cullity, B. D.; Stock, S. R. Elements of X-ray diffraction, 2nd ed.; Prentice-Hall Inc.: 2001; pp 292, 402−403. (54) Jenkins, R.; Snyder, R. L. Introduction to X-ray powder diffractometry; John Wiley & Sons, Inc.: 1996; pp 236−241. (55) Corma, A.; Iborra, S. Optimization of alkaline earth metal oxide and hydroxide catalysts for base-catalyzed reactions. Adv. Catal. 2006, 49, 239−302. (56) Montero, J. M.; Gai, P.; Wilson, K.; Lee, A. F. Structure-sensitive biodiesel synthesis over MgO nanocrystals. Green Chem. 2009, 11, 265−268. (57) McKenna, K. P.; Sushko, P. V.; Shluger, A. L. Inside powders: a theoretical model of interfaces between MgO nanocrystallites. J. Am. Chem. Soc. 2007, 129, 8600−8608. (58) Bond, G. C.; Webb, G.; Malinowski, S.; Marczewski, M. Catalysis by solid acids and bases. In Catalysis: Volume 8; Bond, G. C., Webb, G., Eds.; 1989; p 141. (59) Honji, A.; Cron, L. U.; Chang, J.-R.; Gates, B. C. Ligand effects in supported metal carbonyl: X-ray absorption spectroscopy of Rhenium subcarbonyls on magnesium oxide. Langmuir 1992, 8, 2715−2719. (60) Yacob, A. R.; Mustajab, M. K. A. A.; Samadi, N. S. Calcination temperature of nano MgO effect on base transesterification of palm oil. WASET 2009, 32, 408−412. (61) Di Cosimo, J. I.; Díez, V. K.; Ferretti, C.; Apesteguía, C. R. Basic catalysis on MgO: generation, characterization and catalytic properties of active sites. Catalysis 2014, 26, 1−28. (62) Li, X. H.; Dong, J. L.; Xiao, H. S.; Lu, P. D.; Hu, Y. A.; Zhang, Y. H. FTIR−ATR in situ observation on the efflorescence and deliquescence processes of Mg(NO3)2 aerosols. Sci. China, Ser. B: Chem. 2008, 51, 128−137.

(63) Yao, Y.; Huang, W. An effective regeneration method for CaO/ MgO catalyst used in biodiesel synthesis. Energy Sources, Part A 2011, 34, 261−266. (64) Singh, V.; Yadav, M.; Sharma, Y. C. Effect of co-solvent on biodiesel production using calcium aluminium oxide as a reusable catalyst and waste vegetable oil. Fuel 2017, 203, 360−369. (65) Vahid, B. R.; Haghighi, M. Biodiesel production from sunflower oil over MgO/MgAl2O4 nanocatalyst: effect of fuel type on catalyst nanostructure and performance. Energy Convers. Manage. 2017, 134, 290−300. (66) Sudsakorn, K.; Saiwuttikul, S.; Palitsakun, S.; Seubsai, A.; Limtrakul, J. Biodiesel production from Jatropha Curcas oil using strontium-doped CaO/MgO catalyst. J. Environ. Chem. Eng. 2017, 5, 2845−2852. (67) Su, J.; Li, Y.; Wang, H.; Yan, X.; Pan, D.; Fan, B.; Li, R. Supermicroporous solid base MgO-ZrO2 composite and their application in biodiesel production. Chem. Phys. Lett. 2016, 663, 61−65. (68) Hadiyanto, H.; Afianti, A. H.; Navi’a, U. I.; Adetya, N. P.; Widayat, W.; Sutanto, H. The development of heterogeneous catalyst C/CaO/NaOH from waste of green mussel shell (Perna varidis) for biodiesel synthesis. J. Environ. Chem. Eng. 2017, 5, 4559−4563. (69) Kouzu, M.; Fujimori, A.; Suzuki, T.; Koshi, K.; Moriyasu, H. Industrial feasibility of powdery CaO catalyst for production of biodiesel. Fuel Process. Technol. 2017, 165, 94−101. (70) Romero, R.; Martínez, S. L.; Natividad, R. Biodiesel production by using heterogeneous catalysts. In Alternative Fuel; Manzanera, M., Eds.; 2011; DOI: 10.5772/23908.

N

DOI: 10.1021/acs.iecr.7b02859 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX