Glycerol Esterification for Triacetin Production using Fe3O4@SiO2@PO43- as Heterogeneous Magnetic Catalyst


 To facilitate the magnetic separation, phosphate group is embedded onto silica-coated Fe3O4 magnetic nanoparticles to prepare Fe3O4@SiO2@PO43− solid catalyst for the glycerol esterification with acetic acid. The catalyst was characterized by powder X-ray diffraction (XRD), transmission electron microscopy (TEM), vibrating magnetic spectroscopy (VSM) and Fourier Transform Infrared (FTIR) spectroscopy. The Fe3O4@SiO2@PO43− magnetic catalyst during the glycerol esterification with acetic acid was found to demonstrate excellent glycerol conversion levels (97 %) while retaining 92 % triacetin selectivity. The plausible mechanism of glycerol esterification suggests the initiation of the reaction by the protonation of the acetic acid. The catalyst was recovered from the reaction mixture under the influence of external magnetic field and reused during 4 consecutive reaction cycles.


Introduction
At the beginning of the twentieth century, biodiesel (BD) has emerged as an alternative fuel to combat the issues related to climate change, air pollution, energy demand and energy security. Ever increasing BD production also lead to generate enormous amount of crude glycerol (CGL) which would be approximately one tenth of BD produced. Disposal and puri cation of this excess CGL may pose serious threat to the environment and may cause water as well as soil pollution (Mufrodi et al 2020; Testa et al. 2013).
Therefore, for the past few years, researchers are exploring various methods for e cient utilization of CGL (Okoye et al. 2017). The CGL possesses a meagre economic value because of the impurities present (methanol, catalyst, BD, trig-, di-, and mono-glycerides).
Moreover, puri cation of CGL is quite expensive and may generate huge amount e uents, which is challenging to dispose of and may cause the environmental pollution (Chozhavendhan et al. 2020). One of the most suitable application of this CGL could be as precursor for the synthesis of ne chemicals for nonedible application. Glycerol (GL) esteri cation with acetic acid in presence of acidic acid leads to the formation of acetins viz., monoacetin (MA), diacetin (DA), and triacetin (TA) (Nda-Umar et al. 2020) as shown in scheme 1. TA has a variety of industrial application in pharmaceuticals, cosmetics, polymers, cryogenics, and fuels as an antiknock additive for gasoline, and cold ow and viscosity improver additive for BD (Testa et al. 2019).
For TA synthesis most popular approach is homogeneous Brønsted acid (H 2 SO 4 , H 3 PO 4 , HClO 4 etc.) catalysed GL esteri cation, with acetic acid (AcA), which is a low cost eco-friendly and non-hazardous chemical in comparison to acetic anhydride (AcAn) (Nda-Umar et al. 2020). Homogeneous Acid catalysts demonstrate better reactivity and product selectivity, however, they are highly corrosive, needed neutralization and removal from the reaction mixture, and are non-reusable. Thus, e uents generated during the product washing, to remove the catalyst, need to be disposed of safely to avoid any kind of environmental pollution. To simplify the issues associated with homogeneous catalysts, many researchers, in recent past, have paid additional attention in developing heterogeneous acidic catalysts. Such catalysts are easy to separate from the reaction mixture, stable and reusable (Testa et al. 2019). In case of TA synthesis, the product selectivity was found to be a function of acidic strength of the catalyst, AcA/GL molar ratio and reaction duration. The Brønsted acidic sites (e.g. PO 4 3− , SO 4

Instrumentation
Powder X-ray diffraction data of the prepared samples was collected on a PANalytical's X'Pert Pro using Cu-Kα radiation ( = 0.15406 nm) in the 2θ range of 10 to 80°. The phases present in the samples were identi ed by comparing the diffraction pattern with the JCPDS (Joint Committee of the Powder Diffraction Standards) database les. The Fourier transform-infrared spectra (FTIR) spectra of the samples were recorded in a KBr matrix on an Agilent Cary-660 spectrophotometer in the range of 400-4000 cm -1 .
To study the nature of the catalytic acidic (Brønsted or Lewis acid) sites, the samples were saturated with pyridine at room temperature and then dried at 50°C for 2 h and further heated at 300°C in a mu e furnace for 10 min. Finally, the diffuse re ectance FTIR (DRIFT) spectra of the pyridine treated catalyst samples were recorded in KBr matrix in the mid IR range (400-4000 cm -1 ).
The surface acidity of the samples were calculated by conducting a temperature programmed desorption study of NH 3 (NH 3 -TPD) using a Microtrac-BEL Corporation BELCAT II instrument which utilizes a thermo coupled detector (TCD) for detecting the evolved gases.
The speci c surface area was measured by the Brunauer-Emmett-Teller (BET) method and the pore size by the BJH method using a BEL mini-II, instrument. Prior to the analysis, the samples were heated at 100°C for 3 h under vacuum to remove any adsorbed molecules from the catalyst surface.
Scanning electron microscopy coupled with energy dispersive X-ray spectrometry (SEM-EDS) was performed on a JEOL JSM 6510LV instrument and transmission electron microscopy (TEM) images were recorded on a Hitachi 7500 instrument.
The organic products have been quanti ed by high-performance liquid chromatography (HPLC) over an Agilent In nity 1200 instrument. During the analysis iso-propanol (IPA)/hexane (60/40; v/v) was employed as the mobile phase with a ow rate of 0.6 mL min − 1 , the RX-SIL column (4.6 × 250 mm, 5 µ) as the stationary phase and the peaks were identi ed by using a refractive index (RI) detector. All samples were analysed by maintaining a column temperature of 35°C and injecting a xed sample volume of 20 µL.

Preparation of Fe 3 O 4 MNPs
MNPs were synthesized by following the literature reported procedure with a slight modi cation (Kazemifard et al. 2018). In a typical preparation, as shown in Fig. 1

Esteri cation of glycerol
The esteri cation of GL with AcA was performed in a 50 ml double necked-round bottomed ask equipped with a magnetic stirrer and temperature controller. The optimum reaction conditions for the esteri cation of GL was established by varying the molar ratio of AcA/GL in the range of 1-9, reaction temperature (30-100°C) and catalyst amount (1-5 wt% concerning GL). The products obtained during the reaction were quanti ed by the HPLC technique. A typical HPLC chromatogram of the reaction products obtained under the optimized reaction condition is shown in Fig. S1, indication a maximum of 97 % GL conversion with 92 % TA selectivity.
To evaluate the catalyst reusability, it was recovered from the reaction mixture by applying external magnetic force (Fig. 2 (Fig. 3b), to indicate the presence of amorphous silica over these particles. The FTIR study also con rmed silica coating over the MNPs, as will be discussed in the subsequent section. The average crystallite size of the Fe 3 O 4 @SiO 2 @PO 4 3− particles calculated from Debye-Scherrer equation was found to be 143 nm.

FTIR Analysis
The

Pyridine adsorption study
In the FTIR spectra of pyridine saturated Fe 3 O 4 as well as Fe 3 O 4 @SiO 2 particles (Fig. 5), a peak at 1627 cm − 1 was observed to indicate the covalent interaction between Lewis (L) acidic sites and pyridine molecule. However, the incorporation of phosphate group over the matrix leads to the formation of Brønsted (B) acidic sites. Interaction of these sites with pyridine leads to the formation of pyridinium ion which is characterized by the band at ~ 1546 cm − 1 . The band at ~ 1490 cm − 1 is assigned to the interaction of pyridine on B and L sites over the catalyst surface.

NH 3 -TPD analysis
The strength of acidic sites over the catalyst surface and total acidity was calculated with the help of NH 3

BET analysis
The nitrogen adsorption-desorption isotherms of Fe 3 O 4 and Fe 3 O 4 @SiO 2 @PO 4 3− are shown in Fig. 7, which reveal the surface porosity and pore size behaviour of these nanoparticles.

SEM and TEM analysis
The surface morphology of Fe 3 O 4 and Fe 3 O 4 @SiO 2 @PO 4 3− was compared by SEM technique, as shown in Fig. 9. The SEM images of the Fe 3 O 4 (Fig. 9a) show the formation of ~ 100 nm sized spherical particles which undergo the clustering to form particles of irregular geometry upon phosphate and silica incorporation over Fe 3 O 4 (Fig. 9b). The TEM analysis shows that Fe 3 O 4 spheres ( Fig. 9c) are the clusters of ~ 147 nm sized particles. While Fe 3 O 4 @SiO 2 @PO 4 3− agglomerates are also made up of spherical particles (~ 156 nm) as shown in Fig. 9d. The silica coating of 10.58 nm thickness over Fe 3 O 4 particles could also be observed (Fig. 9d). The crystallite size (143 nm) of Fe 3 O 4 @SiO 2 @PO 4 3− particles calculated from the XRD data is close to the particle size (156 nm) calculated from the TEM imaging to con rm the formation of nano sized catalyst particles.

Catalyst Screening
The catalytic activity of Fe 3 O 4 @SiO 2 @PO 4 3− catalyst was investigated for the GL esteri cation with AcA. The reaction parameters screened includes, AcA/GL molar ratio, reaction temperature, reaction duration, and catalyst amount for obtaining the best catalytic activity. In order to demonstrate the necessity of the catalyst and its various component, various blank and control experiments were performed during the GL esteri cation (Fig. 10). During esteri cation of GL with AcA, one or more -OH groups of the GL molecule can react with AcA. Therefore, up to three esters viz., MA, DA and TA may be formed depending on the conditions set for the reaction. In order to optimize the reaction parameters to achieve the maximum TA selectivity, out of reagent ratio, catalyst concentration, reaction duration and reaction temperature, one variable has been changed at a time as discussed below. The catalyst performance was evaluated on the basis of percentage GL conversion (= moles of all the products/moles of GL taken) × 100) and percentage TA selectivity (= moles of TA/moles of all the products) × 100).

AcA/GL molar ratio
The in uence of AcA/GL ratio on the GL esteri cation over the prepared catalyst was investigated by performing the reaction in the presence of 5 wt% catalyst (with respect to GL) at 80°C, for 80 min and varying the AcA/GL molar ratio 1:1 to 9:1 (Fig. 11a). It can be observed that at 3:1 AcA/GL molar ratio, MA was found to be the main product, which gradually decreases while the TA selectivity increases as the reactant ratio was increased from 3:1 to 6:1. At reactant ratio of 6:1, the observed GL conversion was as high as 97 % and selectivity towards TA was found to be ~ 92 %. A further rise in the AcA/GL ratio was not found to enhance the TA selectivity or GL conversion levels. In earlier reports, the sulphate functionalized CeO 2 -ZrO 2 employed for the GL esteri cation with AcA (Kulkarni et al. 2020). During the study, the catalyst was found to yield higher MA selectivity (52 %) but lower TA selectivity of 5 % while employing AcA/GL molar ratio of 3 at 100°C for 3 h of reaction duration. The TA selectivity gradually increases with increase in AcA concentration and at 10 molar ratio of AcA/GL, TA selectivity was found to be 34 % along with 99.12 % GL conversion. Figure 11b shows the in uence of catalyst amount over GL conversion and selectivity towards MA, DA and TA when the reaction was performed for 80 min, at AcA/GL molar ratio of 6:1, at reaction temperature 80°C and varying the catalyst amount from 1 to 7 wt% (concerning GL). From the result, it was observed that GL conversion increased sharply along with catalyst amount, from 0 to 5 wt%, and then remained constant from 5 wt% to 7 wt%. At 1 wt%, MA was the main product which decreases gradually on increasing the catalyst amount. Thus, the TA selectivity was found to be maximum, 92 %, at 5 wt% catalyst amount in the reaction mixture. In case of AcA/GL (10:1 molar ratio) esteri cation in the presence of SO 4 2− /CeO 2 -ZrO 2 catalyst, the TA selectivity was found to increase gradually from 7.32 to 21.26 % with the increase in catalyst amount (from 1 to 5 wt%, concerning GL) (Kulkarni et al. 2020). Thus extent of GL esteri cation as well TA selectivity was found to be a function of active (acidic) sites, which would be more at higher catalyst concentration in the reaction mixture.

Reaction temperature
The GL esteri cation was found to be endothermic in nature and hence, a higher reaction temperature favours the esteri cation to obtain TA (Li et al.2020). The effect of reaction temperature over acetin selectivity and GL conversion was studied by performing the reaction in presence of 5 wt% catalyst (concerning GL), AcA/GL molar ratio 6:1, at 80 min and varying the reaction temperature from 35 to 100°C as shown in Fig. 11c. At room temperature (35°C), the major product formed was MA which decreases with increasing temperature. As the temperature is increased to 80°C, the TA selectivity touches the highest value of 92 % and remains the same even at an elevated temperature of 100°C. Dizoglu and Sert also reports the same observation, while studying the activated carbon/UiO-66 as catalyst for GL esteri cation with AcA employing 6 wt% of catalyst amount, 6:1 of AcA/GL molar ratio, while the optimum temperature was being 90°C to obtain the TA selectivity of 17.9 % (Dizoglu and Sert 2020). In the literature, very few reports claim

Reaction duration
An increase in reaction time also found to in uence the acetin selectivity when the reaction was performed in the presence of 5 wt% catalyst (concerning GL) at 80°C, employing AcA/GL molar ratio of 6:1 and varying the reaction time from 20 to 80 min as shown in found to increase to 54 % and 34 %, respectively, while the MA selectivity reduced to 12 % after 6 h of reaction duration. Thus our study as well as literature report supported the stepwise GL esteri cation and increase in DA and TA yield on increasing the reaction duration. Thus, the study supports that three -OH groups of GL are esteri ed in a stepwise fashion.

Reusability study
The ease of catalyst separation and reusability are the main advantages of heterogeneous catalysts over homogeneous one. The magnetic catalyst was employed for the GL esteri cation under the optimized reaction conditions and after the completion of the reaction; it was removed from the reaction mixture under the in uence of an external magnetic eld as shown in Fig. 1. The recovered catalyst was washed with ethanol, dried, and calcined at 600°C for 3 h. The regenerated catalyst was further reused during four cycles under similar regeneration and experimental conditions. During the reusability experiments, the TA selectivity was found to decline up to 42 % in the second cycle and then remain 20 % during the next two consecutive cycles (Fig. 12). The reason for the decline in catalyst activity may be the leaching of the active sites from the catalyst surface.
The XRD spectra (Fig. 13) of fresh and reused catalysts revealed that the PO 4 3− moiety has partially detached from the catalyst surface. A comparison of EDX data of fresh and reused catalyst (Fig. S2) also supports the decrease in phosphorous contents (from 10.7 % to 1.6 %) in the used catalyst, to indicate the loss of phosphate group from the catalyst upon its successive reuse. Thus, catalyst decomposition was found to be the primary reason behind the loss in its activity.
3.10 Proposed mechanism for the GL esteri cation with AcA GL esteri cation in absence of catalyst resulted in partial conversion levels into acetins even after a prolonged reaction duration of 4.5 h. However, in the presence of the magnetic Fe 3 O 4 @SiO 2 @PO 4 3− catalyst, reaction duration of 80 min is required to achieve 97 % GL esteri cation. This observation indicates that phosphate group is the catalytically active species in the GL esteri cation with AcA.
Homogeneous phosphoric or sulfuric acid catalyzed esteri cation of mono-as well tri-alcohol (GL) has been reported to follow the rst-order kinetic model (Beula et al. 2013). In such reactions, the rate of the reaction was found to be a function of AcA concentration. During the present study when the AcA concentration was varied, TA selectivity was also found to uctuate. However, the same was not found to be effected when the GL concentration was varied. The effect of time on the course of reaction indicates the stepwise esteri cation of -OH groups of GL.
On the basis above mentioned experimental study as well literature reports, the plausible mechanism should involve the protonation of the AcA, to form a carbocation (II), in the rst step (Kong et al. 2016      N2 adsorption-desorption isotherms and Pore distribution branch of the nitrogen isotherm by the BJH.  SEM images of (a) Fe3O4 and (b) Fe3O4@SiO2@PO43-, and TEM images of (c) Fe3O4 and (d) Fe3O4@SiO2@PO43-.

Figure 10
Effect of the various catalysts over the esteri cation of glycerol. [Reaction conditions: AcA/GL molar ratio = 6:1, catalyst amount = 5 wt% (with respect to GL), 80 °C = reaction temperature, and reaction time = 80 min] Figure 11 Effect of the various reaction parameters on GL esteri cation: (a) AcA/GL molar ratio, (b) catalyst amount, (c) reaction duration and (d) reaction temperature.

Figure 13
Comparison of XRD spectra of fresh and reused catalyst.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.