The epoxidation of canola oil and its derivatives

Abstract

This work explores the epoxidation and subsequent acid catalyzed epoxy ring opening kinetics of canola oil (CanO), canola oil fatty acid methyl esters (CanFAME) and canola oil free fatty acids (CanFFA). Epoxidation reactions were carried out at atmospheric pressure, without any organic solvents and additional catalysts. The formation of epoxide groups and the subsequent acid catalyzed epoxy ring opening reactions and oligomerization were studied over the epoxidation time. An empirical equation describing the in situ epoxidation kinetics including the epoxy ring opening processes is proposed that is applicable to the epoxidation of plant oils and those containing fatty acid derivatives. By comparing the epoxidation kinetics of CanO, CanFAME and CanFFA, it was found that the rates of epoxidation of CanO and CanFAME were similar although the formation of hydroxylated moieties is observed in CanFAME. The epoxidation behaviour of CanFFA was unique since acid groups act as an oxygen carrier (forming peroxy fatty acids with hydrogen peroxide) which leads to extensive epoxide ring opening and formation of oligomeric products. The results of this study have significance in the production of epoxides from plant oils and animal fats, containing FFA and other derivatives.

---

Introduction

Plant oils and animal fats, which are predominantly fatty acids esterified to glycerol as triacylglycerides (TAG) mainly consist of unsaturated C₁₆–C₂₂ fatty acids (especially the unsaturated C₁₈ fatty acids oleic, linoleic and linolenic acids) and lesser amounts of saturated C₁₆–C₂₂ fatty acids (mainly stearic or palmitic acids). The fatty acid profiles of the major oils are well described in the literature.^1–3 Plant oils and animal fats have been used for many years as foods, fuels, lubricants and as a renewable chemical feedstock. Many chemical modifications of oils and fats have been described⁴ including epoxidation, carbonation, esterification and/or transesterification, ozonolysis and others. Often, chemical modification of oils and fats involves reactions at the carbon–carbon double bonds (–C=C–) of unsaturated fatty acids, an important example being epoxidation. Although there are several approaches available for the epoxidation of the double bonds of unsaturated oils and oil derivatives to make biobased epoxides, the most widely used method is the classic epoxidation reaction proposed by Prileschajew⁵ more than a century ago. This takes place in two stages: first a peroxy acid is formed and subsequently the peroxy acid reacts with unsaturated fatty acids to form epoxides. Epoxidation of plant oils is typically carried out on an industrial scale using performic or peracetic acids generated in situ from carboxylic acid and hydrogen peroxide (H₂O₂).⁶

Plant oils with a high unsaturated fatty acid content, such as linseed oil or soy oil, have become important raw materials since numerous modification reactions, including or following epoxidation, can be performed at the alkene-moieties. Several attempts have been made to define the epoxidation reaction rates of different oils and their derivatives, including rapeseed oil,⁷ corn oil,⁸ rubber seed oil,⁹ soybean and sunflower oils^7,8 and fatty acid methyl esters.^7,10 In the majority of these studies, H₂O₂ is used as oxygen donor along with variety of catalysts (oxygen carriers) to form peroxy compounds, which are reactive towards alkenes. Both homogenous and heterogeneous catalysts have been used in the epoxidation of unsaturated oils and their derivatives including formic acid,^7,10 acetic acid,^8–10 organometallic catalysts such as methyltrioxorhenium, peroxophosphatotungstate or molybdenum oxide^7,11 and lipases.^12,13 Some attempts have also been made to study the kinetics of the epoxy ring opening reaction under various acidic conditions.^10,14 A major weakness of some of these reports is that the authors have only studied the epoxidation process itself, but excluded the epoxy ring opening that happens during the epoxidation process.^7–10 In other work, the ring opening of epoxides is studied independently from the epoxidation process.^10,14 More desirable would be a simultaneous study of both the in situ epoxidation reactions and the ring opening reactions occurring during the epoxidation process, in order to provide a more complete picture of epoxidation. To the best of our knowledge, there has not been a comprehensive study considering both the kinetics of epoxidation and the possible competing epoxy ring opening reactions, for both a vegetable oil and its common derivatives.

Here we report on a study of the kinetics of epoxidation, possible epoxy ring opening kinetics and oligomerization processes in CanO, CanFAME and CanFFA.

Experimental

Materials

Food grade canola oil was purchased from the local supermarket (Loblaws Inc., Canada) and was used as supplied. In the following experiments, in order to avoid any differences in fatty acid profiles, the CanO, CanFAME and CanFFA were all prepared from the same batch of canola oil.

Formic acid (85%) and aqueous H₂O₂ (35%) were purchased from Univar Canada. Sodium chloride (NaCl) was a food grade and was purchased from Univar Canada, and used in work-up processes. Reagent grade sodium hydroxide (NaOH) was purchased from Sigma-Aldrich Canada and used for transesterification reactions to make canola oil FAMEs. Reagent grade potassium hydroxide (KOH) was also purchased from Sigma-Aldrich Canada and used for hydrolysis/saponification of canola oil to produce FFAs. Anhydrous sodium sulfate (Na₂SO₄) was technical grade and also was purchased from Univar Canada and used as drying agent. Anhydrous methanol was purchased from Sigma-Aldrich Canada and used in preparation of canola oil FAMEs. Hexanes were reagent grade and were obtained from Sigma-Aldrich Canada, and used in extraction of canola oil FFAs.

Hydrogen bromide (HBr), 33 wt% solution in glacial acetic acid was purchased from Fisher Scientific Canada. Glacial acetic acid (99.7%) was obtained from Caledon Laboratory Chemicals Canada and used to prepare 0.1 N HBr solution for oxirane oxygen content titration of epoxides. Reagent grade hydrochloric acid (HCl) was also purchased from Caledon Laboratory Chemicals Canada and used in work-up process of canola oil hydrolysis products.

Methods

GC-FID analysis of CanFAME. A Perkin Elmer (Waltham, Massachusetts, USA) Clarus 500 Gas Chromatograph equipped with a Flame Ionization Detector (GC-FID) was used for chromatographic analysis of oil derivatives. A 100 m × 0.25 mm SP-2560 Supelco capillary column with film thickness of 0.20 μm was selected for the analysis of CanFAME. The oven temperature was maintained at 45 °C for 4 min, increased to 175 °C at a rate of 13 °C min⁻¹, maintained at 175 °C for 27 min then increased to 215 °C at a rate of 4 °C min⁻¹, and finally held at 215 °C for 35 min. The injector and detector temperatures were 240 and 280 °C, respectively. Hydrogen was used as carrier gas at 1.3 mL min⁻¹ and the injection port was in split mode of 20 : 1. 1.0 μL of sample was injected into the gas chromatograph. The individual FAMES are identified by their retention times and the relative percentage (area%) of the fatty acids using a reference standard mixture of methyl esters of fatty acids. This reference standard mixture was GLC-463, from Nu-Chek Prep, Inc. The analysis of the products were carried out at least in duplicate.

HPLC-ELSD method for TAG analysis. The analysis of triacylglycerols was carried out using an Agilent 1200 series high performance liquid chromatograph (HPLC, Agilent Technologies, Inc. USA) equipped with an evaporative light scattering detector ELSD 2000 (Alltech, USA) and an Eclipse XDB-C18 column (150 mm × 4.6 mm i.d, 5 μm particle size, Agilent). The column oven temperature was maintained at room temperature, the nitrogen gas flow was set at 2 L min⁻¹ and the drift tube temperature was kept at 100 °C. The mobile phase flow rate was set at 1.5 mL min⁻¹. All solvents, including acetonitrile, methanol, isopropanol and hexane were HPLC grade and they were obtained from Fisher Scientific Canada. The sample of oil and FAMEs were prepared as following: about 8 mg of canola oil or canola FAME is dissolved in 10 mL of acetone to make a concentration of 0.8 mg mL⁻¹ 20 μL of sample is injected to the system. Mobile phase solvent A: is a mixture of acetonitrile and methanol (4 : 1), while mobile phase solvent B: is a mixture of hexane and isopropanol (8 : 5). The Table 1 below is showing the gradient elution protocol used for the HPLC-ELSD method.

Table 1 Gradient elution protocol used for HPLC-ELSD method for TAG analysis

Time [min]	% of A solvent	% of B solvent
0	100	0
2.5	100	0
15.5	50	50
20	50	50
25	30	70
26	30	70
27	100	0

---

Size exclusion chromatography (SEC). A 1200 series high performance liquid chromatograph from Agilent Technologies (Agilent Technologies Inc, USA) equipped with an Alltech ELSD 2000 (evaporating light scattering detector) was used for all size-exclusion chromatographic experiments to measure MW distributions of the epoxidation products. A Styragel HR 4E THF column (with the size of 4.6 × 300 mm) from Water Corporation, USA was used along with tetrahydrofuran (THF) as mobile phase and at a flow rate 0.5 mL min⁻¹. For every run, the eluent was the same as the sample buffer. The random run-to-run difference in retention times for our system was <0.1%. All data were processed using the Agilent ChemStation for LC 3D Systems (Rev. B.03.01). The SEC calibration curve for the column was carried out using the lipid and polystyrene standards.

Attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy. The analysis of CanO, CanFAME and CanFFA were carried out using a Bruker Alpha FTIR spectrometer equipped with an ATR accessory. IR spectra (4000–650 cm⁻¹, 32 scans, 4 cm⁻¹ resolution) of the products were obtained by placing ca. 10 μL of the epoxidized product onto the ATR crystal. FTIR data analysis was carried out using OPUS spectroscopy software package, and processed using SigmaPlot v12.5 (Systat Software Inc, US).

Transmission Fourier transform infrared spectroscopy. The hydroxyl numbers of CanO, CanFAME and CanFFA were evaluated using ABB-Bomem's WorkIR (ABB-Bomem, QC, Canada) FTIR spectrometer controlled by Windows-based Universal Method Platform for Infrared Evaluation (UMPIRE®) software (Thermal-Lube, QC, Canada). IR spectra (4000–400 cm⁻¹, 16 scans, 4 cm⁻¹ resolution) of the products were collected using CaF2 transmission cell. This method is based on the measurement of desired amount of p-toluenesulfonyl isocyanate (p-TSI) peak in the toluene solution, as described elsewhere.¹⁵ The difference of the peak area corresponding to p-TSI solution (control) and after the reaction with hydroxyl group of the sample (if any) is being proportional to the –OH value of the product. To minimize moisture vapor variation, the spectrometer was purged with dry air using a Balston Dryer (Balston, Lexington, MA).

Preparation of canola oil and its derivatives

CanO. Food grade canola oil was purchased from the local supermarket and was used as supplied, without any further treatment or modification.

CanFAME. A base-catalyzed transesterification procedure was used to prepare CanFAME, which is a more rapid esterification procedure than using acid-catalyzed reactions.^16,17 CanFAME was prepared in ∼6000 g batches, using sodium hydroxide as the base catalyst and following the usual procedure described elsewhere.¹⁸ The catalyst solution in methanol was prepared by dissolving anhydrous NaOH (30 g) in anhydrous methanol (1470 g or ∼46 mol) in a 2 L flat-bottom flask, at a temperature of 50 °C and under continuous mixing with magnetic stirrer bar (500 rpm). About 6000 g of CanO was loaded into 22 L jacketed glass reactor (Chemglass Life Sciences, NJ, USA) equipped with a bottom drain, a heating/cooling control unit and a speed-controlled overhead mechanical mixer. The temperature of the reactor was set to 60 °C and the NaOH solution in methanol was poured into the reactor to give a molar ratio of CanO to methanol of about 1 : 7, which was agitated vigorously at a speed of 350 rpm. Under these conditions, the reaction between methanol and triacylglycerol (TAG) is very rapid. After 3 hours of transesterification – the reaction between methanol and triacylglycerol (TAG) – a sample was taken to evaluate the disappearance of TAG using HPLC. The sample was washed 3 times with saturated NaCl solution, neutralized using 0.1 N HCl solution and an equal amount of hexane. Finally, the sample was concentrated in the hexane layer using a rotary evaporator giving a test sample to analyse by non-aqueous reverse phase HPLC analysis for the presence of residual TAG, as described above. A canola oil (TAG) standard was used for in the HPLC analysis to allow qualitative estimation of the extent of the transesterification reaction. It was found that after 3 hours the transesterification reaction is complete and no TAG remains. Then, the final product (FAME) was washed 3 times with saturated NaCl solution, neutralized using 0.1 N HCl solution and extracted using equal amount of hexane. FAME was dried from traces of water using anhydrous sodium sulfate, filtered, and hexane was removed using rotary evaporator. The fatty acid composition of the CanFAME is presented in Table 2.

Table 2 Fatty acid profile of canola oil and some important calculated parameters of the CanO, CanFAME and CanFFA and its epoxidized derivatives. The OOC (oxirane oxygen content) values given are the theoretical maximum values

Fatty acids	Amount [%]
C16:0	5.0
C16:1	0.3
C18:0	1.8
C18:1 9c	60.7
C18:1 11c	3.4
C18:2 n6	18.9
C18:3 n3	7.3
C20:0	0.6
C20:1 8C	0.5
C20:1 11C	1.1
C22:0	0.4
Total monounsaturated	66.0
Total polyunsaturated	26.2
Total saturated	7.8

---

Properties (calculated)	Value
Average MW CanO [g mol^−1]	882.0
Average MW CanFAME [g mol^−1]	295.3
Average MW CanFFA [g mol^−1]	281.3
Average MW ECanO [g mol^−1]	943.6
Average MW ECanFAME [g mol^−1]	315.9
Average MW ECanFFA [g mol^−1]	301.9
OOC ECanO [%]	6.34
OOC ECanFAME [%]	6.40
OOC ECanFFA [%]	6.69

---

CanFFA. Canola oil free fatty acids (CanFFA) were prepared by saponification using KOH. From the measured fatty acid composition of CanO, the saponification value of the canola oil was calculated to be ∼190 mg KOH per g oil. Saponification was carried out using 3000 g of CanO loaded into a 22 L jacketed glass reactor equipped with bottom drain, heating/cooling control unit, and speed-controlled overhead mechanical mixer. The temperature of the reactor was set to 80 °C. Desired amount of KOH is dissolved in distillated water, then this solution was loaded into the reactor with CanO and agitated at a speed of 350 rpm. TAG analysis of the product was performed using HPLC periodically to check the progress of the reaction. Hydrolysis was complete after 7–8 hours of saponification and the product was acidified using 2 M HCl. The final CanFFA was washed 3 times with saturated NaCl solution in an equal volume of hexane. CanFFA was dried using anhydrous sodium sulfate, filtered and hexane was removed using rotary evaporator. The average molecular weight of CanFFA was calculated from its fatty acid profile as shown in Table 2.

Epoxidation of CanO, CanFAME and CanFFA

Epoxidation of CanO, CanFAME and CanFFA were carried out using performic acid generated in situ from formic acid and H₂O₂. For epoxidation of CanO, the molar ratio of CanO : formic acid : H₂O₂ used was 1.0 : 1.0 : 6.0, respectively. This means that for every mole of double bonds in CanO there will be ∼0.25 moles of formic acid and 1.5 moles of H₂O₂, since the double bond functionality (i.e. the average number of double bonds per CanO TAG) of CanO is estimated to be ∼4, based on the fatty acid profile of CanO and taking the average molecular weight of CanO as 882 g mol⁻¹.

For epoxidation of CanFAME, the molar ratio of CanFAME : formic acid : H₂O₂ was 1.0 : 0.33 : 2.0. This means that as with the epoxidation of CanO, the ratio of reactive components for epoxidation of CanFAME were such that for every mole of double bonds in CanFAME there are ∼0.25 moles of formic acid and 1.5 moles of H₂O₂. A similar molar ratio of the components was used for epoxidation of CanFFA, to provide an equal reaction conditions.

The epoxidation reactions were carried out using 1000 g of either CanO, CanFFA or CanFAME. One of these was added into a 12 L glass reactor equipped with a bottom drain, a water jacket and attached to a recirculating liquid cooler/heater unit (Julabo F25, Julabo USA, Inc.) at a temperature of 25 °C. The selected amount of aqueous H₂O₂ solution (35%) was loaded into the vessel and the mixture stirred vigorously with an overhead mechanical stirrer (RZR 2021, Heidolph Instruments, Germany) at 350 ± 10 rpm to form a homogenous mixture before proceeding with addition of formic acid. Formic acid (85%) was added dropwise through the addition funnel into the reactor at an addition rate of 8–10 g min⁻¹. After complete addition of formic acid to the mixture, the temperature of the reaction was increased stepwise 5 °C/10 min up to the epoxidation temperature (40, 50 or 60 °C), while carefully monitoring temperature for possible rise due to the exothermicity of reaction.

The extent of epoxide formation was followed by measuring the oxirane oxygen content (OOC) of the epoxidized products according to the ASTM standard titrimetric method.¹⁹ For these purposes the aliquots (∼15 mL) of epoxidized products were taken periodically every 10 min without interrupting the epoxidation process. These samples were diluted with an equal amount of ethyl acetate and washed with an equal amount of saturated NaCl solution at least three times (or until pH ≈ 6) to remove the formic acid and unreacted H₂O₂. Then, the epoxide solution in ethyl acetate was dried with anhydrous Na₂SO₄, and finally concentrated from ethyl acetate using rotary evaporator at low atmospheric pressure (@60 °C, 20 mbar).

The extent of formation of oligomeric structures (if any) were measured by size exclusion chromatography (SEC). Since CanFAME is relatively volatile, the loss in the total amount of unsaturated fractions during the epoxidation process were also monitored using GC-MS. Additionally, FTIR was used to evaluate the structural changes over epoxidation processes. The epoxidation of CanO, CanFFA and CanFAME was carried out for up to 72 hours. However, in most cases the epoxidation process is complete within 24–48 hours depending on conditions, and in this manuscript we report on the results of up to 31 hours epoxidation, as extended epoxidation after 31 hours did not provide any useful insight to the reaction kinetics. Epoxidized canola oil (ECanO) was found to be a viscous product at elevated temperatures that tends to crystallize at room temperature. Epoxidized FAME (ECanFAME) is liquid at room temperature, while epoxidized FFA (ECanFFA) tends to crystallize/solidify at room temperature due to the presence of oligomerized compounds, as discussed below.

Results and discussions

Epoxidation

The objectives of this epoxidation study were twofold. Firstly, a study of the formation of epoxy groups from CanO and its derivatives could identify conditions that achieve a fully epoxidized product within the most favourable periods of time. Secondly, the stability of these epoxy groups towards further reaction should be studied at different epoxidation temperatures, in order to identify conditions that ultimately achieve the lowest degree of ring-opening side-reactions.

The kinetics of formation of epoxidized canola oil (ECanO), epoxidized canola fatty acid methyl esters (ECanFAME) and epoxidized canola free fatty acids (ECanFFA) were studied at temperatures of 40, 50 and 60 °C. Since the majority of epoxidation occurs during the earlier stages of epoxidation, the kinetics of the process was studied using the results at initial stages of epoxidation.

For in situ epoxidation, first the formic acid reacts with H₂O₂ leading to the formation of performic acid, which is the main oxidizing product acting on the double bonds of the oil (Scheme 1). This reaction eventually consumes a stoichiometric amount of H₂O₂.

Scheme 1 Formation of performic acid (A) and epoxidation of double bonds with this performic acid (B). Note that to reduce analytical complexities, epoxidized plant oil structure provided in (B) is substituted with the monounsaturated single-chain platform such as methyl oleate. R₁ and R₂ are the mixture of hydrocarbon and ester parts of FAME; m and n are the number repeating hydrocarbon units in different parts of the chain; k₁ and k₂ are the performic acid and epoxide formation rates, respectively.

Then, performic acid oxidises the unsaturated^8,20 oil, donating an oxygen atom to form the epoxide group and liberating formic acid to the bulk reaction mixture (Scheme 1(B)).⁸ Due to the re-generation of the formic acid catalyst, it is possible to use a low molar amount of the formic acid in the epoxidation processes.

Since oil epoxides are fairly reactive they may readily undergo ring-opening reactions or participate in oligomerization processes. It was reported²¹ that a high molar ratio of acid to H₂O₂ is deleterious to the survival of the epoxy groups. This is because attack of epoxy groups by formic acid may lead to the formation of hydroxylated–formiated products,^22,23 due to acid catalyzed epoxy ring-opening reactions (protonation of epoxy oxygen), as shown in Scheme 2(A).

Scheme 2 Examples of possible epoxy ring opening: with formic acid (A) and water as a nucleophile (B). k₃ and k₄ are the ring opening reaction rates by carboxylic acid and water, respectively.

It has been reported²³ that protons at the surface of catalysts (i.e. ion exchange resin) can cause epoxy ring opening by reaction with compounds present (or formed) in the overall mixture. For example, epoxy ring opening can be caused by water in the presence of a strong acid catalyst^20,24 to form di-hydroxy groups, as indicated in Scheme 2(B). Clearly epoxy ring opening at this stage is an undesirable side reaction if the maximum possible OOC is to be achieved. Thus, it is crucial to monitor and control the epoxidation conditions of plant oils in order to minimize such side-reactions. Usually, in situ epoxidation of vegetable oils needs to be carried out at temperatures below 70 °C as high temperature epoxidation lead to the excessive epoxy ring opening reactions. Moreover, peroxy acids are readily decomposed upon heating, and can detonate upon rapid heating to 80–85 °C. In addition, the decomposition rate of H₂O₂ increases approximately 2.2 times for each 10 °C rise in temperature in the range from 20 °C to 100 °C.²⁵ Taking into account these points, the epoxidation of canola oil and its derivatives were carried out below 70 °C.

All of the epoxidation reactions reported here have used 0.25 moles of formic acid per mole of –C=C– double bonds of the starting lipid, which was found to be optimum ratio after monitoring epoxidation reactions covering a range of formic acid amounts from 0.1 to 10 moles. It was established that higher amounts of formic acid lead to excessive ring opening reactions as discussed earlier, while lower amounts of formic acid require extended times for complete epoxidation, making the epoxidation process less cost-effective. An optimum amount of ∼1.5 mole H₂O₂ per mole of lipid double bonds of oils was also established after multiple epoxidation experiments using different amounts of H₂O₂ at a fixed formic acid concentration. The use of an excess of H₂O₂ is necessary to ensure that the epoxidation reaction proceeds to completion. In addition, the excess H₂O₂ compensates for its loss or decomposition during the exothermic epoxidation reaction. It should be noted the amount of formic acid and H₂O₂ used in the epoxidation of CanO, CanFAME and CanFFA vary slightly due to the differences in their average molar concentrations of double bonds.

Oxirane oxygen content of epoxides

The OOC of an epoxidized oil, expressed as grams of oxirane oxygen per 100 g of epoxidized product, is an important parameter used in designing formulations that incorporate epoxides. Since both the CanFAME and the CanFFA used for epoxidation were made from the same CanO, they have the same profile of unsaturated fatty acids so differences between their epoxidation kinetics might be expected to be minimal.

However, the properties of CanO and its derivatives, such as their interfacial tensions, viscosities, densities and polarities, do differ significantly. For example, the interfacial tension values are in the range of 25–27 mN m⁻¹ for CanO,^26,27 25.5–26 mN m⁻¹ for C16–C18 FAMES and 9.2–9.5 mN m⁻¹ for C16–C18 FFAs²⁸ at temperatures of about 60–70 °C, respectively. The viscosities of CanO, CanFAME and CanFFA are about 20 mPa s, 7–9 mPa s and 2.4 mPa s respectively, also at temperatures of about 60–70 °C.²⁸ In addition, the structural differences between these 3 forms likely also affect the reaction, for example there maybe greater steric hindrance to the epoxidation reaction for CanO due to conformations of the TAG structure, compared to the single fatty acyl chains of CanFAME and CanFFA. Based on these parameters and physical properties, it is reasonable to expect that the epoxidation of oil derivatives should proceed at a faster rate compared to CanO.

Fig. 1 shows the OOC measured during the epoxidation of (a) CanO, (b) CanFAME and (c) CanFFA at reaction temperatures of 40, 50 and 60 °C. It is clear from Fig. 1 that there are differences between the epoxidation behaviour of the three lipid forms. As expected, the overall rate of the epoxidation for all products is temperature dependant, so that a higher temperature initiates faster epoxidation.

Fig. 1 Epoxidation kinetics (OOC vs. epoxidation time) of the CanO (a), CanFAME (b) and CanFFA (c) at different temperatures of epoxidation.

To simplify the discussion, Fig. 2 compares the kinetics of epoxidation for CanO, CanFAME and CanFFA at 60 °C reaction temperature.

Fig. 2 Comparison of kinetics of epoxidation (OOC) for the CanO, CanFAME and CanFFA at 60 °C.

In Fig. 2, the epoxidation of CanO occurs at a faster rate compared to its derivatives, forming ECanO with an OOC of 6.0 wt% after 23 hours of epoxidation. Extending epoxidation of CanO beyond this time, did not significantly affect the OOC values achieved. In addition, epoxidation occurs at the highest rate during the initial stages of epoxidation (up to 5–7 h) beyond which a significant decrease in the rate of epoxidation is observed. The later, slower region of epoxidation of CanO can be attributed to the delayed epoxidation of polyunsaturated fatty acids of the oil. This behaviour was demonstrated in our earlier study²⁹ which showed from LC/MS experiments that epoxidation occurs more rapidly for monounsaturated FAs, while the polyunsaturated FAs take longer periods for complete epoxidation due to the sequential epoxidation process.

The maximum OOC value was achieved for ECanO, despite it having similar theoretical maximum OOC values to ECanFAME and ECanFFA (see Table 2). For ECanFAME, the maximum OOC value of about 5.2 wt% was observed at between 10–24 h of epoxidation. Beyond this time, slight decreases in OOC occurred. This lower maximum OOC value, and the decrease from this value over extended epoxidation times, can be attributed to epoxy ring opening processes, similar to those presented in Scheme 2.

These differences in the behaviour of CanO and CanFAME during epoxidation could be attributed to structural differences, since the epoxy group of ECanFAME is readily accessible to acid attack, whilst ring opening of ECanO could be retarded by steric hindrance caused by the 3-dimensional conformations of the three fatty acyl groups on CanO.³⁰ FT-IR results demonstrating that ECanO oxirane groups are not ring-opened are given below.

For ECanFFA the maximum of OOC of 3.8 wt% is observed after 6 hours of epoxidation and prolonged epoxidation to 31 hours leads to a significant decrease in OOC. This dramatically lower maximum OOC for ECanFFA is likely due to enhanced ring opening of the oxirane by carboxylic acid groups both from the formic acid catalyst and from the CanFFA itself. These result in the formation of various ring-opened monomers and possibly the formation of fatty acid oligomers (e.g. estolides).^19,31–33 The epoxidation process and an enhanced oxirane ring opening in ECanFAME and ECanFFA will be discussed along with the FTIR and SEC data obtained for these products.

Kinetics of epoxidation

Usually the Arrhenius equation³⁴ is used to describe the temperature dependence of the reaction rates. If the in situ formation of performic acid demonstrated in Scheme 1(A) is considered to be constant during the epoxidation reaction, then the reaction rate of Scheme 1(B) can be expressed via following relationship, as described previously:¹⁰

In this equation, [EP] is the epoxide; [H₂O₂] and [RCOOH] is the initial concentration of the H₂O₂ and formic acid, respectively; and k_EP is the epoxidation reaction rate constant. Integration of this equation leads to the following linear relationship,¹⁰ where the reaction rate constant can be estimated from (as slope of the plot) the consumption of the H₂O₂ during epoxidation:

According to eqn (2), the plot ln{[H₂O₂]₀ − [EP]} vs. t (epoxidation time) should yield a straight line, if degradation of the epoxy ring and decomposition of H₂O₂ is negligible.

The reaction rate constant was estimated from the plot of ln{[OOC]_m − [OOC]} vs. t, which basically reproduces the relationship discussed above. Note that [OOC]_m is the maximum expected value of OOC calculated based on the fatty acid profiles of CanO (see Table 2), while [OOC] is the experimental OOC value at different times of epoxidation. Fig. 3(a)–(c) represent the epoxidation kinetics of CanO, CanFAME and CanFFA at different temperatures of reaction, respectively.

Fig. 3 Kinetics of epoxidation for (a) CanO, (b) CanFAME and (c) CanFFA by in situ generated performic acid.

The values of the epoxidation rate constants were estimated from Fig. 3, and tabulated in Table 3.

Table 3 Epoxidation rate constants, k_EP, at various temperatures for CanO, CanFAME and CanFFA. For comparative purposes, the peroxyacid (performic acid) rate constants, k_P, at different temperatures are also given

T, [°C]	Epoxidation rate constant (this work) kEP × 10^−6 [L mol^−1 s^−1]			PFA rate constant (literature) kP × 10^−4 [L^6 mol^−2 s^−1]
	CanO	CanFAME	CanFFA	PFA^a^35
a Note that the data presented in this column is adopted from the ref. 35 and the concentration of H2O2 was 50% w/w in water. Equimolar amounts of formic acid and H2O2 is used.
40	32.1	25.8	23.1	7.2
50	59.4	44.5	37.9	12.0
60	73.9	71.5	59.2	19.6

---

It can be seen from Table 3 that the epoxidation reaction rate constants are in the order 10⁻⁶ L mol⁻¹ s⁻¹, which is similar to the epoxidation rate constants for different oils and oil derivatives, found in the literature.^7–10 For example, the epoxidation rate constant for soybean oil is reported to be 18.0 × 10⁻⁶ L mol⁻¹ s⁻¹ at 65 °C,⁸ while this constant is mentioned to be about 72 × 10⁻⁶ L mol⁻¹ s⁻¹ at 60 °C for palm olein based FAMES.¹⁰

The epoxidation rate constants for CanO and its derivatives depend on temperature, increasing by approximately 2.5 times as the epoxidation temperature increases from 40 to 60 °C.

Table 3 also represents the rate constants, k_P, for the formation of performic acid (PFA) at different temperatures. Note that k_P values constants are reconstructed from the literature data³⁵ in order to demonstrate the rate differences between PFA formation and epoxidation reaction rate of CanO and its derivatives with in situ formed PFA.

Also, an Arrhenius plot for PFA, reconstructed from De Fillipis et al.³⁵ is included in Fig. 4 and used to determine the parameters for PFA formation in Table 4.

Fig. 4 Arrhenius plots for epoxidation of CanO, CanFAME and CanFFA by in situ generated PFA. Arrhenius plot for PFA is adopted from ref. 35.

Table 4 Kinetic parameters for epoxidation of CanO, CanFAME and CanFFA at 60 °C

	Ea [kJ mol^−1]	ΔH [kJ mol^−1]	ΔS [J mol^−1 K^−1]	ΔG [kJ mol^−1]
a Note that the kinetic parameters for PFA at 60 °C are reconstructed from the ref. 35.
CanO	36.4	33.6	−222.7	107.8
CanFAME	44.3	41.5	−200.6	108.3
CanFFA	40.8	38.0	−211.9	108.6
PFA^a	43.5^a	40.8^a	−167.1^a	96.4^a

---

It can be seen from Table 3 that the k_P is about two orders of magnitude higher than k_EP. Thus, the epoxidation rate of CanO and its derivative is not limited by the rate of formation of PFA, assuming that irreversible decomposition of the PFA is negligible.

The activation energy, E_a, for the epoxidation of CanO and its derivatives by in situ formed PFA were evaluated from the slopes in Fig. 4. The enthalpy of activation (ΔH), the entropy (ΔS) and the free energy (ΔG) of activation for epoxidation are calculated using the relationships described elsewhere^35,36 (see ESI†):

As can be seen from the table, the activation energy for the epoxidation of CanO is about 36 kJ mol⁻¹, which is lower than the 44 kJ mol⁻¹ for CanFAME and 41 kJ mol⁻¹ for CanFFA. These activation energy values are consistent with earlier studies that reported activation energies for the epoxidation of various vegetable oils to be in the range of 25–85 kJ mol⁻¹.^7,8,10 For example, the activation energy for the epoxidation of rapeseed oil, soybean oil and sunflower oil (catalysed by formic acid) is estimated to be 26, 45 and 83 kJ mol⁻¹, respectively.⁷ The same values of FAMEs are reported to be about 32 ⁷ and 53 kJ mol⁻¹.¹⁰

It should be noted that the equations that were used to calculate the kinetic and thermodynamic parameters only take into account epoxides that were not lost to processes of epoxy ring degradation. In order to estimate an accurate value of the kinetic and thermodynamic parameters it would be desirable to include a correction to account for any actual loss of epoxides.

Kinetics of epoxy ring opening

The extent of epoxy ring opening that occurs during the epoxidation of the CanO and it derivatives is estimated via the measurement of hydroxyl groups. Epoxy ring opening can occur as a result of acid attack which leads to the formation of one hydroxyl group and/or ring opening with water in the presence of acid which forms two hydroxyl groups (see Scheme 2(A) and (B)). In this work, only the overall extent of hydroxyl group formation is measured, without distinguishing between these possible hydroxyl forming reactions.

The hydroxyl values (OHV) of the epoxidized products were measured using single sample mid-FTIR spectroscopy method, as described elsewhere.¹⁵ Fig. 5 represents the formation and change of the hydroxyl groups as a results of epoxy ring opening reactions during in situ epoxidation processes at different temperatures for (a) CanO, (b) CanFAME and (c) CanFFA.

Fig. 5 Epoxy ring opening kinetics (OHV vs. epoxidation time) of the CanO (a), CanFAME (b) and CanFFA (c) at different temperatures of in situ epoxidation.

It can be seen that the epoxy ring opening depends on temperature, so that higher temperatures result in more ring opening reactions for all three lipid types.

For CanO, a maximum OHV of about 10 mg KOH per g is observed after 31 h of epoxidation (Fig. 5(a)), at all temperatures studied. Here, the measurement of OHV was carried out indirectly by FT-IR measurement of the extent of reaction with p-TSI (see Experimental). The low OHV measured can be attributed to the consumption of p-TSI either by moisture³⁷ and/or possibly by a small fraction of the epoxide groups.³⁸ A qualitative estimation of the hydroxyl group formation using ATR-FTIR experiments indicated that no traces of hydroxylated moieties are present in the epoxidized canola oil (see the FTIR section below). Hence, the measured OHV and FT-IR results indicate little or no hydroxyl formation for CanO.

During CanFAME epoxidation, the OHV (Fig. 5(b)) increases nearly linearly for all temperature ranges, reaching the maximum value of 36 mg KOH per g at 60 °C. This indicates that the epoxidized CanFAME can undergo for ring opening process more easily than their TAG (CanO) counterparts. However, the most dramatic increase in the OHV (up to 240 mg KOH per g) was observed for CanFFA (Fig. 5(c)).

It should also be noted that in all CanFFA samples these extensive ring opening reactions occur during the initial stages of epoxidation, similar to oxirane formation. It can be seen from Fig. 1(c) that the oxirane content of CanFFA decreases beyond about 5 hours, consist with the observed decrease in ring opening reactions.

The enhanced epoxy ring opening observed during epoxidation for CanFFA is largely due to the fatty acids themselves, which may become peroxy acids. This leads to the formation of many possible products, including esterified oligomeric forms (estolides) and hydroxylated or mixed hydroxyl/epoxy forms.^19,31–33

Some attempts have been made to model the acid induced epoxy ring opening reaction in plant oils. It was determined that epoxy ring opening is a first order reaction with respect to epoxide concentration and second order with respect to acid concentration.^10,14 Note that in these studies a significant excess of acid was used for ring opening. The reaction rate equation for epoxy ring opening using formic acid can be expressed as follows:^10,14

where RO refers to ring opening process; [EP] and [RCOOH] are the molar ratios of epoxide and formic acid, respectively; k_RO is the ring opening rate constant; n₁ and n₂ are the reaction orders with respect to epoxide and formic acid, respectively. If the reaction orders with respect to epoxy and acid concentration are n₁ = 1 and n₂ = 2, integration of the eqn (6) should yield the linear relationship of ln{[EP₀]/[EP]} vs. reaction time t, with a slope giving the values of k_RO. However, the epoxy ring opening reactions that occur during in situ epoxidation differ significantly from the post-epoxidation ring opening reactions reported in the literature^10,14 because oxirane formation and ring-opening can occur simultaneously. For this reason, the epoxy ring opening rate constants are beyond the scope of this work.

Although water is considered weak nucleophile in the presence of acid the aqueous medium of epoxidation still contributes to the epoxy ring opening, as shown in Scheme 2. The capability of the water to open an epoxy ring is proportional to the concentration of acid. Moreover, the hydroxy-side-chains generated by epoxy ring opening can also act as nucleophiles and open other epoxy rings forming oligomers of hydroxylated/epoxidized products. The generalized kinetic equation for such nucleophilic ring opening can be written as:

where [NU] is the molar concentration of the nucleophile and n₃ is the reaction order with respect to nucleophiles.

If free fatty acids are used in the process, the acid groups can participate in the epoxidation process via formation of peroxy fatty acids. Free fatty acids can also attack the epoxy ring to initiate homopolymerization of free fatty acids³⁹ to form estolides. Thus, when epoxidizing FFAs, the oligomerization of epoxidized products by other FFA (epoxidized or not) is unavoidable. The occurrence of this reaction depends on the molar concentration of the FFAs and of the epoxides.

Eqn (5) below gives the kinetics of the loss of epoxides due to oligomerization:

where OL refers to oligomerization via ring opening; [FA] is the molar ratio of FFAs, n₄ is the reaction order with respect to FFA; k_OL is the ring opening rate constant initiated by FFAs. Although, both k_RO and k_OL represent epoxide ring opening rates, they originate from different processes.

Thus, the complete kinetic equation for the epoxidation of plant oils and animal fats containing significant amounts of derivatives (such FFAs or FAMEs) using in situ formed performic acid can be re-written as:

The formation of oligomeric compounds during epoxidation can be monitored using size exclusion chromatography, as discussed below.

SEC of the CanO, CanFAME and CanFFA

Fig. 6 gives the size exclusion chromatography (SEC) results for (a) CanO, (b) CanFAME, and (c) CanFFA and their respective epoxides after 31 hours of epoxidation at 60 °C (the full SEC traces are given in ESI†).

Fig. 6 SEC chromatograms of the CanO, CanFAME and CanFFA (bottom) and their epoxides (top) after 31 hours of epoxidation (at 60 °C).

CanO elutes first in the SEC trace in Fig. 6 since it has the highest molecular weight (MW, ∼882 g mol⁻¹). The peak position of CanO (4.87 min) is slightly shifted to a higher MW/shorter retention time for ECanO due to the addition of oxygen. A similar behaviour is seen for CanFAME with average MW of about 295 g mol⁻¹ (RT ∼ 5.70 min). No major formation of oligomeric products (which give rise to earlier eluting peaks) is observed for CanO and CanFAME during the epoxidation process.

The CanFFA peak shows retention time of ∼5.70 min similar to that of CanFAME due to its similar molecular weight. There are no major changes in MW in the first 2 h of CanFFA epoxidation (see ESI†), but beyond this time the formation of high MW compounds can be seen as unresolved peaks at <5.5 min and which greatly increase in intensity as the epoxidation proceeds. This is due to the formation of higher MW compounds such as dimers (∼580 g mol⁻¹) and oligomers, including estolides (oligomeric esters of FA³⁹). Note that the saturated fatty acids also can take part in oligomerization reactions to some extent,^31–33 as demonstrated in the production of biobased lubricants.^39,40 Fatty acid trimers can also be seen in the SEC trace of ECanFFA, with an average MW of ∼880 g mol⁻¹ and retention time of 4.85 min. Some possible structures of these di- and tri-mers of the fatty acids (example is given for oleic acid) during epoxidation presented in Scheme 3.

Scheme 3 Fatty acid catalysed epoxy ring opening to form di- and tri-mers (estolides).

The oligomeric product ECanFFA is a light yellow waxy product, which tends to crystalize at room temperature. Overall, it is abundantly clear that the presence of the free fatty acids or use of free fatty acids for epoxidation purposes will definitely lead to a low net yield of epoxides. Thus, if the process objective is to prepare epoxidized derivatives, then the use of feedstocks containing free fatty acids should be avoided.

FTIR study

The structural changes which occurred during epoxidation were studied using ATR-FTIR spectroscopy. Fig. 7 compares the ATR-FTIR spectra for CanO, CanFAME and CanFFA, prior to epoxidation. Detailed descriptions of the FTIR spectra of plant oils and oil derivatives can be found in the literature.^41–47 An explanation of the interpretation indicated in Fig. 7 is given in ESI.†

Fig. 7 ATR-FTIR spectra of CanO, CanFAME and CanFFA.

Fig. 8 compares key areas of interest in the ATR-FTIR spectra corresponding to (a) double bond stretching, (b) epoxy group stretching and (c) hydroxyl group vibration for CanO, CanFAME and CanFFA at different epoxidation times (plots showing the full spectral range are given in ESI†). The arrows in Fig. 8 indicate the direction in which epoxidation times increase for each main feature.

Fig. 8 Stacked FTIR plots showing change of (a) double bonds, (b) epoxy groups and (c) hydroxyl numbers over epoxidation time for CanO, CanFAME and CanFFA, at 60 °C.

Fig. 8(a) shows that in the FTIR spectra of CanO and its derivatives the disappearance of double bonds is seen by the loss of the =C–H stretching band at 3006 cm⁻¹,⁴³ as epoxidation proceeds. The overall trend of disappearance of the double bonds over epoxidation process is similar for CanO and its derivatives. The transformation of double bonds into epoxy groups is confirmed in all products by the appearance of the band at 826 cm⁻¹ (Fig. 8(b)), which corresponds to the C–O–C stretching from the oxirane vibration.^13,45,48 The increase in the intensity of the epoxy moieties beyond 23 h is negligible for CanO and CanFAME.

Also, as mentioned above, the slow epoxidation of CanO at extended times is attributed²⁹ to the delayed and sequential epoxidation of polyunsaturated fatty acids. For CanFFA the epoxy band increases and is most clearly seen at reaction times of 4–23 h, beyond which it largely disappears indicating loss of the oxirane feature. The FTIR spectral region between 3100–3650 cm⁻¹ in Fig. 8(c) is for hydroxyl group vibration and demonstrates the formation of O–H groups due to the epoxy ring opening. The corresponding plots for CanO and its derivatives significantly differ.

There is little or no change over epoxidation time for CanO indicating no degradation of the oxirane ring.⁴⁹ After a period of epoxidation, for CanFAME a small amount of hydroxyl group formation is seen (Fig. 8(c)).

For CanFAME, at the initial stages of epoxidation, the concentration of epoxides, which might undergo ring opening reactions, is low. At longer epoxidation times the probability of ring opening increases partly due to higher epoxy group concentrations.^22,50 In addition, to some extent, as the concentration of H₂O₂ decreases, there may be more formic acid available to react with the epoxides.

However, this does not explain the different behaviour between CanFAME and CanO, which shows no ring opening. To our knowledge, such a difference in behaviour between epoxidized oils and their epoxidized FAME had not been reported before. This behaviour can be attributed to a variety of factors relating to the different chemical structures of TAG and FAME. These may lead to different interfacial properties with the aqueous phase, differences in viscosities, steric hindrance etc. This behaviour is consistent with the relatively low epoxide content in ECanFAME compared to ECanO, as demonstrated earlier. In addition, it was observed for CanFAME that the height of the oxirane band did not further change after 23 h (Fig. 8(b)) nor was there any further change in residual unsaturates (Fig. 8(a)).

For CanFFA, a significant loss of epoxy groups is observed after 23 hours of epoxidation, which continues to decrease until the end of epoxidation due to oxirane ring opening.^{22,49,21,51,52} The spectra in Fig. 8(c) indicate that the intensity of hydroxyl group stretching band (3100–3650 cm⁻¹) is considerably higher for epoxidized CanFFA compared to epoxidized CanO or CanFAME, which confirms the ring opening reactions.

In most literature,^{22,49,21,51,52} the degradation of epoxy group of FFAs is attributed to ring opening by short chain carboxylic acids (precursors for percarboxylic acids), while the possible contribution of the long chain fatty acids (input product of interest) in ring opening reactions were not described.

For CanFFA (Fig. 9), the band at 1707 cm⁻¹ is due to carbonyl (C=O) stretching in carboxylic acid groups of the fatty acids.

Fig. 9 Carbonyl stretching of carboxylic acids (1707 cm⁻¹) and C=O stretching of newly formed esters (1738 cm⁻¹) due to oligomerization for CanFFA over epoxidation time, at 60 °C.

As can be seen from the Fig. 9 the intensity at 1707 cm⁻¹ decreases over epoxidation time and this can be explained by oligomerization of fatty acids, which is known to occur (Fig. 6). At the same time, formation of a new peak centered around 1738 cm⁻¹ is observed (Fig. 9) and is attributed to the C=O stretching for newly formed esters⁴⁴ of fatty acids. The epoxidation of fatty acids via in situ generated percarboxylic acids is well described in the literature, especially for oleic acid.^21,51,52 These reports have shown a decrease in oxirane content over extended epoxidation, which was explained as due to the occurrence of the epoxy ring opening by short chain acids. Very limited information is available^39,53 on the formation of unsaturated (long chain) percarboxylic acids and their self-epoxidation with unsaturated fatty acids to form epoxy-esters (estolides).

Conclusions

The kinetics and products of the epoxidation process for CanO, CanFAME and CanFFA were evaluated under identical processing conditions. For CanO, negligible amounts of by-products were observed during epoxidation. In the epoxidation of CanFAME some hydroxylated moieties were developed over extended periods of epoxidation as a result of epoxy ring opening reactions. The epoxidation of CanFFA however, resulted in the formation of abundant by-products, such an epoxidized, hydroxylated and oligomerized fatty acids. CanFFA may act as an oxygen carrier to form peroxy fatty acids with H₂O₂ which can both accelerate the epoxidation process significantly whilst also leading to extensive epoxy ring opening. The latter results in the formation of estolides and other oligomers.

A full description of the kinetics of epoxidation must include epoxy ring opening and oligomerization processes where they occur. It would be desirable to apply a more detailed treatment to the empirical equations given above in order to fully model these interrelated reactions.

The results of the study may have significance in the production of epoxides from plant oils and animal fats, containing FFA and other derivatives.

Acknowledgements

The authors acknowledge support for this work provided by the Alberta Crop Industry Development Fund Ltd. (ACIDF) and Alberta Innovates BioSolutions.

Notes and references

1. M. N. Belgacem and A. Gandini, Monomers, Polymers and Composites from Renewable Resources, Elsevier, 2008.
2. F. S. Guner, Y. Yagci and A. T. Erciyes, Prog. Polym. Sci., 2006, 31, 633–670.
3. S. N. Khot, J. J. Lascala, E. Can, S. S. Morye, G. I. Williams, G. R. Palmese, S. H. Kusefoglu and R. P. Wool, J. Appl. Polym. Sci., 2001, 82, 703–723.
4. S. A. Madbouly, C. Zhang and M. R. Kessler, Bio-Based Plant Oil Polymers and Composites, Elsevier Inc., Oxford, UK, 2015.
5. N. Prileschajew, Ber. Dtsch. Chem. Ges., 1909, 42, 4811–4815.
6. H. Dodiuk and S. H. Goodman, Handbook of thermoset plastics, Elsevier, Oxford, UK, 3rd edn, 2014.
7. W. J. Zhang, P. P. Jiang, F. J. Wu and M. Chen, Adv. Mater. Res., 2012, 560–561, 184–190.
8. C. S. Cai, H. G. Dai, R. S. Chen, C. X. Su, X. Y. Xu, S. Zhang and L. T. Yang, Eur. J. Lipid Sci. Technol., 2008, 110, 341–346.
9. F. E. Okieimen, O. I. Bakare and C. O. Okieimen, Ind. Crops Prod., 2002, 15, 139–144.
10. L. H. Gan, S. H. Goh and K. S. Ooi, J. Am. Oil Chem. Soc., 1992, 69, 347–351.
11. Y. Zhang, X. Z. Chen, Y. Zhang and Y. X. Zhang, Macromol. Mater. Eng., 2001, 286, 443–448.
12. M. R. G. Klaas and S. Warwel, J. Am. Oil Chem. Soc., 1996, 73, 1453–1457.
13. T. Vlcek and Z. S. Petrovic, J. Am. Oil Chem. Soc., 2006, 73, 247–252.
14. F. A. Zaher, M. H. Elmallah and M. M. Elhefnawy, J. Am. Oil Chem. Soc., 1989, 66, 698–700.
15. M. H. Tavassoli-Kafrani, J. M. Curtis and F. R. van de Voort, Eur. J. Lipid Sci. Technol., 2015, 117, 65–72.
16. W. W. Christie, Lipid Analysis, The Oily Press, Bridgwater, UK, 3rd edn, 2003.
17. K. Ichihara and Y. Fukubayashi, J. Lipid Res., 2010, 51, 635–640.
18. B. Freedman, E. H. Pryde and T. L. Mounts, J. Am. Oil Chem. Soc., 1984, 61, 1638–1643.
19. ASTM D1652-11E1, Standard Test Method for Epoxy Content of Epoxy Resins, ASTM International, 2011, p. 4.
20. M. A. Aegerter, N. Leventis and M. M. Koebel, Aerogels Handbook, Springer, New York, 2011.
21. F. P. Greenspan and R. J. Gall, US Pat., 2,774,774, 1956.
22. M. C. Brasil, A. E. Gerbase, M. A. de Luca and J. R. Gregorio, J. Am. Oil Chem. Soc., 2007, 84, 289–295.
23. J. Salimon, B. M. Abdullah, R. M. Yusop and N. Salih, Chem. Cent. J., 2014, 8(16), 1–11.
24. M. Smith, Organic Chemistry: An Acid–Base Approach, CRC Press, Boca Raton, FL, 2010.
25. G. Strukul, Catalytic Oxidations with Hydrogen Peroxide as Oxidant, Kluwer Academic Publishers, The Netherlands, 1992.
26. L. R. Fisher, E. E. Mitchell and N. S. Parker, J. Food Sci., 1985, 50, 1201–1202.
27. A. G. Gaonkar, J. Am. Oil Chem. Soc., 1989, 66, 1090–1092.
28. A. T. Gros and R. O. Feuge, J. Am. Oil Chem. Soc., 1952, 29, 313–317.
29. S. T. Anuar, Y. Y. Zhao, S. M. Mugo and J. M. Curtis, J. Am. Oil Chem. Soc., 2012, 89, 1951–1960.
30. S. Ted Oyama, Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis, Elsevier, Amsterdam, The Netherlands, 2011.
31. T. A. Isbell, Grasas Aceites, 2011, 62, 8–20.
32. T. A. Isbell, M. R. Edgcomb and B. A. Lowery, Ind. Crops Prod., 2001, 13, 11–20.
33. T. A. Isbell, R. Kleiman and B. A. Plattner, J. Am. Oil Chem. Soc., 1994, 71, 169–174.
34. S. A. Arrhenius, Z. Physik. Chem., 1889, 4, 226–248.
35. P. De Filippis, M. Scarsella and N. Verdone, Ind. Eng. Chem. Res., 2009, 48, 1372–1375.
36. V. D. Ramos, D. Derouet and L. L. Y. Visconte, Polym. Test., 2003, 22, 889–897.
37. D. C. Allport, D. S. Gilbert and S. M. Outterside, MDI and TDI: Safety, Health and the Environment, A Source Book and Practical Guide, John Wiley & Sons, England, 2003.
38. E. Occhiello and F. Aguirre, US Pat., 2010/0151138 A1, 2010.
39. J. C. J. Bart, E. Gucciardi and S. Cavallaro, Biolubricants: Science and Technology, Woodhead Publishing, 2013.
40. S. C. Cermak and T. A. Isbell, Ind. Crops Prod., 2002, 16, 119–127.
41. M. Chen, W. Ding, Y. Kong and G. Diao, Langmuir, 2008, 24, 3471–3478.
42. P. Larkin, Infrared and Raman Spectroscopy, Principles and Spectral Interpretation, Elsevier, The Netherlands, 1st edn, 2011.
43. D. J. Last, J. J. Najera, C. J. Percival and A. B. Horn, Phys. Chem. Chem. Phys., 2009, 11, 8214–8225.
44. D. H. Lee and R. A. Condrate, J. Mater. Sci., 1999, 34, 139–146.
45. A. R. Mahendran, N. Aust, G. Wuzella and A. Kandelbauer, Macromol. Symp., 2012, 311, 18–27.
46. N. Vlachos, Y. Skopelitis, M. Psaroudaki, V. Konstantinidou, A. Chatzilazarou and E. Tegou, Anal. Chim. Acta, 2006, 573, 459–465.
47. N. Q. Wu, L. Fu, M. Su, M. Aslam, K. C. Wong and V. P. Dravid, Nano Lett., 2004, 4, 383–386.
48. T. Theophanides, Infrared Spectroscopy. Materials Science, Engineering and Technology, InTech, 2012.
49. A. Campanella and M. A. Baltanas, Eur. J. Lipid Sci. Technol., 2004, 106, 524–530.
50. T. I. Cooney, F. Cardona and T. Tran-Cong, Kinetics of in situ epoxidation of hemp oil under heterogenous reaction conditions: an overview with preliminary results, eddBE2011 Proceedings, Queensland University of Tecnology, 2011, pp. 106−111.
51. I. V. Kozhevnikov, G. P. Mulder, M. C. Steverink-de Zoete and M. G. Oostwal, J. Mol. Catal. A: Chem., 1998, 134, 223–228.
52. D. Swern, G. N. Billen, T. W. Findley and J. T. Scanlan, J. Am. Chem. Soc., 1945, 67, 1786–1789.
53. S. Warwel and M. R. G. Klaas, J. Mol. Catal. B: Enzym., 1995, 1, 29–35.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra17732h

---