Heterogeneous alkaline earth metal–transition metal bimetallic catalysts for synthesis of biodiesel from low grade unrefined feedstock

Abstract

A bimetallic alkaline earth metal–transition metal oxide, synthesized through a method of direct low-temperature decomposition of the bimetallic complex, is reported for the synthesis of biodiesel. Due to the high phase purity of the Ca/Fe catalytic system and its catalytic stability and robustness, the Ca/Fe catalyst was selected for further investigation. A transesterification conversion of 99.5% could be achieved in 1 h under the optimal conditions: feedstock to methanol, 1 : 20; catalyst loading, 6 wt%; temperature, 120 °C. ANOVA tests suggested that the reaction temperature was discerned as the most prominent factor which contributed 82.84% to the overall catalytic feedstock conversion. In addition, the Ca/Fe catalytic system demonstrated a high FFA tolerance of 2 wt% and a water tolerance of 1 wt% with remarkable catalytic activity in one-step biodiesel synthesis.

---

1. Introduction

Limited fossil fuel reserves and the associated environmental pollution have prompted scientists to search for a sustainable liquid fuel to support our future needs. Biodiesel is found to be a promising renewable energy source because of its non-toxicity, carbon neutrality with low sulphur content and having similar physical and chemical properties compared to conventional diesel fuel. It may reduce harmful emissions such as NO_x, SO_x, CO, CO₂, unburnt hydrocarbons and particulates.^1–4 Transesterification involves the breaking down of triglycerides with alcohol.^1,2,4 Homogeneous catalysts for traditional biodiesel synthesis can be divided into two main groups, strong bases (e.g. NaOH or KOH) and strong acids (e.g. H₂SO₄ or HCl).^5,6 However, large quantities of fresh water are required for final product purification.

The use of different heterogeneous base catalysts for catalytic transesterification, such as alkaline earth metal oxides and hydroxides, has been widely reported in the literature.^7–11 The catalytic activities of alkaline earth metal oxides, including magnesium oxide (MgO), calcium oxide (CaO), strontium oxide (SrO) and barium oxide (BaO), were closely related to their surface basic strength. The basicity increased with atomic number, which increased with increasing cationic size and decreasing polarizing power. This destabilized the O²⁻ anions.¹² As a result, the catalytic activities of the four catalysts demonstrated the sequence MgO < CaO < SrO < BaO.¹³ BaO is the strongest basic oxide but it is soluble in methanol and exhibits a high toxicity.⁹ Although SrO is insoluble in methanol, it would be deactivated by atmospheric moisture and carbon dioxide (CO₂).¹³

It is believed that CaO is one of the reactive single oxides and has been widely employed for transesterification reactions because of its high catalytic activity and low preparation cost. However, the use of alkaline metal oxides is often confronted with low durability and serious leaching problems due to metal dissolution. This accounts for the loss of the active species from metal oxide catalysts, with the catalytic activity declining significantly upon regeneration.^14–16 Furthermore, the adsorption of water molecules and CO₂ on the active sites of the alkaline earth metal oxides would suppress the esterification and transesterification reactions.^16,17

In order to improve the dissolution of the active metal, various alkali metal oxides supported on alumina,^18–20 CaO supported on silica²¹ and CaO mixed with zinc oxide (ZnO)^22,23 have been reported as heterogeneous catalysts for transesterification. However, it was found that leaching was not improved as these catalysts demonstrated poor reusability and robustness. More recently, mixed metal oxides such as hydrotalcites²⁴ and Ca–La mixed oxides have been explored.²⁵ In our previous work, we studied the transesterification of Camelina sativa oil with methanol to synthesize biodiesel using Na_0.1Ca_0.9TiO₃ nanorods as a heterogeneous catalyst, synthesized through alkali hydrothermal synthesis.²⁶ The leaching was not improved due to the existence of single oxides.¹⁶ Furthermore, low grade alcohols and unrefined feedstocks containing high degrees of water cause the hydrolysis of triglyceride to form diglyceride and free fatty acid (FFA) while the high levels of FFA contamination will subsequently induce saponification and deactivate the base catalysts.^27,28

Owing to the high toxicity of Ba and the low stability of Sr in air, this study focused on Mg and Ca for biodiesel synthesis from low grade feedstocks with high levels of FFA contamination. The most abundant transition metals, such as iron (Fe), manganese (Mn) and chromium (Cr), were proposed to act as stabilizers to retain the alkaline earth metal, Mg and Ca, in a bimetallic catalytic system and further enhance its catalytic stability and robustness. Srebrodolskite type materials are heterogeneous catalysts for use in various applications. This material with a general formula of A₂B₂O₅ consists of divalent cations and trivalent cations. The conventional synthesis of srebrodolskite Ca₂Fe₂O₅ involves the mixing and grinding of CaO and Fe₂O₃ followed by a high temperature calcination at over 1000 °C for 3 h in air.²⁹ These harsh synthetic conditions would surely lead to the formation of highly aggregated particles which decreases the resultant surface area-to-volume ratio. Co-precipitation is another commonly employed preparation method which induces phase separation of metal oxide instead of forming a final single phase mixed metal oxide. Xue and co-workers have reported the synthesis of a Ca₂Fe₂O₄–Ca₂Fe₂O₅-based catalyst for biodiesel synthesis, however, a mixed phase catalyst was observed.³⁰

Herein, a series of alkaline earth metal–transition metal bimetallic oxide catalysts for transesterification with methanol have been investigated. These mixed oxide catalysts have been synthesized through low-temperature decomposition of the corresponding bimetallic complex, in which the bimetallic complex is directly calcinated under atmospheric conditions to yield the desired catalyst. Switching of the inorganic metal salts to organometallic complexes is found to significantly decrease the calcination temperature which prevents the formation of highly aggregated catalysts. The FFA and water tolerance of the catalytic system were also investigated. Meanwhile, the reaction conditions for transesterification were optimized. The optimization is very important to maximize the biodiesel yield and to minimize the production cost. A stepwise approach has been extensively employed, however, it is relatively time consuming and difficult to determine the optimal conditions as some of the reaction conditions simultaneously affect the biodiesel yield. Taguchi analysis is a cost-effective and time saving alternative approach for optimization in which the analysis can be divided into orthogonal array experiments, signal-to-noise (S/N) ratio analysis and range analysis.^31–35 Analysis of variance (ANOVA) was also introduced as a statistical model to evaluate whether the factors were prominent using an F-test under consideration of the experimental error.

2. Materials and method

2.1. Materials

Refined food grade canola oil was obtained from a local store in Hong Kong. Crude rice bran oil was obtained from a local store in China. Crude flaxseed and rapeseed oil were produced from an in-house cold-pressed oil extractor using flaxseed and rapeseed obtained from a local store in Hong Kong. Waste cooking oil was collected from a local restaurant in Hong Kong. Calcium nitrate tetrahydrate (Ca(NO₃)₂·4H₂O, 99%) was purchased in laboratory reagent grade from BDH Chemical Ltd. Iron(III) nitrate nonahydrate (Fe(NO₃)₃·9H₂O, >99%), magnesium nitrate hexahydrate (Mg(NO₃)₂·6H₂O, 99%) and manganese(II) nitrate tetrahydrate (Mn(NO₃)₂·4H₂O) were supplied by Acros. Ethylenediaminetetraacetic acid (EDTA, 99%) was obtained from Research Chemical Limited. Chromium(III) nitrate nonahydrate (Cr(NO₃)₃·9H₂O, 99%), polyvinylpyrrolidone (PVP, average M_w 40 000), methyl yellow, neutral red, bromothymol blue, phenolphthalein, nile blue, tropaeolin O and 2,4-dinitraniline were collected from Sigma Aldrich. Oleic acid (C₁₈H₃₄O₂, 99.9%) was obtained in laboratory reagent grade from Fisher Chemical. Methanol (CH₃OH, 99.8%) and diethyl ether (C₄H₁₀O, 99.5%) were obtained in ACS reagent grade. Aqueous ammonia solution (NH₄OH, 28.0–30.0 wt%) and potassium hydroxide (KOH) were purchased from UNI-CHEM.

2.2. Catalytic preparation and characterization

2.2.1. Catalyst preparation. Mixed oxides were synthesized through a simple direct decomposition of their corresponding bimetallic metal complexes. Ethylenediaminetetraacetic acid (EDTA, 1 mol) was dissolved in milli-Q water followed by the addition of four equivalents of ammonia water (4 mol). Aqueous polyvinylpyrrolidone solution was added into the ammonium EDTA solution in a drop-wise manner. The solution was stirred vigorously at 60 °C for 15 min. The mixed metal solution (1 : 1 mol/mol, 10 mL) was prepared in milli-Q water and was added drop-wisely into the reaction mixture. The solution was stirred for a further 30 min and finally was dried at 110 °C overnight. The dried mixed metal EDTA complex precursor was calcinated at 600 °C for 5 h in air.

2.2.2. Catalytic characterization. The size and morphology of the catalyst were characterized using a Hitachi S-4800 field emission scanning electron microscope (SEM, 5 kV) equipped with energy dispersive spectrometry (EDX) with Horiba EMAX EDS detectors. Powder X-ray diffraction (XRD) patterns were obtained using a Rigaku SmartLab X-ray diffractometer using CuKα (λ = 1.54056, 45 kV, 200 mA) radiation with 2θ ranging from 10° to 80° with a step size of 0.02° at a scan rate of 5° min⁻¹. Hammett indicator analysis was applied to elucidate the surface basic strength of the catalyst. The catalyst (5 mg) was immersed in methanolic Hammett indicator solution (1 mL, 50 μM) under ultrasonic irradiation and was allowed to stand for 1 h to achieve equilibrium. The Hammett indicators used for analysis were methyl yellow (H̲ = 3.3), neutral red (H̲ = 6.8), bromothymol blue (H̲ = 7.2), phenolphthalein (H̲ = 9.7), nile blue (H̲ = 10.2), tropaeolin O (H̲ = 11.0) and 2,4-dinitraniline (H̲ = 15.0). Surface area analysis was performed using a Quantachrome Autosorb iQ gas sorption analyzer. The sample was outgassed at 0.03 torr with a 2 °C min⁻¹ ramp to 130 °C and held at 130 °C for 20 h. The sample was then held under vacuum until the analysis was performed using N₂ at 77 K (P/P₀ range 1 × 10⁻⁵ to 0.995).

2.3. Feedstock evaluation

With reference to ASTM D664, a standard titration method with a titrant of standard KOH solution was applied in determination of the acidity (AD) and acid value (AV) for all of the feedstocks, while the quantity of water was estimated through Karl Fischer titration based on ASTM D4377 using an automated V20 Volumetric Karl-Fischer Titrator. The AD and AV of the various feedstocks range from 0.11 to 2.00 wt% and from 0.22 to 3.98 mg_KOH g⁻¹ respectively, as summarized in Table 1.

Table 1 Acidity, acid value and water content for each feedstock sample

Feedstock	AD (wt%)	AV (mgKOH g^−1)	Water (wt%)
Refined food grade canola oil	0.11	0.22	0.13
Crude flaxseed oil	1.16	2.32	0.12
Crude rapeseed oil	1.65	3.28	0.13
Crude rice bran oil	2.00	3.98	0.11
Waste cooking oil	0.65	1.29	0.12

---

Fatty acid profile of each feedstock was analyzed on a HEWLETT 5890 SERIES II Gas Chromatograph system equipped with a flame-ionization detector and a capillary column (DURABOND-WAX, 30 m × 0.250 mm, film thickness of 0.5 μm, part no.: 122-7033). The operating conditions for the analysis were as follows: the injector temperature was 280 °C; the detector temperature was 280 °C; the temperature program was started at 180 °C at a rate of 2 °C min⁻¹ to 240 °C. All of the feedstocks contained fatty acid chain lengths of 16 to 22, which are desirable for biodiesel synthesis, as summarized in Table 2.

Table 2 Fatty acid composition for each feedstock sample

Feedstock	Fatty acid composition^a (%)
	C16 : 0	C18 : 0	C18 : 1	C18 : 2	C18 : 3	C20 : 1	C22 : 1
a C16 : 0 = palmitic acid, C18 : 0 = stearic acid, C18 : 1 = oleic acid, C18 : 2 = linoleic acid, C18 : 3 = linolenic acid, C20 : 1 = eicosenoic acid and C22 : 1 = erucic acid.
Refined food grade canola oil	5.00	2.68	63.06	22.82	6.44	—	—
Crude flaxseed oil	6.06	2.53	19.19	24.34	47.88	—	—
Crude rapeseed oil	3.91	1.41	17.05	12.94	9.03	14.54	41.12
Crude rice bran oil	19.33	2.79	43.66	34.22	—	—	—
Waste cooking oil	21.88	5.15	56.73	14.48	0.86	0.90	—

---

2.4. Catalytic study

All catalytic reactions were conducted in a stirred batch reactor containing methanol, catalyst and feedstock sample (0.46 g) with different ratios as specified in the detailed results and discussion section. The reaction mixture was heated with constant vigorous stirring at 750 rpm at a specific temperature for a predesigned time. The reaction mixture was separated from the catalyst when the reaction was completed. The feedstock conversion was determined using ¹H-NMR spectroscopy (Bruker, 400 MHz) with CDCl₃. The feedstock conversion was calculated according to the ratio of the integral for the signal of the –OCH₃ in the methyl ester compared to that of the α-CH₂ in the triglyceride and methyl ester as follows,^36–39

The reaction was allowed to cool down to room temperature after completion of the catalytic reaction. The used catalyst was isolated out and was dried at room temperature for the next cycle of the catalytic reaction without any washing steps. The same amount of fresh feedstock and methanol were added to the recycled catalyst and the catalytic study was performed for several cycles under the same conditions.

The quantity of active metals leached into the biodiesel layer from the catalyst was determined through inductively coupled plasma optical emission spectroscopy (ICP-OES) using an Agilent Technologies 700 Series ICP-OES. A sample of 0.1 mL was pre-digested in a mixture of concentrated hydrochloric acid and concentrated nitric acid (3 : 1 v/v) at a temperature of 60 °C for 2 h.

3. Results and discussion

3.1 Characterization of mixed oxide catalysts

The morphologies of all catalytic systems were characterized using SEM as shown in Fig. 1. The Mg/Fe and Ca/Fe catalytic systems are found to adopt nanoparticle morphologies. A TEM micrograph of the Ca/Fe catalytic system also shows a nanoparticle morphology with an average particle size of 110.8 ± 3.5 nm. The other catalytic systems, however, exhibit irregular nanoparticles with different sizes. The EDX analysis was performed to determine the elemental compositions and their corresponding ratios for the six catalytic systems. The results are tabulated in Table 3. It is found that the ratio for all six catalytic systems is found to be nearly 1.

Fig. 1 SEM micrographs of the (a) Mg/Fe, (b) Ca/Fe, (c) Mg/Mn, (d) Ca/Mn, (e) Mg/Cr and (f) Ca/Cr catalytic systems. (g) TEM and (h) HR-TEM of the Ca/Fe nanoparticle catalyst.

Table 3 Characterization and the catalytic performances of a series of catalysts

Catalytic system	Atomic ratio in EDX analysis	Surface basicity
Mg/Fe	1.1	7.2 < H̲ < 9.7
Ca/Fe	1.0	10.2 < H̲ < 11.0
Mg/Mn	1.0	7.2 < H̲ < 9.7
Ca/Mn	0.9	9.7 < H̲ < 10.2
Mg/Cr	0.9	7.2 < H̲ < 9.7
Ca/Cr	0.9	7.2 < H̲ < 9.7
CaO	—	10.2 < H̲ < 11.0

---

The crystal structures of all of the catalysts were characterized using XRD as depicted in Fig. 2. A high phase purity in the XRD pattern is observed for Ca/Fe which matches to the orthorhombic phase Ca₂Fe₂O₅ (ICDD: 01-076-8615). For the Mg/Fe catalytic system, it is suspected that amorphous MgO is present as only cubic phase MgFe₂O₄ (ICDD: 01-088-1936) was found in the XRD analysis with a theoretical Mg-to-Fe ratio of 0.5, which deviated from the atomic ratio found in the EDX analysis. Mixed XRD patterns obtained in the Mg/Mn catalytic system are attributed to tetragonal phase MgMn₂O₄ (ICDD: 01-074-2023), hexagonal phase MgMnO₃ (ICDD: 024-0736) and single orthorhombic Mn₃O₄ (ICDD: 01-072-7943). The diffraction peaks in the Ca/Mn catalytic system show agreement with a mixed phase of orthorhombic CaMn₂O₄ (ICDD: 01-072-6399) and monoclinic Ca₂Mn₃O₈ (ICDD: 034-0469). The XRD patterns of the Mg/Cr and Ca/Cr catalytic systems demonstrate cubic MgCr₂O₄ (ICDD: 010-0351) and tetragonal CaCrO₄ (ICDD: 008-0458) respectively, however, single Cr₂O₃ is found in both catalytic systems. Single phase mixed oxides are obtained only in the Mg/Fe and Ca/Fe catalytic systems due to good matching of the Mg(II) ion or Ca(II) ion to the Fe(III) ion. In contrast, mixed phases of mixed metal oxides are observed in the other catalytic systems, probably due to size mismatching of the alkaline earth metal ion and transition metal ion. It is well known that both Mn and Cr have multiple stable oxidation states. Under atmospheric conditions at 600 °C, Mn³⁺ and Cr³⁺ should be oxidized to higher oxidation states. Therefore, a mixture of single oxide and mixed oxide was obtained for Mn or Cr containing catalytic systems as suggested by the XRD analysis.

Fig. 2 XRD spectra of the (a) Mg/Fe, (b) Ca/Fe, (c) Mg/Mn, (d) Ca/Mn, (e) Mg/Cr and (f) Ca/Cr catalytic systems.

It has been reported that the basic sites on the surface of heterogeneous catalysts are the active sites for catalytic transesterification.^40,41 Therefore, it is important to investigate the correlation between the surface basic strength and the catalytic activity of heterogeneous catalysts. As shown in Table 3, all of the catalytic systems demonstrate a basic property on the catalyst surface as detected through Hammett indicator analysis. The Ca/Fe catalytic system and single CaO exhibit the highest surface basicity with 10.2 < H̲ < 11.0, which could be considered as strong solid bases while the Ca/Mn catalytic system was considered a mild solid base catalyst with 9.7 < H̲ < 10.2. The Mg/Fe, Mg/Mn, Mg/Cr and Ca/Cr catalytic systems were considered as neutral to weak basic catalysts with a low surface basicity of 7.2 < H̲ < 9.7. The surface basicity of the Ca/Fe catalytic system is much higher than that of Ca₂Fe₂O₅ synthesized through high temperature calcination by Kawashima, which is only in the range of 7.2 to 9.3.⁴² Other catalytic systems, such as M/Fe, Mg/Mn, Mg/Cr and Ca/Cr, gave relatively low surface basicity in the range of 7.2 to 9.7.

The BET specific surface area of the Ca/Fe catalytic system was found to be 32.7 m² g⁻¹, as determined from the BET adsorption/desorption isotherm, which is 46 times and 4 times the surface areas reported by Kawashima⁴² and Xue³⁰ respectively.

3.2 Catalytic activities

The catalytic performances of all of the catalytic systems were investigated for transesterification of refined food grade canola oil with the results shown in Table 4. It can be observed that the catalytic activities demonstrated a positive correlation with the surface basicity. Both the Ca/Fe and CaO catalytic systems, having higher surface basicity, gave faster reaction kinetics with conversions of 98.7% and 99.6% in 1 hour respectively. The Ca/Mn catalytic system showed slower kinetics with a conversion of 97.8% in 4 h due to the lower surface basicity. Other catalytic systems, however, still gave moderate catalytic conversions due to their relatively low surface basicities. Biodiesel synthesis with the Ca/Fe catalytic system was at a higher reaction temperature but with a significantly shorter reaction time compared to the literature report by Zhang.⁴³ Furthermore, a simple approach of direct reuse of the catalyst without further washing steps for the used catalyst was employed to investigate the reusability of the catalytic systems.

Table 4 Catalytic performances of the mixed oxide catalysts

Catalytic system	Reaction time (h)	Catalytic conversion^a (%)
		1	2	3	4	5
a Reaction conditions: feedstock-to-methanol molar ratio of 1 : 20, catalyst loading of 6 wt%, and reaction temperature of 120 °C.
Mg/Fe	4	47.3	37.8	25.3	—	—
Ca/Fe	1	98.7	89.0	81.9	71.7	37.5
Mg/Mn	4	71.1	33.9	—	—	—
Ca/Mn	4	97.8	32.4	—	—	—
Mg/Cr	4	79.1	12.9	—	—	—
Ca/Cr	4	80.3	12.5	—	—	—
CaO	1	99.6	—	—	—	—

---

The Ca/Fe catalyst was the only catalytic system which could be reused for over 4 cycles. It demonstrates that Fe, found in most abundance in the bimetallic Ca₂Fe₂O₅ catalytic system, stabilizes the Ca and further enhances its catalytic stability and robustness. In contrast, CaO was completely dissolved at the end of the first cycle which implied that CaO underwent homogeneous catalysis rather than heterogeneous catalysis and cannot be regenerated for the next cycle.

The EDX analysis of the used Ca/Fe catalytic system suggests that the atomic Ca-to-Fe ratio is equal to 0.69 which indicates a significant loss of Ca after the fifth catalysis cycle. In order to further confirm the loss of Ca and Fe from the catalyst, the amounts of metal leaching out were analyzed through ICP-OES of the reaction mixture. As depicted in Table 5, the amount of Ca and Fe leached to the biodiesel layer ranged from 310.5 to 1489.5 mg L⁻¹ and 26.9 to 532.7 mg L⁻¹ which aligns with the decrease in Ca as suggested in the EDX analysis. It is found that the extent of Ca leached from the bimetallic Ca/Fe system was smaller than that from CaO^14,15 as it was totally dissolved during the catalysis.

Table 5 Leaching study of the Ca/Fe catalytic system for each cycle of transesterification

Cycle	Concentration of metal found in the biodiesel layer (mg L^−1)
	Ca	Fe
1	1489.5 ± 0.3	532.7 ± 0.3
2	299.8 ± 0.5	114.9 ± 0.3
3	310.5 ± 0.5	80.3 ± 0.2
4	320.3 ± 0.7	112.1 ± 0.3
5	360.2 ± 0.7	26.9 ± 0.2

---

Apart from the extension of the catalytic activity, magnetic recycling of the used Ca/Fe catalysts was also studied. After ultrasonic treatment of the reaction mixture, a strong rare earth based magnet was placed on the side wall of the reaction flask for 30 min. The Ca/Fe catalyst diffused to one side with a strong magnet (see ESI Fig. S1†). This concludes that the Ca/Fe catalytic system is a magnetic catalyst which favours magnetic separation and recovery for the next catalytic cycle. A similar observation was reported by Zhang.⁴³

3.3 Optimization for Ca₂Fe₂O₅ catalyzed biodiesel synthesis

Since the Ca₂Fe₂O₅ catalytic system gave an excellent robustness towards biodiesel synthesis, its reaction conditions, including the feedstock-to-methanol molar ratio (factor A), catalyst loading (factor B) and reaction temperature (factor C), were then optimized using Taguchi analysis and ANOVA (see ESI Table S1†). Detailed calculations were according to the literature report by Wu.³³ As tabulated in Table 6, the average feedstock conversion and signal-to-noise (S/N) ratios ranged from 2.2% to 78.0% and from 6.71 to 37.84 respectively. The optimal conditions for each factor, as shown in Fig. 3, are clearly distinguished to be a combination of A₂B₃C₃ as follows: feedstock-to-methanol molar ratio of 1 : 20 (26.5), catalyst loading of 6 wt% (27.3) and a reaction temperature of 120 °C (33.5). The reaction temperature (factor C) gives the largest range value which indicates that the feedstock conversion changes significantly with the change of the reaction temperature. The Ca₂Fe₂O₅ catalyzed transesterification under the optimal reaction conditions, as shown in the ESI (Fig. S2†), suggested that a remarkable conversion of 99.5% was achieved after 1 h and this catalytic system is significant as no observable conversion was obtained in the background reaction.

Table 6 Orthogonal array experimental design in a OA₉ matrix

Entry	Factor			Conversion (%)			Average Yi (%)	Standard deviation	S/N ratio
	Feedstock : MeOH molar ratio A	Catalyst loading B (wt%)	Reaction temperature C (°C)	y1	y2	y3
1	1 : 10	2	80	2.2	2.2	2.1	2.2	0.06	6.71
2	1 : 10	4	100	20.4	20.9	21.1	20.8	0.36	26.36
3	1 : 10	6	120	40.5	40.7	41.4	40.9	0.47	32.23
4	1 : 20	2	100	14.8	14.1	14.6	14.5	0.36	23.22
5	1 : 20	4	120	77.4	78.4	78.2	78.0	0.53	37.84
6	1 : 20	6	80	8.4	8.3	8.0	8.2	0.21	18.31
7	1 : 30	2	120	33.9	33.0	32.8	33.2	0.59	30.43
8	1 : 30	4	80	5.6	5.3	5.3	5.4	0.17	14.64
9	1 : 30	6	100	37.5	36.5	36.0	36.7	0.76	31.28

---

Fig. 3 Relationship between (a) feedstock-to-methanol molar ratio, (b) catalyst loading, (c) reaction temperature and their corresponding mean S_ji value. (d) Range values of each factor.

For an increase of the feedstock-to-methanol molar ratio from 1 : 10 to 1 : 30, the feedstock conversion steadily increases and then slightly decreases beyond the optimal ratio (1 : 20). As the transesterification reaction involves three successive reversible reactions, a higher feedstock-to-methanol molar ratio is usually employed to drive the equilibrium to the product side. Other ratios beyond the optimal (1 : 20) do not increase the biodiesel yield as large amounts of methanol decrease the concentration of substrate in the reaction mixture.^44–46 It is clear that there is an increase of catalytic conversion with an increase of catalyst loading from 2 wt% to 6 wt%. This may due to the availability of more active sites on the catalyst surface to generate alkoxide anions and enhance the catalytic transesterification.^46,47 Further increasing the catalyst loading (6 wt%) did not enhance the catalytic conversion significantly as the amount of active sites were already saturated. Also, it is likely to be due to the mass transfer inefficiency of the catalyst in the three separated phase system and the blockage of active sites by methyl ester products.^48,49 The feedstock conversion is found to be enhanced when the reaction temperature is increased from 80 °C to 120 °C. The increase in temperature to 120 °C would increase the kinetic energy of the substrates which leads to a higher diffusion rate and increases the collisions between the Ca₂Fe₂O₅ catalyst and the substrates for faster catalytic transesterification.⁵⁰ Furthermore, higher temperature would reduce the viscosity of the feedstock sample.⁵¹

At the 90% confidence level, the ANOVA results (Table 7) are F_A (5.65) < F_α; F_B (13.90) > F_α and F_C (99.20) > F_α. The catalyst loading (factor B) and reaction temperature (factor C) are prominent factors affecting the biodiesel yield. Furthermore, the reaction temperature (factor C) gives the highest percent contribution of 82.84% to the final catalytic activity, followed by catalyst loading (factor B, 11.61%) and feedstock-to-methanol molar ratio (factor A, 4.72%). The experimental error contributes 0.83% which implies that all of the experimental results collected in this optimization process are reliable and no important factors are omitted.

Table 7 Results of ANOVA

Factor	SSj	dFj	Vj	Fj	F0.1 (2,2) = 9.00	Pj (%)
A	36.60	2	18.30	5.65	<	4.72
B	90.06	2	45.03	13.90	>	11.61
C	642.81	2	321.41	99.20	>	82.84
Error	6.47	2	3.24	—	—	0.83
T	775.94	8	—	—	—	100.00

---

Kawashima et al.⁴² reported the preparation of Ca₂Fe₂O₅ through a conventional solid-state reaction involving mixing metal oxides as a catalyst for transesterification at methanol refluxing temperature with 92% yield. Xue and co-workers have also reported the synthesis of a Ca₂Fe₂O₄–Ca₂Fe₂O₅ based catalyst through co-precipitation for biodiesel synthesis obtaining an 85.4% conversion.³⁰ However, the high biodiesel yield obtained under mild conditions is probably due to the presence of single CaO and CaCO₃ respectively. In comparison, this novel preparation of a high phase purity mesoporous Ca₂Fe₂O₅ catalytic system through the direct decomposition of a bimetallic Ca/Fe EDTA complex minimizes the formation of unwanted single oxides and carbonates and can still catalyze transesterification with faster kinetics at 120 °C.

3.4 Effect of reaction temperature and the reaction kinetics

The reaction kinetics of the Ca/Fe system catalyzed transesterification were further investigated. The reaction temperature is found to be the most significant factor in range analysis, which affects the rate of reaction and the feedstock conversion as the intrinsic rate constant is a strongly temperature-dependent function. Due to the excess of methanol, the overall reaction most likely followed a pseudo-first-order reaction kinetics model. The related equations and the calculation have been proposed in the literature^52,53 and can be found in the ESI.† The activation energy (E_a) for the Ca/Fe catalytic system catalyzed transesterification is found to be 45.52 kJ mol⁻¹.

3.5 Proposed mechanism of Ca/Fe catalyzed transesterification

The proposed mechanism of transesterification catalyzed by the Ca/Fe catalytic system is proposed in Scheme 1. The transesterification reaction takes place on the surface of the Ca/Fe catalytic system. In the first step of the mechanism, O²⁻ on the surface extracts H⁺ from a methanol molecule to generate a methoxide anion, which would then nucleophilically attack the carbonyl carbon of a triglyceride molecule to form a tetrahedral intermediate. Subsequently, rearrangement of the tetrahedral intermediate yields the biodiesel product and diglyceride anion. In the last step, H⁺ on the catalyst surface attacks the diglyceride anion with regeneration of the Ca/Fe catalyst.

Scheme 1 Proposed mechanism of the Ca/Fe catalytic system catalyzed transesterification.

3.6 Study of the FFA tolerance

Low grade alcohols and unrefined feedstocks often possess significant amounts of water and FFAs which lead to hydrolysis and saponification. Most of the heterogeneous catalysts are active only when FFAs are significantly removed in a pre-treatment process. Therefore, the design and modification of the solid catalyst applied to one-step simultaneous esterification and transesterification is gaining more and more attention.^54–56 The Ca₂Fe₂O₅ catalyzed transesterification from various refined and unrefined feedstocks serves as a replacement for refined plant oil. As summarized in Table 8, the Ca₂Fe₂O₅ catalytic system performed well in mediocre FFA contaminated feedstocks.

Table 8 Ca₂Fe₂O₅ catalyzed biodiesel synthesis with different refined and unrefined feedstocks

Entry	Feedstock	Conversion^a (%)
a Reaction conditions: feedstock-to-methanol molar ratio of 1 : 20, catalyst loading of 6 wt%, reaction temperature of 120 °C and reaction time of 1 h.
1	Refined food grade canola oil	98.7
2	Crude flaxseed oil	96.6
3	Crude rapeseed oil	94.8
4	Crude rice bran oil	92.6
5	Waste cooking oil	97.2

---

Further investigation of the FFA tolerance of the Ca/Fe catalytic system was studied as illustrated in Fig. 4a. The refined food grade canola oil and oleic acid were used as a model feedstock and FFA respectively in this study. A series of stimulated feedstock samples with different amounts of oleic acid (Table S2 in the ESI†) mixed with refined food canola oil were used for one-step simultaneous esterification and transesterification. It was found that the catalyst tolerated an oleic acid content of 2.00 wt% with a remarkable conversion of 92.5%. The Ca/Fe catalytic system demonstrated a high robustness towards highly FFA contaminated feedstock. Ca₂Fe₂O₅ showed much higher FFA tolerance than NaOH, in which the NaOH system would induce saponification with a higher FFA content (>1 wt%).^28,57 For comparison, the same reaction using the corresponding amount of CaO was also tested. It was found that CaO could still catalyze the transesterification but it was totally dissolved at the end of the reaction, which pointed out that CaO underwent homogeneous catalysis rather than heterogeneous catalysis.

Fig. 4 Effect of (a) oleic acid and (b) water content on transesterification. Reaction conditions: a feedstock-to-methanol molar ratio of 1 : 20, catalyst loading of 6 wt%, reaction temperature of 120 °C and reaction time of 1 h.

3.7 Study of the water tolerance

The water tolerance of the Ca/Fe catalytic system was examined, the results of which are depicted in Fig. 4b. A series of simulated feedstock samples containing different amounts of water (Table S3 in the ESI†) mixed with refined food canola oil were used for catalytic transesterification. The catalyst could withstand a water content of 1.00 wt% with a noteworthy conversion of 91.6% in which the catalyst can even be applied in low grade methanol. The Ca/Fe catalyst had a higher water tolerance than the existing homogeneous base catalyst. It is comparable to our previous work on sulfated zirconium oxide which showed even higher tolerance.^58,59

4. Conclusions

A series of alkaline earth metal–transition metal catalytic systems were explored for transesterification with methanol. The Ca/Fe catalytic system, due to the stabilization from Fe, gave the highest catalytic stability and robustness towards refined and unrefined feedstocks. Through Taguchi analysis, a conversion of 99.5% was achieved under optimal reaction conditions and ANOVA tests suggested that the reaction temperature is the prominent factor which contributed 82.84% to the overall conversion. The Ca₂Fe₂O₅ catalytic system was applied to a one-step simultaneous esterification and transesterification and tolerated a feedstock containing a significant amount of FFA (2 wt%) and water (1 wt%).

Acknowledgements

The authors grateful acknowledge the financial support from The Hong Kong Research Grants Council, and The Hong Kong Polytechnic University. T. L. Kwong acknowledges the receipt of a postgraduate studentship administrated by The Hong Kong Polytechnic University.

Notes and references

1. F. Ma and M. A. Hanna, Bioresour. Technol., 1999, 70, 1–15.
2. A. Demirbas, Prog. Energy Combust. Sci., 2005, 31, 466–487.
3. R. Jothiramalingam and M. K. Wang, Ind. Eng. Chem. Res., 2009, 48, 6162–6172.
4. A. P. Vyas, J. L. Verma and N. Subrahmanyam, Fuel, 2010, 89, 1–9.
5. J. M. Marchetti, V. U. Miguel and A. F. Errazu, Renewable Sustainable Energy Rev., 2007, 11, 1300–1311.
6. D. Lee, Y. Park and K. Lee, Catal. Surv. Asia, 2009, 13, 63–77.
7. T. F. Dossin, M. Reyniers and G. B. Marin, Appl. Catal., B, 2006, 62, 35–45.
8. S. Gryglewicz, Appl. Catal., A, 2000, 192, 23–28.
9. S. Gryglewicz, Bioresour. Technol., 1999, 70, 249–253.
10. M. L. Granados, M. D. Z. Poves, D. M. Alonso, R. Mariscal, F. C. Galisteo, R. Moreno-Tost, J. Santamaría and J. L. G. Fierro, Appl. Catal., B, 2007, 73, 317–326.
11. E. S. Umdu and E. Seker, Bioresour. Technol., 2012, 106, 178–181.
12. G. Pacchioni, J. M. Ricart and F. Illas, J. Am. Chem. Soc., 1994, 116, 10152–10158.
13. S. Yan, H. Lu and B. Liang, Energy Fuels, 2008, 22, 646–651.
14. M. Kouzu, S. Yamanaka, J. Hidaka and M. Tsunomori, Appl. Catal., A, 2009, 355, 94–99.
15. A. K. Singh and S. D. Fernando, Energy Fuels, 2008, 22, 2067–2069.
16. M. L. Granados, D. M. Alonso, I. Sádaba, R. Mariscal and P. Ocón, Appl. Catal., B, 2009, 89, 265–272.
17. H. Mootabadi, B. Salamatinia, S. Bhatia and A. Z. Abdullah, Fuel, 2010, 89, 1818–1825.
18. J. Jitputti, B. Kitiyanan, P. Rangsunvigit, K. Bunyakiat, L. Attanatho and P. Jenvanitpanjakul, Chem. Eng. J., 2006, 116, 61–66.
19. H. Kim, B. Kang, M. Kim, Y. M. Park, D. Kim, J. Lee and K. Lee, Catal. Today, 2004, 93–95, 315–320.
20. W. Xie, H. Peng and L. Chen, Appl. Catal., A, 2006, 300, 67–74.
21. M. C. G. Albuquerque, I. Jiménez-Urbistondo, J. Santamaría-González, J. M. Mérida-Robles, R. Moreno-Tost, E. Rodríguez-Castellón, A. Jiménez-López, D. C. S. Azevedo, C. L. Cavalcante Jr and P. Maireles-Torres, Appl. Catal., A, 2008, 334, 35–43.
22. J. M. Rubio-Caballero, J. Santamaría-González, J. Mérida-Robles, R. Moreno-Tost, A. Jiménez-López and P. Maireles-Torres, Appl. Catal., B, 2009, 91, 339–346.
23. A. C. Alba-Rubio, J. Santamaría-González, J. M. Mérida-Robles, R. Moreno-Tost, D. Martín-Alonso, A. Jiménez-López and P. Maireles-Torres, Catal. Today, 2010, 149, 281–287.
24. D. G. Cantrell, L. J. Gillie, A. F. Lee and K. Wilson, Appl. Catal., A, 2005, 287, 183–190.
25. S. Yan, M. Kim, S. Mohan, S. O. Salley and K. Y. S. Ng, Appl. Catal., A, 2010, 373, 104–111.
26. L. F. Man, W. T. Wong and K. F. Yung, J. Cluster Sci., 2012, 23, 873–884.
27. G. Guan, K. Kusakabe, N. Sakurai and K. Moriyama, Fuel, 2009, 88, 81–86.
28. H. J. Berchmans and S. Hirata, Bioresour. Technol., 2008, 99, 1716–1721.
29. D. Hirabayashi, T. Yoshikawa, K. Mochizuki, K. Suzuki and Y. Sakai, Catal. Lett., 2006, 110, 155–160.
30. B. Xue, J. Luo, F. Zhang and Z. Fang, Energy, 2014, 68, 584–591.
31. C. Chuanwen, S. Feng, L. Yuguo and W. Shuyun, J. Mater. Sci.: Mater. Electron., 2010, 21, 349–354.
32. S. G. Hosseini and A. Eslami, Prog. Org. Coat., 2010, 68, 313–318.
33. X. Wu and D. Y. C. Leung, Appl. Energy, 2011, 88, 3615–3624.
34. Y. Yamini, N. Bahramifar, F. Sefidkon, M. J. Saharkhiz and E. Salamifar, Nat. Prod. Res., 2008, 22, 212–218.
35. S. M. Pourmortazavi, S. S. Hajimirsadeghi, I. Kohsari, R. Fareghi Alamdari and M. Rahimi-Nasrabadi, Chem. Eng. Technol., 2008, 31, 1532–1535.
36. C. Samart, C. Chaiya and P. Reubroycharoen, Energy Convers. Manage., 2010, 51, 1428–1431.
37. G. Knothe, J. Am. Oil Chem. Soc., 2000, 77, 489–493.
38. G. F. Ghesti, J. L. de Macedo, I. S. Resck, J. A. Dias and S. Dias, Energy Fuels, 2007, 21, 2475–2480.
39. J. K. Satyarthi, D. Srinivas and P. Ratnasamy, Energy Fuels, 2009, 23, 2273–2277.
40. J. V. Gerpen, Fuel Process. Technol., 2005, 86, 1097–1107.
41. H. Hoydonckx, D. de Vos, S. Chavan and P. Jacobs, Top. Catal., 2004, 27, 83–96.
42. A. Kawashima, K. Matsubara and K. Honda, Bioresour. Technol., 2008, 99, 3439–3443.
43. P. Zhang, Q. Han, M. Fan and P. Jiang, Appl. Surf. Sci., 2014, 317, 1125–1130.
44. G. Teng, L. Gao, G. Xiao and H. Liu, Energy Fuels, 2009, 23, 4630–4634.
45. N. Viriya-empikul, P. Krasae, W. Nualpaeng, B. Yoosuk and K. Faungnawakij, Fuel, 2012, 92, 239–244.
46. Y. Li, S. Lian, D. Tong, R. Song, W. Yang, Y. Fan, R. Qing and C. Hu, Appl. Energy, 2011, 88, 3313–3317.
47. E. Lotero, Y. Liu, D. E. Lopez, K. Suwannakarn, D. A. Bruce and J. G. Goodwin, Ind. Eng. Chem. Res., 2005, 44, 5353–5363.
48. H. Zeng, Z. Feng, X. Deng and Y. Li, Fuel, 2008, 87, 3071–3076.
49. W. Xie, H. Peng and L. Chen, J. Mol. Catal. A: Chem., 2006, 246, 24–32.
50. Q. Shu, J. Gao, Z. Nawaz, Y. Liao, D. Wang and J. Wang, Appl. Energy, 2010, 87, 2589–2596.
51. D. Rattanaphra, A. Harvey and P. Srinophakun, Top. Catal., 2010, 53, 773–782.
52. E. Li, Z. P. Xu and V. Rudolph, Appl. Catal., B, 2009, 88, 42–49.
53. D. Kusdiana and S. Saka, Fuel, 2001, 80, 693–698.
54. P. S. Sreeprasanth, R. Srivastava, D. Srinivas and P. Ratnasamy, Appl. Catal., A, 2006, 314, 148–159.
55. J. Park, D. Kim and J. Lee, Bioresour. Technol., 2010, 101, S62–S65.
56. M. G. Kulkarni and A. K. Dalai, Ind. Eng. Chem. Res., 2006, 45, 2901–2913.
57. E. Crabbe, C. Nolasco-Hipolito, G. Kobayashi, K. Sonomoto and A. Ishizaki, Process Biochem., 2001, 37, 65–71.
58. Y. Zhang, W. T. Wong and K. F. Yung, Appl. Energy, 2014, 116, 191–198.
59. Y. Zhang, W. T. Wong and K. F. Yung, Bioresour. Technol., 2013, 147, 59–64.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra13819a

---