2.1 Reagents

Waste cooking soybean oil was obtained from Brazilian restaurant. The other reagents were: Silica beach sand (MPI silica); Hydrochloric acid (HCl, Synth, 36.5%); Sodium hydroxide (NaOH, Vetec, 99%); Potassium hydroxide (KOH, Synth PA); Ethanol (C₂H₅OH, Dynamic PA, 96%); Sodium chloride (NaCl, Dynamic, 99%); Methanol (CH₃OH, Vetec, 99.8%); Sodium sulfate (Na₂SO₄, Vetec, 99%); Benzoic acid (C₇H₆O₂, Dynamic, 99.5%); Phenolphthalein (C₂₀H₁₄O₄, Vetec PA ACS) and Distilled water. The chemicals obtained were used as received.

2.2 Catalyst preparation and characterization

Amorphous silica MPI was synthesized from beach sand using a methodology developed in this research group and described by de Carvalho et al. (2015). In this work, the SPS catalysts were obtained by calcination of mixtures containing alkaline hydroxides (NaOH and KOH) and MPI silica, at 450 °C in a muffle furnace for 4 h. Different molar ratios of NaOH:KOH:MPI silica were used (1:1:1; 2:1:1; 1:2:1, and 1:1:2), to produce the catalysts named SPS, SPS-1, SPS-2, and SPS-3 respectively. The obtained catalysts were applied in preliminary tests for the biodiesel production, using the following synthesis conditions: 3.5% (w/w, catalyst/WCO), 9:1 (molar ratio A/O), 2 h, and 70 °C.

The catalysts were characterized by various techniques. X-ray diffraction analyses (XRD) were performed using a Bruker D2 Phaser device (Bruker AXS, Madison, WI, USA) with CuKα radiation (λ = 1.5406 Å), 30 kV filament, 10 mA current, Ni filter and a LYNXEYE detector in the range from (2θ) 5 to 50° for MPI silica, and 5 to 70° for the catalysts. Elemental analysis was conducted on a Bruker S2 Ranger X-ray fluorescence (XRF) Spectrometer (Bruker AXS, Madison, WI, USA) using Pd or Ag radiation (max. power 50 W, max. voltage 50 kV, max. current 2 mA, XFlash® Silicon Drift Detector). Fourier transform infrared spectroscopy (FTIR) was accomplished in a Shimadzu IRAffinity-1 spectrometer (Columbia, MD, USA) with attenuated total reflectance (ATR). Spectrum analysis variation was in the range of 600–4000 cm⁻¹ with a resolution of 4 cm⁻¹ and 32 scans. The thermal analysis was performed in a thermo microbalance (TG-209-F1-Libra, Netzsch, Selb, Germany) using an alumina crucible for measuring 10 mg of the samples with a continuous heating rate of 10 °C min⁻¹ in nitrogen (N_2(g)) purge gas at a flow rate of 20 mL min⁻¹.

The morphologies and chemical composition of catalysts were obtained using a field emission scanning electron microscope (FESEM, Auriga, Carl Zeiss, Oberkochen, BW, Germany) and energy dispersive X-ray spectroscopy (EDX, XFlash Detector 410-M, Madison, WI, USA), respectively.

The analysis of CO₂ desorption at programmed temperature (CO₂-TPD) consisted of weighing a 200 mg of the SPS catalyst. Next, the pre-treatment was performed with heating at 200 °C under a N₂ flow of 16 mL min⁻¹ for 1 h. At the end of this period, the temperature was reduced to 60 °C and the CO₂ flow (16 mL min⁻¹) was inserted into the reaction line to start the adsorptive process for 30 min. The analysis was subsequently started in a He (g) atmosphere in the temperature range of 40 °C to 500 °C with a heating rate of 10 °C min⁻¹ min under He (g) flow (16 mL min⁻¹). The desorbed CO₂ was then quantified by a thermal conductivity detector.

The deconvolution methodology was applied to the CO₂-TPD, and TG/DTG of catalyst and biodiesel using PeakFit 4.12 software (Systat Software, Inc., Berkshire, UK) by applying Gauss + Lorentz for better curve adjustment, Savitzky-Golay smoothing filter (<20%) and linear baseline.

Hammett’s basicity test was performed using the titration method with acid-base indicators. A phenolphthalein indicator (H_= 9.3) and a 0.01 mol/L methanolic benzoic acid solution were used as the titrant for the experimental procedure. The Hammett basicity test consisted of stirring 0.15 g of the catalyst with 2 mL of methanolic indicator solution at a concentration of 0.1 mg mL⁻¹ for 30 min at 300 rpm. The obtained data were then applied in Eq. (S1) and (S2) (all figures, tables and equations indicated with S are in the supplementary material) to calculate the number of basic sites from the basicity calculation.

Raman spectra were obtained using a confocal Raman microscope (LabHAM HR Evolution, HORIBA Scientific), with laser wavelength of 532 nm, grade: 600 gr mm⁻¹, laser power of 1% and scanning range 400–4000 cm⁻¹. N₂ adsorption–desorption isotherms at 77 K were used to determine the textural parameters of the SPS catalyst in a Micromeritics ASAP 2020 apparatus (Norcross, GA, USA). The specific surface area (S_BET) was calculated using the Brunauer – Emmett – Teller (BET) equation.

2.3 Preparation and characterization of biodiesel from WCO

Biodiesel was synthesized in three consecutive stages: (i) previous filtration of the WCO in a separating funnel to remove impurities; (ii) transesterification reaction at 70 °C with methanol and catalyst in a reflux reactor; and (iii) purification according to the methodology described in the literature (Fernandes et al., 2012), with adaptations. The time, catalyst concentration and molar ratio parameters were varied for synthesis optimization.

Hydrogen Nuclear Magnetic Resonance (NMR ¹H) and Carbon-13 (NMR ¹³C) one-dimensional analyses were obtained by a Bruker Avance III HD NMR SPECT. 300 Spectrometer operating at frequencies of 300.13 MHz for hydrogen (¹H) and 75.47 MHz for carbon (¹³C) respectively. The WCO and biodiesel were dissolved in deuterated chloroform (CDCl₃) in the proportion of 20 mg of sample to 0.5 mL of solvent. The chemical shifts (δ) were expressed in parts per million (ppm) and Tetramethylsilane (TMS) was used as an internal standard. Eq. (S3) was used to calculate the conversion of esters (Gohain et al., 2017).

The thermal analysis (TG/DTG) was accomplished in a thermo microbalance (TG-209-F1-Libra, Netzsch, Selb, Germany) using nitrogen (N_2(g)) as purge gas at a flow rate of 20 mL min⁻¹, alumina crucible, heating rate of 10 °C min⁻¹, 10 mg of sample and final temperature was 600 °C. The deconvolution was performed using the Peakfit v.4.12 software.

The physical–chemical properties were obtained under conditions for density at 20 °C and kinematic viscosity at 40 °C according to the American Society for Testing and Materials (ASTM) D4052 and D7042, respectively. The yields of the transesterification reactions were calculated using Eq. (S4) (Yang et al., 2016). The acidity index was obtained applying Eq. (S5) (AOCS, 2009).

2.4 Catalyst reuse

The regeneration was performed in three stages: (1) washing with mixture of hexane and ethanol (50 mL, 1:1 v/v) under stirring for 1 h, (2) soaking in a new mixture of hexane and ethanol (50 mL, 1:1 v/v) for 4 h to remove excess glycerol and waste cooking oil that could remain on its surface and (3) oven drying for 2 h at 150 °C. This procedure was repeated for five cycles.

The overall methodology employed in this research is presented in a simplified flow diagram depicted in Fig. 1.

Close

Fig. 1

Simplified flow diagram of the overall methodology for catalyst preparation and biodiesel production from waste cooking oil (WCO).

Export to PPT

3.1 Preliminary tests for catalysts syntheses

The catalysts syntheses were performed with different proportions of MPI silica, NaOH, and KOH in order to evaluate the reaction yield under the reaction conditions described in Table S1. (S indicates supplementary material). The selected SPS catalyst was obtained with lower proportion (1:1:1) of the reagents, since the others did not significantly affect the reaction yield for WCO biodiesel production, Table S2. However, the concentrations of the species obtained by XRF that correspond to the oxides of the components showed differences related to the feed molar ratio (Do Nascimento-Dias et al., 2017), which probably occurred due to the utilization of sodium and potassium hydroxides that reacted with silica forming the metal silicates and producing the catalyst, with the possibility of varying the concentrations. In addition, it was possible to verify that pure silica did not have the ability to catalyse the reaction, indicating the need to modify its structure and composition.

3.2 Characterization of MPI silica and SPS catalyst

The XRD pattern for MPI silica (Fig. 2a) exhibited a broad peak centered at 2θ angle 22.8°, characteristic of an amorphous silica structure with the presence of short-range order in atomic clusters (Salakhum et al., 2018; Stanishevsky and Tchernov, 2019), the crystalline plane phases of the SPS catalyst corresponding to Na₂O (2θ = 32.23°, ICSD 060435), KCl (2θ = 40.72°, ICSD 044281), K₂(Si₂O₅) (2θ = 28.57°, 31.77° and 58.87°, ICSD 280480), K₂O (2θ = 41.50° and 66.53°, ICSD 060489), SiO₂ (2θ = 13.10° and 25.87°, ICSD 065497), K₂CO₃ (2θ = 29.87°, 31.75°, 31.77° and 66.47°, ICSD 000662) and Na₂(Si₂O₅) (2θ = 32.80° and 50.47°, ICSD 080378) were shown in Fig. 2b indicating that the modification of the MPI silica to obtain catalytic sites was successful.

Close

Fig. 2

XRD pattern of MPI silica and SPS catalyst.

Export to PPT

The XRF results for the SPS catalyst and MPI silica were described in Table 1and allowed to confirm the elements presence of found in the XRD phases, revealing the presence of alkali oxides as the main components of the produced SPS. These compounds have the ability to promote the catalytic activity of the transesterification reaction (Chouhan and Sarma, 2013), since mixed metal oxides are present as an interesting class of solid heterogeneous catalysts, allowing the association of the various oxide phases that promote appropriate characteristics for the reaction process (Lee et al., 2016).

FTIR spectra of MPI silica and SPS were presented in Fig. S1 and Table S3. For both materials the band at 1417 cm⁻¹ indicates the existence of Na₂CO₃ formed by the reaction of sodium hydroxide with atmospheric CO₂ (Belmokhtar et al., 2016; Simanjuntak et al., 2014). The broadband in the region of 2500 and 3750 cm⁻¹ may be attributed to OH groups from silanol and adsorbed water (Hindryawati et al., 2014). For MPI silica the band at 1314 cm⁻¹ is associated with asymmetrical stretching of the siloxane (Si-O-Si) bonds (Hindryawati et al., 2014). Additional description of the SPS FTIR spectrum were provided in Table S3.

A comparison of both spectra, allowed observing a decrease in the intensity of the bands in the range of 3526 to 2451 cm⁻¹ in MPI silica, probably due to moisture loss during the silica calcination to produce the SPS. Additionally, the decrease of a signal at 1606 cm⁻¹ and disappearance of the band at 1314 cm⁻¹, in the SPS catalyst might have occurred due to a reaction of the silane’s groups polycondensation during SPS preparation. The increase in intensity of the band at 1427 cm⁻¹ is attributed to the carbonates formed after the treatment of MPI silica to produce SPS (Peyne et al., 2017). Moreover an enlargement of the peak at 1000 cm⁻¹ (Si-O-Si group in the MPI silica) in the range of 1000–800 cm⁻¹ is noted, which could be probably due to the overlap of some peaks present in the SPS. The changes in the silicon bands would be caused by depolymerization of the silica network by the potassium ions, which are network-modifying agents and may affect the number of atoms in the first coordination sphere of the atomic silicon (Puligilla et al., 2018).

The results of the thermogravimetric analysis for the MPI silica exhibited three main events characteristics of amorphous silica (Fig. 3a). The first event of mass loss (4.8%) occurred in the range of 27.39 °C to 190 °C, due to the removal of physisorbed water on the silica surface. The second mass loss event occurs over a wide temperature range of 190 to about 632.86 °C (mass loss of 2.22%), probably due to the condensation of less stable silanols and of silane into siloxane. The third event (9.31% of mass loss) starting at 632.86 °C may be attributed to the more stable silanols were dehydroxylated and condensed (De Carvalho et al., 2015; Kin et al., 2009).

Close

Fig. 3

TG/DTG analysis curves for MPI silica (a), SPS catalyst (b), and DTG deconvolution of SPS catalyst.

Export to PPT

The results of the thermal analysis for SPS, Fig. 3b, showed (I) a 5.75% of mass loss attributed to the water molecules adsorbed on the material (34–190 °C) and (II) a second mass loss event of 1.35% attributed to the silanols and silane into siloxane condensation, of the 190 °C to 604.96 °C, and (III) third mass loss event of 10.38%, release of water by condensation/polymerization of the Si-OH groups above 604.96 (He et al., 2010; Kin et al., 2009). The DTG deconvolution result of the SPS catalyst, Fig. 3c and Table S4 confirmed the three mass loss events verified in Fig. 3b. This shows the stability of the material after removing the water by heating for activating the catalyst.

The FESEM micrograph for the MPI silica revealed the presence of irregularly distributed particles (Fig. 4a). The EDX maps showed a homogeneous distribution of the elements Si, Al, O, and Na on the surface of the support (Fig. 4b-e). FESEM micrograph for SPS (Fig. 4f) indicated larger irregular structures in relation to the image obtained for the MPI silica (Fig. 4a). Upon modification of MPI silica with alkali treatment and calcination the surface heterogeneity (with steps and kinks) and particle agglomeration are enhanced. The mapping images (Fig. 4g-j) demonstrated that there is good dispersion of the elements Si, Na, K, and O over the catalyst. The morphological changes in the silica MPI which acts as a support presented formation of the SPS catalyst by adding alkali metals and the adopted synthesis methodology. The EDX maps and spectra for both materials, SPS (Fig. 4k) and MPI silica (Fig. S2) are in agreement with the XRD (Fig. 2) and XRF results (Table 1).

Close

Fig. 4

(a) FESEM; (b–e) elemental mapping images of MPI silica and (f) FESEM, (g-j) mapping images, (k) and EDX spectrum of SPS catalyst.

Export to PPT

The CO₂-TPD results of SPS (Fig. 5a) exhibited several desorption peaks, indicating the presence of with weak (100.67–149.71 °C), medium (149.71–273 °C) and strong (273–405 °C) basic sites, demonstrating the heterogeneity of the material (Eom et al., 2015).

Close

Fig. 5

CO₂-TPD deconvolution analysis for SPS catalyst.

Export to PPT

The deconvolution of the CO₂-TPD (Fig. 5b and Table S5) enabled to verify that the largest proportion of basic sites are strong characteristic, composing about 51.9%, with 31.6% corresponding to sites of medium strength.

The results of the Hammett basicity test were positive to identify that SPS has basic sites in the range of 9.3 < H_ <15 of phenolphthalein (pKb = 9.8), corresponding to 3.2 mmol g⁻¹ of the indicator, which denote the presence of a substantial quantity of active sites at alkaline pH in the SPS (Okoye et al., 2019). According to the specialized literature, sodium and potassium silicates have a basic strength above the phenolphthalein range (range 15 < H_ <18.4) (Hindryawati et al., 2014), justifying the catalytic capacity of the SPS catalyst (this study). The obtained results of basicity are consistent with those of CO₂-TPD.

In the Raman spectrum of SPS catalyst, Fig. 6, there were peaks in the range of 500–700 cm⁻¹ that could be attributed to vibrational stretching and bend of the Si-O-Si (Santos et al., 2019; Partyka and Leśniak, 2016). At 965.20 cm⁻¹ there was a peak that corresponds to the antisymmetric stretching of the Si-O bond (Zhu et al., 2019). At 1059.84 cm⁻¹, a peak of strong intensity was attributed to the symmetrical stretching of the C-O of the K₂CO₃ group (Ma et al., 2021). A peak at 1074.96 cm⁻¹ was assigned to stretching CO₃^2– of the MgCO₃ (Williams et al., 1992).

Close

Fig. 6

Raman spectra of SPS catalyst.

Export to PPT

The N₂ adsorption–desorption isotherms of SPS (Fig. S3) were classified as type III at relative pressures of 0.1 < P/P_o < 0.6, and type IV(a) at 0.6 < P/P_o < 0.98, with an H3 hysteresis cycle (Thommes et al., 2015). The specific surface area (S_BET) obtained for the SPS catalyst is 0.710 m² g⁻¹ and the pore volume is 0.00421 cm³ g⁻¹. This demonstrates a sharp drop when compared to the MPI silica that presented S_BET of 33.54 m² g⁻¹ and pore volume of 0.18 cm³ g⁻¹, according to Carvalho et al. (2015). This decrease may be associated with the addition of K⁺ and Na⁺ and also due to the particle agglomeration upon calcination as seen in FESEM images. Strong basic sites that may occlude the catalyst pores (Farias et al., 2011).

3.3 Proposed mechanism for biodiesel production

In accordance with the obtained XRD, XRF, and Raman results, three species make up the SPS catalyst: potassium carbonate, alkali metal oxides, and sodium and potassium silicates. In this work, a mechanism was proposed for biodiesel synthesis by transesterification reaction, adapted from Guo et al. (2012). This proposal is based on the silicate interactions (Fig. 7), in which the methanol approaches the catalyst surface favoring the ion exchange between the metal silicate (Na or K) and the hydrogen from the alcohol forming the methoxide. There is a subsequent nucleophilic attack of the methoxide on the carbonyl of the triglyceride forming the tetrahedral intermediate that after rearrangement results in the fatty acid methyl esters (FAME). Lastly, an intramolecular rearrangement of protons in the diglyceride occurs to stabilize the charge.

Close

Fig. 7

Proposed mechanism for the H/M ion exchange of methanol with base sites of alkaline silicates.

Export to PPT

3.4 Biodiesel synthesis

3.4.1

3.4.1 Assessment of transesterification yield

The reaction yield (%) for the biodiesel synthesis (calculated by Eq. S4) was assessed at different conditions of time, molar ratio of alcohol to oil (A/O), SPS catalyst concentration and its reuse (Fig. 8). It was observed that the yield progressively increased with increase in time (Fig. 8a), catalyst concentration, and molar ratio of A/O (Fig. 8b-c), as determined through Eq. (S4). The tests for the SPS reuse revealed a good catalyst performance as the reaction yield was approximately the same during in the first four cycles, with a decrease only in the fifth cycle (Fig. 8d). The factors that most influenced the yield were time and A/O molar ratio from 12:1 to 15:1.

Close

Fig. 8

Biodiesel yield (%)as a function of (a) time variation, (b) SPS concentration, (c) molar ratio of A/O and (d) reuse of the catalyst.

Export to PPT

3.4.2

3.4.2 Biodiesel characterization

The ¹H NMR spectra for the waste cooking oil (WCO) (Fig. 9a) and biodiesel (Fig. 9b) were obtained with the purpose of evaluating the biodiesel purity and the conversion to esters in the SPS-catalysed transesterification reaction. The conversion may be confirmed by the disappearance of the peaks at 4.1–4.3 ppm, attributed to hydrogen from triglycerides (WCO) (Ruschel et al., 2016), and the appearing of a signal at 3.7 ppm due to hydrogen from methoxy groups of methyl esters (biodiesel) (Gohain et al., 2017), resulting in a conversion of around 93.89% (Eq.(S3)). In addition, absence of contaminants is observed.

Close

Fig. 9

¹H NMR spectra (a) WCO and (b) biodiesel.

Export to PPT

In the ¹³C NMR spectra of WCO and biodiesel (Fig. S4), carboxyl ester group (signals at 174.3 and 54.43 ppm) and olefinic groups (unsaturated methyl ester, signals in the range of 130.18–129.73 ppm) respectively were observed (Fig. S4a). Furthermore, triglycerides peaks (CH-O and CH₂-O) are found between 68.88 and 61.77 ppm (Fig. S4a) (Tariq et al., 2011). A peak referring to methyl carbon appears at 14.9 ppm, while carbons of methylene groups (hydrocarbon chain) were observed in the range of 34.9–22.48 ppm (Fig. S4b) (Gohain et al., 2017). Comparing the results for WCO and biodiesel, it is possible to observe the disappearance of triglyceride peaks and the appearance of ester carbonyl signal (C-O) at 51.43 ppm in the ¹³C NMR of biodiesel (Fig. S4b) (Tariq et al., 2011).

Thermal analysis of WCO and biodiesel produced from WCO (Fig. 10a and 10b) also confirmed the conversion and quality of biodiesel produced with the SPS catalyst, in agreement with the 1H NMR analysis (Fig. 9b). The DTG deconvolution method was performed on the DTG curve of biodiesel (Fig. S5 and Table S6) to evaluate its components. A mass loss event for WCO is observed in the range of 347–475 °C, (Fig. 10a) which corresponds to the decomposition of triglycerides while for the biodiesel the mass loss (Fig. 10b) occurs at lower temperatures in the region of 163 to 266 °C due to the decomposition of smaller molecules, corresponding to monoalkyl ester (Dantas et al., 2007; Misutsu et al., 2015). The SPS showed similar behaviour as other catalysts studied in the specific literature, which are described in Table 2.

Close

Fig. 10

TG/DTG curves of (a) WCO and (b) biodiesel produced from waste cooking oil using SPS catalyst.

Export to PPT

Table 2 Comparison of the performance of SPS group of catalysts with literature reports.

Catalyst	Catalyst’s preparation	Reaction conditions	Yeld (%)	Ref.
Pure silica	–	12 h, 70 °C, 3.5 (wt.%) catalyst, 400 rpm, A/O 9:1	Not converted	This work
SPS SPS-1 SPS-2 SPS-3	MPI Silica + NaOH + KOH calcination, 4 h, 450 °C	2 h, 70 °C, 3.5 (wt.%) catalyst, 400 rpm, A/O 9:1	92.8 93.6 93.5 93.7	This work
Na2SiO3	Dehydrated at 200 °C for 20 min and calcination, 2 h, 400 °C	Up to 4 h, 50–65 °C, 3–5 (wt.%) catalyst, A/O 1:3–1:15	57–99.6	Long et al., 2011.
Li, Na, K/Silicate	Silica/metallic ions, 2 h, 90 °C	1 h, 65 °C, 3 (wt.%) catalyst, A/O 9:1	96.5–98.2	Hindryawati, 2014.
Na2SiO3 (Sigma–Aldrich (Pty)	(direct use)	4 h, 64 °C, 2.5 wt% catalyst, A/O 6:1	57.92	Daramola et al., 2016
Na2SiO3 (Kermel Corp.)	Calcination, 2 h, 400 °C	1 h, 60 °C, 3 (wt.%) catalyst, A/O 7.5:1,	15–99	Guo et al., 2012.
K2SiO3	Silica from bamboo leaves + KOH, calcination, 2 h, 500 °C	2 h, 60 °C, 3 (wt.%) catalyst, A/O 9:1,	83.35	Manurung et al., 2019.
KAlSiO4	K2O.3.9SiO2 + KOH + Al(NO3)3, calcination,4h, 1000–1200 °C	2–5 min, 120–180 °C, 3 (wt.%) catalyst,	20–54.4	Wen and Yan, 2011.

It was possible to observe four components (Fig. S5) which corresponds to the observed mass loss in the TGA curve of biodiesel. Of these, three are attributed to monoalkyl ester, corresponding to 98% of the integrated area of DTG, and the other peak corresponds to oxidation products formed (about 2%) (Díaz-Ballote, 2018). It is possible to prove the synthesis occurrence via the thermal evaluation of biodiesel, which corroborates the ¹H and ¹³C NMR results. The values calculated for the physicochemical properties such as density at 20 °C (0.890 g cm⁻³), kinematic viscosity at 40 °C (5.061 mm² s⁻¹) and acidity index (between 0.355 and 0.155 mg KOH g⁻¹) were within the specifications of National Agency of Petroleum Natural Gas and Biofuels (ANP N° 45, 2014),. Characteristic bands of biodiesel were observed in the FTIR spectrum of biodiesel (Fig. S6). The band at 1736 cm⁻¹, was attributed to the elongation of the carbonyl bond; absorptions in the 1425–1447 cm⁻¹ correspond to the asymmetric CH₃ flexion, and the bands in the range of 1188–1200 cm⁻¹ correspond to O-CH₃ stretching (Mumtaz et al., 2012, De Morais et al., 2013).