Study of the operational conditions for ethyl esters production using residual frying oil and KF/clay catalyst in a continuous system

2.1. Materials

The substrates used were residual frying oil obtained from a local restaurant (Umuarama - PR) and ethanol (Panreac, 99.9% purity). The composition of the RFO was previously reported by Zempulski et al., 2020Zempulski DA, CP, Milinsk MC, Alves HJ, Silva C. 2020. Continuous transesterification reaction of residual frying oil with pressurized ethanol using KF/Clay as catalyst. Eur. J. Lipid Sci. Technol. 122, 1900315. https://doi.org/10.1002/ejlt.201900315 , and its free fatty acid and water contents were 1.16 wt% and 0.16 wt%, respectively. For the preparation of the catalyst, Brazilian sodium smectite clays were used from deposits located in the states of Paraná and Paraíba, and potassium fluoride KF.2H₂O (Synth, 98% purity). For the characterization analyses methyl heptadecanoate (Sigma-Aldrich, > 99% purity), heptane (Anhydrol), potassium hydroxide, methanol (Panreac, 99.9%), sulfuric acid (Anidrol, 98%), boron trifluoride (Sigma-Aldrich, 14% in methanol), monolein (Sigma Aldrich, ≥ 98%), 1,3-diolein (Sigma Aldrich, ≥ 95%), glycerol trioleate (Sigma Aldrich, ≥ 98.5%), derivative N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA, Sigma Aldrich, ≥ 98.5%) were used.

The catalyst used in this work was based on the study of Alves et al. (2014)Alves HJ, Rocha AM, Monteiro MR, Morwtti C, Cabrelon MD, Schwengber CA, Milinsk MC. 2014. Treatment of clay with KF: New solid catalyst for biodiesel production. Appl. Clay Sci. 91-92, 98-104. https://doi.org/10.1016/j.clay.2014.02.004 , where the clay was crushed (< 60 mesh) and wet-impregnated with potassium fluoride at the concentration of 1.7 mol·L^-1. The suspension, a 15% clay and potassium fluoride solution (mass concentrate), was maintained in a blanket and reflux (80 °C, 30 min), followed by drying (110 °C, 24 h). The mass obtained from clay and KF was de-agglomerated in a mortar until a fine powder was obtained. The final stage consisted of the formation of granules, where the powder was compacted in a hydraulic press with a 5-ton pressure, de-agglomerated, and classified in sieves. The granules used in the work were in the range of 710 μm to 1 mm obtained with the aid of sieves.

2.2. Experimental procedures

The reactions, carried out in duplicate, served to investigate the influence of the oil-to-ethanol mass ratio of 1:1, 1:1.5 and 1:2 (equivalent to the oil-to-ethanol molar ratios of 1:20, 1:30 and 1:40, respectively). The tests were conducted using a catalyst mass of 2 g in the catalyst bed (these conditions correspond to an oil-to-catalyst ratio of 8:1, 6.4:1 and 5.4:1, respectively) at 275 and 300 °C. The catalyst mass and temperatures were defined based on the results obtained by Zempulski et al. (2020)Zempulski DA, CP, Milinsk MC, Alves HJ, Silva C. 2020. Continuous transesterification reaction of residual frying oil with pressurized ethanol using KF/Clay as catalyst. Eur. J. Lipid Sci. Technol. 122, 1900315. https://doi.org/10.1002/ejlt.201900315 , which provided the experiments the highest ester content and in the case of the selected temperature range did not show any loss in catalytic activity (in 3 h of reaction). The pressure was maintained at 20 MPa based on previous studies (Gonzalez et al. 2013Gonzalez SL, Sychoski MM, Navarro-Díaz, HJ, Callejas N, Sainebe M, Vieitez I, Jachmanián I, Silva C, Hense H, Oliveira JV. 2013. Continuous catalyst-free production of biodiesel through transesterification of soybean fried oil in supercritical methanol and ethanol. Energy Fuels 27, 5253-5259. https://doi.org/10.1021/ef400869y ; Trentini et al. 2018Trentini CP, Fonseca JM, Cardozo Filho L., Reis RR, Sampaio SC, Silva C. 2018. Assessment of continuous catalyst-free production of ethyl esters from grease trap waste. J. Supercrit. Fluids 136, 157-163. https://doi.org/10.1016/j.supflu.2018.02.018 ), and the residence times evaluated were 15, 30 and 45 min.

The experimental apparatus where the reactions were performed was previously described (Zempulski et al., 2020Zempulski DA, CP, Milinsk MC, Alves HJ, Silva C. 2020. Continuous transesterification reaction of residual frying oil with pressurized ethanol using KF/Clay as catalyst. Eur. J. Lipid Sci. Technol. 122, 1900315. https://doi.org/10.1002/ejlt.201900315 ), and operated in continuous mode. The reaction mixture (RFO + ethanol) was kept under constant stirring (mechanical stirrer, IKA RW20) and pumped into the system using an isocratic pump (Waters 515 HPLC). The heating and maintenance of the operating temperature was done in a furnace (Sanchis, BTT1050-00), where the reactor was placed. The reactor consisted of 3 parts: pre-heating zone, catalytic tubular reactor (internal diameter of 9.5 ×10^-3 m), and non-catalytic tubular reactor, whose total volume was 35.15 mL. The catalytic tubular reactor was filled with glass beads of medium diameter of 4 mm at the ends and the catalyst (previously activated in a muffle - 48 h, 200 °C). To avoid mass losses to the system, sintered steel filters (2 μm, Allcron) were arranged at the ends. Before collection sampling, the samples were cooled by indirect heat exchange, using a thermostatic bath (Tecnal TE - 184). The residence time was calculated by dividing the empty volume of the reactor (mL) by the flow of substrates (mL min^-1) and counted after the system entered steady state, considering 2.5 times the residence time. After the reaction, unreacted ethanol was recovered using a rotary evaporator (Marconi, MA 120) coupled with a vacuum pump (Quimis®, 0955V).

The stability of the catalyst was performed under the experimental conditions (catalyst mass, oil-toethanol mass ratio, temperature and residence time) that provided not only high yield but also low thermal decomposition, resource savings (energy for temperature rise and ethanol employed). This step was performed from the addition of 2 g of catalyst in the catalytic tubular reactor and as soon as the adopted experimental conditions were reached the reaction mixture (oil + ethanol) was then pumped into the system. The reaction consisted of operating the reactor for a period of 8 h with sample collection every 30 min after reaching steady state.

Under the same experimental conditions as those selected to assess catalyst stability, the reaction was carried out in two steps. The procedure previously described (Abdala et al., 2014Abdala ACA, Colonelli TAS, Trentini CP, Oliveira JV, Cardozo-Filho L, Silva EA, Silva C. 2014. Effect of additives in the reaction medium on noncatalytic ester production from used frying oil with supercritical ethanol. Energy Fuels 28, 3122-3128. https://doi.org/10.1021/ef402253e ; Trentini et al., 2018Trentini CP, Fonseca JM, Cardozo Filho L., Reis RR, Sampaio SC, Silva C. 2018. Assessment of continuous catalyst-free production of ethyl esters from grease trap waste. J. Supercrit. Fluids 136, 157-163. https://doi.org/10.1016/j.supflu.2018.02.018 ) was used, which consisted primarily of collecting the sample and removal of glycerol. The removal of glycerol was performed by adding distilled water and n-hexane to the reaction product obtained to form immiscible phases. The mixture was centrifuged and the phase containing esters was carried to the oven for evaporation of ethanol and n-hexane. Then, ethanol was added to the sample (without glycerol) and the mixture was pumped into the reaction system. In the second step, the system was filled with fresh catalyst and recycled catalyst (used for a period of 3 h). After the reaction, the recycled catalyst was washed with ethanol and n-hexane to remove possible products adhered to the catalyst. It was then dried in an oven (100 °C for 24h) for later re-use.

2.3. Analytical methods

The collected samples were monitored for preparation: evaporation of unreacted ethanol, removal of glycerol, addition of internal standard (methyl heptadecanoate) and dilution in heptane. One μL of the diluted sample was injected into a gas chromatograph (Shimadzu, GC-2010 Plus) equipped with an automatic injector (Shimadzu, AOC-20i Autoinjector), Shimadzu Rtx-Wax capillary column (30 m x 0.32 mm x 0.25 µm) and flame ionization detector. Nitrogen was used as the carrier gas and the samples were injected in split mode (1:20). The injector and detector were kept at a temperature of 250 °C and the heating ramp for the oven was as follows: it started at 120 °C, followed by an increase to 180 °C at 15 °C·min^-1 and then to 240 °C at 6 °C·min^-1, as described by Trentini et al. (2018)Trentini CP, Fonseca JM, Cardozo Filho L., Reis RR, Sampaio SC, Silva C. 2018. Assessment of continuous catalyst-free production of ethyl esters from grease trap waste. J. Supercrit. Fluids 136, 157-163. https://doi.org/10.1016/j.supflu.2018.02.018 .

The ester yield (Equation 1) was obtained through the convertibility of the RFO (89.52± 0.88 wt%) as determined using methodology proposed by Gonzalez et al. (2013)Gonzalez SL, Sychoski MM, Navarro-Díaz, HJ, Callejas N, Sainebe M, Vieitez I, Jachmanián I, Silva C, Hense H, Oliveira JV. 2013. Continuous catalyst-free production of biodiesel through transesterification of soybean fried oil in supercritical methanol and ethanol. Energy Fuels 27, 5253-5259. https://doi.org/10.1021/ef400869y , which refers to the maximum conversion of ethyl esters that can be achieved from the RFO.

Thermal decomposition was determined according to Vieitez et al. (2009)Vieitez I, Silva C, Alckmin I, Borges GR, Corazza FC, Oliveira JV, Grompone MA, Jachmanián I. 2009. Effect of temperature on the continuous synthesis of soybean esters under supercritical ethanol. Energy Fuels 23, 558-563. https://doi.org/10.1021/ef800640t , and consisted of derivatizing of the compounds (mono, di and triglycerides and ethyl esters) present in the samples with methanolic solution of BF₃ and analyzed by gas chromatography. The decomposition of the samples was obtained from Equation 2:

Where was the sum of all ethyl ester percentages, P₁₆ was the percentage of 16 ethyl ester (considered the most stable), and the subscripts 0 and S indicated the original oil and sample, respectively.

For the quantification of mono-, di- and triglycerides, the sample was derivatized with MSTFA (15 minutes at room temperature) and diluted in heptane (Trentini et al., 2019Trentini CP, Postaue N, Cardozo-Filho L, Reis RR, Sampaio SC, Silva C. 2019. Production of esters from grease trap waste lipids under supercritical conditins: Effect of water addition on ethanol. J. Supercrit. Fluids 147, 9-16. https://doi.org/10.1016/j.supflu.2019.02.008 ). The prepared samples were injected (2 μL) into a gas chromatograph (Shimadzu, GC-2010 Plus) equipped with capillary column Zebron ZB-5HT inferno™ (10 m × 0.32 mm × 0.10 µm), flame ionization detector and column injector. The following oven temperature gradient was applied: initially the column was held at 50 °C for 1 minute, and then heated to 180 °C at a rate of 15 °C∙min^-1, then at 230 °C at a rate of 7 °C∙min^-1 and at 380 °C at 10 °C∙min^-1 and held at this temperature for 5 min. The detector temperature was 380 °C and the heating program of the injector was: initial temperature of 60 °C for 1 min. then heated to 380 °C at a rate of 10 °C∙min^-1 and held for 10 min at this temperature. For the quantification of the compounds, a calibration curve with the standards of triolein, diolein and monolein were used (Trentini et al., 2019Trentini CP, Postaue N, Cardozo-Filho L, Reis RR, Sampaio SC, Silva C. 2019. Production of esters from grease trap waste lipids under supercritical conditins: Effect of water addition on ethanol. J. Supercrit. Fluids 147, 9-16. https://doi.org/10.1016/j.supflu.2019.02.008 ).

2.4. Catalyst characterization

After the reaction, the catalyst recovered from the reactor was washed with ethanol and n-hexane and dried in an oven (Zempulski et al., 2020Zempulski DA, CP, Milinsk MC, Alves HJ, Silva C. 2020. Continuous transesterification reaction of residual frying oil with pressurized ethanol using KF/Clay as catalyst. Eur. J. Lipid Sci. Technol. 122, 1900315. https://doi.org/10.1002/ejlt.201900315 ). The characterization of the catalysts was obtained by X-ray fluorescence analysis (XRF, Spectro Philips MagiX); X-ray diffraction (using the XRD apparatus, Siemens Kristalloflex), in the range of 4° < 2Ɵ < 40°, using CuKα radiation (λ = 1.54056 nm, 40 kV, 40 mA as radiation incident) with a nickel filter, and speed of 0.5^o min^-1 and; Fourier transform infrared spectroscopy (spectrometer FTIR Bomem MB-Series) in the range of 4000 and 500 cm^-1, using a resolution of 4 cm^-1; and scanning electron microscopy (SEM) (Quanta 440 microscope FEI), coupled to an energy dispersive spectroscopy (EDS). In the SEM analysis, the powder samples were dispersed onto a double-sided carbon tape, and subsequently metallized with a thin layer of gold by sputtering.

2.5. Analysis of data

All reactions and analyzes were performed, at least, in duplicate. The means and standard deviation of each assay were submitted to ANOVA using Excel® 2010 software and Tukey’s test (95% confidence) was applied to determine whether there was a significant difference between samples from the different operating conditions studied. The principal component analysis (PCA) was carried out using the Statistica 8.0 software (StatSoft™, Inc.). The principal components were derived from a correlation matrix.

3.1. Effect of mass ratio oil to ethanol

Figure 1 shows the values obtained in terms of ethyl ester yield for the reactions conducted by varying the oil-to-ethanol mass ratio (1:1, 1:1.5 and 1:2) at different residence times. It can be observed at 275 ºC (Figure 1a) that the addition of ethanol to the reaction had a greater influence on the initial minutes of the reaction (15 min), for the other evaluated times there was an increase in the ester yield only when the oil-to-ethanol mass ratio was increased from 1:1 to 1:1.5. The highest yield obtained at this temperature was ~90% in 15 min of reaction. At a temperature of 300 ºC (Figure 1b), the highest ester yield, obtained in 15 min, was for the oil-to-ethanol mass ratio of 1:2. Similarly, for other ratios and residence times, the increase in ethanol did not influence the yield of esters. From the general analysis of the data obtained in Figure 1, it appears that higher yields were obtained in short residence times and that the addition of a high concentration of ethanol to the reaction (above 1:1.5) was not necessary to obtain a high yield of esters at the evaluated temperatures, which is advantageous, considering the lower mass ratio of oil-to-ethanol and low reaction time to obtain high ester yields.

Increasing the proportion of ethanol in the reaction mixture benefits the conditions of mass transfer in the reaction medium (Biktashev et al., 2011Biktashev AS, Usmanov RA, Gabitov RR, Gazizov RA, Gumerov FM, Abdulagatov IM, Yarullin RS, Yakushev, IA. 2011. Transesterification of rapeseed and palm oils in supercritical methanol and ethanol. Biomass Bioener. 35, 2999-3011. https://doi.org/10.1016/j.biombioe.2011.03.038 ). It can also increase the contact area between the alcohol and triglycerides, thus shifting the balance of transesterification to the formation of products (Campanelli et al., 2010Campanelli P, Banchero M, Manna L. 2010. Synthesis of biodiesel from edible, non-edible and waste cooking oils via supercritical methyl acetate transesterification. Fuel 89, 3675-3682. https://doi.org/10.1016/j.fuel.2010.07.033 ; Silva and Oliveira, 2014Silva C, Oliveira JV. 2014. Biodiesel production through non-catalytic supercritical transesterification: Current state and perspectives. Brazilian J. Chem. Eng. 31, 271-285.). In addition, the addition of alcohol to the reaction medium decreases the critical temperature of the reaction mixture (Osmieri et al., 2017Osmieri L, Esfahani RAM, Recasens F. 2017. Continuous biodiesel production in supercritical two-step process : phase equilibrium and process design. J. Supercrit. Fluids 124, 57-71. https://doi.org/10.1016/j.supflu.2017.01.010 ). As an example, for the mass ratios of 1:0.5, 1:1 and 1:2 (oil:methanol), the critical temperatures are 345.93, 305.85 and 282.20 °C, respectively (Bunyakiat et al., 2006Bunyakiat K, Makmee S, Sawangkeaw R, Ngamprasertsith S. 2006. Continuous production of biodiesel via transesterification from vegetable oils in supercritical methanol. Energy Fuels 20, 812-817. https://doi.org/10.1021/ef050329b ). Thus, the greater availability of alcohol in the reaction medium, the better the possibility of conducting the reaction at a milder temperature, still resulting in a high reaction rate (Hegel et al., 2008Hegel P, Andreatta A, Pereda S, Bottini S, Brignole EA. 2008. High pressure phase equilibria of supercritical alcohols with triglycerides, fatty esters and cosolvents. Fluid Phase Equilib. 266, 31-37. https://doi.org/10.1016/j.fluid.2008.01.016 ).

Silva et al. (2010)Silva C, Castilhos F, Oliveira JV, Cardozo Filho L. 2010. Continuous production of soybean biodiesel with compressed ethanol in a microtube reactor. Fuel Process. Technol. 91, 1274-1281. https://doi.org/10.1016/j.fuproc.2010.04.009 reported that the mass ratio 1:1 seemed to be the most adequate in terms of ester yield and reagent savings for the reaction between soybean oil and ethanol at 300 °C. He et al. (2007)He H, Wang T, Zhu S. 2007. Continuous production of biodiesel fuel from vegetable oil using supercritical methanol process. Fuel 86, 442-447. https://doi.org/10.1016/j.fuel.2006.07.035 found that the increase in ethanol-to-oil mass ratio resulted in an increase in yield, and this behavior was maintained until the ratio of ~1:1.5. Xu et al. (2016)Xu QQ, Li Q, Yin JZ, Guo D, Qiao BQ. 2016. Continuous production of biodiesel from soybean flakes by extraction coupling with transesterification under supercritical conditions. Fuel Process. Technol. 144, 37-41. https://doi.org/10.1016/j.fuproc.2015.12.018 indicated that when working with excess alcohol in supercritical conditions, it must be considered that alcohol at its critical point acts not only as a reagent, but also as a solvent, which dissolves the oil present in the reaction mixture and produces an almost homogeneous reaction. The same researchers found that an excess of alcohol could impair the reaction yield because the reagents were too diluted. High concentrations of ethanol can lead to a reduction in reaction rates, which prevents the chemical balance of the reaction from being achieved. It is also important to mention that a high mass ratio may be unfavorable, as more intense processes are required for the separation and evaporation of alcohol (Bunyakiat et al., 2006Bunyakiat K, Makmee S, Sawangkeaw R, Ngamprasertsith S. 2006. Continuous production of biodiesel via transesterification from vegetable oils in supercritical methanol. Energy Fuels 20, 812-817. https://doi.org/10.1021/ef050329b ).

Regarding the temperature applied to the reaction, it was noted that in general, raising the temperature from 275 to 300 °C resulted in a low increase in ester yield for the evaluated residence times. Lin and Tan (2014)Lin HC, Tan CS. 2014. Continuous transesterification of coconut oil with pressurized methanol in the presence of a heterogeneous catalyst. J. Taiwan Inst. Chem. Eng. 45, 495-503. https://doi.org/10.1016/j.jtice.2013.06.015 also found that even by raising the operating temperature (from 280 to 300 °C), ester yields remained similar (~95%) for the reaction between coconut oil and methanol (mass ratio 1:1), with 40 g of MnO₂ catalyst and 11 MPa. Visioli et al. (2019)Visioli LJ, Castilhos F, Silva C. 2019. Use of heterogeneous acid catalyst combined with pressurized conditions for esters production from macauba pulp oil and methyl acetate. J. Supercrit. Fluids 150, 65-74. https://doi.org/10.1016/j.supflu.2019.03.023 reported ~60% ester content for the reaction between macauba oil and methyl acetate (mass ratio 1:2.5), with 2 g of γ-alumina catalyst, at 275 and 300 °C.

Figures 1c and 1d show the percentages of thermal decomposition of the fatty acids present in the samples obtained and it was verified from these data that thermal decomposition is influenced by residence time and temperature. For the initial minutes of reaction, there was low decomposition, and by increasing the residence time, a thermal decomposition of up to 15% was noted, similarly, by raising the operating temperature from 275 to 300 °C (45 min); in the mass ratio of 1:2, thermal decomposition went from ~5 to 15%, respectively. It was also possible to verify that a greater amount of ethanol in the reaction mixture contributed to lower percentages of thermal decomposition at the temperatures evaluated. Thermal decomposition for reactions with waste oil and ethanol has also been reported by Trentini et al. (2020)Trentini CP, Postaue N, Cardozo-Filho L, Silva C. 2020. Waste oil/crambe oil blends for ethyl ester production under supercritical conditions. J. Supercrit. Fluids 163, 104889. https://doi.org/10.1016/j.supflu.2020.104889 , with values of ~20% for 275 and 300 °C, and by Abdala et al. (2014)Abdala ACA, Colonelli TAS, Trentini CP, Oliveira JV, Cardozo-Filho L, Silva EA, Silva C. 2014. Effect of additives in the reaction medium on noncatalytic ester production from used frying oil with supercritical ethanol. Energy Fuels 28, 3122-3128. https://doi.org/10.1021/ef402253e , who obtained ~ 6 and 12% decomposition at 275 and 300 °C, respectively.

The observed thermal decomposition intensified at 300 °C when the oil used for the reactions was composed mainly of linoleic fatty acid, which is less stable at high temperatures compared to saturated and monounsaturated esters, as found by Trentini et al. (2019)Trentini CP, Postaue N, Cardozo-Filho L, Reis RR, Sampaio SC, Silva C. 2019. Production of esters from grease trap waste lipids under supercritical conditins: Effect of water addition on ethanol. J. Supercrit. Fluids 147, 9-16. https://doi.org/10.1016/j.supflu.2019.02.008 . They reported ~23% reduction in the amount of linoleic ester used by raising the operating temperature from 275 to 300 °C, thus verifying that the decomposition of esters is strongly dependent on the operating temperature. Therefore, temperatures above 300 °C should be avoided, as close to this temperature, unsaturated esters are consumed due to thermal decomposition.

Regarding the reactions that occur with thermal decomposition, linear thermal dimerization generally occurs in monounsaturated and polyunsaturated esters, forming acyclic structures of the dimers. A mixture of monocyclic and cyclic dimers of six members can cause other compounds to be formed, such as oligomers, polymers, isomers and thermal cracking products (Quesada-Medina and Olivares-Carrillo, 2011Quesada-Medina J, Olivares-Carrillo P. 2011. Evidence of thermal decomposition of fatty acid methyl esters during the synthesis of biodiesel with supercritical methanol. J. Supercrit. Fluids 56, 56-63. https://doi.org/10.1016/j.supflu.2010.11.016 ). In addition to reducing the yield in esters, thermal decomposition can influence the quality of the biodiesel obtained due to the increase in viscosity and crystallization temperature related to the accumulation of decomposition products (Liu et al., 2016Liu J, Shen Y, Nan Y, Tavlarides LL. 2016. Thermal decomposition of ethanol-based biodiesel: Mechanism, kinetics, and effect on viscosity and cold flow property. Fuel 178, 23-36. https://doi.org/10.1016/j.fuel.2016.03.033 ).

In order to reduce the dimensionality of the data set related to the effect of operational variables on the ester yield (EY) and thermal decomposition (TD) of the reaction products, the principal component analysis (PCA) was performed, using the data from the Table 1. Figure 2 shows the biplot (PC1xPC2) of scores and loadings obtained from a correlation matrix. The two main components (PCs) explained a total cumulative variance of 100% of the data. The vectors of the variables under study form an angle of 90 °, which indicates that there is no correlation between EY and TD.

The PCA formed 2 trend groups. Group A is formed by samples that presented higher EY, since it is close to the vector of this variable. This group is mainly composed of samples obtained at 275 °C, in residence times of 15 to 45 min and oil-to-ethanol mass ratio of 1:1.5 and 1:2, except for sample 6 (obtained at 15 min and 300 °C). These results confirm that the increase in residence time (30 to 45 min) and oil-to-ethanol mass ratio (1:1 and 1:2) did not affect EY, since under these conditions, samples with similar characteristics were obtained. Group B contains samples that showed higher TD and was mainly secreted by reaction products obtained at a temperature of 300 °C and residence time of 45 min, regardless of the oil-to-ethanol mass ratio. This indicates that the increase in temperature and residence time contributed to TD. Sample 10 (obtained at 300 °C, 30 min and oil-to-ethanol mass ratio 1:1) is also represented by this group due to the lower mass of ethanol used at a high temperature. Samples 2 and 3 showed lower TD, since the opposite variable was found in the opposite quadrant. This result was attributed to the lower temperature and residence time employed in this case.

In short, from the results obtained (Figures 1 and 2), it was possible to verify that similar ester yields were obtained with oil-to-ethanol mass ratio of 1:1.5 and 1:2. In addition, the thermal decomposition became accentuated by increasing the residence time, as well as the operating temperature from 275 to 300 °C. It was also noted that raising the operating temperature resulted in a maximum percentage increase of only 5% in ester yield and intensified thermal decomposition (2.6 times). Thus, the operating conditions of 275 °C, 15 min and oil-to-ethanol mass ratio of 1:1.5 were promising to conduct the reaction and verify the stability of the catalyst.

3.2. Catalyst stability

The stability of the catalyst was evaluated for the reaction conducted at the residence time of 15 min, for a period of 8 h after reaching a steady state, at 275 ºC, using 2 g of catalyst and oil-to-ethanol mass ratio of 1:1.5. The results for yield in ethyl esters are presented in Figure 3.

In the first 3 hours of operation, the catalyst remained stable, providing average values for ester yield of 84.26 ± 2.75%. After the mentioned period, there was a slight reduction in the yield, ~15%, and the values remained stable at the time of 3 to 6 h of operation, when again a reduction of ~29% was verified in relation to the initial yield (30 min), and from this period the yield remained close to 65.30 ± 1.60% until the evaluated operation time (8h). According to Simões et al. (2020)Simões SS, Ribeiro JS, Celante D, Brondani LN, Castilhos F. 2020. Heterogeneous catalyst screening for fatty acid methyl esters production through interesterification reaction. Renew. Energy 146, 719-726. https://doi.org/10.1016/j.renene.2019.07.023 with a loss of up to 15% in catalyst activity, it can still be reused. Ensuring catalyst stability is of great importance for industrial applications and reducing the cost of biodiesel production (Zhang et al., 2018Zhang Y, Liu H, Zhu X, Lukic I, Zdujic M, Shen X, Shala D. 2018. Biodiesel synthesis and kinetic analysis based on MnCO₃/Na silicate as heterogeneous catalyst. J. Serbian Chem. Soc. 83, 345-365. https://doi.org/10.2298/JSC170612005Z ). Therefore, the results obtained suggest that the KF/clay catalyst has considerable stability and can be used continuously for ~6 hours of operation.

The stability of the catalyst used in the transesterification reactions in a continuous and pressurized reactor was investigated by other researchers. Jesus (2010)Jesus AA. 2010. Síntese de biodiesel em meio contínuo pressurizado empregando hidrotalcitas como catalisadores heterogêneos. Aracaju: Universidade Tiradentes. used 4 g of hydrotalcites in pressurized methanol at 300 ºC, 15 MPa and residence time of 7 min, and observed that the ester content remained unchanged for 24h. Visioli et al. (2019)Visioli LJ, Castilhos F, Silva C. 2019. Use of heterogeneous acid catalyst combined with pressurized conditions for esters production from macauba pulp oil and methyl acetate. J. Supercrit. Fluids 150, 65-74. https://doi.org/10.1016/j.supflu.2019.03.023 analyzed the reaction of macauba oil and methyl acetate using 2 g γ-alumina catalyst and found that the catalyst remained stable in the evaluated period (8 h) at 225 and 275 °C, with no loss in activity of the catalyst.

Regarding the catalyst stability in reactions conducted in a batch reactor, Teo et al. (2015)Teo SH, Goto M, Taufiq-Yap YH. 2015. Biodiesel production from Jatropha curcas L. oil with Ca and La mixed oxide catalyst in near supercritical methanol conditions. J. Supercrit. Fluids 104, 243-250. evaluated the activity of the catalyst for 5 cycles of 10 min and found a reduction of ~25% in the esters yield in the fourth cycle, for the reaction between Jatropha curcas L. oil and methanol, with 1 wt.% of CaLaO catalyst, at 240 °C. In the study by Ribeiro et al. (2018)Ribeiro JS, Celante D, Brondani LN, Trojahn DO, Silva C, Castilhos F. 2018. Synthesis of methyl esters and triacetin from macaw oil (Acrocomia aculeata) and methyl acetate over γ-alumina. Ind. Crops Prod. 124, 84-90. https://doi.org/10.1016/j.indcrop.2018.07.062 the reuse of the γ-alumina catalyst at 300 °C was reported for 8 cycles of 60 min each, with a reduction of 12.47% in ester content after the seventh cycle.

3.3. Two-step reaction

Based on the results obtained in the first step of the reaction with KF/clay catalyst, a second step was conducted to investigate the influence of the recycled catalyst and the new catalyst in the reaction. The first-step sample (time 0) was obtained in 15 min of reaction, 275 °C, oil-to-ethanol mass ratio 1:1.5 and 2 g of catalyst, and used to conduct the second step, under the same operating conditions with residence times of 15, 30 and 45 min. The esters and intermediate compound contents obtained are presented in Figure 4, as well as the values for percentage of thermal decomposition.

Applying the second reaction step, it was possible to verify that using recycled and new catalysts, they provided an increase of ~6.6% in the ester content, respectively. Trentini et al. (2018)Trentini CP, Fonseca JM, Cardozo Filho L., Reis RR, Sampaio SC, Silva C. 2018. Assessment of continuous catalyst-free production of ethyl esters from grease trap waste. J. Supercrit. Fluids 136, 157-163. https://doi.org/10.1016/j.supflu.2018.02.018 carried out the reaction in two steps, using lipids from grease trap waste and ethanol, in which the ester content increased from ~70 to 72%. Regarding thermal decomposition (Figure 4b), this was higher than that found in the first reaction step, reaching values of ~3.8%. However, these decomposition values are still low, since numbers between 6-7% are reported in the literature for reactions with residual oils and the same temperature range as in this study (Abdala et al., 2014Abdala ACA, Colonelli TAS, Trentini CP, Oliveira JV, Cardozo-Filho L, Silva EA, Silva C. 2014. Effect of additives in the reaction medium on noncatalytic ester production from used frying oil with supercritical ethanol. Energy Fuels 28, 3122-3128. https://doi.org/10.1021/ef402253e ; Trentini et al., 2020Trentini CP, Postaue N, Cardozo-Filho L, Silva C. 2020. Waste oil/crambe oil blends for ethyl ester production under supercritical conditions. J. Supercrit. Fluids 163, 104889. https://doi.org/10.1016/j.supflu.2020.104889 ).

Figure 4.  Reaction products obtained in two-step reaction at 275 ºC, 20 MPa, oil-to-ethanol mass ratio 1:1.5 and applying 2 g of recycled catalyst () and new catalyst (): (a) ester content, (b) thermal decomposition, and content of (c) triglycerides, (d) diglycerides and (e) monoglycerides. Values are presented as mean ± standard deviation (n=2).

Figure 4c shows a reduction in the triglyceride content (TG), compared to the first reaction step (time 0), with > 99% consumption. In contrast, increases in diglyceride (DG) and monoglyceride (MG) contents were observed, (Figures 4d and 4e, respectively). The contents in intermediate compounds (DG and MG) obtained using a reaction step was 0.46 wt%, which increased to 2.50-2.53 wt%; while the content in triglycerides decreased from 4.34 to 0.19-0.36 wt%. Santos et al. (2018)Santos KC, Hamerski F, Voll FAP, Corazza ML. 2018. Experimental and kinetic modeling of acid oil (trans)esterification in supercritical ethanol. Fuel 224, 489-498. https://doi.org/10.1016/j.fuel.2018.03.102 reported the presence of ~12 wt% of intermediate compounds, in ~ 30 min, 279 °C and 20 MPa for the reaction between acidic oil and ethanol, and Trentini et al. (2020)Trentini CP, Postaue N, Cardozo-Filho L, Silva C. 2020. Waste oil/crambe oil blends for ethyl ester production under supercritical conditions. J. Supercrit. Fluids 163, 104889. https://doi.org/10.1016/j.supflu.2020.104889 reported that 11.23 wt% unreacted compounds remained from the reaction with residual oil/crambe oil mixture (27/75) at 300 °C, 20 MPa and 30 min. Thus, the low levels found for the components in this study indicate the high conversion of residual frying oil into ethyl esters. The esters present in the samples, together with the intermediate components and the decomposed components total the convertibility content of the residual frying oil.

The execution of the second reaction stage for both tested catalysts was effective, and resulted in an increase in the ester content, low decomposition and still low content of unreacted compounds, proving the efficiency of the application of this process. According to the ANP 45/2014 (Brazil) (ANP, 2014ANP. 2014. Agência Nacional do Petróleo. Resolução ANP N^o 45, DE 25.8.2014 - DOU 26.8.2014.) the limit for MG, DG and TG in biodiesel is 0.70, 0.20 and 0.20 wt%, respectively, so the values obtained for MG and TG are in accordance with the legislation.

3.4. Comparison to the literature

Table 2 shows the comparison of the results obtained in this study with reports regarding the use of heterogeneous catalyst under pressurized conditions using a continuous reactor to obtain esters, as well as research using RFO in pressurized conditions, without using a catalyst.

As seen in Table 2, for reactions conducted using catalysts and refined oils, ester yields close to this research were obtained; however, the use of residual frying oil represents an ecologically better, low-cost option (Hajjari et al., 2017Hajjari M, Tabatabaei M, Aghbashlo M, Ghanavati, H. 2017. A review on the prospects of sustainable biodiesel production: A global scenario with an emphasis on waste-oil biodiesel utilization. Renew. Sustain. Energy Rev. 72, 445-464. https://doi.org/10.1016/j.rser.2017.01.034 ). In the studies carried out by Jesus (2010)Jesus AA. 2010. Síntese de biodiesel em meio contínuo pressurizado empregando hidrotalcitas como catalisadores heterogêneos. Aracaju: Universidade Tiradentes. and Krohn et al. (2011)Krohn BJ, McNeff CV, Yan B, Nowlan D. 2011. Production of algae-based biodiesel using the continuous catalytic Mcgyan® process. Bioresour. Technol. 102, 94-100. https://doi.org/10.1016/j.biortech.2010.05.035 methanol, alcohol of better efficiency in transesterification, was used, due to the lower activation energy compared to ethanol, which implies a higher reaction rate. However, they were not able to obtain results higher to those reported in this study, in addition, the production of biodiesel through ethanol is more attractive, as it is a reagent obtained from renewable sources (Soares and Andreozzi, 2011Soares DZ, Andreozzi SL. 2011. Reflexões sobre o etanol e o biodiesel na matriz energética brasileira. Rev. Geográfica América Cent. 2, 1-17.). McNeff et al. (2008)McNeff CV, McNeff LC, Yan B, Nowlan DT,Rasmussen M, Gyberg AE, Krohn BJ, Fedie RL, Hoye TR. 2008. A continuous catalytic system for biodiesel production. Appl. Catal. A Gen. 343, 39-48. https://doi.org/10.1016/j.apcata.2008.03.019 and Mazanov et al. (2016)Mazanov SV, Gabitova AR, Usmanov RA, Gumerov, FM, Labidi S, Amar MB, Passarello JP, Kanaev A, Volle F, Neindre BL. 2016. Continuous production of biodiesel from rapeseed oil by ultrasonic assist transesterification in supercritical ethanol. J. Supercrit. Fluids 118, 107-118. https://doi.org/10.1016/j.supflu.2016.07.009 reported higher yields than this study, although the reactions were carried out at higher temperatures (440 and 350 °C, respectively).

The researchers Maçaira et al. (2011)Maçaira J, Santana A, Recasens F, Larrayoz MA. 2011. Biodiesel production using supercritical methanol/carbon dioxide mixtures in a continuous reactor. Fuel 90, 2280-2288. https://doi.org/10.1016/j.fuel.2011.02.017 , also using a catalyst, reported a lower yield than that obtained in this research for the reaction at ~200 °C, lower than the critical alcohol temperature, which may have influenced the low ester yield when compared to this study. The esters yield obtained in this study was high when compared to other studies (Table 2) that also used residual frying oil (Abdala et al., 2014Abdala ACA, Colonelli TAS, Trentini CP, Oliveira JV, Cardozo-Filho L, Silva EA, Silva C. 2014. Effect of additives in the reaction medium on noncatalytic ester production from used frying oil with supercritical ethanol. Energy Fuels 28, 3122-3128. https://doi.org/10.1021/ef402253e ; Fonseca et al., 2018Fonseca JM, Cardozo-Filho L,Teleken LG, Silva C. 2018. Ethyl esters from waste oil: Reaction data of non-catalytic hydroesterification at pressurized conditions and purification with sugarcane bagasse ash. J. Environ. Chem. Eng. 6, 4988-4996. https://doi.org/10.1016/j.jece.2018.07.044 ; Gonzalez et al., 2013Gonzalez SL, Sychoski MM, Navarro-Díaz, HJ, Callejas N, Sainebe M, Vieitez I, Jachmanián I, Silva C, Hense H, Oliveira JV. 2013. Continuous catalyst-free production of biodiesel through transesterification of soybean fried oil in supercritical methanol and ethanol. Energy Fuels 27, 5253-5259. https://doi.org/10.1021/ef400869y ; Trentini et al., 2020Trentini CP, Postaue N, Cardozo-Filho L, Silva C. 2020. Waste oil/crambe oil blends for ethyl ester production under supercritical conditions. J. Supercrit. Fluids 163, 104889. https://doi.org/10.1016/j.supflu.2020.104889 ). Fonseca et al. (2018)Fonseca JM, Cardozo-Filho L,Teleken LG, Silva C. 2018. Ethyl esters from waste oil: Reaction data of non-catalytic hydroesterification at pressurized conditions and purification with sugarcane bagasse ash. J. Environ. Chem. Eng. 6, 4988-4996. https://doi.org/10.1016/j.jece.2018.07.044 reported 74% esters through hydro-esterification at 300 ºC. However, a previous hydrolysis step was performed in this study. It is important to mention that not even the addition of co-solvent (n-hexane) and water to the reaction, as performed by Abdala et al. (2014)Abdala ACA, Colonelli TAS, Trentini CP, Oliveira JV, Cardozo-Filho L, Silva EA, Silva C. 2014. Effect of additives in the reaction medium on noncatalytic ester production from used frying oil with supercritical ethanol. Energy Fuels 28, 3122-3128. https://doi.org/10.1021/ef402253e and Gonzalez et al. (2013)Gonzalez SL, Sychoski MM, Navarro-Díaz, HJ, Callejas N, Sainebe M, Vieitez I, Jachmanián I, Silva C, Hense H, Oliveira JV. 2013. Continuous catalyst-free production of biodiesel through transesterification of soybean fried oil in supercritical methanol and ethanol. Energy Fuels 27, 5253-5259. https://doi.org/10.1021/ef400869y respectively, resulted in a higher yield than reported in this study. Therefore, the presence of the KF/clay catalyst in the reaction was essential to achieving high ester yield.

In this work, oil-to-catalyst mass ratios of 8:1, 6.4:1 and 5.4:1 (oil to ethanol mass ratio 1:1, 1:1.5 and 1:2, respectively) were used, and high performance was achieved with the application of a low cost and residual oil for the production of renewable fuel, and under milder temperature conditions. It should be noted that for most cases, reactions using catalysts seek to conduct the reactions under less severe conditions, as it allows to reduce energy consumption (Andreo-Martínez et al., 2020Andreo-Martínez P, Ortiz-Martínez VM, García-Martínez N, Ríos AP, Hernández-Fernández FJ, Quesada-Medina J. 2020. Production of biodiesel under supercritical conditions: State of the art and bibliometric analysis. Appl. Energy 264, 114753. https://doi.org/10.1016/j.apenergy.2020.114753.). In this way, the conditions reported in this study, which promoted high yield in esters, applying residual oil, ethanol and lower temperature are advantageous, highlighting the effectiveness of the KF/clay catalyst in the production of biodiesel. In addition, the catalyst used is easy to obtain and comes from clay, a cheap and natural source which can be of great interest for the various benefits linked to its use.

3.5. Catalyst characterization

The micrographs of the catalyst employed in the reactions, before and after the reactions of stability and two steps (recycled and new catalyst) are set forth in Figure 5. Note that the recovered catalyst samples (Figures 5b-d) resemble the catalyst before being used in the processes (Figure 5a), maintaining the rough-appearance characteristic of this process. Even after exposure to the conditions of temperature and pressure employed, the granules of the catalyst were not destroyed, which suggests that the stages of compaction and granulation were successful, and that the material performed well during the reaction time. Thus, the conditions employed in the studied reactions did not significantly modify the morphological characteristics of the catalyst, indicating that the compacting process was effective for the possibility of its application in pressurized and continuous processes.

The quantification of the elements present in the catalyst before and after the reactions is in Table 3, obtained from the semi-quantitative analysis of energy dispersive spectroscopy (EDS).

The presence of fluoride (19.4%) and potassium (16.5%) in the catalyst sample is due to the impregnation process with potassium fluoride salt. It can be observed that in the recovered samples there was a decrease in these components (mainly potassium), being more accentuated for the catalysts that were exposed for the longest time to the operational conditions (stability reaction and 2 stages with recycle). The activity of the catalysts found is due to the basic sites generated precisely in the impregnation with KF, since the active sites correspond to the fluoride ions themselves and to the negatively charged oxygen atoms located in their vicinity, and the presence of K⁺ cations distributed around the active sites generating the basicity of the catalyst (Alves et al., 2014Alves HJ, Rocha AM, Monteiro MR, Morwtti C, Cabrelon MD, Schwengber CA, Milinsk MC. 2014. Treatment of clay with KF: New solid catalyst for biodiesel production. Appl. Clay Sci. 91-92, 98-104. https://doi.org/10.1016/j.clay.2014.02.004 ; Boz et al., 2009Boz N, Degirmenbasi N, Kalyon DM. 2009. Conversion of biomass to fuel: Transesterification of vegetable oil to biodiesel using KF loaded nano-γ-Al₂O₃ as catalyst. Appl. Catal. B Environ. 89, 590-596. https://doi.org/10.1016/j.apcatb.2009.01.026 ; Endalew et al., 2011Endalew AK, Kiros Y, Zanzi R. 2011. Heterogeneous catalysis for biodiesel production from Jatropha curcas oil (JCO). Energy 36, 2693-2700. https://doi.org/10.1016/j.energy.2011.02.010 ). Thus, the results from the EDS analysis (Table 3) indicate that the leaching of these elements resulted in a drop in yield over the 8 h operation in the stability reaction (Figure 3). When the two-step reactions were compared (Figure 4), the catalyst submitted to reuse showed lower yield than the reaction with a new catalyst, showing the occurrence of leaching of these elements (F and K). The other elements present in the catalyst originated from clay, which is composed of several types of minerals and their composition can vary. The catalyst recovered after the reactions had a higher carbon content, which may be due to remaining oil residues and esters in the sample, even after washing. It is worth noting that the EDS results are semi-quantitative, and that the percentages shown in Table 3 represent the average composition of five selected regions within each analyzed micrograph.

Figure 6 s shows the infrared spectra for the catalyst samples before (a) and after the stability reactions (b), two steps with new catalyst (c) and two steps using recycled catalyst (d). The vibration bands that appear between 650 and 1200 cm^-1 correspond to Si-O and Si-O-Al vibrations, and are typical of quartz and montmorillonite clay minerals, with bands appearing between 700 and 900 cm^-1 to the octahedral layers of clay. These vibrations remain even on the catalysts subjected to the reactions. In the range of 900 to 1100 cm^-1 there is the characteristic SiO bond elongation, being more evident in the catalyst after use in the reactions (b-d), probably due to the leaching of the K⁺ ion, which is responsible for the catalyst activity. Thus, the structure of the clay becomes more evident and for this reason, the increase in the band attributed to the SiO bond present in the minerals of the clay occurs. The 3500 to 3800 cm^-1 range corresponds to the O-H elongation. The bands in the region of 1250-1750 cm^-1 correspond to the treatment of the clay with KF. Bands in the range of 1600-1700 represent H-O-H and correspond to adsorbed water, present only in sample (a).

In Figure 7 the XRD for the catalyst is presented before and after the use of stability reactions, two steps with new catalyst and recycled catalyst. The crystalline phases identified in the samples were montmorillonite (Na-Mg-Al-Si₄O₁₁), quartz (SiO₂) and albite (Na(AlSi₃O₈)), a type of feldspar. These phases were already predicted, as they are crystalline phases commonly found in bentonite clays. The peaks present in the catalyst (Figure 7a) remain similar in the catalysts recovered after use in the reactions (Figure 7b-d). There is also a crystalline phase in the catalyst, corresponding to the treatment with potassium fluoride, and it can be observed that the leaching shown in the EDS (Table 3) and FTIr (Figure 6) of the K ⁺ element is also evident in the XRD, mainly in the reaction which stood exposed for the longest operating conditions (stability, Figure 7b).

It is worth mentioning that there is some possibility for the development of a more stable catalyst in order to reduce the leaching of active species such as K⁺ and F^-. According to the results obtained in this work, given the reactor operating conditions (temperature, flow and pressure practiced), it is believed that there is an “excess” of K⁺ and F^-, which eventually are not interacting properly with the surface of the clay particles (weak interaction), in order to withstand the experimental conditions of temperature and pressure, which in turn, are not mild. Thus, it is necessary to conduct a future study to evaluate the amount of the KF saline precursor used in the impregnation, which can guarantee that the ideal content of K⁺ and F^- ions capable of forming stable active sites on the catalyst surface.