Biodiesel production evaluating the use and reuse of magnetic nanocatalysts Ni_0.5Zn_0.5Fe₂O₄ synthesized in pilot-scale

2.1 Nanocatalysts synthesis

The synthesis of Ni_0.5Zn_0.5Fe₂O₄ by combustion reaction involved the mixing of metal ion salts as oxidizing reagents (nickel nitrates - Ni(NO₃)₂·6H₂O and hexahydrate zinc - Zn(NO₃)₂·6H₂O, nanohydrate iron nitrate - Fe(NO₃)₃·9H₂O and urea - (NH₂)₂CO as a reducing agent to form a redox solution. The initial composition of the solution was calculated based on the total valence of the oxidizing and reducing reactants, using the propellants and explosives chemical concepts (Jain et al., 1981). The redox mixture of metallic nitrates and fuel was subjected to direct heating in conical reactors developed for the combustion synthesis that allows the production in different batches of 10, 100 and 200 g of the product (Costa and Kiminami, 2012). The nanocatalysts samples were designated by the nomenclatures PM, MM and GM (10, 100 and 200 g/product, respectively). To ensure the samples reproducibility on three different containers, we carried out 10 syntheses of each sample, PM, MM and GM.

2.2 Nanocatalysts characterization

Samples as synthesized PM, MM and GM were submitted to X-ray Diffraction (XRD), Textural Analysis (BET/BJH), Transmission Electron Microscopy (TEM), Magnetic Measurements (AGM) and Temperature-Programmed Desorption (TPD) characterizations. All characterizations were performed in triplicates to obtain a better accuracy of the results.

The determination of the present phases, crystallinity and the crystallite size of the MNP of Ni_0.5Zn_0.5Fe₂O₄ were determined from the diffraction data using a Bruker X-ray diffractometer (model D2 Phaser, Cu-Kα radiation), operating with tube of copper target at a voltage of 30.0 kV and 10.0 mA of current, with detector of 55D160. The crystallinity was determined from the ratio between the integrated area of the peak relative to the crystalline phase and the area related to the amorphous fraction using the Difract Eva software. The medium crystallite size was calculated from the X-ray enlargement line (d311) by deconvolution of the polycrystalline cesium secondary diffraction line (used as standard) using the Scherrer Equation (Sekar and Halliyal, 1998) with the Difract Eva software.

The surface area measurement was performed by the nitrogen/helium adsorption method developed by BET. A new equipment model Nova 3200 brand Micromeritics was used. This technique was also used to determine the average size of particle agglomerates (equivalent spherical diameter) by means of the Reed Equation (Reed, 1996). Theoretical density (ρ) used was 5.361 g/cm³ for Ni-Zn ferrite, obtained according to the crystal data sheet JCPDF 08-0278 from the Shimadzu program data package. The pore volume and the pore diameter were determined by the theory developed by BJH. The morphological aspect of the samples was obtained in a transmission electron microscope (TEM).

The magnetic hysteresis cycles (M x H) of the samples were obtained by an Alternate Gradient Magnetometer (AGM). By means of the M x H curves it was possible to determine some magnetic parameters values, such as: coercive field (Hc), remnant magnetization (Mr), saturation magnetization (Ms). The saturation magnetization was determined by fitting the field data applied to the function M = Sm (1–α/H), where M is the magnetization, Sm is the saturation magnetization, α is the fitting parameter and H is the applied field. Then the magnetic characteristics were obtained from the hysteresis graph, observing the curves behavior in the origin proximities of the Cartesian plane.

The active sites analysis was performed by the temperature-programmed desorption (TPD) method of NH₃ molecules. In TPD studies, the solid, previously equilibrated with a gas, under well-defined temperature and partial pressure conditions, is subjected to a heating under temperature and flow schedule of an inert gas, that is, a flow gas that flows out on the sample, by monitoring the continuous gas desorption.

2.3 Catalytic performance

The chemical processes used to obtain biodiesel were the reactions of transesterification and esterification of commercial soybean oil, both by methyl and ethyl routes. For the esterification reactions the raw material of the soybean oil fat was used and artificially acidified with 15% by weight of oleic acid, in order to simulate high acidity, taking as reference some regional fats (castor oil, slaughterhouse fats and frying oil), which present contents between 10 and 20% of free fatty acids. Both reactions were conducted in a stainless steel reactor, which is equipped with a cup of borosilicate with useful volume of 80 mL, pressurized, composed of a duct for thermocouple input and coupled to a manometer. Heating and stirring of the system were promoted by a plate model IKA C-MAG HS 7, the stirring was by a magnetic bar of approximately 2.5 cm.

The catalytic reactions were carried out using 10 g of fatty acid, temperature of 180 °C, oil:alcohol molar ratio of 1:12, with 2% of catalyst for 1 h. These conditions were originated from a reaction optimization, in order to select the reaction and the experimental conditions that offered better conversions of fatty acids to esters. The tests were performed with the PM sample and reproduced in triplicate.

Through the results obtained with the PM nanocatalyst, the methyl and ethyl esterification reaction was selected for continuity of the catalytic tests, this time involving the three samples studied as nanocatalysts (PM, MM and GM), varying the amount of fatty acids from 10 to 30 g, the oil:alcohol molar ratio from 1:12 to 1:15 and the catalyst percentage from 2 to 3%, in order to further maximize the catalytic activity results. These tests were reproduced in septuplicate.

The obtained products were analyzed on a gas chromatograph Varian 450c with Flame Ionization Detection (FID) on a short capillary column of stationary phase Varian Ultimetal “Select Biodiesel Glycerides + RG” (15 m × 0.32 mm × 0.45 μm). The detector temperature was 250 °C and the injector temperature was 240 °C. The oven temperature was programmed from 150 up to 260 °C at a heating rate (ramp) of 10 °C/min. The carrier gas used was H₂ of high purity. The samples preparation consisted in the dilution of 50 mg of these in 5 mL of standard n-hexane UV/HPLC (Vetec P.A./A.C.S.) and subsequent injection of 1 μL of the solution into the equipment. The standard used was the internal standard provided by Varian Inc.

3.1 X-ray diffraction (XRD)

The X-ray diffractograms of the nanocatalysts Ni_0.5Zn_0.5Fe₂O₄ obtained by combustion reaction in different containers (10, 100 and 200 g of production) are shown in Fig. 1(a, b and c). X-ray diffraction curves has shown the presence of the main characteristic peaks of the inverse spinel structure, evidenced by the appearance of the main peak 2θ = 35.5°, according to crystal card JCPDF 52-0278, regardless of the type of vessel used (10, 100 and 200 g of production). All the diffractograms has shown peaks with high intensity and high basal width for all the reflections, revealing that the synthesized materials are crystalline and with nanostructural characteristics.

Close

Fig. 1

X-ray diffractograms of Ni_0.5Zn_0.5Fe₂O₄ nanocatalyst: (a) PM, (b) MM and (c) GM.

Export to PPT

Moreover, it can observe that the diffractograms presented in addition to the ferrite spinel Ni-Zn major phase, the presence of characteristic peaks of hematite (Fe₂O₃) and zinc oxide (ZnO) segregated phases. The phase Fe₂O₃ was identified by the standard card JCPDS 40-1139, and the phase ZnO was identified by standard card JCPDS 36-1451.

As for the presence of these segregated phases, it can say that they were generated possibly by the rapidity of the syntheses, which did not allow the complete formation and crystallization of the spinel phase under study, which does not compromise the syntheses efficiency, but rather indicates that it can change the synthesis conditions and obtain a single-phase product, if it is essentially necessary. However, another possibility for the second phase traces formation can be attributed to the maximum temperatures reached in the combustion syntheses, because the maximum temperatures reached in all the reactions were very close, not exceeding 800 °C.

Then it would probably be necessary, higher temperature to effectively meet the reagents demand to be dissolved for single-phase formation (spinel). Sinthiya et al. (2015), when they synthesized by the hydrothermal method the ferrite ZnFe₂O₄ using different fuels, observed the presence of ZnO as a segregated phase, and also attributed to the temperature.

The batch synthesis reproducibility in pilot scale and the efficiency of the combustion method were demonstrated by the diffraction curves reproduced, since the presence of the majority inverse spinel phase of the Ni-Zn ferrite was evidenced with intensity and basal width of the similar diffraction peaks, indicating good crystallinity and crystallite size very close, despite the discrete traces of hematite (Fe₂O₃) and zinc oxide (ZnO). Moreover, the presence of these oxides may possibly act beneficially for many applications, for example in catalytic reactions, as active metals and promoters.

Structural data such as crystallinity and crystallite size, which were obtained from the XRD results of the nanocatalyst Ni_0.5Zn_0.5Fe₂O₄ are shown in Table 1.

In general, it was observed a low variation in crystallite sizes that varied from 31.1 to 42.6 nm, indicating that the identified basal reflections diffracted with good uniformity, suggesting that the nanoparticles of Ni_0.5Zn_0.5Fe₂O₄ presented characteristic of low anisotropy in diffraction. Thus, it was possible to obtain crystallites at a nanometric scale, without further sintering, confirming the efficiency of the combustion reaction method for the synthesis of these materials. Leal et al. (2018) when they synthesized spinel type MFe₂O₄ ferrites by combustion reaction, also proved the efficiency of the method in obtaining powders with nanometric characteristics, presenting crystallite sizes varying from 12 to 45 nm. Thakur et al. (2016), they produced the ferrite Ni_0.5Zn_0.5Fe₂O₄ by coprecipitation, and only reached crystallite size values of 38.36 to 42.16 nm, when they sintered the material.

Although the crystallite sizes of the three samples were close, the value determined for PM was 23% higher than the value of the MM sample, and approximately 11% in relation to the value of the GM sample, which in turn were closer to each other. This is possibly related to the combustion temperatures reached during the syntheses and can also be attributed to the variation in the amount of the segregated phases reported for the samples.

These assignments are consistent with those reported in the literature, for example, the study of the ferrite Ni-Zn developed by the authors Srinivas et al. (2016), when they concluded that temperature (heat treatment mode), secondary phases formation, preparation method, additives nature, among others, influence crystallite sizes. Saba et al. (2011), they observed an increase in crystallite size and then a decrease with concentration increase of ions Ni²⁺ in thin films of Ni-Zn. They reported that this variation is due to the formation of the secondary phase of α-Fe₂O₃ at higher concentrations of Ni²⁺.

In relation to the crystallinity of the samples, it was also observed that they presented close values ranging from 55.6 to 69.1%, due to the structural organization and the crystals periodicity within the network. In all cases, the combustion reaction process proved to be an excellent technique to obtain crystalline magnetic catalysts, at nanometer scale and large batch.

3.2 Textural analysis by nitrogen adsorption (BET)

As a result of the textural characterization, the adsorption/desorption isotherms of N₂ for samples PM, MM and GM are shown in Fig. 2. The study of the adsorption phenomenon was done aiming to obtain information concerning the specific area and the porous structure of the samples, since the construction of the adsorption isotherm curve is of fundamental importance, because its format reveals details about the morphological characteristic, that is, the curve shape is a function of the type of material porosity.

Close

Fig. 2

Adsorption/desorption isotherms of N₂ in triplicates of the samples (a) PM, (b) MM and (c) GM.

Export to PPT

The surface area of a catalyst determines the reactants accessibility to the active sites, and the magnitude of this area determines that a catalyst satisfactorily promotes a chemical reaction. In majority particles their surfaces are quite irregular and such irregularities can range from atomic scale to relatively large pores. Thus in heterogeneous catalysis the gaseous adsorption method of N₂ is widely used to study the textural properties of a catalyst and is based on the determination of the amount of an inert gas required to form a mono molecular layer on the catalyst surface, at a constant temperature. Thus, the catalyst surface area is equal to the area to be occupied by each gas molecule under pre-established conditions.

Then, regarding to the particle size obtained by the surface area, there is a gas passing between the particles, then, so thinner they are and more interparticle porosity exists, the reading made in the gas passage will indicate a larger surface area, which is the contact area. This context refers to the importance exerted by the porous characteristic of the studied samples.

Then the curves of Fig. 2 show that the samples PM, MM and GM presented characteristics of mesoporous materials, however, PM and GM characteristics were more similar in some parameters when compared to the sample MM. Then PM and GM presented surface structures with profiles of the isothermal adsorption curve, which according to the IUPAC classification (Sing et al., 1985), they are classified as type V, suggesting a mesoporous material characteristic (pores with rays varying from 10 to 250 Å). Also analyzing the hysteresis form that corresponds to different pore geometries, it can say that the nanoparticles of cited samples are represented by hysteresis loop type 3 (H3) (constituting the formation of wedge-shaped pores, cones or parallel plates), and they are also associated with hysteresis type 2 (H2) (open and closed cylindrical pores with strangulations, bottle-like morphology).

The sample MM has shown a surface structure with a profile of the isothermal adsorption curve framed in type V (solids with mesopores) and together type III (nonporous solids associated with mesopores). The hysteresis form was represented by hysteresis loop type 3 (H3).

Table 2 shows the specific surface area (S_BET), particle size (D_BET), pore volume (Pv), pore radius (Pr), relation between particle size and crystallite size (D_BET/Cs) and the adsorbed volume of N₂, corresponding to samples PM, MM and GM.

The samples presented high surface areas. The highest medium specific surface area of 65.29 m²/g was obtained by PM, and the lowest surface area was of 50.94 m²/g, obtained by MM. Consequently, these samples presented, respectively, the lowest (17.14 nm) and the largest (21.97 nm) particle size value. Comparing the particle size results presented by the samples obtained by combustion reaction in different production batches with the particle size reported by Ghasemi and Mousavinia (2014), when synthesizing the same system by the sol–gel method, whose values were between 1 and 4 μm (1000–4000 nm), the technique efficiency is verified by combustion reaction in the pilot-scale nanomaterials production.

It was also sought to correlate the pores volume (which are the voids between the nanoparticles agglomerates, which form their structure) with their respective adsorbed volumes of N₂. Fig. 2 shows that sample MM presented the highest volume of adsorbed N₂, with a value of 109.98 cm³/g, followed by GM and PM, with values of 97.97 and 95.60 cm³/g, respectively. This is probably related to the porosity of each of them, that is, as larger the porosity, as greater the volume of adsorbed N₂, thus confirming the values of obtained pore volumes, which averaged 0.171, 0.152 and 0.162 cm³/g.

Consequently, in the same order, the pore radio values that were 48.042; 32.630 and 42.754 Å, that is, as larger they were, as larger the respective associated volume. Therefore, this corroborates the pore shapes obtained by PM and GM, which according to their hysteresis presented more varied geometries, and even closed pores, which hinders the adsorption/desorption process, possibly leading to a lower porosity, when compared to the sample MM.

Regarding the D_BET/C_S ratio (particle size/crystallite size), it can be said that the closer to 1, the more it suggests that the particle size is close to the size of the crystal, thereby indicating that the particle tends to be monocrystalline. And that the greater than 1, more polycrystalline is the material. However, in the present study this D_BET/C_S ratio was below 1, and this is probably related to the differences between the techniques and methodologies used in the collection of these results. In the X-ray diffraction technique, the Scherrer equation (Sekar and Halliyal, 1998) was used, and the textural analysis by adsorption/desorption of N₂ was performed based on BET/BJH methodologies (Reed, 1996). For the calculation of particle size (D_BET), it was taken into account the surface area of the sample as a whole, ie the crystalline and non-crystalline regions present, in which the N₂ gas flows through and fills all the voids present in the sample, leading to a particle size value closer to the real one. However, for the calculation of crystallite size (C_S), only the main peak 311 of the crystalline phase Ni_0.5Zn_0.5Fe₂O₄ was considered, disregarding the presence of the segregated and/or deleterious phases present in the sample, besides the possible presence of non-crystalline (amorphous) material, possibly leading to less precise (not near real) results of crystallite sizes and with higher values.

Therefore, the obtained results are very attractive for catalysis, because the materials with a greater amount of active chemical sites obtained by the expansion of its surface area, that is, by the particle size reducing to a nanoscale has been the target of several researches, because the increase in the surface area/volume ratio favors high chemical reactivity and high catalytic activity.

3.3 Transmission electron microscopy (TEM)

Fig. 3 shows the light field micrographs of the nanocatalysts PM, MM and GM under study. In these micrographs are observed agglomerates of weakly interconnected nanoparticles with average sizes of 325, 540 and 330 nm, respectively, and consisting of nanometric particles having both irregular and approximately spherical shapes, with average sizes of 18, 23 and 25 nm, respectively. This agglomeration state is probably related to the method of chemical synthesis adopted.

Close

Fig. 3

Transmission electron microscopy of the samples: (a) PM, (b) MM and (c) GM.

Export to PPT

Particle size values were considerably lower than reported by Wang et al. (2016), which synthesized the ferrite Ni_0.8Zn_0.2Fe₂O₄ by the sol gel method, and observed much larger particles with size around 50 nm. Abbas et al. (2014), when studying the nanoferrites Ni_0.5Zn_0.5Fe₂O₄ synthesized by two different routes, the sonochemical method and the polyol, reported that the average size obtained by polyol and sonochemical method were 25 and 40 nm, respectively. In this way, stands out once again the efficiency of the combustion reaction technique to obtain particles at nanometric scale, in addition to its pilot-scale production.

These results also confirm the particle sizes obtained in the N₂ adsorption (BET) textural analysis, whose values ranged from approximately 17 to 22 nm, proving the nanometric characteristic of the samples.

3.4 Magnetic measurements

In the catalysis, the most evident points in the nanomagnetic materials are the high surface area, the ease of recovery through magnetic field, the reuse and the presence of basic or acidic active sites (Guo et al., 2012; Jing et al., 2015). In addition, heterogeneous catalysts are usually separated from the reaction medium by intensive filtration and/or centrifugation steps, thus much research has been devoted to the development of heterogeneous magnetic catalysts which are readily separable. Therefore, the fact that the nanocatalysts PM, MM and GM have high surface areas (which also favors the presence of active sites), as well as having magnetic properties (which effectively helps to prevent catalyst loss and to increase the recovery rate during the separation process), this places the magnetic catalysts as more advantageous than the conventional heterogeneous catalysts as potential products in the catalytic area.

Fig. 4 shows the dependence of the magnetization M as a function of the applied magnetic field H for the samples PM, MM and GM resulting from the synthesis by combustion reaction in different batches. The curves were plotted with their respective magnifications to visualize the determination of remnant magnetization (Mr) and coercive field (Hc). The magnetic properties of ferrites are appreciably dependent on structure, composition, defects, crystallite size, internal stress and cation distribution.

Close

Fig. 4

Hysteresis curves M x H of samples (a) PM, (b) MM, (c) GM and their respective magnifications for determination of Mr and Hc.

Export to PPT

It was observed that the three samples have hysteresis curves of ferrimagnetic behavior, which are characteristics of soft magnetic materials, which magnetizes and demagnetizes with low field values, due to their small values of remnant magnetization and coercivity, but different of zero, thereby revealing the complete formation of the narrow magnetic hysteresis cycle.

In Table 3 it has the magnetic parameters (saturation magnetization (Ms), remnant magnetization (Mr) and coercive field (Hc)) calculated from the hysteresis curves of the samples PM, MM and GM.

The maximum saturation magnetization value for the sample PM was of 36.5702, for MM was of 20.9721 and for GM was of 19.8312 emu/g. The maximum magnetization presented by PM was 42.65 and 45.77% higher than those presented by MM and GM respectively. These magnetization values for PM, MM and GM were 54.29; 73.78 and 75.21% lower than the value of 80.0 emu/g reported by Jacobo and Bercoff (2016), when they synthesized by the sol gel method the ferrite Ni—Zn to investigate the structural and electromagnetic properties of this pure material and when in replacing to the yttrium ions.

This discrepancy is mainly due to the type of method employed for the synthesis, where the combustion reaction was considerably more efficient for nanoscale production, whose particle sizes (17.14; 21.97; 17.44 nm) were lower to the value reported for the sol gel method (350 nm). This statement can be supported by the report by Ramesh et al. (2016), that ferrites Ni—Zn are excellent soft magnetic materials and that it is well known that ferrite properties are highly sensitive to the preparation method, sintering conditions and chemical modifications.

According to Das and Singh (2016), in general, magnetic properties depend on several factors, such as the size, shape and cations distribution. These authors obtained particle sizes of 40 to 110 nm for the ferrite Ni_0.5Zn_0.5Fe₂O₄ synthesized by the self-combustion method, whose magnetization was 73.13 emu/g, which value is higher than the magnetization of 59.0 emu/g reported by Masrour et al. (2014) for particles Ni_0.5Zn_0.5Fe₂O₄ produced in bulk by the solid state method. The authors mentioned that the magnetization increase can be explained by changes in the interchange interactions between the tetrahedral and octahedral subunits due to crystallinity, particle shape and magnetization direction.

Therefore, the magnetization values of this work in function of the obtained particle sizes are consistent with the magnetization and particle sizes reported by other authors, because the magnetization is influenced by the particles size. This happens because for different synthesis methods, the magnetization occurs according to the cations distribution in the available interstices, which affects the exchange interactions and, therefore, the total magnetic moment of the ferrite.

This can be supported by Bohr's magnetic moment, which is related to greater or lesser occupation of ions at the octahedral sites (spin down or spin up, respectively), causing increase or reduction of the total magnetic moment of the network. However, as reported by Castro et al. (2014), on the other hand, extrinsic characteristics such as grain and/or particle size influence the magnetic domain area and may contribute to the magnetization increase because as larger the particle and/or grains size as lower the number of barriers, favoring a greater magnetization.

Thus, it is evidenced that the samples synthesized by combustion reaction in different batches are extremely nanometers, have good saturation magnetization and are very reactive, which is perfectly adequate for many applications, such as catalysis.

3.5 Thermal analysis by temperature-programmed desorption (TPD)

In the catalysis area, the MNP can act as promising catalysts, because the manipulation of nanoscale structures leads to obtain nanocatalysts with a greater number of active chemical sites, thus promoting an expansion of the surface area. Netto et al. (2013) they reported that nanoparticles have large surface area as well as high mass transfer area, ideal criteria, for example, for a catalysis support.

The reactivity of the catalysts surface is a direct consequence of its intrinsic characteristics as well as of its synthesis method; however, the reactivity can be affected by conditions that are imposed on these materials to prepare them for a given application. Therefore, a better accuracy of surface reactivity can validate the materials as catalytic active, and this is done with some specific investigations, such as to measure the surface area, to study the shape and distribution of the pores, and to study the acid and basic sites present in the surface. Therefore, the structure, morphology, chemical and thermal state and surface reactivity exert a strong interaction with the nature, quantity and strength of the active sites.

Table 4 shows the results obtained from the TPD analysis. It should be noted that the final temperature results correspond to the average of three results obtained from GM, as well as the volume of adsorbed NH₃ corresponds to the average of three results obtained from this same sample after respective correction of the results, considering the deconvolution of the desorption peaks.

Thus, it appears that MNP are composed of weak, moderate and strong acid sites, but almost 88% of the volumes of their sites are of a weak and moderate nature, since the results obtained from the TPD, where the desorption events of NH₃, considered as the acidic sites present in the MNP, occurred at the final temperatures of 197, 309 and 478 °C, with volumes of 2.01; 1.18 and 0.44 mLg⁻¹, respectively. Therefore, the total volume of acid sites was of 3.63 mLg⁻¹.

Then we can say that the nanoparticles of the samples PM, MM and GM, synthesized by combustion reaction, which have the same chemical composition, structure and morphology, are formed by acid sites. Kurian et al. (2015) they investigated the effect of the preparation method on the structure, magnetism and acid properties of cobalt ferrite nanoparticles synthesized by two different methods (co-precipitation and sol–gel). The authors confirmed the presence of weak, moderate and strong acid sites in the nanoparticles produced by both methods and related the amounts of the acid sites and their variations as a function of the particles agglomeration. The authors justified this according to the results obtained from XRD and SEM.

3.6 Selection of the catalytic process

Fig. 5 shows the results of the catalytic tests obtained in the transesterification and esterification reactions by means of the methyl and ethyl routes of the soybean oil to obtain biodiesel. The reactions were obtained at 180 °C with a molar ratio of 1:12, in the presence of 2% of the sample PM, and reaction time of 1 h, in order to select the process in which MNP of Ni_0.5Zn_0.5Fe₂O₄ presented more active.

Close

Fig. 5

Results of the catalytic tests with the sample PM, in the methyl and ethyl transesterification and esterification reactions.

Export to PPT

In general, the analysis of these results confirms that the sample was catalytically active for the reactions of methyl and ethyl transesterification and esterification of soybean oil, since they presented average efficiency (between the methyl and ethyl route), in both investigated reactions, of 12.5 and 84.6%, respectively, that is, the reactions conducted without the presence of the sample (blank test) were higher, where the conversion was on average of 2.3 and 16.2%, respectively.

However, it is observed that the catalytic activity was significantly more satisfactory in the esterification reaction, to the detriment of the transesterification reaction. It can also observe that as in transesterification as in esterification, the nanocatalyst was more active when the alcoholizing agent was methanol. This may have occurred because the methanol is most reactive, whereas the ethyl esters have a higher affinity for glycerin, making it difficult to separate them. That is, these results can be attributed to the fact that, the transesterification and esterification reactions by the ethyl route is significantly more complex than by the methyl one, because with the increase of the carbonic chain of the used alcohol, greater will be the spatial impediment of alcohol with the triacylglyceride.

Taking into account the conversion values in the esterification, it can be inferred that there was a difference of approximately 15% in favor of the methyl route. In transesterification, there was a difference of approximately 21% in the conversion values in favor also of methanol. This may have occurred because, in relation to alcohol, methanol is the most used in these reactions for reasons of physical and chemical nature (short chain and polarity), leading to higher conversions.

Then, the combination of these characteristics of short chain and polarity, coupled with the fact that methanol is also free of water, makes it easier to separate the esters and glycerin, leading to greater conversions. Thus, possibly the water present in the ethanol interfered, for example, in the transesterification reaction, reducing the amount of transesterified product, and provided the emulsion formation in the esters mixture due to the vegetable oil saponification.

Therefore, the alcohol type may possibly be attributed as one of the factors for the disparity in conversions from the transesterification and esterification reactions. Murugesan et al. (2009), they reported that ethyl esters present greater difficulty in formation when compared to methyl esters; biodiesel produced from ethanol has the kinematic viscosity higher than the biodiesel produced from methanol. The authors also reported that during the transesterification reaction there may be the emulsions formation, due to the non-miscibility of the ethanol with triacylglycerides, which makes it difficult to separate and purify the esters.

Although ethanol is also polar in nature because it has a hydroxyl group, it is an oil solubilizing agent, weaker than for example methanol and hexane. Dagostin et al. (2015) they reported that ethanol has limited solubility with triacylglycerols and is able to solubilize the oil not only present in the matrix, but also water and other polar components. If the hydrated ethanol is applied as a solvent, its ability to solubilize triacylglycerides is further reduced, so, gradually, the efficiency process is limited.

This fact assumes that probably the esterification results have been highlighted, because in the transesterification may have occurred pores clogging and active surface poisoning caused by organic species, such as triacylglycerols and glycerol, which led to the decrease of the active sites for the reagents, reducing the conversions of the oil into esters.

On the other hand, a factor that must be attributed as a preponderant to the excellent catalytic activity obtained in the esterification reaction is related to the acidity of the MNP given by TPD (that presented on its surface weak, moderate and strong acidic sites, considered as propitious to promote an efficient catalytic activity), added to the soybean oil acidification used as raw material, since for the transesterification reactions neutral soybean oil was used. A previous acidity indication of the reactional product (biodiesel) of the esterification was verified by the pH measurement of the washing water, which confirmed the simulation efficiency of an acidic raw material promoted by the oleic acid addition.

Later, the acidity of the soybean oil with oleic acid mixture was confirmed by titration, presenting acidity of 0.360 ± 0.006 mg KOH/g, thus meeting the maximum limit of 0.5 mg KOH/g, established by law instrument as biodiesel quality standard. Martins et al. (2016) they reported that the quality of the oil used to produce biodiesel, especially in terms of acidity, is an important index to be appreciated, and should be within specifications, because high acidity, for example, can lead to corrosion or deterioration of the engine. Thus, it is important to analyze the oil acidity and then compare the results with the pre-set parameters.

Because of this, and still in relation to the samples acidity, it is possible to consider the presence of Brönsted and Lewis’s acid sites. It is worth mentioning that although the literature reports that the Brönsted’s acid sites presented in the catalysts are more suitable for the esterification reaction (Di Serio et al., 2008), the catalytic activity may be related to the presence of Lewis’s acidic sites, even in a small amount. For the esterification reaction, Lewis’s acids may be more suitable than Brönsted’s acids, especially because in the presence of these sites, dehydration and racemization of the involved alcohol can be avoided (Narasimharao et al., 2007).

In order to evaluate the performance of the three studied samples (PM, MM and GM), the results of this study showed that the PM sample exhibited a greater efficiency in the esterification, under different reaction conditions, whose results are plotted in Fig. 6. The reactions were performed maintaining the same temperature of 180 °C and the same reaction time of 1 h, varying only the molar ratio from 1:12 to 1:15, as well as the amount of nanocatalyst sample, from 2 to 3%.

Close

Fig. 6

Results of the catalytic tests for the samples PM, MM and GM on methyl and ethyl esterification.

Export to PPT

In general, the obtained results express the direct effect promoted by the oil:alcohol molar ratio in the esterification reactions. Firstly, the molar ratio increase was an extremely important variable because it resulted in a very considerable increase in the conversion rate, possibly due to the greater glycerol solubilization remnant by the alcohol. Then the amount of alcohol was predominant for the excellent obtained results, and these results can be understood like have done a more adequate alcohol concentration to execute the esterification reactions.

So much so that even the tests carried out without the presence of the nanocatalyst PM (blank test), realized at a molar ratio between oil and alcohol of 1:12, whose conversions were of 17.0 and 15.4%, increased of about 26% in the methyl route and of 9% in the ethyl route, when compared to the blank tests in ratio of 1:15, whose conversions were 42.70 and 24.32%, respectively. The esterification can be self-catalyzed due to the presence of Brunsted’s functional acids, although weak, in fatty acids and for this reason the catalytic activity should always be compared to a reaction conducted without the presence of the catalyst.

On the other hand, the nanocatalyst increase was also an important variable that contributed to the conversions increase, because probably it propitiated that more nanoparticles per volume were dispersed in the reactional medium where they were involved, since being magnetic a good part of them were retained in the magnetic bar used for stirring.

It was observed that methyl and ethylic routes presented excellent results for the three samples PM, MM and GM as nanocatalysts, whose conversions were respectively 99.54 ± 0.16; 97.50 ± 0.34 and 99.42 ± 0.40% in the methyl route, and 98.59 ± 0.13; 96.70 ± 0.54 and 98.38 ± 0.18% on the ethylic route. Thus, taking into account the blank tests, it is evidenced the strong influence that the presence of nanocatalysts imposed on the conversion of soybean oil to esters, thus revealing the high performance of the studied samples as catalysts for biodiesel production.

Thus, the conversion results, which were all above the minimum limit required for the biodiesel use, which is 96.5%, validate both routes as promising. However, it should be emphasized that the choice, as well as the care with the limits or amount of alcohol involved in the reactions, should be carefully analyzed, because the recovery system of unreacted alcohol is an important variable in the biodiesel production when we have to decide which route to use. For example, in the ethyl route the ethanol recovery makes this process more expensive regarding to energy consumption, to the detriment of the methyl route, due to the dehydration stage of the ethanol in order to be returned to the reaction. Moreover, according to Lora et al. (2012), ethanol must be completely anhydrous, because any presence of water greater than 2% can cause the chemical reaction reverse, and also the triacylglycerides molecular breakdown.

In addition, the type and quantity that make up the mixture species to obtain biodiesel, such as oil, alcohol and catalysts, are closely related to the independent variables such as time, temperature and agitation, thus the properties of the final product are strongly influenced by the involved reactional conditions.

The excellent obtained results can also be attributed to the texture and morphological characteristics presented by these materials, which have high surface areas, composed of mesopores and with presence of active acid sites, which culminated in the high conversion values, because such characteristics are structures for catalysis. The reactions using the samples PM and GM showed conversions a little larger than the reactions with the sample MM, this possibly is related to their respective surface areas, since it is a crucial parameter in catalysis, because it is on it that all reaction is processed, as well as the type and amount of active acid sites present in the nanoparticles.

In addition, it may be that part of the reaction has possibly occurred on the pores surface, and less in its interior, due to its surface shape. This is supported by the argument that in practice, a catalyst does not have its energetically homogeneous surface, that is, with all its equivalent adsorption sites and with the same amount of energy to interact with the reagent molecules. Therefore, that is why in catalysis it is necessary to determine the actually active surface, usually constituted by a set of atoms called sites, which have catalytic activity and are accessible to the reagents (Silva et al., 2008).

Also, the catalytic efficacy of the nanocatalyst can be attributed due to its unique structural properties beyond its thermal and chemical stability. A spinel is formed by the distribution of divalent and trivalent metal ions between the tetrahedral and octahedral coordination sites resulting from the cubic and narrow oxide ion packing. The catalytic activity of the spinel ferrites depends exclusively on the redox properties of the metal ions present in the octahedral sites, since the surface of the spinel structure contains mainly octahedral sites and, consequently, its catalytic activity is directly related to the octahedral cations. Thus, the catalytic activity of these materials is due to the ability of the metal ions to migrate between the subnets without altering the crystalline structure. These properties make this material an efficient catalyst for many chemical reactions (Dantas et al., 2017; Nair and Kurian, 2017).

Then, from the obtained results, it is evident that through combustion reaction, it is possible to start the production from a small scale of 10 g for a production in large scale of 200 g/product, without any loss in the structure of these materials for application in catalysis, in the field of biodiesel production. In addition, the conversions results place the obtained products above the minimum quality limit required by law to use biodiesel in diesel engines. Both esterification reactions by methyl and ethyl route showed excellent conversion values, thus demonstrating the high efficiency that the samples PM, MM and GM can offer to catalyze the biodiesel production reactions, even at the industrial level.

The promising results of achieved catalytic activity corroborate Buller and Strunk (2016) report, that nanostructured materials are widely adequate to face the pressing challenges associated with energy conversion, because they work effectively in the major domains of energy conversion investigation (chemical, thermal, electrochemical and solar). These authors reported that in many cases the catalyst activity and their nanostructure are inseparably correlated.

3.7 Biodiesel characterization

Other quality specifications, in addition to the conversion of triacylglycerides to esters (which is within the standards established by current legislation), should be met as criteria to apply this biodiesel at industrial level, such as viscosity (40 °C), density (20 °C), acidity and iodine index, flammability, peroxides, among others. The first four specifications were investigated for soybean biodiesel catalyzed by the sample GM, and the results are presented in Table 5.

These obtained values place the soybean biodiesel produced within the specifications applicable to the commercial biodiesel quality standards, governed by ANP, ASTM and EN standards.

3.8 Recovery and reuse

Regarding the recyclability of heterogeneous catalysts, it is said that it is one of the most important constraints to consider, and in this context, those with inherent magnetic characteristics are more advantageous because they promote different results when it comes to their recovery and reuse.

Fig. 7 shows the excellent magnetic response of the sample GM, synthesized in large scale, where its nanoparticles involved in the oily reaction medium respond promptly to the stimuli of an external magnetic field (magnet), facilitating its recovery after each reaction, by means of a simple, fast and efficient method.

Close

Fig. 7

Magnetic response of the nanoparticles of the sample GM against a magnet stimulus.

Export to PPT

Therefore, this magnetic property contributes effectively to avoid the nanoparticles loss and to increase their recovery rate during the separation process, because these materials are normally separated from the reaction medium by means of intense filtration and/or centrifugation steps, then, many investigation has been devoted to develop magnetic nanoparticles that are easily separable, and this work not only effectives this research, but it also does the differential to obtain these nanoparticles with highly nanometric characteristics and in large scale.

Bet-Moushoul et al. (2016), they used heterogeneous nanocatalysts CaO—Au to produce sunflower biodiesel, and obtained conversions up to 97% in transesterification. After nanocatalyst recovery by filtration and washing steps, a conversion of 89% was obtained after 10 reuses. The authors attributed the reduction of the catalytic activity to the catalyst solubilization in methanol during the recovery, because it caused loss of catalyst and of the active sites. They also attributed the formation of triacylglycerides and glycerol, which filled the pores, causing the active sites decrease, which consequently reduced the yield.

MNP were reused in the methyl esterification reactions under the same optimized conditions. The conversions results obtained on reuse are illustrated in Fig. 8.

Close

Fig. 8

Results of the reuse tests for the sample GM in the methyl esterification.

Export to PPT

After three cycles, there was no significant loss of efficiency in the catalytic activity, which presented a mean conversion of 98.95 ± 0.44%. Therefore, the magnetic nanocatalyst sample is economically viable for practical application from the industrial point of view. The magnetic nanoparticles recovery for reuse in catalytic processes, in the most varied fields, has been highly reverberated in the specialized literature. In the biodiesel production field, some authors have reported excellent performances, for example, Liu et al. (2016), which obtained 98.3% of conversion in the transesterification, using the magnetic nanoparticles of the catalyst MgFe₂O₄@CaO. The catalyst was recovered by external magnetic field and reused 5 times, and still showed activity above 89%. Feyzi and Norouzi (2016), using magnetic nanocatalysts of Ca/Fe₃O₄@SiO₂, obtained high catalytic activity for biodiesel production, which reached 97%. The authors reported that the catalyst was recovered simply by using an external magnetic field and reused several times without appreciable loss of its catalytic activity.

It can be inferred from the achieved results that the proposed structure could be applied successfully in biodiesel production (except for some adjustments that contribute to a reduction in the operating temperature), because the benefits have surpassed the traditional methods, since that in the last few years the proper choice of raw material and catalyst has been seen as a critical point to determinate the cost of biodiesel production in refineries and try to minimize it.

This contributes to the technical-economic approach of the researchers Abubakar et al. (2015) regarding to the production processes in biodiesel refineries. The authors reported that biodiesel production, like other engineering projects, involves important decisions that have to be made under uncertainties arising from a variety of issues, such as the inherent variation in operating conditions and market forces that characterize the inflation, depreciation factors and variations in equipment and production costs.

That is, the authors listed as a critical point in the biodiesel production chain, operational, financial and management issues in the face of present challenges. In this way, the results obtained with the samples PM, MM and GM express great prospects for, in addition to other studies, to reach some constraint aims in the biofuel production, and eventually to offer additional measures of performance that are necessary for potential investors, governments, engineers and other interested parts in order to ensure the safety of the refineries installations and their cost-effectiveness.

Therefore, it was possible to obtain a final product with differential to be produced on a large scale, with wide possibilities of application, thus providing more arguments in the researches sum about magnetic ceramics, which in turn revolves around to present a new material or for new applications, with relatively low cost, and especially with the possibility of having a technological product directed to the society benefit, regarding to the life and the environment preservation, through the contribution for biodiesel production, complying the true role of science, which must be to promote the common good of society in general.