**Zirconia-Supported Silver Nanoparticles for the Catalytic Combustion of Pollutants Originating from Mobile Sources**

**Maia Montaña 1, María S. Leguizamón Aparicio 1, Marco A. Ocsachoque 1, Marisa B. Navas 1, Ivoneide de C. L. Barros <sup>2</sup> , Enrique Rodriguez-Castellón <sup>3</sup> , Mónica L. Casella <sup>1</sup> and Ileana D. Lick 1,\***


Received: 29 January 2019; Accepted: 16 March 2019; Published: 25 March 2019

**Abstract:** This work presents the physicochemical characterization and activity of zirconia-supported silver catalysts for the oxidation of pollutants present in diesel engine exhaust (propane, propene, naphthalene and soot). A series of silver-supported catalysts AgxZ (x = 1, 5 and 10 wt.%, Z = zirconia) were prepared, which were studied by various characterization techniques. The results show that silver is mainly found under the form of small metal nanoparticles (<10 nm) dispersed over the support. The metallic phase coexists with the AgOx oxidic phases. Silver is introduced onto the zirconia, generating Ag–ZrO2 catalysts with high activity for the oxidation of propene and naphthalene. These catalysts also show some activity for soot combustion. Silver species can contribute with zirconia in the catalytic redox cycle, through a synergistic effect, providing sites that facilitate the migration and availability of oxygen, which is favored by the presence of structural defects. This is a novel application of the AgOx–Ag/ZrO2 system in the combustion reaction of propene and naphthalene. The results are highly promising, given that the T50 values found for both model molecules are quite low.

**Keywords:** silver nanoparticles; zirconia; hydrocarbons; diesel soot; catalytic combustion

#### **1. Introduction**

Mobile emission sources generate a large amount of pollutants, which are emitted into the atmosphere and cause serious environmental problems, both historically and in human health. Among these pollutants are nitrogen oxides (NOx), carbon oxides (COx), volatile organic compounds (VOCs), sulfur oxides (SOx) and particulate matter, usually called soot or diesel soot [1].

Among the VOCs, different light hydrocarbons (C2–C10), known as NMHCs (non-methane hydrocarbons), are found in very low concentrations. The NMHCs that are the most difficult to eliminate are C3-C4, and polyaromatic compounds (PAHs), which present a level of high toxicity [2,3]. On the other hand, diesel soot consists of solid particles of small size, whose main component is carbon [4]. These particles are very small and can penetrate the lungs, which makes soot very harmful to health.

These pollutants can be removed by catalytic oxidation to CO2 and water, avoiding the formation of toxic compounds. This process can be carried out in catalytic converters located in the exhaust pipe of automobiles. Moreover, to remove particulate matter, the converter must retain the particles and then regenerate itself by oxidizing this retained material. In addition, to avoid an extra energy demand, this process must function in a passive way. This implies that the oxidation reaction must occur within the operating temperature range of the exhaust pipe [5–8]. Passive filter regeneration technology is used in the commercial Continuously Regenerating Trap system (CRT, Johnson Matthey, London, UK).

A wide variety of active phases have been reported for the catalytic oxidation of contaminants present in emission sources. In this sense, different types of supports have been studied, such as: simple and compound oxides, and zeolites and clays, among others [9–12]. It is possible to differentiate between catalysts based on oxidic phases, and those based on supported metallic phases. The latter group usually contains Pt, Pd, Rh, and Au, among other noble metals. These systems are called supported platinum group metal (PGM) catalysts and present very good activity, so they can be used in several applications [13–20]. However, due to the high cost and low abundance of these metals, it is necessary to reduce the metal load used, or replace them with other less expensive phases.

In the case of VOCs, and particularly in the case of the oxidation of the remaining hydrocarbons (HC), the following oxidic systems and compounds have been reported as being active: Co3O4, MnO2, CeO2, and CuO [21–31]. The combustion reaction of particulate matter has been extensively studied in recent decades, identifying catalysts with very diverse active phases [32–41]. A meticulous study of them can be found in a review by A. Bueno Lopez et al. [32] and in a more recent review by D. Fino et al. [33]. Catalysts containing MnO2-x, CeO2, or alkali metals have been reported as being very active, like other noble metal free oxides systems. In addition, the metallic promotion of these oxides has been studied, to achieve better catalytic performance. In this context, the development of Ag-containing oxidic catalysts has received attention in recent years [5,42–58]. In the case of the combustion of particulate matter, these catalysts have been reported as very active [42–44,47]. There are also some studies in which different catalysts containing silver are applied in the total oxidation of remaining HC. The silver nanoparticles are of low cost, and they possess high activity and stability. Silver has been widely used as biocide, or in the development of sensors. In this sense, Ag is considered safe for the design of materials, due to its low potential hazardous effects.

In the case of polycyclic aromatic hydrocarbons (PAHs), the number of reports is much lower. In particular, the Ag/CeO2 catalyst presents a great performance in soot oxidation reactions [55], propene and naphthalene [58]. According to the literature, the excellent results should be assigned to the ability of the Ag-containing catalysts to activate oxygen through dissociative adsorption. This capacity can be promoted synergistically with the mobility of oxygen, as provided by the support. This type of mechanism, based on the adsorption of the reactants, has been accepted as one of the feasible mechanisms in the oxidation reactions of soot and hydrocarbons [13,19,28,42,50]. Yamazaki et al. studied soot oxidation over a CeO2–Ag catalyst with a 'rice-ball' morphology, consisting of a center composed of Ag particles, surrounded by fine CeO2 particles. A mechanism for soot oxidation has been proposed, where active oxygen atomic species are formed on the Ag surface, from the dissociative adsorption of O2. These species migrate to the CeO2 interface, and form On <sup>x</sup><sup>−</sup> species; then they further migrate onto soot particles, where oxidation occurs at very low temperatures [50]. A. Serve et al. studied Ag catalysts supported on yttria-stabilized zirconia (YSZ) for soot oxidation. They also proposed that Ag favors the activation of the dissociatively adsorbed oxygen, and the lattice integration of gaseous oxygen. Soot oxidation occurs through bulk O2-YSZ [55].

On the other hand, it is proposed that oxidation reactions can follow a mechanism of the Mars-van Krevelen type. According to this mechanism, the catalytic surface can form a chemical bond with an adsorbate/reactant, for example, HC. These species can be oxidized by a surface lattice atom on the catalytic surface. When a reaction product desorbs, a vacancy is generated on the surface, which must be regenerated [59–63]. It is clear that surface species must have some redox capacity, and also the capacity to form oxygen vacancies [56]. In this context, zirconia is presented as a suitable support, due to its redox capacity [26]. In addition, when zirconia are added to a host cation in its crystalline network, the bulk and sub-surface defects are generated. This facilitates the migration of oxygen species from the bulk to surface. The presence of oxygen vacancies has been correlated with the catalytic activity in oxidation reactions; for example, the oxidation of CO and soot [8,56,58].

Various works concerning oxidation mechanisms indicate that not only the presence of oxygen vacancies is necessary, but also the presence of electrophilic species called "active oxygen Ox −" (x = 1 or 2). Some authors even indicate that even the presence of an excessive number of surface defects can generate the deactivation of these Ox − species, due to the generation of O2 − species, which are nucleophilic. In a recent work, Wang et al. have studied the roles of oxygen vacancies, hydroxyls and Ox − in different oxidation reactions, using CeO2 and Ag/CeO2 as catalysts [56,58]. Moreover, they have shown that the formation of Ox − species can be promoted with the presence of Ag to obtain higher NO and soot oxidation activities. Another desirable characteristic is that the support possesses certain levels of acidity to favor the adsorption of the hydrocarbon. In this sense, there are some reports on the activity of pure zirconia for oxidation reactions [26].

Zirconia is a support that has a certain redox capacity [64]; it presents polymorphs with controllable acid–base properties and good thermal stability. Besides, in its amorphous hydrous state, it presents a high surface area that can favor the dispersion of the supported precursors, and thus avoid agglomeration during the thermal treatments [64–66].

In addition to the different reaction mechanisms that can be proposed, there are several parameters influencing the behavior of the active species in these systems. Thus, a fundamental factor to consider is the size of the metal particles (an optimum value is <10 nm) [67]. Also, the catalytic efficiency of the Ag catalysts can be related to the adsorption of some hydrocarbons on this metal [68].

In this work, it is proposed that the preparation of zirconia-supported Ag catalysts, obtained by the impregnation of ZrO2·nH2O (hydrogel hydrous zirconium oxide) with AgNO3 solutions. The focus will be on the formation of supported Ag nanoparticles, and in the study of this catalytic system in oxidation reactions of pollutants generated in emission sources. The activity of these catalysts is analyzed in the oxidation reactions of model molecules of different hydrocarbons at very low concentrations: propane, propene, and naphthalene. It is worth mentioning that the Ag/ZrO2 system has not yet been used for propene and naphthalene elimination. Also, their activities are analyzed in the combustion of diesel soot/particulate matter reaction. The prepared catalysts are characterized using several physicochemical techniques: N2 adsorption–desorption, scanning electron microscopy, and energy dispersive X-ray spectroscopy (SEM-EDS), X-ray diffraction (XRD), diffuse reflectance spectroscopy (DRS-UV-vis), transmission electron microscopy (TEM), and temperature-programmed reduction (TPR). X-ray photoelectron spectroscopy (XPS) was employed to obtain information about the species at the surface level.

#### **2. Results and Discussion**

Table 1 presents the characterization results for zirconia and silver–zirconia-supported catalysts.


**Table 1.** Specific surface area, TPR results and crystalline phases of the studied samples.

\*<sup>1</sup> obtained by XRD. \*2 theoretical H2 mmol consumed per 100 mg of catalyst (Equation (1)) \*3 experimental mmol H2 consumed per 100 mg of catalyst. \*<sup>4</sup> obtained by TEM.

Textural properties were obtained by N2 adsorption–desorption at 77 K. Figure 1 presents the adsorption–desorption isotherms of the Ag-supported catalysts. In all the cases, the obtained isotherms were type IV according to the IUPAC classification, presenting a hysteresis cycle that is associated with the process of filling of the mesopores.

The BET surface area values corresponding to the original hydrogel and the prepared catalysts are presented in Table 1. The starting hydrogel exhibits a high specific surface area value. Then, the addition of Ag and the subsequent thermal process decreases the surface area, due to the loss of the water molecules of the hydrogel, and also the increase of the metal–host–metal interactions. An increase in the Ag content decreases the specific surface area without decreasing the pore volume. The crystallization process of pure zirconia generates a system with a slightly larger area than that of the Ag-catalysts. This indicates that the thermal treatment is the predominant factor in textural changes.

**Figure 1.** N2 adsorption–desorption isotherms of the catalysts. (**a**) Ag1Z, (**b**) Ag5Z and (**c**) Ag10Z.

Figure 2 shows the XRD patterns of the support and the AgxZr catalysts. The XRD pattern of the undoped ZrO2 support, calcined for 2 h at 600 ◦C presents diffraction lines that are typical of the monoclinic phase of ZrO2, with peaks being located at 28.2, 31.5, 34.2, and 50.2◦ (PDF N◦ 03-065-1025). The diffraction lines of the meta-stable tetragonal phase, located at 30.0◦, 35.0◦, and 50.0◦ (PDF N◦ 01-089-7710) were identified for all AgxZ catalysts. The addition of Ag strongly influences the crystallization of amorphous hydrated zirconium oxide, and promotes the stabilization of the meta-stable tetragonal crystalline phase. This specific phase has been proposed as the active phase on certain reactions [69]. The presence of the meta-stable phase, even in the catalyst with the lowest Ag content, indicates a strong Ag—support interaction, generated in the crystallization and sintering stages.

As it has been widely reported, some ions, called "host ions" or dopants, can be incorporated into the crystal lattice of zirconia by replacing the central ion, Zr (IV). In this sense, the effective ionic radius of Ag (I) is similar to those of the most common cations used in doping zirconia, such as Y (III) and Ca (II). Furthermore, if the dopants have a lower oxidation state than the central ion in the crystalline structure, defects are generated at the bulk and surface levels, with the consequent formation of oxygen vacancies [70].

The zirconia crystallite size, obtained by Debye–Scherrer´s equation, is shown in Table 1. The results indicate that AgxZ catalysts contains nanometric zirconia crystals, with a size slightly larger than those present in pure zirconia.

On the other hand, the patterns do not exhibit diffraction lines of Ag0 crystallites, located at 38.15◦, 44.34◦, and 64.51◦, (PDF N◦ 03-065-2871), or lines of Ag2O, located at 32.88◦ and 38.15◦ (PDF N◦03-065-6811). This indicates that, if these species exist, they are present at a size that cannot be detected by XRD. It should be noted that there are no marked differences between the XRD profiles of AgxZ catalysts. Only a very slight increase in the intensity of the line at 30.0◦, belonging to the support, is observed. This indicates the presence of good stabilization of the metastable tetragonal phase, and a good distribution of the supported species on the surface, since no segregated crystalline phases are observed.

**Figure 2.** XRD patterns of support and catalysts. (**a**) ZrO2, (**b**) Ag1Z, (**c**) Ag5Z, and (**d**) Ag10Z.

Figure 3 presents the SEM images and elemental maps (MAP) of the prepared catalysts. These analyses were carried out in order to determine the distribution of silver on the catalyst surface. The overlapping of zirconium and Ag distribution in the MAP images showed that the structures containing Ag are formed on the particles of the catalysts. Moreover, the Ag species are widely distributed over the Ag5Z and Ag10Z catalysts (Figure 3). Some EDS analyses were carried out for the very small areas, where the highest intensity was observed in the mapping of silver. It was found that the experimental Ag/Zr atomic ratio was greater than the nominal one.

**Figure 3.** SEM images and elemental mapping of Ag and Zr in catalysts: (**A**) Ag5Z and (**B**) Ag10Z.

In order to analyze the presence of reducible phases in the catalysts, TPR analyses were carried out, and the results are depicted in Figure 4. The TPR diagram of the pure ZrO2 (not shown) does not present any signal, which indicates the absence of reducible phases at the studied temperature range, although it is known that it can be partially reduced at high temperature (about 800 ◦C). The TPR diagrams of the catalysts containing 5 and 10 wt.% of Ag presented a signal at very low temperature (approximately 90 ◦C). This can be assigned to the reduction of oxidic or cationic Ag species, in low concentrations, which are well-dispersed on the support, or presenting low interactions with it. This signal also can be assigned to superficial oxygen atoms coming from its migration from zirconia [58]. The integrated area of the signal increases with the increase of Ag content. However, it is noteworthy that all the AgxZ catalysts had an extremely low level of H2 consumption (Table 1), which is lower than the expected if the supported phase corresponds to the stoichiometric reduction of the Ag2O oxide (Equation (1)).

$$\text{Ag}\_2\text{O} + \text{H}\_{2(g)} \leftrightarrow \text{H}\_2\text{O}\_{(g)} + 2\text{Ag}^0\tag{1}$$

These results would indicate that silver is mainly in a reduced state. These species could have been formed during the calcination process, probably due to the reducing environment of the zirconia hydrogel-containing surface OH groups, and the low stability of the Ag oxides at high temperature. Corro et al. propose that the Ag<sup>0</sup> formation is associated with the low enthalpy of formation of Ag2O at high temperature [71]. In previous studies, it has been reported that the reducibility of transition metal species increases on ZrO2 [26,72]. It has also been proposed that the formation of oxygen vacancies affects the redox properties of the supported species. A mode of behavior that similar to that observed in this work was reported by Aneggi et al., who studied the Ag/ZrO2 catalytic system [53]. These authors show that during the calcination process, AgNO3 decomposes almost completely into metallic Ag, although they observed by HRTEM the presence of Ag2O on the surface of the catalysts.

**Figure 4.** H2-TPR profiles of (**a**) Ag1Z, (**b**) Ag5Z, and (**c**) Ag10Z.

In order to obtain more information about the nature of the supported Ag phases, DRS UV-Vis studies were performed. The spectra obtained for AgxZ catalysts are shown in Figure 5. An energy absorption band located between 330 nm and 390 nm can be clearly observed. In this region of the spectrum, the absorption is assigned to Ag clusters Agn <sup>δ</sup><sup>+</sup> (n = 2–7) or particles of metallic silver of small size (3–5 nm). As the size of the particles grows, this band can suffer a bathochromic shift, towards longer wavelengths. This is probably the reason for the slight shift of this band in the Ag10Z catalyst spectrum. The bands located at 220 and 265nm are associated with the presence of the Ag+1 species, and the charge transfer band of the Agn <sup>δ</sup><sup>+</sup> clusters respectively [67]. Although there is absorbance in the region of the plasmon resonance, 400–450 nm, these bands are not well-resolved or defined, due to the presence of energy absorption of the support, ZrO2.

**Figure 5.** DRS UV-Vis signals for the prepared catalysts. (**a**) Ag1Z, (**b**) Ag5Z and (**c**) Ag10Z.

The characterization techniques performed, which provide information at a mass level, suggesting that a part of Ag is present in a metallic state, Ag0. In order to corroborate the presence of this type of species, transmission electron microscopy (TEM) analysis were carried out (Figure 6). It is noticeable that the micrographs corresponding to Ag5Z and Ag10Z catalysts show zirconia structures with highly dispersed silver nanoparticles. In the micrographs of the Ag1Z catalyst, although some Ag particles

were observed, the resolution of the technique does not allow for the visualization of a significant number of them. Ag particle size distribution was analyzed from some selected regions of the TEM micrographs; the results obtained for the mean particle size are listed in Table 1.

The histograms presented in Figure 6 show that for both Ag5Z and Ag10Z catalysts, the Ag particles had sizes of between 2 and 10 nm. The histogram of the Ag5Z catalyst exhibits a relatively narrow particle size distribution, mainly constituted of particles having diameters of between 4 and 5 nm. The increase in the silver loading of the catalysts leads to a greater contribution of the larger particles, and therefore, the average particle size increases. This increase in the silver particle size leads to a loss of metal dispersion (D%), which decreases from 23% to 17% for Ag5Z and Ag10Z, respectively.

**Figure 6.** TEM micrographs and histograms of particle size distribution: (**A**) Ag5Z catalyst; (**B**) Ag10Z catalyst.

The TEM results agree with those obtained by H2-TPR and UV-Vis spectroscopy. Both techniques suggested that part of the Ag was in a metallic state under the form of nano-sized particles, which cannot be observed by XRD. However, there is evidence for the existence of oxidic species in low concentration, due to the presence of signals at low temperature in the TPR analysis, and some absorption bands in UV-Vis, associated with this species.

The oxidation states of the supported species of Ag surface were studied by XPS, although several reports indicate that it is difficult to distinguish between chemical states of Ag, since the binding energies values of Ag3*d*5/2 (BE) for Ag(0) and their oxides are very similar [73]. Thus, the AgMNN Auger signals were also analyzed, in order to obtain more precise information, since the chemical displacements of the Auger peaks were generally greater than the displacements of the photoelectronic peaks, therefore being more appropriate for considering the use of the modified Auger parameter (AP\*) to identify the Ag oxidation states. Therefore, in this work, the XPS spectra of the catalysts in the Ag3*d* region and in the Auger region, AgMNN (Figure 7A,B), were studied. Table 2 presents the binding energies (BE) of the Ag3*d*5/2 peak, the kinetic energy of the AgMNN Auger transition (KE), and the calculated AP\* (AP\* = KE + BE) for the different components of the AgMNN signals.

BE values for the Ag3*d*5/2 peaks are very close to each other, for the three studied catalysts. The values are between 368.2 eV and 367.7 eV, decreasing slightly when the metallic content decreases. In the AgMNN region, the signals are wide, complex, and have several contributions (Table 2). Those whose AP\* are at values of ca. 725 eV and 720 eV, can be associated with the presence of Ag in the metallic state. On the other hand, signals that are located at lower values of KE and AP\* (around 717.5 eV), are associated with the presence of oxidic phases of Ag. The metallic particles are probably oxidized on their surfaces [73].

**Figure 7.** (**A**) Ag 3d XPS spectra, (**B**) AgMNN Auger transition spectra, and (**C**) O1s XPS spectra of the different studied catalysts.

**Table 2.** XPS BE (eV) values for the Ag3d, O1s, and Zr3d, Auger parameters, Ag/Zr atomic ratios, and oxygen species ratio for the studied catalysts.


\*<sup>a</sup> between parenthesis the relative percentages of the different components of the signal are shown.

Table 2 also presents the Zr 3*d*5/2 and O 1*s* binding energy values (in eV) from the core-level spectra signals. The spectra of the Zr 3*d* core level spectrum shows two binding energies about 182 and 184 eV, which are assigned to the doublet Zr 3*d*5/2 and Zr 3*d*3/2, respectively. These peaks are typical of Zr4+ in ZrO2. The O 1*s* signal was decomposed in three components, with binding energies of about 530, 531, and 532–533 eV (Figure 7C), being related to surface oxygen species with different chemical environments. The species with binding energy between 529–530 eV, were attributed to the Oα, which is characteristic of lattice O in ZrO2, [15,24] and those in the 531–533 eV region, were related to Oβ, which are associated with surface oxygen vacancies, oxygen-adsorbed species, surface hydroxyl, adsorbed water molecules, or carbonate species [74]. In our case, however, the presence of carbonates was not evident in the spectra of the C1*s* region (not shown). The Ag5Z and Ag10Z catalysts presented a higher proportion of O<sup>β</sup> species (Table 2), which are proposed as possible active sites.

The Ag/Zr surface atomic ratios have been calculated from XPS analysis, and then compared with the theoretical composition of the bulk (Table 2). As expected, the surface presents an enrichment of Ag, due to the predominant location of Ag on the external surface of the support, both for the Ag5Z and Ag10Z catalysts.

The AgxZ catalysts, and also the pure zirconia were used in the catalytic oxidation of three model molecules of hydrocarbons (HC): propane, as a saturated linear hydrocarbon molecule; propene, as an unsaturated linear hydrocarbon molecule; and naphthalene, as a polyaromatic hydrocarbon molecule. In addition, the prepared catalysts were tested in the catalytic oxidation of particulate matter (diesel soot). Table 3 summarizes the catalytic results obtained in all of the studied reactions. The combustion reactions of hydrocarbons were carried out in fixed-bed micro-reactors. The obtained T50 values in each experiment were presented, with T50 being the temperature where 50% HC conversion is reached. The results corresponding to Tmax for the soot combustion reactions, carried out in a thermogravimetric reactor, are also presented in Table 3. Tmax is defined as being the temperature at which the derivative of the TGA curve shows a minimum. This derivative curve represents the loss of mass by combustion, which is associated with the maximum burning rate.

As can be seen, all of the studied catalysts presented a level of activity for the total oxidation of all of the selected model molecules. In addition, in all the reactions the support contributed to the activity, with its redox characteristics. The influence of Ag and its metallic load depended on the nature of the molecule to be oxidized.


**Table 3.** Catalytic activity of ZrO2 and the three prepared AgxZ catalysts, for the different oxidation reactions studied.

<sup>a</sup> 1000 ppm HC, 8% O2, mcat = 100 mg, Q = 50 mL·min−1; <sup>b</sup> 150 ppm HC, 8% O2, mcat = 100 mg, Q = 30 mL·min−1; <sup>c</sup> mcat = 30, msoot = 3, loose contact, <sup>Δ</sup>T = 10 ◦C min−1, 10% O2, Q = 100 mL·min−1.

In the case of propane combustion, the catalysts exhibited low activity, and the contribution of Ag was not significant (Table 3). Using pure zirconia, T50 was 490 ◦C and the addition of Ag at different concentrations slightly increased the activity. Within the AgxZ series, the most active series the propane combustion resulted was the Ag10Z catalyst. When this catalyst is employed, it is possible to reduce T50 by 150 ◦C, with respect to the uncatalyzed reaction (600 ◦C). A series of materials with greater activity than those studied here for propane combustion, are presented in the bibliography [25,27,30,31,75]. Different reports present catalytic systems where the T50 values are lower than 300 ◦C, even when using high-space velocities. For example, in a recent paper by X. Li et al. [30], the activities of the CoCeOx catalysts was reported and compared with other similar systems, which presented very low values of T50. The poor performance of the AgxZ catalysts can be attributed to the low capacity of the Ag species to promote the activation of propane.

In contrast, the AgxZ catalysts had very good level of activity for the oxidation of propene. The results for propene oxidation reaction, expressed as propene conversion to CO2% vs temperature, are plotted for all the studied catalysts in Figure 8A. No other by-products or CO were observed as reaction products; CO2 selectivity was higher than 99%. In the absence of a catalyst, T50 was 630 ◦C,

and for pure ZrO2, 50% propene conversion was reached at 520 ◦C. The addition of Ag to the system substantially increased the catalytic activity: Ag5Z and Ag10Z catalysts presented great performance, reaching 50% of CO2 conversion at 280 ◦C and at 210 ◦C, respectively. Both catalysts exhibited high levels of conversion (> 90%) at temperatures less than or equal to 330 ◦C. It is evident that the catalysts are more effective for the elimination of propene than propane. The results here presented are original, since there are no reports on the activity of the Ag/zirconia system for the elimination of propone, a molecule that is classified as a VOC.

**Figure 8.** (**A**) Catalytic activity in propene oxidation: (----) without catalysts, (o) ZrO2, ((■) Ag1Z, (●) Ag5Z, (▲) Ag10Z. (**B**) Catalytic activity in naphthalene oxidation: (o) ZrO2, (■) Ag1Z, (●) Ag5Z, (▲) Ag10Z.

On the other hand, the prepared catalysts presented noticeable activity the naphthalene oxidation. Figure 8B shows the results obtained, expressed as naphthalene conversion to CO2% vs temperature. In the absence of a catalyst, naphthalene is oxidized at high temperatures, reaching T50 at 430 ◦C. Pure zirconia presented good activity, and attained 50% conversion at 280 ◦C. The addition of Ag significantly promoted its activity. Thus, the Ag5Z catalyst was the most active of the series, achieving a high level of conversion at low temperatures (250 ◦C). Using this catalyst, T50 decreased by more than 250 ◦C compared to the temperature that was reached in the absence of the catalyst. When increasing the Ag content from 5 to 10%, a slight decrease in activity was observed. Probably, an increase in the size of Ag particles, and the consequent decrease in metal dispersion, generated a lower availability of active sites for the activation of naphthalene.

These results are an unprecedented contribution, taking into account that the silver–zirconia system has not been studied before, for the total oxidation of naphthalene. Besides, these results are promising, given that the T50 reached with the Ag5Z catalyst is 172 ◦C. This value is very low, and also comparable to previously obtained results using transition metal oxide catalysts supported on zirconia [26] and with results recently reported by M. Liu et al. [58] with the Ag/CeO2 catalysts. Even more, as the Ag5Z catalyst was the most active, it was reused three times in the reaction cycles, keeping its activity.

Next, the activity results for diesel soot oxidation are presented. Table 3 shows the maximum oxidation rate temperature (Tmax) obtained for diesel soot combustion in the presence of air, using loose contact conditions, and carried out in a thermogravimetric reactor. Table 3 includes the differences between Tmax in a reaction with and without a catalyst (ΔTmax). The obtained values for ΔTmax indicate that both the support and the catalysts presented levels of activity. While pure zirconia exhibits poor activity, the activity of the AgxZ catalysts depends on the Ag content (Table 3, Supplementary Material). It was found that with the increase in Ag content, the Tmax decreased. In this reaction, Ag10Z catalyst was the most active of the series, reaching a ΔTmax of 225 ◦C. These results indicate that the presence of the Ag species generates a promoting effect.

As stated previously, although the catalysts presented activity, they depended on the Ag content and the nature of the molecule to be oxidized. The systems did not have high efficiency for propane oxidation, as it was being evident that it is not possible to activate this saturated molecule. On the other hand, they showed activity for the diesel soot oxidation, reaching combustion temperatures that were comparable with other reported catalysts containing silver, and other oxidic systems, when measurements were made in "loose contact" mode [53,76]. However, there are numerous systems in the literature that show evidence of higher activity; therefore, the presented formulations could be optimized to achieve better results. Even so the obtained value of Tmax with the Ag10Z catalyst can be considered to be acceptable, since it is in the range of temperatures at which the exhaust pipe operates.

In the oxidation of molecules containing π–π bonds, such as propene and naphthalene, the catalysts exhibited very good performance. It was achieved a significant reduction at T50, especially when using the Ag5Z and Ag10Z catalysts. As previously mentioned, these results are original, since the silver–zirconia system has not been applied to for the elimination of these unsaturated molecules. Moreover, a crystal-size effect of silver nanoparticles has been found in the catalytic performance of the system in the naphthalene combustion, which was not observed for the combustion of propene. As expected, zirconia provided some activity in all the reactions studied, and the addition of high Ag content (5 and 10 wt.%) generated an Ag–zirconia system with a greater availability of superficial oxygen species. This fact may be responsible for the acceleration of the reaction and the regeneration of the active sites, through successive redox cycles. M. Skaf et al. have demonstrated the ability for Ag species (Ag0 and Ag+) supported on ceria to adsorb oxygen molecules through redox cycles [77]. The Ag–zirconia synergic effect has been observed from the characterization techniques, and the formation of a metastable tetragonal phase is observed by XRD, leading to the generation of oxygen vacancies.

The coexistence of both types of species, AgOx and Ag0, may favor the redox capacity of the surface, with the oxidic Ag species being responsible for oxidizing the adsorbed substrate onto the catalytic surface, and hence adopting a lower oxidation state. This situation allows for an explanation of the reaction mechanism, employing Mars-van Krevelen formalism. Oxidic sites must be regenerated, and this step is favored through the capacity of the system, which contains oxygen vacancies and a high availability of adsorbed Ox − species, as evidenced by XPS. According to this technique, the catalysts with the highest metallic loading, Ag5Z and Ag10Z, which are the most active, contain a greater proportion of oxidic species of high availability (Oβ).

According to the results of TEM, DRS UV-vis, and XPS techniques, Ag is preferably found on the surface. Ag<sup>0</sup> nanoparticles co-exist with oxidic Ag species (Agδ<sup>+</sup> and Ag+). According to TPR results, the presence of these oxidic phases is conditioned by the presence of reducible phases in a very low proportion. Supported Ag nanoparticles found are small (<10 nm), and well-dispersed over zirconia.

As previously mentioned, in the literature, various mechanisms that are related to this kind of systems have been proposed for oxidation reactions. Therefore, it is not possible to rule out the dissociative adsorption of molecular oxygen on Ag nanoparticles, or that a combination of both mechanisms occurs. When the activated oxygen (On <sup>x</sup>−) is consumed during the oxidation of the substrate, it is restored by a new round of adsorption of the O2 present in the gas phase, or by the migration of oxygen species through structural defects. These defects are present in the Ag nanoparticles and in the support, which also can chemisorb oxygen and provide mobility. Oxygen vacancies exist not only at the surface level, but also at the bulk level.

The behavior observed in naphthalene oxidation, where the most active catalyst is Ag5Z, and a later increase in the Ag content, generates a less active catalyst. Probably, this can be associated with an increase in the size of the Ag particles, with a resultant decrease of dispersion, and a lower capacity towards the adsorption of naphthalene. This point must be studied more deeply in the future, in order to correlate the catalytic results with experimental evidence. In addition, a possible geometric effect by the Ag ensembles, which cause a decrease in the capacity of naphthalene adsorption, has not yet been ruled out [68]. Probably, given the geometry of the naphthalene molecule, with two conjugated aromatic rings, its adsorption is sensitive to the structure of the catalyst surface. There is also evidence that propene combustion does not follow the same behavior, in comparison to the size of the silver crystallite, marking a difference between both molecules, which must be analyzed with greater depth in the future. All of the studied catalysts were active in the soot combustion reaction, with the Ag10Z catalyst being the most active one. Again, the ability of these catalysts to generate a synergism between the Ag species and zirconia for the activation and regeneration of oxygen is evidenced. This leads to an oxidation on the catalytic interphase, which in this case involves a gas phase and two solid phases.

#### **3. Experimental Section**

#### *3.1. Material Preparation*

Hydrous zirconium oxide, ZrO2·nH2O, was prepared by precipitation from a ZrOCl2·6H2O (Fluka, Buchs, Switzerland) solution with ammonium hydroxide (pH 10). The process was carried out at room temperature with constant stirring for 6 h. The pH was maintained close to 10. After filtration, the solid was washed (Cl− negative test in the solid) and dried at 100 ◦C for 6 h. Portions of ZrO2·nH2O were impregnated with an aqueous solution of AgNO3, to obtain solids with a silver concentration of 1, 5, and 10 wt.% (grams of Ag per 100 g of catalyst). After drying, the samples were thermally treated at 600 ◦C for 2 h. The materials so obtained were named Ag1Z, Ag5Z and Ag10Z, where Z = zirconia.

#### *3.2. Catalyst Characterization*

The textural characterization of the support and catalysts was determined by using the BET method, using a Micromeritics Accusorb 2100 E apparatus (Micromeritics, Norcross, GA, USA). Samples were degassed at 100 ◦C prior to analysis.

Scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM-EDS) analyses were performed using a SEM Philips 505 equipment (Philips Co, Amsterdam, The Netherlands). The energy dispersive X-ray analysis of the samples was performed using an EDAX DX PRIME 10 analyzer (EDAX, New Jersey, NJ, USA) at a working potential of 15 kV.

Temperature programmed reduction (TPR) analysis was performed with a home-made equipment. In a typical run, the sample (30.0 mg) was placed in an electrically heated fixed-bed quartz micro-reactor, and heated from 50 to 800 ◦C, with a heating rate of 10 ◦C min−1. In the experiments, a feed of 10% hydrogen in nitrogen (flow rate of 20 mL min<sup>−</sup>1) was used. Hydrogen consumption was detected by a thermal conductivity detector.

Diffuse reflectance UV–visible spectra (DRS-UV–vis) of AgxZ catalysts were obtained with a Perkin Elmer Lambda 35 UV-VIS spectrometer (Perkin Elmer Inc., Waltham, MA, USA).

The distribution of metal particle sizes was determined by transmission electron microscopy (TEM) using a JEOL 100 CX instrument (Jeol Ltd., München, Germany), having a resolution of 6 A and an accelerating voltage of ca. 100 kV. The samples were ground and ultrasonically dispersed in distilled water. To estimate the mean particle size, the particles were considered to be spherical, and the second moment of the distribution was employed. The expression used for the calculation was:

$$d\_{AV} = \frac{\sum n\_i d\_i^3}{\sum n\_i d\_i^2} \tag{2}$$

where *ni* is the number of particles of *di* size. Over 200 particles were measured. The metal dispersion (D) was estimated from the Ag particle size distribution obtained from the TEM measurements by the following equation [24], assuming spherical particles:

$$D = \frac{6M\_{A\underline{\mathcal{g}}}}{\sigma \rho\_{A\underline{\mathcal{g}}}} \frac{\sum n\_i d\_i^2}{\sum n\_i d\_i^3} \tag{3}$$

where *MAg* and *<sup>ρ</sup>Ag* are the molar mass (107.87 g·mol<sup>−</sup>1) and density of Ag (10.5 g·cm<sup>−</sup>3), respectively, and <sup>σ</sup> is the area that is occupied by 1 mol of Ag at the surface (8.75 × <sup>10</sup>−<sup>16</sup> cm2·mol<sup>−</sup>1).

XPS measurements were carried out, using a Physical Electronics spectrometer (PHI Versa Probe II Scanning XPS Microprobe, Physical Electronics, Chanhassen, MN, USA) with monochromatic X-ray Al Kα radiation (100 μm, 100 W, 20 kV, 1486.6 eV), and a dual beam charge neutralizer. The energy scale of the spectrometer was calibrated, using Cu 2*p*3/2, Ag 3*d*5/2 and Au 4*f* 7/2 photoelectron lines at 932.7, 368.2, and 84.0 eV, respectively. Under a constant pass energy mode under 23.5 eV conditions, the Au 4f7/2 line was recorded with 0.73 eV FWHM at a binding energy (BE) of 84.0 eV. The collected XPS spectra were analyzed using PHI SmartSoft software and processed using MultiPak 9.3 package. The binding energy values were referenced to the adventitious carbon C 1*s* signal (284.8 eV). The recorded spectra were always fitted using Gauss–Lorentz curves. The atomic concentration percentages of the characteristic elements of the surfaces were determined, taking into account the corresponding area sensitivity factor for the differently measured spectral regions.

#### 3.2.1. Catalytic Activity for Propane and Propene Oxidation

The catalytic activity for propane and propene oxidation was measured by using a fixed-bed quartz reactor that was electrically heated. Figure 9 shows a schematic diagram of the reaction system.

**Figure 9.** Flow diagram of the reaction system. Components: unit (1) gas cylinders (He, O2, C3H8,C3H6); unit (2) On–off valves; unit (3) mass flow controllers; unit (4) Tee valves; unit (5) four-way valve; unit (6) quartz reactor; unit (7) furnace; unit (8) thermocouple type k; unit (9) thermostat; unit (10) gas chromatograph; unit (11) injection valve; unit (12) vent.

The feed employed for the propane oxidation consisted of a mixture of 1000 ppm of C3H8, 8% of O2 and He to close the balance. Propene combustion was carried out using a gas mixture, having a composition of 1000 ppm of C3H6, 6% of O2 and He to close the balance. For each experiment, the mass of the catalyst used was 100 mg, and the reaction temperature varied from 150 to 600 ◦C. A total flow rate of 50 mL·min−<sup>1</sup> was used in both processes. The products were analyzed by a Shimadzu GC 2014 (Shimadzu Corporation, Kyoto, Japan) chromatograph with a thermal conductivity detector.

#### 3.2.2. Catalytic Activity for Naphthalene Oxidation

The catalysts were tested for the oxidation of naphthalene, using an electrically heated fixed-bed quartz reactor and 100 mg of catalyst. The reaction feed consisted of 10% O2 and 90% He, and 150 ppm naphthalene. A total flow rate of 30 mL·min−<sup>1</sup> was used. Catalytic activity was measured over a temperature range from 150 to 500 ◦C. Data were obtained at each temperature after a certain time, in order to obtain a stable concentration of naphthalene in the gas phase. The feed flow passes through a thermostated saturator containing naphthalene. The combustion was initiated as soon as a stable vapor pressure was reached. The products were analyzed using a Shimadzu GC 2014 chromatograph with a thermal conductivity detector.

#### 3.2.3. Catalytic Activity for Diesel Soot Oxidation

Diesel soot oxidation tests were performed in a thermogravimetric reactor (TA-50 Shimadzu, Shimadzu Corporation, Kyoto, Japan) over a range of 200 to 650 ◦C, with a heating rate of 10 ◦C·min<sup>−</sup>1. The reaction mixture (8 vol.% O2) was obtained from two feed lines, air and He respectively, individually controlled to close the balance. Printex-U was used as the model diesel soot. Before the reaction, the Printex-U was mixed with the catalyst at a 1/10 ratio, with a spatula (loose contact). The mass loss and the reaction temperature were recorded as a function of time. From the mass loss information as a function of time, the derivative curve (DTGA) was obtained, and from it, the temperature at which the combustion rate is maximum (Tmax).

#### **4. Conclusions**

This paper reported the preparation of a series of Ag-ZrO2 catalysts having three different silver concentrations (1, 5, and 10 wt.%). The prepared catalysts were examined by using various analytical techniques (SEM-EDS, XRD, DRS, TEM, TPR, and XPS). It was observed that part of the silver was in a metallic state, in the form of nano-sized particles, and evidence was also found for the existence of oxidic species at low concentration. The catalysts were tested for the catalytic combustion of propane, propene, naphthalene, and particulate matter. Their activities depended on the Ag content and the nature of the molecule to be oxidized. Particularly, the Ag5Z and Ag10Z catalysts exhibited very good performance in the oxidation of molecules containing π–π bonds (propene and naphthalene). Particularly, these catalysts are innovative and display promising results, given that the resulting silver/zirconia system is very active for the elimination of unsaturated molecules, and can be considered a non-toxic catalyst. All of the studied catalysts were also active in the soot combustion reaction, with the Ag10Z catalyst being the most active one. These results could likely be attributed to an Ag-ZrO2 synergic effect.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4344/9/3/297/s1, Figure S1: TEM micrograph of Ag1Z catalyst, Figure S2: DTGA diagrams for the diesel soot catalytic results.

**Author Contributions:** M.L.C. and I.D.L. conceived and designed the experiments; M.S.L.A., M.B.N. and M.A.O. performed the experiments; I.d.C.L.B. and E.R.-C. contributed with the XPS spectroscopy measurements and analyses; all authors discussed the results; I.D.L. and M.L.C. wrote the manuscript. All authors read, revised, and approved the final manuscript.

**Funding:** This research was funded by Consejo Nacional de Investigaciones Científicas y Técnicas: PIP 0276, Agencia Nacional de Promoción Científica y Tecnológica: PICT 0737, Universidad Nacional de La Plata: X700, and Universidad Nacional de La Plata: X707.

**Acknowledgments:** The authors acknowledge the financial support of CONICET, ANPCyT, and UNLP, and Pablo Fetsis, Mariela Theiller, and María Laura Barbelli for characterization experiments.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Synthesis, Performance and Emission Quality Assessment of Ecodiesel from Castor Oil in Diesel/Biofuel/Alcohol Triple Blends in a Diesel Engine**

#### **Beatriz Hurtado 1, Alejandro Posadillo 2, Diego Luna 1,\* , Felipa M. Bautista 1, Jose M. Hidalgo <sup>3</sup> , Carlos Luna <sup>1</sup> , Juan Calero <sup>1</sup> , Antonio A. Romero <sup>1</sup> and Rafael Estevez <sup>1</sup>**


Received: 13 November 2018; Accepted: 20 December 2018; Published: 3 January 2019

**Abstract:** This research aims to promote the use of second-generation biofuels based mainly on Castor oil, which is not adequate for food use, and Sunflower oil as a standard reference for recycled oils. They have been applied in the production of Ecodiesel, a biofuel that integrates glycerol as monoglyceride, employing sodium methoxide as homogeneous catalyst and ethanol as solvent, but operating in milder conditions than in the synthesis of conventional biodiesel in order to obtain a kinetic control of the selective transesterification. The behavior of biofuels has been evaluated in a conventional diesel engine, operating as an electricity generator. The contamination degree was also evaluated from the opacity values of the generated smokes. The different biofuels here studied have practically no differences in the behavior with respect to the power generated, although a small increase in the fuel consumption was obtained in some cases. However, with the biofuels employed, a significant reduction, up to 40%, in the emission of pollutants is obtained, mainly with the blend diesel/castor oil/alcohol. Besides, it is found that pure Castor oil can be employed directly as biofuel in triple blends diesel/biofuel/alcohol, exhibiting results that are very close to those obtained using biodiesel or Ecodiesel.

**Keywords:** castor oil; biofuel; selective transesterification; ecodiesel; biodiesel; diesel engine; electricity generator; smoke opacity; Bacharach opacity

#### **1. Introduction**

Nowadays, regardless of the introduction of vehicles that incorporate electric or hydrogen engines, the gradual incorporation of biofuels as substitute of fossil fuels is still mandatory [1]. The use of biofuels palliate the depletion of fossil fuels, minimize the negative impact of greenhouse gases, producing less amount of carbon monoxide, sulfur dioxide and unburned hydrocarbons than fossil fuel [2,3], and also allow continued use of the fleet of cars currently existing, estimated at more than a billion, without modifying the compression ignition (C.I.) engines [4]. Furthermore, biofuels and, specifically the biodiesel, can be easily integrated into the logistic of the global transportation system [5,6]. The gradual replacement of fossil fuels by others of renewable nature involves the introduction of blends diesel/biofuel. In this sense, the objectives pursued by the EU are estimated at 20% of biofuel in the blend in 2020 and 30% in 2030. Despite these objectives are apparently not difficult

to achieve, the enormous amount of glycerol produced during the synthesis of biodiesel makes that other approaches can be considered in order to accomplish the fossil replacement to a higher extent. In this respect, a viable option could be the use of unprocessed vegetable oils in double blends with conventional diesel. This is theoretically possible because all the relevant physicochemical properties, for its use as fuels in conventional diesel engines, are comparable to conventional diesel, with the exception of the viscosity, much higher in oils than in diesel [7]. Another approach is related to the use of alcohols in triple blends with diesel and different biofuels. According to EN 14214, the presence of alcohols in fuel and biofuels does not constitute any inconvenience. In fact, according to recent research, the presence of ethanol and other short-chain alcohols has a favorable effect on the emissions of the biofuels [8–11], so it is advisable the addition of certain quantities of ethanol to the diesel No. 2 standard. These mixtures constitute the so-called E diesel, oxidiesel or oxygenated diesel, which apart from reducing the emissions of the C.I. engines, improves the flow properties (viscosity) and the essential parameters that limit the application of diesel when operating at low temperatures [12], like the "cloud point" (CP), "pour point" (PP), cold filter plugging point temperature (CFPP), or point of occlusion of the cold filter (POFF), viscosity, and emission levels of the motors, without any significant negative effect in most of the parameters that define the quality of biodiesel [13–16]. Thus, the use of alcohols in triple mixtures diesel/biofuel/alcohol would allow to replace larger amounts of fossil diesel than those achieved using double diesel/biofuel blends. In fact, the utility of the triple diesel/biodiesel/ethanol blends has been patented under the name of Diesterol [17,18]. These blends reduce emissions, viscosity and flash point, together to a slight reduction of the engine power [13–18].

Considering the biofuel synthesis and taking into account the available technology, the conventional biodiesel production described by the standard EN 14214, present as the main drawback the glycerol generated as byproduct, which is a 10% by weight of the total of biodiesel produced [19]. For instance, the Lurgi's biodiesel fabrication technology is based on two successive transesterification reactions of TG with methanol to form FAME and glycerol in the presence of an alkaline catalyst [20]. At the end of the reaction, the mixture is neutralized by adding hydrochloric acid. A subsequent counter-current washing step removes by-product components and gives a "ready for use" biodiesel after final drying step. The washing step of biodiesel to eliminate glycerol residues, that must be less than 0.02%, provides an additional complication. On one hand, it requires processing with high energy costs. On the other hand, it requires a high consumption of water. A possible solution to this problem is the production of a new type of biodiesel that integrates glycerol in the form of a soluble derivative. Thus, the production of glycerol is avoided, and, at the same time, the atomic performance of the process is increased, since all the reactive raw materials are transformed into a biofuel [21,22]. In this sense, our Research Group has accomplished the transesterification of triglycerides with ethanol to produce monoglycerides (MGs) as soluble derivatives of glycerol employing different lipases as catalysts. Hence, through the partial transesterification of one mole of triglyceride (TG) with ethanol, two moles of ethyl esters (FAEE) and one mole of monoglyceride (MG) are generated, obtaining a biofuel called Ecodiesel, Figure 1 [23–29].

However, the high cost of lipases compels to search for a more cost-effective process for producing Ecodiesel, such as the employ of supported KF or CaO as heterogeneous catalysts [28,29]. Thus, the Ecodiesel, constituted by the 2/1 mixture of FAME/MG can be obtained under kinetic control of the chemical process, using a catalyst less basic than the alkali metals usually employed as homogeneous catalysts in the biodiesel production. However, further attempts to reduce the economic cost of the process should be investigated.

**Figure 1.** Selective transesterification of a triglyceride molecule through the application of enzymatic catalysis to produce Ecodiesel, a biofuel similar to biodiesel, constituted by two molecules of ethyl esters of fatty acids and a monoglyceride molecule.

In addition to the abovementioned drawback of glycerol generation during the biodiesel production, another important issue related to the biodiesel production is the ethical conflicts surrounding the production of biodiesel from edible oils, instead of using these edible oils exclusively for feeding purposes. To overcome this, the production of the so-called second-generation biofuels, which are obtained from non-edible oils and also from recycled waste oils, using procedures with minimum waste generation and high atom efficiency, are being taken into account [30,31]. Among the non-edible oils, Castor oil (*Ricinus communis* L.) should be a promising option because it has a large international market, since it is widely used in chemical and pharmaceutical industry [32,33] and it can be grown on marginal lands and in semi-arid climates [34]. However, it is not considered a good raw material to produce biofuels [35] because of the high kinematic viscosity that it exhibits (241.5 cSt). Likewise, the kinematic viscosity of the biodiesel obtained by its transesterification with methanol is also too high for being employed in combustion engines (15 cSt). In fact, very recent studies [36–39] have determined that Castor oil biodiesel can be only used in a 20–30% in mixtures with fossil diesel, i.e., BCO20 to BCO30. In this respect, some researches, regarding the use of triple diesel/biodiesel/alcohol blends in order to increase the amount of biodiesel of castor oil in these blends, have been initiated [13].

Another possibility, which is barely studied at present, is that Castor oil can be employed in its pure form blending with fossil diesel [40–42]. Castor oil exhibits, in fact, advantageous rheological properties for being employed directly as biofuel, such as high solubility in very diverse organic compounds, compatibility with polar liquids of low viscosity, such as alcohols; high values of Cetane Index (81.1) and Flammability Point (229 ◦C); low cloud point −10 to −18 ◦C, (−23 ◦C in FAME), and crystallization (pour point) −30 ◦C (−45 ◦C in FAME), which allow its use in motor oil high performance, as a lubricant and in hydraulic braking systems [34,43].

Hence, in this research, two different aspects have been addressed. On one hand, to further reduce the economic cost of the Ecodiesel production, a basic homogeneous catalytic process at the experimental conditions that are soft enough to achieve the transesterification of the esters of primary alcohols, positions 1 and 3 of glycerol, without affecting the carbon 2, a secondary alcohol, which is less reactive, has been investigated. To do so, a commercial sodium methoxide is employed as catalyst. Furthermore, in order to avoid the glycerol production that surely occurs employing methanol and an homogeneous catalyst, the possibility of using ethanol as solvent in a methanol/ethanol mixture to attenuate the higher reactivity of methanol has been studied.

On the other hand, to solve the problem of the high viscosity of Castor oil, this research deals with the possibility of increasing the amount of renewable material in biofuel blends, here so-called diesel additive, through the application of diesel/biofuel/alcohol triple blends. In this sense, the most suitable double and triple blends, according to their viscosity values, are applied as a biofuel in a conventional diesel engine, operating as an electric generator. Besides, the quality of the emissions obtained by using these biofuels was also evaluated from the opacity values of the generated smokes.

#### **2. Results**

#### *2.1. Synthesis of Ecodiesel by Kinetic Control of the Conventional Procedure of Biodiesel Production*

As aforementioned, to obtain a selective alcoholysis of triglycerides, the use of milder conditions than those usually employed in conventional biodiesel production (FAME) is required. As can be seen in Table 1, the biodiesel production from two different oils, sunflower oil (SO) and castor oil (CO), reached a total conversion and selectivities higher than 90% to FAME + MG at 60 ◦C, 60 min of reaction time and employing 4 mL of sodium methoxide per 100 mL of oil. For its part, in order to substitute the lipases by a homogeneous catalyst in the Ecodiesel synthesis, the use of ethanol to reduce the glycerol production has been accomplished, due to ethanol is less reactive than methanol and can avoid the break of the ester bond of secondary alcohol). Table 2 shows the results obtained in the selective ethanolysis of Sunflower oil (SO) at room temperature (25 ◦C) and 30 min of reaction time, using different proportions of ethanol (EtOH) and methanol (MeOH).

**Table 1.** Results obtained in the transesterification of different oils. Reaction conditions: 100 mL of oil, 20 mL of methanol and 4 mL of sodium methoxide, 60 ◦C and 60 min of reaction time. In all the cases, the conversion is 100%. Sel. = FAME + MG. In the methanol fraction, 13 mL in this case, 2–4 g of glycerol and 3.5–6.5 g of MG were obtained.


**Table 2.** Results obtained in the transesterification of 500 mL of Sunflower oil. Reaction conditions 25 ◦C, 30 min of reaction time and 300 rpm. The neutralization is carried out with of H3PO4. No phase separation was observed.


As can be seen in Table 2, non-appreciable differences in the viscosity of the mixtures have been obtained. Furthermore, the selectivities of the reactions are between 77% and 90%. Thus, we have selected the reaction conditions more favourable from an economic point of view (Entry 5), since the least amount of catalyst and alcohols are employed. This proportion can be easily transformed to a ratio if we consider the reaction of 100 mL of Sunflower oil. In this case, the reaction mixture would be oil/EtOH/MeOH/NaOMe ratio of 100/5/5/1.

Once the mixture of reactants was fixed, different reaction parameters such as reaction temperature (Table S1) and reaction time were also studied. Table 3 compiles the conversion values and selectivity to different products in the Ecodiesel synthesis from SO (EcoSO) and CO (EcoCO), at different reaction times, 30 ◦C and with a mixture of methanol/ethanol. It is remarkable that, operating at the same experimental conditions as those employed in the Ecodiesel production over different enzymatic extracts [23–27] but with an oil/EtOH/MeOH/NaOMe ratio of 100/5/5/1, it is possible to obtain biofuels that can be mixed with fossil diesel for being employed in diesel C.I. engines, by reducing the starting viscosity of the oil in practically 1/3 (Tables 1 and 3). Besides, an atomic efficiency of practically 100% is obtained. According to complementary studies, the biofuels showed in Tables 2 and 3 can be used in mixtures with diesel fossil, up to 20% of Biofuel in the total mixture for Ecodiesel of sunflower

oil (Table S2), and up to 15% if the Biodiesel is obtained from castor oil [41]. In these blends, the viscosity values drop to the limits established in EN 14214 standards.

**Table 3.** Results obtained in the transesterification of castor oil. Reaction conditions: 100 mL of either Castor oil or Sunflower oil (\*), 5mL of ethanol, 5 mL of methanol and 1.0 mL of sodium methoxide, at 30 ◦C and 300 rpm. The neutralization is carried out with 0.1 mL of H3PO4. No phase separation was observed.


*2.2. Characterization of Biofuel Mixtures, with Fossil Diesel and Alcohols, to Allow Their Use in Conventional Compression Ignition Engines, without Any Modification*

#### 2.2.1. Double Blends: Diesel/Biofuels

The mixtures of oils with diesel exhibit very different viscosity values, depending on the percentages of oil in the diesel, and also depending on the oil employed. However, the rheological properties of these mixtures are hardly affected by the nature of the oils, mainly at low oil concentrations, so it is possible to obtain mixtures with a 10% *v*/*v* of oil in diesel, suitable of being use in conventional diesel engines, conforming to EN 14214. Thus, it is very remarkable that, despite the high viscosity of castor oil, it behaves analogously to sunflower oil when it is blending with diesel. In fact, considering BSO10 blends, the viscosity reached a 3.6 cSt while a BCO10 blend reaches a viscosity value of 4.5 cSt.

Taking into account the use of different mixtures of diesel with Ecodiesel, with lower viscosity than the starting oils, higher values of fossil fuel replacement can be achieved. Thus, with sunflower oil, employed in this research as a reference for waste cooking oils (second generation biofuel), it is possible to obtain Ecodiesel (EcoSO) under the experimental conditions as in Table 3, with viscosities in the range of 12 to 15 cSt, whereas in a blend of diesel with a 30% of EcoSO, a viscosity value of 5.1 cSt is obtained, suitable to be employed directly in conventional diesel engines. In addition, we must not lose the perspective of how the biofuels is considered to be employed, i.e., in different mixtures, 20% in 2020, and 30% in 2030. To do so, the process described in Tables 2 and 3 seems to be the most appropriate to produce the biofuels in the most easy (and economical) way, although obviously they can be obtained with higher quality, even with a similar quality to biodiesel. However, higher concentrations of catalyst and alcohols must be used for this purpose, implying then a higher cost.

Regardless of the procedure employed for the Ecodiesel synthesis, it is important to establish what percentages of mixture with diesel can be employed in every case. Thus, for the Ecodiesel obtained with the experimental conditions indicated in Table 3 (viscosity = 64 cSt), the maximum mixing level for being employed in C.I. engine is 15%, attaining a viscosity value of 5.8 cSt. However, if a lower viscosity Ecodiesel is employed, either EcoCO or EcoSO, obtained by using higher concentrations of catalyst and alcohol, they can be employed even up to 30% blending with diesel, Figure 2. These results show that EcoCO exhibits similar rheological properties as BCO when they are blended with diesel [41], up to 25% *v*/*v*.

**Figure 2.** Kinematic viscosity values (cSt), obtained at 40 ◦C, of mixtures composed for (**a**) increasing amounts of castor oil ecodiesel (EcoCO) in fosil diesel, and (**b**) increasing amounts of sunflower oil ecodiesel (EcoSO) in fosil diessel.

#### 2.2.2. Triple Blends: Diesel/Biofuels/Short Chain Alcohols

In Figure 3, the viscosity values of different triple blends Diesel/EcoSO/alcohol, are collected. Two different alcohols, ethanol and 2-propanol have been employed. First of all, it should be highlighted that the blends were prepared adding increasing amounts of a mixture of Ecodiesel/alcohol 4:1 (*v*/*v*). This 4:1 proportion has the maximum amount of ethanol for the blends to be stable, taking into account that the subsequent adding to diessel fuel will be further dilute the mixture (triple blends). Thus, these Ecodiesel/alcohol mixtures present a viscosity of 7.75 cSt with ethanol y 7.32 cSt with 2-propanol, so they cannot be applied as biofuels because of their high viscosity values (the diesel engine works at viscosities between 3–5 cSt, according to EN 14214). However, by adding different amount of diesel to these double mixtures, a sufficient reduction of the viscosity values was achieved, in the range of 3–5 cSt, which can be perfectly employed in a C.I. engine. Taking into account the viscosity results shown in Figure 3a, with a blend diesel/EcoSO/ethanol 60/30/10, it is possible to incorporate a 40% of renewable compounds as diesel additive, which is almost the double than that achieved with the blends diesel/Ecodiesel, where a maximum of 25% of EcoSO in diesel was obtained (Figure 2b). Regarding the data obtained with 2-propanol, Figure 3a, with only a 40% of diesel in the blend, we are able to obtain a biofuel with a suitable viscosity for being employed in C.I. engines, replacing a 5% more of diesel than using ethanol.

Then, the high capability shown by alcohols to reduce the viscosity of their blends with EcoSO can be also employed to optimize the blends of these alcohols with oils. This can be especially useful in the case of castor oil, which according to the data collected in Figure 4, has a greater mixing capability with alcohols than EcoCO itself. This behavior can be explained by the high content of hydroxyl groups of Castor oil, exceptional among fatty acids, which confer it a high capability for being blended with alcohols, including ethanol, in any proportion. The appropriate viscosity values, lower than 5.0 cSt, are reached when the concentration of alcohols in the blends are higher than 60%. However, in EcoCO/alcohol blends, Figure 4b, the expected decrease in viscosity was not observed, taking into account the much lower viscosity of the Ecodiesel, compared to the starting Castor oil. Therefore, there is no advantages in using EcoCO instead of pure Castor oil.

Furthermore, as can be seen in Figure 4, any blend of CO either with ethanol or isopropanol, in which the alcohol content was above 65% exhibits the appropriate viscosity for being use in conventional diesel engines. This is especially interesting in the case of ethanol, due to its renewable character, which means that these mixtures of castor oil with ethanol, 35/65, are 100% renewable biofuels. However, it is foreseeable that mixtures of CO with alcohols would not have an adequate

behavior in terms of the power reached and the high consumption of biofuel, due to the low calorific power of the alcohols.

(**b**)

**Figure 3.** Viscosity values of the different triple mixtures obtained blending conventional Fossil Diesel, Ecodiesel from sunflower oil (EcoSO) and either ethanol (**a**) or 2-Propanol (**b**). In all the cases, the Ecodiesel/alcohol ratio is constant, with 25% alcohol.

**Figure 4.** Viscosity values of the different double mixtures obtained blending ethanol (1.3 cSt) or isopropanol (1.63 cSt) with castor oil (227.0) cSt (**a**) and with the Ecodiesel from castor oil (EcoCO) (**b**).

This behavior makes possible the assumption that triple blends diesel/EcoCO/alcohol and diesel/CO/alcohol can be employed as biofuels. In this respect, the triple mixtures diesel/EcoCO/ethanol and diesel/EcoCO/2-propanol have been prepared by blending fossil diesel with increasing amounts of a 1:1 blend of EcoCO/alcohol, and the viscosity values obtained for those blends are shown in Figure 5.

According to the results shown in Figure 5a, it is possible to incorporate a 40% diesel additive, e.g., 60/20/20 in the blend diesel/EcoCO/ethanol, so it is possible to achieve a higher substitution of diesel fuel than that achieved with the double blends diesel/Ecodiesel. Furthermore, no differences in the viscosity values obtained for the blends using either ethanol or 2-propanol were observed (Figure 5).

(**b**)

**Figure 5.** Viscosity values of the different triple mixtures obtained blending Ecodiesel from castor oil (EcoCO), a conventional diesel and either ethanol (**a**) or isopropanol (**b**). In all cases, the Ecodiesel/alcohol ratio is 1:1.

However, when Castor oil is directly employed in these triple blends, diesel/CO/alcohol, a different behavior depending on the alcohol employed is observed. As can be seen in Figure 6, for blends diesel/Castor oil/alcohol, the replacement of fossil diesel by diesel additive in amounts higher than 20% is not possible, e.g., 80/10/10. This fact is due to the low solubility of ethanol and diesel. However, if 2-propanol is employed, a 40% of diesel fuel can be replace by diesel additives, as can be seen for the blend 60/20/20, diesel/Castor oil/2-propanol.

(**b**)

**Figure 6.** Viscosity values of the different triple mixtures obtained blending Castor Oil (CO), fossil diesel and either ethanol (**a**) or isopropanol (**b**). In all cases, the Ecodiesel/alcohol ratio is constant, 1:1.

These observations are really important, since Castor oil can be employed directly in the blends mixing with 2-propanol up to 40% of the total blend. Thus, there is no need to carry out the transesterification of this oil in order to get a biofuel from this raw material.

#### *2.3. Evaluation of Different Biofuels from Their Behavior in a Conventional Internal Combustion Engine*

#### 2.3.1. Double Blends Diesel/Biofuels

At this point, it must be indicated that the physico-chemical properties of Ecodiesel (corrosion, calorific value, density, cetane index, viscosity and several properties of biofuels at low temperatures) to be employed as biofuel, are similar to those exhibited by Biodiesel. Therefore, only a very short number of rheological properties are relevant enough for being taken into account, considering its application in motor tests. These properties, i.e., viscosity, pour point (PP) and cloud point (CP), are collected in Table 4 for the different blends here studied. In principle, all the blends shown in Table 4 can be employed as biofuels in conventional diesel engines without any modification. However, the presence of MG in the Ecodiesel slightly increases the viscosity values of the mixtures, being this fact more noticeable in blends with more than 40% of Ecodiesel. In this sense, it is not advisable the use of double blends in which Ecodiesel is present in a percentage higher than 50%.

Once the reological properties were obtained, all the blends collected in Table 4 were tested in an internal combustion engine. For a better comparison, one more experiment at the same conditions but operating with a conventional diesel, was also carried out.

**Table 4.** Rheological properties of diesel/biofuel blends, either with EcoSO, or conventional Biodiesel from sunflower oil (BSO).


The results of power generation and opacity (contamination parameter) obtained with all the double blends, diesel/Biodiesel, are shown in Figure 7. First of all, it can be seen that at 4 kW of Power demand, the highest Power Generation is achieved by the engine, regardless of the biofuel employed. In addition, all the blends exhibited similar values of Power generation, which was also analogous to that exhibited by the fossil diesel. However, considering the opacity parameter, it is highlighted that the lowest opacity value was obtained with the BSO50 blend, which presents also the highest content of Biodiesel.

**Figure 7.** (**a**) Power generated (in Watts), based on the power demanded (in kWatts) and (**b**) Opacity values generated in the smokes (in Bacharach units) as a function of the power demanded of different double mixtures of Fossil Diesel with Biodiesel of Sunflower Oil (BSO).

The double blends diesel/EcoSO also exhibited a better power generated values than the fossil diesel at high power demands, although their behavior at power demands of 1, 2 and 3 kWatts is pretty similar, even with the lowest amount of EcoSO in the blend, Figure 8a. Furthermore, the values of opacity are better than that obtained with fossil diesel and, in general, better than those obtained with the diesel/biodiesel blends, Figure 8b.

**Figure 8.** Results obtained by using different mixtures of Ecodiesel from sunflower oil with diesel fossil: (**a**) Power values generated (in Watts), based on the power demanded (in kWatts). (**b**) Opacity values generated in the smokes (in Bacharach units) as a function of the power demanded.

#### 2.3.2. Triple Blends: Diesel/Biofuels/Short Chain Alcohols

The behavior of the more characteristics triple blends diesel/Ecodiesel/alcohol has been investigated to estimate the influence of the percentages of Diesel Additive (Ecodiesel + alcohol) in the rheological properties abovementioned, i.e., viscosity and pour point and cloud point temperatures. Thus, the results obtained for the most characeristic blends are compiled in Table 5.


**Table 5.** Rheological properties of triple blends diesel/biofuel/alcohol, in the proportions indicated as percentages.

The same as with double blends, the triple ones were tested in a diesel combustion engine and the results of power generated, opacity and consumption are shown in Figures 9 and 10. On one hand, the power generated is similar to that obtained with fossil diesel, mainly up to 3 kW of power demanded. However, two different behaviors have been observed from 4 kW, and it is related to the ethanol taking part of the mixture. As it was explained before, ethanol has a low calorific power. In fact, the higher the amount of ethanol in the blend, the lower power generated at 5 kW (entry 5, Figure 9).

On the other hand, it is noteworthy the results related to the opacity for all the triple blends here studied. These opacity values are quite lower than that obtained with fossil diesel, indicating the great importance of these blends to reduce the emissions of pollutants. Among all of these blends, the behavior of the Castor oil is surprising, no matter with which alcohol is blended. In fact, in addition to the lowest values of opacity, the lowest consumption in L/h were achieved at high power demanded, even better than that obtained with fossil diesel, Figure 10.

**Figure 9.** (**a**) Power generated values (in Watts), based on the power demanded (kWatts) and (**b**) Opacity values generated in the smokes (in Bacharach units) as a function of the power demanded of the blends of Table 5.

**Figure 10.** Consumption values of the blends in Table 5 as function of the power demanded of the engine.

#### **3. Discussion**

#### *3.1. Synthesis of Ecodiesel by Kinetic Control of the Conventional Procedure of Biodiesel Production*

From the results here obtained, we can conclude that among all the available procedures to reduce the viscosity of the oils by chemical reaction, the selective alcoholysis exhibits the greater atomic efficiency (100%), taking full advantage of the triglyceride molecule as biofuel [21,22]. As can be observed from Tables 1–3, by this procedure, the viscosity of the Ecodiesel is slightly higher than that exhibited by the biodiesel. This fact can be explained because the total alcoholysis generates three molecules of FAME. But, in exchange, glycerol generation, in a 10% by weight of the total biodiesel produced, is avoided.

Another important aspect from the economical point of view is that the synthesis of biodiesel needs, in general, more energetic reaction conditions. In fact, complementary studies about the Ecodiesel production [44], Table S2, have also shown that the reaction temperatures in the range of 20 to 60 ◦C, produce very little influence on the biofuel obtained, since the viscosity in this range is reduced by less than 10%. This fact implies that the activation energy in the alcoholysis of the secondary carbon of glycerol (position 2) is higher than corresponding to the activation of the primary alcohols (positions 1 and 3 of glycerol). This, among other things, would justify that both processes take place through a different reaction mechanism. From a practical point of view, this implies that the synthesis of Ecodiesel, regardless of other parameters involved, it should always be performed in the range 20–40 ◦C, usually at room temperature. Furthermore, as can be seen in Table 2, the use of EtOH together with MeOH allow the 1,3-selective transesterification over sodium methoxide, attaining an Ecodiesel similar to that obtained with lipases, improving the process from an economic point of view.

In addition, another important aspect in the evaluation of the homogeneous catalytic synthesis of Ecodiesel here reported is related to its application to waste cooking oils. This process can be applied to waste oils subjected to a very simple cleaning process, since after the synthesis reaction of the Ecodiesel it is necessary to carry out a very simple filtering operation, to eliminate the salts of alkaline phosphates produced by the reaction with phosphoric acid. In the filtering process, the solid residues usually mixed with the wasted oil are also easily eliminated. In other words, biofuels can be "cleaned" in the same synthesis process of Ecodiesel. Therefore, with pure oils, the only waste is practically reduced to the Na3PO4 salts produced in the last neutralization step. Furthermore, the presence of ethanol as a co-reactant or solvent of methanol, works very positively because it improves the viscosity, and prevents that methanol becomes separate of the biofuel.

Everything raised above leads us to conclude that the production of Ecodiesel in mild conditions, employing homogeneous basic catalysts improves by far the previous studies which implies the use of lipases [23–26] or heterogeneous basic catalysts [28,29], not only from a technical point of view but also form the economic one, above all if we consider that this Ecodiesel can be also employed as Biofuel in different double and triple blends.

#### *3.2. Evaluation of Different Biofuels from Their Behavior in a Conventional Internal Combustion Engine*

From the oils here studied as raw materials in the Ecodiesel production, sunflower oil and castor oil, both Ecodiesel produced, EcoSO and EcoCO, exhibited similar rheological properties as those described in the literature for the Biodiesel from Castor oil [41], as they can be mixed in a 25% with fossil diesel, Figure 2, to run in conventional diesel engines. However, it is possible to achieve similar, or even higher, replacement of fossil diesel using pure castor oil in triple mixtures with alcohols. The peculiar structure of Castor oil allows a high solubility with both, fossil diesel and alcohols, including ethanol (Figure 4). Thus, triple mixtures can be obtained with Castor oil (Figure 6) with a high level of fossil diesel replacement, with no need of transforming the starting oil into Ecodiesel (Figures 4 and 5).

Taking into account the two different alcohols proposed in this research, and, according to the results shown in Figures 5 and 6, if ethanol is employed in the triple mixtures, the EcoCO favors a higher substitution of fossil diesel than the use of pure Castor oil, 60/20/20 and 80/10/10, respectively. However, if isopropanol is employed, similar results are obtained with EcoCO and pure CO. Anyway, a 40% of fossil diesel substitution was always achieved.

These double and triple blends have been evaluated according to the viscosity, as this parameter gives information about which ones can be considered for its use in a diesel engine, as the limits of viscosity required by conventional diesel engines are in the range 3–7 cSt, although European regulations (EN 14214) establish the interval 3–5 cSt, Figures 3–6.

Regarding the behavior of the blends at low temperatures, Table 5, conventional biodiesel and ecodiesel, in double and triple mixtures, solidify at temperatures around −10 ◦C and are completely frozen at temperatures around −12 ◦C. In contrast, the corresponding temperatures for diesel fossil are around −16 to −18 ◦C, for pour point and cloud point, respectively. Thus, for the values here obtained, it can be observed that in blends whose percentage of biofuel is higher, the temperatures of pour point and cloud point are higher, solidifying before. Independently on the slightly worst temperatures of pour and cloud point than those exhibited by diesel, the blends studied in the present work gave acceptable results for environmental conditions where the temperatures are not reduced below −10 ◦C. In addition, the use of any alcohol as an additive would influences positively these parameters, since its addition promotes a better performance against low temperatures.

The blends, which meet the requirements for being employed as fuel in diesel engines, were tested at different values of power demanded. As can be seen in Figures 7 and 8, independently of the biofuel tested, a stabilization of the generated power values takes place above 4 kW of demanded power. Likewise, the values of power generated are slightly higher when blends with Ecodiesel were employed, especially in the proportions 20% and 30%, Figures 7a and 8a.

About the pollution produced during the combustion of the different samples, the results show that the higher the power demanded, the higher the emission of pollutants, as it was expected. In fact, an increase in the power demand from a power equivalent to the engine working at idling speed to the highest demand of 4 or 5 kW, the soot formation doubles. However, it is very remarkable that for blends diesel/biodiesel with biodiesel content higher than 10% (BS20, BS30, BS40 and BS50) and blends diesel/Ecodiesel with Ecodiesel content higher than 5%, the opacity values obtained were always lower than that obtained with fossil diesel, Figures 7b and 8b.

Regarding the triple blends diesel/biofuel/alcohol, the power generated values and the opacity as a function of the power demanded are shown in Figure 9. On one hand, it should be noted that all the samples exhibited similar values of power generated than diesel fuel (Figure 9a), although those in which ethanol is employed as alcohol showed the lower ones, especially the blend diesel/CO/EtOH (50/25/25). However, if 2-propanol is employed, independently on the biofuel, EcoSO, EcoCO or CO, very good results of power generated were obtained.

On the other hand, it must be remarked the surprisingly low opacity values obtained with all the triple blends tested, more than 10% lower than the opacity obtained with diesel, Figure 9b.

According to the engine fuel consumption, the results in Figure 10 show that EcoSO in triple blends with isopropanol exhibit higher consumption than diesel, whereas blended with ethanol exhibit lower consumption than diesel. For its part, the blends with EcoCO and CO behave the opposite, i.e., a lower consumption was obtained using 2-propanol. Thus, considering the results previously exposed, the triple blend diesel/CO/2-proOH (50/25/25) exhibit the best performance, with higher power generation, lower emission of pollutants and also lower fuel consumption, indicating its feasibility for being employed in diesel engines.

#### **4. Materials and Methods**

#### *4.1. Evaluation of Sodium Methoxide as Homogeneous Catalyst in the Selective Alcoholysis to Obtain Ecodiesel*

The selective alcoholysis of the Sunflower oil (food quality) and Castor oil (Panreac, Castellar Del Valles, Spain) were carried out over commercial sodium methoxide (Panreac, Sodium methylate solution 30% in methanol PS) as homogeneous catalyst. The reactions were performed in a 250 mL flask immersed in a temperature-controlled water bath, at atmospheric pressure using methanol as alcohol. Different reaction conditions have been studied, such as temperature, 20–60 ◦C, and reaction time, Figure 11.

**Figure 11.** Experimental dispositive for the transesterification reaction.

In general, 120 mL of oil (0.1 mol) were mixed with variable proportions of methanol and sodium methoxide. The changes are easily visualized as a ratio, for example, 100/10/1 indicates 100 mL of oil, 10 mL of methanol and 1 mL of sodium methoxide. Variable amounts of ethanol are sometimes also incorporated in some experiments, which act as a solvent, given its lower activity with respect to methanol. It has been taken as standard measure that 0.1 mol = 91.0 g of oil. The weigh was performed on a Mettler AJ50 precision balance (precision of ± 0.01 g), depositing the sample directly into a 250 mL two-neck round bottom flask.

#### *4.2. Determination of the Content of FAMEs, FAEEs and Glycerides (MG, DG and TG) in the Reaction Products by Gas Chromatography*

The determination of the content of methyl esters, ethyl esters and different glycerides in the biofuel samples has been carried out by a chromatographic method developed in previous researches [23]. Thus, an HP 5890 Series II gas chromatograph with a HT5 (25 m × 0.32 mm I.D × 0.1 μm, SGE, Supelco) Aldrich Chemie capillary column and equipped with a flame ionization detector (FID) was employed. Cetane (n-hexadecane) is used as an internal standard

This method basically consists of a modification and integration of two official methods, UNE EN ISO 14103 (esters) and UNE EN ISO 14105 (glycerides), to quantify the content of glycerol, ethyl esters and glycerides (mono, di and triglycerides), respectively. The ethyl esters of palmitic acid, stearic acid, oleic acid, linoleic acid, and linolenic acid were commercially obtained from AccuStandard (New Haven, CT, USA), and hexadecane (cetane) was obtained from Sigma-Aldrich (St. Louis, MO, USA).

Considering that castor oil or sunflower oil are constituted by a mixture of fatty acids (mainly linoleic, oleic, palmitic, and stearic acids, in sunflower oil, or ricinoleic in castor oil) in variable proportions, the results obtained are expressed as the relative amounts of the corresponding methyl esters (FAME, fatty acid methyl esters), monoglycerides (MG), and diglycerides (DG) that are integrated in the chromatogram. The number of diglycerides (DG) and triglycerides (TG) that have not reacted is calculated from the difference to the internal standard (cetane). Thus, the Conversion includes the total amount of triglyceride transformed (FAEE + MG + DG) in the methanolysis process, and Selectivity refers to the relative amount of FAEE + MG obtained.

#### *4.3. Determination of Kinematic Viscosity of Biofuels*

The kinematic viscosity has been measured in an Ostwald-Cannon-Fenske capillary viscometer (Proton Routine Viscometer 33200, size 150), determining the time required for a certain volume of liquid to pass between two marked points on the instrument, placed in an upright position. From the flow time (t), expressed in seconds, we obtain the kinematic viscosity expressed in centistokes, υ = C·t. Where C is the calibration constant of the measurement system in mm2/s2, which is specified by the manufacturer (0.040350 mm2/s2 at 40 ◦C, in this case). All measures have been carried out in duplicate and are presented as the average of both, proving that there is no greater variation of 0.35% between measures, as required by the standard ASTM (American Society for Testing and Materials) D2270-79, Method for calculating viscosity index from kinematic viscosity at 40 and 100 ◦C.

#### *4.4. Determination of the Pour Point and Cloud Point of Biofuels*

Cloud point and the Pour point are determined by introducing the different double or triple samples, of different composition, in a digitally controlled temperature refrigerator for twenty-four hours; after this time the loss of transparency of the solutions is evaluated. The appearance of turbidity in the samples is indicative that the cloud point temperature has been reached (cloud point). After a progressive decrease in temperature, the samples are kept under observation until they stop flowing (pour point).

#### *4.5. Assessment of Energy Performance and Air Pollutant Emissions in a Diesel Engine Electric Generator, Fueled with Different Blends of Biofuels*

The mechanical and environmental characterization of a compression ignition diesel engine has been carried out, working at a rate of 3000 rpm coupled to an AYERBE electric generator, 5KVA, 230v type AY4000MN, for the generation of electricity, operating at a crankshaft constant rotation rate and under different degrees of demand for electrical power. This is achieved by connecting heating plates of 1000 watts each one (Figure 12a). This diesel engine will operate at a constant rate of rotation of the crankshaft and torque, so that the different values of electrical power obtained will be an exact consequence of the mechanical power obtained after the combustion of the corresponding biofuel. Different tests are obtained by providing to the engine double and triple mixtures of different biofuels in different percentages. The electrical power generated can be easily determined from the product of the potential difference (or voltage) and the electric current intensity (or amperage), equation (1), both obtained by means of a voltmeter-ammeter [45,46].

$$\text{Electrical Power Generalted (Watts)} = \text{voltage (Volts)} \times \text{amperage (Amps)}\tag{1}$$

The consumption of the diesel engine with the different biofuels employed was calculated estimating the speed of consumption of the engine, with a given fuel, when operating under a determined demand of electric power.

On the other hand, the contamination degree is evaluated regarding the opacity of the smoke generated in the combustion process. This is obtained by using an opacimeter—TESTO 308 opacity meter—under the operating conditions studied (Figure 12b). All the results obtained with the biofuels evaluated were compared with the corresponding measurements obtained with conventional diesel. The opacimeter, is a device designed to estimate the amount of soot emitted by diesel engines. Unlike gasoline engines, where the amount of carbon monoxide (CO) and hydrocarbons (HC) is measured to assess the quality of combustion and toxic emissions, only the amount of coal is analyzed in diesel engines (soot). These are tiny particles in suspension which can not be treated as a gas, that is, they can not be quantified through the gas analyzer. The opacimeter is basically composed of three components: measuring chamber, analyzer and a portable terminal. The outlet of the exhaust pipe of the engine is connected, through a pipe and a hose, with the measuring chamber and partial samples of the exhaust gases are taken. It is called partial since only part of the gases enter the machine and the rest is lost in

the atmosphere. The gases enter inside a tube and through a sensor the intensity of the light (turbidity) is measured, to then calculate the density of the particles. The tube has a source of halogen light at one end and a receiver at the other so that when there is no gas inside the tube, the light intensity is not affected. The result of such measurements is reflected visually on a filter paper. Besides, this value can be expressed as a percentage of Opacity (being 100% totally cloudy and 0% totally clear) or as an equivalent number called the k value (Opacity Bacharach) the scale runs from white (0 Bacharach unit) to black (9 Bacharach units), as established by ASTM D 2156-94, Standard Test Method for Smoke Density in Flue Gases from Burning Distillate Fuels [47]. It must be clarified that all the measured were repeated at least three times, attaining an experimental error lower than 5%.

**Figure 12.** (**a**) Electrogenerator AYERBE, 5KVA, 230v tipo AY4000MN, heating plates of 1000 watts of power each and voltmeter-ammeter devise (yellow colour, on the floor); (**b**) TESTO 308 opacity meter, which operates as established by ASTM D 2156-94, Standard Test Method for Smoke Density in Flue Gases from Burning Distillate Fuels.

#### **5. Conclusions**

In this research, some Biofuels have been synthesized employing a commercial basic homogeneous catalyst, sodium methoxide, at milder conditions than those employed to obtain conventional biodiesel, to favor a selective 1,3 methanolysis of triglycerides, achieving a 100% atomic efficiency in the synthesis of Ecodiesel. In this respect, sunflower oil and castor oil have been evaluated. Besides, the behavior of diesel/biofuel blends, as well as the behavior of several triple blends, diesel/biofuel/alcohol have been also studied. Furthermore, the differences between Ecodiesel and conventional biodiesel, as well as the alcohol employed in the blends, in what proportions must be added together with fossil diesel, to obtain the rheological properties to operate in conventional diesel engines without making any modifications have been also determined. The main conclusions obtained in this research can be summarized as follows:

It has been demonstrated, for the first time, that the use of EtOH as solvent in the 1,3-selective transesterification of triglycerides improve the diffusion between methanol and TG, allowing its reaction at milder conditions, since an increase in the temperature of the reaction does not improve the Ecodiesel production.

Ecodiesel from both, sunflower and castor oil, has been successfully obtained at the reaction conditions: Oil/ethanol/methanol/catalyst ratio 100/5/5/1, employing sodium methoxide as homogeneous catalyst at 30 ◦C and at a stirring speed of 300 rpm and 15 min of reaction time.

This research has shown that the chemical route here proposed for the synthesis of Ecodiesel is able to reduce the production costs to a great extent, in comparison with enzymatic routes and heterogeneous catalysed routes.

About the rheological properties of the double and triple blends here studied, it has been verified that the viscosity, Pour Point and Cloud Point values of the different samples allow their use as biofuels in conventional diesel engines.

The results obtained using a compression ignition diesel engine show that 4 kW of power demand gives the greatest engine power generation, independently on the fuel employed.

With the blends diesel/biodiesel, a 50% of fossil diesel can be replaced by a biofuel obtained from a renewable source, whereas in the blends diesel/Ecodiesel, the amount of fossil diesel that can be replaced is lower, 30%. This fact is due to the presence of MG in the Ecodiesel, which increase the viscosity above 6 cSt.

The use of triple blends diesel/biofuel/alcohol allows us to increase the diesel replacement, even employing directly Ecodiesel or Castor oil. Thus, it is possible to use pure Castor oil in the proportions 50/25/25, diesel/CO/2-propanol, in conventional diesel engines, with a performance and level of consumption like fossil diesel, but with an appreciably lower emission of pollutants. In the case of the Ecodiesel of sunflower oil, (and used cooking oils, for use as second generation biofuels) it is possible to obtain triple mixtures with any ethanol in a proportion 50/40/10 diesel/EcoSO/ethanol which also exhibits very good results in terms of consumption and emission of pollutants.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4344/9/1/40/s1, Figure S1: Chromatograms obtained in the sunflower oil alcoholysis (a). The initial sunflower oil is in black (a). Commercial diesel fuel chromatogram (b), Table S1: Viscosity values of different blends diesel/Ecodiesel from Castor oil (EcoCO), Table S2: Viscosity and Selectivity values of Ecodiesel of Sunflower oil (EcoSO) and reaction products obtained at the same reaction conditions as in Table 2, after 60 min. The Conversion was also 100%. Sel. = FAE + MG, being FAE = FAME + FAEE.

**Author Contributions:** This research article is part of the doctoral thesis of B.H., and A.P., directed by D.L., J.M.H and F.M.B., who in general conceived and designed the experiments. C.L., J.C., A.A.R. and R.E. made substantive intellectual contributions to this study, making substantial contributions to conception and design of it, as well as to the acquisition, analysis and interpretation of data. Furthermore, D.L. and R.E wrote the paper. All the authors have been also involved in drafting and revising the manuscript, so that everyone has given final approval of the current version to be published in Catalysts Journal.

**Funding:** This research received no external funding.

**Acknowledgments:** This research is supported by the MEIC funds (Project ENE 2016-81013-R), Junta de Andalucía and FEDER (P11-TEP-7723), that cover the costs to publish in open access.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**


#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
