**Application of Sulfated Tin (IV) Oxide Solid Superacid Catalyst to Partial Coupling Reaction of** α**-Pinene to Produce Less Viscous High-Density Fuel**

#### **Seong-Min Cho 1, Chang-Young Hong 2, Se-Yeong Park 3, Da-Song Lee 1, June-Ho Choi 1, Bonwook Koo <sup>4</sup> and In-Gyu Choi 1,5,6,\***


Received: 19 April 2019; Accepted: 16 May 2019; Published: 18 May 2019

**Abstract:** Brønsted acid-catalyzed reactions of α-pinene have been studied because of their ability to produce various types of fragrance molecules. Beyond this application, dimeric hydrocarbon products produced from coupling reactions of α-pinene have been suggested as renewable high-density fuel molecules. In this context, this paper presents the application of a sulfated tin(IV) oxide catalyst for the partial coupling reaction of α-pinene from turpentine. Brønsted acid sites inherent in this solid superacid catalyst calcined at 550 ◦C successfully catalyzed the reaction, giving the largest yield of dimeric products (49.6%) at 120 ◦C over a reaction time of 4 h. Given that the low-temperature viscosity of the mentioned dimeric products is too high for their use as a fuel in transportation engines, lowering the viscosity is an important avenue of study. Therefore, our partial coupling reaction of α-pinene provides a possible solution as a considerable amount of the isomers of α-pinene still remained after the reaction, which reduces the low-temperature viscosity. On the basis of a comparison of the reaction products, a plausible mechanism for the reaction involving coinstantaneous isomerization and coupling reaction of α-pinene was elucidated.

**Keywords:** solid superacid catalyst; sulfated tin(IV) oxide; α-pinene partial coupling; renewable high-density fuel

#### **1. Introduction**

Turpentine, one of the most widely produced plant-derived secondary metabolites, is a mixture of monoterpenes. It is made up mainly of α-pinene and its isomers, such as β-pinene and camphene. Depending on the production method, it is categorized as gum turpentine (produced from oleoresins of conifers), wood turpentine (produced from aged pine stumps), sulfate turpentine (produced by the kraft pulping process), and sulfite turpentine (produced by the sulfite pulping process) [1]. Concerns about fossil fuel depletion and environmental destruction urge us to develop alternative energy resources. In this regard, turpentine, in which C10 hydrocarbons form major components, has been

reported as a potential candidate for providing biofuel blends for fueling both spark ignition engines and compression ignition engines [2–5].

A coupling reaction by which C20 hydrocarbons can be synthesized from renewable α-pinene has been devised for ramjet propulsion [6–8], but not for conventional jet fuels. This is because the relatively high carbon number of these compared to that of petroleum-derived fuels tremendously increases their low-temperature viscosity, which limits the suitability of using dimeric products alone for transport fuel [9]. Thus, the partial coupling of monoterpene hydrocarbons was suggested as one possible solution [10]. This agrees with the results reported by Harvey et al., who suggested blending dimeric products with monoterpene hydrocarbons such as α-pinene, thereby resolving the viscosity problem of dimeric products [6].

Coupling of monoterpene hydrocarbons has been studied using various types of heterogeneous acid catalysts, such as Nafion, Nafion SAC-13, montmorillonite K10, Al-incorporated MCM-41, Pd-Al-incorporated MCM-41, phosphotungstic acid supported on MCM-41, phosphotungstic acid supported on SiO2, Zeolite Hβ, and silica-alumina aerosol [6–8,11–14]. Given that the isomerization of monoterpene hydrocarbons is usually attributed to Brønsted acid activity and that Brønsted acids can also assist in their coupling reactions undertaken in harsh conditions, it has been thought that the catalytic activities of the above catalysts are caused by Brønsted acid sites and the role of Lewis acid sites is less significant.

Herein, we used sulfated tin(IV) oxide (SO4 <sup>2</sup>−/SnO2) as a solid superacid catalyst to carry out partial coupling of α-pinene for the production of less viscous high-density fuel molecules. SO4 <sup>2</sup>−/SnO2 catalysts have been used in various types of organic reactions, such as dehydration of sorbitol and xylose to isosorbide and furfural, respectively, esterification of free fatty acids, the Penchmann condensation reaction, and deprotection of silyl ether groups [15–19]. Successive chemical precipitation and immersion in diluted sulfuric acid yielded this catalyst from tin chloride pentahydrate in a facile procedure. In this study, the catalyst successfully furnished dimeric products from α-pinene with its isomers in solventless conditions. On the basis of the results, a plausible mechanism for the isomerization and coupling reaction, in which Brønsted acid catalysis plays a central role, was also suggested. To our knowledge, no attempt has been made so far to propose a mechanism considering both reactions together.

#### **2. Materials and Methods**

#### *2.1. Catalyst Preparation*

To prepare the catalysts, tin oxide powder obtained from tin (IV) chloride pentahydrate (SnCl4·5H2O) by chemical precipitation, followed by drying, was used as a precursor. Briefly, a transparent 0.1 M tin chloride solution was prepared by dissolving SnCl4·5H2O in distilled water. To hydrolyze the tin chloride complex, a 28 wt% aqueous ammonia solution was added dropwise under vigorous stirring. The addition was stopped when the pH of the solution reached approx. 8. After a precipitate was separated from clear supernatant liquid, thorough washing was carried out with a 4 wt% ammonium acetate solution by centrifugation. The white product was then dried in an oven at 105 ◦C for 24 h and ground into a fine powder. The prepared tin oxide (SnO2) powder (10 g) was placed in a round flask containing 3 g of sulfuric acid diluted with 20 mL of distilled water. After sufficient stirring at 80 ◦C, water was removed in vacuo and sulfuric acid-treated SnO2 was dried and stored in an oven at 65 ◦C. This precatalyst was dried further at 120 ◦C for 12 h, followed by calcination at 450, 550, 600, and 650 ◦C for 4 h.

#### *2.2. Catalyst Characterization*

To understand the properties of the prepared catalysts and elucidate their catalytic activity, we conducted X-ray diffraction (XRD) analysis, field-emission scanning electron microscopy (FE-SEM), thermogravimetric analysis (TGA), and temperature-programmed desorption of ammonia (NH3-TPD). The XRD patterns of the catalysts were collected to compare the crystal structures. The surface morphology of the catalysts was investigated using FE-SEM. Energy dispersive X-ray spectroscopy (EDS) and elemental mapping were also performed as part of the SEM investigation. To study the thermal properties of the catalysts, TGA was carried out. NH3-TPD was performed to characterize the improvement in the catalyst acidity. The detailed analysis conditions are described in the Supplementary Materials.

#### *2.3. Catalytic Tests*

In a typical experiment, turpentine (5 g), tridecane (2 g, GC internal standard), and catalysts (0.1 g) were added without any solvent to a 50 mL glass flask equipped with a magnetic Teflon-coated stirrer and a reflux condenser. The reactor was then loaded on a preheated aluminum heating block and stirred vigorously. Upon completion of the reaction, the reactor was removed from the heating block and immediately cooled to room temperature using a cold-water bath. After cooling, the crude reaction mixture was diluted with n-hexane (100 mL) and filtered over a Celite pad. To estimate the product composition by the internal standard method, a diluted reaction mixture was analyzed using a gas chromatograph (7890B) equipped with a DB-5ms column (30 m × 250 μm, 0.25 μm thickness), a mass spectrometer detector (5977A), and a flame ionization detector (Agilent Technologies, Santa Clara, CA, USA). The yield of the products and the conversion rate of α-pinene were calculated using the following equations:

$$\text{Yield of product } i \text{ (\%)} \quad = \frac{\text{Weight of product } i}{\text{Weight of circle reaction mixture}} \times 100,\tag{1}$$

$$\text{Conversion rate of } \alpha - \text{pinene } (\%) = \frac{\text{Consumed weight of } \alpha - \text{pinene}}{\text{Initial weight of } \alpha - \text{pinene}} \times 100. \tag{2}$$

#### **3. Results and Discussion**

#### *3.1. Catalyst Characteristics*

#### 3.1.1. Catalyst Surface Morphology

To investigate the surface morphological properties of intact tin oxide (SnO2) and sulfated tin oxide (SO4 <sup>2</sup>−/SnO2), field emission-scanning electron microscopy (FE-SEM) observation was carried out. The SEM images in Figure 1a,b show that intact SnO2 has a rough surface, whereas SO4 <sup>2</sup>−/SnO2 exhibits a smooth one because, although both surfaces consist of globular nanoparticles, the size of the nanoparticles in SO4 <sup>2</sup>−/SnO2 was much smaller than that of the nanoparticles in intact SnO2. Energy dispersive spectroscopy (EDS) analysis confirmed the presence of sulfur-containing groups on the surface of SO4 <sup>2</sup>−/SnO2 (Figure S1a–d). In addition, elemental mapping images clearly show a uniform distribution of sulfur atoms on the surface (Figure 1c and Figure S1e).

#### 3.1.2. Catalyst Crystal Structure

The crystal structures of tin oxide (SnO2) and sulfated tin oxide (SO4 <sup>2</sup>−/SnO2) calcined at 550 ◦C was characterized using powder X-ray diffraction, as shown in Figure 2. There were obvious differences in the diffraction peaks for the species. The diffractogram of intact SnO2 shows a good match to the ICDD powder diffraction file of cassiterite SnO2 (PDF 00-041-1445), which means that it has a tetragonal crystal structure similar to that of cassiterite SnO2 (P42/mnm space group). Major diffraction peaks observed at 2θ = 27◦, 34◦, and 52◦ could be indexed to the (110), (101), and (211) planes of cassiterite SnO2, respectively. Immersion in diluted sulfuric acid before calcination significantly influenced the crystal structure. Even though the characteristic diffraction peaks of cassiterite SnO2 were detected in the diffractogram of SO4 <sup>2</sup>−/SnO2, the intensities of the peaks decreased significantly and their breadth broadened considerably. Because this weakening and broadening denotes diminished crystallinity and crystallite size [16,20], Figure 2 suggests an amorphous structure and small-sized crystallites of SO4 <sup>2</sup>−/SnO2. In accordance with the results presented in previous studies related to sulfated metal oxides such as SnO2 and ZrO2, sulfate groups on the surface of sulfuric acid immersed SnO2 seem to hamper both aggregation and crystallization themselves during calcination [17,21,22]. This result also coincides with the FE-SEM images shown in Figure 1, which shows the difference in the size of globular nanoparticles making up the surfaces.

**Figure 1.** Field-emission scanning electron microscopy (FE-SEM) images of the surface of (**a**) intact SnO2; and (**b**) SO4 <sup>2</sup>−/SnO2; (**c**) SEM-energy dispersive X-ray spectroscopy (EDS) elemental mapping images of SO4 <sup>2</sup>−/SnO2. The samples were prepared by calcination at 550 ◦C for 4 h. In the overlay image of SEM and sulfur atom mapping, the gray color represents an area in shadow or surface facing away from the EDS detector.

**Figure 2.** X-ray diffraction patterns of intact SnO2 (black line) and SO4 <sup>2</sup>−/SnO2 (red dashed line). The samples were calcined at 550 ◦C. On the basis of the ICDD powder diffraction file (PDF 00-041-1445), the peaks were indexed as tetragonal crystal structure as cassiterite SnO2 (P42/mnm space group).

#### 3.1.3. Catalyst Thermostability

The catalyst was designed to be applied to the coupling of α-pinene under somewhat harsh reaction conditions (≥100 ◦C). Our query was whether the catalytic activity over sulfated tin oxide (SO4 <sup>2</sup>−/SnO2) could be sustained during the reaction. Because sulfate groups introduced on the surface of the SO4 <sup>2</sup>−/SnO2 support by immersion in diluted sulfuric acid are responsible for the catalytic activity [23], the thermal decomposition behavior of sulfate groups was investigated by thermogravimetric analysis (TGA). Figure S2 shows the TGA graph of intact SnO2 and SO4 <sup>2</sup>−/SnO2 calcined at 550 ◦C. SO4 <sup>2</sup>−/SnO2 displayed two distinguishable weight loss sections, whereas intact SnO2 presented gradual weight reduction throughout the temperature range. The weight reduction of intact SnO2 and the first weight loss of SO4 <sup>2</sup>−/SnO2 from 120 to 550 ◦C can be explained by the desorption

of chemisorbed water molecules and dehydroxylation on the surface of the SnO2 support [20,22]. The difference between the extents of the weight reduction experienced by them may be attributed to the differences in the surface functional groups, especially sulfate groups, on which water can interact [23]. In addition, significant weight loss was observed at temperatures higher than 600 ◦C in the case of SO4 <sup>2</sup>−/SnO2, which is attributed to the decomposition of sulfate groups on the surface of the SnO2 support [20,22].

#### 3.1.4. Catalyst Acidity

To qualitatively evaluate the acidities of the intact tin oxide (SnO2) and sulfated tin oxide (SO4 <sup>2</sup>−/SnO2) catalysts, we carried out temperature programmed desorption of ammonia (NH3-TPD). The TPD profiles of desorbed ammonia clearly show an improvement in the acid strength of the catalysts when SnO2 was immersed in diluted sulfuric acid before calcination (Figure 3a). Peak deconvolution can help us interpret the meaning of the overlapped peaks. As can be seen in Figure 3b, the profile of intact SnO2 calcined at 550 ◦C consisted mainly of two peaks centered at approx. 372 ◦C (dark yellow) and 513 ◦C (dark cyan), which are attributed to weak and strong chemisorption of ammonia. On the other hand, the SO4 <sup>2</sup>−/SnO2 calcined at 550 ◦C had four kinds of acid sites which present four peaks centered at approx. 240 ◦C (magenta), 353 ◦C (blue), 531 ◦C (purple), and 736 ◦C (orange) (Figure 3c). In particular, the strongest acid sites were attributed to the desorption of ammonia from sulfate groups on the surface of SO4 <sup>2</sup>−/SnO2 [24]. This interpretation coincided with the TGA results for SO4 <sup>2</sup>−/SnO2 shown in Figure S2, which displays a significant weight loss at temperatures higher than 600 ◦C due to the decomposition of sulfate groups on the surface. In addition to the obvious differences in the peak number and overall peak intensities in the TPD profiles, the amount of NH3-uptake by the catalysts also suggests that immersion in diluted sulfuric acid generates much more acid sites than those developed in intact SnO2 (Table 1).

**Figure 3.** (**a**) Temperature programmed desorption of ammonia (NH3–TPD) profiles of intact SnO2 (black line) and SO4 <sup>2</sup>−/SnO2 (red dashed line); (**b**–**c**) Deconvolution of overlapped peaks in the profiles. The samples were prepared by calcination at 550 ◦C for 4 h; (**d**) NH3–TPD profiles of SO4 <sup>2</sup>−/SnO2 calcined at 550 ◦C (red dashed line), 600 ◦C (blue dotted line), and 650 ◦C (orange line).

The effect of the calcination temperature on the acidity of the sulfated catalyst was evaluated based on the amount of NH3 uptake (Table 1). Generally, increasing the calcination temperature reduces the total acid sites of sulfated or phosphated solid acid catalysts [25,26]. Our results coincide with those reported. However, one difference was found above the 600 ◦C calcination temperature, with a new peak centered at approx. 696 ◦C appearing and the disappearance of the peak centered at approx. 736 ◦C in the TPD profiles (Figure 3d).


**Table 1.** NH3 uptake of intact SnO2 and SO4 <sup>2</sup>−/SnO2 calcined at different temperatures.

<sup>1</sup> Intact SnO2 was calcined at 550 ◦C. <sup>2</sup> Calcination temperature. <sup>3</sup> Calculated based on the NH3–TPD results.

#### *3.2. Catalytic Tests and Reaction Mechanism*

#### 3.2.1. Effect of Catalyst Calcination Temperature on Partial Coupling Reaction of α-Pinene

When it comes to preparing a sulfated metal oxide catalyst using sulfuric acid treatment, a sintering process is important because the promotion of sulfate groups on the surface of metal oxide occurs during calcination and the acid sites generated by these sulfate groups offer the essential catalytic activity [23]. Figure 4 shows stark differences in the effect of the calcination temperature on the catalytic activity of sulfated tin oxide (SO4 <sup>2</sup>−/SnO2). The yield of dimeric products from α-pinene (**1**) was the largest for the catalyst calcined at 550 ◦C (48.9 ± 1.2%), while almost no conversion to dimeric products and significantly lowered production were observed below and above this temperature, respectively. When we tried to correlate the yield of dimeric products with the amount of the total acid sites of catalysts, unlike previous papers [25,26], there was a large discrepancy in the calcination temperature range from 450 to 550 ◦C (Table 1 and Figure 4). Although the catalyst calcined at 450 ◦C showed the highest NH3 uptake, it was not able to furnish dimeric products at all. Since a large amount of sulfuric acid can hinder the growth of SnO2 crystals (Figure 2), thereby leading to poor promotion of sulfate groups on the surface of SO4 <sup>2</sup>−/SnO2, the negligible catalytic activity of catalyst calcined at 450 ◦C (α-pinene conversion rate <20%) was attributed to sulfuric acid remaining in large quantities during the calcination process. Zhang et al. reported that increasing the calcination temperature from 150 to 550 ◦C made sulfate groups of SO4 <sup>2</sup>−/CeO2 transition from surface sulfate states to bulk sulfate states; the catalyst mainly possessing sulfate groups as surface sulfate states worked well in catalytic reduction of NO by NH3. According to the Raman spectra in which only the catalyst calcined at 550 ◦C presented the peaks denoting bulk sulfate states, this transition seems to occur abruptly when the calcination temperature increased from 450 ◦C to 550 ◦C [27]. In this respect, the discrepancy between the yield of dimeric products and the amount of the total acid sites of catalysts can be understood by considering that the catalyst possessing sulfate groups as bulk sulfate states is effective in the coupling reaction of α-pinene. On the other hand, calcination temperatures higher than 550 ◦C caused significant decomposition of sulfuric acid and even sulfate groups on the surface of SO4 <sup>2</sup>−/SnO2 (Figure S2). Therefore, when the same amount of sulfuric acid was treated, a much higher calcination temperature made the catalyst lose sulfate groups. In this regard, the less effective catalytic activity of catalysts calcined at 600 and 650 ◦C can be understood. Finally, to compare the effect of the promotion of sulfate groups on the catalytic activity, we conducted the reaction with intact SnO2 calcined at 550 ◦C, which showed negligible α-pinene conversion (data not shown).

In contrast to dimeric products, catalysts prepared at higher calcination temperatures (600 and 650 ◦C) more readily furnished camphene (**6**) from α-pinene (**1**) and the catalyst calcined at 550 ◦C also yielded a considerable amount of compound **6** (Figure 4). In other words, the conversion to compound **6** competed with the production of dimeric products from compound **1**, and once formed, compound **6** was thought to be indifferent to the homocoupling reaction with these catalysts. This is supported by the fact that several catalysts lack the ability to catalyze homocoupling of compound **6** to form dimeric hydrocarbon products [11].

**Figure 4.** Effect of catalyst calcination temperature on the yields of dimeric products (-) and camphene (•) and the conversion rate of α-pinene (). All the other reaction conditions were maintained for 3 h at 120 ◦C.

3.2.2. Effect of Reaction Time and Temperature on Partial Coupling Reaction of α-Pinene

To investigate the effects of reaction time and temperature on the yield of dimeric products, we selected the sulfated tin oxide (SO4 <sup>2</sup>−/SnO2) catalyst calcined at 550 ◦C based on the results shown in Figure 4. As one can see in Figure 5, the conversion of α-pinene (**1**) was almost 100 ± 0% only after 30 min except for the reaction at 100 ◦C. This consumption furnished dimeric products almost entirely at the beginning of the reaction (Figure 5a). The additional increase in the yield continued up to 3 h at all reaction temperatures. A quite interesting feature was that after 3 h, there was no significant difference among the yields of dimeric products at 100, 110, and 130 ◦C, although the increase in the yield of dimeric products during the reaction was the highest at 100 ◦C (from 13.6 ± 3.2% to 45.0 ± 0.7%). A significant difference after 3 h was only observed at 120 ◦C, which produced the highest yield of all (48.9 ± 1.2%). However, when we carried out the reaction further in these conditions, there was a poor improvement (from 48.9 ± 1.2% to 49.6 ± 0.7%). In other words, only around half of compound **1** yielded dimeric products and therefore our catalyst did not seem suitable for the coupling reaction. However, considering the low-temperature viscosity of dimeric products [9], this partial coupling will be even more appropriate for fuel applications [10].

**Figure 5.** Effect of reaction time and temperature on the yields of (**a**) dimeric products and (**b**) camphene: (-) for 100 ◦C; (-) for 110 ◦C; () for 120 ◦C; () for 130 ◦C; (-) for α-pinene conversion rate except for 100 ◦C condition; (+) for α-pinene conversion rate at 100 ◦C condition. The used catalyst was prepared by calcination at 550 ◦C for 4 h.

Figure 5b shows that the yields of camphene (**6**) decreased depending on the reaction temperature and time. Given compound **6** seems to be less reactive to homocoupling with our catalyst, its decrease during the reaction time implies the possibility of heterocoupling of compound **6** with other species, even including dimeric products. Because such heterocoupling can consume dimeric products and

furnish oligomeric products, reaction temperatures higher than 120 ◦C produced a lower yield of dimeric products.

In order to investigate why the yields of both dimeric products and camphene did not show major changes after 3 h, we conducted a reusable test of the catalyst to prove whether or not the deactivation of the catalyst occurred during the reaction. Unfortunately, the used catalyst did not furnish dimeric products (data not shown). We thought that this was due to the sludge generated by the condensation of various substrates covering the acid sites of the catalyst. An interesting result was that thorough washing with acetone also cannot restore the activity of the used catalyst, but rather eliminated the activity. This is because the acetone washing removed not only the sludge but also the sulfate groups of bulk sulfate states [27]. As mentioned before, sulfate groups as bulk sulfate states of catalysts seem to have key role in the coupling of α-pinene (**1**).

#### 3.2.3. Mechanism of Isomerization of α-Pinene over Sulfated Tin Oxide

Finally, the mechanism of isomerization and coupling reaction of α-pinene (**1**) over the sulfated tin oxide (SO4 <sup>2</sup>−/SnO2) catalyst was considered. The generally accepted mechanism of acid-catalyzed isomerization of compound **1** includes two distinct pathways [28]: ring enlargement rearrangement (path **A**), wherein tri- or bicyclic compounds are produced; and ring opening rearrangement (path **B**), in which monocyclic compounds are formed (Scheme 1). Both pathways are initiated by protonation of olefin in compound **1** generating pinanyl cation (**2**). If the highly strained 4-membered ring in carbocation **2** suffers from ring enlargement by Wagner–Meerwein rearrangement, bornanyl cation (**3**) is formed. This carbocation **3** can produce tricyclene (**4**) via direct deprotonation or camphene (**6**) via deprotonation of isocamphanyl cation (**5**) resulting from 1,2-sigmatropic rearrangement of carbocation **3**. Alternatively, carbocation **2** can be stabilized by the opening of the four-membered ring, which results in *p*-mentha-1-en-8-yl cation (**7**). In light of the similarity in structure and stability between carbocation **7** and *p*-mentha-1-en-4-yl cation (**8**), the 1,2-hydride shift between them is considered reversible. These carbocations (**7** and **8**) can be deprotonated to furnish monocyclic compounds (**9**–**12**).

In the above mechanism, the selectivities of camphene (**6**) from path **A** and limonene (**9**) from path **B** are of interest, and it is undeniable that the selectivities depend on what catalyst is used. Because compound **9** can further isomerize to other monocyclic compounds, catalysts showing the higher (but not by much) selectivity for compound **6** compared to that for compound **9** have been reported more commonly. Kitano et al. reported isomerization of α-pinene (**1**) over an Al2O3-supported MoO3 catalyst, which presented a slightly higher selectivity toward compound **6** than for compound **9,** although the conversion of compound **1** was not good enough [29]. SiO2- or MCM-41-supported H3PW12O40 and MSU-S BEA or Y catalysts showed over 90% conversion of compound **1** and a slightly higher selectivity for compound **6** [12,30,31]. Furthermore, very powerful catalysts such as Fe3<sup>+</sup>-exchanged clinoptilolite, sulfated zirconia, SBA-15 supported TiO2, and sulfuric acid-treated montmorillonite clay have been suggested for the production of compound **6** with prominent selectivity [28,32–35]. However, catalysts that showed the higher selectivity for compound **9** compared to compound **6** have been reported much less frequently. Yamamoto et al. developed a SiO2-supported Pr2O3 catalyst, which showed very high selectivity for compound **9** although the conversion of compound **1** was notably low [36]. In addition to this catalyst, a SiO2-supported AlCl3 catalyst showed higher (but not by much) selectivity for compound **9** with the varying conversion of compound **1**. In this study, SO4 <sup>2</sup>−/SnO2 showed much higher selectivity for compound **6** than for compound **9** with 100% conversion of compound **1** (Figure 5b and Figure S3). This tendency can be justified by the difference between compounds **6** and **9** in reactivity for further isomerization as mentioned previously. It has also been reported that Al-MCM-41 lacks the ability to catalyze homocoupling of compound **6** formed by the isomerization of compound **1** [11]. Given the significant amount of compound **6** still remaining after the coupling reaction was over, not only further isomerization but also homocoupling of compound **6** seems to be difficult with our catalyst, as mentioned in the previous section.

**Scheme 1.** Plausible mechanism of isomerization of α-pinene over sulfated tin oxide catalyst: α-pinene (**1**); pinanyl cation (**2**); bornanyl cation (**3**); tricyclene (**4**); isocamphanyl cation (**5**); camphene (**6**); *p*-mentha-1-en-8-yl cation (**7**); *p*-mentha-1-en-4-yl cation (**8**); limonene (**9**); terpinolene (**10**); α-terpinene (**11**); γ-terpinene (**12**); *p*-mentha-4(8)-en-2-yl cation (**13**); allylic carbocation I (**14**); isoterpinolene (**15**); allylic carbocation II (**16**); *p*-mentha-2-en-8-yl cation (**17**); allylic carbocation III (**18**); *p*-mentha-3-en-1-yl cation (**19**); *p*-mentha-3,8-diene (**20**); *p*-cymene (**21**); *p*-menthene isomers (**22a**–**c**).

In some papers, isoterpinolene (**15**) has been also suggested as a co-product. One plausible mechanism for the formation of compound **15** starts from protonation of terpinolene (**10**), resulting in *p*-mentha-4(8)-en-2-yl cation (**13**) [28]. Not only can direct deprotonation of carbocation **13** generate compound **15**, but also deprotonation of allylic carbocation I (**14**) resulting from carbocation **13** via 1,2-hydride shift can do the same. However, this suggestion has been controversial considering various isoterpinolene/terpinolene concentration ratios, either higher or lower than 1 [37]. A second possible mechanism is initiated by protonation of α-terpinene (**11**) or γ-terpinene (**12**), which gives allylic carbocation III (**18**) and *p*-mentha-3-en-1-yl cation (**19**). This proposal was supported by Salacinski's results which showed the chemical equilibria of *p*-menthadiene species under sulfuric acid at 67 ◦C [38]. When compound **11** or **12** reacted under this condition as a sole starting material, the chemical equilibrium consisted of only compounds **11**, **12**, **15**, and a small amount of *p*-mentha-3,8-diene

(**20**), where stabilization by the formation of a conjugated diene was considered the driving force. In addition, the author described reaction coordinate diagrams with allylic carbocations (**14** and **18**) as reaction intermediates. The presence of compound **20** in our results, even though the quantity of it was relatively small, also seems to indicate this mechanism. Additionally, the successive transformation of *p*-mentha-1-en-8-yl cation (**7**) into allylic carbocation II (**16**) and *p*-mentha-2-en-8-yl cation (**14**) via 1,3- and 1,5-hydride shifts, respectively, was suggested as the other possible route for the formation of compound **15** [39]. Behr and Wintzer also reported that compound **15** was formed as a major side product when the hydroaminomethylation of compound **9** was carried out with a [Rh(cod)Cl]2/TPPTS catalyst [40], which means that compound **9** can be a linchpin when it comes to the production of compound **15**. As seen in Figure S3, the yield of compound **15** seems to follow the same trend as that of **9** along the reaction time. This also suggests that the third suggested pathway chiefly occurs during the isomerization of α-pinene (**1**) in the case of our catalyst.

In addition to the above-isomerized products, *p*-cymene (**21**) and *p*-menthene isomers (**22a**–**c**) were detected in the reaction mixture. The simultaneous formation of compounds **21** and **22a**–**c** can be explained by disproportionation between α-terpinene (**11**) and γ-terpinene (**12**) [6,11]. Moreover, dehydrogenation of compounds **11** and **12** was suggested to justify the production of compound **21** with the generation of hydrogen gas [41,42]. Given the para position of methyl and isopropyl groups therein, compound **21** was generally reported as the target product not only when α-pinene (**1**) was used in neat form [41,43] or as a major constituent of crude sulfate turpentine [44], but also when limonene (**9**) was used as a sole starting material [42,45]. The concentration of compound **21** in the reaction mixture gradually increased with reaction time (Figure S3). This is because it did not participate in further reactions, including both the isomerization and coupling reaction in our catalytic system [7,12].

#### 3.2.4. Mechanism of Coupling of α-Pinene over Sulfated Tin Oxide

The lack of knowledge about the molecular structure of dimeric products obtained from monoterpene is attributed to the simultaneous homo- and heterocoupling that occurs for the starting materials and the isomers therein [6,11,12]. Furthermore, it being hard to isolate only one dimeric product from a product mixture, the study of their molecular structure has proven difficult. These phenomena were also observed in our results; although the reaction started with α-pinene (**1**) as a sole substance, besides isomerization, a variety of dimeric products were concurrently produced (Figure S4). Nevertheless, some reports have suggested several possible molecular structures without an understanding of the complicated reaction system [7,11,46].

Acid-catalyzed coupling reactions of monoterpene hydrocarbons generally involve three steps: first, the protonation of olefin in monoterpene giving carbocations; next, the attack of olefin (nucleophile) in another monoterpene on the previous carbocation (electrophile) furnishing a dimeric carbocation with the formation of a new C–C bond between the nucleophile and the electrophile; and, finally, the deprotonation of this carbocation giving dimeric products. Of course, dimeric carbocations formed by nucleophilic attack on monomeric carbocations, or even by the re-protonation of dimeric products, can suffer from isomerization and further coupling reactions, which is one reason for the complexity of the reaction mechanism. In addition to coupling reactions that involve protonation/deprotonation, the Diels–Alder reaction between monoterpenes, especially α-terpinene (**11**), has also been suggested as a possible mechanism for the coupling reaction of monoterpenes [47].

In light of the results obtained for a reaction at a relatively low temperature (100 ◦C, Figure S3a), although the concentration of α-pinene (**1**) in the reaction mixture precipitously decreased with reaction time, considering almost all of the dimeric products were yielded in just 30 min, compound dimeric structure (**PD1**). We can also imagine that further isomerization of this possible dimeric product gives, for example, **PD1a** and **PD1b** via ring enlargement and ring opening, respectively.

The isomers of α-pinene (**1**), such as camphene (**6**), limonene (**9**), terpinolene (**10**), α-terpinene (**11**), γ-terpinene (**12**), and isoterpinolene (**15**), and *p*-mentha-3,8-diene (**20**), all of which have olefin in their structure, can partake in coupling reactions as compound **1** does. As electrophiles, stable allylic carbocations (**14**, **16**, and **18**) derived from these monocyclic isomers were thought to play a pivotal role in the coupling reaction [38,39]. This type of coupling can occur from the beginning to the end of a reaction, particularly in a predominant coupling reaction that occurs after compound **1** is completely consumed. In Scheme 2, we suggest some possible dimeric structures (**PD2**–**4**), showing that the isomers react as nucleophiles and electrophiles (carbocations).

**Scheme 2.** Examples of possible dimeric products of the coupling reaction of α-pinene catalyzed using sulfated tin oxide: (**A**) coupling of bornanyl cation (**3**) and α-pinene (**1**) giving possible dimeric product (**PD1**) and its isomers (**PD1a**–**b**); (**B**) coupling of allylic carbocation III (**18**) and α-terpinene (**11**) giving **PD2**; (**C**) coupling of allylic carbocation I (**14**) and limonene (**9**) giving **PD3**; (**D**) coupling of allylic carbocation II (**16**) and camphene (**6**) giving **PD4**.

Although we are not sure whether camphene (**6**) reacts as a nucleophile or electrophile (as isocamphanyl cation (**5**)) during the coupling reaction, it is clear that compound **6** predominantly participates in a heterocoupling (Scheme 2, **PD4**) rather than a homocoupling reaction considering that a significant amount of compound **6** still remained after the reaction. This is in agreement with the results obtained using Al-incorporated MCM-41 [11]. Meylemans et al. asserted that this phenomenon is attributed to the low basicity of compound **6**, which causes poor interactions between the external olefin and the acid sites of the catalyst, thereby making the protonation of compound **6** difficult [7].

#### **4. Conclusions**

In summary, a sulfated tin(IV) oxide catalyst prepared using a facile procedure was applied to the partial coupling reaction of α-pinene to furnish a renewable and less viscous high-density fuel. To evaluate the catalytic activity of the catalyst, we considered the effect of the calcination temperature, reaction time, and reaction temperature, and attempted to rationalize the results using the catalyst characteristics. The catalyst calcination temperature had an enormous influence on the production of dimeric hydrocarbon products, while reaction times and temperatures exceeding 1 h and 100 ◦C affected the reaction to a lesser extent. The highest yield of dimeric products (49.6%) was obtained when the catalyst was calcined at 550 ◦C and the reaction was carried out at 120 ◦C for 4 h. Although

the yield was less than half, we think this value is enough to consider utilizing the reaction products as renewable fuels, because it is not known that the dimeric products alone have a low-temperature viscosity too high for use as a fuel in transportation engines. In other words, the mixture with the isomers of α-pinene can drag down the low-temperature viscosity to the range of transportation fuels. Finally, we described the possible mechanism of the coinstantaneous isomerization and coupling reaction of α-pinene owing to our catalyst acting as a Brønsted acid.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1996-1073/12/10/1905/s1, Figure S1: (a,b) Energy dispersive spectroscopy (EDS) spectra, (c,d) elemental quantitative data obtained from intact SnO2 and SO4 <sup>2</sup>−/SnO2 and (e) higher resolution of elemental mapping image of SO4 <sup>2</sup>−/SnO2, Figure S2: TGA curves of intact SnO2 (black line) and SO4 <sup>2</sup>−/SnO2 (red dashed line), Figure S3: The yields of the monomeric products along the reaction time at (a) 100 ◦C and (b) 110 ◦C, Figure S4: General chromatogram of dimeric products extracted from GC/FID result.

**Author Contributions:** Conceptualization, S.-M.C. and D.-S.L.; Data curation, S.-M.C., C.-Y.H. and B.K.; Formal analysis, S.-M.C., C.-Y.H. and B.K.; Methodology, S.-M.C. and D.-S.L.; Supervision, I.-G.C.; Writing—original draft, S.-M.C., S.-Y.P. and J.-H.C.; Writing—review & editing, C.-Y.H.

**Funding:** This study was supported by Mid-career Researcher Program in Basic Research of National Research Foundation of Korea grant funded by the Korea government (MSIP) (NRF-2016R1A2B4014222).

**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* **Green Diesel Production over Nickel-Alumina Nanostructured Catalysts Promoted by Copper**

#### **Mantha Gousi 1, Eleana Kordouli 1,2, Kyriakos Bourikas 2,\*, Emmanouil Symianakis 3, Spyros Ladas 3, Christos Kordulis 1,2,4 and Alexis Lycourghiotis <sup>1</sup>**


Received: 15 June 2020; Accepted: 15 July 2020; Published: 18 July 2020

**Abstract:** A series of nickel–alumina catalysts promoted by copper containing 1, 2, and 5 wt. % Cu and 59, 58, and 55 wt. % Ni, respectively, (symbols: 59Ni1CuAl, 58Ni2CuAl, 55Ni5CuAl) and a non-promoted catalyst containing 60 wt. % Ni (symbol: 60NiAl) were prepared following a one-step co-precipitation method. They were characterized using various techniques (N2 sorption isotherms, XRD, SEM-EDX, XPS, H2-TPR, NH3-TPD) and evaluated in the selective deoxygenation of sunflower oil using a semi-batch reactor (310 ◦C, 40 bar of hydrogen, 96 mL/min hydrogen flow rate, and 100 mL/1 g reactant to catalyst ratio). The severe control of the co-precipitation procedure and the direct reduction (without previous calcination) of precursor samples resulted in mesoporous nano-structured catalysts (most of the pores in the range 3–5 nm) exhibiting a high surface area (192–285 m2 g<sup>−</sup>1). The promoting action of copper is demonstrated for the first time for catalysts with a very small Cu/Ni weight ratio (0.02–0.09). The effect is more pronounced in the catalyst with the medium copper content (58Ni2CuAl) where a 17.2% increase of green diesel content in the liquid products has been achieved with respect to the non-promoted catalyst. The copper promoting action was attributed to the increase in the nickel dispersion as well as to the formation of a Ni-Cu alloy being very rich in nickel. A portion of the Ni-Cu alloy nanoparticles is covered by Ni0 and Cu<sup>0</sup> nanoparticles in the 59Ni1CuAl and 55Ni5CuAl catalysts, respectively. The maximum promoting action observed in the 58Ni2CuAl catalyst was attributed to the finding that, in this catalyst, there is no considerable masking of the Ni-Cu alloy by Ni<sup>0</sup> or Cu0. The relatively low performance of the 55Ni5CuAl catalyst with respect to the other promoted catalysts was attributed, in addition to the partial coverage of Ni-Cu alloy by Cu0, to the remarkably low weak/moderate acidity and relatively high strong acidity exhibited by this catalyst. The former favors selective deoxygenation whereas the latter favors coke formation. Copper addition does not affect the selective-deoxygenation reactions network, which proceeds predominantly via the dehydration-decarbonylation route over all the catalysts studied.

**Keywords:** green diesel; renewable diesel; Ni catalyst; biofuel; hydrodeoxygenation; Cu-promotion effect

#### **1. Introduction**

The emission of carbon dioxide from the combustion of fossil fuels (carbon, oil, and natural gas) is responsible for the global warming whereas the increasing demands for fossil fuels is expected to lead to their progressive depletion during the 21st century [1]. The replacement of fossil fuels by renewable ones is actually critical for facing such a double problem [1–3]. Undoubtedly, biomass is an important source of renewable energy. The animal fat and plant oils triglyceride biomass is very attractive because it is much less complex than lignocellulosic biomass whereas the ratio of oxygen to the combustible carbon and hydrogen atoms in a molecule of triglyceride is relatively small. The catalytic transformation of triglycerides into n-alkanes in the diesel range (green diesel) was initially studied over noble metals and conventional NiMo or CoMo/γ-Al2O3 sulfide catalysts [4]. Nickel non-sulphided catalysts have gained much interest in the last decade [5–16]. The previously mentioned transformation over nickel non-sulphided catalysts is realized by hydrotreatment (temperature: 240–360 ◦C, hydrogen pressure: 10–80 bar) whereas the relevant chemistry, stated in the next paragraph, is now well established.

The removal of oxygen without or with very small fragmentation extent of the side chains of triglycerides is called Selective DeOxygenation (SDO). The first step of the SDO network is the rapid hydrogenation of olefinic bonds of side chains of triglycerides. It is followed by the gradual decomposition of the O-C bonds in the glycerol side resulting progressively to di-glycerides and then mono-glycerides and then to free fatty acids and propane. The free fatty acids could be transformed by direct decarboxylation (deCO2) into n-alkanes with an odd number of carbon atoms (mainly n-C17 and n-C15) and carbon dioxide. However, this pathway, if any, has very low probability above nickel catalysts [4]. Much more probable is the second pathway whereby the free fatty acids are reduced to the corresponding fatty aldehydes by water removal. The aldehydes are then decarbonylated, which results in n-alkanes with an odd number of carbon atoms (mainly n-C17 and n-C15) and carbon monoxide. Thus, the SDO via this route takes place by dehydration–decarbonylation (deH2O-deCO). The third pathway involves the reduction of the fatty aldehydes into the corresponding fatty alcohols and their very rapid equilibration. The fatty alcohols are transformed into olefins by dehydration, which are then hydrogenated and leads to n-alkanes with an even number of carbon atoms (mainly n-C18 and n-C16). Thus, the SDO through this route takes place by dehydration (deH2O). In parallel, the fatty alcohols may react with the fatty acids producing long chain esters [6]. These may undergo SDO resulting in hydrocarbons. The above-described SDO network is schematized in Figure S1 (Supplementary Material). In addition to the previously mentioned liquid phase reactions, reactions may take place between CO2, CO, and hydrogen deliberated in the gas phase (reverse water gas shift and methanation).

Recently, we have contributed to this subject by developing a co-precipitation methodology, which ensures severe control of the precipitation parameters. This methodology allowed preparing nickel–alumina catalysts with a high specific surface area, even at high nickel loading [6]. These catalysts are very active in the SDO of sunflower oil (SFO) under solvent-free conditions and a very high SFO volume/catalyst ratio (100 mL/1 g). This work showed that the yield of green diesel over such catalysts is a linear function of the nickel surface exposed. The highest value of the latter and, thus, the maximum catalytic performance was achieved over the catalyst with nickel content at about 60 wt. % in which a good compromise between the specific surface area and nickel loading has been achieved. This preparation methodology was then successfully applied to nickel–zirconia catalysts [7] and to nickel–alumina catalysts promoted by molybdenum [17] and zinc [18]. Concerning the promoted catalysts, the total amount of the nickel content was about 60 wt. % whereas the promoters exhibited their optimum effect at very small loadings (1–6%) compared to the nickel loading (54–59%).

With the idea to extend our study to the copper promoted catalysts, we are surveying the relevant literature. The promoting action of copper concerning nickel catalysts used in the SDO of plant oils, biodiesel, residual fatty raw materials, and related compounds into green diesel has gained the research interest in the last years [19–30]. Yakovlev's group was pioneer in this subject [19–22]. They first studied Ni monometallic and Ni-Cu bimetallic catalysts supported on CeO2, ZrO2, or CeO2-ZrO2 in the SDO of biodiesel [19,20] and then a bimetallic Ni-Cu catalyst supported on γ-Al2O3 in the SDO of methyl palmitate and ethyl caprate [21]. The bimetallic catalysts were proved more attractive due to their efficiency to prevent methane formation. The Ni–Cu/CeO2–ZrO2 catalyst exhibited the

highest performance partly attributed to the presence of a Ni1−xCux (x = 0.2–0.3) solid solution as a constituent of the active center and partly to the development of a mixed cerium-zirconium oxide phase. The formation of the previously mentioned nickel–copper solid solution was also found in the Ni−Cu/γ-Al2O3 catalyst [21]. The Crocker's group has also contributed in the domain [23–26]. In the first work [23], a series of Ni/γ-Al2O3 catalysts were prepared containing 20 wt. % Ni and 0, 1, 2, and 5 wt. % Cu and studied in the SDO of tristearin, stearic acid, and triolein. The maximum performance was obtained over the catalyst 20% Ni–5% Cu/γ-Al2O3. The promoting action of copper was attributed to the increase in the surface of metallic nickel as well as to the suppression of surface coking and, hence, catalyst deactivation. This may reflect the ability of Cu to curb the cracking activity of nickel expressed in the non-promoted catalyst. This was also confirmed in the subsequent articles reported by the group where the un-promoted 20% Ni/γ-Al2O3 catalyst was compared to the promoted 20% Ni–5% Cu/γ-Al2O3 one in the SDO of residual fatty raw materials (yellow grease and hemp seed oil [24], waste free fatty acids, and brown grease [25]). They also compared the 20% Ni–5% Cu/γ-Al2O3 catalyst to the 20% Ni–5% Fe/γ-Al2O3 and 20% Ni-0.5%Pt/γ-Al2O3 ones in the SDO of waste cooking oil (WCO) [26]. A work reported by Jing et al. deals with the copper promoting action in Ni/γ-Al2O3 catalysts containing 20 wt. % Ni and 0, 1, 3, 6, or 10 wt. % Cu in the SDO of biodiesel [27]. The maximum performance was obtained over the catalyst containing 6 wt. % Ni. The decrease in catalytic performance as the copper loading increases from 6 to 10 wt. %, which was attributed to its accumulation on the support surface that leads to pore blocking. As inferred in the first work of the Crocker's group [23], the promoting action of copper was mainly attributed to coking suppression. Two interesting works were reported by the group of Fu in which methanol was used as a hydrogen donor instead of gas hydrogen [28,29]. In the first work [28], copper (30 wt. %), nickel (30 wt. %), and nickel–copper catalysts (15, 30, 60 wt. %, and Cu/Ni weight ratio = <sup>1</sup> <sup>2</sup> ) supported on ZrO2 were studied in the SDO of oleic acid. It was reported that the formation of a Cu-Ni alloy favors SDO and inhibits cracking, which increases the catalytic performance. In the second work [29], copper (60 wt. %), nickel (60 wt. %), copper (40 wt. %)-nickel (20 wt. %), copper (30 wt. %)-nickel (30 wt. %), and copper (20 wt. %)-nickel (40 wt. %) catalysts supported on alumina were studied in the SDO of oleic acid. Cracking of C–C bonds deduced by the presence of cracked paraffins was found over the nickel monometallic catalyst. The formation of the Cu–Ni alloy in the bimetallic catalysts and the presence of partially oxidized copper favor SDO and inhibit the cracking of the C–C bonds, which leads to enhanced catalytic performance. A very recent work reported by Miao et al. [30] deals with two monometallic (10 wt. % Ni/γ-Al2O3, 10 wt. % Cu/γ-Al2O3) and five bimetallic NiCu/γ-Al2O3 catalysts with Cu/Ni ratios equal to 1/9, 3/7, 5/5, 7/3, and 9/1 used in the SDO of methyl laurate. They concluded that the oxide precursors can be effectively reduced at 420 ◦C for 2 h into the corresponding metallic catalysts, which is comprised of Ni0, Cu0, and NiCu alloy-supported species. The formation of the NiCu alloy promotes the electronic interactions between Ni and Cu, which enhances catalytic performance. The catalyst with the Cu/Ni ratio equal to 7/3 was proved to be the most active. The nickel and the copper active centers favor, respectively, SDO through dehydration–decarbonylation and dehydration.

In conclusion, the promoting action of copper has been demonstrated. Some aspects of the promoting action start to emerge. The copper promoting action is expressed through the increase in the nickel dispersion and the formation of nickel-copper alloy with better catalytic behavior than nickel. It seems that this curbs the C-C cracking activity of nickel, which depresses the formation of cracked paraffins, methane, and carbon deposition and favors the formation of diesel range n-alkanes and catalyst stability.

In the present work, we are continuing our research effort on co-precipitated Ni-Al2O3 catalysts by studying the copper and promoting action in these catalysts concerning the SDO of SFO. Four catalysts containing 60, 59, 58, and 55 wt. % Ni and 0, 1, 2, and 5 wt. % Cu, denoted by 60NiAl, 59Ni1CuAl, 58Ni2CuAl, and 55Ni5CuAl, were synthesized following the previously mentioned rigorous co-precipitation methodology. The catalysts were characterized using various methods and evaluated in the SDO of SFO in a semi-batch high-pressure reactor. Our approach differs from

those mentioned above [19–30] in two points. In our study, the Cu/Ni weight ratio ranges from 0.02 to 0.09 whereas such a ratio was much higher in the studies reported so far (0.33 [19,20], 0.38 [21], 0.05–0.25 [23], 0.25 [24–26], 0.05–2. [27], 0.5 [28], 0.5–2.0 [29], and 0.1–9 [30]). The second difference is that the evaluation of the catalysts in the present work was performed under solvent-free conditions and the SFO volume to catalyst ratio is equal to 100 mL/1 g and the reaction time is equal to 9 h. These correspond to an LHSV value equal to 11.1 h−<sup>1</sup> for experiments taken place in fixed bed reactors. These experimental conditions are very hard when compared to the corresponding ones reported in the previous works. The choice of SFO as feedstock was done by taking into account that genetically-modified sunflower grown on marginal land has been identified as sustainable biofuel source because it does not encroach upon arable lands [31].

#### **2. Experimental**

#### *2.1. Synthesis of the Catalysts*

The hydroxide precursors of the catalysts studied were prepared by co-precipitation using an aqueous solution of Al3<sup>+</sup>, Ni2<sup>+</sup>, and Cu2<sup>+</sup> nitrate salts [Al (NO3)29H20, Ni (NO3)26H2O, Cu (NO3)23H2O, E. Merck]. This solution was in a funnel and was added drop-by-drop to a vessel containing 330 mL distilled water in which the pH was adjusted to 8 by NH4OH. NH4OH 30% solution (Carlo Erba Reagents) was used in a pH-control system (Metrhom) for keeping pH equal to 8 in the previously mentioned vessel during co-precipitation. Figure S2 presents the set-up used and gives more details. Even more experimental details have been reported elsewhere [6,7]. The rate of introduction of the mixed nitrate solution into the co-precipitation vessel was equal to 1.2 mL/min instead 0.7 mL/min adopted in Reference [6]. Using this rate, we prepared both the copper promoted and the non-promoted catalysts. The precipitated hydroxides were dried at 110 ◦C for 24 h. The dried hydroxides were decomposed to the corresponding oxides by heating them gradually under argon flow of 30 mL/min for 40 min. This time period was necessary for increasing the temperature from 25 to 400 ◦C. The final catalysts were then synthesized by reduction (activation) of the oxide precursors under hydrogen flow (30 mL/min) at 400 ◦C for 2.5 h.

#### *2.2. Catalysts Characterization*

The physicochemical properties of the catalysts were determined using various techniques. A porosimeter (Micromeritics, Tristar 3000) was used for determining the values of a specific surface area and the pore size distribution. The XRD patterns used for determining the crystal phases and the mean size of nanocrystals (Scherer's relationship) were recorded in a Brucker D8 Advance diffractometer. The catalyst morphology was determined by SEM and their composition by EDS. A SEMJEOL JSM6300 microscope with an ED spectrometer was used in all cases. The catalysts' nanostructure was investigated by TEM using a JEOL JEM-2100 system. The surface analysis of the catalysts was obtained by XPS measurements using a MAX200 (LEYBOLD/SPECS) electron spectrometer. The NH3-TPD experiments for determining the acid sites were carried out in a laboratory-developed set up. The above techniques were applied on the final catalysts. In contrast, the H2-TPR experiments were performed in the oxide precursors. Characterization details have been reported elsewhere [6,7,10,17,18,31–34].

#### *2.3. Catalytic Tests*

A semi-batch reactor was used in all cases. The experiments were carried out at 310 ◦C, H2 pressure, a rate equal to 40 bar and 100 mL/min, respectively, and ratio of sunflower oil volume to catalyst mass equal 100 mL/1 g. The catalytic experiments were monitored for 9 h and performed under solvent-free conditions. Samples withdrawn from the reactor liquid phase were analyzed by GC (Shimadzu GC-2010, column: SUPELCO, MET-Biodiesel, l = 14 m, d = 0.53 mm, tf = 0.16 μm) and GC-MS (GC-MS-QP2010 Ultra). The accuracy of the catalytic results was determined by performing several runs twice. The results differed less than 2% in all cases. Experimental and theoretical mass balance determined for all catalytic tests differ less than ±3%. Experimental details were reported in previous contributions [6,7,10,17,18,31].

#### **3. Results and Discussion**

#### *3.1. Catalysts Characterization*

The SEM microphotographs recorded at various magnifications were similar to those published previously [6] and showed the presence of micro grains of different sizes and interparticle macro pores as well. Typical pictures for the 60NiAl and 58Ni2CuAl catalysts are presented in Figure S3. The elemental analysis, performed by EDS, indicated compositions very close to the nominal ones for the catalysts prepared. In fact, the percentage composition in nickel/copper determined for the 60NiAl, 59Ni1CuAl, 58Ni2CuAl, and 55Ni5CuAl catalysts were, respectively, equal to 59.8, 58.8/1.3, 57.3/2.3, and 55.5/6.0. A typical example of the EDS analysis concerning the 58Ni2CuAl catalyst is illustrated in Figure S4.

Figure 1 illustrates the pore volume distribution curves obtained for the catalysts studied. The curves show a mono-modal or bimodal pore size distribution in the range of 2–100 nm. The first and the second peak are centered at about 3 and 4–5 nm, respectively. More precisely, the pore size distribution curve of 60NiAl catalyst exhibits a single peak in the previously mentioned range centered at about 3 nm. The addition of a small amount of Cu (59Ni1CuAl catalyst) provoked the rise of an additional peak centered at 4–5 nm. Further increase of Cu content (58Ni2CuAl catalyst) resulted in the disappearance of the peak at 3 nm and the intensification of that centered at 4–5 nm. Adding higher Cu content (55Ni5CuAl catalyst) provoked the re-appearance of the bi-modal pore-size distribution curve but with lower intensity. In all catalysts studied, an additional broad but less intensive peak is observed in the range of pore diameter >100 nm. Thus, the solids prepared exhibit mainly mesoporous texture with the most pores concentrated in the range of 3–5 nm. This implies solids with a very high specific surface area and a very small mean pore diameter. Inspection of Table 1 shows that this is the case. The somewhat smaller value of specific surface area obtained for the 58Ni2CuAl catalyst reflects the disappearance of the pore-size distribution peak at 3.0 nm. It is notable that the addition of copper is bringing no negligible changes in the texture of the catalysts studied, which depend on the copper content.

**Figure 1.** Pore volume distribution curves obtained for the catalysts studied.


**Table 1.** BET Specific Surface Area (SSABET), Specific Pore Volume (SPV), and Mean Pore Diameter (MPD) of the catalysts studied.

The XRD patterns obtained after activation of the catalysts are illustrated in Figure 2. Inspection of this figure shows that NiO is predominant in the 60NiAl catalyst (main peaks 2θ: 37.2, 43.3, 62.9, and 75.4◦/PDF-2 2003 # 47-1049). Moreover, the peaks assigned to metallic nickel, Ni0, (2θ: 44.3, 51.6 and 76.1◦/PDF-2 2003 # 01-1258) are also observed.

**Figure 2.** XRD patterns of the catalysts studied. The dashed lines correspond to the most important diffractions of the expected phases: (o) Ni0, (\*) NiO or Ni0.95Cu0.5O, (+) Cu0, (ˆ) CuO. The deconvolution of the peak at 2θ ~37◦ (inset) indicates the presence of CuO (red peak).

The addition of Cu to the nickel–alumina catalyst results in the disappearance of the previously mentioned peaks attributed to Ni0. Cu favors the dispersion of Ni (as it is shown below by XPS) and the too small Ni0 crystallites are not any more detectable by XRD. The formation of small amounts of Ni0.95Cu0.05O with diffraction peaks at 2θ 37.2, 43.3, 62.8, and 75.4◦ (PDF-2 2003 # 78-0644) cannot be excluded in the samples 59Ni1CuAl and 58Ni2CuAl while taking into account that the nominal Cu/Ni ratio in these samples is equal to 0.0157 and 0.0319, respectively, whereas this ratio in the mixed oxide is equal to 0.053. Therefore, copper is not in a sufficient quantity to form exclusively Ni0.95Cu0.5O. The absence of additional peaks relevant to copper phases, for instance copper/copper oxide, could be attributed to the copper entrapment in the Ni0.95Cu0.5O and/or to the formation of nanoparticles below the XRD identification limit. In the sample 55Ni5CuAl with nominal Cu/Ni ratio equal to 0.084, the formation of Ni0.95Cu0.5O is also probable, though a portion of copper is identified as metallic, Cu0, through its characteristic peaks at 2θ 50.4 and 74.1◦ (PDF-2 2003 # 85-1326). The deconvolution of the asymmetric peak at 2θ ~37◦ indicated the presence of a monoclinic phase of CuO in this sample (see inset in Figure 2), which is revealed by its characteristic peak at 35.5◦ (PDF-2 2003 # 80-0076). The absence of any peak due to alumina indicates that this oxide is largely amorphous. The nickel and copper surface speciation is further investigated by XPS.

Based on the XRD peak at 37.2◦, we have calculated the mean size of the NiO/Ni0.95Cu0.5O nanocrystals using the Scherer's relationship. It was found equal to about 5.5 nm. The TEM images of the promoted catalysts showed a rather uniform particle size distribution with a similar mean value. A typical image is illustrated in Figure S4.

Ex-situ XPS was performed on the promoted catalysts and resulted in the expected detection of Ni, Al, Cu, and O. Organic carbon from a surface contamination layer was also detected. Figure 3A shows the Ni2p3/2 spectral region of the promoted catalysts studied by XPS. The main 2p3/2 component at BE equal to 856 eV is attributed to Ni2<sup>+</sup>. These spectra show that only traces of Ni0 (less than 5% of the Ni2<sup>+</sup>) are present in these catalysts, which is in good agreement with the XRD results.

**Figure 3.** The XP spectra of the promoted catalysts studied: (**A**) The Ni2p3/2 spectral region, (**B**) the Cu2p3/2 spectral region, and (**C**) the Al2p–Ni3p spectral region.

Figure 3B shows the Cu2p3/2 region of the Cu-containing catalysts. Since the Cu-related spectral intensity is small, especially for the 59Ni1CuAl and 58Ni2CuAl catalysts, the Cu2p3/2 spectra for the latter two catalysts in Figure 3B represent the average of 5 and 3 scans, respectively, whereas only one scan was collected for Cu2p3/2 in the 55Ni5CuAl catalyst as well as for all other spectra presented in this section. The Cu2p3/2 peaks are relatively broad and could be described as a superposition of two states, ~932.5 eV (Cu0 or Cu<sup>+</sup>1) and ~933.7 eV (Cu<sup>+</sup>2). The low BE state is predominant in the 55Ni5CuAl and 59Ni1CuAl catalysts (FWHM 2.6 and 2.8 eV, respectively), whereas the two states are comparable in the 58Ni2CuAl catalyst (FWHM 3.2 eV). The main Cu Auger LMM peaks (not shown) indicate only very small amounts of Cu<sup>0</sup> in the 55Ni5CuAl sample.

Figure 3C shows the broad Ni3p spectral region, which, in all cases, is superimposed by the sharp Al2p peak at a BE of 73.9 ± 0.1 eV, as expected for Al2O3. The area of this peak can be fairly accurately determined by drawing a linear background, as shown in Figure 3C.

The above, in conjunction with the XRD results, provide a deeper insight concerning the nickel–copper surface speciation. Ni2<sup>+</sup>, Cu0, Cu <sup>+</sup>1, and Cu<sup>+</sup><sup>2</sup> phases, which is very likely among Ni0.95Cu0.5O, are present on the surface of the promoted catalysts. The expected formation of alumina was also confirmed.

Table 2 summarizes quantitative surface analysis results obtained by using XPS data (areas of Ni2p3/2, Cu2p3/2, and Al2p peak). More details were reported in Reference [6]. The results are compared with the corresponding calculated nominal atomic bulk compositions, based on the catalyst preparation procedure. The first row in Table 2 shows the corresponding data from measurements on the non-promoted 60NiAl catalyst [6]. One has to notice that the XPS-derived Ni/Al ratio is always larger than the nominal one in the bulk of the corresponding Cu-promoted catalyst (the more so, the higher the Cu content). Furthermore, this ratio is always larger than the corresponding ratio for the non-promoted catalyst (the more so, the higher the Cu content) even though the nominal Ni content in the Cu-containing catalysts should be, in all cases, somewhat smaller than that in the non-promoted one. This suggests that the presence of Cu species tends to increase the dispersion of the Ni-phase in proportion to its content, which is in agreement with the literature [23].

**Table 2.** Quantitative XPS results and comparison with the respective calculated nominal atomic compositions in the bulk.


(1) Estimated uncertainty ±10% on the stated Ni and Cu values. (2) Estimated uncertainty ±20% on the stated ratios.

Another interesting observation concerns the relative surface concentration of copper in the promoted catalysts reflected in the Cu/Al ratios. The XPS Cu/Al ratio is lower than the nominal one in the 59Ni1CuAl catalyst. This presumably indicates that the much larger amount of NiO nanoparticles in this catalyst partially cover the copper species during the Ni0.95Cu0.5O phase. This is also reflected in the value of XPS Cu/Ni atomic ratio, which is much lower than the nominal value. In the 58Ni2CuAl catalyst, the XPS Cu/Al atomic ratio is comparable to the nominal one and the value of XPS Cu/Ni atomic ratio is close to the nominal value. Both indicate that there is no considerable masking of the Ni0.95Cu0.5O by the NiO nanoparticles. The situation is very different in the 55Ni5CuAl catalyst where both the XPS Cu/Al and Cu/Ni atomic ratios are higher than the corresponding nominal ones. These indicate that the CuO and Cu0 nanoparticles detected by XRD may situate on the top of Ni0.95Cu0.5O nanoparticles. Thus, the joint use of XRD and XPS characterization shed light on the relative location of the nickel and copper nanoparticles formed on the promoted samples in addition to surface speciation.

The previously mentioned ex-situ characterization indicated the formation of nickel and copper non-metallic phases. Only in the 55Ni5CuAl catalyst, metallic copper was, in addition, detected. The absence of metallic phases could be attributed either to no formation of these phases upon activation or to the extensive surface re-oxidation of the Ni<sup>0</sup> and Cu0 due to the atmospheric exposure. In order to clarify this point, we performed H2-TPR. Figure 4 shows the H2-TPR curves of the catalysts studied.

**Figure 4.** H2-TPR profiles of the catalysts studied.

The H2-TPR profile of the sample 60NiAl is characterized by two well-distinguished peaks. The first small size symmetric peak with a maximum at ~300 ◦C is assigned to the reduction of NiO nanoparticles weakly interacting with alumina [31,35,36]. These particles are created in the precursor state during the thermal treatment of the sample under Ar. The addition of Cu up to 2% decreases the size of this peak, which suggests a decrease of the relatively bigger NiO nanoparticles and, thus, weakly interacting with alumina. This is in line with the increase in the nickel dispersion inferred by XPS. The very high peak at about 300 ◦C in the sample with the higher Cu loading (55Ni5CuAl) is attributed to the reduction of CuO to Cu0 [23,27–29]. There is no doubt that both the previously mentioned NiO and CuO reductions occur upon activation taking place at 400 ◦C. It is plausible assuming that the metallic nickel (copper) formed is entirely (partially) re-oxidized upon atmosphere exposure, which is in agreement with the XRD and XPS results. The second broad peak is assigned to various Ni-oxo species strongly interacting with the alumina and resembling the hardly reduced NiAl2O4 phase [31,35,36]. The addition of Cu is causing a slight shift of the maximum of the broad peak toward higher temperatures, which likely indicates the increase in the NiO/Ni0.95Cu0.5O dispersion deduced by XPS. Taking into account that (i) the low temperature extreme of the broad reduction band is located below 400 ◦C, (ii) the catalysts activation-reduction time (2.5 h), and (iii) the transient character of the H2-TPR method, we are arguing that the reduction of the NiO/Ni0.95Cu0.5O and CuO species present in the precursor catalysts are reduced in great extent upon activation. Then these are re-oxidized upon atmospheric exposure. This is in line with the literature. In fact, Miao et al. [30] reported that the oxidic precursors are reduced at 420 ◦C for 2 h into the corresponding metallic catalysts, comprised from Ni0, Cu0, and NiCu alloy-supported species. In conclusion, the NiO, CuO, and Ni0.95Cu0.5O species deduced by the Ex-situ characterization had been presumably transformed in a great extent into Ni0, Cu0, and a Ni-Cu alloy very rich in nickel catalysts upon activation. To further investigate this point, we are performing the following experiment for the 58Ni2CuAl catalyst, taken as an example. The precursor catalyst was activated in the TPR set–up at 400 ◦C for 2.5 h, the temperature was reduced at 100 ◦C under hydrogen flow, and the H2-TPR experiment was performed in situ for the activated catalyst (Figure 5). A comparison of the H2-TPR profiles of the precursor and the activated catalyst clearly shows that inception point of catalyst reduction was shifted from ~200 to ~450 ◦C after catalyst activation whereas the amount of the reduced species was considerably reduced. This corroborates our previous assumption that the NiO, CuO, and Ni0.95Cu0.5O species deduced by the Exsitu characterization had been transformed in a great extent into Ni0, Cu0, and a Ni-Cu alloy very rich in nickel upon activation. This is not the case for the like NiAl2O4 surface phase since the reduction above 700 ◦C is similar for the precursor and activated catalyst. This phase seems to be well dispersed and, thus, it is not detectable by XRD. In this point, one may wonder whether the Ni0, Cu0, and a Ni-Cu species re-oxidized upon exposure in air are actually present in the functional catalysts working

under a very reducible atmosphere (hydrogen pressure 40 bar). There is strong evidence that it is actually the case.

**Figure 5.** H2-TPR profiles of the 58Ni2CuAl precursor catalyst (**A**) and of the same catalyst after its in-situ activation at 400 ◦C for 2.5 h (**B**).

In fact, Totong et al. [37] have recorded the H2-TPR profiles after chemical activation (reduction by NaBH2) of nickel catalyst (10 wt. % Ni) supported on CeO2-ZrO2 and its exposure in air. The H2-TPR profile exhibited re-reduction peaks. Moreover, the formation of the Ni-Cu alloy has been reported several times in the literature [19–23,28–30]. The formation of the Cu–Ni alloy favors SDO and inhibits cracking of the C–C bonds, which leads to enhanced catalytic performance [23,28,29]. Extending the picture, we had outlined the local arrangement of nickel and copper oxide species into the reduced ones. We are visualizing the surface speciation in the functional catalysts. In the 59Ni1CuAl catalyst, the much larger amount of Ni0 nanoparticles covers partially the Ni-Cu alloy, whereas, in the 58Ni2CuAl catalyst, there is no considerable masking of this Ni-Cu alloy by the Ni0 nanoparticles. Lastly, in the 58Ni5CuAl catalyst, a portion of Cu<sup>0</sup> nanoparticles partially cover the Ni-Cu alloy ones.

The acidity of the catalysts has been studied by NH3-TPD. Figure 6 reveals that all catalysts exhibit mainly weak (desorption temperature range <300 ◦C) and relatively low moderate (desorption temperature range 300–450 ◦C) and very low strong (desorption temperature range >450 ◦C) acidity [17,38,39]. A small addition of Cu (59Ni1CuAl) increases total acidity. However, this trend is reversed as the Cu loading is further increased. It is interesting that the sample with the maximum copper content exhibits considerably low weak/moderate acidity and relatively high strong acidity.

**Figure 6.** NH3-TPD curves of the catalysts studied.

#### *3.2. Catalysts Evaluation*

A representative chromatogram taken after sampling the liquid phase of the reactor is depicted in Figure 7.

**Figure 7.** A typical gas chromatogram of a liquid sample received from the reactor upon transformation of SFO into green over the 58Ni2CuAl catalyst (reaction time = 4 h): C15. Pentadecane, C16. Hexadecane, C17. Heptadecane, C18. Octadecane, 1. Palmitic acid, 2. Stearic acid/ethyl stearate, 3. Propyl stearate, 4. Palmityl stearate, 5. Stearyl stearate, 6. 2(octadecyloxy) ethyl stearate, 7. Distearine. (Reaction conditions: 310 ◦C, 40 bar hydrogen pressure, hydrogen flow rate = 100 mL/min (STP), reactant volume to catalyst mass ratio = 100 mL/1 g).

Inspection of the chromatograms in conjunction with the study of GC-MS spectra showed the presence of the non–reacted SFO and normal alkanes in the diesel range (n-C17, n-C18, n-C15, and n-C16). Hydrocarbons with a smaller number of carbon atoms are not detected in the liquid phase of the reaction mixture. We have detected intermediate compounds such as acids (palmitic and stearic acid), propyl and ethyl esters (ethyl stearate and propyl stearate), bigger esters (palmityl stearate, stearyl stearate, 2(octadecyloxy) ethyl stearate), and saturated diglycerides (distearine). In the gas phase, we have detected propane, CO, CO2, and CH4. The same products were detected in the presence of the non-promoted and the promoted catalysts, which suggests no considerable influence of copper on the entire SDO reaction network described in the introduction.

The variation with time of the conversion of SFO and the composition of the liquid reaction mixture in hydrocarbons, acids, and esters, determined over the most active catalyst 58Ni2CuAl, is illustrated in Figure 8.

Inspection of this figure shows that the composition in total hydrocarbons and each hydrocarbon separately increase monotonically with time. In contrast, the composition in the intermediate total acids and esters increases initially with time and then it decreases. Similar kinetic curves were obtained for all the catalysts evaluated in the present work. This suggests no considerable influence of copper on the SDO mechanistic scheme described in the introduction for the nickel-based catalysts.

**Figure 8.** Kinetics of the SFO selective deoxygenation over the most active catalyst (58Ni2CuAl). (Reaction conditions: 310 ◦C, 40 bar hydrogen pressure, hydrogen flow rate = 100 mL/min (STP), reactant volume to catalyst mass ratio = 100 mL/1 g).

Figure 9 presents the conversion of SFO, and the composition of the liquid reaction mixture in total n-alkanes, acids, and esters, determined after 9 h of the reaction.

**Figure 9.** Evaluation parameters obtained for the SDO of SFO over the catalysts studied at reaction time equal to 9 h (Reaction conditions: 310 ◦C, 40 bar hydrogen pressure, hydrogen flow rate = 100 mL/min (STP), reactant volume to catalyst mass ratio = 100 mL/1 g).

A very high conversion is obtained in all cases. The most important intermediate products are esters and fatty acids with their compositions in the reaction mixture in the range of 13–18% and 13–20%, respectively. The most important finding is that copper considerably increases the composition of the liquid reaction mixture in total hydrocarbons in the diesel range, which is the most significant evaluation parameter from the practical point-of-view. The effect is more pronounced in the presence of 58Ni2CuAl catalyst, where an increase of 17.2% has been obtained with respect to the non-promoted 60NiAl catalyst. Therefore, the promoting action of copper is clearly emerged for the first time for catalysts with a very small Cu/Ni weight ratio (0.02–0.09) operating under solvent-free conditions and very high SFO volume-to-catalyst ratio (100 mL/1 g) and a reaction time of 9 h. These correspond to an LHSV value equal to 11.1 h−<sup>1</sup> for experiments taken place in fixed bed reactors. Therefore, the values of the composition of the liquid reaction mixture in total hydrocarbons obtained in the present experiments correspond to the mean value obtained for experiments taken place in a fixed bed reactor at LHSV value equal to 1 h−<sup>1</sup> and time on stream of 100 h. This is in conjunction with the almost linear increase of this evaluation parameter (Figure 8) with time providing a strong evidence for the relatively high stability of the catalysts tested.

At this point, it should be noted that the commercially produced green diesel initially contains saturated hydrocarbons with carbon atoms in the range of C15–C18, mainly C17 and C18, as in the present work. These linear hydrocarbons improve the ignition properties (cetane number) of fuel, but reduce its flow characteristics. Thus, an effort has been undertaken to improve the latter. Yeletsky et al. [40] have reviewed this issue, which presented recent advances. One may also notice that the yield of total hydrocarbons obtained over the most active catalyst studied was 71.5% (the rest being mainly high molecular weight esters and free fatty acids (Figure 9)). However, the hydrocarbons yield can be very easily increased to 100% by a slight decrease in the "volume of SFO to catalyst mass" ratio.

Figure 10 illustrates the composition of the liquid product in n-alkanes obtained over the catalysts studied at 9 h of the reaction.

**Figure 10.** Composition of the liquid product in n-alkanes with odd and even numbers of carbon atoms obtained over the catalysts studied at reaction time equal to 9 h (Reaction conditions: 310 ◦C, 40 bar hydrogen pressure, hydrogen flow rate = 100 mL/min (STP), reactant volume to catalyst mass ratio = 100 mL/1 g).

It is seen that the content of the reaction mixture in hydrocarbons with an odd number of carbon atoms (n-C15 and mainly n-C17) are considerably higher than that of the corresponding ones with even carbon atoms (n-C16 and n-C18). In fact, the values obtained for the ratios (n-C17/(n-C17+n-C18)) and (n-C15/(n-C15+n-C16)) are, respectively, centered in the range of 0.97–0.95 and 0.86–0.77, which indicates that SDO mainly proceeds via deH2O-deCO rather than through deH2O as expected for nickel catalysts [5–8,10,17,18,31]. The above indicate that the presence of small amounts of copper does not disturb the SDO mechanism.

It seems to us reasonable to relate the copper-promoting action to the increase of nickel dispersion brought about by copper, which is deduced by XPS. However, the monotonous increase of Ni dispersion with the Cu content (Table 2) is not compatible with the volcano-like trend of catalytic performance and its maximization over the catalyst with the medium Cu content (58Ni2CuAl). The inferred formation of a Ni-Cu alloy on the base of catalyst characterization should be another crucial factor [19–23,28–30]. The close location of Ni and Cu atoms inside the alloy allows the development of electronic interactions between these two elements, which may influence the intensity of adsorption of hydrogen and several key intermediates and, thus, the catalytic performance to hydrocarbons [30]. It is known that the intensity of adsorption for hydrogen, oxygen, carbon, and several organic molecules is lower on copper when compared to nickel [41] and this could explain the different catalytic behavior exhibited from monometallic nickel and Ni-Cu alloys in several catalytic reactions [42]. As already concluded on the surface of the 59Ni1CuAl and 55Ni5CuAl catalysts, a portion of the more active Ni-Cu alloy nanoparticles is covered by Ni<sup>0</sup> and Cu<sup>0</sup> nanoparticles, respectively. On the contrary, there is no considerable masking of this Ni-Cu alloy in the 58Ni2CuAl catalyst explaining its highest catalytic performance. The latter could be secondarily attributed to the mono-modal pore size distribution exhibited by this catalyst, centered at about 4.5 nm (Figure 1), which presumably facilitates the reactants/products' mass transfer. The relatively low performance of the 55Ni5CuAl catalyst with respect to the 59Ni1CuAl and 58Ni2CuAl catalysts could be partly attributed to its remarkably low weak/moderate acidity and the relatively high strong acidity. The former favors SDO whereas the second catalytic cracking results in coking [5].

A direct comparison of the present catalytic results with those of the literature ones is not an easy task. This is due to the high diversity of the set ups (fixed bed, batch, semi-batch), experimental conditions (temperatures: 250–400 ◦C, hydrogen pressures: 5–40 bar, equivalent LHSV values: 1–7 h<sup>−</sup>1), and feedstocks (biodiesel, methyl palmitate, ethyl caprate, tristearin, stearic acid, yellow grease, hemp seed oil, waste free fatty acids, brown grease, oleic acid, methyl laurate, and sunflower oil) used in the relevant works [19–30]. However, some useful conclusions could be drawn by comparing the present results with the literature results. First, it should be mentioned that, even though the non-promoted catalyst (60NiAl) is very active when compared to the current catalysts, a yield of 61.2% of total hydrocarbons was obtained over this catalyst working under solvent-free conditions and an equivalent LHSV value equal to 11.1 h−1. In spite of this high activity, the presence of a very small amount of copper is sufficient to considerably increase the previously mentioned yield. The maximum promoting action of copper in the present work is obtained in the 58Ni2CuAl catalyst with a Cu/Ni weight ratio equal to 0.034, which is much smaller to those obtained in the other relevant works [19–30]. We are attributing this difference to the preparation method followed in the present study (controlled co-precipitation). This method results initially in the formation of layered double hydroxide precipitates in which the Ni2<sup>+</sup>, Cu2<sup>+</sup>, and Al3<sup>+</sup> ions are atomically mixed in the basal layers of the double hydroxide [43–45]. This, in turn, facilitates the well mixing of very small alumina, NiO, CuO, and Ni0.95Cu0.5O particles formed in the subsequent thermal step (heating under argon) and, thus, the well mixing and high dispersion of the Ni0, Cu0, and Ni-Cu alloy rich in nickel is formed upon activation. The high-level synergy between the Ni<sup>0</sup> hydrogenation sites and Cu<sup>0</sup> sites inside the Ni-Cu alloy seems to be the main reason for the promoting action of copper.

As already mentioned, the presence of copper does not change practically in the SDO network followed in metallic nickel catalysts where the deH2O-deCO pathway predominates. This is in agreement with the literature. In fact, copper has a weak effect, if any, on SDO mechanism. The Ni-Cu bimetallic catalysts still prefer catalyzing the deH2O-deCO pathway [19,23]. A shift of the mechanism from the deH2O-deCO to the deH2O pathway takes place in copper-promoted catalysts above a critical Cu/Ni weight ratio much higher than those studied in the present work. For example, Miao et al. [30] have been recently reported where the SDO network changes from deH2O-deCO to the deH2O pathway at a Cu/Ni weight ratio higher than 2.3. The relatively high concentration of the intermediate esters determined in the present study (Figure 9) was compared to most of the

relevant works [19–30], which can be effortlessly related to the solvent-free conditions adopted in the present study. This facilitates the bimolecular esterification between the intermediates' fatty acids and fatty alcohols.

#### **4. Conclusions**

Copper doping considerably increases the catalytic performance of the nickel–alumina co-precipitated catalysts (60 wt. % Ni) for the transformation of SFO into green diesel whereas it does not affect the SDO reaction network. The effect is more pronounced in the catalyst containing 2 wt. % copper (58Ni2CuAl), where a 17.2% increase of green diesel content in the liquid product has been achieved with respect to the non-promoted catalyst, 60NiAl. The promoting action of copper is demonstrated for the first time for catalysts with a very small Cu/Ni weight ratio (0.02–0.09). This was attributed to the increase in the nickel dispersion caused by copper and to the formation of a Ni-Cu alloy very rich in nickel, which is considered more active than nickel. In the 59Ni1CuAl and 55Ni5CuAl catalysts, a portion of the Ni-Cu alloy is covered by Ni0 and Cu0 nanoparticles, respectively. On the contrary, there is no considerable masking of this Ni-Cu alloy in the 58Ni2CuAl catalyst, which explains its highest catalytic performance. The relatively low performance of the 55Ni5CuAl catalyst could be partly attributed to its remarkably low weak/moderate acidity and the relatively high strong acidity. The former favors SDO whereas the second catalytic cracking results in coking.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1996-1073/13/14/3707/s1. Figure S1: SDO reaction network. Figure S2: Presentation of the setup used for co-precipitation. Figure S3: SEM images. Figure S4: EDS spectrum. Figure S5: TEM image.

**Author Contributions:** Conceptualization, K.B., C.K. and A.L.; Methodology, M.G. and E.K.; Software, S.L.; Validation, E.K., S.L. and E.S.; Investigation, M.G., E.K. and E.S.; Data Curation, M.G., E.K. and E.S.; Writing-Original Draft Preparation, A.L.; Writing-Review & Editing, K.B.; Visualization, K.B., C.K. and A.L.; Supervision, K.B. and S.L. All authors have read and agreed to the published version of the manuscript.

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

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

#### **References**


© 2020 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/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Energies* Editorial Office E-mail: energies@mdpi.com www.mdpi.com/journal/energies

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18