Green diesel production using stearic and palmitic acids on Ni catalysts obtained from Ternary Hydrotalcites Ni-Mg–Al

Abstract

Ni-Mg–Al ternary hydrotalcites (HTs) with molar ratio M²⁺ (Ni²⁺  + Mg^{2 +})/Al = 3, and different content of Ni, Ni0Mg3, Ni2Mg1, Ni2Mg1, Ni3Mg0 were prepared. The HTs were calcined and then reduced to generate catalysts based on metallic Ni, which, were used to produce green diesel from a mixture of stearic and palmitic acids at 320 °C and 1400 psia of H₂. A reaction scheme was proposed; the main products were alkanes n-C17 and n-C15. The prepared catalysts carry out, mainly, reactions of decarboxylation (DCX) and decarbonylation (DCN), and some hydrocracking. The most active catalyst was Ni2Mg1 because it forms small clusters of nickel, apparently stabilized on MgO or NiO-MgO spinel like structures.

Graphical Abstract

1 Introduction

Concern about the environment and global warming are the driving forces behind the need to produce biofuels. Among the fuels for heavy transport uses, diesel has a preponderant role; biodiesel and green diesel are considered as good substitutes for this fuel. Manufacture of biodiesel has a certain degree of maturity; besides, its production requires low capital investment [1], thus there are already several industrial process applications. Biodiesel is synthesized from the transesterification of fatty acids with some alcohol [2], using acid or basic catalysts under non-severe conditions. Since biodiesel contains esters, the oxygen presence decreases its calorific value and limits their storage time; in addition, biodiesel is very sensitive to the presence of water [3]. In recent years, research emphasis has been oriented towards green diesel, also called eco-diesel, synthetic diesel, syndiesel or hydrogenated biodiesel. The production of green diesel occurs through a hydrodeoxygenation process (HDOP); during which HDOP the hydrodeoxygenation (HDO), decarboxylation (DCX), decarbonylation (DCN) and hydrocracking (HYC) reactions occur. The main products from the HDOP are n-alkanes [5]; green diesel is, due to its chemical composition, equal to diesel; and it can be used in motor engines without any modifications or, even more, can be used to improve the cetane number of fossil diesel for sale. Several types of catalysts have been proposed to perform the HDOP, among them sulfide, NiMo/Al₂O₃, CoMo/Al₂O₃, noble metals (Pt, Pd, Rh, Ru etc.), or another metal catalysts (for example, Ni). From the recent reviews on this topic, Arun et al. [6], Douvartzides et al. [7], and Chen et al. [8], some insight can be found: sulfided catalysts are the less expensive but they can suffer deactivation by sulfur leaching; therefore, it is necessary the addition of a sulfiding agent [9], thus products can be contaminated with sulfur. The noble metal catalysts are highly active; but expensive. Hence, in view of its low cost, non-noble reduced metal catalysts have had increased attention as an interesting alternative. In hydrogen presence, nickel-based catalysts have shown good HDO activity. Mortensen [10] reported the activity of Ni/SiO₂ catalysts on the HDO of phenol. Onyestyák et al. [11] studied the HDO of sunflower oil using non pyrolytic Ni-Raney, Ni/Al₂O₃ and NiMo/Al₂O₃ sulfided catalysts. Lui et al. [12] used Ni in metal organic frameworks and achieved up to 98.8% triglycerides conversion. In a recent review Hongloi et al. [13], reported that for Ni catalysts some of the studied supports are Al₂O₃, zeolites, zirconia, activated carbon. Most of these catalysts have acidic supports and coke formation is a concern. In principle, deactivation by coke is a minor effect on basic supports; but there is few information related to catalysts on this kind of supports. Hydrotalcites (HTs) can be considered as precursors of basic supports. In the calcined form they have been studied as catalytic precursors or supports for methanation of C0₂ [14]: hydrogenation and hydrogenolysis [15], hydrodesulfurization HDS [16], dry reforming: [17]. HTs were also used as precursors of catalysts in the HDOP: for example, Shailesh et al. [18] studied Pd and Ru supported on a Mg–Al hydrotalcites; the authors reported conversions up to 95% of Jatropha oil. Abidin et al. [19] studied bimetallic HTs (Ni, Mg, Zn or Fe and Al), and found a conversion near 100% and a selectivity of ~ 89% for green diesel at 300 °C, with Ni-HTs.

The chemical formula of the HTs like materials is [M²⁺_1-xM³⁺_x(OH)₂]^x+ (Aⁿ⁻¹)_x/n*mH₂O where M²⁺ and M³⁺ can be any metal cation with charge 2 + or 3 + , and Aⁿ⁻ is the charge compensating anion (CO₃²⁻, NO₃, etc.) [20]. Ternary HTs can be synthetized using different M²⁺ cations if the cation ratio (M₁²⁺  + M₂²⁺)/M³⁺ is maintained. Then, it is feasible to achieve a successive substitution of Mg²⁺ (M₁²⁺) by any other (M₂²⁺) cation; as a result, it is possible to control the incorporation of variable amounts of a reducible metal (Ni in our case) to the HTs structure, presumably with a high dispersion of Ni because a homogeneous distribution of M1^II and M2^II cations is expected. In summary, the flexibility in the material preparation can potentially provide a way to reach both high Ni loadings as well as high Ni dispersions, thus the use of ternary Ni-Mg–Al HTs as precursors of catalysts are of interest. It should be noted that to obtain the metallic phase from HTs, it is required to calcine and then to reduce them.

Ethical issues have been raised about using edible biomass as a feedstock for producing a bio-substitute for fossil diesel. As a result, there is growing trend to use non-edible oils as feedstock for manufacturing biodiesel or green diesel. Several oils that meet this criterium include jathopa (Jatropha curcas) [21], karanja (Pungamia pinnata) [22], and castor oil (risinus comunis) [23]. Castor oil has valuable applications in the pharmaceutical and chemical industries [24]. Consequently, its industrial production is well established. This makes it an attractive option as a feedstock for producing biodiesel or green diesel. However, the are some challenges associated with the use of castor oil to produce biodiesel. Castor oil is composed of ricinoleic (RA), oleic and linoleic fatty acids [25]. The methyl esters of these fatty acids have low cetane numbers and high viscosity. High viscosity results in large droplets in the fuel injector, leading to incomplete combustion and smoke emissions. Since castor oil has a high content in RA (up to 90%) and linoleic (3–6%) fatty acids [26], castor oil is not suitable as a feedstock for industrial biodiesel production [27]. These issues can be avoided if castor oil is used to produce green diesel instead.

Stearic acid (SA) is an intermediate product of the HDO reaction of ricinoleic acid (RA); in this reaction RA loses a hydroxyl group, and its double bond is easily hydrogenated; thus, producing the stearic acid (SA). The advantages of using SA as model feed molecule are evident in the literature: Snáre et al. [28] employed stearic acid to compare the HDO activity of different single noble metal catalysts (Pt, Pd, Ru orOs) over supports as carbon or Al₂O₃, they also tested the performance of NiMo/Al₂O₃ and Ni on various supports as C, Al₂O₃, and Cr₂O₃. Lestari et al. [29, 30] used SA to investigate the total conversion and selectivity between hydrodeoxygenation (HDO) and decarboxylation (DCX) reactions for Pd/C and Pd/SBA catalysts. Hachemi et al. [31] tested Ni supported on SiO₂, Al₂O₃, and HY zeolite and reported yield and selectivity values for HDO/DCX. Arora et al. [32] used SA to determine the reaction scheme and rate constants for a sulfided NiMo catalysts, and Kumar et al. [33] reported the reaction scheme for SA and evaluated the kinetic coefficients for Ni supported on Al₂O₃, SiO₂, and HZSM5 catalysts.

To our knowledge there is no systematic study on the use of ternary Ni-Mg–Al hydrotalcites as catalyst precursors to produce green diesel through the HDO reaction. Additionally, there is no information on the possible changes on the reaction scheme for basic catalysts nor any data on rate constants.

In this work, highly active catalysts obtained from ternary Ni(x)-Mg(y)-Al hydrotalcites with Ni different content were synthesized, the catalysts were used in the HDO reaction of a mixture of stearic and palmitic fatty acids to investigate the effect of increasing the Ni content on the HDO activity and selectivity and to identify what Ni to Mg ratio generates the most active catalyst. A reaction scheme was proposed and the rate constants for the HDO, DCX, DCN and HYC reactions were evaluated. The catalysts precursors were characterized by XRD to confirm that the different hydrotalcite materials were properly synthesized. XRD was also used to identify the crystallographic phases obtained from the calcined hydrotalcites and catalysts. TPR was used to investigate the effect of the Ni content on the reducibility of the calcined hydrotalcites. HREM was used to determine the morphology of metallic Ni particles and supports. Finally, we explored the existence of a correlation between the quantity of well dispersed Ni particles and activity.

2 Experimental

2.1 Hydrotalcites synthesis

The HTs were obtained using the coprecipitation method at low supersaturation. A solution containing the required amounts of the cations (Ni²⁺, Mg²⁺ and Al³⁺) was prepared in deionized water using reagent grade salts: Mg(NO₃)₂*6H₂O (Sigma-Aldrich, 99%), Ni(NO₃)₂*6H₂O (Sigma-Aldrich 100%) and Al₂(NO₃)₃*9H₂O (J. T. Baker, 99.7%); the total cation (Ni²⁺, Mg²⁺ and Al³⁺) concentration was kept constant at 0.64 M. An additional solution was prepared using 1.25 M NaOH (Reactivos y Productos Químicos Finos, 97%) and 0.5 M (NH₄)₂CO₃ (Mallickrodt, 99.9%). Both solutions were added dropwise into a reactor containing 100 mL of deionized water at 60–70 °C while stirring vigorously, keeping the pH constant at 10.5. The product was aged for 24 h, then filtered and washed with deionized water, to remove any unreacted ions. Finally, the solid was dried overnight at 120 °C and stored until use.

The synthesized materials were named NixMgy, where "x" and "y" represent the relative moles of Ni or Mg cations, respectively, to maintain the molar ratio [Ni (x) + Mg (y)]/Al = 3. Following the above procedure, four HTs (Ni0Mg3, Ni1Mg2, Ni2Mg1 and Ni3Mg0) were prepared.

2.2 Characterization of fatty acids

Technical grade “stearic acid” (Drogueria Cosmopolita) was used as a reactant in the activity experiments. This reactant was characterized by acid transesterification with methanol, followed by identification of the methyl esters by mass spectroscopy. The transesterification reaction was performed at 60 °C using a 0.1 M solution of H₂SO₄ (Meyer, A.R.) for 4 h. The characterization results showed that the “stearic acid” was a mixture of stearic (48.9 mol%) and palmitic (45.3 mol%) fatty acids (see Table 1 in the results section).

Table 1 Identified chemical compounds by the “Stearic acid” characterization

2.3 Characterizations of catalysts and precursors

2.3.1 Textural properties and X-ray diffraction (XRD)

The hydrotalcites and their calcination products were characterized for their textural properties by nitrogen physisorption using a Micromeritics TriStar for specific area and pore size distribution. For the specific area, the BET method was used, and the desorption isotherm and the BJH method were used to obtain the pore size distribution.

The crystalline structure of the samples was analyzed using the XRD patterns obtained in a 2θ range from 5° to 75° with a scan speed of 2°/min. The patterns were collected on a SIEMENS-D5000 diffractometer with CuKα (1.54056 Å) radiation.

2.3.2 Temperature programmed reduction (TPR)

The reduction behavior of the calcined hydrotalcites was analyzed using with an ISRI 1000 apparatus equipped with a thermal conductivity detector. The samples were first calcined at 450 °C and then reduced in a 30 mL/min flow of H₂/Ar (29.7 v/v, Praxair certified gas mixture) using a heating rate of 10 °C/min from 20 to 1000 °C. An estimation of content (mol%) of each reducible species was obtained by deconvolution of the trace signal considering a Gaussian-type functions for the resulting peak areas of each specie. Finally, the signal was converted to moles of consumed H₂ using the reduction of Ag₂O (JT Baker A.R.) as standard.

2.3.3 Scanning electron microscopy (SEM–EDX) and transmission electron microscopy (TEM)

SEM–EDX was used to determine the elemental composition of the calcined materials. The analysis was performed using a JEOL JSM-5900LV scanning electron microscope equipped with an Oxford-Isis Energy Dispersive X-ray analyzer.

TEM was used to determine the presence and size of the metallic nickel particles on the catalyst surface. Analysis was performed using a JEOL 2010 microscope operating at 200 keV.

2.4 Catalytic activity

The Ni0Mg3, Ni1Mg2, Ni2Mg1 and Ni3Mg0 hydrotalcites were calcined for 4 h at 400 °C in air (Infra, dry grade) flow and then reduced for 4 h in H₂ (Infra 99.99) flow at 490° C to obtain the catalysts. The used feedstock was a 2% w/w solution of technical grade “stearic acid”, which actually is a mixture of stearic (48.9 mol%) and palmitic (45.3 mol%) fatty acids (see Table 1 in the results section), in dodecane (JT Baker, R.G.). The reaction was carried out in a 300 mL Parr 4561 M stainless steel reactor using ~ 0.35 g of the reduced catalyst. The reactor was pressurized with H₂. Reaction temperatures was 300 °C and the reaction pressure was 1400 psia. Liquid samples were taken at different times and analyzed on an Agilent Technologies 7890 gas chromatograph with a flame ionization detector and an Ultra 1 column (Agilent). Product conversions were calculated using the following formula:

$$Xi=\frac{{C}_{product\;i}}{{C}_{0\;reagents}}$$

where C_product = Concentration of product i, mol/L.

C_{0 reagents} = Initial concentration of stearic + palmitic fatty acids, mol/L.

3 Results and discussion

3.1 Stearic acid characterization

The analysis of the technical grade “stearic acid” used in the reaction experiments is shown in Table 1. Stearic (C18) and palmitic (C16) acids comprise 94.25% of the feedstock.

3.2 X-ray diffraction for synthesized hydrotalcites

The XRD patterns of the synthesized HT materials are outlined in Fig. 1a. Characteristic sharp and intense peaks are observed at 2Θ = 6.0, 12.1, 38.4, 61 and 62°, which correspond to the (003), (006), (009), (110) and (113) crystallographic planes of a well crystalized HT (JCPDS 22–700). In addition, broad signals for the (015) and (018) planes located at 2q = 39° and 2q = 43° are observed. According to Shishido et al. [34] this suggests that the Ni²⁺ ions are incorporated in the Mg²⁺ sites and can be dispersed uniformly in the brucite layer of the HT structure after calcination.

Fig. 1

XRD patterns for: (a) synthetized hydrotalcites and, (b) calcined hydrotalcite products

3.3 XRD for Calcined Hydrotalcites

The XRD patterns of NixMgy-c at 450 °C are presented in Fig. 1b. The four samples display five peaks located at 2 Q =  ~ 22°, 34.9°, 37°, 43.4° and 63°.

For the Ni0Mg3-C sample, the disappearance of the corresponding hydrotalcite-like structure peaks is noticed; this phenomenon is due to collapse of the layered structure, i.e., dehydroxylation of the brucite-like layers and the removal of the interlayer anions (CO₃²⁻). According to Morkthar et al., [35] heating the Mg–Al hydrotalcite at temperatures between 350 and 450 °C, causes the crystallization of MgO in periclase structure (JCPDS, 45–0946 card, characterized by intense peaks at 43° (200) and 62.5° (220) and a small peak at 37° (111)). Thus, the Ni0Mg3 oxidic precursor has MgO in periclase structure. For Ni–Al HT like materials, as our Ni3Mg0-c sample, Clause et al. [36] reported that upon heating at temperatures in the interval of 50 to 800 °C, the only phase detected by XRD is NiO, this oxide crystallizes, also, with the periclase structure (JCPDS 47–1049, displaying a high intensity peak near 2Q ~ 37°); but the obtained cell parameters are lower than those corresponding to NiO, suggesting the existence of Al(III) ions inside the oxide phases. It should also be noted that g-Al₂O₃ has signals at very similar 2Q values. Then the bimetallic Ni0Mg3 and Ni3Mg0 HTs-c have a periclase like crystallographic structure with Al inside de respective oxides.

The XRD signals for our ternary HTs samples, Ni1Mg2 and Ni2Mg1, are similar to those obtained for the binary calcined hydrotalcites (Ni0Mg3 and Ni3Mg0) oxides, this result is in accord with Manikandan et al. [37] suggesting that in the calcined samples relatively homogeneous periclase-like material exists. However, it is difficult to distinguish using XRD between the possibility of two separated oxides or a solid solution of oxides or even the formation of spinels.

3.4 Scanning electron microscopy-energy dispersive X-Ray (SEM EDX)

The results of the SEM EDX technique applied to calcined hydrotalcites are shown in Table 2. The results show that, after calcination, the ratio between the moles of each cation remains at the proportions of the original hydrotalcite.

Table 2 Analysis SEM–EDX by element for calcined samples

3.5 Catalytic activity

The results of the activity test for the Ni2Mg1 catalyst at 320 °C are reported in Fig. 2. N-alkanes with carbon chains in the C4-C17 range were the main group of products. Some intermediate products as alcohols and aldehydes were found at the detection threshold. The analysis of the produced gas showed the presence of methane, ethane, propane, butane, and pentane. However, the mass balance shows that the total moles of detected n-alkanes were slightly higher than the moles of fatty acids in the feed. This means that cracking occurs, possibly including the solvent (n-dodecane); therefore, compounds with less than twelve carbons were not considered in the analysis here. Practically, there was no detection of n-C18 in the products implying that stearic acid (C18) did not react through the HDO reaction. Likewise, n-C16 that would come from palmitic acid was not significantly detected. Thus, with the catalysts used here, the HDO reaction per se was not carried out and practically all the n-alkanes were obtained from the decarboxylation (DCX) and decarbonylation (DCN) reactions. From our experiments, it is not possible to distinguish between the DCX or DCN reaction paths because the DCX reaction produces terminal n-alkenes, but alkenes are very reactive and in the presence of H₂ and a metallic phase, such as Ni⁰ can produce n-alkanes.

Fig. 2

Concentration of reagents and products versus time in the HDO process for the Ni2Mg1 catalysts, P₀ = 1400 psi, T = 320 °C

From the data in Fig. 2 and considering the global concentration of fatty acids (stearic + palmitic), the total conversion (X_TOT) can be assessed. Figure 3 shows the plot of X_TOT versus time for the different catalysts. The Ni2Mg1 catalyst reaches an almost complete conversion in 80 min.

Fig. 3

Yield (Ci/C0) versus time for the synthesized catalysts

3.6 Reaction scheme

Yield was defined as:

$$Yield=\frac{Concentration\;of\;i\;product}{Initial\;concentration\;of\; key\;reactant}=\frac{Concentration\;of\;i\;product}{oil}$$

A graph of the yield to various hydrocarbons versus global conversion (X_TOT) (stearic + palmitic fatty acids) for the Ni2Mg1 catalyst (reduced hydrotalcite) is shown in Fig. 4. The observed behavior corroborates that n-C17 and n-C15 are the main products and therefore, for the catalysts and the operating conditions used here, the reaction path does not involve the HDO reaction. In accordance with our results, Onyestyák et al. [11] reported that for monometallic Ni-Raney and Ni/Al₂O₃ catalysts the preferred reaction route was the DCX + DCN over the HDO.

Fig. 4

Yield versus total conversion for Ni2Mg1 catalyst, P₀ = 1400 psia, T = 320 °C

At high conversion, the yield of n-C17 and n-C15 showed an abrupt decrease, while simultaneously n-C16 and n-C14 yields increased; also, methane was detected in the gas products. This means that n-C16 and n-C14 come from the hydrocracking of n-C17 and n-C15, respectively. It is well documented by many authors that hydrogenolysis of n-alkanes on Ni-containing catalysts occurs mainly at the terminal carbon–carbon bonds by successive scissions [33, 38,39,40,41,42]. Moreover, using a non-pyrolytic Ni-Raney catalyst, Onyestyák et al. [11] reported the production of n-alkanes and methane in the HDO of sunflower oil, which means that the cracking reaction is a property of the reduced metal since the authors worked with unsupported Ni-Raney.

From the interpretation of Figs. 2 and 4, the proposed reaction scheme is presented in Fig. 5.

Fig. 5

Proposed reaction scheme for the mix of stearic and palmitic fatty acids

From the reaction scheme of Fig. 5, the following set of first order ordinary differential equations was used to estimate the rate constants in the reaction scheme, considering that H₂ was in excess (i.e. pseudo first order for the organic reactants), the statistical analyses of the residuals are presented in Appendix 1.

$$-\frac{d{C}_{1}}{dt}={k}_{1}{C}_{1}$$

(1)

$$\frac{d{C}_{2}}{dt}={k}_{1}{C}_{1}-{k}_{3}{C}_{2}$$

(2)

$$\frac{d{C}_{3}}{dt}={k}_{2}{C}_{3}$$

(3)

$$\frac{d{C}_{4}}{dt}={k}_{3}{C}_{2}-{k}_{4}{C}_{4}$$

(4)

$$\frac{d{C}_{5}}{dt}={k}_{4}{C}_{4}+{k}_{2}{C}_{3}-{k}_{5}{C}_{5}$$

(5)

$$\frac{d{C}_{6}}{dt}={k}_{5}{C}_{5}-{k}_{6}{C}_{6}$$

(6)

$$\frac{d{C}_{7}}{dt}={k}_{6}{C}_{6}-{k}_{7}{C}_{7}$$

(7)

$$\frac{d{C}_{8}}{dt}={k}_{7}{C}_{7}-{k}_{8}{C}_{0}$$

(8)

All the fitted rate constants (ks, L/(min*g_cat)) are shown in appendix 1. In Fig. 6, the. rate constants for the reaction DCN + DCX of stearic (k1) and palmitic (k2 are shown). The DCN + DCX reactions are characterized by the rate constant k1 which produces n-C₁₇ and k2 which produces n-C₁₅. On the other hand, HYC products result from the demethylation reaction; for example: n-C17 yields n-C16 (k3). So, it is possible to evaluate the selectivity between the DCN + DCX reactions versus HYC. These results are shown in Fig. 7. To understand how the reaction system is performing, catalysts characterizations are necessary.

Fig. 6

Rate constants for the reaction DCN + DCX of stearic (k1) and palmitic (k2) fatty acids (T = 320 °C, P₀ = 1400 psia)

Fig. 7

Selectivity [decarboxylation (DCX) + decarbonylation (DCN)]/Hydrocracking (HYC) reactions for stearic (C17/C16) and palmitic (C15/C14) fatty acids

3.7 Temperature programed reduction (TPR)

The TPR results for oxide samples after calcination at 450 °C are shown in Fig. 8. The Ni0Mg3-c oxidic sample, as expected, does not have noticeable reducible species. For the Ni3Mg0 catalyst, three peaks at the temperature ranges of 280 to 400 °C, 400–470 °C and 470–550 °C are observed. There is, also, a shoulder in T > 550 °C.

Fig. 8

TPR profiles for the NixMgy-c samples, after calcination in air flow at 400 °C

In the TPR technique, in general, higher reduction temperatures indicate stronger interactions between reducible species and support. Parmaliana et al. [43], studied the reduction of NiO supported on MgO, and reported peaks comparable to those observed in our nickel-containing samples. According to these authors the first peak in the temperatures from 300 °C to 370 °C (573–640 K) is assigned to NiO crystals with weak interaction with the support. In fact, the authors reported that unsupported NiO is reduced at 370 °C. The peak which maximum is located at ~ 570 °C (870 K) was assigned to Ni²⁺ ions having square-pyramidal coordination in the outmost layer of MgO structures. Another interpretation was given by Jafarbegloo et al. [44] who proposed that the reduction peak at 400–550 °C, corresponds to NiO diffusion into MgO, that occurred during calcination, these TPR peaks are assigned to the reduction of Ni²⁺ ions in the outermost layer and sub-surface layers of the MgO lattice. High temperatures of reduction (T > 750 °C) lead to Ni in subsurface layers of the MgO or spinel like structures [45], with strong interaction with the support.

In our samples in the temperature range of 400–500 °C there is other defined peak; According to Parmeliana [43], this could be caused by NiO well dispersed on NiO/MgO or NiO on the mixed oxide (MgO-NiO), However, we cannot lose sight of the Ni3Mg0 sample because it does not have magnesium and in this sample a peak between 400—475 °C is also observed, very similar to the ones existing in the ternary samples. Therefore, we cannot rule out the presence of an aluminum compound. For example, Ni ions also could diffuse on alumina. Rebours et al. [46] using ²⁷Al-MAS-NMR, and EXAFS, signaled that the thermal decomposition of Ni–Al, and Mg–Al hydrotalcites-does not lead to single phase products and proposed a three-phase model which involve the formation of NiO or MgO crystallites, MgO doped with, Ni or M^IIO Ni- Mg-doped alumina, and an aluminate spinel-type phase at the M^IIO-alumina interface.

With successive additions of Ni (Ni1Mg2, N2Mg1 and Ni3Mg0), there is a decrease in the temperature of the maximum for the first and second reduction peaks, (assigned to bulk NiO T_R 280–400 °C); and well dispersed NiO on the support (T_R 400–500 °C). However, the onset temperature does not move; so, it seems that there is an autocatalytic effect caused with the increment of nickel content, i.e., once the reduction of Ni begins it catalyzes the reduction of the different Ni species in the samples. This seems to be also happening for the high temperature reducible species. Guo et al. [45] suggested that if Ni is in excess respect to the M²⁺Al₂O₄ spinel stoichiometric ratio, upon heating, this Ni is expelled from the structure and consequently reduced at lower temperatures than Ni ions in the spinel structure.

In summary, the TPR and XRD results show that four reducible nickel species exist in the catalysts: (a) Peak I, T_R < 400 °C, associated to the reduction of Ni species with weak interaction with the support, possibly bulk NiO, the appearance of this peak at low nickel content suggests that at least part of the Ni is not in a mixed oxidic phases. (b) Peak II, which appears from 400 < T_R < 500 °C can be related to well dispersed NiO species on NiO-MgO or Al₂O₃ structures (c) Peak III, at 475 < T_R < 590 °C well dispersed Ni species having square-pyramidal geometry on a periclase-like (MgO) [43], or Ni²⁺ ions in the outermost layer and sub-surface layers of the MgO or possibly Al₂O₃ lattices. (d) Peak IV, at T_R > 600–650 Ni species in spinel like structures.

3.8 Transmission electron microscopy (TEM)

TEM was used to get information about the morphology of the metal cumulus on reduced catalysts. In the case of the sample without Ni (Ni0Mg3), some particles with flake like forms are observed (see Fig. 9), these crystallographic structures were reported before [47]; according to our XRD results, this material is composed by MgO periclase-like structures and possibly MgAl₂O₃ spinel with low crystallinity. For the sample with low Ni content (Ni1Mg2), the flake-like structures are not observed, and a small number of metallic nickel clusters appear. With more nickel in the samples, Ni2Mg1 and Ni3Mg0, numerous nickel clusters with sizes from 5 to 15 nm are observed. Using Selected Area Electron Diffraction (SAED), the Ni2Mg1 shows an increase in the crystallinity of the material. However, the sample with the highest nickel content, Ni3Mg0 showed a lost in crystallinity. Using the SAED technique, one could obtain the diffraction pattern of the sample. Then, using the software CrysTBox ringGUI [48], the reciprocal distances in the rings (1/d) and, comparing with data from [49]; it is possible to identify the crystallographic phases. for the Ni2Mg1 sample several rings associated with the crystallographic planes (111), (222), (004) were identified. The crystallographic planes are associated to metallic nickel in FCC (F3-m3) structure. In addition, other low intensity rings related to other crystallographic planes associated to other Ni and even NiO crystallographic structures were also identified.

Fig. 9

TEM micrographs for the reduced catalysts and their SAED. For the Ni2Mg1 sample also the identified diffraction planes by the software CrysTBox ringGUI [47]

The activity of the catalysts reaches a maximum for the Ni2Mg1 catalyst (see Fig. 3). For the Ni0Mg3 (no nickel) sample, which showed absence of any phases susceptible to reduction, as expected, no activity was observed; hence, the activity is related to reducible nickel species (Ni-R). The Ni-R at high reduction temperatures (> 700 °C, peak IV) do not participate in the catalytic activity, considering that before the reactivity test all the catalysts were reduced at 490 °C. Another phenomenon is that the total amount of Ni in the sample is not directly related with the active Ni on the catalyst, because the sample with the higher nickel content, Ni3Mg0, is not the most active. On one hand, the TPR analysis of this sample shows a decrease in Ni R. On the other hand, SEM EDX results show that the molar ratio M^II/Al in the samples is constant; so, there is not Ni lost in the Ni3Mg0 sample during the synthesis of the material. Instead, the Ni-R species decrement is probably due to the diffusion of Ni deep into the spinel-like structures. The decrease extends to Ni-R of peaks III, indicating that Mg has some role not in the activity but in Ni reducibility and dispersion of the Ni species.

Listed in increasing reduction temperature order, the Ni-R that can be considered as source of the active metallic Ni are: (a) peak I, NiO bulk species, (b) peak II, well dispersed NiO on MgO-NiO or Al₂O₃ and, (c) peak III well dispersed Ni species having square-pyramidal geometry or Ni²⁺ ions in the outermost layer and sub-surface layers of the MgO or Al₂O₃ lattices, species with medium interaction with the support. An attempt was made to find out a possible correlation between the H₂ consumption of these three reducible species and the HDOP rate constants. The best correlation was found for the hydrogen consumption associated with peak III in the TPR. This is shown in Fig. 10.

Fig. 10

Kinetic coefficients vs. peak III (400 < Tr < 500 °C) hydrogen consumption by TPR. Reaction conditions 320 °C, P0 = 1400 psias

For the Ni2Mg1 sample, along with enough Ni-R there is a formation of Ni in square-pyramidal coordination on MgO or Ni²⁺ ions in the outermost layer and sub-surface layers of the MgO or Al₂O₃ lattices. Upon reduction these ions give rise to small Ni metallic particles and due to its medium force interaction with the support, formation of large Ni metallic particles is inhibited. Guo et al. [45] worked with Ni/MgAl₂O₄, Ni/MgO and Ni/Al₂O₃ catalysts for the methane reforming. These authors reported that Ni on MgAl₂O₄ has high activity because of a low acidity and sintering-resistance compared to NiO on γ-Al₂O_3, they, also, suggested that MgAl₂O₄ can stabilize tiny Ni crystallites.

4 Conclusions

With the catalysts used here, the reactions that take place in HDOP are decarboxylation (DCX) and decarbonylation (DCN). Also, there are some hydrocracking products.

In the calcined hydrotalcites up to four different reducible species of Ni exist: (1) Peak I, bulk NiO, (2) Peak II, well dispersed NiO on MgO or NiO-MgO or Al₂O₃ mixed oxides, (3) Peak III, dispersed Ni species having square-pyramidal coordination on a periclase-like (MgO) or Ni²⁺ ions in the outermost layer and sub-surface layers of the MgO or possibly Al₂O₃ lattices. (4) Peak IV Ni in spinel like crystallographic structures.

The best activity was found for the Ni2Mg1 catalyst. The catalytic activity correlates well with peak III in the TPR, because the medium force interaction between the Ni metallic particles with the support leads to the formation of small particles and is good enough to inhibit the sintering of the Ni during activation.

Appendix 1

Table 3

To determine the optimal set of the reaction rate coefficients for the reaction scheme and differential equations shown in Fig. 6, Matlab™ software was used, the fitting of the model was performed with the ode45 function to integrate the differential equations and lsqnonlin function, to minimize the non-linear squares and get an estimate of the set of kinetic coefficients. The kinetic fit relies on minimizing the squared difference between calculated results by the model and experimental data.

Table 3 Constant rates for the HDOP of a mixture of palmitic and stearic acid (T = 320 °C, P₀ = 1400 psia)

Figure 11 and Fig. 12

Fig. 11

Parity graphic between production of each product versus predicted production

Fig. 12

Normal probability plots for the residuals of the fitting of a)Ni0Mg3, b)Ni0Mg3, c)Ni0Mg3 and, d)Ni0Mg3

A residual analysis was also made and found a reasonable distribution of the errors occurs (Fig. 11). Additionally, we obtained a parity plot comparing the experimental concentrations versus the predicted concentrations for the reactives and products can be obtained (Fig. 12). The graphics shows that the goodness of the correlation is adequate. The point associated with higher concentration in the three graphics are the initial concentration of fatty acids. Finally, the normal probability plots for the residuals of the fitting which suggest a normal distribution of the errors.