2.1 Catalyst preparation

The γ-alumina support (labeled as Al₂O₃-SG) was prepared by a sol-gel process following a published procedure [13], that we have modified concerning the hydrolysis and peptization steps. In our case, 1 mol of aluminum tri-sec-butoxide was hydrolyzed with 100 mol of water at 60 °C under stirring for 1 hour, followed by a peptization step with 0.1 mol of HCl (12 mol/L aqueous solution). The temperature of the solution was then increased to 80 °C while stirring during 1 hour until gel formation. The resulting gel was dried in an oven at 120 °C for 12 hours. The white powder was ground and calcined in air flow (10 L/h/g) during 5 hours at 500 °C (heating rate of 1 °C/min), so as to yield the transitional alumina, Al₂O₃-SG.

A reference γ-alumina sample supplied by Rhodia (170 m²/g) was also used for comparison as catalyst support (labeled commercial γ-Al₂O₃). The magnesium precursor was then inserted at room temperature into the alumina supports through wet impregnation. An alcoholic magnesium acetate solution (50 mL EtOH) was used to prepare the Mg-loaded alumina samples. The Al₂O₃ support (5 g) was mixed with the Mg acetate solution (0.14 mol/L) and the resulting mixture was stirred for 2 hours. The solvent was slowly evaporated at 50 °C, then the alumina impregnated solids were dried in air for 12 hours at 120 °C, before being activated up to 550 °C under flowing air for 8 hours (heating rate of 1 °C/min), so as to completely decompose the acetate anions and generate the corresponding supported MgO phase. The amount of Mg precursor was adjusted to obtain a final Mg content of 8 wt % after calcination. The resulting alumina-supported MgO samples were labeled as MgO-Al₂O₃-SG and MgO-γAl₂O₃.

Commercial lanthanum oxide (5 m²/g) supplied by Rhodia was also used for comparison as basic catalyst.

2.2 Characterization

Powder XRD patterns of the calcined alumina supported MgO materials were recorded on a Bruker D5005 diffractometer with monochromatized Cu Kα radiation (λ = 1.5418 Å) at 40 kV, 30 mA. The diffraction patterns were recorded in the 2θ values range 20–80° with a rate of 0.02°/sec and a step size of 2 seconds. Bulk chemical analysis for Mg and Al was achieved by ICP. The surface area and pore size analysis of the samples were carried out by adsorption-desorption of nitrogen on a Micromeritics ASAP 2010 instrument (−196 °C). Prior to N₂ adsorption, the samples were degassed under vacuum at 90 °C for 1 hour, followed by a further heating at 300 °C for 6 hours. The BET equation was used to calculate the surface area from adsorption data obtained at p/p₀ between 0.05 and 0.2. The pore diameter was evaluated at the maximum of the pore size distribution determined by the BJH method (adsorption).

TG-DTA study of the decomposition sequence in flowing air of the alumina supported magnesium acetate was performed on a SDT 2926 microbalance (TA instruments) from 20 to 800 °C at a heating rate of 10 °C/min.

TG-MS measurements were carried out with a TGA apparatus SDT Q600 from TA Instruments coupled with a mass spectrometer Prisma QMS 200 (Balzers). Known amounts of samples were disposed in an alumina crucible and heated from 25 to 750 °C with a 10 °C/min temperature slope under a N₂ flow rate of 100 mL/min. Mass spectroscopy measurements were performed at m/z 18 (water) and m/z 44 (CO₂).

IR spectra were recorded on a Nicolet Nexus spectrometer with a spectral resolution of 4/cm. Samples were pressed as self-supported wafers. Discs of 10 mg/cm² were prepared and treated directly in the cell. The samples were evacuated up to 240 °C (10 °C/min) for 30 minutes and then brought back to room temperature prior to contact of probe molecules. Pyridine and CO were introduced on the activated samples at room temperature.

2.3 Catalytic experiments

The transesterification of β-sitosterol with methyl dodecanoate was carried out at atmospheric pressure under a nitrogen flow in a Pyrex batch reactor equipped with a mechanical stirrer. The nitrogen flow was used instead of air flow in order to favor the elimination of the co-product (methanol) generated during the transesterification reaction, as well as to avoid parallel oxidation reactions. In accordance with previous data [11], the N₂ flow was fixed to 5 L/h. Under such conditions the yield to phytosterol esters reached a maximum, while no removal of methyl dodecanoate was observed during all experiments.

First, 3.5 × 10⁻² mol of sterol and 3.5 × 10⁻² mol of methyl dodecanoate were heated to 240 °C. Then the catalyst (5 mg mol%) was added to the mixture. The analysis of the reaction products was performed on a GPC equipped with a FID detector and an on-column injector. The percentage of each compound was determined using hexadecane as an internal standard. The conversions of methyl dodecanoate and β-sitosterol were followed as functions of the time on stream of the microsamples withdrawn periodically and evaluated with respect to the initial and final contents of fatty methyl ester and free sterol in the solution.

3.1 Catalysts characterization

The wide-angle X-ray diffractograms of the calcined sol-gel and commercial alumina support samples (Al₂O₃-SG and commercial Al₂O₃) showed the characteristic diffraction lines expected for γ-alumina (JCPDS file 050-0741). However, smaller full width at half maximum (FWHM) and higher diffraction peaks intensity were noticed for the commercial γ-Al₂O₃ compound, suggesting large crystal size variations among both samples. Table 1 summarizes the textural properties of the parent alumina host support and the MgO impregnated samples. The alumina sample prepared through the sol-gel procedure exhibited a specific surface area of 300 m²/g with a pore volume of 0.28 cm³/g and a mean pore size of 3.5 nm, determined by applying the BJH model to the adsorption branch. By contrast, decreased specific surface area (170 m²/g) and larger mean pore diameter (13 nm) was observed for the commercial γ-alumina sample, indicating that larger and more crystallized alumina particles are present, as confirmed by the XRD pattern.

The two alumina-supported MgO samples displayed well-resolved XRD lines at 2θ ∼ 42.8° and ∼ 62.2°, readily attributed to crystalline MgO (periclase) and suggesting that the size of the particles formed are large enough to be detected by XRD (Fig. 2). Visual comparison with the stick diagrams summarizing information of JCPDS files indicated the presence of only MgO (JCPDS file 86-1627). Indeed, no other diffraction lines due to possible spinel MgAl₂O₄ phase were visible on the XRD patterns. In agreement with previously literature reported results [13], a decrease in specific surface area was logically observed after the magnesium precursor impregnation step, from 300 to 204 m² g⁻¹ (sample MgO-Al₂O₃-SG) and from 170 to 157 (MgO-γAl₂O₃), respectively. This corresponds to a specific surface area decrease of 32 and 8% with respect to that of the parent alumina support.

Fig. 3 compares the TG-DTA patterns of magnesium acetate impregnated on both sol-gel and commercial γ-alumina samples. It is clearly seen from both TG traces and DTA endotherms (non oxidative degradation of acetate ions at low temperature) and exotherms (at higher temperature), that the decomposition sequences in oxidative atmosphere of the impregnated Mg acetate followed a similar trend for both alumina supports. The main DTA exotherms between 270 and 500 °C correspond to the oxidative degradation of the acetate anions, which are in agreement with that of the unsupported magnesium precursor (TG-DTA traces not shown). The DTA peak observed at ∼ 535–540 °C for the impregnated magnesium acetate on the sol-gel alumina sample was attributed to the decomposition of magnesium carbonate species. As for lanthanum oxides, magnesium is also well-known to be sensitive to CO₂ contained in the ambient atmosphere, leading to partial surface carbonation. Such TG-DTA results were confirmed by in situ TGA-DTA coupled to mass spectroscopy experiments (Fig. 4). Thermo-gravimetric analyses combined with mass spectroscopy measurements (TGA-MS) of water (m/z 18) and CO₂ (m/z 44) signals indicated that a simultaneous production of carbon dioxide and water occurred in the temperature range of 300–500 °C. The first peak (300–365 °C) obviously corresponds to the decomposition of anhydrous acetate, while the second peak between 370 and 500 °C can be related to the decomposition of hydrogenocarbonate species. TGA-MS experiments at m/z 44 in the temperature range of 500–555 °C only showed the production of carbon dioxide, while no corresponding release of water was observed. This CO₂ production process is associated with a low weight loss and evidences the decomposition of superficial carbonate species.

The resulting decomposition sequence in oxidative atmosphere of the supported Mg acetate is summarized by the following equations:

3.2 Catalytic behavior of various alumina-supported MgO samples

When the transesterification of methyl dodecanoate with β-sitosterol was performed in the absence of catalyst, the main products are the phytosterol ester and the conjugated diene compound (stigmasta-3,5-diene) which is the side-product of sterol dehydration. Since the selectivity to the phytosterol ester and diene are close to 50%, there is a competition between the transesterification reaction and the sterol dehydration due to the high temperature and the resulting phytosterol ester yield is very low (15%).

In order to replace homogeneous catalysts by basic and reusable solids, two Mg-containing alumina compounds (MgO-Al₂O₃-SG and MgO-γAl₂O₃) were considered for the selective synthesis of phytosterol esters and compared with lanthanum oxide, which was previously used [12].

We first checked that both alumina supports (Al₂O₃-SG and commercial γ-Al₂O₃) were not selective to phytosterol esters (Table 2). While high sterol conversion values were obtained (∼92%) due to the intrinsic alumina basic sites, the side-reaction of sterol dehydration was strongly favored, leading to phytosterol esters yield ranging from 9 to 49% for the commercial and sol-gel alumina, respectively. The fact that the selectivity to sterol esters exhibited by the alumina prepared through the sol-gel route was higher than that of the commercial alumina, may account for different surface properties (nature and strength of acidic centers) due to distinct synthesis and activation procedures.

To our delight, we found that the sol-gel MgO-containing alumina sample (MgO-Al₂O₃-SG) was far more selective (95%) than sole MgO or supported MgO on the commercial alumina catalyst (56%), affording the desired product with 93% yield (Table 2, Fig. 5). We first assumed that such difference of behavior could originate from variation in the nature, number, location and possibly strength of the acidic and basic catalytic centers between the two alumina-supported MgO samples. The determination of the acidity of basic samples is crucial, because the acid sites present can be responsible for undesired side-reaction during the transesterification of natural sterols with fatty methyl esters. Indeed, as already observed with unsupported MgO [11], the presence of acid sites, also present on MgO in high amounts along with the strong basic sites, was shown to favor the formation of diene by-products through a sterol dehydration side reaction. The assessment of the surface acidity of the Mg-containing alumina samples was carried out by infrared spectroscopy of adsorbed probe molecules. IR-sensitive probe molecules, such as pyridine (py), are routinely used to determine the surface acidity of metal oxides. It could form coordinated species on Lewis acid sites (pyL) and pyridinium ions on protonic sites (pyH⁺), whose corresponding spectra are markedly different from each other. Pyridine adsorption was performed on samples evacuated at low temperature (240 °C) in order to work under conditions similar to those used during the catalytic tests. However, in our case, pyridine was not suited to unravel the number, location and strength of both surface acid sites (pyL and pyH⁺), because the useful pyridine absorption region (1650–1300 cm⁻¹) was completely obscured by the signal of carbonate species present on the surface of the alumina supported MgO samples. CO was also used as probe molecule to evaluate the acidity of the samples. Unfortunately, CO was shown to be too weakly bound on the samples activated at 240 °C and proved to be inappropriate to characterize the acid properties of the samples in these conditions.

Using such a low activation temperature typically results in the presence of high amounts of impurities (residual carbonates, hydrated catalyst surface…) on the surface of the catalysts, which hinders further analysis of the samples with common IR acidity probes such as CO and pyridine.

With the aim of understanding the role of acid sites onto the catalyst surface on the occurrence of the side reaction during the transesterification of natural sterols with fatty methyl esters, we next investigated the formation of the conjugated diene by-products from a mechanistic point of view. The reaction was first performed in the presence of the sole sterol and the commercial alumina-supported MgO catalyst (without the fatty methyl ester). In that case, sterol degradation was observed resulting in the formation of diene compounds. Such formation of by-products was also detected when the reaction was carried out with free phytosterol esters and the MgO-γAl₂O₃ catalyst, indicating a possible degradation of the main product during the catalytic reaction. This indicates that the fatty methyl ester is highly adsorbed on the catalyst surface onto the active sites that are responsible for the formation of dienes by dehydration of sterol or sterol esters transformation. Consequently, in the course of the reaction, the coverage of MgO-γAl₂O₃ catalyst with free sterols and sterol esters increased drastically with respect to that with methyl dodecanoate, thereby resulting in a significant formation of conjugated diene by-products especially for high conversion rate values (i.e. 80% for MgO-γAl₂O₃ sample). Note that the side reaction was substantially limited when the reaction was conducted in the presence of the MgO-Al₂O₃-SG catalyst (formation of side compounds not exceeding 5 wt%).

While the catalytic activity of the lanthanum oxide sample is far higher than that of the Mg-containing alumina catalysts, the corresponding phytosterol esters selectivity and yield were significantly lower (88 and 83%, respectively), indicating that the most efficient catalyst is the supported magnesium oxide on the alumina prepared by the sol-gel procedure.

On the course of our work, we next investigated the influence of the calcination temperature on the catalytic behavior of the alumina-supported MgO material prepared by the sol-gel route. The MgO-Al₂O₃-SG sample was calcined at a temperature varying between 450 and 650 °C under flowing air. The catalytic results are indicated in Table 3 with phytosterol esters selectivity values expressed at 80% sterol conversion. The highest selectivity to phytosterol esters value was obtained for the catalyst calcined at 550 °C (99%) compared with 97 and 92% for the samples thermally treated at 450 and 650 °C, respectively. While all samples exhibited a similar (initial) catalytic activity (Table 3), the results reported in Fig. 6 showed major changes in the phytosterol esters selectivity as a function of sterol conversion. Indeed, the selectivity of the MgO-Al₂O₃-SG sample calcined at 450 °C markedly decreased between 80 and 100% sterol conversion, by contrast to that of the corresponding sample calcined at 550 °C (very slight variation from 99 to 95%). A significant drop of selectivity was still observed for the alumina-supported MgO catalyst calcined at 650 °C, with a selectivity value ended at ∼84%. Such results strongly suggest that the calcination temperature impacts on the nature, strength and location of the basic sites at the catalyst surface. It was already mentioned in previous studies that the increase of calcination temperature resulted in a decrease of the catalyst specific surface area, but more importantly in the decrease of the amount of basic centers of various magnesium oxides [14–17]. Moreover, we have also observed that the calcination temperature led to a strong modification of the surface properties of various La-based compounds used in the transesterification reaction (of the fatty methyl ester with β-sitosterol), with as a result a strong decrease of the resulting selectivity and yield to phytosterol esters [18].

As already mentioned in the characterization section, the MgO-Al₂O₃-SG sample calcined at 550 °C exhibited dispersed magnesium oxide particles on the porous alumina surface. However, in the particular case of the compound calcined at the highest temperature (650 °C), the presence of a small fraction of MgAl₂O₄ phase, along with the supported MgO nanoparticles, was also observed in the corresponding XRD pattern (not shown). Such spinel phase obviously stemmed from a stronger acid-base interaction between the two oxidic MgO and Al₂O₃ species (amphoteric character) favored by the higher thermal treatment temperature, confirming that an adequate calcination temperature is necessary for achieving appropriate catalyst acid-base properties well adapted for a particular catalytic reaction.

3.3 Recycling study

Performing the transesterification reaction by heterogeneous catalysis provides solid catalyst reusability and, hence, reduces operational costs by facilitating easy separation of products, as well as improves industrial processes by avoiding wax formation.

To prove the efficiency of the catalyst, a series of four tests was performed with the MgO-Al₂O₃-SG sample calcined at 650 °C. At the end of the first test, the catalyst was centrifuged and washed thoroughly with 2-methoxyethyl-ether, before being reintroduced inside the reactor without having been calcined. After three cycles of reaction, the catalytic reaction behavior (sterols conversion, phytosterol esters selectivity and yield) was shown to be totally similar, confirming that MgO-Al₂O₃-SG is a stable and reusable catalyst. As indicated in Table 4, the phytosterol esters selectivity at 80% conversion varied from 92 to 95% with analogous yield trend (74–77%). Such behavior was already observed by Pouilloux et al. [11] with sole MgO or ZnO, but in that case a significant amount of dienes by-products was generated over MgO and in the case of ZnO, the catalytic activity was lower than that obtained with sol-gel alumina-supported MgO.

A recycle test was also carried out after having regenerated the catalyst by calcination at 650 °C after the first reaction run before reintroduced the recovered catalyst inside the reactor. Here again, no differences in the catalytic properties were observed compared with that of the first run (Table 4).

In all cases, the concentration of leached magnesium species was extremely low (variation between 0.005 and 0.011 wt%), confirming that the Mg-based active sites were stabilized and strongly retained onto the porous alumina surface due to a strong acid-base interaction. Such performances are indeed required for efficient sterols conversion and for a reduced contamination of residues by alkaline-earth metal ions stemming from leached catalysts.