Myller S.
Carvalho
a,
Raísa A.
Lacerda
a,
João P. B.
Leão
a,
Jackson D.
Scholten
b,
Brenno A. D.
Neto
c and
Paulo A. Z.
Suarez
*a
aINCT-CATÁLISE, Laboratório de Materiais e Combustíveis, Instituto de Química, UnB, Brasília, DF, Brazil. E-mail: psuarez@unb.br; Fax: +55 6132734149; Tel: +55 6131073852
bLaboratory of Molecular Catalysis, Chemistry Institute – UFRGS, RS, Brazil
cINCT-CATÁLISE, Laboratory of Medicinal and Technological Chemistry, Instituto de Química, UnB, Brasília, DF, Brazil
First published on 26th April 2011
An important drawback to be overcome in biodiesel technology is its low oxidative stability. One approach to improve the oxidative stability of soybean oil biodiesel is the partial hydrogenation of double bonds. In the current work, an efficient two-phase catalytic system using palladium acetate dissolved in BMI·BF4 ionic liquid to an in situ generation of palladium nanoparticles was developed in order to promote a selective hydrogenation reaction. Upon using this catalytic system it was possible to partially hydrogenate biodiesel into mono-hydrogenated compounds avoiding the formation of saturated compounds. The nanoparticulate system was compared with the traditional heterogeneous Pd/C system and gave far higher selectivity. It was possible to recover and reuse the ionic phase containing the catalyst up to three times without significant loss in its catalytic performance. Indeed, atomic absorption spectroscopy showed an excellent reclaim of the catalyst, which stayed in the ionic phase. Several parameters, such as temperature, hydrogen pressure, metal concentration and reaction time, were also evaluated.
![]() | ||
Fig. 1 Structures of the main fatty acids present in soybean oil: oleic, linoleic and linolenic acids. |
One approach to improve soybean based biodiesel oxidative stability is the use of antioxidant additives,8 which are able to retard the reaction of oxygen triplet with the allylic positions in the fatty acid chains. Regrettably, these additives do not eliminate the double bonds. In some international specifications for biodiesel, such as EN 14214 in the European Community, the iodine index parameter is also specified, which is a direct measurement of olefins.9 Therefore, even using antioxidant additives, this regulation eliminates soybean oil as a potential feedstock, as it generally does not pass the test of the iodine index parameter.10 Besides, it is largely known that the oxidative stability increases with the saturation degree of the fatty acid chains. Therefore, another possible alternative to stabilize biodiesel is the direct hydrogenation of the double bonds.10
One major drawback to the direct hydrogenation of biodiesel is that upon decreasing the unsaturated chains, a straight increase in the melting point of biodiesel is noted, making less interesting its beneficial properties, thus becoming a hard task to match some specified parameters such as a cold filter plugging point.11 Hence, in order to improve the oxidation stability without compromising biodiesel properties, the partial and selective hydrogenation is a main goal and a real challenge to this emerging technology. Preferentially, the process should result in a higher oleic composition, which displays only one double bond. Besides, it is also well known that during hydrogenation processes, the cis double bonds are partially isomerized into trans isomers.12–19 This phenomenon is particularly undesirable due to the highest crystallization point of trans isomers when compared with those of a cis conformation, which becomes also a critical issue for the cold filter plugging point parameter especially for monounsaturated fatty acids.16
In order to overcome major technological problems associated with polyunsaturated biodiesel derivatives, and to avoid selectivity issues in the hydrogenation reaction, we envisage the use of imidazolium-based ionic liquids (ILs, Fig. 2) as a media to promote direct biodiesel hydrogenation. Our main expectation was that this class of reaction medium would be able to allow a higher control in the “oleic-like” ester derivatives formation during the partial hydrogenation process.
![]() | ||
Fig. 2 Examples of imidazolium-based ionic liquids. Note that these are derivatives of the 1-butyl-3-methylimidazolium cation. |
Our group has a large experience in developing and applying ILs,20–23 including biodiesel synthesis in these media.24–27 It is well known that ILs have a beneficial effect on the selectivity of many reactions, and they can also promote reactions that are difficult to occur in classical organic solvents.28–31 The aim of this work is to study the catalytic activity of palladium acetate dissolved in ILs to an in situ generation of palladium nanoparticles that could be used in the selective hydrogenation of soybean fatty acid methyl esters (FAMEs). The study on the influence of palladium nanoparticles on the selective hydrogenation reaction (and on its parameters, e.g. pressure and temperature) and the catalyst/FAME molar ratio, as well as the recycling reactions is presented herein.
The starting FAME composition is shown in entry 0. At the end of the reaction, the formation of a fine black powder was observed for entries 1 to 3. In the case of entry 1, a two phase system was obtained: an organic upper layer containing FAME and an ionic lower phase composed of a dispersion of the black powder in BMI·BF4. For homogeneous conditions (entries 2 and 3) a dispersion of the black powder in FAME was observed. Finally, for entry 4 a typical heterogeneous mixture was observed at the end of the reaction. As depicted from Table 1, all studied systems were active in the hydrogenation of the different unsaturated FAMEs. When comparing entries 1–3, it can be noted that a strong selectivity effect takes place in the presence of IL 1c. Indeed, while 18∶
1 cis and 18
∶
1 trans are the most important hydrogenation products using 1c as the reaction media (Table 1, entry 1), the complete hydrogenation compound (C18
∶
0) is the major product with no ionic liquid (Table 1, entries 2 and 3) or conventional heterogeneous catalysis (Table 1, entry 4). This IL effect will be better discussed in due course. Moreover, although two-phase conditions showed the highest selectivity, it has also the lowest activity, which can be interpreted as a phase mass transfer phenomenon. Note that the highest activity and the lowest selectivity were observed using Pd/C, leading to the total hydrogenation of biodiesel, resulting in a maximum percentage of 18
∶
0, in only 2 hours of reaction.
As already described, in entries 1–3 it was possible to observe the formation of a black precipitate finely divided at the end of the hydrogenation of FAME. This precipitate remains dispersed almost exclusively in the IL phase, which was analyzed by transmission electron microscopy (TEM) to testify palladium nanoparticles formation, shape and size distribution, which is stabilized by IL 1c. The shape of the nanoparticles is discussed later.
In order to study the influence of catalyst concentration, temperature and pressure on the hydrogenation of FAME promoted by palladium nanoparticles anchored in BMI·BF41c, different experiments were carried out using Pd(OAc)2 as the catalyst precursor, which are summarized in Table 2. The catalytic activity could be evaluated by means of FAME composition, iodine value (IV) and selectivity (by FAME composition).
Entry | Pd(OAc)2a | Temp./°C | Press.b | FAME (%)c | IVd | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
|||||
a mmol of palladium acetate. b Pressure (atm). c 1 hour. d Iodine value (IV) = (mg of I2)/(g of biodiesel). e Isomerized methyl linoleate compounds mixture. f Isomerized methyl linoleate compounds mixture. | |||||||||||
0 | — | — | — | 6 | 1 | 47 | 1 | 23 | 0 | 4 | 131 |
5 | 0.1114 | 25 | 5 | 6 | 1 | 48 | 1 | 27 | 2 | 4 | 115 |
6 | 0.0446 | 25 | 5 | 5 | 1 | 46 | 3 | 26 | 1 | 4 | 120 |
7 | 0.0446 | 25 | 10 | 3 | 0 | 45 | 2 | 31 | 5 | 4 | 111 |
8 | 0.1114 | 50 | 5 | 1 | 0 | 37 | 5 | 30 | 12 | 4 | 108 |
9 | 0.1114 | 50 | 10 | 0 | 0 | 30 | 0 | 36 | 16 | 4 | 106 |
10 | 0.0446 | 50 | 1 | 6 | 0 | 48 | 0 | 25 | 2 | 4 | 126 |
11 | 0.0446 | 50 | 5 | 2 | 0 | 40 | 4 | 26 | 7 | 4 | 117 |
12 | 0.0446 | 50 | 10 | 1 | 0 | 35 | 5 | 32 | 12 | 4 | 110 |
13 | 0.0446 | 50 | 50 | 1 | 0 | 10 | 1 | 44 | 16 | 6 | 90 |
14 | 0.0446 | 50 | 75 | 0 | 0 | 12 | 2 | 50 | 1 | 7 | 73 |
15 | 0.1114 | 80 | 5 | 1 | 0 | 32 | 7 | 30 | 16 | 5 | 107 |
16 | 0.1114 | 80 | 10 | 0 | 0 | 0 | 1 | 20 | 48 | 10 | 70 |
17 | 0.0446 | 80 | 1 | 5 | 1 | 45 | 4 | 25 | 3 | 4 | 125 |
18 | 0.0446 | 80 | 5 | 0 | 0 | 33 | 2 | 32 | 17 | 4 | 107 |
19 | 0.0446 | 80 | 10 | 0 | 0 | 1 | 6 | 38 | 32 | 5 | 81 |
20 | 0.0446 | 80 | 75 | 0 | 0 | 0 | 3 | 57 | 2 | 8 | 61 |
21 | 0.0046 | 80 | 75 | 0 | 0 | 0 | 0 | 39 | 24 | 11 | 75 |
As already described elsewhere,19 at all studied temperatures, by increasing the pressure, the catalytic activity is enhanced (compare in Table 2, entries 6 and 7, and entries 10–14, 15–16 and entries 17–20). Indeed, using 0.0446 mmol of Pd(OAc)2 at 25 °C, increasing the pressure from 5 atm (Table 2, entry 6) to 10 atm (Table 2, entry 7), IV decreases from 120 to 111. This behavior was even more intense at 80 °C (Table 2, entries 17–19), when we observed iodine values of 125, 107 and 81, respectively, using 1, 5 and 10 atm of pressure of H2. This effect of pressure can be easily understood in terms of the availability of hydrogen, which enhances with temperature, because the viscosity and density of the IL strongly decrease.32 Thus, phase mass transfer issues are much less important.
In order to have a better evaluation of the effect of pressure in the reaction selectivity, the chart is plotted as shown in Fig. 3, with the percentage of 18∶
1 trans formed in entries 10–14 (Table 2).
![]() | ||
Fig. 3 Effect of hydrogen pressure on 18![]() ![]() |
As depicted in Fig. 3, increasing the hydrogen pressure from 1 to ∼50 atm increases the formation of 18∶
1 trans. At higher pressures, formation of this isomer is diminished. According to a previous study conducted by Coenen,12 the lifetime of the FAME intermediate semi-hydrogenated depends on the concentration of hydrogen atoms adsorbed. For low-pressures (Table 2, entries 10–12), the amount of hydrogen available on the catalytic surface is not sufficient to promote hydrogenation. Thus, the rotation of the CH2R fragment around the C–C bond followed by hydrogen atom elimination results in C
C double bond isomerization. Yet, at higher pressures (Table 2, entries 14 and 20), the amount of hydrogen is sufficient and reductive elimination takes place affording the hydrogenated product and the elimination of an isomerized unsaturated product is inhibited. This behavior was equally noted at 50 °C and 80 °C, being more evident at higher temperatures, despite the fact that it is not easy to rationalize.
A plausible explanation is due to the presence of in situ generated stabilized metallic nanoparticles (as described before), where the catalytic active species consumes high amounts of hydrogen and is favoured by both temperature and pressure. Thus, increasing the pressure up to 50 atm, more active species are formed and due to diffusion issues there is not enough hydrogen available, making the isomerization more competitive than hydrogenation. For high pressures, hydrogen diffusion does not compromise its availability in the catalytic sites and the isomerization becomes less important than hydrogenation. Nonetheless, it is important to highlight that even with this increasing isomer formation for pressures down to 50 atm, the system presented here is highly selective when compared with other works reported previously in the literature. For instance, Deliy et al.14 obtained 40% of 18∶
1 trans at 10 atm and 30 °C and in this current work, at similar conditions, only 5% (Table 2, entry 7) of the trans isomer was observed. Note that even at high temperatures (80 °C and 10 atm—Table 2, entry 19), that usually promote isomerization, only 32% of 18
∶
1 trans was obtained.
It was also observed that the amount of 18∶
0 did not increase significantly upon increasing the pressure. Indeed, it seems that the 18
∶
3 and 18
∶
2 had a greater selectivity by catalytic site than 18
∶
1 chains. As already highlighted, the high selectivity in mono-hydrogenated products from dienes using ILs was claimed to be due to the higher solubility of dienes in BMI·BF4 when compared to mono-enes and paraffins.34 In the last few years some papers have been published relating the hydrogenation of polyunsaturated compounds using metal complexes dissolved in ILs.33,34,39 It was claimed in these papers that the highest solubility of dienes when compared with mono-enes in imidazolium based ILs provides high selectivity using molecular compounds34,39 or nanoparticles (such as in our case),35 always leading preferentially to partially hydrogenated products. Besides enhancing the selectivity of the catalytic systems, ILs have been highlighted as environmental friendly media useful for two phase catalysis, providing easy separation of the catalysts from the products and high activity due to charged intermediates stabilization.36,37 Actually, the IL effect observed for our selective hydrogenation is that mono-enes formed during the reaction are separated (solubility effect) from the reaction medium (ionic phase). Additionally, the IL effect is responsible to (co-)promote the formation and stabilization of ionic intermediates of the hydrogenation reaction through different types of ion pairs formation.
In this work we tried to determine the differences in the solubility of C18∶
0, C18
∶
1, C18
∶
2 and C18
∶
3 in the ionic 1c but the very low solubility and high viscosity and density of the biodiesel mixture led to unprecise results. However, despite the fact that it was not possible to determine with accuracy the comparative solubility of the different C18 FAMEs, by analogy with the different studies available in the literature and discussed before (see ref. 33–35,39) it is more than reasonable to assume that the selectivity in this case can also be attributed to the higher solubility of polyunsaturated FAME in IL 1c in comparison to monounsaturated and saturated FAMEs. It means, the formation of the saturated product was inhibited by diffusion problems of 18
∶
1 isomers to the catalytic site. In order to check this hypothesis, a reaction with no IL was performed and no selectivity was observed. The natural course is to form saturated products that in fact occurs very fast. Another important result shown in Table 2 is for the reactions performed at 75 atm of H2 pressure (entries 14 and 20). We observed a high removal of polyunsaturated FAME without an increasing level of 18
∶
1 trans and 18
∶
0. Thus, it was possible to maintain its beneficial properties, due to a higher oleic composition, naturally, also increasing its oxidation stability.
ILs display both high organization and the presence of ionic channels (for instance, see ref. 38), which can also be responsible for a faster hydrogenation than isomerization. When the substrate (biodiesel) is inside the three dimensional well-organized supramolecular structure of BMI·BF41c, the steric hindrance effect has an enhanced importance, and hydrogen transfer is favored than isomerization (which is avoided), mainly due to the long size chain of the substrate that is stacked inside the used IL (see Scheme 1). Consequently, hydrogen transfer may occur faster than isomerization reaction. Moreover, after mono-ene formation, a solubility effect takes place, favoring its maintenance as a mono-ene compound.
![]() | ||
Scheme 1 Ionic liquid effect favoring hydrogenation reaction faster than isomerization reaction. |
Note in Scheme 1 that the well-organized 3D structure of BMI·BF4 probably avoids an easy isomerization due to the steric hindrance effect. Also, note that hydrogen transfer is facilitated inside the 3D structure of IL since it stabilizes the palladium nanoparticles and the charged intermediates from hydrogen transfer.
The temperature effect on biodiesel hydrogenation was equally studied at 25, 50 and 80 °C as previously summarized in Table 2. Using the same amount of the catalyst, it was observed that by increasing the temperature the catalytic activity was increased (compare entries 5, 8 and 15; entries 6, 11 and 18; entries 7, 12 and 19; entries 9 and 16; entries 14 and 21). Results are in perfect agreement with those related in the literature.19 It was also observed, at least for the first reaction hour, that upon temperature increase, the percentage of 18∶
1 trans raises and has almost no influence on the formation of 18
∶
0. The result confirms previous works described in the literature.12,14,15
In Fig. 4 we see the shape and size distribution of the generated palladium nanoparticles. Note that even at different temperatures both nanoparticles shape and size are statistically the same. Indeed, the diameters were calculated at different temperatures, resulting in 5.1 ± 0.7 nm at 25 °C, 4.7 ± 0.9 nm at 50 °C and 4.2 ± 0.8 nm at 80 °C. It is also important to note that the best nanoparticle distribution was obtained at 50 °C. A long term reaction (96 h) was carried out at 10 atm and 50 °C using 0.0446 mmol of Pd. Fig. 5 shows the composition of samples collected at different times.
![]() | ||
Fig. 4 MET analysis of palladium nanoparticles imbibed in BMI·BF4: (a) 25 °C (5.1 ± 0.7 nm); (b) 50 °C (4.7 ± 0.9 nm); and (c) 80 °C (4.2 ± 0.8 nm). |
![]() | ||
Fig. 5 Biodiesel hydrogenation products as a function of time. Reaction conditions: P(H2) = 10 atm, T = 50 °C; 2 mL of BMI·BF4. |
As can be depicted from Fig. 5, the polyunsaturated compounds are quickly hydrogenated into monounsaturated compounds. Indeed, 18∶
3 and 18
∶
2 are consumed after 60 and 1055 min, respectively. Note that C18
∶
0 starts to be formed only after 340 min of reaction and in 1055 min only 5% of the unsaturated compounds were completely hydrogenated. Thus, using the catalytic system proposed herein, it is possible to convert almost all the polyunsaturated compounds into the so-desired monounsaturated products, avoiding the formation of the paraffinic FAME.
Therefore, it is also possible to depict from Fig. 5 that the hydrogenation of the monounsaturated compounds has a slower velocity when compared with the hydrogenation of polyunsaturated ones. The IV against the time curve shown in Fig. 6, which displays two distinct regions, allows a better visualization of the observed behavior. Indeed, from first to the sixth hour of reaction, a high hydrogenation rate is observed (IV drops by approximately six units per minute, which is almost six times higher to the values observed in an industrial process)15 and after 10 h a very low hydrogenation rate is observed (IV drops by 10−3 units per minute). As already pointed out, this is possibly a result of diffusion issues due to differences in the solubility of the C18∶
3, C18
∶
2 and C18
∶
1. It is interesting to compare the results from this study with those shown in Table 1 in the absence of any IL. Note that with no IL (Table 1, entry 1), biodiesel was completely hydrogenated in 1440 min. To check this hypothesis, at the end of the reaction, the organic phase was separated and mixed with a new charge of Pd(OAc)2 (0.446 mmol) dissolved in 2 mL of BMI·BF4 and the system was once more pressurized with 10 atm of H2 for additional 24 h at 50 °C. The final product composition was 0% of 18
∶
3; 0% of 18
∶
2; 29% of 18
∶
1 cis; 27% of 18
∶
1 trans, and 17% of 18
∶
0; and the IV was 64. It means the IV dropped by only 4%, showing that the low reaction rate is not a result of catalyst deactivation but probably a low hydrogenation activity of the system for monounsaturated compounds as explained before.
![]() | ||
Fig. 6 Iodine value as a function of time. |
As highlighted, one of the advantages in the use of ILs as a reaction medium in a two-phase catalysis is the possibility of recover and reuse of the catalysts. For instance, some of us hydrogenated 1,3-butadiene with the palladium acetylacetonate dissolved in BMI·BF41c and could carry out up to 15 recycles without loss of catalytic activity.39 Recycling of the ionic catalytic phase was achieved in the current work by simple phase separation and recovering the black ionic viscous suspension (containing palladium nanoparticles). It was recharged for further runs and the results are shown in Table 3. Note that this study was conducted using the reaction parameters similar to those previously used (Table 2, entry 14: 10 mg of the catalyst precursor, 2 mL of BMI·BF4, 15 mL of biodiesel, 90 min, 75 atm of H2 and 50 °C).
Entry | Percentage of FAME (%) | IV | ||||||
---|---|---|---|---|---|---|---|---|
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
18![]() ![]() |
||
Starting biodiesel | 7 | 1 | 52 | 1 | 23 | 0 | 3 | 135 |
22 | 0 | 0 | 0 | 3 | 57 | 4 | 8 | 61 |
23 | 0 | 0 | 0 | 2 | 56 | 3 | 9 | 60 |
24 | 0 | 0 | 7 | 4 | 50 | 3 | 7 | 68 |
25 | 1 | 0 | 21 | 0 | 41 | 2 | 7 | 80 |
26 | 0 | 1 | 29 | 2 | 33 | 1 | 5 | 86 |
27 | 3 | 1 | 38 | 1 | 24 | 1 | 4 | 106 |
28 | 6 | 0 | 51 | 1 | 23 | 0 | 4 | 136 |
As can be depicted from Table 3, the catalytic activity and selectivity remained almost the same in the two first charges of biodiesel. However, a continuous drop was observed in the catalytic activity from the third run on, being verified complete deactivation of the catalytic system in the seventh run. This behavior can be better visualized in Fig. 7.
We envisage that part of palladium nanoparticles could be migrating to the FAME phase. Actually, atomic absorption experiments confirmed that 3% of the palladium content can be found in the FAME phase after recycling reaction, which does not explain the loss of activity during recharge reactions. Despite the partial possibility of recycling the system, the obtained results are considerably interesting, mainly because until the moment, to the best of our knowledge, it is the most selective and efficient system to perform selective hydrogenation of biodiesel towards oleic composition of the final product. However, TEM analysis (data not shown) performed after catalysis demonstrated that the palladium nanoparticles are not agglomerating during recharge reactions. To obtain a plausible explanation to the observed behavior (loss of activity), BET analysis was performed. Interestingly, we could observe after the first run a superficial area of 171.685 m2 g−1, a porous volume of 0.3854 cc g−1 and a porous size smaller than 735.8 Å. After the sixth run, the same analysis was performed and we could observe a superficial area of 199.615 m2 g−1, a porous volume of 0.4726 cc g−1 and a porous size smaller than 412.8 Å. If the porous size is decreasing during the reaction, it is now comprehensive the reason for the loss of activity. The substrate cannot access the catalyst and the reaction does not take place, especially because the tested FAMEs are much higher in van der Waals volume than 1,3-butadiene.
![]() | ||
Fig. 7 Biodiesel hydrogenation yields during recycling reactions. |
This journal is © The Royal Society of Chemistry 2011 |