CHEMSUSCHEM COMMUNICATIONS DOI: 10.1002/cssc.201402236
One-pot Aldol Condensation and Hydrodeoxygenation of Biomass-derived Carbonyl Compounds for Biodiesel Synthesis Laura Faba, Eva Daz, and Salvador OrdÇez*[a] Integrating reaction steps is of key interest in the development of processes for transforming lignocellulosic materials into drop-in fuels. We propose a procedure for performing the aldol condensation (reaction between furfural and acetone is taken as model reaction) and the total hydrodeoxygenation of the resulting condensation adducts in one step, yielding n-alkanes. Different combinations of catalysts (bifunctional catalysts or mechanical mixtures), reaction conditions, and solvents (aqueous and organic) have been tested for performing these reactions in an isothermal batch reactor. The results suggest that the use of bifunctional catalysts and aqueous phase lead to an effective integration of both reactions. Therefore, selectivities to n-alkanes higher than 50 % were obtained using this catalyst at typical hydrogenation conditions (T = 493 K, P = 4.5 MPa, 24 h reaction time). The use of organic solvent, carbonaceous supports, or mechanical mixtures of monofunctional catalysts leads to poorer results owing to side effects; mainly, hydrogenation of reactants and adsorption processes.
The preparation of drop-in fuels from lignocellulosic biomass (second-generation biofuels) is nowadays of key interest. Many of these technologies, such as pyrolysis or gasification followed by Fischer–Tropsch synthesis, are trade-offs in terms of energy because of the high reaction temperatures they require.[1] In addition, they lead to complex hydrocarbon mixtures that necessitate distillation units with high energy demands. Consequently, the carbon footprints of these products are similar to fossil fuels.[2] On the other hand, liquid-phase processes have attracted much attention because of the lower reaction temperatures and the higher selectivities that can be obtained.[3] Moreover, the separation is easier because of the immiscibility of the resulting organic and the aqueous phases. At this point, Dumesic and co-workers have proposed a scheme for obtaining dieselquality biofuels from cellulosic and hemicellulosic materials using a cascade of four catalytic steps: (i) hydrogenolysis of the biopolymer; (ii) dehydration of the resulting sugars, yielding carbonylic species (furfural and hydroxymethylfurfural); (iii) [a] Dr. L. Faba, Dr. E. Daz, Prof. S. OrdÇez Department of Chemical and Environmental Engineering University of Oviedo C/Julin Clavera, s/n—33006—Oviedo—Asturias (Spain) Fax: (+ 34) 985103434 E-mail: [email protected] Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cssc.201402236.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
condensation of these species (in the most usual approach, by adding compounds with two carbons with a-hydrogen atoms, such as acetone); and (iv) hydrodeoxygenation of the resulting condensation adducts.[4] The first three steps are done in the aqueous phase, whereas the last one is more usually performed in the organic phase. Concerning the catalysts: the first two steps involve acid catalysis and can be easily integrated using a variety of mineral acids such as HCl, H2SO4, or H3PO4,[5] or solid catalysts (zeolites) at temperatures lower than 580 K.[6] On the other hand, basic catalysts are needed for aldol condensation whereas metal particles catalyze hydrodeoxygenation reactions.[7, 8] Considering the compatibility of both functionalities, the main scope of this Communication is the integration of these last steps is in order to obtain liquid n-alkanes from furfural (a dehydrated derivative of pentoses). This integration leads to a two-step process for obtaining biofuels from cellulosic materials, taking advantage of the hemicellulosic fraction. In the case of actual biomass, a water pretreatment allows to separate the hemicellulose, recovering almost the complete amount of xylo-oligomers at low cost and without any damage to the cellulose and lignin.[9] These fractions can be integrated into other biomass conversion technologies obtaining energy, fuels, and chemicals simultaneously (in the context of biorefineries).[10] Considering the decrease in costs of such coproduction, the main techno-economic disadvantage of this configuration should be the use of hydrogen. However, the hydrogen required (5.5 mol of H2 per mol of furfural, considering total selectivity to n-tridecane) can be obtained in situ by gasification of residual fractions of the biomass, with a considerable reduction in costs.[11] When considering the integration mentioned above more deeply, different problems arise, mainly related to development of the catalyst. The aldol condensation is carried out in the aqueous phase (where encountering the carbonyl precursors is more likely) and requires a very specific distribution of basic sites.[7, 12] This distribution is present in bulk mixed oxides (such as Mg–Zr mixed oxides),[7] being even more favorable when these oxides are dispersed onto inert supports, such as high-surface-area graphites (HSAG).[12] On the other hand, the introduction of active metals in these materials can lead to uncontrolled changes in the basicity patterns and the morphology (decrease in pore volume), affecting the selectivity to the desired compounds.[13] Operating with mixtures of basic catalysts and conventional hydrogenation catalysts, such as noblemetal catalysts supported on alumina or activated carbon, might overcome this problem. ChemSusChem 2014, 7, 2816 – 2820
2816
CHEMSUSCHEM COMMUNICATIONS
www.chemsuschem.org
Another aspect that should be taken into account is the reaction medium. Usually, the aldol condensation is performed in aqueous phase, considering that organic solvents have a negative effect on catalyst performance. Furfural and acetone are soluble enough in water and this solvent promotes abstraction of the a-proton from acetone (starting point of the condensations), and increases the stability of the carbanion intermediate.[7, 14] However, hydrodeoxygenation of the resulting condensation adducts is usually performed in organic solvents, Scheme 1. General mechanism of acetone–furfural condensation and hydrogenation for obtaining alkanes. mainly because of the low water solubility of the condensation adducts.[4, 15] total hydrodeoxygenation yields (formation of n-octane and nFinally, the last difficulty to be overcome in the design of tridecane). the one-pot process is the direct hydrogenation of carbonylic The first set of reactions was performed in organic phase, reactants (acetone and furfural), yielding furfuryl alcohol and considering two different situations: optimal conditions for the isopropanol; compounds that do not undergo further condenslowest step (hydrodeoxygenation), or a two-step process, sation reactions. with each step performed in favorable conditions for aldol Herein, we have tested different approaches to accomplish condensation and hydrodeoxygenation, respectively. In oneintegration of aldol condensation and hydrodeoxygenation of pot processes, reaction times of 24 h have been considered, the resulting adducts into a one-pot process, in such a way according to previous studies on condensation.[7] This time that n-alkanes could be obtained directly from furfural and was extended to 48 h in the two-step process to guarantee acetone, as shown in Scheme 1 and detailed in the Supporting that maximum selectivities were obtained. Information (Scheme S1). The best results achieved for each A one-pot process was performed at typical hydrodeoxygestep in previous studies were considered as initial conditions nation conditions (P = 4.5 MPa, T = 493 K) using a bifunctional for this study.[7, 16] Experiments were carried out in a stirred catalyst (Pt-impregnated Mg–Zr mixed oxide) and a mechanical catalyst mixture (Pt/AC + MgZr/HSAG500). The results are sumbatch reactor, described in the Experimental section. The remarized in Table 1 (entry O1), whereas the temporal profile of sults are discussed in terms of both condensation yields (forreactants and products obtained with Pt/MgZr is detailed in mation of adducts of eight and thirteen carbon atoms), and
Table 1. Summary of the main results obtained with the different reaction schemes tested. Results after 24 h at 493 K and 4.5 MPa of H2 in the case of one-step process in organic “O1” or aqueous phase “A1” and after 48 h (24 h at 323 K and 1 MPa and 24 h at 493 K and 4.5 MPa H2) in the case of twosteps process in organic “O2” or aqueous phase “A2”.[a] Entry
O1 O2
A2
A1
Catalyst Pt/MgZr Pt/AC + MgZr/HSAG500 Pt/MgZr Pt/MgZr/HSAG500 MgZr + Pt/Al2O3 Pt/MgZr MgZr + Pt/Al2O3 MgZr + Pt/AC MgZr/HSAG500 + Pt/AC MgZr/HSAG500 + Pt/Al2O3 Pt/MgZr MgZr + Pt/Al2O3
Conversion [%] C5 C3
Selectivity[b] [%] C3H8O C5H6O2
SC8/SCi0
SC13/SC i0
Carbon balance
47.2 23.1 31.2 19.4 23.5 56.3 63.3 62.7 78.0 42.7 46.2 48.9
2.2 3.1 4.3 3.3 7.7 0.5 0.0 0.0 0.0 0.6 0 0
7.7 0.1 3.7 10.4 19.2 7.1 8.0 37.7 9.5 10.3 40.9 35.7
5.7 0.1 1.1 7.9 43.2 37.6 14.3 31.1 13.4 41.7 38.3 39.7
58.2 62.5 82.4 83.9 94.0 61.5 70.8 49.8 15.6 38.8 99.6 94.0
94.2 41.5 87.5 54.7 72.1 99.1 100 99.7 100 99.9 99.6 99.9
35.2 20.4 23.1 35.8 33.6 0.6 0.1 0.0 0.0 1.0 0.8 0.2
[a] The mathematical expressions of conversion, selectivities and carbon balance are detailed in the Supporting Information. [b] C3H8O and C5H6O2 are the selectivities to isopropanol and furfuryl alcohol, respectively; and SC8/Cio and SC13/Cio are the carbon yields to the formation of C8 and C13 adducts, respectively. More details about how are these selectivities were calculated are given in the Supporting Information.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 2816 – 2820
2817
CHEMSUSCHEM COMMUNICATIONS the Supporting Information (Figure S1). In both cases the selectivity for n-tridecane was almost negligible, but there were important differences between these materials. The results of Pt/ MgZr are conditioned by the condensation step, whereas this material shows a high hydrodeoxygenation activity, mainly for the C8 adduct (chemical structures of the primary condensation adducts and further hydrogenation and hydrodeoxygenation products are given as Supporting Information). The global selectivity for condensation adducts (C8 and C13) is lower than 14 % after 48 h, but more than 40 % of the C8 obtained was completely hydrogenated. The reaction was mainly limited by a fast hydrogenation of furfural (enhanced at high temperature), with more than 35 % of furfuryl alcohol selectivity at these conditions (Supporting Information, Figure S1). On the other hand, results obtained with the mechanical mixture are limited by the hydrogenation of reactants but also by the adsorption of reactants and reaction intermediates on the active carbon surface, leading to carbon balances lower than 63 % after 48 h. Because hydrogenation of reactants inhibits further condensation, the procedure has been improved, operating in two consecutive steps: the first one at 323 K in inert atmosphere (1 MPa N2), and the second one in hydrogen (4.5 MPa) and at 493 K. These results are summarized in Table 1, entry O2, whereas time profiles obtained at these conditions with Pt/ Al2O3 + MgZr are detailed as example in the Supporting Information (Figure S2). Concerning the bifunctional catalysts, and in a good agreement with the previous results for condensation reaction, the carbon-supported catalyst shows a better performance for the condensation reaction than the bulk metal oxide.[12] This better performance is caused by the more favorable distribution of basic sites when the oxides are dispersed onto an inert support. It should be noted that the best performance found in a previous work was observed in the aqueous phase, suggesting similar mechanisms in both phases. In addition, carbon-containing catalysts strongly interact with the C13 adduct, favoring the subsequent hydrodeoxygenation. Better condensation results were obtained with a mechanical mixture of basic and metal catalysts, with more than 56 % of C13 condensated adducts after 48 h reaction time. The most notable differences occurred during the hydrodeoxygenation step. Aldol condensation reactions are enhanced in the absence of metal active phases,[17] so the segregation between these sites favors that the C8-condensated adduct remains adsorbed on the basic site without undergoing hydrogenation. The high temperatures increase the ratio of condensation, obtaining more C13 adducts. Inorganic supports for metal particles lead to poor interaction with C13 adducts, hindering further hydrodeoxygenation. Consequently, in spite of the low activity, this configuration allows obtaining a higher yield of n-tridecane, as can be viewed in Figure 1. The same procedure was repeated using water as solvent instead of n-hexane. Profiles obtained with Pt/MgZr are shown in the Supporting Information (Figure S3). Both reactants (furfural and acetone) are soluble enough in water to perform the experiment in the aqueous phase. In general terms, the selec 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
Figure 1. Comparison of final results obtained with different catalysts in organic solvent for the two-step process (24 h at condensation conditions and 24 h at hydrogenation conditions). Results are given in terms of the ratio between the carbon atoms in the resulting alkanes and the total content of carbon atoms of the reactants (%).
tivities for the condensation reactions increase in the aqueous phase, which is in a good agreement with literature findings,[18] suggesting that the enolate form of the acetone is solvated in the aqueous phase, enhancing its condensation reactions. The highest activity for the condensation reaction was obtained by using a mixture MgZr + Pt/AC as catalyst, reaching almost 70 % of C8 and C13 global selectivity. These reaction conditions also led to an improvement of the hydrodeoxygenation efficiency of the condensation adduct, this effect being more distinct when using the catalyst mixture and inorganic materials as metal support. Under these conditions, more than 50 % of each adduct was completely hydrogenated after 48 h. Considering both steps, the total amount of initial carbon that is transformed into n-octane and n-tridecane as function of the catalyst is shown in Figure 2. As can be observed, more than 12 % of alkanes were obtained with MgZr/HSAG500 + Pt/Al2O3, with a high selectivity of n-tridecane. These results contrast with the values of the carbon balances. Mass balance closure is markedly lower than in the same experiments performed in organic media. This can be attributed to the lower solubility of the condensation adducts in the aqueous phase, especially at the lower temperatures used for the aldol condensation. Indeed, the carbon balances get worse when using carbon-
Figure 2. Comparison of final results obtained with different catalysts in aqueous phase for the two-step process (24 h at condensation conditions and 24 h at hydrogenation conditions). Results are given in terms of the ratio between the carbon atoms in the resulting alkanes and the total content of carbon atoms of the reactants (%).
ChemSusChem 2014, 7, 2816 – 2820
2818
CHEMSUSCHEM COMMUNICATIONS supported catalysts (hydrophobic material), and are even worse when the support is microporous (higher adsorption capacity). These findings led us to exclude the use of carbon-based catalysts for this one-pot process, in spite of the good performance of these materials for each individual reaction.[12, 16] From the results obtained in the two-step processes, and literature findings about the highest reaction rates for the condensation in aqueous medium, a reaction scheme can be proposed, considering one-pot and one-step reactions performed in aqueous phase and at the conditions optimized for the hydrogenation reactions (493 K and 4.5 MPa). Two different catalysts were considered for this experiment: the bifunctional Pt/ MgZr and a mixture of bulk Mg-Zr oxide and Pt/Al2O3 catalyst. Temporal profiles obtained are shown in the Supporting Information (Figure S4). Our obtained results show that carbon balance closures are very high (close to 100 %) under these conditions, especially with the bifunctional catalyst (Table 1, entry A1). The highest average temperature of these experiments increases the solubility of these condensated adducts on water and hinders their adsorption. In spite of the higher temperature, condensation in the aqueous phase prevails over hydrogenation and neither isopropanol nor furfuryl alcohol was detected in these experiments. Thus, selectivities higher than 70 % for condensation adducts were observed for both catalytic systems. Major differences were observed in the hydrogenation of these adducts. In the case of the mechanical mixture, only n-tridecane was observed, obtaining an overall yield of 14.5 % for the formation of this alkane (Figure 3). However,
www.chemsuschem.org to those obtained in the two-steps process, the results are completely different, with a significant improvement when hydrogenation conditions are applied. This suggests that the solubility of these adducts in water plays a key role in the process, being this solubility favored at higher temperatures. In the experiment performed with the bifunctional catalyst (the best combination among the ones we tested), the total yield to n-alkanes was higher than 50 %, resulting in an nalkane mixture with a 66 % of n-tridecane concentration. This mixture is a diesel-range fuel, suitable for direct blending with a mineral diesel fraction. In conclusion, based on using model compounds the reported results show the possibility of transforming furfural and acetone (and eventually other biobased carbonylic compounds) into n-alkane mixtures in a one-pot process, integrating aldol condensation and total hydrodeoxygenation of the condensation adducts. The decrease in H2 pressure during the hydrodeoxygenation does not reach 5 bar, indicating that the reaction is not limited by the availability of H2. Furthermore, the optimal conditions for obtaining the highest n-alkane yields involve performing the process in the aqueous phase, enabling direct recovery of the n-alkanes because they are immiscible with water, thereby avoiding distillation or extraction procedures. In addition, the procedure only involves the use of heterogeneous catalysts, which are easy to separate and reuse. The yield of n-alkanes reported here, 50 %, is especially meritorious if it is compared to yields recently reviewed in the literature, 70 %, for heterogeneously catalyzed two-step processes, with each step independently optimized, and working at higher hydrogen pressures in the latter step.[18] This approach offers a clear advantage for such processes, and should be further validated with real biomass materials.
Experimental Section
Figure 3. Comparison of final results obtained with different catalysts in aqueous phase for the one step process (24 h at hydrogenation conditions). Results are given in terms of the ratio between the carbon atoms in the resulting alkanes and the total content of carbon atoms of the reactants (%).
almost null n-octane selectivity was measured. When the bifunctional catalyst is used, the selectivities for the formation of both alkanes largely increase, being the fraction of adduct completely hydrogenated a 40 and 85 % for C8 and C13 adducts, respectively. These results suggest that desorption of the condensation adducts from the base catalysts and the readsorption on the surface of the hydrogenation catalysts are slow enough to influence the selectivities. If the results obtained in the aqueous-phase one-step procedure are compared 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Reported experiments were performed in a 0.5 L stirred batch autoclave reactor (Autoclave Engineers EZE Seal). Two different solvents were used: distilled water (entries A1 and A2) and n-hexane (entries O1 and O2). Experimental conditions corresponded to a total loading of 5 % of organic compounds in solvent; specifically, 0.324 mol L 1 of acetone and 0.324 mol L 1 of furfural. Previous works noted that this is the optimal molar ratio for maximizing the yield to condensated adducts.[7] Samples were analyzed by GC-FID in a Shimadzu GC-2010, using a 15 m long CP-Sil 5 CB capillary column. Concerning to the catalysts, bifunctional catalysts with basic and metal functionalities (1 % Pt/MgZr, 1 % Pt/MgZr/ HSAG500), mechanical mixtures of basic catalysts (MgZr, MgZr/ HSAG500), and metal catalysts (0.5 % Pt/Al2O3, 1 % Pt/AC) were considered. Reactions were catalyzed by 0.1 g of bifunctional materials or with mechanical mixtures of 0.5 g of basic materials and 0.1 g of metal catalyst. Basic catalysts formulations were optimized for the aldol condensation of furfural–acetone mixtures in previous works,[7, 12] whereas platinum is the most active noble metal in the C8 hydrodeoxygenation.[16] All of the materials were characterized by different techniques (physisorption, TEM, XRD, and CO2-TPD), and characterization results are summarized in Table S1 in the Supporting Information. ChemSusChem 2014, 7, 2816 – 2820
2819
CHEMSUSCHEM COMMUNICATIONS Acknowledgements This work was supported by the Spanish Government (contract CTQ2011-29272-C04-02). L.F. thanks the Government of the Principality of Asturias for a Ph.D. fellowship (Severo Ochoa Program) Keywords: aldol reaction · biofuels · C C coupling · heterogeneous catalysis · hydrodeoxygenation [1] a) A. Demirbas, Appl. Energy 2011, 88, 17 – 28; b) A. L. Villanueva Perales, C. Reyes Valle, P. Ollero, A. Gmez-Barea, Energy 2011, 36, 4097 – 4108. [2] a) H. Michel, Angew. Chem. Int. Ed. 2012, 51, 2516 – 2518; Angew. Chem. 2012, 124, 2566 – 2566; b) S. G. Wettstein, D. M. Alonso, E. I. Grbz, J. A. Dumesic, Curr. Opin. Chem. Eng. 2012, 1, 218 – 224. [3] a) A. V. Subrahmanyam, S. Thayumanavan, G. W. Huber, ChemSusChem 2010, 3, 1158 – 1161; b) E. Furimsky, Catal. Today 2013, 217, 13 – 56. [4] G. W. Huber, J. N. Chheda, C. J. Barrett, J. A. Dumesic, Science 2005, 308, 1446 – 1450. [5] J. N. Chheda, Y. Romn-Leshkov, J. A. Dumesic, Green Chem. 2007, 9, 342 – 350. [6] R. Sahu, P. L. Dhepe, ChemSusChem 2012, 5, 751 – 761. [7] L. Faba, E. Daz, S. OrdÇez, Appl. Catal. B 2012, 113, 201 – 211. [8] L. Faba, E. Daz, S. OrdÇez, Appl. Catal. B 2014, 160 – 161, 436 – 444.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org [9] a) S. V. Vassilev, D. Baxter, L. K. Andersen, C. G. Vassileva, T. J. Mogan, Fuel 2012, 94, 1 – 33; b) R. Xing, A. V. Subrahmanyam, H. Olcay, W. Qi, G. P. van Walsum, H. Pendse, G. W. Huber, Green Chem. 2010, 12, 1933 – 1946. [10] J. Q. Bond, A. A. Upadhye, H. Olcay, G. A. Tompsett, J. Jae, R. Xing, D. M. Alonso, D. Wang, T. Zhang, R. Kumar, A. Foster, S. M. Sen, C. T. Maravelias, R. Malina, S. R. H. Barrett, R. Lobo, C. E. Wyman, J. A. Dumesic, G. W. Huber, Energy Environ. Sci. 2014, 7, 1500 – 1523. [11] a) L. He, D. Chen, ChemSusChem 2010, 3, 1169 – 1171; b) D. M. Alonso, J. Q. Bond, J. A. Dumesic, Green Chem. 2010, 12, 1493 – 1513. [12] L. Faba, E. Daz, S. OrdÇez, ChemSusChem 2013, 6, 463 – 473. [13] a) C. J. Barrett, J. N. Chhedda, G. W. Huber, J. A. Dumesic, Appl. Catal. B 2006, 66, 111 – 118; b) L. Faba, E. Daz, S. OrdÇez, Biomass Bioenergy 2013, 56, 592 – 599. [14] S. Abell, F. Medina, D. Tichit, J. Prez-Ramrez, J. E. Sueiras, P. Salagre, Y. Cesteros, Appl. Catal. B 2007, 70, 597 – 605. [15] A. Corma, O. Torre, M. Renz, N. Villandier, Angew. Chem. 2011, 123, 2423 – 2426. [16] R. D. Cortright, R. R. Davda, J. A. Dumesic, Nature 2002, 418, 964 – 967. [17] J. N. Chheda, J. A. Dumesic, Catal. Today 2007, 123, 59 – 70. [18] M. J. Climent, A. Corma, S. Iborra, Green Chem. 2014, 16, 516 – 547.
Received: March 27, 2014 Published online on August 1, 2014
ChemSusChem 2014, 7, 2816 – 2820
2820
One-pot Aldol Condensation and Hydrodeoxygenation of Biomass-derived Carbonyl Compounds for Biodiesel Synthesis Laura Faba, Eva Daz, and Salvador OrdÇez*[a] Integrating reaction steps is of key interest in the development of processes for transforming lignocellulosic materials into drop-in fuels. We propose a procedure for performing the aldol condensation (reaction between furfural and acetone is taken as model reaction) and the total hydrodeoxygenation of the resulting condensation adducts in one step, yielding n-alkanes. Different combinations of catalysts (bifunctional catalysts or mechanical mixtures), reaction conditions, and solvents (aqueous and organic) have been tested for performing these reactions in an isothermal batch reactor. The results suggest that the use of bifunctional catalysts and aqueous phase lead to an effective integration of both reactions. Therefore, selectivities to n-alkanes higher than 50 % were obtained using this catalyst at typical hydrogenation conditions (T = 493 K, P = 4.5 MPa, 24 h reaction time). The use of organic solvent, carbonaceous supports, or mechanical mixtures of monofunctional catalysts leads to poorer results owing to side effects; mainly, hydrogenation of reactants and adsorption processes.
The preparation of drop-in fuels from lignocellulosic biomass (second-generation biofuels) is nowadays of key interest. Many of these technologies, such as pyrolysis or gasification followed by Fischer–Tropsch synthesis, are trade-offs in terms of energy because of the high reaction temperatures they require.[1] In addition, they lead to complex hydrocarbon mixtures that necessitate distillation units with high energy demands. Consequently, the carbon footprints of these products are similar to fossil fuels.[2] On the other hand, liquid-phase processes have attracted much attention because of the lower reaction temperatures and the higher selectivities that can be obtained.[3] Moreover, the separation is easier because of the immiscibility of the resulting organic and the aqueous phases. At this point, Dumesic and co-workers have proposed a scheme for obtaining dieselquality biofuels from cellulosic and hemicellulosic materials using a cascade of four catalytic steps: (i) hydrogenolysis of the biopolymer; (ii) dehydration of the resulting sugars, yielding carbonylic species (furfural and hydroxymethylfurfural); (iii) [a] Dr. L. Faba, Dr. E. Daz, Prof. S. OrdÇez Department of Chemical and Environmental Engineering University of Oviedo C/Julin Clavera, s/n—33006—Oviedo—Asturias (Spain) Fax: (+ 34) 985103434 E-mail: [email protected] Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cssc.201402236.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
condensation of these species (in the most usual approach, by adding compounds with two carbons with a-hydrogen atoms, such as acetone); and (iv) hydrodeoxygenation of the resulting condensation adducts.[4] The first three steps are done in the aqueous phase, whereas the last one is more usually performed in the organic phase. Concerning the catalysts: the first two steps involve acid catalysis and can be easily integrated using a variety of mineral acids such as HCl, H2SO4, or H3PO4,[5] or solid catalysts (zeolites) at temperatures lower than 580 K.[6] On the other hand, basic catalysts are needed for aldol condensation whereas metal particles catalyze hydrodeoxygenation reactions.[7, 8] Considering the compatibility of both functionalities, the main scope of this Communication is the integration of these last steps is in order to obtain liquid n-alkanes from furfural (a dehydrated derivative of pentoses). This integration leads to a two-step process for obtaining biofuels from cellulosic materials, taking advantage of the hemicellulosic fraction. In the case of actual biomass, a water pretreatment allows to separate the hemicellulose, recovering almost the complete amount of xylo-oligomers at low cost and without any damage to the cellulose and lignin.[9] These fractions can be integrated into other biomass conversion technologies obtaining energy, fuels, and chemicals simultaneously (in the context of biorefineries).[10] Considering the decrease in costs of such coproduction, the main techno-economic disadvantage of this configuration should be the use of hydrogen. However, the hydrogen required (5.5 mol of H2 per mol of furfural, considering total selectivity to n-tridecane) can be obtained in situ by gasification of residual fractions of the biomass, with a considerable reduction in costs.[11] When considering the integration mentioned above more deeply, different problems arise, mainly related to development of the catalyst. The aldol condensation is carried out in the aqueous phase (where encountering the carbonyl precursors is more likely) and requires a very specific distribution of basic sites.[7, 12] This distribution is present in bulk mixed oxides (such as Mg–Zr mixed oxides),[7] being even more favorable when these oxides are dispersed onto inert supports, such as high-surface-area graphites (HSAG).[12] On the other hand, the introduction of active metals in these materials can lead to uncontrolled changes in the basicity patterns and the morphology (decrease in pore volume), affecting the selectivity to the desired compounds.[13] Operating with mixtures of basic catalysts and conventional hydrogenation catalysts, such as noblemetal catalysts supported on alumina or activated carbon, might overcome this problem. ChemSusChem 2014, 7, 2816 – 2820
2816
CHEMSUSCHEM COMMUNICATIONS
www.chemsuschem.org
Another aspect that should be taken into account is the reaction medium. Usually, the aldol condensation is performed in aqueous phase, considering that organic solvents have a negative effect on catalyst performance. Furfural and acetone are soluble enough in water and this solvent promotes abstraction of the a-proton from acetone (starting point of the condensations), and increases the stability of the carbanion intermediate.[7, 14] However, hydrodeoxygenation of the resulting condensation adducts is usually performed in organic solvents, Scheme 1. General mechanism of acetone–furfural condensation and hydrogenation for obtaining alkanes. mainly because of the low water solubility of the condensation adducts.[4, 15] total hydrodeoxygenation yields (formation of n-octane and nFinally, the last difficulty to be overcome in the design of tridecane). the one-pot process is the direct hydrogenation of carbonylic The first set of reactions was performed in organic phase, reactants (acetone and furfural), yielding furfuryl alcohol and considering two different situations: optimal conditions for the isopropanol; compounds that do not undergo further condenslowest step (hydrodeoxygenation), or a two-step process, sation reactions. with each step performed in favorable conditions for aldol Herein, we have tested different approaches to accomplish condensation and hydrodeoxygenation, respectively. In oneintegration of aldol condensation and hydrodeoxygenation of pot processes, reaction times of 24 h have been considered, the resulting adducts into a one-pot process, in such a way according to previous studies on condensation.[7] This time that n-alkanes could be obtained directly from furfural and was extended to 48 h in the two-step process to guarantee acetone, as shown in Scheme 1 and detailed in the Supporting that maximum selectivities were obtained. Information (Scheme S1). The best results achieved for each A one-pot process was performed at typical hydrodeoxygestep in previous studies were considered as initial conditions nation conditions (P = 4.5 MPa, T = 493 K) using a bifunctional for this study.[7, 16] Experiments were carried out in a stirred catalyst (Pt-impregnated Mg–Zr mixed oxide) and a mechanical catalyst mixture (Pt/AC + MgZr/HSAG500). The results are sumbatch reactor, described in the Experimental section. The remarized in Table 1 (entry O1), whereas the temporal profile of sults are discussed in terms of both condensation yields (forreactants and products obtained with Pt/MgZr is detailed in mation of adducts of eight and thirteen carbon atoms), and
Table 1. Summary of the main results obtained with the different reaction schemes tested. Results after 24 h at 493 K and 4.5 MPa of H2 in the case of one-step process in organic “O1” or aqueous phase “A1” and after 48 h (24 h at 323 K and 1 MPa and 24 h at 493 K and 4.5 MPa H2) in the case of twosteps process in organic “O2” or aqueous phase “A2”.[a] Entry
O1 O2
A2
A1
Catalyst Pt/MgZr Pt/AC + MgZr/HSAG500 Pt/MgZr Pt/MgZr/HSAG500 MgZr + Pt/Al2O3 Pt/MgZr MgZr + Pt/Al2O3 MgZr + Pt/AC MgZr/HSAG500 + Pt/AC MgZr/HSAG500 + Pt/Al2O3 Pt/MgZr MgZr + Pt/Al2O3
Conversion [%] C5 C3
Selectivity[b] [%] C3H8O C5H6O2
SC8/SCi0
SC13/SC i0
Carbon balance
47.2 23.1 31.2 19.4 23.5 56.3 63.3 62.7 78.0 42.7 46.2 48.9
2.2 3.1 4.3 3.3 7.7 0.5 0.0 0.0 0.0 0.6 0 0
7.7 0.1 3.7 10.4 19.2 7.1 8.0 37.7 9.5 10.3 40.9 35.7
5.7 0.1 1.1 7.9 43.2 37.6 14.3 31.1 13.4 41.7 38.3 39.7
58.2 62.5 82.4 83.9 94.0 61.5 70.8 49.8 15.6 38.8 99.6 94.0
94.2 41.5 87.5 54.7 72.1 99.1 100 99.7 100 99.9 99.6 99.9
35.2 20.4 23.1 35.8 33.6 0.6 0.1 0.0 0.0 1.0 0.8 0.2
[a] The mathematical expressions of conversion, selectivities and carbon balance are detailed in the Supporting Information. [b] C3H8O and C5H6O2 are the selectivities to isopropanol and furfuryl alcohol, respectively; and SC8/Cio and SC13/Cio are the carbon yields to the formation of C8 and C13 adducts, respectively. More details about how are these selectivities were calculated are given in the Supporting Information.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 2816 – 2820
2817
CHEMSUSCHEM COMMUNICATIONS the Supporting Information (Figure S1). In both cases the selectivity for n-tridecane was almost negligible, but there were important differences between these materials. The results of Pt/ MgZr are conditioned by the condensation step, whereas this material shows a high hydrodeoxygenation activity, mainly for the C8 adduct (chemical structures of the primary condensation adducts and further hydrogenation and hydrodeoxygenation products are given as Supporting Information). The global selectivity for condensation adducts (C8 and C13) is lower than 14 % after 48 h, but more than 40 % of the C8 obtained was completely hydrogenated. The reaction was mainly limited by a fast hydrogenation of furfural (enhanced at high temperature), with more than 35 % of furfuryl alcohol selectivity at these conditions (Supporting Information, Figure S1). On the other hand, results obtained with the mechanical mixture are limited by the hydrogenation of reactants but also by the adsorption of reactants and reaction intermediates on the active carbon surface, leading to carbon balances lower than 63 % after 48 h. Because hydrogenation of reactants inhibits further condensation, the procedure has been improved, operating in two consecutive steps: the first one at 323 K in inert atmosphere (1 MPa N2), and the second one in hydrogen (4.5 MPa) and at 493 K. These results are summarized in Table 1, entry O2, whereas time profiles obtained at these conditions with Pt/ Al2O3 + MgZr are detailed as example in the Supporting Information (Figure S2). Concerning the bifunctional catalysts, and in a good agreement with the previous results for condensation reaction, the carbon-supported catalyst shows a better performance for the condensation reaction than the bulk metal oxide.[12] This better performance is caused by the more favorable distribution of basic sites when the oxides are dispersed onto an inert support. It should be noted that the best performance found in a previous work was observed in the aqueous phase, suggesting similar mechanisms in both phases. In addition, carbon-containing catalysts strongly interact with the C13 adduct, favoring the subsequent hydrodeoxygenation. Better condensation results were obtained with a mechanical mixture of basic and metal catalysts, with more than 56 % of C13 condensated adducts after 48 h reaction time. The most notable differences occurred during the hydrodeoxygenation step. Aldol condensation reactions are enhanced in the absence of metal active phases,[17] so the segregation between these sites favors that the C8-condensated adduct remains adsorbed on the basic site without undergoing hydrogenation. The high temperatures increase the ratio of condensation, obtaining more C13 adducts. Inorganic supports for metal particles lead to poor interaction with C13 adducts, hindering further hydrodeoxygenation. Consequently, in spite of the low activity, this configuration allows obtaining a higher yield of n-tridecane, as can be viewed in Figure 1. The same procedure was repeated using water as solvent instead of n-hexane. Profiles obtained with Pt/MgZr are shown in the Supporting Information (Figure S3). Both reactants (furfural and acetone) are soluble enough in water to perform the experiment in the aqueous phase. In general terms, the selec 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org
Figure 1. Comparison of final results obtained with different catalysts in organic solvent for the two-step process (24 h at condensation conditions and 24 h at hydrogenation conditions). Results are given in terms of the ratio between the carbon atoms in the resulting alkanes and the total content of carbon atoms of the reactants (%).
tivities for the condensation reactions increase in the aqueous phase, which is in a good agreement with literature findings,[18] suggesting that the enolate form of the acetone is solvated in the aqueous phase, enhancing its condensation reactions. The highest activity for the condensation reaction was obtained by using a mixture MgZr + Pt/AC as catalyst, reaching almost 70 % of C8 and C13 global selectivity. These reaction conditions also led to an improvement of the hydrodeoxygenation efficiency of the condensation adduct, this effect being more distinct when using the catalyst mixture and inorganic materials as metal support. Under these conditions, more than 50 % of each adduct was completely hydrogenated after 48 h. Considering both steps, the total amount of initial carbon that is transformed into n-octane and n-tridecane as function of the catalyst is shown in Figure 2. As can be observed, more than 12 % of alkanes were obtained with MgZr/HSAG500 + Pt/Al2O3, with a high selectivity of n-tridecane. These results contrast with the values of the carbon balances. Mass balance closure is markedly lower than in the same experiments performed in organic media. This can be attributed to the lower solubility of the condensation adducts in the aqueous phase, especially at the lower temperatures used for the aldol condensation. Indeed, the carbon balances get worse when using carbon-
Figure 2. Comparison of final results obtained with different catalysts in aqueous phase for the two-step process (24 h at condensation conditions and 24 h at hydrogenation conditions). Results are given in terms of the ratio between the carbon atoms in the resulting alkanes and the total content of carbon atoms of the reactants (%).
ChemSusChem 2014, 7, 2816 – 2820
2818
CHEMSUSCHEM COMMUNICATIONS supported catalysts (hydrophobic material), and are even worse when the support is microporous (higher adsorption capacity). These findings led us to exclude the use of carbon-based catalysts for this one-pot process, in spite of the good performance of these materials for each individual reaction.[12, 16] From the results obtained in the two-step processes, and literature findings about the highest reaction rates for the condensation in aqueous medium, a reaction scheme can be proposed, considering one-pot and one-step reactions performed in aqueous phase and at the conditions optimized for the hydrogenation reactions (493 K and 4.5 MPa). Two different catalysts were considered for this experiment: the bifunctional Pt/ MgZr and a mixture of bulk Mg-Zr oxide and Pt/Al2O3 catalyst. Temporal profiles obtained are shown in the Supporting Information (Figure S4). Our obtained results show that carbon balance closures are very high (close to 100 %) under these conditions, especially with the bifunctional catalyst (Table 1, entry A1). The highest average temperature of these experiments increases the solubility of these condensated adducts on water and hinders their adsorption. In spite of the higher temperature, condensation in the aqueous phase prevails over hydrogenation and neither isopropanol nor furfuryl alcohol was detected in these experiments. Thus, selectivities higher than 70 % for condensation adducts were observed for both catalytic systems. Major differences were observed in the hydrogenation of these adducts. In the case of the mechanical mixture, only n-tridecane was observed, obtaining an overall yield of 14.5 % for the formation of this alkane (Figure 3). However,
www.chemsuschem.org to those obtained in the two-steps process, the results are completely different, with a significant improvement when hydrogenation conditions are applied. This suggests that the solubility of these adducts in water plays a key role in the process, being this solubility favored at higher temperatures. In the experiment performed with the bifunctional catalyst (the best combination among the ones we tested), the total yield to n-alkanes was higher than 50 %, resulting in an nalkane mixture with a 66 % of n-tridecane concentration. This mixture is a diesel-range fuel, suitable for direct blending with a mineral diesel fraction. In conclusion, based on using model compounds the reported results show the possibility of transforming furfural and acetone (and eventually other biobased carbonylic compounds) into n-alkane mixtures in a one-pot process, integrating aldol condensation and total hydrodeoxygenation of the condensation adducts. The decrease in H2 pressure during the hydrodeoxygenation does not reach 5 bar, indicating that the reaction is not limited by the availability of H2. Furthermore, the optimal conditions for obtaining the highest n-alkane yields involve performing the process in the aqueous phase, enabling direct recovery of the n-alkanes because they are immiscible with water, thereby avoiding distillation or extraction procedures. In addition, the procedure only involves the use of heterogeneous catalysts, which are easy to separate and reuse. The yield of n-alkanes reported here, 50 %, is especially meritorious if it is compared to yields recently reviewed in the literature, 70 %, for heterogeneously catalyzed two-step processes, with each step independently optimized, and working at higher hydrogen pressures in the latter step.[18] This approach offers a clear advantage for such processes, and should be further validated with real biomass materials.
Experimental Section
Figure 3. Comparison of final results obtained with different catalysts in aqueous phase for the one step process (24 h at hydrogenation conditions). Results are given in terms of the ratio between the carbon atoms in the resulting alkanes and the total content of carbon atoms of the reactants (%).
almost null n-octane selectivity was measured. When the bifunctional catalyst is used, the selectivities for the formation of both alkanes largely increase, being the fraction of adduct completely hydrogenated a 40 and 85 % for C8 and C13 adducts, respectively. These results suggest that desorption of the condensation adducts from the base catalysts and the readsorption on the surface of the hydrogenation catalysts are slow enough to influence the selectivities. If the results obtained in the aqueous-phase one-step procedure are compared 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Reported experiments were performed in a 0.5 L stirred batch autoclave reactor (Autoclave Engineers EZE Seal). Two different solvents were used: distilled water (entries A1 and A2) and n-hexane (entries O1 and O2). Experimental conditions corresponded to a total loading of 5 % of organic compounds in solvent; specifically, 0.324 mol L 1 of acetone and 0.324 mol L 1 of furfural. Previous works noted that this is the optimal molar ratio for maximizing the yield to condensated adducts.[7] Samples were analyzed by GC-FID in a Shimadzu GC-2010, using a 15 m long CP-Sil 5 CB capillary column. Concerning to the catalysts, bifunctional catalysts with basic and metal functionalities (1 % Pt/MgZr, 1 % Pt/MgZr/ HSAG500), mechanical mixtures of basic catalysts (MgZr, MgZr/ HSAG500), and metal catalysts (0.5 % Pt/Al2O3, 1 % Pt/AC) were considered. Reactions were catalyzed by 0.1 g of bifunctional materials or with mechanical mixtures of 0.5 g of basic materials and 0.1 g of metal catalyst. Basic catalysts formulations were optimized for the aldol condensation of furfural–acetone mixtures in previous works,[7, 12] whereas platinum is the most active noble metal in the C8 hydrodeoxygenation.[16] All of the materials were characterized by different techniques (physisorption, TEM, XRD, and CO2-TPD), and characterization results are summarized in Table S1 in the Supporting Information. ChemSusChem 2014, 7, 2816 – 2820
2819
CHEMSUSCHEM COMMUNICATIONS Acknowledgements This work was supported by the Spanish Government (contract CTQ2011-29272-C04-02). L.F. thanks the Government of the Principality of Asturias for a Ph.D. fellowship (Severo Ochoa Program) Keywords: aldol reaction · biofuels · C C coupling · heterogeneous catalysis · hydrodeoxygenation [1] a) A. Demirbas, Appl. Energy 2011, 88, 17 – 28; b) A. L. Villanueva Perales, C. Reyes Valle, P. Ollero, A. Gmez-Barea, Energy 2011, 36, 4097 – 4108. [2] a) H. Michel, Angew. Chem. Int. Ed. 2012, 51, 2516 – 2518; Angew. Chem. 2012, 124, 2566 – 2566; b) S. G. Wettstein, D. M. Alonso, E. I. Grbz, J. A. Dumesic, Curr. Opin. Chem. Eng. 2012, 1, 218 – 224. [3] a) A. V. Subrahmanyam, S. Thayumanavan, G. W. Huber, ChemSusChem 2010, 3, 1158 – 1161; b) E. Furimsky, Catal. Today 2013, 217, 13 – 56. [4] G. W. Huber, J. N. Chheda, C. J. Barrett, J. A. Dumesic, Science 2005, 308, 1446 – 1450. [5] J. N. Chheda, Y. Romn-Leshkov, J. A. Dumesic, Green Chem. 2007, 9, 342 – 350. [6] R. Sahu, P. L. Dhepe, ChemSusChem 2012, 5, 751 – 761. [7] L. Faba, E. Daz, S. OrdÇez, Appl. Catal. B 2012, 113, 201 – 211. [8] L. Faba, E. Daz, S. OrdÇez, Appl. Catal. B 2014, 160 – 161, 436 – 444.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemsuschem.org [9] a) S. V. Vassilev, D. Baxter, L. K. Andersen, C. G. Vassileva, T. J. Mogan, Fuel 2012, 94, 1 – 33; b) R. Xing, A. V. Subrahmanyam, H. Olcay, W. Qi, G. P. van Walsum, H. Pendse, G. W. Huber, Green Chem. 2010, 12, 1933 – 1946. [10] J. Q. Bond, A. A. Upadhye, H. Olcay, G. A. Tompsett, J. Jae, R. Xing, D. M. Alonso, D. Wang, T. Zhang, R. Kumar, A. Foster, S. M. Sen, C. T. Maravelias, R. Malina, S. R. H. Barrett, R. Lobo, C. E. Wyman, J. A. Dumesic, G. W. Huber, Energy Environ. Sci. 2014, 7, 1500 – 1523. [11] a) L. He, D. Chen, ChemSusChem 2010, 3, 1169 – 1171; b) D. M. Alonso, J. Q. Bond, J. A. Dumesic, Green Chem. 2010, 12, 1493 – 1513. [12] L. Faba, E. Daz, S. OrdÇez, ChemSusChem 2013, 6, 463 – 473. [13] a) C. J. Barrett, J. N. Chhedda, G. W. Huber, J. A. Dumesic, Appl. Catal. B 2006, 66, 111 – 118; b) L. Faba, E. Daz, S. OrdÇez, Biomass Bioenergy 2013, 56, 592 – 599. [14] S. Abell, F. Medina, D. Tichit, J. Prez-Ramrez, J. E. Sueiras, P. Salagre, Y. Cesteros, Appl. Catal. B 2007, 70, 597 – 605. [15] A. Corma, O. Torre, M. Renz, N. Villandier, Angew. Chem. 2011, 123, 2423 – 2426. [16] R. D. Cortright, R. R. Davda, J. A. Dumesic, Nature 2002, 418, 964 – 967. [17] J. N. Chheda, J. A. Dumesic, Catal. Today 2007, 123, 59 – 70. [18] M. J. Climent, A. Corma, S. Iborra, Green Chem. 2014, 16, 516 – 547.
Received: March 27, 2014 Published online on August 1, 2014
ChemSusChem 2014, 7, 2816 – 2820
2820
