Base-Catalyzed Condensation of Levulinic Acid: A

DOI: 10.1002/cctc.201600064

Communications

Base-Catalyzed Condensation of Levulinic Acid: A New Biorefinery Upgrading Approach Laura Faba, Eva Daz, and Salvador OrdûÇez*[a] The base-catalyzed condensation of levulinic acid (LA) under mild conditions (0.1 MPa, 323 K, 0.45 mol L¢1 aq solution of levulinic acid, catalyst/reactant = 1:24 wt.) was proposed as a new approach for the upgrading of this molecule by using mixed oxides as catalysts. Three products were identified: a-angelica lactone (AL) and two aldol condensation adducts, LA–LA and AL–LA. These last ones (obtained with selectivities higher than 90 % at 30 % conversion upon using MgZr) present high value as fuel and chemical precursors. The distribution of basic and acidic sites on the catalyst governs the performance and selectivity of the catalyst, whereas the stability of the catalyst in the reaction media determines the deactivation behavior. Bulk MgZr oxides present a good balance of these properties.

Transformation of biomass into highly valuable products and liquid fuels has received much attention in the last years. Levulinic acid (4-oxopentanoic acid, LA) is nowadays considered one of the most promising bioplatform molecules. This keto acid is obtained in significant yields (> 80 %) by the aqueousphase hydrolysis of cellulosic feedstock if the hydrolytic conditions are acidic enough to yield levulinic acid and formic acid by fragmentation of dehydrated sugars.[1] These conditions are industrially implemented by the BIOFINE process.[2] LA has been proposed as a reactant in different processes for the production of advanced biofuels, monomers, as well as other molecules with industrial interest.[3] Most of these processes involve the transformation of levulinic acid into g-valerolactone.[4, 5] However, this reaction requires the presence of molecular hydrogen and severe conditions (473 K and 3 MPa), which hinder both energy trade-off and process sustainability. Consequently, different improvements, such as using formic acid as the hydrogen source, have been explored.[6] Other acid-catalyzed reactions are proposed, including esterification with different alcohols catalyzed by homogeneous acids such as HCl, H2SO4, and H3PO4[7] or basic salts such as K2CO3 at temperatures over 450 K,[8] not fulfilling the “green chemistry” principles. Most of these previously studied reactions are focused on the reactivity of the carboxyl group of the levulinic acid; references dealing with the reactivity of the carbonyl group are [a] Dr. L. Faba, Dr. E. Daz, Dr. S. OrdûÇez Department of Chemical and Environmental Engineering University of Oviedo C/Julin Clavera s/n, 33006 Oviedo (Spain) E-mail: [email protected] Supporting Information and the ORCID identification number(s) for the author(s) of this article can be found under http://dx.doi.org/10.1002/ cctc.201600064.

ChemCatChem 2016, 8, 1490 – 1494

very scarce. In this framework, aldol condensation is one of the most common reactions involving this functional group. This reaction increases the length of the carbon chain by a basecatalyzed mechanism that begins with abstraction of a proton a to the carbonyl functional group. This reaction has already been proposed to upgrade bioplatform molecules, such as acetone, 5-hydroxymethylfurfural, and furfural.[9–12] Although acetone self-condensation is often studied in the gas phase,[9, 13] most aldol condensations are performed in the aqueous phase, at low pressure, and mild temperatures,[10–12] and it leads to sustainable and environmentally benign biorefinery processes. Despite the fact that the first references about these reactions propose homogeneous catalysts such as NaOH and KOH,[14] there have been important advances in the development of active, selective, and stable solid basic catalysts, such as mixed oxides, aluminophosphate, and carbon materials.[9–12] To the best of our knowledge, the aldol condensation of levulinic acid has not yet been reported. It is expected that the aldol condensation of levulinic acid would yield a mixture of C10–C15 adducts. These adducts could have high value as fuel additives or surfactants, and hydrodeoxygenation of these compounds would result in a good-quality liquid fuel without needing further purification steps. The scarce amount of information on the condensation of this compound is related to its dehydration and subsequent dimerization[15] in addition to its cross-condensation with furfural or 5-hydroxymethylfurfural.[16] In the latter case, good activity has been obtained upon using water as the solvent, whereas condensation in organic solvents has not been observed. In this work, NMR spectroscopy was used to corroborate the assumption that abstraction of the proton a to the aldehyde group, which would allow the subsequent aldol condensation, was feasible under the reaction conditions. The aqueous-phase aldol condensation of levulinic acid catalyzed by mixed oxides was studied in this work. These materials were previously tested in the cross-condensation of furfural and acetone[11, 17] with significant selectivities and conversions. The activity results obtained were analyzed in terms of the physicochemical properties of the catalysts. Finally, a reaction mechanism and kinetic model were proposed for this reaction. The aldol condensation of LA was studied in a batch reactor at 323 K and 0.1 MPa (200 mL of a 0.45 mol L¢1 aq solution of LA, and 0.5 g of catalyst). Three different compounds were identified by LC–MS, and the corresponding spectra can be found in the Supporting Information. The first identified compound was angelica lactone (AL), obtained by intramolecular cyclization of levulinic acid. The presence of a secondary signal at m/z = 98 in the mass spectrum of another reaction product

1490

Ó 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications (with main signal at m/z = 196) corroborates the reaction between a-AL and LA, which resulted in a condensation product with 10 carbon atoms and a furanic ring, labeled AL–LA. The carboxylic group of this molecule is corroborated by secondary peaks corresponding to interactions with cations produced by electrospray ionization (Na + and NH4 + ). The existence of this group in the 10-carbon-atom molecule suggests that this compound is obtained by aldol condensation and not by ketonization. The condensation reactivity of angelica lactone under mild conditions over carbonate salts and other homogeneous catalysts has been previously reported.[18] On the other hand, the direct condensation of two molecules of levulinic acid is also detected by LC–MS. In this case, signals detected in this peak indicate the presence of the levulinic acid structure (linear and complete), and the intensity of the carboxylic groups is two times higher than that for the other compounds, which suggests that this molecule has double the amount of this group. This product is labeled 2 LA. The complete structures and names of all these compounds are summarized in Scheme 1.

Figure 1. Concentration profiles of reactant and products in the aldol condensation of levulinic acid at 323 K by using a) MgZr and b) MgAl as catalysts. Levulinic acid: +; 2 LA: *; AL: &; AL–LA: ~. Dashed lines correspond to the kinetic model predictions

Scheme 1. Chemical structures and full names of compounds detected in the condensation of levulinic aldol catalyzed by mixed oxides at 323 K.

The presence of different compounds suggests that the condensation of levulinic acid is a complex process involving different active sites and, subsequently, different adsorption modes of the levulinic acid on the surface of the catalyst. To identify the relevance of each step of the process, two different mixed oxides (i.e., MgZr and MgAl) with different distributions of acidic and basic sites were tested. The evolution of the concentrations of the reactant and reaction products with reaction time is shown in Figure 1. Upon using MgZr as the catalyst (Figure 1 a), more than 33 % conversion of levulinic acid was obtained, with final selectivities of 77.8, 12.3, and 9.8 % for 2 LA, a-AL, and AL–LA, respectively. In the case of MgAl (Figure 1 b), the conversion of levulinic acid only reached a value of 9.3 %, with final selectivities of 69.6, 18.3, and 12.7 % for 2 LA, a-AL, and AL–LA, respectively. Good carbon balances were obtained in both cases, with final values of 92 and 95 % for MgZr and MgAl, respectively. Considering that no side products were detected in any case and that the values were almost constant from the start of the reaction (the carbon balChemCatChem 2016, 8, 1490 – 1494

www.chemcatchem.org

ance decreased from 100 % to the final value in less than 2 h in both cases), the small deviation in the carbon balance could be caused by strong adsorption of either the reactant or the reaction intermediates on the surface of the catalyst. The concentration profiles of the reactants and products recorded for both catalysts are useful to obtain information about the reaction mechanism. Analyzing the activity results as a function of the characterization parameters (summarized in Table 1 and detailed in our previous work),[9] both acidic and basic sites can have an important role in the process. Despite the fact that mixed oxides are generally considered basic materials, the presence of medium-strength basic sites is always related to the presence of weak acidic sites (acid–base pairs). Thus, they can also catalyze reactions requiring weak-medium acidic sites. According to the obtained profiles, a-AL and 2 LA follow the typical evolution of primary products, with a fast in-

1491

Table 1. Distribution of acidic and basic sites, as measured by temperature programmed desorption analyses.[9, 13]

Catalyst

Basicity [mmolCO2 g¢1] Weak Medium

MgAl MgZr MgZr/HSAG

59.1 0 10.1

79.4 120.1 13.5

Strong

Acidity [mmolNH3 g¢1] Weak Medium

Strong

33.5 13.3 20.2

– 0.8 1.5

0.02 – 0.1

0.08 0.95 –


Communications crease at low conversions. In the case of a-AL, the maximum was obtained after 4-6 h with both materials. This maximum concentration reached a higher value if MgAl was used as the catalyst, which is in good agreement with the higher concentration of the weak and medium-strength acidic sites of this material (Table 1). After this 6 h, the concentration of a-AL remained constant if MgZr was used, whereas it decreased in presence of MgAl. The formation of 2 LA follows the typical trend of a primary product formed in a reversible reaction. This reversibility is more evident in the case of MgAl, for which an almost constant concentration was obtained after a reaction time of 5 h. In the case of MgZr, the increase in the concentration was slower after 4 h, but an increase was still observed even after 10 h, and a final value almost three times higher than that obtained in the case of MgAl was reached. These different behaviors can be explained by considering the active sites that catalyze the aldolization and the retro-aldolization reactions. It was previously reported, for reactions of other bioplatform molecules, that the aldol condensation is catalyzed by good equilibration of the medium-strength basic sites and the mediumstrength acidic sites.[11] This distribution is more favorable in the case of MgZr (Table 1). On the other hand, the retro-aldolization is favored if MgAl is used, because of the high concentration of acidic sites, which stabilizes the adsorption of condensation intermediates and enhances the inverse reaction.[11] Finally, the second condensation product, AL–LA, evolves with a typical profile of a final product: its concentration increases in the last stages of the reaction at the same time as the concentration of its precursor (i.e., AL) decreases. Considering these premises, the proposed mechanism is detailed in Scheme 2. On the one hand, molecules of levulinic acid are transformed into 4 hydroxypenten-3-enoic acid by a keto–enol

Scheme 2. Single steps in the proposed mechanism for the aldol condensation of levulinic acid.

ChemCatChem 2016, 8, 1490 – 1494

www.chemcatchem.org

equilibrium. The second OH group allows intramolecular cyclization to form an ether bond followed by dehydration, and this results in the formation of a-AL. This reaction was previously observed as a side reaction in the aqueous-phase hydrolysis–dehydration of sugars upon using strong homogeneous acid catalysts.[15] The direct condensation of two molecules of levulinic acid is explained by considering the fact that this acid is strong enough to be dissociated into a proton and a levulinate anion in the aqueous phase. The excess amount H + in the aqueous phase enhances the intramolecular cyclization because of the acid media obtained. The presence of mediumstrength basic sites promotes abstraction of the a proton to yield a carbanion. This enolate attacks the carbonyl group of the contiguously adsorbed levulinate anion. The b-hydroxy ketone obtained is so unstable that it is not observed, and it undergoes fast dehydration in the presence of weak acids to yield 2 LA. The presence of AL–LA can only be justified by enolization of the a-AL adduct and subsequent interaction with the carbonyl group of the levulinate anion. Given that this step involves participation of a primary reaction product, yields of AL–LA are markedly lower than those corresponding to the 2 LA adduct, despite the fact that the a-proton of AL is more acidic than the equivalent hydrogen atom of LA. The profiles obtained suggest that longer reaction times would have a positive effect on the reaction. Reactions with both catalysts were performed under the same conditions but the time was increased from 24 to 72 h. Comparing the results obtained after 24 and 72 h, it was concluded that MgAl was completely deactivated, without any improvement in the conversion (9 and 10 % after 24 and 72 h, respectively). This deactivation is justified by the interaction of levulinic acid with both metals, which leads to catalyst leaching by the formation of soluble salts. In good agreement with this fact, the weight of the catalyst recovered after the reaction was 47 % that of the initial catalyst loading. On the contrary, upon using MgZr, this deactivation was not observed, and the conversion of LA increased upon increasing the reaction time (33 and 42 % after 24 and 72 h, respectively), and all the solid catalyst was recovered after 72 h. In both cases, the carbon balance remained almost constant at 24 and 72 h (92 and 84 % with MgAl and MgZr, respectively), discarding other side reactions and corroborating the hypothesis of the initial adsorption of levulinic acid on the surface of the catalyst. These differences in terms of catalytic stability can be justified by considering the different crystalline structures of the catalysts. According to previously reported XRD results,[11] MgZr is more crystalline than MgAl, with crystalline phases involving both cations. However, no peaks related to Al2O3 were observed in MgAl, and most of this metal is located in the bulk material and not in the surface (corroborated by comparing the XRD and inductively coupled plasma MS results). Consequently, the MgO phases are more exposed to the aqueous environment in the MgAl catalyst, and this leads to faster leaching of the active phase. Despite the fact that no lixiviation was observed with MgZr, the activity of this material was also modified with time. After 72 h, the whole selectivity of C10 compounds decreased from 82.3 to 69.3 %, and a mixture with a lower amount of linear dimer

1492


Communications (from 69.6 to 34.6 %) was obtained, which suggests that longer times can have a negative effect on the product distribution. To optimize the catalytic results, the effect of supporting the mixed oxide over mesoporous materials (high surface area graphite, HSAG) was also studied. It has been demonstrated that supporting the active phase on an inert material enhances the dispersion of the active sites, which increases the number of possible interactions between the reactant and the active sites. This support was chosen because of the good results previously obtained in other aldol condensation reactions, for which condensation reaction rates four times higher than those obtained with the bulk material were observed.[16] Results obtained for the condensation of LA with MgZr/HSAG are depicted in Figure 2. A reaction with the use of only the sup-

ces by using the Scientist software for solving the resulting ordinary differential equations (ODE) system. The best results were obtained by considering that the cyclization of LA (k1) and the formation of the self-condensation adduct of levulinic acid (k3) take place as two parallel routes, whereas the AL–LA adduct is obtained as a secondary product (k2). The two condensation steps present first-order reaction kinetics for the species that is able to form the carbanion by a-H abstraction (i.e., levulinic acid), in good agreement with the findings reported in the literature for the kinetics of aldol condensations.[19] Cyclization of LA was considered as an irreversible first-order reaction (k1). The retro-aldolization was also considered, as was as a side reaction for the permanent adsorption of levulinic acid (kD), which can explain the decrease in the carbon balance (92, 95, and 87 % with MgZr, MgAl, and MgZr/ HSAG, respectively). The resulting kinetic model is sketched in Scheme 1. The kinetic data are summarized in Table 2, whereas

Table 2. Kinetic constants for the fitting of the experimental results to the kinetic model.[a]

Catalyst

k1 [h¢1]

k2 [h¢1]

k3 [h¢1]

k¢3 [h¢1]

kD [h¢1]

r2 [–]

MgAl 3.9 Õ 10¢2 5.8 Õ 10¢3 4.7 Õ 10¢3 8.0 Õ 10¢1 1.4 Õ 10¢1 0.9994 MgZr 1.0 Õ 10¢1 4.4 Õ 10¢2 1.6 Õ 10¢2 1.5 Õ 10¢1 1.1 Õ 10¢1 0.992 MgZr/HSAG 2.1 Õ 10¢2 3.6 Õ 10¢5 3.8 Õ 10¢3 9.3 Õ 10¢1 1.5 0.98 [a] See Scheme 1 to identify the constants.

Figure 2. Concentration profiles of reactant and products in the aldol condensation of levulinic acid at 323 K by using MgZr/HSAG as the catalyst. Levulinic acid: +; 2 LA: *; AL: &; AL–LA: ~. Dashed lines correspond to the kinetic model predictions

port as the catalyst (without MgZr) was performed, and no conversion was observed. This result corroborates the fact that the support does not have any active sites that could modify the reaction mechanism, and so results obtained for a reaction performed with the supported catalyst could only be justified by the different dispersion of the MgZr active sites. As expected, the temporal profiles were similar to those observed with the bulk material. The linear condensation adduct was the main product, and significant amounts of AL were also detected. The final conversion was 11 %, which is two times lower than that obtained for the MgZr catalyst; furthermore, the formation of AL–LA was not observed. These results highlight that no activity was observed after 6 h, which suggests either a high influence of the retro-aldolization equilibrium or fast deactivation of the catalyst because of the permanent adsorption of the reaction compounds on the catalyst surface. The concentration profiles obtained for all the catalysts were fitted to different kinetic models. A stirred batch reactor was considered ideal for these purposes. All the diffusional limitations were discarded by controlling the stirring rate and the particle size of the catalyst, so kinetic control was considered. The temporal profiles of the concentrations of all the compounds were fitted by considering unsteady-state mass balanChemCatChem 2016, 8, 1490 – 1494

www.chemcatchem.org

the ability of the model to predict the evolution of the different compounds is shown in the trend lines of Figures 1 and 2. According to the kinetic values, MgZr oxide has the highest values for all the direct reactions, which justifies the higher concentrations of the reaction products obtained with this material. The high influence of the retro-aldolization step upon using MgAl or the supported catalyst was also observed. In the case of MgAl, this reaction is justified by the acid–basic properties, whereas in the case of MgZr/HSAG, high interaction between C=C and the carbonaceous support must also be taken into account.[17] The hypotheses considered in the reaction results section can be corroborated by the dependence of these kinetic constants on the concentration of medium-strength basic sites. As observed in Figure 3, the value of k1 is strongly dependent on this parameter, as well as the other two direct steps. However, there is an opposite effect if the retro-aldolization is considered. In conclusion, fuel precursors and additives can be obtained by the aqueous-phase aldol condensation of levulinic acid (LA) under mild conditions and by using mixed oxides as catalysts. Two different condensation adducts were identified: the first one corresponded to the self-condensation adduct of LA, and the second one corresponded to the condensation of LA with the lactone obtained for the cyclization of LA. The best results were obtained with MgZr, for which a 33 % conversion of levulinic acid was observed after 24 h with almost 83 % selectivity for the formation of condensation products. Medium-strength

1493


Communications fAL ¼

5 ½AL¤ 5 ½AL¤ þ 10 ½AL ¢ LA¤ þ 10 ½2AL¤

fAL¢LA ¼ f2LA ¼ CB ¼

basic sites were identified as the key parameter of this reaction. A parallel mechanism was proposed by assuming a kinetic model with first-order dependence on the involved organic compounds.

Experimental Section Mixed oxides were prepared by following optimized methodologies to obtain materials with high activity in aldol condensations: sol–gel technique for MgZr and co-precipitation under low supersaturation conditions for MgAl. In both cases, the nitrate salts were used, and precipitation of the mixed oxides was obtained by increasing the pH to 10. Materials were filtered, washed, dried in an oven, and calcined in He flow to 873 and 923 K for MgZr and MgAl, respectively. Further details about the preparation procedures are reported elsewhere.[11, 16] The surface, morphological, and physicochemical properties of these materials were characterized by different techniques, including N2 physisorption, XRD, X-ray photoelectron spectroscopy, and temperature programmed desorption of CO2 and NH3. Characterization results are discussed in previous papers, but the main results are summarized in Table 1 to identify the influence of the chemical properties of the catalysts on their activity.[9, 11, 16] The aldol condensation of levulinic acid was performed in a 1 L, double-walled, thermostatically temperature-controlled glass reactor. Levulinic acid (12 g) was dissolved in water (250 mL). The resulting mixture was stirred until it reached the reaction temperature, 323 K. The catalyst (0.5 g) was then introduced into the reactor, and this moment was considered the starting point of the reaction. Liquid samples were collected for 24 h and analyzed by capillary gas chromatography with a Shimadzu GC-2010 equipped with a flame ionization detector by using a 15 m long CP-Sil 5 CB capillary column as the stationary phase. Peak assignment was performed by liquid chromatography with a mass spectrometer detector by using a LC–MS Agilent 6460. Relative responses were calculated by using commercial standards for levulinic acid (Aldrich, > 98 %) and angelica lactone (Sigma–Aldrich, > 98 %) and the effective carbon number concept for the dimers.[20] Quantitative results were used to calculate the conversion, product selectivities (denoted f in the following equations), and carbon balance (CB) closure by calculating this last parameter according to Equations (1)–(4). The reported values are averages of two analyses, and a relative error lower than 5 % was obtained in all cases.

www.chemcatchem.org

10 ½2LA¤ 5 ½AL¤ þ 10 ½AL ¢ LA¤ þ 10 ½2AL¤

5 ½LA¤ þ 5 ½AL¤ þ 10 ½AL ¢ LA¤ þ 10 ½2AL¤ 5 ½LA¤0

ð2Þ ð3Þ ð4Þ

Acknowledgements

Figure 3. Evolution of the kinetic constants as a function of the concentration of medium-strength basic sites. k1: ^; k2 : ~; k3 : &; k¢3 : *.

ChemCatChem 2016, 8, 1490 – 1494

10 ½AL ¢ LA¤ 5 ½AL¤ þ 10 ½AL ¢ LA¤ þ 10 ½2AL¤

ð1Þ

This work was supported by the Spanish Ministry of Economy and Competitiveness (contract CTQ2014-52956-C3-1-R; BIODEACT PROJECT). Keywords: aldol reaction · biofuels · C¢C coupling · heterogeneous catalysis · platform molecules [1] a) Y. D. Dwivedi, K. Gupta, D. Tyagi, R. K. Rai, S. M. Mobin, S. K. Singh, ChemCatChem 2015, 7, 4050 – 4058; b) B. Girisuta, L. P. B. M. Janssen, H. J. Heeres, Ind. Eng. Chem. Res. 2007, 46, 1696 – 1708; c) Y. Zuo, Y. Zhang, Y. Fu, ChemCatChem 2014, 6, 753 – 757. [2] S. F. Daniel, J. Hayes, Michael H. B. Hayes, Julian R. H. Ross in Biorefineries—Industrial Processes and Products, Vol. 1, Wiley-VCH, Weinheim, 2006, pp. 139 – 162. [3] G. Novodrszki, N. R¦tfalvi, G. Dibû, P. Mizsey, E. Cs¦falvay, L. T. Mika, RSC Adv. 2014, 4, 2081 – 2088. [4] a) A. Dutta Chowdhury, R. Jackstell, M. Beller, ChemCatChem 2014, 6, 3360 – 3365; b) E. F. Mai, M. A. Machado, T. E. Davies, J. A. LûpezSnchez, V. Teixeira, Green Chem. 2014, 16, 4092 – 4097. [5] J. Zhang, J. Chen, Y. Guo, L. Chen, ACS Sustainable Chem. Eng. 2015, 3, 1708 – 1714. [6] P. P. Upare, M. G. Jeong, Y. K. Hwang, D. H. Kim, Y. D. Kim, D. W. Hwang, U. H. Lee, J. S. Chang, Appl. Catal. A 2015, 491, 127 – 135. [7] E. Karimi, I. F. Teixeira, L. P. Ribeiro, A. Gûmez, R. M. Lago, G. Penner, S. W. Kycia, M. Schalf, Catal. Today 2012, 190, 73 – 88. [8] A. Caretto, A. Perosa, ACS Sustainable Chem. Eng. 2013, 1, 989 – 994. [9] L. Faba, E. Daz, S. OrdûÇez, Appl. Catal. B 2013, 142, 387 – 395. [10] K. Pupovac, R. Palkolvits, ChemSusChem 2013, 6, 2103 – 2110. [11] L. Faba, E. Daz, S. OrdûÇez, Appl. Catal. B 2012, 113, 201 – 211. [12] A. Bohre, S. Dutta, B. Saha, M. M. Abu-Omar, ACS Sustainable Chem. Eng. 2015, 3(7), 1263 – 1277. [13] A. A. Nikolopoulos, B. W. L. Jang, J. J. Spivey, Appl. Catal. A 2005, 296, 128 – 136. [14] M. Vashishtha, M. Mishra, D. O. Shah, Appl. Catal. A 2013, 466, 38 – 44. [15] R. W. Blessing, L. Petrus, Process for dimerization of levulinic acid, where organic phase comprises levulinic acid that is contacted in the presence of hydrogen with a strong acidic heterogeneous catalyst comprising a hydrogenating metal, WO 2006056591A1. [16] A. S. Amarasekara, T. B. Singh, E. Larkin, M. A. Hasan, H. J. Fan, Ind. Crops Prod. 2015, 65, 546 – 549. [17] L. Faba, E. Daz, S. OrdûÇez, ChemSusChem 2013, 6, 463 – 473. [18] J. Xin, S. Zhang, D. Yan, O. Ayodel, X. Lu, J. Wang, Green Chem. 2015, 17, 1065 – 1070. [19] S. Abellû, F. Medina, D. Tichit, J. P¦rez-Ramrez, J. E. Sueiras, P. Salgare, Y. Cesteros, Appl. Catal. B 2007, 70, 577 – 584. [20] J. T. Scanlon, D. E. Willis, J. Chromatogr. Sci. 1985, 23, 333 – 340. Received: January 19, 2016 Published online on March 23, 2016

1494
