Palladium supported on clays to catalytic deoxygenation of soybean fatty acids

Research paper

Palladium supported on clays to catalytic deoxygenation of soybean

fatty acids

Chaline Detoni

⁎

, Francine Bertella

, Mariana M.V.M. Souza

, Sibele B.C. Pergher

, Donato A.G. Aranda

a

Escola de Química, Universidade Federal do Rio de Janeiro, Cidade Universitária CT Bloco E, Rio de Janeiro, RJ 21945-970, Brazil

b_{Instituto de Química, Universidade Federal do Rio Grande do Norte, Centro de Ciências Exatas, Av. Salgado Filho, 3000, Lagoa Nova, Natal, RN 59078-970, Brazil}

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 12 November 2013 Received in revised form 20 April 2014 Accepted 24 April 2014 Available online xxxx Keywords: Deoxygenation Palladium Clays Hydrocarbons i-Alkanes

This work aims at the soybean free fatty acid deoxygenation using palladium supported (1 wt.%) on different clays as catalysts. The results presented a promising technology for the single-stage production of hydrocarbons in the diesel fuel range. Clays used as palladium support were a natural Brazilian Montmorillonite (BM), this same clay in its pillared form (PILCBM) and two commercial clays (K10 and KSF). Catalysts were characterized by N2adsorption, X-ray diffraction (XRD), FTIR spectra of adsorbed pyridine, CO chemisorption and scanning

electron microscopy (SEM). Reactions were carried out in batch mode, under different H2pressures (10, 20, 30

and 40 bar) at 300 °C. Reactions performed using Pd/K10 as catalyst at 30 bar of H2presented interesting results:

74.5% conversion after 6 h and selectivities to n- and i-alkanes equal to 69% and 29%, respectively.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

The large increase in energy consumption in the recent decades and the growing environmental awareness have turned renewable fuels to be an attractive alternative (Ping et al., 2011). Considerable efforts have been made to develop clean and renewable fuel technologies in order to secure the world energy reserves and gain environmental ben-efits (Lestari et al., 2009). Although biodiesel is a well-established addi-tive to mineral diesel fuel, its use is associated with a number of specific problems. Consequently, there is a need to develop new and improved methods for producing motor fuels from natural raw materials (Kikhtyanin et al., 2010). A technology for converting free fatty acids in hydrocarbons in the diesel range, based on the deoxygenation reaction, has been recently developed (Kubičková et al., 2005; Mäki-Arvela et al., 2007, 2008; Rozmyszowicz et al., 2012; Snåre et al., 2006, 2007). Howev-er, despite the fact that n-paraf_{finic compounds are ideal components} for the mixture with petroleum diesel, due to its high cetane number and environmental benefits, long chain alkanes present relatively high melting points, affecting negatively theflow properties of the fuel. In this way, catalytic reactions that could produce diesel fuel with good flow properties without additives and in a single stage reaction become interesting from economic and environmental point of view. Bifunction-al metBifunction-al-containing catBifunction-alysts could be an interesting option, since these materials possess both metal active sites, which are responsible for

deoxygenating free fatty acids, and acid sites that may perform the sub-sequent conversion of the obtained n-alkanes into i-alkanes.

Clay minerals are among the world's most important and useful in-dustrial materials. Catalysts based on clays have been used in a wide va-riety of chemical reactions for many years. Natural untreated clays have a very low ability to catalyze reactions in either polar or nonpolar media. However, the structural properties of these materials can be modified by various activation methods in order to produce catalysts with high acidity, surface area, porosity, and thermal stability (Moronta et al., 2005). These materials are relatively inexpensive and may be produced on large scales primarily because the basic ingredient used in produc-tion is readily available from natural sources (Paiva et al., 2008). These reasons demonstrate the immense potential of the clay materials in ca-talysis. According to the exposed, four different clays were used as sup-port for palladium catalysts, which were tested in free fatty acid deoxygenation, in a solvent free system and using as feedstock a mix-ture of free fatty acids obtained from non-catalytic hydrolysis of the degummed soybean oil. Results presented in this work are promising since they enable the use of acidic clay materials as catalysts to produce high quality renewable diesel fuel.

2. Experimental 2.1. Catalyst preparation

The four different clays used as support for palladium catalysts were: a natural Brazilian Montmorillonite (BM), this same clay pillared (PILCBM) and two acidic commercial clays (KSF and K10).

Applied Clay Science xxx (2014) xxx–xxx

⁎ Corresponding author at: ITMC-RWTH Aachen University, Worringerweg 1, 52074 Aachen, Germany.

E-mail address:ch_detoni@yahoo.com.br(C. Detoni).

http://dx.doi.org/10.1016/j.clay.2014.04.026

0169-1317/© 2014 Elsevier B.V. All rights reserved.

Contents lists available atScienceDirect

Applied Clay Science

2.1.1. Pillared clay preparation

As mentioned previously, the starting material used to prepare the pillared clay was a Brazilian Montmorillonite (BM). The Al_– PILCBM was synthesized by intercalating the polyoxocations pre-pared in a pillaring solution into the interlamellar space of the raw montmorillonite. This pillaring solution was prepared by slow addition of a 0.6 mol·L− 1 NaOH solution to another 0.6 mol·L−1AlCl3solution under constant stirring for 2 h. Then, the oligomeric solution prepared was aged for 6 days at room temperature. Finally, the aged solution was slowly added to a sus-pension of BM in deionized water. The exchange process was car-ried out at room temperature for 2 h under constant stirring. The resulting Al–PILCBM was separated by filtration and washed with deionized water until pH = 7. The solid obtained was dried at 80 °C for 12 h and calcined for 30 min at 150 °C and 3 h at 450 °C. This pillaring process followed the methodology described

byPergher and Sprung (2005).

2.1.2. Palladium impregnation

Palladium was introduced in the clays by incipient wetness impreg-nation, as described byCañizares et al. (1998). A specific quantity of palladium(II) acetate– Pd(OAc)2– solution in toluene was added to a pretreated support (450 °C for 6 h— 2.5 °C·min−1_{). The quantity} used, equal to the pore volume of the support, was calculated to provide approximately 1 wt.% of palladium. The solvent was removed by evap-oration. The samples were dried at 100 °C for 12 h, and treated at 450 °C (2.5 °C·min−1) for 6 h.

2.2. Catalyst characterization

X-ray powder diffraction (XRD) patterns were recorded in a Rigaku Miniflex II Diffractometer with a monochromator, using CuKα radiation (40 kV and 40 mA) in the range 3b 2θ b 90°. The stepsize was 0.05° with 2 s by step.

Textural characteristics, such as BET specific surface area, pore vol-ume and average pore diameter (BJH method) were determined by N2 adsorption–desorption at −196 °C in a Micromeritics Tristar 3000. Prior to the analysis, samples were outgassed for 24 h at 200 °C to elim-inate the physically adsorbed moisture.

Chemical composition of the materials was determined by X-ray fluorescence (XRF) using a Bruker spectrometer (AXS S4 Explorer). SEM analyses were performed in a MEV— FEI Quanta 200 operating at 20 kV.

The clay acidity was determined by the pyridine adsorption –desorp-tion method, monitored by FT-IR spectroscopy (Spectrum 100 FT-IR Spectrometer). A thin wafer of the pure sample (around 30 mg) was placed in a quartz infrared cell for in situ pyridine adsorption and desorp-tion experiments. The wafer was heated under vacuum (10−4Torr) at 450 °C for 1 h and, after cooling to room temperature, an FTIR spectrum of the vacuum treated sample was recorded as a background spectrum. Adsorption of pyridine on the sample was then carried out at 150 °C for 1 h. Desorption of pyridine was performed under vacuum at different temperatures (200 °C, 300 °C, 400 °C) and the IR spectra was measured at 25 °C. The concentrations of the Brönsted acid sites (BAS) and the Lewis acid sites (LAS) were calculated from the integration of the 1540 and 1450 cm−1intensity bands, and the corresponding molar extinction coefficients used were ε1540 cm−1= 0.059 cm2·mmol−1andε_{1450 cm}−1=

0.084 cm2_·mmol−1_{(Datka, 1981). The results were obtained by the} Lambert–Beer equation.

CO chemisorption analyses were obtained at room temperature using an Autochem 2910 Micromeritics by the CO pulse chemisorp-tion method. The catalysts were reduced in situ at 300 °C with hy-drogen; after the reducing procedure, catalysts wereflushed with He. The CO pulse chemisorption measurements were performed by introducing 10% CO in helium. The dispersion was calculated by

using a stoichiometry of CO/Pd = 1 and assuming Pd particle as a sphere.

2.3. Deoxygenation reactions

Deoxygenation reactions were performed using soybean free fatty acid as feedstock. It was obtained from the non-catalytic hydrolysis of the degummed soybean oil. The composition of the soybean fatty acid was determined by gas chromatography (meth-odology described inSection 2.4). It is composed by a blend of the following free fatty acids (wt.%): 10.3% of palmitic acid, 3.4% of stearic acid, 26.3% of oleic acid, 57.0% of linoleic acid and 3.0% of linolenic acid.

A typical deoxygenation reaction was carried out using 25 g of soybean free fatty acid and 2.5 g of catalyst, added in a batch Parr® reactor (240 mL). The catalyst was added to the fatty acid, the reactor was properly closed and H2pressure was increased to 10–40 bar. The reactions were performed at 300 °C for 6 h. Since the atmosphere in the reaction vessel was always rich in H2(high H2pressures) and Pd is easily reduced from PdO to Pd0_{, the catalysts were not previously} reduced. After 6 h, reaction mixture was solubilized in heptane and filtered for catalyst removal. Solvent was removed using a rotary evaporator and the reaction products were analyzed by gas chromatography.

2.4. Analysis

The feedstock was characterized by dissolving 0.05 g of sample in 1 mL of heptadecanoic acid solution (10 mg·mL−1in heptane). Sam-ples were analyzed in a gas chromatograph (Shimadzu GC-2010) equipped with Carbowax column (30 m × 0.32 mm × 0.25μm) and a flame ionization detector (FID). 1 μL of the sample was injected and the carrier gas (H2)flow rate was 1.9 mL·min−1. The analysis were car-ried out at 200 °C isothermal, with injector and detector temperatures at 250 °C with a split ratio of 1:50.

Typically, the reaction products had to be dissolved in pyridine and silylated with MSTFA (N-methyl-N-(trimethylsilyl)trifluoroacetamide) in order to be analyzed by a gas chromatography technique. Generally 100% of MSTFA was added to the sample. After the addition of the silylation agent, the samples were kept in an oven at 60 °C for 30 min. The internal standard eicosane (C20H42— 10 mg·mL−1in pyridine) was added for quantitative calculations. Samples were analyzed in a gas chromatograph (Shimadzu GC-2010) equipped with a non-polar column (DB5-HT— 15 m × 0.32 mm × 0.1 μm) and a flame ionization detector (FID). 1μL sample was injected into the GC with an column in-jection and the carrier gas (H2)flow rate was 3 mL·min−1. The injector and detector temperatures were 380 °C. The GC oven temperature was programmed as follows: 1 °C·min− 1 ramp from 25 °C to 50 °C; 1 min soak at 50 °C; 5 °C·min− 1 ramp from 50 °C to 380 °C and 10 min at 380 °C. Conversion was calculated as de-scribed in Eq.(1). Cð Þ ¼%
X P−AIS AIS  CIS V m  100% ð1Þ where:

ΣP sum of the areas of the products (hydrocarbons and oxy-genated products);

AIS area corresponding to internal standard (eicosane); CIS concentration of internal standard solution (mg·L−1); V volume corresponding to eicosane solution + solvent +

MSTFA added to sample (mL); m weight of the sample (mg).

Yield and selectivity were calculated as described in Eqs.(2) and (3), respectively. Yð Þ ¼% AP AIS CIS V m  100% ð2Þ Sð Þ ¼% Y_C 100% ð3Þ where:

AP area of the product;

AIS area corresponding to internal standard (eicosane); CIS concentration of internal standard solution (mg·L−1); V volume corresponding to eicosane solution + solvent +

MSTFA added to sample (mL); m weight of the sample (mg).

3. Results and discussion 3.1. Catalyst characterization

Chemical composition, measured by X-ray_{fluorescence (XRF), of the} clays used as support for Pd and palladium catalysts are presented as follow (mass percentage— wt.%): K10: 15.2% Al2O3; 0.4% CaO; 5.1% Fe2O3; 2.2% K2O; 1.3% MgO; 74.9% SiO2; 0.1% SO3. BM: 17.5% Al2O3; 4.3% CaO; 4.7% Fe2O3; 0.5% K2O; 4.0% MgO; 67.8% SiO2. PILCBM: 26.7% Al2O3; 0.1% CaO; 4.3% Fe2O3; 0.5% K2O; 3.3% MgO; 64.2% SiO2. KSF: 13.7% Al2O3; 3.6% CaO; 8.7% Fe2O3; 0.8% K2O; 2.7% MgO; 50.4% SiO2; 19.5% SO3. Pd/K10: 13.3% Al2O3; 0.3% CaO; 5.0% Fe2O3; 2.4% K2O; 1.0% MgO; 77.3% SiO2; 0.0% SO3; 1.0% PdO. Pd/BM: 17.8% Al2O3; 4.3% CaO; 4.8% Fe2O3; 0.6% K2O; 4.1% MgO; 66.7% SiO2: 0.7% PdO. Pd/PILCBM: 26.6% Al2O3; 0.0% CaO; 4.1% Fe2O3; 0.5 K2O; 3.2% MgO; 63.5% SiO2; 1.6% PdO. Pd/KSF: 13.3% Al2O3; 4.0% CaO; 10.2% Fe2O3; 0.9% K2O; 3.4% MgO; 44.5% SiO2; 21.2% SO3; 1.7% PdO.

The XRD patterns of the clays used as supports and of the Pd cata-lysts used in the recycle reactions are presented inFig. 1. Through the analysis of the BM and PILCBM diffractograms (Fig. 1A), it is possible to observe that BM is a smectite, a lamellar material with a 2:1 structure due to the presence of a reflection in 2θ = 5.7° corresponding to 001-plane. When this clay is pillared, this reflection is shifted to lower values of 2_{θ (2θ = 4.8°). The basal spacing d001-values were calculated from} the Bragg's law using the 2θ angle at which the maximum of the reflec-tion corresponding to the 001 reflection is obtained. It can be observed

that the pillaring process increased the basal spacing (Table 1) with re-spect to the starting clay. According toCseri et al. (1995), K10 and KSF commercial clays are obtained from the same Bavarian bentonite, there-fore they present similar XRD patterns. KSF is obtained by treatment with sulfuric acid at room temperature, resulting in the exchange of Na, Ca and Mg cations by protons, with only mild extraction of Al, Mg and Fe from the structure, thus retaining the montmorillonite structure. K10 is activated with a mineral acid at high temperature, resulting in ion exchange and extraction of Fe and Mg. X-ray diffraction patterns of the clays also present characteristic reflections of illite and quartz (Mezni et al., 2011; Moraes et al., 2010) from the raw clays. The XRD patterns were recorded for all Pd catalysts (not shown) and no significant differ-ences compared to the starting supports were observed. No reflection due to palladium was detected, probably thanks to the low metal con-tent, and also no further structural alterations in the supports were ob-served after impregnation and calcination processes.

The basal spacing, BET specific surface areas (SBET), micropore areas (Smicro), external areas (Sext), total pore volumes (Vtotal) and micropo-rous volumes (Vmicro) are summarized inTable 1. Pillaring process im-proved BM specific surface area and basal spacing. K10 clay presents larger specific surface area if compared to other clays. Acid treatment as-sociated to high temperature may have caused dealumination of the structure, creating larger specific surface areas located mainly in mesopores, while microporosity remains negligible. After palladium im-pregnation there was a decrease in specific surface area of the catalysts if compared with bare supports, except for Pd/PILCBM, it can be associ-ated with the blockage of clay pores by palladium particles. Pd/K10 mi-cropore area (Smicro) presented a threefold increase when compared to the K10. Palladium incorporation possibly leads to micropore formation by ordering delaminated layers or by decreasing mesopores by block-age. Pillared clay presented a small increase in the specific surface area after palladium incorporation, possibly some layers have been separat-ed by palladium, leading to micropore formation, this can be confirmed by the increase on the basal space from 9.3 on the PILCBM to 9.8 to Pd/ PILCBM.

The N2adsorption–desorption isotherms of the clays are shown in

Fig. 2. K10 clay presented a type IV isotherm (Fig. 2A), characteristic of mesoporous materials, but also displays type II features, which is indic-ative of high external surface, it is characteristic of lamellar materials, it is in accordance with results presented inTable 1. KSF clay presented characteristics of non-porous material (N2adsorption isotherm not presented). BM and PILCBM not only presented a type I isotherm (Fig. 2B), characteristic of microporous materials, but also displayed type IV features with a pronounced hysteresis loop, which is indicative of partial mesoporosity; these are characteristics of lamellar and pillared materials. By comparing the two isotherms it is possible to observe that the pillaring process yields a solid with the largest N2adsorption capacity,

since for p/p0= 0.2, N2adsorbed capacity for BM is 15.6 cm3·g−1, while for PILCBM it is 23.5 cm3_·g−1_{. It is also possible to observe that palladium} impregnation did not significantly affect isotherms of the clays.

The SEM images (Fig. 3) of the clays show a lamellar structure with crystals in the form of small and thin“sheets”. For BM and PILCBM (Fig. 3A–B) it can be observed that the pillaring process does not affect the shape of the crystals. K10 and KSF (Fig. 3C–D) have been described byMolu and Yurdakoç (2010)as“corn flake like crystals with fluffy ap-pearance revealing its extremelyfine plate structure”. SEM images of the palladium catalysts did not present any different features from the ones presented by the supports (not shown).

Adsorption of pyridine has generally been used to evaluate the nature of the acid sites (Brönsted and Lewis) and the density and strength of these sites as well. When pyridine molecules are adsorbed in solid acids, their interaction with Brönsted or Lewis acid sites produces specific bands in the IR spectra. For the interaction between pyridine and Brönsted acid sites, a band around 1540 cm−1can be observed, bands at 1620 and 1638 cm−1are also attributed to pyridine adsorbed in Brönsted acid sites. The interaction between pyridine and Lewis acid sites shows a band around 1450 cm−1, bands at 1620 and 1577 cm−1 are also attributed to pyridine adsorbed in Lewis acid sites; 1490 cm−1 band is associated with interaction between pyridine and both acid sites (Kiselev and Lygin, 1975; Mériaudeau et al., 1999). The IR spectra of pyr-idine adsorbed/desorbed in the clays used as supports are presented in

Fig. 4. InFig. 4A it is possible to observe a broad band, in the K10 Py-IR spectra, at 1543 cm−1assigned to the pyridinium ion (PyH+_{) in Brönsted} acid sites. Bands at 1447 cm−1presented by the clays could be associated to the physically adsorbed pyridine, while bands located at 1450 cm−1or higher wave-numbers are related to pyridine strongly adsorbed in Lewis acid sites (Yang et al., 2012). The spectra of pyridine desorption of KSF and BM (Fig. 4D–E) showed no pyridine adsorbed at Lewis acid sites after heating at 300 °C and 200 °C, respectively. Desorption of pyridine from

PILCBM (Fig. 4C) showed that the band at 1447 cm−1shifts to higher wave-numbers after heating the sample, indicating that the pyridine physically adsorbed in Lewis acid sites desorbed, and the remaining band is related to pyridine strongly adsorbed in Lewis acid sites (1455 cm−1).Fig. 4A related to pyridine adsorbed in K10 clay showed an intense band at 1447 cm−1. After treatment at high desorption tem-perature, this band shifts to higher wave-numbers (1456 cm−1), confirming that the band at 1447 cm−1_{is related to pyridine physically} adsorbed on Lewis acid site. According toLercher et al. (1996), while the band around 1540 cm−1does not change in wave-number upon varying the acidity of the solid, the band typical for coordinative bound pyridine (between 1445 and 1460 cm−1) increases in wave-number as the strength of interaction increases. The presence of the bands at 1620 and 1638 cm−1in the spectra of K10 (Fig. 4B) con_{firms the presence of} Brönsted acid sites in this material. Even after treatment at high desorp-tion temperature these bands are still present in the spectra, suggesting that the Brönsted acid sites in K10 clay are moderate to strong.

The acidity evaluation of Pd/clays was not performed.Issaadi et al. (2006)found through TPD and IR analyses that the acid properties of the pillared montmorillonite were affected by the Pd impregnation. According to the authors, comparing the spectra of pillared montmorillonite with the spectra of Pd/pillared montmorillonite, there was a decrease in the intensity of the bands attributed to Lewis acidity, while those attributed to Brönsted acidity had their intensity increased. The authors also stated that the treatment adopted reduced Pd2+_{to its} metallic state, and that this Pd0 _{activates the dissociation of} chemisorbed hydrogen, which then spilled over from the pillared montmorillonite and reduces, in part, the iron existent in the composition of the montmorillonite. This iron reduction reaction increases the negative charge on the clay lattice which is balanced by incorporating H+_{ions and increasing the Brönsted acidity. Lewis (LAS)} and Brönsted (BAS) acid site concentrations for the clays were Table 1

Basal spacing, textural properties of the clays and Pd/clay catalysts.

Basal spacing d(001) (Å) SBET(m2·g−1) Smicro(m2·g−1) Sexta(m2·g−1) VTotal(cm3·g−1) Vmicro(cm3·g−1) Pore diameter sizeb(nm)

BM 7.8 51.5 23.0 28.0 0.08 0.010 9.6 PILCBM 9.3 79.5 63.5 16 0.06 0.030 10.7 K10 9.7 230.5 3.5 227.0 0.30 – 5.8 KSF 10.0 7.4 1.9 5.5 0.01 0.001 13.6 Pd/BM 7.6 43.3 20.5 22.8 0.08 0.010 9.7 Pd/PILCBM 9.8 90.8 71.0 19.8 0.07 0.030 9.9 Pd/K10 9.7 212.5 10.3 202.2 0.30 – 6.1 Pd/KSF 10.0 4.7 1.5 3.2 0.01 0.001 26.4 a

Sext= Sext+ mesopores. b _{Average pore diameter.}

measured and are presented: LAS: BM—19.3 μmol·mg−1_{; PILCBM—} 53.3μmol·mg−1_{; K10—34.5 μmol·mg}−1_{; and KSF—22.8 μmol·mg}−1 and BAS: K10—8.8 μmol·mg−1_{. PILCBM presents the highest} concentra-tion of Lewis acid sites among the studied clays. This behavior is associat-ed with the pillaring process since it introduces aluminum polyoxocations in the clay structure. In pillared clays, Lewis acidity is attributed to the charge deficiency generated by low Al3+_{coordination in the crystal} edges (Pergher et al. 1999). K10 presents a moderate Lewis acidity if compared to PILCBM but it also presents Brönsted acid sites. Clays presented the following order of increasing Lewis acidity: BM b KSF b K10 b PILCBM.

Results concerning CO chemisorption are presented inTable 2. Pd/ clay catalysts presented heterogeneous dispersion results. Pd/BM presented 56% of dispersion, the highest dispersion achieved, possibly associated with low palladium content (0.7 wt.%). Pd/PILCBM present-ed dispersion values lower than Pd/BM, but its palladium content is more than twice larger. As presented previously, KSF has a very low spe-cific surface area, so it was not possible to measure the palladium dis-persion. Pd/K10 catalyst presented a palladium dispersion of 28% and its higher specific surface area collaborates to the highest metallic area. Analyzing chemisorption data of the recycled catalysts, it is possi-ble to observe a great loss in the metallic area and palladium dispersion. It is associated to the cleaning procedure, since the catalysts were sub-mitted to high calcination temperatures.

3.2. Deoxygenation reactions

Results concerning free fatty acid deoxygenation are presented

inTable 3. The main products obtained in deoxygenation reactions

were straight hydrocarbons as n-tetradecane, n-pentadecane, n-hexadecane and n-heptadecane, and branched hydrocarbons as 2,6,11-trimethyldodecane, 2,6,10-trimethyltetradecane, 3,3,4-trimethyldecane, 2,10-dimethylundecane and 3,8-dimethyldecane (i-alkanes). Byproducts, oxygenated compounds, are long-chain esters as C17\COO\C18, methyloctadecanoate, ethyl-hexadecanoate and ethyl-palmitate.

The reactions carried out with pure clays do not present any signi fi-cant deoxygenation activity (Table 3). For Pd/clay catalysts, conversions were lower than that obtained in the blank experiment, except for Pd/ K10. These results are associated with textural properties of the clays, since the materials presented low specific surface area limiting the ac-cess of the reagents to the active sites.

Pd/K10 presented the best results among the studied catalysts and it is probably related to its higher specific surface area and acidity charac-teristics, since K10 clay presents both Lewis and Brönsted acid sites. These properties make this catalyst an interesting and promising material to be used in deoxygenation reactions. As can be seen inTable 3, 46% of conversion and almost 99% of selectivity to hydrocarbons were achieved when Pd/K10 was used as catalyst in soybean fatty acid deoxygenation. K10 clay presents high concentration of Lewis acid sites, which are mod-erate to strong, and Brönsted acid sites (Fig. 4). The presence of branched hydrocarbons in the products can improve fuelflow properties and in-crease its cetane number. Isomerization reactions generally take place over bifunctional catalysts containing metallic sites for hydrogenation/ dehydrogenation and acid sites for skeletal isomerization via carbenium ions (Alvarez et al., 1996; Martens et al., 1989). According to the classical isomerization mechanism, n-paraffin dehydrogenates on the metal sites, and the produced linear olefins are protonated on the Brönsted acid sites Fig. 3. SEM micrographs of: A) BM, B) PILCBM, C) K10 and D) KSF.

to the corresponding linear carbenium ions, which subsequently rear-range into alkylcarbenium ions. These alkylcarbenium ions undergo a skeletal rearrangement before desorption by deprotonation and hydro-genation over metal sites to the corresponding iso-paraf_{fins (}Deldari, 2005). As shown inFig. 4all the clays presented Lewis acid sites and K10 clay possesses high concentration of the strong acid sites. Further-more, just K10 clay presents Brönsted acid sites responsible for the

Fig. 4. IR spectra A) after pyridine adsorption at 150 °C for K10, PILCBM, KSF and BM clays and after pyridine desorption at different temperatures for B) K10; C) PILCBM; D) KSF and E) BM.

Table 2

Palladium dispersion by CO chemisorption and Pd0_{particle size of the Pd/clays.}

Catalyst Palladium dispersion (%) Particle size (nm) Palladium surface (m2 ·g−1metal) Pd/BM 56.0 2.0 25.1 Pd/PILCBM 14.0 7.8 63.9 Pd/KSF – – – Pd/K10 28.0 3.9 128.7 Table 3

Deoxygenation results using different Pd/clay catalysts, soybean free fatty acid and stearic acid as feedstock, at 300 °C, 10 bar of H2pressure for 6 h.

Catalyst Conversion (%) Selectivities (%)

n-Alkanes i-Alkanes Oxygenated compounds

Blanc 4.9 0.0 0.0 15.0 BM 1.5 30.0 0.0 0.0 PILCBM 3.5 91.2 0.0 0.0 KSF 3.1 0.0 0.0 0.0 K10 4.5 89.1 0.0 0.0 Pd/BM 2.7 60.0 0.0 38.5 Pd/PILCBM 2.3 84.2 0.0 11.5 Pd/KSF 4.4 34.5 0.0 54.3 Pd/K10 46.0 90.6 8.3 0.0 Pd/K10a _{99.2 88.0 12.0 0.0} Pd/K10b 58.1 61.4 36.9 1.7 a

Stearic acid used as feedstock.

b

rearrangements of carbenium ions and the isomerization of the alkanes. When stearic acid (saturated C18) was used as feedstock, conversion in-creased to 99% with 100% of selectivity to hydrocarbons (using Pd/K10 as catalyst). To investigate this result, an experiment with previously in situ hydrogenated soybean free fatty acid was carried out. In this experiment, soybean free fatty acids were hydrogenated at 120 °C using Pd/K10 cata-lyst for 2 h under 30 bar of H2pressure. After that, and with the reactor at room temperature, the remaining H2was purged and reactor was charged with 10 bar of H2and deoxygenation reaction was carried out as previously described. The reaction performed with previously hydro-genated free fatty acids presented results more similar to the ones pre-sented by the reaction with soybean free fatty acid, than that obtained when stearic acid was used as feedstock. It can be related to the faster cat-alyst deactivation due to the double bonds hydrogenation step.Snåre et al. (2008)have compared deoxygenation results between oleic and linoleic acids using a Pd/C catalyst and they found lower conversion when linoleic acid was used as feedstock. The authors attributed this

difference to the step related to the hydrogenation of linoleic acid to oleic acid. They also observed that the catalyst deactivation by coke for-mation was more pronounced in the reaction carried out using linoleic acid as feedstock.

Based on previous results obtained with Pd/K10, the influence of H2 pressure was investigated. Four different H2pressures were studied (10, 20, 30 and 40 bar) and the results are presented inTable 4. When hydrogen pressure is increased, conversions also increase, except for 40 bar of pressure. At 30 bar of H2pressure the highest conversion was observed and a change in product distribution was also detected with the increase in i-alkane selectivity. As presented previously in the literature (Mäki-Arvela et al., 2007; Rozmysłowicz et al., 2010, 2012), catalyst deactivation is retarded at high H2concentrations, lead-ing to higher conversions. As observed, when H2pressure higher than 30 bar was tested, conversion dropped from 74.5% (30 bar) to 65.5% (40 bar), which may be caused by the high saturation of palladium sur-face by hydrogen molecules, hindering the reagent access to the active sites. Since the catalysts were not previously reduced, one experiment with reduced Pd was performed and it is possible to observe that after 6 h of reaction no differences were observed in conversion or selectivity if comparing pre-reduced and no pre-reduced catalyst.

Deoxygenation results with Pd/K10 catalyst as a function of time-on-stream are presented inFig. 5A–C. The conversion (Fig. 5A) increases with the time, as expected. Until 6 h of reaction, catalyst deactivation was not observed, suggesting that longer reaction times could lead to higher conversions. The selectivity (Fig. 5B) to linear alkanes is maxi-mum at 180 min of reaction and the decrease in the n-alkane selectivity is followed by the increase in branched alkane selectivity. Oxygenated compound concentration is maximum at 30 min (25 wt.%); after this time these compounds were converted in hydrocarbons. This becomes Table 4

Deoxygenation results using Pd/K10 catalyst and soybean free fatty acid as feedstock, at 300 °C, and different H2pressures for 6 h.

H2pressure (bar) Conversion (%) Selectivities (%)

n-Alkanes i-Alkanes Oxygenated compounds 10 46.0 90.6 8.3 0.0 20 61.2 89.9 6.8 2.5 30 74.5 69.0 29.2 1.8 30a 74.4 64.6 34.8 0.0 40 65.5 75.7 19.0 0.0 a_{Pd previously reduced at 150 °C/1 h in H} 2.

Fig. 5. Deoxygenation of soybean fatty acid over Pd/K10 at 300 °C, 30 bar of H2pressure: A) Conversion as a function of stream (wt.%), B) selectivity as a function of

clearer when analyzing selectivity as a function of the conversion (Fig. 5C), where at 20 wt.% of conversion, selectivity to oxygenates is 75%; with the progress of the reaction oxygenates are consumed and hydrocarbons are formed, suggesting that these oxygenated products are also deoxygenated to hydrocarbons.

4. Conclusions

The use of clays as support to palladium catalysts was investigated on the deoxygenation of free fatty acids. Obtained results showed that clay-supported palladium catalysts were not able to deoxygenate free fatty acids, except for Pd/K10. The influence of textural and acid proper-ties was evaluated, and these characteristics of K10 commercial clay (Brönsted and Lewis acidities and large specific surface area) were de-finitive for deoxygenation reactions, since 74.5% of conversion with al-most 100% of selectivity to aliphatic hydrocarbons was achieved when Pd/K10 (1% of Pd) was used as catalyst. Different H2pressures were evaluated and 30 bar was found as the best work pressure. The main products obtained by catalytic deoxygenation of soybean free fatty acid were n- and i-alkanes. This catalytic system has high potential to be used for the synthesis of hydrocarbons in the diesel and kerosene (Jet Fuel) range.

Acknowledgments

The authors are grateful to CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) and FAPERJ (Fundação de Amparo á Pesquisa do Estado do Rio de Janeiro) for the Ph.D. scholarships. The authors would like to acknowledge Mr. B. Rozmysłowicz from Process Chemistry Centre, Åbo Akademi University for CO chemisorption analysis and Professor Victor L.S. Teixeira da Silva from Federal University of Rio de Janeiro (NUCAT/COPPE) for IR-Py analysis.

References

Alvarez, F., Ribeiro, F.R., Perot, G., Thomazeau, C., Guisnet, M., 1996.Hydroisomerization and hydrocracking of alkanes: 7. Influence of the balance between acid and hydrogenating functions on the transformation of n-decane on PtHY catalysts. J. Catal. 162, 179–189.

Cañizares, P., Lucas, A., Dorado, F., Durán, A., Asencio, I., 1998.Characterization of Ni and Pd supported on H-mordenite catalysts: influence of the metal loading method. Appl. Catal. A Gen. 169, 137–150.

Cseri, T., Békássy, S., Figueras, F., Cseke, E., Menorval, L.C., Dutartre, R., 1995. Characteriza-tion of clay-based K catalysts and their applicaCharacteriza-tion in Friedel–Crafts alkylaCharacteriza-tion of ar-omatics. Appl. Catal. A Gen. 132, 141–155.

Datka, J., 1981.Dehydroxylation of NaHY zeolites studied by infrared spectroscopy J. Chem. Soc. Faraday Trans. 77, 2877–2881.

Deldari, H., 2005.Suitable catalysts for hydroisomerization of long-chain normal paraffins. Appl. Catal. A Gen. 293, 1–10.

Issaadi, R., Garin, F., Chitour, C.-E., 2006.Study of the acid character of some palladium-modified pillared clay catalysts: use of isopropanol decomposition as test reaction. Catal. Today 113, 166–173.

Kikhtyanin, O.V., Rubanov, A.E., Ayupov, A.B., Ech, G.V., 2010.Hydroconversion of sun-flower oil on Pd/SAPO-31 catalyst. Fuel 89, 3085–3092.

Kiselev, A.V., Lygin, V.I., 1975.Infrared Spectroscopy of Surface Compounds. John Wiley & Sons, New York.

Kubičková, I., Snåre, M., Eränen, K., Mäki-Arvela, P., Murzin, D.Yu., 2005.Hydrocarbons for diesel fuel via decarboxylation of vegetable oils. Catal. Today 106, 197–200.

Lercher, J.A., Gründling, C., Eder-Mirth, G., 1996.Infrared studies of the surface acidity of oxides and zeolites using adsorbed probe molecules. Catal. Today 27, 353–376.

Lestari, S., Mäki-Arvela, P., Simakova, I., Beltramini, J., Lu, G.Q.M., Murzin, D.Yu., 2009. Cat-alytic deoxygenation of stearic acid and palmitic acid in semibatch mode. Catal. Lett. 130, 48–51.

Mäki-Arvela, P., Kubičková, I., Snåre, M., Eränen, K., Murzin, D.Yu., 2007.Catalytic decar-boxylation of fatty acids and their derivatives. Energy Fuels 21, 30–41.

Mäki-Arvela, P., Snåre, M., Eränen, K., Myllyoja, J., Murzin, D.Yu., 2008.Continuous decar-boxylation of lauric acid over Pd/C catalyst. Fuel 87, 3543–3549.

Martens, J.A., Tielen, M., Jacobs, P.A., 1989.Relation between paraffin isomerisation capa-bility and pore architecture of large-pore bifunctional zeolites. Stud. Surf. Sci. Catal. 46, 49–60.

Mériaudeau, P., Tuan, A.Vu., Nghiem, T., Sapaly, G., Naccache, C., 1999.Comparative eval-uation of the catalytic properties of SAPO-31 and ZSM-48 for the hydroisomerization of N-octane: effect of the acidity. J. Catal. 185, 435–444.

Mezni, M., Hamzaoui, A., Hamdi, N., Srasra, E., 2011.Synthesis of zeolites from the low-grade Tunisian natural illite by two different methods. Appl. Clay Sci. 52, 209–218.

Molu, Z.B., Yurdakoç, Z., 2010.Preparation and characterization of aluminum pillared K10 and KSF for adsorption of trimethoprim. Microporous Mesoporous Mater. 127, 50–60.

Moraes, D.S., Angélica, R.S., Costa, C.E.F., Rocha Filho, G.N., Zamian, J.R., 2010.Mineralogy and chemistry of a new bentonite occurrence in the Eastern Amazon region, North-ern Brazil. Appl. Clay Sci. 48, 475–480.

Moronta, A., Luengo, J., Ramírez, Y., Quiñónez, J., González, E., Sánchez, J., 2005. Isomeriza-tion of cis-2-butene and trans-2-butene catalyzed, by acid- and ion-exchanged smectite-type clays. Appl. Clay Sci. 29, 117–123.

Paiva, L.B., Morales, A.R., Valenzue la-Diaz, F.R., 2008.Organoclays: properties, preparation and applications. Appl. Clay Sci. 42, 8–24.

Pergher, S.B.C., Fornés, V., Corma, A., 1999.Materiales Laminares Pilareados: Preparación y Propriedades. Quim. Nova 22, 693–709.

Pergher, S.B.C., Sprung, R., 2005.Pilarização de uma argila Brasileira com Poliidroxications de alumínio: Preparação, Caracterização e Propriedades Catalíticas. Quim. Nova 28, 777–782.

Ping, E.W., Pierson, J., Wallace, R., Miller, J.T., Fuller, T.F., Jones, C.W., 2011.On the nature of the deactivation of supported palladium nanoparticle catalysts in the decarboxylation of fatty acids. Appl. Catal. A Gen. 396, 85–90.

Rozmysłowicz, B., Mäki-Arvela, P., Lestari, S., Simakova, O.A., Eränen, K., Simakova, I.L., Murzin, D.Yu., Salmi, T.O., 2010.Catalytic deoxygenation of tall oil fatty acids over a palladium-mesoporous carbon catalyst: a new source of biofuels. Top. Catal. 53, 1274–1277.

Rozmyszowicz, B., Mäki-Arvela, P., Tokarev, A., Leino, A.R., Eränen, K., Murzin, D.Yu., 2012.

Influence of hydrogen in catalytic deoxygenation of fatty acids and their derivatives over Pd/C. Ind. Eng. Chem. Res. 51, 8922–8927.

Snåre, M., Kubičková, I., Mäki-Arvela, P., Chichova, D., Eränen, K., Murzin, D.Yu., 2008. Cat-alytic deoxygenation of unsaturated renewable feedstocks for production of diesel fuel hydrocarbons. Fuel 87, 933–945.

Snåre, M., Kubičková, I., Mäki-Arvela, P., Eränen, K., Murzin, D.Yu., 2006.Heterogeneous catalytic deoxygenation of stearic acid for production of biodiesel. Ind. Eng. Chem. Res. 45, 5708–5715.

Snåre, M., Kubičková, I., Mäki-Arvela, P., Eränen, K., Wärnå, J., Murzin, D. Yu, 2007. Produc-tion of diesel fuel from renewable feeds: kinetics of ethyl stearate decarboxylaProduc-tion. Chem. Eng. J. 134, 29–34.

Yang, J., Kikhtyanin, O.V., Wua, W., Zhou, Y., Toktarev, A.V., Echevsky, G.V., Zhang, R., 2012.

Influence of the template on the properties of SAPO-31 and performance of Pd-loaded catalysts for n-paraffin isomerization. Microporous Mesoporous Mater. 150, 14–24.