jou rn al hom epage: www.elsev ier .com/ locate /ca t tod
ontinuous production of biodiesel from low grade feedstock inresence of Zr-SBA-15: Catalyst performance and resistance againsteactivation
ose Iglesiasa,∗, Juan A. Melerob, L. Fernando Bautistab,abriel Moralesb, Rebeca Sánchez-Vázquezb
Department of Chemical and Energy Technology, Universidad Rey Juan Carlos, C/Tulipán s/n, Móstoles E28933, Madrid, SpainDepartment of Chemical and Environmental Technology, Universidad Rey Juan Carlos, C/Tulipán s/n, Móstoles E28933, Madrid, Spain
r t i c l e i n f o
rticle history:eceived 13 November 2013eceived in revised form 1 January 2014ccepted 7 January 2014vailable online xxx
eywords:
a b s t r a c t
Zirconium-containing SBA-15 materials have been used in the production of fatty acid methyl estersfrom low grade oleaginous feedstock. Its resistance against deactivation has been assessed by means ofstudying the effect of conventional impurities present in lipid wastes over the catalytic performance ofthis material. Alkaline metal cations like potassium could interact with Brønsted acid sites, causing theirneutralization by ion exchange and a limited, but not complete, deactivation of the material. Additionally,organic unsaponifiable compounds like retinoids or phospholipids – being studied in this work as retinol
iodieseleterogeneous acid catalystsirconiumr-SBA-15ow grade feedstockixed bed reactor
and lecithin, respectively – strongly interact with the catalyst surface, leading to a strong deactivationof the material, though reversible, since they are fully regenerated by calcination in air. Catalytic assaysin continuous mode in a fixed bed reactor suggest a higher resistance of Zr-SBA-15/bentonite pelletsagainst catalyst deactivation. Bentonite clay, which has been used as binding agent for the preparationof the particulate catalyst, seems to be responsible for this behavior, acting as poison scavenger andpreventing the access of the impurities to the catalytic acid sites and consequently their deactivation.
. Introduction
Biodiesel consists of a mixture of fatty acid alkyl esters obtainedrom renewable resources, typically from vegetables oils and/ornimal fats, by alcoholysis – mainly using methanol – of triglyc-rides present in lipids, conventionally using a basic or an acidatalyst. Currently, the most extended process for biodiesel pro-uction involves the transesterification of high-grade, refinedleaginous feedstock with methanol using homogeneous alkalineatalysts, because of their superior catalytic performance at loweaction temperatures compared to acid catalysts. However, thisperation mode has several important disadvantages, such ashe need of using high-grade feedstock, with low amounts ofmpurities – typically free fatty acids (FFA), water or unsaponifi-ble matter. In this way, these raw materials usually represent aubstantial fraction of the processing cost for biodiesel, reducingts profitability. As an alternative, lower-grade feedstock could
Please cite this article in press as: J. Iglesias, et al., Continuous production of
performance and resistance against deactivation, Catal. Today (2014), http
e used as starting raw materials, but in this case, the processingosts are moved from the feedstock to the conditioning stageseeded for the treatment of the feedstock to decrease the amount
of the impurities, and the purification of biodiesel through theremoval of salts formed during the neutralization of homogeneouscatalysts. Moreover, conventional processes are quite sensitive towater content in the feedstock and other impurities, making evenworse the above described drawbacks of this technology [1].
Overcoming these disadvantages has been the focus of scien-tists’ efforts which have crystallized in the development of newproduction technologies with higher profitability and environmen-tal sustainability. In addition, some environmental politics driveinvestments in the biofuels sector in the same direction. As anexample, the EU Directive 2009/30/CE, which stiffs the legislation interms of greenhouse emissions by means of conventional fuels sub-stitution by biofuels, suggests the use of waste feedstock as startingraw materials for their production. In this way, it is of major impor-tance studying alternative non-edible waste feedstock which couldbe used for the sustainable production of biodiesel.
During the last years, development of heterogeneous catalystsfor biodiesel production has been a very active research area [2]since the use of heterogeneous catalysts involves several advan-tages which are not present in their homogeneous counterparts,
biodiesel from low grade feedstock in presence of Zr-SBA-15: Catalyst://dx.doi.org/10.1016/j.cattod.2014.01.004
such as the production of a higher grade biodiesel and glycerol,the ability to be used in continuous fixed bed reactors, no wastewater generation because the catalyst removal is not required,fewer purification steps and, therefore, fewer complex facilities [1].
n particular, heterogeneous acid catalysts could be of great inter-st to improve biodiesel production process, because these, besideshe multiple advantages described for heterogeneous systems, canlso simultaneously drive both esterification and transesterifica-ion reactions, being not required the conditioning stages for these of low grade feedstocks with high acidity. In this way, hetero-eneous acid catalysts make possible designing integrated processased on packed bed continuous flow reactors with enhanced pro-uctivity, easy catalyst separation and minimum purification stepss compared to conventional biodiesel production processes [3–5].oreover, techno-economic studies suggest that heterogeneous
cid catalyzed processes are more profitable than conventionalrocedures applied to the production of biodiesel [6]. However,he industrial implementation of such catalysts in commercialiodiesel production still needs important improvements, mainlyevoted to the stability of the catalysts and resistance against deac-ivation in presence of high amounts of several impurities which arelenty in low-quality oleaginous raw materials (water, metals andnsaponifiable matter).
In this context, Zr-SBA-15 has revealed to be an interest-ng catalyst for the production of biodiesel. Recently, we haveemonstrated that this catalyst show high catalytic performance
n biodiesel production from low-grade waste cooking oil in a con-inuous packed bed reactor [7]. However, there are still some lacksf information about the response of this material to the presencef some impurities accompanying the oleaginous feedstock, whichould act as potential catalyst poison. This work sheds some lightn the catalytic behavior of Zr-SBA-15 material in biodiesel pro-uction from low-grade raw materials, assessing the influence ofeveral natural substances present in oleaginous feedstock on thetability of the catalyst against deactivation. Finally, this investi-ation provides some answers about the good resistance of thisaterial against catalyst poisoning in long time-on-stream exper-
ments for the transformation of mixtures of low-grade oleaginouseedstock.
. Materials and methods
.1. Materials
Crude palm oil (CPO, Gran Velada), category-1 animal fat (AF-, Ibergrasa), waste cooking oil (WCO, obtained from the cantinef the university), pork lard (Lard, Ibergrasa), mixed animals fatsChicken-Beef-Pork, MAF, Ibergrasa), and methanol (ACS grade,ldrich) were used as feedstock in acid catalyzed methanolysis
ests. Oleaginous feedstocks were filtered to remove suspendedolids prior to their use as the sole conditioning step for theseaterials. Tetraethylorthosilicate (TEOS, 98%, Aldrich) and zir-
onocene dichloride (Cp2ZrCl2, ABCR) were used as silicon andirconium precursors, respectively, together with Pluronic P-123PEO20-PPO70-PEO20, Aldrich) as the structure directing agent forhe synthesis of Zr-SBA-15. Natural bentonite, a montmorillonite-ype clay (Süd-Chemie) was used as binding agent and methylellulose (Aldrich) was employed as additive to control the plas-icity of the mixed catalyst-clay-water in the agglomeration ofr-SBA-15.
.2. Synthesis of Zr-SBA-15 material
Zirconium-functionalized SBA-15 material was preparedccording to the method described in literature [8]. In a typical
Please cite this article in press as: J. Iglesias, et al., Continuous production of
performance and resistance against deactivation, Catal. Today (2014), http
ynthesis, Pluronic 123 (4 g) was dissolved in a hydrochloric acidqueous solution (125 mL of 0.67 N) at room temperature. Afteromplete dissolution, 1.2 g of zirconocene dichloride were addednd the resultant suspension was stirred for 3 h and heated to 40 ◦C.
PRESSay xxx (2014) xxx– xxx
TEOS (8.63 g) was added to the synthesis media under vigorousstirring, which continued for 20 additional hours keeping constantthe synthesis temperature at 40 ◦C. The resultant suspension washydrothermally aged at 130 ◦C under static conditions for 24 h.The materials were finally recovered by filtration and air-driedovernight. Surfactant was removed by calcination at 450 ◦C in airduring 5 h.
2.3. Agglomeration of Zr-SBA-15
Agglomeration of powder Zr-SBA-15 was accomplished with theaim to get a particulate material to be used in a packed bed reactorwhile keeping low the pressure drop during operation. Agglomer-ation of Zr-SBA-15 was performed following a method previouslyreported in literature [8]. Powdered Zr-SBA-15 (65 g) was mixedwith bentonite (25 g), acting as binding agent, methyl cellulose(10 g), which increased the plasticity of the mixture, and ultra-purewater. The mixture was loaded into a kneading machine (Lleal) andthe dough was kneaded during 3 h to achieve a homogeneous mix-ture. The resultant material was extruded to form uniform rods(diameter 1.0 mm) which were dried in a climatic chamber withcontrolled atmospheric humidity. The resultant material was cal-cined in air at 450 ◦C during 5 h. Finally, the rods were cut intoparticles of 1.0 mm.
2.4. Catalyst characterization
The textural properties of the synthesized catalyst samples werecalculated from N2 adsorption–desorption isotherms recorded at77 K in a Micromeritics TRISTAR 3000 unit. Surface area was cal-culated applying the B.E.T. method and pore size distributions byapplying the B.J.H. method using the K.J.S. correction. Total porevolume was assumed to be that recorded at p/p0 = 0.985. Structuralproperties were assessed by means of X-ray powder diffraction(XRD) experiments performed on a Philips XıPert diffractometerusing the Cu K� line in the 2� angle range 0.6–5.0◦, with a stepsize of 0.02◦ for low angle analysis and a step size of 0.04◦ in the2� angle range of 5.0–50.0◦ for high angle analysis. Bulk zirconiumcontents were determined by Inductively Coupled Plasma-AtomicEmission Spectroscopy (ICP-OES). Acid capacity was measured byammonia temperature programmed desorption in a Micromeritics2910 (TPD/TPR) equipment fitted with a TCD detector. X-ray photo-electron spectroscopy (XPS) experiments were performed using aKratos AXIS HSi instruments fitted with a charge neutralizer and AlK� X-ray source. Spectra were recorded at normal emission usinganalyser pass energy of 40 eV and X-ray power of 225 W. Prior to theanalysis, samples were outgassed at 10−11 bar overnight. Bindingenergies were referenced to the C1s line (284.8 eV) and deconvo-lution curves were achieved using the Casa XPS software.
2.5. Catalytic experiments
Catalytic experiments performed in presence of powder Zr-SBA-15 were accomplished in a batch reactor. The experiments werecarried out in a 25 ml stainless-steel autoclave (Autoclave Engi-neers) fitted with temperature controller, mechanical stirrer anda pressure transducer. In a typical experiment, 5 g of low-gradeoleaginous feedstock, methanol (50 methanol to oil molar ratio)and the catalyst (12.45 wt% catalyst loading) are placed togetherinside the reactor vessel. The system is then hermetically closed andthe temperature (209 ◦C) and stirring conditions (2000 rpm) set up.The reaction was allowed to proceed for 6 h before cooling down
biodiesel from low grade feedstock in presence of Zr-SBA-15: Catalyst://dx.doi.org/10.1016/j.cattod.2014.01.004
the reactor using an ice-water bath. The resultant suspension wasfiltered using a nylon-membrane filter to recover the catalyst andminimize solid losses – for catalyst reuse experiments. The excessof methanol was removed by rotary evaporation under vacuum
Fig. 1. FAME yield obtained in the batch-mode methanolysis of various low-grade
ARTICLEATTOD-8836; No. of Pages 8
J. Iglesias et al. / Catalys
t 60 ◦C at 0.2 bar. The recovered catalysts samples were double-ashed with methanol and n-hexane in an ultrasound bath for the
emoval of both polar and non-polar surface-adsorbed compounds.lternatively, recovered catalysts were thermally treated at 450 ◦C
n air atmosphere, depending on the experiment, to completelyurn out the adsorbed organic substances.
Catalytic reaction experiments with agglomerated Zr-SBA-15ere driven in a continuous up-flow fixed-bed reactor (120 cm
ength, 0.9 cm ID). The inner temperature of the reactor (209 ◦C) wasontrolled by an external wrapping metallic resistance, whereashe pressure of the system (70 bar) was controlled by means of aeedle valve placed in the reactor exit, to ensure liquid phase condi-ions. The catalytic bed was initially loaded with 28 g of particulateatalyst and used for all the reaction tests. Methanol and oil wereed to the inlet of the reactor using HPLC pumps, allowing an effec-ive control of the methanol to oil molar ratio (50) and residenceime of the reactants stream inside the reactor of 30 min. The efflu-nt solution coming from the reactor was expanded in a flash vesselt 65 ◦C under vacuum (0.2 bar) to remove some of the methanolxcess and allowed to decant in a tank to separate the FAME andlycerol layers. FAME layer was then recovered and rotaevaporatedo eliminate the remaining methanol.
The molar yield of transformation of fatty acid alkyl chainscomprising both FFA and glycerides) into fatty acid methyl estersYFAME) was calculated by analyzing crude reaction samples by
eans of 1H NMR analysis in a Varian Mercury Plus 400 unit in similar way to that described by Whalen et al. [9].
.6. Biodiesel characterization
The biodiesel product was thoroughly washed twice with dis-illed water in a separation funnel and the resultant product dried at0 ◦C and 0.5 bar for 12 h to remove traces of moisture. The methylsters produced were characterized using the quality standard ISO-4214, in terms of composition (FAME content, acid value, presencef alkaline metals and phosphorus, presence of mono-, di-, andriglycerides, presence of methanol and glycerol), density, viscos-ty, flash point, cetane number, iodine value, and cold propertiesuch as cold filter plugging point (CFPP).
. Results and discussion
.1. Catalytic tests in presence of low grade feedstock: influencef catalyst poisons
The first purpose of the present work was to evaluate theroperties of the starting oleaginous raw materials which couldxert any influence on the catalytic behavior of Zr-SBA-15 cata-yst. Table 1 lists the main properties calculated for CPO, WCO,F-1, Lard and MFA, such as density, viscosity, acid value, metalsontent, water and unsaponifiable matter. The exhaustive charac-erization of these raw materials revealed the presence of severalmpurities which could adversely interfere in the catalytic activ-ty and stability of Zr-SBA-15 to produce fatty acid methyl estersFAME). In particular, high content of free fatty acids (FFA), alkalinend alkaline earth metals, phosphorous, unsaponifiable matter andoisture have been observed. Hence, the processing of these rawaterials in the production of biodiesel by means of acid-catalysis
s expected to be difficult [10] and could probably lead to someatalyst poisoning during the methanolysis reaction tests.
The catalytic activity of Zr-SBA-15 material in the production of
Please cite this article in press as: J. Iglesias, et al., Continuous production of
performance and resistance against deactivation, Catal. Today (2014), http
ethyl esters from low-grade oils and fats was evaluated in batch-ode using previously optimized reaction conditions: 209 ◦C, 45.80ethanol to oil molar ratio, 12.45 wt% catalysts loading, 2000 rpm
feedstocks in presence of fresh and used Zr-SBA-15 samples under optimized reac-tion conditions. Reaction conditions: Time = 6 h; 45.80 methanol:oil molar ratio;12.45 wt% catalyst loading; stirring rate 2000 rpm.
for the different raw materials in three sets of experiments: usingthe fresh catalyst, after a first reuse with intermediate catalystwashing with methanol and n-hexane, and after a second reusewith intermediate thermal treatment (450 ◦C in air). Zr-SBA-15 pro-vided very good behavior under the optimized conditions for allthe experiments leading to molar FAME yields around 92% after6 h. Nonetheless, with regard to the first reutilization experimentafter catalyst washing, the material evidenced a slight deactiva-tion, with a 5% FAME yield decay – on average. However, initialcatalytic activity was fully recovered in the third consecutive reac-tion experiment, after the calcination treatment, which suggeststhat the catalyst deactivation observed during the first recyclingtest was a reversible process, most probably linked to the adsorp-tion of organics onto or around the catalytic sites [11], regardlessthe starting raw material used in the catalytic test.
With the aim to fully understand the deactivation phenomenaobserved in the previous experiments, some catalytic tests wereperformed in the methanolysis of CPO, by modifying this time thestarting raw material with the addition of several of the naturalsubstances and impurities detected in the low grade feedstocks:alkaline metals such as Na and K – through de addition of NaCl andKCl – P – which was added in the form of lecithin – and unsaponifi-able compounds such as cholesterol and retinol, a chemical presentin high concentration in several oleaginous feedstock, such as crudepalm oil [12]. As in the previous catalytic tests, the activity andstability of the Zr-SBA-15 material were evaluated in two consec-utive reactions runs, one of them performed in presence of thefresh catalyst and the second one after washing the used mate-rial with methanol and n-hexane. Fig. 2 depicts the results (FAMEyield) obtained from the described experiments.
First use of the catalyst in presence of unmodified CPO providedgood catalytic results, yielding FAME productivity above 80% of themaximum available after 3 h. Recycling tests evidenced the samecatalytic activity decay (∼5%) previously observed for tests per-formed during 6 h. These FAME yield values were used as referenceto compare with in the study of the deactivating capability of thepreviously described potential poisons. At this respect, most of thecatalytic tests carried out in presence of doped CPO showed sim-ilar reaction results, suggesting lack of deactivating capability formost of the tested substances. However, there is a substance whichdisplays a much higher deactivating capability than any other oneamong the tested potential poisons: lecithin. Lecithin is the com-mon name given for phosphatidylcholine, a phosphatide of fattyacid diglycerides linked to the choline ester of phosphoric acid. Thisis a natural substance, present in both animal and vegetal lipids,
biodiesel from low grade feedstock in presence of Zr-SBA-15: Catalyst://dx.doi.org/10.1016/j.cattod.2014.01.004
which is a key building block of cell membrane bilayers. The firstuse of the catalyst in presence of lecithin-containing CPO yielded65–70% FAME, depending on the initial phosphorus loading, a muchlower yield than that obtained for the reference tests in absence of
4 J. Iglesias et al. / Catalysis Today xxx (2014) xxx– xxx
Table 1Properties of crude palm oil used as feedstock for methyl ester production.
Property Analytical method Unit Value
CPO AF-1 WCO Lard MAF
Acid value UNE EN ISO 660:2000 mgKOH g−1 21.45 33.45 4.06 0.47 3.57Density at 40 ◦C UNE EN ISO 3675:1999 kg m−3 908 829 918 894 920Viscosity at 40 ◦C UNE EN ISO 3104:1996 mm2 s−1 42.9 47.4 66.5 48.7 52Metals content ASTM D5185:2013 mg kg−1
otential poisons. These low FAME yields, together with the dropbserved in the recycling tests, make obvious the deactivation ofhe zirconium-based material in presence of this natural substance.his fact seems to be linked to the high affinity of zirconium oxidesor phosphates. Several authors have described the strong adsorp-ion of these ions onto the surface of zirconium oxide particles [13],nd consequently, the deactivation of the Zr-SBA-15 material coulde related to the preferential adsorption of the phosphate group of
ecithin onto the zirconium acid sites, in an analogous way to thatescribed by Su. In this way, Lewis acid sites would be blockednd the access of the glycerides to these catalytic centers hindered,imiting the extension of the reaction.
With regard to the experiments performed in presence ofodium or potassium chloride (500 mg kg−1), cholesterol andetinol (each at 5 wt%), negligible differences with the resultschieved for the reference catalytic tests were detected, suggestingn absence of catalytic poisoning, at least for these substances at
Please cite this article in press as: J. Iglesias, et al., Continuous production of
performance and resistance against deactivation, Catal. Today (2014), http
he tested concentration levels. Increasing the amount of these sub-tances up to 1000 mg kg−1 for the alkaline cations and to 10 wt% inhe case of the organic chemicals produced almost the same results,nless for the potassium and retinol, which significantly reduced
ig. 2. Results from batch-mode methanolysis reaction tests, performed in presence Zr-SBholesterol and retinol). Time = 3 h; 45.80 methanol:oil molar ratio; 12.45 wt% catalyst loa
the FAME yield achieved when present at these concentration lev-els.
In the case of potassium, the deactivating effect could be relatedto the partial neutralization of the Brønsted acid sites present inZr-SBA-15 materials [14], since Lewis acid sites do not present ionexchange capability and thus, alkaline cations such as sodium orpotassium should not interfere with this type of acid sites. The factthat Zr-SBA-15 activity is conditioned by the presence of potassiumbut not of sodium could be related to the higher solubility of potas-sium chloride in methanol at high temperatures, in contrast to thatfound for sodium chloride, whose solubility in methanol is scarceat temperature levels such as those used in the catalytic tests [15].In this way, the interaction between sodium and the Brønsted acidsites is limited, and thus the neutralization of the same does notoccur during the reaction test.
On the other hand, the deactivating effect of retinol seems to becaused by a similar process to that found for lecithin, the retention
biodiesel from low grade feedstock in presence of Zr-SBA-15: Catalyst://dx.doi.org/10.1016/j.cattod.2014.01.004
over the catalyst surface. Bearing in mind the lipophilic nature ofretinol and its low affinity for silica [16], the deactivation of Zr-SBA-15 materials by retinol could be caused by the interaction of retinolwith the zirconium acid sites. However, this fact has not previously
A-15 material, carried out over CPO doped with some impurities (NaCl, KCl, Lecithin,ding; stirring rate 2000 rpm.
een described, and nevertheless the interaction and deactivationapability of retinoids seems to be weaker than that observed forhe phosphatidylcholine.
In order to fully understand the deactivating effect of the dif-erent substances used as potential poisons for Zr-SBA-15 in the
ethanolysis transformation of oleaginous feedstock into methylsters, several analytical techniques were applied for the charac-erization of the physicochemical properties of the used samples ofhe mesostructured materials.
Results obtained from the thermogravimetric analysis per-ormed on used samples of Zr-SBA-15 suggest a negligible influencef retained matter, since the weight loss during the thermal analy-is was quite similar for every sample (ranging from 14 to 17 wt%).his similar result, regardless the potential poison in whose pres-nce the catalytic material was used, suggests that the amountf glycerides/FAME retained by the catalyst is much higher thanhe held amount of lecithin or retinol, shadowing their presencen thermogravimetric curves. In this way, although lecithin andetinol deactivation capabilities have been ascribed to the selectivedsorption of these substances onto the catalytic acid sites, theirresence cannot be ascertained from TG analysis.
In order to get better insights in the effect of the different sub-tances acting as poisons in the surroundings of zirconium acidites, XPS spectra were recorded for the used Zr-SBA-15 samples
Please cite this article in press as: J. Iglesias, et al., Continuous production of
performance and resistance against deactivation, Catal. Today (2014), http
Fig. 3). Zr-3d XPS spectra evidenced the presence of two groups ofignals corresponding to two different types of zirconium speciesn all the samples. In the case of a fresh sample of Zr-SBA-15 thesewo chemical environments lead to two Zr3d core level signals:
Fig. 3. Zr-3d and O-1S XPS spectra recorded for Zr-SBA-15 catalysts used in methan
PRESSay xxx (2014) xxx– xxx 5
one located at 184.0 eV (and its spin-orbit component at 186.5 eV)and the second one located at 182.9 eV (and its spin-orbit com-ponent at 185.3). The former is usually ascribed to the presenceof isolated Zr4+ species in silica framework [17,18], whereas thelast indicates the presence of zirconium dioxide nanodomains [19].These spectra remain almost unchanged when using the Zr-SBA-15 catalyst in presence of large amounts of sodium and cholesterol(not shown), confirming the absence of interaction between thesesubstances and the zirconium sites. On the contrary, spectra cor-responding to samples used in presence of potassium, retinol andlecithin evidenced some changes with regards to the Zr3d XPS spec-trum achieved for the fresh catalyst. In the case of potassium andretinol, XPS is slightly shifted toward lower binding energies but,in essential, the recorded spectrum remains almost the same thanin the fresh catalyst. A different behavior is found when the cata-lyst is used in presence of lecithin, which leads to a great change inthe chemical environment of zirconium atoms. This is the chemi-cal exerts the highest influence, among the tested substances, onthe modification of the XPS Zr3d core level spectrum of Zr-SBA-15. In this case, both signals, those attributed to isolated zirconiumand zirconia are, by far, more shifted toward lower binding energyvalues than in the rest of the cases under study. This modificationis accompanied by the transformation of the two signals detectedfor Zr-SBA-15 into a single one, suggesting that the interaction of
biodiesel from low grade feedstock in presence of Zr-SBA-15: Catalyst://dx.doi.org/10.1016/j.cattod.2014.01.004
the lecithin with the catalyst occurs through the zirconium atoms.This strong modification of the XPS spectra could be related to theinteraction of hydroxyl functionalities at zirconium sites, leadingto an ionic complexation of phosphate/phosphonate groups with
olysis tests performed in presence of some impurities in the reaction mixture.
a Metal loading calculated by means of ICP-OES.b Acid loading calculated by NH3 temperature programmed desorption analysis.c Specific surface area calculated by the B.E.T. method.
applyiter ob
tczbtactaoarOptnldyaci
i–tbsciemsfsUotb
3p
piv1stthca
from mixing AF-1 and WCO.The properties of the product stream (after a simple wash-
ing stage with ultra-pure water) were assessed according to theEuropean Standard UNE-EN 14214 and listed in Table 3. FAME
Table 3Quality parameters of biodiesel samples analyzed by ISO 14214 standard.
d Total pore volume recorded at p/p0 = 0.985.e Mean pore size calculated as the maximum of the B.J.H. pore sizes distribution
f Unit cell size calculated as 2/(√
3·d100), where d100 is the Bragg’s lattice parame
he zirconium atoms, in a similar way to those described by Blan-hard and co-workers on the immobilization of phospholipids ontoirconated surfaces [20–22]. This interaction has been described toe energetically favorable and chemically stable, and it supportshe previous mentioned conclusions about the preferential inter-ction of phosphate ion with the zirconium sites to be the majorause of catalyst deactivation in presence of lecithin. With regardso the influence of the different substances in O 1S XPS spectra,gain two different environments were detected, one attributed toxygen atoms in silica (∼532.5 eV) and the other one to oxygentoms bonded to zirconium sites (∼530.5 eV). In this case, unlikeetinol and lecithin, potassium does not exert any influence on the
1s XPS spectra. On the other hand, organic impurities lead to a dis-lacement of both signals to lower energy bindings, indicating thathe interaction of these substances with the catalyst surface occurs,ot only with zirconium sites, but also with the silica matrix, at
east with the oxygen atoms of the catalyst support. In this way, theeactivation phenomena detected when performing the methanol-sis tests in presence of retinol and lecithin can be ascribed to thedsorption of these substances onto the catalyst support – in thease of retinol – as well as on the catalytic acid sites – clearly evidentn the case of lecithin.
The individual study of the different impurities accompany-ng low grade oleaginous feedstock indicated that some of them
potassium chloride, retinol and lecithin – caused catalyst deac-ivation by different reasons. Neutralization of Brønsted acid sitesy ion-exchange was ascribed to be the main influence of potas-ium chloride. On the contrary, adsorption onto the surface of theatalyst, as well as onto the catalytic sites, was the major causen the case of unsaponifiable organic compounds. However, thextension of the catalyst deactivation was not large in the case ofethanolysis tests performed in presence of real low-grade feed-
tock. Moreover, calcination revealed to be a suitable method toully recover the starting catalytic activity of Zr-SBA-15 material,uggesting that most of the catalyst deactivation was reversible.nder these assumptions, a deeper insight on the catalytic stabilityf Zr-SBA-15 material in biodiesel synthesis was accomplished, thisime by performing the methanolysis tests in a continuous packeded reactor.
.2. Catalytic tests in presence of low grade feedstock: assays in aacked bed reactor
Table 2 depicts the physico-chemical properties of both theowder and the agglomerated Zr-SBA-15 material to be used
n the fixed bed reactor. Whereas the powder sample displaysalues for the textural and structural parameters typical from SBA-5 mesostructured materials, the agglomerated sample displayslightly reduced textural properties, acid loading and metal con-ent, most likely due to the dilution of the Zr-SBA-15 material with
Please cite this article in press as: J. Iglesias, et al., Continuous production of
performance and resistance against deactivation, Catal. Today (2014), http
he clay used as binding agent [7]. However, previous experimentsave demonstrated that the observed reductions in the physico-hemical properties are not high enough to jeopardize the catalyticctivity of the Zr-SBA-15 material in a packed bed reactor.
ng the K.J.S. correction.tained as (d100 +
√3·d110 +
√4·d200)/3.
The catalytic activity and stability of Zr-SBA-15 material wasassessed in the transformation of category-1 animal fat and wastecooking oil mixtures, ranging from 10 to 30 wt% animal fat in theoverall mixture. Since category-1 animal fat is a solid under ambi-ent conditions, this was mixed with WCO, after melting at 65 ◦C,prior to its injection in the fixed bed reactor. Results from thecatalytic transformation of these mixtures are depicted in Fig. 4.Bentonite agglomerated Zr-SBA-15 material displays an excellentcatalytic behavior in terms of activity and stability in the produc-tion of biodiesel from AF-1:WCO mixtures during, at least, 45 hon stream experiments. This catalytic activity and stability is evenmore outstanding considering that the different experiments wereaccomplished in presence of the same catalyst sample. After almost150 h on stream, the Zr-SBA-15 material provided a product streamwith 96% FAME yield on average, without evidence of catalystpoisoning, even although the amount of sodium, potassium, phos-phorus and unsaponifiable matter, was progressively incrementedin the feed stream insofar as the amount of AF-1 was increased inthe mixture. This outstanding stability of the agglomerated mate-rial in continuous operation mode in the fixed bed reactor could beascribed to the bentonite used as binding agent. Clays have beendescribed to be excellent adsorbents for the purification of veg-etable oils [23], being able to retain inorganic metal ions – suchas iron and copper – phosphorus as well as some unsaponifiableisoprenoids and chlorophyll [24–26]. Bearing in mind this adsorp-tion activity of clays, bentonite could act as a poison scavenger,preventing the deactivation of the Zr-SBA-15 materials by captur-ing the different substances before these reach the catalytic sites. Aprove of this behavior can be the low phosphorus content found inthe product stream, which is substantially lower than that coming
biodiesel from low grade feedstock in presence of Zr-SBA-15: Catalyst://dx.doi.org/10.1016/j.cattod.2014.01.004
J. Iglesias et al. / Catalysis Today xxx (2014) xxx– xxx 7
F ing from condi
c1oart1csurmKsppaadt
4
efamopBoulizt
ig. 4. FAME yield obtained in the transesterification of AF-1/WCO mixtures (rangaterial in a packed bed reactor operating under steady-state conditions. Reaction
ontent in final product was close to that specified in UNE-EN4214 (96.5 wt%), an outstanding result in view of the propertiesf the used low-grade raw materials. In the case of the densitynd viscosity, values are within the range set by the standard,egardless of the composition of the starting raw material. Notehat both parameters decrease with increasing the amount of AF-
in the initial mixture, due to its lower density and viscosity asompared to WCO, so that there is a clear direct influence of thetarting feedstock on the final properties of the biodiesel prod-ct. Flash points were above 160 ◦C, discarding the presence ofesidual alcohol, also confirmed by the direct measurement ofethanol content (<0.2 wt%). Phosphorus and other metals (Na,
, Ca, Mg) contents, were lower, in every sample, than thoseet by the UNE-EN 14214. In conclusion, the obtained biodieselroducts meet most of the requirements established by the Euro-ean standard. Values exceeding the limit could be solved with
small addition of additives, or mixing low-grade raw materi-ls with oils and fats with complementary properties, or blendingifferent biodiesel batches so that the final blend fulfills specifica-ions.
. Conclusions
Zirconium-containing SBA-15 materials have revealed to bexcellent catalysts for the transformation of low-grade oleaginouseedstock into biodiesel through the methanolysis of triglyceridesnd free-fatty acids present in the raw materials. However, theseaterials seem to be sensitive to the presence of certain kinds
f impurities, present in waste lipids, which can act as catalystoisons. Alkaline metals, such as potassium, could interact withrønsted acid sites causing a partial, but limited, deactivationf these heterogeneous acid catalysts. In a similar way, organicnsaponifiable compounds, such as retinoids or phospholipids –
Please cite this article in press as: J. Iglesias, et al., Continuous production of
performance and resistance against deactivation, Catal. Today (2014), http
ike retinol and lecithin, respectively – strongly interact with the sil-ca surface – the catalyst matrix – in the case of retinoids, or with theirconium sites, like lecithin does. These interactions cause the par-ial blockage of the zirconium acid sites preventing the reactants to
m 10 to 30 wt% in AF-1) with methanol performed over agglomerated Zr-SBA-15tions: 209 ◦C, 50 methanol:oil molar ratio, and 30 min residence time.
access them and avoiding the progress of the chemical transforma-tion, being one of the most probable cause of catalyst deactivationwhen these substances are present. On the contrary, methanolysiscatalytic tests performed in a fixed bed reactor suggested a strongerresistance of Zr-SBA-15/bentonite pellets against catalyst deacti-vation. These pellets are able to provide a sustained FAME yieldover 96% during more than 150 h on stream, even starting fromsuch a low grade raw material as WCO/AF-1 mixtures containingup to 30 wt% animal fat. The reason for the higher resistance againstdeactivation could be ascribed to the clay used as binding agent– bentoninte – which could act also as poison scavenger duringthe reaction, avoiding the access of poisons to the catalytic acidsites.
Acknowledgements
The financial support from the Spanish Ministry of ScienceandInnovation through the project CTQ2008-01396 and fromthe Regional Government of Madrid through the project S2009-ENE1743 are gratefully acknowledged. RSV also thanks the SpanishGovernment for a FPI grant. Mr. Mark Isaacs (Cardiff School of Bio-sciences, Cardiff University) is kindly acknowledged for running theXPS experiments used in this work.
References
[1] J.A. Melero, J. Iglesias, G. Morales, Green Chem. 11 (2009) 1285–1308.[2] Y.C. Sharma, B. Singh, J. Korstad, O. Roberts, Biofuel Bioprod. Bior. 5 (2011)
69–92.[3] C.V. McNeff, L.C. McNeff, B. Yan, D.T. Nowlan, M. Rasmussen, A.E. Gyberg, B.J.
Krohn, R.L. Fedie, T.R. Hoye, Appl. Catal. A Gen. 343 (2008) 39–48.[4] M. di Serio, S. Mallardo, G. Carotenuto, R. Tesser, E. Santacesaria, Cat. Today 195
(2008) 54–58.[5] Y. Cheng, Y. Feng, Y. Ren, X. Liu, A. Gao, G. He, F. Yang, J. Li, Bioresour. Technol.
biodiesel from low grade feedstock in presence of Zr-SBA-15: Catalyst://dx.doi.org/10.1016/j.cattod.2014.01.004
[8] J.A. Melero, J. Iglesias, J. Sainz-Pardo, P. de Frutos, S. Blázquez, Chem. Eng. J. 139(2008) 631–641.
[9] B.D. Wahlen, B.M. Barney, L.C. Seefeldt, Energy Fuel. 22 (2008) 4223–4228.10] G. Morales, L.F. Bautista, J.A. Melero, J. Iglesias, R. Sánchez-Vázquez, Bioresour.
Technol. 102 (2011) 9571–9578.11] J.A. Melero, L.F. Bautista, J. Iglesias, G. Morales, R. Sánchez-Vázquez, Cat. Today
195 (2012) 45–53.12] S.C. Yap, Y.M. Choo, C.K. Ooi, A.S.H. Ong, S.H. Goh, Elaeis 2 (1991) 369–378.13] Y. Su, H. Cui, Q. Li, S. Gao, J.K. Shang, Water Res. 47 (2013) 5018–5026.
Please cite this article in press as: J. Iglesias, et al., Continuous production of
performance and resistance against deactivation, Catal. Today (2014), http
14] D.M. Do, S. Jaenicke, G.-K. Chuah, Catal. Sci. Technol. 2 (2012) 1417–1424.15] S.P. Pinho, E.A. Macedo, J. Chem. Eng. Data 50 (2005) 29–32.16] N.G. Eskandar, S. Simovic, C.A. Prestidge, Int. J. Pharm. 376 (2009) 186–194.17] A. Ramanthan, M.C. Castro-Villalobos, C. Kwakernaak, S. Telalovic, U. Hanefeld,
Chem. Eur. J. 14 (2008) 961–972.
[[
[[
PRESSay xxx (2014) xxx– xxx
18] L. Armelao, C. Eisenmenger-Sittner, M. Groenewolt, S. Gross, C.Sada, U. Schubert, E. Tondello, A. Zattin, J. Mater. Chem. 15 (2005)1838–1848.
19] E. Rodríguez-Castellón, A. Jiménez-López, P. Maireles-Torres, D.J. Jones, J. Roz-iére, M. Trombetta, G. Busca, M. Lenarda, L. Storaro, J. Solid State Chem. 175(2003) 159–169.
20] P. Kohli, G.J. Blanchard, Langmuir 16 (2000) 695–701.21] M. Mazur, P. Krysinski, G.J. Blanchard, Langmuir 21 (2005) 8802–8808.22] B.P. Oberts, G.J. Blanchard, Langmuir 25 (2009) 13918–13925.
biodiesel from low grade feedstock in presence of Zr-SBA-15: Catalyst://dx.doi.org/10.1016/j.cattod.2014.01.004
23] W.L. Siew, K.Y. Chah, J. Oil Palm Res. 19 (2007) 356–363.24] M.A. Didi, B. Makhoukhi, A. Azzouz, D. Villemin, Appl. Clay Sci. 42 (2009)
336–344.25] F. Hussin, M.K. Aroua, W.M.A.W. Daud, Chem. Eng. J. 170 (2011) 90–106.26] E.L. Foletto, D. Silva Paz, A. Gündel, Appl. Clay Sci. 83-84 (2013) 63–67.
