Metallic monolithic catalysts based on calcium and cerium for the production of biodiesel

Fuel 182 (2016) 668–676

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Metallic monolithic catalysts based on calcium and cerium for the production of biodiesel Inés Reyero a, Ainara Moral a, Fernando Bimbela a,b, Jelena Radosevic c, Oihane Sanz c, Mario Montes c, Luis M. Gandía a,b,⇑ a Grupo de Reactores Químicos y Biorreactores, Departamento de Química Aplicada, Universidad Pública de Navarra, Edificio de los Acebos, Campus de Arrosadía, 31006 Pamplona, Spain b Institute for Advanced Materials (InaMat), Universidad Pública de Navarra, Campus de Arrosadía, 31006 Pamplona, Spain c Departamento de Química Aplicada, UFI 11/65, Facultad de Ciencias Químicas de San Sebastián, UPV/EHU, Paseo Manuel de Lardizábal 3, 20018 San Sebastián, Spain

a r t i c l e

i n f o

Article history: Received 9 May 2016 Received in revised form 5 June 2016 Accepted 7 June 2016

Keywords: Biodiesel Calcium oxide Cerium oxide Metallic monoliths Methanolysis

a b s t r a c t The present work reports the preparation, characterization and testing of Ca/Ce oxides as heterogeneous catalysts for the transesterification of sunflower oil with methanol (methanolysis), both in powder and structured forms, to produce biodiesel. A series of Ca-based catalysts in powder form were prepared on four different supports (commercial Al2O3, SiO2 and CeO2, along with in-house prepared CeO2) following different techniques. The best catalyst formulation in terms of activity and stability was a Ca/Ce mixed oxide (20 wt% CaO) catalyst prepared under the metallic citrates decomposition technique. Different suspensions could be formulated using this catalyst for further washcoating FecralloyÒ monoliths. The effect of the solvent (water or alcohols) and of the use of additives (polyvinyl alcohol, polyvinylpyrrolidone, and colloidal Al2O3 and CeO2) on the catalytic performance of the catalysts in the transesterification reaction was studied. The best results could be obtained for the structured catalysts prepared using suspensions having isopropanol as solvent medium and 1% of polyvinylpyrrolidone. The monoliths prepared using this formulation yielded the best oil conversions after a second reaction cycle reported so far in the literature concerning the use of structured catalysts for biodiesel production, with oil conversion values of 70% after 6 h of reaction at 60 °C using recovered and thermally regenerated monoliths. The thermal reactivation of the catalysts has been proven to be crucial in order to partially recover the catalyst activity, though significant leaching of the active catalytic layer was found to occur during the first reaction cycle. The results presented here are remarkably superior to the best ones previously reported in the literature concerning the use in a second reaction cycle of structured methanolysis catalysts at atmospheric pressure. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Certain scientific topics suffer from a sudden outburst of research interest by a wide part of the scientific community. Biodiesel production via heterogeneous catalytic transesterification of triglycerides is one of the exemplifying cases. In spite of the very abundant literature concerned with this topic that has appeared in the recent years [1,2], there is still a lack of systematic research studies focusing on the main challenges to be faced such as e.g. the issue of the poor chemical stability of the catalysts proposed

⇑ Corresponding author at: Institute for Advanced Materials (InaMat), Universidad Pública de Navarra, Campus de Arrosadía, 31006 Pamplona, Spain. E-mail address: [email protected] (L.M. Gandía). http://dx.doi.org/10.1016/j.fuel.2016.06.043 0016-2361/Ó 2016 Elsevier Ltd. All rights reserved.

for the production of biodiesel which complicates their reutilization [3–8]. In a recent review by Mittelbach the challenges that the biodiesel industry has to face nowadays have been clearly exposed [9], and definitely process intensification has to be addressed in order to achieve the necessary technology improvement mentioned therein [10,11], an improvement capable of granting lower energy input, higher overall yields and less waste material in the process. As accurately pointed out by Mittelbach, biodiesel is still the cheapest alternative to fossil diesel fuel, and it must be borne in mind that the transition period from fossil-derived energy sources to a 100% non-fossil energy mix scenario is neither going to be easy nor imminent. Structured catalytic reactors such as those based on monolithic catalysts constitute an interesting approach in terms of process

I. Reyero et al. / Fuel 182 (2016) 668–676

intensification of biodiesel production via transesterification with methanol (methanolysis) of triglycerides. Such catalytic reactors are hydrodynamically superior compared to those based on particulate catalyst formulations, and certain basic operation units, namely catalyst separation when using slurry reactors, can be avoided. This could reduce the production costs and improve the quality of both biodiesel and glycerol in industrial operation, yet only a few studies can be found in the literature on the use of structured methanolysis catalysts [12–19]. For structured catalysts to reach a successful implementation in biodiesel production, some challenges still need to be tackled and given proper solutions. Particularly, adherence of the active phase onto the support or substrate is a key issue where there is still a long path to be covered before the problem is finally solved. In this regard, leaching phenomena have been discussed in some studies using Ca-based heterogeneous catalysts in powder form. However, little is still known on the performance of structured catalytic systems containing CaO as active phase. It should be noted that CaO exhibits an outstanding activity in alcoholysis reactions with triglycerides which, along with other advantages such as wide availability, low cost and even ‘‘renewable” origin justify the interest in further investigations. In a previous work, the use of Mg-Al hydrotalcites deposited on FecralloyÒ metallic monoliths by the washcoating method was studied for the methanolysis of sunflower oil [12]. Adherence of the active phase onto the substrate was found to be challenging even though some works in the literature have astonishingly achieved CaO adherences on cordierite monoliths up to 99% [16] in spite of the stability problems exhibited by this active phase related to the formation during reaction of calcium glyceroxide [7,8,20]. Our previous study revealed that adding small amounts (5–10 wt%) of sepiolite as additive into the slurries during catalyst preparation was positive for enhancing adherence. Up to 77% oil conversion could be attained after 10 h of reaction at usual methanolysis conditions, though catalyst stability remained unsolved because the catalyst could not be reutilised after its recovery [12]. Calcium oxide from a variety of both mineral and organic sources is very likely the most investigated solid as methanolysis catalyst [7,8,20–22]. As far as we know, this compound has always been used as methanolysis catalyst in powder form. Only Kwon et al. describe the use of CaO supported over c-alumina washcoated cordierite honeycomb monoliths for fatty acid methyl esters production [15]. However, CaO has poor chemical stability under transesterification reaction conditions, resulting in the leaching of calcium species into the reaction mixture [7,8,20,23]. Several approaches have been followed in order to improve the stability of calcium such as the modification of CaO with metals and the use of perovskites containing Ca. Significant enhancement on the catalyst reusability has been achieved with CaO-CeO2 mixed oxides [24,25] and CaO supported on Ce-based materials [26]. In this work, a series of CaO-containing solids in powder form were prepared under different techniques aiming at achieving catalytic formulation that could eventually be capable of yielding both high catalytic activity and stability. Thus, the first part of the study aimed to identify those catalysts prepared under different techniques that presented promising results in the methanolysis of sunflower oil. Only those samples yielding superior catalytic performances were further studied in the preparation of the structured catalysts. To overcome the adherence and stability issues, different strategies such as using different solvents during the stage of catalyst deposition onto the monoliths, the use of different additives and regeneration by means of thermal reactivation have been studied.

669

2. Materials and methods 2.1. Catalysts preparation Different catalysts were prepared, both in powder form and structured on monolithic substrates. All the samples considered in this study together with the nomenclature used and relevant information regarding their preparation and catalytic testing are compiled in Table 1. A series of calcium catalysts was prepared on four different supports following the incipient wetness impregnation technique; the Ca content in this series was fixed at 10wt%. Calcium nitrate (Ca(NO3)2
4H2O, Sigma Aldrich, 99.0% min.) was selected as Ca precursor. Commercial Al2O3 (Spheralite 505, Procatalyse), SiO2 (Aerolyst 350, Degussa) and CeO2 (99.995% min., Sigma-Aldrich) were directly used as supports. Another ceria support was synthesized in the laboratory from cerium nitrate (Ce(NO3)3
6H2O, Merck, 98.5% min.) by precipitation of cerium hydroxide at 90 °C using NH4OH solution (Merck) as precipitating reagent until a final pH of 8.5 was attained, followed by ageing at 90 °C for 4 h, drying at 100 °C for 12 h and calcination at 700 °C for 5 h in a muffle furnace. On the other hand, a series of Ca-Ce oxides could be prepared by the method of metallic citrates decomposition, based on the reaction of cerium nitrate with citric acid (anhydrous, Panreac, 99.5% min.), following a procedure similar to that described in Samantaray et al. [27]. Four different solids having different calcium contents (0, 5, 10 and 20 wt%) and an additional CaO solid could be prepared following this procedure. Ca-Ce oxides could also be prepared by direct physical mixing of the corresponding nitrate salts and further calcination, following two different procedures. In the first case, the necessary amounts of the metallic nitrate salts were dissolved in aqueous medium and maintained under vigorous stirring for 5 h. Afterwards, the water from the solution was evaporated and the resulting solids were dried at 100 °C for 12 h and subsequently calcined at 700 °C for 5 h. The solids were prepared so as to have a nominal content of CaO in the final solid of 10 wt%. In the other case, the necessary amounts of both nitrate salts were mechanically mixed in a mortar and were directly calcined at 1050 °C for 4 h. Regarding the structured catalysts, a commercial metallic alloy (FecralloyÒ, FeCr22Al5, 50 lm, Goodfellow) served for the preparation of the monoliths as described in Reyero et al. [12]. A washcoating technique was selected for depositing Ca and Ce from a suspension prepared with a mixed Ca-Ce oxide prepared again by the method of metallic citrates decomposition. In this case, the mixed oxide was formulated in order to have a CaO content of 20 wt%. Different suspensions were prepared using this mixed Ca-Ce oxide (20Ca/Ce-cit), both in aqueous and in alcoholic media. The alcohols used for preparing the suspensions were methanol, ethanol, and isopropanol of high purity (Scharlau, anhydrous, 99.8% min.). The solids from these suspensions (slurried catalyst) were also tested after solvent evaporation to compare their activity to that from the original solid precursor (20Ca/Ce-cit). Furthermore, in various formulations different additives were added into the suspensions in order to enhance their adherence properties. Different combinations were tested using 1% polyvinyl alcohol (PvOH, Sigma-Aldrich, 99% min.) or 1% polyvinylpyrrolidone (PVP, average molecular weight of 40,000 g/mol, Sigma-Aldrich) and/or 20% colloidal ceria (NYACOLÒ). Additional formulations were prepared using 20% colloidal alumina (NYACOLÒ) together with 1% PvOH, and with two other PVP contents in isopropanol (0.5 and 2%). After washcoating, the slurry excess was eliminated by centrifugation and then the solids were dried at 120 °C for 30 min and calcined at 700 °C for 5 h.

670

I. Reyero et al. / Fuel 182 (2016) 668–676

Table 1 List of catalysts prepared both in powder and in structured form and catalytic tests carried out. Catalyst sample

Nominal Ca content (wt %)a

Support

Preparation technique

Catalytic tests

Powder catalysts Ca/Al

10

Al2O3 (Spheralite 505, Procatalyse)

Incipient wetness impregnation

Transesterification of sunflower oil in a slurry batch reactor [8] at 60 °C, 2 wt% of catalyst loading, methanol-to-oil molar ratio of 12

Ca/Si Ca/Ce-comm Ca/Ce-precip

10 10 10

Incipient wetness impregnation Incipient wetness impregnation Incipient wetness impregnation

nCa/Ce-citb

0, 5, 10 and 20

SiO2 (Aerolyst 350, Degussa) CeO2 (Sigma-Aldrich) CeO2, prepared by precipitation using NH4OH (final pH = 8.5) CeO2, prepared by the metallic citrates decomposition method

CaO-cit Ca/Ce-nit

100 10

n/a Ca-Ce mixed oxides

Ca/Ce-nit_solids

10 and 25

Ca-Ce mixed oxides

Suspension

Nominal Ca content (wt %)a

Solvent medium

Incipient wetness impregnation

Additional second reaction cycle using 5Ca/Ce-cit, 10Ca/Ce-cit, and 20Ca/Ce-cit. A third reaction cycle was carried out with 20Ca/Ce-cit.

Thermal decomposition of the citrate precursor Thermal decomposition of the resulting precipitate from a solution containing the nitrate salt precursors Direct thermal decomposition of the physically mixed nitrate salt precursors in powder form Additives

Catalytic tests

Powder solids from catalytic suspensions S-w 20

Aqueous

S-w-PvOHcCeO2 S-w-PvOHcAl2O3 S-m S-e S-i S-i-PVP S-i-cCeO2 S-i-PVPc CeO2

20 20 20 20 20 20 20 20

Aqueous Aqueous Methanol Ethanol Isopropanol Isopropanol Isopropanol Isopropanol

1% PvOH + colloidal 20% CeO2 1% PvOH + colloidal 20% Al2O3

Monolith sample

Nominal Ca content (wt %)a

Suspension

Solvent medium and additives

Catalytic tests

Structured catalystsc M-w-PVOHcCeO2

20

s-apc

Aqueous, 1% PvOH + colloidal 20% CeO2

Transesterification of sunflower oil in a monolithic stirrer batch reactor [12] at 60 °C, 2 wt% of catalyst loading and methanol-to-oil molar ratio of 48

M-i M-i-PVP-zd M-i-PVPcCeO2

20 20 20

s-i s-ipvp s-ipvpc

Isopropanol, none Isopropanol, z% PVP Isopropanol, 1% PVP + colloidal 20% CeO2

Same as for the powder catalysts (see above). A second reaction cycle was carried out with the recovered solids obtained from the S-m, S-i, S-e, S-i-PVP, Si-cCeO2 and S-i-PVPc CeO2 suspensions.

1% PVP colloidal 20% CeO2 1% PVP + colloidal 20% CeO2

a

Expressed as CaO. n = 5, 10 or 20, depending on the nominal CaO content (5, 10 or 20 wt%, respectively). c All the structured catalysts were prepared by the washcoating method, using different suspensions containing the 20Ca/Ce-cit solids and FecralloyÒ, FeCr22Al5, 50 lm, Goodfellow for the preparation of the monoliths. d z = 0.5, 1 or 2% PVP added into the suspension. b

2.2. Catalysts characterization and adherence tests

2.3. Catalyst testing

The catalysts were characterized under different techniques, including N2 adsorption, X-ray diffraction (XRD), temperatureprogrammed desorption of CO2 (CO2-TPD), scanning electron microscopy (SEM) and basic strength measurements using Hammett indicators. In the case of structured catalysts, adherence tests were performed by immersing the prepared catalytic monoliths in HPLC-grade methanol and subjecting the samples to ultrasound at room temperature in an ultrasonic bath for 30 min. Afterwards, the recovered samples were subjected to calcination at 700 °C for 5 h and the remaining mass of catalyst deposited on the monoliths surface was calculated.

Different methanolysis reactions using commercial refined sunflower oil (Urzante, Spain) as raw material were carried out in order to evaluate the catalytic performance of the different catalysts. HPLC-grade methanol was used for the transesterification reactions, which were conducted in all cases at 60 °C and catalyst loading of 2 wt% referred to the mass of sunflower oil. The methanol-to-oil molar ratio was set at a value of 12 in the methanolysis reactions carried out using powder solids and at a value of 48 in the case of structured catalysts. The high methanolto-oil ratio used with the monolithic catalysts was needed to guarantee a minimum volume of liquid that allowed to submerge the monoliths in the reaction mixture to have a good liquid-solid

671

I. Reyero et al. / Fuel 182 (2016) 668–676

sunflower oil conversions of ca. 97% at the same operating conditions. However, it must be borne in mind that the catalysts in this work have a CaO loading of only 10 wt% so their performance as methanolysis catalysts is very interesting. Methanolysis runs at the same operating conditions using only the pure CeO2 supports and the physically mixed Ca/Ce oxides permitted to discard any significant catalytic activity of these materials. All these spent samples from the preliminary tests could be recovered and tested in a second reaction cycle. None of them showed any noticeable activity through 4 h of reaction at the above detailed operating conditions. This dramatic loss of activity could be attributed to leaching of Ca into the reaction medium. Thus, these powders were not further taken into consideration for the ensuing study with the structured catalysts although they allowed to confirm the potential of the Ca-Ce catalytic system so a series of Ca-Ce mixed oxides were prepared and tested. The catalytic performance of these powder catalysts is presented in Fig. 1. From the results it is clear that these catalysts were active in the methanolysis reaction, though it can be remarked that the activity of the samples prepared through the citrate method (XCa/Ce-cit, X = 5, 10, 20 wt% Ca) increased with increasing Ca contents. For example, under the reaction conditions studied, oil conversions of 76, 93 and 96.4% could be achieved after 120 min with 5Ca/Ce-cit, 10Ca/Ce-cit and 20Ca/Ce-cit, respectively. The activity of the solids obtained from the decomposition of the metallic citrates was higher than that of the ones synthesized through the decomposition of the nitrates. As a matter of fact, all the samples prepared after calcining the metallic nitrates at 1050 °C were inactive. Only the sample obtained after calcination at 700 °C of the solid resulting from a solution of the metallic nitrates (10Ca/Ce-nit) was active, giving rise to an oil conversion of 86% (see Fig. 1A). Bulk CaO obtained from calcination of the precursor prepared by the citrates method (CaO-cit) yielded similar results to those reported in previous works [8,12], thus again confirming that the performance of bulk CaO, when calcined at high enough temperatures and if carefully handled after calcination, is hardly affected by any differences regarding the preparation procedure or the Ca precursor [8]. Interestingly, the catalytic activity displayed by the 20Ca/Ce-cit catalysts was only a little worse than that of bulk CaO in the first reaction cycle. The activity of the XCa/Ce-cit solids was additionally evaluated in additional reaction cycles (Fig. 1B). Only the 20Ca/Ce-cit sample maintained a relevant activity in a second reaction cycle so it was

contact. The powder catalysts were tested in a slurry batch reactor; further details along with a detailed description of the handling procedure for transferring the freshly calcined catalysts into the transesterification reactor can be found in a previous work [8]. In the case of the monoliths, a 0.25 L monolithic stirrer batch reactor containing two monoliths was used. A twin pair of catalytic monoliths corresponding to each of the different structured catalysts to be tested, was attached to the stirrer shaft in each catalytic run. The stirring speed was adjusted to 200 rpm, which corresponds to a relative centrifugal force of 1.1 times the standard gravity. The setup includes a sampling line that enables the extraction of liquid samples at different reaction times, which are analysed by Size Exclusion Chromatography (SEC). This enables the assessment of the reaction progress over time and helps determining both the fatty acids conversion and selectivity to the different products over time. More details about the setup and the sampling procedure in particular can be found in a previous work [12].

3. Results and discussion 3.1. Preliminary catalyst screening A preliminary catalyst screening was conducted with both the samples prepared in powder form and with the solids from the suspensions prepared for washcoating the structured catalysts so as to determine which catalytic formulations were efficient in the methanolysis of sunflower oil. The CaO/Al2O3 and CaO/SiO2 catalysts were poorly active, yielding only about 13 and 6% oil conversion, respectively, after 6 h of reaction. This behaviour could be explained by the relatively high activation temperatures that are necessary for preparing these catalysts, together with the foreseeable acid nature of the supports, thus very likely leading to the formation of metallic aluminates and silicates, which are inactive in the methanolysis reaction [3,4]. As for the different CaO/CeO2 powder catalysts, their performance was very similar regardless of the preparation method of the support. After 3 h of reaction, the oil conversion was 85% for the catalyst prepared with the commercial ceria support, 93.7% for the CaO catalyst supported on ceria synthesized through the metallic citrate decomposition method (CaO/Ce-cit) and 96.3% for the CaO catalyst supported on ceria obtained from the decomposition of precipitated cerium hydroxide (CaO/Ce-precip). These values were only slightly lower than those previously reported for bulk CaO catalysts [7,8], which could yield 1

1

Oil conversion

A

B

0.8

0.8

0.6

0.6 st

20Ca/Ce-cit, 1 cycle

0.4

0.4

CaO-cit

5Ca/Ce-cit, 2

nd

5Ca/Ce-cit 10Ca/Ce-cit

0.2

10Ca/Ce-nit 0

0

50

100

150

Time (min)

200

10Ca/Ce-cit, 2

nd

cycle

20Ca/Ce-cit, 2

nd

cycle

20Ca/Ce-cit, 3

rd

cycle

0.2

20Ca/Ce-cit

250

0 0

50

100

cycle

150

200

250

Time (min)

Fig. 1. Evolution of the oil conversion over time for methanolysis reactions using the indicated Ca solids in powder form: (A) first reaction cycle and (B) second and third reaction cycles. Reaction conditions: 60 °C, 2% of catalyst load and 12:1 methanol-to-oil molar ratio.

672

I. Reyero et al. / Fuel 182 (2016) 668–676

Table 2 Specific surface area (SBET) and specific pore volume (Vp) of the different slurried catalysts (20 wt% Ca). Sample

Suspension composition

SBET (m2/g)

Vp (cm3/g)

20Ca/Ce-cit S-w-PvOHcCeO2 S-w-PvOHcAl2O3 S-w S-m S-e S-i S-i-PVP S-i-cCeO2 S-i-PVPcCeO2

Base catalyst Water, 1% PvOH, colloidal 20% ceria Water, 1% PvOH, colloidal 20% alumina Water Methanol Ethanol Isopropanol Isopropanol, 1% PVP Isopropanol, colloidal 20% ceria Isopropanol, 1% PVP, colloidal 20% ceria

22 14 19 14 19 18 18 21 27 26

0.14 0.10 0.16 0.10 0.35 0.14 0.15 0.16 0.21 0.17

selected as solid precursor for further preparing the suspensions required for washcoating the structured catalysts. Interestingly, the activity curve over time yielded in this case is very similar to that of transesterifications catalysed by calcium glyceroxide [7]. It was decided to analyse the performance of this catalyst in a third reaction cycle and a significant loss in activity could be observed, which is indicative of Ca leaching. In order to prove this, the biodiesel samples obtained after the reaction cycles using the 20Ca/Ce-cit catalyst were analysed by atomic emission spectroscopy combined with inductively coupled plasma (ICP-AES), whereas the glycerol phases were subjected to complexometric titration using Na2-EDTA as reagent after eliminating the methanol by rotary evaporation. The analyses revealed that around 6% of the Ca from the fresh catalyst could be dissolved into the reaction medium and ended up mainly in the glycerol phase, though ca. 150 ppm of Ca could also be quantified by ICP-AES in the biodiesel rich phase. This reveals that the leaching of Ca could not be completely avoided. 3.2. Characterization and catalytic activity of the slurried catalysts The dried solids obtained from the suspensions prepared for washcoating the monoliths were analysed by XRD and N2 adsorption/desorption and compared to their calcined precursor in powder form (20Ca/Ce-cit). In the XRD patterns of the different samples, the crystalline phases corresponding to CeO2 could be identified in all cases, whereas CaO was not detected in the solids from the suspensions, but it could be detected in the 20Ca/Ce-cit precursor. Presumably, this could be attributed to a loss of crystallinity upon suspensions preparation. Due to the high CaO loading (nominal content of 20 wt%), even assuming that part of the CaO could be leached upon the preparation of the suspension, the relatively high activity of the catalysts (see discussion below) hints that Ca is still present in significant amounts. Hence, both CaO redispersion and loss of crystallinity during suspensions preparation could explain the absence of CaO diffraction peaks in the XRD pattern of the solids from the suspension. The main textural properties of these solids are compiled in Table 2, which also includes the components of the different suspensions and nomenclature used. The base solid (20Ca/Ce-cit) has a modest surface area (SBET) of 22 m2/g whereas the solids recovered from the suspensions exhibit similar values ranging from 14 to 27 m2/g. In general, the solids from suspensions prepared with isopropanol and containing additives such as PVP and colloidal ceria showed the highest values of SBET and specific pore volume. As for the catalytic activity of the slurried catalyst, the results of the sunflower oil conversion over time obtained in the different methanolysis runs are shown in Figs. 2 and 3. For the sake of clarity, all the runs corresponding to the slurried catalysts from suspensions prepared in different solvent media have been

grouped into those derived from aqueous (Fig. 2A) or alcoholic media (Fig. 2B), whereas Fig. 3 has been solely dedicated to depicting the performance of the different slurried catalysts from suspensions prepared in isopropanol, which will be later discussed. The results from Fig. 2 clearly show a significant diminution in the activity of the slurried catalysts prepared in aqueous medium compared to that of the precursor 20Ca/Ce-cit. This is far more evident in the case of the slurried catalysts having colloidal Al2O3 as additive (S-w-PvOHcAl2O3), which could be attributed to the acid nature of the additive. Such acid nature could neutralise the basicity of the active phase, CaO, or even react with it, resulting in an additional loss of activity. As a matter of fact, the supported CaO/Al2O3 catalyst was poorly active as mentioned before. Regarding the slurried catalysts prepared on alcoholic media, all of them outperformed the slurried catalysts from aqueous suspensions, displaying very similar activities between them over time during the first reaction cycle. However, the best catalytic performance, which is essentially that of the original 20Ca/Ce-cit catalyst, is that of the slurried catalysts prepared in isopropanol (S-i). Regarding the different solids prepared in isopropanol (Fig. 3), again the slurried catalysts corresponding to the S-i suspension yielded the best catalytic activity after 120 min of reaction time, with oil conversions above 95%. According to the results in Table 2, the slurried catalyst obtained using isopropanol as solvent have specific surface areas slightly higher than that of the base catalysts or the solids obtained using water, but the difference in surface area cannot justify alone the different catalytic performance exhibited by these materials. When the samples were recovered and reutilised in a second reaction cycle, the differences between the different samples became remarkably augmented, and only the solids slurried in isopropanol yielded a reasonable activity. Again, the S-i sample outperformed the rest, achieving oil conversions around 93% after 180 min, while the rest of the slurried catalysts could only achieve oil conversions of 67% (S-i-PVP), 39% (S-i-cCeO2) and 30% (S-i-PVPcCeO2) after 180 min. Neither the presence of colloidal CeO2 nor the addition of PVP to the suspensions, added both together or separately, could enhance the activity of the slurried catalysts without additives (S-i). 3.3. Ca/Ce monolithic catalysts Monolithic Ca/Ce catalysts were prepared by washcoating using the suspensions S-w-PvOHcCeO2, S-i, S-i-PVP, S-i-cCeO2, and S-i-PVPcCeO2 which resulted in the monoliths M-w-PvOHcCeO2, M-i, M-i-PVP, M-i-cCeO2, and M-i-PVPcCeO2, respectively. In the case of M-i-PVP, three PVP contents, namely 0.5, 1 and 2% were used in the washcoating suspensions that were denoted as M-i-PVP-0.5, M-i-PVP-1 and M-i-PVP-2, respectively. SEM images (Fig. 4) of the metallic surface of the support prior to the washcoating revealed a smooth surface both in the outside

673

I. Reyero et al. / Fuel 182 (2016) 668–676

Oil conversion

1

1

A

0.8

0.8

0.6

0.6

0.4

B

0.4

20Ca/Ce-cit S-w-PvOHcCeO

20Ca/Ce-cit S-m S-e S-i

2

0.2

0.2

S-w-PvOHcAl O 2

3

S-w 0 0

50

100

150

200

250

Time (min)

0

0

50

100

150

200

250

Tiempo (min)

Fig. 2. Evolution of the oil conversion over time using the slurried catalysts prepared on: (A) aqueous suspensions. (B) Alcoholic suspensions. Open symbols: second reaction cycle. Reaction conditions: 60 °C, 2% catalyst load and 12:1 methanol-to-oil molar ratio. See Table 1 for the nomenclature used.

1

Oil conversion

0.8

0.6 20Ca/Ce-cit

0.4

S-i S-i-PVP

0.2

S-i-cCeO

2

S-i-PVPcCeO

2

0

0

100

200

300

400

Time (min) Fig. 3. Evolution of the oil conversion over time using the slurried catalysts prepared using isopropanol. The open symbols correspond to a second reaction cycle. Reaction conditions: 60 °C, 2% catalyst load and 12:1 methanol-to-oil molar ratio. See Table 1 for the nomenclature used.

and in the inner walls of the monoliths. After washcoating, the surface did no longer present a smooth appearance, and a reasonably homogeneous layer of catalytic material could be observed. Hence, the SEM images of the washcoated monoliths confirmed that the metallic surface were successfully covered by a uniform catalytic layer. Then, the different structured catalysts were subjected to an adherence test of the catalytic layer. The results showed a remarkable adherence of the monolith prepared using the aqueous suspension with additives, with almost 98% of the deposited catalytic layer still maintained on the monolith surface after the test. The presence of additives, particularly PVP, was also very important when isopropanol was used as solvent medium. The presence of PVP allowed to enhance the adherence from about 40% for M-i and M-i-cCeO2 up to values of 80% for M-i-PVP-1 and 89% for M-i-PVPcCeO2. These results were used as a criterion for selecting the monoliths to be tested in the methanolysis reaction. The evolution of the oil conversion over time in the transesterification reactions using the different washcoated monoliths is presented in Fig. 5. Unfortunately, the monoliths that exhibited the best adherence (M-w-PvOHcCeO2) showed a very poor catalytic performance, with less than 5% oil conversion even after 360 min of reaction time. This activity is much worse than the one of the

solids obtained from the S-w-PvOHcCeO2 suspension (see Fig. 2A). Contrarily, the monoliths prepared using isopropanol and PVP in the washcoating suspension yielded acceptable results, so oil conversions above 90% could be achieved after 240 min of reaction (see Fig. 5A) despite the variability of the results that becomes apparent from the corresponding duplicated runs. This variability on the replicate runs is also present when analysing the different amounts of PVP added into the suspensions (Fig. 5B), thus complicating to draw any conclusions regarding the effect of the content of PVP although it seems that its effect is positive within the range of values considered in this study. CO2-TPD measurements were conducted with selected solids evaluated in the methanolysis reactions. The TPD patterns are depicted in Fig. 6. The three samples presented a major desorption peak around 475–480 °C. Interestingly, the slurried catalyst obtained from the aqueous suspension did not present any other peaks, while the sample obtained from the suspension having isopropanol and PVP presented two other peaks at 99 and 706 °C, and in the case of the isopropanol suspension using PVP and colloidal ceria as additives also two additional peaks were found at 121 and 718 °C. The total amounts of CO2 adsorbed on these samples were 0.045 mmol/g (S-w-PvOHcCeO2), 0.079 mmol/g (S-i-PVP) and 0.095 mmol/g (S-i-PVPcCeO2). From these results, it is clear that the basicity decreases if water is used as solvent whereas it increases if the solvent is changed to isopropanol and colloidal ceria is present in the suspension. The basic character of CeO2 would explain the increase in basicity observed. However, the basicity values are not so largely different as to explain the very different catalytic performance exhibited by these solids (see Figs. 2 and 3). It seems likely that the basic strength of the active sites plays also an important role. It should be noted in this regard that only the S-i-PVP and S-i-PVPcCeO2 slurried solids show CO2 desorption peaks at high temperatures (706 and 718 °C), which can be tentatively attributed to strong basic sites that seem to be absent in the catalyst prepared using water as solvent. These results could justify the good catalytic performance of the monoliths prepared using isopropanol, PVP and colloidal ceria in the washcoating suspensions. Furthermore, the use of colloidal ceria as additive in the preparation of the alcoholic suspension resulted in an increase of the suspension’s basicity with the ensuing improvement of the monoliths’ basic properties. A second reaction cycle could be carried out using the monoliths that have yielded the best results in the first reaction cycle. The monolith samples were recovered from the methanoly-

674

I. Reyero et al. / Fuel 182 (2016) 668–676

a) catalytic layer metallic substrate

b)

c)

e)

d)

Fig. 4. SEM images of the washcoated monoliths (a) M-w-PvOHcCeO2, (b) M-i, (c) M-i-PVP, (d) M-i-cCeO2, and (e) M-i-PVPcCeO2.

1

1

B

Oil conversion

A 0.8

0.8

0.6

0.6

0.4

0.4

M-w-PVOHcCeO

M-i-PVP-1

2

M-i-PVP-1 0.2

0

M-i-PVP-0.5 0.2

M-i-PVPcCeO

2

0

100

200

300

400

Time (min)

0

M-i-PVP-2

0

100

200

300

400

Time (min)

Fig. 5. Evolution of oil conversion over time using monolithic catalysts indicated (see text for the nomenclature). Dashed lines correspond to replicate runs. Reaction conditions: 60 °C, 2% catalyst load and 48:1 methanol-to-oil molar ratio.

sis reactor and being conscientiously washed with tetrahydrofuran (THF). The recovered samples maintained a white coloured homogeneous covering layer and looked uniform in their surface. The monolith samples were tested following two different procedures: some samples were readily tested in the methanolysis reactor after drying them overnight in an oven at 100 °C, whereas other samples were again calcined at the same operating conditions previously described before using them in the second reaction cycle. The results of the monoliths tested in a second reaction cycle are pre-

sented in Fig. 7. A clear loss of catalytic activity after the first reaction cycle can be observed in all cases, particularly acute in the samples that were not subjected to recalcination. Maximum oil conversions of just 70% could be achieved after 6 h of reaction and only with the recalcined samples corresponding to the M-i-PVP-1 and M-i-PVP-2 monoliths. Nevertheless, the results obtained for the M-i-PVP-cCeO2 monoliths clearly showed that part of the catalytic activity could be recovered after subjecting the recovered samples to recalcination. The oil conversion could

675

TCD signal (a.u.)

I. Reyero et al. / Fuel 182 (2016) 668–676

S-i-PVPcCeO

2

S-i-PVP-2 S-w-PvOHcCeO

2

0

200

400

600

800

Temperature (°C) Fig. 6. CO2-TPD measurements of the solids from the selected suspensions (a.u.: arbitrary units).

1

M-i-PVP-1

0.8

Oil conversion

M-i-PVP-2 M-i-PVPcCeO2

0.6

0.4

0.2

0

0

100

200

300

400

Time (min) Fig. 7. Oil conversion over time of selected monolithic catalysts in a second reaction cycle. Open symbols: monolith samples readily tested after drying. Filled symbols: samples recalcined prior to the second reaction cycle. Same reaction conditions that for the first reaction cycle.

be doubled after 6 h of reaction when the M-i-PVP-cCeO2 monolith samples were thermally regenerated. Again, this goes to show the importance of adequately reactivating the catalysts before reusing them in the methanolysis reaction, in agreement with the findings from our previous work [12] and by other recent studies [15]. To the best of our knowledge, the results presented in this work are remarkably superior to the best ones previously reported in the

literature concerning the reusability of structured methanolysis catalysts at atmospheric pressure and very similar those reported by Xu et al. after three reaction cycles at pressures slightly above atmospheric (1.8 bar-a) [17]. Kwon et al. [15] have recently reported oil conversions of 94% in a second reaction cycle, but it must be noted the rigorous operating conditions used (250 °C and 63 bar). For the sake of comparison, Table 3 compiles the different results obtained in previous works. Despite the relatively satisfactory results presented in this work, the noticeable loss of activity in the second reaction cycle could be attributed to a loss of active phase, either during the first methanolysis reaction cycle or in one of the stages of the catalyst recovery and reactivation. Nevertheless, the concurrence of deactivation phenomena cannot be ruled out. It should be noted that strong adsorption of acylglycerols, particularly triglycerides, has been previously claimed as a possible deactivation process of basic solids such as Mg-Al hydrotalcites used as methanolysis catalysts [6]. In order to clearly sort this out, analyses of the biodieseland glycerine-rich phases obtained in the different reactions carried out using the M-i-PVPcCeO2 monoliths were carried out. About half of the Ca initially present in the fresh monolith was found to be solubilised in the reaction medium during the first methanolysis reaction. This explains the relatively poor results obtained during the second reaction cycle. After the second reaction cycle, the analyses of the spent samples revealed that the amount of Ca lost was much less. Very similar results could be found in the case of the catalysts tested in powder form, hence signalling that the loss of active phase observed cannot be attributed to the structuration of the catalysts. The results from the analyses showed that the Ca concentration was greater in the glycerine-rich phase than in the biodiesel-rich phase, as expected, given the greater polarity of glycerol. On the basis of previous studies with powder catalysts, the solubility of calcium greatly increases when calcium glyceroxide is formed during the course of the methanolysis reaction [7]. This compound is formed by the reaction of CaO with the glycerol resulting as co-product of biodiesel in the transesterification reaction. Therefore, the high amount of calcium leached out during the first reaction cycle can be attributed to the formation of very soluble calcium glyceroxide.

4. Conclusions This work reports, for the first time in the literature so far, the preparation of different catalyst formulations consisting of Ca-Ce mixed oxides deposited on structured catalysts using FecralloyÒ monoliths for their use in the methanolysis of sunflower oil. The

Table 3 Comparison of the catalytic performance of structured catalysts for transesterification reactions reported in the literature. Reference

Catalyst

Methanolysis conditions

Oil conversion (1st cycle)

Oil conversion (2nd cycle)

This work

70% after 6 h

77% after 10 h

15% after 8 h

59% after 6 h

54%

54% after 2 h

Not reported

CaO deposited on cordierite monoliths

99% after 6.5 h at 250 °C and 63 bar 73% after 5 h

94% after 7 h at 250 °C

[16] [17]

KF/Ca-Mg-Al hydrotalcites on monoliths packed in a membrane reactor SrO on cordierite monoliths

Sunflower oil (60 °C, 1 atm, 2 wt% of catalyst, methanol-to-oil ratio of 48) Sunflower oil (60 °C, 1 atm, 2 wt% of catalyst, methanol-to-oil ratio of 48) Soybean oil (120 °C, 0.5 wt% of catalyst, methanol-to-oil ratio of 32) Rapeseed oil (195 °C, 20 bar, methanol-to-oil ratio of 12) Canola oil (250–310 °C, 63–107 bar, methanolto-oil ratio of 38–50) Oil not specified (65 °C, 1 atm, 4–6 wt% of catalyst, methanol-to-oil ratio of 20) Soybean oil (50–70 °C, 1.8 bar, 0.5–1.5 g of catalyst, methanol-to-oil ratio of 24) Rapeseed oil (150–195 °C, 15–20 bar-g, methanol-to-oil ratio of 7)

99% after 4 h

[15]

Ca-Ce mixed oxides on FecralloyÒ monoliths Mg-Al hydrotalcites on FecralloyÒ monoliths K/Al2O3 on honeycomb cordierite monoliths Zn aminoacid complex on cordierite monolith Zn-Na-Mg oxides on cordierite monoliths

91.7% (biodiesel content) after 3 h at 67 °C 65.4% after 100 h at 195 °C and 20 bar

79.8% (biodiesel content) after 3 reaction cycles Not reported

[12] [13] [14]

[18]

Not reported

676

I. Reyero et al. / Fuel 182 (2016) 668–676

different Ca/Ce formulations prepared under different techniques were found to be very active in the first reaction cycle but having serious stability issues in almost every case. A catalyst formulation of CaO-CeO2 mixed oxides prepared by the metallic citrates method was found to yield the best results in terms of activity and stability. The study of different formulations for preparing the suspensions to be used in the washcoating permitted to conclude that the solvent medium used and the use of additives significantly exert an effect on the catalytic performance and also on the adherence of the catalytic layer onto the monolithic substrate. It was found that isopropanol is an adequate solvent for preparing suspensions that subsequently yield highly active catalysts, whereas the use as additives of polyvinylpyrrolidone and colloidal ceria were necessary for achieving acceptable levels of adherence of the resulting catalytic layers, but at the expense of sacrificing part of the catalytic activity of the structured catalysts. To the best of our knowledge, the results presented in this work are remarkably superior to the best ones previously reported in the literature concerning the use in a second reaction cycle of structured methanolysis catalysts at atmospheric pressure. In spite of these relatively good results, it should be recognized that CaO presents huge difficulties to be used as heterogeneous catalyst for biodiesel production. Leaching of Ca species and formation of Ca-containing compounds that are soluble into the reaction medium give rise to a significant contribution of homogeneous catalysis to the overall process. Nevertheless, several approaches exist to tackle these issues than can be considered in future works. Of course there are still opportunities for improving the formulation of the suspensions based on CaO to be used in the preparation of structured metallic catalysts. However, ceramic structured substrates made of e.g. cordierite can be used instead of the metallic ones with the aim of improving adherence. New issues can arise in that case, because an excessive affinity can lead to solid state reactions during certain preparation stages, such as calcination, that could eventually deactivate the catalyst. Therefore, a delicate equilibrium has to be achieved between the properties of the active phase and those of the support and the substrate in order to be able to prepare a suitable heterogeneous catalyst for biodiesel synthesis. Another option is using an active phase different from CaO which is very active but unstable. In this regard, other Ca compounds such as Ca-Mn and Ca-Zn mixed oxides, calcium silicates and Ca12Al14O32 have been proposed as methanolysis catalysts [7]. There is also the possibility of using active phases with acidic character instead of basic properties. Although it is well known that acid catalysts are less active than the basic ones for methanol-

ysis reactions, such approach may be explored if this drawback can be compensated by a better stability. In conclusion, there are many options for continuing the work in the field of heterogeneous structured reactors for the synthesis of biodiesel. Acknowledgements The Spanish Ministerio de Economía y Competitividad (MINECO), the former Ministerio de Ciencia e Innovación and FEDER funding (ENE2012-37431-C03-03 and ENE2015-66975-C3) are thanked for the financial support and for the FPI predoctoral aid awarded to A. Moral (BES-2013-062799). References [1] Lee AF, Wilson K. Catal. Today 2015;242:3–18. [2] Galadima A, Muraza O. Energy 2014;78:72–83. [3] Arzamendi G, Campo I, Arguiñarena E, Sánchez M, Montes M, Gandía LM. Chem. Eng. J. 2007;134:123–30. [4] Arzamendi G, Campo I, Arguiñarena E, Sánchez M, Montes M, Gandía LM. J. Chem. Technol. Biotechnol. 2008;83:862–70. [5] Arzamendi G, Arguiñarena E, Campo I, Zabala S, Gandía LM. Catal. Today 2008;133–135:305–13. [6] Navajas A, Campo I, Arzamendi G, Hernández WY, Bobadilla LF, Centeno MA, et al. Appl. Catal. B: Environ. 2010;100:299–309. [7] Reyero I, Arzamendi G, Gandía LM. Chem. Eng. Res. Des. 2014;92:1519–30. [8] Reyero I, Bimbela F, Navajas A, Arzamendi G, Gandía LM. Fuel 2015;158:558–64. [9] Mittelbach M. Eur. J. Lipid Sci. Technol. 2015;117:1832–46. [10] Qiu Z, Zhao L, Weatherley L. Chem. Eng. Process. 2010;49:323–30. [11] Maddikeri GL, Pandit AB, Gogate PR. Ind. Eng. Chem. Res. 2012;51:14610–28. [12] Reyero I, Velasco I, Sanz O, Montes M, Arzamendi G, Gandía LM. Catal. Today 2013;216:211–9. [13] Tonetto GM, Marchetti JM. Top. Catal. 2010;53:755–62. [14] Kolaczkowski ST, Asli UA, Davidson MG. Catal. Today 2009;147S:S220–4. [15] Kwon K, Vahdat N, Mbah J. Fuel 2015;145:116–26. [16] Azman SR, Ismail M, Kadhum AAH, Yaakob Z. Int. J. Automot. Mech. Eng. 2014;10:1959–70. [17] Xu W, Gao L, Xiao G. Fuel 2015;159:484–90. [18] Firth B PhD Thesis. University of Bath; 2014. [19] Hosseini S, Janaun J, Choong TSY. Process Saf. Environ. Prot. 2015;98:285–95. [20] Lukic´ I, Kesic´ Zˇ, Zdujic´ M, Skala D. Fuel 2016;165:159–65. [21] Kouzu M, Hidaka JS. Fuel 2012;93:1–12. [22] Navajas A, Issariyakul T, Arzamendi G, Gandía LM, Dalai AK. Asia-Pacific J. Chem. Eng. 2013;8:742–8. [23] Granados ML, Alonso DM, Sádaba I, Mariscal R, Ocón P. Appl. Catal. B: Environ. 2009;89:265–72. [24] Yu X, Wen Z, Li H, Tu ST, Yan J. Fuel 2011;90:1868–74. [25] Wong YC, Tan YP, Taufiq-Yap YH, Ramli I, Tee HS. Fuel 2015;162:288–93. [26] Thitsartarn W, Maneerung T, Kawi S. Energy 2015;89:946–56. [27] Samantaray S, Pradhan DK, Hota G, Mishra BG. Chem. Eng. J. 2012;193– 194:1–9.