- Email: [email protected]
Biofuels production by esterification of oleic acid with ethanol using a membrane assisted reactor in vapour permeation configuration
Applied Catalysis A, General 566 (2018) 121–129
Contents lists available at ScienceDirect
Applied Catalysis A, General journal homepage: www.elsevier.com/locate/apcata
Biofuels production by esterification of oleic acid with ethanol using a membrane assisted reactor in vapour permeation configuration
T
⁎
Catia Cannilla , Giuseppe Bonura, Fabio Costa, Francesco Frusteri CNR-ITAE, Istituto di Tecnologie Avanzate per l’Energia “Nicola Giordano”, Via S. Lucia 5, 98126 Messina, Italy
A R T I C LE I N FO
A B S T R A C T
Keywords: Oleic acid Esterification reaction Biodiesel Methanol Ethanol Vapor permeation Membrane
Biodiesel production by performing the oleic acid esterification reaction in a new reactor system, realized by coupling the batch reactor with a tubular permeoselective membrane, has been investigated. Such reactor configuration was adopted to remove the water produced as by-product during the reaction thus increasing the oleic acid conversion and the alkyl ester production. The activity of acid catalysts and zeolites was evaluated in presence of methanol and ethanol. The membrane resulted much more selective towards water separation when ethanol was used, allowing to force thermodynamic equilibrium at higher conversion values. Very interesting results in terms of oleic acid conversion were achieved with both methanol and ethanol (96% and 98% respectively) in presence of a commercial Amberlyst-15 catalyst, at 80 °C, at a low membrane surface to gas volume ratio (0.20 cm−1) and low acid/alcohol molar ratio (2/1 and 1/1 respectively). Over zeolite samples, conversion values higher than 60% were obtained at 100 °C after 5 h in the esterification reaction performed in vapor permeation configuration.
1. Introduction Biodiesel is a fatty acids alkyl esters (FAAEs) mixture, derived from renewable lipid feedstocks (vegetable oils, animal fats or waste cooking oils), which could be considered a suitable substitute for petroleum derivate fuels due to its environmental benefits [1]. Cetane number, energy content, viscosity and phase changes of biodiesel are similar to those of fossil diesel [2–4]. FAAEs are generally produced by transesterification reaction between the triglycerides contained into oils or fats and an alcohol; anyhow, the moisture and the free fatty acids (FFAs) present into the feedstock could react with the homogenous alkali catalyst used in the industrial process, producing soaps and gels. This reduces the FAAEs yield and deeply influences the performance and the economics of biodiesel production [5]. So, to use low-cost feedstocks (waste frying oils or fats and oils with high FFAs amount), a pretreatment esterification step is mandatory to reduce the free acidity level below 1% wt/wt. Therefore, the esterification reaction can be exploited both as a pre-treatment step of a conventional transesterification process [6,7] or as biodiesel direct production [8,9]. Methanol (MeOH), ethanol (EtOH) and butanol (ButOH) are the most employed alcohols [10]. Anyhow, the biological nature of EtOH and its extra-carbon atom,
which slightly increases the heat content and the esters cetane number, make ethyl esters (EEs) more interesting than methyl esters (MEs). Moreover, EtOH is nontoxic and has better solvent properties than MeOH for oil solubility and EEs have cloud and pour points lower than the MEs [10]. The esterification of oleic acid (OlAc) can represent well the biodiesel production, since it is present in most of oil crops [11]. With EtOH, ethyl oleate (OleOEt) is produced and water is the by product, according to the following equation: C8H17CH=CH(CH2)7COOH + CH3CH2OH ⇆ C8H17CH=CH (CH2)7COOC2H5 + H2O
(1)
The industrial esterification processes carried out with strong Brønsted acid catalysts [12–14] are not environmentally benign, since they require special treatments as neutralization involving costly catalyst separation steps from the homogeneous reaction mixtures. Heterogeneous catalysts could be easier separated, they show high activity and stability and could be reused offering milder operating conditions. Ion exchange resins [15,16], zeolites [17–19], heteropolyacids [20] and modified zirconia [21,22] are the most used solid acid catalysts for OlAc esterification. Besides, niobic acid [23], polymers with acid
Abbreviations: Ai, initial acidity; Af, final acidity; FAAEs, fatty acids alkyl esters; FFAs, free fatty acids; EEs, ethyl esters; MEs, methyl esters; MVol, micropore volume; OlAc, oleic acid; OleOEt, ethyl oleate; PV, pervaporation; PVol, pore volume; SA, surface area; Sm, surface of the membrane; TPD, temperature programmed desorption; VP, vapor permeation; Vg, volume of gas phase ⁎ Corresponding author. E-mail address: [email protected] (C. Cannilla). https://doi.org/10.1016/j.apcata.2018.08.014 Received 2 May 2018; Received in revised form 2 August 2018; Accepted 16 August 2018 Available online 18 August 2018 0926-860X/ © 2018 Published by Elsevier B.V.
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
2. Materials and methods
groups [24]; ammonium salt of TPA [25], carbon material [26] were used too. Anyhow, considering the reaction equilibrium, the water produced and that contained into waste oils could hinder the esterification, accelerating the inverse reaction. Amberlyst-15 (A-15), for example, resulted to be a suitable catalyst for OlAc esterification with MeOH [27], but the conversion rate increased only replacing MeOH and catalyst with fresh one or by removing water from the reaction mixture. For this reason, it is of great interest, from a practical point of view, to study separation processes able to eliminate the water during the reaction [28]. On this account, Son et al. [29] obtained high MEs yield by performing the OlAc esterification in a three-phase fixed bed reactor (80 °C–120 °C) coupled with a continuous water evaporation system. Nijhuis et al. [30] exploited a reactive stripping operation in a monolithic reactor; Lucena et al. [5] employed a reactor coupled to an adsorption column filled with zeolite 3A in the OlAc esterification with MeOH using H2SO4 as catalyst and shifted the biodiesel yield from 88.2% to 99.7% at 100 °C. Recently, the integration of chemical reaction with the membrane separation has received great attention: the use of a reactor equipped with a permeoselective membrane significantly improves the process efficiency favoring not only the chemical equilibrium shift but also a longer catalyst lifetime [31–33]. Specifically, pervaporation (PV) and vapor permeation (VP) technologies could be considered as clean technologies which, because of low temperatures and pressures involved, could efficiently replace conventional energy-intensive separation processes [34–37]. They are based on the use of a semipermeable active or passive membrane which allows the separation of one or more species of a mixture by their passage from one site to the other one on the basis of a sorption-diffusion mechanism. Considering that in PV, the feed is a liquid, a liquid-vapour phase change is required involving an enthalpy of vaporization, as a function of the concentration gradient between the phases on the opposite site of the interfacial barrier. Instead, in VP the feed is a vapour mixture. In such a case, no phase change or no significant temperature differences across the membrane are required, the driving force being the partial pressure of the component in the feed. A noteworthy advantage of VP configuration is that the membrane is not in contact with the liquid reaction mixture rather, thus a longer membrane lifetime is assured [38]. So far, PV process has been exploited in some liquid-phase equilibrium reactions for biofuels production, while very low papers deal with the use of VP technique. Kita et al. [39] published one of the first papers regarding gas separation in the carboxylic acids esterification with alcohols, demonstrating that, by selecting suitable permeation conditions, total conversion could be achieved. The water vapor was removed using a polyimide hollow fiber membrane module placed on the top of a stirred batch vessel; p-toluenesulfonic acid was the catalyst [39,40]. Yamamoto et al. [41] and Okamoto et al. [42] exploited the very high permeation flux, water affinity and selectivity of NaA zeolite membrane in VP configuration to improve the esterification reaction performance. In this paper, the efficiency of a coupled reactor-membrane set-up was demonstrated by performing the heterogeneous esterification of OlAc in presence of a commercial permselective membrane in VP configuration. In this case, two separate units, a batch reactor and an external separation module, were employed. This set-up allows to make clean-up or regeneration steps easier and to optimize independently both the catalytic behavior and the separation step. Solid acid catalysts, as A-15 resin and some zeolites, have been used in the esterification either with MeOH or EtOH. The obtained results have been compared with typical runs performed without the use of membrane.
2.1. Reagents OlAc purchased from Sigma Aldrich presented 99% oleic acid and 1% mixture of palmitoleic acid, palmitic acid and myristic acid, with an acid value of 139.9 mg of NaOH/g. Methanol (> 99%) or anhydrous ethanol (> 99%) were supplied from Fluka. 2.2. Catalysts Amberlyst-15 wet ion exchange was purchased from the Dow Chemical Company, whereas the zeolite catalysts were supplied by Alpha Aesar. A-15 resin was used without any pretreatment. Zeolites have been calcined for 6 h at 500 °C. Two solid acid supported catalysts have been prepared by dry impregnation of two microspherical silica (ES70Y and MS3030 by PQ Corporation) in an ethanolic solution with Hyflon® Ion S4X perfluorosulphic ionomer (20 wt%) with an equivalent weight of 730 (Solvay Solexis) [43]. 2.3. Catalyst testing OlAc esterification reaction was carried out in two stainless steel reactors of different volumes (90 and 300 cc), with a constant stirring rate of 1100 min−1. Experiments were performed under autogenous pressure by maintaining the temperature in the range 65–100 °C. The amount of catalyst, in respect to the OlAc (Rcat/OlAc) was varied from 1 to 15 wt.% and the alcohol/OlAc molar ratio (RAlcohol/OlAc) was in the range 1–1 to 20–1. The reactants and the catalysts were loaded into the reactor in the fixed ratio, nitrogen flux was employed for reactor cleaning and then the reactor was heated to the desired temperature. At the end of the reaction, the reactor was cooled down with an ice bath to condense gaseous products. The catalyst was recovered by filtration. The batch reactor and the water separation module were two distinct units and the schematic configuration of the reaction system was shown in Fig. 1. The gas phase (mainly containing the alcohol and the water) was recirculated through the water permselective membrane (HybSi), provided by Pervatech BV [35,36]. The membrane (250*10*7 mm) had an effective area of 0.005 m2, the substrate material was a γ-alumina phase whereas the top layer was a hybrid silica coated on inside of the support tube. It could operate at a maximum temperature of 150 °C and a pressure of 10 bar, in the pH range of 2–8.5, showing a permeability to pure H2O of 19.66 l/m2 bar at 150 °C. To avoid the gas phase condensation, during the reaction, the recirculation loop was maintained at 115 °C. The recirculation of gas phase was maintained constant (2.8 L/min) by using a gas-pump equipped with a variable potentiometer. At the permeate side of the membrane, the permeate was purged with a stream of carrier gas (N2, 2 L/min) and the permeate was finally obtained in the liquid state after condensation to be subsequently analyzed. 2.4. Biodiesel analysis Conversion of OlAc was calculated based on the AOCS (American Oil Chemist’s Society) official methods for acid value (Cd 3d-63) and FFA amount (Ca 5a-40). Samples were analyzed by a standard acid-base titration procedure for the evaluation of the free residual acidity. The analysis repeatability has been improved by removing methanol or ethanol in excess and water formed in a vacuum condenser, prior to submit the samples to titration. A weighed amount of the sample was dissolved in an ethanol/ethylic ether mixture 1/1 wt/wt and then was titrated with a NaOH 0.5 M solution and phenolphthalein as indicator. The volume of alkaline solution consumed was recorded and the acidity of the sample was determined by means of the Eq. (2), referred to the oil phase (oleic acid + ester) with an error less than 1–2% on the free acidity 122
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
Fig. 1. Schematic configuration of batch reactor coupled with the permeoselective membrane. The insert shows the ceramic tubular membrane [A] and the PVM-035 membrane housing [B].
3. Results and discussion
Acidity = Vtitr(mL) * Ctitr (mol/L) * MWOlAc/(1000 * gsample) * 100 (wt. %) (2)
3.1. Oleic acid esterification with MeOH
The acidity was expressed as OlAc weight percent. The OlAc conversion was calculated as the fraction of the removed acid:
The homogenous esterification reaction of oleic acid is generally carried out using H2SO4 as catalyst: a small amount of catalyst (0.01% wt/wt) is enough to promote the reaction [5]. Anyhow, heterogeneous catalysis could be more advantageous from many points of view avoiding reactor corrosion, aqueous waste production and the need of costly neutralization and separation steps. The operating conditions that mainly affect the conversion of OlAc into biodiesel are the temperature, the amount of catalyst and the alcohol/OlAc molar ratio. Specifically, the stoichiometry amount of molar ratio alcohol/OlAc is 1; however, in literature, a very high alcohol to fatty acid molar ratio has also been reported to improve the esterification reaction rate [14,44,45] by mitigating the inhibiting effect of the water formed which tends to reverse the reaction. Some preliminary experiments of esterification with MeOH have been carried out at 80 °C by using solid acid catalysts with a Rcat/OlAc equal to 5 wt.%, lower than that usually used in literature [46]. The results obtained are reported in Table 1 along with main chemicalphysical properties of catalysts used. Specifically, the performance of the A-15 resin was evaluated along with HZSM-5 zeolite with a Si/Al ratio of 23 and two Hyflon-based catalysts on spherical silica (20H730/ MS3030 and 20H730/ES70Y). High stirring rate of the mixture (1100 rpm) was employed to avoid external mass or heat transport limitations. After two hours, by employing a RMeOH/OlAc of 2 mol/mol, the activity of A-15 was the highest. This result is mainly correlable with a higher acidity of A-15 (4.50 meqH+ gcat−1) in respect of the other systems (< 1.00 meqH+ gcat−1) and confirms the good properties of the resin in terms of esterification efficiency already reported in literature [46]. By considering such preliminary results, A-15 was employed for the study finalized to optimize the reaction conditions. First of all, the influence of the alcohol amount and the reaction
OlAc conversion = (Ai − Af)/Ai * 100 where Ai and Af indicate the initial and final acidity of the solution respectively.
2.5. Catalyst characterization The textural properties of catalysts were determined by physical adsorption measurements of nitrogen to its boiling point (−196.15 °C) using a Micromeritics ASAP2020 gas adsorption device. The isotherms were elaborated for assessment of surface area (SA) and porosity (PVol), with the micropore volume (MVol) determined by the t-plot approach. Surface concentration of acidic sites was determined by temperature programmed desorption of ammonia (NH3-TPD) for zeolite. Before TPD experiment, the catalyst sample (∼100 mg) was saturated for 2 h at 150 °C under a gas mixture containing 5 vol.% of NH3/ He (flow rate = 25 stp mL/min). Then, the sample was purged in helium flow until a constant baseline level was attained. The measurement was performed in the temperature range 100–700 °C, at a rate of 10 °C/min using helium (25 stp mL/min) as the carrier flow. The evolved gases were detected by an on-line thermal-conductivity detector, calibrated by the peak area of known pulses of NH3. Potentiometric titrations were performed to evaluate the acid capacity (meq gcat−1) of the Hyflon-based systems according to the procedure elsewhere described [43].
123
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
water formation. In addition, the presence of water in the reaction medium could negatively affect the reaction rate upon adsorption on acid sites of solid catalysts, so blocking the access of the reactants [17,46]. Thus, in the attempt of overcoming such equilibrium restrictions a water permeo-selective membrane was used, coupled with the batch reactor, by exploiting the VP technology. As previously reported [35,36], the membrane was not in direct contact with the reaction liquid mixture, gas phase being forced to circulate through the membrane by a pump. The “driving force” to favor the water permeation was ensured by the pressure difference between the reactor (autogenous pressure) and the environment (permeate side). The preliminary results obtained using the permselective membrane with the methanol in presence of A-15 are reported in Table 2. The removal of water positively affected the reaction favouring the acid conversion, which increased from 67.0% (run 1, without the membrane) up to 80.2% (run 2, with the membrane) after 2 h. Nevertheless, by using a higher RMeOH/ OlAc, the acid conversion followed a sort of volcano shape trend with a maximum value for a molar ratio of 4 (83.7%). To achieve high conversion values in a short reaction time, a fundamental parameter to take into consideration is the water permeation rate. In this regard, from results reported in Table 2, it can be observed that the amount of water collected at the same reaction time (2 h) and the theoretical water permeation rate (evaluated on the OlAc conversion values) changed with the MeOH/OlAc molar ratio, decreasing with the methanol amount. Specifically, when the methanol amount was high (run 4), no permeated water was trapped. This is likely because of lower water partial pressure in the gas phase recirculated trough the membrane when higher MeOH concentration was used. In such a case, the water permeation capability of the membrane resulted to be limited. Anyhow, at low MeOH/OlAc molar ratio, the condensation system resulted to be more efficient, in fact 27.6% of theoretical water was recovered after 2 h (run 2) which increased up to 56.3% after 5 h (run 5). Naturally, when the permeation of water was more efficient, the acid conversion achievable was higher. In fact, in run 5, the acid conversion approached 97.3%, so confirming the benefit obtained in terms of equilibrium shift due to water removal through the membrane. Anyhow, by analyzing the water solution recovered during the reaction, it has been observed that a little amount of MeOH passed through the membrane along with the water (5 wt.%). This was much more evident when the reactor with a lower volume (90 cc) was used. In fact, in this case, the OlAc conversion didn’t increase, rather after 5 h, the OlAc conversion was only 74.0% (run 6) and no water was recovered. This result likely suggests that the membrane performance is not high when the water concentration in the gas phase is too low. Another factor to consider in the configuration of membrane reactor is the recirculation rate of gas phase containing the water vapor, which also plays an important role. Specifically, as shown in Fig. 3, regardless the MeOH/OlAc molar ratio used (2 or 4), higher amounts of water permeated through the membrane and therefore higher oleic acid conversion values were achieved when higher recirculation rate were used. Specifically, limiting the gas phase recirculation, or rather by filling almost completely the reactor (Sm/Vg = 0.71 cm−1, where Sm is the surface of the membrane and Vg is the volume of gas phase evaluated at the beginning of the reaction), the maximum oleic acid conversion reached after 2 h of reaction was about 64.2% and no water was collected. On the contrary, almost total conversion (96.1%) was obtained along with 53.2% of theoretical water when gas phase volume was higher (Sm/Vg ratio of 0.20 cm−1). It is evident that, with high recirculation flow, the turbulence regime necessary to improve water permeability is ensured and high conversion can be achieved.
Table 1 Main physico-chemical properties of catalysts employed in OlAc etherification (TR = 80 °C; RMeOH/OlAc = 2 mol/mol; Rcat/OlAc = 5 wt.%; time = 2 h). Catalyst
SA (m2 g−1)
PVol (cm3 g−1)
Acidity (meqH+ gcat−1)
OlAc conversion (%)
A-15a 20H730/ ES70Y 20H730/ MS3030 HZSM5 (23:1)
45 177
0.40 0.33
4.50 0.88
67.0 51.4
204
0.33
0.77
32.6
498
0.20
0.30
18.3
a Data supplied by Rohm and Haas. SABET= Surface Area. PVol = Pore Volume, determined from N2 ads/des isotherms at −196.15 °C.
Fig. 2. Esterification reaction of oleic acid with MeOH and A-15: [A] Influence of MeOH/OlAc molar ratio (TR = 80 °C; Rcat/OlAc = 5 wt.%; time = 2 h). [B] Influence of reaction time (TR = 80 °C; RMeOH/OlAc = 4 mol/mol; Rcat/OlAc = 5 wt.%).
time were evaluated at 80 °C. Fig. 2A clearly shows that the reaction rate is slightly affected by the MeOH/OlAc molar ratio considering that by increasing it from 2 up to 4, OlAc conversion increases from 67.0% to 72.1% then reaching a plateau (73.1% at RMeOH/OlAc = 8 mol/mol). So, the use of a MeOH/OlAc molar ratio higher than 4 does not produce noteworthy effects. On the other hand, Fig. 2B shows that the reaction does not take place with high reaction rate and equilibrium conditions are reached only after 6–8 h. According to the literature evidences [5,28,46], the total OlAc conversion and the production of high esters yield are limited by the reversibility of the reaction and the equilibrium constraints due to the
3.2. Oleic acid esterification with ethanol In Fig. 4, a comparison of the activity of A-15 catalyst in the oleic acid esterification with MeOH and EtOH is shown in terms of effect of 124
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
Table 2 OlAc esterification reaction in presence of MeOH with and without the membrane (TR = 80 °C; Rcat/OlAc = 5 wt.%). Run
RMeOH/OlAc (mol/mol)
Time (h)
OlAc conversion (%)
Sm/Vg (cm−1)a
H2Oteor. (g)b
H2Oteor. (g)c
H2OColl (%)d
H2O permeation rate (h−1)e
1f 2 3 4 5 6g
2 2 4 8 2 2
2 2 2 2 5 5
67.0 80.2 83.7 79.8 97.3 74.0
0.27 0.27 0.27 0.32 0.72
5.10 4.33 3.25 6.37 0.96
4.09 3.63 2.60 6.04 0.71
27.6 20.1 – 56.3 –
0.066 0.032 0.014 0.038
a b c d e f g
Sm = surface of the membrane; Vg = volume of gas phase evaluated at the beginning of the reaction. Water amount evaluated by theoretical total OlAc conversion. Theoretical water amount evaluated by measured OlAc conversion. Percentage of water collected as a function of theoretical water (b). Value obtained considering the theoretical water amount produced by measured OlAc conversion. Run carried out without membrane. Run carried out by using a 90 cc reactor.
Fig. 3. Influence of Sm/Vg ratio on the esterification reaction of oleic acid with MeOH in presence of membrane (catalyst: A-15; TR = 80 °C; RMeOH/ OlAc = 2 mol/mol; Rcat/OlAc = 5 wt.%; time = 2 h). Triangle symbols are referred to the run with RMeOH/OlAc = 4 mol/mol.
reaction temperature on OlAc conversion and activation energy is shown. Esterification with methanol resulted to be more endothermic than with ethanol [45] and a pseudo first order dependence for fatty acid esterification emerged. From the conversion data obtained at different temperatures, the apparent activation energy was determined by Arrhenius plot. The activation energies found were 31.5 kJ/mol and 28.0 kJ/mol for methanol and ethanol respectively, values lower than that reported by other authors on solid catalysts [20,45] although in the range of the existing results [12,14,47]. In a prevailing kinetic regime, the esterification reaction with ethanol (1.16 * 10−5 molEEgcat−1 s−1) takes place with a lower rate than in presence of methanol (8.03 * 10−6 molMEgcat−1 s−1). This result is confirmed in literature for palmitic acid esterification, the order of effectiveness being methanol > ethanol, due to the different reactivity of the carbon chains of each alcohol [42]. In Table 3 the comparative results of the esterification reaction of OlAc with EtOH carried out in the two batch reactors of different volume (90 and 300 cc) with and without the membrane, are reported. First of all, it is possible to observe that, after 2 h at 80 °C, the OlAc conversion, with membrane by-passed, was 50.0% and 57.8% depending on the reactor capacity (runs 1–7), lower than that obtained with methanol (67.0%, see Table 2) as expected on the basis of the kinetic results. By using the membrane coupled with the reactor of smaller volume, the conversion increased up to 63.2% (run 2). Also in this case, with a higher alcohol/oleic acid molar ratio, 8 mol/mo (run 4), OlAc conversion slightly decreased to 52.2%, likely due to a lower concentration of water vapour in the gas phase. It is interesting to highlight the better efficiency obtained with the membrane when the
Fig. 4. Esterification reaction of oleic acid with MeOH and EtOH: [A] Influence of reaction temperature (Rcat/OlAc = 1 wt.%; time = 1 h; RAlc/OlAc = 5 mol/ mol). [B] Arrhenius plot.
ethanol is the reactant. In fact, after 5 h of reaction, by using the small reactor, the OlAc conversion was almost total (run 5, 98.1%) unlike the reaction with MeOH (74.0%, see Table 2, run 6). This evidence indicates that the membrane has a higher selectivity towards the water permeation when ethanol is employed instead of methanol. Anyhow, by using the reactor at a capacity as low as 90 cc, the water permeated during reaction was not recovered due to its low concentration; for this reason it was preferred to operate with a larger batch reactor (300 cc). In this case, with a REtOH/OlAc of 2 mol/mol, the oleic acid conversion increased from 57.8% (run 7) without membrane up to 78.6% after 2 h (run 9) with the VP system (under the same amount of reactants) but unfortunately only about 10% of theoretical 125
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
Table 3 Oleic acid esterification reaction in presence of EtOH, with and without the permeoselective membrane (TR = 80 °C; Rcat/OlAc = 5 wt.%). Time (h)
OlAcConversion (%)
Sm/Vg (cm−1)a
H2O (g)
H2Oteor. (g)b
H2Oteor. (g)c
H2OColl (%)d
Autoclave 90 cc 2 1e 2 2 3 4 4 8 5 2
2 2 2 2 5
50.0 63.2 60.0 52.2 98.1
0.74 0.72 0.71 0.74
– – – –
0.96 0.67 0.45 0.96
0.60 0.40 0.23 0.94
– – – –
Autoclave 300 cc 6e 1 7e 2 8 1 9 2 10 1 11 1 12 2
2 2 2 2 5 2 2
59.0 57.8 85.0 78.6 98.2 77.6 73.8
0.20 0.20 0.20 0.28 0.28
1.07 0.19 1.80 2.48 1.14
2.74 2.35 2.74 5.87 5.10
2.32 1.83 2.69 4.55 3.76
46.01 10.47 66.90 54.60 30.28
Run
a b c d e
REtOH/OlAc (mol/mol)
Sm = surface of the membrane; Vg = volume of gas phase evaluated at the beginning of the reaction. Water amount evaluated by theoretical total OlAc conversion. Theoretical water amount evaluated by measured OlAc conversion. Percentage of water collected as a function of theoretical water (b). Runs carried out without membrane.
surface area value (1027 m2 g−1), followed by BEA zeolite (∼800 m2/ g); others investigated zeolite samples showed similar values, with an average surface exposure of about 500 m2/g. The determined surface area values reflect the micropores volume resulting higher for Y and BEA zeolites (0.27 cm3 g−1 and 0.21 cm3 g−1 respectively). Surface acid properties are also reported in Table 4 in terms of quantitative data obtained by NH3-TPD measurements (desorption profiles are shown in Fig. S1 of Supplementary data). As expected, for all samples, total acidity was lower than that of Amberlyst-15. Two main peaks characteristic of the interaction of ammonia with weak and strong acid sites at low (150–250 °C) and high temperature (350–450 °C), were observed. As well known, by NH3-TPD profiles it is not possible to discriminate between Brønsted and Lewis acidity, anyhow it is generally accepted that the peak at lower temperature is associated both to the presence of Lewis sites on zeolite framework or to extra-framework aluminum oxide/hydroxide species. Instead, the peak at higher temperature is generally ascribable to the presence of Brønsted acid sites due to the framework Al ions [49]. On this account, the Y zeolite showed the highest total acidity (1211 μmol g−1), whereas in the HZSM-5 samples acidity decreased by increasing the SiO2/Al2O3 ratio. Preliminary experiments carried out with EtOH at 80 °C revealed that, at this temperature, all zeolites investigated were not so active as A-15 catalyst, therefore, using a molar REtOH/OlAc of 2/1, the reaction temperature was increased at 100 °C. Without the use of the membrane, the higher OlAc conversion achieved was around 23%, and the
water was collected. Instead, by employing a REtOH/OlAc of 1 mol/mol, OlAc conversion value increased of about 30% thanks to the membrane effect (run 6, 59.0%; run 8, 85.0%) and the percentage of the collected water was higher (46,01%). The gas chromatographic analysis of the liquid mixture collected showed a very low amount of EtOH (< 0.1 wt. %) permeated through the membrane and thus confirmed a higher selectivity towards ethanol/water separation (than methanol/water). By further increasing the reaction time, it was possible to achieve almost complete OlAc conversion (i.e. 98.2%) by maintaining mild reaction conditions (run 10). As already seen with MeOH, even in this case, the Sm/Vg ratio influences the performance of the membrane; in fact by increasing this value from 0.20 cm−1 up to 0.28 cm−1, the conversion decreased from 85.0 to 77.6% with a REtOH/OlAc of 1 mol/mol and from 78.6% to 73.8% with a REtOH/OlAc of 2 mol/mol. These very interesting results show a significant improvement in the esterification rate with EtOH under milder reaction conditions if compared with recent literature data [16,48].
3.3. Oleic acid esterification with EtOH using zeolites In the attempt to find more resistant solid acid catalysts for this reaction, a series of zeolites have been investigated. Such zeolites were characterized from a structural and morphological point of view and main results are reported in Table 4. As regards textural properties, Y zeolite exhibited the highest
Table 4 Textural properties and quantitative data of NH3-TPD measurements and acid sites distribution of the investigated zeolites samples. Sample
Y FER HZSM5 BEA HZSM5 HZSM5 a b c d e f
SiO2/Al2O3 (mol/mol)
12 20 23 38 50 80
Textural properties
Surface acidic properties
SALang (m2/g)
PVola (cm3/g)
MVolb (cm3/g)
Total sites (μmol g−1)
w-m sitesc
m-s sitesd
TM1e (°C)
TM2f (°C)
1027 449 498 781 539 574
0.52 0.21 0.20 0.34 0.25 0.29
0.27 0.14 0.13 0.21 0.11 0.11
1211 527 542 871 303 73
0.65 0.62 0.34 0.35 0.56 0.66
0.35 0.38 0.66 0.65 0.44 0.34
219 206 223 194 212 216
376 439 407 413 396 377
Cumulative Pore Volume, evaluated in the range 5–600 Å (micro and mesopores). Micropores Volume evaluated by t-Plot. Weak-medium acid site fraction with temperature of maximum desorption of NH3 in the range 100–300 °C. Medium-strong acid site fraction with temperature of maximum desorption of NH3 above 300 °C. Temperature of maximum desorption of NH3 in the range 100–300 °C. Temperature of maximum desorption of NH3 above 300 °C. 126
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
Fig. 5. [A] Experimental results for catalytic screening in OlAc esterification reaction as a function of SiO2/Al2O3 (TR = 100 °C; REtOH/OlAc = 2 mol/mol; Rcat/OlAc = 5 wt.%, time = 2 h). [B] Total acid sites and weak-medium acid site fraction vs. SiO2/Al2O3 ratio.
reactivity of different zeolites was the following: HZSM5 (23:1)
(80:1) > Y > FER > HZSM5
(50:1) > BEA > HZSM5 Fig. 6. OlAc esterification reaction with ethanol. [A] Influence of EtOH/OlAc molar ratio (catalyst: 5 wt.% referred to the OlAc amount; TR = 100 °C, time = 2 h). [B] Influence of catalyst amount (TR = 100 °C; REtOH/OlAc = 2 mol/ mol; time = 14 h; * indicates REtOH/OlAc = 20 mol/mol).
From the results reported in Fig. 5A it was not possible to establish a clear correlation between the catalytic activity and zeolite properties (see Table 4), however, it was observed that an optimum bulk Si/Al molar ratio maximizes the catalytic behavior. This evidence could be correlated to the hydrophobic character of zeolite surface. In fact, as reported in literature, to a lower Si/Al ratio, it is associated a higher capability to coordinate water molecules, with consequent covering of the solid surface which limits the adsorption of organic substrates [50]. Specifically, this was the trend observed for HZSM-5 zeolites, for which a higher Si/Al molar ratio that could be correlated to a higher hydrophobicity also reflected in a lower acid site availability along with a higher amount of weak/medium acid site fraction (Fig. 5B). Anyhow, the fitting reported in Fig. 5A was obtained without considering the Y zeolite which was characterized by a high OlAc conversion of 21.22%, but, at the same time, by a Si/Al ratio of 12. This catalytic result was likely ascribable to the synergic role played by both the acid surface properties and the textural properties. Specifically, the catalyst porosity generally represents a key factor for biodiesel production since it strictly influences the active sites accessibility of big molecules involved in the esterification reaction [3]. Considering the high activity of HZSM-5 with a Si/Al ratio of 80, it was further employed for new experiments in the attempt at optimizing the reaction conditions. Using such a zeolite as catalyst, it was considered important to exclude the occurrence of mass transfer phenomena during the reaction. In fact, as well known, in a liquid phase reaction, two types of mass transfer resistances should be considered: the external mass transfer resistance, which takes place across the solid-liquid interface and the internal mass transfer resistance, associated with the difference in particle size distribution of the catalyst [7]. Considering the high
stirring speed utilized in the experiments, the first one could be overlooked, whereas bigger attention should be paid on the internal mass transfer resistances. For this reason, a set of experiments were conducted using the HZSM5-80 catalyst particle size in the 16–70 μm range. The results summarized in Fig. S2 showed no effect of the particle size on the OlAc conversion suggesting that there was no intraparticle diffusion control. The effect of the ethanol/OlAc molar ratio was investigated in the range 1–20 mol/mol. The results reported in Fig. 6A show a volcanotype trend indicating that the suitable molar ratio was comprised between 2 and 4. This result partially fits with data obtained by Marchetti et al. [17] which observed that, under their reaction conditions (60 °C and 20 wt.% of catalyst), the 1:1 ethanol/oleic acid molar ratio gave the highest reaction rate. This evidence was justified by considering that the reaction was not at the equilibrium level but still in an initial stage. However, it should be also considered that, when an excess of ethanol was used, it could be preferentially adsorbed onto the acid sites thus decreasing their availability. Furthermore an excessive dilution of solution could cause a decrease in acid sites concentration with negative effect on reaction kinetics [51]. Such findings suggested to optimize the reaction conditions by employing a higher zeolite amount for a longer reaction time. Specifically, as shown in Fig. 6B, with a Rcat/OlAc of 15 wt.%, OlAc conversion values of 62.2% and 70.1% were achieved for a EtOH/ OleOH molar ratio of 2 and 20 respectively, after 14 h of reaction. At this point, the VP module was coupled to the batch reactor and different reaction conditions were studied in presence of the HZSM5-80 127
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
no membrane
Sm/Vg = 0.74 cm-1
membrane
Sm/Vg = 0.20 cm-1 Rcat/OlAc = 5 wt% REtOH/OlAc = 2/1 mol/mol
90 Rcat/OlAc = 15 wt% REtOH/OlAc = 20/1 mol/mol time = 14 h
80 70
Fig. 7. OlAc esterification reaction with EtOH. Comparison of data with and without membrane at 100 °C. A) Rcat/OlAc = 5 wt.%; REtOH/OlAc = 2/1 mol/mol; t = 2 h; Sm/ Vg = 0.74 cm−1. B) Rcat/OlAc = 15 wt.%; REtOH/OlAc = 20/1 mol/mol; t = 14 h; Sm/Vg = 0.74 cm−1. C) Rcat/OlAc = 5 wt.%; REtOH/OlAc = 2/1 mol/mol; t = 2 h; Sm/ Vg = 0.20 cm−1. D) Rcat/OlAc = 5 wt.%; REtOH/OlAc = 2/1 mol/mol; t = 5 h; Sm/ Vg = 0.20 cm−1.
time = 5 h
60 50 time = 2 h
40 30
Rcat/OlAc = 5 wt% REtOH/OlAc = 2/1 mol/mol time = 2 h
20 10 0
catalyst. The results reported in Fig. 7 show that by using a Sm/Vg of 0.74 cm−1, at 100 °C, with a REtOH/OleOH of 2 mol/mol (A condition), the oleic acid conversion value didn’t increase using the membrane. The membrane effect was not so evident by performing the etherification reaction under the optimized reaction conditions previously found without the use membrane (see Fig. 6), that is for a long reaction time (14 h), high ethanol/OlAc molar ratio (20/1 mol/mol) and 15 wt.% of catalyst (B condition). Anyhow, a very interesting result was obtained when the recirculation rate of gas phase was increased by employing the optimized value of Sm/Vg ratio equal to 0.20 cm−1. In this case, in fact, at 100 °C, the OlAc conversions were doubled as evidenced by looking at C and D conditions, so achieving a value of 35.4% after 2 h and 62.1% after 5 h of reaction, at low EtOH/OlAc molar ratio (2 mol/ mol) and with 5 wt.% of catalyst. In Fig. 8, three reproducibility tests with HZSM5-80, in presence of membrane, are reported. The runs were carried out at 100 °C, REtOH/ OlAc = 2 mol/mol and 5 wt.% of catalyst. As it is possible to observe, after three experiments, the OlAc conversion remained almost constant. After each experiment, the catalyst was washed by hot distilled water and filtered prior to drying at 110 °C and the calcination at 500 °C for 3 h. Therefore, the obtained results suggest that, before a new run, zeolite samples need a simple regeneration step to restore their initial surface features and so ensuring a complete reproducibility of catalytic activity upon reusing. In any case, irrespective of the catalytic system employed, the potential of the tandem reactor-membrane setup for biofuels production by equilibrium-limited esterification reaction was fully demonstrated. Specifically, the optimization of the catalytic process-membrane system allowed obtaining higher conversion rate of the substrate at relatively short reaction time, by employing low alcohol-to-FFA molar ratios, with consequent several economic and ecological advantages.
Fig. 8. Catalyst stability (TR = 100 °C; Rcat/OlAc = 5 wt.%; REtOH/OlAc = 2/ 1 mol/mol; time = 5 h; Sm/Vg = 0.20 cm−1).
using a batch reactor coupled with a membrane separation system in VP configuration to drive the esterification reaction of oleic acid have been clearly demonstrated. By optimizing the reaction setup, a total oleic acid conversion can be achieved by continuously subtracting the water
4. Conclusions The catalytic and technological benefits that can be obtained by 128
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
formed from the reaction medium. A higher selectivity of the membrane towards ethanol rather than methanol favours an enhanced esterification rate. An efficient water permeation through the membrane is ensured by a recirculation rate of the gas phase in turbulent flow regime. From the catalytic point of view, a higher catalyst acidity of A15 favours the conversion in relatively mild reaction conditions (TR = 80 °C, REtOH/OlAc = 1 mol/mol; Rcat/OlAc = 5 wt%) although a slight increase of the temperature is beneficial if zeolite samples are used.
[17] J.M. Marchetti, A.F. Errazu, Fuel 87 (2008) 3477–3480. [18] A.M. Doyle, T.M. Albayati, A.S. Abbas, Z.T.A. Lismaeel, Renew. Energy 97 (2016) 19–23. [19] K.H. Chung, B.G. Park, J. Ind. Eng. Chem. 15 (2009) 388–392. [20] K. Srilatha, Ch.R. Kumar, B.L.A.P. Devi, R.B.N. Prasad, P.S. Prasad, N. Lingaiah, Catal. Sci. Technol. 1 (2011) 662–668. [21] D.E. Lopez, J.G. Goodwin Jr, D.A. Bruce, S. Futura, Appl. Catal. A 339 (2008) 76–83. [22] Y.M. Park, J.Y. Lee, S.H. Chung, I.S. Park, S.Y. Lee, D.K. Kim, J.S. Lee, K.Y. Lee, Bioresour. Technol. 101 (2010) S59–S61. [23] L.L. Rade, C.O.T. Lemos, M.A.S. Barrozo, R.M. Ribas, R.S. Monteiro, C.E. Hori, Renew. Energy 115 (2018) 208–216. [24] C.S. Caetano, L. Guerriero, I.M. Fonseca, A.M. Ramos, J. Vital, J.E. Castanheiro, Appl. Catal. A 359 (2009) 41–46. [25] B.Y. Giri, K.N. Rao, B.L.A.P. Devi, N. Lingaiah, I. Suryanarayana, R.B.N. Prasad, P.S. Sai Prasad, Catal. Commun. 6 (2005) 788–792. [26] Z. Gao, S. Tang, X. Cui, S. Tian, M. Zhang, Fuel 140 (2015) 669–676. [27] J.Y. Park, D.K. Kim, J.S. Lee, Renew. Energy 35 (2010) 614–618. [28] K. Suwannakarn, E. Lotero, K. Ngaosuwan, J.G. Goodwin Jr., Ind. Eng. Chem. Res. 48 (2009) 2810–2818. [29] S.M. Son, H. Kimura, K. Kusakabe, Bioresour. Technol. 102 (2) (2011) 2130–2132. [30] T.A. Nijhuis, A.E.W. Beers, F. Kapteijn, J.A. Moulijn, Chem. Eng. Sci. 57 (9) (2002) 1627–1632. [31] E. Drioli, A.I. Stankiewicz, F. Macedonio, J. Membr. Sci. 380 (2011) 1–8. [32] F. Gallucci, A. Basile, F.I. Hai, Introduction—a review of Membrane Reactors, Membrane Reactors, John Wiley & Sons, Ltd., Chichester, UK, 2011, pp. 1–61. [33] S.H. Shuit, Y.T. Ong, K.T. Lee, B. Subhash, S.H. Tan, Biotechnol. Adv. 30 (2012) 1364–1380. [34] C. Cannilla, G. Bonura, F. Frusteri, Catalysts 7 (2017) 187. [35] C. Cannilla, G. Bonura, L. Frusteri, F. Frusteri, Environ. Sci. Technol. 48 (2014) 6019–6026. [36] C. Cannilla, G. Bonura, L. Frusteri, F. Frusteri, Chem. Eng. J. 282 (2015) 187–193. [37] F. Lipnizki, R.W. Field, P.K. Ten, J. Membr. Sci. 153 (1999) 183–210. [38] B. Bolto, M. Hoang, Z. Xie, Water Res. 46 (2012) 259–266. [39] H. Kita, K. Tanaka, K. Okamoto, M. Yamamoto, Chem. Lett. (1987) 2053–2056. [40] K.I. Okamoto, M. Yamamoto, S. Noda, T. Semoto, Y. Otoshi, K. Tanaka, H. Kita, Ind. Eng. Chem. Res. 33 (1994) 849–853. [41] M. Yamamoto, N. Munehisa, M. Kaibara, K. Horii, K. Tanaka, H. Kita, K. Okamoto, Membrane 20 (1995) 143–148. [42] K.I. Okamoto, H. Kita, Ind. Eng. Chem. Res. 40 (2001) 163–175. [43] F. Frusteri, L. Frusteri, C. Cannilla, G. Bonura, Bioresour. Technol. 118 (2012) 350–358. [44] A.A. Kiss, A.C. Dimian, G. Rothenberg, Adv. Synth. Catal. 348 (2006) 75–81. [45] K. Srilatha, N. Lingaiah, B.L.A. Prabhavathi Devi, R.B.N. Prasad, S. Venkateswar, P.S.S. Prasad, Appl. Catal. A 365 (2009) 28–33. [46] J.Y. Park, D.K. Kim, J.S. Lee, Bioresour. Technol. 101 (2010) S62–S65. [47] M.C.G. Albuquerque, C.L. Cavalcante Jr., A.E.B. Torres, D.C.S. Azevedo, E.J.S. Parente Jr., Energy Fuels 23 (2009) 1136–1138. [48] Y. Jiang, J. Lu, K. Sun, L. Ma, J. Ding, Energy Convers. Manag. 76 (2013) 980–985. [49] G. Bonura, C. Cannilla, L. Frusteri, A. Mezzapica, F. Frusteri, Catal. Today 281 (2017) 337–344. [50] P.M. Veiga, A.C.L. Gomes, C.O. Veloso, C.A. Henriques, Appl. Catal. A 548 (2017) 2–15. [51] K.C.S. Figueiredo, V.M.M. Salim, C.P. Borges, Braz. J. Chem. Eng. 27 (2010) 609–617.
Funding sources Part of this work was supported by the Bilateral Agreement of Scientific and Technological Cooperation CNR-MTA/HAS (Hungarian Academy of Science). Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.apcata.2018.08.014. References [1] C. Cannilla, G. Bonura, F. Arena, E. Rombi, F. Frusteri, Catal. Today 195 (1) (2012) 32–43. [2] P. Sivakumar, K. Anbarasu, S. Renganathan, Fuel 90 (2011) 147–151. [3] C. Cannilla, G. Bonura, E. Rombi, F. Arena, F. Frusteri, Appl. Catal. A 382 (2) (2010) 158–166. [4] C. Beatrice, G. Di Blasio, C. Guido, C. Cannilla, G. Bonura, F. Frusteri, Appl. Energy 132 (1) (2014) 236–247. [5] I.L. Lucena, G.F. Silva, A.N.F. Fabiano, Ind. Eng. Chem. Res. 47 (2008) 6885–6889. [6] B. Škrbic´, Z. Predojevic´, N. Đurišic´-Mladenovic, Waste Manag. Res. 33 (8) (2015) 723–729. [7] S.Z. Abidin, K.F. Haigh, B. Saha, Ind. Eng. Chem. Res. 51 (2012) 14653–14664. [8] M.E. Borges, L. Diaz, Renew. Sustain. Energy Rev. 16 (5) (2012) 2839–2849. [9] D.A.G. Aranda, R.T.P. Santos, N.C.O. Tapanes, A.L.D. Ramos, O.A.C. Antunes, Catal. Lett. 122 (1–2) (2008) 20–25. [10] J.M. Encinar, J.F. González, A. Rodriguez-Reinares, Fuel Process. Technol. 88 (2007) 513–522. [11] F.A.P. Voll, C. Da Silva, C.C.R.S. Rossi, R. Guirardello, F. de Castilhos, J.V. Oliveira, L. Cardozo-Filho, Biomass Bioenergy 35 (2011) 781–788. [12] M. Chai, Q. Tu, M. Lu, Y.J. Yang, Fuel Process. Technol. 125 (2014) 106–113. [13] M.L. Pisarello, B. Dalla Costa, G. Mendow, C.A. Querini, Fuel Process. Technol. 91 (9) (2010) 1005–1014. [14] M.S. Zanuttini, M.L. Pisarello, C.A. Querini, Energy Convers. Manag. 85 (2014) 407–416. [15] N. Ozbay, N. Oktar, N.A. Tapan, Fuel 87 (2008) 1789–1798. [16] A. Hykkerud, J.M. Marchetti, Biomass Bioenergy 95 (2016) 340–343.
129
Contents lists available at ScienceDirect
Applied Catalysis A, General journal homepage: www.elsevier.com/locate/apcata
Biofuels production by esterification of oleic acid with ethanol using a membrane assisted reactor in vapour permeation configuration
T
⁎
Catia Cannilla , Giuseppe Bonura, Fabio Costa, Francesco Frusteri CNR-ITAE, Istituto di Tecnologie Avanzate per l’Energia “Nicola Giordano”, Via S. Lucia 5, 98126 Messina, Italy
A R T I C LE I N FO
A B S T R A C T
Keywords: Oleic acid Esterification reaction Biodiesel Methanol Ethanol Vapor permeation Membrane
Biodiesel production by performing the oleic acid esterification reaction in a new reactor system, realized by coupling the batch reactor with a tubular permeoselective membrane, has been investigated. Such reactor configuration was adopted to remove the water produced as by-product during the reaction thus increasing the oleic acid conversion and the alkyl ester production. The activity of acid catalysts and zeolites was evaluated in presence of methanol and ethanol. The membrane resulted much more selective towards water separation when ethanol was used, allowing to force thermodynamic equilibrium at higher conversion values. Very interesting results in terms of oleic acid conversion were achieved with both methanol and ethanol (96% and 98% respectively) in presence of a commercial Amberlyst-15 catalyst, at 80 °C, at a low membrane surface to gas volume ratio (0.20 cm−1) and low acid/alcohol molar ratio (2/1 and 1/1 respectively). Over zeolite samples, conversion values higher than 60% were obtained at 100 °C after 5 h in the esterification reaction performed in vapor permeation configuration.
1. Introduction Biodiesel is a fatty acids alkyl esters (FAAEs) mixture, derived from renewable lipid feedstocks (vegetable oils, animal fats or waste cooking oils), which could be considered a suitable substitute for petroleum derivate fuels due to its environmental benefits [1]. Cetane number, energy content, viscosity and phase changes of biodiesel are similar to those of fossil diesel [2–4]. FAAEs are generally produced by transesterification reaction between the triglycerides contained into oils or fats and an alcohol; anyhow, the moisture and the free fatty acids (FFAs) present into the feedstock could react with the homogenous alkali catalyst used in the industrial process, producing soaps and gels. This reduces the FAAEs yield and deeply influences the performance and the economics of biodiesel production [5]. So, to use low-cost feedstocks (waste frying oils or fats and oils with high FFAs amount), a pretreatment esterification step is mandatory to reduce the free acidity level below 1% wt/wt. Therefore, the esterification reaction can be exploited both as a pre-treatment step of a conventional transesterification process [6,7] or as biodiesel direct production [8,9]. Methanol (MeOH), ethanol (EtOH) and butanol (ButOH) are the most employed alcohols [10]. Anyhow, the biological nature of EtOH and its extra-carbon atom,
which slightly increases the heat content and the esters cetane number, make ethyl esters (EEs) more interesting than methyl esters (MEs). Moreover, EtOH is nontoxic and has better solvent properties than MeOH for oil solubility and EEs have cloud and pour points lower than the MEs [10]. The esterification of oleic acid (OlAc) can represent well the biodiesel production, since it is present in most of oil crops [11]. With EtOH, ethyl oleate (OleOEt) is produced and water is the by product, according to the following equation: C8H17CH=CH(CH2)7COOH + CH3CH2OH ⇆ C8H17CH=CH (CH2)7COOC2H5 + H2O
(1)
The industrial esterification processes carried out with strong Brønsted acid catalysts [12–14] are not environmentally benign, since they require special treatments as neutralization involving costly catalyst separation steps from the homogeneous reaction mixtures. Heterogeneous catalysts could be easier separated, they show high activity and stability and could be reused offering milder operating conditions. Ion exchange resins [15,16], zeolites [17–19], heteropolyacids [20] and modified zirconia [21,22] are the most used solid acid catalysts for OlAc esterification. Besides, niobic acid [23], polymers with acid
Abbreviations: Ai, initial acidity; Af, final acidity; FAAEs, fatty acids alkyl esters; FFAs, free fatty acids; EEs, ethyl esters; MEs, methyl esters; MVol, micropore volume; OlAc, oleic acid; OleOEt, ethyl oleate; PV, pervaporation; PVol, pore volume; SA, surface area; Sm, surface of the membrane; TPD, temperature programmed desorption; VP, vapor permeation; Vg, volume of gas phase ⁎ Corresponding author. E-mail address: [email protected] (C. Cannilla). https://doi.org/10.1016/j.apcata.2018.08.014 Received 2 May 2018; Received in revised form 2 August 2018; Accepted 16 August 2018 Available online 18 August 2018 0926-860X/ © 2018 Published by Elsevier B.V.
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
2. Materials and methods
groups [24]; ammonium salt of TPA [25], carbon material [26] were used too. Anyhow, considering the reaction equilibrium, the water produced and that contained into waste oils could hinder the esterification, accelerating the inverse reaction. Amberlyst-15 (A-15), for example, resulted to be a suitable catalyst for OlAc esterification with MeOH [27], but the conversion rate increased only replacing MeOH and catalyst with fresh one or by removing water from the reaction mixture. For this reason, it is of great interest, from a practical point of view, to study separation processes able to eliminate the water during the reaction [28]. On this account, Son et al. [29] obtained high MEs yield by performing the OlAc esterification in a three-phase fixed bed reactor (80 °C–120 °C) coupled with a continuous water evaporation system. Nijhuis et al. [30] exploited a reactive stripping operation in a monolithic reactor; Lucena et al. [5] employed a reactor coupled to an adsorption column filled with zeolite 3A in the OlAc esterification with MeOH using H2SO4 as catalyst and shifted the biodiesel yield from 88.2% to 99.7% at 100 °C. Recently, the integration of chemical reaction with the membrane separation has received great attention: the use of a reactor equipped with a permeoselective membrane significantly improves the process efficiency favoring not only the chemical equilibrium shift but also a longer catalyst lifetime [31–33]. Specifically, pervaporation (PV) and vapor permeation (VP) technologies could be considered as clean technologies which, because of low temperatures and pressures involved, could efficiently replace conventional energy-intensive separation processes [34–37]. They are based on the use of a semipermeable active or passive membrane which allows the separation of one or more species of a mixture by their passage from one site to the other one on the basis of a sorption-diffusion mechanism. Considering that in PV, the feed is a liquid, a liquid-vapour phase change is required involving an enthalpy of vaporization, as a function of the concentration gradient between the phases on the opposite site of the interfacial barrier. Instead, in VP the feed is a vapour mixture. In such a case, no phase change or no significant temperature differences across the membrane are required, the driving force being the partial pressure of the component in the feed. A noteworthy advantage of VP configuration is that the membrane is not in contact with the liquid reaction mixture rather, thus a longer membrane lifetime is assured [38]. So far, PV process has been exploited in some liquid-phase equilibrium reactions for biofuels production, while very low papers deal with the use of VP technique. Kita et al. [39] published one of the first papers regarding gas separation in the carboxylic acids esterification with alcohols, demonstrating that, by selecting suitable permeation conditions, total conversion could be achieved. The water vapor was removed using a polyimide hollow fiber membrane module placed on the top of a stirred batch vessel; p-toluenesulfonic acid was the catalyst [39,40]. Yamamoto et al. [41] and Okamoto et al. [42] exploited the very high permeation flux, water affinity and selectivity of NaA zeolite membrane in VP configuration to improve the esterification reaction performance. In this paper, the efficiency of a coupled reactor-membrane set-up was demonstrated by performing the heterogeneous esterification of OlAc in presence of a commercial permselective membrane in VP configuration. In this case, two separate units, a batch reactor and an external separation module, were employed. This set-up allows to make clean-up or regeneration steps easier and to optimize independently both the catalytic behavior and the separation step. Solid acid catalysts, as A-15 resin and some zeolites, have been used in the esterification either with MeOH or EtOH. The obtained results have been compared with typical runs performed without the use of membrane.
2.1. Reagents OlAc purchased from Sigma Aldrich presented 99% oleic acid and 1% mixture of palmitoleic acid, palmitic acid and myristic acid, with an acid value of 139.9 mg of NaOH/g. Methanol (> 99%) or anhydrous ethanol (> 99%) were supplied from Fluka. 2.2. Catalysts Amberlyst-15 wet ion exchange was purchased from the Dow Chemical Company, whereas the zeolite catalysts were supplied by Alpha Aesar. A-15 resin was used without any pretreatment. Zeolites have been calcined for 6 h at 500 °C. Two solid acid supported catalysts have been prepared by dry impregnation of two microspherical silica (ES70Y and MS3030 by PQ Corporation) in an ethanolic solution with Hyflon® Ion S4X perfluorosulphic ionomer (20 wt%) with an equivalent weight of 730 (Solvay Solexis) [43]. 2.3. Catalyst testing OlAc esterification reaction was carried out in two stainless steel reactors of different volumes (90 and 300 cc), with a constant stirring rate of 1100 min−1. Experiments were performed under autogenous pressure by maintaining the temperature in the range 65–100 °C. The amount of catalyst, in respect to the OlAc (Rcat/OlAc) was varied from 1 to 15 wt.% and the alcohol/OlAc molar ratio (RAlcohol/OlAc) was in the range 1–1 to 20–1. The reactants and the catalysts were loaded into the reactor in the fixed ratio, nitrogen flux was employed for reactor cleaning and then the reactor was heated to the desired temperature. At the end of the reaction, the reactor was cooled down with an ice bath to condense gaseous products. The catalyst was recovered by filtration. The batch reactor and the water separation module were two distinct units and the schematic configuration of the reaction system was shown in Fig. 1. The gas phase (mainly containing the alcohol and the water) was recirculated through the water permselective membrane (HybSi), provided by Pervatech BV [35,36]. The membrane (250*10*7 mm) had an effective area of 0.005 m2, the substrate material was a γ-alumina phase whereas the top layer was a hybrid silica coated on inside of the support tube. It could operate at a maximum temperature of 150 °C and a pressure of 10 bar, in the pH range of 2–8.5, showing a permeability to pure H2O of 19.66 l/m2 bar at 150 °C. To avoid the gas phase condensation, during the reaction, the recirculation loop was maintained at 115 °C. The recirculation of gas phase was maintained constant (2.8 L/min) by using a gas-pump equipped with a variable potentiometer. At the permeate side of the membrane, the permeate was purged with a stream of carrier gas (N2, 2 L/min) and the permeate was finally obtained in the liquid state after condensation to be subsequently analyzed. 2.4. Biodiesel analysis Conversion of OlAc was calculated based on the AOCS (American Oil Chemist’s Society) official methods for acid value (Cd 3d-63) and FFA amount (Ca 5a-40). Samples were analyzed by a standard acid-base titration procedure for the evaluation of the free residual acidity. The analysis repeatability has been improved by removing methanol or ethanol in excess and water formed in a vacuum condenser, prior to submit the samples to titration. A weighed amount of the sample was dissolved in an ethanol/ethylic ether mixture 1/1 wt/wt and then was titrated with a NaOH 0.5 M solution and phenolphthalein as indicator. The volume of alkaline solution consumed was recorded and the acidity of the sample was determined by means of the Eq. (2), referred to the oil phase (oleic acid + ester) with an error less than 1–2% on the free acidity 122
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
Fig. 1. Schematic configuration of batch reactor coupled with the permeoselective membrane. The insert shows the ceramic tubular membrane [A] and the PVM-035 membrane housing [B].
3. Results and discussion
Acidity = Vtitr(mL) * Ctitr (mol/L) * MWOlAc/(1000 * gsample) * 100 (wt. %) (2)
3.1. Oleic acid esterification with MeOH
The acidity was expressed as OlAc weight percent. The OlAc conversion was calculated as the fraction of the removed acid:
The homogenous esterification reaction of oleic acid is generally carried out using H2SO4 as catalyst: a small amount of catalyst (0.01% wt/wt) is enough to promote the reaction [5]. Anyhow, heterogeneous catalysis could be more advantageous from many points of view avoiding reactor corrosion, aqueous waste production and the need of costly neutralization and separation steps. The operating conditions that mainly affect the conversion of OlAc into biodiesel are the temperature, the amount of catalyst and the alcohol/OlAc molar ratio. Specifically, the stoichiometry amount of molar ratio alcohol/OlAc is 1; however, in literature, a very high alcohol to fatty acid molar ratio has also been reported to improve the esterification reaction rate [14,44,45] by mitigating the inhibiting effect of the water formed which tends to reverse the reaction. Some preliminary experiments of esterification with MeOH have been carried out at 80 °C by using solid acid catalysts with a Rcat/OlAc equal to 5 wt.%, lower than that usually used in literature [46]. The results obtained are reported in Table 1 along with main chemicalphysical properties of catalysts used. Specifically, the performance of the A-15 resin was evaluated along with HZSM-5 zeolite with a Si/Al ratio of 23 and two Hyflon-based catalysts on spherical silica (20H730/ MS3030 and 20H730/ES70Y). High stirring rate of the mixture (1100 rpm) was employed to avoid external mass or heat transport limitations. After two hours, by employing a RMeOH/OlAc of 2 mol/mol, the activity of A-15 was the highest. This result is mainly correlable with a higher acidity of A-15 (4.50 meqH+ gcat−1) in respect of the other systems (< 1.00 meqH+ gcat−1) and confirms the good properties of the resin in terms of esterification efficiency already reported in literature [46]. By considering such preliminary results, A-15 was employed for the study finalized to optimize the reaction conditions. First of all, the influence of the alcohol amount and the reaction
OlAc conversion = (Ai − Af)/Ai * 100 where Ai and Af indicate the initial and final acidity of the solution respectively.
2.5. Catalyst characterization The textural properties of catalysts were determined by physical adsorption measurements of nitrogen to its boiling point (−196.15 °C) using a Micromeritics ASAP2020 gas adsorption device. The isotherms were elaborated for assessment of surface area (SA) and porosity (PVol), with the micropore volume (MVol) determined by the t-plot approach. Surface concentration of acidic sites was determined by temperature programmed desorption of ammonia (NH3-TPD) for zeolite. Before TPD experiment, the catalyst sample (∼100 mg) was saturated for 2 h at 150 °C under a gas mixture containing 5 vol.% of NH3/ He (flow rate = 25 stp mL/min). Then, the sample was purged in helium flow until a constant baseline level was attained. The measurement was performed in the temperature range 100–700 °C, at a rate of 10 °C/min using helium (25 stp mL/min) as the carrier flow. The evolved gases were detected by an on-line thermal-conductivity detector, calibrated by the peak area of known pulses of NH3. Potentiometric titrations were performed to evaluate the acid capacity (meq gcat−1) of the Hyflon-based systems according to the procedure elsewhere described [43].
123
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
water formation. In addition, the presence of water in the reaction medium could negatively affect the reaction rate upon adsorption on acid sites of solid catalysts, so blocking the access of the reactants [17,46]. Thus, in the attempt of overcoming such equilibrium restrictions a water permeo-selective membrane was used, coupled with the batch reactor, by exploiting the VP technology. As previously reported [35,36], the membrane was not in direct contact with the reaction liquid mixture, gas phase being forced to circulate through the membrane by a pump. The “driving force” to favor the water permeation was ensured by the pressure difference between the reactor (autogenous pressure) and the environment (permeate side). The preliminary results obtained using the permselective membrane with the methanol in presence of A-15 are reported in Table 2. The removal of water positively affected the reaction favouring the acid conversion, which increased from 67.0% (run 1, without the membrane) up to 80.2% (run 2, with the membrane) after 2 h. Nevertheless, by using a higher RMeOH/ OlAc, the acid conversion followed a sort of volcano shape trend with a maximum value for a molar ratio of 4 (83.7%). To achieve high conversion values in a short reaction time, a fundamental parameter to take into consideration is the water permeation rate. In this regard, from results reported in Table 2, it can be observed that the amount of water collected at the same reaction time (2 h) and the theoretical water permeation rate (evaluated on the OlAc conversion values) changed with the MeOH/OlAc molar ratio, decreasing with the methanol amount. Specifically, when the methanol amount was high (run 4), no permeated water was trapped. This is likely because of lower water partial pressure in the gas phase recirculated trough the membrane when higher MeOH concentration was used. In such a case, the water permeation capability of the membrane resulted to be limited. Anyhow, at low MeOH/OlAc molar ratio, the condensation system resulted to be more efficient, in fact 27.6% of theoretical water was recovered after 2 h (run 2) which increased up to 56.3% after 5 h (run 5). Naturally, when the permeation of water was more efficient, the acid conversion achievable was higher. In fact, in run 5, the acid conversion approached 97.3%, so confirming the benefit obtained in terms of equilibrium shift due to water removal through the membrane. Anyhow, by analyzing the water solution recovered during the reaction, it has been observed that a little amount of MeOH passed through the membrane along with the water (5 wt.%). This was much more evident when the reactor with a lower volume (90 cc) was used. In fact, in this case, the OlAc conversion didn’t increase, rather after 5 h, the OlAc conversion was only 74.0% (run 6) and no water was recovered. This result likely suggests that the membrane performance is not high when the water concentration in the gas phase is too low. Another factor to consider in the configuration of membrane reactor is the recirculation rate of gas phase containing the water vapor, which also plays an important role. Specifically, as shown in Fig. 3, regardless the MeOH/OlAc molar ratio used (2 or 4), higher amounts of water permeated through the membrane and therefore higher oleic acid conversion values were achieved when higher recirculation rate were used. Specifically, limiting the gas phase recirculation, or rather by filling almost completely the reactor (Sm/Vg = 0.71 cm−1, where Sm is the surface of the membrane and Vg is the volume of gas phase evaluated at the beginning of the reaction), the maximum oleic acid conversion reached after 2 h of reaction was about 64.2% and no water was collected. On the contrary, almost total conversion (96.1%) was obtained along with 53.2% of theoretical water when gas phase volume was higher (Sm/Vg ratio of 0.20 cm−1). It is evident that, with high recirculation flow, the turbulence regime necessary to improve water permeability is ensured and high conversion can be achieved.
Table 1 Main physico-chemical properties of catalysts employed in OlAc etherification (TR = 80 °C; RMeOH/OlAc = 2 mol/mol; Rcat/OlAc = 5 wt.%; time = 2 h). Catalyst
SA (m2 g−1)
PVol (cm3 g−1)
Acidity (meqH+ gcat−1)
OlAc conversion (%)
A-15a 20H730/ ES70Y 20H730/ MS3030 HZSM5 (23:1)
45 177
0.40 0.33
4.50 0.88
67.0 51.4
204
0.33
0.77
32.6
498
0.20
0.30
18.3
a Data supplied by Rohm and Haas. SABET= Surface Area. PVol = Pore Volume, determined from N2 ads/des isotherms at −196.15 °C.
Fig. 2. Esterification reaction of oleic acid with MeOH and A-15: [A] Influence of MeOH/OlAc molar ratio (TR = 80 °C; Rcat/OlAc = 5 wt.%; time = 2 h). [B] Influence of reaction time (TR = 80 °C; RMeOH/OlAc = 4 mol/mol; Rcat/OlAc = 5 wt.%).
time were evaluated at 80 °C. Fig. 2A clearly shows that the reaction rate is slightly affected by the MeOH/OlAc molar ratio considering that by increasing it from 2 up to 4, OlAc conversion increases from 67.0% to 72.1% then reaching a plateau (73.1% at RMeOH/OlAc = 8 mol/mol). So, the use of a MeOH/OlAc molar ratio higher than 4 does not produce noteworthy effects. On the other hand, Fig. 2B shows that the reaction does not take place with high reaction rate and equilibrium conditions are reached only after 6–8 h. According to the literature evidences [5,28,46], the total OlAc conversion and the production of high esters yield are limited by the reversibility of the reaction and the equilibrium constraints due to the
3.2. Oleic acid esterification with ethanol In Fig. 4, a comparison of the activity of A-15 catalyst in the oleic acid esterification with MeOH and EtOH is shown in terms of effect of 124
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
Table 2 OlAc esterification reaction in presence of MeOH with and without the membrane (TR = 80 °C; Rcat/OlAc = 5 wt.%). Run
RMeOH/OlAc (mol/mol)
Time (h)
OlAc conversion (%)
Sm/Vg (cm−1)a
H2Oteor. (g)b
H2Oteor. (g)c
H2OColl (%)d
H2O permeation rate (h−1)e
1f 2 3 4 5 6g
2 2 4 8 2 2
2 2 2 2 5 5
67.0 80.2 83.7 79.8 97.3 74.0
0.27 0.27 0.27 0.32 0.72
5.10 4.33 3.25 6.37 0.96
4.09 3.63 2.60 6.04 0.71
27.6 20.1 – 56.3 –
0.066 0.032 0.014 0.038
a b c d e f g
Sm = surface of the membrane; Vg = volume of gas phase evaluated at the beginning of the reaction. Water amount evaluated by theoretical total OlAc conversion. Theoretical water amount evaluated by measured OlAc conversion. Percentage of water collected as a function of theoretical water (b). Value obtained considering the theoretical water amount produced by measured OlAc conversion. Run carried out without membrane. Run carried out by using a 90 cc reactor.
Fig. 3. Influence of Sm/Vg ratio on the esterification reaction of oleic acid with MeOH in presence of membrane (catalyst: A-15; TR = 80 °C; RMeOH/ OlAc = 2 mol/mol; Rcat/OlAc = 5 wt.%; time = 2 h). Triangle symbols are referred to the run with RMeOH/OlAc = 4 mol/mol.
reaction temperature on OlAc conversion and activation energy is shown. Esterification with methanol resulted to be more endothermic than with ethanol [45] and a pseudo first order dependence for fatty acid esterification emerged. From the conversion data obtained at different temperatures, the apparent activation energy was determined by Arrhenius plot. The activation energies found were 31.5 kJ/mol and 28.0 kJ/mol for methanol and ethanol respectively, values lower than that reported by other authors on solid catalysts [20,45] although in the range of the existing results [12,14,47]. In a prevailing kinetic regime, the esterification reaction with ethanol (1.16 * 10−5 molEEgcat−1 s−1) takes place with a lower rate than in presence of methanol (8.03 * 10−6 molMEgcat−1 s−1). This result is confirmed in literature for palmitic acid esterification, the order of effectiveness being methanol > ethanol, due to the different reactivity of the carbon chains of each alcohol [42]. In Table 3 the comparative results of the esterification reaction of OlAc with EtOH carried out in the two batch reactors of different volume (90 and 300 cc) with and without the membrane, are reported. First of all, it is possible to observe that, after 2 h at 80 °C, the OlAc conversion, with membrane by-passed, was 50.0% and 57.8% depending on the reactor capacity (runs 1–7), lower than that obtained with methanol (67.0%, see Table 2) as expected on the basis of the kinetic results. By using the membrane coupled with the reactor of smaller volume, the conversion increased up to 63.2% (run 2). Also in this case, with a higher alcohol/oleic acid molar ratio, 8 mol/mo (run 4), OlAc conversion slightly decreased to 52.2%, likely due to a lower concentration of water vapour in the gas phase. It is interesting to highlight the better efficiency obtained with the membrane when the
Fig. 4. Esterification reaction of oleic acid with MeOH and EtOH: [A] Influence of reaction temperature (Rcat/OlAc = 1 wt.%; time = 1 h; RAlc/OlAc = 5 mol/ mol). [B] Arrhenius plot.
ethanol is the reactant. In fact, after 5 h of reaction, by using the small reactor, the OlAc conversion was almost total (run 5, 98.1%) unlike the reaction with MeOH (74.0%, see Table 2, run 6). This evidence indicates that the membrane has a higher selectivity towards the water permeation when ethanol is employed instead of methanol. Anyhow, by using the reactor at a capacity as low as 90 cc, the water permeated during reaction was not recovered due to its low concentration; for this reason it was preferred to operate with a larger batch reactor (300 cc). In this case, with a REtOH/OlAc of 2 mol/mol, the oleic acid conversion increased from 57.8% (run 7) without membrane up to 78.6% after 2 h (run 9) with the VP system (under the same amount of reactants) but unfortunately only about 10% of theoretical 125
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
Table 3 Oleic acid esterification reaction in presence of EtOH, with and without the permeoselective membrane (TR = 80 °C; Rcat/OlAc = 5 wt.%). Time (h)
OlAcConversion (%)
Sm/Vg (cm−1)a
H2O (g)
H2Oteor. (g)b
H2Oteor. (g)c
H2OColl (%)d
Autoclave 90 cc 2 1e 2 2 3 4 4 8 5 2
2 2 2 2 5
50.0 63.2 60.0 52.2 98.1
0.74 0.72 0.71 0.74
– – – –
0.96 0.67 0.45 0.96
0.60 0.40 0.23 0.94
– – – –
Autoclave 300 cc 6e 1 7e 2 8 1 9 2 10 1 11 1 12 2
2 2 2 2 5 2 2
59.0 57.8 85.0 78.6 98.2 77.6 73.8
0.20 0.20 0.20 0.28 0.28
1.07 0.19 1.80 2.48 1.14
2.74 2.35 2.74 5.87 5.10
2.32 1.83 2.69 4.55 3.76
46.01 10.47 66.90 54.60 30.28
Run
a b c d e
REtOH/OlAc (mol/mol)
Sm = surface of the membrane; Vg = volume of gas phase evaluated at the beginning of the reaction. Water amount evaluated by theoretical total OlAc conversion. Theoretical water amount evaluated by measured OlAc conversion. Percentage of water collected as a function of theoretical water (b). Runs carried out without membrane.
surface area value (1027 m2 g−1), followed by BEA zeolite (∼800 m2/ g); others investigated zeolite samples showed similar values, with an average surface exposure of about 500 m2/g. The determined surface area values reflect the micropores volume resulting higher for Y and BEA zeolites (0.27 cm3 g−1 and 0.21 cm3 g−1 respectively). Surface acid properties are also reported in Table 4 in terms of quantitative data obtained by NH3-TPD measurements (desorption profiles are shown in Fig. S1 of Supplementary data). As expected, for all samples, total acidity was lower than that of Amberlyst-15. Two main peaks characteristic of the interaction of ammonia with weak and strong acid sites at low (150–250 °C) and high temperature (350–450 °C), were observed. As well known, by NH3-TPD profiles it is not possible to discriminate between Brønsted and Lewis acidity, anyhow it is generally accepted that the peak at lower temperature is associated both to the presence of Lewis sites on zeolite framework or to extra-framework aluminum oxide/hydroxide species. Instead, the peak at higher temperature is generally ascribable to the presence of Brønsted acid sites due to the framework Al ions [49]. On this account, the Y zeolite showed the highest total acidity (1211 μmol g−1), whereas in the HZSM-5 samples acidity decreased by increasing the SiO2/Al2O3 ratio. Preliminary experiments carried out with EtOH at 80 °C revealed that, at this temperature, all zeolites investigated were not so active as A-15 catalyst, therefore, using a molar REtOH/OlAc of 2/1, the reaction temperature was increased at 100 °C. Without the use of the membrane, the higher OlAc conversion achieved was around 23%, and the
water was collected. Instead, by employing a REtOH/OlAc of 1 mol/mol, OlAc conversion value increased of about 30% thanks to the membrane effect (run 6, 59.0%; run 8, 85.0%) and the percentage of the collected water was higher (46,01%). The gas chromatographic analysis of the liquid mixture collected showed a very low amount of EtOH (< 0.1 wt. %) permeated through the membrane and thus confirmed a higher selectivity towards ethanol/water separation (than methanol/water). By further increasing the reaction time, it was possible to achieve almost complete OlAc conversion (i.e. 98.2%) by maintaining mild reaction conditions (run 10). As already seen with MeOH, even in this case, the Sm/Vg ratio influences the performance of the membrane; in fact by increasing this value from 0.20 cm−1 up to 0.28 cm−1, the conversion decreased from 85.0 to 77.6% with a REtOH/OlAc of 1 mol/mol and from 78.6% to 73.8% with a REtOH/OlAc of 2 mol/mol. These very interesting results show a significant improvement in the esterification rate with EtOH under milder reaction conditions if compared with recent literature data [16,48].
3.3. Oleic acid esterification with EtOH using zeolites In the attempt to find more resistant solid acid catalysts for this reaction, a series of zeolites have been investigated. Such zeolites were characterized from a structural and morphological point of view and main results are reported in Table 4. As regards textural properties, Y zeolite exhibited the highest
Table 4 Textural properties and quantitative data of NH3-TPD measurements and acid sites distribution of the investigated zeolites samples. Sample
Y FER HZSM5 BEA HZSM5 HZSM5 a b c d e f
SiO2/Al2O3 (mol/mol)
12 20 23 38 50 80
Textural properties
Surface acidic properties
SALang (m2/g)
PVola (cm3/g)
MVolb (cm3/g)
Total sites (μmol g−1)
w-m sitesc
m-s sitesd
TM1e (°C)
TM2f (°C)
1027 449 498 781 539 574
0.52 0.21 0.20 0.34 0.25 0.29
0.27 0.14 0.13 0.21 0.11 0.11
1211 527 542 871 303 73
0.65 0.62 0.34 0.35 0.56 0.66
0.35 0.38 0.66 0.65 0.44 0.34
219 206 223 194 212 216
376 439 407 413 396 377
Cumulative Pore Volume, evaluated in the range 5–600 Å (micro and mesopores). Micropores Volume evaluated by t-Plot. Weak-medium acid site fraction with temperature of maximum desorption of NH3 in the range 100–300 °C. Medium-strong acid site fraction with temperature of maximum desorption of NH3 above 300 °C. Temperature of maximum desorption of NH3 in the range 100–300 °C. Temperature of maximum desorption of NH3 above 300 °C. 126
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
Fig. 5. [A] Experimental results for catalytic screening in OlAc esterification reaction as a function of SiO2/Al2O3 (TR = 100 °C; REtOH/OlAc = 2 mol/mol; Rcat/OlAc = 5 wt.%, time = 2 h). [B] Total acid sites and weak-medium acid site fraction vs. SiO2/Al2O3 ratio.
reactivity of different zeolites was the following: HZSM5 (23:1)
(80:1) > Y > FER > HZSM5
(50:1) > BEA > HZSM5 Fig. 6. OlAc esterification reaction with ethanol. [A] Influence of EtOH/OlAc molar ratio (catalyst: 5 wt.% referred to the OlAc amount; TR = 100 °C, time = 2 h). [B] Influence of catalyst amount (TR = 100 °C; REtOH/OlAc = 2 mol/ mol; time = 14 h; * indicates REtOH/OlAc = 20 mol/mol).
From the results reported in Fig. 5A it was not possible to establish a clear correlation between the catalytic activity and zeolite properties (see Table 4), however, it was observed that an optimum bulk Si/Al molar ratio maximizes the catalytic behavior. This evidence could be correlated to the hydrophobic character of zeolite surface. In fact, as reported in literature, to a lower Si/Al ratio, it is associated a higher capability to coordinate water molecules, with consequent covering of the solid surface which limits the adsorption of organic substrates [50]. Specifically, this was the trend observed for HZSM-5 zeolites, for which a higher Si/Al molar ratio that could be correlated to a higher hydrophobicity also reflected in a lower acid site availability along with a higher amount of weak/medium acid site fraction (Fig. 5B). Anyhow, the fitting reported in Fig. 5A was obtained without considering the Y zeolite which was characterized by a high OlAc conversion of 21.22%, but, at the same time, by a Si/Al ratio of 12. This catalytic result was likely ascribable to the synergic role played by both the acid surface properties and the textural properties. Specifically, the catalyst porosity generally represents a key factor for biodiesel production since it strictly influences the active sites accessibility of big molecules involved in the esterification reaction [3]. Considering the high activity of HZSM-5 with a Si/Al ratio of 80, it was further employed for new experiments in the attempt at optimizing the reaction conditions. Using such a zeolite as catalyst, it was considered important to exclude the occurrence of mass transfer phenomena during the reaction. In fact, as well known, in a liquid phase reaction, two types of mass transfer resistances should be considered: the external mass transfer resistance, which takes place across the solid-liquid interface and the internal mass transfer resistance, associated with the difference in particle size distribution of the catalyst [7]. Considering the high
stirring speed utilized in the experiments, the first one could be overlooked, whereas bigger attention should be paid on the internal mass transfer resistances. For this reason, a set of experiments were conducted using the HZSM5-80 catalyst particle size in the 16–70 μm range. The results summarized in Fig. S2 showed no effect of the particle size on the OlAc conversion suggesting that there was no intraparticle diffusion control. The effect of the ethanol/OlAc molar ratio was investigated in the range 1–20 mol/mol. The results reported in Fig. 6A show a volcanotype trend indicating that the suitable molar ratio was comprised between 2 and 4. This result partially fits with data obtained by Marchetti et al. [17] which observed that, under their reaction conditions (60 °C and 20 wt.% of catalyst), the 1:1 ethanol/oleic acid molar ratio gave the highest reaction rate. This evidence was justified by considering that the reaction was not at the equilibrium level but still in an initial stage. However, it should be also considered that, when an excess of ethanol was used, it could be preferentially adsorbed onto the acid sites thus decreasing their availability. Furthermore an excessive dilution of solution could cause a decrease in acid sites concentration with negative effect on reaction kinetics [51]. Such findings suggested to optimize the reaction conditions by employing a higher zeolite amount for a longer reaction time. Specifically, as shown in Fig. 6B, with a Rcat/OlAc of 15 wt.%, OlAc conversion values of 62.2% and 70.1% were achieved for a EtOH/ OleOH molar ratio of 2 and 20 respectively, after 14 h of reaction. At this point, the VP module was coupled to the batch reactor and different reaction conditions were studied in presence of the HZSM5-80 127
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
no membrane
Sm/Vg = 0.74 cm-1
membrane
Sm/Vg = 0.20 cm-1 Rcat/OlAc = 5 wt% REtOH/OlAc = 2/1 mol/mol
90 Rcat/OlAc = 15 wt% REtOH/OlAc = 20/1 mol/mol time = 14 h
80 70
Fig. 7. OlAc esterification reaction with EtOH. Comparison of data with and without membrane at 100 °C. A) Rcat/OlAc = 5 wt.%; REtOH/OlAc = 2/1 mol/mol; t = 2 h; Sm/ Vg = 0.74 cm−1. B) Rcat/OlAc = 15 wt.%; REtOH/OlAc = 20/1 mol/mol; t = 14 h; Sm/Vg = 0.74 cm−1. C) Rcat/OlAc = 5 wt.%; REtOH/OlAc = 2/1 mol/mol; t = 2 h; Sm/ Vg = 0.20 cm−1. D) Rcat/OlAc = 5 wt.%; REtOH/OlAc = 2/1 mol/mol; t = 5 h; Sm/ Vg = 0.20 cm−1.
time = 5 h
60 50 time = 2 h
40 30
Rcat/OlAc = 5 wt% REtOH/OlAc = 2/1 mol/mol time = 2 h
20 10 0
catalyst. The results reported in Fig. 7 show that by using a Sm/Vg of 0.74 cm−1, at 100 °C, with a REtOH/OleOH of 2 mol/mol (A condition), the oleic acid conversion value didn’t increase using the membrane. The membrane effect was not so evident by performing the etherification reaction under the optimized reaction conditions previously found without the use membrane (see Fig. 6), that is for a long reaction time (14 h), high ethanol/OlAc molar ratio (20/1 mol/mol) and 15 wt.% of catalyst (B condition). Anyhow, a very interesting result was obtained when the recirculation rate of gas phase was increased by employing the optimized value of Sm/Vg ratio equal to 0.20 cm−1. In this case, in fact, at 100 °C, the OlAc conversions were doubled as evidenced by looking at C and D conditions, so achieving a value of 35.4% after 2 h and 62.1% after 5 h of reaction, at low EtOH/OlAc molar ratio (2 mol/ mol) and with 5 wt.% of catalyst. In Fig. 8, three reproducibility tests with HZSM5-80, in presence of membrane, are reported. The runs were carried out at 100 °C, REtOH/ OlAc = 2 mol/mol and 5 wt.% of catalyst. As it is possible to observe, after three experiments, the OlAc conversion remained almost constant. After each experiment, the catalyst was washed by hot distilled water and filtered prior to drying at 110 °C and the calcination at 500 °C for 3 h. Therefore, the obtained results suggest that, before a new run, zeolite samples need a simple regeneration step to restore their initial surface features and so ensuring a complete reproducibility of catalytic activity upon reusing. In any case, irrespective of the catalytic system employed, the potential of the tandem reactor-membrane setup for biofuels production by equilibrium-limited esterification reaction was fully demonstrated. Specifically, the optimization of the catalytic process-membrane system allowed obtaining higher conversion rate of the substrate at relatively short reaction time, by employing low alcohol-to-FFA molar ratios, with consequent several economic and ecological advantages.
Fig. 8. Catalyst stability (TR = 100 °C; Rcat/OlAc = 5 wt.%; REtOH/OlAc = 2/ 1 mol/mol; time = 5 h; Sm/Vg = 0.20 cm−1).
using a batch reactor coupled with a membrane separation system in VP configuration to drive the esterification reaction of oleic acid have been clearly demonstrated. By optimizing the reaction setup, a total oleic acid conversion can be achieved by continuously subtracting the water
4. Conclusions The catalytic and technological benefits that can be obtained by 128
Applied Catalysis A, General 566 (2018) 121–129
C. Cannilla et al.
formed from the reaction medium. A higher selectivity of the membrane towards ethanol rather than methanol favours an enhanced esterification rate. An efficient water permeation through the membrane is ensured by a recirculation rate of the gas phase in turbulent flow regime. From the catalytic point of view, a higher catalyst acidity of A15 favours the conversion in relatively mild reaction conditions (TR = 80 °C, REtOH/OlAc = 1 mol/mol; Rcat/OlAc = 5 wt%) although a slight increase of the temperature is beneficial if zeolite samples are used.
[17] J.M. Marchetti, A.F. Errazu, Fuel 87 (2008) 3477–3480. [18] A.M. Doyle, T.M. Albayati, A.S. Abbas, Z.T.A. Lismaeel, Renew. Energy 97 (2016) 19–23. [19] K.H. Chung, B.G. Park, J. Ind. Eng. Chem. 15 (2009) 388–392. [20] K. Srilatha, Ch.R. Kumar, B.L.A.P. Devi, R.B.N. Prasad, P.S. Prasad, N. Lingaiah, Catal. Sci. Technol. 1 (2011) 662–668. [21] D.E. Lopez, J.G. Goodwin Jr, D.A. Bruce, S. Futura, Appl. Catal. A 339 (2008) 76–83. [22] Y.M. Park, J.Y. Lee, S.H. Chung, I.S. Park, S.Y. Lee, D.K. Kim, J.S. Lee, K.Y. Lee, Bioresour. Technol. 101 (2010) S59–S61. [23] L.L. Rade, C.O.T. Lemos, M.A.S. Barrozo, R.M. Ribas, R.S. Monteiro, C.E. Hori, Renew. Energy 115 (2018) 208–216. [24] C.S. Caetano, L. Guerriero, I.M. Fonseca, A.M. Ramos, J. Vital, J.E. Castanheiro, Appl. Catal. A 359 (2009) 41–46. [25] B.Y. Giri, K.N. Rao, B.L.A.P. Devi, N. Lingaiah, I. Suryanarayana, R.B.N. Prasad, P.S. Sai Prasad, Catal. Commun. 6 (2005) 788–792. [26] Z. Gao, S. Tang, X. Cui, S. Tian, M. Zhang, Fuel 140 (2015) 669–676. [27] J.Y. Park, D.K. Kim, J.S. Lee, Renew. Energy 35 (2010) 614–618. [28] K. Suwannakarn, E. Lotero, K. Ngaosuwan, J.G. Goodwin Jr., Ind. Eng. Chem. Res. 48 (2009) 2810–2818. [29] S.M. Son, H. Kimura, K. Kusakabe, Bioresour. Technol. 102 (2) (2011) 2130–2132. [30] T.A. Nijhuis, A.E.W. Beers, F. Kapteijn, J.A. Moulijn, Chem. Eng. Sci. 57 (9) (2002) 1627–1632. [31] E. Drioli, A.I. Stankiewicz, F. Macedonio, J. Membr. Sci. 380 (2011) 1–8. [32] F. Gallucci, A. Basile, F.I. Hai, Introduction—a review of Membrane Reactors, Membrane Reactors, John Wiley & Sons, Ltd., Chichester, UK, 2011, pp. 1–61. [33] S.H. Shuit, Y.T. Ong, K.T. Lee, B. Subhash, S.H. Tan, Biotechnol. Adv. 30 (2012) 1364–1380. [34] C. Cannilla, G. Bonura, F. Frusteri, Catalysts 7 (2017) 187. [35] C. Cannilla, G. Bonura, L. Frusteri, F. Frusteri, Environ. Sci. Technol. 48 (2014) 6019–6026. [36] C. Cannilla, G. Bonura, L. Frusteri, F. Frusteri, Chem. Eng. J. 282 (2015) 187–193. [37] F. Lipnizki, R.W. Field, P.K. Ten, J. Membr. Sci. 153 (1999) 183–210. [38] B. Bolto, M. Hoang, Z. Xie, Water Res. 46 (2012) 259–266. [39] H. Kita, K. Tanaka, K. Okamoto, M. Yamamoto, Chem. Lett. (1987) 2053–2056. [40] K.I. Okamoto, M. Yamamoto, S. Noda, T. Semoto, Y. Otoshi, K. Tanaka, H. Kita, Ind. Eng. Chem. Res. 33 (1994) 849–853. [41] M. Yamamoto, N. Munehisa, M. Kaibara, K. Horii, K. Tanaka, H. Kita, K. Okamoto, Membrane 20 (1995) 143–148. [42] K.I. Okamoto, H. Kita, Ind. Eng. Chem. Res. 40 (2001) 163–175. [43] F. Frusteri, L. Frusteri, C. Cannilla, G. Bonura, Bioresour. Technol. 118 (2012) 350–358. [44] A.A. Kiss, A.C. Dimian, G. Rothenberg, Adv. Synth. Catal. 348 (2006) 75–81. [45] K. Srilatha, N. Lingaiah, B.L.A. Prabhavathi Devi, R.B.N. Prasad, S. Venkateswar, P.S.S. Prasad, Appl. Catal. A 365 (2009) 28–33. [46] J.Y. Park, D.K. Kim, J.S. Lee, Bioresour. Technol. 101 (2010) S62–S65. [47] M.C.G. Albuquerque, C.L. Cavalcante Jr., A.E.B. Torres, D.C.S. Azevedo, E.J.S. Parente Jr., Energy Fuels 23 (2009) 1136–1138. [48] Y. Jiang, J. Lu, K. Sun, L. Ma, J. Ding, Energy Convers. Manag. 76 (2013) 980–985. [49] G. Bonura, C. Cannilla, L. Frusteri, A. Mezzapica, F. Frusteri, Catal. Today 281 (2017) 337–344. [50] P.M. Veiga, A.C.L. Gomes, C.O. Veloso, C.A. Henriques, Appl. Catal. A 548 (2017) 2–15. [51] K.C.S. Figueiredo, V.M.M. Salim, C.P. Borges, Braz. J. Chem. Eng. 27 (2010) 609–617.
Funding sources Part of this work was supported by the Bilateral Agreement of Scientific and Technological Cooperation CNR-MTA/HAS (Hungarian Academy of Science). Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.apcata.2018.08.014. References [1] C. Cannilla, G. Bonura, F. Arena, E. Rombi, F. Frusteri, Catal. Today 195 (1) (2012) 32–43. [2] P. Sivakumar, K. Anbarasu, S. Renganathan, Fuel 90 (2011) 147–151. [3] C. Cannilla, G. Bonura, E. Rombi, F. Arena, F. Frusteri, Appl. Catal. A 382 (2) (2010) 158–166. [4] C. Beatrice, G. Di Blasio, C. Guido, C. Cannilla, G. Bonura, F. Frusteri, Appl. Energy 132 (1) (2014) 236–247. [5] I.L. Lucena, G.F. Silva, A.N.F. Fabiano, Ind. Eng. Chem. Res. 47 (2008) 6885–6889. [6] B. Škrbic´, Z. Predojevic´, N. Đurišic´-Mladenovic, Waste Manag. Res. 33 (8) (2015) 723–729. [7] S.Z. Abidin, K.F. Haigh, B. Saha, Ind. Eng. Chem. Res. 51 (2012) 14653–14664. [8] M.E. Borges, L. Diaz, Renew. Sustain. Energy Rev. 16 (5) (2012) 2839–2849. [9] D.A.G. Aranda, R.T.P. Santos, N.C.O. Tapanes, A.L.D. Ramos, O.A.C. Antunes, Catal. Lett. 122 (1–2) (2008) 20–25. [10] J.M. Encinar, J.F. González, A. Rodriguez-Reinares, Fuel Process. Technol. 88 (2007) 513–522. [11] F.A.P. Voll, C. Da Silva, C.C.R.S. Rossi, R. Guirardello, F. de Castilhos, J.V. Oliveira, L. Cardozo-Filho, Biomass Bioenergy 35 (2011) 781–788. [12] M. Chai, Q. Tu, M. Lu, Y.J. Yang, Fuel Process. Technol. 125 (2014) 106–113. [13] M.L. Pisarello, B. Dalla Costa, G. Mendow, C.A. Querini, Fuel Process. Technol. 91 (9) (2010) 1005–1014. [14] M.S. Zanuttini, M.L. Pisarello, C.A. Querini, Energy Convers. Manag. 85 (2014) 407–416. [15] N. Ozbay, N. Oktar, N.A. Tapan, Fuel 87 (2008) 1789–1798. [16] A. Hykkerud, J.M. Marchetti, Biomass Bioenergy 95 (2016) 340–343.
129