Kinetics and modeling of fatty acids esterification on acid exchange resins

https://doi.org/10.1016/j.cej.2009.12.050Get rights and content

Abstract

Biodiesel, a renewable fuel of vegetal origin, has been an object of a rapidly growing interest, in the latest years, both as a pure fuel and as blending component to reduce exhaust pollutants of traditional diesel fuel. Biodiesel is conventionally produced through a well-established technology that involves the use of alkaline catalysts and is, therefore, not compatible with the presence of free fatty acids (FFAs) in the feedstock due to the formation of soaps. Also the presence of FFA in small amounts is detrimental, because, formed soaps strongly affect the successive glycerol separation giving place to a long settling time. Normally, highly refined vegetable oils are used as raw materials for biodiesel production. A preliminary stage of acidity reduction is necessary, when the starting material is characterized by a high free acidity (higher than 0.5% by weight). This pre-treatment can be pursued, as example, by means of an esterification reaction of the FFAs with methanol, catalyzed by sulphonic ionic exchange resins. In the present work, a batch reactor has been used for the study of the above-mentioned reaction and different acid ionic exchange resins have been tested as heterogeneous catalysts. Two kinds of substrates have been submitted for esterification with methanol: a model mixture of soybean oil artificially acidified with oleic acid and a commercial high-acidity mixture of waste fatty acids (oleins). A detailed kinetic model has been developed and tested in which the following key phenomena, characterizing the system, have been introduced: (i) the physical phase equilibrium (partitioning equilibrium) of the components between the resin-absorbed phase and the external liquid phase; (ii) the ionic exchange equilibria; (iii) an Eley–Rideal surface reaction mechanism. The developed kinetic model was able to correctly interpret all the experimental data collected, both as a function of the temperature and of the catalyst concentration.

Introduction

Biodiesel represents a valuable alternative to petroleum-derived fuels due to both its renewable nature and its substantially reduced net carbon dioxide emission. This biofuel is conventionally produced through batch or continuous transesterification of highly refined vegetable oils with methanol by using homogeneous alkaline catalysts such as sodium or potassium hydroxides or methoxides [1], [2]. The mentioned technology is not compatible with oils which free fatty acids (FFAs) content exceeding a threshold value of about 0.5% by weight. The main limitation for a wider biodiesel market share is represented by the relatively high cost of the raw material: the steps of production, transportation, storage and refining of vegetable oils affect for more than 85% of the total cost of biodiesel [3].

A possible solution to this drawback could be the development of new technologies enabling to employ waste raw materials such as fried oils or mixture of oils from various sources that cannot be treated in the conventional process for their high content in free fatty acids. This perspective discloses the way toward the development of innovative biodiesel production processes such as those based on supercritical methanol [4], or the two-stage process (esterification and transesterification reaction) [5], [6]. The esterification reaction of acid oils or fats can then be used both as biodiesel direct production (in the case of substrates with very high content of FFAs) and as pre-treatment step in the framework of a conventional transesterification process (for feedstock with moderate free acidity). The generic esterification reaction of a carboxylic acid with methanol, producing methylester and water, is schematically shown below:RCOOH(A)+CH3OH(M)RCOOCH3(E)+H2O(W)

The esterification processes for FFAs abatement are generally promoted by homogeneous acid catalyzed reaction [5], [6] or by ionic exchange acid resins as heterogeneous catalysts. These resins are constituted by a cross-linked polymeric matrix on which the active sites for the esterification reaction are represented by protons bonded to sulphonic groups. However, due to their particular structure, these resins are subjected to a remarkable swelling phenomenon [7] when contacted with polar solvents and, considering the components involved in the FFAs esterification reaction, the resin shows a high tendency to incorporate mostly both the water formed by reaction and methanol used as esterification agent. This occurs for both the relatively high polarity of water and methanol and for their reduced molecular size that correspond to an increased diffusional rate inside the pores of the polymeric matrix. According to this selective mechanism, in the interior of the resin an environment is created that strongly differs in concentration from the bulk phase. This aspect plays a fundamental role both on kinetics and on chemical equilibrium, because, the reaction occurring almost completely on the internal volume of the resin, is strongly affected by the partitioning of the component between the absorbed and bulk liquid phases. The correct description of the kinetics for such systems requires then additional information (separately collected) regarding the phase partitioning of the various components between the liquid and the absorbed phase. A further complication is represented by the lack of partitioning data collected in correspondence to the reaction temperature. The mentioned phenomenon is widely described in the literature by means of more or less complex models but this aspect is neglected in some papers that deal with the esterification of fatty acids and the experimental data are often correlated by using a pseudo-homogeneous model, mathematically much simpler but inadequate to the interpretation of the real reaction mechanism. An example of this kind of approach is reported by Pasias et al. [8] who have investigated the FFAs esterification reaction catalyzed by Purolite resin and have interpreted their kinetic data by using a pseudo-homogeneous equilibrium model.

Marchetti et al. [9] have studied this reaction by using, on the contrary, basic resins as catalysts like Dowex monosphere 550 A and Dowex upcore Mono A-625 obtaining interesting results, but without introducing a modeling approach. Tesser et al. [10] reported the esterification reaction kinetics of oleic acid with methanol in the presence of triglycerides, catalyzed by acid resin Resindion Relite CFS in a batch reactor. Furthermore, Santacesaria et al. [11], [12] have shown that the esterification reaction, performed in a continuous packed bed tubular reactor (PBR), was strongly affected by external mass transfer limitations, while, good results can be achieved by adopting alternative reactor configurations represented by the well stirred slurry reactor (WSSR) and the spray tower loop reactor (STLR).

Many authors have studied the esterification reactions on different cationic exchange resins but mainly related to short chain fatty acids with different alcohols and adopting different kinetic modeling approaches focused on the experimental data correlation but always neglecting the partitioning phenomenon occurring between the interior and the exterior liquid phase with respect to the resins.

For example, Sanz et al. [13] have studied the kinetics of lactic acid esterification reaction with methanol, catalyzed by different acidic resins. These authors have interpreted their results by means of an Eley–Rideal (ER) and a Langmuir–Hinshelwood (LH) model.

Ali and Merchant [14] have reported kinetic results related to the esterification between acetic acid and 2-propanol on different resins interpreting the obtained data with various models: pseudo-homogeneous (PH), Eley–Rideal (ER), Langmuir–Hinshelwood (LH) and a modified Langmuir–Hinshelwood (MLH). The last one contains an empirical exponent, for water concentration, in order to take into account for the greater affinity of water for the catalytic resin. Ali and Merchant [14] have statistically compared the various models tested and the MLH model resulted the best for the description of their experimental data, as a demonstration that the partitioning phenomenon plays a fundamental role in the kinetics description.

Also Yalcinyuva et al. [15] have demonstrated that the phase partitioning and the swelling phenomena cannot be neglected for an accurate description of an esterification process involving ionic exchange resins. In fact, they have investigated the esterification reaction of myristic acid with isopropyl alcohol catalyzed by Amberlyst 15 and they have shown that the water concentration in the reactive mixture resulted always lower than the theoretical value calculated on the basis of the measured acid conversion.

Mazzotti et al. [16] reported a kinetic study of acetic acid esterification with methanol, in the presence of Amberlyst 15 as catalyst in which the partitioning phenomenon has been interpreted on the basis of a Flory–Huggins model.

Popken et al. [17] and Song et al. [18] have studied the reaction of acetic acid with methanol also using Amberlyst 15 as catalyst. A mass-based Langmuir model has been used for describing the absorption, while, the kinetic behavior has been interpreted with LH model. At last, a recent review on properties and uses of exchange resins has been published by Alexandratos [19].

The purpose of our work is, therefore, the study of the esterification reaction performed on the following acid substrates: (i) a model mixture of soybean oil containing controlled amounts of oleic acids; (ii) commercial mixtures of oils and FFA (oleins) characterized by high FFA concentrations (also greater than >95% by wt as oleic acid). Methanol has been chosen as esterification agent, while, selected acid ion-exchange resins are: Amberlyst 15 and Relite CS. The catalyst selection has been based on both a previous catalysts screening activity and on the availability of literature information. As a matter of fact, Amberlyst 15 is one of the more widely used and studied resin and some absorption and kinetic data are available in the literature. Relite CFS has been chosen because it is less expensive.

The experimental runs, performed in a batch reactor, have been interpreted with a new kinetic model based on an ionic exchange reaction mechanism that takes into account also for the physical partitioning effects of the various components of the reacting mixture between the liquid phase internal and external to the resin. This phase partitioning effect has been studied separately by means of some specific measurements, conducted both at 25 and 100 °C for what concerns the binary system methanol–water. A simple partition–absorption model, suitable to be embedded into the kinetic model, has been developed for the description of the experimental binary phase equilibrium data. The absorption model resulted formally equivalent to a Langmuir model based on the resin void volume available for the absorption.

Section snippets

Reactants and methods

The used reactants and the related purities are the following: methanol (Aldrich, purity >99%, w/w), oleic acid (Carlo Erba, purity >90%, w/w), and a commercially available acidity-free soybean oil (acidity <0.3%, w/w). The oleins have been furnished by a local company (Parodi S.r.L.) and their acid composition is shown in Table 1.

The resins Amberlyst 15 and Amberlyst 16 have been purchased by Acros Organics, Amberlyst 131 by Sigma–Aldrich and Relite CFS has been purchased by Resindion.

These

External diffusive phenomena

A preliminary investigation has been conducted in order to evaluate the influence of mass transfer limitations on the measured kinetic. The extent of the external diffusion has been verified by doing experiments at different stirring rates (500, 1000, 1200 and 1500 rpm), and a constant reaction rate has been observed above the threshold value of about 1200 rpm (see Fig. 2). All the experimental runs have been performed in conditions in which all external diffusive phenomena can be neglected (1500 

Conclusions

In this paper the esterification reaction of free fatty acids in high-acidity substrates has been studied and modeled. Different acid ionic exchange polymeric resins have been tested as catalysts in a batch reactor and two of them have been selected after a preliminary screening. On the selected catalyst (Amberlyst 15 and Relite CFS) different kinetic runs have been performed and the influence of temperature and catalyst concentration has been studied. A detailed kinetic/equilibrium model has

Acknowledgement

Thanks are due to the Italian Ministry of Foreign Affairs (MAE) for the financial support.

    List of symbols

    a

    acidity (%wt in oleic acid)

    C

    concentration (mol mL−1)

    Ctitr

    concentration of titrant (mol mL−1)

    EA

    activation energy (kcal mol−1)

    H

    ionic exchange equilibrium constant

    k

    kinetic constant of uncatalyzed reaction (mL2 mol−2 min−1)

    kcat, k−cat

    kinetic constants of the forward and the reverse reaction (mL−1 gcat−1 min−1)

    kabs

    kinetic constant for absorption process (min−1)

    kdes

    kinetic constant for desorption process (mol min−1 cm−3)

References (23)

  • E. Santacesaria et al.

    Kinetics and mass transfer of free fatty acids esterification with methanol in a tubular packed bed reactor: a key pretreatment in biodiesel production

    Ind. Eng. Chem. Res.

    (2007)
  • Cited by (114)

    View all citing articles on Scopus
    View full text