Article pubs.acs.org/jced
Separation of Itaconic Acid from Aqueous Solution onto IonExchange Resins Antonio Irineudo Magalhaẽ s, Jr.,† Júlio Cesar de Carvalho,*,† Elia Natália Meza Ramírez,† Jesus David Coral Medina,† and Carlos Ricardo Soccol† †
Federal University of Paraná, Department of Bioprocess Engineering and Biotechnology, P.O. Box 19011, Curitiba, Paraná 81531-990, Brazil
ABSTRACT: Itaconic acid (IA) is an unsaturated diacid, a promising compound that might replace part of the petrochemicalbased monomers, such as acrylic acid, as a building block for polymers. Recent developments in biotechnology allow the efficient production of IA through fermentation processes. However, further enhancements are necessary in the downstream (recovery) of the product. This investigation examined the separation of IA by adsorption from aqueous solutions, using two types of commercial, strongly basic ion-exchange resins: Purolite A-500P and PFA-300. To evaluate the separation process, the following parameters were tested: pH (from 3.03 to 6.33), temperature (from 10 to 50 °C), and IA concentration (from 0.41 to 6.50 g· L−1). The Freundlich and Langmuir isotherms were shown to be good fits to the experimental data, and the adsorption kinetics for IA was found to follow a pseudo-second-order (PSO) model. After batch tests, continuous adsorption experiments were carried out using a fixed bed column, and a simplified mathematical model was developed and evaluated in order to determine the adsorption parameters. The experimental data obtained from the column tests were aligned with those obtained from the isotherms and batch simulations with PSO. The resin PFA-300 proved to be more efficient for IA recovery through adsorption, with a maximum capacity of 0.154 gIA.gresin−1 when compared to the resin A-500P, with a maximum capacity of 0.097 gIA.gresin−1. Both resins have high affinity for the solute, being half-saturated with equilibrium concentrations below 0.25 g·L−1 of acid. has three different states of protonation (H2IA, HIA−, and IA2−), with two dissociation constants in aqueous solution: pKa1 ≈ 3.85 and pKa2 ≈ 5.55.4,7 Therefore, the charge of IA is pH-dependent, an important characteristic to explore in ionexchange. Biotechnological advances have made possible the production of IA through fermentation using Aspergillus terreus, creating a renewable and environmental friendly substitute to petrochemical-based products.4 However, some methods of separation and recovery of organic acids from the fermented broth are inefficient or expensive, increasing the overall cost of production. The development of an economically viable downstream process is paramount to allow the biobased production of an organic acid.8 Crystallization is the usual unit process for recovery of IA, relying on evaporation and cooling of the solution. Other recovery methods from aqueous solutions or fermented broths have been reported, such as
1. INTRODUCTION The recovery and purification of organic acids from aqueous solutions or fermentation broths is an essential step in several biotechnological processes.1 Separation steps have significant environmental and economic impacts in these bioprocesses; their improvement is essential for the reduction of the environmental burden, energy consumption, and waste generation.2 Organic acids produced via bioprocesses, such as citric, lactic, tartaric, gluconic, and itaconic acids have been widely used as intermediates in various branches of industry because they can be easily transformed into a wide range of different substances.3−5 The purification of these acids is important because it affects the quality of final products, which may be used as food additives, or pharmaceuticals and biodegradable plastic compounds.3 Itaconic acid (IA) can be used as a substitute for petroleumbased compounds, such as acrylic or methacrylic acids, because it is equally monounsaturated, and therefore can be polymerized by addition.4,6 Pure IA is a white, crystalline solid, an unsaturated organic diacid with formula C5H6O4 and molar mass of 130.1 g·mol−1. With its two carboxylic groups, IA © XXXX American Chemical Society
Received: July 20, 2015 Accepted: November 9, 2015
A
DOI: 10.1021/acs.jced.5b00620 J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data
Article
liquid−liquid extraction9−12 and membrane separations.13−16 However, besides the work carried out by Gulicovski et al.,17 little is described about IA recovery through adsorption. Synthetic ion-exchange resins have been studied in the separation and purification of other organic acids.1,3,18−20 These resins can strongly adsorb charged species, while the other components of the solution flow through the adsorbent. The adsorbed product can be recovered subsequently by eluting the loaded bed, which allows a reduction in the amount of eluent used and results in a product with high levels of purity. Despite the existing knowledge about separation of organic acids by ionexchange methods, there are details that must be further developed for specific systems, such as the effective solute load, or the kinetics of adsorption. Therefore, the objective of this study was to evaluate the separation of IA from aqueous solutions by using two cationic ion-exchange synthetic resins: Purolite A-500P and PFA-300. These resins were selected because of their high adsorption capacity, in a wide range of pH, and its high regeneration yield. A mathematical model of the adsorption kinetics in a fixed bed column was used to identify the operational parameters for IA recovery.
2.2. Fixed-Bed Column Adsorption Studies. The experiments in fixed bed column were performed to remove IA from aqueous solution with a concentration of 52.0 g·L−1 and an initial pH 3.85. This was made to simulate the values of pH and IA concentration expected in a fermented broth, besides maintaining the pH close to pKa1 (found to be more adequate for adsorption, as explained in the results).22,23 The schematic diagram of the experimental setup is shown in Figure 1. A glass column with an internal diameter of 1.0 cm was used
2. MATERIALS AND METHODS 2.1. Batch Adsorption Studies. The tests were performed in 250 mL Erlenmeyer flasks containing 100 mL of IA solution and 2.00 g of adsorbent. All assays were made in a thermostatic shaker at 28 °C with agitation at 120 rpm. The resins were activated after serial washing with hydrochloric acid (2 N), deionized water, sodium hydroxide (2 N) and a further washing with deionized water. The initial IA solutions (Aldrich Company Co., acid purity ≥99%) were prepared at concentrations of 6.50 g·L−1. The experiments covered two types of strongly basic resins available on the market: Purolite A-500P and PFA-300. Unless otherwise stated, resin masses in the text refer to the wet resin, as it is usually considered in laboratory and industry. The main physical and chemical characteristics of the resins are described in Table 1.
Figure 1. Experimental fixed-bed column adsorption: (1) feed of IA solution; (2) peristaltic pump; (3) fixed-bed column; (4) outlet collection.
as fixed bed column. The adsorbent bed was packed with Purolite PFA-300 and A-500P resins as follows: 10.0 g of the adsorbent (which had not yet been submitted to the washing and activation steps) were carefully poured into each column until all material was packaged between porous frits. Then the column was washed and activated with a sequence of 500 mL of deionized water, 200 mL of HCl (2 N), 500 mL of deionized water, and 200 mL of NaOH (2 N). Samples were taken every 2 min under a flow of 0.825 mL.min−1. 2.3. Adsorption Isotherm Models. For the modeling of the adsorption isotherms, the models of Freundlich and Langmuir were evaluated. The solid-phase concentrations (q) at equilibrium were calculated from the concentration in the liquid, through a material balance:
Table 1. Typical Physical and Chemical Characteristics of the Resins parameters polymer matrix structure physical form and appearance functional groups shipping weight (g·L−1) particle size range (mm) moisture retention (%) total exchange capacity (equiv L−1 min−1)
Purolite A-500P
Purolite PFA-300
macroporous styrenedivinylbenzene opaque nearwhite spheres R-(CH3)3N+ 655−685 0.850−0.600 63−70 0.8
cross-linked gel polystyrene amber spherical beads R-(CH3)2(C2H4OH)N+ 690 0.710−0.425 40−45 1.4
q=
(C 0 − C )ρ V m
(1)
where C0 and C are, respectively, the initial concentration and the equilibrium concentration of IA in the liquid-phase (mM), ρ is the molecular weight (g mol−1), V is the solution volume (L), and m is the mass of wet resin (g). The Langmuir isotherm is one of the standard models to calculate the adsorption equilibrium parameters, and is defined based on the assumption that distribution of pores in the surface of the adsorbent is homogeneous, with negligible interaction forces between adsorbed molecules. The equation for a fixed temperature is given below:24,25
Preliminary analyses demonstrated that the system reached adsorption equilibrium in about 30 min with the proportion tested; therefore, the subsequent tests were done considering 1 h of contact. The IA concentration in equilibrium solutions was analyzed using a UV−vis spectrophotometer, Shimadzu (UV1601PC), by reading the absorbance at a wavelength of 240 nm, with an extinction coefficient of 0.409 g·L−1·cm−1.21 Samples of 0.5 mL each were collected and diluted with 4.5 mL of phosphate buffer solution (100 mM) at pH 7.0. All experiments in this section were performed in triplicate.
q= B
qS·KL ·C 1 + KL ·C
(2) DOI: 10.1021/acs.jced.5b00620 J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data
Article
where qS is the saturation capacity of the resin, that is, the maximum solid-phase concentration of IA in equilibrium, and KL is the Langmuir equilibrium constant, related to the affinity of the solute to adsorption sites.25 Equation 2 can be linearized in the form of eq 3. Using the least-squares method in Microsoft Excel 2013, the values of qS and KL were determined by linear regression of eq 2, in which the slope is 1/qS and the intercept is 1/(KL·qS). C 1 1 = + C q qS·KL qS
(3)
The other isotherm used was the Freundlich model, eq 4:25
q = KF·C1/ n
(4)
where KF and n are temperature-dependent constants for a specific solute and adsorbent. Equation 4 is exponential, and its linearized form is the logarithm of both sides of eq 5: ln q = ln KF +
⎛1⎞ ⎜ ⎟ ln C ⎝n⎠
Figure 2. Scheme of the main stages and directions in the mass transfer of the fixed bed adsorption column: (1) mass transfer in the bulk liquid; (2) diffusion in the liquid film; (3) intraparticle diffusion; (4) adsorption.
(5)
With the use of the linear regression of eq 5, the intercept is ln KF and the slope is 1/n, so the values of KF and n may be calculated. 2.4. Adsorption Kinetics in Batch Processes. The investigation of the adsorption kinetics in batch was done by collecting 0.5 mL samples every 3 min up to 1 h of equilibration time. These experimental data were analyzed using a pseudo second-order (PSO) model. Ho and McKay deduced the simple linear equation of a PSO model for the analysis of adsorption kinetics from liquid solutions:26−28 dq = k 2·(qe − q)2 dt
(6)
where qe is the amount of solute adsorbed at equilibrium (g· g−1) and k2 is the PSO rate constant of sorption (g·g−1· min−1).29 Integrating eq 6, for the initial conditions q(0) = 0, and rearranging to obtain a linear form gives ⎛ ⎞ t 1 ⎟ ⎛⎜ 1 ⎞⎟ = ⎜⎜ + ⎜ ⎟·t 2⎟ q ⎝ k 2·qe ⎠ ⎝ qe ⎠
Figure 3. Mass transfer in accordance with the movement through the adsorption bed.
It is assumed that the column packing is homogeneous. Then, the solute in the bulk phase flows along the column bed according to the following material balance:
(7)
A linear regression of t/q as a function of t will give, comparing with eq 7, a slope of 1/qe and an intercept 1/(k2·qe2), from which the values of qe and k2 may be isolated. 2.5. Mathematical Model of Adsorption on the Fixed Bed Column. A mathematical model for adsorption in the fixed bed ion exchange column was developed and verified experimentally. The model encompasses four transfer stages: mass transfer in the bulk liquid, diffusion in the liquid film, intraparticle transfer, and adsorption.30 Thus, each mass transfer step will be developed separately in a first approach. These steps will be later combined in the transition boundary of each region: bulk liquid, liquid film, and adsorbent particle (Figure 2). The mass transfer in the bulk liquid describes the complete filling of the adsorbate along the entire fixed bed column, as shown in Figure 3. During the process, while the accumulation occurs on the adsorption sites of the resin, both convective motion and dispersion (axial and radial) occur.
Input − Output + Generation − Consumption = Accumulation
(8)
The following assumptions were made for formulating the model equations: 1. chemical reactions do not occur in the column 2. the process is isothermal and isobaric 3. the particles that make up a solid phase of fixed bed are spherical, uniform in size, and homogeneous 4. the dispersion in the radial direction of the bed is negligible 5. the diffusion is based in the Fick’s first law 6. the flow rate is constant and invariant along the column Thus, the mass conservation equation for the solute in the bulk liquid that flows over the bed is used to represent the relationship between corresponding changes in eq 9: C
DOI: 10.1021/acs.jced.5b00620 J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data A ·[NIA|z − NIA|z +Δz ] = A ·Δz ·
∂C IA ∂t
Article
Table 2. Effect of Initial pH on the Adsorption of IA onto Ion-Exchange Resinsa
(9)
where A is the flow area, NIA is the mass flow of IA, z is distance to the inlet, CIA is the concentration of the IA in the bed, and t is time. Equation 9 can be rearranged as follows:25,31 −
∂NIA ∂C IA = ∂z ∂t
(10)
The mass flow can be divided in two components, diffusion and convection:
NA = JIAz + C IA ·vz
(11)
where JIAz is the diffusive flow (−Dz·∇CIA), Dz is the diffusivity bed constant, ∇CIA is the concentration by cylindrical coordinates C(r, θ, z), and vz is the linear bed velocity.25,31 Combining eq 10 and eq 11, with negligible radial diffusivity, leads to the final relation: ∂C IA ∂ 2C IA ∂C = DZ · − vz IA 2 ∂t ∂z ∂z
This simplified mathematical model describes the concentration of IA along the column length, the z axis, according to the change in time. The initial and boundary conditions are
(13)
⎧ ⎪ t = 0 → C(0, 0) = 0 z = 0⎨ ⎪ ⎩ t > 0 → C(0, t ) = C0
(14)
∂C =0 ∂z
(15)
z=H→
T (°C)
mresin (g)
3.03 3.85 4.68 5.50 6.33
28 28 28 28 28
2.00 2.00 2.00 2.00 2.00
3.03 3.85 4.68 5.50 6.33
28 28 28 28 28
2.00 2.00 2.00 2.00 2.00
C0 (g·L−1) A-500P 6.50 6.50 6.50 6.50 6.50 PFA-300 6.50 6.50 6.50 6.50 6.50
Ce (g·L−1)
q (g·g−1)
4.72 4.45 4.72 4.96 5.06
0.089 0.103 0.089 0.077 0.072
3.28 3.53 4.01 4.50 4.57
0.161 0.149 0.125 0.101 0.097
a (pH) Initial pH of IA solution; (T) adsorption temperature; (mresin) adsorbent mass; (C0) initial concentration of IA solution; (Ce) equilibrium concentration; (q) adsorbed adsorbate. Standard uncertainties (u) are u(pH) = 0.005, u(T) = 0.05 °C, u(m) = 0.001 g, and the combined expanded uncertainties (Uc) are Uc(C0) = 0.005 g·L−1, Uc (Ce) = 0.16 g·L−1, and Uc(q) = 0.008 g.g−1 (level of confidence = 0.95).
(12)
⎧ ⎪C(z , 0) = 0 t = 0⎨ ⎪ ⎩ q(z , 0) = 0
pH
deprotonating, A-500P resin has a higher adsorption capacity. In the case of PFA-300 resin, the adsorption capacity decreases with increasing initial pH. When the initial pH exceeds the pKa2, the ability of the resin to adsorb IA decreases and tends to a lower limit as shown in Figure 4. This may be due to
The partial differential eq 12 are solved numerically by reducing them to a set of nonlinear algebraic equations using an explicit finite difference technique. A mathematical algorithm to solve these equations was developed and implemented in a computer program using the MATLAB software.
3. RESULTS AND DISCUSSION 3.1. Batch Adsorption Equilibrium. The pH and temperature of the system may affect the behavior of both the adsorbent and the solute. In the case of strong anion exchangers, the influence of pH is apparent only at extreme pH values.32 However, for ionizable solutes such as IA, pH affects the species distribution (H2IA, HIA−, and IA2−). Therefore, pH affects the adsorption, because it determines the solute charge and the density of charge on the surface of the adsorbent.32,33 Temperature has a more complex effect, because it alters both equilibrium and kinetics. A rise in temperature changes the chemical potential and the adsorbent density, affecting the affinity of the anions by the adsorbent, decreasing the adsorption capacity. 1,3 Also, the range of temperature recommended for the resins is narrow (typically ambient). Adsorption of IA from aqueous solutions was evaluated at five initial pH values: 3.03, 3.85, 4.68, 5.55, and 6.33, all with 6.50 g·L−1 of IA, in order to evaluate the effect of initial pH on adsorption. It can be observed in Table 2 that when the initial pH is near the value of the first IA dissociation constant, pKa1, that is, when only one carboxyl group of the IA molecule is
Figure 4. Effect of the initial pH on the adsorption of IA onto ionexchange resins, pH range = 3.03−6.33, temperature = 28 °C, solid− liquid ratio = 20 g·L−1, C0 = 6.50 g·L−1, time = 60 min: (●) A-500P; (■) PFA-300.
deprotonation of the two carboxylic groups of the acid, when the same IA molecule potentially binds to more than one of the active sites of the resin.33 These experimental data are in accordance with Gulicovski et al.,17 who concluded that the IA adsorption onto alumina surface is extremely pH dependent and the maximum adsorption occurs at a pH near the value of pKa1. The results of IA adsorption on the resins at different temperatures, 10, 20, 30, 40, and 50 °C, are shown in Table 3 and Figure 5. The IA solution was prepared with an initial pH of 3.85. The results show that the temperature has not a significant impact in adsorption using the PFA-300 resin, in the range of 10−50 °C. In the case of the A-500P resin, the adsorption capacity of the resin slightly decreases with temperature increase. This behavior shows that the reaction is exothermic, as is expected for adsorptions.1 D
DOI: 10.1021/acs.jced.5b00620 J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data
Article
Table 3. Effect of Temperature on the Adsorption of IA onto Ion-Exchange Resinsa pH
T (°C)
mresin (g)
3.85 3.85 3.85 3.85 3.85
10 20 30 40 50
2.00 2.00 2.00 2.00 2.00
3.85 3.85 3.85 3.85 3.85
10 20 30 40 50
2.00 2.00 2.00 2.00 2.00
C0 (g·L−1) A-500P 6.50 6.50 6.50 6.50 6.50 PFA-300 6.50 6.50 6.50 6.50 6.50
Table 4. Effect of Initial Concentration of Acid on the Adsorption of IA onto Ion-Exchange Resinsa
Ce (g·L−1)
q (g·g−1)
pH
T (°C)
mresin (g)
4.52 4.55 4.56 4.58 4.67
0.099 0.098 0.097 0.096 0.092
3.85 3.85 3.85 3.85 3.85
28 28 28 28 28
2.00 2.00 2.00 2.00 2.00
3.55 3.50 3.59 3.62 3.66
0.148 0.150 0.146 0.144 0.142
3.85 3.85 3.85 3.85 3.85
28 28 28 28 28
2.00 2.00 2.00 2.00 2.00
a
C0 (g·L−1) A-500P 0.41 0.81 1.63 3.25 6.50 PFA-300 0.41 0.81 1.63 3.25 6.50
Ce (g·L−1)
q (g·g−1)
0.01 0.05 0.53 1.79 4.60
0.020 0.038 0.055 0.073 0.095
0.02 0.02 0.10 0.96 3.52
0.019 0.040 0.076 0.115 0.149
a (pH) Initial pH of IA solution; (T) adsorption temperature; (mresin) adsorbent mass; (C0) initial concentration of IA solution; (Ce) equilibrium concentration; (q) adsorbed adsorbate. Standard uncertainties (u) are u(pH) = 0.005, u(T) = 0.05 °C, u(m) = 0.001 g, and the combined expanded uncertainties (Uc) are Uc(C0) = 0.005 g·L−1, Uc(Ce) = 0.03 g·L−1, and Uc(q) = 0.002 g·g−1 (level of confidence = 0.95).
(pH) Initial pH of IA solution; (T) adsorption temperature; (mresin) adsorbent mass; (C0) initial concentration of IA solution; (Ce) equilibrium concentration; (q) adsorbed adsorbate. Standard uncertainties (u) are u(pH) = 0.005, u(T) = 0.05 °C, u(m) = 0.001 g, and the combined expanded uncertainties (Uc) are Uc(C0) = 0.005 g·L−1, Uc (Ce) = 0.1 g·L−1, and Uc(q) = 0.004 g.g−1 (level of confidence = 0.95).
Figure 5. Effect of temperature on the adsorption of IA onto ionexchange resins, temperature range = 10−50 °C, pH = 3.85, solid− liquid ratio = 20 g·L−1, C0 = 6.50 g·L−1, time = 60 min: (●) A-500P; (■) PFA-300.
Figure 6. Langmuir isotherm for the adsorption of IA onto ionexchange resins, C0 range = 0.41−6.50 g·L−1, temperature = 28 °C, pH = 3.85, solid−liquid ratio = 20 g·L−1, time = 60 min: (●) A-500P; (■) PFA-300; (− − − ) Langmuir isotherm of A-500P; (---) Langmuir isotherm of PFA-300.
The effect of initial acid concentration on adsorption onto resins was evaluated at five different initial IA concentrations, 0.41, 0.81, 1.63, 3.25, and 6.50 g·L−1. It was observed in Table 4, that when the initial acid concentration is raised the equilibrium concentration increases by an ever smaller extent. This was expected, and is due to the saturation of the ionexchange sites of the resins, preventing more binding from occurring between the free acid and the adsorbent. Figure 6 shows that the equilibrium concentrations increased from 0.02 g·g−1 to 0.10 g·g−1 for A-500P and to 0.15 g·g−1 for PFA-300. Table 5 shows the effect of contact time on the adsorption of IA for each resin, studied over a period of 30 min. The adsorption rate was higher in the early stages when the acid contacts the adsorbent, and subsequently drops when the system approaches equilibrium. This is expected because a larger number of sites are available for adsorption in the surface of the resin at the beginning of the process, and the higher solute concentration in aqueous phase favors the mass transfer.1,34 3.2. Adsorption Isotherms. The results demonstrate that the Langmuir isotherm explains the experimental data especially at low concentrations of IA (
