Removal of succinic acid from fermentation broth by multistage process (membrane separation and reactive extraction)

Removal of succinic acid from fermentation broth by multistage process (membrane separation and reactive extraction)

Separation and Purification Technology 192 (2018) 360–368 Contents lists available at ScienceDirect Separation and Purification Technology journal ho...

846KB Sizes 3 Downloads 524 Views

Separation and Purification Technology 192 (2018) 360–368

Contents lists available at ScienceDirect

Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur

Removal of succinic acid from fermentation broth by multistage process (membrane separation and reactive extraction)

MARK



K. Prochaska , J. Antczak, M. Regel-Rosocka, M. Szczygiełda Institute of Chemical Technology and Engineering, Poznan University of Technology, Berdychowo Str. 4, 60-965 Poznań, Poland

A R T I C L E I N F O

A B S T R A C T

Keywords: Succinic acid Fermentation broth Ultrafiltration Bipolar membrane electrodialysis Reactive extraction

An environmentally friendly process for separation of succinic acid from the model solutions as well as the actual post-fermentation broth left after bioconversion of raw glycerol (which is generated in a large amount as a byproduct during the production of biodiesel) was studied. An integrated system to realize this process was proposed which consisted of: ultrafiltration (UF), bipolar membrane electrodialysis (EDBM) and 3-step reactive extraction (RE) with commercial solvating extractants. Pre-clarification process carried out by UF allowed the removal of high molecular contaminants present in the feed solution, such as: biomass, proteins as well as bacteria cells. Significant reduction in permeate flux during the process was observed due to fouling of ceramic UF membranes. However, the fouling layer was effectively removed by applying hydraulic and chemical cleaning baths. Application of the EDBM process in the proposed integrated system allowed elimination of acidification of broth which usually generates a considerable amount of wastes. The succinic acid, present in the aqueous stream after EDBM was removed in a three-step reactive extraction at more than 90% efficiency. Extraction can support membrane techniques to separate efficiently carboxylic acids from the post-fermentation broth, however, it is not selective enough to separate succinic acid from other acids present in the broths. The only way to reach selective extraction of succinic acid over acetic, lactic acids and glycerol is to decrease pH to 2 and use Cyanex 923 as an extractant.

1. Introduction Biodiesel is mainly produced from vegetable oils with an addition of methanol and in the presence of an alkaline catalyst in a biotechnological process, in which the most important step is the transesterification of triglycerides present in vegetable oils. During the production of biodiesel, a large amount of raw glycerol is generated as a by-product (up to 10 wt% of fuel production). One of the ways of disposal of the excessive unpurified glycerol is through the fermentation process. Depending on the bacterial strains used for bioconversion, post-fermentation solutions with different, but always very complex composition are obtained. They contain besides the main product, a number of other compounds, e.g. mono- and multivalent carboxylic acids, low molecular weight neutral organic compounds (unreacted glycerol, lactose) and a large amount of inorganic salts. The use of impure glycerol phase as the main carbonate source for biotechnological synthesis of different organic compounds, has been proposed in several works [1,2]. The fermentation process generates a broth containing the dissociated and non -dissociated forms of metabolites and residual mineral salts.



Various unit operations are required for separation, concentration and purification of organic compounds from the fermentation broth. The traditional method for isolating carboxylic acid from fermentation broths is precipitation of the acid salts with Ca(OH)2 or Ca(CO)3, and then acid recovery with sulfuric acid. However, large amount of calcium sulfate formed within this process, which should be managed, outbalances the benefits of this process. Carboxylic acids can be separated also by means of porous solid sorbents characterized with large porous surfaces, such as activated carbons or various resins. However, a drawback of this technique is the limited sorption capacity. Recently also some membrane techniques have been proposed for separation of carboxylic acids, i.e. electrodialysis, ultrafiltration, nanofiltration and reverse osmosis [3,4]. Other techniques proposed for carboxylic acid separation from aqueous solutions are bipolar membrane electrodialysis or liquid-liquid extraction, called also reactive extraction. The ultrafiltration is a membrane separation technique in which the driving force of the process is the pressure difference across the membrane. The transmembrane pressure range used in the ultrafiltration processes is between 0.3 and 1 MPa. The transport of molecules through

Corresponding author. E-mail address: [email protected] (K. Prochaska).

http://dx.doi.org/10.1016/j.seppur.2017.10.043 Received 23 July 2017; Received in revised form 19 October 2017; Accepted 19 October 2017 Available online 21 October 2017 1383-5866/ © 2017 Elsevier B.V. All rights reserved.

Separation and Purification Technology 192 (2018) 360–368

K. Prochaska et al.

succinic [21]. Such acids as lactic, propionic, succinic, maleic, butyric, oxalic or acetic can be efficiently transported from the aqueous to the organic phase by organophosphorus solvating extractants [22–24], aliphatic amines [22,25–28], quaternary ammonium [22,29] or phosphonium salts [30–32]. Also an integrated system of nanofiltration, bipolar membrane electrodialysis and reactive extraction has been recently proposed by our team to successfully remove fumaric acid from the fermentation broth [33]. In this investigation, an integrated system consisting of three stages: ultrafiltration (UF), bipolar membrane electrodialysis (EDBM) and reactive extraction (RE) was proposed for separation of succinic acid from the fermentation broth left after bioconversion of glycerol.

the ultrafiltration membrane is based on the sieve mechanism and the size of retained particles defines the (cutoff) limit molecular weight membrane. The ultrafiltration processes applied for clarification of continental water to get drinking water or clarification of industrial wastewater, as well as in dairy industries such for milk dehydration and whey concentration. The pore size of ultrafiltration membranes disqualifies this process for direct separation of low molecular weight organic compounds. However, many literature reports suggest the advantages of applying the ultrafiltration as a pre-treatment step of postfermentation solutions for removal of suspended solids, turbidity and large microorganisms, as well as dissolved macromolecules, colloids and small bacteria [5]. Juang et al. [6] have used UF to remove the residues of biological material from fermentation broth of Serratia marcescens SMDR, while Cho et al. [7] have applied the ultrafiltration as the first step of extracting organic acids (acetic and butyric) to remove micro-organisms and macromolecular compounds from the solution after bioconversion of woodchips. The ultrafiltration process of postfermentation broth obtained during the production of succinic acid from biomass has also been reported [8,9]. One of the most important issues concerning all pressure-driven membrane techniques applied in industrial scale is membrane fouling phenomenon, which is mainly caused by the deposition of organic and inorganic substances on the membrane surface and in its pores. A significant reduction in the permeate flux observed during UF process (caused by deposition of particles and molecules on membrane) is one of the major factors limiting the wider application of ultrafiltration process in industry [10]. The formation of biomass cake layer on the membrane surface in time of pre-treatment of post-fermentation broths could have negative influence not only on the filtration yield, but also on the retention of components present in the working solutions. The way of formation of the fouling layer on membrane surface depends on a number of factors, e.g. the structure of retained particles and molecules (the size, shape, affinity for the membrane surface) as well as the morphology of membrane surface (porosity) and the nature of the membrane material (hydrophilicity) or processing conditions (TMP, T, Qf) of ultrafiltration process [11,12]. Due to the presence of a significant number of byproducts in the actual post-fermentation solutions, such residues as the unreacted glycerol, lactic, formic or acetic acids, magnesium(II) and calcium(II) inorganic salts, and a large amount of lactose or ethanol, it is problematic (or sometimes even impossible) to isolate final product with high purity and good yield using a one-step separation technique. Therefore, the integrated separation processes including pre-treatment, organic acid separation from fermentation broth and conversion of carboxylic salts to organic acid, has attracted lots of attention [13,14]. Electrodialysis process is an example of membrane separation techniques in which the transfer of ions through the ion-selective membranes followed under the influence of an applied electric potential difference. The process uses an electrical driving force to transfer ions from the diluate chamber through the cathode (positively charged ions) and anode (negatively charged ions) to a concentrate chamber, creating a more concentrated stream. The bipolar membranes (BP) consist of two membranes: the anion exchange and cation exchange, separated by a thin water layer of a thickness of about 2 nm [15]. Under the influence of a direct current (DC) voltage, in the catalytic space of the bipolar membrane the water molecules are split to hydrogen and hydroxyl ions [16]. Although the EDBM process has successfully been applied in various areas of technology, e.g. production of food or production of useful chemical compounds from industrial wastewater, such as ammonia [17] or the desalination of industrial saline stream [18], the number of applications of EDBM is continuously increasing and nowadays it is more often applied for environment friendly technologies, e.g. the production of organic acids from the solutions after bioconversion processes [19]. Several studies have indicated that EDBM is available process for producing organic acids, such as acetic, oxalic and citric [16] lactic acid [20] or

2. Material and methods 2.1. Materials For the investigation of separation and concentration of succinic acid from model solutions five organic compounds were used: succinic acid (Suc) (Sigma-Aldrich, Poland), glycerol (Glyc) (POCH S.A, Poland), formic acid (Form) (POCH S.A, Poland), acetic acid (Ac) (POCH S.A, Poland), lactic acid (Lact) (Sigma-Aldrich, Poland). Two synthetic water solutions were subjected to separation by bipolar membrane electrodialysis: (1) 2-component solution, with succinic acid and glycerol and (2) 5-component solution, containing glycerol and 4 carboxylic acids: succinic, formic, acetic and lactic (the composition of 5component model solution was similar to the permeate solution after ultrafiltration of actual post-fermentation broth). Model solutions were prepared by dilution with deionized water of a conductivity not-exceeding 3 μS/cm. The pH of these solutions was adjusted to 8.5 by adding an appropriate amount of sodium hydroxide (Sigma-Aldrich, Poland). The actual post-fermentation broth (pH = 8.5) left after bioconversion of glycerol to succinic acid (preliminary centrifugated to remove biomass) was delivered from the Poznan University of Life Sciences. The composition of the model solutions and the actual postfermentation broth are shown in Table 1. Solvating (mixture of trialkylphosphine oxides - Cyanex 923) or basic (trioctylamine - TOA) extractants were used as received to prepare 0.1 or 0.4 mol/dm3 organic solutions of extractant in octanol or in low-aromatic kerosene Exxsol D 220/230 (aromatic content: 0.05 wt%, distillation range 222–234 °C, Exxon Mobil Chemical, Germany). 2.2. Ultrafiltration (UF) UF process was carried out on a laboratory ultrafiltration unit, which was equipped with two (left – lm and right – rm) commercially available tubular ceramic membranes: Céram INSIDE® (TAMI, France). Technical parameters of the membranes applied in this study are given in Table 2. The volume of the feed tank amounted to 10 dm3. Both membranes before UF process were conditioned until a constant water flux was established. The flux of deionized water is called the “initial water flux” (Jilm/Jirm). Pre-clarification of the post-fermentation solution by ultrafiltration was carried out under the following Table 1 Composition of the model solutions and the actual post-fermentation broth after bioconversion of glycerol to succinic acid. Solution

2-component model solution 5-component model solution/actual postfermentation broth

361

Concentration, C0, g/dm3 Suc

Glyc

Form

Ac

Lact

23.3 23.3

15.2 15.2

– 9.9

– 8.4

– 6.4

Separation and Purification Technology 192 (2018) 360–368

K. Prochaska et al.

The average value of the current efficiency was calculated on the basis of the following equation:

Table 2 Characteristics of applied ceramic ultrafiltration membranes. Parameter

Ultrafiltration membrane

Configuration Cut-off Membrane material Effective surface area Number of channels Internal channel diameter External membrane diameter Membrane length

Tubular 15 kDa Titanium oxide/zirconium oxide 0.0042 m2 3 6 mm 10 mm 250 mm

Initial water flux at 0.4 MPa on: (1) left membrane (2) right membrane

542.9 dm3/m2 h 250.9 dm3/m2 h

Hydrodynamic coefficient for: (1) left membrane (2) right membrane

1357.3 dm3/m2 h MPa 626,5 dm3/m2 h MPa

CE =

EC =

U ·I ·t m

(3)

where: EC – energy consumption needed to produce 1 kg of Suc, kWh/ kg; U – voltage, V; I – current, A; m – mass of the final product, g; t – time, h. 2.4. Reactive extraction (RE) Reactive extraction of the solutions studied was carried out in a typical way: the aqueous feed was mechanically shaken with the same volume of the organic phase for 30 min at 25 °C in glass separatory funnels, and then allowed to stand for phase separation. The influence of pH on succinic acid extraction with various solvating extractant solutions was investigated. A solution of 0.2 mol/dm3 NaOH was used as a stripping phase. Reactive extraction of the solutions resulting from fermentation broth processing, i.e. bipolar membrane electrodialysis (diluate, concentrate) was carried out in three steps with 0.4 mol/dm3 TOA or Cyanex 923 as extractants in Exxsol D 220/230, at w/o = 1. The concurrent extraction was realized by contacting in each step the aqueous raffinate with a fresh organic phase. The extraction efficiency (percentage extraction) was calculated from the equation:

• cleaning with water, t = 5 min, T = 50 °C; • cleaning with 5% sodium hydroxide solution, t = 60 min, T = 60 °C; • cleaning with water, t = 5 min, T = 50 °C; • cleaning with 3% nitric acid solution, t = 10 min, T = 60 °C; • cleaning with water to pH ≈ 5–6. 2.3. Bipolar membrane electrodialysis (EDBM) A 10-chamber laboratory EDBM setup with a stack consisting of 10 bipolar (PC 200bip), 10 anion-exchange (PC 200D) and 1 cation-exchange (PC-SK) membranes produced by (PCCell GmbH, Germany) separated by 0.5 mm spacers, was used in this study. The effective surface area of each membrane was equal to 207 cm2. The electrodialysis stack was connected with a flow pump (Verder, Poland), DC power supply (NDN) and a multifunction meter (Elmetron, Poland) measuring the pH, the conductivity and the temperature of working solutions. The anode was made of titanium plated with iridium and the cathode was made of steel 316. In each experiment the model solution or UF permeate of post-fermentation broth was fed into the diluate compartment while in the concentrate compartment the sodium succinate acid solution (with the concentration of succinate ions corresponding to the initial feed solution) was used to decrease the cell voltage at the beginning of the experiment. Two surface platinized electrodes, made of titanium coated with ruthenium, terminated the setup with sodium sulfate (0.25 M) as a rinsing solution. Flow rate of the feed solutions, circulated through working compartments was set on 100 dm3/h. Bipolar membrane electrodialysis process was performed at 25 ± 2 °C under constant electric field conditions at 120 and 90 A/m2 of current density. EDBM experiments were conducted for 180 min. The desalination degree of succinic acid obtained during bipolar membrane electrodialysis was calculated as follows: t Cdil ·100% 0 Cdil

(2)

where: CE – current efficiency, %; F – Faraday’s constant (96485), C/ mol; I – current, A; z – valence of ions; V – diluate volume, dm3; ΔCdil – change of Suc concentration in diluate chamber, mol/dm3; n – number of cells; Δt – time, s; The average value of energy consumed for 1 kg of succinic acid production was determined using the equation:

conditions: TMP = 0.4 MPa, T = 25 ± 2 oC, Qf = 300 dm3/h. During the ultrafiltration process 10 dm3 of the solution from the feed vessel was pumped through the UF membrane modules, and during the whole process the retentate was circulated in a closed loop with a volume flow rate of 300 dm3/h. Each separation process was performed for 3 h. After the ultrafiltration process of post-fermentation broth, the ultrafiltration unit was cleaned according to the following procedure:

ηdes = 1−

F ·z·V ·ΔCdil ∗100% n·I ·Δt

EE =

C 0−C ·100% C0

(4)

where: EE – extraction efficiency, %; C – initial concentration of the acid in the aqueous phases before extraction, g/dm3; C – concentration of the acid in the aqueous phases after extraction, g/dm3. 0

2.5. Analytical methods The contents of mono- and dicarboxylic acids and their salts in the starting solution and all fractions obtained during separation of succinic acid processes were determined using a high performance liquid chromatography HP Agilent 1100 Series (Germany), equipped with an autosampler, interface (HP 35900), RI Detector (HP 1047A), pump (HP1050), and separating column Rezex ROA-Organic Acid H+ (8%), Phenomenex®. The eluent of 2.5 mM H2SO4 solution was supplied continuously at the rate of 0.9 cm3/min. The column temperature and that at the input to the detector was 40 °C, P = 0.56 MPa. All samples were acidified to pH ≤ 2 by addition of 0.1 cm3 25% H2SO4 to 1 cm3 of sample before analysis. 3. Results and discussion The integrated system consisting of three stages: (i) ultrafiltration (UF), (ii) bipolar membrane electrodialysis (EDBM) and (iii) reactive extraction (RE) was proposed for separation and purification of succinic acid from the post-fermentation broth left after bioconversion of glycerol (Fig. 1). As the first stage the UF process was performed as a pre-treatment to remove biological material residues after bioconversion from the postfermentation broth. At the second stage, EDBM of the clarified

(1)

where: ηdes – desalination degree, %; Ctdil – concentration of succinic acid in diluate chamber after time t, g/dm3; C0dil – initial concentration of succinic acid in diluate chamber, g/dm3; 362

Separation and Purification Technology 192 (2018) 360–368

K. Prochaska et al.

Fig. 1. The integrated system for separation of succinic acid from fermentation broth.

flux during ultrafiltration of succinic acid post-fermentation broth using a ceramic membrane RC with MWCO 10 kDa. After collecting 160 cm3 of the permeate, approximately 55% reduction in permeate flux in reference to the initial water flux was observed. Wang et al. have concluded that the decrease in the value of permeate flux during filtration results from the membrane fouling, and is influenced by the feed solution, as well as such properties of the membrane as hydrophobicity, morphology, thickness or pore size and their distribution. After the ultrafiltration process of post-fermentation broth resulting from bioconversion of glycerol to succinic acid (after a preliminary centrifugation of the biomass), the ultrafiltration membrane modules were washed in the first step of cleaning operation with deionized water. This operation rinsed off the materials which were not chemically adsorbed on the membrane surface and in the pores. In the next “chemical cleaning” stage, 5% sodium hydroxide solution and 3% nitric acid solutions were used. Finally, the cleaning was finished by washing with deionized water until pH 5 ÷ 6 was obtained. The results presented in Fig. 2 indicate that the initial water flux (obtained before UF process of post fermentation broth) and the flux obtained after three step cleaning procedure finally reached nearly the same value. The obtained results are similar to those presented by Waszak et al. [34] on UF (conducted in the time range from 12 to 37.5 h) of post-fermentation broth in a membrane bioreactor using a ceramic tubular membrane with MWCO 8 kDa. After 20 min of broth filtration, approximately 75% reduction of permeate flux in reference to the initial water flux was noted. It is worth noting that the application of hydraulic (water) and chemical (sodium hydroxide) bath for cleaning of the membranes after UF of broth, makes it possible to restore the initial values of the hydrodynamic properties of the membrane. Taking into account the

permeate was carried out after UF, and then RE of both the concentrate and diluate solutions obtained in bipolar membrane electrodialysis was used to recover the residues of succinic acid. 3.1. Ultrafiltration The process of ultrafiltration of the actual post-fermentation broth left after bioconversion of glycerol was performed in order to remove high molecular contaminants present in the feed solution such as: biomass, proteins or bacteria cells. However, prior to the ultrafiltration of the broth the studies on the influence of TMP on the magnitude of initial water flux were carried out. These experiments confirmed that the yield was proportional to the pressure difference applied for both UF ceramic membranes (data not shown). The experimental data, which are presented in Fig. 2, illustrate the fluxes obtained during the UF process in a unit with two (a. left and b. right) membrane modules (the initial water flux(Ji), the flux of postfermentation broth during UF (Jb), water flux after UF (Jw) and water flux after chemical cleaning of the membrane (Jc)). The initial water fluxes amount to 542.9 dm3/m2·h and 250.9 dm3/m2·h, for the left and the right membrane, respectively. After 15 min of UF process (Fig. 2a) the permeate flux of post-fermentation broth decreased to 195.7 dm3/ m2 h, due to the fouling of the membrane surface and its pores. After about 60 min of UF, permeate flux stabilized and finally after 180 min of the process, it reached the value of 82.9 dm3/m2 h, which makes 15% of the initial water flux. The fouling within the membrane pores can cause changes in the apparent pore size, pore size distribution, and pore density of the membrane which reduce the permeate flux [10]. Similarly, Wang et al. [8] indicate a significant reduction in permeate

Fig. 2. The change in fluxes: water flux before UF, post-fermentation broth flux during UF, water flux after UF and water flux after membrane chemical cleaning; TMP = 0.4 MPa.

363

Separation and Purification Technology 192 (2018) 360–368

K. Prochaska et al.

Fig. 3. Change in concentrations of the organic compounds in the EDBM process at different current density: (a) and (b) 2-component solution, (c) and (d) 5-component solution; i = 120 and 90 A/m2, T = 25 ± 2 °C.

Similarly, greater changes in the concentrations of the monovalent carboxylic acids in the concentration chamber have been observed to correspond to higher current density. The transport of carboxylic acids through the anion-exchange membrane depends on the geometry and the structure of these compounds [33]. Formic, acetic and lactic acids (which are monoprotic acids) reveal higher mobility in solution and are more easily transported through the exchange-membrane than succinic acid (diprotic acid), due to smaller diameter of their molecules. Wang et al. [16] has studied the application of EDBM in production of monoprotic, diprotic and triprotic organic acids, and they have observed that the concentration of produced acids in the reaction time increased in the following order: monoprotic (acetic) acid > diprotic (oxalic) acid > triprotic (citric) acid. In addition, an insignificant diffusive transport of glycerol (which in feed solutions is present as undissociated compound and cannot migrate through the membrane under influence of current) was noted, which was probably a consequence of its high concentration in the initial solution [37]. The number of components present in the working solution has also significant influence on both the migration of ions and conversion of succinates to succinic acid. From Fig. 2 it can be seen, that after 180 min of EDBM, growth in the concentration of succinic acid decreased from 16.9 to 15.7 g/dm3 (current density 120 A/m2) and from 15.0 to 13.0 g/dm3 (current density 90 A/m2) for 2- and 5-component solution, respectively. Han et al. [38] have observed a decrease in mass transfer through the anion-exchange membrane for all types of ions with an increase in the number of components in the separated solution, independently of other conditions of the process. Three of the most fundamental parameters describing electromembrane processes, i.e. the desalination degree, current efficiency and energy consumption [39] commonly used to assess both the yield and efficiency as well as the economic aspects of the bipolar membrane electrodialysis, have been calculated from the experimental data of the EDBM process of 2- and 5-component model solutions. The results

results obtained during our research as well as many literature reports [35,36] one can conclude that the pre-treatment of the actual postfermentation broth using UF process is necessary for the next steps in the proposed integrated system: UF-EDBM-RE.UF is an effective operation that permits removal of the residue biological material and macromolecular compounds after the glycerol fermentation process, as well as leads to a significant reduction in solution turbidity. 3.2. EDBM of model solutions and actual post-fermentation broth Several bipolar electrodialysis processes of 2- and 5-component model solutions (pH 8.5) were investigated. The permeate obtained in the ultrafiltration pre-treatment of the post-fermentation broth (pH 8.5) was also subjected to the EDBM process. All these processes were conducted under constant values of the current densities of 120 and 90 A/m2 in operating time of 180 min. 3.2.1. Concentration of succinic acid Fig. 3. presents the variations in the concentration of succinate ions in the concentrate chamber vs. the operating time of EDBM process of 2- and 5-component model solutions. The increase in concentration of succinic acid in the concentrate compartment during the process increases with the higher value of current density (120 A/m2) and equals approximately 16.9 and 15.7 g/ dm3 for 2- and 5 component (Fig. 3a and 3c) solutions, respectively. This effect should be attributed to the fact that the higher the current density, the greater the number of succinate ions is transported to the concentrate compartment through the anion-exchange membrane and simultaneously, more water molecules are dissociated in the intermediate layer of the bipolar membrane. Similar effects have been observed by Fu et al. [21] during a study on the use of bipolar membrane electrodialysis process for production of succinic acid from sodium succinate under current density values of 125, 250 and 375 A/m2. 364

Separation and Purification Technology 192 (2018) 360–368

K. Prochaska et al.

factors: OH− ions competition (their number increased at higher values of current density and also because OH− ions reveal much higher affinity to the anion exchange membrane than the carboxylic ions) and diffusion of acids molecules. The results presented in Table 3 also indicate that the current efficiency decreased in the operation time, independently of the other process conditions. For example, during the EDBM process of 2-component solution at the current density 90 A/m2 the current efficiency decreased from 47.7 to 19.0% after 30 and 180 min, respectively. The observed reduction in the current efficiency in the operation time was related to the fewer number of ions transported through the membrane during the process. Similar effects have been observed by Fu et al. [21], during the EDBM production of succinic acid at 125 A/m2, the current efficiency was inversely proportional to the running time. The current efficiency slightly dropped from 96.5 to 92.8% (BP-A-C-BP configuration), from 96.8 to 88.7% (BP-C-BP configuration) and from 90.0 to 74.4% (BP-A-BP configuration) respectively, after 30 and 180 min. The composition of the separated solution also affects the current efficiency of EDBM process. Increase in the number of components from 2 to 5 causes a decrease in the current efficiency from 19.0 to 16.5% after 180 min of EDBM processes at current density of 90 A/m2.

Table 3 Variations in the degree of desalination, current efficiency and energy consumption during the time of EDBM process of 2- and 5-component model solutions at current density 120 and 90 A/m2, T = 25 ± 2 °C. i, A/m2

t, min

ηdes, %

CE, %

EC, kWh/kg

2-Comp. solution

5-Comp. solution

2-Comp. solution

5-Comp. solution

2-Comp. solution

5-Comp. solution

120

30 60 120 180

22.7 41.2 56.2 72.4

19.7 36.9 51.1 67.4

28.9 26.1 17.8 15.3

25.1 23.4 16.2 14.3

3.2 3.5 5.0 5.8

3.7 3.9 5.5 6.3

90

30 60 120 180

26.6 46.4 59.7 64.4

18.0 40.8 51.1 55.8

47.7 41.2 26.5 19.0

32.5 36.3 22.7 16.5

1.7 2.0 3.1 4.5

2.6 2.3 3.6 5.2

concerning all EDBM processes of synthetic solutions are summarized in Table 3. 3.2.2. Current efficiency The current efficiency after 180 min of EDBM under current density of 120 A/m2 reached 15.3 and 14.3% for 2-and 5-component solutions, respectively, and increased to 19.0 and 16.5% under the current density of 90 A/m2 (Table 3). It can be concluded, that the current efficiency of EDBM is directly related to: (i) the gradient of transported ions, (ii) the current density, (iii) the operation time and (iv) the number of components present in the initial solution. On the one hand, the higher value of current density in the process causes higher migration of succinate ions from the diluate to the concentrate compartment, which reaches 15.7 and 13.0 g/dm3, respectively, for 120 and 90 A/m2, after 180 min of EDBM of 5-component model solution. However, the rise in the content of succinic acid in the concentrate compartment is probably not enough to compensate greater power used in the process. On the other hand, higher value of the current density is associated not only with the rise in the migration of succinate ions, but also with the higher production of hydroxyl ions via splitting of water molecules by the bipolar membranes. The increasing number of hydroxyl ions in the diluate compartment may result in greater competitiveness for the passage of succinate ions through the membrane. Igliński et al. [40] who studied the production of citric acid using the EDBM process in the range of current density from 520 to 1040 A/m2, have reported that for all performed EDBM processes the increase in the current density has a negative influence on the process efficiency. Wang et al. [16] also have proved that during the production of acetic, oxalic and citric acids by EDBM, the current efficiency decreases with increasing current density. The latter authors suppose that this effect was caused by two main

3.2.3. Energy consumption The economic aspects of the EDBM process was evaluated by the energy consumption according to the current density and the operation time. As shown in Table 3, the energy consumption during EDBM of 5component solutions increases with both high value of the current density (e.g. from 5.2 to 6.3 kWh/kg for 90 and 120 A/m2, respectively), the time of operation (e.g. from 3.7 to 6.3 kWh/kg for 30 and 180 min, respectively, 120 A/m2). There is also change with increasing number of compounds in the solution (from 5.8 to 6.3 kWh/kg, respectively, for EDBM of 2- and 5-componenmt solutions, 120 A/m2). These observed effects can be described in the same way as above. The results obtained during the tests of bipolar membrane electrodialysis on model solutions indicate a possibility to apply the EDBM process for the separation of succinic acid from the actual post-fermentation broth. Therefore, in the next stage of the study, the bipolar membrane electrodialysis of actual post-fermentation broth, after the ultrafiltration process as pre-treatment, was investigated. The processes were performed under a constant current densities 120 and 90 A/m2in 180 min. Fig. 4 illustrates the variations in the concentrations of the components of post-fermentation broth in the concentrate chamber in the time of EDBM. The increase in the succinic acid concentration after 180 min of process equals 10.9 and 9.5 g/dm3, for the current density 120 and 90 A/m2, respectively. Comparing the results presented and the experimental data obtained for EDBM of 5-component model solutions (Fig. 3(c) and 3(d)) a significant reduction in the transport of

Fig. 4. Change in concentrations of compounds of post-fermentation broth in concentrate chamber during EDBM process, i = 120 and 90 A/m2, T = 25 ± 2 °C.

365

Separation and Purification Technology 192 (2018) 360–368

K. Prochaska et al.

H2 Suc ⇔ HSuc− + H+

Table 4 Experimental results of desalination, current efficiency and energy consumption during EDBM of post-fermentation broth (a) 120 A/m2 and b) 90 A/m2), T = 25 ± 2 °C. i, A/m2

t, min

ηdes, %

CE, %

E, kWh/kg

120

30 60 120 180

20.6 31.3 39.9 46.8

26.1 19.9 12.7 9.9

3.5 4.6 7.1 9.0

30 60 120 180

21.5 30.9 33.9 40.8

37.8 27.2 14.9 12.0

2.2 3.0 5.4 7.0

90

(5)

and (6)

HSuc− ⇔ Suc2 − + H+

• Transfer of proton to amine: R3 N+ H+ ⇔ R3NH+

(7)

• Formation of ammonium salt with one carboxylic group of the acid: R3NH+org + HSuc−aq ⇔ R3NH2 Sucorg

(8)

Or both carboxylic groups: succinate ions (15.7 g/dm3 for 120 A/m2 and 13.0 g/dm3 for 90 A/m2) is clearly visible. The decrease can be explained by the presence of quantities of mono- and multivalent inorganic salts, which compete with the transport of co-ions due to their small size and high mobility [41]. Small amount of the transported succinate ions leads to low current efficiency (9.9 and 12.0% respectively for 120 and 90 A/m2) and high energy consumption (9.0 and 7.0 kWh/kg respectively for 120 and 90 A/m2) at the end of the process (Table 4) as against to the model solutions results (Table 3). Because of the transport of the other components during EDBM of the post-fermentation broth, in order to recover the largest possible amount of succinic acid, both fractions - diluate and concentrate - were subjected to the reactive extraction at the final stage of separation and purification processes.

2− 2R3NH+org + Sucaq ⇔ (R3NH)2 Sucorg

(9)

Thus, generally, the reaction of Suc extraction with tertiary amine can be expressed as:

R3Norg + H2 Sucaq ⇔ R3NH2 Sucorg

(10)

or

2R3Norg + H2 Sucaq ⇔ (R3NH)2 Sucorg

(11)

On the other hand, solvating extractants, such as Cyanex 923, transport the acid to the organic phase as a result of the acid molecule salvation (Kertes and King, 1986)

H2 Sucaq + 2Cyanex923org ⇔ (H2 Suc·2Cyanex923)org

(12)

Thus, with increasing pH, i.e. with decreasing content of the acidic form, the extraction of carboxylic acids decreases. However, solvating extractants are not likely to be as sensitive to pH changes as amines.The acid was stripped with 0.2 mol/dm3 NaOH from the loaded organic phase, and the efficiency of stripping exceeded 80%. Sodium succinate formed as a result of stripping can be easily converted into Suc, if necessary. The diluate and concentrate after preliminary treatment, i.e. EDBM process, were extracted in three steps with solutions of TOA or Cyanex 923 in Exxsol D 220/230. Exxsol D 220/230 was chosen as a diluent for organic phases used to extract Suc from actual solutions to improve phase separation after extraction. To avoid formation of emulsions or to improve disengagement of the phases after extraction, octanol applied as a diluent of the organic phases for model solutions - was replaced with aliphatic diluent in extraction from actual fermentation broths. The pH of diluate was adjusted from 9.92 to 2 to maintain acids in their acidic form. The pH of the concentrate after EDBM was equal to 4.3 and was not changed prior to extraction. Values of percentage extraction of Suc and other components of the diluate and concentrate after three extraction steps are shown in Table 5. As the broth contains, except from Glyc, a mixture of various carboxylic acids, all of them to some extent are extracted by both extractants. Percentage extraction of the acids from actual solutions both

3.3. Reactive extraction of succinic acid Prior to the three-step extraction of Suc from the solutions left after the actual post-fermentation broth processing, solvating (Cyanex 923) or basic (TOA) extractants were screened for Suc removal from model solution at various initial pH values. Dissociation constants of Suc are equal to pKa1 = 4.21 and pKa2 = 5.64 [42]. It means that at pH < pKa1 Suc exists in the aqueous solutions as an acid, at pH between pKa1 and pKa2 forms HSuc−, and above pKa2 a succinate salt is formed, as shown in Fig. 5. As TOA extracts from 60 to 20% succinic acid at pH range from 2.5 to 9, it means that only acidic form can be extracted efficiently with this tertiary amine. Such an efficient extraction at low pH can be attributed to the high contribution of the acidic form (shown in Fig. 5) and the following reactions of the acid (H2Suc) with the extractant (R3N) [25,43].

• Two stage dissociation of succinic acid:

Table 5 Initial concentrations of components (C0) of diluate and concentrate obtained in EDBM and extraction efficiency of these components after three stage extraction with 0.4 mol/ dm3 Cyanex 923 or TOA in Exxsol D220/230. Component

Diluate, pH adjusted to 2 0

C , g/dm

Suc Form Lact Ac Glyc

Fig. 5. Extraction efficiency of 0.01 mol/dm3 Suc with 0.1 mol/dm3 TOA (■) or Cyanex 923 (▴) in octanol from aqueous model solutions of various pH values.

366

13.8 3.5 4.3 4.6 12.2

3

Concentrate, pH 4.3 C0, g/dm3

E, % C 923

TOA

100 100 49 92 13

51 44 23 40 18

32.8 6.4 2.1 3.8 3.0

E, % C 923

TOA

47 26 23 72 19

6 8 41 24 18

Separation and Purification Technology 192 (2018) 360–368

K. Prochaska et al.

Acknowledgments The authors wish to acknowledge Polish Ministry of Science and Higher Education for the financial support (Grant No. DS-PB/32/ 0701).The research is a continuation of the theme of the project “Biotechnological conversion of glycerol to polyols and dicarboxylic acids” implemented within the Operational Programme – Innovative Economy, 2007 – 2013, co-financed by the European Union. PO IG 01.01.02.074/09. References [1] J.H. Ahn, Y.S. Jang, S.Y. Lee, Production of succinic acid by metabolically engineered microorganisms, Curr. Opin. Biotechnol. 42 (2016) 54–66. [2] C. Gao, X. Yang, H. Wang, C.P. Rivero, C. Li, Z. Cui, Q. Qi, C. Sze, K. Lin, Robust succinic acid production from crude glycerol using engineered Yarrowia lipolytica, Biotechnol. Biofuels 9 (2016) 1–11. [3] T. Kurzrock, D. Weuster-Botz, Recovery of succinic acid from fermentation broth, Biotechnol. Lett. 32 (2010) 331–339. [4] C.S. López-Garzón, A.J.J. Straathof, Recovery of carboxylic acids produced by fermentation, Biotechnol. Adv. 32 (2014) 873–904. [5] V. Bonnélye, L. Guey, J. del Castillo, UF/MF as RO pre-treatment: the real benefit, Desalination. 222 (2008) 59–65. [6] R.-S. Juang, H.-L. Chen, Y.-C. Lin, Ultrafiltration of coagulation-pretreated Serratia marcescens fermentation broth: flux characteristics and prodigiosin recovery, Sep. Sci. Technol. 47 (2012) 37–41. [7] Y.H. Cho, H.D. Lee, H.B. Park, Integrated membrane processes for separation and purification of organic acid from a biomass fermentation process, Ind. Eng. Chem. Res. 51 (2012) 10207–10219. [8] C. Wang, Q. Li, H. Tang, W. Zhou, D. Yan, J. Xing, Y. Wan, Clarification of succinic acid fermentation broth by ultrafiltration in succinic acid bio-refinery, J. Chem. Technol. Biotechnol. 88 (2013) 444–448. [9] W. Tomczak, M. Gryta, The application of ultrafiltration for separation of glycerol solution fermented by bacteria, Polish J. Chem. Technol. 15 (2013) 115–120. [10] M.O. Nigam, B. Bansal, X.D. Chen, Fouling and cleaning of whey protein concentrate fouled ultrafiltration membranes, Desalination 218 (2008) 313–322. [11] R. Fan, M. Ebrahimi, H. Quitmann, P. Czermak, Lactic acid production in a membrane bioreactor system with thermophilic Bacillus coagulans: fouling analysis of the used ceramic membranes, Sep. Sci. Technol. 50 (2015) 2177–2189. [12] N. Hilal, H. Al-Zoubi, N.A. Darwish, A.W. Mohammad, M. Abu, Arabi, A comprehensive review of nanofiltration membranes: treatment, pretreatment, modelling, and atomic force microscopy, Desalination 170 (2004) 281–308. [13] K.K. Cheng, X.B. Zhao, J. Zeng, R.C. Wu, Y.Z. Xu, D.H. Liu, J.A. Zhang, Downstream processing of biotechnological produced succinic acid, Appl. Microbiol. Biotechnol. 95 (2012) 841–850. [14] Y. Song, J. Xu, Y. Xu, X. Gao, C. Gao, Performance of UF-NF integrated membrane process for seawater softening, Desalination 276 (2011) 109–116. [15] J. Wiśniewski, G. Wiśniewska, T. Winnicki, Application of bipolar electrodialysis to the recovery of acids and bases from water solutions, Desalination. 169 (2004) 11–20. [16] Y. Wang, N. Zhang, C. Huang, T. Xu, Production of monoprotic, diprotic, and triprotic organic acids by using electrodialysis with bipolar membranes: effect of cell configurations, J. Memb. Sci. 385–386 (2011) 226–233. [17] Y. Li, S. Shi, H. Cao, X. Wu, Z. Zhao, L. Wang, Bipolar membrane electrodialysis for generation of hydrochloric acid and ammonia from simulated ammonium chloride wastewater, Water Res. 89 (2016) 201–209. [18] K. Ghyselbrecht, A. Silva, B. Van der Bruggen, K. Boussu, B. Meesschaert, L. Pinoy, Desalination feasibility study of an industrial NaCl stream by bipolar membrane electrodialysis, J. Environ. Manage. 140 (2014) 69–75. [19] C. Abels, F. Carstensen, M. Wessling, Membrane processes in biorefinery applications, J. Memb. Sci. 44 (2013) 285–317. [20] X. Wang, Y. Wang, X. Zhang, H. Feng, T. Xu, In-situ combination of fermentation and electrodialysis with bipolar membranes for the production of lactic acid: continuous operation, Bioresour. Technol. 147 (2013) 442–448. [21] L. Fu, X. Gao, Y. Yang, F. Aiyong, H. Hao, C. Gao, Preparation of succinic acid using bipolar membrane electrodialysis, Sep. Purif. Technol. 127 (2014) 212–218. [22] A.S. Kertes, C.J. King, Extraction chemistry of fermentation product carboxylic acids, Biotechnol. Bioeng. 28 (1986) 269–282. [23] A. Labbaci, G. Kyuchoukov, J. Albet, J. Molinier, Detailed investigation of lactic acid extraction with tributylphosphate dissolved in dodecane, J. Chem. Eng. Data 55 (2010) 228–233. [24] M. Wisniewski, M. Pierzchalska, Recovery of carboxylic acids C1–C3 with organophosphine oxide solvating extractants, J. Chem. Technol. Biotechnol. 80 (2005) 1425–1430. [25] Y.K. Hong, W.H. Hong, D.H. Han, Application of reactive extraction to recovery of carboxylic acids, Biotechnol. Bioprocess Eng. 6 (2001) 386–394. [26] L. de S. Moraes, F. de A. Kronemberger, H.C. Ferraz, A.C. Habert, Liquid-liquid extraction of succinic acid using a hollow fiber membrane contactor, J. Ind. Eng. Chem. 21 (2015) 206–211. [27] C. Umpuch, S. Sakeaw, S. Kanchanatawee, K. Jantama, Removal of contaminated organic acids from simulated succinic acid fermentation broth by reactive extraction process: single- and mixed-solute solution, Sep. Sci. Technol. 51 (2016)

Fig. 6. Selectivity of Suc extraction over other components of diluate (pH 2) or concentrate (pH 4.3) after EDBM with 0.4 mol/dm3 TOA or Cyanex 923.

with solvating (Cyanex 923) and basic (TOA) extractants depends on pH of the aqueous phase and is greater from the diluate at pH 2 than from the concentrate at pH near 4. Moreover, the highest efficiency of extraction after three steps is noted for Ac and Suc acids (both from diluate and concentrate) and Form acid (from diluate). The three-step extraction of Suc proves that the acid can be efficiently removed with Cyanex 923 both from the diluate and the concentrate after bipolar membrane electrolysis. However, in most cases studied, Suc cannot be selectively separated from other acids present in the aqueous phases. Reactive extraction enables the carboxylic acids to be separated from Glyc. The above observations are evidenced by values of selectivity of Suc extraction in the presence of other components of diluate or concentrate after EDBM shown in Fig. 6. Selectivity of Suc extraction was defined as the ratio of distribution ratios of Suc and another component in the organic and the aqueous phases:

S=

DSuc Danother component

(13)

The values of S were presented as logarithm to avoid very high differences among them. To ensure selective extraction of succinic acid in three steps it is necessary to decrease pH to 2 and use a solvating extractant.

4. Conclusions A system combining the ultrafiltration, bipolar membrane electrodialysis and reactive extraction is proposed for the separation of succinic acid from the broth left after bioconversion of glycerol. UF was applied as pre-cleaning step to remove the biological material. Significant reduction in the permeate occurred, however the fouling layer might be totally removed by cleaning the UF membrane. The use of EDBM process allowed elimination of the step of broth acidification which usually generates a considerable amount of wastes. In addition, the use of a cell configuration consisting of anion exchange and bipolar membranes enables effective separation of succinates from non-ionic compounds present in the separated solution. The succinic acid, present in the concentrate solution from EDBM, was removed with 90% efficiency in a three-step reactive extraction with a solvating extractant. Reactive extraction can support membrane techniques to separate efficiently carboxylic acids from the post-fermentation broth, however, it is not selective enough to separate succinic acid from other acids present in the broths. The only way to reach selective extraction of succinic acid over acetic, lactic acids and glycerol is to decrease pH to 2 and use Cyanex 923 as an extractant.

367

Separation and Purification Technology 192 (2018) 360–368

K. Prochaska et al.

[36] K. Wang, W. Li, Y. Fan, W. Xing, Integrated membrane process for the purification of lactic acid from a fermentation broth neutralized with sodium hydroxide, Ind. Eng. Chem. Res. 52 (2013) 2412–2417. [37] X. Han, D.J. Chadderdon, J. Qi, L. Xin, W. Li, W. Zhou, Numercial analysis of anionexchange membrane direct glycerol fuel cells under steady state and dynamic operations, Int. J. Hydrogen Energy. 39 (2014) 19767–19779. [38] L. Han, S. Galier, H. Roux-de, Balmann, Ion hydration number and electro-osmosis during electrodialysis of mixed salt solution, Desalination 373 (2015) 38–46. [39] R.C. Wu, Y.Z. Xu, Y.Q. Song, J.A. Luo, D. Liu, A novel strategy for salts recovery from 1,3-propanediol fermentation broth by bipolar membrane electrodialysis, Sep. Purif. Technol. 83 (2011) 9–14. [40] B. Igliński, S. Koter, R. Buckowski, M. Lis, The production of citric acid using electrodialysis with bipolar membrane of sodium citrate solutions, Polish J. Environ. Stud. 15 (2006) 411–417. [41] F.G. Wilhelm, I. Pünt, N.F.A. Van der Vegt, M. Wessling, H. Strathmann, Optimisation strategies for the preparationof bipolar membranes with reduced salt ion leakage in acid-base electrodialysis, J. Membr. Sci. 182 (2001) 13–28. [42] J.A. Tamada, A.S. Kertes, C.J. King, Extraction of carboxylic acids with amine extractants. L. Equilibriaand law mass action modeling, Ind. Eng. Chem. Res. 29 (1990) 1319–1326. [43] G. Kyuchoukov, A.F. Morales, J. Albet, G. Malmary, J. Molinier, On the possibility of predicting the extraction of dicarboxylic acids with tributylphosphate dissolved in a diluent, J. Chem. Eng. Data 53 (2008) 639–647.

629–640. [28] L. Ahsan, M.S. Jahan, H. Liu, Y. Ni, Recovery of acetic acid from pre-hydrolysis liquor of a kraft-based dissolving pulp production process by reactive extraction with tri-octyl amine (toa) and octanol, J. For. 2 (2012) 38–43. [29] G. Kyuchoukov, D. Yankov, J. Albet, J. Molinier, Mechanism of lactic acid extraction with quaternary ammonium chloride (Aliquat 336), Ind. Eng. Chem. Res. 44 (2005) 5733–5739. [30] J. Marták, Š. Schlosser, Extraction of lactic acid by phosphonium ionic liquids, Sep. Purif. Technol. 57 (2007) 483–494. [31] J. Marták, Š. Schlosser, M. Blahušiak, Mass-transfer in pertraction of butyric acid by phosphonium ionic liquids and dodecane, Chem. Pap. 65 (2011) 608–619. [32] F.S. Oliveira, J.M.M. Araújo, R. Ferreira, L.P.N. Rebelo, I.M. Marrucho, Extraction of l-lactic, l-malic, and succinic acids using phosphonium-based ionic liquids, Sep. Purif. Technol. 85 (2012) 137–146. [33] K. Prochaska, K. Staszak, M.J. Woźniak-Budych, M. Regel-Rosocka, M. Adamczak, M. Wiśniewski, J. Staniewski, Nanofiltration, bipolar electrodialysis and reactive extraction hybrid system for separation of fumaric acid from fermentation broth, Bioresour. Technol. 167 (2014) 219–225. [34] M. Waszak, M. Gryta, The ultrafiltration ceramic membrane used for broth separation in membrane bioreactor, Chem. Eng. J. 305 (2016) 129–135. [35] D. Pleissner, R. Schneider, J. Venus, T. Koch, Separation of lactic acid and recovery of salt-ions from fermentation broth, J. Chem. Technol. Biotechnol. 92 (2017) 504–511.

368