Separation and Purification of Lactic Acid from Fermentation Broth

Jun 30, 2017 - Lactic acid is produced industrially from bacterial fermentation of carbohydrates (e.g., sugar, starch) followed by separation processe...
0 downloads 10 Views 4MB Size
Article pubs.acs.org/IECR

Separation and Purification of Lactic Acid from Fermentation Broth Using Membrane-Integrated Separation Processes Hee Dae Lee,†,§ Min Yong Lee,†,§ Yoon Sung Hwang,†,§ Young Hoon Cho,†,‡ Hyo Won Kim,† and Ho Bum Park*,† †

Department of Energy Engineering, Hanyang University, Seoul 04763, Republic of Korea Korea Research Institute of Chemical Technology, Daejeon 34114, Republic of Korea



ABSTRACT: Lactic acid is produced industrially from bacterial fermentation of carbohydrates (e.g., sugar, starch) followed by separation processes such as precipitation, distillation, and reactive extraction. However, these conventional separation processes are energy intensive. In this study, we report an integrated membrane separation process consisting of ultrafiltration (UF) and nanofiltration (NF) for lactic acid recovery from fermentation broth, combined with ion exchange (IEX) and vacuum-assisted evaporation. Most organic and inorganic components in lactic acid fermentation broth, including microbes, glucose, and inorganic salt ions were successfully removed by UF and NF processes. Membrane fouling in the UF process became severe due to the high concentration of microbes and organic compounds. The effects of various UF membranes on the extent of membrane fouling were also studied to enhance separation efficiency. The separation of lactic acid continued using NF membranes, considering both size exclusion and Donnan exclusion effects. Finally, IEX and vacuum evaporation (VE) processes were also used to eliminate residual salt ions and to increase lactic acid purity. The hybrid membrane-separation process produced lactic acid with high purity (>99.5%). acid productivity.9,10 In this sense, a continuous fermentation process would be more favorable than the batch fermentation process. In general, precipitation is used to separate lactic acid from fermentation broth.11 In the precipitation process, lactic acid is recovered from precipitated calcium lactate by adding sulfuric acid. However, precipitation is not suitable from an economic perspective because the purification and separation stages take up 50% of the total production cost.12 Precipitation also generates a large amount of solid waste, a detrimental environmental effect.13 In this regard, alternative separation processes to recover lactic acid more efficiently have been studied. Among these, membrane-based separation processes effectively recover lactic acid from fermentation broth. Membrane separation processes are regarded as economical since they are easy to scale up with high separation efficiency.6 Electrodialysis (ED)14 and liquid surfactant membrane (LSM) extraction15,16 have also been investigated for lactic acid separation. ED shows high separation efficiency but still requires high energy input.11 LSM is a promising candidate separation process owing to easy scale-up and low energy and

1. INTRODUCTION Lactic acid is widely used in foods, detergents, and pharmaceutical and cosmetic applications.1−3 In the chemical industry, it is an essential raw material for resins, textile printing, and adhesives. Production of lactic acid is growing rapidly in the global market.2,4 The annual global market for lactic acid reached 259,000 t in 2012 and is forecasted to reach 367,300 t by the year 2017.5 Lactic acid is a naturally produced organic acid that can be commonly obtained by chemical synthesis or fermentation. For chemical synthesis of lactic acid, lactonitrile is used as an intermediate in industrial production.6−8 Although L-lactic acid is the target material in most cases, L- and D-lactic acids are often coproduced from chemical synthesis.6 Therefore, efficient separation of L-lactic acid from a mixture of the two optical isomers is desired. Chemical synthesis also produces a large amount of byproducts and impurities, and the raw materials for chemical synthesis of lactic acid, including acetaldehyde and hydrogen cyanide, are expensive. For these reasons, microbial fermentation processes have been of primary interest in lactic acid production.5,6 The batch fermentation process has been extensively studied in recent decades because it offers some key advantages, such as easy control of microbial contaminants and product quality per batch.9 On the other hand, batch processes have some drawbacks, such as high capital costs for large-scale lactic acid production. Moreover, batch processes often lead to low lactic © 2017 American Chemical Society

Received: Revised: Accepted: Published: 8301

May 15, 2017 June 27, 2017 June 30, 2017 June 30, 2017 DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research

Figure 1. Integrated membrane processes for lactic acid separation and purification proposed in this study.

capital costs for their operation.17 However, emulsion swelling due to water permeation through the membrane often reduces the separation efficiency because of the dilution of the extractants.18,19 This study proposes an efficient hybrid-integrated membrane separation process to recover lactic acid from the fermentation broth. The fermentation broth contains microbes, lactic acid, glucose, and a number of inorganic salt ions. As lactic acid is the target recovery material, all other components in the fermentation broth must be thoroughly separated. The proposed hybrid-integrated membrane separation process consists of four separation process units of ultrafiltration (UF), nanofiltration (NF), ion exchange (IEX), and vacuumassisted evaporation (VE) with the purity of lactic acid in each purification step, as shown in Figure 1. In the first step, a UF membrane process was used as a pretreatment to remove microbial or suspended colloidal particles. The removal of these components can improve the separation efficiency in subsequent membrane separation processes. Also, microbial species are required to be completely separated and continuously recycled in a fermentation broth.6 The next separation process unit uses NF to separate organic solutes (e.g., glucose and lactic acid) from fermentation broth. In the NF separation process, size sieving and electrostatic repulsion are the major separation mechanisms.20 For the sizesieving mechanism, solutes with molecular weights over the molecular weight cutoff (MWCO) of NF membranes (e.g., 150−250 g/mol for NF membranes) are retained.21 Thus, glucose is effectively separated from lactic acid due to the difference in the molecular weights of glucose (180 g/mol) and lactic acid (90.08 g/mol). The electrical charge of the NF membrane also affects organic molecule separation. The NF membrane is a charged membrane, with different surface electrostatic charges as a function of pH, which influences the transport of charged molecules such as lactic acid and inorganic salt ions.22 Next, the residual monovalent inorganic ions are removed using an IEX resin. Since the ability of NF membranes to reject monovalent ions is much more limited than that of reverse osmosis (RO) membranes, the IEX process is introduced to remove the small amount of residual inorganic salt ions. Finally, high-purity concentrated lactic acid is recovered by VE. The hybrid separation process based on membrane technology proposed in this study for obtaining lactic acid is very energy efficient. In general, RO can be used to separate water from feed stock, but it has drawbacks in terms of energy consumption because higher applied pressure is required. In this regard, the VE process was introduced instead of RO as a final process to concentrate lactic acid. As a result,

highly enriched lactic acid was recovered by VE with low thermal energy.

2. EXPERIMENTAL SECTION 2.1. Materials. Fermented broth solution consisting of microbes, soluble organics, and inorganic salts was supplied by SAIT (Samsung Advanced Institute of Technology, Suwon, Korea). Details of each component in the fermentation broth are summarized in Table 1. The commercial UF membranes Table 1. Specific Information of Components in Fermentation Broth Used in This Study components organic

concentration (ppm) microbial species lactic acid glucose

11.1 (optical density) 54,000 22,000

inorganic cation

sodium (Na+) ammonium (NH4+) potassium (K+) calcium (Ca2+) magnesium (Mg2+)

200 50 1000 3500 80

inorganic anion

chloride (Cl−) nitrate (NO3−) phosphate (PO43−) sulfate (SO42−)

300 50 1000 400

were purchased from Sepromembranes (Oceanside, CA, USA), while the HL membrane and NE 90 membrane were purchased from GE (Feasterville-Trevose, PA, USA) and Woongjin Chemical (Seoul, South Korea), respectively. All membrane specifications are documented in Table 2. The IEX resins were purchased from Samyang Corporation (Seoul, Korea) and Mitsubishi Chemical (Tokyo, Japan). Distilled water was produced from a Milli-Q system (Millipore, Billerica, MA, USA). 2.2. Clarification of Fermentation Broth by Microfiltration and Ultrafiltration (UF). A dead-end filtration experiment was performed to evaluate the cell removal efficiency of MF and UF membranes. MF and UF membranes were installed in a stirred filtration cell (Amicon 8050, Millipore, Billerica, MA, USA). The filtration experiment was operated at 1 bar and pressurized by a nitrogen gas cylinder. Optical density was measured using an optical density meter (Ultrospec10, GE Healthcare, Piscataway, NJ, USA) to evaluate the cell removal efficiency of the MF and UF membranes. Here, 8302

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research Table 2. Specifications of Commercial UF and NF Membranes Used in This Study type

a

material

water flux (L/m2 h bar)

name

MWCO (kDa)

contact angle (deg)

UF

PSF PES PES PES PES PAN

PS 20 PES 5 PES 10 PES 20 PES 900 PAN 200

900 70 100 350 1200 300

20 6 10 20 20 20

70 46 47 53 50 54

NF

PA aromatic PA

HL NE90 (Woongjin)

12.6 11.8

95%a 99.5%a

27 26

MgSO4 rejection for NF membranes

Table 3. Ion Exchange Resins Used in This Study name type ionic form functional group manufacturer exchange capacity operating pH range

strong acid cationic resin

weak acid cationic resin

strong base anionic resin

weak base anionic resin

SCR-BH Gel H+ SO3− Samyang 2.0 mequiv/mL-R 0−14

WK60L Porous H+ COO− Mitsubishi Chem 4.4 meq/mL-R 4−14

SAR10MBOH Gel OH− N+(CH3)3Cl− Samyang 1.3 meq/mL-R 0−14

AW90 Porous OH− (CH2)nN(CH3)2 Samyang 1.5 meq/mL-R 0−9

⎛ C ⎞ Rejection(%) = ⎜1 − P ⎟ × 100 CF ⎠ ⎝

a chlorinated poly(vinyl chloride) (CPVC, Pure Envi-Tech, Korea) MF membrane and PS20 (Sepromembranes, Oceanside, CA, USA) UF membrane were used. The removal efficiency was calculated from the equation below: ⎛ D ⎞ Cell removal (%) = ⎜1 − P ⎟ × 100 DF ⎠ ⎝

(2)

where CP and CF are the permeate concentration (g/L) and the feed concentration (g/L), respectively. A high pressure stirred cell (HP4750, Sterlitech, Kent, WA, USA) was employed for the filtration experiment. First, pure water was filtered through the membrane until the total organic carbon (TOC) of the permeate dropped below 1 ppm in order to minimize the effect of residual impurities in membrane. TOC was measured using a total organic carbon analyzer (Multi N/C 3100, Analytikjena, Jena, Germany). The clarified fermentation broth was then filtered at 13.8 bar. 2.4. Inorganic Ion Removal by Ion Exchange (IEX) Column. Ion chromatography (IC, 883 Basic IC Plus, Metrohm AG, Herisau, Switzerland) was used to detect specific ion species and to determine specific ion concentrations in solution. Ion species and ion concentrations were analyzed based on the difference in the retention time of each ion. IEX resins used in this study are summarized in Table 3. A column filtration method was employed to remove inorganic ions from the clarified broth. A vertical 50 mL IEX column was carefully filled with preswollen IEX resin. Deionized water was continuously fed into the prepared IEX column until the ion conductivity of the permeate dropped below 1 μS cm−1 to eliminate any impurities existing in the ion-exchange resin. The ion conductivity was measured using an ion conductivity meter (inoLab 720, WTW, Weilheim, Germany). After stabilization, the clarified broth was fed to the IEX column, and filtrate solution was periodically sampled. 2.5. Vacuum-Assisted Evaporation (VE). The concentrated, high-purity lactic acid was finally recovered using an evaporation process. Water was evaporated at 80 °C under vacuum. After the VE process, the lactic acid concentration was measured with HPLC, and the concentration factor was calculated from the equation below:

(1)

where DF and DP are the optical density (absorbance) of feed and permeate, respectively. The membrane fouling propensity was also evaluated for the commercial UF membranes shown in Figure 3. For membrane stabilization, pure water filtration was performed for 10 min. Then, the permeate flux of the membrane in the fermentation broth was measured for 1 h. After fermentation broth filtration, the cake layers formed on membrane surface were thoroughly rinsed with pure water. Finally, pure water filtration was performed again for 10 min to estimate the degree of irreversible membrane fouling.23 2.3. Separation of Lactic Acid from Clarified Fermentation Broth by Nanofiltration (NF). The rejection of lactic acid and glucose at different feed pH levels was measured to study the transport behavior of organic solute in the NF membrane. Lactic acid and glucose were purchased from Sigma-Aldrich (St. Louis, MO, USA), and 1000 ppm feed solutions of glucose and lactic acid were prepared. The pH level of feed solution was adjusted using 1 M HCl and NaOH solutions. Glucose must be thoroughly separated from the fermentation broth to enhance the chemical purity of lactic acid so that the NF membrane process was introduced as a multistage filtration process in this study. Accordingly, the clarified broth was filtered three times using the NF membrane process. Lactic acid and glucose concentration in the clarified fermentation broth were measured using high-performance liquid chromatography (HPLC, E2695, Waters, Milford, MA, USA). The NF membrane specifications used in this study are summarized in Table 2. The solute rejection was calculated from the equation below: 8303

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research Concentration factor =

Ct C0

cells and large colloidal particles was evaluated by comparing the colors before (dark brown) and after filtration (light yellow), as shown in Figure 2(b). The turbidity in the fermentation broth was markedly reduced because the cells, and insoluble large particles were completely removed by the pressure-driven membrane separation processes. Although both MF and UF membranes can remove microbes from the fermentation broth with almost 100% rejection efficiency, the UF membrane process was chosen as a pretreatment step in this study. More particles penetrate into the MF membrane than the UF membrane since the pore size of the MF membrane is larger than that of the UF membrane. The propensity of adsorption into the membrane’s inner pores increases when more particles larger than the membrane surface pores penetrate the membrane. Therefore, the internal fouling tendency is more significant in the MF membrane process than the UF membrane process. 3.2. Membrane Fouling Issue. Figure 3(a) shows that the water flux continuously declined over the course of the filtration experiment. Compared with pure water filtration, a severe reduction of permeate flux was observed during the fermentation broth filtration process. The flux reduction is mainly caused by membrane fouling, which depends on not only operation parameters, but also the physicochemical properties and morphologies of the membrane.24 When the solution permeates the membrane, the membrane surface pores are often plugged by foulants larger than the membrane surface pore, such as microbes and colloidal particles. As a result of pore blockage, the membrane resistance increases significantly. As discussed previously, the majority of microorganisms in the fermentation broth were rejected by membrane filtration. In addition, the accumulated solid particles at the membrane surface give rise to a thick cake layer. Therefore, both membrane pore blocking and thick cake layer formation contribute to the observed severe flux reduction and related membrane performance. Figure 3(b) shows the normalized water fluxes of poly(ether sulfone) (PES) UF membranes with different pore sizes. UF membrane pore size is practically defined in terms of MWCO, which is summarized in Table 2. As shown in Figure 3(b), the permeate flux of fermentation broth declines as the MWCO of each membrane increases. The PES 900 membrane, with the largest MWCO and highest flux among the PES membranes tested, shows the lowest flux during the fermentation broth filtration. This measurement result indicates that more microbes and colloidal particles penetrate the UF membrane so that the frequency of particle adsorption into membrane pores increases. Colloidal particles that penetrate the membrane surface often plug the internal pores of the membrane, a process called internal fouling.25 Among PES membranes, the most severe internal fouling occurred in the PES 900 membrane. In contrast to other PES membranes, the permeate flux of the PES 900 membrane did not largely recover after changing the feed solution to pure water, indicating that some portion of the internal fouling was irreversible. Thus, severe internal fouling occurs for increased membrane surface pore size. Figure 3(c) shows the normalized water fluxes of different kinds of UF membranes with the same MWCO. Foulant adhesion is caused by electrostatic interaction, hydrophobic interaction, and charge transfer.24,26 Although various UF membranes have similar MWCOs, the PES membrane showed the highest pure water flux. In addition, the PES membrane

(3)

where Ct and C0 are the lactic acid concentration (g/L) in the permeate solution after VE and in the feed solution, respectively. A thermal evaporation experiment was also performed to evaluate the suitability of the VE process with respect to energy efficiency. The VE experiment was performed at 80 °C. The chemical purity of the lactic acid was calculated from following equation: Lactic acid purity(%) =

Lactic acid concentration(g/L) Total Dissolved Solids concentration(g/L)

(4)

3. RESULTS AND DISCUSSION 3.1. Clarification of Fermentation Broth by Membrane Filtration. In continuous fermentation processes, microbial species and large particles need to be consistently separated from the fermentation broth for continuous lactic acid production. Therefore, it is important to continuously separate microbes from the fermentation broth in the pretreatment step. MF and UF membrane processes are both suitable for pretreatment since the sizes of microbes (1−5 μm) are larger than the membrane pore sizes (0.45 μm for CPVC MF membrane, 20 kDa MWCO for PS20 UF membrane). However, there was no significant difference in the cell removal efficiency between MF and UF membrane processes. The optical density measurement results shown in Figure 2(a) presents the cell separation efficiencies in the MF and UF membrane processes. Almost 100% cell removal efficiency was achieved by both UF and MF membranes. The rejection of

Figure 2. (a) Optical density measurement before and after membrane filtration, representing the cell separation efficiency by MF and UF membrane processes. (Experimental conditions: stirred dead-end filtration, 1 bar, 25 °C. Fermentation broth used as feed solutions.) (b) Turbidity reduction by MF and UF membrane filtration. 8304

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research

Figure 3. (a) Membrane fouling propensity of a UF membrane during fermentation broth filtration, (b) effect of membrane surface pore size, and (c) effect of membrane material on fouling propensity. Experimental conditions: stirred dead-end filtration, 1 bar, 25 °C. Fermentation broth and deionized water were used as feed solutions..

fermentation broth filtration is largely influenced by membrane hydrophilicity. 3.3. Organic Solute Rejection by NF Membrane. Figure 4(a) and (b) shows the transport behavior of organic molecules in the NF membrane process as a function of pH. The experiment was carried out with 1000 ppm of a single solution, and solute concentration was obtained from TOC analysis. The TOC analysis results show that lactic acid rejection is lowest at a pH of 3, while glucose is similarly rejected at all tested pH levels. In NF membrane processes, organic solutes (e.g., lactic acid and glucose) in a mixture can be retained by size sieving and Donnan exclusion.25,28 The clarified fermentation broth contains a number of electrolytes (e.g., inorganic salt ions and lactic acid) and nonelectrolytes (e.g., glucose). If only neutral solutes exist in solution, size exclusion will be the primary separation mechanism because Donnan exclusion has no effect on an electrically neutral solute.21 For fermentation broth, however, both size sieving and Donnan exclusion affect the transport of different solutes because lactic acid, glucose, and inorganic ions co-exist in the feedstock. Size sieving is due to

exhibited a higher flux recovery ratio after membrane cleaning than any other membrane. In this study, microbes and suspended colloidal particles in the feed stock were successfully rejected by the PES membrane, and the interaction between microbes and the PES membrane seems to be relatively weaker than with the others. As mentioned above, both the physicochemical properties of the membrane and operational parameters have an effect on membrane performance during the filtration process. Specifically, membrane surface parameters (e.g., hydrophobicity, charge, roughness, and porosity) have a strong influence on membrane fouling.27 In general, hydrophobic membranes are more susceptible to membrane fouling than hydrophilic membranes because hydrophobic organics tend to adsorb more readily onto a hydrophobic membrane surface.26 On the basis of the contact angle data summarized in Table 2, the PES 900 and PAN 200 membranes have a more hydrophilic surface than the PS 20 membrane. Therefore, the recovery ratio in water flux after membrane cleaning in the PS 20 membrane is much lower than that with the PES 900 membrane. Accordingly, the membrane fouling during 8305

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research

Figure 4. Effect of solution pH on (a) lactic acid rejection and (b) glucose rejection. Experimental conditions: stirred dead-end filtration, 6.9 bar, 25 °C. Solutions of 1000 ppm lactic acid and glucose were used as feed solutions. Comparison of (c) lactic acid rejection and (d) glucose rejection by NF membranes. Experimental conditions: stirred dead-end filtration, 13.8 bar, 25 °C. Clarified broth was used as a feed solution.

membrane was observed at pH 5.5,31 and the NE 90 membrane surface had a positive charge below pH 3.3.32 Therefore, the HL and NE 90 membranes have positive surface charges because the pH of the fermentation broth used in this work is 3.0, which is below the isoelectric point. The surface zetapotentials of HL and NE 90 at pH 3 are 3.5 mV31 and 0 mV,32 respectively. Thus, the positive charge density of the HL membrane is higher than that of the NE 90 membrane at pH 3. Accordingly, the penetration of lactic acid in NF membranes is affected by the electrostatic interaction between lactic acid and the membrane surface at low pH. Thus, the separation of lactic acid molecules largely depends on Donnan exclusion. In general, lactic acid is regarded as a negatively charged molecule because of its carboxylate groups (COO−). However, in acidic conditions, most of the lactic acid is dissolved in water as neutral molecules instead of charged molecules due to protonation of the carboxylate groups. Therefore, lactic acid is not regarded as an electrolyte at pH 3, so the rejection of lactic acid is largely influenced by the sieving mechanism rather than Donnan exclusion. Figure 4(c) and (d) represents the rejection of glucose and lactic acid from the clarified fermentation broth by different NF membranes at the same pH level. The glucose and lactic acid

the size difference between the solute and the membrane surface pore, while Donnan exclusion depends strongly on the surface charge of the membranes.29 In size sieving, large molecules with molecular weights over 150−250 g/mol can be retained by NF membranes.21 The molecular weights of glucose (C6H12O6) and lactic acid (C3H6O3) are 180 and 90.08 g/mol, respectively. Therefore, the separation of glucose from lactic acid is strongly affected by the difference in molecular size. Donnan exclusion also affects the separation of organic molecules because both the NF membrane surface and lactic acid molecules have electrostatic charge. In general, NF membranes are prepared by interfacial polymerization of acyl chlorides (COCl−) and amines,30 similar to interfacial polyamide RO membranes. NF membranes have excess carboxylic acid groups and a few amine groups at the surface. The surface layer of NF membranes is electrically charged by the protonation and deprotonation of the functional groups at the membrane surface. The positive surface charge of the membrane is caused by the protonation of primary amine groups (i.e., NH2 → NH3+), and the negative charge results from the deprotonation of carboxyl acid groups (i.e., COOH → COO −). Comparing the isoelectric points (where the membrane is uncharged), the isoelectric point of the HL 8306

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research

multistage filtration process, while glucose was not completely separated in a single NF membrane process. This represents the correlation between solute concentration and permeate flux (Jw). The permeate flux increased as the glucose concentration decreased owing to the decrease in the trans-membrane osmotic pressure difference at the same applied feed pressure of 13.8 bar. Solute accumulation on a membrane surface leads to another resistance to the permeate flux, the so-called “concentration polarization”, so that permeate flux is low when glucose concentration is high. 3.4. Inorganic Ion Exclusion by NF Membranes. As mentioned above, NF membranes are ampholytic membranes that have different surface charges at various pH due to the protonation or deprotonation of functional groups (e.g., NH2 and COOH). Therefore, NF membranes repel co-ions with the same charge as membrane surface, while counterions can penetrate the membrane by electrostatic force. In general, NF membranes show excellent performance in rejecting sugars and multivalent salts by both size sieving and electrostatic repulsion effect. Figure 6 shows the rejection trends of different inorganic salt ions in the clarified broth by the NF membrane. Like the organic solute separation previously discussed, two separation mechanisms explain the inorganic salt ion rejection in NF membrane filtration. Separation of inorganic ion salts is influenced by the size sieving mechanism. There is a variation in hydrated ion radius of ions existing in an aqueous solution. The ion hydration can also be affected by the attraction between ions and solvent.33 For cations, the divalent cations (Mg2+ and Ca2+) are more easily rejected than monovalent ions (Na+ and K+); this is caused by the size difference between hydrated ions and the membrane pores. Multivalent salts are retained more consistently than monovalent salts because their hydrated radius is larger than that of the NF membrane pores. Donnan exclusion also plays a role in salt ion rejection. The NF membrane has a positively charged surface at low pH due to the protonation of amine groups, so cations are more consistently rejected than anions by the electrostatic repulsion. As presented in Figure 6, cations were highly rejected by NF membranes, while anions still passed through the NF membrane. When rejection of monovalent cations (e.g., K+) is compared with monovalent anions (e.g., Cl− and NO3−),

concentrations were obtained from HPLC analysis. HPLC analysis results indicated that most glucose was rejected by both NF membranes. However, the glucose rejection by the NE 90 membrane is slightly higher than by the HL membrane. On the basis of the NF membrane specifications used in this study and summarized in Table 2, the NE 90 membrane has a much thicker layer than the HL membrane because it shows lower pure water flux and higher solute rejection. Therefore, glucose rejection largely depends on the membrane surface characteristics. For lactic acid, more lactic acid passes through an HL membrane than an NE 90 membrane. As mentioned above, the surface charge density in the HL membrane is higher than in the NE 90 membrane at pH 3. As such, the electrical attraction between lactic acid and the HL membrane is greater than that of the NE 90 membrane so that the rejection of lactic acid was lower when using an HL membrane. Accordingly, the lactic acid rejection is influenced by the charge density of membrane surface. As presented in Figure 5, glucose in the clarified broth was thoroughly separated when NF membranes were arranged as a

Figure 5. Comparison of water flux and organic solute rejection during the multistage NF process. Experimental conditions: stirred dead-end filtration, 13.8 bar, 25 °C. Clarified broth was used as a feed solution.

Figure 6. (a) Cation and (b) anion rejection of NF membranes. Experimental conditions: HL membrane, stirred dead-end filtration, 34.5 bar, 25 °C, clarified broth was used as a feed solution. 8307

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research

Figure 7. (a) Cation adsorption and (b) anion adsorption using various ion exchange resins. Experimental conditions: column filtration, 25 °C. Clarified broth as a feed solution.

Figure 8. Changes in (a) cation and (b) anion concentrations in each separation step.

which have similar hydrated radius, markedly lower rejection is shown for the anions. Accordingly, not only size sieving but also electrostatic repulsion affects the ion separation in NF membrane filtration. 3.5. Inorganic Ion Removal in IEX Process. In the IEX process, inorganic salt ions are selectively adsorbed in a resin as the feed solution passes through a column filled with IEX resin. Figure 7 shows the effect of IEX resin type on the adsorption efficiency of specific salt ions. Among the IEX resins used in this study, strongly acidic resins (SCR-BH) and weakly basic exchange resins (AW90) were the most effective in removing the small amount of residual salt ions. For cationic exchange resins, the weakly acidic exchange resin (WK60L) is a poroustype resin with a high surface area and a high ion exchange capacity. However, it can be operated only in the pH range of 4−14. Therefore, at low pH, a strongly acidic exchange resin is more effective at removing ions than a weakly acidic resin. Weakly basic exchange resins are also porous and have a higher ion exchange capacity (IEC) than a strongly basic resin; that is, a weakly basic exchange resin is more efficient at rejecting anions of interest.

Figure 8 shows the changes in cation and anion concentrations in each separation step. Residual monovalent cations (NH4+ and K+) were completely removed by the cation exchange resin column. Divalent ions can be mostly removed by the NF membrane process, but the ability of an NF membrane to reject monovalent ions is limited. In this regard, the IEX process helps remove the residual monovalent ions. 3.6. Concentration through Vacuum-Assisted Evaporation (VE). In general, osmotically driven membrane processes (ODMPs) such as RO, forward osmosis (FO), and pressure-retarded osmosis (PRO) are widely available in various water treatment processes.34,35 Among them, RO shows sufficiently strong rejection of both sugar and salts that it has been studied as a concentration method.36 However, RO has drawbacks with respect to energy efficiency. Specifically, RO requires a high operating pressure to pass water from a highly concentrated solution (feed solution) to a more dilute solution (permeate solution). In addition, the permeation rate largely depends on the applied pressure, so pressure is a major factor to be considered in operation. In this regard, an alternative to RO is essential to separate water with low 8308

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research

Figure 9. Lactic acid enrichment (dewatering) by (a) thermal evaporation and (b) vacuum-assisted evaporation.

process removed monovalent ions completely. Finally, highpurity lactic acid (>99.5%) was recovered by VE. We envision that the introduction of the dewatering process such as forward osmosis can lead highly concentrated lactic acid in the near future.

pressure and energy efficiency. Therefore, the evaporation process was introduced as a final step to obtain concentrated lactic acid from the lactic acid/water mixture in this study. In contrast to RO, applied pressure for dewatering is not required in the evaporation process, minimizing the energy consumption problem. In addition, evaporation is a simple and reliable process. In this study, VE was introduced to enrich the lactic acid. In contrast to conventional evaporation processes, VE is beneficial for lactic acid separation because lactic acid enrichment is possible with low thermal energy. When the two different evaporation methods were compared, as presented in Figure 9, the more enriched lactic acid was recovered by VE. Thus, VE is advantageous not only in terms of energy efficiency but also in lactic acid recovery ratio. Highly concentrated lactic acid (76 wt %) was obtained through VE as shown in Figure 9. Accordingly, VE is suitable as a final step for obtaining lactic acid with high purity.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +82-2-2220-2338. Fax: +82-2-2291-5982. ORCID

Young Hoon Cho: 0000-0002-8863-1979 Ho Bum Park: 0000-0002-8003-9698 Author Contributions §

H. D. Lee, M. Y. Lee, and Y. S. Hwang contributed equally to this work and should be considered co-first authors. Notes

The authors declare no competing financial interest.



4. CONCLUSIONS In this study, we proposed a hybrid-integrated membrane separation process consisting of UF, NF, IEX, and VE. In UF, the microbes were first removed from the fermentation broth by size exclusion. In continuous fermentation, microbes need to be recycled for continuous lactic acid production. Therefore, UF is a suitable pretreatment step in this process for cell recycling. However, severe membrane fouling was observed during fermentation broth filtration. That is, antifouling UF membranes should be developed for this purpose. Second, NF was used as a purification step. Lactic acid, glucose, and inorganic ions in clarified broth were rejected in the purification step based on size and charge repulsion. While lactic acid passed through the NF membrane, glucose was rejected by the NF membrane process because of its relatively larger size. In addition, the NF membranes used in this study (e.g., HL and NE 90) have a positive surface charge since the pH of the fermentation broth is below the isoelectric point of each membrane. Therefore, electrolyte (lactic acid) easily permeates the NF membranes due to electrostatic attraction, and neutral molecules (glucose) were retained by the NF membrane. Divalent ions were also excluded due to the surface charge of the membrane. The residual inorganic ions were then almost completely removed by the IEX process. The NF process is limited in its ability to remove monovalent ions, but the IEX

ACKNOWLEDGMENTS This work was supported by the project “Development of energy efficiency separation and purification processes of lactic acid” (201200000002177), Samsung Electronics Corporation.



REFERENCES

(1) Martinez, F. A. C.; Balciunas, E. M.; Salgado, J. M.; Gonzalez, J. M. D.; Converti, A.; Oliveira, R. P. D. Lactic acid properties, applications and production: A review. Trends Food Sci. Technol. 2013, 30 (1), 70−83. (2) Rogers, P.; Chen, J.-S.; Zidwick, M. J. Organic Acid and Solvent Production: Acetic, Lactic, Gluconic, Succinic, and Polyhydroxyalkanoic Acids. In The Prokaryotes; Springer, 2013; pp 3−75. (3) Baniel, A. M.; Eyal, A. M.; Mizrahi, J.; Hazan, B.; Fisher, R. R.; Kolstad, J. J.; Stewart, B. F. Lactic acid production, separation and/or recovery process. U.S. Patent No. 5,510,526, 1996. (4) Kulprathipanja, S.; Oroskar, A. R. Separation of lactic acid from fermentation broth with an anionic polymeric absorbent. U.S. Patent No. 5,068,418, 1991. (5) Abdel-Rahman, M. A.; Tashiro, Y.; Sonomoto, K. Recent advances in lactic acid production by microbial fermentation processes. Biotechnol. Adv. 2013, 31 (6), 877−902. (6) Pal, P.; Sikder, J.; Roy, S.; Giorno, L. Process intensification in lactic acid production: A review of membrane based processes. Chem. Eng. Process. 2009, 48 (11−12), 1549−1559.

8309

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310

Article

Industrial & Engineering Chemistry Research (7) Tsao, G. T.; Lee, S. J.; Tsai, G.-J.; Seo, J.-H.; McQuigg, D. W.; Vorhies, S. L.; Iyer, G. Process for producing and recovering lactic acid. U.S. Patent No. 5,786,185, 1998. (8) Narayanan, N.; Roychoudhury, P. K.; Srivastava, A. L (+)lactic acid fermentation and its product polymerization. Electron. J. Biotechnol. 2004, 7 (2), 167−178. (9) Klasson, T.; Clausen, E.; Gaddy, J. Continuous fermentation for the production of citric acid from glucose. Appl. Biochem. Biotechnol. 1989, 20-21 (1), 491−509. (10) Gonzalez, M. I.; Alvarez, S.; Riera, F. A.; Alvarez, R. Lactic acid recovery from whey ultrafiltrate fermentation broths and artificial solutions by nanofiltration. Desalination 2008, 228 (1−3), 84−96. (11) Wasewar, K. L.; Yawalkar, A. A.; Moulijn, J. A.; Pangarkar, V. G. Fermentation of glucose to lactic acid coupled with reactive extraction: a review. Ind. Eng. Chem. Res. 2004, 43 (19), 5969−5982. (12) Eyal, A. M.; Bressler, E. Industrial Separation of Carboxylic and Amino-Acids by Liquid Membranes - Applicability, Process Considerations, and Potential Advantages. Biotechnol. Bioeng. 1993, 41 (3), 287−295. (13) Shreve, R. N.; Brink, J. A., Jr. Chemical Process Industries. McGraw-Hill Book Co.: 1977. (14) Lee, E. G.; Moon, S. H.; Chang, Y. K.; Yoo, I. K.; Chang, H. N. Lactic acid recovery using two-stage electrodialysis and its modelling. J. Membr. Sci. 1998, 145 (1), 53−66. (15) Sirman, T.; Pyle, L.; Grandison, A. S. Extraction of organic acids using a supported liquid membrane. 15. Biochem. Soc. Trans. 1991, 19 (3), 274S−274S. (16) Yuanli, J.; Fuan, W.; Hyun, K. D.; Sook, L. M. Modeling of the permeation swelling of emulsion during lactic acid extraction by liquid surfactant membranes. J. Membr. Sci. 2001, 191 (1), 215−223. (17) Kocherginsky, N.; Yang, Q.; Seelam, L. Recent advances in supported liquid membrane technology. Sep. Purif. Technol. 2007, 53 (2), 171−177. (18) Chaudhuri, J.; Phyle, D. Emulsion liquid membrane extraction of organic acidsI. A theoretical model for lactic acid extraction with emulsion swelling. Chem. Eng. Sci. 1992, 47 (1), 41−48. (19) Bart, H.; Jüngling, H.; Ramaseder, N.; Marr, R. Water and solute solubilization and transport in emulsion liquid membranes. J. Membr. Sci. 1995, 102, 103−112. (20) Schäfer, A.; Fane, A. G.; Waite, T. Nanofiltration of natural organic matter: removal, fouling and the influence of multivalent ions. Desalination 1998, 118 (1), 109−122. (21) Vellenga, E.; Trägårdh, G. Nanofiltration of combined salt and sugar solutions: coupling between retentions. Desalination 1998, 120 (3), 211−220. (22) Bouchoux, A.; Roux-de Balmann, H.; Lutin, F. Nanofiltration of glucose and sodium lactate solutions: Variations of retention between single-and mixed-solute solutions. J. Membr. Sci. 2005, 258 (1), 123− 132. (23) Yamamura, H.; Kimura, K.; Watanabe, Y. Mechanism involved in the evolution of physically irreversible fouling in microfiltration and ultrafiltration membranes used for drinking water treatment. Environ. Sci. Technol. 2007, 41 (19), 6789−6794. (24) Fane, A.; Fell, C. A review of fouling and fouling control in ultrafiltration. Desalination 1987, 62, 117−136. (25) Van der Bruggen, B.; Schaep, J.; Wilms, D.; Vandecasteele, C. Influence of molecular size, polarity and charge on the retention of organic molecules by nanofiltration. J. Membr. Sci. 1999, 156 (1), 29− 41. (26) Hilal, N.; Ogunbiyi, O. O.; Miles, N. J.; Nigmatullin, R. Methods employed for control of fouling in MF and UF membranes: a comprehensive review. Sep. Sci. Technol. 2005, 40 (10), 1957−2005. (27) Evans, P. J.; Bird, M. R.; Pihlajamäki, A.; Nyström, M. The influence of hydrophobicity, roughness and charge upon ultrafiltration membranes for black tea liquor clarification. J. Membr. Sci. 2008, 313 (1), 250−262. (28) Schaep, J.; Van der Bruggen, B.; Vandecasteele, C.; Wilms, D. Influence of ion size and charge in nanofiltration. Sep. Purif. Technol. 1998, 14 (1), 155−162.

(29) Hussain, A.; Abashar, M.; Al-Mutaz, I. Prediction of Charge Density for Desal-HL Nanofiltration Membrane from Simulation and Experiment using Different Ion Radii. Sep. Sci. Technol. 2007, 42 (1), 43−57. (30) Childress, A. E.; Elimelech, M. Relating nanofiltration membrane performance to membrane charge (electrokinetic) characteristics. Environ. Sci. Technol. 2000, 34 (17), 3710−3716. (31) Braeken, L.; Bettens, B.; Boussu, K.; Van Der Meeren, P.; Cocquyt, J.; Vermant, J.; Van der Bruggen, B. Transport mechanisms of dissolved organic compounds in aqueous solution during nanofiltration. J. Membr. Sci. 2006, 279 (1), 311−319. (32) Shah, A. D.; Huang, C.-H.; Kim, J.-H. Mechanisms of antibiotic removal by nanofiltration membranes: Model development and application. J. Membr. Sci. 2012, 389, 234−244. (33) Nightingale, E., Jr Phenomenological theory of ion solvation. Effective radii of hydrated ions. J. Phys. Chem. 1959, 63 (9), 1381− 1387. (34) Cho, Y. H.; Han, J.; Han, S.; Guiver, M. D.; Park, H. B. Polyamide thin-film composite membranes based on carboxylated polysulfone microporous support membranes for forward osmosis. J. Membr. Sci. 2013, 445, 220−227. (35) Li, Z.; Linares, R. V.; Abu-Ghdaib, M.; Zhan, T.; YangaliQuintanilla, V.; Amy, G. Osmotically driven membrane process for the management of urban runoff in coastal regions. Water Res. 2014, 48, 200−209. (36) Xu, Y.; Lebrun, R. E. Investigation of the solute separation by charged nanofiltration membrane: effect of pH, ionic strength and solute type. J. Membr. Sci. 1999, 158 (1), 93−104.

8310

DOI: 10.1021/acs.iecr.7b02011 Ind. Eng. Chem. Res. 2017, 56, 8301−8310