Electrolytic membrane extraction enables production of fine chemicals from biorefinery sidestreams

Article pubs.acs.org/est

Electrolytic Membrane Extraction Enables Production of Fine Chemicals from Biorefinery Sidestreams Stephen J. Andersen,†,⊥ Tom Hennebel,†,‡,⊥ Sylvia Gildemyn,† Marta Coma,† Joachim Desloover,† Jan Berton,§ Junko Tsukamoto,§ Christian Stevens,§ and Korneel Rabaey*,† †

Laboratory of Microbial Ecology and Technology, Ghent University, Coupure Links 653, B-9000 Ghent, Belgium Department of Civil and Environmental Engineering, 407 O’Brien Hall, University of California, Berkeley, California 94720-1716, United States § SynBioC, Department of Sustainable Organic Chemistry and Technology, Ghent University, Coupure Links 653, B-9000 Ghent, Belgium ‡

S Supporting Information *

ABSTRACT: Short-chain carboxylates such as acetate are easily produced through mixed culture fermentation of many biological waste streams, although routinely digested to biogas and combusted rather than harvested. We developed a pipeline to extract and upgrade short-chain carboxylates to esters via membrane electrolysis and biphasic esterification. Carboxylaterich broths are electrolyzed in a cathodic chamber from which anions flux across an anion exchange membrane into an anodic chamber, resulting in a clean acid concentrate with neither solids nor biomass. Next, the aqueous carboxylic acid concentrate reacts with added alcohol in a water-excluding phase to generate volatile esters. In a batch extraction, 96 ± 1.6% of the total acetate was extracted in 48 h from biorefinery thin stillage (5 g L−1 acetate) at 379 g m−2 d−1 (36% Coulombic efficiency). With continuously regenerated thin stillage, the anolyte was concentrated to 14 g/L acetic acid, and converted at 2.64 g (acetate) L−1 h−1 in the first hour to ethyl acetate by the addition of excess ethanol and heating to 70 °C, with a final total conversion of 58 ± 3%. This processing pipeline enables direct production of fine chemicals following undefined mixed culture fermentation, embedding carbon in industrial chemicals rather than returning them to the atmosphere as carbon dioxide.

■

INTRODUCTION Both microbial fermentation and petrochemical production can generate industrially valuable building block chemicals. The petrochemical industry dominates chemical production because of the relatively low cost of the oil and gas substrates and wellentrenched separation and recovery technology, such as liquid− liquid extraction, adsorption, distillation, and more. So-called biorefineries can utilize microorganisms in fermentation processes to generate sustainable products from biological resources. Genetically modified organisms and single-strain cultures can generate products with high specificity at relatively high production rates, but such processes are cost-intensive and thereby limit the application to high-value products only. Mixed culture fermentations can be operated at lower costs on more complex feedstocks, including various bioindustrial wastes and syngas, but generally without specificity at biologically constrained rates.1−5 Example products include acetate, butyrate, caproate, and other short-chain carboxylates, common intermediates in the anaerobic digestion of organic wastes to biogas. These are industrially valuable building block chemicals, though often not valuable enough to overcome the cost of recovery from low titer broths. The separation and recovery of © XXXX American Chemical Society

fermentation products, even from highly specific pure cultures, can account for over 60% of the total plant costs, thus providing an incentive for low cost separation technologies.6 Alternatively, biogas can be generated from a broad range of organic waste, and though it is significantly less valuable, it is easily extracted and utilized, a key factor in the success of anaerobic digestion. Separation and recovery processes tailored to specific compounds and low titer broths can improve the competitiveness of biorefineres for both pure and mixed culture fermentation processes.7,8 We present a bioproduction alternative to digestion leading to nonfuel outcomes in a biorefinery context, a processing pipeline to recover and upgrade the ubiquitous short-chain carboxylate fermentation products as high value esters. The processing pipeline extracts carboxylates directly from a broth and upgrades SCFAs to esters in a second step. Membrane electrolysis (ME) generates a concentrate stream that is delivered to a reactive extraction Received: January 28, 2014 Revised: May 12, 2014 Accepted: May 21, 2014

A

dx.doi.org/10.1021/es500483w | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

Article

Figure 1. Schematic for carboxylate extraction from a fermenter by membrane electrolysis and biphasic esterification. Fermentation generates short chain carboxylates that are driven across the anion exchange membrane by an applied current and are protonated. Alcohol and heat react with the carboxylic acid in the water exclusion phase to generate esters.

step, biphasic esterification (BE) (Figure 1). The ME process is a two-chamber electrochemical water treatment step in which the fermentation broth or carboxylate-rich stream is electrolyzed by a cathode and the applied electrical potential draws carboxylate ions across a single polymeric anion exchange membrane (AEM) into a clean, highly saline, and low pH anolyte, without solids nor biomass. Ion exchange membranes are designed for selective transport of ions, and an AEM is capable of excluding solids, microorganisms, and uncharged molecules larger than the effective pore size of the membrane. A number of carboxylate ions have been demonstrated to cross AEMs: for example, synthetic lactate salts.9,10 ME stands apart from conventional electrodialysis and bipolar electrodialysis in which the treated streams are not engaged in the electrochemical reactions. In addition, only a single membrane is required for ME, whereas electrodialysis requires many.11 The ME process generates hydroxide ions and hydrogen gas in the broth (cathodic electrolysis) and protons and oxygen gas in the extractant (anodic electrolysis). The hydroxide ions can replace caustic soda management of a fermentation broth. In the anode chamber, the extracted ionic fermentation product is protonated to its acidic conjugate by electrolytically generated protons. This product is then esterified in the BE step, increasing the volatility of the product for separation. Conventional esterification is generally performed in the complete absence of water or in organic solvents, as water kinetically constrains the esterification reaction.12 The BE process consists of an unreactive water exclusion phase (e.g., solvent or ionic liquid) and an extractant (anolyte) phase. Esterification proceeds by excluding the water component following the addition of excess alcohol and heat. In a fully realized process, the higher-value, volatile esters are extracted from a low-affinity, nonvolatile, water-exclusion layer, such as an ionic liquid, and the aqueous extractant is returned to the anode to accrue more fermentation product. The objective of this research was to present and demonstrate a viable processing pipeline for extracting and upgrading short-chain carboxylates. As presented, ME and BE are two interdependent processes that enable ester production directly from fermenter broths. Acetate was used as the model

carboxylate for the ME and BE pipeline because acetate is among the most common fermentation products. The ME process was demonstrated at low current density with a lowcost membrane and then with a higher cost, proton-excluding membrane at a greater range of current densities. The anode salinity was investigated as a means of maximizing carboxylate flux in broths in which the relatively low concentration of product results in low Coulombic efficiency extraction. The ME and BE pipeline was ultimately demonstrated with an acid concentrate extracted from biorefinery thin stillage, and with this concentrate, BE was demonstrated with xylene and a xylene/ionic liquid mixture. The ME and BE pipeline successfully extracted acetate from biorefinery thin stillage and upgraded the product to methyl acetate and ethyl acetate.

■

MATERIALS AND METHODS Electrochemical Cells. Electrochemical cells (two chambers separated by a membrane, each with internal dimensions 80 × 80 × 20 mm) were used as previously described,13 with spacer material (ElectroCell A/S, Denmark) between the surface of the electrode and the AEM. Membranes (AM-7001 Anion Exchange Membranes, Membranes International Inc., Ringwood, NJ; and fumasep FAB, FumaTech GmbH, Germany) were pretreated in accordance with manufacturer specifications. The anode was an Ir MMO-coated titanium electrode (IrO2/TaO2; 0.65/0.35), 80 mm × 80 mm, with a centrally attached, perpendicular current collector (Magneto Special Anodes BV, The Netherlands). The cathode was a stainless steel wire mesh with a stainless steel sheet metal current collector. For high-current, high-concentration experiments only (Supporting Information (SI) Figure S2), a lowvolume two-chamber electrochemical cell was used (internal dimensions 70 × 10 × 20 mm with 50 × 10 mm effective membrane area) with a single pass of electrolytes at 1.54 L d−1 per chamber. The anode was an Ir MMO-coated titanium electrode (IrO2/TaO2; 0.65/0.35), 50 mm × 20 mm (Magneto Special Anodes BV, The Netherlands), and the cathode was stainless steel, 60 mm × 40 mm. The anolyte consisted of sodium sulfate, corrected to pH 2 with sulfuric acid. The synthetic catholyte consisted of a sodium acetate, sodium butyrate, or sodium caproate solution (SigmaB

dx.doi.org/10.1021/es500483w | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

Article

Carboxylate concentrations were measured by gas chromatography (GC), as described in the Supporting Information. All ME experiments were performed in triplicate, at a minimum. Biphasic Esterification. Esterification experiments were performed with both a synthetic anolyte and a real (i.e., generated in a previous experiment) anolyte. The synthetic anolyte consisted of 20 g L−1 acetate (as acetic acid, pH 2) and 0.5 M Na2SO4. In all cases, 20 mL of anolyte was used with a xylene (Sigma-Aldrich, Belgium) solvent layer of 20 mL in a 100 mL bottle for approximately 60 mL of headspace. The layered liquids were vacuumed to a slight under-pressure and heated to 70 °C. At t = 0 h, 5 mL of either ethanol, methanol, or distilled water (control) was injected into the biphasic solutions. The reactors were then stirred and maintained at 70 °C for 20 h. The aqueous fraction was sampled at t = 0 and t = 20 h and analyzed for carboxylate concentration by GC. The GC conditions are described in the Supporting Information. These BE experiments were performed in quadruplicate. Biphasic esterification was also performed with an ionic liquid for water exclusion. The ionic liquid was trihexyltetradecylphosphonium bis(trifluoromethylsulfonyl) amide (Sigma-Aldrich, Belgium). Real anolyte (6 mL), xylene (2 mL), and ionic liquid (1 at 1.07 g/mL) were heated to 70 °C and stirred in a high-pressure flask (15 mL). Ethanol (room temperature, 1 mL) was added at t = 0, and the xylene was tested for esters after t = 5 min, 30 min, 3 h, 18 h, and 24 h by GC/MS, as described in the Supporting Information.

Aldrich, Belgium), corrected to pH 5.5 with sulfuric acid prior to operation. Fermenter broth from bioethanol production was sourced from the thin stillage of Alco Bio Fuel NV (Gent, Belgium). To avoid clogging within the laboratory scale connectors and to allow consistency in the reactor design, this broth was centrifuged at 10 000 rpm for 10 min, and suspended solids were removed. This broth was dosed with sodium acetate to bring the concentration to 5 g L−1 acetate to simulate conditions after carboxylate fermentation. For a consistent concentration of acetate, active fermentation was avoided by centrifugation of the thin stillage, a low residence time in the cathode chamber, and daily replacement of the feed. The initial concentration of acetate in the broth was only 0.7 g L−1 because the fermentation process is focused on bioethanol production. In all tests, sodium sulfate salt was used in the electrolyte rather than sodium chloride to prevent the formation of chlorine gas at the anode while increasing the conductivity. For continuous experiments, the electrochemical cells were fed at a rate of 0.8 L d−1 to achieve a residence time of 5 h. The anode and cathode batch experiments were performed with equal volumes of electrolyte at 1 L. Both compartments were continuously stirred at a recirculation rate of 6 L h−1. All electrochemical experiments were controlled with a VSP multipotentiostat (Princeton Applied Research, France) and an Ag/AgCl reference electrode (+0.197 V vs SHE, Princeton Applied Research, France) in the cathode compartment. The applied current is reported as current density, defined as the set current divided by the exposed surface area of the anion exchange membrane. The anode reaction was the oxidation of water (2 H2O → 4 H+ + O2 + 4e−), which replenishes the sulfuric acid/sulfate medium. The cathodic reaction was the reduction of water (2 H2O + 2e− → H2 + 2 OH−). Membrane Electrolysis. ME was tested with continuously fed synthetic solution for flux characterization, as further described below. Acetate, butyrate, or caproate in synthetic broths were tested for flux through an AEM at various current densities. For two synthetic experiments with acetate, the anolyte was altered. In one experiment, the concentration of sodium sulfate was increased in the anode to observe the trend of the flux of acetate relative to anode salinity. Similarly, the concentration of sodium acetate at pH 2 was increased in the anode to observe the trend of the acetate flux relative to acetate accumulation as acetic acid. For these experiments, the fluxes were tested at 20 A m−2 only. Next, the process was tested with thin stillage from a bioethanol plant to demonstrate the extraction process with a real matrix and native ions. By using this broth, we were able to demonstrate the full pipeline from a consistent and practical source fluid. Both the cathode compartment (broth) and the anode compartment (extractant) were first tested in a batch extraction. The cathode was then continuously fed while the anode compartment was run in batch to better mimic realistic operating conditions (accumulation in the anode, excess broth relative to extractant). For these experiments, the current remained set at 20 A m−2. In all measurements of flux, the cell was set at 1.6 mA m−2 (a galvanostatic setting of −0.1 mA controlled at the cathode) for at least 4 h prior to the run to allow for electrode and membrane polarization, then was changed to the set current density of interest. Hydroxide ions generated by water reduction were mitigated by 1 M sulfuric acid by a pH controller (Consort, Belgium).

■

RESULTS Membrane Electrolysis Extraction. In the first phase, we demonstrated the basic ME concept on synthetic solutions with various concentrations of the model carboxylate, acetate (Figure 2a,b), and other carboxylates of interest, butyrate and

Figure 2. Extraction profile for acetate across a Membrane International AMI-7001 anion exchange membrane (MI) from a starting catholyte concentration of 1, 5, and 10 g/L acetate. (A) Mass flux per area of membrane per day, (B) Coulombic efficiency for the same data points. The error bars are occasionally obscured by the medallions. C

dx.doi.org/10.1021/es500483w | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

Article

Figure 3. Coulombic efficiency (molar flux of acetate/mol of applied current) can be improved with membrane selection and diffusion of anions against the applied current. Experimental changes in efficiency of two different anion exchange membranes versus anolyte concentration of Na2SO4 (left). Illustration of the mechanism (right).

caproate (SI Figure S1), from 0 to 30 A m−2 projected membrane surface. For an initial catholyte concentration of 10 g L−1, the flux of acetate was 1.05 kg m−2 d−1 at 20 A m−2, corresponding to a Coulombic efficiency of 99.4 ± 0.1%. At 30 A m−2, the maximum flux of acetate was 1.38 ± 0.03 kg m−2 d−1, corresponding to a Coulombic efficiency of 87.0 ± 1.8%. The extraction efficiency was lower for 1 and 5 g L−1 acetate broth, corresponding to less availability of the target anion in the bulk (molar concentration), and similarly, butyrate and caproate show a low flux rate and Coulombic efficiency (SI Figure S1). Transport limitations that manifest as membrane resistance and ion transport number have been previously demonstrated for low-concentration salt solutions.14 A target molecule at high concentration will flux efficiently at low current densities, as was seen with acetate in Figure 1, and at high current densities for both acetate and caproate, as shown in SI Figure S2. High-current-density extraction (and therefore, higher flux) is preferable in a fully realized system. Acetate and caproate were tested at equal molar concentrations (170 mM: 10 g L−1 acetate and 19.6 g L−1 caproate) and display good extraction for current densities up to 100 A m−2 (SI Figure S2), although caproate did not show efficiency equal to acetate at a current density of