Key words

separations; biorefineries; membranes; resin-wafer-eletrodeionization

5.1 Introduction

Separations technologies play an important role in integrated biorefineries. Separations may account for ~ 50% of the total production costs in bioprocessing operations (Hestekin et al., 2002). Therefore, designing and implementing efficient separations strategies are critical factors in successful bioprocessing. In this chapter we provide an overview of separations technologies commonly employed in existing and future biorefineries. Separations are required for feedstock preparation, product recovery and purification, removal of impurities, and water management, recovery, and reuse. We focus on the chemical and biochemical aspects of separations technologies in liquid process streams.

From a thermodynamic standpoint, separations are inherently energy intensive, and therefore, strategies must consider energy demand to achieve the targeted separations. Based on the Energy Independence and Security Act, EISA 2007 (US Government Printing Office Pub L. 110−140), biofuels must achieve specific targets for reduction in greenhouse gas emissions (GHGs) in the US. Conventional biofuels must achieve a 20% reduction in GHGs, advanced biofuels must achieve a 50% reduction, and cellulosic biofuels must achieve a 60% reduction. Reducing energy use in biorefining in general, and separations in specific, is critical to meeting these mandates. Europe has GHG reduction mandates that also warrant reducing energy use.

Separations are based on physical and chemical differences between species as well as the nature of the mixture. Typical physical factors include size, shape, compressibility, density, and viscosity. Chemical factors include solubility, hydrophobicity/hydrophilicity, polarity, pK_a, boiling and freezing points, and specific molecular interactions. Critical decisions for selection of separations strategies are based on the following questions:

There are several critical factors that help define the most efficient separations scheme. As purity requirements increase, operation scale, energy consumption, and total costs increase exponentially. Therefore understanding purity requirements is essential in designing the separations scheme. The critical first decision to consider is the order of the separations scheme. Should a dilute solute be separated from solvent or should solvent be stripped from the dilute species or solute? Should separation schemes capture several chemically similar targets and then separate them with more specific processes? Should the separation scheme select individual species from the process stream? These decisions are based on knowledge of the process stream and targeted species, purity requirements, species stability, and available technologies. Within technologies important parameters include system footprint, substrate requirements, waste discharge requirements, and total throughput and process time.

Separations require both a method for speciation and a driving force. Speciation creates the materials needs and driving force creates the energy demand. Some common separations technologies include:

Most chemical and biological processes involve multiple reactions and separations processes. With design, integration of a specific reaction with product separations can increase overall performance of both conversion and recovery. Ultimately, choice of separations trains in integrated biorefineries is based on economics. The economics is defined by purity requirements, co-product uses, waste disposal routes, energy and substrate use, capital and operating costs, and system footprint.

There are entire textbooks written on most of these subjects (Coulson et al., 1991; Green and Perry, 2007; Tarleton and Wakeman, 2008; Seader et al., 2010). For the sake of brevity, we provide a few examples of the challenges in chemical and biochemical engineering associated with separations in biorefinery operations. We focus on emerging technologies that we expect to exhibit rapid growth. Specifically, we focus on membrane-based separations because they are becoming the separations platform of choice. In addition, we provide examples where electrochemical-driven processes could significantly reduce energy consumption during separations.

5.2 Separations technologies

5.2.1 Membrane separation technologies

Membrane-based separations technologies are becoming more widely deployed in integrated biorefinery operations due to their versatility, separations efficiency, energy savings, and economic benefits (Ho and Sirkar, 1992; Mulder, 1996; Cheryan, 1998; Baker, 2004; Li et al., 2008). They are used in the food, pharmaceutical, biotechnological, bioprocessing, and chemical industries. A membrane is a porous, semi-permeable separation medium that fractionates different species from a solution based on size, shape, solubility, or molecular interactions. The permeate solution containing the ‘smaller’ species penetrates through the membrane, whereas, the retentate solution containing the ‘larger’ species is rejected by the membrane (Fig. 5.1a). Membranes are fabricated from many materials including inorganics such as alumina or silica or organics such as polyethersulfone, polyamides, or cellulose acetate. Membranes are commercially available in different module formats, including tubular, hollow fiber, flat sheet, spiral wound, etc. Membranes can be fabricated with pore diameters ranging from < 1 nm (virtually non-porous) to 10 μm.

Based on pore size, membrane-based processes are classified as presented in the filtration spectrum (Fig. 5.1b). From a mixture, a particular membrane-based process rejects the species mentioned on the retentate side and allows permeation of the species mentioned on the permeate side of Fig. 5.1(b). Examples of the application of membranes based on pore size are:

• separation of activated carbon from sugar – conventional filtration

• separation of biological cells from proteins – microfiltration (MF)

• separation of proteins from salt – ultrafiltration (UF)

• separation of sugar from salt – nanofiltration (NF)

• separation of salt from water (desalination) – reverse osmosis (RO).

From the right of the filtration spectrum (conventional filtration) to the left (RO), the transmembrane pressure drop required for solvent flux increases due to decreasing membrane pore size. All of the processes from conventional filtration through RO operate under the driving force of chemical potential in the form of either concentration or pressure gradient. Membrane-based processes are rate governed unlike processes like distillation, which are equilibrium governed.

There are some other membrane-based processes, such as electrodialysis (ED), electrodeionization (EDI), pervaporation (PV), vapor permeation (VP), membrane distillation (MD), supported liquid membranes (SLM) that are used frequently, but not included in the filtration spectrum. Among them, ED and EDI are charge-based membrane separations processes that operate under the driving force of electrochemical potential and separate charged species from uncharged species or fractionate multi-charged species. PV and VP operate under the driving force of chemical potential and fractionate organic/water mixtures with the help of a permselective (non-porous for all practical purpose) membrane. The permeate transports across the membrane in the gas phase. Membrane-based processes that are relevant in integrated biorefineries are described below.

Size (or solubility)-based membrane separations

MF and UF are the two widely used membrane filtration processes in biorefineries. The pore diameters of the membranes are in the range of 2 nm to 50 nm for UF and 50 nm to 5 μm for MF. Choice of membrane material and pore diameter depends on characteristics of the species present in the solution. MF is used for filtering coarse materials in biorefineries and often used as a pre-filtration step to the downstream UF. As an example, 0.45 μm polyethersulfone MF membrane (hollow fiber module) was used for removing cellular debris from the fermentation broth of a recombinant enzyme, glucose fructose oxidoreductase (GFOR). The permeate of the MF was then filtered through a 30 kDa molecular weight cut-off (MWCO) UF membrane to remove the nutrients (sugar and salts) in the permeate and collect GFOR in the retentate. The purified stream of GFOR was then used to convert sugar to biobased chemicals as described in detail in a later section. Membrane-based separations of biomolecules are susceptible to severe membrane fouling due to adsorption of biomolecules within membrane pores. A fouled membrane will exhibit dramatically declining permeate flux. A proper clean-in-place (CIP) procedure with suitable solvents and/or surfactants is necessary to restore permeate flux and reuse the fouled membrane. For example, a cleaning solution consisting of 0.2% sodium hydroxide, 25 g/L sodium chloride, and 0.2% Triton X-100 non-ionic surfactant is typically used for regenerating membranes subjected to biological fouling.

RO is typically used for desalination of process water for treatment and reuse, and, therefore, is an important part of biorefineries. In RO, a transmembrane pressure, higher than the osmotic pressure of the solution in the feed side, is used to transfer water preferentially over the solute. This results in a purified water stream on the permeate side of the membrane. A transmembrane pressure as high as 50–80 bar is required for RO, which makes it unattractive to many users. In PV, the membrane acts as a permselective barrier between a liquid and a vapor phase. PV is commonly used to recover alcohols from very dilute aqueous solution as an alternative to energy-intensive distillation. Alcohol is preferentially solubilized in a membrane, diffused through the membrane, and then desorbed as vapor due to the applied vacuum on the permeate side. This results in an alcohol-enriched vapor phase starting from a lean solution of alcohol. An inert sweep gas is used to maintain the driving force on the permeate side.

Charge-based membrane separations

ED and EDI are membrane-based processes that remove ions from solution under an applied electric potential. ED is commercially used for water desalination, water treatment and reuse, and organic acid recovery. ED consists of successive alternative arrangements of cation exchange and anion exchange membranes within two electrodes to form two different types of compartments, diluate and concentrate. A schematic of the configuration of different components present in a typical ED set-up is represented in Fig. 5.2. When an electrolyte is fed through the diluate compartment in the presence of an applied electric potential, cations move towards the cathode, transport across the cation exchange membranes, enter the concentrate compartment, and accumulate there due to the impermeable anion exchange membrane on the other boundary. Similarly, the anions move from diluate to concentrate compartment through anion exchange membranes. So, ED facilitates depletion of ions from diluate solution and accumulation of ions in concentrate solution. An electrolyte is recirculated on the two end electrode chambers to maintain the continuity in current flow.

Electrodeionization (EDI) is a modified version of ED that contains conductive ion exchange (IX) resin beads within the diluate compartment (Fig. 5.3). EDI combines the advantages of ED and IX chromatography. It utilizes in-situ regeneration of the IX resin beads by a phenomenon known as ‘water splitting’. Water splitting on the surface of the IX resin beads regenerates the beads and ensures higher ionic conductivity within the diluate compartment. EDI outperforms ED with dilute solutions, where due to the limited ion concentration, ionic conductivity decreases and electrical energy is wasted in water splitting. In contrast, the conductive IX resin beads in EDI provide sufficient ionic conductivity, even with a dilute solution, and provide an efficient ion transport pathway through the IX resin beads. In conventional EDI, loose IX resin beads are used; however, the researchers at Argonne National Laboratory have improved the technology by using resin wafers (RW) to incorporate the loose ion exchange resin. The modified platform is called RW-EDI. Argonne patented the technology to fabricate and use the resin wafers (Lin et al., 2005a, 2008). The technology offers enhanced flow distribution, higher conductivity, superior pH control, ease of materials handling and system assembly, and a porous solid support for incorporation of catalysts, biocatalysts, and other adjuvants. RW-EDI is used for production and recovery of biobased chemicals (Arora et al., 2007), especially organic acids from fermentation broth (Lin et al., 2005b), post-transesterification glycerin desalting (Datta et al., 2009), conditioning of biomass hydrolysate liquor (Datta et al., 2013), and for CO₂ capture from flue gas (Lin et al., 2013). Argonne deploys three different ED stack sizes (Fig. 5.4) to design experiments and evaluate performance from fundamental and exploratory scale research through pilot-scale and field deployment. A small stack (14 cm² surface area) is used for proof of concept experiments, which is then scaled up to a TS2 stack (195 cm² surface area), for process optimization and more rigorous studies. At the pilot scale (1700 cm² surface area), extended campaigns are conducted to evaluate potential for commercialization.

5.3 Removal of impurities from lignocellulosic biomass hydrolysate liquor for production of cellulosic sugars

Lignocellulosic biomass can be converted to biofuels by biochemical conversion (Fig. 5.5). Enzymatic hydrolysis is used to convert the cellulose component of the lignocellulose to soluble, monomeric sugars (primarily glucose), which are then fermented to biofuels (Aden et al., 2002). Lignocellulose, primarily composed of cellulose, hemicellulose and lignin is a rigid, crystalline structure not directly amenable to enzymatic hydrolysis. Pretreatment is required to disrupt the crystalline structure of the cellulosic moieties and expose them for the downstream enzymatic hydrolysis step. The most common pretreatment technique is dilute acid treatment with sulfuric acid at elevated temperature and pressure. The heat generates unwanted degradation products and impurities (organic acids, furans, phenolics, etc.) along with the solubilized C5 sugars (primarily xylose from hemicellulose) in the hydrolysate liquor (Tucker et al., 2003). These impurities are detrimental to downstream enzymatic hydrolysis and fermentation. The acidity of the hydrolysate liquor (e.g., by sulfuric acid and organic acids) also inhibits enzymatic hydrolysis and fermentation. Hence, removal of these byproducts and reagents is essential for further processing of the insoluble biomass component (cellulose and lignin) that will lead to an efficient biofuels production strategy (Biomass Multi Year Program Plan, 2008). The major impurities of dilute acid pretreatment are acetic acid, produced by deacetylation of hemicellulose, furfural, produced by degradation of C5 sugars, and hydroxylmethyl furfural (HMF), produced by degradation of C6 sugars. Acetic acid and sulfuric acid are ionic impurities, and therefore, could be separated based on their ionization. The non-ionic impurities, such as furfural and HMF, could be removed using adsorption techniques. Ionic impurities from corn stover hydrolysate liquor can be separated using RW-EDI followed by the removal of non-ionic impurities using adsorptive beads as described below.

5.3.1 RW-EDI for removal of ionic impurities

Biomass hydrolysate liquor and the residual solid biomass are first separated by feeding the pretreatment slurry into a solid liquid separator (unpublished data). In conventional systems, sulfuric acid is removed by treating with lime (over-liming) and precipitating sulfate ions as CaSO₄ (gypsum) as shown in Fig. 5.5. Over-liming removes sulfate ions from hydrolysate liquor, albeit increasing the number of unit operations and residence time (overliming tank + pH adjustment tank/settlement tank + separation devices) (Aden et al., 2002). It also requires addition of chemicals and generation of a low-value and potentially toxic byproduct, gypsum. The flash vaporization prior to solid-liquid separation removes only 8% of the total acetic acid. Further reduction in acetic acid concentration is accomplished by aqueous dilution of the hydrolysate liquor.

This is an example of using emerging technology to remove ionic impurities, sulfuric acid and acetic acid. We treat corn stover hydrolysate liquor using resin wafer electrodeionization (RW-EDI) technology. RW-EDI-based deacidification provides a pathway for conditioning of biomass hydrolysate liquor with fewer unit operations, lower operation time, and reduced use of chemicals and water. Similar to ED, the RW-EDI stack is configured for ion transport from one compartment (diluate) to another (concentrate) by alternative arrangement of cation exchange and anion exchange membranes within two electrodes (Fig. 5.6). Diluate compartments contain ion exchange resin beads within a porous solid matrix, called a resin wafer (Lin et al., 2005a, 2008) for enhanced conductivity, particularly for dilute solutions. The electrically-driven membrane process facilitates depletion of ions from one solution (feed solution within the diluate compartment) and accumulation of ions in another solution (recovery solution within the concentrate compartment) under an applied electric field. Depending on operating conditions, RW-EDI can separate the sulfate and acetate present in the corn stover hydrolysate liquor either together or sequentially. Using RW-EDI, > 99% sulfuric acid and > 95% of acetic acid were removed. For the neutral xylose sugar, > 98% was retained (Fig. 5.7) (Lin et al., 2013). By adjusting the operating conditions, selective separation of sulfuric acid and acetic acid was achieved to obtain two separate acid-enriched streams. For a typical case, the sulfuric acid-enriched stream contained around 20 g/L of sulfuric acid and 1 g/L of acetic acid. On the other hand, the acetic acid-enriched stream contained around 0.5 g/L of sulfuric acid and 9 g/L of acetic acid. The sulfuric acid stream could be recycled back for the dilute acid pretreatment, while the acetic acid stream could be recovered as a value-added biobased co-product.

5.3.2 Adsorptive beads for removal of non-ionic impurities

Polymeric ion exchange resin beads (Nilvebrant et al., 2001) and polymeric adsorptive beads (Weil et al., 2002) have been used for the removal of fermentation inhibitory compounds (both ionic and non-ionic) from biomass hydrolysate liquor. Here, an example is presented where adsorptive beads-based removal of non-ionic impurities, such as furfural and HMF, was used as a polishing step to the RW-EDI as shown in Fig. 5.5. Furfural and HMF were removed through hydrophobic interaction with the commercially available functionalized adsorptive beads. Comparative results for removal of furfural and HMF from RW-EDI treated corn stover hydrolysate liquor using different types of commercially available resin beads are shown in Fig. 5.8. Dowex L-493 appears to be superior to the other resin beads primarily due to their very high surface area (~ 1100 m²/g). It removed all furfural and HMF from the RW-EDI-treated corn stover hydrolysate liquor. The RW-EDI-treated and Dowex L-493-polished corn stover hydrolysate liquor is a pure sugar solution containing 40 g/L xylose. The other resin beads that were evaluated are less efficient compared to L-493, as shown by the percentage removal and amount adsorbed (mg/g of beads) values.

5.4 Glycerin desalting as a value added co-product from biodiesel production

The annual production of biodiesel in the United States is increasing from 75 million gallons in 2005 to an EISA 2007 mandate of 1 billion gallons (http://www.biodiesel.org/production/production-statistics). Biodiesel is primarily produced by transesterification of triglycerides (vegetable oil or fat) with an alcohol (methanol or ethanol) using inorganic catalyst (acid or base) in homogeneous phase (Fig. 5.9) (Noureddini and Zhu, 1997). The principal co-product of biodiesel production is glycerin (glycerol) (Thompson and He, 2006). Typically, 100 kg of vegetable oil is reacted with 10 kg of methanol to yield 100 kg of biodiesel and 10 kg of glycerin. The process generates inorganic salts, such as NaCl or KCl, as impurities in the glycerin phase. Removal of these impurities is critical to transform the crude glycerin into a high value, purified glycerin for use in the bioprocessing, food, pharmaceutical, and cosmetics industries. Purified glycerin from biodiesel production facilities could also be used as a starting material for useful chemicals as summarized in Fig. 5.10. The existing technologies for refining crude glycerin, such as distillation and chemical additions, are inadequate and energy intensive. Therefore, an energy efficient and cost-effective glycerin purification technique is necessary to enhance the sustainability and economics of biodiesel production.

RW-EDI could remove salt impurities from the crude glycerin layer (70–75% glycerin, 20–25% methanol, 4–5% inorganic salts and 1–2% water by weight). Crude glycerin solution is fed through the diluate compartment, while the salt recovery solution is fed through the concentrate compartment of the RW-EDI stack as demonstrated in Fig. 5.11. Under an applied electric potential, salt ions (e.g., Na⁺ and Cl⁻) from the feed crude glycerin solution transport towards the respective electrodes, cross the ion exchange membrane barrier, and accumulate in the salt recovery solution. As a result, the glycerin solution becomes depleted of salt and the recovery solution becomes enriched in salt, thereby providing a purified glycerin solution at the diluate exit. A typical batch experiment with simulated crude glycerin solution (1 L) exhibits more than 99% removal of NaCl as demonstrated in Fig. 5.12. There is insignificant loss of glycerin from the feed solution. This technique could be improved further by optimizing process parameters for enhanced efficiency and cost effectiveness.

5.5 Succinic acid production

RW-EDI enables integration of the upstream bioreactor and downstream product separation for organic acids (Fig. 5.13). Direct recovery of organic acids from fermentations offers two immediate benefits: prevention of reactor acidification and avoidance of product inhibition. Rapid organic acid separation during production offers a new way to control reactor pH without the addition of pH buffers. We used succinic acid as a model organic acid to demonstrate the efficiency of the integrated RW-EDI for production and recovery of organic acids from fermentation broth.

Succinic acid is an important building-block chemical feedstock for polymers with wide applications (acyl halides, anhydrides, esters, amides, and nitriles for applications in drug, agriculture, and food products). Succinic acid is considered one of the most highly valued biobased products as a replacement for fossil-based maleic anhydride or butanediol (Werpy and Peterson, 2004). Succinic acid can be produced by fermentation of sugars from different feedstreams by microorganisms, but due to the need for neutralization or pH control during this process, succinate salt is typically the direct product. Purification to the acid form, which is required for further conversion, is expensive and often requires further processing (Fig. 5.13) to meet purity requirements, leaving a solid gypsum waste product.

A mutant strain of Escherichia coli, AFP184 (pfl-, ldh-, and ptsG-), was used for the fermentation of glucose to succinic acid (Donnelly et al., 2004). This strain contains mutations which enable the conversion of glucose to primarily succinic acid with minor amounts of acetic acid and ethanol. The RW-EDI stack was connected to the succinic acid fermentation tank. Fermentation broth was pumped from the fermentation tank to the RW-EDI (diluate compartment) to remove the organic acid salts from the broth. The organic acid salt was converted in situ to the acid in the RW-EDI using a configuration containing an alternating arrangement of bipolar membrane and anion exchange membrane (Lin et al., 2011). The anion exchange membrane allows only transfer of succinate ions from diluate to concentrate compartment. The bipolar membrane splits water electrochemically (water splitting phenomenon) on its surface and produces hydroxyl ions and protons in the diluate and concentrate compartments, respectively. Protons in the concentrate compartment combine with the transferred succinate and form succinic acid. Hydroxyl ions in the diluate help maintain the basic pH required to dissolve CO₂ as bicarbonate ion before the broth is returned to the fermenter. The recovered succinic acid was collected from the concentrate (product) stream in the RW-EDI. The succinic acid was biologically formed by reaction of glucose and CO₂.

The use of the integrated RW-EDI provides several benefits to succinic acid production. Using water splitting EDI, the succinate in the broth was simultaneously removed and converted to the acid form in the recovery stream. The alkalinity in the organic acid-depleted broth was used to dissolve the CO₂ before the broth returned to the fermentation tank. Figure 5.14 illustrates the configuration of the RW-EDI for succinic acid fermentation. Figure 5.15 shows succinic acid production using the RW-EDI system. Succinic acid production was increased significantly over the byproduct acetic acid. Figure 5.16 shows the concentrations of the product and byproducts in the recovery product tank. The RW-EDI provided an efficient strategy for continuous, integrated production and separation of succinic acid.

5.6 Solvent extraction: the example of recovery of value added proteins from distiller’s grains and solubles (DGS)

The US produces ~ 14 billion gallons of ethanol per year, predominantly in corn dry mills, and generates distiller’s grains and solubles (DGS) as the main co-product. A bushel of corn (56 lb) yields around 16 lb of DGS, or 5.4 lb of dried DGS (DDGS) per gallon of undenatured ethanol (Arora et al., 2008). Compositional analysis of DGS reveals significant amounts of protein (30%), polysaccharide (18% glucan, 10% xylan and 6% arabinan) and fat (11%) (Datta et al., 2010). DGS is primarily used as an animal feed for ruminants such as cattle; however, due to high fiber content, it is unsuitable for monogastrics, such as hogs and poultry. Besides that, due to high moisture content (50–60%), it is dried and transported to storage when there is a lack of immediate and local market demand. Drying DGS to produce DDGS is energy intensive and requires about one-third of the energy consumed in ethanol dry mills. Typically, the whole process of drying DGS to DDGS consumes 4.77 MJ of energy in the form of natural gas and 0.5 MJ of energy in the form of electricity per kg of DDGS (McAloon et al., 2000).

To address these challenges, we evaluated a new separations process to extract protein from DGS using biobased solvents and hydrolyze the unrecovered cellulosic sugars using enzymatic saccharification (Fig. 5.17) (Datta et al., 2010). The goal was to extract a high value animal feed with low fiber content and produce additional sugar for ethanol production. Ammonia fiber expansion (AFEX)-treated DGS (Bals et al. 2006) was used because it has a more vulnerable fiber structure than the regular DGS, which enables easy access of the targeted components (protein, cellulosic materials) by the reaction mixture (biobased solvent and cellulases). The protein-rich fraction with residual biobased solvent phase could be used as a high value animal feed. The biobased solvents are food grade materials and increase the nutritional value (fatty acids) of the animal feed. The concentrated protein stream would reduce the energy required associated with drying DGS. The hydrolyzed sugar would enhance the overall ethanol yield. Overall, the process economics of the corn to ethanol production would be improved by higher ethanol yield, reduced energy inputs, and increased value of the animal feed co-products.

The solvents used were citrus-derived D-limonene (DL), soybean-derived distilled methyl esters (DME) and corn starch-derived ethyl lactate (EL). They were formulated and supplied by VertecBioSolvents (www.vertecbiosolvents.com). In the separate protein extraction study, up to 45% of protein was extracted from DGS using biobased solvents. The hydrophobic solvent DL was superior in extracting proteins in comparison to the hydrophilic EL. For the simultaneous protein extraction and enzymatic saccharification study, hydrophobic biobased solvents (DL and DME) were used along with two different enzyme recipes; cellulase (Accelerase) only, and a mixture of cellulase (Accelerase) and amylase (Stargen). Both enzyme recipes were efficient in hydrolyzing cellulose to glucose from AFEX-DGS as demonstrated in Fig. 5.18. A total of 50–80 mg glucose/gm of dry AFEX-DGS was obtained after 120 h of processing by enzymatic hydrolysis under different process conditions. Figure 5.18 also demonstrates the advantage of amylase (in the presence of cellulase) on enzymatic hydrolysis for different solvents. Approximately one-third of the glucan in DGS is starch, indicating the need for amylase in addition to cellulase (Kim et al., 2008).

Further experiments with a mixture of cellulase and amylase reveal an increase in glucose production with increase in time (Fig. 5.19). In comparison to blanks, the biobased solvents did not affect glucose production. This observation implies that the biobased solvents did not adversely affect enzymatic hydrolysis and could open a new avenue for aqueous phase enzymatic reactions in the presence of the biobased solvents. Protein analysis on the AFEX-DGS reveals extraction of around 30% protein from the solid phase after simultaneous extraction and saccharification. However, protein analysis on the biobased solvent and aqueous phases indicates a distribution of less than 10% of extracted protein in the solvent phase and the rest in the aqueous phase. This suggests an inefficient protein extraction by biobased solvents in the presence of the aqueous phase. We attributed this to mass transfer limitations. The hydrophilic biomass was completely surrounded by the aqueous phase, thereby shielding it from the hydrophobic solvent. Improved operating conditions (size reduction, addition of other chemicals) and process engineering (mixing conditions) are necessary to enhance the efficiency of the targeted simultaneous protein extraction and enzymatic saccharification. Nevertheless, this study evaluates an innovative pathway that could potentially lead towards value added corn co-products (animal feed + additional sugar) for improved corn to ethanol economics.

5.7 Biofuels recovery by solvent extraction in an ionic liquid assisted membrane contactor

Development of an efficient separations technique that will lead to an integrated fermentation-separations biofuel system will reduce energy use. A conventional separations process, such as distillation, is used to recover ethanol from fermentation broth (Fig. 5.20) containing 2–20% ethanol. Roughly, 19,000 BTU/gal of energy is consumed for distillation treatment of a 5% ethanol fermentation broth. Most of the energy is consumed to heat/vaporize the 95% water present in the broth. An energy saving alternative strategy would be to shift to separation of the 5% ethanol rather than the 95% water as observed in processes such as pervaporation (PV) or liquid–liquid extraction (LLE). PV consists of a permselective membrane that concentrates the minor component from a solution in the feed side to the vapor phase on the permeate side (under vacuum) as shown in Fig. 5.21. Because of its simplicity and lower energy consumption compared to distillation, PV is frequently used for dehydration of organic solvents, recovery of organics from aqueous solution and separation of azeotropic organic mixtures. In spite of the energy benefits, the low partition coefficient and slow permeation rate of alcohols across the PV membrane limit the application of PV.

Another well-known alternative to distillation for biofuels recovery is LLE by organic solvents (Wittenberg and Arana 2010; Aravani et al., 2010; van der Wielen and Heijnen, 2010). Different organic solvents, such as hexane, acetone, methanol, ethanol, higher chain alcohols, have been evaluated; however, hexane is the preferred solvent for industries. Other specialty solvents, such as supercritical CO₂, have also been reported for biofuels extraction (Bothun et al., 2003), but they have not been implemented at commercial scale. For LLE, typically a series of mixers/gravity settlers are employed (Fig. 5.22), where the solvent phase is dispersed within the aqueous phase, the alcohol is transferred from the aqueous phase to the solvent phase, followed by the separation of the two phases by gravity settling. There are several challenges in using LLE for alcohol recovery (Gawronski and Wrzesinska, 2000; Bothun et al., 2003):

To address the challenges associated with PV and LLE, we developed a method for LLE using a membrane contactor (Snyder et al., 2006). It contains a solvent on the permeate side (and within the membrane pores) that enhances the alcohol selectivity factor between the membrane and the fermentation broth and also improves the permeation rate of extracted alcohol through the membrane (Shukla et al., 1989; Stanojevic et al., 2003). We describe an emerging technology for alcohol recovery by liquid–liquid extraction in an ionic liquid assisted membrane contactor.

5.7.1 Ionic liquid (IL) assisted membrane contactor

Ionic liquids are solvents comprised entirely of cations and anions and are molten salts under atmospheric conditions (Fadeev and Meagher, 2001; McFarlane et al., 2005; Zhao et al., 2005). They are defined as salts whose melting point is below 100 °C. Due to their thermal and mechanical stability, lower vapor pressure and electrochemical properties, they are gradually gaining popularity as industrial solvents, electrolytes, and other specialty applications, such as CO₂ capture. ILs are available commercially. Their high costs have limited deployment in commodity applications. However, the IL-assisted membrane contactor offers the opportunity to minimize IL loss, and therefore, reduce IL operating costs and make them a viable option for commercial applications (Lin and Snyder, 2012). Membrane contactors are established separations systems for small footprint, energy-efficient recovery of target species in continuous mode operation. They consist of a porous membrane with an extracting solvent (IL in this case) on the permeate side and the process solution (fermentation broth) on the feed side as demonstrated in Fig. 5.23. A small pressure differential is applied to wet the membrane with the extracting solvent. Typically, a membrane module with high pore surface area, such as hollow fiber or honeycomb, is used as membrane contactor. Both the membrane material and extracting solvent provide chemical discrimination for the target. Recovery of biofuels by liquid–liquid extraction within an IL-assisted membrane contactor has several advantages:

1. High interfacial surface area in the hollow fiber or honeycomb membrane module significantly increases the liquid–liquid contact area and thus improves mass transfer.

2. The membrane acts as a permselective barrier to isolate the two phases and selectively transfer alcohol through the membrane pores to the solvent phase. It does not rely on density difference.

3. The in-line slipstream evaporation in the IL loop removes the extracted alcohol continuously and ensures a higher driving force for extraction.

4. Overall, the system has a smaller footprint, reduced processing time, and reduced risk of contamination of the fermentation.

5. Near zero vapor pressure of ILs minimizes solvent loss, and therefore significantly reduces make-up solvent requirements and minimizes thermal energy needed for alcohol recovery from the solvent.

6. The reduced solvent volume enables operation of the permeate side at elevated temperatures with reduced energy consumption. Keeping the extracting at elevated temperatures enhances both extraction of the alcohol in the membrane contactor and subsequent stripping from the solvent.

The IL-assisted membrane contactor is used for liquid–liquid extraction (recovery) of ethanol or butanol from a fermentation broth. The chemical characteristics of an IL can be tuned by proper choice of the cation and anion from a vast pool of candidates. We evaluated both hydrophilic and hydrophobic ILs for ethanol recovery. However, it was observed that the hydrophobic IL has higher solubility for ethanol. The IL is used on the permeate side to extract ethanol from the fermentation broth on the feed side (Fig. 5.24). A small positive pressure head is applied on the feed side to avoid contamination of the IL into the fermentation broth. The process relies on three steps: solubility (partitioning) of ethanol in the membrane, diffusion (transfer) of ethanol within the IL-assisted membrane pores, and dissolution (desorption) of ethanol from the permeate IL phase due to a concentration gradient. The raffinate (treated fermentation broth), depleted of ethanol, is recycled back to the storage tank or fermenter. The IL phase containing the extracted ethanol is fed to an evaporator to strip the ethanol from the IL phase. The IL is recycled back to the membrane contactor and the operation is conducted in a continuous mode to maximize the recovery. Countercurrent flow (aqueous and IL phases) is employed to maximize the concentration gradient throughout the length of the membrane module.

Experimental results for the recovery of ethanol from a simulated fermentation broth are presented. Hydrophobic 1-hexyl-3-methylimidazolium bis-trifluoromethylsulfonylimide, ([hmim][Tf2N]), was used for all experiments. The experiments were conducted in a continuous batch mode operation without the evaporator, hence an accumulation of ethanol in the IL phase is observed. Figure 5.25 represents the separation factor for three different cases. It is evident from Fig. 5.25, that the IL-assisted membrane contactor has enhanced the separation factor (for both hydrophilic and hydrophobic membranes) compared to the IL without any membrane. The separation factor is defined as the ratio of the concentration of ethanol to water in IL phase divided by the ratio of the concentration of ethanol to water in aqueous phase. The separation factor has increased from 37 without the membrane contactor to more than 115 with the membrane contactor. This could be attributed to the membrane, which acts as a barrier between the two phases, thereby enhancing the selectivity of ethanol over water. The separation factor is a key process performance index to determine the energy consumption for ethanol recovery. Figure 5.26 illustrates the calculated thermodynamic energy consumption for ethanol recovery using the IL and membrane contactor with different separation factors. The high separation factor shown in Fig. 5.25 demonstrates clear energy savings benefits for the IL/membrane contactor in comparison to distillation or pervaporation. Figure 5.27 represents a typical concentration profile for ethanol in the extracted IL phase as a function of time. As observed in Fig. 5.27, starting from a pure IL, the cumulative concentration of ethanol increases and after around 6 h the concentration reaches 8%. This experiment shows the potential for recovery of ethanol from a fermentation broth.

Potential challenges associated with the alcohol recovery using the IL-assisted membrane contactor and potential solutions are outlined below.

1. Fouling of the membrane module – selection of proper membrane material and proper CIP procedure.

2. Shell side vs. lumen side – need to determine experimentally, but solvent flow in the lumen side and broth in the shell side might be beneficial from operational point of view.

3. Identifying a suitable IL – experiments to determine the partition coefficient of alcohols in different ILs from a vast pool of ILs. Hydrophobic ILs likely the superior candidates.

4. Price of ILs – ILs are expensive, but reduced volumes and avoiding solvent losses and makeup could make them economically feasible.

5. Portability/ease of operation – IL-based extraction in the membrane contactor should require a small footprint.

5.8 Emerging trends in separations technology for advanced biofuels

Advanced biofuels are defined by EISA 2007 as biofuels that are derived from non-food feedstocks and result in at least a 50% reduction in life cycle GHG emissions. To achieve market penetration, advanced biofuels should be compatible with existing fuel infrastructure and ‘drop-in’ to existing fuel production, distribution, and utilization pathways. A simplified schematic containing an overview of advanced biofuels production routes is given in Fig. 5.28.

In general, conversion technologies involve multiple steps that require different separations techniques. Separations technologies tend to be specific to the feedstocks, products, process streams and conversion technologies. Separations are classified based on the types and concentrations of species, solvents, and reaction conditions. There are significant opportunities to improve crosscutting separations technology that will enable deployment in integrated biorefineries. Critical crosscutting separations challenges include:

Some of the critical separations challenges in conversion technologies for advanced biofuels are:

5.9 Performance indices

5.9.1 Reverse osmosis (RO)

For processes with non-porous membranes, such as RO, there are three major performance indices that are commonly used (Ho and Sirkar, 1992):

• flux

• rejection

• recovery

Flux is defined as the transfer rate of a species per unit surface area. Mass flux is generally expressed as mass/area/time in the form of g/cm²/s or mole/cm²/s. Similarly, volumetric flux is expressed as cm³/cm²/s.

For RO, volumetric flux of water (J_w) and solute (J_s) are defined as:

${\mathit{J}}_{\mathit{w}}=\mathit{A}\left(\mathrm{\Delta }\mathit{P}-\mathrm{\Delta }\mathrm{\pi }\right)$

where A = water permeability = $\frac{{\mathit{P}}_{\mathit{w}}}{\mathit{L}}$ and ${\mathit{P}}_{\mathit{w}}=\frac{{\mathit{D}}_{\mathit{w}}{\mathit{C}}_{\mathit{w}}{\overline{\mathit{V}}}_{\mathit{w}}}{\mathit{RT}}$

${\mathit{J}}_{\mathit{S}}=\mathit{B}\left({\mathit{C}}_{\mathit{Fm}}-{\mathit{C}}_{\mathit{P}}\right)$

where B = solute transport number = $\frac{{\mathit{P}}_{\mathit{S}}}{\mathit{L}}$ and P_s = D_sK_s

ΔP = transmembrane pressure drop

Δπ = osmotic pressure gradient between the feed and permeate

L = thickness of the membrane

C_Fm = Concentration of the solute in the feed side at the membrane surface

C_P = concentration of the solute in the permeate side

D_w and D_s = diffusivitities of water and solute, respectively

C_w = concentration of water in the membrane

${\overline{\mathit{V}}}_{\mathit{w}}$ = partial molar volume of water

R = universal gas constant

T = absolute temperature

K_s = solute partition coeffient

The solute rejection, R, is expressed as:

$\mathit{R}=\left(1-\frac{{\mathit{C}}_{\mathit{P}}}{{\mathit{C}}_{\mathit{F}}}\right)$

where C_P and C_F are concentration of the solute in permeate and feed, respectively.

Recovery of water (r) is expressed as:

$\mathit{r}=\frac{{\mathit{J}}_{\mathit{w}}\mathit{·}\mathrm{Area}}{\mathrm{Feed}\mathrm{volumetric}\mathrm{flow}\mathrm{rate}}$

5.9.2 Ultrafiltration (UF) and microfiltration (MF)

For processes with microporous membranes, such as UF and MF, the expression of flux is different than for RO (Ho and Sirkar, 1992). Note that for processes with microporous membranes, the osmotic pressure gradient between the feed and permeate solutions is negligible.

Flux

Hagen-Poiseuille’s equation for laminar, incompressible flow through a uniform cylindrical channel is used to express pressure drop (ΔP) as:

$\mathrm{\Delta }\mathit{P}=\frac{32\mathrm{\mu }\mathit{Lv}}{d_P{}^2}$

The above equation is used to derive an expression for permeate flux (J) through a microporous membrane as given below. J is expressed as volumetric flow rate per unit external surface area of membrane.

$\mathit{J}=\frac{\mathit{q}{\mathit{N}}_{\mathit{P}}}{{\mathit{A}}_{\mathit{m}}}=\mathit{v}\frac{\mathrm{\pi }{d}_P{}^2}{4}\frac{{\mathit{N}}_{\mathit{P}}}{{\mathit{A}}_{\mathit{m}}}=\frac{\mathrm{\pi }{\mathit{r}}_{\mathit{P}}^4\mathrm{\Delta }\mathit{P}}{8\mathrm{\mu }\mathit{L}}\frac{{\mathit{N}}_{\mathit{P}}}{{\mathit{A}}_{\mathit{m}}}=\frac{\mathrm{\epsilon }{\mathit{r}}_{\mathit{P}}^2\mathrm{\Delta }\mathit{P}}{8\mathrm{\mu }\mathit{L}}=\mathit{A}(\mathrm{\Delta }\mathit{P})$

where μ = viscosity of solution, L = thickness of membrane, v = velocity, d_P = diameter of membrane pore, q = volumetric flow rate through a single pore, N_P = number of pores, A_m = membrane external surface area, ΔP = transmembrane pressure drop, A = permeability, and

$\mathrm{\epsilon }=\mathrm{porosity}\mathrm{of}\mathrm{membrane}=\frac{{\mathit{N}}_{\mathit{P}}\mathrm{\pi }{\mathit{r}}_{\mathit{P}}^2}{{\mathit{A}}_{\mathit{m}}}.$

From Darcy’s Law the flux, J, could also be expressed as

$\mathit{J}=\frac{\mathrm{\Delta }\mathit{P}}{\mathrm{\mu }{\mathit{R}}_{\mathit{m}}}$

where R_m = resistance of the membrane.

At higher rejection, the rejected species tend to form a layer on the surface of the membrane and this phenomenon is known as fouling. The expression of J for a fouled membrane changes to:

$\mathit{J}=\frac{\mathrm{\Delta }\mathit{P}}{\mathrm{\mu }\left({\mathit{R}}_{\mathit{m}}+{\mathit{R}}_{\mathit{F}}\right)}$

where R_F = resistance of the layer of the rejected species.

Selectivity (S)

Selectivity of species a over b for membrane-based separation is defined as:

$\mathit{S}=\frac{\mathrm{ratio}\mathrm{of}\mathrm{the}\mathrm{concentration}\mathrm{of}\mathit{a}\mathrm{to}\mathrm{that}\mathrm{of}\mathit{b}\mathrm{in}\mathrm{permeate}}{\mathrm{ratio}\mathrm{of}\mathrm{the}\mathrm{concentration}\mathrm{of}\mathit{a}\mathrm{to}\mathrm{that}\mathrm{of}\mathit{b}\mathrm{in}\mathrm{feed}}=\frac{{\left(\raisebox{1ex}{${\mathit{C}}_{\mathit{a}}$}\!\left/ \!\raisebox{-1ex}{${\mathit{C}}_{\mathit{b}}$}\right.\right)}_{\mathit{P}}}{{\left(\raisebox{1ex}{${\mathit{C}}_{\mathit{a}}$}\!\left/ \!\raisebox{-1ex}{${\mathit{C}}_{\mathit{b}}$}\right.\right)}_{\mathit{F}}}$

5.9.3 Electrodialysis (ED) and electrodeionization (EDI)

For ED/EDI, the critical performance measurement indices are briefly described below.

Separation efficiency

Separation efficiency (α) in an ED/EDI process is defined as:

$\mathit{\alpha }=1-\frac{{\mathit{C}}_{\mathit{o}}}{{\mathit{C}}_{\mathit{i}}}$

where C_i and C_o are the mole concentration of species at the inlet and outlet of the diluate side of ED/EDI stack, respectively.

Current efficiency

Current efficiency (η) is defined as:

$\mathrm{\eta }=\frac{\left({\mathit{C}}_{\mathit{i}}-{\mathit{C}}_{\mathit{o}}\right)\mathit{zQF}}{\mathit{I}}$

where Q = volumetric flow rate of the solution, F = Faraday’s number = 96,485 amp-s/mole, I = current, and z = ionic charge/mole.

Productivity

Productivity (P) is the same as the mass flux of the species and is expressed as:

$\mathit{P}=\frac{\left({\mathit{C}}_{\mathit{i}}-{\mathit{C}}_{\mathit{o}}\right)\mathit{Q}}{{\mathit{A}}_{\mathit{m}}}$

where A_m = the cross-sectional area of the membrane.

5.9.4 Adsorption

Adsorption capacity (q)

$\mathit{q}=\frac{\mathrm{Amount}\mathrm{of}\mathrm{solute}\mathrm{adsorbed}}{\mathrm{Mass}\mathrm{of}\mathrm{adsorbent}\mathrm{used}}\mathrm{or}\frac{\mathrm{Amount}\mathrm{of}\mathrm{solute}\mathrm{adsorbed}}{\mathrm{Porous}\mathrm{surface}\mathrm{area}\mathrm{of}\mathrm{adsorbent}\mathrm{used}}$

5.9.5 Liquid–liquid extraction

Partition coefficient (K)

$\mathit{K}=\frac{\mathrm{Concentration}\mathrm{in}\mathrm{solvent}1}{\mathrm{Concentration}\mathrm{in}\mathrm{solvent}2}={\frac{{\left(\mathit{C}\right)}_1}{\left(\mathit{C}\right)}}_2$

5.10 Conclusion

Separations can account for ~ 50% of the total production costs in bioprocessing operations. Designing and implementing efficient separations strategies are critical factors in successful bioprocessing. We provided an overview of separations technologies commonly employed in existing biorefineries as well as emerging separations technologies that are well poised to exhibit rapid growth in future biorefineries. Success requires consideration of feedstock composition, potential conversion technologies, separations scheme, product composition, and purity requirements. Design of the separations systems will impact energy use, system footprint, and capital costs. Several examples were presented from our work.

5.11 Acknowledgements

Funding for the work is gratefully acknowledged from the following sponsors: US Department of Energy, Office of Energy Efficiency and Renewable Energy, Bioenergy Technologies Office, US Department of Energy, Office of Energy Efficiency and Renewable Energy, Technology Commercialization Fund, US Department of Agriculture – CSREES Grant # 68-3A75-6-505, and the Illinois Corn Marketing Board.