en

gine

Enzymatic hydrolysisFuel ethanolLignocellulosic biomassSimultaneous saccharication andfermentation (SSF)

fromgrownoetstage

umetric productivity, reduced labor costs, and reduced vessel down time for cleaning and lling. On the

the Inof thee to thentrat

demand is projected to grow rapidly as vehicle trafc increasesthroughout the world and even accelerates in Asia. Besides thenegative global warming impact of fossil fuels, volatile oil pricesand dependency on politically unstable oil exporting countries re-sulted in a signicant increase in international interest in alterna-tive fuels and led policy makers in the EU and the US to issueambitious goals for substitution of alternative for conventionalfuels (Galbe and Zacchi, 2002; Wyman, 2007).

as desirable (Farrell et al., 2006; Hahn-Hgerdal et al., 2006). Alter-natively, ethanol can be produced from lignocellulosic materialssuch as agricultural residues, wood, paper and yard waste in muni-cipal solid waste, and dedicated energy crops, which constitute themost abundant renewable organic component in the biosphere (Cla-assen et al., 1999).

Regardless of the feedstock, the nal ethanol selling prize mustbe competitive with that for gasoline, but gasoline benets fromover a century of learning curve improvements and largely paidfor capital. Thus, prot margins in ethanol production processesare low, and returns on capital are uncertain due to the tremendous

* Corresponding author. Tel.: +1 951 781 5703; fax: +1 951 781 5790.

Bioresource Technology 101 (2010) 48624874

Contents lists availab

T

elsE-mail address: charles.wyman@ucr.edu (C.E. Wyman).centrations of carbon dioxide (CO2), the dominant greenhousegas, have increased from a pre-industrial value of about 280 ppmto 379 ppm in 2005, primarily as a result of fossil fuel use (IPCC,2007). Overall, petroleum is the source of about 170 quadrillion(1015) BTUs or quads of energy of the total of more than 460 quadsthe world uses, far more than derived from coal, natural gas,hydroelectric power, nuclear energy, geothermal, or other sources.Over half of petroleum in this total is used for transportation, and

gallons in 2007. However, these raw materials are insufcient tomeet the increasingdemand for fuels, and concernshaveheightenedrecently that competition between the use of agricultural commod-ities for fuelproduction isdrivingup foodcosts. Furthermore, theuseof food crops for fuel productionmay lead to environmentally detri-mental indirect land use changes, e.g. the deforestation of tropicalrainforest to gainmore farmland. In addition, the reductionof green-house gases resulting fromuse of starch-based ethanol is not as high1. Introduction

According to the recent report ofon Climate Change (IPCC) warmingis unequivocal and is very likely duanthropogenic greenhouse gas conc0960-8524/$ - see front matter 2009 Elsevier Ltd. Adoi:10.1016/j.biortech.2009.11.009other hand, these systems are more susceptible to microbial contamination and require more sophisti-cated operations. Despite the latter challenges, continuous processes could be even more important toreducing the costs of overcoming the recalcitrance of cellulosic biomass, the primary obstacle to low costfuels, through improving the effectiveness of utilizing expensive enzymes. In addition, continuous pro-cessing could be very benecial in adapting fermentative organisms to the wide range of inhibitors gen-erated during biomass pretreatment or its acid catalyzed hydrolysis. If sugar generation rates can beincreased, the high cell densities in a continuous system could enable higher productivities and yieldsthan in batch fermentations.

2009 Elsevier Ltd. All rights reserved.

tergovernmental Panelworlds climate systeme observed increases inions. Atmospheric con-

Ethanol made biologically by fermentation from a variety of bio-mass sources is widely recognized as a unique transportation fuelwith powerful economic, environmental and strategic attributes.First generation ethanol made from starch-rich materials such ascorn and wheat or from sugar feedstock is a mature commodityproduct with a worldwide annual production of over 13 billion USKeywords:Continuous fermentation

cellulosic biomass to ethanol. As shown in this review, some continuous fermentations are nowemployed for commercial ethanol production from cane sugar and corn to take advantage of higher vol-Review: Continuous hydrolysis and ferm

Simone Brethauer, Charles E. Wyman *

Center for Environmental Research and Technology and Chemical and Environmental En

a r t i c l e i n f o

Article history:Received 31 August 2009Received in revised form 2 November 2009Accepted 3 November 2009Available online 14 December 2009

a b s t r a c t

Ethanol made biologicallyresidues, grasses, and fasttion fuel with powerful ecocompetitive for these benimportant potential advan

Bioresource

journal homepage: www.ll rights reserved.tation for cellulosic ethanol production

ering Department, University of California, Riverside, CA 92507, United States

a variety of cellulosic biomass sources such as agricultural and forestrying wood is widely recognized as a unique sustainable liquid transporta-

mic, environmental, and strategic attributes, but production costs must beto be realized. Continuous hydrolysis and fermentation processes offers in reducing costs, but little has been done on continuous processing of

le at ScienceDirect

echnology

evier .com/locate /bior tech

as the reagents travel through the PFR with diffusion assumed tobe negligible in the axial direction. Consequently, PFR operationsimply that inoculum has to be constantly fed to the reactor for fer-

strate or the need for recycle. In a two stage system with properlydesigned unequal reactor sizes (see Fig. 2a), the maximum possibleoverall productivity would be lower than in a single stage system,but the substrate conversion at identical productivities would behigher. In the two stage system, a maximum overall productivityof 4.16 g L1 h1 was calculated at a substrate conversion of 92%(Fig. 2b). If a substrate conversion of 99% were the goal, the pro-ductivity in a single stage CSTR would be 2.77 g L1 h1 but3.94 g L1 h1 for a two stage system. Generally, a cascade of fer-mentors would be superior to a single vessel for autocatalytic reac-tions such as cell growth which are product-inhibited, but forsituations with substrate inhibition, a single stage CSTR is oftenmore favourable to remove as much reactant as possible (de Goo-ijer et al., 1996).

3. Continuous ethanol production from starch and sugarfeedstocks

3.1. Industrial continuous ethanol production from sugar cane

Sugar cane is a tropical and subtropical crop that is the primaryfeedstock for ethanol production in Brazil, India, and Colombia. Itcontains mainly sucrose, a dimer of glucose and fructose, which

pr

rce Tmentation processes. Cascading a large number of CSTRs in serieswill have similar performance to a PFR.

In a system with constant overall reaction stoichiometry thatcan be described by a single kinetic equation, performing the reac-tion in two or more bioreactors may lead to a higher product con-centration, a higher degree of conversion, a higher volumetricproductivity, or a combination of these factors compared to opera-tion of a single CSTR. One approach to optimizing a continuous pro-cess is to determine the reactor conguration that gives the lowestresidence time to achieve a certain degree of conversion. If thekinetics are known, a plot of the reciprocal rate against the dimen-sionless substrate concentration S/Sfeed can be employed to esti-mate the reaction residence time and therefore the reactionvolume (de Gooijer et al., 1996). For a CSTR, the area correspondingto a rectangle whose height equals the reciprocal of the rate at thedesired conversion will equal the residence time for reaction tothis conversion, whereas the residence time for a PFR will corre-spond to the area under the curve (see Fig. 1). If the desired conver-sion is higher than the minimum in the curve, a combination ofprice swings in petroleum prices. In this context, costs must be keptas low as possible, and continuous fermentation of cellulosic bio-mass to ethanol can offer important advantages in terms of greaterproductivity and lower costs. Unfortunately, although processdesigns have been conceptualized based on continuous enzymatichydrolysis and fermentations to take advantage of their low costpotential, limited studies have actually been reported from whichto design or advance the technology. Thus, more information issorely needed on this subject to guide the advancement of lowercost approaches to making ethanol and overcome the signicantcost barriers to market entry.

In this paper, we will provide a short introduction to conceptsand characteristics of continuous fermentations. Then a summaryis presented of experiences and research activities with rst gener-ation industrial continuous ethanol fermentations as these providethe foundation for second generation cellulose-based processes.Following that, we review current knowledge of continuous fer-mentation of lignocellulosic material, including those based onchemical and enzymatic hydrolysis of cellulose to glucose.

2. Concept of continuous fermentations

In a true continuous fermentation system, substrate is con-stantly fed to the reaction vessel, and a corresponding ow of fer-mented product broth is discharged to keep the reactor volumeconstant. Furthermore, the balance between feed and dischargeis maintained for long enough times to achieve steady state oper-ation with no changes in the conditions within the reactor. Com-pared to a batch reaction, this mode of operation offers reducedvessel down time for cleaning and lling providing improved vol-umetric productivity that can translate into smaller reactor vol-umes and lower capital investments plus ease of control atsteady state.

Two basic types of continuous reactors can be employed: thecontinuous stirred tank reactor (CSTR) or the plug ow reactor(PFR). In an ideally mixed CSTR, the composition in the reactor ishomogenous and identical to that for the outgoing ow. In an idealPFR, the reactants are pumped through a pipe or tube with a uni-form velocity prole across the radius, and the reaction proceeds

S. Brethauer, C.E. Wyman / Bioresoureactors will require less reaction volume. Thus, for the situationdepicted in Fig. 1, the combination of a CSTR followed by a PFR willbe preferred if a conversion of 98% is targeted.An important performance criteria is the productivity of the fer-mentation system, i.e., the amount of product formed per unit oftime and reactor volume, which depends on several factors includ-ing substrate concentration, cell concentration, and dilution rate.We used a simple model based on Monod growth kinetic that in-cludes terms for product and cell inhibition (Lee et al., 1983) toillustrate the inuence of some operational parameters. Generally,the productivity in a continuous fermentation system is higherthan in a batch reactor. In the model example, a productivity of3.57 g L1 h1 was calculated for a batch process inoculated with1 g L1 yeast cells. In a standard single stage CSTR without cellretention, where the biomass concentration would be xed, a max-imum productivity of 4.24 g L1 h1 was calculated for a dilutionrate of 0.136 h1, however, the substrate conversion was only83%. Generally it is desirable to achieve almost complete substrateconversion at the highest possible productivity to avoid loss of sub-

4

2

0

reci

1.00.80.60.40.20.0[S]/ [Sfeed]

CSTRPFR

Fig. 1. Graphical design method to estimate reactor residence times and corre-sponding volumes. The reciprocal of reaction rate is plotted versus dimensionlesssubstrate concentration S/Sfeed on the basis of a literature kinetic model (de Gooijeret al., 1996; Lee et al., 1983).10

8

6

ocal

rate

echnology 101 (2010) 48624874 4863is readily assimilated by Saccharomyces cerevisiae (Sanchez andCardona, 2008). Both sugar cane juice and molasses normally con-tain sufcient minerals and nutrients for S. cerevisiae to ferment

rce Ta

b

4864 S. Brethauer, C.E. Wyman / Bioresouthem directly to ethanol (Wheals et al., 1999). In Brazil, 7080% ofthe distilleries employ fed-batch processes (Melle-Boinot Process)for fermentors with outputs ranging from 400 to 2000 m3 ethanolper day. Typically, high yeast cell concentrations of between 8%and 17% achieve fermentation times of only 610 h and nal etha-nol concentrations of up to 11% v/v, corresponding to an averageethanol yield of 91%. After each fermentation cycle, the yeast cellsare separated, treated with dilute sulphuric acid to kill contaminat-ing bacteria, and then recycled to start a new fermentation. This se-quence can be repeated up to 200 times and minimizes carbonconsumption for yeast growth while providing very high ethanolproductivities (Dorer and Amorim, 2007; Godoy et al., 2008;Wheals et al., 1999; Zanin et al., 2000). The rst continuous ver-sions of the Melle-Boinot process appeared in the 1970s, but sev-eral operational problems were detected, such as a high level ofcontamination, low productivity, low yields, and problems withsolids ow. Todays continuous fermentation processes are opti-mized based on kinetic models to achieve high productivities (typ-ically 10 mL L1 h1), high process exibility and stability, and lowconsumption of chemicals and are considered to be less expensivefor ethanol production than batch processes (Zanin et al., 2000).

An important feature of state-of-the-art continuous processes isthe use of multiples stages (typically four or ve) of variable sizes.

Fig. 2. Design considerations for a multistage continuous fermentation system asdescribed by the kinetic model of Lee et al. (1983): (a) the volume of the rst CSTRin a series of two CSTRs comprising a total volume of 1 L inuences the productivityof the whole system. The dilution rate was set to 0.1 h1. (b) Inuence of thedilution rate (based on the whole system) on productivity and steady statesubstrate concentration of a volume optimized system of two CSTRs in series. Therst reactor has a volume of 0.83 L and the total volume of the system is 1 L.The sugar substrate is fed to the top of the rst reactor togetherwith the recycled yeast cream and leaves through the bottom,owing then by gravity to the middle of the next stage. Each reac-tor typically uses an external plate-type heat exchanger for coolingof the fermentation broth, with the kinetic energy of the liquidleaving the heat exchanger outlet used to agitate the reactor. Theyeast cells produced are separated from the wine by disk-bowlcentrifuges, forming a yeast cream, which is then sent to acid treat-ment prior to being recycled back to the rst reactor (Zanin et al.,2000).

Guerreiro et al. (1997) described an expert system for the de-sign of such an industrial continuous fermentation plant whichcombines expert knowledge and industrial practices with kineticmodelling. As parameters are taken from industrial fermentations,differences between theoretical calculations and practical resultsare claimed to be minimal. In the example presented, an inputmedium containing 170190 g L1 sugar was fed at a rate of143 m3 h1 to a four stage reactor train with volumes of 215,274, 324, and 213 m3 to maximize performance. In this case, thesteady state concentrations in the rst and last stages were 54and 1 g L1 of sugar, 42 and 66 g L1 of ethanol, and 29 and31 g L1 of yeast biomass, respectively, and the process productiv-ity was given as 7.7 g L1 h1. Generally, the volumes of the tanksinuence the productivity which varied from 6.1 to 7.9 g L1 h1.Through optimization, it was possible to replace a previous fed-batch plant that consisted of 24 fermenters of 200 m3 volume each(total 4800 m3) producing 400 m3 of 96% ethanol per day with acontinuous plant with a total volume of 2500 m3 producing about440 m3 of 96% ethanol per day (Guerreiro et al., 1997). However,larger continuous plants exist with capability to produce up to600 m3 ethanol per day (Zanin et al., 2000).

In some Brazilian distilleries, processes based on occulentyeast strains are employed, with cell separation in settlers to avoidcostly centrifuges. Yeast occulation is a reversible, asexual, cal-cium dependent process of self-aggregation in which cells adhereto form ocs consisting of thousands of cells. Because of theirmacroscospic size and mass, the yeast ocs rapidly settle out ofthe fermenting medium, thus providing natural cell immobiliza-tion (Verbelen et al., 2006). Compared to the classical Melle-Boinotprocess, it is claimed that up to 1.5% higher fermentation efciencyis obtained, ethanol production costs are ca. $7/m3 lower, and con-sumption of chemicals such as antifoam is reduced (Zanin et al.,2000).

Despite such desirable attributes, there are also critical opinionsabout replacing batch fermentations with continuous processes. Inone study of the advantages and disadvantages of continuous andbatch fermentation processes for 62 distilleries over a time span of9 years (19982007), batch processes with yeast recycle wereshown to be less susceptible to bacterial contamination and thecorresponding loss in productivity (Godoy et al., 2008). Lactobacil-lus contaminations, in particular, are regarded as the major factorthat can reduce ethanol yield and also impair yeast centrifugation,and greater quantities of antibiotics are needed to address this is-sue for continuous processes. Also, slightly more sulphuric acidwas consumed in continuous processes. Yet, continuous processeshave the advantages of lower installation costs due to smaller fer-mentor volumes and less heat exchanger demands as well as lowercosts due to greater automation (Godoy et al., 2008).

3.2. Continuous ethanol production from corn

Up to now, corn is the major feedstock for ethanol production inthe US, which surpasses Brazil as the largest ethanol producer.

echnology 101 (2010) 48624874Corn kernels contain about 70% by weight starch on a dry weightbasis. Starch is a D-glucose polymer, consisting of about 30% amy-lose, a linear chain of a-1,4 linked glucose units with a helical

water to form a mash.The isolated starch from wet-milling and the mash from the dry

rce Tgrind process are treated identically to produce ethanol. First, athermostable alpha-amylase, which breaks down the starch poly-mer to soluble dextrins by hydrolyzing a 14 bonds, is added.The mixture is heated to over 100 C to liquefy the mash over aholding time of at least 30 min. Then, glucoamylase is added,which converts liqueed starch to glucose at an optimal tempera-ture of 65 C. In the nal fermentation step, which is performedeither coupled (simultaneous saccharication and fermentation,SSF) or subsequent to glucoamylase treatment (separate hydrolysisand fermentation, SHF), the mash is cooled to 32 C, and yeast isadded as well as ammonium sulphate or urea as a nitrogen source.Alternatively, proteases are added to break down corn protein tofree amino acids for use as a nitrogen source. Fermentation is com-pleted in 4872 h to a nal ethanol concentration of 1012% v/vand higher. Over the course of fermentation, the pH drops to 4.0or lower, which helps to prevent bacterial contamination. Manyplants use simultaneous saccharication and fermentation, be-cause it lowers the risk of contamination, lowers the initial osmoticstress on the yeast, and is generally more energy-efcient (Bothastand Schlicher, 2005).

According to a United States Department of Agriculture (USDA)survey in 2002, 27% of the dry grind distilleries in the US employcontinuous fermentation processes which are more common inlarge plants producing more than 400 m3 ethanol per day (Shapo-uri and Gallagher, 2005). To the best of our knowledge, no perfor-mance data for industrial continuous corn ethanol fermentationsare published. However, Bai et al. (2008) described in their reviewa commercial plant employing a self-occulating yeast with a pro-duction capacity of 680 m3 per day which started operation in2005 in China. In this system, six fermentors with volumes of1000 m3 each were arranged in a cascade, and corn meal hydroly-zate, with a sugar concentration of 200220 g L1, was fed to thefermentation system at a dilution rate of 0.05 h1. The nal ethanolconcentration was reported to be 1112% v/v. Yeast ocs were re-tained within the fermentor by bafes to effectively immobilizethem, and the yeast concentration within the fermentors wasmaintained at 4060 g DCW L1.

4. Continuous production of second generation ethanol fromlignocellulosic materials

Although composition of lignocellulosic materials varies in dif-ferent plants, the three main components are cellulose (3661%),hemicellulose (1339%), and lignin (629%) (Olsson and Hahn-Hgerdal, 1996). Cellulose is a D-glucose polymer, where the sub-units are linearly linked by b-1,4 glycosidic bonds and exists incrystalline and amorphous forms. Hemicellulose is composed oflinear and branched heteropolymers of pentoses (i.e., xylose andarabinose) and hexoses (i.e., mannose, glucose, and galactose). Lig-structure and 70% amylopectin, a highly branched polymer withadditional a-1,6 glycolytic bonds. Ethanol from corn can be pro-duced by either a dry grind (67% of the fuel ethanol) or wet mill(33%) process with recent growth in the industry mostly withdry grind plants due to their lower capital costs (Bothast and Schli-cher, 2005).

In the wet mill process, the grain is separated into its four basiccomponents of starch, germ, ber, and protein to recover highervalue co-products including corn oil, corn gluten meal, corn glutenfeed, and germ. On the other hand, the dry grind process is muchsimpler in that the entire corn kernel is ground and mixed with

S. Brethauer, C.E. Wyman / Bioresounin is a polymer that can consist of three different phenylpropaneunits (p-coumaryl, coniferyl and sinapyl alcohol) that bind theplant together. In order to release fermentable sugar monomers,cellulose and hemicellulose are hydrolyzed chemically, enzymati-cally, or by their combination (Gray et al., 2006; Hendriks and Zee-man, 2009; Wyman et al., 2004).

Lignocellulosics can be hydrolyzed chemically by addition ofacids, with sulphuric acid most often preferred based on priceand toxicity, and acid hydrolysis can be divided in two categories:concentrated acid hydrolysis and dilute acid hydrolysis. Concen-trated acid processes operate at low temperatures, e.g., 40 C, andgive high sugar yields, e.g., 90% of theoretical glucose yield. How-ever, acid consumption is high, a lot of energy is consumed for acidrecovery and recycle, the equipment can suffer from corrosion, andreaction times of 26 h are required. The dilute acid process ischaracterized by a low acid consumption and very short reactiontimes at high temperatures. Hemicellulose is generally much moresusceptible to acid hydrolysis than cellulose, and yields of morethan 85% can be obtained at relatively mild conditions, with onlya small part of the cellulose converted to glucose. More severe con-ditions required to achieve high glucose yields from cellulose,however, lead to degradation of hemicellulose sugars, resultingin low yields and unwanted side-products that are also strong fer-mentation inhibitors. Potential inhibitors that can be formed or re-leased from hemicellulose, cellulose, and lignin during suchthermochemical routes include furfural, 5-hydroxymethylfurfural(HMF), levulinic acid, acetic acid, formic acid, uronic acid, 4-hydroxybenzoic acid, vanillic acid, vanillin, phenol, cinnamalde-hyde, and formaldehyde. To reduce degradation of monosaccha-rides at high temperature, dilute acid hydrolysis is typicallycarried out in two stages, with hemicellulose solubilized in the rstunder relatively mild conditions and the residual solids hydrolyzedin the second under the more severe conditions needed to break-down cellulose. With this procedure, hemicellulose derived sugaryields are in the range of 90%, while glucose yields are only about4060% at realistic residence times. However, it has been reportedthat alternative reactor congurations to classic batch reactors,such as a shrinking-bed reactor give glucose yields of up to 90%(Taherzadeh and Karimi, 2007).

Cellulase enzymes from the fungus Trichoderma reesei canhydrolyze biomass to sugars at near ambient temperatures, result-ing in little degradation. However, because sugar yields from rawbiomass are very low, the biomass is subjected to a pretreatmentstep. Numerous pretreatment methods have been developedincluding pretreatment with steam, liquid hot water, dilute acid,lime, ammonia, and wet oxidation and are discussed in more detailelsewhere (Hendriks and Zeeman, 2009; Mosier et al., 2005; Wy-man et al., 2005, 2009). Effective pretreatments are thought to en-hance enzymatic digestibility of biomass due to several effects:disruption of the lignocellulosic structure by loosening the hemi-cellulose lignin entanglement, hemicellulose hydrolysis, lignin sol-ubilisation and disruption, decrystallization of cellulose, andincreased accessible surface area (Lynd et al., 2002; Zhang andLynd, 2004, 2006). During many pretreatments, fermentationinhibitors such as acetic acid, lignin breakdown products, and fur-fural are released. After pretreatment, several process congura-tions are possible, as recently reviewed in detail (Cardona andSanchez, 2007). In the separate hydrolysis and fermentation(SHF) approach, the liquid and solid phases are separated after pre-treatment, and the solid phase may be subjected to additionalwashing steps. The solids, in case of dilute acid and steam pretreat-ment, contain most of the lignin and cellulose from the raw bio-mass, with the latter hydrolyzed to glucose by addition ofcellulolytic enzymes that are comprised of endo- and exoglucanaseand b-glucosidase activities, often supplemented with additionalb-glucosidase derived from Aspergillus niger. The resulting hexose

echnology 101 (2010) 48624874 4865solution is then fermented to ethanol using conventional yeast orother suitable microorganisms. A suitable pentose fermentingstrain can convert the liquid stream from pretreatment containing

solubilized hemicellulose to ethanol in a separate unit usually aftera detoxication step such as overliming to reduce fermentationinhibitors and make the hydrolyzate fermentable. Hydrolysis andfermentation were initially separated to better match the pH andtemperatures to those that are optimal for each step, with about50 C preferred for enzymatic hydrolysis and about 32 C often bestfor fermentations. Alternatively, enzymes could be added to thewhole pretreatment slurry without separation of the liquid fromthe solids, followed by fermentation of pentoses and hexoses toethanol, a process which we call separate hydrolysis and co-fer-mentation (SHcF). This more integrated approach is economicallyvery attractive, but the fermentation step is much more challeng-ing than in SHF and complicates hydrolyzate conditioning to re-move inhibitors due to the presence of the solids (Cardona andSanchez, 2007).

Because cellulases are inhibited by their hydrolysis productscellobiose and glucose, a favoured processing mode is to combine

often low with values approaching typically not more than 70 g L1

due to challenges in feeding solids concentrations higher thanabout 10% by weight to the fermentors and end-product inhibitionof cellulase enzymes by the sugars released. Thus, a concentrationstep, e.g., vacuum evaporation, might be needed to achieve higherconcentrations, with additional extra costs possibly counterbal-anced by savings in the nal distillation step (Maiorella et al.,1984).

In one study, sugar cane bagasse was delignied by autoclavingin 1% NaOH for 1 h, and the solids were washed several times priorto hydrolysis by T. reesei cellulases. In a single stage continuous fer-mentation of S. cerevisiae with a 16% glucose feed, several dilutionrates were tested to nd a maximum ethanol productivity of4.1 g L1 h1 at a dilution rate of 0.13 h1. At this point, steadystate concentrations of 90 g L1 glucose, 31 g L1 ethanol, and3.8 g L1 biomass were measured. To avoid washout of largeamounts of unfermented glucose, a continuous single stage cell re-

Reac

SingSingwith

SingwithmemBatc

Sing

4866 S. Brethauer, C.E. Wyman / Bioresource Technology 101 (2010) 48624874hydrolysis and the fermentation, a process termed simultaneoussaccharication and fermentation (SSF), thereby keeping sugarconcentrations low (Gauss et al., 1976; Spindler et al., 1987; Takagiet al., 1977; Wright et al., 1987; Wyman et al., 1986). Although ini-tial applications subjected only the washed solids fraction frompretreatment to SSF with the liquid pentose stream processed sep-arately, these two steps can be combined in what is termed simul-taneous saccharication and co-fermentation (SScF) (Wooley et al.,1999). Despite the need to reduce the temperature for SSF from theoptimal levels for enzymes to accommodate fermentative organ-isms available to date, SSF was shown to achieve higher rates,yields, and concentrations than SHF by overcoming the major ef-fects of end-product inhibition (Spindler et al., 1987; Wrightet al., 1987). In addition, SSF, and even more so SScF, reduces fer-mentation equipment demands, and the presence of ethanol im-pedes invasion by unwanted organisms. Thus, until enzymes arefound that can overcome end-product inhibition, SSF or SScF arelikely to be preferred in terms of productivity, yields, and ethanolconcentrations. In the following sections, results for continuousfermentations with hydrolzates from acid and enzymatic hydroly-sis and for SSF applications are summarized.

4.1. Fermentation of hexoses in enzymatic hydrolyzates

Fermentation of hexose sugars derived from enzymatic hydro-lysis of washed pretreated lignocellulosic material generally doesnot pose special difculties (see Table 1), as the inhibitor concen-tration should be very low. However, compared to starch and sug-arcane fermentations, the sugar concentration after hydrolysis are

Table 1Fermentation of hexoses in enzymatic hydrolyzates.

Medium Sugar concentration(g L1)

Sugar cane bagasse pretreated with NaOH; washedsolids enzymatically hydrolyzed, concentrationby vacuum evaporation; addition of cheapnitrogen source, CaCl2, MgSO4

Reducing sugars: 160

Steam exploded oak chips, washed solidsenzymatically hydrolyzed, concentratedby vacuum evaporation, sterilizationfor 120 min at 60 C

Glucose: 180

Glucose: 170

Steam exploded spruce, whole slurryenzymatically hydrolyzed, addition

Glucose: 2550,mannose: 10of complete mineralmedium salts

Singwithcycle fermentation system was set up, and the maximal productiv-ity reached 18.3 g L1 h1 at a dilution rate of 0.3 h1 with a steadystate glucose concentration of 22 g L1 (Ghose and Tyagi, 1979).

Lee et al. (2000) employed an enzymatic hydrolyzate derivedfrom washed steam exploded oak chips (3 min at 215 C). To re-duce fermentation inhibitors, the hydrolyzate was sterilized for120 min at 60 C, rather than at a typical temperature of 121 C.Continuous cultures were performed in a reactor equipped withan internal membrane ltration module to retain cells inside thereactor. At a dilution rate of 0.22 h1 and a feed glucose concentra-tion of 180 g L1, 77 g L1 ethanol was produced, corresponding toa productivity of 16.9 g L1 h1 and a yield of 0.43 g g1. In a batchfermentation in a similar medium containing 170 g L1 glucose,only 57 g L1 ethanol was produced in 210 h, with 35 g L1 glucosenot utilized, leading to a very low productivity of 0.3 g L1 h1. Noproblems were experienced with bacterial contamination despitethe low sterilization temperature.

When the solidliquid separation and solids washing steps areomitted and the whole slurry is enzymatically hydrolyzed, fermen-tations are much more difcult, as exemplied by the work ofPalmqvist et al. (1998) who employed a hydrolyzate of spruce pre-treated by steam explosion for 5 min at 215 C after sulphur diox-ide impregnation. Two different batches of hydrolyzate were used,containing 2550 g L1 glucose and approximately 10 g L1 man-nose, that were both supplemented with mineral media. S. cerevi-siae ATCC 96581, a strain isolated from a spent sulphite liquor (SSL)fermentation plant running since 1940 and showing a 7-fold high-er maximum growth rate on SSL than bakers yeast, was employedfor the study. At a pH of 4.6 in a batch system, cells metabolized

tor type Dilutionrate(h1)

Ethanol(g L1)

cell dryweight(g L1)

Ethanolyield(g g1)

Ethanolproductivity(g L1 h1)

References

le stage CSTR 0.13 31 3.8 0.19 4.1 Ghose andTyagi(1979)

le stage CSTRcell recycle

0.3 58 30 0.36 18.3

le stage CSTRcell retention bybrane module

0.22 77 n.d. 0.43 16.9 Lee et al.(2000)

h 57 0.34 0.3(fermentationtime of210 h)

le stage CSTR 0.05 20 0.9 0.32 0.5 Palmqvistet al.0.1 Washout

(1998)le stage CSTR

cell recycle0.1 23 Maximum

260.51 2.3

rce Tglucose without growing and produced only a minor amount ofethanol. When the pH was raised to 5.0, the cells started to grow,and ethanol productivity increased. At the point that glucose wascompletely consumed, the system was switched to continuousoperation at a dilution rate of 0.05 h1 while raising the pH furtherto 5.5. At steady state, 20 g L1 ethanol was measured, correspond-ing to a yield of 0.32 g g1 and a productivity of 0.5 g L1 h1 at ayeast concentration of 0.9 g L1. Raising the dilution rate furtherto 0.1 h1 resulted in washout of the cells.

In a second continuous fermentation run with a dilution rate of0.1 h1, the pH was adjusted to 5.7, and cells were recycled by l-tration to increase productivity. Cells were removed at a low rateto maintain a constant cell density of approximately 6 g L1 be-tween 115 and 200 h, but when removal was stopped, the cell den-sity reached a maximum value of 26 g L1 after 290 h offermentation. Glucose, which was fed at a concentration of36 g L1 during this run, was depleted over most of the time, andethanol concentrations reached values as high as 23 g L1, corre-sponding to a mass yield of 0.51 g g1. The volumetric productivityof 2.3 g L1 h1 was 4.6 times higher than for the case without cellrecycle. Low growth rates were blamed on the presence of inhibi-tors such as HMF and furfural that are known to induce a lag phasein cell growth and ethanol production while they are being re-duced by the yeast. As in a continuous mode, the inhibitors wereconstantly fed with the result that the resulting low growth ratelimited the maximum possible dilution rate. The strong effect ofthe fermentation pH was explained by the presence of weak acids(e.g., acetic, formic, levulinic, p-hydroxybenzoic, vanillic, and syrin-gic acid), which are growth-inhibiting in their undissociated form(Palmqvist et al., 1998).

4.2. Fermentation of hexoses in acid hydrolyzates

Acid hydrolysis of lignocellulosic materials releases severalkinds of fermentation inhibitors, as described above. Chemicaldetoxication methods such as overliming, ion exchange adsorp-tion, addition of activated carbon, solvent extraction, or steamstripping, can improve the fermentability of an acid hydrolyzate,but these steps are costly and can also result in considerable sugarloss (Olsson and Hahn-Hgerdal, 1996). However, these inhibitors,especially HMF and furfural, can alternatively also be convertedin situ to non-toxic forms by yeast under appropriate process con-ditions. Because continuous cultivations keep such inhibitors atlower concentrations than seen initially for batch operations, wewould expect continuous processes to be better able to cope withthese compounds and avoid an extended lag phase and possiblyloss of cell viability. In situ detoxication is even more effectiveat higher cell densities, which can be achieved by different cellretention techniques in a continuous fermentation setup.

The fermentability of two stage dilute acid hydrolyzates wasstudied by several authors (Table 2). For example, in the rst step,30% w/w wood chips, water, and 5 g L1 H2SO4 were heated bysteam injection and kept at 12 bar for 710 min. After rapiddecompression, the liquid phase was separated, and the solidsheated again to a pressure of 21 bar and held there for 7 min.Brandberg et al. (2005) did not detoxify the hydrolyzate and em-ployed the yeast strain S. cerevisiae ATCC 96581 originally isolatedfrom spent sulphite liquor which proved to be relatively tolerant tothe inhibiting environment of dilute acid wood hydrolyzate. Con-tinuous experiments were performed with and without cell recy-cling by a cross-ow lter unit, and the effects of nutrientaddition were investigated. Recirculation of 90% of the cells inthe outow allowed a specic growth rate of 10% of the dilution

S. Brethauer, C.E. Wyman / Bioresourate and still avoided washout. Furthermore, the use of cell recircu-lation limited the losses of carbon in the form of cell biomass, asthe cell biomass yield was reduced by 50%. However, in continuousfermentations with 90% cell recycle by cross-ow ltration, unsup-plemented hydrolyzate could still not be fermented at dilutionrates of 0.1 and 0.06 h1. However, with supplementation of min-eral media ingredients, 99% of the sugars were converted at a cellconcentration of 6 g L1 leading to an ethanol concentration of17 g L1 and a productivity of 1.6 g L1 h1. Microaerobic condi-tions resulted in higher biomass growth and gave more stableand robust fermentations. Glycerol production, by which cellsreoxidize NADH and maintain the redox balance, was generallylow using acid hydrolyzate, apparently due to the fact that S. cere-visiae can use reduction of furfural as an alternative redox sink.

In a later work, because wheat hydrolyzate is available in largeamounts at a relatively low price, the authors tested it as a cheapalternative to the expensive ingredients of a complete denedmin-eral media including trace metals and vitamins (Brandberg et al.,2007). When dilute acid spruce hydrolyzate was supplementedwith 10% wheat hydrolyzate, only a minimal level of biologicalactivity could be sustained in a continuous cultivation atD = 0.1 h1. However, when 7.5 g L1 (NH4)2SO4 and 20 lg L1 bio-tin were added along with wheat hydrolyzate, steady state fermen-tation of the dilute acid hydrolyzate was achieved with a hexoseconversion of 76%, a cell concentration of 1.9 g L1, and an ethanolproductivity of 1.7 g L1 h1. To improve the productivity, threedifferent types of cell retention were evaluated: cross-ow ltra-tion with 75% recirculation, sedimentation in a settler equally sizedas the working volume of the reactor, and immobilization in cal-cium alginate. When 75% of the cells were retained by ltrationat a fermentation dilution rate of 0.1 h1, the cell concentrationin the reactor tripled compared to the classic chemostat, hexoseconversion increased to 94%, and the ethanol production ratewas 2.3 g L1 h1. Comparable results were achieved with cellrecirculation by a settler and immobilization. When the dilutionrate was increased to 0.2 h1, neither ltration nor sedimentationcould prevent washout of the cells. On the other hand, cultureimmobilization increased the ethanol productivity to 3.5 g L1 h1,although hexose conversion dropped. Ethanol was still produced ata rate of 4.5 g L1 h1 for a dilution rate of 0.3 h1, but hexose con-sumption decreased even more (Brandberg et al., 2007).

A similar dilute acid hydrolyzate supplemented with denedmineral media was fermented by yeast strain S. cerevisiae CBS8066 immobilized in Ca-alginate beads at dilution rates of 0.3,0.5, and 0.6 h1. Glucose consumption was found to drop withincreasing dilution rate from 86% to 79% and mannose consump-tion from 72% to 55%. Ethanol yields based on consumed sugarsvaried between 0.45 and 0.47 g g1. In contrast, fermentation ofthe hydrolyzate with free cells experienced washout at a dilutionrate of only 0.2 h1. However, when the experiments were re-peated with a second batch of hydrolyzate, none of the describedfermentations were successful without detoxication of the hydro-lyzate by overliming (Taherzadeh et al., 2001).

In a follow up work by the same authors with immobilizedyeast cells, different reactor congurations were examined, andexperiments were performed with a dened glucose based mineralmedia in detoxied hydrolyzate. In a single stage CSTR operated atdilution rates between 0.22 and 0.86 h1, conversion of an initialglucose concentration of 20 g L1 in the control mineral media de-creased from 100% to 77%, but productivities increased from 2.1 to6.6 g L1 h1. If an equal sized uidized bed bioreactor (FBBR) wasconnected to the CSTR, glucose conversion was higher than 99%even for the highest dilution rate calculated based on the rst reac-tor only. In contrast, a single FBBR gave a higher glucose conversionof 92% at a dilution rate of 0.86 h1 and a higher ethanol productiv-ity of 7.4 g L1 h1. When hydrolyzate containing initially

1

echnology 101 (2010) 48624874 486728.4 g L sugars was employed as carbon and energy source, con-versions dropped from 98% to 54% as dilution rate increased from0.22 and 0.86 h1 in a single stage CSTR, and ethanol productivities

ilutte (

.06

.1

.1

rce TTable 2Fermentation of hexoses in acid hydrolyzates.

Medium Sugarconcentration(g L1)

Reactor type Dra

Two stage dilute acidhydrolyzate of spruce

3040 Single stage CSTR with 90%cell recirculation

0

Two stage dilute acidhydrolyzate of spruce,supplemented with fullmineral medium salts

3040 0

Two stage dilute acidhydrolyzate of spruce,addition of wheathydrolyzate, (NH4)2SO4,biotin

Glucose: 34,mannose: 4

Single stage CSTR 0Single stage CSTR witheither 75% cell retention byltration, or cellsedimentation, or cellimmobilisation in alginate

4868 S. Brethauer, C.E. Wyman / Bioresouwere in the range of 2.76 g L1 h1. The addition of a FBBR in-creased sugar conversion to 86.6% at the highest dilution rate of0.86 h1. However, productivities were slightly lower in the twostage system, with values between 1.4 and 5.5 g L1 h1. This reac-tor segregation gained about 11.6% in hexose assimilation if hydro-lyzate was fermented but only 1.2% in synthetic media (Purwadiand Taherzadeh, 2008).

Alternatively, yeast cells can also be encapsulated rather thantraditionally entrapped in an alginate matrix, which has the advan-tages that resistance to diffusion of nutrients as well as cell leakageis lower and higher cell concentrations are possible. Talebnia andTaherzadeh (2006) encapsulated S. cerevisiae CBS 8066 cells andsecond stage hydrolyzate, obtained as previously described, wasused for the experiments and supplemented with mineral mediaingredients. In continuous hydrolyzate fermentations at dilutionrates of 0.1, 0.2, 0.3, 0.4, and 0.5 h1, and glucose conversiondropped from 95% to 71% and mannose conversion from 98% to79% over this range. Ethanol productivity increased with increasing

beadsSingle stage CSTR with cellimmobilization in alginatebeads

0.2

Two stage dilute acidhydrolyzate of wood,addition of full mineralmedium ingredients

Glucose: 10.2,mannose:19.8

Single stage CSTR with cellimmobilization in alginatebeads

0.3

0.6

Mineral medium Glucose: 20 Single stage CSTR with cellimmobilization in alginatebeads

0.220.86

Fluidized bed bioreactor(FBBR)

0.86

Equally sized CSTR andFBBR in series

0.86rst r

Two stage dilute acidhydrolyzate of wood,detoxided byoverliming, addition offull mineral mediumingredients

Glucose: 28.4 Single stage CSTR with cellimmobilization in alginatebeads

0.220.86

CSTR and FBBR in series 0.86

Second stage dilute acidhydrolyzate of wood,addition of full mineralmedium ingredients

Glucose: 19.3;mannose 6.6

Single stage CSTR with freecells

0.1

Single stage CSTR withencapsulated cells

0.1

0.5

Two stage dilute acidhydrolyzate

Glucose: 28,mannose: 6,galactose: 2

Single stage CSTR withocculating yeast

0.52

Concentrated sulphuricacid hydrolyzedconiferous wood,supplemented withcorn steep liquor,KH2PO4, MgSO4, CaCl2

Glucose: 106,mannose 11

Single stage tower typereactor with occulatingyeast

0.3

Two tower type reactors inseries

0.2ionh1)

Ethanol(g L1)

Celldryweight(g L1)

Hexoseconversion(%)

Ethanolproductivity(g L1 h1]

References

0.1 Washout Brandberg et al.(2005)

17 6 99 1.6

n.d. 1.9 76 1.7 Brandberg et al.(2007)5.7 94 2.3

echnology 101 (2010) 48624874dilution rate from 1.1 to 4.2 g L1 h1, and cell viability was highunder all conditions with values between 77% and 90%. In contrast,if the hydrolyzate was continuously fermented by free yeast cellsat a dilution rate of 0.1 h1, 90% of the hexoses were converted,the ethanol yield was low at 0.34 g g1, and cell viability was only25%.

To date, almost no industrial processes employ encapsulated orimmobilized cells due to worries about long term stability and theadditional costs for immobilization (Sanchez and Cardona, 2008;Verbelen et al., 2006). In addition, there are signicant challengesin applying immobilized cells at the scales typical of ethanol pro-duction processes, and they will become even more challengingto apply when using hydrolyzates containing lignin and other sol-ids. However, the natural form of immobilization cell occulation has greater potential to be employed on an industrial scale and isalready used for cane sugar fermentations. Thus, Purwadi et al.(2007) investigated the fermentability of the dilute acid hydroly-zate described above with a occulating yeast strain of S. cerevisiae

n.d. n.d. 3.45

n.d. n.d. 86 (glucose),72 (mannose)

n.d. Taherzadeh et al.(2001)

79 (glucose),55 (mannose)

n.d. n.d. 100 2.1 Purwadi andTaherzadeh(2008)

77 6.6

92 7.4

(based oneactor only)

99

98 2.754 6

87 5.5

n.d. n.d. 90 (hexoses) 0.86 Talebnia andTaherzadeh(2006)95 (glucose),

98 (mannose)1.14

71 (glucose),79 mannose

4.2

n.d. 2330 96 n.d. Purwadi et al.(2007)

ca. 60 n.d. 82 20 Tang et al. (2006)

56 (1),63 (2)

100 12.6

rce Tisolated from an ethanol plant. One stage continuous cultivationsof unsupplemented hydrolyzate with the occulating yeast werecarried out successfully at dilution rates of 0.13, 0.24, 0.38, and0.52 h1 with cell concentrations in the reactor between 23 and30 g L1 achieved by separation from the efuent by a settler. Evenat the highest dilution rate, 96% of the sugars were assimilated, andethanol yields based on consumed sugars decreased with increas-ing dilution rate from 0.46 to 0.42 g g1. HMF and furfural were al-most completely removed. Cell growth rate was controlled bylimiting the nitrogen source in the medium, so that ethanol wasproduced over 37 retention times at a constant cell density. Ifnitrogen was supplemented, cell density quickly reached concen-trations of more than 50 g L1. Continuous cultivation was also car-ried out using two bioreactors connected in series, where cellswere recycled from the last reactor to the rst with dilution ratesof 0.24, 0.38 and 0.52 h1 based on the total volume of the reactionsystem. The results were similar to those for single stage experi-ments with respect to sugar assimilation, ethanol yields, and cellconcentrations.

A thermotolerant occulating yeast strain was also investigatedfor the fermentation of a sugar stream produced by hydrolyzingconiferous wood with concentrated sulphuric acid to obtain a solu-tion containing 106 g L1glucose, 11 g L1 mannose, 3 g L1galact-ose, and 12 g L1 xylose. The pH of the acid hydrolyzate wasadjusted to 3.0 by adding Ca(OH)2 and holding overnight beforeadding further media components. Media optimization experi-ments showed that yeast extract and peptone could be replacedby 1% corn steep liquor supplemented with 0.05% KH2PO4, 0.05%MgSO47H2O, and 0.01% CaCl22H2O. Acid hydrolyzate fermentedat temperatures below 35 C with a dilution rate of 0.3 h1 resultedin an ethanol productivity of 20 g L1 h1 and a yield based on totalsugars in the feed of 82% and compounds in the hydrolyzate didnot signicantly inhibit the fermentations. In a scale up experi-ment, two 4.5 L tower reactors were connected in series to fermentacid hydrolyzate. Growth of contaminating bacteria was repressedat pH 4 but not 4.5 for operation at 35 C with a dilution rate of0.2 h1 without adding K2S2O5. However, at this condition, wash-out occurred at a dilution rate of 0.3 h1. At a dilution rate of0.2 h1, ethanol concentrations of 56 and 63 g L1 were measuredin reactors 1 and 2, respectively, corresponding to an overall pro-ductivity of 12.6 g L1 h1. All hexose sugars were converted com-pletely in reactor 2. The system was run for 30 days (Tang et al.,2006).

4.3. Co-fermentation of pentoses and hexoses in solution

To be economical, all sugars, both pentoses and hexoses, mustbe converted to ethanol or other useful products. However, S. cere-visiae does not ferment pentoses, and other strains have to be em-ployed for fermentation of mixtures containing arabinose andxylose (see Table 3). Generally, natural pentose fermenting organ-isms are less inhibited by hemicellulose hydrolyzate than recombi-nant gram negative xylose fermenting strains such as E. coli, K.oxytoca or Zymomonas mobilis (Georgieva and Ahring, 2007). Forexample, although Pichia stipitis yeast can naturally ferment pen-tose sugars to ethanol, hexose sugars are used preferentially, andpentose uptake is competitively inhibited by hexoses. Thus, pen-tose fermentation is only possible at very low glucose concentra-tions. In addition, microaerophilic conditions are required, whichare difcult to maintain in large scale systems, and even thenyields are low. Because a series of reactors has the advantage overa single CSTR in that optimal conditions for each vessel can be ad-justed separately, Grootjen et al. (1991) applied a three reactor sys-

S. Brethauer, C.E. Wyman / Bioresoutem with P. stipitis in which they adjusted the residence time suchthat glucose was depleted in the last reactor and optimal pentoseconversion took place. A synthetic medium containing 10 g L1 xy-lose and 40 g L1 glucose was employed to demonstrate this ap-proach, and at a dilution rate of 0.027 h1 with controlledaeration of the rst two reactors, almost complete mixed sugarconversion was possible with the ethanol productivity being0.51 g L1 h1. However, if the dilution rate was doubled, only20% of the xylose could be converted (Table 3).

Genetically engineered Z. mobilis strains were also investigatedfor their co-fermentation potential. Lawford et al. (1998) used asingle stage CSTR to rst characterize the continuous co-fermenta-tion abilities of the strain with a synthetic medium containing8 g L1 glucose and 40 g L1 xylose. With dilution rates between0.04 and 0.1 h1, approximately constant ethanol yields of 2022 g L1 were achieved, and xylose conversion remained high withvalues between 83% and 93%. Furthermore, long term adaptationexperiments were performed employing a detoxied dilute acidpretreatment hydrolyzate of poplar sawdust, starting with an ini-tial hydrolyzate concentration of 10%. The co-fermentation abili-ties could be preserved, while the concentration of hydrolyzatewas raised over a time period of 50 days to 35%, a level whichwas stated to be substantially inhibiting in batch fermentationthereby showing that long term continuous fermentation is a valu-able tool to adapt the microorganism to toxic conditions. Krishnanet al. (2000) employed an immobilized recombinant Z. mobilisstrain to co-ferment xylose and glucose. In batch experiments itwas shown that fermentation results were comparable for concen-trated acid rice straw hydrolyzate and a synthetic control medium.Continuous experiments in a uidized bed reactor were performedwith the control medium at dilution rates between 0.24 and0.5 h1 and excellent volumetric ethanol productivities of 6.515.3 g L1 h1 were achieved. However, xylose conversion variedfrom 10% to 90%.

Georgieva and Ahring (2007) investigated the ability of thethermophilic anaerobic bacterium Thermoanaerobacter BG1L1 thatwas genetically modied to be lactate dehydrogenase decient toferment xylose in undetoxied hydrolyzate. The National Renew-able Energy Laboratory (NREL) pretreated corn stover in dilute sul-phuric acid at a solids concentration of 30% to produce hydrolyzate(the liquid phase) containing 16 g L1 glucose and 68 g L1 xylose.For the fermentations, the hydrolyzate was diluted 1:2 to 1:12, cor-responding to total solids contents (TS) of 2.515%, and supple-mented with yeast extract, minerals, trace metals, vitamins, andNa2S. The strain was immobilized on a granulated carrier materialand employed in a continuous uidized bed reactor system oper-ated at 70 C and pH 7.0 with a residence time of 2 days for allbut the highest TS concentration, for which the residence timewas increased to 3 days. After start-up of the reactor with syntheticmedia, the hydrolyzate concentration was increased stepwise afterreaching steady state to allow adaptation of the microorganism toinhibitors in the feed. Independent of the substrate concentration,ethanol yields based on sugars consumed were in the range of0.390.42 g g1 (7682% of the theoretical possible), and the max-imum ethanol concentration was 10.4 g L1. Xylose consumptionranged from 89% to 98% at 2.510% TS but dropped to 67% for15% TS. Overall sugar conversion to ethanol was between 70%and 77% at 2.510% TS and 52% at 15% TS. The reactor was operatedfor 135 days without any bacterial contamination.

An identical fermentation system was used to convert wet ex-ploded wheat straw hydrolyzate to ethanol. A 200 g L1 wheatstraw suspension was heated to 170 C, and hydrogen peroxidewas added to reach a nal oxygen concentration of 3% per g ofdry matter content. Pretreatment was stopped as soon as all hydro-gen peroxide had reacted with the biomass, and the resulting sus-pension was diluted with water to a concentration equivalent to

echnology 101 (2010) 48624874 4869dry matter concentrations of 312%. Then, the pH was adjustedto 5.0 with sodium hydroxide, and the mixtures were autoclaved,which increased inhibitor concentrations further. Following enzy-

Dilurate(h

0.02

0.02

0.02

0.10.2

0.24

0.25

0.04

rce Tmatic hydrolysis of the slurry, the remaining solids were removed,and the nal fermentation media was made up as described above.No detoxication or washing steps were included. The glucose con-centration in the wet exploded wheat straw hydrolyzate (WEH)varied between 9 and 27 g L1 and the xylose concentration be-tween 3 and 14 g L1. With increasing WEH concentration, the eth-anol concentration in the efuent increased from 4.6 to 14.4 g L1,

Table 3Continuous co-fermentation of pentoses and hexoses in solution.

Medium Sugarconcentration(g L1)

Reactor type

Growth medium with yeast extract andmineral salts

Glucose: 40,xylose: 10

Three CSTRs inseries

Hydrolyzate of dilute sulphuric acidpretreatment of corn stover at a solidsconcentration of 30%, diluted 1:2 to1:12, supplemented with yeast extract,minerals, vitamins and Na2S

Glucose: 4.4,xylose: 21.1

Fluidized bedreactor withimmobilized cells

Enzymatic hydrolyzate of whole wetexploded wheat straw slurry,supplemented with yeast extract,minerals, vitamins and Na2S

Glucose: 22.2,xylose: 11.2

Fluidized bedreactor withimmobilized cells

Dilute sulphuric acid pretreated sprucehydrolyzate, supplemented with yeastextract, (NH4)2SO4, K2HPO4, CaCl2,MgSO4, vitamins, and trace metalsolution

Glucose: 13.5,mannose: 20.5,xylose: 7.9

Single stage CSTRCSTR with cellretention

Synthetic medium with yeastextract, KCl and K2HPO4

Glucose: 49.9,xylose: 12.9

Fluidized bedreactor withimmobilized cellsGlucose: 68.8,

xylose: 23.135% hydrolyzate of dilute sulphuric

acid pretreatment, detoxied byoverliming, supplemented withcorn steep liquor

Glucose: 8,xylose: 40

Single stage CSTR

4870 S. Brethauer, C.E. Wyman / Bioresoucorresponding to ethanol yields based on consumed sugars in therange of 0.390.42 g g1 (7683% of theoretical). Provided the res-idence time was increased from 2 to 3 days for the highest WEHconcentration, glucose utilization was higher than 90% for alltested conditions, whereas xylose conversion was lower at 7280%. The total experiment lasted 143 days, and no contaminationof the reactor occurred (Georgieva et al., 2008).

The dimorphic lamentous fungus Mucor indicus (formerly M.rouxii) is a promising alternative to S. cerevisiae as it is capable ofxylose fermentation, is safe for humans, and produces ethanol fromhexoses with comparable yields and productivities. In this case,forestry residues mainly from spruce were treated at a solids con-centration of 33% with 0.5 g L1 sulphuric acid for 10 min at 15 barto produce a liquid containing 44.5 g L1 sugars (20.5 g L1 man-nose, 13.5 g L1 glucose, 7.9 g L1 xylose and 3.2 g L1 galactose).This solution was then supplemented with yeast extract,(NH4)2SO4, K2HPO4, CaCl2, MgSO4, antifoam, vitamin solution, andtrace metal solution for continuous cultivations in a standard stir-red tank reactor with synthetic glucose media as well as hydroly-zate at 32 C and pH 5.5. With the pure glucose medium,conversion was more than 98%, an ethanol yield of 0.41 g g1

was reached at dilution rates of 0.1 and 0.2 h1, and washout oc-curred at a dilution rate of 0.3 h1. When hydrolyzate was em-ployed, the fermentations failed, and cells were washed out at adilution rate as low as 0.1 h1. Thus, a new bioreactor was con-structed containing a stainless steel net with 0.5 0.5 mm squareholes placed at the top where uid left the reactor. During contin-uous cultivation, lamentous biomass quickly covered the screen,allowing permeation of liquid while partly retaining cells to pre-vent washout. With this conguration, continuous fermentationsof hydrolyzate were possible up to a dilution rate of 0.3 h1. Thebest results were achieved at a dilution rate of 0.2 h1, with 8799% hexose conversion, only 26% xylose conversion, and an ethanolconcentration of 17 g L1, corresponding to a yield of 0.45 g/g and aproductivity of 3.3 g L1 h1 (Karimi et al., 2008).

4.4. Continuous simultaneous saccharication and fermentation

tion

1)

Strain Ethanol(g L1)

Sugar conversion(%)

Ethanolproductivity(g L1 h1)

References

7 Pichia stipitis 18.8 Almost complete 0.51 Grootjenet al. (1991)

ThermoanaerobacterBG1L1

9.1 Glucose: 91,xylose 89

0.18 Georgievaand Ahring(2007)

ThermoanaerobacterBG1L1

11.6 Glucose: 93,xylose 76

0.23 Georgievaet al. (2008)

Mucor indicus Washout Karimi et al.(2008)17 Hexoses:8799,

xylose 263.3

Zymomonas mobilisCP4(pZB5)

26.9 Glucose: 99.8,xylose: 91.5

6.5 Krishnanet al. (2000)

34.5 Glucose: 99,xylose: 36.4

8.6

Zymomonas mobilis39676:pZB4L

ca. 20 Xylose: 92.5 0.8 Lawfordet al. (1998)

echnology 101 (2010) 486248744.4.1. Experimental dataSSF is one of the most promising process congurations to con-

vert pretreated lignocellulosic biomass to ethanol as it reducesend-product inhibition of cellulase by sugars. However, only lim-ited information has been published about continuous SSF, pre-sumably due to the inherent experimental challenges ofhomogenous delivery of solid substrate, the extended run timesneeded, and the complexities of continuous experimental systems.On the other hand, batch systems are hampered by mixing prob-lems at high solid substrate loadings, which can be avoided byoperation of a CSTR with high conversion of insolubles to ethanol.Furthermore, in a batch reactor, the high amounts of b-glucosidaseadded are only required at the beginning of the reaction when cel-lobiose production is highest, while we hypothesize that b-glucosi-dase loading can be reduced in a continuous system becausecellobiose production slows with conversion. For consolidated bio-processing by strains such as Clostridium thermocellum, the highercell densities possible in a CSTR allows higher cellulase productionlevels.

In a pioneering study, South et al. (1993) presented experimen-tal results for continuous conversion of pretreated hardwood ourto ethanol for two different systems: SSF that utilized the D5Astrain of S. cerevisiae in combination with cellulase enzymes anddirect microbial conversion (DMC) with the cellulose fermentingstrain C. thermocellum. Hardwood our was pretreated for 10 s at220 C with 1% H2SO4, with typical particle sizes of about0.05 mm. A single stage CSTR was set up with a working volumeof 1.25 L, and solid substrate was fed to the reactor from a 20 L feedreservoir by a progressing cavity pump that intermittently addedbiomass to achieve the target ow rate. The SSF system was runwith cellulose loadings from 5 to 60 g L1 at cellulase loadings

ranging from 7 to 25 FPU g1 cellulose but seemed relatively insen-sitive to variations in enzyme loadings and substrate feed concen-trations. With a cellulose feed of 61 g L1, an enzyme loading of12 FPU g1, and a dilution rate of 0.02 h1, an ethanol concentra-tion of 21 g L1 was reached, corresponding to a conversion of60% and a productivity of 0.41 g L1 h1 (Table 4). Generally, con-version in a CSTR was found to be 813% lower than in batch reac-tors at residence or reaction times between 1 and 3 days. Atcomparable substrate concentrations (45 g L1) and residencetimes (12 14 h), substrate conversion in the C. thermocellum sys-tem was 77%, signicantly higher than the 31% achieved in theSSF system.

In a later paper, a semi-continuous laboratory scale fermenta-tion system was applied to convert paper sludge to ethanol in anSSF process with the goal of reaching a nal ethanol concentration

a cascade of three equally sized vessels, with the mean residencetime in each fermentor being 36 h to give an overall dilution rateof 0.009 h1. Although substrate concentration, enzyme loading,and general performance were not specied, an ethanol concentra-tion of approximately 40 g L1 was measured early in the runs be-fore bacterial contamination with Lactobacilli was reported to be aproblem. After a run time of 400 h, lactic acid concentration in-creased to nearly 25 g L1, but liquid culturing techniques detectedcontamination after only 168 h. Concomitantly to the increase inlactic acid concentration, the concentration of arabinose, a sugarwhich cannot be fermented by the yeast employed, decreased from25 g L1 to 2 g L1, mirroring the lactic acid curve. Addition of10 ppm penicillin every 12 h combated this problem temporarilybut was overcome after 100 h. However, three doses of 10 ppm vir-giniamycin over 36 h decreased lactic acid concentrations to pre-

Enz(FPcel

12

2015n.s

10

S. Brethauer, C.E. Wyman / Bioresource Technology 101 (2010) 48624874 4871of more than 4 wt.% to realize cost- and energy-effective distilla-tion. Mixing of unreacted sludge proved to be almost impossibleat the solids concentrations necessary to reach the targeted etha-nol concentration in a batch system, making it necessary to employa fed-batch or continuous process to reduce the viscosity and as-sure reasonable mixing. A solid feeding device was developed, forwhich a motor driven plunger advanced a plug of paper sludge intothe reactor at 12 h feeding intervals. The systemwas operated witha residence time of 4 days (corresponding to a dilution rate of0.01 h1) with a cellulase loading of 1520 FPU g1 cellulose, andS. cervisiae D5A was used to ferment the sugars to ethanol. Afterdifculties with the rst generation system, a retrotted design en-abled stable operation for more than one month. In one run, cellu-lose fed at a concentration of 82 g L1 (corresponding to a solidconcentration of about 12%) achieved an average conversion of92% and 42 g L1 ethanol with an enzyme loading of 20 FPU g1.Based on the reported data, the productivity was calculated to be0.42 g L1 h1. However, at a higher cellulose loading of 120 g L1

and lower cellulase concentration of 15 FPU g1, only 74% conver-sion was reached, corresponding to an ethanol concentration of50 g L1 (Fan et al., 2003).

The National Renewal Energy Laboratory (NREL) conceptualizedtwo base case scenarios for large scale cellulosic ethanol produc-tion based on continuous SSF operation in reactor trains consistingof 56 vessels (Aden et al., 2002;Wooley et al., 1999). Furthermore,a bioethanol pilot plant, equipped with capabilities for feedstockhandling, continuous dilute acid pretreatment, yeast inoculumproduction, and SSF with commercially available cellulase, wasput into operation in 1995 (Schell et al., 2004). The plant was de-signed to process 900 kg dry biomass per day, and initial demon-stration runs were performed using corn ber, a corn wet-millingby-product, as the substrate. After pretreatment, the materialwas neutralized and sent directly to the rst 9000 L fermentor of

Table 4Continuous SSF.

Substrate Celluloseconcentration(g L1)

Reactor type Dilutionrate (h1)

Solids of dilute sulphuricacid pretreatedhardwood our

61 Single stageCSTR

0.02

Paper sludge 82 CSTR with 12 hfeed intervals

0.01120

Whole slurry of dilute acidpretreated corn ber

n.s. 3 CSTRs of9000 L each inseries

0.009

Wastepaper 50 2 equally sized 0.01875

CSTRs in seriescontamination levels. Adding ethanol to the fermentor at the endof the run to increase its concentration to 50 g L1 failed as a strat-egy to reverse contamination. The authors concluded that microor-ganisms that consume arabinose or any other biomass-derivedsugar not utilized by the primary fermentative microorganismpose a signicant challenge to the desired yeast mono culture. Fur-thermore, high levels of contaminant compete with the yeast forglucose, especially as the by-products such as lactic and acetic acidare inhibitory to the fermenting strain. Thus, they concluded that amicroorganism that can ferment all sugars in the feed stream is vi-tal to avoiding invasion by unwanted microbes. Also, separate sac-charication and fermentation might be advantageous because ofthe ability to operate saccharication under conditions not favour-able to contaminant growth and the much shorter fermentationresidence times with sugars than possible with cellulose hydroly-sis. Alternatively, a microorganism with outstanding tolerance tothe fermentation inhibitors produced during pretreatment couldbe employed (Schell et al., 2007).

4.4.2. Kinetic modellingKinetic modelling is a useful tool to gain a deeper insight into

a reaction system if performance of a process can be predictedunder various conditions, but although many models have beendeveloped to describe enzymatic hydrolysis of cellulose or SSFin batch systems, only few have been applied to simulate contin-uous processes. South et al. (1995) sought to develop a kineticmodel of continuous SSF to predict cellulose conversion over awide range of conditions, such as cellulose loading, enzyme load-ing, and reaction time but with minimal complexity. In this case,cellulose was assumed to be hydrolyzed to cellobiose which fur-ther broke down to glucose by the action of b-glucosidase, withno direct conversion of cellulose into glucose considered. Cellu-lase adsorption onto cellulose and lignin was described according

yme loadingU g1

lulose

Ethanol(g L1)

Celluloseconversion(%)

Ethanolproductivity(g L1 h1)

References

21 60 0.42 South et al.(1993)

42 92 0.42 Fan et al.(2003)50 74 0.5

. 40 (lastreactor)

n.s. 0.36 Schell et al.(2007)

8 and 12 43 0.225 Philippidis

and Hatzis(1997)

could result from better pretreatment methods was predicted to

attention: (1) continuous SSF of unwashed substrates, (2) continu-

rce Tto the Langmuir model, and the cellulose hydrolysis rate was ta-ken to be proportional to the concentration of the enzymesub-strate complex divided by the specic capacity of the substratefor cellulase. Inhibition by cellobiose and ethanol was also inte-grated into the model, and the conversion of cellobiose wasmodelled to follow MichaelisMenten kinetics with inhibitionby the glucose released. Cell growth followed Monod kineticsand was taken to be inhibited by ethanol. Furthermore,the decreasing reactivity of cellulose with conversion was de-scribed by a cellulose hydrolysis rate equation of the formk(x) = k * (1 x)n + c in which k is the hydrolysis rate constant,x is the fractional cellulose conversion, n is an exponent of thedeclining substrate reactivity, and c is the conversion indepen-dent component in the rate function. To describe continuousSSF, a particle population model was used to account for the ef-fect of the distribution in particle exit age on the change in par-ticle reactivity with conversion. Comparison of the populationmodel to a soluble substrate model assuming a uniform meanresidence time with experimental data showed that the chosenapproach was necessary to describe the heterogeneous reactionaccurately. Overall the model predicted the experimentallydetermined conversions in a CSTR quite well with a root meansquared difference of 5.2%.

The model was then extended to describe potentially more opti-mal reactor congurations than a single stage CSTR, such as a cas-cade of fermentors or a system that partially retained solids toincrease their residence time relative to liquid residence time.The resulting predictions revealed that conversion in a single stageCSTR was lower than in a batch reactor but that a cascade of 5CSTRs could attain an equal conversion of 95% after 5 days. Retain-ing the solids for a mean residence time 1.5 times longer than thehydraulic residence time reduced the time to reach 90% conversionfrom 2.5 days in a batch reactor to 1.2 days for the ve stage cas-cade, a 47% reduction in overall reactor volume compared to abatch system. Modelling the inuence of ethanol concentrationshowed that inhibition of cellulase was more important than inhi-bition of the fermenting organism for the SSF system, because thegrowth rate is far in excess of the hydrolysis rate (South and Lynd,1994).

Philippidis and Hatzis (1997) developed a kinetic model forsimultaneous saccharication and fermentation based on experi-mental data of a two stage continuous SSF of wastepaper contain-ing 50% cellulose and 20% lignin. The solid loading in theirexperiments was 100 g L1, and the dilution rate was set to0.01875 h1, corresponding to a total mean residence time of53 h in the two reactor system with 8 L working volumes for each.Applying an enzyme loading of 10 FPU g1 cellulose supplementedwith 20 CBU of b-glucosidase resulted in steady state ethanol con-centrations of approximately 8 g L1 in the rst stage and 12 g L1

in the second stage, giving an overall yield of 43% and a productiv-ity of 0.225 g L1 h1. The model assumed that cellulose was simul-taneously converted either directly to glucose or rst to cellobiosewhich was then converted by b-glucosidase to glucose. Glucosewas then catabolized to ethanol, cells, and CO2. The rate equationsfor cellulose decomposition included enzyme inhibition by glu-cose, cellobiose, and ethanol as well as an exponential decay termof the form ektime to account for the time-dependent decline inthe hydrolysis rate due to the loss of effective surface area of cellu-lose. The specic rate constants employed a MichaelisMentendependence on the cellulase concentration, cellobiose hydrolysiswas assumed to be inhibited by only glucose, and loss of enzymeactivity due to adsorption on lignin was accounted for in the spe-cic rate constant. The model t the continuous data well for both

4872 S. Brethauer, C.E. Wyman / Bioresousteady state operation and even for the transition from batch start-up, providing more condence in the model and its usefulness fortheir theoretical parameter study. They then applied the model toous fermentation of mixed pentose and hexose sugars, (3) optimaldesign of cascade systems for continuous SSF, (4) techniques toavoid and overcome bacterial contamination, (5) the effect of oper-ating parameters on reactor stability and washout of organisms,and (6) tradeoffs in yields for continuous vs. batch operations atlower, more economically attractive enzyme loadings. To ensureapplicability to commercial systems, conditions at the laboratoryscale should be chosen that are compatible with an industrial costeffective process, and it should be kept in mind that such factors asmedia additives, fermentation pH, sterilization temperature, andcell density strongly inuence performance.

Acknowledgements

The authors are grateful to the FordMotor Company for sponsor-ing the Chair in Environmental Engineering at the Bourns College ofEngineering at the University of California, Riverside and for provid-ing funds for our research on cellulosic ethanol. Our work onadvancing continuous fermentations is made possible throughfunds from the National Institute of Standards and Technologyincrease ethanol yields from 46% to 59% (a 27% increase), while 5and 10-fold improvements enhanced yields to 70% and 73%. Thus,saturation was projected to be reached quickly, with the result thatother kinetic parameters must be improved to obtain higher etha-nol yields. One of the critical parameters was the specic rate lossfactor k in cellulose reactivity, with a reduction from 0.02 in thebase case to 0.005 boosting ethanol yield to 65% at the base casedilution rate. Microbial parameters of maximal growth rate andethanol resistance had a small effect on performance, supportingthe notion that enzymatic hydrolysis is the rate limiting step forSSF.

5. Summary and conclusions

Continuous fermentations generally show higher productivitiesthan batch processes and exhibit reduced vessel down time forcleaning and lling, thereby enabling a smaller plant size at anequal annual capacity. For conversion of lignocellulosic biomassto ethanol, continuous fermentations provide several additionaladvantages. First, costly conditioning of inhibitory compounds re-leased into acid hydrolyzates can be omitted or at least reduceddue to the in situ detoxication abilities of yeast, especially at highcell densities achievable in cell retention systems such as immobi-lization and self-occulation. By properly adjusting the residencetime, the viscosity of the solids can be reduced substantially, allow-ing continuous enzymatic hydrolysis systems to handle highereffective solids concentrations than could be mixed initially in abatch operation. Furthermore, we hypothesize that b-glucosidaseloading can be reduced in a continuous system as the burst of cel-lobiose production is avoided at steady state. However, experiencewith continuous enzymatic hydrolysis of cellulosic biomass is ex-tremely limited, and several important research topics requireplot ethanol yield as a function of dilution rate to obtain a bellshaped curve with a distinct maximum. Thus, a given ethanol yieldcould be reached at two dilution rates, with the higher one moreattractive in that a much higher productivity was realized. Addingethanol to avoid bacterial contamination was then shown to de-crease the ethanol yield from 46% if no ethanol was recycled to31% if 80 g L1 of ethanol was fed to the reaction. Then, the effectof doubling the specic cellulose hydrolysis rate constants that

echnology 101 (2010) 48624874through our partnershipwithDartmouth College via awardNumber60NANB1D0064 and the USDA National Research Initiative Com-petitive Grants Program through Contract 2008-35504-04596, and

rce Twegreatly appreciate the support by each of these agencies.We alsoacknowledge support by the Bourns College of Engineering at theUniversity of California at Riverside through the Center for Environ-mental Research and Technology (CE-CERT) and the Chemical andEnvironmental Engineering Department.

References

Aden, A., Ruth, M., Ibsen, K., Jechura, J., Neeves, K., Sheehan, J., Wallace, K.,Montague, L., Slayton, A., Lukas, J., 2002. Lignocellulosic Biomass to EthanolProcess Design and Economics Utilizing Co-current Dilute Acid Prehydrolysisand Enzymatic Hydrolysis for Corn Stover, National Renewable EnergyLaboratory, Golden, CO, NREL/TP-510-32438.

Bai, F.W., Anderson, W.A., Moo-Young, M., 2008. Ethanol fermentation technologiesfrom sugar and starch feedstocks. Biotechnol. Adv. 26 (1), 89105.

Bothast, R.J., Schlicher, M.A., 2005. Biotechnological processes for conversion of corninto ethanol. Appl. Microbiol. Biotechnol. 67 (1), 1925.

Brandberg, T., Karimi, K., Taherzadeh, M.J., Franzen, C.J., Grustafsson, L., 2007.Continuous fermentation of wheat-supplemented lignocellulose hydrolysatewith different types of cell retention. Biotechnol. Bioeng. 98 (1), 8090.

Brandberg, T., Sanandaji, N., Gustafsson, L., Franzen, C.J., 2005. Continuousfermentation of undetoxied dilute acid lignocellulose hydrolysate bySaccharomyces cerevisiae ATCC 96581 using cell recirculation. Biotechnol.Prog. 21 (4), 10931101.

Cardona, C.A., Sanchez, O.J., 2007. Fuel ethanol production: process design trendsand integration opportunities. Bioresource Technol. 98 (12), 24152457.

Claassen, P.A.M., van Lier, J.B., Contreras, A.M.L., van Niel, E.W.J., Sijtsma, L., Stams,A.J.M., de Vries, S.S., Weusthuis, R.A., 1999. Utilisation of biomass for the supplyof energy carriers. Appl. Microbiol. Biotechnol. 52 (6), 741755.

de Gooijer, C.D., Bakker, W.A.M., Beeftink, H.H., Tramper, J., 1996. Bioreactors inseries: an overview of design procedures and practical applications. EnzymeMicrob. Technol. 18 (3), 202219.

Dorer, J., Amorim, H.V., 2007. Applied bioethanol technology in Brazil. SugarIndustry (Zuckerindustrie) 132 (9), 694697.

Fan, Z.L., South, C., Lyford, K., Munsie, J., van Walsum, P., Lynd, L.R., 2003. Conversionof paper sludge to ethanol in a semicontinuous solids-fed reactor. BioprocessBiosyst. Eng. 26 (2), 93101.

Farrell, A.E., Plevin, R.J., Turner, B.T., Jones, A.D., OHare, M., Kammen, D.M., 2006.Ethanol can contribute to energy and environmental goals. Science 311, 506508.

Galbe, M., Zacchi, G., 2002. A review of the production of ethanol from softwood.Appl. Microbiol. Biotechnol. 59 (6), 618628.

Gauss, W.F., Suzuki, S., Takagi, M., 1976. Manufacture of alcohol from cellulosicmaterials using plural ferments. US patent 3,990,944.

Georgieva, T.I., Ahring, B.K., 2007. Evaluation of continuous ethanol fermentation ofdilute-acid corn stover hydrolysate using thermophilic anaerobic bacteriumThermoanaerobacter BG1L1. Appl. Microbiol. Biotechnol. 77 (1), 6168.

Georgieva, T.I., Mikkelsen, M.J., Ahring, B.K., 2008. Ethanol production from wet-exploded wheat straw hydrolysate by thermophilic anaerobic bacteriumThermoanaerobacter BG1L1 in a continuous immobilized reactor. Appl.Biochem. Biotechnol. 145 (13), 99110.

Ghose, T.K., Tyagi, R.D., 1979. Rapid ethanol fermentation of cellulose hydrolysate 1:batch versus continuous systems. Biotechnol. Bioeng. 21 (8), 13871400.

Godoy, A., Amorim, H.V., Lopes, M.L., Oliveira, A.J., 2008. Continuous and batchfermentation processes: advantages and disadvantages of these processes in theBrazilian ethanol production. Int. Sugar J. 110 (1311), 175182.

Gray, K.A., Zhao, L.S., Emptage, M., 2006. Bioethanol. Curr. Opin. Chem. Biol. 10 (2),141146.

Grootjen, D.R.J., Jansen, M.L., Vanderlans, R., Luyben, K., 1991. Reactors in series forthe complete conversion of glucose xylose mixtures by Pichia stipitis andSaccharomyces cerevisiae. Enzyme Microb. Technol. 13 (10), 828833.

Guerreiro, M.A., Andrietta, S.R., Maugeri, F., 1997. Expert system for the design of anindustrial fermentation plant for the production of alcohol. J. Chem. Technol.Biot. 68 (2), 163170.

Hahn-Hgerdal, B., Galbe, M., Gorwa-Grauslund, M.F., Liden, G., Zacchi, G., 2006.Bio-ethanol the fuel of tomorrow from the residues of today. TrendsBiotechnol. 24 (12), 549556.

Hendriks, A.T.W.M., Zeeman, G., 2009. Pretreatments to enhance the digestibility oflignocellulosic biomass. Bioresource Technol. 100, 1018.

IPCC, 2007. Summary for policymakers. In: Solomon, S., Qin, D., Manning, M., Chen,Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Ed.), Climate Change 2007:The Physical Science Basis. Contribution of Working Group I to the FourthAssessment Report on the Intergovernmental Panel on Climate Change,Cambridge University Press, Cambridge, United Kingdom and New York, NY,USA.

Karimi, K., Edebo, L., Taherzadeh, M.J., 2008. Mucor indicus as a biolter andfermenting organism in continuous ethanol production from lignocellulosichydrolyzate. Biochem. Eng. J. 39 (2), 383388.

Krishnan, M.S., Blanco, M., Shattuck, C.K., Nghiem, N.P., Davison, B.H., 2000. Ethanolproduction from glucose and xylose by immobilized Zymomonas mobilisCP4(pZB5). Appl. Biochem. Biotechnol. 8486, 525541.

S. Brethauer, C.E. Wyman / BioresouLawford, H.G., Rousseau, J.D., Mohagheghi, A., McMillan, J.D., 1998. Continuousculture studies of xylose-fermenting Zymomonas mobilis. Appl. Biochem.Biotechnol. 7072, 353367.Lee, J.M., Pollard, J.F., Coulman, G.A., 1983. Ethanol fermentation with cell recycling computer simulation. Biotechnol. Bioeng. 25 (2), 497511.

Lee, W.G., Park, B.G., Chang, Y.K., Chang, H.N., Lee, J.S., Park, S.C., 2000. Continuousethanol production from concentrated wood hydrolysates in an internalmembrane-ltration bioreactor. Biotechnol. Prog. 16 (2), 302304.

Lynd, L.R., Weimer, P.J., van Zyl, W.H., Pretorius, I.S., 2002. Microbial celluloseutilization: fundamentals and biotechnology. Microbiol. Mol. Biol. Rev. 66, 506557.

Maiorella, B.L., Blanch, H.W., Wilke, C.R., 1984. Economic evaluation of alternativeethanol fermentation processes. Biotechnol. Bioeng. 26 (9), 10031025.

Mosier, N., Wyman, C., Dale, B., Elander, R., Lee, Y.Y., Holtzapple, M., Ladisch, M.,2005. Features of promising technologies for pretreatment of lignocellulosicbiomass. Bioresource Technol. 96 (6), 673686.

Olsson, L., Hahn-Hgerdal, B., 1996. Fermentation of lignocellulosic hydrolysates forethanol production. Enzyme Microb. Technol. 18 (5), 312331.

Palmqvist, E., Galbe, M., Hahn-Hgerdal, B., 1998. Evaluation of cell recycling incontinuous fermentation of enzymatic hydrolysates of spruce withSaccharomyces cerevisiae and on-line monitoring of glucose and ethanol. Appl.Microbiol. Biotechnol. 50 (5), 545551.

Philippidis, G.P., Hatzis, C., 1997. Biochemical engineering analysis of criticalprocess factors in the biomass-to-ethanol technology. Biotechnol. Prog. 13 (3),222231.

Purwadi, R., Brandberg, T., Taherzadeh, M.J., 2007. A possible industrial solution toferment lignocellulosic hydrolyzate to ethanol: continuous cultivation withocculating yeast. Int. J. Mol. Sci. 8 (9), 920932.

Purwadi, R., Taherzadeh, M.J., 2008. The performance of serial bioreactors in rapidcontinuous production of ethanol from dilute-acid hydrolyzates usingimmobilized cells. Bioresource Technol. 99 (7), 22262233.

Sanchez, O.J., Cardona, C.A., 2008. Trends in biotechnological production of fuelethanol from different feedstocks. Bioresource Technol. 99 (13), 52705295.

Schell, D.J., Dowe, N., Ibsen, K.N., Riley, C.J., Ruth, M.F., Lumpkin, R.E., 2007.Contaminant occurrence, identication and control in a pilot-scale corn ber toethanol conversion process. Bioresource Technol. 98 (15), 29422948.

Schell, D.J., Riley, C.J., Dowe, N., Farmer, J., Ibsen, K.N., Ruth, M.F., Toon, S.T.,Lumpkin, R.E., 2004. A bioethanol process development unit: initial operatingexperiences and results with a corn ber feedstock. Bioresource Technol. 91 (2),179188.

Shapouri, H., Gallagher, P., 2005. USDAs 2002 ethanol cost-of-production survey.United States Department of Agriculture, Agricultural Economic Report Number841.

South, C.R., Hogsett, D.A., Lynd, L.R., 1993. Continuous fermentation of cellulosicbiomass to ethanol. Appl. Biochem. Biotechnol. 39, 587600.

South, C.R., Hogsett, D.A.L., Lynd, L.R., 1995. Modeling simultaneous saccharicationand fermentation of lignocellulose to ethanol in batch and continuous reactors.Enzyme Microb. Technol. 17 (9), 797803.

South, C.R., Lynd, L.R., 1994. Analysis of conversion of particulate biomass to ethanolin continuous solids retaining and cascade bioreactors. Appl. Biochem.Biotechnol. 4546, 467481.

Spindler, D.D., Wyman, C.E., Mohagheghi, A., Grohmann, K., 1987. Thermotolerantyeast for simultaneous saccharication and fermentation of cellulose toethanol. Appl. Biochem. Biotechnol. 17, 279293.

Taherzadeh, M.J., Karimi, K., 2007. Acid-based hydrolysis processes for ethanol fromlignocellulosic materials: a review. BioResources 2 (3), 472499.

Taherzadeh, M.J., Millati, R., Niklasson, C., 2001. Continuous cultivation of dilute-acid hydrolysates to ethanol by immobilized Saccharomyces cerevisiae. Appl.Biochem. Biotechnol. 95 (1), 4557.

Takagi, M., Abe, S., Suzuki, S., Emert, G.H., Yata, N., 1977. A Method for Production ofAlcohol Directly from Cellulose using Cellulase and Yeast, Indian Institute ofTechnology, Delhi, India.

Talebnia, F., Taherzadeh, M.J., 2006. In situ detoxication and continuous cultivationof dilute-acid hydrolyzate to ethanol by encapsulated S. cerevisiae. J. Biotechnol.125 (3), 377384.

Tang, Y., An, M., Liu, K., Nagai, S., Shigematsu, T., Morimura, S., Kida, K., 2006.Ethanol production from acid hydrolysate of wood biomass using theocculating yeast Saccharomyces cerevisiae strain KF-7. Process Biochem. 41(4), 909914.

Verbelen, P.J., De Schutter, D.P., Delvaux, F., Verstrepen, K.J., Delvaux, F.R., 2006.Immobilized yeast cell systems for continuous fermentation applications.Biotechnol. Lett. 28 (19), 15151525.

Wheals, A.E., Basso, L.C., Alves, D.M.G., Amorim, H.V., 1999. Fuel ethanol after 25years. Trends Biotechnol. 17 (12), 482487.

Wooley, R., Ruth, M., Sheehan, J., Ibsen, K., Majdeski, H., Galvez, A., 1999.Lignocellulosic biomass to ethanol process design and economics utilizing co-current dilute acid prehydrolysis and enzymatic hydrolysis: current andfuturistic scenarios, National Renewable Energy Laboratory, Golden, CO,NREL/TP-580-26157.

Wright, J.D., Wyman, C.E., Grohmann, K., 1987. Simultaneous saccharication andfermentation of lignocellulose: process evaluation. Appl. Biochem. Biotechnol.18, 7590.

Wyman, C.E., 2007. What is (and is not) vital to advancing cellulosic ethanol. TrendsBiotechnol. 25 (4), 153157.

Wyman, C.E., Dale, B.E., Elander, R.T., Holtzapple, M., Ladisch, M.R., Lee, Y.Y., 2005.Coordinated development of leading biomass pretreatment technologies.

echnology 101 (2010) 48624874 4873Bioresource Technol. 96 (18), 19591966.Wyman, C.E., Dale, B.E., Elander, R.T., Holtzapple, M., Ladisch, M.R., Lee, Y.Y.,

Mitchinson, C., Saddler, J.N., 2009. Comparative sugar recovery and

fermentation data following pretreatment of poplar wood by leadingtechnologies. Biotechnol. Prog. 25 (2), 333339.

Wyman, C.E., Decker, S.R., Himmel, M.E., Brady, J.W., Skopec, C.E., Viikari, L., 2004.Hydrolysis of cellulose and hemicellulose. In: Dumitriu, S. (Ed.),Polysaccharides: Structural Diversity and Functional Versatility, second ed.Marcel Dekker, Inc., New York, pp. 9951033, invited.

Wyman, C.E., Spindler, D.D., Grohmann, K., Lastick, S.M., 1986. Simultaneoussaccharication and fermentation with the yeast Brettanomyces clausenii.Biotechnol. Bioeng. Symp. 17, 221238.

Zanin, G.M., Santana, C.C., Bon, E.P.S., Giordano, R.C.L., de Moraes, F.F., Andrietta,S.R., Neto, C.C.D., Macedo, I.C., Fo, D.L., Ramos, L.P.and others, 2000. Brazilianbioethanol program. Appl. Biochem. Biotechnol. 8486, 11471161.

Zhang, Y.-H.P., Lynd, L.R., 2004. Toward an aggregated understanding of enzymatichydrolysis of cellulose: Noncomplexed cellulase systems. Biotechnol. Bioeng. 88(7), 797824.

Zhang, Y.H., Lynd, L.R., 2006. A functionally based model for hydrolysis of celluloseby fungal cellulase. Biotechnol. Bioeng. 94 (5), 888898.

4874 S. Brethauer, C.E. Wyman / Bioresource Technology 101 (2010) 48624874

Review: Continuous hydrolysis and fermentation for cellulosic ethanol productionIntroductionConcept of continuous fermentationsContinuous ethanol production from starch and sugar feedstocksIndustrial continuous ethanol production from sugar caneContinuous ethanol production from corn

C