Bioresource Technology 138 (2013) 259–265

Contents lists available at SciVerse ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Case Study

Butanol production by bioconversion of cheese whey in a continuouspacked bed reactor

0960-8524/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biortech.2013.03.180

⇑ Corresponding author. Tel.: +39 081 7682262: fax: +39 081 5936936.E-mail address: giolivie@unina.it (G. Olivieri).

F. Raganati a, G. Olivieri a,⇑, A. Procentese a, M.E. Russo b, P. Salatino a, A. Marzocchella a

a Dipartimento di Ingegneria Chimica, dei Materiali e della Produzione Industriale - Università degli Studi di Napoli Federico II, P.le V. Tecchio 80, 80125 Napoli, Italyb Istituto di Ricerche sulla Combustione – Consiglio Nazionale delle Ricerche, P.le V. Tecchio 80, 80125 Napoli, Italy

h i g h l i g h t s

� Butanol production by bioconversion of renewable resource.� Clostridium acetobutylicum fermentation of unsupplemented cheese whey.� Operation of a biofilm packed bed reactor.� Optimization of operating conditions.

a r t i c l e i n f o

Article history:Received 7 February 2013Received in revised form 25 March 2013Accepted 27 March 2013Available online 4 April 2013

Keywords:Cheese wheyButanolBiofilm fixed bed reactorABE

a b s t r a c t

Butanol production by Clostridium acetobutylicum DSM 792 fermentation was investigated. Unsupple-mented cheese whey was adopted as renewable feedstock. The conversion was successfully carried outin a biofilm packed bed reactor (PBR) for more than 3 months.

The PBR was a 4 cm ID, 16 cm high glass tube with a 8 cm bed of 3 mm Tygon rings, as carriers. It wasoperated at the dilution rate between 0.4 h�1 and 0.94 h�1.

The cheese whey conversion process was characterized in terms of metabolites production (butanolincluded), lactose conversion and biofilm mass. Under optimized conditions, the performances were:butanol productivity 2.66 g/Lh, butanol concentration 4.93 g/L, butanol yield 0.26 g/g, butanol selectivityof the overall solvents production 82 wt%.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction great deal of substrates: pentoses, hexoses, mono-, di- and polysac-

The socio-economic scenario of the beginning of the third mil-lennium revives the interest in a strategy for the bioconversion ofindustrial wastewaters in biofuels and bulk chemicals. In this con-text, acetone–butanol–ethanol (ABE) fermentation is being consid-ered as a way to upgrade renewable resources into valuable basechemicals and liquid fuels (Ezeji et al., 2007a; Dürre, 2007; Friedl,2012). Indeed, butanol offers several advantages over ethanol forgasoline–alcohol blending because of its high energy content, lowmiscibility with water, and low volatility (Bohlmann, 2007; Cas-cone, 2008). In addition, butanol can replace gasoline with no needto modify the current vehicle and engine technologies.

ABE is typically produced during the last stage of batch fermen-tation of some Clostridium strains – saccharolytic butyric acid-pro-ducing bacteria – under appropriate operating conditions(Clostridium saccharoperbutylacetonicum, C. acetobutylicum, C. bei-jerinckii, C. aurantibutyricum). The strains are able to metabolize a

charides (Flickinger and Drew, 1999). Under batch conditions, thefermentation process of solvent-producing clostridia proceeds withthe production of cells, hydrogen, carbon dioxide, acetic acid andbutyric acid during the initial growth phase (acidogenesis) (Jonesand Woods, 1986). As acid concentration increases (pH decreases),the cell metabolism shifts to solvent production (solventogenesis)and acidogenic cells – able to reproduce themselves – shift to thesolventogenesis state with a morphological change (Jones andWoods, 1986). Under solventogenesis, the active cells become endo-spores unable to reproduce themselves. Accordingly, two differentphysiological states must be taken into account for clostridia: onefor the acidogenic phase, and one for the solventogenic phase.

Despite the aforesaid remarkable advantages of butanol as a bio-product, its industrial production via fermentation is not developedfor several issues: feedstock cost and availability, low yield and pro-ductivity, low concentration of butanol in the broth (as a conse-quence of the product-inhibition feature of ABE fermentation),degeneration of butanol-producing strains (Kumar and Gayen,2011). Moreover, the low butanol concentration in fermentationbroth makes its recovery and concentration quite complex (Liuand Fan, 2004; Ezeji et al., 2007b, 2013; Napoli et al., 2012a).

Fig. 1. Outline of the apparatus adopted for continuous test equipped with a packedbed biofilm reactor. F – gas sterilization filter.

260 F. Raganati et al. / Bioresource Technology 138 (2013) 259–265

Reactor performance enhancements were proved by continuousABE production in reactors operating with clostridium cell-confine-ment options: cell-immobilisation (Ezeji et al., 2007b; Lee et al.,2008; Qureshi et al., 2000; Napoli et al., 2010; Lu et al., 2012) orcell-recycling (Meyer and Papoutsakis, 1989; Tashiro et al., 2005;Zheng et al., 2013). As regards cell-immobilized reactors, their highcell density improves butanol yield and butanol recovery. Contin-uous bioconversion in immobilized cell reactors has several advan-tages over batch cultures, which are typically adopted for butanolproduction via fermentation (Qureshi et al., 2000). The mainadvantages are related to the high cell concentrations and to thereactor operability at high dilution rates without cell washout(Welsh et al., 1987). Moreover, the cell support may often be re-used (Krouwel et al., 1980).

As regards the feedstock, abundance and un-competitivenesswith food sources are prerequisites of potential substrates. Gener-ally, lignocellulosic biomass and wastewater streams have theserequisites. The former may be cultivated ad hoc or obtained asagro-industrial by-products. The latter include cheese-whey andwastewaters. The potential of cheese-whey as feedstock for buta-nol production has been pointed out by several authors (Qureshiand Maddox, 1987; Gonzalez-Siso, 1996; Foda and Dong, 2010).The tests have always focused on batch fermentations (Foda andDong, 2010) and continuous bioreactors fed with whey permeatesupplemented with yeast extract (Qureshi and Maddox, 1987).

The aim of this contribution is to investigate the feasibility ofbio-butanol production by continuous conversion of cheese whey,a dairy industry wastewater characterized by high lactose and pro-tein content. The anaerobic solventogenic bacterium C. acetobutyl-icum DSM 792 was adopted for the fermentation processes. Theconversion was carried out in a packed bed reactor (PBR) of Tygonrings as biofilm carriers. The process was characterized in terms ofbutanol production rate, butanol selectivity and butanol yield as afunction of the dilution rate.

2. Methods

2.1. Microorganism

C. acetobutylicum DSM 792 was supplied by DSMZ. Stock cul-tures were reactivated according to the DSMZ procedure. Reacti-vated cultures were stored at �80 �C. The thawed cells wereinoculated in 15 mL Hungate tubes containing 12 mL of syntheticmedium: 20 g/L lactose, 5 g/L yeast extract (YE). Cells were grownunder anaerobic conditions for 48 h at 37 �C. Then pre-cultureswere transferred into fermentation bottles.

2.2. Medium

2.2.1. Synthetic mediumThe synthetic medium consisted of 20 g/L lactose and 5 g/L YE

supplemented with P2 stock solution: buffer – 0.5 g/L KH2PO4,0.5 g/L K2HPO4; 2 g/L ammonium chloride; mineral – 0.2 g/L

Table 1Cheese whey pretreatment methods. Main results.

Pretreatment method Max butano

Heat sterilization inoculum 6.4Control 0

Tyndalization inoculum 6.3Control 0

Dry tyndalization inoculum 8.1Control 0

Deproteinization inoculum 8.9Control 0

MgSO4�7H2O, 0.01 g/L MnSO4�H2O, 0.01 g/L FeSO4�7H2O (Qureshiand Blaschek, 2000). The lactose–YE solution was sterilized inautoclave (20 min at 120 �C).

2.2.2. Cheese whey powderThe cheese whey powder (CWP) was by Sierolat, an Italian com-

pany. The composition of the powder was: lactose 69 wt%, proteins12.5 wt%, ashes 7.5 wt%, lactic acid 4 wt%, moisture 2.5 wt%, fats2.5 wt%, galactose 2 wt%.

The CWP was solubilized in deionized water (concentration40 g/L) and pretreated according to the methods reported in Table 1.

The continuous biofilm reactor was fed with a pretreated solu-tion of CWP characterized by 28 g/L lactose concentration.

2.3. Apparatus and operating conditions

All the experiments were carried out at 37 �C.

2.3.1. Batch fermentationPyrex screw capped bottles (100 mL) containing 75 mL medium

were used as fermenters. Tests were carried out in fermenters atrest without pH control. The medium was inoculated with 6.25%(v/v) suspension of active growing pre-cultures. 3 mL of cultureswere periodically sampled to measure the concentration of lactoseand metabolites.

2.3.2. Continuous bioreactorThe apparatus adopted for the cheese whey fermentation con-

sisted of a fixed bed reactor, liquid pumps, a heating apparatus, adevice for pH control, and on-line diagnostics (Fig. 1). The fixedbed was at the bottom of a 200 mL glass lined pipe (4 cm ID,16 cm high) jacketed for the heat exchange. The liquid phase vol-ume in the reactor, or reaction volume, was modified by changingthe level of the overflow duct. Nitrogen was sparged at the reactorbottom to support anaerobic conditions. The device for pH controlconsisted of a pH-meter, a peristaltic pump, a vessel with NaOH0.3 M solution, and a controller. The reactor with the carriers wassterilized in autoclave at 121 �C for 20 min. The gas stream was

l [g/L] Residual acids [g/L] Clot presence

2.7 Yes02.7 Yes0.12.7 No6.82.3 No0

F. Raganati et al. / Bioresource Technology 138 (2013) 259–265 261

sterilized by filtration. The sterile medium was fed at the bottom ofthe reactor by means of a peristaltic pump.

Tygon rings (3/1 mm OD/ID) were selected as biofilm carriersaccording to previous investigation results (Napoli et al., 2010).

2.4. Analytical procedures

pH of batch cultures was measured off-line in 1.5 mL samplesby a pH-meter (Hanna Instruments).

Analysis of culture samples was carried out after centrifugationat 11,000g for 10 min. The liquid phase was characterized in termsof lactose and metabolite concentrations, total organic carbon(TOC), and total nitrogen (TN). Lactose concentration was deter-mined by high performance liquid chromatography (HPLC) usingan Agilent 1100 system (Palo Alto, CA). Lactose was separated atroom temperature by means of a 8 lm Hi-Plex H, 30 cm � 7.7 mmcolumn and detected with a refractive index detector. Deionizedwater was used as mobile phase at 0.6 mL/min flow rate. A GC appa-ratus Agilent 6890 series (Palo Alto, CA) equipped with a capillarycolumn poraplot Q (25 m � 0.32 mm) and a FID was used for themetabolite analysis. Hexanoic acid was adopted as internal stan-dard to analyze acids and alcohols and to assess their concentra-tions. A Shimadzu TOC 5000A analyzer was used to measureTOC/TN concentration.

2.5. Experimental procedures

300 lL of stock culture were transferred into four 15-mL Hungatetubes containing the culture media (20 g/L of lactose). The cultureswere incubated for 1 day under batchwise anaerobic sterile condi-tions, then 40 mL of active culture were inoculated in the reactor.

The biofilm in the fixed bed reactor was grown adopting a lac-tose-based medium at the beginning of the test at a pH set at a va-lue that was compatible with the acidogenic phase. The feedingwas switched to CWP-based medium and after a pre-set time thepH was definitively decreased for solvent production.

Tests aimed at butanol production were carried out with thebiofilm PBR operated at some specific conditions. The dilution rate(D) – the ratio between the feeding volumetric flow rate and thevolume of the fixed bed – ranged between 0.54 and 0.94 1/h. Eachsteady state was characterized in terms of metabolites and lactoseconcentration. The biomass inside the reactor was measured at theend of the test program by sacrificing the reactor (Qureshi et al.,1988). The mass of biofilm in the reactor was assessed at the endof the run in agreement with the following procedure: (i) the drycarrier was weighted before filling the reactor; (ii) at the end ofthe test, the reactor was rinsed with sterile water to remove lactoseand metabolites; (iii) the carriers with the biomass wereharvested and dried at 40 �C for 1 day and (iv) the dried mass ofthe biomass and carriers were weighted. The dried mass of thebiofilm in the reactor was assessed as the difference between theweight of the carriers-biofilm and the carriers.

2.6. Data analysis

The biofilm reactor performances were assessed by measuringthe concentration of lactose and metabolites provided that thesteady-state had stabilized for at least 10 times the reactors meanresidence time (1/D). Reactor performances were reported in termsof lactose conversion degree (nL), lactose-to-‘‘i-species’’ fractionalyield coefficient (Yi/L), butanol productivity (WB), ABE productivity(WABE), butanol selectivity (U), concentration of acidogenic cellswithin the biofilm (XAcidogenic).

nL, Yi/L, WB, WABE and U were assessed assuming that: (i) thefeeding was aseptic and free of metabolites, and (ii) the gas strip-ping of metabolites was negligible. They were:

nL ¼ðLIN � LOUTÞ

LOUT ð1Þ

Yi=L ¼iOUT

LIN � LOUT ð2Þ

WB ¼ D � BOUT ð3Þ

WABE ¼ D � ABEOUT ð4Þ

U ¼ D � BOUT

D � ðAOUT þ BOUT þ EOUTÞð5Þ

where L, A, B and E are the concentrations of lactose, acetone, buta-nol and ethanol, respectively, measured in the feeding (suffix IN)and in the effluent (suffix OUT).

The concentration of acidogenic cells within the biofilm(XAcidogenic) was estimated according to the theoretical frameworkreported hereinafter. Major issues for the estimation were: (a) thepH gradient in the biofilm; (b) the coexistence of acidogenic cellsand solventogenic cells in the biofilm and (c) the reactions for pro-duction/conversion of acids.

The pH was assumed to decrease in the biofilm from the bulk va-lue at the liquid interface. In particular, the pH gradient was the re-sult of the equilibrium between transport phenomena andmicrobiological conversions (Olivieri et al., 2011; Rai et al., 2012).According to the pH profile within the biofilm, two stratified regionsmay be found in the biofilm: (i) an external layer characterized bylocal value of pH > 4.5 and presence of acidogenic cells; (ii) internallayer characterized by local value of pH < 4.5 and presence of sol-ventogenic cells. Acids were produced by cells present in the exter-nal layer and were reconverted by cells present in the inner layer. Asa result, the acids net production depended on the relative depth ofthe two layers.

The total acids production rate by acidogenic cells (WTOTAcids) in

the biofilm may be estimated taking into account the yields of bu-tyric and acetic acids (BA and AA) with respect to the lactose (YBA/L

and YAA/L). YX/L, YBA/L and YAA/L yields were assessed as a function ofthe specific growth rate according to the relationships reported byNapoli et al. (2012b) and the biomass-to-acids fractional yieldcoefficient (YAcids/X) was:

YAcids=X ¼YAA=L þ YBA=L

YX=Lð6Þ

The total acids production rate by acidogenic cells was:

WTOTAcids ¼ D � XAcidogenic � YAcids=X ð7Þ

A fraction of the acids produced was converted in solvents accord-ing to the reactions:

C12H22O11 �H2Oþ 2CH3COOH !þ4ATP2CH3COCH3 þ 2C2H5OH

þ 6CO2 þ 4H2 ð8Þ

C12H22O11 �H2Oþ 2C3H7COOH !þ4ATP2CH3COCH3 þ 2C4H9OH

þ 6CO2 þ 4H2 ð9Þ

According to the stoichiometry of reactions (8) and (9), the acetoneproduced is equal to the sum of the converted acetic and butyricacids:

½A� ¼ ½AA�UP þ ½BA�UP ð10Þ

where [A] is the acetone concentration, and [j]UP the concentrationof the converted acid ‘‘j’’. Moreover, converted acids are assumed tofulfill the balance proposed by Desai et al. (1998):

Fig. 2. Main data measured during PBR start-up. The vertical dotted line marks the instant at which the pH set-up of the controller was changed from 5.5 to 4.5.

262 F. Raganati et al. / Bioresource Technology 138 (2013) 259–265

½BA�UP

½AA�UP ¼ 0:315½BA�½AA� ð11Þ

where [j] is the concentration of acid ‘‘j’’ in the reactor. Given thedilution rate at which the reactor is operated, the rate of the con-verted acids (WUP

Acids) may be assessed by working out Eqs. (10)and (11). The sum of the estimated rate of converted acids and ofthe rate of produced acids (WNet

Acids) measured is:

WTOTAcids ¼ ðW

NetAcids þWUP

AcidsÞ ð12Þ

The concentration of acidogenic cells within the biofilm was esti-mated by working out Eqs. (7) and (12):

XAcidogenic ¼WTOT

Acids

D � YAcids=Xð13Þ

3. Results and discussion

3.1. CWP pretreatment

Preliminary tests were carried out with solutions of raw CWP(40 g/L) without C. acetobutylicum inoculum (data not reported).Lactose was barely consumed and the conversion after 7 dayswas smaller than 2%. Although lactose uptake rate was inapprecia-ble, the results were tell-tale of the presence of endogenous micro-organisms and a light sterilization was planned.

The study regarding pretreatment of CWP (40 g/L) solutionswas aimed at selecting the optimal conditions in terms of feedstockoperability and butanol production. Denaturation of proteins withformation of aggregate/clots during thermal treatment of CWPshould be avoided. Indeed, large clots may clog both pipes andthe packed bed reactor during the continuous fermentation. Theinvestigated pretreatment methods were:

� Heat sterilization. CWP solution was sterilized in autoclave at120 �C for 20 min;� Tyndalization. CWP solution was sterilized according to the fol-

lowing thermal program: 30 min at 80 �C; 24 h at 37 �C; 30 minat 80 �C; 24 h at 37 �C; 30 min 80 �C;� Dry tyndalization. CWP was sterilized according to the following

thermal program: 60 min at 80 �C; 24 h at 37 �C; 60 min at80 �C; 24 h at 37 �C; 60 min at 80 �C. The sterilized CWP wassolubilized in sterilized deionized water.� Deproteinization. The CWP solution was deproteinized by heat

treatment: 30 min at 90 �C. The precipitate was removed bycentrifugation at 5500g, 30 min at 10 �C. The supernatant wassterilized in autoclave (120 �C, 20 min) so as to be used as fer-mentation medium.

Table 1 reports the results of the analysis of the pretreatmentmethods. The performance of each method was characterized interms of clot formation and sterilization efficiency. The efficiency

Fig. 3. Data measured during the PBR operation as a function of the dilution rate.Feedstock: solution of CWP pretreated. Vertical lines: see caption of Fig. 2.

F. Raganati et al. / Bioresource Technology 138 (2013) 259–265 263

was expressed as the maximum concentration of butanol and resid-ual concentration of acids measured in fermentation tests, as com-pared with those measured in tests carried out withoutinoculum(control).

Table 2Steady-state cultures of C. acetobutylicum (up-phase).

Operating conditions

D [h�1] 0.54pHLactose in the feeding [g/L]

ResultsLactose [g/L] 9.1Lactose conversion degree, nL [%] 68Acetone [g/L] 0.58Butanol [g/L] 4.93Ethanol [g/L] 0.50Acetic acid [g/L] 1.23Butyric acid [g/L] 1.32Butanol productivity, WB [g/Lh] 2.66ABE productivity, WABE [g/Lh] 3.24Butanol yield, YB/L [g/g] 0.26ABE yield, YABE/L [g/g] 0.32Acids yield, YAcids/L [g/g] 0.14Butanol selectivity, U [g/g] 0.82

Table 1 shows that heat sterilization and tyndalization gavegood results in terms of sterilization efficiency. However, bothmethods were discarded because a significant amount of clotsformed. Both dry tyndalization and deproteinization were charac-terized by absence of clot formation and high butanol concentra-tion. However, dry tyndalization was discarded because it wasnot successful in terms of sterilization. Based on the results re-ported in Table 1, the deproteinization process was chosen as thebest method to pretreat the CWP solution.

3.1. Biofilm reactor start-up

Tygon rings – 39.2 g – were adopted to prepare a 8 cm highpacked bed. The volume of the reactor was set at 100 mL by meansof the overflow duct. The reactor was inoculated with actively grow-ing culture. No chemical was adopted to assist cell immobilizationon the selected carriers. The start-up process was carried out withthe synthetic medium with 25 g/L lactose.

Fig. 2 reports the time series of metabolite concentration and pH.The reactor was inoculated at t = 0, operated under batch conditionswith respect to the liquid phase for 24 h (data not reported) andthen under continuous conditions. The dilution rate was set at0.40 h�1 and the pH was gradually increased from 5.0 to 5.5 to oper-ate the fermentation under acidogenesis conditions (Napoli et al.,2011). A visible biofilm layer on carriers formed in about 1 weekand at t = 7 days the dilution rate was increased (D = 0.8 h�1) to pro-mote biofilm production over suspended cell growth. It is interest-ing to note that solvent production started even though the bulk pHwas still 5.5. According to the biofilm schematization reported inSection 2.6, the pH in the biofilm decreased moving from theliquid-biofilm toward the inner region (Qureshi et al., 2005; Napoliet al., 2010) promoting the solventogenic shift of inner cells.

At t = 18.5 days, the carriers were covered by abundant biofilmand a steady state condition was approached. The total acid (ace-tic+butyric) yield was about 0.35 gacid/glactose. The small differencebetween the total yield measured and the value assessed by Napoliet al. (2012b) may be due to the partial solvent production. The feed-ing stream was switched from lactose-synthetic medium to CWPsolution (28 g/L lactose) to accustom cells to the complex medium.A new steady-state regime was approached in about three days.

Altogether, the biofilm reactor start-up took about 21 days anda remarkable amount of active biofilm was formed.

3.2. Butanol production

The bioreactor operating conditions were changed att = 21.7 days (Fig. 3) to produce butanol: pH = 4.7, D = 0.54 h�1.

0.64 0.74 0.84 0.944.728

14.4 15.2 17.0 20.649 46 39 260.07 0.07 0.15 0.083.23 3.24 2.93 1.460.25 0.47 0.21 0.241.65 2.10 2.73 3.571.89 1.97 2.19 2.652.06 2.39 2.46 1.372.27 2.80 2.76 1.670.24 0.25 0.26 0.200.26 0.29 0.30 0.220.26 0.32 0.45 0.840.91 0.86 0.89 0.92

Table 3Steady-state cultures of C. acetobutylicum (down-phase).

Operating conditions

D [h�1] 0.54 0.64 0.74 0.84pH 4.7Lactose in the feeding [g/L] 28

ResultsLactose [g/L] 9.2 13.3 16.3 17.4Lactose conversion degree, nL [%] 67 53 42 38Acetone [g/L] 0.4 0.2 0.13 0.1Butanol [g/L] 3.06 2.3 2.13 1.79Ethanol [g/L] 1.0 0.84 0.62 0.31Acetic acid [g/L] 0.74 0.99 1.24 1.42Butyric acid [g/L] 0.82 1.06 1.95 2.25Butanol productivity, WB [g/Lh] 1.65 1.47 1.58 1.5ABE productivity, WABE [g/Lh] 2.41 2.14 2.14 1.85Butanol yield, YB/L [g/g] 0.16 0.16 0.18 0.17ABE yield, YABE/L [g/g] 0.24 0.23 0.25 0.20Acids yield, YAcids/L [g/g] 0.08 0.14 0.27 0.35Butanol selectivity, U [g/g] 0.67 0.70 0.72 0.85

Fig. 4. Estimated biomass concentration under acidogenic phase – Eq. (13) – vs. thedilution rate.

264 F. Raganati et al. / Bioresource Technology 138 (2013) 259–265

The pH value was set according to previous investigations (Napoliet al., 2010). It was higher than the value at which cells shift to sol-vent production, pH = 4 (Napoli et al., 2009), because pH andmetabolites gradient (see Section 2.6) was expected across the bio-film (Qureshi et al., 2005; Napoli et al., 2010).

The reactor approached a steady-state regime within 7 days. Asexpected, solvents and acids were continuously produced, con-firming the co-existence of both C. acetobutylicum cells voted toproduce ABE and cells committed to produce acids and biomass(Mollah and Stuckey, 1993; Napoli et al., 2009). PBR performanceswere characterized in terms of metabolite concentration, lactoseconversion degree, yield of acids, butanol and ABE, productivityof butanol and of ABE, and butanol to solvents selectivity (Table 2).The performances were assessed by working out the concentrationof lactose and metabolites according the relationships reported inthe Section 2.6, provided that the steady-state was stabilized forat least 10 times the reactors mean residence time (1/D = 1.9 h).The lactose concentration was 9 g/L, about half the affinity con-stant (28 g/L) assessed by Napoli et al. (2009).

Fig. 3 reports the main data measured during the continuous fer-mentation. The dilution rate was steadily increased from 0.54 to0.94 h�1 (up phase). Tables 2 and 3 report the PBR performancesas a function of the dilution rate. The analysis of the results reportedin Fig. 3 and Table 2 highlighted the issues reported hereinafter:

� Butanol and solvent concentration, and lactose conversiondegree (nL) decreased with D. They were characterized by amaximum at D = 0.54 h�1;� Except for the highest value of D investigated, butanol and ABE

productivity as well as butanol and ABE yields were approxi-mately constant with D. At D = 0.94 h�1 an abrupt reduction ofsolvent expression was assessed;� Butanol selectivity was quite high (0.92 g/g) and about constant

with D. It is worthy to note that the selectivity was among thelargest values reported in the literature;� The concentration of acids and their yield and productivity

increased progressively with D.

Results reported in Table 3 refer to steady-state cultures carriedout at D decreasing between 0.94 h�1 and 0.54 h�1 (down-phase).The analysis of the data in the table confirms the effects of D high-lighted with reference to the up-phase investigation. However, PBRperformances assessed during the down-phase were not so high asthose assessed during the up-phase.

The biofilm PBR was stopped at the end of the down-phase andthe overall biomass concentration was 101.7 gDM/L, that is a bio-mass to carrier ratio of 0.1 gDM/g.

It is evident that the PBR performances decreased with the dilu-tion rate under operating conditions investigated. The perfor-mances were almost constant for D varying between 0.64 and0.84 h�1, the PBR experienced its worst performances when Dwas set to 0.94 h�1. Moreover, the acid productivity/concentrationincreased progressively with D and were very high at D = 0.94 h�1.From the butanol/solvent production point of view the best perfor-mances were assessed at the lowest D investigated. These resultsare in agreement with those reported by Qureshi and Maddox(1987) even though performances assessed in the present investi-gation were better when PBRs are operated under close operatingconditions (absence of yeast extract, 28 g/L lactose in the feeding).

The decrease in performances at high dilution rate deservessome comments.

From the kinetic point of view, the production rate of butanol/solvents should be favoured by high dilution rate. Indeed, the highlactose concentration coupled with the high feeding flow rateshould increase the production rate. From the biofilm structurepoint of view, the relationship between the PBR performances

F. Raganati et al. / Bioresource Technology 138 (2013) 259–265 265

and D depends on the distribution of the active cells t in the bio-film. High feeding flow rate is typically coupled with low lactoseconversion, as reported in the present investigation and by Qureshiand Maddox (1987). Therefore, pH profile within the biofilm (seeSection 2.6) may be expected to be less steep and the layerthickness of acidogenic cells within the biofilm (XAcidogenic) shouldincrease: the boundary of the acidogenic layer should movetoward the carrier surface. Indeed, Fig. 4 supports this result: theXAcidogenic – assessed according to Eq. (13) – increases with D.Moreover, the gradual recovery of solvent productivity as Ddecreased points out that biomass was still present in the PBRand ready to produce solvents when operating conditions werecompatible with the conversion. It is worthy to note that the as-sessed acidogenic cells ranged between 1 and 4 wt% of the biomasspresent in the PBR. The biofilm complement consisted of clostrid-ial-form cells, extracellular products, and spores.

The steady state at D = 0.94 h�1 was particularly interesting.Two issues may be listed as responsible for this behavior.

The feeding flow rate – dilution rate 0.94 h�1 – was apparentlytoo high and the stream might strip cells off the biofilm surface.Therefore, low butanol production rate and slightly high biomassin the effluent were observed at the outset of continuous operationat 0.94 h�1 dilution rate. However, high dilution rate also led to ahigh undigested level of lactose, as observed in the PBR effluent(Fig. 3).

4. Conclusions

Butanol production by C. acetobutylicum fermentation of cheesewhey in a packed bed reactor (PBR) was carried out for more than3 months. Un-supplemented deproteinized cheese whey wasadopted as feedstock.

The effects of the dilution rate (D) on the PBR performanceswere investigated. Solvents concentration, productivity rate andlactose conversion decreased with D. Under optimized conditions(D = 0.54 1/h) performances were: butanol productivity 2.66 g/Lh,butanol concentration 4.93 g/L, butanol yield 0.26 g/g, butanolselectivity of the overall solvents production 82 wt%. The perfor-mances decrease with D may be interpreted taking into accountthe biofilm fraction of acidogenic cells.

Acknowledgements

The authors are indebted with Sierolat S.p.A. (Francolisi, IT) forsupplying cheese whey powder. The authors thank Mrs. RobertaCarpine, Mrs. Bernadette Lombardi and Mr. Alessandro Manzi forperforming tests.

References

Bohlmann, G., 2007. Biobutanol. SRI Consulting, California.Cascone, R., 2008. Biobutanol – A Replacement for Bioethanol? Chem. Eng. Prog.,

S4–S9.Desai, R., Nielsen, L.K., Papoutsakis, E.T., 1998. Stoichiometric modeling of

Clostridium acetobutylicum fermentations with non-linear constraints. J.Biotechnol. 71, 191–205.

Dürre, P., 2007. Biobutanol: an attractive biofuel. Biotechnol. J. 2, 1525–1534.Ezeji, T., Qureshi, N., Blaschek, H.P., 2007a. Butanol production from agricultural

residues: impact of degradation products on Clostridium beijerinckii growth andbutanol fermentation. Biotechnol. Bioeng. 97, 1460–1469.

Ezeji, T.C., Qureschi, N., Blaschek, H.P., 2007b. Production of acetone butanol (AB)from liquefied corn starch, a commercial substrate, using Clostridium beijerinckiicoupled with product recovery by gas stripping. J. Ind. Microbiol. Biotechnol. 34,771–777.

Ezeji, T.C., Qureschi, N., Blaschek, H.P., 2013. Microbial production of a biofuel(acetone–butanol–ethanol) in a continuous bioreactor: impact of bleed andsimultaneous product removal. Bioprocess Biosyst. Eng. 36, 109–116.

Flickinger, M., Drew, S.W.(Eds.), 1999. Encyclopedia of Bioprocess Technology, vols.1–5. Wiley-Blackwell, John Wiley & Sons Inc.,

Foda, M.I., Dong, H., Li, Y., 2010. Study of the suitability of cheese whey forbiobutanol production by Clostrodia. J. Americ. Sci. 6, 39–46.

Friedl, A., 2012. Lignocellulosic biorefinery. Env. Eng. and Management J. 11, 75–79.Gonzalez-Siso, M.I., 1996. The biotechnological utilization of cheese whey: a review.

Bioresour. Technol. 57, 1–11.Jones, D., Woods, D., 1986. Acetone–butanol fermentation revisited. Microbiol. Rev.

50, 484–524.Krouwel, P.G., Van der Laan, W.F.M., Kossen, N.W.F., 1980. Continuous production of

n-butanol and isopropanol by immobilized, growing Clostridium butylicum cells.Biotechnol. Lett. 2, 253–258.

Kumar, M., Gayen, K., 2011. Developments in biobutanol production: new insights.Appl. Energy 88, 1999–2012.

Lee, S., Cho, M., Park, C., 2008. Continuous butanol production using suspended andimmobilized Clostridium beijerinckii NCIMB 8052 with supplementary butyrate.Energy Fuels 22, 3459–3464.

Liu, J., Fan, L.T., 2004. Downstream process synthesis for biochemical production ofbutanol, ethanol, and acetone from grains: generation of optimal and near-optimal flowsheets with conventional operating units. Biotechnol. Prog. 20,1518–1527.

Lu, C., Zhao, J., Yang, S.T., Wei, D., 2012. Fed-batch fermentation for n-butanolproduction from cassava bagasse hydrolysate in a fibrous bed bioreactor withcontinuous gas stripping. Bioresour. Technol. 104, 380–387.

Meyer, C.L., Papoutsakis, E.T., 1989. Continuous and biomass recycle fermentationsof Clostridium acetobutylicum. Part 1: ATP supply and demand determinesproduct selectivity. Bioprocess Eng. 4, 1–10.

Mollah, A.H., Stuckey, D.C., 1993. Maximizing the production of acetone–butanol inan alginate bead fluidized bed reactor using Clostridium acetobutylicum. J. Chem.Technol. Biotechnol. 56, 83.

Napoli, F., Olivieri, G., Marzocchella, A., Salatino, P., 2009. Bioenergy II: anassessment of the kinetics of butanol production by Clostridiumacetobutylicum. Int. J. Chem. React. Eng. 7, A45.

Napoli, F., Olivieri, G., Russo, M.E., Marzocchella, A., Salatino, P., 2010. Butanolproduction by Clostridium acetobutylicum in a continuous packed bed reactor. J.Ind. Microbiol. Biotechnol. 37, 603–608.

Napoli, F., Olivieri, G., Russo, M.E., Marzocchella, A., Salatino, P., 2011. Continuouslactose fermentation by Clostridium acetobutylicum – assessment ofacidogenesis kinetics. Bioresour. Technol. 102, 1608–1614.

Napoli, F., Olivieri, G., Russo, M.E., Marzocchella, A., Salatino, P., 2012a. Optimizationof solvent recovery in the production of butanol by fermentation. Environ. Eng.Manag. J. 11, 1499–1504.

Napoli, F., Olivieri, G., Russo, M.E., Marzocchella, A., Salatino, P., 2012b. Continuouslactose fermentation by Clostridium acetobutylicum – assessment of energeticsand product yields of the acidogenesis. Enzyme Microb. Technol. 50, 165–172.

Olivieri, G., Russo, M.E., Marzocchella, A., Salatino, P., 2011. Modeling of an aerobicbiofilm reactor with double-limiting substrate kinetics: bifurcational anddynamical analysis. Biotechnol. Prog. 27, 1599–1613.

Qureshi, N., Maddox, I.S., 1987. Continuous solvent production from whey permeateusing cells of Clostridium acetobutylicum immobilized by adsorption ontobonechar. Enzyme Microb. Technol. 9, 668–671.

Qureshi, N., Paterson, A.H.J., Maddox, I.S., 1988. Model for continuous production ofsolvents from whey permeate in a packed bed reactor using cells of Clostridiumacetobutylicum immobilized by adsorption onto bonechar. Appl. Microbiol.Biotechnol. 29, 323–328.

Qureshi, N., Blaschek, H.P., 2000. Economics of butanol fermentation using hyper-butanol producing Clostridium beijerinckii BA101. Trans. IChem. E 78(Part C),139–144.

Qureshi, N., Schripsema, J., Lienhardt, J., Blaschek, H.P., 2000. Continuous solventproduction by Clostridium beijerinckii BA101 immobilized by adsorption ontobrick. World J. Microbiol. Biotechnol. 16, 337–382.

Qureshi, N., Annous, B.A., Ezeji, T.C., Karcher, P., Maddox, I.S., 2005. Biofilm reactorsfor industrial bioconversion processes: employing potential of enhancedreaction rates. Microb. Cell Factories 4, 24.

Rai, V., Upadhyay, R.K., Thakur, N.K., 2012. Complex population dynamics inheterogeneous environments: effects of random and directed animalmovements. Int. J. Nonlinear Sci. Numer. Simul. 13, 299–309.

Tashiro, Y., Takeda, K., Kobayashi, G., Sonomoto, K., 2005. High production ofacetone–butanol–ethanol with high cell density culture by cell-recycling andbleeding. J. Biotechnol. 4 (120), 197–206.

Welsh, F.W., Williams, R.E., Veliky, I.A., 1987. Solid carriers for a Clostridiumacetobutylicum that produces acetone and butanol. Enzyme Microb. Technol. 9,500–502.

Zheng, J., Tashiro, Y., Yoshida, T., Gao, M., Wang, Q., Sonomoto, K., 2013. Continuousbutanol fermentation from xylose with high cell density by cell recyclingsystem. Bioresour. Technol. 129, 360–365.