hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

9
Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey V. Perna a,b , E. Castello ´ b, *, J. Wenzel a , C. Zampol c , D.M. Fontes Lima c , L. Borzacconi b , M.B. Varesche c , M. Zaiat c , C. Etchebehere a a Department of Microbiology, Faculty of Science, Faculty of Chemistry, General Flores 2124, Montevideo, Uruguay b Chemical Engineering Institute, Faculty of Engineering, University of the Republic, Herrera y Reisig 565, CP.11300, Montevideo, Uruguay c Laborato ´rio de Processos Biolo ´gicos, Centro de Pesquisa, Desenvolvimento e Inovac ¸a ˜o em Engenharia Ambiental, Escola de Engenharia de Sa ˜o Carlos, Universidade de Sa ˜o Paulo, Av. Joa ˜o Dagnone, 1100 e Santa Angelina, 13.563-120 Sa ˜o Carlos, SP, Brazil article info Article history: Received 5 June 2012 Received in revised form 14 August 2012 Accepted 5 October 2012 Available online 21 November 2012 Keywords: Hydrogen Clostridium T-RFLP Microbial community structure Packed bed reactor Real time PCR abstract In this work, an upflow anaerobic packed bed reactor configuration to produce hydrogen using cheese whey as the substrate was tested. The microbiological composition was linked to the reactor operation. Three different organic loading rates of 22, 33, and 37 g COD/L-d were used at a fixed hydraulic retention time of 24 h. The increase of the organic loading rate from the initial value of 22 g COD/L-d to a value of 37 g COD/L-d and the adjustment of the pH to values greater than 5 had a positive effect on the hydrogen production and values of up to 1 L H 2 /L- d were achieved. The production of hydrogen was stable under all of the investigated conditions, and the problems frequently reported in this kind of reactors (bed clogging, methanogenesis and solvent production) were not observed during operation. The highest hydrogen yield was 1.1 mol H 2 /mol lactose far from the maximum theo- retical value (8 mol H 2 /mol lactose), but similar to previous works using cheese whey as substrate. The microbiological analysis showed a mixed population with a low proportion of hydrogen-producing fermenters (Clostridium and Klebsiella) and other non-hydrogen- producing organisms. Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. 1. Introduction Packed bed reactors are a good alternative for the production of hydrogen from wastewaters because of their simple construction and increased cell retention time, which is essential for hydrogen production [1e7]. The support matrices can be made of recycled materials, and neither mechanical agitation nor a recirculation apparatus is required in this configuration, which results in low costs for reactor construction and operation. Moreover, the packing material can help in the selection of microbial populations [8]. Cheese whey is an important byproduct generated during cheese production. It is composed mainly of sugars (lactose, 70e72% dried extract), proteins (8e10%), and mineral salts (12e15% dried extract) [9]. Because of the low buffer capacity of cheese whey, its treatment in a conventional anaerobic reactor frequently leads to acidification and inhibition of methanogenic activity [10]. Therefore, its treatment in * Corresponding author. Facultad de Ingenierı´a, Universidad de la Repu ´ blica, Herrera y Reisig 565, CP.11300, Montevideo, Uruguay. Tel.: þ598 2 711 08 71/44 78; fax: þ598 2 710 74 37. E-mail addresses: elenacas@fing.edu.uy, [email protected] (E. Castello ´ ). Available online at www.sciencedirect.com journal homepage: www.elsevier.com/locate/he international journal of hydrogen energy 38 (2013) 54 e62 0360-3199/$ e see front matter Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijhydene.2012.10.022

Upload: v-perna

Post on 27-Nov-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

ww.sciencedirect.com

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 2

Available online at w

journal homepage: www.elsevier .com/locate/he

Hydrogen production in an upflow anaerobic packed bedreactor used to treat cheese whey

V. Perna a,b, E. Castello b,*, J. Wenzel a, C. Zampol c, D.M. Fontes Lima c, L. Borzacconi b,M.B. Varesche c, M. Zaiat c, C. Etchebehere a

aDepartment of Microbiology, Faculty of Science, Faculty of Chemistry, General Flores 2124, Montevideo, UruguaybChemical Engineering Institute, Faculty of Engineering, University of the Republic, Herrera y Reisig 565, CP.11300, Montevideo, Uruguayc Laboratorio de Processos Biologicos, Centro de Pesquisa, Desenvolvimento e Inovacao em Engenharia Ambiental, Escola de Engenharia de

Sao Carlos, Universidade de Sao Paulo, Av. Joao Dagnone, 1100 e Santa Angelina, 13.563-120 Sao Carlos, SP, Brazil

a r t i c l e i n f o

Article history:

Received 5 June 2012

Received in revised form

14 August 2012

Accepted 5 October 2012

Available online 21 November 2012

Keywords:

Hydrogen

Clostridium

T-RFLP

Microbial community structure

Packed bed reactor

Real time PCR

* Corresponding author. Facultad de Ingenierþ598 2 711 08 71/44 78; fax: þ598 2 710 74 37

E-mail addresses: [email protected],0360-3199/$ e see front matter Copyright ªhttp://dx.doi.org/10.1016/j.ijhydene.2012.10.0

a b s t r a c t

In this work, an upflow anaerobic packed bed reactor configuration to produce hydrogen

using cheese whey as the substrate was tested. The microbiological composition was

linked to the reactor operation.

Three different organic loading rates of 22, 33, and 37 g COD/L-d were used at a fixed

hydraulic retention time of 24 h. The increase of the organic loading rate from the initial

value of 22 g COD/L-d to a value of 37 g COD/L-d and the adjustment of the pH to values

greater than 5 had a positive effect on the hydrogen production and values of up to 1 LH2/L-

d were achieved. The production of hydrogen was stable under all of the investigated

conditions, and the problems frequently reported in this kind of reactors (bed clogging,

methanogenesis and solvent production) were not observed during operation.

The highest hydrogen yield was 1.1 mol H2/mol lactose far from the maximum theo-

retical value (8 mol H2/mol lactose), but similar to previous works using cheese whey as

substrate. The microbiological analysis showed a mixed population with a low proportion

of hydrogen-producing fermenters (Clostridium and Klebsiella) and other non-hydrogen-

producing organisms.

Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights

reserved.

1. Introduction construction and operation. Moreover, the packing material

Packed bed reactors are a good alternative for the production

of hydrogen from wastewaters because of their simple

construction and increased cell retention time, which is

essential for hydrogen production [1e7]. The supportmatrices

can be made of recycled materials, and neither mechanical

agitation nor a recirculation apparatus is required in this

configuration, which results in low costs for reactor

ıa, Universidad de la [email protected], Hydrogen Energy P22

can help in the selection of microbial populations [8].

Cheese whey is an important byproduct generated during

cheese production. It is composed mainly of sugars (lactose,

70e72% dried extract), proteins (8e10%), and mineral salts

(12e15% dried extract) [9]. Because of the low buffer capacity

of cheese whey, its treatment in a conventional anaerobic

reactor frequently leads to acidification and inhibition of

methanogenic activity [10]. Therefore, its treatment in

ublica, Herrera y Reisig 565, CP.11300, Montevideo, Uruguay. Tel.:

(E. Castello).ublications, LLC. Published by Elsevier Ltd. All rights reserved.

Page 2: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

Fig. 1 e Schematic description of the packed bed reactor

used in this study showing the different sampling points.

i n t e r n a t i o n a l j o u rn a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 2 55

a two-phase anaerobic reactor is an important alternative

[11,12].

The production of hydrogen using cheese whey as

a substrate was previously studied using different reactor

configurations, which included a continuous stirred tank

reactor (CSTR) [13e15] and an upflow anaerobic sludge bed

(UASB) reactor [16,17]. The best performance was obtained

using the CSTR (maximum volumetric hydrogen production

rate of 33.7 L H2/L-d), a high organic loading rate

(138.6 g lactose/L-d) and low hydraulic residence time (HRT)

of 6 h [15].

However, this reactor configuration requires an external

sedimentation tank to separate the biomass and a significant

amount of energy for continuous mixing. Consequently,

packed bed reactors may provide an alternative to overcome

these problems. This reactor configuration has been success-

fully tested in the production of hydrogen using different

substrates, most of which were synthetic [2,3], as well as in

a few applications using industrial wastewaters [18]. Addi-

tional investigations of the application of this reactor config-

uration for the production of hydrogen from industrial

wastewaters are still needed.

In this sense, this work sought to test a packed bed reactor

configuration for the production of hydrogen using cheese

whey as the substrate and to evaluate the effect of increasing

the organic loading rate. To deeper understand the process is

crucial to know which microorganisms are involved in

hydrogen production and which are its main competitors. It

is also important to assess the proportion of hydrogen

producing microorganisms during reactor operation to

determine how to enrich in hydrogen producing bacteria. In

this work, we studied the microbial composition of biomass

in samples taken from different parts of the reactor

throughout the operation using a set of different techniques,

including T-RFLP, isolation and characterization of the

predominant organisms and quantification of the Fe-

hydrogenase genes by real time PCR. The potential to

produce hydrogen of the isolates was confirmed. The

microbial community composition was then related to the

operation of the reactor.

2. Materials and methods

2.1. The reactor system and operation

The total volume of the reactor was 3.75 L, the working

volume was 2.50 L and the height was 75 cm (Fig. 1). The

support material was recycled low-density polyethylene

(Interplast Embalagens Plasticas Ltda., Sao Carlos, SP, Brazil),

as described in Peixoto et al. [18]. The support matrix was

sterilized in an autoclave before use (Phoenix Ind. e Com. de

Equipamentos Cientıficos Ltda., Araraquara, SP, Brazil).

The reactor contained 7 sampling points: one at the

homogenization chamber (1P) and 6 points in the bed zone (2P,

3P, 4P, 5P, 6P, 7P) at spacing of 85 mm (Fig. 1).

The system was operated at 30 �C at different organic

loading rates, and with a constant HRT of 24 h, as calculated

for the liquid volume of the reactor at the beginning of

operation. The HRT was fixed at a value higher than those

typically used (less than 12 h) because the OLR was increased

by increasing the concentration of the cheese whey solution.

The start-up was performed with a cheese whey solution

prepared from cheese whey powder, with a concentration of

25 gCOD/L. After the start up, the concentration of the

cheese whey solution was increased maintaining the HRT at

24 h.

2.2. Inoculation

Raw cheese whey from a local cheese production factory

(Salute Prod. e Com. de Leite Ltda, Sao Carlos, SP, Brasil) was

naturally fermented for 3 days in contact with the atmosphere

to generate the inoculum. This method of inoculation was

used because is more feasible to be applied in large-scale

plants as compared to procedures that use methanogenic

inoculum subjected to chemical or thermal treatment. The

fermented wastewater was pumped into the reactor and

recycled for three days to ensure adhesion of the biomass [1].

2.3. Substrate

The reactor was fed with cheese whey powder (CWP), diluted

to a chemical oxygen demand (COD) concentration according

to the required OLR and supplemented with sodium bicar-

bonate (0.3e0.6 g NaHCO3/g COD) to control the pH in the

reactor. The carbohydrate content of the CWP was 72%, and

the protein content was 11%. The solution was stored at 4 �Cbefore being fed to the reactor.

Page 3: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 256

2.4. Analytical methods

Biogas production was measured with a Milligascounter gas

meter (Dr.-Ing. Ritter Apparatebau GmbH & Co., Bochum

Germany).

Chemical oxygen demand, total suspended solids (TSS)

and volatile suspended solids (VSS) were determined accord-

ing to standard methods [19].

Hydrogen levels were measured using a gas chromato-

graph (GC 2010 Shimadzu Corporation, Kyoto, Japan) equip-

ped with a Carboxen (1010 plot, 30 m� 0.53 mm) column and

thermal conductivity detector; argon was used as the carrier

gas. The temperatures of the injector and detector were

220 �C and 230 �C, respectively. The temperature of the

column was increased from 130 �C to 135 �C at a rate of 46 �C/min.

Solvent (acetone, methanol, ethanol, isobutanol, and n-

butanol) concentrations were measured by gas chromatog-

raphy using a GC 2010 Shimadzu� equipped with a flame-

ionization detector and a sample introduction system to

aCOMBI-PAL�headspace (AOC5000modelandHP-INNOWAX�

columnwith a film thickness of 30m� 0.25m� 0.25 mm).

Organic acids were analyzed with a Shimadzu� high-

pressure liquid chromatography (HPLC) system that was

composed of an LC-10ADvp pump, an FCV-10ALvp solenoid

valve, a CTO-10Avp oven (working temperature of 62 �C), anSCL-10Avp controller, an SPDM10Avp UV detector with

a diode array of 205 nm and an Aminex HPX-87H ion-

exchange column (0.3-m long with a 7.8-mm internal diam-

eter). The mobile phase was 0.005 M of H2SO4 and had a flow

rate of 0.8 mLmin�1. Carbohydrate concentrations were

determined by the colorimetric method developed by Dubois

et al. [20].

2.5. Microbial community analysis

Samples from the suspended biomass in the reactor were

collected at days 40, 47 and 54 of operation from the mixing

chamber (1P) and from the middle point in the bed zone

(4P). At the end of operation (day 56), the reactor was

dismantled, and a sample from all the biomass attached to

the support matrix was collected. For this sample, all the

support material with the biomass attached was separated

from the liquor and submerged in distilled water; the

biomass was then detached from the support by mechan-

ical agitation.

The composition of the microbial community was deter-

mined using T-RFLP analysis of the 16S rRNA gene. DNA was

extracted from samples taken from the reactor biomass and

inoculum using an UltraClean Soil DNA Extraction Kit (MO BIO

Laboratories Inc. Carlsbad, CA, USA) according to the manu-

facturer’s protocol. The PCR reaction (using primers 27

forward and 1492 reverse), purification, digestion with

restriction enzyme MspI, fragment separation and analysis

were performed as described elsewhere [16].

2.6. Isolation and characterization of the isolates

Aerobic bacteria were isolated in agar plates with TSA (DIFCO

Laboratories, Detroit, MI, USA) medium. Two different

methods were used to isolate anaerobic bacteria: serial dilu-

tion culturing and anaerobic plate culturing. Serial dilutions

(1/10) of samples with and without heat treatment were

cultured in anaerobic (under argon atmosphere) glucose-

containing medium (SigmaeAldrich, Germany, 10 g/L) liquid

medium (PYG medium) as previously described [16]. The

method was repeated until a single morphology was detected

by microscopic observation. The thermally treated samples

(100 �C, 15 min) were also cultured on plates with PYG agar

medium (supplemented with 2% agar (DIFCO Laboratories,

Detroit, MI, USA), incubated anaerobically (GENbag, Bio-

merieux Marcy l’Etoile, France)).

Lactic acid bacteria were isolated in agar plates with MRS

medium (DIFCO Laboratories, Detroit, MI, USA) incubated in

a CO2-enriched atmosphere.

All of the incubations were performed at 35 �C. The

hydrogen-producing capacity of the fermentative isolates was

tested in anaerobic liquid PYG or MRS medium.

The isolates were characterized by 16S rRNA gene

sequence analysis [16]. The sequences were compared to

those from the Ribosomal Database Project database using the

Seqmatch comparison tool [21].

2.7. Quantification of Fe-hydrogenase by real-time PCR

Real-time PCR was used to quantify the proportion of

hydrogen-producing bacteria that contained the enzyme Fe-

hydrogenase in samples from the biomass. The number of

copies of the Fe-hydrogenase gene in the DNA samples was

determined using the method described by Fang et al. [22].

The PCR reaction mixture consisted of 10 mL of SYBR Green

mix (Rotor Gene SYBR Green RT-PCR kit, Qiagen Inc., Valen-

cia, CA, USA), 1 mL of each primer and 8 mL of template. The

amplification conditions and primers used were the same as

those described by the authors [22]. A calibration curve was

constructed using 10-fold dilutions of a Fe-hydrogenase gene

PCR product from a Fe-hydrogenase clone (clone A10) as

standards. This PCR product was obtained by amplifying the

Fe-hydrogenase inserted into a plasmid using the primers for

the plasmid (T3 and T7). The clone was previously obtained

from a Fe-hydrogenase library [23]. The inserted sequence of

the clone A10 presented 60% homology with the Fe-

hydrogenase of Clostridium thermocellum ATCC 27405

(Accessing number YP 001036773) according to Blast X search

at NCBI [23]. The amplification reactions were performed in

a Rotor Gene 6000 (Corbett Research, Sidney, Australia). The

reaction was performed in duplicate, and the average and

standard deviation were determined. DNA was quantified

using a fluorometer (Qubit� 2.0, Invitrogen, Carlsbad, CA,

USA).

2.8. Accession number of the sequences

16S rRNA gene sequences from the bacterial isolates were

deposited in the NCBI database under the accession numbers:

strain L1: JX047846; strain L2: JX047847; strain P2: JX047850;

strain P3: JX047851; strain P5:JX047852; strain P6: JX047853

strain P7: JX047854 strain P8: JX047855; strain V1: JX047859;

strain V2: JX047860; strain V3: JX047861; strain V4: JX047862;

strain V5: JX047863.

Page 4: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

i n t e r n a t i o n a l j o u rn a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 2 57

3. Results and discussion

3.1. Reactor performance

The reactor operation was divided into 5 phases. The first 6

days were considered as an adaptation phase; gas production

was not detected, and the different operating variables were

adjusted. In phases 1 and 2, the reactor was operated at an

average organic loading rate of 22 g COD/L-d (Table 1). Because

of an operation problem with the controller during the adap-

tation phase and phase 1, the pH dropped to an average value

below 5 which is insufficient for hydrogen production. Then,

the pH was adjusted to values greater than 5 in phase 2

through the addition of NaHCO3 to the influent (Fig. 2a).

During phases 3 and 4, the OLR was increased to 33 g COD/L-

d and 37 g COD/L-d, respectively.

The COD removal values ranged between 14% and 32%,

mainly due to the production of organic fermentation prod-

ucts (acetic, lactic and butyric acids) because the removal of

lactose was always greater than 92%. Hydrogen production

was detected from day 7 onward and increased during oper-

ation to amaximum of 1.2 L H2/L-d detected at day 54 in phase

4 (Fig. 2a).

The increase in the operation pH (phase 2) and the increase

in the organic loading rate (phases 3 and 4) positively affected

both the hydrogen production rate and hydrogen yield

(Fig. 2a). The highest hydrogen production rate was observed

during phase 4 (Fig. 2a), and the production of methane was

always under the detection limit, which indicates that meth-

anogenesis was suppressed during operation.

Table 1 e Mean values of the organic loading rate (OLR), pH, COvolumetric hydrogen production rate (VHPR), yield (Y ) and per

Day OLRa

(g COD/L-d)pHa CODa

removal (%)

Adaptation

phase

0e6 25 4.0� 0.2b 0

Phase 1 7e32 22� 3c 4.8� 0.3d 14� 2e

Phase 2 33e40 5.7� 0.5j 32

Phase 3 41e47 33 6.2� 0.4m 16

Phase 4 48e56 37� 1o 5.6� 0.2p 18

aMean values� standard deviation.

b n¼ 5.

c n¼ 12.

d n¼ 17.

e n¼ 6.

f n¼ 16.

g n¼ 12.

h n¼ 5.

i n¼ 7.

j n¼ 6.

k n¼ 5.

l n¼ 4.

m n¼ 5.

n Two samples.

o n¼ 4.

p n¼ 6.

q n¼ 5.

The effect of the OLR on hydrogen production has been

studied previously. Hafez et al. [24] proposed an optimal

operational OLR of 103 g COD/L-d working with a CSTR reactor

fed with glucose and with the application of an HRT for 8 h.

The extremely high OLRs (154 and 206 g COD/L-d) resulted in

the selection of non-hydrogen fermenters and a decrease in

hydrogen production.

In the present work, hydrogen production was stimulated

by an increase in the OLR, and this trend was observed for

the three loads tested. The OLR was increased to 37 g COD/L-

d, which corresponds to a cheese whey concentration

of 37 g COD/L. At this concentration, neither the lactose

removal efficiency nor the hydrogen production was

inhibited; thus, the OLR could be further increased in future

works. The increase in the OLR should be induced by

working with a higher cheese whey concentration and not

by working at a lower HRT. The raw cheese whey contains

a high concentration of COD, with an average value of

60 g COD/L. The highest concentration of whey powder

solution used to feed the reactor was 37 g COD/L, which is

lower than concentration of the raw cheese whey, so

increasing this concentration would be necessary to avoid

dilutions as the ultimate goal is to find a solution to the

treatment of raw cheese whey.

The production of hydrogen in the packed bed reactor was

lower than those observed in other works that used cheese

whey as the substrate. Working with a CSTR fed with cheese

whey powder, Davila et al. [15] obtained a VHPR of 33.8 L H2/L-

d operating with an OLR of 138.6 g lactose/L-d and an HRT of

6 h. Using the same reactor as Davila et al. [15] but fed with

raw cheese whey instead of CWP, Venetsaneas et al. [25]

D removal, proportion of hydrogen in the gas (%H2),cent lactose conversion in each operation stage.

% H2a VHPRa

(L H2/L-d)Ya (mol H2/mol

lactose)Lactose

conversion (%)

0 0 0 e

6� 2f 0.06� 0.02g 0.05� 0.02h 92� 5i

5� 1j 0.20� 0.08j 0.20� 0.08k 98� 1l

10 m 0.8� 0.1m 0.53/0.60n 98.9/99.1n

10� 1p 1.0� 0.2q 0.668� 0.002q 98.9/9.0n

Page 5: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

0,0

1,0

2,0

3,0

4,0

5,0

6,0

7,0

8,0

0

0,2

0,4

0,6

0,8

1

1,2

1,4

1,6

0 10 20 30 40 50 60

pH

HP

R (

LH

2/L

d), H

Y (

mol

H2/

mol

la

ctos

e)

Time (d)

HPR HY pH

0

2000

4000

6000

8000

10000

12000

0 10 20 30 40 50 60

Con

cent

ratio

n (m

g/L

)

Time (d)

Lactic Acetic Propionic Butiric

a b

Fig. 2 e (a) Hydrogen-production rate (HPR), hydrogen yield (HY) and pH during operation, the different phases of operation

are indicated with dashed lines. (b) Organic acids measured at the outlet of the reactor during different phases of operation.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 258

obtained a maximum VHPR of 2.9 L/L-d for an OLR of 60 g

DQO/L-d and an HRT of 24 h.

In a UASB, Carrillo-Reyes et al. [26] obtained a VHPR of

1.7 L H2/L-d for an OLR of 48 g COD/L-d and an HRT of 8 h.

The UASB, however, exhibited the problem of methane

production.

Inapreviouswork [17] usingaUASB reactor inoculatedwith

kitchen waste compost and operated for more than 200 days,

maximum values of 2 L H2/L-d and a yield of 1.9 mol H2/mol

lactose consumedwere obtained. These valueswere two times

higher than those obtained in the packed bed configuration.

However, the production of hydrogen in the UASB reactor

exhibited high fluctuations, which were partially explained by

the use of raw cheese whey with a variable composition in the

feed. The new configuration tested resulted in a more stable

production of hydrogen, as shown by the lower dispersion of

the values. However, using the same configuration, Peixoto

et al. [18] observed higher HPR values, although these values

were highly dispersed; the authors reported that a decrease in

the HRT due to clogging problems could lead to a decrease in

the HPR value. The clogging problems reported by the authors

were caused by an excessive growth of biomass even though

the reactor was operated for 60 days.

In the packed bed reactor studied here, no clogging was

observed during operation. Awhite substance,which probably

consisted of a mixture of precipitated proteins and biomass,

accumulated in the homogenization chamber. This accumu-

lationof solids causedblocking problemsat the entrance of the

influent. This problem was minimized by purges from the

homogenization chamber (a total of 11 purges of 100-mL each).

This operation avoided the clogging of the bed and blocking of

theentrance.Thiscontrolof thebiomassgrowthbypurgingthe

excess of biomass could be the reason for which the hydrogen

production in this study was stable, whereas instabilities in

hydrogen production have been reported for this type of

reactor when biomass accumulates in the bed [27].

After the last day of operation, the reactor was dismantled,

and the biomass concentrations in the different zones were

determined. A higher proportion of biomass was detected in

the interstitial volume of the bed (48.3% in a volume of 2 L,

33.3 g of VS), whereas only 18.4% was attached to the support

matrix (in 800 g of matrix, 12.7 g of VS). The homogenization

chamber located at the bottom of the reactor contained 33.3%

of the biomass in a volume of 0.5 L (22.9 g of VS). The average

concentration of volatile solids throughout the reactor was

27.6 g VS/L, and low values of solids were detected in the out

stream (less than 1.5 g SSV/L). With respect to the biomass in

the entire reactor, the OLR applied to the biomass was

1.3 g COD/g VS-d. This value was far from the optimal value

obtained by Hafez et al. [24] (between 4.4 and 6.4 g COD/g VSS-

d) in a CSTR and that obtained in a previous work in which

a UASB reactor was used (3 g COD/L-d) [17].

3.2. Fermentation products

During operation, the production of organic acids in the

reactor changed (Fig. 2b). Lactic and acetic acids predominated

during the adaptation phase and phase 1; however, after the

pH values were adjusted to greater than 5, the butyric acid

production increased along with a decrease in the concen-

tration of lactic acid (phase 2).

This trend was also observed after the OLR was increased

and persisted until the end of operation. Clearly, the operation

at pH values of less than 5 favored lactic fermentation. This

effect could be related to the fact that lactate dehydrogenase is

induced under acidic conditions, and therefore, some of the

potential reducing power present in pyruvate was lost by its

diversion to lactate, as has been previously reported [28].

Ethanol was detected during all stages of operation but at

low proportions. The ethanol content exhibited higher values

during phase 3 (mean values of 149, 182, 421, 345 mg/L at

stages 1, 2, 3 and 4, respectively). Butanol was detected in the

effluent at days 40, 42 and 56 at concentrations of 17, 15 and

53 mg/L, respectively. A number of butyrate-producing clos-

tridia form small amounts of n-butanol. With some of these

clostridia, such as Clostridium beijerinckii, a shift from butyrate

production to solvent production has been observed [29]. In

this case, however, the butanol production was very low, and

butyrate fermentation continued.

No increase in the production of propionic acid was

detected during operation. According to Leite et al. [1], the

increase in pH strongly affected the overall production of

hydrogen and organic acids with the exception of propionic

acid, the production of which increased with increasing

alkalinity. The behavior reported by Leite et al. [1] was not

observed in this work.

3.3. Microbial community analysis

The microbial composition of the biomass was studied by T-

RFLP of 16S rRNA genes in samples taken from two different

points of the reactor (1P, the mixing chamber, and 4P, the

Page 6: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

i n t e r n a t i o n a l j o u rn a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 2 59

liquid phase at the midpoint of the bed zone). The samples

were taken at the end of each operational phase. On day 56,

the reactor was dismantled, and a sample from the biomass

attached to the support matrix was analyzed. The microbial

community from the inoculum was more diverse than those

from the reactor samples, suggesting that particular organ-

isms were selected during operation (Fig. 3a). A different

microbial population was observed in the samples taken at

the mixing chamber and suspended biomass in the reactor at

the different OLRs tested. Cluster analysis based on T-RFLP

profiles revealed two groups: onewith the samples taken from

the suspended biomass in the bed zone of the reactor and

anotherwith the samples taken from themixing chamber; the

sample taken from the inoculum forms a separated branch

(Fig. 3b).

There are two dominant peaks in the profiles from the

samples taken from the mixing chamber (177 and 569 nucle-

otides), whereas a peak of 67 nucleotides lengthwas dominant

in the samples taken from the middle of the bed zone.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

I (0) 1P(40)

1P(47)

1P(54)

SM(56)

4P(40)

4P(47)

4P(54)

Rel

ativ

e ab

unda

nce

(%)

Sample

580 569 554

539 515 296

214 212 193

186 181 177

130 124 119

113 110 103

95 92 90

80 76 67

64 57 55

53

01

23

45

67

89

0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1

Similarity

1P_(40)

1P_(47)

1P_(54)

SM_(56)

4P_(40)

4P_(47)

4P_(54)

I_(0)

a

b

Fig. 3 e (A) 16S rRNA gene T-RFLP community analyses in

samples obtained from the inoculum, mixing chamber (1P)

and suspended biomass from the bed zone (4P) during

operation at different organic loading rates. A sample

taken from the biomass attached to the support matrix

(SM) was analyzed at the end of operation. (b) Cluster

analysis using paired-group and Morisita-only peaks that

represented more than 2% of the total intensity were

considered.

The sample taken from the biomass attached to the

supportmatrix showed a high relative abundance of the peaks

that corresponded to 177 and 186 nucleotides and exhibited

greater similarity with the samples taken from the mixing

chamber according to the cluster analysis (Fig. 3b).

To determine the identity of the microorganisms from the

biomass, strains were isolated from different points of the

reactor in samples taken at the end of operation (day 56). The

isolateswere classifiedwithin the generaKlebsiella,Clostridium,

Enterococcus, Streptococcus, Pseudomonas and Lactobacillus

according to the 16S rRNA gene sequence analysis (Table 2).

The capacity to produce hydrogen was tested for the

different isolates in batch cultures. As expected, only the

isolates characterized within the genera Clostridium and Kleb-

siella produce H2 by fermentation (Table 2).

The predicted T-RF lengths were determined from the 16S

rRNA gene sequences and correlated to the T-RFLP profiles.

From the 13 isolates, 11 were correlated to T-RFLP peaks.

These isolates were characterized within the genera Strepto-

coccus, Enterococcus, Pseudomonas, Lactobacillus and Clos-

tridium. In particular, strains characterized as Lactobacillus and

Clostridium were isolated from different zones of the reactor

and detected in several samples in the T-RFLP, which indi-

cated a prevalence of these organisms in the reactor (Table 2).

Themain fermentation products of the isolates (acetic, butyric

and lactic acid) were in accordance with the fermentation

products detected in the reactor. Propionic bacteria were not

selected during operation, and propionic acid was not

produced.

The real-time PCR quantification results showed that the

largest proportion of bacteria with the gene encoding the Fe-

hydrogenase was obtained when the load was 37 g COD/L-

d (Fig. 4). This result corresponded with the increase in the

production of butyric acid and hydrogen, which indicates the

prevalence of hydrogen production via the butyric pathway

during this phase. Hydrogen-producing bacteria represented

by the Fe-hydrogenase gene were detected throughout the

reactor, with a higher proportion present in the sample taken

from the suspended biomass in the midpoint of the bed zone

(4P) than in the mixing chamber (1P) or the biomass attached

to the bed matrix (SM). These results confirm that the distri-

bution of the biomass was not homogeneous.

According to the microbiological analysis, the organisms

involved in the hydrogen production belonged to the genera

Clostridium and Klebsiella. The presence of Klebsiella has been

previously reported to support the competition of Clostridium

with other undesired microorganisms to enhance hydrogen

production [30].

The presence of Klebsiella could be responsible for the low

amounts of ethanol observed in the effluent [29].

Although hydrogen production increased during opera-

tion, fermenters with the capacity to produce hydrogen did

not dominate in the reactor biomass (T-RF correlated to the

Clostridium was always less than 4%, and organisms from the

Klebsiella genus were only detected by isolation). The persis-

tence of a mixed microbial population with a low proportion

of hydrogen-producing bacteria could explain the low yield

compared with the maximum theoretical expected values.

Other organisms without the capacity to produce hydrogen

werealsodetected (Streptococcus,Enterococcus, andPseudomonas).

Page 7: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

Table 2 e Characterization of the isolates according to an analysis of 16S rRNA genes and hydrogen production in batchcultures. Closer relatives were determined according to a comparison of 16S rRNA sequenceswith sequences from the RDPdatabase using the Seqmatch tool; only sequences from isolateswere taken into account for the comparison. The expectedfragment size was calculated from the 16S rRNA gene sequence and correlated to T-RFLP peaks. The production of H2 wasdetermined for each fermentative isolate in anaerobic liquid PYG or MRSmedium. Fermentation products according to thebibliography [29] are shown.

Strain Samplingpoint

Closer relativea H2 CorrelatedT-RFLP peakb

Sample thatpresented this T-RFc

Fermentationproducts

P3, P6 1P, 2P Streptococcus lutetiensis NEM 782 (AJ297215) (1.000) � 555 Inoculum Lactic

P2 1P Klebsiella sp. HL1 (AB074192) (0.995) þ 494 N. d. 2,3-Butanediol, lactic,

acetic, formic, ethanol

P5 SM Enterococcus gallinarum (AF039900) (1.000) � 59 4P (54, 56) Lactic

P7 6P Pseudomonas stutzeri DSM 5190 (AJ288151) (0.997) � 490 N. d. Not a fermenter

P8 7P Pseudomonas aeruginosa; S25 (DQ095913) (1.000) � 143 4P (40, 47, 54) Not a fermenter

V1, L1 1P Lactobacillus casei ATCC334 (D86517.1) (0.997) � 569 1P; 4P (40, 47, 54) Lactic, acetic

L2 SM Lactobacillus brevis; NRIC 1684 (AB024299) (0.982) � 569 1P; 4P (40, 47, 54) Lactic, acetic

V3 SM Clostridium beijerinckii DSM791; (X68179.1) (0.995) þ 515 1P; 4P (47, 54) Acetic, butyric

V2, V4, V5 1P, SM, 4P Clostridium tyrobutyricum NIZO 51; (L08062) (0.994) þ 515 1P; 4P (47, 54) Acetic, butyric

a Accession numbers of the sequences and S-ab, a score that indicates the homology between the sequences by the RDP Seq-match tool (shown

in brackets).

b The predicted T-RF lengths (in nucleotides) were determined from the sequences by “in sillico” digestion using the enzyme Msp I.

c Samples with the presence of the T-RF peak and the day of sampling are indicated. SM: biomass attached to the supportmatrix, 4P: suspended

biomass from the middle point of the bed zone, 1P: mixing chamber. N.d: T-RF not detected in the T-RFLP chromatograms.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 260

These organisms may compete for the substrate, thereby

reducing the efficiency of the process. But, on the other hand,

their presence could also have a positive effect helping the

growth of the hydrogen producing reactors. In a recent review,

Hung et al. [31] summarized the possible role of organisms

described in several hydrogen-producing bioreactors. Accord-

ing to this review, organisms from the genera Streptococcus and

Pseudomonas could contribute to the anaerobiosis of the system

necessary for the growth of strict anaerobic organisms, such as

Clostridium, or to the degradation of complex carbon sources.

The presence of Clostridium bacteria correlatedwith butyric

acid production and accompanied the production of

hydrogen. In particular, strains affiliated with Clostridium

tyrobutyricum were isolated in samples taken from all reactor

zones, including the biomass attached to the matrix. This

organism has been reported to ferment lactate to butyrate,

0

2000

4000

6000

8000

10000

12000

14000

Inoculum day 40 day 47 day 54 SM day 56

Cop

y nu

mbe

r H

ydro

gena

se/n

gDN

A

Sample

Mixing chamber (1P)

Suspended biomass (4P)

Fig. 4 e Quantification of Fe-hydrogenase by real-time PCR

in samples taken from the inoculum, mixing chamber (MC)

and midpoint of sampling in the bed zone (suspended

biomass) at different OLRs. A sample from the biomass

attached to the support matrix (SM) was analyzed at the

end of operation (day 56). The error bars represent the

standard deviation of triplicate measurements.

CO2 and H2 in the presence of acetate [32]. Thus, the presence

of this organism could favor the production of hydrogen by

the transformation of lactic acid (generated by lactic fermen-

tation) into butyric acid with H2 production.

There are two dominant peaks in the T-RFLP profiles of the

samples taken from themixing chamber (corresponding to 177

and 569 nucleotides). These peaks were correlated to organ-

isms from the genus Lactobacillus detected in this study (Table

2) and in a previous study [16]. These organisms represented

a bacterial community proportion of 82% on day 40; this

proportion decreased to 25% on the last day of operation. This

decrease in the relative proportion of Lactobacillus correlated to

the increase inH2production, suggesting that lactic fermenters

were the main competitors with the H2-producing capability.

Nevertheless, the role of members of the Lactobacillus

genus in the production of hydrogen is not yet clear. Hung

et al. [31] have suggested that hydrogen production could not

be associated with the presence of Lactobacillus in the reactors

because no previous reports on the production of hydrogen

from organisms of the Lactobacillus genera have been pub-

lished. Nevertheless, we found two works in which hydrogen

production was associated with lactobacilli [14,33]. In partic-

ular, Lactobacillus bifermentans exhibited the capability to

produce hydrogen and CO2 from lactate [33]. However, the

inhibition of Clostridium growth by Lactobacillus has also been

reported [34]; thus, whether the presence of Lactobacillus has

a positive or negative effect on hydrogen production remains

unknown.

Both the T-RFLP and real-time PCR analysis indicated that

the biomass exhibited different compositions in the different

reactor zones, with a predominance of butyric-producing

organisms in the suspended biomass in the bed zone and

a predominance of lactic acid bacteria in the mixing chamber.

A similar behavior has also been reported by Lee et al. [3], who

determined that suspended cell growth is responsible for

more hydrogen production than is solid-phase cell growth.

Page 8: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

i n t e r n a t i o n a l j o u rn a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 2 61

4. Conclusions

In the present study, the feasibility of producing hydrogen in

a packed bed reactor using cheesewhey powder as a substrate

at three different OLRs was demonstrated. No operational

problems related to bed clogging, the persistence of meth-

anogenesis or a high production of solvents were detected.

Although the yield was not as high as those obtained in other

reactor configurations, the production was stable, with

a tendency to increase at higher loads. The use of a higher OLR

and lower HRT should be further investigated to determine

the adequate conditions for the treatment of cheese whey in

the concentration produced by industry. This optimization

will provide advantages related to the size of the equipment

and water savings, which will make this alternative more

economical for treating this byproduct.

A mixed microbial community composed of low predomi-

nance of hydrogen-producing bacteria (such as Clostridium

and Klebsiella) and several non-hydrogen producers was

selected during operation.

Cheese whey treatment in two stages, an acidogenic phase

with hydrogen production in a packed bed reactor and

a second methanogenic step, could be a feasible and low-cost

alternative for real-scale applications that should be further

studied to improve the efficiency of hydrogen production.

Acknowledgements

This research was funded by a research grant from OPCW and

a collaborative project between Brazil and Uruguay (490967/

2008-6) CNPq/DICYT. J. Wenzel and V. Perna were funded by

a grant from the Agencia Nacional de Investigacion e Inno-

vacion (ANII), Uruguay.

r e f e r e n c e s

[1] Leite J, Fernandes B, Pozzi E, Barboza M, Zaiat M. Applicationof an anaerobic packed-bed bioreactor for the production ofhydrogen and organic acids. Int J Hydrogen Energy 2008;33:570e86.

[2] Chang J, Lee K, Lin P. Biohydrogen production with fixed-bedbioreactors. Int J Hydrogen Energy 2002;27:1167e74.

[3] Lee K, Lo Y, Lo Y, Lin P, Chang J. H2 production withanaerobic sludge using activated-carbon supported packed-bed bioreactors. Biotechnol Lett 2003;25:133e8.

[4] Yokoi H, Aratake T, Hirose J, Hayashi S, Takasaki Y.Simultaneous production of H2 and bioflocculant byEnterobacter sp. BY-29. World J Microbiol Biotechnol 2001;17:609e13.

[5] Rachman M, Nakashimada Y, Kakizono T, Nishio N.Hydrogen production with high yield and high evolution rateby self-flocculated cells of Enterobacter aerogenes in a packed-bed reactor. Appl Microbiol Biotechnol 1998;49:450e4.

[6] Kumar N, Das D. Continuous hydrogen production byimmobilized Enterobacter cloacae IIT-BT 08 usinglignocellulosic materials as solid matrices. Enzym MicrobTechnol 2001;29:280e7.

[7] Lin CN, Wu S, Chang JS. Fermentative hydrogen productionwith a draft tube fluidized bed reactor containing silicone-

gel-immobilized anaerobic sludge. Int J Hydrogen Energy2006;31:2200e10.

[8] Silva A, Hirasawa J, Varesche M, Foresti E, Zaiat M. Evaluationof support materials for the immobilization of sulfate-reducing bacteria and methanogenic archea. Anaerobe 2006;12:93e8.

[9] Panesar P, Kennedy J, Gandhi D, Bunko K. Bioutilisation ofwhey for lactic acid production e review. Food Chem 2007;105:1e14.

[10] Garcıa P, Rico J, Polanco F. Anaerobic treatment of cheesewhey in a two-phase UASB reactor. Environ Technol 1991;12:355e61.

[11] Saddoud A, Hassaıri I, Sayadi S. Anaerobic membranereactor with phase separation for the treatment of cheesewhey. Bioresour Technol 2007;98:2102e8.

[12] Demirel B, Yenigun O, Turgut T. Anaerobic treatment of dairywastewaters: a review. Process Biochem 2005;40:2583e95.

[13] Azbar N, CetinkayaDokgoz FT, Keskin T, Korkmaz KS,Syed HM. Continuous fermentative hydrogen productionfrom cheese whey wastewater under thermophilic anaerobicconditions. Int J Hydrogen Energy 2009;34:7441e7.

[14] Yang P, Zhang RM, Benemann JR. Biohydrogen productionfrom cheese processing wastewater by anaerobicfermentation using mixed microbial communities. Int JHydrogen Energy 2007;32:4761e71.

[15] Davila-Vazquez G, Cota - Navarro CB, Rosales - Colunga LM,de Leon- Rodrıguez A, Razo- Flores E. Continuousbiohydrogen production using cheese whey: improving thehydrogen production rate. Int J Hydrogen Energy 2009;34:4296e304.

[16] Castello E, Garcıa y Santos C, Iglesias T, Paolino G, Wenzel J,Borzacconi L, et al. Feasibility of biohydrogen productionfrom cheese whey using a UASB reactor: links betweenmicrobial community and reactor performance. Int JHydrogen Energy 2009;34:5674e82.

[17] Castello E, Perna V, Wenzel J, Borzacconi L, Etchebehere C.Microbial community composition and reactor performanceduring the hydrogen production in a UASB reactor fed withraw cheese whey inoculated with compost. Water SciTechnol 2001;64:2265e73.

[18] Peixoto G, Saavedra NK, Varesche MB, Zaiat M. Hydrogenproduction from soft-drink wastewater in an upflowanaerobic packed-bed reactor. Int J Hydrogen Energy 2011;36:8953e66.

[19] Standard methods for the examination of water andwastewater. 19th ed.; 1995.

[20] Dubois M, Gilles KA, Hamilton JK, Smith F. Colorimetricmethod for determination of sugar and related substances.Anal Chem 1956:350e6.

[21] Crolla I, Paolino G, Castello E, Etchebehere C. Evaluation ofthe biohydrogen production capacity by anaerobic darkfermentation from different industrial wastewater. In:Proceedings of the World hydrogen Technology Convention,December 2007.

[22] Fang H, Zhang T, Ch Li. Characterization of Fe-hydrogenasegenes diversity and hydrogen-producing population in anacidophilic sludge. J Biotechnol 2006;126:357e64.

[23] Cole JR, Wang Q, Cardenas E, Fish J, Chai B, Farris RJ, et al.The ribosomal database project: improved alignments andnew tools for rRNA analysis. Nucl Acids Res 2009;37:141e5.

[24] Hafez H, Nakhla G, Naggar MHE, Elbeshbishy E,Baghchehsaraee B. Effect of organic loading on a novelhydrogen bioreactor. Int J Hydrogen Energy 2010;35:81e92.

[25] Venetsaneas N, Antonopoulou G, Stamatelatou K,Kornaros M, Lyberatos G. Using cheese whey for hydrogenand methane generation in a two-stage continuous processwith alternative pH controlling approaches. BioresourTechnol 2009;100:3713e7.

Page 9: Hydrogen production in an upflow anaerobic packed bed reactor used to treat cheese whey

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 4e6 262

[26] Carrillo-Reyes J, Celis L, Alatriste-Mondragon F, Razo-Flores E. Different start-up strategies to enhancebiohydrogen production from cheese whey in UASB reactors.Int J Hydrogen Energy 2012;37:5591e601.

[27] Lima DMF, Zaiat M. The influence of the degree of back-mixing on hydrogen production in an anaerobic fixed-bedreactor. Int J Hydrogen Energy 2012;37:9630e5.

[28] Hallenbeck P, Ghosh D. Advances in fermentativebiohydrogen production: the way forward? TrendsBiotechnol 2009;27:287e97.

[29] Gottschalk G. Bacterial metabolism. 2nd ed. New York:Springer; 1986.

[30] Pattra S, Lay C-H, Lin C-Y, O-Thong S, Reungsang A.Performance and population analysis of hydrogenproduction from sugarcane juice by non-sterile continuous

stirred tank reactor augmented with Clostridium butyricum.Int J Hydrogen Energy 2011;36:8697e703.

[31] Hung C-H, Chang Y-T, Chang Y- J. Roles of microorganismsother than Clostridium and Enterobacter in anaerobicfermentative biohydrogen production systems e a review.Bioresour Technol 2011;102:8437e44.

[32] Barlow H, Truper HG, Dworkin M, Hardeer W, Schliefer H.The prokaryotes: a handbook on the biology of bacteria. 2nded. New York: Springer-Verlag; 1991.

[33] Kandler O, Schillinger U, Weiss N. Lactobacillus bifermentanssp. nov., nom. rev., an organism forming CO2 and H2 fromlactic. Syst Appl Microbiol 1983;4:408e12.

[34] Noike T, Takabatake H, Mitzuno O, Ohba M. Inhibition of H2

fermentation of organic wastes by lactic acid bacteria. Int JHydrogen Energy 2002;27:1367e71.