A

WfipwebaastCrfi©

K

1

emesrpiwq

1d

Biochemical Engineering Journal 32 (2006) 33–42

Hydrogen production by indigenous photosynthetic bacteriumRhodopseudomonas palustris WP3–5 using optical

fiber-illuminating photobioreactors

Chun-Yen Chen a, Chi-Mei Lee b, Jo-Shu Chang a,∗a Department of Chemical Engineering, National Cheng Kung University, Tainan 710, Taiwan, ROC

b Department of Environmental Engineering, National Chung Hsing University, Taichung, Taiwan, ROC

Received 14 September 2005; received in revised form 30 May 2006; accepted 21 August 2006

bstract

A novel optical fiber-based photobioreactor was utilized to produce H2 by indigenous purple nonsulfur bacterium Rhodopseudomonas palustrisP3–5 using acetate as the sole carbon source. Plastic cladding of conventional end-light optical fibers was removed to obtain side-light optical

bers (SLOF), which was inserted into photobioreactors as the internal light source. External irradiation by conventional lamps may also berovided for the bioreactor as supplemental light sources. The H2 production performance and light conversion efficiency of the photobioreactorere assessed when various illumination systems were used. The light sources examined included SLOF excited by halogen lamp (OF-HL), SLOF

xcited by metal–halide lamp (OF-MH), tungsten filament lamp (TL), halogen lamp (HL), and binary combinations of the above. Compared withioreactors illuminated by external lamps, the OF-HL system produced more H2 (625 ml), had higher light conversion efficiency (1.80%), andchieved higher H2 yield (1.19 mol H2/mol acetic acid). However, among the single light sources examined, HL gave the highest overall (νH2 )nd specific (νs,H2 ) H2 production rate of 8.68 ml/l h and 3.01 ml/h g cell, respectively, primarily due to enabling better cell growth. Using OF-MHystem resulted in poor H2 production, indicating that emission spectrum of light sources was critical to photo-H2 production. Combination ofwo different light sources appeared to further enhance photo-H production, especially when optical fibers and external lamps were combined.

2

ombination of OF-HL and TL exhibited the highest H2 yield, νH2 , and νs,H2 of 2.64 mol H2/mol acetic acid, 17.06 ml/h l, and 9.47 ml/h g cell,espectively. However, the highest total H2 production (944 ml) and light conversion efficiency (1.42%) were attained when two types of opticalbers were incorporated (i.e., the OF-HL/OF-MH system).2006 Elsevier B.V. All rights reserved.

eudo

sqtod[rfT

eywords: Photohydrogen production; Photobioreactor; Optical fiber; Rhodops

. Introduction

Biological production of H2 is considered the mostnvironment-friendly route of producing H2 [1–3], which is theost promising alternative to fossil fuels because it is clean,

fficient, and recyclable. Microbial conversion of organic sub-trates into H2 by light-dependent (e.g., photosynthetic bacte-ia) or light-independent (e.g., acidogenic bacteria) metabolicathways [4] is of great interest due to the potential of produc-

ng clean energy (H2) from renewable resources (e.g., organicastes). In particular, photosynthetic bacteria have been fre-uently used to produce H2 because they have high theoretical

∗ Corresponding author. Tel.: +886 6 2757575x62651; fax: +886 6 2357146.E-mail address: changjs@mail.ncku.edu.tw (J.-S. Chang).

ca[

ctc

369-703X/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.bej.2006.08.015

monas palustris

ubstrate conversion efficiency and produce a relatively smalluantity of by-products (such as CO2) [5,6]. In addition, pho-osynthetic bacteria can produce H2 from mineralization ofrganic acids (e.g., acetic acid, butyric acid), which are pre-ominant soluble metabolites from dark hydrogen fermentation3]. This raises the possibility of using photosynthetic bacte-ia to degrade effluents from dark H2 fermentation stage forurther H2 production and more complete biodegradation [7,8].herefore, dark and photo-H2 production systems have beenombined in series to produce H2, achieving a higher yieldnd a lower chemical oxygen demand (COD) in the effluent5,7,9].

Hydrogen production by photosynthetic bacteria is mainlyatalyzed by nitrogenase [10], allowing evolution of H2 inhe absence of molecular nitrogen and oxygen, but with theonsumption of ATP and free electrons originating from the

3 ngine

roAeabeptiao[fsbCp

vpteettbcoriaIlpog

Rwplemboodpe

2

2

n

icl0emm2Ho5ttt

2

lCcwtdlletto

2

iiwmhesfiawro2maGdu

4 C.-Y. Chen et al. / Biochemical E

educing power [1]. Thus, sufficient supply of ATP becomes onef the major concerns for efficient photo-H2 production. SinceTP synthesis in photosynthetic bacteria is a light-dependentvent, requiring light energy primarily at the wavelength of 522nd 860 nm [11], it would be critical for a H2-producing photo-ioreactor to use proper light sources that provide sufficient lightnergy with needed wavelengths. However, in conventionalhotobioreactors using external illumination systems (e.g.,ungsten filament lamp or halogen lamp) [5,12–14], the lightntensity tends to decrease rapidly due to the shielding effectsrising from increases in the concentration of cells and productsr from formation of biofilm on the surface of reactor vessels15]. Furthermore, although a short light path is theoreticallyavorable for achieving high light efficiency, conventional lightources cannot be in close contact with the bacterial cultureecause they usually generate a considerable amount of heat.onsequently, the light conversion efficiency of conventionalhotobioreactors has been limited to less than 10% [16,17].

Due to the problems and limitations associated with con-entional light sources, this work aimed to develop a novelhotobioreactor using optical fibers as the internal illumina-ion system. Meanwhile, conventional lamps were also installedxternally as the supplemental light sources. Optical fiber isxpected to markedly enhance light conversion efficiency ofhe photobioreactor because it provides uniform light distribu-ion [18,19] with a high surface-to-volume ratio [20] and cane directly immersed in the bacterial culture to achieve effi-ient light energy transfer without heat generation. Althoughptical fibers have been applied in TiO2-based photo-catalyticeactors [21], the idea of using optical fiber in biological systemss relatively novel. Most of the optical fiber-based photobiore-ctors were used for microbial desulfurization [15,19,20,22].n contrast, to date only two reports describing using end-ight [23] or light-diffusing [18] optical fibers for photo-H2roduction. In both cases, cells were immobilized on theptical fibers by natural gels (e.g., alginate, agar, and gellanum).

In this study, a purple nonsulfur photosynthetic bacteriumhodopseudomonas palustris WP3–5 isolated from central Tai-an [5] was used to produce H2 in an optical fiber-implementedhotobioreactor. The optical fibers were excited by differentight engines (halogen or metal–halide lamps) to investigate theffect of light emission spectra on photo-H2 production. Perfor-ance of the H2-producing photobioreactors was also examined

y using various single light sources and binary combinationsf internal (optical fibers) and external (tungsten filament lampr halogen lamp) light sources. The goal of this study was toevelop a photobioreactor capable of producing H2 at a fastroduction rate, a high H2 yield, and enhanced light conversionfficiency.

. Materials and methods

.1. Organism and medium

The bacterial H2 producer used in this study was an indige-ous photosynthetic bacterium, R. palustris WP3–5, which was

rad(

ering Journal 32 (2006) 33–42

solated from a swine wastewater treatment system located inentral Taiwan [5]. The bacterium was grown with Rhodospiril-aceae medium [5], consisting of (in g/l) K2HPO4, 1.5; KH2PO4,.5; MgSO4·7H2O, 0.2; NaCl, 0.4; CaCl2·2H2O, 0.05; yeastxtract, 0.2; iron citrate solution (1.0 g/l), 5 ml/l; trace ele-ent solution, 1 ml. The trace element solution contained (ing/l) ZnCl2, 70; MnCl2·4H2O, 100; H3BO3, 60; CoCl2·6H2O,

00; CuCl2·2H2O, 20; NiCl2·6H2O, 20; NaMoO4·2H2O, 40;Cl (25%), 1 ml/l. The cells were cultivated at 32 ◦C anaer-bically for 48 h under a light intensity of approximately0 W/m2 (illuminated by tungsten filament lamp). The ini-ial pH value of medium prior to incubation was adjustedo 7.0–7.1. Argon gas was used to create anaerobic condi-ions.

.2. Preparation of optical fiber

Plastic-clad optical fibers (11 mm in diameter, 25 cm inength) purchased from Baycom Optic-Electronic Co. (Hsin-hu, Taiwan) were used in this study. The optical fiber wasomposed of a polymethyl methacrylate (PMMA) core coatedith fluorinated alkyl methacrylate copolymer. The protec-

ive cladding was removed by mechanical polishing to allowirect light emission from the PMMA core (i.e., a side-ight optical fiber). One of the two fiber-ends on which theight is incident was also polished to attain maximum lightmission. Prior to being installed inside the photobioreac-or, the side-light optical fiber was physically polished untilhe desired light intensity and uniform light distribution werebtained.

.3. Fabrication and operation of photobioreactor

The photobioreactor was a sealed glass vessel with a work-ng volume of 500 ml (Fig. 1). The side-light optical fiber wasnserted into the photobioreactor from the top. The optical fiberas excited to achieve a light intensity of ca. 95 W/m2 by aetal–halide lamp (150 W; Gorich Co., Hsin-Chu, Taiwan) or a

alogen lamp (150 W; Gorich Co., Hsin-Chu, Taiwan). In somexperiments, external light sources were also mounted in bothides of the bioreactor (Fig. 1b) using a conventional tungstenlament lamp (100 W) or a halogen lamp (100 W), resulting inlight intensity of ca. 95 W/m2. Cells of R. palustris WP3–5ere inoculated into the reactor with a 10% inoculum. The

eactor was operated at 32 ◦C, pH 7.1, and an agitation ratef 100 rpm. The initial acetate concentration was maintained at667 mg/l in all tests. A gas meter (Type TG1; Ritter Inc., Ger-any) was used to measure the amount of gas products generated

nd the gas volumes were calibrated to 25 ◦C and 760 mmHg.as samples were taken from sampling port by gas syringe atesired time intervals to measure the gas composition. The liq-id sample was also collected from the sealed glass vessel with

espect to time to determine cell concentration, pH and residualcetate concentration. Time-course data of cumulative H2 pro-uction were simulated by modified Gompertz equation (Eq.1)) [24,25] to determine the kinetic parameters of photo-H2

C.-Y. Chen et al. / Biochemical Engineering Journal 32 (2006) 33–42 35

F cal fib

p

H

w

ig. 1. Schematic diagrams of the photobioreactor system (a) with internal (opti

roduction.

= Hmax exp

{−exp

[Rmaxe

p(λ − t) + 1

]}(1)

mHtg

er) light sources and (b) with combination of internal and external light sources.

here H denotes the cumulative H2 production (ml), Hmax the

aximum cumulative H2 production (ml), Rmax the maximum2 production rate (ml/h), t the culture time (h), and λ denotes

he lag time required for the onset of H2 evolution (h). The hydro-en production performance was mainly assessed by maximum

3 ngine

cpa

O

S

H

2W

dbmvi(3H

2

wη

η

wHai2U

2

b(Tct

subtta

(icwaoTf5af

cTu1tr

3

3

w(t((Pmili

3W

iafbibi

6 C.-Y. Chen et al. / Biochemical E

umulative H2 production, overall H2 production rate, specificroduction rate, and H2 yield. The definition of these parametersre indicated as follows:

verall H2 production rate

= Cumulative H2 production (ml)

Culture time for H2 evolution (h) × working volume (l)

(2)

pecific H2 production rate

= Cumulative H2 production (ml)

Culture time for H2 evolution (h) × cell mass (g)

(3)

2 yield = Amount of H2 produced (mol)

Amount of substrate (acetate) consumed (mol)(4)

.4. Absorption spectrum of intact cells of R. palustrisP3–5

The absorption spectrum of cells of R. palustris WP3–5 wasetermined according to the procedures proposed by [26]. Inrief, overnight culture of R. palustris WP3–5 (5 g dry cell/l) wasixed with sucrose (Merck, New Jersey, USA) with a weight to

olume ratio of 5 g sucrose/3.5 ml cell suspension. The result-ng cell/sucrose mixture was subjected to a full-spectrum scanfrom 200 to 1000 nm) against a blank solution of 5 g sucrose in.5 ml deionized water by a spectrophotometer (Model U-2001,itachi, Tokyo, Japan).

.5. Estimation of light conversion efficiency

Light conversion efficiency (η) is defined as the efficiency byhich the light energy can be transformed into H2 energy. Thevalue was calculated according to Eq. (5) [4,27]:

(%) = H2 output × H2 energy content

light energy input

×100 = 33.61ρH2VH2

IAt× 100 (5)

here VH2 is the volume of H2 produced (l), ρH2 the density of2 produced (g/l), I the light intensity (W/m2), A the irradiated

rea (m2), and t is the duration of H2 production (h). The lightntensity (I) was measured with a LI-250 Light Meter with a LI-00SA pyranometer sensor (LI-COR Inc., Lincoln, Nebraska,SA) that gave a unit of W/m2.

.6. Analytical methods

The cell concentration in the photobioreactor was determinedy optical density measurement at a wavelength of 660 nm

i.e., OD660) with a spectrophotometer (Model U-2001, Hitachi,okyo, Japan) after proper dilution with deionized water. Theell dry weights were obtained by filtering 10 ml aliquots of cul-ure through a cellulose acetate membrane filter (0.45 �m pore

bsfe

ering Journal 32 (2006) 33–42

ize, 47 mm in diameter). Each loaded filter was dried at 105 ◦Cntil the weight was invariant (about 72 h). The dry weight oflank filter was subtracted from that of the loaded filter to obtainhe dry cell weight (DCW). The OD660 values were convertedo DCW concentration via proper calibration (i.e., 1.0 OD660pproximately equals 4.0 g dry cell/l).

Acetate concentration was analyzed by gas chromatographyGC-14B, Shimadzu, Tokyo, Japan) equipped with a flame ion-zation detector (FID). Samples were injected into a 15 m longapillary column (Type no. 11052, Restek, Bellefonte, PA, USA)ith an internal diameter of 0.53 mm. Nitrogen was employed

s the carrier gas with a flow rate of 20 ml/min. The temperaturef injector and detector was set at 220 and 230 ◦C, respectively.he oven temperatures were initially set at 110 ◦C, increased

rom 110 to 200 ◦C at a rate of 8 ◦C/min, and held at 200 ◦C formin. Liquid samples were centrifuged (6000 × g for 20 min)nd filtered (0.45 �m membrane) prior to being injected into GCor analysis.

The gas products (H2 and CO2) were also analyzed by gashromatography (Model 9800, China Chromatography, Taipei,aiwan) using a thermal conductivity detector. The carrier gassed was argon and the column (0.53 mm in inner diameter and5 m in height) was packed with Porapak Q. The temperature athe column was initially 50 ◦C and was increased to 200 ◦C at aate of 30 ◦C/min. The temperature at injector was 140 ◦C.

. Results and discussion

.1. Properties of the side-light optical fiber (SLOF)

Conventional plastic-clad end-light optical fibers (Fig. 2a)ere polished mechanically to obtain side-light optical fibers

Fig. 2b). The SEM micrograms show that the surface struc-ure of the original plastic-clad optical fiber was very smoothFig. 2c), while the surface became rough after polishingFig. 2d), allowing light emission from the entire surface ofMMA core. Light distribution of the SLOF was determined byeasuring the light intensity along the fiber surface. As shown

n Fig. 3, the SLOF excited by metal–halide and halogen lightamps displayed a uniform light distribution with a constant lightntensity of ca. 95 W/m2.

.2. Absorption spectrum of intact cells of R. palustrisP3–5

The absorption spectrum of R. palustris WP3–5 is illustratedn Fig. 4a. The major absorption bands are located at 400, 805,nd 850 nm. Light bands at 800 and 850 nm are known to resultrom absorption by bacteriochlorophylls of the purple nonsulfuracterium [16], while the light absorption bands around 400 nms contributed by bacteriochlorophylls and hemes [16]. Sinceacteriochlorophyll is the primary machinery for ATP synthesisn photosynthetic bacteria [28], the light sources applied should

e compatible with the absorption bands of R. palustris WP3–5hown in Fig. 4a to ensure excitation of bacteriochlorophyllor efficient photo-H2 production and to prevent waste of lightnergy [15]. Light emission profiles of the three light sources

C.-Y. Chen et al. / Biochemical Engineering Journal 32 (2006) 33–42 37

F orphoo

[se4am(t

Fe

3

s

ig. 2. (a) The untreated optical fiber, (b) the side-light optical fiber, (c) surface mf the side-light optical fiber after polishing.

15,16,19] used in this work are illustrated in Fig. 4b. The tung-ten filament lamp (TL) and halogen lamp (HL) have similarmission band, which basically cover the wavelength range of00–900 nm. The only difference was that HL has higher rel-

tive intensity at 800–900 nm. In contrast, light emission frometal halide lamp evenly distributes in the range of 400–700 nm

Fig. 4b). Effect of light emission patterns on photo-H2 produc-ion is to be discussed in the following sections.

ig. 3. Light intensity distribution on the surface of side-light optical fibersxcited by metal–halide and halogen light engines.

feieoHHwltl(vsoyetHtcati

logy of the side-light optical fiber before polishing, and (d) surface morphology

.3. Effect of single light source on photo-H2 production

Photobioreactors illuminated with different single lightources at a similar light intensity of ca. 95 W/m2 were examinedor their photo-H2 production performance and light conversionfficiency. Time-course profiles of H2 evolution are illustratedn Fig. 5a, indicating that optical fibers excited by halogen lightngine (OF-HL) exhibited the highest H2 production (Hmax)f 625 ml, followed by 381 and 277 ml obtained from usingL and TL, respectively (Fig. 6a and Table 1). In contrast,2 production was negligible without light source (dark) orith the illumination of optical fiber excited by metal–halide

amp (OF-MH). Kinetic analysis by modified Gompertz equa-ion (Eq. (1)) shows that using OF-HL and HL as the singleight source resulted in similar maximum H2 production rateRmax) of 7.28 and 7.54 ml/h, respectively (Table 1). The Rmalue for TL was slightly lower at 6.02 ml/h. The lag time wasimilar (λ =17–19 h) for OF-HL, HL, and TL (Table 1). More-ver, bioreactors with OF-HL appeared to achieve higher H2ield (1.19 mol H2/mol acetic acid) than the rest of light sourcesxamined (Table 1). Although OF-HL illumination resulted inhe highest Hmax and H2 yield, conventional lamps (especiallyL) gave higher or comparable overall volumetric H2 produc-

ion rate (νH2 ) and specific H2 production rate (νs,H2 ) (Fig. 6b and

) when compared with the OF-HL system. These results can bettributed to better cell growth under illumination by halogen orungsten filament lamps (Fig. 5b), which provided much higherrradiation area (276 cm2) than the optical fiber (138 cm2).

38 C.-Y. Chen et al. / Biochemical Engineering Journal 32 (2006) 33–42

Table 1Performance of photo-H2 production for photobioreactors illuminated with different single light sources

Light source Irradiationarea (cm2)

H2 yield (mol H2/molacetic acid)

Acetateconversion (%)

H2 content(%)

Specific growthrate (h−1)

Modelsimulationa

Hmax (ml) Rmax (ml/h) λ (h) r2

OF-HL 138 1.19 70.1 54 0.0191 625 7.28 17.72 0.999OF-MH 34.5 NA 51.1 0.3 0.0199 2 NA NA NATL 276 0.80 91.6 78 0.0374 277 6.02 17.33 0.976HL 276 1.08 92.6 81 0.0416 381 7.54 19.29 0.981D 0.8

N equat

Hdasfs

Flm

ccl

ark 0 0.002 12.34

A, not applied (since the data cannot be simulated by the modified Gompertza Simulation of time-course data by modified Gompertz equation.

R. palustris strains are known to be capable of producing2 via dark fermentation [29,30]. However, the poor H2 pro-uction from our dark experiments is expectable due to using

cetate as the sole carbon source. Acetate is apparently not auitable carbon source for R. palustris WP3–5 to undergo darkermentation, which in general prefers more complicate carbonources, such as carbohydrates [30]. This is confirmed by poor

ig. 4. (a) Absorption spectrum of intact cells of R. palustris WP3–5 and (b)ight emission spectrum of tungsten filament lamp (�), halogen lamp (- - -), and

etal halide lamp (· · ·).

msifi

Ff

0.0098 0.5 NA NA NA

ion due to negligible H2 production).

ell growth and low substrate (acetate) conversion in the darkultures when compared with those obtained in the presence ofight sources (Fig. 5b and Table 1).

Unlike efficient H2 production with OF-HL, OF-MH illu-

ination did not result in photo-H2 production despite a con-

iderable cell growth (Table 1 and Fig. 5b). This raises themportance of selecting a proper light engine for the opticalber, because the difference in light emission spectrum between

ig. 5. Time-course profiles of (a) cumulative H2 production and (b) cell growthor photobioreactors using different single light sources.

C.-Y. Chen et al. / Biochemical Engineering Journal 32 (2006) 33–42 39

Fr

mrf7to[sAwtsasf(p

9tilto

Ff

al

3l

tseactascp

ig. 6. The effect of different single light sources on (a) the overall H2 productionate, (b) the specific H2 production rate, and (c) the light efficiency.

etal–halide lamp and halogen lamp seemed to play a majorole in the efficiency of photo-H2 production. The emission bandrom metal–halide lamp uniformly distributes between 400 and00 nm but lacks the light at 800–900 nm (Fig. 4b). However,he purple nonsulfur bacteria absorb the light at wavelengthsf 522 and 860 nm (Fig. 4a) to stimulate photo-H2 production11]. In particular, the light at 800–900 nm is required for ATPynthesis in purple nonsulfur bacteria and thus is essential toTP-consuming photo-H2 production. This seems to explainhy H2 production was inefficient when OF-MH was used as

he sole light source. As the metal–halide lamp was able toupport cell growth satisfactorily (Table 1 and Fig. 5b), thebsorption band of R. palustris cells in the range of 400–600 nmeems to correlate closely with cell growth. As the light emissionrom halogen lamp covers the wavelength range of 350–900 nmFig. 4b), OF-HL (or HL) illumination resulted in much betterhoto-H2 production than OF-MH (Figs. 5 and 6).

It is also noticed that with a similar light intensity (about5 W/m2), HL illumination attained superior photo-H2 produc-ion performance over TL (Table 1 and Figs. 5 and 6). Possible

nterpretations for this result include the fact that HL is a face-ight source, thereby offering more uniform light irradiation forhe photobioreactor than TL, which is a spot-light source. More-ver, higher relative intensity of HL at 800–900 nm (Fig. 4b) may

iecp

ig. 7. Time-course profiles of (a) cumulative H2 production and (b) cell growthor photobioreactors using binary combinations of different light sources.

lso contribute to better photo-H2 production efficiency with theight source.

.4. Photo-H2 production using binary combinations ofight sources

Investigation of single-light-source systems shows that theype (optical fibers or external lamps), irradiation area, and emis-ion spectrum of light sources are the key factors affecting thefficiency of photo-H2 production. To further reveal interactivend complementary relations between the light sources, binaryombinations of different light sources were applied to the pho-obioreactors for H2 production. Comparison of single- (Table 1nd Figs. 5 and 6) and binary- (Table 2 and Figs. 7 and 8) light-ource systems shows that with a similar total light intensity,ombination of two light sources exhibited better H2 productionerformance than that of single light sources. This due primar-

ly to the fact that combining two light sources with differentmission spectra could provide a wider light emission range toover the light energy needed for both cell growth and photo-H2roduction.

40 C.-Y. Chen et al. / Biochemical Engineering Journal 32 (2006) 33–42

Table 2Performance of photo-H2 production for photobioreactors illuminated with binary combinations of light sources

Light source Irradiationarea (cm2)

H2 yield (mol H2/molacetic acid)

Acetateconversion (%)

H2 content(%)

Specific growthrate (h−1)

Modelsimulationa

Hmax (ml) Rmax (ml/h) λ (h) r2

OF-MH/HL 310 2.22 89.0 77 0.0351 855 14.4 16.0 0.998OF-MH/TL 310 2.43 75 78 0.0369 841 10.6 18.6 0.988OF-HL/HL 310 2.18 74 66 0.0301 712 10.1 13.6 0.999OF-HL/TL 310 2.64 75 72 0.0318 868 14.3 14.1 0.999OF-HL/OF-MHb 207 1.91 77 74 0.0260 944 9.2 12.7 0.998TL/HL 276 1.85 83 88 0.0316 667 11.2 18.9 0.999

ptica

tew(pltapH(fi(v(latpPeoep

scs

ν

H(epeop

3y

WohcU7lCo1Ha

TC

O

RRRRRRR

a Simulation of time-course data by modified Gompertz equation.b Contained two optical fibers excited by metal–halide light engine and four o

Table 2 shows that combination of optical fibers excited bywo different light engines (OF-HL/OF-MH) gave the high-st H2 production (Hmax) of 944 ml, while the lowest Hmaxas obtained from combination of two external lamps (HL/TL)

Fig. 8a and Table 2). Despite achieving the highest Hmax, theure optical fiber system (OF-HL/OF-MH) had significantlyower irradiation area than that of the rest of combinations,hereby obtaining a considerably lower growth rate (Fig. 7bnd Table 2). This results in a lower overall and specific H2roduction rates (Fig. 8b and c) of reactors illuminated by OF-L/OF-MH. In contrast, the highest νH2 (17.1 ml/h l) and νs,H2

9.47 ml/h g cell) was obtained from combination of tungstenlament lamp with optical fiber excited by halogen light enginei.e., OF-HL/TL) (Fig. 8b and c). Nevertheless, the best light con-ersion efficiency still belonged to the pure optical fiber systemOF-HL/OF-MH) (Fig. 8d). The H2 production performance andight efficiency were quite similar for OF-MH/HL, OF-MH/TLnd OF-HL/TL, whereas the performance of OF-HL/HL sys-em was relatively poor (Table 2 and Fig. 8), probably due toroviding only one type of light emission spectrum (i.e., HL).hotobioreactors with binary combinations of optical fibers andxternal lamps (especially, the OF-HL/TL system) show higherr comparable H2 production rates than those obtained from rel-vant studies (Table 3) using organic acids as the substrate forhoto-H2 production.

It is also worth noting that the binary system of OF-HL/HLtill gave much higher νH2 (13.3 ml/l h) and νs,H2 (6.34 ml/h gell) than those obtained from the individual single-light-sourceystems (i.e., HL and OF-HL). HL illumination only gave a

av

C

able 3omparison of photo-H2 production performance obtained from the present work an

rganism Substrate H2 yielda (%)

. palustris WP3–5 Organic acids 2.7–49.6

. palustris P4 Acetate 60–70

. palustris Acetate 14.8

. palustris DSM 131 Benzoate 62hodopseudomonas capsulata Acetate 32.6hodopseudomonas capsulate NCIB8254 Organic acid 66. palustris WP3–5 Acetate 48–66

a Percentage of H2 yield (mol H2/mol substrate) out of the theoretical H2 yield; NA

l fibers excited by halogen light engine.

H2 of 8.68 ml/l h and a νs,H2 of 3.01 ml/h g cell, while OF-L attained a νH2 of 6.83 ml/l h and a νs,H2 of 2.53 ml/h g

Fig. 6). Moreover, the photobioreactor illuminated only byxternal light sources (HL/TL) was less efficient in photo-H2roduction than all binary combinations of optical fibers andxternal lamps (Fig. 8), indicating the advantage and importancef using optical fibers as the internal light source to enhance H2roduction.

.5. Biogas composition, substrate conversion, and H2

ield

The gas products from anaerobic cultures of R. palustrisP3–5 consisted primarily of H2 (54–88%) with the remainder

f CO2 (Tables 1 and 2). The simple biogas composition andigh H2 content allow easier H2 separation in downstream pro-essing to obtain purified H2 products for fuel cell applications.sing a single light source, the acetate conversion ranged from0% to 90%, while the conversion was 74–89% when binaryight sources were used (Tables 1 and 2). This suggests a highOD removal efficiency during photo-H2 fermentation. On thether hand, the H2 yield obtained from single light source was.19, 0.8, and 1.08 mol H2/mol acetic acid for OF-HL, TL, andL, respectively (Table 1). Bioreactors with binary light sources

ttained higher H2 yields, ranging from 1.9 to 2.6 mol H2/mol

cetic acid (Table 2), which were 48–66% of the theoreticalalue of 4.0 mol H2/mol acetic acid derived from Eq. (6) [7,12]:

H3COOH + 2H2O → 4H2 + 2CO2 (6)

d from comparable studies

Overall H2

production rate(ml/l h)

Specific H2

production rate(ml/g h)

Light conversionefficiency (%)

Reference

NA NA NA [5]NA 9.8 NA [30]2.2 NA 0.1 [12]NA 3.55 NA [31]NA 19.07 4.2 [32]NA 3–16.5 NA [33]17.1 9.47 1.8 This study

, not available.

ngine

TltH

3

vtttFH1eilsM

Foe

tlfistclmsb

4

fs

C.-Y. Chen et al. / Biochemical E

he H2 yields obtained from this work (especially from binary-ight-source systems) are higher than most of the ones reported inhe comparable studies (namely, 2.7–70% out of the theoretical

2 yield) (Table 3).

.6. Light conversion efficiency

In addition to H2 production rate and H2 yield, the light con-ersion efficiency is also an important variable used to evaluatehe biological photo-H2 production process [11]. In this study,he motivation of using optical fibers in photobioreactors waso enhance light conversion efficiency for photo-H2 production.rom the results of using single light sources (Fig. 6d), OF-L displayed the highest light conversion efficiency (η) of ca..80%, which is 2.5–3-fold higher than that obtained from thexternal lamps (HL and TL). Meanwhile, the η value of 1.80%

s also higher than most of the light efficiencies reported in theiterature for photo-H2 production (Table 3). For the binary lightource systems (Fig. 8d), using optical fibers alone (OF-HL/OF-

H) also had the best light conversion efficiency (1.42%), while

ig. 8. The effect of binary combinations of different light sources on (a) theverall H2 production rate, (b) the specific H2 production rate, and (c) the lightfficiency.

sfiHbspsoweittarto

A

NE7

R

ering Journal 32 (2006) 33–42 41

he η value dropped to 1.12% when combination of two externalamps (HL/TL) were used. For binary combinations of opticalbers with external lamps, the light conversion efficiency wasimilar (η = 1.0–1.3%) (Fig. 8d). These results clearly indicatehat optical fibers markedly increased the light conversion effi-iency for photo-H2 production by R. palustris WP3–5. The highight conversion efficiency derived from using optical fibers is

ost likely due to the close contact between cells and the lightource as well as due to the uniform light distribution providedy optical fibers.

. Conclusions

Optical fibers were successfully implemented in bioreactorsor photo-H2 production. Using optical fibers as internal lightources markedly increased total H2 production and light conver-ion efficiency of the photobioreactors. However, using opticalbers alone resulted in lower cell growth rate, leading to lower2 production rates. This limitation may be resolved by com-ination of optical fibers with external lamps. The binary-light-ource systems appeared to further enhance the performance ofhoto-H2 production. Selecting light sources with proper emis-ion spectrum patterns was shown to be critical to the efficiencyf photo-H2 production. Combinations of halogen lamp (HL)ith tungsten filament lamp (TL) or metal halide lamp (MH)

nabled efficient H2 production, while using MH alone resultedn negligible H2 production. The best binary combination washe OF-HL/TL system, giving a H2 yield, overall H2 produc-ion rate, and specific H2 production rate of 2.64 mol H2/molcetic acid, 17.06 ml/h l, and 9.47 ml/h g cell, respectively. Theesults show that the photo-H2 production system developed inhis study displayed excellent H2 production efficiency in termsf H2 yield, H2 production rate, and light conversion efficiency.

cknowledgements

The authors gratefully acknowledge financial supports fromational Science Council of Taiwan (grant no. NSC-93-2211--006-040) and Bureau of Energy of Taiwan (grant nos. 93-ET--006-001-ET and 94-ET-7-006-004-ET).

eferences

[1] J. Miyake, M. Miyake, Y. Asada, Biotechnological hydrogen production:research for efficient light energy conversion, J. Biotechnol. 70 (1999)89–101.

[2] T. Kondo, M. Arakawa, T. Hirai, T. Wakayama, M. Hara, J. Miyake,Enhancement of hydrogen production by a photosynthetic bacteriummutant with reduced pigment, J. Biosci. Bioeng. 93 (2002) 145–150.

[3] D. Das, T.N. Veziroglu, Hydrogen production by biological processes: asurvey of literature, Int. J. Hydrogen Energy 26 (2001) 13–28.

[4] H. Koku, I. Eroglu, U. Gunduz, M. Yucel, L. Turker, Aspects of themetabolism of hydrogen production by Rhodobacter sphaeroides, Int. J.Hydrogen Energy 27 (2002) 1315–1329.

[5] C.M. Lee, P.C. Chen, C.C. Wang, Y.C. Tung, Photohydrogen productionusing purple nonsulfur bacteria with hydrogen fermentation reactor efflu-ent, Int. J. Hydrogen Energy 27 (2002) 1309–1313.

[6] E. Fascetti, E. D’addario, O. Todini, A. Robertiello, Photosynthetic hydro-gen evolution with volatile organic acids derived from the fermentation of

4 ngine

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

2 C.-Y. Chen et al. / Biochemical E

source selected municipal solid wastes, Int. J. Hydrogen Energy 23 (1998)753–760.

[7] J. Miyake, X.Y. Mao, S. Kawamura, Photoproduction of hydrogen fromglucose by a co-culture of a photosynethic bacterium and Clostridiumbutyricum, J. Ferment. Technol. 62 (1984) 531–535.

[8] H. Takabatake, K. Suzuki, I.B. Ko, T. Noike, Characteristics of anaerobicammonia removal by a mixed culture of hydrogen producing photosyntheticbacteria, Bioresour. Technol. 95 (2004) 151–158.

[9] N. Kataoka, A. Miya, K. Kiriyama, Studies on hydrogen production by con-tinuous culture system of hydrogen-producing anaerobic bacteria, WaterSci. Technol. 36 (1997) 41–47.

10] J. Miyake, N. Tomizuk, A. Kamibayashi, Prolonged photo-hydrogenproduction by Rhodospirillum rubrum, J. Ferment. Technol. 60 (1982)199–203.

11] I. Akkerman, M. Janssen, J. Rocha, R.H. Wij1ffels, Photobiological hydro-gen production: photochemical efficiency and bioreactor design, Int. J.Hydrogen Energy 27 (2002) 1195–1208.

12] M.J. Barbosa, J.M.S. Rocha, J. Tramper, R.H. Wijffels, Acetate as a carbonsource for hydrogen production by photosynthetic bacteria, J. Biotechnol.85 (2001) 25–33.

13] A.A. Tsygankov, T.V. Laurinavichene, I.N. Gogotov, Laboratory scale pho-tobioreactor, Biotechnol. Tech. 8 (1994) 575–578.

14] T. Kondo, M. Arakawa, T. Wakayama, J. Miyake, Hydrogen production bycombining two types of photosynthetic bacteria with different characteris-tics, Int. J. Hydrogen Energy 27 (2002) 1303–1308.

15] B.W. Kim, K.P. Chang, H.N. Chang, Effect of light source on the micro-biological desulfurization in a photobioreactor, Bioprocess Eng. 17 (1997)343–348.

16] E. Nakada, Y. Asada, T. Arai, J. Miyake, Light penetration into cell sus-pensions of photosynthetic bacteria and relation to hydrogen production,J. Ferment. Bioeng. 80 (1995) 53–59.

17] E. Fascetti, O. Todini, Rhodobacter sphaeroides RV cultivation and hydro-gen production in a one- and two-stage chemostat, Appl. Microbiol.Biotechnol. 44 (1995) 300–304.

18] A. Yamada, H. Takano, J.G. Burgess, T. Matsunaga, Enhanced hydrogen

production by a marine photosynthetic bacterium Rhodobacter marinusimmobilized onto light diffusing optical fiber, J. Mar. Biotechnol. 4 (1996)23–27.

19] K.H. Lee, B.W. Kim, Enhanced microbial removal of H2S using chlorobiumin an optical fiber bioreactor, Biotechnol. Lett. 20 (1998) 525–529.

[

ering Journal 32 (2006) 33–42

20] J.Y. An, B.W. Kim, Biological desulfurization in an optical-fiber photo-bioreactor using an automatic sunlight collection system, J. Biotechnol. 80(2000) 35–44.

21] K. Hofstadler, R. Bauer, S. Novallc, G. Helsler, New reactor design forphotocatalytic wastewater treatment with TiO2 immobilized on fused silicaglass fibers: photomineralization of 4-chlorophenol, Environ. Sci. Technol.28 (1994) 670–674.

22] P.F. Henshaw, W. Zeu, Biological conversion of hydrogen sulphide to ele-mental sulphur in a fixed-film continuous flow photo-reactor, Water Res.35 (2001) 3605–3610.

23] L. Mignot, G.A. Junter, M. Labbe, A new type of immobilized cell pho-toreactor with internal illumination by optical fiber, Biotechnol. Tech. 3(1989) 98–107.

24] S.V. Ginkel, S. Sung, J.J. Lay, Biohydrogen production as a function of pHand substrate concentration, Environ. Sci. Technol. 35 (2001) 4726–4730.

25] C.Y. Lin, C.H. Lay, Effects of carbonate and phosphate concentrationson hydrogen production using anaerobic sewage sludge microflora, Int. J.Hydrogen Energy 29 (2004) 275–281.

26] H.G. Truper, N.W. Pfennig, Characterization and identification of theanoxygenix phototrophic bacteria, Prokaryotes 1 (2000) 299–312.

27] J. Miyake, X.Y. Mao, S. Kawamura, Efficiency of light energy conversionto hydrogen by the photosynethic bacterium Rhodobacter sphaeroides, Int.J. Hydrogen Energy 12 (1987) 147–149.

28] A. Vermeglio, P. Joliot, The photosynthetic apparatus of Rhodobactersphaeroides—review, Trends microbiol. 7 (1999) 435–440.

29] E.N. Kondratieva, Phototrophic micro-organisms as source of hydrogenand hydrogenase formation, in: H.G. Schlegel, J. Barnea (Eds.), MicrobialEnergy Conversion, Erich Goltze, KG, Gottingen, 1976, pp. 205–216.

30] Y.K. Oh, E.H. Seol, M.S. Kim, S. Park, Photoproduction of hydrogen fromacetate by a chemoheterotrophic bacterium Rhodopseudomonas palustrisP4, Int. J. Hydrogen Energy 29 (2004) 1115–1121.

31] J. Fissler, G.W. Kohring, F. Giffhorn, Enhanced hydrogen production fromaromatic acids by immobilized cells of Rhodopseudomonas palustris, Appl.Microbiol. Biotechnol. 44 (1995) 43–46.

32] X.Y. Shi, H.Q. Yu, Response surface analysis on the effect of cell concen-

tration and light intensity on hydrogen production by Rhodopseudomonascapsulate, Process Biochem. 40 (2005) 2475–2481.

33] L. Segers, W. Verstraete, Conversion of organic acids to H2 by Rhodospiril-laceae grown with glutamate or dinitrogen as nitrogen source, Biotechnol.Bioeng. 25 (1983) 2843–2853.