International Journal of Hydrogen Energy 27 (2002) 1331 – 1338
www.elsevier.com/locate/ijhydene
A pneumatically agitated &at-panel photobioreactor with gas
re-circulation: anaerobic photoheterotrophic cultivation of a
purple non-sulfur bacterium
Sebastiaan Hoekema ∗ , Martijn Bijmans, Marcel Janssen, Johannes Tramper,
Ren1e H. Wij3els
Food and Bioprocess Engineering Group, Department of Agrotechnology and Food Sciences, Wageningen University, P.O. Box 8129,
6700 EV Wageningen, The Netherlands
Abstract
The application of hydrogen as a clean and e4cient energy carrier in the near future becomes more and more evident. Within
the process of photobiological hydrogen production, purple non-sulfur bacteria are an interesting subject of study because
of their high hydrogen producing capacity. In a previous study, the used Rhodopseudomonas sp. had proven to e4ciently
produce hydrogen from acetic acid and light energy. We constructed a pneumatically agitated &at-panel photobioreactor as
a model system for optimization of photoheterotrophic hydrogen production. Batch experiments and a chemostat experiment
were performed to investigate the proper functioning of the new photobioreactor. During the 8rst experiments, argon gas was
sparged through the system for mixing and inhibition of growth was observed. Experimental results indicate that the stripping
of carbon dioxide from the culture liquid caused this inhibition of growth. Possibly, the Rhodopseudomonas sp. used requires
carbon dioxide during growth on a highly reduced substrate like acetate. Recirculating the gas prevented the carbon dioxide
from being stripped from the system. In this mode of operation, growth was supported.
? 2002 International Association for Hydrogen Energy. Published by Elsevier Science Ltd. All rights reserved.
Keywords: Hydrogen; Organic acids; Purple non-sulfur bacteria; Photobioreactors; Pneumatic agitation
1. Introduction
The worldwide energy requirement is growing exponentially, the reserves of fossil fuels are decreasing rapidly and
the combustion of fossil fuels has serious negative e3ects
on the environment. For this reason, much research is aimed
at the exploration of new and sustainable energy production systems that could substitute energy production based
on fossil fuels.
Hydrogen is a clean and e4cient fuel and is de8nitely a
potential substitute. Most hydrogen used today is produced
∗ Corresponding
author. Tel.: +31-317-483396; fax:
+31-317-482237.
E-mail address: [email protected]
(S. Hoekema).
in physical–chemical processes (steam reforming of natural gas) or electrochemical processes (electrolysis of
water). These processes are energy intensive and they are
not sustainable because there is only a limited reserve of
fossil fuels.
Biomass is a potential renewable feedstock for hydrogen production either by physical=chemical treatment,
e.g. gasi8cation followed by steam reforming of the
produced syngas, or by conversion in a biological system. Thermal gasi8cation is best applicable for large-scale
hydrogen production from dry biomass. Wet biomass substrates are favorable for biological systems for hydrogen
production. Biological hydrogen production could play an
important role in developing a renewable hydrogen industry
[1].
Anoxygenic photosynthetic bacteria (purple non-sulfur
bacteria) can produce hydrogen from simple organic
0360-3199/02/$ 22.00 ? 2002 International Association for Hydrogen Energy. Published by Elsevier Science Ltd. All rights reserved.
PII: S 0 3 6 0 - 3 1 9 9 ( 0 2 ) 0 0 1 0 6 - 4
1332
S. Hoekema et al. / International Journal of Hydrogen Energy 27 (2002) 1331 – 1338
molecules like organic acids or alcoholic compounds. Light
energy (sunlight) is used to provide the energy needed for
this thermodynamically unfavorable conversion. The combination of these bacteria and other bacteria (facultative
anaerobes, obligate anaerobes or even thermophiles and aerobes) in a two-step system could provide a system that can
e4ciently produce hydrogen from carbohydrates present in
(waste) biomass. In the 8rst step of such a process bacteria
convert the carbohydrates to organic acids like acetate. In
the second, photoheterotrophic step the organic acids are
converted to hydrogen [2,3].
Examples of full-scale application of biological systems
for the production of hydrogen are not presented in literature, but research on lab-scale is well documented [4,5].
For the successful scale-up of the second step, the e4ciency
at which (solar) light energy is directed to biomass growth
and hydrogen production is the most important optimization
parameter.
In the 8eld of algal biotechnology, research has been focused on optimizing the photosynthetic e4ciency (PE) of
photobioreactors. The PE of phototrophic growth indicates
the energy stored as a product per unit of photosynthetically
active radiation (PAR, 400 –700 nm) absorbed. The PE was
found to be highest in systems with a high surface to volume ratio, thereby minimizing the e3ect of mutual shading. Shallow, plate-type photobioreactors operated at high
biomass densities and intense aeration proved to have a high
PE [6,7]. The PE on biomass production reached in this
kind of system was 20% and close to the theoretical maximum PE of algal photosynthesis of 27%, under ideal conditions [8]. This was demonstrated under outdoor solar irradiation for the cultivation of the cyanobacterium Spirulina
platensis [9].
Following this concept of e4cient light usage, we
developed a lab-scale &at-panel photobioreactor for
the anaerobic cultivation of purple non-sulfur bacteria and concomitant hydrogen production from organic acids. Richmond and co-workers [6] were able
to use air as the gas facilitating turbulent mixing.
For photoheterotrophic hydrogen production, argon
should be used because oxygen (air) and nitrogen gas
lower nitrogenase-based hydrogen production. The continuous throughput of fresh argon gas is expensive
and the hydrogen produced will be diluted strongly.
These e3ects are undesirable and applying a closed gas
re-circulation system could provide a solution for this
problem.
In this study a lab-scale &at-panel photobioreactor design is described. It was tested for its applicability for the
cultivation of the purple non-sulfur bacterium Rhodopseudomonas sp. The experiments were performed using continuous re-circulation of argon gas. It was demonstrated
that the system works well and that a steady state could be
attained. In addition, it was shown that gas re-circulation
might even be essential to maintain optimal growth
conditions.
2. Materials and methods
2.1. Bacterial strain and medium composition
JoAnn Radway of the University of Hawaii kindly provided a Rhodopseudomonas sp. HCC 2037 culture. In a
previous study, this bacterium was selected out of three
promising photosynthetic bacterial strains as the most
e4cient hydrogen producer from acetate [2]. The culture
was maintained in the so-called ‘SyA’ medium under a
nitrogen headspace. The composition of the SyA medium
is given in Table 1. The cultures were illuminated with
90 mol m−2 s−1 of PAR in a day=night cycle of 16=8 h at
◦
25 C.
During the batch experiments the so-called ‘AA-b’
medium was used. During the chemostat experiment the
culture was diluted with the so-called ‘AA-c’ medium.
The composition of the media is given in Table 1. The
media were prepared from deionized water and concentrated stock solutions and autoclaved prior to use. Calcium chloride and magnesium sulfate were autoclaved
together, but separate from the other components to prevent
calcium- and magnesium phosphate depositions. The concentrated vitamin solution used was 8lter sterilized prior to
use.
Initially AA-b1 medium was used, which is similar to the
SyA medium except for the concentrations of the carbonand nitrogen source and no vitamins or yeast extract were
added. After some experiments it was found that the gas
spargers clogged. This was probably due to calcium- and
magnesium phosphate precipitations. In further experiments,
the concentrations of phosphate bu3er and macro nutrients
(calcium chloride and magnesium sulfate) were therefore
reduced to the levels indicated under AA-b2 medium in
Table 1.
The concentrations of elements needed to yield a certain
dry weight biomass concentration (Cx ) were calculated
according to the elemental composition of our bacterium
(data not shown). This composition was CH1:76 O0:38 N0:14 .
The phosphorous content was taken to be 1.61% as reported
by Tsygankov and Laurinavichene [10] for Rhodobacter capsulatus. The sulfur content of the bacteria was
assumed to be 0:0045 mol mol C−1 . Taking this into account, the elemental composition of the biomass yielded
CH1:76 O0:38 N0:14 P0:01 S0:0045 .
On the basis of this elemental composition, the AA-b2
medium was designed to support a Cx of 1:7 g l−1 . Ammonium sulfate was added to assure a high enough
sulfur content. The concentrations of magnesium and
calcium were assumed to be su4cient to support
this Cx although no references were found in literature to support this. The composition of the ‘AA-c’
medium that was used for chemostat cultivation is also
shown in Table 1. It was designed to support a Cx of
4 g l−1 .
S. Hoekema et al. / International Journal of Hydrogen Energy 27 (2002) 1331 – 1338
1333
Table 1
Composition of the di3erent media used
Component class
Component
Concentration (mg=l)
SyA medium
Vitamins
Biotin
Thiamin
p-amino benzoic acid
Vitamin b12
Nicotinamine
Phosphate bu3er
KH2 PO4
K2 HPO4
Macronutrients
MgSO4 · 7H2 O
CaCl2 · 2H2 O
Micronutrients
Na2 EDTA · H2 O
FeSO4 · 7H2 O
H3 BO3
MnSO4 · H2 O
Na2 SO4 · 7H2 O
ZnSO4 · 7H2 O
Cu(NO3 )2 · 3H2 O
C and N sources
Na-succinate
Yeast extract
Na-acetate (mM)
NH4 Cl (mM)
(NH4 )2 SO4 (mM)
AA-b1 medium
AA-b2 medium
—
—
—
—
—
—
—
—
—
—
1732
1466
1732
1466
108.3
91.6
433
366
200
99
200
99
12.5
6.2
50
25
20
11.8
2.8
2.77
0.75
0.24
0.04
20
11.8
2.8
2.77
0.75
0.24
0.04
—
—
40
9.3
0.39
—
100
105
24.3
1.11
1
1
1
1
1
20
11.8
2.8
2.77
0.75
0.24
0.04
8100
1000
20
—
—
20
11.8
2.8
2.77
0.75
0.24
0.04
—
—
40
9.3
—
AA-c medium
1
1
1
1
1
2.2. Acetic acid
2.4. Hydrogen
The acetic acid concentration in the culture medium
was determined by gas chromatography. Samples were
centrifuged and the supernatant was diluted 1:1 with a 3%
(v=v) formic acid solution. Then the samples were stored
◦
at −80 C. The samples were analyzed on a HP 5890
gas chromatograph equipped with a glass packed column
(length 2 m, internal diameter 2 mm, 10% &uorad 431 on
supelco-port, 100 –120 mesh) and a &ame ionization detec◦
tor (FLD). The column and the FID were kept at 130 C and
◦
280 C, respectively. Nitrogen saturated with formic acid
was used as the carrier gas at a &ow rate of 40 ml min−1 .
The volume fraction of hydrogen in the produced gas
was determined using a Chrompack CP9000 series gas chromatograph equipped with a packed column (length 1:8 m,
internal diameter 0:25 in, molsieve 13X, 60 –80 mesh) and
a thermal conductivity detector (TCD). The column and the
◦
◦
TCD were kept at 100 C and 120 C, respectively. The carrier gas was argon at a &ow rate of 20 ml min−1 .
2.3. Ammonium
Samples for ammonium determinations were centrifuged
◦
and the supernatant was stored at −80 C. The ammonium
concentration in the samples was determined using Nessler’s
+
reagent. Nessler’s reagent (HgI2−
4 ) together with NH4 forms
the yellow complex NH4 HgI4 ; potassium sodium tartrate
was added to keep the salts dissolved that might interfere
with the determination. The absorbance was read at 440 nm
on a Spectronic 20 Genesys spectrophotometer and compared to the absorbance of standard solutions.
2.5. Light measurement
The PAR light intensity was measured using a LI SA-190
quantum sensor combined with a LI-250 read-out unit
(Li-Cor, USA). This device measures a light intensity expressed in mol PAR m−2 s−1 . The spectrum of a 500 W
tungsten-halogen lamp used during the experiments (Philips
Halotone R7s, 8tted in a Philips QVF 415n re&ector) was
acquired using an AVS-S2000 8ber-optic spectrometer
(Avantes, USA) in W cm−2 from 400 to 950 nm with a
step-size of 0:3 nm. A measured mol PAR m−2 s−1 value
could thus be converted to a W m−2 (400 –950 nm) value.
The conversion factor was 0.539. All light intensities were
expressed in W m−2 in the 400 –950 nm range, except for
the light intensity in the cabinet in which the cultures and
the small batch experiments were incubated. These were
1334
S. Hoekema et al. / International Journal of Hydrogen Energy 27 (2002) 1331 – 1338
illuminated using &uorescent tubes, for which no spectrum
was measured. This value has the unit mol PAR m−2 s−1 .
2.6. Other analyses
Measuring the optical density of the cultures at 660 nm
(OD660 ) monitored bacterial growth with a Spectronic 20
Genesys spectrophotometer. The cell dry weight was determined by centrifuging 50 ml of cell suspension (7000g
for 10 min), washing the pellet with deionized water, centrifuging again, resuspending in a small volume of deion◦
ized water and drying at 103 C until constant weight. One
OD660 -unit was found to be equal to a Cx of 0:7 g l−1 .
2.7. Chemicals
All chemicals used were reagent grade and produced by
Merck, Darmstadt, Germany.
2.8. Flat-panel photobioreactor
The new design of the &at-panel photobioreactor (PRB)
is depicted in Fig. 1. It consisted of a stainless-steel frame
and three polycarbonate panels. The reactor was composed
of two compartments located behind each other. The front
compartment contained the bacterial culture (3 cm deep).
The culture volume was 2:4 l. Through the hind compartment (2 cm deep) water was circulated via a temperature
controlled water bath in order to maintain the temperature
◦
of the culture at 30 C. Two 500 W tungsten-halogen lamps
(Philips Halotone R7s, 8tted in Philips QVF 415n re&ectors) were placed on one side of the reactor. The lamps were
mounted above each other in a frame and placed on 75 cm
distance from the reactor. The average light intensity at the
reactor surface was 175 W m−2 .
A membrane gas pump circulated the gas through the
spargers (hypodermic needles) at the bottom of the reactor. The produced gas was collected in a gasbag. Two 1-l
pressure vessels prevented pressure &uctuations in the gas
re-circulation system. A pressure valve maintained a constant input pressure to the mass &ow controller. A condenser
prevented water vapor from entering the gas re-circulation
system.
The reactor was autoclaved completely prior to all runs.
The culture medium was autoclaved separately and fed to
the reactor after. Samples for determination of the Cx and
medium component concentrations were taken from the reactor using a sample port, attached to the out&ow tube. Bacterial growth was monitored on-line. On the left-hand side
of the reactor a small tube was attached to the reactor (Fig.
1). The bacterial suspension &owed through this tube as a
result of an airlift e3ect. Within a few seconds the culture
passed the tube, ensuring the representativeness of the sample. A red light emitting diode (LED) peaking at 665 nm
was used as the light source on one side of the tube. On
the other side a homemade PAR light sensor registered the
Fig. 1. Schematical drawing of the photobioreactor setup: (1) membrane gas pump; (2) gasbag for collection of produced gas; (3) two
1-l pressure vessels; (4) pressure valve; (5) mass &ow controller;
(6) condenser; and (7) pH=redox electrode.
remaining PAR light intensity as a mV signal. This signal
was translated to a value similar to the absorbance measured
in a spectrophotometer according to the following of equation:
ABS = −log
I
;
I0
(1)
ABS is the absorbance dimensionless, I the light-induced
signal after passage through the tube (mV), and I0 the reference light-induced signal with only medium without bacteria in the tube (mV).
The validity of the method was checked by correlating the
daily OD660 values of a complete growth experiment in the
reactor to the corresponding absorbance values obtained using Eq. (1). A straight line with an R2 of 0.99 was obtained.
The liquid level was controlled using another homemade
light sensor, similar as described previously. The sensor was
placed behind the reactor at the desired level and the eWuent
pump was controlled on the basis of the signal registered by
this light sensor.
S. Hoekema et al. / International Journal of Hydrogen Energy 27 (2002) 1331 – 1338
2.9. Serum ?ask experiments
In an experiment to assess the in&uence of the ammonium concentration on growth of Rhodopseudomonas sp.
8 closed 100 ml serum &asks were used. These were 8lled
with 50 ml of medium and autoclaved. The &asks were inoculated under an argon headspace. The &asks were stirred
continuously using magnetic stirrers and illuminated with
90 mol PAR m−2 s−1 in a day=night cycle of 16=8 h at
◦
25 C.
2.10. Bubble column experiments
Other experiments were done in small 300 ml glass bubble columns to investigate the e3ect of shear forces on
Rhodopseudomonas sp. The gas was sparged through perforated plates in the bottom of the reactors. The reactors
were equipped with a water jacket for the re-circulation of
cooling water via a temperature controlled water bath. The
◦
temperature was controlled at 30 C. The reactors were autoclaved empty and 8lled with sterile medium afterwards.
Two 300 W tungsten-halogen lamps (Philips Halotone R7s
8tted in Philips QVF 415n re&ectors) were placed on one
side of the reactor. The lamps were mounted above each
other and placed on 30 cm distance from the reactor in
order to yield an average light intensity of 135 W m−2 at
the reactor surface.
3. Results and discussion
3.1. Operation of ?at-panel photobioreactor with
continuous gassing of argon
During this 8rst experiment using the &at-panel photobioreactor AA-b2 medium was used, containing 40 mM
acetate and 9:3 mM ammonium. The reactor was gassed
continuously with argon at a &ow rate of 0:83 l l−1 min−1 .
Argon was used as the agitation gas because oxygen and
nitrogen negatively a3ect the hydrogen evolving capacity
of the bacterial nitrogenase enzyme.
2.0
0.4
0.3
1.5
3
2
0.2
1.0
0.1
0.5
OD660 [-]
Absorbance [-]
A pH electrode and a redox electrode were 8tted into
the reactor for pH control and monitoring purposes, respectively. The pH was controlled at 6.8–7.0 in all experiments
by dosing 0:5 M HCl. During the chemostat experiment the
redox potential was monitored to check for anaerobicity.
Since the redox couple O2 =H2 O has a potential of +820 mV,
a sudden increase in redox potential would indicate oxygen leakage into the system. The redox potential was measured against a built-in Ag=AgCl reference electrode and
converted to a value with the standard hydrogen electrode
as the reference.
The set-up was computer controlled and most signals were
recorded using a data logger.
1335
1
0.0
0.0
0
2
4
6
8
Time [days]
Fig. 2. Light absorbance and OD660 values during all experiments in the &at-panel photobioreactor: (1) on-line light absorbance
measurement during the batch experiment with continuous argon
sparging; (2) on-line light absorbance measurement and o3-line
OD660 measurements () during the batch experiment with gas
re-circulation; (3) on-line light absorbance measurement and the
o3-line OD660 measurements ( ) during the chemostat experiment. The arrow indicates the point where the batch pre-culture
ended and chemostat operation was started.
The recorded on-line absorbance of the bacterial suspension is shown as line 1 in Fig. 2. This absorbance was
measured in a small tube attached to the reactor in which
the culture &owed. It can be seen clearly that the absorbance
of the culture does not change and apparently no growth
occurred.
It was unclear what was the reason for the absence
of growth. Possibly, ammonium inhibition played a role.
No data was available on possible growth inhibition of
Rhodopseudomonas sp. by ammonium. A set of serum
&ask experiments was performed to assess the e3ect of
ammonium on the growth and hydrogen evolution of
Rhodopseudomonas sp.
The results are presented in Fig. 3. Batch incubations were
performed using AA-b1 medium with 30 mM acetate and
0, 1, 2, 5, 10, 15 and 20 mM of ammonium, respectively.
At the end of the growth phase, the OD660 and the volume
fraction of H2 in the headspace were determined. Using this
volume fraction the total amount of hydrogen gas produced
was calculated. It can be seen clearly that the 8nal OD660
value increases with the increasing initial concentration of
ammonium up to 20 mM. A concentration of 9:3 mM of ammonium is therefore much too low to cause full inhibition of
growth. From Fig. 3, it is also clear that ammonium represses
hydrogen production signi8cantly above a concentration of
5 mM. It is well known from literature that ammonium represses the nitrogenase enzyme at millimolar concentrations
already [11–13].
Another explanation for the lack of growth in the
&at-panel photobioreactor continuously gassed with argon,
could be shear stress caused by gas bubbling. In order to
1336
S. Hoekema et al. / International Journal of Hydrogen Energy 27 (2002) 1331 – 1338
1.6
2.5
2.0
1.5
0.8
1.0
H2 [mmol]
OD660 [-]
1.2
0.4
0.5
0.0
0.0
0
5
10
15
20
+
NH [mM]
4
Fig. 3. The in&uence of the initial ammonium concentration on
the 8nal OD660 value () and the total amount of hydrogen gas
evolved ( ) in closed 100 ml serum &asks.
2.5
0.3
2.0
OD660 [-]
1.5
1.0
OD660 [-]
0.2
0.1
0.5
0.0
0.0
0
1
2
3
4
Time [days]
Fig. 4. The in&uence of the type of gas agitation and its &ow
rate on bacterial growth, measured as OD660 in 300 ml bubble
columns. No gas agitation (dimensionless) (not marked); nitrogen 6:66 l l−1 min−1 (); argon 6:66 l l−1 min−1 ( ); argon
1:66 l l−1 min−1 (♦); argon 0:33 l l−1 min−1 (∇).
investigate this further, 8ve experiments were performed
in 300 ml glass bubble columns, using nitrogen or argon
at various &ow rates. AA-b1 medium with 40 mM
of acetate and 9:3 mM of ammonium was used. Five
experiments were done: one experiment without any agitation, another experiment applying 6:66 l l−1 min−1 nitrogen, and three more experiments applying argon at 0.33,
1.66 and 6:66 l l−1 min−1 , respectively. Fig. 4 shows the
optical density (OD660 ) in time for all these experiments.
Only in the experiment in which no agitation was applied
an increase in the OD660 value was observed. The optical
density reached a value of 2.4; this is equal to a Cx of
1:7 g l−1 , which was the dry weight biomass concentration
the medium was designed for. The remaining incubations
did not show any increase in OD660 in time, indicating that
no growth had occurred. From these results, it is clear that
pneumatic agitation with nitrogen or argon inhibits bacte-
rial growth at 6:66 l l−1 min−1 and at any &ow rate ranging
from 0.33 to 6:66 l l−1 min−1 , respectively. The extremely
low &ow rate of 0:33 l l−1 min−1 already inhibited growth
completely, which made shear stress as an explanation for
the absence of growth improbable. Moreover, all bacteria
from the Rhodospirillaceae family are contained by a cell
wall [14] that o3ers protection against shear stress.
The absence of growth could also be related to stripping
of carbon dioxide from the culture medium by the continuous gas &ow. The carbon dioxide dependent growth of members of the Rhodospirillaceae family on acetate as the only
organic substrate can be explained in two ways.
Firstly, when phototrophic bacteria grow on highly
reduced substrates they must have a means for disposing of excess reducing equivalents, in order to retain
the redox balance between the substrates consumed and
their metabolic products [15]. The Calvin cycle enzymes can consume these reduction equivalents during
carbon dioxide 8xation [16]. Dependent on the discrepancy between the degree of reduction of the biomass
and the organic substrate carbon dioxide can be either produced or consumed [17]. The incorporation of
14
CO2 into cell material during growth of Rhodospirillum rubrum on a range of reduced carbon substrates
including acetate was observed. There was only net uptake of carbon dioxide during growth on propionate and
butyrate [18]. Moreover, it was shown that the activity
of ribulose bisphosphate carboxylase (the enzyme responsible for carbon dioxide 8xation in the Calvin cycle) of
Rhodobacter sphaeroides was strongly de-repressed during
growth on highly reduced substrates [19]. Although
the operation of the Calvin cycle may be needed to
a certain extent, it is energetically expensive and assimilation of a broad spectrum of reduced substrates
is apparently favored over large-scale carbon dioxide
8xation [15].
Secondly, a large group of phototrophic bacteria from the
Rhodospirillacae family lack the enzyme isocitrate lyase.
This enzyme is part of the glyoxylate cycle that replenishes
the pool of citric acid cycle intermediates. A large group of
these bacteria are able to grow on acetate as the sole organic
substrate and therefore need another route for the replenishment of used citric acid cycle intermediates. Ivanovskii et al.
[20] proposed an anaplerotic cycle of acetate assimilation in
which citramalate is an intermediate during glyoxylate formation. In this citramalate cycle carbon dioxide is used and
formed again and no net consumption takes place. However,
since propionyl-CoA carboxylase (the enzyme that couples
carbon dioxide to propionate) has a low a4nity for carbon dioxide, a certain carbon dioxide concentration in the
growth medium is needed to maintain a high rate of growth
and acetate assimilation [20].
Possibly most of the carbon dioxide produced was
removed in our pneumatically agitated systems quite ef8ciently due to the continuous stripping with argon gas.
This might have caused the absence of growth in this 8rst
3.2. Operation of the ?at-panel photobioreactor with gas
re-circulation
A batch experiment in the &at-panel photobioreactor system was performed using gas re-circulation. After inoculation, the complete system was &ushed with argon to create
anaerobic conditions and the gas re-circulation was switched
on at 0:83 l l−1 min−1 . AA-b2 medium was used, containing
40 mM acetate and 9:3 mM ammonium. In Fig. 2, the development of both the recorded on-line culture absorbance,
and the OD660 measured o3-line are indicated by number
2. During the course of the experiment the absorbance of
the bacterial suspension increased signi8cantly from 0.01 to
0.25. The o3-line OD660 measurements, also shown in Fig. 2,
show the same trend and increased from 0.15 to 1.2. It is
clear that growth occurred now and apparently our assumption of carbon dioxide depletion in the previous experiment
was correct because during this new experiment with gas
re-circulation growth was supported. The medium was designed to support 1:7 g l−1 of dry weight biomass and as
can be seen from Fig. 2 an OD660 of 1.2 was reached. This
value equals 0:85 g l−1 of dry weight biomass. It is not clear
why the Cx remained lower than anticipated, while acetate
and ammonium were still present in the medium (data not
shown).
After growth was demonstrated in batch culture, an experiment was performed to cultivate Rhodopseudomonas sp. in
chemostat. First, a batch pre-culture was performed similar
to the one described above. A small amount of yeast extract and vitamins (at concentrations indicated under AA-c
medium in Table 1) was added to the AA-b2 medium to
stimulate growth. After the batch pre-culture, the culture
was switched to a chemostat at a dilution rate of 0:035 h−1 .
This rate equals 50% of the maximum growth rate of our
Rhodopseudomonas sp. of 0:07 h−1 measured in batch experiments previously (data not published).
During chemostat operation the AA-c medium (Table 1) was used. The elemental composition of this
medium supports a Cx of 4 g l−1 . In Fig. 2, the development of both the culture absorbance recorded
on-line and the OD660 measured o3-line are indicated by number 3. As can be seen from Fig. 2,
the OD660 was 1.2 at the end of the batch pre-culture,
which equals a Cx of 0:84 g l−1 . This is identical to the
previous experiment. Apparently, the vitamins and yeast
extract added did not contribute to a higher Cx at the end of
the batch culture. After switching to chemostat operation
on day 4, a slight transient decrease in absorbance can be
observed (Fig. 2). Possibly the culture needed some adaptation time. After this the Cx rose again and became more
stable at an OD660 of 1.63, which equals a Cx of 1:14 g l−1 .
1337
80
20
60
15
40
10
20
5
0
+
NH [mM]
4
experiment with the &at-panel photobioreactor. It gave
an extra incentive to start with experiments using the gas
re-circulation system. In this mode of operation all the
carbon dioxide produced is retained in the system.
Acetate [mM]
S. Hoekema et al. / International Journal of Hydrogen Energy 27 (2002) 1331 – 1338
0
0
2
4
6
8
Time [days]
Fig. 5. Concentrations of acetate () and ammonium ( ) during
the chemostat experiment in the &at-panel photobioreactor (line 3
in Fig. 2).
Again it is not clear why the Cx remained lower than
anticipated. Possibly a limitation in the concentrations of
calcium, chloride or magnesium in the medium was the
cause for the lower than anticipated Cx . As described before,
the AA-b1 medium (Table 1) supported a Cx of 1:7 g l−1 .
Applying this medium in the &at-panel photobioreactor
during preliminary experiments resulted in clogging of the
hypodermic needles used as gas spargers in the bottom of
the reactor. We assume that salt deposits near the end of
the needles caused this clogging. These salt deposits were
probably formed because of water evaporation to the dry
air passing. Calcium- and magnesium phosphates dissolve
poorly into water and possibly the concentrations of these
salts exceeded the solubility at the tip of the spargers. In
order to prevent the clogging of the spargers, the calcium
chloride, magnesium sulfate and phosphate bu3er concentrations in the medium were lowered to the levels indicated
in Table 1 as AA-b2 medium. The medium content of phosphorous and sulfur were kept at the level needed to support
a Cx of 1:7 g l−1 . No references were found on the calcium, chloride or magnesium requirements of Rhodopseudomonas sp. and it might well be that the concentrations
of these elements were too low to support 1:7 g l−1 of dry
weight biomass. The same observations are valid for the
AA-c medium used during the chemostat experiment and
designed to support 4 g l−1 of dry weight biomass.
Another possibility is that the light intensity on the reactor surface, 175 W m−2 on average, did not support a Cx
exceeding 0:84 g l−1 .
Fig. 5 shows the concentrations of acetate and ammonium
during the entire experiment. It can be seen that the concentrations decrease fast during the batch pre-culture. After day
4, at the start of continuous chemostat dilution, the concentrations of both acetate and ammonium increase again due
to the fact that more of the substrates is introduced than can
be consumed by the culture. During the entire experiment
the ammonium concentration remains higher than 4 mM.
This is too high to facilitate hydrogen production, as can
1338
S. Hoekema et al. / International Journal of Hydrogen Energy 27 (2002) 1331 – 1338
be seen in Fig. 3. In Fig. 5, we can also see that the consumption of the two substrates follows the biomass composition well. The C=N-ratio in the consumption of substrates
equals 8.3 throughout the experiment. This corresponds well
with the elemental balance shown in Eq. (2), from which
a C=N-ratio of 8 can be calculated. The elemental biomass
composition was determined in previous experiments (data
not published).
[5]
[6]
[7]
3:96 CH3 COOH + NH+
4 → 7:143 CH1:76 O0:38 N0:14
+ 0:79 CO2 + 3:64 H2 O:
[4]
(2)
The redox potential remained around 150 mV during the
entire experiment, indicating that no air leaked into the system and that anaerobic conditions were maintained.
[8]
[9]
[10]
4. Conclusions
It was demonstrated that the newly developed photobioreactor with gas re-circulation functions properly. It was
possible to attain a steady chemostat culture of a photoheterotrophic bacterium under anaerobic conditions.
The observed absence of growth of Rhodopseudomonas
sp. during experiments with continuous sparging of argon
was probably caused by the lack of carbon dioxide, due to the
stripping action of the sparged gas. When gas re-circulation
was applied, growth was observed. Apparently, carbon dioxide is needed when the used Rhodopseudomonas sp. grows
on reduced carbon substrates like acetate. Indications in this
direction were also found in literature.
This new design of a &at-panel photobioreactor with pneumatic agitation opens up a future in which light energy can
be directed to hydrogen gas production with high e4ciency.
Acknowledgements
The Dutch government 8nancially supported this study.
The Ministries of Economic A3airs (EZ), Education, Culture and Science and Housing (OC & W), Spatial Planning and the Environment (VROM) supported the study via
the Economy, Ecology and Technology (EET) foundation
(project number EET-K 99116).
References
[1] Benemann J. Hydrogen biotechnology: progress and prospects
Nat Biotechnol 1996;14(9):1101–3.
[2] Barbosa MJ, Rocha JMS, Tramper J, Wij3els RH. Acetate as
a carbon source for hydrogen production by photosynthetic
bacteria. J Biotechnol 2001;85:25–33.
[3] Claassen PAM, van Lier JB, Lopez Contreras A M, van Niel
EWJ, Sijtsma L, Stams AJM, de Vries SS, Weusthuis RA.
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
Utilisation of biomass for the supply of energy carriers. Appl
Microbiol Biotechnol 1999;52:741–55.
Asada Y, Miyake J. Review: photobiological hydrogen
production J Biosci Bioeng 1999;88(1):16.
ZYurrer H, Bachofen R. Hydrogen production by the
photosynthetic bacterium Rhodospirillum rubrum. Appl
Environ Microbiol 1979;37(5):789–93.
Richmond A. Microalgal biotechnology at the turn of the
millennium: a personal view J Appl Phycol 2000;12:441–51.
Hu Q, Guterman H, Richmond A. A &at inclined
modular photobioreactor for outdoor mass cultivation of
photoautotrophs. Biotechnol Bioeng 1996;51(1):51–60.
Radmer R, Kok B. Photosynthesis: limited yields, unlimited
dreams Bioscience 1977;27:599–605.
Hu Q, Faiman D, Richmond A. Optimal tilt angles of enclosed
reactors for growing photoautotrophic microorganisms
outdoors. J Ferment Bioeng 1998;85(2):230–6.
Tsygankov AA, Laurinavichene TV. In&uence of the degree
and mode of light limitation on growth characteristics of
the Rhodobacter capsulatus continuous cultures. Biotechnol
Bioeng 1996;51:605–12.
Fedorov AS, Laurinavichene TV, Tsygankov AA. Factors
in&uencing nitrogenase switch-o3 by ammonium in the purple
bacterium Rhodobacter capsulatus. Microbiology-USSR
1999;68(4):379–86.
Yagi K, Maeda I, Idehara K, Miura Y, Akano T, Fukatu K,
Ikuta Y, Nakamura K. Removal of inhibition by ammonium
ion in nitrogenase dependent hydrogen evolution of a
marine photosynthetic bacterium, Rhodopseudomonas sp.
strain W-1S. Appl Biochem Biotechnol 1994;45=46:429–36.
Yakunin AF, Hallenbeck PC. Short-term regulation of
nitrogenase activity by NH+
4 in Rhodobacter capsulatus:
multiple in vivo nitrogenase responses to NH+
4 addition J
Bacteriol 1998;180(23):6392–5.
Weckesser J. Anoxygenic phototrophic bacteria: model
organisms for studies on cell wall macromolecules.
In: Blankenship RE, Madigan MT, Bauer CE, editors.
Anoxygenic phototrophic bacteria. Dordrecht: Kluwer, 1995.
p. 207–30.
Ferguson SJ, Jackson JB, McEwan AG. Anaerobic respiration
in the Rhodospirillaceae: characterisation of pathways and
evaluation of roles in redox balancing during photosynthesis
FEMS Microbiol Rev 1987;46:117–43.
Lascelles J. The formation of ribulose 1:5-difosfate
carboxylase by growing cultures of Athiorhodaceae. J Gen
Microbiol 1960;23:499–510.
Sojka GA. Metabolism of nonaromatic compounds. In:
Clayton K, Sistrom WR, editors. The photosynthetic bacteria.
New York: Plenum Press, 1978. p. 707–18.
Ormerod JG. The use of radioactive carbon dioxide in the
measurement of carbon dioxide 8xation in Rhodospirillum
rubrum. Biochem J 1956;64:373–80.
Gibson JL, Tabita FR. Di3erent molecular forms of Dribulose-1,5bisfosfate carboxylase from Rhodopseudomonas
sphaeroides. J Biol Chem 1977;252:943–9.
Ivanovskii RN, Krasil’nikova EN, Berg IA. The mechanism of acetate assimilation in the purple nonsulfur
bacterium Rhodospirillum rubrum lacking isocitrate lyase.
Microbiology-USSR 1997;66(6):621–6.