New York, 14853, USA. E-mail: de54@cornebDepartment of Biological and Environmenta
New York, 14853, USA
Cite this: RSC Adv., 2014, 4, 1460
Received 13th September 2013Accepted 13th November 2013
DOI: 10.1039/c3ra45087b
www.rsc.org/advances
1460 | RSC Adv., 2014, 4, 1460–1468
Hollow fibre membrane arrays for CO2 delivery inmicroalgae photobioreactors
Michael Kalontarov,a Devin F. R. Doud,b Erica E. Jung,a Largus T. Angenentb
and David Erickson*a
Microalgae can serve as a carbon sink for CO2 sequestration and as a feedstock for liquid biofuel
production. Methods for microalgal biomass and biofuel cultivation are progressing, but are still limited
in the efficiency of light delivery and gas exchange within cultures. Specifically, current gas exchange
methods are very energy intensive since they rely on mixing algal cultures at high flow rates. One
method that can improve gas exchange within photobioreactors without excessive mixing is the use of
hollow fibre membranes, which enable simultaneous transport of gases deep into the reactor and rapid
exchange with the culture media. Here we demonstrate the optimal geometric and operational
conditions for CO2 transport to planar cultures of Synechococcus elongatus via hollow fibre membrane
arrays. Specifically, we investigated the effects of inter-fibre spacing and active/passive aeration on the
growth rate, planar surface density, and total biomass accumulation. We show that spacing in excess of
3 times the fibre diameter lead to significant variations in the uniformity of the surface density and
spatially resolved growth rate, whereas spacing of 3 times the fibre diameter supported culture surface
densities nearing 90%, which were maintained for 17 days without decreasing. Active aeration with the
fibres showed an increase in the specific growth rate and the average surface density with respect to
passive aeration by approximately 15% and 35%, respectively, while also eliminating gradients in localized
growth rates along the length of the fibres.
Introduction
Concerns about the impact of climate change, CO2 emissions,and energy security have led to widespread interest in theproduction of biofuels from microalgae.1 Microalgae havehigher CO2 xation efficiencies and growth rates than otherplant-based feedstocks2,3 and the potential to utilize waste-wateror industrial gas wastes as nutrient sources.4 The most devel-oped method for extracting biofuels from microalgae is con-verting their stored lipids into biodiesel,5 which utilizes aseparation process that is very energy intensive.6 This hasprompted the research and development of many engineeredstrains of cyanobacteria to directly secrete fuels such ashydrogen,7,8 ethanol,9,10 isobutyraldehyde,11 and other highvalue products.12
To take advantage of these engineered strains, innovativephotobioreactor (PBR) designs are required that can sustainhigh density cultures while enabling efficient light deliveryand gas exchange.13 The most common reactors used in algalcultivation are open raceway ponds and tubular-type enclosedreactors.14 While these designs do have respective advantages,15
Engineering, Cornell University, Ithaca,
ll.edu
l Engineering, Cornell University, Ithaca,
both are faced with fundamental limitations in delivery ofsufficient light and CO2 and extraction of products to maintainhigh photosynthetic rates.16 The former of these problems hasreceived a signicant amount of attention as of late.17 Unevenlight distribution causes the culture to be overexposed at thesurface and underexposed below the light penetration depth.18
To counteract this problem many approaches have beeninvestigated including the integration of optical bres,19 inter-action with evanescent20 and plasmonic elds,21 and planarwaveguides.22
In parallel to the problem of light delivery, limitations in gasexchange and transport are also being addressed. Traditionally,gas exchange in PBRs is provided by bubbling or passive expo-sure to the atmosphere.23 Though easy to implement, thesemethods constrain optimal PBR geometries, operation andlimit possible culture densities. CO2 delivery is limited byuneven distribution throughout the reactor volume.24 Main-taining a uniform distribution is important for efficient volumeutilization since regions with low CO2 concentration suffer fromlower rates of photosynthesis.25 Turbulent ow mixing is acommon mechanism by which CO2 concentration is equili-brated. This condition requires that a large amount of energy isspent onmixing the algal cultures,26 up to 41% of the cultivationenergy budget in some cases, and contributes to the alreadyhigh energy costs of the algal cultivation process.27 This hasmotivated research into various methods for improving gas
Fig. 1 Two fibre unit cell miniature reactor: (a) schematic of reactor (b)picture of fully fabricated reactor (c) reactors arranged under a fluo-rescent lamp.
Paper RSC Advances
Publ
ishe
d on
13
Nov
embe
r 20
13. D
ownl
oade
d by
Cor
nell
Uni
vers
ity o
n 23
/01/
2014
13:
09:0
5.
View Article Online
exchange coefficients and reducing the mixing energy demandsin PBRs.
A recent advance in enhancing gas exchange has been theintegration of hollow bre membranes (HFMs) modules intoPBRs.28 A HFM consists of hollow bres with membranes thatallow for gas exchange between the media inside and outsidethe bre. HFMs have been used in the chemical, petrochemical,pharmaceutical and galvanic industries,29 and applied in suchvaried applications as wastewater treatment,30 drinking watertreatment,31 tissue engineering,32 and the development of arti-cial organs.33 Recently, HFMs have shown potential to addressthe gas exchange challenges faced by PBRs34,35 and several labscale reactors incorporating HFM modules have been repor-ted.35–38 These studies have veried the potential benets ofintegration of HFMs into PBRs by reporting increased biomassproduction, improved gas exchange, regulation of pH, andpromotion of CO2 xation. In previous work, we have charac-terized the effectiveness with which a single HFM bre, whichwas applied independently of a module, can provide thenecessary gas exchange to locally sustain growth in a carbon-limited reactor with no circulation and only passive gastransport.39
Here, we performed a study to determine the optimal oper-ating and geometric conditions for the use of hollow bre arraysas a method for delivery of atmospheric levels of CO2 to planarcultures of photosynthetic organisms. We report three experi-ments that investigated the effects of inter-bre spacing andactive/passive aeration on the organism growth rate, planarsurface density, and total biomass accumulation. In the rstexperiment, we have characterized what array spacing might bemost effective by studying a two bre unit cell. In the secondexperiment, we fabricated full HFM arrays and measured thebehaviour of the organisms for an extended period of time(30 days). In the third experiment, a study characterizing theeffect of adding active gas ow through the bres was conducted.
Materials and methodsInvestigation of optimal spacing with two bres
To characterize the effects of the spacing in a HFM bre arraywe rst conducted experiments on a unit cell of two bres.Miniature reactors were fabricated, inoculated, and operated inthe manner described in our previous work,39 however, as seenin Fig. 1(a), two bres were inserted instead of one. Thedimensions of the sealed miniature reactors were 40 mm �16 mm � 6 mm and the HFM bres (model no. MHF304KMpurchased from theMitsubishi Rayon Co., Ltd.) were centred onthe bottom of the reactors along the primary axis. The reactorsconsisted of an epoxy sealed polydimethylsiloxane (PDMS)chamber sandwiched between a microscope slide and a largecover slip (Fig. 1(a) and (b)). KwikWeld Epoxy (J-B WeldCompany) was used as a sealant to make the reactors as gastight as possible. The reactors were inoculated with S. elongatusSA665 (obtained from the Liao lab at UCLA11) and modiedBG-11 medium. The medium was modied by the addition of50 mg L�1 thiamine and a 50mM phosphate buffer (pH 7.0), theremoval of bicarbonate, and by sparging with N2 for 30 min in
an anaerobic serum bottle prior to re-suspension and inocula-tion of S. elongatus. This was performed to make the initialculture carbon free and to achieve CO2 limited growth condi-tions in the reactor. The reactors were placed under twouorescent lamp strips (American Fluorescent Plug-in LightStrip), which provided a photon ux density of 75 mE s�1 m�2
(Fig. 1(c)). The ambient temperature in the laboratory wasmeasured to be 25 �C. The ends of the bres where open tothe atmosphere to allow passive gas exchange to occur throughthe lumens of the bres.
Three different unit cell spacing values were tested: 870 mm,1740 mm, and 3480 mm. The diameter of the HFM bres was�290 mm, thus, these spacing distances correspond to 3, 6, and12 bre diameters, respectively. A 3D printed (OBJET Connex500) frame was used to position the bres on the glass slide andensure the proper gap between them; epoxy was used to x thebres in place. Four reactors were fabricated for each spacingdistance. The space between the bres was imaged by usinguorescence microscopy at the time point of inoculation andthen again aer a seven-day operating period. As can be seen inFig. 1(a), the observed region was offset from the edges of thereactor to minimize boundary effects on our measurement (theobserved region was 26 mm long). The captured images were
processed to map the surface density of S. elongatus distribu-tion. Briey, surface density is the percentage of the two-dimensional area, in this case between the two bres, that iscovered by the bacteria (a more thorough description ofimaging setup and processing algorithm is available in ourprevious publication39).
Expanded HFM bre arrays
To conduct experiments with larger arrays of bres, we fabri-cated reactors using the same processes as described above(Fig. 2), except two large cover slides (75 mm � 50 mm � 1 mm)were used to sandwich the epoxy sealed PDMS chamber.
The area of the reactor was increased to 58 mm � 38 mm �6 mm. The bre array was placed on the bottom of the reactorand centred; the bres were arranged parallel to the primaryaxis of the reactor and were 75 mm in length. The bre arrayspanned 20 mm of the reactor width and was offset from thewalls of the reactor by about 1 cm on each side so that edgeeffects did not interfere with the growth in the array. Thedimensions of the observed region were 20 mm � 40 mm, andthe region was marked by a grid placed in the centre of thereactor as seen in Fig. 2(a) and (b). The grid served as a referencefor imaging using a uorescence microscope. We fabricatedthree reactors for two unit-cell types: 19 bres were used tomake the 3-diameter spacing array and 11 bres were used tomake the 6-diameter spacing array (Fig. 2(a)). Aer inoculation(using the same procedure as for the unit-cell experiments), thereactors were placed under the uorescent light strips asutilized in the previous experiment (Fig. 2(b)). The experimentwas conducted for a 30-day operating period and the reactorswere imaged daily on day 0 – day 5, and again on day 10, 17, and
Fig. 2 HFM array reactors: (a) photographs of the two types of reac-tors tested: 3-diameter unit-cell (left) and 6-diameter unit-cell (right).(b) 3 reactors of each type were fabricated, and shown here arrangedunder the fluorescent lamp strip.
1462 | RSC Adv., 2014, 4, 1460–1468
30. Images were taken in the same manner as previously dis-cussed. A total of 360 images were taken per reactor to map thegrowth distribution around the bre array and images werestitched together to produce surface density maps.
Active gas delivery
The goal of this experiment was to study the effect of activeaeration on the growth of S. elongatus in HFM array reactors; allother conditions such as the organism, media, light conditions,etc. were kept the same. To incorporate active ow into thesereactors, we built a setup that allowed us to pump incompressed atmospheric air under controlled ow and pressureconditions (Fig. 3). Three actively aerated reactors werecompared to the same number of passively aerated and non-bre control reactors. The actively aerated reactors were oper-ated at a pressure of 3.45 kPa, which corresponds to a ow rateof �50 mL min�1. The reactors were constructed similarly as inthe previous experiment except the bre ends were bundled andinserted into a coupler to interface them with tubing. The bresused in these experiments were approximately 10 cm long. The3-bre diameter spacing was chosen for the bre array in thesereactors. A cantered 20 mm � 40 mm region was imaged on adaily basis using uorescence microscopy in each reactor andanalysed with the previously described methods. Using theseimages the surface density of the bacteria in each reactor wascalculated. The experiment was allowed to run for 5 days andupon completion the organisms were harvested from the reac-tors. Optical density at 750 nm was measured for these samplesto compare nal bacteria concentration achieved in the threedifferent types of reactors.
Results and analysisInvestigation of optimal bre spacing
Aer seven days of growth, the space between the bres in eachof the two bre unit-cell reactors was imaged and the averagesurface density was calculated. Sample images of the inter-brespacing are presented in Fig. 4. S. elongatus, which appearswhite in these uorescence microscopy images, was inoculatedat a low surface density (Fig. 4(a)). By day 7 of the operatingperiod (Fig. 4(b)), the bacteria covered most of the available areaexcept for a small gap. The local surface density measurementswere averaged together to get the surface density in the observedregion for each reactor, with four reactors for each bre spacing.The dependence of the surface density on the two bre unit-cellspacing is show in Fig. 4(c). The surface density was highest forthe 3-diameter spacing, reaching an average of 80% � 4.5%coverage, while the surface density decreased to 43%� 1.5% forthe 6-diameter gap and 24% � 2.3% for the 12-diameter gap.The amount of reactor surface area not taken up by the bres foreach unit cell type depends on the bre spacing. For each unit-cell type the percentage of reactor area supported for photo-synthetic growth by the bres is 75%, 83%, and 92%, for the 3,6, and 12-diameter gaps, respectively. Multiplying the respectivemeasured surface densities and available area percentagesyielded the better comparison metric to identify which type of
Fig. 3 Setup for flow experiments: the setup consists of a compressed air source, manifold, and pressure regulators to allow for the activeaeration of 3 miniature reactors. The actively aerated reactors are compared to 3 passively aerated reactors, and 3 control reactors with no fibres.
Paper RSC Advances
Publ
ishe
d on
13
Nov
embe
r 20
13. D
ownl
oade
d by
Cor
nell
Uni
vers
ity o
n 23
/01/
2014
13:
09:0
5.
View Article Online
unit cell was more effective. This calculation yields adjustedsurface densities (accounting for the reactor area taken up bythe bres themselves) of 60%, 36%, and 22% for the 3-, 6-, and12-diameter gap unit-cells, respectively. Based on these resultswe decided to further explore the 3- and 6-diameter unit-cells byfabricating bre arrays with these gap distances.
HFM bre arrays
The growth distribution that developed in the 3- and 6-diameterbre arrays differed from each other both spatially andtemporarily. In Fig. 5, we present the surface density maps ofthe growth distribution on day 0, 3, 5, 10, 16, and 30 of theoperating period for both array types. In both cases, the initialdistribution is uniform and at a low surface density of 1.35% �0.4%. By day 3, the surface density maps show clear evidence ofbacterial growth, which was initiated adjacent to the bres inboth cases. However, by day 5 there was greater coverage in the3-diameter spaced array compared to the 6-diameter array.Additionally, the quality of this coverage was also superior asthe bacteria were at a higher surface density in the tighter array.By day 10, most of the available area in the 3-diameter spacedarray is covered by a dense uorescent layer, however, in the 6-diameter spaced array the bacteria are at a comparable surface
Fig. 4 Final surface density for different unit cell fibre spacings: exampleand day 7 (b) of the operating period, illustrating the amount of area takeaverage surface density (averaged over the observed region) for the ddiameter spacing.
density only directly adjacent to the bres. The bacteria layercontinues to get denser in the 3-diameter array until the layercompletely covers all of the available area at an average surfacedensity of 88%. In the 6-diameter array the bacteria were limitedto bands of high surface density directly next to the bres withan area of restricted growth at further distances from the bre.The measurements for day 30 show that surface density of thebacteria distribution decreases in the centre of the 3-diameterarray, while for the 6-diameter array the bacteria bands next tothe bres expand slightly.
To visualize the growth layers, which we have so far repre-sented by surface density maps, we present several uorescentmicroscopy images of S. elongatus in these reactors in Fig. 6. Thesurface density maps are made by processing multiple imagesand stitching the results together to map out the whole20mm� 40mm observed region in each reactor. Looking at theday 17 map for the 3-diameter spaced array (Fig. 6(a)), very highsurface density regions are seen on the le and right panels.The image in Fig. 6(a) depicts a layer that formed in between thebres; note that there is minimal empty space between thebacteria. The right panel is taken at the centre of the observedregion and depicts a layer that has a lower local surface densityregion inside it. For the 6-diameter spaced array (Fig. 6(b)), wewould like to illustrate how the size of the bacteria bands
images of the space between fibres for a 3 diameter spacing on day 1 (a)n up by the bacteria on the initial and final days. (c) Comparison of theifferent unit-cell fibre spacings. Highest percentage reached with 3-
Fig. 5 Surface density maps for two types of arrays over the course of 30 days: local surface density for the (a) 3 fibre diameter spaced array and(b) 6fiber diameter spaced array. A high density uniform bacteria layer develops in the tight fibre array; however, the bacteria grow in band next tothe fibres in the sparse array.
RSC Advances Paper
Publ
ishe
d on
13
Nov
embe
r 20
13. D
ownl
oade
d by
Cor
nell
Uni
vers
ity o
n 23
/01/
2014
13:
09:0
5.
View Article Online
formed next to the bre differs at two locations. On the le, thepanel depicts a location closer to the observed region edge andlarger surface coverage is seen. On the right, an image from thecentre is shown and here a thinner bacterial band is observed.The bands are asymmetric relative to the bres and this isconsistent with our observations of growth around a singlebre.39
In Fig. 5, we illustrate the spatial variation in the bacteriadistributions that developed in the two types of tested arraydesigns. Data pertaining to the development of these distribu-tions with respect to time is presented in Fig. 7. This datadescribes the trends in the total surface density of the bacteriain the observed region for the complete set of reactors (n¼ 3 foreach type). The total surface density is a representation of thefraction of the observed region coved by bacteria. Though theinitial conditions for both types of reactors are virtually iden-tical, the measured values for the surface density quicklydiverge. In the reactors with 3-diameter arrays, a large propor-tion of the available area becomes covered with bacteria in anexponential manner; aer 5 days 63% � 10% of the surface is
1464 | RSC Adv., 2014, 4, 1460–1468
coved by a bacterial layer. Growth is slower in the 6-diameterarray reactors and only 30% � 9% of the surface area is coveredby the bacteria aer 5 days. In the next phase, the rate at whichthe surface density grows decreases in both reactor types. Thebacteria layers in the 3-bre diameter spaced arrays reach amaximum average surface density of 88% � 3%. This highsurface density persists until day 17, but measurements madeon day 30 reveal that the density of these layers does not persistsindenitely and a decrease is observed. In reactors with 6-berdiameter spaced arrays, the average surface density has beenobserved to increase at a slow rate through the course of theexperiment.
By performing an exponential t on the total surface densitydata for days 0–4, we can obtain the specic growth rate for thebacteria that developed in these reactors. The average specicgrowth rate in the 3-bre diameter spaced array and 6-brediameter spaced array reactors was found to be 0.041 � 0.004(h�1) and 0.031 � 0.003 (h�1), respectively. We also mappedthe specic growth rate as a function of location in theobserved region by applying the same method to the local
Fig. 6 Micrographs of bacteria layers in HFM fibre array reactors. Images taken at the edge and centre of the observed region in a 3 fibre diameterspaced reactor (a) and 6 fibre diameter spaced reactor (b).
Fig. 7 Total surface density in the observed region for the 3 fibrediameter spaced and 6 fibre diameter spaced reactors (n ¼ 3 for eachtype). The error bars represent the standard error of the mean.
Fig. 8 Local specific growth rate maps for the two types of fibre arrayswith passive aeration. In the tighter array the growth rate distribution ismore continuous than in the sparser array. Growth is also higher at theedges as expected for passive aeration.
Paper RSC Advances
Publ
ishe
d on
13
Nov
embe
r 20
13. D
ownl
oade
d by
Cor
nell
Uni
vers
ity o
n 23
/01/
2014
13:
09:0
5.
View Article Online
surface density maps. Such maps are presented in Fig. 8.When comparing the maps for the two types of arrays, weobserved two trends. In the tighter arrays the growth rates aredistributed in a continuous manner, however, in the sparserarray the areas of high growth rate are concentrated next to thebres. This of course parallels the observations of the surfacedensity maps presented earlier.
Another trend is that in both cases the growth rates arehigher closer to the edges of the observed region. This is anartefact of the fact that these reactors are passively aerated. CO2
is entering the reactors from the open ends on both sides of the
reactors and a reduced amount is delivered to the centre, thus,limiting the growth rate there. The extent of the higher growthrate region is a function of CO2 diffusion rates in the bre andthe media and the rate of CO2 depletion by the bacteria.
Fig. 10 Local specific growth rate maps for reactors with active andpassive aeration. In the actively aerated reactor the growth rate is moreuniformly distributed along the length of the fibres. In the passivelyaerated array growth is higher at the edges as expected.
RSC Advances Paper
Publ
ishe
d on
13
Nov
embe
r 20
13. D
ownl
oade
d by
Cor
nell
Uni
vers
ity o
n 23
/01/
2014
13:
09:0
5.
View Article Online
Active aeration of HFM array reactors
To determine if the limitations observed in the passive aerationexperiments could be overcome, we compared how bacteriagrew in actively aerated and passively aerated reactors with 3-bre diameter spaced bre arrays versus a set of control reactorswithout bres. The total surface density of the observed regionsfor the complete set of reactors (n¼ 3 for each type) is plotted inFig. 9. The starting density for this experiment was 3.0%� 0.9%.The control reached the lowest surface density aer 5 days, 17%� 3%, due to being limited by gas transport through the liquidvolume and exchange through the reactor walls. The passivelyaerated reactors behaved as before, exhibiting close to exponen-tial growth and reaching 54%� 2% coverage during the course ofthe experiment. The actively aerated reactors reached 74% � 8%average surface density, which was the highest out of this set ofreactors. The specic growth rates for the actively and passivelyreactors were calculated by performing an exponential t on thetotal surface density data for the rst four days. The averagespecic growth rates for the actively and passively reactors were0.037 � 0.002 (h�1) and 0.032 � 0.004 (h�1), respectively.
We also present spatial distribution of the local specicgrowth rates for actively and passively aerated reactors inFig. 10. In the actively aerated reactor, the specic growth rate isuniformly distributed in the direction of gas propagation in theobserved region. However, in the passively aerated reactor wesee higher growth rates toward the le and right edges ofthe observed region as previously discussed. Finally, at theconclusion of the experiment we extracted the liquid volumefrom the reactors and measured the optical density (OD) at750 nm to compare the bacteria concentrations in these reactors.We present this data in Fig. 11. The initial OD for the experiment
Fig. 9 Total surface density in the observed region for the activelyaerated, passively aerated, and control (non-fibre) reactors (n ¼ 3 foreach type). The error bars represent the standard error of the mean.
1466 | RSC Adv., 2014, 4, 1460–1468
was 0.005� 0.001. The control reactors were found to have an ODof 0.05� 0.006, due to growth at the edges of the reactors. In thepassively aerated reactors, the OD of the bacteria solutionreached 0.076� 0.008. The ODmeasurements for these two typesof reactor are closer than the nal surface density values, since
Fig. 11 Optical density measurements of bacteria solution in activelyaerated, passively aerated, and control reactors at the conclusion ofthe experiments.
the bre array only covers�53% of the available reactor area. TheOD found in the actively aerated reactors was 0.214 � 0.010. Thesurface density values for the actively and passively aeratedreactors were much closer together due to the fact that eventhough in both types of reactors the bacteria layers covered asimilar amount of area, the bacteria layer in the actively con-tained a larger number of bacteria due to being stacked to alarger degree in the third dimension.
DiscussionPerformance of bre array reactors
We have shown that the localized effect of a single HFM bre onbacteria growth could be extended to a large area by properlyarraying multiple bres. Table 1 allows us to compare theresults of these experiments. For context, S. elongatus observedin a bubble-aerated, continuously rapidly mixed, culture asksmaintained at a temperature of 35 �C and a similar light ux(50 mE s�1 m�2) as used in our experiments was reported to be�0.058 (ref. 40). The growth rates observed in our experimentsrange from 53–71% of growth rate in well mixed culture asksbut were observed without any mixing. In the single breexperiments, the growth rates observed depended on the initialsurface density.39 The same trend is seen in the array reactors,as the experiments started at 3.0% initial surface density havesomewhat lower growth rates suggesting that CO2 is already alimiting factor at these surface densities. For the same initialsurface density, however, using the 3-bre diameter spacing inthe arrays increased the growth rate by 30% and more thandoubled the surface density aer 5 days with respect to the 6-bre diameter array spacing.
Additionally, our 30-day experiments with the 3 bre diam-eter spaced array reactors showed that the surface density of thebacteria layers increased even further aer day 5 and can bemaintained at these high values without mixing or mediareplenishment for at least 17 days without decreasing. The areaspanned by the array in our experiments was 800 mm2, but itcould be increased if more bres were added to the array whilestill delivering the same performance, as long as the bres aremaintained at the same length. By properly spatially arrangingthe bres we were able to achieve the goals of distributing CO2
and providing channels for gas transport throughout a PBR.
Active aeration to address limitations of passive aeration
To apply HFM arrays over any area both the number of bresand the length of the bres have to be increased. If passive
aeration is used, the bre length is limited by gradients in theCO2 concentration along the bre. The effects of this constraintcan be observed in the specic growth rate maps for thepassively aerated reactors, which were presented in Fig. 8 and10. In both those cases the hotspots for growth were close to theedges of the observed region. Over the length tested in ourreactors (7.5–10 cm) the growth in the middle was diminishedbut still allowed for a uniform distribution of bacteria todevelop over time, as seen in Fig. 5. If longer bres were usedthe gradients in bacteria growth rates would have increased,potentially to the point that growth in the centre would beinhibited. To mitigate this problem we have considered activelyaerating the reactors. When compared to passively aeratedreactors in the same experiment, active aeration improved theinitial specic growth rate and the nal average surface densityby approximately 15% and 35%, respectively (Table 1). Further-more, a more uniform distribution in the growth rate wasobserved (Fig. 10); no discernible gradients along the bredirection indicate that active aeration is a way to overcome thisconstraint on length. The uniformity of the distribution could bemaintained for longer bres if the ow rate, input gas CO2
concentration, and desired bacteria concentration are properlymatched. The proper combinations of these conditions will beexplored in future experiments. Lastly, OD measurementsrevealed that adding aeration also greatly increased the amountof bacteria in the layers that developed around the bres byproviding better access to CO2 than passive aeration, as indicatedby the 3 times greater nal OD in the actively aerated reactors.
Conclusions
Eliminating the need to mix and circulate the bacteria culturethroughout a PBR would greatly reduce the operational energycosts. In this work we have directly applied HFM bres to aplanar culture of cyanobacteria to provide gas exchange andfacilitate growth without circulation or media replenishment.We have demonstrated that a high surface density bacteria layercan be attained andmaintained with passive aeration through aHFM array. While passive aeration can potentially eliminate theneed to spend energy on gas exchange, it limits the area towhich a bre array can be applied. We have shown that activeaeration can be a way to address this limitation since itdecreases the gradients in the growth along the bre length.Though active aeration through the bre array does requireadditional energy we have observed that it also leads to higherreactor productivity. Further work is necessary to understand
how the geometric and operational conditions for the brearrays would change with increased CO2 concentrations, lightintensity, gas ow rates through the bres, and bacteriaconcentrations. In summary, bre arrays have been demon-strated as a possible way to provide gas exchange and facilitatephotosynthetic growth over any surface area.
Acknowledgements
This work was supported by the Advance Research ProjectAgency – Energy and the academic venture fund of the David R.Atkinson Center for Sustainable Future. The authors would alsolike to thank the James Liao group at UCLA for providing the S.elongatus strain used in this study. We also thank MatthewMancuso for helpful discussions regarding the experiment anddata analysis.
References
1 M. K. Lam, K. T. Lee and A. R. Mohamed, Int. J. GreenhouseGas Control, 2012, 10, 456–469.
2 P. J. L. Williams, Nature, 2007, 450, 478–478.3 J. Fargione, J. Hill, D. Tilman, S. Polasky and P. Hawthorne,Science, 2008, 319, 1235–1238.
4 L. Brennan and P. Owende, Renewable Sustainable EnergyRev., 2010, 14, 557–577.
5 A. L. Stephenson, E. Kazamia, J. S. Dennis, C. J. Howe,S. A. Scott and A. G. Smith, Energy Fuels, 2010, 24, 4062–4077.
6 K. Sander and G. S. Murthy, Int. J. Life Cycle Assess., 2010, 15,704–714.
7 M. L. Ghirardi, M. C. Posewitz, P.-C. Maness, A. Dubini, J. Yuand M. Seibert, in Annual Review of Plant Biology, 2007, vol.58, pp. 71–91.
8 K. McNeely, Y. Xu, N. Bennette, D. A. Bryant andG. C. Dismukes, Appl. Environ. Microbiol., 2010, 76, 5032–5038.
9 M. D. Deng and J. R. Coleman, Appl. Environ. Microbiol.,1999, 65, 523–528.
10 D. X. Luo, Z. S. Hu, D. G. Choi, V. M. Thomas, M. J. Realff andR. R. Chance, Environ. Sci. Technol., 2010, 44, 8670–8677.
11 S. Atsumi, W. Higashide and J. C. Liao, Nat. Biotechnol.,2009, 27, 1177–U1142.
12 D. C. Ducat, J. C. Way and P. A. Silver, Trends Biotechnol.,2011, 29, 95–103.
13 D. Erickson, D. Sinton and D. Psaltis, Nat. Photonics, 2011, 5,583–590.
14 D. Aitken and B. Antizar-Ladislao, Energies, 2012, 5, 1613–1647.
15 O. Jorquera, A. Kiperstok, E. A. Sales, M. Embirucu andM. L. Ghirardi, Bioresour. Technol., 2010, 101, 1406–1413.
16 L. Xu, P. J. Weathers, X. R. Xiong and C. Z. Liu, Eng. Life Sci.,2009, 9, 178–189.
1468 | RSC Adv., 2014, 4, 1460–1468
17 P. J. L. Williams and L. M. L. Laurens, Energy Environ Sci.,2010, 3, 554–590.
18 C. G. Lee, Biotechnol. Bioprocess Eng., 1999, 4, 78–81.19 C. Y. Chen, G. D. Saratale, C. M. Lee, P. C. Chen and
J. S. Chang, Int. J. Hydrogen Energy, 2008, 33, 6886–6895.20 M. D. Ooms, V. J. Sieben, S. C. Pierobon, E. E. Jung,
M. Kalontarov, D. Erickson and D. Sinton, Phys. Chem.Chem. Phys., 2012, 14, 4817–4823.
21 M. D. Ooms, L. Bajin and D. Sinton, Appl. Phys. Lett., 2012,101, 253701.
22 E. E. Jung, M. Kalontarov, D. F. R. Doud, M. D. Ooms,L. T. Angenent, D. Sinton and D. Erickson, Lab Chip, 2012,12, 3740–3745.
23 A. Kumar, S. Ergas, X. Yuan, A. Sahu, Q. O. Zhang, J. Dewulf,F. X. Malcata and H. van Langenhove, Trends Biotechnol.,2010, 28, 371–380.
24 A. P. Carvalho and F. X. Malcata, Biotechnol. Prog., 2001, 17,265–272.
25 J. L. G. Sanchez, M. Berenguel, F. Rodriguez, J. M. F. Sevilla,C. B. Alias and F. G. A. Fernandez, Biotechnol. Bioeng., 2003,84, 533–543.
26 C. M. Beal, R. E. Hebner, M. E. Webber, R. S. Ruoff andA. F. Seibert, BioEnergy Res., 2012, 5, 341–362.
27 Y. Chisti, Trends Biotechnol., 2008, 26, 351–352.28 M. S. Abd Rahaman, L. H. Cheng, X. H. Xu, L. Zhang and
H. L. Chen, Renewable Sustainable Energy Rev., 2011, 15,4002–4012.
29 A. K. Pabby and A. M. Sastre, J. Membr. Sci., 2013, 430, 263–303.
30 G. K. Agrahari, A. Rawat, N. Verma and P. K. Bhattacharya,Desalination, 2013, 314, 34–42.
31 N. P. De Souza and O. D. Basu, J. Membr. Sci., 2013, 436, 28–35.
32 D. Mohebbi-Kalhori, A. Behzadmehr, C. J. Doillon andA. Hadjizadeh, J. Artif. Organs, 2012, 15, 250–265.
33 A. A. Palakkan, D. K. Raj, J. Rojan, S. R. G. Raj,P. R. A. Kumar, C. V. Muraleedharan and T. V. Kumary,Tissue Eng., Part A, 2013, 19, 1056–1066.
34 A. Gabelman and S. T. Hwang, J. Membr. Sci., 1999, 159, 61–106.
35 L. H. Fan, Y. T. Zhang, L. H. Cheng, L. Zhang, D. S. Tang andH. L. Chen, Chem. Eng. Technol., 2007, 30, 1094–1099.
36 L.-H. Fan, Y.-T. Zhang, L. Zhang and H.-L. Chen, J. Membr.Sci., 2008, 325, 336–345.
37 H. W. Kim, A. K. Marcus, J. H. Shin and B. E. Rittmann,Environ. Sci. Technol., 2011, 45, 5032–5038.
38 A. Kumar, X. Yuan, A. K. Sahu, S. J. Ergas, J. Dewulf andH. Van Langenhove, J. Chem. Technol. Biotechnol., 2010, 85,387–394.
39 M. Kalontarov, D. F. R. Doud, E. E. Jung, L. T. Angenent andD. Erickson, RSC Adv., 2013, 3, 13203–13209.
40 T. D. B. MacKenzie, R. A. Burns and D. A. Campbell, PlantPhysiol., 2004, 136, 3301–3312.
