Dynamic Microfiltration in Microalgae Harvesting for Biodiesel Production

Published: December 28, 2010

r 2010 American Chemical Society 2455 dx.doi.org/10.1021/ie101070q | Ind. Eng. Chem. Res. 2011, 50, 2455–2460

RESEARCH NOTE

pubs.acs.org/IECR

Dynamic Microfiltration in Microalgae Harvestingfor Biodiesel ProductionSergio D. Rios,† Ester Clavero,‡ Joan Salvad�o,†,‡ Xavier Farriol,† and Carles Torras*,‡

†Departament d'Enginyeria Química, Universitat Rovira i Virgili, Av. Països Catalans, 26, 43007, Tarragona, Catalonia, Spain‡Bioenergy and Biofuels Division, Catalonia Institute for Energy Research (IREC) C/Marcel 3 lí Domingo, 2, 43007, Tarragona,Catalonia, Spain

ABSTRACT: Microalgae is a promising feedstock for biodiesel production because of its lipid content and because it helps todecrease the use of land for nonfood applications (second generation biomass for energy production). To produce biodiesel, severalstages should be followed. An important one is microalgae concentration after the growth stage and prior to lipid extraction.Microfiltration is a possible method if its energy consumption is optimized, considering the total cost of the whole process and thevalue of the final product. Enhanced dynamic microfiltration has been tested as a method for reducing fouling and concentrationpolarization at low transmembrane pressure with promising results. Three variables have been considered so far: rotational speed,transmembrane pressure, and type of membrane (material and mean pore size).

1. INTRODUCTION

Biodiesel can be produced from several well documented rawmaterials.1,2 Among them, microalgae are especially worth study-ing because their productivity is extremely high: some strains arecapable of reproducing more than three times a day3,4 and alsobecause some species are composed of lipids by more than50%5,6 (or up to 80% in cases of genetically modified ones3,7).

The process for producing biodiesel from microalgae consistsroughly of three stages. First, a growth followed by a harvestingprocess has to be performed; this stage includes an operation toconcentrate the microalgae after it has grown. Second, the lipidshave to be extracted from themicroalgae and third, a reaction andseparation process must be performed in order to obtain the finalproduct.3,8 Traditional methods to separate solids from liquidsare membrane processes such as micro- or ultrafiltration pro-cesses. These methods have been used to recover the microalgaebiomass from the growth medium together with other commonmethods such as centrifugation, flocculation, etc., which often arecombined.9-11 Membrane processes can play different roles inthe production of biodiesel, not only on microalgae concentra-tion but also in the transesterification process,12,13 for example.

Because of the nature of the microalgae, a large amount offouling and concentration polarization occurs during the microfiltra-tion process that causes a substantial decrease in the permeation flux,similar to other close applications.14 As will be demonstrated, thisreduction can be up to 95% of the initial flux after less than 2 h.Therefore, a large amount of energy and time are wasted duringmicrofiltration, compromising the feasibility of the process. To reducefouling, dynamic filtration can be used instead, as it maximizes theturbulence over the membrane and thus, the shear stress on it.15,16

There are several typesof dynamicfiltration schemes suchas rotationalsystems, vibratory systems, etc. In this work, a study with a rotationaldisk setup, which contains themembrane, was performed by studyingthe effect of the rotational speed of the disk in order to concentratemicroalgae, thus, removing water from the feed suspension.

The goal of this research note is therefore to provide newexperimental results that show the enhancement that it isachieved in the microfiltration of microalgae through dynamicprocessing.

2. EXPERIMENTAL SECTION

2.1. Biological Materials. The microalgae strain Phaeodacty-lum tricornutum Bohlin has been used in this work (a diatomstrain that has long been used in aquaculture, mainly as a foodsource for fish, crustacean, and bivalve production17). It has afusiform shape, measuring about 40 μm long and 3.5 μmwide. It was chosen due to its ease of growth in large scalevolumes18 and high lipid content.3 It was grown in 300 L ofphotobioreactors with continuous artificial light, at IRTA'slaboratories at Sant Carles de la R�apita (Tarragona, Spain).For all experiments, the cultures were harvested after 6 daysof growth (end of exponential growth) in the same con-ditions and contained a mean concentration of 3.94 �106 ( 0.22 � 106 cells/mL (0.0624 ( 0.0035 g/L). Becauseof the nature of the suspended matter that is alive and incontrast with inorganic solutions, it should be noted that ispractically impossible to achieve exact conditions. Therefore,a similar initial concentration was used in all of the experi-ments.2.2. Methods. The filtration efficiency was evaluated by

measuring the ash-free dry weight (AFDW) and the absorbanceat 750 nm of the culture (initial suspension), retentate and finalpermeate, in order to evaluate the cell concentration. For AFDW,a known volume of sample was filtered through a precombusted

Received: May 12, 2010Accepted: December 13, 2010Revised: November 30, 2010

2456 dx.doi.org/10.1021/ie101070q |Ind. Eng. Chem. Res. 2011, 50, 2455–2460

Industrial & Engineering Chemistry Research RESEARCH NOTE

glass fiber filter, (Whatman GF/F, 0.2 μm). The filtered materialwas washed with ammonium bicarbonate in order to eliminatethe remaining salts while avoiding osmotic shock. It was thendried at 100 �C for 24 h, weighed, and, after being combusted in amuffle furnace for 4 h at 450 �C, reweighed in order todetermine the organic content in weight. The AFDW wascalculated by withdrawing the final filter weight from the driedfilter weight. For absorbance measurements, 1 mL of thesample was fixed with 10 μL of a 37% formaldehyde solution.Absorbance at 750 nm was read with a Synergy HT Multi-Mode Microplate Reader (Biotek). Results are given as cells/mL for better understanding. The cell concentration wasestimated by interpolating absorbance in a least-squaresregression (Y = (65.135 � 106)X þ 640519, r2 = 0.988, whereY is the concentration in cells/mL and X is the absorbancemeasured) obtained by correlating cell counts from a Neu-bauer hemocytometer with absorbance. Additionally, a least-squares regression was used to obtain an expression tocalculate the concentration in terms of mass by interpolatingthe results obtained for AFDW and absorbance (Z = 1.5838�10-11Y, r2 = 0.987, where Z is the concentration in g/mL andY is the concentration in cells/mL).Rejection (also named rejection coefficient) and volu-

metric flux reduction (VFR) have been evaluated in this work.They were calculated by applying eqs 1 and 2, respectively. Inthese equations, “initial” means at time zero and “plateau” isreferred to that period in which no large variation of the flux( J) occurs.

rejection ð%Þ ¼ 100 1-½μ- algae�permeate

½μ- algae�initial

!ð1Þ

VFR ð%Þ ¼ 100 1-JplateauJinitial

� �ð2Þ

The confidence interval has been calculated by applying eq 3,where xh is the mean value, z is the normal distribution, s isstandard deviation, and n the number of samples. In all cases aconfidence level of 95% was used.

confidence interval ¼ x(ZR=2 � sffiffiffin

p ð3Þ

Membrane surface micrographs were obtained by scanningelectron microscopy (SEM). The samples were subjected toa gold sputtering process in order to make them conductive.

The SEM equipment used was a Jeol JSM-6400 scanningmicroscope, operating with a voltage of 15 kV.Shear stress over the membrane can be estimated from the

rotation and operating and geometric parameters using the sameprocedure described in a previous paper,15 by applying eq 4.

τmembrane ¼ 0:057Fv1=5ðkωÞ9=5r8=5 ð4Þ

2.3. Equipment. Figure 1 shows the experimental setup used.It consists of a screw monoblock pump (2.6ID10,PCMMoineau) used specifically with microalgae to minimize theshear inflicted on them.18 The pump motor drive is regulatedby a frequency regulator (VDF-E, Delta Electronics, Inc) in orderto control the flow rate. A back-pressure controller (St€ubbeDHV712R) used to maintain the transmembrane pressure (TMP)constant during the entire experiment was connected to theoutflow of the membrane filtration system (rotational membranemodule). Themodule was a commercial modulemade by KMPT(Krauss Maffei Process Technology) model Dynamic CrossflowFilter DCF 152/0,14, where 152 is the external diameter from themembrane in millimeters and 0.14 is the total membrane surfacearea (square meters). The equipment consists of a closed cavitywith two parallel shafts that each have three overlapping mem-brane disks attached and permeate is collected inside the shaftsand driven out. The disks rotate with a determined angular speedthat can be adjusted by means of a switch/potentiometer unit(E82ZBU for 8200 motec). The membrane module is alsoequipped with a security valve set at 5.5 bar (St€ubbe, DHV-716).Four different commercial membranes made by KMPT were

tested: two ceramics (Al2O3) with mean pore size of 0.5 and2 μm and two polymeric ones made in Teflon (PolyTetra-FluoroEthylene, PTFE) with a mean pore size of 0.5 and1 μm. Each membrane has a filtration area of 0.023 m2.Membrane water permeabilities were tested before using andafter every wash, and the cleaning ratio was evaluated by measuringflux recovery. The manufacturer only provided water perme-ability for the ceramic membranes, which were 1600 L/h/m2/barfor the 2 μm one and 1200 L/h/m2/bar for the 0.5 μm one.2.4. Design of Experiments. All the experiments were

carried out at a recirculation flow rate of 200 L/h and theeffective membrane area was 0.023 m2.The variables studied were the rotational speed, the TMP, and

the type ofmembrane. Four rotational speeds were tested: 0, 560,830, and 1110 rpm (which correspond to 0, 50, 75, and 100%

Figure 1. Experimental setup.



of the module capacity). Whenever rotation is applied, aturbulent regime occurs because Reynolds higher than 1.3 �105 apply. The mean shear stress applied to the membrane canbe estimated by using eq 4.15 Considering the initial conditionof the fluid, assuming that it did not change significantlyduring the experiments, and using a k factor of 0.4,19 themean shear stress on the membrane was 0, 8.93, 18.14, and30.61 Pa, respectively.Two TMPs were tested: 1 and 2 bar. Four different

membranes were tested: two ceramic ones with a mean poresize of 0.5 and 2 μm and two polymeric ones with mean poresize of 0.5 and 1 μm.In general, all of the experiments were carried out from fresh

initial suspension and with extraction of the permeate in orderto evaluate the concentration process. An exception was madefor those in which the rotational speed was studied in the sametest. In these cases, a fresh initial suspension was also used butthe permeate was driven to the feed vessel in order to keep theconcentration constant.At least three experiments per test were conducted to study

reproducibility. After each experiment, membranes werecleaned by soaking them in a 400 mg/L NaClO solution indemineralized water, for 12 h. Afterward, they were assembledagain, and demineralized water was pumped for 1 h at 1 bar.Finally, water permeability was measured again. In those casesthat permeability was less than 90% of the theoretical value(the one measured before the first use of the membrane withtest fluid), the above-mentioned NaClO solution was pumpedfor 1 h at 1 bar followed by demineralized water pumping for1/2 h at 1 bar.The temperature of the bath containing the recirculating fluid

was being measured but not controlled. Fluctuations up to 8 �Coccurred at a rate of 3 �C/h which is expected to have a limitedeffect onmembrane permeability due to viscosity changes. As theexperiments were performed within a similar period of time, asystematic effect of temperature change in the results wasexpected, and the comparative results were analyzed. Moreover,once the plateau was reached, in all experiments and despite ofthe temperature increment, no significant changes in permeabil-ity were observed.

3. RESULTS AND DISCUSSION

The variables studied were the rotational velocity of thedisk, TMP, and the type of membrane. Their effects onpermeability and rejection were evaluated. In these experi-ments, the type of membrane had effects on both parameters,

but rotational velocity and TMP only affected permeability.Table 1 shows the flux results for all of the membranes testedat the different rotational speeds and TMP. Values of waterpermeability measured at 1 bar and for all the membranes usedwere 1769 ( 20 for the 2 μm ceramic membrane, 1445 ( 45for the 0.5 μm ceramic membrane, 1046 ( 186 for the 1 μmpolymeric membrane, and 1302 ( 269 for the 0.5 μmpolymeric membrane. Values indicate that reproducibility inpolymeric membranes is lower than in commercial ones. Also,it is noticeable that for polymeric membranes, permeability issimilar for both.

With regard to the rotational velocity of the disks, severalexperiments were performed at 0, 560, and 1110 rpm for eachTMP. When using the microalgae suspension, at 1110 rpmpermeability was 212 ( 68 L/m2/h/bar, at 560 rpm it was109 ( 48 L/m2/h/bar, and with no rotational speed it was68 ( 25 L/m2/h/bar. Values show the important VFR mea-sured and caused by the particulates that promote fouling andconcentration polarization as well as the enhancement ob-tained by applying rotation. Looking at the mean values of allthe experiments performed, VFR at 1110 rpm was 79% ( 7%,

Table 1. Membrane Fluxes

flux L/h/m2

TMP membrane initial plateau at 1110 rpm plateau at 550 rpm plateau at 0 rpm

2 bar ceramic, 2 μm 2520( 73 460( 73 222( 59 130( 18

ceramic, 0.5 μm 1830( 47 431( 44 268( 28 105( 33

polymeric, 1 μm 1227( 16 200( 62 67( 16 69( 30

polymeric, 0.5 μm 3629( 298 207( 45 99( 17 75( 40

1 bar ceramic, 2 μm 1286( 96 403( 64 211( 35 130( 15

ceramic, 0.5 μm 943( 21 370( 25 265 ( 38 127( 23

polymeric, 1 μm 867( 180 127( 19 44( 4 33( 15

polymeric, 0.5 μm 1154( 162 195( 66 99( 14 54( 27

Figure 2. Decline of the permeation flux with time at differentrotational speeds. Case corresponding to a ceramic membrane with amean pore size of 2 μm, a TMP of 2 bar, and a recirculation volumetricflow rate of 200 L/h.



at 560 rpm it was 91% ( 5%, and at 0 rpm it was 94% ( 2%.Figure 2 shows a representative case of flux decline with timecaused by fouling and concentration polarization at differ-ent rotational speeds (ceramic membrane, mean pore sizeof 2 μm, TMP of 2 bar). The experiment was started with100% of the rotational speed measuring an initial permeabilityof about 1240 L/h/m2/bar. After 40 min, the plateau wasreached with a permeability of about 240 L/h/m2/bar, andtherefore a VFR of 80% was measured. Afterward, the rota-tional speed was reduced stage by stage obtaining a higherdecline of fluxes, up to a VFR of 95% (60 L/h/m2/bar) witha rotational speed of 0. In each reduction of rotational speedand in all the experiments performed, a sharp and substantialdecrease of the flux was observed. It should be stated that inthis experiment, permeate was driven into the feed tank in order tomaintain the concentration constant along all the experiment.

With regard to the TMP, performance was better at 1 barthan at 2 bar, in terms of permeability. Initially and in allexperiments, the permeability was almost double at 2 bar(as theoretically expected), but at the plateau the permeabil-ities at 1 and 2 bar were similar. Considering the mean valuesof all the experiments performed, VFR at 2 bar was 91%( 4%,while at 1 bar it was 84% ( 6%. Table 2 shows the resultsfor each rotational speed. In this case, it can be stated thatfouling effects, which reduce the permeability of the mem-brane, increase when the TMP also increases. In the caseof a ceramic membrane with a mean pore size of 0.5 μm and1110 rpm, the final flux at 1 bar was 370 ( 25 L/h/m2, whileat 2 bar it was 431 ( 44 L/h/m2. Initial fluxes were 943 ( 21and 1830 ( 47 L/h/m2, respectively. It is a promising resultas it indicates that good performance is obtained at a lowerenergy input.

Concerning the type of membrane, polymeric and ceramicmembranes were evaluated (Figure 3). In terms of perme-ability, ceramic membranes had better performance thanpolymeric membranes because of three factors: VFR due tofouling and concentration polarization was less for ceramicones (for example, considering the case of the 0.5 μm meanpore size membranes, initial permeability was higher for thepolymeric membrane, but the ceramic one had a larger valueat the plateau), reproducibility was better, and cleaning waseasier for ceramic membranes. In all cases, the initial perme-ability measured with water recovered after cleaning. A maincause for this behavior could be because of the different typesof porous structure that ceramic and polymeric membraneshave. To show this difference, SEM micrographs were ob-tained and are shown in Figure 4. It can be seen that whilepores in ceramic membranes correspond to the spaces that arebetween the solid particles, in Teflon polymeric membranesthe fibers create a net that constitute the barrier. Although themean rejection of the membrane is the one stated, due to thegeometry of the net, which has holes that are much larger than

the mean theoretical value corresponding to the membranecutoff, more biomass can reach the inside of the membraneand is trapped inside, increasing the level of fouling. Anothercause may be the electrical interactions between the mem-brane material and part of the biomass being filtered, parti-cularly exopolysaccharides.20 Another study focused on thisphenomena would be needed in order to discuss this further,but Teflon membranes are reported to have higher biofoulingdue to their roughness and lower surface potential,21 so it canbe expected that this cause could be of more importance forthe polymeric membrane rather than for the ceramic one. Interms of rejection, Table 3 shows the results as a functionof the material and mean pore size. First, the performanceof the ceramic membranes was better than the polymeric ones.Ceramic membranes offered an almost complete rejec-tion (although the mean rejection is higher for the membranewith a mean pore size of 0.5 μm than the 2 μm one, it isbecause of the experimental error as the statistics show).Second, the reproducibility of ceramic membranes was higher(lower standard deviation) than the polymeric membranes.On the other hand, membranes with the same mean poresize offered different rejections. Therefore, rejection dependslargely on several parameters and can differ greatly fromwhat is to be expected based on the calculation of themembrane's nominal cutoff. In this case, the shape of theparticulate combined with the structure of the pores ofthe membrane can play an important role in the selectivityof the membrane. Figure 5 shows the concentration profilesof the retentate and permeate streams for polymeric andceramic membranes. The increase in the cell concentrationof the retentate was higher for the ceramic membrane than forthe polymeric membrane. Consequently, while the concen-tration of microalgae in the permeate stream was null in thecase of the ceramic membrane, significant values were deter-mined for the same stream for the one concerning thepolymeric membrane.

Table 2. Volumetric Flux Reduction of Membranes at Dif-ferent TMP and Rotational Speeds

volumetric flux reduction (%)

(mean value ( confidence interval)

1110 rpm 560 rpm 0 rpm

1 bar 75.3( 9.7 87.8( 8.2 91.7( 3.7

2 bar 83.6( 8.4 93.4( 4.1 95.2 ( 1.9

Figure 3. Variation of the permeation flux with time of polymeric andceramic membranes with a mean pore size of 0.5 μm at differentrotational speeds, a TMP of 2 bar, and a recirculation volumetric flowrate of 200 L/h.



4. CONCLUSIONS

Microfiltration of microalgae by using dynamic systems pro-vides a clear enhancement of the process performance. As foulingand concentration polarization effects obtained with the use ofthis suspended matter are very important, increasing the shearstress over the membrane significantly improves the permeabilityof the membrane. Optimum conditions can be found by a propercombination of variables such as the TMP or the rotationalspeed. Results demonstrate that a higher TMP does not improvemicrofiltration performance due to fouling effects. Therefore, the

process can be carried out with lower energy costs and can bemade more cost-effective.

Membranes with similar nominal cut-offs but that are made ofdifferent materials offer different performance in terms ofselectivity. For the Phaeodactilum Tricornutummicroalgae strain,ceramic membranes with a mean pore size of two micrometersoffer complete rejection at lowTMP and reasonable permeabilityby using rotational enhancement.

Further experiments are being performed in order to continuethis study, including more microalgae species experiments atdifferent TMPs and other tests, as well as a study of the economicbalance of the process.

’AUTHOR INFORMATION

Corresponding Author*Tel.: þ34.977.202.444. E-mail: [email protected].

’ACKNOWLEDGMENT

This work is supported by the project ENE2007-65033funded by the Spanish Ministry of Science and Innovation andby the project “Fuels from Biomass” under the microalgaeresearch program funded by the Excma. Diputaci�o de Tarragona.The authors wish to thank the researchers and technicians of theAquatic Ecosystems Research Group of IRTA, Sant Carles de laR�apita, Tarragona, Spain.

’REFERENCES

(1) Demirbas, A. Importance of Biodiesel as Transportation Fuel.Energy Policy 2007, 35, 4661.

(2) Vasudevan, P. T.; Briggs, M. Biodiesel Production;Current Stateof the Art and Challenges. J. Ind. Microbiol. Biotechnol. 2008, 35, 421.

(3) Chisti, Y. Biodiesel from Microalgae. Biotechnol. Adv. 2007,25, 294.

(4) Miao, X. L.; Wu, Q. Y. Biodiesel Production fromHeterotrophicMicroalgal Oil. Bioresour. Technol. 2006, 97, 841.

(5) Rosenberg, J. N.; Oyler, G. A.;Wilkinson, L.; Betenbaugh,M. J. AGreen Light for Engineered Algae: Redirecting Metabolism to Fuel aBiotechnology Revolution. Curr. Opin. Biotechnol. 2008, 19, 430.

(6) Packer, M. Algal Capture of Carbon Dioxide; Biomass Genera-tion as a Tool for Greenhouse Gas Mitigation with Reference to NewZealand Energy Strategy and Policy. Energy Policy 2009, 37, 3428.

(7) Spolaore, P.; Joannis-Cassan, C.; Duran, E.; Isambert, A. Com-mercial Applications of Microalgae. J. Biosci. Bioeng. 2006, 101, 87.

Table 3. Rejection of Membranes Calculated from AFDW

mean value ( confidence interval

ceramic(2 μm)

ceramic(0.5 μm)

polymeric(1 μm)

polymeric(0.5 μm)

mean rejection(%)

98.4( 1.0 99.2( 0.6 90.1( 5.7 99.4( 0.5

Figure 5. Evolution of microalgae concentration in the retentate andpermeate streams for polymeric and ceramic membranes, a TMP of2 bar, and a recirculation volumetric flow rate of 200 L/h.

Figure 4. Scanning electron surface micrographs of the (a) 0.5 μm ceramic and the (b) 0.5 μm polymeric membranes.



(8) Meng, X.; Yang, J. M.; Xu, X.; Zhang, L.; Nie, Q. J.; Xian, M.Biodiesel Production from Oleaginous Microorganisms. Renew. Energ.2009, 34, 1.(9) Rossignol, N.; Vandanjon, L.; Jaouen, P.; Quemeneur, F. Mem-

brane Technology for the Continuous Separation Microalgae/CultureMedium: Compared Performances of Cross-Flow Microfiltration andUltra-Filtration. Aquacult. Eng. 1999, 20, 191.(10) Rossi, N.; Jaouen, O.; Legentilhomme, P.; Petit, I. Harvesting of

Cyanobacterium Arthrospira Platensis Using Organic Filtration Mem-branes. Food Bioprod. 2004, 82, 244.(11) Molina Grima, E.; Belarbi, E. H.; Acien Fernandez, F. G.;

Robles Medina, A.; Chisti, Y. Recovery of Microalgal Biomass andMetabolites: Process Options and Economics. Biotechnol. Adv. 2003,20, 491.(12) Cao, P. G.; Tremblay, A. Y.; Dube, M. A. Kinetics of Canola Oil

Transesterification in a Membrane Reactor. Ind. Eng. Chem. Res. 2009,48, 2533.(13) Cao, P. G.; Tremblay, A. Y.; Dube, M. A.; Morse, K. Effect of

Membrane Pore Size on the Performance of a Membrane Reactor forBiodiesel Production. Ind. Eng. Chem. Res. 2007, 46, 52.(14) Sur, H. W.; Cui, Z. F. Enhancement of Microfiltration of Yeast

Suspensions using Gas Sparging;Effect of Feed Conditions. Sep. Purif.Technol. 2005, 41, 313.(15) Torras, C.; Pallares, J.; Garcia-Valls, R.; Jaffrin, M. Y. Numerical

Simulation of the Flow in a Rotating Disk Filtration Module. Desalina-tion 2009, 235, 122.(16) Brou, A.; Ding, L. H.; Boulnois, P.; Jaffrin, M. Y. Dynamic

Microfiltration of Yeast Suspensions Using Rotating Disks Equippedwith Vanes. J. Membr. Sci. 2002, 197, 269.(17) Richmond, A. Handbook of Microalgal Culture: Biotechnology

and Applied Phycology; Wiley-Blackwell: Oxford, OX, UK, 2003.(18) Jaouen, P.; Vandanjon, L.; Quemeneur, F. The Shear Stress of

Microalgal Cell Suspensions (Tetraselmis Suecica) in Tangential FlowFiltration Systems: The Role of Pumps. Bioresour. Technol. 1999,68, 149.(19) Bouzerar, R.; Ding, L. H.; Jaffrin, M. Y. Local Permeate Flux-

Sheer-Pressure Relationships in a Rotating DiskMicrofiltrationModule:Implications for Global Performance. J. Membr. Sci. 2000, 170, 127.(20) Morineau-Thomas, O.; Jaouen, P.; Legentilhomme, P. The

Role of Exopolysaccharides in Fouling Phenomenon during Ultrafiltra-tion of Microalgae (Chlorella sp and Porphyridium Purpureum): Advan-tage of a Swirling Decaying Flow. Bioprocess Biosyst. Eng. 2002, 25, 35.(21) Kang, S.; Hoek, E.M. V.; Choi, H.; Shin, H. Effect ofMembrane

Surface Properties During the Fast Evaluation of Cell Attachment. Sep.Sci. Technol. 2006, 41, 1475.

Documents

Dynamic Microfiltration in Microalgae Harvesting for Biodiesel Production