Bioresource Technology 132 (2013) 293–304

Contents lists available at SciVerse ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Potential pre-concentration methods for Nannochloropsis gaditana and acomparative study of pre-concentrated sample properties

0960-8524/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biortech.2013.01.037

Abbreviations: A, OD750 (optical density at 750 nm) of sample; AFDW, ash-freedry weight; AS, aluminium sulphate; B, OD750 of initial culture; C, algae culture; C,OD750 of reference blanks; Ca, calcium; CF, concentration factor; Df, mean diameterof floc; FC, filtered algae culture; FE, flocculation efficiency; GM, growth medium;HCl, hydrochloric acid; hf, final height of concentrated algae solution; ho, initialheight of examined algae solution; k, constant of viscosity; Mg, magnesium; NaOH,sodium hydroxide; Nng, Nannochloropsis gaditana; OD, optical density; PAC,polyaluminium chloride; Pht, Phaeodactylum tricornutum; SR, sedimentation rate;SSVF, settleable solid volume fraction; s, shear stress; _c, shear rate.⇑ Corresponding author at: Departament d’Enginyeria Química, Universitat

Rovira i Virgili, 43007 Tarragona, Catalonia, Spain. Tel.: +34 977 559 64.E-mail address: jsalvado@irec.cat (J. Salvadó).

Sema S�irin a, Ester Clavero b, Joan Salvadó a,b,⇑a Departament d’Enginyeria Química, Universitat Rovira i Virgili, 43007 Tarragona, Catalonia, Spainb Bioenergy and Biofuels Division, Institut de Recerca de l’Energia de Catalunya (IREC), C/Marcel lí Domingo 2, 43007 Tarragona, Catalonia, Spain

h i g h l i g h t s

" Pre-concentration methods for high lipid content microalga species were studied." Dewatering with autoflocculation is found promising." Characteristic properties of pre-concentrated samples were evaluated and compared." Viscosity, PSD and Ca/Mg ions of pre-concentrated samples were analysed.

a r t i c l e i n f o

Article history:Received 10 October 2012Received in revised form 6 January 2013Accepted 7 January 2013Available online 22 January 2013

Keywords:MicroalgaeBiodieselPre-concentrationpHAutoflocculation

a b s t r a c t

We compared potential pre-concentration techniques for Nannochloropsis gaditana (Nng) by testing nat-ural sedimentation; flocculation with aluminium sulphate, polyaluminium chloride and chitosan; andinduced pH. Promising flocculation efficiencies and concentration factors were obtained in a short timewith alkalinity-induced flocculation at an adjusted pH of 9.7 and with chitosan at an adjusted pH of 9.9using a concentration of 30 mg L�1. The sedimentation rates of alkalinity-induced flocculation were alsoevaluated. Additionally, viscosity, particle size distribution and Ca/Mg ions were analysed for pre-concen-trated samples of N. gaditana (Nng) and the previously studied Phaeodactylum tricornutum (Pht) whichwere obtained by various different harvesting methods under optimal conditions. The rheological prop-erties of the concentrated algae suspensions of two microalgal species showed Newtonian behaviour. Themean diameters of the flocs were between 39 and 48 lm. The Ca/Mg analysis showed that Mg+2 is thetriggering ion for alkalinity-induced flocculation in the conditions studied.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Microalgae are photosynthetic organisms that store solar en-ergy in biomass chemical bonds through photosynthesis with theuse of inexpensive natural resources such as CO2 and H2O. Eversince they were identified as a possible raw material for the

production of biodiesel, among other commercial applications,the growth aspects of microalgal cultures have received consider-able attention.

Of the various biotechnological processes involved in algaefarming, harvesting is especially important in determining the costand quality of the end product. It is estimated that at least 20–30%of the total cost of producing biomass can be attributed to therecovery process (Christenson and Sims, 2011). Furthermore, mic-roalgal removal has long been used in water treatment, in whichthe methods applied are similar to those for biofuel production,although with different concerns about the harvested microorgan-isms. Regardless of the objective of the harvesting process, thesmall size of the algal cells, the scant difference in density betweenalgae and the growth medium, and the dilute concentrations of al-gal cultures make the harvesting process a key challenge, espe-cially on an industrial scale. To overcome these challenges, alow-energy based harvesting method needs to be developed. Thecombination of several separation strategies has been proposed

294 S. S�irin et al. / Bioresource Technology 132 (2013) 293–304

in which the selection of strategies depends on the species and thedesired final product, among other factors.

The first step in the harvesting process (pre-concentration)should essentially be simple, fast, efficient and cost effective. Theaim of the pre-concentration, the primary dewatering operation,is to increase the concentration of algae and to reduce the volumeto be processed for further treatment. The simple sedimentationsystem is suitable for harvesting microalgae which have naturallyhigh sedimentation rates (Danquah et al., 2009). If the species haspoor sedimentation properties (S�irin et al., 2011), the majortechniques that can be used to harvest microalgae cells are centri-fugation, flocculation, filtration and screening, flotation and elec-trophoresis (Uduman et al., 2010).

Of these techniques, flocculation is one of the most commonand its mechanism mostly depends on cell and flocculant charges.The efficiency of flocculation is affected by the concentration, theionic strength, the zeta potential (n), the pH of the culture solution,the dosage of the flocculants and co-flocculants, the pH adjustmentbefore or after the flocculants are added, and the mixing time andspeed. The degree of salinity also has an effect. Numerous coagu-lants or flocculants have been tested, most commonly with metalsalts such as aluminium sulphate, ferric chloride and ferric sul-phate (Papazi et al., 2010), but also with chitosan (Divakaran andSivasankara Pillai, 2002), cationic starch (Vandamme et al., 2011),bioflocculant produced from bacteria (Salim et al., 2011), etc.

Autoflocculation is the spontaneous formation of flocs and thesubsequent settling of microalgae (Uduman et al., 2010). It occursas a result of the precipitation of carbonate salts with algal cells athigh pH, a consequence of algae photosynthetic CO2 consumption(Sukenik and Shelef, 1984). Therefore, prolonged cultivation insunlight with a limited CO2 supply promotes the autoflocculationof algal cells for harvesting. Some cultures naturally reach highalkaline pHs during the exponential growth phase due to photo-synthesis (CO2 consumption) (S�irin et al., 2011) and the simulationof autoflocculation, alkalinity-induced flocculation, using causticsoda, lime (Nurdogan and Oswald, 1995), and other substanceshas also been studied on the laboratory scale (Semerjian andAyoub, 2003; Chen et al., 2011). Autoflocculation is by no meansa new harvesting technique in water treatment (Tenney and Ver-hoff, 1973). However, although autoflocculation might be consid-ered one of the most promising methods because it requires lessenergy, has a high removal efficiency, is an induced flocculationmethod that does not need chemicals, and mostly takes advantageof gravity settling, until recently it has not been given much atten-tion by research articles concerning the production of biofuel frommicroalgae.

Each algal species reacts differently to different processing tech-nologies, and the optimal method for maximising algal biomassrecovery may very likely depend on the strain used (Natural AlgalBiofuels Technology Roadmap). Among the many species of algae,those belonging to the genus Nannochloropsis are particularlyinteresting because of their ability to accumulate large quantitiesof lipids, which can reach concentrations of up to 65–70% of theirtotal biomass (Simionato et al., 2011). However, in field studies,the lipid content of dry biomass is much lower than the high valuesreported on a laboratory scale (S�irin et al., 2011).

Nannochloropsis gaditana is a microalgal species that belongs tothe class Eustigmatophyceae and, in addition to its lipid content, iswidely recognised as an important source of pigments of greatcommercial value (Macias-Sanchez et al., 2005). Therefore, thisspecies is also considered a promising candidate for industrial bio-fuel production applications, and studies on its morphology, ultra-structure and the growth physiology of its system have beenconducted by other authors (Lubián, 1982).

The harvesting of high lipid content Nng algae, low-energy har-vesting procedures and optimal methods for maximising algae

recovery and reducing costs are important research topics. Thisarticle focuses on (I) potential pre-concentration methods by natu-ral sedimentation, and flocculation with commercial flocculants(aluminium sulphate, polyaluminium chloride) and chitosan; and(II) analysing the pre-concentrated samples of Nng and the previ-ously studied Phaeodactylum tricornutum (Pht) (S�irin et al., 2011).Pre-concentrated sample characteristics such as viscosity, particlesize distribution (PSD) and magnesium and/or calcium concentra-tions that comparatively little research has been published wereaimed to provide insight into flocculation behaviours.

2. Methods

2.1. Microalgal cultures

The Nng strain was obtained from the Institut de Recerca iTechnologia Agromentaries (IRTA) in Sant Carles de la Rapita (Tar-ragona, Spain). The composition of culture medium (although herewe did not add sodium silicate), the culture preparation and ash-free dry weight (AFDW) protocols were described by earlier studies(S�irin et al., 2011).The cultures (300 L of slurry) were harvested onthe first day of the stationary phase after 10 days of growth. Celldensity and pHculture were monitored offline by taking daily sam-ples. Cell concentration was estimated by interpolating absorbancein a least squares regression for y = 14.056 ⁄ 107x + 928993(R2 = 0.99), where y represents the cell density (cells mL�1) and xis the absorbance value at 750 nm (the wavelength for turbidity).Formaldehyde solution (37%) was used as a fixative for turbiditysamples (1 mL of sample was fixed with 10 lL of solution).

Microscopic photos were taken using a Zeiss Axio Scope A1microscope. The culture properties of Nng are detailed in Table 1a.

2.2. Natural sedimentation and flocculation experiments

To determine the effects of light and temperature on the sedi-mentation of Nng cultures, natural sedimentation experimentswere conducted under varying conditions: in daylight and dark-ness and at different temperatures.

To determine the effect of pH on algal flocculation, the pH of thestock microalgae solution was adjusted to acidic and alkaline pHs(2 6 pH 6 11) with HCl or NaOH (0.1 and 1 N) solutions.

The algal cells were also flocculated with metal salts (alumin-ium sulphate (AS), pre-hydrolyzed metal salt-polyaluminium chlo-ride (PAC)) and with chitin-derived polysaccharide (chitosan) tocompare the effects of different flocculants. The flocculants ASand PAC were purchased from Kemira Iberica S.A., Spain, and chito-san was purchased from Sigma Aldrich Co., Spain.

The required concentrations of PAC and AS solutions were pre-pared by diluting stock solutions of flocculants to a reasonable andeffective dilution ratio (e.g. considering algae cell damage and thecorrosivity factor). The chitosan stock solution (5 g L�1) was pre-pared by dissolving chitosan in 1% acetic acid solution under con-tinuous agitation until a clear solution was obtained. Otherdilutions of chitosan were made from the stock solution to obtainthe required concentrations.

All experiments on sedimentation and flocculation were per-formed as previously described in S�irin et al. (2011). Flocculationefficiency (FE), the concentration factor (CF) and settleable solidvolume fraction (SSVF) were also calculated in the same way de-scribed previously (S�irin et al., 2011). Cell concentrations for allflocculation samples were monitored with a UV–visible spectro-photometer (Synergy HT Multi-Mode Microplate Reader (Biotek))with absorbance set at 750 nm. Fig. 1 shows the sample heightsfor natural sedimentation and flocculation experiments.

Table 1aCharacteristics of Nannochloropsis gaditana cultures.

Shape ofcell

Mean size(lm)

Cell concentration (cellsmL�1)

Dry weight (DW)(mg L�1)

Ash-free DW(mg L�1)

Salinity(‰)

pHrange

Mg & Ca ions composition(mM)

Spherical 3–4(diameter)

14.03 ± 0.67 � 106 132 ± 3.32 89.92 ± 1.61 �38 8.5–10 Mg:57.80±0.0Ca:9.76± 0.51

Fig. 1. Figure of sample heights for natural sedimentation and flocculationexperiments.

S. S�irin et al. / Bioresource Technology 132 (2013) 293–304 295

2.3. Comparison of pre-concentration process for N. gaditana and thepreviously studied P. tricornutum species

In order to better understand the pre-concentration processesof Nng and the previously studied Pht (S�irin et al., 2011), we useddifferent techniques (Table 1b) to harvest the following three typesof sample under a variety of conditions;

(1) Growth medium (GM), which was obtained by means of theprocedure described in the microalgal cultures section (for300 L slurries of algae), but skipping the algal inoculationstep so that the effect of the growth medium on the harvest-ing processes could be observed.

(2) Microalgal cultures (C), which were obtained as described inthe microalgal cultures section (for 300 L slurries of algae) sothat the effect of the algae species could be observed.

Table 1bExperimental conditions to obtain concentrated samples of Phaeodactylum t

Sample treated Treatment method

(1) Growth medium (GM) Alkalinity induced floccu

(2) Microalgae culture (C) Flocculation with comme

Flocculation with comme

(3) Filtered culture (FC) Flocculation with chitosa

* Data taken from S�irin et al. 2011.� Data obtained at this study.

(3) Filtered culture (FC), which was obtained by filtering micro-algal culture through glass fibre filters (Whatman GF/F GlassMicrofiber filter 4.7 cm, nominal pore size 0.7 lm), cell-freeculture, to observe the effect of other materials in the culturemedium (e.g. extracellular organic matter).

All experiments in this section were conducted using the sedi-mentation experimental procedure in 250 mL cylinders as per-formed in pH-induced flocculation experiments.

2.3.1. Viscosity analysis of samples after harvestingThe rheological properties of algae slurries play a role in agita-

tion and pumping power requirements and must be considered asone of the design aspects for algae production processes. Researchshould be carried out into downstream processes and increasingcell concentrations during harvesting. The rheology of cell culturesis also influenced by cell morphology (e.g. size, shape and aggrega-tion) as well as the biomass concentration and the process applied.

The samples of two marine species pre-concentrated withdifferent treatment methods were tested for rheometric character-istics. Viscosities were measured over a range of shear rates(245–2700 s�1) with a rotational viscometer (Thermo-HaakeVT550) using an NV sensor at 30 �C. According to the shear rates,the range of flows could be assumed to be between 45 and500 L min�1 (�3–30 m3 h�1). The culture densities of both specieswere also similar to productivity rates found in outdoor facilities(especially in raceway ponds). All the rheometric measurementsstudied were taken in triplicate and the viscosities were calculatedusing the averages of the measurements. Under the shear ratesused, the following assumptions were made: (i) algal flocs didnot change their shape; (ii) no cell settling occurred during mea-surements; (iii) the algae were not damaged.

The kinematic viscosities of the residual solutions after everyharvesting technique were monitored by means of an Oswald–Cannon–Fenske viscometer (at 30 �C) to check for any differencesin viscosity in relation to the culture medium.

There are no straightforward techniques for experimentallycharacterising floc strength without destroying the flocs to someextent, and there is little information in the literature on research-ing floc structure and floc strength under different coagulationmechanisms. So, we tried to understand floc strength by applyingdifferent agitation speeds (max 1500 rpm) and shear rates

ricornutum and Nannochloropsis gaditana species.

Condition

lation pHthreshold

pH = 11rcial flocculant PAC 30 ppm for P. tricornutum*

20 ppm for N. gaditana�

rcial flocculant AS 30 ppm for P. tricornutum*

20 ppm for N. gaditana�

n 20 ppm for P. tricornutum*

30 ppm for N. gaditana�

296 S. S�irin et al. / Bioresource Technology 132 (2013) 293–304

(2700 s�1). The settling efficiencies of the samples were checkedafter the viscosity measurements had been taken.

2.3.2. Particle size analysis of samples after harvestingParticle/floc size distribution was analysed in order to make an

in-depth study of the diameters of the particles/flocs. Understand-ing the characteristics of the floc (e.g. size, shape, density) is essen-tial if the settling characteristics are to be understood and designprocesses developed. The particle sizes of the concentrated sam-ples were measured using a Coulter Multisizer 3 (Beckman CoulterInc, USA) with a 1000 lm orifice tube capable of characterisingflocs ranging in size between 20 and 600 lm, depending uponthe density of the flocs being analysed. Gibbs (1982) mentionedthe possibility of floc breakage when the particles pass through asmall orifice. To prevent this from happening, an orifice of1000 lm was used to characterise flocs. Stirring was applied sothat the flocs could not settle. The average/mean diameter of theflocs (Df) was calculated using the overlays of three samples. Sea-water was used as the conductive electrolyte and filtered withglass fibre filters (Whatman GF/F Glass Microfiber filter 4.7 cm,nominal pore size 0.7 lm) prior to use. Multisizer AccuComp1.19 software provided by Coulter Limited was used for furtherdata analysis. All data given in this study are the average of threemeasurements.

2.3.3. Ca and Mg ion analyses of samples after harvestingThe concentration of Ca and Mg ions in residual water were

measured by means of atomic absorption spectroscopy (model3110; Perkin–Elmer) after harvesting the algae culture in orderto better understand the mechanism at work in alkalinity-inducedflocculation. The Ca and Mg contents of the monitored sampleswere compared with the content of the culture medium to helpestablish the mechanism.

3. Results and discussion

3.1. Pre-concentration of N. gaditana

3.1.1. Natural sedimentation and sedimentation with pH-inducedflocculation

In natural sedimentation experiments, no significant sedimen-tation of Nng was detected in either dark or light conditions or atdifferent temperatures (see Fig. S1 in Supplementary data). Thisproves that natural sedimentation is not a useful harvesting tech-nique for the pre-concentration of either this species or Pht (S�irinet al., 2011).

Fig. 2a shows the flocculation efficiency (FE) and concentrationfactor (CF) of N. gaditana cells with pH-induced flocculation toacidic-alkaline values at a settling time of one hour.

No satisfactory flocculation efficiencies were observed at acid-ity-induced pHs (up to pH = 8.7). As the acidity of the algae solu-tion increased, the cell colour visibly changed and becamelighter. Microscopic examinations of acidic samples also show thatunder acidic conditions Nng cells kept their shape and margin;however, the chloroplast shrank (see Figs. S2b; S3a and b in Sup-plementary data). Therefore, we concluded that the culture couldnot tolerate sudden and/or drastic changes in pH to acidic values.

When the pH of the culture was induced to alkaline pH values,the media started to take on a coarse appearance and this was fol-lowed by spontaneous sedimentation. The rapid increase in pHcul-

ture triggered floc formation. A threshold pH, at which flocculationstarted to increase significantly, was reached. Settling was rapidclose to pHthreshold values, although FE was low. At the thresholdpH of Nng cells (pHthreshold = 9.70), flocculation efficiency (FE) in-creased to 89% with a settleable solid volume fraction (SSVF) of

0.124 in 10 min (Table 1c). The CF was highest (6.65) not only be-cause of an increase in FE, but also because of a decrease in SSVF.Below the pHthreshold value, FE and CF were both lower, whereasabove the pHthreshold FE was higher and CF lower. After the pHthresh-

old had been reached, it was difficult to increase the alkalinity of thesolution. Higher flocculation efficiencies were observed up topH = 11, although sedimentation rates were slower.

The height of the interface was recorded at regular time inter-vals. The sedimentation rates (SR) were compared betweenpHthreshold and pH = 11.0 (Fig. 2b). At the threshold pH (pH = 9.7)the sedimentation rate (SR) was as high as 119 cm h�1, but whenpH was increased to 11, SR decreased to 6.3 cm h�1 (Table 1c).

The cells were in good condition after treatment with alkalinity-induced flocculation. It was clear to the naked eye that the Nngcells maintained their colour. Microscopic examination showedthat they kept their shape and the cells were also in better condi-tion than those harvested at acidic pHs (see Figure S2d and e inSupplementary data).

Pht and Nng cultures reach high alkaline pH during the expo-nential growth phase due to photosynthesis (CO2 consumption).Therefore, threshold pH can be used to achieve autoflocculationby adjusting the air or CO2 supply. There must be a certain specificthreshold pH to trigger autoflocculation for each species and med-ium. These findings are parallel to those of Spilling et al. (2011) forPht and of Vandamme et al. (2012) for Chlorella vulgaris. However,if artificial pH manipulation is used to reach alkaline pH, the envi-ronmental impact of that alkalinity on the resulting waste watermust also be considered (S�irin et al., 2011).

3.1.2. Sedimentation with AS and PAC flocculationAluminium sulphate (alum; AS) and poly-aluminium chloride

(PAC) were tested for use in the harvesting of Nng species.Fig. 3a and b show the changes in the flocculation efficiency (FE)

of Nng cells using AS and PAC, respectively. The change in pH dueto the addition of the flocculants is also shown in the figure. Sam-ples were taken after 15 min; 30 min at 1/3 h and 2/3 h and after2 h of settling at 2/3 h.

After 15 min of settling time, no efficient flocculation was foundfor the AS samples, especially above 40 mg L�1 (ppm). The FEs ofthe AS samples did not show any great differences between sam-pling heights either. It seemed not enough time had passed for par-ticles to appear. Very few flocs settled to the bottom of the tubes.When the settling time in the AS experiments was doubled, the in-crease in FEs was satisfactory. For AS concentrations above 50 ppm,2 h of prolonged settling time was also tested. No great differencesin FEs were observed between settling times of 30 min and 2 h ateither sampling height.

AS has an optimum pH range of 4.0–7.0, whereas the optimalcoagulation with AS takes place at pH values between 5 and 7(Wang et al., 2005). PAC is a flocculant with a wide pH adaptionrange of 5.0–9.0 and its flocculation efficiency is better (between6 and 7.5) (S�irin et al., 2011). The optimum dosages found for theflocculants PAC and AS indicate that these final pH values are alsosuitable for Nng cells.

The optimum concentration for AS flocculant with Nng cellswas found to be 20 ppm, regardless of the settling time. At30 min settling time, the FE was around 70%. The FE for an AS con-centration of 20 ppm at 15 min settling time was 30% lower than ata settling time of 30 min.

The FEs of samples taken at 1/3 h after 15 min settling timewere clearly different, especially above 20 ppm. When the settlingtime was doubled (30 min), and then extended to 2 h, no great dif-ferences were observed in the FEs at any of the PAC concentrationsat any sampling height. Therefore, it seems that 30 min of settlingtime is sufficient to determine the optimum concentration of PACfor the flocculation efficiency of the Nng species studied. With

a

b

Fig. 2. (a) Effect of pH change on flocculation efficiency and concentration factor (N) of N. gaditana culture @1 h settling. (b) Graph of (hclarified water/hculture) ratio versussedimentation time to compare sedimentation rate for N. gaditana culture. Data are represented with ±standard error (SE) (n = 3).

S. S�irin et al. / Bioresource Technology 132 (2013) 293–304 297

PAC concentrations between 10 and 20 ppm, FEs tripled for the15 min samples. However, no great difference was found in the30 min samples. Results were most satisfactory with the additionof 20 ppm of PAC and the final pH range was between 6.5 and7.6 as it was in the Pht culture (S�irin et al., 2011).

In the Nng cultures studied, the optimum results for PAC de-pended on the choice of settling time. At 10 ppm of PAC, 30 minsettling time was found to be the optimum time with an algaerecovery range of around 55%, whereas for 15 min settling time,20 ppm was a better choice with an algae recovery range of around60%.

The use of polymerised forms of aluminium (e.g. PAC) has be-come more common because they perform better at higher chargedensities than AS. Additionally, PAC and polyaluminium sulphate(PAS) often result in a decrease in coagulant doses and the associ-ated production of solids (Zhao, 2003). PAC also produces strongerand more rapidly settleable flocs than AS at equivalent dosages(Gregory and Rossi, 2001).

Colour changes in the concentrated algae cells were observed atflocculant concentrations of over 90 ppm for both PAC and AS. Thisis most probably due to the low pH caused by flocculant additionsas observed previously in Pht samples (S�irin et al., 2011). At allother concentrations, the cells maintained their colour. On the ba-

sis of microscopic examinations, the chloroplasts seemed normal(see Supplementary data Fig. 2f and g).

For Nng, PAC yielded moderately faster sedimentation than ASwith better concentration factors and FEs. However, as noted in aprevious study (S�irin et al., 2011), a downstream flocculant re-moval process is probably required for both flocculants. Otherwise,the presence of flocculants further downstream, and extractionand/or fuel conversion processes must also be understood andchecked. Moreover, if the intention is to recycle the waste waterresulting from the process, the effect on culture growth and lipidproduction must be clarified (S�irin et al., 2011).

3.1.3. Sedimentation with chitosan flocculationChitosan, a cationic polyelectrolyte, was tested for use in har-

vesting Nng cultures. Fig. 3c shows the FE and CF of Nng cells ver-sus pH adjustments to different values after the addition of 30 ppmof chitosan. OD measurements were taken for 1/3 h samples. Atacidic and slightly acidic pHs, flocculation was not efficient, whichled to results being unsatisfactory when chitosan was used alone.After the addition of chitosan, acidic and slightly acidic pH adjust-ments also failed to provide adequate FEs. However, when pH wasadjusted so that it was higher than pHculture, efficiency increasednoticeably.

Table 1cResults for harvesting of N. gaditana by inducing the pH to alkaline ([height/diameter]ratio � 7).

pH Max pH induced flocculationefficiency @1 h (%)

Settleable solid volume fraction(SSVF) @10 min (hf/ho)

Max SSVF @1 h(hf/ho)

Concentration Factor(CF) @10 min

Cell conditionafter flocculation

Sedimentationrate (cm/h)

Cellwall

Chloro-plast

9.70 90.61 0.124 0.08 6.65 U U 1199.90 98.78 0.200 0.104 4.20 U U 96

10.16 98.88 0.328 0.142 2.23 U U 7810.25 99.07 0.604 0.182 0.07 U U 5411.00 1.20 0.979 0.723 0.00 U U 6.3

298 S. S�irin et al. / Bioresource Technology 132 (2013) 293–304

When the pH of the solution was induced to alkaline pH valueshigher than that of pHculture after the addition of 30 ppm of chito-san, the algae cells started to flocculate. An appreciable incrementin flocculation efficiency was observed. The highest FEs and CFswere reached by increasing the pH to 9.9 at both sampling heights.We therefore decided to conduct further experiments to find theoptimum chitosan concentrations at an adjusted pH level of 9.9.

Determination of optimum concentrations of chitosan: The floccu-lation efficiencies of different concentrations of chitosan at ad-justed pH (�9.9) are shown in Fig. 3d. Samples were taken at 1/3and 2/3 h after 15 and 30 min of settling.

The flocculation efficiency clearly increased at a chitosan con-centration of 10 ppm. However, FEs and CFs were highest at30 ppm. Additionally, at chitosan concentrations above 30 ppm, re-moval efficiencies and concentration factors did not significantlyimprove. Only at 200 ppm of chitosan was a settling time of15 min insufficient to yield efficient CFs.

No conspicuous changes were observed in cell colour (via thenaked eye) or the chloroplast of the cells (by microscope) (see Sup-plementary data Fig. S2g).

Chitosan flocculation presumably takes place by charge neutral-isation and bridging between algal cells by chitosan chains, as inthe case of other polyelectrolytes (Divakaran and Sivasankara Pil-lai, 2002). As expected, an analysis of the Ca and Mg ion concentra-tions in chitosan samples (Table 2) also suggests that theflocculation mechanism does not depend on the Mg or Ca contentof the medium in alkalinity-induced flocculation.

At chitosan concentrations of above 30 ppm, no significant in-crease was found in FE as the dosage was increased, but rather aplateau was reached. Kaseamchochoung et al. (2006) concludedthat the FE of chitosan depends on its characteristics and the pHand ionic strength of the medium. Our results also suggest thatthe FE of Nng cells with chitosan is more affected by ionic strengththan by flocculant concentration in the same way as the FE is af-fected in Pht (S�irin et al., 2011). Nevertheless, unlike previous Phtresults, the FE of Nng cells did not decrease when the concentra-tion of chitosan was increased to above 90 ppm. Similarly, no dis-tinct difference was found in SSVF after optimum chitosan dosageshad been reached. We might conclude that chitosan concentrationsexceeding the optimal dosage keep the charges stabilised but donot overload them because of a higher initial algae concentrationof N. gaditana. This result also indicates that the type of harvestingmicroorganism also has an important effect on FE (S�irin et al.,2011).

Chitosan is advantageous as a flocculant for waste water treat-ment because it yields high FEs and is known to be low-toxic, bio-compatible and biodegradable. Although chitosan flocculation haswell defined uses in wastewater systems, there is a gap betweenresearch results and industrial algae harvesting applications usingchitosan, most probably due to the price of chitosan itself. How-ever, the chitosan used in the harvesting process does not needto be very high quality. Additionally, flocculation with chitosan al-lows the concentrated biomass to be used without hindering other

processes, which provides a problem-free final product and en-ables the waste water to be directly used via recycling. Althoughthis is beyond the scope of this study, further research is neededif the effects on cultivation of recycled water after flocculationusing chitosan are to be understood.

Several authors (Lavoie and de la Note, 1983; Divakaran andSivasankara Pillai, 2002) have published similar optimum resultsfor chitosan concentrations between 20 and 40 ppm with differentadjusted pHs for several species. In our opinion, this supports theimportance of the pH in chitosan flocculation.

Chitosan tends to be avoided as a flocculant because of its highcost, even though its benefits are widely recognised. Nevertheless,the chitosan used in harvesting processes to maintain the qualityof the algae quality need not be of a very high quality. This makesthe harvesting cost lower than expected.

3.2. Analysis of the pre-concentrated samples of N. gaditana versuspreviously studied P. tricornutum species

3.2.1. Viscosity analysisThe rheological properties of the concentrated algae suspen-

sions of two microalgal species, Pht and Nng, were studied afterbeing harvested by different treatment methods (Table 1b).

For all the samples tested, the apparent viscosity was constantwhen measured at different rotational speeds corresponding to dif-ferent shear rates (Fig. 4a and b). Our experimental results estab-lish that the samples showed the characteristics of a Newtonianfluid in the range of shear rates tested, where the constant of pro-portionality relating the shear rate and shear stress is viscosity. Itwas calculated as:

s ¼ k _c

where s is the shear stress (Pa), _c is the shear rate (s�1) and k is theconstant of viscosity (Pa s).

The samples measured had a dynamic viscosity range ofapproximately 2–2.4 � 10�3Pa s at 30 �C after the pre-concentra-tion process, whereas the dynamic viscosity of seawater is givenas 0.87 � 10�3Pa s (temperature = 30 �C; salinity:38ppt). The linearregression determination coefficients for the viscosity (R2) of thesamples were higher than 0.96.

Samples C, GM or FC treated with the same harvesting methodsdisplayed no major differences in viscosity (Table 2). Additionally,the kinematic viscosity data of residual solutions after the algaehad been harvested were also monitored and compared with kine-matic viscosity data of the culture medium and no differences werefound in viscosities before or after treatment (data not shown). Sim-ilarly, in alkalinity-induced flocculation experiments, no markeddifference in viscosity was observed at pHthreshold and pH = 11.

The viscosities of the supernatant of the algal culture suggestthat no viscous substances such as extracellular polysaccharide(EPS) were excreted by the species during the pre-concentrationprocess due to stress. The microalgae species used in this studyare all unicellular species with negligible EPS production. Hence,

a

b

c

Fig. 3. Flocculation efficiency of N. gaditana (a) with PAC, (b) with AS, (c) Changes in the flocculation efficiency (FE) and concentration factor (CF) (N) of N. gaditana flocculatedwith 30 ppm of chitosan according to pH @15 min settling time, (d) Flocculation efficiency of N. gaditana versus chitosan concentration at pH = 9.9⁄ (e) CF and SSVF accordingto chitosan concentration at adjusted pH = 9.9 @30 min settling time(FEs of samples taken from the top 2/3rds of the tubes were used to calculate CF). Data a–d arerepresented with ±SE (n = 3). The abbreviations 1/3 and 2/3 represent one-third or two-thirds from the top of the tubes.

S. S�irin et al. / Bioresource Technology 132 (2013) 293–304 299

the effect of EPS was not measurable in our viscosity analyses. Be-low a solid concentration of about 4% by weight, most sludges ex-hibit Newtonian fluid behaviour: that is, there is a linearrelationship between shear stress and shear rate where the con-stant of proportionality, l, is the viscosity of the fluid (i.e. water).Algae cultures are generally accepted to behave like Newtonianfluids (Clementi and Moresih, 1998). During batch culture, the rhe-ological behaviour of the medium can change from Newtonian tonon-Newtonian due to EPS production during growth; however,in long-term cultures, the rheology reverts to Newtonian as a resultof the hydrolysis of the polymer (Lupi et al., 1991).

Newtonian fluids are very similar to water, making both pump-ing and mixing relatively easy. Additionally, the mixing times forNewtonian fluids are lower than those of non-Newtonian fluids.

Although the rheology of cell cultures is influenced by cell mor-phology (for example, size, shape and aggregation) as well as bio-mass concentration, after the pre-concentration process nosignificant differences in viscosities were observed in either Nngor P. tricornutum, regardless of the method used. The reason for thismight be that the cultures were grown with the same seawater,that the initial concentration of microalgae cultures was lower orthat the viscosity of Mg(OH)2 was low (even at high concentrationsit flows like water).

To measure floc strength, we applied different agitation speeds(max 1500 rpm) and shear rates (2700 s�1). The settling efficien-cies of the samples were checked after the viscosity measurementshad been taken. No precise differences were observed in terms ofsettling rates. The turbidity of the residual solutions was

d

e

Fig. 3. (continued)

Table 2Dynamic viscosities and Mg and Ca contents of samples.

Species Method of concentration Sample Viscosity* (lPas) R2** Mg content�,� (%) Ca Content�,� (%)

Phaedactylum tricornutum pHthreshold C 2.1 0.9867 84 ± 0.91 97 ± 2.86GM 2.1 0.9792 97 ± 1.77 97 ± 1.92FC 0.9898 94 ± 1.85 97 ± 1.29

pH = 11 C 2.2 0.9840 0 ± 0.00 86 ± 1.29GM 0.9898 5 ± 1.40 81 ± 5.92FC 2.3 0.9681 0.32 ± 0 82 ± 4.34

PAC C 2.2 0.9939 98 ± 1.60 98 ± 1.84GM 0.9898 98 ± 1.80 98 ± 1.80FC 0.9867 98 ± 1.85 98 ± 1.80

AS C 2.3 0.9898 98 ± 1.60 98 ± 0.00GM 2.2 0.9968 98 ± 1.60 98 ± 0.80FC 0.9968 98 ± 1.60 98 ± 1.80

Chitosan C 2.2 0.9730 97 ± 0.91 94 ± 5.75GM 0.9898 98 ± 0.40 95 ± 1.60FC 0.9898 98 ± 1.67 97 ± 0.45

Nannochlorpsis gaditana pHthreshold C 2.1 0.9791 87 ± 1.91 98 ± 0.00GM 2.0 0.9970 92 ± 2.50 99 ± 0.31FC 0.9968 90 ± 3.28 97 ± 0.00

pH = 11 C 2.2 0.9774 0 ± 0.00 77 ± 1.59GM 2.1 0.9923 0.5 ± 0.68 72 ± 5.35FC 2.4 0.9748 0.7 ± 0.91 77 ± 1.81

PAC C 2.0 0.9912 97 ± 0.70 98 ± 0.88GM 0.9730 98 ± 0.40 98 ± 0.88FC 0.9939 97 ± 0.40 98 ± 0.88

AS C 2.1 0.9730 95 ± 2.60 98 ± 0.88GM 0.9856 97 ± 1.60 98 ± 0.00FC 0.9800 97 ± 1.60 97 ± 1.60

Chitosan C 2.2 0.9843 98 ± 0.93 97 ± 1.31GM 2.1 0.9886 97 ± 1.60 97 ± 0.47FC 0.9800 97 ± 1.53 97 ± 1.31

* Viscosity measurements were done at 30 �C with concentrated algae samples after harvesting.** R2 is the linear regression determination coefficient of viscosity.� Mg and Ca ion measurements were done with residual solution samples after harvesting.� Mg and Ca ions percentages were calculated by comparison to culture medium ion contents.

300 S. S�irin et al. / Bioresource Technology 132 (2013) 293–304

S. S�irin et al. / Bioresource Technology 132 (2013) 293–304 301

satisfactory, as it was before the procedure. Microscopic observa-tions backed these settling results.

3.2.2. Particle size analysisCoagulation/flocculation is a well-known process used in water

treatment to remove suspended particles by combining smallerparticles into larger aggregates. The mechanisms involved in thiscan include the compression of the diffuse (double) layer, chargeneutralisation, enmeshment or ‘sweep floc’, and adsorption andinterparticle bridging (Tchobanoglous and Schroeder, 1987). Thesemechanisms result in flocs of different sizes, strengths and struc-tures. However, it is difficult to state and characterise the sizeand shape of flocs because the different generation mechanismsmean they are so irregular. Particle shape affects the behaviourof aggregated particles, particularly in terms of collision efficiency(Wiesner 1992) and settling rates (Li and Logan, 1997). Particles ofdifferent sizes settle at different rates. The larger particles will set-tle more rapidly than the smaller ones (assuming similar densityand shape).

Table 3 shows the variations in the particle diameter of the con-centrated samples after harvesting using the methods shown in

(a)

(c)

(d)

Fig. 4. (a) Example of shear stress (s) as a function of shear rate ( _c) for P. tricornutum and30 �C. (c) Percentage particle size distribution of cultures after harvesting methods P. tric

Table 1b. The particle size distribution graph for Pht and Nng sam-ples can be seen in Fig. 4c and d, respectively.

On average, the diameter of the Pht flocs was greater than thatof the Nng flocs and, because of the larger size of this species, thestandard deviation was higher. Coulter methods use the equivalentresistance diameter, which is the diameter of a spherical particlethat has the same resistance as the particle tested. Therefore,Nng samples yielded results with standard errors that were lowerthan expected. The sphericity of Nannochloropsis cells is higherthan that of Phaedactylum cells, so the standard errors were lower.

The size distribution graph for the Pht samples showed that thepHthreshold samples contained a higher percentage of particles withdiameters in the range 50–80 lm whereas pH = 11 samples con-tained a higher percentage of particles with shorter diameters(30–40 lm). For samples harvested at pH = 11, the Dfs wereapproximately 10% smaller.

The Dfs of Nng were in the range 38–44 lm (i.e. smaller thanthat found for the Pht samples). The most noticeable differencein the particle size distribution of Nng was that no particle wasover 70 lm (especially in pHthreshold). The size distribution graphfor this species showed that pHthreshold samples contained a higher

µm)

µm)

(b)

(b) N. gaditana cultures of growth medium (GM) samples harvested at pHthreshold atornutum and (d) N. gaditana. Data are represented with ±standard error (SE) (n = 3).

302 S. S�irin et al. / Bioresource Technology 132 (2013) 293–304

percentage of particles in the 50–70 lm diameter range, whereaspH = 11 samples contained higher percentages of lower range par-ticles (30–40 lm).

The settling properties of alkalinity induced flocculation for Phtand Nng flocs support the size distribution graph. The pHthreshold

samples showed higher SRs than pH = 11 samples where the FEswere almost the same with high CFs. The mechanism of alkalinityinduced flocculation is the same for both species which is enmesh-ment of algae cells into magnesium hydroxide particles. Alkalinity-induced flocculation can result from an increase in medium ionicstrength, which causes double layer compression. By linking ourviscosity-floc strength results, it is unlikely that this mechanismwas involved in our experiments, as the change in ionic strengthcaused by pH increase is limited. As Vandamme et al.(2012), wealso credit that flocculation in our experiments was caused bythe mechanism of charge neutralisation.

The mean particle size of magnesium hydroxide seems to de-crease as the pH value of the solution and the Mg2+ concentration in-crease under particular experimental conditions (Xu and Deng,2006). Therefore, for pH = 11 alkalinity-induced flocculation sam-ples contained higher percentages of lower range particles. In addi-tion, the Dfs of the flocs of Pht was found higher than Nng mostprobably only due to this species has bigger spherical particle diam-eter than Nng species.

No distinct differences were found in the Dfs of the C and FC sam-ples. However, after a detailed examination, some diversity in the par-ticle size distribution was observed. For instance, for Pht samples, atpH = 11, the C sample contained a higher percentage of particles be-tween 40 and 50 lm while the FC sample contained higher percent-ages in the 30–40 lm range. When compared to the GM particles,they were found to have a lower percentage of 40–50 lm particles aswell. The Dfs of the GM samples were a little smaller. The same nuancesin particle ranges were also observed in the Nng samples. C and FC sam-ples contained microalgae cells and other materials in the culture med-ium (e.g. extracellular organic matter) which might let the magnesiumhydroxide neutralise the negative surface charge of them (Semerjianand Ayoub, 2003) resulting a little higher Dfs.

For AS, PAC and chitosan flocculant treatments of Pht atoptimum concentrations, Dfs were measured as 40.3 ± 3.07,

Table 3Particle diameter variations of concentrated samples after harvesting of cultures with diff

Species Sample Treatment Mean diameter(lm)

Standa(lm)

Phaeodactylumtricornutum

GM pH = 11 41.5 3.64C 43.1 5.40FC 43.1 6.41GM pHthreshold 47.8 6.45C 47.7 6.60FC 47.3 7.43C pH = 9.30 45.8 8.34

pH = 9.54 46.6 9.33AS 40.3 3.07PAC 39.8 3.53Chitosan(1)(pH = 9.85)

39.2 3.18

Chitosan(2)(pH = 9.9)

40.6 3.16

Nannochloropsisgaditana

GM pH = 11 38.5 2.87C 40.5 5.11FC 40.1 2.65GM pHthreshold 40.2 5.06C 42.9 3.99FC 41.3 6.78C AS 40.00 4.50

PAC 44.32 8.18Chitosan(1)(pH = 9.74)

43.00 4.90

Chitosan(2)(pH = 9.9)

44.10 6.50

39.8 ± 3.53 and 40.6 ± 3.16 lm, respectively. The Dfs of Nng sam-ples treated with AS, PAC and chitosan flocculants at optimum con-centrations measured 40.0 ± 4.50, 44.32 ± 8.18 and 44.10 ±6.50 lm, respectively with high SEs. The diameters were similarbecause all three flocculants had almost the same percentage ofparticles in the 30–40 lm range. However, chitosan had more par-ticles in the 40–50 lm range, which resulted in faster sedimenta-tion rates and higher CFs than for other flocculants which mightindicate the mechanisms of flocculation are different. By alumin-ium ions, metal hydroxides bind to the negative surface of the mic-roalgal cells and destabilise the microalgal suspension by chargeneutralisation resulting in enmeshment of microalgae and insolu-ble precipitates Insoluble metal hydroxides can destabilise themicroalgal suspension through a mechanism known as sweepingflocculation (Duan and Gregory, 1996). In the case of chitosan, itcan easily react with microalgal particles by interparticle bridgingbehaviour and may even remove smaller particles by an enmesh-ment mechnaism where the interparticle bridging capabilityshould be the main mechanism (Chung et al., 2005).

3.2.3. Ca and Mg ion analysisThe calcium and magnesium ion concentrations in the residual

solutions after harvesting were monitored and compared with theion content of the culture medium so that the effect of Mg/Ca ionson flocculation mechanisms with different treatment methodscould be estimated (Table 1c).

Several studies have mentioned that under alkaline conditionstwo major reactions are effective: the precipitation of calcium car-bonate (CaCO3) and the precipitation of magnesium hydroxide(Mg(OH)2), depending on the primary particles and the ions con-tained in the solution (Vandamme et al., 2011).

Our analysis of Ca and Mg ion concentrations with alkalinity-in-duced flocculation substantiate the fact that the Mg ion is the pro-tagonist in the flocculation mechanism at alkaline pH. AtpHthreshold, magnesium ion concentrations were lower than in theculture medium, although no distinctive differences in the calciumion concentration were found (Table 2) in comparison with the cul-ture medium. The flocs are assumed to be a mixture of Mg+2 ions

erent methods.

rd error (SE) Coefficient variation(%)

d10 ± SE(lm)

d50 ± SE(lm)

d90 ± SE(lm)

17.5 34.8 ± 0.07 39 ± 0.30 52.4 ± 1.2025.0 35.3 ± 0.07 40.3 ± 0.30 54.2 ± 0.4625.7 35.0 ± 0.14 39.0 ± 1.10 54.7 ± 1.5327.0 35.5 ± 0.07 43.6 ± 0.48 66.4 ± 0.5327.6 35.4 ± 0.20 43.5 ± 0.52 67.0 ± 0.4731.4 35.3 ± 0.16 41.9 ± 1.16 66.80 ± 1.525.7 34.8 ± 0.03 42.1 ± 0.27 62.6 ± 0.8427.0 35.9 ± 0.20 45.1 ± 0.61 63.8 ± 0.3814.9 34.8 ± 0.07 38.2 ± 1.13 48.8 ± 4.1217.7 34.8 ± 0.02 37.6 ± 0.01 47.0 ± 0.7816.2 34.8 ± 0.02 37.4 ± 0.21 45.4 ± 0.27

15.5 35.0 ± 0.02 38.7 ± 0.31 48.6 ± 0.29

25.6 34.8 ± 0.02 36.7 ± 0.13 42.3 ± 0.2717.8 34.9 ± 0.03 38.4 ± 0.05 47.8 ± 0.9113.2 35.0 ± 0.02 38.6 ± 0.08 47.8 ± 0.3117.8 34.9 ± 0.07 38.0 ± 0.16 47.7 ± 0.1718.6 35.3 ± 0.05 40.7 ± 0.16 53.5 ± 0.3723.2 35.0 ± 0.03 38.6 ± 0.27 49.3 ± 0.6316.0 34.8 ± 0.07 38.2 ± 1.13 52.6 ± 4.2023.0 35.4 ± 0.02 40.3 ± 0.01 56.3 ± 0.8019.0 34.9 ± 0.10 38.8 ± 0.14 55.1 ± 1.32

20.9 34.9 ± 0.12 40.5 ± 0.66 57.7 ± 2.37

S. S�irin et al. / Bioresource Technology 132 (2013) 293–304 303

generated by the reaction of the NaOH, which causes the precipita-tion of Mg(OH)2.

Mgþ2 þ 2OH� !MgðOHÞ2 #

When the culture pH was induced to pH = 11, no magnesiumions were detected, and the calcium ion concentration decreased.All magnesium ions converted into Mg(OH)2 and CaCO3 startedto form. This resulted in a time lag for the initiation of settlingand lower CF and SR values at higher alkaline pH at which Mgand Ca precipitated sediment with algae cells. The production of(OH)� ions, especially at pH values higher than pHthreshold, changesthe inorganic carbon equilibria in the seawater and facilitates thefollowing buffering reaction (Turnbull and Ferriss, 1986);

OH� þHCO�3 $ H2Oþ CO2�3

It is also why it is difficult to increase the alkalinity of the solu-tion. As a result of the continuing increase in pH, CaCO3 also pre-cipitates at higher alkaline pH values.

Caþ2 þ CO�23 ! CaCO3 #

The pH is buffered by a set of reactions that take place betweencarbon dioxide and water. The formation of the alkaline scalesCaCO3 and Mg(OH)2 strongly depends on temperature, pH, the re-lease rate of CO2 and the concentrations of HCO�3 , CO2�

3 ,Ca2+, andMg2+ ions.

Because the concentrations of magnesium ions in the culturemedia were approximately the same, it is not surprising that sim-ilar results were found for both Pht and Nng species. This supportsthe hypothesis that the Mg ion is the triggering ion in the floccula-tion mechanism.

After both algae species at pHthreshold had been harvested, theresidual solutions contained lower concentrations of magnesiumions than the culture medium, although no distinctive decreasewas found in the concentration of calcium ions. When pH was in-duced to pH = 11 for harvesting, no magnesium ions were detectedand the calcium ion concentrations in the residual samples de-creased in both algae species compared to initial Ca concentrations(Table 2).

No changes in Mg or Ca ion concentrations were detected in theresidual solutions in the algae cultures harvested with AS, PAC orchitosan flocculants (at the concentrations given in Table 1b). Un-der both conditions, neither the GM nor the FC samples demon-strated a distinct change in Ca or Mg ion concentration.

4. Conclusion

The choice of pre-concentration method in algae depends onthe target product. Our experiments with N. gaditana showed thatalkalinity-pH is a promising pre-concentration method althoughflocculation with chitosan gives better results (higher FE and CF).Viscosity results show that pre-concentration processes make bothpuımping and mixing easier because of the Newtonian behaviourof the samples. Particle size analysis supports settling propertiessufficiently, however further research on density and mechanismof the flocs is recommended. The analysis of Ca and Mg ion concen-trations substantiates the fact that the Mg ion is the protagonist inthe alkalinity-induced flocculation mechanism.

Acknowledgements

This research was supported by the project ENE2011-22761 ofthe Spanish Ministry of Science and Innovation, and the ‘BiomassFuels’ project of the Excma. Diputació de Tarragona. The researchwas also supported by the European Regional Development Funds(ERDF, FEDER Programa Competitividad de Catalunya. 2007-

2013).The authors would like to thank Maria Pilar Rey Valera fromIREC and research group of the Aquatic Ecosystems from IRTA (SantCarles de la Rapita, Tarragon, Spain) for valuable work with the al-gal cultures.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.biortech.2013.01.037.

References

Chen, C.Y., Yeh, K.L., Aisyah, R., Lee, D.J., Chang, J.S., 2011. Cultivation,photobioreactor design and harvesting of microalgae for biodiesel production:a critical review. Bioresour. Technol. 102, 71–81.

Christenson, L., Sims, R., 2011. Production and harvesting of microalgae forwastewater treatment, biofuels, and bioproducts. Biotechnol. Adv. 29, 686–702.

Chung, Y.C., Li, Y.H., Chen, C.C., 2005. Pollutant removal from aquaculturewastewater using the biopolymer chitosan at different molecular weights. J.Environ. Sci. Health A Tox. Hazard. Subst. Environ. Eng. 40 (9), 1775–1790.

Clementi, F., Moresih, M., 1998. Rheology of alginate from Azotobacter vinelandii inaqueous dispersions. J. Food Eng. 36, 51–62.

Danquah, M.K., Gladman, B., Moheimani, N., Forde, G.M., 2009. Microalgal growthcharacteristics and subsequent influence on dewatering efficiency. Chem. Eng. J.151, 73–78.

Divakaran, R., Sivasankara Pillai, V.N., 2002. Flocculation of algae using chitosan. J.Appl. Phycol. 14, 419–422.

Duan, J., Gregory, J., 1996. Influence of soluble silica on coagulation by aluminiumsulphate. Colloids Surf. A 107, 309–319.

Gibbs, R.J., 1982. Floc breakage during HIAC light blocking analysis. Environ. Sci.Technol. 16, 298–299.

Gregory, J., Rossi, L., 2001. Dynamic testing of water treatment coagulants. WaterSci. Technol. Water Supply 1, 65–72.

Kaseamchochoung, C., Lertsutthiwong, P., Phalakornkule, C., 2006. Influence ofchitosan characteristics and environmental conditions on flocculation ofanaerobic sludge. Water Environ. Res. 78, 2210–2216.

Lavoie, A., de la Note, J., 1983. Harvesting microalgae with Chitosan. J. WorldMaricult. Soc. 14, 685–694.

Li, X., Logan, B.E., 1997. Collision frequencies between fractal aggregates and smallparticles in a turbulently sheared fluid. Environ. Sci. Technol. 31, 1237–1242.

Lubián, L.M., 1982. Nannochloropsis gaditana sp. nov., una nueva Eustigmatophyceaemarina. Lazaroa 4, 287–293.

Lupi, F.M., Fernandes, H.M.L., Sa Correia, I., Novais, J.M., 1991. Temperature profilesof cellular growth exopolysaccharide synthesis by Botryococcus braunii. J.Phycol. 3, 35–42.

Macias-Sanchez, M.D., Mantell, C., Rodriguez, M., Martinez de la Ossa, E., Lubian,L.M., Montero, O., 2005. J. Food Eng. 66, 245–251.

Natural Algal Biofuels Technology Roadmap. <www1.eere.energy.gov>.Nurdogan, Y., Oswald, W.J., 1995. Enhanced nutrient removal in high-rate ponds.

Water Sci. Technol. 31, 33–43.Papazi, A., Makridis, P., Divanach, P., 2010. Harvesting Chlorella minutissima using

cell coagulants. J. Appl. Phycol. 22, 349–355.Salim, S., Vermuë, M.H., Wijffels, R.H., 2011. Harvesting of microalgae by bio-

flocculation. J. Appl. Phycol. 23, 849–855.Semerjian, L., Ayoub, G.M., 2003. High-pH–magnesium coagulation–flocculation in

wastewater treatment. Adv. Environ. Res. 7, 389–403.Simionato, D., Sforza, E., Carpinelli, E.C., Bertucco, A., Giacometti, G.M., Morosinotto,

T., 2011. Acclimation of Nannochloropsis gaditana to different illuminationregimes: effects on lipids accumulation. Bioresour. Technol. 102, 6026–6603.

S�irin, S., Trobajo, R., Ibanez, C., Salvadó, J., 2011. Harvesting the microalgaePhaeodactylum tricornutum with polyaluminum chloride, aluminium sulphate,chitosan and alkalinity-induced flocculation. J. Appl. Phycol. 24, 1067–1080.

Spilling, K., Seppälä, J., Tamminen, T., 2011. Inducing autoflocculation in the diatomPhaeodactylum tricornutum through CO2 regulation. J. Appl. Phycol. 23, 959–966.

Sukenik, A., Shelef, G., 1984. Algal autoflocculation–verification and proposedmechanism. Biotechnol. Bioeng. 26, 142–147.

Tchobanoglous, G., Schroeder, E.D., 1987. Water Quality: Characteristics, Modelling,Modification. Addison-Wesley Publishing Co., Reading, Massachusetts.

Tenney, M.W., Verhoff, F.H., 1973. Chemical and autoflocculation of microorganismsin biological wastewater treatment. Biotechnol. Bioeng. 15, 1045–1073.

Turnbull, A., Ferriss, D.H., 1986. Mathematical modelling of the electrochemistry incorrosion fatigue cracks in structural steel cathodically protected in sea water.Corros. Sci. 26, 601–628.

Uduman, N., Qi, Y., Danquah, M.K., Forde, G.M., Hoadley, A., 2010. Dewatering ofmicroalgal cultures: a major bottleneck to algae-based fuels. J. Renew. Sustain.Energy 2, 012701.

Vandamme, D., Foubert, I., Meesschaert, B., Muylaert, K., 2011. Flocculation ofmicroalgae using cationic starch. J. Appl. Phycol. 22, 525–530.

304 S. S�irin et al. / Bioresource Technology 132 (2013) 293–304

Vandamme, D., Foubert, I., Fraeye, I., Meesschaert, B., Muylaert, K., 2012.Flocculation of Chlorella vulgaris induced by high pH: role of magnesiumand calcium and practical implications. Bioresour. Technol. 105, 114–119.

Wang, L.K., Hung, Y.T., Shammas, N.K., 2005. Physicochemical treatment processes.Handbook of Environmental Engineering, vol. 3. Humana Press, New Jersey, p.113.

Wiesner, M.R., 1992. Kinetics of aggregate formation in rapid mix. Water Res. 26,379–387.

Xu, H., Deng, X., 2006. Preparation and properties of superfine Mg(OH)2 flameretardant. Trans. Nonferrous Met. Soc. China 16, 488–492.

Zhao, Y.Q., 2003. Correlations between floc physical properties and optimumpolymer dosage in alum sludge conditioning and dewatering. Chem. Eng. J. 92,227–235.