Bioresource Technology 158 (2014) 217–224

Contents lists available at ScienceDirect

Bioresource Technology

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

An ultra-low energy method for rapidly pre-concentrating microalgae

http://dx.doi.org/10.1016/j.biortech.2014.02.0330960-8524/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 206 296 5535.E-mail addresses: teodora@seattleu.edu (T.R. Shuman), mason@seattleu.edu (G.

Mason), marsolem@seattleu.edu (M.D. Marsolek), liny2@seattleu.edu (Y. Lin),reeved@seattleu.edu (D. Reeve), schachta@seattleu.edu (A. Schacht).

Teodora Rutar Shuman ⇑, Gregory Mason, Michael D. Marsolek, Yizhou Lin, Daniel Reeve,Alexander SchachtSeattle University, 901 12th Ave, P.O. Box 222000, Seattle, WA 98122, USA

h i g h l i g h t s

� ECF uses Ni electrodes and 4 s treatment time with pulsed or continuous DC.� Algae rapidly separate from suspension with input energy density of 0.03 kWh/m3.� Max separation after 2 h is 97%; max separation effectiveness is �30%/(kWh/m3).� Rapid separation occurs even if untreated algae are mixed with treated saltwater.� Process does not cause significant cell damage.

a r t i c l e i n f o

Article history:Received 22 October 2013Received in revised form 8 February 2014Accepted 10 February 2014Available online 17 February 2014

Keywords:AlgaeSeparationDewateringElectro-coagulationFlocculation

a b s t r a c t

This study demonstrates that Nannochloropsis sp. can be effectively separated from its growth medium(0.2–0.3 g/L) using electro-coagulation–flocculation in a 100 mL batch reactor with nickel electrodesand a treatment time of only 4 s. Minimum energy density input for effective separation is 0.03 kWh/m3. Both energy input and treatment time are much smaller than reported elsewhere. The process resultsin rapid separation of microalgae (over 90% in 120 min) with minimal damage to algal cells (>90% stillalive after processing). At around 4 V input, algae can be effectively separated even in very low concen-trations. Pulsing is equally effective in separating microalgae as continuous direct current of same mag-nitude and total exposure time. Algae can separate from their growth medium even if the suspensionitself is not treated, but is mixed with treated saltwater with same conductivity. The described methodhas significant advantages including applicability to continuous processing and water reuse.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Microalgae have a potential to become a renewable resource forfuel production that does not compete with food crops and does notuse large amounts of fresh water, energy, or land. Half of biodieselneeds in the US can be supplied with microalgae grown on only2.5% of the total cropland, compared to 24% of cropland neededfor the same amount of biodiesel from palm oil (Chisti, 2007). Mic-roalgae can also be used to produce or supplement a wide variety offuels, including gasoline and kerosene. In addition, microalgae canbe used to produce animal feeds, fertilizers (Demirbash, 2011),food, and many other chemicals (Zeng et al., 2011; Chisti, 2007;Spolaore et al., 2006a) currently produced from crude oil.

Presently, production volumes of microalgae are very low andcosts of growing and processing are prohibitively high, limitingthe commercialization of algae based biofuels (Cheng and Timilsi-na, 2011). For microalgae to become commercially viable it isimperative to reduce costs in all phases of the algae-to-fuel life-cy-cle. Processing of microalgae to fuels and other products typicallyrequires harvesting, dewatering, and extraction of fuel precursors,i.e., oils and carbohydrates (DOE, 2010), or other chemicals. All ofthese processes are extremely resource intensive and unsustain-able for large-scale production (NRC, 2012). In particular, the smallsizes of the microalgae, usually ranging from one to ten microme-ters, and low densities in the growing ponds, on the order of 1 g/L,present significant challenges to the harvesting and dewateringprocesses.

Microalgae harvesting and dewatering are currently accom-plished through centrifugation, flocculation, sedimentation or fil-tration. Centrifugation is commonly used and is estimated toconsume 8 kWh/m3 of microalgal suspension (Danquah et al.,

Fig. 1. Schematic of the experimental equipment with a photo of the algae reactor.

218 T.R. Shuman et al. / Bioresource Technology 158 (2014) 217–224

2009), although recently a commercial developer claims to havemade a centrifuge that operates at 1 kWh/m3 (Evodos, 2013).

Flocculation requires the use of expensive chemicals but theprocess itself may demand zero or very little external energy input(Uduman et al., 2010). Gravity sedimentation requires largeamounts of land to accommodate very slow batch processing. Fil-tration and screening require at least 0.4 kWh/m3 for a low-vibrat-ing screen filter, and increased operation costs for periodicreplacement of the screens (Uduman et al., 2010). Alternative coag-ulation methods use sonication or electrolysis. Zhang et al. (2009)report significant algal removal rates with sonication-enhancedcoagulation using as little as 0.1 kWh/m3. However, Zhang et al.(2006) report that when ultrasound is used alone, without addedcoagulants, the minimum energy input to achieve significant re-moval rate is several orders of magnitude higher.

The work presented in this paper focuses on electro-coagula-tion–flocculation (ECF) for low-energy dewatering or pre-concen-trating of microalgae. Poelman et al. (1997) showed thatelectrolytic flocculation can remove 95% of the algae from suspen-sion by consuming as little as 0.3 kWh/m3 and without addedcoagulants. Other researchers, including (Alfafara et al., 2002;Emamjomeh and Sivakumar, 2009; Aragon et al., 1992; Azarianet al., 2007; Vandamme et al., 2011; Gao et al., 2010; Udumanet al., 2011) have also reported success using ECF. In published pa-pers, the exposure times of algae mediums to direct current are be-tween 2 and 90 min using batch processing. These short processingtimes are advantageous from the applications standpoint. Pearsallet al. (2011) applied this process to a continuous flow device andreported success using flow rates below 5 mL/s. Continuous-flowprocessing was also attempted by Kim et al. (2012). Most processesuse aluminum electrodes, magnetic stirrers to enhance mixing, andmost have been applied to freshwater algae. Vandamme et al.(2011) demonstrated that power consumption needed for ECF islower for marine than for freshwater microalgae and concludedthat is due to higher conductivity of the marine medium. The low-est power consumption reported by Vandamme et al. (2011) is0.2 kWh/kg of recovered algal biomass. While all of the abovemen-tioned processes can remove at least 80% of the algae from a med-ium, even the best of these processes are still far too energy, waterand nutrient intensive to make microalgae based biofuel a viablefuel alternative, as concluded by NRC (2012). The design of a devicethat separates microalgae and its growth medium, has ultra-lowenergy consumption, and allows reuse of water is a necessary stepand is yet to be developed.

The ECF process described in this paper attempts to addresssome of the shortcomings currently associated with other ECF pro-cesses. It is different than its precursors in several aspects:

1) Microalgae suspension is subjected to direct current for onlya few seconds, where the current is either pulsed or appliedcontinuously. The short processing time is favorable for con-tinuous flow processing and desirable for commercialapplications.

2) The process does not use stirrers during electrolysis. This isan advantage because overall energy input is reduced.

3) The process has a high ratio of electrode surface area to algaesuspension volume. This increases the areas for electro-chemical reactions and is conducive to self-induced mixing.

4) Electrodes are made from pure nickel.5) The process is applied directly to microalgae suspended in

its growth medium at various concentrations, and to puresaltwater prior to mixing with algae suspension.

6) The process does not require that any external coagulantchemicals are added during the process.

7) The process uses a simple device design that can be adaptedto continuous flow processing in commercial applications.

This paper describes the equipment and process used to sepa-rate marine microalgae, Nannochloropsis sp., from their growthmedium, and presents results of varying operational parameters,including voltage, initial algae concentration, electrode material,current pulsing scheme, and treated suspension, and how thesechanges affected the efficiency of the process and cell viability.

2. Methods

2.1. Experimental setup and equipment

A schematic of the experimental setup is shown in Fig. 1. Thesetup consists of four major components: an algae reactor, powerelectronics, a control computer, and separation-monitoringequipment.

The algae reactor is a batch reactor with a 100 mL capacity andfour electrode pairs. The reactor walls are made out of acrylic foreasy viewing. Outer dimensions of the acrylic housing are6 cm � 6.5 cm � 9 cm. The electrodes are slid into and held inplace by milled grooves on the inside surfaces of the acrylic walls.The eight electrodes have surface dimensions of 5.0 cm by 7.6 cmand have alternating offsets at the top for electrical contact. Nickelalloy 200 electrodes (>99% pure nickel) were used for all but oneexperiment and stainless steel 316 electrodes were used in a singleexperiment.

Power for the reactor is supplied by a DC power supply rated at40 V and 128 A (Hewlett Packard 6684A 0-40V/0-128A). Powerfrom the supply is controlled using a high power MOSFET(IXFN180N25T) and driver (27425). The MOSFET is configured asa switch to control the voltage supplied to the reactor electrodes.

A computer equipped with a National Instrument data acquisi-tion card and LabView software controls the MOSFET drivers andmonitors the resulting power to the reactor. Power to the reactoris controlled by pulsing the voltage to the reactor plates withpulses that are rectangular. The computer controls the pulse dura-tion, duty cycle and total number of pulses. Pulse amplitude is con-trolled by setting the output on the DC power supply. Current wasmeasured using a Fluke i30 DC/AC Current Probe. Voltage wasmeasured directly using the data acquisition card and a voltage di-vider. A Fluke 196 ScopeMeter was used to independently verifythe current and voltage values and shape of the pulses.

Algae density was measured using Thermo Scientific Scanning10 UV spectrophotometer at 440 nm. This wavelength was foundto give peak absorbance during a wavelength scan performed with

T.R. Shuman et al. / Bioresource Technology 158 (2014) 217–224 219

the same instrument. Data collected using Nannochloropsis sp. (seeSection 2.3) showed a linear relation (R2 = 0.98) between opticaldensity (OD) at 440 nm and the dry mass of algae:ODinAU ¼ 4:1ðdrymassin g

LÞ � 0:1.Algae suspension conductivity was measured using an Omega

CDH-80MS conductivity meter. Algal suspension conductivity, r,increases with decreasing OD linearly for our range of data: r[mS/cm] = 80 � 46 OD [AU].

Algae photos were obtained with Nikon eclipse TE2000-Umicroscope. Cell viability was determined with SYTOX Green Nu-cleic Acid Stain (Invitrogen, Molecular Probes), a dead-cell stainsuitable for use in liquids with high salinity (Veldhuis et al.,1997), and two instruments: a FlowCAM� cytometer equippedwith a 488 nm laser (Fluid Imaging Technologies), and a ZeissLSM 510 NLO Confocal Microscope.

2.2. Microalgae

All tests reported in this paper were performed using Nanno-chloropsis sp. at pond concentrations, unless otherwise stated. Nan-nochloropsis sp. is a marine unicellular alga belonging to the classEustigmatophyceae. Nannochloropsis has been identified as one ofthe most promising photoautotrophic producers of eicosapentae-noic acid (EPA), important x3 polyunsaturated fatty acid for hu-man consumption. Its widest application is in aquaculture, asfeed for rotifers in fish hatcheries (Spolaore et al., 2006b). Chisti(2007) also identified Nannochloropsis as one of the few high-oilproducing microalgae suitable for biodiesel production. Zhukovaand Aizdaicher (1995) claimed that 75% of the total fatty acidsare comprised of only three components: 16: 0, 16: 1 (n-7) and20: 5 (n-3), which makes its processing for oil production simpler.Nannochloropsis is commonly around two micrometers in diame-ter. The small size of the cells produces problems for cheap har-vesting with filters. Nannochloropsis sp. live cultures wereobtained from a grower in Apopka, FL, through Aquatic Eco Sys-tems, Inc. part #LAC1Q. The algae were grown in Guillard mediaf/2 and suspensions at pond concentrations (OD between 0.78and 1.06) were shipped overnight in thermally insulated contain-ers. The concentration range was due to variation in the algaebatches provided by the supplier. Algae suspensions were kept ina refrigerator when not used for experiments. Algae were consid-ered viable for experiments as long as their optical density re-mained unchanged. The experiments were conducted with algaeat room temperature.

Chlorella minutissima, a salt-water microalga, was also used inthis study. Those results were similar to those obtained using Nan-nochloropsis and therefore are not reported in this paper.

2.3. Experimental method

Algae were treated in the batch reactor to study the effect ofvarying process parameters on separation time. Each treatmentfollowed the same procedure.

The initial algae concentration was measured using OD. Onehundred milliliters of algae at room temperature was placed inthe batch reactor. The operational parameters for the voltage con-trol circuit (nominal voltage, pulsed or continuous, and duration)were established and the algae were processed in the reactor.Nominal voltage was set at the power supply and was between 0and 4 V. When pulsed, voltage was always pulsed 50 ms ‘‘on’’and 50 ms ‘‘off’’ (50% duty cycle). Duration is the total ‘‘on’’ timefor the voltage. For most tests, duration was 4 s. This resulted ina total process time of 8 s for pulsed voltage and 4 s for continuousvoltage. During the process, reactor voltage and current were re-corded at sample rate of 100 kHz.

Immediately after the treatment the solution was placed in abeaker, stirred for 15 s with a thin spatula, and poured into a100 mL graduated cylinder. The separation in each cylinder wasmeasured by taking a 4 mL sample from the middle of the cylinderand measuring its optical density (OD). This procedure was re-peated over a period of several hours to obtain separation versustime data. Separation at time t after pouring into a graduated cyl-inder is reported in percent as ECF Efficiencyat time t and depends onthe OD measured at time t (ODat time t) and the OD of the untreatedalgae (ODinitial), as defined by:

ECF Efficiencyat time t ¼ 1� ODat time t

ODinitial

� �� 100½%�

Energy input was found by integrating the power over the dura-tion of the processing time using the formula:

Ein ¼Xn

i¼0

ViIiDt

where Vi and Ii are the voltage and current recorded at the timei � Dt, and Dt = 10�5 s is the sample period.

The mass of metal used in the process is calculated using Fara-day’s Law:

m ¼ MzF

Xn

i¼1

IiDt

where m is mass of the metal in g, M is molecular weight(M = 58.6934 g/mol for nickel,) z is the valence, i.e., number of elec-trons released per ion (z = 2 electrons/mol), F = 96500 C/mol, I iscurrent in amperes, and Dt = 10�5 s.

Variants of the aforementioned procedure were used to test forspecific effects. The effect of applied voltage was studied by sub-jecting algae to pulsed voltage for a 4 s duration (8 s total processtime), where the nominal voltage was varied between 0 (no treat-ment, control) and 4 V, i.e., 0 V, 1 V, 2 V, 3 V, 3.5 V, and 4 V. Thereactor used nickel electrodes. The algae were tested in theirgrowth medium (0.28 g/L density, OD = 1.06) with an algae sus-pension conductivity of 32 mS/cm. The procedure was repeatedfor the 4 V and 2 V with varying initial algae concentrations (ODbetween 0.78 and 1.0). Effect of duration was examined by repeat-ing the 4 V test with a duration of 2 s (OD = 1.06).

Variations in algae concentration were further examined bytesting algae samples that had been diluted with salt water. Testswere performed for dilutions of 50/50 and 25/75 samples. The 50/50 sample contained 50 mL algae suspension (OD � 1.06) and50 mL saltwater. Similarly the 25/75 sample contained 25 mL algaesuspension and 75 mL saltwater. The saltwater contained table salt(NaCl) mixed with distilled water to the same conductivity as thealgae suspension (32 mS/cm). Each sample was processed with apulsed nominal voltage of 4 V and duration of 4 s. The test wasthen repeated using a 25/75 dilution of algae suspension(OD = 0.94) and salt water with 2 V and a 4 s duration.

The effect of pulsing was investigated as a way of further reduc-ing energy input and extending the lifespan of the electrodes. Apulsed voltage is also less likely than a continuous voltage to dam-age algae through Joule heating (Kandušer and Miklavcic, 2008).The effect was studied by comparing results from 3 V and 4 V testswhere voltage was either pulsed with duration of 4 s as describedpreviously, or applied continuously for 4 s. Note that in all tests,pulsed and continuous, the total ‘‘on’’ time for the voltage was 4 s.

The effect of electrode material on separation was examined bycomparing results from the 4 V tests using nickel electrodes withsimilar tests using marine-grade stainless steel electrodes. Becauseit was apparent from preliminary tests that stainless steel wouldnot be as effective as nickel, stainless steel was tested for longerduration times – 4 V pulsed with 5 s duration and 4 V continuous

220 T.R. Shuman et al. / Bioresource Technology 158 (2014) 217–224

with 5 s duration. The test was then repeated at 3.5 V pulsed volt-age with duration times of 10, 12 and 14 s.

Finally, several tests were performed by processing only saltwater in the reactor and then mixing this with algae samples. Inthis test, six 100 mL samples of saltwater (NaCl mixed with dis-tilled water to a conductivity of 37 mS/cm) were processed with4 V pulsed and a 4 s duration. Immediately following the process-ing, 100 mL of untreated algae suspension were poured into eachof three saltwater samples and stirred for 15 s with a thin spatula.The other three 100 mL samples of treated saltwater were left for24 h; after which, 100 mL of untreated algae suspension werepoured into each and stirred with the spatula for 15 s. Optical den-sity in all six samples was measured immediately after the algaewere mixed into the salt water and in 30 min increments for thenext two hours.

The effect of the ECF process on cell viability was examinedusing optical data collected using a Fluid Imaging Technologies’FlowCAM�. Six samples were shipped overnight to Fluid ImagingTechnologies and tested the next day. The six samples were: con-trol (0 V), 2 V pulsed, 3 V pulsed, 3 V continuous, 4 V pulsed and4 V continuous all with a duration of 4 s. Cell viability was exam-ined using SYTOX Green Nucleic Acid Stain (Invitrogen, MolecularProbes). SYTOX Green penetrates cells with compromised plasmamembranes and does not cross the membranes of live cells. Nucleicacids of dead cells stained with this dye fluoresce bright greenwhen excited with 450–490 nm source, and their fluorescencedoes not overlap with chlorophyll autofluorescence; hence, fluo-rescence of this dye and autofluorescence can be used simulta-neously as markers of dead and live cells, respectively (Satoet al., 2004). Twenty-five microliters of each sample were mixedwith equal amount of SYTOX Green and incubated at room temper-ature for 30 min in the dark; 50 lL sample/dye was mixed into25 mL of HPLC grade H2O. A 300 lL sample was run through50 lm flow cell in FlowCAM�. The FlowCAM� uses two photomul-tipliers (PMTs), and a 488 nm laser to characterize live versus deadcells. Live cells are detected using one PMT (Ch.1). This PMT has a650 nm low pass filter allowing for chlorophyll-b fluorescencedetection. Dead cells are identified with the other PMT (Ch.2). ThisPTM has a 535 nm ± 15 nm band pass filter allowing for recogni-tion of Sytox Green’s emission at 523 nm.

Further cell viability tests were conducted on a laser scanningZeiss LSM 510 NLO Confocal Microscope. Three images were cre-ated to analyze the samples—a light field image, auto fluorescentimage, and dark field image. The light field image uses white lightto illuminate the algae cells and shows all the cells present in thesample. The auto fluorescent image is created using a 458 nm lightwave and shows the auto fluorescence of the algae. The dark fieldimage shows if the SYTOX Green Nucleic Acid Strain has pene-trated the cell membrane. If there are ruptures in the cell mem-brane, the stain will cause those areas of the cell to fluorescebright green when excited by a 488 nm laser.

2.4. Mechanism

The process described in Sections 2.1 and 2.3 is based on ECF ofmicroalga and causes alga to separate from it growth medium. Thisseparation of microalga has been postulated in the literature to bedue to the following effects: (1) charge neutralization, and (2)sweeping flocculation and enmeshment (Wu et al., 2012; Vandam-me et al., 2011; Gao et al., 2010). This study also explored cellmembrane damage as a potential contributor to the separation.These separation mechanisms are summarized in the followingparagraphs.

Separation begins when direct current is applied to the algalgrowth medium (an electrolytic solution) and metal ions are re-leased from the anode (see Eq. (1) for anode made from pure nick-

el). Hydroxide ions, OH�, are formed at the cathode through thereduction of water to hydrogen and hydroxide, see Eq. (3). Floccu-lation occurs when algae stick together once their repulsive nega-tive charge is neutralized by positive metal ions (chargeneutralization). The metal ions, including nickel, also bond withOH� to form metal hydroxides, (see Eq. (4)). In basic solutions,nickel ions precipitate from the solution in the form of light-greennickel hydroxide, Ni(OH)2. The hydroxide has a large adsorptionarea and electrostatically attracts negatively-charged algae cellsthrough fine enmeshment. So, during its precipitation, nickelhydroxide may carry the attracted algae cells out of the solution.Wu et al. (2012) and Gao et al. (2010) postulated that at higherpH the dominant mechanism for algae removal is actually sweep-ing flocculation and (gross) enmeshment. Gross enmeshment oc-curs when large volumes of the precipitating hydroxidephysically trap the algae cells and flocs and sweep them out ofthe solution. This physical process is more effective for removingflocs than individual algal cells from solutions, due to larger vol-ume of flocs. The enmeshment, both fine and gross, creates struc-tures that fall to the bottom of the solution or float to the top ofthe solution in presence of gasses (such as those formed in waterelectrolysis, Eqs. (2) and (3)). Evidence of both charge neutraliza-tion and enmeshment is discussed in Section 3.

Relevant chemical half-reactions occurring at the nickel elec-trodes are summarized below (Eqs. (1)–(3)), as is the relevant reac-tion in the electrolyte, i.e., the suspension (Eq. (4)):

Anode:

Ni! Ni2þ þ 2e� ð1Þ

4OH� ! O2 " þ2H2Oþ 4e� ð2Þ

Cathode:

4H2Oþ 4e� ! 2H2 " þ4OH� ð3Þ

Electrolyte:

Ni2þ þ 2OH� ! NiðOHÞ2 # ð4Þ

Finally, cell membrane damage can cause algae to die and fallout of the growth medium. Numerous electroporation studies haveshown that voltages above the membrane threshold can damagethe membrane (Kandušer and Miklavcic, 2008; Sack et al., 2010).Furthermore, nickel is a toxin and its presence around algae maypoison the cells. Therefore, cell viability studies are merited and in-cluded in Section 3.7.

3. Results and discussion

Results from this study focused on six areas: (1) voltage pulseamplitude; (2) algae concentration; (3) pulsing; (4) electrodematerial, (5) dilution; and (6) electrolysis effects. In addition, thestudy evaluated the minimum energy required for separation andthe effect of the process on cell viability.

3.1. Effect of applied voltage on separation

The effect of applied voltage on separation was studied for 0 V(control sample), 1 V, 2 V, 3 V, 3.5 V, and 4 V. Fig. 2 shows ECF Effi-ciency over time for different levels of pulsed voltage. In all but the1-volt and control samples, 80% of the algae separated after 2 h. Ahigher separation rate is noted at higher voltages, as much as 90%in two hours for 4 V. The maximum measured separation was 97%,and could be reached by algae treated between 2 and 4 V. Rapidseparation occurs for energy densities between 0.03 kWh/m3 at2 V and 0.3 kWh/m3 at 4 V.

Fig. 2. ECF Efficiency over time for Nannochloropsis sp. Treatment samplesprocessed using a pulsed nominal voltage (1–4 V as shown in data-markers,50 ms on-50 ms off) with 4 s duration of applied voltage (8 s total process time)using nickel electrodes. Control sample is not processed; it is allowed to settlenaturally. Initial OD is 1.06.

Fig. 3. Effect of algae concentration on ECF Efficiency over time for Nannochloropsissp. processed using pulsed voltage (50 ms on-50 ms off) with 4 s duration of appliedvoltage (8 s total process time) using nickel electrodes at a nominal 4 V (0.3 kWh/m3) and 2 V (0.03 kWh/m3). Initial OD is indicated next to data curve and is variedbetween 0.78 and 1.06.

T.R. Shuman et al. / Bioresource Technology 158 (2014) 217–224 221

Cylinders containing algae suspension that were processed at4 V and allowed to settle for two hours show three distinctive lay-ers: a thick light green sediment on the bottom; the supernatantthat is mostly clear with a few light-green specs; and, a light greenlayer that floats on the top.

The bottom layer contains algae and nickel hydroxide. The pres-ence of algae was verified using an optical microscope (Fig. A.1).The layer also contains light-green nickel hydroxide, Ni(OH)2. Thiswas verified by testing a sample of the bottom layer with dimethylglyoxine in pH = 10 solution made with ammonium hydroxide. Thesample immediately turns bright strawberry red, indicating largequantities of nickel. Presence of the insoluble nickel hydroxidewas also verified by comparing against Ksp. Nickel ion concentra-tion was estimated using Faraday’s law and OH� concentrationusing measured pH (pH = 8.5). The product [Ni+2][OH�]2 was calcu-lated to be on the order of 10�8. This is significantly larger than thesolubility product constant Ksp = 6 � 10�16 and indicates insolubil-ity of nickel hydroxide under these conditions.

The supernatant contains a few individual algal cells, observedmicroscopically, and small amounts of nickel Ni+2. Presence ofnickel was detected as the sample turned a barely visible pale pinkcolor in the dimethyl glyoxine test.

The top layer contains algae and some nickel hydroxide. Thepresence of algae was verified using an optical microscope(Fig. A.2). Electrolysis of water forms hydrogen and oxygen gasses,which rise to the surface. Formation of the gasses was visible dur-ing treatment as small bubbles forming on the electrodes. Thesegasses carry algal flocs and metal hydroxide to the top surface.The flocs are held on the top by surface tension. If disrupted theseflocs sink to the bottom. It is interesting to note that the upperlayer is thinner at lower voltage inputs. This is because lower volt-ages result in fewer gasses and so fewer of the flocs are pulled tothe surface.

The ratio of nickel released (calculated using Faraday’s Law) todry algal biomass ranges approximately between 0.1 and 0.3 mgnickel/mg alga. Presence of nickel produces safety concerns bothduring biomass processing and using of the final products. Also,it influences the types of processes that can be applied, and may in-hibit or aid them, warranting careful design of the algal biomassprocessing (NRC, 2012).

A key finding is that a voltage between 1 V and 2 V is the min-imum necessary to rapidly separate algae. Corresponding mini-

mum energy input is referred to as the ‘‘threshold’’ in this paper.The threshold in these experiments is 0.03 kWh/m3 or 0.11 kWh/kg (2 V input). This is significantly lower than other researchershave found. This is expected for four reasons: (1) this study usedmarine algae with high medium conductivity, (2) the reactor usedin this study has a much higher surface to volume ratio than inother studies, (3) the proximity of electrodes in the reactor inducesnatural mixing which aids the separation process, and (4) the treat-ment times are much shorter than others report.

Overall these findings correlate with research done by Poelmanet al. (1997). That study, however, used larger voltages (18–85 V),consumed more energy (0.3 kWh/m3), and required longer treat-ment time (up to 75 min). The algae suspension was also different:a mixture of blue–green, green algae and diatoms at a density of0.008 g/L. Finally, tests in that study were performed in a signifi-cantly larger vessel, around 100 L, with flat lead cathodes and alu-minum anode tubes. Nevertheless, conclusions were similar tothose obtained in this study: (1) there is a threshold value thathas to be exceeded for satisfactory coagulation to occur; and (2)maximum achievable separation is 96%. Poelman et al. (1997) alsoconcluded that the threshold value was dependent on the appliedvoltage, electrode surface area, and distance between cathode andanode.

3.2. Effects of algae concentration

One of the challenges facing commercialization of algae pro-cessing is that algae are live organisms and so constantly changing.These changes can affect total algae concentration (as measured byOD) and suspension conductivity (which is influenced by OD). Thiseffect was also seen by Pearsall et al. (2011). They hypothesizedthat current flow is primarily influenced by ion transport in themedium and to a lesser extent by the transport of algal cells.Decreasing algae density results in fewer obstructions to the cur-rent flow, hence higher conductivity. Pearsall et al. (2011) observeda decrease in conductivity with increasing algae density, similar tothat observed in this study, although at much lower, freshwaterconductivities (�1 mS/cm).

Fig. 3 shows the range of ECF Efficiencies over time for differentinitial algae pond concentrations (OD between 0.78 and 1.06) at2 V and 4 V. Note that concentration affects not only total

Fig. 4. ECF Efficiency (at 120 min after processing) divided by energy densityplotted versus energy density for treatment of Nannochloropsis sp. Algae processedusing pulsed voltage (1–4 V as shown in data-markers, 50 ms on-50 ms off) with 4 sduration of applied voltage (8 s total process time). A single case, 4 V with0.13 kWh/m3 was processed for duration of 2 s (4 s total process time).

222 T.R. Shuman et al. / Bioresource Technology 158 (2014) 217–224

separation time but also the slope of the separation versus timecurve. Increasing algae concentration slows the initial separationat 4 V and speeds it at 2 V. For example, algae with initial densityof 0.2 g/L will not separate if treated at 2 V, but will rapidlyseparate at 4 V. The results support the separation mechanisms de-scribed earlier.

At higher voltage there is sufficient current to initiate floccula-tion and to force the formation of hydroxides regardless of concen-tration. At 4 V with low initial algae concentration (OD � 0.8) andcorresponding higher suspension conductivity (42 mS/cm) there isa faster initial separation rate (steeper slope in Fig. 3). The highersuspension conductivity leads to higher current (7A at OD = 1.06to 9A at OD = 0.84), more nickel ions being released, and more flocsand hydroxides being formed, which in turn causes rapid algaeseparation through sweeping flocculation and gross enmeshment.In those experiments, nickel hydroxide was observed to form evenduring the 8 s processing time.

At 4 V with high initial algae concentration (OD � 1.0) and cor-responding lower suspension conductivity (32 mS/cm), the initialseparation rate is slower (shallower slope on Fig. 3). The currentis lower due to low suspension conductivity, and so fewer nickelions are released, causing a small delay in the floc formation andhydroxide precipitation. Once formed, the hydroxides trap algaeand separate them from the suspension through grossenmeshment.

An opposite effect was seen when processing pulsed data at 2 V.At 2 V, there are fewer flocs and less hydroxide is formed due tofewer ions being released than at 4 V. Evidence of fewer and smal-ler flocs was observed using microscope. Lower initial concentra-tion algae (OD � 0.8) is even less likely to form flocs becausealgae are further apart. Even the increase in metal ions releaseddue to higher conductivity may not be enough to overcome the dis-tance between algae cells. At higher algae concentration,(OD � 1.0) flocculation occurs even at the low voltages, althoughat slower rate. It takes time to form, precipitate and enmeshhydroxides with algal cells; once that occurs, settling begins tohappen. In the low voltage experiments, the reduced number andsize of flocs emphasizes the importance of fine enmeshment forseparation.

3.3. Minimum energy density

While previous results indicate that higher voltages result infaster separation, they do not characterize the total energy re-quired for separation. Fig. 4 shows the ratio of measured ECF Effi-ciency two hours after treatment to energy density plotted as afunction of energy density. The pulsed duration for all treatmentswas the same, 4 s, except for a single sample (OD = 1.06) wherethe nominal voltage was 4 V and the pulsed duration was 2 s(0.13 kWh/m3 energy density.) Halving the exposure time in thissample also resulted in over 90% ECF Efficiency after two hours,indicating that even at the shorter duration there was sufficientnickel release to cause effective separation. The benefit of shorten-ing the time is that the ratio of ECF Efficiency to energy densitydoubles (up from 2.5 to 5 as shown in Fig. 4).

The data shows that 4 V cases actually have lower ECF Effi-ciency-to-energy density ratio than do the 3 V cases. In fact, the2 V cases appear to have the highest ratio, even though separationpercent for 2 V cases is lower (see Fig. 2 above). This suggests thatobtaining 90% ECF Efficiency is not necessarily the most efficientmethod for separating algae from the medium. Aiming for nearlycomplete separation is less energy efficient than simply acceptingpartial separation. The caveat is that 2 V pulsing is only effectivefor high algae concentrations (density approximately 0.3 g/L) andenergy densities above a threshold value of 0.03 kWh/m3. At lowerconcentrations, pulsing at 2 V results in very little separation be-

cause the energy density threshold is higher. Finally, increasing en-ergy density above 0.3 kWh/m3 may not result in anyimprovement in ECF Efficiency when measured two hours aftertreatment.

3.4. Pulsing effect on separation

Pulsing effect was studied by comparing separation times underfour conditions, all using nickel electrodes: 3 V and 4 V continuous,and 3 V and 4 V pulsed all with duration of 4 s. Tests showed thatthere was no difference in separation between pulsed and contin-uous voltage, (evidenced also on FlowCAM�, Fig. A.3). Statistically,the difference is negligible; the paired t-test, two-tailed, shows nosignificant difference between the pulsed and no-pulsed data(p = 0.09 for 3 V data). The results from the 3 V test, however, high-light the effect of proximity to the threshold. The shallow slopeindicates smaller or no flocs (evidenced also on FlowCAM�, resultsnot shown), which in turn means that gross enmeshment is lesseffective in removing algae from the solution rapidly. The onsetof rapid separation is also delayed due to delayed precipitation ofthe hydroxide, similar to the 2 V cases. The large error bars around60 min after treatment are caused by small variations in energy in-put for different samples, which, in turn, influence the formationand precipitation of hydroxides and formation of flocs, (see Fig. 5).

3.5. Effect of electrode material

Test comparing the effectiveness of nickel to stainless steelshow the stainless steel is much less effective that nickel. In fact,the separation rate is much lower for stainless steel electrodeseven when the voltage is applied for a longer time. After treatingalgae with 3.5 V pulsed for a 14 s duration (0.4 kWh/m3), onlyabout 25% of the algae had separated from the medium after twohours and it takes many days to reach 90% separation (Figure A.4).In contrast, using nickel electrodes and duration of 4 s, algaeachieved 90% separation in about two hours. The data also showthat separation rate is independent of the treatment length. Thatis, using stainless steel electrodes and treating for 5 s, versus 14 shad the same effect. This indicates that the treatment conditionswere well above the threshold. Furthermore, as in the nickel tests

Fig. 5. Effects of pulsed vs continuous voltage on Nannochloropsis sp. using nickelelectrodes. Nominal voltage (3–4 V as shown in data-markers) was pulsed (50 mson-50 ms off) or continuous (constant on) for a 4 s duration of applied voltage (8 stotal process time for pulsed treatment, 4 s total process time for continuoustreatment).

Fig. 6. Comparison of ECF Efficiency when treating Nannochloropsis sp. directlyversus when mixing untreated suspension with treaded saltwater (50% dilution)immediately or after 24 h. All treatments pulsed with a nominal 4 V (50 ms on-50 ms off) with 4 s duration of applied voltage (8 s total process time).

T.R. Shuman et al. / Bioresource Technology 158 (2014) 217–224 223

(Section 3.4), ECF Efficiency is unchanged if voltage is pulsed or ap-plied continuously.

The low ECF Efficiency of iron was observed by others, includingVandamme et al. (2011), Gao et al. (2010), and Aragon et al. (1992).In this study, as in the aforementioned studies, the solution be-came yellowish throughout and turbidity increased immediatelyafter treatment using stainless steel electrodes. Only in case of veryhigh energy inputs (�3.5 kWh/m3) did turbidity completely disap-pear after 3 days leaving a thick green layer on the top and brown-ish/green layer on the bottom. While a more careful study iswarranted to explain lower ECF Efficiency of stainless steel, thereasons are likely due to iron hydroxides being relatively ineffi-cient coagulants, as postulated by Emamjomeh and Sivakumar(2009).

3.6. Dilution and electrolytic effect

Effect of dilution was tested for 50/50 and 25/75 samples withnickel electrodes using a 4 V and 2 V pulsed signal with a 4 s dura-tion. The data indicate that dilution has a small to negligible influ-ence on separation provided that the voltage is high enough.Previous results showed that low voltages, 2 V, and low concentra-tion (OD � 0.8) did not result in settling, so it is not surprising that2 V test of diluted samples resulted in minimal separation. At 4 V,however, the current is high enough to produce nickel ions to makeflocs and precipitate hydroxides in quantities sufficient to separatea wide range of microalgae densities.

Results for test in which untreated algae were mixed with trea-ted salt water are shown in Fig. 6. These results corroborate findingfrom the dilution test, that the existence of hydroxides alone canbe sufficient to cause separation. The results demonstrate thatthe algae does not need to be processed directly in the reactor inorder for the separation to be achieved. Instead algae can be mixedwith treated saltwater and still achieve similar separation rates. Al-gae samples that were mixed with treated saltwater 24 h aftertreatment at 4 V experienced approximately 25% less separationafter a two hour period. This is explained by two effects. The firsteffect is evidenced by non-existence of the top layer, so all separa-tion was in the form of settling. This is because gas-bubbles that

form during electrolysis are no longer in solution and so cannot liftthe algae and hydroxide to the top. The second effect is related tocharge neutralization and fine enmeshment. Flocculation inducedby charge neutralization with metal ions is restrained because ionshave already precipitated as hydroxides. Charge neutralizationwith hydroxides is also likely delayed since hydroxide no longerforms around algae, hence slowing fine enmeshment. The separa-tion is mostly caused by gross enmeshment since insolublehydroxides are present even 24 h after treatment, and when thoseare mixed in with algae, they settle the flocs. These data emphasizethe importance of gross enmeshment, which is responsible for upto 75% of the removal of algae from the solution. Gross enmesh-ment played a smaller role for the 3 V data, only up to 25% (datanot shown). It is very likely that the lower currents induce lessnickel ion formation. Furthermore, fewer nickel ions mean forma-tion of a smaller number of flocs and reduced effectiveness of grossenmeshment. Algal cells and smaller flocs were too small to be allpulled down with gross enmeshment in 3 V cases.

It should be noted that algae will not separate if algae suspen-sion is poured into treated saltwater without additional stirring.But, it will separate if it is stirred or if saltwater is poured into al-gae. The process of hydroxide formation and precipitation is visu-ally equivalent in saltwater and in algae-suspension samples.These observations further support gross enmeshment as the pri-mary separation mechanism.

Mixing treated saltwater with the microalgae solution is alsomore effective than adding nickel salt to a microalgae solutionand inducing settling by changing the pH. Salts are not as efficientin producing hydroxides as ECF and as previously discussed,enmeshment with hydroxides is the primary separation mecha-nism. To verify this, a 100 mL algae solution sample was placedin a beaker and thoroughly mixed with ample amount of solidnickel chloride. No settling was observed 24 h after treatment. Thisresult is corroborated by the findings of Vandamme et al. (2011).They showed that ECF consumes less aluminum than chemicalcoagulation–flocculation with alum. Furthermore, the proposedapproach to treat saltwater may have practical consequences inthe design of microalgae farms and treatment facilities. For exam-ple, seawater can be treated and then pumped to the settling tankswhere it would be mixed with algae; that way fouling of the

224 T.R. Shuman et al. / Bioresource Technology 158 (2014) 217–224

pumps and ECF devices with algal suspensions could be avoided.While this approach uses more water, most of the water can bere-used. The parameters surrounding the industrial processing willlikely affect the choice of the process that is eventually used toimplement ECF.

3.7. Cell viability

One potential concern with any dewatering method is how it af-fects algae cell viability. Cell viability helps identify if separationwas occurring due to cell damage – one of the separation mecha-nisms discussed earlier. Visual inspection of individual cells andflocs imaged with the FlowCAM� and measurements of cell round-ness indicate no visual damage to the majority of the treated cells.Only heat treated samples showed a significant increase in cellmortality compared to the control. The percent of dead cells rangesfrom 3% for 2 V, pulsed to 7% for 4 V, non-pulsed samples. (0.6% forcontrol and 100% for heat treated samples.) The data also showsthat pulsing does not have a significant effect on cell viability(Fig. A.3). Cell viability tests suggest that it is unlikely that nickelpenetrated the Nannochloropsis sp. membrane, making these stur-dy species suitable for ECF. Finally, FlowCAM� images show thatthe particles are much larger after processing, providing more sup-port for the enmeshment/flocculation mechanism. The results ofthis test performed using the Zeiss LSM 510 NLO Confocal Micro-scope corroborate those from the FlowCAM�.

The key finding is that the electro-coagulation-flocculation usedin this study does not significantly affect cell viability for Nanno-chloropsis sp. In terms of separation mechanism, this suggests thatflocculation and enmeshment are the primary methods for separa-tion and the separation due to cell damage has only a minor effect.However, further research is necessary to study cell viability forother microalgae.

4. Conclusions

Electro-coagulation–flocculation can rapidly separate Nanno-chloropsis sp. from growth medium at pond concentrations (over90% in 120 min) after treatment with ultra-low energy from a DCpower supply for a very short time (4 s) in a batch reactor withnickel electrodes. The threshold for effective separation is closeto 0.03 kWh/m3 for algae suspension density of 0.3 g/L. The processdoes not cause significant cell damage and is effective using pulsedor continuous current. Algae do not have to be subjected directly tothe process and will separate from their growth medium even ifmixed with an independently treated growth medium.

Acknowledgements

Dr. Margaret Hudson, Dr. Cheryl Wotus, and Dr. Dan Smith, ofthe Biology Department at Seattle University.

Dr. Susan Jackels of the Chemistry Department at SeattleUniversity.

Senior design student teams sponsored by Bioalgene, Boeing,and Valicor� Renewables LLC, formerly SRS Energy LLC.

Fluid Imaging Technology, for conducting FlowCAM� tests.

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.2014.02.033.

References

Alfafara, C.G., Nakano, K., Nomura, N., Igarashi, T., Matsumura, M., 2002. Operatingand scale-up factors for the electrolytic removal of algae from eutrophiedlakewater. J. Chem. Technol. Biotechnol. 77, 871–876.

Aragon, A.B., Padilla, R.B., Ros, Fiestas., de Ursinos, J.A., 1992. Experimental study ofthe recovery of algae cultured in effluents from the anaerobic biologicaltreatment of urban wastewaters. Resour. Conserv. Recycl. 6, 293–302.

Azarian, G.H., Mesdaghinia, A.R., Vaezi, F., Nabizadeh, R., Nematollahi, D., 2007.Algae removal by electro-coagulation process, application for treatment of theeffluent from an industrial wastewater treatment plant. Iranian J. Publ. Health36 (4), 57–64.

Cheng, J.J., Timilsina, G.R., 2011. Status and barriers of advanced biofueltechnologies: a review. Renew. Energy 36, 3541–3549.

Chisti, Y., 2007. Biodiesel from microalgae. Biotechnol. Adv. 25, 294–306.Danquah, M.K., Ang, L., Uduman, N., Moheimani, N., Forde, G.M., 2009. Dewatering

of microalgal culture for biodiesel production: exploring polymer flocculationand tangential flow filtration’’. J. Chem. Technol. Biotechnol. 84 (7), 1078–1083.

Demirbash, M.F., 2011. Biofuels from algae for sustainable development. Appl.Energy 88, 3473–3480.

DOE, 2010. National algal biofuels technology roadmap. US Department of Energy,Office of Energy Efficiency and Renewable Energy, Biomass Program. http://www1.eere.energy.gov/biomass/pdfs/algal_biofuels_roadmap.pdf, accessedOctober 11, 2011.

Emamjomeh, M.M., Sivakumar, M., 2009. Review of pollutants removal byelectrocoagulation and electrocoagulation/flotation processes. J. Environ.Manage. 90, 1663–1679.

Evodos company web site. http://www.evodos.eu/, accessed December 4, 2013.Gao, S., Yang, J.X., Tian, J.Y., Ma, F., Tu, G., Du, M., 2010. Electro-coagulation–flotation

process for algae removal. J. Hazard. Mater. 177, 336–343.Kandušer, M., Miklavcic, D., 2008. Electroporation in biological cell and tissue: an

overview. In: Vorobiev, E., Lebovka, N. (Eds.), Electrotechnologies for Extractionfrom Food Plants and Biomaterials. Springer Science, New York, pp. 1–37.

Kim, J., Ryu, B.-G., Kim, B.-K., Han, J.-I., Yang, J.-W., 2012. Continuous microalgaerecovery using electrolysis with polarity exchange. Bioresour. Technol. 111,268–275.

NRC-National Research Council, 2012. Sustainable Development of Algal Biofuels inthe United States. The National Academies Press, Washington, DC.

Pearsall, R.V., Connelly, R.L., Fountain, M.E., Hearn, C.S., Werst, M.D., Hebner, R.E.,2011. Electrically dewatering microalgae. IEEE Trans. Dielectr. Electr. Insul. 18(5), 1578–1583.

Poelman, E., De Pauw, N., Jeurissen, B., 1997. Potential of electrolytic flocculation forrecovery of micro-algae. Resour. Conserv. Recycl. 19, 1–10.

Sack, M., Sigler, J., Frenzel, S., Eing, Chr., Arnold, J., Michelberger, Th., Frey, W.,Attmann, F., Stukenbrock, L., Mueller, G., 2010. Research on industrial-scaleelectroporation devices fostering the extraction of substances from biologicaltissue. Food Eng. Rev. 2, 147–156.

Sato, M., Murata, Y., Mizusawa, M., Iwahashi, H., Oka, Sh., 2004. A simple and rapiddual-fluorescence viability assay for microalgae. Microbiol. Cult. Collect. 20 (2),53–59.

Spolaore, P., Joannis-Cassan, C., Duran, E., Isambert, A., 2006a. Commercialapplications of microalgae. J. Biosci. Bioeng., Volume: 101 Issue: 2 Pages: 87–96, FEB 2006.

Spolaore, P., Joannis-Cassan, C., Duran, E., Isambert, A., 2006b. Optimization ofNannochloropsis oculata growth using the response surface method. J. Chem.Technol. Biotechnol. 81, 1049–1056.

SYTOX Green Nucleic Acid Stain, http://probes.invitrogen.com/media/pis/mp34860.pdf, accessed on 9/22/2013.

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), 1–15.

Uduman, N., Qi, Y., Bourniquel, V., Danquah, M.K., Hoadley, A., 2011. A parametricstudy of electro-coagulation as a recovery process of marine microalgae forbiodiesel production. Chem. Eng. J. 174, 249–257.

Vandamme, D., Pontes, S.C.V., Goiris, K., Foubert, I., Pinoy, L.J.J., Muylaert, K., 2011.Evaluation of electro-coagulation–flocculation for harvesting marine andfreshwater microalgae. Biotechnol. Bioeng. 108 (10), 2320–2329.

Veldhuis, M.J.W., Cucci, T.L., Sieracki, M.E., 1997. Cellular DNA content of marinephytoplankton using two new fluorochromes: taxonomic and ecologicalimplications. J. Phyml. 33, 527–541.

Wu, Z., Zhu, Y., Huang, W., Zhang, Ch., Li, T., Zhang, Y., Li, A., 2012. Evaluation offlocculation induced by pH increase for harvesting microalgae and reuse offlocculated medium. Bioresour. Technol. 110, 496–502.

Zeng, X., Danquah, M.K., Chen, X.D., Lu, Y., 2011. Microalgae bioengineering: FromCO2 fixation to biofuel production. Renew. Sustain. Energy Rev. 15, 3252–3260.

Zhang, G., Zhang, P., Wang, B., Liu, H., 2006. Ultrasonic frequency effects on theremoval of Microcystis aeruginosa. Ultrason. Sonochem. 13, 446–450.

Zhang, G., Zhang, P., Fan, M., 2009. Ultrasound-enhanced coagulation forMicrocystis aeruginosa removal. Ultrason. Sonochem. 16, 334–338.

Zhukova, N.V., Aizdaicher, N.A., 1995. Fatty acid composition of 15 species of marinemicroalgae. Phytochemistry 39 (2), 351–356.