Feasibility study of microfiltration for algae separation in an innovative nuclear effluents decontamination process

Separation and Purification Technology 125 (2014) 126–135

Contents lists available at ScienceDirect

Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur

Feasibility study of microfiltration for algae separation in an innovative nuclear effluents decontamination process D. de Gouvion Saint Cyr a,b,c,d,g, C. Wisniewski e,⇑, L. Schrive f, E. Farhi g, C. Rivasseau a,b,c,d a

CEA, Laboratoire de Physiologie Cellulaire Végétale, 38054 Grenoble, France CNRS, UMR5168, 38054 Grenoble, France c Université de Grenoble, 38000 Grenoble, France d INRA, 38054 Grenoble, France e UMR QualiSud, UFR des Sciences Pharmaceutiques et Biologiques, Université de Montpellier 1, 34093 Montpellier, France f CEA-DEN, Laboratoire des Procédés Supercritiques et de Décontamination, 30207 Bagnols sur Cèze, France g Institut Laue Langevin, Division Science, 38042 Grenoble, France b

a r t i c l e

i n f o

Article history: Received 15 October 2013 Received in revised form 21 January 2014 Accepted 23 January 2014 Available online 4 February 2014 Keywords: Micro-algae Microfiltration Membrane fouling Hydrodynamic actions

a b s t r a c t Bio-remediation technologies often offer efficiency, cost and environmental impact benefits against physico-chemical technologies. Concerning the remediation of radionuclide-containing water, a few bio-based technologies have been proposed but none is currently operational in highly radioactive environments. A new radio-tolerant micro-alga, isolated from a nuclear facility, possesses properties that offer new decontamination prospects for the nuclear industry or for the clean-up of environmental water. A pilot-scale treatment unit based on this alga is currently under development for the decontamination of radioactive water. It includes separation and/or concentration steps relying on membrane filtration. This work aims at verifying the feasibility of microfiltration as separation step for the targeted algae separation. Recommendations about the choice of operating conditions limiting and/or controlling the membrane fouling are provided with the objective to enhance the separation efficiency. Lab-scale dead-end filtration tests were implemented and the key factors involved in the separation performances were investigated. Membrane characteristics, biomass composition, and hydrodynamic conditions were considered. Organic membranes provided adequate filtration performance. Membrane fouling was essentially induced by a rapid reversible algae deposit and to a lesser extent by irreversible pore blockage caused by smaller particles and dissolved organic matter. To cancel the reversible fouling, hydrodynamic actions such as stirring and back-flush efficiently prevented algae deposit, allowing higher filtration productivity. This study demonstrates the feasibility of membrane separation for micro-algae harvesting at laboratory-scale and specifies the suitable working conditions. Ó 2014 Elsevier B.V. All rights reserved.

1. Introduction The nuclear industry generates radionuclides, including radioactive metals, carbon-14 and tritium, which are found in liquid and gaseous effluents. These effluents must consequently be decontaminated before release in order to reduce their activity below controlled thresholds [1]. Technologies frequently used to separate radionuclides from liquid effluents include evaporation, solid/liquid separation by filtration, centrifugation or decantation possibly associated to a chemical precipitation/flocculation process, membrane filtration, sorption, and ion exchange [2,3]. These methods are efficient and robust; however, they are expensive ⇑ Corresponding author. Tel.: +33 411759664. E-mail address: [email protected] (C. Wisniewski). http://dx.doi.org/10.1016/j.seppur.2014.01.039 1383-5866/Ó 2014 Elsevier B.V. All rights reserved.

and do not completely remove some radionuclides such as carbon-14 and tritium. Alternative technologies are needed to reduce radioactive releases in aqueous effluents. Remediation technologies based on micro-organisms are already used in a variety of industrial applications [4–6] and often offer efficiency, low cost and environmental impact benefits against physico-chemical technologies [7]. They may constitute interesting alternatives in the nuclear field as well, but only a few bio-based technologies have been proposed. One study concerns lab scale experiments for the decontamination of a highly radioactive water issuing from Fukushima accident [8]. Real-scale implementation has scarcely been performed and, to our knowledge, only two processes were reported to have been tested at this scale for uranium treatment, whose radioactivity is low [9,10]. The availability of microorganisms adapted to nuclear

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135

conditions, i.e. tolerant to high ionizing radiations, nutrient stress and metallic stress, would facilitate the development of such bio-applications for the nuclear industry. Actually, living photosynthetic biomasses, especially microalgae, are a potential remediation matrix because they can take up inorganic carbon-14 via photosynthesis, organic carbon-14 via metabolic pathways as well as a large panel of radioactive metals such as silver-110 m and cobalt-60. A new autotrophic green unicellular micro-alga was recently isolated from a used fuel storage pool of a nuclear facility. This alga is extremely radio-tolerant since it survives to radiation doses of up to 20 kGy and has a high capacity to concentrate the main radionuclides present in nuclear effluents, including carbon-14 and the gamma-emitters 238U, 137Cs, 110mAg and 60Co [11]. It is therefore of great interest for the development of new remediation solutions for nuclear effluents and for the clean-up of environmental water. The implementation of a pilot-scale treatment unit, based on this micro-alga and including different tasks, is currently under development for the decontamination of radioactive effluents. To ensure the objectives of the process, algae have first to be produced in a growth medium and harvested before ensuring the treatment of the contaminated effluent. Moreover, after the treatment of the effluent, algae must be separated from the decontaminated effluent. The development of such an unusual process is subject to several constraints linked to the use of a biological matrix in a nuclear environment. Firstly, during the harvesting step, algae have to be strictly separated from the growth medium in order to comply with regulatory constraints in nuclear industry. Secondly, after the decontamination step, the separation technology should allow the whole retention of the algae to ensure a perfect treated water quality in terms of suspended solids retention. In microbiological processes, biomass harvesting and/or dewatering are crucial steps; depending on the final objective (e.g. biomass separation from treated water or biomass concentration), the chosen technology can lead to cell concentration or to complete dewatering [12]. The frequently used technologies are based on usual liquid–solid separation techniques, such as centrifugation, filtration, flotation and sedimentation. In most cases, separation must not modify cells integrity and viability. The choice of the separation technique integrates biological constraints, separation efficiency and energy cost. It should be noticed that in algae field, the separation step is currently a technological lock due to high energy cost [13]. Micro-algae harvesting usually relies on centrifugation, sedimentation, flotation or flocculation [12–14]. Recently, membrane filtration has emerged as a promising advanced method for micro-algae harvesting with several advantages not requiring chemical coagulants or flocculants, which reduces the operational cost while ensuring the quality of the biomass [15]. Furthermore, according to the above-mentioned specific constraint of our pilot-scale treatment unit, membrane filtration seems to be the most appropriate separation technology in our specific case. This work aims at demonstrating the feasibility of microfiltration as separation step for micro-algae separation and at optimizing the filtration efficiency with regard to selectivity and permeability constraints and to biomass fate. Laboratory-scale experiments were performed on a frontal filtration unit to identify the key factors involved in the separation performances, notably in membrane fouling. Three factors known to impact membrane permeability were considered and tested: (i) the membrane characteristics, namely pores diameter and material, (ii) the biomass composition and (iii) the hydrodynamic conditions, namely stirring and back-flush conditions. With the objective to provide recommendations about the operating conditions to be employed at pilot-scale, a particular attention was paid to the influence of the biomass characteristics on the filtration efficiency in the context of the decontamination process; filtration conditions permitting a

127

rapid and easy micro-algae harvesting without cell damage were specifically explored. 2. Experimental 2.1. Algal suspension 2.1.1. Strain, culture and storage conditions The micro-alga Coccomyxa actinabiotis used is described in [11]. Algal biomass was grown continuously in a 10 L photobioreactor, sparged with air at 50 ± 10 L h1, under continuous illumination of 100 ± 20 leinstein m2 s1, at 24 ± 2 °C, using a modified Bold Basal Medium (Sigma–Aldrich, Saint Louis, MO) diluted twice with deionized water. In most cases, freshly produced biomass was immediately used for filtration assays. To assess the impact of biomass ageing on filtration, it was also stored for 1–15 days. Three storage modes were tested, corresponding to the fate of biomass in the context of the decontamination process. In storage conditions I (SCI), the freshly produced biomass was stored under light (100 leinstein m2 s1) with air stirring and nutrient addition; in storage conditions II (SCII), it was stored under light (100 leinstein m2 s1), with stirring but without any nutrient addition; in storage conditions III (SCIII), it was stored away from light (15 leinstein m2 s1), without stirring and with no nutrient additions. 2.1.2. Characterization of algal suspensions Algal biomass consists of an aqueous phase and suspended particles, including algal cells. These two phases were characterized by conventional analytic tools set out below. 2.1.2.1. Particles characterization. Algal density was measured using a Mallassez counting cell and correlated to dry weight (DW). The size of algal cells was determined by microscopic observation. Biomass also contains some cell debris and potentially some bacteria depending on storage conditions. Debris was not quantified but bacterial population was estimated by counting. 2.1.2.2. Aqueous phase characterization. The aqueous phase contained nutrients, dissolved organic matter, and cell content released by cell damage. Dissolved organic matter comes from extracellular polymeric substances (EPS) produced by algae, especially when cells grow under stressful conditions [16]. They are mainly carbohydrates and proteins [17]. Free EPS were separated from algal cells by centrifugation at 2000g for 5 min as described in [15]. Protein and polysaccharides content was characterized by colorimetric methods. Extracellular proteins were quantified using Bradford method (Kit UP36858A, Interchim, Monluçon, France); the detection limit for this method is 1 lg mL1. The phenol–sulfuric acid method from [18] was employed for polysaccharides measurement. The average molecular weight of polysaccharides was determined using size-exclusion chromatography; the size distribution was not identified. 2.2. Filtration unit 2.2.1. Experimental device The dead-end filtration apparatus employed for filtration experiments in both membrane selection and assessment of the impact of the suspension nature and hydrodynamic actions consisted of a laboratory pressurized and stirred rig (Spectrum Laboratories Inc, Rancho Dominguez, CA). The rig was composed of a 0.4 L filtration cell providing a surface of 38.4 cm2. Trials were conducted at constant temperature (24 ± 2 °C), in a batch mode under constant pressure (20 kPa). Compressed air was supplied by a valve placed

128

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135

before the tank containing the algal suspension. Shear stress was induced by a magnetic impeller and rotational speed was adjusted by a magnetic stirrer. An outlet valve and a peristaltic pump were installed to flush back deionized water, if necessary. Deionized water, and not the filtrate as usual, was used for the back-flush. This procedure was followed to take into account the regulatory constraints which impose to strictly separate growth medium from algae before the decontamination step (first separation step at pilot-scale). A given volume of algal suspension was instantaneously introduced in the stirred cell using an opening/closing valve. Permeate was collected in a beaker placed on an electronic balance and the filtrate mass was weighted every minute. Filtration was conducted until one of the following parameter was reached: (i) all the liquid has passed through the membrane, it is therefore considered as a dewatering operation (named De in Table 1), (ii) the time of filtration has reached 20 min, it is considered as a concentration operation (named Co in Table 1). In the case of a dewatering operation, the dry weight of algae harvested on the filter was determined and the volume concentration factor was evaluated. In the case of a concentration operation, only the volume concentration factor was calculated. The experimental set-up is shown in Fig. 1; the experimental conditions for the different assays are summarized in the Table 1.

2.2.2. Membrane characteristics Three organic disc filters (i.e. lab-scale flat sheet membranes) and two mineral ones chosen for their hydrophilic properties were initially selected. They had different pore size, from 0.1 to 1 lm. Their characteristics are summarized in Table 2. Experiments were performed to select the most appropriate membrane to retain algae, let the nutrients and dissolved products pass through and allow the collected micro-algae to be reused for further process step. Water permeability (Lp0) of the five membranes was first assessed. Then algae filtrations under defined working conditions were

carried out so as to observe the efficiency of the different membranes. The adhesion of algae to membrane surface, as well as the reversible nature of the fouling, was studied according to an original experimental procedure. After a given filtration time of the micro-algal suspension, the tested membrane was delicately scraped with a spatula and soaked in a smaller deionized water volume to remove the green deposit, ensuring a mechanical action like a turbulence promoter or an automatic scraped filter, which permits to eliminate the external deposit. This operation was repeated three times. During the filtration operation, permeability was measured versus time. At the end, the membrane was washed by a reverse flux of 0.1 M NaOH to recover products retained in and/or on the membrane [19]. Volume concentration factor was calculated, dry weight of re-suspended algae was determined and both biomass filtrate and washing NaOH solution were analyzed to quantify the potential presence of EPS. This protocol may allow distinguishing the reversible fouling part from the irreversible one for the tested membrane. 2.3. Experimental strategy to identify the key factors involved in the separation performances Key factors involved in the separation performances were studied using different testing procedures related to the nature of the suspension to be filtered (for the identification of the fouling compounds) or to different working conditions (for the optimization of the filtration). 2.3.1. Identification of fouling compounds To identify the role of the different fraction of the algal suspension, as well as the role of particles or dissolved compounds in fouling, filtration tests were carried out with different types of algal suspensions, e.g. with or without concentration and/or fractionation by centrifugation, stored under different modes.

Table 1 Experimental conditions for the different assays. Figure

4 5a

5b

6

7 8a 8b 9a 9b

a

Filtration conditions

Algal suspension

Hydrodynamic conditions

Type of suspension filtereda

Dewatering (De) Concentration (Co)

Filtered volume (mL)

Filtration time (min)

Algal dry weight (gDW L1)

Storage condition (SCI, SCII, SCIII)

Storage time (d)

Stirring

Backflush

A A

De Co

100 170

5 20

0.2 0.3

SCIII SCII same results with SCIII

5 12

Yes Yes

No No

B C D A B C D A A A A A A A A A A A A

De De Co Co Co De Co De Co Co Co/De De De De De Co Co De De

175 175 190 60 100 100 100 275 170 115 – 140 135 320 330 225 190 365 365

2.5 2.5 20 20 20 10 3 20 20 20 P20 3 6.5 5 5 20 20 9 14.5

0.1

SCI

6

Yes

No

0.15 0.3 0.55 0.0.5 to 0.8 0.15 0.15 0.05 0.05 0.24 0.24 0.1 0.1

SCIII

6

Yes

No

Both SCII and SCIII –

0 to 14 0

–

0

SCII

5

SCII

6

Yes Yes No Yes No Yes Yes Yes Yes

No No No No No Yes No Yes No

Suspension A: whole culture; suspension B: culture supernatant; suspension C: washing water; suspension D: water washed algae.

129

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135

Air P = 20 kPa

TANK (algal suspension)

Manometer

Opening/closing valve

STIRRED CELL WATER

FILTRATE (waste) Membrane Balance

Magnetic stirrer

Fig. 1. Scheme of the filtration bench scale unit. Fig. 2. Algal suspension fractionation and filtration procedure.

2.3.1.1. Fractionation of the suspension by centrifugation. The algal suspension was directly filtered (suspension A on the Fig. 2) or was fractionated before filtration in order to study the fouling induced by each fraction. The fractionation procedure, described in Fig. 2, first consisted in centrifuging the algal suspension at 2000g for 5 min to gather algae in the pellet while particles (debris, possible bacteria) and soluble substances remain in the supernatant [15,16]. However, if bacteria concentration was high, numerous bacteria were found in the pellet, stuck on algae cells. Culture supernatant, also called crude supernatant (suspension B), was filtered while algae pellet was washed with deionized water and then centrifuged at 2000g for 5 min. Washing water was filtered (suspension C) and washed algal pellet was re-suspended in deionized water (suspension D) and then filtered in the same conditions. It should be noticed that the algae pellet washing method using deionized water does not induce any osmotic stress on algae cells, since they can grow up in demineralized water. This point has been described in [11]. During filtration, permeability change was characterized versus filtered volume, since each fraction has the same volume. Initial permeability, measured after filtration of 30 mL, related to the nature of fouling products in each fraction. Permeability loss with filtered volume was compared to identify the highest fouling fraction.

2.3.1.2. Cell and storage conditions. To test the influence of cell density, fresh biomass was either used as it was or diluted twice with deionized water or concentrated twice by centrifugation. The three samples were filtered under stirring and permeability was measured versus time. To test the impact of biomass composition, algal suspension was filtered after different type and storage time (ST) and permeability was measured. The permeability after 5 and 20 min measured in these assays was compared according to cell density and to storage time. 2.3.2. Assessment of hydrodynamic actions to prevent fouling 2.3.2.1. Stirring. The shearing effect on permeability change versus filtered volume was tested on suspensions at different cell density filtered with and without stirring. In dead-end filtration, stirring drags the suspension along the membrane surface, giving to the liquid a tangential motion, such as in tangential filtration, and thus limiting particles deposit on membrane surface. The shear rate c(s1) is a function of the rotational speed of the impeller vRPM (m s1), the diameter of the impeller D (m) and the distance between the filter and the impeller x (m), as follows [20]:

c¼

mRPM  p D 60



ð1Þ

x

Table 2 Membrane characteristics. Name

Supplier

Commercial name

Material

Pore size (lm)

Lp0

TiO2 PVDF MEC NC FV

Tami Millipore Millipore Whatman Pall

DisRam Durapore HVLP MF-HAWP Optitran-Ba-S-85 Type A/E

Ceramic TiO2 Polyvinylidene fluoride (PVDF) Mixed cellulose ester Nitrocellulose Glass fiber

0.14 0.45 0.45 0.45 1

– 17 400 36 000 – 500 000

Lp0 Lp0

th:

theoretical permeability provided by the supplier (L h1 m2 bar1). water permeability measured experimentally (L h1 m2 bar1).

exp:

th

Lp0

exp

1000 40 600 41 000 28 000 103 000

130

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135

In the case of Newtonian fluids, the shear stress s (Pa) is linked to the shear rate by viscosity l (Pa s), as follows:

s¼lc

ð2Þ

Knowing the shear rate induced by this motion is essential because micro-organisms are susceptible to damage that is dependent on the prevailing shear stress [21,22]. Beyond a threshold value, dependent on the considered algae species [23], viability of algal cells was reported to decrease strongly; a threshold value of about 1.3 Pa was reported by [23]. In the present working conditions, the shear rate ranged from 200 to 1300 s1, depending on the rotational speed of the impeller, corresponding to a shear stress of 0.8 ± 0.5 Pa; the important standard deviation came from the lake of precision on the rotation of the impeller, rotation induced by the magnetic agitator. 2.3.2.2. Back-flushing. Back-flushing is a hydrodynamic action which allows removing fouling caused by solids deposit by reversing temporarily the flux direction. At lab scale, with our filtration device, it was quite difficult to maintain similar frequency and flow rate conditions for back-flush. Thus, in this work, back-flush of deionized water was performed for 1–3 min every 3–5 min, at a flow rate ranging from 4.7 to 13.8 mL min1. Efficiency of the combined effect of back-flush and shear on the change in the filtered volume versus time was evaluated on suspensions at different cell density.

aiming at evaluating the impact of bacteria, bacteria population reached up to 50% of the algal density, in cell mL1. The size of bacteria and debris ranged from 0.1 to 1 lm. 3.1.2. Aqueous phase characterization Free extracellular proteins ranged from about 0.5 to 2 mgPR L1, the lowest value for the fresh and least concentration biomass and the highest values for the longest-stored and the most concentrated biomass, respectively. Free polysaccharides content ranged from 4 to 17 mgPS L1 (i.e. from 7 to 25 mgPS/gDW), depending on growth and storage conditions (Fig. 3). This concentration was quite similar to the EPS composition determined in another micro-alga species, namely Chlorella, which produces 20 mgPS gDW1 and 2 mgPR gDW1 or less [24]. The average polysaccharides molecular weight was assessed at roughly a few hundred of kDa. Storage conditions I (light, stirring, nutrient addition) could be considered as fed-batch growth conditions. Algae grew and reached a very high cell density, synthesizing at the same time EPS, up to 17 mgPS L1. Bacteria could easily grow in such conditions, using extracellular products as organic carbon source [25]. Storage conditions II were similar to a batch culture. Algae grew as long as nutrients were not limiting. Polysaccharides content per unit of culture volume increased with time. During storage conditions III (low light, no nutrient addition, no stirring), photosynthetic metabolism was off and the dry weight was stationary. Proteins and polysaccharides contents were also constant per unit of culture volume (Fig. 3).

3. Results and discussion 3.2. Membrane selection 3.1. Characterization of algal suspension depending on growth and storage conditions 3.1.1. Particles characterization Cell density reached 2 gDW L1 in selected conditions. For filtration assays, cell density was adjusted to 5–80 million cells mL1, corresponding to 0.05 to 0.7 gDW L1. The size of algal cells was estimated close to 3  7 lm. Biomass also contained some debris such as cell wall fragments which were not quantified. When culture were freshly produced or stored in conditions called ‘‘SCII’’ and ‘‘SCIII’’, we considered the biomass as a ‘‘bacteria-free biomass’’ due to the limited amount of bacteria. We observed a high quantity of bacteria only during the storage conditions ‘‘SCI’’; with this kind of culture, washing the biomass with deionized water permits to eliminate a high amount of bacteria which are therefore mainly in the supernatant and the deionized water fraction (after centrifugation at 2000g during 5 min). Most assays were performed with bacteria-free suspensions freshly produced or stored in conditions SCII or SCIII. In assays

Five different membranes whose characteristics are listed in Table 2 were tested. Water permeability of the five membranes, assessed experimentally matched the values provided by the suppliers, which validated the filtration procedure. According to the obtained data of water flux and to the working constraints and, first considerations about membrane choice were proposed: (i) The two mineral membranes, namely TiO2 and FV, have a high roughness, greater than 500 nm [26], which does not allow algal deposit to be harvested. Moreover, the TiO2 membrane has a very low permeability in water (420 ± 10 L h1 m2 bar1) (Table 2), due to its small pore size of 0.14 lm compared to other membranes. Conversely, the FV membrane has a correct permeability (100 000 ± 10 000 L h1 m2 bar1) (Table 2) but its pore size of 1 lm, in the same range of algal cell size range, cannot guarantee a complete algal retention.

Fig. 3. Change in (A) algal dry weight and (B) polysaccharides content that occur with time, depending on the storage conditions (SCI, SCII, SCIII). The composition of freshly produced biomass corresponds to a 0 day storage time.

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135

(ii) The three organic membranes (PVDF, NC and MEC), with the same pore size of 0.45 lm, presented similar initial water permeability around 30 000 ± 4000 L h1 m2 bar1 (Table 2). Although differing in their material, the three of them have obviously a smooth surface. Roughness is lower than 200 nm according to [27], which enables an easy harvest of the algal deposit. Organic membranes were thus chosen owing to more adapted characteristics to the process objectives. Preliminary filtrations of algal suspensions on organic membranes were carried out. Tests led to a very rapid deposit of algae on the membrane surface. The dead-end filtration of biological suspension at constant pressure is classically described by the cake filtration model, associated to a progressive solids deposit on membrane surface and based on Ruth’s equation [28]. The specific cake resistance is then calculated from the slope of the linear part of the curve obtained when plotting the ratio t/V versus V, where V is the volume filtered after a time t. In the shear rate and mixing conditions employed in this study however, the ratio t/V versus V did not present a linear part which meant that fouling was not associated to a progressive cake deposit and could not be described by a cake filtration model. The same result for the filtration of a micro-algal suspension was obtained by [29]. This extremely rapid deposit was of great interest for our application and could be advantageously exploited. Some of these preliminary filtration tests were accompanied of further experiments with the objective to specify the membrane fouling nature. After a 5 min filtration of a stirred biomass (0.2 gDW L1, 5-days storage in SCIII), the green deposit covering the entire surface of the membrane was eliminated, using the scraping procedure described in the Section 2. Algal deposit was scraped and recovered with 5 mL of deionized water. The solids quantified after drying reached 85 ± 5% of the solids initially present in the suspension. According to the very low (and negligible) solids concentration in the filtrate, it is supposed that 15% of solids were either lost through the rustic method of scraping or very strongly stuck on the membrane. In any event, a great majority of algae seemed to be easily recovered after having been deposited on an organic material (e.g. nitrocellulose), which could thus facilitate the algae recovering for further process step. Fig. 4 presents the permeability change with time in the working conditions described above. Permeability quickly decreased from 24 000 to 570 L h1 m2 bar1, corresponding to a 98% flux

Fig. 4. Change in permeability with time during filtration of a 0.2 gDW L1 (SCIII, 5-day storage) stirred algal suspension on a nitrocellulose membrane.

131

loss. Membrane scraping enabled the partial recovery of the initial permeability. The permeability recovered after each scraping reached 10 000 ± 100 L h1 m2 bar1. It then decreased with time at the same rate as in the first filtration. Consequently, 42% of the fouling was reversible, induced by the instantaneous algae deposit on the membrane and 58% was irreversible, owing to other phenomena like adsorption and/or pore blocking. In biological suspensions, EPS, e.g. proteins and polysaccharides, are largely implied in this type of fouling [30–32]. Organic membranes could thus be re-used for numerous filtration assays since the irreversible fouling part did not rise. Moreover, algae could readily and entirely be recovered from the membrane in a minute volume. These results confirmed that organic membranes, like nitrocellulose membrane, provide a behavior which complies well with the selectivity and process constraints defined above. Consequently, organic membranes were selected and only the nitrocellulose membrane was used in following experiments concerning the optimization of the filtration. 3.3. Identification of the key factors involved in membrane fouling The following experiments and the chosen experimental procedures aimed at identifying the origin of the membrane fouling and the hydrodynamic actions to limit it. 3.3.1. Identification of the fouling compounds by modification of the biomass to be filtered Different procedures were implemented to identify the compounds of the suspension involved in membrane fouling. To highlight the role of the various biomass components, the suspension was fractionated. Algae density and biomass composition were also modified, the latter through storage of the suspension. 3.3.1.1. Fractionation of the suspension: role of the different components. A bacteria-free and a bacteria-contaminated suspension were fractionated according to the procedure described in Fig. 2. Filtration of the different fractions of the bacteria-free biomass (fresh or stored in CSII and CSIII, whatever the time) confirmed the role of algal cells in permeability decrease (Fig. 5A). Permeability was similar for both the whole culture (suspension A on the Fig 2.) and water washed algae (suspension D) at about 12 000 L h1 m2 bar1 after 30 mL filtered. It was reduced by 90% after 80 mL filtered, corresponding to 3 min of filtration (Fig. 5A). Fouling by algal cells was clearly the preponderant fouling phenomenon, regardless of the presence of soluble products. Both non-centrifuged particles (cells or debris) and soluble products induced a low fouling after a 100 mL filtered volume (Fig. 5A). Debris could influence fouling on a 0.01 lm membrane [15]. Here, debris size ranged from 0.1 to 1 lm and could induce pore blocking on the 0.45 lm membrane employed. The permeability loss after 180 mL filtered was 35% and 20% respectively in the crude supernatant (suspension B) and in the washing water (suspension C), meaning that washing water contained less fouling products or particles than the crude supernatant. Analysis of polysaccharides content actually revealed that washing water contained in most cases only 10–20% of the polysaccharides determined in the crude supernatant. The filtration behavior was quite different when many bacteria grew in the algal suspension (Fig. 5B). Permeability measured after 30 mL filtered was about 540 L h1 m2 bar1 for the whole culture (suspension A) and 7300 L h1 m2 bar1 for the water washed algae (suspension D). The presence of bacteria with algal cells readily induced fouling by increasing the cake formation, even when algae were washed. The permeability curves of the crude supernatant (suspension B) and washing water (suspension C) also indicate

132

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135

Fig. 5. Change in permeability with time for different fractions of (A) a 0.3 gDW L1 bacteria-free biomass (SCII, 12-day storage) and (B) a 0.1 gDW L1 bacteria-contaminated biomass (SCI, 6-day storage) filtered with stirring on a nitrocellulose membrane.

the presence of compounds with highly fouling properties. However, the EPS (mainly polysaccharides) in the contaminated supernatant did not rise, as described by Fig. 3. Particles, mainly bacteria and debris, were thus responsible for the flux loss observed in these two fractions coming from the biomass containing bacteria. Small particles such as debris and bacteria as well as soluble products are hence responsible for irreversible membrane fouling, observed on Fig. 4. To avoid the irreversible fouling, culture growth should be optimized in order to limit bacterial proliferation, debris and polysaccharides production. It seems rather difficult to limit the EPS concentration since it is already low in the studied biomass but it is important to avoid working conditions supposed to increase the EPS content (i.e. mechanical stress, starvation stress. . .). Otherwise, a membrane with smaller pore size may be considered. But it could imply retention of more particles and soluble compounds and thus a higher decrease in permeability. An alternative would be to use a 0.45 lm membrane with anti-fouling properties, as described in [33]. 3.3.1.2. Impact of cell density and storage time on the suspension filterability. Previous results highlighted the role of algal cells in the reversible flux loss. The irreversible fouling part originates from very small particles (cell wall fragments for example) and soluble products, whose content was modified during storage. It was attempted to determine which of these two parameters, cell density and storage time mostly influenced permeability loss during filtration. Filtration of the same initial biomass (0.3 gDW L1), with and without preliminary conditioning (concentrated twice by centrifugation (0.55 gDW L1) or diluted twice by dionized water adding (0.15 gDW L1)) showed that permeability loss after a defined time of filtration (5, 10 or 20 min) was closely linked to cell density. Permeability measured after 10 min of filtration was 767, 456 and 263 L h1 m2 bar1 for the 0.15, 0.3 and 0.55 gDW L1 suspensions, respectively; the permeability measured after 20 min of filtration was 360, 240 and 200 L h1 m2 bar1 respectively (Fig. 6). As mentioned above, biomass composition changed with storage time and storage conditions and it could thus impact the fouling behavior. Permeability loss was actually not significantly correlated with storage time. Depending on the storage conditions, permeability after 20 min either decreased with storage time (SC II) or remained nearly constant (SC III). The variation, below 15% when biomass was stored for 15 days, was not significant.

Fig. 6. Effect of cell density (in gDW L1) on permeability after 5, 10 or 20 min of filtration with stirring (suspension stored 6 days under SCIII).

To confirm that algal density, irrespective of growth conditions, was behind the permeability loss, all permeability data measured after 5 and 20 min in the numerous filtration assays performed with bacteria-free biomass were gathered in Fig. 7. Measured permeability was clearly linked with algal density (Fig. 7), with an inversely proportional relationship, whatever the polysaccharides content. The relationship between algal density and permeability is therefore of great interest for process extrapolation. It will permit to evaluate the volume that could be filtered in a given time, knowing algal density, filtration surface and working pressure. Membrane fouling, regardless of the type of phenomena (cake deposit, pore restriction, pore blocking, adsorption), was hence predominantly induced by algal cells. Rise in free soluble products only impacted the reversibility rate of the fouling. This conclusion agrees with some investigations like the one of [34] which attributed 5% of the flux loss to soluble material and more than 60% to particles deposit. Opposite results also mentioned [15] come from differences in the physicochemical properties of the biomass, in operating conditions and in membranes properties. 3.3.2. Hydrodynamic actions to prevent fouling It was necessary to enhance permeability in order to keep a suitable productivity of the operation. Literature data [31] and previous observations suggest that hydrodynamic actions might greatly slow algal deposit on the membrane and thus limit the

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135

133

Fig. 7. Correlation between permeability and algal dry weight (DW): permeability measured (A) after 5 min versus DW; (B) after 20 min versus DW. All suspensions come from bacteria-free biomasses (SCII and SCIII), stored during 0–12 days.

rising of membrane resistance with time. Efficiency of two types of mechanical actions, namely stirring and back-flushing, is reported here. 3.3.2.1. Stirring. The effect of stirring during filtration on permeability was examined for suspensions at different micro-algal density (Fig. 8). When filtering a 0.15 gDW L1 suspension under stirring, the permeability loss after 100 mL filtered was only 44%

whereas it rose to 83% without stirring (Fig. 8A). At high cell density, stirring efficiently suspended the numerous particles and hampered the cake formation. Conversely, filtering a 0.05 gDW L1 suspension did not result in any difference related to stirring because there were few particles to suspend (Fig. 8B). 3.3.2.2. Back-flushing. A 0.24 gDW L1 algal suspension was filtered under stirring for a few minutes then back-flush was applied for a

Fig. 8. Effect of stirring on permeability versus filtered volume during the filtration of a suspension (0-day storage) at (A) low (0.05 gDW L1) or (B) medium (0.15 gDW L1) micro-algal density.

Fig. 9. Effect of back-flush on filtered volume with time of micro-algal suspensions at (A) high density (0.24 gDW L1, 5-day storage, SCII) or (B) low density (0.1 gDW L1; 6-day storage, SCII). Stirring was applied during filtration.

134

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135

short time (see Section 2 for details) (Fig. 9A). The first back-flush enabled to recover a permeability of 4000 L h1 m2 bar1, corresponding to 16% of the initial permeability. When repeating the operation, the permeability recovery decreased over cycles. Back-flush and stirring impacted less the reversible fouling than expected from scraping experiments where 42% of the initial permeability was recovered. The much higher, almost infinite, shear rate on the membrane induced by scraping permits to eliminate particles more strongly attached to the membrane than stirring associated to back-flush which was besides performed at a low non-optimized flow rate. Furthermore, micro-algae concentration in the suspension increased during filtration. As the back-flush time and the flow rate were kept constant, this resulted in a drop in permeability recovery. Combined hydrodynamic actions allowed increasing the filtered volume by 10–30%, after 20 min of filtration, depending on the suspension concentration (Fig. 9). This volume gain was lower for more dense suspensions, due to the increase in algal concentration as mentioned above. Numerous studies report that back-flush is efficient to limit the flux loss and yet is not the ultimate solution for curing the bio-fouling [33]. Both stirring and back-flushing revealed promising results for the scale-up of the filtration step. Cycles of filtration, back-flush and relaxation should now be optimized in order to limit the fouling. Periodic back-flush flow rate particularly should play an important role in eliminating the reversible fouling, especially in the re-suspension of the stacked cells. 4. Conclusion The aim of this study was to verify the feasibility of microfiltration as separation step for algae concentration and separation in a nuclear effluents decontamination process. First, some membranes were chosen depending on selectivity, permeability and process constraints. Secondly, the key factors involved in membrane fouling were identified using different testing procedures relating to modification in the characteristics of the suspension to be filtered (i.e. biomass) or to different filtration working conditions (e.g. hydrodynamic actions). This study showed that organic materials with 0.45 lm pore size were the most suitable regarding process constraints. In addition to having a good selectivity and a high permeability, it permitted to recover 85% of initial algal weight and re-use it. Fouling was mostly induced by micro-algal cells, rather than EPS. Micro-algae induced a reversible fouling. The permeability was related to their density. Owing to the reversible nature of the algae deposit, hydrodynamic actions were able to limit fouling and increase filtered volume. Scale-up outlooks will allow filtrating higher volume of algal suspension with a cell density of the same order of magnitude. Others filtration devices will be sought to rise filtration surface and to optimize hydrodynamic on membrane surface. Acknowledgements The authors thanks the Transversal Nuclear Toxicology Program of the CEA and the ILL for financial support, A. Prat (CEA-DEN, LPSD, Bagnols-sur-Cèze) for supplying filtration material and technical support, A. Heyraud (CERMAV, Grenoble) and E. Kaifas (CEA, LPCV, Grenoble) for the determination of the polysaccharide molecular weight, J.-P. Cadoret (IFREMER, Nantes) for discussions on algal culture and J.-C. Jouin (Protolabo, Mauves sur Loire, France) for the photobioreactor making.

References [1] OSPAR Commission, Liquid Discharges from Nuclear Installations in 2009, OSPAR Commission, 2011. [2] V. Efremenkov, La gestion des déchets radioactifs des centrales nucléaires (Management of radioactive wastes from nuclear plant), IAEA Bull. (1989) 37– 42. [3] IAEA, Combined Methods for Liquid Radioactive Waste Treatment, IAEA, 2003. [4] M.N.V. Prasad, R. Prasad, Nature’s cure for cleanup of contaminated environment – a review of bioremediation strategies, Rev. Environ. Health 27 (2012). [5] M.C. Tomei, A.J. Daugulis, M. Ahmad, Ex situ bioremediation of contaminated soils: an overview of conventional and innovative technologies, Crit. Rev. Environ. Sci. Technol. 43 (2013) 2107–2139. [6] S. Wasi, S. Tabrez, M. Ahmad, Use of Pseudomonas spp. for the bioremediation of environmental pollutants: a review, Environ. Monit. Assess. (2013) 1–9. [7] I.M.M. Gillespie, J.C. Philp, Bioremediation, an environmental remediation technology for the bioeconomy, Trends Biotechnol. 31 (2013) 329–332. [8] H. Shimura, K. Itoh, A. Sugiyama, S. Ichijo, M. Ichijo, F. Furuya, et al., Absorption of radionuclides from the fukushima nuclear accident by a novel algal strain, PLoS One 7 (2012) e44200. [9] A. Gerth, A. Hebner, G. Kiessig, A. Zellmer, Passive treatment of minewater at the schlema-alberoda site, in: P.D.B.J. Merkel, D.-G.A. Hasche-Berger (Eds.), Uranium in the Environment, Springer, Berlin, Heidelberg, 2006, pp. 409–414. [10] W.-M. Wu, J. Carley, M. Fienen, T. Mehlhorn, K. Lowe, J. Nyman, et al., Pilotscale in situ bioremediation of uranium in a highly contaminated aquifer. 1. Conditioning of a treatment zone, Environ. Sci. Technol. 40 (2006) 3978–3985. [11] C. Rivasseau, E. Farhi, A. Atteia, A. Couté, M. Gromova, D. de Gouvion Saint Cyr, et al., An extremely radioresistant green eukaryote for radionuclide biodecontamination in the nuclear industry, Energy Environ. Sci. 6 (2013) 1230. [12] D. Vandamme, I. Foubert, K. Muylaert, Flocculation as a low-cost method for harvesting microalgae for bulk biomass production, Trends Biotechnol. 31 (2013) 233–239. [13] S.L. Pahl, A.K. Lee, T. Kalaitzidis, P.J. Ashman, S. Sathe, D.M. Lewis, Harvesting, thickening and dewatering microalgae biomass, in: M.A. Borowitzka, N.R. Moheimani (Eds.), Algae for Biofuels and Energy, Springer, Netherlands, 2013, pp. 165–185. [14] N. Pragya, K.K. Pandey, P.K. Sahoo, A review on harvesting, oil extraction and biofuels production technologies from microalgae, Renew. Sustain. Energy Rev. 24 (2013) 159–171. [15] W. Zhang, W. Zhang, X. Zhang, P. Amendola, Q. Hu, Y. Chen, Characterization of dissolved organic matters responsible for ultrafiltration membrane fouling in algal harvesting, Algal Res. 2 (2013) 223–229. [16] A. Mishra, B. Jha, Isolation and characterization of extracellular polymeric substances from micro-algae Dunaliella salina under salt stress, Bioresour. Technol. 100 (2009) 3382–3386. [17] R.K. Henderson, A. Baker, S.A. Parsons, B. Jefferson, Characterisation of algogenic organic matter extracted from cyanobacteria, green algae and diatoms, Water Res. 42 (2008) 3435–3445. [18] M. Dubois, K.A. Gilles, J.K. Hamilton, P.A. Rebers, F. Smith, Colorimetric method for determination of sugars and related substances, Anal. Chem. 28 (1956) 350–356. [19] M.T. Hung, J.C. Liu, Microfiltration for separation of green algae from water, Colloids Surf., B 51 (2006) 157–164. [20] L.O. Figura, A.A. Teixeira, Food Physics: Physical Properties – Measurement and Applications, Springer, 2007. [21] J.A. Sánchez Pérez, E.M. Rodríguez Porcel, J.L. Casas López, J.M. Fernández Sevilla, Y. Chisti, Shear rate in stirred tank and bubble column bioreactors, Chem. Eng. J. 124 (2006) 1–5. [22] D.A. Ladner, D.R. Vardon, M.M. Clark, Effects of shear on microfiltration and ultrafiltration fouling by marine bloom-forming algae, J. Membr. Sci. 356 (2010) 33–43. [23] M.H.A. Michels, A.J. van der Goot, N.-H. Norsker, R.H. Wijffels, Effects of shear stress on the microalgae Chaetoceros muelleri, Bioprocess Biosyst. Eng. 33 (2010) 921–927. [24] S. Babel, S. Takizawa, H. Ozaki, Factors affecting seasonal variation of membrane filtration resistance caused by Chlorella algae, Water Res. 36 (2002) 1193–1202. [25] W. Bell, R. Mitchell, Chemotactic and growth responses of marine bacteria to algal extracellular products, Biol. Bull. 143 (1972) 265. [26] Z. Zhong, D. Li, B. Zhang, W. Xing, Membrane surface roughness characterization and its influence on ultrafine particle adhesion, Sep. Purif. Technol. 90 (2012) 140–146. [27] E.F. Barbosa, L.P. Silva, Nanoscale characterization of synthetic polymeric porous membranes: scrutinizing their stiffness, roughness, and chemical composition, J. Membr. Sci. 407–408 (2012) 128–135. [28] R. van Reis, A. Zydney, Bioprocess membrane technology, J. Membr. Sci. 297 (2007) 16–50. [29] S. Babel, S. Takizawa, Microfiltration membrane fouling and cake behavior during algal filtration, Desalination 261 (2010) 46–51. [30] S. Ognier, C. Wisniewski, A. Grasmick, Influence of macromolecule adsorption during filtration of a membrane bioreactor mixed liquor suspension, J. Membr. Sci. 209 (2002) 27–37.

D. de Gouvion Saint Cyr et al. / Separation and Purification Technology 125 (2014) 126–135 [31] J. Orantes, C. Wisniewski, M. Heran, A. Grasmick, The influence of operating conditions on permeability changes in a submerged membrane bioreactor, Sep. Purif. Technol. 52 (2006) 60–66. [32] J. Wu, P. Le-Clech, R.M. Stuetz, A.G. Fane, V. Chen, Effects of relaxation and backwashing conditions on fouling in membrane bioreactor, J. Membr. Sci. 324 (2008) 26–32.

135

[33] T. Hwang, S.-J. Park, Y.-K. Oh, N. Rashid, J.-I. Han, Harvesting of Chlorella sp. KR1 using a cross-flow membrane filtration system equipped with an antifouling membrane, Bioresour. Technol. 139 (2013) 379–382. [34] L. Defrance, M.Y. Jaffrin, B. Gupta, P. Paullier, V. Geaugey, Contribution of various constituents of activated sludge to membrane bioreactor fouling, Bioresour. Technol. 73 (2000) 105–112.