Combined colloidal and organic fouling of FO membranes: The influence of foulant–foulant interactions and ionic strength

Journal of Membrane Science 493 (2015) 539–548

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Combined colloidal and organic fouling of FO membranes: The influence of foulant–foulant interactions and ionic strength Machawe M. Motsa a,c,n, Bhekie B. Mamba b, Arne R.D. Verliefde c a

Department of applied chemistry, University of Johannesburg, P.O Box 17011, Doornfontein 2028, South Africa College of Engineering, Science and Technology, University of South Africa, P.O Box 392, Pretoria 003, South Africa c Department of Applied Analytical and Physical chemistry, University of Gent, Coupure links 653, B-9000 Ghent, Belgium b

art ic l e i nf o

a b s t r a c t

Article history: Received 30 December 2014 Received in revised form 12 June 2015 Accepted 20 June 2015 Available online 9 July 2015

In this work we determined the exact foulant–foulant interactions during combined membrane fouling in forward osmosis (FO). FO fouling tests were performed using sodium alginate and colloidal silica (STZL, 139 nm) as model foulants. The fouling potential of each foulant and their mixture was investigated using feed solutions containing fixed concentrations of Na þ and Ca2 þ (total ionic strength of 0.5 M). Changes on the foulant surface charges upon mixing with cationic species were monitored using zeta potential analysis. The inter-foulant interactions were determined by conducting sequential fouling experiments (the membrane was fouled by each foulants in alternating sequences) as well as Quartz Crystal Microbalance with Dissipating monitoring (QCM-D). Fouling tests with the mixture of alginate and silica colloids resulted in a flux decline trend very similar to that of alginate alone, an observation attributed to the primary effect of alginate during the gel layer formation. QCM-D analysis revealed that alginate macromolecules adsorb onto the surface of silica colloids and thus influencing the subsequent colloid– colloid interactions resulting in a cake layer with similar hydraulic resistance to that formed by alginate alone. There was no evidence of synergistic effects when alginate and silica colloids co-existed mixed in same feed matrix. It was also found that inter-foulant interactions were governed by non-electrostatic forces because at seawater level ionic strength there was excessive reduction of the Debyle length and electric double layer (EDL) such that normal electrostatic forces were greatly suppressed and short range non-electrostatic forces became dominant. The Extended-Derjaguin-Landau-Verwey-Overbeek (XDLVO) approach was used to understand foulant membrane interactions during fouling and it could not compliment the flux decline trends observed from fouling with combined foulant feed stream. The newly proposed approach of determining combined fouling, whereby the membrane was fouled with the two different single foulants in alternating sequences revealed that gel layer hydraulic resistance, hindered colloid back diffusion and cake enhanced osmotic pressure (CEOP) were responsible for flux loss during combined alginate–colloid fouling. & 2015 Elsevier B.V. All rights reserved.

Keywords: Alginate Combined fouling Flux decline Sequential fouling Silica colloids

1. Introduction During any membrane-based filtration process, fouling of all kinds (colloidal deposition, organic adhesion, formation and growth of bacterial biofilms and precipitation of sparingly soluble minerals) impedes membrane performance, resulting in generally higher water production costs [1,2]. FO is no exception to this phenomenon and though fouling in FO is moderate and reversible compared to RO processes. However, the coupled presence of fouling and the unique concentration polarisation phenomenon n Corresponding author at: Department of Applied Analytical and Physical Chemistry, University of Gent, Coupure links 653, B-9000 Ghent, Belgium. Fax: +32 9 264 6242. E-mail address: [email protected] (M.M. Motsa).

http://dx.doi.org/10.1016/j.memsci.2015.06.035 0376-7388/& 2015 Elsevier B.V. All rights reserved.

have catastrophic effects on permeate flux. Earlier studies have focused on investigating the fouling behaviour of FO membranes by single, well-characterized foulants with homogeneous physico-chemical properties [3–7]. And they have demonstrated that membrane fouling by single foulants is strongly dependent on foulant properties (type, size and surface charge density), feed solution chemistry (pH, total ionic strength and concentration divalent cations), membrane orientation and operational conditions such as cross-flow velocity and operational flux [8–12]. However, the experimental insights obtained from studies conducted with single foulants cannot be correctly applied to water and wastewater treatment systems where fouling is almost always caused by more than one type of foulant, with different physicochemical characteristics (i.e. colloidal materials and dissolved organic macromolecules) [1,11].

540

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

This work takes fouling studies in FO a step further toward realworld situations where different types of foulants co-exist. The extensively used sodium alginate (as model organic foulant) was mixed with colloidal silica (as model foulant representative of suspended colloidal matter). The co-existence of organic foulants and silica particles in the presence of divalent cations has been discussed in several studies on NF/RO fouling, but may lead to an even more complicated and unique fouling behaviour in FO membrane processes due to the additional occurrence of the ICP phenomenon [13–15]. Several studies conducted on NF and RO processes have revealed that poly-dispersed suspensions form cake-layer structures with resistances different than mono-dispersed solutions and that the interactions between foulants can be correlated to flux decline behaviour. Contreras et al. [16] evaluated the fouling behaviour of nanofiltration membranes using foulants representative of proteins, polysaccharides and inorganic colloidal foulants in water and wastewater. They hypothesised that three mechanisms were responsible for enhanced membrane flux decline when combined foulants were used: 1) increased hydraulic resistance of the mixed cake layer structure; 2) hindered silica colloids back diffusion in the concentration polarisation layer in the presence of sodium alginate and 3) changes in colloid surface properties due to organic adsorption. Cake-enhanced osmotic pressure (CEOP) was also found to have a dominant effect on permeate flux loss when silica colloids and NOM co-existed in the same feed solution [17]. In another work involving silica colloids and NOM by Li and Elimelech [18], permeate flux loss was also attributed to hindered back diffusion of silica colloids in the presence of the complexed NOM. Amongst the few efforts studying combined fouling in FO processes, Lui and Mi determined the effects of combined fouling by alginate and gypsum in forward osmosis processes. A synergistic effect between alginate fouling and gypsum scaling was observed, mainly through aggravated gypsum scaling in the presence of alginate molecules due to cake-enhanced concentration polarisation (CECP). In addition, alginate molecules acted as nuclei in gypsum crystal growth thus significantly increasing the gypsum crystal size and accelerating deposition [19]. In another recent study on FO by Kim et al. [20], the propensity of combined organic and colloidal fouling in forward osmosis under various solutions chemistries (pH, and calcium ion concentrations) was investigated. The obtained findings demonstrated synergistic effects for combined fouling with alginate and silica colloids, where the overall flux decline was more severe with combined foulants compared to the sum of the individual contributions of alginate and silica colloids alone. All these studies have clearly demonstrated that permeate flux loss during the filtration of feed solutions bearing organic and colloidal foulants is mainly due synergistic effects of the mixture that promote CEOP which in turn greatly reduce the effective driving force. However, the interfoulant (foulant–foulant) interactions that influence the synergy between different foulant species are not clearly defined. The interactions between the membrane and these complex mixtures, particularly at seawater level ionic strength where the surface properties of foulants and membrane maybe greatly altered compared to low ionic strength environments. According to our knowledge, this is the first effort to describe the fouling mechanisms during combined colloidal and organic fouling at seawater level ionic strengths. Therefore, the objective of this study was to identify and determine the exact key interactions between organic and colloidal foulants (alginate and silica colloids) when co-existing in the same feed solution. The impact of the resulting inter-foulant interactions on permeate flux during seawater desalination using FO membranes was determined by conducting a series of fouling tests. Membrane and foulant interfacial free energies were determined

using advanced contact angle measurements; the data was then processed using the Extended-Derjaguin-Landau-Verwey-Overbeek (XDLVO) approach to gain more insight into fouling mechanisms. The observed feed water flux decline trends of the complex foulant mixtures were compared to those obtained with single foulants. In addition, we focused on membrane–foulant and foulant-deposited-foulant interactions by proposing a new method of investigating combined fouling, whereby the membrane was fouled with the two different single foulants in alternating sequences – and the fouling trends observed were compared to those observed for combined fouling.

2. Material and methods 2.1. Membrane and characterisation The commercially available cellulose triacetate (CTA) membrane used in this study was supplied by Hydration Technologies Inc. (Albany, OR). The CTA membrane was chosen for this work due to its extensive characterisation in previous applications, thus making it ideal for our new investigations of characterizing fouling at seawater level ionic strength. It has been found to have a total thickness of approximately 50 mm [19,21]. It was previously tested and recorded a salt rejection of above 95%, a pure water permeability (A) of 0.44 70.12 L m  2 h  1 bar  1 and a salt permeability coefficient (B) of 0.261 70.026 L m  2 h  1 [20]. The membrane structural factor (S) is a parameter that describes the effect of membrane support thickness, porosity and tortuosity on mass transport in the support layer, and is around 481 mm [22,23]. The membrane was stored in ultrapure water at 4 °C prior to use. 2.2. Model foulants and their characterisation Colloidal silica (Snowtex ST-ZL) was used as a model foulant representative of suspended colloidal matter and was supplied by Nissan Chemical Industries, Tokyo, Japan. The data supplied by the manufacturer specified an average particle size of 100 nm. The supplied colloidal suspension contained 40% w/w amorphous silica and 60% w/w water. Alginic acid (provided as sodium alginate) was used as a model organic foulant to represent common polysaccharides which are the main constituents of organic matter in wastewater and seawater. This model foulant was provided by Sigma-Aldrich, St Louis, MO, United States, and was received in powder form. The molecular weight of the alginate was about 12– 80 kDa. An alginate stock solution of 2 g L  1 was prepared by dissolving alginate powder in deionised (DI) water, after which this solution was mixed vigorously for 24 h, and thereafter kept at 4 °C. Dynamic light scattering experiments were performed to analyse the particle/aggregate size of silica colloid and alginate as well as their mixture using a Malvern photon correlation spectrometer (Malvern Instruments, UK). A concentration of 0.5 g L  1 (instead of the 0.2 g L  1 used for fouling experiments) was used for alginate aggregate size determination, to enhance instrument detection. It is worth mentioning that this increase in concentration did not significantly alter the obtained values as proven by other experiments performed with 0.3, 0.75 and 1 g L  1 alginate concentrations. The foulant zeta potentials (for silica colloids and alginate) when exposed to the different electrolyte solutions were calculated from the measured electrophoretic mobility, obtained using a Malvern Zetasizer300 HS series (Malvern Instruments, UK). The concentration of alginate was again increased to 0.5 g L  1 during this analysis to enhance instrument detection.

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

Table 1 The various feed solutions, their composition and corresponding draw solutions. Feed solutions

Solution Solution Solution Solution

Feed composition

DS concentration (MNaCl)

Total ionic strength (mM)

1 2 3 4

Alginate Silica colloids Alginate þSilica colloids Alginate þ0.476 M NaClþ 0.008 M CaCl2 Solution 5 Silica colloids þ 0.476 M NaClþ 0.008 M CaCl2 Solution 6 ST-ZL Silica colloids þ Alginateþ 0.476 M NaClþ 0.008 M CaCl2

0

1.8

500

3.5

2.3. Feed and draw solution chemistries Sodium alginate, a representative of polysaccharides in wastewater and seawater, has been previously identified as a major organic foulant [24]. It is also known to form highly arranged complexes in the presence of divalent cations resulting in severe membrane flux decline [25]. Therefore the feed solutions in this study contained fixed concentrations of cationic species (0.476 M Na þ and 0.008 M Ca2 þ ). The concentrations of alginate and ST-ZL silica colloids in the feed waters for the fouling experiments were fixed at 200 mg L  1 and 1 g L  1 respectively for both single and mixed feed solutions. The total ionic strength of the feed solutions was fixed at 0 and 0.5 M as listed in Table 1 and the corresponding draw solution concentrations were adjusted accordingly to give similar initial fluxes (see Table 1). The differences in the degree of internal concentration polarisation (ICP) and external concentration polarisation (ECP) due to the different feed concentrations were the reason for the large deviation among the draw solutions. No further pH adjustments were conducted; the solutions were tested at their ambient pH (6.90 70.45). No large pH variations were observed during the experiments. 2.4. Forward osmosis membrane test unit and fouling protocol The FO membrane fouling tests were performed using a benchscale cross-flow system; a detailed description of the FO unit is given elsewhere [12]. During filtration the draw solution typically

541

becomes diluted due to water permeation and gradually decreases in concentration which tends to reduce the osmotic drive force. Therefore, an easy method to estimate the changes in draw solution concentration was developed using a conductivity metre (Consort, Model C931, Belgium). A programme-controlled (LabVIEW software) 3-way valve was installed on the DS return tube just before it enters the draw solution tank (Fig. 1). The valve temporally directs (at set intervals) the draw solution into a filter funnel containing dry solid salt (NaCl) after being triggered by a decline in draw solution conductivity. The dissolved salt then dripped into the bulk draw solution to correct the dropping solution conductivity and maintain constant draw solute concentration. The electrolyte solution resulting from the dissolved NaCl (which was used as a draw and a feed solution) enabled the use of conductivity to monitor ionic concentration via conductivity measurements. However, due to the non-ideal nature of solution especially at our working concentrations; the effective (measured ionic concentration) is usually lower than the real concentration, thus this method was the closest estimation of the ionic strength of the feed and draw solution. The membrane performance with respect to fouling resilience was tested in two manners: 1) by “traditional” fouling of the membrane with the single foulants and their mixtures (combined fouling), and 2) by fouling the membrane in alternating sequences of the two foulants, as well as their mixture. “Traditional” fouling tests with single foulants and their mixture were performed over a 24 h period while sequential fouling tests were conducted over two 24 h segments (48 h). All fouling experiments were conducted with the membrane oriented with the active layer facing the draw solution (PRO mode) to aggravate membrane fouling and clearly distinguish fouling mechanisms. A new membrane coupon was used for each experiment and prior to each fouling experiment, a baseline experiment was performed using a feed solution of MilliQ with similar ionic strength but lacking the foulant. During the experiment, the draw and feed solutions were circulated on each side of the membrane at a flow of 7.0 cm s  1 counter-current configuration. An electronic balance (Defender, Ohaus, USA) connected to a data-logging system was used to record the water flux at pre-determined time intervals. The average water flux values of experimental runs were then normalised to the initial water flux. The initial permeate flux was constant at 15.84 70.75 L m  2 h  1 for all conducted experiments.

Pump

Feed tank

2

3

FO membrane cell FO membrane

1 4

Pump

Magnetic stirrer

Balance

Relay board Conductivity meter PC 1. Draw solution tank, 2. Filter funnel with salt, 3. 3-way valve, 4. Draw over-flow vessel Fig. 1. Illustration of the laboratory-scale FO test unit.

542

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

2.5. Contact angle measurements Contact angle measurements were used to determine the surface free energy components of the CTA membrane and foulants using the Young-Dupré equation (Eq. (1)). This equation links the contact angle of a drop liquid placed on a solid surfcace wit the surface tension components of the liquid, according to van Oss [26]

γl(1 + cosθ ) = 2( γsLW γl LW +

γs+ γl− +

γs−γl + )

(1)

where:

Δm = −

θ is the measured contact angle, γLW is the Lifshitz–van der Waals free energy component, γ þ is the electron–acceptor component, γ  is the electron donor component; subscripts s and l designate the solid and liquid phases, respectively. Surface tension components of a membrane can be determined by measuring contact angles between the membrane and three well-characterized probe liquids with known surface tension components, such as deionised water, glycerol and diiodomethane. The contact angles due to these probe liquids were measured on filtered lawns of alginate, silica colloids and their mixture using a computerised Krüss DSA 10-MK2 (Krüss, Germany) contact angle goniometer. A full description of how the contact angle analysis was conducted can be found in the Appendix. The total interfacial free energies (interfacial free energy of cohesion and adhesion) per unit area between solid materials 1 and 2 immersed in liquid medium 3 were determined from the sum of the Lifshitz–van der Waals (LW) and acid–base (AB) components as expressed by Eq. (4) LW ΔG132 =2

(

γ3LW −

AB ΔG132 = 2 γ3+

− 2

water that was pumped across the sensors for 1 h. The temperature in the flow chambers was set at 18 °C and the solution were kept at 22 °C in a water bath. Adsorption is indicated by changes in the vibration frequency of the piezoelectric quartz-crystal sensor and the amount of organic foulant macromolecule adsorbed can be calculated from the frequency change using the Sauerbrey equation (Eq. (5)) [27], which relates the mass change per unit area at QCM electrode surface to the observed change in oscillation frequency of the crystal

(

γ1LW

γ1− +

γ1+γ2−

)(

γ2LW −

γ2− −

− 2

)

γ3LW

)

γ3− + 2 γ3−

(2)

(

γ1+ +

γ2+ −

γ1−γ2+

TOT LW AB ΔG132 = ΔG132 + ΔG132

γ3+

) (3) (4)

TOT If surfaces 1 and 2 are the same, then ΔG131 indicates the interfacial free energy of cohesion. The interfacial free energy of cohesion, ΔG131, describes energetic favourability of a solid material (1) interacting through a liquid medium (3) with itself (1). Cohesive free energy offers insight into particles stability as well as particle deposition onto surfaces already covered by the same particles. The interfacial free energy of adhesion, ΔG132, describes the attraction or repulsion of solid material (1) interacting with another solid material (2) through a liquid medium (3). The adhesive free energy relates to adsorption and adhesion of dissimilar materials.

2.6. Measurement of alginate adsorption on silica colloids The adsorption of alginate macromolecules on silica was studied using quartz crystal microbalance with dissipating monitoring (QCM-D) technique with silica sensors (Q-Sense, Sweden). The silica-coated quartz crystal sensors were used to simulate the surface of the silica particles. The same ionic strength (0.5 M) and foulant concentration (200 mg L  1) used for cross-flow fouling tests was used for the adsorption studies. Prior to analysis ultrapure water was pumped across the sensors for 15 min at 0.5 mL/ min after which the foulant solution was introduced. At t¼100 min, the desorption step was initiated using ultrapure

CQCM n

Δf

(5)

where: CQCM is the mass sensitivity constant (17.77 ng cm  2 Hz  1 at f ¼5 MHz). Δf is the change in the resonance frequency. n is the overtone number (n has values 1, 3, 5,…). CQCM is a constant independent of the overtone number/order, n.

3. Results and discussion 3.1. Characterisation of model foulants The resulting foulant sizes and surface charges after both foulants were mixed with the elevated concentrations of Na þ and Ca2 þ are presented in Table 2. Both alginate and colloidal silica recorded negative zeta potential values, indicating their negative surface charge. The colloidal silica exhibited a slightly more negative charge (  54 mV) compared to alginate (  47 mV) in the absence of cations. The presence of cationic species (Na þ and Ca2 þ ) substantially reduced the surface charge of both foulants and their combination, mainly through surface charge neutralization. However, the extent of neutralization was not the same for both foulants, with alginate being mostly neutralised; recording  8.85 versus the 14.2 mV for the colloids. This could be due to the fact that the amount of silica colloids (1 g L  1) used for purposes of this study was too high to be completely neutralised. A similar observation was made when the silica and alginate coexisted, the particles' surface charges was reduced from  72.15 to  15.2 mV. According to the literature, the addition of salts, particularly calcium salts, to alginate-containing solutions should result in the formation of a highly organised gel-type conformation that should increase both the alginate aggregate size [28]. The significant increase in size was evident from the measured hydrodynamic diameter of alginate in the presence of Ca2 þ as listed in Table 2. The size and charge results of the combined or mixture of the two foulants should be viewed with reservation since the Table 2 Measured zeta potential and hydrodynamic diameters for alginate and silica colloids in different solution conditions. Sample

Zeta potential (mV)

Hydrodynamic diameter (nm)

ST-ZL ALG ST-ZL þ ALG *ST-ZL (0.5) *ALG (0.5) *ST-ZL þALG (0.5)

 54.4  47.5  72.15  14.2  8.85  15.2

139.9 66.02 184.9 156.35 535.4 177.8

ST-ZL: Model silica colloids. ALG: Alginate. *

0.5 M feed ionic strength.

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

Normalized flux (J/Jo)

1.0 0.8 0.6 0.4 Baseline ST-ZL SC ALG Combined

0.2 0.0 0

5

10

15

20

25

15

20

25

Time (h)

1.2

Normalized flux (J/Jo)

1.0 0.8 0.6 0.4 Baseline ST-ZL SC ALG Combined

0.2 0.0

0

5

10 Time (h)

Fig. 2. (a) Permeate flux decline curves due to fouling with single foulants (ST-ZL colloidal silica, 1 g L  1and alginate, 200 mg L  1) and their combination without addition of mono and divalent salts. The draw solution was adjusted to give similar flux to that of fouling in the presence of background electrolytes. Fouling experiments were performed in PRO mode at a cross-flow velocity of 4.0 cm s  1 for 24 h. (b) Permeate flux decline curves due to fouling with single foulants and their combination in the presence of 476 mM NaCl and 8 mM CaCl2. The membrane area was deemed representative and fouling experiments were performed once.

colloidal silica was the most active light-scattering species during DLS analysis, mainly due to the particle concentration, size and shape; therefore the recorded particle size of the mixed foulants of 177.8 nm could have been highly influenced by the silica colloids since it was close to that of silica alone (156.35 nm). 3.2. Membrane fouling behaviour caused by single and combined foulants When the membrane was fouled with both foulants and their mixture in the absence of abundant cationic species, no severe fouling was observed and all foulants resulted in almost identical fouling profiles (Fig. 2a). However, when calcium ions together with a high concentration of sodium ions (0.476 M) were added, clear distinctions in fouling rates were observed between the different foulants (Fig. 2b). Fig. 2b also shows that during the earlier stages of filtration the baseline experiment (without a foulant) exhibited a similar flux loss rate as the ST-ZL colloids. This observed flux loss was mainly caused by the increased feed solution concentration which then reduced the effective driving force.

However, after about 8 h of filtration, the silica colloid permeate flux trend deviate from that of baseline indicating continual loss in permeate flux was noticed over the entire filtration period. On the other hand, alginate fouling was distinct resulting in a high initial flux loss rate compared to that of the baseline experiment indicating that flux loss was mainly due to foulant deposition. In most studies on combined fouling, tests involving the mixture of alginate and silica colloids resulted in severe permeate flux decline [29]. This observation has mainly been attributed to the synergy between the two foulants manifested through increased fouling rate compared to that of single foulants. However, interestingly, in this study, when the two foulants co-existed in the same feed stream, they exhibited a fouling trend very similar to that of alginate alone (Fig. 2b), contradicting several observations made in RO studies that reported a clear synergy between the two foulants at lower ionic strength [18,20]. This deviation in fouling behaviour is mainly due to the structures of the fouling layers resulting from the different membrane processes. In FO, the lack of hydraulic pressure allows for the formation of relatively loose cake layer where foulant deposition is dependent on permeation drag, while the RO process results in a compact and cohesive fouling layer that has more resistance to permeate flow, resulting in higher flux decline rates. It also indicates that the resulting gel layer due to combined alginate–silica fouling had a similar hydraulic resistance to that of alginate alone and the alginate aggregates had a primary role in the cake layer formation during combined fouling. One over arching observation was that despite the different initial flux loss rates for alginate, ST-ZL silica colloids, after 20 h fluxes were similar for all the three feed solutions. We therefore, hypothesise that this is due to the fact that at this point the permeate flux had dropped to such low levels due to fouling and ICP that the existing permeate drag is no longer sufficient to cause foulant deposition on the membrane. As a result, the flux levels off (around 4 L h  1 m  2) regardless of the feed type resulting in the stable flux point observed from 20 h. We hypothesise that this is more or less comparable to the “limiting flux” observed in pressure-driven membrane applications. It has been reported that organic fouling is highly influenced by the presence of Ca2 þ ions in the feed stream [28] and the flux loss trends shown in Fig. 3 clearly indicate the effect of calcium ions on alginate during combined alginate–silica fouling. Severe permeate flux loss was observed with the addition of calcium ions due to gel layer formation. The obtained results of combined alginate–silica colloids fouling are complex and have important implications on

1.2 1.0 Normalized flux (J/Jo)

1.2

543

0.8 0.6 0.4 0.2 0.0

0

5

10

15

20

25

Time (h) Fig. 3. Permeate flux decline as a function of time for membranes fouled with a mixture of alginate and ST-ZL colloidal silica with and without 8 mM Ca2 þ . The total ionic strength was fixed at 0.5 M.

544

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

membrane flux decline during seawater treatment, thus motivating the exploration of a new approach of investigating inter-foulant interactions where the membrane was fouled with each single foulant in alternating sequences. Further discussions and observations made on this approach are discussed in detail in Section 3.4. It is also worth mentioning that the feed ionic strength played a significant role in the observed flux loss. The high feed stream ionic strength (0.5 M) greatly reduced the Debye length and electric double layer (EDL) such that the effect of normal particle– particle electrostatic forces was largely suppressed and shortrange non-electrostatic interactions became dominant [30]. The excessive reduction in repulsive forces amongst foulant particles results in the formation of a more densely packed cake layer on the membrane surface, which consequently causes a greater resistance to permeate flow (this is later explained using the Extended Derjaguin–Landau–Verwery–Overbeek approach (XDLVO)) [31]. Thus, the permeate flux loss at seawater level ionic strength can be partly attributed to the significant role of the non-electrostatic forces between the foulants and the membrane, as well as between the on-coming foulant particles and the already deposited cake layer [32,33]. 3.3. Interfacial surface free energies Further insight (elucidate) into the membrane fouling process was attained by computing the surface free energies for the membrane and foulants using measured contact angles of diiodomethane, glycerol and the salt solution containing Na þ and Ca2 þ . The resulting interfacial energies are displayed in Table 3. Alginate exhibited the most negative energy of cohesion (  22.07 mJ m  2) suggesting that the macromolecules would rather form an interface among themselves (strong hydrophobic attraction) in the feed solution, rather than forming an interface with water. This also implies that alginate is more likely to be deposited on a membrane surface already covered with alginate aggregates [34,35]. On the other hand, the ST-ZL colloidal silica exhibited a strongly positive energy of cohesion (þ 25.19 mJ m  2) implying that the colloids would rather form large interfaces with water other than aggregating, and as such indicating that for the silica colloids, the critical coagulation concentration has not been reached and colloidal particle deposition onto the membrane could only be prompted by the presence of a strong permeate drag force. The computed energy of cohesion for the alginate–silica colloids mixture was  2.51 mJ m  2. This energy value—when compared to that of the single foulants—suggests that inter-foulant interactions of foulants were partly dominated by alginate. There were weak hydrophobic attraction forces amongst alginate and silica colloids as shown by the weakly negative cohesion energy value. Similar observations were made with the energy of adhesion; alginate had the most negative value, the colloids recorded a strong positive energy value, while the energy value for the mixture of alginate and silica colloids was slightly negative. This means that during the initial stages of membrane fouling the alginate macromolecules would attach spontaneously upon Table 3 Interfacial free energy of cohesion and adhesion for alginate, colloidal silica and their mixture. Foulant

Energy of cohesion, ΔG131 (mJ m  2)

Energy of adhesion, ΔG132 (mJ m  2)

Initial flux decline rate (h  1)

ALG ST-ZL ALG þ ST-ZL

 22.07 þ 24.19  2.51

 20.16 þ16.74  2.82

0.127 0.0319 0.129

deposition onto the membrane surface, leading to a high initial permeate flux loss rate as shown in Fig. 2b and Table 3. As already indicated by the strong positive adhesion energy, the silica colloids will only be deposited on the membrane surface under conditions of high permeate drag force [36,37]. This was supported by the low flux decline rate (high permeation rate) at the initial stages of fouling experiments (Table 3) as well as the positive energy of adhesion (16.74 mJ m  2).Unlike the alginate or colloidal silica, the adhesion energy value for the mixture of alginate and colloidal silica was slightly negative (  2.82 mJ m  2) implying that the affinity for the membrane surface was low. It could then be suggested that permeation drag forces were dominant in the deposition of the combined alginate–silica aggregates resulting in a higher initial flux decline rate as that observed during alginate fouling. This is further complemented by the size of the primary foulant (alginate; Table 2). This implies that the contact angle approach followed in this work could not accurately define foulant–membrane interactions of multi-component feed streams. In addition, interactions energies determined for the mixture from filtered lawns might not reflect full foulant behaviour in FO applications. Therefore, more insight is required to understand the exact foulant–foulant interactions when alginate co-exists with silica colloids. An attempt to address this was made in the next sections. 3.4. Sequential membrane fouling 3.4.1. Inter-foulant interactions and cake layer formation In a bid to further elucidate possible inter-foulant interactions during combined organic and colloidal fouling, we applied a new approach of investigating membrane fouling. In this approach, the membrane was fouled with the two foulants (colloidal silica and alginate) in alternating sequences. Therefore, the second fouling trend was greatly influenced by the cake layer developed during the first fouling run. There was a significant jump in permeate flux when switching from one foulant species to another. This curious regain in flux was only due a mismatch in the feed ionic strengths, that is, the subsequent fresh feed's ionic strength was lower compared to the previous feed and as such the osmotic driving force was increased temporally restoring permeate flux. The effect of the ionic strength differences is investigated in detail in the next section. However, directly after the start of the second fouling runs, there was a conspicuous decline on the regained permeate flux to an almost stable flux points marked i, ii and iii in Fig. 4 for ALG–ST-ZL SC, STZL SC–ALG and combined–combined sequences. This sharp permeate flux loss might be attributed to increase in gel layer resistance and cake-enhanced osmotic pressure (CEOP) due to the high concentration of salt at the membrane surface that severely reduces effective osmotic pressure. The combined alginate–colloid feed had the steepest slope indication that the formed gel layer could enhance CEOP [38]. Another striking observation was that the stable flux points (flux levelling-off point) were different for the three sequences: the ST-ZL SC–ALG sequence recorded no further flux decline below 40% of the initial flux, 50% was the stable flux point for the reverse sequence (ALG–ST-ZL SC) and the combined–combined sequence was stable at 37% (more or less comparable to ST-ZL SC–ALG). These observations reveal that during combined alginate–colloid fouling hindered colloid back diffusion, hydraulic resistance and CEOP play a major role in permeate flux loss while hydraulic resistance and CEOP were the main mechanisms during alginate fouling. 3.4.2. Influence of Ionic strength on foulant deposition during sequential filtration experiments In order to investigate the influence of feed ionic strength on

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

545

1.2

0.8

Normalized flux

Normalized flux (J/Jo)

1.0

0.6 i ii

0.4 iii Baseline ST-ZL SC - ALG ALG - ST-ZL SC Combined-Combined

0.2 0.0 0

10

20 Time (h)

30

40

Time (h)

Fig. 4. Sequential membrane fouling experiments with single foulants in alternating sequences. A fresh feed solution with an ionic strength of 0.5 M was used for the subsequent fouling run. Fouling conditions were kept the same as in previously conducted experiments.

the partial flux recovery and whether the real flux decline in the consecutive fouling experiments was due to CEOP or hydraulic resistance—or a combination of both, another series of experiments were carried out whereby the ionic strength of the second feed solution was adjusted to the same value as the feed solution at the end of the first fouling run, by matching the conductivity of the feed solution at the start of the second fouling run to that of the final conductivity value of the first fouling run through the addition of sodium chloride. Unlike what was observed in the previous section (Fig. 4), no flux was regained for the fouling sequence of ALG–ST-ZL SC and combined–combined. The flux decline trends followed a clear continuation from the previous experiments. This provides evidence that the temporal regain in flux observed in the previous set of experiments for these combinations of foulants was mainly due to the low ionic strength of the new feed. In contrast, a large proportion of permeate flux (about 80%) was recovered when a second feed of alginate was used to foul the membrane already covered by colloidal silica. This suggests that alginate might have an abrasive effect of on the colloidal silica already deposited, an observation that serves as proof that alginate adsorbs on the silica colloids (and subsequently tearing the silica layer off the membrane surface leading to a significant period of permeation flux recovery). Further proof of alginate's abrasive effect on the silica

Fig. 6. Consecutive membrane fouling experiments performed with colloidal silica. The feed solution was adjusted to the same ionic strength as that of the previous feed solution.

layer is provided by Figs. 5 and 6. When silica is used instead of alginate in the second fouling run to further foul the silica-fouled membrane (after correcting the ionic strength), no permeate flux recovery is observed but a continual flux decline until the stable flux point. This shows that it is through specific adsorption of alginate onto the silica, that the combined fouling layers are formed. This could also be an indication why the fouling trends mainly seem to follow the characteristics of the alginate. The interaction between alginate macromolecules was probed using QCM-D and the results are shown in Fig. 7. After the injection of the alginate solution (at t ¼15 min) there was a rapid change in the frequency to  59 Hz an indication of alginate adsorption on the silica. This was supported by the significant rise in dissipated energy after the introduction of alginate (Fig. B1). This observation interaction between the two model foulants is completely different from what has been previously reported at low/ moderate ionic strength which suggests that the extreme reduction in Debyle screening length allowed for non-electrostatic interactions between alginate and silica colloids.

500

Areal mass of alginate (ng/cm 2)

1.2

Normalized flux (J/Jo)

1.0 0.8 0.6 0.4 0.2 0.0

ST-ZL SC - ALG ALG - ST-ZL SC Combined - Combined

0

10

20 Time (h)

30

40

Fig. 5. Sequential membrane fouling experiments with single foulants in alternating sequences. The feed solution was adjusted to the same ionic strength as the previous feed solution. Fouling conditions were kept the same as in previously conducted experiments.

400

300

200

100 A

C

B

0 0

20

40

60

80

100

120

140

160

Time (min) Fig. 7. Areal mass of the adsorbed alginate layer on the silica particle surface as a function of time. Stage A: ultrapure water; stage 2: introduction of 200 mg L  1 in 0.5 M ionic strength; stage 3: desorption with ultrapure water.

546

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

4. Conclusions The fouling trend obtained from combined alginate–colloids feed was very similar to that of alginate alone, an indication that the combined foulant gel layer had the same hydraulic resistance as that of alginate alone. It was also revealed that alginate had a primary effect on the cake layer formation during combined fouling; that is the presence of alginate resulted in faster permeate flux decline. The new approach of sequential fouling experiments and QCM-D analysis provided proof of the adsorption of alginate onto the surface of silica particles surfaces which led to altered colloid–colloid interactions. There was no clear evidence of synergy between alginate and the silica colloids during combined fouling, since combined fouling layer had the same characteristics as the alginate fouling layer. Hindered colloid back diffusion in the presence of alginate gel layer and cake enhanced osmotic pressure were most likely the mechanisms responsible for the observed flux decline during combined fouling. There was no clear evidence of synergy between alginate and the silica colloids during combined fouling, since combined fouling layer had the same characteristics as the alginate fouling layer. The feed ionic strength also influenced permeate flux loss, manifested through the excessive reduced Debyle screening length and electric double layer forces such that the electrostatic forces were suppressed and only nonelectrostatic forces became dominant in foulant–foulant and

foulant–membrane interactions. The contact angle approach used in this work could not clearly define foulant–membrane interactions of combined foulants feed streams. The new approach of sequential fouling experiments revealed that alginate adsorbs on silica colloids, and it is most likely through this specific adsorption that combined fouling results in fouling layers with similar characteristics to that of the alginate fouling layer. This study has brought different insights into the interpretation of combined fouling behaviour, though improvements could be made in developing newer models that could account for the surface interaction during combined fouling.

Acknowledgements This research work was financially supported by the University of Ghent, Belgium through BOF Special Research fund (Account No. B/12493/01 iv1) and the DST/Mintek Nanotechnology Innovation Centre (DST/NIC), South Africa (Cost centre No. 05.15.075885.15).

Appendix A. : Scanning electron microscopy (SEM) Scanning electron microscopy (SEM) was used to determine the macro-scale surface morphology of the different fouling layers and

Fig. A1. SEM micrographs of FO membrane surface fouled with: (a) ST-ZL colloidal silica, (b) alginate and their mixture (c, d). The feed solution ionic composition was made up of 476 mM NaClþ 8 mM CaCl2.

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

the obtained micrographs showed the appearance of the different cake layers after fouling the FO membrane surface with the different single foulants and their mixtures. The silica resulting silica colloids cake layer is shown in Fig. A1a, while the alginate fouling layer showed is displayed by Fig. A1b. When the two foulants were mixed, a crumb-like cake layer was formed that appeared to be made up of layers of alginate–silica aggregates (Fig. A1 c and d).

Appendix B. : Contact angle measurements The surface tensions of alginate due to the different electrolyte solutions were determined by first dissolving it in ultrapure water and deposited onto a NF membrane using a dead-end membrane filtration system for a period of 8 h a method which was also implemented by Jin et al. [39]. The filtered lawns of alginate were then allowed to dry for 12 h in a dessicator. Contact angle measurements of the different salt solutions, deionised water and diiodomethane were then performed using the sessile drop method. However when determining the clean membrane surface tensions two methods were employed: (1) sessile drop on a dry membrane surface and (2) sessile drop measurement on a hydrated membrane surface as previously conducted by Botton and workers, [40] on their work with NF membranes. After drying the membrane samples in a sealed desiccator for 12 h, they were then attached to a glass slide and a minimum of 10 drops were measured per membrane sample for all three probe

547

liquids. The average contact angle was used for computation of surface tensions and interfacial free energies. The hydrated method aimed at delaying water evaporation during contact angle measurement as well as simulating the liquid–membrane contact during the filtration process. Membrane samples were placed on a wet filter paper to curb the effect of water evaporation. Sessile drops of the three probe liquids were placed on the hydrated membrane surface with the contact angle measured within seconds. Drops were placed at intervals of 3–5 min for a period of about 1.5 h. For each membrane sample a minimum of 10 drops were measured. Measuring the liquid drop contact angle on the dry CTA membrane gave an over-estimation of the contact angles since the contact angle is quantified while the membrane is still wetting leading to higher values that are not accurate representatives of the membrane surface tensions. Therefore only the sessile drops on the hydrated membrane surface were used for the computation of membrane surface tensions.

Appendix C. : Alginate adsorption onto silica particles The interactions between foulants (alginate and silica colloids) were studied with QCM-D and the resulting frequency shifts and dissipated energy are shown in Fig. B1a. To relate the observed changes in dissipation, D to changes in frequency, f, we plotted the D–f (dissipation–frequency) plot shown Fig. B1b and it reveals that the density of points is smaller near the origin. An indication of faster kinetics at the beginning and it slows down as the frequency shift is reduced.

22

10

References

20

0

18 -10

14

Frequency (Hz)

-20

12

-30

10 -40

8 6

-50

Dissipation (1E-06)

16

4

-60

2 -70

0 -2

-80 -20

0

20

40

60

80

100

120

140

160

180

Time (min) 20

15

D

10

∇ 5

0

-5 0

-20

-40

-60

-80

f (Hz)

∇

Fig. B1. Frequency shifts corresponding to interactions between alginate and silica particles, (a) change in frequency and dissipation as a function of time during the adsorption and desorption analysis of alginate and (b) D–f plots showing dissipation against frequency.

[1] B. van der Bruggen, C. Vandecasteele, T. Van Gestel, W. Doyen, R. Leysen, A review of pressure-driven membrane processes in wastewater treatment drinking water production, Environ. Prog. 22 (2003) 46–56. [2] D. Jermann, W. Pronk, M. Boller, Mutual influences between natural organic matter and inorganic particles and their combined effect on ultrafiltration membrane fouling, Environ. Sci. Technol. 42 (2008) 9129–9136. [3] B. Mi, M. Elimelech, Chemical and physical aspects of organic fouling of forward osmosis membranes, J. Membr. Sci. 320 (2008) 292–302. [4] S. Hong, S. Lee, C. Boo, M. Elimelech, Comparison of fouling behaviour in forward osmosis (FO) and reverse osmosis (RO), J. Membr. Sci. 365 (2010) 34–39. [5] K. Katsoufidiou, S.G. Yiantsios, A.J. Karabelas, UF membrane fouling by mixtures of humic acids and sodium alginate: fouling mechanisms and reversibility, J. Membr. Sci. 264 (2010) 220–227. [6] H. Yong Ng, V. Parida, Forward osmosis organic fouling: effects of organic loading, calcium and membrane orientation, Desalination 312 (2012) 88–98. [7] M. Xie, D.L. Nghiem, W.E. Price, M. Elimelech, Impact of organic and colloidal fouling on trace organic contaminant rejection by forward osmosis: role initial permeate flux, Desalination 336 (2014) 146–152. [8] T. Chung, S. Zhang, Y.K. Wang, J. Su, M.M. Ling, Forward osmosis processes: yesterday, today and tomorrow, Desalination 287 (2011) 78–81. [9] A. Altaee, Forward osmosis: potential use in desalination and water reuse, Membr. Sep. Purif. Technol. 1 (2012) 79–93. [10] E. Arkhangelsky, F. Wicaksana, C. Tang, A.A. Al-Rabiah, M.S. Al-Zahrani, R. Wang, Impact of intrinsic properties of foulants on membrane performance in osmotic desalination applications, Sep. Purif. Technol. 123 (2014) 87–95. [11] Y. Xiong, Y. Liu, B. Mi, Y. Leng, Hydrated polyamide membrane and its interaction with alginate: a molecular dynamic study, Langmuir 29 (2013) 11600–11608. [12] M.M. Motsa, B.B. Mamba, A.D. Haese, E.M.V. Hoek, A.R.D. Verliefde, Organic fouling on forward osmosis membranes: the role of feed solution chemistry and membrane structural properties, J. Membr. Sci. 460 (2014) 99–109. [13] T.Y. Chuyang, Y. Kwon, J.O. Leckie, The role of foulant-foulant electrostatic interaction on limiting flux for RO and NF membranes during humic acid fouling – theoretical basis experimental evidence, and AFM interaction force measurements, J. Membr. Sci. 326 (2009) 526–532. [14] P.K. Lee, T.C. Arnot, D. Mattia, A review of reverse osmosis membrane materials for desalination—development to date and future potential, J. Membr. Sci. 370 (2011) 1–22. [15] N.D. Long, S. Hawkes, Effects of membrane fouling on the nanofiltration of trace organic contaminants, Desalination 236 (2009) 273–281. [16] A.E. Contreras, A. Kim, Q. Li, Combined fouling of nanofiltration membranes:

548

M.M. Motsa et al. / Journal of Membrane Science 493 (2015) 539–548

mechanisms and effect of organic matter, J. Membr. Sci. 327 (2009) 87–95. [17] S. Lee, J. Cho, M. Elimelech, Combined influence of natural organic matter (NOM) and colloidal particles on nanofiltration membrane fouling, J. Membr. Sci. 262 (2005) 27–41. [18] Q. Li, M. Elimelech, Synergistic effects in combined fouling of loose nanofiltration membrane by colloidal material and organic matter, J. Membr. Sci. 278 (2006) 72–82. [19] Y. Liu, B. Mi, Combined fouling of forward osmosis membranes: synergistic foulant interaction and direct observation of fouling layer formation, J. Membr. Sci. 407–408 (2012) 136–144. [20] Y. Kim, M. Elimelech, K.K. Shon, S. Hong, Combined organic and colloidal fouling in forward osmosis: fouling reversibility and the role of applied pressure, J. Membr. Sci. 460 (2014) 206–212. [21] T.Y. Cath, A.E. Childress, M. Elimelech, Forward osmosis: principles, applications and recent developments, J. Membr. Sci. 281 (2006) 70–87. [22] K. Gerstandt, K.-V. Peinemann, S.E. Skilhagen, T. Thorsen, T. Holt, Membrane processes in energy supply for an osmotic power plant, Desalination 224 (1–3) (2008) 64–70. [23] M. Park, J.J. Lee, S. Lee, J.H. Kim, Determination of a constant membrane structure parameter in forward osmosis processes, J. Membr. Sci. 375 (1) (2011) 241–248. [24] E. Arkhangelsky, F. Wicaksana, C. Tang, A.A. Al-Rabiah, M.S. Al-Zahrani, R. Wang, Combined organic–inorganic fouling of forward osmosis hollow fibre membranes, Water Res. 46 (2012) 6329–6338. [25] V.R. Linares, V. Yangali-Quintanilla, Z. Li, G. Amy, NOM and TEP fouling of a forward osmosis (FO) membrane: foulant identification and cleaning, J. Membr. Sci. 421–422 (2012) 217–224. [26] C.J. Van Oss, Acid–base interfacial interactions in aqueous media, Colloids Surf. A: Physicochem. Eng. Aspects 78 (1993) 1–9. [27] F. Hook, M. Rodahl, B. Kasemo, P. Brezinski, Structural changes in hemoglobin during adsorption to solid surfaces: effects of pH, ionic strength and ligand binding, Biophysics 95 (1998) 12271–12276. [28] K. Listiarini, L. Tan, D.D. Sun, J.O. Leckie, Systematic study on calcium–alginate interaction in a hybrid coagulation–nanofiltration system, J. Membr. Sci. 370

(2011) 109–115. [29] Z. Xiaohua, M. Elimelech, Colloidal fouling of reverse osmosis membranes: measurements and fouling mechanisms, Environ. Sci. Technol. 31 (1997) 3654–3662. [30] T.Y. Chiu, A.E. James, Electrokinetic characterisation technique on asymmetric microfiltration membranes, Colloids Surf. A: Physicochem. Eng. Aspects 301 (2007) 281–288. [31] M. Mishuchuk, Electric double layer and electrostatic interaction of hydrophobic particles, J. Colloid Interface Sci. 320 (2008) 599–607. [32] J. Hansen, H. Lowen, Effective interactions between electric double layers, Ann. Rev. Phys. Chem. 51 (2000) 209–242. [33] K. Schießl, F. Babick, M. Stintz, Calculation of double layer interaction between colloidal aggregates, Adv. Powder Technol. 23 (2012) 139–147. [34] G. Milojka, G. Sinn, W. Gindl, A. Reiterer, S. Tschegg, A comparison of different methods to calculate the surface free energy of wood using contact angle measurements, Colloids Surf. A: Physicochem. Eng. Aspects 181 (2001) 279–287. [35] J. Troger, K. Lunkwitz, Burger Wolfgang, Determination of the surface tension of microporous membranes using contact angle measurements, J. Colloid Interface Sci. 194 (1997) 281–286. [36] Z. Li, V. Yangali-Quintanilla, R. Valladares-Liares, Q. Li, T. Zhan, G. Amy, Flux patterns and membranes fouling propensity during desalination of seawater by forward osmosis, Water Res. 46 (2012) 195–204. [37] J. Zhang, W.L.C. Loong, S. Chou, C. Tang, R. Wong, A.G. Fane, Membrane biofouling and scaling in forward osmosis membrane bioreactor, J. Membr. Sci. 403–404 (2012) 8–14. [38] C. Boo, S. Lee, M. Elimelech, Z. Meng, S. Hong, Colloidal fouling in forward osmosis: role of reserve salt diffusion, J. Membr. Sci. 390–391 (2012) 277–284. [39] X. Jin, X. Huang, E.M.V. Hoek, Role of specific ion interactions in seawater RO membrane fouling by alginic acid, Environ. Sci. Technol. 43 (2009) 3580–3587. [40] S. Botton, A.R.D. Verliefde, N.T. Quach, E.R. Cornelissen, Influence of biofouling on pharmaceuticals rejection in NF membrane filtration, Water Res. 46 (2012) 5848–5860.