Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

download Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

of 5

Transcript of Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    1/10

    Treatment of highly concentrated wastewater containing multiplesynthetic dyes by a combined process of coagulation/occulationand nanoltration

    Can-Zeng Liang, Shi-Peng Sun n, Fu-Yun Li, Yee-Kang Ong, Tai-Shung Chung n

    Department of Chemical &  Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, Singapore 117576, Singapore

    a r t i c l e i n f o

     Article history:Received 11 April 2014

    Received in revised form

    26 June 2014

    Accepted 29 June 2014Available online 8 July 2014

    Keywords:

    Multiple dye wastewater

    Coagulation/occulation

    Nanoltration

    Hollow  ber membranes

    a b s t r a c t

    The treatments of dyes (acid, basic and reactive dyes) wastewater were studied by applying individualcoagulation/occulation (CF) and nanoltration (NF) processes as well as their combination (referred as

    CF–NF). For the treatment of highly concentrated multiple dyes wastewater (MDW, 1000 ppm),

    polyaluminum chloride (PAC) and polydiallyldimethyl ammonium chloride (PDDA) were found to be

    the most effective coagulant and  occulant, respectively. The CF process can achieve about 90% of dye

    removal at the optimal dosage of PAC/PDDA¼400/200 ppm, and the MDW with pH43 is favorable for

    the CF treatment. A positively charged NF hollow   ber membrane was fabricated and used for NF

    treatment. It is able to remove almost 100% dyes with a permeate  ux of about 1.0 L m2 h1 under an

    operating pressure of 1 bar. The combination of CF and NF can complement each other's strengths and

    overcome their individual limitations. The NF treatment can completely remove the strong color left in

    CF treated dye solutions, while the ef ciency of coagulant/occulant is improved by treating NF

    concentrated streams and subsequently results in much less sludge. In addition, membrane fouling is

    abated and NF permeate   ux is increased by applying the CF process as a pretreatment. Thus, the

    combination of CF–NF improves the overall performance for the dyes wastewater treatment.

    &   2014 Elsevier B.V. All rights reserved.

    1. Introduction

    A dye molecule consists of two components; namely, the dye

    chromophore and the dye auxochrome. When the dye molecule is

    exposed to light, the chromophore structure which includes

    double bonds (CQC) oscillates to absorb light and generates

    visible color   [1,2]. By estimation, more than 100,000 synthetic

    dyes and over 700,000 t of dyestuff are produced annually  [3,4].

    The color that appears in industrial wastewater ef uents caused

    by residual dyes is esthetically undesirable and harmful to the

    environments and the ecosystems. Even a very low concentration

    of dyes can generate strong color   [5]. The dye wastewater is

    generated during dye-related activities such as dye production,textile dyeing, leather tanning and paper production, etc   [6].

    Particularly in the textile industry, about 10–15% dyes are lost in

    dyeing processes, and typically 200–350 m3 wastewater are gen-

    erated to produce one ton   nished product   [7]. It is becoming

    more dif cult to directly discharge this kind of wastewater

    because the legislations related to environments are becoming

    more stringent in many countries  [8].

    Generally, there are two decolorization methods   [9]; namely,

    destruction of dye molecules and separation of dyes from water.

    To destruct or transform the dyes, conventional processes are

    usually applied, such as chemical oxidation, photo-catalysis and

    biodegradation [10]. However, the destruction methods is found to

    be inadequate and require extensive energy to break down the dye

    molecules, most of which are stable to light, oxidizing agents and

    microbiological degradation   [1,6,11]. The separation methods

    include adsorption, coagulation/occulation (CF) and membrane

    separation [9]. Adsorption of dyes on powder activated carbon is

    popular and effective. However, the activated carbon is not cheap,and the adsorption performance is reduced sharply after regen-

    eration or reactivation, which also results in a 10–15% loss of the

    sorbent  [3,7,12].  Coagulation/occulation is widely used for dyes

    removal due to its low capital cost and simple operation  [12–14].

    Coagulation of dye solutions is a process of destabilizing the dye

    solution systems to form agglomerates or  ocs. Flocculation is the

    process of destabilizing suspended particle systems and bridging

    the aggregated  ocs to form larger agglomerates that settle down

    under gravity   [15,16]. Charge neutralization is regarded as a

    prerequisite condition in most of coagulation processes   [17].

    The CF process generates sludge and but sometimes it is ineffective

    Contents lists available at  ScienceDirect

    journal homepage:   www.elsevier.com/locate/memsci

     Journal of Membrane Science

    http://dx.doi.org/10.1016/j.memsci.2014.06.057

    0376-7388/& 2014 Elsevier B.V. All rights reserved.

    n Corresponding authors. Tel.:  þ65 6516 6645; fax:  þ65 6779 1936.

    E-mail addresses: [email protected] (S.-P. Sun),

    [email protected] (T.-S. Chung).

     Journal o f Membrane Science 469 (2014) 306–315

    http://www.sciencedirect.com/science/journal/03767388http://www.elsevier.com/locate/memscihttp://dx.doi.org/10.1016/j.memsci.2014.06.057mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.memsci.2014.06.057http://dx.doi.org/10.1016/j.memsci.2014.06.057http://dx.doi.org/10.1016/j.memsci.2014.06.057http://dx.doi.org/10.1016/j.memsci.2014.06.057mailto:[email protected]:[email protected]://crossmark.crossref.org/dialog/?doi=10.1016/j.memsci.2014.06.057&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.memsci.2014.06.057&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.memsci.2014.06.057&domain=pdfhttp://dx.doi.org/10.1016/j.memsci.2014.06.057http://dx.doi.org/10.1016/j.memsci.2014.06.057http://dx.doi.org/10.1016/j.memsci.2014.06.057http://www.elsevier.com/locate/memscihttp://www.sciencedirect.com/science/journal/03767388

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    2/10

    for some soluble dyes. In fact, it is always challenging in selecting

    appropriate coagulant/occulant due to the large number of exist-

    ing and emerging dyes that have complex structures  [7]. Another

    important separation method is the pressure-driven membrane

    technology including ultraltration (UF), nanoltration (NF) and

    reverse osmosis (RO) [1]. Ultraltration has been successfully used

    for separating high molecular weight and insoluble dyes from

    water. However, UF is not able to remove those water-soluble dyes

    with low molecular weights [1,18]. Although ef cient dye removalcan be achieved by RO, the high pressure required in the RO process

    hinders its wide applications to treat dye wastewater. NF is

    positioned in between UF and RO [9], the nominal molecular weight

    cutoff (MWCO) of NF membranes is in the range of 100 to 1000 Da

    and with the pore size of about 0.5–2.0 nm   [19,20]. NF was

    introduced in the early 1980s and has become popular due to its

    low operating pressures and relatively low capital and operating

    costs [20].

    Currently most NF membranes are negatively charged   at-

    sheet composite membranes, which may have high fouling ten-

    dency because most foulants are positive charged. In recent years,

    there is a growing interest to fabricate hydrophilic and positively

    charged NF membranes which are less prone to organic fouling

    [19–30]. Compared to the sophisticated module fabrication pro-

    cess for  at-sheet or spiral-wound membranes, it is convenient to

    fabricate hollow   ber membrane modules, which also provide a

    high surface area per unit volume and thus to reduce manufactur-

    ing costs   [31]. NF hollow   ber membranes have attracted much

    attention due to its superiorities to the conventional  at-sheet NF

    membranes. Recently, several efforts have been made on the

    development and application of NF hollow   ber membranes for

    removing dyes from water. For instance, Sun et al. employed both

    crosslinking and interfacial polymerization to coat a positively

    charged polyethyleneimine (PEI) layer on the outer surface of a

    negatively charged polyamideimide (PAI) hollow  ber substrate to

    produce a double-repulsive NF membrane that can remove 99.8%

    of the positively charged Safranin O and 98.8% of the negatively

    charged Orange II sodium salt [19,21]. Shao et al. developed a thin-

    lm composite NF hollow  ber membrane by interfacial polymer-ization on the inner surface of polyetherimide (PEI) hollow   ber

    supports for removal of Safranin O and Aniline blue and achieved

    higher than 90% dye rejections  [25]. Wei et al. fabricated a NF

    hollow   ber membrane through interfacial polymerization on a

    polysulfone/polyethersulfone supporting membrane for the treat-

    ment of reactive brilliant blue X-BR and acid red B dye solutions

    and obtained above 99.9% dye rejections   [26]. Zheng et al. devel-

    oped positively charged TFC hollow   ber NF membranes via the

    dip-coating method on polypropylene hollow   ber microltration

    membranes. The resultant NF membrane can achieve dye rejections

    of 99.8%, 99.8% and 99.2% for Brilliant green, Victoria blue B and

    Crystal violet, respectively [27].

    However, the drawbacks of membrane technologies including

    NF are the   ux decline caused by membrane fouling and the

    generation of concentrated streams   [32]. To minimize the   ux

    reduction, one approach is to implement a right pretreatment

    process   [33], the other approach is to produce fouling-resistant

    membranes  [34]. The generation of concentrated streams is an

    intrinsic issue for membrane separation processes since mem-branes only achieve separation rather than destruction or trans-

    formation. For the dye-containing wastewater treatment, the

    concentrated stream is usually an unwanted by-product. It must

    be further treated before discharge   [35,36]. To overcome the

    aforementioned problems, a combination of various separation

    methods is necessary to achieve a high dye removal and high

    separation ef ciency.

    To our best knowledge, there is no study on treatment of 

    synthetic dye wastewater containing multiple dyes of different

    classes by a combined process of CF and NF with a positively

    charged NF hollow   ber membrane. In this work, reactive, acid

    and basic dyes were chosen because of their popularity in textile

    and dye industries as well as environmental concerns  [6,7,12,37].

    The representative dye concentrations in dye wastewater streams

    are in the range of 0.05–0.1 g/L  [38]. For this study, aqueous dye

    solutions with a dye concentration of 1000 ppm (1.0 g/L) were

    prepared to simulate the highly concentrated dye house waste

    ef uent and the concentrated stream after membrane  ltration.

    In order to maximize the ef cacy of the CF process and

    minimize the generation of sludge, proper CF formulations were

    carefully evaluated, screened and selected. To reduce membrane

    fouling, hydrophilic and positively charged NF hollow  ber mem-

    branes were made in our laboratory for this study. The CF, NF and

    the combination of CF–NF were studied to remove dye(s) from

    synthetic dye wastewaters, which consist of either single dye or

    multiple dyes. The objectives of this work are to demonstrate that

    both CF and NF techniques can remove dyes from wastewater

    effectively and the combination of CF–NF is able to improve the

    overall performance. This could lead to a new approach for the dyewastewater treatment by a hybrid system.

    2. Experimental

     2.1. Chemicals and materials

    7 inorganic and 4 organic chemicals, listed in   Table 1, were

    chosen and used as coagulants and  occulants in this study. 5 dyes

    (2 acid dyes, 2 reactive dyes and 1 basic dye, purchased from

     Table 1

    Characteristics of coagulants and  occulants used in this work.

    Name of coagulant/occulant Code name Molecular formula Molecular weight

    (g/mol)

    Purity

    Polyaluminum chloride PAC Aln(OH)mCl(3nm) (0omo3n)   Z115b

    Z30% (as Al2O3)

    Aluminum sulfate-octadecahydrate AS Al2(SO4)3 18H2O 666 Extra pure

    Aluminum potassium sulfate-dodecahydrate APS AlK(SO4)2 12H2O 474 99.5%

    Iron(III) chloride-anhydrous IC FeCl3   162 98%

    Iron(III) sulfate-pentahydrate IS Fe2(SO4)3 5H2O 490 97%

    Calcium oxide CO CaO 56   Z98 %

    Magnesium chloride-anhydrous MC MgCl2   95 pure

    Cationic polyacrylamide CPAM   a(CH2–CH–CONH2)m   800–1000 million   Z90%

    Anionic polyacrylamide APAM   a(CH2–CH–CONH2)m   800–2500 million   Z90%

    Polydiallyldimethyl ammonium chloride (cationic, dissolved in water) PDDA (C8H16ClN)n   200,000–350,0 00 20 wt%

    Cyanoguanidine CYGU NH2C(¼NH)NHCN 84   Z99 %

    a Obtained from the specication of the products.b

    Calculated by assuming  m¼n¼1 for PAC.

    C.-Z. Liang et al. / Journal of Membrane Science 469 (2014) 306 – 315   307

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    3/10

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    4/10

    a   ow rate of 0.15 L/min and a pressure of 1.0 bar. The permeate

    solution was collected from the lumen side of the membrane. The

    retentate stream was circulated back to the feed solution. Each NF

    experiment included 3 steps: (1) the initial pure water permeability

    (PWP, L m2 bar1 h1) was measured using DI water at 1 bar; (2) a

    1.0 L dye solution was used as the feed solution and circulated at

    1 bar. Then permeate samples were taken at each consecutive time

    interval of 1 h. Finally, DI water was used to  ush the system and the

    module thoroughly until the circulating water was clean, then thenal PWP was measured.

    Pure water permeability (PWP, L m2 bar1 h1) was calcu-

    lated according to the following equation:

    PWP¼  Q 

    ΔPAt   ð1Þ

    where Q  is the volume (L) of permeate collected during a specic

    period of sampling time   t   (h),   A   is the effective membrane area

    (m2), and  ΔP  is the transmembrane pressure (bar).

    The permeate   ux (J, L m2 h1) can be determined by the

    following equation:

     J ¼Q 

     At   ð2Þ

    where Q  is the volume (L) of permeate collected during a specicperiod of sampling time   t   (h), and   A   is the effective membrane

    area (m2).

     2.5. Physicochemical analyses

    Two methods were chosen to determine the dye concentration;

    namely, total organic carbon (TOC) method and UV –vis integral

    method. The implementation of these two methods depends on

    specic situations as discussed later.

    The TOC method was based on the TOC concentration deter-

    mined by a TOC analyzer (TOC, ASI 5000A, Shimazu).   Appendix

    Fig. 1 shows the calibration curves of TOC values as a function of 

    dye concentration for various dyes. The TOC value has an almost

    linear relationship with dye concentration in the range of 0 to1000 ppm. The dye removal (Rd, %) was calculated by the following

    equation:

    Rd ¼ðC bC aÞ

    C b 100%   ð3Þ

    where  C b   and Ca  are the TOC concentrations of the dye solution

    before and after the CF treatment, respectively.

    The UV –vis integral method was based on the UV –vis integral

    (integrated range: 350–650 nm) obtained through scanning the

    sample by a UV –vis spectrophotometer (Pharo 300, Merck).

    Appendix Fig. 2   shows the calibration curves of UV –vis integral

    values as a function of dye concentration for various dyes. The UV –

    vis integral has a linear relationship with dye concentration when

    the dye concentration is below 100 ppm. However, the relation-

    ship becomes non-linear when the dye concentration is above

    100 ppm. Based on UV –vis integral, the dye removal (Rd, %) by CF

    was calculated by the following equation:

    Rd ¼ðI bI aÞ

    I b 100%   ð4Þ

    where I b  and  I a are the UV –vis integral values of the dye solution

    before and after the CF treatment, respectively.The dye removal (Rd, %) by the NF membrane was calculated by

    the following equation:

    Rd ¼ðI  f  I  pÞ

    I  f  100%   ð5Þ

    where I  f  and  I  p  are the UV –vis integral values of the feed and the

    permeate solutions, respectively.

    It is worthy to point out that the concentrated dye solutions

    had to be diluted to around 100 ppm when the UV –vis integral

    method was used in order to achieve reasonable accuracy.

    The normalized ef ciency of coagulant/occulant (E c ) is dened

    as

    E c ¼mdmc 

    ð6Þ

    where md is the weight of the dye removed by CF, while  mc  is the

    total weight of coagulant/occulant added for the CF treatment.

    The particle size of the settlement after CF was analyzed by

    using a laser diffraction particle size analyzer (Beckman Coulter

    LS230, analysis range: 0.04  mm to 2000  mm). The solution pH was

    determined by using a pH meter (pH/ion S220, Mettler Toledo).

    The aluminum concentration was measured using Inductively

     Table 4

    Preliminary performance evaluation of coagulants on dye removal.

    Coagulant Dye removala (%)

    INCA RBBR BB-R RB-5 AB-8 MDW

    PAC 95.6 90.0 85.5 87.6 0 90.2

    AS 64.2 42.6 81.0 44.3 0 75.7

    APS 51.1 42.7 84.3 25.6 0 72.1

    IC 72.8 59.2 84.6 22.3 0 69.5

    IS 27.4 44.9 84.1 21.3 0 65.3

    CO 0 0 43.1 0 75.2 5 9.9

    MC 29.2 12.6 24.5 0 0 34.4

    a Dye removal was based on the TOC method. The coagulant dosage was

    1000 ppm, the original dye concentration was 1000 ppm.

    0.0

    0.5

    1.0

    1.5

    2.0

    2.53.0

    3.5

    4.0

    50

    60

    70

    80

    90

    100

    0 400 800 1200 1600 2000

       E  c

       R   d

       (   %   )

    PAC dosage (ppm)

    Rd

    Ec

    0.0

    0.5

    1.0

    1.5

    2.0

    2.5

    3.0

    3.5

    4.0

    50

    60

    70

    80

    90

    100

    0 40 80 120 160 200

       E  c

       R   d

       (   %   )

    PAC dosage (ppm)

    Rd

    Ec

    MDW=1000 ppm

    MDW=100 ppm

    Fig. 1.   Coagulation performance of PAC for MDW treatment at different dye

    concentrations: (a) the dye concentration in MDW was 1000 ppm; (b) the dye

    concentration in MDW is 100 ppm. Dye removal is based on the UV –vis integral

    method.

    C.-Z. Liang et al. / Journal of Membrane Science 469 (2014) 306 – 315   309

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    5/10

    Coupled Plasma-Optical Emission Spectrometry (ICP-OES, iCAP

    6000 Series, Thermal Scientic).

    3. Results and discussions

     3.1. Coagulation and  occulation

    Both single dye and multiple dye wastewaters (referred to asSDW and MDW, respectively, hereafter) were studied in order to

    evaluate the effectiveness of coagulants and  occulants in remov-

    ing various types of dyes, and to select the most proper coagulant

    and  occulant. The following procedures were carried out:  rstly

    to screen and select a proper coagulant; secondly to optimize the

    coagulant dosage; thirdly to identify the suitable   occulant (or

    coagulant aid); and  nally optimize the  occulant dosage.

     3.1.1. Selection of the proper coagulant 

    Table 4  summarizes the dye removal for both SDW and MDW

    based on the TOC method. The reasons of choosing the TOC

    method are: (1) to exclude the color interference introduced by

    ironic ion (Fe3þ); (2) To avoid a large pH change that may change

    the intensity of UV –

    vis absorbance caused by the coagulants. Thisis particularly true for calcium oxide (CO) because 1000 ppm CO

    can dramatically change the pH value of dye solutions; (3) there is

    an almost linear relationship between TOC and dye concentration

    in the range of 0–1000 ppm, which makes it simple and reliable

    for analyses.

    It can be observed from Table 4 that an inorganic coagulant can

    be very effective in treating certain dyes, but totally ineffective for

    the others. In the case of MDW, every coagulant appears to be

    more or less effective, this may suggest that interactions between

    dyes take place and subsequently change the MDW's charge

    characteristics. As a result, when a coagulant is added into the

    dyes solution, charge neutralization initially occurs and promotes

    to formation of particles/ocs; then dyes may be also involved in

    the process of co-precipitation and sorption onto  ocs;  nally the

    coagulation happens and dyes are removed   [7,15,17,40]. The

    ability of removing dyes from the MDW sample by the 7 coagulants

    is in the order of PAC4AS4APS4IC4IS4CO4MC. Therefore,

    PAC was identied as the most proper coagulant for further

    studies.

     3.1.2. Optimization of the coagulant dosage

    The dye removal is also studied by the UV –vis integral method

    to identify the optimal coagulant dosage because of the followingreasons: (1) the PAC solution induces no visible color change;

    (2) the organic   occulant may contribute to the TOC value.  Fig. 1

    shows examples of multiple dyes removal from MDW at different

    dye concentrations. For the 1000 ppm MDW treated by PAC, the dye

    removal increases sharply with an increase in PAC dosage initially,

    but nally approaches a plateau, which indicates that the maximum

    dye removal by coagulation is around 97%. Mathematically, the

    0

    20

    40

    60

    80

    100

    CPAM PAC PAC -CPAM

       R   d

       (   %   ) 

    0

    20

    40

    60

    80

    100

     APAM PAC PAC -APAM

    0

    20

    40

    60

    80

    100

    PDDA PAC PAC -PDDA

    0

    20

    40

    60

    80

    100

    CYGU PAC PAC -CYGU

       R   d

       (   %   ) 

       R   d

       (   %   ) 

       R   d

       (   %   ) 

    Fig. 2.   Performance of PAC and different organic  occulants to treat 1000 ppm MDW. The dosage of each PAC, APAM, CPAM, CYGU and PDDA was 400 ppm; For the PAC-

    occulant combination, the dosage of each component was 400 ppm. Dye removal was based on the UV –vis integral method.

    0.0

    0.5

    1.0

    1.5

    2.0

    2.5

    50

    60

    70

    80

    90

    100

    0 100 200 300 400

       E  c

       R   d   (   %   )

    Dosage of PDDA (ppm)

    Rd

    Ec

    Fig. 3.  Effect of different dosages of PDDA combined with 400 ppm PAC to treat

    1000 ppm MDW.  E c  refers to the overall normalized ef ciency of PAC and PDDA.

    The dye removal was based on the UV –vis integral method.

    C.-Z. Liang et al. / Journal of Membrane Science 469 (2014) 306 – 315310

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    6/10

    removal ef ciency per gram of the coagulant (E c ) declines accord-

    ingly. For the 100 ppm MDW treated by PAC,  Fig. 1(b) shows that

    the dye removal reaches a maximum when about 100–150 ppm

    PAC is added and then decreases with a further increment in PAC

    dosage. The exceeded amount of PAC might lead to the reverse of 

    surface charge of the coagulated particles/ocs, thus resulting in

    particles re-stabilization and lowering the treatment ef ciency [41].

    Before the dye removal reaches the maximum, the  E c  value of  Fig. 1

    (a) is always greater than that of Fig.1(b), which implies coagulationtaking place more favorably in a high dye concentration than a low

    dye concentration. In order to achieve a high dye removal without

    much compromising the coagulant ef ciency, we chose a PAC

    dosage of 400 ppm because it can remove about 85% dyes with

    an   E c   value of about 2.1 for further studies to identify a proper

    occulant.

     3.1.3. Identi cation of the suitable occulant 

    A typical coagulation/occulation process involves two stages of 

    mixing process; namely, rapid stirring (75–700 rpm) for 0.5–3 min in

    order to achieve a good dispersion of the coagulant/occulant, and

    slow stirring (30–150 rpm) for 5–30 min in order to propagate the

    growth of  ocs and minimize the breakdown of formed aggregates

    [16,42]. In this work, the experiments associated with both coagulantand occulant, the rapid and slow stirring processes were combined

    into one single stirring scheme; namely, stirring at 200 rpm for

    1 min. Adopting the single stirring scheme is due to the following

    reasons: (1) to simplify the process; (2) to allow the coagulation and

    occulation take place together. Some organic polyelectrolytes, such

    as CPAM and PDDA, exhibit the ability of functioning as both

    coagulant and   occulant, which is demonstrated in   Fig. 2   with

    following discussions. In terms of dyes removal by coagulation/

    occulation, as depicted in Fig. 2(a) and (c), neither APAM nor CYGU

    is effective in dye removal. The combination of PAC-APAM does not

    make much improvement compared with PAC alone; while the

    combination of PAC-CYGU produces antagonistic effect, which

    reduces the ef cacy of PAC. As shown in  Fig. 2(b) and (d), both

    CPAM and PDDA alone are effective but with less ef ciency than PAC.The combinations of PAC-CPAM and PAC-PDDA generate synergetic

    effects, which improve the total dye removal compared to that of PAC

    alone. Both CPAM and PDDA are positive charged, but PDDA is

    superior to CPAM according to its individual or combined perfor-

    mance. Additionally, the suspension produced by PAC-PDDA was

    easier to  lter out than that by CPAM-PDDA. This is due to the fact

    that CPAM possesses a huge molecular weight (800–1000 million)

    which is prone to entangling with the  lter paper and forming a gel-

    like sticky  lm that blocks the   lter pores. In summary, PDDA was

    identied as the most proper   occulant among the 4 examined

    occulants.

     3.1.4. Optimization of the occulant dosage

    The effects of different PDDA dosages on dye removal werestudied by  xing the PAC dosage at 400 ppm.  Fig. 3 displays that

    the dye removal increases with an increase in PDDA dosage. This

    might be attributed to: (1) PDDA possesses a high positive charge

    density (as shown in Table 3) and water solubility. It can function

    as an effective coagulant in the PAC-dye solution system; (2) PDDA

    serves as a   occulant that enhances the bridging mechanism for

    coagulation and improves the aggregating capacity   [16], thus

    increasing the coagulation ability of PAC. However, the   E c   value

    (i.e., the ef ciency per gram of the total coagulants (400 ppm

    PACþPDDA)) decreases when the PDDA dosage increases. There-

    fore, The CF settlement's particle size was investigated to optimize

    the PDDA dosage.

    Fig. 4 displays the particle size (or diameter) distributions of CF

    settlements when 1000 ppm MDW was treated by 400 ppm PAC

    without and with PDDA. The particle sizes were analyzed after CF

    experiments and 1 h settling. As observed at the bottom of  Fig. 4,

    trace amounts of large-size particles in the range of 50–100 mm are

    formed with the addition of PDDA. In contrast, there are no

    particles with a size larger than 30 mm for the CF settlement

    without adding PDDA. This is probably owing to the fact that the

    ne precipitates formed by PAC are not able to grow bigger

    without the aid of   occulant (PDDA). As shown in the enlarged

    picture of  Fig. 4, the particle size distribution curve shifts towardthe right side when increasing the PDDA dosage. This suggests that

    the mean particle size becomes larger with an increase in PDDA

    dosage. Without PDDA, a mean particle size of 7.4  mm is produced

    by PAC, the mean particle size increases to about 7.8, 9.2, 9.3, or

    9.7 mm when adding 100, 200, 300 or 400 ppm PDDA, respectively.

    The bigger the particle size, the easier the  ltration is. It is found

    that the combination of 400 ppm PAC and 200 ppm PDDA pro-

    duces a settlement with a moderate size of   9.3 mm and an

    acceptable dye removal of  90% with a relatively high normalized

    ef ciency (E c ) of   1.5. Therefore, the (400/200 ppm) PAC/PDDA

    dosage was chosen to study the effects of pH.

    0

    2

    4

    6

    8

    0 20 40 60 80 100

       V  o   l  u  m  e   (   %   )

    Particle size (µm)

    0

    2

    4

    6

    8

    3 6 9 12 15

       V  o   l  u  m  e   (   %   )

    Particle size (µm)

    PDDA=0 ppm

    PDDA=100 ppm

    PDDA=200 ppm

    PDDA=300 ppm

    PDDA=400 ppm

    Fig. 4.   Particle size (or diameter) distributions of CF settlements when 1000 ppm

    MDW was treated by 400 ppm PAC without and with PDDA.

    0

    2

    4

    6

    8

    10

    12

    0

    20

    40

    60

    80

    100

    0 2 4 6 8 10 12

       F   i  n  a   l  p   H   (  -   )

       R   d   (   %   )

    Initial pH (-)

    Rd: PAC

    Rd: PAC-PDDA

    Final pH: PAC

    Final pH: PAC-PDDA

    Fig. 5.  Effects of the initial pH on the dye removal for 1000 ppm MDW when using

    PAC (400 ppm) and PAC-PDDA (400/200 ppm). Dye removal was based on the UV –vis

    integral method.

    C.-Z. Liang et al. / Journal of Membrane Science 469 (2014) 306 – 315   311

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    7/10

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    8/10

    a result, the NF membrane shows excellent rejections for both

    positively and negatively charged dye molecules.

    Fig. 6 displays the permeate  ux and dye removal as a function

    of time for MDW. Two MDW concentrations were employed;

    namely, 100 and 1000 ppm. The 1000 ppm MDW was treated

    with and without the CF treatment. For all cases, the permeate ux

    declines initially but becomes stable after 7 h, while the dye

    removal remains above 99%. For the 100 ppm and 1000 ppm

    MDW, they have almost the same initial permeate  ux and declinein the same speed until the fourth hour. After the fourth hour, the

    permeate   ux of the 100 ppm sample tends to drop slightly,

    whereas that of the 1000 ppm sample drops dramatically. After

    7 h, both permeate  uxes become stable, and the permeate  ux of 

    the 100 ppm sample is approximately 20% higher than that of the

    1000 ppm sample. These phenomena may be explainable as

    follows: (1) the dyes adhered to the membrane surface and

    partially blocked the surface pores in the beginning; (2) membrane

    fouling due to the concentration polarization of the dyes became

    serious for the 1000 ppm sample after 4 h. Since the higher the

    dye concentration, the heavier the concentration polarization is

    [44], the  ux decline of the 100 ppm sample is mainly controlled

    by membrane fouling, while that of the 1000 ppm sample is

    affected by both membrane fouling and concentration polarization

    fouling.

    Since about 90% dyes were removed after the CF treatment, the

    dye concentration in the CF treated sample is around 100 ppm. As

    a result, the relationship between permeate  ux vs. time of the CF

    treated sample is nearly parallel with that of the 100 ppm MDW

    sample, but the former has a lower  ux (i.e., about 10%) than the

    latter. The similar relationship between permeate   ux vs. time

    indicate that membrane fouling for both cases are similar due to

    their similar dye concentrations. The lower  ux of the CF treated

    sample may arise from the osmotic pressure increase in the CF

    treated sample. As shown in Fig. 7, about 24 ppm aluminum ion is

    found in the CF treated solution. As a consequence, the effective

    driving force across the NF membrane is reduced even though

    they have the same transmembrane pressure [45,46]. Since the CF

    treated sample has less fouling propensity, it has a higher  ux thanthat of the 1000 ppm sample after the 7-h test (i.e., about 10%

    higher). Clearly, applying the CF process as a pretreatment, not

    only can it lower membrane fouling but also sustain the

    permeate ux.

    In terms of rejections, nearly 100% dyes were removed by the

    NF for all the 3 cases over 8-h tests. The fate of added aluminum

    was also studied for the CF–NF process. As shown in Fig. 7, about

    58% of the total added aluminum (i.e., 57 ppm) reacted and

    remained in the settlement as the solid stream, while around

    41% dissolved and left in the treated solution as the feed stream for

    the NF process, and about 0.5%  nally went to the permeate  ux as

    the ef uent stream. Ideally, there should be no aluminum in the

    NF feed stream after the CF process since the aluminum ion would

    lead to an increase in osmotic pressure and ultimately a decrease

    in effective driving force across the membrane. Additionally, the

    less aluminum in the ef uent stream, the better it is since the

    ef uent would eventually be discharged or reused.

    The UV –vis was also used to conrm the above conclusion.

    Fig. 8 displays the UV –vis spectra in a wavelength range between

    200 and 800 nm for the original 1000 ppm MDW, the CF treated

    MDW, and the NF permeate of the CF treated sample. The

    absorbance decreases signicantly after the CF treatment because

    of about 90% dyes removal at this stage. The UV absorbance drops

    to almost zero after the NF process. This is consistent with theaforementioned results of almost 100% dye removal. The pictures

    at the up right side of   Fig. 8  show the color changes during the

    processes. The original 1000 ppm MDW is deeply dark blue. After

    the CF treatment, it becomes light blue, and then the color

    completely disappears after the CF-NF treatment.

    4. Conclusions

    The treatments of synthetic dye wastewaters were studied by

    applying coagulation/occulation (CF), nanoltration (NF) and the

    combination of CF–NF. It is found that the removal of dye through

    CF from single dye wastewater (SDW) depends on both dye and

    coagulant. Some coagulants are very effective in treating certainSDW, but totally ineffective for the others. The ef ciency of 

    removing dyes from the highly concentrated multiple dyes waste-

    water (MDW, 1000 ppm) by the 7 coagulants follows the order of 

    PAC4AS4APS4IC4IS4CO4MC. For the treatment of MDW,

    polyaluminum chloride (PAC) is the best coagulant and polydial-

    lyldimethyl ammonium chloride (PDDA) is the best   occulant.

    A dosage of 400/200 ppm PAC/PDDA is the optimal CF formulation,

    which is able to remove about 90% of dyes from 1000 ppm MDW.

    With the optimal dosage, the CF process is more favorable for dye

    solutions with a higher concentration and pH.

    The positively charged hollow   ber NF membrane shows dye

    removal of almost 100% for both anionic and cationic dyes. The NF

    performance is also nearly independent of dye concentration with

    a permeate   ux of about 1.0 L m2

    h1

    under an operating

    0.0

    10.0

    20.0

    30.0

    40.0

    50.0

    60.0

    70.0

    80.0

    90.0

    100.0

    0.0

    10.0

    20.0

    30.0

    40.0

    50.0

    60.0

    Total Solid Feed Effluent

       D   i  s   t  r   i   b  u   t   i  o  n  o   f   A   l   (   %   )

       C  o  n  c  e  n   t  r  a   t   i  o  n  o   f   A   l   (  p  p  m   )

       C   F  s  e   t   t   l  e  m  e  n   t  s   t  r  e  a  m

       C   F   t  r  e  a   t  e   d   M   D   W

       N   F  p  e  r  m  e  a   t  e

       T  o   t  a   l  a   d   d  e   d   A   l   f  o  r   C   F

    Fig. 7.  The fate and distribution of aluminum during the CF–NF treatment process.

    0

    10

    20

    30

    40

    50

    200 300 400 500 600 700 800

       A

       b  s  o  r   b  a  n  c  e

    Wave length (nm)

    Original

     After CF

     After CF-NF

    a

    bc

    Fig. 8.   The UV –vis absorbance spectra of (a) the original MDW (1000 ppm),

    (b) solution after the CF treatment and (c) solution after the NF –NF treatment.

    The UV –vis spectrum of the original MDW was obtained via diluting the original

    MDW by 20 times, then the diluted sample was measured, the resultant absor-

    bance was multiplied by a factor of 20.

    C.-Z. Liang et al. / Journal of Membrane Science 469 (2014) 306 – 315   313

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    9/10

    pressure of 1 bar. Additionally, this NF membrane shows an

    excellent rejection of trivalent ions (e.g., Al3þ).

    The generation of concentrated stream and membrane fouling

    is inevitable in the NF process, while the production of sludge and

    the remaining strongly visible color are encountered in the CF

    process. The combination of CF and NF can complement each

    other's strengths and overcome their individual limitations. The

    NF treatment can completely remove the color left in the CF

    treated MDW, meanwhile the ef ciency of coagulant/occulant issignicantly enhanced by treating the concentrated streams of the

    NF process. As a result, much less sludge is generated. In turn, by

    applying CF process as a pretreatment method, the membrane

    fouling is reduced and the NF permeate   ux is increased. Those

    benets and strengths lead to better performance for the treat-

    ment of synthetic dye wastewater. Future works will aim at

    developing a proper CF formulation and a higher   ux positively

    charged NF membrane for the treatment of real dye wastewater.

     Acknowledgments

    The authors would like to acknowledge the   nancial supports

    provided by National Research Foundation (NRF) of Singaporeunder its NRF Proof-of-Concept 8th Grant Call (NRF2012NRF-

    POC001-059) for the project entitled   ‘Development of advanced

    nanoltration membranes for high removing rate of dyes in textile

    wastewater’  (NUS grant number: R-279-000-389-281) as well as

    the  Singapore-MITAlliance for Research and Technology (SMART)

    Centre under its Innovation Grant (ING12045-ENG) for the project

    entitled   ‘Development of robust high-performance nanoltration

    membranes for textile wastewater treatment’ (NUS grant number:

    R-279-000-377-592). Special thanks are given to Mr. Bai-Wang

    Zhao for his supports during the preparation of this work.

     Appendix. Supporting information

    Supplementary data associated with this article can be found in

    the online version at  http://dx.doi.org/10.1016/j.memsci.2014.06.057.

    References

    [1]   A. Akbari, J.C. Remigy, P. Aptel, Treatment of textile dye ef uent using apolyamide-based nanoltration membrane, Chem. Eng. Process. 41 (2002)601–609.

    [2]  T. Vickerstaff, The Physical Chemistry of Dyeing, second ed., Oliver   &   Boyd,London, 1954.

    [3]   T. Robinson, G. McMullan, R. Marchant, P. Nigam, Remediation of dyes intextile effuent: a critical review on current treatment technologies with aproposed alternative, Bioresour. Technol. 77 (2001) 247–255.

    [4]   H. Zollinger, Colour Chemistry-Synthesis, Properties of Organic, Dyes andPigments, VCH Publishers, New York, 1987.

    [5]  P. Nigam, G. Armour, I.M. Banat, D. Singh, R. Marchant, Physical removal of textile dyes from ef uents and solid-state fermentation of dye-adsorbedagricultural residues, Bioresour. Technol. 72 (2000) 219–226.

    [6]  E. Forgacs, T. Cserhati, G. Oros, Removal of synthetic dyes from wastewaters: areview, Environ. Int. 30 (2004) 953–971.

    [7]   A.K. Verma, R.R. Dash, P. Bhunia, A review on chemical coagulation/ occula-tion technologies for removal of colour from textile wastewaters, J. Environ.Manage. 93 (2012) 154–168.

    [8]   C. Hessel, C. Allegre, M. Maisseu, F. Charbit, P. Moulin, Guidelines andlegislation for dye house ef uents, J. Environ. Manage. 83 (2007) 171–180.

    [9]   A.Y. Zahrim, N. Hilal, Treatment of highly concentrated dye solution bycoagulation/occulation–sand   ltration and nanoltration, Water Resour.Ind. 3 (2013) 23–34.

    [10]  O.J. Hao, H. Kim, P.C. Chiang, Decolorization of wastewater, Crit. Rev. Env. Sci.Technol. 30 (2000) 449–505.

    [11]   O. Tünay, I. Kabdasli, G. Erememktar, D. Orhon, Color removal from textilewastewaters, Water Sci. Technol. 34 (1996) 9–16.

    [12]  M. Riera-Torres, C. Gutiérrez-Bouzán, M. Crespi, Combination of coagulation–occulation and nanoltration techniques for dye removal and water reuse in

    textile ef uents, Desalination 252 (2010) 53–59.

    [13]  A. Bes-Piá, M.I. Iborra-Clar, A. Iborra-Clar, J.A. Mendoza-Roca, B. Cuartas-Uribe,M.I. Alcaina-Miranda, Nanoltration of textile industry wastewater using aphysicochemical process as a pre-treatment, Desalination 178 (2005)343–349.

    [14]   T. Chen, B. Gao, Q. Yue, Effect of dosing method and pH on color removalperformance and   oc aggregation of polyferric chloride–polyamine dual-coagulant in synthetic dyeing wastewater treatment, Colloids Surf. A 355(2010) 121–129.

    [15]   A.Y. Zahrim, C. Tizaoui, N. Hilal, Coagulation with polymers for nanoltrationpre-treatment of highly concentrated dyes: a review, Desalination 266 (2011)1–16.

    [16]  K.E. Lee, N. Morad, T.T. Teng, B.T. Poh, Development, characterization and theapplication of hybrid materials in coagulation/occulation of wastewater: areview, Chem. Eng. J. 203 (2012) 370–386.

    [17]  B. Shi, G. Li, D. Wang, C. Feng, H. Tang, Removal of direct dyes by coagulation:the performance of preformed polymeric aluminum species, J. Hazard. Mater.143 (2007) 567–574.

    [18]   K. Majewska-Nowak, Synthesis and properties of polysulfone membranes,Desalination 71 (1989) 83–95.

    [19]   S.P. Sun, T.A. Hatton, T.S. Chung, Hyperbranched polyethyleneimine inducedcross-linking of polyamide-imide nanoltration hollow   ber membranes foreffective removal of ciprooxacin, Environ. Sci. Technol. 45 (2011) 4003–4009.

    [20]  S. Cheng, D.L. Oatley, P.M. Williams, C.J. Wright, Positively charged nanoltra-tion membranes: review of current fabrication methods and introduction of anovel approach, Adv. Colloid Interface Sci. 164 (2011) 12 –20.

    [21]   S.P. Sun, T.A. Hatton, S.Y. Chan, T.S. Chung, Novel thin-lm compositenanoltration hollow   ber membranes with double repulsion for effectiveremoval of emerging organic matters from water, J. Membr. Sci. 401–402(2012) 152–162.

    [22]  J. Gao, S.P. Sun, W.P. Zhu, T.S. Chung, Polyethyleneimine (PEI) cross-linked P84nanoltration (NF) hollow  ber membranes for Pb2þ removal, J. Membr. Sci.452 (2014) 300–310.

    [23]   W.P. Zhu, S.P. Sun, J. Gao, F.J. Fu, T.S. Chung, Dual-layer polybenzimidazole/polyethersulfone (PBI/PES) nanoltration (NF) hollow   ber membranes forheavy metals removal from wastewater, J. Membr. Sci. 456 (2014) 117 –127.

    [24]   P.S. Zhong, N. Widjojo, T.S. Chung, M. Weber, C. Maletzko, Positively chargednanoltration (NF) membranes via UV grafting on sulfonated polyphenylene-sulfone (sPPSU) for effective removal of textile dyes from wastewater,

     J. Membr. Sci. 417–418 (2012) 52–60.[25]   L. Shao, X.Q. Cheng, Y. Liu, S. Quan, J. Ma, S.Z. Zhao, K.Y. Wang, Newly

    developed nanoltration (NF) composite membranes by interfacial polymer-ization for Safranin O and aniline blue removal, J. Membr. Sci. 430 (2013)96–105.

    [26]  X. Wei, X. Kong, C. Sun, J. Chen, Characterization and application of a thin-lmcomposite nanoltration hollow   ber membrane for dye desalination andconcentration, Chem. Eng. J. 223 (2013) 172–182.

    [27]  Y. Zheng, G. Yao, Q. Cheng, S. Yu, M. Liu, C. Gao, Positively charged thin-lm

    composite hollow 

    ber nano

    ltration membrane for the removal of cationicdyes through submerged  ltration, Desalination 328 (2013) 42–50.[28]   M. Li, J. Xu, C.Y. Chang, C. Feng, L. Zhang, Y. Tang, C. Gao, Bioinspired fabrication

    of composite nanoltration membrane based on the formation of DA/PEI layerfollowed by cross-linking, J. Membr. Sci. 459 (2014) 62–71.

    [29]   W. Fang, L. Shi, R. Wang, Interfacially polymerized composite nanoltrationhollow ber membranes for low-pressure water softening, J. Membr. Sci. 430(2013) 129–139.

    [30]   L. Wang, S. Ji, N. Wang, R. Zhang, G. Zhang, J.R. Li, One-step self-assemblyfabrication of amphiphilic hyperbranched polymer composite membranefrom aqueous emulsion for dye desalination, J. Membr. Sci. 452 (2014)143–151.

    [31]  S.P. Nunes, K.V. Peinemann, Membrane Technology in the Chemical Industry,Wiley-VCH Verlag GmbH, Weinheim, 2001.

    [32]   B. Van der Bruggen, L. Lejon, C. Vandecasteele, Reuse, treatment, anddischarge of the concentrate of pressure-driven membrane processes,Environ. Sci. Technol. 37 (2003) 3733–3738.

    [33]   M. Unlu, H. Yukseler, U. Yetis, Indigo dyeing wastewater reclamation bymembrane-based   ltration and coagulation processes, Desalination 240

    (2009) 178–185.[34]  B. Van der Bruggen, M. Mänttäri, M. Nyström, Drawbacks of applying nano ltra-

    tion and how to avoid them: a review, Sep. Purif. Technol. 63 (2008) 251–263.[35]   L. Lei, X. Hu, P.L. Yue, Improved wet oxidation for the treatment of dyeing

    wastewater concentrate from membrane separation process, Water Res. 32(1998) 2753–2759.

    [36]   I.C. Escobar, B. Van der Bruggen, Modern Applications in Membrane Scienceand Technology, American Chemical Society(ACS), Washington, DC, 2011.

    [37]  A. Mittal, J. Mittal, L. Kurup, Batch and bulk removal of hazardous dye, indigocarmine from wastewater through adsorption, J. Hazard. Mater. 137 (2006)591–602.

    [38]  S. Cheng, D.L. Oatley, P.M. Williams, C.J. Wright, Characterisation and applica-tion of a novel positively charged nanoltration membrane for the treatmentof textile industry wastewaters, Water Res. 46 (2012) 33–42.

    [39]  S.P. Sun, K.Y. Wang, D. Rajarathnam, T.A. Hatton, T.S. Chung, Polyamide-imidenanoltration hollow   ber membranes with e longation-induced nano-poreevolution, AlChE J. 56 (2010) 1481–1494.

    [40]   B.Y. Gao, Y. Wang, Q.Y. Yue, J.C. Wei, Q. Li, Color removal from simulated dye

    water and actual textile wastewater using a composite coagulant prepared by

    C.-Z. Liang et al. / Journal of Membrane Science 469 (2014) 306 – 315314

    http://dx.doi.org/10.1016/j.memsci.2014.06.057http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref2http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref2http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref2http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref2http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref4http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref4http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref4http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref6http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref6http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref6http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref6http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref6http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref10http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref10http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref10http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref10http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref10http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref11http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref11http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref11http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref11http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref11http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref18http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref18http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref18http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref18http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref18http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref31http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref31http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref31http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref36http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref36http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref36http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref39http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref38http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref37http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref36http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref36http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref35http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref34http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref33http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref32http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref31http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref31http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref30http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref29http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref28http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref27http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref26http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref25http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref24http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref23http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref22http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref21http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref20http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref19http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref18http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref18http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref17http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref16http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref15http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref14http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref13http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref12http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref11http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref11http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref10http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref10http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref9http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref8http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref7http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref6http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref6http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref5http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref4http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref4http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref3http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref2http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref2http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref2http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref1http://dx.doi.org/10.1016/j.memsci.2014.06.057

  • 8/19/2019 Treatmentofhighlyconcentratedwastewatercontainingmultiple.pdf

    10/10

    ployferric chloride and polydimethyldiallylammonium chloride, Sep. Purif.Technol. 54 (2007) 157–163.

    [41]   W.P. Cheng, F.H. Chi, C.C. Li, R.F. Yu, A study on the removal of organicsubstances from low-turbidity and low-alkalinity water with metal-polysilicate coagulants, Colloids Surf. A 312 (2008) 238–244.

    [42]   J. Zou, H. Zhu, F. Wang, H. Sui, J. Fan, Preparation of a new inorganic –organiccomposite  occulant used in solid–liquid separation for waste drilling   uid,Chem. Eng. J. 171 (2011) 350–356.

    [43]   Y.K. Ong, F.Y. Li, S.P. Sun, B.W. Zhao, C.Z. Liang, T.S. Chung, Nanoltrationhollow ber membranes for textile wastewater treatment: lab-scale and pilot-scale studies, Chem. Eng. Sci. 114 (2014) 51–57.

    [44]   L. Xu, L.S. Du, C. Wang, W. Xu, Nanoltration coupled with electrolytic

    oxidation in treating simulated dye wastewater, J. Membr. Sci. 409-410

    (2012) 329–334.[45]  C.N. Lopes, J.C.C. Petrus, H.G. Riella, Color and COD retention by nano ltration

    membranes, Desalination 172 (2005) 77–83.[46]   C. Tang, V. Chen, Nanoltration of textile wastewater for water reuse,

    Desalination 143 (2002) 11–20.

    C.-Z. Liang et al. / Journal of Membrane Science 469 (2014) 306 – 315   315

    http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref46http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref45http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref44http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref43http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref42http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref41http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40http://refhub.elsevier.com/S0376-7388(14)00516-X/sbref40