High- ﬂ ux a ﬃ nity membranes based on cellulose nanocomposites for removal of heavy metal ions from industrial e ﬄ uents †

e Fully biobased a ﬃ nity membrane processing and its application in the removal of heavy metal ions from mirror industry e ﬄ uents were successfully demonstrated; indicating the potential use of these membranes in point-of-use or point-of-entry water cleaning products that are cheap, environmentally friendly and e ﬃ cient. Layered cellulose nanocomposite membranes were fabricated using cellulose micro ﬁ ber sludge as a support layer and cellulose nanocrystals (CNC SL , CNC BE or PCNC SL ) in a gelatin matrix as the functional layer. Scanning electron microscopy (SEM) studies revealed the bi-layered morphology of the membrane and well-individualized nanocelluloses in the functional layer. Bubble point measurements con ﬁ rmed the membrane pore structure in the micro ﬁ ltration range (5.0 – 6.1 m m), which provided very high water permeability (900 – 4000 L h (cid:1) 1 m (cid:1) 2 ) at <1.5 bars. A tensile strength of 16 MPa in dry conditions and a wet strength of 0.2 MPa, was considered su ﬃ cient for use of these membranes in spiral wound modules. Mirror industry e ﬄ uent laden with metal ions (Ag + and Cu 2+ /Fe 3+ / Fe 2+ ) when treated with cellulose nanocomposite membranes, showed high ion removal capacity, being 100% for PCNC SL followed by CNC BE than CNC SL . The removal of metal ions was expected to be driven by interactions between negatively charged nanocellulose and the positively charged metal ions.


Introduction
Over the past two decades, environmental regulations have become more stringent on heavy metal concentrations, which demand improved and innovative water treatment technologies. In recent years, a wide range of treatment technologies such as chemical precipitation, coagulation-occulation, otation, ion exchange and conventional membrane ltration, have been developed for heavy metal removal from contaminated wastewater. However, no individual treatment has been found to be universally effective and applicable for heavy metals removal. Furthermore, drawbacks such as high operational costs due to the chemicals used, high-energy consumption, handling costs for sludge disposal and high pressure for ltration are also noticed. 1,2 It is important to note that the selection of the most suitable treatment for metalcontaminated wastewater depends on the initial metal concentration, the overall treatment performance compared to other technologies, plant exibility/reliability and environmental impact. All the factors mentioned above should be taken into consideration in selecting the most effective and inexpensive treatment in order to protect the environment.
Recently, nanostructured affinity membranes were developed for the removal/rejection of contaminants based on physical/chemical properties or biological functions. [3][4][5] Rather than operating purely on the sieving mechanism, adsorption membrane based on the selectivity of the membrane to 'capture' contaminants, by immobilizing specic ligands or specic functional groups onto the surface of membranes. Affinity membrane reects technological advances in both xed-bed liquid separation and membrane ltration, and combines both the outstanding selectivity and the reduced pressure drops associated with ltration membranes. 5 Micro-nano structure poly(ether sulfones)/poly(ethyleneimine) nanobrous membrane fabricated by Chu and coworkers from Stony Brook University with nanocrystals as functional additives have show remarkable adsorption for anionic dyes and heavy metal ions from aqueous solution. 4,5 Our recent studies have shown that nanocellulose in native form or with additional surface functionalization (tempo oxidation, enzymatic phosphorylation, cationization etc.) have signicant adsorption efficiency towards metal ions, nitrates, dyes, humic acid etc. from industry effluents. [6][7][8][9] However the challenge is to nd suitable processing methods to fabricate membranes using nanocellulose to obtain optimal adsorption, mechanical stability and ux with out compromising on the green image of 'nanocellulose'. Karim et al. 8 have previously showed that cellulose nanocrystals based membranes prepared by freeze drying resulted in low water ux and low mechanical performance, in spite of good functionality. Low mechanical properties was attributed to the processing method used, viz. freeze-drying, which reduces H-bonding between nanocrystals and low ux was to the high thickness (250 mm) of membrane.
In the present study, an improved fully bio-based high ux membrane-(membranes/adsorbents) was fabricated using very simple and fast process, viz. vacuum ltration followed by hot pressing, following the 'nanopaper approach'. 10, 11 We have recently shown that nanopaper from native or cationically functionalized nanocellulose can effectively remove metal ions, nitrates, etc. 12,13 However the tight network of nanobers resulted in lower ux (5-50 L m À2 h À1 bar À1 ), which needs improvement. Therefore in the current study a thin layer of native and modied cellulose nanocrystals was used as functional entity on a support layer of microsized cellulose bers to fabricate a fully bio-based membrane. It was expected that the networking potential of microscale bers can be combined with adsorption potential of nanocellulose to tailor efficient water cleaning membranes. The support layer of cellulose sludge micro bers cuts down the cost of membranes/adsorbents, but provides mechanical and dimensional stability due to Hbonding. Impregnation of the support layer with nanocrystals provide a functional surface for the membrane, tailorable based on the surface functionality of cellulose nanocrystals. The morphology of the membranes and dispersion on nanoparticles in the functional layer was studied SEM images. Mechanical testing was performed in dry as well as in wet conditions using tensile testing machine. Flux was measured by cross-ow ltration device equipped with temperature and pressure control. To the best of our knowledge, no reports are available on the processing of layered affinity membranes based on cellulosic paper for the removal of metal ions from water. The adsorption performance was evaluated using two real wastewater samples containing Ag + and Cu 2+ /Fe 3+ /Fe 2+ from mirror industry were collected from southern part of Europe (Spain) and discussed with respect to adsorption mechanism as well as performance compared to adsorbent materials available in market.

Materials
Non-dried dissolving cellulose residue (sludge) from dissolving cellulose was obtained from Domsjö Fabriker AB, Ornskoldsvik mill, Sweden. Sludge was used as received without any pretreatments. Based on the data from the material supplier, the chemical composition of the used sludge was, cellulose content (95%), hemicellulose (4.75%) and very low content of lignin.
Gelatin and polyethylene glycol (PEG 2000) were purchase from Sigma-Aldrich (USA). Acetic acid and sulfuric acid were supplied by VWR (BDH, France).

Isolation and surface modication of nanocellulose
The process routes used to prepare cellulose nanocrystals used in the study as functional entity are schematically shown in Fig. 1. Cellulose nanocrystals (CNC SL ) were isolated from cellulose sludge using sulphuric acid hydrolysis as reported by Karim et al. 8 Cellulose nanocrystal were also prepared using bioethanol pilot plant process. Unbarked wood was hydrolyzed using dilute acid in a bioethanol pilot plant and rened by solvent extraction and bleaching to obtain pure cellulose. Cellulose, aer bioethanol process in a plant was supplied by SP Processum, Ornsköldsvik, Sweden (17 wt%). The puried cellulose from bioethanol process was made into 2 wt% suspensions, mixed by shear mixture and passed through the homogenizer, 10 times to obtain a thick gel of cellulose nanocrystals (CNC BE ) as reported earlier by Mathew et al. 14 The CNC SL were enzymatically modied to introduce phosphoryl groups on the surface reaction was carried out accordingly to Bozic et al. 15 Briey, reaction proceed in 0.9 wt% of CNC SL phosphate buffer (pH 7.6) dispersion in the presence of a 50 mM ATP, 250 mM of MgCl 2 and 35 U ml À1 of hexokinase enzyme for 24 h at 30 C. Aer the protein deactivation, PCNC SL were washed several times (centrifuged at 6000 rpm for 4 min) with distilled water, until no ATP absorption peak at 260 nm could be detected. Fig. 1 shows the AFM images of the used cellulose nanocrystals in the membrane fabrication. CNC SL (Fig. 1a) and CNC BE (Fig. 1b) shows typical cellulose nanocrystal structure and the diameters were found to be in the range of 5-10 nm. AFM image of PCNC SL (Fig. 1c) have whisker like structures but were shorter and broader than the corresponding non-modied ones. The diameters were found to be in the range of 20 nm, indicating the presence of laterally aggregated CNCs aer drying on mica surfaces. The diameters reported are based on the height to compensate the tip broadening effect and were measured using Nanoscope V soware. The accurate determination of the length was not possible from the AFM using this method and therefore not presented here. However, the AFM image shows that CNC BE are longer compared to CNC SL and PCNC SL .

Processing and characterization of membranes/ adsorbents
A very simple and fast method, vacuum-ltration was used to prepare thin-layered cellulose nanocomposite membranes. 2 wt% of sludge was used to form the support layer. The fabrication of the thin functional layer was performed by further ltration of nanocellulose (CNC BE , CNC SL and PCNC SL )-gelatin suspension as shown in Fig. 2. Nanocellulose suspensions and gelatin (1 wt% each) were taken in 1 : 1 ratio to form the functional layer. Aer 12 h of drying in room-temperature, the prepared membranes were pressed between aluminium plates in a compression-molding machine with load of 60-70 kN (Fontune Presses, Elastocon, Sweden) with heating (80 C) to obtain compacted membranes.

Microscopy
An atomic force microscope (Nanoscope V, Veeco Instruments, Santa Barbara, CA, USA) was used to examine the morphologies of the functional nanomaterials. The cantilever resonance frequency and the spring constant were 350 kHz and 10-200 nm À1 , respectively for measurement. The diameter measurements were conducted with the aid of Nanoscope V soware.
The membranes fractured in liquid nitrogen and sputter coated with tungsten were observed in the SEM (MAGELLAN 400, SEM, FEI Company) at an acceleration voltage of 3 kV to understand the morphology and nanostructure.

Mechanical properties
The tensile tests of the layered cellulose nanocomposite membranes in dry as well as in wet condition were performed, using a Universal Testing Machine, Shimadzu Autograph AG-X (Japan), with a load cell of 1 kN and the gauge length was 20 mm and width on 5 mm.

Pore size, porosity and BET surface area
The average pore size was measured for CNC BE and PCNC SL membranes using bubble point method using PMI Capillary Flow Porometer from porous Materials, Inc. USA.
Porosity of the samples was determined by the mercury intrusion technique using an AutoPore III mercury porosimeter (Micromeritics Instrument Co.). The membranes cut into a rectangular shape (1 Â 2 cm) was weighed and used for the study.
Surface area of the membranes was determined using N 2 adsorption following Brunauer-Emmett-Teller (BET) method using (Micromeritics ASAP 2020) automated system.

Measurement of water permeability
Sludge support and CNC BE and PCNC SL based membranes were used for the measurement of water permeability. A schematic diagram of the experimental apparatus for a cross-ow ltration test is presented in Fig. 3. The total surface area of membrane used was approximately 0.0169 m 2 , which was sufficiently compared with the working volume of feed water. The water permeability of membranes was determined up to burst pressure (1.5 bar) starting from 0.5 bars.

Evaluation of heavy metals adsorption
Two wastewater samples from mirror making industry, contaminated with Ag + and a mixture of Cu 2+ /Fe 3+ /Fe 2+ were obtained from southern part of Europe were used for the adsorption experiments. The pHs of effluents contaminated  with Ag + and Cu 2+ /Fe 3+ /Fe 2+ were 9.1 and 2.3, respectively (see ESI S1 † for the typical composition of the used water sample). All membranes were places in dead end cell apparatus and 50 ml of polluted water was lled and passed through the porous membranes at 4.5 bars of applied pressure. The ICP-OES was used to measure the ions concentration before and aer treatment. Control experiment was also performed with sludge/ gelatin nanocomposite membrane. The adsorption capacity of membranes for ions was calculated as described below. 7 where; C o and C i are initial and reaming conc. of metal ions in mg L À1 , V is used volume in litres and W is membrane weight in grams.
Metals ions, adsorbed on nanocomposite membranes were further characterized using a scanning electron microscope (SEM, JSM-6460 LV, JEOL, USA), equipped with energy dispersive X-ray elemental spectrometry (EDS), having soware, Inca Energy, Oxford. The adsorbed membranes were scanned with the depth of around 2 mm to detect the presence of metal ions. The electron beam of 20 kV was applied to determine the presence of ions.

Membrane stability/biodegradability
Bi-layered membranes having CNC BE as functional layer and cellulose sludge as support layer (S-G/CNC BE ) were dipped in water having different pH values (2.3, 6.9 and 9.2) at 37 C. Real wastewater of pHs 2.3 and 9.2 described under section 3.5 and tap water at pH 6.9 were used. In a parallel experiment, membranes were embedded in soil at different temperature values (5, 25 and 37 C). In order to determine the stability/ degradability, membranes were removed from water or soil and weighed at different time intervals for a period of 6 months. Degradation rate percentage was calculated with respect to control membranes weight, taken as 100%. The weight loss of treated membranes was measured aer 24 h drying at room temperature. All samples were imaged using a digital camera to visually compare the degradation.

Micro-nano scaled structure of layered membranes
The thickness of the support layer was measured to be 430 mm (determined by digital calipers), which increased to 440-448 mm, aer the impregnation with the functional layer. The thickness of functional layer of CNC SL , CNC BE , and PCNC SL determined based on the thickness difference were 10, 18 and 11 mm, respectively. The thickness difference for the functional layers may be assigned to the dimensions of nanocrystals and the relative degree of impregnation of gelatin/CNC suspension into the support layer.
The morphology of the support and the layered nanocomposite membranes at the micro/nanometer length scale were observed using SEM and are shown in Fig. 4. In Fig. 4a the surface and the cross-section of the support layer are given where microsized bers were clearly visible and the bers were loosely bound together in a 3D network, with micro-scaled voids in between.
The bi-layered membrane cross-sections for S-G/CNC SL , S-G/ CNC BE , S-G/PCNC SL are shown in Fig. 4b-d. In all SEM images in the le column a surface functional layer is clearly visible on the support layer. However, the depth of impregnation varied from sample to sample and it was not possible to see any nanocrystals impregnated into the support layer at this magnication. SEM images in the middle column shows that the functional layer on the top was found to be made up of layered structures in all cases.
The images on the right column show the functional layer at high magnications. The high-resolution SEM image of functional layer showed individualized nanoscaled dispersed within the gelatin forming a network of functional entity without visible agglomerates. The good dispersion of nanocrystals in gelatin matrix indicates positive interaction between gelatin matrix and nanocrystals. It may be considered that nanocrystals are bound together in a three-dimensional network by gelatin chains; and is represented schematically in Fig. 2. Further more, the functional layer also seems to be denser and the nanocrystals being more embedded in the matrix in the case of CNC BE compared to CNC SL and PCNC SL . A lower degree of infusion (and probably lm formation) was expected for CNC BE compared to CNC SL and P-CNC SL , due to the presence of longer nanocrystals and probably explain the higher thickness of the functional layer based on CNC BE . The micro-scaled porosity of the support layer and low thickness of the functional layer conrmed by SEM was expected to provide high ux during water purication while providing sufficient mechanical strength.
The surface area as obtained from BJH analysis of N 2 gas adsorption isotherm of S-G/CNC BE thin-layered nanocomposite membrane was highest (32.6 m 2 g À1 ) compared to S-G/CNC SL membrane (26.3 m 2 g À1 ) followed by S-G/PCNC SL (18.8 m 2 g À1 ) membranes. The lower surface area of all the fabricated membranes is expected due to the low surface area of the microscaled support as well as the compacting of networks (support and functional layers) during hot pressing. Another reason for the difference in the surface area may be the depth of infusion of nanocellulose within the support layer and is probably related to the dimensions of cellulose nanocrystals. The CNC BE , system with lower degree of infusion, provided membranes with highest surface area.

Mechanical properties
Mechanical properties of layered cellulose nanocomposite membranes measured in tensile mode are summarized in Table  1. The support layer showed a strength of 9 MPa, which increased with the addition of the nanocrystalline functional layer. Higher strength of membranes was observed in the current study (13)(14)(15)(16)(17)(18)(19) MPa) compared to our earlier studies on freeze-dried CNC based nanocomposites (2 MPa) 8 as well as the support used in this study. It was inferred that the ltration followed by hot pressing used for the preparation of membranes facilitate the formation of hydrogen bonds and  thereby provide mechanical stability to the membranes. The tensile strength and the modulus of CNC BE based bi-layered membrane were higher compared to CNC SL and PCNC SL membranes. The strain at break was in the range of 2% and the differences between membranes were not statistically signicant. Recently, ultraltration membrane with CNC infused on nonwoven support of electrospun PAN bers was reported. Mechanical properties were measured and compared with commercial available GS0.22 ultraltration membrane; 5 the stress at break of reported nanocomposite membrane was 14 AE 0.8 MPa and modulus was 375 MPa. The bi-layered membranes in the current study fabricated via vacuum ltration have comparable tensile strength and higher modulus. It is worth mentioning that in the current study the support layer is based on cheap bioresidue and the prepared membrane is fully biobased. The mechanical stability in wet condition follows the same trend as shown in dry condition but with lower values. The higher mechanical properties on CNC BE based membranes may be due to the denser functional layer compared to CNC SL and P-CNC SL found in SEM studies. The reported tensile strength (even in wet condition) of all fabricated membranes was enough to use them in water cleaning application especially in spiral wound modules. 16

Pore structure and water permeability of nanocellulose based membranes
Bubble point method is a very accurate and effective technique for pore size measurement at industrial level. The average pore size of the support was 6.6 mm and decreased to 5.1 and 6.2 mm for S-G/CNC BE and S-G/PCNC SL respectively; which conrmed the microltration range of nanocomposite membranes. The large average pore sizes of nanocomposite membranes are benecial to ensure high ux through the membranes. The large distribution of pore sizes was however noticed and may require further process optimization to improve the pore homogeneity.
The porosity of the support layer (based on mercury porosimetry) was 69.48%. Slightly decreased porosity was observed in the case of S-G/PCNC SL and S-G/CNC BE (z65%). The results conrmed the presence of thin and denser functional layer of nanocrystals on the highly porous support layer. Thicker porous material is known to promote water movement in a zig-zag motion while linear motion has been observed within thinner materials. 17,18 The water permeability of the nanocellulose based membranes was evaluated in cross-ow system for S-G/CNC BE and S-G/PCNC SL and compared with that of the support. Water ux experiments were conducted under the variable rate condition (applied pressure) and Fig. 5 shows the water permeability data as a function of applied pressure during cross-ow. The water permeability was 3417 L m À2 h À1 bar À1 for the support, which decreased to 950 and 2622 L m À2 h À1 bar À1 respectively for S-G/CNC BE and S-G/PCNC SL . Ma et al. 5 had reported that the ux through a microltration membrane using CNCs as infused entity within PAN nanobrous scaffold was 0.59 L m À2 h À1 bar À1 and much higher than commercial membrane, GS0.22 (0.25 L m À2 h À1 bar À1 ). Fabricated layered cellulose nanocomposite affinity membranes in the current study therefore have signicantly higher ux compared to previously reported membranes as well as the commercially available membranes. Also the ux reported in the current study is higher than all our earlier reports where a single thicker layer nanocellulose was used for membrane application. 8,11,12 The support layer showed the highest water permeability at all applied pressure owing to the free water ow through microstructure and decreased when the functional layer is present. The water permeability of S-G/CNC BE is low at all applied pressures compared to S-G/PCNC SL and is probably due to denser lm formation by CNC BE on the support; which is supported by SEM studies. The low water ux of CNC BE based nanocomposite membranes was further supported by the average pore size results; being lowest for S-G/CNC BE .

Adsorption of metal ions in cross-ow mode
The removal efficiency of Ag + , Cu 2+ and Fe 3+ /Fe 2+ ions from industrial effluent using the membranes were studied in cross-ow mode and static mode ( Table 2). Cross-ow mode of operation provides more surface area for adsorption in contrast with static mode operation due to high penetration power of pollutants and bulk adsorption possibility, therefore, it is considered to be the efficient mode while using at industrial level in real wastewater purication. The adsorption capacity in static mode is provided with in parenthesis for comparison.
In the case of Ag + , complete removal (100%) was observed in all cases and was explained based on the low concentration of Ag + ions in the industrial effluent. The metal ions removal efficiency of all three nanocomposite functional membranes, is in the order of PCNC SL > CNC BE > CNC SL for Cu 2+ /Fe 3+ /Fe 2+ water. The removal rate of PCNC SL functional membrane was the highest and varied between 74 and 86% for the metal ions. CNC BE nanocomposite membrane has intermediate removal capability, both for Cu 2+ (36%) and Fe 3+ (33%) compared to CNC SL functional nanocomposite membrane (only 13-14% for Cu 2+ and Fe 3+ ). Thus, the removal rate for CNC BE is two times higher with respect to CNC SL .
The concentration of functional group on the surface of fabricated membranes is the reason for the different percentage removal. The sulphate groups present on the surface of CNC SL are 30-90 mmol kg À1 is lowest compare to carboxylic group on CNC BE (150 mmol kg À1 ) and the highest value of carboxylic and phosphate groups (425 mmol kg À1 ) are recorded on phosphorylated CNC; therefore, phosphorylated CNC have highest removal percentage of metal ions compare to other used nanocelluloses. 14,19 In static mode also the removal behaviour follows the same trends for membranes as shown in static mode but the capacity is lower, in general, for all metal ions studied. It may be noted that the pH of effluent containing silver metal ions was 2.7 and copper/iron metal ions was 9.2, which conrms the effectiveness and versatility of the membranes in acidic as well as basic conditions. Further more the membranes were found to be efficient in different concentration ranges and for different metal ions, which can be of advantage in real applications.
Cu 2+ adsorption from water based on literature reports (Table 3) shows that the best removal rate for chitosan with an efficiency of 70 mg g À1 18 on chitosan followed by carbon nanotube (67.8 mg g À1 ). 19 Bio-based nanomaterials have also been used for the remediation of Cu 2+ ; recently two reports conrmed the use of chitin nanobers and cellulose nanocrystals. The maximum sorption capacity of CNCs was 2.5 mg g À1 20 and chitin nanobers was 141.08 mg g À1 . 21 TEMPO-CNF were also used for the bioremediation of copper ions; 44.2-135 mg g À1 sorption capacity was reported with varying carboxylic group contents (0.6-1.5 mmol g À1 ). 22,23 Our nding conrms the outstanding functionality of the used nanocellulose; the PCNC SL has the highest (233 mg g À1 ) Cu adsorption followed by CNC BE (64 mg g À1 ).
The sorption capacity of Fe 3+ (512 mg g À1 ) is very high for PCNC SL membranes in cross-ow mode. One very interesting literature report shows the use of cow bone charcoal for the removal of iron metal ions from industrial wastewater 24 but the sorption capacity of used adsorbent was very low (31.4 mg g À1 ) compared to PCNC SL (512 mg g À1 ) and also CNC BE (102 mg g À1 ) from the current study.
The Ag + sorption capacity onto the with respect to unit mass of our bi-layered membrane or functional entity was lower compared to adsorption results reported by Herrera et al. 25 on the alfalfa biomass (27.37 mg g À1 ) and other available resins; Duolite GT-73 (155.92 mg g À1 ), Diaion WT01S (45.0 mg g À1 ), Diaion CRB02 (15.83 mg g À1 ) and Dowex 66 (48.44 mg g À1 ) 3,25 (See Table 3). The main reason is the low initial concentration of Ag + where as the adsorption capacity reported in the literature is at the saturation limit.
It is expected that the membranes can be reused aer the recovery of the heavy metal ions by acid wash or incinerated at the end-of-life to recover metal ions. It is also considered that these biobased membranes aer adsorption of Ag + ions can be converted to an antibacterial device for the treatment of drinking water as reported by Theresa et al. 26 The reuse of Ag + ions adsorbed membranes, as antibacterial device will be reported in future.

Mechanism of metal ion removal from water
The bi-layered membranes samples, before and aer incubating with effluent water were used for elemental constitution analysis and the spectrum is shown in Fig. 6. Only Si, S, C and O can be observed in untreated ones which have been known as the principle elements of fabricated membranes. No signals of Ag + , Cu 2+ and Fe 3+ /Fe 2+ elements were found in this spectrum. The EDS spectrums of membranes treated with contaminated water show the introduction of new peaks, which conrms the presence of the respective metal ions on the membrane surface. The Ag + , Cu 2+ and Fe 3+ /Fe 2+ signals can be observed on the spectrum obtained aer treatment with polluted water, indicating that the interaction/immobilization of metal pollutants had occurred on the surface of fabricated Table 2 The removal efficiency of Ag + , Cu 2+ and Fe 2+ /Fe 3+ ions from industrial effluent using the membranes in cross-flow mode and static mode. The adsorption capacity in static mode is provided with in parenthesis for comparison The possible mechanism for the adsorption of metal ions on the surface of fabricated nanocomposite might be electrostatic interactions, ions exchange, microprecipitation or interaction a The values reported in the literature are the saturation limit (determined aer 10 to 15 cycle of reuse) but in our study only one cycle of adsorption evaluation was performed. Thus, the reusability must be performed in future to nd the saturation limit. followed by nucleation effect. The mechanisms are not fully understood and interactions as shown in Fig. 6a can be expected. The existences of functional entities as phosphate, carboxylic and sulphonate, on macromolecules chain of PCNC SL , CNC BE and CNC SL , respectively are the main driving force for negative charges on nanocellulose. 6 The isolation of CNC SL was performed by sulphuric acid hydrolysis 8 and presence of sulfonate groups (SO 3 À ) was conrmed by FTIR in our previous study. 14 The main functional group available on CNC BE aer isolation process was carboxylate (-C]OO À ) and is conrmed by XPS analysis 14 PCNC SL was highly effective for metals ions removal and is due to the introduction of phosphate (PO 3 2À ) group on the surface of CNC SL during hexokinasemediated reaction, as described previously. 14,15 See ESI S2 † for the IR spectrum of the used functional entities. The results are also agrees with our earlier studies where model and industrial waste water contaminated with heavy metals ions has treated with four nanocellulose functional entities; CNC SL , CNC BE , P-CNC and P-CNF. 6 It is worth noticing that the fabrication of these nanoparticles in membranes provides the possibility to use them in cross-ow mode, which further enhances the removal efficiency towards metal ions.

Membrane stability/biodegradability study
The visual illustrations of membranes during different stages of degradation the effluent as well as soil are shown in Fig. 7. In the case of effluent water (Fig. 7a) with metal ions and with different pH conditions, no degradation was noted with respect to weight loss or visual appearance, up to 6 months. However the samples showed a change in color attributable to be the adsorption of metal ions/contaminants present in wastewater. Biodegradability of fabricated membranes is depends on weight/size of used membranes, composition of membranes and the effect of living or dead organism. The used wastewater from mining industry was not a suitable habitat for microorganisms to grow. Therefore, the membranes were stable in real polluted water up to 6 months and conrm the suitability of these membranes in real water purication system.
In the case of degradation studies in soil at different temperatures, a different trend was observed. The soil structure and composition (mineral and organic), temperature, water activity, pH, and the oxygen and carbon dioxide content affect degradation. Therefore the extent of biodegradation of the membranes can be expected to vary with region and from season to season. Moreover degradation by microbial attack is the major mode of degradation of the natural composites in soil. In this study, all the above-mentioned factors were maintained uniform for all the samples and degradation studies were carried out under laboratory conditions. The membranes degraded rapidly in soil conrmed by weight loss as well as via visual appearance (see Fig. 7b). It was observed that at higher temperature degradation occurred at higher rates. Complete degradation of CNC BE membrane was recorded aer four months of incubation. Our results are also agreement with previous published data, where cellulose lm was incubated with soil having fungus microbes. A porous structure with fungal mycelia on the surface of the decayed lm was observed, indicates microbial degradation of cellulose lm. 28

Conclusions
The study demonstrates the successful processing and utilization of fully biobased cellulose nanocomposite membranes for metal ions capture from industry effluents. We report for the rst time, high ux layered cellulose membranes with nanocellulose composite as functional layer, prepared following a very simple, fast and effective process, viz. vacuum ltration followed by hot pressing. The support layer provided mechanical stability to the membrane while the functional layer immobilized metal ions from contaminated water via Fig. 7 Stability/biodegradation test of support-CNC BE nanocomposite membranes in real wastewater (a) and soil (b). Nanocomposite membranes have been incubated as discussed in manuscript, percentage degradation was calculated using control samples as 100%. C, T 2 , T 4 and T 6 denotes incubation time interval as 0, 2, 4 and 6 months.
adsorption. The porous network of microbers with a thin functional layer of nanocellulose resulted in high ux and maximal functionality. Good mechanical stability in wet conditions; favorable pore structure and water permeability as well as high affinity of the developed membranes towards metal ions indicate promising new technology for green water puri-cation technology. PCNC SL based membranes were most effective for the removal of all three metal ions and the adsorption capacity aer single cycle of cross-ow ltration was higher than reference adsorbents found in literature. Biodegradation studies in effluent water and soil indicates the potential of these membranes in real application where long term membrane performance in use conditions and possible biodegradation at end-of-life are crucial economical, efficient and environmental friendly membranes.