Adsorptive removal of arsenic from groundwater using a novel high flux polyacrylonitrile (PAN)–laterite mixed matrix ultrafiltration membrane

Abstract

A flat sheet mixed matrix membrane, made of polyacrylonitrile (PAN) copolymer, impregnated with laterite, was fabricated for the removal of arsenic from water. The permeability and molecular weight cut-off of the selected membrane were 3.4 × 10⁻¹¹ m s⁻¹ Pa and 48 kDa, respectively. Morphological analysis showed macrovoids constricted by laterite particles. Surface characteristics assessed by atomic force microscopy revealed the increase in roughness with laterite concentration. The presence of different forms of iron oxide (laterite) and nitrile groups (polyacrylonitrile) in membrane M25 was confirmed by X-ray diffraction. Incorporation of arsenic within the membrane matrix was demonstrated by subsequent lowering of transmittance peaks at different wavelengths of FTIR. Maximum adsorption capacity of the selected membrane was 1.4 mg g⁻¹ at 298 K. Under the optimum operating conditions, the pristine mixed matrix membrane resulted in a filtrate with a concentration below 10 μg l⁻¹ for 17 hours in cross flow mode with a 0.01 m² filtration area. The stability of the membrane was demonstrated for three regeneration cycles. The effect of pH and coexisting anions like phosphate, sulphate, carbonate and bicarbonate on the removal efficiency of arsenic was studied. The performance of the membrane in the presence of arsenic-contaminated groundwater was also tested.

---

Water impact

A novel, high flux PAN–laterite mixed matrix membrane was prepared, characterized in detail and applied for the removal of arsenic from groundwater. The membrane was in ultrafiltration range with high hydrophilicity. The membrane showed high throughput and, at the same time, high rejection of arsenic. The filtration experiments were conducted using 100 ppb arsenic feed solution as well as real groundwater from the affected area. Permeate flux using contaminated groundwater was as high as 225 l m⁻² h⁻¹ with a filtrate concentration less than 10 μg l⁻¹ (WHO limit). The breakthrough time of producing safe drinking water was 17 hours for a virgin membrane (0.01 m² filtration area) and this time can be enhanced significantly for a scaled-up filter with a high filtration area.

1. Introduction

Arsenic contamination of groundwater is a major concern nowadays. Erosion of alluvial soil, anthropogenic activities like hydraulic fracturing, and industrial effluent (copper and zinc processing industries use metallic arsenic in the smelting process) cause mobilization of arsenic in groundwater. Arsenic contamination (arsenicosis) causes diseases like skin cancer, cancer in the bladder, kidneys, and lungs, high blood pressure and reproductive disorders.¹ Generally, arsenic exists in natural water as arsenate or As(V) and arsenite or As(III) in the ratio of about 70 : 30. Amongst the two present forms, arsenite is more toxic than the former. As(III) exists as uncharged species (H₃AsO₃) and As(V) occurs mainly as monovalent (H₂AsO₄⁻) or divalent (HAsO₄²⁻) species or both. The mobility of arsenite is higher than that of arsenate species, and they contaminate the groundwater more than its coexisting form. However, an additional oxidation step to convert As(III) to As(V) is sometimes adopted to exploit the higher adsorption capacity of the arsenate form.² The World Health Organization (WHO) suggested that the maximum permissible limit of arsenic in drinking water is 10 μg l⁻¹.³

Arsenic contamination of groundwater is prevalent in countries like Argentina, Bangladesh, China, Chile, India and Mexico.^4,5 This problem is acute in the eastern part of India, especially in the gangetic plains. Adsorbents like iron oxide and activated carbon^6–8 were suitable for arsenic removal. However, these adsorbents have high synthesizing cost and show poor efficiency in the presence of other co-existing ions. Membrane processes like nanofiltration⁹ and forward osmosis¹⁰ were effective, but energy intensive. Mixed matrix ultrafiltration membranes can be an attractive alternative in this context. These membranes utilize a polymer with an inorganic additive in the matrix¹¹ that improves their selectivity and productivity.¹² Primarily, the choice of an additive depends on its size, performance and most importantly cost. Gohari et al. used iron–manganese binary oxide in polyethersulfone for the removal of trivalent arsenic.¹³ He et al. used a zirconium nanoparticle-incorporated polysulfone hollow fiber membrane for the removal of pentavalent arsenic.¹⁴ However, the complexity in manufacturing such inorganic additive along with its cost is a major disadvantage. Also, leaching of toxic zirconium and manganese in the permeate water stream may give rise to various diseases like granulomata and manganism,¹⁵ respectively. On the other hand, this issue can be addressed, if a naturally abundant material is used as an additive with comparable selectivity. Laterite, present in the Rarh region of the eastern part of India (Kharagpur, 22.32°N, 87.32°E), has high affinity for arsenic uptake from water.¹⁷ The material has major constituents like silicon dioxide (28.7%), aluminium dioxide (21.1%), and iron oxide (47.3 w/w%).¹⁷ Trace quantities of manganese dioxide (0.8 w/w%) and sodium oxide (0.7 w/w%) are also present.¹⁷ Moreover, its arsenic adsorption capacity can be increased several folds by adopting an optimized chemical treatment method.¹⁸ The production process of treated laterite is also less complex, cost effective and highly scalable.

Selection of the base polymer has an important role in membrane fabrication. Polyacrylonitrile (PAN) was used in this study. It has a good solvent resistance and excellent thermal stability, owing to its high density of cross-linking. Once heated, the nitrile groups crosslink between themselves to impart chemical stability to the membrane.¹⁹ Moreover, PAN-based membranes have high throughput that makes them commercially more viable.²⁰ Polyvinyl pyrrolidone (PVP) was used as a pore former and it also increases the mechanical stability of the membrane.²¹

The present paper deals with the scope of impregnating chemically treated laterite in PAN co-polymer with PVP as an additive. Membranes were characterized in terms of permeability, porosity, molecular weight cut-off (MWCO), contact angle and surface area. The presence of laterite and polymer components in the membrane matrix was confirmed by X-ray diffraction (XRD). Morphology and roughness were assessed by a scanning electron microscope (SEM) and an atomic force microscope (AFM), respectively. The adsorption capacity of the mixed matrix membrane (MMM) was evaluated for both forms of arsenic. The dynamic filtration capability of the selected membrane was evaluated to estimate the breakthrough point. The performance of MMM in the presence of other co-existing anions and arsenic-contaminated groundwater feed was also investigated.

2. Experimental

2.1. Materials

PAN (average molecular weight: 150 kDa) copolymer was purchased from M/s, Technorbital Advanced Materials Pvt. Ltd., Kanpur, India. N,N-dimethylformamide (DMF: density 944 kg m⁻³) and polyethylene glycol (PEG) (molecular weight: 100 kDa, 200 kDa, 35 kDa, 20 kDa, 10 kDa, 6 kDa, 4 kDa, 400 Da) were supplied by M/s, Merck (India) Ltd., Mumbai, India. Polyvinyl pyrrolidone (average molecular weight: 44 kDa) and dextran (molecular weight: 70 kDa) were obtained from M/s, Sigma Aldrich Chemicals, USA. For the preparation of synthetic feed solution, sodium arsenate heptahydrate (Na₂HAsO₄·7H₂O) supplied by M/s, Loba Chemie Pvt. Ltd., Mumbai, India was used. Sodium hydroxide, used for regenerating the membrane, was procured from a local market.

2.2. Production of treated laterite

Laterite particles (obtained from Kharagpur, 22.32°N, 87.32°E) were chemically treated to increase its arsenic uptake capacity. The detailed methodology of the chemical treatment is found in Maiti et al.¹⁸ Treated particles were ground by a rotary shaker and were filtered through a fine cloth to produce laterite powder. It was dried for four hours before it was dissolved in the polymer solution. Size distribution of the laterite particles was measured using a Mastersizer (model: 2000) supplied by M/s, Malvern Instruments, UK, at room temperature. The laser diffraction technique was used to measure the size of the particles. The intensity of scattered laser beam passing through a dispersed particulate sample was measured. These data were analyzed to calculate the size of the particles that created the scattering pattern.

2.3. Fabrication of MMM

The nomenclature of different membranes is given in Table 1. At first, a pre-determined quantity of PVP (heated for four hours) was dissolved in DMF solvent. Subsequently, laterite and PAN were added to the solution (333 K) and the mixture was stirred by a mechanical stirrer for two hours. The solution was then homogenized at room temperature (303 ± 2 K) for six hours in an ultrasonic cleaner (Piezo-U-Sonic, India).²² After this, the solution was cast on a glass plate with the help of a doctor's blade at a drawdown speed of 2 mm s⁻¹ and a gap of 150 micron. The glass plate was immersed in a water tub (kept at room temperature) immediately. A few drops of formalin solution were added to the gelation bath to inhibit bacterial growth on the membrane surface. It was then peeled from the surface of the glass plate after 16 hours. All batch mode filtration experiments were conducted with membranes, without the support of polyester fabric. However, dynamic cross flow experiments were undertaken by the membrane cast on polyester fabric (product number: TNW006013, supplied by M/s, Hollytex Inc., New York, USA). The viscosity of the casting solution was measured by a rheometer (model: Physica MCR 301, Anton Paar, USA) and reported in Table 1.

Table 1 Composition of various membranes

Nomenclature	PAN (w/w%)	PVP (w/w%)	Laterite (w/w%)	Viscosity of casting solution (Pa s)
M0	15	1.5	0	300 ± 10
M10			10	500 ± 12
M15			15	750 ± 15
M20			20	962 ± 16
M25			25	1430 ± 17

---

2.4. Surface characterization of the MMM

Surface characterization²³ of the prepared membranes was carried out using a scanning electron microscope (model: ESM-5800, JEOL, Japan) and an atomic force microscope (model: 5500 AFM, Agilent Technologies, USA). The zeta potential of the membrane was measured using sodium chloride solution (0.01 M) at a temperature of 298 ± 2.0 K, a transmembrane pressure drop of 0–2 bar, solution pH of 3.5–11.5 and a cross flow velocity of 0.12 m s⁻¹ in an indigenously fabricated cross flow ultrafiltration cell. Details of this procedure were available.²⁴ FTIR (Fourier transform infrared spectroscopy, supplied by M/s, Perkin Elmer, CT; model: Spectrum 100) of the MMM was conducted before and after the experiment with As(V) solution. X-ray diffraction patterns of laterite, MMM and pure polymeric membrane were carried out using an X-ray diffractometer (M/s, PANalytical, model: Xpert Pro, The Netherlands). The pore volume distribution and the Brunauer–Emmett–Teller (BET) surface area of the prepared membranes were determined using BET surface analyzer supplied by Quantachrome Instruments, Florida, USA (model AUTOSORB-1), with nitrogen as the adsorption medium (degassing temperature: 343 K, time: 24 h). The contact angle of different membranes (using the sessile drop method at room temperature) was measured by a goniometer (M/s, Rame-Hart Instrument Co., USA, model number: 200-F4 series).

2.5. Permeability, porosity and MWCO of MMM

Membrane permeability, porosity, MWCO and As(V) rejection were determined using the unstirred batch cell. Pure water flux was calculated using the following equation:where J_w is the water flux, Δv is the volume of water permeated in time Δt and A is the membrane area.

After evaluation of pure water flux at different transmembrane pressures (TMP), the obtained results were plotted against TMP (ΔP). The slope of the fitted straight line through the origin gave the permeability (L_p) of the membrane. The subsequent equation is given by

The bulk porosity of the membrane was determined gravimetrically by measuring the weight of the water-soaked membrane samples before and after drying.

where ε is the porosity of the membrane samples, w_o and w_l are the weight of the wet and dry membranes, respectively, ρ_w is the density of water, and A and l are the area and thickness of the membrane samples, respectively.

MWCO is the molecular weight of the solute that is 90% retained by the membrane. The percentage of solute retention can be measured as

where R is the percentage solute retention and C_p and C_f are the concentrations of solute in the permeate and feed stream, respectively.

2.6. Determination of adsorption capacity of MMM

A batch adsorption experiment was conducted in an orbital shaker (speed: 150 rpm; time: 24 hours; pH: 7 ± 0.2). Synthetic solutions of As(V) and As(III) were prepared using sodium arsenate heptahydrate (NaHAs₂O₄·7H₂O) and sodium arsenite (NaAsO₂), respectively, at different concentrations. The amount of arsenic adsorbed by the selected membrane was evaluated according to the following equation:where q_e is the amount of arsenic adsorbed by the membrane, c_o and c_e are the initial and equilibrium concentrations, v is the volume of the solution used and m is the amount of adsorbent used. Two types of linear isotherm equations, i.e., Langmuir and Freundlich,^25,26 were used to fit the adsorption data. Adsorption of As(V) on the membrane matrix was carried out at three different temperatures, 298, 308 and 318 K. The adsorption data have been fitted into Langmuir and Freundlich isotherms. Linear forms of the two isotherms are given as follows:where C_e is the equilibrium concentration, q_e is the amount of arsenic adsorbed in the solid phase, V_m is the maximum adsorption capacity, K is the adsorption equilibrium constant, and k_f and n are Freundlich isotherm constants, indicating the adsorption capacity and intensity. The average relative error (ARE) between the fitted and experimental values were obtained according to the following equation:where q_e,calc is the calculated value of the adsorption capacity from the isotherm fit, q_e,meas is the adsorption capacity measured experimentally, and n¹ is the number of data points.

2.7. Dynamic cross flow ultrafiltration experiments

The dynamic filtration of the selected membrane (with support) was performed in a cross flow cell in total recycle mode.²⁷ The effect of operating conditions (cross flow rate and TMP) was analyzed, and optimum conditions were selected in terms of arsenic removal efficiency. Filtration experiments for long duration were carried out to evaluate the stability of the membrane (both in terms of permeate flux and concentration). Experiments were terminated when arsenic concentration in the permeate exceeded the allowable limit (10 μg l⁻¹) and the membrane was regenerated using 1 M sodium hydroxide solution for 24 hours.¹⁸ The procedure was repeated for two cycles. The permeate stream was passed through a Whatman 41 ashless quantitative filter paper (pore size 20–25 μm) to check the leaching of any laterite particles (48 μm) and no particle was found to be retained.

Effects of coexisting anions (phosphate, sulfate, carbonate and bicarbonate) and pH of feed on arsenic removal ability of the membrane were assessed. Total arsenic concentration in all streams was measured using an atomic absorption spectrophotometer (model: analyst 700 coupled with MHS-15, PerkinElmer Instruments, USA).

2.8. Filtration experiments with an arsenic contaminated groundwater sample

Arsenic removal efficiency of the selected membrane (in dynamic filtration mode) was evaluated for a groundwater feed solution obtained from an affected area in West Bengal, India (viz., Rajarhat, Kolkata, 22.35°N, 88.2°E). The concentration of different ions (both anion and cation) present in the sample was measured by an ion chromatograph (model: 883 Basic IC Plus, Metrohm, Switzerland).

3. Results and discussion

3.1. Membrane morphology

The morphology of the membranes, both top and cross-sectional views, is shown in Fig. 1 and 2, respectively. The top view of the membrane M0 (refer to Table 1) shows the presence of pores. Comparing the surfaces of the two membranes at a higher magnification (1500×), it is observed that laterite particles are deposited on the top surface of the M25 membrane, which block the pores and decrease permeability. However, accumulation of these particles increases the hydrophilicity of the membranes, which is corroborated by contact angle measurements, discussed subsequently.

Fig. 1 Top views of the SEM images of membrane (I) M0 (500×), (II) M0 (1500×), (III) M25 (500×), and (IV) M25 (1500×).

Fig. 2 Cross-sectional views (150×) of the SEM images of membranes (I) M0, (II) M10, (III) M15, (IV) M20, and (V) M25.

The cross-sectional images of the prepared membranes are shown in Fig. 2. Tear drop-shaped voids along the cross section of membrane M0 (without laterite) are present. However, on impregnating laterite (mean diameter over volume or de Brouckere mean diameter: 48 μm) within the matrix, the voids tend to reduce in size. This might be attributed to the increase in viscosity of the casting solution due to addition of laterite. This results in delayed solvent (DMF)–nonsolvent (water) demixing, retarding the phase separation kinetics and leading to the formation of smaller sized macrovoids along the cross section of the membrane.^13,28 However, the structure becomes denser with blockage of macrovoids by laterite particles inside the matrix and the pore mouths at the top surface of the membrane.^14,29,30

Fig. 3 shows the AFM images of the prepared membranes. It is observed that the surface property of the membranes changes due to addition of inorganics.³¹ The surface of membrane M0 (pure polymeric) is almost free of any considerable features, although the size of surface features starts increasing from membrane M10 and a distribution of roughness pattern is observed along the surfaces of membranes M15 and M20. For membrane M25, the surface roughness is maximum, indicating the adherence of laterite on the membrane surface. Increased surface roughness favors adsorption of species.³² The mean square roughness of membrane M0 is 17.2 nm, while that of M25 is 117 nm.

Fig. 3 Atomic force microscopic images of membranes and their roughness. (a) M0; (b) M10; (c) M15; (d) M20; (e) M25.

3.2. XRD analysis

XRD analyses of membranes M0 and M25 and laterite are presented in Fig. 4. Laterite particles showed diffraction peaks at 2θ = 24.4°, 38.7°, 42.7°, 62.5° (PCPDF card no. 290173). The peaks represented intensified iron oxyhydroxide, like goethite. However, the presence of weakly crystalline ferrihydrite could also be identified from 2θ = 41.6°, 47.9°, 53.7°, 62.7°, 71.1°, 74.1°. Silicon dioxide was also present and can be observed from XRD peaks at 2θ = 31.1°, 24.3°, 59°, 81.7°.¹⁶ The diffraction peaks of pure PAN were noticed at 2θ = 16.4°, 29.2° for the M0 membrane.³³ A weak diffraction peak of PVP was also present at 2θ = 20°.³⁴ The existence of all these peaks in the diffraction pattern of membrane M25 suggests that laterite particles are successfully impregnated in the polymeric matrix.

Fig. 4 XRD patterns of the M25 and M0 membranes and laterite particles.

3.3. Membrane porosity and permeability

Variation of porosity and permeability of the prepared membranes with laterite concentration is presented in Fig. 5. The porosity of the M0 membrane is around 65%, whereas it decreases to 46% for the M25 membrane. The phenomenon occurs due to the blockage of pores by laterite particles, which can be observed from the SEM micrographs.

Fig. 5 Variation of porosity and permeability with laterite concentration in MMM.

The membrane permeability decreases with the laterite concentration almost linearly. For a pure polymeric membrane (M0), the permeability is 10.2 × 10⁻¹⁰ m Pa s⁻¹ and it decreases to 3.4 × 10⁻¹⁰ m Pa s⁻¹ for the M25 membrane. This result is in corroboration with the variation of porosity with laterite concentration.³⁵

3.4. Pore size distribution, MWCO, contact angle, and average pore radius

Cumulative pore volume distribution of the prepared membranes with the laterite concentration is shown in Fig. 6(a). Two observations can be made from this figure. First, the gradient of the distribution (indicating the rate of increase of pore volume for a specific pore size) gradually decreases for a specific membrane, and hence, the probability of finding pores beyond 80 Å becomes small. Second, the pore volume decreases with laterite concentration. For example, the cumulative pore volume for a pore radius of 339 Å is 0.16 cm³ g⁻¹ and 0.09 cm³ g⁻¹ for the M0 and M25 membranes, respectively. The surface areas of the prepared membranes are presented in Table 2. A decrease in surface area indicates the reduced pore size as discussed above. Therefore, variation of pore volume distribution of the membranes confirms the blockage of pores by laterite, which is aggravated by increasing its concentration.

Fig. 6 Variation of (a) cumulative pore volume of different membranes and (b) membrane properties (MWCO, pore radius, and contact angle).

Table 2 BET (Brunauer–Emmett–Teller) surface area of different membranes

Membrane	Surface area (m^2 g^−1) (standard deviation)^a
a The number of measurements for each membrane sample is 3.
M0	86.6 ± 2.5
M10	74± 5.5
M15	55 ± 4.5
M20	39 ± 4.0
M25	36 ± 5.0

---

Variation of MWCO, pore radius and contact angle of the prepared membranes with laterite concentration is given in Fig. 6(b). MWCO of the M0 membrane is around 110 kDa and this decreases to 48 kDa for the M25 membrane. The result is directly supported by the reduction in permeability and porosity as discussed earlier. The contact angle of the prepared membranes decreases from 75° (M0 membrane) to 55° (M25 membrane). This suggests that the hydrophilicity of the membranes is improved with laterite concentration.^36,37 The average pore radius of the membranes (estimated from surface area analysis) decreases with laterite concentration, as expected. The average pore radius of the M0 membrane is around 67 Å and that of the M25 membrane is 43 Å.

3.5. Selection of membrane

Experiments were conducted in an unstirred batch cell for three hours with As(V) spiked solution (C_o = 100 μg l⁻¹) and the results are shown in Fig. 7(a). It is observed that a sharp decrease in permeate concentration of As(V) occurs from the M0 to M10 membrane, after which it becomes gradual. For the M25 membrane, As(V) concentration in the permeate is 3 μg l⁻¹ at the end of three hours. Hence, in terms of arsenic removal capability, membrane M25 is selected for all subsequent experiments. It is noticed from Fig. 8(b) that the As(V) concentration of the permeate drops below the permissible limit (10 μg l⁻¹) within 35 minutes. The drop in As(V) concentration of the permeate is fast initially. After 45 minutes, equilibrium is attained and a further change in concentration is not observed. Fig. 7(c) shows FTIR transmittance peaks of the M25 membrane at various wavelengths. The peak at 2275 cm⁻¹ corresponds to the symmetrical stretching of nitrile (R–C≡N), whereas the one at 1775 cm⁻¹ refers to amide (C=O) stretching. Alkynyl stretch is observed at 1500 cm⁻¹. The peak at a lower wavelength, i.e., 990 cm⁻¹, corresponds to the iron oxide and hydroxide of iron.³⁸ Also, the presence of physically adsorbed water molecules is confirmed by the peak at 1630 cm⁻¹. Along with this, subsequent lowering of the peaks at the corresponding wavelengths proves the adsorption of As(V) by laterite in the membrane matrix.

Fig. 7 (a) Variation of concentration of As(V) in the permeate with laterite concentration in MMM; (b) profile of As(V) concentration in the permeate; (c) FTIR analysis of the M25 membrane before and after adsorption of As(V).

Fig. 8 (a) Adsorption isotherm of the M25 membrane for As(V) (298 K, 308 K, 318 K) and As(III) (298 K); (b) adsorption isotherm of the M25 and M0 membranes for As(V) (298 K).

3.6. Batch adsorption

Batch adsorption experiments were carried out in an orbital shaker to determine the adsorption capacity of the M25 membrane for As(V) (at different temperatures). The relevant adsorption parameters are listed in Table 3 and the adsorption isotherm of the best fitted model are presented in Fig. 8(a). Comparison of adsorption capacities of the M0 and M25 membranes is also presented in Fig. 8(b).

Table 3 Langmuir and Freundlich parameters^a

Temperature (K)	Solution medium	Langmuir model				Freundlich model
a Each measurement was conducted three times and standard deviation was reported.
		V m (mg g^−1)	K (l mg^−1)	ARE (%)	R ^2	K f	n	ARE (%)	R ^2
298	As(V)	1.4 ± 0.2	0.3 ± 0.01	18	0.88	0.3 ± 0.01	2 ± 0.1	11.4	0.97
308	As(V)	1.5 ± 0.1	0.1 ± 0.02	22	0.85	0.1 ± 0.02	1.3 ± 0.2	15.3	0.95
318	As(V)	2 ± 0.2	0.7 ± 0.05	26	0.81	0.2 ± 0.05	1.8 ± 0.5	22.1	0.91

---

3.6.1. Effect of temperature. At 298 K, the maximum monolayer adsorption capacity of membrane M25 is 1.4 mg g⁻¹, and it increases marginally with temperature. Adsorption intensity (n) decreases with temperature indicating favourable adsorption. Based on ARE and correlation coefficient values, it can be concluded that the mechanism of arsenic adsorption on laterite (in the membrane matrix) follows the Freundlich isotherm due to heterogeneous surface interaction.³⁹

3.6.2. Comparison of adsorption capacities of membranes M0 and M25. The adsorption capacities of membranes M0 and M25 are presented in Fig. 8(b). The adsorption for the M25 membrane is due to the impregnation of laterite in the membrane matrix (1.4 mg g⁻¹), whereas the adsorption of the pure polymeric membrane (M0) is low (0.3 mg g⁻¹). Hence, it can be concluded that the removal of As(V) by the M0 membrane is solely due to the weak electrostatic repulsion offered by the nitrile group of the polymer.

3.7. Effect of pH and the mechanism of As(V) removal from water

The zeta potential of membrane M25 at various solution pH values (0.01 M sodium chloride) is shown in Fig. 9(a). It is observed from this figure that the isoelectric point⁴⁰ of the membrane is around 7.9. At this pH, the membrane surface is devoid of any charge. At an operational pH below the isoelectric point (pH < pH_zpc), the surface becomes positively charged.⁴¹ Pentavalent arsenic exists in groundwater as negatively charged species, such as H₂AsO₄⁻, HAsO₄²⁻, and AsO₄³⁻. However, the first species is more prevalent in the pH range 4–8.^17,18 Therefore, the adsorption of negatively charged arsenate species is favored on the positively charged membrane surface. Interestingly, it is observed from Fig. 9(b) that the rejection is almost 85% at a neutral pH, when the zeta potential of the membrane is around 2.3 mV. At higher pH, i.e., pH > pH_zpc, the surface is negatively charged and it repels arsenate ions. This is evident from Fig. 9(b), where rejection at pH = 12 is around 50%.

Fig. 9 (a) Variation of the zeta potential of the M25 membrane with solution pH; (b) removal of As(V) with solution pH using the M25 membrane.

The procedure (for generation of acid–base treated laterite) helps in the loading of iron or aluminium oxyhydroxide on the laterite surface. These are fine powder-like particles that increase the surface area of the adsorbent. The increase in adsorption capacity can be attributed to the high surface area, which occurs due to the synergistic effect of the acid and base treatment of the laterite surface.¹⁸ Precipitation of iron and aluminium hydroxides can be shown by the subsequent chemical reactions.

Acid treatment step:

Fe₃O₄(solid) + 8H⁺ → 2Fe³⁺(aqueous) + Fe²⁺(aqueous) + 4H₂O (9)

Fe₂O₃(solid) + 6H⁺ → 2Fe³⁺(aqueous) + 3H₂O (10)

Al₂O₃(solid) + 6H⁺ → 2Al³⁺(aqueous) + 3H₂O (11)

Alkali treatment step:

Fe³⁺ + 3OH⁻ → FeOOH(iron oxyhydroxide precipitate) + H₂O (12)

Fe²⁺ + 2OH⁻ → Fe(OH)₂ (13)

4Fe(OH)₂ + O₂ + 2H₂O → 4FeOOH(iron oxyhydroxide precipitate) + 4H₂O (14)

Al³⁺ + 3OH⁻ → AlOOH(aluminium oxyhydroxide precipitate) + H₂O (15)

It can be well understood that adsorption occurs due to a ligand exchange mechanism as the metals (iron or aluminium) form complex salts with arsenate species (H₂AsO₄⁻) in the pH range of 4–8. The same interaction persists between laterite on the membrane surface and arsenate species in the feed. Further, the nitrile groups present in the polymer on the membrane surface repels the negatively charged arsenic species. This is evident from Fig. 8(a), where rejection by the M0 (pure polymeric) membrane is only around 23%. The interaction of arsenate with iron and aluminium oxyhydroxide of laterite can be represented as follows:

≡M–OH + H₂AsO₄⁻(As(V) in the pH range of 4–8) ↔ ≡MHAsO₄⁻ + H₂O (16)

≡M–OH⁺₂ + H₂AsO₄⁻(As(V) in the pH range of 4–8) ↔ ≡MHAsO₄ + H₃O⁺ (17)

where M stands for the metal (iron or aluminium). The density of the active sites (≡M–OH) increases with acid–base treatment and it increases the adsorption of arsenate on the membrane surface. The schematic of the arsenate interaction with the membrane matrix is represented in Fig. 10.

Fig. 10 Interaction of laterite in the membrane matrix with arsenate.

The ionic radius of arsenic species (0.46 Å) is much smaller than the pore radius of the membrane (67 Å). Therefore, the contribution of the sieving mechanism is insignificant.⁴² This can be verified from Fig. 7(a), where it is observed that the removal percentage of As(V) by the M0 membrane is only around 20%. Hence, it can be concluded that adsorption is the dominant removal mechanism, facilitated by electrostatic attraction.

3.8. Dynamic cross flow filtration

3.8.1. Effect of operating conditions. Removal efficiency of a membrane is affected by the operating conditions, i.e., cross flow rate and TMP.⁴³ In order to select the best operating conditions, the membrane was subjected to a run of three hours for various TMP and cross flow rates. The feed concentration was 100 μg l⁻¹ for all experiments. Variation of As(V) concentration in permeates with TMP and cross flow rate is presented in Fig. 11(a) and (b), respectively. The flux decline profile for the same sets of operating conditions are shown in Fig. 11(c) and 11(d) respectively. It is apparent from Fig. 11(a) that the permeate concentration increases with TMP, as the driving force increases, at a fixed cross flow rate. The reason is due to decreased contact time (at higher TMP) between laterite and arsenic species in water. The permeate concentration (after two hours of operation) increases from 2.9 μg l⁻¹ (276 kPa) to 11.9 μg l⁻¹ (552 kPa). Thus, the best performance is observed in case of 276 kPa, where As(V) concentration in the permeate remains below 10 μg l⁻¹. Increased driving force results in higher permeate flux at higher TMP. For example, the flux at the end of two hours is 660 l m⁻² h⁻¹ and 295 l m⁻² h⁻¹ for 551 kPa and 276 kPa, respectively.

Fig. 11 Effect of operating conditions on the performance of the M25 membrane in continuous cross flow mode. (a) Profiles of permeate concentration at different TMP; (b) profiles of permeate concentration at different cross flow rates; (c) profiles of permeate flux at different TMP; (d) profiles of permeate flux at different cross flow rates.

Adsorption efficiency also changes with cross flow rate. As observed in Fig. 11(b), the permeate concentration is higher for 80 l h⁻¹ than that for 60 l h⁻¹ (TMP: 276 kPa). At a higher cross flow rate, adsorption is less due to reduced contact time, thereby increasing the arsenic concentration in the permeate. Since adsorption resistance is lower, the permeate flux increases at a higher cross flow rate. Thus, the flux increases from 295 l m⁻² h⁻¹ (60 l h⁻¹) to 315 l m⁻² h⁻¹ (80 l h⁻¹).

3.8.2. Stability of the membrane and regeneration studies. The performance of the membrane (both in terms of arsenic removal efficiency and permeate flux) in continuous cross flow mode was evaluated under the selected operating conditions and the results are reported in Fig. 12(a) and (b). The concentration of the permeate stream was plotted against n (number of times multiplied by the membrane bed volume). The relation between different parameters with n can be represented as

Fig. 12 (a) As(V) removal from the permeate with continuous filtration and (b) permeate flux for various regeneration cycles of the M25 membrane; (c) effect of co-existing anions on adsorption of As(V) on the M25 membrane.

It is observed from Fig. 12(a) that the breakthrough volume (volume of water processed having an arsenic concentration below 10 μg l⁻¹) for a pristine membrane is approximately 600 × 10³ times the volume of the membrane bed (1.5 × 10⁻⁶ m³) and it decreases after each regeneration cycle. For example, the breakthrough time after the first regeneration is 550 × 10³ m³ of bed volume and it decreases to 400 × 10³ m³ of bed volume after the fourth regeneration. After each regeneration cycle, complete desorption does not occur resulting in a decrease in breakthrough volume.^44,45

It is observed from Fig. 12(b) that the flux decline remains insignificant for the initial eight hours. Initially, adsorptive fouling does not play an important role in affecting the productivity of the membrane. However, the decline becomes prominent within 8 to 14 hours and gradual thereafter, as adsorptive resistance becomes dominant. For example, after the first regeneration, the permeate flux decreases from 272 to 220 l m⁻² h⁻¹ within 8 to 14 hours. The adsorption sites start getting saturated after 14 hours leading to an increase in arsenic concentration as shown by Fig. 12(a). Since an increase in adsorption resistance after 14 hours is not significant, the reduction in permeate flux is gradual. As discussed in the preceding paragraph, the regeneration of MMM is not complete. Therefore, the permeate flux is reduced after each cycle of regeneration. For example, at the beginning of the experiment, permeate flux was 290 l m⁻² h⁻¹ for pristine membrane and this value is 265 l m⁻² h⁻¹ after the first regeneration. It is reduced further to 170 l m⁻² h⁻¹ after the fourth regeneration.

3.8.3. Effect of coexisting anions. Variation of removal efficiency of As(V) by the M25 membrane in the presence of co-existing anions is presented in Fig. 12(c). The anions selected for this study are phosphate, sulfate, carbonate and bicarbonate. The removal of arsenic without any anion concentration is around 93% at a neutral pH. However, this is affected by phosphate, which brings down the removal percentage to 60% when its concentration is around 400 mg l⁻¹. At the same concentration of the sulphate group, the removal percentage is around 72%. The roles of bicarbonate and carbonate are negligible as they do not affect the removal percentage significantly.⁴⁶ Competitive adsorption of other anions over As(V) by the active sites (laterite) is an important factor in this context.⁴⁷

3.8.4. Performance of MMM for arsenic-contaminated groundwater solution. The dynamic filtration capability of the M25 membrane was tested by arsenic-contaminated groundwater (C_o = 94 μg l⁻¹) for 17 hours, under chosen operating conditions, i.e., cross flow rate of 60 l h⁻¹ and 276 kPa TMP. Profiles of permeate flux and concentration are shown in Fig. 13. Concentrations of other ionic species are presented in Table 4. It can be observed from Fig. 13 that the permeate flux remains steady for 7 hours due to low adsorptive resistance (as discussed previously). During this time, the concentration of permeate almost remains near zero. However, as filtration progresses, permeate concentration exceeds the permissible limit (10 μg l⁻¹). Permeate flux decreases from 260 l m⁻² h⁻¹ to 210 l m⁻² h⁻¹ during this period, as adsorptive fouling becomes significant. Hence, for a filter with the M25 membrane having a filtration area of 0.01 m², the time of producing safe drinking water is around 13–14 hours with a production rate above 200 l m⁻² h⁻¹. For a scaled-up filter, one can have a higher life. The obtained breakthrough time (13.5 hours) for the real-life feed is slightly less than that of the synthetic solution (refer to Fig. 12a). This is due to the presence of several anions (refer to Table 4) that compete for the adsorption sites with arsenic.⁴⁸

Fig. 13 Profiles of permeate flux and arsenic concentration for an arsenic-contaminated real-life feed sample for the M25 membrane.

Table 4 Characteristics of feed and permeate (after 17 hours) of the real-life sample^a

Parameters	Feed	Permeate (1 hour)	Permeate (17 hour)
a Note: BDL—below detection limit; number of measurements is 3 in each case.
pH	7.4 ± 0.1	7.1 ± 0.2	7.4 ± 0.1
Chlorine (Cl^−), mg l^−1	0.03 ± 0.001	0.01 ± 0.002	BDL
Fluoride (F^−), mg l^−1	0.05 ± 0.004	0.02 ± 0.001	BDL
Nitrite (NO2^−), mg l^−1	0.3 ± 0.01	0.2 ± 0.03	BDL
Nitrate (NO3^−), mg l^−1	0.2 ± 0.04	0.15 ± 0.05	BDL
Hydrogen phosphate (HPO4^2−), mg l^−1	0.1 ± 0.02	0.1 ± 0.03	BDL
Iron (Fe), mg l^−1	1.9 ± 0.3	1.5 ± 0.12	BDL
Calcium (Ca), mg l^−1	0.05 ± 0.002	0.04 ± 0.001	0.02 ± 0.005
Magnesium (Mg), mg l^−1	1.3 ± 0.2	1.2 ± 0.1	1 ± 0.09
Sodium (Na), mg l^−1	60.9 ± 2	60.1 ± 1	58.2 ± 2
Potassium (K), mg l^−1	3.6 ± 0.4	3 ± 0.3	2.8 ± 0.2

---

3.8.5. Performance of MMM for arsenic-contaminated groundwater solution. Adsorption capacity of treated laterite with other adsorbents is presented in Table 5. It is observed that all the variants of laterite have a lower adsorption capacity compared to treated laterite. In fact, untreated laterite has the lowest capacity of 0.4 mg g⁻¹.⁴⁴ Fe-modified activated carbon⁸ has a capacity that is almost double with respect to treated laterite. But this adsorbent is a synthetic product.

Table 5 Comparison of As(V) adsorption capacity of different adsorbents with treated laterite

Type	Adsorption capacity (mg g^−1)	pH	Reference
Laterite iron concretions	1.0	7.0	7
Fe-modified activated carbon	51.3	6.0	8
Treated laterite	24.8	7.0	17
Untreated laterite	0.4	5.5	44
Coconut shell carbon	2.4	5	48

---

Comparison of performance of different mixed matrix membranes for the removal of As(V) with the present one is shown in Table 6. Interestingly, no data of naturally occurring micro-particle embedded MMM are available. Two references on zirconia nanoparticle-embedded MMM^14,15 and one on iron(III) oxide-loaded chitosan hollow fiber¹⁶ were found. It is observed from this table that the specific flux of the PAN–laterite MMM is comparable with the zirconia nanoparticle-embedded PSF MMM¹⁴ and it is better than the iron-loaded chitosan hollow fiber.¹⁶ Hydrous zirconia nanoparticle–PVDF MMM showed the maximum specific flux. Considering the energy consumption per unit membrane area (m²) per unit volume (m³) of processed water, PAN–laterite MMM has a comparable value with respect to hydrous zirconia nanoparticle–PVDF MMM. It is also clear from Table 6 that PAN–laterite MMM has the lowest manufacturing cost, i.e., US$ 20 per m² membrane area. Performance index in terms of liter of water processed up to breakthrough point (10 μg l⁻¹) per gram of inorganic in the matrix is also presented in this table. PAN–laterite MMM shows a performance index better than the hydrous zirconia–PVDF membrane¹⁵ but lower than the zirconia–PSF blend membrane. It may be noted here that the iron(III)-loaded chitosan hollow fiber membrane was operated under vacuum (10⁻¹¹ torr). Thus, its performance may not be directly comparable with other pressure-driven systems.

Table 6 Comparison of performances of different mixed matrix membranes for the removal of As(V)

Type	Specific flux (l m^−2 h^−1 kPa^−1)	Energy consumption (kW h per m^2 membrane area per m^3 of processed water)^a	Cost of membrane per m^2 (US$)^b	Performance index (liter of water processed up to breakthrough point (10 μg l^−1) per gram of inorganic in the matrix)^b	Reference
a Calculated according to the following equation:where Q is the cross flow rate (m^3 s^−1), ΔP is the transmembrane pressure drop (Pa), η is the pump efficiency, and Jwp is the permeate flux (m s^−1). b It is assumed that 100 ml of casting solution provides 2 A4 (0.06 m^2) sized flat sheet membranes and 150 hollow fibers.
Zirconia nanoparticle-embedded PSF blend hollow fiber membranes	1.25	—	354	141	14
Hydrous zirconia nanoparticles–PVDF flat sheet membrane	8.4	0.015	667	4	15
Iron(III)-loaded chitosan hollow fiber membrane	2.2 × 10^−4	3.6 × 10^−9	1233	—	16
Treated laterite–PAN mixed matrix membranes	0.9	0.025	20	20.3	This study

---

4. Conclusions

A novel PAN–laterite particle MMM was developed, characterized and used for removal of arsenic from groundwater. The scanning electron micrographs indicate that the membrane becomes denser with laterite concentration. The permeability of MMM decreased from 10.2 × 10⁻¹⁰ m Pa s⁻¹ to 3.4 × 10⁻¹⁰ m Pa s⁻¹ as the laterite concentration increases from 0 to 25 wt%. A corresponding decrease in MWCO of the membrane was from 110 kDa to 48 kDa, associated with a reduction in pore radius from 67 Å to 43 Å. Incorporation of laterite increased the hydrophilicity of the membrane and its contact angle decreased from 75° to 55°. However, the arsenic removal capacity of the MMM increases up to 92% (for 25 w/w% of laterite in the membrane). Adsorption of arsenate within laterite embedded in the membrane matrix was confirmed by the reduction in FTIR peaks of appropriate functional groups. Adsorption capacity of arsenate on MMM was found to be 1.4 mg g⁻¹ at room temperature. The pH at the point of zero charge (pH_zpc) of MMM was found to be 7.9. The membrane surface becomes positively charged at the operating pH and helps in the removal of negatively charged arsenate species by adsorption, facilitated by electrostatic attraction. The optimum operating conditions (in terms of arsenic removal percentage) was found to be 276 kPa TMP and 60 l h⁻¹ cross flow rate. High permeate flux of ~225 l m⁻² h⁻¹ was obtained, both for synthetic solution and arsenic-contaminated groundwater. The breakthrough volume, based on WHO limit of arsenic concentration, was 600 × 10³ m³ of bed volume (1.5 × 10⁻⁶ m³) for the synthetic solution. The limiting time (permeate concentration below 10 μg l⁻¹) was 14 hours for the arsenic-contaminated real-life feed solution.

Nomenclature

A	membrane area, m^2
ARE	average relative error
C p	concentration of solute in permeate, mg l^−1
C f	concentration of solute in feed, mg l^−1
C 0	arsenic concentration in the feed, mg l^−1
C e	equilibrium concentration of arsenic in solution, mg l^−1
i	number of data points
J w	distilled water flux, m s^−1
v	volume of the filtrate collected for batch adsorption, l
V m	maximum monolayer adsorption capacity (Langmuir isotherm), mg g^−1
K	adsorption equilibrium constant (Langmuir isotherm), l mg^−1
k f	adsorption intensity (Freundlich isotherm), mg^((n−1)/n) l^1/n g^−1
l	membrane thickness, m
L p	permeability of the membrane, m Pa s^−1
m	mass of adsorbent (membrane samples) used, g
n ^1	total number of data points
n	adsorption intensity (Freundlich isotherm)
q e	amount of arsenic adsorbed, mg g^−1
q e,meas	experimentally measured value of qe, mg g^−1
q e,calc	fitted value of qe from isotherm, mg g^−1
R	rejection, %
w l	weight of the membrane after drying, k.
w o	weight of the membrane before drying, k.

Greek symbols

ε	membrane porosity
ρ w	density of water, kg m^−3
ΔP	transmembrane pressure drop, P.
Δt	time interval over which permeate samples are taken, s
Δv	volume of permeate collected over time period, m^3

Acknowledgements

This work is partially supported by a grant from the SRIC, IIT Kharagpur under scheme no. IIT/SRIC/CHE/SMU/2014-15/40, dated 17-04-2014. Any opinions, findings and conclusions expressed in this paper are those of the authors. The authors also acknowledge the help of Anirban Roy and Mrinmoy Mondal, research scholars from the Department of Chemical Engineering, Kharagpur, India.

References

1. S. K. Acharyya, S. Lahiri, B. C. Raymahashay and A. Bhowmik, Arsenic toxicity of groundwater in parts of the Bengal basin in India and Bangladesh: the role of quaternary stratigraphy and Holocene sea-level fluctuation, Environ. Geol., 2000, 39, 1127–1137.
2. F. W. Pontius, K. G. Brown and C. J. Chen, Health implication of arsenic in drinking water, J. – Am. Water Works Assoc., 1994, 86(9), 52–63.
3. World Health Organization, Guidelines for Drinking Water Quality, 2006.
4. C. F. Harvey, C. H. Swartz, A. B. M. Badruzzaman, N. Keon-Blute, W. Yu, M. A. Ali, J. Jay, R. Beckie, V. Niedan, D. Brabander, P. M. Oates, K. N. Ashfaque, S. Islam, H. F. Hemond and M. F. Ahmed, Arsenic mobility and groundwater extraction in Bangladesh, Science, 2002, 298, 1602–1606.
5. L. M. Del Razo, M. A. Arellano and M. E. Cebarian, The oxidation states of arsenic in well-water from a chronic arsenicism area of northern Mexico, Environ. Pollut., 1990, 64, 143–153.
6. S. Dixit and J. G. Hering, Comparison of arsenic (V) and arsenic (III) sorption onto iron oxide minerals: implications for arsenic mobility, Environ. Sci. Technol., 2003, 37, 4182–4189.
7. F. Partey, D. Norman, S. Ndur and R. Nartey, Arsenic adsorption onto laterite concretions: Temperature effect, J. Colloid Interface Sci., 1998, 321, 493–500.
8. W. Chen, R. Parette, J. Zou, F. S. Cannon and B. A. Dempsey, Arsenic removal by iron modified activated carbon, Water Res., 2007, 41, 1851–1858.
9. K. Kosutic, L. Furac, L. Sipos and B. Kunst, Removal of arsenic and pesticides from drinking water by nanofiltration membranes, Sep. Purif. Technol., 2005, 42, 137–144.
10. P. Mondal, N. Hermans, A. T. K. Tran, Y. Zhang, Y. Fang, X. Wang and B. V. der Bruggen, Effect of physico-chemical parameters on inorganic arsenic removal from aqueous solution using a forward osmosis membrane, J. Environ. Chem. Eng., 2014, 2, 1309–1316.
11. R. E. Kesting, in Synthetic Polymeric Membranes, McGraw Hill, New York, 1971.
12. L. Y. Ng, A. W. Mohammad, C. P. Leo and N. Hilal, Polymeric membranes incorporated with metal/metal oxide nanoparticles: a comprehensive review, Desalination, 2010, 8, 15–33.
13. R. J. Gohari, W. J. Lau, T. Matsuura and A. F. Ismail, Fabrication and characterization of novel PES/Fe–Mn binary oxide UF mixed matrix membrane for adsorptive removal of As(III) from contaminated water solution, Sep. Purif. Technol., 2013, 118, 64–72.
14. J. He, T. Matsuura and J. P. Chen, A novel Zr-based nanoparticle-embedded PSF blend hollow fiber membrane for treatment of arsenate contaminated water: Material development, adsorption and filtration studies, and characterization, J. Membr. Sci., 2014, 452, 433–445.
15. Y.-M. Zheng, S.-W. Zou, K. G. N. Nanayakkara, T. Matsuura and J. P. Chen, Adsorptive removal of arsenic from aqueous solution by a PVDF/zirconia blend flat sheet membrane, J. Membr. Sci., 2011, 374, 1–11.
16. M. S. S. Dorraji, A. Mirmohseni, F. Tasselli, A. Criscuoli, M. Carraro, S. Gross and A. Figoli, Preparation, characterization and application of iron (III)-loaded chitosan hollow fiber membranes as new bio-based As(V) adsorbent, J. Polym. Res., 2014, 21, 399–412.
17. S. K. Maji, A. Pal and T. Pal, Arsenic removal from real-life groundwater by adsorption on laterite soil, J. Hazard. Mater., 2008, 151, 811–820.
18. A. Maiti, J. K. Basu and S. De, Development of treated laterite for arsenic adsorption: Effects of treatment parameters, Ind. Eng. Chem. Res., 2010, 49, 4873–4886.
19. H. R. Lohokare, M. R. Muthu, G. P. Agarwal and U. K. Kharul, Effective arsenic removal using polyacrylonitrile-based ultrafiltration (UF) membrane, J. Membr. Sci., 2008, 320, 159–166.
20. W. Li, Z. Yang, Q. Meng, C. Shen and G. Zhang, Thermally stable and solvent resistant self-crosslinked TiO₂/PAN hybrid hollow fiber membrane fabricated by mutual supporting method, J. Membr. Sci., 2014, 467, 253–261.
21. H. Basri, A. F. Ismail and M. Aziz, Polyethersulfone (PES)–silver composite UF membrane: Effect of silver loading and PVP molecular weight on membrane morphology and antibacterial activity, Desalination, 2011, 273, 72–80.
22. S. Chatterjee and S. De, Adsorptive removal of fluoride by activated alumina doped cellulose acetate phthalate (CAP) mixed matrix membrane, Sep. Purif. Technol., 2014, 125, 223–238.
23. R. J. Gohari, W. J. Lau, T. Matsuura, E. Halakoo and A. F. Ismail, Adsorptive removal of Pb(II) from aqueous solution by novel PES/HMO ultrafiltration mixed matrix membrane, Sep. Purif. Technol., 2013, 120, 59–68.
24. Y. Zhao, W. Xing, N. Xu and F. S. Wong, Effects of inorganic electrolytes on zeta potentials of ceramic microfiltration membranes, Sep. Purif. Technol., 2005, 42, 117–121.
25. S. Shrestha, G. Son, S. H. Lee and T. G. Lee, Isotherm and thermodynamic studies of Zn (II) adsorption on lignite and coconut shell-based activated carbon fiber, Chemosphere, 2013, 92, 1053–1061.
26. A. Masoumi and M. Ghaemy, Removal of metal ions from water using nanohydrogel tragacanth gum-g-polyamidoxime: Isotherm and kinetic study, Carbohydr. Polym., 2014, 108, 206–215.
27. S. Mondal, Chhaya and S. De, Prediction of ultrafiltration performance during clarification of stevia extract, J. Membr. Sci., 2012, 396, 138–148.
28. G. Arthanareeswaran and P. Thanikaivelan, Fabrication of cellulose acetate-zirconia hybrid membranes for ultrafiltration applications: Performance, structure and fouling analysis, Sep. Purif. Technol., 2010, 74, 230–235.
29. G. Bottino, V. Capannelli, V. D'Asti and P. Piaggio, Preparation and properties of novel organic–inorganic porous membranes, Sep. Purif. Technol., 2001, 22–23, 269–275.
30. G. Bottino, V. Capannelli and A. Comite, Preparation and characterization of novel PVDF-ZrO₂ composite membranes, Desalination, 2002, 146, 35–40.
31. H. Rajabi, N. Ghaemi, S. S. Madaeni, P. Daraei, M. A. Khadivi and M. Falsafi, Nanoclay embedded mixed matrix PVDF nanocomposite membranes: Preparation, characterization and biofouling resistance, Appl. Surf. Sci., 2014, 313, 207–214.
32. M. Khayet, K. C. Khulbe and T. Matsuura, Characterization of membranes for membrane distillation by atomic force microscopy and estimation of their water vapour transfer coefficients in vacuum membrane distillation process, J. Membr. Sci., 2004, 238, 199–211.
33. X. H. Flora, M. Ulaganathan and S. Rajendran, Influence of Lithium Salt Concentration on PAN–PMMA Blend Polymer Electrolytes, Int. J. Electrochem. Sci., 2012, 7, 7451–7462.
34. A. Rawat, H. K. Mahavar, S. Chauhan, A. Tanwar and P. J. Singh, Optical band gap of polyvinylpyrollidone/ polyacrilamide blend thin films, Indian J. Pure Appl. Phys., 2012, 50, 100–104.
35. D. J. Miller, S. Kasemset, L. Wang, D. R. Paul and B. D. Freeman, Constant flux cross flow filtration evaluation of surface-modified fouling-resistant membrane, J. Membr. Sci., 2014, 452, 171–183.
36. N. M. Wara, L. F. Francis and B. V. Velamakanni, Addition of alumina to cellulose acetate membranes, J. Membr. Sci., 1995, 104, 43–49.
37. L. Palacio, J. I. Calvo, P. Pradanos, A. Hernandez, P. Vaisanen and M. Nystrom, Contact angles and external protein adsorption onto UF membranes, J. Membr. Sci., 1999, 152, 189–201.
38. J. Coates, Interpretation of infrared spectra, a practical approach, in Encyclopedia of Analytical Chemistry, ed. R. A. Meyers, John Wiley & Sons Ltd., Chichester, 2000, pp. 10815–10837.
39. J. S. Piccin, G. L. Dott and L. A. A. Pinto, Adsorption isotherms and thermochemical data of FD&C red N0 40 binding by chitosan, Braz. J. Chem. Eng., 2011, 28, 295–304.
40. S. S. Tripathy and A. M. Raichur, Abatement of fluoride from water using manganese dioxide coated activated alumina, J. Hazard. Mater., 2008, 153, 1043–1051.
41. S. Ran, A. R. D. Verliefde, J. Hu, Z. Zeng, J. Lu, A. J. B. Kemperman, H. Deng, K. Nijmeijer, S. G. J. Heijman and L. C. Rietveld, Tight ceramic UF membrane as RO pretreatment: the role of electrostatic interactions on phosphate rejection, Water Res., 2014, 48, 498–507.
42. D. R. Latulippe, A. M. Mika, R. F. Childs, R. Ghosh and C. D. M. Filipe, Flux performance and macrosolute sieving behaviour of environment responsive formed-in-place ultrafiltration membranes, J. Membr. Sci., 2009, 342, 227–235.
43. R. Mukherjee and S. De, Adsorptive removal of nitrate by polyacrylonitrile–alumina nanoparticle mixed matrix hollow fiber membrane, J. Membr. Sci., 2014, 466, 281–292.
44. C. W. Cheung, J. F. Porter and G. Mckay, Sorption kinetic analysis for the removal of cadmium ions from effluents using bone char, Water Res., 2001, 35, 605–612.
45. S. De and A. Maiti, Arsenic Removal from Contaminated Groundwater, TERI Press, New Delhi, India, 2012.
46. X. Meng, G. P. Korfiatis, S. Bang and K. W. Bang, Combined effects of anions on arsenic removal by iron hydroxides, Toxicol. Lett., 2002, 133, 103–111.
47. X. Meng, S. Bang and G. P. Korfiatis, Effect of silicate, sulphate and carbonate on arsenic removal by ferric chloride, Water Res., 2000, 34, 1255–1261.
48. L. Lorenzen, J. V. Deventer and W. M. Landi, Factor affecting the mechanism of adsorption of arsenic species on activated carbon, Miner. Eng., 1995, 8, 557–569.

---