Enhancing the permeation selectivity of sodium alginate membrane by incorporating attapulgite nanorods for ethanol dehydration

Abstract

Hybrid membranes for ethanol dehydration were fabricated by blending sodium alginate with natural hydrophilic attapulgite nanorods, which contained plentiful selective channels and hydrophilic –OH groups. With the incorporation of attapulgite nanorods, the crystallinity of hybrid membranes was gradually decreased and the content of non-freezable water in hybrid membranes was increased, facilitating the solution-diffusion process of water molecules by forming hydration layers along the nanorods. The water uptake of hybrid membranes was ∼10% higher than the pristine alginate membrane while the swelling degree in feed solution was only increased by ∼1%, exhibiting good structural stability in ethanol dehydration. The optimum separation performance with a permeate flux of 1356 g m⁻² h⁻¹ and a separation factor of 2030 for dehydration of a 90/10 wt% ethanol/water feed was achieved using the hybrid membrane with 2 wt% of attapulgite nanorods. Moreover, the influences of feed temperature and feed composition on separation performance were investigated.

---

1. Introduction

Membrane technology has become one of the most effective and energy-saving methods to separate gas and liquid mixtures.^1–6 Pristine polymer membranes often suffer from the trade-off effect between permeability and selectivity.^7,8 Introducing inorganic fillers into polymer membranes as the second phase is an effective strategy to overcome this obstacle.^6,9–12 The introduced inorganic fillers can disrupt polymer chain packing,¹³ increase void volume¹⁴ and interact favorably with permeation molecules.¹⁵ Thus the enhancement of separation performance could be credited to the synergistic effect of polymer structure change and the high selectivity of inorganic fillers.

In the pervaporation dehydration process, both the polymer matrix and the introduced inorganic fillers should interact favorably with water molecules when the permeation molecules are water. Polyvinyl alcohol (PVA), chitosan (CS) and sodium alginate (SA) have been proved as versatile polymer matrices for alcohols dehydration due to their high hydrophilicity and good film-forming property.^16–19 Natural clay minerals such as kaolinite, montmorillonite and halloysite, etc. serving as inorganic fillers have attracted great attention for the fabrication of hybrid membranes due to their unique structure (different layered structures), thermal and mechanical properties. Usually, the clay minerals are hydrophilic and have an excellent water adsorption property,^20–22 which can facilitate the water solution process in membrane separation process. In general, water absorption on clay minerals is multidimensional. One hand, water can absorb within the space between the individual building blocks (e.g., silica and alumina or magnesia) by the hydration steps corresponding to one, two and sometimes three water layers between building blocks.²³ On the other hand, water molecules can bond on the clay external surfaces by hydrogen bond effect to form the initial monolayers,²⁴ which can be treated as the part of clay minerals. After that, water bulk adsorption is triggered and then the hydration layers form along the clay external surfaces, which would facilitate the diffusion process.²⁵ However, the different clay families have the different water adsorption properties due to the various arrangements of building blocks. The smectite clay family, which is considered to be the 2 : 1 type clay in which each aluminosilicate layer contains two long double silica tetrahedral chains sandwiching an octahedral layer of either magnesium (Mg) or aluminum (Al) hydroxide,²⁶ has superior water adsorption strength and adsorption capacity over other clay minerals.²⁷ Most of the introduced smectite clays in hybrid membranes for pervaporation are two-dimension sheets, which lead to the formation of tortuous diffusion channels in hybrid membranes to improve the selectivity. However, the analytical model to describe molecular transport in hybrid membrane demonstrates that the randomly oriented fillers with a high aspect ratio are preferred to provide a high permeability.²⁸ Consequently, one-dimension nanostructure materials with high water absorption property can be exploited as promising fillers to fabricate hybrid membranes for such applications as pervaporative dehydration. As a part of the smectite family, attapulgite (AT) appears to be the one-dimension nanostructure intrinsically. Meanwhile, there exist a lot of channels²⁶ paralleling to the long axis of AT nanorods, which can serve as the diffusion channels to sieve water molecules over other molecules. Thus the AT nanorods may serve as an effective additive (second phase) to construct the straight and selective channels in hybrid membranes.

In water permselective membranes, water states and their ratios are very important according to the well-known solution-diffusion theory.^29–31 The water absorbed in polymers, which can act as a plasticizer to change the packing of polymer chains and enhance separation performance of the considered polymer, is present in three states: free water, freezable bound water and non-freezable water.^29,31–33 Free water, which attributes to the confinement effect of polymer chains, acts like bulk water in polymers and freezes at 0 °C. Freezable bound water, which retains in polymers by the weak interactions of the water molecules with the polar groups, freezes at a temperature lower than 0 °C. And non-freezable water, which retains in polymers by the strong interactions of the water molecules with the polar groups²⁹ or by disrupting the interface regions between crystalline and amorphous region,³⁴ does not freeze at 0 °C or even below −100 °C. Due to the “bound” term, Hirata³⁰ called the bound water (freezable bound water and non-freezable bound water) as hydration water. But the term ‘hydration water’ is not suitable when both freezable bound water and non-freezable bound water are present in the polymers. Thus, in this study, the water retained in membranes is termed as freezable water, which includes free water and freezable bound water, and non-freezable water according to the difference of freeze behavior. And the water retained in hydrophilic polymers as freezable water can only penetrate through the amorphous regions.^34,35 Specially, non-freezable water can attract other water molecules to enhance the solution process and to form hydration layers facilitating the diffusion process of water molecules in membranes.^36,37

In the present study, hydrophilic AT nanorods with selective channels were introduced into sodium alginate (SA) matrix to fabricate hybrid membranes for pervaporative dehydration of ethanol. The water states were measured by differential scanning calorimetry (DSC) and their contents in membranes were calculated and analyzed. X-ray diffraction (XRD) was used to analyze the crystallinity of membranes. Moreover, the water uptake, swelling degree of membranes were investigated. Pervaporation experiments were performed to evaluate the membrane performance in ethanol dehydration.

2. Experimental

2.1 Materials

Attapulgite (AT) was obtained from Zhengzhou University. Sodium alginate (SA) was supplied by Qingdao Bright Moon seaweed Group Co. Ltd. (Shandong, China). Polyacrylonitrile (PAN) ultrafiltration membranes with a molecular weight cut-off of 100 kDa were obtained from Shandong Lanjing Membrane Engineering & Technology Co. Ltd. (Shandong, China). Calcium chloride dihydrate (CaCl₂·2H₂O) and absolute ethanol were purchased from Tianjin Guangfu Technology Development Co. Ltd. (Tianjin, China). All the reagents were of analytical grade and used without further purification. Deionized water was used through the whole experiment.

2.2 Preparation of the hybrid membranes

The hybrid membranes were fabricated by blending and spin coating methods. Typically, a certain amount of AT nanorods was dispersed in 35 ml water. And then 0.70 g SA was dissolved in above solution under stirring for 4 h at 30 °C. After being filtrated and degassed, the homogenous solutions were spin-coated onto the dry PAN substrates. And the rotation process was first set at 500 rpm for 20 s and then increased to 800 rpm for 40 s. The solvents of composite membranes were evaporated after being placed at room temperature for 24 h. The dry membranes were crosslinked by 0.5 M CaCl₂ solution for 10 min and then rinsed three times by pure water. The cross-linked membranes were obtained after drying at room temperature. The resultant composite membranes were named as SA/PAN and SA-AT(X)/PAN, where X meant the mass percentage of AT nanorods to SA matrix, ranging from 0 wt% to 8 wt%. And the corresponding homogenous membranes were prepared on glass plates instead of PAN substrates and denoted as SA pristine and SA-AT(X) for characterizations.

2.3 Characterization of AT nanorods and membranes

The morphology of AT nanorods was observed by transmission electron microscopy (TEM, JEOL JEM-100CXII). The surface morphology and cross-section morphology of membranes were examined by scanning electron microscope (SEM, Nanosem 430).

The chemical property of AT nanorods was characterized by Fourier transform infrared spectra (FT-IR, BRUKER Vertex70). The crystallization property of AT nanorods and membranes was investigated by a D/MAX-2500 X-ray diffractometer (CuKα). TGA (NETZSCH-TG 209F3) was conducted to acquire the thermal properties of AT nanorods and membranes under nitrogen flow. The glass transition temperature (T_g) of membranes was determined by differential scanning calorimetry (DSC, 204 F1 NETZSCH). To investigate the hydrophilicity of membrane surface, the static contact angles of composite membranes were measured by JC2000C Contact Angle Meter.

2.4 Water states

The contents of water in different states were characterized by DSC. Typically, a certain amount (about 5 mg) of dry membranes and a determined amount of water (about 5 mg) were tightly sealed in a pan to ensure that there was no loss of weight because of water vapor during the DSC process. After conditioned at temperature for 24 h, each sample pan was cooled down to −40 °C and then was gradually warmed up to 25 °C at a heating rate of 10 °C min⁻¹. Each sample was measured three times to guarantee the reproducibility. The amount of freezable water (W_f) was estimated from the DSC curves by the following equation,where Q_f was the melting heat of freezable water obtained from the DSC peak area and ΔH_m was assumed to be the melting heat of pure water (334 J g⁻¹).

The amount of non-freezable water (W_nf) was obtained from the difference between the amount of total water (W_c, W_c = (water, g)/(dry polymer, g)) and that of freezable water.

W_nf = W_c − W_f (2)

2.5 Water uptake

The AT nanorods and membranes were first dried in vacuum oven at 40 °C, weighed (M_D) and then immersed in deionized water at room temperature till complete hydration to obtain the wet mass (M_W). The measurements were repeated three times, and the error was within ±5%. The water uptake was calculated by the equation given below.

2.6 Swelling degree

The mass of as-prepared homogenous membranes was measured after they were fully dried in a vacuum oven at 40 °C. And then the dry membranes were immersed into feed solution at 76 °C for 48 h to achieve the sorption equilibrium. The mass of swollen membranes was obtained after carefully blotting the liquid droplets with filter paper. The swelling degree (SD, %) was calculated bywhere M_S and M_D were the mass of swollen and dry membrane, respectively.

2.7 Pervaporation experiments

The pervaporation performance of composite membranes was characterized by the P-28 membrane module (CM-Celfa AG Company, Switzerland) with an effective area (A) of 2.56 × 10⁻³ m² as reported in our previous work.⁹ The water/ethanol feed solution was pumped through membrane surface with a flow rate of 60 L h⁻¹ and the permeate pressure was maintained about 0.1 kPa by a vacuum pump. The pervaporation experiment was run at 76 °C for 2 h to obtain the steady state and subsequently the permeate vapor was collected by liquid nitrogen cold traps at intervals of 0.5 h (t). The permeate solution was weighed to calculate the permeate flux (J, kg m⁻² h⁻¹) and the mass fractions of water and ethanol were analyzed by HP4890 gas chromatography (GC) to obtain the separation factor (α). The permeation flux (J), separation factor (α) and separation index (PSI) were applied to evaluate the pervaporation performance.

PSI = J(α − 1) (7)

where M was the weight of permeate solution collected every 0.5 h, P and F represented the weight fractions of water (W) and ethanol (E) in permeate solution and feed solution, respectively. In addition, to analyze the effect of driving force on pervaporation performance, the permeance of individual components ((P/l)_i, GPU) (1 GPU = 7.501 × 10⁻¹² m³ (STP) per m² s Pa) and the selectivity (β) were obtained by the following equations:where J_i was the permeation flux of component i (g m⁻² h⁻¹), l was the membrane thickness (m), P_io and P_il were the partial pressures (kPa) of component i in feed side and permeate side, respectively. P_il was taken as zero due to the high vacuum degree in permeate side. γ_io and χ_io were the activity coefficient and mole fraction of component i in feed liquid, respectively. P^sat_io was the saturated vapor pressure (kPa) of pure component i. All experiments were performed using water/ethanol aqueous solution and repeated three times to guarantee the validity of experimental data.

3. Results and discussion

3.1 Characterization of AT nanorods

As shown in Fig. 1, AT exhibited the typical rod-like structure with a diameter of ∼20 nm and a length range of 300–1000 nm. The AT nanorod was identified to be the orthorhombic phase with a typical peak position at 8.4° by XRD (Fig. 2). The cell constants were calculated by the diffraction peaks, a = 12.987 Å, b = 17.992 Å and c = 5.203 Å, which were in good agreement with the previous work (a = 12.902 Å, b = 17.870 Å and c = 5.138 Å).²⁹

Fig. 1 TEM image of AT nanorods.

Fig. 2 Wide-angle X-ray diffraction curves of AT nanorods.

Due to the alternating arrangement of sandwich layered structures,^38,39 the channels with the free cross section of about 3.8 × 6.3 Ų⁶ running paralleling to the long axis of AT nanorods are constructed as shown in Fig. 3. Usually, water molecules with different states can retain in AT nanorods. Typically, one half of the water are closely coordinated to the Mg or Al ions as structural water, while the other half are loosely held in the open channels²⁶ as freezable water. In addition, there exist plentiful active –OH groups on the surface of AT nanorods on account of the tetrahedral and octahedral layered structures.⁴⁰ These –OH polar groups are effective to bond water molecules along the rods via hydrogen bonds.²⁹

Fig. 3 Structure of AT nanorod.

As can be seen from the FT-IR spectra of AT nanorods (Fig. 4), the peaks at about 3500 cm⁻¹ were attributed to the –OH stretching vibration of Si–OH and freezable water in AT structure. The peak at 1624 cm⁻¹ represented non-freezable water, and the peak at 1442 cm⁻¹ was assigned to the bending vibration of non-freezable water^40,41 and calcite which is a universal constitute of clays.²⁹ The characteristic peaks at 1033, 986 and 730 cm⁻¹ were attributed to the stretching vibrations of Si–OH, Al–OH and Mg–OH, respectively. The absorption peak at 882 cm⁻¹ was the Si–O–Si symmetric stretching vibration in AT structure.⁴² The thermal stability of AT nanorods was investigated by TGA as shown in Fig. 5. The thermal decomposition of AT nanorods included three stages. The first stage (40–100 °C, weight loss: ∼5%) involved the release of the water adsorbed on AT surface and in open channels. The second stage (100–230 °C, weight loss: ∼3%) was related to the residual water held in AT open channels. The third stage (230–690 °C, weight loss: ∼14%) was the loss of plentiful structural water (non-freezable water) in AT structure.²⁹ Due to the plentiful hydrophilic –OH groups and the unique sandwich structure, AT nanorods could retain water with different states on its external surface and in its open selective channels. And the high water uptake of AT nanorods (57%) proved their high ability of water absorption.

Fig. 4 FT-IR spectra of AT nanorods.

Fig. 5 TGA curves of AT nanorods.

3.2 Characterization of membrane

The surface morphologies of membranes were characterized by FESEM as shown in Fig. 6(a–d). Compared to the smooth surface of SA pristine membrane, rod-like structures were observed on the surface of hybrid membranes incorporated with AT nanorods. The cross-section morphologies of composite membranes were also measured by FESEM. As showed in Fig. 6(e–h), the active layers with thickness of ∼1100 nm were tightly coated on PAN support layers without any defect. To further observe the cross-section morphology in detail, the cross-section morphologies of corresponding homogenous membranes were characterized and the results were shown in Fig. 6(i–l). Compared to the pristine SA membrane, nanorods with the diameter of ∼30 nm which is larger than ∼20 nm of the pristine AT nanorods (as shown in Fig. 1), were randomly dispersed in the SA-AT(1%) and SA-AT(2%) membranes. The large diameter of nanorods in membranes was attributed to the tight coating of SA chains, and the thickness of this coating is ∼5 nm. This coating could improve the compatibility between AT nanorods and SA matrix. Agglomeration of AT nanorods with diameter of ∼120 nm was observed in the membrane with a high AT content (SA-AT(6%) membrane) as shown in the circle of Fig. 6(l).

Fig. 6 (a–d) Surface morphologies and (e–h) cross-section morphologies of SA/PAN and SA-AT(X)/PAN, and (i–l) cross-section morphologies of SA pristine and SA-AT(X) membranes.

As revealed by XRD (Fig. 7), the SA-AT(1%) membranes exhibited a typical diffraction peak at 2θ = 14.2° just as that of SA pristine membrane. No typical diffraction peaks of AT nanorod were observed. This revealed that AT nanorods were homogeneously dispersed in the polymer matrix. As the content of AT nanorods increased, the typical diffraction peak (2θ = 8.4°) of AT was observed and gradually enhanced, while the intensity of SA diffraction peak at 2θ = 14.2° was gradually attenuated, indicating that the crystallinity of SA matrix was reduced with the introduction of AT nanorods. As revealed by previous researches, the change of crystallinity could be attributed to the combination of hydrogen bond effect and agglomeration effect.^9,44 At low content of AT nanorods, the decline of crystallinity could be attributed to the hydrogen bonds between SA and the oxygen-containing groups (–OH) in AT structure, which interfered the ordered packing of SA chains and resulted in the increase content of amorphous regions in membranes. The agglomeration of AT nanorods reduced the interfacial interaction, but the bonding sites of hydrogen bonds were notably increased due to the high AT content. As a result, the crystallinity of hybrid membranes was further decreased.

Fig. 7 Wide-angle X-ray diffraction curves of SA pristine and SA-AT(X) membranes.

The thermal stability of homogeneous membranes was investigated by TGA. As presented in Fig. 8(a), the thermal decomposition trends of SA pristine and SA-AT(X) membranes were almost the same. Two distinct weight-loss stages were observed, the first stage for evaporation of water (40–200 °C) and the second stage for decomposition of –OH and –COOH groups in SA chains (200–300 °C). Obviously, all the membranes were thermally stable up to 200 °C, which were adequate for practical application in pervaporation operation.

Fig. 8 (a) TGA curves and (b) DSC curves of SA pristine and SA-AT(X) membranes.

The glass transition temperatures (T_g) of SA pristine and SA-AT(X) membranes were determined by DSC as shown in Fig. 8(b). The T_g of hybrid membranes was firstly increased and then decreased with the addition of AT nanorods. The increase of T_g was attributed to the limited movement of SA chains which was caused by the tight interaction between SA chains and AT nanorods. But with the further increasing content of AT nanorods, the agglomeration would weaken the interfacial interaction and the limited effect of hydrogen bonds, leading to the decrease of T_g.

3.3 Water state in hybrid membranes

Among various techniques for the study of water states in polymers, DSC is the most convenient and informative approach.³⁰ The freezable water could be detected by the DSC curves, and the non-freezable water could be calculated by the difference between total water and freezable water. As reported, freezable water could also be sub-classified into free water that undergoes similar thermal transitions to that of bulk water, and freezable bound water that possesses thermal phase transitions with a shift to low temperature compared to that of free water.^36,38,39 Usually, the phase transition peaks of free water and freezable bound water are merged together as a single peak on the DSC curves.

The water states in membranes were studied by DSC and the results were shown in Fig. 9(a). The transition peaks of SA-AT(1%) and SA-AT(2%) shifted to low temperature, which was attributed to the increasing content of freezable bound water. And that of SA-AT(6%) shifted to the high temperature, which was ascribed to the decreasing content of freezable bound water. With the incorporation of AT nanorods, the hydrogen bond in AT nanorods would bond water molecules along their surface, increasing the content of freezable bound water. Thus the transition peak of SA-AT(1%) shifted to lower temperature than that of SA pristine membrane, and the transition peak of SA-AT(2%) shifted to the lower temperature than that of SA-AT(1%). The agglomeration of AT nanorods appeared in SA-AT(6%) membrane led to reduced hydrogen bond effect between hydrogen bonds and water and then decreased content of freezable bound water. As a result, the transition peak of SA-AT(6%) shifted to higher temperature than that of SA-AT(2%). To further reveal the influence of AT nanorods on the states and ratio of water in hybrid membranes, the peak areas were integrated to obtain the weight fraction of freezable water and then the contents of freezable water and non-freezable water were calculated. As shown in Fig. 9(b), the content of non-freezable water and the ratio of non-freezable/freezable water were increased with the incorporation of nanorods. The introduced AT nanorods can interrupt a whole crystalline regions to be many small fragments, which increased the amount of interface regions between crystalline region and amorphous region. And the plentiful –OH groups of AT nanorods existing in hybrid membranes could also bond water as non-freezable water in membranes.³⁸ These all were beneficial to the increase of non-freezable water in hybrid membranes.^29,32,34

Fig. 9 (a) DSC thermograms of swelling membranes and (b) freezable water and non-freezable water content in SA pristine and SA-AT(X) membranes.

3.4 Hydrophilicity, water uptake and swelling degree

The hydrophilicity of membrane surface was investigated by static contact angle and the results were shown in Fig. 10(a). The tight SA coating on AT nanorods⁴³ resulted in the similar surface hydrophilicity of hybrid membranes with the SA pristine membrane.

Fig. 10 (a) Water contact angle, (b) water uptake and (c) swelling degree of SA pristine and SA-AT(X) membranes.

In the membranes for separation, adequate water uptake was essential to ensure the efficient transport of small molecules.^30,45 Water uptake of hybrid membranes was measured to investigate the water storage capability. As shown in Fig. 10(b), with the addition of only 2% of AT nanorods, the water uptake of SA-AT(2%) membrane was sharply increased to 59%, while that of SA pristine membrane was 50%. Only a slight increase of water uptake (1%) was obtained along the increasing AT content from 2 wt% to 8% wt%. The increase of water uptake could be inferred by the following three factors, the increase of interface regions between crystalline region and amorphous region in hybrid membranes, the excellent water absorption ability of the introduced AT nanorods, and the high content of non-freezable water in membranes. Due to the limited content of AT nanorods and the slight increase of water uptake at high AT content, the increase of water uptake was mainly attributed to the increase of amorphous region and non-freezable water in membranes.

The weight swelling degree of membrane was measured after being soaked in feed solution at 76 °C for 48 h. As shown in Fig. 10(c), the SA pristine membrane had a low swelling degree (2.3%) owing to the cross-linking effect of Ca²⁺. It was worth noting that the hybrid membranes showed only a slight increase of swelling degree by about 1%, which was favorable to practical application.

3.5 Pervaporation performance of membranes

3.5.1 Effect of AT content on pervaporation performance. The effects of AT content on pervaporation performance of SA-AT(X)/PAN membranes were investigated for a 90 wt% ethanol–water feed solution at 76 °C. As shown in Fig. 11, the permeation flux and separation factor of pristine SA/PAN membrane were 984 g m⁻² h⁻¹ and 574, respectively. The permeation flux was increased to ∼1350 g m⁻² h⁻¹ with the addition of 2 wt% of AT, then remained almost constant as the AT content further increased to 6 wt%, and finally slightly decreased to 1169 g m⁻² h⁻¹ when the AT content was up to 8 wt%. The separation factor was sharply increased to 2030 at the AT content of 2 wt%, i.e. 3.5 times higher than pristine SA/PAN membrane. Then the separation factor decreased with further increase of AT content and reached 530 at 6 wt% AT content.

Fig. 11 Effect of AT content on pervaporation performance of membranes.

As shown in Scheme 1, water could transport through the open channels of AT, through or along the hydration layers besides through the SA matrix. Due to the hydrophilicity of AT nanorods, AT would bond water molecules as the first layer of hydration layers and could be treated as a part of AT nanorods.⁴⁶ Freezable water would be attracted by the first layers to form hydrogen networks as the second layers, which would facilitate the diffusion of other water molecules. Meanwhile, with the incorporation of AT nanorods, water could be retained in hybrid membranes as non-freezable water by the plentiful hydrogen bonds between –OH on AT nanorods during the diffusion process of water. Additionally, AT nanorods decreased the crystallinity of SA, increased the interchain spacing of SA matrix and produced many interface regions (usually contain abundant hydrogen bonds) between amorphous region and crystalline region along AT nanorods. Non-freezable water could also retain in membranes by disrupting these hydrogen bonds, while freezable water would retain and transport in membrane until most of the hydrogen bonds were disrupted.³⁶ The non-freezable water in membranes could attract other water molecules to favor the solution process of water in hybrid membranes. The optimum permeation flux was only increased by ∼1.4 times compared to that of pristine SA/PAN membrane due to the tight interaction between water molecules and AT nanorods. And it was worth noting that the separation factor was increased sharply as the incorporation of AT nanorods, this was attributed to the enhanced solution process of water due to the increase content of non-freezable water and the selective channels in AT structure. As the AT content further increased, the formation of agglomeration (as shown in the circle of Fig. 6(l)) resulted in the decrease of flux and separation factor simultaneously.

Scheme 1 Schematic representation of mechanism for water permeation through the hybrid membrane.

3.5.2 Effect of feed temperature on pervaporation performance. To illustrate the effect of feed temperature on membrane pervaporation performance, the SA/PAN and SA-AT(2%)/PAN were investigated with 90 wt% ethanol aqueous solution at different temperatures. As shown in Fig. 12(a and b), both the total flux and the water flux of SA-AT(2%)/PAN were larger than those of SA/PAN, respectively. In addition, the increments of total flux and water flux for SA-AT(2%)/PAN were larger than those for SA/PAN, while the ethanol flux for SA-AT(2%)/PAN, as well as the increment of ethanol flux, was smaller than that for SA/PAN in the range of operation temperature as shown in Fig. 12(b). As a result, the separation factor of SA-AT(2%)/PAN was higher than that of SA/PAN. The continuous increase of flux with the increased temperature was attributed to the increase of driving force and water molecule diffusion rate.

Fig. 12 Effect of feed temperature on pervaporation performance of SA/PAN and SA-AT(2%)/PAN: (a) total flux and separation factor; (b) water flux and ethanol flux; (c) Arrhenius plots of water and ethanol flux; (d) water permeance, ethanol permeance and selectivity.

The effect of temperature on permeation flux could also be described by Arrhenius equation (eqn (10)) and the apparent activation energy could be figured out according to the equation:

where A_p, J_p, E_p, R and T represent the pre-exponential factor, permeation flux, apparent activation energy, gas constant and feed temperature, respectively. As shown in Fig. 12(c), the ln J_p versus 1000/T curves of SA/PAN and SA-AT(2%)/PAN for water and ethanol were similar to straight lines, respectively, implying that the permeability of both water and ethanol followed the Arrhenius law. As shown in Arrhenius plots, the calculated apparent activation energies of water (38.45 kJ mol⁻¹ and 40.48 kJ mol⁻¹ for SA/PAN and SA-AT(2%)/PAN, respectively) were higher than corresponding apparent activation energies of ethanol (12.87 kJ mol⁻¹ and 9.25 kJ mol⁻¹ for SA/PAN and SA-AT(2%)/PAN, respectively). Therefore, water permeation was more sensitive than ethanol permeation to the increase of temperature resulting in the increase of separation factor. And the difference between water apparent activation energy and ethanol apparent activation energy of SA-AT(2%)/PAN was larger than that of SA/PAN, leading to a high selectivity of SA-AT(2%)/PAN. Fig. 12(d) revealed the effect of feed temperature on permeance and selectivity with normalized driving force. The permeance of water remained almost a constant, while that of ethanol declined continuously, which was attributed to that the weakened solubility played a dominant role with the increasing temperature. Due to the larger difference between the permeance of water and ethanol for SA-AT(2%)/PAN than that for SA/PAN, the selectivity (β) of SA-AT(2%)/PAN was higher than that of SA/PAN.

3.5.3 Feed composition on pervaporation performance. Feed composition is another important parameter in pervaporation. The increased water concentration in feed could improve the partial pressure of water on the feed side, leading to the enhancement of driving force for water and the reduction of that for ethanol. To investigate the effect of feed composition on membrane pervaporation performance, the SA/PAN and SA-AT(2%)/PAN were characterized under the water content range from 5 wt% to 30 wt%. As shown in Fig. 13(a and b), the total flux and water flux for both SA/PAN and SA-AT(2%)/PAN were increased with the increasing amount of water in feed mixture, and the water flux of SA-AT(2%)/PAN was larger than that of SA/PAN when the water content in feed was less than 25 wt%. The ethanol flux of SA/PAN was first decreased and then increased while that of SA-AT(2%)/PAN was decreased when the water content in feed increased from 5 wt% to 30 wt%. The water content in permeate for the SA-AT(2%)/PAN membrane was larger than that of SA/PAN in all cases.

Fig. 13 Effect of feed composition on pervaporation performance of SA/PAN and SA-AT(2%)/PAN: (a) total flux and separation factor; (b) water flux and ethanol flux; (c) water permeance, ethanol permeance and selectivity.

The SA-AT(2%)/PAN membrane had higher total flux and water flux than those of the SA/PAN membrane at the water content less than 25 wt%, which was ascribed to the high non-freezable water content and the low crystallinity of SA-AT(2%)/PAN. But at high water content, the SA polymer chains would be swollen due to the plasticizing effect of freezable water and reduced the resistance to molecules permeation. Thus the ethanol flux of SA/PAN was increased when the water content was larger than 25%. Relatively, the SA-AT(2%)/PAN with AT nanorods containing high content of non-freezable water were less swelling than SA/PAN caused by water. Thus the SA-AT(2%)/PAN exhibited higher pervaporation performance than those of SA/PAN in all cases. To further investigate the effects of feed content on separation performance, the selectivity versus feed water content for SA/PAN and SA-AT(2%)/PAN was characterized and shown in Fig. 13(c). Although the trends of water/ethanol separation selectivity were almost the same, the selectivity for the SA-AT(2%)/PAN were higher than that for the SA/PAN at various feed compositions. This was attributed to the combined effects of the high non-freezable water content and the large water absorption amount by incorporating AT nanorods into SA matrix.

3.5.4 Operation stability. In practical application, the long-term operation stability of membrane is very important. Fig. 14 showed the long-term pervaporation performance of SA-AT(2%)/PAN up to 72 h for 90 wt% ethanol aqueous solution at 76 °C. The water content in permeate remained around 99.5 wt% with slight variations. And the permeation flux was declined due to the relaxation of SA chains and eventually stabilized at about 1250 g m⁻² h⁻¹. Thus, the membrane exhibited desirable operation stability.

Fig. 14 Long-term pervaporation performance of SA-AT(2)/PAN.

3.5.5 Comparison with literature data. The natural clay and SA have been widely applied in the pervaporation dehydration because of their hydrophilic nature and unique structure, respectively. The pervaporation performance of hybrid membranes incorporated with natural clays for organic dehydration and SA-based hybrid membranes for ethanol dehydration in literatures were list in Tables 1 and 2, respectively. These results demonstrated that the SA-AT(2%)/PAN showed high performance for ethanol dehydration in this research.

Table 1 Pervaporation dehydration performance of polymer/clay hybrid membranes in literatures

Membrane	Feed composition	Water content (wt%)	Permeation flux (g m^−2 h^−1)	Separation factor	Reference
SA-AT(2%)/PAN	Water/ethanol	10	1356	2030	This work
CS–Na^+-MMT	Water/isopropanol	10	142	14 992	47
CS–K^+-MMT	Water/acetone	5	1560	2200	25
SA–Na^+-MMT	Water/isopropanol	10	50	∞	16
PVA–Na^+MMT-5	Water/isopropanol	10	750	1116	48
PVA–Na^+MMT-10			510	2241
PVA–Na^+MMT-5	Water/1,4-dioxane	10	76	216
PVA–Na^+MMT-10			93	369
PVA–bentonite	Water/isopropanol	12.5	∼950	46	49

---

Table 2 Comparison of pervaporation performance of hybrid membranes with SA matrix for dehydration of ethanol/water mixture

Membrane	Water content (wt%)	Feed temperature (°C)	Permeation flux (g m^−2 h^−1)	Separation factor	PSI (10^6)	Reference
SA-AT(2%)/PAN	10	76	1356	2030	2.75	This work
SA–zeolite beta	10	30	132	1598	0.21	50
SA–zeolite beta	10	60	178	395	0.70
SA–Al–MCM-41	10	30	129	1089	0.14	17
SA–HPA	10	60	330	1000	0.33	18
SA–HPA	10	30	140	14 991	2.10
SA–mPTA	10	30	111	1866	0.21	51
SA–PVP–PWA	4	70	1400	1750	2.45	52
SA–zeolite 4A	10	70	163	279	0.45	53
SA–zeolite 4A	10	30	138	1334	0.18

---

4. Conclusion

In this study, natural attapulgite nanorods were introduced into alginate matrix to fabricate hybrid membranes for ethanol dehydration. The attapulgite nanorods in hybrid membranes decreased the crystallinity of adjacent alginate matrix and increased the content of non-freezable water, which could form hydration layers and be treated as part of AT surfaces, facilitating the transport of water molecules in hybrid membranes. Thus, the permeation flux of hybrid membrane was increased by about 1.4 times compared to that of pristine SA/PAN membrane. The open selective channels in AT structures improved the selectivity of water molecules and the increase contents of non-freezable water favored the solution of freezable water. As a result, the separation factor of hybrid membranes was remarkably increased up to 2030 when the AT content was 2 wt%, about 3.5 times higher than pristine SA/PAN membrane. This study revealed that the selective channels in AT and the increase content of non-freezable water in membranes could notably affect the performance of membranes for ethanol dehydration.

Nomenclature

Symbols

J	Permeate flux (g m^−2 h^−1)
α	Separation factor
PSI	Separation index
A	Effective area (m^2)
M	Weight of permeate solution (g)
β	Selectivity
γ	Activity coefficient
P/l	Permeance (GPU)
l	Membrane thickness (m)
P^satio	Saturated vapor pressure (kPa)
Ap	Pre-exponential factor
Ep	Apparent activation energy
R	Gas constant
T	Feed temperature (K)
F	Mass fraction in feed solution (wt%)
X	AT content in hybrid membrane
SD	Swelling degree (%)
AT	Attapulgite
SA	Sodium alginate

Subscripts

W	Water
E	Ethanol

Acknowledgements

The authors gratefully thank the financial support from the National Science Fund for Distinguished Young Scholars (21125627), the National Natural Science Foundation of China (21576189, 21490583, 21306131), the Program of Introducing Talents of Discipline to Universities (B06006) and the Open Project of the State Key Laboratory of Chemical Engineering (SKL-ChE-14B03).

Notes and references

1. D. P. Suhas, A. V. Raghu, H. M. Jeong and T. M. Aminabhavi, RSC Adv., 2013, 3, 17120–17130.
2. L. Chai, H. Li, X. Zheng, J. Wang, J. Yang, J. Lu, D. Yin and Y. Zhang, J. Membr. Sci., 2015, 491, 168–175.
3. N. Nordin, A. F. Ismail, A. Mustafa, R. S. Murali and T. Matsuura, RSC Adv., 2014, 4, 52530–52541.
4. D. Hua and T.-S. Chung, J. Membr. Sci., 2015, 492, 197–208.
5. Y. L. Liu, Y. H. Su, K. R. Lee and J. Y. Lai, J. Membr. Sci., 2005, 251, 233–238.
6. H. Sanaeepur, A. Kargari and B. Nasernejad, RSC Adv., 2014, 4, 63966–63976.
7. D. L. Gin and R. D. Noble, Science, 2011, 332, 674–676.
8. Y. Li, G. He, S. Wang, S. Yu, F. Pan, H. Wu and Z. Jiang, J. Mater. Chem. A, 2013, 1, 10058–10077.
9. K. Cao, Z. Jiang, J. Zhao, C. Zhao, C. Gao, F. Pan, B. Wang, X. Cao and J. Yang, J. Membr. Sci., 2014, 469, 272–283.
10. W.-G. Kim and S. Nair, Chem. Eng. Sci., 2013, 104, 908–924.
11. R. W. Baker and B. T. Low, Macromolecules, 2014, 47, 6999–7013.
12. N. C. Su, Z. P. Smith, B. D. Freeman and J. J. Urban, Chem. Mater., 2015, 27, 2421–2429.
13. T. C. Merkel, B. D. Freeman, R. J. Spontak, Z. He, I. Pinnau, P. Meakin and A. J. Hill, Science, 2002, 296, 519–522.
14. J. Ahn, W.-J. Chung, I. Pinnau, J. Song, N. Du, G. P. Robertson and M. D. Guiver, J. Membr. Sci., 2010, 346, 280–287.
15. J. Ploegmakers, S. Japip and K. Nijmeijer, J. Membr. Sci., 2013, 428, 331–340.
16. S. D. Bhat and T. M. Aminabhavi, Sep. Purif. Technol., 2006, 51, 85–94.
17. M. B. Patil, R. S. Veerapur, S. A. Patil, C. D. Madhusoodana and T. M. Aminabhavi, Sep. Purif. Technol., 2007, 54, 34–43.
18. V. T. Magalad, A. R. Supale, S. P. Maradur, G. S. Gokavi and T. M. Aminabhavi, Chem. Eng. J., 2010, 159, 75–83.
19. S. G. Adoor, V. Rajineekanth, M. N. Nadagouda, K. Chowdoji Rao, D. D. Dionysiou and T. M. Aminabhavi, Sep. Purif. Technol., 2013, 113, 64–74.
20. M. Sabard, F. Gouanvé, E. Espuche, R. Fulchiron, G. Seytre, L.-A. Fillot and L. Trouillet-Fonti, J. Membr. Sci., 2014, 450, 487–498.
21. B. V. K. Naidu, S. D. Bhat, M. Sairam, A. C. Wali, D. P. Sawant, S. B. Halligudi, N. N. Mallikarjuna and T. M. Aminabhavi, J. Appl. Polym. Sci., 2005, 96, 1968–1978.
22. S. Sinha Ray and M. Okamoto, Prog. Polym. Sci., 2003, 28, 1539–1641.
23. B. L. Eric Ferrage, L. J. Michot and J.-L. Robert, J. Phys. Chem. C, 2010, 114, 4515–4526.
24. I. B. J. M. Cmes, G. Besson, M. Fran ois, J. P. Uriot, F. Thomas and J. E. Poirier, Langmuir, 1992, 8, 2730–2739.
25. C. Gao, M. Zhang, Z. Jiang, J. Liao, X. Xie, T. Huang, J. Zhao, J. Bai and F. Pan, Chem. Eng. Sci., 2015, 135, 461–471.
26. W. L. Haden and I. A. Schwint, Ind. Eng. Chem., 1967, 59, 58–69.
27. C. D. Hatch, J. S. Wiese, C. C. Crane, K. J. Harris, H. G. Kloss and J. Baltrusaitis, Langmuir, 2012, 28, 1790–1803.
28. D. Y. Kang, C. W. Jones and S. Nair, J. Membr. Sci., 2011, 381, 50–63.
29. R. Giustetto and R. Compagnoni, Clay Miner., 2011, 46, 371–385.
30. Y. Hirata, Y. Miura and T. Nakagawa, J. Membr. Sci., 1999, 163, 357–366.
31. W. Li, Y. Zheng and R. Cheng, Polymer, 2008, 49, 4740–4744.
32. T. Nakaoki and H. Yamashita, J. Mol. Struct., 2008, 875, 282–287.
33. Z. H. Ping, Q. T. Nguyen, S. M. Chen, J. Q. Zhou and Y. D. Ding, Polymer, 2001, 42, 8461–8467.
34. J. Ostrowska-Czubenko and M. Gierszewska-Drużyńska, Carbohydr. Polym., 2009, 77, 590–598.
35. Y. S. Kim, L. M. Dong, M. A. Hickner, T. E. Glass, V. Webb and J. E. McGrath, Macromolecules, 2003, 36, 6281–6285.
36. R. M. Hodge, G. H. Edward and G. P. Simon, Polymer, 1996, 37, 1371–1376.
37. W. Z. Zhang, M. Satoh and J. Komiyama, J. Membr. Sci., 1989, 42, 303–314.
38. S. Amnuaypanich, J. Patthana and P. Phinyocheep, Chem. Eng. Sci., 2009, 64, 4908–4918.
39. W. G. Liu and K. De Yao, Polymer, 2001, 42, 3943–3947.
40. R. M. Barrer and N. Mackenzie, J. Phys. Chem., 1954, 58, 560–580.
41. R. M. Barrer, N. Mackenzie and D. M. Macleod, J. Phys. Chem., 1954, 568–572.
42. J. H. Xu, J. Tao, Y. Gan, C. S. Peng and Z. Li, Mater. Res. Innovations, 2014, 18, S2-377–S2-381.
43. Y. Wang, X. Shi, W. Wang and A. Wang, J. Appl. Polym. Sci., 2013, 129, 1080–1088.
44. J. Li, L. Shao, L. Yuan and Y. Wang, Mater. Des., 2014, 54, 520–525.
45. G. He, Y. Li, Z. Li, L. Nie, H. Wu, X. Yang, Y. Zhao and Z. Jiang, J. Power Sources, 2014, 248, 951–961.
46. M.-J. Wei, J. Zhou, X. Lu, Y. Zhu, W. Liu, L. Lu and L. Zhang, Fluid Phase Equilib., 2011, 302, 316–320.
47. S. K. Choudhari and M. Y. Kariduraganavar, J. Colloid Interface Sci., 2009, 338, 111–120.
48. S. G. Adoor, M. Sairam, L. S. Manjeshwar, K. V. S. N. Raju and T. M. Aminabhavi, J. Membr. Sci., 2006, 285, 182–195.
49. T. Jose, S. C. George, M. G. Maya, H. J. Maria, R. Wilson and S. Thomas, Ind. Eng. Chem. Res., 2014, 53, 16820–16831.
50. S. G. Adoor, L. S. Manjeshwar, S. D. Bhat and T. M. Aminabhavi, J. Membr. Sci., 2008, 318, 233–246.
51. S. G. Adoor, V. Rajineekanth, M. N. Nadagouda, K. C. Rao, D. D. Dionysiou and T. M. Aminabhavi, Sep. Purif. Technol., 2013, 113, 64–74.
52. V. T. Magalad, G. S. Gokavi, C. Ranganathaiah, M. H. Burshe, C. Han, D. D. Dionysiou, M. N. Nadagouda and T. M. Aminabhavi, J. Membr. Sci., 2013, 430, 321–329.
53. S. D. Bhat and T. M. Aminabhavi, J. Appl. Polym. Sci., 2009, 113, 157–168.

---