15

Journal of Membrane Science 238 (2004) 103–115

Pervaporation of hydrazine hydrate: separation characteristics of membranes with hydrophilic to hydrophobic behaviour S.V. Satyanarayana1 , P.K. Bhattacharya∗ Department of Chemical Engineering, Indian Institute of Technology Kanpur, Kanpur 208 016, India Received 22 May 2003; received in revised form 1 March 2004; accepted 1 March 2004

Abstract Anhydrous hydrazine, used as rocket propellant, is an important inorganic compound and is obtained by dehydrating hydrazine hydrate. However, hydrazine forms an azeotrope with water which makes conventional separation processes energy intensive. Pervaporation may act as an alternative process; however, because of the highly alkaline nature of hydrazine, proper selection of polymer plays a vital role. Several polymeric membranes (PERVAP® 2200, 2201 and 2202, ethyl cellulose (EC), acrylonitrile butadiene styrene (ABS), polystyrene, and modified EC) were chosen to study separation characteristics. Reacting EC with phenyl isocyanate and varying its amount, modifications were carried out in order to alter the hydrophilic characteristics of the membrane to hydrophobic. 1 H NMR spectra estimated degree of substitution in terms of carbamate groups. Contact angle measurements were taken to observe hydrophilic/hydrophobic characteristics. The higher contact angles of water with modified EC membranes, compared to unmodified form, indicate increase of hydrophobicity of the membranes. Further, FT-IR, XRD and positron annihilation techniques were employed to observe characteristics of one such modified ethyl cellulose membrane. Sorption studies were also carried out and sorption of both water and hydrazine hydrate in modified EC membrane were observed to be lower compared to unmodified. Pervaporation studies with all chosen polymers revealed selective diffusion of water playing major role compared to sorption. Further, apolar materials exhibited higher separation factors than polar materials. Encouraging results, in terms of higher PSI, were obtained with modified EC (262.4) and ABS (395.9) membranes. © 2004 Elsevier B.V. All rights reserved. Keywords: Pervaporation; Hydrazine hydrate; Modified hydrophobic ethyl cellulose; Membrane characterization

1. Introduction Hydrazine is an important inorganic chemical with high heats of combustion and hence becomes highly useful as rocket fuel. The other main applications of hydrazine include de-oxygenation of boiler feed water, fuel cells and production of blowing agents. However, production of hydrazine Abbreviations: ABS, acrylonitrile butadiene styrene; EC, ethyl cellulose; FT-IR, Fourier transform infra-red; MEC, modified ethyl cellulose; NMR, nuclear magnetic resonance; PAL, positron annihilation lifetime; PS, polystyrene; PSI, pervaporation separation index; XRD, X-ray diffraction ∗ Corresponding author. Tel.: +91-512-2597093; fax: +91-512-2590104. E-mail address: [email protected] (P.K. Bhattacharya). 1 Present address: Department of Chemical Engineering, J.N. Technological University, Ananthapur 515 002, Andhra Pradesh. 0376-7388/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2004.03.025

which is carried out using any of the known various reaction routes still pose problems with regard to its yield and purity [1]. Ordinary distillation provides hydrazine in the form of hydrate (64 wt.% of hydrazine). This form of hydrazine finds wide applications; whereas, use of hydrazine in rocket propulsion requires hydrazine to be in the anhydrous form. Therefore, the removal of water from hydrate state to produce anhydrous hydrazine is essential to make it suitable for such purposes. However, conventional separation techniques for the removal of water experience difficulty as hydrazine forms an azeotrope with water at 71.5 wt.% of hydrazine [2]. Further, hydrazine and water are highly polar by nature and between them there is strong hydrogen bonding. Therefore, conventionally combinations of processes are required to seek dehydration. These are chemical reactions, followed by distillation or azeotropic distillation with aniline as entrainer. Still, a specific problem remains which is the explo-

104

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

sive nature of hydrazine vapours during distillation, apart from other limitations like high energy consumption and cost. Hence, there is a need to search alternative technologies to produce anhydrous hydrazine. Application of membrane technology is one such attempt to address the stated problems [3–5]. Pervaporation process, in particular, is considered because of its potential to separate azeotropic mixtures. Pervaporation is relatively a new membrane separation process that has elements common to reverse osmosis and gas separation through polymeric membranes. In pervaporation, the feed solution is contacted to one side of the membrane and permeate (in vapour form) is obtained on the other side by keeping downstream pressure lower than the saturation vapour pressure of the permeated components. Chemical potential gradient across the membrane is the driving force for mass transfer. It is used for separation of liquid mixtures (particularly azeotropic mixtures [6], isomers [7] and organic–organic solvents [8]). Removal of trace quantities of organics from wastewater is also being considered to be a potential area of application [9,10]. Pervaporation has many advantages over distillation. The process may be operated at ambient temperature with simpler equipment requiring low energy and low capital investment [11]. The ambient temperature operation of pervaporation is of significance; particularly for hydrazine–water system, as during distillation

(around 64%). Further, it was reported that ethyl cellulose is selective for hydrazine during sorption and selective for water during diffusion [12]. Therefore, such a problem may be addressed by selecting a membrane with high water-permselectivity. This may be achieved by increasing the sorption selectivity (water to hydrazine) or diffusion selectivity. For a better sorption ratio, hydrophilic moiety may be introduced into the polymer chain to enhance water sorption but hydrazine sorption may also increase because of high polarity of the compound. In addition to this, excessive swelling may decrease diffusion selectivity as free volume increases and thus overall selectivity decreases [13]. Various hydrophilic chitosan and modified chitosan membranes were previously used [14] and were found to have low selectivities. Again for better diffusion ratio, the hydrophobic moiety may be introduced into the polymer chain. This may, however, decrease permeation rate. Thus, with the introduction of balanced quantities of hydrophilic and hydrophobic moieties, better flux and better selectivity may be obtained. In this regard, the conversion of remaining hydroxyl groups (depending on the degree of deacetylation of ethyl cellulose) to carbamate group may increase the hydrophobicity of the membrane. Such a conversion may be carried out by reacting ethyl cellulose with phenyl isocyanate [15], according to the following reaction:

at high temperature explosive decomposition of hydrazine may occur. The selection of polymer for hydrazine–water separation plays an important role because of the highly alkaline nature of hydrazine (pH > 12.5) and few polymers are available to withstand such solutions. Further, hydrazine has strong reducing and hydrolysing effects. Ravindra et al. [5] have used ethyl cellulose membrane to carry out this separation. Even though, ethyl cellulose was found to be highly selective for water (around 240) at low concentrations of hydrazine (around 10%), selectivity was observed to be poor (around 2) at higher concentrations of hydrazine

The present work is a part of a comprehensive work for dehydration of hydrazine hydrate. Therefore, the objective of the present work is to study and observe separation characteristics of different varieties of hydrophilic and hydrophobic membranes by pervaporation at fixed operating conditions (temperature: 50 ◦ C, downstream pressure: 0.1 mmHg, static feed condition). Commercial (PERVAP® ) as well as laboratory prepared membranes (polystyrene, acrylonitrile butadiene styrene and modified ethyl cellulose) were utilised for the purpose. In order to observe the effect of hydrophobicity over hydrophilicity on separation characteristics, modi-

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

105

fied membranes were prepared by reacting ethyl cellulose polymer with different amounts of phenyl isocyanate.

the polymer from hydrophilic to hydrophobic. It was, therefore, thought appropriate only to employ FT-IR spectrum analysis of filtrate to check the formation of isocyanate.

2. Experimental

2.3. Preparation (casting) of membrane

2.1. Materials

Ethyl cellulose (EC) polymer (10 g) was dissolved in toluene (90 g). The solution was centrifuged (REMI model-R 24, India) at 10,000 rpm for 15 min for the removal of undissolved polymer. The supernatant homogeneous solution was transferred to a conical flask (air tight) and kept for overnight for the removal of entrapped air bubbles. The casting of membrane was carried out on a glass plate using a modified thin film applicator (ACME, India). After around 24 h of solvent evaporation, the membrane was placed in a vacuum oven for another 4 h for the removal of residual traces of solvent. The procedure was repeated for casting acrylonitrile styrene butadiene (ABS) and polystyrene (PS) based membranes (in toluene solvent with 16 wt.% polymer concentration). For modified ethyl cellulose membranes, varied quantities of phenyl isocyanate were added to ethyl cellulose polymer solution (after centrifuging) and casting of the membrane was carried out after 24 h of the reaction. The entire operation was done at room temperature (around 25 ◦ C). Dense modified EC membrane was thus obtained after the evaporation of solvent and unreacted phenyl isocynate. In all, four varieties of modified ethyl cellulose (MEC) membranes were prepared and are denoted with respect to the amounts of isocyanate used (Table 1).

Analytical grade toluene (Ranbaxy, India), benzene (Ranbaxy, India), benzoyl chloride (S.D. Fine, India), hydrazine hydrate (Qualigens, India), sodium azide (Hi media, India), ethyl cellulose (ethoxy content 48–49.5%: Loba-Chemie, India) and acrylonitrile styrene butadiene (Strasis & Co., USA) were used for experimentation. Double distilled water was used for preparation of solution and for the purpose of analysis. The commercial composite membranes (PERVAP® 2200, 2201, 2202) were obtained from Sulzer Chemtech, Germany. Such membranes consist of a porous support layer (70–100 ␮m) on top of a polymer fleece (non-woven fabric of thickness 100 ␮m). Further, on top of the porous support is a very thin (0.5–2 ␮m) polymeric separating layer. Being a proprietary item, the composition of thin top layer could not be obtained; however, as per Jonquières et al. [16], the material is a cross-linked polyvinyl alcohol based membranes. 2.2. Preparation of phenyl isocyanate Phenyl isocyanate is very sensitive towards water and reacts very easily to give the corresponding carbamic acid. Further, carbamic acid is so unstable that it dissociates into amine and carbon dioxide. Therefore, it was decided to use freshly prepared phenyl isocynate before an experiment. Further, precautions were taken to avoid the contact of phenyl isocyanate with water. Benzoyl chloride and sodium azide, in presence of organic medium (benzene), may produce phenyl isocyanate, according to following reaction [17,18]:

Around 60 ml of benzene (dehydrated by keeping in calcium chloride) and 35 g of sodium azide were taken in a 250 ml stoppered conical flask. Further, around 70 ml of benzoyl chloride was taken in a Soxhlet apparatus with a Teflon stopcock. The Soxhlet apparatus was fixed on a conical flask and benzoyl chloride was added dropwise. Flow rate of benzoyl chloride was adjusted in order to add entire benzoyl chloride with in 8 h. The reaction mixture was continuously stirred using magnetic shaker. The flask was kept in chilled ethanol. Thus, reaction temperature was maintained at 3–5 ◦ C by recirculating (Julabo bath, Germany) chilled ethanol. On completion of reaction, the whole mixture was vacuum filtered. The objective behind preparing isocynate is to modify EC polymer and thus alter the characteristics of

2.4. Characterisation of modified EC 2.4.1. FT-IR The FT-IR spectra were obtained for phenyl isocyanate, EC as well as for MEC. Vector 22 FT-IR Spectrometer (Brooker, Germany) was used for the purpose.

2.4.2. NMR The amounts of phenyl isocynate were varied to observe modifications during reaction with EC. Accordingly, the proton NMR (1 H NMR; make—JOEL JNM-LA400, Japan) spectra were recorded for phenyl isocyanate as well as for EC and MEC polymers. Solvent CDCl3 was used for the purpose. TMS (tetra methyl silane) was used as an internal standard for reference to 1 H NMR. Chemical shifts and coupling constants were recorded. 2.4.3. XRD The molecular packing of the EC and MEC membranes were investigated with wide angle X-ray diffraction instru-

106

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

Table 1 Amounts of carbamate groups corresponding to phenyl isocyanate Amounts of phenyl isocyanate (ml in 100 g of polymer solution) Number of carbamate groups per 100 units of ethyl cellulose Notation for modified membrane obtained

0.0 0.0 MEC0

ment (WAXRD). WAXRD curves of the membranes were obtained using an Iso-Debyeflex X-ray powder diffractometer having monochromatic radiation of ␣-rays emitted by Cu at a wave length of 1.54 Å. Scanning conditions: angle (7–50◦ at a rate of 3◦ /min); accelerating voltage (30 kV) and tube current (20 mA). 2.4.4. Positron annihilation lifetime (PAL) spectroscopy The PAL measurements were carried out using a fast–fast system having a resolution of 300 ps (FWHM for the 60 Co prompt ␥-rays, under 22 Na window settings). The positron source was prepared by depositing around 2 ␮Ci aqueous 22 NaCl on a thin aluminium foil (thickness ∼ 12 ␮m) which was covered with an identical foil. Approximately, one million counts were collected in each spectrum and four spectra were measured for each sample. The lifetime data were analysed using PATFIT-88 programs [19]. Source correction was done for all the spectra. The following expression was used to relate o-Ps pick-off lifetime (τ 3 ) and special free volume radius (r) [20]
 −1 1 r 1 2πr (1) 1− + sin τ3 = 2 r + r 2π r + r where r is the electron layer thickness. Further, the fractional free volume f may be estimated from the following empirical relation [20]: f = bVF I3

(2)

where VF is the free volume of the membrane and I3 the intensity corresponding to τ 3 . The scaling factor b is obtained from variation of free volume with temperature. However, in the absence of such data, it may be typically assigned a value of 1.0 nm−3 [21]. Estimation errors are mentioned in Table 2. 2.5. Contact angle measurements Equilibrium contact angles of water and hydrazine hydrate with membranes were measured in saturated environment through sessile drop method using Goniometer (Rame-Hart Inc., Imaging System, USA). Flat sheets were mounted using stainless steel holder and placed in a chamber. A glass

1.0 3.0 MEC1

2.0 11.0 MEC2

3.2 14.3 MEC3

4.0 24.2 MEC4

5.0 – Loss of strength

syringe with a stainless steel needle was used to put the liquid drop on membrane. The angles were measured with RHI software by capturing the image with video camera. Around 5 min stabilisation time was allowed to capture the image. Around 50 readings were then recorded in a time span of 1 s. Procedure was repeated to estimate an average value (within ±3◦ error) of contact angle. 2.6. Sorption measurements Pre-weighed dry membranes were taken in a conical flask containing water or hydrazine hydrate for sorption purpose. The flask was kept on shaker bath (model SW-23, Julabo, Germany) under 200 rpm for 6–7 days at 50 ◦ C. The membranes in conical flasks were taken out at regular intervals and were wiped with tissue paper for the removal of adhered liquid. The wet weight of the membrane was measured. The procedure was repeated until consecutive readings of weight of wet membranes were found equal. The difference of weights was presented with respect to dry weight of membranes as percentage of sorption. 2.7. Analysis Hydrazine is a hygroscopic substance which is prone to air oxidation. Further, under exposure to atmosphere it absorbs carbon dioxide and, therefore, analysis of hydrazine sample requires proper precautions. The tedious procedure involved with Penneman method [22] of employing potassium iodate as titrant is prone to errors. Therefore, gas chromatograph method [23] was employed for analysis of hydrazine. Both the feed and the permeate samples concentrations were measured. Gas chromatograph (Nucon, India equipped with TCD) was employed with chromosorb-103 as column for the purpose. The column is made of stainless steel of dimensions: length 6 ft, outside diameter 1/8 in. and inside diameter of 2 mm. The injector, detector and oven temperatures were set at 210, 220 and 170 ◦ C, respectively. Helium was used as a carrier gas at 20 ml/min. Observed retention time for water was 1.94 and that for hydrazine was 2.46 min for the above stated conditions which were within ±5% error. 2.8. Pervaporation: set-up and procedure

Table 2 Free volume parameters of EC and MEC4 membrane Membrane

τ 3 (ns)

I3 (%)

r (nm)

Vf

EC MEC4

2.63 ± 0.01 2.26 ± 0.01

26.0 ± 0.2 17.5 ± 0.2

0.337 0.308

0.160 0.121

(nm3 )

f 0.042 0.021

A set-up was designed and developed for the pervaporation experimental investigations and a schematic version is shown in Fig. 1. Pervaporation test cell, made of glass, was having specially designed flanges to lodge the membrane with an effective membrane area of 50.6 cm2 . The mem-

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

107

Fig. 1. Schematic diagram of experimental set-up for pervaporation.

brane was kept on highly porous stainless steel support with the shiny polymeric layer (for commercial membranes) facing the feed solution. Initially fixed volume of feed solution (hydrazine hydrate of 64 wt.% of hydrazine) was taken on the upper side (feed side) of the cell. Both upstream and downstream sides of the cell were heated (in order to maintain isothermal conditions around the cell [24]) and hence cell surface was covered by heating mantels. The temperatures of both the sides were controlled at 50 ◦ C, using PID controller device (Fuji, Japan). The membrane upstream side was kept at atmospheric pressure and the downstream side was maintained under vacuum, using a vacuum pump (Vacuum Techniques, Bangalore). The condenser system consisted of two traps that can be used alternately, allowing the permeated pervaporate stream to be sampled continuously without interruption of the operation. The permeated vapours are condensed in the trap, which is kept in Dewar flask, filled with liquid nitrogen. The frozen permeate was collected within a specified time interval. The cold traps were brought to room temperature for measuring its weight, using a five decimal balance, to determine mass flux. Permeate was analysed to determine its hydrazine content. Flux values and hydrazine concentration in permeate were recorded as a function of time. To minimise in measurement errors, an average of two separate consecutive readings were taken after the system reached to steady state. The average error in the total permeation mass flux was estimated to be within ±3% and that for hydrazine concentration was within ±5%. The ability of a membrane to separate a given liquid mixture is defined by separation factor αPWH (water to hydrazine): αPWH =

y/(1 − y) x/(1 − x)

(3)

where y and x are the mass fractions of water in permeate and feed, respectively. Further, P refers to pervaporation. Mass flux data were normalised corresponding to 5 ␮m thickness

to compare the obtained flux data between commercial and laboratory made membranes. This was done using inverse relationship of flux and thickness for fixed operating conditions. Further, to compare the performance of different membrane materials the PSI (product of permeation mass flux and separation factor) value [25] of a membrane was utilised.

3. Results and discussion 3.1. Modified EC membrane: characterisation Ethyl cellulose membrane was reacted with different proportions of phenyl isocyanate to form modified ethyl cellulose membrane (MEC). The reaction has been described in the earlier sections. 1 H NMR spectra estimated degree of substitution in terms of carbamate groups. In the following section, such a modified membrane, however, using a fixed quantity of phenyl isocyanate (4 ml per 100 g of polymer solution), was characterised in order to observe the modifications in comparison to unmodified form of ethyl cellulose. FT-IR, XRD and positron annihilations techniques were also employed for the purpose and results are discussed. However, contact angles were measured for all the membranes used for the present work. 3.1.1. FT-IR FT-IR spectrum of phenyl isocyanate showed a peak at 2135 cm−1 which corresponds to isocyanate [26], confirming the formation of the reaction (Section 2.2). Fig. 2 depicts the FT-IR spectrum of EC and MEC4. Ravindra et al. [27] have used FT-IR spectrum for EC and have explained various functional groups. Comparison between EC and MEC spectrums (Fig. 2) show a decrease in relative intensity of –OH peak (at 3476 cm−1 ) from 0.742 (EC) to 0.585 (MEC4). Further, an extra peak was found in the case of

108

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

Fig. 2. FT-IR spectrums of EC and MEC4 membranes.

MEC4 at 1722.68 cm−1 . This peak corresponds to carbamate group [26], confirming the conversion of hydroxyl groups (in the closed ring structure of EC) into carbamate groups. 3.1.2. NMR Modifications on EC polymers by reacting with phenyl isocyanate were carried out, as mentioned earlier, primar-

ily to alter the characteristics of the said polymer from hydrophilic to hydrophobic. It was, therefore, necessary to estimate the extent of reaction with respect to change in the amounts of isocyanate. Since, the conversion was in terms of carbamate groups, it was thought to estimate the degree of substitution (or amount of carbamate groups) through 1 H NMR. Chemical shifts were recorded. A typical 1 H NMR

Fig. 3. A typical representation of 1 H NMR spectra for EC and MEC4.

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

109

Fig. 4. XRD spectrums of EC and MEC4 membranes.

spectra, obtained for EC and MEC4, is given in Fig. 3 for the purpose. The comparison between the spectra obtained for EC and MEC (Fig. 3) shows multiplet in the region of 7.0551–8.0576 for MEC. This indicates the presence of aromatic ring [28], consisting of five protons. The chemical shift at 1.1569 corresponds to aliphatic –CH3 group of –OC2 H5 (2 × 3 = 6 protons) group. Therefore, if each EC contains one carbamate group, there may be six aliphatic protons equivalent to five aromatic protons. For MEC membranes (all the modified forms, i.e. from MEC1 to MEC4), intensities, corresponding to –CH3 as well as aromatics groups, are noted. Accordingly, calculations provide the number of carbamates per 100 units of ethyl cellulose. Such values with respect to the amounts of phenyl isocyanate (added during reaction), are presented in Table 1. Table depicts, with increase in the amounts of phenyl isocyanate more and more hydroxyl groups, in EC, are converted to carbamate groups. 3.1.3. XRD Fig. 4 shows the WAXRD spectra of EC and MEC4. Both spectra display three sharp peaks at 14◦ , 17◦ and 26◦ . The sharp peaks suggest the semi-crystalline nature of the polymer. There is not much difference in the obtained spectra for EC and MEC membranes. However, one can observe a slight umbrella type shape at lower angles, in addition to intensity changes (corresponding to above angles), for MEC4 membrane. Qualitatively, this change suggests that the MEC membrane is slightly amorphous compared to EC membrane. This may, in a sense increase the flux. Studies on the details of the structure of the polymer through XRD spectrum analysis are not the objective of the present work. 3.1.4. Positron annihilation lifetime spectroscopy A detailed work, communicated separately [29] on positron lifetime spectroscopic analysis was made for vari-

eties of dense and composite membranes in order to obtain primarily free volume parameters. In this paper, results pertaining to EC and MEC membranes are reutilised and presented in Table 2. Ortho-positron lifetime (τ 3 ) and intensity (I3 ) were observed to be smaller for MEC4 compared to EC. Further, Table 2 shows the calculated values of free volume fraction which was also found to be smaller for MEC4 as compared to EC. The obtained free volume parameters were observed to be similar to those for semi crystalline polymers [30]. Therefore, during modification of EC dense film, free volume becomes smaller (decreasing diffusivities of species within the modified membrane), causing reduction in flux. 3.1.5. Contact angle: analysis Contact angles of water and hydrazine hydrate were measured on different membranes and the results are tabulated in Table 3. The contact angles for both hydrazine hydrate and water on PERVAP® and EC were found to be lower than that for MEC4, PS and ABS membranes. This suggests that EC and PERVAP® membranes are hydrophilic by nature; whereas, other membranes are hydrophobic by nature (apolar nature of polymers). Further, expectedly, EC membrane on reaction with phenyl isocyanate (MEC4 membrane) Table 3 Contact angles of water and hydrazine hydrate Membrane

Hydrazine hydrate contact angle (◦ )

Water contact angle (◦ )

Ratio of contact angles of water to hydrazine hydrate

PERVAP® 2200 PERVAP® 2201 PERVAP® 2202 EC MEC4 PS ABS

30.7 23.4 30.0 32.8 66.0 73.7 69.5

65.2 30.5 51.3 61.6 95.5 78.9 78.3

2.12 1.30 1.71 1.87 1.45 1.07 1.13

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115 100

1.9

90

Contact angle

1.8

hydrazine hydrate water Ratio of contact angles

80

70

1.7

60 1.6

50

40 1.5 30

20

Ratio of contact angles (water to hydrazine hydrate)

110

1.4 0

5

10

15

20

25

Number of carbamate groups / 100 units of ethyl cellulose Fig. 5. Influence of amounts of carbamate groups on contact angles water and hydrazine hydrate.

changes its characteristics from hydrophilic to hydrophobic, as evidenced from increase of contact angle. This is due to the introduction of phenyl group. Amounts of phenyl isocyanate (during reaction with EC membrane) were varied to observe the increase in hydrophobicity on MEC membrane through the measurements of contact angles. Accordingly, relationships were observed in terms of contact angles of water, hydrazine hydrate as well as ratio of contact angles of water to hydrazine hydrate against amounts of carbamate groups. Such relationships are shown in Fig. 5. It may be observed from Fig. 5, as the amount of carbamate groups increases; contact angle of both hydrazine hydrate and water increases too, whereas the ratio of angles decreases. Further, low values of contact angles of hydrazine hydrate, compared to water for all membranes, suggest that hydrazine is strongly held by membrane compared to water. FT-IR studies with EC membrane (soaked in hydrazine or water) also led to such conclusions [27]. 3.2. Sorption Sorption experiments, under equilibrium conditions, could only be conducted with dense membranes (EC, MEC, PS and ABS). Sorption experiments could not be conducted for PERVAP® membranes (composite structure) due to the difficulty in differentiating between skin and support layers. Sorption results of water and hydrazine hydrate for all the membranes are reported in Table 4. The temperature of sorption experiments was maintained at 50 ◦ C. This was done in order to compare such sorption results with subse-

quent pervaporation experimental results to be conducted at the same temperature. Apolar membranes (ABS and PS) exhibit relatively low sorption characteristics for both hydrazine hydrate and water, reconfirming hydrophobicity. Similarly, MEC4 membrane which changed its characteristics due to the reaction with isocyanate also exhibits relatively lower sorption of water and hydrazine hydrate as compared to pure EC membrane, indicating increase in hydrophobicity. Sorption results of water and hydrazine hydrate on MEC membranes are shown in Fig. 6 against varying amounts of carbamate groups. There is decrease in sorption with increase in the amounts of carbamate groups and the sorption is negligible at higher amounts of carbamate groups. Sorption selectivity was estimated, assuming linear solubilities [5,31]. This assumption may be considered to be valid for polar components in apolar polymers. Fig. 6 was

Table 4 Sorption characteristics of water and hydrazine hydrate (T = 50 ◦ C) Membrane

Sorption of water (g/g of polymer) (%)

Sorption of hydrazine hydrate (g/g of polymer) (%)

Sorption selectivity (water/hydrazine)

EC MEC1 MEC2 MEC3 MEC4 PS ABS

4.01 3.75 2.70 1.41 0.25 0.11 0.30

5.21 4.95 3.70 2.09 0.39 0.22 0.51

0.681 0.667 0.665 0.570 0.533 0.390 0.482

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115 0.72

Hydrazine hydrate Water sorption selectivity

5

0.68

4

0.64

3

0.60

2

0.56

1

0.52

0

Sorption selectivity (water/hydrazine)

Sorption (g / 100 g of polymer solution)

6

111

0.48 0

5

10

15

20

25

Number of carbamate groups / 100 units of ethyl cellulose Fig. 6. Influence of amounts of carbamate groups on sorption of water and hydrazine hydrate.

drawn to observe a relationship between sorption selectivity and amounts of carbamate groups. Further, Table 4 presents sorption selectivity for all the membranes. Interestingly, from the sorption results, it may be observed that hydrazine hydrate being more selective to all the membranes compared to water (selectivities less than 1).

on obtained selectivity, seems to be a better proposition for the removal of water from hydrazine hydrate. The results (Table 5) clearly show that the PSI index and selectivity of laboratory prepared hydrophobic membranes (MEC, PS and ABS) are better than that of hydrophilic membranes (PERVAP® and EC). Earlier, Ravindra et al. [5] obtained flux, selectivity and PSI values (at room temperature for EC membrane) as 50 g/m2 h (for 5 ␮m thickness), 2.82 and 141, respectively. In the present work, flux has almost doubled but the selectivity has decreased significantly (61% of 2.82). However, the present work obtains a PSI value of 166 which is 1.17 times higher than earlier work [5]. This small increase therefore, may be simply explained because of the choice of higher operation temperature of 50 ◦ C. Further, PERVAP® 2201 membrane gave a reasonably high selectivity of 3.58; but, at the expense of flux which is very low (4 g/m2 h for 5 ␮m thickness). This is because the said membrane is supposed to exhibit a highly cross-linked PVA skin layer [16]. The selectivity, obtained with PS membrane, is the highest but the flux and PSI index are very low. This is because of strong hydrophobic nature of PS membrane. One may, therefore, decrease its hydrophobicity by modifying the polymer or blend with hydrophilic polymer to obtain better flux. In

3.3. Pervaporation Pervaporation of hydrazine hydrate was carried out at 50 ◦ C, 0.1 mmHg with static feed conditions, using different PERVAP® and laboratory prepared membranes. The obtained flux, selectivity and PSI at quasi-steady state are reported in Table 5. It may be mentioned here that the choice of 50 ◦ C temperature, higher than the earlier reported work with EC membrane only [5] at room temperature, was primarily to obtain an appreciable quantity of permeate which could enhance accuracy of analytical estimation. It is encouraging to observe (Table 5) higher separation for hydrazine–water using MEC and other selected hydrophobic polymers. The selectivity of water to hydrazine (of all the membranes except PERVAP® 2200) is higher than the selectivity obtained by distillation (1.4 at 760 mmHg [2]). Therefore, pervaporation in comparison to distillation, based Table 5 Flux, selectivity and PSI index (T = 50 ◦ C; downstream pressure = 0.1 mmHg) Membrane

Thickness (␮m)

Actual flux (g/m2 h)

Normalised flux (g/m2 h for 5 ␮m)

Selectivity

PSI (g/m2 h for 5 ␮m)

PERVAP® 2200 PERVAP® 2201 PERVAP® 2202 EC MEC4 PS ABS

2 2 2 58 75 56 110

306.00 10.00 78.50 8.28 5.45 1.02 3.55

122.4 4.0 31.4 96.5 81.8 11.5 78.1

1.09 3.58 1.93 1.72 3.23 5.45 5.07

133.4 14.32 60.6 166.0 262.4 62.6 395.9

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115 98

3.4

Flux Selectivity

Flux (g/m2h for 5µm)

96

3.2

94

3.0

92

2.8

90

2.6

88

2.4

86

2.2

84

2.0

82

1.8

80

Selectivity

112

1.6 0

5

10

15

20

25

Number of carbamate groups / 100 units of ethyl cellulose Fig. 7. Influence of amounts of carbamate groups on total flux and selectivity.

this regard, an attempt was made to blend PS and EC (both are soluble in toluene) to obtain better PSI value. However, it was observed that there was phase separation between PS and EC in toluene. This may be attributed to the fact that entropy of mixing may be small and there may be a positive enthalpy of mixing which gives rise to positive Gibbs free energy of mixing [32]. Results are, therefore, not reported. One of the interesting polymers (ABS), selected for the present work, gave the highest PSI value with reasonably high flux and better selectivity. ABS is a blend of glassy copolymer with rubbery domain [33]. The component acrylonitrile is hydrophilic whereas styrene and butadiene are hydrophobic by nature. Therefore, the high selectivity may be attributed due to the presence of glassy copolymer (styrene-co-acrylonitrile) and good flux may be due to the

presence of rubbery domain (butadiene). Such a combination of two different contrast properties led to a high PSI index (≈396). Still, there may be a possibility to improve PSI index by varying the composition of ABS polymer. Considering the effect of modifications on EC membrane, it may be observed from Table 5 that MEC4 gave better selectivity and PSI index than that obtained from EC. Increased hydrophobicity may have decreased the solubility and hence swelling; this may have increased the diffusion selectivity of water to hydrazine. These results were, however, obtained with a fixed amount of isocyanate (4 ml per 100 g of polymer solution). Therefore, another set of experiments were carried out to observe the influence of varying amounts of isocyanate (in other words varying amounts of carbamate groups) on the modifications of EC. Fig. 7

Fig. 8. Influence of amounts of carbamate groups on PSI.

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

113

Table 6 Estimated values of diffusion selectivity Membrane

Diffusion selectivity (water/hydrazine)

EC

MEC1

MEC2

MEC3

MEC4

PS

ABS

2.52

2.80

3.64

5.22

6.05

13.97

10.52

shows the effect of amounts of carbamate groups on flux and selectivity. There is a 16% decrease in flux (increase of carbamate groups from 0 to 24) whereas selectivity increased up to 92%. This is more evident from Fig. 8 with respect to PSI value. The figure shows a steep increase in PSI up to 15 carbamate groups, which correspond to 6.4 PSI/carbamate groups. Overall, there is 57% increase in PSI which may be considered to be significant. However, with significant increase in the amounts of isocyanate (beyond 4 ml), the membrane casting was becoming difficult (particularly, during peeling off the membrane from the casting glass plate). Therefore, membrane may have lost its physical strength. It was observed earlier (Section 3.1.5) that with increase in the amounts of carbamate groups, contact angles for both the components increased, suggesting decrease in adhesion force between the membrane material and feed components [33]. Decrease of adhesion force decreases the interaction between the membrane material and the components. This reduces the solubility of water and hydrazine hydrate in the membrane material (Fig. 6). However, such a phenomenon (decrease of adhesion forces) facilitates in easy desorption of components. Overall, in spite of increase in hydrophobicity as well as decrease in free volume fraction (due to isocynate reaction on EC), only around 16% decrease, in the flux, was observed. This may be due to the fact that the modification is providing a membrane (MEC) towards an amorphous structure as

compared to semi crystalline structure of EC (as explained in Section 3.1.3). 3.4. Diffusion selectivity It may be noted while comparing Tables 4 and 5 that sorption selectivity was less than 1; whereas, pervaporation selectivity was greater than 1. According to the solution-diffusion model [35], overall selectivity of pervaporation process αPWH is the product of sorption αSWH and diffusion selectivity’s αD WH , i.e. αPWH = αSWH αD WH

(4)

Use of Eq. (4) is highly restricted as the calculation of sorption selectivity is based on pure component swelling data. However, this may be considered to be valid for very low solubilities of feed components in the membrane. Such an assumption was made only to highlight the importance of diffusion during pervaporation. Eq. (4) thus provides diffuS sion selectivity αD WH , once sorption αWH and overall pervaP poration selectivity’s αWH are known. Such estimated values (diffusion selectivity) are reported in Table 6. Values are found to be greater than 1. This may be attributed due to the fact that water, being the smaller molecule, diffuses easily through the membrane compared to hydrazine. The study, therefore, confirms importance of diffusion selectivity over sorption selectivity in the case of dehydration of hydrazine hydrate. Similar such attributions were made in the

6

α = 27.92exp(-1.526θr)

5

correlation coefficient = 0.9963

Selectivity

4

3

2

1

0 1.0

1.5

2.0

Contact angle ratio (water to hydrazine hydrate) Fig. 9. Contact angles ratio vs. pervaporation separation factor.

2.5

114

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

case of ethanol–water separation using hydrophobic PVC [36]. It may, therefore, be concluded that sorption selectivity may always be less than 1 for a system having highly polar components like hydrazine and water. Accordingly, for higher values of pervaporation selectivity diffusion selectivity has to be higher. Such a desirable property (improvement in diffusion selectivity) may be attained by introducing hydrophobic moiety into the polymer chain as well as by thermal annealing of the membrane. Obviously, such an attempt may be at the expense of flux. 3.5. Contact angle versus pervaporation selectivity Contact angles reflect interaction between membrane material and feed components [37]. Further details on contact angles are nicely described by van Oss [34]. In pervaporation, both sorption and desorption may get reflected by contact angles. Therefore, one may develop a relationship between contact angles and pervaporation selectivity to help to understand the process. An attempt has been made to relate such available data (Table 3, Fig. 5, Table 5 and Fig. 7). Fig. 9 depicts relationship between process selectivity and ratio of contact angles of water to hydrazine hydrate. Following exponential relationship was obtained with a regression coefficient of 0.9963 αPWH = 27.92 exp(−1.526θr )

(5)

where, θr is the contact angle ratio of water to hydrazine hydrate. Eq. (5) may help to estimate separation factor (for a polymer of unknown pervaporation performance) with known contact angles.

out and sorption of both water and hydrazine hydrate for MEC membrane were observed to be lower compared to unmodified. XRD analysis suggested MEC4 to be slightly amorphous compared to EC. Estimated free volume fractions of MEC4 membrane, using PAL technique, were observed to 50% (from 0.042 to 0.021) to that of EC membrane. Pervaporation of hydrazine hydrate experiments provided encouraging results. The selectivity of water to hydrazine (of all the membranes except PERVAP® 2200) is higher than the selectivity obtained by distillation (1.4 at 760 mmHg [2]). Pervaporation studies with all chosen polymers revealed selective diffusion of water playing major role compared to sorption. Further, apolar materials exhibited higher separation factors than polar materials. Encouraging results, in terms of higher PSI, were obtained with MEC4 (262.4) and ABS (395.9) membranes. A useful exponential relationship was obtained between process selectivity and the ratio of contact angles of water to hydrazine hydrate.

Acknowledgements One of the authors (PKB) thankfully acknowledges the funding received from Indo-French Centre for Promotion of Advanced Research (IFCPAR) in partial support to this work. Further, authors wish to thank Professor R.N. Mukherjee, Chemistry Department, IIT—Kanpur, for his help to study NMR spectra.

Nomenclature 4. Conclusions Several polymeric membranes (both hydrophilic and hydrophobic) were chosen and prepared to study separation characteristics during pervaporation of hydrazine hydrate at fixed operating conditions. One such polymer (EC) was modified (MEC1–MEC4) in order to alter the hydrophilic characteristics of the membrane to hydrophobic by reacting EC with phenyl isocyanate and varying its amount. Modifications were ascertained through FT-IR spectrum analysis which showed formation of carbamate group. Further, 1 H NMR spectra estimated degree of substitution in terms of carbamate groups. Contact angle measurements were taken to observe hydrophilic/hydrophobic characteristics of the membranes. XRD and PAL techniques were also employed to observe the characteristics of MEC4. The higher contact angles of water with MEC membrane, compared to EC membrane, indicate increase of hydrophobicity of the membranes. Further, hydrophobicity of the MEC polymer increases with increase in the amounts of carbamate groups. Sorption studies were also carried

C f I3 r r VF x y

constant in Eq. (2) free volume fraction ortho-positrionium intensity spherical free volume radius (nm) electron layer thickness (nm) free volume of the sphere weight fraction of water in feed weight fraction of water in feed

Greek symbols α selectivity τ3 life time of ortho-positrionium Subscripts H hydrazine W water Superscripts D diffusivity P pervaporation S sorption

S.V. Satyanarayana, P.K. Bhattacharya / Journal of Membrane Science 238 (2004) 103–115

References [1] E.W. Schmidt, Hydrazine and Its Derivatives: Preparation, Properties and Applications, Wiley, New York, 1984. [2] R.Q. Wilson, H.P. Munger, J.W. Clegg, Vapour–liquid equilibrium in the binary system of hydrazine/water, Chem. Eng. Prog. Symp. Ser. 3 (1952) 115–117. [3] F. Wolf, P. Pollandt, R. Helmer, Diffusion behaviour of hydrazine through molecular sieve membranes, Z. Chem. 15 (1975) 160– 161 (in German); Chem. Abstr. 83 (1975) 183856. [4] J.I. Dytnerskij, Membranprozesse Zur Trennung flussiger Gemische, VEB Deutscher Verlag fur Grundstoffindustrie, Leipzig, 1977; R. Rautenbach, R. Albrecht (Eds.), Membrane Processes, Wiley, Chichester, UK, 1989, p. 368. [5] R. Ravindra, S. Sridhar, A.A. Khan, Separation studies of hydrazine from aqueous solutions by pervaporation, J. Polym. Sci. Part B: Polym. Phys. 37 (1999) 1969–1980. [6] A. Duggal, E.V. Thompson, Dependence of diffusive permeation rates and selectivities on upstream and downstream pressures. V1. Experimental results for the water/ethanol system, J. Membr. Sci. 27 (1986) 13–30. [7] M. Wessling, U. Werner, S.-T. Hwang, Pervaporation of aromatic C8 -isomers, J. Membr. Sci. 57 (1991) 257–270. [8] B. Cao, H. Hinode, T. Kajiuchi, Permeation and separation of styrene/ethylbenzene mixtures through cross-linked poly (hexamethylene sebacate) membranes, J. Membr. Sci. 156 (1999) 43–47. [9] E.E.B. Meuleman, J. Willemsen, M.H.V. Mulder, H. Strathaman, EPDM as a selective membrane material in pervaporation, J. Membr. Sci. 188 (2001) 235–249. [10] S.V. Satyanarayana, P.K. Bhattacharya, Real coded genetic algorithm for optimization of pervaporation process parameters for removal of volatile organics from water, Ind. Eng. Chem. Res. 42 (2003) 3118– 3128. [11] X. Feng, R.Y.M. Huang, Liquid separation by membrane pervaporation: a review, Ind. Eng. Chem. Res. 36 (1997) 1048–1066. [12] R. Ravindra, S. Sridhar, A.A. Khan, A.K. Rao, Pervaporation of water, Polymer 41 (2000) 2795–2806. [13] H. Fuzita, Fortschr Hoch polym.-Forsch. 3 (1961) 1; R.Y.M. Huang (Ed.), Pervaporation Membrane Separation Processes, Elsevier, The Netherlands, 1991, p. 147. [14] S.V. Satyanarayana, K.K. Rao, R. Ravindra, P.R. Shah, A.A. Khan, Studies on the separation of ethanol–water and hydrazine–water mixtures by pervaporation, in: Proceedings of the Golden Jubilee Symposium on Advances Chemical Engineering, IICT, Hyderabad, 9–11 August 1994. [15] A. Kumar, R.K. Gupta, Fundamentals of Polymers, McGraw-Hill, Singapore, 1998. [16] A. Jonquières, R. Clément, P. Lochon, J. Néel, M. Dresch, B. Chrétien, Industrial state-of-art of pervaporation and vapour permeation in the western countries, J. Membr. Sci. 206 (2002) 87–117. [17] M.J.L. Kishore, Preparation of heterogeneous alumina catalyst covalently binding mono and bimetallic complexes for alkane isomerization, M.Tech. Thesis, Indian Institute Technology, Kanpur, 2003.

115

[18] I.L. Finar, Organic Chemistry, vol. 1: The Fundamental Principles, 6th ed., ELBS/Longman, Singapore, 1973, pp. 268, 359. [19] P. Kirkegaard, N.J. Pedersen, M. Eldrup, PATFIT-88: a data processing system for positron annihilation spectra on mainframe and personal computers, Riso National Laboratory, Denmark, 1989. [20] S.J. Tao, Positronium annihilation in molecular substances, J. Chem. Phys. 56 (1972) 5499. [21] S. Weinhold, G.M. Stack, M.R. Tant, A.J. Hill, Effect of copolymer composition on free volume and gas permeability in poly (ethylene terephthalate)–poly (1,4 cyclohexylenedimethylene terephthalate) copolyesters, Eur. Polym. J. 32 (1996) 843–849. [22] R.A. Penneman, L.F. Audrieth, Quantitative determination of hydrazine, Anal. Chem. 20 (1948) 1058–1061. [23] L.A. Dee, A.K. Webb, Gas chromatographic separation of hydrazine mixtures and water using a stationary phase that is chemically similar to hydrazine, Anal. Chem. 39 (1967) 1165–1167. [24] A. Hillaire, E. Favre, Isothermal and nonisothermal permeation of an organic vapour through a dense polymer membrane, Ind. Eng. Chem. Res. 38 (1999) 211–217. [25] R.Y.M. Huang, J.W. Rhim, Separation characteristics of pervaporation membrane separation processes, in: R.Y.M. Huang (Ed.), Pervaporation Membrane Separation Processes, Elsevier, The Netherlands, 1991. [26] W. Kemp, Organic Spectroscopy, 3rd ed., Pal-Grave Publishers, China, 1991. [27] R. Ravindra, K.R. Krovvidi, A.A. Khan, FT-IR, diffusivity, selectivity and aging studies of interactions of hydrazine, water and hydrazine hydrate with the ethyl cellulose membrane, Macromolecules 30 (1997) 3288–3292. [28] R.M. Silvenstein, F.X. Webster, Spectrometric Identification of Organic Compounds, 6th ed., Wiley, New York, 1998. [29] S.V. Satyanarayana, V.S. Subrahmanyam, H.C. Verma, A. Sharma, P.K. Bhattacharya, Positron annihilation study of pervaporation dense membranes, J. Membr. Sci., Communicated. [30] L.H. Sperling, Introduction to Physical Polymer Science, Wiley, New York, 1986. [31] A.B. Krewinghaus, Solution and transport of organic liquids and vapours in structurally modified in polyethylene, Ph.D. Thesis, MIT, 1966. [32] A. VrŸ, J. Polym. Sci. Part A 2 (1919) 6; C.J. van Oss (Ed.), Interfacial Forces in Aqueous Media, Marcel Dekkar, New York, 1994. [33] G.A. Morneav, W.A. Pavelich, L.G. Roettger, Bory-Warner Chemicals, Kirk and Othmer, Encyclopedia of Chemical Technology, vol. 1, Wiley/Interscience, New York, 1978, pp. 442–456. [34] C.J. van Oss, Interfacial Forces in Aqueous Media, Marcel Dekkar, New York, 1994. [35] J.G. Wijmans, R.W. Baker, The solution–diffusion model: a review, J. Membr. Sci. 107 (1995) 1–21. [36] H. Okuno, T. Nishida, T. Uragami, Preferential sorption and permeation of binary liquid mixtures in poly (vinyl chloride) membrane by pervaporation, J. Polym. Sci. Part B: Polym. Phys. 33 (1995) 299. [37] Y.M. Lee, D. Bourgeois, G. Belfort, Sorption, diffusion and pervaporation of organics in polymer membranes, J. Membr. Sci. 44 (1989) 161–181.