polyacrylonitrile composite membranes for gas dehydration and humidification

polyacrylonitrile composite membranes for gas dehydration and humidification

ARTICLE IN PRESS Chemical Engineering Science 65 (2010) 4672–4681 Contents lists available at ScienceDirect Chemical Engineering Science journal hom...

500KB Sizes 2 Downloads 14 Views

ARTICLE IN PRESS Chemical Engineering Science 65 (2010) 4672–4681

Contents lists available at ScienceDirect

Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces

Using poly(N,N-dimethylaminoethyl methacrylate)/polyacrylonitrile composite membranes for gas dehydration and humidification Jennifer Runhong Du, Li Liu, Amit Chakma, Xianshe Feng n Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1

a r t i c l e in f o

a b s t r a c t

Article history: Received 10 February 2010 Received in revised form 29 April 2010 Accepted 9 May 2010 Available online 31 May 2010

The transport of water vapor through a composite membrane consisting of hydrophilic poly(N,N-dimethylaminoethyl methacrylate) (PDMAEMA) as the active layer and polyacrylonitrile (PAN) as the substrate was investigated, and the performance of the membrane for gas dehydration and humidification applications was evaluated. For gas dehydration, methane/water vapor mixtures were used as feed and vacuum was applied on the downstream side. The feed composition and operating temperature were found to have a significant effect on the membrane performance. The PAN substrate had little effect on the permeation of methane, but the resistance of the substrate to water vapor permeation was significant because of the substantially higher permeability of water vapor in the membrane. For gas humidification, liquid water was brought to be in contact with the active layer of the membrane and nitrogen gas flowed on the other side. With an increase in the gas flow rate, the mass transfer rate of water through the membrane to reach the gas stream increased, and the humidity level of the gas stream decreased. The humidification can be enhanced significantly by operating at a higher temperature. A phenomenological mass transfer equation was derived for membrane humidifiers to correlate the overall mass transfer coefficient and membrane area, and this equation could be used in process design and scale up. & 2010 Elsevier Ltd. All rights reserved.

Keywords: Composite membrane Water permeation Gas dehydration Humidification Natural gas Poly(N,N-dimethylaminoethyl methacrylate)

1. Introduction Many processes involve dehydration or humidification of gases. In natural gas processing, for example, if water vapor in the natural gas is not removed, it not only enhances pipeline corrosion but may also form solid methane hydrates, causing pipeline blockage. Conventionally, the natural gas is dehydrated by scrubbing with glycols at the wellhead. This method is quite efficient for the removal of water vapor from the gas stream, but it involves complex processes for gas dehydration and solvent regeneration, which are unsuitable for offshore applications due to large space requirements. In addition, the BTEX (i.e., benzene, toluene, ethylbenzene and xylene) compounds emitted to the air from the vent stream during glycol regeneration impose an important environmental problem. Membrane process for gas dehydration is a promising alternative to absorption operation. Generally, a hydrophilic membrane is preferred for water vapor to permeate through the membrane preferentially, producing a dried retentate at a pressure that is essentially the same as the feed pressure. Sometimes the membrane can be used in conjunction with absorption to form a hybrid process where the membrane is used for bulk separation prior to absorption for

n

Corresponding author. Tel.: + 1 519 888 4567x36555; fax: + 1 519 746 4979. E-mail address: [email protected] (X. Feng).

0009-2509/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2010.05.005

final finishing, which allows for longer cycles of gas absorption and solvent regeneration, thereby improving the overall efficiency of the process. Opposite to gas dehydration, some applications require humidification of a gas stream. Fuel cells using proton exchange membrane are a typical example, where water management is critical to the performance and reliability of the fuel cell (Hyun and Kim, 2004). If the membrane is too dry, the proton conductance will be reduced, leading to an increased resistive loss and thus a decreased power output. On the other hand, too much water present in the fuel cell assembly will cause cathode flooding, resulting in blockage of the gas flow channels. Therefore, it is crucial to maintain the humidity of the membrane at a proper level to avoid dehydration and flooding. At present, humidifiers based on gas bubbling through water are widely used, and membrane-based humidifiers have attracted significant attention. In a membrane humidifier, the incoming dry gas flows to one side of a water permeable membrane, and hot water flows to the other side so as to induce water vapor transport through the membrane to humidify the gas without any flooding problems. Water vapor transport through membranes is important for gas dehydration and humidification by membranes. Same as gas permeation and pervaporation, the permeation of water vapor through a membrane is also governed by the solution–diffusion mechanism. The permeability of water vapor in polymeric membranes varies by 5 orders of magnitudes, depending on the

ARTICLE IN PRESS J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

membrane materials (Metz et al., 2005a). Hydrophilic polymers with functional groups capable of interacting with water molecules (including hydrogen bonding, ion–dipole and dipole– dipole interactions) are desired for gas dehydration and humidification (Semenova et al., 1997). In general, the stronger the interaction between water and the membrane is, the higher will be the sorption selectivity. Thus the interactions based on hydrogen bonding and ion–dipole interactions are expected to yield a high solubility selectivity. The sorbed water in a hydrophilic membrane will induce membrane plasticization and/or swelling, making molecular diffusion in the membrane easier to occur. On the other hand, water molecules in membrane tend to form clusters due to strong hydrogen bonding, and cluster size depends on the strength of the interaction between water molecules and the polymer matrix. If the water–water interaction dominates over the water–polymer interaction, larger clustering of water molecules will take place. Thus, the size of clusters in hydrophobic polymers is usually larger than that in hydrophilic membranes. The molecules in the clusters are immobilized with a low diffusivity as compared to the unassociated water molecules (Barrie et al., 1975; Mensitieri et al., 1995). Thus, generally, the use of hydrophilic polymer membranes can enhance the solubility and diffusivity of water vapor in a membrane because of the strong interaction between permeating water molecules and hydrophilic groups on the polymer chains. The permselectivity of water vapor in several hydrophilic polymeric membranes has been reported. Landro et al. (1991) reported a H2O/CH4 selectivity in the range of 103–104 in an amorphous polyurethane membrane. Fu et al. (1994) investigated the permeation of water vapor, oxygen and nitrogen through sulfonated poly(phenylene oxide) membranes and found that the selectivity of water vapor over other gases increased remarkably with an increase in the degree of sulfonation. Huang et al. (2003) studied the water vapor permeation through polyimide dense membranes. A membrane from a blend of polyimide and sulfonated polyethersulfone was studied for dehydration of compressed air by Wu et al. (2002). Metz et al. (2005b) studied the water vapor/ nitrogen permeation in block copolymers of poly(ethylene oxide) (PEO) and poly(butylene tererphthalate) (PBT), where the hydrophobic PBT segments provide mechanical strength to the polymer, and the water vapor permeability was shown to depend strongly on the structure and the composition of the copolymer. In the present work, water vapor permeation through poly(N,N-dimethylaminoethyl methacrylate)/polyacrylonitrile (PDMAEMA/PAN) composite membranes relevant to gas dehydration and humidification was studied. The membranes were prepared by interfacial crosslinking. The potential use of the membrane for natural gas dehydration was evaluated in terms of water vapor separation from methane. The effects of operating parameters (e.g., feed composition and temperature) and the resistance of the PAN substrate on the membrane performance was investigated. For gas humidification, nitrogen was used as the sweeping gas and the humidification performance was evaluated at different gas flow rates and operating temperatures. PDMAEMA and its copolymers are suitable materials for gas separation membranes (Zhao et al., 2008; Ji et al., 2009). Because of the amino groups in the polymer, PDMAEMA is expected to have good permeability to water vapor.

ultrafiltration membrane (thickness  70 mm and molecular weight cutoff 20 kDa) supplied by Sepro Membranes was used as the substrate. Both gases CH4 and N2 used in the permeation experiments were of research grade (99.0–99.998% pure) and were supplied by Praxair Specialty Gases and Equipment. 2.2. Membrane preparation The PAN membrane was rinsed with deionized water prior to use as a substrate. The PDMAEMA/PAN composite membranes were prepared by interfacial crosslinking of a thin layer of PDMAEMA coated onto the PAN substrate. The detailed procedure of membrane formation has been reported elsewhere (Du et al., 2007). Briefly, an ethanol solution containing 15 g/L of PDMAEMA was coated onto the surface of the PAN substrate for 15 min. After the excess solution on the membrane was removed, the membrane was air dried and a thin layer of PDMAEMA was formed on the membrane. Then the PDMAEMA layer was allowed to contact with a solution of p-xylylene dichloride dissolved in heptane at a concentration of 10 g/L for 5 h at ambient temperature, during which period PDMAEMA was crosslinked with p-xylylene dichloride at the solid–liquid interface. Then the excess crosslinking solution was removed, and the resulting membrane was rinsed with heptane and then dried in air. The thickness of the PDMAEMA layer was ca 1.5 mm. 2.3. Permeation tests for gas dehydration The schematic diagram of the experimental setup for gas dehydration is shown in Fig. 1. The membrane was mounted in the permeation cell, which had an effective permeation area of 16.6 cm2. A stream of methane gas was bubbled through a water bath, and this moist stream (which was essentially saturated with water) was mixed with a stream of dry methane. The flow rates of both gas streams were controlled by mass flow controllers (Matheson Gas, Model 8270), and the humidity of the binary methane/water vapor mixtures so obtained was controlled by adjusting the flow rates of the two streams. The gas mixture at a given humidity was fed continuously to the membrane unit at a flow rate of 1500 sccm. The operating temperature, which varied in the range of 25 to 55 1C, was controlled using a thermal bath. A paraffin oil liquid seal was used in the retentate line to ensure that the water vapor present in the air will never enter the membrane system during permeation. The water contents in both the feed and retentate streams were measured by a dew point meter (Vaisala Drycap, Model DM70, Finland). The feed side was maintained at atmospheric pressure, while the permeate side was kept at a low pressure (which was 1.7 kPa, unless specified otherwise) using a vacuum pump. The water vapor that permeated through the membrane was condensed in a cold trap immersed in liquid nitrogen, and its permeation rate was determined by weighing the permeate water collected in the condenser for a given period of time. The permeance of water vapor was calculated from the following equations: Qvap ¼ AJvap ðpf Xpp YÞ

ð1Þ

QCH4 ¼ AJCH4 ½pf ð1XÞpp ð1YÞ

ð2Þ

Y¼ 2. Experimental 2.1. Materials PDMAEMA was synthesized by bulk radical polymerization as described in our previous work (Du et al., 2006). PAN

4673

Qvap Qvap þQCH4

ð3Þ

where Q is the permeation rate (cm3(STP)/s), J is the permeance (cm3(STP)/cm2 s cmHg) (1 cm3(STP)/cm2 s cmHg¼106 GPU ¼3.35  10–4 (mol/(m2 s Pa)), and subscripts vap and CH4 represent water vapor and methane, respectively. A is the effective membrane area for permeation (cm2), and X and Y are

ARTICLE IN PRESS 4674

J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

Mass flow controller

Pressure regulator

Paraffin oil seal

Dew point meter Dew point meter

Needle valve

Vacuum gauge

Thermal bath

Permeation cell Mass flow controller

Switch valve

Vacuum pump

CH4 cylinder Collection tube Humidifier

Fig. 1. Schematic diagram of experimental setup for methane dehydration.

the concentrations (mol fraction) of water vapor on the feed and permeate sides, respectively. In spite of the relatively high feed flow rate used, the water concentration at the retentate exit was found to be about 8–10% lower than the feed inlet because of the high selectivity, and thus X was taken as an average of the water concentration. pf and pp are the feed and permeate pressures (cmHg), respectively. Looking into Eqs. (1)–(3), A, pf, pp, Qvap and X are the known quantities from the experiment, and methane permeance JCH4 can be determined separately as shown below. Thus water vapor permeance Jvap and the other two unknown quantities (i.e., QCH4 and Y) can be determined from the above three equations. The permeance of methane JCH4 was measured with the traditional constant pressure-variable volume method, and the procedure and apparatus were the same as those described before (Du et al., 2006). In the feed side, a pressurized methane gas stream was saturated with water vapor by flowing the gas through a porous sintered stainless steel ball immersed in water and then through a packed bed of glass beads to minimize the carryover of water mist. JCH4 was determined from the permeation rate by JCH4 ¼

V At Dp

Water reservoir

Pressure regulator

Mass flow controller

Needle valve

Membrane

Thermal bath

Dew point meter

Paraffin oil seal N2 cylinder

Fig. 2. Schematic diagram of experimental setup for gas humidification.

the membrane unit by QVap ¼ Qgas

yvap 1yvap

ð5Þ

ð4Þ 3. Results and discussion

where V is quantity (cm3(STP)) of methane permeate over a period of time t (s) at a transmembrane pressure differential of Dp (cmHg). The methane permeance so obtained was assumed to be the same as that in the gas dehydration experiments because it is a constant that is independent of pressures.

2.4. Permeation tests for gas humidification The setup for gas humidification is shown schematically in Fig. 2. Nitrogen was used as the sweeping gas in the experiments. Water vapor that permeated through the membrane was carried over by the carrier gas, which was thus humidified. The pressures on both sides were kept at atmospheric pressure. The flow rate of the sweeping gas was controlled by a mass flow controller. The water content in the outlet of the sweeping gas stream was measured by the dew point meter. The water permeation rate (QVap) was calculated from the carrier gas flow rate (Qgas) and the concentration of water vapor (yvap) in the sweeping gas exiting

3.1. Separation of water vapor from methane Water vapor is condensable and is able to interact strongly with itself and with the hydrophilic polymer via hydrogen bonding. Therefore, water vapor generally has a much higher solubility and diffusivity in hydrophilic polymeric materials than permanent gases do. The PDMAEMA membrane prepared in this work is ionic and it contains hydrophilic groups, i.e., tertiary and quaternary amino groups, as shown schematically in Fig. 3 (Du et al., 2006, 2007). Fig. 4 shows the results of the dehydration experiments at an operating temperature of 25 1C when the relative humidity of the feed was changed from 10% to 95% (corresponding to a water content of 0.32–3.1 mol%). As shown in Fig. 4(a), the water vapor is significantly enriched in the permeate stream, with a vapor concentration of over 92 mol%. This confirms the preferential permeation of water vapor in the hydrophilic PDMAEMA membrane. The water flux, shown in Fig. 4(b), experiences a progressive increase with an increase in the water content in the feed, which is not surprising in

ARTICLE IN PRESS J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

N O

OO

O

N

N

N

N

N

N

4675

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

N

N+

N+

N+

N

N

N+

N

N

N+

Cl

N

N+

N

N

N+

N

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

N

N

N

N

N

O

N

O

N

Fig. 3. Schematic diagram of chemical structure of PDMAEMA active layer.

Molar fraction of H2O in permeate

1.00

0.98

0.96

0.94

0.92

Water flux (cm3 (STP)/cm2.s)

0.90 0.0120.0000.0050.0100.0150.0200.0250.030 0.010 0.008 0.006 0.004 0.002 0.000 0.00

0.01 0.02 Molar fraction of H2O in feed

0.03

Fig. 4. Effect of water vapor concentration in feed on (a) permeate concentration and (b) water permeation flux. Operating temperature 25 1C.

consideration of the strong interaction between water and the hydrophilic polymer. It is believed that the high permselectivity to water vapor permeation is primarily derived from the solubility selectivity. There is generally a positive deviation from the Henry’s law with

respect to sorption isotherm of water vapor in polymers, although different degrees of deviations have been observed. Schult and Paul (1996) showed that the sorption isotherms are almost linear at low activities (i.e., obeying Henry’s law), and the Flory–Huggins model applies at high activities due to the plasticization of the polymer and the clustering of water molecules. At intermediate activities, a weak dual-mode sorption combining Henry’s law ‘‘dissolution’’ and Langmuir ‘‘hole filling’’ is often observed. In other studies, water vapor sorption is found to follow a dualmode isotherm at low activities (Stannett et al., 1980; Hayashi et al., 1999), whereas at high activities the clustering effect is so significant that the Flory–Huggins model was not adequate (Stannett et al., 1980). On the other hand, a non-Fickian diffusion behavior of water vapor has also been reported due to the hydrogen bonding clusters (Schult and Paul, 1996; Aranda et al., 1995). Sorption and diffusion experiments can be carried out for membranes with homogeneous structures. Because of the crosslinking gradient in PDMAEMA layer induced by interfacial crosslinking and the asymmetric structure of the composite membrane studied here, a detailed investigation on water vapor sorption and diffusion is difficult. Nevertheless, it appears clear that the membrane is very permselective for water vapor permeation, which will be discussed further. The permeant–permeant and permeant–polymer interactions may cause a coupling effect in the permeation of gas mixtures (Zhu et al., 2005). Water vapor is preferentially sorbed into the membrane, and the sorption of water can result in plasticization or swelling of the polymer matrix. In this case, the permeation rate of other gas component in the mixture will be enhanced. However, too much sorbed water in the membrane will block the pathways in the membrane available for other component to diffuse, which tends to reduce the permeation rate of other gases. On the other hand, because of the strong affinity between water and the hydrophilic membrane, the presence of other gas (e.g., methane) in the mixture is unlikely to affect the water sorption and permeation significantly, especially at low temperatures. In this work, water vapor was found to have little effect on the permeation of methane over the experimental range studied. At 25 1C and a transmembrane pressure difference of 1 atm, the

ARTICLE IN PRESS 4676

J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

10

5000

5

Jvap (GPU)

JCH4 (GPU)

10000

1 10000 3.0

1000 50000.0000.0050.0100.0150.0200.0250.030

4000

3.2

3.3

3.2

3.3

Jvap (GPU)

5000

4

αvap/CH

3.1

3000

2000

1000

1000 0.00

0.01 0.02 Molar fraction of H2O in feed

0.03

3.0

3.1 1000/T (1/K)

Fig. 5. Effect of water vapor concentration in feed on (a) water vapor permeance and (b) vapor/methane selectivity. Operating conditions same as in Fig. 3.

Fig. 6. Permeance of methane and water vapor at different temperatures. Permeate pressure 1.7 kPa. Mole fraction of water in feed: (J) 0.005; (&) 0.015 and (W) 0.025.

permeance of dry methane is 1.06 GPU, which is very close to the permeance of methane in the presence of saturated water vapor (which is 1.13 GPU). The interfacial crosslinking produces an asymmetric crosslink network structure. The high mobility of uncrosslinked side chains in PDMAEMA beneath the thin crosslinked outer surface enables effective gas transport even in the absence of water vapor. Fig. 5 shows the permeance of water vapor in the membrane and the membrane selectivity. The permeance of water vapor is found to increase exponentially with water vapor concentration in the feed side of the membrane, as shown in Fig. 5(a), and it can be described empirically by

depends on the activation energy of diffusion (ED) and the heat of sorption (DHS)

Jvap ¼ 3866 expð10:54XÞ

ð6Þ

where X is the mol fraction of water vapor in the feed. The selectivity of water vapor over methane can reach 5000 when methane is saturated with water vapor (see Fig. 5(b)). The effects of temperature on the permeance of water vapor and methane through the PDMAEMA/PAN composite membrane are shown in Fig. 6, where the permeance was plotted versus the reciprocal of operating temperature on a semi-logarithmic scale. The permeance of methane increases with an increase in temperature, as shown in Fig. 6(a), which suggests that the temperature dependence of methane permeance can be described by an Arrhenius type of relationship. The activation energy for methane permeation was calculated to be 20.1 kJ/mol. The permeance of water vapor, on the other hand, decreases with an increase in the temperature, as shown in Fig. 6(b). This is understandable and can be explained qualitatively. Based on the solution–diffusion model, the permeability is determined by the solubility and diffusivity of the permeant in the membrane, and the overall temperature dependence of permeability (which can be characterized by the activation energy for permeation EP)

EP ¼ ED þ DHS

ð7Þ

The diffusivity increases with an increase in temperature and the activation energy for diffusion is always positive. For a permanent gas, the heat of sorption normally has a small negative value, and the gas permeability through polymeric membranes increases with an increase in temperature. This is found to be the case for methane permeation through the PDMAEMA/PAN membrane, as shown in Fig. 6(a). However, for the permeation of such condensable permeant as water vapor, the heat of sorption is more significant and may dominate over the activation energy for diffusion, resulting in a negative temperature dependence of permeability. This is shown in Fig. 6(b). In addition, at a given concentration of water vapor in the feed, an increase in temperature increases its saturated vapor pressure and thus its activity decreases. Therefore, the negative effect of temperature on water vapor permeation will be more significant at a higher temperature. With an increase in temperature, the decrease in water vapor permeability and the increase in the methane permeability result in a reduction in the membrane selectivity. This is illustrated in Fig. 7, where the water content in permeate is shown to decrease with increase in temperature. It can thus be concluded that a lower operating temperature is more advantageous for natural gas dehydration. It should be noted that in natural gas dehydration, the methane loss due to permeation through the membrane is an important factor that affects the process economics. From an application point of view, the small fraction of methane permeated through the membrane can be recovered upon condensation of water vapor in the permeate for discharge.

ARTICLE IN PRESS

1.00

20000

0.95

15000 JCH4, PAN (GPU)

Mole fraction of H2O in permeate

J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

0.90

4677

10000

5000

0.85

0

0.80

50

30

40

50

60

60

70

80

90

100

pm (kPa)

Temperature (°C)

where l is the thickness of the substrate, r the average radius of the pores, e the area porosity of the substrate, M the molecular weight of the permeant, R the ideal gas constant, pm the average pressure, Z the gas viscosity, T the temperature and PPAN the intrinsic permeability of PAN dense membranes. At a given temperature, the permeance of methane through the PAN substrate membrane at different pressures was found to be constant, as shown in Fig. 8, indicating that the contribution of viscous flow was insignificant. This hypothesis is further supported by experimental data of the permeances of methane and water vapor at different temperatures, as shown in Fig. 9, which shows a linear relationship between the permeance of the PAN substrate and reciprocal of square root of temperature (1/T0.5). From the methane permeation data through the PAN substrate, the e and r of the substrate cannot be determined separately, but the dimensionless lumped parameter (er/l) was found to be 3  10–5. In an ideal composite membrane, separation occurs in the thin dense skin layer, while the microporous substrate functions as a mechanical support. In this case, the mass transfer resistance is

20000

18000 JCH4, PAN (GPU)

Because the concentration of water vapor in the subatmospheric gaseous permeate stream is much higher than the saturated vapor pressure at ambient conditions, the condensation can be accomplished easily by, for example, letting the permeate stream reach atmospheric pressure. In the PDMAEMA/PAN composite membrane, the PAN substrate is meant to provide mechanical support to the PDMAEMA skin layer. Since the permeability of water vapor in the membrane is several orders of magnitude greater than the permeability of methane, it is of interest to investigate whether the resistance of the PAN substrate to gas permeation can indeed be neglected. A simple resistance model approach (Huang and Feng, 1993; Liu et al., 2009) is used here. While it is impossible to study the permeation in the thin skin layer of the composite membrane alone, the substrate PAN membrane can be evaluated independently. For this purpose, the permeation of methane and water vapor through the PAN substrate was also determined experimentally. The gas permeation through a porous substrate can be considered to consist of gas transport in the pores, which is mainly controlled by Knudsen diffusion and viscous flow, and gas permeation in the polymer matrix (Liu et al., 2009)  1=2 er 2 er 32 P ð1eÞ pm þ þ PAN ð8Þ JPAN ¼ 8ZRTl l 9pMRT l

Fig. 8. Permeance of methane in PAN substrate at different temperatures. (&) 35 1C; (J) 40 1C; (W) 45 1C; (  ) 50 1C and () 55 1C.

16000

14000

12000

10000 1600055.055.556.056.557.0

14000 Jvap, PAN (GPU)

Fig. 7. Permeate concentration versus operating temperature. Mole fraction of water in feed: (J) 0.005; (&) 0.015 and (W) 0.025.

12000

10000

8000 55.0

55.55

6.05

6.55

7.0

1000/T0.5 (1/K0.5) Fig. 9. Permeance of (a) methane and (b) water vapor in PAN substrate versus square root of temperature. Mole fraction of water in feed: (J) 0.005 and (W) 0.025.

controlled by the dense skin layer. If the resistance of the substrate is not negligible as compared to the skin layer resistance, the overall performance of the composite membrane will be compromised (Liu et al., 2001). The total resistance of the composite membrane RTotal can be considered to be the sum of the resistances of the skin layer RPDM and the PAN substrate RPAN RTotal ¼ RPDM þRPAN

ð9Þ

ARTICLE IN PRESS 4678

J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

0.00020

10

0.00015 (RPAN/RTotal)CH4

JCH

4, PDM

(GPU)

5

0.00010

0.00005

1 10000 3.0

3.1

3.2

0.00000 0.5 30

3.3

40

50

60

40

50 Temperature (°C)

60

(RPAN/RTotal)Vap

Jvap, PDM (GPU)

0.4

5000

0.3 0.2 0.1 0.0 30

1000 3.0

3.1

3.2

3.3

1000/T (1/K) Fig. 10. Permeance of methane and water vapor in PDMAEMA coating layer. Mole fraction of water in feed: (J) 0.005 and (W) 0.025.

and the resistance is related to the permeance R¼

1 JA

ð10Þ

Using Eqs. (9) and (10), the resistance of the PDMAEMA skin layer in the PDMAEMA/PAN composite membrane to the permeation of water vapor and methane can be evaluated from their permeance in the composite membrane JTotal and the PAN substrate JPAN. Then the permeance of water vapor and methane through the PDMAEMA skin layer can be calculated easily, and the results are presented in Fig. 10. Comparing the data in Fig. 10 with the permeance data of the composite membrane shown in Fig. 6, a similar trend in the permselectivity can be observed. However, while the permeance of methane in the PDMAEMA skin layer is almost the same as its permeance in the composite membrane, the permeance of the composite membrane to water vapor is considerably lower than the PDMAEMA layer. This is more clearly illustrated in Fig. 11 by using the mass transport resistance of the PAN substrate relative to the total resistance of the composite membrane for the permeation of water vapor and methane. The resistance derived from the substrate is negligible for methane permeation. Conversely, for water vapor, which has a much higher permeability, the porous PAN substrate accounts for 30–40% of the total resistance. The above analysis indicates that the composite membrane did not work out its full potential, and the substrate should be optimized to minimize its influence on the mass transfer, especially for the fast-permeating component. The effect of the substrate is expected to be more significant with a further reduction in the top layer thickness for increased permeance. Nonetheless, because of the high water vapor permeability in the PDMAEMA, the potential of using the membranes for natural gas dehydration has been demonstrated. It may be mentioned that a

Fig. 11. The ratio of PAN substrate resistance to the total resistance of the PDMAEMA/PAN composite membrane for the permeation of (a) methane and (b) water vapor. Mole fraction of water in feed: (J) 0.005 and (W) 0.025.

PDMAEMA/polysulfone composite membrane has been studied for use in pervaporation for separation of water from ethylene glycol (Du et al., 2008). Therefore, the PDMAEMA-based composite membranes could be used in a hybrid process for natural gas dehydration that integrates a membrane process for bulk dehydration and an absorption process with ethylene glycol where a glycol regeneration can be accomplished with pervaporation.

3.2. Membrane humidifier The membrane humidification process can also be regarded as a pervaporation process where liquid water is the feed and there is a purging gas stream instead of vacuum on the permeate side. In this section, the humidification performance was investigated using nitrogen as the sweeping gas. The driving force for water vapor transport through the membrane was provided by lowering the vapor pressure on the permeate side with the use of the sweeping gas. The effect of the sweeping gas flow rate, which was varied from 50 to 1750 sccm, on the humidification performance of the composite membrane is shown in Fig. 12. As expected, there is a reduction in the humidity level of the outlet gas stream when the inlet gas flow rate increases. An increase in the sweeping gas flow rate will lower the partial pressure of water vapor on the permeate side, and thus the driving force increases, leading to an increased water transfer rate. Apparently, the increase in water transfer rate is less than proportional with an increase in the gas flow rate. It should be pointed out here that the water transfer rate in Fig. 12 represents the overall transfer rate averaged over the whole membrane area as the local water transfer rate varies

ARTICLE IN PRESS J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

where psat is the saturated water vapor pressure (cmHg) at the operating temperature and Q is the gas flow rate (cm3(STP)/s) on the permeate side. At a given inlet gas flow rate Q0, the flow rate and water concentration of the outlet gas (Qout and Yout, respectively) can be obtained by integrating Eqs. (11) and (12) with the following boundary conditions:

0.04 Average water transfer rate (cm3(STP)/cm2.s) Molar fraction H2O in humidified gas

4679

0.03

0.02

Q¼ Q0 and Y¼0 at A ¼0; Q¼ Qout and Y¼ Yout at A ¼16.6 cm2.

0.01

0.00 0.015 0500100015002000

0.010

0.005

0.000 500

0

1000 Gas flow rate (sccm)

1500

2000

Fig. 12. Effect gas flow rate on (a) water vapor concentration in outlet stream and (b) average mass transfer rate of water. Temperature 25 1C. Symbols represent experimental data, and the solid lines are calculated values based on the cross flow model.

Feed side

psat

H2O

Membrane Qvap

Permeate side

pp Humidified N2 outlet, Qgas+Qvap

Dry N2 inlet, Qgas dA

psat

Feed side

Membrane Permeate side Q

pp, dQ, Y

Q + dQ

Fig. 13. Schematic of cross flow model for the membrane humidifier.

with position on the membrane surface because the gas is gradually humidified along its flow direction. As a first approximation, the water vapor transport through the membrane was evaluated on the basis of a simple cross flow model shown in Fig. 13. Assuming a constant pressure on the permeate side, the following equations can be formulated for a differential unit of membrane area dQ ¼ Jvap ðpsat pp YÞdA

ð11Þ

dQ ¼ dðQYÞ

ð12Þ

In both gas dehydration and humidification processes, the driving force for water vapor permeation through the membrane is the difference in its chemical potential between feed and permeate side created by vacuum or purging. If the membrane permeability to water can be considered as the same in both processes, the permeance of saturated water vapor in the membrane can be estimated from Eq 6 to be 5350 GPU, which was used in the calculation of the humidification performance. The water vapor flux and water content in the humidified gas so calculated are also shown in Fig. 12 (solid lines). It appears that at a relatively low gas flow rate, the membrane performance for gas humidification is overestimated. The reasons for the deviation can be attributed to the following: the permeance of saturated water vapor is in principle the same as the liquid water at an equal activity, but this is not always the case as observed experimentally, presumably due to different swelling of the membrane by liquid and vapor. For example, Kataoka et al. (1991) found that homogeneous PAN membranes showed similar permeability between pervaporation and vapor permeation, while cellulose acetate membranes exhibited a higher flux for pervaporation. Hamada et al. (1997) observed a higher solubility and permeability of water in different hydrophilic membranes from liquid phase than from the vapor phase. Will and Lichtenthaler (1992) found a higher flux in pervaporation than vapor permeation for the separation of methanol–water mixtures using poly(vinyl alcohol) membranes. On the other hand, unlike the vacuuminduced pervaporation and vapor permeation where the permeate is withdrawn instantly under vacuum, in gas humidification the micropores in the substrate of the membrane are filled with the gas and the vapor permeant leaving the skin layer needs to diffuse through the gas in the pores of the substrate to reach the bulk gas stream. This imposes an additional resistance to water vapor transport, which may become significant because of the lack of convective flow in the pores. Another possible reason for the deviation is that the driving force for water permeation is considerably different for gas dehydration (under vacuum) and gas humidification, and the water permeance value used in gas humidification calculations may be overestimated. To address the aforementioned issues, a different approach is used here from an engineering perspective via the use of an overall mass transfer coefficient of water in the membrane humidifier. The driving force for humidification can be taken as Dp¼ psat  pv, where psat is the saturated vapor pressure on the liquid side and pv is the partial vapor pressure of water in the gas stream. Dp varies in the humidifier from point to point along gas flow on the membrane surface, and so does the local water flux. For a differential membrane area dA, the local water flux is dW ¼ ðU DpÞdA

ð13Þ 3

where W is the quantity of water vapor (cm (STP)/s) permeated through the membrane to reach the gas on the permeate side, and U is the mass transfer coefficient (cm3(STP)/cm2 s cmHg). To apply this equation to the entire membrane area, a few assumptions can be made for simplifications: (1) isothermal operation, (2) constant flow rate of the sweeping gas to be humidified and (3) negligible variation in gas pressure along the gas flow. As in pervaporation,

ARTICLE IN PRESS J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

dðDpÞ Dp2 Dp1 ¼ WT dW

ð14Þ

where WT is the total quantity of water (cm3(STP)/s) transferred in the membrane humidifier, and Dp1 and Dp2 correspond to Dp (cmHg) at the gas inlet and outlet of the humidifier, respectively. Substituting Eq. (13) into Eq. (14) and integrating yields the following equation WT ¼ UADpLM

ð15Þ

where DpLM is the overall driving force for water transfer expressed as the logarithmic mean of Dp1 and Dp2,

Dp2 Dp1 DpLM ¼ lnðDp2 =Dp1 Þ

ð16Þ

Eq (15) is similar to the widely used heat transfer equation for heat exchangers. It is an important equation for membrane humidifier as it can be used to (1) determine the overall mass transfer coefficient U from measured quantities WT and DpLM, (2) predict humidification performance of a given humidifier and (3) calculate the membrane area required for a given humidification task. The overall mass transfer coefficient for gas humidification was calculated from the experimental data, and the results are presented in Fig. 14. As expected, the mass transfer coefficient is shown to increase with an increase in the gas flow rate until the gas flow rate is sufficiently high. It appears clear that the mass transfer resistance at the gas side of the membrane is not insignificant especially at low gas flows. Tailoring of the substrate structure using such techniques as those reported by Chung and co-workers (Hosseini et al., 2010) toward reduction of its mass transfer resistance is needed to maximize the potential of the membrane for gas humidification. Fig. 15 shows the effects of operating temperature on the mass transfer rate of water through the composite membrane for gas humidification. As the temperature increases, the saturated vapor

Overall mass transfer coefficient (10-3 cm3 (STP)/cm2.s.cmHg)

10

8

6

4

0.05 Molar fraction H2O in humidified gas

the energy needed for liquid evaporation is provided by thermostatted liquid water and thus the process is isothermal. Because of the relatively low water content in the humidified gas, the total molar flow rate of gas on the permeate side of the membrane can thus be regarded as constant. As a result, the accumulative quantity of water transferred to the gas at a certain point is proportional to the partial vapor pressure of water at this point, that is, pv varies linearly with W. Therefore, at a given temperature

0.04 0.03 0.02 0.01 0.00 0.05 20

Average water transfer rate (cm3 (STP)/cm2.s)

4680

30

40

50

30

40 Temperature (°C)

50

0.04 0.03 0.02 0.01 0.00 20

Fig. 15. Effects of temperature on average mass transfer rate of water and water vapor concentration in the humidified stream. Nitrogen gas flow rate 1000 sccm.

pressure in the feed increases, resulting in an increase in the driving force for water permeation. Thus for a given gas flow rate, the water vapor content in the humidified gas increases. It is interesting, however, to note that when the temperature changed from 25 to 45 1C, there is a very little change in the relative humidity of the humidified gas leaving the membrane humidifier (30.8–31.0% RH) at a constant gas inlet flow rate of 1000 sccm. Such a relative humidity level is considered to be very significant because of the small membrane area of 16.6 cm2 used. The effectiveness of the membrane humidifier can be further seen from the fact that once cooled down to ambient temperature (i.e., 23 1C), a humidified gas at 31% RH and 45 1C would lead to full saturation of the gas with water. Interestingly, the overall mass transfer coefficient for humidification varies slightly from 5.03  10–3 to 5.23  10–3 cm3(STP)/(cm2 s cmHg) when the operating temperature increased from 25 to 45 1C. This small variation (i.e., 4%) is within the experimental error, and thus the overall mass transfer coefficient can be considered as constant at a given gas flow rate in the temperature range studied. It further indicates that the increased water transfer rate with temperature is mainly due to increased saturated vapor pressure (i.e., the driving force).

2

4. Conclusions 0 0

500

1000 Gas flow rate (sccm)

1500

2000

Fig. 14. Overall mass transfer coefficient of water through the membrane to reach the gas stream. Temperature 25 1C.

The use of interfacially formed PDMAEMA/PAN composite membranes for applications involving water vapor permeation was investigated, and the performance of the membrane for gas dehydration and gas humidification were evaluated. For gas dehydration, where water vapor/methane mixtures were

ARTICLE IN PRESS J.R. Du et al. / Chemical Engineering Science 65 (2010) 4672–4681

admitted to the feed side at atmospheric pressure and vacuum was applied on the permeate side: (1) the permeance of water vapor through the membrane was found to increase exponentially with an increase in water vapor concentration in the feed, (2) while the PAN substrate had little effect on the permeation of methane; because of the substantially higher permeability of water vapor in the membrane, the resistance of the substrate to water vapor permeation was significant and (3) optimization of substrate structures to reduce its resistance to water vapor permeation is needed in order for the composite membrane to work out its full separation potential. For gas humidification, where liquid water flowed on one side of the membrane and nitrogen gas flowed on the other side, thereby humidifying the nitrogen stream, (1) the water mass transfer rate increased with an increase in the gas flow rate, but the increase in the water transfer rate was less than proportional, resulting in a reduction in the humidity level of the gas stream, (2) an increase in temperature enhance the gas humidification significantly, (3) at a gas flow rate of 1000 sccm, a relative humidity of 30% was achieved at temperatures 25–45 1C with a small membrane area of 16.6 cm2 and (4) a mass transfer equation was derived for membrane humidifiers, which could be used in process design and scale up.

Acknowledgement Research support from the Natural Sciences and Engineering Research Council (NSERC) of Canada is gratefully acknowledged. References Aranda, P., Chen, W.J., Martin, C.R., 1995. Water transport across polystyrenesulfonate/alumina composite membranes. J. Membr. Sci. 99, 185–196. Barrie, J.A., Machin, D., Nunn, A., 1975. Transport of water in synthetic cis-1,4polyisoprenes and natural rubber. Polymer 16, 811–814. Du, R., Chakma, A., Feng, X., 2007. Interfacially formed poly(N,N-dimethylaminoethyl methacrylate)/polysulfone composite membranes for CO2/N2 separation. J. Membr. Sci. 290, 19–28. Du, J.R., Chakma, A., Feng, X., 2008. Dehydration of ethylene glycol by pervaporation using poly(N,N-dimethylaminoethyl methacrylate)/polysulfone composite membranes. Sep. Purif. Technol. 64, 63–70. Du, R., Feng, X., Chakma, A., 2006. Poly(N,N-dimethylaminoethyl methacrylate)/ polysulfone composite membranes for gas separation. J. Membr. Sci. 279, 76–85.

4681

Fu, H., Jia, L., Xu, J., 1994. Studies on the sulfonation of poly(phenylene oxide) (PPO) and permeation behavior of gases and water vapor through sulfonated PPO membranes. II. Permeation behavior of gases and water vapor through sulfonated PPO membranes. J. Appl. Polym. Sci. 51, 1405–1409. Hamada, T., Taya, M., Tone, S., 1997. Pervaporation and vapor permeation behavior of water and 2-propanol in water-selective membranes. J. Chem. Eng. Jpn. 30, 600–608. Hayashi, Y., Sugiyama, S., Kawanishi, T., Shimizu, N., 1999. Kinetics of sorption and permeation of water in glassy polyimide. J. Membr. Sci. 156, 11–16. Hosseini, S.S., Peng, N., Chung, T.S., 2010. Gas separation membranes developed through integration of polymer blending and dual-layer hollow fiber spinning process for hydrogen and natural gas enrichments. J. Membr. Sci. 349, 156–166. Huang, J., Cranford, R., Matsuura, T., Roy, C., 2003. Water vapor permeation properties of aromatic polyimides. J. Membr. Sci. 215, 129–140. Huang, R.Y.M., Feng, X., 1993. Resistance model approach to asymmetric polyetherimide membranes for pervaporation of isopropanol/water mixtures. J. Membr. Sci. 84, 15–27. Hyun, D., Kim, J., 2004. Study of external humidification method in proton exchange membrane fuel cell. J. Power Sources 126, 98–103. Ji, P., Cao, Y., Zhao, H., Kang, G., Jie, X., Liua, D., Liu, J., Yuan, Q., 2009. Preparation of hollow fiber poly(N,N-dimethylaminoethyl methacrylate)-poly(ethylene glycol methyl ether methyl acrylate)/polysulfone composite membranes for CO2/ N2 separation. J. Membr. Sci. 342, 190–197. Kataoka, T., Tsuru, T., Nakao, S., Kimura, S., 1991. Membrane transport properties of pervaporation and vapor permeation in ethanol–water system using polyacrylonitrile and cellulose acetate membranes. J. Chem. Eng. Jpn. 24, 334–339. Landro, L.D., Pegoraro, M., Bordogna, L., 1991. Interactions of polyether– polyurethanes with water vapor and water–methane separation selectivity. J. Membr. Sci. 64, 229–236. Liu, L., Chen, Y., Li, S., Deng, M., 2001. The effect of a support layer on the permeability of water vapor in asymmetric composite membranes. Sep. Sci. Technol. 36, 3701–3720. Liu, L., Jiang, N., Burns, C.M., Chakma, A., Feng, X., 2009. Substrate resistance in composite membranes for organic vapor/gas separations. J. Membr. Sci. 338, 153–160. Mensitieri, G., Del Nobile, M.A., Sommazzi, A., Nicolais, L., 1995. Water transport in a polyketone terpolymer. J. Polym. Sci. B: Polym. Phys. 33, 1365–1370. Metz, S.J., van de Ven, W.J.C., Potreck, J., Mulder, M.H.V., Wessling, M., 2005a. Transport of water vapor and inert gas mixtures through highly selective and highly permeable polymer membranes. J. Membr. Sci. 251, 29–41. Metz, S.J., van de Ven, W.J.C., Mulder, M.H.V., Wessling, M., 2005b. Mixed gas water vapor/N2 transport in poly(ethylene oxide) poly(butylene terephthalate) block copolymers. J. Membr. Sci. 266, 51–61. Schult, K.A., Paul, D.R., 1996. Water sorption and transport in a series of polysulfones. J. Polym. Sci. B: Polym. Phys. 34, 2805–2817. Semenova, S.I., Ohya, H., Soontarapa, K., 1997. Hydrophilic membranes for pervaporation: an analytical review. Desalination 110, 251–286. Stannett, V., Haider, M., Koros, W.J., Hopfenberg, H.B., 1980. Sorption and transport of water vapor in glassy poly(acrylonitrile). Polym. Eng. Sci. 4, 300–304. Will, B., Lichtenthaler, R.N., 1992. Comparison of the separation of mixtures by vapor permeation and by pervaporation using PVA composite membranes. I. Binary alcohol–water systems. J. Membr. Sci. 68, 119–125. Wu, Y., Peng, X., Liu, J., Kong, Q., Shi, B., Tong, M., 2002. Study on the integrated membrane processes of dehumidification of compressed air and vapor permeation processes. J. Membr. Sci. 196, 179–183. Zhao, H., Cao, Y., Ding, X., Zhou, M., Yuan, Q., 2008. Poly(N,N-dimethylaminoethyl methacrylate)-poly(ethylene oxide) copolymer membranes for selective separation of CO2. J. Membr. Sci. 310, 365–373. Zhu, W., Gora, L., van de Berg, A.W.C., Kapteijn, F., Jansen, J.C., Moulijn, J.A., 2005. Water vapor separation from permanent gases by a zeolite-4A membrane. J. Membr. Sci. 253, 57–66.