Accepted Manuscript

Roll-coating of defect-free membranes with thin selective layer for alcohol

permselective pervaporation: From laboratory scale to pilot scale

Jie Li, Naixin Wang, Hao Yan, Guojun Zhang, Zhenping Qin, Shulan Ji, Guojun
Zhang

PII: S1385-8947(15)01740-4
DOI: http://dx.doi.org/10.1016/j.cej.2015.12.068
Reference: CEJ 14578

To appear in: Chemical Engineering Journal

Received Date: 7 October 2015

Revised Date: 2 December 2015
Accepted Date: 17 December 2015

Please cite this article as: J. Li, N. Wang, H. Yan, G. Zhang, Z. Qin, S. Ji, G. Zhang, Roll-coating of defect-free
membranes with thin selective layer for alcohol permselective pervaporation: From laboratory scale to pilot scale,
Chemical Engineering Journal (2015), doi: http://dx.doi.org/10.1016/j.cej.2015.12.068

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 Roll-coating of defect-free membranes with thin selective

layer for alcohol permselective pervaporation: From

laboratory scale to pilot scale

Jie Li, Naixin Wang*, Hao Yan, Guojun Zhang, Zhenping Qin,

Shulan Ji*, Guojun Zhang

Beijing Key Laboratory for Green Catalysis and Separation, College of

Environmental and Energy Engineering, Beijing University of Technology, Beijing

100124, P.R.China

Abstract: Pervaporation offers advantages over traditional separation processes, but

there is still lack of scale-up methods and manufacturing systems to easily and

controllably obtain the defect-free membranes. In this work,

poly(dimethylsiloxane)/polysulfone composite membranes were scaled up using a

roll-coating method. The thickness of the poly(dimethylsiloxane) separation layer was

successfully controlled on polysulfone substrate by roll-coating different layers. The

membranes are used for separating ethanol/water mixtures. The pervaporation

experiments were performed (i) in the laboratory using a radial cell with a membrane

area of 0.0028 m2; (ii) in a large cell using a plate-and-frame module with membrane

area of 0.36 m2; and (iii) in a pilot plant located at an industrial site, with a

plate-and-frame module and total membrane area of 2.16 m2. On-site operations

showed that the flux of 1000~1200 gm-2h-1 with 60~65 wt% ethanol in permeate

* Corresponding authors: wangnx@bjut.edu.cn (N. Wang) and jshl@bjut.edu.cn (S. Ji), Fax:

+86-10-6739 2393.

1
was achieved by the pilot-scale pervaporation facility, which demonstrated technical

feasibility for industrial application.

Keywords: roll-coating; thin selective layer; alcohol permselective pervaporation;

pilot scale; on site operation.

1. Introduction

In recent years, biofuel produced by biological fermentation has been prompted

as a green alternative to fossil fuels [1, 2]. However, due to the inhibition of high

bio-alcohol concentration in the fermentation broth, the preparation of biofuel is

impeded. Therefore, in situ removal and enrichment of bio-alcohol from the

fermentation broth is a key issue to improve the process efficiency [3]. In order to

achieve this goal, pervaporation (PV) has been widely studied, because of its energy

efficiency and harmless to microorganisms [4]. However, efficient stripping of

volatile organic compounds (VOC) from aqueous is more challenging than

dehydration [5-6]. Previous studies by various authors have identified

poly(dimethylsiloxane) (PDMS) with high separation factors for alcohol-water

separations. Despite offering advantages in selectivity, dense PDMS membranes show

limitations in flux and mechanical strength that limit their application. Over the past

decades, the coating of a very thin top layer on a porous substrate has been applied to

increase PV flux. Four main multilayer strategies have been explored: (i) dip coating

[1, 2], (ii) spin coating [3, 4], (iii) spray coating [7, 8] and (iv) doctor blading [9] at a

lab scale. For example, Osorio-Galindo et al. [10] prepared PDMS-PMHS/PEI

2
composite membranes by doctor blading. The membrane area is 78.54 cm2 and the

selective thickness is 50 m. This composite membrane performed a separation factor

of 3.7 and a flux of 270 gm-2h-1 at 5 wt% ethanol feed concentration (40 0C). Lee et

al. [11] fabricated PDMS/PEI hollow fiber composite membranes by spinning. The

membrane area is 903 cm2 and the selective thickness is 0.07 m. The composite

hollow fiber membrane was stable over the long-term (about 160 days) with an

ethanol permeation flux of a total flux of 231-252 gm-2h-1 and a separation factor of

7-9. More recently, we also prepared PDMS/PSf membranes by dip-coating with the

area of 28 cm2 and the selective-layer thickness of 1 m [1]. The PDMS/PSf

composite membrane performed a separation factor of 8.5 and a flux of 828.6

gm-2h-1 at 5 wt% ethanol feed concentration (60 0C). However, the manual operation

involved in dip-coating brings about many uncontrollable human factors, leading to

instability and low productivity of the membrane performance. For such dense

composite membranes, the mass transport across the membranes is governed by a

solution-diffusion mechanism. Therefore, the thickness of selective layer is one of the

most important controllable factors to achieve high permeate production (flux).

However, it is relatively difficult to obtain thin defect-free selective layer. Therefore,

it is necessary to seek for a facile low-cost scale up fabrication approach to meet

industrial requirements.

Besides membrane fabrication, it is also challenging thing to move from

laboratory to pilot plant operations, since increased membrane area requirement,

3
quantity of feedstocks, time for execution, analytical facilities, technical issues, and

operating personnel are associated with such scale up. Key steps in the development

of a membrane process for a separation of interest to move from lab scale to pilot

plant include: (i) Setting membrane performance targets; (ii) Perform a

proof-of-concept test with a small scale test; (iii) testing the membrane in a pilot-plant

module; (iv) showing stable long-term operations [12]. These arguments in favor of

plat-form composite membranes as their modules have been scaled up industrially.

In the present work, we firstly designed a facile roll-coating system in pilot scale

and prepared poly(dimethylsiloxane)/polysulfone (PDMS/PSf) membranes by a

roll-coating method (Figure 1) which has great potential for economical scale up. The

thickness and optimized operation parameters of the membranes were investigated for

improving separation performance and obtaining a stable pervaporation process.

Scanning electron microscopy (SEM) was used to understand the microtopographical

changes the membrane surfaces as a function of processing changes to correlate with

the PV performance. The effect of feed composition, permeate pressure and feed

temperature on transmembrane flux and separation characteristics has been

investigated to optimize the performance. The stability of the membranes was

monitored for 1000 hours. To the best of our knowledge, such a detailed insight in

up-scaling of PDMS membranes for pervaporation has not been reported yet.

Figure 1

2. Experimental

4
2.1 Materials

PDMS with a viscosity of 2550 mPas was purchased from China Bluestar

Chengrand Chemical Co. Ltd. (China). Tetraethyl silicate (TEOS) and ethanol were

obtained from Beijing Chemical Company (China). Dibutyltin dilaurate, n-heptane

and n-butanol were supplied by Tianjin Fuchen Chemical Reagents Company

(China). In our experiments, all reagents were of analytical grade and were used

without further purification.

2.2 Preparation of lab scale and pilot-scale PDMS/PSf composite membrane

The PSf substrate was soaked for 24 h using pure water. The pores of the

supports were filled with water to minimize penetration of the coating polymer

solution into the pores; and allowed to dry for 30 min to remove excess water on the

support surface before roll-coating the polymer solution [2]. PDMS was dissolved in

n-heptane to form a 10 wt% PDMS solution. After the polymer solutions were stirred

at room temperature for 1 h, the cross-linking agent TEOS and the catalyst dibutyltin

dilaurate were added into the polymer solution (WPDMS:Wn-heptane:WTEOS:Wdibutyltin

dilaurate = 1:10:0.1:0.005), and the resulting mixture was continuously stirred for 3 h

[2]. Air bubbles trapped in the polymer solution were removed by degassing at 100 Pa

for 10 min. The polymer solution was poured into a feed tank, which was placed

under a roller, and roll-coated on the surface of the PSf support with lab scale or

pliot-scale roll-coating system (Figure S1). To produce defect-free membranes,

removal of the solvent at room temperature for 12 h and subsequent curing in the

5
oven at 80 C for 12 h was used to fabricate the PDMS/PSf composite membranes in

a low humidity enclosure.

2.3 Membrane characterization

SEM photographs of PDMS/PSf membranes were taken using a scanning

electron microscope (Hitachi-4300, Japan). The membranes were fractured in liquid

nitrogen. Membrane surfaces and liquid nitrogen fractured cross section were coated

with a conductive layer of sputtered gold to make it conductive. X-ray diffraction

(XRD) experiments were conducted on an X-ray diffractometer (D8 ADVANCE,

Bruker/AXS, Germany).

2.4 Pervaporation experiments

The specification of the composite PDMS/PSf membrane module is given in

Table 1.

Table 1

2.4.1 Laboratory test

The pervaporation performance of the membrane was evaluated using a

pervaporation apparatus fabricated in our laboratory [13, 14]. The selected ethanol

concentration in the feed solution was ~10 wt% and the experiments were conducted

at 60 C. The downstream pressure was maintained at 200~300 Pa and the

pervaporation performance was measured three times by using three membranes

under the same fabrication conditions with the average of the three trials used as a

data point. The permeate vapor was trapped in liquid nitrogen and subsequently

6
analyzed by gas chromatography (GC-14C, Shimadzu, Japan). The permeate flux (J)

was determined according to the following equation:

W
J= (2)
At
where W is the weight of the liquid collected in the cold traps, A is the effective area

of the membrane, and t is the collection time for the pervaporation.

The selectivity of the composite membranes is expressed as a separation factor ,

which is defined as

YEtOH / YH 2O
HEtOH
O = (3)
2
X EtOH / X H 2O

where YEtOH , YH 2O , X EtOH and X H 2O are the weight ratios of ethanol and water

in the permeation and feed sides, respectively.

2.4.2 Pilot test

A pervaporation facility with a total membrane area of about 0.36 m2 was built in

our lab. The membrane module used in this study for evaluation of its performance

was a stainless-steel plate-and-frame configuration module using 6 pieces of plat

sheet PDMS/PSf membranes with 0.06 m2. This membrane module is equipped with a

total membrane area of 0.36 m2. The operating temperature was maintained at 60 C

throughout the experiment with an upstream pressure at the atmospheric pressure

(101.325 kPa) and downstream pressure controlled by a vacuum pump.

A similar pilot scale module was also prepared with 36 pieces of flat sheet

PDMS/PSf membranes and a total membrane area of about 2.16 m2. The schematic

flow inside of the membrane module is shown in Figure S2. The apparatus was

7
located in a factory for ethanol production. A schematic of the pervaporation

separation test unit is shown in Figure 2. The permeate components were trapped in

collection vials, which were cooled by a chiller at -15 ~ -10 C.

Figure 2

3. Results and Discussion

3.1 Roll-coating system and roll-coated membranes

Figure 1 is a schematic diagram of roll-coating system for the preparation of

composite membranes. Referring to Figure 1, the apparatus includes three major parts,

such as the support plate, polymer solution tank and roller. The polymer solution tank

placed on a lift was fixed on the base. The roller with a motor was fixed on the

support plate. Before roll-coating process, the PSf substrate was attached to the roller

and fixed by snaps to keep the substrate smooth and flat. Then, the polymer solution

was poured into polymer solution tank. The tank was raised by a lift to make the

solution contact with PSf substrate. After that, the motor was turned on and the roller

rotated with the driving of motor through shaft and gear. After rotating several circles,

the motor was turned off and the polymer solution tank was descended. The snap was

unsnatched from slot to take the composite membrane from the roller. Due to the

sensitive of the preparation process to the environment, the system was used at a

space with under-controlled humidity and temperature. Photograph of the system is

presented in Figure S1.

By using the newly developed roll-coating system, PDMS/PSf membranes were

8
easily fabricated. Figure 3 shows the cross section and surface morphologies of

PDMS/PSf membranes with different roll-coating cycles. As shown in Figure 3 (a), a

thin layer of 1 m could be obtained by using a roll-coating technique. With

increasing coating cycles from 1 to 60 (Figure 3(b-e)), the thickness of selective layer

increased from 3 m to 5 m (Figure S3). Obviously, it was also noted that the

thickness of the selective layer could be easily tuned by adjusting the roll-coating

cycles. In addition, the surface morphology of PDMS/PSf membrane with 30 coating

cycles was shown in Figure 3(f). It can be seen that a uniform and defect-free surface

was formed on the substrate. Further evidences were provided by XRD spectra of

membranes with different coating cycles. As shown in Figure 4, PDMS membranes

exhibited sharp reflections at 2 of 13.96, and 21.48, which was very similar to the

results of earlier research [15]. It also can be observed that as coating cycles increased

from 10 to 60, the intensities of diffraction peaks are weakened gradually.

Figure 3

Figure 4

In pervaporation, the permeate flux is inversely proportional to the membrane

thickness, and the selectivity may be dependent on the membrane thickness to get a

defect-free membrane. To optimize performance, membranes with different thickness

were fabricated and used to separate the ethanol/water mixture (Figure 5). Clearly, the

selectivity showed an increase followed by a decrease. For example, the separation

factor obtained from rotating 1 cycle and 30 cycles were 6.4 and 11.6, respectively. It

9
was observed that when the membrane thickness increased, the flux decreased from

2936 gm-2h-1 to 1493 gm-2h-1. This suggested that the surface pores of PSf support

may not be completely covered with PDMS within a ultrathin layer, while after

roll-coating with 30 cycles, a defect-free PDMS selective layer could be formed.

However, for example, the separation factor decreased from 11.6 to 7.3 with the total

flux decreased from 1493 gm-2h-1 to 1357 gm-2h-1 when coating cycles increased

from 30 to 50. This indicated that the excessive coating would not benefit for

pervaporation performance and may reflect intrusion into the support pores with

increasing coating cycles. As a result, both the separation factor and the total flux

decreased when the coating cycles exceeded 30 layers. Therefore, the appropriate

coating cycles was selected as 30.

Figure 5

For the PDMS with hydroxyl terminal groups, humidity is an important factor

for PDMS pre-crosslinking and coating. Figure 6 shows the pervaporation

performance of PDMS/PSf membranes prepared in 20%, 30%, 40% relative humidity

(RH) at room temperature. Clearly, the separation factor was dramatically improved

in lower RH. The ethanol permeate composition decreased from 56.3% to 42.3% with

a flux increase from 1493 gm-2h-1 to 1921 gm-2h-1, when the humidity for

preparation increased from 20% to 40%. Therefore, the sorption of water from the

environment could negatively affect the selectivity of the resulting membrane. In the

subsequent experiments, the humidity was strictly controlled as 20%.

10
 Figure 6

In order to investigate the uniform defect-free character within a large area

membrane, a series of laboratory scale modules were prepared with membranes from

different positions (through side or middle of the pilot-scale membranes), which were

cut from a 600 cm2 area membrane (Figure S4). There was no obvious difference

specific to a particular position. From Table 2, it can be seen that the difference in

separation factor and flux was 3% at any position within the membranes.

Table 2

3.2 Scale up of pilot scale module and pervaporation performance

The pervaporation performance of the pilot scale PDMS/PSf membrane was

investigated in our laboratory to study the performance of a plate-and-frame module

with a membrane area of 0.36 m2. As the fermentation processes of corn and sugar are

often conducted in a relatively hermetic container, the ethanol generated from the

fermentation industry may approach a constant concentration in a period which

needed to be removed, so the membrane performance as a function of ethanol

concentration was investigated. As shown in Figure 7, the ethanol in feed tank (10 L

liquid mixtures initially) decreased from 10.44 wt% to 1.03 wt% with the

pervaporation operation in 19.5 h, while the ethanol in permeate decreased from 47

wt% to 17.8 wt%, with the flux suffers a reduction from ~2000 gm-2h-1 to ~1000

gm-2h-1. This observation also indicates that the membrane is efficient for the

treatment of solvent/water waste.

11
 Figure 7

The effects of feed flow rate on the PV performance of PDMS/PSf membrane

are shown in Table 3. As the flow rate increased from 300 L/h to 800 L/h, the flux

increased from 1081 gm-2h-1 to 1981 gm-2h-1, while the separation factor increased

from 7.3 to 8.3. Based on the results, it can be seen that separation performance of the

membrane improved when the flow rate increased. It can be attributed that the

concentration and temperature boundary layers can be minimized by higher flow rates.

The concentration and temperature polarization thus weakened to improve the

separation behavior. However, the ultrahigh flow velocity could destroy the selective

layers under the long-time running conditions. Moreover, the pumping costs will arise

from a corresponding high feed flow rate. Therefore, the moderate feed flow rate

should be considered in the real industry.

Table 3

As pervaporation is a separation process in which a liquid feed comes into

contact with one side of a membrane while vacuum is applied on the other side to

produce a permeate vapor, the downstream pressure creates the driving force for mass

transport across the membrane. Typically, the downstream pressure is kept as low as

possible (most-often reported under vacuum conditions of 0-100 Pa) to maximize the

driving force in academic research. In fact, it is wise to examine the pervaporation

process at relatively higher downstream pressures (up to several thousand Pascal),

because of the technical difficulty of achieving and keeping relatively low vacuum in

12
industrial practice. The effects of downstream pressure on membrane separation were

shown in Table 4. Interestingly, when the downstream pressure increased from 200 Pa

to 2000 Pa, the flux decreased from 1981 gm-2h-1 to 832 gm-2h-1, while the

separation factor increased from 8.3 to 10.5. Similar phenomenon was also found by

other researches [16]. Thus, the downstream pressure was suggested to be kept under

1000~2000 Pa to obtain both high separation factor and flux.

Table 4

To study the stability of the roll-coated membranes, the pilot scale module was

tested using 10 wt% ethanol feed over a period of 1000 h. The membranes flux and

ethanol in permeate are plotted in Figure 8. The fluxes were 1000~2000 gm-2h-1,

while the ethanol in permeate was 46~58 wt.%. This trend indicated that the alcohol

in permeate changed slightly, while the total flux increased with time over a period of

1000 h. Thus, the roll-coated membranes could move forward to the production of

pilot scale module.

Figure 8

Besides ethanol, butanol has also been regarded as one promising biofuels to

meet the needs of sustainable and green energy systems. However, butanol is highly

toxic to the fermenting microorganism, resulting in low product concentration in the

fermentation process. The highest n-butanol productivity by the strain fermentation

reported in literature is 3.0 wt% [17, 18]. Therefore, the pervaporation performance

with n-butanol/water at different concentration below 3 wt% was conducted. As

13
shown in Figure 9, the ethanol in feed tank (10 L liquid mixtures initially) decreased

from 2.9 wt% to 0.05 wt% (500 ppm) with the pervaporation operation in 20 h, while

the n-butanol in permeate decreased from 49.4 wt% to 1.4 wt%, with the flux suffers a

reduction from ~1300 gm-2h-1 to ~600 gm-2h-1. In the case of the pervaporation of

the 2.9 wt% n-butanol mixture, the n-butanol content in the permeate and permeate

flux can reach 49.4% ( = 27.3) and 1370 gm-2h-1 (60 0C), respectively. It is notable

that the roll-coating PDMS/PSf membrane also shows excellent pervaporation

performance for the other solvent-water mixtures.

Figure 9

3.3 On-site long-term stability of pilot-scale modules

Based on the laboratory studies of roll-coated membranes, a pilot scale

pervaporation module with 2.16 m2 membrane area was conducted in a factory for

1000 h on-site operation. Figure 10 shows a view of the ethanol perm-selective

pervaporation pilot plant in on-site operation. The horizontal cylinder contained the

membrane module under vacuum, while the permeate condenser was used to convert

gas-phase permeate to liquid, which then pooled in the bottom and was pumped out.

In order to reduce the total energy consumption for condensation and improve the

ethanol contents in permeate, an unique design for condensation system was

conducted and similar design was also used in other research [19]. As shown in

Figure 2, three permeate condensers were cascade connected with the separate

permeate gathering tank. The first two condensers were installed before vacuum pump

14
with a negative pressure inside, while the third condenser was installed after vacuum

pump with a atmospheric pressure. The condensing temperature of these three

condensers were separately controlled through regulating the amount of condensing

agent. The ethanol and water vapor was expected to condense in the different

condensers according to their different condensation temperature. In this on-site

operation, the downstream pressure was controlled in the range of 1000~1200 Pa. In
B
light of Antoine equation ( lg P = A ), the boiling point for ethanol and water is
t +C
-11.2 C~-8.8 C and 7.0 C~ 9.7 C respectively under 1000~1200 Pa. Therefore, the

condensation temperature for the first and second condenser was set as -5 C~0 C

and -15 C~-10 C respectively, resulting to part of water vapor was condensed

within the first condenser while most of ethanol vapor was condensed within the

second condenser. The third condenser was used to collect the residual components to

reduce waste.

In addition, the feed flow rates, heat input, and vacuum were controlled to obtain

the optimized performance by varying the process conditions. Typically each set of

process conditions were held constant for 24 h or more, with feed and permeate

samples taken every 3-4 h for analyses. The appropriate over-all operation conditions

are shown in Table 5. The pervaporation process was running continuously for 1000 h

and the change of operating conditions were shown in Figure S5. Because of the pilot

plant was placed in the outdoor, the operating conditions was significant affected by

the weather, especially for temperature. However, even though the operating

15
conditions were unsteady, the pilot scale pervaporation module showed an excellent

separation performance, with ethanol content in permeate up to 60~65 wt% and the

flux as high as 1.0~1.2 kgm-2h-1 (Figure 11). There was no indication of

performance decline, in either flux or selectivity.

Figure 10

Figure 11

Table 5

Table 6 compares the PV performance for the ethanol-water system of PDMS

composite membranes prepared by various methods reported in the literature. Our

composite membrane with laboratory scale had a flux of 1493 gm-2h-1 and a

selectivity of 11.6 for ethanol to water in the 10 wt% ethanol solution at a feed

temperature of 60 0C. The pilot-scale membrane modules showed comparative

separation performance with those in lab test. It indicated that the roll-coating method

is easily to scale-up for preparing composite membranes with good uniformity. The

composite membrane have excellent ethanol permselectivity in a relatively wide range.

Given these results, the roll-coated membrane are considered to be sufficiently robust

for an economic process. Considering the large scale of the module, 1000 h long-time

stability, and together with the benefits of easy membrane fabrication, the roll-coating

membrane is a promising candidate for application in membrane preparation.

Table 6

16
4. Conclusions

In this study, a pilot-scale PDMS/PSf membrane was successfully fabricated

using a roll-coating method. The prepared membranes were used for lab-scale and

pilot-scale pervaporation. The membranes showed high and stable pervaporation

selectivity for ethanol, which is attributed to the controlling coating layer and relative

humidity for preparation of the membranes. Batch experiments were first carried out

at these roll-coating membranes with a separation factor of 11.6 and a total flux of

1493 gm-2h-1, which are excellent compared with the corresponding values for the

PDMS/PSf membrane prepared by other methods. The pervaporation performance of

the as-prepared membrane was then successfully evaluated under various operating

conditions in a pilot-scale module (0.36 m2) in laboratory and a pilot-scale plant (2.16

m2) located in a factory. The membrane shows robust performance in terms of flux

and separation factor under 1000 h operation during the pilot-scale evaluations, which

proved the technical feasibility of the pervaporation process.

Acknowledgements

The authors gratefully acknowledge the support of the National High

Technology Research and Development Program of China (No.2012AA03A607, No.

2015AA03A062), National Natural Science Foundation of China (No. 21176007, No.

21406006), and Science and Technology Program of Beijing Municipal Education

Commission (KM201510005010). The authors thank Tsinghua University, Nanjing

Tech University, Jiusi Co., LTD and Shandong megavision Membrane Technology &

17
Engineering Co., LTD for the support in the pilot test. The authors thank Wei Zhang,

Hongwei Fan, Yonggang Zhang, Lin Wang, Guixiong Shi and Hongpan Shen for the

assistance in the on-site long-term stability of pilot-scale modules.

References

[1] J. Li, S. Ji, G. Zhang, H. Guo, Surface-modification of poly(dimethylsiloxane)

membrane with self-assembled monolayers for alcohol permselective
pervaporation, Langmuir, 29 (2013) 8093-8102.
[2] F. Xiangli, Y. Chen, W. Jin, N. Xu, Polydimethylsiloxane (PDMS)/ceramic
composite membrane with high flux for pervaporation of ethanolwater mixtures,
Ind. Eng. Chem. Res., 46 (2007) 2224-2230.
[3] C.A. Page, A.E. Fouda, T. Matsuura, Pervaporation performance of
polyetherimide membranes spin- and dip-coated with polydimethylsiloxane, J.
Appl. Polym. Sci., 54 (1994) 975-989.
[4] J.D. Le Roux, D.R. Paul, Preparation of composite membranes by a spin coating
process, J. Membr. Sci., 74 (1992) 233-252.
[5] T. Zeng, A. J. McCabe, T. C. Frank, J. K. Carpenter, W. A. Arnold, E. L. Cussler,
Membrane-assisted volatile organic compound removal from aqueous acrylic
latex is faster than from aqueous solutions, Ind. Eng. Chem. Res., 53 (2014)
12420-12427.
[6] B. Ulrich, T. C. Frank, A. McCormick, E.L. Cussler. Membrane-assisted VOC
removal from aqueous acrylic latex, J. Membr. Sci., 452 (2014) 426-432.
[7] H. Tang, S. Ji, L. Gong, H. Guo, G. Zhang, Tubular ceramic-based multilayer
separation membranes using spray layer-by-layer assembly, Polym. Chem., 4
(2013) 5621-5628.
[8] H. Fan, Q. Shi. H. Yan, S. Ji, J. Dong, G. Zhang, Simultaneous spray
self-assembly of extremely high loading ZIF-8PDMS nanohybrid membrane
exhibiting exceptionally high biobutanol permselective pervaporation, Angew.

18
 Chem. Int. Ed., 53 (2013) 5578-5582.
[9] H. Zhou, R. Shi, W. Jin, Novel organicinorganic pervaporation membrane with a
superhydrophobic surface for the separation of ethanol from an aqueous solution,
Sep. Purif. Technol., 127 (2014) 61-69.
[10] M. Osorio-Galindo, A. Iborra-Clar, I. Alcaina-Miranda, A. Ribes-Greus,
Characterization of poly(dimethylsiloxane)-poly(methyl hydrogen siloxane)
composite membranes for organic water pervaporation separation, J. Appl. Polym.
Sci., 81 (2001) 546-556.
[11] H.-J. Lee, E.J. Cho, Y.-G. Kim, I.S. Choi, H.-J. Bae, Pervaporative separation of
bioethanol using a polydimethylsiloxane/polyetherimide composite hollow-fiber
membrane, Bioresour. Technol., 109 (2012) 110-115.
[12] L.S. White, Development of large-scale applications in organic solvent
nanofiltration and pervaporation for chemical and refining processes, J. Membr.
Sci., 286 (2006) 26-35.
[13] J. Li, G. Zhang, S. Ji, N. Wang, W. An, Layer-by-layer assembled nanohybrid
multilayer membranes for pervaporation dehydration of acetonewater mixtures,
J. Membr. Sci., 415416 (2012) 745-757.
[14] G.J. Zhang, J. Li, S.L. Ji, Self-assembly of novel architectural nanohybrid
multilayers and their selective separation of solvent-water mixtures, AlChE J., 58
(2012) 1456-1464.
[15] L. Qu, G. Huang, Q. Wang, Z. Xie, Effect of diphenylsiloxane unit content on
aggregation structure of poly(dimethylsiloxane-co-diphenylsiloxane), J. Polym.
Sci., Part B: Polym. Phys., 46 (2008) 72-79.
[16] P. Sukitpaneenit, T.S. Chung, PVDF/nanosilica dual-layer hollow fibers with
enhanced selectivity and flux as novel membranes for ethanol recovery, Ind. Eng.
Chem. Res., 51 (2012) 978-993.
[17] A. Kujawska, J. Kujawski, M. Bryjak, W. Kujawski, ABE fermentation products
recovery methodsA review, Renewable Sustainable Energy Rev. 48 (2015)
648-661.
19
[18] C. Jin, M. Yao, H. Liu, C. F. Lee, J. Ji, Progress in the production and application
of n-butanol as a biofuel, Renewable Sustainable Energy Rev., 15 (2011)
4080-4106.
[19] L. Lin, Y. Kong, J. Yang, D. Shi, K. Xie, Y. Zhang, Scale-up of pervaporation
for gasoline desulphurization: Part 1. Simulation and design, J. Membr. Sci., 298
(2007) 1-13.
[20] X.L. Han, L. Wang, J.D. Li, X. Zhan, J.A. Chen, J.C. Yang, Separation of
ethanol from ethanol/water mixtures by pervaporation with silicone rubber
membranes: effect of silicone rubbers, J. Appl. Polym. Sci., 119 (2011)
3413-3421.
[21] J.X. Guo, G.J. Zhang, W. Wu, S.L. Ji, Z.P. Qin, Z.Z. Liu, Dynamically formed
inner skin hollow fiber polydimethylsiloxane/polysulfone composite membrane
for alcohol permselective pervaporation, Chem. Eng. J., 158 (2010) 558-565.
[22] R. Wang, L. Shan, G. Zhang, S. Ji, Multiple sprayed composite membranes with
high flux for alcohol permselective pervaporation, J. Membr. Sci., 432 (2013)
33-41.
[23] X. Zhan, J.D. Li, J.Q. Huang, C.X. Chen, Enhanced pervaporation performance
of multilayer PDMS/PVDF composite membrane for ethanol recovery from
aqueous solution, Appl. Biochem. Biotechnol., 160 (2010) 632-642.
[24] Y. Luo, S.J. Tan, H. Wang, F.W. Wu, X.M. Liu, L. Li, Z.B. Zhang, PPMS
composite membranes for the concentration of organics from aqueous solutions by
pervaporation, Chem. Eng. J., 137 (2008) 496-502.
[25] L. Li, Z.Y. Xiao, S.J. Tan, P. Liang, Z.B. Zhang, Composite PDMS membrane
with high flux for the separation of organics from water by pervaporation, J.
Membr. Sci., 243 (2004) 177-187.
[26] W. Wei, S. Xia, G. Liu, X. Dong, W. Jin, N. Xu, Effects of polydimethylsiloxane
(PDMS) molecular weight on performance of PDMS/ceramic composite
membranes, J. Membr. Sci., 375 (2011) 334-344.
[27] J. Liu, J. Chen, X. Zhan, M. Fang, T. Wang, J. Li, Preparation and
20
characterization of ZSM-5/PDMS hybrid pervaporation membranes: Laboratory
results and pilot-scale performance, Sep. Purif. Technol., 150 (2015) 257-267.

21
Figure Captions

Figure 1. The apparatus of roll-coating system for the preparation of composite

membranes. 1-PSf substrate, 2-Roller, 3-Slot, 4-Snaps, 5-Polymer solution, 6-Polymer

solution tank, 7- Base, 8- Lifts, 9- Motor, 10- Shaft, 11- Gear, 12- Support plate (a),

13- Support plate (b).

Figure 2. Schematic diagram of pilot scale pervaporation apparatus.

Figure 3. Images of the cross-section of composite membranes with different coating

layers (a) PSf substrate, (b) 1 layer, (c) 10 layers, (d) 30 layers, (e) 60 layers and (f)

surface of composite membrane with 30 layers.

Figure 4. The XRD spectra of PDMS membranes with 10, 30 and 60 coating layers

Figure 5. Pervaporation performance versus number of roll-coating layers.

Figure 6. Effects of relative humidity on pervaporation performance of PDMS/PSf

membranes

Figure 7. Effects of ethanol concentration on pervaporation performance of PDMS

membranes (Experimental conditions: downstream pressure of 200 Pa, at 60 C)

Figure 8. Long-term test of a pilot-scale module in lab (experimental conditions: feed

of 10 wt% ethanol and flow rate of 800 L/h, downstream pressure of 1000~2000 Pa,

at 60 C)

Figure 9. Effects of n-butanol concentration on pervaporation performance of PDMS

membranes (Experimental conditions: downstream pressure of 200 Pa, at 60 C)

Figure 10. Pervaporation equipment of pilot scale located in a factory for demo

22
operation

Figure 11. Long-term test of a pilot-scale module for alcohol recovery from a feed

containing 9~10 wt% ethanol (a) Ethanol content in permeate, (b) Permeate flux.

23
 Table 1 Specification of the PDMS/PSf composite pervaporation membranes
Membrane Specification Dimension
Laboratory cell Coating layer 1~60
Thickness of selective layer 1~5 m
Number of sheets 1
Effective area of module 0.0028 m2
Pilot plant in lab Coating layer 30
Thickness of selective layer ~4 m
Number of sheets 6
Effective area of module 0.36 m2
Pilot plant in factory Coating layer 30
Thickness of selective layer ~4 m
Number of sheets 36
Effective area of module 2.16 m2

24
 Table 2 Pevaporation performance of the laboratory scale modules produced from
different position of the pilot scale membranes
Position Ethanol in permeate Separation factor Flux
(%) () (gm-2 h-1)
1 middle 57.05 12.0 1470
2 middle 55.59 11.3 1514
3 side 56.57 11.7 1485
4 side 55.96 11.4 1432
5 side 56.96 11.9 1521
6 side 55.19 11.1 1560

(Experimental conditions: feed of 10 wt% ethanol and downstream pressure of 200 Pa, at 60 C)

25
 Table 3 Effects of flow rates on membranes performance
Flow rate of feed Ethanol in permeate Separation Flux
(L/h) (%) factor () (g m-2h-1)
300 44.8 7.3 1081
500 46.5 7.8 1442
800 47.0 8.0 1981

(Experimental conditions: feed of 10 wt% ethanol and downstream pressure of 200 Pa, at 60 C)

26
 Table 4 Effects of downstream pressure on membranes performance
Downstream pressure Ethanol in permeate Separation Flux
(Pa) (%) factor () (gm-2h-1)
200 47.0 8.0 1981
1000 48.7 8.5 1166
2000 51.0 10.5 832

(Experimental conditions: feed of 10 wt% ethanol and flow rate of 800 L/h, at 60 C)

27
 Table 5 The parameters of pervaporation operation for a pilot scale module
Parameters Operation parameters

350~450
Flow rateL/h

Downstream pressure (Pa) 1000~1200

Cold trap temperature (C) -15~-10
Feed temperature (C) 55~65
Ethanol in feed (wt%) ~10

28
 Table 6 Comparisons of PDMS composite membranes performance with different preparation methods
Membrane Membrane Membran Selective layer Operating Downstream Feed concentration Flux Separation Temperature Ref.
-2 -1
preparation e area thickness time pressure (ethanol/water) (g m h ) factor (C)
method (cm2) (m) (h) (Pa) ()
PDMS/PVDF Casting 22.4 10 - ~100 5 wt% 500.7 8.3 40 [20]
PPMS/PVDF Casting 22.4 10 - ~100 5 wt% 450.1 9.4 40 [20]
PDMS/PSf Cross-flow - - - ~100 8 wt% 265 6.4 50 [21]
casting
PDMS/PSf Multiple 24.6 <1 - 200 5 wt% 3275 7.5 60 [22]
spraying
PDMS/PVDF Casting - 10 - 200 5 wt% 450 15 60 [23]
PDMS-PMHS/P Casting 78.54 50 - 933 5 vol% 270 3.7 40 [10]
EI
PDMS/PEI Spining 903 0.07 ~3800 667 5 wt% 231~252 6~9 - [11]
PDMS/CA Casting 224 1~2 - 667~1334 5 wt% 1139.6 9.3 40 [24]
PDMS/CA Casting 240 8 - 266 5 wt% 1300 8.5 40 [25]
PDMS/PSf Dip-coatin 28 1 - 200 5 wt% 828.6 8.5 60 [1]
g
PDMS/Ceramic Dip-coatin - 5 720 <500 5 wt% 16500~1750 7~8.9 40 [26]
g 0
ZSM/PDMS Casting 189000 8.9 1000 2300 10 wt% 1170 13.5 60 [27]
PDMS/PSf Roll-coatin 28 4 - 200 10 wt% 149345 11.60.3 60 This
g study
PDMS/PSf Roll-coatin 3600 4 1000 1000~2000 10 wt% 1000~2000 7.7~12.4 60 This
g study
29
PDMS/PSf Roll-coatin 216000 4 1000 1000~1200 10 wt% 1000~1200 13.5~16.7 55~65 This
g study

30
Figure 1

31
Figure 2

32
Figure 3

33
Figure 4

34
Figure 5

35
Figure 6

36
Figure 7

37
Figure 8

38
Figure 9

39
Figure 10

40
Figure 11

41
 PDMS/PSf composite membranes were scaled up using a roll-coating method.
The thickness of the selective layer was successfully controlled by roll-coating.
The membrane shows robust PV performance during the pilot-scale evaluations.

42