Professional Documents
Culture Documents
XU YIMING
(B.Eng. Dalian University of Technology)
A THESIS SUBMITED
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
DEPARTMENT OF CHEMICAL AND
BIOMOLECULAR ENGINEERING
NATIONAL UNIVERSITY OF SINGAPORE
2017
Supervisor:
Professor Neal Tai-Shung Chung
Examiners:
Associate Professor Jianwen Jiang
Associate Professor Jianping Xie
Professor Xianshe Feng, University of Waterloo
DECLARATION
I hereby declare that this thesis is my original work and it has been written by
This thesis has also not been submitted for any degree in any university
previously.
Xu Yiming
21 August 2017
i
ACKNOWLEDGEMENTS
First of all, I would like express my sincere appreciation and token of thanks
learned a lot from him not only in terms of research but also in terms of hard
days.
I would also like to gratefully acknowledge my mentor, Dr. Le Ngoc Lieu and
Dr. Zuo Jian who gave me their unselfish encouragements and detailed
pervaporation group including Dr. Ong Yee Kang, Dr. Shi Guimin, Dr. Tang
Yupan, Dr. Hua Dan and Mr. Peyman Salehian for sharing their research
other research staffs and students in the research group who provided me their
kind help and shared with me all the joys and sorrows. Their company made
i
I would like to sincerely thank my thesis advisory committee (TAC) members,
Prof. Jiang Jianwen and Prof. Xie Jianping, who gave me their valuable
I would also like extend my sincere gratitude to NUS for providing the
Last but not the least, I want to express my thanks to my parents and family
members from the core of my heart. Thanks for their love and care during the
past twenty-six years. No matter what happens, I know they are always
Because of them, I could be able to put all my efforts in research without any
ii
Table of Contents
ACKNOWLEDGEMENTS ................................................................................ i
SUMMARY…………………………………………………………………viii
iii
CHAPTER 3: EXPERIMENTAL ................................................................... 42
iv
3.5.10 Positron annihilation lifetime spectroscopy (PALS) ................. 51
v
5.3.5 Performance benchmarking and stability of C-TR membranes . 98
vi
7.3.4 Pervaporation performance ...................................................... 148
vii
SUMMARY
close boiling point mixtures due to its advantages of low energy consumption,
performance and good stability have been developed for alcohol dehydration
via pervaporation.
performance with the highest water permeability of 0.322 mg m-1 h-1 kPa-1,
which is 2-6 times higher than other three membranes and comparable mole-
based selectivity of 685. Long-term tests reveal the good operating stability of
viii
and positron annihilation lifetime spectroscopy measures were also carried
show that the incorporation of UiO-66 greatly improves the normalized flux of
the MMMs have the water permeability of 0.329 and 0.292 mg m-1 h-1 kPa-1
systematically studied.
ix
Last but not the least, MMMs consisting of two different functionalized UiO-
and at the same time enhance separation factor within the particle loading
x
LIST OF TABLES
Table 4.1 Densities and water contact angles of PESU, T-PESU, PPSU and H-
Table 4.2 Solvent uptake results for PESU, T-PESU, PPSU and H-PESU
membranes ............................................................................................................. 63
Table 4.3 PALS data of dry PESU, T-PESU, PPSU and H-PESU membranes .... 64
Table 4.4 PALS data of wet PESU, T-PESU, PPSU and H-PESU membranes .... 65
................................................................................................................................ 71
isopropanol dehydration......................................................................................... 99
Table 6.1 Solvent uptake analyses for the pristine PI membrane and MMMs .... 117
Table 6.2 PALS data for the pristine PI membrane and MMMs ......................... 118
xi
Table 6.6 A comparison of pervaporation performance of membranes for n-
Table 7.1 Solvent uptake ratios of the pristine PI membrane and UiO-66-NH2
Table 7.2 Comparison of solvent uptake ratio among different UiO-66 MOF based
Table 7.3 PALS data of the pristine PI membrane and UiO-66-NH2 based MMMs
.............................................................................................................................. 147
Table 7.4 Comparison of PALS data between UiO-66-NH2 and UiO-66-F4 based
.............................................................................................................................. 156
LIST OF FIGURES
xii
Figure 2.2 Schematic diagram of pore flow model................................................ 21
imide) ..................................................................................................................... 44
Figure 4.1 Chemical structures of PESU, T-PESU, PPSU and H-PESU .............. 59
Figure 4.2 TGA curves of PESU, T-PESU, PPSU and H-PESU membranes ....... 61
Figure 4.3 XRD patterns of PESU, T-PESU, PPSU and H-PESU membranes..... 61
Figure 4.4 Comparison of FFVs at dry and wet states measured by PALS........... 66
Figure 4.5 (a) Water and (b) ethanol/methanol normalized flux of H-PESU
Figure 4.8 Long-term stability of the PESU and H-PESU membranes for ethanol
Figure 5.5 (a) Top-view 3D TGA-FTIR analyses of the polyimide precursor PI-
10-0 and (b) Heating time of TGA and weight percentage vs. temperature
correlation .............................................................................................................. 91
xiii
Figure 5.6 (a) Top-view 3D TGA-FTIR analyses of the polyimide precursor PI-9-
1 and (b) Heating time of TGA and weight percentage vs. temperature correlation
................................................................................................................................ 91
Figure 5.7 Solvent uptake analyses for PI and C-TR membranes ......................... 93
60 ºC....................................................................................................................... 96
dehydration at 60 ºC ............................................................................................... 99
morphologies of (b) the pristine PI membrane and MMMs containing (c) 10wt%
Figure 6.3 (a) SEM image and (b) EDX mapping of Zr on the cross-section of
Figure 6.4 (a) Top-view of 3D TGA-FTIR analyses of UiO-66 and (b) its
Figure 6.5 (a) Top-view of 3D TGA-FTIR analyses of the MMM containing 20wt%
UiO-66 and (b) its temperature vs. weight percentage and heating time............. 115
Figure 6.6 TGA curves of the pristine PI membrane and MMMs ....................... 116
xiv
Figure 6.9 n-Butanol dehydration performance of the pristine PI membrane and
Figure 6.10 Dehydration performance of the pristine PI membrane and MMMs for
normalized flux and (b) Water concentration in the permeate vs. carbon number
.............................................................................................................................. 123
60 ºC..................................................................................................................... 126
Figure 7.2 FESEM images of (a) UiO-66-NH2 and (b) UiO-66-F4 nanoparticles
.............................................................................................................................. 142
membrane (b) MMM containing 10wt% UiO-66-F4 and MMMs containing (c)
Figure 7.4 (a) SEM image and (b) EDX mapping of Zr on the cross-section of
Figure 7.5 (a) SEM image and (b) EDX mapping of Zr on the cross-section of
Figure 7.7 TGA curves of UiO-66-F4, pristine PI membrane and UiO-66-F4 based
xv
Figure 7.8 Methanol dehydration performance of the pristine PI membrane and
NH2 based MMMs for C1–C3 alcohols at 60 ºC with feed composition of alcohol
/water (85/15wt%): (a) Normalized flux and (b) Water concentration in the
LIST OF SYMBOLS
θ incident angle
xvi
𝜌𝑒 density of hexane
𝜌0 density of membrane
τ3 o-Ps lifetime
d d-spacing
l membrane thickness
𝑛 any integer
I3 o-Ps intensity
J flux of a membrane
Pi permeability of component i
Pj permeability of component j
xvii
pisat saturated vapor pressure of component i
BDC 1,4-benzene-dicarboxylate
DABA 3,5-diaminobenzoicacid
DMAC N,N-dimethylacetamide
DMF N,N-dimethylformamide
DMSO N,N-dimethylsulfoxide
GO graphene oxide
IPA isopropanol
NMP N-methyl-2-pyrrolidone
PAE polyarylether
xviii
PALS positron annihilation lifetime spectroscopy
PBI polybenzimidazole
PBO polybenzoxazole
PBOZ polybenzoxazinone
PDMS polydimethylsiloxane
PESU polyethersulfone
PI polyimide
PPSU polyphenylsulfone
PTFE polytetrafluoroethylene
T-PESU polytrimethylphenylethersulfone
TR thermal rearrangement
xix
CHAPTER 1: INTRODUCTION
molecules to pass while retains others and therefore achieves separation [1].
different fields such as gas separation, waste water treatment, sea water
permeation and vaporization of the permeate. Figure 1.1 shows the schematic
be separated (feed) contacts with the membrane on one side and permeating
1
1.2 Current applications of membrane pervaporation
Table 1.1 lists some potential liquids separations that request technological
main application in the industry. Some organics, such as alcohols with low
2
molecular weight, ethers, and ketones, can be completely miscible with water
such as distillation usually need high energy thus are not economic.
not need third component to break the azeotropes and can avoid cross-
the permeating component consumes the latent heat and operating condition of
produce high-purity alcohols. The most important alcohols are ethanol and
effective cleaner for electronic and optical devices because of its non-toxicity,
hand sanitizer and disinfecting pads [5, 6]. The production of ethanol and IPA
3
attention. Besides, the dehydration of other solvents, such as glycerol, acetone
developed by Membrane Technology and Research (MTR) [9]. After that, the
process was further expanded to various areas: (1) removal of trace amount of
volatile organic compounds (VOCs) from water stream [10]; (2) recovery of
valuable organics from aqueous solutions [11]; (3) removal of certain product
from a fermentation broth, which is helpful to improve the conversion rate and
4
In this application, membrane materials need to be organophilic and
5
poly(tetrafluoro ethylene) (PTFE), and cellulose acetate [13, 14, 17]. The main
limitation for its wide application is lack of suitable membranes and modules
Based on the above review, it can be found that one of the major obstacles for
and high costs limit their applications. Researchers have tried to integrate
6
good mechanical stability, thermal stability and chemical resistance are also
swelling in harsh environment [18]. The other challenge is due to the fact that
a certain material often can only show good separation performance for a
specific mixture but not for other types of mixtures. In other words, different
selection and tests of membrane materials must be done for different mixture
pairs.
reduces defects and minimize transport resistance [2]. Based on the second
a porous supporting layer is desired to achieve high permeance. There are two
made of the same material [19] and (2) composite membranes consisting of
different materials for the dense selective layer and the porous substrate [20].
The second type membranes show great potential for commercialization due to
cost effective advantage. The composite membranes only need a small amount
of expensive material with high performance for the ultrathin selective layer
and the porous substrate can be made from cheap materials. However, the
7
compatibility of these different materials and the defect-free requirement for
the ultra-thin selective layer are the main challenges of composite membranes.
durability.
In the industry, there are three major separation technologies for alcohol
• Distillation
very effective at low alcohol concentrations and can provide a high separation
at large-scale and is easier for mass and energy integration [5]. However, it
8
becomes difficult for azeotrope separation without adding in additional
heat and mass integration becomes very difficult at small scales [5].
• Adsorption
have been developed including molecular sieves e.g. zeolites and bio-based
materials such as starch, corn cobs, cellulose and activated palm stone [5, 21].
Both liquid and vapor phase adsorptions are technically feasible, but the vapor
sieves provide a high selectivity, but water is very strongly adsorbed thus high
• Pervaporation
9
separating azeotropic and close boiling point liquid mixtures [22]. Second, it is
process because only the permeating component consumes the latent heat [23,
feed pressures and temperatures [22, 25]. Last, the compactness in module
one of the major obstacles for the development of pervaporation is the lack of
membranes with high productivity, high selectivity and good stability. The
other challenge is due to the fact that a certain material often can only show
good separation performance for some specific mixtures but not for other
The objective of this thesis is to explore various new membrane materials with
high productivity, high selectivity and good stability, including polymers and
because they have good mechanical strength, excellent thermal and chemical
10
stability. A fundamental study of polymer properties and a systematic study
rearrangement;
membranes;
11
The dissertation is organized into eight chapters. Chapter one gives the
development of pervaporation.
Chapter two presents the fundamentals of mass transport mechanisms, and the
techniques.
structure, free volume, d-spacing, water contact angle, water and ethanol
Long-term stability tests were carried out for both H-PESU and PESU
membranes.
12
Chapter five reports the synthesis of novel polyimide and the development of
also studied.
Chapter six shows the design of novel mixed matrix membranes consisting of
Solvent uptakes and free volume properties were characterized to reveal the
13
UiO-66-NH2 and UiO-66-F4 nanoparticles with similar particle size were
spectroscopy test were carried out to reveal the actual mechanisms for the
the MOFs particles and the effects of particle loading on the alcohol separation
Chapter eight draws the conclusions of the current research work and proposes
1.6 References
[2] L.Y. Jiang, Y. Wang, T.S. Chung, X.Y. Qiao, J.Y. Lai, Polyimides
(2009) 1135-1160.
14
[4] T. Jojima, M. Inui, H. Yukawa, Production of isopropanol by
(2008) 1219-1224.
[5] L.M. Vane, Separation technologies for the recovery and dehydration of
553-588.
(2010) 45-58.
unit: VOC removal from water and surfactant solutions, J. Membr. Sci. 202
(2002) 177-193.
[12] L.M. Vane, A review of pervaporation for product recovery from biomass
15
[13] B. Smitha, D. Suhanya, S. Sridhar, M. Ramakrishna, Separation of
(2004) 1-21.
(2000) 75-78.
(2000) 159-174.
(2010) 12067-12073.
16
[22] P. Shao, R.Y.M. Huang, Polymeric membrane pervaporation, J. Membr.
[24] G. Liu, W.S. Hung, J. Shen, Q. Li, Y.H. Huang, W. Jin, K.R. Lee, J.Y.
4521.
(1999) 183-210.
17
CHAPTER 2: LITERATURE REVIEW
during mass transport through the membrane. The driving force of mass transport
affinities of feed components to the membrane material, and (3) the membrane
formulate one single explanation for the sophisticated transport process. There
pervaporation: (i) the solution-diffusion model and (ii) the pore flow model.
researchers describe the mass transport through a non-porous membrane [1]. The
18
the permeant from the feed liquid onto the membrane surface, (ii) diffusion of
the permeant in the membrane, and (iii) desorption of the permeant in a vapor
phase on the downstream side of the membrane [2]. This process is illustrated
in Figure 2.1. The first two steps are the rate-controlling steps and the difference
in solubility and diffusivity of each species in the mixtures determines the final
membrane should have higher solubility and diffusivity towards the permeate
components.
special interaction and/or high affinity with membrane materials, which could be
significantly affected by the molecular size and shape of the permeate, the
mobility of polymer chains, the interstitial space between polymer chains or the
19
2.1.2 Pore flow model
Pore flow model is proposed more recently by Okada and Matsuura [5].
membranes, pore flow model assumes there are straight and cylindrical pores
The pores are perpendicular to the membrane surface and all the pores are
are filled with liquid with a distance of δ1 while the rest part of the pores is
filled with vapor with a length of δ2. In other words, there is a liquid/vapor
model, the mass transport process also includes three steps: (i) the permeates
transport through the liquid-filled portion of the pores from feed solution, (ii) a
liquid-to-vapor phase transition happens at the phase interface and (iii) the
downstream. In other words, this model suggests that the mass transport can
Thus, this model simplifies the membrane structure. Pore flow model has been
studied and applied for various membranes and different separating mixtures
20
Figure 2.2 Schematic diagram of pore flow model
describe pervaporation process and they can be used to explain and predict
operation conditions.
The flux (J) was determined by the total mass of permeate (Q) divided by the
product of the effective membrane area (A) and interval time (t).
21
Q
J (2.1)
A t
ywi / ywj
(2.2)
xwi / xwj
Flux and separation factor are usually reported in most studies, and are widely
used in the comparison with literature date. However, flux and separation
factor lie on both the intrinsic properties of membrane material and operation
Ji l
Pi (2.3)
pif pip
pif and pip are the partial vapor pressure of component i in feed and permeate,
f
Therefore, pi represents the partial vapor pressure of component i in a
hypothetical vapor phase in equilibrium with the feed liquid, which can be
22
Wherein, xi is the mole fraction of component i in the feed. i and pi are the
sat
Process Modelling software (version 7.2) based on the Wilson equation and
pip yi p p (2.5)
is the total pressure at permeate side, was assumed to be zero due to the
following equation:
Ji l
Pi (2.6)
xi i pisat
Ideal selectivity of the membrane (β) was expressed as the ratio of the
23
The two important aspects of pervaporation membrane development are
usually used due to their preferential sorption for water molecules. However,
fractional free volume gives rise to a higher diffusivity but a lower diffusivity
mechanical strength.
relatively cheap but often encounter the trade-off between permeability and
high performance, but are limited by the expensive costs and fragility. In the
24
following part, both polymeric and inorganic materials as well as mixed
• Polymeric materials
polyvinyl alcohol (PVA) [11], cellulose, chitosan [12] and alginate [13], have
operating temperatures. Therefore, polymers with stiff and rigid chains have
strength due to the rigid polymer chains. Moreover, the glassy characteristic of
membranes.
25
Polyimides have been viewed as a kind of promising polymers due to its
strength, as well as good film forming property [15]. There are some
Torlon, which have been used widely in recent years. Usually, polyimides are
Liu et al. successfully applied P84 co-polyimide hollow fiber membranes for
IPA dehydration [16] and modifications including silicon rubber coating and
post heat treatment are used to enhance membrane performance. Jiang and
hollow fiber membranes using this co-polyimide as the selective layer [18].
26
membrane shows a good separation efficiency with water concentration in
Polyamides known as Nylon are another kind of heat resistance materials with
have extraordinary thermal and chemical stability thus can effectively resist
aggressive chemicals including acids, bases, organic solvents, oils and strong
high permeability due to the large free volumes but relatively low selectivity.
27
In addition, aromatic polymers such as polybenzoxazole (PBO),
chemical and thermal stability. Both PBO and PBOZ membranes were
precursors and were applied in dehydration of alcohols [19, 27, 28]. Ong et al.
and Xu et al. found that the thermally rearranged PBO membranes show a
of 200-250h [19, 27] whereas Pulyalin A et al. found that water diffusivity
was greatly improved in the PBOZ membrane compared to its precursor [28].
Chung and co-workers [29-32]. They have developed PBI membranes both in
such as alcohols, glycols and acetone. In addition, PBI was also used as a filler
[33].
• Inorganic materials
and can thus be applied in harsh chemical environment. Because of the ability to
be used at high temperature, inorganic membranes show high flux with a stable
28
brittleness, high cost and complicated processbility, which restrict their wide
and metal organic frameworks (MOFs) etc. [34-39]. Currently, most of the
Zeolites membranes have high permeability and selectivity due to the highly
ordered and well-defined structure. Zeolites have various structures with different
pore sizes thus can have molecular sieving property for different pairs of mixtures.
and have stable performance under harsh environments. Morigami et al. first
plant [40]. This shows the feasibility of large-scale use of zeolite membranes in
industry.
29
properties and high stability in water due to the hydrogen-bonding interactions
between GO nanosheets. Recently, Yeh et al. and Hung et al. applied the GO
membranes for the dehydration of ethanol and isopropanol, respectively [42, 43].
there are still some issues that need to be addressed to further develop GO
properties [44]. It has been a hot topic for gas separation [45]. In recent years,
MOFs are not stable in water, therefore very few research works develop pure
pervaporation.
30
• Mixed matrix membranes
carbon molecular sieves, silica, metal oxides, carbon nanotubes, silicate, metal
covalent organic frameworks and many others have been investigated and
[14, 44, 45, 48]. Among them, MOFs, have been demonstrated as promising
materials for MMMs due to their advantages such as large surface areas,
tuneable porosity, high adsorption capacity and good affinity with polymer
separation factor [49], while others observed improvements of both flux and
separation factor [50]. The different results can be ascribed to several factors,
including the compatibility between inorganic fillers and the polymeric matrix,
the hydrophilicity of inorganic fillers, the pore size of inorganic fillers and the
selectivity [14]. To improve the compatibility between the polymeric and the
31
inorganic phases and achieve an even dispersion of inorganic fillers, following
approaches have been developed recently: (1) modifying inorganic fillers with
coupling agents which have strong bondings with polymeric phase [51]; (2)
coating inorganic fillers with a thin polymeric layer [52, 53]; (3) employing
polyhedral oligomeric silsesquioxane (POSS) [18, 54] or MOFs [49, 50, 55];
32
Asymmetric membranes are developed to reduce thickness of selective layer
the same material for selective layer and porous substrate and (2) composite
membranes with different materials for selective layer and porous substrate.
porous solid phase and a dense skin layer are formed. This phase inversion
be used for the ultra-thin selective layer, while the porous substrate can be
fiber or tubular membranes. Among them, hollow fiber membranes have been
widely used because hollow fiber membranes have superior advantages such
33
flexibility and ease of fabrication as well as scale-up [1, 59]. Conventional
hollow fiber membranes have been widely developed [60, 61]. Multi-bore
traditional single-bore hollow fiber membranes thus can reduce the chances of
fiber membranes are still in the developing stage and are only commercialized
2.4 References
34
[7] T. Okada, T. Matsuura, Predictability of transport equations for
163-175.
(1991) 151-168.
employing the modified pore-flow concept, J. Membr. Sci. 374 (2011) 67-82.
[14] Y.K. Ong, G.M. Shi, N.L. Le, Y.P. Tang, J. Zuo, S.P. Nunes, T.-S. Chung,
57 (2016) 1-31.
35
[15] L.Y. Jiang, Y. Wang, T.S. Chung, X.Y. Qiao, J.-Y. Lai, Polyimides
(2009) 1135-1160.
[16] R. Liu, X. Qiao, T.S. Chung, The development of high performance P84
(2010) 45-58.
[18] N.L. Le, Y.P. Tang, T.S. Chung, The development of high-performance
[19] Y.M. Xu, N.L. Le, J. Zuo, T.S. Chung, Aromatic polyimide and
325.
polyamide (PA) composite membranes, J. Appl. Polym. Sci. 105 (2007) 3132-
3137.
[22] Y. Wang, L.Y. Jiang, T. Matsuura, T.S. Chung, S.H. Goh, Investigation
36
polyetherimide (PEI) membranes for isopropanol dehydration via
216.
24.
membranes with improved selectivity, Ind. Eng. Chem. Res. 52 (2012) 1141-
1149.
[27] Y.K. Ong, H. Wang, T.S. Chung, A prospective study on the application
(2007) 60-66.
37
[30] Y. Wang, M. Gruender, T.S. Chung, Pervaporation dehydration of
[32] G.M. Shi, Y. Wang, T.S. Chung, Dual‐layer PBI/P84 hollow fibers for
[33] T.S. Chung, W.F. Guo, Y. Liu, Enhanced Matrimid membranes for
[37] P.S. Tin, H.Y. Lin, R.C. Ong, T.S. Chung, Carbon molecular sieve
38
[38] Y.P. Tang, D.R. Paul, T.S. Chung, Free-standing graphene oxide thin
[41] D.R. Dreyer, S. Park, C.W. Bielawski, R.S. Ruoff, The chemistry of
[42] T.M. Yeh, Z. Wang, D. Mahajan, B.S. Hsiao, B. Chu, High flux ethanol
[43] W.S. Hung, Q.F. An, M. De Guzman, H.Y. Lin, S.H. Huang, W.R. Liu,
C.C. Hu, K.R. Lee, J.Y. Lai, Pressure-assisted self-assembly technique for
39
organic frameworks with exceptional stability, J. Am. Chem. Soc. 130 (2008)
13850-13851.
[48] J. Caro, Are MOF membranes better in gas separation than those made of
[50] D. Hua, Y.K. Ong, Y. Wang, T.X. Yang, T.S. Chung, ZIF-90/P84 mixed
40
[55] Y.M. Xu, T.S. Chung, High-performance UiO-66/polyimide mixed
[57] W.S. Hung, M. De Guzman, S.H. Huang, K.R. Lee, Y. Jean, J.Y. Lai,
(1999) 183-210.
[60] P. Sukitpaneenit, T.S. Chung, Fabrication and use of hollow fiber thin
film composite membranes for ethanol dehydration, J. Membr. Sci. 450 (2014)
124-137.
technologies: past, present, and future, Prog. Polym. Sci. 37 (2012) 1401-1424.
41
CHAPTER 3: EXPERIMENTAL
3.1 Materials
3.1.1 Polymers
6FDA, HAB and DABA with four various diamine HAB to DABA ratios of
10:0, 9:1, 7:3 and 5:5 (abbreviated as PI-10-0, PI-9-1, PI-7-3 and PI-5-5) via
diamine (HAB, Tokyo Chemical Industry) was used without further treatments.
Acetic anhydride and pyridine from Aldrich of the reagent grade or higher,
42
(Singapore), tetrafluorobenzene-1,4-dicarboxylic acid from Matrix Scientific
and acetic acid (glacial) from Merck (Germany) were obtained and used as
received.
Scientific (UK) was used as solvents for UiO-66-Type MOFs syntheses and
Fisher Scientific (UK) of the reagent grade or higher, were used to prepare the
aqueous feed solutions for pervaporation studies. Hexane (HPLC grade) from
HAB and DABA at ratios of 10:0, 9:1, 7:3 and 5:5 were accurately weighed
dianhydride was added. The reaction mixture was stirred for half an hour at
polyimide, acetic anhydride and pyridine were poured into the solution.
Further stirring in one day was conducted to achieve the complete imidization.
43
The synthesis scheme is shown in Figure 3.1. The polyimide solution was then
acid (glacial) were added into 20mL DMF in a well-sealed blue cap bottle.
The mixture was stirred at 100°C (for UiO-66 and UiO-66-NH2) or at room
temperature (for UiO-66-F4) for 24h. After cooling to room temperature, the
44
UiO-66-Type MOFs particles were obtained by centrifugation and further
purified by washing with fresh DMF for several times. After that, the UiO-66-
Type MOFs particles were immersed in fresh DMF prior to use. The ligand
The flat dense membranes were prepared by the knife casting and solvent
vacuum oven at 120°C. Firstly, the polymer was dissolved in NMP to prepare
solution was cast on a glass plate using a casting knife with a gap thickness of
guarantee a slow solvent evaporation. After peeled off from the glass plate,
the membrane was further dried in a vacuum oven at 250°C for 15h to remove
45
3.4.2 Fabrication of polyimide dense membranes by ring casting
(PTFE) syringe filter and poured onto silicon wafers. The solution was left to
totally remove the residual solvent, the membranes were further dried under
200 °C at a rate of 25 °C/h, kept at 200 °C for 12h and naturally cooled down
to room temperature. The thickness of the resultant membrane was about 20-
treatment
to keep isothermally for half an hour in a high vacuum level. After thermal
the furnace before tests. The resultant membranes are abbreviated as C-TR-10-
46
3.4.4 Fabrication of mixed matrix membranes by ring casting
The mixed matrix membranes in this study were made by the solution casting
was filtered using a 5 µm PTFE syringe filter before mixing with the UiO-66-
directly mixed with the polymer solution. The mixture was stirred for another
30min and cast onto a silicon wafer. The casting solution was put in an oven at
80°C to ensure slow evaporation of the solvent. After one to two days, the
mixed matrix membranes were peeled off from the silicon wafer. To totally
remove the residual solvent, the membranes were further dried in a vacuum
3.5 Characterizations
spectroscopy (Bio-Rad FTS 135) (FTIR) in the range of 500-4000 cm-1 in both
47
3.5.2 Wide angle X-ray diffraction (XRD)
1.54 Å as the radiation source. The d-spacing (𝑑), representing the average
2𝑑𝑠𝑖𝑛𝜃 = 𝑛𝜆 (3.1)
where, 𝜃 is the incident angle, 𝑛 is the any integer and 𝜆 is the wavelength of
the beam.
rate of 30 mL min-1. The MOFs loading in the mixed matrix membranes was
𝑃𝑎𝑡𝑖𝑐𝑙𝑒 𝑤𝑒𝑖𝑔ℎ𝑡
𝑃𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑙𝑜𝑎𝑑𝑖𝑛𝑔 = ∗ 100% (3.2)
𝑃𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑤𝑒𝑖𝑔ℎ𝑡+𝑝𝑜𝑙𝑦𝑚𝑒𝑟 𝑤𝑒𝑖𝑔ℎ𝑡
48
3.5.4 Themogravimetric analyses-Fourier transform infrared spectrometer
(TGA-FTIR)
In order to investigate the reactions during the thermal treatment, TGA-IR was
membranes and MMMs) with a flow rate of 50 mL min-1. All samples were
The morphologies of the hollow fiber substrates and TFC membranes were
N2 to obtain proper cross-sections and coated with platinum using a JEOL JFC-
49
MMMs using an Oxford INCA energy dispersion of X-ray system together
Pte Ltd. The membrane was first measured in air and then in hexane, and the
where, 𝑚0 and 𝑚0′ represent the masses of the membrane in air and in hexane,
respectively. At least three samples were tested for each kind of membranes to
measure the contact angle at room temperature. Deionized water droplets were
placed onto the membrane surface by a Gilmont microsyringe and the contact
50
3.5.9 Solvent uptake analyses
Before conducting the experiments, the membrane strips were first dried in a
vacuum oven overnight to remove any moisture. The pre-weighed dry strips
swollen strips were removed from the solvent, blotted with tissue papers to
remove the excess solvent on their surfaces and their weights were determined
different time intervals were recorded until there was no further notable
change in their weights which indicates that the swollen membranes have
reached sorption equilibrium. The solvent uptake ratio for each sample was
determined by the weight difference between the swollen strip (𝑀𝑤𝑒𝑡 ) at the
𝑀𝑤𝑒𝑡 −𝑀𝑑𝑟𝑦
𝑆𝑜𝑙𝑣𝑒𝑛𝑡 𝑢𝑝𝑡𝑎𝑘𝑒 𝑟𝑎𝑡𝑖𝑜 = (3.4)
𝑀𝑑𝑟𝑦
picoseconds per channel was used to measure the lifetime and intensity of
22
positron species. Na was selected as the positron source and the counting
rate was around 165 counts s-1. A detailed description of the experimental
procedures can be found elsewhere [1, 2]. To perform the wet sample tests for
51
four polyarylether membranes, the membranes were firstly saturated in an
pervaporation tests, and then hermetically sealed in a plastic bag with the 22Na
[3]. The PATFIT program was employed to analyze the raw data and acquire
the three lifetime components (τ1, τ2 and τ3). The mean free-volume radius R
𝑅 1 2𝜋𝑅
𝜏3−1 = 2[1 − ∆𝑅 + 2𝜋 𝑠𝑖𝑛 ( ∆𝑅 )] (3.5)
where τ3 is the o-Ps lifetime. ΔR represents an empirical constant (1.66 Å). The
equation [6-8]:
4
𝐹𝐹𝑉 = 0.0018𝐼3 (3 𝜋𝑅33 ) (3.6)
The pervaporation system was a lab-scale static cell as depicted elsewhere [9].
Aqueous alcohol solutions were used as the feed solutions. The downstream
pressure was maintained at less than 1 mbar by a vacuum pump. The test
temperatures were controlled at 40°C, 60°C and 80°C. The system was
52
gas chromatography (HP-INNOWAX column, Hewlett-Packard GC 7890)
was much smaller than that of the feed solution and the variation (less than 1
wt%) could be neglected. In order to make sure the good reproducibility, all
𝑄
𝐽= (3.7)
𝐴𝑡
where A, Q and t are the effective membrane area, total mass of the permeate
𝑦 /𝑦
1/2 = 𝑥𝑤,1/𝑥𝑤,2 (3.8)
𝑤,1 𝑤,2
indicate the components’ weight fractions in the feed and the permeate.
To study the intrinsic properties of the MMMs, the permeability and mole-
𝑙
𝑃𝑖 = 𝐽𝑖 × (3.9)
𝑥𝑖 𝛾𝑖 𝑃𝑖𝑠𝑎𝑡 −𝑦𝑖 𝑃𝑝
𝐽𝑖 ,𝑃𝑖𝑠𝑎𝑡 and 𝛾𝑖 are the flux, the saturated vapor pressure of component 𝑖 and the
component 𝑖 in the feed and the permeate, respectively. 𝑃𝑝 is the total pressure
at the permeate side, which could be viewed as zero because the pressure in
the permeate side is much smaller than the saturated vapor pressure in the feed
53
side due to the vacuum condition. By using the AspenTech Process Modelling
software, both 𝛾𝑖 and 𝑃𝑖𝑠𝑎𝑡 were calculated by the Wilson and the Antoine
following equation:
𝑀 𝑃
𝛽𝑚𝑜𝑙𝑒 = ( 𝑀𝑗 ) (𝑃 𝑖 ) (3.10)
𝑖 𝑗
where 𝑀𝑖 and 𝑀𝑗 are the molecular weights of water and alcohol, respectively.
3.7 References
[1] N.L. Le, Y.P. Tang, T.S. Chung, The development of high-performance
[2] Y.P. Tang, H. Wang, T.S. Chung, Towards high water permeability in
[3] Y.P. Tang, D.R. Paul, T.S. Chung, Free-standing graphene oxide thin films
56 (1972) 5499.
[5] S.J. Lue, D.T. Lee, J.Y. Chen, C.H. Chiu, C.C. Hu, Y.C. Jean, J.Y. Lai,
54
[6] V.P. Shantarovich, I.B. Kevdina, Y.P. Yampolskii, A.Y. Alentiev, Positron
annihilation lifetime study of high and low free volume glassy polymers:
[7] H.M. Chen, W.S. Hung, C.H. Lo, S.H. Huang, M.L. Cheng, G. Liu, K.R.
Lee, J.Y. Lai, Y.M. Sun, C.C. Hu, R. Suzuki, T. Ohdaira , N. Oshima, Y.C.
[8] M.L. Williams, R.F. Landel, J.D. Ferry, The temperature dependence of
55
CHAPTER 4: POLYARYLERHER MEMBRANES FOR
4.1 Introduction
technologies that can conserve energy, minimize waste discharge and reduce
technology and has been widely used in various industries because it has the
small footprint and ease of operation [2, 3]. In addition, this process is not
56
and the enhanced solubility selectivity of water to organic solvents [9-11].
stability and selectivity [6, 12-14]. Meanwhile, glassy polymers with rigid and
potential membrane materials for this application because their rigid and stiff
structures could not only improve the diffusivity selectivity but also enhance
polymer has been widely used owing to its good mechanical properties,
have been carried out: namely, (1) bulk modification of the PAE material, e.g.
sulfonation of PAE [23] and grafting hydrophilic groups to its chains [24, 25];
(2) surface modifications of PAE membranes via coating [26, 27] and
interfacial polymerization [28, 29]; and (3) blending directly with hydrophilic
[31]. These modifications not only improve membrane hydrophilicity but also
57
modified PESU membranes using a copolymer Pluronic F127 as a surface
thiophene from n-octane via pervaporation [32]. Both the permeation flux and
Different from previous work which just modified the membrane surface or
investigate the potential of this new H-PESU for dehydration of alcohols via
pervaporation. The second objective of this work aims to study the influence
structures of these four PAE polymers are depicted in Figure 4.1. To meet our
and methanol aqueous solutions is our focus because they are smaller
[33]. Besides, solvent uptake and fractional free volume were characterized to
reveal their solubility and diffusivity preferences and to explain their different
58
term stability tests have also been carried out to give a comprehensive
evaluation about these membranes. This work may provide useful insights to
dehydration.
4.2 Experimental
determined by contact angle tests. Solvent uptake tests were conducted at 60℃
to study the affinities of alcohol and water with the membranes. PALS tests
59
were carried out for membranes at both dry state and wet state to study the
Figure 4.2. PESU and PPSU have better thermal stability than T-PESU and H-
PESU. The former two show no obvious weight loss (less than 1 wt%) until
PESU membrane starts to degrade from 400°C, while the H-PESU membrane
undergoes an earlier weight loss at a lower temperature range from 200 to 400
agreement with previous studies [34, 35]. After that, the decomposition of the
range. Nevertheless, these four membranes have sufficient thermal stability for
60
PESU
100
PPSU
Weight percentage (%) T-PESU
80 H-PESU
60
40
20
0
0 200 400 600 800
Temperature (ºC)
Figure 4.2 TGA curves of PESU, T-PESU, PPSU and H-PESU membranes
Intensity (a.u.)
PPSU
T-PESU
H-PESU
PESU
5 10 15 20 25 30 35 40 45 50
2 theta (º)
Figure 4.3 XRD patterns of PESU, T-PESU, PPSU and H-PESU membranes
61
Figure 4.3 displays the XRD patterns of the four membranes. It is found that
the peak of T-PESU appears at the smallest angle while the other three show
peak angle should have a larger d-spacing value. Therefore, among these four
because the three -CH3 groups in T-PESU can prevent polymer chains from
close packing due to the steric hindrance effect. Table 4.1 shows that the
density of these four membranes follows the trend of T-PESU < PPSU < H-
PESU < PESU. The lowest density of T-PESU agrees well with its highest d-
Table 4.1 Densities and water contact angles of PESU, T-PESU, PPSU and H-
PESU membranes
Membrane Density (g/cm3) Water Contact Angle (º)
PESU 1.354±0.021 88.8±1.4
PPSU 1.266±0.005 85.8±0.9
T-PESU 1.213±0.017 85.4±0.9
H-PESU 1.294±0.013 41.8±1.4
Table 4.1 also summarizes the water contact angles of these four membranes.
PESU, PPSU and T-PESU membranes have close and comparable water
contact angles, while the water contact angle of the H-PESU membrane is
much smaller than those of the other three. Therefore, it can be concluded that
among these four membranes, the H-PESU membrane has the largest
62
4.3.2 Solvent uptake analyses
membranes, the solvent uptake tests were carried out and characterized by the
membrane weight changes due to the sorption of different solvents. Table 4.2
compares the solvent uptake ratios of the membranes in water and ethanol and
methanol. The water uptake ratios follow the trend of PESU < PPSU < T-
inverse trend. This phenomenon can be explained by the contact angle results.
Since the water contact angles follow the trend of PESU < PPSU < T-PESU <
the membranes can absorb more water and less ethanol and methanol
highest water solubility and lowest ethanol and methanol solubilities, followed
towards water of these four membranes obeys the trend of PESU < PPSU < T-
Table 4.2 Solvent uptake results for PESU, T-PESU, PPSU and H-PESU
membranes
Water Ethanol Methanol Uptake ratio Uptake ratio
Membrane uptake uptake uptake ratio of water to of water to
ratio (%) ratio (%) (%) ethanol methanol
PESU 3.1±0.7 9.9±1.6 5.2±1.1 0.31 0.60
PPSU 3.9±0.8 8.1±1.5 4.8±0.7 0.48 0.81
T-PESU 4.1±0.7 7.5±1.5 4.6±1.1 0.55 0.89
H-PESU 7.9±0.8 5.7±1.5 3.3±0.7 1.39 2.39
63
4.3.3 PALS analyses
To further quantitatively analyze the free volume size of the four membranes,
PALS measurements were carried out. The cavity size of free volume and its
lifetime τ3 and its intensity I3, respectively [37, 38]. Table 4.3 shows PALS
raw data, free volume sizes and FFVs of the four membranes at the dry state.
solution for 12 hr before tests. Table 4.4 tabulates the results. Comparing the
PALS data at dry and wet states, increases in τ3 and R are observed for all four
clearly accounts for the significant flux increment after swelling. On the
contrary, I3 decreases for all the membranes, indicating that adjacent small
cavities at the dry state may coalesce to form large cavities at the wet state or
some cavities are filled with the penetrants at the wet state [39].
Table 4.3 PALS data of dry PESU, T-PESU, PPSU and H-PESU membranes
Membrane τ3 (ns) I3 (%) R (Å) FFV (%)
PESU 1.779±0.009 19.52±0.15 2.644±0.008 2.720±0.046
PPSU 1.922±0.008 23.83±0.14 2.782±0.006 3.871±0.049
T-PESU 1.928±0.009 20.19±0.13 2.788±0.007 3.299±0.048
H-PESU 1.839±0.009 18.91±0.14 2.703±0.008 2.816±0.045
64
Table 4.4 PALS data of wet PESU, T-PESU, PPSU and H-PESU membranes
Membrane τ3 (ns) I3 (%) R (Å) FFV (%)
PESU 1.867±0.010 18.88±0.14 2.731±0.008 2.898±0.047
PPSU 2.044±0.009 22.16±0.12 2.896±0.006 4.057±0.050
T-PESU 1.967±0.010 20.12±0.13 2.824±0.007 3.418±0.049
H-PESU 2.117±0.010 18.07±0.11 2.961±0.007 3.535±0.048
A comparison of FFVs at dry and wet states is shown in Figure 4.4. For PESU,
T-PESU and PPSU, the percentage of FFV increment follows the order of
ratio. This is because the FFV increment is caused by swelling and for these
uptake prevails over the water uptake as shown in Table 4.2. Therefore, the
FFV incremental trend is in good agreement with the ethanol uptake ratio
trend for these three membranes. On the contrary, swelling of the H-PESU
implied in Table 4.2. Figure 4.4 shows that the FFV increment for the H-
PESU membrane is much more severe than the other three. As a result, FFV of
the H-PESU membrane becomes the second largest after swelling. In addition,
the incorporation of the Pluronic moiety. This suggests that the swelling of H-
65
5 30
4.5
3.8714.057 25
2.5 15
2
1.5 10
1 5
0.5
0 0
PESU PPSU T-PESU H-PESU
Figure 4.4 Comparison of FFVs at dry and wet states measured by PALS
product of flux and membrane thickness. This is to eliminate the impact of the
66
For these four PAE membranes, the water normalized flux is much higher than
the ethanol normalized flux and accounts for most of the total normalized flux,
higher water solubility while a higher FFV is conducive to the fast diffusion of
higher hydrophilicity and higher FFV are favorable to attain a high permeation
flux. In this regard, the PESU membrane shows the smallest water normalized
flux due to its smallest water solubility and diffusivity since it has the lowest
water uptake ratio and the smallest FFV as aforementioned in Table 4.2 and
Table 4.4, respectively. The increases of water normalized flux for T-PESU
and PPSU are because of the increases in both water solubility and diffusivity.
It's worth noting that the H-PESU membrane shows the largest water
normalized flux, which is 2-6 times higher than those of the other three
membranes. This can be explained by the following reasons: (1) the H-PESU
membrane has the largest water solubility as shown in Table 4.2 due to the
hydrophilic nature of the Pluronic moiety and (2) its FFV value is moderate as
shown in Table 4.4, and its hydrophilic Pluronic moiety may form special
water transport channels for fast water diffusion, which lead to a high water
diffusivity. At the same time, the H-PESU membrane also has the highest
ethanol normalized flux among these four PAE membranes which may be due
67
In addition, the PPSU membrane shows the lowest separation factor which is
mainly ascribed to its smallest diffusivity selectivity since it has the largest
than the others. This is due to the constraint of the hydrophobic and rigid
PESU moiety. In other words, the swelling mainly occurs in the Pluronic
regions, which have the highest solubility selectivity and most likely promote
the transport of water molecules rather than ethanol molecules. These Pluronic
channels are preferentially occupied by water molecules and this possibly adds
the resistance for ethanol transport. Therefore, the water molecules have
priority to diffuse through these channels. On the other hand, the rigid PESU
moiety prevents the membrane from uncontrolled swelling and ensures good
Table 4.6 shows a benchmarking comparison with literature data [40-45] . The
four membranes developed in this study not only exhibit comparable water
permeability but also higher mole-based selectivity than most of the rest in the
68
Table 4.6 A comparison of pervaporation performance for ethanol
dehydration
Water
Membrane Water Mole-
Temp. permeability
Membrane thickness concentration o based Ref.
( C) (mg m-1 h-1
(μm) in feed selectivity
kPa-1)
Crosslinked
20 10% 60 1.136 417 [40]
Chitosan
Crosslinked
29 10% 60 0.823 456 [41]
PVA
P84®
--- 15% 60 0.034 1874 [42]
copolyimide
Pervap® 2201 --- 10% 60 0.022 309 [43]
TR PBO 20-25 10% 25 1.254 186 [44]
Matrimid
45-55 15% 60 0.136 481 [45]
/hPIM-1
This
PESU 21 15% 60 0.049 1111
study
This
T-PESU 20 15% 60 0.061 1326
study
This
PPSU 23 15% 60 0.098 552
study
This
H-PESU 36 15% 60 0.322 685
study
not been widely studied, which may be attributed to the lack of suitable and
since they don’t have the solvent-induced swelling problem. However, their
69
[4, 6, 17]. Therefore, polymeric membranes are still not popular for methanol
dehydration.
In view of the good performance for ethanol dehydration, the four membranes
developed in this study were employed for methanol dehydration. Table 4.7
dehydration, the H-PESU membrane shows both the highest water and
methanol normalized fluxes. On the other hand, all these four PAE membranes
show separation factor higher than 28 and mole-based selectivity more than 82
which is far better than most of the other polymeric membranes listed in
Table 4.8 [33, 46, 47]. Especially, the T-PESU membrane has the highest
70
Table 4.8 A comparison of pervaporation performance for methanol
dehydration
Water Mole-
Membrane Water
Temp. permeability based
Membrane thickness concentration Ref.
(oC) (mg m-1 h-1 selectivity
(μm) in feed
kPa-1)
5%-sPPSU 16 15% 60 0.033 33.8 [33]
2.5%-sPPSU 16 15% 60 0.030 32.1 [33]
Agarose 14-20 14% 30 0.925-1.321 8.3 [46]
Polyamide-
30 10% 20 0.268-0.415 9.2-99.9 [47]
sulfonamide
This
PESU 21 15% 60 0.085 85.1
study
This
T-PESU 20 15% 60 0.089 114.0
study
This
PPSU 23 15% 60 0.089 82.0
study
This
H-PESU 36 15% 60 0.171 89.7
study
The H-PESU membrane is chosen for further studies of the effect of feed
85wt%, 90wt% and 95wt% are investigated. Figure 4.5 shows that the water
normalized flux decreases while the ethanol and methanol normalized fluxes
alcohol concentration results in a lower driving force for water transport but a
higher driving force for alcohol transport [44]. As a result, both total
concentration as shown in Figure 4.6 and Figure 4.7. Similar trends have been
71
a 6000
(gμm/m2h)
4000
3000
2000
1000
0
80 85 90 95 100
Alcohol Concentration (%)
1000
b Ethanol/water system
900
Alcohol normalized flux
600
500
400
300
200
100
0
80 85 90 95 100
Alcohol Concentration (%)
Figure 4.5 (a) Water and (b) ethanol/methanol normalized flux of H-PESU
membranes at different feed concentrations at 60 ºC
6000
Total normalized flux
500
Separation factor
5000
400
(gμm/m2h)
4000
3000 300
2000 200
1000 100
0 0
80 85 90 95 100
Ethanol Concentration (%)
72
6000 40
Separation factor
30
(gμm/m2h)
4000 25
3000 20
2000 15
10
1000 5
0 0
80 85 90 95 100
Methanol Concentration (%)
was selected for comparison. Their fluxes and separation factors were
continuously monitored for 250h. Figure 4.8 shows their results. It can be
found that for H-PESU membrane, the water concentration in the permeate
decreases slightly from 98.7 wt% to 96.5 wt% at the early stage, which is
the H-PESU membrane is quite reliable. This is again due to the combination
of the strengths from both rigid hydrophobic PESU moiety and flexible
73
normalized flux and separation factor during the entire testing duration of
250h. This is because the PESU membrane has higher anti-swelling properties
than the H-PESU membrane because the latter contains the hydrophilic
Pluronic moiety.
100.0
14000
Normalized flux (gμm/m2h)
permeate (wt%)
10000 94.0
92.0
8000 90.0
6000 88.0
4000 86.0
84.0
2000 82.0
0 80.0
100 0 50
150 200 250
Test Duration (h)
PESU H-PESU
Figure 4.8 Long-term stability of the PESU and H-PESU membranes for
ethanol dehydration at 60 ºC with the feed composition of 85wt%/15wt%
ethanol/water
4.4 Conclusions
materials: PESU, PPSU, T-PESU and H-PESU have been explored for
74
PALS measurements reveal that swelling results in a larger free volume radius
and a bigger FFV. In addition, the adjacent small cavities in the membranes at
the dry state may coalesce to form large cavities at the wet state. Among these
four PAE membranes, the H-PESU membrane shows the best ethanol
h-1 kPa-1 and a comparable mole-based selectivity of 685 due to its unique
water molecules. Therefore, the H-PESU membrane has the highest water
solubility and solubility selectivity. On the other hand, its hydrophobic and
rigid PESU moiety can prevent the membrane from uncontrolled swelling and
4.5 References
(1999) 183-210.
[2] L.M. Vane, A review of pervaporation for product recovery from biomass
75
[3] Y.M. Xu, T.S. Chung, High-performance UiO-66/polyimide mixed matrix
[4] Y.K. Ong, G.M. Shi, N.L. Le, Y.P. Tang, J. Zuo, S.P. Nunes, T.S. Chung,
57 (2016) 1-31.
(2010) 12067-12073.
[10] M.E. Dmitrenko, A.V. Penkova, A.B. Missyul, A.I. Kuzminova, D.A.
76
[11] A.J. Toth, A. Andre, E. Haaz, P. Mizsey, New horizon for the membrane
[12] W.S. Hung, S.M. Chang, R.L.G. Lecaros, Y.L. Ji, Q.F. An, C.C. Hu, K.R.
1728-1739.
[14] L.L. Xia, C.L. Li, Y. Wang, In-situ crosslinked PVA/organosilica hybrid
[15] Y.M. Xu, N.L. Le, J. Zuo, T.S. Chung, Aromatic polyimide and
325.
[17] L.Y. Jiang, Y. Wang, T.S. Chung, X.Y. Qiao, J.Y. Lai, Polyimides
(2009) 1135-1160.
[18] M.M. Teoh, T.S. Chung, K.Y. Wang, M.D. Guiver, Exploring Torlon/P84
77
modifications for pervaporation dehydration of isopropanol, Sep. Purif.
(2009) 363-368.
low temperature hydrothermal coating process, J. Membr. Sci. 380 (2011) 98-
113.
78
[27] R.X. Liu, X.Y. Qiao, T.S. Chung, Dual-layer P84/polyethersulfone
2978-2988.
[31] Y.Q. Wang, T. Wang, Y.L. Su, F.B. Peng, H. Wu, Z.Y. Jiang, Protein-
79
pervaporation from sulfonated polyphenylsulfone (sPPSU), J. Membr. Sci.
(2011) 14213-14218.
(2009) 2688-2698.
[36] M. Calle, C.M. Doherty, A.J. Hill, Y.M. Lee, Cross-linked thermally
[37] H.M. Chen, W.S. Hung, C.H. Lo, S.H. Huang, M.L. Cheng, G. Liu, K.R.
Lee, J.Y. Lai, Y.M. Sun, C.C. Hu, R. Suzuki, T. Ohdaira , N. Oshima, Y.C.
[39] S. Zhang, R. Zhang, Y.C. Jean, D.R. Paul, T. S. Chung, Cellulose esters
80
[40] W. Zhang, G. Li, Y. Fang, X. Wang, Maleic anhydride surface-
[43] D. Van Baelen, B. Van der Bruggen, K. Van den Dungen, J. Degrève, C.
[44] Y.K. Ong, H. Wang, T.S. Chung, A prospective study on the application
(2002) 293-300.
[47] W.H. Chan, C.F. Ng, S.Y. Lam-Leung, X. He, Water–alcohol separation
81
[48] Y. Wang, T.S. Chung, B.W. Neo, M. Gruender, Processing and
82
CHAPTER 5: POLYIMDE AND CROSSLINKED
THERMALLY REARRANGED
POLY(BENZOXAZOLE-CO-IMIDE)
5.1 Introduction
effective cleaner for electronic and optical devices because of its non-toxicity,
hand sanitizer and disinfecting pads [1, 2]. In addition, it can be considered as
an alternative fuel to alleviate global warming and slow down the depletion of
fossil fuels [3, 4]. Isopropanol is produced either by the chemical reaction
both processes require the separation of water from IPA to produce high purity
83
Pervaporation, a membrane-based separation technology, may be regarded as
particularly in separating azeotropic and close boiling point liquid mixtures [8].
does not need third components, which are commonly employed in the so-
process because only the permeating component consumes the latent heat [10,
feed pressures and temperatures [8, 12]. Last, the compactness in module
advantages in ease of fabrication and scale-up, low capital costs, and low
strength [15-21]. However, its high solvent resistance reduces its solubility in
have developed an alternative PBO membrane preparation method via the in-
84
situ thermal conversion of hydroxyl-containing polyimide precursors. In this
separating CO2/CH4 mixtures, surpassing the 2008 upper bound [22-25]. The
existence of micropores and high free volume in TR polymers is the reason for
the relatively stable selectivity [22]. Researchers have also investigated the
methods [29], TR protocols [27, 30-32] and TR-PBO membrane thickness [33]
Although TR-PBO membranes have been widely explored for gas separation,
there are limited studies on their applications for pervaporation [34, 35].
Among the limited studies, only one of them was related to alcohol
dehydration where Ong et al. studied the effects of dwell duration and thermal
85
addition, the TR-PBO membranes which were treated at 450°C for half an
hour exhibited the highest flux with a reasonable separation factor for ethanol
dehydration.
acid (DABA). The resultant polyimide precursors are first cast into
The advantages of these three chosen monomers are: (1) 6FDA contains
fluorine groups –CF3, which are expected to prevent polymer chains from
compact packing and increase the free volume of the resultant polymer; (2)
HAB has a hydroxyl group ortho to the amine group, which meets the
for the carboxylic-group containing polyimide was proposed by Qiu et al. [36]
generated from the decarboxylation of the anhydride along with the release of
a CO and CO2 molecule. After that, these adjacent phenyl radicals combine to
86
form linkages that yield biphenyl crosslinking. This would further enhance the
To our best knowledge, this is the first time that this material is synthesized
5.2 Experimental
investigate the reactions during the thermal treatment, TGA-IR was carried
out. PALS measurements and solvent uptake analyses were also conducted to
87
5.3 Results and discussion
polyimide, ATR-FTIR analyses are carried out and the results are shown in
Figure 5.2. The synthesized polymers possess the characteristic peaks at 1774
and 1716 cm-1 which are mainly attributed to typical double bands stretching
of the C=O group in the imide ring [37-40]. In addition, the C–N stretching at
1365 cm-1, the transverse stretching of C–N–C groups at 1084 cm-1, and the
presence of imide groups [34, 37, 38]. This confirms the successful fabrication
In contrast, for the C-TR-PBO membranes, it should be noted that the ATR-
FTIR spectra show weak peak intensities due to the dark brown colour of the
88
membranes. Therefore, a transmission mode of FTIR is used for the C-TR-
PBO membranes. According to Figure 5.3, three new peaks for C-TR-PBO
membranes can be observed at around 1477, 1554 and 1620 cm−1, which
characterize the vibration of benzoxazole rings [32, 41, 42]. Peaks at 1477 and
1554 cm−1 are the absorbing bands of benzoxazoles whereas the characteristic
peak at 1620 cm−1 is due to C=N stretching [32, 42]. These results validate the
Figure 5.4 shows the thermal behavior of the synthesized polyimides. The
from 100 to 700 oC. The initial weight loss (less than 1%) between 100oC to
second stage weight loss at 250-500oC is due to the release of gas molecules
89
produced from crosslinking and thermal rearrangement reactions. From 500 ºC
onwards, polymer chains start to decompose which results in the last stage
weight loss.
100
PI-5-5
90 PI-7-3
Weight percentage (%)
PI-9-1
80 PI-10-0
70
60
50
40
100 150 200 250 300 350 400 450 500 550 600 650 700
Temperature (ºC)
Figure 5.5 and Figure 5.6 show the TGA curve and the simultaneous IR
lowest intensity of a peak, red signifies the medium intensity, and turquoise
and the -C=O characteristic peak (1750 to 1850 cm−1) show up in the TGA-IR
spectra. The detected CO2 comes from the thermal rearrangement process.
When the temperature increases to about 520 oC, additional IR peaks including
H2O (3600-3750 cm−1), CFx (1150-1200 cm−1) and -C-O (1000-1050 cm−1)
groups are evolved. This suggests that the polymer backbone starts to
90
Compared with the IR spectrum of PI-10-0, Figure 5.6 reveals that PI-9-1
experiences an earlier appearance of H2O peak, -C=O peak, -C-O peak and
molecules.
91
According to Figure 5.1, water, CO2 and CO molecules are released during the
reaches about 500 oC, decomposition of polymer chains starts, similar to the
case of PI-10-0.
In order to understand the affinity between the permeate molecules and the
membrane, solvent uptake analyses were carried out at 60 oC. This temperature
is kept the same as the pervaporation test in order to have a fair comparison.
Figure 5.7 shows a comparison of solvent uptakes for the three polyimide
Both the polyimide precursors and C-TR membranes have higher affinity
towards isopropanol than water, which implies that the membranes remain
the water uptake increases while isopropanol uptake decreases for both
water. Comparing the polyimide and its respective C-TR membrane, it can be
observed that the latter absorbs less water and more isopropanol, which
implies C-TR membranes have lower solubility selectivity towards water than
92
polyimide membranes. This phenomenon corresponds well with the
HAB-DABA
quantitatively analyze the free volume size and distribution for the polyimide
and its intensity I3 are correlated to the size and concentration of the free
volume, respectively [44-46]. Figure 5.8 presents the PALS results. It can be
observed that the polyimide membranes have τ3 data in the range of 2.19 to
93
cavities during the thermal rearrangement and thermal crosslinking reactions.
TR membranes also increase, signifying that more pores are created after the
thermal treatment.
6 8
5.5 C-TR
5
PI
6
4.5
τ3(ns)
I3 (%)
4 4
3.5
3 C-TR 2
2.5
PI
2 0
10-0 9-1 7-3 5-5
HAB-DABA
On the other hand, Figure 5.8 also reveals the effect of DABA content on
probably because the carboxylic side chain in DABA is too small to affect
94
Figure 5.9 indicates the calculated fractional free volume as a function of
DABA to HAB ratio. Given that the lifetime τ3 for PI and C-TR membranes
volumes are mainly affected by the change of intensity. The fractional free
those of C-TR membranes exhibit a V-shape in the range from 1.44 to 2.20 %.
2.5
Fractional free volume (%)
2 C-TR
1.5
1 PI
0.5
0
10-0 9-1 7-3 5-5
HAB-DABA
95
4000 100.0
C-TR
3500
2
PI 99.0
3000 98.5
98.0
2500
C-TR 97.5
2000
97.0
PI
1500 96.5
96.0
1000
95.5
500 95.0
10-0 9-1 7-3 5-5
HAB-DABA
factor initially increases from PI-10-0 to PI-7-3 but subsequently decreases for
PI-5-5. The increase in normalized flux is attributed to two factors. First, the
discussed in section 3.2, which allows more water to be absorbed onto the
membranes. This, in turn, results in the loosening of polymeric chains (in the
wet state) and the increase in flux. Second, the membranes with a higher
DABA to HAB ratio have higher fraction free volumes (PI-7-3 and PI-5-5) as
indicated in Figure 5.9, which may lead to a higher flux in the wet state. On
the other hand, when the DABA to HAB ratio increases from 10:0 to 7:3, the
96
increase in separation factor is mainly due to the enhanced solubility
selectivity as discussed in section 3.2. For the membrane PI-5-5, the slight
rearrangement and crosslinking. Figure 5.10 shows that the normalized fluxes
of C-TR membranes are higher than those of PI membranes for all DABA to
larger free volume sizes and higher fractional free volumes. However, when
membranes have slightly lower separation factors. This is due to the decreased
diffusivity selectivity when their free volume sizes are larger and decreased
increases while that of PI membranes decreases, leading to the result that the
former is higher than the latter. This interesting result is attributed to anti-
97
5.3.5 Performance benchmarking and stability of C-TR membranes
and selectivity for the membranes in this study with various pervaporation
membranes available in literatures [34, 47-52]. Both the PI-5-5 and C-TR-5-5
much higher than others. This superior performance opens up the opportunity
for 200 h at 60 ºC. Figure 5.11 shows that the C-TR membrane displays quite
factor can be observed during the entire testing duration of 200 h. The stable
98
Table 5.1 A comparison of pervaporation performance of dense membranes
for isopropanol dehydration
Water Mole-
Water permeability
Temp. based
Membrane concentration -1 -1 Ref.
(ºC) (mg m h selectivity
in feed -1
kPa )
Matrimid 18% 100 0.490 50 [47]
Torlon 15% 60 0.011 3302 [48]
Ultem 15% 60 0.012 683 [48]
BPADA-ODA-
20% 60 0.145 9530 [49]
DABA polyimide
Crosslinked
10% 60 0.835 609 [50]
Chitosan
Matrimid/MgO
18% 100 0.067 1513 [47]
MMMs
PBI/ZIF-8 MMMs 15% 60 0.358 1969 [51]
P84/ZIF-90 MMMs 15% 60 0.166 449 [52]
TR PBO 10% 80 0.299 670 [34]
C-TR-5-5 15% 60 0.150 4019 This study
PI-5-5 15% 60 0.129 882 This study
140 60
(%)
50
120 40
100 30
20
80
10
60 0
0 50 100 150 200
Operating time (h)
99
5.4 Conclusions
form C-TR membranes. The science of both PI and C-TR membranes has
(2) As the DABA to HAB ratio increases, both PI and C-TR membranes
solubility selectivity towards water. On the other hand, the C-TR membrane
(3) C-TR membranes show a higher normalized flux than their PI precursor
membranes. As the DABA to HAB ratio increases, both normalized flux and
100
5.5 References
[1] L.M. Vane, Separation technologies for the recovery and dehydration of
553-588.
(2010) 45-58.
(2008) 1219-1224.
C.A. Eckert, W.J. Frederick, J.P. Hallett, D.J. Leak, C.L. Liotta, J.R. Mielenz,
[6] L.M. Vane, A review of pervaporation for product recovery from biomass
101
[9] B. Smitha, D. Suhanya, S. Sridhar, M. Ramakrishna, Separation of
(2004) 1-21.
[11] G. Liu, W.S. Hung, J. Shen, Q. Li, Y.H. Huang, W. Jin, K.R. Lee, J.Y.
4521.
(1999) 183-210.
[14] L.Y. Jiang, Y. Wang, T.S. Chung, X.Y. Qiao, J.Y. Lai, Polyimides
(2009) 1135-1160.
[16] G.L. Tullos, J.M. Powers, S.J. Jeskey, L.J. Mathias, Thermal Conversion
[17] X.D. Hu, S.E. Jenkins, B.G. Min, M.B. Polk, S. Kumar, Rigid-Rod
102
[18] K.I. Fukukawa, M. Ueda, Recent Development of Photosensitive
[22] H.B. Park, C.H. Jung, Y.M. Lee, A.J. Hill, S.J. Pas, S.T. Mudie, E. Van
Wagner, B.D. Freeman, D.J. Cookson, Polymers with Cavities Tuned for Fast
Selective Transport of Small Molecules and Ions, Science, 318 (2007) 254-
258.
[23] M. Calle, C.M. Doherty, A.J. Hill, Y.M. Lee, Cross-Linked Thermally
[24] M. Calle, H.J. Jo, C.M. Doherty, A.J. Hill, Y.M. Lee, Cross-Linked
[25] H.B. Park, S.H. Han, C.H. Jung, Y.M. Lee, A.J. Hill, Thermally
rearranged (TR) polymer membranes for CO2 separation, J. Membr. Sci. 359
(2010) 11-24.
103
[26] C.H. Jung, J.E. Lee, S.H. Han, H.B. Park, Y.M. Lee, Highly permeable
(2011) 1156-1165.
[28] S.H. Han, J.E. Lee, K.J. Lee, H.B. Park, Y.M. Lee, Highly gas permeable
[29] S.H. Han, N. Misdan, S. Kim, C.M. Doherty, A.J. Hill, Y.M. Lee,
43 (2010) 7657-7667.
[30] D.F. Sanders, Z.P. Smith, C.P. Ribeiro, R. Guo, J.E. McGrath, D.R. Paul,
[31] Z.P. Smith, D.F. Sanders, C.P. Ribeiro, R. Guo, B.D. Freeman, D.R. Paul,
104
[32] H. Wang, T.S. Chung, The evolution of physicochemical and gas
308-312.
[34] Y.K. Ong, H. Wang, T.S. Chung, A prospective study on the application
(2012) 182-193.
[36] W.L. Qiu, C.C. Chen, L.R. Xu, L.L. Cui, D.R. Paul, W.J. Koros, Sub-Tg
51 (2010) 4077-4086.
105
[40] W. Albrecht, B. Seifert, T. Weigel, M. Schossig, A. Holländer, T. Groth,
[41] S.Y. Wu, S.M. Yuen, C.C.M. Ma, Y.L. Huang, Synthesis and properties
725-735.
[44] S.J. Lue, D.T. Lee, J.Y. Chen, C.H. Chiu, C.C. Hu, Y.C. Jean, J.Y. Lai,
[45] H.M. Chen, W.S. Hung, C.H. Lo, S.H. Huang, M.L. Cheng, G. Liu, K.R.
Lee, J.Y. Lai, Y.M. Sun, C.C. Hu, R. Suzuki, T. Ohdaira , N. Oshima, Y.C.
106
[47] L.Y. Jiang, T.S. Chung, R. Rajagopalan, Matrimid®/MgO mixed matrix
[48] Y. Wang, L.Y. Jiang, T. Matsuura, T.S. Chung, S.H. Goh, Investigation
[52] D. Hua, Y.K. Ong, Y. Wang, T.X. Yang, T.S. Chung, ZIF-90/P84 mixed
Y.M. Xu, N.L. Le, J. Zuo, T.S. Chung, Aromatic polyimide and crosslinked
107
CHAPTER 6: UiO-66/POLYIMIDE MIXED MATRIX
6.1 Introduction
The global warming resulting from the heavy utilization of fossil fuels has
mankind can alleviate the reliance on fossil fuels. Biofuels, produced from
biofuels from fermentation broth is the most expensive process, which can
account for 60 to 80% of the total cost [1]. Therefore, the development of an
equilibrium [6].
108
Based on membrane materials, pervaporation membranes can be classified
have superior thermal, chemical and mechanical stabilities. They are not
their fragility and high costs of fabrication limit their wide applications.
utilized for pervaporation due to the advantages such as ease of fabrication and
They also suffer from the problem of long-term stability due to the solvent-
particles into polymeric matrices may result in MMMs that not only have
better separation performance than the original polymeric membranes but also
109
have been investigated and employed in mixed matrix membranes [8, 14-16].
large surface areas, controllable porosity, high adsorption capacity for certain
gases and good affinity with polymer chains, as well as the tuneable chemical
properties [14-16].
2008 [17]. UiO-66 based membranes have shown excellent gas separation
adsorption capacity [18-23]. Nik et al. reported CO2 permeability of the UiO-
membranes also show great potential in water reuse and desalination [24, 25].
of saline solutions [24, 25]. The excellent chemical stability was attributed to
110
was able to effectively remove most of Pb (II) (61.4%) and exhibited
chemical and thermal stability and tunable chemical properties [26-29], this
work aims to explore the feasibility of applying UiO-66 based MMMs for
uptakes and free volume size would be characterized in order to reveal the
actual mechanisms for the performance enhancement. This study may provide
useful insights to develop high performance UiO-66 based MMMs for alcohol
6.2 Experimental
three monomers 6FDA, HAB and DABA with HAB to DABA of 9:1. UiO-66
was synthesized and MMMs were prepared as the methods in Chapter 3. The
111
crystallographic structure of the UiO-66 nanoparticles was explored by XRD
was used to study the distribution of zirconium element. Thermal stability and
measurements and solvent uptake tests were carried out to reveal the
XRD. Figure 6.1 compares the tested XRD result with the simulated one. All
characteristic diffraction peaks match very well with the simulated UiO-66
structure of UiO-66 without other impurities. The peaks of the original MMM
of UiO-66 based MMMs, a MMM with 20wt% loading was soaked in pure
water at 60°C for one week followed by XRD testing. Remarkably, the peaks
of the MMM after soaking in pure water at 60°C for one week are the same as
the original MMM without any treatment, which means that UiO-66 can
maintain its structure in pure water at 60°C for one week. This finding
previous works [24, 25]. Figure 6.2 shows the FESEM images of UiO-66
112
particles after removing the solvent by drying them in a vacuum oven. The
MMM
UiO-66 Simulated result
5 15 25 35 45
2 Theta (º)
113
in the polyimide matrix. There is no apparent particle agglomeration even at
the highest Uio-66 loading (30 wt%). These observations demonstrate good
nanoparticles across the membrane. The result shown in Figure 6.3 further
Figure 6.3 (a) SEM image and (b) EDX mapping of Zr on the cross-section of
MMMs containing 30wt% UiO-66
Figure 6.4 and Figure 6.5 display the TGA curves and the simultaneous IR
spectra of gases released from UiO-66 nanoparticles and the MMM containing
of CO2 (2250-2400 cm−1) and H2O (3600-3750 cm−1) are evolved when the
the UiO-66 framework. These results confirm the superior thermal stability of
UiO-66 nanoparticles.
114
Figure 6.4 (a) Top-view of 3D TGA-FTIR analyses of UiO-66 and (b) its
temperature vs. weight percentage and heating time
For MMMs, the situation is more complicated compared with pure UiO-66
the heat treatment as explained in our previous study [31]. Figure 6.5(b) shows
that the MMM undergoes two stages of weight loss during the heat treatment.
The first stage of weight loss starts from 300°C to 520°C due to the
rearrangement of the pristine polyimide. Figure 6.5(a) reveals that -C=O peak
(1750 to 1850 cm−1) and -C-O peak (1100 to 1200 cm−1) firstly show up at
115
around 300°C, which suggests the beginning of the decarboxylation reaction
[31, 32]. The evolved CO2 peak (2250-2400 cm−1) between 300°C and 520°C
polyimide chains begin, which results in the second stage of weight loss.
Along with these decomposition processes, -C-O, CFx, CO, CO2 and H2O are
Besides the thermal stability, TGA curves can be used to determine the actual
under air atmosphere. Based on the remaining weight of ZrO2, the actual
100
PI (0%)
90 MMM (10%)
Weight percentage (%)
80 MMM (20%)
70 MMM (30%)
60
50
40
30
20
10
0
0 100 200 300 400 500 600 700 800
Temperature (ºC)
116
6.3.2 Solvent uptake analyses
The solvent uptake ratio gives the information about the affinity of water and
organic species towards the membranes. The solvent uptake tests were
by the sorption of tested solvents. Table 6.1 presents the solvent uptake ratios
particle loading increases, the water uptake ratio increases from 3.0% to
14.0%. This suggests that the addition of UiO-66 nanoparticles into MMMs
would provide a higher water solubility. On the other hand, all ethanol,
UiO-66, terephthalic acid, contains the hydrophilic functional groups that may
favour the sorption of water rather than alcohols. As a result, the water uptake
ratio increases while the organic species uptake ratio decreases with an
Table 6.1 Solvent uptake analyses for the pristine PI membrane and MMMs
UiO-66 Water Ethanol Isopropanol n-Butanol
Membrane loading uptake uptake uptake uptake
wt% ratio% ratio% ratio% ratio%
PI 0 3.0±0.8 11.9±0.2 20.0±1.2 14.6±0.3
MMM 10 4.9±0.5 9.5±0.4 15.1±1.0 14.2±0.5
MMM 20 8.4±1.2 8.7±0.4 11.3±0.2 14.0±0.5
MMM 30 14.0±1.5 7.5±0.5 8.5±0.7 13.2±0.1
Test temperature: 60 ºC
117
6.3.3 Positron annihilation lifetime spectroscopy analyses
carried out to investigate the free volume size and its distribution for the
membranes. In PALS, the o-Ps lifetime τ3 and its intensity I3 are related to the
size and concentration of the free volume, respectively [33-35]. Based on the
results of τ3 and I3, fractional free volume (FFV) of the MMMs can be
calculated from equations 2 and 3. Table 6.2 shows that τ3, I3, R and FFV all
matrix can create more and larger free volume, thus leads to an increase in
FFV. Remarkably, the FFV value is almost doubled when embedding 30wt%
product of diffusivity and solubility, and a larger FFV is beneficial for higher
the polyimide would likely increase the overall permeability for alcohol
dehydration.
Table 6.2 PALS data for the pristine PI membrane and MMMs
UiO-
66
Membrane τ3 (ns) I3 (%) R (Å) FFV (%)
loading
wt%
PI 0 1.928±0.029 4.416±0.084 2.788±0.023 0.722±0.032
MMM 10 2.015±0.027 5.041±0.080 2.870±0.020 0.898±0.033
MMM 20 2.164±0.025 5.767±0.074 3.001±0.017 1.175±0.035
MMM 30 2.304±0.023 6.298±0.069 3.119±0.015 1.441±0.037
118
6.3.4 Pervaporation performance
Figure 6.7, Figure 6.8 and Figure 6.9 present the ethanol, isopropanol and n-
exclude the influence of membrane thickness, the normalized flux (i.e., the
10000 100.0
Total normalized flux (gμm/m2h)
9000 90.0
permeate (wt%)
6000 60.0
5000 50.0
4000 40.0
3000 30.0
2000 Normalized Flux 20.0
1000 10.0
Water Concentration in Permeate
0 0.0
0 10 20 30
UiO-66 loading (%)
Figure 6.7 Ethanol dehydration performance of the pristine PI membrane
and MMMs at 60 ºC using ethanol/water (85/15wt%) as the feed
The normalized flux increases rapidly with the increase of UiO-66 content in
119
MMMs. For example, the highest normalized flux for isopropanol dehydration
is 4724.5 gμm/m2h when the MMM contains 30wt% UiO-66. This value is
more than two times higher than that of the pristine polyimide membrane. The
membranes with higher UiO-66 loadings display higher water uptake ratios
and higher water solubilities, which can allow more water molecules to be
PALS analyses. Table 6.2 demonstrates that membranes with higher UiO-66
loadings exhibit higher FFV values, which can help penetrants to diffuse faster
5000 100.0
Total normalized flux (gμm/m2h)
3000 96.0
2500 95.0
2000 94.0
1500 93.0
1000 Normalized Flux 92.0
500 Water Concentration in Permeate 91.0
0 90.0
0 10 20 30
UiO-66 loading (%)
120
5000 100.0
4500 99.0
Total normalized flux (gμm/m2h)
permeate (wt%)
3000 96.0
2500 95.0
2000 94.0
1500 93.0
1000 Normalized Flux 92.0
500 Water Concentration in Permeate 91.0
0 90.0
0 10 20 30
UiO-66 loading (%)
enlarge the free volume size and the FFV value of the membranes, as shown
not favorable for alcohol and water separation. On the other hand, a higher
UiO-66 loading may result in MMMs that possess higher solubility selectivity,
separation efficiency. These two effects compete with each other and may lead
is kept at 99.9% if the MMMs contain UiO-66 less than 20wt%, but drops
121
dehydration (Figure 6.9), the water concentration in the permeate stays
impressively high at 99.9% for all MMMs studied. The fundamental reasons
and n-butanol
Figure 6.10 plots and compares the normalized fluxes and water
system > n-butanol/water system. This trend is consistent with the reverse
molecular size can diffuse through the membrane easier and faster. Therefore,
the ethanol/water system shows the highest normalized flux, while the n-
system. This trend obeys the order of alcohols’ molecular sizes because the
swelling effect induced by the ethanol/water system may lead to the loosening
ethanol [36]. Furthermore, ethanol can form clusters with water and then pass
through the membranes together [6, 36]. This coupling effect also reduces the
122
separation efficiency of MMMs for ethanol dehydration. In summary, except
for the ethanol/water system, the newly developed MMMs are very effective
in the permeate are more than 99.7% for all MMMs studied.
a 10000
PI (0%)
9000
MMM (10%)
8000
Total normalized flux
MMM (20%)
7000 MMM (30%)
(gμm/m2h)
6000
5000
4000
3000
2000
1000
0
1 2 3 4 5
Carbon number of alcohols
b 100
98
Water concentration in the
96
94
permeate (wt%)
92
90
88
86 PI (0%)
84 MMM (10%)
MMM (20%)
82
MMM (30%)
80
1 2 3 4 5
Carbon number of alcohols
123
• The effect of feed temperature and feed composition on
temperatures of 40, 60 and 80°C are shown in Figure 6.11. The normalized
consistent with previous studies [36, 37]. On the other hand, the water
5000 100.0
Water concentration in permeate
Total normalized flux (gμm/m2h)
4500 95.0
4000 90.0
3500 85.0
3000 80.0
(wt%)
2500 75.0
2000 70.0
1500 65.0
Normalized Flux
1000 60.0
Water Concentration in
500 55.0
Permeate
0 50.0
20 40 60 80 100
Feed Temperature (ºC)
124
Table 6.3 Thermodynamic properties of feed mixtures as a function of
operating temperature
Saturated Saturated
vapor vapor Fugacity Fugacity
Feed Activity Activity
pressure pressure of of
Feed temp. coefficient coefficient
of of solvent water
(℃) of solvent of water
solvent water (kPa) (kPa)
(kPa) (kPa)
Isopropanol
/Water 40 1.164 1.901 14.047 7.387 13.898 2.106
MMM with 20wt% UiO-66 loading was tested under three different feed
concentrations: 75/25 wt%, 85/15 wt% and 95/5 wt% IPA/water. Figure 6.12
shows the results. As the feed IPA concentration increases, the normalized
flux shows a decreasing trend. The reduction of the normalized flux is mainly
due to the decrease of driving force as the feed IPA concentration increases.
Since the vacuum condition at the permeate side is maintained, the fugacity at
the permeate side can be neglected. Therefore, the normalized flux mainly
saturated vapour pressure and activity coefficient of species in the feed side.
As shown in Table 6.4, the fugacities of IPA and water at the permeate side
permeate can keep more than 99% when the feed concentration is lower than
85% but decrease to around 96% when the feed concentration increases to
95%. The drop of separation efficiency may be due to the swelling of the
125
6000 100.0
(wt%)
3000 75.0
70.0
2000
65.0
Normalized Flux
60.0
1000
Water Concentration in Permeate 55.0
0 50.0
70 75 80 85 90 95 100
Feed Concentration(%)
ethanol dehydration is presented in Table 6.5, Table 6.6 and Table 6.7 [31, 36-
49]. The newly developed MMM containing 30wt% UiO-66 not only displays
also possesses a much higher water permeability of 0.329 mg m-1 h-1 kPa-1
126
than most of others. For n-butanol dehydration, the 30/70 UiO-66/6FDA-
m-1 h-1 kPa-1 and an extremely high mole-based selectivity of 14214. For
Since permeance is the ratio of permeability to the active layer thickness [56],
Table 6.5 and Table 6.6 also indicate the newly developed membranes have
®
Matrimid 30-60 18% 100 0.490 50 [38]
®
Torlon 23 15% 60 0.011 3302 [39]
®
Ultem 24 15% 60 0.012 683 [39]
BPADA-ODA-
DABA 15-20 20% 60 0.145 9530 [40]
polyimide
Crosslinked
20 10% 60 0.835 609 [41]
Chitosan
®
Matrimid /MgO 30-60 18% 100 0.067 1513 [38]
MMM
P84/ZIF-90
19-22 15% 60 0.166 449 [37]
MMM
TR PBO 20-25 10% 80 0.299 670 [36]
C-TR-5-5 27 15% 60 0.150 4019 [31]
Polyimide/UiO- This
30 15% 60 0.329 2209
66 MMM study
127
Table 6.6 A comparison of pervaporation performance of membranes for n-
butanol dehydration
Water
Membrane Water
Temp. permeability Mole-based
Membrane thickness concentration Ref.
(ºC) (mg m-1 h-1 selectivity
(μm) in feed
kPa-1)
Crosslinked
10 10% 30 0.916 87 [42]
PVA
SYMPLEX 2 10% 25 3.881 153 [43]
®
Pervap 2510 0.5-2 15% 60 0.08 17 [44]
Crosslinked
20 10% 60 1.136 417 [41]
Chitosan
Crosslinked
29 10% 60 0.823 456 [46]
PVA
®
P84 --- 15% 60 0.034 1874 [47]
copolyimide
®
Pervap 2201 --- 10% 60 0.022 309 [48]
TR PBO 20-25 10% 25 1.254 186 [36]
Polyimide/UiO- This
30 15% 60 0.572 34.5
66 MMM study
128
6.4 Conclusions
ratio but a decrease in alcohol uptake ratio. Thus, the MMMs’ solubility
in MMMs.
(3) PALS data have confirmed that both free-volume radius R and
129
concentration in the permeate remains always high at 99.9% for all
MMMs.
(5) The normalized flux follows the trend of ethanol/water system >
6.5 References
C.A. Eckert, W.J. Frederick, J.P. Hallett, D.J. Leak, C.L. Liotta, J.R. Mielenz,
(2004) 1-21.
[4] G. Liu, W.S. Hung, J. Shen, Q. Li, Y.H. Huang, W. Jin, K.-R. Lee, J.-Y.
4521.
[5] L.M. Vane, A review of pervaporation for product recovery from biomass
130
[6] P. Shao, R.Y.M. Huang, Polymeric membrane pervaporation, J. Membr.
[7] T.C. Bowen, R.D. Noble, J.L. Falconer, Fundamentals and applications of
[8] Y.K. Ong, G.M. Shi, N.L. Le, Y.P. Tang, J. Zuo, S.P. Nunes, T.-S. Chung,
57 (2016) 1-31.
[10] L.Y. Jiang, Y. Wang, T.S. Chung, X.Y. Qiao, J.Y. Lai, Polyimides
(2009) 1135-1160.
2692-2696.
131
[16] J. Caro, Are MOF membranes better in gas separation than those made of
organic frameworks with exceptional stability, J. Am. Chem. Soc. 130 (2008)
13850-13851.
[19] M.W. Anjum, F. Vermoortele, A.L. Khan, B. Bueken, D.E. De Vos, I.F.J.
Maurin, Probing the dynamics of CO2 and CH4 within the porous zirconium
of the CO2/CH4 gas mixture within the porous zirconium terephthalate UiO-66
(2011) 13768-13774.
132
[23] S.J.D. Smith, B.P. Ladewig, A.J. Hill, C.H. Lau, M.R. Hill, Post-synthetic
[24] X. Liu, N.K. Demir, Z. Wu, K. Li, Highly water-stable zirconium metal–
9029-9032.
September 2016).
133
[31] Y.M. Xu, N.L. Le, J. Zuo, T.S. Chung, Aromatic polyimide and
325.
[32] W.L. Qiu, C.C. Chen, L.R. Xu, L.L. Cui, D.R. Paul, W.J. Koros, Sub-Tg
[33] S.J. Lue, D.T. Lee, J.Y. Chen, C.H. Chiu, C.C. Hu, Y.C. Jean, J.Y. Lai,
[34] H.M. Chen, W.S. Hung, C.H. Lo, S.H. Huang, M.L. Cheng, G. Liu, K.R.
Lee, J.Y. Lai, Y.M. Sun, C.C. Hu, R. Suzuki, T. Ohdaira , N. Oshima, Y.C.
[36] Y.K. Ong, H. Wang, T.S. Chung, A prospective study on the application
[37] D. Hua, Y.K. Ong, Y. Wang, T.X. Yang, T.S. Chung, ZIF-90/P84 mixed
134
[38] L.Y. Jiang, T.S. Chung, R. Rajagopalan, Matrimid®/MgO mixed matrix
[39] Y. Wang, L.Y. Jiang, T. Matsuura, T.S. Chung, S.H. Goh, Investigation
[42] M.C. Burshe, S.B. Sawant, J.B. Joshi, V.G. Pangarkar, Sorption and
(1997) 145-156.
135
[45] G.M. Shi, T.X. Yang, T.S. Chung, Polybenzimidazole (PBI)/zeolitic
[48] D. Van Baelen, B. Van der Bruggen, K. Van den Dungen, J. Degrève, C.
136
CHAPTER 7: FUNCTIONALIZED UiO-66-TYPE MOFS
PERVAPORATION
7.1 Introduction
attention due to the highly fluctuated oil price and global warming.
membranes are widely used for pervaporation because they have reasonable
between permeability and selectivity for polymeric membranes. They are also
137
separation performance. Compared with polymeric membranes, inorganic
membranes don’t suffer from the swelling problem because of their superior
applications are limited by the fragility and high fabrication cost. In order to
particles with an inherently high permeability and selectivity can help increase
the separation performance, while the polymeric matrices can minimize the
fragility issue and reduce the overall fabrication cost for MMMs [16,17].
capacity [18-20]. Among them, UiO-66 MOFs have been extensively studied
due to their noticeably high thermal, chemical and mechanical stability [21].
By altering ligands with various functional groups, one can easily manipulate
the pore size and chemical properties of UiO-66-Type MOFs [22,23]. UiO-66
MOF pores facilitated the CO2 transport [24]. Nik et al. found that the
138
ideal selectivity and CO2 permeability for CO2/CH4 separation [25]. The study
nitrogen dioxide from air with the aid of amine groups [26]. In addition, UiO-
66-NH2 based membranes also show great potential for water treatment and
aforementioned, there are limited studies for pervaporation [30, 31]. However,
the MMMs because the original polyimide membrane has high separation
the overall flux due to the high porosity and the positive influence of
of this work are to (1) explore the effect of functional groups of UiO-66-Type
MOF on alcohol dehydration and (2) investigate and compare their separation
loading.
139
MMMs for comparison. In addition, solvent uptake and positron annihilation
This study suggests UiO-66-NH2 based MMMs are very promising for
dehydration of alcohols.
7.2 Experimental
the MOFs particles and MMMs were examined by TGA. Solvent uptake and
positron PALS analyses as well as 200-hour stability tests were carried out to
reveal the fundamental science and underline mechanisms for the performance
enhancement.
synthesized particles. Figure 7.1 shows the XRD spectra of UiO-66-NH2 and
140
UiO-66-F4. All the characteristic diffraction peaks agree quite well with the
simulated result in the database [33], indicating that both UiO-66-NH2 and
similar to UiO-66.
40000
UiO-66-Simulated
35000 UiO-66-NH2
30000 UiO-66-F4
Intensity (a.u.)
25000
20000
15000
10000
5000
0
5 15 25 35 45
2 theta (º)
Figure 7.2 shows the morphology of the synthesized UiO-66-NH2 and UiO-66-
uniform with a particle size of 80-90nm while the size of UiO-66-F4 particles
pristine polyimide membrane and MMMs with various UiO-66-NH2 and UiO-
polyimide matrix and the synthesized nanoparticles. Figure 7.4 and Figure 7.5
141
show the EDX mapping of Zr across the MMMs and further confirms the
142
Figure 7.4 (a) SEM image and (b) EDX mapping of Zr on the cross-section of
MMMs containing 30wt% UiO-66-NH2
Figure 7.5 (a) SEM image and (b) EDX mapping of Zr on the cross-section of
MMMs containing 10wt% UiO-66-F4
TGA was carried out to study the thermal stability of the UiO-66-NH2
nanoparticles, pristine PI membrane and MMMs. Figure 7.6 and Figure 7.7
PI membrane and MMMs under air atmosphere. The TGA curves of UiO-66-
NH2 and UiO-66-F4 agree well with literatures [22, 34]. The weight loss from
200 °C is attributed to the removal of the residual DMF solvent in the particles.
the thermal behaviour of the pristine polyimide can be found in our previous
143
weight loss is mainly because of (1) the decomposition of UiO-66-type
pristine polyimide.
100
UiO-66-NH2
90 PI
MMM-10%
Weight percentage (%)
80
MMM-20%
70 MMM-30%
60
50
40
30
20
10
0
0 200 400 600 800
Temperature (℃)
In addition, the TGA results can be applied to calculate the actual MOFs
oxidized at 800°C under air atmosphere and the final ash is ZrO2. the actual
8.8wt%, 19.7wt% and 27.5wt%, respectively while the actual loading of UiO-
144
100
UiO-66-F4
90
PI
Weight percentage (%)
80
70 MMM-F4-10%
60
50
40
30
20
10
0
0 200 400 600 800
Temperature (℃)
solvents towards the membranes, solvent uptake tests were carried out at 60oC,
the same temperature used in the pervaporation tests. Table 7.1 shows the
methanol, ethanol and isopropanol, while Table 7.2 tabulates the solvent
UiO-66-NH2 loading, the water uptake ratio increases but the uptake ratios of
methanol, ethanol and isopropanol decrease. This means that the adding of
145
Clearly, the hydrophilic amine groups in UiO-66-NH2 nanoparticles provide
Table 7.1 Solvent uptake ratios of the pristine PI membrane and UiO-66-NH2
based MMMs
UiO-66- Water Methanol Ethanol
Isopropanol
NH2 uptake uptake uptake
Membrane uptake ratio
loading ratio ratio ratio
%
wt % % % %
PI 0 3.3±0.3 9.2±0.3 11.5±0.7 18.4±0.6
MMM 10 5.6±0.5 8.0±0.6 8.3±0.3 10.4±0.6
MMM 20 9.4±0.5 6.1±0.9 7.1±0.3 9.8±0.8
MMM 30 16.9±0.7 5.0±0.2 6.5±0.6 8.2±0.8
Table 7.2 Comparison of solvent uptake ratio among different UiO-66 MOF
based MMMs
Particle Water Ethanol
Membrane loading uptake ratio uptake ratio
wt % % %
MMM-UiO-66-NH2 10 5.6±0.5 8.3±0.3
MMM-UiO-66-F4 10 4.4±0.2 12.1±1.0
uptake ratio can be further confirmed by comparing the solvent uptake ratios
wt% loading. As shown in Table 7.2, the water uptake ratio follows the order
amine and hydrophobic fluorine functional groups. While the ethanol uptake
ratio follows the opposite order. These orders indicate that UiO-66-NH2 based
MMMs should have greater water solubility and higher solubility selectivity
towards water.
146
7.3.3 Positron annihilation lifetime spectroscopy analyses
investigate the free volume of the MMMs. The free volume size and its
concentration were characterized by the o-Ps lifetime τ3 and its intensity I3,
respectively [35-37]. Based on τ3 and I3, fractional free volume (FFV) can also
calculated. Table 7.3 summarizes the PALS data for the pristine polyimide
membrane and UiO-66-NH2 based MMMs. τ3, I3, R and FFV all increase with
UiO-66-NH2 nanoparticles.
Table 7.3 PALS data of the pristine PI membrane and UiO-66-NH2 based
MMMs
UiO-
66-NH2
Membrane τ3 (ns) I3 (%) R (Å) FFV (%)
loading
wt%
PI 0 1.650±0.038 3.020±0.100 2.510±0.036 0.360±0.028
MMM 10 1.769±0.034 3.449±0.089 2.633±0.030 0.475±0.028
MMM 20 1.773±0.029 4.187±0.092 2.637±0.025 0.579±0.029
MMM 30 1.913±0.029 4.439±0.084 2.774±0.023 0.715±0.032
For comparison, the PALS data of UiO-66-F4 based MMMs at 10wt% particle
loading were also measured. Table 7.4 shows that at the same particle loading,
UiO-66-HN2 based MMMs have a larger free volume size (R) and a higher
FFV than UiO-66-F4 based MMMs because UiO-66-HN2 has a larger pore
147
Table 7.4 Comparison of PALS data between UiO-66-NH2 and UiO-66-F4
based MMMs
Particle
Membrane loading τ3 (ns) I3 (%) R (Å) FFV (%)
wt%
MMM-
UiO-66- 10 1.769±0.034 3.449±0.089 2.633±0.030 0.475±0.028
NH2
MMM-
10 1.660±0.030 3.639±0.105 2.520±0.029 0.420±0.028
UiO-66-F4
dehydration
feed. All MMMs have a particle loading of 10wt%. To exclude the influence
that at the same particle loading, UiO-66-NH2 based MMMs have both higher
normalized flux and higher separation factor than that of UiO-66-F4 based
role on the dehydration performance of these MMMs. Not only the high FFV
of the UiO-66-NH2 MMMs facilitates the water transport, but also the
pores provide high water solubility and help water diffuse across the
membranes. Since the UiO-66-NH2 based MMMs have both the higher
148
normalized flux and separation factor, the UiO-66-NH2 based MMMs are the
alcohols
to two factors: (1) higher water solubility due to the presence of hydrophilic
existence of higher FFV values. As a result, the MMMs with a 30wt% UiO-
66-NH2 loading have the highest normalized flux of 8878 gμm/m2h for ethanol
149
dehydration which is more than three times of the normalized flux of the
12000 100
90
10000
80
70
8000
60
(wt%)
6000 50
40
4000
30
Normalized Flux 20
2000
10
Water Concentration in Permeate
0 0
0 10 20 30
UiO-66-NH2 loading (%)
10000 100
Total normalized flux (gμm/m2h)
9000 90
Water concentration in permeate
8000 80
7000 70
6000 60
(wt%)
5000 50
4000 40
3000 30
2000 Normalized Flux 20
1000 Water Concentration in Permeate 10
0 0
0 10 20 30
UiO-66-NH2 loading (%)
150
4500 100
3500 80
70
3000
60
(wt%)
2500
50
2000
40
1500
30
1000 Normalized Flux 20
500 10
Water Concentration in Permeate
0 0
0 10 20 30
UiO-66-NH2 loading (%)
loading range of 0–10 wt%, the FFV increment degree is relatively small and
increased, the FFV value becomes much higher especially at the highest
151
higher particle loadings. In addition, the up-and-down convex trend between
much small kinetic diameter than ethanol (i.e., 0.38 vs. 0.43 nm).
MMMs-20wt% are higher than that of the pristine polyimide membrane for
The isopropanol/water system shows the best separation performance and the
water concentration in permeate can keep impressively high at 99.9% for all
between water concentration and UiO-66-NH2 loading because IPA has a large
and isopropanol
Figure 7.11 replots the normalized flux and water concentration in permeate as
system. This order is consistent with the reverse order of alcohol molecular
sizes because a small alcohol size tends to diffuse faster across the membrane.
152
size can be easily intercepted and separated by the membrane and (2) an
alcohol with a higher linearity can pass through the membrane network more
easily [38, 39]. Since isopropanol has the largest molecular size and lowest
linearity among these three alcohols, the isopropanol/water system shows the
a
12000
PI (0%)
MMM (10%)
Total normalized flux
10000
MMM (20%)
8000 MMM (30%)
(gμm/m2h)
6000
4000
2000
0
0 1 2 3 4
Carbon number of alcohols
b
100
Water concentration in
80
permeate (wt%)
60
40 PI (0%)
MMM (10%)
20 MMM (20%)
MMM (30%)
0
0 1 2 3 4
Carbon number of alcohols
153
7.3.5 Long stability of UiO-66-NH2 based MMMs
factor to evaluate a membrane. Therefore, a 200-h test was carried out for IPA
7.12 presents the results. The MMMs display very stable performance without
obvious changes of flux and separation factor in the entire testing duration.
The good stability of UiO-66-NH2 based MMMs is mainly because the amine
bonding with imide groups, thus effectively reduce membrane swelling [24].
5000
4000
3500 80
3000
(wt%)
60
2500
2000
40
1500
1000 20
500
0 0
0 50 100 150 200
Test Duration (h)
154
7.3.6 Performance benchmarking
Table 7.6, Table 7.7 and Table 7.8 compare the dehydration performance of
now, methanol dehydration via pervaporation has not been widely used due to
the lack of suitable materials with both affordable price and good anti-swelling
7.6 shows that the MMM comprising 10 wt% UiO-66-NH2 has better
literature data. For isopropanol dehydration, the MMMs with 30wt% UiO-66-
NH2 not only show a comparable water permeability of 0.295 mg m-1 h-1 kPa-1
higher than most of others. Overall, the newly developed UiO-66-NH2 based
155
Table 7.6 A comparison of pervaporation performance for methanol
dehydration
Water Mole-
Membrane Water
Temp. permeability based
Membrane thickness concentration in Ref.
(ºC) (mg m-1 h-1 selectivity
(μm) feed
kPa-1)
Polyimide/UiO-66- This
NH MMM-20% 40 15% 60 0.455 28.8
2 study
Polyimide/UiO-66- This
NH MMM-30% 22 15% 60 0.599 19.8
2 study
Crosslinked
20 10% 60 1.136 417 [44]
Chitosan
Crosslinked PVA 29 10% 60 0.823 456 [49]
®
P84 copolyimide --- 15% 60 0.034 1874 [50]
®
Pervap 2201 --- 10% 60 0.022 309 [51]
156
Table 7.8 A comparison of pervaporation performance of membranes for
isopropanol dehydration
Water
Membrane Water Mole-
Temp. permeability
Membrane thickness concentration based Ref.
(ºC) (mg m-1 h-1
(μm) in feed selectivity
kPa-1)
®
Matrimid 30-60 18% 100 0.490 50 [40]
®
Torlon 23 15% 60 0.011 3302 [42]
®
Ultem 24 15% 60 0.012 683 [42]
BPADA-ODA-
15-20 20% 60 0.145 9530 [43]
DABA polyimide
Crosslinked
20 10% 60 0.835 609 [44]
Chitosan
P84/ZIF-90 MMM 19-22 15% 60 0.166 449 [46]
TR PBO 20-25 10% 80 0.299 670 [41]
C-TR-5-5 27 15% 60 0.150 4019 [32]
Polyimide/UiO-66- This
NH2 MMM-10% 34 15% 60 0.205 66250
study
Polyimide/UiO-66- This
NH2 MMM-10% 47 15% 60 0.265 66250
study
Polyimide/UiO-66- This
NH2 MMM-10% 19 15% 60 0.295 66250
study
7.4 Conclusions
157
matrix without visible agglomerations even for the highest loading of
30wt%.
provide more water sorption sites and work as water selective fillers. A
size (R) and higher FFV than that of UiO-66-F4 based MMMs.
can greatly improve both the normalized flux and separation factor of
20 wt%.
(5) The normalized flux shows the trend of methanol/water system >
158
7.5 References
C.A. Eckert, W.J. Frederick, J.P. Hallett, D.J. Leak, C.L. Liotta, J.R. Mielenz,
(2004) 1-21.
[4] G. Liu, W.S. Hung, J. Shen, Q. Li, Y.H. Huang, W. Jin, K.-R. Lee, J.-Y.
4521.
[5] L.M. Vane, A review of pervaporation for product recovery from biomass
(2015) 9799-9806.
[8] L.L. Xia, C.L. Li, Y. Wang, In-situ crosslinked PVA/organosilica hybrid
159
[9] A. Svang-Ariyaskul, R.Y.M. Huang, P.L. Douglas, R. Pal, X. Feng, P.
Chen, L. Liu, Blended chitosan and polyvinyl alcohol membranes for the
[10] H.A. Tsai, W.H. Chen, C.Y. Kuo, K.R. Lee, J.Y. Lai, Study on the
[11] Y.K. Ong, G.M. Shi, N.L. Le, Y.P. Tang, J. Zuo, S.P. Nunes, T.S. Chung,
57 (2016) 1-31.
[13] L.Y. Jiang, Y. Wang, T.S. Chung, X.Y. Qiao, J.Y. Lai, Polyimides
(2009) 1135-1160.
[14] T.C. Bowen, R.D. Noble, J.L. Falconer, Fundamentals and applications of
2692-2696.
160
[18] Z. Jia, G. Wu, Metal-organic frameworks based mixed matrix membranes
[20] J. Caro, Are MOF membranes better in gas separation than those made of
organic frameworks with exceptional stability, J. Am. Chem. Soc. 130 (2008)
13850-13851.
[24] M.W. Anjum, F. Vermoortele, A.L. Khan, B. Bueken, D.E. De Vos, I.F.J.
161
[26] G.W. Peterson, J.J. Mahle, J.B. DeCoste, W.O. Gordon, J.A. Rossin,
[28] B.J. Yao, W.L. Jiang, Y. Dong, Z.X. Liu, Y.B. Dong, Post‐Synthetic
toward stand-alone membranes for dye removal and separation, Chem. Eur.J.
22 (2016) 10565-10571.
[32] Y.M. Xu, N.L. Le, J. Zuo, T.S. Chung, Aromatic polyimide and
325.
162
[33] Crystallography Open Database, Information card for 4512072.
September 2016).
[34] J.B. DeCoste, G.W. Peterson, H. Jasuja, T.G. Glover, Y.G. Huang, K.S.
(2013) 5642.
[35] S.J. Lue, D.T. Lee, J.Y. Chen, C.H. Chiu, C.C. Hu, Y.C. Jean, J.Y. Lai,
[36] H.M. Chen, W.S. Hung, C.H. Lo, S.H. Huang, M.L. Cheng, G. Liu, K.R.
Lee, J.Y. Lai, Y.M. Sun, C.C. Hu, R. Suzuki, T. Ohdaira , N. Oshima, Y.C.
163
pervaporation from sulfonated polyphenylsulfone (sPPSU), J. Membr. Sci.
[41] Y.K. Ong, H. Wang, T.S. Chung, A prospective study on the application
[42] Y. Wang, L.Y. Jiang, T. Matsuura, T.S. Chung, S.H. Goh, Investigation
[46] D. Hua, Y.K. Ong, Y. Wang, T.X. Yang, T.-S. Chung, ZIF-90/P84 mixed
164
[47] M. Yoshikawa, K. Masaki, M. Ishikawa, Pervaporation separation of
(2002) 293-300.
[48] W.H. Chan, C.F. Ng, S.Y. Lam-Leung, X. He, Water–alcohol separation
[51] D. Van Baelen, B. Van der Bruggen, K. Van den Dungen, J. Degrève, C.
Y.M. Xu, S. Japip, T.S. Chung, Mixed Matrix Membranes with nano-sized
Press.
165
CHAPTER 8: CONCLUSIONS AND ECOMMENDATIONS
8.1 Conclusions
membranes with high separation performance and good stability have been
via pervaporation
166
for both alcohol dehydration cases. Fundamental characteristics of pore
structure, free volume, d-spacing, water contact angle, water and ethanol
PALS measurements show that swelling gives rise to a larger free volume
radius and a bigger FFV. Moreover, the adjacent small cavities in the
membranes at the dry state may coalesce to form large cavities at the wet state.
The H-PESU membrane shows the best ethanol dehydration performance due
PESU and flexible hydrophilic Pluronic moieties. It has the highest water
permeability of 0.322 mg m-1 h-1 kPa-1 which is 2-6 times higher than other
three membranes because the hydrophilic Pluronic moiety may form special
685 due to the rigid hydrophobic PESU moiety that prevents the membrane
167
8.1.2 Polyimide and Crosslinked Thermally Rearranged Poly(benzoxazole-
Pervaporation
results for isopropanol dehydration, have been obtained via in-situ thermal
168
preliminary results achieved in this study may offer useful insights for the
with a particle size of around 100nm were successfully synthesized and they
visible agglomeration even at the highest 30wt% loading. Adding UiO-66 into
volume radius and fractional free volume but also water solubility and
30wt%, the MMMs have the water permeability of 0.329 and 0.292 mg m-1 h-1
data. These experimental results strongly suggest the newly developed UiO-
169
66/polyimide MMMs have great potential for isopropanol and n-butanol
concentration in the permeate obeys the opposite trend. In addition, the study
of feed temperature effects shows that with the increase of feed temperature,
normalized flux increases and water concentration in permeate keeps high and
stable at 99.9%. The study of feed composition effects reveals that with the
Pervaporation
UiO-66-NH2 with particle size of 80-90nm and UiO-66-F4 with particle size
uniformly in the polyimide matrix without visible agglomerations even for the
170
isopropanol dehydration performances were studied and compared.
follows the reverse trend. The effects of functional groups of the MOFs
improve normalized flux of the MMMs due to the enhanced water solubility
and diffusivity, and at the same time enhance separation factor within the
particle loading range of 0-20wt% due to the higher selectivity towards water.
intermolecular bonding among their amine and imide groups. The UiO-66-
8.2 Recommendations
171
Table 8.1 A comparison of pervaporation performance of dense membranes
for ethanol dehydration
Water
Membrane Water Water Mole-based
Temp. permeability
Membrane thickness concentration concentration selectivity
(℃) (mg m-1 h-1
(μm) in feed in permeate (water/ethanol)
kPa-1)
membranes are usually preferred to get a higher flux. But in the asymmetric
difficulty in forming the defect-free selective layer due to its slow non-solvent
with other hydrophobic polymers e.g. PESU and adding non-solvent in dope
may be helpful for fast phase inversion and formation of defect-free selective
layer.
172
The other three PAE membranes i.e. PESU, PPSU and T-PESU are high
barrier materials and show high selectivity but low permeability. Therefore,
first treated with ozone for a certain time then surface initiated radical
hydrophilic layer will be formed on the top layer. The hydrophilic polymer
chains will increase the sorption of the water on the membrane surface and the
PAE layer can provide high selectivity so as to increase its water permeability
improve membrane flux. The porous support and the ultrathin selective layer
can greatly reduce the transport resistance and contribute to high flux while
the highly water-selective PAE layer would make sure the good separation
factor.
173
PI membrane shows moderate permeability and selectivity. On one hand, to
presented in Figure 8.1 is also applicable for PI membrane. Besides this way,
directly reacting with some chemicals with both hydrophilic functional groups
chemicals are shown in Figure 8.2. Polyimide can react with –NH2 in these
functional groups in these chemicals can help further increase the water
needed.
the condition of very high particle loading, particles may agglomerate and
voids may form at the particle-polymer interface. The MMMs become brittle
and difficult for handling and testing. Adding modulator with amine groups
e.g. 4-aminobenzoic acid, may be a good approach to solve this problem. The
174
presence of amine groups on the modulators may create a chemical reaction
between the MOF outer surface and the imide group of the polymer, thus
increase the compatibility between MOF particle and polyimide and result in a
stable membrane.
Last, all the developed membranes in this these are used for dehydration of
alcohols via pervaporation. In the future, one can also extend the application
also be applied for vapor permeation because vapor permeation is very similar
and the feed is in vapor phase. A higher temperature is beneficial for a higher
flux. Therefore, vapor permeation may reduce the required membrane size for
175
BIBLIOGRAPHY
PUBLICATIONS
[1] Y.M. Xu, N.L. Le, J. Zuo, T.S. Chung, Aromatic polyimide and
325.
[3] Y.M. Xu, Y.P. Tang, T.S. Chung, M. Weber, C. Maletzkoc, Polyarylether
[4] Y.M. Xu, S. Japip, T.S. Chung, Mixed Matrix Membranes with nano-sized
In Press.
CONFERENCE PROCEEDINGS
[1] Y.M. Xu, N.L. Le, J. Zuo, T.S. Chung, Aromatic polyimide and
176