 

ﬁg

.. .
3 IL; 1“. .

. ma... 9% m:

, e
.541
.‘.{.vtl.

 

 

 

,1 w DRILL
1

 

 

 

 

 

3g.ﬁqv-§vuw ‘ . 2”: Nu. on". i .

’ mm

 

 

.2
5443 :530
LIBRARY
MiChigan State Th‘ ' t rt'fyth tth
- - ISIS oce l a e
Ul‘llVG rsrty dissertation entitled

 

 

SELECTIVE, ULTRATHIN MEMBRANE SKINS PREPARED
BY DEPOSITION OF NOVEL POLYMER FILMS ON
POROUS ALUMINA SUPPORTS

presented by

Anagi Manjula Balachandra

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Chemistry

 

 

ﬂ ILL-u Lust/‘7

7 Major Professor’s We
1. / i r '
5/ .1 o/ A on 3
' I

Date

 

MSU is an Afﬁrmative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

OCT 2 II 2005

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.p65-p. 15

 

SELECTIVE, ULTRATHIN MEMBRANE SKINS PREPARED BY
DEPOSITION OF NOVEL POLYMER FILMS ON POROUS ALUMINA SUPPORTS

By

Anagi Manjula Balachandra

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHOLOSOPHY
Department of Chemistry

2003

ABSTRACT
SELECTIVE, ULTRATHIN MEMBRANE SKINS PREPARED BY
DEPOSITION OF NOVEL POLYMER FILMS ON POROUS ALUMINA SUPPORTS
By

Anagi Manjula Balachandra

Membrane-based separations are attractive in industrial processes because of their
low energy costs and Simple operation. However, low permeabilities oﬁen make
membrane processes uneconomical. Since ﬂux is inversely proportional to membrane
thickness, composite membranes consisting of ultrathin, selective skins on highly
permeable supports are required to Simultaneously achieve high throughput and high
selectivity. However, the synthesis of defect-free skins with thicknesses less than 50 nm
is difﬁcult, and thus ﬂux is often limited.

Layer-by-layer deposition of oppositely charged polyelectrolytes on porous
supports is an attractive method to synthesize ultrathin ion-separation membranes with
high ﬂux and high selectivity. The ion-transport selectivity of multilayer polyelectrolyte
membranes (MPMS) is primarily due to Donnan exclusion; therefore increase in ﬁxed
charge density should yield high selectivity. However, control over charge density in
MPMS is difﬁcult because charges on polycations are electrostatically compensated by
charges on polyanions, and the net charge in the bulk of these ﬁlms is small. To
overcome this problem, we introduced a templating method to create ion-exchange sites
in the bulk of the membrane. This strategy involves alternating deposition of a Cu”-
poly(acrylic acid) complex and poly(allylamine hydrochloride) on a porous alumina

support followed by removal of Cu” and deprotonation to yield free -COO' ion-exchange

sites. Difffusion dialysis studies Showed that the Cl'/SO42' selectivity of Cu2+-templated
membranes is 4-fold higher than that of membranes prepared in the absence of Cu”.
Post-deposition cross-linking of these membranes by heat-induced amide bond formation

further increased Cl‘/SO42' selectivity to values as high as 600.

Room-temperature, surface-initiated atom transfer radical polymerization (ATRP)
provides another convenient method for formation of utrathin polymer Skins. This
process involves attachment of polymerization initiators to a porous alumina support and
subsequent polymerization from these initiators. Because ATRP is a controlled
polymerization technique, it yields well-deﬁned polymer ﬁlms with low polydispersity
indices (narrow molecular weight distributions). Additionally, this method is attractive
because ﬁlm thickness can be easily controlled by adjusting polymerization time. Gas-
permeability data showed that grafted poly(ethylene glycol dimethacrylate) membranes
have a COz/CH4 selectivity of 20, whereas poly(Z-hydroxyethyl methacrylate) G’HEMA)
ﬁlms grown from a surface have negligible selectivity. However, derivatization of
PHEMA with pentadecaﬂuorooctanoyl chloride increases the solubility of C02 in the
membrane and results in a COZ/CH4 selectivity of 9.

Although composite PHEMA membranes have no signiﬁcant gas-transport
selectivity, diffusion dialysis studies with PHEMA membranes showed moderate ion-
transport selectivities. Cross-linking of PHEMA membranes by reaction with succinyl
chloride greatly enhanced anion-transport selectivities while maintaining reasonable ﬂux.
The selectivities of these systems demonstrate that alternating polyelectrolyte deposition
and surface-initiated ATRP are indeed capable of forming ultrathin, defect-free

membrane skins that can potentially be modiﬁed for speciﬁc separations.

Copyright by
ANAGI MANJULA BALACHANDRA
2003

To my dearest parents, Lalini and Henry Balachandra and my husband, Dhammika. Your
endless support and encouragement made this dissertation and doctoral degree possible.

ACKNOWLEDGMNETS

I would like to extend my deep appreciation and gratitude to my adviser, Prof.
Merlin Bruening, for his excellent guidance, support and encouragement. I was really
lucky to be in your group and enjoyed working with you. Thank you for being such a
wonderful adviser and a mentor.

I would also thankful to Prof. Gregory Baker for his expertise advices during the
polymerization project. I am also thankful for my guidance committee members; Profs,
Pinnavaia, Watson and Broderick for their time and suggestions.

I greatly acknowledge the ﬁnancial support from the NSF Center for Sensor
Materials at Michigan State University.

I am very much thankful to Bruening group members past and present; Dan,
Jeremy, Wenxi, Milind, Jinhua, Kangping, Yinda, Skanth, Sandra, Keith, Jacque, Bo
Young, Matt, Brian. J in, Lei, Sri, Christin and Xiaoyun. Especially, to Dr. Wenxi Huang
for teaching me many of the polymerization techniques. I also want to thank Baker group
members I B, Ying and Bao, it was very nice working with you.

Finally, I like to thank my family, especially, my parents for all their support,
encouragement and love. I owe you very much for every Single achievement in life.
Biggest thanks go to my husband, Dhammika for his constant support, understanding and

love. It want be possible for me to ﬁnish my Ph.D. without you.

vi

TABLE OF CONTENTS

List of Schemes

List of Tables

List of Figures

List of Abbreviations
Chapter 1: Introduction

1.1 Structure of the Dissertation

1.2 Formation and Development of Thin Polymer Films
1.3 Multilayer Polyelectrolyte Films

1.4 Ion Permeation through MPFS

1.5 MPFS as Ion-separation Membranes

1.6 Grafted Polymer Brushes

1.7 Atom Transfer Radical Polymerization (ATRP)

1.8 Formation of membranes by surface-initiated ATRP
1.9 Gas-Separation Membranes

1.10 Mechanism of Gas Transport through Polymer Membranes
1.11 Summary

1.12 References

Chapter 2: Enhancing the Anion-Transport Selectivity of Multilayer
Polyelectrolyte Membranes by Templating with Cu2+

2.1 Introduction

2.2 Experimental
2.2.1 Chemicals and Solutions
2.2.2 Film Preparation
2.2.3 Film Characterization
2.2.4 Ion-Transport Studies

2.3 Results and Discussion
2.3.] Synthesis and Characterization of Cu2+-Templated PAA/PAH
Films
2.3.2 Anion Transport through Cu2+-Templated PAA/PAH Membranes
2.3.3 Cross-linked Cu2+-Templated PAA/PAH Membranes
2.3.4 Changing the Surface Charge of Membranes
2.3.5 Anion-Transport through Partially Cross-linked,
Cu2+-Templated Membranes Deposited at Different pH values
2.3.6 Anion-Transport through Partially Cross-linked, Cu +-Templated
Membranes Deposited with Different Cu2+ Concentrations
2.3.7 Modeling of Anion-Transport through Cu2+-Templated PAA/PAH
Membranes
2.4 Conclusions
2.5 References and Notes

vii

xi
xii

xvi

36

36
39
39
4O
41
42

44
44
47

50
53

54
56
59

68
69

Chapter 3: Preparation of Composite Membranes by Atom Transfer
Radical Polymerization Initiated from a Porous Support

3.1 Introduction
3.2 Experimental
3.2.1 Chemicals and Solutions
3.2.2 Polymerization of EGDMA
3.2.3 Polymerization of HEMA and Subsequent Derivatization
3.2.4 Film Characterization on Gold Wafers
3.2.5 Film Characterization on Alumina Supports
3.2.6 Gas-Permeation Studies
3.3 Results and Discussion
3.3.1 Synthesis and Characterization of PEGDMA ﬁlms
3.3.2 Synthesis and Characterization of PHEMA ﬁlms
3.3.3 Gas Permeation through Polymer Membranes
3.3.4 Gas Permeation through Composite PEGDMA Membranes
3.3.5 Gas Permeation through PHEMA Membranes
3.4 Conclusions
3.5 References and Notes

Chapter 4: Ion Transport through Grafted Poly (2-hydroxyethyl
methacrylate) Membranes and their Derivatives

4.1 Introduction
4.2 Experimental

4.2.1 Chemicals and Solutions
4.2.2 Polymerization of HEMA from Gold-coated wafers and Porous
Alumina
4.2.3 Derivatization of PHEMA with Crown ethers
4.2.4 Chemical Cross-linking of PHEMA
4.2.5 F ilrn Characterization
4.2.6 Ion-Transport Studies with PHEMA Membranes
4.3 Results and Discussion
4.3.1 Synthesis and Characterization of PHEMA Membranes
4.3.2 Ion-Transport Studies with Composite PHEMA Membranes
4.3.3 Chemically Cross-linked PHEMA Membranes and their
Ion-Transport Properties
4.3.4 Crown Ether-Derivatized PHEMA Membranes and their
Ion-Transport Properties
4.4 Conclusions
4.5 References

viii

73

73
75

76
78
80
80
80
82
82
87
89
94
98
108
109

113

113
115

115

116
116
117
117
117
118
118
119

122
129

134
135

Chapter 5: Conclusions and Future Work

5. 1 Conclusions
5.2 Suggestions for Future Work
5.3 References

ix

136

136
138
140

Scheme 1.1

Scheme 3.1

Scheme 3.2

Scheme 3.3

Scheme 4.1

Scheme 4.2

LIST OF SCHEMES

Mechanism of ATRP.
Growth of cross-linked PEGDMA from an anchored initiator.

Growth of PHEMA brushes from a gold surface: (1) formation
of a monolayer of initiator and (2) polymerization from the
initiator-modiﬁed gold surface.

Derivatization of PHEMA with perﬂuorooctanoyl chloride.

Cross-linking of PHEMA by reaction with succinyl chloride.

F unctionalization of PHEMA with 2-hydroxy methyl 18-crown-6.

17

86

91
102

124

132

LIST OF TABLES

Table 2.1 Anion ﬂuxes (moles cm'zs'l) through bare porous alumina and
alumina coated with PAA/PAH and PAA-Cu/PAH ﬁlms cross-linked
at different temperatures. 51

Table 2.2 Thicknesses, ﬂuxes (moles cm‘zs’l), selectivities and Cu2+
concentrations (M) in partially cross-linked, 10.5-bilayer
PAA-Cu/PAH ﬁlms deposited at different pH values. 55

Table 2.3 Anion Fluxes (moles cm’zs’l), selectivities, thicknesses, and
estimated Cu2+ concentrations (M) of partially cross-linked
(130 °C) 10.5-bilayer PAA-Cu/PAH membranes and ﬁlms deposited
using different Cu + concentrations. 58

Table 2.4 Diffusion coefﬁcients obtained from modeling ion transport through
10.5-bilayer PAA-Cu/PAH membranes cross-linked at different
temperatures. 67

Table 3.1 Gas permeability coefﬁcients, estimated ﬁlm thicknesses, and idea
selectivities for PEGDMA membranes. 96

Table 3.2 Gas Permeability Coefﬁcients, Estimated Film Thicknesses and Idea
Selectivities for Underivatized PHEMA Membranes 100

Table 3.3 Estimated ﬁlm thicknesses, gas permeability coefﬁcients and ideal
selectivities for ﬂuorinated PHEMA membranes 107

Table 4.1 Film thicknesses, ﬂuxes and selectivities for composite PHEMA
membranes (porous alumina support). 121

Table 4.2 Hydrated radii and hydration energies of various ions. 123

Table 4.3 Fihn thicknesses, ﬂuxes and selectivities for cross-linked PHEMA
membranes in diffusion dialysis with single-salt solutions. 128

Table 4.4 Film thicknesses, normalized ﬂuxes and selectivities for

cross-linked PHEMA membranes in diffusion dialysis with mixed
salt solutions. . 130

xi

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4
Figure 1.5

Figure 1.6

Figure 2.1

Figure 2.2:

Figure 2.3

Figure 2.4

LIST OF FIGURES

Deposition of a Langmuir-Blodget ﬁlm from a ﬂoating
Langmuir monolayer. Figure adapted from.

Schematic diagram of the “Dip and Rinse” procedure for
alternating deposition of oppositely charged polyelectrolytes
on a charged substrate. In the actual ﬁlm structure, polycations
and polyanions are interdigitated.

Synthesis of Cu2+-templated polyelectolyte ﬁlms. Step 1:
deposition of a polycation (PAH) in the presence of
uncomplexed Cu2+. Step 2: deposition of partially Cu2+
-complexed PAA. Step 3: removal of Cu2+. Step 4:
deprotonation of the free carboxylic acid groups of PAA.
Repetition of steps 1 and 2 produes a multilayer ﬁlm.
Inertwining of layers is not shown for ﬁgure clarity.

Two methods for grafting of polymer ﬁlms onto solid surfaces.
Cartoon showing the different stages of surface-initiated ATRP.

Surface initiated atom transfer radical polymerization from (a)
polyelectrolyte deposited on alumina and (b) Au-coated alumina.

Preparation of Cu2+- templated polyelectrolyte ﬁlms on porous
alumina supports. Step l-adsorption of partially Cu2+-complexed
PAA on porous alumina. Step 2-adsorption of a polycation
(PAH). Step 3- removal of Cu“. Step 4-deprotonation of the
free carboxylic acid groups of PAA. Repetition of steps 1 and 2
produces multilayer ﬁlms. lntertwining of layers is not shown
for ﬁgure clarity.

Apparatus for ion-transport measurements.

Cyclic voltammetry of a MPA-modiﬁed gold electrode coated
with10 bilayers of PAH/PAA-Cu before (solid line) and after
exposing to pH 3.5 water (dotted line).

External reﬂectance FTIR spectra of (a) a 10-bilayer
PAH/PAA-Cu ﬁlm, (b) the same ﬁlm after exposure to

pH 3.5 water, (c) the ﬁhn after exposure to pH 5-6 water,
and (d) a 10-bi1ayer PAH/PAA ﬁlm deposited without Cu2+.
All ﬁlms were deposited on a gold wafer coated with a
monolayer of MPA.

xii

13

15

19

21

37

43

46

48

Figure 2.5

Figure 2.6

Figure 2.7

Figure 2.8

Figure 2.9

Figure 3.1

Figure 3.2

Figure 3.3

Receiving phase concentration as a function of time for a bare
porous alumina membrane (black), an alumina membrane coated
with 10.5 bilayers of Cu2+-templated FAA/PAH (open) and an

alumina membrane coated with 10.5 bilayers of FAA/PAH (grey).

Different symbols represent different salts: circles-NaCl and
squares-NaZSO4. The inset shows 8042' concentration vs. time
for pure PAA/PAH (grey) and templated PAA/PAH (open). The
membrane separates the receiving phase (initially deionized
water) from a 0.1 M salt solution.

External reﬂectance FTIR Spectra of IO-bilayer PAH/PAA-Cu
ﬁlms cross-linked at different temperatures: (a) unheated

(b) heated to 160°C. Films were deposited on MPA-coated
gold.

Cyclic voltammetry of 10-bi1ayer ﬁlms of PAH/PAA-Cu2+
deposited at different deposition pH values on MFA-modiﬁed
gold surfaces: pH 5.5-dashed line, pH 6-solid line, and

pH 6.6-dotted line. Areas of the reduction peaks were
calculated by drawing the base line from the current

value at a potential of 0.3V to the current value at -0.5V.

Cyclic voltammetry of IO-bilayer PAH/PAA-Cu ﬁlms

deposited with different Cu2+ concentrations: 2.5 mM Cu2+
-dotted line, 5.0 mM Cu2+—solid line and 7.5 mM Cu2+-dashed
line. The area of the reduction peak was calculated after drawing

the base line from the current at a potential of 0.3V to that
at -O.4V. Films were deposited at pH 6 on MPA-modiﬁed gold.

Schematic representation of the model used to simulate ion
transport through templated MPMS. The membrane consists

of two charged layers: a surface layer and the membrane bulk.
The line represents a hypothetical concentration proﬁle for

the excluded ion. me' and mez are the ﬁxed charge
concentrations of bulk of the ﬁlm and surface layer respectively.

Apparatus for gas-permeation measurements

Schematic diagram showing polymerization of EGDMA from a
polyelectrolyte surface modiﬁed with an initiator.

External reﬂection FTIR spectra of PAH/PSS/PAH on a
MPA-coated Au wafer before (a) and after reaction with an
acid bromide initiator (b). Spectrum (c) is from a PEGDMA
ﬁlm grown for 20 hours from PAH/PSS/PAH modiﬁed with

initiator.

xiii

49

52

57

60

61

81

83

84

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Figure 3.10

Figure 3.11

Figure 4.1

Figure 4.2

FESEM images of the ﬁltrate side of porous alumina (0.02 pm

surface pore diameter) (a) before depositing polyelectrolytes

(b) after depositing one bilayer of PSS/PAH, and

(c) after growth of PEGDMA from the surface for 20 hours.

Image ((1) is a cross-section of the membrane shown in (c). 88

Reﬂectance F TIR spectra of (a) a monolayer of
BrC(CH3)2COO(CH2)1 13); and (b) a graﬁed PHEMA layer. 90

Ellipsometric thicknesses of PHEMA ﬁlms vs polymerization time.
Thickness values are the average of three different ﬁlms, and the
error bars represent standard deviations. 92

Fluxes of different gases through a PEGDMA membrane (50 nm
thick) as a function of transmembrane pressure drop. The outlet
pressure was 1 atm and the measurements were performed at room
temperature. PEGDMA was deposited on porous almnina. 95

Fluxes of different gases through a PHEMA membrane

(60 nm thick) as a function of transmembrane pressure drop.

The outlet pressure was 1 atm and the measurements

were performed at room temperature. PHEMA was

deposited on porous alumina. 99

Reﬂectance FTIR spectra of a PHEMA ﬁlm grown from
(BrC(CH3)2COO(CH2)1 IS); on gold before (a) and after
reaction with pentadecaﬂuorooctanoyl chloride (b). 103

Transmission FTIR spectra of a PHEMA ﬁlm grown from
(BrC(CH3)2COO(CH2)1 IS); on gold-coated porous alumina

(a) before and (b) after reaction with pentadecaﬂuorooctanoyl
chloride. 104

Fluxes of different gases through a ﬂuorinated-PHEMA

membrane (100 nm thick) as a function of transmembrane

pressure drop. The outlet pressure was 1 atm and the

measurements were performed at room temperature. 106

Plot of receiving phase concentration as a function of time

in diffusion dialysis when the source phase (0.1 M salt) was

separated from the receiving phase (initially deionized water)

by a porous alumina substrate capped with 28 nm of PHEMA. 120

Reﬂectance F TIR spectra of a 28-nm thick grafted PHEMA

ﬁlm on a gold-coated wafer before (a) and after reaction with
succinyl chloride (b) and subsequent exposure to water (0). 125

xiv

Figure 4.3

Figure 4.4

Plot of receiving phase concentration as a function of time in

diffusion dialysis when the source phase (0.1 M salt) was

separated from the receiving phase (initially deionized water)

by a composite membrane of 35 nm cross-linked PHEMA ﬁlm.

The inset expands the concentration scale for 8042' and

Fe(CN)63'. PHEMA was cross-linked by succinyl chloride. 127

Reﬂectance FTIR spectra of (a) a grafted PHEMA layer

(28 nm) (b) CDI-derivatized PHEMA and (c) CDI-derivatized
PHEMA after reaction with 2-hydroxymethyl-18-crown-6

(43 nm). 133

XV

ATRP
bpy

CV
DMF
EGDMA
FESEM
FTIR
HEMA
MPA
MPF
MPM
PAA
PAA-Cu
PAH
PEGDMA
PHEMA
PSS

LIST OF ABBREVIATIONS

atom transfer radical polymerization
bipyridine

cyclic voltammogram
N,N-dimethylformamide

ethylene glycol dimethacrylate

ﬁeld emission scanning electron microscopy
fourier transform infrared spectroscopy
2-hydroxyethyl methacrylate
mercaptopropionic acid

multilayer polyelectrolyte ﬁlm
multilayer polyelectrolyte membrane
poly(acrylic acid)

poly(acrylic acid) copper complex
poly(allylamine hydrochloride)
poly(ethylene glycol dimethacrylate)
poly(2-hydroxyethyl methacrylate)
poly(styrene sulfonate)

xvi

CHAPTER 1

Introduction

1.1 Structure of the Dissertation

Membrane-based separations are attractive because of low energy costs and
simple Operation. However, low ﬂux through selective membranes oﬂen limits their
utility."2 The most common strategy for increasing ﬂux is the use of asymmetric or
composite membranes that consist of an ultrathin, selective skin on a highly porous
support. The minimal thickness of the Skin allows high ﬂux, while the support provides
mechanical strength.3 The selective skin should be aS thin as possible to achieve the
highest ﬂux, but synthesis of defect-free skins with thicknesses less than 50 nm is an
ongoing challenge.4 My research focused on the development of methods for formation
of defect-free, ultrathin polymer Skins and the use of these Skins for separations.
Speciﬁcally, this dissertation discusses the fabrication of ultrathin skins on porous
alumina supports using two techniques: alternating polyelectrolyte deposition and
grafting of polymers from a surface using atom transfer radical polymerization (ATRP).
These techniques afford composite membranes that allow selective ion and/or gas
transport.

Chapter 1 of this dissertation describes previous research on thin ﬁlm formation
with special emphasis on alternating polyelectrolyte deposition and polymerization from
a surface using atom transfer radical polymerization (ATRP). This chapter also contains
background information about membranes for gas and ion separations and sets forth the

motivation behind my research.

Chapter 2 concerns templating of multilayer polyelectrolyte ﬁlms (MPFS) to
enhance their ion-transport selectivities. I ﬁrst discuss some of the limitation of MPFS as
ion-separation membranes and then show how these challenges can be overcome by
using Cu2+ as a placeholder during polyelectrolyte deposition. Speciﬁcally, I utilized
Cu2+-chelated poly(acrylic acid) (FAA) and post-deposition removal of Cu2+ to introduce
ion-exchange Sites into the bulk of FAA/poly(allylamine hydrochloride) ﬁlms. Diffusion
dialysis studies showed 5-fold higher Cl’/SO42' selectivities with Cu2+-templated
membranes than with similar membranes deposited without Cu”. Post-deposition cross-
linking of Cu2+-templated membranes by heat-induced amide bond formation from
carboxylate and ammonium groups further increased C1'/SO42‘ selectivity to values as
high as 600. Ion-transport simulations suggest that both analyte size and analyte charge
are important in effecting selective transport.

The third chapter of the dissertation discusses formation of composite membranes
by polymerization from porous alumina and examines gas transport through these
systems. We synthesized cross-linked poly(ethylene glycol dimethacrylate) (PEGDMA)
and non cross-linked poly(2-hydroxyethyl methacrylate) (PHEMA) ﬁlms by ATRP ﬁom
initiators immobilized on porous alumina. Gas-permeation studies with PEGDMA ﬁlms
showed a COZ/CH4 selectivity of 20, whereas PHEMA ﬁhns exhibited a selectivity of
0.7. However, ﬂuorination of PHEMA through reaction of hydroxyl groups with
pentadecaﬂuorooctanoyl chloride increased the COZ/CH4 selectivity to 9.

Chapter 4 describes the performance of PHEMA and derivatized-PHEMA ﬁlms
as ion-separation membranes. Diffusion dialysis studies indicate that PHEMA

membranes have a Cl’/SO42' selectivity of 15, a K+/Mg2+ selectivity of 47 and a Fe(CN)(,3'

/Cl' selectivity of 160. Unlike polyelectrolyte membranes, PHEMA is a neutral polymer,
and the observed selectivities are probably due to differences in size or hydration
energies among the various ions. Although PHEMA has moderate selectivities, Cl'
ﬂuxes through these membranes were quite low. To achieve high selectivity while
maintaining reasonable ﬂux, we cross-linked very thin (28 nm) PHEMA ﬁlms by
reaction with a di-acid chloride. Diffusion-dialysis studies with these cross-linked
PHEMA membranes showed a Cl’/SO42‘ selectivity of 300 and only a 50% decrease in
C1' ﬂux compared to that of bare alumina.

Finally, chapter 5 contains conclusions and suggestions for future work.

1.2 Formation and Development of Thin Polymer Films

Development of thin organic ﬁlms has been a highly active area of research for
nearly 100 years.5'7 Coating a solid substrate with a thin ﬁlm plays a vital role in
controlling surface properties for applications in optics,8 sensingf”10 corrosion

protection,ll and separations,”14

and many of these applications require well-deﬁned
ﬁlms with uniform morphologies.7 5

Langmuir-Blodgett (LB) ﬁlms were the ﬁrst examples of multilayer coatings
prepared in a controlled layer-by-layer fashion.15 "6 The deposition of LB ﬁlms begins by
mechanically assembling an array of arnphiphilic molecules on a water surface. Once the

molecules are compressed to the desired organization, the ﬁlm can then be transferred to

a solid support as shown in Figure 1.1. This results in deposition of a single layer of

 

71% 359K
WRETEZWI <-

 

 

Figure 1.1: Deposition of a Langmuir-Blodgett ﬁlm from
a ﬂoating Langmuir monolayer. Figure adapted from
hgp://www.nes.coventry.ac.uk/research/cmbe/ﬁlmbal.htm.

 

molecules, and subsequent immersions in the trough result in the deposition of more
layers. LB ﬁhns have been used to investigate phenomena ranging from optical
properties of chromophores to ordered polymerizations on a substrate.7 However, these
coatings have two limitations which greatly restrict their applications. First, LB ﬁlms are
fragile because the layers are linked by weak van der Waals forces.17 Second, fabrication
of these materials requires pro-assembly at the air-water interface, so only arnpiphillic
molecules form ﬁlms.

A more recent step in the development of ultrathin ﬁlms was the discovery of
self-assembled monolayers (SAMS).5 ’18'2' The principle behind formation of these
monolayers is simple; a molecule containing a head group that adsorbs to a surface, e. g.
thiols on gold, assembles on the substrate under the constraints of intermolecular forces
and adsorption site geometry. Unlike LB ﬁlms, formation of SAMS doesn’t require any
pre-assembly, so their synthesis is simple and convenient. Additionally, because they are
bound to a surface, SAMS are more robust than LB ﬁlms.22

The most common family of SAMS is organothiols adsorbed on gold, and the ﬁrst
systematic study of these materials was done by Nuzzo and Allara in 1985.1823 Since
then, organothiol monolayers have been the subject of thousands of investigations. By
employing thiols with different tail groups, SAMS can be easily used to modify surface
properties. In addition to Au, other substrates such as Al/Ale3, Si/SiOz and Cu have
also been used to support SAMS.7’24

The main drawback to SAMS is the limited ﬁlm thickness available from

monolayer formation. Additionally, although self-assembled coatings are more

convenient and stable than LB ﬁlms, the stability of Au—thiol ﬁlms is still an issue at high
temperatures.25

More recently a number of techniques were developed to prepare multilayer ﬁhns.
26'” The most convenient of these methods is probably alternating deposition of anionic
and cationic polyelectrolytes on charged supports.3 "32 This strategy overcomes many of
the limitations imposed by LB ﬁlms and SAMS, although multilayered polyelectrolyte
ﬁlms are not well-ordered structures. Polyelectrolyte ﬁlms are very stable because of the
multiple electrostatic interactions between layers, and synthesis of these coatings requires

only a simple dip and rinse procedure. The versatility and simplicity of alternating

polyelectrolyte deposition are evident from the variety of charged substrates used for

33-36 33.37-44

deposition and the wide range of polyelectrolytes that can form these ﬁhns.
Because polyelectrolyte ﬁlms form the basis of much of my work, I describe them in

more detail below.

1.3 Multilayer Polyelectrolyte Films

Using colloidal alumina and colloidal silica, Iler45 ﬁrst demonstrated the
fabrication of multilayers by sequential adsorption of oppositely charged materials.
Later, Decher and Hong extended this idea and employed alternating deposition of
oppositely charged polyelectrolytes (e. g. poly(allylamine hydrochloride) and sodium
poly(styrenesulfonate)) on charged supports to form multilayer polyelectrolyte ﬁlms
(MPFS).31’32’46 Their technique allowed rapid formation of multilayer ﬁlms with control

over ﬁlm thickness on the nm scale. More recently, alternating polyelectrolyte deposition

has been demonstrated on a variety of charged substrates with polyelectrolytes ranging
from poly(styrenesulfonate)37’46’47 to DNA47 and charged viruses.38
Deposition of MPFS occurs as shown in Figure 1.2.48 The procedure begins with
immersion of a charged substrate into a solution containing an oppositely charged (with
respect to the substrate) polyelectrolyte. A ﬁlm forms due to electrostatic attraction
between the substrate and the polyelectrolyte, but the thickness of this ﬁlm is limited by
electrostatic repulsion of incoming chains by adsorbed polymer. (This picture is
oversimpliﬁed because the main driving force behind ﬁlm formation is actually the
increase in entropy that results when adsorption of a polyelectrolyte chain displaces many
counterions from the substrate surface).49 Rinsing of the substrate with water and
immersion in a second solution containing an oppositely charged polyelectrolyte then
yields another layer on the surface, and repetition of this adsorption sequence results in a
multilayer ﬁlm. Charge overcompensation at each deposition step is the key to
subsequent adsorption of an oppositely charged polyelectrolyte. Typically, ﬁlm thickness
increases linearly with the number of polyelectrolyte layers after deposition of the ﬁrst 3

50-52
or 4 layers.

Several features of alternating polyelectrolyte deposition make it a unique
technique for thin ﬁlm formation. First, this procedure is environmentally friendly
because in most cases, the solvent is water. Second, the only stipulation on the substrate
for ﬁlm formation is that it should contain sufﬁcient charge. Thus, substrates with a wide
range of topologies can support ﬁlm growth. The electrostatic interactions between
polyelectrolytes and surface charge also allow good adhesion between the polymer and

the substrate. Finally, the thickness of ﬁlms can be easily controlled by varying

a:
.3
n:
L
H
U)
.0
3
(I)

 

 

 

  

1. Polyanion ((39990 3. Polycation
O ——
2. Wash (9 Q (E) 4. Wash
03 ..:
G)

 

®®®®®®®®®®

Figure 1.2: Schematic diagram of the “Dip and Rinse” procedure for alternating
deposition of oppositely charged polyelectrolytes on a charged substrate. In the
actual ﬁlm structure, polycations and polyanions are interdigitated. Figure
adapted from reference 46.

50.53

deposition parameters such as the number of adsorption steps, the pH of deposition

solutions,54'55 deposition time,37 and the supporting electrolyte concentration in

deposition solutions.50’56'58

Since the introduction of MPF S, these materials have been subjected to numerous
studies to understand their structure and the mechanism of ﬁlm formation. Small-angle
X-ray scattering38 and UV/visible spectroscopy33'35’40 provided evidence for layer-by-
layer growth, but neutron reﬂectometry showed that multilayer ﬁlms are not highly
stratiﬁed. Strong interdigitation of polycations and polyanions occurs over several

neighboring layerssg’6O A variety of other methods were also used to characterize these

multilayers, including: electrochemical techniques,58 surface plasmon resonance,36’56’61

ellipsometry,58 quartz-crystal-microbalance gravimetr'y,36’62 X-ray photoelectron

57,63 14,34

. . 4 . .
spectroscopy, atomic force mrcroscopy,3 and scannlng electron microscopy.

Many recent studies showed that MPFS have potential applications in areas such

43,66-70 56,71,72

“‘65 non-linear optics, sensors, conductive

as light-emitting devices,

76,77 13,14,78-85

73,74 separation membranes

coatings, surface patterning,75 protective coatings,
and nano-particle formation.86 Some of these applications require impermeable ﬁlms
(e. g., protective coatings), while others necessitate ﬁlms with high or selective
permeabilities (e.g., separation membranes). In the speciﬁc area of separation
membranes, efﬁcient separation requires an understanding of the factors that affect

transport of molecules or ions through MPF S. Below, I discuss literature that relates

speciﬁcally to ion transport through MPFS.

1.4 Ion Permeation through MPFS

Several recent papers reported studies of ion permeation through MPFS.
Schlenoff and co-workers used electrochemical methods to examine the transport of
redox-active ions thorough a thin MPF on an electrode.87 Transport rates of redox-active
ions decreased with increasing charge, suggesting that diffusion through MPFS occurs via
hopping between ion-exchange sites. The need for more ion-exchange Sites to
compensate the charge on multivalent ions should result in fewer hopping sites for these
species, and hence slower transport. MOhwald and von Klitzing studied the transport of
neutral quenchers through polyelectrolyte ﬁlms by total internal reﬂection ﬂuorescence
spectroscopy.88 They Showed that the diffusion of molecules through the bulk of the ﬁlm
is much slower than through the outer layers. This probably occurs because layers in the
bulk of the ﬁlm pack more tightly than outer layers.

In situ ellipsometric and cyclic voltammetric studies done by our group showed that
the permeability of MPFS depends on the pH and ionic strength of polyelectrolyte
deposition solutions, the number of layers in the ﬁlm, and the nature of constituent
polymers.58 More recently, MOhwald and co-workers investigated the release of
ﬂuorescein from multilayered polyelectrolyte capsules and showed that the rate of release
is a function of the number of assembled polyelectrolyte layers.89’90 These studies

provide important background for designing separation membranes containing MPFS.

1.5 MPFS as Ion-separation Membranes

Several recent studies exploited the simplicity and versatility of layer-by—layer

adsorption of polyanions and polycations to form ion-separation membranes. Krasemann

10

and Tieke examined ion permeation through 60 bilayers of poly(sodium styrene
sulfonate) (PSS)/poly(allylamine hydrochloride) (PAH) deposited on a polymeric
support. These membranes exhibited a Nai'lMg2+ selectivity of 112 and a Cl'/SO42‘
selectivity of 45.79 Selectivity increased with both the number of bilayers and the ionic
strength of deposition solutions. Krasemann and Tieke suggested that the increase in
selectivity with the number of layers is due to the multi-bipolar nature of the
polyelectrolyte ﬁlm. They thought that electrostatic repulsion would occur at each
bilayer, resulting in higher monovalent/divalent ion selectivities for ﬁlms with many
layers. However, several studies suggest that polyelectrolyte bilayers are completely
intertwined and that the bulk of these ﬁlms have a net neutral charge.91 Thus, repulsion
of ions should occur only at the ﬁlm-solution interface. Perhaps large numbers of
bilayers resulted in membranes with fewer defects.

Our group Showed that membranes composed of 5 PSS/PAH bilayers on porous
alumina have a Cl'/SO42' selectivity of 6 and that Cl°/SO42’ selectivity does not increase
after deposition of an additional 5 bilayers.14 Selectivity is largely due to the electrostatic
exclusion of multiply charged ions by surface charge, as selectivity depends on whether
the membrane is terminated with a polycation or a polyanion. Stair et.al80 found that
upon changing the surface of PAA/PAH-capped ﬁlms from PAA to PAH, Cl’/SO42'
selectivity decreased from 360 to 2. This observation further conﬁrmed that outer layer
charge plays a dominant role in determining selectivity.

All of these previous studies suggest that electrostatic exclusion is a major factor
behind ion-transport selectivity in multilayer polyelectrolyte membranes. Therefore,

increasing the charge density in the bulk or at the surface of these membranes should

ll

enhance monovalent/divalent ion—transport selectivity. However, as mentioned, control
over charge density in the bulk of MPFS is difﬁcult because polycation charge is
essentially completely compensated by polyarrion charge, giving little net ﬁxed charge
density in interior of the ﬁlm.42'9| Insertion of net charge into MPF interiors likely
requires a post-deposition reaction.

We utilized Cu2+ as a template to create ion-exchange Sites (ﬁxed charge) in the
bulk of poly(acrylic acid) (PAA)/PAH membranes. This strategy involves alternating
deposition of PAA-Cu (1 Cu2+ per 8 COO‘ groups) and PAH on porous alumina supports
followed by removal of Cu2+ and deprotonation to yield free -COO' ion-exchange sites
(Figure 1.3). Difﬁrsion dialysis studies showed that the selectivity of Cu2+-templated
membranes is dramatically higher than that of membranes prepared in the absence of
Cu”, presumably due to the higher concentration of ﬁxed charge in the bulk of the ﬁlm.
Post-deposition cross-linking of these membranes by heat-induced amide bond formation
further increased Cl'/SO42' selectivity to values as high as 600. These remarkable

selectivities are achieved with no signiﬁcant decrease in ﬂux relative to pure FAA/PAH

membranes.

1.6 Grafted Polymer Brushes

Attachment of polymer chains to substrates provides another attractive way of
controlling surface properties.92'94 Assemblies of polymers that are linked to a surface
and yet highly extended into solution are often termed polymer brushes.95 These tethered
polymer chains have potential applications in chemical separations, sensing, stabilization

of colloidal suspensions, control of wetting and adhesion, corrosion resistance,

12

 

NH; [git-COO
coo-4%; coo
NH; 3‘)?!

   
   
 

O = 2 COO‘

000—ng— COO-

 

NH;
+H3

—coo- NH;
+H3 0

—COO- NH3
H3 NH;

—000H

(2)

 

O =§ coo-Cu2+-ooc 6

Figure 1.3: Synthesis of Cu2+-templated polyelectolyte ﬁlms. Step 1: deposition
0 a polycation (PAH) in the presence of uncomplexed Cu2+. Step 2: deposition of
IDar‘iially Cu2+-complexed PAA. Step 3: removal of Cu2+. Step 4: deprotonation of
e fl‘ee carboxylic acid groups of PAA. Repetition of steps 1 and 2 produces a
multilayer ﬁlm. lntertwining of layers is not shown for ﬁgure clarity.

microﬂuidics, fouling resistance and “chemical gating”.9("99 Methods for formation of

polymer brushes include covalent attachment and physisorption of chains to a substrate.
The latter method generally utilizes block copolyrners containing one block that strongly
interacts with the surface and a second block that forms the brush layer.100 In this case,
attachment to the surface occurs via van der Waals forces or hydrogen bonding, and thus
these brushes can be desorbed by good solvents or displaced by other polymers.
Covalent tethering of polymer brushes to a surface greatly enhances ﬁlm stability.
Covalent attachment of polymer brushes to a substrate can occur by either the

'0‘ In the case of “grafting to” a

“grafting to” or the “grafting from” method (Figure 1.4).
surface, functional groups on a pre-synthesized polymer react with groups on the
substrate to anchor the polymer.102 Although this technique provides strong adhesion
between the surface and the polymer, thicknesses of ﬁlms prepared in this way are
generally limited to less than 5 nm. After formation of a relatively thin ﬁlm, a diffusion
barrier prohibits more polymer molecules from reaching reactive surface Sites, resulting

in thin ﬁlms with low grafting densities.
The “grafting from” technique is attractive because of the potential for formation
of long polymer brushes with a high grafting density.93 In this technique, initiators are
ﬁrSt itIllnobilized on the surface, and polymer growth subsequently proceeds from these
Sites- Growth occurs as small monomers diffuse to the surface and reach growing chain
ends, and thus, in contrast to the “grafting to” technique, there is no large diffusion

baxrier to polymer growth.

Among the numerous polymerization techniques, free radical polymerization is

14

“Grafting to” a surface “Grafting from” a surface

Z

{jut—2 l

 

Figure 1.4: Two methods for grafting of polymer ﬁlms onto solid surfaces.
F igure adapted from reference 101.

the most widely used process because it is relatively easy to perform, and a wide range of
monomers can be used.'03 However, free radical polymerization provides limited control
over molecular weight and molecular weight distribution because of rapid radical-radical
termination reactions. This limitation inspired the emergence of new controlled/“living”
radical polymerization techniques.
The concept of “living” polymerization was ﬁrst introduced by Szwarc in 1956.104
The key feature of any “living” process is that chain growth proceeds without the
occurrence of irreversible termination steps, i.e., chain transfer, radical coupling, and
disproportionation, and molecular weight is a linear function of conversion. The ﬁrst
reported “living” procedure was the sequential anionic polymerization of two non-polar
monomers to produce block copolyrners.105 Since then, numerous studies demonstrated
contro lled/“living” polymerization techniques.103 In the mid 19908, two research groups
report ed the discovery of a controlled/“living” radical polymerization method that
employs transfer of a halide atom between a transition metal salt and a growing radical.
This process was termed atom transfer radical polymerization (ATRP).1°6’107 Today, of
the ma11y different “living” polymerization techniques (cationic, anionic, ring-opening,
and ni‘Zl‘oxide-mediated polymerization;108 and reversible addition fragmentation chain

transfer1 09'1”), ATRP is probably the most powerful, versatile and attractive technique.

1 '7 Atom Transfer Radical Polymerization (ATRP)

ATRP, as the name implies, is the reversible formation of a radical by transfer of
a -
hallde atom from an alkyl halide to a transition metal of low oxidation state (Scheme

1 - l 1
) ‘ 11 Upon transfer of the halide atom, the transition metal undergoes a one-electron

l6

oxidation. After formation, radicals can either propagate via reaction with monomer or

reform the dormant Species by abstraction of a halide atom from a metal-ion complex.

(311(1) complexes are the most common ATRP catalysts, but other transition metal ions
have also been used (Ru(u),‘°7 Fear),1 ‘2" ‘3 Ni(II),' '4" '5 Pd(II) and Rh(II)' '5). Monomers
polymerized using ATRP include styrenes, (meth)acrylates, acrylonitriles,
(meth)acrylamides, methacrylic acids and vinylpyridine.l '6 Compounds containing weak

carbon-halogen or hetero-halogen bonds, e.g., a-bromocarbonyl groups or sulfonyl

haI ides, serve as initiators.

 

k
’R° 4' Cu"le Ligand

R—x + CuIXILigand - ——§-
Initiator Catalyst kd Q4) ."~.‘kt
“‘ R-R

kp

Initiator - Alkyl halide M = Monomer

Catalyst — Transition metal (Cu, Fe, Ru, Ni, Pd)
complexed by one or more ligands

X = Halogen atom

Scheme 1.1: Mechanism of ATRP (adapted from reference 11 l).

ATRP is a controlled or “living” process when the atom-transfer equilibrium

Strongly favors the dormant species to give low radical concentrations. Because radical
Coupling and disproportionation kinetics are second order with respect to radical

oncentratron, termination In ATRP can be minimal compared to propagation, leading to

t
he f‘Ol‘mation of well-deﬁned polymers with low polydispersity. To control

Do
1 5"Itlerization, the transition metal/ligand ATRP catalyst is generally selected so that the

17

activation rate constant (k3) is much lower than the deactivation rate constant (kd).
Advantages of ATRP over conventional radical polymerizations include: compatibility
with a variety of functional monomers, tolerance to trace impurities (water, oxygen, and
i nhibitor), control over molecular weights and molecular weight distributions, and
possible block-copolymer formation by sequential activation of the dormant chain end in
t he presence of different monomers.

ATRP initiated from surfaces provides an attractive and convenient way to
synthesize dense, uniform polymer brushes with controllable thickness (Figure 1.5).
When initiators are covalently attached to a surface, atom-transfer results in the formation
of radicals on the substrate but not in solution, limiting unwanted polymerization in

”“21 also helps to avoid

solution. The ability to perform ATRP at room temperature

autopolymerization in solution, and thus extensive extraction of adsorbed polymer after

growth of a polymer brush is not necessary. Miminal solution polymerization is

especi ally important when synthesizing cross-linked polymer ﬁlms because cross-linked,
physisorbed polymer is difﬁcult to remove from a surface.1 '9 Recently, numerous reports
described the use of ATRP to grow polymer brushes from a variety of substrates in a

Well-deﬁned manner.l '7‘120’122'125 The wide range of monomers that can be used in ATRP

Shoald permit tailoring of the properties and composition of polymer brushes.

1 ‘8 Formation of membranes by surface-initiated ATRP

Like alternating polyelectrolyte deposition, surface-initiated ATRP provides a
ethod for formation of ultrathin membrane skins on porous alumina supports. This

Qess Involves Immobilization of an Initiator on porous alumma and subsequent

18

O—C—Br

1 1
O—C -
O—O—C-
O—O—O—c-
O—O—O—O—O—c-
O—O—O—O—O—C—Br

I l (dormant)

O—O—O—O—O—O—C-
O—O—O—O—O—O—O—C-

Time

 

CuBr/ligand

CuBr2 lligand
CuBr2 lligand
CuBr2 lligand
CuBr2 lligand
CuBr lligand

CuBr2 lligand
CuBr2 lligand
CuBr2 lligand

Figure 1.5: Cartoon showing the different stages of surface-initiated ATRP.

l9

polymerization from the anchored initiator sites (Figure 1.6). We used two strategies to
anchor the initiator to alumina. The ﬁrst method employed adsorption of a few layers of
charged polyelectrolytes and subsequent attachment of the initiator to the MPF surface,
while the second procedure involved gold sputtering followed by formation of a self-
assembled monolayer of a disulﬁde initiator. Polymerization from these immobilized
i ni tiators occurred using room-temperature ATRP. Using these procedures, we grew two
kinds of polymer ﬁlms, cross—linked poly(ethylene glycol dimethacrylate) (PEGDMA)
and non cross-linked poly(2-hydroxyethyl methacrylate) (PHEMA). Cross-linked
pol ytner ﬁlms are very attractive in separations because of their mechanical stability and
low free volumem'127 PHEMA ﬁlms are also attractive membrane materials because
their hydroxyl groups can be easily derivatized in high yields to tailor ﬁlms for speciﬁc
separreltions.118 The polymer grth from alumina was monitored by transmission-FT IR
spectroscopy and scanning electron microscopy (SEM). Both top-down and cross-
sectional SEM images of these polymer ﬁlms showed that they effectively covered the
surface pores of the alumina.
Gas permeation studies with cross-linked PEGDMA membranes showed a gas-
tr311513011 selectivity of 20 for CO; over CH4. In contrast, non cross-linked PHEMA
membranes exhibited minimal gas transport selectivities that depended primarily on the
“1013? masses of the permeating gases. However, after derivatization of the hydroxyl
groups of PHEMA with pentadecaﬂuorooctanoyl chloride, COz/CH4 selectivity increased
to N 9 . A detailed description of gas-separation membranes prepared by ATRP from
porous alumina supports will form chapter 3 of the dissertation. Below I give some

ba
cl(ground on gas-separation membranes to provide some context for my studies.

20

5 nm-Au
Coating \

 

Porous Alumina

 

 

 

 

Dnitiator Anchoring I l Initiator Anchoring—l
1* it. t t t} trim ///////

 

 

Polymerization

, l
Grafted Polymer
Film
(5?; {5.9; eteariogmeﬁe

 

 

    

 

Figure 1.6: Surface-initiated atom transfer radical polymerization from (a) a
Polyelectrolyte ﬁlm deposited on alumina and (b) a self-assembled initiator
mOnolayer on Au—coated alumina.

21

1.9 Gas-Separation Membranes

Gas separation with polymer membranes was initially described over a century
ago. Mitchell ﬁrst reported that different gases permeate through natural rubber at

different rates.108

In 1866, Graham demonstrated the enrichment of air with O; by
permeation through a natural rubber membrane.128 He showed that a mixture of gases
could be separated according to their molecular weights by permeation through a
microporous membrane, and the proportionality of gas ﬂux to the reciprocal of the square
root of molecular weight later became the well-known Graham’s Law of Diffusion.
Subsequently, there was little development of gas-separation membranes until the 1960’s
when Loeb and Souirajan129 invented the ﬁrst asymmetric membrane of industrial
interest. This membrane, which was prepared from cellulose acetate by phase inversion,
was originally made for desalination of water and later modiﬁed for gas separation. The
asymmetric membrane contained a thin, dense, selective skin at the surface of a highly
porous material, and this structure allowed much greater ﬂuxes than thick, homogeneous
membranes. The dense skin also exhibited higher selectivities than the porous structures
that behaved according to Graham’s law, while the underlying porous material provided
mechanical strength.

The era of commercial gas separations began in the 1970’s. Monsanto initiated
the ﬁrst large-scale separation of gases in 1977 for the recovery of H; from industrial gas
streams using membranes made of polysulfone. In the 19808, Perrnea introduced the

prism membrane for separation and recovery of hydrogen from purge gas streams of

ammonia plants. This was a polysulfone membrane coated with silicone rubber. In the

22

mid 1980s, Cyanara, Separex and GMS used dried cellulose acetate membranes for
removal of CO2 from natural gas, and in 1982, Generon produced the ﬁrst N2/Air
separation membrane using poly(4-methyl-l-pentene). This system had an O2/N2
selectivity of ~ 4. In the mid 19903, Generon, Praxair and Medal produced a polyimide
membrane for O2/N 2 separation with a selectivity of 6-8.3 Other recent developments in
gas-separation membranes include the commercialization of composite membranes. 3
Composite membranes are made by depositing a thin layer of polymer on a highly porous
substrate. The thin selective layer acts as a discriminating ﬁlm to give selectivity, and the
porous layer provides mechanical stability to the system. Composite membranes have
the advantage that only a small amount of expensive skin material is needed, while the
porous support can be made from an inexpensive polymer.

Membrane geometry is also critical for practical separations, as surface area must
be maximized. Asymmetric membranes can be packed as hollow ﬁbers or in a spiral-
wound conﬁguration. A spiral-wound module consists of series of membrane envelopes,
and each envelope consists of two membrane sheets, which are separated by a feed
spacer. In a hollow ﬁber module, asymmetric hollow ﬁbers are bundled together to
achieve a very high surface area. Some commercial hollow ﬁber modules contain more
than a million hollow ﬁbers. 130’” 1

Current membrane-based gas separations include a wide range of applications
such as recovery of H2 from synthesis gases and petrochemical process streams, removal
of CO2 from mixtures of hydrocarbons and natural gases, N2 or 02 enrichment from air,

S02 removal from smelter gas streams, H28 and water removal from natural gas and air

- - - 107
streams, and NH3 removal from recycle streams In arrunonra synthesrs. These

23

processes use both composite and asymmetric membranes as well as several different

membrane geometries including hollow ﬁber and Spiral wound systemsf”132
Of special relevance to this thesis, several studies demonstrated the possibility of

using various poly(alkyl methacrylates) in gas separation membranes, e. g., poly(ethyl

‘33 and a styrene/methacrylate co-

methacrylate),4 poly(tert-butyl methacrylate),
polymer.134 Yoshikawa and co-workers Showed that the presence of amine moieties in
poly(methacrylate) ﬁlms greatly enhances CO2/N2 separation.135 The use of ﬂuorinated
poly(methacrylates) provides another way to enhance CO2/N2 or CO2/CH4 selectivity.‘36
Most previous studies with poly(alkyl methacrylates) employed cast membranes with
large thicknesses, and decreasing the thickness of these ﬁlms would greatly enhance ﬂux.
Although many successes have been achieved in gas separation membrane research,
fabrication of utrathin (<50 nm) polymer skins is still a challenge, so the focus of this
work was to develop methods for deposition of selective, ultrathin polymer skins.

To develop either thin or thick membrane materials, one needs to understand the
factors affecting gas separation and the mechanism of separation. Below I discuss the

mechanism of gas transport through polymer membranes and the factors that determine

selectivity for one gas over another.

1.10 Mechanism of Gas Transport through Polymer Membranes

Gas-transport selectivity is usually based on one of three mechanisms: Knudsen

'37 Knudsen diffusion dominates when

diffusion, molecular sieving or solution/diffusion.
membrane pores are larger than the gas molecules being separated but smaller than the

mean free path of these gases. Permeation rates are inversely proportional to the square

24

root of the gas molecular weight of the gas so selectivity is the reciprocal of the square
root of the ratio of the molecular weights of the gases being separated.‘38 Because most
gases of interest have similar molecular weights, separations based on Knudsen—diffusion
are not highly selective.

Molecular sieving of common gases occurs when pore diameters in a membrane
are smaller than 7 A. Selectivities between gases of different sizes can be nearly inﬁnite

”9"40 This method, however, is limited by

when one gas is incapable of entering a pore.
relatively low ﬂuxes and the difﬁculty of preparing defect-free membranes with uniform
pore Sizes.

Solution-diffusion is the most common mechanism that operates in practical gas
separations. In this mechanism, transport of gases occurs in three steps: sorption of the
penetrant into the polymer ﬁlm at the high-pressure interface, diffusion of the penetrant
through the polymer ﬁlm, and ﬁnally, desorption at the permeate Side (low-pressure
interface). Thus, gas ﬂux depends on the diffusivity and solubility of the gas in the
polymer as well as the transmembrane partial pressure gradient. F ick’s ﬁrst law
(equation 1.1) describes the transport of a species within a nonporous membrane. Flux, J,
is proportional to the concentration gradient, dc/dx, and the diffusion coefﬁcient, D, for

the molecule in the membrane.

dc
=-D—- 1.1
J dx

According to Henry’s law, the concentration, C, of a speciﬁc gas in the membrane at the
high- or low-pressure interface is proportional to partial pressure, p, and the solubility

coefficient, S (equation 1.2).

25

C=Sp L2
Using Henry’s law to determine concentrations at the two gas-membrane interfaces, and
assuming steady-state ﬂux and constant values of D and S allows transformation of
equation 1.1 to equation 1.3.

J = D S Ap/d 1.3
In this equation, Ap is the partial pressure difference between the feed and permeate, and
6 is the membrane thickness. The permeability coefﬁcient, P, of a particular membrane
for a speciﬁc gas is then deﬁned by equation 1.4.

P = D S = J d/Ap 1.4
Selectivity for one gas over another, (1,413, is given by the ratio of the permeability
coefﬁcients of two gases (equation 1.5). This ratio is a measure of the relative ﬂuxes of
the two gases at the same driving force. Since permeability coefﬁcients depend on both
solubility and diffusion coefﬁcients, selectivity also contains diffusivity (DA/DB) and
solubility (S A/SB) components. Diffusion selectivity generally favors small molecules,

while solubility selectivity favors more condensable gases.

_£-&§_A_

- 1.5
a“ P, DES,

Both selectivity and permeability determine membrane performance, and thus the careful

selection of polymer materials is vital for efﬁcient separations.

26

1.1 1 Summary

As discussed earlier, the main objective of this work is the fabrication of defect-
free, ultrathin polymer skins on porous supports and the use of these composite
membranes in ion and gas separation to achieve high selectivity along with high ﬂux. In
this chapter, I tried to Show the need for ultrathin polymer skins in separation membranes
and the challenges in forming these polymer skins with controllable thicknesses. I also
discussed background on the development of polyelectrolyte ﬁlms and polymerization
from a surface, the theory behind gas transport through polymer membranes and some

relevant past research done on the development of gas and ion separation membranes.

27

1 .1 l
(1)
(2)
(3)
(4)
(5)
(6)

(7)

(8)

(9)

(10)

(11)
(12)

(13)

(14)

(15)
(16)

(17)

(18)

References

Spillman, R. W. Chem. Eng. Prog. 1989, 85, 41-62.

Henis, J. M. S.; T ripodi, M. K. Science 1983, 220, 11-17.

Baker, R. W. Ind. Eng. Chem. Res. 2002, 41, 1393-1411.

Chiou, J. S.; Paul, D. R. J. Membr. Sci. 1989, 45, 167-189.

Ulman, A. Chem. Rev. 1996, 96, 1533-1554.

Swalen, J. D.; Allara, D. L.; Andrade, J. D.; Chandross, E. A.; Garoff, S.;
Israelachvili, J.; McCarthy, T. J .; Murray, R.; Pease, R. F.; et a1. Langmuir 1987,
3, 932-950.

Ulman, A. Introduction to Ultrathin organic ﬁlms: from Langmuir-Blodgett to
self-assembly; Academic Press: Boston, 1991.

Bubeck, C.; Neher, D.; Kaltbeitzel, A. Makromol. Chem, Macromol. Symp. 1990,
3 7, 239-245.

Miller, L. S.; McRoberts, A. M.; Walton, D. J .; Peterson, 1. R.; Parry, D. A.;
Sykesud, C. G. D.; Newton, A. L.; Powell, B. D.; Jasper, C. A. Thin Solid Films
1996, 284-285, 927-931.

Petty, M. C. Biosensors Bioelectron. 1995, 10, 129-134.

Whitesides, G. M.; Laibinis, P. E. Langmuir 1990, 6, 87-96.

Stroeve, P.; Vasquez, V.; Coelho, M. A. N.; Rabolt, J. F. Thin Solid Films 1996,
284-285, 708-712.

van Ackem, F.; Krasemann, L.; Tieke, B. Thin Solid Films 1998, 327-329, 762-
766.

Harris, J. J.; Stair, J. L.; Bruening, M. L. Chem. Mater. 2000, 12, 1941-1946.
Langmuir, I. J. Am. Chem. Soc. 1917, 39, 1848-1906.
Blodgett, K. B. J. Am. Chem. Soc. 1935, 57, 1007-1022.

Embs, F.; Funhoff, D.; Laschewsky, A.; Licht, U.; Ohst, H.; Prass, W.; Ringsdorf,
H.; Wegner, G.; Wehrmann, R. Adv. Mater. 1991, 3, 25-31.

Allara, D. L.; Nuzzo, R. G. Langmuir 1985, 1, 45-52.

28

(19)

(20)

(21)

(22)

(23)

(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)
(32)

(33)

(34)

(35)

(36)

Allara, D. L.; Fowkes, F. M.; Noolandi, J.; Rubloff, G. W.; Tirrell, M. V. Mater.
Sci. Eng. 1986, 83, 213-226.

Bishop, A. R.; Nuzzo, R. G. Curr. Opin. Coll. Int. Sci. 1996, I, 127-136.

Nuzzo, R. G.; Fusco, F. A.; Allara, D. L. J. Am. Chem. Soc. 1987, 109, 2358-
2368.

Knoll, W.; Matsuzawa, M.; Offenhausser, A.; Riihe, J. Isr. J. Chem. 1997, 36,
357-369.

Allara, D. L.; Nuzzo, R. G. Langmuir 1985, 1, 52-66.

Bain, C. D.; Troughton, E. B.; Tao, Y. T.; Evall, J .; Whitesides, G. M.; Nuzzo, R.
G. J. Am. Chem. Soc. 1989, 111, 321-335.

Bain, C. D.; Evall, J.; Whitesides, G. M. J. Am. Chem. Soc. 1989, 111, 7155-
7164.

CaO, G.; Hong, H. G.; Mallouk, T. E. Acc. Chem. Res. 1992, 25, 420-427.

Fang, M.; Kaschak, D. M.; Sutorik, A. C.; Mallouk, T. E. J. Am. Chem. Soc.
1997, 119,12184-12191.

Kim, H.-N.; Keller, S. W.; Mallouk, T. E.; Schmitt, J.; Decher, G. Chem. Mater.
1997, 9,1414-1421.

Kohli, P.; Blanchard, G. J. Langmuir 1999, 15, 1418-1422.

Huc, V.; Bourgoin, J .-P.; Bureau, C.; Valin, F.; Zalczer, G.; Palacin, S. J. Phys.
Chem. B 1999, 103, 10489-10495.

Decher, G.; Hong, J. D. Makromol. Chem, Macromol. Symp. 1991, 46, 321-327.
Decher, G.; Hong, J. D. Ber. Bunsenges. Phys. Chem. 1991, 95, 1430-1434.

Lvov, Y.; Ariga, K.; Ichinose, 1.; Kunitake, T. J. Am. Chem. Soc. 1995, I I 7,
6117-6123.

Caruso, F .; Furlong, D. N.; Ariga, K.; Ichinose, 1.; Kunitake, T. Langmuir 1998,
14, 4559-4565.

F erreira, M.; Rubner, M. F. Macromolecules 1995, 28, 7107-7114.

Caruso, F .; Niikura, K.; Furlong, D. N.; Okahata, Y. Langmuir 1997, I3, 3427-
3433.

29

(37)

(38)

(39)

(40)

(41)
(42)

(43)

(44)

(45)
(46)
(47)
(48)

(49)

(50)

(51)

(52)

(53)

(54)

Lvov, Y.; Decher, G.; MOehwald, H. Langmuir 1993, 9, 481-486.

Lvov, Y.; Haas, H.; Decher, G.; MOehwald, H.; Mikhailov, A.; Mtchedlishvily,
B.; Morgunova, E.; Vainshtein, B. Langmuir 1994, 10, 4232-4236.

Lvov, Y. M.; Lu, Z.; Schenkman, J. B.; Zu, X.; Rusling, J. F. J. Am. Chem. Soc.
1998, 120, 4073-4080.

Sukhorukov, G. B.; MOehwald, H.; Decher, G.; Lvov, Y. M. Thin Solid Films
1996, 284-285, 220-223.

Schlenoff, J. B.; Laurent, D.; Ly, H.; Stepp, J. Adv. Mater. 1998, 10, 347-349.
Laurent, D.; Schlenoff, J. B. Langmuir 1997, 13, 1552-1557.

Lindsay, G. A.; Roberts, M. J .; Chaﬁn, A. P.; Hollins, R. A.; Merwin, L. H.;
Stenger—Smith, J. D.; Yee, R. Y.; Zarras, P.; Wynne, K. J. Chem. Mater. 1999, 11,
924-929.

Lvov, Y.; Munge, B.; Giraldo, 0.; Ichinose, 1.; Suib, S. L.; Rusling, J. F.
Langmuir 2000, 16, 8850-8857.

Iler, R. K. J. Coll. Int. Sci 1966, 21, 569—594.

Decher, G.; Hong, J. D; Schmitt, J. Thin Solid Films 1992, 210/211, 831-835.
Lvov, Y.; Decher, G.; Sukhorukov, G. Macromolecules 1993, 26, 5396-5399.
Decher, G. Science 1997, 277, 1232-1237.

Bertrand, P.; Jonas, A.; Laschewsky, A.; Legras, R. Macromol. Rapid. Commun.
2000, 21, 319-348.

Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153-8160.

Laschewsky, A.; Mayer, B.; Wischerhoff, E.; Arys, X.; Jonas, A. Ber. Bunsenges.
Phys. Chem. 1996, 100, 1033-1038.

Arys, X.; Jonas, A. M.; Laguitton, B.; Laschewsky, A.; Legras, R.; Wischerhoff,
E. Thin Solid Films 1998, 327-329, 734-738.

Hoogeveen, N. G.; Stuart, M. A. C.; Fleer, G. J.; Boehmer, M. R. Langmuir 1996,
12, 3675-3681.

Yoo, D.; Shiratori, S. S.; Rubner, M. F. Macromolecules 1998, 31, 4309-4318.

30

(55)

(56)

(57)

(58)

(59)

(60)

(61)

(62)

(63)
(64)

(65)

(66)

(67)

(68)

(69)

(70)

(71)

Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213-4219.

Caruso, F.; Niikura, K.; Furlong, D. N.; Okahata, Y. Langmuir 1997, 13, 3422-
3426.

Sukhorukov, G. B.; Schmitt, J .; Decher, G. Ber. Bunsenges. Phys. Chem. 1996,
100, 948-953.

Harris, J. J .; Bruening, M. L. Langmuir 2000, 16, 2006-2013.

Schmitt, J .; Griienewald, T.; Decher, G.; Pershan, P. S.; Kjaer, K.; LOesche, M.
Macromolecules 1993, 26, 7058-7063.

LOesche, M.; Schmitt, J.; Decher, G.; Bouwman, W. G.; Kjaer, K.
Macromolecules 1998, 31 , 8893-8906.

Kurth, D. G.; Osterhout, R. Langmuir 1999, 15, 4842-4846.

Lvov, Y.; Ariga, K.; Onda, M.; Ichinose, 1.; Kunitake, T. Collid Surf A 1999,
146, 337-346.

Chen, W.; McCarthy, T. J. Macromolecules 1997, 30, 78-86.
Liu, Y.; Ma, H.; Jen, A. K. Y. Chem Commun. 1998, 2747-2748.

Wu, A.; Yoo, D.; Lee, J. K.; Rubner, M. F. J. Am. Chem. Soc. 1999, 121, 4883-
4891.

Koetse, M.; Laschewsky, A.; Mayer, B.; Rolland, 0.; Wischerhoff, E.
Macromolecules 1998, 31 , 9316-9327.

Laschewsky, A.; Wischerhoff, E.; Kauranen, M.; Persoons, A. Macromolecules
1997, 30, 8304-8309.

Roberts, M. J .; Lindsay, G. A.; Herman, W. N.; Wynne, K. J. J. Am. Chem. Soc.
1998, 120, 11202-11203.

Wang, X.; Balasubramanian, S.; Kumar, J .; Tripathy, S. K.; Li, L. Chem. Mater.
1998, 10, 1546-1553.

Balasubramanian, S.; Wang, X.; Wang, H. C.; Yang, K.; Kumar, J .; Tripathy, S.
K.; Li, L. Chem. Mater. 1998, 10, 1554-1560.

Yang, X.; Johnson, S.; Shi, J .; Holesinger, T.; Swanson, B. Sens. Actuators, B
1997, B45, 87-92.

31

(72)
(73)

(74)

(75)

(76)

(77)
(78)
(79)
(80)

(81)

(82)
(83)

(84)

(85)
(86)
(87)
(88)

(89)

(90)

(91)

Lowy, D. A.; F inklea, H. O. Electrochim. Acta 1997, 42, 1325-1335.
Fou, A. C.; Rubner, M. F. Macromolecules 1995, 28, 7115-7120.

Lukkari, J .; Vinikanoja, A.; Salomaki, M.; Aaritalo, T.; Kankare, J. Synth. Met.
2001, 121, 1403-1404.

Zheng, H.; Rubner, M. F .; Hammond, P. T. Langmuir 2002, 18, 4505-4510.

Dai, J.; Sullivan, D. M.; Bruening, M. L. Ind. Eng. Chem. Res. 2000, 39, 3528—
3535.

F arhat, T. R.; Schlenoff, J. B. Electrochem. Solid-State Lett. 2002, 5, 813-815.
Krasemann, L.; Tieke, B. Mater. Sci. Eng. C 1999, C8-C9, 513-518.
Krasemann, L.; Tieke, B. Langmuir 2000, 16, 287-290.

Stair, J. L.; Harris, J. J.; Bruening, M. L. Chem. Mater. 2001, 13, 2641-2648.

Stroeve, P.; van Os, M.; Kunz, R.; Rabolt, J. F. Thin Solid Films 1996, 284—285,
200-203.

Krasemann, L.; Tieke, B. Chem. Eng. Tech. 2000, 23, 211-213.
Krasemann, L.; Tieke, B. J. Membr. Sci. 1998, 150, 23-30.

Meier-Haack, J .; Lenk, W.; Lehmann, D.; Lunkwitz, K. J. Membr. Sci. 2001, 184,
233-243.

Levaesahni, J .-M.; McCarthy, T. J. Macromolecules 1997, 30, 1752-1757.
Dai, J .; Bruening, M. L. Nano Letters 2002, 2, 497-501.

Farhat, T. R.; Schlenoff, J. B. Langmuir 2001, 1 7, 1184-1192.

Von Klitzing, R.; MOhwald, H. Thin Solid Films 1996, 284-285, 352-356.

Qiu, X.; Donath, E.; MOhwald, H. Macromolecular Materials and Engineering
2001, 286, 591-597.

Antipov, A. A.; Sukhorukov, G. B.; Donath, E.; MOhwald, H. J. Phys. Chem. B
2001, 105, 2281-2284.

Schlenoff, J. B.; Ly, H.; Li, M. J. Am. Chem. Soc. 1998, 120, 7626-7634.

32

(92)
(93)
(94)
(95)
(96)
(97)
(98)

(99)

(100)

(101)

(102)

(103)

(104)

(105)

(106)

(107)

(108)

(109)

(110)

Riihe, J .; Knoll, W. Supramolecular Polymers 2000, 565-613.

Zhao, B.; Brittain, W. J. Prog. Polym, Sci. 2000, 25, 677-710.

Uyama, Y.; Kato, K.; Ikada, Y. Adv. Polym. Sci. 1998, 13 7, 1-39.

Milner, S. T. Science 1991, 251, 905-914.

Ito, Y.; Nishi, S.; Park, Y. S.; Iminishi, Y. Macromolecules 1997, 30, 8096.
Mayes, A. M.; Kumar, S. K. MRS Bull. 1997, 22, 43-47.

Halperin, A.; Tirrell, M.; Lodge, T. P. Adv. Poly. Sci, 1991, 100, 31-71.

Nagale, M.; Kim, B. Y.; Bruening, M. L. J. Am. Chem. Soc. 2000, 122, 11670-
11678.

Belder, G. F .; ten Brinke, G.; Hadziioannou, G. Langmuir 1997, 13, 4102-4105.

Jordan, R.; Ulman, A.; Kang, J. F.; Rafailovich, M. H.; Sokolov, J. J. Am. Chem.
Soc. 1999, 121, 1016-1022. _

Mansky, P.; Liu, Y.; Huang, E.; Russell, T. P.; Hawker, C. Science 1997, 275,
1458-1460.

Matyjaszewski, K. In ACS Symposium Series; Matyjaszewski, K., Ed.; American
Chemical Society: Washington DC, 1998; Vol. 685, pp 2-30.

Szwarc, M. Nature 1956, I 78, 1168-1169.

Van Beylen, M.; Bywater, S.; Smets, G.; Szwarc, M.; Worsfold, D. J. Adv. Polym.
Sci. 1988, 86, 87-143.

Wang, J.-S.; Matyjaszewski, K. Macromolecules 1995, 28, 7901-7910.

Kato, M.; Karnigaito, M.; Sawamoto, M.; Higashimura, T. Macromolecules 1995,
28, 1721-1723.

Husseman, M.; Malmstrtim, E. E.; McNamara, M.; Mate, M.; Mecerreyes, D.;
Benoit, D. G.; Hedrick, J. L.; Mansky, P.; Huang, E.; Russell, T. P.; Hawker, C. J.
Macromolecules 1999, 32, 1424-1431.

Baum, M.; Brittain, W. J. Macromolecules 2002, 35, 610-615.

Tsujii, Y.; Ejaz, M.; Sato, K.; Goto, A.; Fukuda, T. Macromolecules 2001, 34,
8872-8878.

33

(111)

(112)

(113)

(114)

(115)

(116)

(117)

(118)

(119)

(120)

(121)

(122)

(123)
(124)
(125)
(126)
(127)

(128)

Matyjaszewski, K.; Xia, J. Chem. Rev. 2001, 101, 2921-2990.

Wei, M.; Xia, J .; Matyjaszewski, K. Polym. Prepr. (Am. Chem. Soc., Div. Polym.
Chem.) 1997, 38, 233-234.

Wei, M.; Xia, J .; McDermott, N. E.; Matyjaszewski, K. Polym. Prepr. (Am.
Chem. Soc., Div. Polym. Chem.) 1997, 38, 231-232.

Granel, C.; Dubois, P.; Jerome, R.; Teyssie, P. Macromolecules 1996, 29, 8576-
8582.

Uegaki, H.; Kotani, Y.; Kamigaito, M.; Sawamoto, M. Macromolecules 1997, 30,
2249-2253.

Matyjaszewski, K.; Xia, J. Chem. Rev., 101, 2921-2990.

Kim, J .-B.; Bruening, M. L.; Baker, G. L. J. Am. Chem. Soc. 2000, 122, 7616-
7617.

Huang, W.; Kim, J .-B.; Bruening, M. L.; Baker, G. L. Macromolecules 2002, 35,
1 175-1 179. .

Huang, W.; Baker, G. L.; Bruening, M. L. Angew. Chem. Int. Ed. 2001, 40, 1510-
1512.

Matyjaszewski, K.; Miller, P. J .; Shukla, N.; Irnmarapom, B.; Gelman, A.;
Luokala, B. B.; Siclovan, T. M.; Kickelbick, G.; Vallant, T.; Hoffrnann, H.;
Pakula, T. Macromolecules 1999, 32, 8716-8724.

Jones, D. M.; Huck, W. T. S. Adv. Mater. 2001, 13, 1256-1259.

Shah, R. R.; Merreceyes, D.; Husemann, M.; Rees, 1.; Abbott, N. L.; Hawker, C.
J .; Hedrick, J. L. Macromolecules 2000, 33, 597-605.

Huang, X.; Wirth, M. J. Macromolecules 1999, 32, 1694-1696.

Huang, X.; Doneski, L. J .; Wirth, M. J. Anal. Chem. 1998, 70, 4023-4029.
Huang, X.; Doneski, L. J .; Wirth, M. J. Chemtech 1998, 28, 19-25.

Jia, J.; Baker, G. L. J. Polym. Sci, Part B: Polym. Phys. 1998, 36, 959-968.
Eisenbach, C. D. Polymer 1980, 21, 1175-1179.

Graham, T. Philos. Mag. 1866, 32, 401-420.

34

(129)
(130)
(131)

(132)

(133)

(134)

(135)

(136)

(137)
(138)
(139)

(140)

Loeb, S.; Sourirajan, S. Adv. Chem. Ser. 1963, 38, 117-132.
Henis, J. M. S.; Tripodi, M. K. J. Membr. Sci. 1981, 8, 233-246.
Fritzsche, A. K. Polymer News 1988, 13, 266-274.

Meyer, H. S.; Gamez, J. P. In Proceedings of the Laurance Reid Gas
Conditioning Conference, 1995; Vol. 45th, pp 284-307.

Wright, C. T.; Paul, D. R. Polymer 1997, 38, 1871-1878.

Raymond, P. C.; Paul, D. R. J. Polym. Sci, Part B: Polym. Phys. 1990, 28, 2079-
2102.

Yoshikawa, M.; Fujimoto, K.; Kinugawa, H.; Kitao, T.; Ogata, N. Chem. Lett.
1994, 243-246.

Arnold, M. E.; Nagai, K.; Freeman, B. D.; Spontak, R. J .; Betts, D. E.; DeSimone,
J. M.; Pinnau, I. Macromolecules 2001, 34, 5611-5619.

Koros, W. J .; Fleming, G. K. J. Membr. Sci. 1993, 83, 1-80.
Uhlhorn, R. J. R.; Keizer, K.; Burggraaf, A. J. J. Membr. Sci. 1989, 46, 225-241.
Koresh, J. E.; Soffer, A. Sep. Sci. T echnol. 1987, 22, 973-982.

Way, J. D.; Roberts, D. L. Sep. Sci. T echnol. 1992, 27, 29-41.

35

CHAPTER 2

Enhancing the Anion-Transport Selectivity of Multilayer
Polyelectrolyte Membranes by Templating with Cu2+

2.1 Introduction

Alternating adsorption of polyanions and polycations is an attractive method for
forming ultrathin separation membranes because of its versatility and simplicity."2
Previous studies of multilayer polyelectrolyte membranes (MPMS) deposited on porous
supports showed selective separation of monovalent and divalent ions,3'5 modest gas
separations3'f”9 and highly selective pervaporation.3’8’10’ll This study focuses on
enhancing the anion-transport selectivities of MPMS by increasing their ﬁxed negative
charge density through templating with Cu”. Introduction of ion-exchange sites occurs
due to partial Cu2+ complexation by the carboxylate groups of poly(acrylic acid) (PAA)
during the deposition of FAA/poly(allylamine hydrochloride) (PAH) membranes.
Removal of Cu2+ in acidic solution and subsequent-deprotonation of -CO0H groups
yields ﬁxed -C00’ ion-exchange sites as shown in Figure 2.1. Studies of ion transport
through Cu2+-templated PAA (PAA-Cu)/PAH membranes show a 4-fold increase in CI'
/S042' selectivities compared to pure PAA/PAH membranes deposited under Similar
conditions. Cross-linking];l3 of templated ﬁlms through heat-induced arnidation can
yield Cl'/SO42' selectivities as high as 610, and these selectivities can be achieved without
a diminution in ﬂux relative to PAA/PAH membranes that are not templated or cross-
linked.

Such separations of ions of different valence are important in applications such as

14-16 17,18

removal of harmful ions from water, water softening, production of edible

36

 

 

O = Cu2*(COO')2

Figure 2.1: Preparation of Cu2 - templated2 polyelectrolyte ﬁlms on porous alumina
supports. Step 1-adsorption of partially Cu2 -complexed PAA ozn porous alumina.
Step 2- adsorption of a polycation (PAH). Step 3-removal of Cu2 .Step 4-
deprotonation of the free carboxylic acid groups of PAA. Repetition of steps 1 and 2
produces multilayer ﬁlms. lntertwining of layers is not Shown for ﬁgure clarity.

37

salt from sea water,19 and prevention of fouling by cooling water.20 Most of these
applications require high permselectivity among different ions as well as high ﬂux.
Previous research on ion-exchange membranes Showed that permselectivity can be

”’24 or surface

controlled by alteration of hydrophilicity/hydrophobicity,2I’22 cross-linking
charge density.25 To achieve maximum efﬁciency in ion-separation processes, a minimal
membrane thickness is also vital for achieving high ﬂux. Multilayer polyelectrolyte ﬁlms
(MPFS) are attractive for ion separation membranes in part because of their minimal
thickness.

Several studies on ion-exchange membranes already showed an increase in
monovalent/ divalent ion selectivities after adsorption of one layer of polyelectrolyte to
the surface of the membrane.”27 However, even better perrnselectivities might be
obtained when membranes are exclusively composed of multilayer polyelectrolyte ﬁlms
(MPFS). MPFS are attractive as separation membranes because both their thickness and
surface charge density can, in principle, be controlled by varying deposition conditions

30-34

such as pH,”29 salt concentration, and the number of adsorbed layers.35 Recent

”’13 that

studies Show that MPMS exhibit monovalent/divalent ion-transport selectivities
are at least in part due to Donnan exclusion at the charged surface layer of the
polyelectrolyte ﬁlms. Control over the charge density either at the surface or in the bulk
of MPMS should thus yield control over membrane selectivity. Previous
characterization‘1’6’37 of MPFS showed, however, that the bulk of the ﬁlm is intrinsically

charge compensated (i.e., polycations exactly neutralize the charge on polyanions), and

total exchangeable charge resides only at the surface of the ﬁlm. Enhancement of ion-

38

transport selectivity by introduction of charge into the bulk of MPFS will likely require
post-deposition ﬁlm modiﬁcation.

In this study, we use Cu2+-complexed PAA38'40 to control the charge density
within PAA/PAH ﬁlms. Several groups recently integrated metallosupramolecular
polyelectrolytes into polyelectrolyte assembliesims For example, Kurth and coworkers
used iron-coordinated terpyridine as a polycation to form MPFS with poly(styrene
sulfonate).45 However, in the present case, we only partially complex PAA with Cu”, so
that PAA is still deposited as a polyanion. This allows the introduction of cation-
exchange sites after removal of Cu”, and formation of highly selective membranes.
Difﬁrsion-dialysis studies and simple modeling of transport through these membranes

suggest that selectivity is due to both Donnan exclusion and difﬁrsional selectivity.

2.2 Experimental

2.2.1 Chemicals and Solutions

Poly(allylamine hydrochloride) (PAH) (MW = 70,000), poly(acrylic acid) (PAA)
(MW = 2,000) and 3-mercaptopropionic acid (MPA) were used as received from Aldrich.
We used a relatively low molecular weight PAA to increase its solubility when
complexed with Cu”. NaCl, CuCl2'5H20, and Na2SO4 were used as received from
Spectrum. AnodiscTM porous alumina membranes with 0.02 rim-diameter surface pores
(Whatrnan Anodisc) were used as supports for deposition of polyelectrolyte ﬁlms.“47
For cross-linked PAA/PAH membranes, the outer polypropylene support ring of the

alumina membrane was burned off prior to ﬁlm deposition by heating at 400 °C for 18-20

hours. This was done in order to prevent melting of the polymer ring into the pores of the

39

membrane during heat-induced cross-linking. Gold slides (200 nm of sputtered Au on 20
nm Cr on Si (100) wafers) were used as substrates for ellipsometry, external reﬂection

FTIR spectroscopy and cyclic voltammetry.

2.2.2 Film Preparation

Prior to ﬁlm deposition, porous alumina substrates were cleaned in a UV/ozone
cleaner (Boekel UV-Clean model 135500) for 15 min. Deposition ofpolyelectrolyte
ﬁlms began by dipping the substrate into a solution of PAA (0.04 M with respect to the
repeating unit) containing CuCl2 (2.5mM to 7.5 mM) for 5 min followed by rinsing for 1
min with water (Milli-Q, 18 MQ-cm). The substrate was then immersed in a solution of
PAH (0.04 M with respect to the repeating unit and containing the same CuCl2
concentration as PAA) for 5 min and rinsed with» water for l min. The above procedure
was repeated until the desired number of bilayers was deposited. The pH values of the
deposition solutions were adjusted to 5.5, 6, or 6.6 using dilute HCl and NaOH solutions.
Both PAH and PAA solutions contained 0.5 M NaCl as a supporting electrolyte to
increase ﬁlm thickness, and PAH and PAA depositions were always done at the same pH.
The porous alumina membrane is asymmetric such that the permeate side contains 0.2
rim-diameter pores, while pores on the ﬁltrate side are 0.02 um in diameter. Deposition
of polyelectrolytes was limited to the ﬁltrate side by using an o-ring holder. After
deposition of the desired number of layers, membranes were rinsed well with water and
dried with N2. For cross-linking, membranes were placed in a ﬂask that was
subsequently purged with N2 for 30 min and then slowly heated to the desired
temperature (~45 min ramping time). Heating continued at the desired temperature (100-

160 °C) for an additional 2 hours under N2 purging.

4O

Similar to porous alumina supports, gold slides were UV/ozone cleaned prior to
deposition. However, before polyelectrolyte adsorption, slides were immersed in an
ethanolic solution of 2 mM MPA for 30 min and rinsed well with ethanol followed by
deionized water (Milli-Q, 18 MQ-cm). This produces a carboxylic acid-containing
monolayer on the surface that will be charged upon deprotonation. The polyelectrolyte
deposition on gold was the same as for porous alumina except that depositions started
with PAH rather than PAA. We used MPA rather than a long-chain alkanoic acid to

avoid possible blocking of electron transfer at the gold surface.“49

2.2.3 Film Characterization

Film thickness was determined with a rotating analyzer ellipsometer (J .A. Wollam
model M-44), assuming a ﬁlm refractive index of 1.5. The refractive index and
absorption coefﬁcient of the substrates were determined after deposition of the MPA
layer. For each sample, thicknesses at three different spots were taken. External
reﬂectance FTIR spectra were obtained with a Nicolet Magma-560 FTIR Spectrometer
using a Pike grazing angle attachment (80° angle of incidence). The Spectrometer
employs a MCT detector. Electrochemical measurements were performed with a CH-
Instrument Electrochemical Analyzer (model 605) employing a standard three-electrode
cell containing a Ag/AgCl (3M KCl) reference electrode and a Pt wire counter electrode.
The working electrode was a gold slide in a plastic holder that exposed an area of 0.1
cm2. The supporting electrolyte in all electrochemical experiments was 0.1M Na2SO4.
Before measuring cyclic voltarnmograms (scan rate of 0.1 V/s), solutions were purged

with N2 for 20-30 min.

41

2.2.4 Ion-Transport Studies.

Diffusion dialysis was performed using two glass half cells as shown in Figure
2.2. The membrane was clamped between the two half cells, with the ﬁlm side of the
membrane facing towards the feed cell, so as to expose 2 cm2 of membrane to the feed
solution. The permeate cell contained deionized water (90 mL), and the feed cell
contained 0.1 M solutions (90 mL) of NaCl or Na2SO4. After dialysis with a particular
salt, the apparatus was washed well with water, and both cells were equilibrated with
deionized water for 30 min before examining the next salt. Alternating NaCl (pH 5.3)
and Na2S04 ﬁrH 5.6) transport experiments were performed until two successive chloride
ﬂuxes matched within 15%. Once the membrane achieved a steady Cl' ﬂux value (this
usually occurs after 4 to 5 NaCl and Na2S04 runs), it was removed from the permeability
apparatus and dipped in pH 3.5 water (dilute HCl solution) for one hour to ensure that all
the copper was removed from the membrane. Then the membrane was immersed in
deionized water (adjusted to pH 5-6 with a dilute NaOH solution) for 1-2 hours to
deprotonate the -COOH groups that were created upon removal of Cu“. Permeability
experiments were then repeated, and Cl' ﬂuxes differed by less than 5% when performed
before and after a Na2SO4 permeability experiment. The C1‘/S042' selectivities and ﬂux
values reported here are calculated exclusively from permeability studies performed
directly after immersing in pH 3.5 water followed by pH 5-6 water. (In fact, the initial

conditioning runs are probably not necessary).

42

Conductivity
meter

   
 

Mow.m:rpr-lto.dwb

Mechanical‘
Stirrer . ﬂ

\1
‘1‘. \

 

 

 

J

 

l l

 

Receiving Phase
Water (90 mL)

3:117

 

2/

 

/

 

—L

Source Phase
0.1 M Salt Solution

Membrane (90 mL)

Figure 2.2: Apparatus for ion-transport measurements.

43

Magnet

The permeate-cell conductivity values were converted to concentration using a
calibration plot of conductivity versus concentration for a particular salt. The ﬂux (J) for
each permeating ion was calculated using equation 2.1, and the selectivity (a) of one ion

over the other was obtained from equation 2.2.

AQK ,,
At A '
*i 22
or J2 .

In these equations, AC/At is the concentration change in the receiving cell with time
obtained from the Slope of a plot of concentration versus time; V is the volume of the
solution in the receiving cell after 90 min, A is the exposed surface area of the

membrane; and subscripts 1 and 2 refers to the two different permeating ions.

2.3 Results and Discussion

2.3.1 Synthesis and Characterization of Cu2+-Templated PAA/PAH Films.

Figure 2.1 shows schematically the preparation of Cu2+-templated PAA/PAH
ﬁlms on porous alumina supports. The procedure begins by preparing PAA complexed
with Cu2+. To do this, we employ a PAA repeating unit to Cu2+ ratio of 8:1 so that ~25%
of the -C00' groups of PAA will be complexed with Cu2+ (two -C00' groups should
bind with one Cu2+ ion). The pH of the solution must be around 5.5 so that —C00'
groups are mostly deprotonated and Cu(0H)2 does not precipitate. Alternating
adsorption of the CUB-complexed PAA (uncomplexed -C00' groups allow PAA to act as

a polyanion) and PAH molycation) results in a MPF. Although UV/visible Spectroscopy

suggests that PAH does not form a complex with Cu2+ at pH 5.5,50 we use the same Cu2+
concentration in PAH deposition as for PAA to prevent leaching of Cu2+ ﬁom the
deposited PAA-Cu layer during immersion in the PAH solution. After deposition of the
desired number of layers, we expose ﬁlms to an HCl solution (pH 3.5) to exchange
protons for Cu2+ and create free -CO0H groups on PAA chains. Subsequent immersion
in a pH 5.5 solution deprotonates these -CO0H groups (exchange of protons for Na+) and
increases the ﬁxed negative charge density in the bulk of the ﬁhn. The Cu2+-templated
PAA/PAH ﬁlms differ from pure PAA/PAH ﬁlms in that they contain -C00' groups that
are electrically compensated by mobile cations (N a+) rather than neighboring ammonium
groups of PAH. For cross-linked ﬁlms, we use the same deposition procedure (Figure
2.1), except the removal of Cu2+ and the deprotonation of CO0H groups (Steps 3 and 4 in
Figure 2.1) occur after heating the ﬁlms for two hours. Heating results in the formation
of amide cross-links from -C00'-NH3,+ pairs.12'5"52

Cyclic voltammetry (Figure 2.3) of PAH/PAA-Cu ﬁlms deposited on gold wafers
conﬁrms the presence of Cu” in these ﬁlms as well as its removal at low pH. The peaks
due to Curr/Cu completely disappear after immersion of the electrode in water at pH 3.5
(pH adjusted with 0.1 M HCl). Integration of the reduction or oxidation peak allows
estimation of the amount of Cu2+ in the ﬁlm, and this proves useﬁrl in modeling of ion
transport (vide infra). As a comparison, we also tried to put Cu2+ into a 10-bilayer
PAH/PAA ﬁlm (deposited under similar conditions) by immersing the ﬁlm in a 0.1 M
CuCl2 solution for 20 hours.53 Cyclic voltammetry of this ﬁlm showed that the amount of

adsorbed /absorbed Cu2+ is about 1/6 of that in a Cu2+-temp1ated ﬁlm.

45

 

Current

 

 

l I

 

 

0.6 0.4 0.2 0.0 -0.2
Potential (V vs. Ag/AgCl)

Figure 2.3: Cyclic voltammetry of a MPA-modiﬁed gold electrode
coated with 10 bilayers of PAH/PAA-Cu before (solid line) and
after exposure to pH 3.5 water (dotted line).

46

Reﬂectance FTIR spectra also conﬁrm templating of PAH/PAA-Cu ﬁlms with
Cu“. The spectrum of a PAH/PAA-Cu ﬁlm (spectrum a, Figure 2.4) shows a broadening
of the -C00' symmetric stretch compared with the Spectrum of a pure PAH/PAA ﬁlm
(spectrum (1, Figure 2.4). This broadening results from counter-ion-induced changes in
the energy of the -C00' stretch.54 Upon exposure to pH 3.5 water and removal of Cu2+
(Spectrum b, Figure 2.4), the -C00' symmetric stretch looks like that of a pure PAH/PAA
ﬁlm. Further, a 50% increase in the acid carbonyl peak (1715 cm'l) after immersion in
pH 3.5 water suggests that lowering of pH creates free -CO0H groups from the Cu“
complexes, as would be expected. Immersing the ﬁlm in pH 5-6 water deprotonates
-CO0H groups and results in a decrease in the acid carbonyl peak (spectrum 0, Figure

2.4).

2.3.2 Anion Transport through Cu21-Templated PAA/PAH Membranes.
Ion-transport studies Show that PAA-Cu/PAH membranes on porous alumina
supports are signiﬁcantly more selective than similar pure PAA/PAH membranes. Figure
2.5 shows a plot of receiving-phase concentration as a function of time for membranes
sandwiched between deionized water (receiving phase) and 0.1 M NaCl or Na2SO4
(source phase). These plots Show that the Cl' ﬂux through both Cu2+-templated and pure
PAA/PAH membranes is about 40 % of that through bare porous alumina. However,
10.5-bilayer PAA-Cu/PAH membranes (the top layer in the ﬁlm is PAA-Cu) show a 4-
fold decrease in S042‘ ﬂux relative to pure 10.5-bilayer PAA/PAH membranes as Shown
in the inset of Figure 2.5. Overall, Cl'/S042' selectivity increases 4-fold due to templating

of 10.5-bilayer ﬁlms (Table 2.1).

47

 

 

Absorbance

(b)

COO-Cu2+

 

 

I l I l

 

1800 1700 1600 1500 1400 1300
Wavenumbers (cm4)

Figure 2.4: External reﬂectance FTIR spectra of (a) a 10-bilayer
PAI-I/PAA-Cu ﬁlm, (b) the same ﬁlm after exposure to pH 3.5 water,
(c) the ﬁlm after subsequent exposure to pH 5.5 water, and (d) a 10-
bilayer PAH/PAA ﬁlm deposited without Cu2+. All ﬁlms were
deposited on a gold wafer coated with a monolayer of MPA.

48

3.5

 

 

 

 

 

 

 

 

 

0.08
" 0.06
E 3.0 -
v 0.04
8
2:: - 0.02
S 2.5
‘g' 0.00
g 2.0 -
O
o
8 1.5 -
m
.C
o.
P’ 1 0 r ’
c .
0 Q’
(I) 0.5 ‘ “I
m Q/

A/
/O/ .. f. m if,
0.0/L1 123 1! El 1931 1.1 H

 

O 51015202530354045

Time (min)

Figure 2.5: Receiving phase concentration as a function of time for a bare
porous alumina membrane (black), an alumina membrane coated with 10.5
bilayers of Cu21-templated PAA/PAH (open) and an alumina membrane coated
with 10.5 bilayers of PAA/PAH (grey). Different symbols represent different
salts: circles-NaCl and squares-Na2S04. The inset Shows 8042’ concentration
vs. time for pure PAA/PAH (grey) and templated PAA/PAH (open). The
membrane separates the receiving phase (initially deionized water) from a 0.1
M salt solution.

49

2.3.3 Cross-linked Cu2+-Templated PAA/PAH Membranes.

One possible limitation to Cl’/SO42' selectivity is that swelling in water may
decrease charge density and reduce Donnan exclusion of S0421 In an effort to limit ﬁlm
swelling, we cross-linked PAA-Cu/PAH ﬁlms by heating under N2 to form amide bonds
through reaction of the ammonium groups of PAH and the carboxylate groups of PAA
that are not complexed with Cu“. Reﬂectance FTIR spectroscopy conﬁrms that the
cross-linking reaction occurs.12 After heating at 160 °C, external reﬂectance FT IR
spectra of 10-bilayer PAH/PAA-Cu ﬁlms show a large reduction in the intensity of —-
C00' peaks at 1570 cm'1 and 1400 cm’1 and the appearance of amide peaks at 1660 cm’1
and 1550 cm'1 (Figure 2.6). With lower heating temperatures, the amide peaks are more
clearly visible after exposing the cross-linked ﬁlms to low-pH solutions because peaks
due to Cu21-C00‘ complexes also appear in this region of the spectrum. The degree of
cross-linking depends greatly on heating temperature as indicated by amide peaks that
increase with cross-linking temperature.12

Diffusion dialysis studies Show that as heating temperature (and hence the degree
of cross-linking) increases, Cl'/S042' selectivity increases and then peaks at a heating
temperature of 130 °C (Table 2.1, 10.5-bilayer PAA-Cu/PAH ﬁlms). Partially cross-
linked 10.5-bilayer PAA-Cu/PAH membranes (130 °C) show a 10-fold increase in C1”
/S042’ selectivity relative to unheated, templated membranes, and this increase is

achieved with only a 20% decrease in C1' ﬂux. At higher cross-linking temperatures,

50

Table 2.1: Anion ﬂuxesa (moles cm'zs'l) through bare porous alumina and alumina
coated with PAA/PAH and PAA-Cu/PAH ﬁlms cross-linked at different temperatures.

 

 

Film Composition Cross-linking 108 x (31' 108 x S042' Cl'/S.0..;2'c
T ( C) Flux Flux Selectivrty
Bare - 4.211 3.3102 1.310.09
10 PAA/PAH - 1.0102 1.5102 0.710.03
10 PAA-Cu/PAHb - 2.3103 02710.01 911
10.5 PAA/PAH - 1.3104 01110.05 1313
10.5 PAA-Cu/PAHb - 1.6102 0.0310001 5513
10.5 PAA-Cu/PAHb 100 1.6105 00210004 80115
10.5 PAA-Cu/PAHb 120 1.4102 0.00610002 240180
10 PAA/PAH 130 0091003 002710005 310.4
10 PAA-Cu/PAHb 130 2.010.06 003210.005 62111
10.5 PAA/PAH 130 00710.01 00028100009 2617
10.5 PAA-Cu/PAH" 130 1.31005 00021100001 610120
10.5 PAA-Cu/PAH" 140 05110.2 00016100005 330170
10.5 PAA-Cu/PAH" 160 00871004 00003100001 2911.3

 

 

 

3‘ Flux values were calculated from the Slopes of plots of concentration in the receiving
phase vs. time. Errors represent standard deviations of at least three measurements.

Flux was measured after removal of Cu2+ from the membrane and deprotonation of
newly formed —COOH groups.
° Calculated as the average of selectivity values for each membrane and not from average
ﬂux values.

51

 

0.01 C00-

 

Absorbance

fmide l
Amide II

(b)

 

 

 

1 800 1 700 1600 1 500 1400

Wavenumbers (cm-‘)

Figure 2.6: External reﬂectance FTIR spectra of 10-bilayer PAH/PAA-
Cu ﬁlms (a) before and (b) after heating at 160°C. Films were deposited
on MFA-coated gold.

52

Cl" ﬂux drops rapidly, presumably due to a tighter membrane structure. Sulfate ﬂux does
not continue to drop signiﬁcantly at higher cross-linking temperatures, and thus Cl'/S042'
selectivity eventually decreases. Compared with pure PAA/PAH membranes, the
selectivities and ﬂuxes through partially cross—linked PAA-Cu/PAH membranes are
remarkable. Table 2.1 shows that partially cross-linked (130 °C) 10.5-bilayer PAA-
Cu/PAH membranes Show a 20-fold increase in Cl'/SO42’ selectivity relative to similar
cross-linked pure PAA/PAH membranes. Additionally, the Cl' ﬂux through these cross-
linked Cu2+-templated membranes is 20-fold higher than the Cl' ﬂux through pure
PAA/PAH membranes cross-linked at the same temperature. This may be due to the
formation of new transport pathways upon removal of Cu2+ or a lower degree of cross-

linking in the templated ﬁlm.

2.3.4 Changing the Surface Charge of Membranes.

Our previous studiess’l3 of MPMS showed that much of the ion-transport
selectivity in these systems is due to a high charge density at the membrane surface.
However, Cu2+-templated membranes differ from previous MPMS in that they contain
ﬁxed charge throughout the membrane. In an effort to understand more about
selectivities in PAA-Cu/PAH membranes, we changed the terminating layer of these
ﬁlms from PAA to positively charged PAH. If selectivity in these systems is largely due
to charge at their surface, changing the outer layer from a polyanion to a polycation
Should have a dramatic effect on ion transport.

Changing the surface from PAA-Cu (10.5-bilayer ﬁlms) to PAH (IO-bilayer
ﬁlms) in cross-linked (130 °C), templated ﬁlms resulted in a 15-fold increase in 8042'

ﬂux and a 50% increase in CI‘ ﬂux (Table 2.1). Thus Cl’/SO42' selectivity decreased from

53

610 to 60 on going from a 10.5-bilayer to a lO—bilayer cross-linked PAA-Cu/PAH ﬁlm.
In the case of unheated Cu2+-templated membranes, terminating with PAH rather than
PAA-Cu yielded a decrease in C1'/SO42' selectivity from 55 to 9. These data clearly
indicate that Donnan exclusion at the ﬁlm surface plays a large role in determining
selectivity. For unheated, pure PAA/PAH membranes, selectivity actually reverses (from
13 to 0.7) upon changing the top layer from PAA to PAH. However, with Cu“-
templated ﬁhns, we still see a signiﬁcant Cl’/SO42’ selectivity when the surface of the
membrane is positively charged because of ﬁxed negative charge density in the bulk of
the membrane.

2.3.5 Anion-Transport through Partially Cross-linked, Cu2+-Templated
Membranes Deposited at Different pH values.

Variation of the pH at which PAA-Cu/PAH ﬁlms are deposited allows some
control over the amount of Cu2+ in these ﬁlms and may provide a means for controlling
transport selectivity. Table 2.2 gives the Cl'/S042' selectivity values for partially cross-
linked 10.5-bilayer PAA-Cu/PAH membranes deposited from solutions at three different
pH values (5.5, 6, and 6.6). We also tried to deposit membranes at pH values <5.5, but
under these conditions, polymer precipitates from deposition solutions. Selectivity is
highest for ﬁlms deposited at pH 5.5 and decreases at higher deposition pH values. The
ellipsometric thicknesses (Table 2.2) of Similar ﬁlms on gold wafers are independent of
pH over this range of values,55 so selectivity differences are likely due to changes in

charge density. The UV/visible Spectra of the PAH/Cu2+ solutions showed a shift in the

54

Table 2.2: Thicknesses, ﬂuxes (moles cm'zs’1), selectivities and Cu2+ concentrations (M)
in partially cross-linked, 10.5-bilayer PAA-Cu/PAH ﬁlms deposited at different pH
values.

 

 

Deposition Thickness” cr Fluxc 8042' Flux° Cl‘/SO42‘ Cu2+
pHal (A) x 108 x 101 l Selectivityd Concentratione
5.5 170i3 l.3i0.05 2.1i0.1 610i20 1.032006
6 1703.6 1.1i0.2 2.6i0.4 430:90 0.9i0.04
6.6 170i10 1.63:0.3 170i42 11i4 0.3i0.1

 

 

 

aBoth PAA-Cu and PAH were deposited at this pH.

bThiclmesses are for lO-bilayer PAH/PAA-Cu ﬁlms deposited on gold wafers as
described in the experimental section.

CError values represents standard deviations.

dCalculated as the average of selectivity values for each membrane and not from average
ﬂux values.

eCu2+ concentration in the membrane was estimated from the area of the reduction peak
in a cyclic voltammagram of a ﬁlm on gold. This area was converted to number of moles
of Cu21/cm2, and this value was divided by the ellipsometric ﬁlm thickness to obtain the
concentration.

55

Cu2+ absorption peak from 820 nm at pH 5.5 to 780 nm at pH 6 and to 710 nm at pH 6.6,
suggesting that the amine groups of PAH begin to form complexes with Cu2+ at the
higher pH values. In addition, at higher pH values Cu2+ can form hydroxide complexes.
These competing reactions probably reduce the amount of Cu2+ deposited in the
membrane as -C00'-Cu2+ complexes.

To quantitatively investigate ﬁxed-charge density in PAH/PAA-Cu membranes,
we employed cyclic voltammetry to estimate Cu2+ concentrations in analogous ﬁlms
deposited on gold (Figure 2.7). By integrating the area of the reduction peak, Cu2+
concentrations could be estimated. In agreement with transport studies, Table 2.2 Shows
that the maximum amount of copper is deposited at pH 5.5. Hence, after removal of Cu2+

from the ﬁlm, higher charge densities should enhance Cl'/S042' selectivity for ﬁlms

deposited at the lower pH values.

2.3.6 Anion Transport through Partially Cross-linked, Cu21-Templated
Membranes Deposited with Different Cu 1 Concentrations.

Altering the amount of Cu2+ present during deposition should provide another
means for controlling ﬁxed charge and selectivity in membranes. To examine this
possibility, we prepared cross-linked PAH/PAA-Cu membranes using Cu2+
concentrations of 2.5 mM, 5 mM or 7.5 mM in both PAA and PAH deposition solutions.
Table 2.3 gives the Cl'/SO42' selectivities and ﬂux values for these membranes. (We
deposited PAA and PAH at pH 6 because at pH 5.5, higher Cu2+ concentrations resulted

in precipitation).

56

 

 

Current

 

 

L

 

0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6
Potential (V vs. Ag/AgCl)

 

Figure 2.7: Cyclic voltammetry of lO-bilayer PAH/PAA-Cu ﬁlms
deposited at different deposition pH values on MPA-modiﬁed gold
surfaces: pH 5.5-dashed line, pH 6-solid line, and pH 6.6-dotted line. Areas
of the reduction peaks were calculated by drawing the baseline from the
value of current at a potential of 0.3 V to the value of current at -0.5 V.

57

$503738”. 8% 35308
832235 65680326 use weﬂezameo 80.39 90:88:? 0:93 momma eoEEHBoe £3 countenance $39.
.5308 3368536 05 E eontomoe mm 158:
N com 06 02 on weiﬁTmmBo coca Boom? Bow :0 ago :O-<<m\m<m 63:93 com 805 0.8 uptempo 8305256
.82? 56 owﬁozw Soc So: one 285808 some com 82.? 33528 mo owﬁog 65 we 33330.2

.0 no In a we: 28 $30 oo eouebcooeoo 68.8 2: 359:8 muons—om eoEmoooe Em can <<m ﬁome

 

 

 

vodolrod 3.31m 2.11.26 3H: m.0Hm.N m6
oodﬁ: 0.132 omHOmv v.0HoN Ndﬂg ow
Sded oﬂeom mﬂw mum" 20.1%; ON
o omﬂe: NH: NHm 5.36de o
eosmbcooeoo A v DSSoBom 2x Ex 92::
e z. o E: m .4. e 3.. . an: e on” ecoEmomoe E pom:
+N 0 e H .L m 6305—25. -N OmCU E 4. Om 5.5 :ocmbeooeou 5:0

 

 

.meoaabeooeoo $20 Bonﬁre mean eozmoooe age can 855868 Egoéﬁwm 3.92563 Go 33 woo—57380
335a oo 36 meoeebeooeoo £20 36850 98 £88533 $223628 50080 8:25 BOSE cog "QN 03:.

58

Cl'/SO42' selectivity was highest when the concentration of Cu2+ present during
deposition was 5 mM. Cyclic voltammetry (Figure 2.8) and ellipsometric data (Table
2.3) for analogous ﬁlms deposited on gold wafers indicate that the highest Cu”
concentration occurs in ﬁlms deposited with 5 mM Cu”. This happens because ﬁlm
thickness increases greatly when the Cu2+ concentration during deposition is 7.5 mM.
Hence, the highest selectivity is observed for the MPMS deposited in the presence of 5
mM Cu2+. The structure of ﬁlms probably depends on the Cu2+ concentration present
during deposition,”57 and this might explain why ﬁlms deposited with 7.5 mM Cu2+ are
less selective than ﬁlms deposited with 2.5 mM Cu“, even though the former ﬁlms have

a higher concentration of Cu“.

2.3.7 Modeling of Anion Transport through Cu2+-Templated PAA/PAH

Membranes

To better understand the observed selectivities and ﬂuxes through templated
membranes, we began developing a simple model for ion transport based on previous
models of ion-exchange membranessg’59 and MPMS.60 We employed a two-layer model
of the MPF, assuming the bulk of the membrane as one layer and the surface as another
(Figure 2.9). The charge density in these two regions should be different, as several

36’6' show that net charge in MPFS resides primarily at the ﬁlm surface.

recent studies
However, in templated ﬁlms, removal of Cu2+ yields cation-exchange sites in the bulk of

the membrane as well as on the surface. We estimated the ﬁxed negative charge density

in the bulk of templated ﬁlms from the amount of deposited Cu2+ determined from cyclic

59

 

Current

 

 

 

l L I l

 

0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6
Potential (V vs. Ag/AgCI)

Figure 2.8: Cyclic voltammetry of lO-bilayer PAH/PAA-Cu ﬁlms deposited
with different Cu2+ concentrations: 2.5 mM Cu2+ -dotted line, 5.0 mM Cu”-

solid line and 7.5 mM Cu2+-dashed line. The area of the reduction peak was
calculated after drawing the baseline from the current at a potential of 0.3 V
to that at -O.4 V. Films were deposited at pH 6 on MFA-modiﬁed gold.

60

Concentration

Donnan Equilibrium

Porous Bulk MPF v C "‘2
._ . me

X

CE,F

 

Permeate Feed

 

 

Surface
Nernst-Planck Transport Layer

Figure 2.9: Schematic representation of the model used to simulate ion
transport through templated MPMS. The membrane consists of two
charged layers: a surface layer and the membrane bulk. The line
represents a hypothetical concentration proﬁle for the excluded ion.
me' and me2 are the ﬁxed charge concentrations of bulk of the ﬁlm
and surface layer respectively.

61

voltammetry (noting that every Cu2+ binds with two COO' groups) and from the
ellipsometric thicknesses of corresponding ﬁlms deposited on gold. To estimate surface
charge density, we assumed that the surface layer was composed of pure PAA.62

With a structural model in hand, one can examine transport through templated
polelectrolytes by modifying models of transport through ion-exchange membranes. Ion
transport through a membrane can result from convection, diffusion, or migration. In
diffusion dialysis, the convective contribution can be neglected as a ﬁrst approximation
(osmotic flow is not large), and ﬂux through the membrane can be described as a
combination of diffusion (concentration gradient as driving force) and migration

(electrical potential as driving force) using equation 2.3, the Nemst-Planck equation.63

dC,‘ — ZiFCiDi g2

Ji="Dt
dx RT dx

2.3

 

In this equation, J, is the flux 0 fion i, C,- is the ion concentration, Di is the diffusion
coefﬁcient, z,- is the ion charge, F is the Faraday constant, R is the gas constant, T is the
absolute temperature and (1) is the electrical potential. In this study, the presence of ﬁxed
charge in the membrane results in a very low concentration of excluded ion in the
membrane, and the migration term for this ion is negligible. Thus, one can simply use

Fick’s law for steady-state diffusion to calculate ﬂux for the excluded ion (equation 2.4).

2.4

62

In equation 2.4, d is the thickness of the polyelectrolyte region (surface or bulk),
and C Z'R and C1?" are the excluded ion concentrations at the right and left side of the
region.

To calculate mobile ion concentrations in the membrane, we assumed Donnan
equilibria at the interfaces and used the equation 2.4 to calculate concentration proﬁles
within layers. Donnan equilibrium occurs in membranes containing ﬁxed charge because
mobile anions and cations in the membrane are not present in a stoichiometric ratio. This
results in a potential drop at the solution-membrane interface and exclusion of ions whose
charge is of the same sign as the ﬁxed charge. The potential also enhances the

concentration of counter ions in the membrane.

Equating membrane and solution electrochemical potentials for each ion,
neglecting activity coefﬁcients and assuming that the standard state chemical potential is
the same in the membrane and in the solution results in equation 2.5, which describes the

distribution of excluded ion B between the membrane and the solution.64

(lzBl/lel)

215:: IZBlCi;JFI‘ZXIC)?I

CB lzBICB

2.5

 

In this equation, C $2 , Cx, 23 and 2x are the concentrations (in the membrane) and charges

of the excluded ion and the ﬁxed charge respectively. The charge of the co-ion is 2A, and

CB is the concentration of the excluded ion in solution.

63

The procedure for calculating a concentration proﬁle through the membrane
involves several steps. First, we calculate the excluded ion concentration (C 5) in the

porous alumina at the alumina/polyelectrolyte ﬁlm interface using equation 2.6 and the
experimental ﬂux values through bare porous alumina (J1me) and porous alumina coated
with a polyelectrolyte membrane (Jcomed). This equation assumes a negligible receiving-
phase ion concentration so ﬂux is simply proportional to the concentration in the alumina
at the alumina-polyelectrolyte or alumina-feed interface. The linear relationship between
receiving-phase concentration and time suggests that receiving-phase concentration is

indeed negligible. In the case of a bare membrane, the concentration at the alumina/feed

interface is simply the source-phase concentration, C 5 .

J
R d F
C B =—CO—“’e—x C B 2.6
bare
The next step in determining the concentration proﬁle is calculation of the concentration
of excluded ion (C Z'L’ ) in the polyelectrolyte ﬁlm at the alumina/bulk ﬁlm interface

using equation 2.7.
/
c?“ z lzBIcg‘Ll +|2x|C§’2‘ (’ZB'/'z"')

2.7
Cg lZBlCE

 

 

Here, C 3}" is the concentration of the ﬁxed charge in the bulk of MPF and C 5 is the

excluded ion concentration in the porous alumina at the alumina/polyelectrolyte ﬁlm

interface. Subsequently, we solve equation 2.4 to calculate the concentration of excluded

64

ion (C Z’R' ) in the bulk ﬁlm at the bulk ﬁlm/surface layer interface. Donnan equilibrium
at the bulk ﬁlm/surface layer interface is slightly different than at the alumina/bulk ﬁlm
interface because both ﬁlms contain ﬁxed charge, albeit in different amounts. The
Donnan equilibrium assumptions for this interface give equation 2.8, which allows for
calculation of the concentration of excluded ion (C 1;“ ) in the surface layer at the

surface layer/bulk ﬁlm interface.

 

L m mL lzBl/l/lel
Ci? 2_ IZXICX2+|ZB|CB 2

le _ m mR 2'8
CB lZXle‘+|ZB|CB ‘

In this equation, C It" 2 and C ,2" ’ are the concentrations of ﬁxed charge in the surface layer

and in the bulk respectively and C $19 is the concentration of excluded ion in the bulk at

the bulk ﬁlrn/ surface layer interface. Finally, using equation 2.4, we calculate the
concentration of excluded ion in the surface layer at the surface layer/feed interface using

Fick’s law and subsequently solve equation 2.9 for the concentration of the excluded ion

in the source phase (C 5 ).

 

 

Cg : IZBIC; lZBI/lel 2 9
mR2 mR2 r112 '
CB IZBICB +|2X|CX

In this equation C :2 is the concentration of the ﬁxed charge in the surface layer

and CZR” is the concentration of excluded ion in the surface layer at the surface layer/ feed

65

interface. 23, 2A and 2,. are the charges of excluded ion, co-ion and the ﬁxed charge

respectively.

We performed this entire calculation iteratively using the experimental ﬂux values
and varying the diffusion coefﬁcient of the excluded ion in the polyelectrolyte ﬁlm until
the calculated source-phase excluded ion concentration matched the experimental value
(0.1 M). Table 2.4 lists the diffusion coefﬁcients determined for C1' and 8042' in this
way. (We assumed that the diffusion coefﬁcient was the same in the bulk and surface
ﬁlm layers). These calculations on unheated, templated membranes yielded Cl' diffusion
coefﬁcients on the order of 10'8 cmz/s, and the Cl' diffusion coefﬁcient was a factor of 6
higher than that of 8042’. Miyoshi reported58 that diffusion coefﬁcients through ion
exchange membranes are ~10"8 cmz/s for Na+ and Mg2+, reasonably close to the Cl' values
we calculate. The calculated diffusion coefﬁcients suggest that selectivity is about

equally due to Donnan exclusion and diffusivity differences.

Modeling studies on cross-linked, templated membranes show that as the cross-
linking temperature increases, diffusion coefﬁcients generally decrease (Table 2.4). This
is reasonable because cross-linking reduces swelling, and hence free volume. At low
cross-linking temperatures, decreases in the 8042' diffusion coefﬁcient are more dramatic
than those for Cl', and this may be due to size exclusion and/or hydrophobicity effects.
21’65’66 On going from a cross-linking temperature of 140 to 160 °C, the diffusion

coefﬁcient of 8042' actually increased, probably because ﬁxed charge decreased at this

temperature, and we couldn’t take this into account. At a cross-linking temperature of

130 °C, the calculated Cl' diffusion coefﬁcient was 25-fold higher than that of 8042'.

66

Table 2.4: Diffusion coefﬁcients obtained from modeling ion transport through 10.5-
bilayer PAA-Cu/PAH membranes cross-linked at different temperatures.

 

 

 

Bulk Estimated
Cross- Cl- SO42- Fixed Surface Electrostatic
linking Diffusion Diffusion Char e Fixed Diffusional Exclusion
Temperature Coefﬁcient Coefﬁcient .g 3. Charge Selectivityb . . c
(°C) (“11284) (cmzs") Den31ty3 Density“ Select1v1ty
(mol/cm) 3
(mol/cm L
Unheated 8.8><10‘9 1.4x10‘9 1.4x10‘3 1.1><10‘2 6 9
100 8.8><10‘9 1.1x10‘9 1.5><10‘3 1.2x10'2 8 10
120 7.4x10'9 3.6><10"° 1.6><10'3 1.3><10‘2 20.5 11.7
130 6.8><10‘9 1.5><10‘l0 2.0x10'3 1.6x10'2 45 13.6
140 2.4><10‘9 1.1x10"° 1.9><10'3 1.5><10‘2 22 15
160 4.1><10‘l0 2.2x10“° 2.1><10‘3 1.7><1o'2 1.7 17

 

a Determined using cyclic voltammetry and ellipsometry.
b The diffusional selectivity is the ratio of diffusion coefﬁcients obtained from the model.
c Total Cl'/SO42' selectivity divided by diffusional selectivity.

67

 

These simple calculations suggest that the highest selectivities in cross-linked ﬁlms are

about equally due to diffusion and Donnan selectivities. We should note, however, that

this model does not take into account activity coefﬁcients or the effect of hydrophobicity
on partitioning.

Several previous studies showed that diffusion through charged membranes can
be complicated due to electrostatic interactions between the membrane and the diffusing
ions.“68 Additionally, charge distributions in our simulations are oversimpliﬁed and
only approximate. However, the modeling studies do strongly indicate that selectivity is
only partly due to Donnan exclusion. A full understanding of transport through MPMS

will likely require measurement of diffusion and partition coefﬁcients.

2.4 Conclusions

Partial complexation of the -C00' groups of PAA with Cu2+ provides a
convenient method to enhance ﬁxed negative charge density in MPMS. Removal of Cu2+
leaves behind -COOH groups that behaves as ion-exchange sites. Diffusion-dialysis
studies with Cu2+-templated membranes show that templating increases anion-transport
selectivities, and post deposition cross-linking of these membranes further enhances Cl'
/SO42' selectivities to values high as 610. Changing the surface layer from negatively
charged PAA to positively charged PAH greatly reduces Cl'/SO42' selectivity, showing
that selectivity is highly dependent on surface charge. Simulation of ion-transport data
using a simple two-layer model of MPFS suggests that the observed Cl'/SO42' selectivities

are due to both Donnan exclusion and differences in difﬁisivities of ions.

68

2.5 References and Notes

(1)
(2)
(3)
(4)
(5)
(6)
(7)

(8)
(9)
(10)

(11)
(12)

(13)

(14)

(15)

(16)

(17)
(18)

(19)

Decher, G.; Hong, J. D. Ber. Bunsenges. Phys. Chem. 1991, 95, 1430-1434.
Decher, G. Science 1997, 277, 1232-1237.

Krasemann, L.; Tieke, B. Mater. Sci. Eng, C 1999, 8—9, 513-518.
Krasemann, L.; Tieke, B. Langmuir 2000, 16, 287-290.

Harris, J. J .; Stair, J. L.; Bruening, M. L. Chem. Mater. 2000, 12, 1941-1947.
Levasahni, J .-M.; McCarthy, T. J. Macromolecules 1997, 30, 1752-1757.

Stroeve, P.; Vasquez, V.; Coelho, M. A. N.; Rabolt, J. F. Thin Solid Films 1996,
285, 708-712.

van Ackem, F .; Krasemann, L.; Tieke, B. Thin Solid Films 1998, 329, 762-766.
Kotov, N. A.; Magonov, S.; Tropsha, E. Chem. Mater. 1998, 10, 886-895.
Krasemann, L.; Tieke, B. Chem. Eng. T echnol. 2000, 23, 211-213.

Krasemann, L.; Tieke, B. J. Membr. Sci. 1998, 150, 23-30.

Harris, J. J .; DeRose, P. M.; Bruening, M. L. J. Am. Chem. Soc. 1999, 121, 1978-
1979.

Stair, J. L.; Harris, J. J .; Bruening, M. L. Chem. Mater. 2001, 13, 2641-2648.

Hell, F.; Lahnsteiner, J .; Frischherz, H.; Baumgartner, G. Desalination 1998, 11 7,
173-180.

Amor, Z.; Bariou, B.; Mameri, N.; Taky, M.; Nicolas, S.; Elmidaoui, A.
Desalination 2001, 133, 215-223.

Sata, T.; Yamaguchi, T.; Matsusaki, K. J. Chem. Soc., Chem. Commun. 1995, 11,
1 153-1 154.

Mika, A. M.; Childs, R. F.; Dickson, J. M. Desalination 1999, 121, 149-158.

Brett, S. W.; Gaterell, M. R.; Morse, G. K.; Lester, J. N. Environ. T echnol. 1999,
20, 1009-1018.

Takata, K.; Yamamoto, Y.; Sata, T. Bull. Chem. Soc. Jpn. 1996, 69, 797-804.

69

(20)

(21)

(22)

(23)

(24)

(25)
(26)
(27)
(23)
(29)

(30)

(31)
(32)

(33)

(34)
(35)
(36)
(37)

(38)

(39)

(40)

Bader, M. S. H. J. Hazard. Mater. 2000, B73, 269-283.
Sata, T.; Mine, K.; Higa, M. J. Membr. Sci. 1998, 141, 137-144.

Sata, T.; Mine, K.; Tagami, Y.; Higa, M.; Matsusaki, K. J. Chem. Soc., Faraday
Trans. 1998, 94, 147-153.

Sata, T.; Nojima, s. J. Polym. Sci. Polym. Phys. Ed. 1999,37, 1773-1785.

Sata, T.; Emori, S. 1.; Matsusaki, K. J. Polym. Sci. Polym. Phys. Ed. 1999, 37,
793-804.

Matsusaki, K.; Hashimoto, N.; Kuroki, M.; Sata, T. Anal. Sci. 1997, 13, 345-349.
Tsuru, T.; Nakao, S.-I.; Kimura, S. J. Membr. Sci. 1995, 108, 269-278.

Urairi, M.; Tsuru, T.; Nakao, 8.; Kimura, S. J. Membr. Sci. 1992, 70, 153-162.
Yoo, D.; Shiratori, S. S.; Rubner, M. F. Macromolecules 1998, 31 , 4309-4318.
Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213-4219.

Sukhorukov, G. B.; Schmitt, J .; Decher, G. Ber. Bunsenges. Phys. Chem. 1996,
100, 948-953.

Sukhishvili, S. A.; Granick, S. J. Chem. Phys. 1998, 109, 6861-6868.
Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153-8160.

Ldsche, M.; Schmitt, J .; Decher, G.; Bouwman, W. G.; Kjaer, K. Macromolecules
1998, 31, 8893-8906.

Lvov, Y.; Decher, G.; Mohwald, H. Langmuir 1993, 9, 481-486.

Decher, G.; Hong, J. D.; Schmitt, J. Thin Solid Films 1992, 210/211, 831-835.
Schlenoff, J. B.; Ly, H.; Li, M. J. Am. Chem. Soc. 1998, 120, 7626-7634.
Laurent, D.; Schlenoff, J. B. Langmuir 1997, 13, 1552-1557.

Gotoh, Y.; Igarashi, R.; Ohkoshi, Y.; Nagura, M.; Akamatsu, K.; Deki, S. J.
Mater. Chem. 2000, 10, 2548-2552.

Rivas, B. L.; Schiappacasse, N.; Basaez, L. A. Polym. Bull. 2000, 45, 259-265.

Porasso, R. D.; Benegas, J. C.; Hoop, M. A. G. T. v. d. J. Phys. Chem. B 1999,

70

(41)
(42)

(43)

(44)

(45)

(46)
(47)
(48)

(49)

(50)

(51)

(52)

(53)

(54)

(55)

(56)

(57)

103, 2361-2365.
Kurth, D. G.; Osterhout, R. Langmuir 1999, 15, 4842-4846.
Caruso, F.; Schuler, C.; Kurth, D. G. Chem. Mater. 1999, II, 3394-3399.

Xiong, H.; Cheng, M.; Zhou, Z.; Zhang, X.; Shen, J. Adv. Mater. 1998, 10, 529-
532.

Zhang, X.; Shen, J. C. Adv. Mater. 1999, 11, 1139-1143.

Schutte, M.; Kurth, D. G.; Linford, M. R.; Colfen, H.; Mohwald, H. Angew.
Chem. Int. Ed. 1998, 3 7, 2891-2893.

Chen, W.-J.; Aranda, P.; Martin, C. R. J. Membr. Sci. 1995, 107, 199-207.
Chen, W.-J.; Martin, C. R. J. Membr. Sci. 1995, 104, 101-108.
Dai, Z.; Ju, H. Phys. Chem. Chem. Phys. 2001, 3, 3769-3773.

Bain, C. D.; Troughton, E. B.; Tao, Y. T.; Evall, J .; Whitesides, G. M.; Nuzzo, R.
G. J. Am. Chem. Soc. 1989, 111, 321-335.

At ph 5.5, the UV/vis spectrum of PAH/Cu2+ solution is identical to that of a
CuClz solution containing the same Cu2+ concentration.

Dai, J .; Sullivan, D. M.; Bruening, M. L. Ind. Eng. Chem. Res. 2000, 39, 3528-
3535.

Dai, J .; Jensen, A. W.; Mohanty, D. K; Emdt, J .; Bruening, M. L. Langmuir
2001, 1 7, 931-937.

Joly, 8.; Kane, R.; Radzilowski, L.; Wang, T.; Wu, A.; Cohen, R. E.; Thomas, E.
L.; Rubner, M. F. Langmuir 2000, 16, 1354-1359.

Deacon, G. 3.; Phillips, R. J. Coord. Chem. Rev. 1980, 33, 227-250.
Rubner and coworkers reported that thickness of similar MPF's are dependent on
deposition pH (reference 29). However, in our case we use a supporting

electrolyte (NaCl) and Cu2+ in the deposition solution.

Mendelsohn, J. D.; Barrett, C. J .; Chan, V. V.; Pal, A. J .; Mayes, A. M.; Rubner,
M. F. Langmuir 2000, 16, 5017-5023.

Fery, A.; Schoeler, B.; Cassagneau, T.; Caruso, F. Langmuir 2001, 1 7, 3779-
3783.

71

(5 8)
(59)
(60)
(61)

(62)

(63)

(64)

(65)
(66)

(67)

(68)

Miyoshi, H. J. Membr. Sci. 1998, 141, 101-110.

Manzanares, J. A.; Mafé, 8.; Pellicer, J. J. Phys. Chem 1991, 95, 5620-5624.
Lebedev, K.; Ramirez, P.; Mafe, S.; Pellicer, J. Langmuir 2000, 16, 9941-9943.
Lowack, K.; Helm, C. A. Macromolecules 1998, 31 , 823-833.

To estimate the charge density in the surface layer, we ﬁrst determined the
concentration of Cu2+ in a 10-bilayer ﬁlm using cyclic voltammetry and
ellipsometry and then multiplied this value by a factor of 8 to get the -COO‘/
-COOH concentration in the ﬁlm. Then we multiplied this value by a factor of 2
because the surface of the ﬁlm is mostly PAA rather than PAH/PAA. We
assumed that -COOH groups were completely deprotonated.

Higa, M.; Kira, A. J. Phys. Chem. B 1995, 99, 5089-5093.

Peeters, J. M. M.; Boom, J. P.; Mulder, M. H. V.; Strathmann, H. J. Membr. Sci.
1998, 145, 199-209.

Sata, T. J. Membr. Sci. 1994, 93, 117-135.
Sata, T.; Yamaguchi, T.; Matsusaki, K. J. Phys. Chem 1995, 99, 12875-12882.

Beerlage, M. A. M.; Peeters, J. M. M.; Nolten, J. A. M.; Mulder, M. H. V.;
Strathmann, H. J. Appl. Polym. Sci. 2000, 75, 1180-1193.

Robertson, B. C.; Zydney, A. L. J. Membr. Sci. 1990, 49, 287-303.

72

CHAPTER 3

Preparation of Composite Membranes by Atom Transfer Radical
Polymerization Initiated from a Porous Support

3.1 Introduction

Synthesis of practical separation membranes requires methods for creating thin,
selective skins at the surface of highly permeable supports."5 Composite membranes
prepared by depositing an expensive skin on a relatively inexpensive support are
especially attractive in this regard because only a small amount of the selective skin
material is needed.6’7 The most common methods for formation of composite membranes
include interfacial polymerizationf"9 casting,'0plasma polymerization,11 and solution
coating.12 Even with the successes of these methods, synthesis of selective membrane

skins with minimal (<50 nm) thicknesses is still difﬁcult.3'13

This chapter describes our
initial investigations into the possibility of using polymerization from a surface to create
ultrathin membrane skins with unique structures.

Many recent studies demonstrate the use of controlled polymerization techniques
to grow polymer chains from surfaces in a well-deﬁned way.”23 These procedures
generally involve attachment of polymerization initiators to a surface and subsequent
polymerization from these initiators. Of the many types of possible polymerization

l,24 cationic,21 anionic,15 ring-opening,25 and ring-opening

strategies (e.g., radica
metathesis“), atom transfer radical polymerization (ATRP) is especially attractive

because it yields polymers of low polydispersity and is compatible with a variety of

functional monomers. Since the initial discovery of ATRP,”28 we and others have

73

adapted this technique for surface-initiated polymerization.”"19’29’30 The recent discovery
of transition metal complexes that catalyze ATRP from a surface at room temperature is
particularly important because low-temperature polymerization from a substrate can
occur with minimal simultaneous polymerization in solution.'9’30'33 This helps to avoid
physisorption of unbound polymer chains and allows synthesis of cross-linked polymer
ﬁlms. Additionally, the controlled nature of ATRP affords control over skin thickness by
variation of polymerization time.

This work demonstrates the versatility of surface-initiated ATRP for forming
ultrathin membrane skins on a porous support and examines gas permeation through
these membranes. We utilized ATRP to synthesize two kinds of membrane skins: cross-
linked poly(ethylene glycol dimethacrylate) (PEGDMA)3 l and poly(2-hydroxyethyl
methacrylate) (PHEMA).19 The synthesis involves covalent attachment of an ATRP
initiator32 to a modiﬁed porous alumina support followed by room-temperature
polymerization with a suitable monomer..

Cross-linked polymer membranes such as PEGDMA are potentially attractive for
gas separations because they should be able to function in high levels of plasticizing and
condensable vapors that oﬁen degrade membrane performance.”38 Recent methods for
forming cross-linked membranes include UV-irradiation of benzophenone-containing

7.4 .
3 0 and chemlcal

polymers,34’3"r”39 heating of polyimides that contain diacetylene groups,
cross-linking of polyimides with diamino compounds.41 Koros and coworkers
demonstrated that chemical cross-linking of carboxylic acid-containing polyimides with

ethylene glycol greatly increases the C02 plasticization pressure and also increases

COz/CH4 selectivity.10 Preparation of cross-linked membranes can also occur by casting

74

a solution containing cross-linkable monomer and subsequently polymerizing the ﬁlm.
Although this method does not result in ultrathin skins, Hirayama and co—workers showed
that cross-linked polymer ﬁlms containing polyethylene oxide chains have a COz/Nz
selectivity of 65.42’43

This work demonstrates that ATRP from a surface allows controlled synthesis of
ultrathin, cross-linked and derivatizable membrane skins. Gas-permeation studies with
PEGDMA ﬁlms grown on porous alumina supports show that these membranes are free
of defects and have a COz/CH4 selectivity of ~20. In comparison, PHEMA brushes show
selectivity values typical of Knudsen diffusion. One advantage of the PHEMA
membranes, however, is that they can be readily derivatized with a variety of functional
groups. Esteriﬁcation of PHEMA with pentadecaﬂuorooctanoyl chloride increases the
CO; permeability of these membranes, but still yields a COz/CH4 selectivity of only 8.
Future work aims at exploiting the versatility of ATRP for creating membranes for

specialty applications.

3.2 Experimental

3.2.1 Chemicals and Solutions

Poly(allylamine hydrochloride) (PAH) (Mw= 70,000), sodium
poly(styrenesulfonate) (PSS) (MW: 70,000), 3-mercaptopropionic acid (MPA),
pentadecaﬂuorooctanoyl chloride (97%), pyridine, dimethylforrnamide (DMF,
anhydrous, 99.8%), tetrahydrofuran (THF, anhydrous, inhibitor free, 99.8 %), methanol
(anhydrous, 99.8%), 2-bromopropionylbromide (2-BPB), CuCl (99.999%), CuBr

(99.999%), CuBrz (99%) and 2,2'-bipyridine (bpy, 99%) were used as received from

75

Aldrich. MnClz (Acros) and NaBr (Spectrum) were also used as received. Triethylamine
(Spectrum, 98%) was vacuum distilled over CaHz. 2-Hydroxyethyl methacrylate
(HEMA, Aldrich, 98%, inhibited with 300 ppm hydroquinone monomethyl ether
(MEHQ)) and ethylene glycol dimethacrylate (EGDMA, Aldrich, 98%, inhibited with
100 ppm MEHQ) were puriﬁed by passing them through a column of activated basic
alumina (Spectrum). Deionized water (Milli-Q, 18.2 M0 cm) was used for preparation
of solutions and rinsing. The disulﬁde initiator, (BrC(CH3)2COO(CH2)118)2, was
synthesized according to a literature procedure.16 AnodiscTM porous alumina membranes
(Fisher) with 0.02 um-diameter surface pores were used as supports for membrane
formation. Gold slides (200 nm of sputtered Au on 20 nm Cr on a Si (100) wafer) were
used as substrates for ellipsometry and Fourier transform inﬁ'ared (FTIR) external

reﬂection spectroscopy.

3.2.2 Polymerization of EGDMA

The initial step in the polymerization procedure is the attachment of initiating
groups to the substrate. In some cases, we ﬁrst deposited a multilayer polyelectrolyte
ﬁlm on the substrate and subsequently anchored initiators to these modiﬁed surfaces.
The deposition of polyelectrolytes occurred as follows. Au-coated wafers were cleaned
in a UV/ozone chamber (Boekel UV-Clean model 135500) for 15 minutes and
subsequently immersed in a 1 mM ethanolic solution of MPA for 30 minutes, rinsed with
ethanol and deionized H20, and dried with N2. This procedure yields a carboxylic acid-
terminated surface. Substrates were then immersed in a polycation solution (0.02 M
PAH, 0.5 M NaBr, pH 2.3) for 5 minutes and rinsed with deionized water. (Molarities of

polymers are given with respect to the repeating unit.) Subsequent immersion in a

76

polyanion solution (0.02 M PSS, 0.5 M MnCl2, pH 2.1) for 2 minutes and rinsing with
deionized water yielded a polyelectrolyte bilayer on the surface. A second layer of PAH
was then adsorbed on top of the PAH/PSS layer to provide amine functional groups for
initiator anchoring.

Initiator was attached to PAH/PSS ﬁlms via reaction with 2-BPB in the presence
of triethylamine.32 The gold slide was ﬁrst immersed in a solution of triethylamine
(0.242 g in 10 mL THF), and then the initiator solution (0.432 g of 2-BPB in 10 mL
THF) was added drop-wise while stirring. Because the reaction is exothermic, both
solutions were cooled to 0 °C prior to the reaction. The reaction was stopped after 2
minutes by transferring the slide to a THF solution. Initiator attachment was performed
in a glove box because the acid bromide is moisture sensitive. Further rinsing with ethyl
acetate, ethanol, and deionized water followed by drying with N2 was done outside the
glove box.

Polymerization of EGDMA occurred by immersing the initiator-modiﬁed surfaces
in a solution containing EGDMA (monomer), DMF, deionized H20 and the Cu catalyst
system.31 In this procedure, the monomer mixture, 42 mL of solution containing
EGDMA, H20 and DMF (323:8, vzvzv) was ﬁrst degassed in a three necked ﬂask by three
freeze-pump-thaw cycles. Then, CuCl (180 mg, 1.8 mmol), CuBr2 (120 mg, 0.54 mmol),
and bpy (731 mg, 4.68 mmol) were quickly added to the degassed mixture under a
nitrogen atmosphere. This mixture was immediately degassed using another two freeze-
pump-thaw cycles and then warmed to room temperature with continuous stirring until
the solution became a homogeneous dark brown color. The sealed vessel containing the

monomer/catalyst solution was next transferred into a glove bag that was subsequently

77

purged with nitrogen for at least an hour. The polymerization solution was then
transferred to vials containing polyelectrolyte-coated gold wafers modiﬁed with
initiators, and polymerizations were carried out in the glove bag at room temperature for
different times. After polymerization, substrates were removed from the vessels, rinsed
with DMF, sonicated (1 minute) in DMF, rinsed with THF followed by ethanol, and dried
under a ﬂow of N2.

To grow PEGDMA on porous alumina, one bilayer of PSS/PAH was deposited
directly on UV/ozone-cleaned alumina, and the initiator attachment and polymerization
occurred as described above. Polyelectrolyte depositions were limited to the ﬁltrate side
of the alumina membrane by using a holder, and the initiator attachment and

polymerization were done without the holder.

3.2.3 Polymerization of HEMA and Subsequent Derivatization

For the polymerization of HEMA from gold wafers, substrates were ﬁrst cleaned
in a UV/ozone chamber for 15 minutes, immersed in an ethanol/water (50:50, v:v)
solution for 10 minutes, rinsed with water, and dried with nitrogen. These Au-coated
supports were then immersed in a 1 mM ethanolic solution of the disulﬁde initiator,
(BrC(CH3)2COO(CH2)11S)2, for 12 hours to form a monolayer of initiator. After
monolayer formation, the substrates were rinsed with ethanol and dried with N2.

Polymerization of HEMA occurred by immersion of the initiator monolayer-
coated substrates in a methanolic solution containing HEMA and the Cu catalyst system.
To prepare this solution, 42 mL of HEMA and methanol (1:1, v:v) were ﬁrst degassed in
a three-necked ﬂask by three freeze-pump-thaw cycles. Then, 552 mg (3.84 mmol) of

CuBr, 86 mg (0.39 mmol) of CuBr2, and 1329 mg (8.52 mmol) of bpy were quickly

78

added to the HEMA/methanol while ﬂowing N2 over the solution. This mixture was
immediately subjected to another two freeze-pump-thaw cycles and subsequently stirred
until a homogeneous dark brown solution formed. The sealed vessel containing the
polymerization solution was then transferred to a glove bag, which was purged with N2
for ~l hour. The polymerization solution was ﬁnally transferred into vessels containing
substrates modiﬁed with initiators, and polymerizations were carried out for different
times. Aﬂer polymerization, substrates were removed from the vessels, rinsed with
methanol, sonicated (1 minute) in DMF, rinsed with THF followed by ethanol, and dried
under a ﬂow of N2.

For the polymerization of HEMA from porous alumina, substrates were ﬁrst
coated with gold, and the initiator was attached as a self-assembled monolayer as
described above. Prior to gold coating, substrates were immersed in boiling methanol for
10 minutes and subsequently cleaned in a UV/ozone chamber for 10 minutes. Substrates
were then sputter-coated (ﬁltrate side only) with 5 nm of gold and again UV/ozone
cleaned. Initiator anchoring was done, as described above, by immersion in an ethanolic
solution of disulﬁde initiator (this immersion occurred in an air-tight vessel that was
initially purged with N2 gas). After monolayer formation, PHEMA was polymerized
from the initiator surface as described above.

To derivatize PHEMA coatings, ﬁlms were immersed in 7 mL of anhydrous DMF
containing pentadecaﬂuorooctanoyl chloride (0.08 M) and pyridine (0.1 M). After 15
minutes, ﬁlms were removed from the solution, rinsed with DMF followed by ethanol,

and dried with a ﬂow of nitrogen. The ﬂuorination was monitored by FTIR spectroscopy

79

of ﬁlms on gold wafers and alumina membranes. PHEMA membranes were ﬂuorinated

after initial gas transport measurements.

3.2.4 Film Characterization on Gold Wafers

Ellipsometric thickness measurements were obtained using a rotating analyzer
ellipsometer (model M-44; J .A. Woollam), assuming a ﬁlm refractive index of 1.5. For
each polymer ﬁlm, thicknesses were measured at three different spots and averaged. At
least three samples of each ﬁlm were examined. External reﬂection FTIR spectroscopy
was performed with a Nicolet Magna 560 FTIR using a Pike grazing angle (80°)

accessory.

3.2.5 Film Characterization on Alumina Supports

Film growth was monitored by transmission FT IR spectroscopy (Mattson
Instruments, Inﬁnity Gold) and F ield-Emission Scanning Electron Microscopy (FESEM,
Hitachi S-470011, acceleration voltage of 15 kV). Membranes were coated with 5 nm of
gold for imaging purposes. In the case of cross-sectional images, membranes were

freeze-fractured under liquid N2 prior to sputter coating with gold.

3.2.6 Gas-Permeation Studies

Gas-penneation studies were performed using a permeation cell with a pressure
relief valve, and permeate ﬂux was measured as a ﬁmction of inlet pressure (5-45 psig)
using a soap-bubble ﬂow meter (Figure 3.1). 02, N2, H2, He, CH4, and CO2 were used for
permeation studies, and measurements were performed for each gas separately in the

above order. After examining all gases, O2 permeability was remeasured to check the

80

  
  
  
  

h

c.1-

'-

cml.-

 

.11

iiﬁ!i!i!i!i£i£i!i!iﬁ£i§!ﬁi£i§l1iﬁ!ﬁhi ' ' ' ' 'WIHiITiﬁli'jli’ﬂiﬁﬁﬁliliﬁf

ﬂimﬂiJ ' ,' $232555:

L—
93
<1)
E
E
2
u.
E
E»
D

 

 
     
     
 
   
 

 

5: E,
032 1%
En ,
2‘5 r
em
E 3
‘1

t

 

002

He

 

N2

   

Gas
Controller

 

81

Apparatus for gas-permeation measurements.

Figure 3.1

stability of the membrane, and the O2 ﬂux changed by <10%. For each gas, the
permeation cell was purged several times with the gas of interest over a 20-minute period
to obtain a stable ﬂux value. Gas permeation studies were done for each polymer
(PEGDMA or PHEMA) at several different thicknesses, and for each thickness, three
different membranes were tested. The area of the membrane exposed to gas was 2.0 cm2.
The selectivity of one gas over another was obtained from the ratio of the respective

permeability coefﬁcients at 45 psig.

3.3 Results and Discussion

3.3.1 Synthesis and Characterization of PEGDMA films

The ﬁrst step in growing polymer ﬁlms from a substrate is attachment of an initiator to
the surface. We chose to attach initiators to adsorbed multilayer polyelectrolyte ﬁhns
because electrostatic adsorption provides a convenient way to introduce functional groups
on a surface. Deposition of PAH/PSS/PAH ﬁlms results in a surface rich in amine
groups, and attachment of initiators to this surface via amide linkages occurs easily
(Figure 3.2). Initially, we grew ﬁlms on gold-coated Si wafers because this substrate
facilitates ﬁlm characterization by ellipsometry and reﬂectance FTIR spectroscopy. The
ellipsometric thickness of PAH/PSS/PAH ﬁlms on MPA-coated gold was 4.8 i 0.4 nm,
and the reﬂectance FT IR spectra (Spectrum a, Figure 3.3) of these ﬁlms had strong
sulfonate peaks at 1219 and 1177 cm'1 as well as a number of peaks due to aromatic and
—NH3+ modes.“ After reaction of the ﬁlm with the acid bromide initiator, the reﬂectance
FT IR spectrum looked similar to that of PAH/PSS/PAH, but there was a small increase in

the peak intensity in the amide region (1650-1560 cm'l, Spectrum b, Figure 3.3),

82

080:3 :0 4:3 005.008
000080 050500.038 0 8on gnaw .Ho :owﬁtoﬁbom wagon? 8003? 030835 "Nae. 953%

“ES. 2 0050030 .2055 EE
Ll 00:58 025.0020 “ES.

   

“=20 Ea; 2.280
0 I + + v
L - >3 >08 58

h

 

(anteater/timea?»

o. o
a...» Efuovz
5118:0005 0:37.330 05982028

 

 

Cass EE
03.282028 00053.2

83

 

0.05

 

 

Absorbance

 

.<_°.>__,/\ J
WHO M

KNOW

I l l l l

3500 3000 2500 2000 1500 1000

 

 

 

 

Wavenumbers (cm'1)

Figure 3.3: External reﬂection FTIR spectra of a
PAH/PSS/PAH ﬁlm on a MPA-modiﬁed Au surface before (a)
and after reaction with an acid bromide initiator (b). Spectrum
(c) is from a PEGDMA ﬁlm grown for 20 hours from
PAH/PSS/PAH modiﬁed with initiator.

84

indicating initiator attachment. Film thickness increased only slightly (<O.5 nm), as
would be expected, upon initiator attachment.

Growth of PEGDMA ﬁlms from the initiator-containing surfaces occurred upon
exposure of the substrate to a DMF/H20 solution containing monomer and Cu catalyst.
Scheme 3.1 outlines the synthesis of cross-linked PEGDMA ﬁlms from initiator-
modiﬁed gold surfaces using ATRP. Cumulative ﬁlm thickness increased to 20 i 1.4 nm
in 10 hours, 52 i 1.1 nm in 20 hours and 73 i 6.6 nm in 30 hours. Similar to previous
studies, growth of PEGDMA was characterized by the appearance of a strong carbonyl
peak at 1730 cm'I in the reﬂectance FTIR spectrum of the ﬁlm (Spectrum c, Figure
3.2).31 A C=C stretching band (1637 cm!) from unpolymerized vinyl groups was also
present in the spectrum. By taking the ratio of the IR absorbance of remaining
(unpolymerized) vinyl groups to the absorbance of carbonyl groups and comparing this
with the same ratio in the monomer (EGDMA) spectrum, we estimated that 25% of the
total vinyl groups remain after polymerization. This suggests approximately 50% cross-
linking in PEGDMA ﬁlms. Previous studies done by our group showed that the
C=C/C=C absorbance ratio is constant for different PEGDMA ﬁlm thicknesses,
indicating that the amount of cross-linking is independent of polymerization time 3 ‘

As a control experiment, we also attempted polymerization of EGDMA from
PAH/PSS/PAH-coated wafers (no initiator was attached to this surface). Exposure of the
PAH/PSS/PAH surface to EGDMA and Cu catalyst resulted in no detectable carbonyl
peak in the reﬂectance FTIR spectrum 0 f these films and n 0 increase in e llipsometric

thickness. This result provides further evidence that the derivatization of the amine

85

0 CH3 (u3 u
wNH—C—c—Br + CH2=c-C——OCHZCH20—C—(|3=CH2 EGDMA

 

 

 

H CH3 CH3
CuCI/CuBr2 ,BPY / \ / \
DMF/HZO

—N

RT
1 BPY

t? 9‘3 CH3

wwNH—C—CIZ CHz-C CHz-C CKBr) PEGDMA

 

Scheme 3.1: Growth of cross-linked PEGDMA from an anchored initiator.

86

groups of PAH with 2-BPB was successful and that PEGDMA was chemically attached
to the surface. When growing ﬁlms on porous alumina, we deposited a PSS/PAH ﬁlm
rather than PAH/PSS/PAH because the surface of the alumina is positively charge below
pH 8.45’46 We limited the polyelectrolyte deposition to one bilayer to minimize any
blocking effect from PSS/PAH. Top-view FESEM images show that pores are clearly
open aﬁer depositing one bilayer of PSS/PAH (compare Figure 3.4a and Figure 3.4b).
After subsequent polymerization of EGDMA from these surfaces, pores are completely
covered with the polymer (Figure 3.4c and Figure 3.4d). Both Figure 3.4d and images of
the permeate side of the membrane aﬂer polymerization (not shown) demonstrate that the
interiors of the pores are open, indicating minimal deposition inside pores. Because of
the high pore density in the alumina substrates, ellipsometric measurements were not
possible on these surfaces, and thus PEGDMA thicknesses were estimated from the
cross-sectional FESEM images. Thicknesses of ﬁlms on porous alumina (Table 3.1)

were similar (within ~25%) to those of ﬁlms prepared on gold-coated Si.

3.3.2 Synthesis and Characterization of PHEMA films

To compare cross-linked PEGDMA with uncross-linked polymer ﬁlms, we also
synthesized PHEMA by polymerization from a surface.47 In the case of HEMA
polymerization, we chose methanol as a solvent instead of water because aqueous
conditions yield very high ﬁlm thicknesses that would not be conducive to high ﬂux.19
However, PHEMA ﬁlms grown from initiators attached to PAH/PSS/PAH on gold were
only ~20 nm thick. To achieve larger thicknesses that should allow complete coverage of
surface pores, we grew ﬁlms from a self-assembled monolayer of initiator,

(BrC(CH3)2COO(CI-Iz)1 IS)2, on gold.19 This monolayer likely has a higher density of

87

 

Figure 3.4: FESEM images of the ﬁltrate side of porous alumina (0.02 pm
surface pore diameter) (a) before depositing polyelectrolytes (b) aﬁer
depositing one bilayer of PSS/PAH, and (c) after growth of PEGDMA
from the surface for 20 hours. Image (d) is a cross-section of the
membrane shown in (c).

88

surface-initiating groups than the derivatized polyelectrolytes. After immersing a gold-
coated substrate in disulﬁde initiator solution for 12 hours, the FTIR spectrum of the
surface showed the appearance of a carbonyl peak at 1739 cm'l conﬁrming the formation
of an initiator monolayer (Figure 3.5). Polymerization of HEMA was carried out by
immersing initiator-modiﬁed substrates in HEMA/methanol solutions containing the Cu
catalyst system for different times. Scheme 3.2 outlines the synthetic pathway that we
used to grow PHEMA from gold surfaces. Figure 3.6 shows the variation of PHEMA
ﬁlm thickness with polymerization time when growth occurs from the monolayer of
initiators. Although ﬁlm grth levels off aﬂer a few hours, ﬁlms as thick as 100 nm can
be easily produced.

To grow these ﬁlms on porous alumina rather than gold-coated wafers, we
sputter-coated the alumina with 5 nm of Au and then formed the self-assembled initiator
layer on this surface. Similar to PEGDMA membranes, FESEM images of PHEMA ﬁhns
on porous alumina supports show that PHEMA effectively covers the surface without
ﬁlling the pores. Thicknesses of ﬁlms on porous alumina (determined by FESEM, Table
3.2) are similar (within ~30%) to those of ﬁlms on gold-coated wafers (determined by

ellipsometry, Figure 3.6).

3.3.3 Gas Permeation through Polymer Membranes

Flux through polymeric gas separation membranes, F, generally depends on
transmembrane pressure drop and membrane thickness as shown in equation 3.1, where P
is the permeability coefﬁcient of the material, Ap is the pressure drop, and l is the

. 4
membrane thickness. 8

89

Absorbance

 

0.02

 

M)

 

(a)x1(1

 

 

 

3500 3000 2500 20(1)

Wavenumbers (cm")

90

1500

1000

 

 

Au

 

 

 

 

0 CH3

S-—(CH2)11—O-C— (:3— Br

CH3
0 H3 HEMA

S-—(CH2)11—O-C— C— Bl’

(1)

0 CH3

AU —-S-—-——(CH2)11—O—CI—C::—Bf +

 

 

 

Au

CH3

CH3

0
n
CH2=(|3-—C—O—CHZCH20H
CH3

CuBr/10 mol.% CuBr2/B PY

2’1 CH30H, RT

 

 

 

Scheme 3.2: Growth of PHEMA brushes from a gold surface: (1)

IO CH3 EH3
—-S—(CH2)11—O—C—C:C:H3 CH2— PHEMA

=<-—o —CH2CH20H

/’ \\‘§\____5/
\;_ N/ \
B PY

formation of a monolayer of initiator and (2) polymerization from the

initiator-modiﬁed gold surface.

91

Thickness (nm)

 

2000

 

 

 

1600 ~ I
1200 ~ }’
I

800 - I
400 -

I

0 1 1 l
O 10 20 30 40

Time (hours)

Figure 3.6: Ellipsometric thicknesses of PHEMA ﬁlms as a function of
polymerization time. Thickness values are the average of measurements on
three different ﬁlms, and the error bars represent standard deviations.

92

F=——- 3.1

Theoretically, gas permeation through polymeric membranes is usually described by the
solution-diffusion model. The gas dissolves in the high-pressure side ofthe membrane,
diffuses through the membrane because of a concentration gradient, and desorbs at the
low-pressure side. Hence, the permeability of a gaseous penetrant, P, can be written as a
product of an average diffusivity, D, and an effective solubility coefﬁcient, S, for the

penetrant in the polymer matrix.49

P =D><S 3.2

The solubility coefﬁcicient depends on the condensability of the penetrant, polymer-
penetrant interactions, and the free volume of the polymer matrix. Diffusivity provides a
measure of the mobility of the penetrant in the polymer matrix and is determined by the
packing of polymer chains and the size and shape of the penetrating gas. The ideal

selectivity of a membrane, (IA/.8 , is the ratio of the permeabilities of the two gases.

CIA/B = ~—-- 3.3

By substituting equation (2) into equation (3),

— 3.4

93

where DA/DB is the diffusivity selectivity and S A/SB is the solubility selectivity. Thus to
improve the permselectivity of one gas over the other, one must increase either the

diffusivity selectivity or the solubility selectivity or both.

3.3.4 Gas Permeation through Composite PEGDMA Membranes

Figure 3.7 shows gas ﬂuxes through a PEGDMA membrane as a function of
transmembrane pressure drop. Fluxes of different gases through the membrane increase
in the approximate order N2,CH4,02<He,H2<COz. Flux increases linearly with pressure
drop for all gases except C02. The nonlinear increase in C02 ﬂux with increasing
pressure suggests a small amount of plasticization, which is surprising considering that
pressures are less than 45 psi g and the membrane is partially cross-linked. Table 3.1
summarizes the gas permeation data for several gases at three different PEGDMA ﬁlm
thicknesses. As the ﬁlm thickness increases, selectivity initially increases and plateaus at
a ﬁlm thickness of ~50 nm. (Selectivity appears to decrease slightly for the 80 nm-thick
ﬁlm, but the small difference between selectivities of 50 and 80 nm-thick ﬁlms is still
within experimental error.) The lower selectivity of 30 nm-thick PEGDMA is likely due
to areas where the ﬁlm does not completely cover the substrate. Presumably, thicker
ﬁlms yield a more complete surface coverage.

Selectivity in these membranes is a function of both polymer and gas properties.
Cross-linking reduces the polymer chain mobility and should enhance chain packing and
the rigidity of the polymer.41 ’48 Tighter packing should favor the diffusion of smaller
molecules. For example, C02, which is a linear molecule with a kinetic diameter of 3.3
A, should diffuse faster than CH4, which is a spherical molecule with a kinetic diameter

of 3.8 A.“ However, the possible plasticization noted above suggests that the PEGDMA

94

 

 

12

 

 

 

 

 

 

 

0 CO2 .
A H2
10* C] He O
o 02
v CH4 .
5‘ 8' ' N2
’5
Tm O
a 6-
.2 o
f». 4- . ‘
g A E]
s; A C]
s ' a 0
LT. 2- . a O
0 D O
. O
O
0_ 2 9 Q 9 9 3 i 3 3
O 10 20 30 40 50

Pressure (psig)

Figure 3.7: Fluxes of different gases through a PEGDMA membrane (50
nm thick) as a function of transmembrane pressure drop. The outlet
pressure was 1 atm and the measurements were performed at room
temperature. PEGDMA was deposited on porous alumina.

95

.mmma mmv 00.50008 00 :05

man 210 0:8: .x. 3 5000 0.0 00538000 ~Z\~OU 0:0 £U\~OU 8030820003 mo 00:000m 600.8000 000
m020> 023000—00 0w000>0 05 080 .00805808 8000.030 00:: 0o.“ 000030—00 000B Ami; m0 00v 00333020?

00805808 8000.006 008: 08 08065000

325008000 («0 0w000>0 08 000 0020> 00003 A 850000 was: 0003200 0003 08068000 0923008005“0

.00w08m Simmm 88.0 00008300 0003 000005—0500

8380 0 ~88 \80 3.09 M80 2-3 x“ n 005m _

 

 

 

 

 

 

 

 

mﬁom :3 036; N02 ﬁon”: 0.30.0 0600.0 No3.— modﬁmg ow om
000m mi: v.oH~.~ 000m foam; mdﬁwd Noﬁmw Nomad ﬁoﬁam om om
0.0.3.0 0.0030 Noam; Sham N0 w 3”? @303 73 30. em N4
£000 £0000 ~23 60 .00 N: 0: N2 N0 .005 Awﬁwmv
0 z: 00 0 A8080 v m :22 0o 0 0 _ 008000 00 000 :0 0 000MH030 8200:
u 0. .0 _ m m 00 . b U 5.0 00 _ _ U _.m 008.com
00805808

<EQOmm 08 005330200 002 080 00000535 85 000080000 .08065000 055008008 000 36 030,—.

96

is not especially rigid. Solubility selectivity in PEGDMA is also possible due to
interactions between polarizable C02 and the polar carbonyl groups of PEGDMA. Koros
showed that COz/CHa selectivity increases with increasing concentration of carbonyl
groups in a polymer due to the attraction between C02 and polar carbonyl groups.50
Freeman and coworkers also demonstrated that the introduction of polar nitro groups into
the backbone of polysulfone enhances COz/CH4 selectivity due to interaction of the nitro
groups with C02.51

Diffusivity and solubility selectivities should be much more effective in
separating COz/CH4 than OZ/Nz.48 The kinetic diameters of 02 and N2 differ by only
0.018 nm.48 Accordingly, we didn’t observe large OZ/NZ selectivities. A recent study of
polyimide membranes showed that cross-linkingincreases COZ/CH4 selectivity but has
little effect on 02/N2 separation.10

Calculated permeability coefﬁcients for PEGDMA (Table 3.1) are considerably
higher than for poly(methyl methacrylate) (PMMA), which has a C02 permeability
coefﬁcient of about 0.6 Barrers. However, as might be expected, the COZ/CHa selectivity
of PMMA ﬁlms is about 5-fold higher than that of PEGDMA.52’53 Poly(ethyl
methacrylate) (PEMA), in contrast, has a C02 permeability coefﬁcient of 7 Barrers and a
COz/CHa selectivity of 20,54 and thus is similar in selectivity to PEGDMA, but slightly
less permeable. These data suggest that the bulkier side groups in both PEMA and
PEGDMA yield higher permeabilities and lower selectivities than PMMA. (All of these
polymers should be glassy at room temperature.) Although PEGDMA is cross-linked, its
bulky side chains still likely result in a higher free volume and, hence, higher

permeability than that of either PMMA or PEMA. Studies done on various

97

methacrylates also showed that the free volume of the polymers depends on the side
chain.54'57
Hirayama and coworkers recently reported fabrication of partially cross-linked
ﬁlms from poly(ethylene oxide) (PEO) dimethacrylate. C02 permeabilities of these ﬁlms
are about 2-fold higher than those of PEGDMA,43 and C02/N2 selectivity is 65. The high
selectivity likely stems from interaction between C02 and PEO, and the high
permeability probably results from the bulky PEO groups, which are amorphous.
However, the PEO dimethacrylate ﬁlms were 100-200 pm thick, and thus the PEGDMA

membranes grown from a surface still allow ZOO-fold higher ﬂuxes. In the future, growth

of PEO dimethacrylate from a surface could result in very high C02 ﬂux.

3.3.5 Gas Permeation through PHEMA Membranes

To compare cross-linked PEGDMA membranes with linear polymer ﬁlms, we
studied the gas-permeation properties of PHEMA that was grown from the surface of
porous alumina. Figure 3.8 shows the ﬂuxes of several gases through a PHEMA
membrane as a function of transmembrane pressure. Gas permeation through these
membranes is typical of Knudsen diffusion, as ﬂuxes of the different gases are inversely
proportional to the square root of molar mass. Unlike PEGDMA, PHEMA ﬁlms show
very little selectivity for CO2 over CH4. The observed C02/CHa selectivity was 0.7 and
02/N2 selectivity was only 0.9. Selectivities and calculated permeability coefﬁcients for
PHEMA at three different ﬁlm thicknesses are given in Table 3.2. As mentioned above,
PEMA has a C02/CH4 selectivity of 20 with a C02 permeability coefﬁcient of 7 Barrers
at 35 °C. 54 Both PHEMA and PEMA are glassy polymers with similar Tg values (T8 of

PHEMA is 85 °C and T8 of PEMA is 65 °C58), hence Tg shouldn’t have a dramatic effect

98

 

 

 

 

 

 

 

 

1o - ‘ ”2 A
[3 He
V CH4
I N2 A
.g‘ 8 " 0 02 A
5 0 C02 0
'70)
A A
E 6 r- U
9 n
m A
E
U
v D
8 A v
‘2? 4 ' D
x v
E A I: v
v 5 5
2 ~ ‘ V §
[:1 V G a
A V a
D 8
5 5
0 l l I L
0 10 20 30 40 50

Pressure (psig)

Figure 3.8: Fluxes of different gases through a PHEMA membrane (60
nm thick) as a function of transmembrane pressure drop. The outlet
pressure was 1 atm and the measurements were performed at room
temperature. PHEMA was deposited on porous alumina.

99

008000.: 0.0 0020> 0m0u0>0 05 0:0 080.5808 80:00.00 00:5 00.: 00003000 0:03 3me 3 00V 00_:>000_0m a
00:05808 8000020 00:5 00.: 08205000

053008000 .00 0w000>0 05 0:0 0020> 00:05 A 500:3 800 0000—0200 0003 08065000 002500809” 0
.m0w08_ EMmmm 80.0 00008000 0003 000005050 0

 

 

 

 

 

 

 

 

ad 0.0 ad 700 mduuwxz VHMN gum— m.oﬂ¢.0 mdﬂNS 00 v
ad ed m6 700 ONﬁ: WHMN MH©~ 700 700. mm N
0.6 0.6 ad 3% NﬂN~ mﬂaN mﬂom Nﬂa 70w w~ g
£000 £0000 N200 N00 £0 a: 0: N2 N0 .005 0320
30¢on DEC.
_ -
033000—0m $00809 008005000 323008000 000 00003000 87m :0008
008.com
00:05:08

0.0/Ema 0000030008: :0.“ 000300200 #002 0:0 000005—030 EE 00008000 08003000 055008.800 000 "Nd 030,—.

100

on permeability. The permeability of C02 obtained in our study is similar to the reported
C02 permeability for PEMA, but the observed low selectivity of PHEMA compared with
the selectivity reported for PEMA suggests that the hydroxyl groups in PHEMA
dramatically alter polymer packing.

To improve the COz/CH4 or 02/N2 selectivity of PHEMA membranes, we

explored derivatization of PHEMA with ﬂuorinated compounds. One attractive feature

 

of PHEMA is that its hydroxyl groups can be easily derivatized with various acid E
chlorides or carbonyldiimidazole to introduce different functional groups.19 We reacted ’
PHEMA with pentadecaﬂuorooctanoyl chloride in the presence of a base to obtain r
PHEMA with perﬂuorinated side chains (scheme 3.3). Disappearance of the alcohol %

peak (3500-3300 cm") in the reﬂectance FTIR spectrum of PHEMA (Spectrum b, Figure
3.9) indicates conversion of the hydroxyl groups to ﬂuorinated esters. The appearance of
a ﬂuorinated ester peak at 1800 cm" and CFx peaks at 1250 cm”1 also conﬁrm the
esteriﬁcation of PHEMA. Based on the density of HEMA and poly(1,l ’-
dihydroperﬂuorooctyl methacrylate, we would expect to see a more than 100 % increase
in thickness upon ﬂuorination. Nevertheless, the ellipsometric thicknesses of PHEMA
ﬁlms increased only by ~70 % after reaction with pentadecaﬂuorooctanoyl chloride,
suggesting <100 percent derivatization of hydroxyl groups or very dense ﬁlms.
However, the disappearance of the alcohol peak in the IR spectrum of PHEMA points to
virtually quantitative derivatization. In addition to examining PHEMA ﬁlms on gold
wafers, we also characterized the ﬂuorinated ﬁlms on alumina with transmission FTIR
(Figure 3.10). Appearance of a carbonyl peak around 1800 cm‘1 suggests the

incorporation of perﬂuorinated groups. The relatively small increase in ﬁlm thickness

101

 

S? W 1*“
Au ——S—(CH2)11—O—C—(I3 CHz—C Br
CH3 o:\/ n PHEMA

O— CH2CH2 OH

 

 

 

S0

 

45A \ (I? i
kg CF3{CF2%C——Cl 5:

 

 

 

 

 

DMFR T
ﬁr
Fl uon' nated-PHEMA
9 0 9%
AU —S—(CH2)11—O—C—IC CHz—C Br
CH3 0:< n
O— CHzCHzO‘nzf Csz’eCF3

0

Scheme 3.3: Derivatization of PHEMA with perfluorooctanoyl chloride.

102

 

 

 

 

 

 

 

0. 1
8
C
N
g
<

(a) M
N
I 1 1 l l 1
3500 3000 2500 2000 1500 1000

Figure 3.9: Reﬂectance FTIR spectra of a PHEMA ﬁlm grown
from (BrC(CH3)2COO(CH2)1 IS)2 on gold (a) before and (b) after

Wavenumbers (cm'l)

reaction with pentadecaﬂuorooctanoyl chloride.

103

 

 

 

 

(b)

Absorbance

(a)

 

l L l l l

4000 3 500 3000 2500 2000 l 500 1000

 

 

Wavenumbers (cm")

Figure 3.10: Transmission FTIR spectra of a PHEMA ﬁlm
grown from (BrC(CH3)2COO(CH2)1 18); on gold-coated
porous alumina (a) before and (b) aﬁer reaction with
pentadecaﬂuorooctanoyl chloride.

104

. ' Q.__' J‘

after ﬂuorination may indication a high ﬁlm density (low free volume) aﬁer
derivatization. This is consistent with the fact that ﬂuorinated PHEMA is less permeable
than poly(1,1’—dihydroperﬂuorooctyl methacrylate (vide infra), and the brush-like
structure of these ﬁlms might account for this high density. .

Gas-permeation studies with PHEMA were repeated after ﬂuorination, and Figure

3.11 shows the ﬂuxes of several gases as a function of transmembrane pressure drop.

 

3‘
After ﬂuorination, the ﬂuxes of various gases decrease in the order C02>He>H2>02>N2, l
CH4 and are no longer dependent solely on the molar masses of the gases. Because ’
ﬂuorination enhances C02 ﬂux relative to other gases, C02/CH4 selectivity increases
~10-fold compared to non-ﬂuorinated ﬁlms. Similar to PEGDMA, ﬂuorinated PHEMA 1%

also shows an increase in C02/CH4 selectivity with increasing ﬁlm thickness. The
highest selectivity was obtained for a lOO-nm thick ﬂuorinated ﬁlm (Table 3.3). At ﬁhn
thicknesses higher than 100 nm, ﬂux values for some gases were lower than the detection
limits of our ﬂow meter.

The increase in C02/CH4 selectivity and C02 permeability after ﬂuorination
probably occurs due to an increase in C02 solubility in the polymer matrix. Compared
with the other gases we tested, C02 has a high polarizability and quadrupole moment48
that should allow it to interact with the polar ﬂuorinated side chains (the C-F dipole
moment is 1.39 D”). Thus C02 solubility is higher after ﬂuorination compared to other
gases hence we observe higher C02/CH4 selectivity.

Arnold and coworkers reported the gas permeation properties of membranes
prepared from poly(1,1’-dihydroperﬂuorooctyl acrylate) (PF 0A) and poly(1,1’-

dihydroperﬂuoro methacrylate) (PFOMA).60 These membranes are similar to ﬂuorinated

105

 

 

 

 

 

 

 

 

 

6
0 CD2 0
D He
5 " a H2 o
0 02
o
a," I N
S 4' c; ‘73
Tm V 4 0 D i
F; El
E’l 3 - o
n D ,
E g
8 o D ‘
‘2? 2 L D A
3 ° C. + A
u. A O
. D A <> 0
1 - A O
D ‘ 0 <7
0
‘3 3 o v V V e
6 V V O
OH ‘ ‘ ‘
O 10 20 30 40 50

Pressure (psig)

Figure 3.11: Fluxes of different gases through a ﬂuorinated-
PHEMA membrane (100 nm thick) as a function of transmembrane
pressure drop. The outlet pressure was 1 atm and the measurements
were performed at room temperature.

106

 

 

 

008000: 0:0 m00:0> 0m00>0 050:0 a00:0:0808 800.80 00:5 :8 00:23:00 003 80:: 00 :00 000300009
00:0:0808 800.80 00:5 :8 08000000
30500800 00 00890 05 0:0 m00:0> 000:: .: :000000 0:00 0808800 00>» 08000000 35000800..

.0008: $5.0m": 800 00:08:00 003 0000:0050.w

 

 

 

 

 

 

 

0000.0 00000 0.000.: 00mm 00000 NON: 00m: 0000. m :00 00:
:.:0\r.0 0000.0 :.00w.: 00: 0 0000.0 ~00: m0:~ N.00m.0 0.0000 00
0000.0 0000.0 :.000.: m0m~ N00 0.00: ~00: :00 :00 00
~Z\~OU £0000 NZ\~O Nov 0::0 mm: 0:: N7: N0 aA85
00508:.
o3:>::00:0m @0803 0080000000 30500800 000 0808800 8:“:
005808 0.0/man:

00:08:03: :8 0003:0000 :000: 0:0 08000000 3050080: 000 00000585 8:: 00:08:00 "ﬁn 0:00P

107

PHEMA and have similar C02/CH4 and 02/N 2 selectivities. However, the permeability

coefﬁcients of C02 in PF 0A and PF OMA are, respectively,18-fold and 5-fold higher

than that of ﬂuorinated-PHEMA.. The extra ester group in ﬂuorinated PHEMA probably

reduces the chain mobility and lowers the permeability, and as mentioned above,
derivatized ﬁlms may have a high density.. Other studies also showed that ﬂuorinated

side chains greatly enhance C02 permeability, and hence C02/CH4 selectivity.6"62

3.4 Conclusions

Surface ATRP provides a convenient way of synthesizing ultrathin, cross-linked
ﬁlms and linear polymer brushes on porous supports. FTIR and F ESEM conﬁrm ﬁlm
formation on these surfaces. Gas permeation studies with these ﬁlms indicate that
PEGDMA has a C02/CH4 selectivity of ~20 and a C02 permeability coefﬁcient of 20
Barrers. Unlike cross-linked PEGDMA, linear PHEMA ﬁlms show only Knudsen-
diffusion based selectivity. However, esteriﬁcation of the hydroxyl groups of PHEMA
with pentadecaﬂuorooctanoyl chloride increases the C02/CH4 selectivity to ~8 and the
C02 permeability coefﬁcient to ~20 Barrers. The derivatizability of PHEMA may make

it a suitable candidate for specialty separations.

108

 

 

.305

(1)

(2)
(3)
(4)

(5)

(6)

(7)

(8)
(9)
(10)
(11)

(12)

(13)

(14)

(15)

(16)

(17)
(18)

References and Notes

Ferjani, E.; Lajimi, R. H.; Deratani, A.; Roudesli, M. S. Desalination 2002, 146,
325-330.

Dutta, B. K.; Sikdar, S. K. Environ. Sci. & Tech. 1999, 33, 1709-1716.
Li, X.-G.; Huang, M.-R. J. Appl. Polym. Sci. 1997, 66, 2139-2147.

Karnada, K.; Kamo, J .; Motonaga, A.; Iwasaki, T.; Hosokawa, H. Polym. J.
1994, 26, 833-839.

Marek, M., Jr.; Brynda, E.; Houska, M.; Schauer, J .; Hynek, V.; Sipek, M.
Polymer 1996, 37, 2577-2579.

Baker, R. W. Ind. Eng. Chem. Res. 2002, 41, 1393-1411.

Pinnau, 1.; Freeman, B. D. In Polymer Membranes for Gas and Vapor Separation;
Pinnau, 1., Freeman, B. D., Eds.; American Chemical Society: Washington, DC,
2000; Vol. 744, pp 1-22.

Liu, C.; Martin, C. R. Nature 1991, 352.50-52.

Petersen, R. J. J. Membr. Sci. 1993, 83, 81-150.

Staudt-Bickel, C.; Koros, W. J. J. Membr. Sci. 1999, 155, 145-154.

Hayakawa, Y.; Terasawa, N.; Hayashi, E.; Abe, T. J. Appl. Polym. Sci. 1996, 62,
951-954.

Yanagishita, H.; Kitamoto, D.; Haraya, K.; Nakane, T.; Okada, T.; Matsuda, H.;
Idemoto, Y.; Koura, N. J. Membr. Sci. 2001, 188, 165-172.

Pinnau, 1.; Koros, W. J. Ind. Eng. Chem. Res. 1991, 30, 1837-1840.
Jordan, R.; Uhnan, A. J. Am. Chem. Soc. 1998, 120, 243-247.

Jordan, R.; Ulman, A.; Kang, J. F.; Rafailovich, M. H.; Sokolov, J. J. Am. Chem.
Soc. 1999, 121, 1016-1022.

Shah, R. R.; Merreceyes, D.; Husemann, M.; Rees, 1.; Abbott, N. L.; Hawker, C.
J .; Hedrick, J. L. Macromolecules 2000, 33, 597-605.

Huang, X.; Wirth, M. J. Macromolecules 1999, 32, 1694-1696.

Huang, X.; Doneski, L. J .; Wirth, M. J. Chemtech 1998, 28, 19-25.

109

 

 

(19)

(20)

(21)

(22)

(23)

(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)

(32)

(33)
(34)

(35)

Huang, W.; Kim, J .-B.; Bruening, M. L.; Baker, G. L. Macromolecules 2002, 35,
1 175-1 179.

Ejaz, M.; Yamamoto, S.; 0hno, K.; Tsujii, Y.; Fukuda, T. Macromolecules 1998,
31, 5934-5936.

Zhao, B.; Brittain, W. J. J. Am. Chem. Soc. 1999, 121, 3557-3558.

Huang, W.; Skanth, 0.; Baker, 0. L.; Bruening, M. L. Langmuir 2001, 17, 1731-
1736.

Zhao, B.; Brittain, W. J .; Zhou, W.; Cheng, 8. Z. D. J. Am. Chem. Soc. 2000, 122,
2407-2408.

Prucker, 0.; Riihe, J. Macromolecules 1998, 31 , 602-613.

Husemann, M.; Mecerreyes, D.; Hawker, C. J .; Hedrick, J. L.; Shah, R.; Abbott,
N. L. Angew. Chem. Int. Ed. 1999, 38, 647-649.

Jeon, N. L.; Choi, I. S.; Whitesides, G. M.; Kim, N. Y.; Laibinis, P. E.; Harada,
Y.; Finnie, K. R.; Girolarni, G. S.; Nuzzo, R. G. Appl. Phy. Lett. 1999, 75, 4201-
4203.

Wang, J .-S.; Matyjaszewski, K. Macromolecules 1995, 28, 7901-7910.

Kato, M.; Kamigaito, M.; Sawamoto, M.; Higashimura, T. Macromolecules 1995,
28, 1721-1723.

Huang, X.; Wirth, M. J. Anal. Chem. 1997, 69, 4577-4580.

Matyjaszewski, K.; Miller, P. J .; Shukla, N.; Immaraporn, B.; Gelrnan, A.;
Luokala, B. B.; Siclovan, T. M.; Kickelbick, G.; Vallant, T.; Hoffmann, H.;
Pakula, T. Macromolecules 1999, 32, 8716-8724.

Huang, W.; Baker, G. L.; Bruening, M. L. Angew. Chem. Int. Ed. 2001, 40, 1510-
1512.

Kim, J.-B.; Bruening, M. L.; Baker, G. L. .1. Am. Chem. Soc. 2000, 122, 7616-
7617.

Jones, D. M.; Huck, W. T. S. Adv. Mater. 2001, 13, 1256-1259.
Kita, H.; Inada, T.; Tanaka, K.; Okamoto, K. J. Membr. Sci. 1994, 87, 139-147.

Bos, A.; Pont, I. G. M.; Wessling, M.; Strathmann, H. J. Polym. Sci., Part B:
Polym. Phys. 1998, 36, 1547-1556.

110

 

 

(36)
(37)

(38)

(39)

(40)

(41)

(42)

(43)

(44)
(45)

(46)

(47)

(48)
(49)
(50)

(51)

(52)

(53)

(54)

McCaig, M. 8.; Paul, D. R. Polymer 1999, 41 , 629-637.
Rezac, M. E.; Schoberl, B. J. Membr. Sci. 1999, 156, 211-222.

Bos, A.; Punt, I. G. M.; Wessling, M.; Strathmann, H. Sep. Purif. Tech. 1998, 14,
27-39.

Wright, C. T.; Paul, D. R. J. Membr. Sci. 1997, 129, 47-53.

Rezac, M. E.; Sorensen, E. T.; Beckham, H. W. J. Membr. Sci. 1997, 136, 249-
259.

Liu, Y.; Wang, R.; Chung, T. S. J. Membr. Sci. 2001, 189, 231-239.

Hirayama, Y.; Tanihara, N.; Kusuki, Y.; Kase, Y.; Haraya, K.; Okamoto, K. i. J.
Membr. Sci. 1999, 163, 373-381.

Hirayama, Y.; Kase, Y.; Tanihara, N.; Sumiyarna, Y.; Kusuki, Y.; Haraya, K. J.
Membr. Sci. 1999, 160, 87-99.

Stair, J. L.; Harris, J. J.; Bruening, M. L. Chem. Mater. 2001, 13, 2641-2648.
Parks, 0. A. Chem. Rev. 1965, 65, 177-198.

Bluhm, E. A.; Bauer, E.; Chamberlin, R. M.; Abney, K. D.; Young, J. S.;
Jarvinen, G. D. Langmuir 1999, 15, 8668-8672.

Actually, PHEMA is partially branched due to intermolecular transesteriﬁcation
and in some cases is lightly cross-linked.

Ghosal, K.; Freeman, B. D. Polym. Adv. Tech. 1994, 5, 673-697.
Stern, S. A. J. Membr. Sci. 1994, 94, 1-65.
Koros, W. J. J. Polym. Sci., Part B: Polym. Phys. 1985, 23, 1611-1628.

Ghosal, K.; Chem, R. T.; Freeman, B. D.; Savariar, R. J. Polym. Sci, Part B:
Polym. Phys. 1995, 33, 657-666.

Hoover, J. M.; Smith, S. D.; DeSimone, J. M.; Ward, T. C.; McGrath, J. E.
Polymer Preprints 1987, 28, 390-392.

Robeson, L. M. J. Membr. Sci. 1991, 62, 165-185.

Chiou, J. 8.; Paul, D. R. J. Membr. Sci. 1989, 45, 167-189.

111

 

(55)

(56)

(57)

(58)

(59)

(60)

(61)

(62)

Raymond, P. C.; Paul, D. R. J. Polym. Sci, Part B: Polym. Phys. 1990, 28, 2079-
2102.

Wright, C. T.; Paul, D. R. Polymer 1997, 38, 1871-1878.

Chiou, J. S.; Barlow, J. W.; Paul, D. R. J. Appl. Polym. Sci. 1985, 30, 1173-1186.
Andrews, R. J .; Grulke, E. A. In Polymer handbook; Brandrup, J ., Immergut, E.
H., Grulke, E. A., Eds; John Wiley and Sons: New York, 1999; pp VI 193- VI
277.

Yampolskii, Y. P.; Alentiev, A. Y.; Shishatskii, S. M.; Shantarovich, V. P.;
Freeman, B. D.; Bondar, V. I. Polymer Preprints 1998, 39, 884-885.

Arnold, M. E.; Nagai, K.; Freeman, B. D.; Spontak, R. J .; Betts, D. E.; DeSimone,
J. M.; Pinnau, I. Macromolecules 2001, 34, 5611-5619.

Nagale, M.; Kim, B. Y.; Bruening, M. L. J. Am. Chem. Soc. 2000, 2000, 11670-
1 1678.

Xiao, K. P.; Ham's, J. J .; Park, A.; Martin, C. M.; Pradeep, V.; Bruening, M. L.
Langmuir 2001, 1 7, 8236-8241.

112

 

VE‘: '. '.. V‘) v

5'-

 

CHAPTER 4

Ion Transport through Grafted Poly(2-hydroxyethyl methacrylate)
Membranes and their Derivatives

4.1 Introduction

Chapter 3 demonstrated the fabrication of ultrathin skins on porous supports using
surface-initiated atom transfer radical polymerization. This chapter investigates the ion
permeability of poly(2-hydroxyethyl methacrylate) (PHEMA) membranes that were also
prepared by ATRP from a surface. Although such membranes did not prove highly
selective in gas separations, they do allow selective transport of monovalent ions.
Derivatization of PHEMA with a cross-linking agent, succinyl chloride, results in C1’
/SO42' selectivities as high as 300.

Such selectivities may be important in nanoﬁltration (NF), which has become an
important area of research in membrane-based separations. This technique is widely used
in applications such as water softening and puriﬁcation, removal of heavy metals from
water streams, waste-water reclamation, and separation of organic solutes."2 NF is
sometimes preferable to reverse osmosis because it occurs at lower pressures and hence,
has lower energy costs. The separation characteristics in NF fall between reverse
osmosis and ultraﬁltration, and separation is based on sieving and electrostatic effects?”4
Typical NF membranes are synthesized from polymers such as polysufones, polyamides,
modiﬁed aromatic polyamides, and derivatives of polyvinyl alcohols.S

In 1965, Baddour and coworkers demonstrated that hydrogels made from
PHEMA are capable of desalinating brackish or sea water.6’7 They showed that

membranes synthesized by copolymerization of HEMA and ethylene glycol

113

 

 

dimethacrylate have NaCl rejections up to 87.6%.(”7 Haldon and Lee examined the
permeability of PHEMA membranes that were prepared by copolymerization with
different cross-linkers (ethylene glycol dimethacrylate, trimethylolpropane
trimethacrylate (TPT) or pentaerythritol tetramethacrylate) and showed that the water-
permeability of the membrane depends on the cross-linking density.8 A similar study by
J adwin and coworkers found that water permeability through cross-linked PHEMA
membranes decreases from 6X104’ m3-m/m2-day to 6><10’9 m3-m/m2-day, and salt
rejection increases from 78% to 94% as the amount of cross-linking with TPT increases
from 0 to 11 mole percent.7 However, such membranes were prepared by solution
casting followed by photoirradiation, and thus, the thicknesses of these materials were
relatively high (100-500 pm). This, of course,_results in unacceptably low ﬂuxes.

The main objective of this work was to determine whether ultrathin, PHEMA
skins prepared by ATRP from a surface can successfully separate different ions. Ion-
transport studies with composite PHEMA membranes showed moderate selectivities (Cl'
/S042'selectivity of 15, K+/Mg2+ selectivity of 47 and Cl'/Fe(CN)63' selectivity of 164),
but Cl' ﬂuxes in diffusion dialysis were quite low (15% of that through the bare alumina
support). (Higher selectivities were obtained with even thicker ﬁlms, but ﬂux was even
lower.) The relatively low ﬂuxes occurred in spite of the fact that ﬁhn thickness was less
than 100 nm, suggesting that ﬁlms are relatively dense. One advantage of PHEMA is
that it can be easily tailored for speciﬁc applications through derivatization of its
hydroxyl groups.9 In chapter 3, I showed that derivatization of PHEMA with
pentadecaﬂuorooctanoyl chloride enhanced gas-transport selectivities. This chapter

shows that reaction of PHEMA with succinyl chloride dramatically increases ion-

114

 

 

transport selectivities, while still allowing reasonable ﬂux values with 28 nm-thick ﬁlms

(Cl‘ ﬂux was 50% of that through bare alumina).

4.2 Experimental

4.2.1 Chemicals and Solutions

Poly(allylamine hydrochloride) (PAH) (MW: 70,000), sodium
poly(styrenesulfonate) (PSS) (MW: 70,000), 3-mercaptopropionic acid (MPA), pyridine,
dimethylformamide (DMF, anhydrous, 99.8%), tetrahydrofuran (THF, anhydrous,
inhibitor free, 99.8%), ethyl acetate, ethanol, 2-bromopropionylbromide (2-BPB), CuCl
(99.999%), CuBr (99.999%), CuBr2 (99%), 2,2'-bipyridine (bpy, 99%), 1,1'-
carbonyldiimidazole (CD1, 98%), 2-hydroxymethyl-18-crown-6 (90%) and succinyl
chloride (90%) were used as received from Aldrich. MnCl2 (Acros) and NaBr
(Spectrum) were also used as received. Triethylamine (Spectrum, 98%) was vacuum
distilled over CaH2. 2-Hydroxyethyl methacrylate (HEMA, Aldrich, 98%, inhibited with
300 ppm hydroquinone monomethyl ether (MEHQ)) was puriﬁed by passing it through a
column of activated basic alumina (Spectrum). Deionized water (Milli-Q, 18.2 M0 cm)

was used for preparation of solutions and rinsing. AnodiscTM porous alumina membranes

(Fisher) with 0.02 um-diameter surface pores were used as supports for membrane
formation. Gold-coated slides (200 nm of sputtered Au on 20 nm Cr on a Si (100) wafer)
were used as substrates for ellipsometry and Fourier transform infrared (FTIR) external

reﬂection spectroscopy.

115

 

4.2.2 Polymerization of HEMA from Gold-coated wafers and Porous Alumina

For p olymerization o f H EMA from gold w afers, 2 -BPB w as immobilized o n a
multilayer polyelectolyte ﬁlm (PAH/PSS/PAH) using the procedure described in Chapter
3. In the case of porous alumina, the initiator was attached to a PSS/PAH bilayer.
Polymerization of HEMA occurred by immersion of the initiator-coated substrates in an
aqueous solution containing HEMA and a Cu catalyst system.9 To prepare the catalyst
solution, 42 mL of HEMA and deionized water (1:1, v:v) were ﬁrst degassed in a three-
necked ﬂask by three freeze-pump-thaw cycles. Then, 55 mg (0.55 mmol) of CuCl, 36
mg (0.16 mmol) of CuBr2, and 244 mg (1.56 mmol) of bpy were quickly added to the
HEMA/water solution under a ﬂow of nitrogen. The mixture was immediately subjected
to another two freeze-pump-thaw cycles and subsequently stirred until a homogeneous
dark brown solution formed. The sealed vessel containing the polymerization solution
was then transferred to a glove bag, which was purged with N2 gas for ~1 hour. The
polymerization solution was ﬁnally transferred into vessels containing substrates
modiﬁed with initiators, and polymerizations were carried out for different times. After
polymerization, substrates were removed from the vessels, rinsed with deionized water,
sonicated (1 minute) in DMF, rinsed with THF followed by ethanol, and dried under a

ﬂOW Osz.

4.2.3 Derivatization of PHEMA with Crown ethers

To couple crown-ethers to PHEMA, a ﬁlm-coated substrate was immersed into a
0.2 M solution of CD1 in DMF for 12 h and subsequently rinsed with DMF. CDI-
functionalized PHEMA substrates were then immersed in a 10-mL DMF solution

containing 2-hydroxymethyl-18-crown-6 (0.1 M) and triethylamine (0.1 M). The

116

 

 

reaction was carried out at 70 °C for 4.5 days, after which the substrate was rinsed with
DMF, followed by ethanol and dried with a ﬂow of nitrogen. Derivatization was

performed after ion-permeability studies with the underivatized PHEMA membrane.

4.2.4 Chemical Cross-linking of PHEMA

PHEMA-coated substrates were immersed in 10 mL of DMF containing succinyl
chloride (0.1 M) and pyridine (0.1 M). After 10 minutes, substrates were removed from
the reaction solution, rinsed with DMF, and dried with N2. Derivatization was performed

after ion-permeability studies.

4.2.5 Film Characterization

Ellipsometric measurements were obtained with a multiwavelength, rotating
analyzer ellipsometer (model M-44; J .A. Woollam) using WVASE32 software at an
incident angle of 75°. The refractive index of the ﬁlms at all wavelengths was assumed
to be 1.5. For each substrate, thicknesses were measured at three different spots and
averaged. Reﬂectance FTIR spectroscopy was performed using a Nicolet Magna-IR 560
spectrometer containing a PIKE grazing angle (80°) attachment. The spectra were

collected with 256 scans using a MCT detector.

4.2.6 Ion-Transport Studies with PHEMA Membranes

Diffusion-dialysis studies with PHEMA membranes were performed using two
glass half cells (Figure 2.2) as described in Chapter 2 (section 2.2). The permeate cell
contained deionized water (90 mL), and the feed cell contained 0.1 M salt solutions (90
mL) of KCl, K2S04, K3Fe(CN)6 and MgCl2. After dialysis with each salt, the entire

apparatus was rinsed well with deionized water and subsequently ﬁlled with water for 30

117

 

 

minutes to remove any adsorbed ions. Salts were examined in the same order as given
above, and after MgCl2 dialysis, a second KCl dialysis was performed to check the
integrity of the membrane. (The difference between KCl ﬂuxes in the ﬁrst and second
dialyses was <10%). Conductivity (Orion model 115) of the receiving side was recorded
at every 5 minutes for a period of 45 minutes, and conductivity values were converted to
concentration using a calibration curve of conductivity vs. concentration for each salt.
Fluxes were calculated from the slopes of concentration vs. time plots using equation 2.1.
Selectivity of monovalent over divalent ions was calculated from the ratio of respective
ﬂux values (equation 2.2).

For chemically cross-linked PHEMA membranes, I also examined dialysis with
solutions containing 1000 ppm Cl' and 1000 ppm SO42' or 500 ppm Cl' and 2500 ppm
8042' (solutions were prepared with KCl and K2SO4.). Dialysis was carried out for 90
min, and 2 mL samples were withdrawn from both sides at 10-min intervals. These
samples were analyzed with an ion chromatograph (Dionex 600) using an ASl4A anion
column and an 8 mM Na2CO3/1 mM NaHCO; eluent. Normalized ﬂuxes were calculated
from the respective slopes of normalized concentration vs time plots, and the anion-
selectivities were determined from the ratio of normalized ﬂuxes. Normalization was

performed by dividing by the source-phase concentration.

4.3 Results and Discussion

4.3.] Synthesis and Characterization of PHEMA Membranes
Initiator anchoring and polymerization were initially performed on gold—coated

wafers to facilitate ﬁlm characterization. First, 2-BPB was covalently attached to the

118

. ‘ 4
'.,
It. .
. 7v“

 

 

polyelectrolytes through amide linkages. As previously discussed (section 3.3.1), we did
not observe a prominent change in the reﬂectance FTIR spectrum of polyelectrolytes
after initiator anchoring, but there was a small absorbance increase in the amide region
(1650-1560), conﬁrming initiator attachment. Polymerization was carried out by
immersing the initiator-modiﬁed substrates in HEMA/water solutions containing
CuCl/CuBr2/bpy. Unlike gas-permeability studies, the HEMA polymerization was
carried out in water, rather than methanol, because aqueous polymerization is a more
controlled process that yields a‘relatively linear relationship between ﬁlm thickness and
polymerization time. Several previous studies showed that aqueous conditions also

9‘” The appearance of a strong carbonyl peak

accelerate ATRP of hydrophilic monomers.
at 1730 cm"1 in the reﬂectance FTIR spectrum of ﬁlms after polymerization indicated

formation of PHEMA on the surface.

4.3.2 Ion-Transport Studies with Composite PHEMA Membranes

Figure 4.1 shows results of diffusion dialysis through a composite membrane
containing a 28 nm-thick PHEMA skin. The linear relationship between receiving-phase
concentration and time shows that ﬂux is constant and conﬁrms that ion concentrations in
the receiving phase were negligible compared to those in the source phase. Table 4.1
gives the cumulative thicknesses of PHEMA ﬁhns, ion-transport selectivities and ion
ﬂuxes through PHEMA membranes. As the PHEMA ﬁlm thickness increases, selectivity
increases, presumably because the ﬁlm more completely covers the substrate.

Since PHEMA is neutral, the observed selectivities among ions are probably due

to differences in hydration energies and hydrated radii. Both hydration energy and

119

.v ‘ u I" 4
~,:'_.;;

 

 

Receiving Phase Concentration (mM)

 

 

 

 

 

 

 

 

2.5
O KCI
I KZSO4 .
' MgCl2
2.0- A K3E'-'e(CN)6 .
o
1.5- I
o
1.0- .
I
I I
I
0.5“ . . V V
I
' v A ‘ ‘
I . ‘
0.0?
0 10 20 30 40
Time (min)

Figure 4.1: Plot of receiving phase concentration as a function of time
in diffusion dialysis when the source phase (0.1 M salt) was separated
from the receiving phase (initially deionized water) by a porous
alumina substrate capped with 28 nm of PHEMA.

120

:_ no?
\
.' I'
.«‘-'—n_.'..

 

 

 

:00 505 00:88:00 00>» 80880000 :05 00 02.: -:U 00:00: 05 00 0800 05 003 :0: 8:00:00 .:0 00:00

.m0=:0>

02.: 00000 800 :0: 0:0 0:0:0808 0000 :8 029 3:300:00 .:0 000:0>0 05 00 00:23:00 .00::0> 03:: .:0 02:02:00: .:0 8:00:o
0808080008 00.5 :000: :0 .:0 0:000:00

0:00:00 800080: 80::m: 0:08 08:: .0> 000:: 083000: 8 85880800 .:0 000:0 05 808 0:080:00 0:03 0020:: 00:00.

0:05 8: 0.0 003 00:03 .80.: :0:0:::::\0:30::00:0300 05 0020:: 000005082. :00: :05 000: 3: 00::0> 00:03 0808080008
03: .:0 00000 05 800080: 0030:» 0:80? 0:00 :0 08:0 Emma: 00 080800820 3: 0088000 0:03 0000:0083. a

 

 

 

 

 

 

 

 

 

 

000mm: 0:. :0000 0:000 00000000 :00000000 00000000 0000.0 :0 00:
m :000 2000: 00m : 00.00:N0 :000000 :0000 :00 00 00:
:00 000 0000.0 0000.0 :00 000: 0000 00 00
£025: 00:20:00-0 0.000 .002 00:20:»: -000 :0 :50 :58: 25:
«0005:0000 85000839:
U3350000 00 : xAm~80\m0:080 A:02": 8:0:

 

 

 

 

 

 

0:00:00 08820 30:03 005808 0.0/Ema: 0:009:00 :8 000050000 0:0 0002.: 00000585 80...: 3.0 030.0

121

hydrated radii decrease in the order Mg2+ > SO42' > C1' > K+ (See Table 4.2).l2 Since
hydration energy and hydrated radius are higher for Mg2+ than SO42", one would expect to
observe higher selectivity values for KVMg2+ than for Cl'/SO42'. This was indeed the
case for all three PHEMA thicknesses. The Cl'/Fe(CN)63' selectivity is higher than
monovalent/divalent selectivities probably because of the large hydration radius and
hydration energy of Fe(CN)63'.

Although we were successful in obtaining reasonable selectivities with PHEMA in:
membranes, Cl' ﬂux values were relatively low for thicker ﬁlms. For 28 nm-thick i i

PHEMA ﬁlms, Cl- flux was 80% of that through bare alumina (Cl' ﬂux through bare

 

alumina was 5.2><10’8 moles/cmzs'), but ﬂux decreased dramatically with increasing 9
PHEMA thickness (Table 4.1). We thought that cross-linking of thin PHEMA
membranes might allow both high ﬂux and high selectivity. Cross-linking should reduce
swelling and may allow thin ﬁlms to fully cover a substrate.
4.3.3 Chemically Cross-linked PHEMA Membranes and their Ion-Transport
Properties

We cross-linked PHEMA by derivatizing ﬁlms with a di-acid chloride that would
react with hydroxyl groups in adjacent PHEMA chains. Scheme 4.1 outlines the reaction
of graﬁed PHEMA layer with succinyl chloride. Reﬂectance F TIR spectra of PHEMA
ﬁlms after derivatization showed the disappearance of the hydroxyl peak and a ~2-fold
increase in the intensity of ester carbonyl peak, suggesting quantitative conversion of
hydroxyl groups to esters (Spectrum b, Figure 4.2). However, the appearance of a small

shoulder at 1819 cm'1 suggested the presence of unreacted acid

122

 

Table 4.2: Hydrated radii and hydration energies of various ions.‘2

 

 

 

 

 

 

 

Ion Hydrated Hydration Energy
Radius (pm) (kJ/mol)
K+ 212 330
CI' 224 365
S047“ 278 h 1035
Mg2+ 299 1945
Fe(CN)63' 396

 

 

 

 

 

123

 

 

 

 

 

9 9H3 9H3 ° .
WWNH—C—Cl: CHz—C CKBF) + (”W
O=§
H n O
DMF, RT
it
(H) CH3 (EH3 CH3
WW‘NH--C—C|3 CHz—C CHz—C C|(Br)
H 0:< n O \ n-1
0 g0 30
O O O
—O
o 0
Cl OCHZCHZO—

Scheme 4.1: Cross-linking of PHEMA by reaction with succinyl
chloride.

124

__ s - ,:-_
A I

 

 

0.02 .

 

X 0.8 (c)

:

 

x 0.8 (b) '

M (a)

 

Absorbance
W‘r

 

l l l l l l

3500 3000 2500 2000 1500 1000

 

 

Wavenumbers (cm")

Figure 4.2: Reﬂectance FT IR spectra of a 28-nm thick grafted PHEMA
ﬁlm on a gold-coated wafer before (a) and after reaction with succinyl

chloride (b) and subsequent exposure to water (c).

125

chlorides, implying less than 100% cross-linking. Disappearance of the 1819 cm'l peak
upon exposure to water conﬁrmed that this peak is probablydue to unreacted acid
chloride (spectrum c, Figure 4.2). Nevertheless, the small size of the acid chloride peak
suggests that cross-linking is >50%.

Figure 4.3 shows results of diffusion dialysis through a composite membrane
containing a 35 nm-thick cross-linked PHEMA skin. (The PHEMA ﬁlm was ~28 nm-
thick prior to cross-linking, so these results can be compared to those in Figure 4.1.)
Table 4.3 summarizes the ﬂuxes and selectivities of chemically cross-linked PHEMA
membranes. For the 35 nm-thick ﬁlm (initially 28 nm of PHEMA), reaction with
succinyl chloride increased Cl’lSO42' selectivity by a factor of 100 and Cl’/Fe(CN)63'
selectivity by a factor of 90. The large increase in monovalent/multivalent anion
selectivities suggests that in addition to cross-linking, which reduces membrane swelling
and increases size-based selectivity, we also introduced ﬁxed negative charges into the
PHEMA ﬁlm. As reﬂectance FTIR spectra indicate, these negative charges probably
result from hydrolysis of unreacted acid chlorides. When the membrane is negatively
charged, Donnan potentials cause substantial exclusion of multiply charged anions such
as S042” and Fe(CN)63' and hence, higher monovalent! multivalent anion selectivities. In
contrast to anion transport, K+/Mg2+ selectivity increases only 4-fold after cross-linking.
Negative charges in the membrane should reduce monovalent/divalent cation selectivity,
and thus we see only a small increase in Ki'fMg2+ selectivity after cross-linking. The 4-
fold increase in selectivity is likely due to a reduction in ﬁlm swelling. Consistent with

decreased ﬁlm swelling, large increases in anion-transport selectivities with thin PHEMA

126

 

 

 

Receiving Phase Concentration (mM)

 

 

 

 

 

 

 

 

 

 

 

 

1.6
o
0.06 ~ . "
' o
0.04 r .
I
1.2 ~ 0.02 ~ . ' '
- A
0 ‘* i T ‘ f ‘ f ‘ T 0
0 10 20 30 40
o
0.8 -
0
O C KCI
MgClz
0.4 -
' . K2304
. A K3Fe(CN)6
v V V
o o T x x I X X X n n n
0 10 20 3O 40

Time (min)

Figure 4.3: Plot of receiving phase concentration as a function of
time in diffusion dialysis when the source phase (0.1 M salt) was
separated from the receiving phase (initially deionized water) by a
composite membrane of 35 nm cross-linked PHEMA ﬁlm. The inset
expands the concentration scale for SO42' and F e(CN)63 ’. PHEMA

was cross-linked by succinyl chloride.

127

 

 

- . .....i..1,..p.

000:? 08:: 00 8:00”v

.:0::000

0808:0000 0:: 8 000::0000 00 :0:: 08:: .0> 000:: 088000: :0 50880800 :0 00:80 0:: 80:: 00:83:00 0:03 00::0> 03:”?
00::000: 0:0 0000058581.: 0803800 800000 0808.898

8 0080000 00 0:0003 0:00 :0 08:0 080:0:00800 :0 00000538: 080800050 0080008 80:: 00:08:00 00 00000585. 0

 

 

 

 

 

 

 

 

 

128

 

 

 

 

 

NN how 00m , N: mod mod mm mm
N 00 mm 0: de 0.: 0m :N
.0025: 0220086 00020 :02 022000: -000 :0 :55
«00088:?
$838200 o0:x 00880208: 003:: 8:0:
0:88:00

::00-0:080 :83 80380 :0:0:.:::0 8 00:0:0808 <:>:m:::n: 0088:0080 :0: 0088:0200 0:0 0003:: 00000808: 8:”: "0.0 030,—.

ﬁlms are accompanied by a 40% decrease in C1' ﬂux (See Tables 4.1 and 4.2). However,
ﬂux is still about 50% of that through bare alumina.

The ion-transport studies described above were performed with source-phase
solutions that contained single salts. Actual ion separations occur from mixed salts,
however, so we brieﬂy examined ion-transport with both KCl and K2S04 in the source
phase. Table 4.4 contains preliminary selectivities and ﬂux values obtained with cross-
linked PHEMA membranes. The Cl'/SO42' selectivity obtained with solutions of mixed
salts is a factor of 1.6 to 4.5 higher (depending on the ratio of C1' to 8042') than the
selectivity obtained from single-salt experiments. The increase in selectivity with mixed
solutions is presumably due the fact that the diffusion potential is lower in mixed
KCI/K2S04 solutions than it is when only K2S04 is present. In K2SO4 solutions,
diffusion of K” is faster than diffusion of S042} so a diffusion potential develops to resist
current ﬂow. The difﬁision potential decreases K+ transport and increases SO42'
transport. When KCl is present, a smaller diffusion potential develops because Cl' has a
higher mobility than SO42" The no-current condition can be reached with a smaller
potential that increases Cl’ ﬂux much more than SO42' ﬂux.

4.3.4 Crown Ether-Derivatized PHEMA Membranes and their Ion-Transport
Properties

We hoped to improve the selectivity of PHEMA membranes by derivatization
with a complexing agent that can selectively bind a particular ion. Selective hopping of
ions between crown-ether binding sites could facilitate the transport of one ion over
another. Crown ethers are well known for both selective binding and fast release of alkali

metal

129

 

 

Table 4.4: Film thicknesses, normalized ﬂuxes and selectivities for cross-linked
PHEMA membranes in diffusion dialysis with mixed salt solutions.

 

 

 

 

Normalized F luxd X109 Selectivity°
Film
Thickness“l 2
(mm) or so4 ‘ cr/soi'
21b 450 4 106
3 5c 440 0.9 490

 

 

 

 

 

 

aThicknesses are estimated from measured ellipsometric thicknesses of corresponding
ﬁlms on gold wafers as described in experimental section. Thicknesses after
derivatization are reported.

bDiffusion dialysis was done using a 1000 ppm Cl' and 1000 ppm S0421

°Diffusion dialysis was done using 500 ppm Cl' and 2500 ppm S0421

dNormalized ﬂuxes are reported.

°Ratio of noramalized ﬂuxes.

130

v

 

 

cations.l3 We derivatized PHEMA with 2-hydroxymethyl-18-crown-6 because the size
of the crown ether cavity matches with the ionic radius of K+.M’15

In the derivatization of PHEMA, ﬁlms were ﬁrst reacted with CD1 for 12 hours
(Step 1, Scheme 4.2).9 After coupling of PHEMA with CD1, the reﬂectance FTIR
spectrum of the ﬁlm showed complete disappearance of the hydroxyl peak (3500 cm"-
33000m'1) and the appearance of a strong carbonyl peak at 1771 cm"1 due to the
imidazole carboxylic acid intermediate, suggesting quantitative reaction (Spectrum b,
Figure 4.4).9 Since imidazole groups can be displaced with nucleophiles such as amines
and alcohols, CDT-derivatized PHEMA is attractive for the immobilization of a variety of
functional groups. In this case imidazole-derivatized PHEMA ﬁlms were reacted with 2-
hydroxymethyl-l 8-crown-6 to introduce crown ether functionalities into the ﬁlm (Step 2,
Scheme 4.2). Since the reaction was very slow, the solution was heated to 70 °C for 4.5
days. Disappearance of the carbonyl peak at 1771 cm'1 and the appearance of peaks at
1265 cm'1 and 1155 cm'1 (C-O-C stretches of the crown ether) in the reﬂectance FTIR
spectrum of derivatized ﬁlms conﬁrmed the reaction with the crown ether (spectrum c,
Figure 4.3). Additionally, the ellipsometric thickness increased by 55 %, indicating
incorporation of crown ether moieties.

However, diffusion dialysis yielded a K+ ﬂux (2.3 x 10''0 molescm'zs'l) through
crown ether derivatized membranes that was smaller than Na+ ﬂux (2.9 x 10'10 molescm'
2s"). The lack of K‘i/Na+ selectivity could be due to crown ether moieties that are not in
the right conformation for K+ ion binding, or perhaps hopping between binding sites is
slower than diffusion. A better way of performing this derivatization would be to use a

K+-salt of the crown ether and subsequently remove the K+ ions aﬁer derivatization.

131

 

 

(H) CH3 / \
wwNH— C— C %Hz— Cl(Br) + (\N—C—N/W
N C”) \éN

OCHchZOH con

 

 

Step 1 DMF
RT, 12h

CH3
0... 'c' (30%05' +00.)

__ :<—o CHZCHZOC— 51::

/CH20H
Step 2 %

 

ﬂ

 

 

 

 

 

DMF, 70°C
0 CH3
WWNH— C— C—ECHOz—C :+CI(Br)
CHzCHZOCOCHz

Scheme 4.2: Functionalization of PHEMA with 2-hydroxy methyl 18-
crown-6.

132

 

 

0.02 N

 

 

Absorbance

 

. 0.8 (b) J

(a)
M

l 1 J l l l

3500 3000 2500 2000 1500 1000

 

 

 

 

 

Wavenumbers (cm")

Figure 4.4: Reﬂectance FTIR spectra of (a) a grafted PHEMA layer
(28 nm) (b) CDI-derivatized PHEMA and (c) CDI-derivatized
PHEMA after reaction with 2-hydroxymethyl-l 8-crown-6 (43 nm).

133

 

4.4 Conclusions

Surface-initiated ATRP is an attractive and versatile method of forming ultrathin
polymer skins on porous supports. Ion—permeability studies with composite PHEMA
membranes showed moderate selectivities for monovalent over divalent and trivalent
ions. However, reaction with succinyl chloride enhanced Cl'/SO42' and Cl'/ Fe(CN)63'
selectivities by 2 orders of magnitude, presumably due to the introduction of cross-
linking and negative charges into the membrane. Using mixed-salt solutions, even better

selectivities were obtained because of a reduction in diffusion potentials.

134

 

4.5

(1)
(2)
(3)
(4)
(5)

(6)
(7)

(8)
(9)

(10)

(11)

(12)
(13)

(14)

(15)

References

Xu, X.; Spencer, G. Desalination 1997, 113, 85-93.

Wang, X.-L.; Zhang, C.; Ouyang, P. J. Membr. Sci. 2002, 204, 271-281.
Bowen, W. R.; Mohammad, A. W.; Hilal, N. J. Membr. Sci. 1997, 126, 91-105.
Bowen, W. R.; Mukhtar, H. J. Membr. Sci. 1996, 112, 263-274.

Haraya, K.; Li, S.; Mizoguchi, K. Springer Series in Materials Science 1999, 35,
95-124.

Baddour, R. F .; Graves, D. J .; Vieth, W. R. J. Colloid Sci. 1965, 20, 1057-1069.

Jadwin, T. A.; Hoffman, A. S.; Vieth, W. R. J. Appl. Polym. Sci. 1970, 14, 1339-
1359.

 

Halden, R. A.; Lee, B. E. Br. Polym. J. 1972, 4, 491-501.

Huang, W.; Kim, J .-B.; Bruening, M. L.; Baker, G. L. Macromolecules 2002, 35,
1 175-1 179.

Wang, X. S.; Lascelles, S. F.; Jackson, R. A.; Armes, S. P. Chem. Commun.
1999, 1817-1818.

Robinson, K. L.; Khan, M. A.; de Banez, M. V.; Wang, X. S.; Annes, S. P.
Macromolecules 2001, 34, 3155-3158.

Marcus, Y. Biophys. Chem. 1994, 51, 111-127.

Noble, R. D. J. Chem. Soc. Faraday Trans. 1991, 8 7,
2089-2092.

Thunhorst, K. L.; Noble, R. D.; Bowman, C. N. J. Membr. Sci. 1999, 156, 293-
302.

Reusch, C. F.; Cussler, E. L. AIChE Journal 1973, 19, 736-741.

135

CHAPTER 5

Conclusions and Future Work

5.1 Conclusions

The work reported in this dissertation demonstrated the versatility of
multilayer polyelectrolyte deposition and surface-initiated atom transfer radical f
polymerization (ATRP) in the formation of defect-free, ultrathin membrane skins.
These skins are quite selective in both ion and gas separations. Chapter 2 described a

method of introducing ﬁxed negative charges into the bulk of multilayer

 

polyelectrolyte membranes (MPMS) to enhance anion-transport selectivities. This
method relies on complexation of Cu2+ by the —C00' groups of poly(acrylic acid)
(PAA) during the deposition of PAA/ poly(allylamine hydrochloride) (PAH) ﬁlms on
porous alumina supports. Subsequent removal of Cu2+ and deprotonation results in
ion-exchange sites in the bulk of the membrane skin. Diffusion dialysis studies with
Cu2+-templated PAA/PAH membranes showed a 4-fold increase in Cl'/SO42'
selectivity compared to pure PAA/PAH membranes deposited under similar
conditions. Post deposition cross-linking of these membranes further increased Cl'
/SO42' selectivity to values as high as 600. These remarkable selectivities are
presumably due to increased ﬁxed negative charge density in the bulk of the
membrane, which increases the Donnan potential to give greater exclusion of divalent
than monovalent anions. However, modeling of ion-transport data suggested that

selectivity is due to both Donnan exclusion and diffusivity differences among ions.

136

The second method I utilized in thin ﬁlm formation was room-temperature,
surface-initiated ATRP. Chapter 3 showed the versatility of this technique in the
synthesis of cross-linked polymer ﬁhns from a modiﬁed porous alumina support.
Cross-linked poly(ethylene glycol dimethacrylate) (PEGDMA) membranes exhibited
a COz/CH4 selectivity of 20. In comparison non cross-linked poly(2-hydroxyethyl
methacrylate) (PHEMA) grown form a modiﬁed porous alumina support showed
minimal gas-transport selectivities. However, derivatization of the hydroxyl groups

of PHEMA with ﬂuorinated acid chloride yielded moderate gas-transport selectivity.

 

":1 ' VA -._.—

Polymer growth was monitored by scanning electron microscopy (SEM) and
transmission FTIR spectroscopy. Both top-down and cross-sectional SEM images
showed that these polymer ﬁlms effectivelycover the surface pores of the alumina
support.

Chapter 4 discussed the promise of PHEMA membranes in ion separations.
PHEMA by itself showed moderate ion-transport selectivities. However, cross-
linking PHEMA via reaction with succinyl chloride increased Cl'/SO42' and Cl'
/Fe(CN)63' selectivity by IOO-fold. This large increase in anion-selectivities is
probably due to reduced ﬁlm swelling and introduction of negative charge by
hydrolysis of unreacted acid chlorides.

Overall, these studies demonstrate novel methods for the formation of
ultrathin membranes. Although new procedures for ﬁlm deposition may not be
practical on a large scale, they should allow development of membrane for speciﬁc
small-scale (i.e., analytical) separations. Work with both multilayer polyelectrolyte

and grafted polymer ﬁlms shows that the minimal thickness of these systems allows

137

high ﬂux. Additionally, the wide variety of functional groups that can be included in

these membranes allows tailoring of transport properties.

5.2 Suggestions for Future Work

My success in growth of a cross-linked ﬁlm from a porous substrate should
now allow investigation of new types of membrane systems such as imprinted
polymers. Cross-linked polymers are attractive materials for separations because of
their low free volume (which will reduce ﬁlm swelling) and ability to withstand
drastic separation conditions. An interesting area of research will be the examination
of PEGDMA membranes in ion separations. However, ion-transport through
PEGDMA itself is quite low, and thus, incorporation of non cross-linkable monomers
into PEGDMA ﬁlms will likely be necessary to achieve a desirable ﬂux. Introduction
of charged functionalities into these ﬁlms could be achieved by co-polyrnerizing
ethylene glycol dimethacrylate (EGDMA) with t-butyl methacrylate and subsequently
hydrolyzing t-butyl groups to —COOH groups. Deprotonation of these —COOH
groups will yield ion-exchange sites in the membrane. Such membranes have the
potential to be extremely selective in anion separations.

In chapter 4, I showed the potential of derivatized poly(2-hydroxyethyl
methacrylate) (PHEMA) ﬁlms in anion separations. Similarly PHEMA could be
derivatized for cation separation by reaction with a di-arnine (e. g. ethylene diamine).
This d erivatization c ould b e e asily p erformed b y ﬁrst coupling P HEMA with 1 ,1'-
carbonyldiimidazole (CD1) and then reacting CDI-derivatized PHEMA with the di-

amine.‘ Similar to reaction of PHEMA with succinyl chloride, derivatization with a

138

 

diamine should result in cross—linking along with residual free amine groups that can
give ﬁxed positive charge to the membrane. It will also be interesting to test PHEMA
and its derivatives in neutral-molecule separations. This will provide valuable
information in understanding the separation mechanism in these systems. As
previous studies demonstrated2 that thick PHEMA membranes can be used in
nanoﬁltration, it will deﬁnitely be interesting to test ultrathin PHEMA and cross-
linked PHEMA (HEMA copolymerized with EGDMA) ﬁlms in nanoﬁltration.

Another very interesting and challenging area of research will be the synthesis
of ultrathin, imprinted polymer membranes. In imprinting, a template, a monomer
and a cross-linker are polymerized together, and the subsequent release of the
template results in a polymer that contains cavities that are selective for the template.
3‘5 Imprinted polymers could be prepared by copolymerizing EGDMA (cross-linker)
with HEMA that was hydrogen bonded with a template molecule. For templates, we
could use any molecule of interest (possibly a high molecular weight species) that can
interact with HEMA through hydrogen bonding. Removal of the template should
leave behind recognition sites in the membrane through which transport could take
place. These imprinted membranes may be useful in selective separation of the
template molecule.

In summary, the development of new techniques for forming ultrathin
membrane skins has the potential to yield a wide variety of separation membranes

ranging from imprinted to ion-exchange systems.

139

I; 1'1 -'.v.’.::i'ﬂT.a‘ “km

5.3

(1)

(2)

(3)
(4)
(5)

References

Huang, W.; Kim, J.-B.; Bruening, M. L.; Baker, G. L. Macromolecules 2002,
35, 1175-1179.

Jadwin, T. A.; Hoffman, A. S.; Vieth, W. R. J. Appl. Polym. Sci. 1970, 14,
1339-1359.

Shea, K. J. Trends Polym. Sci. 1994, 2, 166-173.

Wulff, G. Angew. Chem. Int. Ed. 1995, 34, 1812-1832.

Kriz, D.; Rarnstrom, 0.; Mosbach, K. Anal. Chem. 1997, 69, 345A-349A.

140

 

Wthvruiuihﬂm.