:3" '-

‘1 '7“

..v‘:.'_'.
my .. ‘

L"
.45

’5;

 

goon

5&097i*

.*.

This is to certify that the
dissertation entitled

ULTRATHIN, SELECTIVE POLYIMIDE MEMBRANES
PREPARED FROM LAYERED POLYELECTROLYTES

presented by

DANIEL M. SULLIVAN

has been accepted towards fulfillment
of the requirements for the

 

 

 

 

Doctoral degree in Chemistry
7 o
major Proféssor’s Siggajui‘e
Ma \ / / I A 02 O O k/
T T T
Date

MSU is an Affirmative Action/Equal Opportunity Institution

 

LIBRARY I

Michigan State
University

 

 

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

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

 

Ultrathin, Selective Polyimide Membranes Prepared from Layered
Polyelectrolytes

BY

Daniel M. Sullivan

A Dissertation
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Chemistry

2004

ABSTRACT

ULTRATHIN, SELECTIVE POLYIMIDE MEMBRANES PREPARED FROM
LAYERED POLYELECTROLYTES

By

Daniel M. Sullivan

The work presented in this dissertation applies layer-by-layer techniques
and subsequent heat-induced imidization to form ultrathin polyimide films. Two
layer-by-layer methods are discussed, one that employs polymers that covalently
bond to each other, and another that uses electrostatic interactions for film
buildup. The polyimide films presented here were tailored for two areas of
application; anti-corrosion coatings and high flux, selective membranes.

Ultrathin, passivating films are attractive for protecting metal surfaces
without completely masking substrate properties. Layer-by-layer covalent
deposition of GantrezTM/poly(allylamine) films yields ultrathin amic acid-linked
layers that can be imidized by heating. lmpedances of Al electrodes coated with
Gantrezm/poly(allylamine) films depend on the number of deposited bilayers,
and the imidized films increase aluminum oxide resistance by up to two orders of
magnitude. These results suggest that increases in film resistance play a role in
oxide passivation in this system.

Low fluxes often restrict practical applications of membrane-based
separations. A reduction in the thickness of the membrane increases flux, but
currently, thicknesses below 50 nm are difficult to achieve. Alternating

polyelectrolyte deposition (APD) offers a simple and convenient method to

fabricate ultrathin membrane skins, but the selectivity of polyelectrolyte films is
generally low.

To develop discriminating, ultrathin polyelectrolyte films, we first adsorb
poly(amic acid) salts and poly(allylamine hydrochloride) (PAH) on porous
alumina. Subsequent heating yields highly selective polyimide membranes.
Remarkably, after imidization at temperatures ranging between 150 and 180 °C,
membranes composed of poly(p-phenylenepyromellitamic acid) and PAH have
Cl'lSO42' selectivities as high as 1000 and K"/Mg2+ selectivities of 100-300. The
minimal thickness (4 to 9 nm) of these membranes allows fluxes that are 50% of
those through the bare porous alumina support.

For gas separations, we utilized fluorinated polyimides because these
materials offer both high selectivity and high permeability. Selectivities (Oz/N2 up
to 6.9 and Cog/CH4 up to 68) and perrneabilities of three different, fully imidized,
poly(amic acid)/PAH membranes are comparable to literature values for the
corresponding bulk polyimides, even when films are 35 to 40 nm thick.

We also investigated pervaporation separations of water from alcohol
solutions using cross-linked polyimide membranes. The poly(amic acids)
employed in these studies contain diaminobenzoic acid groups that can form
amide cross-links with the polycation when heated. Water-selective
pervaporation separations using cross-linked polyimide membranes are highly
selective (water/alcohol selectivities are 1100 and 6100 for solutions containing
10% and 90% isopropanol, respectively) and the fluxes are also high, ranging

from 2 to 11 kgm‘Z-h".

To my wife, Sarah

ACKNOWLEDGEMENTS

I would like to first thank my advisor Dr. Merlin Bruening for giving me the
opportunity to work in his group. Over the years, he taught me how to be
thorough in both research and writing. He is not only a good advisor, but a good
friend (and yes, a better golfer).

I also need to pay gratitude to my fellow group members, both past and
present. They were helpful and supporting at each stage of my graduate career.
I also enjoyed the many stimulating conversations and the cultural experiences
they were able to provide. Along with my group members, I also would like to
thank the many friends that I made in the Chemistry Department. The most
notable people include Jeremy, who taught me the art of sensitivity (sarcasm at
the fullest) and both Jason and Carl, with who I spent many hours watching bad
movies (intentionally) and brewing good beer (well, most were good).

My family was also very helpful during my many, many years in college. I
would especially like to thank my wife, Sarah, for being very supportive of the
long hours and the stressful times. I don’t think I could have done this without
her.

Lastly, thanks for reading this, you are probably the only one who ever has

or will.

TABLE OF CONTENTS

List of Tables
List of Figures
Chapter 1 Introduction and Background

1.1 Thin polymer films

1.2 Polyimides

1.3 Ultrathin polyimide anti-corrosion coatings

1.4 Multilayer polyelectrolyte films

1.5 Ion-selective polyimide membranes

1.6 Ultrathin polyimide membranes for gas separations

1.7 Pervaporation separations using cross—linkable polyimides
1.8 References

Chapter 2 Ultrathin, Layered Polyimide Coating on Aluminum

2.1 Introduction
2.2 Experimental
2.2.1 Materials
2.2.2 GantrezTM/PAAm Film Synthesis
2.2.3 Film Characterization
2.2.4 Electrochemical Studies
2.3 Results and Discussion
2.3.1 Synthesis of GantrezTM/PAAm Films
2.3.2 Electrochemical Impedance Spectroscopy
2.3.3 Protection of Al by GantrezTM/PAAm Coatings
2.4 Conclusions
2.5 References

Chapter 3 Ultrathin, Ion-Selective Polyimide Membranes
Prepared from Layered Polyelectrolytes

3.1 Introduction

3.2 Experimental
3.2.1 Chemicals and Materials
3.2.2 Substrate Preparation
3.2.3 Synthesis of PMDA-PDA
3.2.4 Film Characterization
3.2.5 Membrane Preparation
3.2.6 Membrane Transport

3.3 Results and Discussion

vi

29

29
3O
30
32
33
33

34
36
38
42
43

46

46
49
49
49
50
50
50
51
53

3.3.1 Deposition of PMDA-PDA/PAH Films on Aland Au
3.3.2 Coverage of porous alumina
3.3.3 PSS/PAH base layer
3.3.4 Unheated PMDA-PDA/PAH membranes
3.3.5 Heated membranes
3.4 Conclusion
3.5 References

Chapter 4: Ultrathin, Gas-selective Polyimide Membranes
Prepared from Multilayer Polyelectrolyte Films

4.1 Introduction
4.2 Experimental
4.2.1 Materials
4.2.2 Substrate Preparation
4.2.3 Polymer Synthesis
4.2.4 Solution Preparation and Film Deposition
4.2.5 Film lmidization
4.2.6 Film Characterization
4.2.7 Gas-Transport Studies
4.3 Results and Discussion
4.3.1 Film Formation and Composition
4.3.2 Film lmidization
4.3.3 FESEM Studies of Membrane Formation
4.3.4 Gas-Transport Measurements
4.3.5 Gas Transport as a Function of Film Composition and the
Degree of lmidization
4.3.6 Gas Transport as a Function of Film Thickness
4.4 Conclusion
4.5 References

Chapter 5: Ultrathin, Cross-linked Polyimide Pervaporation
Membranes Prepared from Polyelectrolyte Multilayers

5.1 Introduction
5.2 Experimental
5.2.1 Materials
5.2.2 Substrate Preparation
5.2.3 Polymer Synthesis
5.2.4 Solution Preparation and Film Deposition
5.2.5 Film lmidization
5.2.6 Film Characterization
5.2.7 Pervaporation Measurements
5.3 Results and Discussion
5.3.1 Film Deposition and Characterization
5.3.2 Pervaporation Experiments with non Cross-linkable

vii

53
60
60
62
66
71
72

74

74
79
79
79
80
80
81
81
82
83
83
85
88
88

92
94
95
96

99

99
104
105
105
105
106
107
107
107
111
111
115

Membranes
5.3.3 Cross-linkable Membranes

5.3.4 Comparison with related membrane systems

5.4 Conclusion
5.5 References

viii

118
122
123
124

Table 2.1

Table 3.1

Table 3.2

Table 3.3

Table 4.1

Table 4.2

Table 5.1

Table 5.2

Table 5.3

LIST OF TABLES

Coating thicknesses and equivalent circuit parameters for Al
electrodes coated with various layers of GantrezTM
/poly(allylamine) films.

Fluxes and anion selectivity coefficients for porous alumina
coated with a 5-bilayer PSS/PAH base capped with 1.5 to
3.5 bilayers of PMDA-PDA/PAH.

Cation selectivity coefficients for porous alumina coated with
a 5-bilayer PSS/PAH base capped with 1.5 to 3.5 bilayers of
PMDA-PDA/PAH.

Ionic and hydrated radii, and hydration energies of the ions
used in the dialysis transport experiments.

Permeability (barrers) and selectivity coefficients for the
transport of various gases through several polyimide
membranes containing ~10% PAH.

Gas permeability and selectivity values for 6FDA-
mPDA/PAH membranes prepared with different heating
temperatures and thicknesses.

Bilayer thicknesses (nm) and Water Contact Angles for
Unheated and Heated (250 °C) Polyelectrolyte Films
Deposited on Al-coated Si.

Permeate Water Concentrations, Selectivity Coefficients
(listed in parenthesis), and Fluxes (listed in italics, in kg-
m'Z-h") for Pervaporation through several Poly(amic
acid)/PAH Membranes at 50 c-’C.

Permeate Water Concentrations, Selectivity Coefficients
(listed in parenthesis), and Fluxes (listed in italics, in kg-
m'z-h") for Pervaporation through several Poly(amic
acid)/PDADMAC and Poly(amic acid)/LPEI Membranes at
50 9C.

40

65

67

90

93

110

116

117

Figure 1.1

Figure 1.2

Figure 1.3

Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4

Figure 3.1

Figure 3.2

LIST OF FIGURES

Polyimide synthesis via a two step polycondensation
reaction. An equimolar mixture of dianhydride and diamine
reacts to form a poly(amic acid) that is then converted to the
polyimide by removal of water.

Schematic diagram of multilayer film formation using
alternating polyelectrolyte deposition.

Heat-induced imidization of multilayer poly(amic acid)
/poly(al|ylamine) films.

Synthesis of a GantrezTM/poly(allylamine) bilayer on Al and
subsequent conversion to an imide-containing film.

Reflectance FT IR spectra of GantrezTM/PAAm films. Figure
A shows spectra after the deposition of GantrezTM on Al (Gz
1) and on one (Gz 2) and two (Gz 3) GantrezTM/PAAm
bilayers. Figure B shows the spectra after deposition of
PAAm (1, 2, and 3 GantrezTM/PAAm bilayers) as well as the
spectrum of a 3-bilayer film after heating. The spectrum of
the heated film is multiplied by 0.5 for clarity.

Equivalent circuits for impedance data. Circuit A was used
to fit "bare" Al and Al coated with GantrezTM/PAAm films.
Circuit B would be used if the organic film were highly
resistive. The physical meaning of the different symbols is
Rs, solution resistance, Flax, oxide resistance, Rf, film
resistance, Cf, film capacitance, and Cox, oxide capacitance.

Impedance plots for bare Al (circles, both 2' and Z' ' were
multiplied by 5 for clarity) and Al coated with 3 bilayers of
GantrezTM/PAAm before (squares) and after heating
(triangles). All impedance data were measured in 0.5 M
NaCI at pH 3 after a 4 h immersion time.

(a) Membrane consisting of a porous alumina support, a
PSS/PAH base layer, and a PMDA-PDA/PAH selective
capping layer (b) PMDA-PDA structure and heat induced
imidization and (c) PSS and PAH structures.

Dialysis setup used to measure ion-permeability of
polyelectrolyte membranes: magnetic stir bar (a), O-ring
and membrane (b), mechanical stir prop, motor not shown
(c), and conductivity probe (d).

14

31

35

37

39

48

52

Figure 3.3

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Figure 4.1

Figure 4.2

Figure 4.3

Reflectance FTIR spectra of 1, 3, 5, 7, and 9 bilayers of
PMDA-PDA/PAH on Al-coated Si wafters.

Ellipsometric thickness and absorbance at 1580 cm'1 of
PMDA-PDA/PAH films on Al-coated Si wafers as a function
of the number of bilayers in the film.

Reflectance FTIR spectra of 2.5 bilayers of PMDA-
PDA/PAH, 4.5 bilayers of PSS/PAH, and the composite film
(same number of bilayers for each) deposited on gold-
coated substrates modified with MPA.

Reflectance FTIR spectra of films containing a PAH/PSS
base of 4.5 bilayers and a PMDA-PDA/PAH topcoat of 2.5
bilayers before and after heating for 2 h at several
temperatures. Films were deposited on gold-coated
substrates modified with MPA.

Top-down (a) and cross-sectional (b) FESEM images of
porous alumina coated with 5 bilayers of PSS/PAH and
capped with 2.5 bilayers of PMDA-PDA/PAH.

Plots of normalized receiving phase conductivity vs. time for
diffusion dialysis experiments in which a heated 2.5-bilayer
PMDA-PDA/PAH + 5-bilayer PSS/PAH membrane
separates a 0.1 M source phase from the receiving phase.
The inset shows MgClz and K2804 plots magnified by 2
orders of magnitude.

Plots of the diffusion dialysis flux of KCI (top) and K2804
(bottom) as a function of the imidization temperature of
composite membranes consisting of 5 bilayers of PSS/PAH
and a capping layer of 1.5 (circles), 2.5 (squares), and 3.5
(triangles) bilayers of PMDA-PDA.

Heat-induced imidization of a poly(amic acid)/PAH film on a
porous support. Neutralization of PAH occurs when it
contributes a proton for the formation of water. Intertwining
of neighboring layers is not shown for figure clarity.

Structures of the polyimides used in this study.
Brewster-angle transmission FTIR Spectra of 9.5-bilayer

6FDA-mPDA/PAH films on Si before and after heating at
150 °C or 250 °C for 2 h to induce imidization.

xi

55

57

58

59

61

68

70

77

78

86

Figure 4.4

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

Figure 5.5

Figure 5.6

Cross-sectional field-emission scanning electron microscopy
images of (a) a bare alumina support and (b) an alumina
support coated with an imidized 9.5-bilayer 6FDA-
mPDA/PAH film.

Structures of the poly(amic acids) and other polyelectrolytes
used in this study.

Heat-induced imidization and cross-linking of BPDA-
DABA/PAH films.

Schematic diagrams of (a) the pervaporation system and (b)
a cross-sectional view of the membrane cell.

Cross-sectional FESEM image of an alumina support coated
with an imidized 12.5-bilayer BPDA-DABA/PAH film.

Reflectance FTIR spectra of various poly(amic acid)/PAH
films deposited on Al-coated Si. The spectra of heated films
were normalized to the imide peak of BPDA-DABA at 1725
cm“, while spectra of heated and unheated BPDA-
DABA/PAH films are plotted on a common scale.

Reflectance FTIR spectra of various BPDA-DABA/polycation
films deposited on AI-coated Si. The spectra of heated films
were normalized to the imide peak of BPDA-DABA/PAH at
1725 cm“.

xii

89

102

103

108

113

114

120

Chapter 1

Introduction and Background

1.1 Thin polymer films
Thin polymer films play an increasingly important role in a range of

technological areas, including protective coatings,“3 adhesives,4 lithography,5'6

1 12-14

organic light-emitting diodes,7'8 sensorsf”1 membranes, and optical
coatings.‘5'16 Properties such as permeability, refractive index, wetting, and film
functionality are easy to control by selecting the coating material, and polymer
films are, in general, inexpensive to produce. In several ways, thin film
technologies have become an indispensable part of everyday life. As an
example, thin coatings of photoresist polymers are essential in fabrication of
microelectronic devices. Polymer films are spin-coated onto silicon substrates,
and after treatment with UV light and chemical etching, these films give rise to
the three-dimensional structures that make up microchips.17

The most common techniques for forming thin polymer films include
spraying and dip-, flow-, or spin-coating. These methods allow for good control
over film thickness down to 100 nm, but some applications would benefit from
much thinner polymer films. Deposition of “ultrathin” films (thicknesses <100 nm)
requires different coating strategies that allow regulation of film thickness at the
macromolecular level. Control over both thickness and the properties of polymer

films is an ongoing challenge; therefore, new methods for depositing polymer

films are constantly being sought.

Various methods have been developed to better control polymer film
thickness. The Langmuir-Blodgett (LB) technique was the first method used to
deposit single layers of organic molecules on solid surfaces.18 Although this
technique was originally employed for deposition of small molecules, it has been
further applied to amphiphilic, polymeric materials.‘9'20 Film deposition occurs by
dipping a substrate perpendicularly though a film suspended at a liquid/air
interface. Multiple layers can be applied through repeated dipping cycles, and
this coating method affords good control over the film thickness since
monolayers are deposited during each dipping step. Unfortunately, the LB
method has many limitations that prevent it from becoming a practical film
deposition technique. The biggest problem is the need for complicated
deposition instrumentation. By its nature, the deposition procedure restricts
substrate dimensions to those of the Langmuir trough. Additionally, polymers
used to form LB films must be amphiphilic, and this requirement greatly reduces
the variety of films that can be produced. Finally, since van der Waals forces
hold LB films together, coatings are not very stable.

More recently, other layer-by-layer methods have been developed for
controlled deposition of ultrathin films, and some of these techniques are
essentially free of the complications that arise in LB deposition.‘°"'24 Such layer-
by-layer techniques typically involve cycles of dipping a substrate into a polymer
solution, rinsing with an appropriate solvent, and then dipping once again into
another polymer solution. Each layer is typically bound to the previous layer by

either covalent or ionic interactions, making these films much more stable than

those prepared by LB. Additionally, many types of polymers can be utilized in
Iayer-by-layer techniques, so the chemistry of films can be easily adjusted to suit
specific applications.

The work presented in this dissertation applies layer-by-Iayer techniques
and subsequent heat-induced imidization to form ultrathin polyimide films. The
ultrathin polyimide films used as the anti-corrosion coatings described in Chapter
2 were fabricated using polymers that form covalent bonds between layers.
Films as thin as 30 nm increase the resistance of aluminum oxide by 2 orders of
magnitude. Chapters 3 — 5 discuss a different method for polyimide film
fabrication, alternating polyelectrolyte deposition. In this work, ultrathin,
polyimide membranes were prepared for the selective separation of ions, gases,
and liquids. The highly selective ion-separation membranes (Cl'/SO42' selectivity
of 1000) described in Chapter 3 contained selective polyimide skins (4-9 nm
thick) deposited on a highly permeable gutter layer. Chapter 4 presents the
fabrication of ultrathin gas-separation membranes composed of fluorinated
polyimides. Even though these polyimide films are as thin as 35 nm, they
provide selectivities and perrneabilities that are identical to the bulk, cast
materials. Using cross-linkable materials, the pervaporation membranes
described in Chapter 5 provide selectivities and fluxes as high as 6100 and 2
kg-m'zoh", respectively, for the removal of water from 90% isopropanol solutions.
A common thread through all of these chapters is the exploitation of the layer-by-
layer process to achieve high-performance, ultrathin polyimide coatings. In all of

these areas, the thickness and chemistry of the films are easily controlled. The

remainder of this introduction provides important background concerning
polyimides, corrosion, multilayer polyelectrolyte films, ion separations, gas

separations, and pervaporation.

1.2 Polyimides

Polyimides are an attractive family of polymers because of their high
thermal stability, resistance to chemical degradation, mechanical strength, and
low dielectric constant.25 Due to these properties, these materials have industrial

26,27

applications in electronics (insulators and photoresists), optical

28,29

waveguides, composites for aerospace applicationsf’o'31 high temperature

3233 and membrane-based separations.34

adhesives,

The majority of polyimides are infusible and insoluble. Thus, the synthesis
of these polymers takes place in a two step process, where the first step involves
the formation of a soluble poly(amic acid), and the second step is imide formation
(Figure 1.1).25 Poly(amic acids) are usually formed by a simple condensation
polymerization of dianhydrides and diamines, which are often aromatic. The
reaction is performed in a dipolar, aprotic solvent and proceeds by a nucleophilic
attack of the amino group on the carbonyl carbon of the anhydride.

Cyclization to the imide occurs by the reaction of the carboxylic acid and
the amide with the release of water as a by product (Figure 1.1). This reaction is
induced by either thermal or chemical means, but the thermal method is more

common. In the thermal reaction, heating poly(amic acids) at 250 to 350 °C will

effect full imidization. This method is usually employed with “solid state”

.2283 Lo _m>oE2 3 2363.3 on.
2 noto>coo 55 m_ 65 38 oEman m E22 2 20mm. oEEmfi new 20:95:86 Lo 9338

5.0653 c< .cozowm. cozmmcoucoozoa amam oz: m «3 $8593 mEEEon. u: 0.59”.

23::an
o
f... / z \ _
In /
O

\

 

/

 

 

 

 

 

AEom 2:528

poly(amic acids), but the imidization proceeds faster in the presence of residual
dipolar, amide (e.g., N,N-dimethylfonnamide) solvent.25 Chemical imidization is
achieved in solution with an excess of a dehydrating agent such as acetic
anhydride and is catalyzed with an amine base. Both chemical and thermal

imidization techniques have been applied to thin films of polyamic acids.35'38

1.3 Ultrathin polyimide anti-corrosion coatings

The United States alone spends an average of $278 billion dollars per
year to combat corrosion.39 The majority of corrosion countermeasures employ
either a thick organic coating (paint) or galvanization to protect steel.
Nevertheless, there are some specialized areas in which an ultrathin, passivating
film is attractive for protecting surfaces without completely masking substrate
properties. Examples of these include passivation of heat exchangers,“41
electronic materials,” and high Tc superconductors.43

In Chapter 2, I show that layer-by-layer covalent deposition of
GantrezTM/poly(allylamine) films on aluminum yields ultrathin amic acid-linked
layers that can be imidized by heating to form cross-linked polyimide coatings
that protect aluminum from Cl'-induced corrosion.44 Although the electrical
resistance of these films in water is relatively small because of their minimal
thickness, they do form a passivating layer on the surface oxide and increase
oxide resistance by 2 orders of magnitude. lmpedances of AI electrodes coated
with GantrezTM/poly(allylamine) films depend on the number of deposited bilayers

and increase by an order of magnitude after imidization. These results suggest

that small increases in film resistance play a role in oxide passivation by this

system.

1.4 Multilayer polyelectrolyte films

In 1966, ller published a paper entitled “Multilayers of Colloidal Particles”
which described the use of oppositely charged particles such as alumina and
silica to build up multilayered films.45 He deposited these films on glass in a
layer-by-layer fashion and estimated their thickness by the color of the
interference pattern they produced. It wasn’t until twenty-five years later that
Decher and Hong reported the use of bipolar amphiphiles and polyelectrolytes to
deposit multilayered films on charged surfaces.”47 They showed that ultrathin
polymer films could be formed by a simple dip and rinse methodology. Their
work has since triggered hundreds of publications in many different fields of
research.

Layer-by-layer (LBL) assembly of polyelectrolytes is a simple method to
build up ultrathin polymer films. Film formation begins by the adsorption of a
polyelectrolyte from solution onto an oppositely charged substrate. Useful

substrates include essentially all materials that support sufficient surface charge;

21 ,48,49 51,52

commonly used supports are glass, quartz,50 silicon wafers, mica,53

4454’“ and some polymers.”59 The first adsorbed polyelectrolyte

metal coatings,
layer over-compensates the charge of the substrate, leaving an excess charge
for subsequent adsorption of a polyelectrolyte of opposite sign (Figure 1.2).

Rinsing of the substrate with water and a second immersion in the oppositely

charged surface

i+i+i+i+i+i+i+i+i

1. Immerse in

polycation @
solution + + + +
2. Rinse

\

++++++
4Ffifl§2fl§l3flfl§2flms
+++++++++++++++++

(writer: struts

+ +
+ c +
+ +
+
+, ,‘u‘
+1
+r

'r
+»
+

+

1. Immerse in
polyanion
solution

- 2. Rinse

fifilleEMllwfiZei

itititititititiri

repeat steps to add
multiple bilayers

Figure 1.2: Schematic diagram of multilayer film formation using alternating

polyelectrolyte deposition.

charged polyelectrolyte results in one bilayer on the substrate, and deposition of
additional layers occurs similarly. The driving force for polyelectrolyte adsorption

6° When a single ionic

is not a change in enthalpy, but an increase in entropy.
bond is formed between two polyelectrolytes, two counter ions are displaced into
the solution. Considering that each polyelectrolyte chain contains many charged
groups that can undergo ionic cross-linking with an oppositely charged strand, it
can easily be seen that a large increase in entropy occurs upon adsorption due
to the increased freedom of many counterions.

Film thickness can be controlled over a wide range by changing either the
number of bilayers in the film or by varying the deposition solution’s ionic

371‘” Increasing the ionic strength usually

strength, pH, or solvent composition.
increases the layer thickness because the electrolyte ions screen charges on the
polyelectrolytes and allow formation of coiled, rather than extended chains. The
presence of many loops and tails in the film results in thicker layers. Deposition
pH also has a pronounced effect on the thickness of films containing weak-acid
or weak-base polyelectrolytes because protonation or deprotonation of the
polyelectrolytes influences their charge density and conformation.

The layer-by-layer method is not limited to simple polyelectrolytes, and

many types of charged materials such as dendrimers,62'E55 biomolecules,52'5“"52166

68,69 58,70

dyes,67 metal colloids, clays, and other inorganic particles71 '72 have been
included in multilayer films. Existing polyelectrolytes can also be derivatized to
incorporate desired functional groups.“74 This vast range of available charged

materials allows tailoring of ultrathin, multilayer films for many possible

applications. Several recent papers describe attempts to apply LBL films as
conducting layers,75 active enzyme-containing films,76 electrocatalysts,77

8

electrochromic films,”7 sensors,79 light-emitting thin films,80 patterned

surfaces,81 anti-corrosion coatings,44 and perrnselective membranes.59'82'83

1.5 Ion-selective polyimide membranes
Many industrial applications require the removal of ions from solution or
selective extraction of specific ions. The removal of salt from water, for example,

84,85

is important in applications such as food processing and the production of

potable water.“88 Salt purification methods, such as the production of table salt
from sea water89 and the removal of heavy metals from electroplating waters,”91
are also of great importance. Ion-separation methods that employ membranes
include reverse osmosis, nanofiltration, and diffusion dialysis, but electrodialysis
is the most widely used technique.14 As an example, caustic soda is produced in
the USA at a rate of 14 million tons per year, almost entirely by the electrodialysis
of brine.92 In this process chlorine is produced at the anode and NaOH at the
cathode in stoichiometric quantities, and the electrodes are separated by ion-
selective membranes.14

High fluxes are vital to any membrane application so that both the size of
the membrane and energy costs can be kept to a minimum. For a particular
system, flux is typically maximized by making the active separation layer of the

membrane as thin as possible, as flux is usually inversely proportional to

thickness.13 However, because the separation layer is very thin, it must be

10

reinforced by a highly perrneable support that provides mechanical strength.
Fabrication of such a membrane is generally achieved by either casting an
asymmetric membrane”94 to create a dense surface layer on an otherwise
porous polymer or by forming a composite membrane through deposition of a

‘3'95 Unfortunately, the formation

thin polymer coating on a porous support.
defect-free selective skins less than 100 nm is often difficult.

The layer-by-layer deposition method provides a new and versatile
technique for fabricating the ultrathin skins of composite membranes. Because
such films form by adsorption, surface roughness does not greatly affect
adhesion, and defect-free films can be obtained on porous supports since each
added layer covers defects in the previous layer. Polyelectrolyte films are also
well suited for the separation of ions in solution since the charged nature of these
films allows for Donnan (electrostatic) exclusion of ions with the same charge as
the outer layer of the membrane.

Krasemann and Tieke were the first to show that multilayer polyelectrolyte
films could be used to make ion-selective membranes.59 They used some of the
more common polyelectrolyte combinations to form films on
polyacrylonitrile/polyethylene terephthalate (PAN/PET) supports. They reported
that the addition of supporting salt to the polyelectrolyte deposition solutions
greatly enhances ion-transport selectivity. Membranes composed of 60 bilayers
of poly(allylamine hydrochloride)/poly(styrene sulfonate) (PAH/PSS) provided

selectivities of 45 for CI'ISO42' and 113 for K"/Mgz*. In another publication,

Toutianoush and Tieke showed that some multiply charged ions permanently

11

adsorb into poly(vinyl amine)/po|y(vinyl sulfate) (PVA/PVS) membranes,
changing the flux of monovalent ions.96 They later used PVA/PVS membranes to
remove salts from water in reverse osmosis.97

Much of the ion separation work using multilayer polyelectrolyte films has
come out of the Bruening group. Early studies showed that as few as 4 bilayers
of PSS/PAH will cover the pores of porous alumina supports.98 These composite
membranes had CI'ISO42° selectivities of only ~5, but the fluxes of KCI through
such films were nearly identical to those through the bare alumina support
(diffusion dialysis of single-salt solutions). Thus, even though the ultrathin
PSS/PAH films were not very selective, they provided a convenient way to cover
underlying pores without hindering flux. Hybrid membranes can be fabricated
using PSS/PAH as a “gutter layer" and capping this layer with a selective skin
composed of a few bilayers of more selective polyelectrolytes. Since the
selective polyelectrolyte skin is extremely thin, the flux of monovalent species will
be high. Stair et. al. reported that membranes capped with 2.5 bilayers of cross-
linked poly(acrylic acid) (PAA)/PAH exhibit selectivities as high as 360 for CI‘
ISO42' and allow fluxes of KCI that are only ~60% less than those through the
bare alumina support.83 When applied to nanofiltration, membranes containing a
PAA capping layer allow water fluxes that are similar to those of current
commercial membranes, with selectivities as high as 80 for Cl'/SO42'.99

As previously mentioned, Donnan exclusion is probably the major factor in

the ion-transport selectivity of multilayer polyelectrolyte membranes. Therefore,

high charge densities are vital to achieving large selectivities. In typical

12

polyelectrolyte membranes, the film surface is the major source of the
membrane’s fixed charge, as the interior of the film is often net neutral due to
complete charge compensation in the ionic cross-links.100 To further enhance
the understanding and selectivity of these systems, two methods were developed
to add charge to the interior of the polyelectrolyte film after film deposition. The
first of these methods makes use of polyelectrolytes that are partially derivatized
with photocleavable protecting groups.73 Since the polyelectrolytes are only
partially derivatized, there is still sufficient charge for film formation. Once the
film is deposited, UV light is used to photolyze the protecting group, leaving
either positive or negative charges in the bulk of the film, depending on the
polyelectrolyte employed. PAA/PAH membranes containing PAA with 0, 33, 50,
and 63% of it carboxylic acid groups converted to negative groups only after film
formation show CI'ISO42' selectivities of 9, 100, 150, and 170, respectively. The
second method for forming net charge in the bulk of films involves the deposition
of PAA partially complexed with Cu“.101 Removal of the Cu“ after deposition
can yield Cl'/SO42' selectivities as high as 610 for cross-linked PAA/PAH
membranes. Both of these methods show that Donnan exclusion plays a large
part in the selectivity of these membranes.

Chapter 3 of this dissertation describes highly selective ion separations,
which are achieved primarily by size selectivities, and not by Donnan
exclusion.102 The membranes that effect these separations are formed using

layer-by-layer adsorption of a poly(amic acid) and PAH on porous alumina, and

subsequent heat-induced imidization yields ultrathin, polyimide films (Figure 1.3).

13

 

Figure 1.3: Heat-induced imidization of multilayer poly(amic

acid)/poly(allylamine) films.

14

These membranes allow remarkably selective transport of singly over doubly
charged ions, and their minimal thickness permits high flux of the monovalent
species. However, selectivity is based primarily on the larger size of the divalent
ions because the polyimide films are essentially neutral materials with relatively
low free volume. The membranes consist of a base of five bilayers of PSS/PAH
attached electrostatically to the alumina support and a selective capping film of
poly (p-phenylenepyromellitamic acid) (PMDA-PDA)/PAH. The extent of
imidization can be controlled by the heating temperature, which has a direct
impact on the flux of the ions through the membrane. Membranes imidized at
temperatures ranging between 150 to 180 °C have CI‘ISO42' selectivities around
1000 and K“/Mg2+ selectivities of 300. Increasing the number of PMDA-

PDA/PAH layers results in increases in selectivity and decreases in ion fluxes.

1.6 Ultrathin polyimide membranes for gas separations

In the past 20 years, membranes have been widely investigated for the
separation of gases for potential applications such as oxygen enrichment for
medicinal use or enhanced combustion, carbon dioxide removal from natural gas
streams, and large scale purification of hydrogen for fuel cells.12 The solution-
diffusion mechanism of selective gas permeation that generally operates in these
films is based on the solubility and diffusion of a gas within a polymeric
membrane. Differences in solubilities and diffusivities among gases result in

separation. Such differences depend on the polymer used to make the

15

membrane, and the gas permeability of a polymer is often inversely related to its
selectivity.103

There have been some investigations of polyelectrolyte multilayers as gas
separation membranes. Polyelectrolyte films can increase the gas selectivity of
non-porous supports,58"°4'105 but deposition of these films on porous supports

‘05-‘06 The earliest work was reported by

yields little increase in selectivity.
Stroeve et al., in which they coated porous CelgardTM (polypropylene) and non-
porous silicone membranes with up to 50 bilayers of PSS/PAH.105 Transport
through the coated CelgardTM membranes was still dominated by Knudsen
diffusion, and the coated silicone membranes showed at most a 20% increase in
C02/N2 selectivity over the bare silicone. This small selectivity enhancement may
be due to coverage of defects in the support. Leva'salmi and McCarthy later
showed that PSS/PAH films could increase the selectivity of non-porous poly(4-
methyl-1-pentene) membranes.104 Krasemann and Tieke prepared films
composed of PSS and various polycations and applied these to gas permeation
experiments.107 The best selectivity they achieved was only 1.5 (COz/Nz using
PSS/poly(4-vinylpyridine) multilayers). Standard gas-separation membranes with
similar permeability would have selectivity values greater than 20 for COz/Nz.
We suspect that the lack of gas-transport selectivity in polyelectrolyte films is
inherent in the structure and packing of the polyelectrolytes thus far employed.
As mentioned in section 1.2, polyimide membranes can provide highly

selective gas separations. The selectivity results from a combination of low free

volume, low segmental motion and small interchain distances. Some general

16

trends relate structure to gas permeability and selectivity, but there is no clear
methodology to predict these properties.34 One such trend is that m,m’ imide
linkages (determined by the diamine) within the polymer generally decrease the
permeability and increase the selectivity relative to p,p’ linkages. This is most
likely a result of a differences in the segmental motion of the glassy polymer, as
m,m’ linkages put a kink in the polymer chain and do not allow for rotational
motion. Increasing the rigidity of the monomers has a similar effect.
Permeability increases when bulky CF3 or dimethylsiloxyl groups are
incorporated into polyimides, but dimethylsiloxyl groups also significantly reduce
the selectivity. The bulky groups likely disrupt packing and increase the free
volume of the membrane. The CF3 groups still allow selectivity because the
rotational mobility about the flexible linkages is also slightly inhibited.108

Chapter 4 of this dissertation describes the formation of gas-selective,
fluorinated polyimide membranes on porous alumina supports by the alternating
electrostatic deposition of poly(amic acid) salts and PAH followed by heat-
induced imidization.109 These membranes are appealing because of the stability
and selectivity of polyimides and the fact that alternating polyelectrolyte
adsorption allows formation of ultrathin (as low as 35-40 nm), defect-free
membranes that allow high flux. FT IR spectroscopy shows that heating of
poly(amic acid)/PAH films at 250 °C for two hours completely converts the
poly(amic acid) to the corresponding polyimide, and scanning electron

microscopy reveals uniform films with minimal deposition in substrate pores.

Permeability coefficients and selectivities (Oz/N2 up to 6.9 and COZICH4 up to 68)

17

of three different imidized poly(amic acid)/PAH membranes are comparable to
literature values for the corresponding bulk polyimides, provided that the ratio of

poly(amic acid) to PAH in the film is high (9:1 ).

1.7 Pervaporation separations using cross-linkable polyimides.
Pervaporation is a relatively new membrane-based technique for the
separation of liquid mixtures.110 In this process, a liquid mixture is exposed to the
surface of a membrane whose permeate side is at reduced pressure, and
selective transport of one component through the membrane allows for solvent
purification or analyte collection. Pervaporation is attractive because it allows for
the separation of azeotropic mixtures and often requires less energy than
conventional distillationm'112 Many of the successes with pervaporation involve
the removal of water from organic solvents, but the reverse separation has also

been demonstrated.‘ ‘3” ‘5

Even with advances in pervaporation technology,
however, there is still a need for membrane systems with increased selectivity
and flux.

Tieke and coworkers were the first to show that multilayer polyelectrolytes

“6"" Their membranes

could serve as selective pervaporation membranes.
consisted of 60 bilayers of polyelectrolytes deposited on PAN/PET, and an
ethanol solution containing 6.2 wt% water was used as the feed for
pervaporation. They found that selectivities and flux values varied greatly with

the polyelectrolytes used to form the film and that deposition pH had a large

effect on the performance of the membrane. They suggested that the optimal

18

deposition pH is the mean pKa value of the cationic and anionic groups of the
polyelectrolytes. The larger selectivity values at these pH values may occur
because of a high ionic cross-link density that results in a very compact film
structure. Although not specifically stated, this high degree of ionic cross-linking
should keep film swelling to a minimum. Additionally, annealing of films at 90 °C
increases the selectivity of all of polyelectrolyte pervaporation membranes,
presumably because it allows for formation of even more ionic cross-links.

Tieke and coworkers also investigated how the membrane performance
depends on the addition of supporting salt to the polyelectrolyte deposition
solutions. They found that addition of 1 M NaCl yields a 4-fold decrease in
membrane flux, presumably because film thickness increases with ionic strength
in deposition solutions. The selectivity also increased with the ionic strength.
For example, PVA/PVS films deposited without and with 1 M NaCl had selectivity
values of 35 and 280, respectively. I think that the large variations in the
selectivity of the different films may be to due incomplete pore coverage or
defects in thinner films. No SEM images or thickness data for films deposited on
PAN/PET supports are given, so it is difficult to know whether films are thick
enough to cover the 20 to 200 nm-diameter pores in the underlying supports.

Meier-Haack and coworkers deposited polyelectrolytes on an asymmetric
membrane composed of carboxyl-funotionalized polyamide-6.82 This support
membrane contains a dense, 0.5 um-thick skin layer, but it provides little
selectivity in pervaporation of alcoholic solutions. Deposition of as few as 6-

bilayers of polyelectrolytes on the support completely covers the substrate, to

19

give very high water/alcohol selectivities. Membranes coated with
poly(ethylenimine)/alginic acid films exhibit the highest selectivities; 90% ethanol
feed solutions had a water/alcohol selectivity of 1400. They credited the
selectivities to the hydrophilicity of the film and the high degree of ionic cross-
links.

In Chapter 5, I describe the formation of covalently cross-linked polyimide
membranes that have high pervaporation selectivities and fluxes for the removal
of water from alcohol solutions.118 These films were produced using the
alternating deposition of poly(amic acids) and polycations as described in
sections 1.4 and 1.5. Moreover, incorporation of additional carboxylic acid
groups in the poly(amic acid) allows cross-linking via reaction of these groups
with the amines of the polycation. A similar cross-linking reaction was previously
shown in PAA/PAH films.55'83 FTIR spectroscopy confirms both full imidization
and formation of amide cross-links after heating at 250 °C for 2 h, while scanning
electron microscopy reveals uniform, ~50 nm thick films (7.5 to 12.5 poly(amic
acid)/polycation bilayers). Pervaporation was investigated as a function of cross-
linking by varying either the polycation (PAH, polyethylenimine, or
poly(diallyldimethylammonium chloride)) or the number of cross-linkable groups
in the polyamic acid. Maximum cross-linking and selectivities occur with imidized
films prepared from PAH and a poly(amic acid) that contains diaminobenzoic
acid in every repeat unit. Such membranes exhibit water/alcohol selectivities of

1100 and 6100 for solutions containing 10% and 90% isopropanol, respectively,

20

and the minimal thickness of the multilayer films still allows these selectivities to

occur at fluxes of 11 and 2 kg-m'Z-h", respectively.

21

1.8 References.

(1)

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

(5)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

Spencer, M.; Ruben, K.; Li, C.; Williams, P.; Flaim, T. D. Proceedings of
SPIE- The International Society for Optical Engineering 2003, 4979, 79-86.

Delucchi, M.; Barbucci, A.; Cerisola, G. Mater. Eng. 1999, 10, 223-235.
De Rosa, R. L.; Wagner, S. R. J. Adhesion 2002, 78, 113-127.
Jovanovic, R.; Dube, M. A. J. Macromol. Sci. Polym. R. 2004, C44, 1-51.

Yamada, S.; Mrozek, T.; Rager, T.; Owens, J.; Rangel, J.; Willson, C. G.;
Byers, J. Macromolecules 2004, 37, 377-384.

Riehn, R.; Charas, A.; Morgado, J.; Cacialli, F. Appl. Phys. Lett. 2003, 82,
526-528.

Nguyen, T. P.; Le Rendu, P.; Long, P. D.; De Vos, S. A. Surf. Coat. Tech.
2004, 180-181, 646-649.

Fujita, K.; lshikawa, T.; Tsutsui, T. Mol. Cryst. Liq. Cryst. 2003, 405, 83-
88.

Levit, N.; Pestov, D.; Tepper, G. Sensors Actuat. B-Chem. 2002, B82,
241-249.

Matsumoto, T.; Ohashi, A.; Ito, N.; Fujiwara, H. Biosens. Bioelectron.
2001, 16, 271 -276.

Shepherd, R. L.; Barisci, J. N.; Collier, W. A.; Hart, A. L.; Partridge, A. C.;
Wallace, G. G. Electroanalysis 2002, 14, 575-582.

Koros, W. J. In Membrane Separation Systems Recent Developments and
Future Directions; Baker, R. W., Cussler, E. L., Eykamp, W., Koros, W. J.,
Riley, R. L., Strathmann, H., Eds.; Noyes Data Corp.: Park Ridge, New
Jersey, 1991, pp 189-241.

Pinnau, l.; Freeman, B. D. In Membrane Formation and Modification;
Pinnau, l., Freeman, 8. D., Eds.; American Chemical Society:
Washington, DC, 2000, pp 1-22.

Mulder, M. Basic Principles of Membrane Technology, 2 ed.; Kluwer
Academic Publishers: Dordrecht, The Netherlands 1997.

22

(15)

(16)

(17)

(18)
(19)

(20)

(21)

(22)

(23)

(24)

(25)
(26)

(27)

(28)

(29)

(30)

(31)

(32)

Masson, F.; Decker, 0.; Andre, S.; Andrieu, X. Prog. Org. Coat. 2004, 49,
1-12.

Toyooka, T.; Kobori, Y. J. Photopolym. Sci. Technol. 2000, 13, 301-306.

Van Zant, P. Microchip Fabrication: A Practical Guide to Semiconductor
Processing, 4 ed.; McGraw-Hill Professional, 2000.

Roberts., G. G. Langmuir-Blodgett Films“, Plenum Press: New York, 1990.
Li, T.; Mitsuishi, M.; Miyashita, T. Thin Solid Films 2004, 446, 138-142.

Budianto, Y.; Aoki, A.; Miyashita, T. Macromolecules 2003, 36, 8761-
8765.

Decher, G. Science 1997, 277, 1232-1237.

Zhou, Y.; Bruening, M. L.; Bergbreiter, D. E.; Crooks, R. M.; Wells, M. J.
Am. Chem. Soc. 1996, 118, 3773-3774.

Major, J. S.; Blanchard, G. J. Langmuir2001, 17, 1163-1168.

Liu, Y.; Bruening, M. L.; Bergbreiter, D. E.; Crooks, R. M. Angew. Chem.
Int. Ed. Engl. 1997, 36, 2114-2116.

Sroog, C. E. Prog. Polym. Sci. 1991, 16, 561-694.
Senturia, S. D. Polym. Mater. Sci. Eng. 1986, 55, 385-389.

Duran, J.; Viswanathan, N. S. Organic Coatings and Applied Polymer
Science Proceedings 1983, 48, 283-286.

Baek, S. H.; Kang, J.-W.; Li, X.; Lee, M.-H.; Kim, J.-J. Opt. Lett. 2004, 29,
301-303.

Miyadera, N.; Kuroda, T.; Takahashi, T.; Yamamoto, R.; Yamaguchi, M.;
Yagi, S.; Koibuchi, S. Mol. Cryst. Liq. Cryst. 2003, 406, 233-243.

Johnson, W. S.; Cobb, T. Q.; Lowther, S.; St. Clair, T. L. Proceedings of
the Annual Meeting of the Adhesion Society 1998, 215t, 33-35.

Kourtides, D. A. Flame Retard. Polym. Mater., Proc. Conf. Recent Adv.
Flame Retard. Polym. Mater. 1990, 285-292.

da Silva, L. F. M.; Adams, R. D.; Gibbs, M. Int. J. Adhes. Adhes. 2004, 24,
69-83.

23

(33) Millington, 8.; Shaw, S. J. MRS Bull. 2003, 28, 428-433.

(34) Langsam, M. In Polyimides Fundamentals and Applications; Ghosh, M. K.,
Mittal, K. L., Eds.; Marcel Dekker: New York, 1996, pp 697-741.

(35) Kilian, H.-G.; Bronnikov, S.; Sukhanova, T. J. Phys. Chem. 32003, 107,
1 3575-1 3582.

(36) Cho, D.; Yang, G.; Drzal, L. T. Macromol. Res. 2003, 11, 297-302.

(37) Baur, J. W.; Besson, P.; O'Connor, S. A.; Rubner, M. F. Mater. Res. Soc.
Symp. Proc. 1996, 413, 583-588.

(38) Srinivasan, M. P.; Gu, Y.; Begum, R. Colloid Surface A 2002, 198-200,
527-534.

(39) Koch, G. H.; Brongers, M. P. H.; Thompson, N. G.; Virrnani, Y. P.; Payer,
J. H. “Corrosion Cost and Preventive Strategies in the United States," CC
Technologies Laboratories, Inc., 2001.

(40) Flemming, H.-C.; Schaule, G. Werkst. Korros. 1994, 45, 29-39.

(41) Melo, L. F.; Bott, T. R. Exp. Therm. Fluid Sci. 1997, 14, 375-391.

(42) Bellucci, F.; Nicodemo, L.; Monetta, T.; Kloppers, M. J.; Latanision, R. M.
Corros. Sci. 1992, 33, 1203-1226.

(43) Ritchie, J. E.; Wells, C. A.; Zhou, J.-P.; Zhao, J.; McDevitt, J. T.; Ankrum,
C. R.; Jean, L.; Kanis, D. R. J. Am. Chem. Soc. 1998, 120, 2733-2745.

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

(45) Iler, R. K. J. Colloid Interface Sci. 1966, 21, 569-594.

(46) Decher, G.; Hong, J. D. Ber. Bunsen-Ges. Phys. Chem. 1991, 95, 1430-
1434.

(47) Decher, 6.; Hong, J. D. Makromol. Chem, Macromol. Symp. 1991, 46,
321-327.

(48) Ferreira, M.; Rubner, M. F. Macromolecules 1995, 28, 7107-7114.

(49) Klitzing, R. v.; Mohwald, H. Macromolecules 1996, 29, 6901-6906.

24

(50)

(51)

(52)

(53)

(54)

(55)

(55)

(57)
(58)
(59)

(50)

(51)

(62)

(63)

(64)

(55)

(55)

Yang, X.; Johnson, S.; Shi, J.; Holesinger, T.; Swanson, B. Sensors and
Actuators 1997, 45, 87-92.

Sukhishvili, S. A.; Granick, S. J. Chem. Phys. 1998, 109, 6861-6868.

Lvov, Y.; Ariga, K.; lchinose, l.; Kunitake, T. J. Am. Chem. Soc. 1995, 117,
6117-6123.

Moriguchi, l.; Teraoka, Y.; Kagawa, S.; Fendler, J. H. Chem. Mater. 1999,
11, 1603-1608.

Caruso, F.; Niikura, K.; Furlong, D. N.; Okahata, Y. Langmuir1997, 13,
3422-3426.

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

Delcorte, A.; Bertrand, P.; Wischeroff, E.; Laschewsky, A. Langmuir1997,
13, 5125-5136.

Chen, W.; McCarthy, T. J. Macromolecules 1997, 30, 78-86.
Kotov, N. A.; Magonov, 8.; Tropsha, E. Chem. Mater. 1998, 10, 886-895.
Krasemann, L.; Tieke, B. Langmuir 2000, 16, 287-290.

Arys, X.; Jonas, A. M.; Laschewsky, A.; Legras, R. In Supramolecular
Polymers, 2000, pp 505-563.

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

Anzai, J.; Kobayashi, Y.; Nakamura, N.; Nishimura, M.; Hoshi, T.
Langmuir 1999, 15, 221 -226.

Bergbreiter, D. E.; Franchina, J. G.; Kabza, K. Macromolecules 1999, 32,
4993-4998.

Dennody, D. L.; Peez, R. F.; Bergbreiter, D. E.; Crooks, R. M. Langmuir
1999, 15, 885-890.

Watanabe, H.; Regen, S. L. J. Am. Chem. Soc. 1994, 116.

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

25

(57)

(58)
(59)
(70)

(71)

(72)

(73)

(74)

(75)

(75)

(77)
(73)

(79)

(80)

(81)

(82)

(83)

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.

Yonezawa, T.; Onoue, S.; Kunitake, T. Adv. Mater. 1998, 10, 414-416.
Liu, Y.; Claus, R. O. J. Appl. Phys. 1999, 85, 419-424.

Fendler, J. J. Std. Surf. Sci. Cafe]. 1997, 103, 261-276.

Kovtyukhova, N. l.; Ollivier, P. J.; Martin, B. R.; Mallouk, T. E.; Chizhik, S.
A.; Buzaneva, E. V.; Gorchinskiy, A. D. Chem. Mater. 1999, 11, 771-778.

Cassagneau, T.; Fendler, J. H. Adv. Mater. 1998, 10, 877-881.

Dai, J.; Balachandra, A. M.; Lee, J. I.; Bruening, M. L. Macromolecules
2001 , 35, 3164-3170.

Dai, J.; Jensen, A. W.; Mohanty, D. K.; Emdt, J.; Bruening, M. L. Langmuir
2001, 17, 931-937.

Cheung, J. H.; Fou, A. F.; Rubner, M. F. Thin Solid Films 1994, 244, 985-
989.

Onda, M.; Lvov, Y.; Ariga, K.; Kunitake, T. Biotechnol. Bioeng. 1996, 51,
163-167.

Stepp. J.; Schlenoff, J. B. J. Electrochem. Soc. 1997, 144, L155-L157.
Laurent, 0.; Schlenoff, J. Langmuir1997, 13, 1552-1557.

Sun, Y.; Zhang, X.; Sun, 0.; Shen, J. Macromol. Chem. .Physic1996, 197,
147-153.

Fou, A. C.; Onitsuka, O.; Ferreira, M.; Rubner, M. F.; Hsieh, B. R. J. Appl.
Phys. 1996, 79, 7501-7509.

Hammond, P. T.; Whitesides, G. M. Macromolecules 1995, 28, 7569-
7571.

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

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

26

(84)

(35)
(85)

(87)

(38)
(89)

(90)

(91)

(92)

(93)

(94)

(95)

(95)

(97)

(98)

(99)

Wang, X.; Spencer, H. G. Trends in Polym. Sci. 1997, 5, 38-39.

Pothakamury, U. R.; Barbosa-Canovas, G. V. Fluid Part. Sep. J. 1993, 6,
110-118.

Meier-Haack, J.; Booker, N. A.; Carroll, T. Water Res. 2003, 37, 585-588.
Fusaoka, Y.; lnoue, T.; Murakami, M.; Kurihara, M. Membrane Technology
Conference: The Future of Purer Water, Proceedings, San Antonio, TX,
United States, Mar. 4-7, 2001 2001, 1369-1373.

Weber, W. J., Jr.; LeBoeuf, E. J. Water Sci. Technol. 1999, 40, 11-19.
Saracco, G.; Zanetti, M. C. Ind. Eng. Chem. Res. 1994, 33, 96.

Ventura, X. A. Proceedings - AESF SUR/FIN Annual lntemational
Technical Conference 2003, 477-492.

Li, H.; He, X.; Liang, Y. Sep. Sci. Technol. 2003, 38, 1633-1648.

Genders, J. D.; Hartsough, D.; Thompson, J. Proc. - Electrochem. Soc.
1994, 94-22, 457-466.

Kim, S.-G.; Kim, Y.-l.; Yun, H.-G.; Lim, G.-T.; Lee, K.-H. J. Appl. Polym.
Sci. 2003, 88, 2884-2890.

Loeb, S.; Sourirajan, S. Adv. Chem. Ser. 1963, 38, 117-132.

Polotskaya, G. A.; Kuznetsov, Y. P.; Goikhman, M. Y.; Podeshvo, l. V.;
Maricheva, T. A.; Kudryavtsev, V. V. J. Appl. Polym. Sci. 2003, 89, 2361-
2368.

Toutianoush, A.; Tieke, B. In Novel Methods to Study Interfacial Layers;
Elsevier: Amsterdam, 2001, pp 416-425.

Jin, W.; Toutianoush, A.; Tieke, B. Langmuir 2003, 19, 2550-2553.

Harris, J. J.; Stair, J. L.; Bruening, M. L. Chem. Mater. 2000, 12, 1941-
1946.

Stanton, B. W.; Harris, J. J.; Miller, M. D.; Bruening, M. L. Langmuir2003,
19, 7038-7042.

(100) Riegler, H.; Essler, F. Langmuir 2002, 18, 6694-6698.

27

(101)

(102)

(103)
(104)

(105)

(106)

(107)

(108)

(109)

(110)

(111)

(112)

(113)

(114)

(115)

(116)

(117)

(118)

Balachandra, A. M.; Dai, J.; Bruening, M. L. Macromolecules 2001, 35,
3171-3178.

Sullivan, D. M.; Bruening, M. L. J. Am. Chem. Soc. 2001, 123, 11805-
11806.

Robeson, L. M. J. Membrane Sci. 1991, 62, 165-185.
Levasalmi, J.-M.; McCarthy, T. J. Macromolecules 1997, 30, 1752-1757.

Stroeve, P.; Vasquez, V.; Coelho, M. A. N.; Rabolt, J. F. Thin Solid Films
1 996, 284-285, 708-712.

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

Krasemann, L.; Tieke, 8. Mat. Sci. & Eng. C1999, 8-9, 513-518.

Coleman, M. R.; Koros, W. J. J. Polym. Sci, Part B: Polym. Phys. 1994,
32, 1915-1926.

Sullivan, D. M.; Bruening, M. L. Chem. Mater. 2003, 15, 281-287.

Huang, R. Y. M. Pervaporation Membrane Separation Processes; Elsevier
Science: Amsterdam, 1991; Vol. 1.

Lipnizki, F.; Field, R. W. Environ. Prog. 2002, 21, 265-272.

Aminabhavi, T. M.; Khinnavar, R. S.; Harogoppad, S. B.; Aithal, U. S. J.
Macromol. Sci. R. M. C. 1994, C34, 139-204.

Liu, Q.-L.; Xiao, J. J. Membrane Sci. 2004, 230, 121-129.

Volkov, V. V.; Fadeev, A. G.; Khotimsky, V. S.; Litvinova, E. G.;
Selinskaya, Y. A.; McMillan, J. D.; Kelley, S. S. J. Appl. Polym. Sci. 2004,
91, 2271 -2277.

Slater, C. S.; Hickey, P. J.; Juricic, F. P. Sep. Sci. Technol. 1990, 25,
1063-1077.

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

Krasemann, L.; Toutianoush, A.; Tieke, B. J. Membrane Sci. 2001, 181,
221-228.

Sullivan, D. M.; Bruening, M. L. submitted to Langmuir2004.

28

Chapter 2

Ultrathin, Layered Polyimide Coating on Aluminum

2.1 Introduction

In 1999, Harris and Bruening reported the synthesis of ultrathin, nylon-like
coatings through cross-linking of poly(acrylic acid)/poly(allylamine hydrochloride)
(PAA/PAH) films deposited on gold surfaces.1 They showed that such films have
the potential to passivate surfaces, and subsequent studies examined protection
of aluminum by these coatings.2'3 Here, I report on the possibility of using similar
ultrathin polyimide coatings to protect Al from corrosion. In contrast to PAA/PAH
films, these coatings are constructed using a layer-by-layer method based on
formation of covalent, rather than ionic, bonds. Such polyimide films protect the
native oxide layer of Al, and even a 30 nm-thick film increases the impedance of
Al by 2 orders of magnitude.

Nevertheless, passivation by the films employed in this work is
significantly less than that of typical coatings because surface passivation
generally increases with coating thickness. However, there are applications
where ultrathin coatings are required. In heat exchangers, for example, coatings
should be thin enough that they do not drastically reduce heat-transfer
coefficients.“*5 Thin coatings are also important for protecting electronic
materials6 and high Tc superconductors.7

In spite of the fact that thin films are not as passivating as thick coatings,

they can provide substantial protection. This should not be too surprising, as

29

thin-film formation is the mechanism of corrosion prevention by inhibitors such as
1H-benzotriazole on copper.8 In a similar way, fatty acids can form a protective
layer on Alf"11 and alkanethiolates form protective monolayers on copper.‘2'14
The ultrathin cross-linked films discussed here should have a stability advantage
over monolayer films, but they do have the disadvantage of needing a heat
treatment. Such heat treatments are common, however, in the deposition of
practical epoxy coatings.15

I should note that the film-formation method used in this work is similar to
a procedure reported by Liu and co-workers. They described a covalent process
in which polyamidoamine (PAMAM) dendrimers and GantrezTM (poly(methyl vinyl
ether-aIt-maleic anhydride» are covalently deposited in a layer-by-layer
fashion.16 Heating of these films produces a passivating polyimide film.17 In this
study, I utilize poly(allylamine) and GantrezTM in an analogous fashion to prepare
layered polyimide films (Figure 2.1). Although such films provide some

protection for Al, under the conditions of this study, they are slightly less

protective than ultrathin nylon-like films prepared from multilayer PAA/PAH films.

2.2 Experimental Section

2.2.1 Materials. Poly(allylamine hydrochloride) (MW ~ 70 000), sodium metal,
methanol, tetrahydrofuran, and dimethyl sulfoxide were obtained from Aldrich.
The GantrezTM (MW ~1,100,000, Mn ~310,000) was a gift from ISP Technologies
(Wayne, NJ). THF and DMSO were dried using molecular sieves. Al-coated

substrates were made by sputtering 200 nm of Al on Si(100) wafers. This

30

 

l ALUMINUM I
O - O

CH3O

n CH30 1-n
o O

 

 

heat

 

I ALUMINUM l

 

Figure 2.1: Synthesis of a GantrezTM/poly(allylamine)
bilayer on Al and subsequent conversion to an imide-

containing film.

31

substrate has a native oxide layer with an ellipsometric thickness of about 40 A
after 15 min of cleaning in a UV/03 cleaner. In this chapter, when we refer to Al
or “bare“ Al, we mean Al with its native oxide.

Poly(allylamine hydrochloride) was converted to the unprotonated
poly(allylamine) (PAAm) by reacting it with freshly generated sodium methoxide.
In a typical procedure, 1g of sodium metal was placed into a nitrogen-purged
three-neck round-bottom flask fitted with a condenser. Then, 70 mL of methanol
was added dropwise while stirring and cooling in an ice bath. After dissolution of
the sodium, a stoichiometric amount of PAH (4.07 g) was added to the solution
and allowed to react at room temperature for 1 h. The resulting solution was
then filtered to remove some of the insoluble sodium chloride. The methanol was
removed in a vacuum and the product was dissolved in chloroform. This solution
was filtered to remove the remaining sodium chloride, and the chloroform was
removed to yield the product. The complete conversion to poly(allylamine) is
confirmed by NMR studies: 1H NMR (300 MHz, 020): 60.84.45 (m), 2.25-2.55
(s).

2.2.2 GantrezTMlPAAm Film Synthesis. The Al substrate was placed in a
UV/03 cleaner for 15 min and then immersed in the GantrezTM solution (25 wt %
in THF) for 20 min. Next, the wafer was rinsed with 3 mL of THF, placed into a
20-mL THF bath, sonicated for 3 min, and dried with N2. The wafer was then put
into a PAAm solution (0.1 wt % in DMSO) for 30 min, rinsed with 3 mL of DMSO,
and sonicated in DMSO for 3 min. Residual DMSO was removed by rinsing

thoroughly with THF. The procedure was repeated until the desired ellipsometric

32

thickness or number of bilayers was achieved. To convert amic acid groups to
imides, the samples were heated at 150 °C under N2 for 2 h.18

2.2.3 Film Characterization. External reflectance FTIR spectra of films were
obtained with a Nicolet Magna-IR 560 spectrometer using a Pike grazing angle
(80°) attachment. Film thicknesses were measured with a M-44 rotating analyzer
spectroscopic ellipsometer (J. A. Woollam), assuming a film refractive index of
1.5.

2.2.4 Electrochemical Studies. Impedance data and polarization curves were
obtained with a CH Instruments electrochemical analyzer (Model 604).
Measurements were made using a Ag/AgCI (3 M KCI) reference electrode and a
platinum wire counter electrode. The working electrode was ”bare" or coated Al
contained within an O-ring holder that exposed 2.2 cm2 of the sample. We
intentionally used a rather corrosive electrolyte solution (0.5 M NaCl adjusted to
pH 3.0 with HCI) for impedance measurements to differentiate between the
impedances of bare and film-coated AI. At neutral pH, the impedance of bare AI
is as high as 107 0 cm2 after a 4 h immersion in 0.5 M NaCl due to the native
aluminum oxide layer. Acquisition of impedance data started only after 4 h of
immersion in NaCl solutions in order to achieve stable values. For bare Al, the
impedance data were acquired at the open circuit potential (ca. -1.20 V vs
Ag/AgCl) using a sinusoidal voltage of 5 mV. Other impedance measurements
were made at a dc potential of -0.75 V vs Ag/AgCl because coated Al has a
much more positive open circuit potential (048 to -0.75 V) than bare Al.

Applying potentials higher than -0.7 V (pitting potential of Ala) causes rapid

33

corrosion in some cases. The frequency range for impedance measurements
was 105 Hz to 10 mHz. Impedance data fits were performed using LEVM 7.0
software written by J. Ross MacDonald (available from Solatron).19 All
electrochemical experiments were performed on at least three different

electrodes.

2.3 Results and Discussion

2.3.1 Synthesis of GantrezmlPAAm Films. Synthesis of GantrezTM/PAAm
films occurs by alternating deposition of the two polymers as shown in Figure 2.1.
In the initial deposition of GantrezTM on Al, we speculate that anhydride groups
react with the surface to form carboxylate salts. The presence of large anhydride
peaks (1860 and 1790 cm") in the FTIR spectrum of the first GantrezT'“ layer (G2
1, Figure 2.2a) shows that most of the anhydride groups did not react with the
surface and are available for subsequent attachment of PAAm. Upon exposure
to PAAm, most of the anhydride groups on the surface react with amine groups,
covalently linking PAAm to the surface. FTIR spectroscopy (spectrum PAAm 1,
‘Figure 2.2b) gives evidence for this reaction as the peaks corresponding to cyclic
anhydrides diminish after reaction with PAAm, and amide peaks (1630 and 1540
cm") appear. Deposition of multilayers proceeds by alternating reactions of
GantrezTM with surface amine groups and PAAm with unreacted surface
anhydride groups as shown by FTIR spectra (Figure 2.2). This chemistry is
analogous to that used previously in preparing films containing alternating layers

of GantrezTM and PAMAM dendrimers.16

34

 

 

b
.002 l .002 R After
heating

 

 

 

Absorbance

 

 

 

 

 

 

“M

1900 1800170016001500 19001800170016001500
Wavenumbers (cm'I)

 

 

 

Figure 2.2: Reflectance FT IR spectra of GantrezTM/PAAm films. Figure A
shows spectra after the deposition of GantrezTM on Al (G2 1) and on one
(G2 2) and two (G2 3) GantrezTM/PAAm bilayers. Figure B shows the
spectra after deposition of PAAm ( 1, 2, and 3 GantrezTM/PAAm bilayers) as
well as the spectrum of a 3-bilayer film after heating. The spectrum of the
heated film is multiplied by 0.5 for clarity.

Conversion of GantrezTM/PAAm films to polyimides occurs by heating of the film
to convert amic acids to imides as shown in Figure 2.1. Reflectance FT IR
spectra confirm this conversion as seen from the appearance of the cyclic imide
bands at 1770 and 1710 cm'1 and the disappearance of amide bands at 1630
and 1540 crn'1 (Figure 2.2b). In addition to creating a polyimide film, this process
likely anneals the structure, resulting in a more passivating film. Unlike previous
studies with heated dendrimer/GantrezTM films,‘6'18 GantrezTM/PAAm films will
not undergo retro-Michael additions to form additional internal cross-links. This is
because PAAm does not contain the secondary and tertiary amines that exist in
PAMAM dendrimers.

2.3.2 Electrochemical Impedance Spectroscopy. Electrochemical impedance
spectroscopy is one of the most powerful methods for studying corrosion
because it often provides mechanistic information about corrosion processes.”29
In this technique, one determines the impedance of an electrochemical cell over
a wide range of frequencies and then uses an equivalent circuit to describe
impedance as a function of frequency. Various equivalent circuit models were

22.26-28.30-32 and

proposed to simulate the impedance behavior of coated aluminum,
most of these are similar to the circuits shown in Figure 2.3. For bare Al, the
insulating oxide film forms a leaky capacitor, resulting in circuit A. The presence
of an organic coating that is highly blocking complicates the circuit by the
introduction of another leaky capacitor as shown in circuit B. As a parallel RC

component results in a semicircle in a plot of real versus imaginary components

of impedance, the introduction of an organic coating often results in a plot with

36

 

 

 

 

 

Rf —’WV\1—
ROX

Figure 2.3: Equivalent circuits for impedance data. Circuit A was used to
fit "bare" Al and AI coated with GantrezTM/PAAm films. Circuit B would be
used if the organic film were highly resistive. The physical meaning of the
different symbols is Rs, solution resistance, R0,, oxide resistance, R,, film

resistance, Cf, film capacitance, and Cox, oxide capacitance.

37

two partial semicircles.2°"°"'26 Note that the magnitude of the diameter of each
semicircle is proportional to the resistance in the parallel RC circuit.33 Thus,
larger semicircles in impedance plots represent better corrosion protection.
Scully and Hensley suggest that coating resistances need to be at least 107 0
cm2 to provide effective protection of Al.34
2.3.3 Protection of AI by GantrezTMlPAAm Coatings. Heated
GantrezTM/PAAm films protect the underlying oxide (Table 2.1), but average
values for R0,, are a factor of 5 to 10 lower for 2- and 3-layer films than for related,
heated PAA/PAH films with similar thicknesses.3 The heated PAA/PAH films
form amide cross-links upon heating rather than imides. Additiionally, the
GantrezTM/PAAm films may yield less surface protection because of a lower
cross-linking density or weaker binding to the AI surface. Figure 2.4 shows
representative impedance plots for Al coated with a 3-bilayer GantrezTM/PAAm
film before and after heating. Unlike heated PAA/PAH films, there is only one
semicircle in the impedance plot for imidized GantrezTM/PAAm.3 Plots of phase
angle versus frequency also show only one RC time constant for these
electrodes. This suggests that R, values for GantrezTM/PAAm systems are
minimal, so we used circuit A (Figure 2.3) to model these data.35

The impedance data explain why even ultrathin films may reduce the rate
of Al corrosion. Binding of a monolayer of GantrezTM to the surface passivates
the oxide layer and, in turn, protects Al (Table 2.1) by increasing the oxide
resistance. The film resistance due to a single GantrezTM layer or unheated

GantrezTM/PAAm films should be small, but the film still apparently inhibits

38

 

 

 

0.6
(QT
E 0.4 A
O
C} A I I I I
g n
=1}! 0.2 - '
11'. ..O...\. 0
g.” X5
0.0 g

 

0.0 0.1 02 003 0.4
2' (M0cm2)

Figure 2.4: Impedance plots for bare Al (circles, both 2' and Z' '
were multiplied by 5 for clarity) and Al coated with 3 bilayers of
GantrezTM/PAAm before (squares) and after heating (triangles).
All impedance data were measured in 0.5 M NaCl at pH 3 after a

4 h immersion time.

39

no 9 me no A we to 9 3 No 9 we to u. E, - 3 9 3 59.8; .552: so

No H 3. No 9 3 we .1. we no 9 3 3 9 It, to H 3. - 5952c: .reome so
N 9 2 m .1. m a H m 0.0 .1. I No 9 3 - mod .1. 3.0 598; heads; gm
5.0 0.. E to .1. No mo 9. no to U... to to H No no 9 3.. - 598%: $5.03 at
EN oi 09 3 R - - g 5552 5cm 355.25 8:80

SN of 2: m2 8 om 2v 8:8; 998 880.25 8:80

 

296.3 298.3 296.5 .035 293.3 Ewmbcmo

m ON N m; F 203.80.: _< 2mm

 

.mEE $55.22.mvzoEEchEmO
Co 295 maotm> 53> 8600 38:00.0 2 .2 wLEmEEmQ 3526 E06230 new $30525 9580 gnu 035...

40

dissolution of the oxide. Thus, the oxide resistance is equivalent to what it would
be in a less corrosive, neutral solution. Previous studies showed that copolymers
of maleic acid and styrene protect pigments in solution,36 but it is somewhat
surprising that pure GantrezTM provides protection because it is not particularly
hydrophobic (the 4 h immersion time in water should convert the remaining
anhydrides to diacids). We surmise that a protective layer of Al carboxylates is
forming on the surface similar to protective layers formed by corrosion
inhibitorsf”10 The layer inhibits CI' adsorption,37 thus protecting the oxide. This
layer will not be as ordered as protective films of self-assembled monolayers, but
it does passivate the oxide film

Oxide resistances of AI electrodes coated with GantrezTM/PAAm films
depend substantially on the number of layers in the film and whether the film is
imidized. Interestingly, reaction of the GantrezTM with a layer of PAAm results in
a 12-fold decrease in oxide resistance (unheated films, Table 2.1). This may be
due to the fact that PAAm is more hydrophilic than Gantrezm. lmidization of
Gantrezm/PAAm films increases R0,, by an order of magnitude.

As the number of GantrezTM/PAAm layers increases, oxide resistance
increases significantly (Table 2.1). For heated films, R0,, increases by an order of
magnitude on going from 1 to 3 bilayers of GantrezTM/PAAm. Multilayers
increase R0,, values because they protect the oxide from degradation, even
though their resistance may not be extremely high. An increasing number of

layers may result in fewer defects in the coating. Cox does not vary much with

41

the number of layers because the thickness and dielectric constant of the

aluminum oxide are relatively independent of coating thickness.

2.4 Conclusions

GantrezTM/PAAm films can be synthesized on Al by a simple "dip and
rinse", Iayer-by-layer method. Each added layer binds covalently to the previous
layer via reaction of an anhydride (Gantrezm) with a primary amine (PAAm).
Heating of these films results in imidation of GantrezTM/PAAm, and even a single
monolayer of GantrezTM attached to the surface protects the aluminum oxide and
increases the oxide resistance by 35-fold. For GantrezTM/PAAm films on Al,
electrode impedance depends on both the number of bilayers and whether the
film is imidized. lmidization increases oxide resistance by an order of magnitude

over unheated films.

42

2.5 References

(1)

(2)
(3)

(4)
(5)
(5)

(7)

(9)
(1 O)

(11)
(12)

(13)

(14)

(15)

(15)

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

Harris, J. J.; Bruening, M. L. Langmuir2000, 16, 2006-2013.

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

Flemming, H.-C.; Schaule, G. Werkst Korros 1994, 45, 29-39.
Melo, L. F.; Bott, T. R. Exp. Therm. Fluid Sci. 1997, 14, 375-391.

Bellucci, F.; Nicodemo, L.; Monetta, T.; Kloppers, M. J.; Latanision, R. M.
Corros. Sci. 1992, 33, 1203-1226.

Ritchie, J. E.; Wells, C. A.; Zhou, J.-P.; Zhao, J.; McDevitt, J. T.; Ankrum,
C. R.; Jean, L.; Kanis, D. R. J. Am. Chem. Soc. 1998, 120, 2733-2745.

Brusic, V.; Frisch, M. A.; Eldridge, B. N.; Novak, F. P.; Kaufman, F. 8.;
Rush, 8. M.; Frankel, G. S. J. Electrochem. Soc. 1991, 138, 2253-2259.

Allara, D. L.; Nuzzo, R. G. Langmuir1985, 1, 45-52.

Ogawa, H.; Chihera, T.; Taya, K. J. Am. Chem. Soc. 1985, 107, 1365-
1369.

Fischer, E. R.; Parker, J. E., III Corrosion 1997, 53, 62-64.

Haneda, R.; Nishihara, H.; Aramaki, K. J. Electrochem. Soc. 1997, 144,
1215-1221.

Jennings, G. K.; Laibinis, P. E. Colloid SurfaceA 1996, 116, 105-114.

Jennings, G. K.; Munro, J. C.; Yong, T.-H.; Laibinis, P. E. Langmuir1998,
14, 6130-6139.

Grandle, J. A.; Taylor, S. R. Corrosion 1994, 50, 792-803.

Liu, Y.; Bruening, M. L.; Bergbreiter, D. E.; Crooks, R. M. Angew. Chem.
Int. Ed. Engl. 1997, 36, 2114-2116.

43

(17)

(18)

(19)
(20)

(21)

(22)

(23)

(24)

(25)
(25)

(27)

(28)
(29)

(30)

(31)

(32)

(33)

(34)

Zhao, M.; Liu, Y.; Crooks, R. M.; Bergbreiter, D. E. J. Am. Chem. Soc.
1999, 121, 923-930.

Zhao, M.; Zhou, Y.; Bruening, M. L.; Bergbreiter, D. E.; Crooks, R. M.
Langmuir1997, 13, 1388-1391.

Macdonald, J. R. Solid State Ionics 1992, 58, 97-107.
Deflorian, F.; Fedrizzi, L.; Bonora, P. L. Prog. Org. Coat. 1993, 23, 73-88.

Deflorian, F.; Miskovic-Stankovic, V. B.; Bonora, P. L.; Fedrizzi, L.
Corrosion 1994, 50, 438-446.

Kendig, M.; Scully, J. Corrosion 1990, 46, 22-29.

Mansfeld, F.; Chen, C.; Lee, C. C.; Xiao, H. Corros. Sci. 1996, 38, 497-
513.

Bonora, P. L.; Deflorian, F.; Fedrizzi, L. Electrochim. Acta 1996, 41, 1073-
1082.

Griffin, A. J. J.; Brotzen, F. R. J. Electrochem. Soc. 1994, 141, 3473-3479.
Su, P.-C.; Devereux, O. F. Corrosion 1998, 54, 419-427.

Mertens, S. F.; Xhoffer, C.; DeCooman, B. C.; Temmerrnan, E. Corrosion
1997, 53, 381-388.

Scholl, H.; Jimenez, M. M. D. Corros. Sci. 1992, 33, 1967-1978.
Lin, 8.; Shih, H.; Mansfeld, F. Corros. Sci. 1992, 33, 1331-1349.

Bessone, J.; Mayer, C.; Jtittner, K.; Lorentz, W. J. Electrochim. Acta.
1983, 28, 171-175.

Bellucci, F.; Kloppers, M.; Latanision, R. M. J. Electrochem. Soc. 1991,
138, 40-48.

Mansfeld, F.; Kendig, M. W.; Tsai, S. Corrosion 1982, 38, 478-485.

Bard, A. J.; Faulkner, L. R. Electrochemical Methods Fundamentals and
Applications; John Wiley 81 Sons, Inc.: New York, 1980.

Scully, J. R.; Hensley, S. T. Corrosion 1994, 50, 705-716.

44

(35)

(36)

(37)

There is a small possibility that film resistance is high, but the time
constants (I = 1/RC) for the two parallel RC components in circuit B

(Figure 2.3) are similar. This would also result in only one semicircle in
impedance plots

Mueller, B.; Mebarek, D. Angew. Makromol. Chem. 1994, 221, 177-185.

EI-Awady, A. A.; Abd-El-Nabey, B. A.; Aziz, S. G. J. Chem. Soc, Faraday
Trans. 1993, 89, 795-802.

45

Chapter 3

Ultrathin, Ion-Selective Polyimide Membranes Prepared from Layered

Polyelectrolytes

3.1 Introduction

Membrane-based separations are attractive because of their convenience
and low energy costs."2 Separation occurs by simply flowing a mixture past a
membrane, and by either a pressure or concentration gradient, primarily one
component of the mixture will permeate through the membrane. In spite of the
appeal of such separations, development of highly selective membranes that
allow practical fluxes is an ongoing challenge. This is particularly true because
flux is often inversely proportional to selectivity.3

The tradeoff between flux and selectivity generally requires that practical
membrane systems consist of an ultrathin, selective skin on a highly permeable
support. The minimal thickness of the selective layer allows a reasonable flux,
while the support provides needed mechanical strength. Such membranes are
often prepared by phase-inversion processes that result in a polymeric material
with a dense surface layer and a highly porous bulk."'5 Composite membranes
provide an appealing alternative to phase inversion, because deposition of a thin,
selective layer on an inexpensive support allows use of more expensive, and
potentially more discriminating, skin materials.6'7

This chapter describes the formation of selective polyimide skins on

porous alumina by alternating polyelectrolyte deposition (APD) and subsequent

46

imidization (Figure 3.1). Deposition of the polyimide film on a relatively
permeable PSS/PAH base layer allows coverage of the surface with an
extremely thin (4 — 9 nm) polyimide film. Additionally, the APD/heating method is
convenient and affords control over the selectivity of the polyimide film.8'10 Such
polyimide membranes show remarkably selective transport of singly over doubly
charged ions (CI/8042' selectivities reach values as high as 1000), and their
minimal thickness permits high flux of the monovalent species.

Aromatic polyimides are, in general, attractive membrane materials
because of their mechanical strength, thermal stability, selectivity, and wide
tunability.”13 With common processing techniques, however, it is difficult to form
ultrathin selective membranes with thickness less than 40 nm. APD, on the other
hand, offers an easy way to deposit ultrathin multilayered films. A few previous
studies combined APD of poly(amic acids) and a polycation with post-deposition
heating to form polyimide films.”16 Here we utilize this approach to form highly
selective, ultrathin polyimide membranes on porous supports coated with
multilayered polyelectrolyte gutter layers. Because the polyimide portion of such
systems is ultrathin, these membranes simultaneously allow high flux and high
selectivity. Although several groups studied layered polyelectrolyte membranes
for various applications,”23 the polyimide films described here show unique ion-

transport selectivities with deposition of only a few layers.

47

PM DA-PDA/ PAH

3:” "3 5 1. 5 if; ’3- i-‘g (:93 I} 3": '9‘: 15?: 'IL';;'- '15:}?3 3.3? 1: Fl:
. '- 9.1 5'! 1+ 14:» .-

‘ ~‘ " tr.“ > ;

'2? 5' (a 7 Tr. '3 71 'j? “

a PSS/PAH —>

 

 

 

 

PorousT Alumina
/ _o o o 3 r o o\
b --.—NH 0 HN-- I39; -—.—M --
K o o/n \ o o/n

so;

C 0 NH;

PSS PAH

Figure 3.1: (a) Membrane consisting of a porous alumina support, a
PSS/PAH base layer, and a PMDA-PDA/PAH selective capping layer (b)
PMDA-PDA structure and heat induced imidization and (0) P88 and PAH

structures.

48

3.2 Experimental

3.2.1 Chemicals and Materials. Poly(sodium 4-styrenesulfonate) (PSS, Mw =
70,000), poly(allylamine hydrochloride) (PAH, Mw = 70,000), pyromellitic
dianhydride, 1 ,4 phenylenediamine, N,N-dimethylacetamide, and
mercaptopropionic acid (MPA) were used as received from Aldrich. Solutions of
KCI, K2804, K3Fe(CN)5, BaClz, CBCI2, and MgClz were prepared to 0.1 M in 18.2
MQ-cm deionized (Milli-Q) water. Alumina support membranes (Whatman
Anopore, 0.02 pm pore diameter skin layer, 25 mm diameter) were purchased
from Fisher. Silicon(100) wafers (Silicon Quest) sputter coated with either 200
nm of Al or 20 nm of Cr and then 200 nm of Au were used as supports for
ellipsometry and reflectance FTIR measurements.

3.2.2 Substrate Preparation. In the case of porous alumina supports, the
polypropylene support ring on the alumina was removed to prevent it from
melting into substrate pores during heat-induced imidization. This was
accomplished by cutting off as much of the polymer as possible with scissors,
and then burning off the remaining ring at 400 °C for 18 h in a furnace.
Subsequently, the alumina supports were rinsed with acetone, dried with N2, and
cleaned for 15 min in a UV/Oa cleaner (Boekel Industries, model 135500). Au-
coated wafers were first cleaned with “piranha” solution (70% H2804(conc)/30%
H202(aq): caution: piranha is a strong oxidizer and should not be stored in closed
containers), rinsed in water, and dried with N2. They were then immersed into a
2 mM MPA in ethanol solution for 30 min, rinsed with ethanol, and dried with N2.

Al-coated wafers were simply cleaned by UV/Oa prior to use.

49

3.2.3 Synthesis of PMDA-PDA. The poly(amic acid), poly(pyromellitic
dianhydride- phenylenediamine) (PMDA-PDA),24 was prepared using a typical
literature procedure.25 1,4 phenylenediamine (0.66 g) was dissolved in MN-
dimethylacetamide (50 mL) and pyromellitic dianhydride (1.34 g) was added to
the solution in small increments over a ~15 min period. This mixture was purged
with nitrogen and stirred for 24 h. The PMDA-PDA was precipitated into 1.5 L of
ethanol and was subsequently filtered. The product was dried by vacuum,
dissolved in an additional 50 ml of MAI-dimethylacetamide, and precipitated into
ethanol once again. The final product was dried by vacuum overnight and stored
in a desiccator until needed.

3.2.4 Film Characterization. Ellipsometric thickness measurements on Al-
coated silicon wafers were made using a rotating analyzer ellipsometer (model
M-44, J. A. Woollam) and WVASE32 software. A film refractive index of 1.5 was
assumed in all thickness determinations. Reflectance FTIR spectra were
obtained using a Nicolet Magna-560 FTIR spectrophotometer and a Pike grazing
angle (809) attachment. A UV/Oa-cleaned substrate was used as a background.
Field-emission scanning electron microscopy (FESEM) images were obtained
with a Hitachi S-4700 ll instrument using an acceleration voltage of 15 kV.
Samples were fractured in liquid nitrogen and sputter-coated with 5 nm of gold
before imaging.

3.2.5 Membrane Preparation. Specially designed holders that expose only one
side of the membrane to solution were used during the deposition of the

polyelectrolytes. This limits film growth to the top of the membrane and keeps

50

polymer from filling the pores from the back side. Film preparation starts with
exposure of the cleaned membrane surface to a solution of P88 (20 mM, with
respect to the repeating unit, in 0.5 M MnClz, pH adjusted to 2.3, Figure 3.10) for
2 minutes. A single layer of negatively charged PSS remains on the surface after
rinsing with deionized water for 1 minute. Exposure to a PAH solution (20 mM in
0.5 M NaBr, pH adjusted to 2.3, Figure 3.1c) for 5 min followed by water rinsing
for 1 min yields a second, positively charged layer on the alumina. This
procedure was repeated until 5 PSS/PAH bilayers were deposited.

PMDA-PDA solutions for APD were made by first dissolving ~16 mg
PMDA-PDA in 1 ml of 0.15 M triethylamine. This solution was then diluted to 10
ml (5 mM), with 0.5 M NaCl, and the pH was adjusted to 4.5. This solution was
prepared only when needed to reduce the risk of hydrolysis of the poly(amic
acid). The PMDA-PDA solution was deposited on the 5 bilayers of PSS/PAH for
3 min and rinsed to yield the first PMDA-PDA layer. PAH (0.02 M, 0.5 M NaCl,
pH adjusted to 4.5) was added in the same manner. This procedure was
repeated until 1.5, 2.5, or 3.5 bilayers were built up. The extra half bilayer
indicates that the membranes terminated in the poly(amic acid). The samples
were dried with N2 only after all of the layers were added.

3.2.6 Membrane Transport. Dialysis was performed using a holder composed
of two identical glass cells (100 ml volume) connected by a flange with a 2.1 cm2
hole (Figure 3.2). The membrane was centered over the hole and an O-ring was
placed on its top side to provide a seal and some cushion. The membrane

separates a source phase solution (0.1 F salt, unbuffered, pH 5.3-6.2) from a

51

 

 

 

69

Source Phase Receiving Phase
(0.1 N Salt) (DI Water)

Figure 3.2: Dialysis setup used to measure ion-permeability of
polyelectrolyte membranes: magnetic stir bar (a), O-ring and membrane

(b), mechanical stir prop, motor not shown (c), and conductivity probe (d).

52

receiving phase that initially contains deionzed water.26 Both phases were stirred
vigorously to reduce concentration polarization at the two membrane surfaces.
The receiving phase conductivity was measured at 5 min intervals for 45 min to
monitor the change in concentration, and these values were divided by the
source phase conductivity to normalize the data. The conductivity data were
converted to concentration via fitting to conductivity calibration curves of standard
salt solutions (5e'7 to 0.1 M). Selectivity coefficients were determined by dividing
KCI flux values by those of another salt.

In diffusion dialysis experiments, membrane conditioning was required to
obtain steady flux values. (We witnessed an increase in KCI flux for initial
transport experiments.) We repeated measurements of KCI, K2804, KaFe(CN)6,
BaClz, CaC|2, and MgClz fluxes twice in this order, with one final KCI
measurement. The last two KCI flux measurements differed by less then 10%,
so we utilized the second set of flux values to calculate selectivity. Additionally,
experiments were performed in which the flux of KCI was measured between
experiments with each of the above salts, and the incremental change in KCI flux
values did not indicate an effect from the salts used. The increase in flux during

conditioning is most likely due to swelling and reorganization of the polymers.

3.3 Results and Discussion
3.3.1 Deposition of PMDA-PDAIPAH Films on Al and Au. One challenge in
depositing PMDA-PDA/PAH films by APD is making a poly(amic acid) solution for

deposition. Poly(amic acids) are not water soluble as synthesized, so extra steps

53

are needed to make the polyelectrolyte solution. One method is to first dissolve
the poly(amic acid) in an organic solvent, such as DMF, and then dilute with
water.14 Another way to do this (without organic solvents) is to convert the
protonated poly(amic acid) into the readily water soluble salt form (Figure
3.1b).15'16 We accomplished this by simply dissolving the polymer in a solution
containing a base (0.15 M triethylamine), and then further diluting this mixture
with water. The solution properties, such as pH and salt concentration, can then
be adjusted to control the thickness of the film. We found that the PMDA-PDA
would precipitate out of a 0.5 M NaCl solution when the pH was lowered below
4.0. Therefore, we made these solutions at pH 4.5 to minimize the risk of non-
electrostatic polymer adsorption at the membrane surface.

We first investigated the alternating deposition of PMDA-PDA and PAH
directly on Al-coated Si wafers. These substrates are convenient to use for APD
because the oxide layer is net positively charged and therefore does not require
treatment to introduce charged groups on its surface. Film preparation begins
with an immersion into the poly(amic acid) solution, during which a layer of
PMDA-PDA attaches electrostatically to the oxide surface. After rinsing the
substrate with water, immersion into a PAH solution yields another layer (two
layers, PMDA-PDA/PAH, are referred to as one bilayer). Subsequent alternating
immersions into PMDA-PDA and PAH solutions produce a multilayer film.
Reflectance FTIR spectra demonstrate the layer-by-Iayer growth of these films,
as absorbances increase regularly with the number of deposited bilayers (Figure

3.3). Ellipsometric studies also indicate a linear growth for PMDA-PDA/PAH

54

 

9 bilayers

 

7 bilayers

5 bilayers

 
 
  
 

Absorbance

3 bilayers

‘ 1 bilayer

\‘Ib

 

 

 

 

3500 3000 2500 2000 1500 1000
Wavenumber (cm'1)

Figure 3.3: Reflectance FTIR spectra of 1, 3, 5, 7, and 9 bilayers of PMDA-
PDA/PAH on Al-coated Si wafters.

55

films with a 26 A increase in thickness per bilayer (Figure 3.4). Heating of these
films at 150 °C reduces the bilayer thickness to 23 A.

We also investigated whether PMDA-PDA/PAH films will deposit on top of
a PSS/PAH coated substrate. We used Au-coated Si wafers modified with a
monolayer of MPA because the lower pH solutions employed in PSS/PAH film
deposition corroded the Al-coated substrates. Deprotonation of the carboxylic
acid groups of MPA yields a negatively charged surface, so deposition began
with PAH. The ~170 A thick PSS/PAH films (4.5 bilayers) terminate with PAH, so
PMDA-PDA/PAH deposition occurred as described above. Figure 3.5 shows the
separate spectra of both PMDA-PDA and PSS/PAH films, and a spectrum of a
composite PSS/PAH + PMDA-PDA/PAH film.

Heating PMDA-PDA under N2 yields a polyimide membrane (Figure 3.1b).
To examine the extent of imidization, we prepared PAH/PSS (4.5-bilayers) +
PMDA-PDA/PAH films on gold supports and measured their reflectance FTIR
spectra. Figure 3.6 shows spectra of films containing a 2.5-bilayer top coat of
PMDA-PDA/PAH before and after heating for 2 h under a N2 atmosphere.
Unheated films have a broad absorbance band from 1680 to 1520 cm", primarily
due to the overlap of the amide and asymmetric —COO’ stretches of the
poly(amic acid). The symmetric -COO' stretch appears at 1407 cm". Heating of
these films results in the appearance of asymmetric and symmetric imide
carbonyl peaks (1775 and 1730 cm", respectively) and the disappearance of
amide and carboxylate peaks. Higher heating temperatures result in higher

intensities of imide peaks, showing that one can control the extent of imidization.

56

Ellipsometric Thickness (A)

250

 

5

.1.

01

O
1

 

'8'

 

50*

 

 

 

Number of Bilayers

Figure 3.4: Ellipsometric thickness and absorbance at 1580
cm'1 of PMDA-PDA/PAH films on Al-coated Si wafers as a

function of the number of bilayers in the film.

57

 
  
  
  

eouquosqv 3|

Absorbance

 

  

Composite
film

    

 

 

PSSI PAH

PM DA-PDAIPAH

 

 

 

I T T l

3500 3000 2500 2000 1500 1000
Wavenumber (cm'1)

Figure 3.5: Reflectance FTIR spectra of 2.5 bilayers of PMDA-
PDA/PAH, 4.5 bilayers of PSS/PAH, and the composite film (same
number of bilayers for each) deposited on gold-coated substrates
modified with MPA.

58

Absorbance

 

 

 

 

.\
I.
.‘ ............ Unheated
0.02 lj ————— 130°C
1 | 150°C
.‘j —-—.—- 180°C
l i '
. i 0.. o...o
I\. ,- :
l ’ :
l" g
j I":
”/‘J\ ’ :
l / / .
I I \\,/ _ 1.
l 7! _I\ t, .
iwpvfi ./'\ '°,,.°".... ::I’\JP:\ a...
/ A) '/ \. .\".:' I \"
I‘ ’v/ “of -\’f..,-“"\.
'\/ .J x...
‘.--'
1 800 1 700 1 600 1 500 1400 1 300

Wavenumbers (cm'1)

Figure 3.6: Reflectance FTIR spectra of films containing
a PAH/PSS base of 4.5 bilayers and a PMDA-PDA/PAH
topcoat of 2.5 bilayers before and after heating for 2 h at

several temperatures. Films were deposited on gold-

coated substrates modified with MPA.

59

3.3.2 Coverage of porous alumina. The first challenge in forming ultrathin
polyelectrolyte membranes is to show that they completely cover a porous
substrate without filling underlying pores. Previous FESEM studies of 5-bilayer
PSSIPAH coatings attached electrostatically to porous alumina supports showed
complete pore coverage with little penetration of polymer into the underlying
cavities.26 Figure 3.7 shows examples of typical top-down and cross-sectional
FESEM images of porous alumina coated with 5 bilayers of PSSIPAH and
capped with PMDA-PDA/PAH. Top-down images (Figure 3.7a) suggest
complete pore coverage (no pores are visible), and since the film deposition was
restricted to the top-side, images of the uncoated alumina backside show
completely open pores (not shown). There also appear to be cracks in the
membrane surface, but these are likely a result of the gold that was sputter-
coated onto the surface to provide a conductive layer for imaging or contraction
of the film in the SEM vacuum. Cross-sectional images (Figure 3.7b) show 40 to
50 nm-thick films and, this and other images suggest that films are‘relatively
uniform across the support. These images show very little, if any, deposition of
polyelectrolytes on the walls of the 0.2-um-diameter pores that make up the
majority of the porous alumina.

3.3.3 PSSIPAH base layer. Previous diffusion dialysis studies on PSSIPAH
films show little difference in the flux of singly charged ions through film-coated
and bare alumina supports. The porous alumina substrate is effectively covered

with as few as 4.5 bilayers of PSS/PAH (anionic top layer), but KCI flux only

60

4g

500nm
Ill||,l||»|l|

Polymer film

 

Figure 3.7: Top-down (a) and cross-sectional (b)
FESEM images of porous alumina coated with 5
bilayers of PSSIPAH and capped with 2.5
bilayers of PMDA-PDA/PAH.

61

decreases by 25% relative to the bare support. Addition of more bilayers has
little effect on the flux, as 9.5-bilayer films reduce KCl flux by only 33%. In the
case of K2804, flux drops by ~80% compared to bare alumina for both 4.5 and
9.5 bilayer PSSIPAH films. The difference in KCI and K2804 fluxes results in a
CI'ISO42' selectivity of 5 to 7. The relatively small changes in flux that occur
when doubling the thickness of the film suggest that Donnan exclusion near the
surface is the major factor in the selectivity. The excess charge at the surface
hinders transport of divalent ions whose charge is of the same sign as the
surface charge, and this explains the larger flux for singly charged Cl' relative to
doubly charged 8042'.

The Donnan exclusion model is also consistent with the fact that if the
outer layer is changed to the positively charged PAH (changing from a 4.5 to 5
bilayer membrane) the K2804 flux increases by 120%. Therefore, we think these
PSSIPAH films have a large free-volume that permits high fluxes, while their
highly charged surface allows for Donnan selectivity between single and multiply
charged ions. Thus, when using PSSIPAH as a gutter layer, the polyimide
portion of the membrane will still be the dominant factor in controlling transport,
as PSSIPAH provides little resistance to mass transport . A similar approach
was previously used to form a thin, discriminating skin layer of poly(acrylic
acid)/PAH on top of PSSIPAH.18
3.3.4 Unheated PMDA-PDAIPAH membranes. When membrane skins are
composed entirely of imidized PMDA-PDA/PAH (9.5 bilayers), salt flux is about 2

orders of magnitude lower than that through bare alumina supports. These

62

results are similar to those for heated PAA/PAH membranes.18 Therefore, we
used PSSIPAH as a highly permeable gutter-layer to allow for a reduction in the
polyimide film thickness. Since the pores in the alumina are already covered by
PSSIPAH, only a very thin skin of PMDA-PAH/PAH is needed form a defect-free,
selective coating that controls transport properties.

We prepared alumina membranes coated with 5 bilayers of PSSIPAH
(terminating with PAH) and capped these systems with 1.5, 2.5, or 3.5 bilayers of
PMDA-PDA/PAH (terminating with PMDA-PAH). The KCI flux for these
unheated membranes is essentially the same as that through the PSSIPAH
gutter-layer and is independent of the thickness of the PMDA-PDA/PAH film
(Table 3.1).18 The CI‘ISO42' and CI'lFe(CN)63' selectivities are also similar to
those of PSSIPAH films.“"26 (CI'lFe(CN)53' selectivities are high because the
Fe(CN)53‘ is large and triply charged). An intriguing aspect of unheated
membranes with PMDA-PDA/PAH layers is that they are highly cation selective
(Table 3.2). The K“/Mg2+ selectivity value for unheated films is relatively high at
120, and somewhat lower for K“/Ca2+ and K”/Ba2+ at about 40 and 27,
respectively. This value for K*/Mgz+ selectivity is comparable to that achieved
previously with a 60 bilayer PSSIPAH membrane.22 Cation selectivity seems to
also be independent of the number of PMDA-PDA/PAH bilayers. However,
capping an unheated film with an amine rather than the poly(amic acid) (3-bilayer
film, Table 3.2) does give a 2-fold increase in cation-transport selectivity. I
should note that other studies suggest that termination of films with a polycation

rather than a polyanion yields tighter films.”28

63

.=mm .2 rd 859:8 among 858 on... a
.A 9 8:239 E2: F..w.~.Eo.m._oE so? x md v6 5: -_0 mEEEm msocod m

 

 

 

on door do» 03 - 09 mm mod Nd Nd o om?
HdVN H83 H82. H05 H00: H03 Hmd Hoe Hod o
05 oomv comm dmm - dNN mm Pd d d Nd 0 mm:
H 003 H ooov H OONm H door H ONm H 08 H me H dN H wN 0
do? own ddv dNF MN m? w H 5 ad m o md o 03
H eddN H ddmN H ddN_. H d9. H can H 08 H N.N H mN H EH 0
dNVH o: 3 av - . I . 1 rd d d md
82 H82 H2: H2: mm+v~3+9 Him Ham H? 0.02
com dot ooN . l Nd l . Nd md md Nd
H 03 H 88 H 93 m o + o H 3 T. H v H 3. H on H mm H man 8625
996.5 998.5 996.5 whim? 226.5 993.5 896.5 926.5 $38.5 $38.5
md mN me md m mN m... md mN me .anP
95mm:

-szovwnio .s

H.820 .5

7008920.: so?

xzc 5x

 

«1528.35 Ho eggs 3 2 3 £5, 8&8 can 38 :Eamn. 5.6.55 m es,
858 «9:835 328 50:95 m_w>_m_u 563:6 .2 £56580 33:00.3 :25 new 0.93.“. n...» 53m...

64

odem mev vKHmm

VN

H03. 9me m+nm

NHONF mHFm VHVN

Qm Hom NNH NN m.m H 3

9.8 23.8.50 .2 rd 5.3 0958.9... 9oz, 3:08:33 ..< m

2H3 HHS ~va :Hoo - «HHS 2H; 0.2:
«How? SHE mHmm Hwy - Haunt Haw: 00mm:
mHooH NHNV mHR HM? mHo? me: «~va 0.02
mHHmm HNHR “How mefl - Hwy 2H5 0.02

mm Em 5 NF

 

 

 

2.me NKHdN FHmN :Hmv dedm ddem Ho: HoFN HdNF HdNF cmumocc:
998.5 998.5 936.5 996.5 228.5 293.5 298.5 398.5 996.5 996.5
md mN me md mN me md m mN me .aEm...
manna...

@me .s

+N.85. a H.522 .s

 

..:<n=<on.-<o§n_ .0 236.3 on 9. 3 EE 5880 us... 38 :5.me .235

-m m 5.3 858 mEEBm 390a .3395 m_m>.m_n 87.2% .5. 35.0580 3.2623 cozmo "Nd was...

65

3.3.5 Heated membranes. Heating of membranes converts the poly(amic acid)
to the polyimide (Figure 3.1b), and the extent of imidization has a large effect on
the membrane selectivity and flux. Tables 3.1 and 3.2 show the selectivity
coefficients, a, for membranes top-coated with 1.5, 2.5, 3, and 3.5 PMDA-
PDA/PAH bilayers that were heated at various temperatures. Heating
membranes increases CI’ISO42' selectivities to values as high as 1100. For
comparison, Neosepta membranes coated with a single layer of polyanion
showed a maximum Cl'ISO42' selectivity of 20.29 In contrast, the K“/Mg2+
selectivity is relatively high (120) for unheated films, but heating does not
increase this selectivity to the same degree as seen for anions. Figure 3.8 shows
typical normalized conductivity vs. time plots for KCI, K2804, and MgClg.

Donnan exclusion due to surface charge, which is a major factor in the

182236 is not the main reason for

selectivity of some polyelectrolyte membranes,
the large selectivity values in the present case. This can be seen by changing the
sign of the outer-layer charge through termination with PAH rather than PMDA-
PDA (Tables 3.1 and 3.2, 3-bilayer PMDA-PDA/PAH). This change in sign has
only a small effect (at most a factor of 2) on selectivity with either cations or
anions.

The selectivity among ions most likely reflects differences in hydration
energies and ion sizes. Both hydration energy and hydrated ion radius decrease
in the order Mg“ > Ca2+ > [.an+ > 8042' > or > K“ (Table 3.3).30 Either loss of

waters of hydration upon partitioning into the membrane or slower diffusion of

large hydrated ions would result in K"/Mg2+ and Cl'/SO42' selectivity. The K*/Mg2+

66

Table 3.3: Ionic and hydrated radii and hydration energies of
the ions used in the diffusion dialysis experiments?

 

lon Ionic Radius Hydrated Radius Hydration Energy

 

cr 181 223 340
5042' 240 278 1080
Fe(CN)53‘ 380 396 -
K+ 138 212 295
Mg2+ 72 299 1830
Ca2+ 100 271 1505
Be” 136 254 1250

a Data from reference 30.

67

Normalized Conductivity

 

 

 

 

 

 

 

 

.-
db

0 10 20 3O 40 50
Time (min)

Figure 3.8: Plots of normalized receiving phase conductivity vs. time for

diffusion dialysis experiments in which a heated 2.5-bilayer PMDA-

PDA/PAH + 5-bilayer PSSIPAH membrane separates a 0.1 M source

phase from the receiving phase. The inset shows MgCl2 and K2804 plots

magnified by 2 orders of magnitude.

68

selectivity should also be greater than that for CI'ISO42', as is seen for unheated
films, because K”/Mg2+ hydration energy differences are greater than those of Cl”
ISO42'. Selective transport of K” over Ba2+ or Ca2+ also occurs through unheated
membranes, but these selectivity values are only about 30% of those for K”/Mgz*,
as would be expected from trends in hydrated radii and hydration energies.

After heating and imidization, the membrane should become denser and
more hydrophobic. This would explain the increase in Cl'lSO42' selectivity upon
heating. However, we would expect the K*/Mg2+ selectivity to increase with
heating to an even greater extent. We speculate that Mg2+ may be diffusing
through the membrane with only a few waters of hydration, and thus increases in
film density may not slow its diffusion as much as that of 8042', which has a
larger unhydrated radius.

The selectivity depends greatly on the heating temperature applied. A plot
of the flux vs. heating temperature (Figure 3.9) shows that the 8042' flux
decreases rapidly with heating temperature up to 150 °C and levels off after this
temperature. For Cl', flux usually decreases over the entire heating range we
tested. The selectivity, therefore, reaches a maximum point at a particular
heating temperature and decreases at higher heating temperatures because Cl'
flux is further reduced with higher heating temperature while 8042' flux is
constant.

Membrane selectivity also varies with the number of PMDA-PDA/PAH
bilayers. Adding more bilayers appears to decrease the heating temperature at

which the onset of high CI‘ISO42' selectivity occurs. Due to interpenetration of

69

5e-8

 

4e-8 .

 

2e-8 - “*5-

+ 1.5 PMDA—PDA/PAH
16-8 1 + 2.5 PMDA-PDA/PAH
+ 3.5 PMDA—PDA/PAH

 

 

 

T

 

(DO

   
   

K2804

+ 1.5 PMDA-PDA/PAH
+ 2.5 PMDA-PDA/PAH
+ 3.5 PMDA-PDA/PAH

Flux (mole cm'2 s“)

c»
‘P
co

2e-9 .

 

 

 

0 130 150 155 180
Heating Temperature

Figure 3.9: Plots of the diffusion dialysis flux of KCI (top) and
K2804 (bottom) as a function of the imidization temperature
of composite membranes consisting of 5 bilayers of
PSSIPAH and a capping layer of 1.5 (circles), 2.5 (squares),
and 3.5 (triangles) bilayers of PMDA-PDA.

7O

polymers, the 1.5-bilayer PMDA-PDA/PAH top-coat likely contains some PSS
and may be less selective than thicker top-coats. Remarkably, even a 2.5-bilayer
PMDA-PDA/PAH top-coating that is only 70 A thick can give Cl'/SO42'
selectivities of over 1000 after heating at 180 °C. Selectivities as high as 680 for
Cl'lSO42' can be achieved with Cl- flux that is still 50% of that through bare

alumina.

3.4 Conclusion.

In summary, ultrathin polyimide membranes can be formed on porous
alumina supports coated with a polyelectrolyte gutter layer, and these
membranes exhibit extremely large selectivity coefficients for separation of
monovalent and divalent ions. The selectivity varies with both the thickness of
the selective skin layer and the extent of imidization. Also, these polyimide
membranes are selective for both anions and cations. This phenomenon may be
best explained by the differences in the hydration energies and sizes of the ions,

as well as a small amount of selectivity due to the negative membrane surface.

71

3.5 References.

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

(6)

(7)

(8)
(9)
(10)
(11)
(12)

(13)

(14)

(15)

(15)

(17)

Lipnizki, F.; Field, R. W. Environ. Prog. 2002, 21, 265-272.
Koros, W. J. Chem. Eng. Prog. 1995, 91, 68-81.

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

Loeb, S.; Sourirajan, S. Adv. Chem. Ser. 1963, 38, 117-132.

Kim, S.-G.; Kim, Y.-l.; Yun, H.-G.; Lim, G.-T.; Lee, K.-H. J. Appl. Polym.
Sci. 2003, 88, 2884-2890.

Pinnau, l.; Freeman, B. D. In Membrane Formation and Modification;
Pinnau, l., Freeman, B. D., Eds.; American Chemical Society:
Washington, DC, 2000, pp 1-22.

Polotskaya, G. A.; Kuznetsov, Y. F.; Goikhman, M. Y.; Podeshvo, l. V.;
Maricheva, T. A.; Kudryavtsev, V. V. J. Appl. Polym. Sci. 2003, 89, 2361-
2368.

Decher, G. Science 1997, 277, 1232-1237.

Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213-4219.
Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153-8160.
Huang, R. Y. M.; Feng, X. J. Membrane Sci. 1993, 84, 15-27.

Clausi, D. T.; Koros, W. J. J. Membrane Sci. 2000, 167, 79-89.

Sroog, C. E. In Polyimides; Wilson, D., Stenzenberger, H. D.,
Hergenrother, P. M., Eds.; Chapman and Hall: New York, 1990, pp 252-
284.

Baur, J. W.; Besson, P.; O'Connor, S. A.; Rubner, M. F. Mater. Res. Soc.
Symp. Proc. 1996, 413, 583-588.

Liu, Y.; Wang, A.; Claus, R. 0. Appl. Phys. Lett. 1997, 71, 2265-2267.

Moriguchi, l.; Teraoka, Y.; Kagawa, 8.; Fendler, J. H. Chem. Mater. 1999,
11, 1603-1608.

Harris, J. J.; Bruening, M. L. Langmuir2000, 16, 2006-2013.

72

(18)

(19)

(20)
(21 )
(22)
(23)

(24)

(25)

(26)

(27)

(28)

(29)

(30)

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

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

Levasalmi, J.-M.; McCarthy, T. J. Macromolecules 1997, 30, 1752-1757.
Kotov, N. A.; Magonov, S.; Tropsha, E. Chem. Mater. 1998, 10, 886-895.
Krasemann, L.; Tieke, B. Langmuir 2000, 16, 287-290.

Krasemann, L.; Tieke, B. Chem. Eng. Technol. 2000, 23, 211-213.

Li, 0.; Yamashita, T.; Horie, K.; Yoshimoto, H.; Miwa, T.; Maekawa, Y. J.
Polym. Sci, Part A: Polym. Chem. 1998, 36, 1329-1340.

Sroog, C. E. J. Polym. Sci, Macromol. Rev. 1976, 11, 161-208.

Harris, J. J.; Stair, J. L.; Bruening, M. L. Chem. Mater. 2000, 12, 1941-
1946.

Smith, R. N.; Reven, L.; Barrett, C. J. Macromolecules 2003, 36, 1876-
1881.

Stanton, B. W.; Harris, J. J.; Miller, M. D.; Bruening, M. L. Langmuir2003,
19, 7038-7042.

Sata, T.; Yamaguchi, T.; Matsusaki, K. J. Membrane Sci. 1995, 100, 229-
238.

Marcus, Y. Biophys. Chem. 1994, 51, 111-127.

73

Chapter 4

Ultrathin, Gas-selective Polyimide Membranes Prepared from Multilayer

Polyelectrolyte Films

4.1 Introduction

Polyimides are attractive membrane materials because of their
mechanical and thermal stability,"2 and compared with other polymers, certain
polyimides exhibit a favorable combination of permeability and gas-transport
selectivity.3'9 Practical application of these materials, however, requires
formation of a defect-free polyimide layer on a highly permeable support, and this
layer should be as thin as possible to allow high flux through the membrane.
Methods for creating ultrathin polyimide skins include phase-inversion processes
and formation of composite membranes by deposition of polyimide films on
highly permeable substrates. For example, Koros and coworkers prepared cast-

‘0'11 and hollow-fiber polyimide membranes12 by a phase-inversion process

sheet
that yields a dense, selective layer at the surface of a porous material.

Composite membranes offer the advantage that only a thin, selective film
of the separation material is used, and thus more expensive, and hopefully more
selective, polymers can be employed.13 In the case of polyimides, composite
membranes can be prepared by deposition of a poly(amic acid) followed by

imidization to form a selective film. Ding and coworkers coated porous

polysulfone hollow fibers with a solution of a poly(amic acid) salt and showed that

74

 

heating the fibers at 150 °C for 24 h yields fully imidized films that are as thin as
50 nm.14

Even with these successes in creating ultrathin membranes, deposition of
defect-free polyimide skins that are less than 50 nm thick is still difficult.‘5'16 New
methods for forming ever-thinner skins will allow creation of membrane systems
with higher fluxes, and the recently developed technique of alternating
polyelectrolyte deposition (APD) may prove useful in this regard. This method
simply involves alternating immersions of a charged substrate into oppositely
charged polyelectrolytes,17 and thus film thickness is easily controlled by
changing either the number of adsorbed polyelectrolyte layers or deposition
variables such as pH, supporting salt concentration, and solvent.”‘20 Because
the films self-assemble, surface roughness does not greatly affect adhesion, and
defect-free films can be obtained on porous supports since each added layer
covers defects in the previous layer.”21

A few recent studies demonstrate membrane formation using APD on
permeable supports. Successful separation of monovalent and divalent ions with
such membranes occurs with various polyelectrolytes and deposition

22-24

conditions, and highly selective separations by pervaporation are also

25"” In contrast, gas-transport selectivities due to multilayer

possible.
polyelectrolyte films are relatively modest. Polyelectrolyte films can increase the
gas selectivity of non-porous supports,28'30 but deposition of these films on

porous supports yields little increase in selectivity?“31 We suspect that the lack

75

 

of gas-transport selectivity in polyelectrolyte films is inherent in the structure and
packing of the polyelectrolytes thus far employed.

In this work, we utilize poly(amic acids) as polyelectrolytes for APD so that
these materials can be imidized to produce polyimides that are known to exhibit
high gas-transport selectivities and perrneabilities. Several previous studies
combined APD of poly(amic acids) and a polycation with postdeposition heating
to form polyimide films or ion-selective membranes,"’2'3""35 but this chapter
represents the first utilization of this method to form membranes for gas
separations. Ionized precursor poly(amic acids) serve as polyanions that form
films with poly(allylamine hydrochloride) (PAH) on porous alumina36 (Figure 4.1),
and subsequent heating yields the polyimide membrane. We chose to study the
three polyimides shown in Figure 4.2 because their gas-transport properties were
previously investigated.7‘9 (In this chapter, we use the acronyms in Figure 4.2 to
refer to both the imidized and poly(amic acid) forms of the polymers.) One of the
main points that we demonstrate in this study is that the presence of a small
amount of PAH in the membrane does not alter the selectivity of the polyimide
films. Gas-transport studies clearly demonstrate that 35 to 50 nm-thick polyimide
films prepared using APD have selectivities and perrneabilities that are the same
as those of the corresponding bulk polyimides. Thus APD indeed provides a new

tool for development of ultrathin, gas-selective membranes.

76

 

FHCCF

i©1

n NH3+ mag;t

 

 

 

 

 

Figure 4.1: Heat-induced imidization of a poly(amic acid)/PAH film on a

porous support. Neutralization of PAH occurs when it contributes a proton for
the formation of water. Intertwining of neighboring layers is not shown for

figure clarity.

77

F3C CF3

O F3C CF3 O
CF3
6FDA-BAHF {an
CF3 n
O O

o 0 o
CF3
BTDA-BAHF N O O N O O
cr=3
0 0

Figure 4.2: Structures of the polyimides used in this study.

78

4.2 Experimental Section

4.2.1 Materials. 4,4’-(hexafluoroisopropylidene)diphthalic anhydride (6FDA),
3,3’,4,4’-benzophenonetetracarboxylic dianhydride (BTDA), 4,4’-
(hexafluoroisopropylidene) dianiline (BAHF), 1,3-phenylenediamine (mPDA), (3-
aminopropyl)triethoxysilane and poly(allylamine hydrochloride) were purchased
from Aldrich. DMF (Spectrum) was dried with molecular sieves for at least 24 h,
and the 3,3’,4,4’-benzophenonetetracarboxylic dianhydride was purified by
vacuum sublimation prior to use. All other chemicals were used as received.
Acetone (Vorpak), ethanol (Pharrnco, 200 proof) and deionized water (Milli-Q,
18.2 MQcm) were used for rinsing and preparation of polymer-containing
solutions. Porous alumina supports were 25 mm Whatman Anodiscs with 0.02
pm-diameter surface pores (Fisher Scientific), and silicon(100) wafers (Silicon
Quest) were used as supports for ellipsometry (single-side polished) and
Brewster-angle transmission FTIR spectroscopy (double-side polished).

4.2.2 Substrate Preparation. Prior to film deposition, silicon wafers were
silanized with (3-aminopropyl)-triethoxysilane to prepare a controlled surface
capable of supporting a positive charge. Silanization was accomplished using
the method of Petri et al.37 In the case of porous alumina supports, the
polypropylene support ring on the alumina was removed to prevent it from
melting into substrate pores during heat-induced imidization. This was
accomplished by cutting off as much of the polymer as possible with scissors,

and then burning off the remaining ring at 400 °C for 18 h. Subsequently, the

79

alumina supports were rinsed with acetone, dried with N2, and cleaned for 10 min
in a UV/ozone cleaner (Boekel Industries, model 135500). Film formation
occurred directly on the alumina, so silanization was not necessary.

4.2.3 Polymer Synthesis. The poly(amic acids) were prepared using a typical
literature procedure.38 First, the diamine was dissolved in an appropriate solvent
(acetone for 6FDA-BAHF and DMF for both 6FDA-mPDA and BTDA-BAHF). A
stoichiometric quantity of the dianhydride was then added to this solution
gradually over a ~15 min period, and the 15% by weight (total dissolved solids)
solution was stirred for 24 h. The poly(amic acids) were precipitated twice in
either 1:1 hexanes/ethyl acetate (6FDA-BAHF and BTDA-BAHF) or 3:1
chloroform/hexanes (6FDA-mPDA), filtered, and then dried under vacuum for 24
h.

4.2.4 Solution Preparation and Film Deposition. Due to the insolubility of the
poly(amic acid) in water, solutions containing 0.005 M 6FDA-BAHF and 0.5 M
NaCI were prepared by dissolving 0.039 g of 6FDA-BAHF in 7 mL of ethanol, and
then adding 3 mL of 1.67 M aqueous NaCl. The molarities of the polyelectrolytes
are given with respect to the repeating unit. The pH was adjusted to 5.0 by
adding 0.2 M triethylamine in 7:3 (vzv) ethanoVwater. Film deposition began with
immersion of the substrate (silanized silicon or porous alumina) into the
poly(amic acid) solution for 3 min. (The alumina membrane was placed in an o-
ring holder so that only the face with 0.02 pm-diameter pores was exposed to
solutions.) The substrate was removed from solution and rinsed with copious

amounts of ethanol and then Milli-Q water. Subsequently, the substrate was

80

immersed in an aqueous 0.02 M PAH solution (0.02 M NaCI, adjusted to pH 5.0
with NaOH) for 3 min and rinsed with Milli-Q water and then ethanol. This
process was repeated until the desired number of bilayers was deposited, and all
films were terminated with a poly(amic acid) layer. 6FDA-mPDA/PAH films were
prepared similarly except that the rinsing solvent was a 1:1 mixture of ethanol
and water, and the PAH solution did not contain NaCI. BTDA-BAHF solutions
were made by first dissolving BTDA-BAHF in acetone, and then adding water to
make 0.005 M BTDA-BAHF in 7:3 (v:v) acetone/water. The pH of this solution
was adjusted to 5.0 by adding 0.2 M triethylamine in 7:3 acetone/water. For
BTDA-BAHF/PAH films, the poly(amic acid) layers were rinsed with acetone and
neither solution contained NaCI.

4.2.5 Film lmidization. Films were heated using a home-built apparatus
consisting of a temperature controller, thermocouple, glass chamber, and heating
mantle. The heating apparatus was purged with nitrogen for 2 h prior to heating
as well as during heating and cooling. The temperature was ramped at a rate of
5 °C per min and was held at the desired heating temperature for 2 h.

4.2.6 Film Characterization. Ellipsometric thickness measurements on amine-
tenninated silicon wafers were made using a rotating analyzer ellipsometer
(model M-44, J. A. Woollam) and WVASE32 software. Approximate film
compositions were determined by measuring the change in thickness upon
adsorption of each polymer and assuming that each polyelectrolyte layer had the
same density. (For Iayer-by-Iayer monitoring of growth, films had to be dried with

N2 after deposition of each layer, but in all other cases films were dried only after

81

deposition of the entire film.) A film refractive index of 1.5 was assumed in all
thickness determinations. Transmission FTIR spectra of films on double-side
polished Si wafers were measured using a Nicolet Magna-550 FT IR
spectrophotometer with a Brewster angle attachment set at 759 (p-polarized
light). A UV/Oa-cleaned Si substrate was used as a background. Field-emission
scanning electron microscopy (FESEM) images were obtained with a Hitachi S-
4700 instrument using an acceleration voltage of 15 kV. Samples were fractured
in liquid nitrogen and sputter-coated with 5 nm of Au before imaging.

4.2.7 Gas-Transport Studies. Room-temperature (23-24 °C) gas-transport
measurements were carried out using a home-built permeation cell containing of
a chamber that sealed onto a membrane via an o-ring. The membrane was
supported by a stainless-steel frit (Mott), and gas flux was measured using a
soap-bubble flow meter (Fisher Scientific, model 420). Using a pressure relief
valve, the permeation cell was purged many times with the gas of interest and all
measurements with that gas were then taken after a steady-state flux was
observed. Fluxes were measured at pressures of 5 — 45 psig (in 5 psig
intervals), and the area of the membrane exposed to the gas was 2.0 cm2. The
fluxes of 02, N2, H2, CH4, and CO2 (in that order) were determined for each
membrane, and O2 and N2 were then tested again to insure that membrane
properties did not change upon exposure to different gases. Penneabilities of

the individual gases were calculated for each pressure and then averaged.

82

4.3 Results and Discussion

4.3.1 Film Formation and Composition. The selective layer of gas-separation
membranes usually consists of a single, gas-selective polymer,39 while formation
of membranes by APD, in contrast, requires two oppositely charged polymers. In
the present case, the polyanion is a poly(amic acid) that can be heated to form a
gas-selective polyimide, while the cation, PAH, is probably not a selectively
permeable material. Thus formation of gas-selective polyimide/PAH membranes
will likely require keeping the amount of PAH in the film to a minimum. One of
the goals of this study was to obtain multilayer polyelectrolyte films that contain
as much as 90% polyimide. We wanted to make the poly(amic acid) layer as
thick as possible to both reduce the number of layers that need to be deposited
and increase the overall fraction of polyimide in the final film.

Poly(amic acid) layer thickness can be controlled by varying the solvent
composition, deposition pH, and supporting salt concentration in deposition
solutions. The fluorinated poly(amic acids) employed in this study must first be
dissolved in a good solvent (e.g. ethanol or acetone) and then some water can
be added to the solution to enhance the deprotonation of -COOH groups.
Addition of too much water, however, results in precipitation of the polymer if it is
not fully deprotonated. Thus we deposited films from solutions containing 30%
water. Although the addition of salt to polyelectrolyte deposition solutions
increased the thickness of some multilayer polyelectrolyte films by as much as an

19,40

order of magnitude, in the poly(amic acid) case, film thickness increased by

83

at most 50% with the addition of up to 0.5 M NaCI. Although small, this
difference can increase the overall fraction of poly(amic acid) in the film by 5-
10%. Unfortunately, added salt can also change the solubility of the poly(amic
acid). This was most apparent for solutions of BTDA-BAHF. When these
solutions contained 0.5 M salt, they quickly became cloudy if any of the solvent
evaporated. Without the added salt, the solution would stay clear for over two
hours if left uncapped. Thus BTDA-BAHF was deposited in the absence of salt
while 6FDA-BAHF and 6FDA-mPDA were deposited in the presence of 0.5 M
NaCI.

For weakly acidic or basic polyelectrolyte systems, adjusting the
deposition pH can also have a large effect on layer thickness.2°"‘1 In the case of
poly(amic acids) lower pH values should decrease the fraction of acid groups that
are ionized and reduce the number of charged groups per polymer chain. With
fewer available charged sites, more polymer strands are required to compensate
the charge on the substrate surface, and hence more polymer will be deposited.
20"" For fluorinated poly(amic acids), pH also affects film thickness by altering
polymer solubility. Because deposition solutions are composed of both a good
solvent (ethanol or acetone) and water, the polymer is soluble in both the fully
protonated and the fully deprotonated forms. However, at intermediate pH
values, fluorinated poly(amic acids) are partially protonated and only sparingly

soluble. At points near solubility limits, the driving force for film deposition is

greatest,19 and the thickest poly(amic acid) layers form. Thus we deposited

84

poly(amic acid) films at a pH of 5, because this is the lowest intermediate pH at
which polymers reproducibly remain in solution.

The same thought process applies to deposition of PAH, except that we
wanted to deposit as little PAH as possible. In this case, the ideal solution might
be one that contains the polycation dissolved in a very good solvent, water, at a
low pH in the absence of supporting electrolyte. 20"" However, at some point the
amount of PAH deposited is too small to permit deposition of the next poly(amic
acid) layer. Thus for controlled film formation with minimal PAH adsorption, we
adjusted the pH of the PAH solution to be the same as that of the poly(amic
acid). Also, in the case of films containing 6FDA-BAHF, a small amount of NaCl
(0.02 M) was necessary to form layers with consistent thicknesses.

Overall, control over pH and solvent yielded bilayer thicknesses of 34, 38,
and 50 A for 6FDA-BAHF/PAH, BTDA-BAHF/PAH, and 6FDA-mPDA/PAH,
respectively. Increases in film thickness after adsorption of PAH were about 11%
of those resulting from adsorption of poly(amic acid). Thus, control over
deposition conditions allowed formation of films that were ~90% poly(amic acid).
4.3.2 Film lmidization. To examine the extent of imidization after heating, we
measured the Brewster-angle transmission FTIR spectra of poly(amic acid)/PAH
films on amine-modified double-side polished Si supports. Figure 4.3 shows the
spectra of films before and after heating at 150 or 250 °C. Unheated films have a
broad absorbance band from 1670 to 1520 cm’1 primarily due to the overlap of
amide and asymmetric -COO' stretches of the poly(amic acid). Heating of these

films results in the appearance of asymmetric and symmetric imide carbonyl

85

 

 

 

 

0.02 {‘1 _ """""" unheated
1I ------ Heated 150 °C r.
1‘ -—-— Heated 250 °C i ‘:|
I in"
a) II "\- lr' ?
o I l \ I"
g 1 l I r: =l .
.o - ' " /" ‘1. l
L, I \.. ,- '
8 I \ / \ , .I l
l! \ ./"\ \ .-/ ll
/ l| l‘ \ :4. / If»,
lMM ’4‘ l \..] \ I,
' \Afl’~’ \ «IN ‘ I ~
\ \VJV 12‘ \\ \\
\I

 

 

 

 

1800 1700 1600 1500 1400 1300
Wavenumbers (cm")

Figure 4.3: Brewster-angle transmission FTIR Spectra of 9.5-bilayer
6FDA-mPDA/PAH films on Si before and after heating at 150 °C or

250 °C for 2 h to induce imidization.

86

peaks (1775 and 1730 cm") and the reduction of amide and carboxylate peaks.
In principle, a few amide bonds could form by reaction of the amine groups of
PAH and the carboxylate groups of the poly(amic acid), but we see no spectral
evidence for this reaction.

Decreases in the carboxylate and amide peak intensities suggest that
~50% and ~85% imidization occur for samples heated at 100 °C (not shown) and
150 °C, respectively. Spectra obtained after heating at 250 and 300 °C were
essentially identical, implying that full imidization occurs at 250 °C. Thus
complete imidization takes place at temperatures lower than the glass transition
temperature of the polyimides (298 °C for 6FDA-mPDA, 298 °C'for 6FDA-BAHF,
and 305 °C for BTDA-BAHF),7'8 which is typically considered to be the
temperature needed to induce full imidization.42 Several studies show that
complexation of poly(amic acids) with an amine base lowers imidization
temperatures and decreases the time needed for full imidization.”43 Similarly,
we form poly(amic acid) salts when electrostatically attaching the layers, and
thus lower imidization temperatures occur. Low heating temperatures will prove
essential if poly(amic acid) films are to be imidized on temperature-sensitive
polymeric supports.

For most gas-transport studies, we employed heating temperatures of 250
°C to achieve full imidization. Temperatures above 250 °C present some risk of
decomposition of PAH, which may lead to membrane defects. After heating at
250 °C, film thickness decreases by ~30%, and this change is probably due to

both imidization and removal of adsorbed water from the film.

87

4.3.3 FESEM Studies of Membrane Formation. Figure 4.4 shows cross-
sectional FESEM images of porous alumina substrates before and after
deposition of 6FDA-mPDA/PAH films. These images show very little, if any,
deposition of polyelectrolytes on the walls of the 0.2-um pores that make up the
majority of the alumina support. There may be some film penetration (10 - 20
nm) into the 0.02 um pores at the alumina surface. Cross-sectional images
taken at several points along the diameter of the membrane indicate a uniform
film thickness, while top-view images of the membrane show complete coverage
of surface pores. The uncoated back side of the alumina exhibited open pores
with no evidence of polyelectrolyte deposition.

Using FESEM images we were able to estimate polyimide film thicknesses
to within ~5 nm. The thicknesses of 6FDA-mPDA/PAH (9.5 bilayers), 6FDA-
BAHF/PAH (14.5 bilayers), and BTDA-BAHF/PAH (10.5 bilayers) films were all
about 50 nm. These thicknesses are about 30% higher than ellipsometric
thicknesses of similar films deposited on amine-functionalized silicon wafers.
The higher thicknesses on the porous alumina support may be due to a small
amount of deposition in the surface pores or differences between charge
distributions and roughnesses of the two substrates. The similarity of
permeability coefficients determined using these thicknesses and permeability
coefficients of bulk polyimides suggests that the thickness values are reasonable
(see below).

4.3.4 Gas-Transport Measurements. Table 4.1 lists gas permeability

coefficients and the ideal O2/N2 and CO2/CH4 selectivities for polyimide

88

5:00 111 '
111111111;

'5.
A
.9
' .r.
' =5
’ )
- .
. I
'1

 

Figure 4.4: Cross-sectional field-emission scanning
electron microscopy images of (a) a bare alumina support
and (b) an alumina support coated with an imidized 9.5-
bilayer 6FDA-mPDA/PAH film.

89

0.. mm am 59369:...
0.. vm Cm 8586:...
2m:.Eo.m.NEov\Eo.E._.wv 880.23 0 20:86

 

 

me we 2 Rd 8 med 3 3 512.935 5:8
. . . . r . r . r . . r . .2225 no:
8 mm ~33 cool; m+mm 8380 3+3 cigar/5935
9. Z. 8 3 - on m: 3. 218-3% 5:8
mm 6.5 B 3 o: 3 E 5 5.1.3-365 55m
. r . r . r . .. . . r .2855 3.:
mm mm 6+8 moi: mg: mo+o~ 3+2 51.456.14.958
8 B we 2.0 ow 93 em 8 565-8% 5:8
. I . r . r . r . . r . .2965 we.
8 me I: 8320 5+9 80$; omoiwm 515.93.065-58
fontmoom NZnteoa NGen. fun. «In. ~21 ~01 5:50..

 

....<n. $07. 3.59:8 mmcmBEwE 8.8.28

.993 3:95 momma 30.29 Co toamcg 05 5.. 95.0508 3.2.06.3 .ucm 6.20:3. 5:58.503 “—3 036...

90

membranes formed by alternating polyelectrolyte deposition and imidization at
250 °C. Penneabilities are presented in barrers (1O‘°-cm3
(STP)-cm/(cm2-s-cm(Hg)) and were determined using film thicknesses obtained
by FESEM. Selectivities were calculated by dividing the perrneabilities of two
different gases. All of the membranes were made to be about 50 nm thick, and
all data shown represent an average of measurements on 3 replicate
membranes.

Within experimental uncertainty, the permeability and selectivity values of
the tested membranes are the same as those of cast, pure polyimide membranes
with um thicknesses. We should note that the transport experiments with the
membranes prepared by APD were performed at 24 °C, while the literature data
for bulk polyimides were acquired at 35 °C. However, the 10 °C difference in the
present case should have little effect on permeability.44 The high selectivities in
these membranes (Table 4.1) clearly show that the films are free of defects, as
even a small fraction of defects would negate selectivity.45 Annealing that occurs
during heat-induced imidization may help to heal defects in these films.

The agreement between the permeability of membranes prepared by APD
and analogous bulk polyimide membranes is somewhat surprising considering
that the films prepared by APD contain a mixture of the polyimide and PAH. One
would expect lower selectivities and different perrneabilities if the two layers were
as interpenetrated as is the case for other polyelectrolyte films.17
lnterpenetration may not be as prevalent in these films because each

polyelectrolyte is insoluble in the deposition solvent of the other polyelectrolyte.

91

Phase segregation could also occur during heating. Further experiments are
required to determine if phase segregation occurs.

4.3.5 Gas Transport as a Function of Film Composition and the Degree of
lmidization. To investigate the relationship between extent of imidization and
gas transport, we measured gas fluxes through 6FDA-mPDA/PAH membranes
before and after heating at 150 °C or 250 °C (Table 4.2). The unheated
membranes exhibited Knudsen diffusion-like selectivities and relatively low
perrneabilities for 02 and C02 (5- and 18—fold less, respectively, than through
membranes heated at 250 °C). Knudsen diffusion occurs in pores with diameters
smaller than the mean free path of the gas, and flux in such a system is inversely
proportional to the square root of the molecular mass of the gas.46 Thus
Knudsen diffusion selectivities suggest that transport occurs primarily through a
few small defects in unheated films. The relatively low flux through unheated
membranes may be due to the ionic cross-links in the film or the presence of
residual solvent.

Heating at 150 °C for 2 h increased selectivity relative to unheated films
(Table 4.2), but not to the level found in similar films heated at 250 °C. Fluxes of
02 and 002 were also 5-10-fold lower than for films heated at 250 °C. Increasing
heating time (150 °C) to 12 or 24 h did not yield increases in flux, but selectivity
appeared to increase. (The fluxes of N2 and CH4 through the films heated at 150
°C for 12 or 24 h were too low to be measured with our system, so we could only
determine limits for 02/N2 and COg/CH4 selectivities.) The differences between

membranes heated at 150 and 250 °C could result from a lower (10-15°/o) degree

92

.2... 03m... E9.— 9m?
36:: 26: 9: .6 :E: 9: 65 .950. 995 moxafin

 

 

93

muCQEmLDmmmE 025 mp: Cmmémfl wucmhmtfiv 0r:- .WMCNLDEGE O>>u CO wucmethmNGE *0 WQDNEG>MYMWNFVWWNND>M
mm as : Ed 93 em Iamwmcmwmwmflmwmmfld
B B 9 mmd med 3 I<wwmmwwwmmmmcmwmfimws
m. 3. 3 as as a; hawfimmvfimflwmma
3. 3 3 8.0 8.0 Bo I<fimmwwwmmfimmfim£md
8.0 one Ed mad mod 8.0 I<azoawwqmfinggs ma
VIDENOOQ NZQmOQ «001 fun Nzu ~01 o:9oEo_>.

 

.mommmcxofa ocm $59389 grow: E9ot6
5:5 9989: 8:93:68 I<n_\<on_E-<ou_o :8 mo:_m> 33:00.3 ucm mbzfimoEEa mow ”we aim...

of imidization, caged solvent that cannot be completely removed at 150 °C, or a
change in film structure (polymer segregation) at the higher heating temperature.
To better understand the effect of the polycation on gas transport, we
prepared 6FDA-mPDA membranes that contained a higher fraction of PAH (25 —
30%). This was accomplished by increasing the deposition pH for PAH to 9.0
and adding NaCl (0.5 M) to this solution. In relation the 6FDA-mPDA/PAH
membranes containing only 10% PAH and having the same total thickness,
02/N2 selectivity decreased from 6.9 to 2.2 and Cog/CH4 selectivity decreased
from 68 to 5.8, although the perrneabilities of 02 and 002 did not change by an
appreciable extent. This change in selectivity could result from a greater extent
of interpenetration between the layers due to the higher thickness of PAH. The
high pH of the PAH solution should also deprotonate the remaining -COOH
groups in the poly(amic acid) layer on the surface, allowing for a greater degree
of interpenetration.
4.3.6 Gas Transport as a Function of Film Thickness. Although ion-transport
studies suggested that as little as 25 nm of polyelectrolyte film is required to
cover the pores of alumina supports,47 the gas-selectivity of polyimide
membranes decreased for films less than 35 nm thick. When polyimide
thickness was reduced from 50 nm to 25-30 nm by reducing the number of
6FDA-mPDA/PAH bilayers from 9.5 to 5.5, 02/N2 selectivity decreased from 6.9
to 1.9 and COz/CH4 selectivity decreased from 68 to 4.1 (Table 4.2). Decreases

in selectivity result primarily from increases in N2 and CH4 flux. The presence of

94

some selectivity suggests that the density of defects is not high enough for
transport to occur simply by diffusion through pores, but we think that a few
defects result in greatly reduced selectivity by increasing N2 and CH4 transport.
By making the film two layers thicker (7.5 bilayers), we achieved the selectivities
similar to those of 9.5-bilayer films, presumably because defects are covered.
Thus we find that a minimum thickness of 35 — 40 nm is necessary to achieve the
highest selectivities. Even thinner selective films might be constructed if non-

porous supports or gutter layers are used.

4.4 Conclusion

Alternating electrostatic adsorption of poly(amic acids) and PAH followed
by post-deposition heating provides a convenient method for forming ultrathin,
gas-selective polyimide films at the surface of porous alumina. By controlling
deposition conditions, films can be tailored to contain primarily polyimide, and
fully imidized membranes exhibit permeability coefficients and selectivities that
are comparable to literature values for the corresponding bulk polyimides. These
selectivities can be achieved even with films as thin as 3540 nm. Future work
aims at further reducing the minimum film thickness required for high selectivity
and developing gentler imidization conditions that will allow use of polymeric

substrates.

95

4.5 References

(1)
(2)

(3)
(4)

(5)
(6)
(7)
(8)

(9)

(10)
(11)
(12)

(13)

Feng, X.; Huang, R. Y. M. J. Membrane Sci. 1996, 109, 165-172.
Sroog, C. E. In Polyimides; Wilson, 0., Stenzenberger, H. D.,
Hergenrother, P. M., Eds.; Chapman and Hall: New York, 1990, pp 252-
284.

Liu, Y.; Pan, 0.; Ding, M.; Xu, J. Polym. Int. 1999, 48, 832-836.

Robeson, L. M.; Burgoyne, W. F.; Langsam, M.; Savoca, A. C.; Tien, C. F.
Polymer 1994, 35, 4970-4978.

Koros, W. J.; Fleming, G. K. J. Membrane Sci. 1993, 83, 1-80.
Maier, G. Angew. Chem. Int. Edit. 1998, 37, 2961-2974.
Tanaka, K.; Kita, H.; Okano, M.; Okamoto, K. Polymer1992, 33, 587-592.

Tanaka, K.; Okano, M.; Toshino, H.; Kita, H.; Okamoto, K. J. Polym. Sci,
Part B: Polym. Phys. 1992, 30, 907-914.

Koros, W. J.; Walker, D. R. B. Polym. J. 1991, 23, 481-490.

Pinnau, |.; Koros, W. J. J. Appl. Polym. Sci. 1992, 46, 1195-1204.
Pinnau, l.; Koros, W. J. Ind. Eng. Chem. Fies. 1991, 30, 1837-1840.
Clausi, D. T.; Koros, W. J. J. Membrane Sci. 2000, 167, 79-89.
Pinnau, |.; Freeman, B. D. In Membrane Formation and Modification;
Pinnau, |., Freeman, B. D., Eds.; American Chemical Society:
Washington, DC, 2000, pp 1-22.

Ding, Y.; Bikson, B.; Nelson, J. K. Macromolecules 2002, 35, 912-916.

Chung, T. 8.; Tech, S. K.; Hu, X. D. J. Membrane Sci. 1997, 133, 161-
175.

Hachisuka, H.; Ohara, T.; lkeda, K. In Membrane Formation and
Modification; Pinnau, l., Freeman, B. D., Eds.; American Chemical
Society: Washington, D. C., 2000, pp 65-78.

Decher, G. Science 1997, 277, 1232-1237.

96

(13)

(19)
(20)
(21 )

(22)

(23)

(24)

(25)
(26)

(27)

(28)
(29)

(30)

(31)

(32)

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

Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153-8160.
Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213-4219.
Kleinfeld, E. R.; Ferguson, G. S. Chem. Mater. 1996, 8, 1575-1578.

Sullivan, D. M.; Bruening, M. L. J. Am. Chem. Soc. 2001, 123, 11805-
11806.

Krasemann, L.; Tieke, B. Langmuir2000, 16, 287-290.

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

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

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

Levasalmi, J.-M.; McCarthy, T. J. Macromolecules 1997, 30, 1752-1757.
Kotov, N. A.; Magonov, S.; Tropsha, E. Chem. Mater. 1998, 10, 886-895.

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.

Baur, J. W.; Besson, P.; O'Connor, S. A.; Rubner, M. F. Mater. Res. Soc.
Symp. Proc. 1996, 413, 583-588.

Anderson, M. R.; Davis, R. M.; Taylor, C. 0.; Parker, M.; Clark, 8.; Marciu,
D.; Miller, M. Langmuir 2001, 17, 8380-8385.

Moriguchi, l.; Teraoka, Y.; Kagawa, 8.; Fendler, J. H. Chem. Mater. 1999,
11, 1603-1608.

Liu, Y.; Wang, A.; Claus, R. 0. Appl. Phys. Lett. 1997, 71, 2265-2267.

Martin, C. R. Science 1994, 266, 1961-1966.

97

(37)

(38)

(39)

(40)

(41)

(42)

(43)

(44)
(45)

(46)

(47)

Petri, D. F. S.; Wenz, (3.; Schunk, P.; Schimmel, T. Langmuir1999, 15,
4520-4523.

Sroog, C. E. J. Polym. Sci, Macromol. Rev. 1976, 11, 161-208.

Merkel, T. C.; Freeman, B. D.; Spontak, R. J.; He, 2.; Pinnau, l.; Meakin,
P.; Hill, A. J. Science 2002, 296, 519-522.

Arys, X.; Jonas, A. M.; Laguitton, B.; Legras, R.; Laschewsky, A.;
Wischerhoff, E. Prog. Org. Coat. 1998, 34, 108-118.

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

Harris, F. W. In Polyimides, Wilson, 0., Stenzenberger, H. D.,
Hergenrother, P. M., Eds.; Chapman and Hall: New York, 1990, pp 1-37.

Zhou, Y.; He, P.; Chen, K.; Li, 0.; Hong, K. Zhongguo Kexue Jishu Daxue
Xuebao 1995, 25, 21 -26.

Kim, T.-H.; Koros, W. J.; Husk, G. R. J. Membrane Sci. 1989, 46, 43-56.
Henis, J. M. S.; Tripodi, M. K. Science 1983, 220, 11-17.

Kesting, R. E.; Fritzche, A. K. Polymeric Gas Separation Membranes;
John Wiley & Sons: New York, 1993.

Harris, J. J.; Stair, J. L.; Bruening, M. L. Chem. Mater. 2000, 12, 1941-
1946.

98

Chapter 5

Ultrathin, Cross-linked Polyimide Pervaporation Membranes Prepared from

Polyelectrolyte Multilayers

5.1 Introduction

Pervaporation is a promising, membrane-based technique for the
separation of liquid mixtures. In this process, the liquid is exposed to the surface
of a membrane whose permeate side is at reduced pressure, and selective
transport of one component through the membrane allows for solvent purification
or analyte collection.1 Pervaporation is attractive because it allows for the
separation of azeotropic mixtures and often requires less energy than
conventional distillation?3 Many of the successes with pervaporation involve the
removal of water from organic solvents, but the reverse separation has also been
demonstrated.“"6 Even with advances in pervaporation technology, however,

there is still a need for membrane systems with increased selectivity and flux.

Membranes with high water-pervaporation selectivities generally contain
hydrophilic materials that preferentially absorb water. Nevertheless, many
hydrophilic polymers swell dramatically in aqueous solutions to give large fluxes
but little or no transport selectivity.7'9 Hence, to achieve selective water
pervaporation, several research groups reduced the extent to which hydrophilic

1°43 or materials that contain

membranes swell by using cross-linked polymers
both hydrophobic and hydrophilic regions.9"4"8 The use of mildly hydrophobic

polymers, e.g., polyimides, also reduces membrane swelling to yield highly

99

selective pervaporation membranes.‘9'21

All of these strategies for decreasing
swelling can increase selectivity, but at the same time, they do reduce permeate
flux.

When selecting membrane materials, there generally is a compromise

between selectivity and flux because highly selective materials are usually not

highly permeable.22 The most common method employed to overcome this

23,24 25,26

limitation is the use of asymmetric or composite membranes that consist
of a thin, highly selective material on a highly permeable support. Because the
thin separation layer provides most of the resistance to mass transport, flux is
often inversely proportional to the thickness of this layer.25 The recently
developed technique of alternating polyelectrolyte deposition (APD) provides a
particularly versatile method for depositing ultrathin membrane skins whose

”‘32 This method involves

thickness can be controlled on the nm scale.
alternating immersions of a charged substrate into polycation and polyanion
solutions, and thus film thickness is easily controlled by changing either the
number of adsorbed polyelectrolyte layers or deposition variables such as pH

33-36

and supporting salt concentration. To form membranes using APD, the

polyelectrolyte multilayers are simply deposited on porous supports, and
32,33,37

because each added polyelectrolyte layer covers defects in prior layers,

films as thin as 20 nm can completely cover underlying pores.38

Tieke and coworkers used typical polyelectrolyte multilayers, e.g.,
poly(styrene sulfonate)/poly(allylamine), to prepare pervaporation membranes for

the removal of water from ethanol.”40 They found a direct relationship between

100

the separation selectivity and the charge density of the polyelectrolytes,
presumably because higher charge density increases the extent of ionic cross-
linking and reduces swelling. They attempted to further increase ionic cross-
linking by annealing the films at 90 °C and achieved water/ethanol selectivities as
high as 450. Although these membranes were very selective in the removal of
water from organic solvents, they also had relatively low fluxes, most likely
because of the high thickness of the separating layer (sixty bilayers were used to
make these membranes). Meier-Haack and coworkers showed similar results
using polyethylenimine/poly(acrylic acid) films with as few as 6 bilayers.28
However, even in that case, fluxes in highly selective separations were less than
1 kg-m'z-h".

In this C, we combine the versatility of APD with covalent interlayer cross-
linking and the use of semi-hydrophobic polymers (polyimides) to prepare high-
flux, high selectivity pervaporation membranes. Ionized precursor poly(amic
acids) serve as the polyanions in multilayer polyelectrolyte films, and subsequent
heating yields a polyimide coating.”44 Using this method, we previously
prepared polyimide membranes for gas and ion separations,‘°‘9'30 but the
poly(amic acids) employed in this study contain diaminobenzoic acid (DABA)
groups that can form amide cross-links when heated (Figures 5.1 and 5.2).
DABA was previously incorporated into polyimide pervaporation membranes to

17.45

increase hydrophilicity or to allow for cross-linking upon addition of diols or

46.47

diamines. In the case of poly(amic acid)/polycation multilayers, the

polycation, e.g., protonated poly(allylamine), serves as the cross-linking agent,

101

x = 1 BPDA- x = 0.5 BPDA-ODA- x = 0 BPDA-

ODA DABA DABA
H c CH
NH3+cr 3 \E’ 3
13% 1 v1 fl
PAH LPEI PDADMAC

Figure 5.1. Structures of the poly(amic acids) and other
polyelectrolytes used in this study.

102

 

Figure 5.2. Heat-induced imidization and cross-
linking of BPDA-DABA/PAH films.

103

and the cross-linking density can be controlled by varying the fraction of DABA

groups in the poly(amic acid).

By depositing ~50 nm thick films on porous alumina supports,48 we are
able to selectively dehydrate ethanol and isopropanol solutions while maintaining
high flux. Both amide cross-links and the hydrophobic polyimide backbone aid in
reducing the swelling of selective films, and water-selective permeation
presumably occurs because the polycation and non-cross-linked acid groups add
a hydrophilic component to the membrane. We investigated the effect of
imidization temperature, polyimide structure, and the amount of cross-linking on
selectivity and flux. We also utilized three different polycations to examine the
effects of polycation structure and composition on transport. Fully imidized
membranes composed of 3,3’,4,4’-biphenyltetracarboxylic dianhydride-DABA
and poly(allylamine) give the largest water-pervaporation selectivities: 1100 for
10% isopropanol solutions and 6100 for 90% isopropanol, with fluxes of 11 and 2
kg-m'z-h", respectively. Such selectivities are typical of high-performance

pervaporation membranes, but fluxes are, in general, several times larger. “9'51

5.2 Experimental Section

5.2.1 Materials. 3,3’,4,4’-biphenyltetracarboxylic dianhydride (BPDA), 4,4’-
oxydianiline (ODA), 3,5-diaminobenzoic acid (DABA), poly(allylamine
hydrochloride) (PAH), and poly(diallyldimethylammonium chloride) (PDADMAC)
were purchased from Aldrich. Linear polyethylenimine (LPEl) was purchased
from Polysciences. DMF (Spectrum) was dried with molecular sieves for at least

24 h before use. The ODA and BPDA were purified by vacuum sublimation, and

104

DABA was recrystallized twice from water and dried in vacuum prior to use. All
other chemicals were used as received. ACS grade isopropanol (CCI) and 200-
proof ethanol (Pharrnco) were employed in pervaporation studies, and deionized
water (Milli-Q, 18.2 M!) cm) was used for rinsing and preparation of polymer-
containing and pervaporation solutions. Porous alumina supports were 25 mm
Whatman Anodiscs with 0.02 um-diameter surface pores (Fisher Scientific), while
silicon(100) wafers (Silicon Quest) sputter coated with 200 nm of AI served as
supports for ellipsometry, reflectance FT IR spectroscopy, and contact-angle

measurements.

5.2.2 Substrate Preparation. In the case of porous alumina supports, the
polypropylene support ring on the alumina was removed to prevent it from
melting into substrate pores during heat-induced imidization and cross-linking.
This was accomplished by cutting off as much of the polymer as possible, and
then buming off the remaining ring at 400 °C for 18 h in a furnace.
Subsequently, the alumina supports were rinsed with acetone, dried with N2, and
cleaned for 10 min in a UV/Oa cleaner (Boekel Industries, model 135500). Al-

coated wafers were simply cleaned by UV/Oa prior to use.

5.2.3 Polymer Synthesis. The poly(amic acids) were prepared using a typical
literature procedure.‘52 The diamine was dissolved in DMF, a stoichiometric
quantity of the dianhydride was gradually added to this solution over a ~15 min
period, and the 10% by weight (total dissolved solids) solution was stirred for 24
h under a nitrogen atmosphere. The poly(amic acids) were precipitated twice in

0.01 M HCl, filtered, and then dried under vacuum for 24 h.

105

5.2.4 Solution Preparation and Film Deposition. To avoid the possibility of
hydrolysis, poly(amic acid) solutions were made just prior to film formation, and
the dry polymers were always kept in the freezer. Due to the insolubility of the
poly(amic acids) in water, solutions containing ~0.005 M poly(amic acid) and 0.5
M NaCI were prepared by dissolving the poly(amic acid) in 0.75 ml of 0.2 M
triethylamine, diluting to 10 ml with water and adding 0.29 9 NaCl. (The
molarities of the polyelectrolytes are given with respect to the repeating unit.)
The pH was adjusted to ~50 for DABA-containing poly(amic acids) and ~6.0 for
BPDA-ODA by adding 0.1 M HCl. As mentioned in a previous publication, the
deprotonated poly(amic acids) are deposited under conditions near their solubility
limits to achieve films of appreciable thickness.30 Poly(amic acids) typically
precipitate out of aqueous solutions when either the pH is low enough to

protonate most of the carboxylic acids or the concentration of salt is high.

Film deposition began with immersion of the substrate (Al-coated Si or
porous alumina) into the poly(amic acid) solution for 3 min. (The alumina
membranes were placed in an o-ring holder so that only the face with 0.02 pm-
diameter pores was exposed to solutions.) The substrate was then rinsed with
copious amounts of Milli-Q water for about 1 min. Subsequently, the substrate
was immersed in an aqueous 0.02 M PAH, LPEI, or PDADMAC solution
(adjusted to pH 5.0 with NaOH) for 3 min and rinsed with Milli-Q water. This
process was repeated until the desired number of bilayers were deposited, and

all films were terminated with a poly(amic acid) layer.

106

5.2.5 Film lmidization. Films were heated using a home-built apparatus
consisting of a temperature controller, thermocouple, glass chamber, and heating
mantle. The heating apparatus was purged with nitrogen for 2 h prior to heating
as well as during heating and cooling. The temperature was ramped at a rate of

5 °C per min and was held at the final heating temperature for 2 h.

5.2.6 Film Characterization. Ellipsometric thickness measurements on Al-
coated silicon wafers were made using a rotating analyzer ellipsometer (model
M-44, J. A. Woollam) and WVASE32 software. A film refractive index of 1.5 was
assumed in all thickness determinations. Reflectance FTIR spectra were
obtained using a Nicolet Magna-560 FTIR spectrophotometer and a Pike grazing
angle (80‘-’) attachment. A UV/Oa-cleaned Al-substrate was used as a
background. Static water contact-angle measurements (Firsttenangstroms
contact angle analyzer) were also performed before and after film heating. Field-
emission scanning electron microscopy (FESEM) images were obtained with a
Hitachi S-4700ll instrument using an acceleration voltage of 15 kV. Samples
were fractured in liquid nitrogen and sputter-coated with 5 nm of gold before

imaging.

5.2.7 Pervaporation Measurements. All pervaporation measurements were
taken using the home-built apparatus shown in Figure 5.3a. The apparatus
consists of a vacuum pump, peristaltic pump, collection traps, and a heated
water bath, which is not shown in the figure. The membranes are first placed in a

membrane cell (Figure 5.3b) where they are supported by a stainless steel frit

107

 

  
 

membrane *
cell if I f
trap
' feed \‘ ~

solution

l?
l;

 

 

 

 

 

 

 

 

 

 

l.—

 

 

 

vac

flow flow
out

b (\ r/q

    
 

 

 

 

 

 

 

 

 

 

vacuum OUt

Figure 5.3. Schematic diagrams of (a) the
pervaporation system and (b) a cross-sectional view
of the membrane cell.

108

(Mott Corporation) and sealed with an o-ring. The cell is sealed with an
additional o-ring to maintain vacuum, and solution is pumped across the
membrane at a rate of 10 mL/min to minimize concentration polarization. The
membrane cell is connected to a coiled stainless steel feed tube, and both the
membrane cell and coiled tube were immersed into a therrnostated water bath to
control the feed solution temperature. The bath was maintained at 50 °C unless
othenrvise stated. Vacuum (pressure of 0.06 mbar) was applied to the backside
of the membrane, and the permeate was collected in a liquid nitrogen-cooled
trap. Tubing on the permeate side was warmed with heating tape to avoid any
condensation. Between experiments with different concentrations of the same
alcohol, the feed side of the membrane was flushed with 300 mL of the feed
solution. When changing from isopropanol to ethanol, the feed side was flushed
with 1.5 L of water followed by 300 mL of the next feed solution. At all times,
permeate was collected in a trap. Finally, prior to sample collection,
pervaporation was performed for at least 1 h to achieve stable, steady-state
fluxes. The collected samples were analyzed by gas chromatography (Shimadzu
GC-17A equipped with a Restek RTx-BAC1 column) using methanol as an
internal standard. The selectivity,a, was calculated using equation 1 where Pw,
Fw, PA, and FA are the concentrations of water in the permeate and the feed and
alcohol in the permeate and the feed, respectively. Reported selectivities and
fluxes are averages of measurements performed on at least three different

membranes.

109

Table 5.1:

Polyelectrolyte Filma

bilayer thickness

Bilayer thicknesses (nm) and Water Contact Angles for Unheated
and Heated (250 °C) Polyelectrolyte Films Deposited on AI-coated Si.

contact angle

 

unheated heated unheated heated

BPDA-ODA/PAH 4.0 1 0.3 2.8 1 0.3 41 1 6 66 1 5
BPDA-ODA-DABA/PAH 4.1 1 0.1 3.3 1 0.2 27 1 3 49 1 1
BPDA-DABA/PAH 3.5 1 0.1 2.7 1 0.1 27 1 1 52 1 4
mat/E6233” 5.2 1 0.1 4.2 1 0.1 39 1 3 53 1 5
BPDA-DABA/PDADMAC 4.2 1 0.3 2.8 1 0.2 29 1 3 53 1 2
BPDA-ODA-DABA/LPEI 6.7 1 0.6 3.5 1 0.1 39 1 2 57 1 1
BPDA-DABA/LPEI 6.3 1 0.1 3.7 1 0.1 31 1 2 58 1 3

3 Measurements were taken on films containing 5.5 bilayers.

110

PW FA

e .1 a=—
q PAFW

5.3 Results and Discussion

5.3.1 Film Deposition and Characterization. To evaluate how polymer
structure and cross-linking affect pervaporation selectivity and flux, we prepared
polyelectrolyte multilayers using three different poly(amic acids) as well as three
polycations (Figure 5.1). The poly(amic acids) contain varying amounts of
DABA, which is the primary cross-linking moiety in these polymers, while the
polycations PAH and LPEl contain protonated primary and secondary amines,
respectively. These amine groups can form amide cross-links via heat-induced
reaction with the carboxylic acid groups of DABA (Figure 5.2). In contrast,
PDADMAC contains permanently charged quaternary amines that do not react

with carboxylic acids.

We first performed ellipsometric and reflectance FT IR studies with films
deposited on Al-coated Si wafers to characterize film thickness and the heat-
induced imidization reaction. Film thickness increase linearly with the number of
bilayers deposited, and Table 5.1 presents the thickness per bilayer for each of
the films. Bilayer thicknesses are about the same (within a factor of 2) for all of
the materials. From these measurements, we determined the number of bilayers
(unheated) needed to form 40 to 50 nm-thick films for each of the polyelectrolyte
pairs examined. Previous FESEM images suggest that at least 20 nm of film is

required to cover porous alumina, so we deposited 40 to 50 nm-thick films to

111

ensure full coverage of underlying pores.38 lmidization at 250 °C results in a 20

to 50% decrease in film thickness, presumably because of the loss of water.

Thicknesses determined by ellipsometry on Al-coated Si and by FESEM
on porous alumina are in reasonable agreement, although in general, the films
on porous alumina appear to be about 30% thicker.30 Cross-sectional FESEM
images of imidized films on porous alumina showed 40 to 60 nm-thick films for all
of the polyelectrolyte combinations. (Tables 5.2 and 5.3 give the number of
bilayers in different films.) Figure 5.4 shows a typical image of 12.5 bilayers of
imidized BPDA-DABA/PAH on alumina, and this and other images suggest that
films are relatively uniform across the support. Top-down images also suggest
complete pore coverage. Very little, if any, deposition of polyelectrolytes is
noticeable on the walls of the 0.2-pm pores in the bulk of the alumina, and

images of the uncoated backside of the membrane show completely open pores.

Reflectance FTIR spectroscopy confirmed imidization and cross-linking
within polyelectrolyte multilayers. Figure 5.5 shows spectra of three different
poly(amic acid)/PAH films that were heated at 250 °C as well as the spectrum of
one unheated film. After imidization at 250 °C, absorbances due to amic acids
vanished (1750-1500 cm"), and imide peaks appeared (1775 and 1725 cm").
The spectrum of the heated BPDA-DABA/PAH film contains a shoulder at 1670
cm", which is most likely due to the amide cross-links formed by reaction of the
primary amine groups of PAH and the carboxylic acid groups of the DABA

segment (Figure 5.2). Assuming that the extinction coefficient for these amides

112

 

Figure 5.4. Cross-sectional FESEM image of an alumina
support coated with an imidized 12.5-bilayer BPDA-DABA/PAH
film.

113

 

 

BPDA-DABA/ PAH
----- BPDA-ODA-DABA/ PAH
0.0.0.0000. BPDA-ODA/PAH

- ----- Unheated BPDA-
DABA/PAH

Absorbance (a.u.)
\
I

 

 

'OoeoOO"
F l T l I I I

1800 1750 1700 1650 1600 1550 1500
Wavenumber (cm‘)

Figure 5.5. Reflectance FTIR spectra of various poly(amic
acid)/PAH films deposited on Al-coated Si. The spectra of
heated films were normalized to the imide peak of BPDA-DABA
at 1725 cm", while spectra of heated and unheated BPDA-
DABA/PAH films are plotted on a common scale.

114

is comparable to that of the amides in the unheated poly(amic acid) (1660 cm"),
the spectra suggest that roughly 60% of the DABA groups react to form cross-
links. As expected, the absorbance of the 1670 cm'1 shoulder decreases by
~50% on going from BPDA-DABA/PAH to BPDA-ODA-DABA/PAH films.
However, even BPDA-ODA/PAH films show a small shoulder in this region,
which may be due to a small amount of cross-linking with the amic acid or
incomplete imidization. The spectrum of imidized BPDA-DABA/PDADMAC is
similar to that of BPDA-ODA/PAH in this region (supporting information, Figure
5.1), further suggesting that shoulders at 1670 cm'1 in the spectra of BPDA-
DABA/PAH and BPDA-ODA-DABA/PAH are due to amide cross-links.

5.3.2 Pervaporation Experiments with non Cross-Iinkable Membranes. To
examine the effect of film imidization on pervaporation selectivity and flux, we
initially compared unheated and heated membranes that are unlikely to form.
amide cross-links. Such systems contain either BPDA-ODA or PDADMAC.
Unheated BPDA-ODA/PAH membranes exhibited very low water/alcohol
pervaporation selectivities (<4) for any of the alcohol solutions, and fluxes were
extremely high, ranging from 50 to 70 kg-m"°'-h'1 (Table 5.2). FESEM images
suggest complete coverage of the support, so we think that the high fluxes result
from water-swollen membranes rather than incomplete support coverage.
Swelling by water is consistent with the lower water/alcohol selectivities for 10%
alcohol solutions relative to 90% alcohol solutions. However, we do not

understand why the flux is ~20% higher for the 90% alcohol solutions.

115

QdHe. dedd HHN:
88. E H .18 .85. 3d H d8 .8: 8d H 88
dedd HdeH .HHH
.8: .1 H 5.8 .88 8d H :8 .8. 8d H 5.8
:dHQN demd de5H
.8. m... H :8 68. 3 H 8.8 .2. 8d H 8.8
©de: .ded 93 a
a: dd H :5 88. 3 H 5.8 .1. 5 8d H 5.8
NdHem deHd mHNH
.8. ed H 1.8 .8: ed H «.8 a: 3 H 1.8
v H 5 v H 5 v H 8

ad. 2 H 5.8 .3”. w: H QR .d. 5 3 H 58

HHHH
82 a mood H 8.8
dded
88. 8.8 8.8

HHoH
.8: 8d H 8.8

.Hd.
5d. 8d H «.8

N H dd
.8. 3d H 5.8

VHHm
a: 9d H 5.8

5555 5.9. o. 08 859.
15.83.38

5555 em: o. 8.0. 8.8..
11.985.50-35

.5555 5.9. 0.. of 858
IE. 85.30-35

.5555 5d: 859.5
1.11.53-30-35

.2965 Add: 0.. O8 .858
1528-35

.2825 dd: 8.815
15200-55

 

5:95:68. 588 .0555 $9

:o_S_om :oz9odm>5n.

.ocmfim £50m

5:39:05. 89

5.55:. 5:95:55.

 

.05 on 5 5:95:55. 1.5.0.35 6.89266 .9555 .5395 :o:9oam>5n. .8 A.-:.~-E.mv. :_ 52.9.
:_ .55... 55x3... 65 855555: c. 55... 95.05500 3.2.5.3 65.9.5050 55.5 555556 ”NA... 53:...

116

NHVH
.8. «.0 H 1.8
dded
.03. 00.0 H 0.8

.dH0H
.9. 8.0 H :8

«Hm.
.8. 0:0 H :8

0Hkm
.cddH000

.5555 m... 0.. 08 858
59.20.10-310

.5555 0:. o. 08 850;
53220-501000

.5535 0d: 0.. 08 850..
0<20<0a.<m<0-<0n_m

.5535 0.0.. o. 08 858
0<20<0020<0-<0o-<0n_m

555.5 0.0: 8580:
0.120502031001000

 

eded «.33 :9
.8. 0.0 H 0.: .09. 0... H N. 8 .0: «.0 H 5.8
deHN Vded ~H0d
.0». 0.9 H N8 .00.. E H 0.8 .E. 00.0 H 0.8
0Hm0 de.d .de.
.9. 0.0 H :5 .05. 8d H 0.8 .3. 0d H :8
.H05 00H.~ 0H0.
.9. :0 H 0.8 .00.... :H H 0.8 .05. 0.. H :8
NH00 0Hmd «Hum
.00. :0 H :8 .t. 0.: H :8 .H. 0.. H :8
588 01.00 550058. 058 5095 1.0.

:0.S_0m 5.5500200

5500500. 08.:

5.55E 0:90:55.

.00 on 5 00:99:05.

50.1.08 205.50: 0.5 01.20.101.055 0.8.50: 5555 1085. 00.5588: .5. .1550. 0. .85..
c. 005... 00.3.... 0:0 5.005550 :. 005... 05055000 3.2.00.0m .0:o_5:.:00:00 55>> 050E5n. ”a...“ 0.5...

117

lmidization of BPDA-ODA/PAH membranes yields enhanced selectivities
ranging from 16 to 160 (Table 5.2) along with permeate fluxes that are only 3 to
20% of those through the corresponding unheated membranes. Water contact
angles (Table 5.1) show that imidization increases hydrophobicity, which should
reduce swelling and, hence, increase selectivity and decrease flux. Although
heated films are slightly hydrophobic, the polyimides are still sufficiently
hydrophilic to allow preferential permeation of water. Pervaporation membranes
prepared by casting of pure KaptonT'“, a similar polyimide, showed water
selectivities of 50 for 90% ethanol solutions.17 Thus, selectivities do not appear

to be adversely affected by the presence of PAH.

Similar results occur with BPDA-ODA-DABA/PDADMAC, where unheated
membranes show no selectivity in 10% isopropanol and ethanol solutions and
modest selectivities in 90% alcohol solutions (Table 5.3). lmidizing the BPDA-
ODA-DABA/PDADMAC membranes greatly increases the selectivity, with the
most notable result being 98% water in the permeate from a 90% isopropanol
feed. We also examined pervaporation through heated BPDA-
DABA/PDADMAC, and these systems generally have slightly greater fluxes and
slightly lower selectivities than heated BPDA-ODA-DABA/PDADMAC. The
presence of more DABA groups in BPDA-DABA than in BPDA-ODA-DABA likely

allows an increase in film swelling.

5.3.3 Cross-linkable Membranes. To evaluate the dependence of selectivity

on cross-linking density, we compared imidized BPDA-DABA/PAH, BPDA-ODA-

118

DABA/PAH, and BPDA-ODA/PAH membranes. For all solvent mixtures, the
selectivity increases with the amount of DABA in the imidized (250 °C) polymer
film (Table 5.2). As an example, the water/alcohol selectivities for 90%
isopropanol feed solutions are 160, 690, and 6100 for heated BPDA-ODA/PAH,
BPDA-ODA-DABA/PAH, and BPDA-DABA/PAH films, respectively. Remarkably,
within experimental error, the flux values are the same for all three of the
membranes in nearly every case. We expected the flux to decrease with higher
cross-linking densities, but the effect of DABA content on flux appears to be
minimal. Remaining non-cross-Iinked DABA most likely adds a hydrophilic

component to the membrane, which increases water transport.

We also used the BPDA-ODA-DABA/PAH system to investigate the effect
of the feed solution temperature on transport. Consistent with other
pervaporation studies,53 permeate flux increased with temperature. For 10%
isopropanol feeds at 35, 50, and 65 °C, fluxes were 5.0, 9.7, and 17 kg-m'z-h“,

respectively. Selectivity, however, did not depend on the feed temperature.

LPEI provides a second polycation that should be capable of forming
cross-linked films. We thought that the LPEI would be more hydrophilic than
PAH and, hence, provide cross-linked membranes with higher water/alcohol
selectivity.S4 Reflectance FTIR spectroscopy (Figure 5.6) confirms that the
secondary amine in LPEI allows for cross-linking, although the amide shoulder at
1670 cm'1 is about half the intensity of that for PAH-containing films. The lower
cross-linking density appears to have a large impact on selectivity. In the case of

imidized BPDA-DABA films, the use of LPEI rather than PAH results in 10 to 50-

119

 

— BPDA-DABA/ PAH
--- BPDA-DABA/LPEI

BPDA-DABA/
PDADMAC

 

Absorbance (a.u.)
.9aouuuwu awnufivu

 

 

 

 

 

1800 1750 1700 1650 1600 1550 1500
Wavenumber (cm“)

Figure 5.6. Reflectance FTIR spectra of various BPDA-
DABA/polycation films deposited on Al-coated Si. The spectra

of heated films were normalized to the imide peak of BPDA-
DABA/PAH at 1725 cm".

120

fold lower selectivities. The effect is less dramatic for BPDA-ODA-DABA, but
selectivities are still 2 to 4-fold lower, except in the case of 10% ethanol.
Interestingly, like the PDADMAC but unlike the PAH systems, the BPDA-ODA-
DABA/LPEI membranes have higher selectivities than the BPDA-DABA/LPEI
membranes. Without as many cross-links, the added hydrophilicity of the DABA

may again allow for increased swelling that permits some non-specific transport.

Variation of heating temperature should provide some control over both
imidization and cross-linking and, hence, allow manipulation of transport
properties. For pure, dip-cast poly(amic acid) membranes, Kang and coworkers
showed that partial (60%) imidization gave the highest selectivity in the removal

of water from ethanol.20

They attributed the higher selectivity to a balance
between increased hydrophilicity due to remaining amic acid groups and a
reduction in swelling due to imidization. In our previous papers we showed that
the extent of imidization of multilayer poly(amic acid)/polycation films can be
controlled by varying heating temperature?9 FT IR spectra of films heated at 150
°C for 2 h showed 85% imidization, while full imidization occurred at 250 °C.
Thus, to see how membrane performance changes with the extent of imidization,
we compared BPDA-ODA-DABA/PAH membranes heated at 150 and 250 °C
(Table 5.2). Partial imidization does not provide any advantage in the separation,
as it yields selectivities that are as much as 2.4-fold lower than those achieved by
heating at 250 °C. Moreover, substantially higher fluxes were observed for the

partially imidized membranes only with 10% alcohol feed solutions, and even in

that case flux increased by only ~50%. We suspect that a further decrease in

121

heating temperatures would only result in lower water/alcohol selectivities, as
selectivities for unheated membranes are lower than for membranes heated at

150 °C.

5.3.4 Comparison with related membrane systems. Kang and coworkers
used pure pyromellitic dianhydride (PMDA)-ODA-DABA membranes that
contained various amounts of DABA to dehydrate ethanol solutions.17 However,
they did not use DABA as a site for cross-linking, but only to increase the
hydrophilicity of the membrane. Maximum water selectivities and fluxes (320 and
0.06 kg-m‘Z-h", respectively for 90% ethanol) occurred when 60% of the
polyimide repeat units contained DABA. This selectivity is almost 20-fold higher
than that with BPDA-ODA-DABA/PDADMAC films, which contain a similar
fraction (50%) of DABA groups, suggesting that PDADMAC does alter film
packing to reduce selectivity. In contrast, cross-linked BPDA-DABA/PAH films, in
which 100% of the repeat units contain DABA (approximately 60% of these
groups participate in cross-linking), show selectivities similar to bulk PMDA-ODA-
DABA. Although Kang and coworkers used PMDA rather than BPDA as a
monomer, this comparison still suggests that cross-linked polyelectrolyte
multilayers are capable of achieving selectivities similar to those in pure

polyimide films.

Compared to the majority of pervaporation membranes in the literature,
ultrathin BPDA-DABA/PAH composite membranes allow for at least lO-fold

49-51

larger water fluxes while maintaining comparable selectivities. The greater

fluxes result from the minimal thickness of the polyelectrolyte multilayers.

122

Yanagashita showed that polyimide asymmetric membranes can give
pervaporation selectivities and fluxes as high as 900 and 1 kg-m'zh‘,
respectively, for 95% ethanol feed solutions at 60 °C.19 These values are
comparable to those achieved with BPDA-DABA/PAH films because the
asymmetric membranes were prepared by a phase-inversion process that also
results in extremely thin separation layers. Compared with the phase-inversion
process, APD should allow for greater control over the selective layer thickness.
Additionally, less of the relatively expensive polyimide is needed in APD than in
the phase-inversion process because in the latter case, the polyimide serves as
both the support and the separation layer. However, the high heating
temperatures required to drive imidization and cross-linking reactions do greatly

restrict the supports that can be used for forming polyimide membranes by APD.

5.4 Conclusions

Alternating electrostatic adsorption of poly(amic acids) and polycations on
porous alumina supports followed by post-deposition heating provides ultrathin,
polyimide films that selectively remove water from alcohols by pervaporation. By
including a pendant carboxylic acid group in the poly(amic acid), amide cross-
links can be formed between the poly(amic acid) and PAH or LPEI. In the case
of polyimide/PAH films, increasing the cross-linking density results in higher
water/alcohol selectivities, and incorporation of a cross-linking group in each
repeat unit of the polyimide (BPDA-DABA/PAH) yields a selectivity of 6100 for
the removal of water from 90% isopropanol. This selectivity is achieved with

films as thin as 50 nm, so fluxes are still 2 kg-m'z-h'1 at 50 °C.

123

References:

(4)

(5)

(6)

(7)

(8)

(9)
(1 0)

(11)

(12)

(13)

(14)

(15)

Huang, R. Y. M. Pervaporation Membrane Separation Processes", Elsevier
Science: Amsterdam, 1991; Vol. 1.

Lipnizki, F.; Field, R. W. Environ. Prog. 2002, 21, 265-272.

Aminabhavi, T. M.; Khinnavar, R. S.; Harogoppad, S. B.; Aithal, U. S. J.
Macromol. Sci. R. M. C. 1994, C34, 139-204.

Liu, Q.-L.; Xiao, J. J. Membrane Sci. 2004, 230, 121-129.

Volkov, V. V.; Fadeev, A. G.; Khotimsky, V. S.; Litvinova, E. G.;
Selinskaya, Y. A.; McMillan, J. D.; Kelley, S. S. J. Appl. Polym. Sci. 2004,
91, 2271-2277.

Slater, C. S.; Hickey, P. J.; Juricic, F. P. Sep. Sci. Technol. 1990, 25,
1063-1077.

Karakane, H.; Tsuyumoto, M.; Maeda, Y.; Honda, 2. J. Appl. Polym. Sci.
1991, 42, 3229-3239.

Xu, J. B.; Bartley, J. P.; Johnson, R. A. J. Appl. Polym. Sci. 2003, 90, 747-
753.

Kurkuri, M. D.; Aminabhavi, T. M. J. Appl. Polym. Sci. 2003, 89, 300-305.
Huang, R. Y. M.; Shieh, J.-J. J. Appl. Polym. Sci. 1998, 70, 317-327.

Rhim, J.-W.; Yeom, C.-K.; Kim, S.-W. J. Appl. Polym. Sci. 1998, 68, 1717-
1723.

Du Prez, F. E.; Goethals, E. J.; Schue, R.; Qariouh, H.; Schue, F. Polym.
Int. 1998, 46, 117-125.

Yu, J.; Lee, C. H.; Hong, W. H. Chem. Eng. Process. 2002, 41, 693-698.

Chanachai, A.; Jiraratananon, R.; Uttapap, D.; Moon, G. Y.; Anderson, W.
A.; Huang, R. Y. M. J. Membrane Sci. 2000, 166, 271-280.

Chen, X.; Li, W.; Shao, 2.; Zhong, W.; Yu, T. J. Appl. Polym. Sci. 1999,
73, 975-980.

124

(16)

(17)

(18)

(19)

(20)

(21 )
(22)
(23)

(24)

(25)

(25)

(27)

(28)

(29)

(30)

(31)

(32)

Yoshikawa, M.; Yukoshi, T.; Sanui, K.; Ogata, N. J. Polym. Sci, Part A:
Polym. Chem. 1986, 24, 1585-1597.

Kang, Y. S.; Jung, B.; Kim, U. Y. Mol. Cryst. Liq. Crys. A 1993, 224, 137-
146.

Rafik, M.; Mas, A.; Guimon, M. F.; Guimon, C.; Elharfi, A.; Schue, F.
Polym. Int. 2003, 52, 1222-1229.

Yanagishita, H.; Maejima, C.; Kitamoto, D.; Nakane, T. J. Membrane Sci.
1994, 86, 231-240.

Kim, H. J.; Jung, B.; Hong, J.-M.; Kang, Y. S.; Kim, U. Y. Korea Polym. J.
1994, 2, 148-151.

Kim, J.-H.; Lee, K.-H.; Kim, S. Y. J. Membrane Sci. 2000, 169, 81-93.
Robeson, L. M. J. Membrane Sci. 1991, 62, 165-185.
Loeb, S.; Sourirajan, S. Adv. Chem. Ser. 1963, 38, 117-132.

Kim, S.-G.; Kim, Y.-l.; Yun, H.-G.; Lim, G.-T.; Lee, K.-H. J. Appl. Polym.
Sci. 2003, 88, 2884-2890.

Pinnau, l.; Freeman, B. D. In Membrane Formation and Modification;
Pinnau, l., Freeman, B. D., Eds.; American Chemical Society:
Washington, DC, 2000, pp 1-22.

Polotskaya, G. A.; Kuznetsov, Y. P.; Goikhman, M. Y.; Podeshvo, l. V.;
Maricheva, T. A.; Kudryavtsev, V. V. J. Appl. Polym. Sci. 2003, 89, 2361-
2368.

Liu, X.; Bruening, M. L. Chem. Mater. 2004, 16, 351-357.

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

Sullivan, D. M.; Bruening, M. L. J. Am. Chem. Soc. 2001, 123, 11805-
11806.

Sullivan, D. M; Bruening, M. L. Chem. Mater. 2003, 15, 281-287.

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

Kotov, N. A.; Magonov, S.; Tropsha, E. Chem. Mater. 1998, 10, 886-895.

125

(39)

(40)

(41)

(42)

(44)
(45)

(46)

(47)
(48)
(49)
(50)

(51)

Decher, G. Science 1997, 277, 1232-1237.

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

Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213-4219.
Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153-8160.
Kleinfeld, E. R.; Ferguson, G. S. Chem. Mater. 1996, 8, 1575-1578.

Harris, J. J.; Stair, J. L.; Bruening, M. L. Chem. Mater. 2000, 12, 1941-
1946.

Krasemann, L.; Toutianoush, A.; Tieke, B. J. Membrane Sci. 2001, 181,
221-228.

Toutianoush, A.; Tieke, B. Mat. Sci. Eng. C-Bio. 8.2002, 022, 459-463.

Anderson, M. R.; Davis, R. M.; Taylor, C. D.; Parker, M.; Clark, 8.; Marciu,
D.; Miller, M. Langmuir 2001, 17, 8380-8385.

Baur, J. W.; Besson, P.; O'Connor, S. A.; Rubner, M. F. Mater. Res. Soc.
Symp. Proc. 1996, 413, 583-588.

Moriguchi, |.; Teraoka, Y.; Kagawa, S.; Fendler, J. H. Chem. Mater. 1999,
_11, 1603-1608.

Liu, Y.; Wang, A.; Claus, R. 0. Appl. Phys. Lett. 1997, 71, 2265-2267.
Xu, W.; Paul, D. R.; Koros, W. J. J. Membrane Sci. 2003, 219, 89-102.

Schleiffelder, M.; Staudt-Bickel, C. React. Funct. Polym. 2001, 49, 205-
21 3.

Pithan, F.; Staudt-Bickel, C. Chemphyschem 2003, 4, 967-973.
Martin, C. R. Science 1994, 266, 1961-1966.

Shieh, J.-J.; Huang, R. Y. M. J. Membrane Sci. 1997, 127, 185-202.
Wang, X.-P.; Li, N.; Wang, W.-Z. J. Membrane Sci. 2001, 193, 85-95.

Sun, 8.; Zou, J. Ann. N. Y. Acad. Sci. 2003, 984, 386-400.

126

(52) Sroog, C. E. J. Polym. Sci, Macromol. Rev. 1976, 11, 161-208.
(53) Huang, R. Y. M.; Pal, A.; Moon, G. Y. J. Membrane Sci. 2000, 167.

(54) Clark, S. L.; Hammond, P. T. Langmuir2000, 16, 10206-10214.

127

   

l1131111111111111111111