CO2-SELECTIVE POLYMER BRUSH MEMBRANES AND ANTIBACTERIAL
POLYELECTROLYTE MULTILAYER MEMBRANES: SYNTHESIS,
CHARACTERIZATION, AND APPLICATION
By
Sebastian T. Grajales

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirement
for the degree of
Chemistry - Doctor of Philosophy
2014

ABSTRACT
CO2-SELECTIVE POLYMER BRUSH MEMBRANES AND ANTIBACTERIAL
POLYELECTROLYTE MULTILAYER MEMBRANES: SYNTHESIS,
CHARACTERIZATION, AND APPLICATION
By
Sebastian T. Grajales
Composite membranes consist of a selective skin on a highly permeable support. The
minimal thickness of the skin layer affords relatively high flux along with selectivity, whereas the
support provides mechanical strength. This dissertation presents methods for improving the
function of composite membranes prepared by either polymerization from a porous substrate or
alternating polyelectrolyte deposition on ultrafiltration membranes. Specifically, this work aims
to create gas-separation membranes that selectively remove CO2 from H2 streams and
nanofiltration membranes that resist biofouling.
Steam reforming of hydrocarbons followed by the water gas shift reaction yields H2
streams that contain 20 % CO2 as a byproduct. Ideally, in membrane-based purification of these
streams, the CO2 should preferentially pass through the membrane so the H2 remains at high
pressure. Because H2 is smaller than CO2, membranes that selectively pass CO2 must have a high
CO2/H2 solubility selectivity and minimal diffusive resistance. Amorphous poly(ethylene glycol)
(PEG) possesses these properties, but crystallization of PEG chains dramatically decreases flux
and eliminates selectivity. One goal of this research is to create ultrathin, PEG-containing
membrane skins in which PEG chains do not crystallize. Surface-initiated atom transfer radical
polymerization of PEG-methyl ether methacrylate (PEGMEMA) monomers produces thin (<200
nm) membrane skins on porous substrates, but crystallization of long PEG side chains (22-23
ethylene oxide units) restricts both flux and selectivity. However, copolymerization of monomers

containing 8-9 and 22-23 ethylene glycol units yields membranes that maintain a CO2/H2
selectivity of 13 or more, which is equal to the highest reported room-temperature CO2/H2
selectivities.

The short PEG chains in these films frustrate crystallization, and the high PEG

content in the larger monomers leads to high selectivity.
Formation of nanoparticles in membrane skins also inhibits room-temperature PEG
crystallization. The incorporation of tetraisopropoxytitanate into a poly(PEGMEMA) film and
subsequent condensation to TiO2 nanoparticles yields a nanocomposite membrane that maintains
a CO2/H2 selectivity of 6 over time. A small amount of crystallization in these films likely
decreases selectivity from ~13 to 6. Unfortunately, the use of higher nanoparticle concentrations
to better frustrate crystallization leads to defect formation and elimination of selectivity.
Membranes employed in aqueous environments commonly exhibit a decrease in
permeability over time due to fouling. Biofouling, the formation of a microbial foulant layer, is
particularly intrusive because even a few cells can multiply to create a film that reduces
permeability. One goal of this research is to create skin layers that both kill bacteria and provide
useful ion-transport selectivities. Alternating layer-by-layer deposition of poly(styrene sulfonate)
and polycations containing antibacterial guanidine groups generates a surface that resists
biofouling. Reflectance FTIR spectra confirm the layer-by-layer growth of these polyelectrolyte
films on Al-coated Si wafers, and filtration of bacterial solutions show that multilayer films can
be 100% bactericidal at short times. However, preliminary dead-end filtration studies reveal
MgCl2 and NaCl rejections <50%. Future work will examine the permeability of these membranes
in cross-flow filtration and investigate methods to improve ion-transport selectivity.

Copyright by
SEBASTIAN T. GRAJALES
2014

This Dissertation is dedicated to all of my family, and in particular my wife, Ann, and my
daughters, Liliana and Isabel.

v

ACKNOWLEDGEMENTS

First, I would like to thank my advisor, Merlin Bruening. My relationship with him started
when I attended one of his presentations and experienced his energy, enthusiasm, and passion for
good science. I wish that I could better communicate my gratitude for him accepting me into his
group, supporting my research, and strengthening many of my weaknesses. He is a friend and a
mentor to me, not to mention the fact that his ability to earn grant money funded my research!
I thank my committee members Greg Baker, Volodymyr Tarabara, and Gary Blanchard for
helping me. I would also like to thank earlier committee members Mercouri Kanatzidis for giving
me advice and guidance, Mitch Smith for helping me through the transition of changing advisors,
and Tom Pinnavaia for teaching me many concepts in the field of inorganic synthesis, as well as
how to prepare for presentations and improve my technical writing ability.
Next, I would like to thank the Bruening group members: Jamie, Elizabeth, Sri, Somnath,
David, Maneesha, Weihan, Xiaojie, Yujing, and Seth for supporting me and being good colleagues
and friends.
Finally, I thank my family. My wife, Ann, supported me for all these years and my
daughters, Liliana and Isabel, gave my life a new level of meaning. I wouldn’t be here without
the support of my Mom, who has always believed that I could do anything. I thank my Dad and
Stepmom who always encouraged me. I am very grateful for the support of my family.

vi

TABLE OF CONTENTS
LIST OF TABLES ......................................................................................................................... ix
LIST OF FIGURES .........................................................................................................................x
LIST OF SCHEMES......................................................................................................................xv
KEY TO ABBREVIATIONS ...................................................................................................... xvi
Chapter One. Separations Using Polymer Membranes...................................................................1
An Overview ........................................................................................................................1
Classification of Membranes Processes ...............................................................................4
Solution-diffusion permeation through dense membranes ..................................................6
Penetrant Solubility............................................................................................................16
Permeation through porous membranes.............................................................................22
Non-Ideal Membrane Performance....................................................................................25
Concentration Polarization.....................................................................................26
Biofouling ..............................................................................................................29
Synthesis of Selective, High-Flux Membranes ..................................................................30
Atom transfer radical polymerization ....................................................................31
Research Overview ............................................................................................................33
REFERENCES ..................................................................................................................34
Chapter Two. Effects of Monomer Composition on CO2-Selective Polymer Brush Membranes 41
Introduction ........................................................................................................................41
Experimental ......................................................................................................................45
Chemicals. ..............................................................................................................45
Polymerization .......................................................................................................46
Characterization methods.......................................................................................47
Gas Permeation ......................................................................................................47
Results and Discussion ......................................................................................................49
Polymerization Rates as a Function of Monomer Composition ............................49
Determining the Composition of Copolymer Films ..............................................51
Crystallization of PEG Side Chains .......................................................................56
Formation of Gas-Separation Membranes by ATRP .............................................58
Gas Permeation through Poly(PEGMEMA) Membranes ......................................59
Summary ............................................................................................................................63
Acknowledgements ............................................................................................................64
APPENDIX ........................................................................................................................65
Determination of absorption coefficients in reflectance IR spectra. ..................................66
Conclusions ........................................................................................................................72
REFERENCES ..................................................................................................................73
Chapter Three. The effect of nanoparticles on CO2-selective polymer membranes.....................77

vii

Introduction ........................................................................................................................77
Experimental ......................................................................................................................82
Chemicals ...............................................................................................................82
Polymerization .......................................................................................................83
Formation of the Polymer-nanoparticle Composite Film ......................................84
Characterization methods.......................................................................................84
Gas Permeation ......................................................................................................85
Results and Discussion ......................................................................................................86
Membrane Synthesis ..............................................................................................86
Inhibition of Crystallization by TiOx Nanoparticles ..............................................87
Changes in Membrane Permeability due to Nanoparticle Formation ....................92
Conclusions ........................................................................................................................95
REFERENCES ..................................................................................................................97
Chapter Four. Guanidine-Based Antifouling, Ion-selective Membranes ...................................102
Introduction ......................................................................................................................102
Materials and Method ......................................................................................................105
Preparation of films..............................................................................................105
Characterization of solute rejection .....................................................................106
Evaluation of membrane antimicrobial activity...................................................107
Results ..............................................................................................................................108
Membrane Surface Characterization....................................................................108
Zeta Potentials ......................................................................................................114
Antibacterial Activity...........................................................................................116
Transport properties .............................................................................................118
Conclusions ......................................................................................................................122
REFERENCES ................................................................................................................124

viii

LIST OF TABLES

Table 1.1: Physical/chemical properties exploited in chemical separations and the corresponding
separation techniques. Adapted and modified from Mulder.3 ........................................................2
Table 1.2: Selected milestones in understanding gas separations. Modified from reference 5. ....7
Table 1.3: The three possible combinations of solubility and diffusivities that lead to selective gas
transport, along with gas separations that employ these scenarios. Adapted from Pinnau and
Freeman.10 ......................................................................................................................................11
Table 1.4: Kinetic diametersa and critical temperatures of various gases.13 .................................14
Table 2.1: Composition of poly(PEGMEMA-475-co-PEGMEMA-1100) filmsa as a function of
the PEGMEMA-475 mole percentages in the polymerization solution.b ......................................56
Table A.1: Absorption coefficients, k, of homopolymer films. The k values were calculated using
Eq. (A.1-A.5) and the peak absorbances corresponding to C-O-C and C=O stretches. The
uncertainty is one standard deviation of the values obtained from films with a range of thicknesses.
........................................................................................................................................................70
Table 3.1: Selected nanoscale fillers in polymer materials and their corresponding applications.
POSS is polyhedral oligomeric silsesquioxane..............................................................................78
Table 3.2: Permeation properties of poly(PEGMEMA) membranes and poly(PEGMEMA)
membranes containing TiO2 nanoparticles, poly(PEGMEMA)-TiO2. Transport measurements
were made immediately after membrane preparation and one day later. ......................................88
Table 4.1: Strong infrared absorption modes for PSS/PHMG and PSS/PHMB films. ...............108

ix

LIST OF FIGURES

Figure 1.1: Schematic drawing of a membrane separating two phases. .........................................3
Figure 1.2: Classification of membrane processes according to the size of the species being
rejected. Adapted from reference 9. ................................................................................................5
Figure 1.3: Schematic diagram of different types of transport through porous and non-porous
membranes. Adapted from Ho and Sirkar.4 ....................................................................................6
Figure 1.4: Concentration profile of an analyte during solution-diffusion through a membrane,
where c1, c2, c3, and c4 represent the concentrations of the analyte in the feed, upstream face of the
membrane, downstream face of the membrane, and permeate, respectively. (This diagram neglects
concentration polarization, which is discussed below.) ...................................................................8
Figure 1.5: Concentration profile of a penetrant gas in a membrane from c1 at the feed face to c2 at
the permeant face. The partial pressure levels are shown for the feed (p1) and permeant (p2)
gas. .................................................................................................................................................10
Figure 1.6: Schematic diagram of a size-selective separation with a composite membrane. The
membrane has higher permeability for the smaller molecules due to diffusivity selectivity or
molecular sieving. The low pressure side (right) of the membrane contains a higher fraction of
small molecules than the feed side.................................................................................................12
Figure 1.7: Schematic diagram of reverse selectivity. In this case, the membrane has higher
permeability for larger molecules due to solubility selectivity. The lower pressure permeate side
(right) of the membrane contains a higher fraction of the larger molecules than the feed side.....13
Figure 1.8: Polymer volume relative to temperature. Modified from Zolandz and Fleming.14 ....15
Figure 1.9: Overview of various phenomena that generate resistance to convective flux. The
various resistances stem from the ideal membrane-Rm, cake layer formation-Rc, pore blocking-Rb,
concentration polarization-Rcp, and adsorption-Ra. .......................................................................26
Figure 1.10: Simplified scheme of a concentration profile (a) under a concentration driven
membrane separation with perfect mixing and (b) with concentration polarization at the feed
interface under steady-state conditions. The concentration at c1 does not necessarily equal c2, and
c3 does not necessarily equal c4. Note that in the case of concentration polarization the
concentration begins to increase at a distance δ, along the x-axis, from the membrane surface. ..27
Figure 1.11: Schematic drawings of (a) a dead-end membrane and (b) a cross-flow membrane
module. In the dead-end mode the concentrations of rejected species continually increase on the
feed side of the membrane, whereas the cross-flow module allows for rejected species to flow
along the membrane face and out of the module via the retentate path. ........................................29

x

Figure 1.12: Decreases in transmembrane flux over time, due to concentration polarization and
fouling. ...........................................................................................................................................30
Figure 1.13: Reactions involved in ATRP. The ATRP mechanism contains an equilibrium where
a dormant chain (R-X) activates at a rate of kact to form a radical (R●) and deactivates at a rate of
kdeact to return to the protected state. Transition metals (TM) at two different oxidation states (“a”
and “a+1”) complexed with ligand(s) mediate the process. The active radical reacts with monomer
to form polymer with a rate constant of kp or terminates with a neighboring radical with a rate
constant of kt. .................................................................................................................................32
Figure 2.1: Evolution of film thickness with time in surface-initiated polymerization of pure and
mixed monomer solutions. For copolymers, the legend shows the solution PEGMEMA-475 mole
percentage with respect to the total monomer (PEGMEMA-475 plus PEGMEMA-1100). The
overall monomer concentration was 0.67 M, and the catalyst system was 2.0 mM CuCl, 0.60 mM
CuBr2, and 6.0 mM HMTETA. The error bars, which in many cases are obscured by the symbols,
represent one standard deviation. ...................................................................................................50
Figure 2.2: FTIR spectra of 152 nm-thick poly(PEGMEMA-475) (black line) and 118 nm-thick
poly(PEGMEMA-1100) (dashed red line) films on gold-coated substrates. Spectra were taken
immediately after polymerization before the PEG side chains in poly(PEGMEMA-1100)
crystallized. The spectra are offset for clarity. ..............................................................................52
Figure 2.3: Ratios of C-O-C to C=O peak heights (A) and peak areas (B) in reflectance FTIR
spectra of poly(PEGMEMA-1100), poly(PEGMEMA-475), poly(PEGMEMA-300) and
poly(PEGMEMA-475-co-PEGMEMA-1100) films on gold-coated substrates. The ratios are
plotted as a function of the ellipsometric thickness of the film, and for the copolymers, the legend
shows the mole fraction of PEGMEMA-475 (relative to total monomer) in the polymerization
solution. All polymerizations employed a total monomer concentration of 0.67 M, and the
experimental section describes the polymerization conditions. .....................................................54
Figure 2.4: Ratios of C-O-C (1149 cm-1) and C=O (1731 cm-1) absorption coefficients, k, as a
function of the average number of ethylene oxide units per repeat unit, EOavg, in poly(PEGMEMA300), poly(PEGMEMA-475), and poly(PEGMEMA-1100) films. The ethylene oxide units per
monomer were calculated from 1H-NMR spectra of monomers, and the standard deviations in the
absorption coefficient ratios were determined using films with thicknesses ranging from 5 to 454
nm. The line is a fit to the data with a zero intercept. ...................................................................55
Figure 2.5: Reflectance FTIR spectra of (A) 60 nm-thick poly(PEGMEMA-1100) immediately
after polymerization (solid black line) and 24 hours later (dashed red line) and (B) 97 nm-thick
poly(PEGMEMA-475-co-PEGMEMA-1100) immediately after polymerization (solid black line)
and 404 days later (dashed red line). The copolymer was prepared from a solution containing 25
mole% PEGMEMA-475 and 75 mole% PEGMEMA-1100. ........................................................57
Figure 2.6: SEM image of the cross section of a gold-coated porous alumina membrane modified
with a poly(PEGMEMA-475-co-PEGMEMA-1100) film. The alumina exhibits unfilled vertical

xi

pores covered by a 130 nm-thick film that was grown from an aqueous solution containing 25
mole% PEGMEMA-475 and 75 mole% PEGMEMA-1100. ........................................................59
Figure 2.7: Single-gas fluxes through gold-coated porous alumina membranes capped with (A)
poly(PEGMEMA-1100) or (B) poly(PEGMEMA-475-co-PEGMEMA-1100) grown from a
solution containing 50 mole% PEGMEMA-475. (Fluxes are normalized by multiplying by the
SEM thickness of the membrane skin.) The circles represent CO2 data, and the triangles represent
H2 data; filled symbols represent measurements taken immediately after synthesis, and open
symbols represent data obtained 24 h later. The inset in Figure A shows the data on an expanded
ordinate for clarity..........................................................................................................................60
Figure 2.8: Single-gas CO2/H2 selectivity vs. mole fraction of PEGMEMA-475 in the monomer
solution used to create amorphous poly(PEGMEMA-475-co-PEGMEMA-1100) films on porous
alumina membranes. The triangles represent films grown from silylated alumina, and the squares
represent films grown from initiators on gold-coated alumina. For the poly(PEGMEMA-1100)
homopolymer on silylated alumina, the open circle near the origin represents selectivity after
crystallization. Error bars are one standard deviation. ..................................................................62
Figure A.1: Schematic drawing of the reflection of light from (A) a bare gold surface and (B) a
film-covered gold surface. .............................................................................................................66
Figure A.2: Ratios of the C-O-C (1149 cm-1) and C=O (1731 cm-1) absorption coefficients in
poly(PEGMEMA) homo and copolymer films as a function of film thickness. In the case of
copolymer films, the legend shows the PEGMEMA-475 mole percentage of total monomer in the
polymerization solution, and the rest of the monomer was PEGMEMA-1100. The polymerization
was performed as described in the experimental section. ............................................................69
Figure A.3: Proton NMR spectra of (A) PEGMEMA-300, (B) PEGMEMA-475, and (C)
PEGMEMA-1100. Peak areas in these spectra were averaged with areas from two additional
replicate spectra to generate values for the number of average ethylene oxide units per
monomer. .......................................................................................................................................71
Figure 3.1: Cross-sectional scheme of the tortuous path of a permeant molecule traversing a
polymer film containing dispersed, impermeable 2-D nanoparticles. ...........................................79
Figure 3.2: Permeability of a composite film as a function of filler volume fraction as predicted
by the Maxwell equation. The polymer matrix has a permeability of 1 Barrer, and the filler is
impermeable. ..................................................................................................................................80
Figure 3.3: Schematic drawing of the formation of nanoparticles to generate a nanocomposite
polymer brush. ...............................................................................................................................82
Figure 3.4: Reflectance FTIR of poly(PEGMEMA) immediately after synthesis (dashed red line)
and after 24 hours at room temperature (black line). .....................................................................89

xii

Figure 3.5: Surface of poly(PEGMEMA) after a day at room temperature viewed through a crosspolarized optical microscope. ........................................................................................................90
Figure 3.6: Reflectance FTIR spectra of a 44 nm-thick poly(PEGMEMA) film before and after
treatment with TPOT to form nanoparticle films. Spectra were obtained immediately after film
formation before crystallization could occur. ................................................................................91
Figure 3.7: Reflectance FTIR spectra of a 44 nm-thick poly(PEGMEMA)-TiO2 film generated by
soaking in 0.1 M TPOT solution overnight. The spectra taken immediately upon removal from
the TPOT solution (dashed red) and 24 hours later (black) showed little difference. ...................92
Figure 3.8: Permeances of poly(PEGMEMA)-TiO2 membranes as a function of the TPOT
concentration employed for nanoparticle formation. The measurements were taken after
membrane aging overnight at room temperature. The black triangles correspond to CO2 and the
red circles to H2. (The CO2 permeance of the unmodified membrane, 0 M TPOT, was below the
detection limit of the system.) ........................................................................................................94
Figure 3.9: CO2/H2 selectivities of poly(PEGMEMA-1100) membranes as a function of the TPOT
concentration used to modify the membranes prior to aging for one day at room temperature. ...95
Figure 4.1: Reflectance FTIR spectra (3600-2800 cm-1) of (PSS/PHMG)x films adsorbed on
aluminum-coated Si wafers..........................................................................................................109
Figure 4.2: Reflectance FTIR spectra (1800-1000 cm-1) of (PSS/PHMG)x films adsorbed on
aluminum-coated Si wafers..........................................................................................................110
Figure 4.3: Peak heights (1643 and 1008.6 cm-1) in reflectance FTIR spectra of (PSS/PHMG)x
films as a function of the number of adsorbed bilayers, x. ..........................................................111
Figure 4.4: Reflectance FTIR spectra (1800-1000 cm-1) of (PSS/PHMB)x films adsorbed on Alcoated Si wafers. ..........................................................................................................................112
Figure 4.5: Reflectance FTIR spectra of 0.5-6 bilayers of (PSS/PHMB)x films adsorbed on
aluminum-coated Si wafers..........................................................................................................113
Figure 4.6: The peak heights of reflectance FTIR spectra of (PSS/PHMB)x films as a function of
the number of adsorbed bilayers, x. Absorption increases with every sequential layer. ............114
Figure 4.7: Zeta potentials of (PSS/PHMG) multilayers deposited on 50 kDa PES membranes.
Each data point represents an average zeta potential of at least two samples generated on the same
day using the same solutions. The pH refers to the electrolyte deposition solution as described in
the experimental section. (Films with integer numbers of bilayers terminate with PHMG, whereas
fractional bilayer numbers indicate termination with PSS.) ........................................................115

xiii

Figure 4.8: The bactericidal activity of 20 kDa PES membranes coated with (PSS/PHMB) x and
(PSS/PHMG)x films. Activity was calculated by counting the number of colony forming units on
bare and coated membranes [see Eq. (1.2)]. ................................................................................117
Figure 4.9: Membrane pure water permeability as a function of the number of bilayers, x, for
(PSS/PHMG)x- and (PSS/PHMB)x-modified 20 kDa PES ultrafiltration membranes. The pure flux
was measured under a transmembrane pressure of 300 kPa. .......................................................119
Figure 4.10: NaCl rejection in dead-end filtration of 0.01 M NaCl solutions through 20 kDa PES
ultrafiltration membranes coated with (PSS/PHMB)x (black triangles) and (PSS/PHMG)x (red
circles) films.................................................................................................................................120
Figure 4.11: MgSO4 rejection in dead-end filtration of 0.01 M MgSO4 solutions through 20 kDa
PES ultrafiltration membranes coated with (PSS/PHMB)x (black triangles) and (PSS/PHMG)x (red
circles) films.................................................................................................................................121
Figure 4.12: (NH4)3PO4 rejection in dead-end filtration of 0.01 M (NH4)3PO4 solutions through
20 kDa PES ultrafiltration membranes coated with (PSS/PHMB)x (black triangles) and
(PSS/PHMG)x (red circles) films. ................................................................................................122

xiv

LIST OF SCHEMES

Scheme 2.1: Methods for growth of a poly(PEGMEMA) membrane from porous alumina. ......44
Scheme 2.2: Copolymerization of PEGMEMA-475 and PEGMEMA-1100 to provide a membrane
skin. ................................................................................................................................................45
Scheme 4.1: Preparation of an antibacterial polyelectrolyte multilayer film using (a)
poly(styrenesulfonate) (PSS), and
(b) poly(hexamethylene guanidine) (PHMG) or (c)
poly(hexamethylene biguanidine) (PHMB). ................................................................................103
Scheme 4.2: Mechanism of PHMB antibacterial action. (a) The healthy bacterial membrane
contains a mixture of acidic and neutral phospholipids. (b) PHMB adsorbs to the bacterial
membranes through interaction with phospholipds. (c) The affinity of PHMB for acidic
phospholipids leads to sequestration of these phospholipids under the polymer. (d) Lipid
sequestration leads to membrane deformation and eventually cellular leakage and death.
Reproduced with permission from 37 ...........................................................................................104

xv

KEY TO ABBREVIATIONS

ATRP

atom transfer radical polymerization

b

ratio of the rate constants for adsorption and desorption

cb

bulk concentration

c'H

gas capacity constant

cmHg

centimeters of mercury

CRP

controlled radical polymerization

D

diffusivity coefficient

Da

dalton

DK

Knudsen diffusion coefficient

DMF

dimethylformamide

FTIR

fourier transform infrared spectroscopy

GC

gas chromatography

H

enthalpy

HMTETA

1,1,4,7,10,10-hexamethyltriethylenetetramine

i.d.

inner diameter

J

flux

K

partition coefficient

k

mass transfer coefficient

k

absorption coefficient

kact

rate constant for the radical activation formation

KATRP

equilibrium constant for the activation/deactivation reaction

xvi

kdeact

rate constant for the deactivation reaction

kp

rate constant for polymerization

kt

rate constant for termination

l

thickness

L

ligand

[M]

monomer concentration

Mn

number average molecular weight

mol

mole

N

number of species

ñ

complex refractive index

N2

nitrogen

NMR

nuclear magnetic resonance

P

permeability coefficient

p

partial pressure

Pa

Pascal

PEG

poly(ethylene glycol)

PEO

poly(ethylene oxide)

poly(PEGMEMA)

poly(ethylene glycol) methyl ether methacrylate

ppm

parts per million

r

radius

R•

radical terminated chain

Ra

non-ideal adsorption resistance

xvii

Rb

non-ideal pore blocking resistance

Rc

non-ideal cake layer resistance

RCO

absorbance ratio for copolymer

Rcp

non-ideal concentration polarization resistance

Rm

ideal membrane resistance

RO

reverse osmosis

Rtot

total resistance to mass transport

R-X

halogen-capped dormant chain

S

solubility coefficient

s

distance away from pore wall

S

position at a distance x along a pore and s from a pore wall

STP

standard temperature and pressure

T

temperature

Tc

critical temperature

Tg

glass transition temperature

THF

tetrahydrofuran

TM

transition metal

ū

molecular velocity

Vg

glassy volume

Vl

theoretical densified volume

x

mole fraction of a given molecule

x

distance along a pore

z

coordination number around a given molecule

xviii

α

selectivity

β

a given angle from the pore wall

δ

solubility parameter

δ

stagnant layer thickness

ΔGM

Gibbs' free energy of mixing

ΔHM

enthalpy of mixing

ΔHvap

enthalpy of vaporization

Δp

pressure gradient

ΔSM

entropy of mixing

ε

porosity

η

viscosity

λ

mean free path

σ

number of segments in a given polymer chain

τ

tortuosity

φ

volume fraction of a given component

ΦN2

sweep gas flow rate

χ

Flory-Huggins interaction parameter

Ω

number of distinguishable arrangements of molecules

ω

interaction energy between two molecules

xix

Chapter One
Separations Using Polymer Membranes
An Overview
Chemical separations are at the heart of the chemical and petrochemical industries and
represent some of the most costly and energy-intensive steps in industrial production. For
example, a recent U.S. DOE target for CO2 sequestration set a goal of 90% CO2 sequestration at a
cost increase of not more than 30% of the cost of power plant electricity. 1 The CO2 separation
process accounts for 60-80% of the overall isolation, transportation, and storage process needed to
accomplish this goal.2
Industrial separations make use of a wide range of methods, but the physical and chemical
differences among the species being separated limit the choice of the purification technique. Table
1.1 lists some important separation processes and the properties that they exploit.3 Combinations
of separation techniques are also possible.
Over the past 40 years, membrane-based separations have emerged as industrially viable
alternatives to traditional processes such as distillation.4 In these separations, a membrane divides
two bulk phases between which selective transport occurs (Figure 1.1).4

1

Table 1.1: Physical/chemical properties exploited in chemical separations and the corresponding
separation techniques. Adapted and modified from Mulder.3
physical/chemical property

separation process

size

microfiltration, ultrafiltration, dialysis, gas separation, gel
permeation chromatography

vapor pressure

distillation, membrane distillation

freezing point

crystallization

critical point

gas separation

charge

ion exchange,
electrophoresis

density

centrifugation

affinity

extraction, adsorption, absorption, reverse osmosis, gas
separation, pervaporation, affinity chromatography

reactivity

complexation, carrier mediated transport

electrodialysis,

diffusion

dialysis,

The membrane interphase can consist of a nonporous or porous solid, a porous solid with
filled pores, or a liquid or gel, and transport through the membrane can take place under driving
forces such as pressure, temperature, electrical potential, and concentration gradients.4, 5
For a specific separation, the applicability of a given membrane depends on its selectivity
and permeability. High permeabilities are important for achieving a reasonable throughput at
moderate driving forces, whereas high selectivities are vital for obtaining a pure product. In many
cases, permeability and selectivity are inversely related, so optimization of a separation involves
choosing the membrane material with the most economical combination of selectivity and
permeability.

2

Membrane
Phase 1

Phase 2

Figure 1.1: Schematic drawing of a membrane separating two phases.
Quantitative comparison of the energy and operating costs of membrane processes with the
corresponding costs in other separation methods is sometimes difficult and often depends on the
scale of the operation and the required purities. Generally, the advantages of membranes are low
energy consumption, continuous operation, and a small footprint.3 On a small scale, membranes
are typically much easier to install than processes such as pressure swing adsorption. 6 In water
desalination, the low energy requirements of reverse osmosis make it more attractive than
distillation. However, the drawbacks of membrane separations can include low selectivity,
fouling, and concentration polarization.3
This dissertation focuses on developing new membrane materials that may overcome
drawbacks in specific membrane separations, including low selectivity in CO2 removal from H2
streams (chapters 2 and 3) and membrane biofouling in water purification (Chapter 4). To put this
work in perspective, this introduction provides an overview of membrane theory and methods of
transport. The solution diffusion model is discussed as well as permeation though porous
membranes. That section is followed by a discussion of non-ideal permeation highlighting
polarization and fouling, followed by membrane synthesis.

3

Classification of Membranes Processes
Depending on their pore sizes, membranes can separate a wide range of species, and Figure
1.2 shows the standard classifications of membrane separations along with the corresponding
diameters of the retained species. Figure 1.3 illustrates how the mode of permeation varies with
both the membrane pore size and the diameter of the species passing through the membrane. There
are essentially four mechanisms of transport through a membrane: solution-diffusion, molecular
sieving, Knudsen flow, and Poiseuille flow.
The research presented here focuses on dense, nonporous membranes as well as
nanofiltration membranes. Nanofiltration membranes represent the transition between porous
media and dense polymers, and proposed transport models include both solution diffusion and
molecular sieving.7, 8 In contrast, permeation through dense, gas-separation polymers most likely
occurs though a solution diffusion mechanism (see below), although in special cases facilitated
transport may occur and provide very high selectivities. Because of the emphasis of this work on
dense membranes, the following sections provide a detailed analysis of the solution diffusion
model and much shorter descriptions of transport through porous membranes.

4

5

Non- Porous Membranes
Molecules adsorb into,
diffuse across, and desorb
from the membrane.
Porous Membranes
Molecular sieving
excludes molecules larger
than pore diameter.
Knudsen flow through
pores with diameters less
than the mean free path
of gases provides
preferential diffusion of
lighter molecules.
Poiseuille flow through
large pores with no
separation.

Figure 1.3: Schematic diagram of different types of transport through porous and non-porous
membranes. Adapted from Ho and Sirkar.4

Solution-diffusion permeation through dense membranes
RO and membrane-based gas separations typically employ dense membranes to achieve
selective transport of specific ions or small molecules such as water and gases. The development
of increasingly selective polymeric membranes for gas separation accompanied advances in the
theory of gas transport, and the table below shows a few historical highlights that lead to our
current understanding of gas separation membranes, described below.

6

Table 1.2: Selected milestones in understanding gas separations. Modified from reference 5.
Event

Investigator

Year

First gas separation recorded

Graham

1829

Law of mass diffusion

Fick

1855

Permeability coefficient defined

von Wroblewski

1879

Knudsen diffusion defined

Knudsen

1908

Time-lag method to determine
coefficients developed

Daynes

1920

Permeability and diffusivity
models developed

Barrer

1939

Independent arrival at dual-mode
sorption concept

Barrer, Barrie, and Slater

1958

In the solution-diffusion model for both liquid and gas separations, the permeation process
involves dissolution of the transporting species into the membrane, diffusion across the membrane,
and transfer into the receiving phase from the permeate face of the membrane (Figure 1.4). The
amount of a species that flows through a given area of a membrane over a given time period is
termed the flux, J, which has typical units of (mol m-2 s-1) or (m3 m-2 s-1). In the absence of
convection and migration, flux occurs by diffusion and follows Fick’s first law, which is shown in
Eq. (1.1) for a one-dimensional concentration gradient,

J  D

dc
dx

dc
, and a diffusion coefficient, D.
dx
(1.1)

The assumptions of a constant diffusion coefficient across a membrane of thickness, l, along with
steady state flux lead to Eq. (1.2),

7

JA 

DA (c2, A  c3, A )

(1.2)

l

where c2,A is the concentration of A just inside the feed face of the membrane and c3,A is the
concentration of A just inside the permeate face of the membrane. If the feed and permeate faces
of the membrane are in equilibrium with the feed and permeate solutions, respectively, a partition
coefficient KA, Eq. (1.3), describes the ratio of the concentrations in the faces of the membrane
and the neighboring solutions. (See Figure 1.4 for the definition of the symbols).

KA 

c1, A



c4, A

(1.3)

c3, A

c2

Concentration

c1

c2, A

Flux

c4
Feed

c3

Membrane Permeant

Figure 1.4: Concentration profile of an analyte during solution-diffusion through a membrane,
where c1, c2, c3, and c4 represent the concentrations of the analyte in the feed, upstream face of the
membrane, downstream face of the membrane, and permeate, respectively. (This diagram neglects
concentration polarization, which is discussed below.)
Combining Eq. (1.3) and Eq. (1.2) yields Eq. (1.4), where PA, the product of the diffusion and
partition coefficients, is the permeability for species A.

8

JA 

DA K A (c1, A  c4, A )
l



PA (c1, A  c4, A )
l

(1.4)

This expression shows that increases in diffusion and partition coefficients, increases in
concentration gradients, and smaller membrane thicknesses all lead to higher fluxes.
Eq. (1.5) defines the selectivity (  A ) of membrane separations, and in the ideal case where
B

the permeate concentration is negligible compared to the feed concentration, this selectivity is also
equivalent to the ratio of the permeabilities of components A and B.

A 
B

c4,A / c4,B

(1.5)

c1,A / c1,B

However, when the permeate concentrations are not negligible, Eq. (1.6) provides a more general
expression for the ratio of the permeability coefficients.

c4,A / c4,B
(c1,A  c4,A ) / (c1,B  c4,B )



PA
PB

(1.6)

In the case of gas separations Henry’s law, Eq. (1.7), often replaces the partition coefficient
expression in Eq. (1.3) to afford a description of flux as a function of feed and permeate vapor
pressures. Figure (1.5) describes the specific symbols in Henry’s law.

c2,A  SA p1,A

(1.7)

c3,A  SA p4,A
Substitution of Eq. (1.7) into Eq. (1.2) yields

JA 

DA S A ( p1, A  p4, A )
l



PA ( p1, A  p4, A )
l

(1.8)

where the permeability is a product of the diffusion coefficient, DA, and the Henry’s law solubility
constant, SA, as shown in Eq. (1.9).

9

PA  S A  DA

(1.9)

The solubility is a thermodynamic quantity that depends on the interaction of the gas
molecule with the membrane polymer, whereas the diffusion coefficient reflects the kinetics of gas
movement through the polymer. An increase in the intermolecular interactions between the
penetrant and the membrane causes an increase in solubility and, hence, an increase in
permeability. Similarly, an increase in diffusivity due to greater

Concentration

p1

c1
Flux

p2

Feed

c2
Membrane Permeant

Figure 1.5: Concentration profile of a penetrant gas in a membrane from c1 at the feed face to c2 at
the permeant face. The partial pressure levels are shown for the feed (p1) and permeant (p2) gas.
free volume in the polymer also causes an increase in permeability. In the case of Eq. (1.9),
permeability has SI units of (m3×m×m-2×s-1×Pa-1), but the most common engineering unit for P is
the Barrer (10-10 cm3×cm×cm-2×s-1×cmHg-1).
Eq. (1.10) defines the ideal selectivity (negligible permeant partial pressure) for gas
molecule (A) over (B) where

10

 A/ B 

PA S A  DA

PB S B  DB

(1.10)

S 
the ratio of permeabilities is the product of solubility selectivity  A  and diffusivity selectivity
 SB 
 DA 

 . By convention component A is the more permeable species. There are three possible
 DB 
causes for a given selectivity,

 A ,: 1) A has a larger diffusivity, 2) A has a larger solubility, or
B

3) A has both a larger solubility and diffusivity. Table 1.3 presents these scenarios for several
different gas separations.
Table 1.3: The three possible combinations of solubility and diffusivities that lead to selective gas
transport, along with gas separations that employ these scenarios. Adapted from Pinnau and
Freeman.10
Scenario
SB > SA, DA >> DB

Examples
H2/N2, H2/CO, H2/CH4

Membrane type
size sieving/size selective

SA >> SB, DB > DA

CO2/H2, H2S/H2

reverse selective

SA ≥ SB, DA ≥ DB

O2/N2, H2O/Air,
H2S/CH4, CO2/N2

CO2/CH4, either size selective or reverse
selective

The diffusivity selectivity-dominated scenario, row 1 in Table 1.3, employs traditional
size-selective membranes to achieve a higher permeability of gas A than gas B, as shown in Figure
1.6. Diffusivity selectivity nearly always favors the smaller gas. In contrast, the solubility
selectivity-dominated case, row 2 in Table 1.3, uses a “reverse selective” membrane to allow
passage of the larger of two gases (Figure 1.7). The third scenario, row 3 in Table 1.3, represents
a very desirable situation where both solubility and diffusivity selectivity favor the overall
selectivity.

11

Feed

Dp

Permeant

Selective skin
Porous support
Figure 1.6: Schematic diagram of a size-selective separation with a composite membrane. The
membrane has higher permeability for the smaller molecules due to diffusivity selectivity or
molecular sieving. The low pressure side (right) of the membrane contains a higher fraction of
small molecules than the feed side.
.

12

Feed

Dp

Permeant

Selective skin
Porous support
Figure 1.7: Schematic diagram of reverse selectivity. In this case, the membrane has higher
permeability for larger molecules due to solubility selectivity. The lower pressure permeate side
(right) of the membrane contains a higher fraction of the larger molecules than the feed side.
As an example of different selectivity mechanisms, CO2/CH4 separations can utilize both
diffusivity and solubility selectivity. Cellulose acetate membranes exhibit CO2/CH4 selectivities
of 20-30 in natural gas purification primarily because of the high solubility of CO2 in cellulose
acetate.11 In contrast, the high diffusivity selectivity of glassy polyimide membranes provides the
same separation with a remarkable overall CO2/CH4 selectivity of about 90.12 The high diffusivity
selectivity of the polyimide membranes exploites a size difference of only 0.5 Å (Table 1.4). In
fact, Table 1.4 shows that a size range of 2.6 to 3.8 Å covers many common gases.
Typically, glassy polymers are effective in size-selective separations, whereas rubbery
polymers may show reverse selectivity. Glassy polymers have an “unrelaxed” volume (Figure
1.8) that stems from polymer chains trapped in nonequilibrium conformations upon quenching
below

Tg.

13

This

“unrelaxed” volume leads to two types of sorption: Henry’s law adsorption in areas of dense
polymer (similar to adsorption in rubbery polymers) and Langmuir adsorption in the “holes”
between the polymer chains.
Table 1.4: Kinetic diametersa and critical temperatures of various gases.13

a

Molecule

Kinetic Diametera (Å)

Critical Temperature (○C)

He

2.6

-268

H2

2.89

-240

CO2

3.3

31

O2

3.46

-119

N2

3.64

-147

CO

3.76

-140

CH4

3.8

-82

The Kinetic diameter is the minimum zeolite window diameter that permits sorption.

14

V (cm3/mol)

Glassy volume (Vg)
Vg-Vl

Theoretical densified
volume (Vl)

Glass Rubber

Tg
Temperature
Figure 1.8: Polymer volume relative to temperature. Modified from Zolandz and Fleming.14
Thus, Eq. (1.11) describes the total concentration, c, of adsorbed gas as the sum of

c  cD  cH

(1.11)

the sorbed concentration in dense regions of the polymer, cD, and the sorbed concentration in holes,
cH, between the dense regions.14 The concentration in the dense region follows Henry’s law, Eq.
(1.7), and the concentration of adsorbed gas in preexisting holes follows the Langmuir isotherm in
Eq. (1.12) 14

cH 
where

cH

cH bp
1  bp

(1.12)

is the capacity constant, b is the ratio of the rate constants for adsorption and

desorption, and p is the pressure of a single gas. This additional sorption mode results in glassy
polymers typically possessing higher permeabilities than rubbery polymers.10

15

(The main

exception is poly(dimethyl siloxane), which is a rubbery polymer with a CO2 permeability over
3,000 Barrer.)
One important consideration when using glassy polymer membranes that rely on diffusivity
selectivity is whether extensive sorption of a condensable gas such as CO2 will plasticize the
membrane and alter the selectivity. For example, in poly(vinyl chloride), a highly selective glassy
polymer, the diffusivities of gases with critical volumes ranging from 65 – 370 cm3/mol (hydrogen
to n-hexane) vary by ten orders of magnitude.15, 16 However, in plasticized poly(vinyl chloride)
membranes, the same molecules have diffusivities that vary by only two orders of magnitude.15
Penetrant Solubility
Gas condensability, as reflected by critical temperatures, strongly correlates with penetrant
solubility in gas separations. Low critical temperatures indicate strong intermolecular interactions
among gas molecules, and such gases also interact strongly with polymers. Several studies suggest
that at a given temperature the solubilities of different gases in a particular polymer vary
logarithmically with the critical temperature (Tc) of the gas according to Eq. (1.13),17, 18

ln S  M  NTc

(1.13)

where M and N are constants. Table 1.4 shows that the critical temperatures of common gases
vary tremendously, from -268 to 31○C, and solubility selectivity can easily reach values of 100.19
More recently, Freeman and coworkers presented data indicating that the logarithm of solubility
is proportional to the square of the critical temperature, shown in Eq. (1.14).20-22 They suggested
that solubility is influenced by both entropy and enthalpy
ln S A  M  N Tc 

16

2

(1.14)

which causes a deviation from the linear behavior in Eq. (1.13).23 Larger penetrant species require
larger cavities, which leads to an entropic penalty opposing solubility. However, polymerpenetrant interactions are stronger for larger species, so enthalpy changes favor the solubility of
larger gases. These conflicting contributions may explain why trends in solubility for certain gases
follow Eq. (1.13), whereas other groups of gases follow Eq. (1.14).
The solubility can be expressed in terms of the energy required to transfer one mole of
penetrant gas from the bulk gas phase into the membrane according to Eq. (1.15)23

S

T0
exp[DGM / RT ]
TP0

(1.15)

where S is solubility, T0 is 273.15 K, T is the temperature, P0 is 1 atm, and DGM is the Gibbs’ free
energy of mixing. The following section examines the thermodynamic parameters that give rise to
the Gibbs’ free energy of mixing and influence the solubility of a penetrant in a polymer.
Once a penetrant approaches a membrane surface, the probability of dissolution into the
membrane matrix depends on the Gibbs free energy of mixing ( GM ). Mixing spontaneously
occurs when the change in energy is negative ( DGM  0 ). For the mixing reaction A  B  AB
the change in Gibbs free energy is

DGM  GAB  (GA  GB )

(1.16)

where GAB is the Gibbs free energy of the mixture of A and B, and GA and GB are the Gibbs free
energies of pure A and pure B, respectively. Of course the Gibbs free energy, has both enthalpic
and entropic contribution as shown in Eq. (1.17),

DGM  DH M  T DSM

17

(1.17)

where HM is the enthalpy of mixing, T is temperature, and SM is the entropy of mixing. The entropy
change upon mixing is not ideal due to the different sizes of polymer and solvent molecules, as
well as chain connectivity. The most accurate derivation of the entropy of mixing considers all
distinguishable spatial arrangements of the penetrant and polymer molecules.

Boltzmann’s

equation, Eq. (1.18)24

 
DSmix  k ln  AB 
 B 

(1.18)

considers a lattice model which relates the entropy to the number of distinguishable arrangements
of pure polymer molecules (  B ) and mixture of penetrant and polymer molecules (  AB ) in the
lattice, where k is Boltzmann’s constant. The expression for  AB is a continuous product of all
possible conformations of polymer molecules, and the expression has been simplified by Flory to
yield the following Eq. (1.19)24

DSM  k[ N1 ln 1  N2 ln 2 ]

(1.19)

where k is Boltzmann’s constant, NA and NB are the number of solvent species A and polymer
species B respectively, and  A and B are the volume fraction of component A and B, respectively.
The volume fraction is given by the fraction of lattice points occupied, Eq. (1.20)

A 

NA
 N A   NB 

 NB
B 
 N A   NB 

(1.20)

Where NB polymer molecules contain  segments in the polymer chain. This assumes that
each segment,  , is equal in size to a penetrant molecule. Eq. (1.19) can be rewritten in terms of
moles, n, as shown in Eq. (1.21)24

18

DSmix   R[nA ln A  nB ln B ]

(1.21)

Having derived an expression for entropy, the second stage is to derive an expression for
enthalpy. The enthalpy of pure gases is shown in Eq. (1.22)
zN A AA
2
zN 
 B BB
2

H AA 
H BB

(1.22)

where HAA and HBB are the enthalpies of pure A and pure B, respectively, z is the coordination
number, N is the number of molecules, and  AA and BB are the interaction energy between two A
molecules and two B molecules, respectively. The 2 accounts for the fact that one interaction
involves two species. The enthalpy for species A adjacent to species B is

H AB  N AB AB 

( zN A  N AB ) AA ( zN B  N AB )BB

2
2

(1.23)

where HAB is the enthalpy of an AB mixture and  AB is the interaction energy between A and B.
To calculate the enthalpy of mixing, Eq. (1.22)and (1.23) are inserted into Eq. (1.24),

DH mix  H AB  H AA  H BB

(1.24)

which gives the difference in the enthalpy of the products and reactants to generate Eq. (1.25)

DH mix  N AB AB 

( zN A  N AB )AA ( zN B  N AB )BB zN AAA zN BBB



2
2
2
2

(1.25)

Simplification of Eq. (1.25) yields the following enthalpy of mixing,
DH mix  N AB [ AB 

 AA  BB
2

19

]  N AB D AB

(1.26)

which states that the enthalpy of mixing is the number of interaction sites multiplied by the mixture
interaction energy less the pure-substance interaction energy. The number of interaction sites NAB
is approximated below

N AB  zN A xB

(1.27)

By multiplying the interaction sites around an A molecule (z), the number of A molecules (NA),
and the mole fraction of B molecules (xB).
Having shown the enthalpy and entropy, substitution of Eq. (1.26) and Eq. (1.21) into Eq.
(1.17) yields the expression for the Gibbs free energy of mixing in Eq. (1.28).

DGmix  RT [nA ln xA  nB ln xB 

znA N Avo xB DAB
]
RT

(1.28)

This is usually rewritten in terms of the Flory-Huggins interaction parameter, χ, as defined below.

 AB 

zN AvoDAB
RT

AB
 zDkT

(1.29)

It follows from Eq. (1.29) that  is a dimensionless ratio of interaction energy of the polymer
with the penetrant to the kinetic energy RT. Now DGmix is rewritten as Eq. (1.30).

DGmix  RT [nA ln xA  nB ln xB  nA xB  AB ]

(1.30)

This interaction parameter, χ, has the physical meaning of a measure of the affinity of a polymer
for a penetrant.
The Flory-Huggins interaction parameter of the penetrant is useful in the selection of
membrane materials for specific separations. Of particular relevance to this dissertation, the CO2
interaction parameter provides some of the rationale for the selection of the CO2-selective
membrane materials described in chapters 2 and 3. Several studies determined CO2 interaction

20

parameters with different polymers, but there are substantial discrepancies in the reported  values.
The interaction parameter for a penetrant is usually determined by dissolving the penetrant in a
series of solvents with known solubility parameters and plotting the solubility against these solvent
parameters to find a maximum.

The solubility parameters for liquids are defined as

  (DH vap  RT ) / V where DH vap is the enthalpy of vaporization and V is the molar volume at
25 °C. This definition is not valid near the critical temperature and therefore not appropriate for
CO2 near room temperature. Lawson and coworkers estimated the CO2 interaction parameter to
be 6.8 MPa0.5 by separately measuring the solubility of CO2 in a range of solvents with solubility
parameters ranging from 12 to 20 MPa0.5 and finding the maximum on a plot of solubility vs.
solubility parameter.25 This value of the interaction parameter is less than that of H2 and N2, which
is counterintuitive given that CO2 contains a quadrupole moment. In a different study, the CO2
interaction parameter obtained by dissolving CO2 in toluene was 12.2 MPa0.5.26 Two studies in
2005 from Benny Freeman et al. reported a  value of 26.8 MPa0.5 for CO2 in cross-linked
poly(ethyleneglycol)diacrylate from temperature studies,27 and 21.8 MPa0.5 from dissolving CO2
in organic solvents with solubility parameters ranging from 15 to 30 MPa0.5.19 An accurate
determination of the CO2 solubility parameter is necessary to determine a membrane material with
maximum CO2 solubility.
Fractional free volume of the polymer also correlates with permeability because increases
in free volume sometimes lead to higher solubilities and diffusivities. Introduction of side chains
that disrupt polymer packing typically increases free volume and enhances permeability. For
example, Benny Freeman introduced phenyl groups to disrupt chain packing of rigid polyamides
and increased the fractional free volume by up to 7 %.28 Additionally, cross-linking increases
fractional free volume in PEG-based membranes, causing a modest CO2 solubility increase from

21

1.3 cm3(STP)/(cm3atm) in amorphous PEO to 1.4 cm3(STP)/(cm3atm) in crosslinked PEOdiacrylates.27

However, the major achievement of crosslinking is the prevention of PEO

crystallization. Crystalline PEO has a very low fractional free volume and a CO2 solubility of only
0.4 cm3(STP)/(cm3atm). Additionally, permeability can decrease with decreasing fractional free
volume. Permeability decreased by a factor of three through decreases of fractional free volume
by nitrating polysulfone membranes.29
Permeation through porous membranes
Permeation through a porous membrane occurs through either Knudsen diffusion or
Poiseuille flow. In Knudsen diffusion, collisions of molecules with the pore walls control the rate
of transport, whereas in Poiseuille flow the channel is large enough that laminar flow occurs. The
relative contributions of these two transport mechanisms depend on the pore radius and the mean
free path of the permeant (Figure 1.4). Specifically, when the ratio of the pore radius (r) to the
mean free path (λ) is much less than one, Knudsen flow occurs; when r/λ is approximately two,
the Knudsen and Poiseuille flows contribute equally to transport; and when r/λ is much larger than
one, Poiseuille flow occurs (r/λ of 5 is >90% Pouiseuille flow).5 Eq. (1.31) provides an expression
for the mean free path, , of a gas as a function of

λ=

3η πRT
2P 2M

(1.31)

the molecular mass (M), pressure (P), viscosity (), and temperature (T) of the permeating species,
where R is the gas constant. At atmospheric pressure, the mean free path of gases is around 100200 nm,5 so if the membrane pore diameter is less than 50 nm, transport occurs primarily due to
Knudsen flow.

22

The frequency of collisions between a gas and the pore wall varies with the molecular
velocity, u , which is a function of the gas molecular mass (Eq. 1.32).

u=

8RT
πM

(1.32)

Eq. (1.33) describes Knudsen’s diffusion coefficient, DK,5

DK 

u
4

r 2 


0

x 2 dx
2
0  (s 2  x2 )2 s d  ds /  dS

(1.33)

in an infinitely long pore with radius r, where the position S is at a distance x along the pore, that
is a distance s away from the pore wall at an angle (β) from the pore wall. Integration of Eq. (1.33)
yields Eq. (1.34),5
DK 

2ur
3

(1.34)

And substitution of Eq. (1.32) into Eq. (1.34) yields the final expression for DK.

DK 

4r 2 RT
3 M

(1.35)

Eq. (1.35) is valid for a single cylindrical pore, but most membranes contain an assembly of pores
that may offer a tortuous path through the membrane. In this case, Eq (1.36) describes the effective
Knudsen diffusion coefficient,

DK 

4r
3

2 RT
M

(1.36)

where ε is porosity and τ is tortuosity. Porosity, the fraction of the membrane volume occupied
by pores, ranges from 0 to 1, whereas tortuosity is greater than 1. Unfortunately, tortuosity is
difficult to calculate in complex membranes and frequently serves as a correction factor.

23

The combination of the Knudsen diffusion coefficient [Eq. (1.36)] and Fick’s law [Eq.
(1.1)] leads to Eq. (1.37), an expression for flux through the membrane

J

where

4r
3

2 RT
M

 n0 n1  1
  
V V l

(1.37)

n0
n
is the moles of gas per volume at the feed end of the pore and 1 is the moles per volume
V
V

at the permeate end of the pore. Using the ideal gas law, pV  nRT , Eq. (1.38) expresses flux

J

4r
3

2 RT p0  p1
 M lRT

(1.38)

in terms of the partial pressure at the feed side of the pore, po, and at the permeate side of the pore,
p1. Application of this equation to calculate the ideal selectivity for one gas over another, results
in the cancellation of all variables except molecular mass, so selectivity is simply inversely
proportional to the square root of molecular masses, Eq. (1.39).

A 
B

MB
MA

(1.39)

Knudsen selectivity is relatively low, particularly, if H2, which has the lowest mass of any gas, is
not one of the gases. Even in this case, the Knudsen selectivity for H2/CO2 is only 4.7.
As the the ratio of r/λ becomes much larger than one, Poiseuille flow dominates transport,
and Eq. (1.40) describes the total volume flux through the membrane.
2
2
2
r 2  P0  P1  P0  P1  r  P0  P1 
J

8lRT
8lRT

(1.40)

In Poiseuille flow, separation of molecules no longer occurs because they interact with each other
more than with the pore walls.

24

In the application of all selective membranes, whether porous or nonporous, there are
several complications that cause deviations from ideality. Below I describe the most common
complications, concentration polarization and fouling.
Non-Ideal Membrane Performance
In most membrane separations, particularly those that employ liquids, selectivity and flux
decline with time. The pressure-driven convective flux can be written as3

J

Dp
 Rtot

(1.41)

where  is the viscosity and Rtot is the total resistance to mass transport. The value of Rtot is the
sum in Eq. (1.42)

R tot  R m  R c  R b  R cp  R a

(1.42)

the ideal membrane resistance (Rm), and the non-ideal resistances due to cake layer formation (Rc),
pore-blocking (Rb), concentration polarization (Rcp), and adsorption (Ra).3 Figure 1.9
schematically shows the origin of these resistances, and a membrane system could contain any
combination of these resistance types. Cake layer formation typically occurs with proteins, or
other solutes such as colloids, that form gels at increased concentrations. In porous membranes a
solute may physically block a pore and generate Rb or adsorb in a pore without complete blocking
to give Ra. The next section discusses resistance due to the occurrence of concentration
polarization, followed by the section which describes biofouling, or adsorption of microorganisms.

25

Feed

Membrane

Permeant

Rb
Ra

Rcp Rc Rm
Figure 1.9: Overview of various phenomena that generate resistance to convective flux. The
various resistances stem from the ideal membrane-Rm, cake layer formation-Rc, pore blocking-Rb,
concentration polarization-Rcp, and adsorption-Ra.
Concentration Polarization. In many membrane separations, the accumulation of rejected
species near the feed side of the membrane decreases both rejection and flux. This phenomenon,
known as concentration polarization, is particularly problematic in liquid-based separations where
diffusion coefficients are relatively low. In a typical membrane separation, convection moves all
species toward the membrane where some of the species are rejected more than others.
During filtration, the concentrations of the rejected species at the membrane-feed interface
initially increase until the flux due to permeation is equal to the sum of convective flux to the
membrane and back diffusion from the membrane surface (Figure 1.10).

26

(a)

c3
Feed

c4

Bulk
Feed

Permeant

Membrane

cb

c 1 c2
Boundary Layer

Concentration

c2
Membrane

Concentration

c1

(b)

c3

c4
Permeant


Figure 1.10: Simplified scheme of a concentration profile (a) under a concentration driven
membrane separation with perfect mixing and (b) with concentration polarization at the feed
interface under steady-state conditions. The concentration at c1 does not necessarily equal c2, and
c3 does not necessarily equal c4. Note that in the case of concentration polarization the
concentration begins to increase at a distance  , along the x-axis, from the membrane surface.
Eq. (1.43) describes this mathematically
Jc  D

dc
 Jc4
dx

(1.43)

where J is the convective volume flow toward the membrane, c is the local concentration, - D

dc
dx

is the diffusive backflow away from the feed face, and c4 is the concentration in the permeate. The
volume flux is assumed constant throughout the system. Integration of this equation over the
stagnant layer thickness, , from c=cb to c=c1 yields Eq. (1.44)
ln

c1  c4 J 

cb  c4
D

27

(1.44)

where the ratio

D



is often termed the mass transfer coefficient, k. (In fact, the mass-transfer

coefficient is more general than

D



.) In the case of a theoretically perfect membrane where the

solute is completely rejected by the membrane and the permeate concentration (c4) is 0, Eq. (1.44)
becomes

c1
J
 exp  
cb
k

(1.45)

This equation illustrates that the extent of concentration polarization depends on the solute
diffusion coefficient and the hydrodynamic properties that determine k, as well as the flux, which
varies with both pressure and the type of membrane. In general concentration polarization
decreases rejection because of the high solute concentration at the surface.3 (The consequences of
concentration polarization can be quite complicated in certain applications, i.e. when retained
macromolecules form a dynamic secondary membrane that increases rejection of lower molecular
weight solutes.)3 In the case of gas separations, the large diffusion coefficients of gases prevent
significant concentration polarization, but microfiltration and ultrafiltration suffer large flux
declines due to this phenomenon.
Partial remedies for concentration polarization include an increase in operating temperature
or mixing near the surface to increase the mass transfer coefficient, k. A dead-end membrane
module (Figure 1.11a) is the simplest separation design, but most membrane separations employ
a cross-flow process (Figure 1.11b) to limit the thickness of the boundary layer and reduce
concentration polarization. The cross-flow method generates both retentate and permeate streams,
where the product can be either the concentrated solute in the retentate or the purified permeate.

28

Industrial applications typically utilize spiral wound modules or hollow fiber bundles to maximize
the area of membrane per module volume.
(a)
dead-end membrane module

(b)
cross flow membrane module
feed

feed

retentate

membrane

membrane

permeate

permeate

Figure 1.11: Schematic drawings of (a) a dead-end membrane and (b) a cross-flow membrane
module. In the dead-end mode the concentrations of rejected species continually increase on the
feed side of the membrane, whereas the cross-flow module allows for rejected species to flow
along the membrane face and out of the module via the retentate path.

Biofouling. The decrease in membrane performance due to concentration polarization may
reach a steady state, but fouling of leads to a continuous decline in flux (Figure 1.12). The two
phenomena are often related, however, as the stagnant solution in the fouling layer may lead to
increases in concentration polarization.

29

Flux

Concentration
polarization

Fouling

Time
Figure 1.12: Decreases in transmembrane flux over time, due to concentration polarization and
fouling.
Fouling mechanisms include cake or scale formation3 and microorganism adsorption.30
Fouling due to microorganisms, biofouling, is particularly problematic because bacteria multiply
and cause significant decreases in membrane permeability.30 Moreover, bacterial deposition of
extracellular polysaccharides further decreases permeability and allows attachment of additional
microorgranisms.30 The initial step in biofouling, cell adhesion, depends on substrate properties
such as roughness,31-33 topography,34 free energy,35, 36 and polymer composition.37-39 Methods to
modify membranes to resist biofouling include incorporation of antibacterial metal particles,40-42
and surface modification with ammonium groups.43-47 Complementary methods to externally
combat membrane fouling include pretreatment of the feed solution48,

49

and intermittent

membrane cleaning.50, 51 Chapter 4 discusses our attempts to create antifouling membranes.
Synthesis of Selective, High-Flux Membranes
The inverse relationship between flux and thickness, Eq. (1.2), translates to lower operating
pressures for thinner membranes, so commercial separation membranes consist of a thin, selective

30

skin on a highly permeable porous support layer. The support provides mechanical strength,
whereas the skin imparts selectivity while allowing a reasonable flux because of its small
thickness.

Methods for preparing such gas separation membranes with thin skins include

casting,52, 53 spin-coating,54 spray-coating,55, 56 phase-inversion,57, 58 polyelectrolyte deposition,59
interfacial polymerization,60-62 and surface-initiated polymerization.63

Surface-initiated

polymerization, the main focus of this research, involves attachment of an initiator to a surface and
subsequent polymerization from the immobilized initiator. The thickness of the membrane skin
varies with polymerization time and can range from tens of nanometers to several microns.
Initiator attachment typically occurs via reaction of a silane initiator with an oxide/oxidized surface
or adsorption of a polyelectrolyte macroinitiator.64 For research purposes, initiator layers can form
through self assembly of thiol monolayers of initiators on Au-coated membranes.65
Growth of polymers from surfaces can occur via anionic,66 cationic,65, 67 free radical,68-70
ring opening,71 ring opening metathesis,72-76 or controlled radical polymerization (CRP).77-85
CRPs, which are discussed in a number of review articles86, 87 and books,88 include nitroxide
mediated polymerization,78-81 reversible addition fragmentation transfer,82-84 and atom transfer
radical polymerization (ATRP).85

Surface-initiated ATRP is our preferred method of

polymerization because it occurs at room temperature and provides minimal solution
polymerization, control over thickness, and the ability to polymerize a wide range of functional
polymers.
Atom transfer radical polymerization. Fig. 1.13, shows the reactions that constitute the
ATRP mechanism. A number of transition metals including Ti,89 Fe,90-95 Co,96 Ni,97-99 Mo,100-102
Ru,103-107 Rh,108, 109 Pd,110 Re,111 and Os,112 can mediate ATRP, but Cu is the most common
catalyst for this reaction.

31

Figure 1.13: Reactions involved in ATRP. The ATRP mechanism contains an equilibrium where
a dormant chain (R-X) activates at a rate of kact to form a radical (R●) and deactivates at a rate of
kdeact to return to the protected state. Transition metals (TM) at two different oxidation states (“a”
and “a+1”) complexed with ligand(s) mediate the process. The active radical reacts with monomer
to form polymer with a rate constant of kp or terminates with a neighboring radical with a rate
constant of kt.
The first step in the polymerization is transfer of a halide atom to the transition metal (TM)
complex along with oxidation of the TM (Figure 1.13). The rate constant for the reverse reaction,
kdeact, is much higher than the rate of activation, so growing chains are predominantly in the
halogen-capped, dormant state. This decreases the rate of termination relative to polymerization
because polymerization kinetics are first order with respect to radical concentration whereas
termination is a second-order reaction. Minimal termination and the low radical concentrations
lead to slow, controlled polymerization. In the case of Cu-mediated polymerization, Eq. (1.46)
shows the equilibrium constant for the activation/deactivation reaction.

K ATRP

II

•

X/L][R ]
act
 kkdeact
 [Cu
[R-X][Cu I /L]

(1.46)

Solving this equation for the radical concentration yields Eq. (1.47),

 [Cu I /L] 
[R  ]=K ATRP [R-X] 

II
 [Cu X/L] 

(1.47)

and substitution of the radical concentration into the polymerization rate equation yield the
expression on the right of Eq. (1.48).

32

 [Cu I /L] 
R p  k p [M][R  ]  k p [M]K ATRP [R-X] 

II
 [Cu X/L] 

(1.48)

This expression shows that increasing the concentration of monomer or the CuI complex increases
the rate of polymerization, whereas increasing the concentration of the CuII complex should
decrease the reaction rate and increase control over the polymerization.

Research Overview
The following chapters of this dissertation focus on developing new membrane materials
that may enhance performance in existing separations, or increase applications of membranes for
specific separations. Gas separations, including CO2 removal from H2 streams are performed in
chapters 2 and 3 using surface-initiated atom transfer radical polymerization to generate thin
poly(ethylene glycol)-based membranes. Chapter 4 focuses on the separation of ions using a
membrane that resists biofouling in water purification.

33

REFERENCES

34

REFERENCES

(1)

Favre, E. J. Membr. Sci. 2007, 294, 50-59.

(2)

Davison, J.; Thambimuthu, K. Proc. Inst. Mech. Eng. A 2009, 223, 201-212.

(3)

Mulder, M. Basic Principles of Membrane Technology. second ed.; Kluwer Academic
Publishers: Boston, 1996.

(4)

Ho, W. S. W.; Sirkar, K. K. Overview. In Membrane Handbook; Ho, W. S. W.; Sirkar, K.
K., Eds.; Kluwer Academic Publishers: Norwell, MA, 1992; pp 1-25.

(5)

Kesting, R. E.; Fritzsche, A. K. Polymeric Gas Separation Membranes. John Wiley and
Sons, Inc.: New York City, NY, 1993.

(6)

Zolandz, R. R.; Fleming, G. K. Economics. In Membrane Handbook; Ho, W. S. W.;
Sirkar, K. K., Eds.; Kluwer Academic Publishers: Norwell, 1992.

(7)

Hong, S. U.; Malaisamy, R.; Bruening, M. L. Langmuir 2007, 23, 1716-1722.

(8)

Farhat, T. R.; Schlenoff, J. B. J. Am. Chem. Soc. 2003, 125, 4627-4636.

(9)

Corp., A. P. E. Particle size removal range by filtration.
http://www.freedrinkingwater.com/water-education/quality-water-filtrationmethod.htm#Anchor-Pulling-44867 (July 31).

(10)

Freeman, B. D.; Pinnau, I. Gas and Liquid Separations Using Membranes: An Overview.
In Advanced Materials for Membrane Separations; Freeman, B. D.; Pinnau, I., Eds.;
Oxford University Press: Washington D.C., 2004; Vol. 876; pp 1-23.

(11)

Sridhar, S.; Smitha, B.; Aminabhavi, T. M. Sep. Purif. Rev. 2007, 36, 113-174.

(12)

Bos, A.; Punt, I.; Strathmann, H.; Wessling, M. AIChE J. 2001, 47, 1088-1093.

(13)

Breck, D. W. Zeolite Molecular Sieves. Wiley: New York, 1974; p p. 636.

(14)

Zolandz, R. R.; Fleming, G. K. Theory. In Membrane Handbook; Ho, W. S. W.; Sirkar,
K. K., Eds.; Kluwer Academic Publishers: Norwell, MA, 1992; pp 25-53.

(15)

Berens, A. R. J. Appl. Polym. Sci. 1989, 37, 901-913.

(16)

Prabhakar, R. S.; Freeman, B. D. Fluoropolymer-Hydrocarbon Polymer Composite
Membranes for Natural Gas Separation. In Advanced Materials for Membrane

35

Separations; Pinnau, I.; Freeman, B. D., Eds.; American Chemical Society: Washington
D.C., 2004; pp 106-128.
(17)

Dhoot, S. N.; Freeman, B. D.; Stewart, M. E.; Hill, A. J. J. Polym. Sci., Part B: Polym.
Phys. 2001, 39, 1160-1172.

(18)

Prabhakar, R. S.; Freeman, B. D.; Roman, I. Macromolecules 2004, 37, 7688-7697.

(19)

Lin, H.; Freeman, B. D. J. Mol. Struct. 2005, 739, 57-74.

(20)

Lin, H.; Freeman, B. D. Macromolecules 2005, 38, 8394-8407.

(21)

Stern, S. A.; Mullhaup.Jt; Gareis, P. J. AIChE J. 1969, 15, 64-&.

(22)

Bondar, V. I.; Freeman, B. D.; Yampolskii, Y. P. Macromolecules 1999, 32, 6163-6171.

(23)

van der Vegt, N. F. A.; Kusuma, V. A.; Freeman, B. D. Macromolecules 2010, 43, 14731479.

(24)

Chandra, M. Introduction to Polymer Science and Chemistry: A Problem Solving
Approach. Taylor and Francis Group: Boca Raton, 2006.

(25)

Lawson, D. D. Appl. Energy 1980, 6, 241-255.

(26)

Fogg, P. G. T.; Gerrard, W. Solubility of Gases in Liquids. Wiley: New York, 1991.

(27)

Lin, H. Q.; Freeman, B. D. Macromolecules 2005, 38, 8394-8407.

(28)

Singh, A.; Ghosal, K.; Freeman, B. D.; Lozano, A. E.; de la Campa, J. G.; de Abajo, J.
Polymer 1999, 40, 5715-5722.

(29)

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

(30)

Frank, B. P.; Belfort, G. Langmuir 1997, 13, 6234-6240.

(31)

Chambers, L. D.; Stokes, K. R.; Walsh, F. C.; Wood, R. J. K. Surf. Coat. Technol. 2006,
201, 3642.

(32)

Schumacher, J. F.; Carman, M. L.; Estes, T. G.; Feinberg, A. W.; Wilson, L. H.; Callow,
M. E.; Callow, J. A.; Finlay, J. A.; Brennan, A. B. Biofouling 2007, 23, 55.

(33)

Hovgaard, M. B.; Rechendorff, K.; Chevallier, J.; Foss, M.; Besenbacher, F. J. Phys.
Chem. B 2008, 112, 8241.

(34)

Genzer, J.; Efimenko, K. Biofouling 2006, 22, 339.

(35)

Callow, J. A.; Callow, M. E.; Ista, L. K.; Lopez, G.; Chaudhury, M. K. J. R. Soc.
Interface 2005, 2, 319.

36

(36)

Schmidt, D. L.; Brady, R. F.; Lam, K.; Schmidt, D. C.; Chaudhury, M. K. Langmuir
2004, 20, 2830.

(37)

Krishnan, S.; Ayothi, R.; Hexemer, A.; Finlay, J. A.; Sohn, K. E.; Perry, R.; Ober, C. K.;
Kramer, E. J.; Callow, M. E.; Callow, J. A.; Fischer, D. A. Langmuir 2006, 22, 5075.

(38)

Krishnan, S.; Wang, N.; Ober, C. K.; Finlay, J. A.; Callow, M. E.; Callow, J. A.;
Hexemer, A.; Sohn, K. E.; Kramer, E. J.; Fischer, D. A. Biomacromolecules 2006, 7,
1449.

(39)

HoipkemeierWilson, L.; Schumacher, J.; Carman, M.; Gibson, A.; Feinberg, A.; Callow,
M.; Finlay, J.; Callow, J.; Brennan, A. Biofouling 2004, 20, 53.

(40)

Grunlan, J. C.; Choi, J. K.; Lin, A. Biomacromolecules 2005, 6, 1149-1153.

(41)

Jansen, B.; Kohnen, W. J. Ind. Microbiol. 1995, 15, 391-396.

(42)

Lee, D.; Cohen, R. E.; Rubner, M. F. Langmuir 2005, 21, 9651-9659.

(43)

Fu, J.; Ji, J.; Yuan, W.; Shen, J. Biomaterials 2005, 26, 6684-6692.

(44)

Rabea, E. I.; Badawy, M. E. T.; Stevens, C. V.; Smagghe, G.; Steurbaut, W.
Biomacromolecules 2003, 4, 1457-1465.

(45)

Cen, L.; Neoh, K. G.; Kang, E. T. Langmuir 2003, 19, 10295-10303.

(46)

Nakagawa, Y.; Hayashi, H.; Tawaratani, T.; Kourai, H.; Horie, T.; Shibasaki, I. Appl.
Environ. Microbiol. 1984, 47, 513-518.

(47)

Tiller, J. C.; Liao, C.-J.; Lewis, K.; Klibanov, A. M. Proc. Natl. Acad. Sci. U. S. A. 2001,
98, 5981-5985.

(48)

Dudley, L. Y.; Darton, E. G. Desalination 1997, 110, 11-20.

(49)

Ma, W.; Zhao, Y.; Wang, L. Desalination 2007, 203, 256-259.

(50)

Kwak, S.-Y.; Kim, S. H.; Kim, S. S. Environ. Sci. Technol. 2001, 35, 2388-2394.

(51)

Zhao, Y.-J.; Wu, K.-F.; Wang, Z.-J.; Zhao, L.; Li, S.-S. J. Env. Sci. 2000, 12, 241-251.

(52)

Murali, R. S.; Sridhar, S.; Sankarshana, T.; Ravikumar, Y. V. L. Ind. Eng. Chem. Res.
2010, 49, 6530-6538.

(53)

Chafin, R.; Lee, J. S.; Koros, W. J. Polymer 2010, 51, 3462-3471.

(54)

Shen, Y.; Lua, A. C. J. Appl. Polym. Sci. 2010, 116, 2906-2912.

(55)

Krogman, K. C.; Lowery, J. L.; Zacharia, N. S.; Rutledge, G. C.; Hammond, P. T. Nat.
Mater. 2009, 8, 512-518.

37

(56)

Bruening, M.; Dotzauer, D. Nat. Mater. 2009, 8, 449-450.

(57)

Pinnau, I.; Koros, W. J. J. Appl. Polym. Sci. 1991, 43, 1491-1502.

(58)

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

(59)

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

(60)

Lai, C. L.; Huang, S. H.; Lin, W. L.; Li, C. L.; Lee, K. R. J. Membr. Sci. 2010, 361, 206212.

(61)

Hung, W. S.; De Guzman, M.; Huang, S. H.; Lee, K. R.; Jean, Y. C.; Lai, J. Y.
Macromolecules 2010, 43, 6127-6134.

(62)

Jahanshahi, M.; Rahimpour, A.; Peyravi, M. Desalination 2010, 257, 129-136.

(63)

Grajales, S. T.; Dong, X.; Zheng, Y.; Baker, G. L.; Bruening, M. L. Chem. Mater. 2010,
22, 4026-4033.

(64)

Jain, P.; Dai, J.; Grajales, S.; Saha, S.; Baker, G. L.; Bruening, M. L. Langmuir 2007, 23,
11360-11365.

(65)

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

(66)

Priftis, D.; Sakellariou, G.; Mays, J. W.; Hadjichristidis, N. J. Polym. Sci., Part A: Polym.
Chem. 2010, 48, 1104-1112.

(67)

Coles, M. P.; Jordan, R. F. J. Am. Chem. Soc. 1997, 119, 8125-8126.

(68)

Schuh, K.; Prucker, O.; Ruhe, J. Macromolecules 2008, 41, 9284-9289.

(69)

Raghuraman, G. K.; Dhamodharan, R.; Prucker, O.; Ruhe, J. Macromolecules 2008, 41,
873-878.

(70)

Raghuraman, G. K.; Ruhe, J.; Dhamodharan, R. J. Nano. Res. 2008, 10, 415-427.

(71)

Priftis, D.; Sakellariou, G.; Hadjichristidis, N.; Penott, E. K.; Lorenzo, A. T.; Muller, A.
J. J. Polym. Sci., Part A: Polym. Chem. 2009, 47, 4379-4390.

(72)

Trnka, T. M.; Grubbs, R. H. Acc. Chem. Res. 2001, 34, 18-29.

(73)

Grubbs, R. H.; Chang, S. Tetrahedron 1998, 54, 4413-4450.

(74)

Schrock, R. R. Acc. Chem. Res. 1990, 23, 158-165.

(75)

Grubbs, R. H. Tetrahedron 2004, 60, 7117-7140.

38

(76)

Nguyen, S. T.; Johnson, L. K.; Grubbs, R. H.; Ziller, J. W. J. Am. Chem. Soc. 1992, 114,
3974-3975.

(77)

Braunecker, W. A.; Matyjaszewski, K. Prog. Polym. Sci. 2007, 32, 93-146.

(78)

Hawker, C. J.; Bosman, A. W.; Harth, E. Chem. Rev. 2001, 101, 3661-3688.

(79)

Benoit, D.; Chaplinski, V.; Braslau, R.; Hawker, C. J. J. Am. Chem. Soc. 1999, 121,
3904-3920.

(80)

Hawker, C. J.; Barclay, G. G.; Orellana, A.; Dao, J.; Devonport, W. Macromolecules
1996, 29, 5245-5254.

(81)

Fukuda, T.; Terauchi, T.; Goto, A.; Ohno, K.; Tsujii, Y.; Miyamoto, T.; Kobatake, S.;
Yamada, B. Macromolecules 1996, 29, 6393-6398.

(82)

Chong, Y. K.; Krstina, J.; Le, T. P. T.; Moad, G.; Postma, A.; Rizzardo, E.; Thang, S. H.
Macromolecules 2003, 36, 2256-2272.

(83)

Coote, M. L. J. Phys. Chem. A 2005, 109, 1230-1239.

(84)

Achilleos, M.; Legge, T. M.; Perrier, S.; Patrickios, C. S. J. Polym. Sci., Part A: Polym.
Chem. 2008, 46, 7556-7565.

(85)

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

(86)

Kamigaito, M.; Ando, T.; Sawamoto, M. Chem. Rec. 2004, 4, 159-175.

(87)

Edmondson, S.; Osborne, V. L.; Huck, W. T. S. Chem. Soc. Rev. 2004, 33, 14-22.

(88)

Matyjaszewski, K., Advances in Controlled/Living Radical Polymerization. American
Chemical Society: Washington, DC, 2003.

(89)

Kabachii, Y. A.; Kochev, S. Y.; Bronstein, L. M.; Blagodatskikh, I. B.; Valetsky, P. M.
Polymer Bulletin 2003, 50, 271-278.

(90)

Fuji, Y.; Ando, T.; Kamigaito, M.; Sawamoto, M. Macromolecules 2002, 35, 2949-2954.

(91)

Onishi, I.; Baek, K. Y.; Kotani, Y.; Kamigaito, M.; Sawamoto, M. J. Polym. Sci., Part A:
Polym. Chem. 2002, 40, 2033-2043.

(92)

Kotani, Y.; Kamigaito, M.; Sawamoto, M. Macromolecules 1999, 32, 6877-6880.

(93)

Matyjaszewski, K.; Wei, M.; Xia, J.; McDermott, N. E. Macromolecules 1997, 30, 81618164.

(94)

O'Reilly, R. K.; Gibson, V. C.; White, A. J. P.; Williams, D. J. Polyhedron 2004, 23,
2921-2928.

39

(95)

Ishio, M.; Katsube, M.; Ouchi, M.; Sawamoto, M.; Inoue, Y. Macromolecules 2009, 42,
188-193.

(96)

Wang, B.; Zhuang, Y.; Luo, X.; Xu, S.; Zhou, X. Macromolecules 2003, 36, 9684-9686.

(97)

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

(98)

Uegaki, H.; Kamigaito, M.; Sawamoto, M. J. Polym. Sci., Part A: Polym. Chem. 1999,
37, 3003-3009.

(99)

Uegaki, H.; Kotani, Y.; Kamigaito, M.; Sawamoto, M. Macromolecules 1998, 31, 67566761.

(100) Brandts, J. A. M.; van de Geijn, P.; van Faassen, E. E.; Boersma, J.; Van Koten, G. J.
Organomet. Chem. 1999, 584, 246-253.
(101) Le Grognec, E.; Claverie, J.; Poli, R. J. Am. Chem. Soc. 2001, 123, 9513-9524.
(102) Maria, S.; Stoffelbach, F.; Mata, J.; Daran, J.-C.; Richard, P.; Poli, R. J. Am. Chem. Soc.
2005, 127, 5946-5956.
(103) Baek, K. Y.; Kamigaito, M.; Sawamoto, M. J. Polym. Sci., Part A: Polym. Chem. 2002,
40, 1937-1944.
(104) Baek, K. Y.; Kamigaito, M.; Sawamoto, M. J. Polym. Sci., Part A: Polym. Chem. 2002,
40, 2245-2255.
(105) Kamigaito, M.; Watanabe, Y.; Ando, T.; Sawamoto, M. J. Am. Chem. Soc. 2002, 124,
9994-9995.
(106) Terashima, T.; Ouchi, M.; Ando, T.; Sawamoto, M. J. Polym. Sci., Part A: Polym. Chem.
2010, 48, 373-379.
(107) Yoda, H.; Nakatani, K.; Terashima, T.; Ouchi, M.; Sawamoto, M. Macromolecules 2010,
43, 5595-5601.
(108) Percec, V.; Barboiu, B.; Neumann, A.; Ronda, J. C.; Zhao, M. Macromolecules 1996, 29,
3665-3668.
(109) Moineau, G.; Granel, C.; Dubois, P.; Jerome, R.; Teyssie, P. Macromolecules 1998, 31,
542-544.
(110) Lecomte, P.; Drapier, I.; Dubois, P.; Teyssie, P.; Jerome, R. Macromolecules 1997, 30,
7631-7633.
(111) Kotani, Y.; Kamigaito, M.; Sawamoto, M. Macromolecules 2000, 33, 6746-6751.
(112) Braunecker, W. A.; Itami, Y.; Matyjaszewski, K. Macromolecules 2005, 38, 9402-9404.

40

Chapter Two
Effects of Monomer Composition on CO2-Selective Polymer Brush
Membranes
This work was performed in collaboration with Xiaojie Dong and Ying Zhang and published in
Chemistry of Materials: Grajales, S. T.; Dong, X.; Zheng, Y.; Baker, G. L.; Bruening, M. L. Chem.
Mater. 2010, 22, 4026-4033.

This chapter describes the formation of composite, PEG-containing membranes through
atom transfer radical polymerization of poly(ethylene glycol methyl ether methacrylate) from
initiator-modified, porous substrates. The resulting membrane skins are only 50-500 nm thick,
and copolymers that contain a mixture of short and long PEG side chains do not readily crystallize.
The smaller PEG chains (8-9 ethylene oxide units) prevent crystallization, while the presence of
longer side chains (23-24 ethylene oxide units) allows the membranes to maintain a CO2/H2
selectivity of 12 at room temperature. This work examines the effect of side-chain length on
polymerization rate as well as the permeability, selectivity, and crystallinity of copolymer films.
Reflectance FTIR spectroscopy reveals the fraction of different monomers incorporated into
copolymer films and demonstrates when crystallization occurs
Introduction
Selective removal of CO2 from gas streams is becoming increasingly important for
applications such as CO2 sequestration,1 fuel cell operation,2 and H2 synthesis.3 In the case of H2
production, steam reforming of methane produces 9 million tons of H2 per year in the U.S. alone.4
However, the product stream from steam reforming theoretically contains 20 mole% CO2, so

41

purification of the H2 is an important and expensive part of the production process.5 A number of
studies examined membrane-based gas separation as a possible alternative, or as a pre-purification
step, to the pressure swing adsorption process currently used in H2 purification.6, 7 The viability
of membrane processes for purifying H2 streams depends critically on achieving both high flux
and high selectivity, and a recent review summarizes the types of membranes employed for H 2
purification.8 Processes such as H2 recycle in ammonia synthesis have employed H2-selective
membranes for several decades.9 In the case of CO2 removal from H2 streams, however, CO2selective membranes are attractive because purified H2 remains on the high pressure side of the
membrane and does not need to be recompressed.
The solution diffusion model for transport through membranes Eq. (2.1) illustrates how the
flux, J, of a gas through a membrane depends on the partial pressure drop across the membrane,
Δp, the membrane thickness, l, and the gas permeability, P, which is the product of the solubility
coefficient, S, and diffusivity, D, for the gas of interest.10 Eq. (2.2) shows that the ideal selectivity
of gas A over gas B, αA/B, depends on the solubility coefficients and diffusivities of the two gases.
Because of its small size, H2 has a greater diffusivity than CO2, so CO2/H2 selectivity requires high
solubility selectivity for CO2 over H2. Fortunately, CO2 is the more condensable of the two gases,
and solubility selectivity typically favors CO2 transport.10
J

 A/ B 

PDp
l

P  SD

PA S A  DA

PB S B  DB

(2.1)

(2.2)

Membrane materials employed for selective passage of CO2 over H2 should thus have a
higher affinity for CO2 than H2 along with a relatively high free volume that minimizes the
difference between CO2 and H2 diffusivities. One method of increasing CO2 solubility is the

42

incorporation of protonated amines within the membrane.11, 12 In this case, facilitated transport
can lead to CO2/H2 selectivities over 100, but carrier saturation often limits the application of these
systems to low feed pressures (less than 0.05 bar).13-15

At room temperature, amorphous

poly(ethylene glycol) (PEG)-containing membranes exhibit CO2/H2 selectivities ranging from 5
to 12, presumably because of quadrupole-dipole interactions between CO2 and the ether functional
groups of PEG.16-19 Unfortunately, however, crystallization of PEG chains decreases both the
permeability and selectivity of such membranes. To overcome this challenge, several studies
employed cross-linked, solution-cast, PEG-containing membranes to maintain a high free volume,
decrease crystallinity, and increase chemical stability.7, 20-22 In the best case, the cross-linked
membranes show CO2 permeabilities of up to 600 Barrers [1 Barrer = 10−10 cm3 (STP) ·cm /
(cm2·s·cmHg) = 7.5×10−18 m3 (STP) ·m/(m2·s·Pa)] and CO2/H2 selectivities of approximately 12
at room temperature.23
Nevertheless, the high thickness (>100 m) of solution-cast membranes yields a permeance
(permeance is defined as flux/∆p) that is too low for practical separations. This work examines
the use of atom transfer radical polymerization (ATRP) from porous substrates to create composite
membranes with thin (50-500 nm) skins of PEG-containing films that allow selective removal of
CO2 from H2 streams. ATRP is attractive for synthesizing these membrane skins because it often
affords control over film thickness along with polymer chains with relatively low polydispersity.24,
25

Scheme 2.1 shows two permutations of the ATRP process for forming membranes on porous

alumina substrates, and similar processes on porous polymer supports are also possible. Initiator
attachment occurs either via adsorption of a disulfide initiator to gold-coated alumina to generate
A (scheme 2.1), or attachment of a silane directly to the alumina to generate B. Subsequent atom
transfer radical polymerization with a Cu+/1,1,4,7,10,10-hexamethyltriethylenetetramine

43

(HMTETA) catalyst yields the membrane skin. Because of the small thickness of the membrane
skins, the permeances of the modified alumina membranes are several orders of magnitude greater
than those of solution-cast membranes.8
Additionally, this work examines the use of copolymers prepared from poly(ethylene
glycol methyl ether methacrylate) (PEGMEMA) monomers with different PEG chain lengths
(Scheme 2.2) to inhibit the crystallization of the PEG side chains. The combination of monomers
with 8-9 and 23-24 ethylene oxide repeat units yields films that do not crystallize over many
months and membrane skins with a room temperature CO2/H2 selectivity of around 12.

Scheme 2.1: Methods for growth of a poly(PEGMEMA) membrane from porous alumina.

44

Scheme 2.2: Copolymerization of PEGMEMA-475 and PEGMEMA-1100 to provide a membrane
skin.

Experimental
Chemicals. PEGMEMA monomers with Mn = 300 Da (PEGMEMA-300), Mn = 475 Da
(PEGMEMA-475), and Mn = 1,100 Da (PEGMEMA-1100) were obtained from Aldrich. These
monomers contain both 100 ppm 4-methoxyphenol and 300 ppm 2,6-ditert-butyl-4-methylphenol
as inhibitors. Proton NMR spectra (see Appendix A) show that the average numbers of ethylene
oxide units in PEGMEMA-300, PEGMEMA-475, and PEGMEMA-1100 are 4.4, 8.8, and 23.1,
respectively. In some cases, the inhibitors were removed from the monomers using flash column
chromatography with basic alumina, but this did not have a significant effect on polymerizations,
so in most reactions the monomers were used as received. The CuCl, CuBr2, and 1,1,4,7,10,10hexamethyltriethylenetetramine (HMTETA) were purchased from Aldrich, and deionized water
was obtained using a Millipore system (Milli-Q, 18.2 MΩ·cm). THF was distilled over sodium
metal with benzophenone, and DMF was distilled over 4 Å molecular sieves. Silicon(100) wafers
were obtained from NOVA Electronic Materials and sputter-coated with 20 nm of chromium
followed by 200 nm of gold by LGA Thin Films (Santa Clara, CA). The porous alumina substrates
were Anodisc membrane filters (25 mm disks with 0.02 µm surface pores) purchased from

45

Whatman,

and

the

disulfide,

[Br-C(CH3)2-COO(CH2)11S]2,

and

silane,

SiCl(CH3)2(CH2)11OCOC(CH3)2Br, initiators were synthesized as described previously.26-29
Polymerization. Porous alumina substrates were initially rinsed with ethanol, dried with
N2, UV/O3 cleaned (Boekel UV-Clean Model 135500) for 15 min, and sputter coated with 5 nm
of gold in a Pelco SC-7 sputter coater. The thickness of the gold coating was monitored with a
Pelco FTM-2 quartz crystal microbalance. Attachment of a monolayer of the disulfide initiator to
gold-coated Si wafers or porous alumina occurred during an overnight immersion of the substrate
in an ethanolic solution containing 1 mM disulfide initiator. The resulting sample was rinsed with
5 mL of ethanol and dried in a N2 stream. Attachment of the silane initiator occurred during
overnight immersion of an uncoated, UV/ozone-cleaned porous alumina substrate in a solution
containing 8 mM silane in THF. The resulting sample was rinsed with 5 mL THF and dried in a
N2 stream. The two different types of initiator-modified membranes were handled identically in
the ensuing polymerization steps.
Following a previous procedure, the catalyst stock solution was prepared by first dissolving
0.06 g (0.6 mmol) of CuCl and 0.04 g (0.2 mmol) of CuBr2 in 30.0 mL of distilled, degassed
DMF.30, 31 This solution was further degassed via three freeze, pump, thaw cycles. In a N2-filled
glove bag, 490 µL (1.8 mmol) of degassed 1,1,4,7,10,10-hexamethyltriethylenetetramine
(HMTETA) ligand was added to this solution, which subsequently turned a dark green color during
several hours of stirring. During this initial stirring, a small amount of material precipitated, but
the resulting catalytic activity of the solution remained constant for at least several weeks. The
nominal mole ratio of Cu+:Cu2+:HMTETA was 3:1:9.
The monomer solution, which contained 0.75 M PEGMEMA in water, was initially
degassed via three freeze, pump, thaw cycles. (When different molecular weight monomers were

46

used, the total monomer concentration was maintained at 0.75 M). The polymerization solution
was prepared in a N2-filled glove bag by combining degassed monomer and catalyst solutions in a
9:1 volume ratio to produce a blue-green solution containing 0.67 M monomer, 2.0 mM CuCl,
0.60 mM CuBr2, and 6.0 mM HMTETA. The initiator-modified membranes or wafers were
immersed in the polymerization solution for designated periods of time, and the resulting polymer
films were rinsed with 5 mL water and soaked in water for at least two hours before rinsing with
5 mL of ethanol and drying under a N2 stream.
Characterization methods.

Reflectance FT-IR spectroscopy was performed using a

Nicolet Magna-IR 560 spectrometer with a Pike grazing angle (80°) attachment. A UV/ozonecleaned, gold-coated Si wafer was used to obtain the background spectrum. Thicknesses of films
formed on gold-coated Si were determined with a rotating analyzer ellipsometer (J.A. Woollam
model M44) at an incident angle of 75°.

A two-term Cauchy equation was employed to

simultaneously fit film thickness and the Cauchy constants needed to model the wavelengthdependence of the film refractive index. Thicknesses were determined at a minimum of three
locations on each substrate. Scanning electron microscopy was performed with the JEOL semiin-lens cold cathode field-emission scanning electron microscope, model JSM-7500F, operating
with a r-filter in signal maximum mode. Samples were sputter coated with 5 nm of gold prior to
imaging.
Gas Permeation. Single gas permeation experiments were performed by exposing the
membrane to individual gases at varying pressures in an Advantec/MFS UHP-25 cell with a
pressure relief valve. The gas flow rate through the membrane was measured with either an
Optiflow 420 electronic soap bubble meter or a manual soap bubble meter. Mixed-gas experiments
were performed by loading the membrane into a custom membrane holder (all connections utilized

47

Swagelok fittings and were tested to ensure that they maintained pressure over a time-scale longer
than the experiment) that allowed cross-flow of the feed gas as well as a sweep gas on the permeant
side. A backpressure valve was employed to sustain the feed gas pressure, and the feed flow rate
was high enough to maintain constant composition at the face of the membrane (the stage-cut, or
ratio of permeant flow to feed flow, was <1%). The N2 sweep gas/permeant stream was connected
to an automated six-port injector valve on a Hewlett-Packard 6890 GC equipped with a thermal
conductivity detector (TCD) and an Agilent GS-CarbonPLOT capillary column (i.d. = 0.53 mm,
length = 30 m, 3 µm coating). The mixed gas permeances as well as pure gas permeances were
calculated from GC results using Eq. (2.3),

Permeance 

B   N

2

 N  A  Δp

(2.3)

2

where χB is the mole fraction of a particular permeate gas, B, in the gas mixture injected into the
GC column,  N2 is the sweep gas flow rate,  N 2 is the mole fraction of N2 in the gas injected into
the GC column, Δp is the transmembrane partial pressure difference for the gas of interest, and A
is the membrane area. (  N2 /  N2 is the sum of the permeate and sweep gas flow rate, and  N 2
was determined by subtracting the mole fractions of the permeates from unity. This assumes that
the amount of N2 passing from permeate to feed was negligible.) The sweep gas flow rate was
programmed into a mass flow controller, pressure and area were measured, and the mole fractions
were determined from the integrated GC spectra and a calibration curve.

48

Results and Discussion

Polymerization Rates as a Function of Monomer Composition.

The

first

step

in

developing surface-initiated ATRP for creating membrane skins, especially for forming copolymer
skins, is examination of polymerization from model flat substrates such as gold-coated Si wafers.32
Such substrates facilitate ellipsometric and spectroscopic characterization of polymer growth and
composition. We are particularly interested in the relative film growth rates for PEGMEMA-475,
PEGMEMA-1100, and their mixtures because the short side chains of PEGMEMA-475 do not
crystallize, whereas the long chains of PEGMEMA-1100 should promote CO2/H2 selectivity.33 As
Figure 2.1 shows, polymerization of PEGMEMA-475 by itself gives the most rapid initial growth
in film thickness, but the growth rate decreases dramatically after 15 min of polymerization. In
contrast, polymerization of PEGMEMA-1100 is slower, but the film growth rate is essentially
constant for 120 min. Interestingly, surface-initiated polymerizations of mixtures in which
PEGMEMA-475 constitutes either 25 or 75 mole% of the total monomer show relatively high and
steady growth rates.

49

350

poly(PEGMEMA-475)
75 % PEGMEMA-475

Dry Thickness (nm)

300

25 % PEGMEMA-475
poly(PEGMEMA-1100)

250
200
150
100
50
0
0

20

40

60 80 100 120 140
Time (min.)

Figure 2.1: Evolution of film thickness with time in surface-initiated polymerization of pure and
mixed monomer solutions. For copolymers, the legend shows the solution PEGMEMA-475 mole
percentage with respect to the total monomer (PEGMEMA-475 plus PEGMEMA-1100). The
overall monomer concentration was 0.67 M, and the catalyst system was 2.0 mM CuCl, 0.60 mM
CuBr2, and 6.0 mM HMTETA. The error bars, which in many cases are obscured by the symbols,
represent one standard deviation.
Decreases in growth rates with polymerization time, particularly for poly(PEGMEMA475), suggest that the HMTETA-Cu+/2+ catalyst system gives a high concentration of radicals that
results in significant termination by radical-radical coupling.

The large side chains of

PEGMEMA-1100 may provide steric hindrance to both polymerization and termination. In
solution polymerization studies, Haddleton et al. attributed the high ATRP activity of

50

PEGMEMA-475 relative to PEGMEMA-1100 to ethylene oxide coordination with the Cu catalyst
in close proximity to the double bond.34 However, steric issues seem equally likely to affect
polymerization.33

In the case of homopolymerization from surfaces, the PEGMEMA-1100

solutions are also more viscous and contain less water than PEGMEMA-475 solutions, and both
of these factors should also lead to lower polymerization rates for the PEGMEMA-1100. Mixtures
of PEGMEMA-475 and PEGMEMA-1100 have intermediate polymerization rates, and the side
chains are apparently still sufficiently large to reduce termination relative to pure PEGMEMA475. For PEGMEMA-475, PEGMEMA-1100, and their mixtures, film growth is rapid compared
to polymerization of most other monomers and non-aqueous polymerizations.35-37 A number of
studies demonstrated that the presence of water frequently enhances the rate of ATRP.38, 39
Determining the Composition of Copolymer Films. The molar ratio of PEGMEMA-475
to PEGMEMA-1100 in a film may not correspond to the monomer ratio in the polymerization
solution. In particular, the smaller monomer may be more reactive because of faster diffusion to
the surface or less steric hindrance to polymerization. In an attempt to determine the relative
amounts of each monomer incorporated in poly(PEGMEMA-475-co-PEGMEMA-1100) films, we
compared the intensities of carbonyl (1731 cm-1) and C-O-C (1149 cm-1) stretches in both
homopolymer and copolymer films. The reflectance FTIR spectra in Figure 2.2 show that for
amorphous homopolymer films, the intensity of the C-O-C stretch relative to the C=O stretch is
much higher for poly(PEGMEMA-1100) than poly(PEGEMEMA-475), as would be expected
because of the longer PEG chain in the higher molecular weight monomer.

51

2000

0.05

C=O

Absorbance

poly(PEGMEMA-1100)

C-O-C

poly(PEGMEMA-475)

1750

1500

1250

1000

750

Wavenumbers (cm-1)
Figure 2.2: FTIR spectra of 152 nm-thick poly(PEGMEMA-475) (black line) and 118 nm-thick
poly(PEGMEMA-1100) (dashed red line) films on gold-coated substrates. Spectra were taken
immediately after polymerization before the PEG side chains in poly(PEGMEMA-1100)
crystallized. The spectra are offset for clarity.
We initially planned to estimate the mole ratios of monomers incorporated into a
copolymer film using Eq. 2.4, where x475 is the mole fraction of PEGMEMA-475 in a copolymer
film and Rco, R1100, and R475 are the ratios of the C-O-C peak height (1149 cm-1) to the C=O peak
height (1731 cm-1) for copolymer, poly(PEGMEMA-1100), and poly(PEGMEMA-475) films,
respectively.

Rco  x475 R475  (1  x475 ) R1100

52

(2.4)

However, Figure 2.3 suggests that in some cases peak height and peak area ratios for the C-O-C
to C=O absorbance increase with film thickness for homopolymer and copolymer films. This
increasing ratio of C-O-C to C=O absorbance likely occurs because reflectance measurements plot
log(Ro/R), where Ro and R are the reflectivities of bare and film-covered substrates, respectively.
These reflectivities are complicated functions of the film thickness and the complex refractive
indices, n  n  ik , of the substrate and the film. To account for these complications in reflectance
FTIR spectroscopy, we performed Fresnel calculations to determine the absorption coefficient, k,
for different films at the wavelengths of interest. Appendix A provides the details of these
calculations and shows that the calculated absorption coefficients do not vary significantly with
the thicknesses of the films (Figure A.2).

53

54

We used the k values for three homopolymer films, poly(PEGMEMA-300), poly(PEGMEMA475), and poly(PEGMEMA-1100), to create a calibration curve (Figure 2.4) of the kC-O-C / kC=O
ratio versus the average number of ethylene oxide units per repeat unit, EOavg, and then calculated

kC-O-C / kC=O

EOavg in copolymer films.

14
12
10
8
6
4
2
0
0

5
10
15
20
25
Ethylene Oxide Units per
Monomer

Figure 2.4: Ratios of C-O-C (1149 cm-1) and C=O (1731 cm-1) absorption coefficients, k, as a
function of the average number of ethylene oxide units per repeat unit, EOavg, in poly(PEGMEMA300), poly(PEGMEMA-475), and poly(PEGMEMA-1100) films. The ethylene oxide units per
monomer were calculated from 1H-NMR spectra of monomers, and the standard deviations in the
absorption coefficient ratios were determined using films with thicknesses ranging from 5 to 454
nm. The line is a fit to the data with a zero intercept.
For poly(PEGMEMA-475-co-PEGMEMA-1100), this EOavg value is a direct function of the
fraction of each monomer in the copolymer and the average number of ethylene oxide units in each
monomer Eq. (2.5).
EOavg  8.8x475  23.1(1  x475 )

(2.5)

The reflectance FTIR data and the Fresnel calculations show that the ratios of PEGMEMA475 to PEGMEMA-1100 in poly(PEGMEMA-475-co-PEGMEMA-1100) films are approximately

55

the same as those in the solutions from which they grow (Table 2.1). This is somewhat surprising
considering the large differences in the initial growth rates as well as termination rates for surfaceinitiated homopolymerization of the two monomers (Figure 2.1). However, in the case of
copolymers, solutions containing 25 mole% and 75 mole% PEGMEMA-475 give essentially the
same initial growth rate (Figure 2.1), which is consistent with the similarity between solution and
film compositions. This direct correlation between solution and film compositions affords
straightforward control over the ratio of long and short side chains to minimize PEG crystallization
and maximize CO2/H2 selectivity in membranes.
Table 2.1: Composition of poly(PEGMEMA-475-co-PEGMEMA-1100) filmsa as a function of
the PEGMEMA-475 mole percentages in the polymerization solution.b
Feed Solution,
mole% PEGMEMA-475
0

Polymer Composition,
mole% PEGMEMA-475
0

23 ± 1

25

33 ± 2

18 ± 1

50

54 ± 3

15 ± 1

75

71 ± 7

13 ± 1

100

100

8.7 ± 0.6

EOavg

a

Film compositions were determined from reflectance FTIR spectra and Fresnel
calculations (see text) and are expressed as both mole percent and the average number of ethylene
oxide (EOavg) groups per repeat unit. The uncertainties, which were calculated using propagation
of error, represent one standard deviation. bThe mole percentage in the feed solution is relative to
the total amount of monomer in solution.
Crystallization of PEG Side Chains. PEG chains crystallize in helices that contain two
turns per 7 EO units over a length of 19.3 Å,40-42 and helix formation leads to splitting of the 1150
cm-1 C-O-C IR band into peaks at 1120 and 1148 cm-1 and a narrowing of the CH2 wagging peak
at 1360 cm-1.40, 41 In PEGMEMA films, long PEG side chains crystallize similarly, and Figure

56

2.5A demonstrates the changes in reflectance FTIR spectra that occur upon crystallization of a
poly(PEGMEMA-1100) film.

poly(PEGMEMA-1100) (freshly prepared)

25 % PEGMEMA-475 (freshly prepared)

poly(PEGMEMA-1100) (1 day later)

25 % PEGMEMA-475 (404 days later)

1600

(B)
Absorbance

Absorbance

(A)

0.02

1400

1200

1000

800 1600

Wavenumbers (cm-1)

0.02

1400

1200

1000

800

Wavenumbers (cm-1)

Figure 2.5: Reflectance FTIR spectra of (A) 60 nm-thick poly(PEGMEMA-1100) immediately
after polymerization (solid black line) and 24 hours later (dashed red line) and (B) 97 nm-thick
poly(PEGMEMA-475-co-PEGMEMA-1100) immediately after polymerization (solid black line)
and 404 days later (dashed red line). The copolymer was prepared from a solution containing 25
mole% PEGMEMA-475 and 75 mole% PEGMEMA-1100.

Previous reports of PEG crystallization in related coatings suggest that the helical axis
preferentially orients perpendicular to the surface in films less than 100 nm thick, and both
perpendicular and parallel to the surface in thicker films.27, 43-46
However, the presence of crystallinity, rather than its orientation, is the primary concern
for creating selective membranes with high permeabililties, and prior studies reveal that the PEG
chain length, not film thickness, is the primary factor in determining whether crystallization
occurs.47, 48 In contrast to poly(PEGMEMA-1100), poly(PEGMEMA-475) films do not show any

57

evidence of crystallization in reflectance FTIR spectra, presumably because neighboring PEG side
chains are too short to crystallize in this environment.49
Importantly, copolymers prepared from PEGMEMA-1100 and PEGMEMA-475 show no
sign of crystallization, even when the fraction of PEGMEMA-475 in the film is only ~25%. Figure
2.5B shows that the reflectance FTIR spectrum of a poly(PEGMEMA-475-co-PEGMEMA-1100)
film remains unchanged even after more than a year at room temperature. The presence of the
short chains is apparently sufficient to inhibit crystallization of the longer side chains, and this
effect occurs for films prepared from solutions containing PEGMEMA-475 as 25, 50, and 75
mole% of the total monomer.
Formation of Gas-Separation Membranes by ATRP.

Composite membranes are

attractive for gas separations because the minimal thickness of the thin, selective skin allows for
high flux, and the underlying support provides mechanical strength. Hence, the goal in using
ATRP to create a membrane skin is to completely cover a porous substrate with a thin film without
filling the underlying pores. The first step in the formation of membranes by ATRP is attachment
of an initiator to the surface. To be consistent with our work on gold wafers, we sputtered 5 nm
of gold on porous alumina and subsequently adsorbed a monolayer of the disulfide initiator on the
surface (Scheme 2.1A). In a separate procedure, to avoid the deposition of gold on the membrane
surface, we also attached a silane initiator directly to the alumina (Scheme 2.1B). Because alumina
membranes are expensive and fragile, we are also working on growing membrane skins on
polymeric ultrafiltration membranes. This is more challenging, however, because the solvents
employed in initiator attachment may damage the underlying membrane. Aqueous adsorption of
macroinitiators may overcome this problem.38

58

Subsequent polymerization of PEGMEMA from initiators on alumina yields the desired
composite membranes. Figure 2.6 shows a SEM image of a composite membrane containing a
poly(PEGMEMA-475-co-PEGMEMA-1100) skin on a porous alumina support. The film is thin
(130 nm) and clearly covers the pores without filling them. However, selective gas permeation
studies are needed to demonstrate that the films are defect-free and non-crystalline.

100 nm

Film

Figure 2.6. SEM image of the cross section of a gold-coated porous alumina membrane modified
with a poly(PEGMEMA-475-co-PEGMEMA-1100) film. The alumina exhibits unfilled vertical
pores covered by a 130 nm-thick film that was grown from an aqueous solution containing 25
mole% PEGMEMA-475 and 75 mole% PEGMEMA-1100.
Gas Permeation through Poly(PEGMEMA) Membranes. Figure 2.7 shows the CO2 and
H2 fluxes through porous alumina membranes coated with either a poly(PEGMEMA-1100) film
or a copolymer film grown from a solution containing a 1:1 mole ratio of PEGMEMA-475 and
PEGMEMA-1100. For poly(PEGMEMA-1100), the initial CO2/H2 selectivity is 12 ± 1. The

59

highest room-temperature CO2/H2 selectivity of PEG-containing membranes in the literature is
also 12.7, 21-23 However, the CO2/H2 selectivity of the poly(PEGMEMA-1100) film drops to less
than 1 within a day because the CO2 flux decreases 12-fold (Figure 2.7A). This drop in CO2 flux
is almost certainly due to crystallization of the PEG side chains, which as noted above occurs
within one day of synthesis of poly(PEGMEMA-1100) films.

Figure 2.7: Single-gas fluxes through gold-coated porous alumina membranes capped with (A)
poly(PEGMEMA-1100) or (B) poly(PEGMEMA-475-co-PEGMEMA-1100) grown from a
solution containing 50 mole% PEGMEMA-475. (Fluxes are normalized by multiplying by the
SEM thickness of the membrane skin.) The circles represent CO2 data, and the triangles represent
H2 data; filled symbols represent measurements taken immediately after synthesis, and open
symbols represent data obtained 24 h later. The inset in Figure A shows the data on an expanded
ordinate for clarity.
By comparison, poly(PEGMEMA-475) films exhibit an initial CO2/H2 selectivity of 9 ± 2.
On silylated supports, poly(PEGMEMA-1100) and poly(PEGMEMA-475) films show
selectivities of 13 ± 1 and 5.8 ± 0.5, respectively, demonstrating that there is a significant
difference between the gas selectivities of these two polymers. The poly(PEGMEMA-475) films

60

contain a smaller overall fraction of PEG than poly(PEGMEMA-1100), which likely explains their
lower selectivity. However, the selectivity of the poly(PEGMEMA-475) does not decrease with
time because the shorter chains do not crystallize.
The goal of using copolymers as membranes is to achieve the initial selectivity of
poly(PEGMEMA-1100) films while avoiding decreases in selectivity and flux due to film
crystallization.

Figure 2.7B shows that a poly(PEGMEMA-475-co-PEGMEMA-1100) film

(prepared on gold-coated alumina using a solution containing 50 mole% PEGMEMA-475) has an
initial CO2/H2 selectivity of 12-13, which is essentially the same as that of the pure
poly(PEGMEMA-1100). Moreover, the selectivity of the copolymer membranes is constant over
at least several weeks at room temperature, again presumably because the side chains of
PEGMEMA-475 inhibit crystallization. Figure 8 shows the dependence of CO2/H2 selectivity on
the mole fraction of PEGMEMA-475 in poly(PEGMEMA-475-co-PEGMEMA-1100) films. As
long as there is 75 mole% or less PEGMEMA-475 in the polymerization solution, membranes
formed on gold-coated alumina maintain average CO2/H2 selectivities of 12 at room temperature.
The films grown from silylated alumina show a maximum CO2/H2 selectivity of 17 ± 2 with
copolymers generated from a 50 mole% PEGMEMA-475 polymerization solution, but a
selectivity of only 7 when using 75 mole% PEGMEMA-475. Silanization is more likely to
produce initiators within the pores of the alumina, which might somewhat alter film structure at
the support surface and change selectivity relative to films on gold-coated alumina. Moreover, the
initiator packing densities might be different for the gold-coated and silylated surfaces.

61

Figure 2.8: Single-gas CO2/H2 selectivity vs. mole fraction of PEGMEMA-475 in the monomer
solution used to create amorphous poly(PEGMEMA-475-co-PEGMEMA-1100) films on porous
alumina membranes. The triangles represent films grown from silylated alumina, and the squares
represent films grown from initiators on gold-coated alumina. For the poly(PEGMEMA-1100)
homopolymer on silylated alumina, the open circle near the origin represents selectivity after
crystallization. Error bars are one standard deviation.
Determination of the permeability, P, of the films on porous alumina is challenging because
of the difficulty in obtaining accurate values for film thickness [Eq. (2.1)]. Film thicknesses in
multiple cross-sectional SEM images of a given membrane typically show a relative standard
deviation (rsd) of 10-30%, although sometimes we see even larger variations. Based on SEM
images of membranes with a rsd in thickness <25%, the H2 permeability of poly(PEGMEMA-475co-PEGMEMA-1100) films ranges from 2 to 6 Barrers, and the CO2 permeability ranges from 20
to 60 Barrers. (Given the uncertainty in thickness, we could not differentiate among permeabilities
of films with different compositions.) Typical CO2 permeabilities of PEG-containing films range

62

from 12 Barrers in pure PEG50 to 570 Barrers in cross-linked PEG-based acrylates.49, 51 This wide
range of permeabilities in different PEG-containing polymers stems from differences in fractional
free volume and the presence of different polar groups that alter the solubility of gases in the
membrane.23, 52, 53 Although the poly(PEGMEMA-475-co-PEGMEMA-1100) films do not exhibit
as high a permeability as some cross-linked films, the ultrathin skin maximizes permeance for a
given permeability. To improve permeability, future studies will examine cross-linking to increase
flux in these films.
Selectivities in gas mixtures can differ significantly from those with single gases,
particularly when absorption of a gas such as CO2 plasticizes the membranes.54 In mixed-gas
experiments, plasticization may increase the flux of H2 and decrease CO2/H2 selectivity relative to
single-gas measurements. With feed gas streams of 79.7 mole% H2 and 20.3 mole% CO2, the
CO2/H2 selectivity was only 13% lower on average than the pure gas CO2/H2 selectivities.
Summary
In their amorphous state, composite poly(PEGMEMA-1100) membranes provide high CO2
permeances and CO2/H2 selectivities, but crystallization can decrease flux by an order of
magnitude and reduce CO2/H2 selectivity to below 1. Copolymerization of two PEGMEMA
monomers, one containing 8-9 and the other 23-24 ethylene oxide units per side chain, avoids the
problem of crystallization and yields films that maintain an amorphous state for more than a year.
Reflectance FTIR spectra indicate that the ratio of the PEGMEMA-475 and PEGMEMA-1100
incorporated into copolymer films is within 8% of the ratio of the monomers in the polymerization
solution. Membranes consisting of porous alumina supports coated with copolymer films exhibit
stable single-gas CO2/H2 selectivities around 12 at room temperature, which is attractive for CO2
removal from H2 streams.

63

Acknowledgements
We thank the U.S. Department of Energy Office of Basic Energy Sciences for funding and
S. Koszalka and Dr. Kathryn Severin for their assistance.

64

APPENDIX

65

APPENDIX

Effects of Monomer Composition on CO2-Selective Polymer Brush
Membranes
Determination of absorption coefficients in reflectance IR spectra
We plot reflectance IR spectra as log(Ro/R), where Ro is the fraction of the incident light
intensity reflected by a bare gold substrate (Figure A.1A) and R is the fraction of incident light
intensity reflected by a film-coated substrate (Figure A.1B).

A
Air, n1
~
Au, n3

B
Io

IR
θ1

Air, n1
~
Film, n
2

Io

θ1

IR
...

~
Au, n
3

Figure A.1: Schematic drawing of the reflection of light from (A) a bare gold surface and
(B) a film-covered gold surface.
The value of Ro is simply the Fresnel reflection coefficient, r13, multiplied by its complex
conjugate Eq. (A.1), where the Fresnel coefficient is given by Eq. (A.2). In equation S2, n~3 is the
complex refractive index of gold, n-ik, (assumed to be 3-30i), 1 is the angle of incidence (80 º),

~
n1 is the refractive index of air (assumed to be 1.000), and  Au is the complex angle of refraction

66

~
in the gold film. The value of  Au was calculated using Snell’s law, Eq. (A.3). (Tilda superscripts
indicate a complex number.)

(A.1)

(A.2)

(A.3)
Eq. (A.4) allows calculation of the reflectance, R, of the film-covered substrate, where the *

r12 is the Fresnel reflection coefficient for the
represents the complex conjugate. In this equation, ~
r23 is the Fresnel reflection coefficient for the film-gold interface Eq.
air-film interface, Eq. (A.5), ~
(A.6), λ is the wavelength of the incident light in air, n~2 is the complex refractive index of the film,

~
~
l is the film thickness, and cos  2 and cos  3 can be calculated using Snell’s law for the appropriate
interfaces.

(A.4)

(A.5)

(A.6)

67

To calculate the absorption coefficient, k, for a specific polymer film, we determined the
absorbance peak heights at 1731 cm-1 (C=O stretch) and at 1149 cm-1 (C-O-C stretch) from
reflectance IR spectra. It is important to note that the presence of a film on the gold substrate
changes the reflectivity of the surface even in the absence of absorption (when k=0), meaning that
the film-covered gold gives rise to a baseline shift that must be taken into consideration when
calculating peak heights. Specifically, this correction was made by calculating the absorbance
when k=0 and subtracting this from the calculated peak absorbance to generate a corrected peak
height. This correction was used along with the above equations to iteratively calculate the
absorption coefficient, k, which gives rise to the measured peak height value for a particular film.
The ratios of calculated absorption coefficients, kC-O-C/kC=O, were independent of thickness for a
given film composition, as Figure A.2 shows below.

68

20

poly(PEGMEMA- 1100)
25 % PEGMEMA- 475
50 % PEGMEMA- 475
75 % PEGMEMA- 475
poly(PEGMEMA- 475)
poly(PEGMEMA- 300)

kC-O-C / kC=O

15

10

5

0
0

100
200
300
Dry Thickness (nm)

400

Figure A.2: Ratios of the C-O-C (1149 cm-1) and C=O (1731 cm-1) absorption coefficients in
poly(PEGMEMA) homo and copolymer films as a function of film thickness. In the case of
copolymer films, the legend shows the PEGMEMA-475 mole percentage of total monomer in the
polymerization solution, and the rest of the monomer was PEGMEMA-1100. The polymerization
was performed as described in the experimental section.
We averaged the ratios of calculated absorption coefficients for homopolymer films of
different thicknesses (see Table A.1) to plot the dependence of kC-O-C/kC=O on the average ethylene
oxide units per monomer (Figure A.4). This calibration curve allows the thickness-independent

69

conversion from kC-O-C/kC=O to the average number of EO groups per repeat unit in copolymer
films.
Table A.1: Absorption coefficients, k, of homopolymer films. The k values were calculated using
Eq. (A.1-A.5) and the peak absorbances corresponding to C-O-C and C=O stretches. The
uncertainty is one standard deviation of the values obtained from films with a range of thicknesses.
Homopolymer

Wavenumber, cm-1

Bond
Stretch

Absorption
Coefficient, k

1149

C-O-C

0.30±.08

1731

C=O

0.14±0.03

1149

C-O-C

0.42±0.05

1731

C=O

0.11± 0.01

1149

C-O-C

0.48±0.08

1731

C=O

0.045± 0.006

PEGMEMA-300

PEGMEMA-475

PEGMEMA-1,100

Average numbers of ethylene oxide units in each of the different monomers (PEGMEMA300, -475, and -1100) were determined from NMR spectra (Figure A.3). Averaging three NMR
spectra of each monomer, the integrals of the peaks with a chemical shift between 3.4 ppm and 4.5
ppm (arising from the four hydrogen atoms on the ethylene oxide repeat units) were summed and
divided by 4 to give the average number of EO units per monomer. (The intensity of the alkene
protons was set to 1 for normalization.)

70

e

(A) PEGMEMA- 300

g

h

a,b

CH3

H2C
O

g

O
O

CH3
n

c,d,e,f

a

6.5

h
c

b

6.0

5.5

1.00

1.00

5.0

4.5

d

f

4.0

3.5
3.0
11.39 2.63
2.03
2.24 1.94

2.5

2.0

ppm

2.76

e

(B) PEGMEMA- 475

g
a b

10

8

cd f

h

6
4
2
1.00 2.02 28.33 2.48
2.63 1.87 2.64
1.00

0

ppm

Figure A.3: Proton NMR spectra of (A) PEGMEMA-300, (B) PEGMEMA-475, and (C)
PEGMEMA-1100. Peak areas in these spectra were averaged with areas from two additional
replicate spectra to generate values for the number of average ethylene oxide units per monomer.

71

Figure A.3 (cont’d)

(C) PEGMEMA- 1100

a

8

b

6
1.00
1.01

e

c dfg

h

4
2
2.27 89.43 1.66
1.92 2.55
5.18

0

ppm

Conclusions
We performed Fresnel calculations to determine the absorption coefficient, k, for different
films at the wavelengths of interest.

We used the k values for three homopolymer films,

poly(PEGMEMA-300), poly(PEGMEMA-475), and poly(PEGMEMA-1100), at two different IR
absorptions, one corresponding to the carbonyl stretch on the monomer unit in the polymer
backbone and one corresponding to the ether stretch along the PEG side chain, to enable the
determination of the average number of ethylene oxide units per monomer unit.

72

REFERENCES

73

REFERENCES

(1)

Scholes, C.; Kentish, S.; Stevens, G. Sep. Purif. Rev. 2009, 38, 1-44.

(2)

Prakash, S.; Kohl, P. A. J. Power Sources 2009, 192, 429-434.

(3)

Shao, L.; Low, B. T.; Chung, T. S.; Greenberg, A. R. J. Membr. Sci. 2009, 327, 18-31.

(4)

U.S. Department of Energy, Energy Information Administration. The Impact of Increased
Use of Hydrogen on Petroleum Consumption and Carbon Dioxide Emissions.
http://www.eia.doe.gov/oiaf/servicerpt/hydro/pdf/oiafcneaf%2808%2904.pdf
(accessed May 10, 2010).

(5)

Molburg, J. C.; Doctor, R. D. Proc. Ann. Int. Pitt. Coal Conf. 2003, 20, 1388-1408.

(6)

Favre, E. J. Membr. Sci. 2007, 294, 50-59.

(7)

Lin, H. Q.; Van Wagner, E.; Freeman, B. D.; Toy, L. G.; Gupta, R. P. Science 2006, 311,
639-642.

(8)

Ockwig, N. W.; Nenoff, T. M. Chem. Rev. 2007, 107, 4078-4110.

(9)

Perry, E. Process for the Recovery of Hydrogen from Ammonia Purge Gases. U.S. Patent
4,172,885, Oct. 30, 1979.

(10)

Kesting, R. E.; Fritzsche, A. K. Polymeric Gas Separation Membranes. John Wiley and
Sons, Inc.: New York City, NY, 1993.

(11)

Matsuyama, H.; Teramoto, M.; Matsui, K.; Kitaura, Y. J. Appl. Polym. Sci. 2001, 81, 936942.

(12)

Matsuyama, H.; Teramoto, M.; Sakakura, H.; Iwai, K. J. Membr. Sci. 1996, 117, 251-260.

(13)

Kim, M.; Park, Y.; Youm, K.; Lee, K. J. Membr. Sci. 2004, 245, 79-86.

(14)

Noble, R. D.; Pellegrino, J. J.; Grosgogeat, E.; Sperry, D.; Way, J. D. Sep. Sci. Technol.
1988, 23, 1595-1609.

(15)

Way, J. D.; Noble, R. D. I. Facilitated Transport. In Membrane Handbook; Ho, W. S. W.;
Sirkar, K. K., Eds.; Van Nostrand Reinhold: New York, 2001; pp 833-866.

(16)

Koros, W. J. J. Polym. Sci., Part B: Polym. Phys. 1985, 23, 1611-1628.

(17)

Zhong, J.; Lin, G.; Wen, W.-Y.; Jones, A. A.; Kelman, S.; Freeman, B. D. Macromolecules
2005, 38, 3754-3764.

(18)

Bondar, V. I.; Freeman, B. D.; Pinnau, I. J. Polym. Sci., Part B: Polym. Phys. 2000, 38,
2051-2062.

74

(19)

Kusuma, V. A.; Matteucci, S.; Freeman, B. D.; Danquah, M. K.; Kalika, D. S. J. Membr.
Sci. 2009, 341, 84-95.

(20)

Patel, N. P.; Miller, A. C.; Spontak, R. J. Adv. Funct. Mater. 2004, 14, 699-707.

(21)

Kalakkunnath, S.; Kalika, D. S.; Lin, H.; Freeman, B. D. Macromolecules 2005, 38, 96799687.

(22)

Priola, A.; Gozzelino, G.; Ferrero, F.; Malucelli, G. Polymer 1993, 34, 3653-3657.

(23)

Lin, H.; Freeman, B. D.; Kalakkunnath, S.; Kalika, D. S. J. Membr. Sci. 2007, 291, 131139.

(24)

Balachandra, A. M.; Baker, G. L.; Bruening, M. L. J. Membr. Sci. 2003, 227, 1-14.

(25)

Sun, L.; Baker, G. L.; Bruening, M. L. Macromolecules 2005, 38, 2307-2314.

(26)

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

(27)

Zheng, Y.; Bruening, M. L.; Baker, G. L. Macromolecules 2007, 40, 8212-8219.

(28)

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

(29)

Bao, Z.; Bruening, M. L.; Baker, G. L. Macromolecules 2006, 39, 5251-5258.

(30)

Kim, J.-B.; Huang, W.; Miller, M. D.; Baker, G. L.; Bruening, M. L. J. Polym. Sci., Part
A: Polym. Chem. 2003, 41, 386-394.

(31)

Cheng, N.; Azzaroni, O.; Moya, S.; Huck, W. T. S. Macromol. Rapid Commun. 2006, 27,
1632-1636.

(32)

Zhou, F.; Zheng, Z.; Yu, B.; Liu, W.; Huck, W. T. S. J. Am. Chem. Soc. 2006, 128, 1625316258.

(33)

Brown, A. A.; Khan, N. S.; Steinbock, L.; Huck, W. T. S. Eur. Polym. J. 2005, 41, 17571765.

(34)

Lad, J.; Harrisson, S.; Haddleton, D. M., Mechanistic Aspects of Copper-mediated Living
Radical Polymerization. American Chemical Society: Washington, DC, 2003; Vol. 854, p
148-160.

(35)

Huang, W.; Kim, J.-B.; Bruening, M. L.; Baker, G. L. Macromolecules 2002, 35, 11751179.

(36)

Wang, X. S.; Armes, S. P. Macromolecules 2000, 33, 6640-6647.

(37)

Wang, X. S.; Lascelles, S. F.; Jackson, R. A.; Armes, S. P. Chem. Commun. 1999, 18171818.

75

(38)

Jain, P.; Dai, J.; Grajales, S.; Saha, S.; Baker, G. L.; Bruening, M. L. Langmuir 2007, 23,
11360-11365.

(39)

Tsarevsky, N. V.; Matyjaszewski, K. Chem. Rev. 2007, 107, 2270-2299.

(40)

Miyazawa, T.; Ideguchi, Y.; Fukushima, K. J. Chem. Phys. 1962, 37, 2764-2776.

(41)

Yoshihara, T.; Murahashi, S.; Tadokoro, H. J. Chem. Phys. 1964, 41, 2902-2911.

(42)

Takahashi, Y.; Tadokoro, H. Macromolecules 1973, 6, 672-675.

(43)

Li, X.; Hsu, S. L. J. Polym. Sci., Part B: Polym. Phys. 1984, 22, 1331-1342.

(44)

Schönherr, H.; Frank, C. W. Macromolecules 2003, 36, 1188-1198.

(45)

Schönherr, H.; Frank, C. W. Macromolecules 2003, 36, 1199-1208.

(46)

Wang, H. P.; Keum, J. K.; Hiltner, A.; Baer, E.; Freeman, B.; Rozanski, A.; Galeski, A.
Science 2009, 323, 757-760.

(47)

Bitter, J. G. A. Desalination 1984, 51, 19-35.

(48)

Preston, W. E.; Barlow, J. W.; Paul, D. R. J. Appl. Polym. Sci. 1984, 29, 845-852.

(49)

Lin, H.; Kai, T.; Freeman, B. D.; Kalakkunnath, S.; Kalika, D. S. Macromolecules 2005,
38, 8381-8393.

(50)

Lin, H.; Freeman, B. D. J. Membr. Sci. 2004, 239, 105-117.

(51)

Lin, H.; Freeman, B. D. Macromolecules 2005, 38, 8394-8407.

(52)

Lin, H.; Freeman, B. D. J. Mol. Struct. 2005, 739, 57-74.

(53)

Kelman, S. D.; Raharjo, R. D.; Bielawski, C. W.; Freeman, B. D. Polymer 2008, 49, 30293041.

(54)

White, L. S.; Blinka, T. A.; Kloczewski, H. A.; Wang, I.-f. J. Membr. Sci. 1995, 103, 7382.

76

Chapter Three
The effect of nanoparticles on CO2-selective polymer membranes
Introduction
Dispersion of nanoparticles in polymers began in 1987 when the Toyota group exchanged
the surface cations of montmorillonite with quaternary ammonium surfactants, causing the
hydrophilic clay to become compatible with hydrophobic polymers.1 Compared to traditional
inorganic polymer fillers (particle sizes on the micron and larger scale), nanoscale fillers provide
significantly higher increases in transition temperatures (glass transition, melting, and critical
temperatures), toughness, stiffness, and strength, at lower loadings. A series of studies show that
loading of <5 wt% clay nanoparticles in a polymer composite can double the modulus and strength,
increase the heat distortion temperature by almost 100˚C, and enhance barrier properties and flame
retardancy compared to the pristine polymer.2-7 Table 3.1 summarizes some of the property
enhancements and potential applications available with nanocomposites, and several review
articles examined this subject.8-11
With respect to the focus of this dissertation on separation membranes, incorporation of
high aspect ratio (high anisotropy) nanoparticles into films can enhance barrier properties,12, 13 or,
conversely incorporation of low aspect ratio (low anisotropy) nanoparticles into rigid, high freevolume polymers can enhance permeation and selectivity.14, 15 These seemingly contradictory
effects of nanoparticles on permeation properties stem from the size and shape of the particle filler
as well as the rigidity of the polymer chain.

77

Table 3.1: Selected nanoscale fillers in polymer materials and their corresponding applications.
POSS is polyhedral oligomeric silsesquioxane.
Property
enhancement(s)
Flame retardancy, gas
barriers
Viscosity modification,
permeability

Ag

Nanofiller

Application

Refs.

Coatings

16-19

Paint, adhesives,
membranes

14, 20-23

Antimicrobial activity

Coatings

24-31

Carbon nanotubes

Electrical conductivity,
charge transport

Electronics,
optoelectronics

32-35

CdSe, CdTe, TiO2

Charge transport

Photovoltaic cells

36-42

Electronics

43-46

Sensors, LEDs

47-50

Exfoliated clay
SiO2, MgO

Graphene
POSS

Electrical conductivity,
gas barriers, charge
transport
Improved stability,
flammability resistance

HfO2

Refractive index

Photoresists, optics

51, 52

Zeolite, mol. sieve

Permselectivity

Mixed matrix
membranes

15

Barrier enhancement occurs when impermeable, long particles in a film create tortuous paths for
permeation through the membrane (Figure 3.1). Several models were used to describe penetrant
diffusion in composite membranes, some of which consider particle dimension and dispersion.53,
54

78

Figure 3.1: Cross-sectional scheme of the tortuous path of a permeant molecule traversing a
polymer film containing dispersed, impermeable 2-D nanoparticles.
The most common models rely on the Maxwell equation (Eq. 3.1), which relates the
particle loading to the change in permeability.20 In this equation,
 1  F 
PC  PM 

 1   F / 2  

(3.1)

PC and PM are composite and matrix permeabilities, respectively, and F is the filler volume
fraction when the filler consists of dispersed, impermeable, spherical particles. Figure 3.2 plots
the permeability of the composite as a function of filler loading for a completely impermeable
filler and a polymer matrix with a permeability of 1 Barrer.

79

Permeability (Barrer)

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Filler volume fraction

Figure 3.2: Permeability of a composite film as a function of filler volume fraction as predicted by
the Maxwell equation. The polymer matrix has a permeability of 1 Barrer, and the filler is
impermeable.
The Maxwell model effectively describes permeation through composite membranes that
contain traditional fillers, but it does not work for many nanocomposite films. Contradictory to
Maxwell’s equation, incorporation of nanoparticles into a rigid polymer matrix can increase
permeability, probably because the filler disrupts polymer chain packing to increase fractional free
volume.20-23 Freeman and coworkers incorporated MgO nanoparticles into poly(1-trimethylsilyl1-propyne) to increase CO2 permeability by a factor of 17 and H2 permeability by a factor of 40,
which decreased CO2/H2 selectivity by a factor of 2.20 In addition to the effects that nanoparticles
have on fractional free volume, they may also interfere with polymer crystallization. Inhibition of
the room temperature crystallization of poly(ethylene glycol) (PEG) side chains in gas separation
membranes was previously reported by cross-linking55-57 and by copolymerizing two monomers
with different PEG chain lengths as described in Chapter 2.58, 59 Several studies showed that
incorporation of montmorillonite, saponite, TiO2, or Al2O3 in bulk PEG inhibits crystallization.60,

80

61

Additionally, PEG crystal morphology changed from spherical to disk-like upon incorporation

of 1 wt.% carbon nanotubes.62
This chapter examines the effects of titania nanoparticles on polymer films and membranes
generated using atom transfer radical polymerization (ATRP) of poly(ethylene glycol) methyl
ether

methacrylate (PEGMEMA).

ATRP is attractive for synthesizing thin (50-500 nm)

membrane skins from porous substrates because it affords control over film thickness along with
polymer chains with relatively low polydispersity.63, 64 Moreover, as Chapter 2 demonstrates,
poly(PEGMEMA) membranes selectively remove CO2 from H2 streams if crystallization of PEG
chains can be prevented. This chapter explore whether the presence of TiO2 nanoparticles in
PEGMEMA films can inhibit PEG crystallization to maintain high selectivities in PEGMEMA
membranes. The ATRP process for forming membranes on porous alumina substrates, and similar
processes on porous polymer supports are discussed in Chapters 1 and 2.
Figure 3.3 shows the method we employ to generate titania in the poly(PEGMEMA) grown
from porous supports. The tetraisopropoxytitanate (TPOT) adsorbs into the polymer brush and
subsequent exposure to humidity generates a nanocomposite polymer brush with TiO2
nanoparticles. Yuan and coworkers described the use of polymer brushes as templates to generate
titania nanowires,65 but to our knowledge this is the first example of a nanocomposite polymer
brush membrane. (With regard to terminology, composite membranes consist of multiple layers,
typically a support layer with large pores and a selective layer. Composite films contain a mixture
of two species, in this case polymer and TiO2. In this chapter the term nanocomposite will only
apply to a polymer/TiO2 film, which may serve as the skin layer of a membrane.)

81

Porous
Alumina

Polymer Brush

ethanol

+
Poly(PEGMEMA)

Tetraisopropoxytitanate
(TPOT)
H2O

=

=

TiO2 nanoparticle

Poly(PEGMEMA)-TiO2

Figure 3.3: Schematic drawing of the formation of nanoparticles to generate a nanocomposite
polymer brush.

Experimental
Chemicals. PEGMEMA with Mn = 1,100 Da (PEGMEMA) was obtained from Aldrich
and used as received. This monomer contains both 100 ppm 4-methoxyphenol and 300 ppm 2,6ditert-butyl-4-methylphenol as inhibitors. Proton NMR spectra (see Appendix A) show that the
average number of ethylene oxide units in PEGMEMA is 23.1.

The CuCl, CuBr2, and

1,1,4,7,10,10-hexamethyltriethylenetetramine (HMTETA) were purchased from Aldrich, and
deionized water was obtained using a Millipore system (Milli-Q, 18.2 MΩ·cm). Toluene was used
as received. Silicon(100) wafers were obtained from NOVA Electronic Materials and sputtercoated with 20 nm of chromium followed by 200 nm of gold by LGA Thin Films (Santa Clara,
CA). The porous alumina substrates were Anodisc membrane filters (25 mm disks with 0.02 µm

82

surface pores) purchased from Whatman, and the disulfide, [Br-C(CH3)2-COO(CH2)11S]2, and
silane, SiCl(CH3)2(CH2)11OCOC(CH3)2Br, initiators were synthesized as described previously.6669

Polymerization. Porous alumina substrates were initially rinsed with ethanol, dried with
N2, and cleaned with UV/O3 (Boekel UV-Clean Model 135500) for 15 min. Attachment of a
monolayer of the disulfide initiator to gold-coated Si wafers or porous alumina occurred during an
overnight immersion of the substrate in an ethanolic solution containing 1 mM [Br-C(CH3)2COO(CH2)11S]2. The resulting sample was rinsed with 5 mL of ethanol and dried in a N2 stream.
Attachment of the silane initiator to alumina occurred during overnight immersion of the uncoated,
cleaned porous alumina substrates in a solution containing 8 mM silane in THF. The resulting
sample was rinsed with 5 mL THF and dried in a N2 stream. The two different types of initiatormodified membranes were handled identically in the ensuing polymerization steps.
Following a previous procedure, the catalyst stock solution was prepared by dissolving
0.06 g (0.6 mmol) CuCl and 0.04 g (0.2 mmol) CuBr2 in 30.0 mL of distilled, degassed DMF.70, 71
The solution was further degassed via three freeze, pump, thaw cycles. In a N2-filled glove bag,
490 µL (1.8 mmol) of degassed 1,1,4,7,10,10-hexamethyltriethylenetetramine (HMTETA) ligand
was added to this solution, which subsequently turned a dark green color during several hours of
stirring. During this initial stirring, a small amount of precipitation occurred, but the resulting
catalytic activity of the solution remained constant for at least several weeks. The mole ratio of
Cu+:Cu2+:HMTETA was 3:1:9.
The monomer solution, which contained 0.75 M PEGMEMA in water, was initially
degassed via three freeze, pump, thaw cycles. The polymerization solution was prepared in a N2filled glove bag by combining degassed monomer and catalyst solutions in a 9:1 volume ratio to

83

produce a blue-green solution containing 0.67 M monomer, 2.0 mM CuCl, 0.60 mM CuBr2, and
6.0 mM HMTETA.

The initiator-modified membranes or wafers were immersed in the

polymerization solution for designated periods of time, and the resulting polymer films were rinsed
with 5 mL of water and soaked in water for at least two hours before rinsing with 5 mL of ethanol
and drying under a N2 stream.
Formation of the Polymer-nanoparticle Composite Film. Preparation of the ethanolic
TPOT solution at predetermined concentrations was performed in a N2-filled glove bag.
PEGMEMA-modified alumina membranes were immersed in ethanol overnight to ensure that the
polymer film was completely swollen, and the membrane was then rinsed with the TPOT solution
three times, and immersed in this solution for one hour. After removing the membrane from
solution and allowing it to dry in the glove bag for approximately 10 minutes (without rinsing), it
was placed in air for several hours to generate TiO2. Once the hydrolysis of TPOT was completed
the membrane could be rinsed with ethanol without changing its properties.
Characterization methods.

Reflectance FT-IR spectroscopy was performed using a

Nicolet Magna-IR 560 spectrometer with a Pike grazing angle (80°) attachment. A UV/ozonecleaned, gold-coated Si wafer was used to obtain the background spectrum. Thicknesses of films
formed on gold-coated (100) silicon were determined with a rotating analyzer ellipsometer (J.A.
Woollam model M44) at an incident angle of 75°. A two-term Cauchy equation was employed to
simultaneously fit film thickness as well as the A and B Cauchy constants used to model the
wavelength-dependence of the refractive index. Thicknesses were determined at a minimum of
three locations on each substrate. Scanning electron microscopy was performed using the JEOL
semi-in-lens cold cathode field-emission scanning electron microscope, model JSM-7500F,
operating with a r-filter in signal maximum mode. Samples were sputter coated with 5 nm of gold

84

prior to imaging. Cross-polarized microscope images were obtained with a Nikon Optiphot 2POL polarizing optical microscope equipped with a Mettler FP82HT hot stage and a Sony DXC151A CCD camera.
Gas Permeation. Mixed-gas permeation experiments were performed by loading the
membrane into a custom holder (all connections utilized Swagelok fittings and were tested to
ensure that they maintained pressure over a time-scale longer than the experiment) that allowed
cross-flow of the feed gas as well as a sweep gas on the permeant side. A backpressure valve was
employed to maintain the feed gas pressure, and the feed flow rate was high enough to maintain
constant composition at the face of the membrane (the stage-cut, or ratio of permeant flow to feed
flow, was < 1%). The feed gas streams used for CO2 permeances were either pure CO2 or 20%
CO2, 80% H2, and the H2 permeances were measured using a feed gas stream of 80% H2, 20% CO2
to simulate an industrially realistic gas stream. The gas flow rates were measured with either an
Optiflow 420 electronic soap bubble meter or a manual soap bubble meter. The N2 sweep
gas/permeant stream was connected to an automated six-port injector valve on a Hewlett-Packard
6890 GC equipped with a thermal conductivity detector (TCD) and an Agilent GS-CarbonPLOT
capillary column (i.d. = 0.53 mm, length = 30 m, 3 µm coating). The mixed gas permeances as
well as pure gas permeances were calculated from GC results using Eq. (3.2),

Permeance 

B   N

2

 N  A  Δp

(3.2)

2

where

 B is the mole fraction of a particular permeate gas, B, in the gas mixture injected into the

GC column,  N2 is the sweep gas flow rate,  N 2 is the mole fraction of N2 in the gas injected into
the GC column, Δp is the transmembrane partial pressure difference for the gas of interest, and A
is the membrane area. (  N2 /  N 2 is the sum of the permeate and sweep gas flow rate, and  N 2 was

85

determined by subtracting the mole fractions of the permeates from unity. This assumes that the
amount of N2 passing through the membrane was negligible.) The sweep gas flow rate was
programmed into a mass flow controller, pressure and area were measured, and the mole fractions
were determined from the integrated GC spectra and a calibration curve.
Results and Discussion
Membrane Synthesis. The properties of composite polymer films depend on the filler
shape and loading, interfacial interactions between the filler and polymer, and the extent of
dispersion of filler particles. Dispersion of the particles in the polymer often requires modification
of the nanoparticle surface or addition of compatibilizers to the polymer matrix. For example,
dispersion of montmorillonite clays in nylon occurs only after surface exchange of Na+ ion with
anionic surfactants,1 and blending of maleic anhydride grafted polymers with polyethylene aids in
the nanoparticle dispersion in polyethylene.72, 73 The use of either of these techniques in a polymer
film that is less than 500 nm thick poses the added complication that the surface-modifier or
compatibilizer might contribute to the film properties.
Incorporation of nanoparticles in a polymer can occur through (1) polymerization of a
monomer mixture that contains inorganic nanoparticles, (2) simultaneous condensation of
inorganic precursors and polymerization of monomer, or (3) polymerization of monomer to
generate a film followed by incorporation of the inorganic nanoparticles in the film.74 The different
approaches have specific benefits and drawbacks for different applications. Typically, method (1)
is appropriate when filler particles disperse well in the monomer. The second case is not common
because nanoparticle synthesis often occurs at high temperatures and pressures that are
incompatible with polymerization. The third method is common when the filler particles disperse

86

in the polymer matrix and is attractive because the nanoparticle and polymer syntheses are
independent.
In the case of high-flux gas separation membranes, in addition to the requirement of
nanoparticle dispersion, the nanocomposite film must form a defect free skin layers to maintain
reverse selectivity. Preliminary experiments suggested that the presence of inorganic precursors
or particles in the polymerization solution changes the rate of ATRP and increases the probability
of generating defects. Thus, the synthesis strategy that we employ in this work is a variation of
(3), generation of the polymer skin prior to incorporation of inorganic precursor into the film and
subsequent hydrolysis to produce dispersed nanoparticles, as shown in Figure 3.19.
Inhibition of Crystallization by TiOx Nanoparticles. Immediately after preparation,
poly(PEGMEMA) membranes are amorphous and exhibit a CO2/H2 selectivity of 13. However,
even after one day, crystallization of PEG side chains decreases the CO2 permeance by >85% and
the CO2/H2 selectivity by >70%. (The CO2 flux drops below the detection limit of the GC, which
is optimized for H2 detection.) PEG crystallizes in helices containing seven CH2CH2O repeat units
in two turns (a 72 helix) over a length of 19.3 Å, and helix formation causes a notable splitting of
the PEG C-O-C IR band (1150 cm-1) into peaks at 1120 and 1148 cm-1 as well as a narrowing of
the CH2 wagging peak at 1360 cm-1.75, 76 In PEGMEMA films, PEG side chains can also crystallize
in this manner, and reflectance infrared spectra taken after storage of films overnight at room
temperature reveal the crystallization of homopolymers when PEG chains contain more than 9
CH2CH2O repeat units. Figure 3.4 demonstrates the changes in reflectance FTIR spectra that occur
upon crystallization of a 44 nm thick poly(PEGMEMA) film.67, 77, 78

87

88

crystalline

amorphous

Absorbance

0.005

1600

1400

1200
1000
Wavenumbers (cm-1)

800

Figure 3.4: Reflectance FTIR of poly(PEGMEMA) immediately after synthesis (dashed red line)
and after 24 hours at room temperature (black line).
Cross-polarized microscopy provides an additional method of qualitatively identifying the
presence of crystalline polymer because only the crystalline material rotates light prior to its
passage through the second polarizer. The appearance of “maltese-cross” patterns in crystallized
poly(PEGMEMA) films indicates the formation of circular crystalline domains similar to slices of
spherulites. After one day the cross-polarized microscopy image of a poly(PEGMEMA) film is
completely covered with Maltese cross patterns, shown in Figure 3.5.

89

Figure 3.5: Surface of poly(PEGMEMA) after a day at room temperature viewed through a crosspolarized optical microscope.
FTIR spectra of poly(PEGMEMA) and poly(PEGMEMA)-TiO2 films on gold-coated silicon
confirm the presence of TiO2 nanoparticles in the film after treatment with TPOT and hydrolysis
(Figure 3.6). After the film was soaked in THF with TPOT overnight, a broad peak from 880 to
1084 cm-1 appears.

This stems from Ti-O stretching and confirms the presence of TiOx

nanoparticles.

90

poly(PEGMEMA)
poly(PEGMEMA)-TiO2

Absorbance

0.02

1600

1400
1200
1000
Wavenumbers (cm-1)

800

Figure 3.6: Reflectance FTIR spectra of a 44 nm-thick poly(PEGMEMA) film before and after
treatment with TPOT to form nanoparticle films. Spectra were obtained immediately after film
formation before crystallization could occur.
After crystallization of poly(PEGMEMA) films, the peak height at 1120 cm-1 typically
increases relative to that at 1148 cm-1 (Figure 3.4). Comparison of the spectra in Figure 3.6 shows
that the ratio of these two peak heights changes only marginally for poly(PEGMEMA)-TiO2 films,
indicating that the presence of the nanoparticles retards crystallization of PEG chains. (In fact, the
small increase in the 1120 cm-1 peak relative to the 1148 cm-1 peak may be due to further formation
of TiO2 due to hydrolysis.)

91

as made

24 hours later

Absorbance

0.02

1600

1400

1200
1000
Wavenumbers (cm-1)

800

Figure 3.7: Reflectance FTIR spectra of a 44 nm-thick poly(PEGMEMA)-TiO2 film generated by
soaking in 0.1 M TPOT solution overnight. The spectra taken immediately upon removal from
the TPOT solution (dashed red) and 24 hours later (black) showed little difference.
Changes in Membrane Permeability due to Nanoparticle Formation. In addition to
inhibiting PEG crystallization, the presence of TiO2 in poly(PEGMEMA) membranes alters their
permeability. However, membrane properties vary with the concentration of TPOT used to
introduce the Ti. As Figure 3.6 shows, the H2 and CO2 permeances of day-old poly(PEGMEMA)TiO2 membranes are relatively constant when forming nanoparticles using TPOT concentrations
ranging from 0.01 to 0.05 M. Most importantly, the CO2 permeance is at least 5-fold greater than
the corresponding permeance through day-old poly(PEGMEMA), suggesting that the presence of
TiO2 inhibits crystallization. (After one day, the CO2 permeance of poly(PEGMEMA) membranes

92

is <6×10-8 cm3 (STP) cm-2 s-1 cmHg-1. This value is <4% of the permeance through a fresh
membrane.)
If the TPOT concentration in the modification solution is 0.7 or 1 M, the membrane
permeance increases by several orders of magnitude, and selectivity drops to 1. The TPOT
treatment and hydrolysis presumably lead to an increase in the free volume until defects form in
the presence of high TPOT concentrations.
For nanocomposite films made from deposition solutions containing less than 0.05 M
TPOT, the CO2 permeances of the as-prepared membrane is similar to that of a freshly prepared,
untreated poly(PEGMEMA) membrane.

After one day of aging, the CO2 permeance of

membranes modified with 0.01 to 0.05 M TPOT declines by 85% due to some crystallization, but
the CO2 permeance of untreated membranes declines by > 96%. Thus low amounts of TiO2
nanoparticles are likely insufficient to inhibit crystallization.

93

(cm3 STP

Permeance
cm-2 s-1 cmHg-1)

1 10+0
CO2

1 10-1

H2

1 10-2
1 10-3
1 10-4

1 10-5
1 10-6
1 10-7
1 10-8
0

0.02
0.04
0.06
0.08
TPOT concentration (M)

0.1

Figure 3.8: Permeances of poly(PEGMEMA)-TiO2 membranes as a function of the TPOT
concentration employed for nanoparticle formation. The measurements were taken after
membrane aging overnight at room temperature. The black triangles correspond to CO2 and the
red circles to H2. (The CO2 permeance of the unmodified membrane, 0 M TPOT, was below the
detection limit of the system.)
The initial CO2/H2 selectivity was 12-13 for all the membranes prior to crystallization or
treatment with TPOT. Immediately after formation of TiO2 the membrane selectivity remains at
values of 12-13 for samples made using a modification solution containing less than 0.05 M TPOT.
In contrast, for the membranes modified using solutions containing 0.07 and 1 M TPOT, the
CO2/H2 selectivities decrease to ~1 immediately after TPOT treatment, showing the formation of
defects. Thus, effective membrane modification is limited to solutions containing ≤0.05 M TPOT.
After storage at room temperature for one day, the selectivity of the pristine polymer drops
to less than 1 due to crystallization, but the TiO2-loaded nanocomposites maintain selectivities of
around 6 if the TPOT concentration in the modification solution is 0.05 M or less (Figure 3.8).

94

Thus the TiO2 nanoparticles likely reduce the rate of crystallization, although they may not prevent
it. Longer term studies are needed to address whether incorporation of nanoparticles can truly
prevent most crystallization to retain some level membrane selectivity.

12

CO2/H2 selectivity

10
8
6
4
2
0
0

0.02

0.04

0.06

0.08

0.1

TPOT concentration (M)

Figure 3.9: CO2/H2 selectivities of poly(PEGMEMA-1100) membranes as a function of the TPOT
concentration used to modify the membranes prior to aging for one day at room temperature.
Conclusions
The PEG side chains in poly(PEGMEMA) films grown from porous alumina provide a
high initial CO2/H2 selectivity of 12-13, but crystallization in less than a day at room temperature
results in a drastic reduction of permeance and selectivity. The incorporation of TiO2 nanoparticles
frustrates some of the crystallization and allows the membrane to retain CO2/H2 selectivities of
around 6 while maintaining the permeability typical of an amorphous film. With a TPOT
deposition solution containing 0.01 M - 0.05 M TPOT, the loading of TiO2 nanoparticles was high

95

enough to stop some crystallization and low enough to avoid the formation of defects that negate
selectivity.

96

REFERENCES

97

REFERENCES

(1)

Fukushima, Y.; Inagaki, S. Journal of Inclusion Phenomena 1987, 5, 473-482.

(2)

Usuki, A.; Kojima, Y.; Kawasumi, M.; Okada, A.; Fukushima, Y.; Kurauchi, T.;
Kamigaito, O. J. Mater. Res. 1993, 8, 1179-1184.

(3)

Usuki, A.; Kawasumi, M.; Kojima, Y.; Okada, A.; Kurauchi, T.; Kamigaito, O. J. Mater.
Res. 1993, 8, 1174-1178.

(4)

Kojima, Y.; Usuki, A.; Kawasumi, M.; Okada, A.; Kurauchi, T.; Kamigaito, O. J. Poly.
Sci. Part A-Poly. Chem. 1993, 31, 983-986.

(5)

Kojima, Y.; Usuki, A.; Kawasumi, M.; Okada, A.; Fukushima, Y.; Kurauchi, T.;
Kamigaito, O. J. Mater. Res. 1993, 8, 1185-1189.

(6)

Kojima, Y.; Fukumori, K.; Usuki, A.; Okada, A.; Kurauchi, T. J. Mater. Sci. Lett. 1993,
12, 889-890.

(7)

Gilman, J. W.; Kashiwagi, T.; Lichtenhan, J. D. Sampe Journal 1997, 33, 40-46.

(8)

Paul, D. R.; Robeson, L. M. Polymer 2008, 49, 3187-3204.

(9)

Powell, C. E.; Beall, G. W. Current Opinion in Solid State & Materials Science 2006, 10,
73-80.

(10)

Ray, S. S.; Okamoto, M. Prog. Polym. Sci. 2003, 28, 1539-1641.

(11)

Pinnavaia, T.; Beall, G., Polymer–clay Nanocomposites. John Wiley & Sons Ltd.: 2000.

(12)

Kim, S.; Kim, S. J. Appl. Polym. Sci. 2007, 103, 1262–1271.

(13)

Choi, W.; Kim, H.; Yoon, K.; Kwon, O.; Hwang, C. J. Appl. Polym. Sci. 2006, 100, 4875–
4879

(14)

Merkel, T.; Freeman, B.; Spontak, R.; He, Z.; Pinnau, I.; Meakin, P. Science 2002, 296,
519–522.

(15)

Zimmerman, C.; Singh, A.; Koros, W. J. Membr. Sci. 1997, 137, 145–154.

(16)

Manias, E.; Touny, A.; Wu, L.; Strawhecker, K.; Lu, B.; Chung, T. C. Chem. Mater. 2001,
13, 3516-3523.

(17)

Morgan, A. B. Polym. Adv. Technol. 2006, 17, 206-217.

(18)

Peila, R.; Malucelli, G.; Lazzari, M.; Priola, A. Polym. Eng. Sci. 50, 1400-1407.

(19)

Alexandre, B.; Colasse, L.; Langevin, D.; Mederic, P.; Aubry, T.; Chappey, C.; Marais, S.
J. Phys. Chem. B 114, 8827-8837.

98

(20)

Matteucci, S.; Kusuma, V. A.; Kelman, S. D.; Freeman, B. D. Polymer 2008, 49, 16591675.

(21)

Matteucci, S.; Kusuma, V. A.; Sanders, D.; Swinnea, S.; Freeman, B. D. J. Membr. Sci.
2008, 307, 196-217.

(22)

Matteucci, S.; Kusuma, V. A.; Swinnea, S.; Freeman, B. D. Polymer 2008, 49, 757-773.

(23)

Matteucci, S.; Raharjo, R. D.; Kusuma, V. A.; Swinnea, S.; Freeman, B. D.
Macromolecules 2008, 41, 2144-2156.

(24)

Bryaskova, R.; Pencheva, D.; Kale, G. M.; Lad, U.; Kantardjiev, T. J. Colloid Interface
Sci. 2010, 349, 77-85.

(25)

Dai, J.; Bruening, M. L. Nano Lett. 2002, 2, 497-501.

(26)

Gray, J. E.; Norton, P. R.; Alnouno, R.; Marolda, C. L.; Valvano, M. A.; Griffiths, K.
Biomaterials 2003, 24, 2759-2765.

(27)

Grunlan, J. C.; Choi, J. K.; Lin, A. Biomacromolecules 2005, 6, 1149-1153.

(28)

Ignatova, M.; Labaye, D.; Lenoir, S.; Strivay, D.; Jerome, R.; Jerome, C. Langmuir 2003,
19, 8971-8979.

(29)

Jansen, B.; Kohnen, W. J. Ind. Microbiol. Biotechnol. 1995, 15, 391-396.

(30)

Lee, H. J.; Yeo, S. Y.; Jeong, S. H. J. Mater. Sci. 2003, 38, 2199-2204.

(31)

Shi, Z.; Neoh, K. G.; Kang, E. T. Langmuir 2004, 20, 6847-6852.

(32)

Collins, P. G.; Bradley, K.; Ishigami, M.; Zettl, A. Science 2000, 287, 1801-1804.

(33)

Huang, Y.; Duan, X. F.; Cui, Y.; Lauhon, L. J.; Kim, K. H.; Lieber, C. M. Science 2001,
294, 1313-1317.

(34)

Yao, Z.; Postma, H. W. C.; Balents, L.; Dekker, C. Nature 1999, 402, 273-276.

(35)

Poncharal, P.; Wang, Z. L.; Ugarte, D.; de Heer, W. A. Science 1999, 283, 1513-1516.

(36)

Kwong, C. Y.; Choy, W. C. H.; Djurisic, A. B.; Chui, P. C.; Cheng, K. W.; Chan, W. K.
Nanotechnology 2004, 15, 1156-1161.

(37)

Katsaros, G.; Stergiopoulos, T.; Arabatzis, I. M.; Papadokostaki, K. G.; Falaras, P. Journal
of Photochemistry and Photobiology a-Chemistry 2002, 149, 191-198.

(38)

He, J. A.; Mosurkal, R.; Samuelson, L. A.; Li, L.; Kumar, J. Langmuir 2003, 19, 21692174.

(39)

Huynh, W. U.; Dittmer, J. J.; Alivisatos, A. P. Science 2002, 295, 2425-2427.

(40)

Nozik, A. J. Physica E-Low-Dimensional Systems & Nanostructures 2002, 14, 115-120.

99

(41)

Leschkies, K. S.; Divakar, R.; Basu, J.; Enache-Pommer, E.; Boercker, J. E.; Carter, C. B.;
Kortshagen, U. R.; Norris, D. J.; Aydil, E. S. Nano Lett. 2007, 7, 1793-1798.

(42)

Levy-Clement, C.; Tena-Zaera, R.; Ryan, M. A.; Katty, A.; Hodes, G. Advanced Materials
2005, 17, 1512-1515.

(43)

Kim, H.; Macosko, C. W. Polymer 2009, 50, 3797-3809.

(44)

Kim, H.; Macosko, C. W. Macromolecules 2008, 41, 3317-3327.

(45)

Verdejo, R.; Barroso-Bujans, F.; Rodriguez-Perez, M. A.; de Saja, J. A.; Lopez-Manchado,
M. A. J. Mater. Chem. 2008, 18, 2221-2226.

(46)

Raghu, A. V.; Lee, Y. R.; Jeong, H. M.; Shin, C. M. Macromol. Chem. Phys. 2008, 209,
2487-2493.

(47)

Phillips, S. H.; Haddad, T. S.; Tomczak, S. J. Current Opinion in Solid State & Materials
Science 2004, 8, 21-29.

(48)

Zheng, L.; Farris, R. J.; Coughlin, E. B. Macromolecules 2001, 34, 8034-8039.

(49)

Jeon, H. G.; Mather, P. T.; Haddad, T. S. Polym. Int. 2000, 49, 453-457.

(50)

Waddon, A. J.; Coughlin, E. B. Chem. Mater. 2003, 15, 4555-4561.

(51)

Bae, W. J.; Trikeriotis, M.; Sha, J.; Schwartz, E. L.; Rodriguez, R.; Zimmerman, P.;
Giannelis, E. P.; Ober, C. K. J. Mater. Chem. 2010, 20, 5186-5189.

(52)

Zimmerman, P. A.; Rice, B.; Rodriguez, R.; Zettel, M. F.; Trikeriotis, M.; Wang, D. Y.;
Yi, Y.; Bae, W. J.; Ober, C. K.; Giannelis, E. P. J. Photopolym. Sci. Technol. 2008, 21,
621-629.

(53)

Xue, L.; Borodin, O.; Smith, G. D. J. Membr. Sci. 2006, 286, 293–300.

(54)

Barrer, R. M.; Barrie, J. A.; Rogers, M. G. J. Polym. Sci. A: Polym. Chem. 1963, 1, 2565–
2586.

(55)

Kalakkunnath, S.; Kalika, D. S.; Lin, H.; Freeman, B. D. Macromolecules 2005, 38, 96799687.

(56)

Patel, N. P.; Miller, A. C.; Spontak, R. J. Adv. Funct. Mater. 2004, 14, 699-707.

(57)

Priola, A.; Gozzelino, G.; Ferrero, F.; Malucelli, G. Polymer 1993, 34, 3653-3657.

(58)

Grajales, S. T.; Dong, X.; Zheng, Y.; Baker, G. L.; Bruening, M. L. Chem. Mater. 2010,
22, 4026-4033.

(59)

Grajales, S. T.; Dong, X.; Zheng, Y.; Baker, G. L.; Bruening, M. L. Polymer Preprints
2009, 50, 757.

(60)

Loyens, W.; Jannasch, P.; Maurer, F. H. J. Polymer 2005, 46, 915-928.

100

(61)

Croce, F.; Appetecchi, G. B.; Persi, L.; Scrosati, B. Nature 1998, 394, 456-458.

(62)

Jin, J.; Song, M.; Pan, F. Thermochim. Acta 2007, 456, 25-31.

(63)

Balachandra, A. M.; Baker, G. L.; Bruening, M. L. J. Membr. Sci. 2003, 227, 1-14.

(64)

Sun, L.; Baker, G. L.; Bruening, M. L. Macromolecules 2005, 38, 2307-2314.

(65)

Yuan, J.; Lu, Y.; FSchacher, F.; Lunkenbein, T.; Weiss, S.; Schmalz, H.; Muller, A. H. E.
Chem. Mater. 2009, 21, 4146-4154.

(66)

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

(67)

Zheng, Y.; Bruening, M. L.; Baker, G. L. Macromolecules 2007, 40, 8212-8219.

(68)

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

(69)

Bao, Z.; Bruening, M. L.; Baker, G. L. Macromolecules 2006, 39, 5251-5258.

(70)

Kim, J.-B.; Huang, W.; Miller, M. D.; Baker, G. L.; Bruening, M. L. J. Polym. Sci., Part
A: Polym. Chem. 2003, 41, 386-394.

(71)

Cheng, N.; Azzaroni, O.; Moya, S.; Huck, W. T. S. Macromol. Rapid Commun. 2006, 27,
1632-1636.

(72)

Chiu, F. C.; Yen, H. Z.; Chen, C. C. Polym. Test. 29, 706-716.

(73)

Koo, C. M.; Ham, H. T.; Kim, S. O.; Wang, K. H.; Chung, I. J.; Kim, D.-C.; Zin, W.-C.
Macromolecules 2002, 35, 5116-5122.

(74)

Azzaroni, O.; Brown, A. A.; Cheng, N.; Wei, A.; Jonas, A. M.; Huck, W. T. S. J. Mater.
Chem. 2007, 17, 3433-3439.

(75)

Miyazawa, T.; Ideguchi, Y.; Fukushima, K. J. Chem. Phys. 1962, 37, 2764-2776.

(76)

Yoshihara, T.; Murahashi, S.; Tadokoro, H. J. Chem. Phys. 1964, 41, 2902-2911.

(77)

Wang, H. P.; Keum, J. K.; Hiltner, A.; Baer, E.; Freeman, B.; Rozanski, A.; Galeski, A.
Science 2009, 323, 757-760.

(78)

Li, X.; Hsu, S. L. J. Polym. Sci., Part B: Polym. Phys. 1984, 22, 1331-1342.

101

Chapter Four
Guanidine-Based Antifouling, Ion-selective Membranes
This work was performed in collaboration with Dr. Viktoriya Konovalova of the Membrane
Research Center at the Kiev-Mohyla Academy where she measured flux, solute rejections and
antimicrobial activity.
Introduction
Reverse osmosis and nanofiltration are now commercially viable methods for desalination
and water softening, respectively.1,

2

Other membrane processes such as ultrafiltration and

microfiltration have also emerged as standard methods in municipal water treatment.3
Nevertheless, the formation of foulant layers on membrane surfaces is an ubiquitous problem in
membrane-based water purification, and fouling reduction and remediation procedures add
significant costs to these operations.4 Biofouling, fouling due to microorganisms, is particularly
problematic because even a few bacteria on a surface can multiply and cause significant decreases
in membrane permeability.5 The initial step in biofouling, cell adhesion, depends on substrate
properties such as roughness,6-8 topography,9 free energy,10,

11

and polymer composition.12-14

However, even on carefully crafted surfaces, deposition of extracellular polysaccharides
eventually allows attachment of microorgranisms.5
Existing measures to combat membrane fouling include pretreatment of the feed solution,15,
16

intermittent membrane cleaning,17,

18

incorporation of antibacterial metal particles in

membranes,19-21 and surface modification of membranes with ammonium groups.22-26 Surface
modification can in principle occur in existing membranes and avoid the need to reengineer a
process.

Of particular interest to this work, layer-by-layer adsorption of polycations and

102

polyanions (Scheme 4.1) provides a simple, well-controlled method for membrane modification,2733

and adsorption of polycationic biocidal polymers, e.g. polyguanidines and polybiguanidines,

creates films with antibacterial properties. These polycations exhibit low toxicity to humans along
with high antibacterial activity in aqueous solutions34,35 and on cotton.36

Step 1

a

+ + + + + +

Step 2
-

-

-

-

-

+ + + + + +

Substrate

-

b or c

Substrate

+

+
-

-

+
-

+
-

-

+-

+

-

+ + + + + +

Substrate
+

+
+
+
+
- -+
- - -+ + +
+
+
+
+
-

Repeat steps 1 and 2

…

…

+ + + + + +

Substrate
a

b

(CH 2 )6

H
N

c
H
N

NH
SO3 -

PSS

(CH 2 )6
10-12

H
N

H
N

NH

H
N

NH

10-12

n

PHMG

PHMB

Scheme 4.1: Preparation of an antibacterial polyelectrolyte multilayer film using (a)
poly(styrenesulfonate) (PSS), and
(b) poly(hexamethylene guanidine) (PHMG) or (c)
poly(hexamethylene biguanidine) (PHMB).
Scheme 4.2 shows the proposed mechanism by which PHMG kills bacteria.37 Interactions
between guanidinium groups and acidic phospholipids lead to aggregation of these phospholipids
and eventually membrane deformation and cell death.37-39
guanidinium groups in the PHMB repeat unit makes this

103

The presence of neighboring

Scheme 4.2: Mechanism of PHMB antibacterial action. (a) The healthy bacterial membrane
contains a mixture of acidic and neutral phospholipids. (b) PHMB adsorbs to the bacterial
membranes through interaction with phospholipds. (c) The affinity of PHMB for acidic
phospholipids leads to sequestration of these phospholipids under the polymer. (d) Lipid
sequestration leads to membrane deformation and eventually cellular leakage and death.
Reproduced with permission from 37

104

polymer potentially more antibacterial than PHMG. The minimum solution concentration to reach
an antibacterial activity of 100% is 5.0 ppm for PHMG and 1.25 ppm for PHMB.38, 40-42
Considering the need for antifouling coatings in membrane applications43 and the
previously reported ion-selective transport through polyelectrolyte multilayer membranes,32, 44 this
chapter explores the potential application of guanidine-containing polyelectrolyte multilayers as
membrane skins that resist biofouling and provide selective ion transport. We report the influence
of the number of PSS/PHMG or PSS/PHMB bilayers on surface charge, ion rejections, and
antibacterial properties.
Materials and Method
Preparation of films. Poly(sodium 4-styrenesulfonate) (PSS, Mw = 70,000 Da) was
purchased from Sigma-Aldrich, and poly(hexamethylene biguanidine) (PHMB, Vantocil IB) was
used as received from Arch Biocides.

The poly(hexamethylene guanidine), PHMG, was

synthesized by Dr. Viktorya Konovalova according to a previously published procedure.40, 45
Deionized water (Millipore purification system, 18.2 MΩcm) was used for membrane rinsing and
preparation of the polyelectrolyte solutions. PES membranes with a molecular-weight cutoff of 50
kDa were kindly provided by Pall Corporation (Port Washington, NY) and PES membranes with
a molecular-weight cutoff of 20 kDa were purchased from Microdyn-Nadir. All PES membranes
were initially soaked in 0.1 M NaOH for 1 h, then soaked in DI water for 24 h and rinsed with DI
water. For ellipsometry and reflectance FTIR studies, films were formed on Si wafers that were
sputter-coated with 200 nm of Al.
PSS deposition solutions contained 0.02 M PSS (polymer concentrations are always given
with respect to the repeating unit) in 0.5 M NaCl, and the pH of this solution was either left at 5.7
or adjusted to 3.4 with 0.1 M HCl. The poly(hexamethylene guanidine) deposition solution

105

contained 3.0 % (w/v) PHMG (0.169 M) in 0.5 M NaCl, and the pH of this solution was either left
at 7.4 or adjusted to 2.4 with 0.1 M HCl. PHMB was adsorbed from a solution containing 3%
(w/v) PHMB (0.15 M) in 0.5 M NaCl at an unadjusted pH of 5.4.
Formation of polyelectrolyte multilayer films involved alternating immersions of the
substrate in PSS and PHMG or PHMB solutions with a 1-min rinse with deionized water after
adsorption of each polymer. The adsorption time was 10 min. for the initial PSS layer and 5 min
for each subsequent layer.
Zeta potential values for coated PES membranes were calculated from streaming potentials
measured using an Anton Paar ElectroKinetic Analyzer. Reflectance FT-IR spectroscopy of films
on Al-coated Si was performed using a Nicolet Magna-IR 560 spectrometer with a Pike grazing
angle (80°) attachment. A UV/ozone cleaned aluminum-coated wafer served as the background
sample. Thicknesses of films formed on the coated wafers were determined with a rotating
analyzer ellipsometer (J.A. Woollam model M44) at an incident angle of 75°. The data were fit to
a model of an Al substrate covered by an Al2O3 layer and then the polymeric film. Thicknesses
were determined at a minimum of three locations on each aluminum-coated substrate.
Characterization of solute rejection. The following flux and solute rejection experiments
were performed by Dr. Konovalova. Pure water trans-membrane flux and solute rejection were
determined using an Amicon 8050 cylindrical dead-end filtration cell (Millipore) at room
temperature. The operating pressure was 300 kPa. Solute rejection was determined with solutions
containing 0.01 M NaCl or MgCl2, and conductivity measurements were used to determine
electrolyte concentrations in feed and permeate solutions. Solute rejection, R, was calculated using
Eq. (4.1),

106

 Cp
R(%)  1 
 C
f



 100%


(4.1)

where C f and C p are solute concentrations in the feed and permeate, respectively.
Evaluation of membrane antimicrobial activity.

Dr. Konovalova performed the

antibacterial studies described in this chapter. Future studies should be performed on 50 kDa PES
membranes to verify the findings. The microorganisms were gram-negative Escherichia coli BE
strains from the Ukrainian Collection of Microorganisms. The bacteria were cultivated in broth
overnight in an incubator shaker (250 rpm) at 37 °C, and the suspension was then diluted (1 mL
in 100 mL) with fresh, autoclaved broth, and further cultivated to mid-logarithmic phase as defined
by an optical density of 0.6-0.8 at a wavelength of 600 nm. The number of bacterial cells was then
adjusted to approximately 5×108 cells per mL (optical density at 600 nm of ~0.5) in broth.46 Cells
were collected by centrifugation (10 min, 3000 rpm), and then resuspended in sterile, deionized
water.
To test the antibacterial activity of modified and bare membranes, 3 mL of this bacteria
suspension was immediately filtered through a membrane in an Amicon dead-end filtration cell at
room temperature and 300 kPa. The membrane pore size is less than the dimensions of the cell, so
all bacterium should be deposited on the face of the membrane after filtration. After filtration, the
membranes were incubated on Endo Agar medium for 24 hours at 30oC.
Bactericidal activity, η, was calculated as follows:



N1  N 2
100%
N1

107

(1.49)

where N1 and N2 are the numbers of viable colonies on control and modified membranes,
respectively. Optical microscopy was used to count the bacterial colonies, and the control
membranes were unmodified PES.
Results
Membrane Surface Characterization. The goal of this work is to use layer-by-layer
adsorption of PSS/PHMG or PSS/PHMB films to create selective, highly permeable nanofiltration
membranes that resist bacterial growth. The first step in these studies is characterization of the
adsorption of PSS/PHMG or PSS/PHMB films on model substrates. Reflectance FTIR spectra of
these films deposited on aluminum-coated silicon wafers show strong absorption bands in the
1500-1700 cm-1, 1100-1250 cm-1, and 2800-3500 cm-1 regions (Figures 4.1, 4.2, and 4.4). Table
4.1 lists the polyelectrolyte vibrational modes that give rise to these absorbances.
Table 4.1: Strong infrared absorption modes for PSS/PHMG and PSS/PHMB films.
Wavenumbers (cm-1)

Absorption mode

Functional group

Ref.

3320

ν sym(H-N-H)

guanidine

47

3180

ν (N-H)

guanidine

47

guanidine

47

guanidine

47

2800 - 3000
1580 - 1690

ν asym and sym (H-N+H)
(N—C=N)
NH and NH2 deformation

1550 - 1640

ν (asym) (NCN)

guanidine

48

1588

ν (C=N)

guanidine

48

1550

ν (C=N)

biguanidine

48

1400 - 1500

ν (C-N) + δ (N-H)

guanidine

48

1120 - 1250

ν (OSO)

sulfonate

48

1009

Ring stretch

Monosubstituted ring

48

108

Figures 4.1 and 4.2 show the reflectance FTIR spectra of PSS/PHMG films from 36002800 and 1800-1000 cm-1, respectively. Absorbances increase regularly, suggesting controlled
layer-by-layer growth of these films. Figure 4.3 shows peak height versus the number of
PSS/PHMG bilayers for two specific IR absorbances, 1643 cm-1 and 1009 cm-1, that are
characteristic of PHMG and PSS, respectively.

(PSS/PHMG)66
(PSS/PHMG)55

0.004

Absorbance

(PSS/PHMG)44
(PSS/PHMG)3
3
(PSS/PHMG)2
2
PSS/PHMG

3600

3400

3200

3000

2800

Wavenumbers (cm-1)

Figure 4.1: Reflectance FTIR spectra (3600-2800 cm-1) of (PSS/PHMG)x films adsorbed on
aluminum-coated Si wafers.
With the exception of the 3-bilayer coating, the films show an approximately linear
increase in film thickness with the number of deposited layers. The figure may suggest an increase
in the amount adsorbed per layer after deposition of the third bilayer.

109

(PSS/PHMG)6
6
(PSS/PHMG)5
5
(PSS/PHMG)4
4
(PSS/PHMG)3
3
(PSS/PHMG)2
2
PSS/PHMG

Absorbance

0.004

1800

1600

1400

1200

1000

Wavenumbers (cm-1)

Figure 4.2: Reflectance FTIR spectra (1800-1000 cm-1) of (PSS/PHMG)x films adsorbed on
aluminum-coated Si wafers.
Often in the deposition of polyelectrolyte mulitlayers, deposition of a few priming layers
on the substrate precedes regular growth.32, 49 The spectra of (PSS/PHMB) films show trends
similar to those of (PSS/PHMG) films (Figures 4.4-4.6).

110

0.025
peak height at 1643 cm-1
peak height at 1008.6 cm-1

Absorbance

0.02

0.015
0.01

0.005
0
0

2

4

6

8

Bilayers of (PSS/PHMG)x
Figure 4.3: Peak heights (1643 and 1008.6 cm-1) in reflectance FTIR spectra of (PSS/PHMG)x
films as a function of the number of adsorbed bilayers, x.

111

0.005

(PSS/PHMB)6
(PSS/PHMB)5
(PSS/PHMB)4
(PSS/PHMB)2

Absorbance

PSS/PHMB

1800

1600

1400
Wavenumbers (cm-1)

1200

1000

Figure 4.4: Reflectance FTIR spectra (1800-1000 cm-1) of (PSS/PHMB)x films adsorbed on Alcoated Si wafers.

112

(PSS/PHMB)6
(PSS/PHMB)5

0.002

(PSS/PHMB)4
(PSS/PHMB)2

Absorbance

PSS/PHMB

3600

3400

3200
Wavenumbers (cm-1)

3000

2800

Figure 4.5: Reflectance FTIR spectra of 0.5-6 bilayers of (PSS/PHMB)x films adsorbed on
aluminum-coated Si wafers.

113

0.02

peak height at 1556 cm-1

0.018

peak height at 1008.6 cm-1

Absorbance

0.016

0.014
0.012
0.01
0.008

0.006
0.004
0.002
0
0

2

4

6

8

Bilayers of (PSS/PHMB) x

Figure 4.6: The peak heights of reflectance FTIR spectra of (PSS/PHMB)x films as a function of
the number of adsorbed bilayers, x. Absorption increases with every sequential layer.
Zeta Potentials. The rejection of ions in nanofiltration depends in part on Donnan
(electrostatic) exclusion, which is a function of the fixed charge in the membrane. The membrane
zeta potential, the potential at the slip plane outside the surface relative to the potential in the bulk
solution, provides an indication of the surface charge on modified membranes. Figure 4.7 shows
how the zeta potentials of modified PES membranes vary with the number of adsorbed PSS/PHMG
or PSS/PHMB layers.

114

50

Zeta Potential, mV

40

30
20
10
0
deposition pH = 7.4

-10

deposition pH = 2.4

-20
-30
0

1

2

3

4

5

6

Bilayers of (PSS/PHMG)x
Figure 4.7: Zeta potentials of (PSS/PHMG) multilayers deposited on 50 kDa PES membranes.
Each data point represents an average zeta potential of at least two samples generated on the same
day using the same solutions. The pH refers to the electrolyte deposition solution as described in
the experimental section. (Films with integer numbers of bilayers terminate with PHMG, whereas
fractional bilayer numbers indicate termination with PSS.)
The unmodified membranes are negatively charged, and thus they have an inherent
negative zeta potential. Deposition of the first PSS layer via hydrophobic interactions introduces
more negative charge on the surface and decreases the zeta potential. Subsequent adsorption of
PHMG at pH 7.4 changes the zeta potential from -25 mV to +43 mV, which is consistent with
charge reversal due to polycation adsorption. However, all further deposition steps at pH 7.4, up
to the formation of 5.5-bilayer films, give a zeta potential around 20 mV. The lack of negative
zeta potentials after PSS deposition could indicate a lack of polyelectrolyte adsorption, but this is
inconsistent with the nanofiltration flux declines after deposition of each layer (see below).
Another possible explanation for the consistently positive zeta potentials is that these film undergo

115

exponential growth where the polycation diffuses throughout the film, so that zeta potentials are
always positive.32 However, the IR data given above suggest a regular film growth after adsorption
of each layer. At this point, we do not understand the trends in zeta potentials.
Deposition pH may also affect the zeta potentials of PSS/PHMG films by altering the extent
of protonation of the guanidine groups of PHMG. The typical pKa1 of guanidine groups is 13.6
and in PHMB the pKa2 is 2.8.33 To increase the charge density on PHMG groups, we deposited
films at pH 2.4, and this results in a zeta potential of -20 mV after deposition of the second PSS
layer. However, for films consisting of 2 and 2.5-bilayers, zeta potentials were similar to those for
films deposited at pH 7.4, i.e., the zeta potential stays at 20 mV.
Antibacterial Activity. The dependence of the deposition solution on the antibacterial
activity of the films against E. coli was tested. A recent publication compared the bactericidal
activity of different oligomers (5 – 50 units) of bisphenol-A, propoxide, or methoxide, all chainend-modified with guanidine (end-caps).50

In solution, the oligomer chains with only two

guanidine moieties per chain showed a minimal inhibitory concentration of 180 ppm for
Pseudomonas aeruginosa and 140 ppm for Staphylococcus aureus.50 When these oligomers were
deposited on 0.05 µm pore size polyethylene terephthalate (PET) track-etched membranes, in a
manner similar to the one in this chapter, oligomer-modified membranes showed 100 %
bactericidal activity when the deposition solution was 1000 ppm guanidine-containing oligomer.50
Despite only having antibacterial active sites at the end of the oligomers, it was interpreted that
high activity was due to the ability of the guanidine groups to dangle away from the surface. In a
separate study, the adsorption of PHMB on cellulose with deposition solutions at these
concentrations have been attributed to a combination of electrostatic interactions and hydrogenbonding forces.33 The PHMG and PHMB used in this study contain significantly higher density

116

of guanidine units, and therefore should provide high bactericidal activity. Presumably, the
number of functional groups needed to provide sufficient adhesion is only a small fraction of the
guanidine units, which will leave guanidine moieties available as antibacterial sites.

100
Bactericidal Activity (%)

90
80

70
60
50
40

30

(PSS/PHMB)x

20

(PSS/PHMG)x

10
0
0

1

2

3
4
5
6
7
Number of Bilayers, x

8

9

Figure 4.8: The bactericidal activity of 20 kDa PES membranes coated with (PSS/PHMB) x and
(PSS/PHMG)x films. Activity was calculated by counting the number of colony forming units on
bare and coated membranes [see Eq. (1.2)].

Figure 4.8 shows the antibacterial activities of (PSS/PHMB)x and (PSS/PHMG)x films on PES
membranes. In both cases the antibacterial activity increases with the number of adsorbed bilayers,
and remarkably, the (PSS/PHMB) films reach 100 % bactericidal activity after deposition of 6
bilayers. The (PSS/PHMB) films show higher antibacterial activity than (PSS/PHMG) regardless
of the numbers of layers in the film, which is consistent with the higher density of guanidinium
groups in PHMB.

117

Transport properties. Figure 4.9 shows the pure water permeability, m3×m-2×d-1×MPa-1
at a feed pressure of 300 kPa, for PES membranes coated with different numbers of (PSS/PHMG)
and (PSS/PHMB) bilayers. As expected, flux decreases with the number of adsorbed bilayers, but
even after deposition of 4 bilayers, the membrane permeability is about 9-10% of that through
typical commercial membranes.51 Deposition of 1-bilayer films leads to a greater flux decline for
PSS/PHMB than PSS/PHMG, but with adsorption of more bilayers, the fluxes were essentially the
same for the two systems. To maximize flux, one should deposit as few bilayers as possible, but
the need for high NF selectivities and antibacterial properties will likely dictate the optimal number
of bilayers for a particular application.

118

18

(PSS/PHMG)x

Pw (m3×m-2×d-1MPa-1)

16
14

0.4

12

0.3

10

(PSS/PHMB)x

0.2

8
6

0.1

4

0
4

2

6

8

0
0

2
4
6
Number of Bilayers, x

8

Figure 4.9: Membrane pure water permeability as a function of the number of bilayers, x, for
(PSS/PHMG)x- and (PSS/PHMB)x-modified 20 kDa PES ultrafiltration membranes. The pure flux
was measured under a transmembrane pressure of 300 kPa.
Figures 4.10 and 4.11 show the NaCl and MgSO4 rejections of 20 kDa PES ultrafiltration
membranes coated with different numbers of PSS/PHMB or PSS/PHMG bilayers. Both the NaCl
and MgSO4 rejections are higher for PSS/PHMB than for PSS/PHMG films with the same number
of bilayers. However, in all cases, the rejection is less than 50%, suggesting that these membranes
are not very selective. Nevertheless, this conclusion must be viewed with some caution because
filtration in the dead-end cell can lead to very high concentration polarization, which will limit
rejection. Future studies will examine these membranes in cross-flow nanofiltration where the
degree of concentration polarization will be much less. Additionally, our previous studies show

119

that rejection depends on the substrate, and the 50 kDa Pall membranes do not lead to as high of
rejections as 50 kDa substrates from Millipore.44

45

(PSS/PHMB)x

NaCl Rejection (%)

40

(PSS/PHMG)x

35
30
25
20
15
10

5
0
0

2

4

6

8

10

Number of Bilayers, x
Figure 4.10: NaCl rejection in dead-end filtration of 0.01 M NaCl solutions through 20 kDa PES
ultrafiltration membranes coated with (PSS/PHMB)x (black triangles) and (PSS/PHMG)x (red
circles) films.

120

MgSO4 Rejection (%)

50
45

(PSS/PHMB)x

40

(PSS/PHMG)x

35
30
25

20
15
10
5
0
0

2

4

6

8

10

Number of Bilayers, x
Figure 4.11: MgSO4 rejection in dead-end filtration of 0.01 M MgSO4 solutions through 20 kDa
PES ultrafiltration membranes coated with (PSS/PHMB)x (black triangles) and (PSS/PHMG)x (red
circles) films.

121

60

(NH4)3PO4 Rejection (%)

(PSS/PHMB)x
50

(PSS/PHMG)x

40
30
20

10
0
0

2

4

6

8

10

Number of Bilayers, x
Figure 4.12: (NH4)3PO4 rejection in dead-end filtration of 0.01 M (NH4)3PO4 solutions through
20 kDa PES ultrafiltration membranes coated with (PSS/PHMB)x (black triangles) and
(PSS/PHMG)x (red circles) films.
Figure 4.12 shows the (NH4)3PO4 Rejections by PSS/PHMG and PSS/PHMB-modified
membranes. In this case, the PHMG and PHMB ammonium rejections were comparable, but the
rejections are still very low, suggesting that the membranes skins are highly swollen.52
Conclusions
Surface modification of porous PES membranes with either (PSS/PHMB)x or
(PSS/PHMG)x creates surfaces with antibacterial properties. Reflectance FTIR spectra confirm
the layer-by-layer growth of the polyelectrolyte films on aluminum-coated Si, but the zeta
potentials of coated membranes do not oscillate between positive and negative values after
deposition of polycations and polyanions, as would be expected. The 20 mV zeta potential of
films with more than 2 bilayers suggests an excess of PHMB or PHMG in these films. Decreases

122

in water flux with an increasing number of (PSS/PHMB) or (PSS/PHMG) bilayers on PES
membranes confirms regular film growth in spite of the constant zeta potential. Preliminary
filtration studies show low NaCl, MgSO4, and (NH4)3PO4 rejections, but this may occur due to a
high degree of concentration polarization in the dead-end filtration cells.

123

REFERENCES

124

REFERENCES

(1)

Bhattacharyya, D.; Hestekin, J.; Shah, D.; Ritchie, S. J. Chin. Inst. Chem. Eng. 2002, 33,
61-66.

(2)

Reissmann, F. G.; Uhl, W. Desalination 2006, 198, 225-235.

(3)

Kontek Ecology Systems, I. Microfiltration Systems.
http://www.kontekecology.com/files/micro.pdf (accessed Sept. 2, 2010).

(4)

Susanto, H.; Ulbricht, M. J. Membr. Sci. 2005, 266, 132-142.

(5)

Frank, B. P.; Belfort, G. Langmuir 1997, 13, 6234-6240.

(6)

Chambers, L. D.; Stokes, K. R.; Walsh, F. C.; Wood, R. J. K. Surf. Coat. Technol. 2006,
201, 3642.

(7)

Schumacher, J. F.; Carman, M. L.; Estes, T. G.; Feinberg, A. W.; Wilson, L. H.; Callow,
M. E.; Callow, J. A.; Finlay, J. A.; Brennan, A. B. Biofouling 2007, 23, 55.

(8)

Hovgaard, M. B.; Rechendorff, K.; Chevallier, J.; Foss, M.; Besenbacher, F. J. Phys.
Chem. B 2008, 112, 8241.

(9)

Genzer, J.; Efimenko, K. Biofouling 2006, 22, 339.

(10)

Callow, J. A.; Callow, M. E.; Ista, L. K.; Lopez, G.; Chaudhury, M. K. J. R. Soc.
Interface 2005, 2, 319.

(11)

Schmidt, D. L.; Brady, R. F.; Lam, K.; Schmidt, D. C.; Chaudhury, M. K. Langmuir
2004, 20, 2830.

(12)

Krishnan, S.; Ayothi, R.; Hexemer, A.; Finlay, J. A.; Sohn, K. E.; Perry, R.; Ober, C. K.;
Kramer, E. J.; Callow, M. E.; Callow, J. A.; Fischer, D. A. Langmuir 2006, 22, 5075.

(13)

Krishnan, S.; Wang, N.; Ober, C. K.; Finlay, J. A.; Callow, M. E.; Callow, J. A.;
Hexemer, A.; Sohn, K. E.; Kramer, E. J.; Fischer, D. A. Biomacromolecules 2006, 7,
1449.

(14)

HoipkemeierWilson, L.; Schumacher, J.; Carman, M.; Gibson, A.; Feinberg, A.; Callow,
M.; Finlay, J.; Callow, J.; Brennan, A. Biofouling 2004, 20, 53.

(15)

Dudley, L. Y.; Darton, E. G. Desalination 1997, 110, 11-20.

(16)

Ma, W.; Zhao, Y.; Wang, L. Desalination 2007, 203, 256-259.

(17)

Kwak, S.-Y.; Kim, S. H.; Kim, S. S. Environ. Sci. Technol. 2001, 35, 2388-2394.

(18)

Zhao, Y.-J.; Wu, K.-F.; Wang, Z.-J.; Zhao, L.; Li, S.-S. J. Env. Sci. 2000, 12, 241-251.

125

(19)

Grunlan, J. C.; Choi, J. K.; Lin, A. Biomacromolecules 2005, 6, 1149-1153.

(20)

Jansen, B.; Kohnen, W. J. Ind. Microbiol. 1995, 15, 391-396.

(21)

Lee, D.; Cohen, R. E.; Rubner, M. F. Langmuir 2005, 21, 9651-9659.

(22)

Fu, J.; Ji, J.; Yuan, W.; Shen, J. Biomaterials 2005, 26, 6684-6692.

(23)

Rabea, E. I.; Badawy, M. E. T.; Stevens, C. V.; Smagghe, G.; Steurbaut, W.
Biomacromolecules 2003, 4, 1457-1465.

(24)

Cen, L.; Neoh, K. G.; Kang, E. T. Langmuir 2003, 19, 10295-10303.

(25)

Nakagawa, Y.; Hayashi, H.; Tawaratani, T.; Kourai, H.; Horie, T.; Shibasaki, I. Appl.
Environ. Microbiol. 1984, 47, 513-518.

(26)

Tiller, J. C.; Liao, C.-J.; Lewis, K.; Klibanov, A. M. Proc. Natl. Acad. Sci. U. S. A. 2001,
98, 5981-5985.

(27)

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

(28)

Hollman, A. M.; Bhattacharyya, D. Langmuir 2004, 20, 5418-5424.

(29)

Krasemann, L.; Tieke, B. Mat. Sci. & Eng. C 1999, 8-9, 513-518.

(30)

Malaisamy, R.; Bruening, M. L. Langmuir 2005, 21, 10587-10592.

(31)

Smuleac, V.; Butterfield, D. A.; Bhattacharyya, D. Langmuir 2006, 22, 10118-10124.

(32)

Adusumilli, M.; Bruening, M. L. Langmuir 2009, 25, 7478-7485.

(33)

Blackburn, R. S.; Harvey, A.; Kettle, L. L.; Payne, J. D.; Russell, S. J. Langmuir 2006,
22, 5636-5644.

(34)

Lim, N.; Goh, D.; Bunce, C.; Xing, W.; Fraenkel, G.; Poole, T. R. G.; Ficker, L. Amer. J.
Ophthal. 2008, 145, 130-135.

(35)

Kim, B. R.; Anderson, J. E.; Mueller, S. A.; Gaines, W. A.; Kendall, A. M. Water Res.
2002, 36, 4433-4444.

(36)

Kawabata, A.; Taylor, J. A. Dyes Pigm. 2006, 68, 197-204.

(37)

Gilbert, P.; Moore, L. E. J. Appl. Microbiol. 2005, 99, 703.

(38)

Broxton, P.; Woodcock, P. M.; Gilbert, P. J. Appl. Bacteriol. 1983, 54, 345.

(39)

Ikeda, T.; Ledwith, A.; Bamford, C. H.; Hann, R. A. Biochim. Biophys. Acta 1984, 769,
57.

(40)

Wei, D. F.; Ma, Q. X.; Guan, Y.; Hu, F. Z.; Zheng, A. N.; Zhang, X.; Teng, Z.; Jiang, H.
Mater. Sci. Eng. C-Mater. Biol. Appl. 2009, 29, 1776-1780.

(41)

Qian, L. Y.; Guan, Y.; He, B. H.; Xiao, H. N. Polymer 2008, 49, 2471-2475.

126

(42)

Bazaron, L. U.; Stel'makh, S. A. Russ. J. App. Chem. 2008, 81, 2021-2025.

(43)

Xiong, Y.; Liu, Y. Appl. Microbiol. Biotechnol. 2010, 86, 825-837.

(44)

Ouyang, L.; Malaisamy, R.; Bruening, M. L. J. Membr. Sci. 2008, 310, 76-84.

(45)

East, G. C.; McIntyre, J. E.; Shao, J. Polymer 1997, 38, 3973-3984.

(46)

Lehtinen, J.; Nuutila, J.; Lilius, E.-M. Cytometry, Part A 2004, 60A, 165-172.

(47)

Lin-Vien, D.; Colthup, N. B.; Fateley, W. G.; Grasselli, J. G. The Handbook of Infrared
and Raman Characteristic Frequencies of Organic Molecules. Academic Press: San
Diego, 1991.

(48)

Barker, J.; Kilner, M. Coord. Chem. Rev. 1994, 133, 219-300.

(49)

Bruening, M. L.; Dotzauer, D. M.; Jain, P.; Ouyang, L.; Baker, G. L. Langmuir 2008, 24,
7663-7673.

(50)

Vakuliuk, P.; Burban, A.; Konovalova, V.; Bryk, M.; Vortman, M.; Klymenko, N.;
Shevchenko, V. Desalination 2009, 235, 160-169.

(51)

The Dow Chemical Company. Filmtec Membranes.
http://www.dow.com/PublishedLiterature/dh_006b/0901b8038006bfea.pdf?filepath=liqui
dseps/pdfs/noreg/609-00346.pdf&fromPage=GetDoc (accessed Sep. 2, 2010).

(52)

Miller, M. D.; Bruening, M. L. Chem. Mater. 2005, 17, 5375-5381.

127