MEMBRANE COATINGS FOR CATALYTIC WET-AIR OXIDATION, MEMBRANE DISTILLATION, AND
OIL-WATER SEPARATIONS
By
Seth R. Hogg

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

Chemistry – Doctor of Philosophy
2013

ABSTRACT
MEMBRANE COATINGS FOR CATALYTIC WET-AIR OXIDATION, MEMBRANE DISTILLATION, AND
OIL-WATER SEPARATIONS

By
Seth R. Hogg
Thin polymeric films often impart attractive properties to a substrate without
substantially altering the performance of the bulk material. This dissertation demonstrates
three potential applications of polymeric films to enhance the functionality of membranes for
water production and wastewater remediation, which are becoming increasingly important for
combatting water scarcity.
Layer-by-layer adsorption of polyelectrolytes and citrate-stabilized nanoparticles creates
a catalytically active film for wet-air oxidation of pollutants in water. Incorporation of these
catalytic films in membranes allows for improved oxygen availability at the catalyst surface in a
three phase reaction. In this work, the use of tubular catalytic membranes with small inner
diameters (< 1 mm) reduces diffusion limitations relative to similar systems based on larger
ceramic membranes. The reduction in the diffusion distance required for aqueous pollutants to
reach the catalyst surface leads to a 5- to 10-fold increase in the single-pass pollutant oxidation
compared to previous work with ceramic membranes.
Deposition

of

thin,

rough

polypropylene

(PP)

coatings

rapidly

produces

superhydrophobic surfaces. Remarkably, addition of such a PP coating to a poly(vinylidene

difluoride) (PVDF) membrane increases the advancing water contact angle from 124ᵒ to 155ᵒ.
These superhydrophobic membranes are well suited for membrane distillation (MD), where a
nonwetted membrane allows water vapor to transfer from a warm feed solution to a chilled
permeate phase. During MD with highly fouling humic acid, membranes coated with porous PP
show three-fold higher salt rejections than unmodified PVDF membranes. Moreover, the PPcoated membranes resist pore wetting for at least 20 hours during MD of a solution containing
dairy whey. In contrast, bare PVDF membranes show pore wetting within 3.5 h of MD with
dairy why solutions.
Finally, adsorption of charged polyelectrolyte multilayers (PEMs) on membranes
increases their hydrophilicity. This increase in hydrophilicity decreases fouling when filtering
aqueous solutions with hydrophobic foulants, such as oil and grease.

During cross-flow

microfiltration of a 0.2% hexadecane-in-water emulsion, flux declines 80% after two hours for a
PEM-modified membrane compared to a >90% flux decrease for an untreated membrane.
Fouling resistance should be especially high if the foulant and the polyelectrolyte coating have
the same charge. Exploiting electrostatic repulsions, the addition of anionic surfactants to oil
emulsions causes a nearly two-fold increase in oil droplet rejection by anionically modified
membranes.

Copyright by
SETH R. HOGG
2013

To Shana, my teammate in life and love

v

ACKNOWLEDGEMENTS

I want to express my sincerest gratitude to the many people who assisted me in
completing this degree. I appreciate the countless hours, plentiful advice, and encouragement
to think on my own provided to me by my advisor, Dr. Merlin Bruening. Dr. Volodymr Tarabara
was also instrumental in this project assisting with membrane knowledge and engineering
advice while serving on my committee along with Dr. Gary Blanchard and Dr. Tom Hamann. I
also wish to thank Dr. Baokang Bi for training and assistance with use of the SEM-EDS system.
Without the kind encouragement of Dr. Lynn Melton I would never have known the thrill which
was graduate school.
Likewise, I wish to thank Dr. David Dotzauer, Dr. Lu Ouyang, Dr. Sebastian Grajales, and
Jason Armstrong for their guidance and friendship as well as the rest of the members of the
Bruening Research Group, both past and present, for their suggestions and encouragement.
Finally, I would be remiss if I failed to thank the people who assisted me most in my
studies, my family. I thank my parents for raising me in a nurturing home where I was
encouraged to explore my interests and pursue my dreams as well as for their continuing
support and encouragement. I also am forever grateful to Shana Santos for her help on many
late nights, constant encouragement, and her open mind to always listen to my ideas, no
matter how farfetched. Without her strength and perseverance, this project would not be
feasible.

vi

TABLE OF CONTENTS

LIST OF TABLES

x

LIST OF FIGURES

xi

LIST OF ABBREVIATIONS

xvi

Chapter 1: Introduction
1.1 Water concerns
1.2 Membrane applications for water production and treatment
1.2.1 Reverse osmosis
1.2.2 Catalytic wet air oxidation
1.2.3 Membrane distillation
1.2.4 Filtration of oil-water emulsions
1.3 Catalytic nanoparticles
1.4 Functional nanoparticles on porous membranes
1.5 Layer-by-layer assembly of polyelectrolyte multilayer films
1.6 Superhydrophobic coatings
1.7 Scope of this work
REFERENCES

1
1
3
4
8
11
15
18
19
21
23
26
27

Chapter 2: Catalytic wet air oxidation of formic acid using modified polysulfone hollow fibers
as membrane contactors
38
2.1 Introduction
38
2.2 Experimental methods
41
2.2.1 Materials
41
2.2.2 Synthesis of citrate-stabilized Pt nanoparticles
41
2.2.3 Modification of hollow fiber membranes
42
2.2.4 Membrane characterization
43
2.2.5 Catalytic reactions
44
2.2.6 Modeling
44
2.3 Results and discussion
47
2.3.1 Membrane modification with Pt nanoparticles
47
2.3.2 Catalytic performance of PS hollow fiber modules
50
2.3.2.1 CWAO of formic acid using a membrane contactor
50
2.3.2.2 Fiber longevity
52

vii

2.3.2.3 Dependence of oxidation rates on formic acid concentration and O2
overpressure
2.3.2.4 Effect of Pt loading on conversion
2.3.3 Calculated formic acid concentration profiles
2.3.4 Comparison to previous work
2.4 Conclusions
REFERENCES

53
56
58
64
67
68

Chapter 3: Antifouling performance of a porous polypropylene coating in membrane
distillation
3.1 Introduction
3.2 Experimental
3.2.1 Materials
3.2.2 Deposition of crystalline polypropylene films
3.2.3 Membrane characterization
3.2.4 DCMD
3.3 Results and discussion
3.3.1 Membrane modification with polypropylene films
3.3.2 Humic acid DCMD
3.3.2.1 Bare PVDF membranes
3.3.2.2 Membranes with thin PP coatings
3.3.3 Whey DCMD
3.3.3.1 Bare PVDF membranes
3.3.3.2 Membranes with thin PP coatings
3.4 Conclusions
REFERENCES

74
74
77
77
78
78
79
82
82
84
84
87
89
89
91
93
94

Chapter 4: Preliminary investigation of anionic polyelectrolyte multilayer films as coatings
that resist fouling by negatively charged oil drops in oil-in-water emulsions
98
4.1 Introduction
98
4.2 Experimental
103
4.2.1 Materials
103
4.2.2 Modification of nylon membranes
103
4.2.3 Membrane characterization
104
4.2.4 Emulsion preparation
104
4.2.5 Electrostatic oil rejection from large pores
105
4.2.6 Microfiltration flux decline assessment
108
4.3 Results and discussion
109
4.3.1 Membrane modification with PSS and PAH films
109
4.3.2 Formation of oil emulsions using sonication
111
4.3.3 Oil rejection by membranes with large pores
115
4.3.4 Flux decline during microfiltration of oil emulsions
120
4.4 Conclusions
124

viii

REFERENCES

126

Chapter 5: Conclusions and future work
5.1 Conclusions
5.2 Future work
5.2.1 Polypropylene hollow fibers for CWAO
5.2.2 Antifouling MD membranes for the food and beverage industry
5.2.3 Polishing of oil-in-water emulsions using membrane filtration
5.3 Summary
REFERENCES

129
129
132
132
135
136
137
138

ix

LIST OF TABLES

Table 1.1: Properties and applications of commercially available membranes.

4

Table 2.1: Pt loading per length of fiber after LBL modification of a single-fiber module and a
five-fiber module with PSS/[PEI/Pt]2 films. Loadings were determined by dissolution of the Pt in
aqua regia and analysis by ICP-OES. The segments were approximately 3.75 cm long for the
single fiber and 5 cm long for the 5-fiber module.
50
Table 4.1: GC analysis parameters.

107

x

LIST OF FIGURES
Figure 1.1: Membrane feed flow modes in dead-end (A) and cross-flow (B) filtration.

4

Figure 1.2: Schematic drawing of membrane pores in an interfacial contactor that creates a gasliquid interface for a reaction occurring between aqueous component A and gaseous
component B to produce product C.
10
Figure 1.3: Common MD flow configurations showing A) direct contact MD, B) sweeping gas
MD, C) air gap MD, and D) vacuum MD.
12
Figure 1.4: LbL deposition via alternating adsorption of polycations and polyanions.

22

Figure 1.5: Rough surface wetting behavior in A) Wenzel and B) Cassie-Baxter models.

25

Figure 2.1: Schematic drawing of a cylindrical membrane contactor for CWAO with liquid feed
passing through the lumen and gas supplied to the shell side.
39
Figure 2.2: Apparatus for fiber modification with polyelectrolyte multilayers. Deposition
solutions enter the fiber lumen before exiting under cross-flow. A valve located at the lumen
exit creates backpressure to force most solution through the high surface area pores to the
shell. Rinsing after depositions proceeds through passage of water from the shell side through
the lumen in a dead-end setup. For interpretation of the references to color in this and all other
figures, the reader is referred to the electronic version of this dissertation.
43
Figure 2.3: Modeled fiber system dimensions and boundary conditions for inlet and outlet
formic acid concentrations showing only half of the lumen and a single wall.
47
Figure 2.4: Representative SEM images of the lumen (A,B) and shell (C,D) surfaces of PS hollow
fibers before (A, C) and after (B,D) adsorption of a PSS/[PEI/Pt nanoparticle] 2 film.

49

Figure 2.5: Ratio of the outlet to inlet concentrations of formic acid during CWAO in three
different fiber modules coated with PSS/[PEI/Pt]2 films. Triangles represent an O2-saturated
solution as the lone source of O2 using a single Pt-coated PS fiber with N2 (0.69 bar
overpressure) on the shell side. Diamonds and squares represent a single and 5-fiber contactor
reactor, respectively, where an O2 overpressure (0.69 bar) was applied on the shell side at 0
min and subsequently replaced with a N2 overpressure (0.69 bar) at 360 min. The solution flow
rate was 0.1 mL/min in all cases.
52
Figure 2.6: Ratio of the outlet to inlet concentrations of formic acid during CWAO for 10 trials of
8 hours using a single PS fiber module modified with a PSS/[PEI/Pt] 2 film. An O2 overpressure of
0.69 bar was applied at 0 min on the shell side and replaced with N2 at 360 min.

xi

53

Figure 2.7: Concentration of formic acid oxidized (inlet - outlet concentration) in CWAO with a
single-fiber module and an O2 overpressure of 0.69 bar on the shell side of the membrane. The
line represents a linear fit (forced through the origin, slope of 0.55) to the first three data
points. Data are an average of two experiments with two modified single fibers.
54
Figure 2.8: Concentration of formic acid oxidized during CWAO with single fiber modules at
varying overpressures. Membranes were modified with PSS/[PEI/Pt]2 films and the feed
solution contained 50 mM formic acid. Data are an average of three replicate experiments on
each of two different fibers.
56
Figure 2.9: Concentration of formic acid oxidized (inlet - outlet concentration) in CWAO with a
single-fiber module and various formic acid concentration in the inlet solutions. Fibers 1 and 3
were replicate fibers modified with PSS/PEI/Pt (1-Layer) films while fibers 2 and 4 were
modified with PSS[PEI/Pt]3 (3-Layer) films. The error bars in the figure represent standard
deviations with an n value of more than 40.
57
Figure 2.10: Diffusion-limited formic acid concentration profile in a single-fiber reactor as a
function of the distances from the inlet (z,Y-axis) and the center of the fiber (r, X-axis). The fiber
inner radius was 375 μm, and reaction was assumed to occur instantaneously at the fiber wall,
where r=375 μm. The results predict 96% oxidation of the 50 mM formic acid by the time the
solution reaches the outlet. The flow rate was 0.1 mL/min.
59
Figure 2.11: Calculated formic acid concentration profile during oxidation in a single-fiber
reactor. The plot shows the concentration by color at any point along the length of the reactor
(z, Y-axis) and at any radial distance between the center of the fiber to the outer wall (r, X-axis).
The results predict ~52% oxidation of the formic acid by the time the solution reaches the
-1

outlet. The assumed first-order rate constant was 0.01 s . The vertical line at r=375 µm
represents the fiber wall.
60
Figure 2.12: O2 concentration profiles in a polymeric hollow fiber membrane as a function of the
distances from the inlet (z,Y-axis) and the center of the membrane (r, X-axis). The simulation uses
-9

2

an O2 diffusion coefficient of 1.96 x 10 m /s in solution. The inner fiber radius was 375 µm, and
reaction was assumed to occur throughout the fiber wall with a first-order rate constant for formic
-1

acid oxidation of 0.01 s . We assumed that the reaction (equation 2.4) is first order with respect to
formic acid and zero order with respect to O2. The formic acid solution enters the fiber lumen
saturated with atmospheric O2 (0.25 mM) while the shell of the fiber is exposed to (A) 0.1 bar O 2
overpressure (1.42 mM in solution), (B) 0.69 bar O2 overpressure (2.18 mM in solution), and (C) 1.0
bar O2 overpressure (2.58 mM in solution). Note than in Figure A there is a small region of negative
concentration, showing that the assumption of zero-order reaction with respect to O2 is an
oversimplification.
61

xii

Figure 2.13: Simulated mixed cup conversion of formic acid for a single fiber reactor assuming
first-order rate constants for formic acid wet air oxidation from .001 to 1000 (1/s).
63

Figure 2.14: Simulated formic acid conversion in a single-fiber module with membranes having
different wall thicknesses. The inner radius of the fiber was 375 μm, and the rate constant was
-1

0.01 s .

64

Figure 2.15: Diffusion-limited formic acid concentration profile in a tubular ceramic membrane
as a function of the distances from the inlet (z,Y-axis) and the center of the membrane (r, Xaxis). The inner radius was 3.5 mm, and reaction was assumed to occur instantaneously at the
lumen wall, where r = 3.5 mm. The results predict 13% oxidation of the 100 mM formic acid by
the time the solution reaches the outlet. The flow rate was 10 mL/min. The low conversion
stems from high diffusion distance resulting from the large diameter of the ceramic
membrane.
66
Figure 3.1: Illustration showing the increase of apparent water contact angle on a rough surface
due to the presence of trapped air.
76
Figure 3.2: Cross-sectional diagram of the custom-built crossflow membrane holder. The
membrane is sandwiched between the two halves of the holder and sits on a porous stainless
steel frit that provides mechanical support. Warm feed solution enters the cell and travels
across the skin of the membrane before exiting the module as retentate returning to the feed
reservoir. Likewise, the cool permeate stream enters the other half of the module, passes along
the stainless steel frit, and exits the module.
80
Figure 3.3: Diagram of the DCMD apparatus. Both feed and permeate streams pass through
heat-exchange coils to achieve their desired temperatures prior to entry into the membrane
module. The feed retentate is recycled to the feed reservoir, while water vapor diffuses through
the membrane and condenses in the permeate stream. The analytical balance monitors
increases in the mass in the permeate reservoir.
81
Figure 3.4: SEM images of 0.45 μm PVDF membranes before (A) and after (B,C) modification
with porous PP. For the membrane in image (C), the coating was rinsed with ethanol prior to
heating in vacuum.
83
Figure 3.5: SEM images of cross-sections of 0.2 μm alumina membrane that were modified with
thick or thin PP films. Image A shows the thick PP coating, which was prepared without an
ethanol rinse, and image B shows the the PP coating whose deposition included ethanol
rinsing.
83

xiii

Figure 3.6: Change in permeate mass (lines) and conductivity (triangles) over time during DCMD
of humic acid solutions through five bare PVDF membranes. (A) For three membranes, flux
declines 31 ± 18% over 20 hours but salt rejection remains at 99.94 ±0.05%. (B) Two other
membranes show significant increases in permeate conductivity after an initially stable region.
The conductivity spike in (B) suggests entrance of the solution into some membranes pores
within about 200 min of operation. The breakthrough leads to an overall 88.9 ± 2.0% salt
rejection and negligible flux decline.
86
Figure 3.7: Change in permeate mass and conductivity over time during DCMD of humic acid
solutions through 3 PVDF membranes with thin PP coatings. The constant conductivity values
indicate 99.9% salt rejection at a flux only 45% lower than that through an untreated
membrane.
88
Figure 3.8: Change in permeate mass and conductivity over time during whey DCMD through 2
bare PVDF membranes. Both membranes show an increase in permeate conductivity within the
first 400 min, suggesting pore wetting. The reversal of mass flow likely results from osmotic
flow from permeate to feed through wetted pores.
90
Figure 3.9: Untreated PVDF membrane (feed surface) after DCMD of whey solution.

90

Figure 3.10: Change in permeate mass (lines) and conductivity (triangles) over time during
DCMD of whey solutions through 4 PVDF membranes with thin PP coatings. High rejection (>
95%) results in all four trials with an initial flux (100 minutes) only 10% lower than through an
untreated membrane.
92
Figure 3.11: PP coated PVDF membrane (feed surface) after DCMD of whey solution.
Figure 4.1: Membrane fouling without (A) and with (B) hydrophilic modifications.

92
101

Figure 4.2: Oil-membrane interactions in traditional size exclusion (A) and the proposed
electrostatic exclusion (B) systems.
102
Figure 4.3: Diagram of the cross-section of a custom built cross-flow membrane holder. The
membrane is sandwiched between the two halves of the holder atop at porous stainless steel
frit the serves as a mechanical support.
108
Figure 4.4: Scheme of the microfiltration apparatus.

109

Figure 4.5: SEM images of the top face of 5.0 μm nylon membranes before (A), and after
coating with [PAH/PSS]4 (B), [PAH/PSS]6 (C), and [PAH/PSS]8 (D) films. The images show no
obvious signs of pore blockage from film deposition.
111
Figure 4.6: Oil droplet size distributions obtained within 2 min of sonication of 10% hexadecane
in water. Emulsions were stabilized with SDS (A), CTAB (B), and Triton (C).
113
xiv

Figure 4.7: Oil droplet size distributions obtained 20 min after sonication of 10% hexadecane in
water. Emulsions were stabilized with SDS (A), CTAB (B), and Triton (C).
114
Figure 4.8: Control experiment design showing no oil rejection through large, unmodified pores
and electrostatic exclusion with membranes coated with PEMs.
116
Figure 4.9: Oil rejections in the first replicate of dead-end filtration of a 10% hexadecane-inwater emulsion through 5 µm nylon membranes modified with 0, 4, 6, and 8 PAH/PSS bilayers.
Emulsions were stabilized with either SDS (anionic droplets) or Triton (neutral droplets). Error
bars represent the standard deviation of average oil rejection in three replicate trials using the
same membrane.
117
Figure 4.10: Oil rejections in the second batch of membranes prepared for dead-end filtration of
a 10% hexadecane-in-water emulsion through 5 µm nylon membranes modified with 0, 4, 6,
and 8 PAH/PSS bilayers. Emulsions were stabilized with either SDS (anionic droplets) or Triton
(neutral droplets). Error bars represent the standard deviation of average oil rejection in three
replicate trials using the same membrane.
119

Figure 4.11: Oil rejections in the first replicate of dead-end filtration of a 0.5% hexadecane-inwater emulsion through 5 µm nylon membranes modified with 0, 4, 6, and 8 PAH/PSS bilayers.
Emulsions were stabilized with either SDS (anionic droplets) or Triton (neutral droplets). Error
bars represent the standard deviation of average oil rejection in three replicate trials using the
same membrane.
120
Figure 4.12: Permeate mass as a function of time during cross-flow filtration of a 0.2 % (v/v)
hexadecane emulsion through a bare and and two different membranes modified with

[PAH/PSS]4 films. The emulsions were stabilized with 0.1 mM SDS in 0.2 mM NaCl.

122

Figure 4.13: Relative flux values over time for a bare and two trials of PEM modified membranes
during cross-flow filtration of a 0.2 % (v/v) hexadecane emulsion in a 0.1 mM SDS 0.2 mM NaCl
solution.
122

Figure 4.14: Comparison of irreversible membrane staining from uncharged, hydrophobic oilsoluble dye for bare and PEM modified membranes after oil emulsion cross-flow filtration at 80
mL/min and an applied transmembrane pressure of 2 bar.
123
Figure 5.1: Cartoon of the potential catalyst locations resulting from adsorption of [PSS/PEI/Pt]
coatings in hydrophilic PS membranes and chemically treated polypropylene membranes. 134

xv

LIST OF ABBREVIATIONS
RO

Reverse osmosis

WAO

Wet air oxidation

CWAO

Catalytic wet air oxidation

MD

Membrane distillation

LbL

Layer-by-layer

PSS

Poly(styrene sulfonate)

PAA

Poly(acrylic acid)

PAH

Poly(allylamine hydrochloride)

PP

Polypropylene

PS

Polysulfone

PEI

Poly(ethyleneimine)

SEM

Scanning electron microscope

ICP-OES

Inductively coupled plasma optical emission spectroscopy

RSD

Relative standard deviation

DCMD

Direct contact membrane distillation

PVDF

Poly(vinylidine difluoride)

MEK

Methylethylketone

PEM

Polyelectrolyte multilayer

CTAB

Cetrimonium bromide

SDS

Sodium dodecyl sulfate

Triton

Triton X-100

DCM

Dichloromethane

GC

Gas chromatography

xvi

CHAPTER 1: Introduction
1.1 Water concerns
Few resources on earth play as important a role to the sustenance of life as water. No
known species remain viable with prolonged isolation from water.

1,2

Despite the major

scientific and technological advancements of the past two centuries, 780 million individuals still
3

lack access to clean water. This value represents over 10% of the world’s population relying on
unimproved sources of water, and areas within developing countries and regions of high
poverty disproportionately suffer from water scarcity issues. The absence of reliably clean
water results in 3.4 million deaths each year with more than 99% occurring in developing
4

nations.

Recognizing the global importance of water issues, the United Nations assembled in
September 2000 to adopt the Millennium Declaration committing participating nations to a
new global partnership aimed at reducing extreme poverty and improving world health. As part
of this declaration, Goal 7 focuses on improving worldwide environmental sustainability. By
2015, the United Nations desires to halve the proportion of the population without sustainable
5

access to safe drinking water and basic sanitation. Despite achieving the targeted increase in
safe drinking water availability five years ahead of schedule in 2010, current estimates project
an inability to reach the desired sanitation goals, and millions still lack drinking water primarily
3

in sub-Saharan Africa and Oceana. Thus, a great need remains for advances in wastewater
sanitation along with improved clean water access.

1

While the United States and other highly developed nations generally do not suffer from
widespread sanitation or water safety issues, increasing worldwide water demand is creating
higher incidences of water scarcity. Such scarcity occurs when the local total demand for water
exceeds either the quantity available or the achievable production rate. Agriculture accounts
6

for 90% of global freshwater use and plays a large role in economies around the world. The
United States produces the most corn in the world while also growing soybeans and wheat at
levels exceeding 50 and 10% of world supply, respectively. Altogether, agricultural exports from
the United States contribute greater than 290 billion dollars to the economy while providing
7

food products to the world. Clearly, reduction of water stress through improved water
production or agriculturally appropriate treatment methods requires further study.
This dissertation focuses on three different membrane-based water treatment options
to assess growing water contamination concerns. Specifically the use of catalytic membranes
provides an attractive solution to the treatment of industrial waste water, whereas the use of
thin, polymeric films supported on membranes should reduce fouling in both membrane
distillation and oil filtration. The remainder of this chapter discusses recent developments in
membrane applications for water production and treatment. First, I review four common
membrane water treatment procedures related to the work presented in this dissertation.
Next, I give relevant information on catalytic nanoparticle incorporation into membranes along
with a discussion on polyelectrolyte and thin polymer film deposition procedures. Finally, the
introduction concludes with a brief overview of the dissertation research.

2

1.2 Membrane applications for water production and treatment
Membrane-based processes are probably the most promising paradigm for water
treatment and production. Membranes allow transport of specific components in a mixture,
8

with selectivity based on size, charge, and/or chemical solubility. Feed solutions approach
membranes in either a dead-end or cross-flow mode (Figure 1.1). In dead-end filtration, the
feed solution reaches the membrane surface, and components either pass through the
membrane into a permeate stream or remain in the feed solution and give rise to increased
concentrations. To help avoid the accumulation of rejected feed components at the membrane
surface, in cross-flow filtration the feed stream continuously passes tangentially along the
membrane surface. Components of the mixture still pass through the membrane into a
permeate stream, but the continual replenishment of feed solution at the membrane surface
limits the accumulation of rejected components. The availability of membranes with a wide
range of pore sizes allows for tunable rejection and product flow rates suitable for many
applications (Table 1.1). Industrial applications typically utilize membranes due to their low
material costs, low energy requirements, absence of additional adsorbents or solvents, and
easy scale up ability for increased production capacity.

3

Figure 1.1: Membrane feed flow modes in dead-end (A) and cross-flow (B) filtration.

Table 1.1: Properties and applications of commercially available membranes. Table adapted
8

from Mulder.

Membrane
Type

Pore Size
(nm)

Pressure Range
(bar)

Flux Range

Microfiltration

100 - 5000

0.1 - 2

(L/m h bar)
>50

Ultrafiltration

10 - 100

1-5

10 - 50

Nanofiltration

1 - 10

5 - 20

1.4 - 12

Reverse Osmosis

0.1 - 1

10 - 100

0.05 - 1.4

2

Filtration Size Equivalent
Bacteria, yeast cells
Colloidal silica, virus,
proteins
Dye, multivalent ions,
antibiotics
Aqueous salts

1.2.1 Reverse osmosis
Desalinated sea water is the largest sustainable source of water for addressing the
9

growing needs from agriculture, human consumption, and industry. Of the wide range of
available technologies, reverse osmosis (RO) is the most economical process for desalination of

4

10,11

seawater to potable water and accounts for 44% of the world’s desalination capacity.

Additionally, the Desalination Market 2010 report from the Global Water Intelligence
organization predicts exponential growth of RO over the next four years, whereas the
utilization of traditional thermal desalination will likely remain stagnant.

12

This projected

exponential growth assumes that fossil fuel costs will limit the feasibility of thermal processes,
and membrane separations will continue to improve.
2

RO provides fresh water fluxes exceeding 40 L/m h with more than 99% removal of salt
12

but requires feed pressures above 5.5 MPa for treating seawater.

Although the high pressure

requirements contribute substantially to the cost of RO desalination, water production prices
3

3

fell from $1-2.00/m of seawater in 1998 to $0.50-0.70/m in 2005, slightly below the price of
13,14

the most common thermal processes.

In RO operation, pressure applied to the salty feed solution drives water through the
semi-permeable polyamide film on the membrane surface and into the permeate. The applied
pressure must be greater than the osmotic pressure of the seawater feed solution to reverse
the thermodynamically favored flow of water from low to high solute concentration. The Morse
equation provides an estimate for the osmotic pressure,

, of ideal solutions (equation 1.1).
(1.1)

The osmotic pressure depends on the sum of the molarity, M, of all dissolved solutes, the
dimensionless van’t Hoff factor, i, which accounts for solute dissociation, the universal gas
5

15

constant R, and the temperature of the solution in Kelvin, T.

As solute concentrations

increase in contaminated water, osmotic pressure also increases necessitating higher feed
pressure to continue permeate flow. The osmotic pressure dependence on solute
concentration traditionally limits single-stage RO processes to only 35-40% water recovery,
although new advances incorporating a second stage high pressure membrane system allow for
up to 60% water recovery.

16

While RO provides a promising method to efficiently produce fresh water, the technique
still suffers from membrane fouling and instability to chlorine, as well as insufficient boron
rejection.

12

Fouling affects the performance of all membrane processes, and biofouling in

particular remains the largest limitation of seawater RO desalination.

17

Microorganism

attachment to RO membrane surfaces leads to growth of biofilms containing extra-cellular
polymers, and these films reduce membrane water permeability and increase concentration
polarization.

18,19

Additionally, biological microorganisms release acidic byproducts that may

degrade membrane polymers and reduce RO operational lifetimes.

18,20

21

Currently, there are three main procedures to reduce biofouling.

First, biocides, such

as ozone, chlorine, or ultraviolet light can reduce biological activity in the feed solution. Second,
pretreatment of RO feed solutions using microfiltration membranes removes common
biological contaminants before they reach the RO membranes. Finally, work continues on

6

developing RO membranes and coatings with biofouling resistant surfaces to decrease the need
for feed pretreatment.

22-24

The polyamide film responsible for the high salt rejection of RO membranes suffers from
chlorine addition at the amide linkages, limiting chlorine’s application in water pretreatment.
Chlorine replaces the hydrogen on the amide nitrogen to form n-chlorinated amides that
25

undergo an irreversible Orton rearrangement.

Addition of chlorine to the polyamide films
26

alters their morphology from a crystalline to an amorphous state to decrease salt rejection.

The presence of chlorine under high pressures also promotes Orton rearrangement and causes
maximum polyamide restructuring. Therefore membrane cleaning should occur under passive
conditions to minimize damage.

27

Boric acid and borates find use in many industrial products including glass, soaps, flame
retardants, antiseptics, cosmetics, pharmaceuticals, pesticides, and fertilizers. Surface runoff
and industrial effluents may increase groundwater concentrations of boron, and some borates
and borosilicates naturally leach from rocks and soils into water reservoirs. However, the
majority of the Earth’s boron accumulates in oceans, which contain on average 4.5 mg of boron
28

per L.

Groundwater boron levels vary from 0.3 to >100 mg/L depending on geographic

location. The highest levels of boron occur in regions with highly mineralized, carbonated
groundwater such as Turkey and deserts in the United States southwest region. Current World
Health Organization standards call for a maximum boron concentration of 2.4 mg/L in drinking
water to avoid undesired health effects.

29

As this targeted safe maximum concentration is less
7

than the average boron concentration in ocean water, the production of safe drinking water
requires efficient boron removal. Unfortunately, uncharged boric acid easily permeates through
RO membranes due to its small size and diffusion through the polyamide film using hydrogen
30-32

bridges within membrane active groups.

Currently, the best method for boron rejection

relies on feed solutions with pH values >11 to dissociate boric acids into polynuclear ions that
do not pass through RO membranes.

33,34

However, high pH solutions also promote salt scaling
35

on membrane surfaces and increased hydraulic resistance.

To achieve necessary reductions

in boron concentrations, blending with low boron water or multiple passes through RO
elements must occur to overcome the limited rejection possible with current RO technologies.

1.2.2 Catalytic wet air oxidation
Many industrial processes produce concentrated waste streams not amenable to RO
and thus require other treatment options. Advanced oxidation procedures can potentially
provide an effective solution to the removal of organic solutes from wastewater by producing
primarily carbon dioxide and water as the final products. Among advanced oxidation
procedures, wet air oxidation (WAO) is attractive due to its simplicity and low cost, which
results from utilizing air as an oxidant instead of chemicals like peroxide, ozone, or Fenton’s
reagent. WAO is best suited for treatment of waste streams too concentrated for biological
treatment and too dilute to justify the expense of incineration. In a process known as catalytic
wet air oxidation (CWAO), most industrial WAO plants currently utilize fixed-bed catalysts to

8

treat water from chemical, petrochemical, municipal water-treatment, and pharmaceutical
36

industries.

At temperatures and pressures from 180 ◦C and 20 bar to 315 ◦C and 150 bar, WAO
uses solution residence times between 15-120 minutes to oxidize 75-90% of organic
contaminants while producing primarily carbon dioxide and water without generation of
36

dioxins, hydrochloric acid, sulfur dioxide, or other hazardous products.

However, although

WAO effectively oxidizes hazardous organic and inorganic solutes to benign products, the high
temperatures and pressures required limit widespread application. The introduction of
heterogeneous catalysts in CWAO reduced the temperatures and pressures required but
introduced new challenges to the process. These challenges include diffusion limitations in the
transport of gaseous reactants to dispersed catalysts surfaces, the necessity for recovery of
catalysts made from precious metals, and catalyst leaching.

37

Incorporation of catalysts within interfacial membrane contactors allows separation of
the product and reactant streams on opposite sides of the membrane while facilitating a gas38-42

liquid interface within the membrane pores (Figure 1.2).

The interfacial contact within the

membrane pores reduces diffusion distances gaseous reactants to reach the catalyst’s surface,
while the incorporation of catalyst within a membrane eliminates the need for catalyst recovery
addressing two concerns of heterogeneous CWAO.

9

Figure 1.2: Schematic drawing of membrane pores in an interfacial contactor that creates a gasliquid interface for a reaction occurring between aqueous component A and gaseous
component B to produce product C.

Previous work utilizing tubular ceramic membranes loaded with platinum nanoparticle
catalysts demonstrated 3-6 times higher conversion in formic acid oxidation when compared to
a stirred tank reactor under the same conditions. The membranes also exhibited a five-fold
41

increase in catalyst specific activity compared to the stirred tank reactor.

After these

encouraging lab-scale tests that suggested potential industrial membrane CWAO applications,
pilot-scale testing using single and multi-channel ceramic membranes demonstrated toxic
waste treatment on a large scale but also showed that low reaction rates in the system limited
the economic benefits of the process. Importantly though, membrane CWAO systems
40

facilitated competitive conversion rates for industrial effluents under mild conditions.

While

most studies focus on formic acid as a simple model compound, some research assessed the
oxidation of phenol, another common industrial water pollutant. Phenol oxidation occurred
with the formation of no intermediates when using membrane CWAO over cerium oxide
10

38

catalysts.

One other study demonstrated oxidation of phenol, oxalic acid, and acetic acid at

room temperature and less than 4 bar of pressure indicating the possibility of treating aromatic
43

and aliphatic organic contaminants.

Incorporation of catalysts into membranes most commonly occurred through anionic
impregnation or evaporation crystallization, but one study examined the performance of
membranes loaded with catalyst using layer-by-layer deposition of polyelectrolytes and
charged nanoparticles. Although the reaction rates in membranes treated with layer-by-layer
deposition were 33-50% lower than in membranes modified with traditional methods, specific
catalyst activity increased five-fold due to precise localization of platinum at the air-liquid
interface.

44

Finally, one paper examined the use of polymeric membranes in place of ceramic

membranes and found that in mild conditions, high single pass conversions occurred because of
small fiber diameters, despite low catalyst specific activity due to delocalization of catalyst from
45

the air-liquid interface.

While CWAO treats contaminated industrial waste streams, the

resulting water typically is not suitable for human consumption or agricultural use due to the
potential presence of hazardous inorganic contaminants. CWAO therefore requires coupling
with other treatment options to finish water purification to acceptable levels for consumption.

1.2.3 Membrane distillation
Distillation produces highly pure water from sources contaminated with non-volatile
solutes. Unfortunately, most distillation processes require a high-temperature heat source to
11

drive evaporation of feed water into the product. However, over the last two decades
membrane distillation (MD) emerged as a growing technique for desalination or water
treatment at milder temperatures. MD utilizes a porous, non-wetted membrane to separate a
warm feed solution from a cool permeate stream. The membrane must avoid condensation of
permeate vapor within membrane pores, and at least one side of the membrane must directly
46

contact the process liquids in one of the four common orientations in Figure 1.3.

Figure 1.3: Common MD flow configurations showing A) direct contact MD, B) sweeping gas
MD, C) air gap MD, and D) vacuum MD.

The original goal in developing MD was to reduce the space requirements of traditional
distillation by replacing large distillation columns with smaller membrane modules. Unlike
conventional distillation, MD only requires aqueous feed temperatures between 60-90 ◦C,

12

which may reduce energy requirements and allow for incorporation of low grade, waste, or
alternative energy sources of heat.

47

Coupling of solar energy with MD water production might

prove cost competitive with RO in rural areas.

48,49

The low temperatures in MD also allow

dehydration of food and pharmaceutical products while avoiding thermal damage to
temperature-sensitive components, thereby preserving flavor, aroma, or active ingredient
functionality.

34,50-52

Direct contact MD remains the most common configuration due to its simplicity, with
the feed solution on one side of the non-wetted membrane and the permeate solution on the
other side (Figure 1.3A). Water vapor from the feed stream diffuses through the hydrophobic
membrane pores due to the water partial pressure difference that arises from the temperature
gradient across the membrane. The water vapor condenses upon reaching the chilled aqueous
stream passing along the permeate side of the membrane. Heat loss through the membrane
into the permeate stream reduces the temperature gradient between the feed and permeate
streams, however, lowering the driving force and reducing flux. Current studies of direct
contact MD focus on applications in food processing or acid manufacturing.

53-58

Replacing the aqueous permeate stream with an inert sweep gas (sweeping gas MD,
Figure 1.3B) leads to increased mass transport through the membrane pores by maintaining a
low water vapor pressure on the permeate side of the membrane. The sweep gas also reduces
the heat loss through the membrane, increasing efficiency and flux.

59

However, sweeping gas

MD requires condensation of the permeate water vapor downstream from the membrane. The
13

condenser must be large to accommodate the high volume of sweep gas containing relatively
60

small amounts of water vapor permeate.

Despite the proposed increases in flux predicted for

sweeping gas MD, no industries currently utilize the technique, and less than five percent of
published MD papers discuss it.

61

To avoid the complications associated with condensing permeate vapors downstream
from the membrane in sweeping gas MD, air gap MD (Figure 1.3C) employs a condensing
surface separated by a small, stationary air gap from the permeate side of the membrane.
Water vapor diffuses through the membrane, across the air gap, and subsequently condenses
on the chilled surface for collection. This configuration reduces heat loss through the
membrane compared to direct contact MD, but mass transfer resistance increases due to the
stagnant vapor barrier. Thus air gap MD has low fluxes compared to other MD methods.

47

Nevertheless, air gap MD allows easier recovery of trace volatile contaminants from aqueous
62

streams than direct contact MD.

Other uses for air gap MD may include produced water

treatment and concentration of non-volatile solutes.

63-65

Vacuum MD (Figure 1.3D) utilizes reduced pressure at the permeate side of the
membrane to increase vapor flux. This is similar to pervaporation, but the membrane plays no
active role in the separation other than supporting the liquid-vapor interface. Energy and
capital expenses for maintaining the vacuum increase the cost of vacuum MD compared to the
other methods but once again, heat loss through the membrane is negligible. Similar to
sweeping gas MD, the permeate vapors must also be condensed downstream from the
14

membrane. Of all the available MD techniques, vacuum MD most effectively removes volatile
66

organic components from aqueous streams.

Current studies of vacuum MD focus on

desalination or removal of organic contaminants such as benzene, aroma compounds, and
polyphenols from water.

67-70

1.2.4 Filtration of oil-water emulsions
Over the last 15 years, hydraulic fracturing, colloquially referred to as “fracking”, has
emerged as an effective method to extract oil and natural gas from non-porous rock
formations. After drilling an appropriate well, injection of high pressure fracturing fluids into
rock formations creates cracks and channels for oil and natural gas release and capture.
Subsequent addition of solutions containing propping agents, including sand or ceramic beads,
fixes the fractures in place to allow continued oil and gas release upon removal of the pumping
pressure. Traditionally, fracturing fluids contain up to 99% water with various chemical
additives, and fracturing coalbed and shale formations requires up to 350,000 gallons or
5,000,000 gallons, respectively. As the water for hydraulic fracturing primarily comes from local
surface or groundwater, the process results in water stress in areas with multiple wells.
Additionally, wells typically recover only 15-80% of injected volumes, and the recovered water,
known as produced water, contains a dangerous mixture of toxic components and needs
further treatment.

71

15

The composition of the dissolved and dispersed components in produced water varies
with geography and even the age of a particular well. Dissolved organic compounds in
produced water may include formic and propionic acid, benzene, toluene, ethylbenzene,
xylene, and phenols. The latter 4 compounds are relatively toxic. In addition to dissolved
compounds, insoluble oil dispersed in the solution primarily contains aliphatic hydrocarbons
along with dangerous polyaromatic hydrocarbons and heavy alkyl phenols.

72

Dissolved

minerals further complicate produced water. Cations and anions dissolved from the formation
+

+

2+

2+

2+

2+

2+

-

2-

2-

-

include Na , K , Ca , Mg , Ba , Sr , Fe , Cl , SO4 , CO3 , and HCO3 . These ions all
increase salinity, and some of them enhance the scaling potential of the solution. Incredibly,
73

salt concentrations in produced water can exceed 300 g/L.

Other minerals present even

greater danger including heavy metals such as cadmium, chromium, copper, lead, mercury,
nickel, silver, and zinc, as well as naturally occurring radioactive materials consisting primarily of
radium deposits in scale formation.

74,75

Finally, further complicating treatment procedures,

produced water also contains production chemicals such as scale inhibitors, biocides, and
emulsion breakers.

76

In 2009, the average oil and natural gas well in the United States created seven barrels
of produced water for every barrel of oil recovered.

77

Each barrel creates a possible

environmental hazard due to proximity to natural water sources as well as an economic burden
for proper treatment. Recent challenges in treating the Deepwater Horizon oil spill in 2010 only
highlight the need for further developments of systems that remove oil from contaminated
16

waters rapidly without significant economic impact. Membrane filtration provides one of the
best methods for large-scale separation of dilute, oily wastewaters. Rapid production of highly
purified water can occur with readily deployed pre-built modules.

78

Membrane-based separations of oil and water generally rely on size exclusion of oil
droplets rather than chemical selectivity, but formation of an oil film on the membrane reduces
water flux and oil rejection. Frequent cleaning procedures must occur to either flush the oil
79

from the membrane or chemically renew its surface.

A number of recent studies explored

increasing membrane hydrophilicity to reduce oil fouling. Segregation of hydrophilic groups to
the membrane surface and graft polymerization yield extremely hydrophilic membranes that
resist oil wetting.

80-82

In an aqueous solution, these extremely hydrophilic surfaces adsorb a

layer of water that prevents contact with the high fouling oil components that plague
hydrophobic membrane processes. Additionally, the adsorbed water layer creates an entropic
barrier to oil droplet passage through membrane pores, allowing for high oil rejections even in
82

membranes with pore sizes larger than the oil droplets.

Despite the gains in oil rejection achieved through creating hydrophilic membrane
surfaces, most studies only assess oil rejection using concentrations exceeding 1000 mg/L and
droplet diameters greater than 50 µm. Under these conditions, oil exists in an unstable state
that is simpler to remove from water than more stable droplets of less than 20 µm in
diameter.

83

Traditional oil-treatment methods including gravity separation or chemical
84,85

emulsion breaking are also inefficient for removal of oil droplets below 10 µm in size.
17

While membranes may promote coalescence of small oil droplets into larger ones that are
treatable by conventional technologies, further studies must prove this theory conclusively
through addressing rejection of small, stable oil droplets.

1.3 Catalytic nanoparticles
Incorporation of catalytic particles, as earlier discussed for CWAO, represents a
promising method for future water treatment advances. Catalysts play important roles in the
synthesis of most important industrial products and many vital biological processes. Typically,
catalysts lower the activation energy of reactions to increase their rate relative to the noncatalyzed reaction at the same temperature. Unlike the reactants, reactions do not consume
catalysts so they participate in multiple reactions.

86

Enzymatic catalysts participate in biological reactions such as phosphorylation,
oxidation, and esterification that are critical in signal transduction and cellular regulation.

87

With the need for high specificity in biological processes, enzymatic catalysts unsurprisingly
afford the highest selectivity of all catalysts. Alternatively, homogeneous small-molecule
catalysts provide tunable chemoselectivty, regioselectivity, or enantioselectivity while residing
in the same phase as the reactants. However, applications of homogeneous catalysts are
mostly limited to the synthesis of fine chemicals because of the difficulty of recovering the
expensive catalyst from the final products.

88

On the other hand, solid catalysts containing

transition metals facilitate a wide range of industrial reactions in gases and liquids. Reaction on
18

these heterogeneous catalysts takes place at solid-gas or solid-liquid interfaces. However, the
variability in the active sites present on heterogeneous catalysts results in relatively low
86

specificity compared to enzymes and homogeneous catalysts.

Despite the lack of selectivity,

heterogeneous catalysts dominate the global market and find use in many industrial synthetic
processes.
Although nanoparticles have facilitated various processes for many years, only in the
last 30-40 years did the scientific community focus on controlled design of nanoparticles for
89-92

catalysis.

Interest in imparting unique properties to nanoparticle catalysts arose from the

discovery of gold nanoparticles with high catalytic activity despite the relative inactivity of bulk
93,94

gold.

Further studies revealed that the optimal size for catalytically active nanoparticles

falls between 1 and 10 nm.

95,96

At this small diameter, the particles maintain a high surface

area to volume ratio, allowing efficient use of precious noble metal catalysts while affording
unique electronic and catalytic properties.

96-98

However, the small particle diameters also lead

to high particle surface energies that cause rapid aggregation and loss of activity unless a ligand
99-101

or support stabilizes the particles.

1.4 Functional nanoparticles on porous membranes
Incorporation of nanoparticles within membranes prevents particle aggregation and
avoids time-consuming catalyst recovery steps. Additionally, immobilization of nanoparticles in
19

membranes allows for continuous operation of the reactor unlike what is possible for
102

traditional batch or stirred-tank reactors.

Compared to other supports such as porous beads,

membranes have the potential to improve heterogeneous catalysis by reducing the distance
reactants must diffuse before adsorbing to the catalyst surface. Diffusion limitations commonly
challenge the efficiency of reactions involving heterogeneous catalysts in liquids, so catalytic
membranes may lead to gains in observed reaction rates.
Another potential advantage of catalytically active membranes is the possibility of
separating reactant from newly formed products. Continuous removal of product may
eliminate further post-reaction separations. More importantly, removal of products from the
reaction mixture shifts the equilibrium towards products to increase the possible degree of
102

conversion.

Highly porous membranes with small pore diameters contain a large surface area for
incorporation of catalysts, and high catalyst concentrations allow increased conversion or the
use of high reactant concentrations. Along with porosity and pore size, the selection of the
membrane composition for a specific application allows a balance of performance and cost.
Current membrane materials include both hydrophilic and hydrophobic polymers as well as
temperature- and solvent-resistant ceramic membranes. For reactions at temperatures below
100 ◦C in mild conditions, polymeric hollow fiber membranes are likely the best choice due to
their inexpensive production, small diameter, and high packing density. However, for more
extreme catalytic environments, ceramic membranes provide the necessary durability despite

20

increased costs. Potential applications for catalytic nanoparticles immobilized in membranes
include the reduction of nitroaromatic compounds in water, hydrogenation of organic
compounds, and the synthesis of hydrogen peroxide from methanol.

103-106

In order for

nanoparticles to function on membrane surfaces, effective incorporation through covalent or
electrostatic attachment must occur.

1.5 Layer-by-layer assembly of polyelectrolyte multilayer films
Despite the strength of covalent attachment of nanoparticles to surfaces, electrostatic
interactions also provide a suitable immobilization method without as many limitations from
chemical reactivity. The layer-by-layer (LbL) method is the preeminent choice for electrostatic
film attachment to surfaces. LbL assembly techniques employ alternating adsorption of
complementary species to form thin films one “layer” at a time. The most common LbL
technique exploits alternating electrostatic adsorption of polyanions and polycations as Figure
1.4 illustrates. Nearly any surface which allows adsorption of an initial polymer either through
107-109

hydrophobic or electrostatic interactions may support the LbL approach.

Surface

geometries previously modified include flat surfaces, membrane pores, cylindrical particles, and
carbon

nanotubes.

44,105,110-113

Adsorption

of

each

polyelectrolyte

chain

typically

overcompensates the charge of the modified surface so the final surface charge depends on the
composition of the terminating polyelectrolyte layer. In addition to typical synthetic

21

polyelectrolytes, LbL adsorption can employ other charged species such as catalytic
nanoparticles, dendrimers, enzymes, and carbon nanotubes.

114-120

Figure 1.4: LbL deposition via alternating adsorption of polycations and polyanions.

Along with the versatility in choosing constituents for LbL deposition, the properties of
the polymer films vary with deposition conditions including pH, salt concentration,
polyelectrolyte concentration, or solvent composition. While some polymers, such as
poly(styrene sulfonate) (PSS), are strong electrolytes with high charge at nearly all pH
conditions, others are weak electrolytes and pH strongly affects their deposition. Rubner
demonstrated that the thicknesses of films comprised of poly(acrylic acid) (PAA) and
poly(allylamine hydrochloride) (PAH) depends on the pH of deposition solutions. At a pH of 7
where both the carboxylic acid groups of PAA (pKa ~ 4.5) and the amine groups in PAH (pKa ~
9.5) are ionized, a single bilayer produces a compact, ~5 Å-thick film. However, at a pH of 5,
partial protonation of the carboxylic groups of PAA causes deposition of a less extended
polymer and a 14-fold increase in thickness.

121

Related effects occur upon increases in the

ionic strength of the polyelectrolyte deposition solutions because higher salt concentrations
screen the charge within the polyelectrolyte chains to create more coiled polymers and thicker
22

121,122

films.

In summary, tailoring of LbL polyelectrolyte adsorption affords control over

surface charge, swelling, and thickness. Thus, the LbL technique offers a versatile platform for
immobilizing catalytic nanoparticles in membranes and modifying membrane surface to create
low-fouling membranes that reject oil.

1.6 Superhydrophobic coatings
In addition to varying charge and chemical functionality, surface scientists also seek to
control surface wetting. For instance, water repellency plays a vital role in the operation of
common products such as self-cleaning traffic indicators, self-cleaning windows, frictional drag
reducing coatings for naval vessel hulls, and stain resistant textiles.

123,124

Similar coatings are

also attractive in the emerging field of biomedical fluid handling without adsorption of blood or
125

other biological materials.

Quantitative exploration of surface wettability typically involves determining the
contact angle a droplet of a liquid, either water or oil, makes when in contact with the surface.
The equilibrium contact angle,  results from a balance of interfacial energies that give rise to
126

the Young Equation (equation 1.2),

(1.2)

23

where γsv is the solid-vapor interfacial energy, γsl is the solid-liquid interfacial energy, and γlv
is the liquid-air interfacial energy. For a smooth surface, the maximum known water contact
angle is 120◦ on a

127

Teflon coating.

However, the incorporation of rough features on

hydrophobic surfaces allows water contact angles to exceed this value. When the water contact
angle on a surface exceeds 150◦ and does not suffer from large hysteresis (differences >10◦
between advance and receding contact angles), the surface is considered superhydrophobic.
128,129

There are two wetting modes possible with rough, hydrophobic surfaces (Figure 1.5).

In

the Wenzel model, the apparent contact angle, θ*, of a surface increases with the increased
surface area available due to the surface roughness, r (equation 1.3). In this equation, r is the
ratio of total surface area to the apparent area.

(1.3)

As the Wenzel surface becomes increasingly rough, the apparent contact angle continues to
increase.
The Cassie-Baxter model assumes that small air pockets remain trapped within surface
roughness features to reduce the liquid-solid interfacial area. Thus, this model calculates the
apparent contact angle using equation 1.4 based on the fraction of solid in contact with the
liquid, ϕS , and the fraction in contact with air (1-ϕS).

(1.4)

24

As the numbers or size of air traps within the roughness features increases, the fraction of solid
contacting the liquid decreases to give a net increase in the contact angle. The incorporation of
air in the substrate topology to reduce the surface-liquid contact within the Cassie-Baxter state
provides intriguing antifouling possibilities. In many areas, fouling components of solutions
never contact surfaces due to trapped air.

Figure 1.5: Rough surface wetting behavior in A) Wenzel and B) Cassie-Baxter models.

Current methods for creating superhydrophobic surfaces include hydrophobic polymer
recrystallization, LbL assembly with incorporation of hydrophobic nanoparticles, deposition of
fluorinated silsesquioxane films, and adsorption of lubricating omniphobic fluids within
125,130-132

roughness features.

These unique surfaces may find applications in oil-water

separation membranes, stain resistant surfaces, optically transparent self-cleaning glass, lowfouling microfluidic surfaces, and marine bio-fouling resistant coatings.

25

125,133-135

1.7 Scope of this work
This dissertation discusses the development of new membrane coatings that may
enhance water production or pollution remediation. The next three chapters each present a
specific thrust within the overlying goal of developing new water-treatment membranes.
Chapter 2 discusses the incorporation of platinum nanoparticles within polymeric hollow fibers
for low temperature CWAO of dilute formic acid waste streams. Studies of the effect of oxygen
overpressure, solution residence time, and catalyst loading on CWAO conversion rates explore
the performance of the polymeric membranes in comparison to previously utilized ceramic
membranes. Chapter 3 covers the development of superhydrophobic, porous polypropylene
(PP) membrane coatings for membrane distillation. When compared to unmodified
membranes, despite a slight decline in flux resulting from the additional mass transfer
resistance of the porous PP, these coatings decrease the rate of fouling by humic acid and other
foulants to give longer operational stability with less frequent mandatory cleaning. Chapter 4
introduces the concept and initial results of oil-water separations with membranes whose
surface charge matches the charge naturally present on small, emulsified oil droplets. The
electrostatic repulsion of droplets provided by the charged membranes leads to higher oil
rejection and presumably less membrane fouling from oil adsorption. Finally, chapter 5
discusses conclusions from the sum of the work and attempts to predict future challenges and
areas for this work to expand. Overall, this dissertation shows the promise of thin, polymeric
films for enhancing fresh-water production capabilities.

26

REFERENCES

27

REFERENCES

(1)

Rothschild, L. J.; Mancinelli, R. L. Nature 2001, 409, 1092.

(2)

Alpert, P. Integrative & Comparative Biology 2005, 45, 683.

(3)
UNICEF; WHO Progress on Drinking Water and Sanitation: 2012 Update; WHO
Press: New York, NY, 2012.
(4)
Prüss-Üstün, A.; Bos, R.; Gore, F.; Bartram, J. Safer Water, Better Health: Costs,
Benefits and Sustainability of Interventions to Protect and Promote Health; WHO Press:
Geneva, Switzerland, 2008.
(5)
United Nations. Millenium Development Goals: Environmental Sustainability.
http://www.un.org/millenniumgoals/environ.shtml (accessed 03/09/2013).
(6)
Food and Agriculture Organization of the United Nations, Coping with Water
Scarcity: An Action Framework for Agriculture and Food Security, Rome, Italy, 2012,
http://www.fao.org (accessed 03/09/2013).
(7)
United States Environmental Protection Agency. Major Crops Grown in the
United States. http://www.epa.gov/agriculture/ag101/cropmajor.html (accessed
03/09/2013).
(8)
Mulder, M. Basic Principles of Membrane Technology; Second ed.; Springer: New
York, NY, 1996.
(9)
Uemura, T.; Henmi, M. Thin-film Composite Membranes for Reverse Osmosis;
John Wiley & Sons, Inc.: Hoboken, NJ, 2008.
(10) Greenlee, L. F.; Lawler, D. F.; Freeman, B. D.; Marron, B.; Moulin, P. Water
Research 2009, 43, 2317.
(11) Gupta, V. K.; Ali, I.; Saleh, T. A.; Nayak, A.; Agarwal, S. RSC Advances 2012, 2,
6380.
(12)

Misdan, N.; Lau, W. J.; Ismail, A. F. Desalination 2012, 287, 228.

(13)

Dore, M. H. I. Desalination 2005, 172, 207.

28

(14)

Wilf, M.; Bartels, C. Desalination 2005, 173, 1.

(15) Amiji, M. M.; Sandmann, B. J. Applied Physical Pharmacy; McGraw-Hill
Professional, 2002.
(16) von Gottberg, A.; Pang, A.; Talavera, J. L. Optimizing Water Recovery and Energy
Consumption for Seawater RO Systems, GE Water & Process Technologies, 2005.
(17) Flemming, H. C.; Schaule, G.; Griebe, T.; Schmitt, J.; Tamachkiarowa, A.
Desalination 1997, 113, 215.
(18) Abd El Aleem, F. A.; Al-Sugair, K. A.; Alahmad, M. I. International Biodeterioration
& Biodegredation 1998, 41, 19.
(19)

Kim, D.; Jung, S.; Sohn, J.; Kim, H.; Lee, S. Desalination 2009, 238, 43.

(20)

Flemming, H. C. Experimental Thermal and Fluid Science 1997, 14, 382.

(21) Vrouwenvelder, J. S.; Van Loosdrecht, M. C. M.; Kruithof, J. C. Water Research
2011, 45, 3890.
(22) Worthley, C. H.; Constantopoulos, K. T.; Ginic-Markovic, M.; Pillar, R. J.;
Matisons, J. G.; Clarke, S. Journal of Membrane Science 2011, 385-386, 30.
(23) Matin, A.; Ozaydin-Ince, G.; Khan, Z.; Zaidia, S. M. J.; Gleason, K.; Eggenspiler, D.
Desalination and Water Treatment 2011, 34, 100.
(24)

Zhang, Z.; Wang, Z.; Wang, J.; Wang, S. Desalination 2013, 309, 187.

(25)

Glater, J.; Hong, S.-K.; Elimelech, M. Desalination 1994, 95, 325.

(26)

Avlonitis, S.; Hanbury, W. T.; Hodgkiss, T. Desalination 1992, 85, 321.

(27) Antony, A.; Fudianto, R.; Cox, S.; Leslie, G. Journal of Membrane Science 2010,
347, 159.
(28) CRC Handbook of Chemistry and Physics; 69th ed.; Weast, R. C.; Astle, M. J.;
Beyer, W. H., Eds.; CRC Press, Inc.: Boca Raton, FL, 1985.
(29) World Health Organization Press, Boron in Drinking-water, Geneva, Switzerland,
2009, http://whqlibdoc.who.int/hq/2009/WHO_HSE_WSH_09.01_2_eng.pdf (accessed
03/09/2013).

29

(30) Cengeloglu, Y.; Arslan, G.; Tor, A.; Kocak, I.; Dursun, N. Separation and
Purification Technology 2008, 64, 141.
(31) Tu, K. L.; Nghiem, L. D.; Chivas, A. R. Separation and Purification Technology
2010, 75, 87.
(32) Rodriguez Pastor, M.; Ferrándiz Ruiz, A.; Chillón, M. F.; Prats Rico, D. Desalination
2001, 140, 145.
(33)

Glueckstem, P.; Priel, M. Desalination 2003, 156, 219.

(34)

Hilal, N.; Kim, G. J.; Somerfield, C. Desalination 2011, 273, 23.

(35) Dominguez-Tagle, C.; Romero-Temero, V. J.; Delgado-Torres, A. M. Desalination
2011, 265, 43.
(36)

Luck, F. Catalysis Today 1996, 27, 195.

(37)

Levec, J.; Pintar, A. Catalysis Today 2007, 124, 172.

(38) Gutiérrez, M.; Pina, P.; Torres, M.; Cauqui, M. A.; Herguido, J. Catalysis Today
2010, 149, 326.
(39) Iojoiu, E.; Miachon, S.; Landrivon, E.; Walmsley, J.; Rader, H.; Dalmon, J. Applied
Catalysis B: Environmental 2007, 69, 196.
(40) Iojoiu, E. E.; Landrivon, E.; Raeder, H.; Torp, E. G.; Miachon, S.; Dalmon, J.-A.
Catalysis Today 2006, 118, 246.
(41) Miachon, S.; Perez, V.; Crehan, G.; Torp, E.; Ræder, H.; Bredesen, R.; Dalmon, J.
A. Catalysis Today 2003, 82, 75.
(42) Ræder, H.; Bredesen, R.; Crehan, G.; Miachon, S.; Dalmon, J.-A.; Pintar, A.; Levec,
J.; Torp, E. Separation and Purification Technology 2003, 32, 349.
(43) Alame, M.; Abusaloua, A.; Pera-Titus, M.; Guilhaume, N.; Fiaty, K.; Giroir-Fendler,
A. Catalysis Today 2010, 157, 327.
(44) Dotzauer, D. M.; Abusaloua, A.; Miachon, S.; Dalmon, J.-A.; Bruening, M. L.
Applied Catalysis, B: Environmental 2009, 91, 180.
(45) Hogg, S. R.; Muthu, S.; O'Callaghan, M.; Lahitte, J. F.; Bruening, M. L. ACS Applied
Materials & Interfaces 2012, 4, 1440.
30

(46)

Smolders, C. A.; Franken, A. C. M. Desalination 1989, 72, 249.

(47)

Lawson, K. W.; Lloyd, D. R. Journal of Membrane Science 1997, 124, 1.

(48)

Hogan, P. A.; Sudjito; Fane, A. G.; Morrison, G. L. Desalination 1991, 81, 81.

(49) Saffarini, R. B.; Summers, E. K.; Arafat, H. A.; Lienhard V, J. H. Desalination 2012,
299, 55.
(50) Sakai, K.; Ano, T. K.; Muroi, T.; Tamura, M. The Chemical Engineering Journal
1988, 38, B33.
(51) Calabro, V.; Jiao, B. L.; Drioli, E. Industrial & Engineering Chemistry Research
1004, 33, 1803.
(52)
539.

Ding, Z.; Liu, L.; Yu, J.; Ma, R.; Yang, Z. Journal of Membrane Science 2008, 310,

(53)

Laganà, F.; Barbieri, G.; Drioli, E. Journal of Membrane Science 2000, 166, 1.

(54)

Alves, V. D.; Coelhoso, I. M. Journal of Food Engineering 2006, 74, 125.

(55)

Hsu, S. T.; Cheng, K. T.; Chiou, J. S. Desalination 2002, 143, 279.

(56)

Gunko, S.; Verbych, S.; Bryk, M.; Hilal, N. Desalination 2006, 190, 117.

(57)

Khayet, M.; Mengual, J. I. Desalination 2004, 168, 373.

(58) Tomaszewska, M.; Gryta, M.; Morawski, A. W. Journal of Membrane Science
1995, 102, 113.
(59)

Charfi, K.; Khayet, M.; Safi, M. J. Desalination 2010, 259, 84.

(60)

Alkhudhiri, A.; Darwish, N.; Hilal, N. Desalination 2012, 287, 2.

(61)

Khayet, M. Advances in Colloid and Interface Science 2011, 164, 56.

(62) Garcia-Payo, M. C.; Izquierdo-Gil, M. A.; Fernández-Pineda, C. Journal of
Membrane Science 2000, 169, 61.
(63) Izquierdo-Gil, M. A.; García-Payo, M. C.; Fernández-Pineda, C. Journal of
Membrane Science 1999, 155, 291.

31

(64) Thiruvenkatachari, R.; Manickam, M.; Kwon, T. O.; Moon, I. S.; Kim, J. W.
Separation Science and Technology 2006, 41, 3187.
(65)

Alkhudhiri, A.; Darwish, N.; Hilal, N. Desalination 2013, 309, 46.

(66)

Bandini, S.; Saavedra, A.; Sarti, G. C. AIChE Journal 1997, 43, 398.

(67) Garcia-Castello, E.; Cassano, A.; Criscuoli, A.; Conidi, C.; Drioli, E. Water Research
2010, 44, 3883.
(68)

Li, B.; Sirkar, K. K. Journal of Membrane Science 2005, 257, 60.

(69)

Banat, F. A.; Simandl, J. Chemical Engineering Science 1996, 51, 1257.

(70) Hasanoglu, A.; Robolledo, F.; Plaza, A.; Torres, A.; Romero, J. Journal of Food
Engineering 2012, 111, 632.
(71) United States Environmental Protection Agency, Hydraulic Fracturing Research
Study
Fact
Sheet,
Washington,
D.C.,
2010,
http://www.epa.gov/safewater/uic/pdfs/hfresearchstudyfs.pdf (accessed 03/09/2013).
(72)

Ekins, P.; Vanner, R.; Firebrace, J. Journal of Cleaner Production 2007, 15, 1302.

(73) Jacobs, R. P. M. W.; Grant, R. O. H.; Kwant, J.; Marquenie, J. M.; Mentzer, E. In
Produced Water: Technological/Environmental Issues and Solutions; Ray, J. P.,
Engelhardt, F. R., Eds.; Plenum Publishing Corp.: New York, 1992, p 13.
(74)
176.

Hansen, B. R.; Davies, S. H. Chemical Engineering Research and Design 1994, 72,

(75)
193.

T. Gäfvert; I. Færevik; Rudjord, A. L. Radioactivity in the Environment 2006, 8,

(76) Stephenson, M. T. In Produced Water: Technological/Environmental Issues and
Solutions; Ray, J. P., Engelhardt, F. R., Eds.; Plenum Publishing Corp.: New York, 1992, p
1.
(77) Argonne National Laboratory - Environmental Science Division, Produced Water
Volumes
and Management
Practices in the
United
States, 2009,
http://www.ead.anl.gov/pub/doc/ANL_EVS__R09_produced_water_volume_report_24
37.pdf (accessed 03/09/2013).
(78) Ahmadun, F. l.-R.; Pendashteh, A.; Abdullah, L. C.; Biak, D. R. A.; Madaeni, S. S.;
Abidin, Z. Z. Journal of Hazardous Materials 2009, 170, 530.
32

(79) Mansouri, J.; harrisson, S.; Chen, V. Journal of Materials Chemistry 2010, 20,
4567.
(80) Koh, J. K.; Kim, Y. W.; Ahn, S. H.; Min, B. R.; Kim, J. H. Journal of Polymer Science,
Part B: Polymer Physics 2010, 48, 183.
(81)

Zhao, Y. H.; Wee, K. H.; Bai, R. Journal of Membrane Science 2010, 362, 326.

(82)

Maguire-Boyle, S. J.; Barron, A. R. Journal of Membrane Science 2011, 382, 107.

(83) Chakrabarty, B.; Ghoshal, A. K.; Purkait, M. K. The Chemical Engineering Journal
2010, 165, 447.
(84)

Abdessemed, D.; Nezzal, G.; Ben Aim, R. Desalination 2000, 131, 307.

(85) Benito, J. M.; Rios, G.; Ortea, E.; Fernhdez, E.; Cambiella, A.; Pazos, C.; Coca, J.
Desalination 2002, 147, 5.
(86) Swiegers, G. F. Mechanical Catalysis: Methods of Heterogeneous, Homogenous,
and Enyzmatic Catalysis; John Wiley & Sons: Hoboken, New Jersey, 2008.
(87)

Davis, B. G.; Boyer, V. Natural Product Reports 2001, 18, 618.

(88)

Cole-Hamilton, D. J. Science 2003, 299, 1702.

(89) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M. J.; Delmon, B. J.
Journal of Catalysis 1993, 144, 175.
(90) Bond, G. C.; Sermon, P. A.; Webb, G.; Buchanan, D. A.; Wells, P. B. Journal of the
Chemical Society, Chemical Communications 1973, 444.
(91)

Bond, G. C.; Sermon, P. A. Gold Bulletin 1973, 6, 102.

(92)

Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chemistry Letters 1987, 405.

(93)

Hammer, B.; Norskov, J. K. Nature 1995, 376, 238.

(94)

Haruta, M. Chemical Record 2003, 3, 75.

(95)

Valden, M.; Lai, X.; Goodmand, D. W. Science 1998, 281, 1647.

(96) Astruc, D.; Lu, F.; Aranzaes, J. R. Angewandte Chemie International Edition 2005,
44, 7852.
33

(97)

Pool, R. Science 1990, 248, 1186.

(98)

Pachón, L. D.; Rothenberg, G. Applied Organometallic Chemistry 2008, 22, 288.

(99)

Mirkin, C. A. Small 2005, 1, 14.

(100) Doyle, A. M.; Shaikhutdinov, S. K.; Jackson, S. D.; Freund, H. J. Angewandte
Chemie International Edition 2003, 42, 5240.
(101) Schmid, G.; Maihack, V.; Lantermann, F.; Peschel, S. Journal of the Chemical
Society, Dalton Transactions: Inorganic Chemistry 1996, 589.
(102) Armor, J. N. Catalysis Today 1995, 25, 199.
(103) Melada, S.; Pinna, F.; Strukul, G.; Perathoner, S.; Centi, G. Journal of Catalysis
2006, 237, 213.
(104) Gao, H.; Xu, Y.; Liao, S.; Liu, R.; Liu, J.; Li, D.; Yu, D.; Zhao, Y.; Fan, Y. Journal of
Membrane Science 1995, 106, 213.
(105) Dotzauer, D. M.; Dai, J.; Sun, L.; Bruening, M. L. Nano Letters 2006, 6, 2268.
(106) Liu, C.; Xu, Y.; Liao, S.; Yu, D. Applied Catalysis A: General 1998, 172, 23.
(107) Decher, G.; Schlenoff, J. B. Multilayer Thin Films: Sequential Assembly of
Nanocomposite Materials; Wiley-VCH: Weinheim, Germany, 2003.
(108) Decher, G. Science 1997, 277, 1232.
(109) Decher, G.; Schmitt, J. Thin Solid Films 1992, 210/211, 831.
(110) Kidambi, S.; Bruening, M. L. Chemistry of Materials 2005, 17, 301.
(111) Cho, J.; Caruso, F. Chemistry of Materials 2005, 17, 4547.
(112) Gheith, M. K.; Sinani, V. A.; Wicksted, J. P.; Matts, R. L.; Kotov, N. A. Advanced
Materials 2005, 17, 2663.
(113) Schönhoff, M. Journal of Physics: Condensed Matter 2003, 15, R1781.
(114) Tan, Y.-J.; Wang, W.-H.; Zheng, Y.; Dong, J.; Stefano, G.; Brandizzi, F.; Garavito, R.
M.; Reid, G. E.; Bruening, M. L. Analytical Chemistry 2012, 84, 8357.
34

(115) Srivastava, S.; Kotov, N. A. Accounts of Chemical Research 2008, 41, 1831.
(116) Bergbreiter, D. E.; Franchina, J. G.; Kabza, K. Macromolecules 1999, 32, 4993.
(117) Cavalieri, F.; Postma, A.; Lee, L.; Caruso, F. ACS Nano 2009, 3, 234.
(118) Chirea, M.; Pereira, C. M.; Silva, F. Journal of Physical Chemistry C 2007, 111,
9255.
(119) Chirea, M.; Garcia-Morales, V.; Manzanares, J. A.; Pereira, C.; Gulaboski, R.; Silva,
F. Journal of Physical Chemistry B 2005, 109, 21808.
(120) Porcel, C. H.; Izquierdo, A.; Ball, V.; Decher, G.; Voegel, J. C.; Schaaf, P. Langmuir
2005, 21.
(121) Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213.
(122) Ruths, J.; Essler, F.; Decher, G.; Riegler, H. Langmuir 2000, 16, 8871.
(123) Nakajima, A.; Hashimoto, K.; Watanabe, T. Monatshefte fuer Chemie 2001, 132,
31.
(124) Coulson, S. R.; Woodward, I.; Badyal, J. P. S.; Brewer, S. A.; Willis, C. Journal of
Physical Chemistry B 2000, 104, 8836.
(125) Wong, T.-S.; Kang, S. H.; Tang, S. K. Y.; Smythe, E. J.; Hatton, B. D.; Grinthal, A.;
Aizenberg, J. Nature 2011, 477, 443.
(126) Young, T. Philosophical Transactions of the Royal Society 1805, 95, 65.
(127) Nishino, T.; Meguro, M.; Nakamae, K.; Matsushita, M.; Ueda, Y. Langmuir 1999,
15, 4321.
(128) Wenzel, R. N. Industrial & Engineering Chemistry Research 1936, 28, 988.
(129) Cassie, A. B. D.; Baxter, S. Transactions of the Faraday Society 1944, 40, 546.
(130) Erbil, H. Y.; Demirel, A. L.; Avci, Y.; Mert, O. Science 2003, 299, 1377.
(131) Zhai, L.; Cebeci, F.; Cohen, R. E.; Rubner, M. F. Nano Letters 2004, 4, 1349.
(132) Tuteja, A.; Choi, W.; Ma, M.; Mabry, J. M.; Mazzella, S. A.; Rutledge, G. C.;
McKinley, G. H.; Cohen, R. E. Science 2007, 318, 1618.
35

(133) Kobaku, S. P. R.; Kota, A. K.; Lee, D. H.; Mabry, J. M.; Tuteja, A. Angewandte
Chemie 2012, 51, 10109.
(134) Genzer, J.; Efimenko, K. Biofouling 2006, 22, 339.
(135) Shirtcliffe, N. J.; Roach, P. Methods in Molecular Biology 2013, 949, 269.

36

Some of the text and figures contained in chapter two appeared in the following publication.
Hogg, S.R.; Muthu, S.; O’Callaghan, M.; Lahitte, J.F.; Bruening, M.L. ACS App. Mater. Interfaces
2012, 4, 1440-1448.

37

Chapter 2: Catalytic wet air oxidation of formic acid using modified polysulfone hollow fibers
as membrane contactors

2. 1 Introduction
As discussed in chapter 1, CWAO is potentially attractive for remediating aqueous
streams containing organic pollutants that are too dilute for efficient incineration and too
concentrated for biological processing.

1,2

Due to the low solubility of O2 in water, however,

CWAO requires high air pressure to achieve appreciable oxidation. Gas-liquid membrane
contactors control the introduction of multiple reactants to a fixed catalyst surface

3,4

and can

potentially overcome O2-solubility limitations to provide higher CWAO conversions than
5

conventional batch reactors operating under similar gas pressures. Such contactors utilize
membrane pores to create an interface between liquid and gas streams (Figure 2.1), and
management of the gas overpressure in these systems can sometimes control the degree of
6

pore wetting to position the gas-liquid interface near an area rich in catalyst. In specific cases,
the increased availability of gaseous reactant at the catalyst surface gives conversion rates
three to six times higher in membrane contactors than in trickle bed reactors.

2,5

Hollow fiber

membranes provide an attractive choice for contactor systems due to their large surface-areato-volume ratio and high packing density,
hollow fibers as membrane reactors.

4,7-10

11-17

and several recent studies employed catalytic

This chapter examines whether immobilization of

38

nanoparticle catalysts in polymeric hollow fibers enhances the oxidation of a model pollutant
compared to oxidation in ceramic fibers with much larger diameters. The small lumen diameter
in the polymeric fibers should greatly reduce diffusion distances to increase pollutant
conversion relative to typical ceramic fibers with much larger inner diameters.

Figure 2.1: Schematic drawing of a cylindrical membrane contactor for CWAO with liquid feed
passing through the lumen and gas supplied to the shell side.

Metal nanoparticles are promising catalysts for hollow fiber membrane contactors
because of their high surface area to mass ratio.

18-26

Additionally, nanoparticles often possess

unusual electronic and catalytic properties that stem from their unique size between the bulk
and molecular regimes.

18,19,27

Entrapment of such nanoparticles in a hollow fiber both

prevents catalyst aggregation and allows for continuous reactions. Although several methods
exist for forming nanoparticles by reduction of metal ions in membranes,

39

28-30

synthesis of

nanoparticles in solution prior to loading onto supports typically yields more uniform particle
21,28,31-38

shapes and diameters.

However, loading of preformed nanoparticles in membranes

requires strong adhesion between the nanoparticle and the support.
We employ layer-by-layer (LbL) adsorption of nanoparticles and oppositely charged
polyelectrolytes to immobilize preformed catalytic nanoparticles in membrane pores.

39-45

LBL

deposition occurs on many substrate materials with geometries that include flat surfaces,
29,35,36,38,39,46,47

membrane pores, cylindrical particles, and carbon nanotubes.

Thus, this

technique is amenable to hollow-fiber modification. Moreover, the amount of immobilized
particles increases with the number of adsorbed layers.
Miachon and coworkers examined tubular ceramic membranes as interfacial contactors
for CWAO and found increased activity compared to a stirred tank reactor because of control
over the position of the gas-liquid interface within the membrane.

5,48-52

Notably, LBL

deposition of polyelectrolyte multilayers and Pt nanoparticles yielded a membrane reactor with
the highest Pt activity among those tested.

29

However, the pollutant conversion was relatively

low, and the price of the ceramic membranes might limit their potential application.

53

Polymeric hollow fibers may provide a much less expensive option for CWAO of organic
pollutants if these membranes are sufficiently stable.
This chapter describes LBL deposition of polyelectrolytes and citrate-stabilized Pt
nanoparticles in polysulfone (PS) hollow fibers to create membrane reactors for CWAO of
aqueous formic acid. We compare catalytic activities and conversions to previous work with
40

ceramic hollow fiber membranes and model concentration profiles in membrane reactors to
assess the potential benefits of the polymeric hollow fibers. The small diameters of these fibers
give rise to higher single-pass conversions compared to previous, larger ceramic membranes,
but catalyst activity is more modest in the polymeric hollow fibers.

2.2 Experimental methods
2.2.1 Materials
Poly(sodium 4-styrenesulfonate) (PSS, Mw = 70,000 Da), branched poly(ethyleneimine)
(PEI, Mw = 25,000 Da), hexachloroplatinic acid hexahydrate, sodium chloride, and sodium
citrate were used as received from Sigma-Aldrich. Formic acid from Mallinckrodt Baker and
compressed O2 (industrial purity, 99.5%) from Airgas were also used as received. Deionized
water (Milli-Q purification system, 18.2 MΩ·cm) was used for rinsing the membranes and
preparation of the polyelectrolyte and formic acid solutions. The pH values of the
polyelectrolyte solutions were adjusted with dilute solutions of HCl and NaOH.
A PS microfiltration hollow fiber membrane module containing 270 fibers was obtained
from GE Lifesciences (Model # CFP-6-D-6A). Individual fibers were removed from the module
and repotted in 15 cm lengths. Membranes have a stated maximum pore size of 650 nm, an
inner diameter of 750 μm, and a wall thickness of 225 μm.

2.2.2 Synthesis of citrate-stabilized Pt nanoparticles
Synthesis of Pt nanoparticles followed a literature procedure that uses citrate to both
reduce the Pt and prevent particle aggregation through charge repulsion.
41

54

Forty mg of platinic

acid dissolved in 255 mL of water was stirred and heated to reflux in a 500 mL round bottom
flask before addition of 30 mL of a 1% (w/w) sodium citrate solution. The Pt solution slowly
darkened from amber to brown and eventually to black over four hours of reflux. After cooling
to room temperature over ice, the Pt colloid solution was stored in an amber glass bottle in a
refrigerator. Immediately prior to deposition, the Pt colloids were diluted 12.5:1 with deionized
water. Our previous transmission electron microscopy images of Pt nanoparticles prepared
using this method showed a size distribution centered around 3 nm.

29

2.2.3 Modification of hollow fiber membranes
Following a literature procedure, a 1.3 cm-diameter PVC tube housed the hollow fibers
potted in epoxy resin (Loctite E-00NS Hysol Epoxy Adhesive).

55

Modules contained either one

centered fiber or five fibers arranged in a “X” geometry with accessible fiber lengths of 15 cm
between the epoxy resin at each end. Immediately prior to introduction into the module, all
deposition solutions passed through a fresh filter (Ahlstrom grade 601 2.5 μm qualitative filter
paper) to remove any particles that might block fiber pores. The polyelectrolyte solutions
contained either 0.8 mM PSS or 0.8 mM PEI along with 20 mM NaCl. (Polymer concentrations
are with respect to the repeating unit.) Initially, the PS hollow fibers were rinsed with distilled
water for 30 min in a dead end setup with water flowing from shell to lumen. During
polyelectrolyte and nanoparticle adsorption, solutions passed from the lumen to the shell side
using lumen cross-flow under an applied pressure of 0.34 bar. A partially closed valve at the
lumen exit forced most of the flow through the membrane pores (Figure 2.2). Modification

42

began by passage of 250 mL of PSS solution through the fiber to adsorb PSS via hydrophobic
interactions with PS.

56,57

A deionized water rinse from shell through lumen (until the

conductivity of the permeate was less than 10 μS, 100 to 800 mL of rinsing) removed excess
polyelectrolyte from the membrane. Next, passage of 250 mL of PEI solution through the
membrane, followed by rinsing, completed deposition of the first polyelectrolyte bilayer. Pt
nanoparticles were subsequently adsorbed to the PEI layer during passage of 250 mL of the
colloid solution (~0.02 mM in Pt atoms) prior to another rinse. Repetition of the PEI and Pt
deposition process yielded additional bilayers. During the colloid adsorption, the fiber changed
color from white to gray, indicating entrapment of Pt nanoparticles within the membrane.

Figure 2.2: Apparatus for fiber modification with polyelectrolyte multilayers. Deposition
solutions enter the fiber lumen before exiting under cross-flow. A valve located at the lumen
exit creates backpressure to force most solution through the high surface area pores to the
shell. Rinsing after depositions proceeds through passage of water from the shell side through
the lumen in a dead-end setup. For interpretation of the references to color in this and all other
figures, the reader is referred to the electronic version of this dissertation.

2.2.4 Membrane characterization
Scanning electron microscopy images of hollow fibers before and after modification
were obtained using a Hitachi S-4700 II field-emission scanning electron microscope (SEM).
Deposition of 8 nm of sputtered gold rendered the samples conductive for imaging. To prepare
43

membrane cross sections, the fiber segments were soaked in ethanol for two minutes to
completely wet the fiber before fracturing in liquid nitrogen and mounting for imaging. Because
the Pt nanoparticle size is below the resolution of the SEM, the Pt loading onto fibers was
determined through analysis by inductively coupled plasma optical emission spectroscopy (ICPOES). Fibers were cut into several pieces, sonicated in aqua regia for 10 min, and the resulting
solutions were diluted 100-fold prior to analysis with a Varian 710-ES ICP-OES at 214.424 nm.

2.2.5 Catalytic reactions
Wet air oxidation of formic acid served as a model reaction for assessing the catalytic
activity of PS hollow fibers. Initially, 250 mL of formic acid feed solution was pushed through
the membrane using cross-flow at 0.34 bar to wet all pores. Again, a partially closed valve at
the membrane outlet forced most of the flow from the lumen to the shell. Subsequently, a
peristaltic pump pulled the formic acid solution through the fiber lumen at 0.1 mL/min while an
O2 or N2 overpressure was applied to the shell side to prevent transmembrane flux. Four initial
feed samples along with permeate samples at specified intervals were diluted and analyzed for
formic acid using ion chromatography with a Dionex LC20 instrument equipped with an Ionpac
AS16 column.

2.2.6 Modeling
The membrane reactor consists of a tubular fiber with solution introduced through the
lumen, and the incoming formic acid reacts only in the fixed porous catalyst bed within the
fiber walls. An overpressure on the shell side prevents liquid transmembrane flow but is not
44

sufficient to de-wet the pores. The model used for calculation couples the free fluid and porous
media flow through the Navier-Stokes equations and Brinkman’s extension of Darcy’s law. Due
to the symmetry of the system, only half of the fiber needs to be modeled using a twodimensional axisymmetric geometry. The Navier-Stokes equation describes the fluid flow in the
lumen according to equation 2.1, where

)

.

)

)

(2.1)

Within the porous wall, the Brinkman equation describes the flow according to equation 2.2,
. In these equations, η denotes the viscosity of

where

)

)

(2.2)

2

the fluid (Ns/m ), εp is the dimensionless porosity, u is the velocity (m/s), ρ is the density
3

2

(kg/m ), p is the pressure (Pa), κ is the permeability (m ), and I is the identity matrix.
A Fickian approach is suitable for formic acid diffusion, and equation 2.3 provides the
mole balance for formic acid mass transport and reaction. In equation 2.3, cA denotes the
concentration of formic

)
3

(2.3)
2

acid (mol/m ), DA is the diffusivity (m /s), and RA is the reaction rate for formic acid oxidation
3

2

58

[mol/(m ·s)]. We estimated DA as 1.516 x 10-9 m /s.

The reaction only takes place in the

porous walls, so the reaction term is zero in the lumen, and formic acid oxidation in the pores

45

occurs according to equation 2.4. Because the reaction rate did not vary significantly with
external O2 pressure within the range tested, we
(2.4)
where k is the reaction rate constant.

assumed a first-order reaction rate,

Upon the solution’s entry into the lumen, a fully developed laminar flow is assumed and
equation 2.5 gives the velocities where n is the boundary normal vector and r is the distance
from the

[

( ) ]

(2.5)

center of the circular cross-section with a total radius R. At the fiber outlet, the boundary
conditions for the Navier-Stokes equations are

where t is any

and

tangential vector to the boundary. The formic acid concentration at the inlet was fixed as

)

. At the outlet, convection dominates the formic acid mass transport yielding

)

. This implies the gradient of cA in the direction

perpendicular to the outlet boundary is negligible, a common assumption for tubular reactors
with a high degree of transport by convection in the direction of the main reactor axis (Figure
2.3). Thus, this condition eliminates the need for specifying a concentration or a fixed value for
the flux at the outlet boundary. At all other boundaries, insulation conditions apply as given by
equation 2.6.

)

(2.6)
46

Equation 2.7 describes the formic acid conversion where cA is the mixing cup concentration at
the outlet determined from equation 2.8. All calculations were performed by Satish Muthu in
the chemical engineering module of COMSOL Multiphysics 3.5a.

(

)

(2.7)

∫

(2.8)

∫

Figure 2.3: Modeled fiber system dimensions and boundary conditions for inlet and outlet
formic acid concentrations showing only half of the lumen and a single wall.

2.3 Results and discussion
2.3.1 Membrane modification with Pt nanoparticles
Alternating adsorption of polyelectrolytes and Pt nanoparticles yields coated hollow
fiber membranes with open pores. Adsorption occurs during permeation of polyelectrolyte or
47

nanoparticle solutions from the lumen to the shell, so films deposit throughout the membrane
wall, and the membrane turns gray after nanoparticle deposition. The SEM images in Figure 2.4
(A,B) show no obvious changes in the membrane lumen pore sizes after deposition of a
PSS/[PEI/Pt]2 film, presumably because the thickness of the polyelectrolyte/nanoparticle films
is less than 10 nm.

38,59

Meanwhile, the images of the smaller pores on the shell side (Figure 2.4

C,D) show some changes in the surface morphology after film deposition, but pores remain
open. The absence of a cross-flow rinse on the shell side and solution dripping from this surface
during modification likely leads to additional deposition on the shell compared to the lumen.
When a cross-flow rinse was incorporated on the shell side during modification, no significant
pore size reductions were visible in SEM images on either the lumen or shell surfaces. Overall,
pore size reduction when depositing without cross-flow decreased O2 permeability <20% at 0.2
bar, so the open pores on the shell surface will afford the gaseous reactant access to the
catalyst-solution interface.

48

Figure 2.4: Representative SEM images of the lumen (A,B) and shell (C,D) surfaces of PS hollow
fibers before (A, C) and after (B,D) adsorption of a PSS/[PEI/Pt nanoparticle]2 film.

To examine the uniformity of Pt loading along the length of a PS fiber modified with a
PSS[PEI/Pt]2 film, we cut a 15 cm-long fiber into four 3.75 cm-long sections, dissolved the Pt in
aqua regia, and analyzed the aqua regia solutions by ICP-OES. The relative standard deviation
(RSD) in the Pt loading for the four segments (Table 2.1, row 1) was less than 10%. The uniform
loading of Pt nanoparticles should ensure similar catalyst availability over the entire length of
the fiber. We also simultaneously modified five fibers in a single module, cut each fiber into
three 5 cm-long sections and determined the amount of Pt in each segment. In this case the
relative standard deviation in the Pt content of the 15 segments was 17% (Table 2.1, rows 2 to
6). Overall, the loading/cm in the five-fiber module is about 60% of that in a single fiber. The

49

lower deposition solution cross-flow rate during adsorption in the five-fiber module might lead
to lower loadings.
Table 2.1: Pt loading per length of fiber after LBL modification of a single-fiber module and a
five-fiber module with PSS/[PEI/Pt]2 films. Loadings were determined by dissolution of the Pt in
aqua regia and analysis by ICP-OES. The segments were approximately 3.75 cm long for the
single fiber and 5 cm long for the 5-fiber module.

2.3.2 Catalytic performance of PS hollow fiber modules
2.3.2.1 CWAO of formic acid using a membrane contactor
Oxidation of formic acid is an ideal model reaction to test the catalytic activity of
immobilized Pt nanoparticles because the only products are CO2 and water, and Pt is a common
catalyst for this reaction.

60-62

In a control experiment, we bubbled a 50 mM formic acid

solution with O2 and then passed this solution through the fiber lumen while applying a N2
overpressure (0.69 bar) to the shell side of the membrane. One pass of the solution through a
single, Pt-modified PS hollow fiber at 0.1 mL/min (membrane residence time of 40 s) resulted in
oxidation of only 6 ± 3% of the formic acid (Figure 2.5, triangles). The low conversion stems
50

from the limited O2 solubility in the feed solution. The room-temperature solubility of O2 in
29,63

water is ~ 1.25 mM at 1 atm of O2,

and one molecule of O2 can oxidize two molecules of

formic acid. Thus, the O2 in solution can oxidize a maximum of ~2.5 mM formic acid, which is
similar to the observed reaction of 6 ± 3% of the 50 mM formic acid during passage through the
membrane.

In contrast, application of an O2 overpressure (0.69 bar) to the shell side of the hollow
fiber resulted in steady-state oxidation of 63 ± 3% of the formic acid in a 50 mM feed solution
during a single pass through the membrane lumen with a residence time of 40 s. (In this case,
the solution was not saturated with O2 prior to passing through the membrane.) As Figure 2.5
shows (diamonds), after changing the gas in the module from 0.69 bar N2 overpressure to 0.69
bar O2 overpressure with no prior O2 saturation of the feed, the concentration of formic acid
exiting the membrane decreased to the steady state value of 37% of the feed concentration
after displacing the dead volume (~4 mL) at the exit of the hollow fiber module, demonstrating
the O2 dependence of the reaction. After 360 min of oxidation, we changed the shell gas from
0.69 bar O2 to 0.69 bar N2 (overpressure), and conversion declined again confirming the
importance of the membrane contactor for delivering O2 to enhance conversion.

To increase the fraction of formic acid oxidized without decreasing the overall
volumetric flow rate, we increased the number of fibers in the module. Notably, with modules
51

containing 5 fibers operating at the same nominal flow rate, formic acid oxidation reaches 94 ±
3% in a single pass (Figure 2.5, squares) without saturating the solution with O2 prior to passing
it through the membrane. The increase in conversion results from the higher residence times
(198 s) in modules with more fibers. Further increases in the number of fibers will allow even
higher conversions or similar conversions at higher flow rates through the module.

Figure 2.5: Ratio of the outlet to inlet concentrations of formic acid during CWAO in three
different fiber modules coated with PSS/[PEI/Pt]2 films. Triangles represent an O2-saturated
solution as the lone source of O2 using a single Pt-coated PS fiber with N2 (0.69 bar
overpressure) on the shell side. Diamonds and squares represent a single and 5-fiber contactor
reactor, respectively, where an O2 overpressure (0.69 bar) was applied on the shell side at 0
min and subsequently replaced with a N2 overpressure (0.69 bar) at 360 min. The solution flow
rate was 0.1 mL/min in all cases.

2.3.2.2 Fiber longevity
The stability of modified fibers is a major concern because both oxidation of the
polymer and shear stress may leach nanoparticles from the membrane. However, ten 8-hour
replicates of CWAO (50 mM formic acid) with the same fiber show a RSD in formic acid
oxidation of less than 20%, suggesting no continuous leaching of Pt from the system (Figure
52

2.6). While there is some spread in the net conversions in different experiments, the general
trends remain the same in every trial. More importantly, after the first three trials, no clear
trend of decreasing conversion with each subsequent trial occurred. The conversion decline
after the first three trials does not appear to result from Pt leaching since the conversion did
not decrease significantly during the trials. Instead, an inactivation of some catalyst sites may
occur during drying of the fiber between experiments. The outlying data points from trial six are
probably the result of IC instrument error rather than increased performance. Therefore, the
fiber remained relatively stable during the eighty hours of testing in this experiment.

Figure 2.6: Ratio of the outlet to inlet concentrations of formic acid during CWAO for 10 trials of
8 hours using a single PS fiber module modified with a PSS/[PEI/Pt] 2 film. An O2 overpressure of
0.69 bar was applied at 0 min on the shell side and replaced with N2 at 360 min.

2.3.2.3 Dependence of oxidation rates on formic acid concentration and O 2 overpressure

To explore the factors limiting the rate of CWAO in hollow fiber membranes, we varied
both the formic acid concentration and the O2 overpressure during reactions. Figure 2.7 shows
53

the average concentration of formic acid oxidized (the difference between feed and outlet
concentrations) as a function of the feed solution formic acid concentration passed through a
single-fiber module. Error bars represent the standard deviation of the average oxidized formic
acid concentration over a period of 5 hours for two replicate membranes with identical
modification (n=2). At feed concentrations from 10 to 50 mM, the change in formic acid
concentration upon passing through the membrane varies approximately linearly with the feed
concentration. In contrast, the concentration of formic acid oxidized in a single pass through
the membrane plateaus at 40 mM for feed concentrations from 80 to 150 mM. These data
suggest that formic acid concentration limits the reaction rate at low feed concentrations (due
to either diffusion or kinetic limitations), whereas either the availability of O 2 in the catalyst
layer or the amount of catalyst limit the reaction at high formic acid feed concentrations.

Figure 2.7: Concentration of formic acid oxidized (inlet - outlet concentration) in CWAO with a
single-fiber module and an O2 overpressure of 0.69 bar on the shell side of the membrane. The
line represents a linear fit (forced through the origin, slope of 0.55) to the first three data
points. Data are an average of two experiments with two modified single fibers.
54

Figure 2.8 shows the average concentration of formic acid oxidized when using a 50 mM
formic acid feed and various O2 overpressures on the shell side of single fiber modules. Within
experimental error, the amount of formic acid oxidized is independent of the O 2 overpressure
tested. The highest overpressure of ~1 bar represents the maximum recommended operating
pressure from the manufacturer of this PS fiber. One concern in these experiments, however, is
whether changing the overpressure alters the wetting of the pores. The Young-Laplace
equation

(2.9),
)

(2.9)

provides an estimate for the critical pressure, Δp, required to prevent pore wetting, where θ is
the solution contact angle with the material, γ is the surface tension of the liquid, and r is the
64

pore radius. Based on a water contact angle of 67◦ on polysulfone,

and a pore radius of 325

µm, this equation suggests that the pressure required to overcome capillary wetting would be
around 1.7 bar for aqueous solutions. This is consistent with the fact that we observed no gas
bubbles breaking through the membranes. If the pores are indeed wetted at all the pressures in
these experiments, we would expect that increased O2 solubility in water might enhance
oxidation rates. The constant reaction rate at various O2 pressures may stem from the fairly
narrow range of total pressures (1.1-2 bar absolute pressure) and the fairly complicated kinetics
of the formic acid oxidation. Harmsen et al. suggest that equation 2.10 describes the rate of
65

formic acid oxidation, R.

55

[
[

]
]

[

]
[

(2.10)

])

In this expression, K, K’ and K’’ represent products of different equilibrium constants and
surface site densities. At most, the reaction would show a square root dependence on the
concentration of O2 in solution.

Figure 2.8: Concentration of formic acid oxidized during CWAO with single fiber modules at
varying overpressures. Membranes were modified with PSS/[PEI/Pt]2 films and the feed
solution contained 50 mM formic acid. Data are an average of three replicate experiments on
each of two different fibers.

2.3.2.4 Effect of Pt loading on conversion
Low Pt loading will limit the rate of formic acid oxidation when the catalyst is saturated
with reactants. To examine the effect of catalyst loading on oxidation rate, we compared singlefiber membrane reactors modified with one and three PEI/ Pt bilayers atop a base layer of PSS.
Analysis of Pt loading in these fibers using ICP-OES shows that adsorption of each PEI/Pt bilayer
56

adds approximately 500 μg Pt to the membrane. Figure 2.9 demonstrates that the amounts of
formic acid oxidized in fibers containing 1 PEI/Pt bilayer (replicate fibers 1 and 3) are
significantly less than in fibers containing 3 bilayers (replicate fibers 2 and 4) (p=0.99) for feed
solutions containing 50 and 100 mM formic acid. At feed concentrations of 25 mM, the two
types of membranes show similar conversions, presumably because the catalyst is not
saturated with reactant. As in Figure 2.5, the concentration oxidized increases with the feed
(inlet) concentration. The average ratios of the concentrations oxidized in membranes modified
with 3 and 1 PEI/Pt bilayers were 1.14, 1.28, and 1.31 at feed concentrations of 25, 50, and 100
mM, respectively. Thus, although higher Pt loading enhances formic acid oxidation, the increase
of ~1.30 in the amount oxidized is significantly less than the 3-fold increase in Pt loading.
Perhaps Pt in the inner layers of the coating does not participate in oxidation due to slow
formic acid diffusion into the coating.

Figure 2.9: Concentration of formic acid oxidized (inlet - outlet concentration) in CWAO with a
single-fiber module and various formic acid concentration in the inlet solutions. Fibers 1 and 3
were replicate fibers modified with PSS/PEI/Pt (1-Layer) films while fibers 2 and 4 were
57

modified with PSS[PEI/Pt]3 (3-Layer) films. The error bars in the figure represent standard
deviations with an n value of more than 40.

2.3.3 Calculated formic acid concentration profiles
We first calculated the diffusion-limited formic acid conversion. In this calculation, we
assumed laminar flow throughout the fiber lumen and that formic acid oxidation occurs
instantaneously upon encountering the fiber wall, i.e., the concentration of formic acid at the
lumen wall is zero. Figure 2.10 shows the resulting concentration profile. For a 15 cm-long fiber
and a flow rate of 0.1 mL/min, the mixing cup concentration exiting the fiber is only ~4% of the
feed concentration (>96% oxidation of formic acid). Thus the reaction we observe in single
fibers (40-60% oxidation) is significantly slower than the diffusion-limited case where reaction
occurs instantaneously at the lumen wall. The diffusion-limited oxidation in the five-fiber
module is even higher (>99.9%) because of the longer residence time, and this conversion is
again much greater than the 94% we observed.

58

Figure 2.10: Diffusion-limited formic acid concentration profile in a single-fiber reactor as a
function of the distances from the inlet (z,Y-axis) and the center of the fiber (r, X-axis). The
fiber inner radius was 375 µm, and reaction was assumed to occur instantaneously at the fiber
wall, where r=375 µm. The results predict 96% oxidation of the 50 mM formic acid by the time
the solution reaches the outlet. The flow rate was 0.1 mL/min.

To more reasonably model the system, we allowed for diffusion to reaction sites within
the membrane walls and varied the first-order rate constant for formic acid oxidation. We
assumed fully developed laminar flow in the interior of the fiber and no convective flow in the
membrane walls. Figure 2.11 shows the concentration profile in a single-fiber membrane
-1

reactor assuming a rate constant within the fiber walls of 0.01 s . The vertical line indicates the
-1

lumen wall. In this case, a rate constant of 0.01 s

leads to a formic acid concentration profile

very different from that in the diffusion-limited case. Reaction occurs throughout the wall, but
is more prevalent near the lumen. The mixing cup outlet concentration is about half that of the
feed concentration, which agrees well with the observed conversion. Note that we assumed a
59

first order rate constant because the O2 pressure outside the fiber did not affect conversion,
see Figure 2.7. In modeling the reaction in the five fiber module using the same rate constant,
we also see agreement between experimental and observed conversion (94 ± 3 % experimental
versus 97 % in the simulation). Additionally, we modeled O2 concentration profiles for the
polymeric fiber module at overpressures of 0.1, 0.69, and 1 bar, see Figure 2.12. As expected,
the oxygen concentration decreases from the shell to the lumen, so a first-order reaction is an
oversimplification. Near the lumen reaction rates will likely be lower than simulated, whereas
near the shell wall they will be higher than simulated.

Figure 2.11: Calculated formic acid concentration profile during oxidation in a single-fiber
reactor. The plot shows the concentration by color at any point along the length of the reactor
(z, Y-axis) and at any radial distance between the center of the fiber to the outer wall (r, X-axis).
The results predict ~52% oxidation of the formic acid by the time the solution reaches the
-1

outlet. The assumed first-order rate constant was 0.01 s . The vertical line at r=375 m
represents the fiber wall.

60

Figure 2.12: O2 concentration profiles in a polymeric hollow fiber membrane as a function of
the distances from the inlet (z,Y-axis) and the center of the membrane (r, X-axis). The
-9

2

simulation uses an O2 diffusion coefficient of 1.96 x 10 m /s in solution. The inner fiber radius
was 375 m, and reaction was assumed to occur throughout the fiber wall with a first-order
-1

rate constant for formic acid oxidation of 0.01 s . We assumed that the reaction (equation 2.4)
is first order with respect to formic acid and zero order with respect to O 2. The formic acid
solution enters the fiber lumen saturated with atmospheric O2 (0.25 mM) while the shell of the
fiber is exposed to (A) 0.1 bar O2 overpressure (1.42 mM in solution), (B) 0.69 bar O2
overpressure (2.18 mM in solution), and (C) 1.0 bar O2 overpressure (2.58 mM in solution).
Note than in Figure A there is a small region of negative concentration, showing that the
assumption of zero-order reaction with respect to O2 is an oversimplification.

61

Figure 2.12 (cont’d)

Figure 2.13 shows the predicted conversions for a single fiber module at a range of first-1

-1

order rate constants from 0.001 to 1000 s . At values above 1 s , the reaction becomes
62

limited by diffusion of formic acid to the walls of the fiber, and the concentration near the
lumen wall is nearly zero. Rate constants below 0.01 s

-1

give much lower reaction than we
-1

observed. Conversion of the pseudo-homogeneous rate constant (k, 0.01 s ) to a
heterogenous rate constant, k’, requires normalization to the amount of Pt surface area, A, per
solution volume, V, according to equation 2.11.

(11)

Given the total Pt loading of 1 mg in a single fiber and assuming 3 nm-diameter particles, the
-7

value of k’ is 8 x 10 cm/s.

Figure 2.13: Simulated mixed cup conversion of formic acid for a single fiber reactor assuming
first-order rate constants for formic acid wet air oxidation from .001 to 1000 (1/s).

Figure 2.14 shows the predicted effect of membrane wall thickness (over a range of 100300 µm) on the extent of formic acid oxidation. We should note again that this calculation

63

neglects any effect the wall thickness may have on the availability of O2 for the reaction. The
results suggest that the thickness of the wall plays a modest role in the oxidation process, as
the extent of formic acid oxidation only increases from 32% to 57% on increasing the wall
thickness from 100 µm to 300 µm. Most of the reaction takes place within the first 100 µm of
the lumen surface. The model assumes that the pores of the membrane are completely wetted,
which, as mentioned above, is likely given the size of the pores and the water contact angle of
the membrane.

Figure 2.14: Simulated formic acid conversion in a single-fiber module with membranes having
different wall thicknesses. The inner radius of the fiber was 375 µm, and the rate constant was
-1

0.01 s .

2.3.4 Comparison to previous work
The best tubular ceramic membranes for CWAO in Miachon’s work catalyzed oxidation
of about 10% of a 100 mM formic acid solution during a single pass at a flow rate between 7
and 10 mL/min (linear velocity of 0.43 cm/s, residence time of 58 s, membrane length of 25
49

cm).

Those membranes had a lumen diameter of 7 mm, and our simulation of diffusion64

limited oxidation in such large membranes gives a conversion of 13%, similar to the measured
value (see Figure 2.15). The innermost 3 µm layer of the lumen wall in the ceramic membranes
contained 20 nm pores which likely allow a high Pt loading near this wall, as determined by
2

SEM-EDS mapping. The high catalyst loading at the lumen wall may lead to a rate that is closer
to the diffusion limit than in the case of the hollow fiber membranes.

Figure 2.15: Diffusion-limited formic acid concentration profile in a tubular ceramic membrane
as a function of the distances from the inlet (z,Y-axis) and the center of the membrane (r, Xaxis). The inner radius was 3.5 mm, and reaction was assumed to occur instantaneously at the
lumen wall, where r = 3.5 mm. The results predict 13% oxidation of the 100 mM formic acid by
the time the solution reaches the outlet. The flow rate was 10 mL/min. The low conversion
stems from high diffusion distance resulting from the large diameter of the ceramic membrane.

Nevertheless, at similar residence times, the polymer single fiber modules in this study
give a higher single-pass conversion of formic acid than the previous ceramic hollow fibers,
presumably because of the smaller average diffusion distances between the molecules in

65

solution and the polymer fiber. (For the large fiber, the depletion layer around the lumen never
reaches the center of the fiber, Figure 2.15.) Use of ceramic membranes with smaller lumen
diameters (3 mm) should also reduce formic acid diffusion distances and lead to somewhat
higher single-pass conversions. However, the cost of the ceramic membranes may still limit
their practicality. Additionally, while smaller ceramic fibers still contain larger pores on the shell
surface than the lumen, they may lack the distinct multilayered wall composition that afforded
66

high catalyst density near the lumen surface in the work of Miachon.

Remarkably, the five-fiber PS modules in this study catalyzed oxidation of 94 % of the
formic acid in a 50 mM feed solution in a single pass at a flow rate of 0.1 mL/min (linear velocity
of 0.08 cm/s, residence time of 198 s, 15 cm fiber). Assuming performance remains the same
with further scale up and the same linear velocity, a module containing 500 PS fibers would
oxidize 94 % of the formic acid in a single pass, whereas a single 7 mm (inside diameter)
ceramic fiber would oxidize only about 10 % of the formic acid at the same total flow rate. (The
cost of a 500-fiber module is about the same as that of a ceramic fiber.)
One major disadvantage of the polymeric fibers, however, is the relatively low activity of
the immobilized Pt. With a feed solution containing 100 mM formic acid, the modified PS fibers
exhibited a normalized activity of 0.07 mmol formic acid/s·g Pt, whereas the best ceramic
membranes showed an activity of 1.5 mmol formic acid/s·g Pt . The higher activity in the
ceramic membranes likely stems from the 3 layer geometry of the fibers, and the high
concentration of catalyst near the lumen. Moreover, because outer pores do not wet, nearly all
of the catalyst resides near the liquid-air interface to provide readily accessible O2 for

66

oxidation. In contrast, for the polymeric membrane Pt deposition and wetting occur throughout
the fiber wall, so oxygen availability may significantly limit the reaction. Further refinement of
catalyst deposition and wetting properties, probably using membranes with small pores at the
lumen interface, should lead to increased activity in polymeric fiber reactors.

2.4. Conclusions
CWAO in single polymeric hollow fibers at room temperature leads to over 50%
oxidation of the formic acid in 50 mM feed solutions during a single pass through a hollow fiber.
Additionally, 94% oxidation occurs for a single pass through five parallel fibers at the same
volumetric flow rate but lower linear velocity. The PS hollow fibers show significantly higher
conversions than previous ceramic membranes because the small diameter of the hollow fibers
decreases radial diffusion distances. Modeling the diffusion of formic acid within the lumen
indicates the oxidation is significantly below the diffusion-limited rate. A simple first-order
model of the reaction suggests an apparent rate constant of 0.01 s

-1

and shows that reaction

occurs throughout the fiber wall.
One major challenge to the use of the catalytic PS hollow fibers is the low catalyst
activity caused by a disperse loading of Pt nanoparticles throughout the wall of the fibers
instead of in a tightly concentrated region near the lumen surface. With increased nanoparticle
deposition localized near the lumen walls and more controlled pore wetting, the activity might
improve without decreasing conversion.

67

REFERENCES

68

REFERENCES

(1)
Besson, M.; Descorme, C.; Bernardi, M.; Gallezot, P.; di Gregorio, F.; Grosjean, N.;
Pham Minh, D.; Pintar, A. Environmental Technology 2010, 31, 1441.
(2)
Vospernik, M.; Pintar, A.; Berčič, G.; Levec, J.; Walmsley, J. C.; Ræder, H.; Iojoiu,
E. E.; Miachon, S.; Dalmon, J.-A. Chemical Engineering Science 2004, 59, 5363.
(3)

Fritsch, D.; Bengtson, G. Advanced Engineering Materials 2006, 8, 386.

(4)

Gabelman, A.; Hwang, S.-T. Journal of Membrane Science 1999, 159, 61.

(5)
Miachon, S.; Perez, V.; Crehan, G.; Torp, E.; Ræder, H.; Bredesen, R.; Dalmon, J.A. Catalysis Today 2003, 82, 75.
(6)

Westermann, T.; Melin, T. Chemical Engineering and Processing 2009, 48, 17.

(7)

Kleinert, A.; Feldhoff, A.; Schiestel, T.; Caro, J. Catalysis Today 2006, 118, 44.

(8)
Peters, T. A.; Fontalvo, J.; Vorstman, M. A. G.; Keurentjes, J. T. F. Chemical
Engineering Research and Design 2004, 82, 220.
(9)
907.

Wang, H.; Tablet, C.; Schiestel, T.; Caro, J. Catalysis Communications 2006, 7,

(10) Mulder, M. Basic Principles of Membrane Technology; Kluwer Academic
Publishers: Dordrecht, 1996.
(11) Ouyang, L.; Dotzauer, D. M.; Hogg, S. R.; Macanas, J.; Lahitte, J.; Bruening, M.
Catalysis Today 2010, 156, 100.
(12) Wang, H.; Feldhoff, A.; Caro, J.; Schiestel, T.; Werth, S. AIChE Journal 2009, 55,
2657.
(13) Van der Vaart, R.; Lebedeva, V. I.; Petrova, I. V.; Plyasova, L. M.; Rudina, N. A.;
Kochubey, G. F.; Tereshchenko, G. F.; Volkov, V. V.; van Erkel, J. Journal of Membrane
Science 2007, 299, 38.
(14) Peirano, F.; Vincent, T.; Quignard, F.; Robitzer, M.; Guibal, E. Journal of
Membrane Science 2009, 329, 30.

69

(15) Blondet, F. P.; Vincent, T.; Guibal, E. International Journal of Biological
Macromolecules 2008, 43, 69.
(16)

Israni, S. H.; Nair, B. K. R.; Harold, M. P. Catalysis Today 2009, 139, 299.

(17) Caro, J.; Caspary, K. J.; Hamel, C.; Hoting, B.; Kolsch, P.; Langanke, B.; Nassauer,
K.; Schiestel, T.; Schmidt, A.; Schomacker, R.; Seidel-Morgenstern, A.; Tsotsas, E.; Voigt,
I.; Wang, H.; Warsitz, R.; Werth, S.; Wolf, A. Industrial & Engineering Chemistry Research
2007, 46, 2286.
(18) Astruc, D.; Lu, F.; Aranzaes, J. R. Angewandte Chemie, International Edition 2005,
44, 7852.
(19)

Pool, R. Science 1990, 248, 1186.

(20) Mei, Y.; Sharma, G.; Lu, Y.; Ballauff, M.; Drechsler, M.; Irrgang, T.; Kempe, R.
Langmuir 2005, 21, 12229.
(21) Comotti, M.; Li, W. C.; Spliethoff, B.; Schüth, F. Journal of the American Chemical
Society 2006, 128, 917.
(22)

Jiang, Y.; Gao, Q. Journal of the American Chemical Society 2006, 128, 716.

(23)

Meyer, D. E.; Bhattacharyya, D. Journal of Physical Chemistry B 2007, 111, 7142.

(24) Brandão, L.; Fritsch, D.; Mendes, A. M.; Madeira, L. M. Industrial & Engineering
Chemistry Research 2007, 46, 5278.
(25)

Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647.

(26) Park, K. H.; Jang, K.; Kim, H. J.; Son, S. U. Angewandte Chemie, International
Edition 2007, 46, 1152.
(27)

Pachón, L. D.; Rothenberg, G. Applied Organometallic Chemistry 2008, 22, 288.

(28) Bhattacharjee, S.; Dotzauer, D. M.; Bruening, M. L. Journal of the American
Chemical Society 2009, 131, 3601.
(29) Dotzauer, D. M.; Abusaloua, A.; Miachon, S.; Dalmon, J.-A.; Bruening, M. L.
Applied Catalysis, B: Environmental 2009, 91, 180.
(30) Perez, V.; Miachon, S.; Dalmon, J. A.; Bredesen, R.; Pettersen, G.; Ræder, H.;
Simon, C. Separation and Purification Technology 2001, 25, 33.
70

(31) Zhong, L. S.; Hu, J. S.; Cui, Z. M.; Wan, L. J.; Song, W. G. Chemistry of Materials
2007, 19, 4557.
(32)

Gates, B. C. Chemical Reviews 1995, 95, 511.

(33)

Wolf, A.; Schüth, F. Applied Catalysis, A: General 2002, 226, 1.

(34) Zheng, N.; Stucky, G. D. Journal of the American Chemical Society 2006, 128,
14278.
(35) Dotzauer, D. M.; Bhattacharjee, S.; Wen, Y.; Bruening, M. L. Langmuir 2009, 25,
1865.
(36)

Dotzauer, D. M.; Dai, J.; Sun, L.; Bruening, M. L. Nano Letters 2006, 6, 2268.

(37)

Bhattacharjee, S.; Bruening, M. Langmuir 2008, 24, 2916.

(38)

Kidambi, S.; Bruening, M. L. Chemistry of Materials 2005, 17, 301.

(39)

Cho, J.; Caruso, F. Chemistry of Materials 2005, 17, 4547.

(40) Ostrander, J. W.; Mamedov, A. A.; Kotov, N. A. Journal of the American Chemical
Society 2001, 123, 1101.
(41) Kidambi, S.; Dai, J.; Li, J.; Bruening, M. L. Journal of the American Chemical
Society 2004, 126, 2658.
(42)

Jiang, C.; Markutsya, S.; Tsukruk, V. V. Langmuir 2004, 20, 882.

(43) Joly, S.; Kane, R.; Radzilowski, L.; Wang, T.; Wu, A.; Cohen, R. E.; Thomas, E. L.;
Rubner, M. F. Langmuir 2000, 16, 1354.
(44)

Caruso, F.; Caruso, R. A.; Möhwald, H. Science 1998, 282, 1111.

(45) Datta, S.; Cecil, C.; Bhattacharyya, D. Industrial & Engineering Chemistry Research
2008, 47, 4586.
(46)

Schönhoff, M. Journal of Physics: Condensed Matter 2003, 15, R1781.

(47) Gheith, M. K.; Sinani, V. A.; Wicksted, J. P.; Matts, R. L.; Kotov, N. A. Advanced
Materials 2005, 17, 2663.
(48) Iojoiu, E. E.; Landrivon, E.; Ræder, H.; Torp, E. G.; Miachon, S.; Dalmon, J.-A.
Catalysis Today 2006, 118, 246.
71

(49) Iojoiu, E. E.; Miachon, S.; Landrivon, E.; Walmsley, J. C.; Ræder, H.; Dalmon, J.-A.
Applied Catalysis, B: Environmental 2007, 69, 196.
(50) Ræder, H.; Bredesen, R.; Crehan, G.; Miachon, S.; Dalmon, J.-A.; Pintar, A.; Levec,
J.; Torp, E. Separation and Purification Technology 2003, 32, 349.
(51) Iojoiu, E. E.; Walmsley, J. C.; Ræder, H.; Miachon, S.; Dalmon, J.-A. Catalysis
Today 2005, 104, 329.
(52)

Iojoiu, E. E.; Miachon, S.; Dalmon, J.-A. Topics in Catalysis 2005, 33, 135.

(53) Dalmon, J.-A.; Cruz-Lopez, A.; Farrusseng, D.; Guilhaume, N.; Iojoiu, E. E.; Jalibert,
J.-C.; Miachon, S.; Mirodatos, C.; Pantazidis, A.; Rebeilleau-Dassonneville, M.;
Schuurman, Y.; van Veen, A. C. Applied Catalysis, A: General 2007, 325, 198.
(54) Brugger, P. A.; Cuendet, P.; Gratzel, M. Journal of the American Chemical Society
1981, 103, 2923.
(55)

Rouaix, S. C., C.; Aimar, P. Journal of Membrane Science 2006, 277, 137.

(56)

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

(57)

Kotov, N. A. Nanostructured Materials 1999, 12, 789.

(58) Bidstrup, D. E.; Geankoplis, C. J. Journal of Chemical and Engineering Data 1963,
8, 170.
(59)

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

(60)

Fein, D. E.; Wachs, I. E. Journal of Catalysis 2002, 210, 241.

(61)

Palaikis, L.; Wieckowski, A. Catalysis Letters 1989, 3, 143.

(62) Fonseca, I.; Jiang, L.; Pletcher, D. Journal of the Electrochemical Society 1983,
130, 2187.
(63) IUPAC Solubility Data Series Volume 7: Oxygen and Ozone; Battino, R., Ed.;
Pergamon Press Inc: New York, 1981; Vol. 7.
(64)
135.

Kim, K. S.; Lee, K. H.; Cho, K.; Park, C. E. Journal of Membrane Science 2002, 199,

72

(65) Harmsen, J. M. A.; Jelemensky, L.; Van ANdel-Scheffer, P. J. M.; Kuster, B. F. M.;
Marin, G. B. Applied Catalysis, A: General 1997, 165, 499.
(66) Dutczaka, S. M.; Luiten-Oliemanb, M. W. J.; Zwijnenbergc, H. J.; BolhuisVersteegd, L. A. M.; Winnubstb, L.; Hempeniuse, M. A.; Benesa, N. A.; Wessling, M.;
Stamatialisd, D. Journal of Membrane Science 2011, 372, 182.

73

Chapter 3: Antifouling performance of a porous polypropylene coating in membrane
distillation

3.1 Introduction
To address the rapidly growing worldwide water demands discussed in Chapter 1, MD
provides an alternative to reverse osmosis for the purification of water from feeds containing
non-volatile contaminants. Whereas reverse osmosis utilizes high pressures (> 25 bar) to
selectively pass water through dense, non-porous polymeric films, direct contact membrane
distillation (DCMD) relies on a vapor pressure difference between two liquid phases separated
1

by a porous hydrophobic membrane. The vapor pressure difference usually arises due to a
temperature gradient across the membrane. Unlike traditional multistage flash distillation, MD
does not require heating the feed stream until boiling occurs. Theoretically, complete rejection
of non-volatile compounds occurs in MD because only water vapor passes through the
unwetted membrane. Additionally, the low temperature requirements of MD allow for
economic coupling to waste heat or solar energy. Through a combination of high separation
efficiency and low operating temperatures and pressures, MD may provide a competitive
option for water desalination and treatment.

2-6

MD is also attractive for dehydration of foods

and beverages, where low processing temperatures may preserve flavor and avoid
denaturation.

74

In all liquid-based membrane separations, and especially in MD, fouling control requires
constant attention. Membranes for DCMD must be hydrophobic to adequately separate the
aqueous feed and permeate streams and allow only vapor transport through the pores. The
buildup of amphiphilic foulants within MD membranes leads to progressive wetting of pores
7,8

and a crippling drop in solute rejection.

Several groups explored delaying humic acid fouling

of microporous membranes through increasing the hydrophilicity of the membrane.

9-13

However, such treatments are incompatible with the need for hydrophobic membranes in MD.
One previous study explored the addition of an oil- and water-repellant fluorocarbon polymer
emulsion to poly(vinylidine difluoride) (PVDF) polyester composite membranes and found
2

improved humic acid fouling resistance. Unfortunately, the water vapor flux values (0.08 g/m s)
14

were too low for practical application.

Fouling is an especially difficult challenge in food and beverage dehydration with MD.
Despite the many studies on potential MD applications in food production,

15-19

only a few of

these looked at dairy processes within the last decade, presumably because of fouling
20-22

challenges.

For example, application of MD to dehydration of whey solutions will only be

practical with improved resistance to fouling by the feed solution fats and amphiphilic whey
proteins. The food and beverage industry formerly considered whey a waste product of cheese
production, but now it is a common food additive and even contributes to ethanol production
through lactose fermentation.

21

The recovery of whey proteins traditionally includes

75

concentration of dissolved solids to the osmotic limit using ultrafiltration, diafiltration to
remove dissolved sugars and salts, and thermal evaporation and spray drying to reach a final
whey protein product.

20

The development of a membrane coating that resists whey solution

fouling could expand applications of MD in the dairy industry.
23

We hypothesize that superhydrophobic films on surfaces
afford some fouling resistance due to increased liquid slip

25-27

24

and membranes

may

and decreased contact area with

the membrane. Liquid slip increases due to air plastrons retained within the rough features of
the coating (Figure 3.1). The air plastrons also reduce the membrane surface area available for
foulant deposition.

Figure 3.1: Illustration showing the increase of apparent water contact angle on a rough surface
due to the presence of trapped air.

This chapter describes the application of rough PP films as superhydrophobic, foulingreducing coatings on microporous PVDF membranes. We explore the effect of the thin porous
PP coating on vapor flux and salt rejection during DCMD of a model humic acid fouling feed
76

solution and a dairy whey solution. The porous PP coatings demonstrate improved fouling
resistance as they provide on average (n=3) >99% salt rejection after 20 hours of operation for
humic acid solutions compared to untreated membranes which only reject 95% on average
(n=5). These coatings also help prevent pore wetting by whey solutions. During whey filtration,
the time required for fouling-induced pore wetting and salt breakthrough increases by an order
of magnitude when using membranes coated with PP.

3.2 Experimental
3.2.1 Materials
Isotactic PP pellets (Mw=190,000 Mn=50,000), humic acid sodium salt, and calcium
chloride dihydrate (99%) were obtained from Sigma-Aldrich. Sodium chloride (99%),
hydrochloric acid (38%), and methylethylketone (MEK) (99%) were purchased from Columbus
Chemical Industries, and p-xylene (99.6%) was obtained from Fisher Scientific. These reagents
were used as received. Whey solution was obtained after removing casein from store-bought
whole milk using dilute citric acid and calf rennet. Whey was filtered through Whatman #1 filter
paper to remove suspended solids prior to use. HVHP PVDF membranes (25 mm diameter, 0.45
µm nominal pore size) were purchased from Millipore (#HVHP02500), and 25 mm diameter 0.2
µm nominal pore size anodized alumina membranes were obtained from Whatman (#68096022). Deionized water (Milli-Q purification system, 18.2 MΩ·cm) was used for rinsing the
membranes and preparation of the fouling and permeate solutions. The pH of the humic acid
solutions was adjusted with 1 N HCl.
77

3.2.2 Deposition of crystalline polypropylene films
23,24,28

Deposition of superhydrophobic PP films was based on literature procedures.

Isotactic PP (72 mg) was added to 4 mL of stirred p-xylene in a 25 mL round bottom flask. The
flask was heated to 130 ◦C to dissolve the PP. Once the solution was homogeneous, 1.6 mL of
warm MEK was added as a porogen. The MEK non-solvent induces PP crystal nucleation while
also creating voids that lead to pores in cured films. Enough of the PP dispersion to evenly coat
the membrane surface (~0.4 mL) was then deposited onto flat sheet membranes before
spinning at 400 rpm for 30 s to remove excess solution. After the coating was deposited on the
surface of the membrane, an ethanol rinse using approximately 10 mL sprayed from a wash
bottle was sometimes used to remove excess PP from the membrane surface to create a
thinner film. Finally, the deposited PP films were cured in a vacuum oven at 70 ◦C and 1 bar
vacuum for 3 hours before further use.

3.2.3 Membrane characterization
Both treated and bare membranes were characterized by their liquid entry pressure as
well as their apparent water contact angle. To assess the liquid entry pressures, each
membrane was placed atop a porous stainless steel frit serving as a mechanical support within
a modified Amicon 25 mm cell. The base of the Amicon cell was hollowed out to allow
incorporation of a stainless steel frit to support the polymeric membrane under pressure. The
cells were filled with water, and the pressure above the membranes was slowly increased until
78

water permeation was observed. Contact angles were determined with deionized water and a
FTA200 Dynamic Contact Angle Analyzer system using an advancing droplet rate of 0.1 mL/min.
SEM images of membranes before and after modification were obtained using a Hitachi S-4700
II field-emission scanning electron microscope (SEM) operated at an accelerating voltage of 15
keV. Deposition of 8 nm of sputtered gold rendered the samples conductive for imaging. Due to
difficulties in obtaining a fracture suitable for cross-sectional imaging of PVDF membranes, PP
coatings were also applied to alumina membranes that fracture with retention of pores. The
film morphology is visually similar on PVDF and alumina membranes based on SEM images.

3.2.4 DCMD
Membranes were placed in a custom built holder that allows cross flow of both feed and
permeate streams (Figure 3.2). The feed model for a brackish fouling solution contained 100
2+

mg/L humic acid, 0.1 M NaCl, and 3.8 mM Ca

at pH 4. Previous studies showed that this
9,10,13

solution rapidly fouls polymeric polyamide nanofiltration and PVDF MD membranes.

Calcium ions complex with humic molecules to cause aggregation that favors deposition on
membrane surfaces. Additionally, the pKa of the carboxylic groups in humic acids is around 4,
so the low feed pH reduces the humic acid charge state to increase fouling rates. The feed
solution (2 L for humic acid, 500 mL for whey) was placed in a stainless steel tank and circulated
through heat exchange coils within a warm water bath set to 70 ◦C (humic) or 50 ◦C (whey) for
MD (Figure 3.3). The permeate reservoir, which initially contained distilled water (200 mL), was
79

placed on an analytical balance that was connected to a lab PC via LabView to record the
increase in mass due to condensation of vapor flux through the membrane. The permeate
solution was pumped through a chiller to cool the stream to 10 ◦C prior to passing over the
membrane support and returning to the reservoir. Both the feed and permeate streams were
recycled through their respective flow loops for at least one hour at approximately 30 mL/min
prior to introduction to the membrane to ensure accurate temperature control. The
temperatures of feed solutions entering and exiting the membrane were measured and found
to differ by only 3 ◦C on average with a feed temperature of 70 ◦C.

Figure 3.2: Cross-sectional diagram of the custom-built crossflow membrane holder. The
membrane is sandwiched between the two halves of the holder and sits on a porous stainless
steel frit that provides mechanical support. Warm feed solution enters the cell and travels
across the skin of the membrane before exiting the module as retentate returning to the feed
reservoir. Likewise, the cool permeate stream enters the other half of the module, passes along
the stainless steel frit, and exits the module.

80

Figure 3.3: Diagram of the DCMD apparatus. Both feed and permeate streams pass through
heat-exchange coils to achieve their desired temperatures prior to entry into the membrane
module. The feed retentate is recycled to the feed reservoir, while water vapor diffuses through
the membrane and condenses in the permeate stream. The analytical balance monitors
increases in the mass in the permeate reservoir.

A home-built inline flow-through conductivity cell was also incorporated in the
permeate stream immediately after the membrane module to monitor increases in conductivity
due to solute passage using an Orion Model 115A+ conductivity meter. Real-time inline
monitoring allows us to observe changes in conductivity with time to monitor membrane salt
rejection. At the start of each trial, the permeate solution conductivity was negligible (< 5
µS/cm). Extracted feed samples from before and after MD were also analyzed through dilution
of 1 mL aliquots with 6 mL of distilled water before vortex mixing. Conductivity values were
measured in triplicate for each sample. To account for the nonzero permeate volume at the
beginning of the DCMD experiments, a dilution factor (α) was calculated by dividing the initial
volume of water by the volume of water which passed through the membrane during the trial.
Utilizing the change in conductivity of the permeate solution from before and after each

81

distillation trial (ΔCP), the conductivity of the feed solution (CF) samples, and α, the rejection
was calculated from the equation 3.1.

(

)

(3.1)

3.3 Results and discussion
3.3.1 Membrane modification with polypropylene films
Deposition of PP films on 0.45 µm PVDF membranes causes no visually apparent
changes. However, SEM images obtained before (Figure 3.4 A) and after modification (Figure
3.4 B,C) show the presence of PP in a beads-on-a-string morphology. For thicker films deposited
without the ethanol rinse step (Figure 3.4 B), surface porosity decreases and the PP layer
retains only minimal roughness. However, the films deposited with the ethanol rinse step
(Figure 3.4 C) show minimally reduced porosity and high PP roughness, which should provide
relatively open pores that can trap air plastrons. Films prepared without an ethanol rinse have a
heterogeneous average thickness of 4.8 ± 1.8 µm with many peaks and valleys (Figure 3.5 A).
Incorporation of the ethanol rinse reduced the average thickness to 1.8 ± 0.4 µm and improved
homogeneity (Figure 3.5 B). Thinner, more uniform films are desirable to reduce fouling
without greatly increasing mass-transport resistance in MD. All DCMD experiments with PPmodified membranes employed coatings prepared with the ethanol rinse step to minimize the
reduction in flux due to the coating.

82

Figure 3.4: SEM images of 0.45 µm PVDF membranes before (A) and after (B,C) modification
with porous PP. For the membrane in image (C), the coating was rinsed with ethanol prior to
heating in vacuum.

Figure 3.5: SEM images of cross-sections of 0.2 µm alumina membranes that were modified
with thick or thin PP films. Image A shows the thick PP coating, which was prepared without an
ethanol rinse, and image B shows the PP coating whose deposition included ethanol rinsing.

The advancing water contact angle increases from 124 ± 1◦ on the bare PVDF
membranes to 154 ± 1◦ and 157 ± 3◦ for membranes coated with the thick and thin PP films,
respectively. The increased roughness imparted by the film allows air entrapment at the surface
to decrease the polymer-water contact area and increase contact angles. The liquid entry
pressure of bare HVHP membranes was around 1.1 bar in accordance with previous studies.29,30
83

Addition of either the thin or thick PP coating increases the liquid entry pressure to ~2 bar.
Equation 3.2 gives the theoretical expression for liquid entry pressure, ΔP, where γ is the
surface tension of an air-water interface (72.8 mN/m at 25 ◦C), θ is the water contact angle
with the membrane, and r is the membrane pore radius.

(3.2)

Based on this equation, an increase in contact angle from 120◦ to 160◦ without altering pore
radius causes a nearly two-fold increase in the wetting pressure. The SEM images in Figure 3.4
suggest that pore size is approximately the same in coated and bare membranes, thus the twofold increase in wetting pressure is consistent with contact angles and SEM images.

3.3.2 Humic acid DCMD
3.3.2.1 Bare PVDF membranes
Control experiments with PVDF membranes provide a baseline for examining whether
PP-coated membranes increase fouling resistance. During filtration of feed solutions containing
humic acid through bare membranes, increases in conductivity demonstrate substantial feed
solution breakthrough within 200 minutes of operation (Figure 3.6 B) for two membranes. The
buildup of hydrophilic humic acid deposits on the membrane likely allows progressive entrance
of the feed solution into the pores and causes decreased solute rejection. In three other
identical bare membranes tested, limited salt passage occurred (99.94 ± 0.05% average
84

rejection) but flux declined 31 ± 18% over 20 hours due to pore coverage by dissolved humic
components. The differences in humic acid interactions with the bare membrane surface likely
arise from the high complexity and compositional variability of humic acid.

31

Despite using a

standard humic acid powder (Sigma-Aldrich), differences in humic solutions were visible and
probably lead to the variable effects humic acid fouling on bare membranes. Nevertheless, both
salt passage and flux decline are undesirable effects of fouling. For all five bare membranes, the
average flux before breakthrough (as suggested by permeate conductivity increase) was 8.7 ±
2

0.9 g/m s, and the overall average rejection was 95.5 ± 6.3% based on conductivity
measurements at the end of each distillation. The untreated, hydrophobic surface of PVDF
provided a favorable location for foulant adsorption, which lead to rapid membrane failure by
pore wetting (in the two membranes where breakthrough occurred) or cake layer formation.
The initial flux was ~15% less than that through the same membrane with distilled water as the
2

feed (~10 g/m s). The ~15% decrease is much greater than the < 1% that Raoult’s Law would
predict in reduced vapor pressure from salt addition. The extra flux decline likely arises from
concentration polarization at the membrane surface. This concentration polarization not only
will reduce the vapor pressure of water locally at the membrane surface, but may also lead to
initial rapid fouling of the membrane skin to reduce mass transport.

85

Figure 3.6: Change in permeate mass (lines) and conductivity (triangles) over time during DCMD
of humic acid solutions through five bare PVDF membranes. (A) For three membranes, flux
declines 31 ± 18% over 20 hours but salt rejection remains at 99.94 ±0.05%. (B) Two other
membranes show significant increases in permeate conductivity after an initially stable region.
The conductivity spike in (B) suggests entrance of the solution into some membranes pores
within about 200 min of operation. The breakthrough leads to an overall 88.9 ± 2.0% salt
rejection and negligible flux decline.

86

Figure 3.6 (cont’d)

3.3.2.2 Membranes with thin PP coatings
We hypothesized that thin, porous PP coatings will resist fouling due to their
superhydrophobicity (trapped air plastrons) without greatly decreasing flux relative to an
untreated membrane. As Figure 3.7 shows, treated membranes showed essentially no increase
in permeate conductivity for 20 h of filtration of a humic acid solution. The average flux over
2

the first 100 minutes was 4.9 ± 1.8 g/m s, and the overall salt rejection based on conductivity
measurements was 99.9 ± 0.01%. The slightly elevated (but constant) conductivity values in the
87

trial 3 resulted from traces of salt in the permeate solution at the start this experiment. Relative
to bare membranes, the flux was ~45% lower for membranes with thin PP coatings, presumably
due to blockage of PVDF membrane pores during PP film deposition. While lower flux is
undesirable, the stable salt rejection and wetting resistance over 20 hours make PP-coated
membranes attractive for MD with highly fouling solutions. Further reductions in the PP film
thickness through increasing the rotation rate during spin coating may provide similar rejection
while reducing the flux lost to modification.

Figure 3.7: Change in permeate mass (lines) and conductivity (triangles) over time during DCMD
of humic acid solutions through 3 PVDF membranes with thin PP coatings. The constant
conductivity values indicate 99.9% salt rejection at a flux 45% lower than that through an
untreated membrane.
88

3.3.3 Whey DCMD
3.3.3.1 Bare PVDF membranes
As with studies of humic acid fouling, in DCMD with whey solutions we first assessed the
fouling properties of bare PVDF membranes. In whey solutions, these membranes
demonstrated little resistance to fouling (Figure 3.8). Within 200 min, permeates from all three
tested membranes showed increases in conductivity and mass flux through the membranes
declined significantly. In fact, Figure 3.8 shows mass flux from the permeate to the feed after
times as short as 150 min for two of the trials. Most likely, the pores wetted due to fouling and
allowed osmotic water transport from the permeate to the feed. The eventual recovery of
positive mass flux in the second membrane likely resulted from a decrease in osmotic pressure
with time and thermally induced water transport from feed to permeate. In the case of a third
membrane, osmotic flow mass flux always flowed from feed to permeate, but the permeate
conductivity increased rapidly. Overall, bare membranes showed an average initial flux of 2.8 ±
2

0.8 g/m s for the first 100 minutes, but salt rejections cannot be calculated for the first two
trials due to the presumed presence of both MD and osmotic flow preventing us from
establishing a realistic dilution factor. However, the third trial showed only a ~40% salt rejection
which is likely in line with the true value for the first two trials. Nonetheless, the bare
membranes are clearly unsuitable for DCMD of whey solutions under the tested conditions.
Visual inspection of bare membranes after whey MD revealed pore wetting from fats or
proteins (Figure 3.9).

89

Figure 3.8: Change in permeate mass (lines) and conductivity (triangles) over time during whey
DCMD through 3 bare PVDF membranes. All membranes show an increase in permeate
conductivity within the first 400 min, suggesting pore wetting. The reversal of mass flow in two
of the trials likely results from osmotic flow from permeate to feed through wetted pores.

Figure 3.9: Untreated PVDF membrane (feed surface) after DCMD of whey solution.
90

3.3.3.2 Membranes with thin PP coatings
Contrary to observations with bare PVDF membranes, DCMD of whey solutions using
four membranes coated with thin PP films shows only minimal salt breakthrough over 20 hours
2

(Figure 3.10). The average initial flux (100 minutes) was 2.5 ± 0.5 g/m s, and the average
rejection over the entire experiment was 96.5 ± 4.6%. For two trials, the permeate conductivity
increased negligibly (<3 µS/cm) over time, suggesting negligible pore wetting. Moreover, the
flux through these membranes declined <55% over the 20-h filtration. In contrast, with two
other replicate trials, significant surface fouling from whey components lead to a drastic (~
99%) decline in flux. However, unlike bare membranes, despite the high amount of whey
fouling, conductivity increased only mildly (<30 µS/cm). Additionally, the presence of the
antifouling PP layer may allow for easier cleaning of adsorbed whey components to restore flux
values. The initial flux is only 10% less than that through the untreated membrane, and the
resistance to fouling and pore wetting over 20 h will allow for removal of more than an order of
magnitude more water than DCMD with the bare membranes. Examination of the membrane
feed surface after DCMD showed minimal signs of pore wetting, further supporting the notion
that only a small amount of pores leaked during the trial (Figure 3.11). The improved whey
fouling resistance to pore wetting provided by the PP coating may allow for further membrane
use in the dairy industry for whey concentration. Nevertheless, the variable membrane
performance observed through the four trials in this study likely demonstrates the need to
improve whey pretreatment procedures to reduce deviation in the whey solution composition.

91

Figure 3.10: Change in permeate mass (lines) and conductivity (triangles) over time during
DCMD of whey solutions through 4 PVDF membranes with thin PP coatings. High rejection (>
95%) results in all four trials with an initial flux (100 minutes) only 10% lower than through an
untreated membrane.

Figure 3.11: PP coated PVDF membrane (feed surface) after DCMD of whey solution.
92

3.4 Conclusions
Deposition of porous PP coatings on PVDF membranes improves fouling resistance to
avoid pore wetting during DCMD of humic acid and whey solutions. Both whey and humic acid
solutions caused severe fouling of most untreated PVDF membranes within the first several
hours of DCMD. Treated membranes resisted fouling to provide stable rejections for at least 20
hours. Due to fouling susceptibility, in two out of 5 cases bare membranes rejected only 88.9 ±
2.0% of feed salts from humic acid solutions and were unstable in whey solutions. Coating
membranes with PP exhibited salt rejections of 99.9 ± 0.01% and 96.5 ± 4.6% for humic acid
and whey solutions.
PP-treated membranes allowed fluxes 10-45% lower than those through bare
membranes, presumably due to the additional mass transport resistance of the PP layer. While
the PP coatings generally produce porous films with large (0.5-5 µm) pore diameters, the film
morphology is spongy and likely blocks some PVDF membrane pores. Further optimization of PP
deposition conditions may reduce the flux decline from coatings, but some decline will always
occur. Even with the small flux decline, however, the greatly improved membrane performance
in extreme environments makes these membranes attractive for some MD applications.

93

REFERENCES

94

REFERENCES

(1)
Younos, T.; Tulou, K. E. Journal of Contemporary Water Research and Education
2005, 132, 10.
(2)

Alkhudhiri, A.; Darwish, N.; Hilal, N. Desalination 2012, 287, 2.

(3)

Pangarkar, B. L.; Sane, M. G.; Guddad, M. ISRN Materials Science 2011, 2011, 1.

(4)
El-Bourawi, M. S.; Ding, Z.; Ma, R.; Khayet, M. Journal of Membrane Science
2006, 285, 4.
(5)

Curcio, E.; Drioli, E. Separation & Purification Reviews 2005, 34, 35.

(6)

Lawson, K. W.; Lloyd, D. R. Journal of Membrane Science 1997, 124, 1.

(7)

Gryta, M. Separation Science and Technology 1996, 41, 1789.

(8)
Tun, C. M.; Fane, A. G.; Matheickal, J. T.; Sheikholeslami, R. Journal of Membrane
Science 2005, 257, 144.
(9)

Hong, S.; Elimelech, M. Journal of Membrane Science 1997, 132, 159.

(10) Asatekin, A.; Kang, S.; Elimelech, M.; Mayes, A. M. Journal of Membrane Science
2007, 298, 136.
(11) Hamid, N. A. A.; Ismail, A. F.; Matsuura, T.; Zularisam, A. W.; Lau, W. J.; Yuliwati,
E.; Abdullah, M. S. Desalination 2011, 273, 85.
(12) Abuhabib, A. A.; Mohammad, A. W.; Hilal, N.; Rahman, R. A.; Shafie, A. H.
Desalination 2012, 295, 16.

95

(13) Srisurichan, S.; Jiraratananon, R.; Fane, A. Journal of Membrane Science 2006,
277, 186.
(14)

Huo, R.; Gu, Z.; Zuo, K.; Zhao, G. Journal of Applied Polymer Science 2010, n/a.

(15)

Gunko, S.; Verbych, S.; Bryk, M.; Hilal, N. Desalination 2006, 190, 117.

(16) Madaeni, S. S.; Sasanihoma, A.; Zereshki, S. Journal of Food Process Engineering
2011, 34, 1535.
(17) Onsekizoglu, P.; Bahceci, K. S.; Acar, M. J. Journal of Membrane Science 2010,
352, 160.
(18)

Bélafi-Bakó, K.; Koroknai, B. Journal of Membrane Science 2006, 269, 187.

(19)

Cojocaru, C.; Khayet, M. Separation and Purification Technology 2011, 81, 12.

(20) Christensen, K.; Andresen, R.; Tandskov, I.; Norddahl, B.; du Preez, J. H.
Desalination 2006, 200, 523.
(21) Tomaszewska, M.; Białończyk, L. Desalination and Water Treatment 2013, 51,
2362.
(22) Hausmann, A.; Sanciolo, P.; Vasiljevic, T.; Ponnampalam, E.; Quispe-Chavez, N.;
Weeks, M.; Duke, M. Membranes 2011, 1, 48.
(23)

Erbil, H. Y.; Demirel, A. L.; Avci, Y.; Mert, O. Science 2003, 299, 1377.

(24) Franco, J. A.; Kentish, S. E.; Perera, J. M.; Stevens, G. W. Journal of Membrane
Science 2008, 318, 107.
(25) Aljallis, E.; Sarshar, M. A.; Datla, R.; Sikka, V.; Jones, A.; Choi, C.-H. Physics of
Fluids 2013, 25, 025103/1.
(26)

Lee, D. J.; Cho, K. Y.; Jang, S.; Song, Y. S.; Youn, J. R. Langmuir 2012, 28, 10488.
96

(27)

McHale, G.; Newton, M. I.; Shirtcliffe, N. J. Soft Matter 2010, 6, 714.

(28)

Lv, Y.; Yu, X.; Jia, J.; Tu, S.-T.; Yan, J.; Dahlquist, E. Applied Energy 2012, 90, 167.

(29) Khayet, M.; Matsuura, T. Membrane Distillation: Principles and Applications;
Elsevier: Amsterdam, 2011.
(30) García-Payo, M. C.; Izquierdo-Gil, M. A.; Fernández-Pineda, C. Journal of Colloid
and Interface Science 2000, 230, 420.
(31)

Hayes, M. H. B.; Clapp, C. E. Soil Science 2001, 166, 723.

97

Chapter 4: Preliminary investigation of anionic polyelectrolyte multilayer films as coatings
that resist fouling by negatively charged oil drops in oil-in-water emulsions

4.1 Introduction
Rising global energy demands are spawning new methods to extract previously
unavailable natural resources. In particular, hydraulic fracturing is one of the most significant
developments in energy production over the last 15 years. As discussed in Chapter 1, when
coupled with horizontal drilling practices, hydraulic fracturing greatly improves oil and natural
1

gas recovery from difficult formations. However, hydraulic fracturing employs large quantities
of water to crack rock formations and release oil. In fact, in 2007 88% of all liquid volumes
collected at wells within the United States consisted of contaminated water, known as
produced water. Additionally, the ratio of oil recovered to produced water declines with well
age to such an extent that produced water increases up to 98% of all the collected volume near
2

the end of a well’s life.

Although current industry practices include reusing recovered produced water to
continue extracting oil, eventually the water accumulates too much contamination for further
use. Some companies choose to pump the highly contaminated water back underground into
non-producing formations for storage, but the potential risk for environmental contamination is
2

high. Hence, other companies attempt to decontaminate produced water through a complex
multistep process to reduce possible environmental impacts. Generally, produced water
98

treatment first combines physical, chemical, and biological methods to de-oil the water before
removing soluble organics and disinfecting. Subsequent treatment steps include removal of
suspended solids, release of dissolved gasses, and finally desalination and water softening to
remove excess salts and produce a water stream suitable for irrigation of crops, dust control,
vehicle washing, power plant makeup water, and fire control systems.

3,4

Current United States regulations govern the maximum offshore discharge of oil and
grease at a daily maximum of 42 mg/L and a monthly average of less than 30 mg/L. Onshore
regulations generally prohibit produced water discharge unless for direct agricultural or wildlife
5

use with oil and grease levels below 35 mg/L and low salinity. Regulations do not govern the
salinity of offshore produced water discharge into the seas. Unfortunately, coagulation,
flocculation, air flotation, gravity separation, and other common municipal water treatment
6

procedures do not meet these discharge limits for produced water. Membranes have emerged
7-10

as an attractive treatment option in a number of studies
11

water and require minimal space.

because they produce high quality

However, direct membrane exposure to produced water

generally causes rapid flux declines due to surface and pore fouling by rejected oil and grease
components. Furthermore, despite the wide range of treatment options currently available, the
removal of small (< 10 µm), well dispersed oil droplets remains limited.

4

To reduce oil fouling of membranes, many groups explored new membrane coatings
and formulations aimed to increase hydrophilicity.

99

2,10,12-14

Water adsorption resulting from

increased hydrophilicity restricts oil interactions with the wetted surface (Figure 4.1). The
incorporation of polyethylene glycol polymer chains alone or among copolymer blends
represents a common method to increase hydrophilicity. High chain density combined with film
swelling upon exposure to water leads to the innate antifouling ability of polyethylene
15,16

glycol.

In a less common method, Tuteja and coworkers reduced the surface energy on

membranes through addition of fluorinated polyhedral oligomeric silsequioxane features
combined with a unique surface reentrant geometry tailored to generate omniphobic surfaces
by producing a metastable Cassie-Baxter state even for liquids with low surface tensions such
7,17,18

as oils.

While increasing hydrophilicity may reduce membrane fouling in produced water

treatment, unless pore sizes are small such membrane will still not reject small oil particles.
Omniphobic membranes that allow water passage while preventing oil transport are attractive
but current systems have limited throughput and only function under precisely controlled
conditions (<0.07 bar feed pressure).

100

Figure 4.1: Membrane fouling without (A) and with (B) hydrophilic modifications.

Improved membrane filtration of oil-in-water emulsions will require a simple, cost
effective surface modification or new membrane material. We propose exploiting the
antifouling abilities of hydrophilic surfaces formed through adsorption of highly charged
polyelectrolyte multilayers (PEMs). Counterintuitive to current procedures, we also propose the
addition of either anionic or cationic surfactants to feed solutions. Naturally, the inclusion of
surfactants will reduce the interfacial energy between oil and water to stabilize droplets and
reduce coalescence. Traditionally, rejection of oil in membrane processes relies on size-based
exclusion of oil droplets at the membrane surface, which leads to a local increase in oil
concentration. As the droplets become more concentrated at the membrane surface,
coalescence occurs and the average droplet size increases (Figure 4.2A). In our method, the
surface charge of the PEM deposited on the membrane will match the charge of the surfactant
used to stabilize the oil. The similar charge on droplets and the membrane surface should lead
101

to electrostatic repulsion of the droplets to help prevent them from reaching the membrane
(Figure 4.2B). Furthermore, in contrast to traditional methods, oil droplets with diameters
smaller than membrane pores may still undergo rejection due to electrostatic exclusion. We
think the combination of this electrostatic rejection with the antifouling abilities of extremely
hydrophilic surfaces will lead to coatings with superior performance for removal of oil from
water during water polishing steps. This chapter focuses on the deposition of polyanionterminated PEMs to rejection anionic emulsion droplets. Simultaneous work in our laboratory is
examining polycation-terminated PEMs and cationic emulsions.

Figure 4.2: Oil-membrane interactions in traditional size exclusion (A) and the proposed
electrostatic exclusion (B) systems.

102

4.2 Experimental
4.2.1 Materials
Poly(sodium 4-styrenesulfonate) (PSS, Mw = 70,000 Da), poly(allylamine
hydrochloride) (PAH, Mw = 56,000 Da), hexadecane, cetrimonium bromide (CTAB), sodium
dodecyl sulfate (SDS), and Triton X-100 (Triton) were used as received from Sigma-Aldrich.
Sodium chloride (99%, CCI), manganese chloride dihydrate (Alfa Aesar), sodium bromide (Jade
Scientific), dichloromethane (DCM, Mallinckrodt), decane (Alfa Aesar), phosphoric acid (JT
Baker), and formaldehyde (CCI) were also used as received. Deionized water (Milli-Q
purification system, 18.2 MΩ·cm) was used for rinsing the membranes and preparation of the
polyelectrolyte solutions and oil-in-water emulsions. The pH values of the polyelectrolyte
solutions were adjusted with dilute solutions of HCl and NaOH. Nylon microfiltration
membranes with a nominal pore size of 5.0 µm and a diameter of 25 mm were obtained from
GE Osmonics (#R50SP02500) and Sterlitech (#NY5025100). Nylon microfiltration membranes
with a nominal pore size of 0.45 µm and a diameter of 25 mm were also obtained from
Millipore (#HNWP02500).

4.2.2 Modification of nylon membranes
Nylon membranes were immersed in a solution containing

phosphoric acid in a

formalin solution (50 mL formaldehyde and 1 mL 85% phosphoric acid) at 60 ◦C to introduce
hydroxyl and anionic surface groups.

19

During modification with PEMs, the membranes were
103

placed in a dead-end holder similar to an Amicon cell. First, a 10 mL solution containing 0.2 M
PAH, 0.5 M NaBr at pH 2.3 was circulated through the membrane pores using a peristaltic pump
at 3 mL/min for 5 minutes. Next the membrane was rinsed thoroughly with water at 5 mL/min
to remove and unbound or weakly bound PAH. PSS adsorption then occurred during circulation
of 10 mL of 0.2 M PSS, 0.5 M MnCl2 at pH 2.1 for two minutes at 3 mL/min followed by
subsequent rinsing again at 5 mL/min. Additional bilayers were adsorbed in the same way to
obtain films with 4, 6, or 8 PAH/PSS layers. Based on our prior work, polyelectrolyte supporting
20

salts and deposition pH values were selected to produce charged surfaces.

4.2.3 Membrane characterization
SEM images of nylon membranes before and after modification with 4, 6, or 8 PAH/PSS
bilayers were obtained using a Hitachi S-4700 II field-emission scanning electron microscope.
Deposition of 8 nm of sputtered gold using a Pelco SC-7 auto sputter coater rendered the
samples conductive for imaging at 15 keV. Additionally, the water permeability of each
membrane was tested at a range of pressures (0-1.4 bar) before and after LbL deposition to
qualitatively observe polymer film growth.

4.2.4 Emulsion preparation
Three different emulsion solutions were prepared to produce droplets with cationic,
anionic, or neutral charge, depending on the surfactant used to stabilize the droplets. In all
104

cases, a 10% (v/v) solution of hexadecane in 0.01 M NaCl was prepared with the appropriate
surfactant at 5 mM. Hexadecane was first saturated with Oil-Red-O dye (Sigma-Aldrich) by
adding several milligrams of dye to 50 mL of hexadecane until no more could be dissolved. To
disperse the oil, each solution was sonicated using a Misonix ultrasonic liquid processor (3000
Series) with active sonication for a total of 2 minutes consisting of four 30 s periods at power
level 8.0 separated by three 10 s periods without sonication. The droplet size distribution was
determined with dynamic light scattering using a Malvern Mastersizer (Hydro 2000SM) to
ensure that the majority of droplets were below 5.0 µm in diameter. Additionally, droplet
charge was measured using a Brookhaven Instruments ZetaPals Electro Kinetic Analyzer system.

4.2.5 Electrostatic oil rejection from large pores
Wet membranes containing 0, 4, 6, or 8 polyelectrolyte bilayers were first placed inside
a Kimwipe to remove excess surface moisture while still leaving the film hydrated. The
membranes were then placed into a holder (similar to an Amicon cell) equipped with a
suspended magnetic stir bar to ensure feed solution homogeneity. Emulsion solutions were
shaken immediately prior to use just like before particle size determination to redistribute any
settled of oil drops. One mL of the feed solution was removed and stored in a test tube for later
analysis by gas chromatography (GC). Next, 5 mL of the oil emulsion was added to the stirred
membrane holder. The pressure was then increased to 0.68 bar in a dead-end mode and the
permeate was collected. Permeate collection continued for 10 minutes or until all the feed
solution passed through the membrane. Any remaining solution in the membrane holder was
105

then collected and stored for GC analysis. After each dead-end filtration, membranes were
rinsed with isopropanol (to remove any adsorbed oil films) followed by water. Furthermore,
membranes were stored immersed in water to maintain hydration between trials. The same
procedure was followed to assess the rejection of 20-fold diluted oil emulsions to determine
whether oil concentration influenced droplet rejection.
Analysis of feed, permeate, and retentate aliquots included oil extraction and GC. One
mL of each sample (feed, permeate, or retentate if present) was placed into a small test tube,
and 10 drops of 1 M HCl were added to Triton- and SDS-stabilized emulsions. Due to the
enhanced stability provided by charged surfactants, 1 mL of NaCl saturated brine was added to
the SDS emulsions to break the emulsion through screening the surfactant charge. All solutions
were then briefly (~10 s) vortex mixed to ensure incorporation of salt and acid throughout the
sample. Next, 1 mL of DCM was added to each sample before vortex mixing an additional 30
seconds to extract hexadecane into the organic layer. The layers separated within 1 minute,
and 0.5 mL of each DCM layer was removed and placed in a GC vial along with 10 µL of decane
(internal standard). Samples were then analyzed with a Shimadzu GC-17a gas chromatograph
using the parameters from Table 4.1.

106

Table 4.1: GC analysis parameters.
Parameter

Setting

Column
Initial Temperature

Restek RTX-5 (15 m)

Ramp Rate

35 ◦C/min

70 ◦C for 1 minute

Final Temperature

310 ◦C
1 µL
Splitless

Injection Volume
Injection

Areas were obtained for all hexadecane peaks and normalized by dividing by the area of
each decane peak to remove variations in sample injection and detection. Rejections, R, were
determined based on the normalized hexadecane peak areas from the permeate (AP) and feed
(AF) solutions (equation 4.1).

(4.1)

Similarly, hexadecane enrichment, E, in retentate solutions was calculated based on the
normalized hexadecane peak areas from the retentate (AR) and feed (AF) solutions (equation
4.2).

(4.2)

107

4.2.6 Microfiltration flux decline assessment
Similar to oil rejection experiments, the surfaces of wet membranes were dried with
Kimwipes before placing the membrane in a custom built cross-flow holder (Figure 4.3). Next,
10 mL of the 10% hexadecane emulsion with 5 mM SDS was added to 500 mL of water and
vigorously stirred to create a feed solution. The feed solution was placed into a feed tank
connected to a centrifugal pump (Figure 4.4). A 500-mL Erlenmeyer flask aligned to receive the
permeate was placed on top of an analytical balance connected to a computer via LabView
software to capture mass values every 5 seconds. Once mass collection began, the feed
solution was pumped across the membrane surface at 80 mL/min under an applied pressure of
2 bar, and flux was recorded over a period of 2 or more hours.

Figure 4.3: Diagram of the cross-section of a custom built cross-flow membrane holder. The
membrane is sandwiched between the two halves of the holder atop a porous stainless steel
frit that serves as a mechanical support.

108

Figure 4.4: Scheme of the microfiltration apparatus.

4.3 Results and discussion
4.3.1 Membrane modification with PSS and PAH films
Adsorption of highly swollen PAH/PSS films in membrane pores significantly reduces
membrane permeability. With a constant transmembrane pressure, flux declines after
modification are 22 ± 14%, 58 ± 16% and 73 ± 18% after adsorption of 4, 6, and 8 PAH/PSS
bilayers (uncertainties are standard deviation and n = 4). This decline in flux suggests pore
blockage or restriction by the deposited polyelectrolyte films. For the proposed study, pores
need to remain large (~ 5.0 µm) to assess the membrane’s ability to reject small oil droplets
(<5.0 µm) electrostatically without size exclusion. Hence, film growth should ideally not
constrict pore size. Previous ellipsometric measurements and SEM measurements of similar 4.5
bilayers PSS/PAH films deposited on gold wafers and alumina membranes showed a film
109

thickness of only 20 nm.

20

Thus, the large flux decline observed for the 6 and 8 bilayer modified

membranes was unexpected. Flow through the membrane may trap some polyelectrolyte to
give thicker films.
SEM images of the 5.0 m nylon membranes coated with 0, 4, 6, and 8 PAH/PSS bilayers
(Figure 4.5) show no obvious blocking or coverage of membrane pores after film deposition.
Film swelling could lead to a reduction in pore size, but previous studies showed that [PSS/PAH]
films barely swell (at most by ~120%) upon exposure to salt solutions.

21

Therefore, the high flux

decline most likely results from blocking of small pores within the heterogeneous interior of the
spongy membrane.

110

Figure 4.5: SEM images of the top face of 5.0 µm nylon membranes before (A), and after
coating with [PAH/PSS]4 (B), [PAH/PSS]6 (C), and [PAH/PSS]8 (D) films. The images show no
obvious signs of pore blockage from film deposition.

4.3.2 Formation of oil emulsions using sonication
Upon sonication, dispersion of oil into water results in a transition from two transparent
fractions to a single opaque layer. Examination of the droplet size distribution after sonication
reveals the relative stability of the emulsion. Large droplets quickly coalesce due to inherent
instability, whereas small droplets (<5 µm) persist for longer periods of time (days). Dynamic
light scattering revealed that within 2 min of sonication, some large (> 10 m) droplets were
present in the emulsion as seen by the peaks corresponding to large droplets in the size
distributions (Figure 4.6). However, after 20 minutes light scattering did not detect any
111

remaining large droplets in emulsions prepared with three different surfactants (Figure 4.7).
Large droplets likely coalesced due to instability and floated to the surface of the emulsion
away from the sampled region near the bottom of the container. Light-scattering
measurements showed no evidence for droplet coalescence in these three well-dispersed
emulsions for 10 days. However, the feed solution droplet size distribution was always assessed
before and after any rejection experiments to rule out changes in droplet exclusion due to
coalescence. These stable emulsions with droplet sizes less than 10 µm allow us to study
oil/water separations with solutions that typically show low oil rejection in microfiltration.

112

Figure 4.6: Oil droplet size distributions obtained within 2 min of sonication of 10% hexadecane
in water. Emulsions were stabilized with SDS (A), CTAB (B), and Triton (C).

113

Figure 4.7: Oil droplet size distributions obtained 20 min after sonication of 10% hexadecane in
water. Emulsions were stabilized with SDS (A), CTAB (B), and Triton (C).

114

Zeta potential measurements with each of the three emulsions give an indication of the
charge present on oil droplets. The cationic CTAB emulsions have particles with an approximate
zeta potential of 67 mV, whereas the anionic SDS emulsions contain particles with zeta
potential of -58 mV. Triton emulsions, which contain a neutral surfactant showed a zeta
potential of only -8 mV. While the zeta potentials for SDS- and CTAB-stabilized emulsion may
seem high, the zeta potential for the SDS emulsion is consistent with a literature value (-60
22

mV).

Therefore, the droplets in SDS- and CTAB-stabilized emulsions are highly charged so

electrostatic repulsion from charged membrane coatings may be effective.

4.3.3 Oil rejection by membranes with large pores
Filtration using membranes with nominal pore sizes greater than the average oil droplet
diameter allows us to examine whether electrostatic exclusion can contribute to rejection of
charged droplets. Unmodified control membranes with minimal surface charge should allow
complete oil passage, regardless of the droplet charge (Figure 4.8). In contrast, modified
membranes should electrostatically exclude only droplets whose charge has the same sign as
the charge on the membrane surface.

115

Figure 4.8: Control experiment design showing no oil rejection through large, unmodified pores
and electrostatic exclusion with membranes coated with PEMs.

Initial results for three repeated passes of new neutral and anionic 10% hexadecane
emulsions through the same membrane modified with [PAH/PSS] bilayers show encouraging
results (Figure 4.9). The bare membrane shows no neutral oil rejection and only ~5% anionic oil
rejection. The membrane likely has a small negative native charge, so a low percentage of
anionic oil rejection is not surprising. Importantly, rejection of the neutral (triton-stabilized)
emulsion remains zero for membranes coated with 4 and 6 PAH/PSS bilayers. After adsorption
of 8 PAH/PSS bilayers the oil rejection for the triton-stabilized emulsion increases to ~24%.
Meanwhile, anionic oil rejection increases to ~45%, ~60%, and ~95% upon increasing from 4 to
6 to 8 PAH/PSS bilayers in a membrane, respectively. Although the 95% rejection observed for

116

the membrane modified with 8 PAH/PSS bilayers is remarkable, the partial rejection of neutral
oil through this particular membrane suggests some size exclusion contributes to rejection.
Unfortunately, we cannot easily determine what mechanism is responsible for oil rejection in
this case. However, the 60% oil rejection with the SDS-stabilized (anionic) emulsion and the
membrane coated with the [PAH/PSS]6 film is equally impressive due to the lack of neutral oil
(triton-stabilized) rejection. This promising result suggests proper matching of surfactant and
membrane charge may prevent oil accumulation at the membrane surface and lead to higher
oil rejections with reduced fouling.

Figure 4.9: Oil rejections in the first replicate of dead-end filtration of a 10% hexadecane-inwater emulsion through 5 m nylon membranes modified with 0, 4, 6, and 8 PAH/PSS bilayers.
Emulsions were stabilized with either SDS (anionic droplets) or Triton (neutral droplets). Error
bars represent the standard deviation of average oil rejection in three replicate trials using the
same membrane.

117

Figure 4.10 shows oil rejections from a second batch of membranes used for dead-end
filtration of the 10% hexadecane emulsions. Although the general trend of increasing SDS
emulsion rejection with increasing film thickness remains, the oil rejections for the neutral
Triton emulsions are nearly as high as those for the anionic emulsion. Rejection of the ~neutral
droplets rose from a negligible value for a [PAH/PSS]6-coated membrane in the first replicate to
~27% in the second replicate. Likewise, the rejection of neutral droplets by the [PAH/PSS]4coated membrane was essentially equal to oil rejection with the anionic rejection. The
increased triton rejection in this second replication may result from the emergence of a second
small peak in the droplet size distribution centered at 30 µm. The presence of oil droplets larger
than the membrane pores might lead to increased neutral oil rejection and help explain
differences between the two trials. Additionally, the first trial used membranes from Millipore
while the second trial employed membranes from Sterlitech. However, LbL deposition should
be possible for a wide variety of substrates. Regardless, further is needed on the reproducibility
of concentrated oil rejections with treated membranes.

118

Figure 4.10: Oil rejections in the second batch of membranes prepared for dead-end filtration
of a 10% hexadecane-in-water emulsion through 5 m nylon membranes modified with 0, 4, 6,
and 8 PAH/PSS bilayers. Emulsions were stabilized with either SDS (anionic droplets) or Triton
(neutral droplets). Error bars represent the standard deviation of average oil rejection in three
replicate trials using the same membrane.

We also assessed rejection of oil droplets prepared by a 20-fold dilution of the stock
10% hexadecane-in-water emulsions to explore the role oil concentration plays in this rejection
mechanism. Dead-end filtration of these dilute solutions through the same membranes used
for replicate 2 did not show increased rejection of anionic oil droplets (Figure 4.11). Both
neutral and anionic emulsions show noticeable oil rejection even with bare membranes. By
diluting the stock solution 20 fold, the SDS concentration decreases to only 0.25 mM, well
below the critical micelle concentration of 8 mM.

23

At concentrations below the critical micelle

concentration, the SDS may not stabilize oil droplets and coalescence may occur. Studying the
average droplet size over time by dynamic light scattering for these dilute SDS solutions should
119

provide us with information about the droplet stability. However, the critical micelle
concentration for Triton is only 0.25 mM so dilution should not destabilize Triton solutions as
much as SDS.

24

Nonetheless, rejection still results for both anionic and ~neutral emulsions.

Figure 4.11: Oil rejections in the first replicate of dead-end filtration of a 0.5% hexadecane-inwater emulsion through 5 m nylon membranes modified with 0, 4, 6, and 8 PAH/PSS bilayers.
Emulsions were stabilized with either SDS (anionic droplets) or Triton (neutral droplets). Error
bars represent the standard deviation of average oil rejection in three replicate trials using the
same membrane.

4.3.4 Flux decline during microfiltration of oil emulsions
Although the preliminary results in section 4.3.3 are not conclusive, Figure 4.9 suggests
that electrostatic repulsion of charged oil droplets using PEMs occurs. However, filtration
through polyelectrolyte-coated membranes with extremely large pores (5 µm) is not intended
as a practical method for removal of dispersed oil and grease components. Instead, these
120

studies attempt to demonstrate the concept of enhanced rejection through tailored
electrostatic repulsion. For more practical applications, deposition of [PAH/PSS] films in
membranes with smaller pores may reduce fouling arising from oil adsorption onto membrane
surfaces and pores.
We propose that both the hydrophilicity of the PEM and the surface charge contribute
to antifouling behavior. The use of two complimentary methods for fouling reduction should
better stabilize flux compared to other hydrophilic antifouling treatments. The water treatment
industry already doses organic and inorganic additives into water streams during coagulation
and flocculation processes. Therefore, addition of surfactants or polyelectrolytes to increase oil
droplet charge is a foreseeable fouling prevention paradigm.
PEM layers were adsorbed on nylon membranes with nominal 0.45 m pore sizes to
compare flux declines in cross-flow filtration of oil-in-water emulsions. The plot of permeate
mass versus time in Figure 4.12 shows a decline in water passage over time for both bare
membranes and membranes coated with [PAH/PSS]4 films. The second PEM modified
membrane and the bare membrane have a similar initial flux but the bare membrane fouls
more extensively as demonstrated by the plateau in permeate mass. The first membrane
coated with a [PAH/PSS]4 films showed a smaller initial flux, but a relatively small flux decline.
Figure 4.13 shows the ratios of flux to initial flux for each of the membranes. The membranes
with polyelectrolyte coatings clearly foul less extensively. However, this comparison is most
valid for membranes with similar initial fluxes, because the permeate flow brings foulants to
the membrane.
121

Figure 4.12: Permeate mass as a function of time during cross-flow filtration of a 0.2 % (v/v)
hexadecane emulsion through a bare and and two different membranes modified with
[PAH/PSS]4 films. The emulsions were stabilized with 0.1 mM SDS in 0.2 mM NaCl.

Figure 4.13: Relative flux values over time for a bare and two trials of PEM modified
membranes during cross-flow filtration of a 0.2 % (v/v) hexadecane emulsion in a 0.1 mM SDS
0.2 mM NaCl solution.
122

In addition to the flux decline differences observed between bare and PEM modified
membranes during filtration of dilute oil emulsions, hydrophobic dye adsorption readily occurs
on bare membranes whereas treated membranes remain relatively unchanged after each trial
(Figure 4.14). Unfortunately, subsequent attempts to reproduce flux decline behavior with bare
membranes have yet to produce a steady flux value. In each case, unstable flux values have
resulted which lack noticeable trends. Current work is ongoing to discover the cause of this
irreproducibility. Nevertheless, despite the unusual flux values, rejection for three different
bare membranes was consistent with an average hexadecane rejection of 45 ± 13% (n=3).
Addition of PEM multilayers increases the average oil rejection to 84 ± 15% (n=2) while also
reducing flux decline and demonstrating the benefits of our proposed antifouling scheme. After
searching the literature, no papers have been discovered detailing the use of polyelectrolyte
multilayers to repel charged oil droplets to date.

Figure 4.14: Comparison of irreversible membrane staining from uncharged, hydrophobic oilsoluble dye for bare and PEM modified membranes after oil emulsion cross-flow filtration at 80
mL/min and an applied transmembrane pressure of 2 bar.

123

4.4 Conclusions

Deposition of highly charged [PAH/PSS]n films on porous membranes produces
hydrophilic surfaces that may have innate antifouling properties. Meanwhile, additions of
charged surfactants to feed solutions increase the surface charge of dispersed oil droplets. The
preliminary combination of hydrophilic, charged polyelectrolyte multilayer and charged oil
droplets promises as a new strategy for crafting fouling resistant membranes for oil-in-water
filtration. When using membranes with pores much larger than oil droplets (~15 x), membranes
treated with anionic terminated PEM layers appear to show enhanced rejection of anionic
emulsions compared to similarly sized neutral oil droplets. The enhanced rejection
demonstrates the possibility of adding charged surfactants or polymers to feed solutions as a
pretreatment step to aid water recovery by reducing membrane fouling. Similarly, the inclusion
of the same [PAH/PSS] PEMs as a surface treatment for cross-flow microfiltration membranes
does not appear to reduce flux significantly and appears to resist surface fouling to mitigate flux
decline.
While still preliminary in nature, the results presented in this chapter encourage further
exploration of the abilities and limitations of our proposed fouling mitigation scheme. The low
cost and high availability of many charged surfactants and polyelectrolytes suitable for addition
to feed solutions suggests that such a technique might by industrially practicable for removing
oil-in-water emulsions from aqueous streams. Notably, the application of promising antifouling
properties of hydrophilic films combined with our concept to match dispersed particle charge

124

with membrane surface charge represents an exciting opportunity for advancement of
membrane treatment technology.

125

REFERENCES

126

REFERENCES

(1)
United States Environmental Protection Agency, Hydraulic Fracturing Research
Study Fact Sheet, Washington, D.C., 2010,
http://www.epa.gov/safewater/uic/pdfs/hfresearchstudyfs.pdf (accessed 03/09/2013).
(2)
Wandera, D.; Wickramasinghe, S. R.; Husson, S. M. Journal of Membrane Science
2011, 373, 178.
(3)
Technical Summary of Oil and Gas Produced water Treatment Technologies,
Tulsa, OK, 2005, http://w.all-llc.com/publicdownloads/ALLConsultingWaterTreatmentOptionsReport.pdf (accessed 04/05/2013).
(4)
Ahmadun, F. l.-R.; Pendashteh, A.; Abdullah, L. C.; Biak, D. R. A.; Madaeni, S. S.;
Abidin, Z. Z. Journal of Hazardous Materials 2009, 170, 530.
(5)
National Energy Technology Laboratory. Produced Water Management
Technology Descriptions: Fact Sheet - Discharge.
http://www.netl.doe.gov/technologies/pwmis/techdesc/discharge/ (accessed
04/01/2013).
(6)

Cheryan, M.; Rajagopalan, N. Journal of Membrane Science 1998, 151, 13.

(7)
Kota, A. K.; Kwon, G.; Choi, W.; Mabry, J. M.; Tuteja, A. Nature communications
2012, 3, 1025.
(8)
Sagle, A. C.; Van wagner, E. M.; Ju, H.; McCloskey, B. D.; Freeman, B. D.; Sharma,
M. M. Journal of Membrane Science 2009, 340, 92.
(9)

Xu, P.; Drewes, J. E. Separation and Purification Technology 2006, 52, 67.

(10) Yi, X. S.; Yu, S. L.; Shi, W. X.; Sun, N.; Jin, L. M.; Wang, S.; Zhang, B.; Ma, C.; Sun, L.
P. Desalination 2011, 281, 179.

127

(11)

Xu, P.; Drewes, J. E.; Heil, D. Desalination 2008, 225, 139.

(12)

Faibish, R. S.; Cohen, Y. Journal of Membrane Science 2001, 185, 129.

(13) Xu, Z.-L.; Chung, T.-S.; Loh, K.-C.; Lim, B. C. Journal of Membrane Science 1999,
158, 41.
(14) Chakrabarty, B.; Ghoshal, A. K.; Purkait, M. K. Journal of Membrane Science 2008,
325, 427.
(15)

McPherson, T.; Kidane, A.; Szleifer, I.; Park, K. Langmuir 1998, 14, 176.

(16)

Hamming, L. M.; Messersmith, P. B. Material Matters 2008, 3, 52.

(17) Tuteja, A.; Choi, W.; Mabry, J. M.; McKinley, G. H.; Cohen, R. E. Proceedings of
the National Academy of Sciences 2008, 105, 18200.
(18) Kobaku, S. P. R.; Kota, A. K.; Lee, D. H.; Mabry, J. M.; Tuteja, A. Angewandte
Chemie 2012, 51, 10109.
(19)

Xu, F. J.; Zhao, J. P.; Kang, E. T.; Neoh, K. G.; Li, J. Langmuir 2007, 23, 8585.

(20) Stanton, B. W.; Harris, J. J.; Miller, M. D.; Bruening, M. L. Langmuir 2003, 19,
7038.
(21)

Dubas, S. T.; Schlenoff, J. B. Langmuir 2001, 17, 7725.

(22) Tucker, I. M.; Petkov, J. T.; Jones, C.; Penfold, J.; Thomas, R. K.; Rogers, S. E.;
Terry, A. E.; Heenan, R. K.; Grillo, I. Langmuir 2012, 28, 14974.
(23) Domínguez, A.; Fernández, A.; González, N.; Iglesias, E.; Montenegro, L. Journal
of Chemical Education 1997, 74, 1227.
(24) SigmaAldrich, Sigma Product Information: Triton X100,
http://www.snowpure.com/docs/triton-x-100-sigma.pdf (accessed 04/17/2013).
128

Chapter 5: Conclusions and future work

5.1 Conclusions
This dissertation focused on developing coatings that can potentially improve the
performance of membranes in remediation of industrial wastewater, desalination of brackish or
seawater, and removal of oil from oil-in-water emulsions. As the demand for water continues to
increase to sustain agricultural, residential, and industrial needs, new methods for producing
high quality water must improve to keep pace. Future directions in water treatment need to
focus on the decontamination and sanitization of toxic or otherwise challenging sources. Rapid
developments in membrane technology over the last half century have greatly increased our
ability to produce safe water. In fact, Singapore, which has very limited sources of fresh water,
has developed a large-scale wastewater reclamation project to produce drinking water through
1

use of advanced membrane technology. Realization of a cost effective method to quickly and
reliably treat challenging water sources would greatly reduce global water stress in the coming
years.
Chapter 2 focused on the application of PEMs containing Pt nanoparticles as polymer
membrane coatings for CWAO of industrial waste. Several other groups demonstrated potential
benefits of the CWAO process using membranes as interfacial contactors.

2-4

These studies

employed ceramic membranes due to their temperature and chemical resistance. However,
ceramic membranes are generally an order of magnitude more expensive than polymeric

129

options. Our work utilized inexpensive polymeric hollow fibers with inner diameters several
times smaller than what is currently available with practical ceramic membranes.
Remarkably, we observed over 50% oxidation of formic acid during a single pass through
polysulfone (PS) membranes modified with PEMs containing Pt nanoparticles. Using modules
with five fibers resulted in over 90% formic acid oxidation in a single pass due to a longer
residence time in the membrane. Modeling of the reaction suggested that the membrane was
operating well below the diffusion-limited rate with an apparent reaction rate constant of 0.01
-1

s . Despite these positive observations, the PEM-modified polymeric membranes have low Pt
catalytic activity compared to similarly modified ceramic membranes. The low Pt activity in the
polymeric membranes likely results from the loading of Pt throughout the entire wall thickness
rather than in a concentrated region near the air-water interface.
Chapter 3 described porous PP coatings that decrease fouling and prevent pore wetting
during MD of humic acid and whey solutions. As an attractive, low-temperature water
desalination technique, MD may couple with solar energy or waste heat to economically
provide safe water in underdeveloped or water-stressed regions. To reduce the number of
people lacking access to safe drinking water by 2015 as proposed in the Millennium
Development Goals, freshwater production must increase in underdeveloped and remote areas
despite significant infrastructural deficiencies.

5

A portable MD system capable of low-

maintenance operation may be an ideal method for meeting such goals.

130

In studies using humic acid solutions that simulate brackish water samples with high
fouling capability, unmodified PVDF membranes rapidly fouled and wetted to allow salt passage
and an average rejection of only 32 ± 12% after 20 hours. In contrast, despite a 20-30% drop in
flux relative to unmodified membranes, porous PP-modified PVDF membranes maintained a
steady flux while resisting pore wetting to give 99.9 ± 0.1% salt rejection over 20 hours. The
operational stability of the PP-treated membranes more than compensates for the slight flux
decline resulting from the increased mass transport resistance of the added PP. Similarly, when
using dairy whey as a feed solution, bare PVDF membranes did not resist surface and pore
fouling by the proteins and oils present in whey. The conductivity in the permeate rose ~150
µS/cm during MD, and the flux actually reversed for bare membranes with pure water moving
backwards through membrane pores under osmotic flow. Porous PP films deposited on the
PVDF membranes, on the other hand, prevented pore wetting, and the modified membranes
gave 97 ± 0.5% salt rejection over a 20 hour period. Although the porous PP coating resisted
humic acid and dairy whey fouling, the coating was not effective in the presence of surfactants,
so the composition of feed solutions should be well-characterized before using PP-modified
membranes.
Finally, chapter 4 discussed preliminary work on a new strategy for removing oil from
oil-in-water emulsions. The goal was to modify membranes with PEM films and create charged
hydrophilic surfaces that resist adsorption of oil droplets with the same surface charge as the
PEM. Addition of charged surfactants to create charged oil droplets should increase rejection
and decrease fouling due to electrostatic repulsion of the droplets. Many membranes consist of
hydrophobic polymers that rapidly adsorb oil to cause fouling during filtration of oil-in-water
131

emulsions. In the case of PEM-modified membranes, the combination of a tightly held water
boundary layer and electrostatic repulsion of charged oil droplets may provide a new paradigm
for fouling management during oil removal from water.
Initial work using large membrane pores generally incapable of rejecting small oil
droplets demonstrates the ability of negatively charged PEMs to electrostatically repel similarly
charged oil droplets. While further work remains to better define differences in rejection based
on the number of PEMs deposited, rejections of anionic oil droplets generally exceed those of
neutral droplets when oil emulsions pass through anionically terminated PEMs. Additionally, we
presented promising initial results that are consistent with our hypothesis of reduced fouling
with hydrophilic coatings that electrostatically repel oil droplets. During two h of cross-flow
filtration of a 0.2% hexadecane-in-water emulsion, 2 [PAH/PSS]4-coated membranes showed an
80% decline in flux, but flux through a bare PVDF membrane declined more than 90%.
Additionally, PEM-treated membranes resisted dye staining during the two hours of filtration
while showing nearly 2-fold higher oil rejection than bare membranes.

5.2 Future work
5.2.1 Polypropylene hollow fibers for CWAO
As the most expensive component of our CWAO system, Pt must be used efficiently for
practical applications. In an effort to overcome the low catalyst activity in our PS hollow fiber
membranes, we propose the use of similarly sized polypropylene membranes that will maintain
132

high single pass conversions while increasing catalytic activity due to increased control over Pt
location. PS membranes are hydrophilic so PEM and nanoparticle deposition solutions fill the
entire pore volume to deposit nanoparticles throughout the membrane wall (Figure 5.1, top).
However, the high catalytic activity of previous ceramic membranes arose from the ability to
focus the gas-liquid interface to the innermost region of the membranes where all the catalyst
was loaded at high density. The diffuse loading of catalyst throughout our fiber walls precluded
our ability to localize the reaction to catalyst rich areas. Additionally the hydrophilicity of PS
resulted in complete wetting of membrane pores at all available oxygen overpressures.
The hydrophobicity of polypropylene fiber membranes could prevent pore wetting and
limit polyelectrolyte and nanoparticle adsorption to regions near the inner wall of the
membrane (Figure 5.1, bottom). However, adsorption of PEMs to PP surfaces is difficult due to
the hydrophobicity and lack of functional groups that interact with PEMs. Therefore, localized
polyelectrolyte adsorption requires a way to increase the hydrophilicity of the PP surface near
the inner wall of the membrane. Plasma and photo-irradiation techniques are not suitable since
neither will modify the lumen surface effectively. However, flow of chemical oxidants, such as
ammonium persulfate, through the membrane lumen may increase the hydrophilicity of PP
6

membranes only near the lumen surface. Short oxidant residence times should prevent
modification inside the pores. The localization of PEM and nanoparticles adsorption in this
region may allow Pt loading only at the gas-liquid interface to achieve superior catalyst activity
(Figure 5.1)

133

.
Figure 5.1: Cartoon of the potential catalyst locations resulting from adsorption of [PSS/PEI/Pt]
coatings in hydrophilic PS membranes and chemically treated polypropylene membranes.

The fabrication of polymeric membranes with high activity for CWAO at low
temperatures and pressures could increase the feasibility of CWAO for industrial waste
treatment. Polymeric membranes are much less expensive than ceramic membranes.
Additionally, the small fiber diameters reduce the diffusion distance for pollutants to give high
single-pass conversions at short residence times and hopefully avoid the need to recycle feed
solutions through membranes multiple times for high conversion. Similarly, the small fiber size
allows for high packing density within membrane modules to increase throughput while
maintaining a small footprint.

134

5.2.2 Antifouling MD membranes for the food and beverage industry
MD is principally used for either the production of high quality water or the
concentration of non-volatile aqueous solutions. The dairy industry is well suited to use both
functions of MD operation simultaneously. In this way, the distillation of whey solutions
produces both fresh water for use in other areas and concentrated whey solutions suitable for
further drying before sale. Obtaining value from whey solutions, at one point considered a
waste product requiring disposal costs, improves dairy industry profitability.
While we demonstrated improved whey fouling resistance using porous PP coatings,
further studies should examine the duration of resistance to pore wetting in partnership with a
local dairy industry representative. For superhydrophobic films to maintain antifouling
properties, they must retain air trapped within surface roughness features. Long-term studies,
should also explore air back-flushing steps at specified intervals as a simple cleaning procedure.
If short periods of air exposure help maintain air entrapment within surface roughness features,
duration of the PP antifouling properties should increase. Additional MD applications within the
food and beverage industry include other Michigan products such as apple juice and wine.
Apple juice primarily consists of carbohydrate sugars and apple oils extracted during the juicing
process. The removal of water from apple juice allows for the formation of juice concentrate for
shipment in larger quantities and lower prices to markets by removing a majority of the water
and allowing consumers to replace the water later. The use of porous PP coatings may also
reduce fouling from apple solutions, which would increase plant productivity and reduce
membrane treatment costs. When excess alcohol is present in wines, winemakers need to

135

remove it MD may be effective in selective removal of ethanol over water due to ethanol’s
higher vapor pressure.

5.2.3 Polishing of oil-in-water emulsions using membrane filtration
Membranes provide many advantages over alternative techniques for the removal of
small, well dispersed oil droplets from aqueous solutions. However, oils rapidly foul membrane
surfaces to reduce fluxes and require frequent cleaning procedures. Based on our preliminary
investigation in Chapter 4, the surprising addition of charged surfactant to oil emulsions leads
to higher oil rejections by charged membranes despite also stabilizing oil droplets from
coalescence. The next logical step for this project would be the inclusion of these films on
nanofiltration or other membranes with small pores capable of rejecting many sizes of oil
droplets.
The ability to reduce fouling over time during oil filtration experiments while increasing
oil rejection due to electrostatic influences could significantly impact oil removal operations.
Additionally, the use of inexpensive polyelectrolytes and surfactants should avoid high
treatment costs for membranes. Moreover, if PEMs are used to increase the hydrophilicity of a
chemically resistant membrane, such as a ceramic one, the coatings can be considered
sacrificial. Over time as their performance possibly degrades due to oil buildup on treated films,
7

the PEMs can be eluted from the membrane using acidic or basic conditions. The coatings

136

could then be reapplied within a matter of minutes (depending on the number of bilayers) to
reestablish antifouling properties and high oil rejections.

5.3 Summary
In summary, the use of inexpensive PEMs and thin porous PP films can positively alter
the properties of membrane substrates. PEMs can incorporate catalytically active nanoparticles
within membranes, with tunable catalyst loading for industrial waste remediation using CWAO.
Porous PP films deposited on substrates in a controlled environment rapidly produce a reliably
superhydrophobic surface that prevents humic acid and whey protein fouling during MD.
Finally, the incorporation of PEMs on membranes while dosing charged surfactants to feed
solutions leads to improved fouling resistance and oil droplet rejection when compared to
untreated membranes. In total, this dissertation covers three avenues in which thin, polymeric
coatings may assist in developing technologies for the next generation of water treatment and
production facilities.

137

REFERENCES

138

REFERENCES

(1)
Siemens Water Technologies. Membrane Filtration System Improves Water
Sustainability at NEWater Plant in Changi, Singapore.
http://www.water.siemens.com/en/about_us/corporate%20information/Pages/default.
aspx (accessed 04/17/2013).
(2)
Iojoiu, E. E.; Landrivon, E.; Raeder, H.; Torp, E. G.; Miachon, S.; Dalmon, J.-A.
Catalysis Today 2006, 118, 246.
(3)
Miachon, S.; Perez, V.; Crehan, G.; Torp, E.; Ræder, H.; Bredesen, R.; Dalmon, J.
A. Catalysis Today 2003, 82, 75.
(4)
Dotzauer, D. M.; Abusaloua, A.; Miachon, S.; Dalmon, J.-A.; Bruening, M. L.
Applied Catalysis, B: Environmental 2009, 91, 180.
(5)
United Nations. Millenium Development Goals: Environmental Sustainability.
http://www.un.org/millenniumgoals/environ.shtml (accessed 03/09/2013).
(6)
Piletsky, S.; Piletska, E.; Bossi, A.; Turner, N.; Turner, A. Biotechnology and
Bioengineering 2003, 82, 86.
(7)
Shan, W. Q.; Bacchin, P.; Aimar, P.; Bruening, M. L.; Tarabara, V. V. Journal of
Membrane Science 2010, 349, 268.

139