i253”: #
1 g) LIBRARY
Michigan State
University

 

 

 

 

This is to certify that the
dissertation entitled

THE ENVIRONMENTAL APPLICATIONS AND
IMPLICATIONS OF NANOTECHNOLOGY IN MEMBRANE-
BASED SEPARATIONS FOR WATER TREATMENT

presented by

Wenqian Shan

has been accepted towards fulﬁllment
of the requirements for the

 

 

PhD. degree in Environmental Enﬂeenng
/ f“ C"
% ~ WM

 

Major Professor's Signature
ﬂat/7L /;z, 2 0/0
I I

Date

 

MSU is an Afﬁrmative Action/Equal Opportunity Employer

 

 

 

 

 

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

 

DATE DUE DATE DUE DATE DUE ‘

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KzlProiIAccaPresICIRCIDateDmindd

 

 

THE ENVIRONMENTAL APPLICATIONS AND IMPLICATIONS OF
NAN OTECHNOLOGY IN MENIBRANE-BASED SEPARATIONS FOR
WATER TREATMENT

By

Wenqian Shan

A DISSERTATION
Submitted to
Michigan State University

in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Environmental Engineering

2010

ABSTRACT
THE ENVIRONMENTAL APPLICATIONS AND IMPLICATIONS OF
NANOTECHNOLOGY IN MEMBRANE-BASED SEPARATIONS FOR
WATER TREATMENT
By

Wenqian Shan

This dissertation presents results of three related projects focused on the
applications of membrane separation technology to water treatment: 1) Experimental
design and evaluation of polyelectrolyte multilayer ﬁlms as regenerable membrane

coatings with controllable surface properties; 2) Modeling of the interactions of

nanoscale TiOz and NOM molecules in aqueous solutions of environmentally relevant

compositions; 3) Experimental design and preliminary testing of a membrane-based
crossﬂow ﬁltration hydrocyclone process for the separation of oil-in-water
dispersions.

Chapter 2 describes the design of polyelectrolyte multilayers as nanoscale
membrane coatings and their application in nanoﬁltration of feed waters that contain
suspended colloids and dissolved species. Layer-by-layer deposition of anionic and
cationic polyelectrolytes was employed to prepare membrane coatings allowing for a
ﬁne control over their surface properties. This approach to membrane design also
affords a possibility of regenerating coatings after they are fouled by colloids. This
project demonstrated, for ﬁrst time, the possibility of designing nanoﬁltration
membranes with regenerable skin.

Chapter 3 describes a study on the mechanisms of natural organic matter (NOM)

adsorption onto the surface of titania nanoparticles. Titania (TiOz) is often used in the

fabrication of ceramic membranes and understanding how NOM interacts with Ti02

can help to better predict ceramic membrane fouling by NOM-containing waters. The

combined effect of pH and calcium on the interactions of nonozonated and ozonated
NOM with nanoscale Ti02 was investigated by applying extended Derjaguin — Landau
— Verwey - Overbeek (XDLVO) modeling. XDLVO surface energy analysis predicted

NOM adsorption onto TiOz in the ozone-controlled regime but not in the

calcium-controlled regime. In both regimes, short range NOM-NOM and NOM-TiOZ

interactions were governed by acid-base and van der Waals forces, whereas the role of
electrostatic forces was found to be relatively insigniﬁcant. Ozonation increased the

surface energy of NOM, contributing to the hydrophilic repulsion component of the

NOM-NOM and NOM—Ti02 interactions. In the calcium-controlled regime,

non-XDLVO interactions such as intermolecular bridging by calcium were
hypothesized to be responsible for the observed adsorption behavior.

Chapter 4 describes research on the crossﬂow ﬁltration hydrocyclone separation of
oil-in-water dispersions wherein a ceramic tubular membrane was used as the
permeable wall of the hydrocyclone. Air sparging was applied to mitigate oil fouling.
A dual membrane system consisting of an outer hydrophilic ceramic membrane and
an inner hydrophobic polymeric membrane was evaluated to test the possibility of
separating the dispersion into two streams: 1) oil with zero or very low concentration
of water and 2) water with zero or very low concentration of oil. The performance of
the dual membrane system indicated the possibility of using membranes with different
chemical aﬁ'mities to cost-effectively separate the oil-water dispersion into two
separate phases. The incorporation of air sparging to membrane ﬁltration was found to

be effective in mitigating oil fouling with improved permeate ﬂux.

To my father, Jinbao Shan, and my mother; Yanjiang Wang,
who have given me the best they have,

support and encourage me everyday

iv

Acknowledgements

I would like to express my sincere appreciation to my research and academic advisor,
Dr. Volodymyr V. Tarabara for his diligent, insightful and professional guidance
throughout my graduate studies, without which this dissertation would have been
impossible. I am grateful that I had the opportunity to work with him on several grand
projects during my Ph.D. study.

I would like to thank my committee members: Dr. Merlin L. Bruening, Dr. Mantha S.
Phanikumar, Dr Andre Benard and Dr. Patrice Bacchin for their invaluable comments,
suggestions and advice to improve this thesis to its ﬁnal form.

I would like to express my sincere appreciation for NSF which has provided the
ﬁnancial support for my Ph.D. study and given me precious experience during the
collaboration with students and professors in some foreign universities.

I would like to thank Dr. Charles A. Petty, Dr. Susan J. Masten, Dr. Pierre Aimar who
have made valuable suggestions in my different research projects.

I would like to thank previous and current group members who have provided a
helpful and comfortable lab environment that I cannot expect more.

Finally I want to thank my husband Mr. Chaopeng Shen for his endless care and
dedicated support, which has given me much of the encouragement and support at all

stages of my Ph.D. study.

Table of Contents

List of Tables ......................................................................................................... viii
List of Figures .......................................................................................................... ix
Abbreviations .......................................................................................................... xii
Chapter 1. Introduction ............................................................................................... l
1.1. Nanotechnology and the environment .......................................................... 1
1.1.1. Nanotechnologies for a cleaner environment ..................................... 3

1.1.2. Nanotechnology applications in water treatment and quality control..3

1.1.3. Potential environmental risks of nanotechnology ............................... 5

1.2. Membranes nanotechnologies ...................................................................... 6
1.2.1. Nanoparticle-enabled membranes ...................................................... 7

1.2.2. Novel nanoscale membrane coating — polyelectrolyte multilayer ﬁlms
................................................................................................................... 7

1.3. Nanoparticle interactions in aqueous media as a key to elucidating
environmental impacts of nanomaterials ........................................................... 10

Chapter 2. Polyelectrolyte multilayer ﬁlms used as renewable membrane coatings

with tunable surface properties ................................................................................. 12
2.1. Introduction .......................................................................................... 12
2.2. Background and approach ..................................................................... 14

2.2.1. Colloidal fouling ............................................................................. 14
2.2.2. Approach ......................................................................................... 16
2.3. Experimental ......................................................................................... 23
2.3.1. Materials ......................................................................................... 23
2.3.2. Preparation of PEM membranes ...................................................... 24
2.3.3. Membrane characterization ............................................................. 27
2.3.4. Bench-scale crossﬂow ﬁltration system ........................................... 28
2.3.5. Experimental protocol for crossﬂow ﬁltration ................................. 30
2.4. Results and discussion .......................................................................... 34

2.4.1. Charge, hydrophilicity, and water permeability of PEM membranes 34
2.4.2. Determining the salt permeability coefﬁcient of PEM membranes...36
2.4.3. Rejection and permeate ﬂux in colloidal fouling experiments: overall
comparative assessment of NF27O and PEM membranes .......................... 43
2.4.4. Membrane performance in colloidal fouling experiments: Transient

behavior of permeate ﬂux, MgSO4 rejection, and cake resistance ............. 45
2.4.5. Regeneration of [PSS/PAH]4 and [PSS/PAH]4.5 ﬁlms ..................... 52
2.5. Conclusions .......................................................................................... 56

Chapter 3. Interactions of aqueous NOM with nanoscale TiOZ: Implications for

ceramic membrane ﬁltration-ozonation hybrid process ............................................. 58
3. l . Introduction ............................................................................................ 5 8
3.2. Approach ................................................................................................ 60

vi

3.3. Material and methods .............................................................................. 62

3.3.1. Materials ......................................................................................... 62
3.3.2. Characterization of NOM molecules and Ti02 nanoparticles ........... 63
3.4. Results and Discussion ............................................................................ 68
3.4.1. Fractionation of Suwannee River NOM (SRNOM) ......................... 68
3.4.2. Hydrophilicity of SRNOM .............................................................. 71
3.4.3. Free energy of SRNOM-SRNOM cohesion and SRNOM-Ti02
adhesion ................................................................................................... 72
3.5. Conclusions ............................................................................................ 80

Chapter 4. Hybrid dual media membrane crossﬂow ﬁltration hydrocyclone system for

produced water treatment ......................................................................................... 82
4.1. Introduction ............................................................................................ 82

4.2. Basics of hydroclone and oil water separation ......................................... 85
4.2.1. The separation principle of hydrocyclones ....................................... 85

4.2.2. Critical pressure .............................................................................. 87

4.3. Material and methods .............................................................................. 88
4.3.1. Materials ......................................................................................... 88

4.3.2. Characterization of oil-water dispersions and membranes ................ 89

4.3.3. Hybrid crossﬂow ﬁltration system ................................................... 90

4.4. Results .................................................................................................... 95
4.4.1. The effect of membrane pore size, and the presence of surfactant on
membrane ﬁltration performance .............................................................. 95

4.4.2. The performance of hybrid membrane ﬁltration system ................... 99

4.4.3. The performance of dual membrane system ................................... 102

4.5. Conclusions .......................................................................................... 104
Chapter 5. Summary and future work ..................................................................... 105
References ............................................................................................................. 108

vii

List of Tables

Table 2.1. Ionic strength (Ic) and pH of the PEM solutions used to prepare PEM
membranes. The adsorption time for each layer is listed in parentheses. ................... 25

Table 2.2. Transmembrane pressures and resulting permeate ﬂuxes during SiOz
ﬁltration by PEM membranes. Also indicated are transmembrane pressures used to

compact membranes prior to ﬁltration tests .............................................................. 31
Table 2.3. Surface and transport properties of the membranes ................................... 35
Table 2.4. Surface energy of probe liquids and different membranes ......................... 36

Table 3.1. Contact angles measured and surface energy parameters calculated for
SRNOM, ozonated SRNOM, and Ti02 at different pH and concentrations of Ca2+. .65

Table 3.2. Free energy of adhesion of SRNOM on the surface of TiOz nanoparticles

and SRNOM-SRNOM cohesion ............................................................................... 74
Table 4.1. Hybrid separation processes resulting from combining air sparging,
hydrocyclonic separation and membrane ﬁltration .................................................... 94
Table 4.2. Water permeability of different ceramic membranes ................................. 96

viii

List of Figures

Figure 2.1. Conceptualized drawing of the application and regeneration of a

polyelectrolyte multilayer (PEM) nanoﬁltration membrane. ..................................... 13
Figure 2.2. Particle diffusivity as a function of radius for conditions typical of hollow
ﬁber UF membranes [136] ....................................................................................... 15
Figure 2.3. Structure of polymers used in PEM ﬁlms (a) PDADMAC, (b) PAA, (c)
PSS, (d) PAH. .......................................................................................................... 26
Figure 2.4. The schematic of the crossﬂow ﬁltration system ..................................... 29

Figure 2.5. Intrinsic (a) and observed (b) rejection of MgSO4 by NF270 and ﬁve PEM
membranes as a function of permeate ﬂux during ﬁltration of colloid-free MgSO4
electrolyte. ............................................................................................................... 38

Figure 2.6. Dependence of the MgSO4 permeability coefﬁcient on MgSO4
concentration at the membrane surface for NF270 and ﬁve PEM membranes. .......... 39

Figure 2.7. Experimentally determined (symbols) and ﬁtted (lines) concentration
dependence (Cm) of the salt permeability coefﬁcient (1;) and G ratio (see equation
2.15) for each membrane. ......................................................................................... 42

Figure 2.8. Steady state values of the speciﬁc permeate ﬂux (a), observed MgSO4
rejection (b), and concentration polarization factor (0) for membranes before fouling

(ca. J=2.8 x 10'5 m/s) and after fouling with Si02 colloids. ...................................... 45

Figure 2.9. Transient behavior of the normalized permeate ﬂux (a), observed MgSO4

rejection (b) and colloidal cake resistance (c) in experiments on the ﬁltration of SiOz
colloids. ................................................................................................................... 47

Figure 2.10. Evolution of the concentration polarization factor during ﬁltration of
Si02 suspensions by different membranes. ............................................................... 49

Figure 2.11. Hydraulic resistance (Rm) and observed salt rejection (Robs) values of
a UP 50 kDa support before (1) and after the following sequential steps: (2)

modiﬁcation by a PEM; (3) fouling by SiOz colloids for 20 h; (4) backﬂushing with
water for 1 h; (5) soaking in pH 10 buffer for 10 min; (6) redeposition of a PEM layer.
(In cases where no data are visible, corresponding steps were omitted.) PEM

membranes employed included (a) as prepared [PSS/PAH]4.5; (b) a [PSS/PAH]4.5
ﬁlm with Si02 ﬁltration and backﬂushing prior to soaking in pH 10 buffer. ............. 53

Figure 2.12. SEM images of membranes at the different steps described in section

ix

2.3.5: (a) UF 50 kDa support (step 1); (b) after the UP support was modiﬁed by
[PSS/PAH]4.5 (step 2); (c) after the membrane was fouled by SiOz for 20 h (step 3);

(d) after the SiOz-fouled [PSS/PAH]4.5 membrane was soaked in pH 10 buffer and
regenerated with a new PEM. ................................................................................... 55

Figure 3.1. SEM micrographs of the cross-section of (a) the bare membrane and
membranes with the layer of ﬁltered (b) NOM, (c) ozonated NOM, and (d) ozonated

NOM in the presence of Ca2+. The fouled membranes are representative of samples
used in contact angle measurements. Arrows point to the NOM cake ........................ 64

Figure 3.2. TOC fraction of (a) NOM, (b) O3-NOM, (c) NOM(Ca2+) and (d)

O3-NOM(Ca2+) that has a smaller molecular weight than the number indicated on the
x-axis. ...................................................................................................................... 69

Figure 3.3. TOC fraction of (a) NOM, (b) 03-NOM, (c) NOM(Ca2+) and (d)

O3-NOM(Ca2+) that has molecules in the size range indicated on the x-axis. The TOC
content in each MW class was computed as the diﬁerence between the TOC contents
of the ﬁltrates from the two membranes used in the isolation of the corresponding size
fraction ..................................................................................................................... 70

Figure 3.4. Combined effect of pH, Ca2+ and pre-ozonation on hydrophilicity of
SRNOM. .................................................................................................................. 71

Figure 3.5. LW, AB, EL and Total energy of SRNOM-TiOz interactions for
non-ozonated SRNOM (a, c, e) and ozonated SRNOM (b, d, f) at pH 3 (a, b) pH 5 (c,
d), and pH 8 (e, f) ..................................................................................................... 76

Figure 3.6. LW, AB, EL and Total energy of SRNOM-SRNOM interactions for
non-ozonated SRNOM (a, c, e) and ozonated SRNOM (b, d, f) at pH 3 (a, b) pH 5 (c,
d), and pH 8 (e, f) ..................................................................................................... 77

Figure 3.7. Combined effect of pH, Ca2+ and pre-ozonation on the XDLVO energy of
interaction between (a) SRNOM molecules, and (b) SRNOM molecule and TiOz

surface. .................................................................................................................... 79
Figure 4.1. Diagram of a reverse flow (a) and a forward ﬂow hydrocyclones (b) [230]
................................................................................................................................. 87
Figure 4.2. Calibration curve for copper ................................................................... 90
Figure 4.3. Schematic diagram of the hybrid crossﬂow ﬁltration system .................. 92

Figure 4.4. Schematic diagram of the arrangement of hollow ﬁber membrane within a
ceramic membrane. .................................................................................................. 92

Figure 4.5. Schematic diagram of (a) hybrid crossﬂow membrane ﬁltration unit and (b)
the ﬂow pattern inside the crossﬂow membrane ﬁlter in a hybrid ﬁltration system. ..93

Figure 4.6. Permeate ﬂux decline (a) and oil rejection (b) of three membrane during
the ﬁltration of oil-water dispersion without PVA .................................................... 96

Figure 4.7. Oil droplet size distribution with and without the presence of PVA in the
retentate during the ﬁltration of 3.5 pm ﬁlter ............................................................ 97

Figure 4.8. Permeate ﬂux decline (a) and oil rejection (b) of three membranes during
the ﬁltration of oil-water dispersion with 0.2 g/L PVA ............................................. 98

Figure 4.9. Oil droplet size distribution measured for the retentate at 5 min (a) and at
30 min (b) of ﬁltrations with 0.14 um ﬁlter at different hybrid conditions. ............. 100

Figure 4.10. Permeate ﬂux decline during the ﬁltrations with 0.14 um ﬁlter at
different hybrid conditions. .................................................................................... 101

Figure 4.11. SEM images of the outer surface (a) and cross-section (b) of the PT FE
membrane .............................................................................................................. 103

Figure 4.12. The images of samples of the retentate (sample on the left), the permeate
from 1.4 pm ceramic membrane (3 samples in the middle) and the permeate from the
hollow ﬁber membrane (sample on the right). ........................................................ 104

xi

ACF F

ACFFH

CCF

CCFH

NOM

PAA

PAH

PDADMAC

PEM

PSS

PTFE

PVA

RO

SRNOM

XDLVO

Abbreviations

Atomic adsorption

Air sparged crossﬂow ﬁltration

Air sparged crossﬂow ﬁltration hydrocyclone
Crossﬂow ﬁltration

Crossﬂow ﬁltration hydrocyclone
Hydrophilic

Hydrophobic

Molecular weight cutoff

Nanofrltration

Natural organic matter

Poly(acrylic acid)

Poly(allylamine hydrochloride)
Poly(diallyldirnethyl ammonium chloride)
Polyelectrolyte multilayer

poly(styrene sulfonate)
polytetraﬂuoroethylene

Polyvinyl alcohol

Reverse Osmosis

Suwannee River NOM

Extended Derjaguin — Landau — Verwey - Overbeek

xii

Chapter 1. Introduction

1.1. Nanotechnology and the environment

Drawing historical parallels with the invention of steam engines and railways in the

19th century and the emergence of automobiles and computers in the 20th century, the

recent advent and the explosive growth of nanotechnologies promise to have a large

. . . . . st
Impact on the screntrﬁc and economic developments in the 21 century [1].

Nanotechnology is a general term for the design, characterization, production and
application of structures, devices and systems in which the shape and size is
controlled at the nanoscale [2]. The concept of “nanotechnology” was ﬁrst introduced
by Richard Feynman in 1956 [3]. Nanoobjects can take different forms and shapes;
examples include nanoparticles, nanorods and nanoplatelets [4].

Data show that western Europe, Japan and US. were all experiencing steady
increases in public expenditures for the promotion of nanotechnology from 1997 to
2003 [5]. In the US. alone, expenditures doubled from 400 million USD in 2001 to
800 million USD in 2003. Besides the public investment, manufacturers and
enterprises have invested heavily in the research, development and production of
nanotechnology. As of 2004, 1 billion USD has been invested in companies that are
working with nanomaterials [6]. Among that, nearly half of that money was spent in
2003 and 2004 alone. Worldwide, the total investment in nanotechnology in 2005 was
in the 5 to 8 billion USD range [7]. Investments made in nanotechnology have
resulted in many key interests and developments in this area. AS a result, the number

of the publications and patents found when searching with keywords

“nanotechnology” and “nanoparticles” in PUBMED [8] have increased enormously
over the last few decades [1]. For example, less than 500 articles related to
nanoparticles were published between 1995 and 1999, while over 8,000 articles
focused on nanoparticles were published between January I, 2005 and April 24, 2008
[1]. A market volume of 890 billion USD in nanotechnology is expected by 2015
based on the actual market volume in 2003 and the conservatively estimated steady
rise of 8 percent per year for the next 10 years and further [9].

Nanoscale materials have been introduced into people’s daily lives for the purpose
of saving energy and resources, as well as for improving environmental conditions
and the quality of life. The small size, high surface area and reactivity of
nanomaterials make them uniquely suitable for various purposes and as a result there
has been a rapid expansion in the development of nanomaterials for various industries
such as textile, medicine and health, computing, transportation, aeronautics and space
exploration, and environment [10]. However, these very characteristics of
nanomaterials have also given rise to concerns about human exposure to these
materials. People who may be exposed include not only those who are involved in
nanomaterial manufacturing and handing but also consumers and end users of
products. An inadvertent environmental exposure is also a possibility; there is
growing evidence that under certain conditions nanomaterials can be mobile in soil,
air and water [11-13]. The environmental fate and transport of the nanomaterials are
indeed affected by their characteristics (e.g., size, charge, and agglomeration rate)
[14].

It is important to address environmental and health related issues and uncertainties
while this technology is still in early stages of development. Thus, research into

applications of nanotechnology should go hand-in-hand with pro-active health and

environmental implications studies.

1.1.1. Nanotechnologies for a cleaner environment

Nanotechnology has had a signiﬁcant impact on the development of novel
environmental engineering processes and technologies. The impact has been reﬂected
by the beneﬁts of using nanotechnology to deal with legacy environmental pollution
and to predict and prevent future enviromnental problems [7,15,16]. Here are some
speciﬁc examples. Nutt et al. [17] synthesized palladium/gold nanoparticles with
diameters less than 20 nm and found that such nanoparticles have a high catalytic
activity in the hydrodechlorination of trichloroethylene in water, which is a common
contaminant in groundwater. Mayo et al. [18] used magnetic properties of iron oxide
nanocrystals to efficiently absorb and desorb arsenic from drinking water.

Pummakamchana er al. [10] developed solid state gas sensors based on

nanotechnology with increased sensitivity in detecting air pollutants such as NOx.

1.1.2. Nanotechnology applications in water treatment and quality control

Water pollution has become one of the most serious environmental challenges that
humans are facing. Pollution can be due to long term routine contamination and
short-term catastrophic events. Produced water generated during offshore drilling
operations, municipal sewage discharge and agricultural runoff are examples of long
term water contamination. Oil spills, chemical Spills, and ﬂooding, despite the fact
that they are short term events, may have environmental impacts that are as severe as
or worse than long term routine contamination has. A 2008 report indicated that 884
million people lacked access to safe water supplies, which is approximately one in

eight people worldwide [19]. One of the urgent challenges of the environmental

engineering profession is to develop new technology to improve the quality of
drinking water to minimize the likelihood of water-related disease outbreaks. The fact
that water scarcity and related sanitary problems will become only more severe over
time makes it imperative for our society to seek solutions for improved water
treatment processes and for remediating wastewater to meet water quality standards.
We also need to achieve these objectives with high efficiency and at lowest possible
cost [20].

There are established wastewater treatment processes that can be and are used to
treat and reclaim polluted water [21]. The limited energy efﬁciency of existing
methods and, in some cases, concerns with respect to the quality of the treated water
bring questions of the sustainability of these traditional technologies. For example,
while chlorination is known to be effective in deactivating most pathogenic
water-bom microorganisms, it was determined that chlorination byproducts are toxic
and some are carcinogenic. To give another example, some of the conventional water
treatment processes are not effective at removing emerging contaminants (e.g.,
pharmaceuticals, surfactants) from municipal and industrial wastewater [22-24].

A range of accepted water treatment technologies (e.g., adsorption, ion exchange,
redox processes, membrane ﬁltration, and disinfection processes) have beneﬁted from
nanotechnology-enabled discoveries [25]. Nanotechnology has been used to improve
the effectiveness and efﬁciency of water treatment, remediate contaminated water and
to increase the sensitivity of detection of various water toxic contaminants. Some
speciﬁc examples of the use of nanomaterials to improve water quality are: microbial

disinfection with silver nanometallic particles [26], photocatalytical destruction of
biological toxins in wastewater using Ti02 nanoparticles [27], adsorption of heavy

metals from water using self-assembled functional ligands that form a nano-coating

[28], removal of toxic dissolvable metal (copper, uranium) by dendritic polymers [29],
removal of radioactive cesium with nanocomposite ﬁlm containing carbon nanotubes
[30], application of metal and oxide nanoparticles in the removal of pesticides [31],
and remediation of groundwater contaminated by chlorinated solvents and heavy
metals using reactive zero-valent iron nanoparticles [32].

The application of nanotechnology for water treatment and quality control is

expected to become a major industry over the next 10-15 years [20].

1.1.3. Potential environmental risks of nanotechnology

As brieﬂy outlined above, there are a growing number of nanotechnology-based
solutions to environmental protection and remediation of environmental pollution. At
the same time, there are also concerns that nanomaterials may pose environmental
challenges of their own. As noted by Breggin and Pendergrass [33], certain classes of
nanomaterials may, in the future, be considered hazardous waste due to their as-of—yet
unknown toxicological properties, creating potential legal liabilities for manufacturers,
investors, and insurers. So far there is insufficient evidence to support the broad claim
that nanoparticles with undesirable effects (toxicity, and radioactivity, for example)
constitute a new class of non-biodegradable pollutants [34]. However, it has been
determined that if removal of nanomaterials from the environment (soil, water and air)
is necessary, it would be difficult to accomplish due to the extremely small Size and,
in some cases, high mobility of nanomaterials.

It is clear that a better understanding of nanoparticles’ interactions in the
environment is necessary. Studies have been published focusing on the interactions of
nanomaterials with microorganisms, which are an integral component of

environmental media such as soils and surface waters. There is already some evidence

showing that nanoparticles may have an adverse effect on microorganisms depending

on the physicochemical factors (Size, charge, morphology) and material properties of

the nanoparticles. For example, in the study of antibacterial activity of aqueous nC60

suspension with bacterium Bacillus subtilis, larger nC60 particles (> 100 nm in

diameter) appeared to be 100 times less toxic than smaller (< 100 nm) particles [3 5].
Another study Showed that cationic fullerene derivatives were bacteriostatic while
anionic derivatives were not, and that the surface coating on quantum dots of CdS and
CdSe is relevant in determining their toxicity [36]. Another ecotoxicity study has
already found that nanoparticles may affect marine organisms by contaminating their
food web; it has been demonstrated that gold nanorods can easily be transferred from
the water column to the estuarine food web [37]. Further, it has been shown that
nano-silver can disrupt key soil microbial communities [3 8].

In summary, given the uncertainty with respect to the balance between useful
applications and adverse impacts of nanotechnology, the environmental effects of
engineered nanomaterials and nanomaterial-based technologies should be proactively

assessed prior to their mass introduction to the markets.

1.2. Membranes nanotechnologies

Nanotechnology has great potential in the development of novel membrane
materials (with improved permeability and selectivity, with additional functionalities,
etc) and for the development Of novel approaches to the mitigation of membrane
fouling, which is the most egregious problem precluding wider applications of

membranes in water treatment.

1 .2. l. Nanoparticle-enabled membranes

The incorporation of nanoparticles into reverse osmosis [39], nanoﬁltration [40]
and ultrafrltration membranes [41,42] has already been shown to be effective in
improving membrane properties. Nanoparticle-enabled membranes combine
traditional membrane materials and preparation methods with novel nanomaterials
and methods. Two typical methods used to incorporate nanoparticles into the
membrane matrix are: decoration of the membrane surface with nanoparticles [43,44]

and incorporation of nanoparticles within the membrane matrix during membrane

preparation [45]. For example, membranes self-assembled with Ti02 nanoparticles

have been used to modify both ultraﬁltration [44] and reverse osmosis [43]
membranes by self-assembly technology; such membranes have been used as
photocatalytic membranes to oxidize organic contaminants and to reduce membrane
fouling by these organics. As another example, membranes modiﬁed with aligned
carbon nanotubes were used to remove 25 run sized polio viruses from water as well
as larger pathogens such as E. coli and Staphylococcus aureus bacteria [46].

Nanoscale silver particles have been incorporated into ultraﬁltration membranes to

inhibit membrane bioﬁlm growth [42]. Titanium dioxide (TiOz) nanoparticles have

also been embedded with polyamide membranes to decrease biofouling [43].
Generally speaking, nanoparticle—enabled membranes can be designed to be

multifunctional with enhanced features such as functional/catalytic properties with

respect to certain reactions, increased resistance to fouling potential and improved

chemical, mechanical or thermal stability [20].

1.2.2. Novel nanoscale membrane coating — polyelectrolyte multilayer ﬁlms

Polyelectrolyte multilayer ﬁlms (PEMs) are prepared by a layer-by-layer adsorption

of polycations and polyanions. These ﬁlms have been applied on different types of
substrates including membranes. These nanoscale surface ﬁlms have a controllable
molecular architecture [47,48] and are typically in the nanoscale thickness range. The
resultant ultrathin membrane skin allows a high permeation rate [49] and can be used
to alter the surface chemistry of the membrane.

The control over surface chemisz (e.g., roughness, charge, hydrophobicity) of the
membrane is important from the fouling standpoint. For example, the major foulant of
nanoﬁltration membranes — natural organic matter (NOM) — fouls hydrophilic
membranes less than hydrophobic membranes [50]. Similar effects of membrane
hydrophilicity have also been observed in protein ﬁltration experiments [51]. Flux
reduction during the removal of colloids [52,53] and toxic chemicals [54] has been
correlated to membrane roughness. Surface charge of membranes is an important
factor in controlling salt rejection by membranes [55,56].

The development of PEM ﬁlms is an example of the application of nanotechnology
to control membrane surface properties. The interactions of PEM ﬁlms with proteins
[57—66], mammalian cells [62,63,67-71], and bacteria [64,68,72,73] have been studied.
The attributes of the PEM surface do affect the interaction with these substances. For
example, it was found that a PEM surface with higher roughness results in lower
adhesion of human gingival ﬁbroblast cells [74], while another study found that a
PEM ﬁlm with lower stiffness results in lower cell adhesion [70].

In addition, PEM ﬁlms have also been used to promote self assembly of proteins

[75], dyes [75], nanoparticles (e.g., quantum dots [75], polystyrene [76-78], ferritin
[79], Si02 [80], latex particles [SI-83]), and microsized particles (e.g., 0.5-1 um Si02

[84]). Depending on the combination of absorbent and PEM ﬁlms, self assembly

processes can be dominated by short-range hydrophilic interaction [75], surface

roughness [76,77], or surface charge [76,77]. When using a polyelectrolyte coating to
prepare a surface for the self assembly of nanoparticles, the aggregation state of
nanoparticles will be different depending on the preparation methods of the PEM
coatings, and water chemistries of nanoparticle suspensions and polyelectrolyte
solutions. Such dependence can be reﬂected by the fact that the same type of
polyelectrolyte ﬁlm surface (e.g., PEM ﬁlm with poly(diallyldimethyl ammonium
chloride) as the top layer) will result in either monolayers of nanoparticles [80] or
aggregation of the nanoparticles [85]. The resulting form of self-assembly is chemical
dependent [84] and is also affected by the experimental washing and drying processes
[79]. A review paper has summarized the experiments and theory of the irreversible
adsorption of particles (such as latex particles, globular proteins) on heterogeneous
surfaces such as PEM-modiﬁed surfaces [86].

Membranes coated with PEMs have been designed to provide a selective barrier to

aqueous ionic species [87]. A Cl-/SO42- selectivity of 30 has been achieved with 4.5

bilayers PSS/PAH (poly(styrene sulfonate)/ poly(allylamine hydrochloride» ﬁlms
[88]. When changing the PEM compositions to FAA/PAH (poly(acrylic acid)/PSS),
the selectivity could be increased to 80 by sacriﬁcing half of the solution ﬂux [88].
These results indicate that it is easy and feasible to control membrane properties by
changing the compositions of the PEM ﬁlm. High separation performance of mono-,
di-, and multivalent cations and anions [88-95] or neutral molecules [96-99] has
achieved with PEM modiﬁcation on a surface [88,90,91,94,95] or within the inner
pore structure [100] of a porous membrane support. Besides having a controllable
surface chemistry, the PEM ﬁlms have another attractive feature that they are
sensitive to chemical conditions [101], which allows the ﬁlm dissociate from the

support under certain chemical conditions and a new ﬁlm can be reapplied to the same

support.

1.3. Nanoparticle interactions in aqueous media as a key to

elucidating environmental impacts of nanomaterials

Interactions of nanoparticles in the natural water matrix determine nanoparticle
transport, fate, and ecotoxicity in aqueous environments. The likelihood of
nanomaterials entry into the environment is increasing with the growth of the
nanotechnology industry. The natural bodies receive runoff and effluent from
wastewater treatment facilities, and nanomaterials may be present in each of these
sources. In addition, nanomaterials are increasingly used in environmental
remediation and can enter the environment via this route [102]. At this point the
understanding of the transport and fate of nanomaterials in aquatic environments is
lacking [103].

Aggregation and deposition are the two main processes that control the transport
and fate of nanomaterials in aqueous conditions [104]. The major principles of colloid
science (e.g., Smoluchovsky’s equations and DLVO theory [IDS-107]) may still be
applicable for engineered nanoparticles [108], except that the BrOnsted concept needs
to be applied for nanoparticles that are smaller than 10 nm [108].

The processes of aggregation and deposition of nanomaterials are governed by their
physicochemical properties, such as particle size, particle charge, morphology, and
hydrophobicity [109]. These physicochemical properties of nanoparticles suspended
in water can be a function of the chemical makeup of the water (i.e., ionic strength,
pH) and can affect the stability of nanoparticles [110-112]. In a recent review, Wiesner
et al. pointed out that it is not always true that the small size of nanoparticles will

yield a high mobility in porous media (e.g., groundwater aquifer) [103]. Their

10

relatively large diffusivity, a direct consequence of their small size, would also
enhance their chance to attach/deposit on the surfaces of porous media and thus lose
their mobility. Lecoanet et al. studied the mobility of fullerene-based materials in

porous media and showed that single-walled nanotubes and fullerol particles can pass

through the porous medium more rapidly than nC60 nanoparticles [113]. The

relatively slow transport of nC60 was attributed to its hydrophobic interactions with

porous media.

The aggregation and deposition of nanomaterials in aqueous media would also
affect their bioavailability and ecotoxicity. Farre’s review paper on ecotoxicity of
nanomaterials [102] summarized the studies on ecotoxicity of nanomaterials such as
silver nanoparticles [114,115], fullerene [116,117], and single—walled carbon
nanotubes [118-120]. These materials elicited toxic response in species such as

zebraﬁsh, Daphnia magna.

11

Chapter 2. Polyelectrolyte multilayer ﬁlms used

as renewable membrane coatings with tunable

surface properties

This work has been recently published in the Journal of Membrane Science [121].

2.1. Introduction

Polyelectrolyte multilayer (PEM) ﬁlms are prepared by alternately adsorbing
oppositely charged polyelectrolytes onto supports using a layer-by-layer technique
[122-124] and can serve as regenerable surface coatings with controllable
physicochemical properties (e.g. charge, hydrophilicity, swellability, stiffness) that
regulate adhesion to the surface [65,125,126]. The resistance to adsorption of different
substances (e.g., proteins, mammalian cells or bacteria) and the selectivity to aqueous
ionic species of PEM ﬁlms were discussed in section 1.2.2.

When assembled on a surface [88,90,91,94,95] or within the inner pore structure
[100] of a porous membrane support, PEMS can function as nanoﬁltration
[88,90,91,94,95] or reverse osmosis [93] membranes to separate mono-, di-, and
multivalent cations and anions [88—95] or neutral molecules [96-99]. For example,
membranes composed of ﬁve bilayers of poly(styrene sulfonate)/poly(allylamine

hydrochloride) (PSS/PAH) on porous supports allow a high ﬂux at regular
nanoﬁltration pressures and exhibit 95% rejection of MgClz along with a NaJr/Mg2+

selectivity of 22 [127]. As another example, 4.5-bilayer PBS/PDADMAC (poly

12

(diallyldimethylammonium chloride)) ﬁlms on porous supports Show Cl-/F- and

Br-/F- selectivities larger than 3 along with solution ﬂuxes that are 3-fold higher than

those of commercial membranes [128].

The PEM approach to membrane design is highly versatile in that separation and
antiadhesive properties of PEMS can be adjusted through the choice of the constituent
polyelectrolytes, the number and sequence of polyelectrolyte layers in the ﬁlm, and
the deposition conditions (solution pH and ionic strength, e. g., [91]). A unique
advantage of some PEM membranes is that when appropriately constructed, the ﬁlm
can be removed from the porous support via exposure to solutions with high pH
values (IO-12) or ionic strengths [129-131]. The ﬁlm can then be regenerated through
the layer-by—layer process. Thus, a PEM ﬁlm assembled on a UF support can combine

separation and antiadhesive properties with the ability to renew the surface (Fig. 2.1).

PEM , PEM
deposition Fouling removal

          

 

   

I ‘VS'PV‘V‘ i "ECO “ELIE-Ll; l n " ”7-CZQC>C~C> ‘ l. T '
"""""' :00 cm“ : ’ C‘J ‘
((38% 0,6892%; 3885s; 0 (38%
,’ PEM-based NF UF membrane
UF membrane i; PEM-based membrane (support)
(support) / NF membrane (fouled)
" l
.6. , __ Salt-rejecting layer (antiadhesive,
sacriﬁcial PEM) .-.:.:.‘.: new.“
aspirin '5’
' DC“ r" a T A
Mrcroporous layer Qtﬁgﬁﬁa
(support)
PEM-based
NF membrane
(regenerated)

Figure 2.1. Conceptualized drawing of the application and regeneration of a

polyelectrolyte multilayer (PEM) nanoﬁltration membrane.
The support thickness is not drawn to scale.

Nearly all studies on the separation properties of PEMs employed synthetic feed

solutions with only one or two compounds in solution. The performance of PEM

l3

membranes challenged by suspensions of colloids (other than proteins) and
regeneration of fouled PEMS have not been investigated. Very little is known about
the effects of operational variables on PEM separation properties. In the only
published study on the topic, Tieke and coworkers examined rejection with a very
dense 60—bilayer poly(vinylamine)/ poly(vinylsufate) PEM membrane as a function of
transmembrane pressure [93]. They observed pronounced concentration polarization
effects, but a rigorous analysis was not possible because the study was conducted in a
dead-end geometry. Importantly, while PEMS have been effective in simultaneously
increasing protein retention and reducing protein adhesion [58], there have been no
reports on the design of PEM membranes that combine resistance to colloidal fouling
with desirable ion separation properties. This study aims at ﬁlling some of these
knowledge gaps. Speciﬁc objectives of this work are:

1) to evaluate water and solute permeabilities of a diverse set of PEM membranes
to understand the dependence of rejections on pressure and solute concentration;

2) to determine how the permeability and rejection of PEM membranes are affected
by concentration polarization and colloidal fouling;

3) to assess PEM regeneration alternatives, including backﬂushing, for as-prepared

PEMS and PEMS fouled by colloids.

2.2. Background and approach

2.2.1. Colloidal fouling

Although source waters usually undergo pretreatment that is designed to remove
the colloidal fraction, in practice both dissolved and residual suspended phases are
present in nanoﬁltration (NF) or reverse osmosis ﬁltration (RO) feed waters [132,133].

The colloids that persist through the pretreatment stage can form membrane deposits

14

with a high hydraulic resistance on membrane surfaces, contributing to the permeate
ﬂux decline. The size of a colloid has a signiﬁcant effect on its diffusivity and thus
fouling potential. It was experimentally determined [134] and mathematically
calculated [135] that the minimum in particle back-transport diffusivity is occurred
with colloids with a diameter of approximately 100 nm, and thus particles of this size
have the highest membrane fouling potential (see Figure 2.2). Therefore, a colloid of
this size yields a minimum in back-transport diffusivity and thus a maximum in
deposition propensity [136]. In addition, colloids deposited on a nanoﬁltration
membrane surface tend to inhibit back-diffusion of dissolved species, resulting in
precipitation of sparingly soluble salts in the pores of the deposited layer, and also
causing cake-enhanced osmotic pressure [137]. Colloids cause fouling either
externally, on the membrane surface (a build-up of a cake/gel-like layer on the
upstream face of a membrane) or internally, inside of the pores of a membrane [138].
When the colloids get into the internal pores of the membrane, they are hard to

remove and this condition is considered as permanent/irreversible fouling.

        

2;. -2 — Diffusion
E -3 - Diffusion : dominated by shear
é dominated by ' Induced motion
5 g '4 — Brownian motion : _—)
2 52 -5 _ ( I
.0 NE
E o I
m V -6 -1 I
Q l
3’ -7 - I
_.r
-3 , , ! , . .

 

 

0.1 1 10 100 1000 10000
Particle racﬁus (um)

Figure 2.2. Particle diffusivity as a function of radius for conditions typical of
hollow ﬁber UF membranes [136]

15

2.2.2. Approach
Nanoﬁltration properties of PEM-coated membranes were compared with the

corresponding properties of a commercial membrane, NF270. Firstly, permeability to

deionized water and MgSO4 rejection were measured for all membranes in crossﬂow

ﬁltration experiments. Solute transport to the membrane surface and transport across
the membrane were modeled using a thin ﬁlm model and Kedem—Katchalsky

equations, respectively. Measured values of the permeate ﬂux and the concentration

of solute in the permeate were used to determine the MgSO4 permeability coefﬁcients

of the membranes, which is an important criterion for nanoﬁltration membranes. By
performing the crossﬂow experiments over a range of transmembrane pressure

differentials, a range of concentration polarization conditions were tested, and the

MgSO4 permeability coefﬁcients as a function of salt concentration were recorded for

each membrane.

Secondly, the performance of membranes under conditions of colloidal fouling was

evaluated with experiments on the crossﬂow ﬁltration of Si02 colloids suspended in
an electrolyte solution. Based on i) the measured values of permeate flux and Mg2+

concentration in the permeate and ii) the previously determined MgSO4 permeability

coefﬁcient as a function of concentration, the resistance to the permeate ﬂow due to
the deposited layer of colloids was computed for each membrane and used as a
measure of the extent of colloidal fouling.

Determingtion of intrinsic rejection

To characterize the selectivity of membranes in the colloidal fouling study, MgSO4

rejection was measured. The mass transfer coefficient, k , for MgSO4 in the

16

membrane cell channel was estimated from the Sherwood correlation ([139]):

_1_ (2.1)
Sh =-k—d’l = [3.663 +1.613Re-sc.d—"]3
D L

where Sh is the Sherwood number,L is the channel length, D is the diffusion

coefﬁcient of MgSO4 in water (8.5 -10_10m2/s [140]), Re is the Reynolds number,

So is the Schmidt number, and d h is the hydraulic diameter of the channel. For the

membrane cell used in this study, d}, z2h, where h is the channel height. The
value of [Re - Sc %) is a criterion in using the Sherwood correlation of eq. 2.1, and

the value is computed as follows:

dh J _ ,0th 77 db _ thz (2'2)

where v is the crossﬂow velocity. The value of [Re-50%] was in the (9,020 to

10,300) range, which was close to the upper bound of the applicability range

[0.1<(Re-Sc-%h—]<10,000] ofeq. 2.1.

The thin ﬁlm model was used to estimate the concentration of MgSO4 at the

membrane surface, Cm [141]:

—C 2.3
E"__P;=exp(‘lj ( )

Cf—Cp k

and to determine the intrinsic rejection, R-

m , as a function of the permeate ﬂux, J ,

l7

for all permeate sampling times:

p :1_ cp (2.4)

J
Cp+(Cf—Cp)-exp[7(—]

where C p, C f , and Cm are the concentrations of MgSO4 in the permeate, in

the bulk feed, and in the portion of the feed directly adjacent to the membrane,
respectively. In this study, permeate ﬂux and salt rejection were measured at different

transmembrane pressure differentials, AP , and values of Rm were determined as a
function of J using eq. 2.4. These experiments were conducted using colloid-free

MgSO4 electrolyte as the feed.

Determgrgtion of solute permeability coefficient
The coupled transport of solute and solvent in a membrane is described by the
Kedem-Katchalsky equations [142]:

J = Lp(AP—0'A7r) (2.5)

J, = wArr+(l-O')JC (2-6)

where J and J S are volume and solute ﬂuxes across the membrane, respectively,

L p is the hydraulic permeability of the membrane, 0' is the reﬂection coefﬁcient,

and a) is the solute permeability. Volume ﬂux is given byJ = JWVW +J 5175, where

VW and I73 are molar volumes of water and solute, respectively and JW is the

permeate water ﬂux across the membrane. In eq. 2.6, C is the logarithmic mean of

the average concentration of solute within the membrane:

18

C=(C,,, —Cp)-1n’1(c,,, mp) (2.7)

For large volume ﬂows and high concentration gradients across the membrane, the
changing concentration proﬁle inside the membrane can be taken into the account by

recording eq. 2.6 in the differential form. Noting that

J, = JCP (2.8)
and substituting the expression for the local osmotic pressure differential
An' = ngoRT[C(x)—C(x+Ax)] (2.9)

into eq. 2.6, one obtains the Kedem-Katchalsky expression for the volume ﬂux across
a differential element of the membrane:

JCP = —F-‘£+(1—0)JC (2'10)
dx

where P = n¢RTcoAx is the local solute permeability coefﬁcient, n is the total
number of constituent ions in the salt (n = 2 for MgSO4), C now has the meaning
of the logarithmic mean of the average concentration Of solute within the differential

element, and (15 is the osmotic coefficient, which is generally a function of the solute

concentration [143] and can be calculated using the Pitzer equation [144].
Note that the physical meaning of a" can be deduced from the Spiegler-Kedem
relationship [145] that is obtained by integrating the differential form (eq. 2.10) of the

Kedem-Katchalsky expression (eq. 2.6) across the membrane in the presumption of

concentration-independent phenomenological coefﬁcients 0' and P:

19

=0(1—F) (2.11)

 

_ 2.12
erxp[_.l(1 0)] < >

where PS is the solute permeability coefficient (139 = n¢RTro = g). It follows from

eq. 2.11 and 2.12 that the reﬂection coefﬁcient, 0' , represents the limiting value of
the intrinsic rejection achieved at J —> 00.

Experimental evidence indicates that If, is generally concentration-dependent
(e.g., [146-149]). Accordingly, in this study we used the Kedem-Kachalsky model (eq.
2.6) to determine the MgSO4 permeability coefﬁcient as a function of concentration,
PS = 11(C). To compute 13, = IMC) , eq. 2.6 was used in its modiﬁed form:

J, =PS(Cm—Cp)+(1—0)JC (2.13)

In deriving eq. 2.13 the following expressions for the solute permeability and

 

osmotic pressure differential were used: a): P3 and Arr = n¢RT(Cm —Cp). By
ngoRT

incorporating eq. 2.7 and 2.8, eq. 2.13 can be rewritten as:

Cm —CP (214)
JCp =P,(c,,, —Cp)+(l-a)J——————
1n(c,,, /C,,)

The refection coefﬁcient 0' was presumed to be concentration independent and its

value was approximated by the maximum value of intrinsic rejection, R513" . More

Speciﬁcally, the value of 0' was obtained by measuring rejection at the highest

20

 

experimental permeate ﬂux in each conditioning experiment and calculating R5,?“
from eq. 2.4. With 0' available, the only unknown in eq. 2.14 is 19,. By measuring
J and Cp in an experiment on the ﬁltration of colloid-free electrolyte (i.e.
membrane conditioning stage of experiments; see stage 3 in section 2.3.5) and
computing Cm using eq. 2.3 and the Sherwood correlation (eq. 2.1), we applied eq.
2.14 to determine PS for the given Cm. By performing the above procedure at a
series of different transmembrane pressure differentials AP (and, correspondingly,
different permeate ﬂuxes, J), the dependence P5 = PAC) was determined for a
wide concentration range for each membrane.

To evaluate the error introduced into the computation of P5 = R,(C) by presuming

the reﬂection coefficient to be concentration independent and equal to the value of
intrinsic rejection at the maximum permeate ﬂux, we evaluated the ratio, G, of the two
terms on the right hand side of eq. 2.13 (note that this ratio is a measure of the

importance of the coupling between solvent and solute transport).

_ (l-a)J5 (2.15)
"‘ PAC," —Cp)

 

By determining this ratio, we can have the idea of the possible error introduced into

the computation of P, =PS(C) by presuming the reﬂection coefficient to be

concentration independent.

Quantifying concentration polarization and resistance of colloidal cake in colloidal

fouling experiments

Under conditions of colloidal fouling, concentration polarization is enhanced due to

21

the formation of a colloidal cake that hinders back-diffusion of rejected salt [137,150].
This process is called cake enhanced concentration polarization and typically occurs
for salt rejecting membranes. The accumulation of particles forms a barrier to the
backtransport of salt species and as a result, a higher osmotic pressure will be created
due to the increased difference in salt concentration on the upstream face of a
membrane and the salt concentration in the permeate. With mass transfer correlations
such as eq. 2.1 available only for well-defmed geometries and with the value of the

hindered diffusion coefﬁcient in the cake not known, thin ﬁlm theory (eq. 2.3) can no

longer be used to determine Cm . To overcome this difficulty, we ﬁtted the
experimental Ps(Cm) data obtained in experiments on the ﬁltration of colloid-free
electrolyte solution (see Fig. 2.7) to determine an analytical R,(Cm) expression for
different PEM membranes as well as the NF270 membrane. Then, eq. 2.14 was used
to determine the value of Cm for each sampling time in experiments on colloidal

fouling (awas assumed to be R3,)” as mentioned above). Dividing Cm by the

concentration of MgSO4 in the feed, C f, gives the concentration polarization factor,

Cm / C f , which can be determined at different times during the colloidal ﬁltration

experiment.

The hydraulic resistance exerted by the colloidal deposit, Rd , was computed using

the following equation for the permeate ﬂux:

_ AP-OAﬂm (2.16)
#(Rm + Rd)

where AP: Pb —Pp is the pressure differential between the bulk feed and the

22

is the transmembrane osmotic pressure differential, R

permeate, Arrm = 7r," —7r m

P

is the hydraulic resistance of the membrane, and ,u is the water viscosity.

2.3. Experimental

2.3.1 . Materials

All reagents were of ACS analytical grade or higher purity and were used without
further puriﬁcation. Ultrapure water was supplied by a commercial RO/DI system
(LabFive, USFilter Corp., Hazel Park, MI) equipped with a terminal 0.2 pm capsule
microﬁlter (PolyCap, Whatrnan Plc., Sanford, ME). The resistivity of the water was
greater than 16 MO -cm.

Silica colloids (SnowTex-ZL, Nissan Chemical America Corp., Houston, TX) were
received in the form of a concentrated suspension. Dynamic light scattering (BI-MAS

particle sizing module, ZetaPALS, Brookhaven Instrument Corp., Holtsville, NY) was

used to measure the particle size distribution in SiOz suspensions. Samples were

diluted with 0.1 mM MgSO4 to reach the recommended count rate. The Q-potential
of particles was measured by a zeta potential analyzer (ZetaPALS). The pH of the
Si02 suspension in 0.1 mM MgSO4 was in the 6.1—6.6 range. Si02 colloids were (140

:t 1) nm in diameter with a 6 -potential of (-30 d: 5) mV.

Magnesium sulfate salt was used to evaluate the salt rejection in the presence of

. . . . 2+ . .
collords ﬁltering wrth different membranes. Mg concentrations were deterrnrned

using ﬂame atomic absorption (AA) spectroscopy (Perkin-Elmer 1100, Waltham,

MA). A stock solution of lanthanum chloride (Fisher Scientiﬁc) was added to all

samples and standards to achieve a LaCl3 concentration of 0.1% by weight. The

23

[Tr—J

calibration range for Mg2+ concentration was (0.1 to 0.6) mg/L.

2.3.2. Preparation of PEM membranes

A polyethersulfone (PES) membrane (Pall Corp, East Hills, NY) with a MWCO of
50 kDa served as the support for PEM ﬁlms. The permeability of this UF membrane
is considerably higher than that of the PEMS, but its surface porosity is sufﬁciently
low to allow complete coverage of the support (i.e. complete bridging of the support’s
surface pores by the polyelectrolyte molecules) by PEMS with only a few adsorbed

bilayers [94,151]. Prior to the PEM deposition, the support membranes were soaked
ﬁrst in 0.1 M NaOH for 3 h and then in deionized water for 24 h at 4 0C with water
exchanged after the ﬁrst 12 h of storage, as recommended by the manufacturer.
Anionic and cationic polymers were alternately adsorbed on the UF substrate by
immersing the substrate in the corresponding polyelectrolyte solutions with a 1 min

water rinse after the deposition of each layer. Table 2.1 speciﬁes the conditions for

deposition of each layer for all of the polyelectrolyte membranes in this study.

PDADMAC (MW = (100,000 to 200,000», PAH (Mw = 70,000), and P88 (MW

=70,000) were purchased from Aldrich, and PAA (MW = 90,000, 25% aqueous

solution) was obtained from Polysciences. Figure 2.3 shows the structure of the

polyelectrolytes used in the study.

24

Table 2.1. Ionic strength (10) and pH of the PEM solutions used to prepare PEM

 

mn
u Fuel—o he:

a
Augean—ion :5» v

an
N .5630 \uz

o
Ann—83mg :50 v

o
Hanamauman Swan :0 :5

m
Amine—ﬁne: 35o V

 

 

 

 

 

 

 

 

 

. . a a a
. Emma—633.5: amazes eo>oz>Qoasa 3202.352 .3
a 35 a 35 a 95V
. . . Emacs; $5030 EVENSO
_. ~1mm\ nv >55. .. AN 335 Am 3.5.—v G 3.5.;
. . .._. m . Ewan? SEBSO Emcee;
.emm%>5...u s 35 a 35 AN 95
Emacs; $5350 vivasa
Fair—5:35. 3 SE a 55 a is
_ o
.wmmagsrrﬁEE. Emacs; 3.22.505 viii
6 3 3.3 G :33 .
H u . G 3.3

 

 

membranes. The adsorption time for each layer is listed in parentheses.

a The subscript of 4.5 means that one single layer of PS8 was deposited on top of the four

[PSS/PAH] bilayers.

The pH of the PAH solution used in depositing the [PAA/PAH]1.5 was 3.5. The subscript of

1.5 means that one additional layer of PAA was deposited on top of one bilayer of

[PAA/PAH].

25

The ionic strength of the solution used for depositing the ﬁrst PAA layer in [PAA/PAH]1_5
was 0.5 mol/L.

‘1 The pH was that of the as-prepared PEM solution and was not adjusted.

e The ionic strength of the solution during the deposition of the terminating layer was increased
to 2.5 mol/L [88,127] except when PDADMAC was the terminating layer, because
PDADMAC ﬁlms dissociate at high ionic strength [131].

f For strong polyelectrolytes, 3 min (for PDADMAC) or 2 min (for PSS) adsorption time was
sufﬁcient, while for weak polyelectrolytes such as PAH and PAA, a 5 min adsorption time was

used.

(a) 0))

/""\°" 19::

(C)

if

803' Na"

((1)

ti;
H3+ cr

Figure 2.3. Structure of polymers used in PEM ﬁlms (a) PDADMAC, (b) PAA, (c)
PSS, ((1) PAH.

The polyelectrolyte solutions were prepared at a repeat unit concentration of 0.02

mol/L, and the pH and ionic strength of polyelectrolyte solutions were adjusted using

26

1M HCl, 1M NaOH and 1M NaCl solutions. The ionic strength of the solution during
the deposition of the terminating layer was increased to 2.5 mol/L [88,127] except
when PDADMAC was the terminating layer, because PDADMAC ﬁhns dissociate at
high ionic strength [131]. The deposition was always initiated with PSS to ensure the
attachment of the multilayer membrane to the PES support due to hydrophobic

interactions between PSS and PBS [94]. The additional PAA-containing layers of

[PSS/PAH]2+[PAA] and [PSS/PAH]2+[PAA/PAH]1_5 were added to increase the

hydrophilicity of these membranes.

In order to compare the performance of PEM ﬁlms to commercially available
nanoﬁltration membranes with similar water permeability during colloidal fouling,
coupons of commercial polyamide thin-ﬁlm composite NF270 membrane (FilmTec,
Dow Chemical Company, Midland, MI) were cut from the as-received membrane
sheet and soaked in ultrapure water for 24 h at room temperature prior to being

characterized and used for ﬁltration.

2.3.3. Membrane characterization

The streaming potentials of membranes were measured using an electrokinetic
analyzer (BI-EKA, Brookhaven Instrument Corp., Holtsville, NY). Before the test,
membranes were soaked in deionized water for 24 h. The KCl (pH 4) electrolyte

solution used in these measurements had an ionic strength of 0.4 mM, which was the

same as that of the 0.1 mM MgSO4 solution used in the ﬁltration experiments.
To examine the hydrophilicity of the membranes, water contact angles were
measured using a FT A 200 contact angle analyzer (First Ten Angstroms, Portsmouth,

VA). A 5 uL drop of ultrapure water was formed on the tip of a stainless steel syringe

needle and placed onto the membrane surface by raising the membrane until contact

27

was made. An image of the drop was taken two seconds after the drop formed on the
surface, and the left and the right contact angles were determined. To measure surface
energy of different PEM membranes, contact angle measurements were also carried
out with the other two different probe liquids - glycerol (Columbus Chemical
Industries, Inc., Columbus WI) and diiodomethane (Spectrum Quality Products Inc.,
Gardena, CA). At least three membrane coupons were tested with ﬁve images taken
for each membrane.

Scanning electron microscope images were recorded using a Hitachi S-47OOII ﬁeld
emission SEM operated in ultrahigh resolution mode. Samples were mounted on
aluminum SEM specimen stubs and made conductive by sputtering pure osmium
(NEOC-AN, Meiwa Shoji Co. Ltd, Japan) on the samples for 30 S at a current of 10

mA.

2.3.4. Bench-scale crossﬂow ﬁltration system

The schematic of the crossﬂow ﬁltration system is shown in Fig. 2.4 The high
pressure membrane ﬁltration cell (Sepa CF 11, GE Osmonics, Minnetonka, MN) in a
medium/high fouling conﬁguration was pressurized using an external hand pump
(P19-1000, SPX powerteam, Rockford, IL). A positive displacement pump
(Hydra-cell M-03, Wanner Engineering, Minneapolis, MN) equipped with a ﬂow rate
control unit (model MM231001C, Minarik Drive, South Beloit, IL) was used to
deliver the feed water to the ﬁltration cell, and a pulsation dampener (Hydra-cell
110-065, Wanner Engineering, Minneapolis, MN) was installed immediately
downstream from the pump outlet. A back pressure regulator (BP-3, Circle Seal
Controls Inc., Corona, CA) was used to maintain the transmembrane pressure

differential at a constant value in the range of (10 ﬂ: 5) psi ((0.07 i 0.04) MPa) to (250

28

:l: 5) psi ((1.72 d: 0.04) MPa). A pressure gauge (111.11, Wika Instruments Ltd.,
Oakville, ON, Canada) was installed in the retentate line. Two in-line digital
ﬂowmeters (101-7, McMillan Co., Georgetown, TX and L-lOOccm-D, Alicat
Scientiﬁc Inc., Tucson, AZ) were used to record the retentate and permeate ﬂuxes

every 30 s, respectively.

 

 

 

 

 

 

 

 

 

 

 

LabVIew ’
b data acquisition [:l I
mem rane s stem
module y I
~--- back pressure 5
regulator
r 7”:
pressure :
au 9 __;
pulsation [IF 9 g IE3}
dampener 1 ﬂowrneter flowmeter
. Elli ......................................................
variable ﬂow .

rate pump

...................................................................

 

      

stirrin . plate

Figure 2.4. The schematic of the crossﬂow ﬁltration system

Both permeate and retentate ﬂows were directed back into the feed tank, and small
(ca. 10 ml) samples of the feed suspension and the permeate were periodically
collected to determine the value of salt rejection. In all colloidal fouling experiments,
the retentate ﬂow rate was maintained at (1.04 i 0.07) L/min, which corresponds to a
crossﬂow velocity of ca. 0.1 m/s and a Reynolds number of ca. 353:1: 24. The data
from the ﬂowmeters were logged to the computer via a data acquisition module
(PCI-6221/SC-2345, National Instruments Corp., Austin, TX). The temperature of the

water in the 28.4 L (7.5 US gal) high density polyethylene feed tank (Nalgene

29

Labware, Rochester, NY) was maintained at (20.0 :t 0.5) 0C using a programmable

circulating chiller (model 9512, PolyScience, Niles, IL) with an external temperature
probe. The chiller circulated a 1:1 (vol: vol) mixture of distilled water and ethylene
glycol through a custom-made stainless steel chilling element immersed in the feed
tank. The temperature of the feed suspension was recorded, and the measured values

of permeate ﬂux were adjusted based on eq. 2.17 to account for the change in

viscosity due to the oscillations of temperature in the (19.5 to 20.5) 0C range. After

each ﬁltration experiment, the retentate and permeate lines of the crossﬂow system
were ﬂushed with detergent solution once and then with deionized (DI) water four
times. Equation 2.17 showed the correlation between temperature and permeate ﬂux

that we used in the computation.

. . 1.777 —0.052-(T—-273.15)+0.000625o(T— 273.15)2 (2- 17)
0.987

 

J=J

where J ' is measured ﬂux, J is the normalized ﬂux and T is the absolute

temperature in the permeate [152] .

2.3.5. Experimental protocol for crossﬂow ﬁltration
To make sure that the different PEM membranes and the commercial nanoﬁltration
membrane undergo the same ﬁltration procedures for comparison purposes, each
crossﬂow ﬁltration experiment was carried out in the following stages:
Stage 1. Membrane compaction
Ultrapure water was ﬁltered through the membrane for 24 h to ensure that
irreversible compaction would not contribute to the ﬂux decline observed in the

colloidal fouling experiment. The transmembrane pressure differentials during

30

compaction of membranes were set to exceed the pressures used in the fouling tests
(Table 2.2). The impact of the compaction of the UP support on the separation

properties of the overlying PEM layer was evaluated in experiments with

[PSS/PAH]4.5 as a representative PEM. It was found that the water permeability and

salt rejection were similar for both PEM membranes deposited on compacted and
non-compacted PES supports. In view of this result, the PEM membranes used in all
the colloidal experiments in this study were prepared by depositing the

polyelectrolytes onto uncompacted UF membranes.

Table 2.2. Transmembrane pressures and resulting permeate ﬂuxes during SiOz
ﬁltration by PEM membranes. Also indicated are transmembrane pressures used to
compact membranes prior to ﬁltration tests

 

 

 

 

 

 

 

 

 

 

 

Initial Initial speciﬁc
M b Transmembrane Transmembrane permeate permeate ﬂux,
em rane pI'CSSUI'C, pressure, ﬂux’ _11
AP (psi) AP (psi) J IJﬁ/SttPP-IO
-10‘5(m/s) ( a)
during . . .
compaction during colloidalfoulmg tests
[PSS/PDADMAC]4 40 11 2'8 36.3
[PSS/PAH]4 250 215 2‘7 1.9
[PSS/PAH)“ 200 110 3‘0 4.0
[PSS/PAH]2+[PAA] 120 52 3'1 8.8
[PSS/PAH]2+[PAA/PAH]1_5 200 110 2'8 3.7
NF 270 220 122 2.8 3.3

 

 

 

 

 

 

Stage 2. Measurement of membrgne hydraulic resistance
After compaction, pure water permeate ﬂux was recorded at several transmembrane

pressure differentials: 80 psi (0.55 MP3), 120 psi (0.83 MPa), 160 psi (1.10 MPa) and

200 psi (1.38 MPa) for all membranes except [PSS/PDADMAC]4. For

[PSS/PDADMAC]4, a sequence of lower pressures was used: 10 psi (0.07 MPa), 20

psi (0.14 MPa), 30 psi (0.21 MPa) and 40 psi (0.28 MPa). The hydraulic resistance,

31

Rmo , of the clean membrane to water was determined using linear least squares
ﬁtting of J (AP) from equation 2.18:

AP

J =
“R1710

 

(2.18)

Stage 3. Membrane conditioning and characterization

5 mL of 0.4 M MgSO4 was added to 20 L of ultrapure water in the feed tank to

adjust the magnesium concentration to 0.1-10‘3 M. The pH of the feed water was in

the (6.1-6.6) range. Membranes were conditioned by ﬁltering the electrolyte at the
same pressure used in the 24 h water compaction stage (stage 1) until the permeate
ﬂux stabilized (ca. 20 h). When the permeate ﬂux became stable after approximately
20 h of conditioning, the transmembrane pressure was changed in increments to

achieve different permeate ﬂuxes, J . During this process, permeate and feed water

samples were periodically collected to determine values of the MgSO4 rejection for a
series of different values of J. At the end of conditioning, the transmembrane

pressure differential needed to achieve a permeate ﬂux of (2.8:L.0.2)-10_5 m/s was

determined for subsequent use in the colloidal fouling experiment. Setting the initial
permeate ﬂux to the same value in all colloidal fouling tests ensured the same initial

colloid deposition conditions in experiments covering membranes with different

values of water permeability, L p .

Stage 4. Membrane fouling experiments

15 g of SiOz ST—ZL stock solution was added into the 20 L of feed electrolyte to

32

achieve a colloid loading of 300 mg/L. Crossﬂow ﬁltration of the colloid-containing

solution was then carried out for 20 h at the pressure determined in stage 3 to give an

initial ﬂux of (2.81‘0.2)-10_5 m/s. Small amounts of permeate and feed water were

collected periodically to determine observed MgSO4 rejection, RobS' Each ﬁltration

experiment was conducted twice, and the ﬂux proﬁles for the two membranes were

reproducible with a maximum deviation of 11%.

Stage 5. PEM regeneration and backﬂushing test

PEM removal and regeneration tests were performed with [PSS/PAH]4 and

[PSS/PAH]4.5 membranes only. To remove the PEMS from the UP support, the

PEM-coated support was immersed for 10 min in a pH 10 buffer solution containing

0.060 M NazCO3 and 0.596 M NaHCO3. To evaluate the inﬂuence of colloidal

deposition on the efficiency of PEM regeneration, both as-prepared and fouled PEM
membranes were immersed in the buffer solution. Backﬂushing step was added prior
to chemical treatment step to assist the regeneration of a new PEM ﬁlm on a fouled
PEM membrane. The experimental procedure was described as follows: i) membranes
were placed upside down in the membrane cell and backﬂushed with water for l h at
an applied pressure differential that was 30 psi higher than that used in the preceding
fouling experiments; ii) membranes were then soaked in the buffer solution and rinsed
with water; and iii) ﬁnally a new PEM was applied to the surface of the cleaned

membrane using the same layer-by-layer procedure as before. After each step of the

modiﬁed regeneration procedure, the hydraulic resistance and MgSO4 rejection were

measured.

33

2.4. Results and discussion

2.4.1. Charge, hydrophilicity, and water permeability of PEM membranes

As discussed in section 1.1.2, PEM technology provides a simple way to vary the
membrane surface chemistry in terms of surface charge, hydrophobicity and
swellability. Meanwhile, this alternation also allows membranes to have different
ﬁltration properties such as salt rejection and water permeability.

Table 2.3 summarizes properties of the ﬁve PEM membranes and the commercial
nanoﬁltration membrane (NF270) employed in this study. Both, the hydrophilicity
and the surface charge of PEM membranes were primarily determined by the choice
of the terminating polyelectrolyte and the ionic strength of the polyelectrolyte
deposition solution (Table 2.1). As expected, terminating the polyelectrolyte ﬁlm with
a polycation (PAH or PDADMAC) produced a PEM with a positive surface charge,
while terminating the ﬁlm with a polyanion (PSS or PAA) resulted in a negative
surface charge. To maximize the magnitude of the surface charge of the PEMS, the

terminating layer was deposited from a solution with a high ionic strength (2.5 mol/L)

for all membrane types except [PSS/PDADMAC]4. The reason we don’t use high

ionic strength for [PSS/PDADMAC]4 ﬁlm is that the ﬁlm gets dissociable at ionic

strength that is higher than 0.6 mol/L [131].
Additionally, Table 2.3 shows variations in the water contact angle among both

positively and negatively charged PEM ﬁlms. Notably, one PEM membrane,

[PSS/PAH]2+[PAA/PAH]1_5, was more hydrophilic and more negatively charged than

the commercial NF270 membrane.

34

Table 2.3. Surface and transport properties of the membranes

 

 

 

 

 

 

 

 

 

b

Membrane Contact angle a, Streaming potential, Hydraulic resistance ,
0(°) Cm ("M Rm09'1013(m-1)

[PSS/PDADMAC]4 76 i 6 21-4 i 0-9 0.27 :t 0.09
[PSS/PAH]4 34 i 3 13-4 i 0-3 4.3 i 0.7
[PSS/PAH]4_5 36 i 2 '10-8 i 0-7 2.9 :l: 0.4
[PSS/PAH]2+[PAA] 27 i 4 -0-9 i 0-6 0.84 a: 0.32

‘ [PSS/PAH]2+[PAA/PAH]1_5 20 i 3 -7-7 i 0-3 3.6 A 1.1
NF 270 29 it 3 -6.1 i 0.5 2.5 :t 0.2

 

 

 

 

a
Contact angle for a water droplet on the membrane surface.
The hydraulic resistance of the 50 kDa ultraﬁltration membrane used as the support for PEM

13 -
membranes was ca. RmO = 0.1 x 10 m after compaction.

Varying the composition of PEMS also modiﬁes the water permeabilities of these

ﬁhns. For example, PDADMAC-terminated coatings are known for their propensity

to swell [127,153], so [PSS/PDADMAC]4 ﬁlms Show the highest permeability to

water (Table 2.3) of all PEMS studied. The number of deposited layers also affects

ﬂux as shown by a comparison of [PSS/PAH]2+[PAA/PAH]1.5 and

[PSS/PAH]2+[PAA] ﬁlms (Table 2.3).

We also evaluated the surface energy components (Table 2.4) of different
membranes based on the contact angle experiments with three different probing

liquids. The high contact angle correlates to the high hydrophobicity of the

[PSS/PDADMAC]4 ﬁlm. The investigation of surface energy components for

different membranes Showed that [PSS/PDADMAC]4 has a low value of the electron

donor component of free energy ( 7‘ ), which is due to the high hydrophobicity of

35

PDADMAC polymer.

Table 2.4. Surface energy of probe liquids and different membranes

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Surface energy, mJ/m2
YLW 7+ 7— 7A3 TOT

Probe liquid

Ultrapure water 21.8 25.5 25.5 51.0 72.8
Glycerol 34.0 3.9 57.4 30.0 64.0
Diiodomethane 50.8 0.0 0.0 0.0 50.8
Membranes

NF270 43.52 0.25 50.43 7.03 50.56
[PSS/PDADMAC]4 46.12 0.01 8.94 0.64 46.76
[PSS/PAH]4 48.30 0.40 41.68 8.13 56.43
[PSS/PAH]4_5 48.71 1.08 33.12 11.95 60.66
[PSS/PAH]2+[PAA] 48.22 0.04 51.55 2.74 50.97
[PSS/PAH]2+[PAA/PAH]1_5 49.54 0.00 55.75 0.07 49.61

 

2.4.2. Determining the salt permeability coefﬁcient of PEM membranes

PEM membranes are used as nanoﬁltration membranes, therefore, the salt
permeability of such membranes is important from the point of desalination
applications. Here we use the Sherwood correlation and thin ﬁlm model to compute

the intrinsic rejection as a function of transmembrane pressure.

Expectedly, the intrinsic rejection, Rm , increased with an increase in the

36

 

transmembrane pressure and the corresponding increase in the permeate ﬂux (Fig. 2.5

(a)), although the observed rejection, Robs , decreased with an increase in the

permeate ﬂux for NF270 and all the PEM membranes (Fig. 2.5 (b)). This indicates

that a higher Robs can be achieved under a lower concentration polarization

condition by lowering the permeate ﬂux [127], or adding a spacer onto the membrane

to create turbulent ﬂow and thus reduce concentration polarization [132]. The increase

of Rm with higher ﬂux is due to an increase in the amount of water, relative to the

amount of salt, transported across the membrane at higher transmembrane pressures.
When considering observed rejection, its decrease with an increase in the
transmembrane pressure was due to the overcompensation of the better intrinsic
rejection at higher permeate ﬂuxes by the higher concentration polarization that led to
a higher salt concentration gradient across the membrane and, consequently, higher

salt ﬂux.

37

(a)

1.00 - """""""""""" .. ‘3‘;‘;3':'-"-'~'-‘O*‘Or'*“"'""".";'.".' """"""
.0 “ -EI
. A'A .3: . _ o ’
AA' . A- ' '
.A' ',~.
A a '
0.95 ‘ """""""""""""""""""""" .: 1.49:3 """""""""""""""""""
.s A ,r‘ﬁ
I! . ,I':
0.90 ~ --------------------------- . 'f ----- .3255"; - a - NF270

.' .' -ci- SSIPDADMACH
x 5; 1:3 --A- PssiPAI-u4

.' a ' .I ' O " PSSIPA 4.5
.‘ .' .° : - I - SSIPA PAA
' ' ' ,' - -A - SSIPA PA PAI-l]1.5
2

 

 

 

 

0.85 a .' (III. 1
o 50 100 150 200 50 300
2
J (le Ih)
b .
( ) 1.0 ‘l[ """""""""""""""""""""""""""""""""""""""""""""""
. ' '.'.-.-. _ . .
l ."' 1
0.8 """"""‘"""'"“"""""""'“'"""" “2" """"""""""""""""""
A.
M _ .......................... g -. ................... - 9- NF 270
.§ A.‘ - £1 - [PSS/PDADMAC]4
n: A.~ AA, - -A- [PSSIPAI-Il4
”-4 ‘ """""""""""" ~‘2'j """""" A‘ IA""- 0- [PSS/PAl-Ij45
0&9 _O ., a- [PSS/PAHJMPAA]
0,2 _ ........................ 113:: ........... A- [PssPAi-nzﬁpANPAr-urs
E1 “1|. :.
[:1 E9 ------- e ‘.
-EIIEfGli-rda‘I-RAIT: .-_ _
0.0 I I I I I l
o 50 100 150 200 250 300
2
J (le Ih)

Figure 2.5. Intrinsic (a) and observed (b) rejection of MgSO4 by NF270 and ﬁve
PEM membranes as a function of permeate ﬂux during ﬁltration of colloid-free

MgSO4 electrolyte.

The MgSO4 permeability coefficient, PS, of all membranes decreased with an

38

increase in the MgSO4 concentration near the membrane surface, C m (Fig. 2.6).

This decrease in PS somewhat mitigated decreases in the observed rejection due

to the concentration polarization. The concentration dependence of the MgSO4
permeability coefﬁcient for NF270 was reported earlier by Al-Zoubi et al. [148]; in

that paper P, was reported to increase with increasing MgSO4 concentration, which

is opposite to what we observed in our experiments. In our work, 0.19 < r < 0.83 for

all membranes, including NF270, for which I s 0.42.

 

 

 

‘°°°°1 +NF270
. --E--[PSSIPDADMAC]4
a j -- A [PSS/PAHJ4
@131 a --o-- PSS/PA .5
lion +~--—-—--— a ----I PSS/PA 2+[PAA]
E 'ffjgpi‘bil ------ [PSS/PAI-llzéi-[PAAIPAl-ﬂts
..03‘
q .4, ........ -----.-.---.._..";:.:-._._ ............................ ...... . .........
Om AAA-AAA. ““0~~--a
‘- ; A :
5 . .
O
0- 1o - ,,,,,,
1 i i
0.1 1

cm (MM) 10 100

Figure 2.6. Dependence of the MgSO4 permeability coefﬁcient on MgSO4
concentration at the membrane surface for NF270 and ﬁve PEM membranes.

For all membranes, the values of Rm determined for successively higher

transmembrane pressure differentials, AP , asymptotically converged to a value

close to 1 (Fig. 2.5), indicating that the reﬂection coefficient, 0', was close to 1.

Although the permeability of PEM membranes with respect to MgSO4 was higher

39

than that of the NF270 membrane (Fig. 2.6), PEM membranes were highly selective

(a z 1) so that nearly complete MgSO4 rejection by the membranes could in principle

be achieved at sufficiently high ﬂuxes if concentration polarization could be
minimized.
Fig. 2.7 shows experimentally determined and ﬁtted concentration dependence

(Cm) of the salt permeability coefficient (P,) and G ratio for each membrane. The
grey areas indicate the range of Cm where PS(Cm)correlations were applied in
computing Cm in corresponding colloidal fouling stage. Under the concentration

polarization conditions observed in our colloidal fouling experiments, G ratio (eq.

2.15) is relatively low: G301 for all membranes except [PSS/PAH]4 and

[PSS/PAH]2+[PAA/PAH]1.5, for which G S 0.2. Because R32” provides the lower

bound for o , the above values are upper bounds for G values for different
membranes; the true G values are even smaller. Therefore, the second term in eq.
2.14 is small in comparison with the ﬁrst term and the error introduced into the

computed salt permeability coefficientPS = P,(C) dependence is at most 17%. The

maximum error would occur if 0' is in reality 1 for the highest values of G. It
should be noted that by approximating (l-O')JC as (l——R,-'},lax )JC we

over-estimated this term and, therefore, under-estimated P, (see Fig. 2.6).

40

y=13.405x0'4254 o NF270
R2 = 0.9784 '

l

I
P
A
m

l

I
P
O
(’1

 

 

 

 

2500 ------- , ------------------------------ 7» .................. 4 0.2
' CI IPSSIPDADMACI4

2... = 93249.: .........................
”2000 y 383.22x ,_ 0.15

...........................................................

l

I
P
o
01

 

 

 

. .130

 

   

7,770 — ----------- - -------- , -. -------------------

s. I.

E 60 _ ........... .. . ....... 3 .......................
O

0 .-~' ; E

,-

5,30 _....__.- ,
[£20 4-...-...y = 77.268x0'1f45-..-,,.._,________-__,,
10 - ----------- R2 = 0.9519. ........ , ..... ..... 7' 0.05

0 l ‘ I I I 0

o 1 2 3
Cm (mM)

 

 

 

41

  
  

 

 

 

  

 

 

 

 

  

 

 

600 '— “N'H’HHHH-H'" O [PSSIPAH'I4.5 ..___ 0.35
: _ 0.3448 .2-

$500 _ . ‘. -..y _ 240.47x 5 o 25
E 400 .l. ’ ...... R-08603 ..... 4 —) ................. .
q, . ‘ 2
o 300 - --------- .-----__--_-____ ------------------------------------- 4 0.15 g
\- <—- 0
5200 _ ............ .QG ................................................

'" _~" 0 -- 0 05

n' 100 .. -T' ................................................... . ............
O
0 u u u -0.05
4
Cm (mM)

700 p --------- . ------------------------------------------------------ 0.14
A600 -------------- !--~l-'-’-§f£?lﬂ?fl?éﬁ ------- 0 12
1‘2 500 — my - 3235-35" ------- i ------------------ -- 0.10
E R = 0.982 —> o

.9 400 - ---------------------------- ;,-.- ------------------------------ —- 0.08:
v- 300 1‘” -- 0.050
7200 ----'- ----- :,._,.:: -------------------------------------------- -L 0 04

100 - _~’-'-'--l ----------------------------------------------- -- 0.02

I" I
0 . . r ' 0.00
0 5 0 15 20
cm (mM)

350 l A—- IPSS’PAH]2+IPANPAHI1.5 [ 0.30
A300 _.- y = 2294x0268“L 0.25
W .. R2 = 0.7606 ..............

E250 /'—-) J- 0.20 —
O
o; 200 _ ............................................................. '5
O J "o T” 0.15 E
. 3:150 - .1 ~ 0100
v -0.
£100 _ Hf, ................................................. '
5o .. .737 .............................................. tum-"J” 0.05
0 i i i . 0.00
0 6 10
CI“ (mM)

Figure 2.7. Experimentally determined (symbols) and ﬁtted (lines) concentration

42

dependence (C m ) of the salt permeability coefficient (P9) and G ratio (see
equation 2.15) for each membrane.

2.4.3. Rejection and permeate ﬂux in colloidal fouling experiments: overall

comparative assessment of NF270 and PEM membranes
Figure 2.8 summarizes values of speciﬁc permeate ﬂux, MgSO4 rejection, and
concentration polarization factor for all membranes before and after they were fouled

by Si02 colloids. Several general observations can be made regarding the ﬂux and

rejection performance of NF270 and PEM membranes during the ﬁltrations of a
simple salt solution followed by a complicated aqueous mixture of salt and colloids:

1) Initially, NF270 membranes exhibited the highest rejection (Fig. 2.8 (b)) because

of the low MgSO4 permeability of NF270 (Fig. 2.6,) and the relatively low value of

the initial permeate ﬂux (ca. 100 L/mZ/h) set for all membranes (Fig. 2.6). As
discussed above, at higher ﬂuxes, rejection of PEM membranes greatly increases.

2) [PSS/PDADMAC]4 stood out as a membrane with the highest speciﬁc permeate
ﬂux (Fig. 2.8 (a)), which was due in part to the high hydraulic permeability of this
membrane (Table 2.3) and in part due to its high permeability to MgSO4 (Fig. 2.6)

and the resulting low osmotic pressure.
3) Although there were signiﬁcant differences between different membranes in the
salt rejection initially (Fig. 2.8 (b)), it is evident that these differences evened out after

20 h of colloidal fouling.

4) Remarkably, the observed salt rejection of all PEM membranes (except for

[PSS/PAH]4) fouled by colloids was higher than that of the same membrane before

colloidal fouling occurred (Fig. 2.8 (b)). This improvement in rejection was

43

accompanied by a decrease in concentration polarization for those membranes (see

Fig. 2.8 (0)). The most attractive PEM ﬁlm is the [PSS/PDADMAC]4 membrane.

With respect to NF270, the [PSS/PDADMAC]4 membrane’s combination of

comparable rejection and superior speciﬁc permeate ﬂux clearly indicates that a

membrane with highly beneﬁcial properties is formed as a result of deposition of

negatively charged colloids onto the surface of the [PSS/PDADMAC]4 ﬁlm (Fig.

 
   

     

 
 
  

 

 

 

   

2.8).
(a) A . . , .. ... ., .
g g: . . I'IimeOh
:: {’4 ,,: “*5 fl
S 7 ~ --
as g ‘
’2‘ 4 —
(3' NF270 [PSS/PDAD]4 [PSS/PAH]4 [PSS/PAH]4. 5 [PSS/PAH]2 [PSSlPAH]2
+[PAA] +[PAA/PAH]1.5
(b)
100 “ ”’.-Time 0 h
.425 Dv‘l'lme 20 h
80 .. , .. ...-.

60 - ~ 2;” "

    

Robs (0A3)

\‘
4 \\\ \. .
x \» \ -_ .-
. ‘ \ \ I“ ‘
. \\.\
\\\ '

40-I

..\ .‘.\ _._\. .. \ ‘
. . . \ .

‘ .n_ \\,"\ .\A\\>\.\;‘ _. \ ‘ “
. \ .

20-

 

 

 

 

 

 

 

NF270 [S/PDAD]4 [PSS/P] [PSS/P145 [PSS/PAH]2 [PSS/PAH]2 ' _'
+[PAA] +[PAA/PAH]1.5

, El'I'ime 20 h

  

 

 

 

o . r . 9 I . I . , /' 2.4.. .
NF270 [PSS/P0A014 [PSS/PAH14 [PSS/PAH14. 5 [PSS/PAH]2 [PSS/PAH]2
+[PAA] +[PAA/PAH]1. 5

Figure 2.8. Steady state values of the speciﬁc permeate ﬂux (a), observed MgSO4
rejection (b), and concentration polarization factor (c) for membranes before

fouling (ca. J=2.8 X 10‘5 m/s) and after fouling with Si02 colloids.

Note that the initial rejection values are given for the initial permeate ﬂux that
varied only very Slightly (27 m/s to 31 m/S) from one membrane to another (Table

2.2). The permeate ﬂux after fouling (i.e. t = 20 h) was approximately the same (ca.

40% of the initial permeate ﬂux) for all membranes expect for [PSS/PDADMAC]4

and [PSS/PAH]2+[PAA]. For the latter two membranes, the ﬂux after colloidal

fouling was ca. 20% of the initial value. This has to be taken into account when

comparing observed rejections of different membranes at 20 h.

2.4.4. Membrane performance in colloidal fouling experiments: Transient behavior of

permeate ﬂux, MgSO4 rejection, and cake resistance

The effect of membrane fouling on different PEM membranes was evaluated
together with the fouling effect on a commercial nanoﬁltration membrane under the
same feed water and hydraulic conditions.

Figure 2.9 illustrates the transient behavior of normalized permeate ﬂux, J ,

observed salt rejection, Robs , and colloidal cake resistance, Rd , in colloidal fouling

45

experiments. (The non-normalized values of the initial and ﬁnal permeate ﬂux are

given in Fig 2.8a.) Permeate ﬂux and rejection data were recorded experimentally,

while the Rd values were calculated aﬁer the contribution of the cake-enhanced

osmotic pressure to the overall ﬂux decline was accounted for.

  
 
 
 
     

 
 
     

 

 

a)
( 1.0 l ....................................................................
g ”5270
a: , :1 PSSIPDADMACM
a '. _______ A SSIP
130-8 . ; o PSS/P .5 s
g g I PSS/PA 2 AA] .
a Z A 93311:): 24 AAIPAI-l11.5
0.0.64 ° m- -- - 1-.
.3 3““.2 1 .
.5 CE: . ., ‘ .... i
.304 4%”.-.3‘ ---------- ‘27.. o“¢o.««'.'::’:# ‘1'" i W.
E Q5233- é ' 2
z QM s
0.2 l i i M
0 5 1o 15 20 25
Time (h)
(b)

    
     

............................................................................

+ M: 270
SSIPDADMACH
SSIPAH]45

-0- [PSSIPAH]4

 
 

[PssggﬁmZHPAAIPAI-m 5

..................................................

 

 

0 5 1o 15 20 25
Time (h)

46

(C) 16 _ + NF 270
4» [PSS/PDADMAC]4
+ [PSS/PAH“

‘4 ‘ +IPssrpAHI4.5 If """"""""
2 , +IPSSIPAHIz-I-IPAAI ______
1 + [PssrpAruHPAAIPAIqu g 5

 

R¢+Rm (x 1013 m")

 

 

 

Figure 2.9. Transient behavior of the normalized permeate ﬂux (a), observed
MgSO4 rejection (b) and colloidal cake resistance (c) in experiments on the
ﬁltration of SiOz colloids.

There are two considerations that need to be taken into account when analyzing the
data presented in Fig. 2.9.

i) Because a higher rejection results in higher osmotic pressures, the temporal
evolution of the permeate ﬂux should be interpreted together with the rejection data.
In turn, the Observed rejection is a function of the salt concentration at the membrane
surface, which depends on the amount of deposited colloids brought to the membrane.
At the same time, the rate of colloidal deposition depends on the permeate ﬂux. Thus,

the dynamics of all three variables - J ,Robs , and Rd - are interdependent and

Should be analyzed together.

ii) Only at the very early stages of the fouling experiments is the unfouled
membrane surface exposed to the permeate ﬂow. With the formation of a layer of
colloidal particles on the membrane, the properties of the surface (charge and
hydrophilicity) with which depositing colloids interact will be the properties of the

already deposited colloids, and not properties of the as-prepared membranes.

47

Generally speaking, the membranes fell into two categories: (i) membranes with

anticipated MgSO4 rejection behavior wherein the rejection decreased with the
growth of the colloidal cake (NF270, [PSS/PAH]4) and (ii) membranes with Mg
rejection that increased with ﬁltration time ([PSS/PDADMAC]4, [PSS/PAH]4.5,

[PSS/PAH]2+[PAA], and [PSS/PAH]2+[PAA/PAH]1_5).

Performance of NF270, IPSS/PAHl4_

For NF 270 and [PSS/PAH]4 membranes, MgSO4 rejection decreased with ﬁltration

time as expected (Fig. 2.9 (b)). This decrease in observed rejection was due to (i) the

decrease in the permeate ﬂux (Fig. 2.9 (a)) and (ii) cake—enhanced concentration

polarization (Fig. 2.10 (a)). The concentration polarization factor, %’1 , was

f
calculated as described in section 2.2. EA for these two membranes increased
f

signiﬁcantly at ﬁrst and then gradually declined to partially offset the initial increase
(Fig. 2.10 (a)); the corresponding trend in observed rejection (Fig. 2.9 (b)) is

consistent with the behavior reported earlier for LFC-l [132,150,151] and BW30

[132,150,151] reverse osmosis membranes fouled by Si02 colloids. In experiments

with Si02 colloids, the concentration polarization factor for NF270 was ca. 15 times

higher than that for the [PSS/PAH]4 membranes after 20 h of colloidal ﬁltration.

48

   
  
   

(a) 30" ‘ ' ... NF270

2“ ‘ ‘ ” ‘ ‘ +[PSSIPAH14
2004. . , ,, .

 

 

 

150 «
100 -
Q
E
U
25
Time (h)
(b) 1, , , A
9 43— [PSSI PDADMACJ4
8 ' -<->—[PSSIPAH]4.5
7 ,3 -I- [PSS/PAH]2+[PAA]
6 +[PSSIPAH]2+[PAAIPAH]1.5
5
\E 5
0 4
3
2
1 ’ ..........
o l I r I I I
0 5 10 15 20 25
Time (h)

Figure 2.10. Evolution of the concentration polarization factor during ﬁltration of
SiOz suspensions by different membranes.

In experiments with Si02 colloids, the observed salt rejection of NF270 was

signiﬁcantly higher than that of [PSS/PAH]4 during the initial stages of ﬁltration (Fig.

2.9 (b)), which explains the more than 15-fold higher polarization factor for NF270

49

during the early stages of the experiment (Fig. 2.10 (a), t < 5 h). During the ﬁltration
stage that followed, however, the rejection of all membranes was similar and yet the

polarization factor of NF270 was still up to 15 times higher than that of the

[PSS/PAH]4 membrane (Fig 2.10). Considering that the resistance of the cake formed

on the membrane surface is higher for [PSS/PAH]4 than that for NF270 (Fig. 2.9 (c)),

the large difference in the polarization factor can not be explained solely in terms of

colloid and MgSO4 transport. We hypothesize that the deposition of colloids alters the

structure of the PEM ﬁlm so that the extra resistance to the permeate ﬂux due to the
deposited colloids translates into a smaller increase in concentration polarization than
for the more crosslinked commercial membranes such as NF270. It is possible that
such changes in the membrane structure entail changes in the salt permeability

coefﬁcient PS =1§(C) dependence (Fig. 2.5), in which case the results on the

concentration polarization factor (Figures. 2.8 c and 2.10) and resistance of colloidal
deposit (Fig. 2.9 (c)) for PEM membranes would need to be interpreted with caution.

In evaluating ﬂux and observed rejection data (Fig, 2.9 (a), (b)), one can see that, at

steady state, the performance of the [PSS/PAH]4 membrane under conditions of
fouling by Si02 colloids is very similar to that of the NF270 membrane. Given the

possible regeneration of PEM membranes, the [PSS/PAH]4 ﬁlms might provide an

nanotechnology assisted alternate for controlling fouling by negatively charged

colloids.

Performance of [PSS/PAijﬂPAA], [PSS/PAH]7+[PAA/PAH]1_5, and

[PSS/PDADMAClA, [PSS/PAH145 membranes

50

The second category of membranes consists of the positively charged

[PSS/PDADMAC]4, neutral [PSS/PAH]2+[PAA], and negatively charged

[PSS/PAH]4.5 and [PSS/PAH]2+[PAA/PAH]1_5 membranes. The rejections of these

four PEM membranes unexpectedly increased with ﬁltration time. Furthermore, for

these membranes the calculated concentration polarization factor decreased with the

ﬁltration time. For [PSS/PDADMAC]4 (and, at the very end of ﬁltration, for

[PSS/PAH]2+[PAA/PAH]1.5) the concentration polarization factor, computed

assuming a constant PHCm) dependence, was found to decline over the time of

ﬁltration to below 1, which is in the domain of unphysical values (Fig. 2.10 (b)). We

attribute this phenomenon to the possible decrease of the salt permeability coefﬁcient

PS with the deposition of signiﬁcant amount of SiOz colloids onto the PEM surface or
even inside the PEM ﬁlm. The Ps(Cm) correlation was examined with
[PSS/PDADMAC]4 membrane that is covered by only a submonolayer of Si02
colloids. In this case, there was no evidence showing signiﬁcant changes in Ps(Cm)

on the same type of PEM ﬁlm after it was covered with Si02 deposition purely by

adsorption.

Generally, for a net depositional system such as a membrane ﬁlter, prior to the
attainment of steady state ﬂux, the trend of increasing rejection with time can not be
explained without invoking changes in the membrane properties. A modiﬁcation of
the salt-rejecting properties of these membranes due to changes in the PEM structure
upon colloidal deposition is a likely explanation of such a trend. Thus, the above

results bring into question the model’s basic assumption that the membrane transport

51

coefﬁcients remain unchanged with time.
We hypothesize that a PEM-colloid nanocomposite is formed on the UP support
surface during colloidal ﬁltration. This hypothesis is based on the anomalous behavior

of rejection and concentration polarization factors as well as on the observations that

PEM ﬁlms can swell. PDADMAC-capped [PSS/PDADMAC]4 ﬁlms in water have a

swollen thickness of ca. 100 nm [153], which is comparable to the diameter of SiOz

particles and could allow the particles to be embedded inside the PEMS. In this regard,
it is interesting to note that during colloidal ﬁltration [PSS/PDADMAC]4 exhibited
the most rapid ﬂux decline among the six membranes, but the increase in cake

resistance was not as rapid as it was for the other membranes, resulting in the lowest

steady state Rd value among all membranes. One possible reason for the improved

rejection by such PEM-colloid nanocomposite ﬁlms is that the colloids create a

charged layer that contributes to rejection.

2.4.5. Regeneration of [PSS/PAH]4 and [PSS/PAH]4.5 ﬁlms

PAH becomes weakly charged at high pH and thus dissociation of multilayers
composed of [PSS/PAH] (and, thereby, removal of the PEM from the UP support
[153,154]) can be achieved by increasing the pH of the external solution
[126,131,155]. PEMS can potentially then be re-deposited (i.e. regenerated at the

support surface) from low pH solutions. The removal of 1) as-deposited and 2)

colloid-fouled [PSS/PAH]4 or [PSS/PAH]4.5 PEM ﬁlms from the UP support was

monitored by recording changes in hydraulic resistance, Rm =(u-Lp )_1 , and

observed salt rejection, Robs , of the membrane at different steps of the regeneration

52

procedure (Fig. 2.11).

(a) (b)

 

 

 

 

 

 

 

05 _ ........................ . ...............................
I
o.4~
\:\\\\\\
m 03 4-
.D '
“.\‘\
o 7;.‘1:\
tr
\\
0.2 ‘i . t
t
\ \‘
«t
‘\5§~‘Ef‘>
§_.f‘.:-.
01‘.._.._... _\‘\-‘~:“ ....._........ ..........
. k \
\ \‘
\‘ \\\
0.0

   

Figure 2.11. Hydraulic resistance (Rm) and observed salt rejection (Robs) values of
a UF 50 kDa support before (1) and after the following sequential steps: (2)

modiﬁcation by a PEM; (3) fouling by Si02 colloids for 20 h; (4) backﬂushing
with water for l h; (5) soaking in pH 10 buffer for 10 min; (6) redeposition of a
PEM layer. (In cases where no data are visible, corresponding steps were

omitted.) PEM membranes employed included (a) as prepared [PSS/PAH]4.5; (b)

a [PSS/PAH]4.5 ﬁlm with SiOz ﬁltration and backﬂushing prior to soaking in pH
10 buffer.

After soaking as-prepared PEM membranes (Fig. 2.11 (a); [PSS/PAH]4_5

membrane) in the pH 10 buffer solution (step 5 in Fig. 2.11 (a)), the values of Rm
and Robs both decreased to the level typical of the UP membrane (step 1 in Fig. 2.11

(a)), suggesting removal of the polyelectrolyte ﬁlm. (Note that the relatively low

53

values of observed rejection by the PEM membranes are due to concentration

polarization.) Reapplication of the PEM (step 6) returned Rm and Robs to nearly the
levels characteristic of an as-prepared PEM-coated membrane. The same procedure

was followed with [PSS/PAH]4_5 and [PSS/PAH]4 membranes fouled by SiOz, it

seems that the fouling did inhibit the regeneration of PEMS by the low values of Rm
Rm and Robs after step 6 (data not shown). Therefore, the simple soaking of the

fouled membrane in the buffer solution apparently is not sufficient to completely

remove the PEMS and the foulants. SEM images taken for this membrane after step 6

(Fig. 2.12) conﬁrm the presence of residual SiOz colloids on the membrane surface

and offer further evidence that regeneration by soaking alone was not succesful for

[PSS/PAH]4.5 and [PSS/PAH]4 fouled by sioz.

54

(a) (b)

 

 

 

 

 

 

 

 

 

Figure 2.12. SEM images of membranes at the different steps described in section
2.3.5: (a) UF 50 kDa support (step 1); (b) after the UP support was modiﬁed by

[PSS/PAH]4. 5 (step 2); (c) after the membrane was fouled by SiOz for 20 h (step

3); (d) after the SiOz-fouled [PSS/PAH]4.5 membrane was soaked in pH 10 buffer
and regenerated with a new PEM.

The backﬂushing step greatly improved the efﬁciency of PEM regeneration (Fig.

2.11 (b), step 6). The membranes were backﬂushed with ultrapure water (step 4 in Fig.

2.11 (b); [PSS/PAH]4_5 membrane) to remove [PSS/PAH]4.5 ﬁlms from membranes

severely fouled by Si02 colloids (300 mg (Si02)/L, 20 h ﬁltration). After

backﬂushing-assisted PEM removal and PEM regeneration, the value of Rm

returned to the level characteristic of an as-prepared PEM membrane. The value of

Robs increased as well although only to about 50% of the rejecting capability of an

55

as-prepared PEM membrane. While the composition of the soaking solution and
backﬂushing duration need to be optimized to improve the efﬁciency of regeneration,
the demonstrated feasibility of backﬂushing points to the possibility of using PEM
ﬁlms as regenerable nanoﬁltration coatings with controllable charge, hydrophilicity,

and permeability.

2.5. Conclusions

By choosing constituent polyelectrolytes and by adjusting the conditions of their
deposition, supported PEM membranes with controllable surface charge,
hydrophilicity and permeability to water and salt were designed and characterized in
terms of their ion transport properties and resistance to colloidal fouling. It was found
that:

1. Highly hydrophilic and charged PEMS can be designed.

2. The designed nanoscale PEM membranes were highly selective and could

achieve nearly complete intrinsic rejection of MgSO4 at sufﬁciently high

ﬂuxes, which is also indicated by the fact that reﬂection coefﬁcient of PEM
membranes were estimated to be close to 1.

3. Salt permeability coefﬁcients of NF270 and all PEM membranes exhibited
power law dependence on concentration: PS = C ”ft , 0.19 < 2' < 0.83.
4. Under the highly fouling conditions employed in this study, certain PEMS

([PSS/PAH]4) had steady-state performance similar to that of the commercial

NF270 membranes, especially in the longer term (> 5 h)
5. The separation properties of certain PEMS improved dramatically with the

deposition of colloids onto their surface. For these membranes, the

56

concentration polarization decreased and MgSO4 rejection increased with an

increase in the amount of deposited colloids. We hypothesize that a
PEM-colloid nanocomposite was formed on the UP support surface as a result
of colloidal fouling of the PEM ﬁlm.

6. The feasibility of regenerating the PEM coating with “snake skin effect” has
been demonstrated. Although regeneration of highly fouled membranes by
soaking alone was inhibited by the incomplete removal of deposited colloids,
an additional backﬂushing step resulted in an almost complete removal of the
fouled PEMS and enabled reassembly of a PEM ﬁlm with permeabilities
similar to those of the initial PEM membrane but with 50% lower rejection
capacity.

In summary, ultrathin PEM membranes prepared using the layer-by-layer
deposition method showed high salt selectivity, high water ﬂux and could be
regenerated under appropriate conditions. Such membranes can potentially be
designed to use in many different membrane applications where the characteristics of
high ion selectivity, anti-adhesive surface properties and regenerability of the

separation layer are desirable.

57

Chapter 3. Interactions of aqueous NOM with

nanoscale T iOzz Implications for ceramic

membrane ﬁltration-ozonation hybrid process

This work was done in collaboration with Dr. Jeonghuan Kim et al., and has been

published in the Environmental Science and Technology journal [156].

3.1. Introduction
TiOz is one of the most abundant, man-made, and commercially available

nanoparticles. It has been applied to pharmaceutical products, personal care products

and auto parts. Nanoscale Ti02 particles have also been used to fabricate

nanocomposite membranes [43,157-159] and used as catalysts in advanced oxidation

processes [160-166]. In previous work, Kim et al. used Ti02 ceramic membranes that

are resistant to oxidation by ozone to act as a catalyst for the decomposition of ozone
into highly reactive OH-radicals that can react with deposited NOM and thus mitigate
fouling [166].

The reason why it is important to control NOM fouling is that NOM has been
identiﬁed as a major foulant especially in nanoﬁltration of drinking water treatment
[167-171]. NOM is of great concern in water treatment systems because it is
ubiquitous in the environment and is known to react with chlorine, the most

commonly used disinfectant in the US, to form carcinogenic chlorinated compounds

58

[172]. It has been suggested that ultraﬁltration and nanoﬁltration can be effective in
removing NOM; however, in most cases, extensive NOM fouling limits membrane
performance signiﬁcantly. To understand how to prevent or mitigate NOM fouling of
membranes, a better understanding of the NOM fouling mechanisms is needed;
therefore, studies of NOM-membrane interaction are of signiﬁcant practical relevance
to membrane ﬁltration. A number of previous studies describe NOM fouling of
polymeric membranes in terms of the surface charge and molecular conformation of
the NOM which is controlled largely by solution properties, such as pH and ionic

strength [173,174]. The strong dependence on the water environment was also

observed in the TiOz-NOM adsorption and TiOz ceramic membrane ﬁltration in this

study [156].
On the other hand, the investigation into such interactions is very important from

the perspective of environmental implications of nanotechnology; this is so because 1)

TiOz nanoparticles ﬁnd increasingly wide usage in various technologies and therefore,

are very likely to enter natural water systems; 2) NOM is a common component in
natural water; 3) NOM is known to easily adsorb at solid surfaces in water and
consequently this adsorption modiﬁes the surface with a different charge and
hydrophilicity [175-178]; and 4) one type of NOM — humic acid was reported to be

effective in stabilizing aqueous nanoparticle suspensions [179,180]. Therefore, by

understanding the interaction between NOM molecules and Ti02 nanoparticles, we

will be able to understand the resultant structure, surface properties, stability, and

fate/transport of Ti02 in natural water.
To summarize, the speciﬁc objectives of this work were: 1) experimentally

characterize Ti02 nanoparticles and (ozonated and non-ozonated) NOM with respect

59

to hydrophobicity and size distribution; 2) analyze NOM-Ti02 interactions under

different water chemistries using XDLVO modeling. With the modeling study of the

interaction, we would understand the mechanism in improved ﬁltration performance

with hybrid ozonation TiOg catalytic ceramic membrane ﬁltration. It should also give

us clues to understanding the behavior of TiOz nanoparticles when they enter natural

water and interact with NOM under different chemical conditions.

3.2. Approach

The extended DLVO (XDLVO) theory predicts the energy of particle-particle and
particle-surface interaction [106,181,182]. The XDLVO model has been applied to
describe how aqueous colloids interact with surfaces of polymeric membranes
[183-185]. When a particle is suspended at a close afﬁnity to another particle or
interface, besides gravity and hydrodynamic forces acting on that particle, particles

also experience non-covalent forces that include van der Waals forces (FLW ),

electrostatic double layer forces (F EL ), and Lewis acid-base forces (F A B) exerted by

the other particle or interface. XDLVO is an extended DLVO model wherein
acid-base (polar) interactions are also taken into consideration. Incorporation of the
polar interactions has been shown to result in a signiﬁcant change in the predicted
energies of short range (< 5 nm) particle-particle and particle-surface interactions.

In this work we determine the interaction energy (free energy of adhesion) per

unit area between two planar surfaces with surface tension parameters and charges of

NOM and TiOz. Although the model does not take into the account the size

distribution of NOM molecules and the likely differences in hydrophilicity and charge

60

between various size fractions of NOM, this approach should adequately describe

effects, averaged NOM sizes, of ozonation-, pH-, and calcium induced changes in

NOM chemistry on the NOM-TiOz interactions.

The free energy of adhesion between two materials (NOM, N and TiOz, T)

immersed in a liquid (water, W) due to van der Waals forces is calculated as follows

[186]:

 

AGL—W2[\/)’,1&W—\/7LW)(W_W) (3.1)

where yLW, 3%”, , and 7%,”, are LW components of the surface tension for water,

the Ti02 nanoparticles, and NOM molecules respectively. Here, AGfOW is the LW

free energy of adhesion between a TiOz nanoparticle and an NOM molecule that are

at a distance of yo , which is the minimum equilibrium cut-off distance and is usually
assigned a value of 0.157 nm [186].

The free energy of adhesion between a TiOz particle and NOM molecules due to

EL (electrostatic) interaction can be estimated using the following expression [187]:

EL
AGyO =503'CNCT ”EL (32)

—l
where AEL = (3.28 x109,/Ce, ) , 5,50 is the dielectric permittivity of the bulk ﬂuid;
4 N and C T are surface potentials of NOM molecules and TiOz nanoparticles,

respectively; C e, is the concentration of the background electrolyte in mol/L; AEL is

61

the characteristic decay length for EL interactions in water.
The free energy of adhesion per unit area between two different materials immersed

in the same liquid due to acid-base (AB) interaction is given by:

803:3 =2 7:; [JEME—Jﬂw r; “gram—M)
-2(\/rirXL Nye/t]

where y+ is the electron acceptor component and y’ is the electron donor

(3.3)

component of the free energy. The surface tension parameters of NOM molecules

7X1, 77V, yNAB, yNLW, 7N TOT) and TiOz nanoparticles (7T, 7T, 77/18,

yTLW , 77T0T ) can be determined from the extended Young equation after

measuring contact angle data for three probe liquids with known surface tension

parameters (ylLW , 71+ , 71-)

(1+6089)71T0T =2(\/r.”’r 1”” +er‘ 4752/?) (3.4)
7’13 =2\/r+r’ (3.5)

yTOT :7AB +7LW (3.6)

 

where 6 is the contact angle for a given probe liquid. The subscripts s and 1

correspond to the solid surface and the liquid, respectively.

3.3. Material and methods
3.3.1. Materials

TiOz particles (Ti02 nanopowder, <100 nm (BET), mixture of rutile and anatase,

99.9% metals basis, Aldrich) were used as received. Suwannee River NOM

62

(SRNOM), isolated by reverse osmosis, was purchased from International Humic
Substances Society (Denver, CO). A 20 mg/L colloidal suspension of SRNOM was
prepared by dispersing the SRNOM in deionized (DI) water. The total organic carbon
(TOC) content of the SRNOM feed solution was 10 i 0.5 mg C/L. The pH of the
solution was adjusted by adding either HCl or NaOH and a borate (0.0625 mmol/L)

buffer solution. The borate buffer was prepared by mixing aqueous solutions of

Na2B4O7-10H20 (100 mL, 0.025 M) and HCl (41 mL of 0.1 M). Prior to each

ﬁltration experiment, the total ionic strength of the solution was adjusted to 7.510-3
. -2

M usrng 10 M NaCl.

3.3.2. Characterization of NOM molecules and Ti02 nanoparticles

Contact angle measurements were conducted to determine the hydrophobicity of

nonozonated SRNOM molecules, ozonated SRNOM molecules and Ti02

nanoparticles, The TiOz, non-ozonated SRN OM and ozonated NOM suspensions

were ﬁltered through 1 kDa polyethersulfone UF membranes (Pall Corp., East Hills,
NY) using a stainless steel ﬁltration cell (HP4750, Sterlitech Corp., Kent, WA)
without stirring. Prior to the ﬁltration, membranes were soaked in DI water for at least
10 h at room temperature and 200 ml of DI water was ﬁltered through each membrane

to remove trace chemicals, as recommended by manufacturer. The ionic strength, pH,

and Ca2+ concentration of all the suspensions (Table 3.1) were adjusted prior to

ﬁltration. The SEM micrographs (Fig 3.1) show that the material deposited on the

membrane surface formed a cake layer is approximately 0.2 um thick. For the

ozonated SRN OM solutions with and without Ca2+, the ﬁrst 600 mL were ﬁltered

twice to improve NOM recovery. The membranes with deposits were dried in a

63

desiccator before the contact angles were measured. The contact angles for three
materials (TiOz, ozonated SRNOM, and non-ozonated SRNOM) were determined at
three different pH values (3, 5, and 8) and two concentrations of Ca2+ (0 mM, 1 mM).

Thus, contact angles were measured for a total of 18 different material/solution

chemistry combinations.

(8) (b)

 

 

 

 

 

 

Figure 3.1. SEM micrographs of the cross-section of (a) the bare membrane and
membranes with the layer of ﬁltered (b) NOM, (c) ozonated NOM, and (d) ozonated

. + .
NOM 1n the presence of Ca . The fouled membranes are representative of samples
used in contact angle measurements. Arrows point to the NOM cake.

Table 3.1. Contact angles measured and surface energy parameters calculated for
SRNOM, ozonated SRNOM, and TiOz at different pH and concentrations of Ca2+.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 0
Surface energy parameters (mJ/m ) Contact angle ( )
TOT
LW + - AB
r r Y 7 7
a
Probe liquid
Ultrapure water 21.8 25.5 25.5 51.0 72.8
34.0 3.9 57.4 30.0 64.0
Diiodomethane 50.8 0.0 0.0 0.0 50.8
C
l a Diiod
mate 2+
1, H20 Glycerol omet
rial
mM hane
50.8 :1: 1.05 i 46.3 d: 13.9 64.7
T102 0 0 16 :1: 1.8 0
0 0.12 0.77 :t 0.8 :t 0.8
(250 50.8 :1: 1.05 :1: 46.3 :1: 13.9 64.7
m] 0 0 16 i 2.7 0
’ 0 0.18 1.17 :t 1.2 :t 1.2
0‘5 50.8 :1: 1.00 i 46.6 i 13.7 64.5
0 0 17 :t 2.2 0
3”“) 0 0.15 1.01 e 1.0 3:10
50.8 :1: 0.96 :1: 46.9 :1: 13.4 64.2
T102 1 0 18 :1: 1.1 o
0 0.07 0.52 :1: 0.5 :1: 0.5
(25° 50.8 :e 1.08 :1: 46.1 :t 14.1 64.9
m] l 0 16 :t 3.4 0
’ 0 0.22 1.41 i 1.4 :1: 1.4
0'5 50.8 3: 1.18 i 45.4 d: 14.6 65.4
1 0 14 :l: 2.4 0
gm) 0 1.14 0.86 i 0.9 e 0.9
NOM 48.6 i 0.02 :t 40.0 :t 1.7 i 50.3 38 :1: 17 i
0 46 :1: 3.0
(300 1.4 0.07 4.51 3.3 :1: 3.6 2.5 3.9
1111. 48.5 :t 1.0E-4 40.9 i 0.1 i 48.6 39 :t 18 :t
0 48 i 1.3
20 0.9 $0.003 4.98 1.9 :1: 2.2 3.5 2.4

 

 

 

 

 

 

 

 

 

 

 

65

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mg/L Table 3.1 (cont’d)
) 47.3 3 0.004 34.1 3 0.7 3 48.1 46 3 21 3
51 3 2.2
1.3 3 0.03 4.00 2.6 3 2.9 2.3 2.8
48.6 3 0.02 3 41.8 3 2.0 3 50.6 36 3 17 3
NOM 45 3 1.7
0.7 0.06 4.27 2.4 3 2.5 2.9 1.8
(300
48.3 3 0.12 3 40.0 3 4.4 3 52.8 35 3 18 3
ml, 42 3 1.8
0.6 0.10 3.01 1.9 3 2.0 2.0 1.6
20
3315-5
mg/L 48.6 3 39.0 3 0.1 3 48.7 40 3 17 3
3 49 3 1.6
) 0.6 4.18 2.0 3 2.0 2.8 1.5
0.002
0 47.5 3 1.3E-4 54.8 3 0.2 3 47.7 24 3 21 3
3 44 3 2.7
1.2 3 0.01 4.89 3.4 3 3.6 3.6 2.8
NOM
1.6E-5
(1600 46.9 3 51.8 3 0.1 3 46.9 29 3 23 3
3 46 3 3.0
ml, 1.4 5.72 3.8 3 4.1 3.8 3.0
0.002
20
mg/L 48.3 3 0.01 3 35.2 3 1.4 3 49.7 43 3 18 3
49 3 3.5
) 1.2 0.07 4.20 3.7 3 3.9 1.6 3.0
03 47.5 3 0.07 3 47.1 3 3.7 3 51.2 29 3 21 3
42 3 3.4
1.5 0.15 5.7 3.9 3 4.2 3.9 3.5
NOM
46.9 3 115-4 3 42.3 3 0.1 3 47.0 39 3 23 3
(1000 49 3 2.9
1.5 0.01 5.81 3.7 3 3.9 3.5 3.1
ml,
20 48.6 3 0.53 3 39.6 3 9.2 3 57.8 30 3 17 3
34 3 2.3
mm 0.9 0.25 4.3 2.2 3 2.4 3.6 2.5
)

 

 

A FT A 200 analyzer (First Ten Angstroms) was used to measure contact angles.
Measurements were carried out with three different probe liquids — DI water, glycerol

and diiodomethane. The ionic strength, pH and calcium concentration of the DI water

66

were adjusted to match the water chemistry in the droplet to that of the suspensions

from which the Ti02 and SRNOM deposits on membrane ﬁlters were formed.

Measurements were made by forming, at the rate of 0.5 uL/s, a 5 [LL drop of liquid on
the tip of a stainless steel syringe needle and then placing the drop onto the membrane
surface by raising the membrane until contact was made. An image of the drop was
taken and the left and the right contact angles were measured. For each probe
liquid-sample combination at least two membrane coupons were analyzed with ﬁve
images recorded for each membrane coupon.

To evaluate the size distribution of NOM molecules in aqueous suspensions with

different ionic strength, pH and Ca2+ concentrations (Table 3.1), we fractionated

SRNOM using a series of ultraﬁltration membranes with different molecular weight
cut off (MWCO). Basically, the procedure of size fractioning of NOM molecules is

described in Mellema’s Master thesis [188]. In our study, a 40 mg/L colloidal aqueous

suspension of SRNOM was prepared. The pH, ionic strength and Ca2+ concentration

of the suspension were adjusted in the same way as described in Table 3.1. Five types
of ultraﬁltration membranes with different MWCO were used in sequence from large
MWCO to small MWCO. YM30, YMlO, YM3, W1 and YCOS have a MWCO of 30,
10, 3, 1, and 0.5 kDa, respectively. Filtration was conducted with Amicon 8200
ultraﬁltration stirred cells pressurized with nitrogen gas at 60 psi. To clean the
cellulose acetate membranes (YC05, YMI, YM3, YMlO, YM30, Amicon) prior to
ﬁltration, YC membranes were soaked in l M NaCl for 30 min while YM membranes
were soaked in 0.1 M NaOH for 30 min. In order to remove the TOC interference
from these membranes, the membranes were pre—ﬁltered with ca. 200 ml deionized

water. The ﬁltration started with 4 L of SRNOM solution (40 mg/L) through a YM30

67

membrane. During each ﬁltration, 500 mL of permeate was collected, and the rest of
the permeate solution was ﬁltered with the membrane of smaller MW CO. The
ﬁltration took place under stirring conditions to prevent concentration polarization as
much as possible. Then we measured the TOC content in the permeate samples that
were collected from each ﬁltration using a TOC analyzer (OI Analytical Model 1010

Analyzer, College Station, TX).

3.4. Results and Discussion
3.4.1. Fractionation of Suwannee River NOM (SRNOM)

We have plotted the fractioning results in two different ways. Fig 3.2 shows the
original data of TOC that were collected and measured from different permeate
samples. The bars in each plot stand for the concentration of TOC which belongs to
the SRNOM molecules that are smaller than the MW of the corresponding membrane.
Fig 3.3 is another way to present the TOC results. Each bar stands for the TOC
concentration of the SRNOM molecules in the MW range that are indicated in the
X—axis. The fractioning results (Fig. 3.3) tell us that ozonation did break the large

NOM molecules (10-30 kDa) into smaller sized NOM and made the suspension more

polydispersed. By adding only Ca2+, not much difference was observed in the size

charts (Fig. 3.3 (a) and (c)). For ozonated NOM, the presence of Ca2+ helped make a

suspension with even a smaller size distribution.

68

 

 

   

 

 

 

3.100 ’3
E: 80.1 2\’
C C
.9 60- '29;
0 (D
g 40 ~=
g 20 8
‘—
" 0
<05 <1 <3 <10 <30

MW (kDa)

(c) (d)
@100 @100
2, 80 a, 80
c 8
g 60 '8 60
g 40 g 40 . 2
0 20 8 20 :
O 1— § 7’7“
1- o 0

.5

<1 <3 <10 <30
MW(kDa)

 

Figure 3.2. TOC fraction of (a) NOM, (b) O3-NOM, (c) NOM(Ca2+) and ((1)

2+ . . .
O3-NOM(Ca ) that has a smaller molecular weight than the number mdrcated on
the x-axis.
The error bars represent 90% conﬁdence interval.

69

(a) (b)

 

 

 

 

 

00.5 0.51 1-3 3-10 1030 >30 0-0-5 0-51 1-3 310 10-30 >30
MW (kDa) MW (kDa)
(c) (d)

50 _ .............................................................
pH 3 ....................................

//// //)
////// /

9::

\\ \\\

\\\\‘R\\‘
/

/

  
  

\\\\.‘

  

 

 

 

 

/
%
%
%

  

\\\\\\\\\\\\\\\\ 1

x\\\\\\\\\\\\\\\: ,

////~//////////.:o
///////////’/////x

.54 ,
/\ 145‘ / bx

10-30 >30

   

Figure 3.3. TOC fraction of (a) NOM, (b) O3-NOM, (c) NOM(Ca2+) and (d) O3-NOM(Ca2+)
that has molecules in the size range indicated on the x-axis. The TOC content in each MW
class was computed as the difference between the TOC contents of the ﬁltrates from the two
membranes used in the isolation of the corresponding size fraction.

70

 

 

 

 

 

 

 

 

 

 

 

 

Cl NOM
03-N M 4;] 3
E3 03-N OM (Ca++)
45 ‘ j ‘-
c 1
2
U) ;
e 35 ‘1 ‘1
‘H z
0 E
B 1
C
o 25 7 -~r-~ i
15 ’
B

 

 

 

 

Figure 3.4. Combined effect of pH, Ca2+ and pre-ozonation on hydrophilicity of
SRNOM.

The error bars correspond to 90% conﬁdence intervals.

3.4.2. Hydrophilicity of SRNOM
Figure 3.4 illustrates the dependence of the water contact angle of the ozonated and

non-ozonated SRNOM on pH and calcium content. The measured contact angles for

SRNOM were in the range of 350 to 460 indicating that SRNOM is less hydrophobic

than Aldrich NOM ( 0 =740, [177]). Both the addition of calcium and ozonation

resulted in a decrease in the contact angle of SRNOM in water; the effect of calcium
was more pronounced at higher pH values, while the effect of ozonation was greater
at lower pH values. The same effect of ozonation on the hydrophilicity of NOM was
reported by Reckhow and co-workers [189]. The result is consistent with the fact that
the ozonation of the SRNOM produces lower molecular weight, polar, oxygen-rich
compounds with a higher content of hydroxyl, carbonylic, and carboxylic groups

[190]. It is not clear, however, why the addition of calcium rendered SRNOM more

71

 

hydrophilic and why the same trend was not observed when calcium was added to the
ozonated SRNOM solution.

The reaction of ozone with the NOM is pH dependent and the predominant reaction

mechanisms of ozonation depend upon pH (i.e., reactions involving molecular O3

predominate at lower pH and those involving OH-radicals and other secondary
oxidants predominate at higher pH) [191]. The chemical nature of the ozonated NOM
will depend upon reaction pH (see Fig. 3.4). As a result, the interpretation of the
contact angle data for ozonated SRNOM is difﬁcult. For example, SRNOM oxidation
by molecular ozone should generate more hydrophilic NOM species than those
produced during the oxidation of NOM by OH. This difference, however, is countered

by the protonation of hydrophilic functional groups at lower pH.

3.4.3. Free energy of SRNOM-SRNOM cohesion and SRNOM-Ti02 adhesion

The hydrophilicity of the SRNOM determines the magnitude of the forces of

hydrophilic repulsion or hydrophobic attraction (i.e., acid-base forces) that, together

with electrostatic and van der Waals forces, deﬁne SRNOM interactions with other

SRNOM colloids and with the Ti02 nanoparticles. To quantitatively evaluate

acid-base interactions, contact angle measurements with two more probe liquids with
known surface tension parameters were conducted. Based on measured contact angles
and estimated surface tension parameters (Table 3.2), free energies of adhesion (Eqs.
3.1, 3.2, 3.3) corresponding to van der Waals, electrostatic, and acid-base forces were

calculated for the cases of adhesion of l) ozonated and 2) non-ozonated SRNOM to

the surface of a Ti02 nanoparticle at different pH values (Table 3.2, Fig. 3.5 and Fig.

3.6). The negative free energy represents a thermodynamically unstable state

72

 

(attractive to each other), while the positive free energy means a thermodynamically

stable state (repulsive to each other).

73

Table 3.2. Free energy of adhesion of SRNOM on the surface of T102 nanoparticles

and SRNOM—SRNOM cohesion

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- 2
Solution Surface energy, mJ/m
Sample chemistry
7+
P" [Ca 1, AGLW AGAB AGEL A TOT
yo )0 YO 3V0
mM
NOM- TiOz interaction (adhesion)
SRNOM 3 0 -ll.3 3 0.5 27.5 3 3.7 -0.16 3 0.01 16.0 3 3.7
SRNOM 5 0 -11.3 3 0.3 28.5 3 3.7 -0.01 :L- 0.01 17.2 3 3.7
SRNOM 8 0 -10.9 3 0.5 24.1 3 3.5 0.45 3 0.03 13.7 3 3.5
SRNOM 3 1 -11.3 3 0.2 29.13 3.2 -0.12 3 0.01 17.7 3 3.2
SRNOM 5 1 -11.2 3 0.2 26.6 3 2.6 -0.02 at 0.003 15.3 3 2.6
SRNOM 8 1 -ll.3 3 0.2 26.5 3 3.2 0.04 3 0.004 15.3 3 3.2
03-SRNOM 3 0 -10.9 3 0.4 36.6 3 3.4 -0.21 3 0.01 25.4 3 3.4
03-SRN0M 5 0 -10.7 3 0.5 35.0 3 4.1 -0.01 i 0.01 24.3 3 4.1
03-SRNOM 8 0 -ll.2 3 0.4 24.7 3 4.0 0.46 3 0.01 13.9 3 4.0
03-SRN0M 3 1 -10.9 d: 0.6 31.9 3 4.2 -0.10 3 0.01 20.9 3 4.2
03-SRN0M 5 1 -10.7 3 0.5 29.2 3 4.6 00230002 18.5 3 4.6
03-SRN0M 8 1 -ll.3 3 0.3 24.4 3 3.2 0.03 :1: 0.01 13.1 3 3.2
NOM-NOM interaction (cohesion)

SRNOM 3 0 -10.6 3 0.9 25.13 8.1 0.10 3 0.02 14.7 3 8.2
SRNOM 5 0 -10.5 3 0.6 27.1 3 8.5 0.20 3 0.02 16.8 3 8.5
SRNOM 8 0 -9.8 3 0.8 15.7 3 7.4 0.42 3 0.06 6.4 3 7.4
SRNOM 3 1 -10.6 :1: 0.4 27.7 3 7.3 0.05 3 0.01 17.2 3 7.3
SRNOM 5 1 -10.4 3 0.4 23.9 3 5.1 0.04 3 0.01 13.5 3 5.1

 

 

 

 

 

 

 

74

 

 

Table 3.2 (cont’d)

 

 

 

 

 

 

 

 

SRNOM -10.6 3 0.4 24.1 3 7.4 0.03 3 0.004 13.6 3 7.4
O3-SRNOM -9.9 3 0.8 47.4 3 8.5 0.17 3 0.01 37.7 3 8.5
O3-SRNOM -9.5 3 0.9 43.4 3 9.9 0.28 3 0.01 34.2 3 9.9
O3-SRNOM -10.4 3 0.8 17.4 3 8.0 0.44 3 0.01 7.4 3 8.0
O3-SRNOM -9.9 3 1.0 34.7 3 9.6 0.03 3 0.01 24.8 3 9.7
O3-SRNOM -9.5 3 0.9 29.3 3 10.3 0.02 3 0.004 19.83104
O3-SRNOM -10.6 3 0.6 21.4 3 6.5 0.01 3 0.004 10.8 3 6.5

 

 

 

 

 

 

 

75

 

(b)

(a)

 

 

 
  

 

 

BO3-NOM

IO3-NOM (Ca++)

 

 

'II.III_IIIIT

r'.’"""' m H

u\\\\\\\\\\\\\\\\.\\\\\\\
rIIIIIIIIIIII

"_"""""'

   

     
    

 

 

 

 

 

 

 

 

   
   

. “““““““
w «nowhowowomowowowowo.

..,,,///.//////////////.

 

 

 

 

0 0 0 0 0 0 0 0
5 4 3 2 4| 4.| 7.3
NE\_..E .9 0:0 mowtaw

(d)

(C)

 

 

 

    
 

 
 

 

 

 

 
   

    

 

 

 

 

 

 

 

 

    

    
 

 

 

 

w; .
+ .3 _
% :\\\\\\\\\\\\ m
( "3333 T a
M M .
m m M
33 .... M a.
O O .
...vmmwwwvvwvww
0 0 0 0 0 0 0 0
5 4 3 2 1 1 al..
NEE: .>m ocm oomtsm
”I ..
+
a
m .. v. .... ..
w w “L
N N E N
z m...
. unnumﬁmmm. B _.
x///////////////// /A h
H m w
m _ L
0 0 0 0 0 0 0 0
~83: >920 mowtsw

(f)

(e)

 

 

 

    
 

 

 

 

 

    

 

 

 

 

 

 

 

     

  

 

 

1+}
1a ..\,\\\\\\\\.O
m
M M
O 0
N. N.
3 3
O 0 _. .
a s
H wvvvwwvww
0 0 0 0 0 0 0 0
5 4 3 2 1 1 0.
NEEE .>9 oco 885w
1+). _ ”....x ........ M
+ .
a .
WNW .H
M M _
O O
N N
a E
. .....“mmuuummuuu
. ,././/////////////
. w .
. L
0 0 0 0 0 0 0 0
5 4 3 2 4| 1 2
NEEE >9 oco oomtzm

Figure 3.5. LW, AB, EL and Total energy of SRNOM-Ti02 interactions for
non-ozonated SRNOM (a, c, e) and ozonated SRNOM (b, d, f) at pH 3 (a, b) pH 5 (c, d),

and pH 8 (e, f)

76

 

 

 

 

 

 

 

 

        

 
   

 

 

 

 

 

 

(b)

(a)

 

 

   
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

77

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

and pH 8 (e, f)

 

 

 
 

..+) _. H. .3 .3
+ . + . .
a N3 . T a w\\\\. T . l
w .mwmmmwxwxww m . w mmmmmmmm m H Ha
O O . O 0 W M M
N N L _ N N L O O
3 3 E 3 3 E M N N
O O _ 0 O .. ...s ...s
E s . . ) E N O O
\...\\\\\\\\\\\\ a. d am my. a u
9.33wa A ( . A .
w M
L
w m m 0 m. m w m m 0 m. m mo. m m o m. m
NE:.E >986 885m N826: >905 825m NE\_.E >925 momtzw
) . M. .1 _ T
H a m. . a o ....
a T_ +a T Hm
M M M M L M M
O O _ a H. O O m E . O 0
N N H .._ N N N u N N L
. a 6.1:... -... ..-T. .... a E V..... .. . a m
.% ... n ...
ww
L ._
w 3 _ _ _ l 3 .
m m. .... o m. m m 4 2 ._. 4 m .... _ 3
~83: >920 meatsw NEEE >926 oomtzw NEE: >905 oomtzw

Figure 3.6. LW, AB, EL and Total energy of SRNOM-SRNOM interactions for
non-ozonated SRNOM (a, c, e) and ozonated SRNOM (b, d, t) at pH 3 (a, b) pH 5 (c, d),

Short range NOM-NOM and NOM—Ti02 interactions were governed by polar
(acid-base) and van der Waals forces, while the contribution of electrostatic

interaction (A655) to the overall free energy of adhesion was found to be minor

(Figures 3.5 & 3.6). The most salient trend in the interaction energy was the effect of
ozonation at pH 3 and pH 5, which, in the absence of calcium, increased the relative

contribution of hydrophilic repulsion between SRNOM molecules and between

SRNOM and TiOz. Basically, at low pH, ozonation made the SRNOM-SRNOM and

SRNOM-TiOz interactions less favorable. Dramatic changes were observed in the

electron donor and electron acceptor components of free energy of the SRNOM

interactions (Table 3.2). Ozonation led to an increase in the value of the electron

donor component of free energy (y- ), resulting in more positive values of the
acid-base interaction energy ([1ij). This increase, reﬂected by the increase in

hydrophilicity determined in contact angle measurements, was more pronounced at
lower pH.

Combined surface energies due to the three non-covalent interfacial energies (LW,
AB, EL) were presented in Fig 3.7 The addition of calcium decreased differences in
interaction energies for both ozonated and nonozonated SRNOM. In view of the
observed strong effect of calcium on adsorption and permeate ﬂux [156], this

observation indicates that non-XDLVO forces (e.g., steric effects, bridging by calcium)

play an important role in SRNOM-SRNOM and SRNOM-TiOz interactions.

78

 

 

-
m
3
..

 

 

 

NOM (Ca++)
---...4 03-NOM
E1 03-N OM (C a++)

 

_ D NOM

 

 

 

4.

7... ...“;
.... .. .4

«:7

“whit
.3..
_

 

//////////// 5w

 

 

  

 

 

 

 

 

     

a: .....1
3 a ..v

.r .1... .473...

r .\.l . .L

...» .1...i_m$.ut

1 TL 2 .11.: w
4.........

.7, t;

3......
/
7/////.

        

      

  

I
...

/

.

 

 

 

 

 

ANEEEV 628550 202.202 ho >9mcm moi

 

 

(a)

 

 

E1 NOM

 

I. .

   

 

 

 

 

 

O3-NOM (c a++)

NOM (Ca++)
03-N0M

rill-Illilttaoti. . l

 

 

. .3 .. . .
... ..~.,., .

%.%%

           

.. . ......1

 

   

u

.. “5.1.1.1....
“
/ on I.// n I . .. n
1"; a l I la y
. . . I . I? / . .
WI/ ..ZwuN,//./.vv..uv7n..
. .Wn .r. . ..

... :
..L...
v.u

 

 

 

 

 

 

 

 

.. ... . .4.. <
.. .. ,, .... . .11. .
.. . .3... .. . .... .
- ..- ..
L ....l. X .. . ... .

 

..
w
,.

b

 

 

 

 

m

15 r
10
5
0

~E\_.E 60628 «9.75.02 .6 >995 moi

 

 

(b)

Figure 3.7. Combined effect of pH, Ca2+ and pre-ozonation on the XDLVO energy
of interaction between (a) SRNOM molecules, and (b) SRNOM molecule and TiOz

surface.

The error bars correspond to 90% conﬁdence intervals.

The trends observed for the free energy of the SRNOM-TiOz adhesion as a

79

function of ozonation, pH and calcium content (Table 3.2, upper half) were

qualitatively the same as those for the energy of cohesive interaction of a pair

SRNOM molecules implying that the SRNOM-TiOz interaction energy is determined
by the properties of the SRNOM molecules and not properties of the Ti02 surface. It
follows that the SRNOM sorption onto TiOz is insensitive to the amount of SRNOM

already absorbed on the TiOz surface. This indicates that when Ti02 nanoparticles
enter natural water, as long as the interaction between NOM molecules and interaction
between NOM and Ti02 are favored, Ti02 nanoparticles will be covered by

multilayers of NOM molecules.

3.5. Conclusions
The surface energy analysis described in this study provides a reasonable

interpretation of the effect of ozonation in reducing NOM fouling of Ti02 ceramic

membranes. When natural water is being ozonated, the average size of NOM

molecules become smaller, and their surface becomes more hydrophilic. Given this

change in surface chemistry, the interaction between Ti02 and NOM becomes less

favorable mainly due to hydrophilic repulsive forces between the NOM and Ti02.
This ozonation effect is also observed in the adsorption study where adsorption of
ozonated SRNOM on Ti02 nanoparticles was signiﬁcantly less compared to the

adsorption of SRNOM on the same particles, with all the other water conditions being

the same (e.g., ionic strength, pH and presence of calcium). It is also true that the

adsorption of the ozonated SRNOM onto the membrane is improved when TiOz has

80

 

__ .934 v.M"' ‘ry
ﬁ‘
1 t f 5‘
.
_ __ _. -.-

positive charge (at low pH). This ﬁnding indicates that when Ti02 particles enter into
natural water, NOM will have lesser tendency to adsorb on Ti02 if the water has been

oxidized to some extent. Therefore, it is likely that Ti02 particles will keep their

natural surface chemistry and exists as isolated particles without much adsorption of
NOM in such pre-ozonated natural water.

There is no obvious and general trend for the pH effect on changing surface

energies between NOM/NOM and NOM/TiOz when there is no calcium present,

although the surface charge of NOM and TiOz particles do show a strong correlation

to the pH of the suspensions. However, the adsorption data does show that the amount
of sorbed SRNOM decreases with pH in the absence of calcium. This phenomenon is
attributed to the fact that SRN OM adopts a more compact conformation at lower pH,

which is partly proved by the fractioning results. Thus at lower pH, SRNOM can pack

more densely at the Ti02 surface.

In view that the strong observed effect of calcium on adsorption and permeate ﬂux
is not consistent with the analyzed calcium effect by XDLVO modeling, this

inconsistency indicates that non-XDLVO forces (e.g., steric effects, bridging by

calcium) may play an important role in SRNOM-SRNOM and SRNOM-Ti02

interactions. Calcium ions are considered to serve as bridges between NOM-TiOz and

NOM-NOM and allow for a multi-layer adsorption to occur.

81

Chapter 4. Hybrid dual media membrane

crossflow ﬁltration hydrocyclone system for

produced water treatment

4.1. Introduction

Produced water is the water that comes to the surface with oil and gas during
offshore drilling operations [192]. The chemical makeup of produced water depends
on water composition in the location of drilling and on the chemical additives used at
the offshore platform; dissolved salts and organics, inorganic particles, and, of course
oil and grease, are main constituents. The typical range of oil concentration in

produced water is in the (100 to 5000) mg/L range [193]. Produced water is generated

in extremely large quantities. For example, more than 1.6 billion m3 of produced

water were generated in 2002 in the US. alone [194,195]. This is equivalent to a
volume of 1.15 billion gallons per day. A recent report on produced water

management estimated the total volume of produced water generated in 2007 to be

about 21 billion barrels, which equals to 3.2 billion m3 per year [196]. Nearly 1

million wells that actively produce oil and gas were present in the US. by the time the

report was published [196]. This high rate of increase in generating produced water

(1.6 billion m3 in 2002 and 3.2 billion In3 in 2007) was caused by both the increasing

demand for oil and by the decreasing production rate of developed oil wells with time

as the water-to-hydrocarbon ratio increases over the life of the well. A typical

82

 

oil-producing platform produces large quantity of produced water from daily

operations (0.66 m3/s [197]). Besides the US, other countries also produce extremely

large amounts of produced water every day. It has been reported that more than 2,000
million tons of oily wastewater were produced by oil reﬁneries in the European Union
countries [198]. Based on the allowable oil concentration in discharged water (15 to
40 mg/L [199-202]), these waters cannot be directly discharged back to the water

system due to their high oil content. The high oil concentration as well as the large

quantity of produced water from daily operations (0.66 m3/s [197]) makes

cost-effective treatment necessary. The need for cost-efﬁcient technologies for
oil-water separation has become more urgent in light of the recent oil spill events in
the Gulf of Mexico.

Oil with droplet sizes smaller than 20 pm is recognized as emulsified oil while the
other two categories of oil droplets are dispersed oil (20 pm to 150 um) and ﬁ-ee oil
(>150 pm) [203]. Membrane separation has been studied as an approach to oily
wastewater treatment [204]. While there are other conventional ways to achieve oil
water separation, such as centrifugation [205], dissolved air ﬂotation [206,207] and
hydrocyclone separation [208], their efficiency decreases dramatically as the size of
the oil droplets drops below approximately 20 um [209,210]. Given that a substantial
fraction of oil in produced water may exist as emulsiﬁed oil (droplet size less than 20
pm), a multi-step treatment is often required to reduce oil concentration to comply
with environmental regulations. For example, coagulation is employed as a
pretreatment step to destabilize the emulsiﬁed oil [205-207].

One of the beneﬁts of using membrane ﬁltration is that small oil droplets (< 20 um)
can be removed from the oil-water dispersion [211]. However, fouling of the

membranes by oil prevents the acceptance of membranes as a standard means of

83

 

de-oiling water [212].

Membrane-based separation techniques have been used to separate oil-water
dispersions generated in a range of industrial processes. For example, dehydration of
oil emulsion by pervaporation utilizes hydrophilic cellulose ultraﬁltration membrane
with oil retained in the feed [213]. Hydrophobic polypropylene microﬁltration
membranes were applied in membrane distillation process to extract water vapor from
a feed dispersion with low oil concentration [214]. Flocculation was applied together
with microﬁltration to attain higher sustainable water permeate ﬂux [215,216]. There
have been studies that used a one-step membrane treatment to separate oil-water
dispersions [217-220]. In these studies, the choice of membranes and operating
conditions is varied because of two main reasons. Firstly, characteristics of the
oil-water dispersion in terms of droplet size, concentration, stability and chemical
composition vary considerably due to large differences in the preparation procedures,
the oil/water formulation, and the chemical conditions used in preparing the oil-water
dispersion [221]. Secondly, membranes are often modiﬁed or hybridized by other
technologies to enhance their performance.

The inclusion of centrifugal separation into membrane ﬁlters was implemented by
placing a helical insert inside a cylindrical metallic membrane [197]. With the insert
there was sustainable water ﬂux at a constant transmembrane pressure, while the
membrane without the helical insert required a large increase in pressure to maintain a
similar permeate ﬂux. This improvement was attributed to the creation of a forced
swirl/vortex that helps to separate oil from water and thereby mitigating oil fouling
[197].

Adding air sparging into membrane ﬁltration is known to be effective in reducing

concentration polarization and fouling in various membrane applications (e.g.,

84

wastewater treatment [222,223] and surface water treatment [224]) In Li’s study [225],
the injection of air into a tubular polymeric membrane rapidly increased the permeate
ﬂux and decreased the permeation of protein. These phenomena were attributed to the
enhanced local mixing near the membrane surface or the secondary ﬂow generated by
the air bubbles. An air sparging system was incorporated into the hydrocyclonic
separator and it was demonstrated to be effective in solid-liquid separation [226].

The speciﬁc objectives of this work were: 1) to evaluate the oil water separation
performance of Ti02 ceramic membranes with different pore sizes for dispersions that
may contain a surfactant; 2) to design a hybrid ﬁltration/separation system that
incorporates membranes with opposite affinities; 3) to evaluate the efﬁciency of air

sparging and the effect of hydrocyclone characteristics on the membrane ﬁltration

performance.

4.2. Basics of hydroclone and oil water separation

The proposed hybrid crossﬂow ﬁltration system has some of the operating

characteristics of a centrifugal separation device such as a hydrocyclone.

4.2.1. The separation principle of hydrocyclones
The hydrocyclone was invented as a water puriﬁcation device in 1891 by Eugene
Bretney [227]. The initial idea was to separate water or other liquids from impurities

by employing the centrifugal force, caused by the swirling motion of the dispersion

introduced under pressure into the apparatus. In the 20th century, hydrocyclone

designs have been suggested for liquid-liquid separation [228] with reverse ﬂow [229]

and forward ﬂow conﬁgurations (Fig. 4.1). The separation is based on the speciﬁc

85

 

gravity difference between phases in liquid-liquid emulsions. The separation principle
can be described as follows: the emulsion enters the chamber in a tangential direction
through an inlet tube connected to the sidewall of the cylindrical portion of the
cyclone and creates a primary vortex ﬂow. The liquid with the higher density is driven
by centrifugal force to the outer layer of the liquid with lower density; therefore, the
heavier liquid ﬂowing next to the interior surface of the sides of the cyclone is forced
to ﬂow toward the apex of the cone by the pressure exerted from the air cushion that
forms above the water level. This heavier liquid exits the bottom of the cyclone
through the underﬂow oriﬁce. The lighter liquid migrates to the centerline of the
cyclone and exits from the overﬂow oriﬁce. The equation that describes the relative
radial velocity of a droplet of the dispersed phase in the continuous secondary phase is
given by:
2

(Pd-P )d2 “d,t
[ud,r-c,.]= my: - r (4.1)

where x is the droplet diameter, pd is the density of dispersed phase, pc is the

density of continuous phase, and r is the radial position of the drop within the

hydrocyclone, ud’, and “car is the radial velocity of the dispersed phase and the
continuous phase respectively, and um is the tangential velocity of the dispersed

phase.

86

(a) (b)

 

 

 

 

    
 

 

 

 

 

Overﬂow —'
Feed
C:
t Vortex ﬁnder
3.
Feed D
Vortex
stabilizer
Inner helical flow Outer helical
‘ ' ﬂow and
boundary layer Underﬂow
Vortex ﬁnder
—->
rApex
1 Overﬂow
Underﬂow

Figure 4.1. Diagram of a reverse ﬂow (a) and a forward ﬂow hydrocyclones (b) [230]

The hydrocyclone’s shape of an inverted cone increases the angular velocity of the
ﬂuid as the diameter is reduced (for the angular momentum to be conserved). This
increase in angular velocity results in an improved separation efﬁciency of the

dispersed and continuous phases.

4.2.2. Critical pressure

The Young-Laplace equation [231] was applied to compute the critical pressure.
The critical pressure sets the upper limit of the transmembrane pressure that can be
applied to the membrane where the liquid can be eliminated to break into membrane
pores.

20' cos (9
Pcrit = -—-—- (2)

r pore

87

where c is the interfacial tension between the two liquids, 0 is the contact angle of

membrane with that liquid and rpore is the pore radius of the membrane.

When the droplet size was taken into account, the equation of critical pressure is

 

 

 

 

 

 

calculated as [210,232]:
_ l .
K W3
20' cos (9' 2 + 3 cos 0' — cos3 6'
P... -—-—— 1- 3 <3)
pore r . v I
4 ﬂ cos36 -(2-3sin6 +sin36’)
\ r pore )

where rdmp is the droplet radius and 6' = 180 —— 6.

4.3. Material and methods

4.3.] . Materials

Kerosene was chosen as the model oil in this study. The oil concentration was
determined through the measurement of copper concentration via Atomic Absorption
spectrometer (AA, see section 4.3.2). The copper was exchanged from a water phase
to an oil phase with a reagent called Accorga (5050, Cytec Industries Inc, New Jersey).
To prepare the oil stock solution, 100 g of kerosene was gently stirred with 20.5 g of

Accorga solution for 5 min. Then a solution of copper sulphate was prepared by

dissolving 5.6 g of CuSO4-5H20 in 50 g of deionized water. From this solution 35 g

were added to the oil/Accorga mixture and this was gently stirred for another 30 min.

The color of the oil phase turned from yellowish to brown, which indicated that the

88

copper was being transferred to the oil phase. Stirring was then stopped, the phases
rapidly separated, and the aqueous phase was removed from the oil phase using a
separating funnel. This stock kerosene solution is ready to be added into water for an
oil-water dispersion at different concentrations. In a subset of experiments, polyvinyl
alcohol (PVA, MW=25,000, 88% hydrolysis, Polyscience Inc, Warrington, PA) was
used to stabilize the oil-water dispersion [210,233]. For dispersions with PVA present,
the stock oil was mixed with a solution of PVA in water. The PVA-water solution was
prepared by dissolving PVA crystals in approximately 60 °C DI water at the
concentration of 0.2 g/L.

The tubular ceramic membranes made from titania with different pore sizes were

manufactured by TAMI Industry (Filtanium, TAMI North America, St. Laurent,

Quebec). The polytetraﬂuoroethylene (PTFE, 1.1 g/cm3, Teﬂon) hollow ﬁber

membranes were generously provided by Markel Corporation (Plymouth Meeting,

PA).

4.3.2. Characterization of oil-water dispersions and membranes

Light scattering system (Malvem Mastersizer, Westborough, MA) was used to
determine oil (refractive index = 1.45) droplet sizes in the retentate water. The sample
was mixed with 1000 mg/L piperazine (Aldrich) to “freeze” the oil droplets before the
measurements.

Flame atomic absorption (AA) spectroscopy (Perkin-Elmer 1100) was used to
determine oil concentration via the measurement of copper concentration. In order to
measure the concentration of copper in the oil phase, the copper was stripped back
into an aqueous solution using sulphuric acid with pH of 0.5. The copper

concentration in the aqueous solution was analyzed at the wavelength of 324.7 nm. To

89

ensure that all copper was transferred to the aqueous solution, samples were sonicated
in the sonication bath for 5 min before the measurements. A calibration curve (Fig.
was obtained by plotting the absorption versus the concentrations of the diluted

copper standards (SC194-500, Fisher Scientiﬁc, Pittsburgh, PA).

...

Absorption

 

 

 

 

0 5 10 15
Concentration of copper (ppm)

Figure 4.2. Calibration curve for copper

Scanning electron microscope images were recorded using a Hitachi S-47OOII
ﬁeld emission SEM operated in ultrahigh resolution mode. Both the cross section and
the skin surface were sputter coated with 5 nm of gold (Pelco SC-7 auto sputter coater)

before imaging.

4.3.3. Hybrid crossﬂow ﬁltration system

The scheme of our hybrid crossﬂow ﬁltration setup is shown in Fig. 4.3. The feed
oil-water dispersion was prepared in a 5 gallon stainless steel reservoir (Alloy
products corp., Waukesha, WI). The stock oil solution was diluted with deionized
water in ratios of 1:808 (vol: vol). The resultant oil concentration was close to that
used in studies of oil/water hydrocyclone separation [234,235]. The 8 L dispersion

was stirred with an electric powered stirrer and continuously mixed by circulating the

90

dispersion using a diaphragm pump (Hydra-cell M-O3) in order to obtain a dispersion
with desirable oil droplet size distribution. The diaphragm pump continued pumping
the dispersion during the entire ﬁltration. The temperature of the feed water was

maintained at 20 3 2° C.

The feed water was sent to the membrane with a low shear Moyno pump (33201,

Springﬁeld, OH). Three tubular ceramic Ti02 microﬁltration membranes with pore

sizes of 0.14 pm, 1.4 pm and 3.5 pm were selected in the study. All membranes had

inner diameter of 6 mm, length of 25 cm and surface area of 47 cm2. The membranes

were housed in a stainless steel ﬁlter holder (TAMI). In the dual membrane ﬁltration
experiments, a hydrophobic PTFE hollow ﬁber was installed inside and along the
centerline of the ceramic membrane.

To sweep the oil off from the inner wall of the hollow ﬁber, air was applied
continuously and injected into the lumen of the hollow ﬁber at the rate of
approximately 500 ml/min. (Fig. 4.4). A back pressure regulator (BP-3, Circle Seal
Controls Inc., Corona, CA) was used to maintain the transmembrane pressure
differential at a constant value during the experiments. A pressure gauge was installed
in the retentate line to measure the pressure at that location. One ﬂow meter was
installed in the permeate line, and one ﬂow meter and one pressure sensor were
installed in the retentate line. Thus, two ﬂow rates and the transmembrane pressure
could be recorded automatically to the computer, which has a Labview program

installed.

91

LabView

data acquisition

pressure

  

 

AAA‘ AAA‘ AAAAAAAAAAAAAA - A

 

 

 

vvvv vv v vvvvvvv
AAAAA AAAAAA
vvv‘vvv vvvvvvv

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I
‘..v-.'- h- »- -’.,
- V: r ' i '- u
' ”.2.- ;r; .N'
c I er 2»:
n. r: (Y‘- - -~‘
I" \‘I_’ ~
’I":l" -
ran-K
\ u I

 

 

diaphragm pump feed tank

Figure 4.3. Schematic diagram of the hybrid crossﬂow ﬁltration system

 

 

 

   

 

Arr ->_ 5 -> Permeate
Eigﬁ§$53335534 ' ” (0")
i Concentrate
(oil/water)
Permeate
(water)

Figure 4.4. Schematic diagram of the arrangement of hollow ﬁber membrane
within a ceramic membrane.

The hybrid membrane ﬁltration system that incorporates ﬂow management to
introduce rotational ﬂow of the oil-water dispersion was assembled by installing a
rotating tube upstream from the membrane unit (Fig. 4.3 and Fig. 4.5 (a)). The two
swivels (Rotary Systems, Inc., RAMSEY, MN) were installed at the two ends of the
rotating tube allowing for the free rotation of the tube but keeping other parts of the

assembly stationary. The rotating tube was rotated by an electric motor via a

92

belt-and-pulley arrangement. Pulleys of several sizes were used to control the
rotational speed of the tube. A unique attribute of this system is that the axial and
rotational velocity of the ﬂow can be changed independently of each other by
changing the crossﬂow speed and the rotational speed, respectively. In Fig. 4.5 (b),
the ﬂow pattern inside the crossﬂow membrane ﬁltration system is illustrated. The
forced swirl of the oil-water dispersion helps separate the oil from water. The water
permeates through the outer hydrophilic membrane while the oil droplets migrate

towards the centerline where the inner hydrophobic membrane is located.

(a)

 

 

  

(b)

concentrate

     
   

stream
feed stream
under imposed .
rotation: _ , . , , ' permeate
produced water ’ stream #2:
(oil-in-water dewatered oil
mixture)

 

membrane #2: membrane #1:

superhydrophobic hydrophilic
hollow fiber ceramic
membrane membrane

Figure 4.5. Schematic diagram of (a) hybrid crossﬂow membrane ﬁltration unit and
(b) the ﬂow pattern inside the crossﬂow membrane ﬁlter in a hybrid ﬁltration
system.

By injecting air into the feed water immediately upstream from the membrane unit

(see Fig. 4.3), the efficiency of air sparging in reducing membrane fouling by oil was

93

 

evaluated. During these experiments, an air injection line was installed close to the
inlet of the rotational tube; air injected under pressure formed bubbles that were
interspersed with the feed. Air pressure was controlled by a pressure regulator; air
speed was monitored by an air ﬂowmeter; and the air line was connected to a check
valve and a shutoff valve.

The hybrid crossﬂow membrane ﬁltration system as described above combines air
sparging, hydrocyclonic separation and membrane separation processes (Table 4.1).
As a result, the individual effects of each of the three processes, and effects of the
processes applied in various combinations can be evaluated. For example, by injecting
air into the membrane ﬁltration system, we can evaluate the effect of air sparging on
membrane ﬁltration. By rotating the tube, we expect swirl motion of the emulsion to

produce centrifugal separation similar to a hydrocyclonic operation.

Table 4.1. Hybrid separation processes resulting from combining air sparging,
hydrocyclonic separation and membrane ﬁltration.

(H) implies hydrocyclonic separation and (A) implies air sparging

 

 

 

 

Fouling mitigation
strategy Without air sparging With air sparging
Flow type
Crossﬂow membrane Air sparged crossﬂow
Without swirl ﬁltration membrane ﬁltration
(CFF) (ACFF)
Crossﬂow membrane Air sparged crossﬂow
With swirl ﬁltration hydrocyclone membrane ﬁltration
(CF FH) hydrocyclone (ACFF H)

 

 

 

 

94

 

4.4. Results

4.4.1. The effect of membrane pore size, and the presence of surfactant on membrane
ﬁltration performance

The experimental results presented in this section were obtained with only
membrane ﬁltration without adding other hybrid technologies. A series of ﬁltration
experiments was carried out with ceramic membranes of three different pore sizes to
evaluate the effect of membrane pore size on permeate ﬂux and oil rejection. The
experiments were conducted at different transmembrane pressures to make sure that
the three membranes with different permeabilities have a similar initial permeate ﬂux
during oil-water separation tests. The measured permeability values for the three
membranes are given in Table 4.2. The cross-ﬂow velocity was 0.59 m/s with
Reynolds number of 3,524. The produced oil droplet size was in the range of (l to 100)
pm when there was no surfactant present in the dispersion The oil concentration in
the retentate was 170-200 mg/L. Figure 4.6 (a) shows clearly that the 3.5 pm ﬁlter
experienced the highest ﬂux decline over the 15 min of ﬁltration while in the tests
with the 1.4 pm ﬁlter 30% permeate ﬂux decline was observed over 15 min of
ﬁltration. The smallest pore size membrane (0.14 pm) has shown complete oil
rejection, however, this good rejection is a tradeoff between its lower performance in
maintaining a high permeate ﬂux as compared to the 1.4 pm ﬁlter. As for the oil
rejection, the two larger pore sized ﬁlters rejected more than 90% of the oil, while the

0.14 um ﬁlter completely rejected the oil (Fig. 4.6 (b)).

95

Table 4.2. Water permeability of different ceramic membranes

 

 

 

 

 

 

 

 

 

    
  

 

 

 

Membrane pore size Permeability (L/mZ/hr/bar)
3.5 pm 1.1E+04
1.4 pm 9.6E+03
0.14 pm 1.9E+03
(a) (b)
250 -0-3.5 umﬁlter 100 “ME
+ 1.4 um ﬁlter
: + 0.14 um ﬁlter Q 80 7 .......................... . ...................... . ......
................. E 60 _I
E40 -
§ —G— 3.5 um ﬁlter
20 - +1.4 um ﬁlter
0 I I I I 0 +0.14 um filter
0 517018 (minio 15 0 5Time (mini0 15

Figure 4.6. Permeate ﬂux decline (a) and oil rejection (b) of three membrane during
the ﬁltration of oil-water dispersion without PVA

Another set of ﬁltration experiments was conducted to evaluate the effect of PVA
on permeate ﬂux and oil rejection. The addition of PVA resulted in the following
changes: 1) modiﬁcation of oil droplet size and the oil concentration in aqueous
dispersions; 2) ﬁltration performance of hydrophilic ceramic membranes in terms of
oil rejection and permeate ﬂux; and 3) modiﬁcation to the hydrophobicity of a
hydrophobic polymeric membrane.

It was found that the addition of PVA does create an oil-water dispersion with a
higher oil concentration in the aqueous phase and smaller oil droplet size. The
difference in droplet size caused by adding PVA into the dispersion is illustrated in

ﬁgure 4.7. Nanosized particles (< 0.1 pm) were found in the oil-water—PVA

dispersion.

96

 

 

 

12 _ ..........................................................................
—-oil/water

g 8— -------------------------------------------------------------------------

‘e’ 6 -

3

'o ,r-

> 4- --------------------------------------------------------
2 3 ........... f, ............................... v ...........................
O i T I ... l j l
001 01 1 10 100 1000

Particle size (um)

Figure 4.7. Oil droplet size distribution with and without the presence of PVA in
the retentate during the ﬁltration of 3.5 pm ﬁlter

The three membranes (pore sizes 3.5 pm, 1.4 pm and 0.14 pm) were tested with a
dispersion that had PVA concentration of 0.2 g (PVA)/L. In experiments with the 3.5
pm ﬁlter, the permeate ﬂux decline was smaller than what was observed in the
ﬁltration of the PVA-free dispersion. But considering oil rejection data, it is clear that
the improvement in permeate ﬂux was due to the fact that a large proportion of oil
passing through the membrane and not contributing to membrane fouling. In
dispersions with PVA present (Fig. 4.7), most of the oil droplets were smaller than 1
pm. As a result in the ﬁltration with 1.4 pm ﬁlter, oil droplets could potentially pass
through the membrane pores and come out from the permeate stream; this is
consistent with the observed oil rejection of less than 40% over the ﬁrst 15 min of
ﬁltration. In contrast, the 0.14 pm ﬁlter had the most severe ﬂux decline among the
three membranes and the ﬁnal permeate ﬂux after 15 min of ﬁltration was much

smaller compared to the permeate ﬂux after the same duration of ﬁltration of

97

 

PVA-free oil-water dispersion. However, there is still no oil permeating through,
which means a 100% complete rejection of oil can be achieved with membrane
ﬁltration that has the nominal pore size smaller or equal to 0.14 um. So the signiﬁcant
fouling of 0.14 pm ﬁlter occurred because all the oil droplets rejected by the
membrane are being retained on the surface. This series of ﬁltrations has shown a
clear trend that gains in oil rejection can be realized but at the expense of large ﬂux
declines. And removal of the small oil droplets, which are produced when PVA is

added, requires using the membranes that have pore size in the 0.1 pm size range.

 

   

 

 

 

(a) (b)
250 100 1 ------ ‘r a a 1‘
3200 um ﬁlter \’3 80 ~ --------------------- -D— 3.5 um ﬁlter
'g 9.. +1.4 urn ﬁlter
T5150 5 60 - ----------------------- +0.14 umﬁlter -
E100
“—- 50 m 20 _. ......... .W
0 I r 17 r 0 I l l
0 15 0 5 10 15

5T: me (min i 0 Time (min)

Figure 4.8. Permeate ﬂux decline (a) and oil rejection (b) of three membranes
during the ﬁltration of oil-water dispersion with 0.2 g/L PVA

In addition to the effect of PVA on ceramic membrane ﬁltration, an study of PVA
adsorption onto a hydrophobic PVDF membrane (GVHP, 0.22 pm, Millipore) surface
showed that the adsorbed PVA signiﬁcantly changes the surface contact angle of the
membrane. In one series of experiments, an original membrane had an average water
contact angle of 110°. Then the membrane was soaked in 0.2 g/L PVA aqueous
solution for 24 h and rinsed with water gently and dried. The average contact angle
became 40°. The exposure to PVA essentially changed a hydrophobic membrane into

a hydrophilic one. We also tested oil permeability of this type of membrane. It was

98

 

found that the original PVDF membrane can easily ﬁlter out pure oil from an
oil-water dispersion in a deadend ﬁltration experiment. However, when the oil-water

dispersion had PVA present, the membrane was no longer effective in removing oil.

4.4.2. The performance of hybrid membrane ﬁltration system

The hybrid system combines membrane ﬁltration with two other processes to
improve the oil and water separation performance (Table 4.1). The combination of
membrane ﬁltration and hydrocyclonic separation has a potential to achieve better oil
water separation for two reasons. First, the membrane with appropriate pore size and
surface chemistry can serve as a barrier for smaller oil droplets (e.g., < 20 um) ,which
are difficult to remove from water by centrifugal separation within a typical deoiling
hydrocyclone (also see section 4.4.1). Second, the rotational ﬂow would result in
centrifugal transport of oil away from the membrane surface leading to reduced
membrane fouling by oil during separation. Basically, the two technologies would
synergistically enhance one another.

The addition of air to the feed water upstream from the membrane module should
result in a further mitigation of membrane fouling due to ﬂotation and sparging effects.
By attaching to oil droplets (ﬂotation principle), air bubbles would facilitate the
centrifugal transport in the hybrid system [226,236]. Sweeping of oil off the
membrane surface by the air bubbles is also expected to contribute to fouling control
[237].

To test the hybrid treatment system in different conﬁgurations, we have conducted
ﬁltration experiments under four different conditions (Table 4.1). During the ﬁltration,
we monitored the change in the size of oil droplets in the retentate stream (Fig. 4.9).

Imposing rotation (CFFH process) did not seem to affect the size distribution of oil

99

 

droplets. In contrast, the addition of air to the feed resulted in a shift of the peak of the

size distribution of oil droplets to a larger size value.

(a)

(b)

 

 

 

 
 
    

  

     
 

 

 

".
....................... It .
I, lllfrr
09- lm'ETis-vﬁirﬂlul

  

  
  

 

 

0.1 1 10 100 1000 10000

Oil droplet size (um)

Figure 4.9. Oil droplet size distribution measured for the retentate at 5 min (a) and

at 30 min (b) of ﬁltrations with 0.14 um ﬁlter at different hybrid conditions.

CFF - crossﬂow ﬁltration; ACFF — Air sparged crossﬂow ﬁltration; CFFH — crossﬂow ﬁltration

100

hydrocyclone; AC FFH — air sparged crossﬂow ﬁltration hydrocyclone

 
  
 
 

.0
oo
i

.0
O)
1

Normalized Flux

 

 

.0
4:.

F ilfpation time (iiiin)

Figure 4.10. Permeate ﬂux decline during the ﬁltrations with 0.14 pm ﬁlter at
different hybrid conditions.

The initial permeate ﬂux is approximately 170 Urn /hr. Transmembrane pressure is 11 psi.

The permeate ﬂux appears to be affected by the introduction of additional features
(Fig. 4.10). It should be mentioned that the oil concentration in the control experiment
(160 mg/L) is lower than that in the other three experiments (280-330 mg/L). Thus we
would expect a more signiﬁcant permeate ﬂux decline than what would happen in the
control experiment if the oil concentration was similar to that in the other ﬁltration
experiments. Imposing rotation to the feed water did not help to the decrease the
initial (i.e., over the ﬁrst 20 min) permeate ﬂux decline (Fig. 4.10) and the data
actually shows that the permeate ﬂux declined faster when rotation was introduced.
After 20 min of ﬁltration, there was no further permeate ﬂux decline indicating that at

this point, that membrane oil fouling had reached an equilibrium. As for the effect of

101

air sparging on crossﬂow membrane ﬁltration, an improvement in ﬂux decline
became apparent after 15 min of ﬁltration and air sparging became more beneﬁcial in
retaining a higher ﬂux when rotation was applied. The ﬁnal permeate ﬂux values
observed in experiments with the hybrid processes were higher than the ﬁnal
permeate ﬂux value recorded in the control experiment (membrane ﬁltration only),
even though the oil concentration in the control experiment was smaller than the oil
concentrations in the other experiments with hybrid processes. In none of the
experiments was oil detected in the permeate line. The observation in oil droplets can
be correlated to the results in permeate ﬂux during the ﬁltration experiments. The
larger size oil droplets resulted from the addition of air (Fig. 4.9) can partially explain
why air sparging is effective in maintaining relatively high permeate ﬂux. Air bubbles
seem to promote the coalescence of oil droplets, which in turn helps to reduce
membrane fouling and thus maintain high permeate ﬂux.

These preliminary results make a good starting point for further studies. It is also
evident that the major ﬂux decline occurred during the ﬁrst 15 min of ﬁltration;
reduction of this ﬂux decline rate to some extent by judiciously combining individual
processes into one hybrid unit would perhaps result in the largest improvement of the

overall performance.

4.4.3. The performance of dual membrane system

The dual membrane system has been tested to explore the potential of using
membranes with different chemical afﬁnities to separate an oil-water dispersion into
two separate phases. The difference between a typical deoiling membrane ﬁltration
system and the dual membrane system is that the latter could extract oil from the

dispersion. By doing so, oil concentration in the feed dispersion is lowered and thus

102

help to mitigate oil fouling of the hydrophilic membrane.

50 ml kerosene stock solution was mixed in l L deionized water (4,400 mg/L) and
ran the dispersion through dual membrane system (1.4 um ceramic membrane ﬁlter,
PTFE hollow ﬁber (Fig. 4.11) operated at the transmembrane pressure of 10 psi and
with the sweeping air ﬂow rate of approximately 500 ml/min). The permeate coming
out from the lumen of the hydrophobic membrane was approximately 2 ml of pure
kerosene for the initial 15 min. Figure 4.12 shows the retentate, the permeate from the
outer ceramic membrane, and the permeate from the inner hollow ﬁber membrane

after 45 min of ﬁltration.

 

Figure 4.11. SEM images of the outer surface (a) and cross-section (b) of the PTFE
membrane

103

 

 

Figure 4.12. The images of samples of the retentate (sample on the left), the
permeate from 1.4 pm ceramic membrane (3 samples in the middle) and the
permeate from the hollow ﬁber membrane (sample on the right).

4.5. Conclusions
Three ceramic (TiOz) ceramic membranes of different pore sizes were tested as

porous walls of a sparged hydrocyclone separating oil-water dispersions. The pore
size was found to be a critical factor in determining permeate ﬂux and oil rejection.
The addition of PVA also had a profound impact on the separation performance.
Although the complexity of the chemical makeup of produced water generated in the
ﬁeld necessitates tests of the developed hydrocyclone system with speciﬁc water
compositions; the general trends in the separation performance with the membrane
pore size and surfactant content are expected to hold.

The feasibility of afﬁnity-based membrane separations of oil-water dispersions was
also demonstrated. The demonstrated possibility of the recovery of oil can be prone to

use in oil spill clean up application and produced water treatment operations.

104

Chapter 5. Summary and future work

The development of nanotechnology can greatly beneﬁt membrane-based
separations for water treatment. In our study, we found that it was possible to design
high performance polyelectrolyte multilayer membrane with high permeability and

high selectivity using layer-by-layer deposition method. We also studied the

interactions of nanoscale Ti02 and NOM molecules in aqueous solutions to explain

the effect of ozonation in reducing NOM fouling of Ti02 ceramic membranes. Such

understanding can result in better predictions of ceramic membrane fouling by
NOM-containing waters as well as the fate and transport of TiOz in the natural
aqueous environment. Some prospects of future work are discussed below.

In the PEM study (Chapter 2), we have hypothesized that the improved separation

efﬁciency with nanoparticle-fouled PEM ﬁlms could be due to a change in the surface
charge induced by the deposition of Si02 nanoparticles. To test this hypothesis,
positively charged or neutrally charged nanoparticles can be ﬁltered through PEM
ﬁlms and the change in rejection by the PEM ﬁlms can be evaluated. In addition,
PEM ﬁlms can be modiﬁed with nanoparticles of different sizes and shapes (e. g.,

nanotubes, and nanoplatelets) to evaluate the importance of surface morphology on

membrane separation performance. Considering that swellability of PEM ﬁlms such

as [PAA/PAH]n [62,238] and [PSS/PDADMAC]n [153] is strongly dependent on

PEM deposition conditions (e.g., pH and ionic strength of the polyelectrolyte

solutions), it is proposed to deposit a monolayer of nanoparticles with an average

105

diameter comparable to the maximum swelling thickness of the ﬁlm. In view of the
possible use of deposited nanoparticles as catalysts in certain reactions, controlling the
swellability can be used to adjust the access of nanoparticles to permeating species
during the reaction. Finally, to improve the efficiency of PEM regeneration the
backﬂushing can be optimized by a purposeful choice of the backﬂushing solution
and by using auxiliary techniques (e.g., sonication) to remove the fouled PEM ﬁlm

before or after backﬂushing.

The NOM-TiOz interaction study (Chapter 3) can be taken further by applying

XDLVO analysis to the interactions of TiOz nanoparticles with 1) various NOM size
fractions and 2) under different water chemistries. To understand the contribution of

non-XDLVO forces (steric forces and Ca2+ bridging effect) that are hypothesized to

dominate TiOz-NOM interaction in the presence of calcium, modeling at the

molecular scale is needed.

The design of the membrane-based crossﬂow ﬁltration hydrocyclonic separation
process can be greatly facilitated by using computational ﬂuid dynamics as a
predicting and diagnostic tool. Such simulation may be helpful to optimize the
geometry of the system and identify the optimal combination of the many
experimental variables (crossﬂow velocity, rotational velocity performance, air inﬂow
rate, bubble size, etc). The dual membrane conﬁguration has not been tested in the
crossﬂow ﬁltration hydrocyclone framework. It remains to be seen if the rotational
ﬂow can help in alleviating the fouling of the hydrophilic membrane by oil and
enhancing the oil recovery by hydrophobic membrane. Perhaps most importantly,
designing a fouling resistant superhydrophobic surface with nanotechnology can be

what will ensure the commercial viability of the dual membrane system. The CFFH

106

 

system should be tested with oil-water dispersions that contain additional components

(i.e., salts, colloids) as well as the real produced water.

107

References

[1] Kaluza, S.; Balderhaar, J. K.; Orthen, B.; Honnert, B.; Jankowska, E.; Pietrowski,
P.; Rosell, M. G; Tanarro, C.; Tejedor, J .; Zugasti, A. "Literature review - workplace
exposure to nanoparticles," 2009.

[2] "Nanoscience and nanotechnologies: opportunities and uncertainties," The Royal
Society and the Royal Academy of Engineering, 2004.

[3] Feynman, R. P., "There's plenty of room at the bottom", In American Physical
Society, 1956.

[4] ISO/DTS, "Nanotechnologies. Terminology and deﬁnitions for nanoparticles",
2008.

[5] Organization for Economic Co-operation and Development, "Nanotech is not
small", 2008. -

[6] Lauterwasser, C. "Small sizes that matter: opportunities and risks of
nanotechnologies. report in co-operation with the OECD International Futures
Programme," 2008.

[7] Hannah, W.; Thompson, P. B. Nanotechnology, risk and the environment: a review.
J. Environ. Monit. 2008, 10, 291-300.

[8] PUBMED, hEp://www.ncbi.nlm.nih.gov/sites/entrez

[9] Helmut Kaiser consultancy, "Silent growth of nanotechnology application in every
industry worldwide", 2007.

[10]Pummakamchana, O.; Tripatlri, N.; Dutta, J. Air pollution monitoring and GIS
modeling: a new use of nanotechnology based solid state gas sensors. Sci. Technol.
Adv. Mater. 2005, 6, 251-255.

[11]Theron, J.; Walker, J. A.; Cloete, T. E. Nanotechnology and water treatment:
Applications and emerging opportunities. Crit. Rev. Microbiol. 2008, 34, 43-69.

108

 

[12]Fang, J.; Shan, X. Q.; Wen, B.; Lin, J. M.; Owens, G Stability of titania
nanoparticles in soil suspensions and transport in saturated homogeneous soil columns.
Environ. Pollut. 2009, I5 7, 1101-1109.

[13]Wang, Y. G; Li, Y. 8.; Former, J. D.; Hughes, J. B.; Abriola, L. M.; Pennell, K. D.
Transport and retention of nanoscale C-60 aggregates in water-saturated porous media.
Environ. Sci. Technol. 2008, 42, 3588-3594.

[14]Darlington, T. K.; Neigh, A. M.; Spencer, M. T.; Nguyen, O. T.; Oldenburg, S. J.
Nanoparticle characteristics affecting environmental fate and transport through soil.
Environ. Toxicol. Chem. 2009, 28, 1191-1199.

[15]Narr, J.; Viraraghavan, T.; Jin, Y. C. Applications of nanotechnology in
water/wastewater treatment: A review. F resenius Environ. Bull. 2007, 16, 320-329.

[16]Kanr, B.; Kuiken, T.; Otto, M. Nanotechnology and in situ remediation: A review
of the beneﬁts and potential risks. Environ. Health Perspect. 2009, II 7, 1823-1831.

[17]Esterkin, C. R.; Negro, A. C.; Alfano, O. M.; Cassano, A. E. Air pollution
remediation in a ﬁxed bed photocatalytic reactor coated with TiO2. Aiche J. 2005, 51,
2298-2310.

[18]Mayo, J. T.; Yavuz, C.; Yean, S.; Cong, L.; Shipley, H.; Yu, W.; Falkner, J.; Kan,
A.; Tomson, M.; Colvin, V. L. The effect of nanocrystalline magnetite size on arsenic
removal. Sci. Technol. Adv. Mater: 2007, 8, 71-75.

[19]UNICEF/WHO "Progress on drinking water and sanitation: Special focus on
sanitation," 2008.

[20]Diallo, M.; Duncan, J.; Savage, N.; Street, A.; Sustich, R. Nanotechnology
applications for clean water William Andrew Inc.: Norwich, NY, 2009.

[21]Crittenden, J. C.; Trussell, R. R.; Hand, D. W.; Howe, K. J.; Tchobanoglous, G
Water treatment: Principles and design; John Wiley & Sons 2005.

[22]Miege, C.; Choubert, J. M.; Ribeiro, L.; Eusebe, M.; Coquery, M. Fate of
pharmaceuticals and personal care products in wastewater treatment plants -
Conception of a database and ﬁrst results. Environ. Pollut. 2009, 15 7, 1721-1726.

[23]Bolong, N.; Ismail, A. F.; Salim, M. R.; Matsuura, T. A review of the effects of
emerging contaminants in wastewater and options for their removal. Desalination
2009, 239, 229-246.

109

[24]Petrovic, M.; Gonzalez, S.; Barcelo, D. Analysis and removal of emerging
contaminants in wastewater and drinking water. Trac-Trends Anal. Chem. 2003, 22,
685-696.

[25]Savage, N.; Diallo, M. S. Nanomaterials and water puriﬁcation: Opportunities
and challenges. J. Nanopart. Res. 2005, 7, 331-342.

[26]Nangmenyi, G; Xao, W.; Mehrabi, S.; Mintz, E.; Economy, J. Bactericidal
activity of Ag nanoparticle-impregnated ﬁbreglass for water disinfection. .1. Water
Health 2009, 7, 657-663.

[27] Choi, H.; Stathatos, E.; Dionysiou, D. D. Photocatalytic TiOZ ﬁlms and
membranes for the development of efficient wastewater treatment and reuse systems.
Desalination 2007, 202, 199-206.

[28]Vickaryous, W. J.; Herges, R.; Johnson, D. W. Arsenic-pi interactions stabilize a
self-assembled A32L3 supramolecular complex. Angew. Chem-Int. Edit. 2004, 43,
5831-5833.

[29]Diallo, M. S.; Christie, S.; Swaminathan, P.; Johnson, J. H.; Goddard, W. A.
Dendrimer enhanced ultraﬁltration. 1. Recovery of Cu(II) from aqueous solutions
using PAMAM dendrimers with ethylene diamine core and terminal NH2 groups.
Environ. Sci. Technol. 2005, 39, 1366-1377.

[30]Lin, Y. H.; Cui, X. L. Novel hybrid materials with high stability for electrically
switched ion exchange: carbon nanotube-polyaniline-nickel hexacyanoferrate
nanocomposites. Chem. Commun. 2005, 2226-2228.

[31]Nair, A. S.; Pradeep, T. Halocarbon mineralization and catalytic destruction by
metal nanoparticles. Curr. Sci. 2003, 84, 1560-1564.

[32] Phenrat, T.; Liu, Y. Q.; Tilton, R. D.; Lowry, G V. Adsorbed polyelectrolyte
coatings decrease Fe-O nanoparticle reactivity with TCE in water: Conceptual model
and mechanisms. Environ. Sci. Technol. 2009, 43, 1507-1514.

[33]Breggin, L. K.; Pendergrass, J. "Where does the nano go? End-of—life regulation
of nanotechnology," 2007.

[34]Berger, M. "Nanotechnology risks - the real issues", In Nanowerk, 2007.

[35]Lyon, D. Y.; Adams, L. K.; Falkner, J. C.; Alvarez, P. J. J. Antibacterial activity of
fullerene water suspensions: Effects of preparation method and particle size. Environ.

110

Sci. Technol. 2006, 40, 4360-4366.

[36]Mashino, T.; Okuda, K.; Hirota, T.; Hirobe, M.; Nagano, T.; Mochizuki, M.
Inhibition of E-coli growth by fullerene derivatives and inhibition mechanism. Bioorg.
Med. Chem. Lett. 1999, 9, 2959-2962.

[37] Ferry, J. L.; Craig, R; Hexel, C.; Sisco, P.; Frey, R.; Pennington, P. L.; Fulton, M.
H.; Scott, I. G; Decho, A. W.; Kashiwada, S.; Murphy, C. J .; Shaw, T. J. Transfer of
gold nanoparticles from the water column to the estuarine food web. Nat.
Nanotechnol. 2009, 4, 441-444.

[38] Senjen, R. "Nanosilver - a threat to soil, water and human health?", In Friends of
the Earth, 2007.

[39]Jeong, B. H.; Hock, E. M. V.; Yan, Y. S.; Subramani, A.; Huang, X. F.; Hurwitz,
G; Ghosh, A. K.; Jawor, A. Interfacial polymerization of thin ﬁlm nanocomposites: A
new concept for reverse osmosis membranes. J. Membr: Sci. 2007, 294, 1-7.

[40] Majumder, M.; Chopra, N.; Andrews, R.; Hinds, B. J. Nanoscale hydrodynamics -
Enhanced ﬂow in carbon nanotubes. Nature 2005, 438, 44-44.

[41]Xu, J.; Bhattacharyya, D. Fe/Pd nanoparticle immobilization in microﬁltration
membrane pores: Synthesis, characterization, and application in the dechlon'nation of

polychlorinated biphenyls. Ind. Eng. Chem. Res. 2007, 46, 2348-2359.

[42]Taurozzi, J. S.; Arul, H.; Bosak, V. Z.; Burban, A. F.; Voice, T. C.; Bruening, M.
L.; Tarabara, V. V. Effect of ﬁller incorporation route on the properties of
polysulfone-silver nanocomposite membranes of different porosities. J. Membr. Sci.
2008, 325, 58-68.

[43]Kwak, S. Y.; Kim, S. H.; Kim, S. S. Hybrid organic/inorganic reverse osmosis
(RO) membrane for bactericidal anti-fouling. 1. Preparation and characterization of
Ti02 nanoparticle self-assembled aromatic polyamide thin-ﬁlm-composite (TFC)
membrane. Environ. Sci. Technol. 2001, 35, 2388-2394.

[44]Luo, M. L.; Zhao, J. Q.; Tang, W.; Pu, C. S. Hydrophilic modiﬁcation of
poly(ether sulfone) ultraﬁltration membrane surface by self-assembly of Ti02
nanoparticles. Appl. Surf Sci. 2005, 249, 76-84.

[45]Bae, T. H.; Tak, T. M. Effect of Ti02 nanoparticles on fouling mitigation of
ultraﬁltration membranes for activated sludge ﬁltration. J. Membr. Sci. 2005, 249, 1-8.

111

 

 

 

[46] Srivastava, A.; Srivastava, O. N.; Talapatra, S.; Vajtai, R.; Ajayan, P. M. Carbon
nanotube ﬁlters. Nat. Mater. 2004, 3, 610-614.

[47]Jiang, B. B.; Barnett, J. B.; Li, B. Y. Advances in polyelectrolyte multilayer
nanoﬁlms as tunable drug delivery systems. Nanotechnology, Science and Application
2009, 2, 21-27.

[48]Jiang, B. B.; Li, B. Y. Tunable drug loading and release from polypeptide
multilayer nanoﬁlms. Int. J. Nanomed. 2009, 4, 37-53.

[49]Bruening, M. L.; Dotzauer, D. M.; Jain, P.; Ouyang, L.; Baker, (1 L. Creation of
functional membranes using polyelectrolyte multilayers and polymer brushes.
Langmuir 2008, 24, 7663-7673.

[50]Roudman, A. R.; DiGiano, F. A. Surface energy of experimental and commercial
nanoﬁltration membranes: Effects of wetting and natural organic matter fouling. J.
Membt: Sci. 2000, 175, 61-73.

[51]Kull, K R.; Steen, M. L.; Fisher, E. R. Surface modiﬁcation with
nitrogen-containing plasmas to produce hydrophilic, low-fouling membranes. J.
Membr. Sci. 2005, 246, 203-215.

[52]Zhu, X. H.; Elimelech, M. Colloidal fouling of reverse osmosis membranes:
Measurements and fouling mechanisms. Environ. Sci. Technol. 1997, 31, 3654-3662.

[53]Vrijenhoek, E. M.; Hong, S.; Elimelech, M. Inﬂuence of membrane surface
properties on initial rate of colloidal fouling of reverse osmosis and nanoﬁltration
membranes. J. Membr Sci. 2001, 188, 115-128.

[54]Tang, C. Y.; Fu, Q. S.; Criddle, C. S.; Leckie, J. 0. Effect of ﬂux (transmembrane
pressure) and membrane properties on fouling and rejection of reverse osmosis and
nanoﬁltration membranes treating perﬂuorooctane sulfonate containing wastewater.
Environ. Sci. Technol. 2007, 41, 2008-2014.

[55]Wang, J. W.; Yue, Z. R.; Economy, J. Novel method to make a continuous
micro-mesopore membrane with tailored surface chemistry for use in nanoﬁltration. J.
Membr. Sci. 2008, 308, 191-197.

[56] Ba, C. Y.; Langer, J.; Economy, J. Chemical modiﬁcation of P84 copolyimide
membranes by polyethylenimine for nanoﬁltration. J. Membr Sci. 2009, 32 7, 49-58.

[57] Muller, M.; Rieser, T.; Lunkwitz, K.; Meier-Haack, J. Polyelectrolyte complex

112

layers: a promising concept for antifouling coatings veriﬁed by in-situ ATR-FTIR
spectroscopy. Macromol. Rapid Comm. 1999, 20, 607—611.

[58]Meier—Haack, J.; Rieser, T.; Lenk, W.; Lehmann, D.; Berwald, S.; Schwarz, S.
Effect of polyelectrolyte complex layers on the separation properties and the fouling
behavior of surface and bulk modiﬁed membranes. Chem. Eng. Technol. 2000, 23,
114-118.

[59] Rieser, T.; Lunkwitz, K.; Berwald, S.; Meier-Haack, J.; Muller, M.; Cassel, F.;
Dioszeghy, 2.; Simon, F. Surface modiﬁcation of microporous polypropylene
membranes by polyelectrolyte multilayers. In Membrane formation and modification
(ACS Sym. Sen), 2000; Vol. 744, pp 189-204.

[60] Muller, M.; Rieser, T.; Dubin, P. L.; Lunkwitz, K. Selective interaction between
proteins and the outermost surface of polyelectrolyte multilayers: Inﬂuence of the
polyanion type, pH and salt. Macromol. Rapid Comm. 2001, 22, 390-395.

[61]Meier—Haack, J .; Muller, M. Use of polyelectrolyte multilayer systems for
membrane modiﬁcation. Macromol. Symp. 2002, 188, 91-103.

[62]Mendelsohn, J. D.; Yang, S. Y.; Hiller, J.; Hochbaum, A. I.; Rubner, M. F.
Rational design of cytophilic and cytophobic polyelectrolyte multilayer thin ﬁlms.
Biomacromolecules 2003, 4, 96-106.

[63]Yang, S. Y.; Mendelsohn, J. D.; Rubner, M. F. New class of ultrathin, highly
cell-adhesion-resistant polyelectrolyte multilayers with micropatterning capabilities.
Biomacromolecules 2003, 4, 987-994.

[64]Boulmedais, F.; Frisch, B.; Etienne, 0.; Lavalle, P.; Picart, C.; Ogier, J .; Voegel, J.
C.; Schaaf, P.; Egles, C. Polyelectrolyte multilayer ﬁlms with pegylated polypeptides
as a new type of anti-microbial protection for biomaterials. Biomaterials 2004, 25,
2003-2011.

[65] Muller, M.; Meier-Haack, J.; Schwarz, S.; Buchhammer, H. M.; Eichhom, E. J .;
Janke, A.; Kessler, B.; Nagel, J.; Oehnann, M.; Reihs, T.; Lunkwitz, K.
Polyelectrolyte multilayers and their interactions. J. Adhesion 2004, 80, 521-547.

[66] Salloum, D. S.; Schlenoff, J. B. Protein adsorption modalities on polyelectrolyte
multilayers. Biomacromolecules 2004, 5, 1089-1096.

[67] Elbert, D. L.; Herbert, C. B.; Hubbell, J. A. Thin polymer layers formed by
polyelectrolyte multilayer techniques on biological surfaces. Langmuir 1999, 15,
5355-5362.

113

 

 

[68]Richert, L.; Lavalle, P.; Payan, E.; Shu, X. Z.; Prestwich, G D.; Stoltz, J. F.;
Schaaf, P.; Voegel, J. C.; Picart, C. Layer by layer buildup of polysaccharide ﬁlms:
Physical chemistry and cellular adhesion aspects. Langmuir 2004, 20, 448—458.

[69]Thompson, M. T.; Berg, M. C.; Tobias, 1. S.; Rubner, M. F.; Van Vliet, K. J.
Tuning compliance of nanoscale polyelectrolyte multilayers to modulate cell adhesion.
Biomaterials 2005, 26, 6836-6845.

[70] Schneider, A.; Francius, G; Obeid, R.; Schwinte, P.; Hemmerle, J.; Frisch, B.;
Schaaf, P.; Voegel, J. C.; Senger, B.; Picart, C. Polyelectrolyte multilayers with a
tunable Young's modulus: Inﬂuence of ﬁlm stiffness on cell adhesion. Langmuir 2006,
22, 1193-1200.

[71]Lu, Y. X.; Sun, J. Q.; Shen, J. C. Cell adhesion properties of patterned
poly(acrylic acid)/poly(allylamine hydrochloride) multilayer ﬁlms created by
room-temperature imprinting technique. Langmuir 2008, 24, 8050-8055.

[72]Fu, J. H.; Ji, J.; Yuan, W. Y.; Shen, J. C. Construction of anti-adhesive and
antibacterial multilayer ﬁlms via layer-by-layer assembly of heparin and chitosan.
Biomaterials 2005, 26, 6684-6692.

[73]Lichter, J. A.; Thompson, M. T.; Delgadillo, M.; Nishikawa, T.; Rubner, M. F.;
Van Vliet, K. J. Substrata mechanical stiffness can regulate adhesion of viable bacteria.
Biomacromolecules 2008, 9, 1571-1578.

[74]Mhamdi, L.; Picart, C.; Lagneau, C.; Othmane, A.; Grosgogeat, B.;
J affrezic-Renault, N.; Ponsonnet, L. Study of the polyelectrolyte multilayer thin ﬁlms'
properties and correlation with the behavior of the human gingival ﬁbroblasts. Mat.
Sci. Eng. C-Bio. S. 2006, 26, 273-281.

[75]Kotov, N. A. Layer-by-layer self-assembly: The contribution of hydrophobic
interactions. Nanostruct. Mater: 1999, 12, 789-796.

[76] Serizawa, T.; Takeshita, H.; Akashi, M. Electrostatic adsorption of polystyrene
nanospheres onto the surface of an ultrathin polymer ﬁlm prepared by using an
alternate adsorption technique. Langmuir 1998, 14, 4088-4094.

[77] Serizawa, T.; Kamimura, S.; Akashi, M. Electrostatic adsorption of polystyrene
particles with different surface charges onto the surface of an ultrathin polymer ﬁlm.
Colloid Surf A-Physicochem. Eng. Asp. 2000, 164, 237-245.

[78]Ahn, J. S.; Hammond, P. T.; Rubner, M. F.; Lee, I. Self-assembled particle
monolayers on polyelectrolyte multilayers: particle size effects on formation, structure,

114

and optical properties. Colloid Surf A-Physicochem. Eng. Asp. 2005, 259, 45-53.

[79]Uto, K.; Yamamoto, K.; Kishimoto, N.; Muraoka, M.; Aoyagi, T.; Yamashita, I.
Electrostatic adsorption of ferritin, proteins and nanoparticle conjugate onto the
surface of polyelectrolyte multilayers. J. Mater. Chem. 2008, 18, 3876-3884.

[80] Hattori, H. Anti-reﬂection surface with particle coating deposited by electrostatic
attraction. Adv. Mater: 2001, 13, 51-54.

[81]Michna, A.; Adamczyk, Z.; Zembala, M. Deposition of colloid particles on
polyelectrolyte multilayers. Colloid Surf A-Physicochem. Eng. Asp. 2007, 302,
467-472.

[82]Adamczyk, Z.; Michna, A.; Szaraniec, M.; Bratek, A.; Barbasz, J.
Characterization of poly(ethylene imine) layers on mica by the streaming potential
and particle deposition methods. J. Colloid Interface Sci. 2007, 313, 86-96.

[83]Adamczyk, Z.; Zembala, A.; Michna, A. Polyelectrolyte adsorption layers studied
by streaming potential and particle deposition. J. Colloid Interface Sci. 2006, 303,
353-364.

[84]Chen, K. M.; Jiang, X. P.; Kimerling, L. C.; Hammond, P. T. Selective
self-organization of colloids on patterned polyelectrolyte templates. Langmuir 2000,
16, 7825-7 834.

[85]Hattori, H. Two-step assembly technique for preparation of polymer-particle
composite ﬁlms. Thin Solid Films 2001, 385, 302-306.

[86]Adamczyk, Z.; Jaszczolt, K.; Michna, A.; Siwek, B.; Szyk-Warszynska, L.;
Zembala, M. Irreversible adsorption of particles on heterogeneous surfaces. Adv.
Colloid Interface Sci. 2005, 118, 25-42.

[87]von Klitzing, R.; Tieke, B. Polyelectrolyte membranes. In Polyelectrolytes with
Deﬁned Molecular Architecture 1; Springer-Verlag Berlin: Berlin, 2004; Vol. 165, pp
177-210.

[88]Stanton, B. W.; Harris, J. J.; Miller, M. D.; Bruening, M. L. Ultrathin,
multilayered polyelectrolyte ﬁlms as nanoﬁtlration membranes. Langmuir 2003, 19,
7038-7042.

[89] Krasemann, L.; Tieke, B. Composite membranes with ultrathin separation layer
prepared by self-assembly of polyelectrolytes. Mat. Sci. Eng. C-Biomimetic Supramol.

115

Syst. 1999, 8-9, 513—518.

[90]Harris, J. J.; Stair, J. L.; Bruening, M. L. Layered polyelectrolyte ﬁlms as
selective, ultrathin barriers for anion transport. Chem. Mater: 2000, 12, 1941-1946.

[91]Krasemann, L.; Tieke, B. Selective ion transport across self-assembled alternating
multilayers of cationic and anionic polyelectrolytes. Langmuir 2000, 16, 287-290.

[92]Toutianoush, A.; Tieke, B. Selective transport and incorporation of highly charged
metal and metal complex ions in self-assembled polyelectrolyte multilayer
membranes. Mat. Sci. Eng. C-Biomimetic Supramol. Syst. 2002, 22, 135-139.

[93]Jin, W.; Toutianoush, A.; Tieke, B. Use of polyelectrolyte layer-by-layer
assemblies as nanoﬁltration and reverse osmosis membranes. Langmuir 2003, 19,
2550-2553.

[94] Malaisamy, R.; Bruening, M. L. High-ﬂux nanoﬁltration membranes prepared by
adsorption of multilayer polyelectrolyte membranes on polymeric supports. Langmuir
2005, 21, 10587-10592.

[95]El-Hashani, A.; Toutianoush, A.; Tieke, B. Layer-by-layer assembled membranes
of protonated 18-azacrown-6 and polyvinylsulfate and their application for highly
efficient anion separation. J. Phys. Chem. B 2007, 111, 8582-8588.

[96] Liu, X.; Bruening, M. L. Size-selective transport of uncharged solutes through
multilayer polyelectrolyte membranes. Chem. Mater: 2004, 16, 351-357.

[97]Miller, M. D.; Bruening, M. L. Controlling the nanoﬁltration properties of
multilayer polyelectrolyte membranes through variation of ﬁlm composition.
Langmuir 2004, 20, 11545-11551.

[98]Jin, W.; Toutianoush, A.; Tieke, B. Size- and charge-selective transport of
aromatic compounds across polyelectrolyte multilayer membranes. Appl. Surf Sci.
2005, 246, 444-450.

[99] Hong, S. U.; Bruening, M. L. Separation of amino acid mixtures using multilayer
polyelectrolyte nanoﬁltration membranes. J. Membr: Sci. 2006, 280, 1-5.

[100] Hollman, A. M.; Bhattacharyya, D. Pore assembled multilayers of charged
polypeptides in microporous membranes for ion separation. Langmuir 2004, 20,
5418-5424.

116

[101] Sukhorukov, G B.; Antipov, A. A.; Voigt, A.; Donath, E.; Mohwald, H.
pH-controlled macromolecule encapsulation in and release from polyelectrolyte
multilayer nanocapsules. Macromol. Rapid Commun. 2001, 22, 44-46.

[102] Farre, M.; Gajda-Schrantz, K.; Kantiani, L.; Barcelo, D. Ecotoxicity and
analysis of nanomaterials in the aquatic environment. Anal. Bioanal. Chem. 2009, 393,
81-95.

[103] Wiesner, M. R.; Lowry, G V.; Alvarez, P.; Dionysiou, D.; Biswas, P. Assessing
the risks of manufactured nanomaterials. Environ. Sci. Technol. 2006, 40, 4336-4345.

[104] Wiesner, M.; Bottero, J. Y. Environmental Nanotechnology: Applications and
Impacts of Nanomaterials; McGraw-Hill: New York, 2007.

[105] Smoluchowski, M. Versuch einer mathematischen Theorie de
Koagulationkinetic Kolloider Lo"sungen. Z. Phys. Chem. 1917, 92, 129-168.

[106] Derjaguin, B. V.; Landau, L. D. Theory of the stability of strongly charged
lyophobic sols and of the adhesion of strongly charged particles in solutions of
electrolytes. Acta Phys-chim. URSS 1941, 14, 633-662.

[107] Verwey, E. J. W.; Overbeek, J. T. G Theory of the stability of lyophobic colloids:
the interaction of particles having an electric double layer; Elsevier: Amsterdam,
1948.

[108] Kallay, N.; Zalac, S. Stability of nanodispersions: A model for kinetics of
aggregation of nanoparticles. J. Colloid Interface Sci. 2002, 253, 70-76.

[109] Navarro, E.; Baun, A.; Behra, R.; Hartrnann, N. B.; Filser, J .; Miao, A. J.; Quigg,
A.; Santschi, P. H.; Sigg, L. Environmental behavior and ecotoxicity of engineered
nanoparticles to algae, plants, and fungi. Ecotoxicology 2008, I 7, 372-386.

[110] Brant, J.; Lecoanet, H.; Wiesner, M. R. Aggregation and deposition
characteristics of fullerene nanoparticles in aqueous systems. Journal OfNanoparticle
Research 2005, 7, 545-553.

[111] Van Hyning, D. L.; Klemperer, W. G; Zukoski, C. F. Characterization of
colloidal stability during precipitation reactions. Langmuir 2001, I 7, 3120-3127.

[112] Guzman, K. A. D.; Finnegan, M. P.; Banﬁeld, J. F. Inﬂuence of surface potential
on aggregation and transport of titania nanoparticles. Environ. Sci. Technol. 2006, 40,
7688-7693.

117

[113] Lecoanet, H. F.; Bottero, J. Y.; Wiesner, M. R. Laboratory assessment of the
mobility of nanomaterials in porous media. Environ. Sci. Technol. 2004, 38,
5164-5169.

[114] Asharani, P. V.; Wu, Y. L.; Gong, Z. Y.; Valiyaveettil, S. Toxicity of silver
nanoparticles in zebraﬁsh models. Nanotechnology 2008, 19.

[115] Yeo, M. K.; Kang, M. Effects of nanometer sized silver materials on biological
toxicity during zebraﬁsh embryogenesis. Bull. Korean Chem. Soc. 2008, 29,
1179-1 184.

[116] Oberdorster, E. Manufactured nanomaterials (Fullerenes, C60) induce oxidative

stress in the brain of juvenile largemouth bass Environ. Health Perspect. 2004, 112,
1058-1062. ‘

[117] Isaacson, C. W.; Usenko, C. Y.; Tanguay, R. L.; Field, J. A. Quantiﬁcation of
fullerenes by LC/ESI-MS and its application to in vivo toxicity assays. Anal. Chem.
2007, 79, 9091-9097.

[118] Templeton, R. C.; Ferguson, P. L.; Washbum, K. M.; Scrivens, W. A.; Chandler,
G T. Life-cycle effects of single-walled carbon nanotubes (SWNTs) on an estuarine
meiobenthic copepod. Environ. Sci. Technol. 2006, 40, 7387-7393.

[119] Smith, C. J.; Shaw, B. J.; Handy, R. D. Toxicity of single walled carbon
nanotubes to rainbow trout, (Oncorhynchus mykiss): Respiratory toxicity, organ
pathologies, and other physiological effects. Aquat. Toxicol. 2007, 82, 94-109.

[120] Cheng, J. P.; Flahaut, E.; Cheng, S. H. Effect of carbon nanotubes on
developing zebraﬁsh (Danio rerio) embryos. Environmental Toxicology and
Chemistry 2007, 26, 708-716.

[121] Shan, W. Q.; Bacchin, P.; Aimar, P.; Bruening, M. L.; Tarabara, V. V.
Polyelectrolyte multilayer ﬁhns as backﬂushable nanoﬁltration membranes with
tunable hydrophilicity and surface charge. J. Membr. Sci. 2010, 349, 268-278.

[122] Decher, G; Hong, J. D. Buildup of ultrathin multilayer ﬁlms by a self-assembly
process: 2. Consecutive adsorption of anionic and cationic bipolar amphiphiles and
polyelectrolytes on charged surfaces. Ber. Bunsen. Phys. Chem. 1991, 95, 1430-1434.

[123] Decher, G; Hong, J. D.; Schmitt, J. Buildup of ultrathin multilayer ﬁlms by a
self-assembly process: 3. Consecutively alternating adsorption of anionic and cationic
polyelectrolytes on charged surfaces. Thin Solid Films 1992, 210, 831-835.

118

[124] Decher, G Fuzzy nanoassemblies: toward layered polymeric multicomposites.
Science 1997, 277, 1232-1237.

[125] Bertrand, P.; Jonas, A.; Laschewsky, A.; Legras, R. Ultrathin polymer coatings
by complexation of polyelectrolytes at interfaces: suitable materials, structure and
properties. Macromol. Rapid Comm. 2000, 21, 319-348.

[126] Schénhoff, M. Layered polyelectrolyte complexes: physics of formation and
molecular properties. J. Phys.-Condens. Mat. 2003, 15, 1781-1808.

[127] Ouyang, L.; Malaisamy, R.; Bruening, M. L. Multilayer polyelectrolyte ﬁlms as
nanoﬁltration membranes for separating monovalent and divalent cations. J. Membr:
Sci. 2008, 310, 76-84.

[128] Hong, S. U.; Malaisamy, R.; Bruening, M. L. Separation of ﬂuoride from other
monovalent anions using multilayer polyelectrolyte nanoﬁltration membranes.
Langmuir 2007, 23, 1716-1722.

[129] Yoo, D.; Shiratori, S. S.; Rubner, M. F. Controlling bilayer composition and
surface wettability of sequentially adsorbed multilayers of weak polyelectrolytes.
Macromolecules 1998, 31 , 4309-4318.

[130] Dubas, S. T.; Schlenoff, J. B. Factors controlling the growth of polyelectrolyte
multilayers. Macromolecules 1999, 32, 8153—8160.

[131] Dubas, S. T.; Schlenoff, J. B. Polyelectrolyte multilayers containing a weak
polyacid: construction and deconstruction. Macromolecules 2001, 34, 3736-3740.

[132] Wang, F.; Tarabara, V. V. Coupled effects of colloidal deposition and salt
concentration polarization on reverse osmosis membrane performance J. Membr: Sci.
2007, 293, 111-123.

[133] Boussu, K.; Belpaire, A.; Volodin, A.; Haesendonck, C. V.; Meeren, P. V. d.;
Vandecasteele, C.; Bruggen, B. V. d. Inﬂuence of membrane and colloidal
characteristics on fouling of nanoﬁltration membranes. J. Membr: Sci. 2007, 289,
220-230.

[134] F ane, A. G Ultraﬁltration of suspensions. J. Membr. Sci. 1984, 20, 249-259.

[135] Wiesner, M. R.; Clark, M. M.; Mallevialle, J. Membrane ﬁltration of coagulated
suspensions. J. Environ. Eng. -ASCE 1989, 115, 20-40.

119

[136] Letterman, R. D. Water quality and treatment : a handbook of community water
supplies; 5th ed.; McGraw-Hill: New York, 1999.

[137] Hoek, E. M. V.; Elimelech, M. Cake-enhanced concentration polarization: a
new fouling mechanism for salt-rejecting membranes. Environ. Sci. Technol. 2003, 3 7,
5581-5588.

[138] Knyazkova, T. V.; Maynarovich, A. A. Recognition of membrane fouling:
testing of theoretical approaches with data on NF of salt solutions containing a low
molecular weight surfactant as a foulant. Desalination 1999, 126, 163-169.

[139] Rautenbach, R.; Albrecht, R. Membrane processes; John Wiley & Sons Ltd:
New York, 1989.

[140] Newman, J. S. Electrochemical systems; Prentice Hall, USA, 1991; Vol. 2nd ed.

[141] Michaels, A. S. New separation technique for CPI. Chem. Eng. Prog. 1968, 64,
31-43.

[142] Kedem, 0.; Katchalsky, A. Thermodynamic analysis of the permeability of
biological membranes to non-electrolytes. Biochim Biophys Acta 1958, 27, 229-246.

[143] Pitzer, K. S. Thermodynamics of electrolytes .1. Theoretical Basis and General
Equations. J. Phys. Chem-US 1973, 77, 268-277.

[144] Phutela, R. C.; Pitzer, K. S. Heat-capacity and other thermodynamic properties
of aqueous magnesium-sulfate to 473-K. J. Phys. Chem. 1986, 90, 895-901.

[145] Spiegler, K. S.; Kedem, 0. Thermodynamics of hyperﬁltration (reverse
osmosis): criteria for efficient membranes. Desalination 1966, I, 311-326.

[146] Wang, X. L.; Tsuru, T.; Togoh, M.; Nakao, S.; Kimura, S. Evaluation of pore
structure and electrical-properties of nanoﬁltration membranes. J. Chem. Eng. Jpn.
1995, 28, 186-192.

[147] Yaroshchuk, A. E. Rejection of single salts versus transmembrane volume ﬂow
in R0/NF: thermodynamic properties, model of constant coefﬁcients, and its
modiﬁcation. J. Membr: Sci. 2002, 198, 285-297.

[148] Al-Zoubi, H.; Hilal, N.; Darwish, N. A.; Mohammad, A. W. Rejection and
modelling of sulphate and potassium salts by nanoﬁltration membranes: neural

120

 

network and Spiegler—Kedem model. Desalination 2007, 206, 42-60.

[149] Kedem, 0.; Freger, V. Determination of concentration-dependent transport
coefﬁcients in nanoﬁltration: Deﬁning an optimal set of coefficients. J. Membr: Sci.
2008, 310, 586-593.

[150] Hoek, E. M. V.; Kim, A. S.; Elimelech, M. Inﬂuence of crossﬂow membrane

ﬁlter geometry and shear rate on colloidal fouling in reverse osmosis and
nanoﬁltration separations. Environ. Eng. Sci. 2002, 19, 357-372.

[151] Ng, H. Y.; Elimelech, M. Inﬂuence of colloidal fouling on rejection of trace
organic contaminants by reverse osmosis. J. Membr. Sci. 2003, 244, 215-226.

[152] Chellam, S.; Zander, A. Membrane science and theory. In Microﬁltration and
ultraﬁltration membranes for drinking water--A WWA manual M53; lst ed. ed.;
American Water Works Association: Denver, 2005.

[153] Miller, M. D.; Bruening, M. L. Correlation of the swelling and permeability of
polyelectrolyte multilayer ﬁlms. Chem. Mater: 2005, 1 7, 5375-5381.

[154] Hanis, J. J.; Bruening, M. L. Electrochemical and in situ ellipsometric
investigation of the permeability and stability of layered polyelctrolyte ﬁlms.
Langmuir 2000, 16, 2006-2013.

[155] Rmaile, H. H.; Schlenoff, J. B. "Internal pK(a)'s" in polyelectrolyte multilayers:
coupling protons and salts. Langmuir 2002, 18, 8263-8265.

[156] Kim, J.; Shan, W.; Davies, S. H. R.; Baumann, M. J.; Masten, S. J.; Tarabara, V.

V. Interactions of aqueous NOM with nanoscale Ti02: Implications for ceramic
membrane ﬁltration-ozonation hybrid process. Environ. Sci. Technol. 2009.

[157] Molinari, R.; Palmisano, L.; Drioli, E.; Schiavello, M. Studies on various
reactor conﬁgurations for coupling photocatalysis and membrane processes in water
puriﬁcation. J. Membr: Sci. 2002, 206, 399-415.

[158] Kim, S. H.; Kwak, S. Y.; Sohn, B. H.; Park, T. H. Design of Ti02 nanoparticle
self-assembled aromatic polyamide thin-ﬁlm-composite (TFC) membrane as an
approach to solve biofouling problem. J. Membr: Sci. 2003, 211, 157—165.

[159] Bae, T. H.; Kim, I. C.; Tak, T. M. Preparation and characterization of
fouling-resistant Ti02 self-assembled nanocomposite membranes. J. Membr: Sci. 2006,

121

275, 1-5.

[160] Karnik, B. S.; Davies, S. H.; Chen, K. C.; Jalowski, D. R.; Baumann, M. J.;
Masten, S. J. Effect of ozonation on the permeate ﬂux of nanocrystalline ceramic
membranes. Water Res. 2005, 39, 728-734.

[161] Kamik, B. S.; Davies, S. H.; Baumann, M. J.; Masten, S. J. Fabrication of
catalytic membranes for the treatment of drinking water using combined ozonation
and ultraﬁltration Environ. Sci. Technol. 2005, 39, 7656-7661.

[162] Kamik, B. S.; Davies, S. H.; Baumann, M. J.; Masten, S. J. The effects of
combined ozonation and ﬁltration on disinfection by-product formation. Water Res.
2005, 39, 2839-2850.

[163] Choi, H.; Stathatos, E.; Dionysiou, D. D. Sol-gel preparation of mesoporous

photocatalytic Ti02 ﬁlms and Ti02/A1203 composite membranes for environmental
applications. Appl. Catal. B-Environ. 2006, 63, 60-67.

[164] Kamik, B. S.; Davies, S. H.; Baumann, M. J.; Masten, S. J. Use of salicylic acid
as a model compound to investigate hydroxyl radical reaction in ozonation-membrane
ﬁltration hybrid process. Environ. Eng. Sci. 2007, 24, 852-860.

[165] Huang, X. H.; Leal, M.; Li, Q. L. Degradation of natural organic matter by
Ti02 photocatalytic oxidation and its effect on fouling of low-pressure membranes.
Water Res. 2008, 42, 1142-1150.

[166] Kim, J.; Davies, S. H.; Baumann, M. J.; Tarabara, V. V.; Masten, S. J. Effect of
ozone dosage and hydrodynamic conditions on the permeate ﬂux in a hybrid
ozonation-ceramic ultrafrltration system treating natural waters. J. Membr: Sci. 2008,
311,165-171.

[167] Lainé, J. M.; Hagstrom, J. P.; Clark, M. M.; Mallevialle, J. Effects of
ultraﬁltration membrane-composition. J. Am. Water Works Ass. 1989, 81, 61-67.

[168] Pu, P.; Ruiz, H.; Thompson, K.; Spangenberg, C. Selecting membranes for
removing NOM and DBP precursors. J. Am. Water Works Ass. 1994, 86, 55-72.

[169] Allgeier, S. C.; Summers, R. S. Evaluating NF for DBP control with the
RBSMT. J. Am. Water Works Ass. 1995, 87, 87-99.

[170] Nilson, J. A.; DiGiano, F. A. Inﬂuence of NOM composition on nanoﬁltration. J.
Am. Water Works Ass. 1996, 88, 53-66.

122

[171] Braghetta, A.; DiGiano, F. A.; Ball, W. P. Nanoﬁltration of natural organic
matter: pH and ionic strength effects. J. Environ. Eng. -ASCE 1997, 123, 628-641.

[172] Zularisam, A. W.; Ismail, A. F.; Salim, M. R.; Sakinah, M.; Ozaki, H. The
effects of natural organic matter (NOM) fractions on fouling characteristics and ﬂux
recovery of ultraﬁltration membranes. Desalination 2007, 212, 191-208.

[173] Hong, S.; Elimelech, M. Chemical and physical aspects of natural organic
matter (NOM) fouling of nanofrltration membranes J. Membr: Sci. 1997, 132,
159-181.

[174] Braghetta, A.; DiGiano, F. A.; Ball, W. P. NOM accumulation at NF membrane
surface: Impact of chemistry and shear. J. Environ. Eng. -ASCE 1998, 124, 1087-1098.

[175] Jucker, 0; Clark, M. M. Adsorption of aquatic humic substances on
hydrophobic ultrafrltration membranes. J. Membr: Sci. 1994, 97, 37-52.

[176] Cho, J.; Amy, G; Pellegrino, J.; Yoon, Y. Characterization of clean and natural
organic matter (N 0M) fouled NF and UP membranes, and foulants characterization.
Desalination 1998, 118, 101-108.

[177] Yuan, W.; Zydney, A. L. Humic acid fouling during ultraﬁltration. Environ. Sci.
Technol. 2000, 34, 5043-5050.

[178] Violleau, D.; Essis-Tome, H.; Habarou, H.; Croué, J. P.; Pontié, M. Fouling
studies of a polyamide nanoﬁltration membrane by selected natural organic matter: an
analytical approach. Desalination 2005, 173, 223-23 8.

[179] Chen, K. L.; Elimelech, M. Inﬂuence of humic acid on the aggregation kinetics
of fullerene (C-60) nanoparticles in monovalent and divalent electrolyte solutions. J.
Colloid Interface Sci. 2007, 309, 126-134.

[180] Hyung, H.; Kim, J.-H. Natural organic matter (NOM) adsorption to
multi-walled carbon nanotubes: effect of NOM characteristics and water quality
parameters. Environ. Sci. Technol. 2008, 42, 4416-4421.

[181] Verwey, E. J. W.; Overbeek, J. T. G Theory of the stability of lyophobic colloids.
The interaction of particles having an electric double layer; Elsevier: Amsterdam,
1948.

[182] Van 085, C. J. Acid-base interfacial interactions in aqueous media. Colloid Surf
A-Physicochem. Eng. Asp. 1993, 78, 1-49.

123

 

[183] Brant, J. A.; Childress, A. E. Assessing short-range membrane-colloid
interactions using surface energetics. J. Membr. Sci. 2002, 203, 257-273.

[184] Kang, S.-T.; Subramani, A.; Hoek, E. M. V.; Deshusses, A.; Matsumoto, M. R.
Direct observation of biofouling in cross-ﬂow microﬁltration: mechanisms of
deposition and release. J. Membr: Sci. 2004, 244, 151-165.

[185] Kim, S.; Marion, M.; Jeong, B.-H.; Hoek, E. M. V. Crossﬂow membrane
ﬁltration of interacting nanoparticle suspensions. J. Membr: Sci. 2006, 284, 361-372.

[186] Van 038, C. J. Interfacial forces in aqueous media; 2nd ed.; Boca Raton, Fla. :
Taylor & Francis, 2006.

[187] Hoek, E. M. V.; Agarwal, G K. Extended DLVO interactions between spherical
particles and rough surfaces. J. Colloid Interface Sci. 2006, 298, 50-58.

[188] Mellema, J. A. In Department of civil and environmental engineering; Michigan
State University: East Lansing, 1998; Vol. Master of science.

[189] Bose, P.; Bezbarua, B. K.; Reckhow, D. A. Effect of ozonation on some physical
and chemical-properties of aquatic natural organic-matter. Ozone-Sci. Eng. 1994, 16,
89-1 12.

[190] Moon, J.; Kim, S. H.; Cho, J. Characterizations of natural organic matter as
nano particle using ﬂow ﬁeld-ﬂow fractionation. Colloid Surface A 2006, 287,
232—236.

[191] Langlais, B.; Reckhow, D. A.; Brink, D. R. Ozone in water treatment .'
Application and engineering : Cooperative research report; Lewis Publishers:
Chelsea, Mich, 1991.

[192] Silalahi, S. H. D.; Leiknes, T. Cleaning strategies in ceramic microﬁltration
membranes fouled by oil and particulate matter in produced water. Desalination 2009,
236, 160-169.

[193] Hargreaves, J. H.; Silvester, R. S. Computational ﬂuid dynamics applied to the
analysis of deoiling hydrocyclone performance. Chem. Eng. Res. Des. 1990, 68,
365-383.

[194] Rawn-Schatzinger, V.; Arthur, D.; Langhus, B. Coalbed natural gas resources:
water rights and treatment technologies. GasTIPS 2003, 9, 13-18.

124

 

[195] Rawn-Schatzinger, V.; Arthur, D.; Langhus, B. Coalbed natural gas resources:
beneﬁcial use alternatives. GasTIPS 2004, 10, 9-14.

[196] Clark, C. E.; Veil, J. A. "Produced water volumes and management practices in
the United States," 2009.

[197] Holdich, R. G; Cumming, I. W.; Smith, I. D. Crossﬂow microﬁltration of oil in
water dispersions using surface ﬁltration with imposed ﬂuid rotation. J. Membr: Sci.
1998, 143, 263-274.

[198] Wake, H. Oil reﬁneries: a review of their ecological impacts on the aquatic
environment. Estuar: Coast. Shelf S. 2005, 62, 131-140.

[199] International Maritrne Organization; The Paris Convention 1974 and The Oslo
Convention 1972 (OSPAR), Albert Embankment, London, UK, 1974; Vol. 4.

[200] Bevis, A. The treatment of oily water by coalescing Filtr. Sep. 1992, 29,
295-298.

[201] OSPAR- Oslo/Paris convention (for the Protection of the Marine Environment
of the North-East Atlantic). 2001.

[202] Oil in the sea 11]; National Academies Press: Washington, DC, 2003.

[203] Rhee, C. H.; Martyn, P. C.; Kremer, J. G In Proceedings of the 42nd Purdue
Industrial Waste Conference: Purdue University, 1987; p 143.

[204] Mohammadi, T.; Kazemimoghadam, M.; Saadabadi, M. Modeling of membrane
fouling and ﬂux decline in reverse osmosis during separation of oil in water
emulsions. Desalination 2003, I 5 7, 369-375.

[205] Cambiella, A.; Benito, J. M.; Pazos, 0; Coca, J. Centrifugal separation
efficiency in the treatment of waste emulsiﬁed oils. Chem. Eng. Res. Des. 2006, 84,
69-76.

[206] Al-Shamrani, A. A.; James, A.; Xiao, H. Separation of oil from water by
dissolved air ﬂotation. Colloid Surf A 2002, 209, 15-26.

[207] Bensadok, K.; Belkacem, M.; Nezzal, G Treatment of cutting oil/water
emulsion by coupling coagulation and dissolved air ﬂotation. Desalination 2007, 206,
440-448.

125

[208] Hashmi, K. A.; Hamza, H. A.; Wilson, J. C. CANMET hydrocyclone: an
emerging alternative for the treatment of oily waste streams. Miner. Eng. 2004, I 7,
643-649.

[209] Frankiewicz, T. In The Produced Water Seminar, 2001.

[210] Cumming, I. W.; Holdich, R. G; Smith, I. D. The rejection of oil by
microﬁltration of a stabilised kerosene/water emulsion. J. Membr: Sci. 2000, 169,
147-155.

[211] Frankiewicz, T. In 11Lh produced water seminar: Houston, 2001.

[212] Mondal, S.; Wickramasinghe, S. R. Produced water treatment by nanoﬁltration
and reverse osmosis membranes. J. Membr: Sci. 2008, 322, 162-170.

[213] Deng, S. 2.; Sourirajan, S.; Chan, K.; Famand, B.; Okada, T.; Matsuura, T.
Dehydration of oil-water emulsion by pervaporation using porous hydrophilic
membranes. J. Colloid Interface Sci. 1991, 141, 218-225.

[214] Gryta, M.; Karakulski, K. The application of membrane distillation for the
concentration of oil-water emulsions. Desalination 1999, 121, 23-29.

[215] Benito, J. M.; Rios, G; Ortea, E.; Fernandez, E.; Cambiella, A.; Pazos, 0; Coca,
J. Design and construction of a modular pilot plant for the treatment of oil-containing
wastewaters. Desalination 2002, 14 7, 5-10.

[216] Zhong, J .; Sun, X. J .; Wang, C. L. Treatment of oily wastewater produced ﬁ'om
reﬁnery processes using ﬂocculation and ceramic membrane ﬁltration. Sep. Purif.
Technol. 2003, 32, 93-98.

[217] Bodzek, M.; Konieczny, K. The use of ultraﬁltration membranes made of
various polymers in the treatment of oil-emulsion wastewater. Waste Manage. 1992,
12, 75-84.

[218] Ohya, H.; Kim, J. J .; Chinen, A.; Aihara, M.; Semenova, S. 1.; Negishi, Y.; Mori,
0.; Yasuda, M. Effects of pore size on separation mechanisms of microﬁltration of
oily water, using porous glass tubular membrane. J. Membr: Sci. 1998, 145, l-14.

[219] Cheryan, M. Ultrafiltration handbook; Technomic publishing company:
Lancaster, 1986.

126

 

[220] Cheryan, M.; Rajagopalan, N. Membrane processing of oily streams.
Wastewater treatment and waste reduction. J. Membr. Sci. 1998, 151, 13-28.

[221] Hua, F. L.; Wang, Y. J.; Tsang, Y. F.; Chan, S. Y.; Sin, S. N.; Chua, H. Study of
microﬁltration behaviour of oily wastewater. J. Environ. Sci. Heal. A 2007, 42,
489-496.

[222] Psoch, C.; Schiewer, S. Long-term ﬂux improvement by air sparging and
backﬂushing for a membrane bioreactor, and modeling permeability decline.
Desalination 2008, 230, 193-204.

[223] McAdam, E. J .; Eusebi, A. L.; Judd, S. J. Evaluation of intermittent air sparging
in an anoxic denitriﬁcation membrane bioreactor. Water Sci. Technol. 2010, 61,
2219-2225.

[224] Psoch, C.; Schiewer, S. Direct ﬁltration of natural and 'simulated river water

with air sparging and sponge ball application for fouling control. Desalination 2006,
197, 190-204.

[225] Li, Q. Y.; Cui, Z. R; Pepper, D. S. Fractionation of HSA and IgG by gas sparged
ultraﬁltration. J. Membr: Sci. 1997, 136, 181-190.

[226] Miller, J. D.; University of Utah, Salt Lake City, Utah: US, 1981; Vol.
4,279,743.

[227] Bretney, E. In US. Patent, 1891; Vol. 453,105.

[228] Wolbert, D.; Ma, B. F.; Aurelle, Y.; Seureau, J. Efficiency estimation of liquid -
liquid hydrocyclones using trajectory analysis. Aiche J. 1995, 41 , 1395-1402.

[229] Colman, D. A.; Thew, M.; US, 1988; Vol. 4764287.

[230] Benard, A.; Chase, G; Davis, T.; Gaustad, M.; Koppula, K.; Mohan, R.; Muthu,
S.; Petty, C. A.; Pascheco, J.; Rieck, R.; Shan, W.; Tarabara, V. V. In American
Filtration & Separations Society Annual Conference, 2010.

[231] Kong, J.; Li, K. Oil removal from oil-in-water emulsions using PVDF
membranes. Sep. Purif Technol. 1999, 16, 83-93.

[232] Nazzal, F. F.; Wiesner, M. R. Microﬁltration of oil-in-water emulsions. Water
Environ. Res. 1996, 68, 1187-1191.

127

 

[233] Cumming, I. W.; Holdich, R. G; Smith, I. D. The rejection of oil using an
asymmetric metal microﬁlter to separate an oil in water dispersion. Water Res. 1999,
33, 3587-3594.

[234] Colman, D. A.; Thew, M. T. Correlation of separation results from light
dispersion hydrocyclones. Chem. Eng. Res. Des. 1983, 61, 233-240.

[235] Bednarski, S.; Listewnik, J. Hydrocyclones for simultaneous removal of oil and
solid particles from ship's oily waste. F iltr: Separat. 1988, 25, 92-97.

[236] Atwood, R. L.; Advanced Processing Technologies, Inc.: US, 1991; Vol.
4,997,549.

[237] Ducom, G; Matamoros, H.; Cabassud, C. Air sparging for ﬂux enhancement in
nanoﬁltration membranes: application to O/W stabilised and non-stabilised emulsions.
J. Membr: Sci. 2002, 204, 221-236.

[238] Tanchak, 0. M.; Barrett, C. J. Swelling dynamics of multilayer ﬁlms of weak
polyelectrolytes. Chem. Mat. 2004, 16, 2734-2739.

128