137
857

 

THS

THESIS

- 2004'. LIBRARY
24% a I l 3 0 Michigan State
University

This is to certify that the
thesis entitled

Ultrathin, Multilayered Polyelectrolyte Films as Nanofiltration
Membranes

presented by

Brian W. Stanton

has been accepted towards fulfillment
of the requirements for the

MS. degree in Chemistrx

 

77M, mm

Major P'fofessor’s Si at

lQ/lll/Qg
Date

MSU is an Affirmative Action/Equal Opportunity Institution

._-—o.-.-.-.--¢—--n—.-.—.-A

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 CJCIRC/DateDUGpSS-DJS

 

 

Ultrathin, Multilayered Polyelectrolyte Films as Nanofiltration Membranes

By

Brian W. Stanton

A THESIS

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

MASTER OF SCIENCE

Department of Chemistry

2003

ABSTRACT
Ultrathin, Multilayer Polyelectrolyte Films as Nanofiltration Membranes
By

Brian W. Stanton

This thesis shows that alternating polyelectrolyte deposition on porous supports
yields nanofiltration membranes that allow high water flux along with selective ion
transport. Membranes composed of 4.5 to 5 layer pairs of poly(styrene
sulfonate) (PSS)/poly(allylamine hydrochloride) (PAH) on porous alumina allow
water fluxes of 1 to 2 m‘q’m’zd‘1 at 4.8 bar while exhibiting MgSO4 rejections of
96%. In general, divalent-ion rejection increases when the charge of the outer
layer of the membrane matches that of the ion being rejected. Increasing the
concentration of supporting electrolyte present during deposition of the
terminating PSS layer results in a higher surface charge density, and hence
higher NaZSO4 rejections. The use of PSS/poly(diallyldimethylammonium
chloride) bilayers as a gutter layer underneath the selective PSS/PAH skin allows
for higher water fluxes on both alumina and polymeric supports. Preliminary
research shows that the use of polyelectrolyte films for nanofiltration may allow
for repair of fouled membranes. Removal of the surface layer (along with fouling
material) from the membrane using high pH buffer, and adsorption of a new

surface layer may regenerate the membranes.

Dedicated to my wife Angela and to my parents.

ACKNOWLEDGMENTS

I gratefully acknowledge financial support of this work by the US Department of
Energy Office of Basic Energy Sciences and Pall Corporation. We also thank
Keith Harris of DOW Chemical for fruitful discussions and the design of the
nanofiltration apparatus. I would also like to thank Dan Sullivan and Matt Miller

for taking FESEM pictures for this research.

TABLE OF CONTENTS

List of Tables

List of Figures

List of Abbreviations

Chapter 1. Introduction

Chapter 2. Nanofiltration of Salt Solutions Using Multi-layered
Polyelectrolyte Membranes on Alumina Supports

Chapter 3. Development of Composite Polyelectrolyte NF
Membranes on Polymeric Supports

Chapter 4. Fouling and Repair of NF Membranes (Future Work)

vi

vii

viii

11

33

53

Table 2.1.

Table 2.2.

Table 2.3.

Table 2.4.

Table 3.1.

Table 3.2.

Table 3.3.

Table 3.4.

Table 4.1.

Table 4.2.

LIST OF TABLES

Ion rejection and water flux for various polyelectrolyte
nanofiltration membranes using a 1000 ppm salt
concentration in the feed solution.

Ion rejection and water flux for various polyelectrolyte
nanofiltration membranes using a 50 ppm salt
concentration in the feed solution.

Percent rejection values for individual anions and
water fluxes from NF experiments with mixed salt
solutions.

Cl'/SO42’ selectivities from single-salt and mixed-salt
NF experiments.

Ellipsometric thicknesses for PSS/PDADMAC films on
aluminum or gold-coated substrates.

Cl' and 8042' % rejections, water fluxes, and CI‘ISO42'
selectivities for NF with MPF membranes deposited on
alumina using a PSS/PDADMAC gutter layer.

Water flux and % rejection values in NF of CaClz and
MgSO4 solutions by MPFs containing a PSS/PDADMAC
gutter layer on alumina.

Percent rejection of Cl' and 8042' as well as water flux
in NF with various membranes prepared by MPF deposition
on PES 100 supports.

Ellipsometric thickness values for 5-bilayer PAH/PSS
films before and after submersion in buffer solutions.

Absorbances for the 3 main peaks in the spectra in
Figure 4.4.

vi

22

23

28

29

39

41

44

47

62

65

Figure 1.1.

Figure 1.2.

Figure 2.1.

Figure 2.2.
Figure 2.3.

Figure 3.1.

Figure 3.2.

Figure 3.3.

Figure 3.4.

Figure 4.1.

Figure 4.2.

Figure 4.3.

Figure 4.4.

LIST OF FIGURES

Schematic diagram of the alternating deposition
process used to form a polyelectrolyte film on a
porous alumina support.

Cross-sectional view of a fouled membrane on a
porous support.

The chemical structures of the polyelectrolytes used in
this study.

Experimental setup for nanofiltration experiments.
Schematic diagram of the nanofiltration membrane cell.

Field emission scanning eletron microscopy (FESEM)
image of the surface of a bare Pall Corporation 100 kD
ultrafiltration membrane.

Shows the chemical structure of PDADMAC.

Cross-sectional FESEM image of
[PSS/PDADMACMPSS/PAH]PSS deposited on porous
alumina.

FESEM image of a [PSS/PDADMACMPSS/PAH]PSS
film (final PSS layer deposited from 2.5 M MnClz)
deposited on PES100

Plot of water flux vs. time for the NF of a 50 ppm Mg2+
solution (added as MgSO4) containing 20 mg/L of humic
acid.

Plot of the Mg2+ concentration in the permeate vs. time
when filtering a 50 ppm Mg” (from MgSO4) solution
containing 20 mg/L humic acid.

Plot of 8042' concentration in the permeate solution vs.
time when filtering a 50 ppm Mg” (from MgSO4) solution
containing 20 mg/L of humic acid.

ERFTIR spectra of PAH/PSS films on a gold-coated
silicon wafer.

vii

13
16

17

34

37

41

48

57

58

59

64

Abbreviation
NF

MPF

PSS

PAH

NOM

PAA

MPA
FESEM
PDADMAC
Abs

AFM

PES

ERFTIR

LIST OF ABBREVIATIONS

Meaning of abbreviation

Nanofiltration

Multilayer polyelectrolyte film

Poly(styrene sulfonate)

Poly(allylamine hydrochloride)

Natural Organic Matter

Poly(acrylic acid)

3-mercaptopropionic acid

Field emission scanning electron microscopy
Poly(diallyldimethylammonium chloride)
Absorbance

Atomic force microscopy

Poly(ethersulfone)

External reflectance Fourier transform infrared

spectroscopy

viii

Chapter 1

INTRODUCTION

Nanofiltration (NF) is an important membrane-based separation process
for applications such as water softening, wastewater reclamation, and dye-salt
separations."6 Fluxes and analyte rejections in NF are between those of
ultrafiltration and reverse osmosis, and thus, NF affords rejection of large
molecules or multivalent ions along with a relatively high water flux at low
operating pressures (<14 bar).2'7 This thesis investigates the use of alternating
polyelectrolyte deposition as a method of creating NF membranes for potential
industrial applications.

To achieve high water flux, the majority of NF membranes consist of a
thin, selective skin deposited on a thick, permeable support, e.g., an ultra or
microfiltration membrane. The composition of the selective layer determines the
separation characteristics of the membrane, while the support provides
mechanical stability to the system. Typically, solute transport through NF
membranes is a function of the charge and pore size in the skin layer.7'1o Thus,
in NF of ions, the membrane skin is usually highly charged and has an effective
pore size between 1 and 5 nm.‘"13

Adsorption of multilayered polyelectrolyte films (MPFs) on porous supports
is a relatively new technique that provides a simple way to create charged,

14-19

ultrathin membrane skins. This technique is attractive because it affords

 

 

—+ I;\ j —+ ' '1. +
_+ T _+' £77.. "2+
303' '-, Q) '+_
—+ > D++
:3“ —+' j. . +
+ (PSS) ’2 +_. +
+ + X).
_+ —+'Q(\I+. +

 

repeated as many times as needed to prepare multilayer films.

control over the thickness, charge density, and composition of the active skin
layer.20 Deposition of MPFs simply involves alternating immersions of a charged
substrate into solutions containing oppositely charged polyelectrolytes (figure
1.1), and film thickness depends on the number of polyelectrolyte layers as well
as the pH and supporting salt concentration in the polyelectrolyte solutions.”27

As virtually any polyelectrolyte can form MPFs,28’32

alternating polyelectrolyte
deposition is capable of forming a wide variety of membranes.

Several groups performed fundamental studies of transport through
MPFs.15'23'33'39 Mohwald and von Klitzing examined the transport of neutral

36,37 and

quenchers in polyelectrolyte films using time-dependent fluorescence,
other groups investigated the ion-permeability of MPFs using electrochemical
techniques.23'3&4° lon transport through the membrane occurs through a set of
tortuous pores with partitioning of ions through the bilayers of the film partially
reponsible for rejection. The rate of ion diffusion through MPFs depends on ion
charge, the number of bilayers in the film, charge density at the film surface, ionic
strength, and individual bilayer thicknesses. Krasemann and Tieke recently
showed that MPFs composed of 10-60 bilayers exhibit Cl'lSO42‘ selectivities as
high as 45 in diffusion dialysis experiments,15 and recent work in the Bruening
group demonstrated that MPFs with as few as five bilayers can effectively cover
porous supports. The selectivity of such ultrathin films depends greatly on the
charge and composition of the MPF.”"““‘4 However, very little work has been

performed using MPFs in NF. In one recent report, Krasemann and Tieke

demonstrated that 60-bilayer films of poly(vinylamine)/poly(vinylsulfate) could

reject 95% of NaSO4 and up to 98% of MgSO4.16 However, in that case MPFs
were quite thick, and water flux was several orders of magnitude lower than that
of commercial NF membranes.

The goal of my research is to investigate the possibility of using alternating
deposition procedure to create ultrathin NF membranes that will have greater
efficiencies than current commercial membranes. Previous experiments have
shown the ability of poly(styrene sulfonate) (PSS)/poly(allylamine hydrochloride)

“’41 Based on

(PAH) ultrathin films to reject divalent ions in diffusion dialysis.
these results, it may be possible to apply these MPFs to NF applications to allow
for greater water flux values while exhibiting competitive ion rejection properties.
Because of the ultrathin nature (<100 nm thick) of these PSS/PAH membranes,
they should yield high water fluxes along with divalent ion rejection. My research
also explores development of an industrially suitable support for MPFs and
fouling of these membranes.

Chapter 2 of this thesis begins the discussion of MPF NF membranes by
evaluating the NF performance of ultrathin PSS/PAH MPFs on alumina.
Manipulation of deposition conditions and membrane composition allows control
over membrane charge and affords Cl'lSO42' selectivities as high as 80.
Additionally, these membranes permit water fluxes that are as much as twice as
large as those of current commercial membranes. Typical water fluxes through
commercial membranes are about 0.9 m3m'2m’1. 45'“

In spite of the promise of these membranes, there are a number of

challenges that must be overcome before they can be applied industrially. One

of these is the development of an inexpensive, porous substrate for film
deposition. The porous alumina discs that I initially employed are very brittle and
do not provide a high surface area. Although alumina tubes may be more robust
and provide higher surface area,”48 those materials are still prohibitively
expensive for large scale application. Most likely, the development of practical
membrane systems will require replacement of alumina with a porous polymeric
support. Chapter 3 describes my research in that direction. I deposited
multilayer polyelectrolyte films on an ultrafiltration membrane and evaluated the
NF performance of the system. Unfortunetly, membranes composed of 4.5-
bilayers of PSS/PAH on the ultrafiltration support, give fluxes that are
considerably lower than those of similar membranes on alumina supports. Also
the ion rejection decreased, leading me to believe that the film was not as thick
as that on the alumina. This led to an investigation of the use of a non-selective,
highly permeable gutter layer beneath the selective skin. Such a layer may allow
full coverage of a polymeric support while maintaining the high flux and ion
rejection of polyelectrolyte membranes.

Another major challenge in the development of any NF membrane is
fouling, the deposition of insoluble material on the surface of the membrane. The
presence of this unwanted material often decreases both flux and selectivity.
This problem is complicated by the fact that there are a number of different
fouling processes including deposition of natural organic matter (NOM),"9‘51

52.53

colloidal fouling, scaling,54 and adsorption of proteins or bacteria.“56 The

rates of the first three processes increase with both the roughness of the

membrane surface and permeate flux.53 Colloidal and NOM fouling occur when
insoluble particulates get caught in the hills and valleys of a rough surface, while
scaling happens when cations and anions are rejected by the membrane such
that they reach a surface concentration greater than the solubility of a particular
salt. A common example of this is precipitation of calcium sulfate on membrane
surfaces. All types of fouling lead to inefficiencies due to the formation of a cake
layer on the surface of the membrane (Example shown in Figure 1.2). In order
for polyelectrolyte NF membranes to become widely used commercially, these
fouling issues must be explored, and methods for cleaning the membranes will
likely need to be developed. Chapter 4 of this thesis suggests future work to
determine the fouling rate of MPFs and potential methods for repairing fouled

polyelectrolyte membranes.

 

 

Figure 1.2: Cross-sectional view of a fouled membrane on a porous support.
The circles represent colloids or NOM, which may get caught in the hills and

valleys of a rough membrane surface to begin the formation of a cake layer.

References:

(1) Xu, X.; Spencer, H. G. Desalination 1997, 114, 129.
(2) Yaroshchuk, A.; Staude, E. Desalination 1992, 86, 115.

(3) Hofman, J. A. M. H.; Beerendonk, E. F.; Folmer, H. C.; Kruithof, J. C.
Desalination 1997, 113, 209.

(4) Bohdziewicz, J.; Bodzek, M.; Wasik, E. Desalination 1999, 121, 139.
(5) Berg, P.; Hagmeyer, G.; Gimbel, R. Desalination 1997, 113, 205.
(6) Rautenbach, R.; Linn, T.; Eilers, L. J. Membr. Sci. 2000, 174, 231.

(7) Bhattacharyya, D.; Williams, M. E. In Membrane Handbook, Ho, W. S. W.,
Sirkar, K. K., Eds.; Van Nostrand Reinhold: New York, 1992, pp 263.

(8) Childress, A. E.; Elimelech, M. Environ. Sci. Technol. 2000, 34, 3710.

(9) Afonso, M. D.; Hagmeyer, G.; Gimbel, R. Separation Pun'f. Technol. 2001,
22-23, 529.

(10) Bowen, W. R.; Mohammad, A. W.; Hilal, N. J. Membr. Sci. 1997, 126, 91.
(11) Kim, C. K.; Kim, J. H.; Roh, l. J.; Kim, J. J. J. Membr. Sci. 2000, 165, 189.
(12) Kosutic, K.; Kastelan-Kunst, L.; Kunst, B. J. Membr. Sci. 2000, 168, 101.
(13) Shibata, M.; Kobayashi, T.; Fujii, N. J. Appl. Polym. Sci. 2000, 75, 1546.
(14) Harris, J. J.; Stair, J. L.; Bruening, M. L. Chem. Mater. 2000, 12, 1941.
(15) Krasemann, L.; Tieke, B. Langmuir 2000, 16, 287.

(16) Jin, W.; Toutianoush, A.; Tieke, B. Langmuir 2003, 19, 2550.

(17) Dubas, S. T.; Farhat, T. R.; Schlenoff, J. B. J. Am. Chem. Soc. 2001, 123,

(18) Lvov, Y. In Protein Architecture; Lvov, Y., Mohwald, H., Eds.; Marcel
Dekker, Inc: New York, 2000, pp 125.

(19) Schlenoff, J. B.; Ly, l-l.; Li, M. J. Am. Chem. Soc. 1998, 120, 7626.
(20 Decher, G. Science 1997, 277, 1232.
(21
(
(

vv

Yoo, D.; Shiratori, S. 8.; Rubner, M. F. Macromolecules 1998, 31, 4309.
22) Shiratori, S. S.; Rubner, M. F. Macromolecules 2000, 33, 4213.
23) Harris, J. J.; Bruening, M. L. Langmuir 2000, 16, 2006.

(24) Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153.

(25) Caruso, F.; Niikura, K.; Furlong, D. N.; Okahata, Y. Langmuir 1997, 13,
3422.

(26) Decher, 6.; Hong, J.-D.; Schmitt, J. Thin Solid Films 1992, 210/211, 831.

(27) Hoogeveen, N. G.; Cohen Stuart, M. A.; Fleer, G. J. Langmuir 1996, 12,
3675.

(28) Bertrand, P.; Jonas, A.; Laschewsky, A.; Legras, R. Macromol. Rapid
Commun. 2000, 21, 319.

(29) Kotov, N. A.; Dékany, l.; Fendler, J. H. J. Phys. Chem. 1995, 99, 13065.
(30) Lvov, Y.; Haas, H.; Decher, G.; M6hwald, H. Langmuir 1994, 10, 4232.

(31) Lvov, Y.; Ariga, K.; Ichinose, l.; Kunitake, T. J. Am. Chem. Soc. 1995, 117,
6117.

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

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

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

(36) Klitzing, R. v.; Mohwald, H. Thin Solid Films 1996, 352.

(37) Klitzing, R. v.; Mohwald, H. Macromolecules 1996, 29, 6901.

(38) Han, S.; Lindholm-Sethson, B. Electrochimica Acta 1999, 45, 845.
(39) Farhat, T. R.; Schlenoff, J. B. Langmuir 2001, 17, 1184.

(40) Farhat, T. R.; Schlenoff, J. B. J. Am. Chem. Soc. 2003, 125.

(41) Stair, J. L.; Harris, J. J.; Bruening, M. L. Chem. Mater. 2001, 13, 2641.

(42) Balachandra, A. M.; Dai, J.; Bruening, M. L. Macromolecules 2002, 35,
3171.

(43) Dai, J.; Balachandra, A. M.; Lee, J. l.; Bruening, M. L. Macromolecules
2002, 35, 3164.

(44) Dai, J.; Jensen, A. W.; Mohanty, D. K.; Erndt, J.; Bruening, M. L. Langmuir
2001, 17,931.

(45) Performance of commercially available membranes can be found at
www.dow.com/quuidseps/pc/product.htm.

(46) Bhattacharyya, D.; Williams, M. E.; Ray, R. J.; McCray, S. B. In Membrane
Handbook; Ho, W. S. W., Sirkar, K. K., Eds.; Van Nostrand Reinhold: New York,
1992, pp 281.

(47) Aust, U.; Moritz, T.; POPP, U.; Tomanal, G. J. Sol-Gel Sci & Tech. 2003,
26, 715.

(48) Li, Y.; Wang, J.; Shi, J.; Zhang, X.; Lu, J.; Bao, Z. Sep. & Pur. Tech. 2003,
32, 397.

(49) Cho, J.; Amy, G.; Pellegrino, J. Wat. Res. 1999, 33, 2517.

(50) Yuan, W.; Zydney, A. L. Environ. Sci. Technol. 2000, 34, 5043.

(51) Seidel, A.; Elimelech, M. J. Membr. Sci. 2002, 203, 245.

(52) Zhu, X.; Elimelech, M. Environ. Sci. Technol. 1997, 31 , 3654.

(53) Vrijenhoek, E. M.; Hong, S.; Elimelech, M. J. Membr. Sci. 2001, 188, 115.
(54) Le Gouellec, Y. A.; Elimelech, M. Env. Eng. Sci 2002, 19, 387.

(55) Sadr Ghayeni, S. B.; Beatson, P. J.; Schneider, R. P.; Fane, A. G. J.
Membr. Sci. 1998, 138, 29.

(56) Pieracci, J. P.; Crivello, J. V.; Belfort, G. Chem. Mater. 2002, 14, 256.

10

Chapter 2

NANOFILTRATION OF SALT SOLUTIONS USING MULTl-LAYERED
POLYELECTROLYTE MEMBRANES ON ALUMINA SUPPORTS

Preparation of membranes using MPFs requires deposition of these films
on a porous support, which provides mechanical strength. This chapter
describes the NF performance of simple MPFs deposited on porous alumina
(Anodiscm) substrates, while Chapter 3 focuses on polymeric supports. Porous
alumina supports are ideal for initial NF studies because they have a well-
defined, highly porous structure that allows a very high water flux. This enables
the study of flux and rejection properties of the MPFs with minimal effects from
the support. Additionally, below pH 8 alumina disks have a positively charged
surface1 that allows for polyelectrolyte deposition, and the small diameter of
surface pores (20 nm) easily permits full coverage of the support without filling of
the pores.1

The main parameters of importance in nanofiltration are water permeance
and ion rejection. Recently, Jin and coworkers examined NF with 60-bilayer
poly(vinylamine)/poly(vinylsulfate) films,2 but water flux through those
membranes was very low. In contrast, the minimal thicknesses of the MPFs
employed in this work allow water fluxes comparable to those achieved with
commercial membranes. Additionally, this study shows that ion rejection can be
manipulated by altering the outer polyelectrolyte layer to change packing and
surface charge. Such manipulations afford enhancements in Cl’lSO42'

selectivities that occur without an accompanying decrease in CI' flux. Our

11

studies of NF with both mixed and single-salt solutions suggest that selectivity is
a complicated function of solution composition, surface charge, packing, and ion

adsorption.

Experimental

Materials

Poly(allylamine hydrochloride) (PAH) (Aldrich, Mw=70,000), poly(styrene
sulfonate) sodium salt (PSS) (Aldrich, Mw=70,000), poly(acrylic acid) (PAA)
(Mw=90,000, 25 wt% solution, Alfa Aesar), MnClz (Acros), NaBr (Aldrich), NaCI
(Spectrum and CCI), CaClz (CCI), NaZSO4 (Baker), MgSO4-7H20 (Spectrum),
NaHC03 (Mallinckrodt), Na2C03 (Spectrum), CaSO4 (CCI), 3-mercaptopropionic
acid (MPA) (Aldrich), mesityl oxide (Aldrich) and Congo Red (Fisher) were used

as received.

Polyelectrolyte Deposition on Porous Alumina Supports

Prior to film deposition, the porous alumina supports (Whatman®
AnodiscTM 0.02 pm membrane filters) were UV/O3 cleaned for 15 minutes
(Boekel UV_C|eanTM model 135500). The alumina membrane has two distinct
sides: the filtrate side, which has a skin layer of 0.02 um-diameter pores, and the
permeate side, which has 0.2 um-diameter pores. The polyelectrolyte deposition
was limited to the filtrate side using a holder. Polyelectrolyte deposition began
with a 2-min immersion of the alumina support in a 0.02 M solution of PSS.

Polymer molarities are given with respect to the repeating unit. The PSS solution

12

0" “5i”

NH3+CI‘
SO3'Na+ Poly(allylamine hydrochloride)
Poly(styrene sulfonate) PAH
PSS

E :1 n
‘O O
Poly(acrylic acid)

PAA

Figure 2.1: The chemical structures of the polyelectrolytes used in this study.

13

contained 0.5 M MnClz and was adjusted to pH 2.1 with HCI. Following a 1-min
water rinse, the support was then immersed for 5 min in a 0.02 M PAH solution
that contained 0.5 M NaBr (pH adjusted to 2.3 with HCI). Subsequent bilayers
were deposited in the same way. Deposition conditions were chosen based on
literature preparations of polyelectrolyte filmsf"6 and immersion times are not
necessarily sufficient to achieve equilibrium adsorptionf"6 In some cases, after
deposition of 4-bilayers of PSS/PAH, the membranes were immersed in a PSS
solution that contained 0 or 2.5 M MnClz. Another group of membranes had 1
layer of PAA deposited on 4-bilayers of PSS/PAH (PAA deposition: 5-min
immersion in a 0.02 M PAA solution containing 0.5 M NaCl and adjusted to pH
4.5 with NaOH). A 1-minute rinse with water (Milli-Q, 18 MQ-cm) followed all
polyelectrolyte deposition steps. After deposition of the desired number of
polyelectrolyte layers, membranes were stored in water until use. Some
membranes were dried before use, and there was no noticeable difference in

performance after drying.

Ellipsometgr

Ellipsometric thickness determinations (J.A. Woollam model M-44 rotating
analyzer ellipsometer) were performed using Au-coated substrates (200 nm of
Au sputtered on 20 nm of Cr on Si(100) wafers) that were treated with 0.002 M 3-
mercaptopropanoic acid (MPA) in ethanol for 30 minutes and rinsed with ethanol.
This process primes the surface with —COOH groups that can be ionized for

subsequent polycation deposition. Formation of films on these substrates was

14

accomplished using the same deposition conditions as with alumina supports,
but PAH was deposited first rather than PSS. Prior to film deposition, optical
constants of the MPA-coated substrates were determined at the 44 wavelengths
of the ellipsometer, and for calculation of MPF thickness, the refractive index of
the MPF was assumed to be 157'8 The thickness of the top PSS layer was
determined as the difference between film thicknesses after deposition of 4.5 and
5 PAH/PSS bilayers. Previous work showed that films on Au and porous alumina

have similar thicknesses.1

Nanofiltration

NF experiments were conducted in the home-built, cross-flow system
shown schematically in Figure 2.2. The system was pressurized to 4.8 bar with
Ar, and the feed tank was used to store the various brine solutions (4 L), which
contained 50 or 1000 ppm of Na*, 8042', Ca2+, or Mg”. Solutions were prepared
from NaCl, Nast4, CaClz, and MgSO4, respectively. A stainless steel frit (Mott
Corporation) was used as a prefilter to remove rust or particulate matter from the
feed stream. The flow meter was set to 18 mL/min, which is 100-fold greater
than the permeate flow rate. This cross flow is high enough to minimize
concentration polarization, as increasing the cross flow rate to 40 mL/min had
little, if any, effect on ion rejection or permeate flux. The membranes were
housed in a home-built stainless steel/o-ring cell that exposed an active

membrane area of 1.5 cm2 (Figure 2.3). The active area was determined by

15

 

 

 

 

 

 

 

 

 

 

k_J

Figure 2.2: Experimental setup for nanofiltration experiments: Ar tank (1),
stainless steel feed tank (2), centrifugal pump (3), prefilter (4), flowmeter (5),
pressure gauges (6), membrane cells (7), and graduated cylinders (8). There is
a water bath between the pump and the flow meter to dissapate heat from the

pump and prefilter. (Figure courtesy of Jeremy Harris)

16

 

 

 

 

 

 

 

 

 

 

 

Figure 2.3: Schematic diagram of the nanofiltration membrane cell. (A) shows
how the cell is placed together: inlet/outlet ports (1&2), threads (3), rubber o-ring
(4), membrane being analyzed (5), porous stainless steel frit (6), cap that holds
the frit and presses the membrane tight against o-ring (7), and threaded bottom
part of cell that top part screws into (8). The cell is assembled in the direction of
the arrow. (B) shows a bottom view of the upper portion of the cell: recessed

inlet/outlet channels (1&2) and o-ring (3). (Figure courtesy of Jeremy Harris)

17

performing NF with 0.1% congo red dye, and then measuring the diameter of the
stained area of the membrane with calipers. Pressure gauges were used to
ensure that no significant pressure drops occurred across the membrane cells.
Salt rejection in single-salt NF experiments was determined using conductivity
measurements and appropriate calibration curves.

Once the system was pressurized, the membranes were allowed to be in
contact with the flowing solution for 60 min, and permeate aliquots were collected
every 30 min for 1.5 h. After a total of 18 h of pressurized operation, permeate
aliquots were again collected every 30 min for three hours. The 18-h
conditioning period is sufficient to achieve stable membrane performance. In
some cases, NF was performed for 48 h, and no significant change in flux or
rejection relative to that after 18 h was noticed. These results demonstrate that
no significant delamination of the polyelectrolyte layers occurs over short time

penods.

CI‘ISO4‘2' selectivitv measurements

Solutions with both Cl' and SO42“ (sodium salts) were prepared at anion
concentrations of 50 or 1000 ppm. NF with these solutions occurred as
described above, and ion concentrations were determined with a Dionex 600 ion
chromatograph using an AS14A anion column and an 8 mM Na2003/1 mM
NaHC03 eluent. The concentrations were determined using a set of five

standards bracketing the expected feed and permeate concentrations.

18

Capillary Electrophoresis

Capillary electrophoresis experiments were performed with an Agilent
G1600A instrument that employed a UVNisible detector. Bare silica capillaries
(100 um inside diameter, purchased from BIOTAQ) were first treated with 0.1 M
NaOH for 30 min and rinsed with water for 5 min. The first PAH layer was
deposited onto the inside of the silica column by passing a PAH solution through
the capillary for 10 min using a pressure-driven (950 mbar) water flow. All
subsequent PAH and P88 layers were deposited with 5-min flows using the
solutions described previously. The capillary was rinsed with water for 5 min
between each step. Mesityl oxide (500 ppm) was used as a marker to detect
electroosmotic flow. Flow experiments with 1000 ppm 8042‘ were performed
using a potential of 15 W in positive mode, while experiments with 1000 ppm

Ca2+ utilized 5 W in negative mode.

Results and Discussion

Single-salt Nanofiltration with PSS/PAH Membranes

One of the main advantages of MPFs in NF is their small thickness, which
should result in high water fluxes. However, to achieve high rejections of analyte
molecules along with high water permeabilities, MPFs must be thick enough to
completely cover the pores of the material on which they are deposited. Our
previous studies showed that full coverage of a porous alumina support (0.02
um-diameter surface pores) by a PSS/PAH film requires adsorption of at least

4.5 bilayers.“9 Therefore, NF experiments focused on films consisting of 4.5 or 5

19

PSS/PAH bilayers. Films with 4.5 bilayers terminate in 3 P88 layer
([PSS/PAH]4PSS), while 5-bilayer films terminate with PAH ([PSS/PAH]5). Both
ellipsometric studies and scanning electron microscope images indicate that 4.5-
bilayer PSS/PAH films have thicknesses of ~20 nm.”7 In NF experiments,
coverage of porous alumina by PSS/PAH is evident from a dramatic decrease in
water flux after deposition of 4.5-bilayers. Under 4.8 bar of pressure, water flux
through the bare alumina membrane is about 180 m3m'2d4, but after PSS/PAH
deposition, the flux decreases to between 1 and 2.3 m3m'2d'1 (Tables 2.1 & 2.2).
Fluxes through PSS/PAH films are comparable to those of commercial NF
membranes.”11

In most cases, the water flux decreases with an increasing salt
concentration (compare Tables 2.1 and 2.2), presumably because of increased
osmotic pressure. Calculated osmotic pressures for 1000 ppm Mg2+ (MgSO4),
Ca2+ (CaClz), SO42“ (Na2804), and Na+ (NaCl) are 2.0, 1.8, 0.75, and 2.1 bar,
respectively, for 100% rejection. Water flux with 1000 ppm Na“ is similar to that
with 50 ppm Na”, but in this case, osmotic pressure effects are minimized by low
rejections. Increased adsorption of ions in the polyelectrolyte membrane at
higher ionic strength (see below) could also affect water flux.

PSS/PAH membranes show impressive ion-rejection values that vary with
the charge of the outer layer of the membrane. As seen in the top two rows of
Table 2.1, ion-rejection increases when the outer layer charge matches the

charge of the divalent ion of a 2:1 salt. For example, CaClz rejection increases

from approximately 86% to 96% after deposition of a terminating PAH layer. The

20

opposite trend occurs for Na2804. This suggests that electrostatic (Donnan)
exclusion plays a major role in salt rejection.12 One would expect little
dependence of NaCl or MgSO4 rejection on the charge of the outer layer of the
membrane because the cations and anions in these salts have the same
magnitude of charge. However, rejection of NaCI increases on going from a 4.5-
bilayer to a 5-bilayer PSS/PAH membrane. The higher NaCI rejection of 5-
bilayer membranes is accompanied by a 40% decrease in water flux relative to
4.5-bilayer membranes. This decrease is not simply due to increased film
thickness, as water flux through a 5.5-bilayer film is still 30% higher than that for
a 5-bilayer membrane. The lower water flux and higher NaCI rejection of
membranes terminated with PAH suggest that the PAH surface layer is more
dense than PSS. In the case of MgSO4, water flux is nearly the same for 4.5-
and 5-bilayer PSS/PAH membranes, probably because of Mg2+ adsorption to the
PSS layer. This adsorption may equalize MgSO4 rejection by PSS- and PAH-
terminated films.

Adsorption of divalent ions may also explain why MgSO4, Nazso4, and
CaClz rejections do not vary greatly with electrolyte concentration (compare
Tables 2.1 and 2.2, rows 1 and 2).“"14 Increased adsorption of divalent ions at
higher electrolyte concentrations should enhance electrostatic exclusion by
increasing membrane charge. This effect may compensate the greater
screening of surface charge that should occur at higher ionic strengths. NaCI
rejection does increase with decreasing concentration, suggesting that

adsorption of monovalent ions is minimal. The high CaClz rejection by 4.5-bilayer

21

Table 2.1: lon rejection (percent) and water flux values (listed in parentheses in
m3m'2d'1) for various polyelectrolyte nanofiltration membranes. The salt solutions

were 1000 ppm in the ion of interest, and nanofiltration was performed at 4.8 bar.

 

 

Membrane Na+ (NaCI) (N23842O4) Ca2+ (CaCI2) (Ragga)
[PSS/PAH]4Pss* (2:59:53) (£5631) (1:6:023) (19263.2)
[PSS/PAH]5 (3435064) (3055.0) $105093 (13632)
lPSS/PAHhPSS“ (2.2184502 ) (553:091) (1911:0131) (192545022)
lPSS/PAHI‘IPSSW (2.15332) (1:5:022) (1:38:086) (3003.2)
[PSS/PAH]4PAA (1.21431) (302562) (1815:021) (£331)

* 0.5 M MnClz in final PSS layer deposition solution

** 0 M MnClz in final PSS layer deposition solution

*** 2.5 M MnClz in final PSS layer deposition solution

22

Table 2.2: lon rejection (percent) and water flux values (listed in parentheses in

m3m'2d'1) for various polyelectrolyte nanofiltration membranes. The salt solutions

were 50 ppm in the ion of interest, and nanofiltration was performed at 4.8 bar.

 

 

Membrane Na+ (NaCI) ($138424) Ca2+ (CaCI2) ($33,264)
PSS/PAHhPSS" (500:0?) (2343.3) (2?34+i01-3) (3523-1)
[PSS/PAH]5 (1.815%) (132:50) (9126:6917) (917163.15)
[PSS/PAH]4PSS** (2.52132) (1:832) (2.904iiC)1.1) (1:624)
lPSS/PAHI4P33m (3.11.3352) (S7131) (1?:071) (1:8:053)
[PSS/PAH14PAA “if” (1.70031) (37:02.1) (121631)

*

**

0.5 M MnClz in final PSS layer deposition solution
0 M MnClz in final PSS layer deposition solution

*** 2.5 M lVlnClz in final PSS layer deposition solution

23

PSS/PAH membranes also provides evidence for ion adsorption. Calcium ions
should not be rejected by the negative surface presented by PSS-terminated
films, but rejection is about 90%, suggesting that adsorption of Ca2+ creates a
positively charged surface. It should be noted, however, that high Ca2+ rejections
may be partly due to large size of this ion. The hydrated radius of Ca2+ is 271
pm, while that of Na+ is 228 pm.15

To confirm that adsorption of divalent ions occurs, we performed
measurements of electroosmotic flow in silica capillaries coated with PAH/PSS
films.16'" For films terminated with PSS (5 PAH/PSS bilayers), electroosmotic
flow moves toward the negative electrode when the electrolyte contains 1000
ppm 8042' (NaZSO4). However, when the electrolyte contains 1000 ppm Ca2+
(CaCI2), a negative potential had to be applied in order to get the mesityl oxide
off the column. This shows that the direction of electroosmotic flow reverses,

confirming Ca2+ adsorption.

ControllingLReiection bv Manipulating the Outer Polvelectrohrte Layer

The results given above indicate that charge at the membrane-solution
interface plays a major role in ion rejection by MPFs. One way to control the
charge density at this interface is to vary the amount of supporting electrolyte in
the polyelectrolyte deposition solution. High concentrations of salts
electrostatically screen charged groups in the polymer, leading to a film with
increased thickness and a more loopy structure.‘3'"3'19 In general, increasing the

supporting electrolyte concentration present during deposition of the outer

24

polyelectrolyte layer should yield a higher density of fixed charge at the MPF

surface.”21

(When the outer layer is thick and has a loopy structure, its surface
is less intermingled with underlying layers and thus has a higher net charge
density”) To examine the effect of supporting electrolyte, a 4-bilayer PSS/PAH
film was prepared, as described in the experimental section, and subsequently
capped with a final layer of PSS that was deposited from solutions containing
various amounts of MHCIz. Ellipsometric measurements with films deposited on
Au yielded PSS capping layer thicknesses of 21:3, 28:3, and 5117 A for films
deposited in the presence of 0, 0.5, and 2.5 M MnCI2, respectively. These
thicknesses are consistent with literature valuesf"8

Tables 2.1 and 2.2 show that ion rejection depends significantly on the
concentration of supporting electrolyte present during the deposition of the outer
polyelectrolyte layer. For both 50 and 1000 ppm 8042‘, Na2804 rejections
increase from about 50% to >90% when the concentration of MnClz present
during deposition of the PSS capping layer increases from 0 to 2.5 M.
Remarkably, this increase in rejection occurs with no substantial change in water
flux. In contrast to N32804, CaClz and MgSO4 rejections decrease when the top
PSS layer is deposited from 2.5 M rather than 0 M MnClz. All of the ion-rejection
results are consistent with increased negative surface charge due to the
presence of 2.5 M MnClz during PSS deposition. MnClz was chosen as a
supporting electrolyte because it provides a high ionic strength, but one concern

in utilizing this salt is that adsorption of Mn2+ could affect membrane properties.

25

However, high water flux and rejection of negatively charged anions suggest that
any Mn2+ adsorbed in the membrane is removed during rinsing or NF.

Another way to control surface charge is to change the composition of the
polyanion (or polycation) used to cap the membrane. Previous studies suggest
that the high charge density of PAA-capped 4-bilayer PSS/PAH films should
increase 8042‘ rejection relative to that seen with 4.5-bilayer PSS/PAH films.9
Tables 2.1 and 2.2 show that films capped with PAA do exhibit increased 8042'
rejection compared to 4.5-bilayer PSS/PAH membranes (PSS capping layer
deposited in the presence of 0 or 0.5 M MnCl2). However, the high density of the

PAA layer also results in decreased water fluxes.

cr/soj' Selectivity of PSS/PAH membranes

To further investigate the selectivity of PSS/PAH membranes, NF was
performed with solutions containing both NaCl and NaZSO4. Table 2.3 contains
NF results for 50 ppm CI'I50 ppm 8042' and 1000 ppm Cl'/1000 ppm 8042' feed
solutions. In nearly every case, Cl' rejection decreases compared to single-salt
experiments, while 8042' rejection increases. In some cases, the concentration
of CI' in the permeate even exceeds the concentration of CI' in the feed. This
occurs because the concentration of Na+ in the mixed feed solutions is greater
than that of Cl', and thus diffusive flux of Na” is greater than that of CI'. The
difference in diffusive fluxes results in a diffusion potential that enhances Cl'
transport. In solutions containing only Na2804, a diffusion potential that

enhances the 8042' flux also develops because the diffusivity of Na+ in the

26

membrane is much greater than that of 8042'. The presence of CI' in the mixed
salt solutions reduces this diffusion potential, and thus, 8042' rejection increases

in mixed solutions. CI'/SO42 selectivities were calculated according to equation 1,

where C” and Cf represent the concentrations of the analyte in the permeate

and the feed, respectively.

Selectivity = (C5,- mg, )/(C” /C£)}_ ) (1)

503‘
For single-salt solutions, this selectivity can be rewritten as equation 2, where
%R represents the percent rejection of the appropriate salt.

Selectivity = (1 oo-O/o Rm, )/(100-% R,,,_,,,, ) (2)

Table 2.4 shows that in all cases, Cl'lSO42' selectivities are higher with mixed salt
solutions, presumably because of diffusion potentials, as mentioned above.
Additionally, selectivities depend greatly on the outer polyelectrolyte layer.
Deposition of the final layer of PSS from 2.5 M MnClz, rather than 0.5 M MnClz,
increases selectivity to approximately 30 (a 5 to 10-fold increase) for both 50
ppm and 1000 ppm mixed solutions.

Along with high selectivity, practical Cl'ISO42' separations will require both
high water flux and low Cl' rejections. Remarkably, the increase in selectivity due
to the presence of 2.5 M MnClz in the PSS deposition solution is not
accompanied by a decrease in water flux. Additionally, the low Cl‘ rejections by
these films (-6 and 40%) could allow a high recovery of NaCl. All films with 8042‘
rejections greater than 90% showed Cl’ rejections of 40% or lower.

Termination of membranes with PAA rather than PSS also yields very high

selectivities. With a 50 ppm mixed solution, PAA-capped membranes show a

27

Table 2.3: Percent rejection values for individual anions and water fluxes (in

m3m'2d‘1) from NF experiments with mixed salt solutions.

1000 ppm Cl'l1000 ppm

 

 

50 ppm Cl'/50 ppm 8042‘ 2-
so,
Membrane Cl' 8042' water CT 8042' water
flux flux
[PSS/PAH]4PSS (final
'aye'hzrfiggis M 91:6 73w 1.6_+_0.1 -2i5 81i8 1.3:02
[PSS/PAH]4PSS (final
'ayergggjf’ M 40i8 98:1 18:02 43:1 97:1 1.8i0.3

[PSS] PAH14PAA 17i3 95i2 1.0iO.1

28

 

5i1 98.9i0.1 O.9i0.1

Table 2.4: CI'ISO42' selectivities for single-salt and mixed-salt (value in
parentheses) NF experiments. These values were calculated from the average

% rejections shown in Tables 1, 2, and 3. Solutions were prepared using NaCI

 

 

and N32804.
50 ppm Cl'*/50 ppm 1000 ppm C|'*/1000
Membrane 8042' ppm 8042'
[PSS/PAH]4PSS (final layer from 0.7 :0.1 16:04
0.5 M MDCIz) (35:06) (6:3)
[PSS/PAH]4PSS (final layer from 7.7:1.3 17.4:2.7
2.5 M MnClz) (28:5) (33:8)
18:03 95:02
[PSS/PAHj4PAA (16:4) (84:4)

*Single—salt data for chloride-containing solutions (NaCI) were actually measured
using 50 ppm or 1000 ppm Na+ (Tables 1 and 2). Thus, for single-salt

measurements, Cl’ concentration was 77 or 1540 ppm.

29

Cl’/SO42' selectivity of about 16, but when Cl' and SO42" concentrations are 1000
ppm, selectivity increases to 84. This result parallels single-salt experiments, but
overall selectivities are much larger. However, water flux through PAA-capped
membranes is only ~60% of that through PSS/PAH.

In most solutions found in nature, the amount of CI' would be much
greater than that of SO42“. To simulate such a situation, we performed NF
experiments with 1000 ppm C|‘/50 ppm 8042' solutions using a PSS/PAH
membrane capped with PSS deposited from 2.5 M MnClz. Selectivity in this
case is 20 and Cl' rejection is 20%. Selectivity is only slightly lower than that

found for 50 ppm Cl°/50 ppm 8042' solutions.

Conclusions

Alternating polyelectrolyte adsorption on porous supports is a convenient,
versatile method for constructing NF membranes. These membranes provide
high water fluxes and rejections of divalent ions along with CI'ISO42' selectivities
as high as 80. Electrostatic exclusion is a major factor in transport selectivity,
and thus, by changing the outer layer charge in PSS/PAH membranes, rejection,
flux, and selectivity can be tailored to fit the needs of a particular separation.
These attributes make MPFs potentially attractive for NF applications such as

water and salt purification.

3O

References

(1) Harris, J. J.; Stair, J. L.; Bruening, M. L. Chem. Mater. 2000, 12, 1941.
(2) Jin, W.; Toutianoush, A.; Tieke, B. Langmuir 2003, 19, 2550.

(3) Caruso, F.; Niikura, K.; Furlong, D. N.; Okahata, Y. Langmuir1997, 13,
3422.

(4) Decher, G.; Lvov, Y.; Schmitt, J. Thin Solid Films 1994, 244, 772.
(5) Yoo, D.; Shiratori, S. S.; Rubner, M. F. Macromolecules 1998, 31, 4309.

(6) Casson, J. L.; Wang, H.-L.; Roberts, J. B.; Parikh, A. N.; Robinson, J. M.;
Johal, M. S. J. Phys. Chem. B 2002, 106, 1697.

(7) Harris, J. J.; Bruening, M. L. Langmuir 2000, 16, 2006.
(8) Tronin, A.; Lvov, Y.; Nicolini, C. Colloid Polym. Sci. 1994, 272, 1317.
(9) Stair, J. L.; Harris, J. J.; Bruening, M. L. Chem. Mater. 2001, 13, 2641.

(10) Performance of commercially available membranes can be found at
www.dow.com/liquidseps/pc/product.htm.

(11) Bhattacharyya, D.; Williams, M. E.; Ray, R. J.; McCray, S. B. In Membrane
Handbook; Ho, W. S. W., Sirkar, K. K., Eds.; Van Nostrand Reinhold: New York,
1992, PP 281.

(12) Peeters, J. M. M.; Boom, J. P.; Mulder, M. H. V.; Strathmann, H. J.
Membr. Sci. 1998, 145, 199.

(13) Schlenoff, J. B.; Ly, H.; Li, M. J. Am. Chem. Soc. 1998, 120, 7626.
(14) Toutianoush, A.; Tieke, B. Mat. Sci. and Eng. C 2002, 22, 135.

(15) Marcus, Y. Biophysical Chemistry 1994, 51, 111.

(16) Graul, T. W.; Schlenoff, J. B. Anal. Chem. 1999, 71, 4007.

(17) Capillary Electrophoresis measurements kindly done by Matt Miller.
(18) Shubin, V.; Linse, P. J. Phys. Chem. 1995, 99, 1285.

(19) Steeg, H. G. M. v. d.; Stuart, M. A. C.; Keizer, A. d.; Bijsterbosch, B. H.
Langmuir 1992, 8, 2538.

31

(20) Phuvanartnuruks, V.; McCarthy, T. J. Macromolecules 1998, 31, 1906.
(21) Farhat, T. R.; Schlenoff, J. B. Langmuir 2001, 17, 1184.

(22) Schlenoff, J. B.; Dubas, S. T. Macromolecules 2001, 34, 592.

32

Chapter 3

Development of Composite Polyelectrolyte NF Membranes on Polymeric

Suppons

High throughput applications of filtration membranes, e.g., NF, generally
utilize spirally wound or hollow fiber membranes that offer higher surface areas
and, hence, higher fluxes than sheet membranes. Such membrane geometries
are not practicle with ceramic materials, so polymers such as polysulfone are
usually employed as skin layer supports. In addition to being more flexible than
ceramics, polymeric supports are also much less expensive. This chapter
describes our initial investigations of MPF deposition on porous polymer
substrates and the NF properties of the resulting composite membranes.

Any material employed as a support in a composite MPF membrane must
permit high water flux while having pore sizes small enough to be coated by an
ultrathin polyelectrolyte film. Ultraflltration membranes may be capable of
fulfilling both of these requirements. Figure 3.1 shows that pores in a Pall
Corporation 100 kD ultrafiltration membrane have diameters ranging from 45 to
70 nm. Although these pores are 2 to 4 times larger than those in the alumina
substrates described in Chapter 2, they still may be small enough to allow full
coverage of the support. Additionally, the Pall membranes likely have positively

charged surfaces that should support polyelectrolyte deposition.

33

           

. t . ..
.. See:
_._ . . 1% 1. .W.a.nwa ....

     

500 nm

 

Field emission scanning eletron microscopy (FESEM) image of the

surface of a bare Pall Corporation 100 kD ultrafiltration membrane. The applied

potential was 10 kV. Pore sizes range from about 45 nm to 70 nm.

Figure 3.1

Unfortunately, however, although the uncoated ultrafiltration membranes
allow a reasonably high water flux, deposition of 5 bilayers of PSS/PAH on these
materials decreases the permeate flux to values well below those of current
commercial NF membranes. Along with this decrease in flux, ion rejection is
much less than that with polyelectrolyte films deposited on alumina, suggesting
that either there are defects in the film or the film structure is different on the
ceramic and polymeric supports. Thus, to utilize polymeric supports, we would
have to either modify the support or alter the polyelectrolyte film.

To better cover the polymer support and still achieve a reasonable flux, we
decided to first deposit a highly permeable, nonselective polyelectrolyte film on
the support and then coat this film with a selective skin layer. This non-selective
base film is often refered to as a gutter layer. Recently Tieke et al.1'2 showed that
PSS/poly(diallyldimethylammonium chloride) (PDADMAC) films on
polyacrylonitrile/polyethyleneterephthalate supports provide high flux, low-
selectivity membranes for pervaporation separations of ethanol and water. In
water, PSS/PDADMAC films tend to swell much more than PSS/PAH films,3'4
and for NF, this results in a much higher rate of water passage as well as a lower
ion rejection. The use of an ultrathin, selective layer on top of the
PSS/PDADMAC may result in both high flux and high ion rejections. This
chapter explores the use of PSS/PDADMAC films as gutter layers on both
alumina and polymeric supports. The use of gutter layers on alumina results in a
33% increase in flux with little change in selectivity relative to pure PSS/PAH

films. Gutter layers on polymer supports also allow formation of selective

35

membranes, but these membranes seem to be greatly affected by the presence

of divalent cations.

Experimental

Materials

Poly(diallyldimethylammonium chloride) (PDADMAC; Mw 100,000 —
200,000; 20 wt% in water; Aldrich) (stucture shown in Figure 3.2) was used as

received. All other polyelectrolytes and salts were described in Chapter 2.

Polyelectrolee Deposition on Porous Alumina and Polymeric Supports

To deposit PSS/PDADMAC gutter layers, a substrate was first immersed
in a solution of 0.02 M PSS (molarity is calculated with respect to monomer
repeat unit for all polyelectrolytes ) that contained NaCI as a supporting salt. The
deposition time was 3 min and was followed by a 1 min water rinse. The
substrate was then immersed for 3 min in a solution of 0.02 M PDADMAC (NaCI
as supporting salt) and rinsed for 1 min with water. Subsequent bilayers were
deposited in the same fashion. The concentration of the supporting salt was
varied to explore its effect on the gutter layer properties. All other polyelectrolyte
depositions occurred as described in Chapter 2. The ultrafiltration membranes
used for polymeric supports were cut into 2.2 cm-diameter circles from a sheet
provided by Pall. Before polyelectrolyte deposition, I soaked the polymeric
supports in water for about 30 min to wet and pre-swell them. The alumina

supports were cleaned as described previously.

36

NCI'
/ \

Poly(diallyldimethylammonium chloride)
(PDADMAC)

Figure 3.2: Shows the chemical structure of PDADMAC.

37

Ellipsometry and Nanofiltratio_n

Ellipsometric thickness determinations for PSS/PDADMAC films were
initially performed on aluminum-coated Si wafers (200 nm of Al on Si(100)).
However, because of aluminum corrosion, determination of the thicknesses of
films deposited from solutions containing high salt concentrations (greater than
0.5 M NaCI) had to be done on gold substrates coated with 3-mercaptopropionic

acid.5 Ellipsometric and nanofiltration procedures were described in Chapter 2.

Results and Discussion
Gutter layer on alumina

The idea behind using a gutter layer of PSS/PDADMAC is to cover the
porous surface of a support with a highly permeable, continuous layer that can
be easily covered with an ultrathin, selective layer. The first step was to
determine an appropriate thickness for the gutter layer. Previous work in our
group showed that 4 to 5 bilayers of PSS/PAH were required to completely cover
an alumina support. The thickness of such films was about 20 nm as determined
from SEM images. Therefore, it was assumed that the total film (gutter layer plus
selective layer) should be at least this thick. To determine the thicknesses of
PSS/PDADMAC films, optical ellipsometry measurements were performed on
aluminum-coated Si (when using high concentrations of salt, i.e. 2 M, gold coated
with MPA had to be used as a substrate since the aluminum corroded).
Aluminum-coated substrates are most appropriate for these experiments

because they contain a charged aluminum oxide coating on their surface that

38

Table 3.1: Ellipsometric thicknesses (A) for PSS/PDADMAC films on aluminum-
or gold-coated substrates as a function of the number of layers and the salt
concentration present during deposition.

Salt in deposition 2 bilayers 3 bilayers 4 bilayers 5 bilayers

 

solution
0.1 M NaCI3 63:7 103:9 142:10
0.5 M NaCIa 56:1 118:1.6 2:24:21
2 M NaCI” 187:2 535:43c

aFilm was deposited on aluminum-coated Si.

bFilm was deposited on gold-coated Si modified with a layer of MPA. Because
this surface is negatively charged, these films contain PDADMAC/PSS bilayers
rather than PSS/PDADMAC

CIn this case both the refractive index and thickness of the film were fit from

ellipsometric data.

39

should be representative of the alumina anodiscs. Table 3.1 presents the
ellipsometric thicknesses of PSS/PDADMAC films. After applying 1.5 bilayers of
PSS/PAH on top of the 3 bilayer gutter layer deposited from 0.1 M NaCl or 0.5 M
NaCI, the total film thickness on porous alumina is about 300 A as shown in
Figure 3.3. This is somewhat thicker than the film on aluminum measured by
ellipsometry (210 A), however, it should be enough to coat the porous surface
and allow for separation. Table 3.1 also shows that gutter layer thickness
increases with NaCl concentration, as anticipated.6 The use of NaCI in a
deposition solution should thus allow for a full substrate coverage with fewer
deposition steps.

Initial NF experiments with membranes prepared using a PSS/PDADMAC
gutter layer utilized alumina as the support. This allows for comparison with the
NF membranes that were prepared without a gutter layer (Chapter 2). As shown
in Table 3.2, the use of 3 bilayers of PSS/PDADMAC capped with 1.5 bilayers of
PSS/PAH (the final layer of PSS was deposited from 2.5 M MnClg) yields a water
flux of 2.5 m3m'2d'1, which is about 33% higher than the flux through a 4.5-bilayer
PSS/PAH membrane with the final layer being deposited from 2.5 M MnClz (see
Table 2.1). Even with the increased permeate flux, the Cl‘lSO42' selectivity is
essentially the same for the two types of membranes. Thus, for CI'ISO42‘
separations, the PSS/PDADMAC gutter layer is quite promising. Careful control
over the thickness of the gutter layer is important, however, as shown by the fact
that flux decreases more than two-fold when using a 5-bilayer rather than a 3-

bilayer gutter layer. The salt concentration present during gutter layer deposition

4o

'1 .

~ 7 ‘W'MM‘NM

.nwwugth, me. e.

}3
i
. :7
i
I
E
IE

it

..u

 

500 nm

Figure 3.3: Cross-sectional FESEM image of [PSS/PDADMAC]3[PSS/PAH]PSS

deposited on porous alumina. The applied potential for this photo was 15 W.

The final PSS layer was deposited from 2.5 M MnClz. The thickness of the film is

about 300 A.

41

Table 3.2: CI‘ and 8042' % rejections, water fluxes and CI'ISO42' selectivities for
NF with MPF membranes deposited on alumina using a PSS/PDADMAC gutter

layer. The feed solution contained 1000 ppm CH 1000 ppm 8042' pressurized to

 

4.8 bar (70 psi).
Cl' 8042’ water flux in Cl'/SO42'
Membrane % rejection % rejection m3m'2d'1 selectivity
[ESSSS/liafiwggi 2:3 94:2 36:06 17:6
[PSS/PDADMAC]3
[PSS/PDADMAC]5
[PSS/PDADMAC]3
[PSS/PDADMAC},
[PSS/PDADMACh -1 :1 96:1 18:01 28:6

[PSS/PAH] PAA“
* Final layer PSS was deposited from a solution containing 2.5 M MnCl2
** PAA layer deposited from 0.5 M NaCI
* Gutter layer was deposited from 0.5 M NaCl

1 Gutter layer was deposited from 0.1 M NaCl

42

should also affect flux. Changes in the ionic strength of the deposition solution
affect both film thickness and film structure. A high salt concentration will yield a
much thicker film that is very loopy in nature because there is a high degree of
charge screening during the deposition process.7 This also causes there to be a
lesser degree of interpenetration between bilayers, since the charge screening
reduces the charge density in each individual layer.7 The benefit of using higher
ionic strengths during deposition would be that it would take fewer bilayers to get
a defect free coating over the porous support. However, there are negative
effects. Using 2 M NaCI solutions to deposit the PSS/PDADMAC gutter layer
causes the water flux to decrease to 1 m3m'2d'1 when only 2 bilayers of this
gutter layer are deposited beneath a PSS/PAH/PSS (final PSS from 2.5 M
MnClz) capping layer. However, fluxes were similar when gutter layers were
deposited using either 0.1 M or 0.5 M NaCI as a supporting electrolyte (Table
3.2). For the most part, I employed 0.1 M NaCI as it should yield thinner gutter
layers.

The membrane capped with PAA yielded less encouraging results than
membranes capped with PSS/PAH. In the PAA case, the flux increased only
slightly over the non-gutter layer membranes discussed previously (compare
Table 3.2 with Table 2.1) Additionally, the CI'ISO42' selectivity dropped from 80
to 28. The may reflect a change in the PAA morphology that is induced by the
underlying film.8

To further examine the potential utility of membranes containing a

PSS/PDADMAC gutter layer, I performed NF using feed solutions containing

43

Table 3.3: Water flux and % rejection values in NF of CaClz and MgSO4

solutions by MPFs containing a PSS/PDADMAC gutter layer on alumina. The

system was pressurized to 4.8 bar (70 psi).

 

 

1000 ppm Ca2+ source 1000 ppm Mg2+ source
Membrane water flux 0 . . water flux 0 . .
(m3m'2d'1) /o rejection (m3m'2d'1) /c rejection
[PSS/PDADMAC]3
[PSS/PDADMAC]3 1 .6:0.4 79:2 1 5:08 93:1

[PSS/PAH] PSS”

 

*PSS layers were deposited from 0.5 M MnClz solutions, and PAH was deposited

from 0.5 M NaBr solutions.

TGutter layer was deposited from 0.1 M NaCI solutions.

44

Ca2+ and Mg”. A comparison of Table 3.3 and Table 2.1, shows that the
rejection of these two cations by gutter-layer-containing membranes is within 8%
of the rejection with 4.5-bilayer PSS/PAH membranes. Unfortunately, however,
the permeate flux is essentially the same for the two types of membranes. Thus,
the gutter layer is not serving its purpose of allowing an increased flux. This may
be due to adsorption of Ca2+ or Mg2+ by PSS/PDADMAC and a resultant

decrease in the swelling of the gutter layer.9

Multilayered Polyelectrolyte Films on Polymeric Supports

After analyzing the properties of alumina-supported PSS/PDADMAC films
capped with PSS/PAH bilayers, I examined the possibility of using such films on
a polymeric support. Because there is less charge to support film growth on a
polymer surface than on alumina, development of gutter layers that can fully
cover polymer supports may be especially important. In examining the potential
polymer supports provided by Pall, the pure water flux of the bare supports was
first measured. The poly(ethersulfone) (PES) 100 kD ultrafiltration membrane
had a pure water flux of about 10 m3m‘2d". This is about 5 times the flux through
coated alumina membranes so use of this support may be feasible. However, it
should be noted that flux through the PES 100 membrane is only about 6% of
that through bare porous alumina. Using a feed solution of 1000 ppm CW 1000
ppm 8042', the flux decreased to about 5 m3m'2d'1, and there was no selectivity

between Cl' and 8042'. The flux decline suggests that the salt solutions may

45

modify the membrane because there should be a minimal effect of osmotic
pressure as rejections of both of the ions were less than 10%.

After investigating the properties of the bare polymer support, I began
preparing MPFs on these materials. Initially a gutter layer of 5 bilayers was
employed, rather than 3 bilayers of PSS/PDADMAC because I was concerned
about fully covering the support. On top of the 5 bilayers, 1.5 bilayers of
PSS/PAH was deposited, with the final layer of PSS being deposited from 2.5 M
MnClz. Table 3.4 shows the 07803“ selectivities and flux values for these
membranes in NF experiments with a feed solution containing 1000 ppm Cl'
/1000 ppm 8042' (sodium salts). The permeate flux of 1.8 m3m'2d'1 compares
well with the 2.5 M MnCIz-terminated PSS membranes on alumina. However, the
selectivity of 16 is about half of that for a similar membrane on alumina. I though
that this might be due to defects in the film, so I prepared a similar membrane
with a 7-bilayer gutter layer. The selectivity remained 16, but the flux decreased
by 11%. Deposition of an additional PAH/PSS bilayer on the capping layer also
did not achieve better selectivity, only a lower flux. Comparison of Figure 3.4 and
Figure 3.1 suggests that deposition of 7-bilayers of PSS/PDADMAC and 1.5
bilayers of PSS/PAH does fully cover PES 100. The pore structure seen in
Figure 3.1 is completely obscured by the polyelectrolyte film. Although there is
some structure seen on the surface of the coated membrane in Figure 3.4, such
patterns were previously observed with polyelectrolyte films deposited on both
solid and porous substrates.3'8'1o Full coverage of the support suggests that the

lower selectivity of films deposited on polymer supports may be due to a modified

46

Table 3.4: Percent rejection of Cl' and 8042' as well as the water flux in NF with
various membranes prepared by MPF deposition on a PES 100 support. The
calculated Cl'lSO42' selectivities are also shown. The feed solution contained

1000 ppm CH 1000 ppm 8042’ pressurized to 4.8 bar (70 psi).

 

Membrane CT 8042' water flux in Cl'/SO42'

% rejection % rejection m3m'2d'1 . selectivity
Bare PES 100 -1:1 7:1 39:19 1:0
[PSS/PDADMAC]; -8:2 68:2 29:02 3:0.1
[[Ezss’liafiwggf 2:3 93:2 1 .8:0.3 16:5
[[ESSSS/iPtAf-‘ITIPASASC‘J? 2:1 94.0:02 16:02 16:1
[Egg/[Eggiflggh 6:2 92:3 1 .6:0.2 14:6
[Egg/[Egfififlggh 2:4 93:3 13:02 16:6
PSS/PDADMAC15 -2:1 77:1 1 1 3:01 5:2

(PSS/PAH] PAA**T
The gutter layer was deposited from a 0.1 M solution of NaCl.

*The final PSS layer was deposited from a 2.5 M MnClz solution.

**The PAA layer was deposited from a 0.5 M NaCI solution.

47

 

 

500 nm
Figure 3.4: FESEM image of a [PSS/PDADMACMPSS/PAH]PSS film (final PSS

layer deposited from 2.5 M MnCIz) deposited on PESIOO. The applied potential

was 10 kV.

48

film structure that arises from the initial support charge density and/or
morphology. However, although the Cl'lSOtz' selectivity on the polymer support
is lower than that when using coated porous alumina, a selectivity of 16 may still
be acceptable in industrial applications if flux is high.

I also attempted to use a PSS/PAH/PAA cap on 5-bilayers of
PSS/PDADMAC to increase selectivity. However, these films yielded a
selectivity of only 6, as opposed to selectivities of up to 80 observed for PAA
hybrids on alumina (Table 2.1). This may again be due to a change in film
morphology.

To further assess the NF properties of MPFs on PES 100 supports,
solutions containing 1000 ppm Ca” and 1000 ppm Mg2+ were filtered. Results
with these ions were somewhat disappointing. For a
[PSS/PDADMAC]5[PSS/PAH]PSS film, the rejection of Ca2+ dropped from 79% to
59 % on going from an alumina to a PES100 support (final layer of PSS was
deposited from 0.5 M MnCl2). The Mg2+ rejection fell from 93% to 58%.

Further research is needed to better understand these results, and more
work must be done in order to obtain higher rejection with PES supports. If the
hydrophobicity of the polymer support is hindering development of selectivity,
perhaps one could UV irradiate the PES support before depositing the
polyelectrolytes. Belfort et al. showed that UV modification at 254 nm increases
surface hydrophilicity by breaking the sulfur-carbon bond in the main chain,
creating radicals which terminate with hydroxyl groups.11 They also showed that

UV irradiation helps to increase the water flux of the PES by widening the pores.

49

 

The use of such a pretreatment methods may yield thin polyelectrolyte films that
fully cover the support.

It may also be possible to increase the rejection of NF membranes by
annealing them at a slightly elevated temperature for a short period of time.
Tieke et al showed that heating polyelectrolyte membranes at 90°C for 1 hour
greatly increases their selectivity in pervaporation.2 This process may remove
residual water, allowing more ion pairs to form and causing the film to become
more dense. There is, however, a price to paid for the higher selectivity.
Annealing caused the pervaporation flux of the membrane to decrease
significantly. I performed one experiment using PES 100 coated with
[PSS/PDADMAC]7[PSS/PAH]2. The membrane was heated overnight (18 hours)
at 60°C, since heating it at higher temperatures caused the PES 100 to curl up
and become deformed. NF of a 1000 ppm Mg2+ solution by this membrane
resulted in a Mg2+ rejection of 85%. The flux, however, decreased to 0.5 m3m'2d'
1, which may be unacceptable low. It may still be possible to adjust heating

temperature and annealing times to find a situation that leads to increased

rejection and a reasonable flux.

Conclusions

This research shows that it is possible to increase the flux through MPF

films deposited on alumina by depositing an ultrathin, selective layer on a highly

permeable, nonselective gutter layer. Such gutter layers should help in the goal

50

of creating high flux NF membranes using polymeric supports. However, much
more work will need to be done to make such membranes more efficient and

industrially attractive.

51

 

201
5669.

 

 

References:

(9)

Krasemann, L.; Tieke, B. Chem. Eng. Technol. 2000, 23, 211.
Krasemann, L.; Toutianoush, A.; Tieke, B. J. Membr. Sci. 2001, 181, 221.
Dubas, S. T.; Schlenoff, J. B. Langmuir 2001, 17, 7725.

Farhat, T. R.; Schlenoff, J. B. Langmuir 2001, 17, 1184.

 

E?
Harris, J. J.; Bruening, M. L. Langmuir 2000, 16, 2006. :
Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153.
Schlenoff, J. B.; Ly, H.; Li, M. J. Am. Chem. Soc. 1998, 120, 7626. '-

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

Heyde, M. E.; Peters, C. R.; Anderson, J. E. J. Colloid Interface Sci 1975,

50, 467.

(10)

Mendelsohn, J. D.; Barrett, C. J.; Chan, V. V.; Pal, A. J.; Mayes, A. M.;

Rubner, M. F. Langmuir 2000, 16, 5017.

(11)

Kilduff, J. E.; Supatpong, M.; Pieracci, J. P.; Belfort, G. Desalination 2000,

132, 133.

52

20.
5889

 

 

 

Chapter 4

Fouling and Repair of NF Membranes (Future Work)

Membrane fouling is a common problem encountered when performing
NF with natural feed solutions. Fouling leads to inefficient membrane processes
by both decreasing the permeation rate and diminishing the rejection of ions. In
order to completely understand how MPF NF membranes will perform in practical
applications, one must study their fouling properties. In this chapter, I will
discuss preliminary measurements of membrane fouling, as well as a possible
method for regeneration of MPFs.

Chapter 1 of this thesis described the typical events that lead to a fouled
membrane. Essentially all fouling events occur at the surface of the membrane
and result in the formation of a cake layer (Figure 1.2). Rinsing the membrane
with a cleaning solution and then placing it back into operation can reverse many
types of fouling, but typically, the membrane does not completely regain its
original flux.1 Irreversible fouling significantly reduces the efficiency of membrane
processes.

Using MPFs as NF membranes may allow for complete repair of fouled
membranes. In my envisioned regeneration, the polyelectrolyte at the surface of
the membrane would be removed along with the cake layer. Subsequent
adsorption of a new surface layer could restore the original performance of the
membrane. Recent work shows that exposure of PAH/PSS films to pH 10 buffer

does result in a decrease in film thickness, presumably because of the loss of

53

 

20.
5665‘

 

 

 

polyelectrolyte layers.2 Additionally, Schlenoff et al. describe using high
concentrations of salt to delaminate MPFs.3 Such studies suggest that it may be
possible, using the right combinations of pH and ionic strength, to remove the
surface polyelectrolyte layer of a from the membrane, and presumably this would
also remove the fouling layer as well. After removing the top layer, it should be
possible to replace it with a new skin and, therefore, create a new membrane
with its original flux and rejection properties. This chapter provides evidence that
the top layer(s) of MPFs on gold wafers can be removed under appropriate
conditions. Additionally, I briefly discuss the possible application of these data to

regeneration of NF membranes.

Experimental
Materials

NazCoa (CCI), NaH003 (Mallinckrodt), 3-mercaptopropionic acid
(MPA)(Aldrich), PSS, PAH, PAA (polyelectrolytes as described previously), NaCI
(Spectrum), MnClz (Acros), NaBr (Spectrum), MgSO4 (Spectrum)
methanesulfonic acid (Mallinckrodt), and humic acid sodium salt (Aldrich) were
used as received. Buffers contained the following compositions: pH 9 buffer-
0.025 M N32CO3 + 0.25 M NaH003; pH 10 buffer- 0.025 M Na2C03 + 0.025 M
NaHCOs; pH 11 buffer- 0.25 M Na2C03 + 0.025 M NaHCOa, buffers were
adjusted to the exact pH using a pH meter by adding an insignificant (less than

one percent of the original mass) a percent amount of carbonate or bicarbonate

54

 

to raise or lower the pH. The gold-coated silicon wafers used were described in
chapter 2.
Polyelectrolyte deposition and Ellipsometry

The polyelectrolyte deposition on the gold-coated silicon wafers and
alumina supports was done as discussed in Chapter 2. Ellipsometric thickness
measurements were also done as described in previous chapters.
NF measurements

NF was performed using 20 mg/L of humic acid in a 50 ppm Mg” (from
MgSO4) feed solution. NF was run until the flux value dropped to nearly 50% of
original value, at which point, the membrane was considered fouled. Ion
Chromatography (IC) was performed to determine 8042' and Mg” concentrations
in the NF permeate solutions. The 8042' analysis was done as previously
described in chapter 2, while the Mg” was done using a Dionex CS12A cation
column, a CSRS-ULTRA suppressor, and a methanesulfonic acid eluent. The
Mg” was quantified using a 5 point calibration curve which bracketed the

expected concentration of Mg”.

External Reflectance FTIR

External reflectance Fourier Transform Infrared (ERFTIR) spectra were
collected using a Nicolet Magna-IR 560 spectrometer and a Pike grazing angle
(80° angle of incidence) attachment. Spectra were collected using 128 scans,

and a UV/ozone-cleaned gold-coated wafer was used as a background.

55

 

 

 

 

 

Pcmrelectrolvte delamination by exposure to buffer

After 5-bilayers of PAH/PSS were deposited on the surface of the gold
wafers, ellipsometry and FTIR measurements were performed. Then the
samples were placed in the appropriate buffer for the desired time, rinsed with
milIi—Q water (18 MQ/cm) for one minute, and the thickness and IR spectrum
were again obtained. Subsequently, new layers of polyelectrolytes were

deposited on the samples, and ellipsometry and ERFTIR were performed again.

Results and Discussion
Preliminary data on fouling of MPF NF membranes

To gain more understanding of how MPF membranes will perform in non-
Iaboratory settings, fouling of these materials needs to be investigated. Thus, I
performed NF with a 4.5-bilayer PSS/PAH membrane on alumina using a 50 ppm
Mg” feed solution that contained 20 mg/L of humic acid. The humic acid
simulates insoluble natural organic matter (NOM) found in natural water sources.
Both flux and ion rejection (both Mg” and 8042') were measured until the water
flux through the membrane was 50% of the initial flux (initial flux was measured
after a 1 hr equilibration period). When the flux decreased to half the original
value (68 hrs of nanofiltration), the membrane was considered fouled. The
recorded water fluxes and ion rejections are plotted in Figures 4.1, 4.2, 4.3. The
data show a steady decline in both flux and rejection over time. The rejection
decline can be partially explained by an increase in the ion concentration in the

feed solution. Over time water is leaving the system, but a high percentage of

56

 

 

 

 

Figure 4.1: Plot of water flux vs. time for the NF of a 50 ppm Mg” solution
(added as MgSO4) containing 20 mg/L of humic acid. Nanofiltration was

performed using a feed pressure of 4.8 bar (70 psi).

 

1:00—4 i! ,3 “,35fi37 '
g 0.80 J ' :EE'LZI -

(100-4 - —~ —— ~— .
5000

0 1 000 2000 3000
time (min)

4000

57

Figure 4.2: Plot of the Mg” concentration in the permeate when filtering a
solution of 50 ppm Mg” with 20 mg/L humic acid. Nanofiltration was performed

using a feed pressure of 4.8 bar (70 psi).

7 .
6 1 O
5 5.231?
E 4 j ' . ’lcell2‘l
a 1 l,__ __.____-
l . I
2 1 I I
1 gig "
0 L : - - _.: :_: - . __¥ - _ - M- __
0 1000 2000 3000 4000 5000
time (min)

58

 

 

 

 

 

Figure 4.3: Plot of 8042' concentration in the permeate solution vs. time when

filtering a 50 ppm Mg” (from MgSO4) solution containing 20 mg/L of humic acid.

Nanofiltration was performed using a feed pressure of 4.8 bar (70 psi).

25 ,
20 5
‘ O
E 15 - ..
a , 3 '
Q 10 "
I. U
5 i
O L : - ._.- ,__. __-. : A r
0 1000 2000 3000
time (min)

59

 

' 1:01:71?
\ {9913
4000 5000

20.
5685

 

 

 

 

the ions are being rejected. These leads to a 34% increase in the feed
concentration over the 68 h period. However, this does not account for all the ion
rejection decline because the permeate Mg” concentration increased by 200 to
350%. Thus, the fouling layer must be mostly responsible for the rejection
change. This change in rejection may be from a decrease in the charge from the
PSS caused by the adsorbed material on the surface. In the future, this
assessment of the membranes will need to be carried out with other potential
fouling agents such as silica particles or bacteria.

Vrijenhoek et al. demonstrated that the rate of membrane fouling is often
related to surface roughness of the membrane.4 Thus, AFM measurements of
the rougheness of MPFs on alumina also need to be performed. This will allow

comparison of the fouling data with that for other NF or ultrafiltration membranes.

Repairinq Fouled MPF membranes

As discussed earlier, fouling occurs primarily at the membrane surface.
To clean and repair a fouled MPF NF membrane, it may be possible to remove
the surface layer and replace it with a new layer. To investigate this concept, 5-
bilayer PAH/PSS films on gold surfaces were placed in high pH buffers for
various time periods. The high pH should cause deprotonation of PAH, and
hence loss of its positive charge. Once this occurs, the electrostatic interactions
between PAH and PSS will decrease so that the PAH can be rinsed away, as

long as there is minimal interpenetration between the PAH and PSS.

6O

 

Table 4.1 and Figure 4.4 show the results of experiments aimed at
delaminating films using high pH solutions. Table 4.1 demonstrates that films do
become thinner after exposure to the buffers, presumably due to delamination.
However, the amount of thickness lost does not necessarily increase with buffer
pH. The pH 9 buffer removed a greater thickness of the film than the pH 10
buffer. One reason for this may be the greater ionic strength of the pH 9 buffer.
Both buffers were made from Na2C03 and NaHCOa using 0.025 M Na2C03.
However, to achieve the appropriate pH value, the pH 9 buffer contained 0.25 M
NaHCOa, 10-fold more than the amount in the pH 10 buffer.

The higher salt concentration in the pH 9 buffer may cause a
polyelectrolyte film to come apart by increasing the osmotic pressurea. An
experiment performed using a pH 10 buffer with ten times the carbonate and
bicarbonate concentrations showed about 70% removal of the film in 5 min, thus
confirming the ionic strength effect. I also added 0.25 M NaCI to the original pH
10 buffer. This resulted in a loss of 20% of the original film thickness during a 5
min immersion time. The pH 10 buffer with 0.25 M NaCI yields nearly identical
results in 5 min as those seen for the pH 9 buffer in 10 min of immersion time.

Since the goal in repairing a fouled membrane is the removal and
replacement of only the surface layer (or bilayer), conditions that result in loss of
about 20% of the thickness of a 5-bilayer film are desirable. The results in Table
4.1 show that submersion in pH 9 buffer for 10 minutes gives the desired
removal. Furthermore, re-depositing the PAH/PSS layer restores the original

thickness, and therefore, may yield a rejuvenated membrane with its original

61

 

Table 4.1: Ellipsometric thickness values (A) for 5-bilayer PAH/PSS films before
and after submersion in buffer solutions for the indicated time, and after a

PAH/PSS layer was re-deposited to see if the original thickness could be

 

obtained.
Thickness Thickness f
Thickness After % of film after ‘
Buffer Immersion b exposure removed redeposition
. . efore
pH time (min) buffer to buffer by the of a
and buffer PAH/PSS
rinsing layer

9a 2 198:8 179:16 10:6 219:7

9° 5 193:9 161:10 17:4 211:25
9a 10 195:13 159:10 18:2 195:13
9° 20 201 :2 1 56:7 23:2 1 89:3

93 30 197:1 1 142:8 28:7 185:3
10b 2 190:8 184:7 3:3 226:4
10b 5 189:11 177:10 6:1 215:17
10b 10 195:4 181:6 7:4 223:2
10b 20 193:6 178:8 8:3 216:5
10b 30 194:12 177:9 9:3 216:8

1 1° 2 197:4 139:13 29:6 182:19
11° 5 195:13 134:7 31:2 170:12

1 1° 1 0 200:4 1 30:3 35:2 1 67:4

1 1c 20 194:15 121 :5 38:3 168:6

1 1° 30 205:15 128:7 38:1 176:7

a pH 9 buffer contained 0.025 M NazC03 + 0-25 M Ncho3

b pH 10 buffer contained 0.025 M Na2003 + 0.025 M NaHCOa

° pH 11 buffer contained 0.25 M Na2C03 + 0.025 M NaH003

62

 

properties.

FTIR spectra (Figure 4.4) also show that absorbance bands return to their
original levels after immersion of a film in pH 9 buffer and adsorption of a new
PAH/PSS bilayer. (See table 4.2 for peak intensities) This provides further
evidence for the removal and replacement of the surface layer.

Submerging the sample in a pH 9 buffer solution for longer times (20 to 30
min), probably starts removal of a second layer. In this case, the re-deposition
step yields a film that is thinner than the original. Using pH 11 buffer appears to
remove two bilayers almost immediately, as shown by the ellipsometric thickness
and FTIR measurements. This buffer may be too harsh for the goal of removing
the surface. Further research also needs to be done on films capped with PAA.
Immersion in pH 9 buffer for 2 min resulted in the removal of one bilayer from a
PAA-capped film. Longer periods of time resulted in at least 2 bilayers being
removed from the surface. The time has to be monitored carefully in this case,
since the PAA layer appears to be removed very quickly in high pH buffer. This
likely allows the buffer solution to permeate the film more quickly and remove
more layers.

The next step in this research will be to remove the surface layer from a
membrane in the NF setup. The experiment that is planned for this case
involves flowing a buffer solution across the membrane in the cell for a set
amount of time, followed by a rinse and subsequent re-deposition steps. The
difficult part will be determining the amount of time to flow the buffer across the

membrane to obtain the desired results. This will be especially challenging

63

 

:3

AwEov 20383552,
003. com? oovw oomN oovm

 

 

 

 

 

SOUBCIJOSQB

(:3 6:8 6%.
Eton 980m

sr/ 1.3:wa .6

5368900.. Lmt<

 

 

 

 

 

5.52, coo___w “0068-200 m :0 wEE wmnzIdi U5 25:0on m_._.n_mm ”1v 8:9...—

64

20
54,65

 

 

 

Table 4.2: Absorbances (abs) for 3 main peaks in the spectra in Figure 4.4.

 

Peak Abs before buffer Abs after buffer Abs after re-
deposition
1180 cm'1 0.0117 0.0085 0.0116
1220 cm'1 0.0116 0.0083 0.0115
2920 cm" 0.0069 0.006 0.0072

65

 

because of the difficulty of characterizing membranes on porous supports. I
attempted to use Brewster angle FTIR to examine layer removal from alumina,
but the alumina support absorbs below the 1250 cm'1 region, masking the
strongest bands (sulfonates) of PSS/PAH films. Therefore, the timing of the
removal and re-deposition steps may have to be done empirically by measuring

flux and rejection both before fouling and after re-deposition.

Conclusion

Preliminary measurements show that MPF membranes foul significantly in
the presence of humic acids. However, MPF NF membranes may offer a very
effective method for membrane regeneration. Experiments with films on gold-
coated wafers show that it is possible to remove the surface layer of the MPF
using high pH buffer solutions. The removed layers can be replaced by re-
immersing the wafers in the PAH and PSS deposition solutions. Future research

will involve regeneration of fouled membranes in NF cells.

66

 

References:

(1) Kilduff, J. E.; Mattaraj, S.; Sensibaugh, J.; Pieracci, J. P.; Yuan, Y.; Belfort,
G. Environmental Engineering Science 2002, 19, 477.

(2) Harris, J. J.; Bruening, M. L. Langmuir2000, 16,2006.
(3) Dubas, S. T.; Schlenoff, J. B. Macromolecules 2001, 34, 3736.

(4) Vrijenhoek, E. M.; Hong, S.; Elimelech, M. J. Membr. Sci. 2001, 188, 115.

67

20
ea

 

 

 

l11111111111111111111