Www.netl.Doe.gov Kmd Cds Disk23 D-Water Management Projects Produced Water%5C15547%5C15547-FG083108

7/31/2019 Www.netl.Doe.gov Kmd Cds Disk23 D-Water Management Projects Produced Water%5C15547%5C15547-FG083108

1/46

Novel Fouling-Reducing Coatings for

Ultrafiltration, Nanofiltration, and Reverse

Osmosis Membranes

Final Scientific/Technical Report

Reporting Period Start Date: 1 September 2004

Reporting Period End Date: 31 August 2007

Principal Author: Dr. Benny D. Freeman

Date of Report: 29 November 2007

DOE Award Number: DE-FC26-04NT15547

Submitting Organization: The University of Texas at Austin

Center for Energy & Environmental Resources

and Department of Chemical Engineering

10100 Burnet Road, Building 133

Austin, TX 78758


2/46


3/46


4/46

TABLE OF CONTENTS

EXECUTIVE SUMMARY ......................................................................................1EXPERIMENTAL METHODS ...............................................................................3

Task 1: Emulsion Selection, Preparation, and Characterization.......................3

Task 2: Selection of Commercial Membrane Supports ....................................4Task 3: Synthesis and Characterization of Fouling-Resistant Materials...........4Task 4: Preparation and Characterization of Coated and Surface-ModifiedMembranes .......................................................................................................7Task 5: Characterization of Fouling and Separation Performance ...................9

RESULTS AND DISCUSSION ...........................................................................11Task 1: Emulsion Characterization.................................................................11Task 2: Selection of Commercial Membrane Supports ..................................12Task 3: Synthesis and Characterization of Fouling Resistant Materials.........13Task 4: Preparation and Characterization of Coated and Surface-ModifiedMembranes .....................................................................................................18

Task 5: Characterization of Fouling and Separation Performance .................25CONCLUSION....................................................................................................36GRAPHICAL MATERIALS..................................................................................37REFERENCES ...................................................................................................39

ACRONYMS AND ABBREVIATIONS.................................................................40APPENDIX: Report of produced water analysis.................................................41


5/46


6/46

carefully controlled in order to measure accurate, reliable values of water flux andNaCl rejection in accordance with the manufacturers specifications. Feed pHand the use of prefiltration of the feed water were found to be critical variables inmembrane testing. The LE/XLE and AG membranes must be tested underdifferent conditions (pH 8 with unprefiltered feed and pH 7 with prefiltered feed,

respectively) to obtain their best performance. Concentration polarization alsomust be accounted for to find the true salt rejection capabilities of themembranes.

Before applying coatings to membranes, thorough characterization of the coatingmaterials was performed. Three series of PEG-based copolymers weresystematically studied to relate chemical composition and structure to polymerproperties such as water and NaCl permeability. Acrylic acid, 2-hydroxyethylacrylate, or poly(ethylene glycol) acrylate were each copolymerized withpoly(ethylene glycol) diacrylate to form a highly hydrophilic, crosslinked hydrogel.

All of the copolymers exhibited large water uptake, with the PEGA copolymers

having the largest uptake amounts. Water permeability was directly proportionalto the water uptake; higher uptake materials also had higher water permeability,regardless of chemical composition. Ethylene oxide content and crosslinkdensity were major contributors to water sorption and transport behavior.Increasing ethylene oxide content and decreasing crosslink density bothpromoted increased water transport. NaCl diffusion and partition coefficientswere also measured. The salt transport properties were similar to the watertransport properties; higher water uptake materials also had a larger salt uptake,and high water permeability lead to large diffusion coefficients and high saltpermeability. Contact angle measurements confirmed the hydrophilic nature ofthe copolymer surfaces. All copolymer contact angles were less than the contactangle of a commercial RO membrane, indicating that the copolymer surfaceswould be less conducive to oil adhesion.

Preparation of composite membranes using the synthesized hydrogels proveddifficult. Progress was made in elucidating optimal coating conditions andtechniques. A basic model was applied to use composite water flux to gaugecoating thickness. SEM images confirmed the viability of this model. Initialcrossflow fouling tests showed the coated membrane to foul less than anuncoated membrane.

Grafting of hydrophilic PEG molecules was performed by reaction of membranesurface amine groups with the epoxide endgroups of poly(ethylene glycol)diglycidyl ether (PEG diepoxide). XLE membranes were dip coated in heatedsolutions of PEG diepoxide for 10 minutes to ensure reaction with the membranesurface. Dead end water flux testing revealed the effect of PEG diepoxidemolecular weight and concentration on surface coverage and pure water flux(more surface coverage, lower pure water flux). Crossflow testing of high waterflux candidates demonstrated that PEG diepoxide-grafted XLEs have betterfouling resistance than their unmodified counterparts (LE and XLE membranes).

2


7/46

EXPERIMENTAL METHODS

Task 1: Emulsion Selection, Preparation, and Characterization

Two produced water emulsion samples were obtained and analyzed in concertwith the U.S. Bureau of Reclamation, also interested produced water membranefouling studies. Two produced water samples from the southwest region of theUnited States were obtained, one from a natural gas well and one from an oilwell. The samples were analyzed by Crystal Solutions (Laramie, WY) using EPAand AWWA standards.

Model emulsions for fouling studies consisted of 9 parts oil (n-decane, orn-dodecane) to one part surfactant. Dow Corning Fluid 193,dodecyltrimethylammonium bromide (DTAB), and sodium dodecyl sulfate (SDS)were chosen as representative nonionic, cationic, and anionic surfactants,

respectively (cf., Figure 1). Emulsion preparation consisted of mixing the oil,surfactant, and deionized water (Millipore MilliQ, 18 M, 1.2 ppb) in a Waringblender on high speed for 3 minutes.

N+

CH3

CH3

H3C

Br-

dodecyltrimethylammonium bromide (DTAB)

O

S

O

O-

Osodium dodecyl sulfate (SLS)

Na+

Figure 1. Chemical structures of surfactants.

Si

CH3

CH3

O CH2CH2O

x y

Dow Corning Fluid 193 (DCF 193)

Emulsion droplet size distribution and stability were estimated using a Carl Zeiss

AxioSkop Optical Microscope. Droplet sizes were measured using AxioVision LEsoftware. Latex particles of known diameter were used to calibrate themicroscope. Droplet number average diameter, dN, was calculated using theequation:

=i

ii

NN

dNd (1)

3


8/46

Droplet volume average diameter, dv, was calculated using the equation:

=3

4

ii

ii

vdN

dNd (2)

The polydispersity index, PDI, an indication of droplet size distribution, wascalculated from dN and dv:

N

v

d

dPDI = (3)

Task 2: Selection of Commercial Membrane Supports

Flat-sheet AG RO membranes manufactured by GE Infrastructure Water Processand Technologies were used for characterization and modification studies. The

AG membrane is a commercial brackish water desalination RO membrane (50L/(m2h) (LMH) water flux at 225 psig, 98-99% NaCl rejection)1 and is a polyamidethin film composite membrane representative of the vast majority of commercialRO and NF membranes available today. The generic structure of the polyamidelayer, which provides salt rejection in this membrane, is:

C N

O

H

C N

O

H

NC N

O H

C

O

H

C O

x y

Flat-sheet membranes from Dow FilmTec were also obtained for use incharacterization and modification studies. They include the LE (low energy, 49LMH flux at 150 psi, 99.0-99.3% NaCl rejection)2 and XLE (extra low energy, 67LMH at 150 psi, 98.0-99.0% NaCl rejection)3 reverse osmosis membranes.These membranes all share the same general chemical structure shown above.The XLE membrane will be used for modification studies, while the LE will beused as a control for comparison with the modified XLE membranes. Surfacemodification decreases water flux, so comparing a control membrane with a purewater flux similar to that of the modified XLE membrane will provide a better

gauge of the fouling resistance of the modified membranes.

Task 3: Synthesis and Characterization of Fouling-ResistantMaterials

UV-polymerized hydrogels were synthesized as potential fouling-resistantcoatings. The crosslinking agent, poly(ethylene glycol) diacrylate (PEGDA), was

4


9/46

obtained from Sigma-Aldrich (Milwaukee, WI) and used as received. Themonomers, poly(ethylene glycol) acrylate (PEGA), 2-hydroxyethyl acrylate(HEA), and acrylic acid (AA) (cf., Figure 2), were obtained from Sigma-Aldrichand used without further purification. These three monomers were chosen tosystematically study the effects of comonomer chain length, i.e., the number of

ethylene oxide (EO) units, on material properties. Each monomer has an acrylicendgroup and a hydroxyl endgroup. However, AA does not contain any EO,HEA contains one EO unit, and PEGA contains seven units. The photoinitiatorwas 1-hydroxycyclohexyl phenyl ketone (HPK), also obtained from Sigma-

Aldrich.

H2C CH

OCH2CH213

C

O

C

O

CH

CH2O

Poly(ethylene glycol) diacrylate (PEGDA)

H2C CH

OCH2CH27

C

O

OH

Poly(ethylene glycol) acrylate (PEGA)

H2C CH

OCH2CH2C

O

OH

2-Hydroxyethyl acrylate (HEA)

H2C CH

C

O

OH

Acrylic Acid (AA)

Figure 2. Chemical structures of materials used.

Prepolymerization mixtures were prepared by combining desired amounts ofcrosslinker and monomer with a specified amount of deionized (DI) water and 1.0wt% (based on solids content) of photoinitiator. The crosslinker, monomer, andphotoinitiator were first mixed in an amber glass jar and stirred with a magneticstir bar for approximately one hour until the photoinitiator dissolved. An amberglass jar was used to minimize the solution exposure to light. After thephotoinitiator dissolved, the appropriate amount of DI water was added, and thesolution was stirred for an additional hour before use. For the materials preparedin this study, the water content in the prepolymerization mixture was 60 wt%,based on total mixture weight. The films were named based on monomercontent and monomer type. For example, a 60PEGA film was polymerized froman initial mixture containing 60 mole% PEGA and 40 mole% PEGDA, with 60wt% water in the prepolymerization mixture.

Free-standing dense films were prepared by first placing the prepolymerizationmixture between two quartz plates, using spacers to control film thickness. Then,

the mixture was exposed to 312 nm wavelength UV-light for 90 s at 3000 W/cm2in a Fisher-Scientific UV-Crosslinking chamber (Pittsburgh, PA). Films wereremoved from the quartz plates, rinsed, and soaked in DI water until used. Solfractions in the hydrogels were found to be negligible after a 5 day extraction inwater. This behavior is supported by previous work on similar copolymermaterials

4. Therefore, no further extraction was performed apart from rinsing and

storing in DI water.

5


10/46

Hydrogel water sorption was measured. Free-standing hydrogel films wereequilibrated in DI water for a minimum of one hour. The films were then patteddry with ChemWipe tissues and weighed using a balance with milligramprecision. After weighing, the samples were dried under vacuum overnight and

then weighed again. The water volume fraction, vs, is calculated by:

polymer

dry

OH

drywet

OH

drywet

OH mmm

mm

v

2

2

2

+

= (4)

where mwet is the mass of the wet film, mdry is the mass of the dry film, H2O is thedensity of water, and polymeris the polymer film density.

Hydrogel water flux was measured using a dead-end filtration test stand operatedat different low transmembrane pressures. Advantec MFS, Inc. UHP 43

(diameter 43mm) dead-end stirred filtration cells (Dublin, CA) were used atpressures ranging from 1.7 to 4.5 bar (10-50 psig) at 25C. Permeate volume, V,

as a function of time, t, was recorded, and water flux, , was calculated as

follows:

OH2J

A

1

t

VJ OH2

= (5)

whereA is active membrane area. Then, water flux was used to calculate water

permeability, :OH2P

p

JP

OH

OH2

2 =

l(6)

where is film thickness, and p is the applied transmembrane pressuredifference. Ultrapure water from a Millipore MilliQ system (18.2 M, 1.2 ppb)was used in all experiments. Film thickness was measured using a Mitutoyo

Absolute micrometer (Model ID-C112E, Mitutoyo USA, Aurora, IL) followingguidelines in ISO 9339-2

l

5. Water flux was measured at a minimum of threepressures, and a plot was made of water permeability as a function of appliedpressure. Average water permeability was calculated by averaging the individualpermeabilities.

Salt transport in hydrogel films was evaluated using kinetic salt desorption

experiments

6,7

. A hydrogel film was immersed in 50 cm

3

of 5 wt% NaCldissolved in DI water (i.e., the so-called donor solution). After 24 hours, the filmwas removed from the donor solution and quickly patted dry. The film was thenplaced in a beaker containing 50 cm3 of DI water, and the beaker was sealedwith Parafilm to minimize CO2 absorption and changes in solution conductivitythat accompany CO2 absorption. The solution was stirred at approximately 300rpm to ensure a constant salt concentration throughout the liquid in the beaker.The conductivity of the solution was measured as a function of time at 25C

6


11/46


12/46

before spinning. Then, membranes were blotted dry and taped to a siliconwafer, ensuring the membrane was completely flat. Three mL ofprepolymerization mixture were added to the membrane. Varying amounts ofhigh molecular weight PEO were added to the prepolymerization mixture tochange solution viscosity and to prevent pore penetration. Samples were spun

at a given speed for a specified time after a 5 s acceleration period. The filmswere crosslinked in a nitrogen-purged chamber using UV light at 312 nm, 3000

W/cm2, for 90 s.

In addition to creating a dense coating layer, modification by grafting was alsoinvestigated. Surface modification was performed on commercial Dow FilmTecXLE RO membranes. Before modifications were made, membrane sampleswere soaked in 25%(v) isopropanol for twenty minutes, then rinsed in deionizedwater, to remove glycerin from the membrane pores. Poly(ethylene glycol)diglycidyl ether, commonly referred to as PEG diepoxide, was the main focus ofchemical modification studies (cf., Figure 3). Membranes were dip coated in

aqueous solutions of PEG diepoxide, where several PEG diepoxide chainlengths (n values in Figure 3, n = 5, 9, 14 and 23), and concentrations (0-0.2%(mol)) were used10.

The dip coating procedure began with heating deionized water to 40oC using aBarnstead Electrothermal heating unit with stirrer. The appropriate amount ofPEG diepoxide, allowed to equilibrate to room temperature, was then added tothe water. The stirrer was turned off and the membrane was immediatelyimmersed in the solution, to reduce the possible reaction time of PEG diepoxideand water. The membrane was left in the solution (keeping the temperatureconstant) for ten minutes. Finally, the membrane was removed from solution,triply rinsed in deionized water to remove unreacted PEG diepoxide, and storedin deionized water until use.

OCH2CH2H2CO OCH2 CH CH2

O

HCH2C

On

Figure 3. Chemical structure of poly(ethylene glycol) diglycidyl ether (PEG diepoxide) (n 5,9, 14, or 23).

Attenuated total reflectance Fourier Transform Infrared Spectroscopy (ATR-

FTIR) was performed on surface modified membranes using a Thermo NicoletNexus 470 FTIR (Madison, WI) with an Avatar Smart MIRacle ATR accessory(Zinc Selenide crystal). Data were collected and analyzed using Omnic software.Spectra were collected using 128 scans at resolution of 4 cm -1 between 600 and4000 cm-1. Prior to analysis, samples were placed under vacuum overnight toremove excess water.

8


13/46

X-ray photoelectron spectroscopy (XPS) was used to detect membrane surfacegrafting. Samples are dried in a vacuum oven overnight before analysis. A PHI5700 XPS (Physical Electronics, Chanhassen, MN) equipped with amonochromatic Al K1,2 X-ray source performs a surface scan to detect carbon,nitrogen, and oxygen. Operating conditions are: 1x10-9 Torr chamber pressure;

14 kV; 250 W for the Al X-ray source. Steps are not taken to prevent themembrane from collecting charge; this creates a shift in XPS spectra. ThereforeXPS spectra only provide relative amounts of C, N, and O on the surface and donot give insight into bond type.

Characterization of coated and grafted membrane water permeabilities used highpressure dead-end filtration. These measurements were done using SterlitechHP4750 stainless steel high pressure cells (Sterlitech Corporation, Kent, WA) at200 psig. Additionally, a flux resistance model was used to evaluate coatingthickness:

)( pLJ pTOH2 = (9)

where is the measured water flux, and p is the applied pressure. LOH2J pT isgiven as:

+=

wPEGpRO

pTPL

1L

l(10)

where LpRO is the RO membrane permeance, lis the coating thickness, and

PwPEG is the hydrogel water permeability, previously measured for free-standingfilms.

Atomic Force Microscopy (AFM) was used to examine surface topology of neatand modified membranes. A Digital Instruments Dimension 3100 with aNanoscope IV controller was used in tapping mode to image the surface andcollect roughness data. Silicon tips for 300 kHz were used. Samples were patdry before testing.

Scanning electron microscopy (SEM) was also done on composite membranecross-sections to evaluate coating thickness. Samples were freeze-fractured inliquid nitrogen and mounted to an SEM stage. The samples were sputter coatedfor 20 s using an Au target. Images were taken on a LEO 1530 SEM (Carl ZeissSMT Inc) at 10 kV.

Task 5: Characterization of Fouling and Separation Performance

Baseline water flux and NaCl rejection testing of commercial RO membraneswas conducted in crossflow filtration using an industry-standard 2000 ppm NaClfeed, prepared using deionized water from the Millipore system, and the optimumpressure and flowrate for each material, as specified by the manufacturer.Membranes were supplied on a roll, and several rotations of material werediscarded before taking samples for testing. Feed pH was adjusted by addition

9


14/46

of NaHCO3, HCl, or NaOH. The feed was either prefiltered using an activatedcarbon + particle filter, or unprefiltered, depending on the test conditions of themanufacturer.

Concentration polarization was studied in unmodified GE AG RO and Dow

FilmTec XLE membranes. When membranes are subjected to feed solutionscontaining solutes such as salts, a boundary layer develops near the membranesurface in which the solute (e.g. NaCl) concentration (cso(m)) is significantly higherthan in the bulk feed solution (cso)

11. Figure 4 is an illustration of the boundarylayer formed near the membrane surface.

Boundarylayer

Bulk solution

Me

mbrane

Direction

of flow

Permeate

cso(m)

cso

csl

Figure 4. Boundary layer formation near the membrane surface.

Since it is impossible to measure concentration at the membrane surface (cso(m)),the bulk concentration (cso) is used to calculate observed rejection. Theconcentration polarization modulus, M, is a factor which can be used to correct

the observed performance values obtained using bulk concentration, therebyobtaining the inherent material properties of the membrane, independent ofpolarization effects caused by the fluid dynamics of the system.

The concentration polarization level in our crossflow system was determinedusing a model based on the fact that the water permeability of a membrane is amaterial property and therefore independent of the feed12. The concentrationpolarization modulus, M, is given by this model as12:

( ) ( )

( )

sso m w NaCl

so s so s w pw

JpM

J

= =

l

l l

1 (11)

The osmotic pressure at the membrane surface, so(m), is given by:

10


15/46

( )

( )w NaCl

sso m

w(pw)

Jp 1

J

= +

l (12)

where p is the applied transmembrane pressure, so(m), s, and so are theosmotic pressures at the membrane surface, in the permeate, and in the bulkfeed, respectively, Jw(NaCl) is the water flux in 2000 ppm NaCl feed, and Jw(pw) isthe water flux in pure water feed. To determine the polarization levelexperimentally, pure water flux is measured (Jw(pw)), then 2000 ppm NaCl isadded to the feed and water flux (Jw(NaCl)) and bulk and permeate concentrationsare measured. After the concentrations are converted to osmotic pressures (so,and s), the polarization modulus and membrane surface salt concentration canbe calculated. The actual salt rejection of a membrane is given by:

( )

sactual

so m

R 1 100%

=

l

(13)

where s and so(m) are the salt concentrations in the permeate and at themembrane surface, respectively. The observed salt rejection is calculated usingthe bulk concentration (so) instead of the membrane surface concentration(so(m)), and is always lower than the actual salt rejection. Thus, the modelallows simple calculation of the surface concentration, polarization modulus, andactual salt rejection from easily measured experimental quantities.

Fouling studies were also conducted in crossflow filtration mode as described

above. Model foulants and emulsions, described above, were used, andrejection of organic carbon was measured using a Shimadzu TOC 5050A TotalOrganic Carbon Analyzer.

RESULTS AND DISCUSSION

Task 1: Emulsion Characterization

Two produced water emulsion samples from the southwest region of the UnitedStates were obtained and analyzed, one from a natural gas well and one from an

oil well. A detailed analysis can be found in Appendix I. As expected, bothsamples contained a large amount of total dissolved solids. The oil well sampleTDS was 230,000 mg/L, and the natural gas sample TDS was 16,000 mg/L. Thelargest component in each of these was sodium chloride. Calcium carbonateand sulfate were also found in significant amounts in each sample. Surprisingly,the natural gas sample had a higher total organic carbon count, 1650 mg/L, thanthe oil well, 106 mg/L, although the oil well sample contained more aromatichydrocarbons such as benzene, toluene and phenol. The analysis also showed

11


16/46


17/46

Task 3: Synthesis and Characterization of Fouling ResistantMaterials

Free-standing PEG hydrogel films were successfully synthesized. ATR-FTIRwas used to gauge reaction conversion (data not shown here). Reaction of all

copolymers was assumed to be complete.

Water transport properties of free standing PEG films were evaluated todetermine the suitability of these materials as membrane coatings. Figure 6apresents water sorption as a function of comonomer content for each copolymerseries. PEGA copolymers exhibited the largest sorbed volume fraction of water,and HEA copolymers sorbed more water than a pure PEGDA hydrogel. Thewater volume fraction in AA copolymers was slightly higher than that of PEGDA,but changed little with comonomer content.

The observed water sorption behavior is presumably derived from the monomer

content and crosslink density of each copolymer. Crosslink density decreases asmonomer content increases. Decreasing crosslink density could increase wateruptake by giving the polymer network more freedom to swell in water. However,swelling will occur only within the confines of the water solubility limit in thepolymer. In this regard, PEO content is expected to influence water uptakebecause water has a high affinity for ethylene oxide units13. Even high molecularweight PEO is highly soluble in water at room temperature14. Figure 6b showsthe water volume fraction as a function of PEO weight percent in eachcopolymer. PEGDA and PEGA are both approximately 82 wt% PEO, so PEOcontent remains relatively constant for the entire copolymer series. Therefore,the increase in water volume fraction can be solely attributed to the decrease in

crosslink density. Conversely, AA does not contain any PEO, and HEA is 38wt%PEO, so PEO content decreases as comonomer content in AA and HEAcopolymers increases. This decrease in PEO content changes thehydrophilic/hydrophobic balance of the material. Higher comonomer contentresults in less PEO, and therefore, a more hydrophobic material. However,despite their decreasing hydrophilicity, AA and HEA copolymer water volumefractions remain relatively constant over a range of PEO contents. One wouldexpect that for a given crosslink density, a lower PEO content would lead tolower water sorption. This assumption is supported by comparing HEAcopolymer sorption to AA sorption at each comonomer composition (cf., Figure6a). However, since further decrease of PEO content does not decrease the

sorption, the decrease in crosslink density must be offsetting the increasinghydrophobicity as monomer content increases.

13


18/46


19/46

Figure 7b presents a correlation of water permeability with water volume fraction.In general, higher hydrogel water volume fraction leads to higher hydrogel waterpermeability, regardless of chemical composition.

NaCl uptake and transport properties of the free-standing PEG films were also

measured.As shown in Figure 8, NaCl desorption behavior is largely consistentwith that expected from Fickian diffusion models of solute release from a film ofuniform thickness7. A short induction period was observed for many of thesamples, indicated by the non-linear relationship between mass uptake and thesquare root of time at the beginning of the experiment. This induction period maybe partly attributed to delays in data collection, since the equipment used in thisstudy could only collect data in 5 s intervals. Additionally, the induction periodmight be due to boundary layer effects at the beginning of the experiment, beforethe solution in which the sample was soaking became well-mixed. Such effectsare known in the literature, and they were assumed to be small in this study.

The NaCl diffusion coefficient, Ds, for each copolymer was calculated from theslope of the linear portion of the desorption curve using Eq. 7, and the results arepresented in Figure 9. NaCl diffusivity does not change significantly withchanging AA or HEA content, but it does increase slightly with increasing PEGAcontent. Also, the difference in diffusivity between most of the copolymers is notsignificant, except for 80mole% copolymers, where 80PEGA is larger than 80AA.Ds of NaCl in pure water at 25C is approximately 1.6 x 10

-5 cm2/s16. All of themeasured values fall below this value. This result was expected because eventhough NaCl is in an aqueous solution, both NaCl and water must be transportedthrough the hydrogel. Therefore, measured Ds values less than Ds of NaCl inpure water indicate that the NaCl is passing through the hydrogel network andnot through any film defects.

0

0.2

0.4

0.6

0.8

1

1.2

0 100 200 300 400 500 600

Mt/

Minf

t1/2

/L

2x10-6

4x10-6

6x10-6

8x10-6

1x10-5

0 20 40 60 80 10

NaClDiffusionCoefficient,D

s(cm

2/s)

Mole % Comonomer

AA

PEGA

HEA

0

Figure 8.Example salt desorption curve usedto calculate NaCl diffusion coefficient in ahydrogel.

Figure 9. Diffusion coefficients of NaCl inhydrogel copolymers.

NaCl partition coefficients, Ks, were also determined from the desorptionmeasurements, and the results are presented in Figure 10a. Trends in Ks with

15


20/46

varying monomer content are reminiscent of the trends in water sorption forthese materials. Like their water sorption behavior, AA and HEA copolymersshow relatively little change in NaCl uptake with increasing comonomer content.However, PEGA copolymers show salt uptake increasing with increasingcomonomer content, similar to its water sorption behavior. These results are

reasonable since, to a first approximation, the amount of salt sorbed by thepolymer network is often found to be sensitive to the water uptake. To determineif the NaCl concentration of the water sorbed by the hydrogel is the same as theNaCl concentration in the surrounding water, the partition coefficient wasmodeled as follows7:

OHOHpolymerpolymers 22KKK vv += (14)

where Ks is the measured NaCl partition coefficient, Kpolymerand are the

partition coefficients of the polymer and water, respectively, and v

OH2K

polymerand

are the volume fractions of polymer and water, respectively, in the polymer. IfNaCl is not sorbed by the polymer, i.e., if the polymer is impenetrable to NaCl,

then K

OH2v

polymerwould be zero. Also, is, by definition, one. In this case, Eq. 14reduces to

OH2K7:

OHs 2K v= (15)

Therefore, the measured NaCl partition coefficients are plotted versus . As

shown in Figure 10b, all of the measured partition coefficients fall below the linegiven by Eq. 15, so the hydrogels sorb more water than NaCl. These resultsindicate that the materials exhibit some solubility selectivity for water over NaCl.

OH2v

0.35

0.40

0.45

0.50

0.55

0 20 40 60 80 1

NaClPartitionCoefficient,K

s

Mole % Comonomer

AA

PEGA

HEA

(a)

00

0.3

0.4

0.5

0.6

0.7

0.60 0.65 0.70 0.75

NaClPartitionCoefficient,K

Water Volume Fraction, Vs

K = Vs

PEGA

HEA

AA

PEGDA

(b)

Figure 10. NaCl Partition coefficients as a function of(a) copolymer composition and (b) watervolume fraction.

16


21/46

1x10-6

2x10-6

3x10-6

4x10-6

5x10-6

0 20 40 60 80 10

SaltPerme

ability,P(cm

2/s)

Mole % Comonomer

PEGA

AA

HEA

Finally, NaCl permeability coefficientswere calculated using the measured Dsand Ks values along with Eq. 8. Theresults are presented in Figure 11. Asseen with both the salt uptake and

diffusion behavior, AA and HEA copolymerNaCl permeabilities are relatively constantfor all comonomer contents. PEGAcopolymer NaCl permeability increaseswith increasing comonomer content.

Contact angles of the PEG-based filmsand a GE AG membrane were measuredusing a pendant drop technique. Thesamples were immersed in DI water, andn-Decane was used as the probe liquid. The

reported angle was measured through theaqueous phase, so angles of 90 and greaterrepresent hydrophobic surfaces, and anglesless than 90 represent hydrophilic surfaces.The data are presented in Table 1. There isnot a discernable trend relating contact angleto copolymer content or monomer type.Experimental error was large in some cases,which is a typical challenge of contact anglemeasurements17. Most notably, the majorityof the contact angles were significantly lessthan the contact angle reported for the AGmembrane. This result indicates that thecopolymers are more hydrophilic than the ROmembrane, and that oil has less affinity forhydrogel than for RO membrane surfaces.

0

Figure 11. NaCl permeability as afunction of copolymer composition.

Table 1. Oil-in-water pendant drop

contact angles of hydrogel filmsSample Contact Angle (

0)

PEGDA 49 2

20AA40AA60AA80AA

56 462 754 432 3

20HEA40HEA60HEA80HEA

46 356 851 149 4

20PEGA40PEGA

60PEGA80PEGA

48 140 3

37 344 4

GE AG 67 3

After hydrophilicity, surface charge has also been shown to play a key role inmembrane fouling. Zeta potential measurements are a means of quantifyingsurface charge and were done on a representative series of PEG hydrogel films.The measurements were taken using hydrogel films that were fully hydrated in DIwater before testing. As Figure 12a demonstrates, all of the PEG films exhibitedsimilar zeta potential values, and their zeta potential values appeared to remainrelatively constant over the entire pH range. Only one of the films exhibited anisoelectric point. An isoelectric point (IEP) is the pH where the zeta potential iszero. These results suggest that the films do not contain groups that are readilyor significantly ionized.

17


22/46

-10

-5

0

5

2 3 4 5 6 7 8 9

PEGDA

AA 50 mol

HEA 50 mol

PEGA 50 mol

ZetaPotential,,(mV)

pH

(a)

-30

-25

-20

-15

-10

-5

0

5

10

2 3 4 5 6 7 8 9 10

ZetaPotential,,(mV)

pH

Dow XLE

GE AG

(b)

Figure 12. Zeta potential measurements of(a) PEG-based hydrogel filmsand (b) commercialRO membranes.

Zeta potential measurements were also conducted on GE AG and Dow XLE ROmembranes. As seen in Figure 12b, both membranes showed similar isoelectricpoints. The GE AG membrane appeared to possess a slightly larger negativecharge than the Dow XLE membrane at higher pH values, but overall there wasnot a significant difference between the two membranes. This similarity wasexpected due to their identical surface chemistries. Comparing the zetapotentials of commercial membranes to those of the PEG-based hydrogels,PEG-based hydrogels are significantly less negatively charged than the two ROmembranes. This result indicates that a PEG-coated membrane could be lesslikely to foul than an uncoated membrane when subjected to charged foulants

Task 4: Preparation and Characterization of Coated andSurface-Modified Membranes

Preparing composite membranes with a dense PEG layer coating proved to bemore difficult than anticipated. Initially, coatings were made using a slot coatingtechnique, where a bar is drawn down across the membrane, spreading theprepolymerization mixture as it moves. However, these coatings were oftenuneven, and pore penetration caused drastic declines in water flux. It was alsoextremely difficult to produce coatings less than 40 microns thick.

Surface characterization was done on these initial composite membranes. AFMwas used to image the coating surface and to determine the effects of coatingson surface topology, i.e., surface roughness. Figures 13a and b show AFMimages of an uncoated and coated membrane, respectively. The differencesbetween the two surfaces are striking; distinct rough features can be seen on theuncoated image, but the coated image is smooth. Figures 14a and b showheight profiles across a section of each sample surface. Again, a stark contrastexists between the rough features of the uncoated membrane and thesmoothness of the coated one.

18


23/46

(a) (b)Figure 13. AFM height images of(a) neat AG RO membrane and (b) PEG-coated AG ROmembrane.

(a)igure 14. Profile of surface roughness for(a) AG RO membrane and (b) PEG-coated AG ROembrane.

lthough the significant change in surface topology was promising, otherroblems persisted with the slot coating technique. Therefore, spin coating wasied as an alternative to slot coating. First, the effect of spinning on the

was investigated. The water permeability of a water-but not coate ith a polymerization mixture, fell far

and causes the pores to collapse. Accordingly, spin time waser

oncan

APEO content

(b)Fm

Aptrmembrane performancesoaked membrane spun, d wbelow manufacturer-specified values, suggesting that spinning dries the

embranemreduced, and water was added to the membrane during the spin. The watpermeability of these samples agreed with specified values.

Next, the effect of PEO content in the prepolymer mixture was examined. High

molecular weight PEO (Mn = 1,000,000) was added to change the prepolymersolution viscosity because previous work had shown that a more viscous solutiprevented pore penetration during the coating process18. Pore penetrationdecrease the composite membrane water flux, and therefore should beprevented if possible. Samples were spun at 3000 rpm for 10 s, using PEGD

ith 2, 0.5 or 0 wt% PEO. Figure 15a shows that increasingwdecreases the water flux of the composite membrane, indicating that porepenetration is not a problem in spin coating. This behavior could have two

19


24/46

possible explanations: (1) higher viscosity leads to a more uniform coating or (higher viscosity leads to a thicker film. From this analysis, 2 wt% PEO waseliminated as a testing condition because of the low water flux samples.

Next, the effect of spin speed on coating thickness was tested. PEGDA solution

with either 0.5 wt% PEO or 0 wt% PEO were spun for 15 s at either 1000, 2or 3000 rpm. Figure 15b shows that 1000 rpm produces composite membrwith a lower flux, indicating a thicker coating layer. Samples spun at 2000 a3000 rpm showed similar water flux values. Based on these results, 1000 rpmwas also eliminated as a spinning variable.

2)

s

000,anesnd

Finally, the effect of spin time on membrane flux was studied. Time is animportant variable because the solution must be spun long enough to spread intoa thin layer, but not too long to evaporate water from the prepolymer solution, or

e, both of which could cause a drastic decrease in water

15 s, especially the samples coated

to dry out the membranflux. PEGDA samples with either 0.5 wt% or 0 wt% PEO were spun at 2000 rpmfor either 10, 15, or 20 s. As Figure 15c demonstrates, samples spun for 20 shad lower fluxes than those spun for 10 or

0

10

20

30

40

rFlux

at200psi(LMH)

50

0

5

10

30

Water 0 wt PEO 0.5 wt PEO 2 wt PEO

Wate

(a)

15

20

25

Flux

at200psig(LMH)

0 wt% PEO

0.5 wt%PEO

1000 rpm 2000 rpm 3000 rpm

Water

Spin Speed

(b)

0

5

10

15

20

25

30

35

10 s 15 s 20 s

WaterFluxat200psig(LMH)

Spin Time (s)

0 wt% PEO

0.5 wt% PEO

(c)

Figure 15. Composite membrane water flux as a function of(a) PEO content at 10 s and 3000rpm, (b) spin speed with a spin time of 15s, and (c)spin time at a spin speed of 2000 rpm. Allsamples were coated with PEGDA in 60wt% H2O.

20


25/46

with 0.5 wt% PEO in the prepolymer solution. Therefore, 20 s of spin time was

g

ngFigure 16a is a plot of water flux as a function of coating

ickness, calculated using the water permeability of a PEGDA film with 60 wt%gure

the least effective condition, and was no longer considered as a viable spincoating condition.

Although this work did not result in one optimal set of spinning conditions, a

better understanding of the effects of PEO content, time, and speed were gained.Based on these results, 0 wt PEO, 2000 rpm and 10 s were the chosen spinninconditions for future tests.

After finding satisfactory spin conditions, focus was shifted towards quantifyicoating thickness.thwater, and the resistance model given by Eqs. 9 and 10. Using the plot in Fi16a and the highest measured water fluxes from Figures 15a, b, and c, most ofthe coating thicknesses are on the order of 3 microns.

0

10

50

20

30

40

terFluxat200psig(L/m2hr)

0 10 20 30 40 50

Wa

Coating Thickness (microns)

Figure 16. (a) Model prediction of water flux as aSEM image of a composite coated membrane. CoatiTo corroborate the model calculations, SEMthe coating thickness. A sample is shown in Figure 16b.prepared using PEGDA with 0 wt% PEO andcoating thickness, measured using tmicrons. The calculated coating thi

xide

owater

times. For example, XLE membranes dip coated with a 10%(w) solution of MW

function of PEGDA coating thickness. (b)ng thickness is approximately 3 microns.

images were also taken to estimateThis sample was

was spun for 10 s at 2000 rpm; thehe SEM software, was approximately 3ckness based on the measured water flux

was also approximately 3 microns. The agreement between these twotechniques confirms the ability of the model to predict coating thicknesses.

In addition to surface coating with a dense layer, surface modification by grafting

was also studied. Direct chemical surface modification studies focused onpoly(ethylene glycol) diglycidyl ether, or PEG diepoxide, as the grafting molecule.Dip coating has been the method of choice for surface grafting of PEG diepoto the XLE membrane surface. The extent of PEG diepoxide grafting appears tbe directly proportional to the reaction time, as evidenced by lower pureflux (i.e., more PEG surface coverage) of membranes dip coated for longer

200 PEG diepoxide at 40oC for 1 minute and 10 minutes had pure water fluxes of

21


26/46

75 and 45 L/(m2h), respectively.

Several chain lengths of PEG diepoxide (n = 5, 9, 14 and 23, or molecularweights = 200, 400, 600 and 1000) have been grafted to the XLE membranesurface, using a reaction time of ten minutes. The molecular weight and

concentration of PEG diepoxide strongly influence the resulting pure water flthe modified XLE membrane, as seen in Figure 17. Figure 17 gives the purewater flux as a function of PEG diepoxide concentration. One interesting resultseen in Figure 17 is that MW 400

ux of

PEG diepoxide gives grafted membranes withwer water flux than either MW 600 or 1000, which could indicate a change of

de

loattachment method (e.g., reaction at one chain end to form a brush versusreaction at both chain ends to form a loop) as the chain length of PEG diepoxiincreases.

0

20

40

60

80

100

rflux(L/(m

2h)

)

control XLE flux = 100 LMH

MW200

0 2 4 6 8 10 12 14 16

Purewate

PEG diepoxide concentration (wt%)

MW400

MW1000

MW600

Figure 17. Pure water flux (dead end filtration, p = 150 psig) of XLE membranes dip coated(40oC, 10 minute reaction) with increasing concentrations of PEG diepoxide (molecular weights200, 400, 600 and 1000).

The pure water flux data shown in Figure 17 suggest that low concentrations ofPEG diepoxide will maximize water flux of the modified membranes. XLEmembranes were then dip coated with several chain lengths (MW 200, 600, and1000) and low concentrations (0.04-0.16 %(w)) of PEG diepoxide. Fluxes of the

o

iepoxide chain length and concentration increased, fluxecreased. The data presented in Figure 18b indicate that the PEG diepoxide-

treated membranes and three controls (heated to 40 C in water) were measuredusing a 2000 ppm NaCl solution in a crossflow filtration test; results are given inFigure 18a. As PEG ddgrafted membranes and the control membranes had comparable NaCl rejectioncapabilities.

22


27/46

25

30

35

40

45

50(a)

98.6

98.8

99

99.2

99.4

99.6

)

average control NaClrejection = 99.09 +/- 0.02%

(b)

0.04 0.06 0.08 0.1 0.12 0.14 0.16

Waterflux

(L/(m

h))


MW 200

MW 600

MW 1000

average control flux =66.5 +/- 1.7 LMH

2

0.04 0.06 0.08 0.1 0.12 0.14 0.16

ObservedNaC

lrejection(%


MW 200

MW 600

MW 1000

Figure 18. (a) Water flux and (b) NaCl rejection of FilmTec XLE membranes grafted (40oC, 10minute reaction) with PEG diepoxides of different chain length (MW 200, 600, and 1000) andconcentrations.

Contact angle analysis was perform E AG RO membranes dip coated

and the membrane, or a more hydrophilic membrane surface. Results indicate

ed on Gwith increasing concentrations of PEG diepoxide (MW 400). The results aregiven in Figure 19. The contact angle of a n-decane drop in water wasmeasured; a smaller contact angle indicates less contact between the oil droplet

that the most drastic decrease in contact angle, or increase in hydrophilicity,occurred for the lowest PEG diepoxide concentrations (0-2 vol%). Thisobservation corroborates the decision to use low concentrations of PEGdiepoxide for grafting.

20

30

40

50

60

70

o)

0 10 20 30 40 50

ContactAngle(

PEG diepoxide Concentration (vol%)

Figure 19. Decane in water contact angles for GE AG RO membranesdip coated with PEGdiepoxide (n = 9, 40

oC, 10 minute reaction).

to verify the presence of PEG diepox e on the membranesurface after grafting to XLE membranes using X-ray photoelectron spectroscopy

Attempts were made id

23


28/46


29/46

50010001500200025003000350040000

0.2

0.4

0.6

0.8

1

Absorbance

Wavenumber (cm-1

)

2870cm-1

(alkaneC-C)

1100 cm-1

(ether C-O-C)

Figure 20. FTIR spectrum of MW 200 (n=4-5) PEG diepoxide.

-0.05

0

0.05

0.1

0.15

0.2

0.25

10001050110011501200

Absorbancediffe

rence

(Samplespectrum-XL

Espectrum)

Wavenumber (cm-1)

0.04%(w)

0.5%(w)

1.0%(w)

10%(w)

Figure 21.FTIR spectra of MW 200 PEGdiepoxide-grafted XLEs subtracted from anunmodified XLE.

Task 5: Characterization of Fouling and Separation Performance

Before testing coated and grafted RO membranes, the water flux and NaClrejection of unmodified RO membranes had to be determined. Obtaining waterflux and NaCl rejection values (in industry-standard 2000 ppm NaCl) ofcommercial RO membranes in agreement with the manufacturers specificationsproved difficult. Systematic adjustments were made to the crossflow apparatus

and to the operating procedures to bring tested values into agreement with themanufacturers values.

The first major roadblock to obtaining manufacturer-specified membraneperformance parameters was the inability to maintain constant flux in thecrossflow filtration test system. This problem was solved by the addition of acarbon + particle prefilter, which prevents biofouling due to naturally occurringbiological growth in the feed tank over time, as well as particulate fouling due tofine contaminants present in the feed water. As seen in Figure 22, the flux ofXLE membranes in pure water and 2000 ppm NaCl feeds was steady overseveral days when the feed was continuously run through the prefilter. At the

end of each run, 20 ppm dodecyltrimethylammonium bromide (DTAB) was addedto the feed to measure the performance of XLE membranes subjected to thisharsh foulant. In both runs, membrane flux decreased by 75% after addition ofthe foulant. Thus, the same steep initial flux decline as previously reported on

AG membranes exposed to DTAB was also noted with XLE membranes.

25


30/46

0

20

40

60

80

100

120

0 50 100 150 200

Averageflux(L/(m

2h))

Permeation time (hrs)

2000 ppmNaCl added

20 ppmDTAB added

cell 3

cell 2

pure water

pH = 7.5

avg NaClrej. = 99.2%avg TOCrej. = 93.0%

pH = 7.5averageNaCl rej.= 97.2%

(a)

20

40

60

80

100

120

0 10 20 30 40 50 60 70 80

Averageflux(L/(m

2h))


cell 1

cell 2 cell 3

pure water2000 ppmNaCl added

20 ppmDTAB added

pH = 6.5 pH = 7.0averageNaCl rej.= 96.7%

pH = 7.0avg NaClrej= 99.0%

avg TOCrej= 91.6%

(b)

Figure 22. Crossflow filtration performance of unmodified XLE membranes. Cell numbers referto order in which feed flows through membrane test cells. In run (a), only two cells were used in

series, while in run (b), three cells were used in series.

Although the membrane flux was steady after installation of the carbon + particlepre-filter, the target flux and rejection values for XLE membranes still were notachieved. Vertical bars in Figure 22 indicate the expected flux range for XLEmembranes subjected to pure water and 2000 ppm NaCl feeds. Experimentalfluxes were slightly high, and observed NaCl rejection values were lower thanexpected (expected observed rejection is >98%), which suggested that leaksmay be allowing a small portion of feed to bypass the membranes and mix withthe RO permeate. However, leak testing using 1000 ppm MgSO4 feed and XLEmembranes (which should completely reject this salt) showed MgSO4 rejection

greater than 99.5%, indicating that leaks were not the cause of the high flux andlow rejection. Thus, other possible explanations were sought for this behavior.

One variable which was thought to potentially affect membrane performance wasthe feed pH. The effect of feed pH on observed NaCl rejection of XLE and GE

AG membranes was determined by measuring the rejection at several pH values(pH was adjusted with dilute HCl or NaOH). The pH was alternated betweenacidic and basic values to ensure that any observed changes in rejection weredue to the feed pH and not to damage caused to the membranes by acidifying orbasifying the feed. Water flux was also monitored to determine the effect of feedpH on both membrane performance parameters. Figure 23a shows that

observed NaCl rejection is a strong function of feed pH in the range 3-10.Rejection decreases significantly if the feed pH is lower than the pKa of thecarboxylic acid and amine groups on the membrane surface (pKa~5). Figure 23bfocuses on a narrower pH range (5-9), and demonstrates that observed rejectionincreases linearly with feed pH in this range, while water flux is independent offeed pH. Thus, feed pH appears to be a feasible cause of the lower thanexpected observed NaCl rejection values. Figure 23c presents flux and rejectiondata as a function of pH for the GE AG membrane. In general, the trends are the

26


31/46


32/46

A series of experiments was then performed to determine the effects of feed pHand prefiltration on membrane performance. Pure water flux was measuredbefore adding the 2000 ppm NaCl, so that the observed rejection values could becorrected for concentration polarization using the model discussed in theexperimental section. The water flux in pure water and 2000 ppm NaCl, along

with the bulk permeate NaCl concentration, were used in Eqs. 12 and 13 to findthe actual NaCl rejection. Figure 24a shows the effect of feed pH (6.0 vs. 7.9) onthe water flux and actual NaCl rejection of XLE membranes, using unprefilteredfeed water for both experiments. Although water flux is fairly unaffected by feedpH, rejection is much higher at the higher feed pH value. Figure 24b shows theeffect of prefiltration (at pH 7.8-7.9) on XLE performance. Significant differencesin flux and rejection are seen for unprefiltered versus prefiltered feeds.Prefiltered feed water leads to high, nearly constant water flux and only a slightincrease in NaCl rejection over 24 hours. Membranes subjected to unprefilteredfeeds show lower water flux and higher NaCl rejection than those subjected tocontinuously prefiltered feeds. In addition, membranes tested in unprefiltered

feed show a much stronger dependence of flux and rejection on time, as fluxdeclines steadily over 24 hours, and rejection increases steadily (3x moreincrease in rejection (0.6% vs. 0.2%) for membranes tested in unprefiltered feed).Thus, prefiltration of the feed could explain both higher water flux and lower NaClrejection.

60

65

70

75

80

98

98.5

99

99.5

0 5 10 15 20 25

Waterflux

(L/(m

2h))


ActualNaC

lrejection(%)

filled symbols: pH 6.0unfilled symbols: pH 7.9

flux

Ractual

(a)

60

65

70

75

80

85

98.2

98.4

98.6

98.8

99

99.2

99.4

99.6

0 5 10 15 20 25

filled symbols:prefiltered

unfilled symbols:

unprefiltered

Waterflux

(L/(m

2h))


ActualNaC

lrejection(%)

flux

Ractual

(b)

Figure 24. Water flux and NaCl rejection of XLE membranes versus time (2000 ppm NaCl, p =150 psig, feed flowrate = 1.0 gpm, 25

oC), demonstrating the effects of(a) feed pH (unprefiltered

feed) and (b) prefiltration of the feed (pH 7.8-7.9) on membrane performance.

Once the effects of these two important variables, feed pH and prefiltration, weredetermined, the precise conditions under which the manufacturer obtains itsspecified performance values were used to test the membranes. Themanufacturer of the XLE and LE membranes tests their membranes at pH 8using unprefiltered feed water. Figures 25a and b show the water flux and NaClrejection (observed and actual) of the LE and XLE membranes, respectively,during the course of a 24 hour crossflow experiment run at 150 psig and a

28


33/46


34/46

The effects of feed flowrate on polarization modulus and actual (true) NaClrejection of the XLE and AG membranes (at their manufacturer-specified optimalpressures, 150 psig for the XLE and 225 psig for the AG) were also determinedusing the concentration polarization model discussed above. Feed wascontinuously filtered through the carbon/particle prefilter to ensure the

polarization was not affected by membrane fouling. Figure 26a shows theobserved salt rejection, polarization modulus, and actual salt rejection as afunction of flowrate for the XLE membrane. The observed salt rejectionincreases from 97.2 to 98.3% and the polarization modulus decreases from 1.65to 1.2 as flowrate is increased from 0.3 to 1.0 gallons per minute. As flowrateincreases, the feed is circulated across the membranes faster so less salt buildsup at the membrane surface (i.e., lower polarization modulus). Also, since saltflux is proportional to salt concentration, lower surface concentration correspondsto less salt transport through the membrane, or higher salt rejection. Similarbehavior was observed for the AG membranes, as shown in Figure 26b. Bothexperiments showed similar concentration polarization modulus values, further

supporting the assumption that modulus values depend more on the fluiddynamics of the experimental system than on the membrane itself.

1

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

97.5

98

98.5

99

99.5

0.2 0.4 0.6 0.8 1 1.2 1.4

CPModulus(M=cm

/cb

) NaClRejection(%)

Flowrate (gpm)

Rtrue

Robserved

Modulus

(b)

Figure 26.Observed salt rejection, concentration polarization modulus, and actual salt rejectionat several flowrates for(a) the XLE membrane at 150 psig and (b)the AG membrane at 225 psig.

1

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

97

97.5

98

98.5

99

0.2 0.4 0.6 0.8 1 1.2

CPModulus(M=cm

/cb

)

Flowrate (gpm)

NaClrejection(%)

Rtrue

Robserved

(a)

After being able to thoroughly understand the baseline behavior of unmodified,commercial RO membranes, focus turned to evaluating water flux, NaCl

rejection, and fouling behavior of grafted and coated membranes. Severalcrossflow fouling studies were performed on XLE and AG membranes graftedand coated with various PEG compositions. The fouling performance of XLEmembranes grafted with MW 600 PEG diepoxide was studied and compared totwo control membranes: the XLE (heated to 40oC in water) and LE (not heated)membranes. The lower flux membrane (LE) was chosen because its flux in 2000ppm NaCl solution is comparable to the flux exhibited in the previous experiment(Figure 18a) by the MW 600 PEG diepoxide-treated XLE membranes.

30


35/46

Therefore, a better comparison could be made of their fouling performance. Twosamples of each control membrane were tested. Two concentrations of MW 600PEG diepoxide were used in the fouling study: 0.04 %(w) and 0.12%(w). Threemembranes were made with each concentration. Also, an attempt was made totreat only the top surface of two 0.12%(w) PEG diepoxide-grafted membranes,

since all previous treatments had been done by submerging the membrane in thePEG diepoxide solution. The flux was measured for a 2000 ppm NaCl solution,then a 25 ppm dodecane/25 ppm SLS emulsion was added, and the flux wasmonitored over the next two hours. Results are illustrated in Figure 27a.

Although the initial fluxes of the treated XLE membranes were only half that ofthe control XLE membrane, the control membrane showed a significant amountof fouling, while the treated membranes did not experience any fouling. Thetreated XLE membranes also showed better fouling resistance than the controlLE membranes chosen for their comparable initial flux. Salt rejection was alsomonitored as a function of time, as seen in Figure 27b. The PEG diepoxide-treated XLE membranes had the highest rejections. Interestingly, the flux and

rejection of the top surface-treated membranes were almost identical to those ofthe submerged membranes (0.12 wt% MW 600 PEG diepoxide), indicating thatisolation of the membrane surface for grafting may be unnecessary.

25

30

35

40

45

50

55

60

0 0.5 1 1.5 2 2.5

Averageflux(LMH)


LE

XLE

MW 600 0.04 wt% (XLE)

MW 600 0.12 wt% (XLE)

MW 600 0.12 wt%top surface (XLE)

(a)

97

97.5

98

98.5

99

99.5

100

0 0.5 1 1.5 2 2.

AverageNaClrejection(%)

Permeation time (hours)

LE

XLE

n = 600 0.04%(w) (XLE)

n = 600 0.12%(w) (XLE)

n = 600 0.12%(w) top surface (XLE)

5

(b)

Figure 27. (a)Average flux and (b) average NaCl rejection of PEG diepoxide-treated XLEand control LE and XLE membranes in 2000 ppm NaCl plus 25 ppm dodecane/25 ppm SLSemulsion.

Figure 28a shows the water flux versus time for XLE membranes grafted with0.04 and 1.0%(w) MW 200 PEG diepoxide. It was thought that using a shorterchain length PEG diepoxide than the experiment shown in Figure 27 wouldcause the modified XLE membrane to have higher water flux. The test wasconducted at 150 psig and a flowrate of 0.85 gpm. The membranes were firsttested in pure water feed for approximately 20 hours, then 2000 ppm NaCl wasadded and flux was monitored for an additional 20 hours. Finally, 20 ppm of

31


36/46

DTAB foulant was added to the feed to monitor the fouling resistance of themodified membranes. The feed pH was 7 for both the NaCl and NaCl + DTABfeeds. The flux profile for a control XLE membrane tested previously under thesame conditions is shown in Figure 28 for comparison. While the flux of themodified membranes did not surpass that of the unmodified membrane in the

fouling feed, one of the modified membranes did have slightly better foulingresistance than the unmodified membrane. Figure 28b gives the flux of eachmembrane in the fouling feed normalized to the flux in 2000 ppm NaCl, whicheliminates flux differences attributed to the different levels of mass transferresistance of modified and unmodified membranes. From this perspective it isclear that the 1.0%(w) PEG diepoxide-grafted XLE retains more of its initial flux inthe fouling feed than the unmodified XLE. All membranes, modified andunmodified, had observed salt rejections better than 98%.

20

30

40

50

60

70

80

90

0 10 20 30 40 50 60 70

0.04%(w) MW 200PEG diepoxide XLE


Averageflux(L/(m

2h))


XLE

20 ppm DTAB

addedpure water

2000 ppmNaCl

Figure 28a. Water flux as a function of timefor two MW200 PEG diepoxide-grafted XLEmembranes and one unmodified XLE.

0.5

0.6

0.7

0.8

0.9

1

40 45 50 55 60 65 70

Normalizedflux


0.04%(w) MW 200

PEG diepoxide XLE


XLE

Figure 28b. Flux oftwo MW200 PEGdiepoxide-grafted XLE membranes and oneunmodified XLE in 2000 ppm NaCl + 20 ppmDTAB feed, normalized to their respective waterfluxes in 2000 ppm NaCl feed.

LE membranes were also used as controls for comparison to PEG diepoxide-grafted XLE membranes. The LE membrane is very similar to the XLEmembrane, but with lower flux than the XLE. The extent of fouling a membraneexperiences is likely dependent on the amount of water it processes. Therefore,comparing modified XLE membranes, whose flux is lower because of theadditional mass transfer resistance provided by the grafted molecules, to LE

32


37/46

membranes could eliminate differences in water throughput between themodified and control membranes. The water flux properties of a PEG diepoxide-grafted XLE (dip coated in a 40oC solution of 0.04%(w) MW 200 PEG diepoxidefor 10 minutes) and an unmodified LE membrane are compared in Figure 29a.Both membranes have similar water fluxes in pure water and 2000 ppm NaCl

feeds, so the LE is a good choice of control membrane for comparison to thePEG diepoxide-grafted XLE. The membranes fouling behavior when 20 ppmDTAB is added to the feed is also shown, indicating the modified XLE has higherflux than the control LE. The flux decline experienced by each of the membranesis another means of comparison of their fouling resistance. In Figure 29b, theflux after addition of DTAB is normalized by the average flux observed in 2000ppm NaCl feed. This method of comparison clearly shows that the PEGdiepoxide-grafted XLE retains approximately 5% more of its flux and experiencesless fouling than the control LE membrane.

30

40

50

60

70

80

90

0 5 10 15 20 25

0.04%(w)MW 200 PEGdiepoxide XLE

LE


Waterflux(L/(m

2h))

purewater

2000 ppmNaCl added 20 ppm DTAB

added

pH7.1

pH7.6

pH 7.6(a)

0.5

0.6

0.7

0.8

0.9

1

5 10 15 20

0.04%(w)MW 200 PEGdiepoxide XLE

LE

pH 7.6

20 ppm DTABadded

Normalizedflux(flux/2000ppmNaClflux)


25

(b)

Figure 29. (a) Water flux of a PEG diepoxide-grafted XLE and control LE in pure water, 2000ppm NaCl, and 2000 ppm NaCl + 20 ppm DTAB feeds (pure waterand 2000 ppm NaClfeeds runthrough carbon/particle prefilter, p = 150 psig, feed flowrate = 1.0 gpm, 25

oC); (b) Fouling

resistance of the two membranes upon addition of DTAB to the feed (where flux is normalized toaverage flux in 2000 ppm NaCl).

Table 4 gives the observed NaCl rejection values for the two membranes in the2000 ppm NaCl and 2000 ppm NaCl + 20 ppm DTAB feeds introduced during thecourse of the experiment depicted in Figure 29. The control LE and PEGdiepoxide-grafted XLE membranes have similar salt rejection properties,indicating grafting has no negative effect on rejection. Additionally, bothmembranes show a marked increase in NaCl rejection upon addition of DTAB,which is likely caused by DTAB blockage of the membrane surface, preventing

33


38/46

NaCl from reaching and diffusing through the membrane. Organic rejection of100% was measured for both the control LE and PEG diepoxide-grafted XLEmembranes.

Table 4. Observed NaCl rejection values of control LE and PEG diepoxide-grafted XLEmembranes in 2000 ppm NaCl and 2000 ppm NaCl + 20 ppm DTAB feeds, corresponding to theexperiment shown in Figure 29.

Observed NaCl rejection (%)

2000 ppm NaCl feed 20 ppm DTAB + 2000 ppmNaCl feed

LE 98.8 99.5

PEG-grafted XLE 98.7 99.3

Crossflow fouling studies of PEG diepoxide-grafted GE AG RO membranes werealso conducted to enable comparison of the two surface modification methods,

grafting and coating, for the same base membrane. Figure 30a shows the waterflux versus time for an unmodified AG membrane (heated to 40oC for bettercomparison to the modified membranes, which are treated at 40oC) and AGmembranes grafted with 0.04 and 10%(w) MW 200 PEG diepoxide. One high(10%(w)) concentration-grafted membrane was tested in this fouling experimentto determine whether extensive surface coverage provides significantly improvedfouling resistance which overcomes the negative effect on water flux. The testwas conducted at 225 psig (optimum pressure for AG membrane operation) anda flowrate of 1.0 gpm. The membranes were first tested in pure water feed forapproximately 4 hours, then 2000 ppm NaCl was added and flux was monitoredfor an additional 4 hours. Finally, 20 ppm of DTAB foulant was added to the feed

to monitor the fouling resistance of the three membranes. The feed pH wasbetween 7 and 9 for both the NaCl and NaCl + DTAB feeds. Similar to theresults of the experiments discussed previously, the flux of the modified AGmembranes did not surpass that of the unmodified membrane in the fouling feed,but the modified membranes again had slightly better fouling resistance than theunmodified membrane. Figure 30b gives the flux of each membrane in thefouling feed normalized to the flux in 2000 ppm NaCl, which clearly shows thatboth modified membranes had better fouling resistance than the unmodifiedmembrane, with higher concentration PEG diepoxide leading to higher foulingresistance. However, the flux of the 10%(w) PEG diepoxide-grafted membranewas too low to be useful, even with its slightly higher fouling resistance, so low

concentrations still appear to be the most promising for optimizing both water fluxand fouling resistance.

34


39/46

10

20

30

40

50

60

70

80

0 5 10 15 20 25 30

Averageflux(L/(m

2h))


AG (40oC)

0.04%(w) MW 200PEG diepoxide AG

10%(w) MW 200PEG diepoxide AG

pure water2000 ppm NaCl 20 ppm

DTAB added

Figure 30a. Water flux as a function oftime for two MW200 PEG diepoxide-grafted

AG membranes and one unmodified AG.

0.4

0.5

0.6

0.7

0.8

0.9

1

5 10 15 20 25 3


Normalizedflu

x

AG (40oC)

0.04%(w) MW 200PEG diepoxide AG

10%(w) MW 200PEG diepoxide AG

0

Figure 30b. Flux oftwo MW200 PEGdiepoxide-grafted AG membranes and oneunmodified AG in 2000 ppm NaCl + 20 ppmDTAB feed, normalized to their respective waterfluxes in 2000 ppm NaCl feed.

Due to the difficulty in obtaining reproducible dense layer coatings, few foulingtests were performed on coated membranes during the duration of this project. APEG-coated membrane was compared to a water spun control membrane incrossflow filtration (cf., Figure 31). Initially, the PEG-coated membrane water flux

was significantly lower than the control membrane water flux. However, afteraddition of DTAB, an anionic surfactant, the control water flux fell slightly belowthe PEG-coated water flux, proving the PEG-coated membrane to be slightlymore resistant to fouling. NaCl rejection for both samples was >98% and thecoating thickness was calculated to be approximately 2.3 microns. These resultsare encouraging for several reasons. First, PEGDA water permeability is lowerthan all the copolymer water permeabilities. Therefore, coatings made with thecopolymers should produce higher fluxes.

35


40/46


41/46

GRAPHICAL MATERIALS

TablesTable 1. Oil-in-water pendant drop contact angles of hydrogel filmsTable 2. XPS elemental analysis of XLE membranes grafted with MW 200 PEG diepoxide.

Table 3. Comparison of membrane performance values obtained in our laboratory tomanufacturers specifications at same testing conditions.

Table 4. Observed NaCl rejection values of control LE and PEG diepoxide-grafted XLEmembranes in 2000 ppm NaCl and 2000 ppm NaCl + 20 ppm DTAB feeds, corresponding to theexperiment shown in Figure 29.

FiguresFigure 1. Chemical structures of surfactants.Figure 2. Chemical structures of materials used.Figure 3. Chemical structure of poly(ethylene glycol) diglycidyl ether (PEG diepoxide) (n 5,9, 14, or 23).

Figure 4. Boundary layer formation near the membrane surface.Figure 5. Microscopic image of oil emulsion.Figure 6. Hydrogel water volume fraction as a function of (a) copolymer composition and (b)PEO content.

Figure 7.Copolymer water permeability as a function of(a) copolymer composition and (b)water volume fraction.

Figure 8.Example salt desorption curve used to calculate NaCl diffusion coefficient in ahydrogel.

Figure 9. Diffusion coefficients of NaCl in hydrogel copolymers.Figure 10. NaCl Partition coefficients as a function of(a) copolymer composition and (b) watervolume fraction.

Figure 11. NaCl permeability as a function of copolymer composition.

Figure 12. Zeta potential measurements of(a) PEG-based hydrogel filmsand (b) commercialRO membranes.Figure 13. AFM height images of(a) neat AG RO membrane and (b) PEG-coated AG ROmembrane.

Figure 14. Profile of surface roughness for(a) AG RO membrane and (b) PEG-coated AG ROmembrane.

Figure 15. Composite membrane water flux as a function of(a) PEO content at 10 s and 3000rpm, (b) spin speed with a spin time of 15s, and (c)spin time at a spin speed of 2000 rpm. Allsamples were coated with PEGDA in 60wt% H2O.

Figure 16. (a) Model prediction of water flux as a function of PEGDA coating thickness. (b)SEM image of a composite coated membrane.

Figure 17. Pure water flux (dead end filtration, p = 150 psig) of XLE membranes dip coated

(40

o

C, 10 minute reaction) with increasing concentrations of PEG diepoxide (molecular weights200, 400, 600 and 1000).

Figure 18. (a) Water flux and (b) NaCl rejection of FilmTec XLE membranes grafted (40oC, 10minute reaction) with PEG diepoxides of different chain length (MW 200, 600, and 1000) andconcentrations.

Figure 19. Decane in water contact angles for GE AG RO membranesdip coated with PEGdiepoxide (n = 9, 40

oC, 10 minute reaction).

Figure 20. FTIR spectrum of MW 200 (n=4-5) PEG diepoxide.

37


42/46

Figure 21.FTIR spectra of MW 200 PEG diepoxide-grafted XLEs subtracted from anunmodified XLE.

Figure 22. Crossflow filtration performance of unmodified XLE membranes. Cell numbers referto order in which feed flows through membrane test cells. In run (a), only two cells were used inseries, while in run (b), three cells were used in series.Figure 23. (a) Observed NaCl rejection as a function of feed pH in the range 3-10 and (b)water flux and observed NaCl rejection as a function of feed pH (XLE membranes, 2000 ppmNaCl, p =150 psig, feed flowrate = 1.0 gpm, 25C, feed run through carbon/particle prefilter).(c)Observed NaCl rejection and water flux as a function of feed pH for GE AG membrane (2000ppm NaCl, p = 225 psig, feed flowrate = 0.5 gpm, 25C, feed run through carbon/particleprefilter).

Figure 24. Water flux and NaCl rejection of XLE membranes versus time (2000 ppm NaCl, p =150 psig, feed flowrate = 1.0 gpm, 25

oC), demonstrating the effects of(a) feed pH (unprefiltered

feed) and (b) prefiltration of the feed (pH 7.8-7.9) on membrane performance.Figure 25. Water flux and NaCl rejection of unmodified (a)LE and (b) XLE membranes (2000ppm NaCl, pH 7.9, unprefiltered, p = 150 psig, feed flowrate = 1.0 gpm, 25

oC).

Figure 26.Observed salt rejection, concentration polarization modulus, and actual salt rejection

at several flowrates for(a) the XLE membrane at 150 psig and (b)the AG membrane at 225psig.Figure 27. (a)Average flux and (b) average NaCl rejection of PEG diepoxide-treated XLEand control LE and XLE membranes in 2000 ppm NaCl plus 25 ppm dodecane/25 ppm SLSemulsion.Figure 28. (a) Water flux as a function of time for two MW200 PEG diepoxide-grafted XLEmembranes and one unmodified XLE. (b) Flux oftwo MW200 PEG diepoxide-grafted XLEmembranes and one unmodified XLE in 2000 ppm NaCl + 20 ppm DTAB feed, normalized totheir respective water fluxes in 2000 ppm NaCl feed.

Figure 29. (a) Water flux of a PEG diepoxide-grafted XLE and control LE in pure water, 2000ppm NaCl, and 2000 ppm NaCl + 20 ppm DTAB feeds (pure waterand 2000 ppm NaClfeeds runthrough carbon/particle prefilter, p = 150 psig, feed flowrate = 1.0 gpm, 25

oC); (b) Fouling

resistance of the two membranes upon addition of DTAB to the feed (where flux is normalized toaverage flux in 2000 ppm NaCl).Figure 30. (a) Water flux as a function of time for two MW200 PEG diepoxide-grafted AGmembranes and one unmodified AG. (b) Flux oftwo MW200 PEG diepoxide-grafted AGmembranes and one unmodified AG in 2000 ppm NaCl + 20 ppm DTAB feed, normalized to theirrespective water fluxes in 2000 ppm NaCl feed.

Figure 31. Crossflow filtration of a coated and uncoated membrane. Operating conditions:225 psig, 1.0 gpm, 25

0C.

38


43/46

REFERENCES

[1] Product Information: GE Osmonics AG Desal Membranes,http://www.desal.com, (2004).

[2] Dow FILMTEC BW30LE-440 Product Specification,http://www.dow.com/liquidseps/prod/bw30le_440.htm , (2007).

[3] Dow FILMTEC XLE-440 Product Specification,http://www.dow.com/liquidseps/prod/xle_440.htm, (2007).

[4] H. Lin, E. Van Wagner, J. S. Swinnea, B. D. Freeman, S. J. Pas, A. J. Hill,S. Kalakkunnath and D. S. Kalika, Transport and structural characteristics ofcrosslinked poly(ethylene oxide) rubbers, Journal of Membrane Science, 276(2006) 145-161.

[5] Optics and optical instruments: contact lenses-determination of thickness

Part II: Hydrogel contact lenses, ISO 9339-2, (1998).

[6] K. Nagai, et al., Solubility and diffusivity of sodium chloride in phase-separated block copolymers of poly(2-dimethylaminoethyl methacrylate),poly(1,1'-dihydroperfluorooctyl methacrylate), and poly(1,1,2,2-tetrahydroperfluorooctyl acrylate), Polymer, 42 (2001) 9941-9948.

[7] H. Yasuda, C. E. Lamaze and L. D. Ikenberry, Permeability of solutesthrough hydrated polymer membranes. Part I: Diffusion of sodium chloride, DieMakromolekulare Chemie, 118 (1968) 19-35.

[8] R. B. Bird, W. E. Stewart and E. N. Lightfoot, Transport Phenomena, 2nded., John Wiley & Sons, Inc., New York, (2002).

[9] J. Crank and G. S. Park, Methods of Measurement, in J. Crank and G. S.Park (Eds.), Diffusion in Polymers, Academic Press, Inc., New York, (1968).

[10] W. E. Mickols, Composite Membrane with Polyalkylene Oxide ModifiedPolyamide Surface., US Patent US 6,280,853 B1, (2001).

[11] R. W. Baker, Membrane Technology and Applications, 2nd ed., JohnWiley & Sons, Ltd., Chichester, (2004).

[12] I. Sutzkover, D. Hasson and R. Semiat, Simple technique for measuringthe concentration polarization level in a reverse osmosis system, Desalination,131 (2000) 117-127.

[13] N. B. Graham, Poly(ethylene oxide) and related materials, in N. A. Peppas(Ed.), Hydrogels in Medicine and Pharmacy, vol II: Polymers, CRC Press, BocaRaton, (1987), pp. 95-113.

39


44/46

[14] H. S. Ashbaugh and M. E. Paulaitis, Monomer hydrophobicity as amechanism for the LCST behavior of poly(ethylene oxide) in water, Industrial andEngineering Chemical Research, 45 (2006) 5531-5537.

[15] O. Okay, Macroporous copolymer networks, Progress in Polymer Science,

25 (2000) 711-779.

[16] C. J. D. Fell and H. P. Hutchinson, Diffusion coefficients for sodium andpotassium chlorides in water at elevated temperatures, Journal of Chemical andEngineering Data, 16 (1971) 427-429.

[17] A. W. Adamson and A. P. Gast, Physical Chemistry of Surfaces, 6th ed.,John Wiley & Sons, Inc., New York, (1997).

[18] H. Ju, B. D. McCloskey, A. C. Sagle, Y. H. Wu, V. Kusuma and B. D.Freeman, Crosslinked poly(ethylene oxide) fouling resistant coating materials foroil/water separation, Journal of Membrane Science, In Press (2007).

ACRONYMS AND ABBREVIATIONS

AA acrylic acidAFM atomic force microscopyDCF193 Dow Corning Fluid 193DTAB dodecyltrimethylammonium bromideEO ethylene oxideFTIR Fourier transform infrared spectroscopyHEA 2-hydroxyethylacrylateHPK 1-hydroxycyclohexylphenylketone

NF nanofiltrationPDI polydispersity indexPEG poly(ethylene glycol)PEG diepoxide poly(ethylene glycol) diglycidyl etherPEGA poly(ethylene glycol) acrylatePEGDA poly(ethylene glycol) diacrylatePEO poly(ethylene oxide) same as PEG, means multiple EO

unitsRO reverse osmosisSDS sodium dodecylsulfateSEM scanning electron microscopy

TOC total organic carbonUF ultrafiltrationXPS x-ray photoelectron spectroscopy

40


45/46

APPENDIX: Report of produced water analysis

Sample NG-1 is natural gas wellSample OW-1 is oil well

41


46/46

Date post:	05-Apr-2018
Category:	Documents
Upload:	ijaslyn-lek
View:	215 times
Download:	0 times

Www.netl.Doe.gov Kmd Cds Disk23 D-Water Management Projects Produced Water%5C15547%5C15547-FG083108

Documents