Silane surface modified ceramic membranes for the treatment and recycling of SAGD produced water

Accepted Manuscript Silane surface modified ceramic membranes for the treatment and recycling of SAGD produced water Charbel Atallah, André Y. Tremblay, Saviz Mortazavi PII:

S0920-4105(17)30568-5

DOI:

10.1016/j.petrol.2017.07.007

Reference:

PETROL 4093

To appear in:

Journal of Petroleum Science and Engineering

Received Date: 31 March 2017 Revised Date:

30 June 2017

Accepted Date: 4 July 2017

Please cite this article as: Atallah, C., Tremblay, André.Y., Mortazavi, S., Silane surface modified ceramic membranes for the treatment and recycling of SAGD produced water, Journal of Petroleum Science and Engineering (2017), doi: 10.1016/j.petrol.2017.07.007. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

1 2

Silane Surface Modified Ceramic Membranes for the Treatment and Recycling of SAGD Produced Water

3

Charbel Atallaha, André Y. Tremblaya,*, Saviz Mortazavib

4 5

a

6

b

7

Abstract

RI PT

Department of Chemical and Biological Engineering, University of Ottawa, 161 Louis Pasteur, Ottawa, ON, Canada K1N 6N5 CanmetMINING, Natural Resources Canada, 555 Booth St., Ottawa, ON, Canada K1A 0G1

The extraction of bitumen using oil extraction and recovery processes such as SAGD (steam assisted gravity drainage) produces oily process waters that must be treated and recycled when possible. Ceramic membranes are well suited for this task. However, ceramic membranes in aqueous media have a pH dependent surface charge. It was hypothesized that these surface charges are responsible for the high fouling of ceramic membranes in treating wastewaters containing bituminous fines. To maintain desirable hydrophilic properties without surface charges, a highly hydrophilic and neutral organosilane was used to modify the surface of ceramic membrane disks. Membranes having pore sizes of 150 kDa, 300 kDa and 0.14 µm were modified using this organosilane. The ceramic membranes were then used in the filtration of SAGD produced water. Results indicate that the modification was successful in mitigating the irreversible fouling caused by bituminous ultrafines. The permeate flux of the 150 and 300 kDa membranes more than doubled after modification in a 20% silane solution. Furthermore, the filtered water obtained from the modified membranes was of superior quality, to that of the untreated membrane, as evidenced by total organic carbon analysis. All of the ceramic membranes tested were shown to reduce the particle sizes in the produced water from >200 nm in the feed to <40 nm in the permeate.

24 25

Keywords: ceramic membrane, surface modification, organosilane, SAGD, oil sands produced water.

TE D

M AN U

SC

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

EP

26

*Corresponding author: tel: +1-613-562-5800 ext. 6108

28

e-mail: [email protected] (André Y. Tremblay)

29

1. Introduction

30

Ever since the early 20th century, petroleum has been the primary exploited resource globally

31

utilized for energy generation. A prominent issue in oil and gas exploration and production

32

processes is the production of large volumes of waste in the form of produced water. These

33

waters are produced in large quantity and must be treated before recycling or disposal.

AC C

27

1

ACCEPTED MANUSCRIPT

Canada’s oil sands, located in Alberta, represent the third largest oil reserves found on our planet.

35

In fact, an average of roughly 2.3 million barrels of crude oil per day was produced from the oil

36

sands in 2014 alone (“Alberta Official Statistics: Total Bitumen Production, Alberta,” 2015). The

37

leading technology used for bitumen extraction from the oil sands is a process known as Steam

38

Assisted Gravity Drainage (SAGD). In 2014, in-situ extraction techniques like the SAGD

39

method accounted for 55% of the total crude bitumen production rate (“Alberta Official

40

Statistics: Total Bitumen Production, Alberta,” 2015). This extraction technique consists of

41

drilling a pair of horizontal wells 4 to 6 meters apart. Steam is injected into the top well, known

42

as the injection well. The steam heats the surrounding heavy oil, thus reducing its viscosity and

43

allowing it to flow by gravity into the bottom well (producing well). This results in an emulsion

44

of oil and condensed water that is then pumped to the surface for separation and treatment.

45

Presently, about 2-3 barrels of water, injected as steam, are necessary to produce one barrel of oil

46

(Swenson et al., 2012). Furthermore, river water is often used with the SAGD process, which

47

could potentially lead to water shortages during future winters (Jordaan, 2012). Hence, the

48

treatment and recycling of this process water is an issue that requires increasing attention in

49

order to maintain the sustainability of the oil sands industry, as well as to ensure the preservation

50

of natural freshwater resources.

51

For many years, membrane separation technologies have been studied by the oil and gas industry

52

in order to remove oil and other contaminants from produced water (Allen, 2008). Currently, it is

53

widely accepted that membrane technology is the most promising candidate for the treatment of

54

oily wastewater. This is attributed to the high separation efficiency and the relatively simple

55

operational process associated with membrane modules (Padaki et al., 2015). In this application,

56

ceramic membranes are preferred for their chemical and thermal stability. They are less

57

vulnerable to the surrounding media than polymeric membranes, and are suitable for use over a

58

wide range of temperatures and pH (Sondhi et al., 2003). However, universal acceptance of

59

membranes for produced water treatment is presently hindered by the fouling observed in

60

treating these feed streams.

61

Membrane fouling results from the adhesion of non-permeating species onto the membrane

62

surface, or from the deposition of these particles into the pores of the membrane. Membrane

63

fouling obstructs the flow of the permeating species, thus resulting in a significant decline of the

AC C

EP

TE D

M AN U

SC

RI PT

34

2

ACCEPTED MANUSCRIPT

permeate flux. Moreover, the flux decline leads to an increase in transmembrane concentration

65

and pressure gradients (Allen, 2008). These problems have led to researchers developing various

66

fouling remediation techniques. These include the application of ultrasonic fields to break up

67

surface deposits, backflushing with permeate and the use of many different chemical cleaning

68

agents (Chen et al., 2006; Silalahi and Leiknes, 2009; Zhao et al., 2002). However, these

69

methods necessitate longer periods of process shutdown and inactivity in order to be

70

implemented. Naturally, this leads to an unwanted increase of the associated operational costs of

71

the process.

72

An indirect method that can also be used to reduce membrane fouling when treating produced

73

water is through surface chemistry modifications of the membranes themselves. This is done

74

through the addition of chemical compounds that adhere to the membrane surface, thus changing

75

the physicochemical properties of the selective layer (Allen, 2008). The chosen modification

76

would ideally result in a reduction of the attractive forces to the components that have been

77

identified as foulants. In the case of SAGD produced water, prominent foulants likely include a

78

multitude of both organic and inorganic substances that possess high scale forming potential.

79

Ceramic membranes have been modified using many different agents. These include metal oxide

80

nanoparticles such as alumina, titania and zirconia (Chang et al., 2014, 2010; Zhou et al., 2010),

81

grafted polymers (Castro et al., 1996; Faibish and Cohen, 2001a, 2001b) and zeolites (Cui et al.,

82

2008). All of these common surface modifications have been proven successful in forming

83

hydrophilic layers on the membrane surface. Consequently, the application of these modifying

84

agents for the filtration of oil and water emulsions improved membrane flux and mitigated any

85

irreversible fouling. Another group of low surface energy modifying agents are organosilanes.

86

Organosilanes are capable of rendering active membrane layers that possess either hydrophilic or

87

hydrophobic properties. They have been used to modify various hydroxyl group-bearing

88

surfaces, having initially been used on inorganic minerals (Kessel and Granick, 1991). They have

89

since been applied to ceramic membranes with selective layers composed of alumina, zirconia

90

and titania (Hardin et al., 2010; Krajewski et al., 2004; Kujawa and Kujawski, 2016). However,

91

the use of these silane-modified ceramic membranes has been heavily focused on gas separation

92

(Abidi et al., 2006; Bothun et al., 2007; Faiz et al., 2013; Javaid et al., 2005; Ostwal et al., 2011),

93

pervaporation (Alami Younssi et al., 2003; Kujawa et al., 2014; Kujawski et al., 2007) and

94

membrane distillation (Khemakhem and Amar, 2011; Kujawa et al., 2016, 2014; Kujawski et al.,

AC C

EP

TE D

M AN U

SC

RI PT

64

3

ACCEPTED MANUSCRIPT

2016; Larbot et al., 2004). Silane modifying agents have been sparsely explored as an application

96

for the treatment of oil and water emulsions. Moreover, in the few studies that have been

97

performed, alkyl or fluorosilanes were used in order to render the ceramic membranes

98

hydrophobic and thus remove water as an oil contaminant (Ahmad et al., 2013; Gao et al., 2016,

99

2013, 2011). Evidently, this is likely not the ideal situation when dealing with produced water

100

from processes such as SAGD, in which it is preferred to recuperate the purified water as the

101

permeate.

102

The surfaces of unmodified ceramic membranes are hydrophilic due to the charged hydroxyl

103

groups that populate the top layer. For metal oxide ceramics, the predominant surface charge is

104

dependent on the pH. If the pH is above the isoelectric point (IEP) of the ceramic material, then

105

the predominant surface charge is negative, while at a pH that is below the IEP, the predominant

106

surface charge is positive. The pH is thus an important feed parameter, since it determines the

107

surface charge of the membrane selective layer. The IEPs of common ceramic membrane

108

materials in water are given in Table 1 below.

109

Table 1: Isoelectric points of various ceramic materials (Kosmulski, 2011) SiO2

TE D

Ceramic material

M AN U

SC

RI PT

95

ZrO2

TiO2

Isoelectric point at 25°C 1.7-3.5 4-11 3.9-8.2

γ-Al2O3 α-Al2O3 7-8

8-9

When identifying possible foulants in the filtration of SAGD produced water, bitumen associated

111

solids must be taken into consideration. These solids include aggregates of quartz, clays and

112

heavy minerals that are bound together by carbonates, iron oxide and humic matter (Sparks et al.,

113

2003). However, the most problematic of the bitumen associated solids have been determined as

114

being ultrafine clays (Kotlyar et al., 1999; Sparks et al., 2003). These ultrafines are very thin

115

aluminosilicate clay crystallites onto which organic bitumen molecules and humic matter are

116

adsorbed. Therefore, major potential foulants in this application are silicates and adsorbed

117

carbon-rich solids. The surfaces of heavy bitumen molecules, such as asphaltenes, have been

118

shown to possess acidic, basic and amphoteric functional groups (Jada and Salou, 2002). This

119

indicates that bituminous fines can exhibit both negative and positive surface charges.

120

Combining this with the fact that the IEPs of the ceramic materials discussed above can vary

121

significantly, it becomes evident that maintaining the electrostatic repulsion of the identified

AC C

EP

110

4

ACCEPTED MANUSCRIPT

foulants is a difficult task. Hence, the attraction of these foulants to the ceramic membrane

123

surface is a significant hindrance that needs to be addressed.

124

The aim of this work was to verify the hypothesis that charge neutral hydrophilic ceramic

125

membrane surfaces would enhance the performance of ceramic membranes in treating produced

126

waters containing bituminous fines. In order to test the hypothesis, an organosilane modifying

127

agent was selected that would both maintain the hydrophilicity of the metal oxide selective layer

128

and reduce the number of hydroxyl groups on the oxide surface. Ceramic membrane disks were

129

modified using this organosilane and subsequently challenged with SAGD produced waters.

130

2. Experimental Section

131

2.1 Materials

132

The ceramic membrane disks studied in this work were purchased from Sterlitech Corporation

133

(Kent, WA, USA). The membranes had a diameter of 47 mm, a thickness of 2.5 mm, and offered

134

an overall mass transfer surface area of 13.1 cm2. The support layer of these membranes was

135

composed of titania. The selective layer of the 0.14 µm microfiltration membrane was made

136

from alumina, while the selective layer of the 150 kDa and 300 kDa Molecular Weight Cut-off

137

(MWCO) membranes contained a mix of ZrO2 and TiO2. The membrane specifications provided

138

by the manufacturer were determined by mercury porosimetry for the 0.14 µm microfiltration

139

membrane and the MWCO of the ultrafiltration membranes were determined according to the

140

NFX45-103 standard.

141

Anhydrous ethanol (Fisherbrand) and acetic acid (ACS plus, Fisherbrand) were used in the silane

142

surface grafting process. All silane solutions were supplied by Gelest, Inc. (Morrisville, PA,

143

USA). A hydrophilic silane containing polyethylene oxide (PEO) (PEOTMS), was selected as

144

the prime candidate because it offered an extremely low water contact angle of 15-16°, one of

145

the lowest among commercially available organosilanes (Arkles et al., 2009).

146

The feed used to conduct the filtration tests was a sample of SAGD produced water obtained

147

from CanmetMINING (NRCan). It was originally supplied to NRCan from SAGD operations

148

located in Canada’s oil sands in Alberta. To prevent clogging the gear pump used to circulate

149

feed in the test system, the SAGD process water was coarsely filtered using a 50 micron paper

150

filter prior to use.

AC C

EP

TE D

M AN U

SC

RI PT

122

5

ACCEPTED MANUSCRIPT

2.2 Surface Modification of Ceramic Membranes

152

The ceramic disk membranes obtained from Sterlitech were used, as received, in the silanation

153

procedure. The silanating method chosen was the deposition from aqueous alcohol solutions.

154

Firstly, a solution of 95% ethanol / 5% water by weight is prepared. The ethanol solution is then

155

adjusted to a pH of 4.5-5.5 using acetic acid. The silanating agent is then added with stirring to

156

yield a desired final concentration. When treating large objects, a silane concentration of 2 wt%

157

typically results in the deposition of trialkoxysilanes as 3 to 8 molecular layers. In this work, the

158

silane concentration was varied from 0.5 to 30 wt%. Following the addition of the silane, 5

159

minutes were allowed for hydrolysis and silanol formation. The membranes were then immersed

160

into the solution for 2 minutes while being gently agitated. Once removed from the silane

161

solution, the membranes were briefly rinsed in ethanol in order to remove any excess materials

162

and unreacted silane. Finally, the modified membranes were then placed in an oven, under

163

nitrogen and at 110˚C, for 10 minutes in order to cure the silane layer.

M AN U

SC

RI PT

151

164

2.3 Characterization of Membranes, Feed and Permeate

166

In order to confirm the success of the surface silylation procedure, FTIR analyses of the ceramic

167

membrane surfaces were conducted using a Cary 630 FTIR with an Attenuated Total Reflectance

168

(ATR) diamond accessory (Agilent Technologies, Canada). Small amounts of material were

169

scraped off of the surface of both unmodified and modified membranes, resulting in a powder

170

that was then used to completely and evenly cover the diamond accessory.

171

The elemental composition of inorganics in the SAGD produced water samples was determined

172

by inductively coupled plasma atomic emission spectroscopy (ICP-AES) and is provided in

173

Table 2 below. Samples were taken at four different locations ranging from the top to the bottom

174

of the barrel. The standard deviations in Table 2 were determined based on these four

175

measurements.

176

Table 2: Average inorganic composition of SAGD produced water feed

AC C

EP

TE D

165

Component

Na

Si

B

K

S

Ca

Mg

Sr

Ba

6

ACCEPTED MANUSCRIPT

Composition 476 (ppm) with standard

94.1

27.7

15.2

14.2

2.30

0.491

0.159

0.0284

±14.9 ±4.41 ±0.778 ±0.928 ±1.86 ±0.192 ±0.042 ±0.00592 ±0.00332

RI PT

deviations Both the SAGD wastewater feed and the permeate samples were subjected to a particle size

178

analysis. This allowed for the verification of the molecular weight cut-off of the ceramic

179

membranes. The samples were all analyzed in a Zetasizer Nano S90 (Malvern Instruments Ltd,

180

MA, USA). Samples of 3 mL were placed in the particle size analyzer, and the refractive index

181

of diluted bitumen was taken as 1.58 (Taylor et al., 2001). The total organic carbon (TOC) of all

182

feed and permeate samples was determined using a Total Organic Carbon analyzer.

183

For the experiments conducted in this study, five samples of wastewater feed were extracted

184

from the same drum on different dates. SAGD wastewater, like most produced water, undergoes

185

certain transformations upon aging during long term storage, which can affect filtration

186

performance (Ku et al., 2012). It is important to track feed sample usage during testing to be able

187

to feasibly compare results. In Table 3, these samples are assigned a label and their pH and TOC

188

is given.

189

Table 3: Label, pH and TOC of SAGD wastewater feed samples, extracted from the same drum,

190

used for filtration tests

pH

191

AC C

TOC (ppm)

F1

F2

F3

F4

F5

7.93

8.84

7.58

7.75

7.99

445

495

316

376

506

EP

Feed Sample

TE D

M AN U

SC

177

192

2.4 Filtration of SAGD Produced Water

193

The filtration experiments were conducted using a cross-flow filtration system. A schematic

194

diagram of the test system can be found in Figure 1 below. The entire setup was placed in a

195

walk-in fume hood in order to reduce emissions from the SAGD produced water recirculated in

196

the system. All of the filtration tests were conducted at temperatures ranging from 84 to 88 °C. 7

ACCEPTED MANUSCRIPT

The experiments using the 0.14 µm membranes were carried out at a transmembrane pressure

198

(TMP) of 68.95 kPa (10 psi), while all other membranes were subjected to a TMP of 172.6 kPa

199

(25 psi). The TMP of each run was maintained constant through the use of a pressure relief valve.

200

In order to remediate fouling, the system was backflushed with permeate for 5 seconds every 295

201

seconds at 103.4 kPa (15 psi) when using the 0.14 µm membrane, and at 206.8 kPa (30 psi) for

202

all other membranes. Initially, 3.7 L of SAGD produced water was loaded into the feed tank and

203

fed to the membrane with a cross-flow velocity of 1.2 m/s. The system was operated with total

204

retentate recycling, while the permeate was collected in a separate container and periodically

205

recycled back to the feed tank. The flowrate of the permeate stream was monitored using a

206

Coriolis flow meter. The duration of each run ranged from approximately 3.5 to 4 hours, as this

207

amount of time was considered to be sufficient in order to establish any differences between the

208

modifications.

M AN U

SC

RI PT

197

AC C

EP

TE D

209

8

ACCEPTED MANUSCRIPT

3. Results

211

3.1 Surface silylation of ceramic membranes

212

An FTIR spectrum of a 300 kDa membrane surface before and after the surface modification

213

procedure is shown in Figure 2. The modified membrane was modified using a 20 wt% silane

214

solution. Analysis of this spectrum indicates that the ceramic surface was successfully silylated.

215

The strong peak at 1090 cm-1 is characteristic of the Si-O-Si bonds in siloxanes(Launer, 1987),

216

as well as the C-O-C bonds in the PEO chains of the silane, indicating the presence of silane

217

PEOTMS on the ceramic surface. Moreover, the peak at 950 cm-1 is indicative of Si-O-Metal

218

bonding, which is further evidence of the reaction between the silane and the metal oxide surface

219

of the ceramic membrane. The peaks seen between 1350 and 1450 cm-1 represent the -C-H

220

bending found in the silane’s organic components, while the peak at 2850 cm-1 represents C-H

221

stretching.

222

3.2 Membrane flux and lot number

223

After a number of tests to compare the flux of modified and unmodified membranes, it was

224

observed that the flux of the unmodified membranes varied from one manufactured lot to

225

another. While the pore size of the membrane was a manufacturer’s criterion defining the

226

membrane disks, membrane flux was not. The variations in flux could easily be due to different

227

sol concentrations or coating dynamics during the deposition of the successive selective layers

228

on top of the membrane substrate. In order to remove variations between lots, membrane flux

229

was reported along with the lot number.

230

3.3 Membrane integrity on backflushing

231

The membranes tested in this work were cast on flat ceramic supports. Initially, membranes with

232

a MWCO of 50 kDa were to be included. However, the supports for the 50 kDa membranes were

233

different than those of the 150 kDa, 300 kDa and 0.14 µm membranes. The 50 kDa supports did

234

not withstand the backflushing pressure of 206.8 kPa (30 psi). All of the 50 kDa membranes

235

cracked during filtration and could not be adequately tested. All other membranes could be

236

backflushed with the exception of the unmodified 150 kDa membrane, which showed signs of

237

failure in its selective layer after approximately 42 minutes. All other membranes were operated

238

without showing signs of failure.

AC C

EP

TE D

M AN U

SC

RI PT

210

9

ACCEPTED MANUSCRIPT

3.4 Membrane flux and silane concentration

240

The run conditions of the modified membranes using agent PEOTMS and the unmodified

241

membranes are summarized in Tables 4 to 6 below. The tables are sectioned according to the

242

manufacturer’s membrane lot numbers. As previously mentioned, the increase or reduction in

243

membrane flux as a result of treatment was compared within a lot and for the same feed. The

244

average flux over the entire filtration period for the 150 kDa, 300 kDa and 0.14 µm membranes

245

can be found in Figure 3 at different modifying agent concentrations. The permeate flux is given

246

by the following equation, where J is flux in m/s, ṁ is mass flowrate of permeate in kg/s, ρ is

247

permeate density in kg/m3 and A is active membrane surface area in m2. The flux given in

248

equation 1 was multiplied by 3.6 x 106 to give units of L/m2h (Lmh).

SC

( ⁄ ) =

M AN U

249

RI PT

239

∙

(1)

The dashed lines in the plot demonstrate the correlation between flux and silane concentration

251

for filtration tests that were run using the same feed sample, as well as membranes originating

252

from the same batch.

253 254 255

Table 4: Test conditions for the unmodified and modified 150 kDa membrane using agent PEOTMS. Operating pressure 172.6 kPa (25 psi) Trans Membrane Pressure, backflushing with permeate for 5s every 295s at 206.8 kPa (30 psi).

257 258 259

Membrane Lot #

SAGD wastewater feed sample

Feed Temperature (°C)

Unmodified 5% 10% 20% 30%

131113U150 131113U150 131113U150 131113U150 131113U150

F4 F4 F4 F4 F5

85-87 85-88 85-88 85-88 85-88

EP

Surface Modification

AC C

256

TE D

250

Table 5: Test conditions for the unmodified and modified 300 kDa membrane using agent PEOTMS. Operating pressure 172.6 kPa (25 psi) Trans Membrane Pressure, backflushing with permeate for 5s every 295s at 206.8 kPa (30 psi). Surface Modification

Membrane Lot #

Membrane Lot Label

Unmodified

191110U300

B1

SAGD wastewater feed sample F2

Feed Temperature (°C) 87 10

ACCEPTED MANUSCRIPT

191110U300 191110U300 140312U300 140312U300 140312U300 140312U300 140312U300

B1 B1 B2 B2 B2 B2 B2

F2 F2 F3 F3 F2 F3 F3

260

SC

Table 6: Test conditions for the unmodified and modified 0.14 µm membrane using agent PEOTMS. Operating pressure 68.95 kPa (10 psi) Trans Membrane Pressure, backflushing with permeate for 5s every 295s at 103.4 kPa (15 psi). Surface Modification

Membrane Lot #

SAGD wastewater feed sample

Feed Temperature (°C)

Unmodified 2% 10%

190310M014 190310M014 190310M014

F1 F1 F2

84-87 87 84-86

M AN U

261 262 263

84-87 85-87 86-88 86-87 85-88 86-87 85-87

RI PT

0.5% 1% Unmodified 2% 3% 15% 20%

264

All membranes modified in a bath containing 20% of PEOTMS or less showed an increase in

266

flux performance compared to the unmodified membranes. The greatest increase in performance

267

was observed for the 150 kDa and 300 kDa membranes at a concentration of 20%, where the flux

268

of the modified membrane was 2.5 and 2.2 times that of the unmodified membrane, respectively.

269

The flux of the 0.14 µm membrane also showed an increase in flux along the trend of the 150

270

and 300 kDa membranes. The 30% PEOTMS 150 kDa membrane was run with a different feed

271

sample and was not retained for comparison. Similar cases are seen with the 3% PEOTMS 300

272

kDa and 10% PEOTMS 0.14 µm membranes. All membranes were tested with SAGD produced

273

water originating from a single drum. However, the feeds had slightly different pHs and TOC

274

content. When membranes of the same batch were run using a feed having a higher pH and TOC,

275

lower fluxes were observed. There are three experiments that illustrate this: the 30% modified

276

150 kDa membrane tested with feed F5 in Table 4, the 3% modified 300 kDa membrane tested

277

with feed F2 in Table 5, and the 10% modified 0.14 µm membrane tested with feed F2 in Table

278

6. As seen in Table 3, feed sample F5 has a higher pH and TOC than feed sample F4, while feed

279

sample F2 has a higher pH and TOC than both feed sample F3 and F1. The higher TOC values

280

indicate a greater concentration of organics in the feeds that led to lower fluxes.

AC C

EP

TE D

265

11

ACCEPTED MANUSCRIPT

It has been shown that the dissociation of soluble silica, in the form of silicic acid, grows more

282

prominent as the pH surpasses neutral (Sanciolo et al., 2014). Silicic acid dissociates into silicate

283

anions, which can then react with the calcium or magnesium found in the wastewater feed to

284

produce insoluble silicates. These have high scale forming potential, and could account for the

285

observed decline in permeate flux at higher pH values.

286

Nonetheless, the flux of all membranes from the same batch and run with the same feed

287

increased as a result of a treatment with agent PEOTMS. This trend was even observed for silane

288

concentrations as low as 0.5%, as evidenced by the 300 kDa membranes. This indicates that the

289

modification is effective in releasing bitumen fines from the surface of the modified membrane

290

and thus reducing the severity of the fouling phenomenon.

291

3.5 Flux as a function of time and concentration factor

292

The accumulation of colloids and fines on the surface of the membrane is a natural consequence

293

of the filtration process which results in a decline of permeate flux over time. The accumulation

294

of particles is mitigated by the presence of tangential flow over the surface of the membrane and

295

backflushing. Particles on the surface of the membrane are partly swept away by feed circulating

296

across the surface of the membrane and released from the surface by backflushing. Over time,

297

the flux will stabilize, where an equal amount of particles are deposited on and released from the

298

surface of the membrane. The number of particles arriving on the surface of the membrane will

299

be related to the volume of process water filtered and the concentration of particulates and

300

colloids in the process water.

301

During a run, permeate from the membrane was collected in a container placed on the permeate

302

balance shown in Figure 1. It was recycled periodically back to the feed tank. Initially, the feed

303

tank contained 3.7 L of SAGD produced water. As permeate was collected, the concentration of

304

particulates in the feed increased. It is important to determine the effect of this increase in

305

concentration on membrane flux. Hence, membrane flux was plotted as a function of the

306

concentration factor in Figure 4 for the 150 kDa and Figure 5 for the 300 kDa pore sized

307

membranes. The concentration factor is defined as follows, where VF is the initial volume of

308

produced water in the feed tank and VP is the volume of permeate collected at a given time

309

during a filtration test.

AC C

EP

TE D

M AN U

SC

RI PT

281

12

ACCEPTED MANUSCRIPT

310



=

(2)

From Figure 4, it can be seen that the modified 150 kDa membranes maintained a high flux even

312

when the feed water increased in concentration. The flux was also steady in the case of the 15%

313

300 kDa membrane shown in Figure 5. After an initial decline, the flux of the 15% 300 kDa

314

membrane was stable for the second collection of permeate following its recycling. The flux of

315

the 20% 300 kDa modified membrane was much higher than those of the 15% 300 kDa and the

316

untreated membrane. However, at the 20% level, the 300 kDa membranes showed some decline

317

without an indication of stability.

318

3.6 Analysis of water quality

319

3.6.1 Total organic carbon analysis

320

The total organic carbon content of the feed SAGD produced water, as well as the permeate of

321

each run, was measured for use as an indicator of water cleanliness. The results of this analysis

322

can be found below. The fraction of the TOC removed in the permeate relative to the feed SAGD

323

wastewater was plotted against both the silane modification concentration (Figure 6) and the

324

permeate flux (Figure 7). This was done for the 150 kDa, 300 kDa and 0.14 µm membranes.

325

In Figure 6, the dashed lines indicate the correlation observed for membrane filtration tests using

326

the same feed sample. Here, membranes originating from different manufactured batches could

327

be compared as long as they used the same feed because TOC separation depends on the

328

membrane pore size, and not the flux. It was observed that, in all cases, the silane surface

329

modification lead to an increase in TOC separation when compared to the respective unmodified

330

membranes. While the 10% PEOTMS 0.14 µm membrane showed a poorer TOC separation

331

relative to the unmodified 0.14 µm membrane, this can be attributed to the fact that the filtration

332

test of this particular membrane was conducted using a different feed. This feed sample

333

possessed a higher concentration of bituminous organic compounds. The higher concentration of

334

foulants in the feed would have thus lead to a reduction in TOC removal in comparison to the

335

unmodified membrane that used a different feed. The same explanation can be offered for the

336

30% 150 kDa membrane, which showed a decrease TOC separation.

AC C

EP

TE D

M AN U

SC

RI PT

311

13

ACCEPTED MANUSCRIPT

In Figure 7, the dashed lines and arrows indicate the general trend observed for filtration tests

338

using the same feed sample and membrane batch. Since permeate flux was a parameter of

339

interest in this comparison, both the membrane batch and the feed sample had to be taken into

340

consideration. It can be observed that surface modification with silane PEOTMS leads to an

341

increase in both permeate flux and TOC removal for all three sets of the tested ceramic

342

membranes. Moreover, increasing the concentration of the silane in the modifying solution

343

resulted in further improvement in separation and permeate flux. Consequently, for the 150 and

344

300 kDa membranes, the highest silane concentration tested of 20% was found to be optimal in

345

terms of maximizing flux and separation performance. The results of the TOC analysis indicate

346

that the modification is successful in improving the rejection of bituminous organics during

347

filtration of SAGD produced water.

348

3.6.2 Particle size analysis

349

As a method of characterising the performance of the ceramic membranes tested in this study, the

350

particle size distributions of both the feed and permeate samples were determined. The particle

351

distribution of all feeds was measured. They did not vary significantly, containing particles

352

ranging from 220 to 710 nm in diameter. The distributions of all five feed samples were plotted

353

in Figure 8 below. Since the bituminous ultrafines in the tar sands are known to possess

354

dimensions that are inferior to 300 nm (Sparks et al., 2003), the larger particles seen in the

355

produced water feed are likely aggregates of ultrafines. The high salt content in the wastewater

356

could be causing the compaction of flat clay platelets to form larger clusters of three-dimension

357

structures. The particle size distribution of the permeates obtained from the 150 kDa, 300 kDa

358

and 0.14 µm membranes are given in Figures 9, 10 and 11, respectively. These distributions are

359

reported along with the corresponding concentration of the silanating agent. All of the

360

distributions show that the particle diameters in the permeates do not exceed 40 nm. The

361

corresponding pore diameters of the 150 kDa and 300 kDa MWCO membranes are

362

approximately 21.5 nm and 30 nm, respectively, based on models developed using dextran and

363

pullulan molecules (Rolland-Sabaté et al., 2011). The distributions observed for permeate

364

samples from these membranes are, therefore, in accordance with their pore sizes. The fact that

365

filtration with the 0.14 µm membranes resulted in permeates with similar distributions to those

AC C

EP

TE D

M AN U

SC

RI PT

337

14

ACCEPTED MANUSCRIPT

obtained from the 150 and 300 kDa membranes is due to the characterization method as

367

previously mentioned in the materials section.

368

In the case of the 300 kDa membranes, it can be seen in Figure 10 that the silane surface

369

modified membranes all produced permeate particle size distributions that were lower compared

370

to the unmodified membrane. However, this was not a general trend as seen in Figures 9 and 11

371

for the 150 kDa and 0.14 µm membranes respectively. The discrepancy can be attributed to the

372

fact that the SAGD feed contained particles that could aggregate after they had permeated

373

through the membrane. Altogether, the particle size analysis of the permeates demonstrates that

374

all of the ceramic membranes tested were effective in terms of rejecting particles of diameter

375

larger than 40 nm in the five SAGD produced water samples.

SC

RI PT

366

M AN U

376 4. Conclusions

378

Ceramic flat ultra- and micro-filtration membrane disks with pore sizes of 150 kDa, 300 kDa and

379

0.14 µm were modified using an organosilane surface modifying agent. The results successfully

380

demonstrated that it was possible to modify ceramic membranes to exhibit fouling-resistant

381

properties to bituminous solids and consequently enhance their performance in treating SAGD

382

produced waters. Modification agent PEOTMS more than doubled the flux of the 150 kDa and

383

300 kDa ceramic membranes in treating these process waters relative to the unmodified

384

membranes. The modified ceramic membranes were also found to successfully remove up to

385

72% of the total organic carbon found in SAGD produced water ensuring water cleanliness for

386

recycling. Furthermore, all of the ceramic membranes tested were shown to reduce the particle

387

sizes in the produced water from >200 nm in the feed to <40 nm in the permeate. The observed

388

trends indicated that higher concentrations in the modification bath lead to both higher permeate

389

flux and water quality. The results and protocols developed in this work can be used to modify

390

commercial membranes. They will also permit us to better identify other surface modifying

391

agents in order to improve membrane performance in treating SAGD produced waters.

AC C

EP

TE D

377

392 393

15

ACCEPTED MANUSCRIPT

Acknowledgements

395

The authors gratefully acknowledge funding support from Natural Resources Canada through the

396

Program for Energy Research and Development (PERD) and eco-Energy Innovation Initiative

397

(ecoEII).

AC C

EP

TE D

M AN U

SC

RI PT

394

16

ACCEPTED MANUSCRIPT

References

399 400 401

Abidi, N., Sivade, A., Bourret, D., Larbot, A., Boutevin, B., Guida-Pietrasanta, F., Ratsimihety, A., 2006. Surface modification of mesoporous membranes by fluoro-silane coupling reagent for CO2 separation. J. Memb. Sci. 270, 101–107. doi:10.1016/j.memsci.2005.06.054

402 403 404

Ahmad, N.A., Leo, C.P., Ahmad, A.L., 2013. Superhydrophobic alumina membrane by steam impingement: Minimum resistance in microfiltration. Sep. Purif. Technol. 107, 187–194. doi:10.1016/j.seppur.2013.01.011

405 406 407 408

Alami Younssi, S., Iraqi, A., Rafiq, M., Persin, M., Larbot, A., Sarrazin, J., 2003. γ Alumina membranes grafting by organosilanes and its application to the separation of solvent mixtures by pervaporation. Sep. Purif. Technol. 32, 175–179. doi:10.1016/S13835866(03)00031-5

409

Alberta Official Statistics: Total Bitumen Production, Alberta, 2015.

410 411

Allen, E.W., 2008. Process water treatment in Canada’s oil sands industry: II. A review of emerging technologies. J. Environ. Eng. Sci. 7, 499–524. doi:10.1139/S08-020

412 413

Arkles, B., Pan, Y., Kim, Y., 2009. The Role of Polarity in the Structure of Silanes Employed in Surface Modification. Silanes and Other Coupling Agents 5, 51–64.

414 415 416

Bothun, G.D., Peay, K., Ilias, S., 2007. Role of tail chemistry on liquid and gas transport through organosilane-modified mesoporous ceramic membranes. J. Memb. Sci. 301, 162–170. doi:10.1016/j.memsci.2007.06.011

417 418

Castro, R.P., Cohen, Y., Monbouquette, H.G., 1996. Silica-supported polyvinylpyrrolidone filtration membranes. J. Memb. Sci. 115, 179–190. doi:10.1016/0376-7388(96)00019-1

419 420 421 422

Chang, Q., Zhou, J.E., Wang, Y., Liang, J., Zhang, X., Cerneaux, S., Wang, X., Zhu, Z., Dong, Y., 2014. Application of ceramic microfiltration membrane modified by nano-TiO2 coating in separation of a stable oil-in-water emulsion. J. Memb. Sci. 456, 128–133. doi:10.1016/j.memsci.2014.01.029

423 424 425

Chang, Q., Zhou, J.E., Wang, Y., Wang, J., Meng, G., 2010. Hydrophilic modification of Al2O3 microfiltration membrane with nano-sized γ-Al2O3 coating. Desalination 262, 110–114. doi:10.1016/j.desal.2010.05.055

426 427 428

Chen, D., Weavers, L.K., Walker, H.W., 2006. Ultrasonic control of ceramic membrane fouling by particles: Effect of ultrasonic factors. Ultrason. Sonochem. 13, 379–387. doi:10.1016/j.ultsonch.2005.07.004

429 430 431

Cui, J., Zhang, X., Liu, H., Liu, S., Yeung, K.L., 2008. Preparation and application of zeolite/ceramic microfiltration membranes for treatment of oil contaminated water. J. Memb. Sci. 325, 420–426. doi:10.1016/j.memsci.2008.08.015

432 433 434 435

Faibish, R.S., Cohen, Y., 2001a. Fouling and rejection behavior of ceramic and polymermodified ceramic membranes for ultrafiltration of oil-in-water emulsions and microemulsions. Colloids Surfaces A Physicochem. Eng. Asp. 191, 27–40. doi:10.1016/S0927-7757(01)00761-0

436

Faibish, R.S., Cohen, Y., 2001b. Fouling-resistant ceramic-supported polymer membranes for

AC C

EP

TE D

M AN U

SC

RI PT

398

17

ACCEPTED MANUSCRIPT

437 438

ultrafiltration of oil-in-water microemulsions. J. Memb. Sci. 185, 129–143. doi:10.1016/S0376-7388(00)00595-0 Faiz, R., Fallanza, M., Ortiz, I., Li, K., 2013. Separation of olefin paraffin gas mixtures using ceramic hollow fiber membrane contactors. Ind. Eng. Chem. Res. 52, 7918–7929.

441 442 443

Gao, N., Fan, Y., Quan, X., Cai, Y., Zhou, D., 2016. Modified ceramic membranes for low fouling separation of water-in-oil emulsions. J. Mater. Sci. 51, 6379–6388. doi:10.1007/s10853-016-9934-3

444 445 446

Gao, N., Ke, W., Fan, Y., Xu, N., 2013. Evaluation of the oleophilicity of different alkoxysilane modified ceramic membranes through wetting dynamic measurements. Appl. Surf. Sci. 283, 863–870. doi:10.1016/j.apsusc.2013.07.034

447 448 449

Gao, N., Li, M., Jing, W., Fan, Y., Xu, N., 2011. Improving the filtration performance of ZrO2 membrane in non-polar organic solvents by surface hydrophobic modification. J. Memb. Sci. 375, 276–283. doi:10.1016/j.memsci.2011.03.056

450 451 452

Hardin, J.L., Oyler, N.A., Steinle, E.D., Meints, G.A., 2010. Spectroscopic analysis of interactions between alkylated silanes and alumina nanoporous membranes. J. Colloid Interface Sci. 342, 614–619. doi:10.1016/j.jcis.2009.10.083

453 454 455

Jada, A., Salou, M., 2002. Effects of the asphaltene and resin contents of the bitumens on the water-bitumen interface properties. J. Pet. Sci. Eng. 33, 185–193. doi:10.1016/S09204105(01)00185-1

456 457 458

Javaid, A., Krapchetov, D.A., Ford, D.M., 2005. Solubility-based gas separation with oligomermodified inorganic membranes: Part III. Effects of synthesis conditions. J. Memb. Sci. 246, 181–191. doi:10.1016/j.memsci.2004.08.017

459 460

Jordaan, S.M., 2012. Land and water impacts of oil sands production in Alberta. Environ. Sci. Technol. 46, 3611–3617. doi:10.1021/es203682m

461 462 463

Kessel, C.R., Granick, S., 1991. Formation and characterization of a highly ordered and wellanchored alkylsilane monolayer on mica by self-assembly. Langmuir 7, 532–538. doi:10.1021/la00051a020

464 465 466

Khemakhem, S., Amar, R. Ben, 2011. Modification of Tunisian clay membrane surface by silane grafting: Application for desalination with Air Gap Membrane Distillation process. Colloids Surfaces A Physicochem. Eng. Asp. 387, 79–85. doi:10.1016/j.colsurfa.2011.07.033

467 468

Kosmulski, M., 2011. The pH-dependent surface charging and points of zero charge. J. Colloid Interface Sci. 353, 1–15. doi:10.1016/j.jcis.2010.08.023

469 470 471

Kotlyar, L.S., Sparks, B.D., Woods, J.R., Chung, K.H., 1999. Solids associated with the asphaltene fraction of oil sands bitumen. Energy and Fuels 13, 346–350. doi:10.1021/ef980204p

472 473 474

Krajewski, S.R., Kujawski, W., Dijoux, F., Picard, C., Larbot, A., 2004. Grafting of ZrO2 powder and ZrO2 membrane by fluoroalkylsilanes. Colloids Surfaces A Physicochem. Eng. Asp. 243, 43–47. doi:10.1016/j.colsurfa.2004.05.001

475

Ku, A.Y., Henderson, C.S., Petersen, M.A., Pernitsky, D.J., Sun, A.Q., 2012. Aging of Water

AC C

EP

TE D

M AN U

SC

RI PT

439 440

18

ACCEPTED MANUSCRIPT

476 477 478

from Steam-Assisted Gravity Drainage (SAGD) Operations Due to Air Exposure and Effects on Ceramic Membrane Filtration. Ind. Eng. Chem. Res. 51, 7170–7176. doi:10.1021/ie3005513 Kujawa, J., Cerneaux, S., Kujawski, W., 2014. Investigation of the stability of metal oxide powders and ceramic membranes grafted by perfluoroalkylsilanes. Colloids Surfaces A Physicochem. Eng. Asp. 443, 109–117. doi:10.1016/j.colsurfa.2013.10.059

482 483 484 485

Kujawa, J., Cerneaux, S., Kujawski, W., Bryjak, M., Kujawski, J.K., 2016. How To Functionalize Ceramics by Perfluoroalkylsilanes for Membrane Separation Process? Properties and Application of Hydrophobized Ceramic Membranes. ACS Appl. Mater. Interfaces 8, 7564–7577. doi:10.1021/acsami.6b00140

486 487 488 489

Kujawa, J., Kujawski, W., 2016. Functionalization of Ceramic Metal Oxide Powders and Ceramic Membranes by Perfluroalkylsilanes and Alkylsilanes Possessing Different Reactive Groups –Physicochemical and Tribological Properties. ACS Appl. Mater. Interfaces 8, 7509–7521. doi:10.1021/acsami.5b11975

490 491 492

Kujawski, W., Krajewska, S., Kujawski, M., Gazagnes, L., Larbot, A., Persin, M., 2007. Pervaporation properties of fluoroalkylsilane (FAS) grafted ceramic membranes. Desalination 205, 75–86. doi:10.1016/j.desal.2006.04.042

493 494 495 496

Kujawski, W., Kujawa, J., Wierzbowska, E., Cerneaux, S., Bryjak, M., Kujawski, J., 2016. Influence of hydrophobization conditions and ceramic membranes pore size on their properties in vacuum membrane distillation of water-organic solvent mixtures. J. Memb. Sci. 499, 442–451. doi:10.1016/j.memsci.2015.10.067

497 498

Larbot, A., Gazagnes, L., Krajewski, S.R., Bukowska, M., Kujawski, W., 2004. Water desalination using ceramic membrane distillation. Desalination 168, 367–372.

499 500 501

Launer, P.J., 1987. Infrared Analysis of Organosilicon Compounds: Spectra-Structure Correlations, in: Silicone Compounds: Register and Review. Petrarch Sytsems, Bristol, PA, pp. 100–103.

502 503 504

Ostwal, M., Singh, R.P., Dec, S.F., Lusk, M.T., Way, J.D., 2011. 3-Aminopropyltriethoxysilane functionalized inorganic membranes for high temperature CO2/N2 separation. J. Memb. Sci. 369, 139–147. doi:10.1016/j.memsci.2010.11.053

505 506 507

Padaki, M., Surya Murali, R., Abdullah, M.S., Misdan, N., Moslehyani, A., Kassim, M.A., Hilal, N., Ismail, A.F., 2015. Membrane technology enhancement in oil-water separation. A review. Desalination 357, 197–207. doi:10.1016/j.desal.2014.11.023

508 509 510 511 512

Rolland-Sabaté, A., Guilois, S., Jaillais, B., Colonna, P., Rolland-Sabate, A., 2011. Molecular size and mass distributions of native starches using complementary separation methods: Asymmetrical Flow Field Flow Fractionation (A4F) and Hydrodynamic and Size Exclusion Chromatography (HDC-SEC). Anal. Bioanal. Chem. 399, 1493–1505. doi:10.1007/s00216010-4208-4

513 514 515

Sanciolo, P., Milne, N., Taylor, K., Mullet, M., Gray, S., 2014. Silica scale mitigation for high recovery reverse osmosis of groundwater for a mining process. Desalination 340, 49–58. doi:10.1016/j.desal.2014.02.024

AC C

EP

TE D

M AN U

SC

RI PT

479 480 481

19

ACCEPTED MANUSCRIPT

Silalahi, S.H.D., Leiknes, T., 2009. Cleaning strategies in ceramic microfiltration membranes fouled by oil and particulate matter in produced water. Desalination 236, 160–169. doi:10.1016/j.desal.2007.10.063

519 520

Sondhi, R., Bhave, R., Jung, G., 2003. Applications and benefits of ceramic membranes. Membr. Technol. 2003, 5–8. doi:10.1016/S0958-2118(03)11016-6

521 522 523

Sparks, B.D., Kotlyar, L.S., O’Carroll, J.B., Chung, K.H., 2003. Athabasca oil sands: effect of organic coated solids on bitumen recovery and quality. J. Pet. Sci. Eng. 39, 417–430. doi:10.1016/S0920-4105(03)00080-9

524 525 526

Swenson, P., Tanchuk, B., Bastida, E., An, W., Kuznicki, S.M., 2012. Water desalination and deoiling with natural zeolite membranes - Potential application for purification of SAGD process water. Desalination 286, 442–446. doi:10.1016/j.desal.2011.12.008

527 528

Taylor, S.D., Czarnecki, J., Masliyah, J., 2001. Refractive index measurements of diluted bitumen solutions. Fuel 80, 2013–2018. doi:10.1016/S0016-2361(01)00087-4

529 530 531

Zhao, Y., Zhong, J., Li, H., Xu, N., Shi, J., 2002. Fouling and regeneration of ceramic microfiltration membranes in processing acid wastewater containing fine TiO2 particles. J. Memb. Sci. 208, 331–341. doi:10.1016/S0376-7388(02)00314-9

532 533 534

Zhou, J.E., Chang, Q., Wang, Y., Wang, J., Meng, G., 2010. Separation of stable oil-water emulsion by the hydrophilic nano-sized ZrO2 modified Al2O3 microfiltration membrane. Sep. Purif. Technol. 75, 243–248. doi:10.1016/j.seppur.2010.08.008

M AN U

SC

RI PT

516 517 518

AC C

EP

TE D

535

20

ACCEPTED MANUSCRIPT

All figure captions are below. Figures are uploaded separately in PDF format:

Figure 1: Schematic of filtration system

RI PT

Figure 2: FTIR spectra of the surface of an unmodified 300 kDa membrane and a 20% PEOTMS modified 300 kDa membrane. Figure 3: Permeate flux as a function of the concentration of silane PEOTMS in the modification bath for the 150 kDa, 300 kDa and 0.14 µm membrane disks. Labels indicate which runs were conducted using the same membrane batch and wastewater feed.

SC

Figure 4: Permeate flux vs concentration factor for the 150 kDa membranes modified with agent PEOTMS. The solid line at a flux of 142 lmh represents the flux of the unmodified membrane in the initial 42 minutes of its run.

M AN U

Figure 5: Permeate flux vs concentration factor for the 300 kDa membranes modified with agent PEOTMS. Figure 6: Total organic carbon separation as a function of the concentration of silane PEOTMS. Labels indicate which runs were done using the same batch of membranes and the same feed. Figure 7: Total organic carbon separation as a function of permeate flux for the 150 kDa, 300 kDa and 0.14 um membranes. Labels indicate the silane modification concentration, the membrane batch and the feed sample of each run.

TE D

Figure 8: Particle size distributions of the SAGD produced water feeds. Figure 9: Particle size distributions of permeate samples from the 150 kDa membranes.

EP

Figure 10: Particle size distributions of permeate samples from the 300 kDa membranes.

AC C

Figure 11: Particle size distributions of permeate samples from the 0.14 µm membranes.

21

ACCEPTED MANUSCRIPT

N2

Pump

SC

Permeate tank

Feed tank

M AN U

Permeate balance

RI PT

Membrane test cell

Flow meter

AC C

EP

TE D

Figure 1: Schematic of filtration system

Feed balance

ACCEPTED MANUSCRIPT

50 Unmodified (F3) 2% (F3) 3% (F2) 15% (F3) 20% (F3)

45

RI PT

40

30

SC

25 20

10

5 0 5

10

15 20 25 Particle Diameter (nm)

TE D

0

M AN U

15

30

35

EP

Figure 10: Particle size distributions of permeate samples from the 300 kDa membranes.

AC C

Number %

35

40

ACCEPTED MANUSCRIPT

40

35

RI PT

30

Unmodified (F1) 2% (F1) 10% (F2)

SC

20

10

5

0 5

10

15 20 25 Particle Diameter (nm)

TE D

0

M AN U

15

30

35

EP

Figure 11: Particle size distributions of permeate samples from the 0.14 µm membranes.

AC C

Number %

25

40

ACCEPTED MANUSCRIPT

100 Unmodified 90

Modified

80

60 50

4000

3500

3000

M AN U

SC

40

2500 2000 Wavenumber (cm-1)

1500

1000

Transmittance %

RI PT

70

30 20 10 0 500

AC C

EP

TE D

Figure 2: FTIR spectra of the surface of an unmodified 300 kDa membrane and a 20% PEOTMS modified 300 kDa membrane.

1

ACCEPTED MANUSCRIPT

450 0.14 um

F1

300 kDa

400

150 kDa B1F2

F1

F4

B1F2

300

B2F3

F4 F2

250

B2F3

200

150 F4 B2F3

B2F2

100 0

5

10

SC

B2F3

RI PT

350

F4

M AN U

Average Permeate Flux (Lmh)

B1F2

15 20 Silane wt% in modification bath

25

F5

30

35

AC C

EP

TE D

Figure 3: Permeate flux as a function of the concentration of silane PEOTMS in the modification bath for the 150 kDa, 300 kDa and 0.14 µm membrane disks. Labels indicate which runs were conducted using the same membrane batch and wastewater feed.

ACCEPTED MANUSCRIPT

450

5% a 150 kDa

400

10% a 150 kDa

350 300

RI PT

Permeate Flux (Lmh)

500

250 200 150

SC

100 50 0 1.05

1.1

1.15 1.2 Concentration Factor

M AN U

1

1.25

1.3

AC C

EP

TE D

Figure 4: Permeate flux vs concentration factor for the 150 kDa membranes modified with agent PEOTMS. The solid line at a flux of 142 lmh represents the flux of the unmodified membrane in the initial 42 minutes of its run.

ACCEPTED MANUSCRIPT

500

20% a 300 kDa unmodified 300 kDa 15% a 300 kDa

450

RI PT

350 300

250 200 150

SC

Permeate Flux (Lmh)

400

100

0 1

1.05

1.1

M AN U

50

1.15 Concentration Factor

1.2

1.25

1.3

AC C

EP

TE D

Figure 5: Permeate flux vs concentration factor for the 300 kDa membranes modified with agent PEOTMS.

ACCEPTED MANUSCRIPT

0.75

F4

0.7

F1 F4

F4

F4 B2F3

F2

F1 B2F2 B1F2

0.5 B1F2

0.45

B2F3

B2F3

0.4 0.35 0.3 5

10

M AN U

B2F3

0

RI PT

F5

0.6

SC

Fraction of TOC removed

0.65

0.55

0.14 um 300 kDa 150 kDa

15 20 Silane wt% in modification bath

25

30

35

AC C

EP

TE D

Figure 6: Total organic carbon separation as a function of the concentration of silane PEOTMS. Labels indicate which runs were done using the same batch of membranes and the same feed.

ACCEPTED MANUSCRIPT

0.75 20%, F1

0.7

2%, F1

0.6

5%, F1

0%, F1

20%, B2F2

0.55

0%, F1

10%, F2

1%, B1F1

3%, B2F1

0.5 15%, B2F2

0.45 0%, B2F2

0.4 0.35 0.3 150

200

0%, B1F1

M AN U

2%, B2F2

100

RI PT

10%, F1

30%, F2

SC

Fraction of TOC removed

0.65

250 300 Average Permeate Flux (Lmh)

350

0.14 um 300 kDa 150 kDa

400

450

AC C

EP

TE D

Figure 7: Total organic carbon separation as a function of permeate flux for the 150 kDa, 300 kDa and 0.14 um membranes. Labels indicate the silane modification concentration, the membrane batch and the feed sample of each run.

ACCEPTED MANUSCRIPT

25

F1 F2 F3 F4 F5

RI PT

15

SC

10

0 0

100

200

M AN U

5

300 400 500 Particle Diameter (nm)

600

EP

TE D

Figure 8: Particle size distributions of the SAGD produced water feeds.

AC C

Number %

20

700

800

ACCEPTED MANUSCRIPT

40 Unmodified (F4) 10% (F4) 20% (F4) 30% (F5)

35

RI PT

30

SC

20

10

5

0 5

10

15 20 25 Particle Diameter (nm)

TE D

0

M AN U

15

30

35

EP

Figure 9: Particle size distributions of permeate samples from the 150 kDa membranes.

AC C

Number %

25

40

ACCEPTED MANUSCRIPT

EP

TE D

M AN U

SC

RI PT

A ceramic membrane that is fouling resistant to bitumen was developed. Surface modification of membrane achieved using charge neutral PEO-based silane. Ceramic membranes were challenged with SAGD produced water. Modified membranes doubled water permeate flux. Improved TOC and particle size reduction were also observed.

AC C

• • • • •