Integrated approach to characterize fouling on a flat sheet membrane gravity driven submerged membrane bioreactor

Accepted Manuscript Integrated approach to characterize fouling on a flat sheet membrane gravity driven submerged membrane bioreactor Luca Fortunato, Sanghyun Jeong, Yiran Wang, Ali R. Behzad, TorOve Leiknes PII: DOI: Reference:

S0960-8524(16)31394-3 http://dx.doi.org/10.1016/j.biortech.2016.09.127 BITE 17150

To appear in:

Bioresource Technology

Received Date: Revised Date: Accepted Date:

8 August 2016 28 September 2016 29 September 2016

Please cite this article as: Fortunato, L., Jeong, S., Wang, Y., Behzad, A.R., Leiknes, T., Integrated approach to characterize fouling on a flat sheet membrane gravity driven submerged membrane bioreactor, Bioresource Technology (2016), doi: http://dx.doi.org/10.1016/j.biortech.2016.09.127

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.

1

Integrated approach to characterize fouling on a flat sheet membrane

2

gravity driven submerged membrane bioreactor

3 4

Luca Fortunatoa, Sanghyun Jeonga, Yiran Wanga, Ali R. Behzadb, TorOve Leiknesa, *

5 6

a

7

(BESE), King Abdullah University of Science and Technology (KAUST), Thuwal 23955-6900, Saudi

8

Arabia,

9

b

Water Desalination and Reuse Center (WDRC), Biological and Environmental Science & Engineering

Advanced Nanofabrication Imaging and Characterization Laboratory, King Abdullah University of

10

Science and Technology (KAUST), 23955-6900 Thuwal, Saudi Arabia

11

*Corresponding author: Tel. +966 12 808 2193; Email: [email protected]

12 13 14

Abstract Fouling in membrane bioreactors (MBR) is acknowledged to be complex and

15

unclear. An integrated characterization methodology was employed in this study to

16

understand the fouling on a gravity-driven submerged MBR (GD-SMBR). It involved the

17

use of different analytical tools, including optical coherence tomography (OCT), liquid

18

chromatography with organic carbon detection (LC-OCD), total organic carbon (TOC),

19

flow cytometer (FCM), adenosine triphosphate analysis (ATP) and scanning electron

20

microscopy (SEM). The three-dimensional (3D) biomass morphology was acquired in a

21

real-time through non-destructive and in-situ OCT scanning of 75% of the total

22

membrane surface directly in the tank. Results showed that the biomass layer was

23

homogeneously distributed on the membrane surface. The amount of biomass was

24

selectively linked with final destructive autopsy techniques. The LC-OCD analysis

1

25

indicated the abundance of low molecular weight (LMW) organics in the fouling

26

composition. Three different SEM techniques were applied to investigate the detailed

27

fouling morphology on the membrane.

28 29

Keywords: Biomass; Fouling; Gravity driven submerged membrane bioreactor; Optical

30

coherence tomography; Scanning electron microscopy

31 32 33

1.

Introduction Membrane bioreactors (MBRs) are becoming more popular in treating wastewater

34

for water reuse or water reclamation. The use of membranes in the MBRs enables high

35

mixed liquor suspended solids (MLSS) concentration reducing the size of the plant, and

36

their complete solid rejections without additional settling process, thus enhancing the

37

effluent quality compared to conventional biological process (Drews, 2010; Le-Clech et

38

al., 2006).

39

MBRs have been highlighted by their potential for improved removal of

40

hazardous substances from wastewater, essentially eliminating solids in the effluent, and

41

their disinfection capabilities. As such, their importance for the aquatic environment in

42

terms of removal efficiencies and water effluent qualities that can be obtained are well

43

recognized. For this reason, ultrafiltration (UF) membranes are increasingly being applied

44

in MBRs for wastewater treatment plants (WWTPs). Especially in regions with no

45

suitable receiving waters or where treated wastewater is used for groundwater infiltration,

46

UF based MBRs are one of the alternatives to conventional treatment systems. Membrane

47

fouling is considered one of the main drawbacks in operation of MBR systems for

2

48

wastewater treatment, resulting in decreasing permeate flux during filtration and

49

subsequently increase in operation costs of MBR processes (Naessens et al., 2012). In

50

order to control membrane fouling and maintain a stable flux, various forms of air

51

scouring and regular backflushing, disinfection, and chemical cleaning are employed,

52

further increasing the operation costs of MBR processes (Meng et al., 2009).

53

When operating UF filtration systems under dead-end with ultra-low pressure

54

(40–100 mbar) conditions, there is a potential to maintain a stable flux without any

55

backflushing, regular chemical cleaning, and if designed as a gravity-driven system no

56

external energy supply is required for the membrane filtration component during long-

57

term operation. Low-pressure operation of gravity-driven systems is quite feasible using a

58

water head at height differences of 40–100 cm (Peter-Varbanets et al., 2010).

59

In wastewater applications of MBR, fouling is still recognized to be complex and

60

ambiguous. The fouling formation that occurs on membranes in a gravity driven

61

submerged MBR (GD-SMBR) configuration is also not well understood, considering that

62

this mode of operation is a relatively new concept, beneficial for small-scale WWTPs

63

(Fan and Huang, 2002; Meng et al., 2008). However, understanding the mechanisms of

64

fouling formation and their in-depth characterization will help to increase the capacity of

65

GD-SMBR systems.

66

In addition, precise understanding of the biofilm structure and architecture is

67

necessary to select the optimal operating conditions for mitigation and control of fouling

68

in MBR systems (Lee et al., 2008). In literatures, different approaches are proposed to

69

study the biofilm architecture, however, most of them are destructive involving the

70

removal of the membrane from the system and sampling of the fouled surface. Hence, the

3

71

real-time monitoring or assessment is highly required. Recently, optical coherence

72

tomography (OCT) has gained popularity in the study of biofouling in membrane

73

filtration systems because it allows an in-situ and non-destructive assessment and

74

analysis. Moreover, OCT is being highlighted as a suitable fouling monitoring technique

75

since the fouling formation and response is extremely dynamic (Gao et al., 2013; Li et al.,

76

2016). This study, therefore, investigated biofouling formation and evolution on flat

77 78

sheet membrane in a GD-SMBR. It was focused on the characterization of biofouling

79

formation on a small-scale flat sheet membrane GD-SMBR for treating wastewater

80

secondary effluent (WWSE). An integrated approach involving different techniques was

81

applied aiming to have a better understanding of the fouling formed and to validate the

82

data obtained from the OCT device installed. Three-dimensional (3D) biomass

83

morphology was analyzed through OCT using non-destructive in-situ scanning of the

84

submerged membrane module directly in the tank. At the end of the experimental period,

85

the amount and nature of the biomass observed was compared with results obtained from

86

commonly used final destructive autopsy techniques. In particular, biomass and organic

87

characterization of the biofilm extracted from the fouled membrane were carried out;

88

including three different advanced scanning electron microscopic techniques to help gain

89

an in-depth understanding of the biofilm structure relative to results obtained from the

90

OCT scans.

91 92

2.

Materials and methods

93

2.1

Gravity driven submerged membrane bioreactor (GD-SMBR)

4

94 95

2.1.1 Set-up and operation A flat sheet membrane module was submerged in a Plexiglas tank of 0.70 m x

96

0.15 m x 0.02 m (height x width x depth), with a wall thick 0.005 m. The system was run

97

under a gravity driven mode of operation by maintaining a specific water head above the

98

membrane module. A customized flat sheet membrane, manufactured by MemSis

99

Turkey, was submerged in the tank (Figure 1a), and referred to as a gravity driven

100

submerged membrane bioreactor (GD-SMBR). The module was constructed with a

101

Polysulfone (PS) UF membrane (PHILOS, Korea) with molecular weight cut-off

102

(MWCO) of 20 KDa. Effective membrane area was 0.0045 m2 (0.09 m × 0.05 m).

103

The system was representative of a SMBR, treating synthetic secondary

104

wastewater effluent with a chemical oxygen demand (COD) of 7.5 mg/L. The detailed

105

characteristics of feed water can be found elsewhere (Wei et al., 2014). The feed solution

106

was refreshed every week.

107

The feed was pumped continuously to keep a constant level of water in the tank

108

using a level regulator, resulting in a constant pressure head of 0.50 m above the

109

membrane (corresponding to 50 mbar or 5.0 kPa). A 1 mL of activated sludge was added

110

at the beginning of the operational period into the filtration tank as inoculum for the

111

biofilm formation and biomass deposition on the membrane. Experiment was started after

112

one day of stabilization time. The whole system was covered with aluminum foil to avoid

113

the growth of algae. The GD-SMBR system was continuously operated for 43 d.

114

2.1.2 Water flux

115 116

The permeation water (or membrane effluent) was collected from the bottom of the tank in a container on a scale. Water flux was obtained by dividing the weight (kg) of

5

117

the collected permeation water over the effective membrane area (0.0045 m2) and

118

corrected time (h). Water flux (kg/m2 h) was calculated on a daily basis. The normalized

119

flux was defined as the actual flux divided by the initial flux.

120 121 122

2.2 In-situ biofilm monitoring or observation Optical coherence tomography (OCT) was used for in-situ biofilm monitoring in

123

GD-SMBR as shown in Figure 1a. This enabled the observation of biofilm formation on

124

the membrane without removing the module from the tank, allowing a non-destructive

125

and real-time assessment of the biofilm development. Fifteen positions (1.5 cm x 1.5 cm

126

each) were marked at the beginning of the experiment on the virgin membrane (Figure

127

1b). After 43 d of operation period, a stable biofilm had formed upon which a three-

128

dimensional (3D) biofilm analysis was carried out using the OCT for each marked

129

position. By using this approach, analysis of the biofilm covering 75% of the whole

130

membrane area (33.75 cm2 out of 45 cm2) could be achieved. In addition, it was possible

131

to calculate the actual biofilm volume (or biovolume) non-destructively through the OCT

132

scans. Detailed OCT procedure can be found in section 2.4 below.

133 134 135

2.3 Biofilm characterization Once the operation was completed, the membrane module was carefully removed

136

from the GD-SMBR. Each marked position (15 positions) was then cut from the

137

membrane module for characterization of the biomass deposited on the membrane

138

(Figure 1b).

139

6

140

2.3.1 Biofilm extraction

141

For extraction of the biofilm from the fouled membrane, twelve (12) of the fifteen

142

(15) coupons (2.25 cm2 each) were taken from the flat sheet membrane and placed in 12-

143

capped tubes containing 30 mL of autoclaved 1X phosphate-buffered saline (PBS,

144

containing disodium hydrogen phosphate, sodium chloride, potassium chloride and

145

potassium dihydrogen phosphate). The tubes were then placed in an ultrasonic bath.

146

Sonication was repeated three times (for 5 min each) followed by 10 s of vortex mixing.

147

Finally, the aliquots were collected for microbial analyses and remaining solution was

148

filtered through a 0.45 µm pore size syringe filter for organic analyses. The detailed

149

procedure can be found elsewhere (Jeong et al., 2016).

150

2.3.2 Microbial analysis

151

Active biomass in the biofilm on the membrane was determined by measuring the

152

adenosine triphosphate (ATP) concentration in extracted biofilm samples. ATP was

153

measured using the ATP Analyzer (Celsis, USA), which employs the luciferin−luciferase

154

bioluminescence reaction method. Total ATP was divided into extracellular and

155

intracellular (or microbial) ATP by filtering the sample through a 0.22 µm filter using a

156

sterile 25-mL syringe. In other words, extracellular ATP was filtered through the

157

membrane while microbes were retained on the 0.22 µm filter. Thus, microbial ATP was

158

calculated by subtracting extracellular ATP from the total ATP (Jeong et al., 2013).

159

Actual total number of cells in the biofilm extracts was measured using flow

160

cytometric analysis (FCM; Accuri C6 Flow Cytometer, USA). Cells in samples were

161

stained with a cell membrane permeable nucleic acid fluorescent probe (1×SYBR Green

7

162

I, incubation time 15−20 min) followed by a flow cytometric method adapted from

163

Nocker et al. (2010).

164

2.3.3 Organic analysis

165

Total organic carbon (TOC) concentration of the biofilm extracts was measured

166

with a TOC analyzer (TOC-V-CSH, Shimadzu, Japan). The non-purgeable organic

167

carbon (NPOC) method was used (Xu et al., 2006).

168

The dissolved organic carbon (DOC) concentration of the extracted biofilm

169

samples was characterized using liquid chromatography with organic carbon detection

170

(LC-OCD, DOC-Labor, Germany) after filtering samples through a 0.45 µm filter. In the

171

LC-OCD measurement, samples were supplied by a mobile phase (phosphate buffer, 2.5

172

g/L KH2PO4 (Fluka, USA) + 1.5 g/L Na2HPO4
2H2O (Fluka, USA)) at a flow rate of 1.5

173

mL/min to the chromatographic column, which is a weak cation exchange column based

174

on polymethacrylate. A dual column was used with 2,000 µL injection volume and 180

175

min retention time. The measurements were done in duplicate, and the mean value was

176

reported (variation was less than 5%). The main fractions identified by the LC-OCD

177

technique are biopolymers (BP), humic substances (HS), building blocks (BB), low

178

molecular weight (LMW) acids and neutrals (Huber et al., 2011).

179

In this study, the BP was divided into polysaccharides (PS) and protein (PN),

180

where the PN concentration was estimated based on the concentration of organic nitrogen

181

as detected by the organic nitrogen detector (OND). More details can be found elsewhere

182

(Jeong et al., 2014).

183 184

2.4 Optical coherence tomography (OCT)

8

An OCT (Thorlabs GANYMEDE spectral domain OCT system with a central

185 186

wavelength of 930, Thorlabs, GmbH, Dachau, Germany) equipped with a 5X telocentric

187

scan lens (Thorlabs LSM 03BB) was used to investigate the biofilm formation (or

188

growth) on a flat sheet membrane in a GD-SMBR. The OCT probe was mounted on a

189

motorized 3-axis stage slide connected to a Velmex Motor Controllers VMX. This

190

allowed moving the camera in a system of coordinates with an accuracy of 5 µm along

191

the X and Y axes and 0.79 µm along the Z axis. Each marked position of 15 mm x 15

192

mm was scanned at the end of the experiment (43rd day of operation) with a

193

corresponding axial resolution of 2.7 µm and lateral resolution of 20 µm, with a resulting

194

OCT stack of 600 × 600 × 900 (x × y × z).

195 196

2.5 Image analysis and biofilm descriptor calculation

197

Image analysis of the OCT scans was preprocessed using FiJi software. A multi-

198

sequence step was applied: (i) the images were filtered, (ii), contrast and brightness were

199

adjusted and (iii) the images were thresholded and binarized. The physical proprieties of the biofilm were calculated using a customized

200 201

MATLAB code. Each pixel corresponds to a certain length related to the setting

202

acquisition. The mean biofilm layer thickness (Z in µm) was calculated measuring the

203

number of pixels from the membrane to the top layer. The absolute roughness coefficient

204

(Ra in µm) and the relative roughness coefficient (R’a) were calculated using following

205

equations (Eq. (1) – (3)) reported by Derlon et al. (2012).

206 207




′ =  ∑ 


(1)

9



208

 =  ∑  |
 −
′|

209

′ =  ∑  



|| 

(2)



(3)

210 The volume of the 3D OCT stack was given by calculating the biomass

211 212

thresholded voxels for each slice in the volume fraction (Eq. (4)). The biovolume (B) can

213

be expressed in mm3/cm2 as the total biomass volume (V) divided by the monitored area

214

(A).

215 216

B=

V A

(4)

217 218 219 220

A color lookup table was applied to the 2D grey scale images only for visualization (Figure S1a). The commercial software Avizo was used for the analysis and visualization of a

221

single 3D OCT data set. The use of this software allows importing the original set of data

222

without any adjustment. The membrane and the biomass were defined using Avizo’s

223

segmentation editor. A 3D volume corresponding to the biomass was created to visualize

224

the selected structure and to perform a detailed image analysis.

225 226 227

2.6 Scanning electron microscope (SEM) In this study, three of the fifteen marked positions were imaged with three

228

different SEM techniques (environmental SEM (ESEM), cryo-SEM, and freeze-drying

229

SEM) to gain a better understanding of the biofilm on the membrane surface.

10

230 231 232

2.6.1 Environmental SEM (ESEM) FEI Quanta 200FEG SEM equipped with a cold stage and variable chamber

233

pressure was used to observe the biofilm on the membrane surface in normal hydrated

234

condition. The stub was then mounted on a cold stage set at 2.0 °C. To maintain the

235

sample in hydrated state, the SEM chamber humidity was held at 100%. The sample was

236

captured at accelerating voltage of 5-10 kV and working distance of 5-6 mm with

237

gaseous secondary electron detector (GSED).

238

2.6.2 Cryo-SEM

239

To examine the sample in frozen hydrated state, low temperature treatments were

240

carried out using a Quorum PP2000T cryo-transfer system (Qurorum Technologies, UK)

241

that was fitted to an FEI Nova Nano630 SEM with a field emission electron source and

242

through the lens electron detectors. The stub was then secured on the specimen holder,

243

rapidly plunged into liquid nitrogen slush at -210 °C and under vacuum transferred into

244

the PP2000T preparation chamber pre-cooled at -180°C and allowed to equilibrate for 5

245

min. The sample was then sublimed at -90 °C for 2 min to eliminate any condensed ice

246

from the surface which might occur during the transfer. To avoid charging problems, the

247

sample temperature was then reduced to -150°C and sample was sputter-coated with 5

248

nm-thick platinum in an argon atmosphere. The sample was then transferred to the SEM

249

cryo stage which was held at -130°C. The imaging was performed using an accelerating

250

voltage of 2-5 kV, working distance of 5 mm and with cryo stage held at -130 °C.

251

2.6.3 Freeze-drying SEM

11

Freeze-drying was used to remove the water content and to preserve the structure

252 253

of the biofilm with minimal damage before SEM imaging. Samples were frozen in

254

nitrogen slush at -210 °C and transferred to a freeze drier (K775X, Quorum

255

Technologies, UK) with stage held at -120 °C. For freeze-drying the sample, the

256

temperature of the stage was increased gradually from -120 °C to room temperature over

257

48 h under high vacuum. To dissipate charging, the samples were then coated with 5 nm-

258

thick iridium in a sputter coater (Q150T, Quorum Technologies, UK). A Magelan

259

400SEM (FEI, The Netherlands) equipped with secondary electron detector (Everhart-

260

Thornley detector) and through the lens detector was used. The images were taken at an

261

accelerating voltage of 1-3 kV and working distance of 2-4 mm.

262 263

3.

Results and Discussion

264

As stated by Wu et al. (2012), a better understanding of the foulants development

265

and their interaction on membranes can be achieved through the combined use of various

266

techniques. The combination of assorted techniques can provide more insights into

267

membrane fouling mechanisms on the submerged membrane. In this study, an integrated

268

approach involving the use of different techniques (Table 1) was employed to

269

characterize the biofilm or biofouling on the membrane in a GD-SMBR, which will be

270

discussed in following sections.

271 272 273 274

3.1

In-situ biofilm observation using OCT A key aspect of biofilm studies involves the analysis of biofilm structure, which

can predict the biofilm behavior, and thus, the impact on membrane filtration

12

275

performance (Halan et al., 2012). Biofilm structure analysis implicates the use of imaging

276

technique (i.e. confocal laser scanning microscopy, CLSM), where the sample is

277

destructively removed from the membrane (Valladares Linares et al., 2016). For the ex-

278

situ analysis, the sample has to be removed from the reactor, dried or stained before

279

imaging. Therefore, the biofilm structure will be altered by ex-situ analysis associated

280

with drying (or dehydration) (for SEM) or staining (for CLSM). On the other hand, the

281

OCT can be used as a tool to acquire non-destructively online information about the

282

biofilm development at a meso-scale level (Wagner et al., 2010).

283

In this study the OCT was used as the initial step in the characterization of the

284

biofilm on a flat sheet membrane. The approach adopted allowed investigating a selected

285

membrane area and coupling the structural information acquired non-destructively with a

286

fouling composition analysis acquired destructively by membrane autopsy. This way it is

287

possible to combine information about the biomass structure with fouling analysis for a

288

specific and wide area. Despite that CLSM provides a higher image resolution, the

289

observed area is too small to fully explain and correlate the biofilm behavior on a larger

290

area (Sun et al., 2011). As stated by a previous study (Morgenroth et al., 2009), the meso-

291

scale is essential to understand the biofilm behavior. In the present study, the customized

292

membrane module (9 cm x 5 cm) was scanned with OCT in fifteen different positions,

293

each 2.25 cm2, directly in the tank. The acquisition time for each OCT scan was only 50

294

s, a significantly less amount of time for each scan required compared to other commonly

295

used imaging techniques used to study biofilm structures.

296 297

An OCT cross sectional scan of one of the fifteen (15) positions is shown in Figure S1a. The biomass layer (dark orange) is homogeneously deposited on the

13

298

membrane surface (bright orange). The biovolume of the fifteen positions was calculated

299

from each OCT 3D dataset using a customized MATLAB code. The 3D OCT dataset can

300

be directly imported into any advanced 3D analysis software for scientific data. Figure

301

S1b shows an analyzed area of the rendered file in Avizo®. The 3D analysis software

302

visualizes a 3D object allowing a detailed observation of biomass morphology. Once the

303

OCT stack is segmented, it is possible to create a sub-volume of the biomass, and to

304

perform a specified structural analysis.

305

Based on the results obtained, OCT is considered a suitable technique to study the

306

biofilm distribution in membrane filtration processes, and can be used as an aid for final

307

characterization of fouling. The possibility of acquiring scans in the centimeter (cm)

308

range allows advanced assessment of the fouling distribution and homogeneity over a

309

larger membrane surface. Furthermore, the OCT acquisition is very fast, does not require

310

any sample preparation, and can be performed in-situ under continuous operation. Due to

311

the complex nature of biofouling in MBRs, OCT can potentially be used as the first step

312

before conducting a membrane autopsy thereby determining the number of samples

313

required for a representative membrane autopsy.

314 315

3.2 Fouling distributions

316

In this study, a detailed fouling analysis was performed aiming to evaluate the

317

biofilm distribution on a flat sheet membrane in a GD-SMBR operated for 43 d. GD-

318

SMBR is a novel MBR configuration where only the pressure head above the membrane

319

module drives the filtration with no additional energy required for suction. Flux under

320

gravity driven operation has been reported to tend to stabilize over time (Peter-Varbanets

14

321

et al., 2010). Operating the system with low permeate flux has also been considered an

322

approach to reduce the membrane biofouling. Peter-Varbatens et al. (2010) linked the

323

flux stabilization to the biofouling layer properties, resulting in a constant thickness and

324

hydraulic resistance being achieved after 7 d of operation. Tests were conducted with

325

different types of feed water. Recently, Fortunato et al. (2017) monitored and assessed in

326

a real-time the fouling layer evolution in gravity driven membrane filtration system. A

327

similar behavior was observed in this study where a stable flux is achieved after 30 d

328

operation (Figure S2).

329

Analyses on fouling structure and fouling composition in this study are presented

330

in Table 2. Image analysis from OCT scans showed that a highly homogeneous layer of

331

biomass formed over the membrane after 43 d of operation, without backwash or

332

relaxation. The biomass presented a regular morphology along the whole membrane,

333

where analysis of the biomass on 75 % of the available space had an average biofilm

334

thickness of 171 µm (STD 5 µm), and average absolute roughness value of 5 µm (STD 1

335

µm) (Figure 2). Under low-pressure gravity-driven operation, the biomass expanded over

336

time reaching a kind of steady state (homogeneous) covering uniformly all the available

337

area on the membrane. In this specific case, where the biomass covered 100% of the

338

membrane, the biovolume and the average biofilm thickness are linearly correlated as

339

shown in Figure 2.

340 341 342 343

3.3 Biomass analysis After 43 d of operating the GD-SMBR, a membrane autopsy was performed on the fifteen marked positions used for the OCT analysis. Membrane autopsy is a

15

344

commonly used tool to investigate and understand the fouling nature in membrane

345

filtration process, where different approaches and analytical targets are reported in

346

literature. Vrouwenvelder et al. (1998) used ATP and TOC as biofouling indicators in

347

spacer filled channel studies. In MBR filtration systems, the extracellular polymeric

348

substances (EPS) has received considerable attention as it is considered one of the main

349

significant factors in biofouling (Chang et al., 2002; Ishiguro et al., 1994). However,

350

detailed analysis (i.e. fractionation) of the composition of organic foulants deposited on

351

membranes in MBR systems is lacking, especially for GD-SMBR.

352

In Table 2, the autopsy values support the result provided by the structural

353

analysis, confirming the homogeneity of the fouling layer on the membrane surface.

354

Results are expressed according to the common practice as value by the area (cm2) of the

355

analyzed membrane (Table 2b and 2c). In this study non-destructive structural analysis

356

was performed precisely for each position on the membrane in the tank, thereby making

357

it possible for the first time to relate all the values based on the biomass volume (Table

358

2a). In fact, the proposed approach enabled acquiring the structural information and

359

fouling composition on the exact same sample. This is the first time that the autopsy

360

values can be expressed as values/volume (X/mm3 ). In addition, intracellular ATP (intra

361

ATP) concentration by filtering the sample through the 0.22 µm filter could be

362

distinguished by excluding the extracellular ATP (extra ATP) excreted by cell lysis.

363

Results indicate that 1.6±0.4e+06 cells/mm3 can be related to 60±7 pg/mm3 of intra ATP.

364

Moreover, the proposed approach provides a tool to better understand the biomass nature,

365

allowing evaluating the biological activity of the biomass deposited on the membrane,

366

and normalizing the analytical values for the biomass volume. For example, it allowed

16

367

quantifying the number of cells for a volume of biomass deposited on the membrane.

368

Hence, it could be used to distinguish between different types of fouling.

369

In the biofilm membrane reactor (BF-MBR), one of the major foulants is

370

constituted by the submicron particles (Ivanovic et al., 2008). Recently, Matar et al.

371

(2016) found the dominance of LMW-N in the characterization of MBR fouling on

372

different membrane surfaces. During the ultra low pressure (gravity-driven) UF, LMW

373

organics were found to be one of the dominant factors responsible for the membrane

374

fouling (Peter-Varbanets et al., 2011). As shown in Figure 3, the LC-OCD results

375

confirm the abundance of LMW-N in the foulant composition on the fouled membrane in

376

GD-SMBR. The GD-SMBR system in this study combines the bioreactor with a low-

377

energy membrane-based separation. Thus, the biological degradation of organic matter

378

occurs in the reactor due to the high retention time provided by the GD-SMBR system,

379

generating more LMW organic compounds. It was reported that the LMW organics,

380

released from substrate metabolism and biomass decay, were found to have a great

381

impact on organic and biofouling (Barker and Stuckey, 1999).

382

Another reason of the organic distribution analyzed by LC-OCD is due to the use

383

of a UF membrane (20 kDa of MWCO) in the GD-SMBR. As reported by Raspati et al.

384

(2011), the hydrophobic-hydrophilic interaction represented the dominant mechanism in

385

natural organic matter (NOM)-membrane interaction and fouling. The molecular shape

386

(structure) and the MW (size) constitute the important factors in affecting the flux

387

decline. The UF membrane used in this study (20 kDa) has smaller pore size than these

388

macromolecular organic forms, so it generally may be caused by sequential or

389

simultaneous processes of surface (gel-layer) coverage during filtration (i.e. cake

17

390

filtration). Macromolecules of a relatively hydrophilic character (e.g. PS and PN) were

391

effectively rejected by UF membranes, suggesting that macromolecular compounds

392

and/or colloidal organic matter in the hydrophilic fraction may be a problematic foulant

393

on UF membranes (Carroll, 2000; Fan et al., 2001; Lin, 2000). PS and/or PN in

394

macromolecular and/or colloidal form have been reported in literature (Filloux et al.,

395

2012; Lee et al., 2004) to contribute to significant organic fouling. Interestingly, PN

396

contents in BPs detected on the membrane in the GD-SMBR studied were about 2-3

397

times higher than PS. As shown in Figure 3a, the BPs constituted 12% of the total DOC

398

of the foulant. This result is due to the separation effect provided by the UF membrane.

399

However, it was found that filtration by UF membranes removed negligible DOC,

400

particularly the MW fraction of organic foulants (88%) with a smaller size than the UF

401

membrane pore size (20kDa). In fact, the MWs of HS and BB are approximately 1kDa

402

while LMW organics are even less than 350 Da (Huber et al., 2011; Jeong and

403

Vigneswaran, 2015). Hence, theoretically the UF membrane cannot remove these

404

compounds. However, it is known that the fouling by high MW organic occurs during UF

405

filtration, based on adsorption inside the membrane pores and on the membrane surface

406

(Cho et al., 2000). This effect can results in a decreased passage path for water molecules

407

while it can also lead to an increase in removal efficiency of medium-MW and low-MW

408

compounds (i.e. HS/BB and LMW-A and LMW-N). The superior removal efficiency was

409

achieved in the phase of the stable flux (i.e. 70% DOC with 88% BPs, 93% HS, 60% BB

410

and 58% LMW-N). The cake layer (or biofilm) on the membrane therefore includes these

411

organic compounds. As observed in the flux pattern, the transition into the cake filtration

412

was relatively fast. This indicates that the dominant fouling mechanism in GD-SMBR

18

413

was cake filtration, which is mainly by LMW organics (LMW-A=11±1% and LMW-

414

N=55±5%) as shown in Figure 3a and b. Interestingly, it was found to be homogeneous

415

organic fouling deposition on the membrane (Figure S3), which was similar to the

416

biomass structural analysis by OCT. This indicates that GD-SMBR led to an even organic

417

fouling distribution on the membrane.

19

418 419

3.4 Biomass imaging In this study the biofouling structure was analyzed directly in the tank using OCT,

420

acquiring information about the biomass structure and its distribution on the membrane.

421

The OCT can obtain a depth analysis based on backscattered light allowing the

422

acquisition of structural information such as biofilm thickness and biovolume. On the

423

other hand, the OCT is only able to provide information about the total structure and

424

cannot distinguish between components in the structure. Hence, OCT is not sufficient to

425

provide qualitative information about the nature and composition of the fouling layer

426

(Halan et al., 2012).

427

SEM is one of the most common methods used to characterize fouling in

428

membrane filtration system, since it allows a direct observation of the fouling layer with

429

a high resolution (Malaeb and Ayoub, 2011). Hence, in this study, three different SEM

430

techniques: ESEM, freeze-drying SEM and cryo-SEM, were employed as conventional

431

air-dried SEM had a limitation for the study of biofouling considering that a biofilm

432

consists of a high portion of water.

433

SEM images can be found in supplementary information (Figure S4). ESEM can

434

be used to image the biofilm in normal hydrated conditions. With this technique, the

435

biofilm structure can be observed in a state as close as possible to the in situ environment

436

in the reactor, but without the pressure gradient across the biofilm. The ESEM requires

437

minimum sample preparation and therefore, the sample can be imaged directly after the

438

membrane autopsy. As mentioned in section 2.5.1, the biofilm could be imaged using

439

ESEM while still hydrated. The ESEM analysis at lower magnification (Figure S4a)

440

confirmed the homogeneous structure observed in-situ with the OCT. ESEM imaging of

20

441

the biofilm at higher magnification showed that the hydrated surface was covered with

442

bacteria and EPS in a granular form (Figure S4b). The disadvantage of this technique,

443

however, is the lower resolution capability compared to conventional SEM.

444

Cryo-SEM is used to image the biofilm in a frozen hydrated condition. The

445

advantage of this technique is the ability to take images in an ultra-high resolution mode.

446

Cryo-SEM imaging of the biofilm shows the presence of bacteria on the surface and in

447

the matrix. As shown in Figure S4c and S4d, it is possible to notice the presence of

448

different sizes or shapes of bacteria on the surface, confirming a diverse bio-community.

449

To image the surface of a dried biofilm using conventional SEM in high vacuum

450

environment, the water content of the biomass is removed using a freeze-drying

451

technique. Unlike the air drying procedure commonly used (Baker et al., 1995), where

452

the surface tension of water damages the structure of biological samples and causes

453

collapsing of the structure (i.e. biofilm in this case), freeze-drying removes water by

454

sublimation and minimizes the damage which may occur during drying. The advantage of

455

this technique is that samples are imaged in a high vacuum environment allowing ultra-

456

high resolution imaging of the surface of the matrix. However, this technique requires a

457

freeze drying machine and long sample preparation time (more than 24 h) to remove the

458

water from the sample.

459

SEM analysis of the freeze-dried membrane showed rod-shaped bacteria attached

460

to a scaffolding of biomass (Figure S4e and S4f) indicating that the microorganisms in

461

the biomass were interconnected through the fibrils EPS matrix. Freeze-drying SEM

462

imaging also showed large amounts of voids in the dried biomass matrix confirming

463

previous findings that water is the major component of biomass (99%) (Jing-Feng et al.,

21

464

2012). Moreover, from the SEM image, it was possible to notice that the EPS matrix

465

constitutes the majority of the dry mass. As stated by Flemming et al. (2010), the EPS

466

matrix can account for over 90% of the dry mass while the microorganisms account for

467

less than 10%. This specific SEM technique is suitable for the analysis of the EPS matrix,

468

in fact it is the only technique that allows imaging the dry fraction of the biomass

469

deposited on the membrane surface preserving the fouling structure. Although this

470

technique improves the drying condition by eliminating the surface tension during

471

sublimation, the artifacts associated with drying are still present. Thus, the SEM image of

472

the bacteria showed shrinkage and folding of its surface (Figure S4f).

473

The SEM is a high-resolution technique capable to provide detailed information

474

about the fouling surface. The combination of three different SEM techniques provides

475

more details about the nature of the fouling layer deposited on the membrane. In this

476

study, the SEM investigation confirms the presence of a homogeneous fouling layer on

477

the flat sheet membrane, although the portion of fouled membrane imaged is too small

478

with respect to the OCT. Furthermore, the SEM doesn’t provide depth resolved

479

information about the fouling layer necessary to study the fundamental biofilm

480

information (i.e. biofilm structure and hydraulic resistance). Moreover, the SEM is time

481

consuming, requires qualified personnel and can only be performed ex-situ destructively.

482 483 484

4.

Conclusions The OCT was used as a new technique for final characterization of fouling on the

485

membrane in GD-SMBR. The biomass layer was homogeneously deposited on the

486

analyzed membrane surface with an average biofilm thickness value of 171±5 µm, and

22

487

absolute roughness of 5±1 µm. The dominant fouling mechanism in a GD-SMBR was

488

found to be a cake layer deposition, comprised mainly of LMW organics. The use of

489

three different SEM techniques provided a better understanding of the biofilm formed.

490

The freeze-drying SEM allowed imaging the biofilm matrix in the preserving of fouling

491

structure.

492 493

Acknowledgement

494

This study was supported by funding from King Abdullah University of Science and

495

Technology (KAUST).

496 497

References

498

1.

Baker, J., Stephenson, T., Dard, S., Cote, P., 1995. Characterisation of fouling of

499

nanofiltration membranes used to treat surface waters. Environ. Technol. 16, 977–

500

985.

501

2.

Barker, D.J., Stuckey, D.C., 1999. A review of soluble microbial products (SMP)

502

in wastewater treatment systems. Water Res. 33, 3063–3082. doi:10.1016/S0043-

503

1354(99)00022-6

504

3.

Carroll, T., 2000. The fouling of microfiltration membranes by NOM after

505

coagulation treatment. Water Res. 34, 2861–2868. doi:10.1016/S0043-

506

1354(00)00051-8

507

4.

Chang, I.-S., Le Clech, P., Jefferson, B., Judd, S., 2002. Membrane Fouling in

508

Membrane Bioreactors for Wastewater Treatment. J. Environ. Eng. 128, 1018–

509

1029. doi:10.1061/(ASCE)0733-9372(2002)128:11(1018)

23

510

5.

Cho, J., Amy, G., Pellegrino, J., 2000. Membrane filtration of natural organic

511

matter: factors and mechanisms affecting rejection and flux decline with charged

512

ultrafiltration (UF) membrane. J. Memb. Sci. 164, 89–110. doi:10.1016/S0376-

513

7388(99)00176-3

514

6.

Derlon, N., Peter-Varbanets, M., Scheidegger, A., Pronk, W., Morgenroth, E.,

515

2012. Predation influences the structure of biofilm developed on ultrafiltration

516

membranes. Water Res. 46, 3323–3333.

517

doi:http://dx.doi.org/10.1016/j.watres.2012.03.031

518

7.

Drews, A., 2010. Membrane fouling in membrane bioreactors—Characterisation,

519

contradictions, cause and cures. J. Memb. Sci. 363, 1–28.

520

doi:10.1016/j.memsci.2010.06.046

521

8.

Fan, B., Huang, X., 2002. Characteristics of a Self-Forming Dynamic Membrane

522

Coupled with a Bioreactor for Municipal Wastewater Treatment. Environ. Sci.

523

Technol. 36, 5245–5251. doi:10.1021/es025789n

524

9.

Fan, L., Harris, J.L., Roddick, F.A., Booker, N.A., 2001. Influence of the

525

characteristics of natural organic matter on the fouling of microfiltration

526

membranes. Water Res. 35, 4455–4463. doi:10.1016/S0043-1354(01)00183-X

527

10.

Filloux, E., Labanowski, J., Croue, J.P., 2012. Understanding the fouling of

528

UF/MF hollow fibres of biologically treated wastewaters using advanced EfOM

529

characterization and statistical tools. Bioresour. Technol. 118, 460–468.

530

doi:10.1016/j.biortech.2012.05.081

531 532

11.

Flemming, H.-C., Wingender, J., 2010. The biofilm matrix. Nat Rev Micro 8, 623–633.

24

533

12.

Fortunato, L., Qamar, A., Wang, Y., Jeong, S., Leiknes, T., 2017. In-situ

534

assessment of biofilm formation in submerged membrane system using optical

535

coherence tomography and computational fluid dynamics. J. Memb. Sci. 521, 84–

536

94. doi:10.1016/j.memsci.2016.09.004

537

13.

Gao, Y., Haavisto, S., Tang, C.Y., Salmela, J., Li, W., 2013. Characterization of

538

fluid dynamics in spacer-filled channels for membrane filtration using Doppler

539

optical coherence tomography. J. Memb. Sci. 448, 198–208.

540

doi:10.1016/j.memsci.2013.08.011

541

14.

Halan, B., Buehler, K., Schmid, A., 2012. Biofilms as living catalysts in

542

continuous chemical syntheses. Trends Biotechnol. 30, 453–465.

543

doi:10.1016/j.tibtech.2012.05.003

544

15.

Huber, S.A., Balz, A., Abert, M., Pronk, W., 2011. Characterisation of aquatic

545

humic and non-humic matter with size-exclusion chromatography--organic

546

carbon detection--organic nitrogen detection (LC-OCD-OND). Water Res. 45,

547

879–85. doi:10.1016/j.watres.2010.09.023

548

16.

Ishiguro, K., Imai, K., Sawada, S., 1994. Effects of biological treatment

549

conditions on permeate flux of UF membrane in a membrane/activated-sludge

550

wastewater treatment system. Desalination 98, 119–126. doi:10.1016/0011-

551

9164(94)00136-7

552

17.

Ivanovic, I., Leiknes, T., Ødegaard, H., 2008. Fouling control by reduction of

553

submicron particles in a BF-MBR with an integrated flocculation zone in the

554

membrane reactor. pp. 1871–1883. doi:10.1080/01496390801974704

555

18.

Jeong, S., Naidu, G., Vigneswaran, S., 2013. Submerged membrane adsorption

25

556

bioreactor as a pretreatment in seawater desalination for biofouling control.

557

Bioresour. Technol. 141, 57–64. doi:10.1016/j.biortech.2013.01.021

558

19.

Jeong, S., Naidu, G., Vollprecht, R., Leiknes, T., Vigneswaran, S., 2016. In-depth

559

analyses of organic matters in a full-scale seawater desalination plant and an

560

autopsy of reverse osmosis membrane. Sep. Purif. Technol. 162, 171–179.

561

doi:10.1016/j.seppur.2016.02.029

562

20.

Jeong, S., Rice, S.A., Vigneswaran, S., 2014. Long-term effect on membrane

563

fouling in a new membrane bioreactor as a pretreatment to seawater desalination.

564

Bioresour. Technol. 165, 60–68. doi:10.1016/j.biortech.2014.03.098

565

21.

Jeong, S., Vigneswaran, S., 2015. Practical use of standard pore blocking index as

566

an indicator of biofouling potential in seawater desalination 365, 8–14.

567

doi:10.1016/j.desal.2015.02.015

568

22.

Jing-Feng, W., Zhi-Gang, Q., Zhi-Qiang, C., Jun-Wen, L., Yi-Hong, Z., Xuan,

569

W., Bin, Z., 2012. Comparison and Analysis of Membrane Fouling between

570

Flocculent Sludge Membrane Bioreactor and Granular Sludge Membrane

571

Bioreactor. PLoS One 7, e40819. doi:10.1371/journal.pone.0040819

572

23.

Le-Clech, P., Chen, V., Fane, T.A.G., 2006. Fouling in membrane bioreactors

573

used in wastewater treatment. J. Memb. Sci. 284, 17–53.

574

doi:10.1016/j.memsci.2006.08.019

575

24.

Lee, C.H., Park, P.K., Lee, W.N., Hwang, B.K., Hong, S.H., Yeon, K.M., Oh,

576

H.S., Chang, I.S., 2008. Correlation of biofouling with the bio-cake architecture

577

in an MBR. Desalination 231, 115–123. doi:10.1016/j.desal.2007.10.026

578

25.

Lee, N., Amy, G., Croué, J.-P., Buisson, H., 2004. Identification and

26

579

understanding of fouling in low-pressure membrane (MF/UF) filtration by natural

580

organic matter (NOM). Water Res. 38, 4511–4523.

581

doi:10.1016/j.watres.2004.08.013

582

26.

Li, C., Felz, S., Wagner, M., Lackner, S., Horn, H., 2016. Investigating biofilm

583

structure developing on carriers from lab-scale moving bed biofilm reactors based

584

on light microscopy and optical coherence tomography. Bioresour. Technol. 200,

585

128–136. doi:10.1016/j.biortech.2015.10.013

586

27.

Water Res. 34, 1097–1106. doi:10.1016/S0043-1354(99)00273-0

587 588

28.

Malaeb, L., Ayoub, G.M., 2011. Reverse osmosis technology for water treatment: State of the art review. Desalination 267, 1–8. doi:10.1016/j.desal.2010.09.001

589 590

Lin, C., 2000. Effects of humic substance characteristics on UF performance.

29.

Matar, G., Gonzalez-Gil, G., Maab, H., Nunes, S., Le-Clech, P., Vrouwenvelder,

591

J., Saikaly, P., 2016. Temporal Changes in Extracellular Polymeric Substances on

592

Hydrophobic and Hydrophilic Membrane Surfaces in a Submerged Membrane

593

Bioreactor. Water Res. 95, 27–38. doi:10.1016/j.watres.2016.02.064

594

30.

Meng, F., Chae, S.-R., Drews, A., Kraume, M., Shin, H.-S., Yang, F., 2009.

595

Recent advances in membrane bioreactors (MBRs): membrane fouling and

596

membrane material. Water Res. 43, 1489–512. doi:10.1016/j.watres.2008.12.044

597

31.

Meng, F., Yang, F., Shi, B., Zhang, H., 2008. A comprehensive study on

598

membrane fouling in submerged membrane bioreactors operated under different

599

aeration intensities. Sep. Purif. Technol. 59, 91–100.

600

doi:10.1016/j.seppur.2007.05.040

601

32.

Morgenroth, E., Milferstedt, K., 2009. Biofilm engineering: linking biofilm

27

602

development at different length and time scales. Rev. Environ. Sci.

603

Bio/Technology 8, 203–208. doi:10.1007/s11157-009-9163-1

604

33.

Naessens, W., Maere, T., Nopens, I., 2012. Critical review of membrane

605

bioreactor models – Part 1: Biokinetic and filtration models. Bioresour. Technol.

606

122, 95–106. doi:10.1016/j.biortech.2012.05.070

607

34.

Nocker, A., Richter-Heitmann, T., Montijn, R., Schuren, F., Kort, R., 2010.

608

Discrimination between live and dead cellsin bacterial communities from

609

environmental water samples analyzed by 454 pyrosequencing. Int. Microbiol. 13,

610

59–65. doi:10.2436/20.1501.01.111

611

35.

Peter-Varbanets, M., Hammes, F., Vital, M., Pronk, W., 2010. Stabilization of

612

flux during dead-end ultra-low pressure ultrafiltration. Water Res. 44, 3607–16.

613

doi:10.1016/j.watres.2010.04.020

614

36.

Peter-Varbanets, M., Margot, J., Traber, J., Pronk, W., 2011. Mechanisms of

615

membrane fouling during ultra-low pressure ultrafiltration. J. Memb. Sci. 377,

616

42–53. doi:10.1016/j.memsci.2011.03.029

617

37.

Raspati, G.S., Høvik, H.N., Leiknes, T., 2011. Preferential fouling of natural

618

organic matter (NOM) fractions in submerged low-pressure membrane filtration

619

34, 416–422. doi:10/5004/dwt.2011.2901

620

38.

Sun, C., Fiksdal, L., Hanssen-Bauer, A., Rye, M.B., Leiknes, T., 2011.

621

Characterization of membrane biofouling at different operating conditions (flux)

622

in drinking water treatment using confocal laser scanning microscopy (CLSM)

623

and image analysis. J. Memb. Sci. 382, 194–201.

624

doi:10.1016/j.memsci.2011.08.010

28

625

39.

Valladares Linares, R., Fortunato, L., Farhat, N.M., Bucs, S.S., Staal, M.,

626

Fridjonsson, E.O., Johns, M.L., Vrouwenvelder, J.S., Leiknes, T., 2016. Mini-

627

review: novel non-destructive in situ biofilm characterization techniques in

628

membrane systems. Desalin. Water Treat. 1–8.

629

doi:10.1080/19443994.2016.1180483

630

40.

Vrouwenvelder, H.S., van Paassen, J.A.M., Folmer, H.C., Hofman, J.A.M.H.,

631

Nederlof, M.M., van der Kooij, D., 1998. Biofouling of membranes for drinking

632

water production. Desalination 118, 157–166. doi:10.1016/S0011-

633

9164(98)00116-7

634

41.

Wagner, M., Taherzadeh, D., Haisch, C., Horn, H., 2010. Investigation of the

635

mesoscale structure and volumetric features of biofilms using optical coherence

636

tomography. Biotechnol. Bioeng. 107, 844–853. doi:10.1002/bit.22864

637

42.

Wei, C.-H., Harb, M., Amy, G., Hong, P.-Y., Leiknes, T., 2014. Sustainable

638

organic loading rate and energy recovery potential of mesophilic anaerobic

639

membrane bioreactor for municipal wastewater treatment. Bioresour. Technol.

640

166, 326–34. doi:10.1016/j.biortech.2014.05.053

641

43.

Wu, B., Fane, A.G., 2012. Microbial relevant fouling in membrane bioreactors:

642

influencing factors, characterization, and fouling control. Membranes (Basel). 2,

643

565–84. doi:10.3390/membranes2030565

644

44.

Xu, P., Drewes, J.E., Kim, T.-U., Bellona, C., Amy, G., 2006. Effect of

645

membrane fouling on transport of organic contaminants in NF/RO membrane

646

applications. J. Memb. Sci. 279, 165–175. doi:10.1016/j.memsci.2005.12.001

647

29

648

List of Tables

649

Table 1

Fouling characterization techniques used in this study.

650

Table 2

Fouling distribution on fouled membrane after 43d of operation. a)

651

Biomass architecture analysis performed on OCT scans, b) Fouling composition analysis

652

(TOC and LC-OCD), and c) Biological activity (FCM and ATP).

30

653

Table 1

Fouling characterization techniques used in this study.

Techniques

Acquired information

Advantages

Disadvantages

Target: fouling structure and architecture

OCT

Structure and cross-section

Non destructive

No information about composition

Freeze-drying SEM

Surface details

High resolution

Artifacts due to the drying

ESEM

Surface details

Hydrated sample

Lower resolution when compared to conventional or Cryo-SEM

Cryo-SEM

Surface details

Frozen hydrated state, high resolution

Labor intensive

Target: fouling composition

TOC

Total organic carbon concentration of foulant

Fast

Only total value

Fractionized values

Relatively long analysis time

Fast and sensitive

Relative value

Fast and easy

Not covered all the cell, limitation on cell size

Dissolved organic carbon concentration of foulant LC-OCD and detailed hydrophilic fraction

Target: biological activity

Live biomass concentration (total = intracellular + ATP extracellular)

FCM

Cell number (total = live + dead)

654

31

Fouling distribution on fouled membrane after 43d of operation. a) Biomass architecture analysis performed on

655

Table 2

656

OCT scans, b) Fouling composition analysis (TOC and LC-OCD), and c) Biological activity (FCM and ATP).

657

(a) Biomass architecture analysis

658

659

Average Thickness

Biovolume

Biomass Volume

Absolute Roughness

Relative Roughness

(µm)

(mm3/cm2)

(mm3)

(µm)

171±5

15.6±0.7

35.1±1.6

5±1

0.028±0.006

(n=15)

(n=15)

(n=15)

(n=15)

(n=15)

(b) Fouling composition analysis TOC

DOC

BP

HS

BB

LMW-N

LMW-A

(n=12)

(n=12)

(n=12)

(n=12)

(n=12)

(n=12)

(n=12)

133±16

112 ± 9

6.3±0.7

9.4±1.6

2.3±0.7

28.7 ± 2.6

6.0±0.7

(µg/cm2)

(µg/cm2)

(µg/cm2)

(µg/cm2)

(µg/cm2)

(µg/cm2)

(µg/cm2)

5.50±0.78

4.63 ± 0.36

0.26±0.03

0.39±0.07

0.09±0.02

1.19±0.11

0.25±0.03

(µg/mm3)

(µg/mm3)

(µg/mm3)

(µg/mm3)

(µg/mm3)

(µg/mm3)

(µg/mm3)

(c) Biological activity Live cell

Dead Cell

Intr ATP

(n=12)

(n=12)

(n=12)

3.9± 0.9e+07 2

0.2±0.1e+07

2130±400

2

(events/cm )

(events/cm )

(pg/cm2)

1.6±0.4e+06

0.9±0.3e+05

60±7

(events/mm3)

(events/mm3)

(pg/mm3)

32

660

List of Figures

661

Figure 1.

662

coherence tomography (OCT) device. b) Procedure used to characterize the fouling on

663

the flat-sheet GD-SMBR system. The biomass morphology analysis was carried-out non-

664

destructively directly in the filtration tank on the fifteen-marked position; afterwards, a

665

final autopsy was performed on twelve of the fifteen positions. Biomass imaging with

666

three different SEM techniques was performed on three of the fifteen positions.

667

Figure 2.

668

tank using OCT monitoring: a) Average thickness, b) Biovolume, and c) Relative

669

roughness.

670

Figure 3.

671

fractions at 12 positions; (b) average values in organic fractions; and (c) a LC-OCD

672

chromatogram.

a) Schematic drawing of GD-SMBR set-up combined with the optical

Biomass analysis on the biomass morphology performed directly in the

Organic foulant composition in biofilm analyzed by LC-OCD: (a) organic

33

673 a) Schematic drawing of GD-SMBR set-up combined with the optical coherence tomography (OCT) device. b)

674

Figure 1.

675

Procedure used to characterize the fouling on the flat-sheet GD-SMBR system. The biomass morphology analysis was carried-

676

out non-destructively directly in the filtration tank on the fifteen-marked position; afterwards, a final autopsy was performed

677

on twelve of the fifteen positions. Biomass imaging with three different SEM techniques was performed on three of the fifteen

678

positions

34

679

(a)

(b)

35

(c) Biomass analysis on the biomass morphology performed directly in the tank using

680

Figure 2.

681

OCT monitoring: a) Average thickness, b) Biovolume, and c) Relative roughness.

36

(a) 9 BB

8

LMW-A

OCD signal

7 6

BP

HS

LMW-N

5 4 3 2 1 0 30

50

70

90

110

130

Retention time (min)

(b) Organic foulant composition in biofilm analyzed by LC-OCD: (a) average

682

Figure 3.

683

values in organic fractions; and (b) a LC-OCD chromatogram.

684

37

685 686

38

687

Highlights

688 689

•

Gravity driven-submerged MBR (GD-SMBR) is tested for water reclamation.

690

•

An integrated approach is used to characterize the biofouling on GD-SMBR.

691

•

Accurate biofilm structure is obtained non-destructively and in-situ using OCT.

692

•

This approach provides quantitative characterization of biomass on the membrane.

693 694 695

•

This study introduces advanced imaging techniques for membrane biofilm structure.

696

39