Influence of Internal Biofilm Growth on Residual Permeability Loss in

Jan 26, 2010 - Probing Membrane Fouling via Infrared Attenuated Total Reflection Mapping Coupled with Multivariate Curve Resolution. Wei Chen ...
2 downloads 0 Views 3MB Size
Environ. Sci. Technol. 2010, 44, 1267–1273

Influence of Internal Biofilm Growth on Residual Permeability Loss in Aerobic Granular Membrane Bioreactors YU-CHUAN JUANG,† SUNIL S ADAV,† D U U - J O N G L E E , * ,†,‡ A N D J U I N - Y I H L A I § Department of Chemical Engineering, National Taiwan University, Taipei 10617, Taiwan, State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology, Harbin 150090, China, and Center of Membrane Technology, Department of Chemical Engineering, Chung Yuan Christian University, Chungli, Taiwan

Received August 12, 2009. Revised manuscript received January 8, 2010. Accepted January 13, 2010.

Membrane fouling results in flux decline or transmembrane pressure drop increase during membrane bioreactor (MBR) operation. Physical and chemical cleanings are essential to keep an MBR operating at an appropriate membrane flux. Considerable residual membrane permeability loss that cannot be removed by conventional cleaning requires membrane replacement. This study demonstrates that an internal biofilm can develop inside a hollow-fiber membrane and can probably account for up to 58.9 and 81.3% of total membrane resistance for aerobic granular MBR operated in sequencing batch reactor (SBR) mode or continuous-fed mode, respectively. The Arthrobacter sp. (accession no. AM900505 in GenBank) corresponded to internal biofilm development by polymerase chain reaction-denaturing gradient gel electrophoresis (PCRDGGE) analysis and the agar-plating technique. This study also identifies a single strain, Arthrobacter sp., generates the internal biofilm. The Arthrobacter sp. is a rod-shaped bacterium with a size close to that of membrane pores, and can secrete excess bound proteins, hence can penetrate and attach itself inside the membrane and grow. Internal biofilm growth could contribute significantly to membrane resistance during long-term MBR operation.

Introduction Membrane fouling is generally caused by plugged and blocked pores, resulting from deposition of particles and colloids on the membrane surface and precipitation of relatively smaller dissolved materials in membrane pores and on the membrane surface (1-3). Membrane fouling typically results in a flux decline or a trans-membrane pressure (TMP) drop increase across the membrane (4, 5). Some recent studies have investigated membrane fouling (6-11). Yamamoto et al. (12) concluded that concentrated sludge in a mixed liquor formed a membrane fouling layer. Soluble * Corresponding author phone: +886-2-33663028; fax: +886-223623040; e-mail:[email protected]. † National Taiwan University. ‡ Harbin Institute of Technology. § Chung Yuan Christian University. 10.1021/es9024657

 2010 American Chemical Society

Published on Web 01/26/2010

microbial products (SMP), extracellular polymeric substances (EPS), and biopolymer clusters (BPC) can cause membrane fouling (13-16). Wisniewski and Grasmick (17) demonstrated that soluble constituents in a mixed liquor accounted for >52% of total resistance during membrane filtration. Bouhabila et al. (18) indicated that the colloids fraction was the principal foulant on a membrane. Defrance et al. (19) determined that suspended solids are the key foulants in membrane bioreactors (MBR). Seo et al. (20) suggested that the hydrophobic fraction of organic compounds fouled membranes more than did the hydrophilic fraction. Proteins and polysaccharides have also been identified as major organic foulants (21-24). Ramesh et al. (25, 26) determined the potentials of individual components in EPS to foul an MBR membrane. The combined effects of colloids and organic matter on membrane fouling have been investigated (27). An operationally defined categorization scheme is based on cleaning ease. Fouling that can be removed by physical cleaning, such as backwashing, is called reversible fouling, while fouling that must be removed by chemical cleaning is called irreversible fouling (28). The primary advantage of MBR technology over other conventional biological processes is its ability to produce quality water from municipal wastewater, meeting the need to conserve water, particularly in regions that suffer from water shortages (29). The shortcomings of membrane-based systems, such as high installation costs, low permeate flux, and membrane degrading and fouling, also apply to MBR (30, 31). Aerobic granules (AG) are considered a special case of self-immobilized cells (32, 33). Liu and Tay (34) and Adav et al. (35) examined the state-of-the-art aerobic granulation process. A combination of an MBR and AG processes (AGMBR) was developed (36-39). Frequent physical (and chemical) cleaning is always required to retain an acceptable flux over time. In practice, membrane permeability is not recovered entirely by chemical cleaning. In other words, some membrane fouling cannot be eliminated, and is considered a consequence of permanent internal pore blockage that cannot be dissolved or removed. When the so-called “residual loss of membrane permeability” becomes excessive, a membrane must be replaced. This study for the first time demonstrates the formation of an internal biofilm inside a hollow-fiber membrane that could account for most residual membrane permeability loss for an AGMBR. This study also identifies a single strain, Arthrobacter sp., generates the internal biofilm. The significance of internal biofilm development inside an MBR membrane has been overlooked in previous studies.

Materials and Methods Granule Cultivation. The activated sludge obtained from local municipal wastewater treatment plant in Taipei, Taiwan. The sludge sample (1 L) was inoculated in column-type sequential batch reactors (120 × 6 cm). The reactor was fed with synthetic wastewater containing acetate as the sole carbon source with the following media composition (in g/L-1): sodium acetate, 0.4; (NH4)2SO4, 1.0; NaCl, 0.2; MgSO4 · 7H2O, 0.2; FeCl3, 0.02; CaCl2 · 2H2O, 0.01; K2HPO4, 1.65; KH2PO4, 1.35; and micronutrients, 1.0 mL L-1 (40). The organic loading rate (OLR) of this initial feed was about 1.7 kg COD m-3d-1. After 1 week the feed OLR was step increased to 9.0 kg COD m-3 d-1, 12.6 kg COD m-3 d-1 and 16.7 kg COD m3d-1 on week 2, 4, and 6 by proportionally adjusting the concentration of each chemical ingredient except buffer constitutes (K2HPO4 and KH2PO4). VOL. 44, NO. 4, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1267

Fine air bubbles for aeration and mixing were fed through the reactor bottom at superficial velocity of 3.4 cm s-1. The reactor was operated sequentially in 4 h cycles with 5 min settling time and 10 min effluent withdrawal; the remaining time in a cycle was reaction time. The volumetric exchange ratio of liquid was 50%. The matured granules having COD removal potential of 95-96% at 16.7 kg COD m-3d-1 were samples and inoculated into MBR system. MBR Setup and Operation. Two identical laboratoryscale column reactors (R2 and R4), each with a volume of 1 L, were utilized. In each reactor, a 0.03 m2 membrane module made of polyethylene hollow-fiber membranes 500 µm diameter with 0.4 µm pores was installed (Mitsubishi Rayon Co., Tokyo, Japan). With a clean water filtration flux of 5.56 × 10-6 ms-1, the TMP drop was 9.8 kPa. An increase in the TMP drop at the same flow rate was considered membrane fouling. The synthetic wastewater used in the AGMBR test has the same composition in the granule cultivation phase, except that the acetate concentration was fixed to generate an organic loading rate (OLR) of 7.0 kg chemical oxygen demand (COD) m-3d-1. The R2 reactor was operated in 4 h cycles, as stated in granule cultivation phase, and at a liquid volumetric exchange ratio of 50%. Membrane filtration for R2 occurred only with a 10 min permeate withdrawal period. On the other hand, the fresh medium was added to and withdrawn continuously from reactor R4, such that its hydraulic retention time and OLR were the same as those of R2. Thus, the membrane permeate flux for R4 was only 1/24 of that of R2. Membrane Cleaning. The membrane module was periodically removed from the reactor and cleaned using the following protocol. (1) The entire module was placed in a pool filled with clean water and rinsed for 10 min to remove sludge trapped between hollow fibers. (2) The rinsed module was then physically cleaned by backwashing with 1 L clean water flowing at 5 mL min-1, and vibrated with ultrasound at 45 kHz and 100 W in an ice bath. (3) The physically cleaned module was soaked in 1 L 0.1 M NaOH and backwashed using 1 L 0.1 M NaOH at a flow rate of 5 mL min-1, and ultrasound vibration at 45 kHz and 100 W in an ice bath. The TMP of the module before and after each washing step was measured. The fibers at the central regime of the membrane module before and after each cleaning were sampled carefully, such that the fouling layers on and inside the fibers were not disturbed. Membrane Staining and Scanning. The sampled fibers were stained and imaged using confocal laser scanning microscope (CLSM) (Leica TCS SP2 Confocal Spectral Microscope Imaging System, Gmbh, Germany) with a 10× objective using FLUOVIEW version 3.0 software. All probes except for calcofluor white (from Sigma) were purchased from Molecular Probes ( Carlsbad, CA). The SYTO 63 stain (20 µM, in 1× PBS buffer) was first added to the sample and placed on a rotary shaker (100 rpm) for 30 min. The excess dye solution was removed by washing with 1× PBS buffer. The sodium bicarbonate buffer (0.1M, pH 8.3) was added to maintain the amine group in nonprotonated form that was followed by a fluorescein isothiocyanate (FITC) solution (1 mg mL-1 in DMSO), and the sample was incubated on a rotary shaker for 30 min. After staining the excess stain was removed by PBS buffer. The sample was incubated with Concanavalin A-labeled tetramethylrhodamine conjugate solution (0.25 mg mL-1 in 1× PBS buffer) for another 60 min. After staining the excess stain was removed by PBS buffer. The sample was successively stained by calcofluor white (30 mg mL-1, in 1× PBS buffer) for 30 min. After staining the excess stain was removed by PBS buffer. The Nile red solution (1 mg mL-1, in acetone and then dilute to 10 mL with PBS buffer) was added and samples were incubated for 25 min. Excess stain was washed with PBS buffer. Before observation, 1268

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 4, 2010

FIGURE 1. Typical time courses of trans-membrane pressure of MBR tests. SBR: reactor R2; Continued: reactor R4. In SBR mode, TMP was recorded over the withdrawal period (10 min) through membrane filtration. In continuous mode, the TMP was recorded continuously. The period revealed was the no. 3 washing stage from virgin membrane. the sample was stained with SYTOX Blue (2.5 µM in PBS buffer) for 5 min without further washing. The excitation and emission wavelengths for the applied stains were detailed (41). DNA Isolation, PCR, DGGE, and Strain Isolation. The DNA from granules, and from fouling layer collected at surface and inside the hollow fiber membrane were extracted via enzymatic lysis using extraction buffer (100 mM Tris-HCl at pH 8.0, 100 mM EDTA at pH 8.0, and 1.5 M NaCl) containing Proteinase K (10 mg mL-1) (Amresco Inc., Solon, OH), as described previously (42). Polymerase chain reaction (PCR) amplification of the 16S rRNA gene was conducted using extracted DNA with forward primer P1 and reverse primer P2, as described in (42). The GCrich sequence of 40 nucleotides (GC clamp) was attached at the 5′ end of primer P1. The 50 µL PCR mixture consisted of 5 µL PCR buffer (10 ×), 2.5 µL MgCl2 (25 mM), 2.5 µL dNTP (2.5 mM), 1.0 µL (10 mM) each of the primer, 0.5 µL Taq polymerase (Promega, Madison, WI) and 100 ng of the DNA template. The PCRamplification was performed using an eppendorf mastercycler (Eppendorf AG, Hamburg, Germany) by denaturation at 94 °C for 3 min and 25 cycles consisting of 94 °C for 30 s, 55 °C for 60 s, 72 °C for 90 s, and final extension at 72 °C for 7 min. Denaturing gradient gel electrophoresis (DGGE) tests were conducted using Bio-Rad universal mutation detection system with 10% (w/v) polyacrylamide gels. The range of denaturants (100% denaturant corresponds to 7 M urea and 40% (v/v) deionized formamide) was 35-65%. The DGGE was performed at 60 °C for 18 h at 100 V. Gels were stained with ethidium bromide and photographed using a UV transilluminator. The PCR-amplified 16S rRNA was sequenced using the ABI Prism model 3730 version 3.2 DNA sequencer. Aerobic granules in the mixed liquor, the fouling layer on the membrane surface, and the fouling layer insider the membrane fiber were ascetically collected. The granules or fouling layer samples were aseptically broken and the supernatant was serially diluted with sterilized water (105-107) and plated on MBR fed reactor medium agar plate. Similarly, the surface and inside foulants were dilute with sterilized water (101-103) and plated on agar plate. Morphologically distinguished visible colonies of bacteria were selected and purified by several cycles of replating on

FIGURE 2. CLSM images for the stained hollow-fiber membrane in AGMBR. (a) After MBR test; (b) after physical cleaning; (c) after chemical cleaning. Bar ) 100 µm. Six fluorochromes: FITC (green); Con A (light blue); CW (deep blue); Nile red (yellow); red (SYTO 63); pink: (SYTOX Blue). Bottom and left: merged image. Internal biofilm was noted in all images.

FIGURE 3. Magnified CLSM images for the internal fouling layer inside the stained hollow-fiber membrane in reactor 2 after water rinsing, and physical and chemical cleaning. Bar ) 20 µm. Six fluorochromes: FITC (green); Con A (light blue); CW (deep blue); Nile red (yellow); red (SYTO 63); pink: (SYTOX Blue). Bottom and left: merged image. Bottom and middle: regime marked for magnification. medium. The selected bacterial strains were identified by PCR-amplification of 16S rRNA. Analytical Methods. The dry weights of granules, cell biomass, and volatile suspended solids (VSS) in the mixed liquor were measured using Standard Methods (43). The size of sludge granules was determined by a laser particle sizer (Mastersizer Series 2600, Malvern, UK), or by an image analysis system, depending on granule size. The isolated pure culture was prepared for scanning electron microscopy (SEM) observations by fixing it with 2.5% glutaraaldehyde for 2 h, and dehydration via successive passages through 30, 50, 75, 85, 90, 95, and 100% ethanol, followed by critical drying in a critical point dryer (HCP-2, Hitachi Co., Ltd. Tokyo, Japan).

The autoaggregation index of isolated strains was determined as described in ref 44. The size and ζ-potential of isolated cells were determined using a zeta sizer (Nano-ZS, Malvern Co., UK). The content and strains of extracellular polymeric substances (EPS) were determined using the procedures in (45). The carbohydrate content in the EPS was measured using the Anthrone method (46) with glucose as the standard. The protein content in the EPS was measured using the modified Lowry method (47) with bovine serum albumin as the standard. The fluorescence excitation emission matrix (EEM) spectra of different EPS fractions were collected according to the procedure developed by Lu ¨ et al. (48). Particulates in samples VOL. 44, NO. 4, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1269

were removed using a 0.45 µm polytetrafluoroethylene (PTFE) membrane (Shanghai Mosu Scientific Equipment Co., China) prior to EEM tests.

Results Reactor Performance and Membrane Fouling. In the tests the AGMBR and MBR all had >95% removal rates of COD and total organic carbon (TOC) during the 210 d test. Figure 1 shows the typical time courses of TMP for reactors R2 and R4. The membrane module was chemically cleaned on day 67. The TMP of the first SBR cycle for R2 started at roughly 20 kPa and increased to about 31 kPa during a 10 min withdrawal period. During the second SBR period, the TMP increased from 22 to 42 kPa. The “upper” and “lower” TMP increased over time, reaching >80 kPa and roughly 60 kPa, respectively, on day 74, when the next chemical washing was employed. Correspondingly, the TMP for R4 was low (