Environ. Sci. Technol. 2009 43, 8821–8826
Occurrence, Source, and Fate of Dissolved Organic Matter (DOM) in a Pilot-Scale Membrane Bioreactor
supernatant might originate from bound EPS, which can be rejected by membranes. The LC-OCD analysis, together with the results obtained from batch tests, suggested bound EPS might be the most important source of DOM in the sludge suspension.
F A N G A N G M E N G , * ,†,‡ A N J A D R E W S , § RENATA MEHREZ,| VERA IVERSEN,† MATHIAS ERNST,⊥ FENGLIN YANG,# MARTIN JEKEL,| AND MATTHIAS KRAUME† Chair of Chemical and Process Engineering, Technische Universita¨t Berlin, Str. des 17. Juni 135, MA 5-7, 10623 Berlin, Germany, School of Environmental Science and Engineering, Sun Yat-Sen University, Guangzhou 510275, PR China, Chair in Chemical Engineering, Engineering II, HTW Berlin, Wilhelminenhofstr. 75a, 12459 Berlin, Germany, Chair of Water Quality Control, Technische Universita¨t Berlin, Sekr. KF 4, Str. des 17. Juni 135, D-10623 Berlin, Germany, Centre for Water in Urban Areas, Technische Universita¨t Berlin, Sekr. KF 4, Str. des 17. Juni 135, D-10623 Berlin, Germany, and Key Laboratory of Industrial Ecology and Environmental Engineering, MOE, School of Environmental and Biological Science and Technology, Dalian University of Technology, Dalian 116024, PR China
Introduction
Received July 7, 2009. Revised manuscript received October 5, 2009. Accepted October 10, 2009.
In this study, the fate of carbohydrates, proteins, and humic substances in feedwater, sludge supernatant, and permeate of a pilot-scale membrane bioreactor (MBR) was investigated. Over 10 months, carbohydrates were observed to have a lower bioelimination degree (45%) and higher rejection degree (79%) than those of proteins (81% and 44%, respectively), which led to a high carbohydrate/protein ratio of dissolved organic matter (DOM) in sludge supernatant. The batch tests showed that DOM derived from feedwater and bound extracellular polymeric substances (EPS) was eliminated by activated sludge via biosorption and biodegradation. The proteins in bound EPS and feedwater were also found to have much higher biosorption potential (27% and 31%, respectively) than humic substances (11% and 17%, respectively) and carbohydrates (16% and 14%, respectively), indicating that proteins had a high affinity with sludge flocs. The results also showed that carbohydrates and humic substances in bound EPS were more difficult to be eliminated by activated sludge. In addition, the batch tests confirmed that feedwater was mainly composed of readily biodegradable matter, and bound EPS was mainly composed of slowly biodegradable matter. Size exclusion chromatography with continuous organic carbon and UV254 detection (LC-OCD) showed that large-size substances (i.e., carbohydrates and macromolecular proteins) in sludge * Corresponding author e-mail:
[email protected]; phone: +8615924076990. † Chair of Chemical and Process Engineering, Technische Universita¨t Berlin. ‡ Sun Yat-Sen University. § Chair in Chemical Engineering, Engineering II, HTW Berlin. | Chair of Water Quality Control, Technische Universita¨t Berlin. ⊥ Centre for Water in Urban Areas, Technische Universita¨t Berlin. # Dalian University of Technology. 10.1021/es9019996 CCC: $40.75
Published on Web 11/05/2009
2009 American Chemical Society
Dissolved organic matter (DOM) is a group of potential fouling-causing substances in membrane bioreactors (MBRs) used for wastewater treatment (1-4), and it also plays a great role in conventional membrane processes used for drinking water treatment or wastewater reuse (5, 6). Because of the role of membrane rejection, some DOM with large size can be retained, leading to an accumulation of DOM in the bioreactor (7-9), which can consequently result in severe membrane fouling (4, 10). DOM substances may exert negative impacts on biomass activity (7). For example, the accumulation of DOM inhibited the activities of the Nitrosomonas and the Nitrobacter group (11). In addition, the discharge of DOM-rich water brings additional problems to the local environment (e.g., occurrence of dissolved organic nitrogen). Now, combined membrane bioreactor and reverse osmosis (MBR+RO) technology is being actively used for wastewater reuse, in which MBR is used as pretreatment for RO (12). But, the remaining DOM in MBR effluent can foul the subsequent RO membranes (12). Therefore, a comprehensive investigation on the fate of DOM is helpful for the control of MBR/RO fouling, improvement of MBR performance, and implementation of posttreatment for water recycling. To date, most related investigations were focused on the characteristics and fouling potential of DOM (4, 13-16), bioelimination and membrane rejection of DOM (17, 18), and influence of operating conditions on DOM fate (8, 19). For example, Liang et al. (4) reported that hydrophobic aquatic humic substances exhibited higher fouling potential than hydrophilic fractions in DOM. In addition, the shear stress induced by aeration can lead to the release of bound extracellular polymeric substances (EPS) and generate more DOM with biological nature (i.e., large molecular weight) (20). For a MBR process, feedwater and bound EPS are the main source of DOM in sludge supernatant. So far, however, there is no investigation to explain which source is more important to DOM formation in sludge suspension. It is also of high interest to know how DOM in sludge suspension is produced during the complex biological processes. The purpose of this study, therefore, is to understand the fate of DOM based on long-term observation of a pilot-scale MBR and batch tests. In this study, proteins, carbohydrates, and humic substances (characterized by UV254) of feedwater, sludge supernatant, and membrane permeate in the MBR plant were regularly monitored. Batch tests were carried out to elucidate the elimination mechanism of DOM substances in activated sludge. The size distribution of DOM was analyzed by size exclusion chromatography with continuous organic carbon and UV254 detection (LC-OCD).
Materials and Methods Pilot-Scale MBR. The MBR consisted of one anoxic tank and one aerobic tank (working volume of 0.8 m3 each, Figure S1 of the Supporting Information). Combined municipal wastewater, which is a mixture of domestic wastewater, industrial wastewater, and rainwater, was used as feedwater. In addition to this use of a typical real wastewater, the fact that investigations were performed on different days, i.e., with VOL. 43, NO. 23, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8821
FIGURE 1. Statistical graphs showing protein and carbohydrate concentrations of feedwater, sludge supernatant, and membrane permeate (n ) 31, meaning of the statistical box, Figure S4 of the Supporting Information). differing compositions, means that results can be considered representative of similar applications. A 22 m2 membrane module (PVDF, 0.2 µm, A3 Water Solutions GmbH, Germany) was submerged in the aerobic tank; the membrane flux was set at around 16 L/(m2h). The hydraulic retention time (HRT), sludge retention time (SRT), and aeration rate were 7-8 h, 13 days, and 17 m3/h, respectively (21). Total solids (TS) concentration ranged from 5 to 8 g/L. The performance of the MBR in COD and nutrient elimination can be found in Table S1 of the Supporting Information (21). Batch Test. The bioelimination mechanism of DOM derived from feedwater and bound EPS was studied using batch tests (Figure S2 of the Supporting Information). A total of 300 mL of activated sludge was taken from the pilot-scale MBR and washed 3 times with 1000 mL of tap water to exclude the original DOM in the sludge. Then, the sludge was transferred to the batch test vessel, and feedwater or bound EPS, which are extracted by using cation exchange resin, was dosed. The initial organic loading rate was controlled at: carbohydrate concentration ) 10-18 mg/L, protein concentration ) 80-110 mg/L, and UV254 ) 90-110 1/m. The sludge concentration was controlled at around 1.5 g/L via dilution with tap water or feedwater. The batch was equipped with a pitched blade stirrer (200 rpm) and was constantly kept at 20 °C using a water bath. Fine bubble aeration was employed through a porous rubber tube, and the dissolved oxygen (DO) concentration was automatically controlled at 4-6 mg/L by a control valve. The proteins, carbohydrates, and humic substances in soluble form (DOM) and bound form (bound EPS) were analyzed at 0 min (before starting the batch test), 5 min, 20 min, 40 min, 1.5 h, 3 h, 7 h, and 24 h. Analytical Methods. The extraction of bound EPS from sludge flocs was performed with cation exchange resin (Naform, Dowex, U.S.A.) according to Frølund et al. (22). The sample of feedwater and sludge supernatant was prepared by filtering the feedwater and activated sludge with filter paper (black ribbon, Whatman GmbH, Germany). Carbohydrate concentrations of feedwater, sludge supernatant, and membrane permeate were analyzed according to Dubois et al. (23). The influence of nitrite and nitrate on carbohydrate measurement was corrected according to Drews et al. (18). Protein concentrations were determined according to Lowry et al. (24). All samples were analyzed in duplicate, and the results were given as an average value. The COD values of feedwater, sludge supernatant, and membrane permeate were measured using the vial test LCK 314/514 by Hach Lange, Germany. UV254 of the samples was evaluated according to standard methods (25), and SUV254 absorbance was calculated by dividing the UV254 absorbance by the COD concentration of the sample. Size Exclusion Chromatography. To characterize the composition of DOM, the LC-OCD system (DOC-Labor Dr. 8822
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 23, 2009
Huber, Karlsruhe, Germany; SEC column, Toyopearl HW55S by Tosoh Bioscience, Tokyo, Japan) was used according to Haberkamp et al. (26). A typical LC-OCD chromatogram of DOM in activated sludge supernatant is represented in Figure S3 of the Supporting Information. Prior to the measurement with the LC-OCD system, the samples were filtered through a 0.45 µm cellulose nitrate filter (Sartorius, Go¨ttingen, Germany), and the samples were diluted to satisfy the measurement range of 1-5 mg TOC/L.
Results and Discussion Fate of DOM during the MBR Process. To understand the fate of DOM substances during the MBR process, protein and carbohydrate concentrations of feedwater, sludge supernatant, and membrane permeate were regularly monitored over a period of 10 months. Figure 1 shows the protein and carbohydrate concentrations of feedwater, sludge supernatant, and membrane permeate. It can be seen that the sludge supernatant contained much lower protein and carbohydrate concentrations than feedwater, indicating that most DOM derived from feedwater could be removed by biological processes and membrane discharge. In fact, the origins of DOM in the sludge suspension are very complex. On the one hand, feedwater usually contains a great deal of DOM (e.g., proteins, carbohydrates, humic substances, and organic acids), most of which can be eliminated during contacting with biomass (17). The remaining compounds (or nonbiodegradable matter) will contribute to the formation of DOM in sludge suspension. On the other hand, during the biological process, a part of the bound EPS can be hydrolyzed or excreted to produce soluble microbial products (SMP). Strictly speaking, DOM in sludge supernatant is composed of SMP and nonbiodegradable matter derived from feedwater. In this study, therefore, the net bioelimination was used to characterize the degradation of DOM during the biological process, which was obtained by comparing DOM in feedwater and sludge supernatant. The bioelimination degree (BED) and membrane rejection degree (MRD) of DOM were calculated by using the following equations BED )
DOMfeedwater - DOMsludge DOMfeedwater
(1)
MRD )
DOMsludge - DOMpermeate DOMsludge
(2)
Where, BED is the bioelimination degree (%); DOMfeedwater is the DOM concentration in feedwater (mg/L); DOMpermeate is the DOM concentration in membrane permeate (mg/L); DOMsludge is the DOM concentration in sludge supernatant (mg/L); and MRD is the membrane rejection degree (%).
FIGURE 2. Statistical graphs of UV254 and SUV254 of feedwater, sludge supernatant, and membrane permeate (n ) 31, meaning of the statistical box, Figure S4 of the Supporting Information). It was found that on average 81% of proteins and 45% of carbohydrates in feedwater could be eliminated. As mentioned above, however, when DOM derived from feedwater is being eliminated by activated sludge, a part of bound EPS on sludge flocs is also being hydrolyzed simultaneously; namely, DOM elimination can always be accompanied by DOM generation. Therefore, the actual elimination degree of DOM derived from feedwater should be higher than that obtained here. Anyway, we can conclude that carbohydrates should have either low elimination degree by biodegradation and membrane discharge or high generation degree by EPS release. This will be explained in detail by using batch tests and LC-OCD analysis in the following sections. In addition, rejection of DOM compounds by the membrane was 44% for proteins and 79% for carbohydrates on average, suggesting that proteins have a smaller size and can thus be discharged into the effluent more easily. These results are in agreement with Drews et al. (18), who reported that the rejection of DOM components by a PAN membrane (37 nm) ranged from 20% to 70% for proteins and from 75% to 100% for carbohydrates. A previous study using a membrane of a larger pore size (0.4 µm) reported that the rejection of carbohydrates and proteins was only 28-52% and 21-32%, respectively, depending on the imposed SRT (27). While rejection varies with pore size, protein rejection clearly is always lower than that of carbohydrates. Typically, because of the higher affinity of DOM compounds to the membranes, the transport velocity of DOM through the membranes can be lower than that of a water molecule (28). Finally, the retarded transport of DOM across the membranes would lead to its deposition or accumulation in the membrane pores or fouling layer. The carbohydrate/protein ratio of DOM in feedwater, sludge supernatant, and membrane permeate was 0.18, 0.50 and 0.11, respectively, on average. The high carbohydrate/ protein ratio of sludge supernatant further confirms that carbohydrate had a lower bioelimination degree or a higher membrane rejection degree. The carbohydrate/protein ratio was presumably determined by the following (1): proteins and carbohydrates have different biodegradability (2), proteins and carbohydrates have different affinity with sludge flocs (3), the release of bound EPS can change the carbohydrate/protein ratio, and (4) the membrane rejection of proteins and carbohydrates differs significantly. In this study, the carbohydrate/protein ratio of sludge supernatant was found to have no correlation with that of feedwater, indicating that the formation of DOM in sludge suspension is not solely determined by feedwater or not all of the DOM fractions are influenced by feedwater. In order to elucidate the abovementioned hypothesis, LC-OCD analysis and batch tests were performed in this study (see Bioelimination Mechanism of DOM-Source and LC-OCD Analysis). Here, UV254 was used to indicate the presence of humic substances (27). The bioelimination degree of humic sub-
FIGURE 3. Illustration showing the definition of biodegradation and biosorption in the batch tests, taking proteins in feedwater as an example. stances was about 70% (Figure 2a). In addition, on average 30% of humic substances in the sludge supernatant were rejected by membrane separation, implying that the humic substances could pass the membrane readily because of the small molecular weight. In fact, the colloids or proteins in the samples can disturb the UV254 measurement because of their UV254 absorption or light scattering. However, the fouling layer formed on the membranes can capture or retain humic substances when they pass through membranes. Ng et al. (27) showed that the membrane rejection degree of UV254 was observed to be about 11.5% regardless of the change of SRT. This study confirms that humic substances have a smaller size than carbohydrates and proteins, and their rejection degree might be unaffected by operating conditions. From Figure 2b, it can be noticed that the SUV254 is given in the order of feedwater