Sensitive and Versatile Detection of the Fouling Process and Fouling

Feb 3, 2011 - This material is available free of charge via the Internet at .... NPs) are extensively used as an antibacterial additive in commercial ...
1 downloads 0 Views 1MB Size
Download PDF
ARTICLE pubs.acs.org/ac

Sensitive and Versatile Detection of the Fouling Process and Fouling Propensity of Proteins on Polyvinylidene Fluoride Membranes via Surface-Enhanced Raman Spectroscopy Li Cui,† Meng Yao,† Bin Ren,‡ and Kai-Song Zhang*,† †

Key Laboratory of Urban Environment and Health, Institute of Urban Environment, Chinese Academy of Sciences, Xiamen 361021, China ‡ State Key Laboratory of Physical Chemistry of Solid Surfaces and Department of Chemistry, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005, China

bS Supporting Information ABSTRACT: Membrane fouling is the major drawback of membrane-based technologies because it will lead to severe flux declines and the need to clean or replace the fouled membrane. A technique capable of early diagnosis, process monitoring, and evaluation of the role of different foulants playing in the fouling process is crucial for the fouling control. We develop surface-enhanced Raman spectroscopy (SERS) as a new and versatile tool to investigate the fouling process of protein on PVDF (polyvinylidene fluoride) membranes as well as the fouling propensity of three different proteins. We optimized the aggregation level and volume of SERS-active Ag sol and the spectra acquisition method combined with a statistical analysis method to ensure a high detection sensitivity, signal uniformity, and stability. We then used SERS for the early diagnosis of the fouling process and determining when the membrane pores would be blocked. The fouled area was visualized by a combination of the silver staining and Raman mapping. The fouling propensity of different proteins was studied by comparing the relative SERS band intensities of different proteins on a glass slide and after membrane filtration. Compared with fluorescence-based techniques, the narrow, well-resolved Raman band, especially the use of the same excitation line and laser power, endows SERS the ability to compare the fouling propensity in a very simple way.

W

ith the growing global water shortage, membrane-based technologies are gaining increasing popularity in wastewater treatment and reclamation, such as membrane bioreactor (MBR) and reverse osmosis, due to their notable advantages over the conventional biological wastewater treatment process such as the highly improved effluent quality, a smaller footprint, and less sludge production.1 However, a major drawback to the wide application of membrane technology is membrane fouling,2 which will induce a significant flux decline and a reduction in the productivity of clean water. The need to clean or replace the fouled membranes will greatly increase the operation cost and diminish its economic viability. In addition, fouling induced by microbial adhesion followed with the biofilm formation is irreversible and can cause permanent damage to membranes.3 Moreover, fouling caused by protein adsorption is a major problem in protein purification and artificial organs.4 For example, when blood flows through biomaterials, the initial adsorption of serum proteins often induce platelet adhesion, activation of coagulation pathways and then the risk of thrombus.5 A comprehensive and deep understanding of fouling mechanism will help us develop an effective strategy to prevent and control fouling. r 2011 American Chemical Society

To achieve this goal, numerous techniques have been utilized to characterize the membrane fouling process.6 One common way to study fouling is to record the permeate flux with time.7-10 Fouling mechanisms can be inferred by fitting the flux decline curve to the mathematical models presenting standard fouling, complete fouling, intermediate fouling, and cake filtration. The model manifesting the best fit is regarded as the dominant fouling mechanism.7-9 The fouling propensity of different species can also be induced by comparing the flux decline rate or resistance increasing rate.9-11 Ultrasonic time-domain reflectometry (UTDR) can monitor the amount of foulants during filtration based on the acoustic impedance difference at the bulk solution/membrane interface.12 Some electrical parameters were also used to trace the membrane fouling process. For example, the trans-membrane zeta potential measurement can reflect the presence of foulant into the membrane pores.13,14 The impedance spectroscopy technique can determine the membrane electrical resistance Received: November 4, 2010 Accepted: January 11, 2011 Published: February 03, 2011 1709

dx.doi.org/10.1021/ac102891g | Anal. Chem. 2011, 83, 1709–1716

Analytical Chemistry

ARTICLE

Scheme 1. Scheme of the Experimental Setup for Water, Protein, and Ag Sol Filtration

change caused by modification of membrane porosity/pore radius which is related to fouling.13 Although the above methods can give real-time information about the development of the fouling, they have some limitations. One is that they cannot distinguish different molecules and thus fail to identify the role of different species playing in the fouling process, such as the fouling propensity and reversible and irreversible fouling. The composition of foulants is very complicated. Extracellular polymer substances (EPS) secreted by microbes mainly composed of proteins, polysaccharides, and humic substances are generally considered to have significant contribution to the membrane fouling.15,16 In addition, various interactions between membrane and molecules, including the electrostatic and hydrophobic/hydrophilic interaction, also play an important role in the extent of membrane fouling.7,9 Due to this complexity, the ability of techniques in recognizing species and assessing their roles is crucial for the fouling control. The second limit is the low detection sensitivity. The above methods usually use feed solution with a high concentration and involve a long time monitoring. Important information like the fouling process and fouling propensity can only be inferred after recording the whole process. Not only is it time-consuming, but also the low sensitivity makes them unable to characterize the early stage fouling process, especially at a low concentration of feed solution. Some fluorescence-based techniques, like confocal laser scanning microscopy (CLSM) and multiphoton microscopy (MPM), can overcome the above limitations to some degree and have been utilized to study the fouling process.3,11 These techniques require that the foulants be labeled with fluorophores. The differentiation of fouling species is realized by an appropriate choice of fluorophores. However, due to the nature of a broad fluorescence spectrum, fluorescence is limited as a multiplex reporter due to spectral overlap. In addition, photo bleaching is also a problem which limits the long-term monitoring. Surface-enhanced Raman scattering (SERS) has great potentials as a highly sensitive, selective, and even quantitative tool for the analysis of biological or chemical molecules.17-21 With the aid of the extremely strong enhancement generated on the metallic nanoparticles, Raman signal of molecules can be enhanced by many orders of magnitude, which makes SERS an ultrasensitive detection tool, even down to the single molecule level.22 In addition, Raman bands are much narrower and can provide richer information than that of fluorescence. Spectral overlap can be well overcome, which makes SERS a powerful tool in multiplex analysis. Therefore, SERS is receiving more attention in various fields, particularly for biomedical and environmental testing, such

as DNA sequence,23,24 protein assay,25,26 immunoassay,27,28 bacterial or pathogen identification,20,29 biofilm,30 organic pollutants,31,32 etc. The high detection sensitivity of SERS will enable the early diagnosis of fouling and trace the whole fouling process. Moreover, the multiplex capability will help recognize different species and evaluate their roles in the fouling process. The above information will be greatly helpful for a deep understanding of the fouling mechanism and facilitate us to develop effective strategies to control fouling and decrease the operation cost of MBR. In this work, we concentrate on developing SERS methodology, including the optimization of aggregation state of Ag sol, spectral acquisition, and data analysis, for studying the fouling process and evaluating the fouling propensity of three different proteins on PVDF (polyvinylidene fluoride) membranes. Moreover, the ability of the early diagnosis of fouling, judgment of the pore blockage, and visualization of the fouled area by Raman mapping will be also presented.

’ EXPERIMENTAL SECTION Membrane and Protein. The PVDF membrane used in this study was purchased from Millipore (cat. no. GVWP 02500), and the pore size is 0.22 μm. The diameter of the membrane is 25 mm, and the effective filtration area is 4.1 cm2. Myoglobin (Mb) from equine skeletal muscle (Sigma-Aldrich, cat. no. M 0630), bovine serum albumin (BSA)-tetramethylrhodamine (Invitrogen, cat. no. A23016), and ovalbumin-Texas Red (Invitrogen, cat. no. O23021) were used as the model protein foulants. Protein suspensions were prepared in a 10 mM phosphate buffer at a pH of 7.3. The concentration of myoglobin used in the fouling process study was 2 mg/L. The total concentration of mixed protein solutions used in the fouling propensity study was 10 mg/L. In order to remove any protein aggregate that may have formed during storage, the protein solution was centrifuged twice at 5000 rpm; only the supernatants were used in the following filtration experiments. Dead-End Stirred Cell Filtration Module. All filtration experiments were conducted in a 25 mm diameter Millipore stirred cell with a volume of 10 mL (Model 8100, Amicon Corp). The filtration cell is connected with a reservoir and a nitrogen tank. A constant pressure of 0.05 MPa is allowed throughout the filtration (see Scheme 1), and the stirring speed is kept at 300 rpm for filtration. The permeate flux data were recorded automatically every second using an electronic balance connected to a personal computer via a Sartorius Software Wedge. Prior to protein filtration, 120 mL of ultrapure water was filtered through the fresh membrane. The flux of ultrapure water 1710

dx.doi.org/10.1021/ac102891g |Anal. Chem. 2011, 83, 1709–1716

Analytical Chemistry

ARTICLE

Figure 1. UV-vis spectra of Ag sols diluted three times (a), Ag aggregate after centrifugation (b), Ag aggregate after centrifugation and placement in air for 24 h (c), and after filtration through a 0.22 μm PVDF membrane (d).

was recorded and analyzed. Only those membranes with less than 5% pure water flux variation was selected for the further protein filtration. Preparation of Silver Colloid and Aggregated Silver Colloid. Colloidal silver was prepared using the Lee and Meisel’s method.33 Briefly, 72 mg of AgNO3 was dissolved in 400 mL of H2O and brought to boiling under vigorous stirring; a solution of 1% trisodium citrate (8 mL) was added. The solution was kept boiling for ca. 1 h. Aggregated Ag sol was obtained using a method of centrifugation and redispersion-in-water, which will be described in detail in the next section. A UV-vis spectrometer (SHIMADZU UV-2450) was used to characterize the aggregation of Ag sol. Assembly of Silver Nanoparticles (NPs) on the Fouled Membrane. Assembly was performed by filtration of 3 mL of the above aggregated Ag colloid onto the fouled membrane using the filtration cell under a pressure of 0.1 MPa. The stirring magneton was removed from the cell when performing Ag colloid filtration. SERS Measurement. SERS spectra were acquired from a Renishaw inVia confocal micro-Raman system equipped with a Leica microscope, charge coupled device detector, an 1800 L/mm grating, and a holographic notch filter. Excitation was provided by a He-Ne 632.8 nm laser with the power of 65 μW on the sample. A 50 objective (Leica) with a numerical aperture of 0.55 and a working distance of 8 mm was used to focus the laser beam and collect the Raman signal. The line focus mode was used in all Raman measurements. It not only minimizes the sample damage but also increases the uniformity of signals at different spots due to a larger sampling area (30 μm  1.5 μm) and a lower laser power density than the point focus mode.

’ RESULTS AND DISCUSSION Optimization of the Aggregation of Ag NPs for the SERS Measurement. The protocol of the use of SERS to monitor the

protein fouling process is simple, which is performed by filtration of Ag sol through the fouled membrane. However, the PVDF membrane is of porous structure with a pore size of 0.22 μm, which is larger than the diameter of Ag NPs (∼70 nm). The Ag NPs will pass through the PVDF membrane almost completely under pressure. The membrane with a higher fouling degree will block more Ag NPs on the membrane. The number of Ag NPs and the aggregation states of Ag NPs will significantly influence

Figure 2. SEM images of PVDF membranes before (A) and after (B) filtration of Ag sol.

the SERS enhancement. Therefore, it is important to ensure the same amount of Ag NPs on membranes with different fouling degrees. A proper control of the aggregation state of Ag sol will fulfill the above demand. Aggregation of Ag sol was obtained by centrifuging Ag sol three times at 6500 rpm for 6 min. In each cycle, the supernatant was removed and the same amount of ultrapure water was added. In this process, the capping reagent will be removed successively, which promotes the aggregation of Ag sol. The obtained Ag sol was finally dispersed in water by ultrasonication and placed in air for another 1-3 days before use. A UV-vis spectrum was used to illustrate this process (Figure 1). Compared with the Ag sol, a new broad band, appearing at around 800 nm after centrifugation, is characteristic of aggregated Ag sol.26 After being placed in air for 24 h, this broad band became even broader, which indicates that Ag aggregation increases with time. A placement time of 24 h is usually used; however, for the freshly synthesized Ag sol or Ag sol that has been synthesized for a long time, the placement time needs to be extended or shortened accordingly. The above aggregated Ag sol was then subjected to filtration through a 0.22 μm PVDF membrane. The permeate solution appears totally clear by the naked eyes, and the UV-vis spectra of the solution gives absorption values of less than 1% of the aggregated Ag sol, indicating that more than 99% of the aggregated Ag sol has been blocked by the membrane. If 1711

dx.doi.org/10.1021/ac102891g |Anal. Chem. 2011, 83, 1709–1716

Analytical Chemistry

ARTICLE

Figure 3. (A) From top to bottom: SERS spectrum of Mb on the PVDF membrane after filtrating 150 mL of 2 mg/L Mb solution and 3 mL of aggregated Ag sol, Raman spectrum of Mb in the solid state, Mb on PVDF membrane without Ag sol. (B) Histogram of the height of the peak at 752 cm-1 after baseline subtraction from 125 spots on the membrane.

the aggregated Ag sol can be blocked by a blank PVDF membrane, it will certainly be blocked by a fouled membrane. Thus, we can easily obtain fouled membranes with the same amount of Ag sol. Different from the use of ions like Cl- or SO42- to induce aggregate,26,34 the combination of centrifugation and redispersion-in-water cannot only clean the Ag sol but also give a proper control of aggregation. SEM images of a clean PVDF membrane and one after filtration of Ag sol are presented in Figure 2. The PVDF membrane shows a reticular structure. After filtration, some white dots appear in the SEM image. An EDX (energy dispersive X-ray) spectrum confirms that they are from silver (see Figure S1 in the Supporting Information). A closeup image (inset of Figure 2B) shows Ag NPs with spherical, elliptical, and rodlike shapes have obviously different contrast from PVDF. Ag NPs occupy most of the pores sites, and some were even trapped inside the pores. SERS Detection and the Uniformity of the SERS Signal. Figure 3A shows the spectra from myoglobin (Mb) in the solid state, membrane fouled by Mb with and without Ag sol assembled on it. Without Ag NPs, no signal from Mb was obtained. With Ag NPs, the SERS spectrum of Mb with a good signal-tonoise ratio was obtained and corresponded well with the Raman bands of solid Mb and published data,25,34 demonstrating the feasibility of SERS in the detection of protein foulant on the membrane. The uniformity of SERS intensity at different spots of the membrane is a prerequisite for the fouling process study. For this purpose, we performed Raman mapping (see Figure 3A) at five areas (80 μm  80 μm, 20 μm step) evenly distributed over the PVDF membrane and 125 spectra (1 s acquisition time) were acquired to obtain statistical data to allow a reliable comparison. Figure 3B presents the intensity of the peak at 752 cm-1 from these 125 spots. Statistical analysis using Origin software shows that the mean intensity is 4583 cps and the standard deviation is 434 cps, which is only about 10% of the mean value. Such deviation is sufficient for the following fouling process study. The SEM image (Figure 2B) shows that the Ag NPs do not evenly distribute on the membrane surface due to the inhomogeneous structure of PVDF membrane. However, the uniformity becomes better if observed at a larger scale (see Figure S2 in the Supporting Information, at the dimension of the spot size of the line focus laser), which allows us to obtain uniform SERS signals over the membrane. Raman mapping combined with the line focus was thus employed throughout the protein fouling process study. Optimization of the Volume of Ag Sol Used for SERS Study. The amount of Ag sol on the membrane was found to

Figure 4. Dependence of the SERS intensity of the peak at 752 cm-1 on the volume of aggregated Ag NPs. The corresponding photographs of PVDF membranes after Ag NP filtration are also shown below.

influence significantly the SERS intensity of the protein. In order to find the optimal volume of Ag sol, four pieces of PVDF membrane fouled with the same volume of Mb solution were prepared. Aggregated Ag sol with volume of 1, 2, 3, and 4 mL was then filtered through the four fouled membranes. The SERS intensity of the peak at 752 cm-1 was compared in Figure 4. As can be seen, 3 mL of Ag sol produces the strongest SERS intensity. Such a trend is easy to understand. With the increase of Ag sol volume, the color of the membrane became darker (see the picture in Figure 4). It indicates an increase of the coverage of Ag NPs on the membrane, resulting in a stronger SERS signal. However, a further increase of Ag NPs leads to a decrease of the SERS signal. As we know, the enhancement of nanoparticles is greatly distance dependent and shows an exponential decay with the distance,35 so the topper layers of Ag NPs will bring less or even no enhancement to the protein molecules underneath the thick Ag NP layer. Instead, due to a thicker Ag NP layer, more severely the excitation and Raman scattering light will be blocked. Therefore, the SERS signals decrease when a larger volume of Ag NPs is filtered. SERS Study of the Protein Fouling Process. The SERS study of the Mb fouling process is shown in Figure 5A. Each data point was obtained by statistically analyzing 125 spectra from 1712

dx.doi.org/10.1021/ac102891g |Anal. Chem. 2011, 83, 1709–1716

Analytical Chemistry

ARTICLE

Figure 5. (A) Plot of SERS intensity vs filtration time of 2 mg/L Mb solution. (B) SEM image of a fouled membrane with filtration time of 40 min. (C) Flux decline curve during the filtration of 2 mg/L Mb at a constant pressure of 0.05 MPa and stirring speed of 300 rpm. (D) Flux decline at the first 3 min extracted from (C).

Figure 6. (A) Optical image of a fouled membrane after Ag NP filtration. Dashed circle outlines the area of light color. (B) Raman map of the Mb peak at 752 cm-1 on the same area. Raman mapping setting: 80 μm  80 μm, 10 μm pet step, 1 s acquisition time. Excitation line: 633 nm.

different spots. With the increase of the filtration time, SERS intensity of Mb increases quickly at the first 5 min then decreases slowly until 75 min. The flux decline curve was extensively used to monitor the fouling process and predict the fouling mechanism by fitting the curve to some mathematic models.7-9 It was also monitored here during the Mb filtration (Figure 5C). By comparing the flux and SERS results, we can see that no flux decline can be observed at the first 3 min because there are too few Mb on the membrane to decrease the flux (Figure 5D). Therefore, the flux curve is not

able to monitor the launch of fouling. However, the rapid increase of SERS intensity at this early stage undoubtedly proves the initiation of the membrane fouling and the increasing amount of Mb on the membrane. This convincingly demonstrates that SERS can be effectively used for the early diagnosis of the fouling process. However, the SERS intensity starts to decrease after 10 min instead of keeping an increase with time. We acquired the SEM image (Figure 5B) of a membrane that has been filtered for 40 min to understand the phenomenon. Some pores of the membrane 1713

dx.doi.org/10.1021/ac102891g |Anal. Chem. 2011, 83, 1709–1716

Analytical Chemistry

ARTICLE

Figure 7. SERS intensity of Mb and BSA-TMR (A, B) or Mb and OT (C, D) with different ratios on a glass slide (A, C) and after filtration through a 0.22 μm PVDF membrane (B, D). The total concentration of each protein mixture solution is 10 mg/L, except that the concentration of pure BSA-TMR solution used in the membrane filtration is 50 mg/L. At each time, 10 mL of protein solution was filtered through the membrane.

shrink or are even fully blocked by Mb and very few or even no Ag NPs stay. During the filtration process, Ag NPs tend to flow to less fouled areas rather than the fully fouled areas under a pressure of 0.1 MPa. Accordingly, SERS intensity in seriously fouled areas will decrease due to the very few Ag NPs. The density of these areas will increase with the filtration time, resulting in a decrease of SERS intensity and a deterioration of the uniformity. By plotting the total SERS intensity against the filtration time, we can identify the time period that the pores of membrane start to get fully blocked via the turning point on the curve. This information is important because, once the pores are blocked, they are difficult to recover by physical cleaning which will significantly shorter the lifetime of membrane.15 In addition, the following fouling process can also be deduced from the intensitytime curve: protein absorptionf pore shrinkingf pore completely blockedf area fully fouled increasing. Moreover, the addition of Ag NPs onto the fouled membrane produces an Ag-staining effect as in biological analysis, which enables us to identify the fouled area and their distribution on the membrane simply by microscopy. Figure 6A shows the video image of the fouled membrane with Ag NPs on it. The area with light color in the video image (outlined by the dashed circle) is the severely fouled area and has less Ag NPs. This conclusion is supported by the Raman mapping shown in Figure 6B, where the SERS intensity in the corresponding area is lower than other areas. Therefore, the combination of SERS and Ag staining provides a facile way to visualize the fouled area. In summary, SERS is a powerful tool in the early diagnosis of the fouling process. It can easily identify when the pores are blocked and visualize the fouled area. We also compare SERS with fluorescence-based techniques like MPM.11 A direct comparison of the amount of protein on the

membrane is impossible for the two techniques. Indeed, we can compare the concentration of feed solution and the threshold time to observe the protein foulant. The values are 2 mg/L and 0.1 min in SERS and 50 mg/L and 0.25 min in MPM,11 respectively. Therefore, SERS is more sensitive than MPM. In addition, samples used in MPM have to be carefully handled and cannot be kept for a long time due to the fluorescence quenching. In contrary, samples used in SERS can be very “casual”. We did not find any obvious difference in the SERS intensity from the same sample newly prepared and stored in air at room temperature for 4 months. This feature not only simplifies the experimental handling but also makes SERS suitable for long-term monitoring or analysis of samples from different periods. SERS Study of the Fouling Propensity of Different Proteins. Recognition of the fouling propensity of proteins on membranes is very important, as it will not only deepen the understanding of the fouling mechanism and influencing factors but also benefit the adoption of suitable pretreatment before membrane filtration and design of proper membranes for selective filtration. Here, the fouling propensity of three proteins, namely, Mb, BSA-tetramethylrhodamine conjugate (BSA-TMR), and ovalbumine-Texas Red conjugate (OT), was studied by SERS. Due to the weak Raman signals of BSA and ovalbumine, they were labeled by Raman-active dye molecules, tetramethylrhodamine and Texas Red, respectively. The corresponding SERS features will be used to distinguish BSA and ovalbumine. A simple strategy was employed to judge the fouling propensity by comparing the relative SERS intensities of protein mixture on a glass slide and after the mixture solution was filtered through a PVDF membrane. The relative SERS signal of the protein causing severe fouling will be larger on the membrane than that on a glass slide, i.e., the propensity can be directly reflected by their 1714

dx.doi.org/10.1021/ac102891g |Anal. Chem. 2011, 83, 1709–1716

Analytical Chemistry relative intensities. To obtain the SERS signal of the protein mixture on the glass, the protein mixture was first dropped on a clean glass slide and allowed to dry. Then, concentrated Ag sol was dropped on top of the protein mixture and allowed to dry in air. Figure 7A shows the SERS results from the mixed Mb and BSA-TMR solution with different concentration ratios on a glass slide. As the SERS signal of BSA-TMR is much stronger than that of Mb, only at the Mb/BSA-TMR ratio of 9:1 can we observe the typical SERS bands of Mb, appearing at 752, 1126, and 1624 cm-1 together with the strong bands from BSA-TMR. At lower ratios, e.g., 7:3 and 5:5, only bands of BSA-TMR were observed. However, on the PVDF membrane, only SERS bands from Mb were observed (Figure 7B). No bands from BSATMR were detected. This proves that Mb fouls the PVDF membrane much more severely than BSA. Due to the much weaker fouling ability of BSA, its SERS signal on the membrane can only be detected after filtration of a high concentration of BSA-TMR solution (50 mg/L) without stirring. Fouling ability of Mb and ovalbumin was also studied. On the glass slide, the SERS signal of Mb starts to appear at a ratio of 7:3 (Mb/OT = 7 mg/L:3 mg/L) and increases at a ratio of 9:1 (Figure 7C). After filtration through the membrane, the relative intensity of Mb and OT does not show an observable difference from that on the glass slide (Figure 7D), which indicates that Mb and ovalbumin have similar fouling propensity on the PVDF membrane. In summary, the sequence of the fouling propensity of these three proteins is Mb ≈ ovalbumin > BSA. This result is in good agreement with the literature that Mb and ovalbumin fouled a membrane more severely than BSA using other techniques like MPM or the flux decline curve.9,11

’ CONCLUSIONS We have developed SERS as a new and versatile tool to study the fouling process of proteins on PVDF membranes as well as the fouling propensity of different proteins. Suitable aggregation of Ag sol and selection of proper volume of Ag sol to filter optimize the detection sensitivity of SERS. The utilization of line focus mode and statistical analysis of 125 SERS spectra, acquired by Raman mapping on different spots, ensure the relatively good uniformity of SERS signal on the membrane, which is very important for the following fouling process study. In the fouling process study, SERS demonstrates its great ability in the early diagnosis of fouling with its high detection sensitivity, identifying when membrane pores are blocked based on the turning point of the intensity-time curve, visualizing the fouled area by a combination of Raman mapping and Ag staining. SERS is also capable of recognizing the fouling propensity of different proteins by simply comparing the relative SERS intensities of different proteins on a glass slide and after membrane filtration. The fouling propensity of three proteins follows the order of Mb ≈ ovalbumin > BSA. Compared with fluorescencebased techniques, the narrow, well-resolved Raman band, especially the use of the same excitation line and laser power, endows SERS the ability to fulfill the fouling propensity comparison in an easy way. To our knowledge, this is the first time that SERS spectroscopy has been employed to study the membrane fouling. The results shown here indicate a promising perspective of SERS to study fouling in more complex systems with different feed solution and operation conditions, as well as the interactions between foulant species and membrane, which are of great significance for

ARTICLE

the deep understanding of the fouling mechanism and, more importantly, to provide clues for the fouling control and the development of antifouling membranes.

’ ASSOCIATED CONTENT

bS

Supporting Information. EDX spectrum and SEM image of PVDF membrane after Ag sol filtration in a large scale. This material is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Fax: þ86-592-6190534.

’ ACKNOWLEDGMENT We are grateful to Prof. Zhong-Qun Tian and Mr. Bi-Ju Liu in Xiamen University for providing the facility and assistance with the Raman spectrometer. This work was supported by the National Natural Science Foundation of China (Grant No. 20903076), Natural Science Foundation of Fujian Province (No. 2009J05032), and the Knowledge Innovation Project of Chinese Academy of Science (Project No. KZCX2-YW-452). ’ REFERENCES (1) Judd, S. The MBR Book: Principle and Application of Membrane Bioreactors in Water and Wastewater Treatment; Elsevier: Oxford, 2006. (2) Le-Clech, P.; Chen, V.; Fane, A. G. J. Membr. Sci. 2006, 284, 17–53. (3) Juang, Y. C.; Adav, S. S.; Lee, D. J.; Lai, J. Y. Environ. Sci. Technol. 2010, 44, 1267–1273. (4) Stamatialis, D. F.; Papenburg, B. J.; Girones, M.; Saiful, S.; Bettahalli, S. N. M.; Schmitmeier, S.; Wessling, M. J. Membr. Sci. 2008, 308, 1–34. (5) Lee, J. H.; Ju, Y. M.; Kim, D. M. Biomaterials 2000, 21, 683–691. (6) Chan, R.; Chen, V. J. Membr. Sci. 2004, 242, 169–188. (7) Shen, Y.; Zhao, W.; Xiao, K.; Huang, X. J. Membr. Sci. 2010, 346, 187–193. (8) Zheng, X.; Ernst, M.; Jekel, M. Water Res. 2009, 43, 238–244. (9) Loh, S. T.; Beuscher, U.; Poddar, T. K.; Porter, A. G.; Wingard, J. M.; Husson, S. M.; Wickramasinghe, S. R. J. Membr. Sci. 2009, 332, 93–103. (10) Maximous, N.; Nakhla, G.; Wan, W. J. Membr. Sci. 2009, 339, 93–99. (11) Hughes, D. J.; Cui, Z.; Field, R. W.; Tirlapur, U. K. Langmuir 2006, 22, 6266–6272. (12) Xu, X. C.; Li, J. X.; Xu, N. N.; Hou, Y. L.; Lin, J. B. J. Membr. Sci. 2009, 341, 195–202. (13) de Lara, R.; Benavente, J. Sep. Purif. Technol. 2009, 66, 517–524. (14) Tang, C. Y.; Kwon, Y.-N.; Leckie, J. O. Environ. Sci. Technol. 2007, 41, 942–949. (15) Meng, F.; Chae, S.-R.; Drews, A.; Kraume, M.; Shin, H.-S.; Yang, F. Water Res. 2009, 43, 1489–1512. (16) Metzger, U.; Le-Clech, P.; Stuetz, R. M.; Frimmel, F. H.; Chen, V. J. Membr. Sci. 2007, 301, 180–189. (17) Kneipp, K.; Moskovits, M.; Kneipp, H., Eds. Surface-enhanced Raman scattering-Physics and Applications; Springer: Heidelberg and Berlin, 2006. (18) Li, J. F.; Huang, Y. F.; Ding, Y.; Yang, Z. L.; Li, S. B.; Zhou, X. S.; Fan, F. R.; Zhang, W.; Zhou, Z. Y.; Wu, D. Y.; Ren, B.; Wang, Z. L.; Tian, Z. Q. Nature 2010, 464, 392–395. (19) Han, X. X.; Zhao, B.; Ozaki, Y. Anal. Bioanal. Chem. 2009, 394, 1719–1727. 1715

dx.doi.org/10.1021/ac102891g |Anal. Chem. 2011, 83, 1709–1716

Analytical Chemistry

ARTICLE

(20) Knauer, M.; Ivleva, N. P.; Liu, X.; Niessner, R.; Haisch, C. Anal. Chem. 2010, 82, 2766–2772. (21) Halvorson, R. A.; Vikesland, P. J. Environ. Sci. Technol. 2010, 44, 7749–7755. (22) Nie, S. M.; Emery, S. R. Science 1997, 275, 1102–1106. (23) MacAskill, A.; Crawford, D.; Graham, D.; Faulds, K. Anal. Chem. 2009, 81, 8134–8140. (24) Mahajan, S.; Richardson, J.; Brown, T.; Bartlett, P. N. J. Am. Chem. Soc. 2008, 130, 15589–15601. (25) Han, X. X.; Jia, H. Y.; Wang, Y. F.; Lu, Z. C.; Wang, C. X.; Xu, W. Q.; Zhao, B.; Ozaki, Y. Anal. Chem. 2008, 80, 2799–2804. (26) Han, X. X.; Huang, G. G.; Zhao, B.; Ozaki, Y. Anal. Chem. 2009, 81, 3329–3333. (27) Grubisha, D. S.; Lipert, R. J.; Park, H. Y.; Driskell, J.; Porter, M. D. Anal. Chem. 2003, 75, 5936–5943. (28) Qian, X.; Peng, X.-H.; Ansari, D. O.; Yin-Goen, Q.; Chen, G. Z.; Shin, D. M.; Yang, L.; Young, A. N.; Wang, M. D.; Nie, S. Nat. Biotechnol. 2008, 26, 83–90. (29) Rule, K. L.; Vikesland, P. J. Environ. Sci. Technol. 2009, 43, 1147–1152. (30) Ivleva, N. P.; Wagner, M.; Horn, H.; Niessner, R.; Haisch, C. Anal. Chem. 2008, 80, 8538–8544. (31) Jones, C. L.; Bantz, K. C.; Haynes, C. L. Anal. Bioanal. Chem. 2009, 394, 303–311. (32) Guerrini, L.; Garcia-Ramos, J. V.; Domingo, C.; SanchezCortes, S. Anal. Chem. 2009, 81, 953–960. (33) Lee, P. C.; Meisel, D. J. Phys. Chem. 1982, 86, 3391–3395. (34) Xu, H. X.; Bjerneld, E. J.; Kall, M.; Borjesson, L. Phys. Rev. Lett. 1999, 83, 4357–4360. (35) Tian, Z. Q.; Ren, B.; Li, J. F.; Yang, Z. L. Chem. Commun. 2007, 3514–3534.

1716

dx.doi.org/10.1021/ac102891g |Anal. Chem. 2011, 83, 1709–1716