e21224a
Colloidal-NOM Fouling of
Hydrophobic UF Membranes with Negative Surface Charge
Sangyoup Lee, Yongki Shim, In S. Kim, Jaeweon Cho*
Department of Environmental Science and Engineering, K-JIST
1 Oryong-dong, Buk-gu, Gwangju 500-712, Korea
*e-mail:
ABSTRACT
Recent studies have shown that natural organic matter (NOM) is a major membrane foulant during the ultrafiltration of surface water. However, no simple relationship has been established between the various structural fractions of NOM, including hydrophobic, transphilic, hydrophilic, and colloidal (with molecular weights ≥ 3500) constituents and their different fouling potentials with respect to the heterogeneous characteristics of NOM and the complexities of membrane fouling phenomena. Moreover, it is still controversial as to which fraction of NOM is primarily responsible for flux decline during ultrafiltration. In this study, colloidal and hydrophobic NOM constituents isolated from the same source water were ultrafiltered through a hydrophobic UF membrane (with a molecular weight cutoff of 10,000 mass units) with a negative surface charge, using a bench-scale cross-flow filtration unit. Zeta potential, contact angle, and ATR-FTIR spectra were investigated for clean and NOM-fouled membranes. It was shown that colloidal-NOM resulted in more significant fouling, during 24-hour cross-flow filtration tests, conducted under the same; solute concentration, pH, ionic strength, and hydrodynamic operating conditions, as used for hydrophobic NOM. On the other hand, zeta potential, and contact angle results, and ATR-FTIR spectra of clean and fouled membranes did not provide reasons for the more significant flux decline associated with colloidal NOM filtration than hydrophobic NOM filtration, which implies the involvement of complex interactions of heterogeneous NOM and the membrane surface.
KEY WORDS
Colloidal NOM, Hydrophobic NOM, Flux decline, Fouling, UF membrane
INTRODUCTION
‘Loose’ NF and ‘tight’ UF membranes are increasingly being used for drinking water treatment because they meet both the strict water quality regulations and the water production requirement. The successful application of these membranes is often limited by natural organic matter (NOM) fouling. While the mechanism of particle and colloid fouling of microfiltration (MF) and loose UF membranes have been rigorously investigated and are well understood, those of NOM fouling of ‘loose’ NF and ‘tight’ UF membranes remain still unclear. The objectives of this study were to determine the foulants causing flux declines among the various NOM constituents, such as, colloidal, hydrophobic, transphilic, and hydrophilic NOM fractions, and to compare the fouling characteristics of each NOM constituent regarding the tight hydrophobic UF membrane with a negative surface charge.
HYPOTHESES
The characteristics of colloidal NOM can be summarized as follows: 1) larger molecular size than other NOM fractions, and includes hydrophobic acids, hydrophilic acids, and hydrophilic neutrals/bases (Thurman and Malcolm, 1981); 2) colloidal NOM includes amino sugars and polysaccharides (Cho et al., 2000; Leenheer et al., 2000), which are relatively polar and hydrophilic constituents; 3) Colloidal NOM is neutrally or positively charged in natural waters (pH 6-8) due to its NH2/NH content, as primary and secondary aliphatic amide groups (Leenheer et al., 2000); and 4) the mass fraction of colloidal NOM to total NOM is higher in low-humic waters (specific UV absorbance (SUVA) < 3Lmg-1m-1) than in high-humic waters (SUVA > 3Lmg-1m-1) (Hwang et al., 2000). Based on these characteristics it is hypothesized that colloidal NOM is a major foulant of hydrophobic UF membranes with negative surface charge.
MATERIAL AND METHODS
NOM Isolation and Feed Waters
Various NOM fractions were isolated from Nakdong river surface water (NR-SW) using the method developed by Leenheer et al. (2000), with some modifications (marked in Figure 1). The modified protocol is described in Figure 1. After the first step (RO concentration) in Figure 1, the SUVA value increased from 1.79 Lm-1mg-1 (for raw NR-SW) to 2.70 Lm-1mg-1 (for RO concentrated NR-SW), which means that some portions of hydrophilic NOM were missed. This result is attributed to the following supposition; 1) some portions of hydrophilic NOM were not retained during the concentration process due to their relatively small molecular size; and 2) non-charged hydrophilic NOM was adsorbed onto the RO membrane during the concentration process due to its relatively high affinities to adsorb or to form a gel layer (Cho et al., 1998). To ensure flux-decline tests were performed at the same conditions, each feed solution was adjusted to ensure it contained almost the same values of pH, ionic strength, and DOC using 0.1N HCl (or NaOH), NaCl, and pure water, for further details of the feed water characteristics see Table 1.
Figure 1 Experimental protocol used for NOM isolation.
Table 1 Feed water characteristics
Feed water / pH / Conductivity,µS/cm / DOC,
mg/L / UVA254,
cm-1 / SUVA,
L mg-1m-1
Raw / 7.15 / 1517 / 2.29 / 0.0614 / 2.70
HP / 7.12 / 1504 / 2.27 / 0.0691 / 3.02
CD / 7.14 / 1509 / 2.25 / 0.0408 / 1.81
Flux-decline Tests
Experiments were conducted using a bench-scale cross-flow filtration unit fitted with a commercially available UF membrane. The filtration unit and the properties of the membrane used are illustrated and described in Figure 2 and Table 2, respectively. The zeta potential was represented as a function of pH (see Figure 3). In this study, the J0/k ratios (the ratio of the initial pure water permeate flux [J0] to the back diffusional mass transfer coefficient [k] (Cho et al., 2000)) were adjusted to the same value for each filtration test to provide the same hydrodynamic operating conditions. The adjusted hydrodynamic operating conditions used for tests are summarized in Table 3.
Figure 2 Schematics of the cross-flow filtration unit. Figure 3 Zeta potential of the PW membrane.
Table 2 Membrane properties
Membrane / Material / MWCO,Dalton / Contact angle,
° / Zeta potential,
mV @pH 7.0l / Pure water permeability,
L m-2 day-1 kPa-1
PW / Polysulfone / 10,000 / 73.4 (1.92)1 / -29.9 (2.56) / 37.5
1The numbers in parenthesis are standard deviations.
Table 3 Hydrodynamic operating conditions for membrane filtrations
Feed water / Feed flow rate,mL/min (@50 psi) / Cross-flow velocity, cm/sec / J0,
cm/sec / k,
cm/sec / J0/k ratio
Raw / 980 / 34.1 / 1.71ⅹ10-2 / 2.23ⅹ10-3 / 7.66
HP / 950 / 33.3 / 1.58ⅹ10-2 / 2.07ⅹ10-3 / 7.63
CD / 1,000 / 34.7 / 1.44ⅹ10-2 / 1.87ⅹ10-3 / 7.71
During each flux-decline test, NOM removal was determined at filtration times of 30min, 90min, 1200min, and 1260min using DOC (in the unit of mg/L) (SIEVERS® 820, Ionics, US), values of samples (both retentate and permeate) were recorded at each filtration time.
Analytical Methods (Clean versus NOM-Fouled Membranes)
The following analytical methods were used to determine changes in membrane surface properties resulted from NOM fouling. Flux-decline tests were conducted for 21 hours with each fractionated NOM. Two stage hydraulic washing was then performed, as follows, Stage 1: 2-hour hydraulic washing using pure water with the same cross-flow velocity as that used in the flux decline test, Stage 2: 1-hour hydraulic washing using the pure water with at twice the cross-flow velocity of Stage 1. The stage numbers (1 & 2) are marked on Figure 6 (a)). Both clean and NOM-fouled membranes were analyzed in terms of zeta potential measurements (ELS-8000, Photal, Japan), and ATR-FTIR (FT/IR-460 Plus, Jasco, Japan), and contact angle measurement (Half-Angle, Tantec™, US).
RESULTS AND DISCUSSION
Characteristics of Each NOM Constituent
Mass fractions, DOC versus specific UVA (UVA/DOC) values, and UV spectra of various NOM constituents are shown in Figures 4 and 5.
Figure 4 Mass fractions of various NOM constituents of RO concentrated NR-SW (CD: colloidal NOM, HP: hydrophobic NOM, TL: transphilic NOM, HL: hydrophilic NOM, Missing: non- recovered NOM).
(a) (b)
Figure 5 Feed water characteristics: (a) DOC vs. SUVA and (b) UV spectra of different NOM solutions of approximately the same concentration.
The mass fractions of colloidal and hydrophobic NOM accounted for approximate 20 and 40% of the total mass. The SUVA value was much lower for colloidal NOM than for hydrophobic NOM, as hypothesized. The fraction of NOM lost during the dialysis and the XAD-8/4 resins isolations accounted for around 10 and 5% of the total mass. It was noticeable that the recovery (95%) obtained using the XAD-8/4 resins was much better than those obtained not using the colloidal-NOM isolation procedure (70-85%) (Aiken et al., 1992).
Flux-decline Tests (Colloidal NOM versus Hydrophobic NOM)
Results of the 21-hour filtration tests using the different NOM constituents, followed by two steps of 3-hour hydraulic washing, exhibited somewhat different trends of flux decline and flux recovery, as presented in Figure 6 and Table 4.
(a) (b)
Figure 6 (a) Comparison of flux-decline during the filtration of each feed water and (b) the removal efficiencies of each NOM constituent versus filtration times.
Table 4 Comparisons of permeability reductions
Permeability reduction categories / Permeability reduced (PR), L m-2 day-1kPa-1Raw / HP / CD
Total / 27.18 / 23.18 / 31.81
1Equilibrium / 6.60 (24.3)5 / 6.60 (28.5) / 6.60 (20.8)
2CP layer / 1.88 (6.9) / 5.26 (22.7) / 0.67 (2.1))
3Gel layer / 0.80 (2.9) / 2.91 (12.6) / 0.07 (0.2)
4Adsorption / 17.90 (65.9) / 8.40 (36.2) / 24.47 (76.9)
1This was defined as a permeability reduction that occurred during filtration with ambient pure water, adjusted to provide the same pH and ionic strength values at the samples. 2This is reversible by first-step hydraulic washing. 3Also reversible by second-step hydraulic washing. 4Irreversible by hydraulic washing. 5The numbers in parenthesis are percentage ratios (%) of the respective permeability reductions with respect to the total reduction
These results imply that colloidal NOM has a higher fouling potential than hydrophobic NOM. They also show that irreversible (to hydraulic washing) fouling plays a more important role in flux decline than reversible fouling in the cases of colloidal NOM, as compared to the hydrophobic NOM. These observations are in accord with the hypotheses. Figure 6 (b) shows several important NOM removal behaviors; 1) during the initial rapid flux decline (within 2 hours), the NOM removal efficiencies increased in all cases, because accumulated NOM solutes on the membrane surface inhibited solute transport through the membrane pores, 2) the removal efficiency of colloidal NOM was somewhat higher than that of hydrophobic NOM, which is probably due to the larger molecular size of colloidal NOM than hydrophobic NOM (steric hindrance), and 3) colloidal NOM removal efficiency decreased as filtration time increased, while the opposite was true for hydrophobic NOM. The diffusion rate of colloidal NOM through the membrane pores increased with time because the accumulated and adsorbed colloidal NOM solutes on/near the membrane surface and in the pores increase the concentration gradient required for solute diffusion in the absence of charge repulsion such as that caused by hydrophobic NOM.
Clean and Each NOM-fouled Membranes
The results of zeta potential, contact angle, and ATR-FTIR spectra using clean and NOM-fouled membranes are illustrated in Figure 7.
Figure 7.Comparsion between clean and fouled membranes: (a) zeta potential, (b) contact angle, and (c) ATR-FTIR spectra
The effect of hydrophobic-NOM fouling on the surface charge of a membrane was more influential than that of colloidal NOM, which behaved in a manner contrary to expectation (it was expected that membrane fouled with colloidal NOM would exhibit decreased zeta potential of the membrane due to its neutral characteristic). Contact angle measurements showed the increased and decreased hydrophobicities of membranes fouled with hydrophobic NOM and colloidal NOM, respectively. From the ATR-FTIR analyses, we were unable to find any specific peaks (i.e., N-acetyl amino sugars: Leenheer et al., 2000) characterizing the colloidal NOM, however, the IR spectra of a membrane surface fouled with colloidal NOM, exhibited a significantly lower level of colloidal NOM than the other NOM constituents.
CONCLUSION
Based on flux-decline tests with several different NOM constituents, colloidal NOM was found to exhibit higher fouling potential than hydrophobic NOM, as hypothesized. In general, no simple relationship could be established between the different structural characteristics of the different NOM constituents and their fouling potentials in terms of the analytic observations made, which included zeta potential, contact angle, and ATR-FTIR measurements. However, it was found that source waters containing high levels of colloidal NOM have higher fouling potential, and this should be considered prior to the application of membrane filtration.
ACKNOWLEDGEMENT
This work was supported by the Korea Science and Engineering Foundation (KOSEF) through the Advanced Environmental Monitoring Research Center (ADEMRC) at K-JIST, and also supported by a grant (code 4-1-1) from Sustainable Water Resources Research Center of 21st Century Frontier Research Program.
REFERENCES
Aiken G.R., McKnight D.M., and Thurman E.M. (1992) Isolation of hydrophilic organic acids from water using nonionic macroporous resins, Org. Geochem,. 18, 567-573.
Cho J., Amy G.A., Pellegriono J., and Yoon Y. (1998) Characterization of clean and natural organic matter (NOM) fouled NF and UF membranes, and foulants characterization, Desalination, 118, 101-108.
Cho J., Amy G.A., and Pellegrino J. (2000) Membrane filtration of natural organic matter: comparison of flux decline, NOM rejection, and foulants during filtration with three UF membranes, Desalination, 127, 283-298.
Hwang C.J., Sclimenti M.J., and Krasner S.W. (2000) Disinfection by-product formation reactivities of natural organic matter fractions of a low-humic water, ACS Symposium Series 76, 173-187.
Leenheer J.A., Croue J-P., Benjamin M., Korshin G.V., Hwang C.J., Bruchet A. and Aiken G.R. (2000) Comprehensive isolation of natural organic matter from water for spectral characterizations and reactivity testing, ACS Symposium Series 76, 68-83.
Thurman E. and Malcolm R. (1981) Preparative isolation of aquatic humic substances, Environ. Sci. Technol. 15, 436-466
CONTACT DETAILS
Name: Jaeweon Cho, Organization: Kwangju Institute of Science and Technology
Tel: +82 (62) 970-2572, Fax: +82 (62) 970-2434, Email: