Minimization of Membrane Organic Fouling and Haloacetic Acids Formation by Controlling

Minimization of membrane organic fouling and haloacetic acids formation by controlling amino sugars and/or polysaccharide-like substances included in

colloidal NOM

Boksoon Kwon, Sangyoup Lee, Man Bock Gu, and Jaeweon Cho*

Department of Environmental Science and Engineering, Kwangju Institute of Science and Technology (K-JIST), Oryong-dong, Buk-gu, Gwangju 500-712, Korea, *corresponding author: Tel. 82-62-970-2443; Fax 82-62-970-2434; e-mail

ABSTRACT

The objective of this study was to evaluate the effect of colloidal NOM on the disinfection by-products (DBPs) formation potentials and membrane organic fouling. From various analyses of a NOM-fouled UF membrane surface, and large amount of colloidal NOM werefound in the analyzed foulants. From FTIR spectra and pyrolysis analyses, colloidal NOMswerefound to include amino sugars and polysaccharides,as indicated by N-acetyl groups in the FTIR spectra and from pyrolysis GC-Mass analyses. Colloidal NOMs have two problematic aspects for membrane applications,which may induce bio-fouling due to their low biostabilities, and relatively high DBPs reactivities. The amino sugars and polysaccharides can be utilized by heterotrophic bacteria, and are identified as colloidal NOM with high fractions of BDOC (approx. 39%),so, their removal is expected to involve some form of biological pre-treatment process. Colloidal NOM also exhibit relatively high formation potentials of DBPs, especially haloacetic acids (HAA).

KEYWORDS: Colloidal NOM,Foulant, BDOC, DBPs, membrane filtration

INTRODUCTION

Pilot tests, and actual membrane plants,utilizing source waters with natural organic matter (NOM) have been observed toexperience a significant flux decline, which is believed to be as a result of organic- and/or bio-fouling. Some studies reported that the hydrophobic NOM constituents are most likely responsible for this membrane fouling, while others suggested hydrophilic NOMs (including amino sugars and polysaccharides featuring N-acetyl groups from FTIR spectra and pyrolysis GC-Mass analyses) are the major foulants. Two pieces of evidence support the latter; firstly, most NOM-fouled membranes obtained from pilot- and full-scale filtration systems exhibited strong IR peaksfor N-acetyl groups (an evidence of polysaccharide-like or amino sugar substances), when compared to clean membranes, and secondly, both the membrane surface and NOM acid constituents (both hydrophobic and hydrophilic acids) are negatively charged, suggesting stronger electrostatic repulsion, resulting in less fouling or organic adsorption than with other NOM constituents (such as NOM neutrals and bases). NOM has been categorized into three different fractions, including; hydrophobic, transphilic and hydrophilic NOM constituents; in addition to colloidal NOM.

Colloidal NOM can be characterized in to three problematic areas: potential membrane foulants, bio-stability, and DBPs reactivity. From previous studies, it was found that colloidal NOM contain N-acetyl groups, typical of amino sugars and/or polysaccharides (Leenheer et al., 2000). The neutral properties of colloidal NOM can influence foulants during membrane filtration (Cho et al., 1998). Little research has been carried out on the characterization of colloidal NOM with respect to their bio-stabilities and DBPs reactivities; this information could provide a good insight to the optimization of water treatment processes, targeting minimization of biodegradable dissolved organic carbon (BDOC) levels and DBPs formation.

Biodegradable dissolved organic matter (BDOM) is relatively difficult to remove, probably due to its small molecular weight and neutral properties,to minimal BDOC levels in finished water (LeChevallier et al., 1987). BDOC levels (prior to distribution), and their treatability (prior to chlorination), are also important aspects in the bio-stability of distribution systems and DBP formation potentials (Mathieu et al., 1992; Van der Kooij et al., 1992; Servais et al., 1995).

The objectives of this study was to characterize colloidal NOM in terms of DBP formation potentials, membrane fouling and bio-stability to determine the optimum conditions for membrane filtration (including pre-treatment processes).

MATERIALS AND METHODS

Foulants sample and NOM preparations

Fouled UF membranes (MWCO of 2500) were obtained from a pilot-scale membrane filtration unit at in the Gwangju drinking water treatment plant,which had been operated for 6 months with Juam Lake source water. Foulants were scraped and collected from the fouled membrane surface, poured into a 5 L flask contaning pure water, stirred for at least 5 days, and then filtered through a 0.45μm micro-filter.Filtered NOM solution was used for all characterization analyses,which included; colloidal NOM fractionation, structure analysis (i.e., XAD-8/4 resin isolation), functionality measurements with FTIR, and membrane filtration tests.

Colloidal NOM was isolated using a regenerated cellulose dialysis membrane (Spectra/Por 3) that had a molecular weight cutoff (MWCO) of 3,500 Daltons. Prior to isolation, an appropriate length of membrane was cut to accommodatethe sample volume to be dialyzed, and the membrane washed by soaking in pure water overnight. Dialysis was conducted with other chemical processes, as suggested by Leenher et al. (2000). Some of the colloidal NOM was freeze-dried into powder,and was used for FTIR spectral analysis.

Membrane characterizations

Zeta potentials, of both the clean and organic fouled membranes, were measured by an electrophoretic method using an electrophoretic light scattering apparatus (ELS8000, Otzca, Japan). The contact angle (an index of hydrophobicity) was measured by a sessile drop method. An attenuated total refractive-Fourier transform infrared (ATR-FTIR) spectroscopic method was used to determine the functional groups of both the membrane surface and NOM solutes (either foulants or isolated powders) using a FTIR apparatus (Perkin–Elmer IR 2000 series with a 45 degree ZnSe crystal). Table 1 lists a summary of the membrane characterization.

Table 1. Membrane properties

Code / Material / MWCO
(mass unit) / Zeta potential
(mV) @ pH 7 / Manufacturer / Contact angle()
PW / Polyethersulfone / 10,000 / -29.10 / Desal. / 66.0
GH / Polyamide TFC / 2,500 / -30.50 / Desal. / 38.6

Measurements of disinfection by products (DBPs) formation potential

For the measurement of DBP formation potentials (i.e., haloacetic acids (HAAs) and trihalomethanes (THMs)), chlorine was added to sample solutions containing different NOM fractions. Chlorinated samples were stored in a 20oC incubator for three days prior to solvent extraction and DBPs measurements. Gas chromatography,employing electron capture detection (GC-ECD, 5890 Series Ⅱplus, Hewlett Packard) was used to measure DBP concentrations.

Biodegradable dissolved organic carbon (BDOC) measurement

The biodegradable organic matter (BOM) in the water samples was analyzed by the method suggested by Joret et al. (1989). Water samples (300mL) were filtered through a 0.45m filter and poured into a bottle,containing 100g of sand (diameter 0.5mm, uniformity coefficient 1.7), attached with heterotrophic bacteria, which were acclimatized with Nakdong river surface water (sand-water mixtures were incubated at 25oC until the DOC stabilized to a minimum value). The BDOC concentration was calculated from the difference between an initial DOC (average of DOC of sample and DOC of sample contacted with sand) and the minimum DOC obtained during the incubation period.

RESULTS AND DISCUSSION

Foulants obtained from the fouled GH membrane surface were evaluated by a flat-sheet membrane filtration test, and for colloidal NOM and BDOC fractions and NOM structure, as a comparison to NOM included in Juam Lake raw water. It can be hypothesized that foulant solutions may induce a greater flux decline than the raw Juam lake water. Tested UF membranes (GH) are negatively charged and exhibits low hydrophobicity, based on contact angles. From these membrane properties, it is also hypothesized that hydrophilic NOM constituents (especially hydrophobic NOM neutrals) could easily foul the membrane surface. Thus, the constituentsof fouled membranes are anticipated to contain a high fraction of hydrophilic NOM components. These hypotheses were investigated by foulant structural analyses, and zeta potential measurements, of both the clean and fouled membranes.These results are summarized in Figure 1 and Table 2. No significant differences in NOM fractions, between Juam raw water and foulants, were obtained, which was contrary to our hypothesis. However, the fouled GH membrane exhibited low negative zeta potentials compared to the clean GH membrane due to foulants (i.e., neutrals) coating/screening the membrane surface. From flux-decline tests of the PW membrane with both foulant solutions, and raw Juam river water (see Table 3);no significant difference in the flux decline, between the two filtration tests, was observed, as shown in Figure 2.

Figure 1. NOM structure of Juam lake and foulant: H-PHO; hydrophobic NOM, T-PHIL: transphilic NOM, H-PHIL: hydrophilic NOM

Table 2. Comparison of Zeta potential between clean membrane and fouled membrane

Clean GH / Fouled GH membrane not washing / Fouled GH membrane slightly physical washed with pure water
Zeta potential (mV) / -30.51 / -6.14 / -17.95

Table 3. Properties of samples

Sample / pH / Conductivity
(S/cm) / UV254
(cm-1) / DOC
(mg/L)
Juam lake / 7.23 / 83.4 / 0.0211 / 2.08
Foulants / 7.25 / 83.8 / 0.047 / 2.59

Figure 2. Comparison of Flux decline trends between Juam lake and foulant (J0/k=2.5)

FTIR spectra of the clean and fouled membranes and the foulants (as powder) are shown in Figure 3. The clean membrane exhibits IR peaks for aromatic double bonded carbons (around 1500 and/or 1600 cm-1), carboxylic groups (around 1250 cm-1), and a C-O bond of either ethers or carboxylic acids (1250-1050 cm-1) (Skoog and Leary, 1992; Bellamy, 1975). All the peaks of the clean membranes had reduced absorbance intensitiesfollowing organic fouling. The aromatic and carboxylic groups of the clean membrane surface almost disappear due to foulant coatings. The fouled membrane and foulants exhibit different IR spectra trends. N-acetyl peaks (at 1045 cm-1) was found in the IR spectra of the fouled membrane, which is evidence of fouling caused by neutral constituents of NOMs such as amino sugars and polysaccharides. Foulant powers, in comparison to the clean membrane, exhibited a strong peak near 1100 cm-1; indicative of alcohol groups in carbohydrates (probably from amino sugars and polysaccharides). Broad peaks near 1400 (hydrophobic neutral fractions) and 1600 cm-1 (carboxylate groups for the colloid) were also present with the fouled membrane.

Figure 3. FTIR spectra of clean membrane, fouled membrane, and foulants powder

For the rigorous characterization of membrane foulants, colloidal versus non-colloidal NOM foulants, were evaluated.Foulants were identified as containinga relatively high fraction of thecolloidal NOM (62%),compared to the non-colloidal NOM (38%), as shown in Figure 4. Colloidal NOM consists mostly of neutral hydrophilic NOM with a relatively high molecular weight (≥3500 daltons), causing organic-fouling of the membrane due to the neutral properties of the amino sugars and polysaccharides included in the colloidal NOM (Figure 5). Moreover, these amino sugars and polysaccharides can be utilized by heterotrophic bacteria, suggesting the potential for bio-fouling; the colloidal NOM consisted of a high fraction of BDOC (approx. 39%: measured by aeroboc BDOC tests with activity and inhibition controls). FTIR spectra showed an amide (1655 cm–1), a methyl (1382 cm–1) peaks, and a broad C-O (1045cm-1) peak for the colloid fractions,which are all indicative of N-acetyl amino sugars. From the flat-sheet membrane filtration tests with the UF membrane, the colloidal NOM exhibiteda significant flux decline due to organic fouling, as opposed to fouling by other NOM constituents (see Figure 6). Colloidal NOM had relatively high DBP formation potential reactivities in comparison to the other NOM fractions (see Figure 7).

Figure 4. Fraction of colloidal NOM and non-colloidal NOM in foulants from fouled membrane surface

Figure 5. FTIR spectra of colloidal NOM in foulants

Figure 6. Flux-decline trends of UF membrane with different NOM solutions

Figure 7. The reactivities of DBPs of various different NOM constituents

CONCLUSIONS

From the analyses performed; FTIR spectrum, BDOC measurements, membrane filtration tests, NOM and DBPs characterization, colloidal NOM exhibited many problematic inferences: high organic/bio-fouling potential, low bio-stability, and high DBPs reactivity, giving rise to some questions; how can the colloidal NOM be removed efficiently prior to membrane filtration, chlorination, and the distribution system? Theseanswers should be determined in conjunction with the neutral property, and the relatively high molecular weight and biodegradability of colloidal NOM.

ACKNOWLEDGEMENTS

This work was supported by the Korea Science and Engineering Foundation (KOSEF) through the Advanced Environmental Monitoring Research Center (ADEMRC) at Kwangju Institute of Science and Technology (K-JIST).

REFERENCES

Colleen E. Rostad, Jerry A. Leenheer, Brian Katz, Barbara S. Martin, and Ted I. Noyes. (2000) Characterization and disinfection by-product formation potential of natural organic matter in surface and ground waters from Northern Florida. American Chemical Society, 154-172.

Jaeweon Cho, Gary Amy, John Pellegrino, Yeomin Yoon (1998) Characterization of clean and natural organic matter (NOM) fouled NF and UF membranes, and foulants characterization. Desalination 118, 101-108.

Jerry A Leenheer, Jean-Philippe Croue, Mark Benjamin, Gregory V. Korshin, Cordelia J. Hwang, Auguste Bruchet, and George F. Aiken. (2000) Comprehensive isolation of natural organic matter from water for spectral characterizations and reactivity testing. American chemical Society, 68-83.

Sibille L. Mathieu, J.L. Paquin, D. Gatel, and J.C. Block. (1997) Microbial characteristics of a distribution system fed with nanofiltered drinking water. Wat. Res. 30, 2318-2326.

Sylvia E. Barrett, Stuart W. Kransner, and Gary L. Amy. (2000) Natural organic matter and disinfection by-products; characterization and control in drinking water-an overview. American Chemical Society, 2-14.

Presenter Information:

Name: Boksoon Kwon (Student)

Affiliation: Department of Environmental Science and Engineering, K-JIST

Phone: +82-62-970-2449

Fax: +82-62-970-2434

e-mail: