Water Research 39 (2005) 3849-3862 coagulation of natural organic matter with a ferric salt: An electron energy loss spectroscopy investigation. Experiments were first conducted with a reconstituted water containing either synthetic or natural extracts of humic substances. At acid pH, Fe / C remains close to 3 over the whole range of iron concentrations investigated, while a much lower atomic ratio is expected from the value of optimal coagulant dosage.
Water Research 39 (2005) 3849-3862 coagulation of natural organic matter with a ferric salt: An electron energy loss spectroscopy investigation. Experiments were first conducted with a reconstituted water containing either synthetic or natural extracts of humic substances. At acid pH, Fe / C remains close to 3 over the whole range of iron concentrations investigated, while a much lower atomic ratio is expected from the value of optimal coagulant dosage.
Water Research 39 (2005) 3849-3862 coagulation of natural organic matter with a ferric salt: An electron energy loss spectroscopy investigation. Experiments were first conducted with a reconstituted water containing either synthetic or natural extracts of humic substances. At acid pH, Fe / C remains close to 3 over the whole range of iron concentrations investigated, while a much lower atomic ratio is expected from the value of optimal coagulant dosage.
Coagulation of humic substances and dissolved organic matter
with a ferric salt: An electron energy loss spectroscopy investigation A.-V. Jung a, , V. Chanudet a , J. Ghanbaja b , B.S. Lartiges a , J.-L. Bersillon a a Laboratoire Environnement et Mineralurgie (LEM), UMR 7569 CNRS-INPL-ENSG, 15 avenue du Charmois, BP 40, 54501 Vanduvre, France b Service commnun danalyses (MET), Universite Nancy I, BP 239, 54 506 Vanduvre, France Received 5 February 2005; received in revised form 7 July 2005; accepted 11 July 2005 Available online 19 August 2005 Abstract Transmission electron microscopy (TEM) coupled with electron energy loss spectroscopy (EELS) and energy dispersive X-ray spectroscopy (EDXS) was used to investigate the coagulation of natural organic matter with a ferric salt. Jar-test experiments were rst conducted with a reconstituted water containing either synthetic or natural extracts of humic substances, and then with a raw water from Moselle River (France). The characterization of the freeze-dried coagulated sediment by EELS in the 250450 eV range, showed that Fe-coagulant species predominantly associate with the carboxylic groups of organic matter, and that this interaction is accompanied by a release of previously complexed calcium ions. The variation of Fe/C elemental ratio with iron concentration provides insightful information into the coagulation mechanism of humic substances. At acid pH, Fe/C remains close to 3 over the whole range of iron concentrations investigated, while a much lower atomic ratio is expected from the value of optimal coagulant dosage. This suggests that a charge neutralization/complexation mechanism is responsible for the removal of humic colloids, the aggregates being formed with both iron-coagulated and proton-neutralized organic compounds. At pH 8, the decrease in Fe/C around optimal coagulant concentration is interpreted as a bridging of stretched humic macromolecules by Fe- hydrolyzed species. Aggregation would then result from a competition between reconformation of humic chains around coagulant species and collision of destabilized humic material. EELS also enabled a ngerpriting of natural organic substances contained in the iron-coagulated surface water, N/C elemental analyses revealing that humic colloids are removed prior to proteinic compounds. r 2005 Elsevier Ltd. All rights reserved. Keywords: Coagulation; Humic substances; Natural organic matter; Electron energy loss spectroscopy (EELS) 1. Introduction Dissolved organic matter (DOM) represents the heterogeneous mixture of organic colloids (excluding living organisms) and truly dissolved organic molecules present in aquatic systems. It comprises biochemically dened compounds such as proteins and carbohydrates, ARTICLE IN PRESS www.elsevier.com/locate/watres 0043-1354/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2005.07.008
Corresponding author. Private domicile, 84 chemin du
Klettenberg, 68350 Brunstatt, Brunstatt, France. Tel.: +0033 6 32464105; fax: +0033 3 89647787. E-mail address: av1jung@hotmail.com (A.-V. Jung). operationally dened substances such as fulvic and humic acids, and any supramolecular assembly of those molecules and macromolecules (Leenheer and Croue , 2003). DOM has largely been studied in recent years, as it may play a major role in the mobility and bio- availability of anthropic contaminants in the environ- ment (Thurman and Malcolm, 1981; Tipping, 1993). In drinking water treatment, the main motivation for studying DOM stems from its role as precursor of carcinogenic chlorination by-products during the disin- fection stage (Oliver and Visser, 1980). However, most problems encountered in the treatment process can be related to the presence of DOM, including reduced adsorption capacity in activated carbon beds, taste and odour of the nished water, and biological regrowth in the distribution networks (Owen et al., 1995). The removal of DOM from water supplies is carried out in the rst place by coagulation-occulation with a hydrolyzable metal salt such as aluminum or iron (Dempsey et al., 1984; Edwards and Amirtharajah, 1985; Gregor et al., 1997). When added to water, aluminum and iron coagulants yield hydrolysis products that destabilize the organic material present in the raw water. The destabilization mechanisms most often quoted in the literature include charge neutralization, enmeshment, and adsorption (Dempsey et al., 1984). Charge neutralization preferentially applies at acidic pH, while sweep-occulation and adsorption are com- monly referred to in pH-coagulant concentration domains where abundant precipitation of metal hydro- xide is expected (Edwards and Amirtharajah, 1985). However, DOM-specic characteristics, such as its molecular weight distribution, carboxylic acidity, and humic substance content, have been shown to signi- cantly inuence the efciency of DOM removal (Collins et al., 1985; Bose and Reckhow, 1998; Matilainen et al., 2002). Recent extended X-ray absorption ne spectro- scopy (EXAFS) at the Fe K-edge experiments also questioned the coagulation mechanisms usually invoked in the literature, as poorly polymerized Fe species were evidenced in a wide range of operational conditions (Vilge -Ritter et al., 1999a, b; El Samrani et al., 2004). As an optimized DOM coagulation is a key to better drinking water treatment, the interaction behavior between organic compounds and coagulant species should be better understood. The approach taken in this work was to investigate, within ocs, both the nature of DOM and the speciation of an iron-based coagulant by using electron energy loss spectroscopy (EELS) combined with energy dispersive X-ray spectro- scopy (EDXS). EELS provides a quantitative analysis at the nanometer scale of light elements (carbon, nitrogen, oxygen) through the K-shell excitation, and of other important elements in water treatment such as iron, calcium, phosphorus, and sulfur, through the L-shell excitation. It also enables the identication of organic chemical bonds within polymeric systems (Varlot et al., 1998), and can then be used to ngerprint the various organic macromolecules contained in coagulated organ- ic matter. In environmental science, EELS was rst used by biochemists to describe in situ natural materials degradation with C/N ratio measurements (Villemin et al., 1995; Watteau et al., 1996). It has also been used to investigate the association of organic matter with clay particles (Furukawa, 2000). In this paper, we present the results obtained by C- EELS. Fe-EELS results will be reported in a companion paper (Jung et al., 2005a). To avoid problems associated with changes in the quality of the river water, experi- ments were rst carried out with a reconstituted river water containing either humic-like synthetic compounds or natural aquagenic humic extracts. In this way, the inuence of pH on humics removal could be investi- gated. The method was then assessed on raw Moselle river water. 2. Experimental section 2.1. Origin of natural organic matter (NOM) and extraction of humic substances Three sources of organic matter were selected for this study. A synthetic humic-like substance (SHS) was prepared by polymerization of catechol with glycine (Andreux et al., 1980; Jung et al., 2005b). Catechol (6.6 g) was dissolved in 500 mL of a solution of sodium phosphate buffer NaH 2 PO 4 /Na 2 HPO 4 at pH 8 to obtain a 0.03 M catechol solution. This was then mixed with an equal volume of buffered glycine 0.03 M solution. The resulting mixture was then allowed to react in a 2 L ask in the dark at 25 1C and under a pure oxygen constant pressure (0.9 bar) for 5 days. After dialysis, (Spectra/ Por s 7 membrane- 8 kDa molecular weight cut-off Roth-Sochiel Lauterbourg, France), the synthetic mate- rial in a sodium form was freeze-dried and stored at 4 1C in the dark. Extensive characterization of SHS and relevance to natural humic substances have been discussed in a previous work (Jung et al., 2005b). Elemental analysis gave a carbon to nitrogen ratio (C/N) of 18.3 (%C w/w 40.7). Potentiometric titration showed three acidities at pK A 4.6 (2.3 meq/g), 5.1 (0.8 meq/g) and 7.2 (0.6 meq/g). A natural aquagenic humic substance, hereafter referred to as NHS, was extracted from the Moselle river suspended matter. The sampling site was located at Richardme nil (France) in the vicinity of the pumping facility supplying Grand Nancy Urban Community drinking water treatment plant in surface water. Approximately, 12 m 3 of river water were pumped (1 m 3 /h) into a ow-through centrifuge operating at 17,000 rpm. The humic substances contained in the ARTICLE IN PRESS A.-V. Jung et al. / Water Research 39 (2005) 38493862 3850 suspended material were then extracted following the procedure of Thurman and Malcolm (1981), adapted to yield the natural extract in a sodium form. NHS was also puried by dialysis (Spectra/Por s 7 membrane 8 kDa molecular weight cut-off) and freeze-dried. Total exchangeable protons and C/N ratios determined for NHS were 3.0 meq/g and 9.2 (%C w/w 44.2), respectively. Grab samples of the Moselle river water (9.3 mg/L total organic carbon content) were also taken at the sampling site, thus providing a raw water with unfractionated NOM. 2.2. Coagulation procedure Synthetic waters were prepared by dissolving 30 mg of humic material (either SHS or NHS) in 1 L of Milli-Q s ultra-pure water. 336 mg of NaHCO 3 (4 10 3 mol/L) (Sigma-Aldrichanalytical grade) was also added to the suspension to provide a carbonate alkalinity similar to that of natural river waters. Prior to coagulant injection, the pH of synthetic waters was adjusted to pH 6 or 8 with dropwise addition of 0.1 M HCl. The surface water was used as collected (natural pH 7.4). Jar-tests were performed in 1 L bafed reactors of known power dissipation characteristics (Lartiges et al., 1997). Suspen- sion was stirred at 100 rpm which corresponds to a mean velocity gradient of 135 s 1 . The coagulant, a 0.1 M ferric nitrate solution prepared from crystallized Fe(NO 3 ) 3 , 9 H 2 O (95% pureSigma-Aldrich), was injected under agitation using a 5000 mL micro-pipette (Eppendorf, Hamburg, Germany), and mixing was continued for 20 min. The suspension pH was not re- adjusted after coagulant addition. After 24 h of quies- cent settling in graduated Imhoff cones, the residual turbidity of the supernatant was measured on a HACH Ratio XR turbidimeter. The settled volume was recorded, and the sediment was then collected and freeze-dried (48 hEL105 Sentry) to perform TEM observations, EDXS and EELS analysis. 2.3. Transmission electron microscopy (TEM) Electron microscopy observations were performed with a CM20 Philips transmission electronic microscope (TEM) using a 200 kV accelerating voltage. The electron microscope is coupled with an EDAX EDXS, and with a GATAN 666 EELS. The freeze-dried sediment was re- suspended in ethanol (99.5% spectroscopy grade) under ultrasonication, and a drop of suspension was evapo- rated on a carbon network-like holey support lm placed on a 200 mesh copper grid (Euromedex Mundolsheim, France). EDX spectra were recorded with a 10 nm spot diameter and a 30 s counting time. EELS spectra were collected in the 250450 eV region with the same spot size (analyzed area of about 80 nm 2 ). The spectrometer entrance aperture, the half-angle of collection and the dispersion were 2 mm, 10 mrad, and 0.2 eV/channel, respectively. The energy resolution, determined at the full-width at half-maximum of the zero-loss peak is 1.4 eV. The time acquisition was 2.5 s for the low energy region and 8 s for the C K-edge region. The relative thickness (ratio of sample thickness to the free electronic inelastic diffusion pathway), was kept about 0.570.2 for all measurements. The electron dose, calculated from the Egerton relationship (Egerton, 1996), was about 101.9 C/m 2 which is less than the critical value-inducing beam damage within organic polymers (Varlot et al., 1998). For each sample, 15 EEL spectra were recorded. All data processing was carried out with GATAN EL/P 2.1 software. The spectra were rst corrected for dark current, gain variations, and double excitation by applying a Fourier-ratio deconvolution with weak loss energy spectra. Background tting was performed with a power law in order to extract the core-excitation peaks for each element. The amounts of C, Ca, and N were then obtained by integrating the surface area of a 50 eV energy window after having dened the lower-limit in energy for each element, that is 283.8, 346.4 and 401.6 eV for carbon (C K-), calcium (Ca L 2,3 -) and nitrogen (N K-) edges, respectively. Integration results are given by the software in atomic concentration/nm 2 . Hence, Ca/C and N/C atomic ratios were directly calculated from EELS measurements, whereas Na/C and Fe/C ratios were inferred from the amounts in Na, Ca, and Fe, provided by EDXS microanalyses. When Ca was barely detectable on EEL spectra, which is the case for most coagulation experiments carried out at pH 6, Fe/C and Na/C ratios obtained from EDXS spectra were corrected by a proportional constant determined from EELS and EDXS results at pH 8. 3. Results 3.1. Jar-test results Fig. 1 presents the results obtained from jar-tests carried out at pH 6 and 8 with SHS and NHS. All turbidity and sediment volume curves display a classical behavior: the residual turbidity steadily increases at low coagulant concentration to reach a maximum, and then steeply decreases upon further addition of ferric coagulant. The point where the extrapolated steep portion of the turbidity curve intersects the x-axis is selected as the optimum coagulant concentration (OCC), and is indicated with an arrow on the graphs. At higher iron concentrations, the residual turbidity remains low and then strongly increases as the restabilization of the suspension is induced. The inverse pattern is observed for sediment volume: at low iron ARTICLE IN PRESS A.-V. Jung et al. / Water Research 39 (2005) 38493862 3851 concentration, no settleable aggregate is observed up to the maximum in supernatant turbidity. The sediment volume then rapidly builds up just before the OCC, and it levels off with further addition in coagulant. At pH 6, the sediment volume gradually diminishes as the humic suspension is restabilized. OCC values, expressed in molar concentrations, and pHs at OCC are given in Table 1. In all cases, the pH at OCC falls within the 4.56 pH range for best removal of organic substances (Dempsey et al., 1984). Surprisingly, OCCs obtained for NHS coagulated under acid and alkaline conditions are equivalent. As fresh stock solutions of iron coagulant were prepared for each experiment, this result could be due to the proteina- ceaous nature of NHS (Jung et al., 2005b). 3.2. Morphology of coagulated humic substances Typical micrographs of coagulated humic substances are shown in Fig. 2. The sample preparation used in this study precludes artifact free observations. Indeed, freeze-drying is known to drastically modify the structure of aggregated colloids (Hung et al., 1996), whereas redispersion in ethanol may induce secondary ARTICLE IN PRESS Iron concentration (M) 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 0 2 4 6 8 10 12 0 10 20 30 40 50 60 0 2 10 -4 4 10 -4 6 10 -4 8 10 -4 10 -3 SHS pH 8 SHS pH 6 NHS pH 8 NHS pH 6 (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) (l) (m) ( n) (o) (p) (q) Iron concentration (M) 0 2 10 -4 4 10 -4 6 10 -4 8 10 -4 Iron concentration (M) 0 2 10 -4 4 10 -4 6 10 -4 8 10 -4 10 -3 Iron concentration (M) 0 4 10 -4 8 10 -4 1.2 10 -3 Fig. 1. Evolution of residual turbidity (K) and sediment volume () with coagulant concentration for NHS and SHS at pHs 6 and 8. Arrows indicate optimal coagulant concentrations. The letters refer to freeze-dried sediment examined by TEM. A.-V. Jung et al. / Water Research 39 (2005) 38493862 3852 agglomeration effects upon solvent removal (Stevenson and Schnitzer, 1982). Actually, techniques such as atomic force microscopy in aqueous solution (Maurice and NamjesnikDejanovic, 1999; Plaschke et al., 1999) or TEM observation of ultracentrifuged and resin-coated colloids (Wilkinson et al., 1999), seem more appropriate to provide a description of hydrated humic material. Nevertheless, as the prepara- tion conditions remained identical for all samples, the morphology of coagulated humic substances could be compared. Coagulated SHS appears as compact aggregates of overlapping spheroids about 30 nm in size, regardless of pH and iron concentration (Figs. 2a and b). The appearance of NHS aggregated at pH 8 is similar with smaller spheroids (1015 nm) (Fig. 2c). In contrast, electron micrographs of NHS coagulated at pH 6 reveal a network of lament or sheet-like structures at the micron scale (Fig. 2d). No signicant change of the latter morphology could be detected with coagulant concentration. Despite the structural changes certainly induced by the preparatory technique, our TEM observations evidence individual spheroids which aver- age dimension is consistent with the range of sizes previously reported in the literature for humic colloids (Wilkinson et al., 1999; Plaschke et al., 1999). 3.3. EELS spectra of coagulated SHS and NHS To the authors knowledge, EELS investigations of NOM remain relatively scarce in the literature (Watteau et al., 1996; Furukawa, 2000). The spectra of well- characterized compounds such as Bovine serum albumin and Dextran (polysaccharideM W 100; 000) were then recorded as standards, and the identication of EELS spectral features was conducted in agreement with recent near-edge X-ray absorption spectroscopy (NEX- AFS) studies (Myneni, 2002 and references herein). EELS spectra were calibrated in energy using the 346.4 eV Ca L 2,3 -edge as an internal reference. This yielded a 401.6 eV for the N K-edge energy which is consistent with the attribution usually proposed for nitrogen in NOM (Watteau et al., 1996). When calcium was not detected in our samples, the nitrogen K-edge was used instead. An example of energy loss spectrum recorded in the 250450 eV region (SHS coagulated at pH 8 with [Fe] 7 10 4 mol/L), is illustrated in Fig. 3. Three main features can be observed at the carbon edge: the well-dened shoulder at 285 eV can be assigned to the 1s-p* transition of aromatic carbons (CQC); the small peak around 288 eV is attributed to the 1s-p* transition of the CQO bond in carboxylic groups, while the broad peak centered at 300 eV is attributed to overlapping s* transitions for CQC, CC, CH, CQO, and CO bonds (Myneni, 2002). At higher energy, the two peaks observed at 346 and 358 eV correspond to the scattered electrons of the L23 shell of calcium (Watteau et al., 1996). The presence of calcium corresponds to a contamination brought about by the addition of NaHCO 3 ([Ca 2+ ]2.7 10 4 mol/L). It is likely associated with the functional groups of SHS as calcium has been shown to compete with trace elements for the binding sites of humic substances (e.g., Tipping, 1993; van den Hoop et al., 1995). Finally, in accordance with the elemental composition of SHS, a small amount of nitrogen can be detected at the N K-edge around 401 eV. Increasing the coagulant concentration at pH 8 signicantly modies the electron loss spectrum of coagulated SHS (Fig. 4a). At the carbon edge, the sharp peak at 288 eV assigned to p* (CQO) drastically decreases, while the large peak at 300 eV broadens further. In addition, the peak at the calcium edge rapidly drops. A recent NEXAFS study of pedogenic humic acid also showed a similar decrease of the p* resonance of the carboxyl group double bond upon aggregation with Fe 3+ at pH 2 (Thieme et al., 2002). Such observations can be readily explained by an exchange of Fe-coagulant species with previously bonded Ca 2+ ions onto carboxylic groups of SHS. The same interac- tion mechanism has also been inferred from infrared spectroscopy results in the case of adsorption/desorption of NOM on iron oxide particles (Gu et al., 1994), coagulation of sewage with ferric chloride (El Samrani et al., 2004), and more generally, such a reaction scheme is consistent with the strong inuence of carboxylic group ARTICLE IN PRESS Table 1 Optimum coagulant (iron) concentrations (OCC) determined by jar-test procedures OCC pH 8.0 pH 6.0 pH 7.4 SHS NHS SHS NHS NOM Molar concentration (mol/L) 8.3 10 4 6.4 10 4 2.2 10 4 8.1 10 5 1.7 10 4 Reduced concentration (mg Fe/mg C org ) 3.83 2.69 1.01 0.34 1.02 pH at the respective OCC 6.5 6.3 5.1 5.2 5.9 A.-V. Jung et al. / Water Research 39 (2005) 38493862 3853 content on coagulant demand (Lefebvre and Legube, 1990; Randtke, 1993; Collins et al., 1986; Specht et al., 2000). The inset of Fig. 4a presents the evolution of EELS spectra of freeze-dried humic material remaining in the supernatant after settling. For each iron concentration, the spectrum closely resembles that of a sediment aggregated with a lower coagulant dosage, and can be characterized by a higher calcium content and by a slightly sharper peak at 288 eV (p*CQO). This suggests ARTICLE IN PRESS Fig. 2. Electron micrographs of freeze-dried coagulated sediment: (a) SHS pH 8[Fe] 8.0 10 4 mol/L, (b) SHS pH 6 [Fe] 2.0 10 4 mol/L, (c) NHS pH 8[Fe] 7.0 10 4 mol/L, (d) NHS pH 6[Fe] 8.0 10 4 mol/L. A.-V. Jung et al. / Water Research 39 (2005) 38493862 3854 that the unsettled humic material contains a lesser amount of hydrolyzed Fe species. The EELS spectra of NHS coagulated at pH 8 as a function of coagulant concentration exhibit similar features (Fig. 4c): however, the peak characterizing the 1s-p* transition of the CQO bond in carboxylic groups is much less resolved at low iron concentration, and it is gradually replaced by a weak shoulder at higher coagulant dosage. Similarly, the two peaks at the Ca L 2,3 edge, clearly dened at low coagulant concentration, rapidly weaken with an increase in iron concentration. The EELS spectra of SHS and NHS coagulated at pH 6 as a function of iron concentration are shown in Figs. 4b and d. In that case, the relative intensities of the various contributions do not vary noticeably with coagulant concentration. At the carbon edge, the 1s-p* transition of aromatic carbons (CQC) at 285 eV remains as a distinct shoulder, whereas the contribution at 288 eV (p* CQO) is now present as a weakly resolved shoulder. The large dominant feature corre- sponding to the s* transitions for CQC, CC, CH, CQO, and CO bonds has shifted around 297 eV. Finally, a large contribution in the 325 eV region can be evidenced, but could not be assigned. However, the freeze-dried humic material from the supernatant yield EELS spectra similar to those obtained at pH 8 (inset of Fig. 4b). In particular, the p* resonance of the CQO double bond appears as a peak at low iron concentra- tion, and this peak decreases at higher coagulant dosage. Likewise, the Ca edge which was virtually absent in the coagulated sediment, can be clearly observed except at the highest iron concentration. Such observations suggest that the residual turbidity is formed from a less coagulated humic material also at pH 6. 3.4. Elemental analysis of coagulated SHS and NHS Fig. 5 shows the evolution of N/C, Ca/C, Na/C, and Fe/C atomic ratios of coagulated SHS and NHS with iron concentration at pHs 6 and 8. N/C remains constant as a function of coagulant concentration and pH, thus indicating that both the extraction procedure and the synthesis yielded a homogeneous humic material for coagulation experiments. N/C atomic ratios calcu- lated from EELS spectra are equal to 0.0770.02 and 0.1770.02 for SHS and NHS, respectively. Such values are consistent with previous N/C elemental ratios 0.054 and 0.167obtained from chemical analysis (Jung et al., 2005b). Similar trends are also observed for the evolution of Ca/C and Na/C. At pH 8, Ca/C markedly decreases just before OCC to reach an almost constant elemental ratio at higher coagulant concentration, ARTICLE IN PRESS N o r m a l i z e d
i n t e n s i t y
( A . U . ) 270 320 370 420 Energy Loss (eV) 285 eV 1s C=C 288 eV 1s C=C 296 - 297 eV 1s C-N, C=O, C-O C=C, C-C, C-H C-K D e r i v a t i v e
o f
n o r m a l i z e d
i n t e n s i t i e s
( A . U . ) 293-294 299-300 286 285 288 297 275 310 345 380 415 C EELS binding energies Energy Loss (eV) Ca-L 2,3 N-K Standard 1 Polysaccharide Standard 2 Bovine serum albumin 346.4 SHS 401.6 Ca L 2,3 - N-K Fig. 3. Typical EELS spectrum (SHS pH 8[Fe] 8.0 10 4 mol/L). The inset shows the double derivative for (i) Bovine serum albumin, (ii) Polysaccharide, (iii) SHS at pH 8 and at [Fe] 8.0 10 4 mol/L. A.-V. Jung et al. / Water Research 39 (2005) 38493862 3855 whereas Na/C remains close to 0.1 for both SHS and NHS. At pH 6, Ca is not detected within the coagulated sediment, whereas Na/C continuously decreases with iron concentration in accordance with an exchange of sodium ion for hydrolyzed Fe species onto anionic groups of humic substances. The lower Na/C ratios obtained at pH 8 are also consistent with an increased deprotonation of SHS and NHS functional groups, even though Na/C values may be modied during freeze-drying. The evolution of Fe/C atomic ratios as a function of coagulant concentration follows two different patterns according to pH. At pH 8, for both SHS and NHS, Fe/C is about 3 before OCC, it surprisingly drops to a value close to 1 at OCC (0.8 (c) and 1.25 (m)), and then re-increases to an intermediate value at higher coagulant dosage. At pH 6, Fe/C slightly increases with iron concentration from a value of 1 at underdosage for NHS, to reach a Fe/C ratio of about 3 at OCC for both humic compounds, and ARTICLE IN PRESS Fig. 4. EEL spectra of freeze-dried coagulated humic substances as a function of iron concentration: (a) SHS pH 8, (b) SHS pH 6, (c) NHS pH 8, (d) NHS pH 6. A.-V. Jung et al. / Water Research 39 (2005) 38493862 3856 values greater than 4 when the suspension is restabilized. Interestingly, recent Fe K-edge EXAFS measure- ments carried out on Seine river NOM coagulated with iron chloride (Vilge Ritter et al., 1999a, b), revealed that (i) iron hydrolysis is hindered to at most the trimer stage, (ii) each iron atom is surrounded at optimum dosage by three C and 1 C at pH 5.5 and pH 7.5, respectively. Such results would yield Fe/C elemental ratios in the 13 range which is consistent with our data. Assuming a complete removal of humic material by coagulation, the determination of elemental ratios from chemical analysis predicts that, at pH 8, Fe/C should be equal to 0.8 and 0.66 for SHS and NHS, respectively, which is in relatively good agreement with the values obtained from EELS and EDX spectra. In contrast, at pH 6, the same calculations yield Fe/C atomic ratios of 0.22 and 0.7 which is much less than the measured values. Such a discrepancy may be explained by the aggregation behavior of humic substances with pH. The hydrolysis of iron coagulant implies a decrease in pH; hence, at OCC, the nal pH of the suspension initially coagulated at pH 6, was about 4.5. At the latter pH, a small degree of aggregation of humic substances has been clearly demonstrated from the measurement of their diffusion coefcients by uorescence correlation spectroscopy (Lead et al., 2000). In situ AFM observa- tions of adsorbed humic colloids as a function of pH (Plaschke et al., 1999; Balnois et al., 1999), also indicate an increased surface aggregation with a decrease in pH. The iron-coagulated sediment should then include some loosely attached humic colloids that may become rapidly ARTICLE IN PRESS Fig. 5. Evolution of elemental ratios as a function of coagulant concentration: N/C (), Fe/C * 1/10 (&), Na/C * 1/10 for NHS pH 8 (K) and Ca/C (J). A.-V. Jung et al. / Water Research 39 (2005) 38493862 3857 volatilized under the electron beam; the microanalysis being performed mainly with the iron coagulated humic substances. The almost constant Fe/C atomic ratio obtained at pH 6 is then in accordance with previous studies suggesting a complexation mechanism between oligomeric coagulant species and functional groups of fulvic and humic colloids around pH 5 (van Breemen et al., 1979; Jeckel, 1986; Lefebvre and Legube, 1990). The effect of pH on Fe/C elemental ratio can be explained by a dynamic reorganization of humic macromolecules upon interaction with Fe coagulant species. At pH 8, the behavior of deprotonated humic material is generally compared with that of a exible anionic polymer in a stretched conguration (Gosh and Schnitzer, 1980; Vermeer et al., 1998). In that case, at low coagulant concentration, the partial neutralization of humic macromocules by Fe-hydrolyzed species can induce a reconformation of polymer chains, thus leading to a compact structure and values of Fe/C close to 3. Recent uorescence spectroscopy using pyrene as a polarity-sensitive uoroprobe also suggests such a rearrangement of humic macromolecules upon associa- tion with multivalent cations (Engebretson and von Wandruszka, 1994; Kazpard et al., 2005). At higher coagulant dosage, the amount of hydrolyzed Fe species increases relative to the number of humic colloids, and aggregation between stretched polymer chains may proceed before folding and reconformation of humic material take place. In that case, the coagulant species may form bridges between stretched humic chains. In accordance with such an aggregation mechanism, the increase in sediment volume observed with pH above OCC, is proportional to C/Fe ratios for both SHS and NHS. Therefore, the increase in sludge volume at pH 8 is here attributed to a change in shape factor of humic colloids, and not to the formation of abundant iron hydroxide precipitate as predicted by a sweep-occula- tion mechanism (Gregory and Dupont, 2001). 3.5. Coagulation of Moselle river raw water Experiments with Moselle river water were run in duplicate, similar trends being observed for the two grab samples. Four kinds of NOM compounds were identi- ed from TEM examination (Fig. 6): humic-like colloids (Fig. 6a (A)), by far the most frequent compounds in the freeze-dried coagulated sediment, were about 20 nm in diameter with a N/C ratio in the 0.020.05 range; proteinic substances (Fig. 6a (B)), appeared as large grayish areas 300 nm in size with N/C values greater than 0.1; polysaccharidic material was characterized by micron size spheroids with neither nitrogen nor iron detected, a CEEL spectrum equivalent to that of a sugar backbone, and it was easily degraded under the electron beam (Fig. 6b); small organic colloids intimately associated with smectite clay material and with a sharp 299 eV peak, were also found (Fig. 6c). Such organic matter is reminiscent of that observed by Furukawa in Jourdan river sediments (Furukawa, 2000). It should be noted that, for NOM, the interpretation of EELS spectra at the carbon edge is complicated by the presence of L23 peaks at 297 and 300 eV of potassium (Henke et al., 1993). Fig. 7 shows the jar-test results obtained with Moselle river water coagulated with iron at natural pH (7.4). The supernatant turbidity continuously increases with iron concentration up to [Fe]1.2 10 4 mol/L, then steeply decreases at OCC ([Fe] 1.7 10 4 mol/L), slightly re- increases above OCC, and then decreases again at higher coagulant dosage. The evolution with coagulant con- centration of C-EELS spectra corresponding to humic- like colloids is illustrated in Fig. 8a. This kind of material shows a similar reactivity to that of SHS and NHS at pH 8, with a strong decrease of the p*CQO peak at 288 eV with iron concentration. Fig. 8b presents the variation of N/C, Ca/C, and Fe/C elemental ratios of coagulated NOM with iron concentration. Fe/C in- creases continuously with coagulant dosage to reach a value close to 1 above OCC, which is consistent with the results obtained with SHS and NHS. Likewise, Ca/C elemental ratio decreases in accordance with a substitu- tion of Ca for Fe-hydrolyzed species onto functional groups of organic matter. N/C values reveal that NOM removal occurs in two main stages: up to OCC, the N/C atomic ratios are less than 0.03, thus suggesting a preferential coagulation of humic substances; above OCC, N/C increases signicantly to reach values in the 0.1 range which indicates the presence of proteinic matter in the coagulated sediment. Such results are consistent with previous pyrolysis-gas chromatography- mass spectrometry (Py-GC-MS) experiments showing that polyhydroxyaromatics are preferentially removed by ferric chloride (Vilge Ritter et al., 1999a, b; Jung, 2004). More generally, the hydrophobic fraction of NOM is known to be more easily destabilized than hydrophilic compounds (e.g., proteinic matter) (Croue et al., 1993; Bose and Reckhow, 1998). In particular, the slight increase of residual turbidity observed above OCC could correspond to the restabilization of some humic colloids (Fig. 7). 4. Conclusion This study shows that TEM observations combined with EDXS and EELS analyses, represent a powerful tool to investigate the coagulation of NOM. The nanoscale resolution of EELS and EDXS is well- adapted to the size range of organic colloids encoun- tered in surface waters. In particular, a ngerprinting of coagulated NOM can be achieved, even though further ARTICLE IN PRESS A.-V. Jung et al. / Water Research 39 (2005) 38493862 3858 ARTICLE IN PRESS 250 300 350 400 450 Energy Loss (eV) 250 300 350 400 450 Energy Loss (eV) 285 B 300 346.4 401.6 B : protein-like substances B A 0.5 m TEM photo 401.6 285 A 289 300 C.K EELS A : humio-like substances A 0 1 2 6 7 8 9 10 0 : C K a 1' : Fe L a 0.7 2.1 3.5 4.9 6.3 Energy (keV) EDXS Elmt At% 1 2 6 7 8 9 10 11 O Na P S Cl K Ca Fe 55.6 3.5 1.0 1.1 2.5 2.6 2.1 31.6 1 2 6 7 8 9 10 11 O Na P S Cl K Ca Fe 81.6 3.4 1.1 1.1 2.6 2.5 6.1 1.6 Elmt At% B 0 1' 2 6 7 8 9 10 0 : C K a 1' : Fe L a 11 1 0.7 2.1 3.5 4.9 6.3 250 300 350 400 450 285 300 C-K EELS TEM photo 0.5 m EDXS 0 1 Elmt (K) At% 0 1 C 0 55 45 (indicative) 0.7 2.1 3.5 4.9 6.3 C-K EELS 298 299 285 346.4 300 350 400 450 0.2 m TEM photo EDXS 1 0 1' 2 3 4 5 67 8 9 10 11 0.7 2.1 3.5 4.9 6.3 0 : C K a 1' : O L a 1 2 6 7 8 9 10 11 O Na P 3 Mg 4 Al 5 Si S Cl K Ca Fe 58.9 3.4 3.4 6.6 9.5 0.5 0.5 2.8 1.8 2.2 10.5 Elmt At% Energy (keV) Energy (keV) Energy Loss (eV) Energy Loss (eV) Energy (keV) 250 (a) (b) (c) Fig. 6. Typical NOM compounds identied by combining TEM observation with EDXS and EELS spectra: (a) Humic-like substances (A) and protein-like material (B), (b) Polysaccharide-like substances, (c) Organo-mineral species. A.-V. Jung et al. / Water Research 39 (2005) 38493862 3859 work is still required to improve the distinction of natural organic substances. EELS and EDXS examination of coagulated sediment also provides a useful insight into the destabilization mechanism of humic substances and NOM: Fe-hydro- lyzed species associate with carboxylic groups of humic colloids at both pH 6 and 8, thus inducing a release of previously complexed calcium ions. An overall charge neutralization/complexation mechanism seems to be responsible for the removal of humic substances in our experimental conditions as no Fe-hydroxyde precipitate could be detected in the coagulated sediment. The association of non-coagulated humic colloids within ocs may also occur at pH 6. On the other hand, at alkaline and neutral pH, our results suggest that a reorganization of humic macromolecules from a stretched to a coiled conformation takes place upon interaction with coagulant species. Such phenomenon is worth exploring, as the optimal coagulant dosage should then depend on the complex interplay between the rate of collisions of organic colloids/coagulant species and the rate of macromolecules reconformation. Acknowledgments This work was achieved within the framework of Zone Atelier du bassin de la Moselle, with nancial support from Region Lorraine and CNRS (Contrat de Plan Etat-Re gion and Programme Environnement Vie et Socie te ). Agence de lEau Rhin-Meuse is gratefully acknowledged for providing the Field-centrifuge. The authors would also like to thank T. Jacquin (Agence de lEau Rhin Meuse, Rozerieules, France) for providing technical assistance during eld measurements. References Andreux, F., Golebiowska, D., Metche, M., 1980. Oxydative polymerization of O-diphenols in presence or absence of amino-acids. Topics on (catechol-glycocolle) and (catechol- diglycylglycine). General Assembly of Polyphenols Group Report, Logrono, Spain, Bulletin, vol. 9, pp.178188. Balnois, E., Wilkinson, K.J., Lead, J.R., Bufe, J., 1999. Atomic force microscopy of humic substances: effects of pH and ionic strength. Environ. Sci. Technol. 33 (21), 39113917. Bose, P., Reckhow, D.A., 1998. Adsorption of natural organic matter on preformed aluminium hydroxide ocs. J. Envir- on. Eng. 124, 803810. Collins, M.R., Amy, G.L., King, P.H., 1985. Removal of organic matter in water treatment. J. Environ. Eng. 11, 850854. ARTICLE IN PRESS N o r m a l i z e d i n t e n s i t y
( A . U . ) NOM natural pH 7.4 0.08 0.1 0.12 (v) (r) (s) (t) (u) Energy Loss (eV) 250 350 400 450 (a) 0 0.02 0.04 0.06 1.1 10 -4 2.3 10 -4 1.4 10 -4 1.7 10 -4 2.0 10 -4 Iron concentration (M) E l e m e n t a l
a t o m i c
t a l
(r) (s) (t) (u) (v) (b) 300 NOM natural pH 7.4 Fig. 8. (a) EEL spectra for NOM as a function of coagulant dosage. (b) Elemental atomic ratio: N/C (), Fe/C * 1/10 (&), Na/C (K) and Ca/C (J). 0 1 2 3 4 5 6 7 8 9 10 0 10 -4 1.5 10 -4 2 10 -4 2.5 10 -4 NOM natural pH 7.4 (r) (s) (t) (u) (v) Iron concentration (M) 5 10 -5 Fig. 7. Jar-test results for NOM at natural pH (7.4). (K residual turbidity (NTU) sediment volume (mL)). A.-V. Jung et al. / Water Research 39 (2005) 38493862 3860 Collins, M.R., Amy, G.L., Steelink, C., 1986. Molecular weight distribution, carboxylic acidity and humic substances content of aquatic organic matter: implications for removal during water treatment. Environ. Sci. Technol. 20 (10), 10281032. Croue , J.-P., Lefebvre, E., Martin, B., Legube, B., 1993. Removal of dissolved hydrophobic and hydrophilic organic substances during coagulation/occulation of surface waters. Water Sci. Technol. 27, 143152. Dempsey, B.A., Ganho, R.N., OMelia, C.R., 1984. The coagulation of humic substances by means of aluminium salts. J. Am. Water Works Assoc. 74, 141150. Edwards, G.A., Amirtharajah, A., 1985. Removing color caused by humic acids. J. Am. Water Works Assoc. 77, 5057. Egerton, R.F., 1996. Electron Energy-Loss Spectroscopy in the Electron Microscope, second ed. Plenum Press, New York 423pp. El Samrani, A.G., Lartiges, B.S., Montarge` s-Pelletier, E., Kazpard, V., Barre` s, O., Ghanbaja, J., 2004. Clarication of municipal sewage with ferric chloride: the nature of coagulant species. Water Res. 38 (3), 756768. Engebretson, R.R., von Wandruszka, R., 1994. Microorganiza- tion in dissolved humic acids. Environ. Sci. Technol. 28 (11), 19341941. Furukawa, Y., 2000. Energy-ltering transmission electron microscopy (EFTEM) and electron energy-loss spectro- scopy (EELS) investigation of clay-organic matter aggre- gates in aquatic sediments. Org. Geochem. 31 (7/8), 735744. Gosh, K., Schnitzer, M., 1980. Macromolecular structures of humic substances. Soil Sci. 129, 266276. Gregory, J., Dupont, V., 2001. Properties of ocs produced by water treatment coagulants. Water Sci. Technol. 37, 3543. Gregor, J.E., Nokes, C.J., Fenton, E., 1997. Optimizing natural organic matter removal from low turbidity waters by controlled pH adjustment of aluminum coagulation. Water Res. 31 (12), 29492958. Gu, B., Schmitt, J., Chen, Z., Liang, L., McCarthy, J.F., 1994. Adsorption and desorption of natural organic matter on iron oxide: mechanisms and models. Environ. Sci. Technol. 28 (1), 3846. Henke, B.L., Gullikson, E.M., Davis, J.C., 1993. X-ray interactions: photoabsorption, scattering, transmission, and reection at E 5030000 eV, Z 192. At. Data Nucl. Data Tables 54, 181342. Hung, W.T., Chang, I.L., Lin, W.W., Lee, D.J., 1996. Unidirectional freezing of waste-activated sludges: effects of freezing speed. Environ. Sci. Technol. 30 (7), 23912396. Jeckel, M.R., 1986. Interactions of humic acids and aluminium salts in the occulation process. Water Res. 20 (12), 15351542. Jung, A.-V., 2004. Interactions des micropolluants anthropi- ques avec la matie` re organique naturelle et devenir lors des ope rations de traitement de leau. Ph.D. Thesis INPL, Nancy, 427pp. Jung, A.-V., Chanudet, V., Ghanbaja, J., Lartiges, B.S., Abdelmoula, M., Bersillon, J.-L., 2005a. Iron populations evolution during organic matter coagulation by a nitrate salt: energy-loss spectroscopy (EELS) and Mo ssbauer investigations. C.R. Acad. Sci. Chim. submitted for publication. Jung, A.-V., Frochot, C., Parant, S., Lartiges, B., Selve, C., Viriot, M.-L., Bersillon, J.-L., 2005b. Synthesis of amino- phenolic humic-like substances and comparison with natural aquatic humic acids: a multi-analytical techniques approach. Org. Geochem. in press. Kazpard, V., Lartiges, B.S., Jung, A-V., Frochot, C., Espinose de la Caillerie, J.B., Portal, J.M., Go rner, T., Viriot, M.L., Bersillon, J.L., 2005. Aggregation of a humic macromole- cules with polycations (Al 13 and PEI): a combined pyrene uorescence, electrophoretic mobility and SEC investiga- tions. Water Res. submitted for publication. Lartiges, B.S., Bottero, J.Y., Derrenger, L.S., Humbert, B., Tekely, P., Suty, H., 1997. Flocculation of colloidal silica with hydrolysed aluminium: an 27 Al solid state NMR investigation. Langmuir 13 (2), 147152. Lead, J.R., Wilkinson, K.J., Starchev, K., Canonica, S., Bufe, J., 2000. Determination of diffusion coefcients of humic substances by uorescence correlation spectroscopy: role of solution conditions. Environ. Sci. Technol. 34 (7), 13651369. Leenheer, J.A., Croue , J.-P., 2003. Characterizing aquatic dissolved organic matter. Environ. Sci. Technol. 1 (1), 19A26A. Lefebvre, E., Legube, B., 1990. Iron (III) coagulation of humic substances extracted from surface waters: effect of pH and humic substances concentration. Water Res. 24 (5), 591606. Matilainen, A., Lindqvist, N., Korhonen, S., Tuhkanen, T., 2002. Removal of NOM in the different stages of the water treatment process. Environ. Int. 28 (11), 457465. Maurice, P.A., Namjesnik-Dejanovic, K., 1999. Aggregates structures of sorbed humic substances observed in aqueous solution. Environ. Sci. Technol. 33 (9), 15381541. Myneni, S.C.B., 2002. Soft X-ray spectroscopy and spectro- microscopy studies of organic molecules in the environ- ment. Reviews in Mineralogy and Geochemistry (Applications of Synchrotron Radiation in Low-Tempera- ture Geochemistry and Environmental Sciences) 49, 485579. Oliver, B.G., Visser, S.A., 1980. Chloroform production from the chlorination of aquatic humic material: the effect of molecular weight, environment and season. Water Res. 14 (9), 11371141. Owen, D.M., Amy, G.L., Chowdhury, Z.K., Paode, R., McCoy, G., Viscosil, K., 1995. NOM characterization and treatability. J. Am. Water Works Assoc. 87, 4663. Plaschke, M., Ro mer, J., Klenze, R., Kim, J.I., 1999. In situ AFM study of sorbed humic acid colloids at different pH. Colloids Surfaces A: Physicochem. Eng. Aspects 160, 269279. Randtke, S.J., 1993. Coagulation of NOM: an overview of the science and U.S. practice. NOM in drinking water, origin, characterization and removal. Workshop Proceedings of AWWA about Natural Organic Matter in Drinking Water. AWWA Research Fundation Ed., pp.155163. Specht, C., Kumke, M., Frimmel, F., 2000. Characterization of NOM adsorption to clay minerals by size exclusion chromatography. Water Res. 34 (16), 40634069. ARTICLE IN PRESS A.-V. Jung et al. / Water Research 39 (2005) 38493862 3861 Stevenson, I.L., Schnitzer, M., 1982. Transmission electron microscopy of extracted fulvic and humic acids. Soil Sci. 133 (3), 179185. Thieme, J., Schmidt, C., Abbt-Braun, G., Specht, C., Frimmel, F.H., 2002. X-ray microscopy studies of refractory organic substances. In: Frimmel, Abbt-Braun, Heumann, Hock, Lu demann, Spiteller (Eds.), Refractory Organic Substances in the Environment. Wiley-VCH, Weinheim, pp. 239248. Thurman, E.M., Malcolm, R.L., 1981. Preparative isolation of aquatic humic substances. Environ. Sci. Technol. 15 (4), 463466. Tipping, E., 1993. Modeling the competition between alkaline earth cations and trace metal species for binding by humic substances. Environ. Sci. Technol. 27, 520529. Van Breemen, A.N., Nieuwstad, T.J., Van der MeentOlieman, 1979. The fate of fulvic acids during water treatment. Water Res. 13 (8), 771780. Van den Hoop, M.A.G.T., van Leeuwen, H.P., Pinheiro, J.P., Mota, A.M., Simo es Gonc- alves, M.L., 1995. Voltametric analysis of the competition between calcium and heavy metals for complexation by humic material. Colloids Surfaces A: Physicochem. Eng. Aspects 95, 305313. Varlot, K., Martin, J.M., Quett, C., 1998. Physical and chemical changes in polystyrene during electron irradiation using EELS in the TEM: contribution of the dielectric function. J. Microsc. 191 (2), 187194. Vermeer, A.W.P., van Riemsdijk, W.H., Koopal, L.K., 1998. Adsorption of humic acid to mineral particles. 1-Specic and electrostatic interactions. Langmuir 14 (10), 28102819. Vilge -Ritter, A., Masion, A., Boulange , T., Rybacki, D., Bottero, J.Y., 1999a. Removal of natural organic matter by coagulation-occulation: a pyrolysis-GC-MS study. Environ. Sci. Technol. 33 (17), 30273032. Vilge -Ritter, A., Rose, J., Masion, A., Bottero, J.-Y., Laine , J.- M., 1999b. Chemistry and structure of aggregates formed with Fe-salts and natural organic matters. Colloids Surfaces A: Physicochem. Eng. Aspects 147 (3), 297308. Villemin, G., Mansot, J.L., Watteau, F., Ghanbaja, J., Toutain, F., 1995. Etude de la biode gradation de la matie` re organique ve ge tale du sol par la spectroscopie des pertes de nergie de lectrons transmis (EELS: Electron Energy Loss Spectroscopy): re partition du carbone, de lazote et e valua- tion du rapport C-to-N au niveau ultrastructural in situ. C.R. Acad. Sci. Paris 321 (IIa), 861868. Watteau, F., Villemin, G., Mansot, J.L., Ghanbaja, J., Toutain, F., 1996. Localization and characterization by electron energy loss spectroscopy (EELS) of the brown cellular substances of beech roots. Soil Biol. Biochem. 28 (10/11), 13271332. Wilkinson, K.J., Balnois, E., Leppard, G.G., Bufe, J., 1999. Characteristic features of the major components of fresh- water colloidal organic matter revealed by transmission electron and atomic force microscopy. Colloids Surfaces A: Physicochem. Eng. Aspects 155, 287310. Further reading Bruchet, A., Rousseau, C., Mallevialle, J., 1990. Pyrolysis-GC- MS for investigating high-molecular-weight THM precur- sors and other refractory organics. J. Am. Water Works Assoc. 71, 6674. Hitchcock, A.P., Urquhardt, S.G., Rightor, E.G., 1992. Inner- shell spectroscopy of benzaldehyde, terephtalaldehyde, ethylbenzoate, terephtaloyd chlorides and phosphogen: models for core excitation of poly(ethylene) terephtalate. J. Phys. Chem. 96 (22), 87368750. Johnson, P.N., Amirtharajah, A., 1983. Ferric chloride and alum as single and dual coagulants. J. Am. Water Works Assoc. 54, 232239. Kikuna, J., Tonner, B.P., 1996. XANES spectra of a variety of widely used organic polymers at the C K-edge. J. Electron Spectrosc. Relat. Phenom. 82 (1/2), 5360. Lefebvre, E., Legube, B., 1993. Coagulation-occulation by ferric chloride of some organic compounds in aqueous solutions. Water Res. 27 (3), 433447. OMelia, C.R., 1972. Coagulation and occulation. In: Weber, Jr., W. (Ed.), Physicochemical Processes for Water Quality Control. Wiley Interscience, New York 14pp. Packam, R.F., 1965. Some studies of the coagulation of dispersed clays with hydrolysing salts. J. Colloid Sci. 20, 8192. Rose, J., Vilge , A., Lauquet-Olivie, G., Masion, A., Frechou, C., Bottero, J.Y., 1998. Iron speciation in natural organic matter colloids. Colloids Surfaces A: Physicochem. Eng. Aspect 136 (1/2), 1119. Semmens, M.L., Field, T.K., 1980. Coagulation: experiences in organics removal. J. Am. Water Works Assoc. 72, 476483. Stumm, W., OMelia, C.R., 1968. Stoechiometry of coagula- tion. J. Am. Water Works Assoc. 60, 514539. ARTICLE IN PRESS A.-V. Jung et al. / Water Research 39 (2005) 38493862 3862