Inﬂ uence of the apparent molecular size of humic substances on the efﬁ ciency of coagulation using Fenton’s reagent

This work used Fenton’s reagent as a coagulating agent in the treatment of water samples with high true colour caused by humic substances (HS) extracted from peat. In addition, the effects of the apparent molecular size of HS on coagulation, ﬂ occulation and ﬂ otation were studied. To that end, four distinct water samples having the same true colour were prepared using HS with different molecular sizes, which were obtained by ultraﬁ ltration fractioning. Through optimisation of coagulant dosage and coagulation pH, as well as posterior construction of coagulation diagrams for each water sample, it was veriﬁ ed that the sample prepared with the smallest apparent molecular size of HS was the most difﬁ cult to treat, requiring higher coagulant (Fenton’s reagent) dosages than samples prepared with larger HS molecular sizes. Furthermore, ﬁ ltration experiments after dissolved air ﬂ otation (DAF) were carried out in an attempt to simulate conventional treatment. The most representative results in ﬁ ltered water were: apparent colour ≤ 3 HU; turbidity < 0.5 NTU; and residual iron concentration < 0.005 mg/L. The absorbance and total organic carbon values of ﬁ ltered water samples were also very low, suggesting that the formation of disinfection by-products with chlorine would likely be insigniﬁ cant.


INTRODUCTION
Humic substances (HS) are heterogeneous mixture of dark-coloured organic macromolecules of complex composition; they are present in soil and aquatic ecosystems and function as effective carbon reservoirs that prevent carbon dioxide emission.Humic substances comprise 60-70% total soil carbon and 60-90% total carbon dissolved in natural waters (Richard et al. 2008).HS are not well-defi ned substances but can generally be subdivided in three fractions, namely: humins, representing the insoluble humic components of aqueous solutions at all pH values; humic acids (HA), which are soluble in alkaline and weakly acidic solutions but precipitate at or below pH 2.0; and fulvic acids (FA), which are soluble in aqueous solutions at all pH values (Stevenson 1994).Humic materials may be specifi cally targeted for removal 833-847 MARCELO DE JULIO, TATIANE S. DE JULIO and LUIZ DI BERNARDO from potable water supplies because they can adversely affect appearance and taste and they can react with chlorine to form potentially carcinogenic chlorinated organic compounds (Lin andWang 2011, Selcuk et al. 2011).Furthermore, the presence of macromolecular dissolved organic matter may reduce the effectiveness of water treatment processes that employ membranes or microporous adsorbents (Zularisam et al. 2006, Zheng et al. 2011).Even when not specifi cally targeted for removal, macromolecular dissolved organic matter has been shown to compete with low molecular weight synthetic organic chemicals, reducing their adsorption rates and equilibrium capacities (O'Melia et al. 1999, Katsumata et al. 2008, Vreysen and Maes 2008).The removal of (or the reduction in concentration of) such organic matter is therefore an important factor in water treatment.
The fl ocs formed by HS coagulation are relatively small and have low specifi c masses (Fusheng et al. 2008), mainly with low turbidity and high true colour.As a result, their removal by sedimentation is limited.Thus, the use of dissolved air fl otation (DAF) is an interesting clarifi cation technology for water with these characteristics.
The use of Fenton's reagent, which entails adding Fe +2 ions and hydrogen peroxide (H 2 O 2 ) under acid conditions, is advantageous for drinking water treatment because it provides both pre-oxidation and coagulation in a single process.As shown in Equation ( 1), Fenton's reaction forms the hydroxyl free radical (OH • ), which is a highly reactive and non-selective intermediate oxidant capable of effectively oxidising numerous organic substances (Peyton 1988, Nogueira andGuimarães 2000).Because both Fe +2 and Fe +3 ions form hydrolysed species that act as coagulants, Fenton's reagent can have the double function of oxidation and coagulation during treatment processes.
The strong oxidative power of Fenton's reagent that results from the hydroxyl radical (OH • ) is also advantageous as the pre-oxidation step becomes increasingly necessary to address poor raw water quality.With Fenton's reagent, undesirable halogenated by-products are not formed (since the raw water does not have the precursor's agents, such as bromide and iodine).Fenton's reagent is further advantageous for natural organic matter (NOM) removal because the coagulants traditionally used in water treatment (alum or iron salts) are frequently incapable of removing recalcitrant NOM (Fabris et al. 2004), making alternative treatment processes necessary.Advanced oxidative processes (AOP), including the application of Fenton's reagent, have thus been introduced as alternatives.
To our knowledge, there is a dearth of reports on the degradation of HS using Fenton's reagent (Fukushima et al. 2001, Murray and Parsons 2004ab, Katsumata et al. 2008).Some studies have been carried out showing the infl uence of apparent HS molecular sizes on coagulation with traditional coagulants, fl occulation and sedimentation (Ratnaweera et al. 1999, Campos et al. 2005, 2007), but data on the use of DAF and Fenton's reagent are limited.
Thus, the main objective of this study was to evaluate the effects of apparent HS molecular size on the effi ciency of the coagulation, fl occulation and fl otation of water samples with the same true colour using Fenton's reagent as coagulant.In addition to that, the aim of this study was to investigate the resulting fi ltered water quality by means of sand fi ltration experiments following DAF, using each water sample prepared with HS of different apparent molecular sizes.

COLLECTION AND EXTRACTION OF HS
The HS were extracted from peat soil collected from the banks of the Mogi Guaçu River in Brazil HUMIC SUBSTANCE MOLECULAR SIZE INFLUENCE ON FENTON'S COAGULATION (latitude 21.5° S and longitude 47.9° W).Peat samples were dried and then coarsely ground in an agate mortar.The HS extraction method was as follows: the peat was added to a KOH extraction solution (0.5 mol/L) in a 1:20 (m/v) ratio; this solution was mixed for 4 hours at room temperature (~25°C) and then allowed to sediment for 48 hours, after which the supernatant was stored in semipermeable paper bags.These bags were placed in a hydrochloric acid solution (HCl 1%) until the HS extracts reached a pH close to 6.0.Finally, residual Cl − was removed via dialysis in non-chlorinated water until HS samples tested negative for chloride (using an AgNO 3 -based test).
Following extraction, the HS solution was frozen and stored in plastic bottles.

ULTRAFILTRATION (UF) FRACTIONATION OF HS
The extracted HS were diluted to a concentration of 1.0 mg/mL using ground water.The solution was vacuum filtered using a 0.45 mm (Millipore) membrane for subsequent UF fractionation (polyethersulfone (polyethersulfone ( membranes, Vivaflow 50, Satorius group, tangential flow, Vivascience).During HS fractionation, a sample concentration method using recirculation was adopted (Duarte et al. 2001).The maximum applied flow rate was 300 mL/min with 1.5 bar pressure.The following apparent molecular sizes fractions were obtained: between 100 kDa and 0.45 mm; between 30 and 100 kDa; and < 30 kDa.A composite sample containing all size fractions was also obtained, filtered only through the 0.45 mm membrane.

WATER SAMPLES
Four distinct water samples were prepared for each of the three HS size fractions (100 kDa-0.45mm, 30-100 kDa and < 30 kDa) and the composite fraction (0.45mm fi ltered).Each HS fraction was independently added to ground water.The four water samples exhibited true colour values of 100 ± 5 HU (Hazen units), and the pH of each sample was adjusted to 5.0 ± 0.1.

COAGULATION, FLOCCULATION AND FLOTATION TESTS
Coagulation diagrams were built for the four water samples by varying the coagulant dosage (Fenton's reagent) and the coagulation pH; 0.1 M sodium hydroxide (NaOH; A.R., Mallinckrodt) and hydrochloric acid (HCl; 36.5-38%,P.A., Synth) solutions were used to vary the coagulation pH.The temperature of the water samples was maintained at 20 ± 1°C.
Fenton's reagent was applied as a coagulant as follows.First, FeSO 4 x 7H 2 O was dosed, followed by H 2 O 2 .Immediately after the addition of these chemical products, a sample was collected for oxidation pH measurement: 1.5 min of oxidation time (the average velocity gradient of this step was the same as for the rapid-mix step).After this time, the alkalinising agent was dosed (when started to count rapid-mix time) and another sample was collected for coagulation pH measurement.The hydrogen peroxide dosage was equal to three times (3x) the stoichiometric dosage required for Fenton's reagent (Equation 1), being this ratio employed for all tested ferrous sulphate dosages.For example, the mass of 1 mol of FeSO 4 x 7H 2 O is 278.02 g, which corresponds to a Fe +2 mass of 55.85 g.According to Equation ( 1 ); fl occulation (T fl = 15 min; G fl = 25 s -1 ); and DAF (Saturation chamber pressure = 400 kPa; Saturation time = 8 min; Recirculation rate = 15%; Flotation velocities [V f ] of 15, 10 and 5 cm/min).The control f ] of 15, 10 and 5 cm/min).The control f parameters during these experiments were as follows: oxidation pH; coagulation pH; fl otation water pH (to choose which curve would be utilised to make the colour measurement) (Digimed); remaining apparent colour (subnatant); and remaining UV 254 absorbance (DR/4,000 U, Hach, spectrophotometer).During all pH measurements, samples were stirred using a magnetic stirrer (Quimis).In all colour and absorbance measurements, the dilution caused by the introduction of saturated water was taken into account.For colour measurements, (DR/4,000 U, Hach) spectrophotometer calibration curves were constructed with the extracted HS using the successive dilution method, as described in Standard Methods (1998).
For each water sample, a pair of values 'coagulant dosage versus coagulation pH' was selected in the respective diagrams based on remaining apparent colour and UV 254 absorbance, and taking into account the three fl otation velocities studied.However, although 24 coagulation diagrams were generated, only the diagrams for apparent colour and V f = 5 cm/min are presented.f = 5 cm/min are presented. f In the coagulation diagrams, the abbreviation CI indicates that coagulation was ineffi cient.Floc formation was not observed for these points, and consequently, no colour or absorbance measurements were done.

TESTS
With the selected pair of values 'coagulant dosage versus coagulation pH' for each water sample two coagulation, fl occulation, fl otation and fi ltration tests were carried out in order to simulate a complete treatment process.In addition to the control parameters already mentioned it was also measured on fi ltered water: pH; apparent colour; UV 254 absorbance; turbidity (2100P, Hach, Turbidimeter); total organic carbon (TOC analyser 5000A, Shimadzu); and total iron concentration (Atomic absorption spectrophotometer, model AA-1275, Varian).
The fi ltration step employed a fi lter column coupled immediately downstream from the fl otest fl ask.The fi lter column was a 19-mm internal diameter acrylic tube containing sand as the fi lter media (10 cm deep).It was used three types of  Types 1, 2 and 3) with granulometric ranges between 0.27 and 0.59 mm, 0.42 and 0.84 mm, and 0.59 and 1.41 mm, respectively.The average fi ltration fl ow rate was 16 mL/min (corresponding to a fi ltration rate of 80 m 3 /m 2 .d),and samples of the fi ltered water were collected 30 min after the fl otation time (fl otation velocity of 5 cm/min).Because the fi lters were backwashed with tap water, it was necessary to fi lter for 30 min to fl ush this water out of the fi lters before collecting samples.
Figure 1 shows schematic fl owchart of experiments.According to classical theories, there are two predominant mechanisms of metal ion coagulation, 'charge neutralisation' and 'sweep coagulation'; the former occurs at lower pH conditions with the coagulant in its cationic form (e.g., Fe[OH] 2+ , Fe[OH] 2 + ), and the latter at higher pH and higher doses where the coagulant precipitates as a metal hydroxide (Duan and Gregory 2003).MARCELO DE JULIO, TATIANE S. DE JULIO and LUIZ DI BERNARDO An analysis of Figures 2-5 reveals two regions in which coagulation with Fenton's reagent and subsequent fl occulation and fl otation produce removal effi ciencies greater than or equal to 50% and 60% for the fi rst (lower pH) and second (higher pH) region, respectively.Similar behaviour was observed by Edwards and Amirtharajah (1985), who also worked with water sample with true colour of 100 HU (caused by the addition of humic acid) and turbidity equal to zero, using aluminium sulphate (alum) as a coagulant.These researchers concluded that in the fi rst region, corresponding to approximate coagulation pH values of 3.6 to 4.6 in the current study (Figures 2-5), the hydrolysing species adsorption mechanism was probably dominant, causing charge neutralisation of HS molecules.In the second region, which is approximately between pH 5.6 and 6.6 in the current study, Fe(OH) 3 precipitate formation was observed, and removal possibly occurred via the adsorption of HS molecules and their incorporation into the precipitate.Furthermore, coagulant dosages (Fenton) were lower in the fi rst region than for the second region.As observed by Edwards and Amirtharajah (1985), it seems to have a restabilisation region between the two regions in which the coagulation resulted effi cient.

COAGULATION DIAGRAMS
Even knowing that Fenton reaction is a combination of oxidation and coagulation, these results suggest that coagulation using Fenton's reagent produces behaviour similar to that observed in studies using other coagulants, such as ferric chloride and alum.Moreover, oxidation and coagulation probably are removing HS -the former changes the chemical structures of HS molecules, As shown by Katsumata et al. (2008), in the photo-Fenton system, the molecular size of HA was decreased as a result of irradiation.In a series of photo-Fenton processes, therefore, the generation of CO 2 and the formation of ring opening products may contribute to a decrease in molecular size.However, the degradation mechanism of HA during the photo-Fenton process is not presently clear.
For the water sample with the HS fraction fi ltered through a 0.45 µm membrane and with an apparent HS molecular size > 100 kDa, it is evident that the regions formed in the coagulation diagram are wider, and it is observed that the isoeffi cient curves containing the best results (lower remaining apparent colour) comprise a larger area in Figure 3 than in Figure 2. Furthermore, the highest colour removal effi ciency was observed for a ferrous sulphate dosage f MARCELO DE JULIO, TATIANE S. DE JULIO and LUIZ DI BERNARDO of 7.5 mg/L for the water sample with an apparent HS molecular size fraction between 100 kDa and 0.45 µm, whereas this dosage was between 10 and 15 mg/L for the water sample with the HS fraction only fi ltered through a 0.45 µm membrane (with similar coagulation pH values).These results can be explained by the hypothesis that for this water sample (between 100 kDa and 0.45 µm), the smallest HS molecules (< 100 kDa) were removed.According to Campos et al. (2005Campos et al. ( , 2007)), these molecules have higher fulvic acid concentrations than humic acid and have higher proportions of bound oxygen groups.These authors concluded that these groups therefore have greater amounts of negative charge due to the non-bonded electrons, which favour repulsion between these electrons and colloidal particles, thereby negatively infl uencing the coagulation process.Campos et al. (2005Campos et al. ( , 2007) ) worked with the same HS used in the present research (same research group) and characterised the different fractions using nuclear magnetic resonance (NMR) and infrared spectroscopy (IR), in addition to determining the percentage of humic and fulvic acids.
For the water sample with the HS apparent molecular size fraction between 30 and 100 kDa (Figure 4), it can be observed that the isoeffi ciency curves leading to the most effective coagulation were shifted to a slightly lower coagulation pH range (for the second region on the diagrams).For the size fraction < 30 kDa (Figure 5), these curves showed a more accentuated shift, indicating that the smallest HS molecules -which present higher fulvic acid concentrations in relation to humic acid and higher proportions of bound oxygen groups f HUMIC SUBSTANCE MOLECULAR SIZE INFLUENCE ON FENTON'S COAGULATION (Campos et al. 2005(Campos et al. , 2007) ) -required a somewhat lower coagulation pH in comparison to the larger apparent HS molecular sizes.For the fraction < 30 kDa, ferrous sulphate dosages up to 15 mg/L (for the second region on the coagulation diagrams) were required for fl oc formation, indicating that lower dosages were insuffi cient to destabilise the HS.However, for the larger apparent HS molecular sizes (Figure 3), higher removal effi ciencies were reached at dosages lower than or equal to this value.
It was also observed that, for the water sample with an HS size fraction < 30 kDa, the fl otation velocity of 15 cm/min was not effi cient for removing apparent colour (data not presented), indicating that the smaller HS molecular sizes require lower fl otation velocities and, consequently, lower superfi cial application rates must be used in continuous fl ow treatment plants.
As previously mentioned, the coagulation diagrams were constructed for fl otation velocities of 15, 10 and 5 cm/min and for remaining apparent colour and UV 254 absorbance.All of these factors were taken into account when the selected point was chosen (which corresponds to a pair of values coagulation dosage versus coagulation pH), and because of this, the lowest remaining apparent colour found in the diagrams for V f = 5 cm/min was f = 5 cm/min was f not always selected.
A general analysis of all the coagulation diagrams (Figures 2-5) reveals that the selected point for the HS fraction with apparent molecular size between 30 and 100 kDa required a coagulant dosage (Fenton) close to 166% higher than that for the selected point for the HS fraction between 100 kDa and 0.45 µm, and 33% higher than that required for the water sample having the HS fraction only fi ltered through a 0.45 µm membrane.The selected point for the HS fraction < 30 kDa required twice the dosage needed for the fraction between 30 and 100 kDa and approximately 432% higher than that for the fraction between 100 kDa and 0.45 µm.Ratnaweera et al. (1999) also fractionated HS (composed of NOM extracted from natural water by reverse osmosis) using UF (employing polyethersulfone membranes with tangential fl ow) in different apparent molecular sizes (< 10 kDa, between 10 and 50 kDa, between 50 and 100 kDa, and > 100 kDa) and observed that the HS of larger apparent molecular size required lower coagulant dosages in comparison with those of smaller apparent molecular size.These experiments were performed in a jartest apparatus and included coagulation, fl occulation and sedimentation.Campos et al. (2005Campos et al. ( , 2007) also performed jartest experiments using coagulation, fl occulation and sedimentation with alum as a coagulant and found similar results.Thus, it is possible to conclude that, regardless of the coagulant used (even Fenton's reagent, which is both an oxidant and a coagulant) and the clarifi cation technology (sedimentation or fl otation) employed, waters predominantly containing HS molecules of smaller apparent size are more diffi cult to treat.
As verifi ed by Campos et al. (2005Campos et al. ( , 2007) ) the smallest HS fraction presented a higher percentage of fulvic acids in relation to humic acids.The fulvic acids have smaller chains, with structures dominated by aliphatics, a higher number of functional carboxylic groups, and phenolic and alcoholic hydroxyls.These characteristics make the HS more hydrophilic and acidic (Stevenson 1994).These fi ndings suggest that not only the apparent molecular size but also the structural characteristics of HS fractions play a signifi cant role in the coagulation process.

FILTRATION EXPERIMENTS AFTER DAF
Tables I and II show the results of fi ltration experiments following DAF (two replicate experiments for each water sample).The chemical product dosages applied correspond to the dosages of the selected points on the coagulation diagrams (Figures 2-5) for each water sample, and its characterization is also shown in Figures 2-5 ); (A/A 0 )*100 -A is the remaining absorbance (cm -1 ) and A 0 is the absorbance (cm -1 ) of the water sample; *** TOC -Total Organic Carbon.); (A/A 0 )*100 -A is the remaining absorbance (cm -1 ) and A 0 is the absorbance (cm -1 ) of the water sample; *** TOC -Total Organic Carbon.MARCELO DE JULIO, TATIANE S. DE JULIO and LUIZ DI BERNARDO An analysis of Tables I and II shows that, for the eight experiments and the three sand granulometry types, the remaining apparent colour, turbidity and total iron concentrations after fi ltration were always less than or equal to 3 HU, 0.50 NTU (except for experiment 7, in which this value was ≤ 0.60) and 0.005 mg/L (below the detection limit for all samples), respectively.Brazilian drinking water standard (Brazil 2004) set a maximum value for apparent colour after fi ltration of 15 HU (the same value recommended by WHO 2006) and a maximum turbidity value of 1 NTU after rapid fi ltration (the same value require by the Council Directive 1998, for the treatment of superfi cial waters) and recommend a goal of achieving a maximum turbidity of 0.5 NTU after rapid fi ltration.The maximum total iron concentration for Brazilian (Brazil 2004) and European (Council Directive 1998) drinking water standards are 0.3 and 0.2 mg/L, respectively.Thus, the coagulation, fl occulation, DAF and fi ltration treatment employing Fenton's reagent as coagulant meet Brazilian and European drinking water standards (in relation to the measured parameters) for the four water samples studied.
The remaining apparent colour, UV 254 absorbance and turbidity values were very close each other for the three sand types.Curiously, the highest TOC values were obtained for sand of Type 1 (except for experiment 8), which had the smallest granulometry.
Thus, Type 3 sand seems to be the most appropriate for the treatment of water containing a range of apparent HS molecular size fractions because, in addition to producing the best results, this sand has the largest granulometry among the tested sands.This factor would result in a lower head loss in an actual treatment installation and would therefore result in longer fi ltration run times.
Water samples prepared with the smallest apparent HS molecular sizes also had higher TOC concentrations in the fi ltered water (sand Type 3) compared to water prepared with larger apparent HS molecular sizes, indicating that the smallest HS size fractions were more diffi cult to remove.This fact supports the results found in the coagulation diagrams.
According to AWWA (1999a), disinfection byproduct (DBP) concentrations depend on several factors, chief among which are TOC level or UV 254 and disinfectant type and concentration.One way to minimise by-product formation is to limit the TOC concentration of fi ltered water.To adequately reduce the risk of DBP formation, according to the AWWA (1999b), TOC concentrations (in fi ltered water) must be less than 2.0 mg/L.Thus, in this work, the TOC concentrations were well below safe levels (always < 1.0 mg/L), particularly for experiments using Type 3 sand, with an average removal effi ciency for all HS fractions of 80%, indicating that disinfection by-product formation would likely be insignifi cant (Brazilian standards also require that water be disinfected with chlorine and an adequate residual be maintained).
The results of this study suggest that the application of Fenton's reagent in water treatment plants (WTP) may be benefi cial as this product does not contribute to the formation of halogenated pre-oxidation by-products.In addition, Fenton's reagent accomplishes both pre-oxidation and coagulation, and as observed, conventional treatment (coagulation, fl occulation, DAF and fi ltration) using Fenton's reagent as a coagulant results in lower TOC and UV 254 absorbance values for the fi ltered water produced.Therefore, Fenton's reagent seems to be a promising alternative for the treatment of waters containing high concentrations of organic compounds.
With its high oxidative potential (2.33 V; Huang et al. 1993), the free hydroxyl radical (OH • ) generated during Fenton's reaction may have other benefi ts, such as the mineralization of hazardous compounds in the water, which can minimize the amount and toxicity of the sludge to be treated and disposed.The fi rst author have already applied HUMIC SUBSTANCE MOLECULAR SIZE INFLUENCE ON FENTON'S COAGULATION Fenton's reagent in a real scale WTP in Brazil, which treated raw water that received big contribution of wastewater, and obtained good performance concerning the removal of surfactant agents, but the modifi cation in the plant was not adopted because of the high costs of chemical products.

CONCLUSIONS i)
Water samples prepared with the same true colour but with different apparent HS molecular sizes and different humic and fulvic acids percentages showed different coagulation conditions.For water samples with the smallest apparent molecular size fractions, higher dosages of coagulant were needed (up to 432% higher), mainly because these water samples contained higher concentrations of fulvic acids, which have a larger number of negativelycharged groups.
ii) It is recommended that the coagulation conditions of waters from different sources never be extrapolated, even when such waters have similar true colour values.It is necessary to conduct treatability studies to determine appropriate coagulation conditions and design parameters for each water type.iii) Coagulant dosage versus coagulation pH values were optimised, and posterior coagulation diagrams were constructed.Two distinct regions were observed in these diagrams in which coagulation, fl occulation and DAF using Fenton's reagent as a coagulant produced HS removal effi ciencies greater than or equal to 50-60%.The fi rst region occurred at a lower pH range (approximately 3.6 to 4.6) and required lower coagulant dosages (Fenton).The second region occurred at a higher pH range (approximately 5.6 to 6.6) and required higher coagulant dosages.In the fi rst region, coagulation was dominated by the hydrolysing species adsorption mechanism, resulting in charge neutralisation of HS molecules.In the second region, Fe(OH) 3 precipitate formation was the likely removal mechanism, resulting in the adsorption of HS molecules and their incorporation into this precipitate.A restabilisation region also appears to exist between the two regions of effi cient coagulation.iv) With fi ltration experiments following DAF, the fi ltered water had remaining apparent colour, turbidity and total iron values of less than or equal to 3 HU, 0.50 NTU (except for experiment 7, in which the value was ≤ 0.60) and 0.005 mg/L, respectively.These results comply with Brazilian and European drinking water standards (with respect to these parameters), and suggest that Fenton's reagent can be effectively used as a coagulant and not only as oxidant (as vastly applied in the literature).Furthermore, the replacement of alum (one of the most widely used coagulants) by an iron salt (such as Fenton's reagent) may yield additional benefi ts because researchers have questioned the use of alum on the grounds that this product may cause health problems.It is therefore recommended that Fenton's reagent be added to the list of chemical products to be tested in treatability studies.
v) The application of Fenton's reagent in WTP may be benefi cial because this product is unlikely to contribute to the formation of halogenated preoxidation by-products (since the raw water does not contain the precursors agents).Furthermore, Fenton's reagent accomplishes both pre-oxidation and coagulation, and conventional treatment using Fenton's reagent was found to effectively lower TOC (always < 1.0 mg/L and nearly 80% removal effi ciency) and UV 254 absorbance values in fi ltered water.Therefore, Fenton's reagent seems to be a promising alternative for the treatment of waters with high concentrations of organic compounds.MARCELO DE JULIO, TATIANE S. DE JULIO and LUIZ DI BERNARDO fi rst author (process # 02/07680-2).The authors are also very grateful to Professor Eduardo Fausto de Almeida Neves (in memoriam).

Figures 2 -
Figures 2-5 show the coagulation diagrams for the four water samples, which were built after coagulation, fl occulation and DAF experiments.Hydrogen peroxide and ferrous sulphate dosages are shown on the left-hand y axes, while the correspondent iron-II dosages are shown on the right.The coagulation pH is plotted as the abscissa.The main characteristics of each water sample are also shown.

Figure 2 -
Figure 2 -Fenton dose versus pH coagulation diagram -remaining apparent colour (HU) for the water sample with the HS fraction fi ltered through a 0.45 µm membrane (V f = 5 cm/min).f = 5 cm/min).f Note: CI indicates that the coagulation was ineffi cient.

Figure 3 -
Figure 3 -Fenton dose versus pH coagulation diagram -remaining apparent colour (HU) for the water sample with the HS fraction fi ltered through a 0.45 µm membrane with apparent molecular size > 100 kDa (V f = 5 cm/min).f = 5 cm/min).f

Figure 4 -
Figure 4 -Fenton dose versus pH coagulation diagram -remaining apparent colour (HU) for the water sample with the HS fraction with apparent molecular size between 30 and 100 kDa (V f = 5 cm/min).f = 5 cm/min).

Figure 5 -
Figure 5 -Fenton dose versus pH coagulation diagram -remaining apparent colour (HU) for the water sample with the HS fraction with apparent molecular size < 30 kDa (V f = 5 cm/min).f = 5 cm/min).

TABLE I Results for sand fi ltration tests after DAF for the water sample with the HS fraction fi ltered through a 0.45 µm membrane and for the water sample with the HS fraction between 100 kDa and 0.45 µm.
. MARCELO DE JULIO, TATIANE S. DE JULIO and LUIZ DI BERNARDO * Rem.ap.col.-Remaining apparent colour; ** % rem.Abs -Percentage remaining of Absorbance at 253.7 nm (cm -1

TABLE II Results for sand fi ltration tests following DAF for the water samples with HS fractions with apparent molecular sizes between 30 and 100 kDa and < 30 kDa.
HUMIC SUBSTANCE MOLECULAR SIZE INFLUENCE ON FENTON'S COAGULATION *Rem.ap.col.