flocculation treatment

Journal of Colloid and Interface Science 342 (2010) 26–32

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science www.elsevier.com/locate/jcis

Effect of organic matter on arsenic removal during coagulation/flocculation treatment Virginie Pallier a, Geneviève Feuillade-Cathalifaud a,*, Bernard Serpaud b, Jean-Claude Bollinger b a b

Université de Limoges–Groupement de Recherche Eau Sol Environnement (GRESE), ENSIL, Parc d’Ester Technopôle, 16 rue Atlantis, 87068 Limoges, France Université de Limoges–Groupement de Recherche Eau Sol Environnement (GRESE), Faculté des Sciences et Techniques, 123 avenue Albert Thomas, 87060 Limoges, France

a r t i c l e

i n f o

Article history: Received 22 June 2009 Accepted 30 September 2009 Available online 4 October 2009 Keywords: Arsenic Speciation Coagulation/flocculation Iron Precipitation Adsorption Zeta potential

a b s t r a c t The aim of this study is to evaluate the influence of organic matter on arsenic removal by coagulation/ flocculation on both a model water with low mineral content and a natural water sample. Ferric chloride was used as coagulant at concentrations avoiding the preoxidation step usually required to oxidize As(III) and increase its removal. Arsenic removal was accomplished by combining evaluation of arsenic residual concentrations and speciation analysis with zeta potential measurements. A preliminary study evaluated the influence of coagulant dose, coagulation pH, and organic matter on As(III) and As(V) removal. The main conclusions were: (i) As(III) removal depended on coagulant dose and on the number of sites available on hydroxide surfaces rather than on coagulation pH; (ii) As(V) removal depended on the zeta potential of colloidal suspension and was more influenced than As(III) by coagulation pH and the presence of organic matter; (iii) organic matter removal followed As(V) removal. This allowed determination of adsorption as the main mechanism occurring during As(V) and organic matter removal and supposing precipitation/coprecipitation as an important As(III) removal mechanism. Adsorption on preformed ferric hydroxide flocs experiments allowed then confirmation of these hypotheses. Ó 2009 Elsevier Inc. All rights reserved.

1. Introduction Health problems caused by the presence of arsenic in the environment have been a major concern for years because of its high acute and chronic toxicity. Inorganic arsenic compounds are more toxic than organo-arsenic ones [1] and As(III) is more toxic than As(V) [2,3]. While these compounds are released in surface and ground waters by both natural processes and anthropogenic activities, consumption of drinking water represents the main source of exposure to inorganic arsenic [4–7]. As a consequence the maximum contaminant level of total arsenic in drinking water was reduced from 50 to 10 lg L1 (from 0.67 to 0.13 lmol L1) in many countries. Existing processes must thus be optimized and new processes developed to remove arsenic efficiently and to comply with this new drinking water standard. The most promising processes for arsenic removal from highly contaminated waters are coagulation/ flocculation and adsorption because of their low cost and high efficiency [8,9]. Coagulation/flocculation, using mainly iron or aluminum salts as coagulants, is a well-known method for arsenic removal that can convert soluble arsenic species into insoluble reaction products [8]. Many studies have already been done and several conclusions can be drawn: (i) As(V) is more effectively * Corresponding author. Fax: +33 555423680. E-mail address: [email protected] (G. Feuillade-Cathalifaud). 0021-9797/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2009.09.068

removed than As(III). An oxidation step is consequently required to better remove As(III) [10–13]. However, in the presence of natural organic matter (NOM), all the oxidative reagents cannot be used because of the formation of toxic oxidation by-products [11]. These matrix conditions need thus to adapt drinking water treatment and we wondered if higher coagulant concentrations could remove efficiently As(III) without an oxidation step. (ii) Iron coagulants are more effective than aluminum, titanium, and zirconium ones [14–16]. (iii) The process efficiency is not dependent on the effluent initial arsenic concentration [10]. (iv) Silicate, phosphate, and sulfate interfere with arsenic removal [11,17,18]. Indeed, phosphate and arsenate strongly compete for the surface sites of the sorbents and this competitiveness is influenced by 3 the nature of the sorbents, the pH, the initial ½AsO3 4 =½PO4  molar ratio, and the residence time [19,20]. However, Fe, Mn, and Ti oxi3 des are more effective in sorbing AsO3 4 than PO4 . Few authors have been interested in arsenic removal mechanisms during coagulation/flocculation and three main mechanisms have been suggested: (i) Adsorption involving the formation of surface complexes between soluble arsenic and active sites of formed hydroxides; (ii) Coprecipitation with incorporation of soluble arsenic species into a growing hydroxide phase by inclusion, occlusion, or adsorption [21]; (iii) Precipitation and formation of insoluble compounds (like FeAsO4) between As(V) and Fe(III) [8]. However, data are still not sufficient and more experiments are required for a better knowledge of As(III) and As(V) removal.

V. Pallier et al. / Journal of Colloid and Interface Science 342 (2010) 26–32

Besides, as noted previously, As(III) removal by coagulation/ flocculation in the presence of organic matter requires using specific treatment conditions. The most important parameters controlling organic matter removal by coagulation/flocculation are now well mastered [22–24] as well as the mechanisms occurring during the removal [25–28] (charge neutralization, adsorption, and entrapment). However, data concerning arsenic removal in the presence of organic matter are still insufficient and thus its influence on arsenic removal by coagulation/flocculation needs to be evaluated. The purpose of this work was thus to use zeta potential measurements to understand As(III) and As(V) removal by coagulation/flocculation of both model and natural waters with low mineral content, under specific coagulant and pH conditions and in the presence of organic matter. Preliminary conclusions about the effect of coagulant dose and coagulation pH on arsenic removal could thus be made. Adsorption experiments on preformed ferric hydroxide flocs allowed then confirmation of these first hypotheses. 2. Materials and methods All the reagents used were of the purest grade available with the lowest arsenic and iron content. The solutions were prepared in ultrapure grade water (Milli-Q system: resistivity 18.2 MX cm, TOC < 10 lg L1). 2.1. Characteristics of water samples 2.1.1. Model water The model water used was previously developed by Lenoble et al. [29] and presented a low mineral content (Table 1). It was spiked with 1.33 lmol L1 As(III) or As(V) (100 lg L1) because it is the maximum arsenic concentration acceptable before drinking water production in Europe [30]. In order to increase its mineral suspended solid concentration (i.e., turbidity), 2 mL of 10 g L1 kaolinite suspension was added. Adsorption experiments were conducted at pH 6.0 ± 0.1 in batch or under coagulation/flocculation experimental conditions (see Section 2.5) to check that neither 1.33 lmol L1 As(III) nor 1.33 lmol L1 As(V) can significantly sorb

Table 1 Chemical and physico-chemical characteristics of model and natural waters before coagulation/flocculation treatment. Model water

Natural water

6.0 ± 0.1 70 ± 1 0.6 ± 0.1 <0.2 <10 <10 9±2 60.010

5.7 ± 0.1 80 ± 2 6.5 ± 0.1 3.9 ± 0.2 130 ± 3 80 ± 2 13 ± 1 1.35 ± 0.03

SO2 4

8.1 9.9 3.4

4.3 2.8 1.8

PO3 4 Na+ + K Mg2+ Ca2+ SiO2



6.9 0.8 1.3 3.2 9.0

7.8 3.4 0.7 3.0 NDc

pH Conductivity (lS cm1) Turbidity (NTU) DOC (mg C L1) Total iron (lg L1) Ferrous iron (lg L1) Alkalinity (mg CaCO3 L1) As(V) (lmol L1) 1 a

Inorganic species (mg L ) Cl NO 3

a

Anion and cations concentrations (±5%) were measured with a Dionex DX-120 ionic chromatography, after sample filtration on 0.2-lm cellulose nitrate filters. b QL: quantification limit. c ND: not determined.

27

onto 20 mg L1 kaolinite suspension and that their speciation was maintained. In batch experiments, after a contact time of 480 min, As(III) and As(V) residual concentrations were 1.4 ± 0.1 and 1.33 ± 0.06 lmol L1, respectively. Under coagulation/flocculation experimental conditions, less than 5% of As(III) and As(V) was adsorbed onto kaolinite. Under these conditions, arsenic speciation was maintained as in the study of Goldberg [31]. However, literature studies on arsenic adsorption on clay minerals underlined significant arsenic adsorption onto kaolinite. For example, 100% of 20 lmol L1 As(V) was adsorbed onto 40 g L1 kaolinite at around pH 5 [31] and 60% of 0.4 lmol L1 As(III) was adsorbed onto 25 g L1 kaolinite between pH 8.5 and pH 9.5 [32]. At pH 6.0 ± 0.1, these authors showed that around 25% of As(III) and 100% of As(V) was adsorbed under the same experimental conditions. In these studies, the ratio [kaolinite]/[arsenic] was ca. 30,000 and 800,000 whereas in our study, it was only 200. The different results obtained could thus be explained by a very small kaolinite concentration compared to arsenic concentration. In order to study the effect of organic matter on arsenic removal, tests were realized with a Dissolved Organic Carbon (DOC) concentration of 10 mg C L1. Usual surface water matrix conditions [33] were reproduced by extracting fulvic acids from peat humic substances (humic acid sodium salt from Aldrich) according to the Schnitzer and Khan protocol [34]. These compounds are mainly encountered in natural water samples and represent more easily coagulated species than hydrophilic compounds. The influence of fulvic acids on arsenic adsorption onto kaolinite was also studied because Saada et al. [35] showed better As(V) adsorption onto kaolinite coated with humic acids. They suggested that adsorption occurred first on the humic acids sites and then, once they were saturated, on the remaining kaolinite sites. In our study, arsenic adsorption onto kaolinite was not affected by the presence of fulvic acids certainly because for the same arsenic concentrations the kaolinite and fulvic acid concentrations were respectively 1000 and 60 times lower than in the study of Saada et al. [35]. The ratio [AH-coated kaolinite]/[arsenic] applied for adsorption experiments was between 500,000 and 10,000 whereas in our study, it was only 200. 2.1.2. Natural water sample A natural surface water sample was collected in the Limousin region (France). In this region, soils are known to be granitic and water weakly mineralized. It contained only As(V) at a concentration of 1.35 ± 0.03 lmol L1. As expected this surface water from granitic origin presented a significant DOC content and low dissolved iron amounts (Table 1). Besides, the absence of phosphate ions avoided competition with arsenic for adsorption on Fe sorbents. 2.2. Physico-chemical parameters Sample pH was measured under magnetic stirring with a Cyberscan 510 pH meter from Eutech Instruments equipped with a combined Ag/AgCl/KCl 4 M glass electrode and with a platinum temperature probe. Turbidity was measured with a HI 98703 turbidimeter from HANNA Instruments. 2.3. Analytical methods for quantification 2.3.1. Arsenic As(III) and Astotal analysis were done after sample filtration on 0.45-lm cellulose acetate filters. Sample conservation (content and speciation) was achieved by adding 500 mg EDTA L1 and 8.7 mol L1 acetic acid to adjust pH between 3.3 and 3.5 [36].

28

V. Pallier et al. / Journal of Colloid and Interface Science 342 (2010) 26–32

The quantification of Astotal was realized with a Varian SpectrAA 880Z atomic absorption spectrometer equipped with a GTA 100Z graphite furnace, with Zeeman background correction. An arsenic high-intensity boosted discharge hollow-cathode lamp (UltrAALamp Varian, serial number OOHO 770) was used. Optimizations allowed achieving a quantification limit of 0.01 lmol Astotal L1 [37]. A Varian SpectrAA 220 flame atomic absorption spectrometer equipped with a Varian VGA-77 (Vapor Generation Assembly) for the hydride generation was used for As(III) measurements [38]. The quantification limit is 0.008 lmol As(III) L1. The As(V) concentration was obtained by difference between Astotal and As(III) concentrations.

(380 s1) for the coagulation step, 15 min of slow mixing (35 s1) for the flocculation step, and 30 min without mixing for the settling step. A 60-min settling step was adapted in presence of NOM according to authors [23,24]. pH influence on arsenic removal was evaluated for 6.0 6 pH 6 8.0 and [Fe3+] = 38 mg Fe3+ L1 ([Fe3+] = 680 lmol Fe3+ L1) because this coagulant dose let meet the residual Astotal concentration of 1.33 lmol L1 whatever arsenic speciation and matrix conditions. The same coagulation/flocculation/decantation experimental conditions as above were applied.

2.3.2. Iron In order to maintain speciation, iron analysis by the standardized 1,10-phenantroline spectrometric method (NF T 90-017) required a 1% (v/v) acidification of the sample with 2 mol L1 HCl. Fe(II) and Fetotal (after Fe(III) reduction) were quantified in the range 0.01–5 mg L1. Absorbances were measured with a UV1700 Shimadzu spectrophotometer.

Adsorption experiments consisted in evaluating As(III) and As(V) removal by adsorption on ferric hydroxide flocs. Ferric hydroxide flocs were formed by using the Jar Test method under the same coagulation and flocculation experimental conditions as during coagulation/flocculation experiments on the model water spiked with 20 mg L1 kaolinite. Just after the flocculation step, 1.33 lmol As(III) or As(V) L1 was added to the colloidal suspension and the slow mixing (35 s1) was maintained for 1 h to allow arsenic adsorption. After the adsorption step, the settling step was applied. FeCl3 was used to form flocs under the same concentrations as during coagulation/flocculation experiments to be able to compare arsenic removal by coagulation and adsorption.

2.3.3. DOC DOC was quantified, in the range 0.1–20 mg C L1, with a Tekmar–Dohrmann Phoenix 8000 TOC analyzer, after sample filtration on 0.45-lm cellulose acetate filters (Whatman).

2.6. Adsorption experiments

2.4. Zeta potential measurements 3. Results and discussion Zeta potential measurements were realized on colloidal suspensions with a Zetaphoremeter IV Model Z4000 from CAD Instrumentation. It measures the zeta potential of colloidal particles by determining their mobility in an electric field. The colloids to be characterized were placed in an electrophoresis chamber and a voltage (95 ± 5 mV) was applied between two electrodes to produce a uniform electric field in the connecting chamber. The way and the speed of the particle movements are directly proportional to the sign and the magnitude of their charge. The values are then translated into zeta potential by using the Helmholtz–Smoluchowski equation which links the zeta potential to the electrophoretic mobility (Zetaphorometer software). Ionic strength and pH were kept constant during the measurement and the effective length of the cell was determined with a standard 0.01 mol L1 KCl solution. Each result was an average of three readings. 2.5. Coagulation/flocculation experiments Coagulation/flocculation experiments were carried out by using the Jar Test method. The Numeric Flocculator 10409 (Fisher Bioblock Scientific) was equipped with stainless-steel paddles (7.5  2.5 cm) and allowed to work simultaneously with six 1-L tall-form cylindrical beakers. The influence of coagulant dose on As(III) and As(V) removal was studied by using FeCl3 as coagulant [16,39,40] at concentrations between 25 and 150 mg L1 (9.2 6 [Fe3+] (mg L1) 6 55.0 or 165 6 [Fe3+] (lmol L1) 6 985). pH was fixed at different values (with NaOH 1 mol L1), 6.0 ± 0.2, 6.9 ± 0.3, and 7.7 ± 0.1, to study its influence on arsenic removal under these specific coagulant conditions. Previous authors usually tested lower concentrations (1 6 [FeCl3] (mg L1) 6 26) [8,10,15,16,41–44] because their experiments were conducted on waters presenting higher turbidities and ionic strengths. The treatment by coagulation/flocculation thus required less coagulant dose than the treatment of the model and natural waters weakly mineralized chosen in this study. Appropriate contact times consist of 3 min of rapid mixing

3.1. The coagulation/flocculation process 3.1.1. Coagulant dose influence As(III) residual concentrations decreased with increasing coagulant dose and residual concentrations were higher in the presence of fulvic acids under similar coagulation pH conditions (Fig. 1). As(III) removal depends on the number of sites available on ferric hydroxide surfaces [8]. To respect the maximum total arsenic concentration of 0.13 lmol L1 after treatment, the coagulant concentration required was 18 mg Fe3+ L1 (322 lmol Fe3+ L1) without NOM; it was doubled in the presence of fulvic acids. As(V) residual concentrations were under the quantification limit of the spectrometric method for both experiments on model and natural waters except in the presence of fulvic acids in the model water matrix for the lowest coagulant concentration (Fig. 1). Under these conditions, As(V) residual concentration was 0.28 ± 0.01 lmol L1 and the coagulation step was not optimal. Indeed, settleable flocs did not form and this could be explained by a more negative zeta potential of colloidal suspension induced by the presence of organic matter (Fig. 2). In this way, the coagulant dose required for colloids destabilization and flocs formation was higher. However, such results were not observed for the treatment of natural water for which the DOC concentration ([DOC] = 3.9 ± 0.2 mg C L1) was lower than that in the model water ([DOC] = 10.0 ± 0.2 mg C L1). For the lowest treatment dose, the zeta potential of colloidal suspension was slightly lower than for the treatment of model water spiked with fulvic acids and As(V) was, however, completely removed. This result could be explained by the nature of organic compounds present in the natural water matrix. Indeed, for the model water, experiments were conducted with 100% of commercial fulvic acids extracted from peat whereas surface water samples are composed of around 80% of natural fulvic acids and 20% of humic acids which are more hydrophobic compounds. These compounds are generally easily coagulated [43] and could thus improve the coagulation step and the formation of ferric hydroxide flocs. Lower coagulant concentra-

29

V. Pallier et al. / Journal of Colloid and Interface Science 342 (2010) 26–32

As(III) pH 5.9 As(V) pH 5.8 As(III) + FA pH 6.1 As(V) + FA pH 6.0 Natural water (As(V) + DOC) pH 6.2

0.6

-1

(µmolL )

Residual arsenic concentrations

0.8

0.4 0.2 0 0

10

20

Fe

3+

30

40

50

60

-1

concentration (mgL )

Fig. 1. Evolution of residual As(III) and As(V) concentrations after coagulation/flocculation treatment of both model water spiked or not with fulvic acids (FA) and natural water.

40

Zeta potential (mV)

30 20 10 0 -10 0

10

20

30

40

60

50

-20 -30

Fe

-40

3+

-1

concentration (mgL )

FA pH 6.1 As(V) + FA pH 6.0 As(V) pH 5.8

As(III) + FA pH 6.1 As(III) pH 5.9 Natural water pH 6.2

Fig. 2. Evolution of zeta potential of colloidal suspension after coagulation/flocculation treatment of model and natural waters.

tions were thus required to form flocs and completely remove As(V). The higher residual As(V) concentration coincided with the absence of settleable flocs. Thus, sorption on flocs seems to be the main mechanisms occurring during As(V) removal. For lower coagulant doses (2 6 [Fe3+] 6 8 mg L–1 (36 6 [Fe3+] 6 143 lmol L1)), arsenate removal depended on coagulant dose as a result of floc formation and surface area [44]. Fulvic acid removal followed As(V) removal. Indeed, the highest DOC concentrations were obtained for the lowest coagulant dose and then for [Fe3+] > 9.2 mg L1 ([Fe3+] > 165 lmol L1), residual DOC concentrations were stable (Fig. 3). Organic matter seems thus

6

1 0.8 0.6

4 0.4 2

0.2

0 0

20

40

Fe

3+

60

80

Residual As(V) -1 concentrations (µmolL )

DOC DOC - As(III) DOC - As(V) DOC - Natural water As(V)

8

Residual DOC -1 concentrations (mgL )

to be removed by sorption on flocs as shown by Dennett et al. [23]. In addition, residual DOC concentrations were 30% higher after As(V) treatment than after As(III) treatment. On the contrary, As(III) treatment did not influence the organic matter removal. Natural organic matter is usually present in surface waters and probably competed with As(V) for adsorption on flocs. These results are confirmed by the treatment of natural water. Indeed, DOC concentrations after treatment of natural water were stable ([DOC] = 1.2 ± 0.1 mg C L1) whatever coagulant concentration and As(V) was completely removed. Thus, as soon as the coagulation step was effective and flocs formed, As(V) and organic matter were removed by adsorption.

0 100

-1

concentration (mgL )

Fig. 3. Residual dissolved organic carbon and As(V) concentrations after coagulation/flocculation treatment of model and natural waters.

30

V. Pallier et al. / Journal of Colloid and Interface Science 342 (2010) 26–32

Turbidity and iron removal also followed the As(V) behavior. Highest turbidity and total iron concentration were obtained for the lowest coagulant dose for which no settleable flocs were formed. For highest coagulant doses, these parameters were stable. As shown at Fig. 2, colloidal suspension destabilization did not require the complete neutralization of surface charge of colloids. Indeed, for high coagulant doses, even if the zeta potential of colloidal suspension was positive, no colloidal suspension restabilization was observed because residual turbidities were stable. During coagulation/flocculation, Fe3+ hydrolyzed and formed different cationic complexes at acidic pH. Near neutral pH, ferric hydroxide precipitates dominated whereas at more alkaline pH, anionic complexes predominated. Thus, in these pH and high coagulant dose conditions, ferric hydroxide precipitates formed because the water is supersaturated at several orders of magnitude above the solubility of the metal salts. The mechanisms occurring during turbidity removal were thus rather sweep coagulation than neutralization of surface charge of colloids. Dissolved contaminants were thus removed by adsorption to the solid precipitate and particulate contaminants were removed by enmeshment or entrapment within a mass of the solid precipitate [23]. Under these coagulant conditions, pH had little effect on As(III) removal whereas As(V) residual concentrations increased with increasing coagulation pH for [Fe3+] = 9.2 mg L1. Cathalifaud et al. [25] showed that zeta potentials of hydroxide ferric flocs decreased and became negative with increasing pH. In our study, f = 2.2 ± 0.8 and 8 ± 4 mV for pH 5.8 ± 0.1 and 6.9 ± 0.3, respectively, and As(V) exists in anionic species from pH > 2. Its removal was thus not favored when pH increased. Indeed, the pH increase led to the decrease of numbers of positively charged active sites on hydroxide ferric surfaces and to the increase in numbers of OH– ions competitive with arsenic for adsorption on hydroxide ferric surface [40]. This experiment thus confirmed sorption as the predominant As(V) removal mechanism.

hydroxide precipitates [25]; removal of organic matter by sorption was limited and hydroxide precipitates were poorly sorbing compounds. This effect of pH on NOM removal confirmed that under such conditions, organic matter was removed by sorption as in the study of Sharp et al. [46]. The decrease in organic matter removal was accompanied by an increase in As(III) removal. For the pH studied, adsorption being the mainly organic matter removal mechanism [23], there was at least competitive adsorption between As(III) and organic matter on the coagulated iron hydroxide. Besides, As(III) removal was less affected by a negative zeta potential of colloidal suspension than As(V) removal. It seemed thus to be also removed by coprecipitation and entrapment in accordance with the fact that it is present under the form of a neutral species under these pH conditions ðpKðH3 AsO3 =H2 AsO Þ ¼ 9:2Þ [49]. 3 Residual total iron concentration and turbidity were highly dependent on coagulation pH. When the zeta potential of colloidal suspension decreased, these parameters increased. Therefore, the mechanisms occurring during turbidity removal under these conditions were rather neutralization of surface charge of colloids than sweep coagulation as in the study of Jarvis et al. [47]. Besides, higher residual turbidities were obtained by varying pH than by varying coagulant dose for which sweep flocculation occurred. Thus, sweep flocculation improved particle removal when compared to charge neutralization [48]. These experiments allowed drawing preliminary hypotheses concerning As(III) and As(V) removal during coagulation/flocculation. Sorption seemed to be the As(V) removal mechanism and As(III) seemed to be removed both by adsorption and by precipitation/coprecipitation as shown by Daus et al. [50]. 3.2. Confirmation of the mechanisms occurring during arsenic removal by coagulation/flocculation According to Hering et al. [11], at high coagulant concentrations, adsorption of inorganic arsenic contaminants onto precipitated metal hydroxides is likely to be the predominant removal mechanism. Ion activity product calculation taking into account the existence of all protonated or hydroxylated complexes for either Fe(III) or As(V) was realized. Simultaneous coprecipitation of ferric hydroxide with arsenic can occur at low contaminant concentrations, for example, as amorphous ferric arsenate FeAsO4 (log Ksp = 23.0 ± 0.3 at 25 °C and 1 bar [51]). However, the coagulant doses were higher than those chosen in previous studies and the ratio [Fe3+]/[As] ranged from 92 to 550 for an initial arsenic concentration of 1.33 lmol L1. Removal of As(III) and As(V) by adsorption on precipitated iron hydroxide was thus expected as the main arsenic removal mechanisms. An evaluation of the

-20

100

-23

90 80

-26

70 -29 Zeta potential

60

DOC

-32

50

As(III) -35

DOC and As(III) removal percentage (%)

Zeta potential (mV)

3.1.2. Coagulation pH influence Whatever pH and matrix conditions, As(V) residual concentration was under the quantification limit of the GF-AAS method as settleable flocs were formed. The pH increase did not favor organic matter removal as the zeta potential of colloidal suspension decreased (Fig. 4). This decrease was more pronounced for pH > 6.8 as in the study of Kaleta and Elektorowicz [45]. According to Dennett et al. [23], for pH < 6.0, dissolved organic matter is mainly removed by precipitation with iron species whereas for pH P 6.0, adsorption reactions at the surface of ferric hydroxides and coprecipitation with ferric hydroxides predominate. Increasing pH led to the deprotonation of organic matter and to the decrease of the zeta potential of

40 6

6.2

6.4

6.6

6.8

7

7.2

7.4

7.6

7.8

8

Coagulation pH Fig. 4. Evolution of zeta potential of colloidal suspension, DOC, and As(III) removal percentage as function of coagulation pH ([Fe3+] = 38 mg L1).

31

Residual arsenic concentrations -1 (µmolL )

V. Pallier et al. / Journal of Colloid and Interface Science 342 (2010) 26–32

0.7

As(III) Adsorption on preformed flocs pH 7.2

0.6

As(V) Adsorption on preformed flocs pH 7.2

0.5

As(III) Coagulation/Flocculation pH 6.9

0.4

As(V) Coagulation/Flocculation pH 6.9

0.3 0.2 0.1 0 0

10

20

Fe

3+

30

40

50

60

-1

concentration (mgL )

Fig. 5. Comparison of As(III) and As(V) removal by coagulation/flocculation and adsorption on preformed flocs.

adsorption of As(III) and As(V) on preformed hydroxide flocs was thus realized. For 9.2 6 [Fe3+] (mg L1) 6 55.0 and pH 6.0 ± 0.2, removal of As(III) and As(V) by adsorption on preformed ferric hydroxide flocs did not allow achieving as high removal percentage as during the coagulation/flocculation process (Fig. 5). Indeed, for [Fe3+] = 9.2 mg L1, only 55% and 87% of 1.33 lmol L1 As(III) or As(V), respectively, were removed by adsorption on ferric hydroxides flocs whereas 100% As(V) and 77% As(III) were removed during coagulation/flocculation treatment. Similar results were obtained by Ghurye et al. [40] and Mercer and Tobiason [52] who observed that Fe(OH)3 formed in situ were more efficient for removing arsenic than preformed Fe(OH)3. However, in their studies, experimental conditions differed between adsorption and coagulation/ flocculation tests. As(V) added to preformed Fe hydroxide precipitates did not prevent the formation of crystallized iron (hydr)oxides as effectively as when As(V) is coprecipitated with Fe(III) [53]. According to Hering et al. [11], in situ formed precipitated ferric hydroxides present a higher surface area than the preformed ones. Moreover, when Fe(OH)3 formed in situ was used, arsenic sorption occurred as well as precipitation and coprecipitation with Fe(OH)3 in formation. In addition, more adsorption sites were certainly available when compared to preformed Fe(OH)3 for which only the external surface area of the flocs was available for arsenic sorption [40]. As previously observed during coagulation/flocculation, As(III) and As(V) were nevertheless removed but to a smaller extent when the coagulant dose was too low to allow the destabilization of the colloidal suspension and the formation of flocs. Other mechanisms than sorption must then have taken place during coagulation/flocculation such as precipitation and/or coprecipitation with iron or hydroxide precipitates. During adsorption on preformed flocs, As(III) and As(V) removal depended on the coagulant dose and consequently on the number of active sites on the hydroxide ferric surface. The more numerous the active sites on hydroxide ferric surface, the lower the residual arsenic concentrations. In this experiment, no precipitation and coprecipitation mechanisms occurred because dissolved total iron concentration was under 20 lg L1 after treatment. As(V) removal by adsorption was more efficient than the As(III) one (Fig. 5). The differences in As(V) residual concentrations between adsorption and coagulation/flocculation were low. Thus, this experiment confirmed that As(V) was removed by adsorption during coagulation/flocculation. The differences in As(III) residual concentrations between adsorption and coagulation/flocculation were more important. However, adsorption on preformed flocs allowed a decrease of the initial As(III) content from more than 56% to 96%, depending on coagulant concentrations. Adsorption was thus important in As(III) removal but precipitation/coprecipitation also

played an important role for the lowest treatment doses. Indeed, adsorption was insufficient to reduce As(III) concentrations below acceptable levels without applying high coagulant doses. 4. Conclusion This study allowed us to specify As(III) and As(V) removal during coagulation/flocculation treatment in the presence of organic matter through zeta potential measurements. Experiments were conducted on both model and natural waters with low mineral content under high coagulant doses conditions. This allowed evaluation of coagulation/flocculation efficiency for the treatment of waters contaminated both with As(III) and organic matter and presenting characteristics not suitable for a coagulation/flocculation treatment. Coagulant dose, coagulation pH, and matrix composition influenced differently the arsenic species removal. As(V) removal was independent of increasing coagulant dose in the high range of coagulant concentrations tested, at pH 5.8 and without organic matter. For a given treatment dose, its removal was dependent on the coagulation pH and of the zeta potential of the colloidal suspension. On the contrary, As(III) removal strongly depended on coagulant dose and to a smaller extent on coagulation pH. Zeta potential measurements and adsorption on preformed ferric hydroxide flocs experiments underlined that As(V) and organic matter seemed to be removed by adsorption whereas precipitation and coprecipitation seemed to be important for low coagulant dosage conditions during As(III) removal. Acknowledgments The authors thank the ‘‘Regional Council of the Limousin” for the grants assigned to the development of this project and Dr Philippe Chazal for his help in improving the English language. V.P. thanks the French Ministry of Universities for the Ph.D. allocation received. This paper was a part of the Ph.D. thesis by V.P., whose work was distinguished by the ‘‘Prix de Thèse 2009” from ASTEE (Association Scientifique et Technique pour l’Eau et l’Environnement). References [1] M.F. Hughes, Toxicol. Lett. 133 (2002) 1. [2] J.S. Petrick, F. Ayala-Fierro, W.R. Cullen, D.E. Carter, H.V. Aposhian, Toxicol. Appl. Pharmacol. 163 (2000) 203. [3] L. Vega, M. Styblo, R. Patterson, W.R. Cullen, G. Wang, D. Germolec, Toxicol. Appl. Pharmacol. 172 (2001) 225. [4] B.K. Mandal, K.T. Suzuki, Talanta 58 (2002) 201. [5] P.L. Smedley, D.G. Kinniburgh, Appl. Geochem. 17 (2002) 517. [6] M.F. Hughes, E.M. Kenyon, K.T. Kitchin, Toxicol. Appl. Pharmacol. 222 (2007) 399. [7] I. Villaescusa, J.C. Bollinger, Rev. Environ. Sci. Biotechnol. 7 (2008) 307.

32

V. Pallier et al. / Journal of Colloid and Interface Science 342 (2010) 26–32

[8] M. Edwards, J. Am. Water Works Assoc. 86 (9) (1994) 64. [9] S. Song, A. Lopez-Valdivieso, D.J. Hernandez-Campos, C. Peng, M.G. MonroyFernandez, I. Razo-Soto, Water Res. 40 (2006) 364. [10] R.C. Cheng, S. Liang, H.C. Wang, M.D. Beulher, J. Am. Water Works Assoc. 86 (9) (1994) 79. [11] J.C. Hering, P.Y. Chen, J.A. Wilkie, M. Elimelech, S. Liang, J. Am. Water Works Assoc. 88 (4) (1996) 155. [12] M. Bissen, F.H. Frimmel, Acta Hydrochim. Hydrobiol. 31 (2003) 97. [13] Y. Lee, I.H. Um, J. Yoon, Environ. Sci. Technol. 37 (2003) 5750. [14] Y.S. Shen, J. Am. Water Works Assoc. 65 (8) (1973) 543. [15] K.N. Scott, J.F. Green, H.D. Do, S.J. McLean, J. Am. Water Works Assoc. 87 (4) (1995) 114. [16] D. Lakshmanan, D. Clifford, G. Samanta, J. Am. Water Works Assoc. 100 (2) (2008) 76. [17] X. Meng, S. Bang, G.P. Korfiatis, Water Res. 34 (2000) 1255. [18] X. Meng, G.P. Korfiatis, S. Bang, K.W. Bang, Toxicol. Lett. 133 (2002) 103. [19] A. Violante, M. Pigna, Soil Sci. Soc. Am. J. 66 (2002) 1788. [20] M. Pigna, G.S.R. Krishnamurti, A. Violante, Soil Sci. Soc. Am. J. 70 (2006) 2017. [21] D.A. Lytle, T.J. Sorg, V.L. Snoeyink, J. Water Supply: Res. Technol. 54 (2005) 545. [22] G.S. Wang, S.F. Kang, H.J. Yang, S.Y. Pai, H.W. Chen, Environ. Technol. 23 (2002) 1415. [23] K.E. Dennett, A. Amirtharajah, T.F. Moran, J.P. Gould, J. Am. Water Works Assoc. 88 (4) (1996) 129. [24] K.N. Exall, G.W. vanLoon, J. Am. Water Works Assoc. 92 (11) (2000) 93. [25] G. Cathalifaud, M.T. Wais Mossa, M. Mazet, Water Sci. Technol. 27 (1993) 55. [26] A.V. Jung, V. Chanudet, J. Ghanbaja, B.S. Lartiges, J.L. Bersillon, Water Res. 39 (2005) 3849. [27] J.M. Sieliechi, B.S. Lartiges, G.J. Kayem, S. Hupont, C. Frochot, J. Thieme, J. Ghanbaja, J.B. d’Espione de la Caillerie, O. Barrès, R. Kamga, P. Levitz, L.J. Michot, Water Res. 42 (2008) 2111. [28] M. Yan, D. Wang, J. Ni, J. Qu, C.W.K. Chow, H. Liu, Water Res. 42 (2008) 3361. [29] V. Lenoble, C. Chabroullet, R. Al Shukry, B. Serpaud, V. Deluchat, J.C. Bollinger, J. Colloid Interface Sci. 280 (2004) 62.

[30] EU Council Directive 98/83/EC of 3 November 1998 Relative to the Quality of Drinking Water, Official Journal of the European Union L330, p. 32. [31] S. Goldberg, Soil Sci. Soc. Am. J. 66 (2002) 413. [32] B.A. Manning, S. Goldberg, Environ. Sci. Technol. 31 (1997) 2005. [33] B. Legube, F. Xiong, J.P. Croué, M. Doré, Rev. Sci. Eau 3 (1990) 399. [34] M. Schnitzer, S.U. Khan, Humic Substances in the Environment, Dekker, New York, 1972. [35] A. Saada, D. Breeze, C. Crouzet, S. Cornu, P. Baranger, Chemosphere 51 (2003) 757. [36] P.A. Gallagher, C.A. Schwegel, A. Parks, B.M. Gamble, L. Wymer, J.T. Creed, Environ. Sci. Technol. 38 (2004) 2919. [37] J. Michon, V. Deluchat, R. Al Shukry, C. Dagot, J.C. Bollinger, Talanta 71 (2007) 479. [38] V. Pallier, B. Serpaud, G. Feuillade-Cathalifaud, J.C. Bollinger, Int. J. Environ. Anal. Chem, in press, doi:10.1080/03067310902977534. [39] J.D. Chwirka, C. Colvin, J.D. Gomez, P.A. Mueller, J. Am. Water Works Assoc. 96 (3) (2004) 106. [40] G. Ghurye, D. Clifford, A. Tripp, J. Am. Water Works Assoc. 96 (4) (2004) 143. [41] B. Han, T. Runnells, J. Zimbron, R. Wickramasinghe, Desalination 145 (2002) 293. [42] S.R. Wickramasinghe, B. Han, J. Zimbron, Z. Shen, M.N. Karim, Desalination 169 (2004) 231. [43] E. Lefebvre, B. Legube, Water Res. 24 (1990) 591. [44] M.B. Baskan, A. Pala, J. Hazard. Mater. 166 (2009) 796. [45] J. Kaleta, M. Elektorowicz, Environ. Technol. 30 (2009) 119. [46] E.L. Sharp, S.A. Parsons, B. Jefferson, Sci. Total Environ. 363 (2006) 183. [47] P. Jarvis, B. Jefferson, S.A. Parsons, Water Res. 40 (2006) 2727. [48] J. Duan, J. Gregory, Adv. Colloid Interface Sci. 100–102 (2003) 475. [49] B.J. Lafferty, R.H. Loeppert, Environ. Sci. Technol. 39 (2005) 2120. [50] B. Daus, H. Weiss, R. Wennrich, Talanta 46 (1998) 867. [51] D. Langmuir, J. Mahoney, J. Rowson, Geochim. Cosmochim. Acta 70 (2006) 2942. [52] K.L. Mercer, J.E. Tobiason, Environ. Sci. Technol. 42 (2008) 3797. [53] A. Violante, S. Del Gaudio, M. Pigna, M. Ricciardella, D. Banerjee, Environ. Sci. Technol. 41 (2007) 8275.