Fenton-like oxidation of 2,4,6-trinitrotoluene using

2 downloads 0 Views 395KB Size Report
a US EPA priority compound and has been recognized as a mutagenic ..... ophenols, while neutral compounds such as TNT are not affected (Kwan and Voelker, ...
Science of the Total Environment 385 (2007) 242 – 251 www.elsevier.com/locate/scitotenv

Fenton-like oxidation of 2,4,6-trinitrotoluene using different iron minerals Roger Matta a,b , Khalil Hanna b,⁎, Serge Chiron a b

a Laboratoire Chimie et Environnement FRE2704 3, Place Victor Hugo 13331 Marseille cedex 3, France Laboratoire de Chimie Physique et Microbiologie pour l'Environnement, LCPME, UMR 7564 CNRS-Université Henri Poincaré Nancy 1, 405, rue de Vandoeuvre, 54600 Villers-les-Nancy, France

Received 14 February 2007; received in revised form 7 June 2007; accepted 18 June 2007 Available online 26 July 2007

Abstract Degradation of 2,4,6-trinitrotoluene (TNT) was investigated in presence of different oxidants (Fenton's reagent, sodium persulfate, peroxymonosulfate and potassium permanganate) and different iron minerals (ferrihydrite, hematite, goethite, lepidocrocite, magnetite and pyrite) either in aqueous solution or in soil slurry systems. Fenton's reagent was the only oxidant able to degrade TNT in solution (kapp = 0.0348 min− 1). When using iron oxide as heterogeneous catalyst at pH 3, specific reaction rate constants per surface area were ksurf = 1.47.10− 3 L min− 1 m− 2 and ksurf = 0.177 L min− 1 m− 2 for magnetite and pyrite, respectively while ferric iron minerals were inefficient for TNT degradation. The major asset of iron mineral catalyzed Fenton-like treatment has been the complete oxidation of the pollutant avoiding the accumulation of possible toxic by-products. In soil slurry systems, 38% abatement of the initial TNT concentration (2 g/kg) was reached after 24 h treatment time at neutral pH. Rate limiting steps were the availability of soluble iron at neutral pH together with desorption of the TNT fraction sorbed on the clay mineral surfaces. © 2007 Elsevier B.V. All rights reserved. Keywords: Fenton-like oxidation; 2,4,6-trinitrotoluene; Iron minerals; Water

1. Introduction 2,4,6-trinitrotoluene (TNT) has been the most widely used nitroaromatic explosive and TNT concentrations have been reported to range from 10 to 12,000 mg/kg at contaminated sites (Rodgers and Bunce, 2001). Most of these sites contain also contaminated groundwater. TNT is a US EPA priority compound and has been recognized as a mutagenic chemical (Lachance et al., 1999). Consequently, the health advisories for TNT have been issued to less than 0.002 mg/L by the US EPA (US EPA, 2002). In response to the need for clean up of contaminated soils and waters, a ⁎ Corresponding author. Tel.: +33 3 83 68 52 42; fax: +33 3 83 27 54 44. E-mail address: [email protected] (K. Hanna). 0048-9697/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2007.06.030

variety of physico-chemical remediation techniques has been implemented. Alkaline hydrolysis and zero valent iron treatments have been successfully applied for TNT degradation in water and soil treatment (Bandstra et al., 2005; Emmrich, 1999). However, these two technologies proceed through reductive pathways leading to the formation of an uncharacterized polymeric material, mainly in soil, upon prolonged TNT treatment (Thorn et al., 2004). The toxicities and susceptibilities to microbial and chemical degradation of the polymeric materials remain unknown and this issue is still a matter of concern for the broad application of these technologies. Fenton's reaction has also been used to treat TNT contaminated aqueous solution often resulting in complete degradation and partial mineralization with accumulation of short chain carboxylic

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251

acids (e.g. oxalic acid) at the end of the reaction time (Chen et al., 2005; Yardin and Chiron, 2006). Analogous data on Fenton-like reactions using iron minerals as heterogeneous catalysts is scarcely available (Hess and Scharder, 2002), although the efficiency of such processes has been clearly demonstrated for others organic pollutants (Huang et al., 2001; Kwan and Voelker, 2003). Heterogeneous catalysis offers significant advantages. Unlike Fenton's reagent, the reaction of iron oxides with H2O2 can effectively catalyse the oxidation of organic contaminants at circumneutral pH. Use of iron oxides instead of dissolved iron may be especially advantageous for in situ remediation of contaminated groundwater where pH cannot be adjusted. The catalyst can be easily recovered by sedimentation or filtration for further uses. Finally, results obtained in heterogeneous Fenton's reaction may be exploited for in situ chemical oxidation (ISCO) treatment where endogenous iron oxides have been already used for the elimination of few recalcitrant chemical families mainly, chlorinated ethenes (Kim and Gurol, 2005), PAHs (Flotron et al., 2005), pesticides (Mecozzi et al., 2006), and hydrocarbons (Kang and Hua, 2005). In this paper, the catalytic activity of six iron minerals for TNT oxidation was evaluated in aqueous solution. These iron minerals were selected because of their widespread abundance in natural soils and sediments, where they might be exploited for in situ degradation of TNT. The main purpose of this work was to investigate the feasibility of Fenton-like chemical oxidation processes for TNT remediation in water and soil slurry. The scope includes the following steps: 1. The assessment of different oxidants in aqueous solution. 2. The assessment of different iron oxides in aqueous solution. 3. The assessment of the efficiency of modified Fenton's reagent in soil slurry systems. 2. Experimental 2.1. Chemicals 2,4,6-trinitrotoluene (N 98%) was from the chemical stock of the University of Provence. Ferrous sulfate heptahydrate (FeSO4.7H2O) N 97%, oxalic acid, formic acid, and salicylic acid were purchased from Fluka (Buchs, Switzerland). Oxone® (2KHSO5-KHSO4-K2SO4,N 99%), and sodium persulfate (Na2S2O8, N 99%) from SigmaAldrich (St Quentin-Fallavier, France). Potassium permanganate (KMnO4, 95%) from Prolabo (Fontenay-sous-Bois, France), stabilized hydrogen peroxide (30% w/w) with NaH2PO4.7H2O from Fischer Scientific (Illkirch, France).

243

All solvents were gradient grade. Solutions and HPLC eluents were prepared using 18 mΩ cm− 1 deionized water from a Millipore system (Bedford, USA). 2.2. Synthesis and characteristics of iron oxides Iron (oxyhydr)oxides (two-lines ferrihydrite, hematite, goethite, lepidocrocite), were synthesized while pyrite (FeS2, 99.9%) was purchased from Alfa Aesar (Karlsruhe, Germany), and magnetite (Fe3O4, 99.9%) from Sigma (St Quentin-Fallavier, France). 2-line ferrihydrite sample was synthesized according to the method of Cornell and Shwertmann (1996). Briefly, ferrihydrite (Fe2(OH)6) was prepared by neutralizing a 0.2 M ferric chloride solution with 1 M NaOH to a pH of 7–8. The precipitate was freeze dried and stored as a solid. The goethite sample was prepared by neutralizing 500 mL of a 0.5 M ferric nitrate solution (Fe(NO3)3. 9H2O) with 400 mL of 2.5 M sodium hydroxide solution, as described by Villalobos and Leckie (2000). Sodium hydroxide was added quickly in a nitrogen atmosphere in a glove box and the solution was stirred vigorously. The precipitate obtained was aged in an oven for 24 h at 60 °C. Lepidocrocite (γ-FeOOH) was synthesized by oxidation of the (0.228 M FeCl2·4H2O + 0.4 M NaOH) aqueous mixture in the presence of an excess of dissolved oxygen at neutral pH. The precipitate was collected and air-dried. Hematite (∝Fe2O3) was synthesized according to a previously published procedure (Cornell and Shwertmann, 1996). Briefly, ferric chloride was placed in 0.002 M HCl and incubated at 98 °C in a closed vessel for 7 d. After centrifugation, the solid was washed with distilled water and then dried. Iron oxides were characterized using X-ray powder. XRD data was collected with a Philips PW1710 diffractometer using Co K radiation (35 kV, 30 mA). The spectra of XRD were found to be identical to the expected ones. Mössbauer spectroscopy was used to confirm the nature of all precipitates. The specific surface areas of the iron oxides were determined by multipoint N2-BET analysis using a Coulter (SA113) surface area analyzer and are reported in Table 1. 2.3. Soil characteristics Soil samples were collected from an uncontaminated shallow subsurface agricultural field. The soil was air dried at 40 °C for a month and was ground and sieved at 2 mm. To measure pH with a glass pH electrode, 5.0 g of soil was mixed with 5.0 mL reagent water. Soil carbon content was quantified with a C/N analyzer (NA 1500NC, Fisons instrument). To determine the exchangeable Fe (Fe bound to carbonate, Fe bound to oxide and Fe bound to

244

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251

Table 1 Some characteristics of the studied iron minerals Iron oxidation state

BET area (m2/g)

α-FeOOH Fe2 (OH)6

III III

50 327

γ-FeOOH Fe2O3 Fe3O4 FeS2

III III II, III II

59 11 2 0.8

Iron minerals

Formula

Goethite Two-line Ferrihydrite Lepidocrocite Hematite Magnetite Pyrite

organic matter fractions), a sequential extraction procedure was implemented (Tessier et al., 1979). The major characteristics of the soil are reported in Table 2. The iron fraction was also identified by Mössbauer spectroscopy which showed that the iron content of this soil was mainly present as oxide forms (hematite, magnetite and maghemite) or sorbed to the clay fraction. Analysis of this latter fraction revealed that the content of Fe(III) and Fe(II) was 80% and 20%, respectively. Mercury chloride (HgCl2) was added at a concentration level of 400 mg/kg in order to inhibit biological activities. TNT was loaded to the soil by spiking TNT dissolved in analytical grade methanol, and then the solvent was air dried at ambient temperature. Spiking level was 2 g/kg. 2.4. Preparation of reagents and reactions 0.11 mM TNT was allowed to dissolve in distilled water over night in an ultrasonic bath. In the case of Fenton's reagent, persulfate, and peroxymonosulfate experiments, stock solutions of iron catalyst (Fe(II)) were prepared with great caution. A pre-determined mass of iron was dissolved in degassed reagent grade water while slowly stirring (∼ 100 rpm) the solution and immediately the pH was adjusted to 3 using dilute HClO4. All iron catalyst solutions were prepared immediately prior to the initiation of the reactions in order to avoid flocculation and precipitation of iron salts or iron colloids via reactions with dissolved oxygen. Fe (II) and H2O2 were added at 1.55 mM and 80 mM, respectively at pH 3. No attempt was achieved to optimize the Fe(II)/H2O2 ratio to improve the Fenton's reaction efficiency. KMnO4 solution (80 mM) was added to the well-stirred TNT solution at pH 3. Different persulfate or peroxymonosulfate/Fe(II)/TNT ratios (30/ 10/1, 30/20/10, 60/10/1, 60/20/1, and 400/20/1) were tested at pH 3. To verify the efficiency of mineralcatalyzed Fenton-like oxidation, investigations were conducted in batch reactors under continuous stirring and under aerobic conditions. Each reactor was prepared by adding 1.76 g/L iron mineral and 80 mM H2O2 to the

contaminated water. pH was adjusted with concentrated HClO4. Sorption experiments of TNT on iron minerals were conducted in the same conditions but without adding H2O2. Soil slurries were prepared by placing 20 g of soil in a borosilicate Erlen Meyer with 100 mL of distilled water. The resulting slurry was allowed to equilibrate for 24 h to allow desorption of the pollutant. Two Fe(II)/H2O2 molar ratios were used: 1/160 and 1/ 320 ([H2O2] = 0.8 M and 1.6 M). pH was not adjusted (pH = 6.7). Sorption isotherms were developed by equilibrating 3 g of uncontaminated soil with 20 mL of phosphate buffer to give TNT concentrations of 1, 10, 25, and 50 mg/L. Sorption reached equilibrium in 24 h. Soil slurries were then centrifuged and filtered, before analysis. During all the oxidation reactions, 10 mL aliquots were withdrawn at selected time intervals for analysis. The reaction was quenched by adding 1 mL of a 1 M sodium sulfite solution. All experimental runs were performed within a temperature of 20–25 °C in the absence of light. Each experiment was achieved in triplicate, all results were expressed as a mean value of the 3 experiments. Blank experiments were conducted in absence of H2O2. 2.5. Chemical analysis Nitrate ions were determined by ion chromatography (detection limit 10− 6 M), HCO3−/CO32− as inorganic carbon with a Shimadzu TOC-5050 Total Organic Carbon analyser, non-purgeable organic carbon with the same instrument upon sample acidification with HClO4 and 20 min purge with zero-grade air to eliminate CO2, total iron by inductively coupled plasma/atomic emission spectrophotometry (detection limit 0.02 μM, JY 2000 Ultrace, Jobin Yvon). Dissolved ferrous iron concentrations were measured according to the 1, 10-phenanthroline method (American Public Table 2 Some characteristics of the studied soil pH Sand (%) Silt (%) Clay (%) CEC (meq/100) Organic matter (%) Density (g/cm3) Porosity (%) Total Fe (mg/kg) Fe Easy exchangeable (mg/kg) Fe bound to carbonates (mg/kg) Fe bound to oxides (mg/kg) Fe bound to organic fraction (mg/kg)

6.7 54.4 23 22.6 23.5 6.5 2.22 47.5 1150 6.8 400 190.4 535.8

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251

Health Association et al., 1992), using an Agilent 8543 spectrophotometer at λ = 510 nm. TNT decay was followed by reversed phase liquid chromatography. The HPLC consisted of a LC Shimadzu pump 10AT (Touzard & Matignon, France) equipped with a Varian UV detector selected at λ = 254 nm and fitted with a LiChrosphere RP-18 column 250 mm × 4.6 mm i.d., 5 μm particle size (Merck). The system was operated in an isocratic mode (methanol/water; 60/40, v/v) at a flow rate of 1 mL/min. The retention time of TNT in these conditions was 9.4 min (± 30 s). Short chain carboxylic acids were identified and quantified by ion exclusion chromatography using a LC-10AT Shimadzu chromatograph fitted with a 300 × 6.5 mm i.d. ICSep-ORH-801 column (Transgenomic) in conjunction with a 9050 Varian UV detector selected at λ = 204 nm. The mobile phase consisted of water containing 290 μg/L sulfuric acid at a flow rate of 0.6 mL/min. TNT by-product identification was carried out by HPLC-MS using an electrospray interface in negative ionization mode. The HPLC system consisted of a Metachem C-18 column 150 × 2 mm i.d., 3 μm particle size (Varian) and an Esquire 6000 ion trap mass spectrometer (Bruker, Germany). The mobile phase used in chromatographic separation consisted of a binary mixture of solvent A (methanol) and B (0.1% formic acid solution) at a flow rate of 0.2 mL/min. The gradient was operated from 5 to 100% A for 30 min and then back to initial conditions in 5 min. For TNT analysis in soil slurries, samples were centrifuged at 5000 rpm during 20 min. Solid fraction was dried at 40 °C and extracted by means of a pressurized fluid extraction (PFE) procedure while aqueous fractions were injected via a syringe filter (0.45 μm) into a 20 μL injection loop. TNT concentration in soil slurries was expressed as the sum of TNT in solid and in aqueous phases. The instrument used for PFE was the accelerated solvent extractor 300 (ASE 300, Dionex). The solvent was 100% methanol. A pressure of 1500 psi, a temperature of 150 °C, a static time of 10 min, two static cycles, a flush volume of 200 s were selected as recommended (Marple and LaCourse, 2005). In these conditions, percent recovery for TNT was found to be 75 (± 8) %. 3. Results and discussion 3.1. Assessment of different oxidants in aqueous solution Four different oxidants, permanganate, persulfate, peroxymonosulfate (Oxone®) and hydrogen peroxide (Fenton's reaction) were evaluated for TNT abatement in aqueous solution at pH 3. Kinetic results are reported

245

Fig. 1. Evolution of 2,4,6 trinitrotoluene (C0 = 0.11 mM) concentration under four different oxidants at pH 3.0. (♦) Fenton's reagent H2O2:Fe2+ = 51:1; (×) Permanganate [MnO−4 ] = 80 mM; (•) Peroxymonosulfate 2+ [PMS]:[Fe2+] = 48:1, (▵) Persulfate [S2O2− 8 ]:[Fe ] = 60:20.

in Fig. 1. Fenton's reagent was the only oxidant able to degrade TNT in solution. TNT degradation was completed in 120 min and a pseudo first order rate constant of kapp = 0.0348 min− 1 was established. This result is consistent to previous published data (Liou et al., 2003) where a kapp = 0.0588 min− 1 was reported. Under our experimental conditions, the persulfate (S2O82−) and the peroxymonosulfate (HSO5−) anions have been chemically activated by Fe(II) to generate the intermediate sulfate radical (SO4.− ) (Anipsitakis and Dionysiou, 2004). Under acidic conditions the sulfate radical prevails in solution while at neutral and basic pH, sulfate radical undergoes a radical interconversion reaction to produce hydroxyl radicals (Liang et al., 2006). Although comparable standard reduction potential values are reported for hydroxyl and sulfate radicals at acidic pH (E° ~ 2.7 V), sulfate free radical reaction with organic pollutants mainly involves an electron transfer mechanism while, with hydroxyl radical, hydrogen abstraction and addition reaction on aromatic ring are predominant mechanisms (Minisci and Citterio, 1983). This difference in reactivity between both radicals might explain the absence of TNT degradation upon persulfate and peroxymonosulfate oxidation conditions. In presence of three nitro electron withdrawing groups, addition of a sulfate radical to the TNT aromatic ring was seriously precluded. Lack of degradation was observed whatever the persulfate or peroxymonosulfate/Fe(II)/TNT ratio used (see experimental part). Similarly, TNT degradation upon permanganate oxidation was poorly expected. The generally accepted mechanism of oxidation of alkenes with MnO4− is an electrophilic attack by MnO4− on the carbon–carbon double bond, with formation of a cyclic hypomanganate diester as a reaction intermediate. Consequently, in presence of nitro electron withdrawing groups, this mechanism is definitively not favored and accordingly, Waldemer and Tratnyek

246

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251

Fig. 2. (♦) Total organic carbon, (■) oxalic acid, and (▴) nitrate concentration evolution during homogeneous Fenton's reaction of a 0.11 mM TNT aqueous solution. [H2O2] = 80 mM, [Fe2+] = 1.55 mM, pH = 3.

(2006) reported a very low second-order rate constant k = 1.6.10− 1 M− 1 s− 1 for TNT degradation in presence of permanganate. Fig. 2 reports the TOC, nitrate ions and oxalic acid concentration evolution trends against the TNT treatment time under Fenton's reaction. After 48 h treatment time, 58% abatement of the initial TOC value was reached and oxalic acid accounted for the remaining TOC. Consequently, none possible toxic aromatic by-product was accumulated after 48 h. Nitrogen mass balance could not be properly achieved, possibly due to N2 losses during experiments. 3.2. Assessment of mineral catalyzed Fenton-like oxidation in aqueous solution Mineral-catalyzed reactions were first carried out at pH 3 using H2O2 and different iron minerals (goethite, hematite, pyrite, magnetite, ferrihydrite and lepidocrocite), as an alternative to soluble iron-catalyzed reaction, to oxidize TNT. TNT adsorption tests were carried out and showed that this process was of a very minor significance (less than 5%) for all the studied iron minerals. Oxidation kinetic data are reported in Fig. 3. At pH 3, the catalytic activities of iron minerals decreased in the order of pyrite N magnetite N ferrihydrite ∼ lepidocrocite ∼ hematite ∼ goethite. Fenton-like reaction is a surface controlled reaction that depends on H2O2 concentration, on the iron mineral surface area and on solution chemistry (pH, ionic strength). Kwan and Voelker (2003) have suggested the following chain of reactions in mineral catalyzed oxidation system: ≡FeðIIIÞ þ H2 O2 →≡FeðHO2 Þ2þ þ Hþ

ð1Þ

≡FeðHO2 Þ2þ →≡FeðIIÞ þ HO2

ð2Þ

≡FeðIIÞ þ H2 O2 →≡FeðIIIÞ þ OH− þ d OH

ð3Þ

d OH þ H2 O2 →H2 O þ HOd2

ð4Þ

≡FeðIIÞ þ Od− 2 →≡FeðIIIÞ þ O2

ð5Þ

≡FeðIIIÞ þ HO2d →≡FeðIIÞ þ HO−2

ð6Þ

≡FeðIIÞ þ HO−2 →≡FeðIIIÞ þ HOd2

ð7Þ

In mineral catalyzed reaction, the dominant reaction is first a chain of reactions occurring on the mineral surface. Secondly and at pH below 4, a propagation reaction by dissolved iron ions may operate in solution due to the proton promoted dissolution of Fe from the oxide surface. If only Fe(III) is originally present, Fe(II) is slowly generated by reaction (1) and (2) initiating oxidation reactions. In our experimental conditions (0.08 M H2O2) ferric oxides (hematite, goethite, lepidocrocite, ferrihydrite) were found to be inactive for TNT degradation (see Fig. 3). Iron concentration in reaction solutions were measured below limit of detection (0.02 μM) which endorsed the assumption of the very limited solubility of the synthesized ferric iron oxides at pH 3. In contrast, previous findings stated degradation of chlorinated pollutants (e.g. chlorophenols) by Fenton-like reaction using goethite as catalyst at acidic pH (Kwan and Voelker, 2004). This discrepancy probably results from electrostatic effects that increase the adsorption and the oxidation rates of charged dissolved organic compounds such as chlorophenols, while neutral compounds such as TNT are not affected (Kwan and Voelker, 2004). Pyrite and magnetite, two Fe(II) bearing iron minerals were more effective catalysts than ferric oxides for TNT transformation underlying the higher catalytic activity of Fe(II) over Fe (III) in Fenton-like processes (see Fig. 3). 85% TNT

Fig. 3. TNT (C0 =0.11 mM) decay as function of treatment time in presence of different iron minerals; [H2O2]=80 mM, [iron mineral]= 1.76 g/L. At pH 3: (♦) goethite; (□) hematite; (▴) pyrite; (×) Magnetite; (+) ferrihydrite; (⋄) lepidocrocite, and at pH 6.8: (■) magnetite; (▵) pyrite.

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251

elimination was observed after 6 h magnetite-catalyzed reaction while total elimination of TNT was observed after 30 min treatment time in presence of pyrite. Specific reaction rate constants per surface area were ksurf = 1.56.10− 3 L min− 1 m− 2 (kapp = 0.0055 min− 1) and ksurf = 0.180 L min− 1 m− 2 (kapp = 0.2535 min− 1) for magnetite and pyrite, respectively. Before and after (5 min) addition of H2O2, the measured Fe(II) concentration were 0.16 mM/0.24 mM and 0.92 mM/1.38 mM in the magnetite and pyrite system. Therefore, iron oxidation state and Fe(II) dissolution rate appeared to be the key parameters for TNT removal in mineralcatalysed reaction. The better efficiency of pyrite catalyst (vs. magnetite) was due to the high ability of pyrite to dissolve into water in presence of H2O2, O2 and Fe(III) species. The autocatalytic dissolution and oxidation of pyrite has been described in the following reactions (Arienzo, 1999): 2FeS2 þ 15H2 O2 →2FeðIIIÞ þ 2Hþ þ 4SO2− 4 þ 14H2 O ð8Þ þ 2FeS2 þ 7O2 þ 2H2 O→2FeðIIÞ þ 4SO2− 4 þ 4H

ð9Þ

FeS2 þ 14FeðIIIÞ þ 8H2 O→15FeðIIÞ þ 16Hþ þ 2SO2− 4 ð10Þ In a first step, H2O2 reacts with pyrite to produce Fe (III) (Eq. (8)). In a second step, pyrite is oxidized by Fe

247

(III) to produce Fe(II) (Eq. (10)) and to initiate the Fenton's reaction since at acidic pH, pyrite oxidation by Fe(III) is faster than by molecular oxygen. Protons generated by Eq. (10) dissolve more Fe(III) from the oxidized surface and accelerate the reaction. TNT oxidation could be attributed to a heterogeneous catalysis occurring at the iron sulfide surface together with a homogeneous catalysis in the aqueous phase. In presence of pyrite and magnetite, elimination efficiency had a decline trend with the increase in pH value (see Fig. 3). At pH 6.8, specific reaction rate constants per surface area were ksurf = 5.68.10− 5 L min− 1 m− 2 and ksurf = 2.13.10− 4 L min− 1 m− 2 for magnetite and pyrite, respectively. This decrease in the constant rate values (vs. pH 3) was related to a decrease in iron dissolution rate in both cases. Since no TOC abatement was recorded during the pyrite mineral-catalyzed reactions (data not shown), byproduct identification was required. In order to elucidate the different pathways leading to the degradation of TNT in presence of pyrite, samples were analysed by LC/MS. LC/MS in negative electrospray ionization mode revealed four major transformation products (see Fig. 4). Their molecular weight, chromatographic retention times and their main ions obtained by MS detection are reported in Table 3. By product molecular weight was assigned on the basis of the pseudo molecular ion [M–H] − . The by-product structures were tentatively elucidated according to their mass

Fig. 4. LC/MS total ion chromatogram (TIC) under negative electrospray ionization mode obtained after 2 h pyrite-catalyzed reaction. [TNT]0 = 0.11 mM, pH = 3.

248

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251

Table 3 Molecular weight (MW), main ions and relative abundances in LC (−) ESP-MS of TNT by-products detected under pyrite-catalysed Fentonlike oxidation Compound MW RT (min)

Main ions in MS (relative abundance %)

I

488

7.2

II III IV TNT

275 259 243 227

17.8 19.2 23.4 24.8

212 (100); 425 (60); 230 (15); 244 (10); 487 (5) 211 (100); 241 (80); 274 (50); 258 (40) 241 (100); 211 (70); 258 (20) 226 (100); 242 (30) 226 (100); 244 (5)

By-product number as in Fig. 4.

fragmentation pattern. The MS spectrum of IV (MW 243) revealed losses of 16 mass units (m/z 226) possibly related to O losses and IV was assigned the structure of an epoxide. III (MW 259) and II (MW 275) with shorter chromatographic retention time than II might correspond to more polar compounds originating from further hydroxylation of IV. The MS spectrum of I (MW 488) was characterized by a base peak at m/z 212 accounted for the unmodified 2,4,6-trinitrobenzene structure. Signal at m/z 230 represented its deprotonated ion water adduct. The MS spectrum of I showed also losses of 63 mass units (m/z 425) possibly due to the simultaneous losses of nitro and hydroxy moieties. It could be also assumed from the fragment ion at m/z 244 (227 + 17) that I might derive from a hydroxylated TNT derivative and I was tentatively identified as a dimer of 1-hydroxy-2,4,6-TNT. Mineralization of TNT could be hypothesized as follows (Fig. 5). Oxidative degradation of TNT involves a hydroxyl radical attack on the C1 of TNT. This position is favoured possibly due to the inductive effect of the methyl moiety. The resulting

radical which is stabilized by resonance might evolve in two different ways. The first possibility would be a dimerization. The second possibility would be the formation of an epoxide followed by further hydroxylation of the aromatic ring prompting ring opening. In order to obtain better insights into the identity of the end products, short chain carboxylic acids were analysed by ion exclusion chromatography (results not reported; for further information see Yardin and Chiron, 2006). Two major peaks corresponding to oxalic acid and formic acid were identified by comparison of their retention time with those of authentic standards. Note that the detected aromatic intermediates totally disappeared after 24 h treatment time. Note also that none by-product originating from a reductive pathway (hydroxylamine and amine) was detected during the treatment time. 3.3. Assessment of Fenton-like reaction in soil slurry systems. The efficiency of Fenton-like reaction was also evaluated in soil slurry systems at circumneutral pH. The pH value was not optimized to 3 in the soil slurry system mainly because lowering pH to 3 is impracticable in real situations due the high buffering ability of most natural soils and because lowering pH to 3 may prompt heavy metal mobility towards groundwater. Sorption isotherms were developed and the experimental isotherm data was fitted to the Freundlich equation (q = KCn, where q is the amount of TNT sorbed in μg per g of soil, C is the equilibrium concentration of TNT (mg/L) in solution phase, K is the Freundlich sorption constant and n is the heterogeneity factor). The adsorption constant K was established to be 10.6 L/kg

Fig. 5. Proposed reaction scheme for the oxidation of TNT under Fenton-like processes.

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251

(R2 = 0.94). This value was near to that previously reported for a top soil with high organic matter content (8%, Sheremeta et al., 1999). It is well known that TNT is extensively and specifically adsorbed by negatively charged 2:1 clay mineral surfaces saturated with low hydrated cations (Haderlein and Schwarzenbach, 1993). Organic matter via non-specific hydrophobic interactions seems to play a less significant role in TNT sorption processes providing that nitro groups are not reduced into amine group (Sheremeta et al., 1999). Moreover, amorphous iron oxides have been shown to be capable of reducing TNT into amine derivatives. Although the studied soils contained iron oxides, it has been assumed that such reduction reactions probably did not occur during the time frame of our experiments since these experiments were carried out under sterile and aerobic conditions. TNT dissolution/desorption kinetics were investigated using distilled water (see Fig. 6). 65% TNT release was observed after only 10 min, and apparent equilibrium was reached after 150 min. At this time, 80% of the initial concentration of TNT was detected in the aqueous phase. It could be hypothesized that this fraction might correspond to the fraction weakly associated to the organic matter fraction of the soil and the fraction precipitated in the soil. The overall kinetic data obtained during the Fenton-like reaction experiments are depicted in Fig. 6. Note that the TNT spiking level was 2 g/kg which corresponds to a value encountered in heavily contaminated sites. Mineralcatalyzed reaction (with no addition of ferrous iron) contributed to 12% abatement of the initial TNT concentration. The mineral catalyzed oxidation mechanism was possibly limited by the low iron (II) content in soil (20% of the total iron sorbed to clay fraction). When

Fig. 6. TNT concentration evolution profiles against treatment time in soil slurry systems. (■) Fenton's reagent: H2O2: Fe2+ = 160:1; (▵) Fenton's reagent: [H2O2] = 0.8 M with sequential addition of Fe2+ (5 mM) at t = 0, t = 4 h and t = 8 h (arrows); (♦) [H2O2] = 0.8 M without adding Fe2+; (×) TNT dissolution/desorption profile in distilled water. m and m0 stand for TNT weight in soil slurries at t and t0, respectively. [TNT]0 = 2 g/kg of soil.

249

Fe(II) was added in the system, TNT degradation occurred with a very fast initial rate (t b 1 h) followed by a much slower degradation rate. After 1 h treatment time, 28% abatement of the initial concentration of TNT was achieved and 38% after 24 h. The slow down in reaction rate could not be ascribed to the lack of aqueous phase TNT. A much more sound reason should be that under highly oxidizing conditions resulting from an excess in H2O2 concentration, Fe(II) was quickly converted into Fe(III) due to its oxidation by H2O2 and Fe(III) precipitated at pH 6.7, prompting a high decrease in the Fenton's reaction efficiency. Consequently, we hypothesized that the TNT elimination rate was limited by the availability of soluble Fe. Experiments aiming at adding Fe(II) at different times during the time course of the reaction were performed (see Fig. 6). Amendment with two fold 5 mM Fe(II) was needed to reach 80% TNT removal. Further addition of Fe(II) and an increase in H2O2 concentration up to 5% did not result in an increase of TNT abatement. The absence of total TNT removal might be also related to the occurrence of non-pollutant species which might constitute a significant hydroxyl radical sink. Relevant hydroxyl radical scavengers in the aqueous phase of the soil slurry system were inorganic anions such chloride, sulfate and bicarbonate ions and dissolved organic matter (DOM). Although a rate constant is not available for DOM reaction with hydroxyl radical, an average value of (5.0 ± 1.9) × 104 L (mg of C)− 1 s− 1 has been recently reported (Vione et al., 2006). Bicarbonate ions which is the main carbonate specie at pH 6.7 has a well established second order rate constant with hydroxyl radical of 1 × 107 M− 1 s− 1 (Buxton et al., 1988). The scavenging of hydroxyl radical by chloride is operational in acidic solution only and not at neutral to basic pH (Chiron et al., 2006). The possible effect of sulfate ions originating from the selected iron salt (FeSO4) was much more difficult to evaluate. It is known that sulfate ions form complexes with Fe(II) and Fe(III), decreasing the formation of Fe-peroxo complexes and inhibiting the decomposition rate of H2O2. However, such a complexation reaction has appeared to be no longer effective at neutral pH (Park and Choi, 2003). In addition, Kwan and Voelker (2003) did not find any relation between the presence of sulfate ions and the efficiency decrease of Fenton-like systems. The effect of sulfate ions was therefore assumed to be insignificant. Accordingly, only DOM and bicarbonate ion contents were determined in the aqueous phase of the soil slurry system and were found to be 10 mg C/L and 4 mg/L on average. In order to assess the significance of the hydroxyl radical scavenging effect, the absolute rate constant kabs (TNT)

250

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251

of the reaction of TNT with hydroxyl radical had to be determined. For this purpose, competitive kinetic experiments were performed in the presence of equal concentrations of TNT and salicyclic acid (SA), a compound with a well known absolute constant (kabs(SA) = 2.2 × 1010 M− 1 s− 1). The TNT and SA concentrations were monitored during the Fenton's reaction and Ln(C0/C) was plotted against time (results not reported, for more experimental details see Hanna et al., 2005). Assuming a pseudo first order kinetic for both SA and TNT reactions with hydroxyl radical, apparent rate constants are given by the slope of the corresponding straight lines. Kapp(SA) and kapp(TNT) were found to be 0.0078 and 0.0052 min− 1, respectively. The absolute rate constant was calculated as follows: kabsðTNTÞ ¼ kabsðSAÞ ðkappðTNTÞ= kappðSAÞ Þ ¼ 1:5 1010 M1 s1

Since this latter constant value is far above (1000 folds) those reported for the reaction of hydroxyl radical with bicarbonate ions and DOM, mass removal of TNT in the soil slurry system should remain unaffected by the occurrence of these scavengers. Consequently, the observed slow down in the TNT degradation under Fenton-like reaction could be attributed to the time required for further desorption of TNT when all the aqueous TNT has been eliminated in addition to the unavailability of soluble iron. Note that the 20% TNT fraction left in the soil slurry system after subsequent additions of Fe(II) are compatible with the 20% fraction of TNT strongly adsorbed on the soil. Note also that the aromatic by-products identified in part 3.2 were not detected by MS detector in soil slurry systems after 24 h treatment time. 4. Conclusions Fenton-like chemical oxidation processes using iron minerals and H2O2 have turned out to be a promising treatment option for TNT contaminated water remediation. Oxidation state of the iron catalyst and the iron dissolution rate were found to be the key parameters for effective performance of Fenton-like reactions. Accordingly, Fe(II) bearing minerals (pyrite and magnetite) were more effective than ferric oxides (hematite, goethite, lepidocrocite and ferrihydrite) for TNT transformation at acidic and neutral pH. One of the major assets of heterogeneous catalysis is the complete oxidation of the pollutant avoiding the accumulation of possible toxic by-products in treated waters. In soil slurry systems, 38% abatement of the initial TNT concentration (2 g/kg) was achieved after 24 h treatment

time at neutral pH. Rate limiting steps for total TNT removal involved the availability of soluble iron at neutral pH together with desorption of the TNT fraction strongly sorbed on the clay mineral surfaces. It can be reasonably hypothesized that in presence of chelating agents able to both complex TNT and iron in solution such as cyclodextrins, Fenton-like oxidation might be effectively used for improving TNT remediation in contaminated soils. Acknowledgements M. Abdelmoula is thanked for his help in Mössbauer analysis. We thank the anonymous reviewers for their valuable comments. References American Public Health Association (APHA), American Water Works Association (AWWA), Water Pollution Control Federation (WPCF). In: Greenberg AE, Clesceri LS, Eaton AD, editors. Standard methods for the examination of water and wastewater, vol. 18. Washington DC: APHA; 1992. Anipsitakis GP, Dionysiou DD. Radical generation by the interaction of transition metals with common oxidants. Environ Sci Technol 2004;38:3705–12. Arienzo M. Oxidizing 2,4,6-trinitrotoluene with pyrite H2O2 suspensions. Chemosphere 1999;39:1629–38. Bandstra JZ, Miehr R, Johnson R, Tratnyek P. Reduction of 2,4,6trinitrotoluene by iron metal: kinetic controls on product distributions in batch experiments. Environ Sci Technol 2005;39:230–8. Buxton GV, Greenstock CL, Helman WP, Ross AB. Critical review of rate constant for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (·OH/·O−) in aqueous solution. J Phys Chem Ref Data 1988;17:513–883. Chen WS, Juan CN, Wei KM. Mineralization of dinitrotoluenes and trinitrotoluene of spent acid in toluene nitration process by Fenton oxidation. Chemosphere 2005;60:1072–9. Chiron S, Minero C, Vione D. Photodegradation processes of the antiepileptic drug carbamazepine, relevant to estuarine water. Environ Sci Technol 2006;40:5977–83. Cornell RM, Shwertmann U. The iron oxides: structure, properties, reactions, occurrence and uses. Weinheim: VCH; 1996. Emmrich M. Kinetics of the alkaline hydrolysis of 2,4,6-trinitrotoluene in aqueous solution and highly contaminated soils. Environ Sci Technol 1999;33:3802–5. Flotron V, Delteil C, Padellec Y, Camel V. Removal of sorbed polycyclic aromatic hydrocarbons from soil, sludge and sediment samples using the Fenton's reagent process. Chemosphere 2005;59:1427–37. Haderlein SB, Schwarzenbach RP. Adsorption of substituted nitrobenzenes and nitrophenols to mineral surfaces. Environ Sci Technol 1993;39:1811–8. Hanna K, Chiron S, Oturan M. Coupling enhanced water solubilization with cyclodextrin to indirect electrochemical treatment for pentachlorophenol contaminated soil remediation. Water Res 2005;39:2763–73. Hess TF, Scharder PS. Coupled abiotic–biotic mineralization of 2,4,6trinitrotoluene (TNT). J Environ Qual 2002;3:736–44.

R. Matta et al. / Science of the Total Environment 385 (2007) 242–251 Huang HH, Lu MC, Chen JN. Catalytic decomposition of hydrogen peroxide and 2-chlorophenol with iron oxides. Water Res 2001;35:2291–9. Kang N, Hua I. Enhanced chemical oxidation of aromatic hydrocarbons in soil systems. Chemosphere 2005;61:909–22. Kim K, Gurol MD. Reaction of nonaqueous phase TCE with permanganate. Environ Sci Technol 2005;39:9303–8. Kwan WP, Voelker BM. Rates of hydroxyl radical generation and organic compound oxidation in mineral-catalyzed Fenton-like systems. Environ Sci Technol 2003;37:1150–8. Kwan WP, Voelker BM. Influence of electrostatics on the oxidation rates of organic compounds in heterogeneous Fenton systems. Environ Sci Technol 2004;38:3425–81. Lachance B, Robidoux PY, Hawari J, Ampleman G, Thiboutot S, Sunahara GI. Cytotoxic and genotoxic effects of energetic compounds on bacterial and mammalian cells in vitro. Mutat Res 1999;444:25–39. Liang C, Wang Z, Mohanty N. Influences of carbonate and chloride ions on persulfate oxidation of trichloroethylene at 20 °C. Sci Total Environ 2006;370:271–7. Liou MJ, Lu MC, Chen JN. Oxidation of explosives by Fenton and photo-Fenton processes. Water Res 2003;37:3172–9. Marple RL, LaCourse WR. Application of photoassisted electrochemical detection to explosive containing environmental samples. Anal Chem 2005;77:6709–14. Mecozzi R, Di Palma L, Merli C. Experimental in situ chemical peroxidation of atrazine in contaminated soil. Chemosphere 2006;62:1481–9. Minisci F, Citterio A. Electron transfer processes: peroxydisulfate, a useful and versatile reagent in organic chemistry. Acc Chem Res 1983;16:27–32.

251

Park H, Choi W. Visible light and Fe(III) mediated degradation of acid orange 7 in the absence of H2O2. J Photochem Photobiol A Chem 2003;159:241–7. Rodgers J, Bunce N. Treatment methods for the removal of nitroaromatic explosives. Water Res 2001;35:2101–11. Sheremeta TW, Thiboulot S, Ampleman G, Paquet L, Halasz A, Hawari J. Fate of 2,4,6-trinitrotoluene and its metabolites in natural and model soil systems. Environ Sci Technol 1999;33:4002–8. Tessier A, Campbell PG, Bisson M. Sequential extraction procedure for the speciation of particulate trace metals. Anal Chem 1979;51:844–51. Thorn KA, Thorn PG, Cox L. Alkaline hydrolysis/polymerization of 2,4,6-trinitrotoluene: characterization of products by 13C and 15N NMR. Environ Sci Technol 2004;38:2224–31. U.S. EPA, drinking water standards and health advisories. EPA 822-B00-001, USA, 2002. Villalobos M, Leckie JO. Carbonate adsorption on goethite under closed and open CO2 conditions. Geochim Cosmochim Acta 2000;4:3787–802. Vione D, Falletti G, Maurino W, Minero C, Pelizzetti E, Malandrino M, et al. Sources and sinks of hydroxyl radicals upon irradiation of natural water samples. Environ Sci Technol 2006;40:3775–81. Waldemer R, Tratnyek P. Kinetics of contaminant degradation by permanganate. Environ Sci Technol 2006;40:1055–61. Yardin G, Chiron S. Photo-Fenton treatment of TNT contaminated soil extract solutions obtained by soil flushing with cyclodextrin. Chemosphere 2006;62:1395–402.