interaction of 2-chloronaphthalene with high carbon iron filings (hcif): adsorption, dehalogenation...

10
Journal of Colloid and Interface Science 314 (2007) 552–561 www.elsevier.com/locate/jcis Interaction of 2-chloronaphthalene with high carbon iron filings (HCIF): Adsorption, dehalogenation and mass transfer limitations Alok Sinha, Purnendu Bose Environmental Engineering and Management Programme, Department of Civil Engineering, Indian Institute of Technology Kanpur, Kanpur 208016, India Received 13 January 2007; accepted 17 May 2007 Available online 23 May 2007 Abstract Interaction of 2-chloronaphthalene (2-CN) with high-carbon iron filings (HCIF) was studied in anaerobic batch systems, both under well- mixed and poorly-mixed conditions. In well-mixed conditions, partitioning of 2-CN between solid and aqueous phases was fast, resulting in rapid attainment of equilibrium. Equilibrium partitioning could be described by a Freundlich isotherm, C s = K.[C a ] m , where C s (μmoles g 1 iron) and C a (μmoles L 1 ) were the solid and aqueous phase 2-CN concentrations, respectively. Isotherm parameters, m and K were determined to be 0.76 and 5.6 × 10 2 (μmole g 1 iron)/(μmole L 1 ), respectively. Sorption (k 2 ) and desorption (k 3 ) rate constants were determined to be 5.60 × 10 1 h 1 g 1 iron L and 10 h 1 , respectively. Reductive dehalogenation of aqueous phase 2-CN occurred concurrently but at a slower rate, and could be described by the expression (dC T /dt) =−k 1 .M.(C a ) N , where C T (μmoles L 1 ) was the total 2-CN concentration and M (g iron L 1 ) the concentration of HCIF. The values of k 1 and N were determined to be 1.09 × 10 2 h 1 g 1 ironL and 1.647, respectively. In poorly mixed conditions, adsorption (k 2 ) and desorption (k 3 ) rate constants were 3.92 × 10 5 h 1 g 1 iron L and 7 × 10 4 h 1 , respectively, i.e., several orders of magnitude less than in well-mixed systems. The dehalogenation rate parameters, k 1 and N were determined to be 2.22 × 10 4 h 1 g 1 iron L and 0.986, respectively, suggesting slower dehalogenation. These results highlight how mass-transfer limitations during the interaction between HCIF and 2-CN in poorly mixed systems, such as permeable reactive barriers (PRBs), can potentially impact the dehalogenation process. © 2007 Elsevier Inc. All rights reserved. Keywords: 2-Chloronaphthalene; Iron; Reductive dehalogenation; Mixing; Mass transfer limitations 1. Introduction Many halogenated organic compounds (HOCs), i.e., chlori- nated solvents, organo-chlorine pesticides and herbicides, etc. are toxic to humans and other flora and fauna [1]. However, such compounds are widely used in industrial applications and as pesticides and herbicides in household and agricultural ap- plications. Uncontrolled release of such compounds to the en- vironment causes widespread air, water, and soil pollution, and ingestion by human being and other flora and fauna. Of spe- cial concern is the pollution of groundwater resources by chlo- rinated organic compounds. Traditional remediation method for groundwater contaminated with halogenated organic com- pounds involves pumping the contaminated groundwater to the surface and passing it through a treatment process train. Dur- * Corresponding author. Fax: +91 512 2597395. E-mail address: [email protected] (P. Bose). ing such treatment, the contaminant is either degraded, as in the case of advanced oxidation systems [2], or transferred to an- other medium, as in case of air stripping and granular activated carbon adsorption. The decontaminated water is then returned to the subsurface through surface recharge [3]. Of late, issues of long term effectiveness and economic feasibility have gradu- ally shifted remediation trends away from such ‘pump and treat’ systems and towards in situ methods. The main advantage of in situ treatment is that it allows ground water to be treated with- out being brought to the surface, resulting in significant cost savings [4]. Dehalogenation of dilute aqueous solutions of halogenated organic compounds (HOCs) by metallic or zero-valent iron (ZVI) filings was first reported by Sweeny [5,6] and Senzaki and Kumangai [7]. Since then, use of ZVI filings for dehalo- genation of HOCs in water has been a subject of considerable interest. Numerous studies have shown that a wide variety of HOCs undergo reductive dehalogenation through interaction 0021-9797/$ – see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2007.05.045

Upload: alok-sinha

Post on 26-Jun-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Journal of Colloid and Interface Science 314 (2007) 552–561www.elsevier.com/locate/jcis

Interaction of 2-chloronaphthalene with high carbon iron filings (HCIF):Adsorption, dehalogenation and mass transfer limitations

Alok Sinha, Purnendu Bose ∗

Environmental Engineering and Management Programme, Department of Civil Engineering, Indian Institute of Technology Kanpur, Kanpur 208016, India

Received 13 January 2007; accepted 17 May 2007

Available online 23 May 2007

Abstract

Interaction of 2-chloronaphthalene (2-CN) with high-carbon iron filings (HCIF) was studied in anaerobic batch systems, both under well-mixed and poorly-mixed conditions. In well-mixed conditions, partitioning of 2-CN between solid and aqueous phases was fast, resulting in rapidattainment of equilibrium. Equilibrium partitioning could be described by a Freundlich isotherm, Cs = K.[Ca]m, where Cs (µmoles g−1 iron)and Ca (µmoles L−1) were the solid and aqueous phase 2-CN concentrations, respectively. Isotherm parameters, m and K were determined to be0.76 and 5.6 × 10−2 (µmole g−1 iron)/(µmole L−1), respectively. Sorption (k2) and desorption (k3) rate constants were determined to be 5.60 ×10−1 h−1 g−1 iron L and 10 h−1, respectively. Reductive dehalogenation of aqueous phase 2-CN occurred concurrently but at a slower rate, andcould be described by the expression (dCT/dt) = −k1.M.(Ca)

N , where CT (µmoles L−1) was the total 2-CN concentration and M (g iron L−1)the concentration of HCIF. The values of k1 and N were determined to be 1.09 × 10−2 h−1 g−1 iron L and 1.647, respectively. In poorly mixedconditions, adsorption (k2) and desorption (k3) rate constants were 3.92×10−5 h−1 g−1 iron L and 7×10−4 h−1, respectively, i.e., several ordersof magnitude less than in well-mixed systems. The dehalogenation rate parameters, k1 and N were determined to be 2.22 × 10−4 h−1 g−1 iron Land 0.986, respectively, suggesting slower dehalogenation. These results highlight how mass-transfer limitations during the interaction betweenHCIF and 2-CN in poorly mixed systems, such as permeable reactive barriers (PRBs), can potentially impact the dehalogenation process.© 2007 Elsevier Inc. All rights reserved.

Keywords: 2-Chloronaphthalene; Iron; Reductive dehalogenation; Mixing; Mass transfer limitations

1. Introduction

Many halogenated organic compounds (HOCs), i.e., chlori-nated solvents, organo-chlorine pesticides and herbicides, etc.are toxic to humans and other flora and fauna [1]. However,such compounds are widely used in industrial applications andas pesticides and herbicides in household and agricultural ap-plications. Uncontrolled release of such compounds to the en-vironment causes widespread air, water, and soil pollution, andingestion by human being and other flora and fauna. Of spe-cial concern is the pollution of groundwater resources by chlo-rinated organic compounds. Traditional remediation methodfor groundwater contaminated with halogenated organic com-pounds involves pumping the contaminated groundwater to thesurface and passing it through a treatment process train. Dur-

* Corresponding author. Fax: +91 512 2597395.E-mail address: [email protected] (P. Bose).

0021-9797/$ – see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.jcis.2007.05.045

ing such treatment, the contaminant is either degraded, as in thecase of advanced oxidation systems [2], or transferred to an-other medium, as in case of air stripping and granular activatedcarbon adsorption. The decontaminated water is then returnedto the subsurface through surface recharge [3]. Of late, issuesof long term effectiveness and economic feasibility have gradu-ally shifted remediation trends away from such ‘pump and treat’systems and towards in situ methods. The main advantage of insitu treatment is that it allows ground water to be treated with-out being brought to the surface, resulting in significant costsavings [4].

Dehalogenation of dilute aqueous solutions of halogenatedorganic compounds (HOCs) by metallic or zero-valent iron(ZVI) filings was first reported by Sweeny [5,6] and Senzakiand Kumangai [7]. Since then, use of ZVI filings for dehalo-genation of HOCs in water has been a subject of considerableinterest. Numerous studies have shown that a wide variety ofHOCs undergo reductive dehalogenation through interaction

A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561 553

with ZVI filings [8–19]. Interest in ZVI-mediated dehalogena-tion process is primarily due its potential application in perme-able reactive barriers (PRBs) for in situ remediation of ground-water contaminated with HOCs [8]. Pilot and full scale PRBshave been constructed in a number of locations [20,21] with en-couraging results.

Gillham and O’Hannesin [8] suggested that during ZVI-mediated dehalogenation in well-mixed batch systems, the rateof dehalogenation was controlled by the ratio between surfacearea of iron filings and volume of solution, with rate of dehalo-genation being diffusion-controlled, i.e., mass transfer limitedwhen this ratio was low. Results of other batch experiments in-dicated that even in well-mixed systems, the rate of diffusionof HOCs to the ZVI surface may, in some cases, control therate of dehalogenation [22]. Deng et al. [23] reported that inwell-mixed ZVI–HOC systems, reactions having activation en-ergies of 20 kJ/mol or less were diffusion-controlled. In thiscontext, studies indicated that ZVI-mediated dehalogenation oftrichloroethylene (TCE) (activation energy 15–18 kJ/mol) isdiffusion controlled [24], while dehalogenation of vinyl chlo-ride (activation energy of 42 kJ/mol) is controlled by the rate ofreactions on the iron surface [23]. In general, higher activationenergies are reported for saturated chlorinated hydrocarbonsand for this class of compounds, overall dehalogenation rateis likely controlled by reaction rather than by mass transfer lim-itations [24,25]. However, Matheson and Tratnyek [26] showedthat the pseudo-first order dehalogenation rate constant (kobs)for carbon tetrachloride was unaffected by temperature over therange of 4–35 ◦C. Since the reaction controlled rates are tem-perature dependent, it was concluded that the dehalogenationin this case was diffusion controlled.

In some ZVI–HOC systems, HOC dehalogenation rate maybe controlled by reaction at the ZVI surface under well-mixedconditions, but might become mass transfer limited underpoorly mixed conditions. In this context, several researchers[26–28] have reported that during interaction of several HOCswith ZVI, the kobs value increased with the degree of mixing,which suggested that mass transfer is an important contribu-tor to the dehalogenation rate in such cases. Scherer et al. [24]noted that in well-mixed ZVI–HOC systems, dehalogenationrates vary widely depending on the HOC being dehalogenated.However, in systems with mass transfer limitations due to lim-ited mixing, the dehalogenation rates are likely to be slowerand less variable, considering the relatively narrow range indiffusivities of HOCs. In experiments involving the applica-tion of aqueous solutions of HOCs to columns containing ZVI,pseudo-first-order dehalogenation rate of HOCs was indepen-dent of the flow velocity through the column, suggesting thatrate of degradation was not mass transfer limited [8,28,29]. Insimilar experiments involving other HOCs, dehalogenation ratewas observed to increase with increasing flow rates, indicat-ing mass transfer limitations [25,30]. In this context, Rajagopaland Burris [25] have reported that at low initial ethylene di-bromide (EDB) concentration, kobs increased when flow rateincreased while at higher initial EDB concentrations kobs val-ues were lower and less dependent on flow rates. Consideringthe diversity of views, it is clear that discussion is still ongoing

regarding the extent to which mass transfer limitations impactthe overall rate of dehalogenation of HOCs.

The impact of mass transfer limitations on the rate and ex-tent of dehalogenation of 2-chloronaphthalene (2-CN) by high-carbon iron filings (HCIF) was evaluated in the present study.HCIF was chosen considering its attractiveness as the reactivematerial in PRBs due to its low cost [22]. 2-CN has been identi-fied as a priority pollutant by U.S. Environmental ProtectionAgency [31]. However dehalogenation of 2-CN by HCIF iscomplicated by the fact that 2-CN being a hydrophobic com-pound (LogKow = 3.9, Kow is the octanol–water partition co-efficient), may substantially partition to the graphite inclusionspresent on HCIF surface while simultaneously undergoing re-ductive dehalogenation through interaction with metallic iron[32]. Matheson and Tratnyek [26] reported that vigorous mix-ing can decrease the mass transfer limitations and hence en-hance the dehalogenation rates. However, in systems wheresorption occurs, mixing may increase the extent of sorption,thereby decreasing the aqueous concentration, thus affectingthe extent of dehalogenation adversely [8,22,28]. Objective ofthis study was to evaluate and model the interaction of 2-CNwith HCIF in well-mixed as well as poorly-mixed batch sys-tems, so that the effect of mass transfer limitation on dehalo-genation and adsorption rates in this system is elucidated.

2. Materials and methods

2.1. Materials

Commercially available high carbon iron (purchased in Kan-pur, India) was chipped on a lathe machine and then ground intoiron filings in a ball mill. The fraction of HCIF passing through425 µm (40 mesh) sieve and retained on 212 µm (80 mesh)sieve was used in this study. Carbon content of HCIF was de-termined using the Strohlein Apparatus (Adair Dutt & Co. Pvt.Ltd., India) to be 2.72 percent by weight. HCIF was digestedin aqua-regia and analyzed by Atomic Absorption Spectrome-ter (VARIAN, Spectra AA, 220FS, Australia) for determinationof metal content. The filings contained 93% iron, 0.05% cop-per and 0.3% manganese by weight. Silica content of the HCIFwas not measured. Before use, HCIF was washed in nitrogen-sparged 1 N HCl with periodic shaking for 30 min, then rinsed10–12 times with N2-sparged deionized (Milli Q) water, anddried for 2 h at 100 ◦C under nitrogen atmosphere. This treat-ment yielded black metallic filings with no visible rust on thesurface. Surface area of the treated filings was determined byBET (N2) analysis using a BET surface area analyzer (CoulterSA 3100, USA) to be 3.418 m2 g−1.

X-ray diffraction (XRD) analyses were carried out for sur-face characterization of HCIF. An X-Ray Powder Diffractome-ter (Model ISO-Debyeflex 2002, Rich Seifert and Co., Ger-many) using Cu-Kα radiation (with λ = 1.541841 Å) was em-ployed for this purpose. The scan rate used was 3◦ min−1. Thediffraction patterns were analyzed using DIFFRACplus (Release2001 Eva version 7.0) software (Bruker Advanced X-Ray Solu-tions, Germany) with the aid of JCPDF database available withthe software DIFFRACplus (Release 2001 PDF Maint version

554 A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561

7.0, Bruker Advanced X-Ray Solutions, Germany). Transmis-sion Mössbauer Spectroscopy was also used for surface charac-terization of HCIF. The gamma-ray source was 57Co embeddedin Rh matrix. The initial source activity was ∼25 mCi. The linewidth of the outer line of Mössbauer spectrum of a pure ironabsorber with this source was 0.22 mm/s. Argon-filled pro-portional counter obtained from Wissenschaftliche EleKtroniKGmbH was used as detector. This detector was characterizedby its typical resolution of 10% for 14.41 keV energy. The biasvoltage (+1900 V) was supplied by a high voltage power sup-ply. A Canberra model 2006 preamplifier was used with theproportional counter for this purpose. Drive control unit was aprecision oscillator/amplifier, procured from WissenschaftlicheEleKtroniK GmbH. Mössbauer data acquisition and analysiswas done using a multi-channel scaler (MCS-pci) procuredfrom ORTEC, U.S. Sample was prepared by spreading it evenlybetween two layers of a transparent tape fixed on a 0.5 mm thickcopper ring of 1.2 cm diameter. Mössbauer spectra were plottedas % transmission vs velocity of the source (mm/s). The spectrawere analyzed using least square method assuming Lorentzianline shapes.

Other chemicals used were 2-chloronaphthalene (2-CN)99.9% (5000 µg mL−1 in methanol) (Supelco, USA), 1,4-dichlorobenzene (1,4-DCB) 99+% (Sigma Aldrich, USA),n-hexane (Merck HPLC Grade), acetonitrile (Merck HPLCGrade), pyrite (Ward’s Natural Science Est., Inc., USA), HCl(AR Grade, Thomas Baker) and HNO3 (AR Grade, ThomasBaker). All chemicals except pyrite were used as is. The pyritewas ground into a fine powder before use. 1 g of ground pyritewas digested in aqua-regia and analyzed by Atomic AbsorptionSpectrometer for iron content. This analysis revealed the purityof pyrite to be 94%.

2.2. Experimental procedure

Time series experiments were carried to determine the rateand extent of 2-CN dehalogenation. All experiments were car-ried out in 16 mL glass vials with screw caps equipped withteflon lined rubber septa (Wheaton Science, USA). Either 2 or5 g of HCIF was added to a vial in a typical experiment. In allcases, 0.1 g of pyrite was added to each vial, with the exactweight of iron and pyrite added being determined gravimet-rically. Pyrite addition was required for pH control. Aqueoussolution of 2-CN was prepared by adding the required vol-ume of the stock 2-CN solution in methanol (5 mg mL−1) tonitrogen-sparged deionized (MilliQ) water. This solution wasthen transferred to a 1-L separatory funnel and was purged withN2 to maintain anoxic conditions. Vials containing iron filingsand pyrite were filled from the funnel such that no headspaceexisted, and then sealed using the vial screw caps. Aqueous vol-umes in these vials were determined gravimetrically. Average2-CN mass added to each vial was between 25 and 30 µg. Con-trol vials, containing 2-CN but no HCIF or pyrite, were also pre-pared. In experiments involving mixing, the vials were placedon a roller drum which rotated at 15 rpm such that the vial axisremained horizontal at all times. In experiments involving nomixing, vials contents were placed on a platform without dis-

turbance such that the vial axis remains vertical. Temperaturewas 25 ± 1 ◦C during all experiments. Vials were removed induplicate (along with a control) at specified times for samplingand analysis.

Both aqueous and solid-phase 2-CN concentration was mea-sured in each vial. To measure aqueous 2-CN concentration,1 mL aliquots of aqueous phase were sampled from each sealedvial using a micro syringe pierced through the septa, whilepurging nitrogen into the vial using another needle piercedthrough the septa. Each 1 mL aliquot was added to a GC autosampler vial (Wheation Science, USA) containing 0.5 mL n-hexane as solvent and 1,4-dichlorobenzene (1 mg/L) as internalstandard, and sealed. This mixture was then thoroughly mixedon a vortex mixer for 10 min to ensure partitioning of 2-CNto the solvent phase and the solvent analyzed by gas chro-matography (GC). Hexane phase was concentrated, if required,by purging N2. For determining solid phase 2-CN concentra-tion, aqueous content of the 16 mL vial was transferred, as faras practicable, to a pre-weighed and sealed 60 mL vial by airdisplacement using a cannula. The 60 mL vial was weighedafter the transfer for determination of the weight and hencevolume of the transferred aqueous phase. A 3 mL aliquot ofn-hexane containing internal standard was added to the original16 mL vial by injection through the septum. Vial contents werevortex mixed for 5 min to ensure transfer of the solid phase2-CN to the solvent. The solvent was transferred to a sealed10 mL vial. Another 3 mL aliquot of n-hexane with internalstandard was added to the original vial and the contents vor-tex mixed for 5 min to transfer any residual 2-CN. This extractwas transferred to same 10 mL sealed vial. The combined ex-tract was analyzed by GC for determining 2-CN concentration.Using this value, along with the aqueous phase concentrationmeasured earlier, and through application of associated cor-rections to account for, (1) 2-CN mass in the aqueous phaseremoved from the original 16 mL vial to measure the aqueousphase 2-CN concentration, and (2) 2-CN mass in aqueous phasewhich could not be separated from the solid phase, mass of 2-CN adsorbed on HCIF could be calculated. Comparison of thetotal added mass of 2-CN with the sum of measured aqueousand solid-phase mass of 2-CN in a vial containing HCIF, 2 hafter 2-CN addition showed 99% recovery, thus validating theanalytical procedure. Sample pH and oxidation–reduction po-tential (ORP) in each vial was measured both at the start andafter the completion of the experiment.

2.3. Analytical procedures

A Perkin Elmer Clarus 500 gas chromatograph equippedwith an Elite-5 column (30 m × 0.32 mm × 0.25 µm) was usedfor measurement of 2-CN concentration. Carrier gas was ni-trogen. Injector and detector (ECD) temperature were 250 and350 ◦C, respectively. The oven temperature program was as fol-lows: Starting at 80 ◦C ramp at 10 ◦C min−1 to 120 ◦C ramp at20 ◦C min−1 to 220 ◦C and hold for 5 min. Samples (1 µL) wereinjected in split-less mode. Calibration standards in the 100–5000 pg range were prepared. Detection limit for 2-CN was25 pg µL−1, and the corresponding signal/noise (S/N) ratio was

A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561 555

Fig. 1. X-ray diffraction (XRD) and Mössbauer spectra of un-rusted HCIF surface. (A) XRD spectra. (B) Mössbauer spectra.

greater than 4. Sample pH and redox potential were measuredin situ in all vials by piercing the septa, using pH and ORP mi-croelectrodes (Microelectrodes Inc., USA), respectively.

3. Results and discussion

X-ray diffraction (XRD) spectra of HCIF surface (Fig. 1A)showed prominent peaks consistent with metallic iron (JC-PDF:06-696). No peaks attributable to any iron oxide phases wereobserved. Mössbauer Spectroscopy of HCIF (Fig. 1B) indicatedthe presence of pure iron, triolite (FeS) and cementite (Fe3C).Based on both XRD and Mössbauer spectra presented in Fig. 1it was concluded that HCIF surface was not coated with ironoxides.

As depicted schematically in Fig. 2, interaction of 2-CNwith un-rusted HCIF was involved two simultaneous processes[32], (1) adsorption/desorption of 2-CN to graphite inclusionspresent on HCIF surface, and (2) reductive dehalogenation ofresidual aqueous 2-CN through interaction with metallic iron.Thus, at any time,

(1)

CT (µmoles L−1) = [Cs (µmoles g−1 iron).M (g iron L−1)

]+ Ca (µmoles L−1)

where Ca, Cs and CT are the aqueous, sorbed and total 2-CNconcentrations in the vial, and M is the concentration of HCIF.The total 2-CN concentration in the vials, CT, decreased with

time in all experiments, due to dehalogenation through interac-tion with HCIF. In all vials, the pH remained near 7 throughthe experimental duration. Redox potentials were highly reduc-ing (−600 to −700 mV) in all vials through the experimentalduration.

3.1. Model development

Assuming decline in CT to be due to reductive dehalogena-tion, i.e.,

N.Ca + M → Dehalogenation By-Products.

Rate of change of total 2-CN concentration,

(2)dCT

dt= −k1.M.(Ca)

N .

Where, k1 and N are rate constant and reaction order respec-tively of the dehalogenation process. Equation (2) can be lin-earized as below,

(2a)Ln

[−dCT

dt

]= Ln[M.k1] + N.Ln[Ca].

Further, it was assumed [29] that partitioning of 2-CN to carbonpresent in cast iron is nonspecific in nature, i.e., the numberof adsorption sites on the carbon was constrained only by thenumber of 2-CN molecules that could be fitted on the carbon

556 A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561

Fig. 2. Schematic of the interaction of 2-CN with HCIF surface (adapted from Sinha and Bose [32]).

surface. At low surface coverage, such partitioning can be rep-resented by the general equation,

m.Ca (µmoles L−1) + M (g iron L−1)

k2−→k3←−

cs (µmole L−1),

where cs = M.Cs.

Where m is Freundlich exponent and k1 and k2 are sorptionand desorption rate constants. Rate of change in aqueous 2-CNconcentration, dCa

dt, is represented as,

dCa

dt= (Rate of desorption of 2-CN from the solid phase

to aqueous phase)

− (Rate of adsorption of 2-CN from the aqueous to

the solid phase)

− (Rate of dehalogenation of 2-CN in aqueous phase)

or,

(3)d[Ca]

dt= k3.M.[Cs] − k2.M.[Ca]m − k1.M.(Ca)

N .

3.2. Well-mixed case

Assuming that in well-mixed systems, the rate of adsorp-tion/desorption of 2-CN to graphite inclusions on HCIF surfaceis fast in comparison to the dehalogenation rate, adsorptionequilibrium was assumed to be maintained at all times greaterthan 2 h after 2-CN addition to HCIF (see Sinha and Bose [32]for further discussion on this subject). The corresponding equi-librium constant describing 2-CN partitioning is K = k2

k3. Under

such conditions, partitioning of 2-CN between solid and aque-ous phases can be represented by a Freundlich isotherm,

(4)Cs = K.[Ca]m.

A plot of Ca vs Cs of 2-CN in well-mixed systems is presentedin Fig. 3. The experimental data could be adequately repre-sented by the Freundlich isotherm at all times greater than 2 h,with m = 0.76 and K = 0.056 (µmole g−1 iron)/(µmole L−1).

Fig. 3. Freundlich isotherm describing equilibrium partitioning of 2-CN be-tween solid and aqueous phases during contact with HCIF in well-mixed batchreactors (M = 128 g L−1).

Experimental data on the decline in total 2-CN concentra-tion with time in well-mixed systems is presented in Fig. 4A.HCIF concentration (M) in these experiments was 128 g L−1.A plot of Ln[− dCT

dt] versus Ln[Ca] for this data is presented in

Fig. 4B. A linear fit to the data (see Eq. (2a)) yielded values ofk1 = 1.09 × 10−2 h−1 g−1 iron L and N = 1.647. Experimen-tal data on the decline in total 2-CN concentration in systemswhere M = 323 g L−1 is also presented in Fig. 4B. A plot ofEq. (2a) with k1 = 1.09 × 10−2 h−1 g−1 iron L, N = 1.647 andM = 323 g L−1 was observed to agree with experimental datacorresponding to M = 323 g L−1 quite well.

Equations (1), (2) and (3) were solved simultaneously toobtain the variation of Ca, CT and Cs with time. Followinginitial conditions were used for this purpose, at time, t = 0,Ca = CT = [CT]0 = 11.00 µmole L−1, where [CT]0 is the ini-tially added 2-CN concentration. M was taken as 128 g L−1.Values of k1, m and N required for the simulation were as de-termined earlier. The values of adsorption and desorption rateconstants, k2 and k3 respectively, were unknown, but K = k2

k3was known. Numerical solution of the above equations for vari-ous assumed values of k3 were compared with the experimental

A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561 557

Fig. 4. Reductive dehalogenation of 2-CN by HCIF in well-mixed batch reactors. (A) Decline in total 2-CN concentration. (B) Linearized plot of the rate of 2-CNdehalogenation.

data. The simulation corresponding to k2 = 5.6 × 10−1 h−1 g−1

iron L and k3 = 10 h−1 was observed to fit the experimentaldata adequately. Experimental data and corresponding modelsimulations for CT, Ca and Cs are shown in Figs. 5A, 5Band 5C, respectively. The initial data points in Fig. 5C (upto t = 40 min) have also been shown in the inset diagram inFig. 5C, such that the agreement between the model and thedata is more apparent.

3.3. Poorly-mixed case

The decline in total concentration of 2-CN with time dur-ing 2-CN interaction with HCIF under poorly mixed conditionsis presented in Fig. 6A. HCIF concentration (M) in these ex-periments was 128 g L−1. A plot of Ln[− dCT

dt] versus Ln[Ca]

for this data is presented in Fig. 6B. A linear fit to the data (seeEq. (2a)) yielded values of k1 = 2.22×10−4 h−1 g−1 iron L andN = 0.986. Thus the dehalogenation rate in this case was ob-served to be much slower than the well-mixed case discussedearlier. This was attributed to mass transfer limitations on thedehalogenation process in the absence of mixing.

As in case of the well-mixed system, Eqs. (1), (2) and (3)were solved simultaneously to obtain the variation of Ca, CT

and Cs with time. Following initial conditions were used forthis purpose, at time, t = 0, Ca = CT = [CT]0 = 9.8 µmole L−1.M was taken as 128 g L−1. Values of k1, m and N used in thesimulation were 2.22 × 10−4 h−1 g−1 iron L, 0.76 and 0.986,respectively. The values of adsorption and desorption rate con-stants, k2 and k3 respectively, were unknown, but K = k2

k3was

taken the same as in the well-mixed case, i.e., 0.056 (µmole g−1

iron)/(µmole L−1). Numerical solution of the above equationsfor various assumed values of k3 were compared with the cor-responding experimental data. The simulation correspondingk2 = 3.92 × 10−5 h−1 g−1 iron L and k3 = 7 × 10−4 h−1 wasobserved to fit the experimental data adequately. Experimentaldata and corresponding model simulations for CT, Ca and Cs

are shown in Figs. 7A, 7B and 7C, respectively. Comparisonof the adsorption/desorption rates in well-mixed and poorly-mixed cases indicate that the latter are slower by several ordersof magnitude. As in case of dehalogenation rates, this mayalso be attributed to mass transfer limitations on the adsorp-tion/desorption process. Further, in the well-mixed case the rateof adsorption/desorption was fast compared to the rate of de-halogenation, resulting in adsorption equilibrium being main-tained. However, in the poorly-mixed case, the opposite is true.A plot of Cs vs Ca in the poorly mixed case (Fig. 8) indicate that

558 A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561

Fig. 5. Degradation of 2-CN contacted with HCIF in well-mixed batch reactors (M = 128 g L−1). (A) Total 2-CN. (B) Aqueous phase 2-CN. (C) Solid phase 2-CN.

the data does not correspond to the adsorption isotherm, sug-gesting nonequilibrium partitioning of 2-CN to the solid phase.

4. Discussion

Reductive dehalogenation of a HOC at the HCIF surface is acomplex reaction [33], that involves, (1) transport of the HOCmolecule to a reactive site, i.e., a cathodic site on the HCIF sur-face, (2) release of electrons at a corresponding anodic site bythe oxidation of metallic iron, (3) transport of the electrons fromthe anodic site to the cathodic site (either through the metal orthe surrounding solution) and finally (4) the acceptance of theelectrons by the HOC molecule at the cathodic site resultingin reductive dehalogenation. The complex reaction process de-scribed above is approximated by a mixed order reaction in thisstudy. During interaction of 2-CN with un-rusted HCIF sur-face, it appears that the final step as described above controlsthe overall reaction rate in well-mixed systems. However, inpoorly-mixed conditions, the first step, i.e., transport of the 2-CN molecule to a reactive site, appears to control the overallreaction rate. Based on this reasoning, it is entirely possiblethat the dehalogenation reaction, which appears mixed order(k1 = 1.09 × 10−2 h−1 g−1 iron L and N = 1.647) in well-mixed conditions, may well appear to be considerably slowerand approximately first order (k1 = 2.22 × 10−4 h−1 g−1 iron Land N = 0.986) in poorly-mixed conditions.

The rate of solid–liquid partitioning of any solute is moreconventionally described using a mass transfer term rather thanadsorption (k2) and desorption (k3) rate constants used in themodel described in this paper. A mass transfer term is howeverimplicit in the developed model, as described below. It is as-sumed that HCIF is surrounded by a stagnant aqueous film ofthickness L, and the solid phase concentration (Cs) is always inequilibrium with the aqueous 2-CN concentration (Cf

a), at theaqueous film boundary adjacent to the solid surface, i.e.,

(4a)Cs = K.[Cf

a

]m.

The aqueous 2-CN concentration at the far end of the aqueousfilm is equal to the bulk aqueous phase concentration (Ca). Un-der the circumstances, the differential equation describing therate of change of Ca due to adsorption/desorption and dehalo-genation,

(3)d[Ca]

dt= k3.M.[Cs] − k2.M.[Ca]m − k1.M.[Ca]N

can be rewritten as follows by substituting Eq. (4a) in Eq. (3),

(3a)d[Ca]

dt= k3.K.M.

[Cf

a

]m − k2.M.[Ca]m − k1.M.[Ca]N.

Also, K = k2k3

, and hence, k2 = K.k3.Therefore,

(3b)d[Ca] = k2.M.

[Cf

a

]m − k2.M.[Ca]m − k1.M.[Ca]N

dt

A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561 559

Fig. 6. Reductive dehalogenation of 2-CN by HCIF in poorly-mixed batch re-actors (M = 128 g iron L−1). (A) Decline in total 2-CN concentration. (B) Lin-earized plot of rate of 2-CN dehalogenation.

or,

(3c)d[Ca]

dt= KM.

{[Cf

a

]m − [Ca]m} − k1.M.[Ca]N,

where KM = k2.M .In well-mixed conditions, using the appropriate k2 value

(k2 = 5.60×10−1 h−1 g−1 iron L) and for M = 128 g iron L−1,KM was calculated to be 71.7 h−1. Due to the relatively highvalue of KM , equilibrium between solid phase 2-CN concen-tration (Cs) and aqueous phase 2-CN concentration (Ca) isachieved in a time scale which is considerably faster than de-halogenation rate of 2-CN in well-mixed conditions. Hence Caand Cs appear to be in equilibrium, and under such conditions[Ca] = [Cf

a].In poorly mixed conditions, using the appropriate k2 value

(k2 = 3.92×10−5 h−1 g−1 iron L) and for M = 128 g L−1, KM

was calculated to be 5 × 10−3 h−1. Due to the relatively lowvalue of KM , equilibrium between solid phase 2-CN concen-tration (Cs) and aqueous phase 2-CN concentration (Ca) is notobserved in the time scale of 2-CN dehalogenation in poorlymixed conditions. Hence Ca and Cs are not in equilibrium, and[Ca] is always greater than [Cf

a] under these circumstances.Surface area of the HCIF used in this study was determined

to be 3.418 m2 g−1. Therefore, for M = 128 g L−1, total sur-face area (A) of HCIF is 0.437 m2 m−3. Hence the values of

mass transfer coefficients (KSL = KM/A) were calculated to be4.56 × 10−2 and 3.18 × 10−6 m s−1 in well-mixed and poorly-mixed conditions respectively corresponding to M = 128 giron L−1. Assuming KSL = D/L, where D is the molecular dif-fusivity of 2-CN, the difference in KSL values in well-mixedand poorly-mixed conditions is attributed to much lower valuesof L in well-mixed conditions.

Another issue of concern is the reduction in the rates ofHCIF mediated reductive dehalogenation of HOCs due to ac-cumulation of iron corrosion products on the HCIF surface [8,34–36]. However, this issue is not directly relevant to the ob-jectives of the present study, where experiments were carriedout with un-rusted HCIF. Moreover, under anoxic conditionsemployed in this study, any deactivation of the HCIF surfacethrough deposition of iron corrosion products is expected tooccur over time scales that are considerably greater than theduration of the experiments reported in this paper. Nonetheless,the long-term impacts of the rusting of HCIF surface on HOCsdehalogenation rates is an active and interesting area of ongoingresearch.

5. Summary and conclusions

This study was designed to delineate the differences in theinteraction between 2-CN and HCIF in well-mixed and poorly-mixed systems. The results of this study indicate the following:

1. Contacting 2-CN with HCIF in well-mixed batch systemsresulted rapid partitioning of 2-CN to graphite inclusionson the HCIF surface. The residual aqueous phase 2-CN un-derwent dehalogenation at a slower rate. Decline in aque-ous phase 2-CN concentration due to dehalogenation re-sulted in gradual desorption of 2-CN initially adsorbed tothe solid phase. The dehalogenation reaction was deter-mined to be mixed order, with reaction rate constant (k1)

and order (N ) being 1.09×10−2 h−1 g−1 iron L and 1.647,respectively. Adsorption (k2) and desorption (k3) rates con-stants were determined to be 5.60 × 10−1 h−1 g−1 iron Land 10 h−1, respectively.

2. In poorly-mixed batch systems, partitioning of 2-CN to thesolid phase was much slower, resulting in maintenance ofa higher aqueous phase 2-CN concentration. The rate ofdehalogenation was also slower. The dehalogenation reac-tion was determined to be nearly first order, with reactionrate constant (k1) and order (N ) being 2.22×10−4 h−1 g−1

iron L and 0.986, respectively. Adsorption (k2) and des-orption (k3) rates constants were determined to be 3.92 ×10−5 h−1 g−1 iron L and 7 × 10−4 h−1, respectively.

3. Since partitioning reactions were faster than dehalogena-tion reaction in well-mixed batch systems, equilibrium be-tween solid and aqueous phase 2-CN concentration wasmaintained at all times, except just after 2-CN addition tothe system. Since partitioning reactions were slower thandehaogenation reaction in poorly-mixed systems, equilib-rium was never maintained between aqueous and solidphase 2-CN concentration. Mass transfer coefficients (KSL)for solid–liquid partitioning of 2-CN were calculated to

560 A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561

Fig. 7. Degradation of 2-CN contacted with HCIF in poorly-mixed batch reactors (M = 128 g iron L−1). (A) Total 2-CN. (B) Aqueous phase 2-CN. (C) Solid phase2-CN.

Fig. 8. Aqueous vs sorbed 2-CN in poorly-mixed batch reactors (M =128 g L−1).

be 4.56 × 10−2 and 3.18 × 10−6 m s−1 in well-mixedand poorly-mixed conditions respectively corresponding toM = 128 g iron L−1.

Interestingly, interaction of 2-CN with HCIF in both well-mixed and poorly-mixed systems could be modeled using thesame conceptual framework. This ensured that the experimen-tal data obtained from both well-mixed and poorly-mixed sys-tems could be simulated by merely changing values of dehalo-genation, adsorption and desorption rate constants in the same

model. It is thus conceivable that the model developed can beextended to describe 2-CN dehalogenation by HCIF in poorly-mixed flow-through systems like PRBs, where mass transferlimited partitioning and dehalogenation of HOCs are likely tooccur.

References

[1] T.M. Vogel, C.S. Criddle, P.L. McCarty, Environ. Sci. Technol. 21 (1987)722–736.

[2] J.J. Pignatello, K. Baehr, J. Environ. Qual. 23 (1994) 365–370.

[3] D.M. Mackay, J.A. Cherry, Environ. Sci. Technol. 23 (1989) 630–636.

[4] R.W. Gillham, D.R. Burris, Subsurface Restoration Conference, in: 3rdInternational Conference on Ground Water Quality Research, Dallas, TX,June 21–24, 1992.

[5] K.H. Sweeny, AIChE Symp. Ser. 77 (209) (1981) 67–71.

[6] K.H. Sweeny, AIChE Symp. Ser. 77 (209) (1981) 72–78.

[7] T. Senzaki, Y. Kumangai, Kogyo Yousi 357 (1988) 2–7 (in Japanese).

[8] R.W. Gillham, S.F. O’Hannesin, Ground Water 32 (1994) 958–967.

[9] C.G. Schreir, M. Reinhard, Chemosphere 29 (1994) 1743–1753.

[10] F. Chuang, R.A. Larson, M.S. Wessman, Environ. Sci. Technol. 29 (1995)2460–2463.

[11] B.R. Helland, P.J.J. Alvarez, J.L. Schnoor, J. Hazard. Mater. 41 (1995)205–216.

[12] A.L. Roberts, L.A. Totten, W.A. Arnold, D.R. Burris, T.J. Campbell, Env-iron. Sci. Technol. 30 (1996) 2654–2659.

[13] D.R. Burris, R.M. Allen-King, V.S. Manoranjan, T.J. Campbell, G.A. Lo-raine, B. Deng, J. Environ. Eng. 124 (1998) 1012–1019.

A. Sinha, P. Bose / Journal of Colloid and Interface Science 314 (2007) 552–561 561

[14] G.F. Slater, B.S. Lollar, R. Allen King, S. O’Hannesin, Chemosphere 49(2002) 587–596.

[15] C.J. Clark II, P.S.C. Rao, M.D. Annable, J. Hazard. Mater. 96 (2003) 65–78.

[16] T. Kohn, K.J.T. Livi, A.L. Roberts, P.J. Vikesland, Environ. Sci. Tech-nol. 39 (2005) 2867–2879.

[17] J. Gotpagar, S. Lyuksyutov, R. Cohn, E. Grulke, D. Bhattacharyya, Lang-muir 15 (1999) 8412–8420.

[18] D.J. Gaspar, A.S. Lea, M.H. Engelhard, D.R. Baer, R. Miehr, P.G. Trat-nyek, Langmuir 18 (2002) 7688–7693.

[19] H.L. Lien, W.X. Zhang, Colloids Surf. Physicochem. Eng. Aspects 191(2001) 97–105.

[20] S.F. O’Hannesin, R.W. Gillham, Ground Water 36 (1998) 164–170.[21] R.W. Puls, D.W. Blowes, R.W. Gillham, J. Hazard. Mater. 68 (1999) 109–

124.[22] D.R. Burris, T.J. Campbell, V.S. Manoranjan, Environ. Sci. Technol. 29

(1995) 2850–2855.[23] B. Deng, D.R. Burris, T.J. Campbell, Environ. Sci. Technol. 33 (1999)

2651–2656.[24] M.M. Scherer, J.C. Westall, M. Ziomek-Moroz, P.G. Tratnyek, Environ.

Sci. Technol. 31 (1997) 2385–2391.[25] V.K. Rajagopal, D.R. Burris, Environ. Toxicol. Chem. 18 (1999) 1779–

1782.

[26] L.J. Matheson, P.G. Tratnyek, Environ. Sci. Technol. 28 (1994) 2045–2053.

[27] K.D. Warren, R.G. Arnold, T.L. Bishop, L.C. Lindholm, E.A. Betterton, J.Hazard. Mater. 41 (1995) 217–227.

[28] T.L. Johnson, M.M. Scherer, P.G. Tratnyek, Environ. Sci. Technol. 30(1996) 2634–2640.

[29] J. Farrell, M. Kason, N. Melitas, T. Li, Environ. Sci. Technol. 34 (2000)514–521.

[30] P. Zhang, X. Tao, Z. Li, R.S. Bowman, Environ. Sci. Technol. 36 (2002)3597–3603.

[31] U.S. E.P.A. Water Quality Standards Database, U.S. EPA Internet web-site, http://oaspub.epa.gov/wqsdatabase/wqsi_epa_criteria.rep_parameter(2002).

[32] A. Sinha, P. Bose, Water Air Soil Pollut. 172 (2006) 375–390.[33] J.R. Davis (Ed.), Corrosion: Understanding the Basics, ASM International,

Metals Park, OH, 2000.[34] J.F. Devlin, J. Klausen, R.P. Schwarzenbach, Environ. Sci. Technol. 32

(1998) 1941–1947.[35] M.S. Odziemkowski, T.T. Schuhmacher, R.W. Gillham, E.J. Reardon,

Corros. Sci. 40 (1998) 371–389.[36] S.L.S. Stipp, M. Hansen, R. Kristensen, M.F. Hochella, L. Bennedsen, K.

Dideriksen, T. Balic-Zunic, D. Leonard, H.J. Mathieu, Chem. Geol. 190(2002) 321–337.