Pc

HN

a

ARRAA

KPSCS2

1

foboEmasfotbce

t

0d

Journal of Hazardous Materials 165 (2009) 860–866

Contents lists available at ScienceDirect

Journal of Hazardous Materials

journa l homepage: www.e lsev ier .com/ locate / jhazmat

CB congener sorption to carbonaceous sediment components: Macroscopicomparison and characterization of sorption kinetics and mechanism

yeok Choi, Souhail R. Al-Abed ∗

ational Risk Management Research Laboratory, U.S. Environmental Protection Agency, 26 West Martin Luther King Drive, Cincinnati, OH 45268, USA

r t i c l e i n f o

rticle history:eceived 12 August 2008eceived in revised form 15 October 2008ccepted 15 October 2008vailable online 5 November 2008

eywords:olychlorinated biphenylsorptionarbonaceous component

a b s t r a c t

Sorption of polychlorinated biphenyls (PCBs) to sediment is a key process in determining their mobility,bioavailability, and chemical decomposition in aquatic environments. In order to examine the validityof currently used interpretation approaches for PCBs sorption, comparative results on 2-chlorobiphenylsorption to carbonaceous components in sediments (activated carbon, carbon black, coal, soot, graphite,flyash, wood) were macroscopically correlated with the structural, morphological, crystallographic, andcompositional properties of the carbonaceous components. Since the Freundlich sorption constant, KF

(L kg−1) spanned several orders of magnitude, ranging from log KF of 6.13–5.27 for activated carbon, 5.04for carbon black, 3.83 for coal to 3.08 for wood, organic carbon partitioning approach should be morespecifically categorized, considering the various forms, nature and origins of organic carbon in sediment.

ediment-chlorobiphenyl

Sorption rate constants and fraction parameters, which were numerically defined from empirical kineticmodel with fast and slow sorption fractions, were closely related to the physicochemical properties ofthe carbonaceous components. Sorption interpretation approaches with a specific property and view-point, such as organic carbon partitioning, soot carbon distribution, or surface area correlation, did notproperly explain the overall results on sorption capacity, fast and slow sorption kinetics, and partitioningcoefficient. It is also important to emphasize the heterogeneous nature of sediment and the difficulties of

ning

tcnbmRst[atKran

encompassing the partitio

. Introduction

Most polychlorinated biphenyls (PCBs) in the environment areound in sediment which acts as a source for long-term releasef PCBs to aqueous phase [1,2]. In addition to their high sta-ility and low aqueous solubility, the high affinity of PCBs forrganic substances in sediment makes them difficult to treat [3,4].ven though the use of reactive metallic particles has been docu-ented to work efficiently for the dechlorination reaction of PCBs

nd other chlorinated compounds in aqueous phase [4–6], PCBstrongly adsorbed to sediment limit their reaction with metal sur-aces [7–10]. Consequently, sorption of PCBs and other hydrophobicrganic compounds to sediment is a key process in determiningheir fate and risk in aquatic environments. Many studies haveeen conducted to estimate PCBs partitioning to geosorbents, and

haracteristic sorption kinetics and mechanism have been wellstablished experimentally and conceptually [8,11–17].

PCBs adsorb to mainly organic substances in sediment and thushe organic carbon fraction (fOC) is a direct indication for the sorp-

∗ Corresponding author. Tel.: +1 513 569 7849; fax: +1 513 569 7879.E-mail address: al-abed.souhail@epa.gov (S.R. Al-Abed).

ti[

onsg

304-3894/$ – see front matter. Published by Elsevier B.V.oi:10.1016/j.jhazmat.2008.10.100

among its carbonaceous components.Published by Elsevier B.V.

ion capacity of sediment, introducing organic carbon partitioningoefficient (KOC) [7]. In addition to natural organic matter (alsooted as amorphous carbon elsewhere), sediment organic car-on consists of soot carbon (hard carbon or black carbon) withuch higher sorption capacity such as coal, coke, and soot [11,17].

ecently, some concrete evidence has shown that the presence ofoot-like materials enhances sorption of PCBs compared to predic-ions based on dissolution of organic compounds in geosorbents12–15]. Accardi-Dey and Gschwend proposed considering bothbsorption into a biogenic soft organic carbon fraction and adsorp-ion onto a combustion-derived black carbon fraction [15]. SinceOC values for different forms of organic carbon are expected toange several orders of magnitude, PCBs adsorbed to soot carbonsre much less available in aqueous phase than those associated withatural organic matter [14]. As a result, the organic carbon parti-ioning approach might be less helpful to predict PCBs distributionn sediment matrix with heterogeneous forms of organic carbon16].

Rather, PCBs sorption to sediment matrix is a complex functionf the physicochemical properties of numerous sediment compo-ents, including structural (e.g., surface area, pore volume, poreize), morphological (grain size, surface morphology), crystallo-raphic (amorphous and crystal), and compositional properties

zardo

(oPtapicaeitsewsa

srewvtwpaooa

2

2

moIb1e(fia(s

2

r(dpaapsuXmnw

OaTd[

2

bcipcetcoltwa4stctaulca

2

paa1aowvlsd6cid

3

3

cmo

H. Choi, S.R. Al-Abed / Journal of Ha

soot carbon, amorphous carbon) [8]. In addition, surface chemistryf sediment components has been reported to play a critical role inCBs sorption [9,18]. However, along with the organic carbon parti-ioning concept mentioned previously, the surface area correlationpproach has been common to most simply explain sorption resultsarticularly obtained from comparative experiment [19,20]. Focus-

ng on such specific properties (this might be plausible in mostases) might make us to overlook other crucial properties, whichre not under consideration but might actually rule the sorptionvent. For instance, soot and carbon black, which are very similarn nature and composition, are obviously different with respect toheir surface functional groups as determined through molecularcale-spectroscopic observation. On the other hand, when consid-ring their characteristic morphology, micron size carbon blackith nanostructured surface structure is unique compared to nano-

ize soot with nonporous surface structure, which significantlyffects sorption kinetics and capacity [8,21].

Many organic carbon–PCB congener combinations have beentudied both thermodynamically and kinetically before, but theesults have been interpreted in fragments, focusing on some prop-rties of sorbent in each case study. Consequently, in this study,e investigated sorption of aqueous 2-chlorobiphenyl (2-ClBP) to

arious carbonaceous organic materials, and the comparative sorp-ion kinetics and isotherms among the carbonaceous componentsere systematically correlated with their extensively characterizedhysicochemical properties. Through the macroscopic comparisonnd characterization of 2-ClBP sorption, we addressed the validityf currently used PCBs sorption interpretation approaches, such asrganic carbon partitioning, soot carbon distribution, and surfacerea correlation.

. Experimental

.1. Carbonaceous components

General aspects of the carbonaceous components used are sum-arized in Table S1 in Supplementary material. Briefly, two types

f activated carbons: microporous one (GAC300, Norit America,nc.) and mesoporous one (HD3000, Norit America, Inc.), carbonlack (Monarch 800, Carbot Corporation), soot (NIST reference, SRM650), graphite (EC100, Graphite sales, Inc.), and coal (BCR refer-nce, CRM 460) were selected along with wood (Sigma) and flyashBCR reference, CRM 038). GAC300 and HD3000 were further puri-ed with deionized water several times to remove any impuritiesnd heterogeneous debris. After grinding, some of the materialsactivated carbon, graphite, and wood) were sieved with a #100ieve if necessary.

.2. Carbonaceous component characterization

For the characterization of carbonaceous materials used, X-ay diffraction analysis using a Kristalloflex D500 diffractometerSiemens) with CuK� (� = 1.5406 Å) radiation was employed toetermine their crystal nature. A Tristar 3000 (Micromeritics)orosimetry analyzer was used to determine their structural char-cteristics, including Brunauer, Emmett, and Teller (BET) surfacerea, Barrett, Joyner and Halenda (BJH) pore size distribution, andorosity using nitrogen adsorption and desorption isotherms. Theurface morphology of sediment components was investigated

sing environmental scanning electron microscope (ESEM, PhilipsL 30 ESEM-FEG) at the micro-level and high resolution trans-ission electron microscope (HR-TEM, JEM-2010F, JEOL) at the

ano-level. Point of zero charge of the carbonaceous componentsas measured using a Zetasizer 3000 HSA (Malvern Instruments).

etnaI

us Materials 165 (2009) 860–866 861

rganic carbon content was measured using a total organic carbonnalyzer (TOC VCS, Shimadzu) with solid module (SSM-5000A).he fractions of amorphous organic carbon and soot carbon wereetermined by whether the organic carbon is combustible at 375 ◦C11].

.3. Batch sorption studies

As a model PCB, 2-ClBP (AccuStandard, PCB-#1) was usedecause of its highest aqueous solubility at 5–6 mg/L among PCBongeners. The experimental design was based on batch tests,nvolving sorption equilibrium of 2-ClBP between the aqueoushase and individual carbonaceous component. The volume andoncentration of 2-ClBP solution were fixed while the amount ofach carbonaceous component was varied. A 2-ClBP aqueous solu-ion at around 4.0 mg L−1 was prepared once for all experiments. Aalculated amount of carbonaceous component followed by 100 mLf water was added in a 250 mL borosilicate glass bottle with Teflon-ined caps (Fisher). Due to no direct contact of 2-ClBP solution andhe Teflon cap, undesired loss of 2-ClBP caused by sorption to Teflonas minimized. The solution pH was adjusted at 7.0 using 1N nitric

cid (Fisher) or 1N sodium hydroxide (Fisher). Then, 100 mL of.0 mg L−1 2-ClBP solution was added to the pH-adjusted 100 mLolution containing carbonaceous component in order to bring ini-ial 2-ClBP concentration to 2.0 mg L−1 and make carbonaceousomponent/2-ClBP ratio of 5–5000 in 200 mL solution. The bot-les were packed in a Gyrotory shaker (New Brunswick Scientific)t room temperature of 20 ± 1 ◦C at 150 rpm for a period of 4 weeksnder dark conditions. Considering the hydrophobic nature and

ow concentration of 2-ClBP and high adsorptive capacity of thearbonaceous components, the equilibration was expected to bechieved within a short time [22].

.4. Analytical methods

During the batch sorption studies above, 1 mL of slurry sam-le was taken at each time interval of 0, 0.5, 2, 8, 24, 48, 96, 168,nd 672 h. Total loss of reaction volume was less than 5%. The solidnd liquid phases were separated by centrifuging at 6000 rpm for0 min (Sorvall RC-5B Superspeed Centrifuge). After the phase sep-ration, 0.5 mL of supernatant was collected and mixed with 1.5 mLf extraction solvent hexane (Fisher) in 4 mL micro-reaction vesselith PTFE liner (Supelco). The vessel was vigorously agitated in a

ortex for 2 min and 0.5 mL of hexane phase with 2-ClBP was col-ected. Extracted samples were spiked with 10 �L of the internaltandard (D-8 naphthalene stock at 250 ppm in hexane (AccuStan-ard)), and analyzed using a gas chromatograph (Hewlett Packard890) (EPA SW-846 Method 8270C) with a SPB-5 column (Supelco),oupled with a mass spectrometer (Hewlett Packard 5973). The cal-bration curve had R2 of greater than 0.999 based on eight-pointata and the 2-ClBP detection limit was 5 �g L−1.

. Results and discussions

.1. Carbonaceous component properties

The physicochemical properties of the carbonaceous organicomponents are summarized in Table 1 and Table S2 in Supple-entary material. The carbonaceous components are composed

f mainly amorphous organic carbon and partially soot carbon,

xcept graphite with high soot carbon content. The carbon frac-ion, fOC of the carbonaceous components reflects their origin andature. Surface area for PCBs sorption is originated from the bound-ry (inter-particle) and porous structure of particle (intra-particle).n case of nonporous materials, surface area is a strong function of

862 H. Choi, S.R. Al-Abed / Journal of Hazardous Materials 165 (2009) 860–866

Table 1Physicochemical characteristics of carbonaceous organic componentsa.

Component GSb (�m) fOCc fSC

c Crystald Porouse SBETf (m2 g−1) VPore

g (cm3 g−1) PSh (nm) PSdish (nm)

GAC300 1–150 0.81 0.38 Amorp Micro+meso 787 0.506 3.69 <3HD3000 1–150 0.84 0.36 Amorp Meso+micro 474 0.543 5.42 <6, 8–40Carbon black 2–7 0.99 0.25 Amorp Meso 184 0.915 20.4 >10Soot 0.03–0.04 0.90 0.42 Amorp Meso 62.4 0.223 15.6 >20Graphite 10–150 1.00 0.69 Crystal Non 2.79 0.011 15.4 >10Coal 5–20 0.69 0.04 Amorp Non 2.89 0.011 17.5 >15Wood 100–150 0.48 0.02 Amorp Non 0.588 – 18.5 >20Flyash 0.5–10 0.12 0.10 Crystal Non 4.40 0.013 14.2 >25

a For more detailed properties, note Tables S1 and S2 in Supplementary material.b GS (grain size): large particles such as GAC300, HD3000, graphite and wood were ground and sieved with 150 �m sieve.c fOC and fSC: total organic carbon and soot carbon fractions, respectively.d Even amorphous (amorp) materials here, except wood, have crystallinity at some extent.e

gfcl7ttwtsg

gpw

ca

Fs

Porous structure (micro: microporous, meso: mesoporous, non: nonporous).f BET surface area.g Single point adsorption total pore volume at P/Po = 0.99.h PS: BJH adsorption average pore size and PSdis: pore size distribution.

rain size and 2-ClBP sorption to their grain boundary is straight-orward. Surface area of porous materials such as activated carbon,arbon black and soot is determined by their structure, regard-ess of their grain size. GAC300 has the highest surface area at87 m2 g−1 mainly in microporous regime less than 1.7 nm. Evenhough microporous structure is beneficial for sorption capacity,

he structure limits mass transfer of 2-ClBP. Meanwhile, HD3000ith surface area of 474 m2 g−1 has dominantly mesoporous struc-

ure at 7–40 nm. Carbon black is very similar to soot but it has higherurface area and pore volume and much more chemisorbed oxy-en complexes such as carboxylic, quinonic, lactonic, and phenolic

ppaws

ig. 1. Environmental scanning electron microscope and high resolution transmission eltudy: (a) carbon black, (b) graphite, (c) flyash, (d) soot, and (e) HD3000 activated carbon

roups in its surface. Crystal graphite, nonporous coal, and amor-hous wood show low surface area. Flyash has the lowest fOC at 12%hich is mostly soot carbon.

The micro- and nanoscale morphology of some carbonaceousomponents is shown in Fig. 1. Even though both carbon blacknd soot are formed during incomplete combustion of petroleum

roducts, carbon black (Fig. 1a) has well-defined 2–7 �m sphericalarticles with nanostructured surface (high microporous surfacerea) while soot (Fig. 1d) has 30–40 nm spherical nanoparticlesith nonporous surface. As a result, the origin of their adsorptive

ites is different: inner porous structure for carbon black vs. par-

ectron microscope images of some carbonaceous organic components used in this.

H. Choi, S.R. Al-Abed / Journal of Hazardous Materials 165 (2009) 860–866 863

F r cond( nd (h)a n modt ning

tsssmtmsscgss

3

urCw

Q

E

iEttspe(

Q

Q

F

F

t

TK

C

GHCSGCWF

ig. 2. Kinetics of 2-ClBP sorption to various carbonaceous organic components undea) GAC300, (b) HD3000, (c) carbon black, (d) soot, (e) graphite, (f) coal, (g) wood, and slow sorption fractions shown in Eq. (3). For comparison, the first order reactiohat X-axis is in log scale to visualize the fast and slow sorption kinetics in the begin

icle boundary for soot. Graphite (Fig. 1b) has nonporous densetructure, forming ordered graphitic layers. Flyash (Fig. 1c) con-ists of 0.5–10 �m finely divided spherical mineral residue. Theurface of HD3000 (Fig. 1e) at nano-level has a distinct micro- andeso-porous structure from disordered graphitic layers. In some

heoretical sorption kinetic models explaining external, pore andatrix diffusion phenomena, the term “pore” is unclearly used. In

ome cases (e.g., activated carbon), particle itself is porous micro-copically. Meanwhile, even nonporous particles are sometimesharacterized as porous macroscopically because of their aggre-ated nature. That is more prominent in case of aggregates of smallize nanoparticles, where their aggregates look porous even micro-copically (e.g., soot in Fig. 1d).

.2. Sorption kinetics

Kinetics of 2-ClBP sorption to various carbonaceous componentsnder condition of sorbent loading at 100 mg L−1 (sorbent/2-ClBPatio of 50) is shown in Fig. 2. The time evolution of the amount of 2-lBP adsorbed to carbonaceous component Q can be characterizedith first order reaction (Eqs. (1) and (2)) [23]:

= Qe × (1 − e−k×t) (1)

ads = Qe

Qe,max(2)

riTT

able 2inetic parameters of 2-ClBP sorption to carbonaceous organic components under condit

omponent Overall sorption Fast sorption

Qe (�mol g−1) Eads R2 Qe,fast (�mol g−1) Ffast

AC300 21.1 1.00 0.999 20.6 0.98D3000 20.9 0.99 0.999 20.1 0.96arbon black 20.9 0.99 1.000 20.4 0.98oot 13.7 0.65 0.999 10.5 0.77raphite 13.5 0.64 0.994 6.84 0.51oal 17.7 0.84 0.999 13.8 0.78ood 8.58 0.41 0.994 5.24 0.61

lyash 13.8 0.65 0.999 10.1 0.74

ition of sorbent loading at 100 mg L−1 and 2-ClBP concentration at around 2.0 mg/L:flyash. The solid lines are fitting curves to the first order reaction model with fastel fitting with one term shown in Eq. (1) is also presented with dotted lines. Note

of sorption event.

where Q is the amount of 2-ClBP adsorbed (�mol g−1 solid), Qe

s Q at equilibrium, k is first order rate constant (h−1), t is time (hr),ads is sorption efficiency at equilibrium, and Qe,max is Qe when allhe 2-ClBP provided is adsorbed. This model well represents ini-ial fast sorption followed by immediate equilibrium, but distortslow sorption kinetics typically shown in nonporous organic com-onents. The sorption results can be interpreted with the followingquation with fast sorption Qe,fast and slow sorption, Qe,slow termsEqs. (3)–(7)) [24]:

= Qe,fast × (1 − e−kfast×t) + Qe,slow × (1 − e−kslow×t) (3)

e = Qe,fast + Qe,slow (4)

fast = Qe,fast

Qe(5)

slow = Qe,slow

Qe(6)

50 = − ln(0.5)kfast

(7)

where Ffast and Fslow are fractions of fast and slow sorption,espectively, and half of sorption time, t50 (i.e., t at Qe = Qe,fast/2)s based on Qe,fast considering two-fold higher kfast than kslow.he kinetic parameters are summarized in Tables 2 and 3 andable S3 in Supplementary material. As shown in Fig. 2, the two

ion of sorbent loading at 100 mg L−1 and 2-ClBP concentration at around 2.0 mg/L.

Slow sorption

kfast (h−1) t50 (min) Qe,slow (�mol g−1) Fslow kslow (×10−2, h−1)

4.98 8.35 0.52 0.02 1.973.98 10.4 0.79 0.04 7.307.59 5.48 0.48 0.02 4.745.52 7.53 3.21 0.23 5.130.49 84.8 6.61 0.49 4.081.22 34.1 3.97 0.22 3.710.81 51.3 3.35 0.39 3.143.62 11.5 3.65 0.27 4.11

864 H. Choi, S.R. Al-Abed / Journal of Hazardo

Table 3Organic carbon (OC)-, BET surface area (SA)-, and pore volume (PV)-normalized Qe

of carbonaceous organic componentsa.

Material Qe (�mol/unit below)

Solid (g) OC (g) SA (m2) PV (cm3)

GAC300 21.1 26.0 0.03 41.7HD3000 20.9 25.0 0.04 38.5CB 20.9 21.1 0.11 22.9Soot 13.7 15.3 0.22 61.5Graphite 13.5 13.5 4.82 (1223)Coal 17.7 25.7 6.14 (1612)Wood 8.58 17.9 (14.6)b –Flyash 13.8 120 3.12 (1058)

a −1

a

t

tdawg

btecsgwEhtetawtnatCtdswrpsTs

3

cHaip

Q

(

t1Fbgbsf(p2f(

3

sr2weisgbehHsbiiihthimmsb

mhttimoeofospctas

Under condition of sorbent loading at 100 mg L and 2-ClBP concentration atround 2.0 mg/L.b The values in parentheses might not be valid due to the too small SA and PV of

he carbonaceous components.

erms (or two compartments) model was fitted to the experimentalata with slight deviation most probably due to the portion associ-ted with very slow sorption [21]. Eads of carbonaceous componentsas in order of GAC300, HD3000, carbon black � coal > soot, flyash,

raphite > wood.Due to their high surface area, GAC300, HD3000, and carbon

lack even at low loading adsorbed all 2-ClBP. Comparable sorp-ion capacity of carbon black to HD3000 might be because of itsxtremely high pore volume, as observed in sorption of certainhemicals [25]. Sorption efficiency of carbon black compared tooot was much higher due to its hydrophobic surface functionalroups. In most cases, kfast was two-fold higher than kslow and Ffastas dominant over Fslow. The sediment components with higher

ads were characterized by faster sorption. As expected from theirigh surface area, GAC300, HD3000, and carbon black were charac-erized by fast sorption while the other carbonaceous components,specially, graphite and wood had significant Fslow. The slow sorp-ion could be attributed to slow pore diffusion of 2-ClBP into innerdsorptive sites. That is at least true for carbonaceous componentsith porous structure or agglomerated nanosize particle charac-

eristics such as soot. In the case of carbonaceous components withonporous dense structure, such as graphite, coal, wood, and fly-sh, matrix diffusion might be important, and 2-ClBP adsorbedo the sites can affect further sorption of 2-ClBP (stacking effect).arbon black and soot exhibited significantly higher kfast due toheir hydrophobic surface functional groups which are introduceduring their manufacturing processes. Even though the rate con-tants and fraction parameters for the fast and slow terms hereere numerically defined from the model fitting, they were closely

elated to the physicochemical properties of carbonaceous com-onents. The characteristic findings and tendency in comparativeorption kinetics among the components, shown in Fig. 2 andables 2 and 3, and S3, were also consistently observed under otherorbent loading conditions than 100 mg L−1.

.3. Sorption isotherms

The sorption isotherms are shown in Fig. 3. The carbonaceousomponents can be grouped into high adsorptive group (GAC300,D300, carbon black) and intermittent group (coal, soot, fly-sh, graphite, wood). The Freundlich equation was fitted to thesotherms (Eq. (8)) and the Freundlich isotherm parameters areresented in Table 4.

e = KF × (Ce)1/n (8)

where Qe is 2-ClBP adsorbed to carbonaceous component�mol kg−1), Ce is 2-ClBP remaining in liquid (�mol L−1), KF is

sattP

us Materials 165 (2009) 860–866

he Freundlich sorption constant ((�mol/kg)/(�mol/L)1/n, L kg−1 if/n = 1), and 1/n is the Freundlich exponent at equilibrium. Thereundlich sorption constants, KF of GAC300, HD3000, and carbonlack with high surface area and hydrophobic surface functionalroups were 2–3 order higher in magnitude than the other car-onaceous components. The KF values for different organic formspanned several orders of magnitude, ranging from log KF of 6.13or activated carbon to 3.08 for wood. The Freundlich exponent1/n) for most of carbonaceous components particularly with non-orous structure was around 1.0, suggesting linear characteristic-ClBP sorption. As observed in many porous adsorbents, isothermsor GAC 300 (1/n = 0.799), HD3000 (1/n = 0.824), and carbon black1/n = 0.727) exhibited slightly nonlinear characteristic [23].

.4. Characteristic 2-ClBP sorption

Based on the sorption kinetics and isotherms, characteristicorption of 2-ClBP to the carbonaceous components is summa-ized in Table S4 in Supplementary material. The comparative-ClBP sorption results were qualitatively in good agreementith the macroscopic observation of the physicochemical prop-

rties of the components [8]. Organic carbon-normalized Qe wasn order of flyash � GAC300, HD3000, coal > carbon black, wood,oot, graphite while surface area-normalized Qe was high for coal,raphite, and flyash, which are considerably different from solidased Qe. Carbonaceous components with high surface area hadxceptionally high sorption capacity (except soot) and dense andard nonporous components had relatively low sorption capacity.owever, their surface area was not a direct indication of their

orption capacity (e.g., soot vs. coal). Because of their hydropho-ic nature, PCBs predominantly adsorb to the hydrophobic regions

n sediments, organic carbon, and thus sorption of PCBs to sed-ment is commonly described as a function of fOC [8,18]. Thiss true only in case of varying organic carbon content with aomogeneous organic source in composition. The order of sorp-ion capacity and Freundlich constant of the organic componentsere did not follow their organic carbon content. Based on the

sotherm parameters reported in Table 4, PCBs in a sedimentatrix with 10% fOC originated from graphite might be muchore partitioned to aqueous phase, compared to those in another

ediment matrix with even 1% fOC originated from activated car-on.

In addition, numerous studies have revealed that PCBs oftenore strongly adsorb to sediments than we expect based on

ydrophobic interactions according to the organic carbon parti-ioning [19,26]. In order to explain this, it is generally hypothesizedhat contaminants slowly diffuse into mineral micropores locatedn hard glassy organic carbon (denoted as soot carbon here). It

ight be true in case of graphite. The highest soot carbon contentf graphite among the components might have induced the high-st slow sorption fraction. However, only soot carbon content in therganic components did not explain either their sorption capacity,ast and slow sorption kinetics, or partitioning coefficient. In spitef its highest soot carbon fraction, graphite exhibited the lowestorption capacity. Generally, one assumes absorption into amor-hous carbon (soft carbon) and adsorption onto soot carbon (hardarbon) [15]. Moreover, the term “partitioning” relates to absorp-ion processes whereas the term “distribution” is considered moreppropriate in case of adsorption [27]. Van Noort et al. [28] demon-trated that maximum PCBs adsorption capacity of a wide range of

oot-like materials can be estimated from sorbent specific surfacerea, sorbent organic carbon content, and sorbate–sorbent con-act area. The three factors are, of course, critical in determininghe sorption capacity but not universal to explain whole results.CBs sorption to sediment matrix with various carbonaceous com-

H. Choi, S.R. Al-Abed / Journal of Hazardous Materials 165 (2009) 860–866 865

Fig. 3. Isotherms of 2-ClBP sorption to carbonaceous organic components: (a) GAC300, HD3000, carbon black, soot and (b) graphite, coal, wood, flyash. The lines are fittingcurves to the Freundlich sorption model.

Table 4Freundlich isotherm parameters of 2-ClBP sorption to carbonaceous organic componentsa.

Component log KF 1/n R2

Solid (L kg−1)b OC (L kg−1 OC) SA (L cm−2) PV (L cm−3)

GAC300 6.127 6.217 −3.760 3.423 0.799 0.999HD3000 5.274 5.352 −4.402 2.539 0.824 0.996Carbon black 5.037 5.041 −4.228 2.076 0.727 0.998Soot 3.589 3.636 −5.206 1.241 0.925 0.992Graphite 3.476 3.476 −3.970 2.435 0.951 0.987Coal 3.830 3.990 −3.631 2.789 0.993 0.999Wood 3.080 3.400 −3.690 – 0.933 0.992F

ore vo

ppb

4

btcecPitcesbowte

A

MpEd

tetfYad

A

i

R

lyash 3.582 4.521

a KF is normalized with organic carbon content (OC), BET surface area (SA), and pb Actual unit is (�mol/kg)/(�mol/L)1/n .

onents is a complex function of the numerous physicochemicalroperties of the carbonaceous components and surface chemistryetween the components and PCBs.

. Conclusions

The kinetics and the isotherm of 2-ClBP sorption to various car-onaceous organic components in sediment were investigated andhe results were systematically correlated with the properties of theomponents. Since the Freundlich sorption constant spanned sev-ral orders of magnitude depending on the nature and origin of thearbonaceous materials, organic carbon partitioning approach forCBs sorption should be more specifically categorized. The empir-cal kinetic model with fast and slow sorption fractions explainedhe 2-ClBP sorption event well. Organic carbon partitioning, hardarbon distribution, or surface area correlation alone did not prop-rly explain comparative results on sorption capacity, fast andlow sorption kinetics, and partitioning coefficient among the car-onaceous components. By recognizing the heterogeneous naturef sediment with various carbonaceous components exhibiting aide range of PCB sorption capacity and kinetics, we might bet-

er understand and predict PCBs sorption event in the sedimentnvironment.

cknowledgements

This research was funded and conducted by the National Riskanagement Research Laboratory of U.S. EPA, Cincinnati, Ohio. This

aper has not been subjected to internal policy review of the U.S.nvironmental Protection Agency. Therefore, the research resultso not necessarily reflect the views of the agency or its policy. Men-

−4.062 2.468 0.932 0.995

lume (PV).

ion of trade names and commercial products does not constitutendorsement or recommendation for use. The authors recognizehe support of Mr. Eric Graybill of Pegasus Technical Services, Inc.or sample preparation and analysis using GC/MS and Dr. Qiujingang at University of Cincinnati for sample characterization. Theuthors greatly appreciate Norit Americas Inc. for their generousonation of the activated carbon.

ppendix A. Supplementary data

Supplementary data associated with this article can be found,n the online version, at doi:10.1016/j.jhazmat.2008.10.100.

eferences

[1] K.N. Dimou, T.-L. Su, R.I. Hires, R. Miskewitz, Distribution of polychlorinatedbiphenyls in the Newark Bay Estuary, J. Hazard. Mater. 136 (2006) 103–110.

[2] Y.S. Kim, H. Eun, H.S. Cho, K.S. Kim, E. Watanabe, K. Baba, T. Katase, The char-acterization of PCDDs, PCDFs and coplanar PCBs during the past 50 years inGwangyang Bay, South Korea, J. Hazard. Mater. 154 (2008) 756–765.

[3] F. Kastanek, P. Kastanek, Combined decontamination processes for wastes con-taining PCBs, J. Hazard. Mater. 117 (2005) 185–205.

[4] Y. Liu, H. Choi, D.D. Dionysiou, G.V. Lowry, Trichloroethene hydrodechlorinationin water by highly disordered monometallic nanoiron, Chem. Mater. 17 (2005)5315–5322.

[5] S. Agarwal, S.R. Al-Abed, D.D. Dionysiou, Enhanced corrosion-based Pd/Mgbimetallic systems for dechlorination of PCBs, Environ. Sci. Technol. 41 (2007)3722–3727.

[6] H. Choi, S.R. Al-Abed, S. Agarwal, D.D. Dionysiou, Synthesis of Reactivenano-Fe/Pd bimetallic system-impregnated activated carbon for the simul-

taneous adsorption and dechlorination of PCBs, Chem. Mater. 20 (2008)3649–3655.

[7] C.T. Chiou, L.J. Peters, V.H. Freed, A physical concept of soil-water equilibria fornonionic organic compounds, Science 206 (1979) 831–832.

[8] R.G. Luthy, G.R. Aiken, M.L. Brusseau, S.D. Cunningham, P.M. Gschwend, J.J. Pig-natello, M. Reinhards, S.J. Traina, W.J. Weber Jr., J.C. Westall, Sequestration of

8 azardo

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

kerogen in sediments and soils: mechanisms and consequences for distribu-tion, bioaccumulation, and biodegradation, Environ. Sci. Technol. 39 (2005)

66 H. Choi, S.R. Al-Abed / Journal of H

hydrophobic organic contaminants by geosorbents, Environ. Sci. Technol. 31(1997) 3341–3347.

[9] N. Saleh, K. Sirk, Y. Liu, T. Phenrat, B. Dufour, K. Matyjaszewski, R.T. Tilton, G.V.Lowry, Surface modifications enhance nanoiron transport and NAPL targetingin saturated porous media, Environ. Eng. Sci. 24 (2007) 45–57.

10] P. Varanasi, A. Fullana, S. Sidhu, Remediation of PCB contaminated soils usingiron nanoparticles, Chemosphere 66 (2007) 1031–1038.

11] O. Gustafsson, F. Haghseta, C. Chan, H. Macfarlane, P. Gschwend, Quantifica-tion of the dilute sedimentary soot phase: implications for PAH speciation andbioavailability, Environ. Sci. Technol. 31 (1997) 203–209.

12] T.D. Bucheli, O. Gustafsson, Ubiquitous observations of enhanced solid affinitiesfor aromatic organochlorines in field situations: are in situ dissolved expo-sures overestimated by existing partitioning models? Environ. Toxicol. Chem.20 (2001) 1450–1456.

13] T.D. Bucheli, O. Gustafsson, Soot sorption of non-ortho and ortho substitutedPCBs, Chemosphere 53 (2003) 515–522.

14] U. Ghosh, Z.R. Zimmerma, R.G. Luthy, PCB and PAH speciation among particletypes in contaminated harbor sediments and effects on PAH bioavailability,Environ. Sci. Technol. 37 (2003) 2209–2217.

15] A. Accardi-Dey, P.M. Gschwend, Reinterpretation of literature sorption dataconsidering both absorption into organic carbon and sorption onto black car-bon, Environ. Sci. Technol. 37 (2003) 99–106.

16] Z. Bohacek, J. Bezdek, M. Kovarova, J. Hanak, J. Toul, P. Muller, Characteris-tics of organic matter and contents of some ubiquitous hydrophobic organicpollutants in selected soils and sediments, Bull. Geosci. 78 (2003) 179–204.

17] J. Im, C.M. Lee, J.T. Coates, Comparison of two reference black carbons using aplanar PCB as a model sorbate, Chemosphere 71 (2008) 621–628.

18] T.H. Yoon, K. Benzerara, S. Ahn, R.G. Luthy, T. Tyliszczak, G.E. Brown Jr.,Nanometer-scale chemical heterogeneities of black carbon materials and theirimpacts on PCB sorption properties: soft X-ray spectromicroscopy study, Env-iron. Sci. Technol. 40 (2006) 5923–5929.

19] M.T.O. Jonker, A.A. Koelmans, Sorption of polycyclic aromatic hydrocarbonsand polychlorinated biphenyls to soot and soot-like materials in the aque-

[

us Materials 165 (2009) 860–866

ous environment: mechanistic considerations, Environ. Sci. Technol. 36 (2002)3725–3734.

20] C. Cornelissen, M. Elmquist, I. Groth, O. Gustafsson, Effect of sorbate pla-narity on environmental black carbon sorption, Environ. Sci. Technol. 38 (2004)3574–3580.

21] J. Pignatello, B. Xing, Mechanisms of slow sorption of organic chemicals tonatural particles, Environ. Sci. Technol. 30 (1996) 1–11.

22] K.M. McDonough, J.L. Fairey, G.V. Lowry, Sorption of polychlorinated biphenylsto activated carbon: equilibrium isotherms and a preliminary assessment of theeffect of dissolved organic matter and biofilm loadings, Water Res. 42 (2008)575–584.

23] G. Cornelissen, P.C.M. van Noort, H.A.J. Govers, Mechanisms of slow desorptionof organic compounds from sediments: a study using model sorbents, Environ.Sci. Technol. 32 (1998) 3124–3131.

24] R.A. Griffiths, Sorption and desorption by ideal two-compartment systems:unusual behavior and data interpretation problems, Chemosphere 55 (2004)443–454.

25] K.J. Choi, S.G. Kim, C.W. Kim, S.H. Kim, Effects of activated carbon types and ser-vice life on removal of endocrine disrupting chemicals: amitrol, nonylphenol,and bisphenol-A, Chemosphere 58 (2005) 1535–1545.

26] M.T.O. Jonker, F. Smedes, Preferential sorption of planar contaminants in sedi-ments from Lake Ketelmeer, The Netherlands, Environ. Sci. Technol. 34 (2000)1620–1626.

27] G. Cornelissen, O. Gustafsson, T.D. Bucheli, M.T.O. Jonker, A.A. Koelmans, P.C.M.Van Noort, Extensive sorption of organic compounds to black carbon, coal, and

6881–6895.28] P.M. Van Noort, M.T.O. Jonker, A.A. Koelmans, Modeling maximum adsorp-

tion capacities of soot and soot-like materials for PAHs and PCBs, Environ. Sci.Technol. 38 (2004) 3305–3309.