Sorption of Buckminsterfullerene(C60) to Saturated SoilsC H I A - Y I N G C H E N A N DC H A D T . J A F V E R T *

Purdue University, School of Civil Engineering,West Lafayette, Indiana 47907

Received April 1, 2009. Revised manuscript received July31, 2009. Accepted August 10, 2009.

With the increasing use of C60 in many industrial andcommercial sectors, it is likely that it will eventually appear inthe environment; however, its environmental fate and transportis still largely unknown. The extent to which C60 partitions to soilwill contribute to its environmental fate and bioavailability.Because C60 is extremely hydrophobic, in this study the distributionbetween soil and mixtures of ethanol (EOH) and water weremeasured at ethanol mole fractions ranging from XEOH ) 1.0-0.4for two soils. By measuring Kp at XEOH ) 1.0 for a series ofsoils that ranged in organic carbon and clay mineral content,possible mineral contribution to the overall partition process wasfound for some of the soils. After correcting for any mineralcontribution to sorption, the organic carbon normalized partitioncoefficient, Koc, at each value of XEOH was calculated fromthe measured Kp values. Through a classical thermodynamicrelationship, the Koc values determined at XEOH ) 1.0-0.4 wereextrapolated to estimate the pure water (i.e., XEOH ) 0) Koc

value of 107.1 (L/kg). Accounting for any dissolved organic matter(DOM) in any pure water-soil mixtures may lower thisestimate by over a factor of 2, placing this estimate in goodagreement with C60’s octanol-water partition coefficient, Kow

()106.7).

Introduction

Since Buckminsterfullerene (C60) was discovered in 1985 (1),the unique physicochemical properties of it and its modifiedderivatives are motivating factors behind industry to findnew and unique applications for this class of nanomaterials(2). With the widespread use of C60, there is no doubt thatC60 will eventually appear in the environment. Indeed, evenwithout intended production, it is found in particulatesemitted from coal-burning power plants (3). Although itsenvironmental fate remains largely unknown, some potentialadverse effects on health and the environment have beenreported. Interpreting the results of many of these studieshowever is confounded by the fact that addition of C60 to themedia generally occurs via addition of aqueous phase“clusters” which are nanoparticles of C60 with diametersgenerally in the range of 100-500 nm. Clusters, known asnC60, form a stable colloidal suspension in water, and aregenerally prepared by solvent exchange techniques (4, 5) orby long-term stirring of solid material in pure water overseveral weeks, or short-term ultrasonication in water (6, 7).C60 cluster preparations and some of its derivatives have beenshown to have antibacterial or inhibitory effects to a broad

range of bacteria (5, 8). A study by Oberdorster indicatedthat nC60 can cause oxidative damage and depletion ofglutathione (GSH) in vivo in largemouth bass (9). AlthoughC60 administered as aqueous clusters were found to bepotentially harmful both in vitro and in vivo, several authorshave suggested the toxicity is due to the method of clusterpreparation (7, 10), due to solvation by the solvent used toprepare the clusters (tetrahydrofuran) or due to a solventdegradation product (γ-butyrolactone) (10).

Because nC60 forms stable aqueous colloidal suspensions,information on the environmental fate and transport of C60

reported in the literature has emphasized nC60 transport, yetC60 is a molecule with a molecular weight less than that ofmany common chemicals of concern, including brominatedflame retardants. It is quite soluble in numerous organicsolvents, and its molecular aqueous solubility recently wasreported at 7.96 ng/L (11). Hence, molecular C60 does existin water, and it is through this dissolved concentration oractivity that its environmental distribution will be regulated(i.e., thermodynamically controlled), including its self-association to form clusters or nanoparticles. In addition toaqueous solubility, the organic carbon content normalizedsoil-water partition coefficient, Koc, is an important param-eter that regulates distribution in the aquatic environment,and through which a bulk soil-water partition coefficient,Kp, can be estimated.

In the present study, we report on the phase distributionof molecular C60 between soil and ethanol-water mixturesand through a classical thermodynamic relationship, ex-trapolate the measured partition coefficients to the soil-waterpartition coefficient (Kp) that would occur in the absence ofthe ethanol cosolvent. The use of the water-miscible cosolventwas necessary as the aqueous solubility of C60 is lower thanmost instrument detection limits unless several liters of waterare concentrated prior to analysis. The presence of cosolventeffectively increases the solubility of C60 to facilitate themeasurements of partition coefficients and eliminatethe inaccuracy from C60 lost to the vessels. Based on thethermodynamic correlation, the partition coefficient of C60

between soil organic carbon and pure water (Koc) wasestimated.

Materials and MethodsMaterials. Sublimed C60 (99.9%) was purchased from MERCorp. (Tucson, AZ). Ethanol, methanol, and toluene were ofHPLC grade or better and all chemicals were used as received.Water was purified with a Barnstead Nanopure system(Dubuque, IA).

Ethanol-Water Mixtures. To prepare C60 solutions atvarious ethanol-water ratios, C60 was first precipitated onthe walls of glass test tubes via solvent evaporation from atoluene solution. Ethanol and water were mixed to givesolutions at ethanol mole fractions (XEOH) of 0.42, 0.55, 0.64,0.74, 0.85, and 1.0. Each solution was adding to a tube containC60 precipitate and mixed on a horizontal shaker at low speedat 25 °C in a constant temperature room for g1 week. Toprevent cluster formation and to ensure complete dissolutionof the C60, the mass of C60 added to each tube was less thanor equal to the amount required to reach one-half thesolubility at each respective ethanol-water ratio. As aprecaution, after equilibration each solution was centrifugedat 5000 rpm (i.e., 2,380 g) for 30 min in a Sorvall SA 600 rotorto remove any residual suspended particles. The supernatantC60 concentrations were determined by HPLC. Direct UV/visanalyses of all ethanol-water mixtures showed the strongabsorption peaks of molecular C60 from 200-350 nm, whereas

* Corresponding author phone: (765) 494-2196; fax: (765) 496-1107; e-mail: jafvert@ecn.purdue.edu.

Environ. Sci. Technol. 2009, 43, 7370–7375

7370 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009 10.1021/es900989m CCC: $40.75 2009 American Chemical SocietyPublished on Web 08/25/2009

C60 aggregates (nC60) would display a red-shifted and broaderless intense peak (4). Additionally, the characteristic broadabsorption of nC60 between 400 and 500 nm is not observedfor these solutions (5). The resulting ethanol-water stocksolutions had initial C60 concentrations ranging from 0.01 to0.9 mg/L. These solutions were further diluted with the C60-free solvent mixtures at the same respective solvent molefraction ratios, prior to adding to soil.

Partition Coefficient Measurement. Table 1 lists somebasic properties of the soils including the cation exchangecapacity (CEC), organic carbon (OC) content, and the massratio of clay mineral to organic carbon (cm/oc). Thedistribution of C60 between soil and the ethanol-watermixtures were measured at 25 °C using two soils (3 and 15)at all the reported XEOH fractions, and for all soils at XEOH )0.85 and 1.0 (i.e., pure ethanol). The latter experiments wereperformed to investigate possible mineral contribution tosorption, as the group of soils spans a wide range in cm/oc.To construct sorption isotherms, the soil mass (g) to solutionvolume (mL) ratio of each ethanol-water and soil mixturewas adjusted until recovery of C60 in the liquid phase wasbetween 40 and 60%. Duplicate and sometimes triplicatesoil-solution mixtures at various C60 concentrations wereequilibrated in 15 mL screw-capped centrifuge tubes on anangular rotator for 72 h, as test samples equilibrated longerthat 72 h showed no additional sorption to the soil phase.At 72 h, samples were centrifuging at 5000 rpm for 30 minin a Sorvall SA 600 rotor, and the concentration of C60 in eachsupernatant was analyzed directly by HPLC. To determinepossible loss to the test tube walls, control tubes with no soilwere prepared, equilibrated, centrifuged, and analyzed alongwith sample tubes, and showed no significant loss occurredto the glass walls.

C60 Analysis. All stock ethanol-water mixtures and samplesupernatants were analyzed directly by reverse-phase HPLCusing a UV/vis detector set at 336 nm. Either a DiscoveryC-18 column (15 cm × 3 mm I. D., 5 µm particle size) witha mobile phase of methanol/toluene (60:40) or a Cosmosil5PBB column (25 cm × 4.6 mm I.D., 5 µm particle size) with100% toluene as the mobile phase was used. It should benoted that many HPLC detector flow-through cells are notcompatible with toluene as a mobile phase solvent.

Theory. Sorption equilibrium of an organic chemical isdefined as the state at which the sorbed and solution phasespecies have equal fugacities (12-14). In the followingderivation, a basic assumption is that chemical concentra-tions in all phases are dilute. Some parameters (e.g., activitycoefficients) were interpreted accordingly. Typically, thepartition coefficient, Kp, is defined by,

where the subscripts s and l denote the sorbed and liquidphases, respectively, and the asterisk signifies mutual

saturation of the phases. Cs* and Cl* denote sorbed and liquidphase concentrations, respectively, where Cl* conventionallyis referenced to aqueous or liquid phase volume and Cs*to the dry mass of the soil or sediment. Therefore, Kp generallyis reported with units of L/kg. �s* and �l* are fugacitycoefficients, which are equal to the product of the corre-sponding activity coefficient (γ*) and reference state fugacity.Vl and Vs are the molar volumes of liquid and sorbent phase,respectively; hence the ratio Vl/Vs is simply a unit adjustmentfactor converting the mole fraction-based partition coefficient(Kp’ ) X s*/X l* ) γl*/γs*) to the mass fraction-based partitioncoefficient, as the density of water ≈ 1.0 kg/L and theconcentrations are sufficiently dilute. For both phases, aconvenient reference state is the true liquid, or for solids thehypothetical subcooled liquid state of the compound forwhich the chemical activity is 1. Since the “liquid” chemicalat its solubility in water is in equilibrium with this referencestate, the corresponding aqueous activity coefficient isdefined by

where Xl,sat and Sl are the solubility or hypothetical subcooledliquid solubility of the compound in mole fraction and mol/Lunits, respectively. For solutions that are under-saturated (X< Xsat), and because we assume low chemical concentrationsin both phases, γ values are assumed constant at any chemicalconcentration for any combination of sorbent-liquid system(i.e., Henry’s constant domain). Combining eqs 1 and 2 yields

or:

For the case of ambient temperature solids like C60, a crystalenergy term that accounts for the energy required to meltthe compound can be used to convert the ambient tem-perature solubility of the solid to the subcooled liquidsolubility (15):

where Sl,c is the aqueous solubility of the crystalline material;f L and f c are the fugacities of the hypothetical liquid stateand pure crystalline state, respectively; ∆hfus and ∆Sfus arethe enthalpy of fusion and entropy of fusion, respectively;Tt is the triple-point temperature (K), and cp is the heatcapacity of the solid phase. Since, the first term on the right-hand side is the dominant term and the triple-point tem-perature is usually close to the normal melting temperaturein most cases, eq 6 generally is simplified to

ln f L /f c in eq 7 is generally referred to as the ‘crystal energyterm’ or Ec and was estimated to be 6.24 for C60 (16). Notehowever that here this term is defined as a positive (+)number to decrease any ambiguity regarding its sign.Equation 4 therefore can be expressed as:

TABLE 1. Physical and Chemical Properties of Soils (29)

particle size (%)

soil ID CEC (meq/100 g) sand silt clay OC (%) cm/oc

3 12.70 42.4 33.5 24.1 1.53 15.756 33.01 0.2 31.2 68.6 0.72 95.289 12.40 7.1 75.6 17.4 0.11 158.1810 14.58 6.3 71.9 21.8 2.1 10.3811 13.86 1.7 48.8 49.4 1.5 32.9315 11.30 15.6 48.7 35.7 0.95 37.5824 6.28 20.5 58.7 20.8 0.95 21.8925 8.86 41.9 37.6 20.5 0.76 26.97

Kp )Cs*

Cl*)

φl*

φs*)

Vlγl*

Vsγs*(1)

γl )1

Xl,sat) 1

SlVl(2)

Kp ) 1SlVsγs*

·γl*

γl(3)

log Kp ) -log Sl - log Vs - log γs* + log(γl*/γl) (4)

Sl ) Sl,c ·f L

f c(5)

lnf L

f c)

∆hfus

RTt(Tt

T- 1) -

∆cp

R (Tt

T- 1) +

∆cp

Rln

Tt

T(6)

lnf L

f c)

∆hfus

RTm(Tm

T- 1) )

∆Sfus

R (Tm

T- 1) (7)

VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 7371

In this study, partition coefficients were determined withmixtures containing significant ethanol. Previous workindicates that ethanol does not form solvates with C60 (16),suggesting constant Ec as a function of mole fraction ethanol.The term Vs can be viewed as the average molar volume ofthe sorbent phase and constant, however for dilute solutionsit is strictly a remnant of the mole fraction unit conventionand may be considered an intercept-correction term. Thisis especially true when the sorbent phase is a polymer ofequivocal molecular weight; and as we will show, calculationsperformed with weight fraction units are much less awkward(15). For example, if the sorbent phase is the organic carbonin the soil, Vs is the average molar volume of the organiccarbon, γs* is the activity coefficient of C60 in the organiccarbon phase, and eq 8 defines log Koc. Additionally, we willshow that for nonhydrogen bonding neutral chemicals, valuesof γs* for “dissolution” into soil organic carbon increase onlyby about an order of magnitude over an increase in aqueoussolubility of about 8 orders of magnitude. For Kp (or Koc)values measured with the same solute and soil with decreas-ing ethanol volume fraction, some change in γs* may result,however, these changes are likely small and proportional toSl. With increasing dissolved organic matter (DOM) contentand increasing solute hydrophobicity, the ratio γl*/γl decreasesdue to greater interactions of solute molecules with DOM.At the ethanol-water ratios employed in this study, any DOMeffect will be minimal, and so, γ

l* ≈ γl, resulting in good

correlation between log Kp and log Sl (or log Sl,c) valuescalculated at different XEOH values.

In a previous study, the solubility of C60 was measured inethanol-water mixtures at ethanol volume fractions from0.5 to 1.0, and an estimate of the pure aqueous solubility wasgiven by ref 11. Using Wohl’s equation, described in detailelsewhere (11, 16), to model the activity coefficient of C60 asa function of mole fraction composition, the reportedsolubilities were used to train Wohl’s model by adjusting thesize parameter for water (qwater ) 0.75) until accuratepredictions results, as shown by the line in Figure 1. Theresulting model was used to calculate the solubilities of C60

at each value of XEOH used in this study.

Results and DiscussionsSorption isotherms of C60 distributing between EOH-watermixtures and two different soils are shown in Figure 2. Kp

values, equal to the slopes of the isotherms, sharply increasewith decreasing XEOH from 1.0 to 0.4, from Kp ) 0.9-78.2, and2.4-129.4 (L/kg) for Soil 3 and 15, respectively. Similar trendsin Kp values of other hydrophobic compounds in water-miscible cosolvents mixtures have been reported in theliterature (17, 18).

Although partitioning to clay minerals and other soilconstituents may be important in some cases, it is widelyrecognized that at low swelling clay mineral to organic carbonratios, sorption to soil organic matter (as measured by organiccarbon content) dominates the overall phase distributionprocess (13), resulting in a nearly constant value for theorganic carbon content normalized partition coefficient, Koc,for any nonhydrogen bonding compound,

where foc is the mass fraction of organic carbon in the soil.Values of Koc for all sorption isotherms shown in Figure 2 arereported in Table 2 and show that Koc values for Soil 15 werehigher than those for Soil 3 at each respective XEOH. As a

result, sorption isotherms were measured for the six ad-ditional soils listed in Table 1 at XEOH ) 1 and 0.85 to test forpotential contribution to sorption by the swelling clay mineralcontent of Soil 15. Among these soils, the ratio of the massfraction of swelling clays in the soil, fcm (g/g) to mass fractionof organic carbon in the soil (foc) ranges from 10 to 158, withSoil 15 having fcm/foc ) 37.6. The observed sorption partitioncoefficient Kp,obs can thus be expressed as a sum of itscomponent contributions (13),

where Km is the partition coefficient for solute sorption tothe clay minerals (L/kg), and fa,cm and fa,oc are the active oravailable fraction (0 < fa < 1) of each respective phase. These

log Kp ) -log Sl,c -Ec

2.3026- log Vs - log γs* + log(γl*/γl)

(8)

Koc )Kp

foc(9)

FIGURE 1. C60 solubility as a function of mole fraction ethanol(XEOH) in ethanol-water mixtures with measured values fromreference (11) displayed as circles and Wohl’s model, withqwater ) 0.75, displayed as a line. q is “the effective volume” ofthe pure or hypothetical pure liquid of each respectivecomponent.

FIGURE 2. Sorption isotherms of C60 after 72 h of equilibrationfor (a) Soil 3 and (b) Soil 15 at different values of XEOH.

Kp,obs ) Kmfcmfa,cm + Kocfocfa,oc (10)

7372 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009

latter terms account for the fact that each phase may shieldor block available sorption sites on the other phase. Inparticular, sufficient soil organic carbon may coat some ofthe clay minerals preventing sorption of the chemical to asignificant fraction of the minerals (13). As a result, thedifferences between true Koc values that account for sorptiononly to OC and observed values, Koc.obs, calculated frommeasured Kp values assuming sorption occurs only to OC,generally become significant at higher fcm/foc ratios,

All values of the apparent Koc determined with eq 9 forall the soils at XEOH ) 1 and 0.85 are shown in Figure 3 andSupporting Information (SI) Figure S1, respectively, as afunction of fcm/foc. Especially at XEOH ) 1, results are quitesimilar to those found by Karickhoff et al. (13) who found anupper limit of fcm/foc ) 30, below which Koc,obs ) Koc for twocompounds with Koc valuese104. In our case, a threshold offcm/foc ≈ 20 is observed above which the Koc,obs values appearto be scattering, also similar to that reported by Karickhoffet al. Hence, the higher ratio of fcm/foc ()37.58) for Soil 15suggests the first term on the right-hand-side of eq 11 issignificant for this soil and accounts for the higher observedcarbon normalized partition coefficients reported in Table2, whereas partitioning to Soil 3 is dominated by sorption toorganic carbon.

Therefore, for Soil 3, eq 4 can be normalized to the organiccarbon content of the soil,

where variation in log Sl dominates any change in log Koc

(13). As we will show for a series of compounds, log γoc* slightlydecreases with increasing log Sl. Chiou et al. (19) have invoked

the Flory-Higgins polymer model to account for this changein log γoc* with the resulting change proportional to thelogarithm of the molar volume of each compound. In ourcase, the size of the solute molecule (C60) is constant, and sounless the activity of ethanol in the organic carbon phasesignificantly effects the solute-sorbent interactions, the valueof log γoc* should remain fairly constant and independent ofXEOH. As discussed above, Voc can be considered an interceptcorrection factor. For the values reported in Table 2,regression of log Koc,obs versus log Sl (eq 12) results in (logKoc.obs ) -1.10 × log Sl - 1.487) for Soil 3; and (log Koc,obs )-0.92 × log Sl - 0.171) for Soil 15. Because the slopes of bothregressions are ∼-1 as expected, recalculating the interceptswith the slopes set to -1 results in: (log Koc,obs ) -log Sl -1.101) for Soil 3; and (log Koc,obs ) -log Sl - 0.467) for Soil15. Again, the 0.63 log unit shift in the intercept for Soil 15is attributable to sorption to clay minerals. After adjustingthe log Koc,obs values for Soil 15 by this constant shift ()0.63),the regression of log Koc versus log Sl (with intercept)-1.101)for both soils is shown in Figure 4, were Sl are the subcooledliquid solubilities of C60 calculated at each experimental valueof XEOH (11). Because the aqueous solubility of C60 has beenreported at 7.96 ng/L (1.11 × 10-11 M) (11), the regressionequation can be used to extrapolate to XEOH ) 0, where thepartition coefficient is for the chemical distribution betweensoil and pure water, and is equal to (Koc )) 107.1 (L water/kgOC).

Because the extreme hydrophobicity of C60 makes thedirect experimental determination of Koc quite challenging,our value calculated by extrapolation can be compared toKoc values of other compounds through its octanol-waterpartition coefficient, Kow, and estimated aqueous solubility.Water solubilities and Kow and Koc values reported in theliterature for 28 other compounds are given in the SI. Tosimplify comparisons, subcooled liquid solubilities of ambi-ent temperature solids are used in all subsequent calculationsand figures. Additionally, since Voc is an ambiguous term forsoil organic carbon, and because chemical concentrationsin both phases are sufficiently dilute, activity coefficientscan be conveniently defined on a mass-fraction basis, wherethe water phase activity coefficient γl,m* equals the reciprocalmass fraction based solubility, Sl,m (g/g) (15, 20),

and

TABLE 2. Hypothetical Subcooled Liquid Solubilities (Sl) of C60and observed Organic Carbon-Normalized PartitionCoefficients (Koc,obs) for Two Soils

log Koc,obs

XEOH (mol/mol) VEOHa (L/L) log Sl (mol/L) Soil 3 Soil 15

1 1 -2.91 1.79 2.400.85 0.95 -3.20 1.97 2.850.74 0.90 -3.50 2.42 3.090.64 0.85 -3.81 2.60 3.380.55 0.80 -4.11 3.13 3.570.42 0.70 -4.71 3.71 4.13

a VEOH is the ethanol volume fraction.

FIGURE 3. Observed Koc values as a function cm/oc ratio atXEOH ) 1 for all soils reported in Table 1.

Koc,obs ) Km · fa,cm · (fcm/foc) + Koc (11)

log Koc ) -log Sl - log Voc - log γoc* + log(γl*/γl)(12)

FIGURE 4. Correlation between log Koc and the subcooled liquidsolubility of C60 (∆ Soil 15; 0 Soil 3, --- model line, 2 predictedKoc at the subcooled liquid solubility in pure water).

γl,m* ) 1Sl,m

(13)

Koc,m )C oc,m

∗

C l,m∗ )

γl,m∗

γoc,m∗ (14)

VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 7373

where concentrations in organic carbon and water (Coc,m*and Cw,m* , respectively) have units of g/g. With this unitconvention, the corresponding equation to eq 12 is

Since the density of water at 25 °C (Fw ) 0.998 g/mL) is nearunity, Koc,m ) Koc.; however, conventional volume-based Kow

values (M/M) are converted to Kow,m by multiplying by (Fw/Fo), where the density of octanol, Fo is 0.824 g/mL, and where

Since log Kow values are generally compared (not Kow values),this ≈20% correction on Fo results in log Kow,m ≈log Kow. Witheqs 13, 15, and 16, the unambiguous relationship betweenactivity coefficients for a chemical in water, soil organiccarbon, and octanol can be examined. Figure 5 shows themagnitude of these values regressed against log Kow,m (g/g)values for those chemicals listed in the SI, and for C60

determined in this study (γoc,m* ) and our previous study (γo,m

* ,γw,m

* ) (11). Figure 5 also shows log γo,m* and log γoc,m

* regressedagainst log Sl,m, for which,

Regressed directly or by combining eqs 15 and 18, theKoc,m - Sl,m relationship for these data is defined,

and because Fw ≈ 1.0 g/mL, Sl,m with units of g/mL producesKoc with conventional units of L/kg OC (i.e., Koc ) Koc,m).Previously, Karickhoff has suggested the relationship Koc ≈R ·Kow where the proportionality constant is R ) 0.63 (21) orR ) 0.411 (12). Taking the ratio of eq 17 to 18, where Rm )(γo,m* /γoc,m* ) ) (Koc,m/Kow,m), and solving at log Sl,m ) -3 and-5 (1 mg/L and 10 µg/L), results in Rm ) 0.569 and 0.352,

or with conventional units on Kow after correcting for thedensity of octanol,R)0.690 and 0.428, respectively, providingsupport for the notion that Koc ≈ 0.5 ·Kow for moderatelyhydrophobic compounds.

The value of Koc for C60 (107.1 L/kg) determined byextrapolating the line in Figure 4 to C60’s water solubilityvalue does not account for DOM effects in the absence of theethanol cosolvent. Solubility enhancement by DOM has beenaddressed by many authors and is often quantified byconsidering it a separate phase to which partitioning occurs(22, 23),

where the Koc is the partition coefficient measured in waterwith a specific DOM concentration (kg/ L), and KDOC is thepartition coefficient between DOC and pure water, andKoc,pure water is the Koc value in the absence of DOM. DOM inmost natural waters is usually below 60 mg/L (24), and log KDOC

values range between 4 and 6 for hexa- to octa-chlorodibenzo-p-dioxins (25) and is between 4 and 5 for p,p’-DDT (22). Thus,a maximum correction on our value of Koc for C60 can beestimated at ≈ 0.9 log units, (log KOC

C60 ) 6.2 to 7.1 in naturalwaters), placing it in good agreement with C60’s reported logKow of 6.67 (11).

Compared to many compounds of environmental con-cern, such as p,p’-DDT (26), PCBs (19), PAHs (21), and manypesticides (27), the organic carbon normalized partitioncoefficient of C60 is extremely high. Our estimated Koc valuefor C60 is of a similar magnitude to those of hexa- throughocta-chlorinated dibenzo-p-dioxins and dibenzofurans, whichalso have extremely low water solubilities (0.4-36.1 ng/L forhexa- to octa- chlorodibenzo-p-dioxins) (28).

AcknowledgmentsThe financial support by the National Science Foundationunder award EEC-0404006 is acknowledged. We thank Dr.Changhe Xiao for technical assistance.

Supporting Information AvailableAdditional information on data in Figures 3 and 5 is provided.This material is available free of charge via the Internet athttp://pubs.acs.org.

Literature Cited(1) Kroto, H. W.; Heath, J. R.; Obrien, S. C.; Curl, R. F.; Smalley, R. E.

C60 - Buckminsterfullerene. Nature. 1985, 318, 162–163.(2) Kadish, K., Ruoff R. Fullerenes: Chemistry, Physics, And Technol-

ogy; Wiley-Interscience: New York, 2000.(3) Utsunomiya, S.; Jensen, K. A.; Keeler, G. J.; Ewing, R. C. Uraninite

and fullerene in atmospheric particulates. Environ. Sci. Technol.2002, 36, 4943–4947.

(4) Deguchi, S.; Alargova, R. G.; Tsujii, K. Stable dispersions offullerenes, C60 and C70, in water. Preparation and characteriza-tion. Langmuir. 2001, 17, 6013–6017.

(5) Fortner, J. D.; Lyon, D. Y.; Sayes, C. M.; Boyd, A. M.; Falkner,J. C.; Hotze, E. M.; Alemany, L. B.; Tao, Y. J.; Guo, W.; Ausman,K. D.; Colvin, V. L.; Hughes, J. B. C60 in water: nanocrystalformation and microbial response. Environ. Sci. Technol. 2005,39, 4307–4316.

(6) Andrievsky, G. V.; Kosevich, M. V.; Vovk, O. M.; Shelkovsky,V. S.; Vashchenko, L. A. On the production of an aqueouscolloidal solution of fullerene. J. Chem. Soc. Chem. Commun.1995, 1281–1282.

(7) Brant, J.; Lecoanet, H.; Hotze, M.; Wiesner, M. Comparison ofelectrokinetic properties of colloidal fullerenes (nC60) formedusing two procedures. Environ. Sci. Technol. 2005, 39, 6343–6351.

(8) Lyon, D. Y.; Adams, L. K.; Falkner, J. C.; Alvarez, P. J. J.Antibacterial activity of fullerene water suspensions: effects ofpreparation method and particle size. Environ. Sci. Technol.2006, 40, 4360–4366.

FIGURE 5. Comparison of chemical activity coefficients inwater (0), octanol (O), and organic carbon (∆) phasesregressed against log Kow and (actual or subcooled) liquidsolubility. The solid symbols are values for C60 with all otherdata reported in SI Table S1 for the other 28 compounds.

log Koc,m ) -log Sl,m - log γoc,m∗ (15)

Kow,m )γw,m

∗

γo,m∗ (16)

log γo,m* ) -0.107log Sl,m + 0.104 (17)

log γoc,m* ) -0.211log Sl,m + 0.037 (18)

log Koc,m ) -0.79log Sl,m - 0.037 (19)

Koc ) Koc,purewater(1 + KDOC[DOM])-1 (20)

7374 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009

(9) Oberdorster, E. Manufactured nanomaterials (fullerenes, C60)induce oxidative stress in the brain of juvenile largemouth bass.Environ. Health Perspect. 2004, 112, 1058–1062.

(10) Henry, T. B.; Menn, F. M.; Fleming, J. T.; Wilgus, J.; Compton,R. N.; Sayler, G. S. Attributing effects of aqueous C60 nano-aggregates to tetrahydrofuran decomposition products in larvalzebrafish by assessment of gene expression. Environ. HealthPerspect. 2007, 115, 1059–1065.

(11) Jafvert, C. T.; Kulkarni, P. P. Buckminsterfullerene’s (C60) octanol-water partition coefficient (Kow) and aqueous solubility. Environ.Sci. Technol. 2008, 42, 5945–5950.

(12) Karickhoff, S. W. Semiempirical estimation of sorption ofhydrophobic pollutants on natural sediments and soils. Chemo-sphere. 1981, 10, 833–846.

(13) Karickhoff, S. W. Organic pollutant sorption in aquatic systems.J. Hydraul. Eng.-ASCE. 1984, 110, 707–735.

(14) Mackay, D. Finding Fugacity Feasible. Environ. Sci. Technol.1979, 13, 1218–1223.

(15) Prausnitz, J. M.; Lichtenthaler, R. N.; de Azevedo, E. G. Molecularthermodynamics of fluid-phase equilibria; Prentice-Hall: Engle-wood Cliffs, NJ, 1986.

(16) Kulkarni, P. P.; Jafvert, C. T. Solubility of C60 in solvent mixtures.Environ. Sci. Technol. 2008, 42, 845–851.

(17) Rao, P. S. C.; Hornsby, A. G.; Kilcrease, D. P.; Nkedikizza, P.Sorption and transport of hydrophobic organic-chemicals inaqueous and mixed-solvent systemssModel development andpreliminary evaluation. J. Environ. Qual. 1985, 14, 376–383.

(18) Walters, R. W.; Guiseppielle, A. Sorption of 2,3,7,8-tetrachlo-rodibenzo-para-dioxin to soils from water-methanol mixtures.Environ. Sci. Technol. 1988, 22, 819–825.

(19) Chiou, C. T.; Porter, P. E.; Schmedding, D. W. Partition equilibriaof non-ionic organic-compounds between soil organic-matterand water. Environ. Sci. Technol. 1983, 17, 227–231.

(20) Patterson, D.; Tewari, Y. B.; Schreibe, H. P.; Guillet, J. E.Application of gas-liquid chromatography to thermodynamicsof polymer solutions. Macromolecules. 1971, 4, 356–359.

(21) Karickhoff, S. W.; Brown, D. S.; Scott, T. A. Sorption ofhydrophobic pollutants on natural sediments. Water Res. 1979,13, 241–248.

(22) Chiou, C. T.; Malcolm, R. L.; Brinton, T. I.; Kile, D. E. Watersolubility enhancement of some organic pollutants and pes-ticides by dissolved humic and fulvic-acids. Environ. Sci. Technol.1986, 20, 502–508.

(23) Gschwend, P. M.; Wu, S. C. On the constancy of sediment waterpartition-coefficients of hydrophobic organic pollutants. En-viron. Sci. Technol. 1985, 19 (1), 90–96.

(24) Chiou, C. T. Partition and Adsorption of Organic Contaminantsin Environmental Systems; John Wiley & Sons, Inc.: New York,2002.

(25) Broman, D.; Naf, C.; Rolff, C.; Zebuhr, Y. Occurrence anddynamics of polychlorinated dibenzo-para-dioxins and diben-zofurans and polycyclic aromatic-hydrocarbons in the mixedsurface-layer of remote coastal and offshore waters of the baltic.Environ. Sci. Technol. 1991, 25, 1850–1864.

(26) Van den Hoop, M.; Kreule, P.; Loch, J. P. G. Sorption kineticsand transformation of DDT in sediment. Water Air Soil Pollut.1999, 110, 57–66.

(27) Delle, S. A. Factors Affecting sorption of organic compounds innatural sorbent/water systems and sorption coefficients forselected pollutants. A Review. J. Phys. Chem. Ref. Data. 2001,30, 187–439.

(28) Tanaka, F.; Fukushima, M.; Kikuchi, A.; Yabuta, H.; Ichikawa,H.; Tatsumi, K. Influence of chemical characteristics of humicsubstances on the partition coefficient of a chlorinated dioxin.Chemosphere. 2005, 58, 1319–1326.

(29) Means, J. C.; Hassett, J. J.; Barnwart, W. L.; Woods, S. G.; Khan,A. Sorption properties of sediments and energy-related pol-lutants, EPA 68-03-2555; U.S. Environmental Protection Agency:Washington, DC, 1978.

ES900989M

VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 7375