Environ. Sci. Technol. 2008, 42, 5945–5950
Buckminsterfullerene’s (C60) Octanol-Water Partition Coefficient (Kow) and Aqueous Solubility CHAD T. JAFVERT* AND PRADNYA P. KULKARNI Purdue University, School of Civil Engineering, West Lafayette, Indiana 47907
Received November 19, 2007. Revised manuscript received May 21, 2008. Accepted May 21, 2008.
To assess the risk and fate of fullerene C60 in the environment, its water solubility and partition coefficients in various systems are useful. In this study, the log Kow of C60 was measured to be 6.67, and the toluene-water partition coefficient was measured at log Ktw ) 8.44. From these values and the respective solubilities of C60 in water-saturated octanol and watersaturated toluene, C60’s aqueous solubility was calculated at 7.96ng/L(1.11×10-11 M)fortheorganicsolvent-saturatedaqueous phase. Additionally, the solubility of C60 was measured in mixtures of ethanol-water and tetrahydrofuran-water and modeled with Wohl’s equation to confirm the accuracy of the calculated solubility value. Results of a generator column experiment strongly support the hypothesis that clusters form at aqueous concentrations below or near this calculated solubility. The Kow value is compared to those of other hydrophobic organic compounds, and bioconcentration factors for C60 were estimated on the basis of Kow.
Introduction To understand the environmental fate of C60, data on potential exposure concentrations are required. In the environment, water is one of the major media in which transport and reaction occurs and therefore aqueous solubility is an important parameter for characterizing environmental fate of any chemical. An experimentally determined aqueous solubility of C60 has not been reported previously. Another important property and potentially confounding issue is the formation of C60 clusters in water. Researchers have reported formation of clusters when C60 was introduced to water dissolved in a water miscible organic solvent, followed by removal of the organic solvent. The C60 clusters in water have been formed using organic solvents, such as toluene, tetrahydrofuran, or a mixture of toluene, tetrahydrofuran, and acetone (1–5). Alternatively, clusters can be formed via mixing of solid C60 with water for several days (4). Because C60 is expected to be extremely insoluble in water as discretely hydrated molecules, various methods have been employed to increase the concentration of C60 in water and include formation of host-guest complexes with γ-cyclodextrin, dispersion with surfactants solutions, and chemical modification or functionalization resulting in a new chemical species (6–8). In this study, three different experimental approaches were employed in an attempt to determine the solubility of * Corresponding author telephone: (765) 494-2196; fax: (765) 4961107; e-mail:
[email protected]. 10.1021/es702809a CCC: $40.75
Published on Web 07/02/2008
2008 American Chemical Society
C60 in water. The three different methods were (1) a partition coefficient method, (2) a cosolvent method, and (3) a generator column method. In the first method, the aqueous solubility was estimated from partitioning of C60 between water and the water immiscible solvents of toluene and octanol. Although the mutual saturation of the phases may increase the concentration of C60 in the water, the data provide a useful reference for comparison to other measurements. Moreover, the octanol-water partition coefficient (Kow) is used widely to predict partitioning of organic contaminants between aqueous and natural organic phases and to estimate bioaccumulation (9). In the second approach, the solubility of C60 was measured in mixtures of water and a water miscible organic solvent. The data were modeled with Wohl’s equation to evaluate the aqueous solubility estimate obtained via the partition coefficient method. In the third approach, we attempted to measure the solubility of C60 in water directly using a generator column technique. The generator column method has been used widely for determination of the aqueous solubility of sparingly soluble compounds like PAHs and PCBs (10).
Materials and Methods Sublimed C60 (99.9%) was purchased from MER Corp., Tucson, AZ. All solvents were HPLC grade or better and included distilled octanol, water-saturated octanol, and water-saturated toluene prepared in the laboratory. To prepare the distilled octanol, octanol was extracted with 0.01 M NaOH (once) and distilled water (twice) prior to distillation. In the partition coefficient method, the distribution of C60 between octanol (OCT) or toluene (TOL) and water was measured. For this, 3.5 L volumes of water were equilibrated with 5-10 mL solutions of C60 in octanol or toluene in 4 L glass bottles. All phases were mutually saturated before equilibration with C60. The solutions were stirred with a glass stirring bar at low speed (∼50-60 rpm) in a constant temperature room at 25 ( 1 °C. The concentration in each phase was measured after 4-13 days, sacrificing the contents of an entire bottle for the measurement. Approximately 2.5 L of the water phase was extracted with 8-10 mL of hexane, and 5 mL of hexane was analyzed by HPLC on a Cosmosil 5PBB column with 100% toluene as the mobile phase with the detector wavelength at 336 nm. The same HPLC method was used to measure the concentration in each organic phase with 50 µL sample volumes. In the cosolvent method, solubility of C60 was measured in aqueous mixtures of tetrahydrofuran (THF) or ethanol (EOH) at organic cosolvent volume fractions ranging from 0.5 to 1.0. The water-cosolvent mixtures were equilibrated with C60 that was precipitated on the walls of tubes from a toluene solution via solvent evaporation. The samples were equilibrated on a wrist-action shaker for 5-7 days at 25 ( 1 °C. After equilibration, the samples were centrifuged at 4000 rpm for 90 min in a Sorvall SA 600 rotor. The supernatants were analyzed by HPLC on a Supelcosil C-18 column with a mobile phase of THF-acetonitrile (75:25) at a detector wavelength at 336 nm. In the third approach, the generator column method, a stainless steel preparatory HPLC column (25 cm × 2.12 cm i.d.) was filled with 60-80 mesh glass beads coated with C60. The glass beads (Sigma-Aldrich) were coated by addition of C60 dissolved in toluene to the beads and evaporation of the toluene. The porosity of the column was ∼0.40. Distilled water was passed through the column at 0.1, 0.25, 0.5, and 1 mL/ min, resulting in mean hydraulic retention times of 353, 141, VOL. 42, NO. 16, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5945
TABLE 1. Solubility of C60 in Various Solvents solvent
measured solubility (mg/L)
ethanol octanol distilled octanol water-saturated octanol tetrahydrofuran toluene water-saturated toluene
1.4 42.9 38.9 34.7 11 3,000 2,430
a
Ref 11.
b
Ref 12. c Ref 1.
d
reported solubility (mg/L) 0.8a, 1b 47a 0b, 9c, 37d 2,800b
Ref 13.
70, and 35 min, respectively. C60 in the effluent was extracted on an in-line C-18 guard column. For each sample, the guard column was analyzed by HPLC and a Supelcosil C-18 analytical column. A gradient elution was used with 100% methanol for 3 min, followed by 50:50 methanol-toluene for 30 min, 100% toluene for 30 min, and 100% methanol for 30 min. C60 was detected at a wavelength of 336 nm.
Solubilities of C60 measured in a variety of pure solvents at 25 °C are given in Table 1 (1, 11–13). Our measurements are in good agreement with the values previously reported in the literature. The solubility in water-saturated toluene and water-saturated octanol were measured to be 2.43 g/L and 34.7 mg/L, respectively. The solubility of water in octanol has been reported to be 2.3 M (mole fraction ) 0.27) by Leo et al. (14). The presence of water at its saturation level in octanol decreased C60’s solubility from 38.9 to 34.7 mg/L. This 10.8% decrease appears reasonable, although somewhat low, based on decreases of 23% and 18% for p,p′-DDT and hexachlorobenzene at approximately the same temperature (15). Table 2 shows the results of the toluene-water and octanol-water partition experiments. From the measured concentrations of C60 in the aqueous phases (Cw) and in each respective organic layer of toluene or octanol (Ct or Co, respectively), the organic solvent-water partition coefficient (Ktw or Kow) was calculated. The organic solvent-water partition coefficient for each respective solvent was approximately the same for equilibration times g6 days, with an average log Ktw ) 8.44 and an average log Kow ) 6.67, for all data collected after six or more days of equilibration. The solubility of C60 in water can be calculated from these data via eq 1 Kiw )
Ci/
( ) ( ( A21 ′ A12 ′
/ Cw
ln γ1 )
2
+ x23A13 ′
( ) A31 ′ A13 ′
2
+ x2x3
A13 A31 ′ A21 ′ A31 ′ 2 x1 + x 2 + x3 A12 ′ A13 ′
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 16, 2008
)
)
A21 ′ A31 ′ × A12 ′ A13 ′ (2)
where A12′ ) 2q1a12, A21′ ) 2q2a12, A13′ ) 2q1a13, A31′ ) 2q3a13, A23′ ) 2q2a23, A32′ ) 2q3a23. In eq 2, xi refers to the mole fraction of the solute, C60 (i )1), and the two solvents (i ) 2, 3). The values of qi, defining i as above, represent “effective volumes” and often are assumed to equal the surface area of the molecule or the molar volumes of the pure liquids or hypothetical pure liquid of each respective component. The aij terms are the interaction parameters between paired components i and j, and can be obtained by applying Wohl’s equation to the activity data for the respective binary solutions. Since the solubility of C60 in the mixtures is a small value, x1 in the denominator can be ignored. This results in an explicit function (eq 3) through which the activity coefficient of C60 may be calculated from known interaction and size parameters. x22A12 ′
( ) ( A21 ′ A12 ′
2
+ x23A13 ′
( ) A31 ′ A13 ′
2
A12 ′ + A13 ′ -A32 ′
ln γ1 )
(1)
where Kiw is the partition coefficient between organic solvent i and water, Ci/ is the solubility of C60 in the water-saturated organic solvent i, and Cw/ is the solubility in the organicsolvent saturated water phase. From the solubility in watersaturated toluene of 2.43 g/L and Ktw of 108.44, the solubility of C60 in toluene-saturated water is calculated to be 8.49 × 10-6 mg/L. Similarly, the solubility in water-saturated octanol of 34.7 mg/L and Kow of 106.67 gives a solubility of C60 in octanol-saturated water to be 7.42 × 10-6 mg/L. Although the presence of the organic solvent in water may enhance the aqueous solubility of C60, the volume percentage of the solvent in water is very low, that is, at saturation the volume percentage of toluene in water is 0.061%, while that of octanol in water is 0.071%. Although, the solubility of C60 in pure water should be slightly less than the solubility in octanolsaturated water, the solubility in each organic-solvent saturated water phase is a reasonable estimate of the pure aqueous phase solubility of C60 in equilibrium with crystalline 9
x22A12 ′
A12 ′ + A13 ′ -A32 ′
Results and Discussion
5946
material. The log Kow (6.67) for C60 is higher than that of p,p′-DDT (log Kow ) 6.36) indicating a very high affinity for biological lipids and soil organic carbon (15). Results of the second approach are shown in Figures 1 and 2, where the solubility measurements in the mixtures of ethanol-water and tetrahydrofuran-water are reported, respectively. The solubility in these mixtures decreases with decreasing organic solvent fraction. The solubility data were modeled with Wohl’s equation, and the model is represented as the lines on Figures 1 and 2. The pure aqueous phase solubility was set to be equal to the average of the values obtained from the octanol-water and toluene-water partitioning data of 7.96 × 10-6 mg/L. Wohl’s model, described in detail elsewhere (16, 17), estimates the activity coefficient of a component (i.e., unique chemical species) (γi) on the basis of the composition of the system, size of the components, and interaction parameters between the components, with the specific form of Wohl’s model used herein provided as eq 2
(
A21 ′ A31 ′ x2 + x3 A12 ′ A13 ′
+ x2x3 A13 ′ A31 ′
)
2
)
A21 ′ A31 ′ × A12 ′ A13 ′ (3)
For real solutions, the solubility of a solid solute at the temperature of consideration is a function of the solute’s activity coefficient (γ) and its entropy of fusion (∆Sfus) or enthalpy of fusion (∆Hfus). When the i (1) subscript is dropped for clarity, lnS ) -lnγ +
(
)
(
)
Tm ∆Hfus 1 ∆Sfus 1 1) -ln γ + (4) R T R Tm T
where S is the mole fraction solubility of the chemical, Tm is the melting point, T is the temperature at which the solubility is measured, and R is the universal gas constant. The second term on the right-hand side of eq 4 referred to as the “crystal energy term” or Ec accounts for the energy required to melt the chemical. The heat capacity differences between the solid and the supercooled liquid solute are usually ignored in calculation of Ec (17). Since the melting point and enthalpy of fusion of C60 at atmospheric pressures are unknown, the Ec of C60 can not be calculated. In a recent paper, we estimated the value of Ec of unsolvated C60 to be
TABLE 2. Organic Solvent-Water Partitioning of C60 expt no.
equilibration days
Cw (mg/L)
log Cw
Ct (or Co)(mg/L)
log Ktw (or log Kow)
1 2 3 4 5
4 4 6 10 11
toluene-water 0.80 × 10-6 -6.10 1.57 × 10-6 -5.80 1.24 × 10-6 -5.91 3.49 × 10-6 -5.46 1.24 × 10-6 -5.91
978.33 992.94 441.04 715.35 367.08
9.08 8.80 8.55 8.31 8.47
1 2 3 4
7 12 12 13
octanol-water 3.49 × 10-6 -5.46 2.84 × 10-6 -5.54 3.18 × 10-6 -5.49 3.45 × 10-6 -5.46
12.15 16.43 14.64 19.71
6.54 6.76 6.66 6.75
-6.24 (16). We also showed that C60 forms solvates in THF, and the estimated Ec of the solvated C60 in THF is -7.8 (16). The C60 solid phase in equilibrium with EOH exists as unsolvated crystals, and thus to model the solubility in EOH-water systems, Ec equals -6.24 over all ethanol mole fraction values. The activity coefficient of C60 in each pure solvent was calculated via eq 4 from the solubility of C60 in the pure solvents and the estimated crystal energy (Ec ) -6.24 or -7.8). Next, the interaction parameter between C60 and each pure solvent (a12 and a13) can be calculated with eq 5 assuming that only binary interactions occur in the pure solvents. ln γ1 )
2q1a1j q1x1 1+ qjxj
(
)
2
(5)
The values of these interaction parameters, a1j, for C60-ethanol, C60-tetrahydrofuran, and C60-water are 0.95, 0.57, and 2.30, respectively. The interaction parameters between EOH-water and THF-water (a23 ) 0.45 and 1.78, respectively) were calculated by applying Wohl’s model (eq 6) to literature data (18, 19) on excess free energy values for each respective binary system, assuming that higher-order interactions are negligible gE 2a23x2x3q2q3 ) RT x2q2 + x3q3
(6)
For each binary system, the a23 parameter was adjusted to minimize the squared residuals between calculated (eq 6) and reported values (18, 19). The values for the size parameters, qi, used in this study for EOH, THF, and C60 (1.4, 1, and 5, respectively) were those obtained in our previous study (16). The size parameter for water (q ) 0.6) was adjusted to minimize the sum of squared residuals between Wohl’s model-calculated C60 solubilities and those measured in both the EOH-water and THF-water systems; hence this coefficient was used as the only adjustable parameter to fit the model to both data sets. With the size and interaction parameters and the solubility of C60 in the solvent mixtures can be calculated with eqs 3 and 4. The estimated solubility values are shown as lines on Figures 1 and 2. For the EOH-water system, C60’s solubility conforms to a continuous function. With q ) 0.6, the calculated solubilities in the EOH-water system are slightly overpredicted; however a size parameter of 0.75 results in excellent correlation between measured and predicted values as shown by the dotted line in Figure 1. These values (0.6-0.75) are consistent with the other size parameters, and this value generally agrees with the correlation between molecular surface area and size parameter. AsshownpreviouslyforthesolubilityofC60 inTHF-acetonitrile and THF-toluene mixtures, the solubility of C60 in THF-water
should have a discontinuity on the solubility curve because of solvate formation with THF, assuming that no such solvates form with water or that the crystal energy of such solvated crystals is not the same as that for THF-C60 crystals. In our previous study (16), we showed that in the THF-acetonitrile and THF-toluene systems, the discontinuity occurs where the activity of THF in the solution is ∼0.17. In the THF-acetonitrile and THF-toluene systems, the activity of THF is 0.17 at XTHF ≈ 0.1 and 0.2, respectively. In the THF-water system the activity of THF is close to 0.17 at XTHF ≈ 0.17. Thus, at XTHF > 0.17, the solid -phase C60 in equilibrium with THF-water mixtures is predicted to be THF-solvated crystals. Hence, to plot the predicted solubilities in THF-water, again using q ) 0.6 as the only adjusted (i.e., calculated) parameter, the value of the crystal energy term, Ec, was set to -7.8 at XTHF > 0.17 and to -6.24 at XTHF < 0.17. Figure 3 shows the generator column effluent concentrations of C60 at each point in time when the guard column was analyzed. After the first 20 days, at a flow rate of 0.5 mL/min, the effluent reached an apparent steady-state concentration of 10-6 mg/L (1 ng/L). The initial drop in the concentration is likely attributable to weakly attached material being washed out in the water. At 0.25 and 0.1 mL/min flow rates, the concentration in the effluent water increased to about 10-4.5 mg/L, suggesting saturation was not achieved at the higher flow rate. However, upon returning the flow rate to 0.5 mL/ min, the concentration reached ∼10-4 mg/L, and after the flow rate was increased to 1 mL/min, the C60 effluent concentration increased another log unit. A likely reason for these increases may be the formation of C60 clusters. Because of the relatively low C60 effluent concentrations, the presence of clusters could not be confirmed by dynamic light scattering analysis. However, the temporal trend of increasing C60 in the effluent of the generator column mimics a broad breakthrough curve, with C60 clusters being continuously formed, adsorbing to the coated glass beads, detaching, and readsorbing as they grow in size and move through the column. With C60 as the solid surface on the beads, sorption should significantly increase cluster retention time, especially for small clusters with minimal surface charge. Another factor that may contribute to the observed trend is the increased pressure in the generator column because of back pressure generated by the guard column, and potential solvation of the C60 crystals with water overtime because of the higher pressure. Sawamura and Fujita (20) have reported large increases in the solubility of C60 in hexane with increasing pressure up to 400 MPa. Although the back pressure exerted by the guard column in these studies were not as high (∼ 30 atm or 3 MPa), over time it may lead to the inclusion of water molecules in the C60 crystal lattice structure, producing solvated crystals that have higher solubility than the pure crystals, although this does not explain the stability of the effluent solutions. VOL. 42, NO. 16, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5947
FIGURE 1. Solubility in ethanol-water mixtures (s Wohl’s model with qwater ) 0.6, --- Wohl’s model with qwater ) 0.75).
FIGURE 2. Solubility in tetrahydrofuran-water mixtures (s Wohl’s model, --- transition region).
FIGURE 3. Concentration of C60 in the generator column effluent (] 0.5, 9 0.25, 20.1, ] 0.5, and b1 mL/min). To examine the stability of the effluent solutions, a 500 mL effluent sample at the flow rate of 1 mL/min was collected in a volumetric flask, bypassing the guard column, and was stored quiescently for two weeks in the dark at room temperature. A 100 mL aliquot of this sample was extracted with toluene and was analyzed by HPLC. Other studies on C60 aqueous clusters have showed that to efficiently extract C60-clusters with an organic solvent, a destabilizer such as magnesium perchlorate generally is required (2). However, no such destabilizer was required to efficiently extract the C60 from the generator column effluent. The concentration of C60 in the extracted sample was the same (i.e., 1.0 µg/L) 5948
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 16, 2008
FIGURE 4. Comparison of supercooled aqueous solubility and Kow of C60 with other substituted aromatic compounds (s ideal line (log Kow° ) -log Sw + 0.92), 2 C60, 9 p,p′-DDT). All data, except C60, are from ref 15. as that measured using the guard column as an in-line extractor. In addition, adsorption of the C60 to the walls of the volumetric flask was not observed. A possible explanation is that very small C60 clusters (i.e.,
