溶剂溶解度参数

J. Phys. Chem. 1993, 97, 3379-3383 3379 Solubility of Ca in a Variety of Solvents R. S. Ruoff,’ Doris S. Tse, Ripudaman Malhotra,’ and Donald C. Lorents Molecular Physics Laboratory, SRI International, Menlo Park, California 94025 Received: November 10, 1992; In Final Form: January 7 , 1993 The room temperature solubility of pure c 6 0 has been determined in 47 solvents. The solubilities cover a wide range, from 0.01 mg/mL in methanol to 50 mg/mbin 1-chloronaphthalene. The solubilities in CS2, toluene, and hexane, three of the commonly employed solvents, are 7.9,2.8, and 0.04 mg/mL, respectively. An examination of the solubilities of c 6 0 as a function of the solvent properties such as index of refraction, dielectric constant, molecular size, Hildebrand solubility parameter, and H-bonding strength reaffirms the century-old principle ‘like dissolves like”. No single solvent parameter can uniformly predict the solubility of C ~ O , but a composite picture of solvents with high solubility for C ~ O emerges: large index of refraction, dielectric constant around 4, large molecular volume, Hildebrand solubility parameter equal to 10 and tendency to act as a moderate strength nucleophile. Introduction The solubility of c 6 0 and other fullerenes is of both practical and fundamental interest. Extraction and purification of solutes by solvents are significant and costly processes in the production of chemicals and can constitute on the order of 50% of the overall cost of production.’ Extraction and chromatographic techniques currently employed for the isolation of C60,2-5 and of other pure fullerenes, are much too expensive to allow large scale production and utilization by thechemical industry. A deeper understanding of the interaction of fullerenes with various solvents will alrow rational choice of solvents capable of purifying the fullerenes by cheaper and scalable methods. Solubility data can also play a useful role in the choice of stationary phase and type of chromatography employed. The fullerenes, with their unique cage structures, will interact with solvents in interesting ways that provide new information on the mechanisms of solute-solvent interactions. The fullerenes have rigid, well-defined geometries, in contrast to other solutes whose shapes undergo conformational changes and whose intramolecular vibrational partition functions may undergo large-and solvent dependent-changes. A zero-order picture of the interaction afforded by a fullerene solute with a solvent would include two aspects of the geometry of the fullerene: the molecular surface area and the molecular volume. The molecular volume is the volume of the cavity created in the solvent liquid, and the molecular surface area is the surface available for interaction with the solvent. Adams and Ruoff have recently derived a method for calculation of the van der Waals surface and volume of any fullerene.6 The solubilities of fullerenes will be determined not only by these geometric factors but also by other factors involving the specific molecular interactions between the fullerene and the solvent molecules. The possible formation of solid solutions of the solvent with c 6 0 may also play a key role. These questions can now be addressed for c 6 0 and in the future will be addressable for larger fullerenes as they become available in substantial quantities. To develop an understanding of these interactions for C60, we report here measurements of the room temperature solubility of c 6 0 in a range of solvents. The solubilities of C ~ O and higher fullerenes will be reported later. Experimental Section Pure Cm was obtained by using column chromatography with neutral alumina to purify ‘extract” obtained from primary arc- * Author to whom correspondence should be addressed. 0022-3654/93/2091-3319S04.00/0 synthesized soot. It was dried at 350 K in a vacuum oven for 24 h, and its purity was determined by high-performance liquid chromatography (HPLC) and surface analysis by laser ionization (SALI) mass spectrometry to be 99.95%. Solvents used were generally of reagent grade (99% or better) and were used as received. Those of less than 99% purity are as follows: chloroform contained 1% ethanol stabilizer; nitromethane was 98.1% pure; pyridine was Baxter reagent grade; cyclopentane contained 98% pentanes, with at least 75% cyclopentane; 2-methylthiophene was 98% pure; 1,1,2-trichlorotrifluoroethane was industrial grade from Blaco-tron; 1-chloronaphthalene contained about 10% of the 2-chloro isomer; 1-phenylnaphthalene contained about 5% of the 2-phenyl isomer; tetralin was freshly distilled. The xylenes sample was a 22:63:15 mixture of the ortho, meta, and para isomers, and the decalins contained cis and trans isomers, in a 3:7 ratio. A small quantity of c 6 0 was added to each solvent to obtain an indication of how much would be required to ensure that extra solid would be present. If a strong color developed following a brief sonication, about 100 mg of c 6 0 would be placed in a 1 -dram bottle and 1-2 mL of solvent added. If only a faint color developed, only about 20mg of Cm was used.’ A 3/~-in. Teflon-coated stirring bar was used to agitate the solutions in the dark for periods of not less than 24 h to ensure equilibration. The laboratory temperature varied between 293 and 298 K during this period. As a further test of attainment of equilibrium, the solubility of Cm in two commonly employed solvents, CS2 and toluene, was measured after 24,48, and 168 h. The solubility in each solvent at each of these time periods was found to be the same within experimental error. For several weak solvents, additional mixing was effected by sonicating the mixtures in an ultrasound cleaning bath for 5 min, followed by another 24-h stirring period. We did not observe any change in the measured solubilities as a result of the additional stirring, and hence the initial 24-h stirring was considered sufficient for equilibration. The saturated solution was filtered through a 0.45-~m poly(tetrafluoroethy1ene) (PTFE) filter (Gelman Acrodisc); a clean glass syringe was used to transfer the solution. The filtrate bottle was quickly closed with a Teflon- lined screw cap. Care was taken to ensure that solid residue was left in thebottletoserveasproofthat thesolutionwasnot deficient in c60(s) prior to filtering. We initially intended to measure solubilities spectrophoto- metrically. However, we observed that the wavelength of the maximum of each Cm absorbance peak is significantly solvent- dependent. For example, the ‘328-nm” peak occurs at 328 nm in hexane, but at 335 nm in benzene. Also, many of the solvents strongly absorb in the 328-nm spectral region, such as those Q 1993 American Chemical Society 3380 The Journal of Physical Chemistry, Vol. 97, No. 13, 1993 TABLE I: Solubility of Ca in Various Solvents Ruoff et al. solvent [C,], mg/mL mole fraction (XI@) n t V, cm3 mol-' 6, ~ m - ' / ~ alkanes n-pentane cyclopentane n-hexane cyclohexane n-decane decalins cis-decalin rruns-decalin haloalkanes dichloromethane chloroform carbon tetrachloride 1,2-dibromoethane trichloroethylene tetrachloroethylene Freon T F (dichlorodifluoroethane) 1 , I ,2-trichlorotrifluoroethane I , I ,2,2-tetrachloroethane methanol ethanol nitromethane nitroethane acetone acetonitrile N-methyl-2-pyrrolidone benzene toluene xylenes mesitylene tetralin o-cresol benzonitrile fluorobenzene nitrobenzene bromobenzene anisole chlorobenzene 1,2-dichlorobenzene 1,2,4-trichlorobenzene I-methylnaphthalene dimethylnaphthalenes 1-phenylnaphthalene 1 -chloronaphthalene carbon disulfide tetrahydrofuran tetrahydrothiophene 2-methylthiophene pyridine See ref 18. polars benzenes naphthalenes miscellaneous 0.005 0.002 0.043 0.036 0.07 1 4.6 2.2 1.3 0.26 0.16 0.32 0.50 1.4 1.2 0.020 0.0 14 5.3 0.000 0.001 0.000 0.002 0.001 0.000 0.89 1.7 2.8 5.2 1.5 0.0 14 0.4 1 0.59 0.80 3.3 5.6 7 .O 8.5 16 27" 33 36 50 51 7.9 O.OO0 0.030 6.8 0.89 containing nitro groups, CS2, and several of the halogenated solvents. Consequently, we decided to use calibrated HPLC, a method free of such solvent effects. The [CbO] in the saturated filtered solution was analyzed by HPLC using a Waters System 6000 chromatograph equipped with a dinitroanilinopropyl (DNAP) column (9 mm X 250 mm; E. S. Industries), an ultraviolet (UV) detector set at 340 nm, and a Hewlett-Packard Model 3390 integrator. A solution of toluene (20% by volume) in hexane was used as the eluant. Detector response for [ C ~ O ] was calibrated using several standard solutions in the range 0.01 to 0.20 mg/mL. If needed, the saturated solutions were diluted to bring the [c60] within the calibration range. Typically, several dilutions of each solvent were used, and measured solubilities were within 10% of each other. The average values computed are reported below. Results The measured solubilities are reported (in mg/mL and in mole fraction) in Table I, which is organized by solvent type. Several 0.008 0.003 0.073 0.059 0.19 9.8 4.6 2.9 0.27 0.22 0.40 0.60 1.7 1.7 0.042 0.017 7.7 O.OO0 0.001 O.Oo0 0.002 0.001 O.Oo0 1.2 2.1 4.0 8.9 3.1 0.029 0.71 0.78 1.1 4.8 8.4 9.9 31 53 15 68 78 131 97 6.6 O.OO0 0.036 9.1 0.99 1.36 1.41 1.38 1.43 1.41 1.48 1.48 1.47 1.42 1.45 1.46 1.54 1.48 1.51 1.36 1.44 1.49 1.33 1.36 1.38 1.39 1.36 1.34 1.47 1 .so 1 s o 1 .so 1.50 1.54 1.54 1.53 1.47 1.56 1.56 1.52 1.52 1.55 1.57 1.62 1.61 1.67 1.63 1.63 1.41 1 .so 1.52 1.51 1.84 1.97 1.89 2.02 1.99 2.20 - - 9.08 4.81 2.24 4.79 3.40 2.46 - - 8.20 33.62 24.30 35.90 28.00 20.70 37.50 - 2.28 2.44 2.40 2.28 2.76 1 1 .so 25.60 5.42 35.74 5.40 4.33 5.71 9.93 3.95 2.92 2.90 2.50 5.00 2.64 7.60 2.28 2.26 12.30 115 7.0 93 8.6 131 7.3 108 8.2 195 8.0 154 8.8 154 8.8 158 8.6 60 9.7 86 9.3 80 8.6 72 10.4 89 9.2 102 9.3 188 - 118 - 64 9.7 41 14.5 59 12.7 81 12.7 105 11.1 90 9.8 52 11.8 96 11.3 89 9.2 106 8.9 123 8.8 139 8.8 136 9.0 103 10.7 97 8.4 94 9.0 103 10.0 105 9.5 109 9.5 102 9.2 113 10.0 125 9.3 142 9.9 156 9.9 155 10.0 136 9.8 54 10.0 81 9.1 88 9.5 96 9.6 80 10.7 trends are apparent. c 6 0 is essentially insoluble in polar and H-bonding solvents like acetone, tetrahydrofuran, acetonitrile, nitromethane, methanol, and ethanol. It is sparingly soluble in alkanes like pentane, hexane, and decane, with the solubility increasing with the number of carbons. The solubility in cyclopentane and in cyclohexane is also very low. In view of the rather poor solubility in alkanes and cycloalkanes, the solubility of 4.7 mg/mL in the 3:7 mixture of cis- and trans-decalins is truly remarkable. Replicate measurements were conducted to confirm the relatively high solubility. To test if the curved cavity of the cis decalin was particularly responsible for the high solubility, we determined the solubility of c60 in the cis- and trans-decalins separately. Interestingly, while the solubility of c 6 0 is significantly higher in the cis isomer, it is still higher in the mixture than in either of the pure isomers. The solubility in chloroalkanes (with the exception of the Freons) is generally higher than in alkanes, although it does not increase monotonically through the series CH2C12, CHCI3, and CC14 (the values were 0.26,0.16, and 0.32 mg/mL, respectively). Solubility of c 6 0 in a Variety of Solvents Among the C-2 chlorocarbons, the solubility in the ethylene derivatives is less than in the ethane derivative: solubility in trichloroethylene and tetrachloroethylene is a little over 1 mg/ mL, but in 1,1,2,2-tetrachloroethane it is 5.3 mg/mL, the highest among this group of solvents. C~isappreciablysolubleinaromaticsolvents (with thenotable exception of the H-bonding o-cresol). Substitution with electron- donating groups such as methyl and methoxy groups increases the solubility, while substitution with electron-withdrawing groups such as nitro and nitrile reduces it. Solubility increases in going from benzene to toluene and from toluene to xylenes, but introduction of the third methyl group (mesitylene) results in decreased solubility. Solubility of C a in mesitylene is even less than in benzene. Substitution with chlorineand bromineincreases solubility, while that with fluorine decreases it. Introduction of a second chlorine results in a substantial increase in solubility. Indeed, of the one-ring aromatic solvents tested here, o-dichlo- robenzene has the highest solubility for Cb0; the solubility approaches that achieved with naphthalene derivatives. However, as with the methyl substitutions, introduction of a third chlorine (1,2,4-trichlorobenzene) results in a sharp decrease in solubility. The solubility in 2-methylthiophene, a heteroaromatic, is comparable to that in chlorobenzene. However, in pyridine, also a heteroaromatic, it is substantially less. In the nonaromatic heterocycles (tetrahydrothiophene and tetrahydrofuran), the solubility is markedly low. Among the nonaromatic solvents, the solubility is highest is carbon disulfide. Increasing the size of the aromatic system (benzene to naphthalene) results in increased solubility. The relative effect of methyl and chloro substitution on naphthalene is similar to that on benzene. In accord with the general expectation, substitution with a phenyl has an effect similar to substitution with a chlorine. The *champion" solvent appears to be l-chlo- ronaphthalene, but when solubility is expressed as mole fraction, [c,] in 1-phenylnaphthalene is slightly higher. Solubility is influenced by several solvent properties. The H-bonding character of the solvent is an obvious discriminating factor. c60 will have a lower solubility in solvents that organize themselves through polar or H-bond interactions because of the disruption in the solvent structure that would result from dissolution of c60, which is nonpolar and does not participate in H bonds. Many solvent parameters have been developed for a quantitative prediction of solubility of various solutes. Reichardt8 has discussed several quantitative treatments of solvent param- eters, and the reader is directed to this reference for a detailed discussion. Solvent properties that might be expected to influence the solubility of C60 are polarizability, polarity, molecular size, and cohesive energy density. To investigate the influence of these properties, we chose the following solvent parameters: (nZ - 1)/ (n2 + 2), where n is the index of refraction (Na D line); (e - l ) / ( c + 2), where e is the DC dielectric constant; 6, which is the Hildebrand solubility parameter (defined below); V, the molar volume, which is equal to the molecular weight divided by the density at 298 K. These parameters are a measure of the solvent polarizability, solvent polarity, cohesive energy density, and molecular size, respectively. As will be seen below, we have not found any one solvent parameter that universally explains the solubility of C ~ O . This result is not atypical of the interaction between a particular solute and a range of solvents. The Hildebrand solubility parameter is defined as 6 = (hE /V) ' /Z , where hE is the cohesive energy and V the molar v ~ l u m e . ~ The cohesive energy is the energy required to convert a mole of liquid at 298 K to a mole of noninteracting gas. Normally, the enthalpy of vaporization, AHvapr is the available The Journal of Physical Chemistry, Vol. 97, No. 13. 1993 3381 4 + + I I I I I I I I I I I I I I I I I I I t + t I I I -t I+ , , I , +#+ 0 0. I 0.2 0.3 0.4 0.5 0.6 (n2-I~/(n2t2~ I Figure 1. The 298 K solubility of C6O as a function of the solvent polarizability parameter. The polarizability parameter value for C60 is indicated by a dashed vertical line. I20 I I I O I I 0 1 I I I 0 1 4o t I I n , I - 0 - - - [ € - I I / [ c re1 Figure 2. The 298 K solubility of C60 as a function of the solvent polarity parameter. The polarity parameter value for C6O is indicated by a dashed vertical line. experimental quantity and 6 is obtained as 1 ' I 2 P(AH" , , - RT) where p is the density at 298 K and M the molecular weight. Scott has derived an empirical method for correcting AHvap values that are not 298 K values.'O In calculating the 6 values, we have taken recourse to this empirical correction factor when AHvap (298 K) literature values were not readily available. The values for n, e, V, and 6 are listed in Table I, along with the measured solubilities. Blank entries (-) result from our not finding literature values. In the figures presented below, the solubilities have been converted to mole fractions to remove the influence of density and molecular weight. Figure 1 shows thevariation of solubility with the polarizability parameter, (nZ - l)/(nz + 2). Clearly, the higher the value of n, the higher thesolubility. One notable outlier is carbondisulfide, which does not have as high a solubility (expressed as mole fraction) as might be expected from its high refractive index and a polarizability parameter of 0.36. For comparison purposes, the refractive index of c 6 0 is 1.96, which gives a value of 0.49 for the polarizability parameter. Figure 2 is a plot of the solubility as a function of the polarity parameter, (e - l ) / ( c + 2). Note that polarizability and polarity parameters are close to identical for solvents not possessing a permanent dipole moment. There is clearly more scatter in Figure 2 than in Figure 1, but in general the nonzero solubilities cluster in the polarity parameter range 0.3-0.8, Le., that for solvents 3382 The Journal of Physical Chemistry, Vol. 97, No. 13, 1993 Ruoff et al. t x i IZOi- X X X X X x x x X 0 -, " I I I 40 72 I04 I36 I68 200 101 a r YO I ume ( C R ~ - ~ O I - I I Figure 3. The 298 K solubility of Cm as a function of the molar volume of the solvent. The molar volume of solid c60 is 429 cm3 mol-' and is not explicitly shown in the figure.. I20 1 I I I I I I T I X - I , I I I P I I x I I I X I I I- I Y " - I. I " 15 0 5 7 9 I I 13 H i ldebrand solubi I i ly parameter ( c a l 1'z-cm-3'zl Figure 4. The 298 K solubility of as a function of the Hildebrand solubility parameter. The square root of the cohesive energy density of Coo solid at 298 K is indicated by a dashed vertical line with e ranging from 2.5 to 10. Within this range, the solubilities do not display any systematic variation. For comparison purposes, for c 6 0 with t = 3.61, (e - l)/(e + 2) = 0.47. Figure 3 shows the solubility as a function of V, the solvent molar volume. The data clearly indicate that an increase in solvent molecular size favors solvation of c60. The molar volume of fcc c60(s) is 429 cm3 mol-', significantly higher than that of any solvent used here. The solubility of c 6 0 as a function of the Hildebrand solubility parameter, 6, is depicted in Figure 4. From Hildebrand's theory, one would expect the solubility to be greatest in a solvent v:hose 6 value matches that of c 6 0 and to decline progressively with increasing disparity in the 6 values. Solvents with appreciable solubilities, such as toluene, phenylnaphthalene, chloronaphtha- lent, and CS2, have 6 values between 9.0 and 10.0, and from Figure 4 one might expect c 6 0 to have a 6 value of 10. However, some solvents, like acetone, nitrobenzene, and methylene c