3990
Ind. Eng. Chem. Res. 2008, 47, 3990–3995
Correlation and Prediction of Partition Coefficients From the Gas Phase and from Water to Alkan-1-ols Michael H. Abraham,* Asadollah Nasezadeh, and William E. Acree, Jr. Department of Chemistry, UniVersity College London, 20 Gordon Street, London WC1H 0AJ, United Kingdom and Department of Chemistry, P.O. Box 305070, UniVersity of North Texas, Denton, Texas 76203-5070
Partition coefficients, PS, have been obtained from the literature for partitions from water and from the gas phase to wet butan-1-ol, pentan-1-ol, hexan-1-ol, heptan-1-ol, nonan-1-ol, and decane-1-ol. Equations based on the Abraham descriptors have been constructed for the correlation of the 12 sets of coefficients to yield equations for the prediction of further coefficients. The number of compounds used in the regressions ranges from 54 (the decan-1-ol system) to 106 (the pentan-1-ol system), and the standard deviation (SD) in the regression equations is from 0.13 to 0.18 in log PS, except for the water-to-butan-1-ol system, where it is 0.23 log units. For the gas-to-wet alcohol partitions, the SD values are about 0.02 log units larger than for the corresponding water-to-wet alcohol partitions. The coefficients in the regression equations alter quite regularly with increase in size of the alcohol and can be used to deduce the chemical properties of the wet alcohols that influence partition. Introduction The choice of solvent in water-to-solvent extractions is of ongoing importance. Bilgin et al.1 have recently studied the extraction of butanoic acid by 17 water-to-solvent systems, and Offeman et al.2 have investigated the screening of solvents for extraction of ethanol and have determined partition coefficients, PS, for ethanol in a wide variety of water-alcohol systems.3 Since the possible combinations of solutes and solvent systems is so large, there have been a number of attempts to fit and then predict partition coefficients. For the specific case of the water-to-octan-1-ol system, there are available a very large number of methods,4 many of which are commercially available, but for other water-to-solvent systems, few methods are available. Torrens et al.5 have devised a “universal” model for the calculation of partition coefficients in any water-to-solvent system. Typical results are an average absolute deviation, AAD, of 1.48 log units and a standard deviation of 1.66 log units for log PS values of 15 solutes in a number of water-solvent systems,6 but the results are not good enough for the method to be of any real practical use. A statistically much better procedure for the correlation of log PS values was reported by Meyer and Maurer.7 They used a linear solvation energy relationship to correlate log PS values for a variety of solutes in 30 water-solvent systems with an absolute average deviation of 0.12 log units. However, in many of the systems, there were very few solutes. For 14 systems, the number of data points was 15 or less, so that their equation with five descriptors would be expected to yield excellent fits. Subsequently, Meyer and Maurer8 extended their method to include properties of both solutes and solvents in a general equation with a total of 36 properties, denoted as LSER36. For 826 data points over 20 water-solvent systems, the log PS values could be fitted with AAD ) 0.16 log units only, an excellent result. It must be said, however, that the 101 solutes studied were structurally quite simplesonly nine compounds were difunctional, and no heterocyclic com* To whom correspondence should be addressed. Tel.: +44 (0)207679-4639. Fax: +44 (0)20-7679-7463. E-mail: m.h.abraham@ ucl.ac.uk.
pounds at all were included. Indeed, the fragmentation system for the estimation of solute descriptors described by Meyer and Maurer is limited to aliphatic and aromatic compounds, with no provision for heterocyclic compounds at all.8 Since water-alcohol systems are important partitioning systems,2 we wished to study such systems other than the wellknown water-to-octan-1-ol system, especially the other wateralkan-1-ol systems. Torrens gave no such systems as examples, but Meyer and Maurer listed the water-to-pentan-1-ol system, although only for 14 solutes. Our aim is, thus, to collect log PS values at 298 K for as wide a range of solutes as possible in the water-to-alkan-1-ol systems from butan-1-ol up to decan1-ol and to set out equations for the correlation and prediction of log PS values. Methodology. We have devised9,10 two general linear free energy relationships, LFERs, for the correlation and prediction of a variety of water-solvent and gas-solvent partition coefficients, eqs 1 and 2. log PS ) c + eE + sS + aA + bB + νV
(1)
log KS ) c + eE + sS + aA + bB + lL
(2)
The independent variables in eqs 1 and 2 have been described several times.10–20 In brief, they are solute descriptors as follows: E is an excess molar refraction in cm3 mol-1/10. S is a combined dipolarity/polarizability descriptor. A is the overall solute hydrogen bond acidity, and B is the overall solute hydrogen bond basicity. V is McGowan’s characteristic molecular volume in cm3 mol-1/100, and L is the gas-to-hexadecane partition coefficient at 298 K. The set of coefficients, c, e, s, a, b, and V (or l), characterize the system and are determined by multiple linear regression analysis. An amendment to eq 1 is needed for partitions from water to solvents that take up considerable quantities of water, for example ethyl acetate and the alcohols. The standard B-descriptor is replaced by the descriptor Bo. This differs from B only for certain specific compounds: aniline and methylated anilines, pyridine and methylated pyridines, and sulfoxides (but not sulfones). In the tables in the Supporting Information, both B and Bo are given for all the solutes studied. An example of
10.1021/ie800020s CCC: $40.75 2008 American Chemical Society Published on Web 04/22/2008
Ind. Eng. Chem. Res., Vol. 47, No. 11, 2008 3991 Table 1. Coefficients and Statistics for Correlations of Water-to-Alcohol Partitions (log PS, coefficient W) and Gas-to-Alcohol Partitions (log KS, Coefficient l) against Abraham Descriptorsa alcohol Butanol log PS log KS Pentanol log PS log KS Hexanol log PS log KS Heptanol log PS log KS Octanol log PS log KS Nonanol log PS log KS Decanol log PS log KS a
c
e
s
a
b
V
0.369 0.071 -0.123 0.070
0.426 0.077 0.220 0.095
-0.719 0.071 1.414 0.083
-0.091 0.066 3.430 0.069
-2.346 0.080 2.600 0.085
2.689 0.081
0.185 0.049 -0.107 0.043
0.367 0.062 -0.001 0.089
-0.732 0.057 1.188 0.066
0.105 0.052 3.614 0.056
-3.100 0.059 1.671 0.066
3.395 0.078
-0.006 0.090 -0.302 0.088
0.460 0.082 -0.046 0.121
-0.940 0.069 0.880 0.084
0.142 0.058 3.609 0.066
-3.284 0.099 1.785 0.111
3.792 0.074
0.041 0.061 -0.159 0.057
0.497 0.067 0.018 0.098
-0.976 0.052 0.825 0.068
0.030 0.052 3.539 0.061
-3.438 0.054 1.425 0.065
3.859 0.076
0.088 -0.222 0.029
0.562 0.088 0.050
-1.054 0.701 0.055
0.034 3.478 0.060
-3.460 1.477 0.057
3.814
-0.041 0.064 -0.197 0.059
0.562 0.059 0.141 0.095
-1.103 0.052 0.694 0.075
0.090 0.057 3.616 0.070
-3.540 0.046 1.299 0.059
3.922 0.086
-0.136 0.077 -0.302 0.100
0.542 0.099 0.233 0.168
-0.989 0.085 0.741 0.126
0.046 0.064 3.531 0.087
-3.722 0.099 1.177 0.140
3.996 0.107
l
0.523 0.025
0.721 0.024
0.824 0.031
0.830 0.025
0.851 0.011
0.827 0.030
0.835 0.040
N
SD
R2
78
0.23
0.954
298
78
0.24
0.997
4130
106
0.18
0.976
823
106
0.20
0.998
13149
105
0.19
0.970
630
105
0.23
0.991
2104
78
0.14
0.990
1374
78
0.17
0.998
6387
613 395
0.12 0.21
0.995 0.988
23255 6363
82
0.13
0.990
1515
82
0.17
0.998
6261
54
0.13
0.989
843
54
0.18
0.994
1699
F
The Bo-descriptor is used for correlations of log PS and the B-descriptor for correlations of log KS.
an application of the amended eq 1 is the water-to-octan-1-ol system, where log Poct is given by the following equation.21 log Poct ) 0.088 + 0.562E-1.054S + 0.034A 3.460Bo + 3.814V (3) where N ) 613, R2 ) 0.995, SD ) 0.12, and F ) 23255. Here and elsewhere, N is the number of data points (the number of solutes), R is the correlation coefficient, SD is the standard deviation, and F is the F-statistic. Once log PS values are available, they can be transformed into gas-to-alcohol partition coefficients (or more exactly gasto-wet alcohol partition coefficients), KS, through eq 4, where KW is the gas-to-water partition coefficient. The gas-to-solvent (water) partition coefficients are defined through eq 5; if the concentration units in the gas phase and in solution are the same, then KS and KW are unitless and correspond to the Ostwald solubility coefficient. log PS ) log KS - log KW
(4)
KS ) concentration of solute in solution/ concentration of solute in the gas phase (5) Having to hand a series of log KS values in wet alcohols, we aim to set out equations for their correlation and prediction, thus characterizing additional separation systems. The standard eq 2 can be used for partitions from the gas phase to wet solvents. All the log PS values that we used were taken from the valuable compilation of Leo22 except for a few values for
butanoic acid1 and ethanol.3 Most of the log KW values required to convert log PS values into log KS values were taken from the literature.23–28 When these were not available from the literature, they were obtained by an analysis of partition coefficients in various systems (not water-to-wet alcohols, except for the water-to-wet octanol system) or of solubilities in dry solvents exactly as outlined by Acree et al.29 The values of log PS, log KW, and log KS that we used are given in Tables S1-S6 in the Supporting Information. Results Butan-1-ol. The solute descriptors and the log PS, log KW, and log KS values for 81 solutes are in Table S1. Regressions using the LFERs in the amended eqs 1 and 2 yield the coefficients and statistics given in Table 1. The SD values for the coefficients are given in the row below the coefficients. The statistics of the equations in Table 1 are quite reasonable, considering that the solutes used range from simple compounds such as ethylamine to drugs (cocaine, morphine), steroids (estrone, estratriol), and heterocyclic compounds (barbiturates, triazoles). The two equations in Table 1 can, thus, be considered to apply quite generally to a very wide range of solutes. There were a few outliers to the equations, especially the carboxylic acids propanoic acid (n ) 3), butanoic acid (n ) 4), and octanoic acid (n ) 8). In Figure 1 are plotted the observed and calculated log PS values for the n-carboxylic acids formic acid (n ) 1) to octanoic acid (n ) 8); the regression line is also shown. It seems clear that the log PS
3992 Ind. Eng. Chem. Res., Vol. 47, No. 11, 2008
Figure 1. Observed and calculated log P values for the carboxylic acids formic acid (n ) 1) to octanoic acid (n ) 8) in the water-to-butan-1-ol system. The line shown is the regression line.
values for the three acids, propanoic acid, butanoic acid, and octanoic acid, are in error. There are no previous correlations of log PS or log KS values with which to compare our equations. Pentan-1-ol. The compound descriptors and data that we use are in Table S2 for 106 compounds; the coefficients in the equations and the statistics are in Table 1.The equations are slightly better than the corresponding ones for butan-1-ol; this must be due to a rather better quality of the experimental data. As for the water-to-butan-1-ol and gas-to-butan-1-ol systems, the solutes cover a very wide range of compounds. These include mercury(II) chloride, with descriptors taken from our work24 on descriptors for several mercury(II) species. We can only include mercury(II) chloride, because log PS values were not reported for any other mercury(II) species. The general LFERs, eqs 1 and 2, are not restricted to organic solutessthey apply to any solute for which descriptors can be obtained. Meyer and Maurer7 have obtained an equation for water-to-pentan-1-ol partitions. They record AAE ) 0.09 log units but for an equation that fitted only 14 data points with five descriptors. For comparison, AAE ) 0.13 log units for the water-to-pentan-1ol equation summarized in Table 1, which fits 106 data points with five descriptors. Hexan-1-ol. The compound descriptors and values of log PS, log KW, and log KS are in Table S3, and the corresponding regression equations are in Table 1. The statistics are reasonable considering the wide range of compound studied, which includes multifunctional compounds and heterocyclic compounds. The term eE in the equation for log KS is not statistically significant, but we retain it for purposes of comparison. There are no previous regression equations for either the water-to-hexan-1ol or gas-to-hexan-1-ol systems with which to compare our equations. Heptan-1-ol. The compound descriptors and values of log PS, log KW, and log KS are in Table S4, and a summary of the regression equations is in Table 1. The two equations are statistically good and cover a quite large range of compound type. Octan-1-ol. A regression equation for log PS has already been constructed21 for 613 compounds, eq 3, and the corresponding equation for log KS, based on the same set of solutes,25 is summarized in Table 1. The number of compounds in the equation for log KS is less than the number for log PS in eq 3 because values of log KW were not available to convert log PS into log KS. There are no previous equations with which to compare our equation for log KS, but Meyer and Maurer7 have obtained an equation for log PS for the water-to-octan-1-ol system. For 263
solutes, they obtain AAD ) 0.13 log units. Equation 5 with 631 solutes has SD ) 0.12, and the corresponding AAD value will be lower, perhaps about 0.10 log units. Nonan-1-ol. We have used data on 82 solutes in the waterto-nonan-1-ol and gas-to-nonan-1-ol systems; details are in Table S5. The regression equations are again summarized in Table 1. The statistics of the two equations are quite good, and there is a wide variety of compoundd, including a number of derivatives of acyclovir. Decan-1-ol. The last alcohol in the series for which there are enough log PS values to construct a reliable regression equation is decan-1-ol, and for the water-to-decan-1-ol and gasto-decanol-1 systems, we have 54 values. These are in Table S6, together with the respective compound descriptors. The regression equations are given in Table 1. The statistics are in line with those for the other water-alcohol systems, but the variety of compound types is substantially less. For example, there are only three heterocyclic compounds in the data set. Discussion The regression equations for log PS and log KS all have a reasonable number of data points, at least 54, for equations with five descriptors, and so can be regarded as soundly based. The SD values are all less than 0.20 log units, except for the waterto-butan-1-ol system (0.23). This is possibly due to the very large quantity of water taken up by butan-1-ol, which makes the actual experimentation rather difficult. All the SD values refer to fits and not to predictions. We can take a typical system, water-to-heptan-1-ol, and estimate the predictive capability of the equation by dividing the 78 compound data sets into a training set and a test set, each of 39 compounds. The 78 compounds were listed in order of increasing value of log PS, and every other compound in the list was taken as the training set. This yields the same spread of log PS values for the test set and the training set. The training set yielded the regression equation, eq 6. This is very close to the full equation, Table 1, and suggests that the training set is a representative sample of the entire set. Equation 6 was then used to predict log PS for the 39 compounds not used to construct the training equation. log PS for the 39-compound test set was predicted with AD ) 0.011, AAD ) 0.128, and SD ) 0.157 log units. There is no bias in the predictions, with AD ) 0.011 only, and the SD for the predictions, 0.157, is but little more than the SD for the entire 78-compound regression (0.138). We can then suggest that the predictive capability of the various regression equations we have set out will be around 0.02 log units more than the given SD values of the total regression equations. log PS ) -0.021(0.084) + 0.557(0.083)E 0.986(0.063)S + 0.141(0.071)A 3.485(0.073)Bo + 3.821(0.102)V
(6)
where N ) 39, SD ) 0.13, R ) 0.991, and F ) 744. As mentioned in the Introduction, there are a number of specific compounds for which the descriptor Bo is not the same as the general hydrogen bond basicity descriptor, B. Taylor et al.30 first pointed out that there were a number of compounds for which the hydrogen bond basicity varied with the solvent system and did not include these in their analysis of partitioning. Abraham31 then suggested that, if such compounds were assigned two hydrogen bond basicity parameters, the standard basicity B and a new descriptor, Bo, that was necessary for partition from water into solvents that contained considerable amounts of water, the compounds could be incorporated into a 2
Ind. Eng. Chem. Res., Vol. 47, No. 11, 2008 3993
general system. The main variable basicity compounds identified by Abraham were aniline and alkylanilines, pyridine and alkylpyridines, and sulfoxides (but not sulfones). In addition, some nitrogen containing heterocyclic compounds showed variable basicity. The solvents for which the Bo descriptor is necessary are wet alcohols, wet ether, and wet ethyl acetate. Both the basicity descriptors are given in Tables S1-S6. It should be noted that the predictive area of chemical space for our equations is defined quite differently from that of methods that use structural fragments. In the latter case, in order to predict a value for a “new” compound, all the structural features must be included in the list of fragments. Thus, in Table S6, no aliphatic nitrile is listed, and therefore, a fragment scheme based on the compounds in Table S6 could not be used to predict a value for acetonitrile, because the scheme will lack the CN fragment. However, our method is not based on chemical classes but on chemical properties, and it is these that define the chemical space. Since the descriptors for acetonitrile lie within the range of descriptors used to set up the equations, acetonitrile is within our chemical space, and a prediction is justified. This is a marked advantage of a property-based scheme as compared to a fragment-based scheme. The range and type of compounds we have used is considerable and is reflected in the range of descriptors. For example, the A-descriptor varies from 0 up to 1.63 (3,4,5-trihydroxybenzene), and the B-descriptor varies from 0 to 2.04 (morphine) and 2.80 (mitoxantrone), so that predictions for a huge range of compounds can be made. The only restriction, of course, is that descriptors for the compounds must be available. Although the system of Meyer and Maurer7,8 is statistically very good, it is much more restricted; the range of hydrogen bond acidity is from 0 to 0.69 or 1.19 (estimated), and the hydrogen bond basicity range is from 0 to 1.00 only. Thus, important compounds with large hydrogen bond basicity are excluded, for example, caffeine (1.28), barbiturates (>1.16), propranolol (1.31), cocaine (1.50), acyclovir (2.00), or morphine (2.04). As we have pointed out before, the coefficients in the regression equations are not just fitting coefficients but can be interpreted in terms of chemical interactions. The coefficients in the log PS equations refer to differences between water and the wet alcohols as follows. The e-coefficient gives the difference in ability to interact with σ- and π-electrons, the s-coefficient gives the difference in dipolarity/polarizability, the a-coefficient gives the difference in hydrogen bond basicity (because an acidic solute will interact with a basic solvent), the b-coefficient gives the difference in hydrogen bond basicity (because a basic solute will interact with an acidic solvent), and the V-coefficient gives the relative tendency of water and the wet alcohol to interact with solutes purely on the basis of their size. The larger the V-coefficient, the more the hydrophobic solutes will partition into the alcoholic layer. If our equations are consistent, we would expect the various coefficients to alter regularly with increasing size of the alcohol, that is, also with decreasing water content of the water-saturated alcohol. This is shown in Figure 2. The coefficients all behave systematically with the size of the alcohol. The e-coefficient remains almost constant, but the s-coefficient becomes slightly more negative as the alcohol becomes larger, presumably because wet butan-1-ol is more dipolar than wet decan-1-ol and, hence, is that much nearer to water. The behavior of the a-coefficient is very interesting, in that it is close to zero for all the systems studied, so that hydrogen bond basicity of water and all the wet alcohols must be very nearly the same. The
Figure 2. Plots of the coefficients in eq 1 against the carbon number of the alcohols: 2, V-coefficient; O, e-coefficient; 9, a-coefficient; b, s-coefficient; (, b-coefficient. The point at N ) 0 represents water itself. Table 2. log PS and log KS Values for Propylbenzene in Water-to-Solvent and Gas-to-Solvent Systems log PS or log KS system
Obsa
Calc
Obs - Calc
water to butan-1-ol water to pentan-1-ol water to hexan-1-ol water to heptan-1-ol water to octan-1-ol water to nonan-1-ol water to decan-1-ol water to heptane water to hexadecane water to olive oil gas to water gas to water at 310 K gas to hexadecane gas to olive oil
2.70 2.74 2.79 2.86 3.72c 2.90 3.45 4.11 3.85 3.77 0.39 0.11 4.22 4.00
2.98 3.44 3.63 3.73 3.73 3.68 3.69 4.16 4.00 3.95 0.37 0.12 4.23 3.81
-0.28 -0.70b -0.84b -0.87b -0.01 -0.78b -0.24b -0.05 -0.15 -0.18 0.02 -0.01 -0.01 0.19
a Values from ref 22.b Not used in this work.c Recommended value from ref 22.
b-coefficient is always very negative, showing that water is a much stronger hydrogen bond acid than the wet alcohols; also, the larger the alcohol and the less water there is in the saturated solution, the less acidic the wet alcohol becomes. Finally, as expected, the V-coefficient increases markedly with an increase in the size of the alcohol. The larger the alcohol, the better it will accommodate hydrophobic solutes. Exactly the same conclusions can be made by inspecting the coefficients for the gas-alcohol partitions. We noted earlier that some of the carboxylic acids were outliers in the water-to-butan-1-ol system. There are two particular compounds that are outliers in several of the systems, not just in one system. The log PS values for propylbenzene fit our equations in the butan-1-ol and octan-1-ol systems and in many other systems that we have studied; see Table 2. However, the log PS values in the pentan-1-ol, hexan-1-ol, heptan-1-ol, and nonan-1-ol systems are always much smaller than those we calculate and are not compatible with the recommended value in the octan-1-ol system. 22 The calculated values in Table 2 are based on E ) 0.604, S ) 0.50, A ) 0.00, B ) 0.15, L ) 4.230, and V ) 1.1391. We can only assume that there are systematic experimental errors in the log PS determinations for the four water-alcohol systems, above. Another problem is that of the log PS values for 2,6dihydroxybenzoic acid; see Table 3. Values in the octan-1ol system, either the recent value of Dearden et al.32 or the recommended value of Leo,22 are far larger than the reported
3994 Ind. Eng. Chem. Res., Vol. 47, No. 11, 2008 Table 3. log PS Values for 2,6-Dihydroxybenzoic Acid in Water-to-Alkan-1-ol Systems alkan-1-ol
log PS (obs)a
butan-1-ol pentan-1-ol hexan-1-ol heptan-1-ol octan-1-ol decan-1-ol
0.64b 0.71b 0.63b 0.69b 0.69,b 1.65,c2.20d 0.63b
a All values from ref 22 except where shown.b Listed as uncertain or unreliable in ref 22.c Ref 32.d Recommended value, ref 22.
oct ) octanol s ) solvent w ) water Definitions AAD ) (1/N)Σ(|y(calc) - y(obs)|) AD ) (1/N)Σ[(y(calc) - y(obs)] SD ) Σ{[(y(calc) - y(obs)]2}/(N - D - 1) F ) R2(N - D - 1)/(1 - R2)D
Literature Cited values in the other systems. Leo has referenced the lower log PS values as “uncertain or unreliable”, and so we did not use 2,6-dihydroxybenzoic acid in any of our regression equations. We have concentrated on the water-to-wet alcohol partitions, rather than partitions from the gas phase to the wet alcohols, because the former are the more important. There have been no previous published equations for partition from the gas phase to wet alcohols, but both Abraham et al.33 and Li et al.34 have set out equations for the partition of solutes from the gas phase to dry alcohols. Conclusion We have constructed equations for the partition of compounds from water to (wet) alcohols and from the gas phase to (wet) alcohols in terms of the Abraham descriptors. The range of compounds used is very large and, hence, so is the range of descriptors; the equations can be used to predict further partition coefficients for a huge number of compounds. There is no restriction on compound type; partition coefficients can be predicted for any compound for which the descriptors are known, no matter whether organic or inorganic. For example, we include hydrogen peroxide, iodine, and mercury(II) chloride in some of our equations. The set of equations and descriptors represents the most comprehensive method for the prediction of partition coefficients from water to wet alcohols and from the gas phase to wet alcohols yet reported. Supporting Information Available: Tables S1-S6 are available as Supporting Information. This material is available free of charge via the Internet at http://pubs.acs.org. Nomenclature A ) LFER constant AAD ) average absolute deviation AD ) average deviation B ) LFER constant D ) number of descriptors E ) LFER constant F ) Fisher F-statistic S ) LFER constant SD ) standard deviation KS ) gas-to-solvent partition coefficient KW ) gas-to-water partition coefficient L ) LFER constant N ) number of data points PS ) water-to-solvent partition coefficien R ) correlation coefficient V ) LFER constant Subscripts
(1) Bilgin, M.; Kirbas¸lar, S¸. I.¨; Ozcan, Ö.; Dramur, U. Distribution of Butyric Acid between Water and Several Solvents. J. Chem. Eng. Data 2006, 51, 1546. (2) Offeman, R. D.; Stephenson, S. K.; Robertson, G. H.; Orts, W. J. Solvent Extraction of Ethanol from Aqueous Solutions. I. Screening Methodology for Solvents. Ind. Eng. Chem. Res. 2005, 44, 6789. (3) Offeman, R. D.; Stephenson, S. K.; Robertson, G. H.; Orts, W. J. Solvent Extraction of Ethanol from Aqueous Solutions. II. Linear, Branched, and Ring-Containing Alcohol Solvents. Ind. Eng. Chem. Res. 2005, 44, 6797. (4) Waterbeemd, H. van de Mannhold, R. Programs and Methods for Calculation of log P-values. Quant. Struct.-Act. Relat. 1996, 15, 410. (5) Torrens, F.; Sanchez-Marin, J.; Nebot-Gil, I. Universal model for the calculation of all organic solvent-water partition coefficients. J. Chromatogr., A 1998, 827, 345. (6) Torens, F. J. Universal Organic Solvent-Water Partition Coefficient Model. Chem. Inf. Comput. Sci. 2000, 40, 236. (7) Meyer, P.; Maurer, G. Correlation of Partition Coefficients of Organic Solutes between Water and an Organic Solvent. An Application of the Linear Solvation Energy Equation. Ind. Eng. Chem. Res. 1993, 32, 2105. (8) Meyer, P.; Maurer, G. Correlation and Prediction of Partition Coefficients of Organic Solutes between Water and an Organic Solvent with a Generalized form of the Linear Solvation Energy Equation. Ind. Eng. Chem. Res. 1995, 34, 373. (9) Abraham, M. H. Scales of hydrogen bonding: Their construction and application to physicochemical and biochemical processes. Chem. Soc. ReV. 1993, 22, 73. (10) Abraham, M. H.; Ibrahim, A.; Zissimos, A. M. The determination of sets of solute descriptors from chromatographic measurements. J. Chromatogr., A 2004, 1037, 29. (11) Abraham, M. H.; Du, C. M.; Platts, J. A. Lipophilicity of the nitrophenols. J. Org. Chem. 2000, 65, 7114. (12) Abraham, M. H.; Zissimos, A. M.; Acree, W. E., Jr. Partition of solutes from the gas phase and from water to wet and dry di-n-butylether: A linear free energy relationship. Phys. Chem. Chem. Phys. 2001, 3, 3732. (13) Abraham, M. H.; Zissimos, A. M.; Huddleston, J. G.; Willauer, H. D.; Rodgers, R. D.; Acree, W. E., Jr. Some novel liquid partitioning systems: Water-ionic liquids and aqueous biphasic systems. Ind. Eng. Chem. Res. 2003, 42, 413. (14) Abraham, M. H.; Zissimos, A. M.; Acree, W. E., Jr. Partition of solutes into wet and dry ethers; An LFER analysis. New J. Chem. 2003, 27, 1041. (15) Abraham, M. H.; Acree, W. E., Jr. Correlation and prediction of partition coefficients between the gas phase and water, and the solvents dodecane and undecane. New J. Chem. 2004, 28, 1538. (16) Abraham, M. H.; Acree, W. E., Jr. Characterisation of the waterisopropyl myristate system. Int. J. Pharm. 2005, 294, 121. (17) Abraham, M. H.; Zhao, Y. H. Characterisation of the water/onitrophenyl octyl ether system in terms of the partition of nonelectrolytes and ions. Phys. Chem. Chem. Phys. 2005, 7, 2418. (18) Abraham, M. H.; Acree, W. E., Jr. The Correlation and Prediction of Butane/Water and Gas/Butane Partition Coefficients. Can. J. Chem. Eng. 2005, 83, 362. (19) Sprunger, L.; Clark, M.; Acree, W. E., Jr.; Abraham, M. H. Characterization of room-temperature ionic liquids by the Abraham model with cation-specific and anion-specific coefficients. J. Chem. Inf. Model. 2007, 47, 1123. (20) Abraham, M. H.; Ibrahim, A.; Acree, W. E., Jr. Partition of compounds from gas to water and from gas to physiological saline at 310K; linear free energy relationships. Fluid Phase Equilib. 2007, 251, 93. (21) Abraham, M. H.; Chadha, H. S; Whiting, G. S.; Mitchell, R. C. Hydrogen bonding. 32. An analysis of water-octanol and water-alkane partitioning, and the ∆logP parameter of Seiler. J. Pharm. Sci. 1994, 83, 1085.
Ind. Eng. Chem. Res., Vol. 47, No. 11, 2008 3995 (22) Leo, A. J. The MedChem data base 2007; BioByte Corp. and Pomona College, Daylight Chemical Information Systems: 27401 Los Altos, #360 Mission Viejo, CA 92691. (23) Abraham, M. H.; Andonian-Haftvan, J.; Whiting, G. S.; Leo, A.; Taft, R. W. Hydrogen bonding. Part 34: The factors that influence the solubility of gases and vapours in water at 298 K, and a new method for its determination. J. Chem. Soc., Perkin Trans. 2 1994, 1777. (24) Abraham, M. H.; Gil-Lostes, J.; Acree, W. E., Jr.; Cometto-Mu˜niz, J. E.; Cain, W. S. Solvation parameters for mercury and mercury(II) compounds: Calculation of properties of environmental interest. J. EnViron. Monit. 2008, 10, 435. (25) Abraham, M. H.; Gola, J. M. R.; Cometto-Muñiz, J. E.; Cain, W. S. Solvation properties of refrigerants, and the estimation of their watersolvent and gas-solvent partition coefficients. Fluid Phase Equilib. 2001, 180, 41. (26) Abraham, M. H. The determination of air/water partition coefficients for alkylcarboxylic acids by a new indirect method. J. EnViron. Monit. 2003, 5, 747. (27) Cabani, S.; Gianni, P.; Mollica, V.; Lepori, L. Group contributions to the thermodynamic properties of non-ionic organic solutes in dilute aqueous solution. J. Solution Chem 1981, 10, 563. (28) English, N. J.; Carroll, D. G. Prediction of Henry’s Law constants by a quantitative structure property relationship and neural networks. J. Chem. Inf. Comput. Sci. 2001, 41, 1150.
(29) Hoover, K. C.; Acree, W. E., Jr.; Abraham, M. H. Mathematical correlation of phenothiazine solubilities in organic solvents with the Abraham solvation parameter model. Phys. Chem. Liq. 2006, 44, 367. (30) Leahy, D. E.; Morris, J. J.; Taylor, P. J.; Wait, A. R. Model solvent systems for QSAR. Part 3. An LSER analysis of the “critical quartet”. New light on hydrogen bond strength and directionality. J. Chem. Soc., Perkin Trans. 2 1992, 705. (31) Abraham, M. H. Hydrogen bonding. 31. Construction of a scale of solute effective or summation hydrogen-bond basicity. J. Phys. Org. Chem. 1993, 12, 660. (32) Dearden, J. C.; Bresnen, G. M. Thermodynamics of water-octanol and water-cyclohexane partitioning of some aromatic compounds. Int. J. Mol. Sci. 2005, 6, 119. (33) Abraham, M. H.; Le, J.; Acree, W. E., Jr. The solvation properties of the aliphatic alcohols. Collect. Czech. Chem. Commun. 1999, 64, 1748. (34) Li, J.; Zhu, T.; Hawkins, G. D.; Winget, P.; Liotard, D. A.; Cramer, C. J.; Truhlar, D. G. Extension of the platform of applicability of the SM5.42R universal solvation model. Theor. Chem. Acc. 1999, 103, 9.
ReceiVed for reView January 04, 2008 ReVised manuscript receiVed February 19, 2008 Accepted February 27, 2008 IE800020S
