430
Anal. Chem. 1084, 56,430-432
Benzonitrile Nitrogen- 15 Nuclear Magnetic Resonance Chemical Shift as a Probe of Solvent Dipolarity Orland W. Kolling Chemistry Department, Southwestern College, Winfield, Kansas 67156
The Kamlet-Abboud-TaR description of solvent effects upon chromatographic separatlons and chemlcal equlllbrla uses the solvent dlpolarlty number (a*)as one fundamental parameter In llnear free energy equations. Although the a* scale has been recently revlsed by using solvent-lnduced NMR chemlcal shlfts for 13C benzohalldes, the smaller sensltlvlty coefflclents for these probes glve rlse to greater uncertalntles In some Instances for the derlved a* values. By contrast, It Is demonstrated that the larger "N NMR chemlcal shlRs are more responslve to changes In solvent dlpolarlty as well as to hydrogen bond donor behavlor, and these shlft magnltudes can be effectlvely used for purposes of verlflcatlon of a" values. The speclflc case of ["N]benzonltrlle was examlned In detall as an NMR probe In 17 pure solvents and In the aprotlc cosolvent system N,N-dlmethylformamlde-acetone.
In its most general form the Kamlet-Abboud-Taft interpretation of medium effects upon equilibria and reaction mechanisms employs the linear free energy function in eq 1 (I). Here, the influence of the solvent upon the observable
P = Po + S(A*
+ d 6) + UCY + bo
(1)
( P )is determined by the three major parameters related to the solvent species: its hydrogen bond donor acidity (a); its hydrogen bond acceptor basicity (p); and the dipolarity number ( A * ) related to the dipole moment of the solvent molecule (2). (The d6 term is a polarizability correction making a nonzero contribution in eq 1 for aromatics and polyhalogenated alkanes.) Clearly, the A* values are the most fundamental quantities characterizing both hydrogen bonding and nonhydrogen bonding solvents, and the experimental basis for refinement of the whole ?F* scale as well as for its theoretical justification has been expanded greatly beyond the earlier empirical studies on simple UV-visible solvatochromic molecules as the indicator solutes (3). The solvent-induced chemical shifts in the aromatic (para) 13C NMR spectra for benzotrifluoride (BTF) and phenylsulfurpentafluoride (PSPF) as the probe species have been used by Chawla et al. to establish more reliable ?F* numbers for the difficult cases of weak-to-strong hydrogen bond donors and the highly self-associated solvents (4). Although the functional groups in BTF and PSPF exhibit no detectable hydrogen bond acceptor behavior, one important limitation of para 13C NMR responses for monosubstituted benzenes is that the weighting coefficient (s) in eq 1 is modest in most instances (C1.5) and the limiting uncertainty of f0.06 in the corresponding A* values is somewhat larger than desirable (4). Therefore, other solvent-sensitive NMR probes need to be examined as potential secondary indicators for the verification of solvent dipolarity numbers; and the relevant new data for the case of benzonitrile are summarized herein. An earlier reinvestigation of the pyridine 15NNMFt chemical shifts in nonaqueous media demonstrated that the sizable shifts for the solute could be represented by a linear regression in A* and a where a > s (5). Thus, even though hydrogen bond 0003-2700/84/0358-0430$01.50/0
acceptor behavior by such nitrogen bases is significant, one would expect a larger sensitivity coefficient (s) for 15N probes than the para I3C species where the labeled atom is not a part of the primary polar group within the probe. The three major factors supporting the final selection of benzonitrile as the nitrogen base for this study are as follows: (a) it is a significantly more polar molecule than pyridine and the alkylamines and similar in dipolarity to the aromatic amines (A* N 0.9); (b) by contrast to the alkyl and arylamines as well as the alkylnitriles and pyridine, the dipolarity ( A * ) of benzonitrile far exceeds its basicity (0 N 0.4); and (c) as the simplest arylnitrile, benzonitrile exhibits no detectable hydrogen bond donor behavior nor special steric properties to complicate its solute-solvent dipolar interactions.
EXPERIMENTAL SECTION All aprotic solvents were reagent or spectral grade, and an initial drying of each liquid for 1 week over anhydrous calcium sulfate was followed by final purification according to standard methods (6). Boiling points and densities were the criteria used to verify the purity of solvents by comparison to accepted literature values (6).
Dielectric constants for the DMl-acetone and benzene-acetone binary systems were determined from changes in the dielectric cell capacitance at 25 "C with a Sargent Model 5 (5 MHz) cathode-coupled oscillator. Procedural details have been reported previously (7). Similarly, conductance measurements on 10-3-10-5 M solutions of tetra-n-butylammonium iodide in the acetoneDMF cosolvent system were made by using the methods of Bodenseh and Ramsey (8);and ion pair association constants (&) for the solute were evaluated by the computational methods of Fuoss and Hirsch (9). The solvent-induced 15NNMR chemical shifts for benzonitrile were determined by the procedures of Duthaler and Roberts (IO) in which 15N (1 M) nitric acid served as the external reference source. The upfield 15N shifts were obtained for 0.2-2 M benzonitrile in the solvents listed in Table I, and the final composite precision in the chemical shifts in pure solvents was f0.05 (std dev) by this method when solvent diamagnetic susceptibility corrections were applied.
RESULTS AND DISCUSSION For initial comparisons, the two select aprotic solvents, cyclohexane and dimethyl sulfoxide, were chosen to represent the interval 0.0 to 1.0 in T*. In this polarity range the upfield 15N shift for benzonitrile exceeds 4.0 ppm whereas the para 13C shifts for BTF and benzonitrile are 1.2 and 1.4 ppm, respectively (4). For benzonitrile in the alkanols, the 15Nshift is further increased to more than 6.0 ppm relative to n-hexane. The 15N chemical shifts of similar magnitude have been reported by Olah and Kiovsky for acetonitrile (13). Thus, one may infer that the very polar and linear cyano group having the nonbonding pair on its nitrogen atom is the primary determinant of the solute-solvent interactions in both aprotic and hydrogen bonding media. Dipolarity of Pure Solvents. The chemical shift (6 15N) for benzonitrile was determined in 17 anhydrous solvents, including five hydrogen bond donors, and these results are summarized in Table I. Because of the limited number of solvents used, aromatics and polyhalogenated aprotic solvents were excluded in order that the d6 term in eq 1 would be zero. 0 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 3, MARCH 1984
431
~
Table 11. Summary of Experimentally Derived Parameters at 2 5 C for the Binary Solvent: N ,N-Dimethylformamide- Acetone
Table I. Solvent Parameters and ”N NMR Shifts for Benzonitrile (at 25 C) i
~
*
a b~
6”N
8l5N
Aprotic Solvents acetone 2-butanone n-butyl acetate cyclohexane cyclohexanone dimethoxyethane N,N-dimethylformamide dimethyl sulfoxide p-dioxane ethyl acetate n-hexane sulfolane
X
0.71 0.67 0.46
117.8 117.7 116.8 115.0 118.5 117.1 118.5 118.9 117.2 117.3 114.8 118.9
0.00
0.88 0.53 0.88 1.00
0.55 0.55 -0.08 0.98 0.85 0.82 0.54 0.60 0.46
0.15 0.22 0.85 0.98 0.77
119.0 119.6 120.8 121.7 120.4
1.0000
egd
ec
20.68 21.43 22.24 23.11 23.89 25.48 27.06 28.70 30.38 32.05 33.69 34.37 35.26 35.98 36.42 36.70
0.3935 0.3968 0.4003 0.4039 0.4069 0.4127 0.4181 0.4234 0.4282 0.4328 0.4371 0.4388 0.4409 0.4425 0.4436 0.4443
R * ~
0.710 0.718 0.726 0.732 0.740 0.763 0.774 0.796 0.804 0.827 0.848 0.855 0.863 0.873 0.878 0.881
a Nitro en 1 5 NMR chemical shift in ppm for benzoEiperimental density values for the mixed nitrile. solvent. Dielectric constant data with an uncertainty of i 0.03 unit. Reaction field energy. e Solvent dipolarity values derived from the solvatochromic behavior of Phenol Blue ( I 4 ) . Experimental uncertainty in this set is i0.005 (std dev).
As an NMR probe, the quantitative response of benzonitrile clearly distinguishes between hydrogen bond donor (HBD) behavior and simple dipolarity influences by the solvent. Although the graph is not included here, a plot of P(G15N)vs. T* bisects the data in Table I to show a single linear function for the aprotic solvents and the points for HBD solvents as exhibiting significant positive deviations from that line. The final specific multiple regression based upon all of the solvents is given in eq 2, and this function has a 0.988 correlation coefficient. The uncertainties in the statistically derived
+ 3.90?r*+ 4.5a
dlzb
117.8 0.7847 117.8 0.7918 0.8011 117.9 0.8078 0.8169 118.0 0.8339 118.1 0.8499 0.8655 118.2 0.8804 118.3 0.8988 0.9152 118.4 0.9225 0.9304 0.9375 118.5 0.9420 118.5 0.9442
0.0479 0.0959 0.1446 0.1927 0.2908 0.3892 0.4886 0.5890 0.6902 0.7929 0.8371 0.8958 0.9476 0.9791
a Dipolarity numbers from the more recent papers of HBD acidity values from Kamlet and Taft (1, 4, 11 ). Taft et al. ( 1 2 ) .
615N(PhCN) = 115.1
shifta ~ ~
0,0000
Hydrogen Bonding Solvents acetonitrile dichloromethane ethanol methanol 2-propanol
D
r
50 L
€
I
,
,
(2)
quantities are as follows: intercept, f0.02; s, f0.02;a, fO.l; a* (calcd) f0.01-0.02 (std. dev). By contrast to the W-visible solvatochromic indicators where a / s is often somewhat less than unity ( 3 , 1 2 ) ,the a/s ratio of 1.15for eq 2 suggests that benzonitrile like pyridine should be a useful probe of HBD behavior as well as solvent dipolarity. Dipolarity in a Cosolvent System. Since little attention has been given to the responses of either NMR or ESR probes in binary solvents, one pair of aprotic solvents was selected for a detailed examination of the 15N NMR chemical shift for benzonitrile as a function of solvent mole fraction. These experimental NMR results along with other parameters of the cosolvent system, Nfl-dimethylformamide-acetone,are listed in Table 11. Electrolytic studies on quaternary ammonium salts in each of the pure solvents indicate similar ionic solvation and sequences of limiting cationic conductances in acetone and DMF (15);however, as seen in Table 11, the a* and dielectric constant values for each of the pure solvents are substantially different. Although DMF as a solvent shows only polar and Lewis base properties, acetone is both a polar base and an extremely weak hydrogen bond donor because of the very small population of its enolic tautomer. However, the latter characteristic is usually undetected by the “solvatochromic comparison method” (3)and must be considerably lower as an HBD liquid than acetonitrile ( a = 0.15). Nevertheless, a preliminary review of macroscopic properties for acetone-containing cosolvents was made in order to identify those types of solvent pairs in which nonideal solvent-solvent interactions occur.
x2
Figure I. Trend in dielectric constant (e) as a function of mole fraction (X,) for binary solvents containing acetone. Labeled curves are for acetone paired with (A) water, (B) N,N-dimethyiformamide, (C) 1propanol, (D) benzene, and (E) n-hexane.
Graphs of dielectric constant (e) vs. solvent mole fraction for representative binary systems containing acetone are shown in Figure 1. Nonlinear functions are observed not only for those systems having a strong hydrogen bond donor (Le., water and 1-propanol) but also for the cases of the nonpolar hydrocarbons, n-hexane and benzene (16,17). For the HBD systems, the pronounced curvature would be expected to arise from both donor-acceptor interactions involving hydrogen bonding to acetone and disruption of the structure of the self-associated hydroxylic solvent by the added acetone (17). The curvature for the acetone-hydrocarbon pairs occurs at the lower mole fractions of acetone and these plots become linear beyond 0.60 mole fraction of the less polar component. By contrast, the new results for the polar-polar pair, acetone-DMF, show good linearity from 0.0 to 0.7 mole fraction with only a slight departure from linearity in the acetone-poor range. Plots of a number of more complex e functions vs. mole fraction for the acetone cosolvents were made as well. Among the more important of these were the total polarization (P1J
432
ANALYTICAL CHEMISTRY, VOL. 56, NO. 3, MARCH 1984
calculated by the Clausis-Mosotti equation and the reaction field function (0,) derived by Abboud and Taft (18). Although the figures are not included here, the DMF-acetone system again shows nearly ideal graphs with no new features appearing over the total mole fraction range. Thus, any slight departures from linearity over the full composition range are insignificant enough to permit the cosolvent mixtures to be treated as "select solvents" (18) over broad composition intervals. Experimental T* values for the DMF-acetone system were determined independently using phenol blue as the solvatochromic indicator (14). An observed linearity between T* and mole fraction for the cosolvent must be interpreted in terms of the limiting precision in T* values generally. With the normal uncertainties in T* of *0.01-0.03 unit, it can be demonstrated that minor departures from linearity comparable to those found in e vs. mole fraction curves would be hidden by the scatter in the data points. However, with the higher experimental precision (*0.005 std dev) for this cosolvent system, the linearity in r* vs. mole fraction acetone must be considered real. Therefore, the T* numbers can be estimated from eq 3. The 15NNMR chemical shift for benzonitrile was measured in mixtures of DMF-acetone and these results are given in Table 11. The linear regression in eq 4 applies to the NMR spectral shift, and for the liquid mixtures the precision in PN(calcd) is *0.02 ppm with a 0.98 correlation coefficient. 615N(PhCN) = 117.8
+ 4 . 1 ( ~ *- 0.71)
(4) There is no evidence of abnormal deviations from linearity predicted by eq 4 in acetone-rich mixtures. As was noted above, any significant hydrogen bonding from the enolic form should cause a larger upfield I5N shift for PhCN than is expected for a nonhydrogen bonding component, and the larger a / s ratio in eq 2 amplifies the spectral sensitivity of the aryl nitrile to such perturbations. Ion Pair Association and Solvent Dipolarity. A final test of the consistency of the preceding arguments supporting the essentially ideal description of the cosolvents is based upon data for the ion pair association of tetra-n-butylammonium iodide in DMF-acetone mixtures. Bodenseh and Ramsey (8) demonstrated that the Fuoss relationship in eq 5 should be valid for those binary solvent mixtures whose electrolytic characteristics do not differ significantly from pure solvents having the same dielectric constant. Here, the critical em-
KA = h N a 3 exp( 3000
")
fakT
pirical test is that In KA vs. l / c will yield a linear plot, although precise linearity appears to be restricted to cosolvent systems with dielectric constants larger than 19 (8). Newly evaluated association constants for Bu4NI over the mole fraction range of 0.0-0.5 XDMF are listed in Table 111, and the value for pure acetone is in close agreement with the result (154) reported by Reynolds and Kraus (19). However, the literature K Avalue (15)for Bu4NI in pure DMF could not be confirmed nor could constants of any meaningful reproducibility be determined in DMF-rich solvent mixtures be-
Table 111. Ion Pair Association of Bu,NI in DMF-Acetone Mixtures at 2 5 " C XAcet
1.000
0.952 0.904 0.855 0.801 0.709 0.602 0.561
fa
20.68 21.43 22.24 23.11 23.80 25.48 27.25 27.82
KA
155 i 5 113 I 6 90 I 8 60 i 5 45i 7 31. i 6 22 i 4 17 f 5
Experimental values. Mean and standard deviation based on five results in each solvent mixture. cause of the low degree of association of the solute ion pair. When the values for In KA are plotted as a function of 1 / ~ , the linearity predicted by eq 5 is observed within the stated uncertainties given in Table 111. For moderately polar solvents like acetone-rich cosolvents, the fundamental determinant of the extent of ion pair association by large and weakly solvated ions appears to be the medium dipolarity. Thus, the cohesive experimental evidence above supports the conclusion that PhCN is an effective and reliable analytical probe for the measurement of solvent dipolarity numbers ( T * ) for pure solvents, ideal aprotic cosolvents, and hydrogen bond donor liquids.
ACKNOWLEDGMENT Facilities allowing the completion of portions of this study were made available through the generosity of the Department of Chemistry, Kansas State University. Registry No. Bu4NI,311-28-4;benzonitrile, 100-47-0;acetone, 67-64-1; 2-butanone, 78-93-3; n-butyl acetate, 123-86-4;cyclohexane, 110-82-7;cyclohexanone,108-94-1;1,2-dimethoxyethane, 110-71-4;Nfl-dimethylformamide, 68-12-2;dimethyl sulfoxide, 67-68-5;p-dioxane, 123-91-1;ethyl acetate, 141-78-6;n-hexane, 110-54-3; sulfolane, 126-33-0; acetonitrile, 75-05-8; dichloromethane, 75-09-2;ethanol, 64-17-5;methanol, 67-56-1;2-propanol, 67-63-0. LITERATURE CITED (1) Taft, R.; Abboud, J.; Kamlet, M. J . Am. Chem. SOC. 1981, 103, 1080. (2) Brady, J.; Carr, P. J . fhys. Chem. 1982, 86,3053. (3) Kamlet, M.; et al. J . Chem. SOC.,ferkin Trans. 2 1979, 342. (4) Chawla, B.; et al. J . Am. Chem. SOC. 1981, 103, 6924. (5) Kolling, 0. W. Anal. Chem. 1979, 51, 1324. (6) Riddlck, J.; Bunger, W. "Organic Solvents", 3rd ed.; Why-Interscience: New York, 1970; Chapter 5. (7) Kolllng, 0. W. Trans. Kans. Acad. Sci. 1979, 82,218. (8) Bodenseh, H.; Ramsey, J. J . fhys. Chem. 1983, 6 7 , 140. (9) Fuoss, R.; Hirsch, E. J . Am. Chem. SOC. 1960, 8 2 , 1013. (10) Duthaler, R.; Roberts, J. J . Am. Chem. SOC.1978, 100, 4969. (11) Kamlet, M.;Taft, R.; Carr, P.; Abraham, M. J . Chem. SOC.,Faraday Trans. 11982, 78, 1689. (12) Taft, R.; Plenta, N.; Kamlet, M.;Arnett, E. J . Org. Chem. 1981, 46. 661. (13) Olah, G.; Kiovsky, T. J . Am. Chem. SOC. 1988, 9 0 , 4666. (14) Kolling, 0. W. Anal. Chem. 1981, 5 3 , 54. (15) Sears, P.; Wilhoit, E.; Dawson, L. J . Phys. Chem. 1955, 5 9 , 373. (18) Johari, G. J. Chem. Eng. Data 1988, 13, 541. (17) Evans, D.; et al. J . fhys. Chem. 1971, 7 5 , 1714. (18) Abboud, J.; Tan, R. J . fhys. Chem. 1979, 83, 412. (19) Reynolds, M.; Kraus, C. J . Am. Chem. SOC. 1948, 7 0 , 1709.
RECEIVED for review September 6,1983. Accepted December 5, 1983.