Association Between Polar Molecules Part 1 .-Nuclear Magnetic Resonance Study of the Dipole Association of Hexamethylphosphoramide with p-Substituted Nitrobenzenes BY HIDEAKI FUJIWARA,* TOSHIKAZU TAKABA, YUTAKA YAMAZAKI Faculty of Pharmaceutical Sciences, Osaka University, Yamadakami, Suita, Osaka, Japan AND YOSHIO SASAKI Received 17th March, 1978 Dipole association between hexamethylphosphoramide (HMPA) and p-substituted nitrobenzenes was investigated in non-polar solvents. N,N-dimethyl, 0-methyl, methyl and chloro groups were considered as substituents, and carbon tetrachloride and methylcyclohexane were used as solvents. The data are analysed assuming the formation of a 1 : 1 associate between HMPA and nitrobenzenes. A significant substituent effect on both the equilibrium constant and the induced shift of association is confirmed, and these results are discussed in relation to the dipole moment or the linear combination of the substituent constants ci and u,f.Association between polar molecules in solution is of interest in the study of the solvent and substituent effects on the reactivity of organic compounds. Although such association, often called dipole association, has been investigated by several physico-chemical methods of analyses,1-8 it is too weak for its character to have been clarified in detail. For example, although dipole association of DMSO and DMF with substituted benzonitriles has been investigated by analysing the concentration dependence of integrated intensities of i.r. bands in carbon tetrachloride,2 relatively large experimental errors were considered to have masked the expected substituent effects on the equilibrium constant.Thermodynamic parameters of the self- association of dipoles have been estimated with some aliphatic nitriles, but the values obtained by different groups of investigators 4* are incompatible with each other. N.m.r. studies of the self-association of some ketones, nitriles and nitro compounds reveal that the association constant is very In the present study we considered hexamethylphosphoramide (HMPA), which is 9 p lo the most remarkable of the aprotic polar solvents ; it has a dipole moment of 5.37 D and is soluble in many organic solvents. We studied the association of this substance with p-substituted nitrobenzenes in non-polar solvents by n.m.r. spectroscopy.The results are discussed assuming the formation of a 1 : 1 associate. EXPERIMENTAL Commercial methylcyclohexane was shaken with 10 vol % of the mixed acid to remove a small amount of toluene, and washed successively with water, 20 % sodium hydroxide, and water (three times). It was then dried over Pz05 and % 30 % was discarded prior to distillation, to remove a trace amount of cyclohexane. Commerical carbon tetrachloride of J.I.S. G.R. grade was dried over PzOs and distilled. HMPA was dried over barium oxide and distilled under reduced pressure. p-substituted nitrobenzenes purified by recrystallization were stored in a desiccator containing silica gel and the solvents and HMPA were stored over molecular sieves 4A. 7980 ASSOCIATION BETWEEN POLAR MOLECULES [2,4,6-2H3]Nitrobenzene (isotope purity 95 %) was synthesized from [2,4,6-2H3]aniline by the decomposition of the diazonium salt with Na3Co(N0&.l [2,4,6-2H3]Aniline was obtained by the acid catalysed exchange reaction of H with D in D20.12 N.m.r.spectra were observed with a Hitachi R-22 spectrometer operating at 90 MHz and at 34.loC, and the shift was measured by a frequency counter within an error of +O.l Hz. As an internal reference 0.02 vol % TMS was added to the solvent. No discernible differ- ence was observed in the parameter calculated below, when a solvent peak (a low field peak of the methyl doublet of methylcyclohexane) was used as the reference. RESULTS AND DISCUSSION Charge transfer and hydrogen bonding l3 interactions have been suggested to explain the interaction between HMPA and nitrobenzenes, in addition to the electro- static attraction between dipoles.However, negligible contribution from hydrc en bonding may be proved from the fact that addition of pyridine and triethylar.L ie, which are stronger bases than HMPA, to dilute solutions of p-nitroanisole causes far smaller changes in the l H shift than does HMPA (table 1). If a charge transfer force is dominant in the present case, high-field 13C shifts of the nitrobenzenes, TABLE EFFECT OF ADDITION OF THREE SUBSTANCES ON THE lH SHIFT OF p-NITROANLSOLE IN METHYLCYCLOHEXANE' added substance A2.6b A3.P AOCH3 HMPA 0.046 0.338 0.158 pyridine - 0.01 5 0.024 - 0.010 triethylamine 0.008 0.014 0.006 a Observed changes (p.p.m.) in the shift of 0.03 rnol dm-3 p-nitroanisole on addition of the specified substances by 0.7 mol dm-3 are listed.Plus sign denotes the low field shift on addition. b The numbering corresponds to that for nitrobenzene. A = S(HMPA added)- S(HMPA free). electron acceptors, would occur on association with HMPA, an electron donor. l4 However, this is not the case, since addition of HMPA to dilute solutions of p- substituted nitrobenzenes causes down-field shifts for both lH and 13C shifts in the latter (table 2). Furthermore, no new bands which might be assigned to charge transfer were observed in the electronic spectra of HMPA solutions of the nitro- benzenes. From these results, we may conclude dipole association to be dominant in the interaction between HMPA and the nitrobenzenes.This is also supported by the facts that tetramethylurea and nitroethane, with smaller dipole moment than TABLE 2.-EFFECT OF ADDITION OF HMPA ON THE 'H AND l3C SHIFTS OF p-NITROANISOLE AND P-NITROTOLUENE IN METHYLCYCLOHEXANEa compound citeb A1Hlp.p.m." A*3C/p.p.mec p-nitroanisole 296 0.076 0.18 3, 5 0.350 1.08 OCH3 0.148 1 .oa p-nitro t oluene 2,6 0.071 - 0.01 3, 5 0.258 0.83 CH3 0.064 d a Observed changes in the shift of 0.2 mol dm-3 substituted nitrobenzenes on addition of 1 mol dm-3 HMPA are listed. 13C n.m.r. spectra were measured with a Hitachi R-22 CFT n.m.r. spectro- meter using methyl signal of the solvent as an internal reference. Plus sign denotes the low field shift on association. C A = S(HMPA The numbering corresponds to that for nitrobenzene.added) - S(HMPA free). d Methyl carbon signal was masked by the solvent peak.H. FUJIWARA, T . TAKABA, Y . YAMAZAKI AND Y . SASAKI 81 HMPA, cause smaller changes in the shifts, and that molecules with no appreciable dipole moment, such as benzene, xylene and hydroquinone dimethyl ether, undergo only a small change in the l H shift (< 0.02 p.p.m.) on addition of HMPA under similar conditions to those in table 1. Although association between polar molecules is often represented by an anti- parallel pair of dipoles, determination of the composition of the associate is necessary. The Job plot, a well-known method of determining the composition of the complex, supports the existence of a l(HMPA):l(nitrobenzene) associate in all cases, as is shown in fig.1. 8 a X CA+ CB FIG. 1 .-Job plots for the p-nitroanisole + HMPA system in methylcyclohexane. A = &,sd- 8 ~ . 0, Experimental point for the meta protons, with regard to the nitro group, of p-nitroanisole. A, Experimental point for the methoxy protons of p-nitroanisole. To determine the association constant, concentration shifts were measured, maintaining the concentration of the nitrobenzene as low as possible (x 0.03 mol dm-3) and varying that of HMPA from 0 to 0.7moldm-3; this gave enough changes in shifts for data processing. Carbon tetrachloride and methylcyclohexane were used as an inert solvent. The results are shown in fig. 2 for the p-nitroanisole + HMPA system as an example. In the interpretation of these results, equilibrium of eqn (1) was assumed and root mean square deviations of the calculated [eqn (2)] and observed shifts were minimized by the curve fitting method.? Assumption of eqn (1) may be reasonable because we have treated a dilute solution, where binary collision is dominant, and because of the results of the Job method mentioned earlier.R A+B = AB In eqn (2), dcalcd = calculated shift of A, dA = shift of A in the free state, J A B = shift of A in the associated state, AAB = JAB-aA, C, = initial concentration of A, CB = initial concentration of B and K = equilibrium constant. t The calculation was performed with a NEAC 2200 model 700 computer at the Osaka University Computer Centre. A library program for the minimization of functions using derivatives was employed.82 ASSOCIATION BETWEEN POLAR MOLECULES As for the value of aA, data for samples without HMPA were used.K and aAB were reproducible to within M 5 % and f 1 Hz, respectively, on repeated runs, and the root mean square deviation was always < 0.2 Hz. The shift of ortho protons with regard to the nitro group exhibited insufficient concentration dependence to allow data processing, therefore, the data discussed below are for meta protons unless otherwise specified. In the cases of p-nitroanisoIe and N,N-dimethyl-p- nitroaniline, exceptionally, the methyl protons showed such a large concentration dependence (x 10 Hz) in methylcyclohexane as to make the calculation reliable, 7.0- 6.81 , , 0 0.4 0.8 CHMpA/mOI dm--3 FIG. 2.-Concentration shifts of the meta protons, with regard to the nitro group, of p-nitroanisole.0, Experimental point; -. calculated curve with K = 3.50 dm3 mol-l, 6m = 7.302 p.p.m., and 6~ = 6.827 p.p.m. and the K values thus determined were comparable with those obtained for the meta protons (table 3). The K and AAB values for each substituted nitrobenzene were almost equal in carbon tetrachloride and methylcyclohexane. Thus, the following equations were obtained by the least squares calculation based upon linear relation- ship, K(Cc1,) = 0.3 K(c6Hl1CH3), A.4B(CC14) = O.78 AAB(C6H1 1CH3), r.m.s.d. = 0.12 dm3 mol-l, r.m.s.d. = 0.04p.p.m., where r.m.s.d. denotes root mean square deviation between the observed and the calculated values. Small values of K in carbon tetrachloride, compared with those TABLE 3.-ASSOCIATION CONSTANT (K) AND BOUND SHIFT (AAB) DETERMINED FOR THE DIPOLE ASSOCIATION OF HMPA WITH P-SUBSTITUTED NITROBENZENES IN CARBON TETRACHLORIDE AND METHYLCYCLOHEXANEu K A+B K AAB substituent (in carbon tetrachloride) (in methylcyclohexane) [2,4,6-'H31 0.47 0.388 1.65 0.448 P-CH, 0.57 0.362 1.43 0.481 p-c1 0.66 0.460 2.52 0.542 p-OCH3 1.10 0.350 3.40 0.482 (3.37 0.2 1 4) p-N(CH312 0.97 0.252 2.46 0.392 (2.30 0.206)b a K and A m are determined from the data of the meta protons with regard to nitro group, the These values are estimated units being dm3 mol-I for K and p.p.m.for AAB. Am = SAB- 8 ~ . from the data for the methyl protons in the substituent.H. FUJIWARA, T . TAKABA, Y . YAMAZAKI AND Y . SASAKI 83 in methylcyclohexane, suggest the preferred solvation of polar molecules with the solvent molecules in carbon tetrachloride.As shown in fig. 3, AAB was approximately linearly related to the dipole moment p of p-substituted nitrobenzenes, AAB being small for the strong electron donating group. This suggests that the substituent partly compensates for the decrease in 0.6 d 0 . 4 - 2 -5 a 0.2- 0.61 1 ' I I d 4 a a -2 O.Zl*, 0 3 6 PIDebYe FIG. 3.-Correlation between the induced shift (Am) and the dipole moment (p). p is cited in ref. (17) ; the value in benzene is adopted consistently. Am = 8m- SA. Solvent : 0 and -, methylcyclohexane ; A and - - -, carbon tetrachloride. Substituent for nitrobenzenes : (1) [2,4,6-2H31, (2) P-CH3, (3) P-CL (4) P-OCH3, (5) p-N(CH3)2. electron density induced by the association. Plots of AAB against a$ and oi l 5 (fig.4) show that AAB approximately increases with increasing o$ and oi values. The two parameter treatment of AAB given by eqn (3) is summarized in table 4 (3) AAB = acr, +boz +c. The small values of the fitting parameter show that the treatment applies satisfactorily. This result suggests both inductive and mesomeric effects for AAB. / / / / / A' 0.6 E' 0.4 4 3 a d 1 3.; A 0 0.2 0.4 ai FIG. 4.-Plots of Am against (a) 0: and (b) oi. The numbering corresponds to that in fig. 3. Solvent : 0 and -, methylcyclohexane ; A and - - -, carbon tetrachloride.84 ASS 0 CIA T I 0 N BETWEEN P 0 LA R MO I, E C UL E S TABLE 4.-TWO PARAMETER TREATMENT OF THE VALUES BY THE EQUATION Am= aai+ba,++c solvent U fa carbon methyl- tetrachloride 0.25 0.29 0.39 0.01 cyclohexane 0.22 0.16 0.48 0.04 root mean square deviation root mean square of the data' a Fitting parameter f = Three possible explanations of the association shift have been so far pointed out :6* (1) electric field effect of the partner molecule, (2) magnetic anisotropy of the partner molecule and (3) variation in the magnetic anisotropy of the intramolecular group by the association.In the present case, (3) may be excluded, since AAB is 1 -0.6 -0.3 0 0: FIG. 5.-Plots of In K against a t The numbering corresponds to that in fig. 3. Solvent : 0 and -, methylcyclohexane ; A and - - -, carbon tetrachloride. measured at the meta position of the nitro group. If (2) is a dominant factor, associ- ation shift will be governed by the relative orientation of the two molecules participat- ing in the association, and a correlation such as eqn (3) may not be realized.In this manner the low field shift of AAB and its substituent effect are explained by the TABLE 5.-TWO PARAMETER TREATMENT OF THE 1nK VALUES BY THE EQUATION h K = pai+qa,+ + r solvent seriesa P r f b carbon I 0.80 - 1.34 -0.68 0.30 tetrachloride I1 0.43 -2.73 -0.75 0.01 methyl- I 1.58 -0.87 0.44 0.17 cyclohexane I1 1.27 - 2.01 0.38 0.1 1 a Data for N,N-dimethyl-p-nitroaniline are excluded from series 11, but are included in I. root mean square deviation root mean square of the data' b Fitting parameter f =H . FUJIWARA, T . TAKABA, Y . YAMAZAKI A N D Y . SASAKI 85 electron density decrease in the phenyl ring, which is induced by the electric field effect of HMPA, and is partly compensated for by electron donation from the sub- stituent.In contrast to AAB, there is no discernible correlation between K and p, but a In K against a,+ plot suggests a linear correlation. These facts indicate a local contribution to the association, since the Hammett 0 constants are related to the polarization of the nitro group.16 As a model of the associate, an antiparallel pair of dipoles, such as those below, is consistent with the results obtained above ; a parallel pair model seems improbable, since a 31P n.m.r. study l 8 does not prove high field shift by association which is expected for the model. Further study on the structure of the associate and the substituent effect is in progress by the measurement of 3C n.m.r. spectra, including several p-substituted benzonitriles.This work was supported by a grant-in-aid for scientific research from the Ministry of Education, Japan. B. J. Bulkin, Helv. Chim. Acta, 1969, 52, 1348. C. D. Ritchie, B. A. Bier1 and R. J. Honour, J. Amer. Chem. Soc., 1962, 84,4687. A. V. Sechkarev and G. E. Trostentsova, Optics and Spectroscopy, 1973, 34, 707. W. Dannhauser and A. F. Flueckinger, J. Phys. Chem., 1964,68, 1814. T. A. Renner and M. Blander, J. Phys. Chem., 1977,81,857. T. Yonezawa and I. Morishima, Bull. Chem. SOC. Japan, 1966,39,2346. J. A. Riddick and W. B. Bunger, Organic Solvents (John Wiley, New York, 1970). ti H. Saito, Y. Tanaka, S. Nagata and K. Nukada, Canad. J. Chem., 1973,51,2118. * R. W. Taft, G. B. Klingensmith and S. Ehrenson, J. Amer. Chem. Soc., 1965, 87, 3620. lo H. Normant, Bull. SOC. chim. France, 1968, 791. l 1 F. Lagenbucher, R. Mecke and E. D. Schmid, Lieb. Ann. Chem., 1963, 669, 11. l2 Y. Sasaki, A. Takahata, M. Yoritaka, H. Kawaki and Y. Okazaki, Chem. Pharm. Bull., 1974, l 3 A. J. Dale, Acta Chem. Scand., 1970, 24, 3403. l4 I. Prins, J. W. Verhoeven and Th. J. de Boer, Org. Magnetic Resonance, 1977,9, 543. l 5 Y. Yukawa and Y. TSU~O, Nippon Kagaku Zasshi, 1965, 86, 873. 22, 50. M. Avram and GH. Mateescu, Infvared Spectroscopy Applications in Organic Chemistry, trans- lated by L. Birladeanu (Wiley, New York, 1972), p. 308. l7 A. L. McClellan, Tables of Experimental Dipole Moments (W. H. Freeman, San Francisco, 1963). '* H. Fujiwara, unpublished results. (PAPER 8/500)