Correlation analysis of IR, 1 H NMR and UV spectral data of alkyl and aryl 4,6-disubstituted-3-cyano-2-pyridones. Part I

Spectral characteristics of the two series of previously synthesized and identified 4,6-disubstituted-3-cyano-2-pyridones were determined, and corresponding 1 H NMR chemical shifts and IR and UV frequencies were correlated with LFER parameters. A variety of substituents were employed for both alkyl and aryl substitution, and fairly good correlations were obtained, using simple Hammett and Hammett-Taft dual parameter equations, as well as the more sophisticated multiparameter regression approaches. It was established that both polar and steric effects influence the spectra of the investigated compounds.


Introduction
Our previous investigations into the chemistry of substituted 3-cyano-2-pyridones included the synthesis and identification of new compounds [1][2][3][4][5] , the study of new catalysts for the reaction, and the mechanisms of both chemically and enzyme catalyzed reactions 3,[6][7] .These and other papers and communications were summarized in a review 8 .In a recent publication, the UV/VIS absorption spectra of arylazo substituted 3-cyano-2-pyridones in various solvents have been investigated 9 .Other investigators studied the tatomerism of some substituted 2-pyridones by UV/vis, 1 H and 13 C NMR spectroscopic methods. 10t is well known from the literature that 2-pyridones of structure similar to those investigated here, have a broad spectrum of biological activity, and to mention only a few, exhibited vasodilatory, antioxidant, antidiabetic, antimicrobial and antiviral activity 8,11 .Therefore we considered it worthwhile to investigate the effects of substituents in the positions 4 and 6 in the title compounds (Fig. 1) on the spectral features of the groups in investigated molecules, which might be either directly or indirectly involved in molecular interactions in living organisms.A large number of spectral data relating to substituted 3-cyano-2-pyridones, accumulated in the previous investigations, were correlated in the present work using a variety of LFER models, conventionaly used for the study of structure-reactivity and structure-property relationships.It was assumed that it should be possible to find an adequate approach to study the transmission of substituent effects in the multi-substituted 2-pyridone nuclei, considering that their aromaticity has been established 12 .Problems arose from the difficult assignement of the IR stretching frequency bands, low solubility of investigated compounds in the common solvents used for recording the 1 H NMR spectra, and the close values of the UV absorption maxima.However, some fairly good correlations were obtained, considering the possibility of interactions between substituents and probable deviations from planarity in some of the investigated molecules.It was always initially attempted to use a simple Hammett Equation as presented in Eq. 1.a., but it was frequently more appropriate to use other approaches, like the Hammett-Taft (Extended Hammett Equation) DSP model, Eq. 1.b., and Swain-Lupton Eq. 1.c., which are usually given in the literature in their general form 13,14 .In these models, s is the measured spectral characteristics, σ m/p 15 , σ I , σ R and υ 16 , F and R 15 are substituent constants, ρ, ρ I , ρ R , f and r are the corresponding calculated proportionality constants, which in a broad sense reflect the sensitivity of the spectral characteristics to substituent effects, and s o is the intercept.On certain occasions, when other models failed, combined multiparameter equations were applied, the method known to be used before 17,18 , and with the same precision as obtained here.
For the correlation of spectral data for 4-alkyl substituted-6-methyl-3-cyano-2-pyridones we also applied Charton's steric constants υ corrected for the -C-C-branching in the the corresponding alkyl group 19 .The υ corr parameter stands for υ(n 1 +n 2 ), n 1 and n 2 being the number of branch points at the alpha-and beta-C-atoms in the alkyl groups, respectively.In some instances, we considered it appropriate to take into account the conformational effects, because of the close proximity of the alkyl groups and the cyano group.

4-Alkyl-6-methyl-3-cyano-2-pyridones
The IR stretching frequencies of characteristic groups in alkyl substituted 3-cyano-2-pyridones 1a-g are shown in Table 1.As the correlation with Hammett σ constants failed, it was considered appropriate to correlate first the very distinctive CN group frequency with steric constants υ and Es, because of its ortho position to the 4-alkyl group, as the steric effect was to be expected.However, the best correlation has been obtained with υ corr and this is presented in Eq. 2. It is evident from the failure of correlations with the electronic effect substituent constants, and fairly good one with υ corr steric parameter, that the steric effect of the alkyl groups in the ortho-position to the CN group is dominant.Considering that inductive effect of the alkyl groups is weak and practically equal, as examplified by the σ I values (Table 5, section A), this prompted us to introduce the Swain-Lupton parameters as in Eq. 3, together with the υ corr , particularly because of the recent oppinion which supports the preferential use of the field effect.
The above calculation is not statisticaly perfect, in view of the high standard deviation of the F term, although the standard deviation value and the Fisher test (38.82)are affirmative.Evidently, the steric effect of the ortho cyano group restricts the free rotation of the bulkier alkyl groups and hence the iregular field effect and the negative sign of the R-term.It should be mentioned that correlation of the same data with only F and R parameters was completely unacceptable (R=0.40,s=3.10,F=0.36).
Correlation of the ν CO values for the 1a-g series from Table 1 was logically tried with Hammett σ m constants, as in Eq. 4, but this was improved by the inclusion of the steric parameter υ shown in Eq. 5. )

ARKAT
The neccessity to include the steric factor in the calculation of the CO group vibration frequencies shows that an interference of the bulky alkyl groups with the CN substituent occurs.The correlation is even better with υ corr as in Eq. 6, which further confirms the importance of conformational effects.It is possible that this causes attenuation of the resonance effect transmission, which is evident also from negative signs of the steric parameters in Equations 5 and 6.
Correlations of the pyridone ring stretching frequencies as influenced by alkyl groups in position 4 of the heterocyclic 2-pyridone nuclei in the 1a-g series, were expected to be reasonably good with the corresponding steric constants constants υ and the Es values, considering the rather weak electronic effects of this groups.However, these models failed, but fairly good results were obtained with the corrected Charton constant υ corr which is shown in Eq. 7. It appears that these constants, corrected for the C-C branching at the alpha-and beta-carbons in the substituent alkyl groups, also account for the hyperconjugation and its effect on the electron density in the heterocyclic ring.This is consistent to the statement relating to the Eq. 6, that the restriction of the free rotation of the bulkier alkyl groups also affects hyperconjugational resonance interaction.

ARKAT
For the recording of 1 H NMR chemical shifts in the series 1a-g the only available hydrogens were those of the 6-methyl group and the heterocyclic 5H.The values for the 6methyl hydrogens and the 5H were very close in the spectra taken in CF 3 COOH, almost equal, and the correlations were pointless.The investigation was also hindered by the low solubility of the compounds in DMSO-d 6 and CDCl 3 , and it was possible to determine the 5H chemical shifts for four compounds only (Me, Et, i-Pr, t-Bu).These results are presented in Table 1.The data were correlated with Charton's constants's υ, but the regression coefficient was very low (R= 0.909).It was much better with υ corr as shown in Eq. 8.An even better result was obtained with a two parameter equation using both υ and υ corr , R=0.998, s=0.02, but the use of that model was not justified for such a small number of data points.Nevertheless, the qualitative conclusion about the effects of the alkyl groups in the ortho-position both upon the hydrogen in the position 5 and to the cyano group is the same as for the IR spectra, namely that the governing factor is the branching in the alkyl group.The UV spectra for 4-alkyl-6-methyl-3-cyano-2-pyridones were recorded in ethanol in the interval 270-400 nm , but the ν max values were very close and impossible to correlate.

-pyridones (2a-f and 3a-f)
There was initially a dilema what sets of parameters to use for correlations of the effects of substituents in the phenyl ring which are substituents itself on the positions 4 and 6 in the pyridone nucleus.The obvious choice would be the same as those for substituted phenyl groups (Table 5, section C) 15,16 but the corresponding calculation results were not always satisfactory, with few exceptions.Comparison of the two sets of parameters, m/p and F/R scales, 15 both for "ordinary" Hammett substituent constants and for phenyl substituted rings, show that the values of the later are smaller.Except for the unsubstituted phenyl group, for which the substituent constant is determined from the dissociation of substituted benzoic acids, 13 the substituent constants for other substituted rings were determined from dissociation of phenylacetic acids, 16 and this may be the reason for the differences in parameter values.For the substituted phenyl groups, both inductive and resonance effects are attenuated, and probably not proportionally transmitted through the methylene group, and hence the unsatisfactory correlations.Alternatively, one could take substituent effects as transmited from m/p positions in the phenyl ring to the atoms 4 and 6 in the pyridone ring.In these calculations, for the unsubstituted phenyl groups σ constants and steric constants had to be zero, as for hydrogen, taking that the steric effect of the phenyl group is constant.This approach is again not sufficiently precise, as the observed pyridone substituents are in ortho-, meta-and para-positions.This is probably the ARKAT reason why the correlations with the F/R scale are statistically more satisfactory, because this model describes substituents in the meta-and para-positions in the same way.The characteristic IR stretching frequencies of the CN, CO and ring as influenced by the substituted aryl groups in positions 4 and 6 in the 2-pyridone nucleus are given in Table 2 (2a-f and 3a-f series).In all the equations which follow, the measured frequency item has a superscript which denotes the position of the aryl substituent in the 2-pyridone nucleus, and the subscript which shows the group in the molecule whose frequency has been measured.
Correlations for the CN group were first done with σ m/p constants for the phenyl group (Table 5, section C), for compounds in the 2a-f and 3a-f series, but the results were poor.The use of Hammett σ m/p constants (Table 5, section B), according to position of substituents in the phenyl groups in 4 and 6 phenyl substituted 2-pyridones, gave satisfactory result for the 6isomer, and with the steric factor υ for the 4-isomer, as shown in Eqs. 9 and 10.

ARKAT
The correlation was also performed with F and R parameters ( With both sets of parameters (Table 5, section B), the results are better for compounds 3af, even with steric υ included for the compounds 2a-f (Eq.9), to compensate for the strong steric interaction with the orto phenyl substituent.Besides, by taking the value υ=0 for the unsubstituted phenyl group, it is assumed that the steric effect of the phenyl nucleus is constant in the series, which is probably not entirely true.
It was logical to expect that the correlations for the CO frequencies should be first tried with the Hammett σ m constants, but the results were poor.Better correlations were obtained with the extended Hammett equation but σ m and Charton's steric parameter υ had to be included (Table 5, section B) but the results were unacceptable, on the ground of statistical insufficiency, because of too many parameters in the multiparameter equations for only six data points.Swain-Lupton approach was indicated but it was successful only for the 6-aryl isomer (3a-f series) with the substituent parameters for the phenyl group (Table 5, section C), and this is presented in Eq. 12. ) This equation is also not statistically perfect, the scattering of the data points being high, but at least an idea of the effects of substituents on the CO group in 2-pyridones could be obtained.
Because of the nature of the probable interaction of substituted aryl nuclei with the 2pyridone heterocyclic ring, it was considered appropriate to correlate corresponding IR frequencies (Table 2) with the extended Hammett equation (DSP equation).The correlations, for the series 2a-f, were only fair, but became better, by inclusion of the steric parameter υ, which lowered its statistical reliability.Therefore the calculation for this series was performed using the Swain Lupton parameters ( This correlation is better in view of correlation coefficient, overal standard deviation, as well as the overal reliability (Fisher test 40.82 as compered with 25.18 for Eq.14) but less favourable if standard errors of F and R terms are considered.
There were not enough data for 1 H NMR spectra in DMSO-d 6 , for 2a-f and 3a-f series of the substitution types due to the insolubility of the compounds, and the spectra for both were taken in CF 3 COOD. 1 These data are given in Table 3.However, correlation for the heterocyclic 5H was not statistically acceptable for series 2a-f, and not at all possible for series 3a-f.The correlations for the 1 H NMR data for methyl hydrogens for both series, were attempted with σ m constants for the substituted phenyl groups (Table 5, section C) but only that for the 4-methyl group in 6-phenyl substituted isomer was successful and is presented in Eq.16:The calculations using DSP equation and Swain-Lupton equation, gave rather inferior correlations of the 1 H NMR data, the regression coefficients amounting only up to 0.850 and with high standard errors.The results were much better if Charton's steric parameter was included, but these calculations were objectionable because of statistic insuficiency.However, this should be noted because it is still another proof of how important are the steric effects in investigated compounds.
The UV spectra of 4-and 6-substituted aryl, 6-and 4-methyl derivatives were determined in ethanol in the interval 200-400 nm, and the corresponding ν cm -1 10 -3 values are given in Table 4.The correlation for 4-substituted aryl-6-methyl-3-cyano-2-pyridones (series 2a-f) was fairly good with Hammett σ p values, Eq. 17  The correlations of the 4 UV ν for the series 2a-f were done also with the three parameter equations comprising σ I , σ R , F and R, but needed the inclusion of the steric parameter υ.In both ARKAT calculations the correlation coefficients were acceptable but the standard errors were large.However, all calculations for 6-substituted aryl-4-methyl-3-cyano-2-pyridones (series 3a-f) were poor.
The apparent differences in the behaviour of the 4-and 6-aryl substituted derivatives in the transmission of substituent effects, as visible in their UV spectra, could be explained by greater symmetry of the 4-aryl substituted molecules and the electronic effect disturbances because of the close proximity of the three substituents, namely 6-substituted aryl, NH and CO group in the 6-aryl substitution.
It was of interest to try the intercorrelation of the data for IR ring stretching frequencies from Table 2. and the UV absorption maxima from Table 4, which we believed to reflect the similarity of substituents' interaction with π-electrons in the 2-pyridone nuclei.The correlation for the 4aryl substitution (series 2a-f) is fairly good, as presented in Eq. 18, while for the 6-aryl substitution the correlation was poor, probably due to less symetry of the 6-aryl substituted molecules 3a-f.Experimental Section General Procedures.All substances whose spectral characteristics are presented in the present paper and were used for correlations with the LFER parameters, were synthesised by literature methods or by analogy to those, sometimes by modified procedures.They were fully characterized by IR, NMR and UV spectra in our previous communications [1][2][3][4][5] and some of the spectra were rechecked or redetermined on more modern equipment.
The LFER equations 13 and substituent constants 15,16 used for calculation were taken from the standard texts on the subject.Those actually used are presented in Table 5.
The instruments used in the investigation are as follows: IR spectra were recorded on a Bomem MB 100 FTIR spectrophotometer.UV data were obtained using a Shimadzu UV A 160 spectrophotometer. 1 H NMR spectra were determined in an appropriate solvent on a Varian-Gemini 200 MHz spectrometer using TMS as internal standard.

Conclusions
The pyridone nucleus of relatively low aromaticity, is very heavily substituted and, practically, only one position the 5H, is free.There are the possibilities for substituent interactions, both electronic and particularly steric, not always adequately accounted for by the used substituent constants.However, the general trends can be estimated.One conclusion is that in both 4-and 6-ARKAT aryl substitution, the phenyl and heterocyclic nuclei are not co-planar, and that deviations vary with the type of substitution.Hence, the high standard errors in some of the models employed, although the regression coefficients were fairly satisfactory.We believe that these shortcomings will be overcomed in our further investigations of the effect of substituents on the NH group in the 3-cyano-2-pyridone nucleus, and the study of the 13 C NMR spectra.

Table 5 ,
section B), but it was statistically acceptable only for the 6-phenyl substituted compounds (3a-f series):

Table 5 ,
secton C) as in Eq.13: satisfactory from statistical point of view.The correlation for 3a-f series with DSP equation is fairly good, probably because of less steric perturbancies, Eq.14: