The Nitro Group as Substituent Otto Exner Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic, 166 10 Praha 6, Czech Republic Tadeusz Marek Krygowski Department of Chemistry, University of Warsaw, 02 093 Warsza wa, Pasteura I, Poland 1 Introduction If a structural unit is denoted as a ‘substituent’, this usually has one of the two following meanings: (1) The substituent is a smaller part of a molecule which can be introduced by a simple chemical operation, particularly when it can directly replace a hydrogen atom. (2) The substituent is a smaller and less important part of a mole- cule which influences the properties of the molecule in a quantita- tive sense but does not alter its general chemical character: the latter is controlled by another group present: the functional group’ (or the reaction site).The nitro group is a substituent par excellence, both typical and important, according to either definition. For instance, 4-nitrophe- no1 will be viewed in most circumstances as a substituted phenol, much less often as a substituted nitrobenzene. It can be prepared easily from phenol, but not simply from nitrobenzene. Its chemical and physicochemical properties are more closely related to those of phenol than to those of nitrobenzene. This view, however, may change if an appropriate physicochemical property is studied. For instance in electroreduction the nitro group will act as the functional group. On the other hand, the nitro group is a ‘strong’ substituent since the differences in acidity, reaction rates and spectral shifts between nitrophenol and phenol are large compared to the effects of other subs tituents.In this review the term nitro group is used in a narrow sense as NO, bonded to a carbon atom. We shall deal only with the second aspect of substitution, and particularly with the quantitative mea- sures of the substituent strength estimated by substituent constants u1 -3 as well as by other parameters; the electronic and geometric structure of the nitro group must be also mentioned. We give much attention to nitrobenzene since it has been more extensively inves- tigated than any other nitro compound. Otto Exner, born in Praha, formerly Czechoslovak Republic, received a PhD degree from the Institute of Chemical Technology in I951 and a DSc degree from the Czechoslovak Academy of Sciences in 1961.He has worked in several Institutes of this Academy, and has been forced several times to change his employment on polit- ical grounds, and partly also because his interests lay between organic and physical chemistry. He has lectured at several Universities in Czechoslovakia, since 1969 as Professor of Organic Chemistry, and has had appointments as visiting Professor in Ituly, Sweden and France. His main scientific interest is in correlation analy- sis, particularly the Hammett equation, isokinetic relation-ships, and reactivity-selectivity principles: secondary interests are dipole moments, conforma- tions and hydroxylamine deriv- atives.2 Properties and Electronic Structure of the Nitro Group The nitro group may be derived formally from N,O, which is a strong oxidizing agent. If the nitro group is attached to a hydro- carbon of any kind, it changes the electron affinity of the molecule significantly. Alkanes and benzene are not reduced electrochem- ically, whereas their nitro derivatives are, with rather low values of formal potentials of reduction. The earlier formulated structure 1 of the nitro group has been replaced by the resonance of two degenerate forms 2 *3 which describe in a somewhat cumbersome way the simple fact that the two oxygen atoms are equivalent. In particular also the canonical structure 4 comes into consideration; its significance is that the second positive charge is delocalized over the moiety to which the nitro group is attached.The formally positive charge on the nitro- gen atom explains the strong electron-attracting power of the whole substituent. This is manifested by the high value of the group electronegativity and of the dipole moments (see later sections). Recently also a contribution from the biradical form 5 was consid- ered4 and estimated in the case of p-nitro derivatives of aniline and phenol to be ~ignificant.~ Even the old structure 1 could again be taken into account since in fact the NO bond lengths are required for a double bond.4 For chemical consequences the form 4 is par-ticularly important and instructive; it is postulated in textbooks, but its weight in p-nitrophenol and p-nitroaniline was found recently5 to be only about 1%.This is in line with many other theoretical and experimental studies of nitrobenzene and its derivatives,6--” which all have shown a limited importance of the form 4. c13 C 23 c 33 +,Oe /Om-N@ -N \oe \om c 43 c 5) Tadeusz Marek Krygowski was born in Poznah, Poland and received his PhD degree in 1969 and a DSc degree in 1973, both from the Department of Chemistry of Warsaw University. Since I964 he has been working at this University, from 1983 as a Professor of Chemistry. He has lectured at many universities in many countries, serving as an invited Professor in Canada and France. Elected as presi- dent of the Polish Chemical Society in 1994, his muin inter- ests are: ion-pairing in organic electrochemistry, solvent and substituent efsects, structural organic chemistry, and studies on aromaticity.71 Table 1 Geometry of the nitro group attached to various moieties Bond Molecule Bond length Id/& angle (") CN NO NO ON0 Nitrobenzene,Io EP 1.486(2) 1.2234(4) 125.32(8) Nitrobenzene," XD (low temp.) 1.467(1) 1.227(1) 123.2(1) Corrected for liberation 1.477 1.229 1.233 p-Nitroaniline,12 XD 1.434(2) 1.227(2) 121.6(2) N,N-Diethyl-p-nitroaniline,13XDh 1.437(5) 1.234 1.223 12 1.9(3) 1.429(5) I .221 1.232 12 1.7(4) Corrected for liberation 1.441 1.243 1.232 1.434 1.229 1.241 p-Nitrophenolate anion' XD 1.418(5) 1.241(2) 1.247(2) Nitromethane,I4 MW 1.489 1.224 125.3 XD: X-ray diffraction; ED: electron diffraction; MW: microwave measurements. Two independent molecules in the asymmetric unit.3 Geometry of the Nitro Group The geometry of the nitro group does not depend significantly on the nature of the moiety to which it is attached. Data collected in Table 1 present this clearly. The most sensitive parameter is the length of the C-N bond, which may serve as an approximate measure of the resonance effect of the nitro group, interacting via this bond with the moiety to which it is attached. Thus if a partial CN double bond is induced, it means that the canonical structure 4 contributes significantly to the description of the molecule. The C-N length in nitrobenzene is almost the same as in nitromethane, in spite of the sp3 character of the Catom in the methyl group.In electron diffraction'" or low-temperature X-ray diffraction measurements" on nitrobenzene on the one hand, and microwave determination of the geometry of nitromethane14 on the other, the difference is statistically insignificant. It might be concluded that the resonance effect of the nitro group in nitrobenzene is very low or even practically zero. In other words, the structure is described sufficiently by the canonical structures 6 and 7 while structures 8-10 have low weights. As mentioned this need not apply when the nitro group is attached to strongly electron-donating moieties; then structure 11 may be of importance.C 63 c 73 c 7a) C 83 4 Electronic Substituent Effects and Substituent Constants 4.1 Hammett Constants The electron-attracting character of the NO, group is shown in a simple and convenient way by means of Hammett constant^,^ urn and up.Defined originally as the substituent effect on the acidity of substituted benzoic acids in water, these constants can be viewed as experimental quantities. Subsequently, they have been obtained also from other reactions or as statistical mean values from many reactions.' Table 2 gives some examples and the values agree quite well. The electronic effect of NO, is thus regular and well predict- able. In addition it is strong, approaching almost the end point of the scale. (Only 7% of uncharged substituents, show higher values, and CHEMICAL SOCIETY REVIEWS, 1996 Table 2 Some selected values of substituent constants CT for the nitro group Assumed substituent Representative Kind of effect values determination" I + reduced M 0.7 1 Statistical3 0.7 1 'Preferred*' 0.7 1 pK in water3 I+M 0.81 Statistical' 0.78 'Preferred2' I + enhanced M 0.78 1.23 pK in water3 pK aniline3 1.28 pK p heno13 I or F 0.76 pK acetic acid' 0.68 pK quinuclidine' 0.78 h 0.65 Indirect2 0.64 I9F NMR shift2 0.66 Calculated] M 'normal' 0.17 IR intensity3 0.13 Indirect2 0.0 Indirectt9 0.16 IyFNMR shift2 0.19 Calculatedz0 M enhanced 0.46 Indirect's M reduced 0.0 Empiricalz2 Electronegativity 0.40 Calculated26 0.46 Calculatedz2 Polarizability -0.26 Calculated2* a Secondary literature sources are cited wherever possible.From 4-nitrobicyclo- [2.2.2]octanecarboxylic acid.' most of them are rather exotic.) Since nitro compounds are easily available, this group has been of decisive importance in formulat- ing and verifying the validity of the Hammett equation; in fact an irregular effect of some weaker substituents, may be masked by the strong effect of the nitro group. This group has always been present in any set of substituents recommended for studies of reactivity or other properties, the so-called minimum basis set. '5 The use of meta-and para-substituted derivatives of benzene as model compounds has the rationale that direct steric interaction is eliminated.' An additional constraint for the validity of the Hammett equation is the absence of direct resonance between the substituent and the reaction centre ('through resonance' as in p-nitroaniline). If this constraint is abandoned, the so-called dual con- stants are derived,' denoted @ or u-.In the case of the nitro group only u-values come into consideration.According to Table 2 they are rather different from normal up:the difference is usually explained in terms of resonance described by the canonical stmc- ture 11. Accordingly, the normal constants upare also assumed to be composed of an inductive and a smaller resonance part and enormous effort has been given to their quantitative separation.',2 In the case of the nitro group the inductive effect is evidently much more important and any kind of separation must begin with it.4.2 Inductive Substituent Effect Evaluation of a more or less purely inductive effect requires the replacement of the benzene ring considered for the Hammett equa- tion by a rigid alicyclic system' such as bicyclo[2.2.2]octane or quinuclidine. Remarkably, even simple aliphatic systems, for instance derivatives of acetic acid, give concordant results: the direct steric interactions are negligible for common substituents.16 The resulting constants are denoted as inductive, a,,but some authors2 prefer the term field constants, uF.However the question of how the effect is transmitted is immaterial and in fact ill formu- lated.17 More important is the fact that several model systems give the same relative result, as shown for the NO, group in Table 2.To obtain numerically concordant results each kind of the substituent constant must be multiplied by normalizing factors. Therefore, the agreement in Table 2 depends also on these factors which in turn depend in part also on the nitro group itself. The agreement for one THE NITRO GROUP AS SUBSTITUENT-O. EXNER AND T. M. KRYGOWSKI substituent does not tell much; any disagreement could be observed only if the behaviour of the nitro group were rather different from that of the other substituents. This is not the case: again the effect of NO, is strong and regular. A purely theoretical approach18 to the inductive effect uses pseudo-molecules with a non-bonded substituent at a fixed dis- tance: for instance the enthalpy of the isodesmic reaction, equation (l), can be calculated. Even here the resulting constants must be normalized.Examples of these results are given in Table 2 as 'cal- culated' values. NH? NH, NH: NH,'f4.sA w H+H H+H1 I I IX H H X 4.3 Mesomeric (Resonance) Effect Since the inductive effect is omnipresent, any evaluation of the mesomeric effect always means that two similar systems need to be compared.' When the inductive effects can be assumed to be equal, the mesomeric effect is obtained by subtracting the two values. In the case of the NO, group the inductive effect is strong, and sub- tracting two almost equal values is not dependable, as stated in clas- sical textbooks.However, the pioneering work of Taft15 yielded a resonance constant uR,small compared to uIbut not negligible (Table 2). The calculation was complex; the two systems compared were substituted acetic acid esters and substituted benzoic acids; the inductive effects were not equal and were normalized with refer- ence to bicyclo[2.2.2]octane- 1-carboxylic acids. One of the present authors tried to improve this normalization with the result that the mesomeric effect of the nitro group is near to zero in benzoic acids;I9 it was concluded that even the resonance in nitrobenzene corresponding to the canonical structures 8,9 and 10 has almost zero weight with respect to the accuracy of common experimental approaches.This finding was strongly opposed1s,21 but was later rediscovered by Taft himself and used particularly for gas-phase acidities.,, With benzoic acids as models, it is probably not possible to obtain more accurate results, nor to decide whether the effect is small or 'practically zero'. This picture is changed when substituted phenols or anilines are used as model compounds. From the constants a;one can define constants u;which are certainly not zero (Table 2). These constants describe the substituent effect of the nitro group in several similar systems and can be compared with the effects of other acceptor groups; hence they could be accepted as a measure of the resonance effect. Evidently, the canonical structures 8-11 are much more important for derivatives with a conjugated electron-donating group.These results were independently confirmed by a novel approach, based on determination of the weights of canonical struc- tures from experimental bond lengths,23 the so-called HOSE model. The results obtained for p-nitroaniline, its N,N-diethyl derivative and p-nitrophenolate anion are given in Table 3 and compared with those for nitrobenzene. In order to use nitrobenzene as a reference, for which there is no possibility of forming structure 11 by a through-resonance effect, only structures 6-10 were considered for all the compounds. Since the N-0 bonds are rather insensitive to substituent effects in which the nitro group is involved, the N-0 bonds were not considered in the calculations.Even if the HOSE estimates of the weights of canonical structures are only approxi- mate, it is clear that a considerable change of the resonance effect is observed when comparing nitrobenzene and the p-nitrophenolate anion: an increase from 40 to 51% for the sum of the weights of 8, 9 and 10 is observed. It is significant that structure 10, with a sym- metrical (C,) localization of double bonds, increases most signifi- cantly (from 13 to 21 %). This may be connected with two additional kinds of interactions: (i)a contribution of structure 11 which yields the same changes in geometry of the system under study, and (ii) an important contribution of a structure, in which the field effect of the nitro group causes the lone pairs of electron-donating substituents to interact more strongly with the n-system of the ring.4,5,22 We con- clude that the real structure of these para-derivatives is described Table 3 Weights of canonical structures indicating resonance interactions in nitrobenzene and its derivatives (HOSE Molecules Weights of canonical structures 6+6a 7+7a 8 9 10 NB 29.9 29.9 13.3 13.3 13.5 PNA 26.1 26.9 14.4 14.5 18.2 pNPhA 24.7 24.0 15.2 14.9 21.3 DpNA 3,5-DNXy 2,6-DNXy 26.3 28.6 30.1 26.3 28.6 30.1 14.1 13.6 12.9 14.1 13.6 12.9 19.1 15.8 14.1 a Abbreviations:NB: nitrobenzene; pNA: p-nitroaniline;pNPhA: p-nitrophenolate anion; DpNA: NjV-diethyl-p-nitroaniline;3.5-DNXy: NjV-dimethyl-4-nitro-3,5-xylidine; 2,6-DNXy: N,N-dimethyl-4-nitro-2,6-xylidine.not only by structures 6,6a, 7 and 7a, which indicate no resonance between the nitro group and the ring, but also by relatively large contributions of the canonical structures 8,9,10 and 11. An analy- sis of the geometry of nitrobenzene] 1 reveals negligible resonance, even if the weights of the appropriate canonical structures are not exactly equal to zero (Table 3). The same applies for several other acceptor substituents. Note that some physicochemical values have been used to evalu- ate the mesomeric effect directly. For instanct the IR intensity of the vI6band in mono-substituted benzenes24 was used thus. The value of a"Rfor NO, is not zero (Table 2) but it is not certain whether it is not influenced slightly also by the inductive effect: this could be of importance just for NO,.A theoretical model for the resonance effect may be based on the extent of withdrawal of electronic charge from the ring by substitu- ents in mono-substituted derivatives of benzene. STO-3G calcula- tions2s for the nitro group yield a non-zero value of 0.13 which is lower than that in Table 2 which was obtained from nitroethene as a model.20 As well as depending on the model, these kinds of values must depend also on the basis set used for calculations. For more reliable results it may be necessary to take electron correlation into account. 4.4 Other Substituent Effects Electronegativity constants26 are still somewhat mysterious and have been used very rarely, as have polarizability constants22 also (Table 2).The steric constants of the nitro group have been less extensively studied, the hydrophobic constants a little more: for the values see ref. 27. All these effects are overshadowed by the strong polar effect. 5 Pro and Cons for Resonance Effects in Nitrobenzene Derivatives 5.1 Problem Statement As seen already in previous sections, there is a serious dispute con- cerning the importance of resonance in nitrobenzene derivatives. Many arguments have been brought together from various areas outside the framework of the theory of substituent effects. For this reason we give a brief review in a separate section. Evidently the results will be different for nitrobenzene itself and for its derivatives with an electron-donating substituent in a conjugative position, 5.2 Theoretical Calculations The effect of the nitro group on the rest of the molecule and on global or local properties may be described by the energies of the lowest unoccupied and highest occupied molecular orbitals, LUMO and HOMO, and by the global charge distribution.These may all be obtained from molecular orbital calculations. Ab initio calculations on nitrobenzene with optimization of geometry at the 6-31G gave the following extents of with- drawal of the electronic charge from the ring: 0.521 for u electrons, 0.067 for Telectrons, with a total charge transfer of 0.588. Other calculations using different basis sets25,28-30 have given a similar picture.On considering the m-electron charges, it is immediately clear that typical resonance structures which may be drawn for nitrobenzene 6-10 are realistic but the effect is not too large. However this statement does not fit with the total electron charges: of all carbon atoms only C-1 is positively charged. The point is that all hydrogen atoms are a source of electrons for the carbon atoms to which they are attached. 5.3 Structural Arguments Analysis of the geometry of nitrobenzene'' leads to the conclusion that the mesomeric effect of the nitro group on the ring is very small. This conclusion is essentially in line with the above-mentioned quantum chemical calculations which either oppose the existence of this effect29 or indicate its rather low ~alue.~~.~~ The situation is quite different for p-nitrobenzene derivatives with electron-donating substituents leading to a push-pull effect.Valence bond calculations for p-nitroaniline and p-nitrophenol yielded very low weights for structure 11(<1.0%)5but, on the other hand, an analysis of the geometry of N,N-diethyl-p-nitroaniline 12, N,N-dimethyl-4-nitro-3,5-xylidine13 and NJV-dimethyl-4-nitro- 2,6-xylidine 14 leads to the conclusion8 that the steric effect of methyl groups in positions 3 and 5 disturbs the geometry of the N-Ph-N fragment significantly less than their action from posi- tions 2 and 6. pe2 c12> C133 ci4> Thus, the only conclusion may be that the contribution of the structure 11is important in descriptions of the molecular geometries of 12,13 and 14.This is well illustrated by a significant decrease of the sum of the weights of (8 + 9 + 10) from 47% for 12 to 43% for the 3,Sderivative and down to 40% for the 2,6-derivative, which is comparable to 40.1 % for nitrobenzene itself. Further, indirect support for the through-resonance effect in para-substituted derivatives of nitrobenzene comes from the good linear dependencies of the weights of various canonical structures on upor u;of the conjugated s~bstituents.~ For the canonical struc- ture 11the correlation coefficient r is -0.934 for 14 data points; not much worse are r-values for similar scatter plots with other canon- ical structures. 5.4 Discussion of Dipole Moments A classical argument for resonance originates from the dipole moment of nitrobenzene, which is greater than that of nitromethane or 2-nitro-2-methylpropane (Table 4). The difference was called the mesomeric dipole moment.31 Moreover, in 2,4,6-trimethyl-nitrobenzene possible resonance is suppressed and the dipole moment is actually reduced.32 However convincing this reasoning appears, it can be challenged when a more complete series of deriv- atives is compared33 (Table 4).With increasing size of molecules of aliphatic and alicyclic nitro compounds, their dipole moments increase owing to pure induction; ultimately they exceed the value for nitrobenzene. On the other hand, the dipole moments of 2,4,6- trialkylnitrobenzenes decrease continuously with increasing size of the alkyl group, although a further efficient hindrance to resonance is no longer possible.(At a twisting angle of say 70" the resonance is practically inhibited.) These facts can be understood in terms of induction, either within the alkyl group of RNO, or in the ortho-CHEMICAL SOCIETY REVIEWS, 1996 Table 4 Experimental dipole moments of some nitro compounds (in Debye)' R RNO, 2,4,6-R3C6H2N02 H - 3.97 Me 3.16 3.65 Et 3.21 - Pr* 3.31 3.59 Bur 3.42 3.45 1-Adamantyl 4-Adamanty l 1-Diadamantyl Cyclo-C,H, I 3.53 3.55h 3.76h 4.03 3.46 --- Ph 3.97 3.40 a Benzene solution, 298 K, data from ref. 33 unless otherwise noted. Ref. 34. alkyl groups of C,H,R,NO,, but a quantitative estimation is hardly possible.In any case the simple difference between an aliphatic and an aromatic derivative cannot be taken as a measure of the mesomeric effect: any resonance in nitrobenzene has not been proven from dipole moments. 5.5 Correlation of Reactivity Data This matter was summarized recently,35 and has been a subject of much disp~tation,~~.~~,l~,*~ proceeding mostly in terms of substitu-ent constants and not always in an understandable way. The indis- putable experimental fact is an approximate linear relationship between pK values of substituted benzoic acids, for para- vs. meta-derivatives with the same substituent. It is valid for substituents NO,, CN, S02X, CF,, CCl,, CH2Hal and others (only with slight deviations also for COX), in aqueous systems,19 non-aqueous sol- vent~,~~and the gas phase,35 not only for pK values, but also for rate constants of various reactions.19 Conjugated substituents, like OR, NR, and halogens, deviate very distinctly, The slopes of corre- sponding linear plots show that the substituent effect is somewhat stronger from the para- than from the meta-position.The simplest explanation is that all these substituents act by a single main mech- anism: it is merely a question of terminology whether it is called simply the inductive effect. Within the benzene nucleus the so-called .rr-inductive effect19 may be operative, explaining the greater effect in the para-position; this is in agreement with some quantum chemical calculation^.^^ The opponents of this idea argued that the mesomeric effect must be present but that it is proportional to the inductive effect for all named substituents.This cannot be l53I6 directly disproved: the problem is how such a general proportional- ity can come to exist for such different structures. Essential for the reasoning is the linearity of the plot, not its slope. Attempted proof that the slope is sometimes less than unityI6 has a statistical defect: omitting the point for hydrogen (origin of the coordinates) made it impossible to determine the slope reliably. Concluding this discussion, we offer the opinion that the pres- ence or absence of resonance must be considered with respect to an observable quantity and to its accuracy.With this in mind we find practically no resonance in nitrobenzene but can observe it in deriv- atives like 4-nitroaniline or 4-nitrophenolate anion, the extent of the observed resonance effect being dependent on the electron-donat- ing power of the para-substituent. An evident resonance effect of the nitro group is observed in the case of nucleophilic substitution but particularly in the case of vicarious nucleophilic substitution of hydrogen,38 for which it is a sine qua nun. 6 Conclusions The nitro group is an outstanding substituent which should be included whenever possible in studies of substituent effects: its effects are strong and can be quantitatively estimated with reliabil-ity. On the other hand, the theoretical interpretation of these effects is not always unambiguous, particularly the existence of resonance.THE NITRO GROUP AS SUBSTITUENT-0 EXNER AND T M KRYGOWSKI The idea of separating inductive and resonance effects evidently has its limits and the term resonance should generally be used in a quan-titative sense with respect to certain observable quantities Resonance is certainly not present, or at least not observable, in every single case where resonance formulae have been written in the literature Acknowledgements Both of us extend particular thanks to John Shorter who has kindly read and commented on the manuscript T M K wishes to acknowledge support of the research grant BST 24/94 7 References 1 0 Exner, Correlation Analysis of Chemical Data, Plenum Press, New York, 1988 2 C Hansch, A Leo and R W Taft, Chem Rev, 1991,91, 165 3 0Exner, in Correlation Analysis in Chemistry, ed N B Chapman and J Shorter, Plenum Press, New York, 1978, p 439 4 S D Kahn, W J Hehre and J A Pople, J Am Chem Soc , 1987,109, 1871 5 P C Hiberty and G J Ohanessian, J Am Chem Soc , 1982,104,66 6 K B Lipkowitz, J Am Chem Soc ,1982,104,2647 7 P Politzer, L Abrahamsen and P Sjoberg, J Am Chem Soc ,1984,106, 855 8 T M Krygowski and J J Maunn, J Chem Soc ,Perkin Trans 2,1989, 695 9 T M Krygowski and I Turowska-Tyrk, Coll Czech Chem Commun 1990,55, 165 10 A Domenicano, G Schultz, I Hargittai, M Colapietro, G Portalone, P George and C W Bock, Struct Chem ,1989,1, 107 11 R Boese, D Blaeser, M Nussbaumer and T M Krygowski, Struct Chem , 1992,3,363 12 M Colapietro, A Domenicano, C Marciante and G Portalone, Z Naturforsch Ted B, 1982,37, 1309 13 J Maurin and T M Krygowski, J Mol Struct , 1988,172,413 14 A P Cox and S J Wanng, J Chem Soc ,Faraday Trans 2, 1972,68, 1060 15 S Ehrenson, R T C Brownlee and R W Taft, Progr Phys Org Chem , 1973,10, 1 16 M Charton, Progr Phys Org Chem , 1981,13,119 17 0Exner and Z Fnedl, Progr Phys Org Chem , 1993,19,259 18 S Mamott and R D Topsom, J Am Chem SOC , 1984,106,7 19 0Exner, Collect Czech Chem Commun ,1966,31,65 20 S Mamott and R D Topsom, J Chem Soc, Perkin Trans 2, 1985, I045 21 V A Palm, Osnovy Kolichestvennoi Teorii Organicheskikh Reaktsir, Izd Khimya, Leningrad, 1967 ch IX 2, C J Hine, Structural Effects on Equilibria in Organic Chemistry, Wiley, New York, 1975, sect 3-312, Y Tsuno, M Sawada, T Fuji and Y Yukawa, Bull Chem Soc Jpn , 1979, 52, 3033 W Adcock, M J S Dewar and B D Gupta, .I Am Chem Soc , 1973,95,7353 22 R W Taft and R D Topsom, Progr Phys Org Chem , 1987,16, 1 23 T M Krygowski, J Kruszewski and R Anulewicz, Acta Crystallogr Sect B, 1983,32,732 24 R T C Brownlee, R E Hutchinson, A R Katntzky, T T Tidwell and R D Topsom, J Am Chem Soc , 1968,90, 1757 25 T M Krygowski, G Haefelinger and J Schuele, 2 Naturforsch ,Teil B, 1986,41,895 26 S Mamott, W F Reynolds, R W Taft and R D Topsom, J Org Chem , 1984,49,959 27 L Hansch and A J Leo, Substituent Constants for Correlation Analysis in Chemistry and Biology, J Wiley, New York, 1979 28 C W Bock, M Trachtman and P George, J Mol Struct (Theochem),l985,155,122 29 J P Ritchie, Tetrahedron, 1988,44,7465 30 T M Krygowski, K Wozniak C W Bock and P George, J Chem Res (S), 1989,396 31 L E Sutton, in Determination of Organic Structures hy Physical Methods, ed, E A Braude and F C Nachod, Academic Press, New York, 1955,p 373 32 H Kofod, L E Sutton, P E Verkade and B M Wepstser, Rec Trav Chim Pays Bas, 1959,78,790 33 V VSeteEka and 0 Exner, Collect Czech Chem Commun , 1974,39, 1140 34 0 Exner, V JehliEka, L VodiCka and P Jakoubek, Collect Czech Chem Commun , 1980,45,2400 35 M Decouzon, 0 Exner, J F Gal andP C Maria, J Phys Org Chem 1994,7,615 36 0Exner and K Kalfus, Collect Czech Chem Commun ,1976,41,569 37 E R Vorpagel, A Streitwieser, Jr and S D Alexandratos, J Am Chem Soc , 198 1,103,3777 38 M Makosza and J Winiarski, Acc Chem Res ,1987,20,282