STRUCTURE AND REACTMTY OF PYRIDINE
April 20, 1959
the solution stirred at room temperature overnight. The solution was concentrated under reduced pressure, diluted with water and extracted with methylene chloride. The extract was dried with sodium sulfate and concentrated to dryness. The residue (0.80 g . ) was recrystallized from benzene and sublimed for analysis, m.p. 188-189O, [a]D
[CONTRIBUTION FROM THE &OOL
1935
-87.3’; AEzE 246 (8,400)and 303 m p (2,100), shld. 312 mp (1,900); Xmi. 224 (2,900) and 271 mp (590). A ~ J Calcd. . for c ~ ~ H ~ ~ N c,~ 76.47; o: H, 8-78; N, 9.39. Found: c, 76.76; H,8.78; N, 9.50. SWIT. N. J.
OF CHEMISTRY,
RUTGERS, THE STATE UNIVERSITY]
A Comparison of the Structure and Reactivity of Pyridine and Pyridine-1-oxide BY RODERICK A. BARNES RECEIVED OCTOBER23, 1958 The electron distribution and atom localization energies of pyridine and pyridine-1-oxide have been calculated by the molecular orbital method using a consistent set of parameters for both molecules. The results have becn compared with the available experimental data and satisfactory agreement is observed if the resonance integral of the oxygen-nitrogen bond of pyridine-1-oxideis approximately 0.758.
lations for systems containing nitrogen atoms assumed that the coulomb integral of nitrogen was represented by a~ = ac ZP, but in this work the suggestion of Coulson‘ that OIN = (UC ’/& has been followed. Recent calculations by Brown6 have demonstrated that for attack of the phenyl radical on the pyridine nucleus, only the parameters used here, (YN= ac ‘/dand BCN = B,-produced calculated reactivities which agree with the experimentally observed rate factors.e Other calculations for pyridine’ and pyridine-l-oxides have been made using rather different values for the coulomb + N N+ integral of nitrogen. ‘ I I1 0Because there was little basis for choosing a value for the resonance integral of the nitrogen-oxygen greater stability’ of structure 11; but, whether the bond (@NO), the calculations for pyridine-1-oxide contributions of the two structures are really suf- were repeated using three different values covering ficiently different to produce a finite charge a t this the range in which the true value is almost cerposition could not be predicted with any degree of tainly located. The charge distributions resultconfidence. ing from these calculations are given Fig. 1. The failure of the valence-bond procedure for It is apparent that the change in value of PNOhas a quantitative descriptions of heterocycles results from profound effect on the charge distribution of pyrithe fact that there is no generally acceptable method dine-1-oxide. For the lower value of PNO, the for determining exactly the contribution of charged charges on the carbons are much as in pyridine, but structures such as I and II.2 a t the higher values a part of the negative charge on In contrast, the simple molecular orbital method, the oxygen atom appears a t positions 2 and 4. although subject to some limitations3 affords a In principle it should be possible to determine exconvenient procedure for calculating the charge dis- actly the value of @NO by calculating the dipole tribution and relative reactivity of heterocyclic moment for various values of @NO, and comparing molecules. The main di5culty is in deciding on the it with the experimental value. The part of the correct parameters to be used in the calculations. moment which arises from the ?r-electrons is readThe ideal situation, in which one set of parameters ily calculated, but there is difficulty in deciding on would serve for a quantitatively accurate descrip- the values to be used for the u-bond moments, tion of all heterocycles, is quite unlikely to be opera- which certainly must vary with the charge on the tive, but the definition of a set of parameters which two atoms of the bond. In spite of this uncergive a good qualitative description of a wide variety tainty, these calculations have been made in a manof heterocycles may be realizable. (4) Coulson, “Valence,” Oxford University Press, Amen House, The purpose of this paper is to compare the re- London E. C. 4, 1952, p. 242. activity and electron distribution of pyridine and (5) R. D. Brown, J . Chcm. Sac.. 272 (1956). pyridine- 1-oxide, as calculated using a consistent (6) The only uncertainty about these calculations is that the set of parameters for both molecules with the avail- coefficient a of the equation RT log k i / k : u ( A : - A i ) . which rerate constants to atom localization energies ( A i ) , was assumed able experimental data. Many of the earlier calcu- lates to be the same for phenyl radicals as for trichloromethyl radicals. The resonance or valence-bond method for approximating the true electron distribution works very well for aromatic systems containing no heteroatoms. However, in heterocycles such as pyridine-1-oxide this method may be of doubtful value even for a qualitative description. For example, from contributing structures I and I1 it could be assumed that pyridine-1-oxide has a small negative charge a t carbon two because of the somewhat
0
+
+
+
0 4
-
(1) The smaller separation of charge and the greater number of bonds in I1 are the reasons for the prediction of its greater stability. (2) B. Bak, Acta Chim. Scund.. 9, 1355 (1955). has proposed a formula for determining the contribution of charged and uncharged structures from accurate values for the bond lengths, but this procedure has been tested for only a few examples. (3) R. D. Brown, Quart. Rcos., 6, 63 (1952).
(7) P. Yvan, Comfit. rend., 449, 622 (1949). used .ZN = ac -I-Band @CN = B but neglected inductive effects. (8) H. H. Ja66, THISJOURNAL,1 6 , 3527 (1954), calculated the coulomb integrals a N ac 2.0048 and ao = ac 1.016B from a relationship he has developed between the coulomb integral and the substituent constants of the Hammett equation; see H. H. Jaffe, J . Chcm. Phys., 40, 279 (1952).
-
+
+
RODERICK A. BARNES
193G
VOl. 81
ry+ )+
" .I
I.,,
jsu
=(I
i;
=l.Op
36,,
=1.51
Fig, 1
iier siniilar to that of Orgcl. Cottrell, Dick and Suttoii9 assuniing t!ic lcngtli of thc liitrogcn. osygcn bond to be 1.3 A. ; the u Iiioiiicnt for this bond was calculxted froni the cx~icrii~icnt:il iiioiiierit of triiuethylainine osidc a i i c l tlicrdurc :tlso coiitains ;L correction factor. TAB1.L. r-
Compound
1 2
Pyridiiie Pyridine-I-oxide (Bso,
0.W)
3
Pyidine-1-oxide
4
Pj,ridiiie-l-oxidc
(ha,
a
1.03)
hlomeiit
-o\ide 2 . -1; 3.54 2 . 50 2 , :is 2,33 2.44 2.28 2.56 2.42
1 0-
Slunieut
'l'otol
E x 1'1 1, iiitiniet~t
1.33
0,s;
?.US
?,21
0.4;
1.30"
3.44
-1- '78
4.77
1.30"
3.47
428
0.80 1.30" ( P m 1.53) 210 This moment is i i i oppositiuii to tlie s-inomelit.
-i,z
From Table I it niay be estiniatcd that when the rcsunance integral 8x0 is equal to 0.83, tliere would be agrcenient between the calculated and expcriiriental niomerits of pyridine-1-oxide. The aromatic substitution reactions are uf greatest interest for these two niolecules, and since these are usually rate controlled processes, calculation of the energies of activation would be the ideal way of comparing reactivities. The atom localization criergies'O which may be readily calculated are believed to differ from the true activation energies by a factor which is essentially constant for a given type of substitution reaction. The atom localization energy may be described as the amount of energy required t o bring the aroniatic molecule into the transition state while still keeping i t insulated from any interactions with the approaching reagent. Thus either two, one or zero electrons are localized at a given position depending on whether the reagent is electrophilic, a radical or nucleophilic. In Table I1 are listed the values in terms of -/3 for the atom localization energies a t the three positions for each molecule. The atom localization energies of benzene, A , = A , = A n = 2.54, and naphthalene, A, = A , = A , = 2.30, can serve as reference points for the values in the table. It is also noteworthy that for the radical reactions of pyridine a difference in the value of A , of 0.03,3 between two positions is equivalent to a factor of two in the rates (at 91 0 ) . 5 The ncarly perfect correlation between experimental reactivities and localization energies for attack of phenyl radicals on the pyridine nucleus has already been mentioned. Unfortunately the radical reactions of pyridine-1-oxide have not been sufficiently studied to furnish data for coinparison with (9) I-. 13. Orgcl, T. L. Cottrell. IT.Dick a n d L. E. S u t t o n , T r a m F a r o h , * S O L . 4, 7 , 113 (1051). (10) 'The a t o m localization energy is essentially t h e same a s A W which w a s calculatcd b y Wheland a n d considered t o be a p a r t of t h e activation energy; see G. W . Wheland, TnIs J O U R N A L . 64, 901 (1942).
2 42 2.50 3.39 2.54 2.49 2.54 2.58 2.52
2.75
the calculations. Table I1 predicts that for all !dues of 8x0 the order of reactivity sliould be 2 > 3 > 3, the saiiie as for pyridine, with the magnitude of the differences between the positions somewhat greater than for pyridine. These predictions are represented graphically in Fig. 2 . In these as well as the other reactions of pyridine-1-oxide, the atom localization energies would be expected to predict a greater reactivity for the 2-position than that observed because the calculations take no account of the steric ef-iect of the osyx:eri atom of the adjaccnt iiitrogeri atom.
@ +
2.8
0.5
1.0
1.5
PNO.
Fig. 2.-I
