ccdures and the ethanol buffers arc dcscribcd clscaherr.fi The ricgative logarithm of thc dissociation coiistaiit ( p K ) for each of the acids usccl i l l thc buflcrs serves t o describe the itcidity of the solutiotl ~vhci1the huficr r;ctio is kept at uiiity. Materials.-The iiitroaniliiics, ~iitroplicnols and iiitrorncsitylenc mere coinmcrcislly availablc matcrinls ;tiit1 rwre
recrystallized from ethanol. o-Nitrophciiol was stcam distillcd before recrystallization. Melting points of the nitro compounds wcrc identical with those reported in the literdture. Nitrobenzene and nitroinethane, after drying ovcr calcium chloride, were vacuum distilled employing a WidlnCr
PinLmmrmA
4, PENNSYLVANIA
[ c O \ l l W 3 ~ i l*ROM I ~ ~ IIIE I ~ L P A R l h l ET ~OIi CHEMISTRY OE 1’URDUE UAIVERbIIY]
Solubility of Hydrogen Chloride at Low Temperatures. A Measure of the Basic Properties of Aromatic Nuclei; T- and a-Complexes and Their Role in Aromatic Substitut ion U Y I h K H E R T c. BROWN AND JAMES L). URADY’’ e2
Tlic solubility of hytlrogcn chloride at -78.51’ in dilute solutions of aromatic compounds iii toluetie and i n 11-licptaiic varies in a inaniier which indicates that the magnitude of the solubility is a measure of the basic propertics of the aromatic nuclci. Henry’s law constants for these solutions of representative aromatic coinpounds reveal the following order of increasing solubility: benzotrifluoride < chlorobenzene < benzene < toluene < nz-xylene < mesitylene. The technique distinguishes between isomeric methylbenzenes: p-xylene < o-xylene < m-xylene < pseudocumene < hemimellitene < mesitylene < prehnitene < isodurene. The monosubstituted benzenes yield the order: iodo- < bromo- < chloro- < fluoro- < benzene- < methyl- < ethyl- < isopropyl- < t-butyl-. Thiophene appears less basic than benzene by this technique. Olefins neither add hydrogen chloride nor undergo isomerization under the conditions of the measurements. These compounds modify the solubility of hydrogen chloride in a manner as to suggest the following order of increasing basicity: C1,C=_CCIz RCH=CHz < RZC=CHz. RCH==CHR < R,C=CHR. Equilibrium constants arc calculated for the reaction a t (8.51 : A r S H C l e Ar HCl. The results obtained by the hydrogen chloride technique agrec, for the most part, with the relative basicities established by the use of HCl-AIClr or HF-BFJ, or deduced from the ease of nuclear substitution by electrophilic reagents. Four discrepancies are noted between the relative basicities established by the hydrogen chloride technique and those established by the other procedures mentioned. These discrepancies are nicely accounted for on the assumption that aromatic complexes cxist as two separate species or classes, with distinct chemical and physical prcperties: rr-complexes and onium or cr-cornplcxes. I t is further proposed that the hydrogen chloride and HX-MXa techniques supplement each other-the former furnishing a measure of electron density in the ground state of the molecule, the latter in the transition state. A linear free energy relationship exists between the rate of halogenation of aromatic compounds and the stability of ucomplexes (HF-BFa). A similar relationship involving the stability of *-complexes (HC1) does not exist. This argucs against Dewar’s proposal that the rate of aromatic substitution is determined by the rate of formation and the stability of thc a-complcx. Instead, the rate of aromatic substitution appears to depend primarily upon the stability of the u-complex.
-
+
Thc observation by Klatt4 that aromatic hydrocarbons dissolve in liquid hydrogen fluoride was interpreted by Hammett5 to involve a proton transfer to thc aromatic nucleus. C6Hb
% C6EI;
+ 1‘-
lieceiitly, considerable attention has been devoted to studies of the basic properties of aroinatic nuclei. Thus Fairbrother proposed that the change in the dipole moment of iodine dissolved in various aromatic and unsaturated hydrocarbons was related to the postulated basic properties of these substances6 Fairbrother also pointed out that the shift in the color of iodine solutions in hydrocarbon solvents from violet to red to brown could also be correlated with the basic properties of the solvents. More recently, Benesi and Hildebrand7 have demonstrated that the absorption spectra of these iodine (1) The Catalytic IIalides. 111. .I preliminary Communication 71, 3573 (1949). aplieared in THISJOURNAL, ( 2 ) This paper is based upon a thesis submitted by James D. Brady in partial fulfillment of the requirements for the degree o f Doctor of Philosophy. (3) Standard Oil Company (Indiana) Fellow at Purdue University, 1947-1949. (4) W. Klatt, Z.anorg. allgem. Chcm., 234, 189 (19371. (5) L. P. Hammett, “Physical Organi,c Chemistry,” McGraw-Hill Book Co., Inc., N e w York, N. Y., 1940, pp. 293-294. (fi) P. Fairbrother, Nature, 160, 87 (1947); J , Chcin. Soc.. IO51 (1948). (7) H. A. Benesi and J . €1. Kildebrand, THISJOURNAL, 70, 2832 (1948); 71, 2703 (1049). For a discussion of the structure of thc cornplexes of halOgC11 molecules \\it11 rromalic c o u ~ ~ o u n dscc s , R. S. hlulliken, ibrd., 71, GOO (1050).
5
soIutions show regular changes in the ultraviolet region which can again be correlated with the postulated basic nature of aromatic hydrocarbons. Siiiiilar shifts in the spectra of such solutions of broniineSaand of chlorinesb have also been observed. Finally, Andrews and Keefer have examined the formation of coinplexes between silver ion and aroinatic compounds and have correlated the stability of these coniplexes with the ability of the aromatic nuclei to function as basesg While phenomena based upon the absorption spectra o f halogen solutions or the stability of silver ion complexes are presumably related to acid-base interactions, it appears desirable to obtain a more direct experimental basis upon which to develop our understanding of the basic properties of aromatic hydrocarbons. For this purpose a study of the proton affinity of aromatic nuclei by the use of methods related to that introduced by Klatt4 offers more promise.’” We had recently observed that toluene, alunii(8) (a) R. hf. Keefer and L. J. Andrews, ibid., 72, 4677 (1950); (13) i b i d . , 73,462 (1951). (9) L. J. Andrews and R . M. Keefer, i b i d . , 71, 3644 (1949). (10) Unfortunately, Klatt’s own results cannot be used without modification to estimate the basic properties of various aromatics. Apparently the solubility of aromatic hydrocarbons in liquid hydrogen fluoride must depend upon factors other than the proton affinity of the ring. Thus Klatt reports that the solubility changes in the order, nt-xylene toluene benzene, whereas there are both theoretical and experimental reasons to believe that the basic properties of these hydrocarbons must vary i n the ojiposite order, with in-xylene being milch more basic thAn bciizcnc.
tecl were dctermined graphically. 0 334 A26 ,888
This discrepancy between the high reactivity and low basicity of thiophene constitutes a fourth isomers. Since 2,4,4-trimethyl-l-pentene is a repanomaly requiring explanation. Olefins.-A number of representative olefins resentative of the class K2C=CH2 and 2,4,4-triwere examined in toluene solution in order to methyl-2-pentene is a representative of the class establish whether the basicity of these compounds K2C=CIIR, an appreciable difference in basicity, agreed with theoretical predictions of base strength. and therefore of hydrogen chloride solubility, should Winstein and L ~ c a reported s ~ ~ a detailed study of be expected. The freezing points of the two isomeric octenes lie the stability of complexes of olefins with silver ion below -75.51'. Accordingly it was decided t o and observed that the stability decreased in the ormeasure the solubility of hydrogen chloride in the der : n-BuCH=CH2 > Me2C=CH2, EtCH= CHMe > Me2C=CHMe. On theoretical grounds pure compounds. The results are presented in the basicity of the olefins and therefore the stability Table IX. TABLE 1); of thecomplexes should increase in theopposite order : RCH=CH2 < 112C=CH2 < R2C=CHR. Win- DATA FOR THE SOLUBILITY OF HYDROGEN CHLORIDEIU stein and Lucas suggested that steric effects were a t 2,4,4-TRIMETHYL-l-PEsTEYE AND -2-PESTENE ( - 78 51 ") the basis of the disagreement between the theoHenry s Iiydrogrn chloride law retical and observed order. No other experimental Mole conhllllimolr, fraction, Pressure, stant.* procedure has been applied to the determination of Compounda n X X 108 mm k(mm) the base strengths of a series of olefins. 2,4,4-TrimethyI-1f ) 325 7 I8 3 83 5,50 Experiment revealed that under theconditionsernpet] trne 490 10 78 5 95 ployed in the solubility experiments hydrogen chlo840 18 34 10 74 ride neither adds t o nor isomerizes olefins, even 48.3 10 63 4 49 430 those distinguished by reactivity and lability of high 2,4,4-Trimethyl 2 pentene 774 16 93 7 40 order. I n one experiment involving 2,4,4-tri910 19 84 8 72 methyl-2-pentene, hydrogen chloride was main44 96 rnrnoles. * Pmm = kiV. Estimated graphically. tained in solution a t -23.51'. No pressure drop graphs show a slight curvature, possibly the result of 3. occurred in 24 hours. The hydrogen chloride was The small quantity of a more basic impurity in the olefin. removed by distillation a t -78". Recovery was quantitative. The olefin recovered exhibited a reThe results definitely indicate that the 2-isomer fractive index identical with that of the original ma- is a considerably better solvent for hydrogen chloterial. The solubility data are summarized in Table ride than the 1-isomer and the order of increasing VIII. Data for 12-heptane are included to facili- solvent power tate comparison. H It is evident from the results that the solubility c1 of hydrogen chloride is strongly affected by the c 1 11 \H Ir \Cl structure of the olefin. It was, however, surprising that within the experimental error there appeared t o be no difference between the t w o diisobutylene is in good agreement both with the predicted effect of 6 2 3 ) S n'6nStemand I1 J Lucas, T w r s J o u ~ v b 60,836 ~.. (1938)
BASICPROPERTIES OF AROMATIC NUCLEI:T- AND U-COMPLEXES
July 20, 1952
structure on base strength and the known reactivity of the double bonds toward electrophilic reagents. The hydrogen chloride technique thtrdore offers promise not only as a method of determining the base strengths of olefins, but also as a means of establishing the degree of substitution of the carbons of a double bond in a complex hydrocarbon. Further study of the basic properties of olefins by this technique is underway.24 The Nature of T - and a-Complexes.-It was previously proposed" that the complex formed by the reaction of the catalyst couple, hydrogen chloride-aluminum chloride, with toluene is a carbonium ion salt (I).
1
H-C-H
H-$
L 3 structures
7 H + 1I
3 structures
J
Presumably isomeric molecules with the proton in the ortho position would be essentially equally probable. A similar structure has been proposed by McCaulay and Lien for the complexes of aromatics with hydrogen fluoride-boron trifluoride. l6 The structure of these carbonium ion salts is clearly similar to the structure now widely accepted for the intermediate involved in aromatic substitution by electrophilic reagentsz5 CHI I
r
H
H
l
The interaction of hydrogen chloride with aromatic nuclei must be quite different. Whereas the complexes of aromatic hydrocarbons and HCIAlCls or HF-BFa are deeply colored, conduct the electric current, and readily exchange nuclear hydrogen for deuterium, solutions of hydrogen halides in these hydrocarbons are colorless, the solutions do not conduct the current, and no exchange is observed with deuterium halides. Furthermore, in our experiments on the toluene-HC1-AlC& system a t -80' we observed that both formation and dissociation of the complex occurred relatively slowly, in a matter of hours; the reaction apparently involves an appreciable energy of activation." In the studies reported in the present paper, equilibrium between hydrogen chloride and the aromatics was established very rapidly-within a matter of a few minutes. It is therefore proposed that the interaction of hydrogen chloride and aromatic nuclei does not proceed through the formation of similar carbonium ion salts, but instead involves a weak interaction between the *-electrons of the ring and the polar hydrogen chloride molecule (111).
6 CHa I
(111)
intermediates.
fi I
I
A The latter formulation resembles the n-complexes proposed by Dewar.26 However, Dewar would prefer to consider all interactions between aromatic nuclei and electrophilic reagents as involving Tcomplexes of this kind.27 Thus in contrast to our view, he would formulate the complexes of aromatic hydrocarbons and HCl-AICl3 or HF-BFs as not involving the transfer of the proton to any particular carbon atom (IV). I
(24) Work in progress with L. Domash: (25) There has been considerable discussion as to whether these should be considered t o be transition states in aromatic substitution or intermediates. See L. Melander [Aikiwfor Kcmi, 3 , 213 (1950)l for a critical discussion and pertinent references. The isolation of stable carbonium ion salts with HC1-AlCl: and HF-BFa, products which must be essentially identical with those involved in the deuteration of aromatic nuclei with DCl-AlUa [A. Klit and A. Langseth, 2. ghysik. Chcm., 176, 65 (1036)], lends strong support to their interpretation as
CHJ
+ HC1 J_ d----HClor
CHS
A strong argument that the structures of these aromatic complexes with HCl-NC4 and HF-BF3 should be closely related to the structures of the intermediates in aromatic substitution is furnished by the simple quantitative relationship between the relative basicity (HF-BR) of the rnethylbenzenes and their relative rates of halogenation (Table IV), A plot of the logarithm of the relative basicity (HF-BFs) versus the logarithm of the relative reactivity defines an excellent straight line.
3577
CHI I
In our opinion the experimental evidence favors the view that aromatic complexes are of two distinct types, separated by a significant potential energy barrier. One class consists of the carbonium ion type complexes with the electrophilic atom or group united to a definite carbon atom. The second class consists of complexes of less well-defined structures, involving weak interactions between the electrophilic atom or group and the a-electrons; here there is no definite bond between the electrophilic atom or group and any particular carbon atom. The benzene molecule is presently pictured as a planar ring with a relatively high concentration of (26) M . J. S. Dewar, J . Chcm. Soc., 406, 777 (1946). 127) Private communication.
electrons on both faces of the plane.28 An electrophilic atom or group would presumably be attracted t o the electron cloud a t the positions of greatest density. Since the electron density is postulated to be low near the center of the ring and high immediately above and below the ring of carbon atoms, the electrophilic atom should be associated with the electrons in this position and, presumably, can move readily around this annulus of electron cloud without greatly altering its distribution. An increase in the electrophilic nature of the reagent is considered to result in a penetration oi this cloud of 9-electrons, with the formation of a covalent a-bond with one of the six carbon atoms of the ring. This penetration and distortion of the n-electron cloud must involve an appreciable activation energy and accounts for the separation of aromatic complexes into two distinct classes proposed in this paper. It is suggested that the term “n-complex” be restricted to those complexes in which the n-electron cloud is not seriously distorted (e.g., the hydrogen chloride complexes), and the complexes of the remaining class be termed “oniuni ion-complexes” or, more conveniently, “u-coniplexes.” This interpretation affords a simple explanation of the four anomalies observed in this study. I.
The m-xylene anomaly Relative basicity (hydrogen chloride) +Xylene < o-xylene < In-xylene Relative basicity (HF-BF3) p-Xylene < o-xylene benzene
Considerable evidence is now available that the full electron contributing properties of a methyl group can only be realized in an electron deficient system (e.g., carbonium ion) which permits hyperconjugation. Thus a methyl group introduced into the benzoic acid or aniline molecules has only a small effect upon the ionization constant of the acid or base. Moreover, the effect is not very different with the methyl group in either the meta or para position. However, a methyl group introduced into the nucleus of phenyldimethylcarbinyl chloride produces a large effect upon the solvolysis rate in the para position, hut only n small effect in the meta (2s) hl J S D e w x , “ I l e c t r o n i c Theory of O r l o r ( l lliii\er%it\ l’rrii Yrw \ r - r h V
y
OrgdtilL
1~140
Chenii\tr\ ’
I n the hydrogen chloride “a-complex” there is only a relatively small distortion of the x-electrons so that the electron releasing qualities of the two methyl groups are only partially realized. In the same way one can account for the very high basicity and reactivity of mesitylene. Here three methyl groups are in position maxinially to stabilize the u-complex. On the other hand, the other two trimethylbenzenes and both durene and prehnitene have only two methyl groups each in such favored positions. Again in the hydrogen chloride n-romplex the methyl groups contribute (29) IJnput,lished work with Tames D Hrd,ly Sjniilar r f l r t l i c t l methyl w t , ~ t i t i i r n 111 l ~ treiizhyilryl cliloridcs wcre oI>vI \ c t l l i \ I I h‘orrl~,in Et > i-Pr > t-Bu, with the decreasing number of a-hydrogen atoms available for h y p e r c o n j u g a t i ~ n ~this ~ ~ ~is~ ;in contrast to the order observed by the hydrogen chloride technique. In the latter the alkyl groups must contribute t o the electron density of the aromatic ring primarily through their inductive effect.31 Finally, the hydrogen chloride technique forces us to conclude that the apparent electron density of the thiophene ring is less than that of the benzene ring. The great reactivity of the thiophene molecule toward electrophilic reagents must be primarily the result of a transition state of low energy. The thiophene ion (IX) must be very stable compared with the corresponding benzene derivative3* (X) (IX)
TJ,FI
\S/\H +
(XI
0
\/ \/ C C (XIII) I/ + HCl J_ /I ---HCI
/-\
(30) J. W.Raker and W. S. ”athan, J . Chcm. SOL.,1844 (1935). (31) It was pointed out earlier that from the relative electronegativities of the four halogens, one would predict the order of basic properties, FCiHs < ClCiHr < BrCaHs < IC8Hr. Since the opposite order is given by the hydrogen chloride technique (Table VI), it is necessary to consider that resonance effects profoundly modify the electron availability of the aromatic nucieus. This is in sharp contrast to the conclusion reached above that hyperconjugation effects are of relatively minor importance i n the ground state of the alkylbenzenes. (32) We can justify the postulated greater stability of the thiophene ion ( I X ) by comparing the reintvely high stabilities of simple sulfonium ions, RIS +, with the much lower stabilities of the corresponding carbonium ions, R.
