Electron Distribution in Molecules. I. F19 Nuclear ... - ACS Publications

Electron Distribution in Molecules. I. F19 Nuclear Magnetic ... High-Resolution Nuclear Magnetic Resonance And Molucular Structure. S. Brownstein. Che...
0 downloads 0 Views 2MB Size
Oct. 5 , 1952

ELECTRON DISTRIBUTION I N BENZENE DERIVATIVES [CONTRIBUTION FROM

THE

4809

NOYESCHEMICAL LABORATORY, UNIVERSITY OF ILLINOIS]

Electron Distribution in Molecules. I. Flg Nuclear Magnetic Shielding and Substituent Effects in Some Benzene Derivatives1 BY H. S. GUTOWSKY, D. W. MCCALL,B. R. MCGARVEY AND L. H. MEYER RECEIVED FEBRUARY 8, 1952 The influence of substituents on the electron distributiou in benzene produces changes in the nuclear magnetic shielding of fluorine atoms in the molecule. A comparison of the F19 nuclear magnetic shielding in fluorobenzene with that in a substituted fluorobenzene is used in defining a ¶meter. Experimental &values are reported for a considerable number of mono- and polysubstituted fluorobenzenes. An empirical correlation of &values for meta and para substituents with the corresponding Hammett substituent u-constants reveals systematic differences which are attributed to the dependence of the &values on the nature of the electronic interactions of the substituent. Thereby, a detailed analysis of the 6- and U-values for particular substituents permits evaluation of the nature of their electronic effects. Consideration of ortho substituents suggests that besides the usual inductive and electromeric effects a t the meta and para positions, an additional interaction contributes to the ao-valuesin some cases. The potential uses and limitations of 8,-values in evaluating ortho-effects and entropies of reaction are discussed. I t is concluded from &values for polysubstituted fluorobenzenes that the electronic effects of substituents are usually remarkably additive but with significant non-additivity demonstrating interactions between substituents in about a third of the compounds investigated. A 6’-parameter, analogous to 6, is defined in terms of the change produced by a substituent in the ring on the F 1 g nuclear magnetic shielding in benzotrifluorides. The 6’-values are about a tenth of the &values for the same substituent, indicating the extent of attenuation in the side chain of the substituent effects. Less conclusive discussions and comparisons are given of various other aspects of the data and experimental method.

Introduction applied field and is usually in the opposite direction. Several features of the microwave and radio- The effect may be considered an internal diamagfrequency spectra of molecules are associated with netism or a magnetic shielding of the nucleus. The magnitude of this nuclear magnetic shielding their electronic structure. Nuclear electric quadrupole splittings of microwave rotational spectra2a depends upon the nucleus and the electronic strucand the radiofrequency pure quadrupole spectraZbtC ture of the sample .in which it is observed.? A are determined b y the variation in electrostatic general theory has been proposed* for nuclear field a t the nucleus. Electric field gradients have magnetic shielding in molecules, but the complexity been calculated for assumed electronic structures of the calculations has restricted application thus and compared with experimental values to give far to the hydrogen molecule. However, in an considerable information about bond hybridiza- experimental survey of the binary covalent fluotion.2a This report is concerned primarily with rides, the F 1 9 magnetic shielding was found3 to another similar effect, the magnetic shielding of decrease approximately linearly with increasing fluorine nuclei3 and its dependence upon sub- electronegativity of the atom to which the fluorine is bonded. A reasonably accurate physical picture stituents in fluorobenzenes. Nuclei with magnetic moments exhibit radio- is that the more tightly electrons are held by the frequency spectra in the one to forty megacycle atom bound to the fluorine, the less effective they range when placed in magnetic fields of 5,000 to are in shielding the fluorine. These empirical 10,000 gauss.4 Resonance absorption, correspond- results suggested that F 1 9 magnetic shielding ing to the reorientation of the nuclear magnetic values might serve to evaluate electronegativities, moment with respect to the applied magnetic field or electron densities. At least, the fluorine magd i r e ~ t i o n ,occurs ~ a t a frequency given by the netic shielding in a molecule is related directly Larmor equation, v = gpH,/h. g is the nuclear to the electron distribution of the group bound to gyromagnetic ratio, ,B is the nuclear magneton the fluorine. and Ho is the magnetic field a t the nucleus. HowSome preliminary measurements have been reever, the magnetic field a t a nucleus differs to a portedg for an F 1 9 shielding parameter, 6 , defined in small but measurable extent from the magnetic terms of the change in shielding caused by a subfield in the macroscopic sample.6 The applied stituent in fluorobenzene. Substituents in benzene magnetic field interacts with the motions of the of course influence a considerable variety of other electrons in the system, which thereby contribute a In most instances it is generally component to the net magnetic field a t the nucleus. (7) (a) W.I1 Knight, ibid.. 76, 12.59 (1949); ( b j W. G. Proctor This electronic component is proportional to the ( I ) Supported in part hy OKR. (2) (a) C . H. Townes and R. 1’. Dailey, J . Chcm. P h y c . , 17, 7x2 (1949); (h) €I. G. Dehmelt and H . Kruger, h’nlurwisscnschnfle,2, 97, 111 (1950); ( c ) R. Livingston, Phys. Rev.. 82, 289 (1951). (3) €1. S. Gutowsky and C. J. Hoffman, J. Chen. Phys., 19, 1259 (1!25l), and references cited therein. (4) (a) E. M. Purcell, H . C. Torrey and R. V. Pound, Phys. Rev., 69, 37 (19461; (b) F. Bloch, W. W. Hansen a n d M . Packard, ibid., 69, 127 (1946). ( 5 ) (a) E. M. Purcell, Scicncc, 107, 433 (1948), gives a n excellent introductory description of nuclear magnetism; (b) G. E. Pake, A n .J . Phys., 18,438, 473 ( 1 9 5 O i , has published a moredetailed discussion and a bibliography. ( 6 ) (a) W. E. I a m b , J r . , Phys. K c v . , 60, 817 (1941); (b) W. C Dickinauu, ibiJ,,80, 563 (lY5U).

and F. C. Yu,ibid., 77,717 (19.70); rc) W. C.Dickinson, ibid., 77,736 (ism. (8) N. 1’. Ramsey, ibid.,78,tiSR (1950) 19) ( a ) IT. S. Giltowsky, D. W. McCall, B. R. McGarvey and L. H. hleyer, J . Chcm. P h y T . , 19, 1328 (1951); (h) a n earlier suggestion u.as given hy €1. S . Gutowsky and C . J. Hoffman, Phys. Rev., 80, 110 (1950). (10) ( a ) G. E. K. Branch a n d h i . Calvin, “ T h e Theory of Organic Chemistry,” Prentice-Hall, Inc., New York, N. Y., 1941, p. 183; (b) M. J. S. Dewar, “The Electronic Theory of Organic Chemistry,” Oxford a t t h e Clarendon Press, London, 1949, pp. 50, 160; ( c ) L . P. Hammett, “Physical Organic Chemistry,” 1st E d . , hlcGraw-Hill Book Co., Inc., New York, N. Y.,1940. (11) C. K. Ingold, Chcm. R e v s . , 16, 223 (1934). (12) ( a ) I. b:. S u t t o n , Proc. Roy. Soc. (London), 1338, 668 (1931), ( b i 1,. E. Sutton, Truns. Faroduy Soc.. SO, 789 (1934).

4810

H. S. GUTOWSKY, D. W. MCCALL, B. R. MCGARVEY AND L. H. MEYER

agreed that the effects result primarily from the perturbing influence of the substituent on the molecular electronic distribution. Among the properties affected are reactivities and orientation in substitution reactions," dipole mornents,l2 the dissociation constants" or relative rates of reaction'0c of substituted compared to unsubstituted compounds, spectroscopic momentsI4 and vibrational frequencies.'s A comparison of the Hammett substituent constants,loc u, with the corresponding preliminary 6-values revealed" a fairly linear correlation. In this article are presented the experimental details, 6 values for a wide variety of monosubstituted and polysghstituted fluorobenzenes, and similar data for several benzotrifluorides. A detailed comparison of u and some of the other substituent effects with the considerable number of &values now available confirms their general similarity, and develops several interesting exceptions.

Vol. i 4

netic shielding in the sample. Relative absorption intensities are proportional to the number of nuclei. Figure 2 is an oscilloscope photograph of the FIQ absorption line in 2,3,5-trifluorobenzotriRuoride. The strong line is from the CFs group; the three weaker lines, with equal intensities, are from the three structurally distinguishable F" nuclei attached directly to the benzene ring. I t is apparent that

onthe oscill&cope first bne resonance line 2nd then the other. Calibration of the field hinsing current in terms of magnetic field has been d e s ~ r i b e d . ~6, the F'O magnetic shielding parameter is definedaa as lI? X ( H , - H,)/H. where H. is the applied magnetic field for the F1*rrcsonancein the given substituted fluorobenzene and H. is that for the reference, fluorobenzene itself.

Experimental Apparatus and Procedure.-Figure 1 is a block diagram of the experimental arrangement. The spectroscope used a radiofrequency bridge" and was identical with that described in detail elrewhere,' except for the manner of field modulation and sample comparison, The signal generator was set a t a fixed frequency of about 25.5 megacycles/ second, the F'O magnetic resonance frequency in the 6365 gauss field of the permanent magnet. The sawtooth voltage generated in P Dumant 304-H oscilloscope was applied at a frequency of 2 cycles/second to the horizontal axis and also to a cathode follower. The current of the latter war fed through a field modulation coil" about one magnet pole. By connecting the d.c. output of the detector in the communications receiver to the vertical axis of the oscilloscope the absorption line was displaced on the oscilloscope as a function of the magnetic field applied to the sample. I n addition, the position on the oscilloscope of the ahsorption lines was adjustable by passing a direct current through another coil about the other pale of the magnet.

u

II

4 4 D C OUTPUT

Fig. 2.-Oscilloscope photogrqih of the F1* magnetic resonance ahsorption Iincn in 2.:1,5-trifluorohenzotrinuoride. The strong line is assigned to t h r I"' nuclei in the CFa group; the three weak lines. from left to right, to the 5.. 3- and 2fluorines. The horizontal sweep is about 0.9 gauss. increas ing from left t o right; total applied magnetic field. 6365 gauss. Samples were placed in short lengths of 5-mm. a d . Pyrex tubing, sealed a t one end. These t u b a fitted snugly in the r.f. coil of the bridge element in the magnetic field. Most of the samples observed were liquid a t room temperature. In this event one drop was added to a drop of a reference compound and the F'$ magnetic rmonance observed on the

HODULATI

Fig. 1.-Block diagram of the radiofrequency speetroscope with slow sweep modulation far measuring molecular differences in nuclear magnetic shielding. Under otherwise fixed conditions, the position of an F" absorption line on'the oscilloscope depends upon the Flu mag(13) 1. P. 1. Dippy, Chrm. RIVI.. 28. 151 (1939). ( 1 4 ) 1.R. Platt, J . Chrm. Phys., 19, 263 (1951). (1.5) (a) M. Sf. C. Flcff. T i o m Forodoy Sor.. 44. 767 (1948); (b) A. H. Soloway and S. L. Frirss. Txrs J o o r ~ a 13. ~ . 5000 (1951). (161 N. Bloembergen, E. M.Pureell and R. V. Pound, Phyr. Riv., 73, 679 (1948). 117) This arranzcment is similar to that used earlier bv 1. T . Arnold. .. S. S. DharmPtti and M. E. Pneknrd. . I Chnn. . Phyr.. IS. 507 (1951).

substituted fluarohenzener unless the separation of the absorption lines was less than ahout 11.007 gauss, the limiting resolution. I n such cases. p-fluomanisole was used as a secondary reference, and the measurements reduced to the fluorobenzenescale. For the substituted benzotrifluorides, p-iodofluorabenzene was used as a reference, and the values then converted to a benzotrifluoride scale, denoted as 6'. Data on the binary covalent fluorides' have been referred These data were reto fluorine gas as an arbitrary ported as relative shielding values defined as ( H e - Hm)/ H% 6- and 6'-values can be converted to this scale by using the fluambenzene and benrotrifluaride values of 54.32 X 10-&and49.33 X respectivcly, on the F2 scale. One should note that the signs of 6 and 6' are reversed from that of the F? scale. (18) Trifluoroaectie acid is sn excdlent reference comwund which we have used lor moat m c ~ r u r c m e n t on i non-aromatic comvoundr: its shielding value on the PI srale is 60.71; X 10..

Oct. 5, 1952

In a few cases, the samples were solid and solutions were required t o obtain the narrow lines essential for accurate measurements.3 Aromatic solvents, principally nitrobenzene, were used when possible because small solvent effects were found with dissimilar substances. The fluorobenzoic acids and fluoroacetanilides required ethanol t o obtain a solution concentrated enough for measurement. The solutions were handled in the same manner as the liquids. All measurements were made a t room temperature. l9 Sources and magnitudes of errors have been discussed3for the somewhat different sample interchange and field modulation methods used earlier.3,qa The new slow sweep modulation and combined utiknowti-reference samples improved the consistency of these nieasurements. Low positive biasing currents, less than 0.5 gauss, were maintained to reduce hysteresis effects. Also, a double-pole bias switch made it possible to center the two resonances with separate rheostats. This permitted a rapid change from one resonance to another, virtually eliminating error introduced by drift in the signal generator frequency. At least one series of five to ten measurements was made on each compound, and two or three on most. The probable errors of the reported 8and 8'-values are less than =t0.03. Materials. -The compounds used were obtained from several sources. The purity was not critical as long as there was no question regarding the identity of the major constituent. In fact, observation of the F19 absorption lines was found quite useful on several occasions in determining the nature and extent of some fluorine-containing impurities. The o-fluorochlorobenzene, p-fluoroanisole, pfluorobenzonitrile, p-fluorobenzenesulfonyl chloride, and 4fluorobiphenyl were the purest grade available from Eastman Kodak Co. The o-fluoroanisole, and a-, m - and p fluorophenetoles were from Custom Chemical Co. The pfluorodimethylaniline and S-phenyl-p-fluoroaniline were furnished by Dr. Selson J . Leonard, of this department. The o-, m- and p-trifluoromethyl benzotrifluorides were furnished by Dr. 0. R. Pierce of Purdue University. The vt-fluorobromobenzene, nt-fluoroiodobenzene and m-fluorobenzonitrile were synthesized. The remainder of the samples were supplied by Dr. Glenn C. Finger, head of the Fluorospar Division of the Illinois State Geological Survey.

Results and Discussion Monosubstituted Benzene Derivatives A Comparison of U- and 6-Values.-Table I lists 6-values determined for a variety of 0-,m- or pmonosubstituted fluorobenzenes. In Fig. 3, 6values from Table I are plotted against Hammett's substituent constantloCr where the latter are available. It is seen that there is a general correlation between the two parameters' a'though scatter Of individual points from a straight line is often greater than experimental error. In particular, the meta substituents follow a trend differing from that of the para substituents. A linear least squares solution for all the meta and para points gives u = 0.6006 0.243, with a residual sum of 2.38. Separate solutions for the meta and para points give u , ~= 1.696, and u p = 0.5606, 0.271 with residual sums of 0.26 and 0.71, respectively. The improvement in fitting the data by separate solutions demonstrates a fundamental difference in the particular effects operative a t the meta and para positions. Of course, the general correlations between u- and &values confirm the accepted view that the effects of substituents on reactivity are basically electronic in nature. In addition, the previously established dependence of the F19 magnetic shielding on the electronegativity of the attached atom suggests that the substituent effects may be interpreted as changes in electronegativity I

4811

ELECTRON DISTRIBUTION I N BENZENE DERIVATIVES

+

+

(19) T h e Flushielding >slues appear to be temperature independent in fluorocarhonq see, P 4 reference 91, and 8

11

0 50

,

t

3

-

-PIRl

h -

5-

I

UETA

PROQLBLE E R R O R S

- , -200 r n I _ . L - - -100 -1 I-- .

d

Pig. 3.-A plot of Hammett's u versus values of 6 from Table I, for nieta and para substituents in benzene. TABLEI FL9NUCLEARMAGNETIC SHIELDING ~ - \ - A L U E S FOR MosoSUBSTITUTED BENZENE DERIVATIVES~

.

c1 F

Ortho 6 -0.56 .52 .35 +1.93 +0.55 - .27 -2.59

CHI CH3CONHb OH CHsO CPH6O NHz

-1.28 -2.50 -2.24 -2.17 -2.31

Substituent NO2

CN COzH I Br

Substituent #-SOzCl* P-CCla #-Phenylb p-&-CsHr-F p-NH-CsHab

+ +

-0.50

Meta

Para

6

a

6

, J

4-0.33 .30 .05 .26 .24 .21 4- .31 .09 .10 .09

+0.710 .608d ,355 ,352 ,391 ,373 ,337 - ,069

+1.08 4-0.96 .69 - .12 - .23 - .24 - .64 - .55 - .57 -1.06 -1.14 -1.15 -1.46

+0.778 .656d ,728 ,276 ,232 ,227 ,062 - ,170

+ + + + + -

+ + . + -

,.

.

.13 .02

+ + + + + +

. ... . .

+ .loe ...... + .l5

-.

,161

+

+ + + + + +

...

-

-

.. .

.36' ,268 .25 ,660

6 a Substituent 6 r +1.26 P-N(CHd2 - 1 . 6 8 -0.720f + O , 26 O-CH=CHCO~HC - 27 .28 ,418 - .27 +0.009 m-CFj - .67 m-CH(OH)CH3 . 00 - .94 m-O-CsH~-F - .05

+

n.

+

u values are from reference 1Oc unless otherwise indi0

CH3)2solution. cated. b I n ethanol solution. c In H-C-X( I/ J . D. Roberts and E. A , McElhill, T~~~J ~72, 628 ~ (1950). e Computed from the ionization constants for mand P-hydroxybenzoic acids from reference 13. C . C. Price and D. C. Lincoln, THISJOURNAL, 73, 5838 (1951). J. D. Roberts, R. L. LVebb and E. A. McElhill, i b i d . , 72, 408 (1950). d

or electron density a t the point of reaction. A total increase in electronegativity of 2 3 units was found3 to correspondz0to an increase in F19 magnetic shielding of BO X The observed range of 6 in the monosubstituted benzene derivatives is 4.5, or a range in electronegativity of about 0.2, if the same relation holds. This is &tbest a rough estimate, particularly in view of the differences between the meta and para positions. Jaff6 has computedz1 theoretical electron densities by molecular orbital methods for several monosub(20) T h e dependence of t h e F'Qmagnetic shielding on electronegativity varies somewhat for different groups in t h e periodic table. For all reported binary covalent fluorides dEn/d6 is about 4.1 X 10-2 while for Group I V fluorides it is about 6.5 X 10-2. (21) H. H. JaffC, J. Chcm. P h y s . , SO, 778 (1952). We are indebted to Dr. Jaffe for sending us a copy of his article prior to puhlication,

~

stituted benzenes and obtaiiied results also agrccing with the u-values.'? The differences between the u- and 6-values raise some interesting questions. The line for meta substituents has a slope three tinies larger than that for para. Moreover, while the meta line includes the origin, the para scale is displayed systematically by +0.2;. ,And, finally, deviations of several of the points from the appropriate meta or para curve exceed experimental error by significant atnounts. Both u arid 6 have operational definitions so these effects must arise from factors implicit in the nature of the experiments. A possible major factor is that 6 is determined with the niolecule in its normal state, while u is evaluated froin rate or equilibrium data and depends therefore on the character of the transition state.?' H,tnimett's basic definitioii of u is

of substituted and unsubstituted reactants ; also, the last term in eq. 4 becomes puf,'2..3RT since the rate constant in only one direction enters the derivation. It was suggested earlierBa that the differences between u and 6 might be caused by the effects of substituents on the entropies of activation. In ey. 4,6 should be related directly only to c f . Any entropy differences would combine with the u' term to give the observed u, but would not coiitribute to 6. However, the relatively few rcportetl experimental entropy differences2? are gciierally too small and not systematic enough to explain the recent observations. Still, some of thc deviations may be attributed in part to entropy effects. Llnotherway in which the normal molecule differ.; froin the transition state is the polarization of the niolccule by the attacking group. Transiiiissioti of polarizability effects to the iiieta position is by the u - lug K t - log 11" (1, rclati\rely poor ititluctive mechanism, while tlic where fila n t 1 A:''Lre the ioiiiiation coilstdnts for much stronger resonance, or electromeric, inechthe substituted arid unsubstitutcd benzoic acids. anism operates a t the ortho and para positions. However, a more general development is required A t first glance, this might appear to account for the to determine the relation of u to the transition different meta and para curves. Unfortunately state. Accordingly, the following transition state, for the present purpose, the polarizability and expressions are proposed for the rate constants of static electromeric effects are in the same the forward and reverse ionization of the substi- direction.lOb So the u, '6, ratio should be larger tuted benzoic acid than the urn'6m ratio, since U, incorporates both effects while urn and both &values depend primarily on the static effect. Actually, the reverse is true. Thus, it does not appear that the main u,6 differences can be ascribed to transition ?'erw\ normal state differences; so norrnal state effects would appear responsible. A great deal of indirect with similar expressions, without the p u r , for the evidenceIuindicates that substituents influence the unsubstituted benzoic acid. In the equations, K ineta positiori chiefly by the inductive mechanism, is the transmission coeficient ; k, Roltzmann's the pdra by the electromeric, and the ortho by 1' constant; h, Planck's constant; A H * , the enthalpy combination of inductive, electronieric arid steric of activation exclusive of substituent effects ; mechanisms. There 15 no basis in Ramsey's AS*, the entropy of activation. uf isrelated to the theory of nuclear magnetic shielding,8or otherwise, substituent effect on electron distribution and p for supposinq that the fundamentally different is a reaction constant such that --pur gives the inductive and electromeric effects would produce effect of the substituent on the enthalpy of activa- changes in 6 in a ratio identical to the changes in tion. Substituting the rate constant expressions u. In fact, it may be conrluded that the separate in the definition of U , and assuming the trans- meta and para curves are a direct demonstration mission coefficients to be unity,23or a t least to of the different character of the main electronic cancel effects at the meta :tiid the para positions. The sinallcr slope of the para curve then means that tlic clectronieric effects change 6 relatively more than where ASi and AS,O are entropies of ionization, u, as compared to the inductive effects. Fortunately, this is consistent with the para intcrcept i.e., AS?, - AS?. Ry defining p = 1 for the sub- .I11 substituents other than hydrogen have 7r or stituted benzoic acid ionization reaction, and mak- unsharerl p-electrons and therefore produce elecing the usual assumption that entropies of ioniza- tromeric effects a t the ortho and para positions. tion are the same for substituted acids as for the 'The ufi zero value corresponds to a balance (if unsubstituted, u = 2ur/2.3RT. Analogous equa- opposing inductix-c ( + I ) and electromeric (-13 tions are obtained if u is determined from the in- effects.") However, for 6, the electroineric effect is fluence of the substituent on the rate of a reaction, larger than the inductive, with a resultant negative the entropy term in eq. 4 then becoining AS;? - 6,, as observed. The inductive effect must be AS: the difference in the entropies of activation positive for CT,, = 0 hecause the substituents for which g b is about x r o havc positive C T ! ! ~values. ( 2 2 ) Another terminology c a n l i e liasrd o n the relation I~etwcen partial ionic character of a hond a n d t h e electronegativity diifrretice of t h e bonded atoms, 8-Values and thereby ~ - ~ a l i ! er ai n tic, i i i l c r preteil as expresiitil: t h e i t i i i r i ? n c . ? c,f silhstitiients o i l i h c pdrtif '

ril r )

IlICllO"lCI,i

t h i l

111

1Jcn

it

Oct. 5 , 1932

4513

ELECTRON DISTRIBUTION I N BENZENE DERIVATIVES I

I

I

I

I

\

100-

t CI

-2 w

t 7 I

-020

I

OW

I

I

0 20

I

I

0 40

1 I

AtE ,200



Am.

Fig. 4.-A

plot of 6,- versus 6,-values for substituted benzenes.

Interdependence of the 6-Values.--In Fig. 4, the Gp-values are plotted against the 6,-values for the same substituent. I t is seen that the data fall into two distinct classes, in each of which 6, is roughly proportional to 6,. The CH3*groupstands somewhat alone. In Fig. 5 the 8,-values are plotted against the 6,. The same substituents are demonstrated clearly to be divided into the same two classes. Also, it is evident that the CH3 group falls in the class with the halogens and other substituents without multiple bonds. The six 6p-values in Table I for which the corresponding d,-values have not been observed follow the same rules as those in Fig. 4, insofar as the 6,-values can be estimated from the observed u,. Several conclusions can be drawn from the results graphically presented in the two figures. The electromeric effects are generally proportional to the inductive. This follows from the approximate linearity of both 6?4, curves and their positive slopes, in conjunction with the previous conclusion that &,-values are determined mainly by electromeric effects and ,6 by inductive. The separate curves prove the existence of two different types of electromeric interaction. The class with negative &-values includes only substituents with unshared $-electrons; these substituents have been considered generally lob to be - E electromeric in nature. The class with positive &values includes all substituents with r-electrons, NOz, C N and COOH, the +E substituents. The 6,-6, curve in Fig. 5 for -E substituents supports the view that both the E and the I interactions are more important a t the ortho than each is, respectively, a t the para and meta positions. For example, NH2 has the weakest I effect, with a 6, of -0.02, so the 6,-value of -2.31 is mainly the -E effect. This is to be compared with the 6,value for NH2 of - 1.46. As the +I character of the - E substituents increases, the &-values become proportionately more positive compared to the 8;. The methyl group is anomalous again; it is -I, yet the &,-value is even more positive than many of the substituents which are +I. The fluorine substituent with its relatively large +I effect presents another anomaly in its large negative &-value compared to its value for 6., The 69-6, curve in Fig. 5 for +E substituents is

1

0 - L

Fig. 5.-A

I -2w

,

I

I

-100

L

I

I

la,

I

200

plot of 6,- versus 6,-values for substituted benzenes.

also exceptional. The +E substituents exhibit +I interactions, as is seen from the positive 6mvalues in Fig, 4. However, 6,-N02 is negative and the &,-values for COOH and CN are less than the G,-values. The implication is that these substituents have some variety of negative effect on 6 which falls off much more rapidly than do the +I and +E upon going from the ortho to the meta and para positions. The methyl group is a -I, - E substituent and exhibits a similar but positive effect a t the ortho position. This question will be discussed in more detail in the section on the halogen derivatives. The interdependence of the &values is related also to the differences in the behavior of the F- and 6values discussed above. If one plots cP versus urn, a fairly good single line is obtained. The contrasting fact that +E substituents have a separate 6,-6, curve, with larger, positive, values than the -E, supports the conclusion of the previous section that E interactions change 6 proportionately more than do I interactions. The scatter of points from the 6-, curves indicates varying proportions of I and E interactions, which in turn show up in deviations from the F-8 curves. &Values for the Halogens.-The &values of the halogens suggest differences from the other - E substituents. I n Fig. 6, there is shown the dependence of the &values on the electronegativity of the halogen; the urnand data are included for comparison. The main point of interest is the 6, curve, in which 6, becomes more positive in the sequence F, C1, Br and I with &,-values of -2.59 and 4-1.93 for fluorine and iodine, respectively. The +I effects depend on the electron affinity of the halogen and decrease in the order F > C1 > Br > I. The - E effects depend ordinarily on the ionization energy and on this basis would decrease in magnitude in the order I > Br > C1> F. According to this argument, iodine has the weakest +I and the strongest -E effects, and therefore the strongest net negative substituent effect of the halogens. Actually the &values, as well as the F , show that the -E effects are reversed for the halogens,26decreasing in the same order as (26) See, C . K . , A. E. Remick, “Electronic Interpretations of Organic Chemistry,” John Wiley and Sans, Inc., New York, N. Y.,1943, and articles cited therein.

appear to give a negative rather than R positive effect, or else be in the 'i7'roiig sequence. One way in which the halogens inight withdraw electrons selectively from the ortho position is by a sinall overlap C J ~the lacant low-lying d-orbitals of the halogen with the ~-orbital5 in benzene. This ~vould give the sequence I > E3r > C1 sirice the d-orbitals are kJLVeSt lying for iotliiie, a ~ i c lahsent of course i i i the valence shell of fluorine. &,-Values and Entropy Effects.- -The hope wCis voiced earlier that &,-\ dues in coiribination with rate or equilibrium constrints would permit evaluation of the entropy teriiis. Thus, if 6, could be related directly to U' iii eq. 4, and the reaction coii- 00.\ stant p applied to ortho ds well as nieta and para 1 \ substituents, then experimental rate or equilibrium constants could be used to compute entropies of activation or of ionization. It is now clear, h o w ever, that the varying influence on 6 of the several electronic interactions severely limits such an approach. Nonetheless, it may be of some interest to give a comparison of 6, with several "~,"-values, Fig. 1; --a- and U-values compared with the electronegativity where the "uY"-valuesare obtained in the s a n e way of the halogens. as G~~~and uj,from the influence of an ortho substituent on a rate or equilibrium constant. the +I effects, F > C1 > Br > I. With the two Values of "u,, ' were obtained from the equation effects in opposing directions but in the same order, " G ~ " = ( 1 p ) log (K' K f Lusing ) Dippy's tabulationlj then by varying their relative magnitudes, almost of dissociation constants, which includes a limited any desired sequence of resultant effect can be number of ortho-substituted benzoic and phenylobtained. 27 The justification given for reversing acetic acids and the o-halophenols. Reaction the order of the - E effects is that the larger atoms, constant p's oi +1.00, $0.47 and +2.01, respecparticularly those past the first row, do not readily t i ~ e l y , were ' ~ ~ used for the three series of cornincrease their covalencies,2sas required for the - E pounds. The resulting values of "ro" are plotted effect. in Fig. 7 :Lgainst the &-values. I t is seen that the The and Gp-values are in partial accord with data for tlie halogens fall upon separate smooth this argument. It has been remarked in the dis- curves for the three series of compounds. The cussion of the Gfl versus 6, plot in Fig. 4 that fluorine parallelism and large separation of the curves appeared somewhat anomalous, being separated suggests an interaction between the ortho substitufrom the other halogens. It has the largest a,,ents and the reacting group,'Oa the interaction value (+I) and yet the largest negative Gp-value being characteristic of the reacting group and rela( - E ) , as predicted. However, the positive Gnz- tively insensitive to the particular halogen. values for the other halogens are I > Br > C1, -the reverse of the predicted +I order. Xnd then, there is the previously noted exceptionally large c 31 1 &-value for iodine. These observations are in1 A i I compatible with the small range, from -0.24 to -0.12 in 6, on going from chlorine to iodine. -I The indicated small corresponding range in - E effects cannot account for the change in 6, from -0.27 to $1.93; and the predicted direction of the +I sequence, C1 > Br > I, is the reverse of that wrequired to explain the trend in 6,. * Thus, there seems to be an additional positive &-interaction for the halogens in order I > Br > C1 > F, similar perhaps to that noted above for the +E substituents and for CH3. This b,-interaetion may be electromeric in nature since, when present, its sign is opposite to that of the usual electromeric effect, while there is no such correlation with the inductive effect. There are several mechanisms possible, but there is not sufficient evidence for a Fig 7 -A comparison of &,-values with "u,"-values cal critical choice. However, a t least in the case of the halogens, the usual sorts of interaction would culated froin the effect of ortho substituents on dissociatioti Id

~

2 03-

,------,-

~

VO?

BEN20

3

4c

ihEhc'iS

i

" ECN ID5 A L'CS-

I

SOL-

I

(27) J. F..1. Dippy and R.I