LCAO Molecular Orbital Computation of Resonance Energies of Benzene

and Butadiene, with General Analysis of Theoretical Versus

Thermochemical Resonance Energies
Robert S. Mulliken and Robert G. Parr

Citation: J. Chem. Phys. 19, 1271 (1951); doi: 10.1063/1.1748011

View online: http://dx.doi.org/10.1063/1.1748011
View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v19/i10
Published by the AIP Publishing LLC.

Additional information on J. Chem. Phys.

Journal Homepage: http://jcp.aip.org/
Journal Information: http://jcp.aip.org/about/about_the_journal
Top downloads: http://jcp.aip.org/features/most_downloaded
Information for Authors: http://jcp.aip.org/authors

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
 THE JOURNAL OF CHEMICAL PHYSICS VOLUME 19. NUMBER 10 OCTOBER. J951

LCAO Molecular Orbital Computation of Resonance Energies of Benzene and Butadiene,

with General Analysis of Theoretical Versus Thermochemical Resonance Energies
ROBERT S. MULLIKEN, Departmmt of Physics, University of Chicago, Chicago, Illinois*

AND

ROBERT G. PARR, Departmmt of Chemistry, Carnegie IlIstitute of Technology, Pittsburgh, Pmllsylvania

(Received June 22, 1951)

The decrease in 1I"z-electron energy for the change from a Kekule to a proper benzene structure is computed
purely theoretically by the method of antisymmetrized products of MO's (molecular orbitals), in LCAO
approximation, using Slater 2P1l"z AO's (atomic orbitals) of effective charge 3.18, and assuming a carbon-
carbon distance of 1.39A. The result (73.1 kcal/mole) is a theoretical value for the gross (vertical) resonance
energy of benzene taken for constant C-C distances of 1.39A. In order to make a comparison with the net
or ordinary empirical resonance energy, several corrections to the latter are required. The principal one is for
the "compression energy" required to compress the single and stretch the double bonds of the Kekule
structure from normal single and double-bond distances to 1.39A. The others (not hitherto clearly recognized)
involve hyperconjugation and related effects. The corrections are discussed and their magnitudes estimated,
but a reliable value can be obtained only for the compression energy. Allowing for this alone, the computed
net resonance energy is 36.5 kcal. This agrees, within the uncertainties due to the omitted correction terms,
with the value (41.8 kcal) of the "observed" resonance energy 6 based on thermochemical data.
,0
Here 6 is the departure of the actual heat of formation 6H of benzene from the value given by a standard
formula for nonresonating hydrocarbons. A new standard formula containing corrections for the mutual
effects of neighboring carbon-carbon bonds is given; this is of interest also for itself, and its significance is
briefly discussed.
The analysis given in the paper serves to clarify hitherto existing obscurities in what is meant by "reso-
nance energy." An analysis of the nature and significance of "nonresonating" structures (like for example
Kekule benzene) is also included, using He2, 1,3 butadiene, and benzene as examples. Repulsion terms in the
LCAO MO theory occur for nonresonating structures, which appear to be the counterpart of exchange
repulsions of the valence-bond theory. When resonance occurs, it more or less overcomes these repulsions.
The gross resonance energies of cis- and trans-l,3-butadiene are computed by the same method. For trans-
butadiene, the computed value corrected for compression is 3.7 kcal/mole, while the "observed" thermo-
chemical value 6 is 6.5.

1. INTRODUCTION The concept of resonance energy has been used in

ECENT successful calculations of electronic excita- both the LCAO MO and in the VB (valence-bond)
R tion energies and vibrational force constants for
the unsaturated organic molecules ethylene,t-3 acety-
methods. In the VB method, the resonance energy W R
of benzene is conceived as the difference between the
lene,3.4 1,3-butadiene, 6 allene, 6 and benzene 7- lo by the energy W K of a single Kekule VB wave function VtK and
method of antisymmetrized products of MO's in LCAO the actual ground-state energy W N,
approximationll have led us to hope that this method (1)
might be of value for computation of resonance energies.
We have accordingly employed it to compute the In the LCAO MO method, following Huckel, conju-
resonance energies of benzene and 1,3-butadiene. gation or resonance energy in olefinic or aromatic
molecules may be conceived as the difference W R be-
* This work was assisted in part by the ONR under Task Order tween the energy W K of a wave function in which the
IX of Contract N60ri-20 with the University of Chicago.
I R. S. Mulliken and C. C. J. Roothaan, Chern. Revs. 41, 219 7r", electrons are assigned to two-center or "localized"
(1947). . LCAO MO's (like those in ethylene) and the energy W N
2 R. G. Parr and B. L. Crawford, Jr., J. Chern. Phys. 16, 526
(1948). of one in which they are assigned to the best n-center
3 R. S. Mulliken, J. chim. phys. 46, 497 (1949). LCAO MO's obtained by solving a secular equation
• G. R. Taylor, D.Sc. thesis, Carnegie Institute of Technology, (n=number of atoms carrying 7r", electrons).12.13 The
1951.
6 R. G. Parr and R. S. Mulliken, J. Chern. Phys. 18, 1338 computation 6f W N for benzene is discussed in Sec. 2,
(1950).
6 R. G. Parr and G. R. Taylor, J. Chern. Phys. 19,497 (1951).
7 B. L. Crawford, Jr., and R. G. Parr, J. Chern. Phys. 17, 726 12 E. HUckel, Z. Physik 70, 204 (1931), 76, 628 (see table on p.
(1949). 636). F. Hund, Z. Physik 73, 1565 (1932)-introducing localized
8 c. c. J. Roothaan and R. G. Parr, J. Chern. Phys. 17, 1001 MO's for polyatomic molecules. J. E. Lennard-Jones and C. A.
(1949). Coulson, Trans. Faraday Soc. 35, 811 (1939) and references given
9 C. c. J. Roothaan, Revs. Modern Phys. 23, 69 (1951). there.
10 Parr, Craig, and Ross, J. Chern. Phys. 18, 1561 (1950). 13 Concerning 11"% and 1t'y electrons, conjugation, and hypercon-
11 M. Goeppert-Mayer and A. L. Sklar, J. Chern. Phys. 6, 645 jugation, see Mulliken, Rieke, and Brown, J. Am. Chern. Soc. 63,
(1938). 41 (1941).
1271

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
 1272 R. S. MULLIKEN AND R. G. PARR

TABLE I. The vertical (or gross) resonance energy of benzene." however, and so cannot be regarded as purely theo-
retical. IS
Contribu-
tion to The results obtained for butadiene in the present
Ground Kekule vertical work are given in Sec. 7.
state structure resonance
Energy term N K energy

100r 10 -40.26 b -40.26 b 2. THE NORMAL STATE OF BENZENE

.~.S6
It =1-1 or 1[=1_1 -37.32b -36.43 b The ground state of benzene in terms of the LCAO MO
JooorJoo 8.63
8.52
8.63
8.52 +0.18
method has been discussed in detail elsewhere.Il-lI.19 One
J01 =Jo-l or J01 =JO_I
.111 =J -1-1 =Jl_lorJ/l =J -1-1 =J1_1 8.47 8.51 assumes that W N may to a reasonably good approxima-
K01 =KO_l or KOI =KO-I 2.58 2.37 tion be written as a sum of a core energy, W N. core and a
-0.57
Kl_10r KI-I 1.45 2.16 11' ,,-electron energy, W N r,

WN7rl.390rWK'/I"1.39 -llS.40 b -112.23 b 3.17

(73.1 (2)
kcal/mole)
and that the wave function iftNr for the six 11'" electrons
• Numerical values are taken or derived from reference 10. All are in ev.
b The energy zero for each 1< electron is taken as the energy of an electron
may be treated separately and written in normalized
in a 2p" orbital of a carbon atom in the tr'" tetravalent valence state form as follows,
(reference 3) (tr = trigonal hybrid orbital).
iftN= <lcpo(l)a(l)cpo(2)(3(2)CPl(3)
that of W K and W R in Sec. 3 and Sec. 4, in each case
X a(3)cpI(4) (3(4)cp_l(6)a(S)cp_l(6)(3(6J. (3)
using antisymmetrized LCAO MO wave functions.
If a computed resonance energy is to be compared Here the"antisymmetrizer" <l denotes (61)-1
with a thermochemically determined one, considerable XL> (-l)"PP, a and (3 are one-electron spin func-
care must be exercised in defining the reference state K tions, and CPo, CPI, CP-I are the 1I',,-electron MO's, formed
to be used in the computation. This phase of the benzene as orthonormal linear combinations of 2p1l'" AO's
question is discussed in some detail in Sec. 3. Further, Xl, X2' •• X6 on the six carbon atoms:
several corrections to the thermochemical resonance
energy (Sec. 5) must be made (Sec. 6), particularly one cPj= Lp CjpXp·
for "compression energy" (Sec. 4).
The compression energy correction was evaluated For W Nl1' one then finds
some time ago, and used to obtain an empirical reso-
nance· energy corrected for compression.t3.14.16 A dis- WNl1'=2IO+2Il+2LI+(4JOI-2Kol) }
cussion along somewhat similar lines has recently been + (4J 0-1- 2KO-- I)+ (4J 1_ 1 - 2K 1_ 1) (4)
given by Hornig. 16 Van Dranen and Ketelaar17 have
recently proposed a method for theoretical calculation +Joo+J u+J-I-I,
of the resonance energy of benzene (and other mole-
where Ii is the energy of an electron in the MO CPi in the
cules) which is closely connected with ours. Their
field of the core, Jij is the coulomb integral between the
method is based on an inconsistent set of equations,
MO's CPi and cP;, and K ij is the exchange integral be-
14 Reference 13, Sec. 13, especially footnote 28, and Table VI. tween these MO's.
15 The revised (unpublished) computations of Mulliken and The coefficients C jp are found by minimizing the
Rieke (Abstract in Revs. Modern Phys.14, 259 (1942)) were based energy W N" with respect to them, subject to ortho-
on force constant values kc_c=4.50X106 and kc_c=9.87X106
dynes/em from analyses of the vibrational spectra of C:JIs and normalization of the cpj.20 The results for the cPj are up
C 2H 4, values which were recommended by Professor B. L. to normalizing factors,
Crawford, Jr., in 1942. However, since these k values were based
on empirical frequencies uncorrected for the effect (due to zero-
point vibration) of anharmonicity, Mulliken and Rieke first in-
creased them by 5 percent to allow very roughly for this effect.
(5)
They then set up Morse curve equations for C-C and for C=C,
using these augmented k values and valence state dissociation
energies D. where w= exp(21!'i/6).
Valence state D's were used because it is to be expected that for Formulas and values for the various quantities which
not too large vibrations, the bonds should tend to dissociate into
carbon atoms in their (tetravalent) valence states. The valence enter Eq. (4) when the cPj are given by Eq. (5) have been
state D values later published in reference 3' (see reference 3, given elsewhere1°-the values are listed in Table 1. These
footnotes** to Tables V and VI) were used; two sets of values were are for the normal1.39A C-C distances and are based
tried, corresponding to the two conflicting current values of the
heat of sublimation of graphite. Fortunately, the final value of CK upon Slater 2p1r AO's with effective charge Z = 3.18.
is so insensitive to the assumed D values, and to moderate varia-
tions (within the range of credibility) of the values of k used, that 18 R. G. Parr, J. Chern. Phys. 19, 799 (1951). Eq. (5) of this
the probable uncertainty in CK due to these causes is estimated as reference should read F=Heor.+G.
only about ±1 kcal. A more serious difficulty is discussed in Sec. 6. 19 R. S. Mulliken, J. chim. phys. 46, 675 (1949).
18 D. F. Hornig, J. Am. Chern. Soc. 72, 5772 (1950). 20 Actually computation is not necessary here, symmetry argu-
17 J. Van Dranen and J. A. A. Ketelaar, J. Chern. Phys.17, 1338 ments giving the same result more simply. In the general case
(1949); 18, 151 (1950). computation is necessary, however.

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
 RESONANCE ENERGY COMPUTATIONS 1273

3. THE LCAO MO KEKULt STRUCTURE OF forms:

BENZENE
!/IN=&alSa(l)a(1)1Sa(2)~(2l1sb(3la(3l1sb(4l~(4l; (7a)
For benzene, the localized LCAO MO's (see Sec. 1)
for a Kekule structure can be chosen as follows, !/IN = acpB(1la(!),pB(2l~(2l,pA (3l a (alcpA (4l~(4). (7b)

Tl=N(Xl+X2); T2=N(X3+X4); Ta=N(X5+X6), (6) The symbols ls a and ls b denote 1s AO's on the two
atoms a and b, while ,pB and ,pA denote (respectively
, N being a normalizing factor. The Kekule wave function bonding and antibonding) LCAO MO's of the forms,
!/IK" can be constructed by replacing the,pj in Eq. (3) by
the T j, and then placing an extra normalizing factor in CPB= [2(1 +S)J-i(ls a + 1s b),
(8)
front of the a (this is needed because the Tj, unlike the
CPj, are not mutually orthogonal). Instead of Eq. (6), one
could equally well use Tl'=N(X2+Xa) and so on, corre- S being the non-orthogonality or overlap integral
sponding to the other Kekule structure. f1s a1s bdv. Because 1s a and 1s b are not mutually
In the earlier LCAO MO calculations of resonance or orthogonal, in contrast to CPB and CPA which are orthogo-
conjugation energy,12,la this was taken as the difference nal, the extra normalizing factor :n is required in Eq.
between sums ::EEj of one-electron energies for the 1r", (7a). The complete identity of the functions described
electrons assigned on the one hand to localized (2- by Eqs. (7a) and (7b), if:n is taken as 1/(1-S2), may be
center) LCAO's, on the other to the best n-center verified by putting both into determinant form.
LCAO's. In this procedure, interelectronic interactions Evaluation of the energy f!/lH!/Idv using Eq. (7a)
were not considered explicitly; instead, an average field leads to the familiar exchange repulsion of VB theory
governing the MO's was assumed.l 2 The relation be- (plus multiple-exchange terms which are usually neg-
tween the ::EEj method and the theoretically more lected in VB discussions) which always occurs between
obviously safe antisymmetrized-product method used nonbonded electrons on two atoms. As will now be
here has been discussed,t9 but not adequately clarified; shown, a corresponding repulsion is obtained in the
a recent note by one of the present authors contains LCAO MO method using Eq. (7b), even in the simplified
procedure12 which uses the resonance terms contained in
some further analysis of this problem.ls
::EEj as an approximation to the change in the total
In the antisymmetrized-product method, the total
energy.22 In the present example, ::EEj becomes 2E8
energy of the 1r '" electrons in the field of the core (nuclei
+2EA, with23
plus non-1r", electrons) is computed from the relation
W ,,= f!/l" *H!/I"dv, where H is the hamiltonian in the EB=a+~/(l+S), EA=a-~/(l-S),}.
form Vcore+H", where Vcore is an effective potential (9)
::EEj= 4a- 4~S/ (1- S2)
corresponding to the field of the core. This can be done
first using !/INr of Eq. (3), then using a similar expression Since ~<O, the sum -4~S/(1-S2) of the resonance
for !/IK described above, the difference being the 1r '" terms makes the total interaction energy positive,
resonance energy. However, a relatively convenient corresponding to a repulsion, if we assume (as seems
expression for W r in terms of ['s, J's, and K's as in justified by experience23 ) that the a's plus other neces-
Eq. (4) is obtained only if the 1r:t LCAO MO's used form sary terms correspond approximately to the energy of
an orthonormal set. This is true of the cp/s of !/IN" as the separate atoms.
given by Eq. (5), but not of the localized MO's Tj of Taking 1,3 butadiene H 2C=CH-CH=CH 2 as a
Eq. (6) used above for !/IKr' The difficulty is overcome second example, a !/IKr using localized bonding MO's
if the T /s are replaced by an equivalent orthonormal set Tl=N(Xl+X2), T2=N(X3+X4);
of delocalized (n-center) LCAO's-which, however, it
N = [2(1+S12)]-t, (10)
should be expressly noted, is not at all equivalent to the
orthonormal set ,pj of best n-center LCAO's. Since the can be constructed as follows,
transformation from a localized to an equivalent !/IK,,= :naTl(1'a(!)Tl(2'~(2'T2(3'a(3'T2(4' ~(4l. (11a)
delocalized set of MO's has implications of general
interest for an understanding of the significance of "non- Equation (lla) is formally analogous to Eq. (7a) for
resonating" structures, it will be instructive at this He2, with the MO's Tl and T2 playing the same roles as
point to consider two other examples before carrying out the AO's ls a and 1s b there. Equation (lla) is related to
the transformation for benzene. 21 the following Eq. (llb) in exactly the same way as (7a)
The wave function for the ground state of two helium is to (7b):
atoms close together, conceived of as an He2 molecule, !/IK,,= a,pKl (lla(ll,pKl (2l ~(2lcpK2(3la(3lcpK2(4) ~(4l, (llb)
can be written in either of the following equivalent
22 This line of thought has been developed somewhat further in a
recent paper: R. S. Mulliken, J. Am. Chern. Soc. 72 4493 (1950)'
21 See also C. A. Coulson, J. chim. phys. 46, 5 (1949) for some and in J. Chern. Phys. 19, 900 (1951). ' ,
instructive remarks on delocaJization of localized MO's. 23 See reference 3, Secs. 7-8.

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
 1274 R. S. MULLIKEN AND R. G. PARR

where the 4-center delocalized MO's CPKj, unlike Tl and shall designate as CPo, CPl, and CP-I, may be taken, up to
T2, are mutually orthogonal i in normalized form, they normalizing factors, as
are as follows (see Eq. (8»,
(14)
CPKl= [2(1 +S23)]-!h+T2),
(12)
CPK2= [2(I-S23)]-!h- T2)' The 7r ,,-electron energy corresponding to this choice of
The energy of 1/IK .. must of course be identically the CPJ is now
same whether computed from Eq. (lla) or Eq. (llb). WK,,=210+211+2Ll+(4JOl-2Kol)
However, comparing the discussion of Eqs. (7a) and
(7b), Eq. (lla) will show an exchange repulsion between +
(4J 0-1- 2Ko_ I) + (4J l-l- 2Kl _l )
the electron-pairs in Tl and T2, while Eq. (llb) with the +JOO+Jrl+J-l-l. (15)
LEj method will show a corresponding repulsion ex-
Equation (15) is formally identical with Eq. (4), but
pressed in the form -'4,623S23/(I-S232). But if the CPKj
differs because the set CPo, CPl, CP-l is different from
in Eq (llb) are replaced by the best 4-center LCAO's,
CPo, CPI, CP-l (except that the MO CPo turns out to be the
which are of the form: same in both).26
CPl. 2= (ajXI+bjX2)±(bix3+ajX4), (13)
4. COMPRESSION ENERGY CORRECTIONS AND THE
with bl>al and the plus sign for CPI, but b2 <a2 and the THEORETICAL LCAO RESONANCE ENERGY
minus sign 6 for CP2, an attraction is superimposed on the OF BENZENE
repulsion just discussed. Nearly (see Sec. 6) the whole For any assumed values of the C- C distances, the
of this attraction corresponds to conjugation energy. quantities in Eq. (15) can be computed by the method
The reason for it is easily understood (b l > al increases used previously for Eq. (4). The difference W K,,- W Nr
bonding in CPI and decreases antibonding in CP2, as can then be obtained. However, for a suitable theo-
compared with Eq. (12) which is a special case of (13) retically computed resonance energy W R to be compared
with aj= bj), and it is easily computed by the LEj with an empirical counterpart from thermochemical
method.13 The antisymmetrized MO-product method data (Sec. 5), we presumably should use a Kekule
gives the conjugation energy in the form W K,,- W NT, structure in which the bonds have the lengths (1.54A
from 1/IK .. as in Eq. (llb), and 1/IN . . as in Eq. (llb) and 1.34A) of normal single and double bonds alter-
modified by substituting the cp/s of Eq. (13). nately, whereas for normal benzene all the bonds are of
The discussion of 1/1 K of benzene now more or less
1f
length 1.39A. That is, we should take
parallels that of butadiene. 24 Using the three non-
orthogonal 2-center LCAO MO's of Eqs. (6), a six- (16)
electron 1/IK .. analogous to that of Eq. (lla) could be
constructed. But instead of the T j, an orthonormal set of where superscripts denote bond lengths, and set out to
three delocalized cPj (analogous to those of Eqs. (12) for compute this quantity. Direct approach through the
butadiene) can be formed as linear combinations of the formula,
Tj, and these can be used to construct a 1/IK" analogous 1.34,1.64 1.39
to that of Eq. (llb). These delocalized cPj, which we W R= (W K,core - WN,core)

24 In the case of benzene (see reference 29), if a is the 7r x exchange

integral (negative in sign), the 7r x bond energy of three isolated
double bonds according to approximate VB theory (see end of is unfortunately not practicable,-the variation of core
Sec. 6) is -3a and the exchange repulsion is +(3/2)a, giving a energy with bond distances is too hard to handle.
net 7r x bond energy of -(3/2)a for the Kekule structure; and the
computed resonance energy WK .. -WN .. is -1.11a, giving an We therefore adopt an alternative approach, by
over-all net computed 1I"x bond energy of -2.61a. (The signs here writing Eq. (16) in the form,
are confusing, but correspond to the usual definition of bond
energies and resonance energies as intrinsically negative.) It seems
anomalous here that even with strong resonance the 7r x bond
energy is less than for three isolated double bonds, but this is what
the theory gives, and there is no reason to suppose that the LCAO with
MO theory would give a different result. The resolution of the
anomaly must lie in the fact that bond energies in molecules with (18)
more than two electrons always include considerable exchange
repulsions, so that the exchange repUlsion term +(3/2)a men- and
tioned above is not really added to the total energy, but is substi-
tu ted for other exchange repulsion terms of the same order of
magnitude which would have been present even if the double bonds
had been associated only with single bonds, as for example in
ethylene (C=C exchange repulsions with C-H bond electrons) 25 Equation (14) corresponds to a symmetry D3h (all single bonds
or propylene or butylene (C=C exchange repulsions with C-C equivalent, all double bonds equivalent, with the single and double
and C-H bond electrons). (See Syrkin and Dyatkina, reference bonds possibly but not necessarily equal in length). Like <1>1 and
26, p. 242, for a survey of exchange repulsion and other terms <1>-1 of Eq. (3), <1>1 and <1>-1 are degenerate, hence could be replaced
included in the normal bond energies of the paraffins.) by any mutually orthogonal pair of their linear combinations.

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
 RESONANCE ENERGY COMPUTATIONS 1275

resonance with which we are now concerned. It is

desirable to make some allowance for these before
arriving at .1. Following are suitable formulas for
hydrocarbons, not containing triple bonds, in which no
strong steric effects occur :26.27
.1Hi= xNc-H+yNc-c+zNc=c
1.39 1.39
since the quantity is doubtless so
(W K.core- W N.core) -'YlnC-C-'Y2nc=C-.1 (23)
small that it can safely be neglected. In Eq. (18), CK .1Hl(OOK) = -3,55Nc_H-5.00Nc_c
denotes "compression energy." W K .. 1.39_ W N.-1.39 can be
computed from Eqs. (4) and (15), and CK estimated +28.72Nc=c-0.78n-.1. (23a)
rather precisely from empirical (primarily force constant) .1H i here is in kcal/mole. In Eqs. (23), (23a), ne-c is
data if the Kekule bonds are assumed to behave under the sum, taken over all C - C single bonds, of the
compression or stretching like the single and double numbers of bonds going out from the two carbon atoms
bonds in ethane and ethylene. of the given bond to other carbon atoms; nc=c is a
Results of the calculation of W K".1.39 are given in similar sum for C= C double bonds, and n is nc-c+nc=c.
Table 1. From these, For example, n=O for CH 4, C 2H 4, or C 2H 6 ; n=2 for
propane; nc-c=2 and nc=c=l, so n=3, for propylene;
and n= 12 for neopentane (C(CH 3)4)' The term n is
= 3.17 ev= 73.1 kcal/mole. (20) designed to take into account, as well as can be judged
from data on aliphatic molecules, the mutual effect,
The quantity W R1.39 may be called the gross or 1.39A exclusive of 7C" z resonance, of first-neighbor carbon-
vertical resonance energy, the latter because it corre- carbon bonds on .1H fO. This type of correction is of
sponds to a "vertical" process in the terminology used in obvious relevance for Kekule benzene. 28
applications of the Franck-Condon principle, that is, an For benzene, Eq, (23a) yields .1=41.82 kcal/mole,
electronic change with no accompanying change in differing little from the (25°C) value of 41 kcal/mole
internuclear distances. On the other hand, W R is a net obtained by Pauling's method. 29 This value of .1 is in
resonance energy.
26 The basis and significance of Eqs. (23), (23a) may be discussed
For C K , Mulliken, Rieke, and Brown 14 obtained 35.0
more fully in a later paper by R. S. Mulliken. Equation (23) is an
kcal, while in an improved computation, Mulliken and improvement on a somewhat similar equation given in Sec. 12 and
Rieke15 obtained 36.6 kcal with an estimated uncertainty Eqs. (23) and (27) of reference 13. The coefficients in Eq. (23a)
of ±1 kcal. Using the latter value, we find have been adjusted to fit the most recent AHl o (OOK) data given
in the A.P,I. tables.
The term -0.78n is interpreted as arising from substitution-
W R (the'oretical) =36.5 kcal/mole. (21) induced variations in second-order hyperconjugation and other
second-order resonance effects, tempered by steric repulsions.
S. COMPARISON OF THERMOCHEMICAL AND Equation (23a) fits observed data well for most not too much
THEORETICAL RESONANCE ENERGIES branched hydrocarbons not containing triple bonds: IAI ~0.5 for
OF BENZENE acyclic aliphatic hydrocarbons of this category, except A= 1.8 for
methane (this is much better than by most formulas). For dia-
We now seek an empirical value of W R which can mond, A=0.06. For cyclohexane (n= 12), A= -1.95, for cyclo-
pentane (n = 10) A = -7.62; these negative A'S may be attributed
properly be compared with the theoretical value ob- to ring strain. In molecules with first-order 1I"x hyperconjugation,
tained in Sec. 4. Since W N is known empirically, the Eq. (23a) yields small positive A'S (e.g., 1.6 for propylene), and for
essential question is how to obtain a valid "empirical" molecules with 1I"x conjugation, larger positive A'S (e.g. 6.48 for
1,3-butadiene),
WK. Pauling answered this by setting up a standard See also J. R. Platt, J. Chern. Phys. 15,419 (1947); H. Wiener,
bond energy formula of additive type designed to fit J. Chern. Phys. 15, 766 (1947); J. G. M. Bremner and G. D.
normal molecules believed to be "nonresonating." For Thomas, Trans. Faraday Soc. 44, 338 (1948); and particularly
Y. K. Syrkin and M. E. Dyatkina, Structure of Molecules and the
any unsaturated hydrocarbon, this may be written Chemical Bond, translated and revised by M. A. Partridge and D.
O. Jordan (Interscience Publishers, New York, 1950), pp. 243-4,
B=xNc-H+yNe-c+zNc=c+.1, (22) for relatd consiierations.
27 Since our theoretical W R is purely electronic, a OOK equation
like (23a) is preferable to the usual 25°C kind of equation. Still
B being the observed bond energy, and .1 any deviation better would be an equation corrected for zero-point energy, but
from the standard energy,.1 if positive being interpreted we doubt that this would make an important difference.
28 Instead of using an equation like (23a), a common alternative
as resonance energy. Similar formulas can be given for
procedure (see reference 29) is to take A as the difference between
heats of combustion or hydrogenation, or for heats of the heat of hydrogenation of one mole of benzene and that of three
formation .1H i, it being a matter of convenience which moles of cyclohexene. In our opinion, however, this involves more
is used, since all if made equally accurate should give the doubtful points than the use of Eq. (23). In particular, there seems
to be no reason to expect ring strain in either actual or Kekule
same .1. benzene, with either normal single and double bond lengths or
However, when a formula such as Eq. (22) is used, 1.39A bond lengths; in all these cases, a planar structure with 120°
bond angles should be free from strain.
small systematic deviations .1 occur, even in the par- 2. See for example, G. W. Wheland, The Theory of Resonance
affins, which cannot be attributed to the kind of 7C" z (John Wiley and Sons, Inc., New York, 1944).

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
 1276 R. S. MULLIKEN AND R. G. PARR

satisfactory agreement with the theoretical value 36.5 energy, including some u and 'Try hyperconjugation
kcal obtained for W R at the end of Sec. 4. energy; for example, a fairly large amount of 1l'y
However, a more careful analysis shows that we hyperconjugation energy is .believed to be associated
should have made two corrections of some importance, with every C= C double bond. l3 In "resonating"
the one to .1, the other to C K as given in Sec. 4, before unsaturated and aromatic molecules, the second-order
comparing empirical and theoretical resonance energies. 'Tr x hyperconjugation energy in W Sr is superseded by
These will be considered in the next section. energy of first-order 'Trx resonance (that is, of first-order
hyperconjugation as for example in propylene, or of
6. THEORETICAL ANALYSIS OF MEANING OF
THERMOCHEMICAL RESONANCE ENERGIES, conjugation as in butadiene, or of benzene-like reso-
WITH APPLICATION TO BENZENE nance), it being this change from higher-order to first-
order 'Tr x resonance energy which corresponds to .1; but
In order to make, so far as possible, a clean-cut com-
W s, core should be affected only indirectly and very
parison between theory and experiment, it will be useful
slightly.
now to write an equation relating the theoretical
quantity W R,,1.39 which we have actually computed to a We need now to examine carefully how W K .. for a
sum of empirically based terms including C K (cf. Eqs. hypothetical "non-resonating" KekuI6-type structure is
(18)-(19» and the other correction terms mentioned at related to W s". Since by its definition, the Kekule
the end of Sec. 5. It will prove useful first to introduce structure contains no 'Tr x resonance energy whatever
the concept of a "standard hydrocarbon," defined as one (see Sec. 3), while, as we have seen, the "standard"
for which Eqs. (23), (23a) hold with .1=0. Here also we hydrocarbon contains an amount 5. of 'Tr x resonance
note that Eq. (23) should be valid not only for hydro- energy per single bond of length s, we have
carbons with carbon-carbon bonds of the normal (25)
unconjugated equilibrium lengths 1.34A and 1.54A (in
this case Eq. (23a) applies) but also, after suitable the (positive) energy being the greater for W K"d .••
adjustments in the values of the coefficients, for hydro- For benzene at its actual bond lengths, Eq. (25)
carbons with the C= C and C-C bonds stretched or becomes
compressed to any values d and s, respectively. The
values of the coefficients should then be functions of d
and s. We shall denote the corresponding "standard
energy" by W Sd.,. Our theoretical resonance energy W R".1.39 is then
We next note that W Sd •• , even though it corresponds
to a "nonresonating" hydrocarbon (that is, one with no 1.39
conspicuous 'Trx conjugation or resonance energy), must = W Kr1.39- (W N1.39_ W N,core)
actually include a perhaps very appreciable amount of 1.39 1.39
what may be called higher order resonance energy, = (W S1.39- W N1.39)+3 51.39+ (WN, core- W S,core)'
corresponding to numerous minor contributions of
Or, assuming the last term in parentheses to be
miscellaneous VB structures present in the actual wave 1.39 1.39
function. Part of this has been recognized under the negligible, like W K,core- W.v,core in Eqs. (19),
name of second-order hyperconjugation energy, this in
turn usually being described in terms of u and 'Tr
hyperconjugation energy, and the latter being divided + (W S1.39- W S1.34. l.54)+3(J1.39
into 'Trx and 'Try hyperconjugation energy.13
The most important point in the present connection =.1+C s +351.39, (26)
is that W Sd •• includes second-order 'Tr x hyperconjugation
energy. Further, if we write W s as a sum of core and 'Trx since .1 is by definition (see Eq. (23» equal to the first
energy, like W in Eq. (2), that is, term in parentheses above, and where Cs denotes the
energy required to stretch-compress a structure like a
(24) Kekule structure but with standard single and double
carbon-carbon bonds from s= 1.54A, d= 1.34A to
then this second-order 'Trx hyperconjugation energy is s=d=1.39A.

I
part of W Sr, the remainder and major portion of W s .. From Eq. (23) with the values Nc-c=Nc_c=3,
being the ordinary nonresonant 'Tr '" energy of any double nc-c= 12, nc_c= 6, appropriate to benzene, Csis
bonds which are present. The amount of second-order
'Trx hyperconjugation energy is believed to be approxi- s so
mately proportional to the number of C- C single C =c -.1r,
bonds; the amount per single bond of length s (normal where
(27)
or compressed) has been called 5".13 Cso= 3[(y1.39_ y l.54)+ (Z1.39_ Z1.34)J,
The term W S,core in Eq. (24) contains (among other
things) the remainder of the second-order resonance .1r= 12("(11.39- "(11.54)+ 6("(21.39- "(21.34).

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
 RESONANCE ENERGY COMPUTATIONS 1277

TABLE II. The vertical (or gross) resonance energy of 1,3-butadiene."

II N on~resonating" Contribution to
Energy Ground state N structure K vertical resonance energy
term cis~ I,ans- cis· trans- cis- trans-

I, or I r -26.79 b -26.38 b -26.20b -25.74b

0.72 0.73
12 or 1 II -23.45 b -22.91b -23.67 b -23.18 b
JllorJII 10.10 9.99 9.68 9.56
J 22 or JII II 9.55 9.36 9.58 9.42 0.15 0.08
J'2 or Jr II 9.40 9.28 9.53 9.40
·K 12 or Kr II 3.28 3.42 3.55 3.68 -0.54 -0.53

WN or WK -49.78 b -48.93 b -49.44 b -48.65 b 0.34 0.28

(7.8 kcal/mole) (6.5 kcal/mole)

• Numerical values are taken or derived from reference S. All are in ev.
b See Table I. footnote b.

The term Gso represents the energy to compress three empirical or unknown quantities on the right is thus well
standard C - C single bonds, of type nc-c = 0 as in within the uncertainties of the latter.
ethane, and to stretch three standard C = C double It seems worth pointing out here that Eq. (28) for
bonds, of type nc=c=O as in ethylene, to 1.39A. Since W Rr1.39 is equally as valid for calculations made by the
Eq. (23a) fits either ethane or ethylene with .1=0, Gso VB as for those made by the LCAO MO method. How-
may be set equal to the actual energy required to com- ever, existing calculations by the VB method appear to
press the carbon-carbon bonds in three ethane and be confined to the early crude result W R,,= 1. 11 a, where
stretch those in three ethylene molecules to 1.39A. A a is the exchange integral for 1r., electrons on adjacent
reliable empirical value (36.6± 1 kcal) has been given carbon atoms. 24 These calculations took into account
for this quantity at the end of Sec. 4. Equation (26) now only Kekule and Dewar structures, neglected higher
becomes exchange integrals, and involved no attempt to compute
a itself theoretically. In the usual discussions,29 a is
W Rr 1.39= ~+G so+ (351.3L .1r) ; (28)
determined empirically by setting W R .. = 1.11a equal to
or, introducing numerical values from Secs. 4 and 5, .1, ignoring the large correction term Gso as well as the
73.1 = 41.8+ 36.6+ (3151.39- .1r). (29) smaller ones which follow it in Eq. (28).
Unfortunately, no reliable estimate is available for the 7. RESONANCE ENERGIES OF cis- AND trans-
1,3-BUTADIENE
correction term in parentheses. The sign of the hyper-
conjugation energy term 3151.39 is surely positive. Em- The vertical or gross resonance energies of cis- and
pirically, 1'11.64 and 1'21.34 are positive (see Eq. (23a». If trans-1,3-butadiene, for molecular dimensions equal to
the terms 'Y1nC-C and 'Y2nC=C have been interpreted those in the actual ground states, can be computed from
correctly in Sec. 5 as enhancements of second-order W Nr- W K" following the same general procedure as for
resonance energy, 26 they would be expected to increase, benzene. We have
. and fairly fast, with decreasing s or d. This would make
.1r positive. W N .. = 2I1+2I2+Jll+4J12- 2K12 (30)
A quantitative value has been given for 151.39 in from Eq. (5) of reference 5 (corresponding to use of
reference 13, from which 3151.39= 11.8 kcal. This value "best" LCAO MO's), while
cannot be accepted as accurate, although it is probably
correct as to order of magnitude. 3o For .1r, a plausible WK .. =2Ir+2In +h r+4h n-2Kr n, (31)
guess would be that 1'11.39= 21'11.64, hence 1'11.39- 1'11.64 with the integrals I, J, K based on the delocalized but
=0.8 (see Eq. (23a», and .::lr about 8 kcal. Thus it is "non-resonating" butadiene MO's of Eq. (12) above.
likely that (351.39-.1r) in Eq. (29) may be near zero, W K.. is the 1r-electron energy corresponding to iflK" of
although possible that it may amount to ±1O kcal or Eq. (l1b) above. The basic integrals required may be
even more. The agreement between the theoretical found in reference 5. The results are given in Table II.
W Rr1.39 on the left of Eq. (29) and the sum of the The computed gross or vertical resonance energies,
30 Reference 13, Table VII. However, since the computations
after a correction 31 of - 2.8 kcal for "compression
were made by the ~Ei method with neglect of the overlap integral energy" to obtain a theoretical value for the ordinary
S, and involve also other assumptions, they cannot be accepted as (net or nonvertical) resonance energy W R, as was done
accurate. Moreover, as has been found by one of the authors
(RSM) they would be altered (though not changed as to order of for benzene in Sec. 4, may be compared with the
magnitude) if revised to correct an error in one of the equations thermochemical resonance energy of 6.5 kcal obtained 26
of key importance in reference 13, namely, Eq. (32), where liB and
II'should be substituted for 1i!.64 and 1i1.33 (footnote 35 is incorrect). 31 Reference 13, Table VI.

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions
 1278 GEORGE PORTER

in the manner described in Sec. 5, using Eq. (23a). Since Or more accurately, when additional small corrections
ordinary 1,3-butadiene is probably nearly all in the of the type considered for benzene in Sec. 6 are included,
trans-form,32 we use the trans-result from Table II, and we may write in analogy to Eq. (29) for benzene,
obtain
6.5 = 6.5+ 2.8+[01.46_4( 1'11.46- 1'11.64)
computed W R=6.5-2.8=3.7;
(32) - 2(1'21.3L 1'21.34)]' (33)
observed W R = 6.5 kcal! mole.
32 Aston, Szasz, Woolley, and Brickwedde, J. Chern. Phys. 14,
The agreement between theory and experiment is not
67 (1946). quite as good as that for benzene.

THE ]OUR"'AL OF CHEMICAL PHYSICS VOLUME 19. NUMBER 10 OCTOBER. 1951

Isotope Effect in the Band Spectrum of HN0 2

GEORGE PORTER
Physical Chemistry Laboratory, University of Cambridge, Cambridge, England
(Received April 23, 1951)

The band system between 3100 and 3900A which has been attributed both to HN0 2 and enhanced N0 2 is
shown to belong to the former molecule by means of the isotope effect. Measurements of the spectrum of
DN0 2 , and a partial interpretation of the band structure and isotope shift are given.

T HERE has been considerable discussion about the

identity of the molecule responsible for the system
of bands in the region 3100 to 3900A observed by
is little doubt that the two systems are identical but
that they may be due either to HN0 2 or the carrier,
-N0 2•
Melvin and WUlfl and by Bone, Newitt, ~nd Outridge. 2 It would clearly be helpful if the origin of these bands
Melvin and Wulf discovered the system in the products could be settled, for the molecule must in any case be
of photochemical decomposition of N 20 and originally rather unstable and not easily studied by other methods.
attributed it to N 20 3; but they subsequently obtained The main lack of agreement is as to whether or not
the same spectrum in mixtures of NO, N0 2, and H 20, hydrogen is necessary for their appearance, and it was
found that all three molecules are necessary to produce thought possible that the isotopic molecular spectrum,
the bands, and suggested HONO as the molecule re- obtained with deuterium in place of hydrogen would
sponsible. They mentioned, however, that the bands solve the problem unequivocably as well as assist in the
agreed very closely in wavelength with some of the interpretation of the bands. It will be seen that this is
stronger bands of N0 2, suggesting the possibility that the case.
the system could be due to selective enhancement of
EXPERIMENTAL
certain bands of this molecule.
Newitt and Outridge pointed out that the measure- Absorption spectra were taken of mixtures of NO,
ments of Melvin and Wulf agreed with those obtained N0 2 and H 20, or D 20 at various partial pressures. The
earlier by Bone and Newitt in self-absorption of ex- nitric oxide was prepared from sodium nitrite and 10 .
plosions of CO/NO/N 20 mixtures at high pressure. percent sulfuric acid and the nitrogen dioxide by
They repeated this work with very dry gases and also thermal decomposition of lead nitrate, both being
showed that the spectrum was present in CO/N 20 fractionally distilled in vacuum before use. The heavy
flames. In view of the fact that the intensity was not water was 99.7 percent D 20 supplied by the Norsk
much diminished by intensive drying they challenged Hydro Company. The absorption cell was of quartz and
the interpretation of Melvin and Wulf and attributed 100 cm in length, the spectrograph was a quartz Littrow
the spectrum to selective enhancement of certain N0 2 instrument with resolving power of 0.15A at 3000A and
bands. Thompson3 points out the close similarity be- 0.6A at 4000A and the source of continuum was a
tween these bands and those of the alkyl nitrates which xenon-filled flash discharge tube. 6 When using a sub-
supports the assignment to HN0 2, but suggests that the stance with a high absorption coefficient, like N0 2, there
system of Newitt and Outridge may be different from is danger that in the process of recording its spectrum
that of Melvin and Wulf. Gaydon4 concludes that there the absorbed light may initiate photochemical reactions
IE. H. Melvin and O. R. Wuif, J. Chern. Phys. 3, 755 (1935). which, especially if chains are possible, may produce
2 D. M. Newitt and L. E. Outridge, J. Chern. Phys. 6, 752 sufficient product to be detected in the spectra. Nearly
(1938).
3 H. W. Thompson, J. Chern. Phys. 7, 136 (1939). Pearse, The Identification of M oleeular S peetra (Chapman and
4 A. G. Gaydon, Spectroscopy and Combustion Theory (Chapman Hall, Ltd., London, 1950).
and Hall, Ltd., London, 1949); A. G. Gaydon and R. W. B. 6 G. Porter, Proc. Royal Soc. (London) A284, 200 (1950).

Downloaded 19 Jun 2013 to 128.230.234.162. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions