Comparative Calculations of LCAO-MOs of

the Allyl Radical

Cite as: J. Chem. Phys. 26, 151 (1957); https://doi.org/10.1063/1.1743241
Submitted: 21 March 1956 • Published Online: 06 October 2004

Jiro Higuchi

ARTICLES YOU MAY BE INTERESTED IN

Dielectric Behavior of Water at Microwave Frequencies

The Journal of Chemical Physics 26, 156 (1957); https://doi.org/10.1063/1.1743242

Relative Rates of Some Very Rapid Gaseous Bimolecular Reactions

The Journal of Chemical Physics 26, 162 (1957); https://doi.org/10.1063/1.1743243

On the Relaxation of Vibrational Nonequilibrium Distributions. III. The Effect of Radiative

Transitions on the Relaxation Behavior
The Journal of Chemical Physics 26, 137 (1957); https://doi.org/10.1063/1.1743238

J. Chem. Phys. 26, 151 (1957); https://doi.org/10.1063/1.1743241 26, 151

© 1957 American Institute of Physics.

THE JOURNAL OF CHEMICAL PHYSICS VOLUME 26. NUMBER 1 JANUARY. 1957

Comparative Calculations of LCAO-MOs of the Allyl Radical

JIRa HIGUCHI
Laboratory of Physical Chemistry, Tokyo Institute of Technology, Ookayama, Meguro-ku, Tokyo
(Received March 21, 1956)

Comparative calculations of LCAO-MOs are made for the ".-electron system of the allyl radical, the
LCAO-MOs obtained by various methods being discussed from the point of view of form, charge distribution
and energy. For reference, the results using both the MO method with configuration interaction (CI) and
the valence-bond method are given.

INTRODUCTION dition minimizing the energy of the state in question

SINCE the self-consistent field (SCF) procedure was
introduced in the LCAO-MO method by Roothaan,l
this treatment has directly been applied to the ground
state of many molecules without singly occupied or-
bitals. Also, modified treatments have been made on
free radicals or ions with some singly occupied orbitals.
The latter treatments were, however, carried out in vari-
ous ways under the orthonormality condition for orbitals:
(1) the determination of coefficients of LCAO-MOs is
made, one kind for each orbital irrespective of spin func-
tion, under the condition minimizing the energy of the
state in question, asin the Hartree-Fock method for atoms
with singly occupied orbitals2 ; an application of this
method was made to CHa and other radicals4- 6 by the
present author; (2) the determination of coefficients of
LCAO-MOs is made, one kind for each orbital with
each spin function, under the condition minimizing the
energy of the state in question, as introduced by
Berthier7 and by Pople and N esbet8 ; an application of
this method was made on allyl and other radicals by
Berthier9 and on the excited state of l,3-butadiene by
NesbetlO (In the present paper, orbitals determined by
this method will be called BPN orbitals); (3) the co-
efficients of LCAO-MOs are approximated by those of
the corresponding positive ion with no singly occupied
orbital owing to the difficulty of the procedure (1); an
application of this method was made on benzyl radical
by Komatsu, Mori, and Tanaka.!l
In addition, (4) the coefficients of LCAO-MOs are
approximated by those of the corresponding negative
ion with no singly occupied orbitals; (5) the coefficients
of LCAO-MOs are approximated by those for the other
state located near the state in question; (6) the coeffi-
cients of LCAO-MOs are determined under the con-
I C. C. J. Roothaan, Revs. Modern Phys. 23, 69 (1951).
2 D. R. Hartree and W. Hartree, Proc. Roy. Soc. (London)
AI54, 588 (1936).
3 Jira Higuchi, J. Chern. Phys. 22, 1339 (1954).
4 Jiro Higuchi, J. Chern. Phys. 22, 1467 (1954); Bull. Chern.
Soc. Japan 28, 238 (1955).
6 Jiro Higuchi, J. Chern. Phys. 23, 2197 (1955).
6 Jiro Higuchi, J. Chern. Phys. 24, 535 (1956).
7 G. Berthier, Compt. rend. 238,91 (1954); J. chim. phys. 51,
363 (1954).
8 J. A. Pople and R. K. Nesbet, J. Chern. Phys. 22, 571 (1954).
9 G. Berthier, J. chim. phys. 52,141 (1955).
10 R. K. Nesbet, Proc. Roy. Soc. (London) A230, 322 (1955).
II Komatsu, Mori, and Tanaka, J Chern. Soc. Japan 77, 643
(1956); I. Tanaka and C. Komatsu, J. Chern. Phys. 23, 976 (1955).
without exchange terms for MOs, as in the Hartree
method for atoms; (7) the coefficients of LCAO-MOs
are determined by the simple LCAO-MO method
without overlap integrals and (8) with overlap
integrals. 12
No comparative calculations of LCAO-MOs by these
various methods have been made; the present work
takes up the 1!'-electron system of the allyl radical as the
simplest example.
OUTLINE OF CALCULATIONS
The allyl radical HZCa-CbH-CcHz has been investi-
gated theoretically by many authors using either the
valence-bond method18 or the MO method.9.12.14 In the
present calculation, it is assumed that the radical is
planar (C 2v ) and the internuclear distances and the
bond angles are the same as those of the benzene
molecule. This makes possible use of the energy integrals
from the paper by Parr, Craig, and Ross.!n
The 1!'-electron LCAO-l\lOs c/>'s are given by
1b1= all (Xa+Xc)+ a12Xb
2b 1 = aZl (Xa+Xc) - aZZXb (1)
Here, the orbital energy of 1b 1 is lower than that of
2b 1, and the coefficients alp's of atomic orbitals x's are
chosen to satisfy the orthonormality conditions.
The procedure (2) can be seen from reference 7 and
8; the procedures (3) and (4) are the same as in the
common case. I In the present calculation, the procedure
(5) was carried out for the 2Bl states (lb 1)2(2b 1 ) and
(lb l )(laz)2 by using procedure (1). Procedures (7) and
(8) were carried on the assumption that all/ alZ= 2-!
and [2(1+S ac )]-1, respectively. Procedures (1) and (6)
will be described below.
If the doubly occupied, singly occupied and vacant
orbitals are represented by c/>i(i<O), c/>o and c/>j(j>O),
respectively, the electron configuration of the state in
question here can generally be described as
12 W. Moffitt, Proc. Roy. Soc. (London) A2I8, 486 (1953).
13 Lefkovits, Fain, and Matsen, J. Chern. Phys. 23,1690 (1955).
14 O. Chalvet and W. Daudel, J. chim. phys. 49, 629 (1952).
15 Parr, Craig, and Ross, J. Chern. Phys. 18, 1561 (1950).

151
152 lIRO HIGUCHI

Accordingly, the electronic energy of the state is as Another condition that is necessary is
follows:
-1

-1 -1 (H:iO)+ L [2(iO:kk)-(ik:kO)]+(iO:00)=0 (14)

E=2 L (H:ii)+ L [2(ii:kk)- (ik:ki)] k-n
i=- n i,k=-- n
which is also satisfied in the corresponding negative ion
-1
+(H:OO)+ L [2(ii:OO)-(iO:Oi)] (2) with configuration (l/>_n)2(1/>_n+1)2 ..• (1/>_2)2(1/>_1)2(1/>0)2.
i=-n For the vacant orbital I/>j(j> 0), if Ejj is also repre-
sented by Eq. (10), the one electron excitation energies
where H is the Hamiltonian for the core field. are as follows:
If the molecular orbitals I/>/s are represented by
K'+i= Ejj- Eii- [(ii: jj)- (ij: ji)]
LCAO-MOs
(3) +t[(iO: Oi)+ (OJ: jO)]
- (iO: Oi)- (ij: ji)- (OJ: jO)
the self-consistent value of the aip's, which are the (to the quintet state) (15)
elements of the column vectors a's, are determined by
the condition oE=O subject to the orthonormality + { ±[t{ (iO:Oi)- (ij: ji)}2+t{ (ij: ji)- (OJ: JOW
condition. Introducing the matrices with elements +H (OJ: jO)- (iO: Oi)P]!
(to the doublet states) (16)
(4) and
EO...i= Eii- Eoo-t[2(00: jj)- (OJ: jO)]
(5) (to the doublet state). (17)
If the distribution of electrons is not changed by
ionization, orbital energies are good approximations to
(6) the ionization energies of the corresponding orbitals in
molecules without singly occupied orbitals, since the
and ionized state in question is only one and the energy
difference between these states can be represented by
(7) the orbital energy. In the ionization of the doublet
state in question here, however, there is one singlet and
the SCF condition is reduced to one triplet state arising from each doubly occupied
orbital and one singlet state from the singly occupied
[H+ k I: (2J(k)-K(kl)+H2J(OLK(Ol) ]ai
n
orbital. Accordingly, the approximate ionization ener-
gies must be given by

= EiiSai+tEiOSaO (i <0) (8)

J i = - E'i+ { ~t} (i0 :iO)
I: I:
[ H+ k~n (2J(kL K(kl)]ao= k~n EkOSak+EOOSaO. (9)
to the triplet ionized state t
{ (18)
to the singlet ionized state r
Accordingly, the orbital energies are given by and
-1
J0= - fOO. {to the singlet ionized state}. (19)
Ei,= (H:ii)+ L [2(ii:kk)- (ik:ki)]
In this case, however, these orbital energies are of
k=-n
course the measure representing the ionization energies
+t[2(ii:00)- (iO:Oi)] (i<O) (10) and also the excitation energies. 16 ,17
-1
The Hartree SCF-LCAO-MO treatment can be
EOO= (H:OO)+ L [2(00:kk)- (Ok:kO)] (11) performed similarly but without exchange matrix K;
k~n that is, the Eqs. (8) and (9) are replaced by the fol-
and
16 For vacant orbitalsct>;'s, the use of E;/i) = Eii- [(ii:jj) - (ij:ji) ]
-1
instead of Eii may be more convenient, as discussed by Mulliken
E.o=2(H:iO)+2 L [2(iO:kk)-(ik:kO)]+(iO:OO) (12) [R. S. Mulliken, J. chim. phys. 46, 497 (1949)].
k~n
17 In the previous paper on NH, NH., and NH, (reference 6),

-1
the author omitted the relation in Eqs. (15) to (19) for the NH z
radical, and similar relations for NH radical. The ionization energy
= (H:iO)+ L [2(iO:kk)- (ik:kO)]. (13) calculated for NH z (p. 540 I. 8) should read 13.55 ev instead of
k~n
14.16 ev.
 LCAO MOs OF THE ALLYL RADICAL 153

(ev]

-10

-IS

-2.0

a.,) 4A2. (I 4.)(ICla)(a

-lS~--~~--------------~------~------------~~

0.+ 0.5' 0.6 0.7 0.8

FIG. 1. Energy levels without CI as a function of the charge distribution: energy zero is 3W.p •
154 JIRO HIGUCHI

TABLE I. Coefficients of LCAO-MOs' and orbital energies. b

all au a21 a22 a" £ lb l 1b1 £21.11 21.11 Elo2Ia2 £lbl%bl

(I) The ground state 'A, (lbtJ'(la,) Mas 0.4095 0.6220 0.6210 0.87.13 0.7213 -0.332 16.426 2.281 0.000
ground state 'A, (lbla)(lbls)(la,a) BPN MOs{~
(2) The 0.5928 0.2957 0.4492 1.0306 0.7213 -1.878 16.969 0.522 0.000
0.1846 0.8727 0.7205 0.6228 0.7213 -1.043 18.160 15.546 0.000
(3) The positive ion ground state IA I (I btJ' Mas 0.4033 0.6305 0.6250 0.8672 0.7213 -8.848 7.741 2.286 0.000
(4) The negative ion ground state IAI (lbl)'(1a,)' Mas 0.4154 0.6136 0.6170 0.8792 0.7213 8.183 25.113 13.712 0.000
(4a) The negative ion excited state IAI (lbl)'(2btJ' Mas 0.5054 0.4710 0.5457 0.9632 0.7213 11.283 19.877 19.712 0.000
(5a) The 'Bl state (lbtJ'(2bll Mas 0.5054 0.4710 0.5457 0.9632 0.7213 1.213 6.688 10.959 2.242
(5b) The 'Bl state (lbl)(la,)' Mas 0.2967 0.7628 0.6821 0.7535 0.7213 -3.716 5.008 5.219 0.000
(6) The ground state 'A, (lbtJ'(la,) Hartree MOs 0.3421 0.7101 0.6605 0.8034 0.7213 1.325 9.581 4.570 0.000
(7) The "simply determined" Mas [all/ a12 =2-'J 0.4245 0.6004 0.6108 0.8883 0.7213
(8) The "simply determined" Mas [au/au = [2 (1 +5a,) )-'J 0.4205 0.6062 0.6135 0.8844 0.7213

Ibl =all (Xa +X,) +a12Xb

a
{ 2bl =a21 (Xa +Xc) -a22Xb
la, =a"(Xa -x,).
b Values of orbital energies in ev; energy zero W2p = -11.28 ev of a 2P1t electron of a carbon atom in its valence state.

lowing equations: MOs (2) are somewhat different from the majority,
but the 2A2 state MOs (1), the positive and negative
[H+2 k~ /(k)-J(i)+J(O)')ai=~i,Sai+hioSao (20) ion ground state MOs, (3) and (4), and the "simply
determined" MOs, (7) and (8), resemble each other
very much. Furthermore, in the pair of the positive
[H+2 kEn J(kl]ao= k~n ~kOSak+~OOSao. (21) and negative ion ground state MOs, that of the 2Bl
states MOs and the pair of the BPN MOs, the devi-
ations from the 2A2 state MOs are nearly the same
Besides these, the calculations by using both the l\I0
degree in each pair, due to the fairly small ionic con-
method with CI and the valence-bond method were
tribution in the ground state 2A 2 and the first excited
carried out. state 2B 1•
RESULTS
Compared with the ion ground state MOs and the
LCAO-MOs obtained are given in Table 1. In these, "simply determined" MOs, the Hartree SCF MOs (6)
two of the 2BI state MOs, (5a) and (5b), and the BPN are not so good, because the total Hartree wave function
does not perfectly satisfy the Pauli principle and con-
sequently the charge distribution changes somewhat.
~-8.9g On the other hand, the better approximation of the
-10
"simply determined" MOs is due to the small ionic
----IO.lS 'A
'6. --'--'0.81 contribution to the ground state 2A2 and the small
overlap integrals. Both the ion ground state MOs are
extremely good approximations and the mean value of
'8. -13.44 the charge distribution of the 2A2 state given by them
=-I~.S'
's
--:---1 .... 81 ~ -1".76
~-14.48 is nearly the same as that given by the 2A2 state MOs.
-IS +~_ aA, -I"." These facts are helpful in obtaining the best LCAO-
~-15.~7 ~-IS.% MOs for the ground state of radicals, but not always
-,---".53
----16.51
'A, for the excited states. For example, the 2Bl state
~-17.4q j!l.,
(1b l )2(2b l ) MOs are those of the lAl state (1b l )2(2b l )2
-,---17.98
B. of the negative ion with the same effective nuclear
~-i~.lq -,---19.08
B, charge. 1s
-zo ----19.q8
'A, The orbital energy of the 1a2 MO is obtained as
~-ZI.H -9.00 ev and -10.28 ev by the procedures (1) and
'B, -21. 7:1. (2), respectively,19 while the ionization energy observed
is 8.16 ev. 20 Strictly speaking, however, the ionization
energy should be calculated as the energy difference
between the normal and the ionized states by using for
each the best MOs with CI, in which the calculation
for the ion must be carried out by changing the effective
----27.1S'
nuclear charge of atom. Accordingly, the value of
~-27.n ----28.03 'A,
'A, In this case, Eq. (14) is clearly satisfied, and <;o's except for
IS
Elb12bl are evidently zero because of the different symmetry of

-~''---------------------------------~
orbital.
VB .. IHR NO w;tIoo.tC I
;.h 19 W 2p was taken as -11.28 ev [R. S. Mulliken, J. chim. phys.
VB .. tIHR
$tructu1'e C
ah•• MO)
NO .. ah I ('I\~
46, 497 (1949)]'
FIG. 2. Energy levels calculated by both the valence-bond and 20 Lossing, Ingold, and Henderson, J. Chern. Phys. 22, 1489
the MO method: energy zero is 3W2p • (1954).
 LCAO-MOs OF THE ALLYL RADICAL 155

-9.00 ev would be good, whereas the valence-bond

calculation with lOmc homopolar resonance (IHR)
gives -9.77 ev with the unchanged effective nuclear
charge. 13
Figure 1 shows the electronic energies calculated
without CI as a function of the charge distribution and
Table II gives them for the various LCAO-MOs. Figure
2 shows the companson between the energy levels
g..,f<u")\OOO"'l'<l't-\ONO'-OO
calculated by both the valence-bond and the MO \0 0\00\0000 r-t')oo O\N'I""""I 0
ci-.:t1...or...:.nd......;......;f'f'5~o\~
method. It can also be seen here that two of the ion
ground state MOs and the "simply determined" MOs
,,, ,,,, ,,, ,
~N'I""""I'I""""I'I""""INN'I""""I'I""""IN'I""""I

are good approximations for the 2A2 state MOs, while

the energy levels without CI obtained by using the
LCAO-MOs with fairly different charge distributions
are certainly not good. As has been mentioned by many
authors, the effect of CI is very significant as can be
seen from Fig. 2. The results for the lower states ob-
tained by the valence-bond method without IHR were
better than those by the MO method without CI, as
in the CH radical. 21 ,22 In both the valence-bond and
the .1\10 treatment, it may be noted the energies ob-
tained become coincident with each other upon the
inclusion of IHR and CI, respectively (to the third
decimal in ev).
In the 1\10 calculation without CI, the two lowest
2Bl states, (lb 1)2(2b 1) and (lb1)(1a2)2, are so close to
each other that the effect of CI is very significant. The
2Bl state cannot, however, be calculated lower than the
4A 2 state by the interaction between these two configur-
ations using the same LCAO-MOs, although the result
with CI shows the lower location. For example, the
above calculated value by the 2A 2 state MOs is 3W 2p
-24.57 ev, while the 4A2 state and the lowest 2Bl state 8..,f<'<I'Ot-.-<t-O-'<I'O'-<"'l
lI')O\'I""""I~oOOJ:"--..lI')lI')lI)<o:::f'I~
with CI are 3W 2p -24.94 ev and 3W 2p -26.48 ev, d~r...:..o...od~""";~t")O\~
NN'I""""I'I""""I..-I~N'I""""I'I""""IN'I""""I
respectively. It is, however, noted that the energy of
the 2Bl state located lower than the 4A 2 state is obtained ""'"'"' +
as 3W 2p -25.79 ev by the interaction between two of u
the lowest 2Bl states by using the respective best MOs. u
The charge distribution of the lowest 2Bl state is ob-
tained as nearly homo polar (C a -Hl·OO5C b-O,OlOC c-Hl·OO5)
upon the inclusion of CI. This accords with the result
of the small energy depression by IHR in the valence-
bond method in Fig. 2 and with the triplet-like property
of this state.
The ground state 2A2 calculated by the BPN MOs
without CI (3W 2p -28.29 ev) is comparatively lower
than that of the 2A2 state MOs (3W 2p -27.15 ev), and
this can be explained by the consideration that the
BPN 1b 1 MOs are one kind of the alternant orbital,
partly taking into account the correlation energy. That
is, the energy of the 2A 2 state calculated by the 2A 2 state
MOs with partial CI of (1b 1)2(1a2) and (lb1)(1a2)(2b1)
(3W 2p -28.18 ev) is fairly close to that calculated by
21 In Table I of the paper on CH (reference 3), D,(ev) of
LCAO SCF without CI for the 2II state (2,70) should read (2,30).
22 Such a relation does not always exist. See, for example, the
paper on C 2H2 [Nakamura, Ohno, Kotani, and Hijikata, Progr.
Theoret. Phys. 8, 387 (1952)]'
156 lIRO HIGUCHI

the BPN MOs without CI. But the BPN total wave
functions are not eigenfunctions of the total spin oper-
ator S2, though eigenfunctions of the spin component
S.,23 and so are less suitable in simply investigating
the energy levels of radicals, since the forms and
the energies of MOs with a-spin function are not
equal to the corresponding ones with ~-spin func-
tion. The charge distribution obtained by the BPN
method may, however, be fairly satisfactory, since the
charge distribution of the ground state 2A2 with CI
(C a -Hl·O I6C b-O.032C c-Hl·016) is intermediate between that of
the 2A2 state MOs without CI (Ca-Hl·019Cb-o.038C-Hl·OI9)
and that of the BPN MOs (C a -Hl·OI2C b-O.024C c-Hl·OI2).
Finally, it may be noted that the BPN procedure (2)
is not always as easy as the procedure (1) irrespective
23 H. C. Longuet-Higgins and J. A. Pople, Proc. Phys. Soc.
(London) A68, 591 (1955).
of spin function; the case when the symmetry of the
singly occupied orbital differs from the others, as in the
ground state 2A2 (1b 1)2(1a2); and the case when the
occupied orbitals are each filled by two electrons with
paired spins except for the one electron deficient highest
orbital of some species, as in the 2Bl state (1b 1 )2(2b 1).24
ACKNOWLEDGMENTS
The author wishes to thank Professor S. Shida for
his kind advice and Mr. Y. Mori for helpful discussion.
24 Representing the symmetry of singly occupied orbital by P,
a calculation for the MOs without symmetry P can ordinarily be
carried out. The calculation for the MOs with symmetry P must,
in general, be performed by solving the determinantal equation
the order of which is twice that of the case without singly occupied
orbitals. In this case, however, the highest orbital with symmetry
P is singly occupied, then this procedure can be replaced with that
for the negative ion with full occupied highest orbital with sym-
metry P, in obtaining the P orbitals.

THE JOURNAL OF CHEMICAL PHYSICS VOLUME 26. NUMBER 1 JANUARY. 1957

Dielectric Behavior of Water at Microwave Frequencies

E. H. GRANT, T. J. BUCHANAN,* AND H. F. COOK
Department of Physics Applied to Medicine, The Middlesex Hospital Medical School, London, W.l, England
(Received March 1, 1956)

An extensive examination of the dielectric properties of water at microwave frequencies has been made.
The results have been fitted to various functions which have been derived from the Cole and Cole dispersion
equation and it is concluded that the relaxation process in water is characterized by a narrow spectrum of
relaxation times.

INTRODUCTION ment specifically suited to investigation of these liquids

T HE dielectric behavior of water at microwave
frequencies has been reported on several times in
recent years. Inadequate techniques and apparatus led,
in early work, to serious differences between the results
published by various workers. However, with the
development of more suitable apparatus and of better
techniques more recent workersl - 4 have reported results
in substantial agreement with one another. Most of
these authors show that their results satisfy the Debye
equations with a single relaxation time. Cook,3 how-
ever, pointed out that his results were slightly more
consistent with a narrow spectrum of relaxation times.
The purpose of this paper is to give further evidence in
support of this suggestion.
In this Department we have now had experience in
microwave measurements on water and aqueous sub-
stances dating back to 1948, and techniques of measure-
* Now at Marischal College, University of Aberdeen, Scotland.
1 Collie, Hasted, and Ritson, Proc. Phys. Soc. (London) 60, 145
(1948).
2 J. A. Saxton, Proc. Roy. Soc. (London) A213, 473 (1952).
3 H. F. Cook, Brit. J. App!. Phys. 3, 249 (1952).
4 ] . B. Hasted and S. H. M. El-Sabeh, Trans. Faraday Soc. 49,
Pt. 9, 1003 (1953).
have been developed. 6- 7
Lately, one of us (E.H.G.) has been mainly con-
cerned with extension of the previous work in this
Department to lower frequencies. The additional results
obtained, coupled with still further checks of our pre-
vious measurements, have now enabled us to revise
previous analyses of the results.
EXPERIMENTAL
Method
Three sets of microwave apparatus have been em-
ployed to obtain measurements at free-space wave-
lengths of 1.26 em, 3.23 em, 8.22 cm, 10.00 cm, 17.2 em,
and 52.0 cm.
At 1.26 em and 3.23 em separate wave-guide appa-
ratus for each wavelength was used, and is that de-
scribed by Buchanan." The principle of the method of
measurement is that the output of a signal generator
6 T. ]. Buchanan, Proc. Inst. Elec. Engrs. (London) 99, Pt.
III, 61 (1952).
6 T. ]. Buchanan and E. H. Grant, Brit. ]. App!. Phys. 6, 64
(1955).
7 E. H. Grant, Brit. ]. App!. Phys. 6, 181 (1955).