The Principle of Least Nuclear Motion

The Principle of Least Nuclear Motion
JACK HINE
Department of Chemistry, The Ohio State University,

Columbus, Ohio 43210, U.S.A.
1. Introduction . . 1
2. T h e Intersecting Morse Curve Approach . . 4
3. Prediction of t h e Magnitude of PLNM Effects . - 9
4. Free Radical Reactions . . 16
Hydrogen Transfer Reactions . 16
Radical Decomposition Reactions . . 25
5. Multicenter Reactions . . 33
6. Polar Reactions . 36
Proton Transfers from Carbon . . 37
Proton Transfers t o Resonance-Stabilized Carbanions . 48
Lewis Acid-Base Reactions . . 52
Elimination Reactions . . 54
7. Concerted a n d Stepwise Reaction Mechanisms . . 56
8. Conclusion . . 57
Acknowledgement . . 57
References . . 57
1. INTRODUCTION
Organic chemists ordinarily think about structural effects on

reactivity in terms of transition states for the various individual steps
of a reaction. The structure of the transition state is always
intermediate, in at least some respects, between that of the
reactant(s) and that of the product(s). For this reason it is often
useful to express the properties of transition states in terms of those
of the reactants and products (cf. Lefflcr, 1953). However, the
reactivity depends on the free enerLy o f the transition state, which is
1
2 JACK HINE
never intermediate between that of the reactant(s) and product(s).

Nevertheless, one of the most commonly used generalizations con-
cerning structural effects on reactivity is that for sufficiently' closely
related reactions 6~ AG* [the change in activation energy brought
about by a change in reactant structure (Leffler and Grunwald,
1963)J has the same algebraic sign as 6~ AGO. If this generalization is
to apply t o the reverse as well as the forward reaction &RAG*
must be between zero and ~ R A G O This . tendency of structural
changes to change rate constants in the same direction that they
change equilibrium constants has been called the product stability
principle (Hine, 1971). It was widely recognized by chemists long
before it was applied quantitatively t o proton-transfer reactions in
the form of the Br4nsted equation (Br4nsted and Pedersen, 1924) or
rationalized by the theoretical treatments of Evans and Polanyi
(1936, 1938) and others (Horiuti and Polanyi, 1935; Bell, 1936,
1941). It was also recognized t o be a rule t o which there are many
exceptions, including some cases of fairly closely related reactions.
Many of the exceptions may be explained in terms of (1) the
principle of least motion or (2) imperfect synchronization of the
various changes that take place during the reaction (which in severe
cases results in non-monotonic changes in structural or other features
with progress along the reaction coordinate).
The principle of least motion, in the words of Rice and Teller
(1938), states that those elementary reactions will be favored that
involve the least change in atomic position and electronic configura-
tion. This generalization is closely related to the principk o f
minimum structure change (Hiickel, 1934; Wkeland, 1960), a name
given to the usually implicit basis for using the results of chemical
reactions to deduce the structures of reactants or products. Thus the
germ of the principle dates back to the early days of the science of
chemistry. Like the product stability principle, thc principle of least
motion deals with reactivity in terms of the reactant(s) and pro-
duct(s) of the reaction. Rice and Teller's words suggest a division of
the principle into two parts, the principle of least nuclear motion and
the principle of least changc in electronic configuration. The latter
concerns the slowness of reactions involving changes in multiplicity,
changes in orbital symmetry, ctc. Current interest in the Woodward-
Hoffmann (1970) rules has led to many articles and reviews that deal
' This word makes the grnrralization infallible.

LEAST NUCLEAR MOTION 3
with the principle of least change in electronic configuration.

Therefore the present treatment will be devoted largely to the
principle of least nuclear motion (PLNM). However, the product
stability principle and asynchronization effects will often be necess-
ary parts of such a treatment. That is, we shall act 8s though all
structural effects on reactivity can be divided into various compon-
ents. When product stability effects have been accounted for, the
residue will contain the PLNM effects, but it will also contain
asynchronization effects and effects of differences in changes in
electronic configuration.
Students of elementary organic chemistry are commonly told that
the product of the monobromination of an alkane molecule by a
free-radical mechanism depends on which hydrogen atom is removed
from the alkane by an attacking bromine atom. When the alkane is
n-butane [(eqn (l)] the secondary hydrogen atoms are said t o be
H ~ CH3CHCH2CH3+ HBr
C H ~ C H ~ C H Z +C Br- + (1:
removed more rapidly than the primary hydrogen atoms because the
resultant secondary radicals are more stable than primary radicals.
Thus the product stability principle is used, although ordinarily only
implicitly, to rationalize the relative rate of the two reactions.
There is a third method of removal of hydrogen by bromine atoms
that should be even faster than either of the two already mentioned
if the product stability principle were infallible. If the removal of
secondary hydrogen were accompanied by migration of the other
hydrogen atom on the same carbon and of the methyl group on an
adjacent carbon atom, as shown in (2), the product will be a t-butyl
Rr
I
H
---+ CH~-C-CH~
I
CH3
H
radicaI, which is more stable than s-butyl or n-butyl radicals. The fact
that elementary students so rarely’ raise this possibility suggests that
Never, in t h e author’s experience.

4 JACK HINE
they already have the PLNM, or something like it, in their intuition.
The difference in the amounts of change in atomic positions between
the reactions in eqns (1) and (2) is thus so drastic that no chemist
would be likely to overlook it. There are other cases, however, in
which rather substantial differences in the amounts of change in
atomic positions are fairly easily overlooked. Before including such
cases in a survey of a wide variety of the organic reactions to which
the PLNM may be usefully applied, we shall discuss the principle
qualitatively in terms o f what may be called the intersecting Morse
curve approach.
2. THE INTERSECTING MORSE CURVE APPROACH
What may be called the intersecting Morse curve approach has

given valuable insight into the effect of structure on reactivity. It is
conveniently applied to atom-transfer reactions, and we shall apply it
here t o hydrogen-transfer reactions (3). Consider the reaction to
consist of bringing the reactants together with the three atoms
XH + Ye -+ X. + HY (3)
collinear until X and Y are separated by the distance that they will be
separated in the transition state. The hydrogen atom is then moved
from its reactant position at its covalent bonding distance from X t o
its product position at its covalent bonding distance from Y. Then
the products separate. In Fig. 1, a plot of energy vs. the position of
the hydrogen atom, curve HX describes the energy of the H-X bond.
A minimum at the covalent bond distance, the energy rises sharply as
the bond is compressed; it rises and then approaches a limit as the
bond is stretched. At the limit, indicated by the dashed line, the
energy exceeds that at the zero-point vibrational level by the bond
dissociation energy D,(HX). For curve HY, which describes the
energy of the H-Y bond, motion of hydrogen to the right is bond
compression and to the left is stretching. Since the dashed line
corresponds t o the energy of the three dissociated atoms, H - , X.,
and Y . , the zero-point vibrational level of the HY curve lies below
the dashed line by D o(HY).
To a first approximation the energetic requirements for reaction
may be described by starting with the hydrogen at the zero-point
I I
X Y
Position of h y d r o g e n
Figure 1. Intersecting Morse curve representation of the reaction of HX with Y - to give
HY + X..
level of the HX curve and moving t o the right along this curve until it
intersects the HY curve. After this point the system is better
approximated if it is considered to consist of X. + HY, and so we
descend along the HY curve to its zero-point level, the overaI1 energy
of reaction being E,. In a second approximation we allow for
resonance stabilization, which should be a maximum near the point
of intersection of the two curves, where the two valence-bond
structures contribute equally. The dotted line lies below the lower of
the two curves HX and HY by an amount we shall call the resonance
energy. In this approximation the activation energy is that labelled
E* in the figure. In a third approximation, not illustrated in the
figure, we could allow for the energy required t o bring HX and Y *
together with the X-Y distance characteristic of the transition state.
G JACK HINE
>
e
c
W
X Y
Position of h y d r o g e n
Figure 2. Reactions of HX with Y1. and Yz., where HY2 is more stable than HYI.
Let us now discuss the effect of structure on reactivity by

considering the reactions of HX with a number of different Y’s,
Y , *, Y, .,
Y3 *, etc. To do so we would be interested in only that
part of Fig. 1 lying between hydrogen positions x and y. Figure 2
shows an enlarged version of this region. Since we shall discuss
largely relative activation energies n o indications of zero-point
energies or of resonance-energy corrections are shown. If these
energies are the same for each of the different reactions being
compared their neglect is permissible. We shall also assume that the
X-Y distances, and hence the energies required t o bring HX and Y.
to the appropriate distance from each other, are the same in each
reaction. In Fig. 2, Yz is taken t o be similar t o Y1 except that it
a
forms a stronger bond t o hydrogen, i.e., Do (HY,) >Do (HY, ).

Therefore the minimum of the HY, curve lies below that of the HY,
curve by the amount Do(HY2) - D o ( H Y , ) . Since the HY1 and HY2
curves must approach each other at large H-Y distances they have
been shown as tending t o do so in the figure. However, as long as
they do not come together until after they cross the HX curve, the
activation energy for reaction of HX with Y2 will be lower than
that for reaction with Y * ; that is, the product-stability principle will
be applicable.
When X * and Y are not atoms but polyatomic radicals, combina-
tion with a hydrogen atom may involve important changes in
geometry other than just the length of the H-X or H-Y bond. For
such cases we should redefine the abscissa of Fig. 1 and 2 t o refer to
changes in geometry within X andlor Y as well as change in the
position of hydrogen. At position x the geometry is that which is
optimum for the system HX + Y. and at y it is optimum for
X. + HY. Consider, for example, two reactions in which X is an
atom. One is with Y3 *, a radical whose internal geometry in Y3 * is
the same as it is in HY,. The second is with Y4., whose internal
geometry in HY4 differs significantly from what it is in Y,..
-
Otherwise, Y, * and Y4 are quite similar, the dissociation energies
Do(HY3) and Do(HY4) being identical. The curve in Fig. 3 having a
minimum at x and referring t o the energy of HX + Y, rises on going
to the right only by the amount of energy required to stretch the
.,
H-X bond. The curve for the enercgy of HX + Y, however, must
represent the energy required to stretch the H-X bond plus that
required to distort Y4 * from its optimum geometry toward the
geometry that is optimum for the Y4 part of HY4. This factor alone
makes the intersection of the HX + Y. curve and the X. + HY curve
higher for Y, than for Y, and would therefore make the Y4 reaction
slower, as expected from the PLNM. A second factor wilI cause the
X. + HY4 curve to lie above the X. + HY, curve when the H-Y
bond is partly but not completely broken. For example, if we
compare propene, where Y4 * is an allyl radical, with an HY, having
the same bond dissociation energy, we see that the resonance
stabilization of the allyl radical is necessary t o make Do(HY4) as
small as D,(HY,). When the H-Y, bond is partly broken, delocaliza-
tion of the electron left on Y, is diminished only by the fact that
this electron is still interacting with the hydrogen nucleus and
another electron near the hydrogen nucleus. In the case of propene,
8 JACK HINE
X Y
Geometric coordincte
Figure 3. Reaction of HXwith Y , * where there is no accompanying change in the internal
geometry of Y3 and the reaction with Y;1*in which there is a change in the internal geometry
of Y4'.
however, delocalization of this electron is also diminished by geo-
metric factors. The carbon from which the hydrogen atom is being
removed has not yet become sp2 -hybridized; the bond that attaches
this carbon to the adjacent carbon has not yet become short enough
to permit more efficient overlap with the n-electrons of the double
bond, etc. A third factor is the smaller amount of resonance
stabilization that would be expected in the transition state for the
Y 4 reaction (cf. the dotted line in Fig, 3) than in the transition state
for the Y , reaction (cf. the dashed line). This follows from the fact
that at a given point along the geometric coordinate the geometry of
the Y4 system differs more from the optimum for HX + Y . and from
the optimum for X . + HY, (The differences in the position of the
hydrogen atom are the same but there are differences in the
geometry of Y in the case of Y 4 . )
3 . PREDICTION OF THE MAGNITUDE OF PLNM EFFECTS
Although most reactions may be thought of as being slowed by

PLNM effects, in many cases the differences between the magnitudes
of the effect expected in two processes being compared are large and
qualitatively obvious. The formation of s-butyl radicals by removal
of a hydrogen atom from n-butane should be greatly favored by
PLNM effects, relative t o the formation of a t-butyl radical. In other
cases, however, it is less obvious which reaction will have the larger
PLNM effect, or it may be obvious that the effect will be larger in one
case but not obvious whether the difference will be significant. No
method has been devised to obtain precise numerical measures of
I (4)
PLNM, and indeed the concept, like many others that are useful in
organic chemistry, has not been defined precisely. (It is not clear that
a definition that is both precise and broadly useful is possible.)
Nevertheless, consideration of the problem by several workers has
yielded methods for obtaining numbers that are useful for some
purposes.
One study dealt with the formation and reactions of resonance-
stabilized species (Hine, 1966a). The formation of an allylic species
for example [eqn (4)] is accompanied by the shortening of the
carbon-carbon single bond and the lengthening of the carbon-carbon
double bond as each becomes a bond and a half; that is, the bond
number (Pauling, 1960) is 1.5 in the allylic species. In addition, as
the methyl group loses a hydrogen atom the other two hydrogen
atoms move into the plane defined by the three carbon atoms. Is it
10 JACK HINE
plausible that these changes in molecular geometry are large enough

to have a significant effect on the reaction rate? How does one
compare the PLNM effect with that to be expected in some other
reaction?
In the calculations that were made the treatment described in the
preceding section was simplified by changing the intersecting Morse
curves to intersecting parabolas. The estimated PLNM effects were
quite small when the transition state lay very near the reactants or
products. According to a Hooke's law approximation, the effect of
changing bond lengths and bond angles should be proportional to the
square of the magnitude of the change. The reaction treated in detail
was the protonation of cyclohexadienyl anions t o give 1,3- or
1,4-~yclohexadiene,but the treatment is rather directly relevant to

the reaction of any pentadienyl cation, anion, or radical at its two
types of reactive positions [eqn ( 5 ) ] . In terms of Fig. 3 , the
calculations allowed only for the difference between curve HX + Y , *
and curve HX + Y4 '-not for the difference between curves X. +
HY, and X. + HY, and not for differences in resonance stabilization
of the transition state. Also, allowance was made for stretching and
compressing bonds but not for changes in bond angles and planarity.
These calculations gave a maximum PLNM effect, when the
transition state lav about two-thirds of the way along the reaction
coordinate, of 1-2 kcal mol-', with the effect being smaller by as
much as 1 kcal mol-' for reaction at the middle carbon atom of the
7r system. Allowance for the neglected factors would give a larger
estimated effect and one whose maximum would occur when the
transition state lay nearer halfway along the reaction coordinate.
The approximations used in the calculations described in the
preceding paragraph restrict us t o the concltision that PLNM effects
on the reactions ( 5 ) of pentadienyl species may increase the activa-

tion energies by as much as several kcal mol-' with reaction at the
middle carbon atom being the pathway favored by the PLNM. To get
a better idea as t o the magnitude of PLNM effects we should
examine experimental data in cases where we have information on
the importance of product stability effects. Since the examination of
such data in later sections of this article appears t o show substantial
PLNM effects, it seems worthwhile t o have a method (Hine, 1966a)
for predicting the magnitude of PLNM effects that is quicker and
easier (but less reliable) than the calculations that have been
described. Changes in bond lengths being very nearly proportional to
changes in bond numbers over the range in which changes are usually
observed, the sum of the squares of the changes in bond numbers will
be proportional to the sum of the squares of the changes in bond
lengths. To the extent t o which all bonds have the same stretching
force constants, this sum is a measure of how far the curve of the
type of HX + Y , * lies above HX + Y3 in Fig. 3. It is therefore a
crude measure of that part of the PLNM effect arising from changes
in bond lengths. To illustrate use of this method for a reaction of the
type shown in ( 5 ) , let us assign bond numbers to the pentadienyl
radicaI on the basis of a simple Pauling superposition. Equal contri-
butions of valence bond structures [ 11- [3] give hybrid radical [4],
-
.CH2 CH2 CH2
I II II
CH CH CH
II I I
CH -CH CH
I I I1
CH CH CH
/I ll I
CH2 CH2 *CH2
[11 [21 [31
in which the end carbon-carbon bonds are 13 bonds and the middle
ones If bonds. Since this pattern of longer middle bonds and
shorter end bonds is nearer that of a 1,4-pentadiene derivative than
that of a 1,3-pentadiene derivative, reaction at the middle of the 1~
system is favored by the PLNM. Such reaction involves a total of
12 JACK HINE
(2 - 1$)2+ (14 - 1)2 + (14 -1)2 + (2 - 18)’ or $ in the squares of

the changes in bond numbers. Reaction at the end of the rr system to
give a 1,3-pentadiene derivative involves a total of (18 - 1)*+
( 2 - 14)’ + ( I f - 1)’ + ( 2 - 1%)’or in the squares of the changes
in bond numbers. Comparision with the more detailed calculations
on the same type of system and with various experimental observa-
tions shows that sums as large as $ or I9O can be expected to be
significant, as can differences as large as $O--$.
Figure 4. Changes in geometry in the dehydrochlorination of vinyl chloride (-) to

give acetylene via a trans elimination ( - - ) or a cis elimination (-----).
The preceding quick method of assessing PLNM effects will be

applied to a number of specific reactions in subsequent sections.
Tee (1969) generalized a method of predicting the relative magni-
tudes of PLNM effects in which the minimum value of the sum of the
squares of the nuclear displacements brought about by the reaction is
calculated. The method had been applied previously in a less general
form only to certain aspects of the stereochemistry of elimination
reactions (Hine, 196613). Figure 4 illustrates the application of this
method to the E2 dehydrochlorination of vinyl chloride to give
acetylene. In both cis- and trans-elimination a base removes one
hydrogen as a proton and the chlorine comes off as an anion; any
PLNM effects on these processes are assumed t o be the same in the
two different reactions. The remaining two hydrogen atoms become
collinear with the carbon atoms in the reaction product. For both
the cis- and trans-eliminations the position of the atoms in the
product is taken t o be that which involves the smallest sum of the
LEAST NI1C:LEAR MOTION 13
squares of the resulting displacements of the nuclei. The dashed line

in Fig. 4 was calculated to contain the four atoms of acetylene
resulting from a trans-elimination and the dotted line to contain thc
atoms of acetylene resulting from a cis elimination. The truns-
elimination moves the atoms by 0.320 A2 and the cis-elimination
moves them by 1.237 A2 (Tee, 1969). According to this approach
the PLNM will favor trans-elimination.
The cis- and trans-dehydrochlorinations of vinyl chloride represent
a relatively attractive pair of reactions to which the PLNM may be
applied because the product stabilities are probably so nearly the
same when an anionic base is used. The ultimate products, in fact,
are the same for cis- and for trans-elimination, regardless of the
nature of the base. It is the immediate product, however, whose
stability is most reasonably associated with the reaction rate and [ 51 ,
the immediate product of cis-elimination, has the BH molecule
nearer the chloride ion than does [ 6 ] , the immediate product of
BH C1-
H-CZC-H
I c1-
H-C=C-H
RH
[61
trans-elimination. The difference between the interaction of BH with
chloride ion in [5] and in [6] seems unlikely to be very large when
BH is electrically neutral. The rates will be affected by interactions
between the partly negative 3 and the partly negative chloride ion in
the transition state, but this is not a product stability effect. It is an
asynchronization effect, arising from the nonmonotonic change in
electrostatic interactions between B and chlorine that occurs on
moving along the reaction coordinate; these interactions are initially
small because the chlorine atom bears little charge, they reach a
maximum when both B and chlorine are about equally charged, and
they then become small again as B become relatively uncharged.
In assessing the PLNM effect by calculating the sum of the squares
of the nuclear displacements care should be taken to keep the
reactions being compared closely analogous. For example, if the
14 JACK HINE
dehydrochlorination of 2-chloropropene to give propyne were

compared with the dehydrochlorination of vinyl chloride, a larger
sum of the squares o f the nuclear displacements would be obtained
for propyne formation by a given type of elimination process. This is
because a larger number of atoms are moving as the methyl carbon
atom moves into collinearity with the other two carbon atoms than
when collinearity is achieved in acetylene formation. However, this
larger number of atoms is still being moved by the bending of one
bond, and it is the bond bending upon which energy is being
expended.
In three subsequent papers (Tee and Yates, 1972; Tee et al., 1974;
Altmann et al., 1976) the number of reactions to which Tee applied
his method originally (Tee, 1969) has been considerably expanded.
The results of a number of these calculations will be described when
we discuss the type of reaction in question.
Ehrenson (1974a, b) has given the most fundamental treatment of
the PLNM that has appeared t o date. He pointed out that the
minimum sum o f the squares of the nuclear displacements becomes a
more reasonable measure of the magnitude of the PLNM effect if the
individual displacements are appropriately weighted and he derived
equations for calculating the minimum sum of the weighted squares
of the nuclear displacements. When properly weighted by masses and
force constants the minimum sum of the squares of the displace-
ments is proportional to the energy required for the given displace-
ments as calculated by the harmonic oscillator approximation. Even
without weighting it turns out that PLNM-favored paths usually
correspond to vibrational modes requiring less energy. An example of
this fact is found in the dehydrochlorination of vinyl chloride. In
simple terms we may say that either cis- or trans-dehydrochlorination
would produce a distorted acetylene molecule if the four atoms of the
acetylene molecule did not move from the positions they had in the
vinyl chloride reactant. Figure 4 shows that, in terms of the
minimum sum of the squares of the nuclear displacements, the
acetylene molecule produced by trans-dehydrochlorination without
additional changes in geometry is less distorted than the one
produced by cis-dehydrochlorination. A simplistic counter-argument
may be given that in each distorted acetylene there are two C-C-H
bond angles that are bent by 60" and that therefore the amount of
distortion is the same in each case. This argument, however, ignores
interactions between the two angle bends in a given molecule. This is

automatically taken care of in a vibrational normal coordinate
analysis. The bending vibrations of acetylene are of two symmetry
types, xg and 7ru. Comparison with Fig. 4 shows that it is the rg
vibration that tends to give the type of distorted acetylene being

formed in the trans-dehydrochlorination of vinyl chloride and the xu
vibration that gives the distorted acetylene being formed in cis-
dehydrochlorination. The smaller frequency of the xg vibration
(612 cm-’ compared to 730 cm-’ for the xu vibration) or, more
relevantly, its smaller force constant (0.122 mdyn 8-l compared t o
0 . 2 7 0 m d y n 8 - ’ for ru),shows that the rg vibration is much the
“softer” of the two types. To the extent to which these force
constants, based on vibrations of relatively small amplitude, give us
reliable information about the relative amounts of energy required
for the bending (presumably by 30 k 30’) of the C-C-H bonds in the
transition states for acetylene formation, they tell us that the
distorted acetylene being formed in trans-dehydrochlorination is less
distorted energetically than that being formed in cis-dehydro-
chlorination.
Ehrenson has also pointed out that it would ordinarily be more
rigorous to discuss the PLNM in terms of motion t o or from the
transition state. Even in cases where it is difficult to estimate the
energy content of the transition state we may have useful informa-
tion as to its geometry. To the extent to which the transition state
geometry is a linear combination of the geometries of the reactant(s)
and product(s), motion from the reactant toward the transition state
is entirely equivalent to motion toward the product. However, when
the geometric changes that accompany reaction are imperfectly
correlated with each other, and especially when there are important
changes in geometry that d o not even occur monotonically, motion
toward the transition state may differ markedly from motion toward
the product. It is in this type of case that discussion in terms of
motion toward the transition state is most clearly preferable to
16 JACK HINE
discussion in terms of motion toward the products. Such a treatment

in terms of the transition state may not be feasible when there is too
much ignorance about the nature of the transition state. Treatment
in terms of the reaction product, a more familiar type of species than
a transition state, is also usually easier. For these reasons and the fact
that most commonly the two approaches give the same result, most
of the PLNM treatments in this chapter will be in terms of motion
toward the product of the elementary reaction in question.
Ehrenson (1974b) applied his methods in some detail t o reactivity
and stereochemistry in the abstraction and displacement reactions of
protons and hydrogen atoms with methane. He has also treated
unimolecular decomposition reactions, showing that the relative dis-
placement of the centers of gravity of the separating fragments is
usually a better choice of the critical coordinate than is the degree of
stretching of the bond being broken (Ehrenson, 1976).
4. FREE RADICAL REACTIONS
Hydrogen Trunsfer Reactions
In order to explain structural effects on reactivity convincingly in

terms of the PLNM we must assess the product stability effects that
are also present. Product stability effects are best seen in log-log plots
of rate vs. equilibrium constants. Such simple generalizations as the
product stability principle are not very reliable unless the reactions
being compared are quite similar t o each other. Therefore in atom
transfer reactions we shall compare only those in which the atom
being transferred is the same. W e shall consider hydrogen transfers
here because more reliable data are available than for any other
type of atom transfer. To help make the reactions similar t o each
other let us first consider hydrogen transfers t o a methyl radical (6).
XH + Me- --f X- + MeH (6)
Since not as many free energies are available as desired, log k will be
plotted against A H o rather than log K. (The free energies that are
available suggest no changes in any of the quaIitative conclusions that
we shall reach.) Values of AHo were calculated from a recent
collection (Hine, 1975) of representative bond dissociation energies
and values for radicals (not including those data listed as
approximate). Rate constants at 200° (a temperature in or near the
ranges over which the various reactions were studied) were obtained
from the collections of E and log A values3 by Trotman-Dickenson
et al. (Trotman-Dickenson and Milne, 1967; Ratajczak and Trotman-
Dickenson, 1969). For all the HX's covered in both collections of
2-
Figure 5. Plot of log k (k in M-' s-I per hydrogen atom) for reactions of variousHX9swith
methyl radicals in the gas phase at ZOO0 to give methane and X. vs. A@ for the reaction.
The solid circles are for HX's that are saturated hydrocarbons and the X's, in order of
increasing log k, are methyl, cyclopropyl, ethyl, n-propyl and isobutyl (coincident), cyclo-
butyl, cyclohexyl, cyclopentyl, isopropyl, cycloheptyl, and t-butyl. The open circles refer to
other HX's in which the hydrogen atom is attached to carbon, and, like the triangular
points, arc labeled with the appropriate X's.
data log k is plotted against A@ in Fig. 5 (except for HI, whose log
k value of 8.4 and @ value of -32.7 kcal mol-' would give a
point more than 10 kcal mol-' beyond the left-hand boundary of
the figure).
For X = i s o p r o p y l , data on (CH,)*CDz were combined with the assumption that
k H / k D is the same as the value t h a t may be calculated from data listed for C2H6 and
C2D6. For allyl, a-methylallyl, and benzy! it was assumed that the attacking methvl
radical removes hydrogen atoms only from the allylic or bcnzylic positions of propene,
I-butene, and toluene. This assumption tends to make the PLNM e f f e c t appear smaller than
it actually is.
18 JACK HINE
The plot shows that the reactions covered are still not closely
enough related t o give a good rate-equilibrium correlation. However,
restriction of the HX’s t o saturated hydrocarbons gives the 10 points
(representing 11 compounds, since the points for X = n-propyl and
X = isobutyl coincide) shown as solid circles, which fit a straight line
with a correlation coefficient of 0.98. The largest deviations from
this line are by some of the triangular points, which represent HX’s
in which the hydrogen atom is not attached t o carbon. It seems clear
that polar effects are contributing t o some of these deviations. The
reaction rate would be expected t o be increased by a polar inter-
action between Me. and X. in the transition state when X - is an
electron-withdrawing radical (i.e., when X- is relatively stable)
relative to Me.. (Note that this results from the fact that the
interaction between Me. and X. must change non-monotonically on
passing from the reactants, where there is no such interaction,
through the transition state, where there is interaction, t o the
products, where there is again no interaction.) For example, methyl
radicals abstract hydrogen atoms from hydrogen chloride about 3000
times as fast as from cyclopentane (on a per hydrogen basis) in spite
of the fact that the latter reaction is about 7.3 kcal mol-‘ more
exothermic. Trifluoromethyl radicals, which should donate electrons
much more weakly to the strongly electron-withdrawing chlorine
atoms in the transition state, abstract hydrogen atoms from the two
sources at essentially the same rate (Gray et al., 1971). However, not
all the deviations seem explicable in terms of polar effects. There
should be positive deviations in log 12 when X is M e 0 and when it is
MeNH, and there are; however, the deviation by M e 0 should be
larger, yet actually the two deviations are of essentially the same size.
All in all, it should probably be admitted that, for reasons that are
only partly understood, when the nature of the atom to which the
hydrogen atom is attached in HX is varied large deviations from a
linear rate-equilibrium correlation may result.
Having decided not to consider the triangular points in Fig. 5
further, we may wonder about the HX’s in which the hydrogen is
attached to s p 2 -hydridized carbon. It is not clear whether this results
in large deviations from the line in the figure or not. Of the five
points referring to such reactions, two refer t o the formation of
acetyl and formyl radicals (from the corresponding aldehydes). These
two points lie somewhat below the line, but they deviate only
slightly more than do some of the points for saturated hydrocarbons.

A third point, for formation of the carbomethoxy radical, lies on the
line. The points for vinyl and phenyl radical formation lie well above
the line, but it is not clear how reliable the experimental data are in
these cases. The reaction of methyl radicals with benzene t o give
methane directly is particularly liable to complications resulting from
addition of the radical to the ring t o give a methylcyclohexadienyl
radical from which hydrogen can be very easily abstracted by a
second methyl radical. There have been two determinations of the
rate constant at various temperatures for the reverse reaction,
abstraction of a hydrogen atom from methane by phenyl radicals
(Trotman-Dickenson and Milne, 1967). When combined with the
equilibrium constant calculated from the thermodynamic properties
of the species involved, both of these give rate constants for the
formation of phenyl radicals from methyl radicals and benzene that
are smaller, one of them by 15-fold, than the rate constant upon
which the point for phenyl in Fig. 5 is based. The rate constant for
the formation of vinyI radicals was obtained in a study of the
reaction of methyl radicals with ethylene, in which addition t o the
ethylene and polymerization also appeared to be taking place
(Trotman-Dickenson and Steacie, 1951). These side reactions, which
led the authors to conclude that “. . . the results with ethylene are
probably less accurate than those for the other compounds,” would
give products that would contain much more labile hydrogen atoms
than does ethylene. For this reason the rate constant is probably too
large.
Narrowing our attention to the removal of hydrogen from
sp3 -hydridized carbon, we see that the halomethanes appear t o be
too reactive t o fit the linear free energy relationship described by the
saturated hydrocarbons. Polar effects explain the direction of these
deviations but not their relative magnitudes. It is plausible that the
points for fluoromethanes should deviate less than those for chloro-
methanes because a-chloro substituents stabilize carbanions more
than do a-fluoro substituents. However, polar effects should give
smaller deviations for one or two a-chloro substituents than for
three. Unfortunately, it is not clear how reliable some of the relevant
data are. The rate constants for the fluoromethanes that are plotted
appear to be reliable. They are based on data reported in 1965 and
are smaller, by as much as 70-fold, than values that may be obtained
20 JACK HINE
from some reports in 1952 and 1955. Since the rate constants for
bromo- and chloro-methanes are based on these earlier reports, we
may wonder if some of them may not also be too large. The polar
effects of substituents could be discussed in terms of the overall
balance of electron-donor and -withdrawer ability as measured by
Hammett para-substituent constants. Since the values for hydroxy
and alkoxy substituents are negative, that for fluorine is slightly
positive, and those for chlorine and bromine more strongly positive,
it may not be surprising that the small deviations from the line in
Fig. 5 seen with hydroxy and alkoxy substituents are in the opposite
direction from those for haIogen substituents. Speaking alternatively,
in terms of carbanion stabilizing ability it is noteworthy that in some
cases an a-alkoxy substituent can decrease the rate of carbanion
formation, relative t o a-hydrogen (Hine et al., 1967).
The only three reactions that have not yet been discussed are
those in which the allyl, cv-methylallyl, and benzyl radicals are
formed. These might be expected to give better agreement with the
line in Fig. 5 defined by the data on saturated hydrocarbons than
any of the other classes of reactions we have discussed. The reactants
in these reactions are hydrocarbons, which are less polar than are the
reactants in any of the other classes of reactions. The Hammett
substituent constant for p-phenyl is, and that for p-vinyl would be
expected to be, much smaller in absolute magnitude than p-alkoxy or
p-halogen substituent constants. Hence polar effects should be
negligible unless one uses carbanion-stabilizing ability as a measure,
in which case a positive polar effect would be expected. Experi-
mentally, however, the three reactions are too slow, by 30- t o
250-fold, to fit the saturated hydrocarbon line. Empirically, these
three reactions are the only ones besides the 11 reference reactions in
which hydrogen is being removed from an sp3-hybridized carbon
atom that has only carbon atoms and other hydrogen atoms attached
to it. However, instead of giving the closest agreement, these
reactions give some of the largest deviations from the line. The best
explanation for these deviations may be the large change in the
structure of X that accompanies the removal of hydrogen from HX in
these cases. Not only does the hybridization at the carbon atom from
which hydrogen is removed change from s p 3 t o s p 2 , there are also
major changes in bond number and hence in bond length as the
resonance-stabilized allylic and benzyl radicals are formed. It is true
that there should also be a PLNbI effect on the formation o f at least

some of the saturated hydrocarbon radicals. There seems to be good
evidence (Dobbs ct al., 1971, 1972; Krusic and Bingham, 1976) that
replacing the hydrogens of a methyl radical by alkyl groups (or other
resonance-electron-donating groups) tends t o change the geometry
from planar, as in methyl radicals (Herzberg, 1961; Fessenden,
1967; Tan et al., 1972), toward tetrahedral, as in t-butyl radicals
(Wood et al., 1972). Since moving to the left among the solid circles
in Fig. 5 tends to correspond to replacing the a-hydrogen atoms on
the radicals being formed by alkyl substituents, the rates are
expected to be depressed by a PLNM effect t o an increasing extent as
we move t o the right among these points. That is, if all the radicals
being formed were tetrahedral we would expect the points further to
the right t o be higher so that the absolute value of the slope of the
least-squares line would be less. This would decrease the magnitude
of the estimated PLNM effect for the formation of allyl,
a-methylallyl, and benzyl radicals (if we are willing to accept the
PLNM as a reason for changing the slope of the line). However, this
change in the estimated PLNM effect should be small because the
PLNM effect for the saturated hydrocarbon radicals should be small.
Methyl radicals, which are presumably the most strongly coplanar of
the group, have only a rather weak preference for planarity. In fact,
they are said to have the smallest out-of-plane force constant of any
known gaseous D 3 h molecular species (Tan et al., 1972). Further-
more, the carbon- hydrogen bond lengths in methyl radicals
differ from those in methane by only about 0.015 A (Herzberg,
1961), which is far smaller than the changes in carbon-carbon
bond lengths that must accompany the formation of allyl or benzyl
radicals.
Additional, but rather weak, evidence that the line in Fig. 5 has
been steepened only slightly by PLNM effects comes from the
absolute magnitude of the slope, which is 0.51 when expressed in
dimensionless form (by multiplying Alog tz values by 2.3RT to
transform them t o AAG* values in kcal mol-'). The intersecting
Morse curve approach may be approximated by use of two inter-
secting congruent parabolas with the difference in the relative heights
of their minima being proportional t o AGO for the reaction. Express-
ing AG' as a function of AGO shows that dAG'/dAGo is 0.5 when
AGO is zero (as for the point for X = M e in Fig. 5 ) and slowly
22 JACK HINE
decreases as ACo becomes more negative (as it does for the other
solid circles in the figure).
The 30- to 250-fold deviations in k from the line in Fig. 5 seen for
the rate constants for the formation of allylic and benzyl radicals
correspond to AAGt values of 3.2-5.2 kcal mol-' . These effects
accompany changes in the sum of the squares of the bond numbers
of 0.50 for thc allylic cases and 0.58 for the benzyl case. In an
estimate of the effect of bond-distance changes on the rates of
reaction o f cyclohexadienyl anions referred to in an earlier section
(Hine, 1966a), AAG' values as large as 0.9 kcal mol-' were obtained
for a difference of 0.66 in the sum of the squares of the changes in
bond numbers when the transition state lay at the optimum position
along the reaction coordinate. The total estimated PLNM effect
could be considerably larger than this if allowance were made for
changes in bond angles and for loss of resonance stabilization of the
transition state. Hence the order of magnitude of the observed
effects on the reaction rate (AAG' values) is believed to be plausible
for a PLNM effect.
It might be suggested that attachment of s p 2 -hybridized carbon to
the sp3-hybridized carbon from which hydrogen is removed, as has
been done in the reactions in which ally1 and benzyl radicals are
formed, can give deviations from the linear free energy relationship
that holds for the formation of saturated hydrocarbon radicals
because of a decrease in the carbon-hydrogen stretching force
constant. However, this possibility is disposed of by the fact that n o
such decreases are observed. The methyl carbon- hydrogen stretching
frequencies in isobutene, for example, are not significantly different
from those in ethane, propane, or the butanes (Shimanouchi, 1972).
There may be small PLNM effects operating in some of the
non-hydrocarbon reactions covered by Fig. 5. The fluoromethyl,
difluoromethyl, and trifluoromethyl radicals, for example, are non-
planar to an extent that increases with the number of fluorine atoms
(Fessenden and Schuler, 1965). Since the points for the formation of
the various fluoromethyl radicals lie near the methyl end of
the line describing the rates of formation of saturated hydrocarbon
radicals, and since this end of the line refers t o the formation of
planar radicals, it is likely that the positive deviations arise in part
from the smaller change in geometry that accompanies formation of
the fluorinated methyl radicals. The formation of acetyl and formyl
radicals, on the other hand, may be slowed somewhat by PLNM

effects. Microwave measurements show that the carbon-oxygen
bond in a formyl radical is only 1.17 a in length (Austin et al.,
1974), about 0 . 0 4 a shorter than in formaldehyde (Kato et al.,
1969). Considering the rather large force constants characteristic of
carbon-oxygen double bonds this may be a large enough change to
have a significant effect.
The formation of cyclohexadienyl and cycloheptatrienyl radicals
by hydrogen abstraction from certain precursors can result in larger
changes in nuclear geometry than those that accompnay the forma-
tion of simple allylic and benzylic radicals. Reliable data on the
appropriate hydrogen abstractions by methyl radicals do not appear
to be available, but data for ethyl radicals have been reported
(Trotman-Dickenson and Milne, 1967). To estimate how fast these
reactions should be in the absence of a PLNM effect, data on
reactions of ethyl radicals with saturated hydrocarbons are needed.
Log k for abstraction of hydrogen from cyclohexane by ethyl
radicals at 200' is too Iarge by 0.40 t o fit on the plot of log k. vs. AH
for methyl radicals shown in Fig. 5. Total rates of hydrogen
abstraction from n-butane, isobutane, n-hexane, and n-heptane have
been reported but not subdivided into rates for reaction with the
various different types of hydrogen atoms. Nevertheless, if the
carbon- hydrogen bond dissociation energies for the primary,
secondary, and tertiary hydrogen are assumed to be the same as those
for n-Pr-H, i-Pr-H, and t-Bu-H, respectively, the total rate
constants required t o fit the line in Fig. 5 may be calculated. The
actual values of log k 200' are Iarger by 0.25 t o 0.48. Thus, from five
values of k we estimate that the rates of hydrogen abstractions from
saturated hydrocarbons by ethyl radicals are fitted satisfactorily by
the line in Fig. 5 but somewhat better by a Iine 0.4 log unit higher.
We then find that log k for hydrogen abstraction by ethyl radicals
from 1,3-cyclohexadiene, 1,4-cyclohexadiene, and 1,3,5-cyclohepta-
triene is too small by 3.63, 3.18, and 2.80, respectively, to fit on
the line in Fig. 5. It seems clear that these reactions are being
slowed by a PLNM effect, since their rates are about the same as that
of the abstraction of tertiary hydrogen from isobutane by methyl
radicaIs, a process that is 15- 18 kcal mol-' less exothermic,
according to the available bond dissociation energies (Egger and
Cocks, 1973). However, it is much less clear how large the effect is.
24 JACK HINE
Although no curvature seems to be required by the points upon

which the line in Fig. 5 was based, the line must eventually level off
as it continues t o the left. Otherwise, it would reach log k values
exceeding those for diffusion-controlled reactions. The points for the
reactions in question, which are exothermic by 24.8- 28.2 kcal mol-' ,
would lie outside Fig. 5 t o the left. The line would not have reached
log k values characteristic of diffusion control at AH values of this
magnitude but its slope may well have decreased in absolute value,
corresponding t o smaller estimated PLNM effects. Therefore it is not
clear whether the PLNM effects are larger or smaller than those for
the formation of simple allylic and benzyl radicals, whose magni-
tudes are based on points whose AH values are much closer t o those
for the points from which the line in Fig. 5 was drawn. Smaller
effects could be explained by the more exothermic nature of the
reactions under consideration; larger effects can be explained in two
of the three cases by the larger changes in nuclear geometry. The rate
of abstraction of hydrogen from 1,4-~yclohexadieneis faster than
from 1,3-cyclohexadiene, for which AH is essentially the same. This
is as expected from the PLNM in view of the change in the sum of
the squares of the bond numbers of $ for the lY4-dienecompared to
L$ for the 1,3-diene, as described previously. The reaction of
1,3,5-cycloheptatriene is surprisingly rapid, however. It is about
3 kcal mol-' less exothermic than that of the cyclohexadienes. The
sum of the squares of the changes in bond numbers is 1.71, and the
markedly non-planar triene (Traetteberg, 19 64) changes t o a radical
that is probably very nearly planar. Yet, on a per hydrogen basis,
ethyl radicals are reported t o abstract hydrogen atoms from
1,3,5-~ycloheptatrieneas fast as they do from lY3-cyclohexadiene
(Brown and James, 1965). Perhaps these reactions should be
reinvestigated to be sure that such complications as hydrogen
abstraction from primary reaction products are absent.
When both the radical being formed and the one that is reacting
are undergoing major changes in internal geometry, larger PLNM
effects would be expected than when only one of the two radicals
was changing geometry significantly. One such reaction (7) is
CH2"CH-CHz
I
CHz -CH=CH2
+ PhCH3 --f CH2=CHCH3 + PhCH2 (7)
abstraction of a hydrogen atom from toluene by an allyl radical,

which is exothermic by 3.6 kcal mol-' and whose statistically
corrected value of log k (Ratajczak and Trotman-Dickenson, 1969) is
too small by 3.0 to lie on the line shown in Fig. 5. (In this reaction
statistical correction includes not only division by three t o put the
rate constant on a per hydrogen basis but also, in view of the
symmetrical character of the allyl radical, division by two t o put it
on a per carbon basis.) The deviation might be expected to be
approximately the sum of the deviations that may be seen in the
figure for the formation of benzyl (2.3 log units) and allyl (1.5 log
units) radicals by attack of methyl radicals on toluene and propy-
lene, respectively. However, we have already mentioned that the
tendency of the radicals being formed to be increasingly pyramidal as
we move to the left along the line in Fig. 5 may make the line
somewhat steeper than it would otherwise have been. If the right end
of the line were raised slightly to obtain a better approximation of
the ideal line for hydrogen transfer to methyl radicals from saturated
hydrocarbons without significant change in internal geometry, the
deviation of the reaction of allyl radicals with toluene would increase
and those for the reactions of methyl radicals with propylene and
toluene would decrease, with the former deviation being even more
nearly equal to the sum of the latter two.
No large solvent effects on the reactions of methyl radicals with
hydrocarbons would be expected. Therefore it is not surprising that
the abstraction of tertiary hydrogen atoms from saturated hydro-
carbons in the liquid phase is faster than abstraction of primary
benzylic or primary or secondary allyl hydrogen (Pryor et al., 1972),
even though the latter types of reactions are 3- 10 kcal mol-' more
exothermic. A number of other data on hydrogen transfer reactions
in the liquid phase (Hendry et al., 1974) give the same type of
evidence for the PLNM as the gas phase data we have already
discussed.
Radical Decomposition Reactions

Although free radical decomposition reactions do not offer the
range of reliable data that is found for hydrogen-transfer reactions,
they probably come closer to doing so than any other type of radical
reaction. Some of these data seem interpretable in terms of the
PLNM, but other factors are certainly important also.
26 JACK HINE
Data on reactions, such as those shown in equations (8)-(11),in

which a radical decomposes t o give a stable molecule and a new
RCO. + R. + CO (8)
radical by cleavage of a carbon-carbon or carbon- oxygen bond have

been collected in Table 1. All the “preferred” values of log A and E
listed in the compilation by Benson and O’Neal (1970) were used
except the following. The two reactions in which the measurements
quoted were carried out at only one temperature (and that tempera-
ture far from the temperature at which rate constants were needed)
were ignored. No preferred values were listed for the decomposition
of isopropoxy and 2-butoxy radicals, but values from the latest
references have been quoted, the log A value for 2-butoxy in the
reference (East and Phillips, 1966) being different from that quoted
by Benson and O’Neal; a number of other values are from more
recent literature than that covered in their compilation. The rate
constants reported by Cadman et al. (1970) for the decomposition of
butyryl, isobutyryl, and propionyl radicals seemed unreasonably
large, and the value for propionyl is more than 1 0 times as large as
that of Watkins and Thompson (1973). These values have been
ignored, agreeing with Perkins and Roberts (1974) that the assumed
rate constant of 10’ ‘ M -* s - l for the combination of *NF2 and acyl
radicals is unreasonably large. Also ignored were the decarbonylation
rate constants for acyl radicals given by Perkins and Roberts; their
assumption that acyl radicals add to 2-methyl-2-nitrosopropane with
the same rate constant that t-BuQCQ* radicals do was thought t o be
an unsatisfactory approximation.
Values of AH for the various reactions were taken from Benson
and O’Neal (1970) or calculated from the enthalpies of formation of
the compounds (Cox and Pilcher, 1970) and radicals (Egger and
Cocks, 1973); some enthalpies of formation of compounds were
calculated by adding group contributions (Benson, 1968). For two
TABLE 1
Rates of Decomposition of Radicalsa

-- ~
Reaction Phase logAb EC log kb*d AHc

~ ~~
n-Pr. + M e . + CHzzCH2 Gas 13.6 33.1 -4.80 25.8

n-Bu. -+Et. + CHz=CHz Gas 13.6 29.0 -2.52 22.5
2-Bus Me. f propene
-+ Gas 14.16 33.9 -4.68 26.2
i-Bu- + Me. + propene Gas 14.2 32.7 -4.28e 25.1
EtO. + M e - + CHzO Gas 12.15-f 22.11 -0.16 11.2
i-Pro. -+ Me: + MeCHO Gas 11.8g 17.2g 1.94e 6.6
t-BuO. -+ Me. + MezCO Gas 13.5 16.5 3.85h 3.8
2-Bu0. -+Eta + MeCHO Gas 13.4' 17.5' 3.67 6.2
MeCO. + M e . + CO Gas 10.3 . 15.0 1.96 13.4
EtCO- Et. + CO
-+ Gas 12.77' 14.4 4.77 10.2)
c-C~HSCO. c - C ~ H aSk + CO
-+ Gas 11.7' 17.9' 1.75 13.grnsn
t-BuCO. -+ t-Bu* + CO Liquid 11 .go 9.3O 6.73 4.6m2"
PhCO. -+ Ph. + GO
--
Gas 14.6p 29.Q -1.74 26.0p
MeCOz Me. + COz
-+ Liquid 9.49 -14.7m
PhCO2 + Ph. + COz Liquid 5.3' 0.6"'J
t-BuOCO. + t-Bu. + co2 Liquid 1 3.4r 12.lf 6.67 -22.2rn*
EtOCHMe -+ Eta + MeCHO Gas 10.91 23.5 -2.15 3.7
MeCOCHZ- Me. + CH2CO
-+ Gas 12.5 40.0 -9.73 25.6
a Data from Benson and O'Neal (1970) unless otherwise noted.

Time units, seconds.
In kcal mol-'.
At 120°C.
k divided by a statistical factor of 2.
f Leggett and Thyme, 1970.
Cox et al., 1966.
h k divided by a statistical factor of 3.
!East and Phillips, 1967.
I Watkins and Thompson, 1973.
' Cyclopropyl.
Kerr et al. 1969.
Calculated from enthalpies of formations of stable compounds (but not radicals)
estimated using group contributions (Benson, 1968), and/or enthalpies of formation of
radicals from the compilation of Egger and Cocks (1973).
Assuming A H for RCHO RCO. + H. is 86.8 kcal mol-'.
-+
Schuh et al., 1974.

p Solly and Benson, 1971.
Kaptein et al., 1972.
From the report that k is between lo4 and l o 5 s-l at 80 (DeTar, 1967) and the
assumption that log A is 12, a roughly average value for radical decompositions.
Using the enthalpy of formation of the benzoyloxy radical (-17 kca1 mol-') reported by
Carson et d., (1975).
'
'
Griller and Roberts, 1972.
Using the assumption that the energy of the carbonyl-hydrogen bond of t-butyl form-
ate is the same as for methyl formate, namely 95.3 kcal mol-' (Egger and Cocks, 1973).
28 JACK HINE
acyl radicals, AH of formation was based on an enthalpy of

86.8 kcal mol-’ for reaction (12). Values within 0.8 kcal mol-’ of
RCHO -+ RCO. + H - (12)
86.8 kcal mol-‘ are reported for the cases where R is methyl,
phenyl, and hydrogen.
-I
-8
I I
-20
I I
-10
I I
0
1 I
10
A H , kcol mol-‘
I
MeCOCH,.
I
20
IV
Figure 6. Log k for gas phase decompositions of radicals vs. AH of reaction, o decarbonyla-
tion of acyl radicals; 0 decomposition of alkyl radicals to give smaller alkyl radicals plus
alkenes, A decomposition of alkoxy radicals to give alkyl radicals and aldehydes or ketones;
Adecarboxylation of acyloxy radicals.
The data in Table 1 include log k values at 120°, a temperature

near the middle of the range over which measurements were made.
When the bond broken was one of several equivalent bonds, k was
divided by an appropriate statistical factor (but log A was not
corrected). The reactions covered are taken t o include four families
of related reactions plus three odd reactions. Values of log k are
plotted against A H in Fig. 6. For each family of reactions the points
are within the experimental uncertainty of a straight line ( a fact that
is meaningless in the case of the decarboxylation of acyloxy radicals,
a family of reactions that contains only two members). According to
a very simple picture of the product stability principle, in which
“uphill” reactions have transition states that resembIe the products
and "downhill" reactions have transition states that resemble the

reactants, such plots should be smooth curves with slopes (in
dimensionless terms) between zero and 0.5 when AH is negative and
between 0.5 and 1.0 when AH is positive. The line for decarboxyla-
tions of acyloxy radicals, which lies largely in the exothermic region
of the figure, has a slope of 0.48 (when the log k values are
multiplied by -2.3RT to turn them into kcal mol-' and put the
slope on a dimensionless basis). The slope of the line for the
decarbonylation of acyl radicals is also plausible, being 0.72. The
slopes of the lines for the decomposition of alkoxy and alkyl
radicals, being 1.02 and 1.13, respectively, are somewhat larger than
expected. However, these slopes are probably less reliably known,
being based on families of reactions covering ranges of only 7.4 and
3.1 kcal mol-', respectively, in contrast to the 15.3 and 21.4
kcal mol-' ranges of AH values for the other two families. The fact
that the one decarbonylation reaction studied in the liquid phase fits
the correlation as well as any of the other reactions suggests that
solvent effects are riot significant, in the case of the decarbonyla-
tions, at least.
Thus, it appears that the relative reactivities within the various
families of reactions can be explained largely in terms of the product
stability principle. Some of the differences in reactivity between
reactions in different families are in the direction that would be
expected from the PLNM. For example, the fastest reactions for 2
given enthalpy of reaction are the decarbonylations. Geometrically
these reactions involve going from a carbon-oxygen double bond in
the acyl radical to what is often written as a triple bond in carbon
monoxide. However, since the carbon-oxygen double bond in acyl
radicals is unusually short, if one may judge from data on the formyl
radical (Austin et al., 1974), this is probably a fairly small change in
bond distance. Decomposition of an alkoxy radical involves going
from a carbon-oxygen single bond in the radical t o a double bond in
the carbonyl product and going from carbon- carbon- oxygen bond
angles of about 109" t o about 120'. Decomposition of an alkyl
radical also involves changes in bond angles adjacent t o the double
bond being formed in addition to changing a bond from single to
double. Decarboxylation of acyloxy radicals also involves such a
change, this time from about 120" t o 180". Thus, the decomposi-
tions of alkyl, alkoxy, and acyloxy radicals all involve a type of
30 JACK HINE
change in internal geometry that has no parallel in the decarbonyl-

ation of acyl radicals. Log k for these three types of reactions falls
short of log k for acyl radical decomposition by roughly the same
amount. In view of the larger change in internal bond angles that
accompanies decarboxylation of acyloxy radicals, larger deviations
might have been expected for reactions of this type.
No convincing explanation is apparent for why the deviations are
no larger, but it would be valuable t o have data on the gas-phase
decarboxylation rates; perhaps solvent effects make the reactions
faster in solution than in the gas phase. A large deviation from the
decarbonylation rates is seen for the decomposition of acetonyl
radicals (to ketene and methyl radicals), which, like the decarboxy1-
ations, involves a change in bond angle from about 120" t o 180". It
should be noted that, in addition to the PLNM effect, there are
stereoelectronic reasons why the decomposition of acetonyl radicals
should be slower than that of alkoxy or alkyl radicals having the
same enthalpy of reaction. In the transition state, the orbital(s)
containing the unpaired electron in the reactant should overlap with
the orbital that will contain the unpaired electron in the product.
This is accomplished when the bohd being broken is on an atom
adjacent to the one bearing the unpaired electron and nearly parallel
to the orbital containing that electron. If the acetonyl radical were a
u radical with the unpaired electron localized on oxygen, decomposi-
tion could proceed readily via a transition state in which all the
atoms except some of the methyl hydrogen atoms were coplanar as
in [ 7 ] , Or if it were a u radical with the unpaired electron on carbon
it should use a transition state [8] in which the plane of the

methylene group was perpendicular t o the plane of the other carbon
atoms and the oxygen atom. However, it is much more likely that
the acetonyl radical is a rr radical, as is the somewhat similar radical

formed by loss of a hydrogen atom from urea (Bower et al., 1971).
In such a rr radical [9] the carbon-carbon bond t o be broken is
.c-c //o
H\ H, ,O.
,c=c,
t-f
H’ ‘CH3 H CH3
[91
perpendicular to the overlapping p orbitals that form the rr system
containing the unpaired electrons. This should make reaction much
slower than in the case of alkoxy and alkyl radicals which do not
have this stereoelectronic disadvantage. The same type of stereo-
electronic factors should also slow the decarboxylation of acyloxy
radicals. However, it is likely that the acyloxy u radical with the
unpaired electron localized on oxygen is much more stable relative to
the radical than is the case with the acetonyl radical. In the case of
an acyloxy radical such a o radical should be greatly stabilized by
interactions between the carbonyl group and an unshared electron
pair on the singly bonded oxygen atom as in [ l o ] . This should
- -
0
0.
R-CG- t--., R-C,-
!?I 01
-
[lo1 0
increase the concentration of such radicals and hence the rate of

decarboxylation.
The two radical decomposition reactions in Table 1 and Fig. 6 that
involve cleavage of carbon-oxygen bonds are much slower than the
carbon- carbon bond-cleaving radical decompositions that have
similar enthalpies of reaction. Perhaps these differences arise from
some difference in the properties of carbon-oxygen and carbon-
carbon bonds, but there seem to be stereoelectronic factors that at
least partially explain the results. The decomposition of a-ethoxy-
ethyl radicals, for example, is a million times too slow t o fall on the
line for the decomposition of alkoxy radicals, although it, like them,
leads to the formation of an alkyl radical and a carbonyl compound.
There seems to be no reason why the PLNM effect should be larger
for the former reaction than for the decomposition of alkoxy
radicals. In the decomposition of alk-ox): radicals the bond angles at
32 JACK HINE
the carbon atom to which oxygen is attached must change from

about 109.5" t o about 120". Judging from other oxygenated radicals
(Dobbs et al., 1971, 1972) the a-ethoxyethyl radical is probably not
planar at the trivalent carbon atom but the bond angles are probably
significantly larger than 109.5". There seems to be no reason to
expect the carbon-oxygen single bond that is becoming a double
bond to be any shorter in an alkoxy radical than in an a-alkoxy alkyl
a
radical; the 0-H bond in the hydroxy radical is 0.971 long and
that in water is 0.958 a (Sutton, 1958). The strong stabilizing
interaction between resonance-electron-donating substituents and a
radical center makes stereoelectronic effects plausible. The fact that
the hydroxymethyl radical has a pK, of 10.7 (Asmus et al., 1966)
whereas methanol has a pK, of 15.5 (Ballinger and Long, 1960)
shows that, relative to the -OH substituent, the -0- substituent
stabilizes an unpaired electron (relative t o a bond t o hydrogen) on
the carbon atom to which it is attached by about 2.3RT(l5.5 -
10.7), or 6.5 kcal mol-' . Such a stabilizing interaction leads one to
expect the orbital containing the unpaired electron to be nearly
parallel t o an orbital containing an unshared pair of electrons. Thus,
the cu-ethoxyethyl radical would be expected to exist largely in a
conformation like that shown in 1111 as a Newman projection down
the bond from oxygen to trivalent carbon. This should be much
more stable than [ 121 , in which the orbital containing the unpaired
electron is a n t i and parallel to the carbon-oxygen bond to be
broken.
In the decomposition pf the radical formed by removal of
hydrogen from the carbonyl carbon atom of t-butyl formate (to give
carbon dioxide and t-butyl radicals) it is conformation [ 131 that has
the bond to be broken anti to the orbital containing the unpaired

electron. Yet in this conformation there should be marked repulsion
between the t-butyl group and the carbonyl oxygen atom. This and
the large change in geometry at the carbonyl carbon atom that must
accompany decarboxylation give a possible explanation for why
t-butyl groups are lost from t-BuOCO. at about the same rate that
they are lost from t-BuCO. even though the former reaction is more
exothermic by about 27 kcal mol-’ . If this explanation is correct the
decarboxylation of MeOCO., where the conformation analogous to
[ 131 would be less strained, would give a smaller deviation from the
lines in Fig. 6.
Although a number of the observed deviations from the product
stability principle thus seem qualitatively rationalizable in terms of
the PLNM and stereoelectronic effects, it seems likely that polar
effects are probably also important in some of the decompositions of
oxygen-containing radicals.
5. MULTICENTER REACTIONS
Multicenter reactions, like free radical reactions, are relatively well

suited to the application of the principle of least motion in that
solvent effects are commonly small if the reactants and products are
not ionic. It is therefore fitting that the principle has probably been
applied more often t o multicenter reactions than t o any other type
of reaction. Most of the applications have dealt with the conservation
of orbital symmetry, a corollary of the principle of least change in
electronic configuration. In some reactions conservation of orbital
symmetry seems t o be the most important factor, but in others it is
the PLNM that is most important. When the PLNM effect is very
large and obvious its importance is often recognized implicitly, as in
the statement “Antarafacial processes are obviously impossible for
transformations which occur within small or medium-sized rings”
(Woodward and Hoffmann, 1970). In a number of cases where the
PLNM effect is not as large and obvious its importance has been
discussed explicitly .
It seems that the one of two or three competing multicenter
reactions of a given type that is favored by the PLNM is usually also
the one that best conserves orbital symmetry. Thus, in the rearrange-
34 JACK HINE
ment of cyclopropyl cations to allyl cations, the disrotatory pathway

(13) is favored over the conrotatory pathway (14) according t o the
PLNM (Tee and Yates, 1972; Ehrenson, 1974b). Disrotatory re-

arrangement is also favored by conservation of orbital symmetry
(Woodward and Hoffmann, 1970).
For the concerted formation of an allyl cation from a cyclopropyl
halide,4 we may imagine a conrotatory process and two different
disrotatory processes. The two disrotatory processes differ in that in
one the two substituents that are moving toward each other are cis to
the departing halogen( 15), and in the other the two substituents that
are moving apart are cis to the departing halogen. The principles of
least nuclear motion (Tee and Yates, 1972; Ehrenson, 1974b) and of
conservation of orbital symmetry (Woodward and Hoffman, 1970)
agree that the disrotatory pathway shown in (15) should be the most
favored of these three alternative mechanisms of reaction. Relevant
experimental data can be found in the observation of Cristol,
Sequeira, and DePuy (1965) on the solvolysis of the epimeric
7-chloronorcaranes. The cis isomer [ 141 , whose solvolysis by the
disrotatory mechanism shown in (15) would give carbonium ion
[16] , undergoes acetolysis at a reasonable rate at 125’. The
acetolysis of the trans isomer [15] by the same mechanism, which
Although it is not a multicenter reaction, this process is so closely related t o the

rearrangement of cyclopropyl cations discussed in the preceding paragraph that it seems to
be appropriate t o treat it here.
would give the highly strained carbonium ion [ 1 7 ] , is much slower,

even at 210°. Thus, the experimental results are in accord with the
PLNM and the principle of conservation of orbital symmetry; they
also illustrate the tremendous importance of the product stability
principle.
In its simplest form, at least, the PLNM does not correctly predict
the stereochemistry of the cyclization of cis-1,3,5-hexatriene t o
1,3-~yclohexadiene,and it disagrees with conservation of orbital
symmetry arguments, which do predict the stereochemistry
correctly. The reaction can take place in either a conrotatory (16) or
a disrotatory fashion. When rotation around the single bonds of the

reactant is taken as being completely free and the optimum conform-
ation is taken as the starting point, the unweighted sum of the
squares of changes in nuclear positions is less for conrotatory than
for disrotatory cyclization (Tee and Yates, 1972). Experimentally
the thermal cyclization of derivatives of cis-l,3,5-hexatriene has been
found to be a disrotatory reaction in a number of cases, as expected
because of better conservation of orbital symmetry (Woodward and
Hoffmann, 1970). Ehrenson (1974b) has suggested that the relative
force constants for motion in the direction of conrotatory and
disrotatory cyclization are such that a force-constant weighted
36 JACK HINE
treatment of nuclear motions would show the disrotatory process to

be favored.
The Cope rearrangement, as in the case of the degenerate re-
arrangement of 1,5-hexadiene (17), can proceed by a conrotatory
process, with a chair-like transition state, or by a disrotatory process,

with a boat-like transition state. An unweighted PLNM treatment led
t o essentially identical estimated PLNM effects for the two different
mechanisms (Tee and Yates, 1972). Consideration of the situation in
terms of the conservation of orbital symmetry led t o the conclusion
that expectations of differences in the ease of the two processes
could be based only on secondary effects. Such effects appear t o
favor chair-like transition states and indeed, a preference, albeit a
weak one, for the chair-like transition state is found experimentally
(Woodward and Hoffmann, 1970).
One of the most striking successes in the application of conserva-
tion of orbital symmetry to organic reactions has been its ability t o
rationalize data on photochemical as well as thermal reactions. Our
ability to apply the PLNM to such processes is limited by our
widespread ignorance about the geometry of electronically excited
states. Nevertheless, it would be worthwhile to estimate such geo-
metries in the best ways available and to use them in PLNM
treatments.
6. POLAR REACTIONS
There are probably more reliable rate and equilibrium data
available for polar reactions in solution than for all other types of
reactions combined. Solvation almost always has a profound effect
on these data and often the formation of ion pairs and higher
aggregates is also quite important. We have discussed radical and
multicenter reactions first in order to minimize such complications.
The fact that the PLNM often seems useful in rationalizing the data
on reactions of these types may encourage us t o seek its manifesta-
tions in the more dimly lit region of polar reactions. Solvation effects
on polar reactions seem unavoidable, but to minimize the gegen-ion
effects that are often so important in poorly ion-solvating media we

shall prefer to examine data obtained in aqueous solution when such
data are available. We shall start with proton transfer reactions,
which provide us with the largest number of reactions for which both
rate and equilibrium constants have been determined,.
Proton Transfers f r o m Carbon
In looking for PLNM effects on proton transfer reactions let us

consider proton transfers from carbon. Only for proton transfers to
and from carbon are there data on a large number of reactions in
which the proton transfer (rather than diffusion of the reactants
together) is the rate-controlling step. The largest body of such data
refers to proton transfers t o the bases water and hydroxide ion in
aqueous solution. In order t o allow for product stability effects we
shall consider only those acids whose ionization constants have been
determined in aqueous solution, estimated pK, values above 16, for
example, often being subject t o very large uncertainties. There are
three families of m- and p-substituted aromatic acids that meet the
preceding criteria, which give excellent rate-equilibrium correlations.
These are the rates of proton transfer from acids of the type
ArCH(N02)2 to water (Dronov e t al., 1969) and from acids of the
type ArCHMeN02 and ArCH2 NO2 to hydroxide ion (Bordwell and
Boyle, 1972). There are also three families of reactions involving
changes in reactant structure closer t o the acidic proton than the m-
or p-position of a phenyl substituent for which fairly good correla-
tions can be found. One of these is obtained by weeding Dronov and
Tselinsky’s (1970) data on 58 gem-dinitro compounds. All com-
I
pounds not of the type -qCH2CH(NO2), were neglected, as were
I
compounds containing carboxy, amino, o?%mide substituents (there

being evidence for proton removal from carbon with neighboring
group participation by such groups in several cases), and the
compound t-BuCH, CH(N02) 2 (perhaps the most hindered of the
remaining compounds). A plot of log K for proton removal by water
vs. pK, for the remaining 18 compounds covered a pK, range of 4.1
and fitted a straight line of slope 0.77 with a standard deviation of
0.10 in the log Iz values, This slope is about the same as that (0.74)
38 JACK HINE
TABLE 2
Rates of Proton Removal from Carbon in Aqueous Solution at 25OC
lzhC
Acid M-1 s-l
HC(N02 ) 3 0.2 2.0 x 104d7e

H2C(N02 12 3.5 0.7ld8f
PhCH(N0Z)z 3.8 2.2 x 10-2d,g
Barbituric acid 4.0 1Oh. i
02NCHzhc 5.1 3.7 x 10-2i
Meldrum's acid 5.3 2.4h9i
EtCH(NO& 5.4 5.9 x 1 0 4 ~ J
0 2 NCH2 C02 Et 5.8 1.6 x 1.5 x 1 0 5 ~
PhCHzNOz 6.9 160'
BrCHAcZ 7 .oi 3.4 x 10-2"
PhCH(Me)N02 7.4 6.5'
Me2 CHNOz 7.70 1.5 x 0.36p
HC(C02Me)3 7.8 0.1 l k
Nitrocyclohexane 8.3 0.364
MeCH2N02 8.6 3.7 x 10-8; 5.24
PhCOCH2 Ac 8.7' 1.1 x 10-2;
H~CAC~ 9.0 1.3 x 4 104ki
HCN 9.1 3.7 x lo9'
3-Thianaphthenone 9.7 7.9 x 10-4d83
2-Acetylcyclohexanone 9.9 4.6 x 10-4k
H,CNO;? 10.2 4.3 x 10-8; 2 84
2-Carbethoxycyclopentanone 10.5 2.3 x 1 0 - ~ i
AcCHzCOzEt 10.7 1.2 x 10-3;
MeCHAcz 11.0 8.3 x 10-5i
Hz C ( W 2 11.2 2.9 x
H2 W O 2 Ph)Z 11.2 0.23'
2-Carbe thoxy cyclohexanone 11.5 9.7 x 10-6;
PhCH(S02 Et)2 12.1 5.9 x 1 0 4 '
H2 C(SO2 Et)2 12.2 1.7 x
HzC(S02Me)2 12.5 1.1 x 10-2'
AcCH(Et)C02Et 12.7 7.5 x 10-6;
t-BuCH(CN)z 13.1 5 . 4 ~1 0 4 r 4.3 x 105'
H2 C(CO2Et)2 13.3 2.5 x 10-5;
P-O~NC~H~C CN
HZ 13.4 8 x 10-7v*w 200v.w
AcCH2S03- 13.6 2.0 x 10-6k 260k
MeCH(SO2 Ph)2 13.8 6.4 x 10-4'
MeCH(SO2 Et)? 14.6 3.5 x 10-5x 4.3 x 1 0 5 ~
a Not corrected for enolization or hydration. Obtained from the same source as k , or
k h unless otherwise noted.
By the base water.
By the base hydroxide ion.
At 20°C.
Chaudhri and Asmus, 1972.
obtained in the analogous plot for compounds of the type

ArCH(N0, ),. Another family of reactions is taken t o consist of the
rates of deprotonation of six compounds of the type XCOCH, COY
by water. Data on these reactions are listed in Table 2, as are data on
deprotonations of six disulfones of the type RCH(S0, R),. Also
listed in Table 2 are data on the parent reactions for the two series of
gem-dinitro compounds and for the series of ArCH,NO, and
ArCHMeNO, compounds. In addition, other rates of removal of
protons from carbon by water or hydroxide ions in aqueous solution
at or near 25” are listed, where the ionization constant of the acid in
aqueous solution is known. Many values are those listed by Pearson
and Dillon (1953), but we have omitted their values (and values from
other sources) for trifluoromethyl ketones and the k w values that
were estimated from k h values or from data on other compounds.
The trifluoromethyl ketones would exist very largely as hydrates in
aqueous solution, but this complication could not be corrected for
since the equilibrium constants for hydration are not known.
The rate constants for proton removal by hydroxide ion listed in
Table 2 and those for all the reactions in the two series for which
only data on the parent compound are listed in the table are plotted
logarithmically against the pKa values in Fig. 7. To provide an
improved basis for illustrating and allowing for product stability
effects the “intrinsic barrier” concept of Marcus (1968, cf. Cohen
and Marcus, 1968) has been used. It is assumed that in the absence of
f Dronov and Tselinsky, 1970.

g Dronov et al., 1969.
At 12°C.
f Eigen, 1964.
JPearson and Dillon, 1953.
Barnes and Bell, 1970.
Bordwell and Boyle, 1972.
Bell and Crooks, 1965.
Turnbuli and Maron, 1943.
Extrapolated from data obtained at 54-60°C (Belikov et al., 1968).
p Davies, 1974.
4 Bell and Goodall, 1966.
In 7% EtOH-H20 (Eidinoff, 1945).
Rubaszewska and Grabowski, 1969.
‘Hibbert, et al., 1971.
Hibbert, 1973.
Assuming k H / k T is 8.0.
Hibbert and Long, 1972.
Bell and Cox, 1971.
40 JACK H M E
Figure 7 . Plot of logkh for deprotonationby hydroxide ions inaqueous solution at 25' vs.
PK,. Ar refers to m- and p-substituted phenyl groups. The dashed lines are based on eqn.
(18) and refer t o the intrinsic barrier ( A ) shown.
a barrier the reactions (all of which have equilibrium constants

substantially larger than 1 .O) would proceed with diffusion control,
having rate constants of 1 0 " ~- l s - l (cf. Eigen, 1964). Using the
simplest form of the Marcus treatment (which may be derived on the
basis of intersecting congruent parabolas), this gives (18) for k h . The
log k h = 10 - -
2 . 3AR T [,t
2 . 3 R T ( p K a - 15.74)
412 1
intrinsic barrier A is expressed in kcal mol-' . Dashed lines are drawn
in Fig. 7 for barriers of 5, 10, 15, and 20 kcal mol-'. No statistical
corrections have been made in the plot. For members of families of
similar reactions, such corrections would not change the correlation
because the correction would be the same for each reaction. The
deviations of dissimilar reactions from a linear correlation are far
larger than could be removed by any plausible statistical corrections.
The acid with the smallest barrier t o deprotonation by hydroxide

ions is hydrogen cyanide, clearly the one for which the smallest
change in internal geometry would be expected to accompany
deprotonation. The next smallest barriers are for a disulfone and a
dicyano compound. All the other compounds, which have much
larger barriers, have their acidic hydrogen atoms activated by
carbonyl groups, nitro groups, o r both. These facts are also ration-
alizable in terms of the PLNM. The predominating retention of
configuration accompanying the base-catalyzed deuterium exchange
of sulfones of the type RSOzCHRR’ (Cram et al., 1960; Corey and
Kaiser, 1961), the relatively high acidity ( p K , 3 . 3 ) of the bicyclic
trisulfone [ 181 (Doering and Levy, 1955), and other observations
show that carbanions stabilized only by a-sulfone substituents
certainly do not have a strong preference for coplanarity and may
even have some preference for a pyramidal structure. Thus, de-

protonation of a sulfone need not be accompanied by as much
change in bond angles as would be the case for a carbonyl o r nitro
compound. There is probably also a smaller change in bond lengths
in the sulfone reaction. The triketone [ 191, in contrast t o [18],
shows little acidity from its triply activated bridgehead hydrogen
atom (Theilacker and Schmid, 1950; TheiIacker and Wegner, 1963).
&-Cyan0 substituents also show less encouragement of carbanion
planarity than a-carbonyl substituents do. For example, optically
active 1-cyano-2,2-diphenylcyclopropane gives deuterium exchange
with 99% retention of configuration in the presence of sodium
methoxide (Walborsky and Motes, 1970) whereas optically active
1-benzoyl-2,2-diphenylcyclopropane gives only 24% retention (Motes
and Walborsky, 1970). This probably results from the substituent
effect of the cyan0 group being more largely inductive in character
relative to that of a carbonyl substituent, which operates largely by a
42 JACK HINE
resonance effect. The greater electronegativity of oxygen permits a

larger contribution of valence-bond structures in which the negative
charge is delocalized from the carbanion center on to the substituent
in the case of carbanions alpha to carbonyl groups. The greater
electronegativity of sp-hybridized carbon relative to that of
sp2-hybridized carbon gives the cyano substituent a larger inductive
effect. (The Taft substituent constant for CH2 CN is more than twice
as large as that for CH2 Ac.) The largely inductive substituents,
phenylsulfonyl, cyano, and trifluoromethyl, have been shown to
comprise a category of substituents that give cyclopropyl derivatives
that form carbanions more rapidly than do the corresponding
2-propyl derivatives; with the other category, composed of nitro and
various carbonyl substituents, the reverse is true (van Wijnen et al.,
1972). Since cyclopropyl anions have a stronger preference for a
pyramidal structure than 2-propyl anions do, the cyclopropyl anions
are better stabilized by substituents that do not tend to impose
planarity.
Many of the relative sizes of barriers to carbanion formation
shown in Fig. 7 are not explained by the PLNM, of course. The
slopes of the lines through the points for arylnitromethanes and
I-arylnitroethanes are not only larger than would be expected for
constant barriers throughout a series, they are larger than 1.0, so that
in the reverse reaction it is the most basic carbanions that are
protonated most slowly (Fukuyama et al., 1970; Bordwell, et al.,
1970). This well-known anomaly does not seem explicable using the
PLNM. The generally larger barriers for the l-arylnitroethanes relative
to those for the arylnitromethanes are presumably at least partly of
steric origin. However, it is not clear why the same replacement of
a-hydrogen by methyl increases the barriers by so much larger
amounts on going from nitromethane to nitroethane or from nitro-
ethane to 2-nitropropane.
Figure 8 is a logarithmic plot of k, values against pK,. To avoid
cluttering the graph the data for two highly populated series of
gem-dinitro compounds have been plotted as the two least squares
lines. Also, data on four P-dicarbonyl compounds (not of the type
XCOCH2COY) that do not seem capable of adding anything to the
discussion were not plotted. Aside from these, all the lz, values in
Table 2 are plotted. For use of Marcus barriers the rate constants for
diffusion of a hydrogen ion and a carbanion together were assumed
Figure 8. Plot of log kw for deprotonation by water in aqueous solution at or near 2 5 O vs.
pKa. A, disulfones of the type RCH(S02R)z; 0 , Dicarbonyl compounds of the type
XCOCH*COY, including cases where X and Y are identical and cases where X and Y are con-
nected to give a ring. Ar refers to m- and p-substituted phenyl groups. The dashed lines are
based o n eqn. (19) and refer to the intrinsic barrier (A) shown.
to be 3 x 10'' M - ' s-' (larger than for diffusion together of a

hydroxideionand anacid becauseof thegreatermobility of the hydrogen
ion and the attraction of the opposite charges for each other). The
appropriate relation for rate constants with a given barrier as a
function of pK, is then eqn (19). Dashed lines corresponding to
log k , = 12.22 - -
2.3RT
2.3RT(pKa + 1.74)
4A
barriers of 0, 5 , 10, 15, and 20 kcal mol-' are shown in Fig. 8. No
1
statistical corrections have been made.
The rate constants for deprotonation by water plotted in Fig. 8
show some of the same trends already discussed in connection with
the rate constants for deprotonation by hydroxide ions plotted in
Fig. 7. The lowest barriers are for the disulfones, with protonation of
the carbanions by hydronium ions being essentially diffusion-
controlled for five of the six compounds of the type RCH(S02R)2.
The sixth compound, PhCH(S02 E t ) 2 , is only slightly less reactive
and is the one for which proton transfers would be expected to be
44 JACK HINE
subject t o the largest amount of steric hindrance. The barriers for the
gem-dicyanides are almost as low as those for the disulfones. The
other acids, all of which have higher barriers, all have the acidic
hydrogen atoms activated by nitro and/or carbonyl groups. For the
three series of acids [ArCH(NO, ), , RCH, CH(N02),, and
XCOCH, COY] in this category there are rate-equilibrium correla-
tions that correspond to rather small changes in intrinsic barriers
within a series. However, the structural changes in the XCOCH, COY
series, which includes cyclic species such as barbituric acid [ 201 and
Meldrum’s acid [21] ,are so much larger and the number of members
of the series smaller than those of the two series of gem-dinitro

compounds that one may wonder if the good agreement of the
P-dicarbonyl data with a straight line is partly coincidence. Steric
hindrance to deprotonation of an acid, which will also constitute
steric hindrance to protonation of the carbanion, should increase the
intrinsic barrier. Such hindrance may be important in causing barriers
to increase in the orders: MeNO, < EtNO, < i-PrN02; CH2 (NO,),
< RCH, CH(N0, ), ; XCOCH, COY < MeCHAc, and AcCHEtCO, Et.
There is somewhat ambiguous evidence for another trend in
relative barrier heights, one that is in the direction expected from the
PLNM. Other factors being equal, a smaller barrier to deprotonation
is expected when the acidic proton is activated by two substituents
capable of resonance delocalization of the negative charge than when
it is activated by only one such substituent. Calculations leading t o
this conclusion have already been described for the case of the
deprotonation of 1,3-~yclohexadiene, in which the acidic protons are
on a carbon to which one butadienyl substituent is attached, and
1,4-cyclohexadiene, in which the acidic protons are on a carbon t o
which two vinyl substituents are attached. The calculations that
showed it PLDjhl effect favoring protonation of cyclohexadienyl
anions to give 1,4-cyclohexadierie rather than 1,3-cyclohexadiene plus
the principle of microscopic reversibility lead to the conclusion that
a PLNM effect will favor deprotonation of the 1,4.-isomer.The quick

method of estimating PLNM effects described earlier may be
applied to carbanion formation from a monocarbonyl compound and
a 0-dicarbonyl compound as follows. Consider the carbanion derived
from the monocarbonyl compound to be a resonance hybrid of
-
contributing structures [ 2 2 ] and [23].Assign a weight of w to the
,c-c p \ ,0e
,c=c
\Q
\ \
~ 3 1
structure [ 2 3 ] with the negative charge on oxygen, relative to a

weight of 1.0 for [22].This gives a bond number of (1 + 2w)/(l+ w )
for the carbon-carbon bond and a bond number of (2 + w)/(l+ w )
for the carbon-oxygen bond in the hybrid enolate anion, which gives
a value of 2w2/( 1 + w ) for
~ the sum of the squares of the changes in
bond numbers that accompany carbanion formation. For a
0-dicarbonyl compound the carbanion has three contributing
structures. A weighting of w for the two structures with the negative
charge on oxygen leads to the bond numbers for the resonance-
stabilized carbanion shown in [24].The sum of the squares of the
0_ -c
~_ _ _ _ I GiG
1 + 3w
__ c
_ _ _ _~ 1 ::;:
1 G2W
_ _0 _ _
2 + 3w 1 + 3w
1 + 2w
~ 4 1
changes in bond numbers will be 4 w 2/( 1 + 2 ~ ) This
~ . is smaller than
~ any w larger than 1/.\/2. Since the negative charge
2w2/(1+ w ) for
on the resonance-stabilized anions should lie on the more electro-
negative oxygen atoms to a considerably larger extent than on the
carbon atoms, w should be considerably larger than 1.0. Therefore,
the PLNM will tend to give a larger intrinsic barrier to carbanion
formation by a monocarbonyl compound than by a 0-dicarbonyl
compound. Unfortunately, there is no really simple monocarbonyl
compound listed in ‘Table 2 upon which t o test these calculations
concerning the PLNM. The closest approach is 3-thionaphthenone,
whose deprotonation generates a new aromatic ring. Like carbanion
formation from a 0-dicarbonyl compound this may generate a larger
46 JACK HINE
number of smaller changes in bond number and hence a smaller sum

of the squares of the changes in bond number than carbanion
formation by a simple monocarbonyl compound. Nevertheless, the
barrier for 3-thionaphthenone is seen in Fig. 8 t o be somewhat
larger than for a XCOCH,COY of the same basicity. The same type
of calculation gives a larger barrier for nitromethane than for
dinitromethane or a larger barrier for nitroethane than for an
RCH2CH(N02)2,in agreement with what may be seen in Fig. 8.
Such calculations also lead to smaller PLNM effects for three
resonance electron-withdrawing substituents than for two. This fits
the smaller barrier seen for nitroform compared t o dinitromethane.
The barrier for CH(CO,Et), is only slightly smaller than for
XCOCH,COY, but it should be noted that there is a large opposing
steric effect that should tend t o make the CH(C02Et)3 barrier much
larger.
The PLNM, however, does not explain why mononitroalkanes
have the largest barriers of any of the compounds listed and why the
barriers for gem-dinitro compounds are Iarger than for analogous
carbonyl compounds. In fact, the kind of calculations described in
the preceding paragraph give slightly smaller PLNM effects for the
nitro compounds than for their carbonyl analogs. There appears to
be no convincing explanation for the high barriers for carbanion
formation from nitroalkanes. Thus there appears t o be at least one
major factor that affects carbanion formation barriers that has not
been taken into account in rationalizing the data in Table 2 and Figs
7 and 8. This puts the rationalizations that have been made on B
weaker footing than would otherwise be the case.
The data on reactions of acids of known pK, in aqueous solution
listed in Table 2 can be supplemented by data on acids whose pKa
values are not known in aqueous solution or by data in non-aqueous
solutions that also illustrate PLNM effects. For example, tritium is
removed from the triple bonded carbon of phenylacetylene by
hydroxide ions with a rate constant of 270 M -' s-l in water at 25"
(Halevi and Long, 1961). This is larger than most of the rate
constants plotted in Fig. 7 even though phenylacetylene is too weak
an acid for its pK,, which has been estimated to be 21 (McEwen,
1936), to be measured in aqueous solution. The deprotonation of
phenylacetylene, like that of hydrogen cyanide, should be
accompanied by very little change in internal geometry. There are a
number of five-membered-ring heterocycles, including cationic

species, in which protons are removed from sp2 -hybridized carbon
by deuteroxide ions with rate constants as large as l o 6 M-’s - l (not
counting one reported value of 2.6 x lo9 M -’s-l in which the
deprotonation may be due to other bases) (Olofson and Landesberg,
1966; Olofson et al., 1966). It is likely that most of the protons
removed are more weakly acidic than any of the compounds listed in
Table 2, but unfortunately their pK, values do not appear to have
been determined (and would be difficult t o determine). Margolin and
Long (1973) have described evidence that the intrinsic barrier for
deprotonation of chloroform by hydroxide ions, which should be
accompanied by little change in internal geometry, is quite small.
The linear dependence of the basic hydrolysis rate for chloroform on
hydroxide ion concentration up to 0.09 M (Hine, 1950) shows that
chloroform can have a pK, no lower than about 14, since the
hydrolysis proceeds by the reversible formation of the carbanion.
PLNM effects in the base that removes the proton from carbon
have also been suggested. For example, N-methylimidazole has been
found t o remove deuterium from isobutyraldehyde-2-d too slowly,
by about four-fold, to agree with a Br$nsted line based on 3- and
4-substituted pyridines (Hine et al., 1965). This has been attributed
to the change in carbon-nitrogen bond lengths that must accompnay
the formation (20) of the imidazolium ion from the imidazole. No
such change accompanies protonation of a pyridine.
There are many observations in nonaqueous solutions t o which the

PLNM is applicable. Benkeser and coworkers, for example, found
that the metallation of ethylbenzene (Benkeser et al., 1962) or
cumene (Benkeser et al., 1963) with n-amylsodium gives ring-
metallated compounds as the kinetically controlled products but
benzylsodium derivatives as the thermodynamic products. This
shows that there is a smaller intrinsic barrier to the removal of a
proton from sp2 -hybridized carbon, with no significant change in
geometry, than from the benzylic carbon, which is rehybridized from
sp3 to sp2 as the proton is removed.
48 JACK HINE
Proton Transfers to Resonance-Stabilized

Carbanions
Much information about the PLNM has come from the study of
resonance-stabilized carbanions. Many reactions in which such
carbanions are formed were discussed in the preceding section; many
others have been studied largely from the direction of the proton-
ation of the carbanions. One such carbanion protonation is that of
cycIohexadieny1 anion, which takes place most rapidly in the middle
of the 7r system even though this gives the less stable product. This
observation was rationalized in Section 3 in terms of the PLNM.
Similar observations have been made for the protonations of
carbanions that yield a mixture of a$- and 0,y-unsaturated nitriles,
sulfones, and carbonyl compounds. There have been many reports in
which a 0,y-unsaturated ketone is the kinetically controlled product
but the a$-unsaturated ketone is the thermodynamically controlled
product (cf. Bauer, 1914; Dauben and Eastham, 1951; Dauben et al.,
1951; Malhotra and Ringold, 1964). Diethyl cyclopentenylmalonate
is formed more rapidly than the more stable diethyl cyclo-
pentylidenemalonate in the protonation of the common anion (Hugh
and Kon, 1930). The observation that cyclohexenylacetonitrile [ 251
undergoes deuterium exchange in the presence of base much more
rapidly than it is isomerized to the more stable cyclo hgxylidene-
acetonitrile [ 261 shows that the intermediate carbanion yields the
N N N
Ill Ilj 111
C CQ C
I I:
CH
I
v.51 [261
less stable of the two products more rapidly than it yields the more
stable one (Ingold et al., 1936). Similarly, the base-catalyzed
deuterium exchange of the vinylacetate anion is much faster than its
relatively irreversible isomerization to crotonate anion (Ives and
Rydon, 1935). Calculations of the sum of the squares of the changes
in bond numbers that accompany protonation of such catbanions
give results that depend on how heavily valence-bond structures with

the negative charge on oxygen or nitrogen are weighted relative to
structures with the negative charge on carbon. For all plausible
weightings, however, the PLNM favors protonation t o give the
observed P,y-unsaturated products. Concerning reactions of the type
being considered, Ingold has stated “when a proton is supplied b y
acids to the mesomeric anion of weakly ionizing tautomers of
markedly unequal stability, then the tautomer which is most quickly
formed is the thermodynamically least stable” (Catchpole et al.,
1948; Ingold, 1969). Although this generalization, which is essenti-
ally opposite to the product stability principle, fits the reactions
discussed so far in this paragraph, there are a number of cases in
which it fails. For example, the protonation of carbanions of the
type [ 271 has been found t o give P,y-unsaturated sulfones faster than
[R-CH-CH=CHS02R] -
~ 7 1
it gives &,@unsaturatedsulfones, even though, in the case of sulfones,
it is the P,y-unsaturated isomers that are the more stable. Hammond
has suggested that when the transition state in carbanion protonation
comes very early in the reaction the proton may be deposited most
readily near the geometric center of charge (Hammond, 19 55).
Although this is true for the cases described so far, it is not true for
protonation of t-cumyl anions, which also provides another excep-
tion to Ingold’s generalization. Russell observed that t-cumyl-
potassium reacts with DzO, DOAc, and DCI most rapidly at the
exocyclic carbon atom; only for the latter two acids was ring attack
observed, and in each case reaction at the p-position was faster than
at the o-position, even though the latter position is very probably
nearer the center of negative charge in the carbanion (Russell, 1959).
I n this case the I’LNM favors reaction at the p-position over reaction
at the o-position and favors reaction at the exocyclic carbon atom
most of all (Hine, 1966a). Hammond (1955) also suggested that
protonation may take place preferentially at that carbon atom that
bears the largest negative charge, and this point of view has been
adopted b y several other workers (cf. Streitwieser, 1961;
Zimmerman, 1963). The charge densitites obtained for the various
atoms in a carbanion with a delocalized charge depend on the
method of calculation, of course, b u t they d o seem t o b e clearly
50 JACK HINE
correlated with the relative rates of protonation. As Zimmerman has

noted, a simple LCAO MO treatment of a pentadienyl anion assigns
equal charges to the middle and end carbon atoms, as ir. [28], b u t
CN2 -0.333 CH2 -0.316
2 B
HC
BCH --.0333
"$
,CH -0.368
3 //
HC HC,
bCH2 --0.333 'C112 -0.316
[281 ~ 9 3
allowance for differences in bond orders puts a larger charge on the

middle carbon atom, as in [ 291. This then leads t o the expectation
of a greater rate of protonation at the middle of the 7r system in such
pentadienyl anions as the cyclohexadienyl anion. The same con-
clusion can be reached qualitatively from a valence bond resonance
picture. In the first approximation the three contributing structures
[ 301, [ 311 , and [ 321 are weighted equally, giving a hybrid structure
POI [Sll [321
[ 331 with equal charges on the middle and end carbon atoms, but
with larger bond orders for the end carbon-carbon bonds than for
[331
the middle ones. I n the second approximation contributing structure

[31] is seen to have an expected geometry closer t o that of t h e
hybrid [33] than do structures [30] and 1321. Therefore the
contribution of [31] is weighted more heavily than that of [30] and
[ 321 , and now the resulting hybrid has a larger negative charge o n
the middIe carbon atom than on an end carbon. The resulting hybrid
also has carbon-carbon bond orders that differ from each other more
than do those in [33]. This means that the use of such a second
approximation would give larger estimated PLNM effects for proton-
ation of the cyclohexadienyl anion than did the first approximation,
which was described briefly in Section 3.
To the extent to which a delocalized carbanion has a negative
charge on a given carbon atom, the electron pair required for a bond
to a proton is already there. This will favor protonation at that
carbon atom according to that part of the principle of least motion
we have called the principle of least change in electronic configur-
ation. The tendency, described in the preceding paragraph, for the
PLNM t o favor protonation of that atom in a delocalized carbanion
that has the largest negative charge means that the two parts of the
principle of least motion ordinarily favor reaction at the same
position. To learn something about how to weight these two parts of
the principle of least motion relative to each other, we need a case in
which protonation at the most negative carbon does not give the
product favored by the PLNM. There does not appear to be a case in
which all the desired facts are known, but the studies of Knight and
coworkers on the carbanion derived from 6,6-dimethyl-
fulvene seem to provide the closest approach t o such a case (Hine
and Knight, 1970; Knight et al., 1972). Protonation of the iso-
propenylcyclopentadienyl anion [34] can give the 6,6-dimethyl-
fulvene [35] from which the carbanion is ordinarily generated, or
the linearly conjugated triene [36], or the cross conjugated triene
[ 371 , or the deconjugated isomer [38]. A simple Pauling super-
52 JACK HINE
position estimate of the structures of [ 341 - [ 381 gives sums of the

squares of the changes in bond numbers of 0.89, 1.22, 1.55, and 2.22
for the formation of [38], [37], [36], and [35], respectively. Other
methods of estimating the relative magnitudes of the PLNM effects
give about the same results except that in some cases essentially the
same effects were calculated for the formation of [36] and [37].
Thus the PLNM favors formation of [38] most and formation of
[ 351 least. A simple Huckel molecular orbital calculation gives
charge densities of -0.187, -0.182, -0.162, and -0.116 for the
carbon atoms in [34] whose protonation gives [36], [37], [ 381, and
[ 351, respectively. Unfortunately, however, an SCF MO calculation
gave a much different result, with the greatest negative charge on the
carbon whose protonation gives [35]. When the potassium salt of
(341 in diglyme was added to aqueous acetic acid at about 5 O the
observed products were about 63% [ 361, 26% [ 371, and 11%[35].
Unfortunately, [38] is not a known compound and therefore it is
not known whether it would be stable to the reaction conditions;
hence nothing was really learned about the relative rate of formation
of [38]. The results suggest that the intrinsic barriers to the
formation of the three observed isomers stand in the order [36]
< [ 371 < [ 351. This order only partly agrees with that which would
be expected from the PLNM; it may agree better with the principle
of least change in electronic configuration. However, it is possible
that much of the protonation of [34] gives the unstable product
[ 381, which then rearranges by a sigmatropic migration of hydrogen
to give [36].
Lewis Acid-Base Reactions
Some of the structural features associated with apparent PLNM

effects in Lewis acid-base reactions are quite analogous to those
aIready discussed in relation to other types of reactions. An example
of a conversion of sp’-hybridized carbon to sp-hybridized carbon
that is slow relative to analogous conversions of s p 3 - to spZ-carbon
was given in the decomposition of acetonyl radicals to ketene
(Section 4). Among polar reactions it was found that although a plot
of log k vs. log K for the addition of hydroxide ions to the carbonyl
group of aldehydes and ketones gives a fairly smooth line, the rate
constant for addition of hydroxide ions to carbon dioxide is too

small, by about 104-fold, to fall on this line (Hine, 1971). This
deviation is in the direction that would be expected if attack on
carbon dioxide were more hindered (Hine e t al., 1976), but actually
it must be less hindered than attack on most aldehydes and ketones.
The major PLNM effect in this case, of course, arises from the change
from 180" to about 120" bond angles at the electrophilic carbon in
the case of carbon dioxide, compared with a change from about 120'
to about 109.5" in the case of aldehydes and ketones.
The sum of the squares of the changes in bond numbers for
reaction of a cinnamyl cation, like that for reaction of a cinnamyl
anion (Hine, 1966a), may be calculated to b e 129/98 for reaction
with X t o give cinnamyl X but only 45/98 for reaction with X t o give
a-phenyIallyI X For this reason, the fact that cinnamyl cations have
been found in several cases to give a-phenylallyl products more
rapidly than cinnamyl products (Meisenheimer and Link, 1930;
Valkanas e t al., 1963; Pocker, 1959; Goering and Dilgren, 1959,
1960), even though the cinnamyl products are more stable, has been
rationalized in terms of the PLNM. However, the studies of Pocker
and Hill (1969) on reactions involving the 1-phenyl-3-methylallyl
cation [ 391 as an intermediate suggest that other factors may also be
quite important. In this case the PLNM would favor attack on the
phenylated carbon atom (to give [40] ) t o about the same extent as
I+
-
CH3 CH3 CH3
I I I
CH CH CHOH
I/ II I
YH
I
t
CH
/I
6h
II
CHOH CH CH
I I
i
in the reactions of cinnamyl cations, and product stability should not
favor the formation of the conjugated product [41] as much as in
the cinnamyl case (because in the present case the alternative
product [40] has the double bond stabilized by a methyl group that
is absent in the cinnamyl case). Nevertheless, the formation of the
54 JACK HINE
conjugated product is now reported t o be much t h e faster of the two

alternative reactions. A simple HMO calculation would give the same
charge densities for the various atoms in [39] as for the correspond-
ing atoms in a cinnamyl cation and hence would give no charge-
density basis for the difference in behavior of the two species. The
added methyl substituent in [39] would not increase the charge
density on the carbon to which it is attached. However, the
tendency of the electron-donating methyl group to take positive
charge upon itself would increase the charge density in the vicinity of
the carbon to which the methyl is attached. This could aid reaction
at that carbon. A similar situation prevails in the somewhat simpler
case of the 1,l-dimethylallyl cation, which reacts with water t o give
largely 1,l-dimethylallyl alcohol (DeWolfe and Young, 1956), even
though the alternative product 3,3-dimethylallyl alcohol is probably
the more stable product (de la Mare, 1963).
Elimination Reactions
The tendency of the PLNM t o favor trans /3 elimination relative t o

cis elimination in vinyl compounds was described in Section 3. For
0 eliminations that transform single bonds t o double bonds the
dihedral angle between the bonds leading t o the two groups being
eliminated can be anywhere from 0' to 180'. Calculations of the
kind described in Section 3 give an estimated PLNM effect that is
smallest for a dihedral angle of 180°, increases with decreasing
dihedral angle passing through a maximum near go", and decreases t o
a second minimum at O", which is higher than the one at 180' (Hine,
1966b; Tee, 1969). These estimated effects conform with the
experimentally observed tendency of 0 elimination at saturated
atoms to proceed most rapidly when the dihedral angles are near 0'
and 180'.
A 1,4 elimination of X and Y to give a conjugated diene can start
with a cis alkene derivative and give a cisoid diene (21), or it can start
I I
\c/ ci-x \c/k
It - I (21)
/ C L Y ,c*c/
I I
with a trans alkene derivative and give a transoid diene (22). Tee et
al., (1974) have estimated that in the former case the PLNM will
1
' 'c l
X-C\c
II
/
'c-Y
- I
/%C/
I
/C%(-/
(22)
I I
favor the process in which the X and Y groups are cis t o each other
but in the latter case trans elimination will b e preferred. There is
experimental evidence for a preferred cis pathway for eliminations
like those shown in (21) (Cristol et al., 1955; de la Mare et al., 1966),
but unfortunately the preferred stereochemistry of reactions like
(22) does not seem t o have been adequately studied.
The preceding discussion deals only with concerted elimination
reactions, but the PLNM may also b e applied t o stepwise processes.
Thus, just as the PLNM would favor the concerted elimination of HX
from [42] relative to that of HY, it would also favor loss of X- from
1421 "l31
[43], the intermediate carbanion in an ElcB elimination. This is
because less motion is required to bring R, C, C , and Y into
colhearity than R, C, C, and X Observations on diazonium cations,
however, show that the PLNM effect is certainly not the only
important factor, and perhaps not even the most important factor, in
influencing the stereochemistry of the elimination reactions of [ 421.
The syn-isomer has been found t o be the kinetically controlled
product and the anti-isomer the thermodynamically controlled
product in the reactions of aromatic diazonium cations with several
nucleophilic reagents, including hydroxide and cyanide ions (cf.
Smith, 1966). That is, the intrinsic barrier for equilibrating the
diazonium cation with the syn-product [44] is lower than it is for
equilibration with the anti-product [ 451. There should not b e much
56 JACK HINE
difference in the geometric change that the ArN=N group has to

undergo in these two processes. Hence t h e PLNM does n o t seem t o
explain the lower barrier to the formation of the syn-isomer.
However, as Littler (1963) has pointed out, the X group in [44] is
anti t o a pair of unshared electrons on an adjacent atom (just as the
X group in [43] is). That is, the sp2 orbital containing the unshared
pair of electrons is parallel t o the sp2 orbital forming the bond t o X.
This insures maximum overlap throughout the reaction between the
filled orbital and the orbital whose original electron pair is departing
with the leaving X group. I n the anti-isomer [ 451 the X group is in a
position analogous to that of the Y group in [43] ;its bonding orbital
is not parallel to the orbital on the adjacent carbon containing the
unshared pair of electrons.
7. CONCERTED AND STEPWISE REACTION MECHANISMS
Although, in principle, all reactions proceed by a11 possible

mechanisms simultaneously, if there is a pathway by which reaction
is much faster than by any other pathway it is usually referred to as the
reaction mechanism. Hence the problem of predicting t h e mechanism
of a given reaction is one of estimating the relative rates of reaction
by the various conceivable mechanisms. The product stability
principle, both parts of the principle of least motion, and other
generalizations are widely used for this purpose in various specific
cases, but it is worthwhile t o point out that a combination of these
first two principles can be used to obtain a generaIization concerning
trends in the type of reaction mechanism t o be expected.
Let us use the terms “concerted” and “stepwise” t o classify
reaction mechanisms. The way in which changes in bonding are
combined into one step or spread over several steps will determine
how concerted or how stepwise a reaction is. A fully stepwise
reaction will have only one change in bonding per step. Because
changes in bonding involve motion the principle of least motion will
ordinarily favor completely stepwise reactions. However, for many
reactions any completely stepwise mechanism would have to involve
the intermediacy of highly unstable species. That is, the product
stability principle often favors concerted mechanisms. Combining
these two thoughts we may say that the making and breaking of
bonds will tend t o take place in separate steps except where the
formation of a sufficiently unstable intermediate m a y b e avoided b y
combining several such bond changes into one concerted step (Hine,
197 2).
a. CONCLUSION
Discussions of the effect of structure on reactivity can be carried
out more easily and made clearer and more reliable by making
explicit some of the generalizations that have long been used
implicitly. Two of these generalizations are the product stability
principle and the principle of least motion. It is hoped that future
discussions will sharpen the definitions and illuminate the inter-
actions between and limitations of these principles. Further attention
to motion toward the transition state instead of toward the product
may be valuable. The effect of imperfect synchronization of changes
and of non-monotonic changes taking place during a reaction should
also be studied further.
ACKNOWLEDGEMENT
I should like to acknowledge support by the National Science

Foundation of research in the area covered by this chapter.
REFERENCES
Altmann, J. A., Tee, 0. S., and Yates, K. (1976).J. Amer. Chem. SOC. 98,7132.
Asmus, K. D., Henglein, A., Wigger, A., and Beck, G. (1966). Ber. Bunsenges.
Phys. Chem. 70, 756.
Austin, J. A., Levy, D. H., Gottlieb, C. A., and Radford, H. E. (1974). J. Chem.
Phys. 60, 207.
Ballinger, P., and Long, F. A. (1960). J. Amer. Chem. SOC.82, 795.
Barnes, D. J., and Bell, R. P. (1970). Proc. R o y . SOC.A 318, 421.
Bauer, E, (1914). Ann. chim. [9] 1, 342.
Belikov, V. M., Belokon’, Y. N., Kerchemnaya, T. B., and Faleev, N. G. (1968).
Bull. A c a d S c i USSR, Div. Chem. Sci. 317.
Bell, R. P. (1936). &oc. Roy. SOC.A 154, 414.
Bell, R. P. (1941). “Acid-Base Catalysis”, Oxford University Press, London.
Ch. VIII.
Bell, R. P., a n d c o x , B. G. (1971). J. Chem. Soc. (B) 652.
58 JACK HINE
Bell, R. P., and Crooks, J. E. (1965). P / o c . R o y . SOC. A 286, 285.

Bell, R. P., and Goodall, D. M. (1966). Proc. R o y . SOC.A 294, 273.
Benkeser, R. A., Hooz, J., Liston, T. V., and Trevillyan, A. E. (1963).J. Amer.
Chem. SOC.8 5 , 3984.
Benkeser, R. A., Trevillyan, A. E., and Hooz, J. (1962). J. Amer. C h e m SOC.84,
4971.
Benson, S. W. (1968). “Thermochemical Kinetics”. J. Wiley and Sons, New
York pp. 177-182.
Benson, S. W., and O’Neal, H. E (1970). “Kinetic Data on Gas Phase
Unirnolecular Reactions”, Nat. Stand. Ref. Data Ser.-Nat. Bur. Stand. 21,
U.S. Govt. Printing Office, Washington, D.C.
Bordwell, F. G., and Boyle, W. J., Jr. (1972). J. Amer. Chem. SOC.94, 3907.
Bordwell, F. G., Boyle, W. J., Jr., and Yee, K. C. (1970). J. Amer. C h e m SOC.
92, 5926.
Bower, H., McRae, J., and Symons, M. C. R. (1971). J. Chem. SOC.[ A ) 2400.
Br$nsted, J. N., and Pedersen, K. (1924). 2. Phys. Chem. 108, 185.
Brown, A. C. R., and James, D. G. L (1965). Can J. Chem. 4 3 , 660.
Cadman, P., Dodwell, (1, Trotman-Dickenson, A. F., and White, A. J. (1970). J
Chem. SOC. [ A ) 2371.
Carson, 4.S., Laye, P. G., and Morris, H. (1975).J. Chem. Thermodynamics 7,
993.
Catchpole, A. G., Hughes, E. D., and Ingold, C. K. (1948). J. Chem. SOC. 11.
Chaudhri, S. A., and Asmus, K.-D. (1972). J C. S. Faraday Z 385.
Cohen, A. O., andMarcus, R. A. (1968). J. Phys. Chem. 72, 4249.
Corey, E. J., and Kaiser, E. T. (1961). J. Amer. Chem. SOC. 83, 490.
Cox, D. L., Livermore, R. A., and Phillips, L. (1966). J. Chem. SOC. (B) 245.
Cox, J. D., and Pilcher, G . (1970). “Thermochemistry of Organic and Organo-
metallic Compounds.” Academic Press, London. Ch. 5.
Cram, D. J., Nielsen, W. D., and Rickborn, B. (1960). J. A m e r . C h e m SOC. 82,
6415.
Cristol, S. J., Barasch, W., and Tieman, C. H. (1955). J. Amer. C h e m SOC. 77,
583.
Cristol, S. J., Sequeira, R. M., and DePuy, C. H. (1965). J. Amer. C h e m SOC. 87,
4007.
Dauben, W. G., and Eastham, J. F. (1951). J. Amer. Chem. SOC. 73, 4463.
Dauben, W. G., Eastham, J. F., and Micheli, R. A. (1951).J. A m e r . Chem. SOC.
73, 4496.
Davies, M. H. (1974).J. C. S. PerkinII 1018.
de la Mare, P. B. D. (1963). “Molecular Rearrangements” (Ed. P. de Mayo).
Interscience, New York p. 3 7.
de la Mare, P. B. D., Koenigsberger, R., and Lomas, J. S. (1966). J. Chem. SOC.
(B) 834.
DeTar, D. F. (1967). J. A m e r . Chem. SOC.89, 4058.
DeWolfe, R. H., and Young, W. G. (1956). Chem. Rev. 56, 753.
Dobbs, 4.J., Gilbert, B. C., and Norman, R. 0. C. (1971). J. Chem. SOC. [ A )
124.
Dobbs, A. J., Gilbert, B. C., and Norman, R. 0. C. (1972). J. C. S. Perkin11 786.
Doering, W. v. E., and Levy, L. K. (1955). J. Amer. Chem. SOC. 77, 509.
Dronov, V. N., and Tselinsky, I. V. (1970). Reakts. Sposobnost Org. Soedin. 7,
263.
Dronov, V. N., Tselinsky, I. V., and Shokhor, I. N. (1969). Reakts. Sposobnost

Org. Soedin. 6, 948.
East, R. L., and Phillips, L. (1967). J. Chem. SOC. ( A ) 1939.
Egger, K. W., and Cocks, A. T. (1973). Helv. Chim. A c t a 56, 1516.
Ehrenson, S. (1974a). J. Amer. Chem. SOC. 96,3778.
Ehrenson, S. (1974b). J. Amer. Chem. SOC. 96, 3784.
Ehrenson, S. (1976). J. Amer. Chem. SOC. 98, 6081.
Eidinoff, M. L. (1945). J. Amer. Chem. SOC. 67, 2072.
Eigen,M. (1964). Angew. Chem. Intern. Ed. Engl. 3, 1.
Evans, M. G., and Polanyi, M. (1936). Trans. Faraday SOC. 32, 1333.
Evans, M. G., and Polanyi, M. (1938). Trans. Faraday SOC. 34, 22.
Fessenden, R. W. (1967).J. Phys. Chem. 71, 74.
Fessenden, R. W., and Schuler, R. H. (1965). J. Chem. Phys. 43, 2704.
Fukuyama, M., Flanagan, P. W. K., Williams, F. T., Jr., Frainier, L., Miller, S. A.,
ayd Shechter, H. (1970). J. Amer. C h e m SOC.92, 4689.
Goering, H. L., and Dilgren, R. E. (1959). J. Amer. Chem. SOC.81, 2556.
Goering, H. L., and Dilgren, R. E. (1960). J. Amer. Chem. SOC. 82, 5744.
Gray, P., Herod, A. A., and Jones, A. (1971). Chem. Rev. 71, 247.
Griller, D., and Roberts, B. P. (1972). J. C. S. Perkin II 747.
Halevi, E. A., and Long, F. A. (1961).J. Amer. Chem. SOC.83, 2809.
Hammond, G. S. (1955).J. Amer. Chem. SOC. 77, 334.
Hendry, D. G., MU, T., Piszkiewicz, L., Howard, J. A., and Eigenmann, H. K.
(1974). J. Phys. Chem. Ref; Data 3, 937.
Herzberg, G. (1961). Proc. Roy. SOC.A 262, 291.
Hibbert, F. (1973). J. C. S. Perkin IZ 1289.
Hibbert, F., and Long, F. A. (1972). i. Amer. Chem. SOC. 94, 2647.
Hibbert, F., Long, F. A., and Walters, E. A. (1971). J. Amer. Chem. S O C . 93,
2829.
Hine, J. (1950). J. Amer. Chem. SOC. 72, 2438.
Hine, J. (1966a). J. Org. Chem. 31, 1236.
Hine, J. (1966b). J. Amer. Chem. S O C . 88, 5525.
Hine, J. (1971). J Amer. Chem. SOC. 93, 3701.
Hine, J. (1972). J. Amer. Chem. SOC. 94, 5766.
Hine, J. (19 75). “Structural Effects on Equilibria in Organic Chemistry.”
Wiley-Interscience, New York pp. 310- 315.
Hine, J., Green, L. R., Meng, P. C., Jr., and Thiagarajan, V. (1976). J. Org.
C h e m 41,3343.
Hine, J., Houston, J. G., Jensen, J. H., and Mulders, J. (1965). J. Amer. C h e m
SOC. 87, 5050.
Hine, J., and Knight, D. B. (1970). .I. Org. Chem. 35, 3946.
Hine, J., Mahone, L. G., and Liotta, C. L. (1967). J. Amer. Chem. SOC.89,
5911.
Horiuti, J., and Polanyi, M. (1935). Acta Physicochim. U S S R 2, 505.
Huckel, W. (1934). “Theoretische Grundlagen der Organischen Chemie.”
Akademische Verlagsgesellschaft Leipzig. Second edn., pp. 139-142.
Hugh, W. E., and Kon, G. A. R. (1930). J. Chem. SOC. 775.
Ingold, C. K. (1969). “Structure and Mechanism in Organic Chemistry.” Cornell
Univ. Press, Ithaca, New York. Second edn., pp. 826-828.
Ingold, C. K., d e Salas, E., and Wilson, C. L. (1936). J. Chem. SOC.1328.
Ives, D. J. G., and Rydon, H. N. (1935). J. Chem. SOC.1735.
60 JACK HINE
Kaptein, R., Brooken-Zijp, J., and de Kanter, F. J. J. (1972). J. Amer. Chem.

Soc. 94, 6280.
Kato, C., Konaka, S., Iijima, T., and Kimura, M. (1969). Bull. Chem. Soc. Japan
42, 2148.
Kerr, J. A., Smith, A , and Trotman-Dickenson, A. F. (1969).J. Chem. Soc. (A).
1400.
Knight, D. B., Hartless, R. L., and Jarvis, D. A. (1972). J. Org. Chem. 37, 688.
Krusic, P. J., and Bingham, R. C. (1976).J. Amer. Chem. SOC.98, 230.
Leffler, J. E. (1953). Science 117, 340.
Leffler, J. E., and Grunwald, E. (1963). “Rates and Equilibria of Organic
Reactions.” J. Wiley and Sons, New York. p. 26.
Leggett, C., and Thynne, J. C. J. (1970). J. Chem. Soc. (A). 1188.
Littler, J. S. (1963). Trans. Faraday Soc. 59, 2296.
McEwen, W. K. (1936). J. Amer. Chem. SOC. 58, 1124.
Malhotra, S. K., and Ringold, H. J. (1964). J. A m e r Chem. SOC.86, 1997.
Marcus, R. A. (1968).J. Phys. Chem. 72,891.
Margolin, Z.,and Long, F. A. (1973). J. Amer. Chem. SOC.95, 2757.
Meisenheimer, J., and Link, J. (1930). Justus Liebigs A n n . Chem. 479, 211.
Motes, J. M., and Walborsky, H. M. (1970). J. Amer. Chem. Soc. 92, 3697.
Olofson, R. A., and Landesberg, J. M. (1966). J Amer. Chem. Soc. 88, 4263.
Olofson, R. A., Landesberg, J. M., Houk, K. N., andMichelman, J. S. (1966). J.
Amer. Chem. SOC. 88, 4265.
Pauling, L. (1960). “Nature of the Chemical Bond.” Cornell University Press,
Ithaca, New York. 3rd edn., pp. 235- 240.
Pearson, R. G., and Dillon, R. L (1953). J. Amer. Chem. S O C . 75, 2439.
Perkins, M. J., and Roberts, B. P. (1974).J. C. S. Perkin I1 297.
Pocker, Y. (1959). Chem. I d . 195.
Pocker, Y., and HiU, M. J. (1969).J. Amer. Chem. SOC.91, 3243.
Pryor, W. A., Fuller, D. L., and Stanley, J. P. (1972). J. Amer. C h e m SOC.94,
1632.
Ratajczak, E., and Trotman-Dickenson, A F. (19 69). “Supplementary Tables of
Bimolecular Gas Reactions.” Univ. Wales Inst. Sci. Tech., Cardiff.
Rice, F. O., and Teller, E. (1938). J. Chem. Phys. 6, 489.
Kubaszewska, W., and Grabowski, 2. R. (1969). Tetrahedron 25, 2807.
Russell,G. A. (1959). J. Amer. Chem. SOC.81, 2017.
Schuh, H., Hamilton, E. J., Jr., Paul, H., and Fischer, H. (1974). Helv. C h i m
Acta 57, 2011.
Shimanouchi, T. (1972). “Tables of Molecular Vibrational Frequencies.”
Consolidated Vol. I, Nat. Stand. Ref. Data Ser., Nat. Bur. Stand. 39, U.S.
Govt. Printing Office, Washington, D.C.
Smith, P. A. S. (1966). “Open-Chain Nitrogen Compounds.” Benjamin, New
York. Vol. 11, Ch. 11.
Solly, R. K , and Benson, S. W. (1971).J. Amer. Chem. SOC. 93, 2127.
Streitwieser, A., Jr. (1961). “Molecular Orbital Theory for Organic Chemists.” J.
Wiley and Sons, New York pp. 379, 418.
Sutton, L. E. (1958). “Tables of Interatomic Distances and Configuration in
Molecules and Ions.” Special Publication No. 11, The Chemical Society,
London. p. M67.
Tan, L. Y., Winer, A. M., and Pimentel, G. C. (1972). J. Chem. Phys. 57, 4028.
Tee, 0. S. (1969). J. Amer. Chem. SOC.91, 7144.
Tee, 0. S., and Yates, K. (1972). J. Amer. Chem. SOC.94, 3074.

Tee, 0. S., Altmann, J. A., and Yates, K. (1974).J. Amer. Chem. SOC.96, 3141.
Theilacker, W., and Schmid, W. (1950). Justus Liebigs A n n Chem 570, 15.
Theilacker, W., and Wegner, E. (1963). Justus Liebigs Ann. Chem. 664, 125.
Traetteberg, M. (1964). J. Amer. Chem. SOC.86, 4265.
Trotman-Dickenson, A. F., and Milne, G. S. (1967). “Tables of Bimolecular Gas
Reactions.” Nat. Stand. Ref. Data Ser.-Nat. Bur. Stand. 9, U . S Govt. Printing
Office, Washington, D.C.
Trotman-Dickenson, A. F., and Steacie, E. W. R. (1951). J. Chem. Phys. 19,
169.
Turnbull, D., and Maron, S. H. (1943). J. Amer. Chem. SOC.65, 212.
Valkanas, G., Waight, E. S., and Weinstock, M. (1963). J. Chem SOC.4248.
van Wijnen, W. T., Steinberg, H., and de Boer, T. J. (1972). Tetrahedron 28,
5423.
Walborsky, H. M., and Motes, J. M. (1970). J. Amer. Chem. SOC. 92, 2445.
Watkins, K. W., and Thompson, W. W. (1973). Int. J. C h e m Kinet 5, 791.
Wheland, G. W. (1960). “Advanced Organic Chemistry.” J. Wiley and Sons, New
York. 3rd edn., S e c 2.3.
Wood, D. E., Williams, L. F., Sprecher, R. F., and Lathan, W. A. (1972). J.
Amer. Chem SOC. 94,6241.
Woodward, R. B., and Hoffmann, R. (1970). “The Conservation of Orbital
Symmetry.” Academic Press, New York
Zimmerman, H. E. (1963). “Molecular Rearrangements.” (ed. P. d e Mayo). Part
I, Interscience, New York. Ch. 6.

The Principle of Least Nuclear Motion

Uploaded by

Copyright:

Available Formats

The Principle of Least Nuclear Motion

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Principle of Least Nuclear Motion

Uploaded by

Copyright:

Available Formats

The Principle of Least Nuclear Motion

Department of Chemistry, The Ohio State University,

Organic chemists ordinarily think about structural effects on

never intermediate between that of the reactant(s) and product(s).

' This word makes the grnrralization infallible.

with the principle of least change in electronic configuration.

Never, in t h e author’s experience.

2. THE INTERSECTING MORSE CURVE APPROACH

What may be called the intersecting Morse curve approach has

Let us now discuss the effect of structure on reactivity by

forms a stronger bond t o hydrogen, i.e., Do (HY,) >Do (HY, ).

3 . PREDICTION OF THE MAGNITUDE OF PLNM EFFECTS

Although most reactions may be thought of as being slowed by

plausible that these changes in molecular geometry are large enough

1,4-~yclohexadiene,but the treatment is rather directly relevant to

on the reactions ( 5 ) of pentadienyl species may increase the activa-

[11 [21 [31

(2 - 1$)2+ (14 - 1)2 + (14 -1)2 + (2 - 18)’ or $ in the squares of

Figure 4. Changes in geometry in the dehydrochlorination of vinyl chloride (-) to

The preceding quick method of assessing PLNM effects will be

squares of the resulting displacements of the nuclei. The dashed line

dehydrochlorination of 2-chloropropene to give propyne were

interactions between the two angle bends in a given molecule. This is

vibration that tends to give the type of distorted acetylene being

discussion in terms of motion toward the products. Such a treatment

4. FREE RADICAL REACTIONS

Hydrogen Trunsfer Reactions

In order to explain structural effects on reactivity convincingly in

slightly more than do some of the points for saturated hydrocarbons.

that there should also be a PLNbI effect on the formation o f at least

radicals, on the other hand, may be slowed somewhat by PLNM

Although no curvature seems to be required by the points upon

abstraction of a hydrogen atom from toluene by an allyl radical,

Radical Decomposition Reactions

Data on reactions, such as those shown in equations (8)-(11),in

radical by cleavage of a carbon-carbon or carbon- oxygen bond have

Rates of Decomposition of Radicalsa

Reaction Phase logAb EC log kb*d AHc

n-Pr. + M e . + CHzzCH2 Gas 13.6 33.1 -4.80 25.8

a Data from Benson and O'Neal (1970) unless otherwise noted.

Schuh et al., 1974.

acyl radicals, AH of formation was based on an enthalpy of

The data in Table 1 include log k values at 120°, a temperature

and "downhill" reactions have transition states that resemble the

change in internal geometry that has no parallel in the decarbonyl-

it should use a transition state [8] in which the plane of the

the acetonyl radical is a rr radical, as is the somewhat similar radical

increase the concentration of such radicals and hence the rate of

the carbon atom to which oxygen is attached must change from

the bond to be broken anti to the orbital containing the unpaired

Multicenter reactions, like free radical reactions, are relatively well

ment of cyclopropyl cations to allyl cations, the disrotatory pathway

PLNM (Tee and Yates, 1972; Ehrenson, 1974b). Disrotatory re-

Although it is not a multicenter reaction, this process is so closely related t o the

would give the highly strained carbonium ion [ 1 7 ] , is much slower,

a disrotatory fashion. When rotation around the single bonds of the

treatment of nuclear motions would show the disrotatory process to

process, with a chair-like transition state, or by a disrotatory process,

effects that are often so important in poorly ion-solvating media we

Proton Transfers f r o m Carbon