вход по аккаунту


Memory Effects and Stereochemistry in Multiple Carbonium Ion Rearrangements.

код для вставкиСкачать
or a r e only “transferred” in a transition complex. Evidence
against t h e formation of free carbene in t h e reaction of
TAE2+ with strong alkali is provided by t h e unsuccessful
attempt t o convert a mixture of t h e dibromide of T D A E a n d
tetramorpholinoethylene into a n unsymmetrical reaction
product, 1,l-bis(dimethylamino)-2,2-morpholinoethyleneby
t h e action of sodium hydroxide solution [cf. reaction
(S)] [1011.
Though diaminocarbenes have considerable resonance
stabilization, it is not sufficient for the cleavage of TAE
into two carbenes. Stable carbenes are evidently formed only if both the electron vacancy and the free electron pair on the carbene carbon are chemically
saturated 11021:
.r‘ .r‘
x ~ - c - Yt+
tetraaminoethylenes deserve (e.g. for the qualitative
and quantitative detection of oxygen and of electron
acceptors with TDAE, the generation of electricity
with TDAE, etc.), a detailed investigation of the electronic structure of TAE, TAE+, TAE*+, and TAE*,
as well as of electron donor-acceptor complexes of
tetraaminoethylene, should be of great value to the
theory of x systems. However, the chemistry of tetraaminoethylenes is only a small part of the chemistiy
of peramino compounds of carbon. The preparation or investigation of compounds such as
(R2N)zC. (R2N)4C L1051, ( R z N ) ~ C - C ( N R ~ ) ~and
(RzN)C=C(NRz) 1651 may be eagerly awaited.
I am indebted to A. N. Flechter, C. A. Heller, D. Lemal,
and H. E. Winberg for valuable exchanges of informaHowever, it is not yet certain whether compounds of
tion and for the communication of unpublished results.
this type (e.g. R2N@=C=B@R2, R3P==C==PR3, Ishould also like to express my gratitude to J . W. Buchlrr
R2C--C:-=CR2)can still be regarded as carbenes [1031.
for his cooperation.
Received: November 16, 1967
[A 663 IEI
German version: Angew. Chem. 80, 809 (1968)
Translated by Express Translation Service L o n d o n
6. Outlook
Tetraarninoethylenes are still a relatively new topic.
While many surprising results for ethylene chemistry
have appeared in the past six years, further surprises
may be expected from these reactive compounds in the
future. Apart from the practical interest that the
W . Carpenter, unpublished
I1021 Cf. also R. A . Olofson and J. M . Landesbeig, J. Amer.
chern. SOC. 88, 4263 (1966).
[lo31 In the case of the ethylene (C6H$)2C=C(sC&~)2, Seebach [lo41 recently observed cleavage to a carbene.
11041 D . Seebach, Angew. Chem. 79, 469 (1967); Angew. Chem.
internat. Edit. 6, 443 (1967).
11051 H. Weingarten, J. Amer. chern. SOC.88, 2885 (1966).
Memory Effects and Stereochemistry
in Multiple Carbonium Ion Rearrangements I**]
The stereochemical configuration at the initial site of heterolysis in a carbonium ion process in solution controls the specificity of migration two steps later in many sequential
multiple rearrangements. These “memory effects” are attributable to the presence on the
potential energy surface of small, hitherto undetectable minima. A variety of structural,
stereochemical, and isotopic marking techniques are used to demonstrate that memory effects are a general feature of the ring-expansion route to bicyclic carbonium ions. The
intermediates formed in these reactions, although necessarily symmetrical or quasi-symmetrical at equilibrium, act unsymmetrically. The memory effect uses the rate of an intramolecular carbonium ion rearrangement as a calibration with which to compare the rate
of symmetrization. In this way, the system .functions as a sensor of sub-species.
1. Introduction
Much of our understand,ng of organic
mechanism derives historically from the search for
solutions to intriguing puzzles of stereochemistry.
The study of the phenomena we call “memory effects”
is now at the point where experiment has progressed
well beyond
theory, and new principles therefore must emerge.
Angew. Chem. internot. Edit.
VoI. 7 (1968) J No. I0
Stereochemistry controls the course of reaction not
only when bond-making and bond-breaking are confined to a single carbon atomic center as, for example,
in the familiar inversions that accompany nucleo-
A. Berson
Department of Chemistry,
University of Wisconsin, Madison, Wisconsin (USA)
Part of this material was presented in a lecture a t the Second
EUCHEM Conference on Stereochemistry (Burgenstock,
Switzerland, May 1966).
philic substitution[ll, but also when the process involves a center vicinal to that at which bond-breaking
is initiated, as in the neighboring group effect 121.
In multiple carbonium ion rearrangements, such control can be exercised over sites of reaction even more
remote. We call these phenomena memory effects and
recognize their existence whenever two requirements
are met: (i) a bond to a configurationally specifiable
center of the reactant is broken, initiating a sequence
of migrations, and (ii) the configuration of this center
determines which of the potential migrating groups actually moves at each stage of the multi-step process.
Superficially, the specificity of migration seems to be
based on “memory” in that the set of instructions
(or to pursue the biological parallel, the “engram”)
embodied in the original configuration guides the
sequence even after one of the bonds at the initiating
site no longer exists.
From the purely descriptive point of view, the study of
these effects opens up a broad field of new rearrangements and intriguing stereochemical phenomena. But
ultimately we are concerned with mechanism, that is,
with the intimate structure of the energy-reaction coordinate surface. Memory effects serve as probes for
the detection of small potential energy pockets which
are impalpable by other means. The structures of the
subspecies that occupy these energy minima and the
precise way in which the original stereochemical “engram” is processed are subjects that have not yet passed
much beyond the phenomenological stage. Before
examining these questions, however, we may find it
useful to discuss the bases for the characteristic stereochemical phenomena of three-center displacements and
neighboring group effects. Then, after a description of
several experiments demonstrating memory effects,
their fundamental origin can be speculated upon.
2. Theoretical Basis of Stereochemical ControI in
Three-Center Displacements and Neighboring
G r o u p Displacements
Birnolecular nucleophilic (SN2) displacement occurs
with Walden inversion 131, even in the face of strongly
countervailing electrostatic forces as in onium salt
decompositions, where an anionic nucleophile attacks
a cationic substrate 141. In sharp contrast, electrophilic
substitutions, including some specifically believed to
be bimolecular (SE2) “1, frequently give retention of
[11 For historical and critical reviews, see a) C. K. Ingold Structure and Mechanism in Organic Chemistry. Cornell University
Press, Ithaca 1953; b) A . Streitwieser j r . : Solvolytic Displacement Reactions. McCraw-Hill, New York 1962.
[2] For a review, see S. Winstein, Bull. SOC.chim. France (5) 18,
55C (1951).
131 C. K . Ingold Structure and Mechanism in Organic Chemistry. Cornell University Press, Ithaca 1953, p. 380.
[4] a) S . H . Harvey, P. A . T. Hoye, E. D . Hughes, and Sir Christopher Ingold, J. chem. SOC.(London) 1960, 800; b) H. R. Snyder
and J . H . Brewster, J. Amer. chem. SOC.71, 29 (1949); c) J . Read
and J. Walker, J. chem. SOC.(London) 1934, 308; d) Inversion
also occurs in oniurn salt formation, S. Siege1 and A . F. Graeje,
3. Amer. chem. SOC.75, 4521 (1953).
configuration 1 5 ~ 6 1 . This extraordinary pattern seems
to correlate with the number of electrons involved in
the reacting bonds. One therefore is stimulated to
seek a common quantum-mechanical basis for the
two kinds of result. A treatment of the transition state
by molecular orbital methods 181 provides a simple
theoretical picture [9,101.
The substitution reaction can be approached naively as
a three-orbital problem; the reacting centers form an
array in which the central atom (B) and its two attached groups - the substituting agent (A) and the
leaving group (C) - either are disposed linearly
(A-B-C) or else form an acute angle. These correspond to the transition states of reactions giving,
respectively, configurational inversion or retention at
B. Regardless of what orbitals are chosen for the basis
set, the idealized case will be one with equal bond
integrals (p) between B and both A and C. For the
linear, inversion-producing relationship of these centers, one can assume a zero A-C integral, which
leads to an allyl-like set of energy levels: one bonding,
one non-bonding, and one anti-bonding fib, 111.
It is not obvious why this arrangement is so favorable
in the nucleophilic case (SN2) except by comparison
with the set derived from the alternative geometry,
in which the A-B-C angle is acute, corresponding to
front-side substitution. This inevitably will produce a
non-zero bond integral ( k p ) between A and C. Inclusion of this off-diagonal element in the secular determinant produces a cyclopropenyl-like arrangement
of energy levels, one bonding and two anti-bonding.
For any given value of k, the energy of the bonding orbital of the primitive linear case is lowered by an increment n m b , but that of the non-bonding level is
raised (to anti-bonding, see Fig. 1) by a greater
amount Amn. Therefore, the two-electron (electrophilic) system, in which only the bonding level is occupied, will prefer the triangular geometry. In the
four-electron (nucleophilic) case, the stabilization of
the bonding level produced by the finite A-C interaction in the triangular system is overbalanced because the anti-bonding level is doubly occupied. The
linear, inversion-producing geometry therefore is
preferred for nucleophilic displacement 1121. Of course,
[S] F. R. Jensen, L. D . Whipple, D . K. Wedegaertner, and J. A.
Landgrebe, J. Amer. chem. SOC.81, 1262 (1959).
[6] a) S . Winstein and T. G. Traylor, J. Amer. chem. SOC. 78,
2591 (1956); b) S. Winstein, T. G. Traylor, and C. S . Garner,
ibid. 77, 3741 (1955).
[7] a) Sir Christopher Ingold, Rec. chem. Progress (KresgeHooker Sci. Lib.) 25, 145 (1964); Helv. chim. Acta 47, 1191
(1964); b) see also H . B. Charman, E. D. Hughes, and Sir
Christopher Ingold, 3. chern. SOC.(London) 1959, 2523, 2530
[ S ] The argument given here is an expansion of one briefly
presented elsewhere [9, 101.
[9] E . L. Eliel, N . L . Allinger, S. J . Angyal, and G . A. Morrison:
Conformational Analysis. Interscience, New York 1965, p. 483.
[lo] See also J . A. Berson, J. H. Hammons, A . W . McRowe,
R. G . Bergman, A. Remanick, and D . Houston, J. Amer. chem.
SOC.89, 2590 (1967).
1111 H . Jaffe, J. chem. Physics 21, 1618 (1953).
I121 This treatment predicts that free-radical three-center displacements should show a rather low degree of stereospecificity,
since the difference in energy between the linear and triangular
systems in the three-electron case is small.
Angew. Chem. internat. Edit.
Vol. 7 (1968)
No. I0
factors such as electrostatic forces, steric repulsions,
and solvation effects, which are not taken into account by the simple LCAO approach, may dominate
in some situations, but the basic quantum mechanical
influences always will be at work.
group. As is borne o u t 1101 by simple LCAO considerations, the preference for rear-side attack shown by external nucleophiles also should extend to such cases.
This predicts that in the multiple rearrangement of
(3), shift of R3 would be followed (or accompanied)
by shift of the farther (R1) rather than that of the
nearer (R*) group. The preference would hold regardless of whether R1 and R3 moved in discrete steps [with
( 4 ) as an intermediate] or simultaneously [one fivecenter transition state with structure (6)].
Fig. I . Energy levels in units of B of the three-center system A-B-C a s
a function of the b o n d integral ( k p ) between A and C. I n general,
> l>mbl.
The same destabilizing effect of a front-side interaction
occurs in the case of nucleophilic attack on a nonclassical carbonium ion or on the bridged intermediates responsible for the characteristic stereochemistry
of neighboring group effects. These may be treated [lo]
as four-electron, four-center systems in which the
nucleophile ()’
the bridging atom (R) from a
terminus of the bridge with either inversion [as in (111
or retention [as in (211.
The bond integral ( k p ) in case (2) will be greater than
in case ( I ) , and just as in the three-center system, the
destabilization of the lowest anti-bonding level is
greater than the stabilization of the bonding level.
Even when nonclassical bonding is absent and hence
front-side migration can occur readily, other influences may still enforce strong rear-side selectivity in
the rearrangement steps that follow the initiating one.
Conformational effects are potent factors of this kind
and may well be influential in the remarkable multistep “backbone” rearrangements of polycyclic terpenoids 1131, in which only the extra-annular groups
An example involving several such steps is that of the friedelany[ cation (71, which rearranges to a cation ( 8 ) of the 9amvrin series. T h e selectivitv of these transformations does
not necessarily signify a “synchronous” or “concerted” multiple migration viri a single transition state ( 6 ) o r a step-wise
process via discrete nonclassical intermediates (41, but may
also result from a series of classical ions. A t each stage of
such a process, the extra-annular group (H or CH3) at t h e
position vicinal t o t h e electron-deficient center is bound
by a n orbital whose axis is nearly parallel to, and hence in
excellent position t o interact with, that of the empty porbital, whereas the corresponding orbital binding a ring
carbon is badly displaced from this favorable orientation.
These relationships are shown in (91, which is a n excision of
part of t h e “backbone” of (7). Migration of CH3 from C-5
t o C-4 leaves the C-10 hydrogen in a more favorable posi-
3. Multiple Rearrangements
The role of Y in nucleophilic attack on bridged ions
can be assumed, in a formal sense, by some internal
nucleophile, such as a migrating hydrogen or alkyl
1131 For a review, see inter niia J . F. King and P . de Mayo in
P. de Mnyo: Molecular Rearrangements. Interscience, New York
1963, Part Two, Chapter 13.
Angew. Chem. internnt. Edit. 1 VoI. 7 (1968) / No. I0
tion than either C-1 or C-9 for a subsequent migration. The
rearrangements of the extra-annular groups therefore are
expected to be preferred, regardless of whether t h e intermediates are classical or nonclassical, and the observed
memory effects are explicable without recourse to any special
theoretical formulation. The truly novel aspects of memory
effects emerge only when the commonplace stereoelectronic
and conformational features are absent.
rearrangement. Neither of these is in a good position
to interact with the p-orbital left vacant if the first
ring-expanding rearrangement leads to a classical
cation, nor can either readily attack the rear side of a
mesomeric bridged ion if the ring-expansion leads to a
nonclassical intermediate. Such systems therefore put
forward a new problem: Is memory preserved despite
the attenuation of the normally dominant stereoelectronic factors? The following examples show that
it is.
5. Ring-Expansions of 2-Norbornylmethyl
4. Forced Migration of Improperly Situated
Incorporation of the fragment (9) in a bridged bicyclic
system, e.g., (10) or ( I I ) , prohibits migration of the
bridgehead hydrogen [at C-4 of (10) and C-1 of ( I I ) ,
Ring-expansion route
(12), R'-X
The geometric restrictions embodied in structure ( I 1 )
are met in 2-endo- and 2-exo-norbornylmethyl
systems [(12) and (13)]. Studies by Alder and his coworkers 1141 show that the deaminative ring-expansion
of the endo-amine gives a different set of products than
that obtained from the exo-amine. Later work on the
deaminations and on solvoryses of arenesulfonates
elucidates further mechanistic details and leads to a
forrnulationt15.161 which is shown in Scheme 1. The
systems derived from (12) and (13) are prototypes of
a number of other examples. Scheme 2 summarizes
the common phenomena of memory effects in sym-
(16), XI"
(lP), M'" -X
(IS), M'" -X
(21). M'Ia
Scheme I . A specific example of memory effect.
which correspond stereoelectronicallY to the most
favorable group, the C-10 hydrogen of (9)j because
of the inability of the bridgehead carbon to
trigonal configuration. One of the ring carbons thereby
is forced to migrate in the second step of the multiple
1141 a) K . Alder and E. Windemuth, Ber. dtsch. chem. Ges. 71,
2404 (1938); b) K. Alder, H. Krieger, and H . Weiss, Chem. Ber.
8 8 , 1 4 4 (1955); c) K. Alder and R . Reubke, ibid. 91, 1525 (1958).
[151 J. A. Berson and p . Reynolds- Warnhuff,J. Amer. chern. SOC.
84, 682 (1962); 86, 595 (1964).
flsl J . A . Berson and D , Willner, J, Amer. them, sot. 84, 575
(1962); 86,609 (1964).
Angew. Chem. internat. Edit. 1 Val. 7 (1968) / No. I0
bolic form. The dashed arrows represent reactions
observed in some but not all of the systems studied.
To aid in comparisons, the reactants and unstable
entities of Scheme 1 are labeled both with ordinal
numbers and with boldface letters, which are also
used in Scheme 2.
Ring-expansion route
Scheme 2. Generalized scheme of memory effects.
octyl, or 2-bicyclo[2.2.2]octy1 precursors [(20) [201,
(21) [181, or (22) [18,19J].
From the left in Schemes 1 and 2, the entry is by ringexpansion. In the specific case of Scheme 1, the firstformed ring-expanded cations (15) and (14) are conveniently represented as boat-chair conformational
isomers which interconvert by way of a transition
state (I5a) ROO. In other cases, however, Roo is a
quasisymmetric or truly symmetric intermediate from
which products may be formed directly. This path is
indicated in Scheme 2. In the generalized scheme,
transformations of R’a to M’O and of R”@ to M”@
occur by second rearrangements or bond delocalization processes, which compete with crossover (interconversion of R’@ and R”@).
The competition between intramolecular trapping and
crossover, measured by the ratio of rates of the reactions R’O
M’@ ( k l ) and R’@ + R O O ( k z ) gives a
quantitative estimate of the magnitude of the memory
effect, that is, the efficiency with which the system follows the “instructions” embodied in the specific
stereochemistry R’-X. Ideally, one would prefer that
the only crossover occur at the stage of the firstformed ring-expanded cations R’@ and R”@. Then
the desired ratio of rates can be deduced from the
product distribution data. In this kind of system, those
cations which have suffered two successive rearrangements remain on a given side of the dotted line, that
is, M’@ -+ M ” @ interconversion is negligible, and all
of the selectivity inherent in the ratio of rates kl/kz remains in the observed products.
The conditions for the ideal situation are that R’@ + M ” @ be
The first cationic intermediate (14) from the endo ring
expansion has a chair cyclohexane ring, whereas the
cation (15) from the exo has a boat. If conformational
interconversion of (14) and (15) were fast, crossover
between the two systems would occur, and the products would be the same or nearly so from either endo
or exo starting material [(12) and (13) respectively].
In fact, however, crossover is not extensive, and the
conformationally isomeric cations (14) and ( 1 5 ) are
trapped, both by external nucleophiles and by further
intramolecular bond rearrangements [(14) -+ (16)
and (15)
The specificities of the second rearrangement steps
constitute memory effects. Although the stereoelectronic alignments of the potential migrating
bonds, (C-l)-(C-8) and (C-l)-(C-7), are not particularly good in either (14) or ( 1 5 ) , models suggest that
the (C-l)-(C-8) bond in (14) and the (C-l)-(C-7)
bond in (15) would be favored for further rearrangement 115,161, in accord with the observations.
Two kinds of access into the Schemes 1 and 2 may be
recognized. From the right one forms cationic intermediates by interaction of the electron-deficient center
with a nearby x- or a-system. This mode of entry
provides no opportunity for a memory effect to
operate, since only one rearrangement or bonddelocalization step is involved, but it permits the
generation by “conventional” routes of the cations
(M’@ and M”@) which are the products of the fwostep path in the ring-expansion route. The behavior of
the cations formed by the x- or a-routes serves as a
standard of comparison for those formed by the ringexpansion route. For example, in Scheme 1, (16) is
generated from 4-cycloheptenylmethyl or 2-endo-bicyclo[3.2.l]octyl precursors [(18)
or (19) 11811, and
(17) from 3-cyclohexenylethyl, 2-exo-bicyclo[3.2.1][17] G . Le Ny, C . R. hebd. Seances Acad. Sci. 251, 1526 (1960).
[lS] a) H. L. Goering and M . F. Sloan, J. Amer. chern. SOC. 85,
1397 (1963); b) H. L. Goering and G . Fickes, unpublished work;
G . Fickes, Ph.D. Dissertation, University of Wisconsin, 1964.
1191 a) A. A. Youssef, M . E. Baum, and H . M . Walborsky, J.
Amer. chern. SOC.81, 4709 (1959); b) ibid. 83, 988 (1961);
c) H. M . Walborsky, J. Webb, and C. G . Pin, 3. org. Chemistry
28, 3214 (1964).
Angew. Chem. internut. Edit. 1 Vol. 7 (1968) No. I0
essentially irreversible and that direct as well as indirect interconversion of M’O and M ” @ be very slow relative to capture of the cations as final products. Although a few systems
meet the ideal requirements, many of them form doubly rearranged cations (M’e, M”@) that drift across the barrier to
some extent. The product ratios then give too pessimistic a n
estimate of the memory effect. This is not a n insuperable
difficulty, and corrections for it can be made from independent studies of the properties of M’O and M”@ generated in
the TC- or a-route entry (from the right side of Scheme 2).
The characteristic stereochemical feature of the memory effect is not given by the symbolism of Scheme 2
and must be stated separately. In all cases, the preferred migrating group in the second rearrangement
step is the one that was originally remote from the site
of the leaving group in the starting material [e.g.,
C-7 rather than C-6 in the endo-norbornylmethylsystem (12)].
6. Persistence of the Memory Effect in a System of
Low Conformational Stability
If the memory effect were solely caused by a conformational barrier inhibiting rapid interconversion of the
isomeric cations R’@ and R”@(Scheme l), it should
diminish greatly or completely disappear in systems so
rigid that “conformations” represent very shallow
[20] S. Winstein and P. Carrer, J. Amer. chem. SOC.83, 4485
energy minima. The cations (25) and (26) produced in
the ring-expansions of syn- and anti-norbornenylmethyl derivatives [21,221 (23) and (24) are drawn here
in “twisted” representations merely as a graphic
device to distinguish them from the quasi-symmetric
cation (27) ROO. Although physical evidence on the
conformations of bicyclo[2.2.2]octenes is sparse,
molecular models of this ring system are rather resistant to twisting distortions, and it seems unlikely that
there can be a large barrier separating the two cations
from each other or from the quasi-symmetric cation
(27). Nevertheless, memory is preserved.
The products in this system are derived from cations G
(R = H) and L, which do not interconvert appreciably
under the conditions and thus constitute an ideal pair
for the preservation of whatever memory inheres in
(25) and (26). Cations G and L already had been encountered in solvolyses, G from endo- and L from
exo-2-bicyclo[2.2.2]oct-5-en-2-y1 substrate [(28) [231
and (29) 12411. Cation G generated from optically active (28) gives largely the completely racemic alcohol
bicyclo[3.2.l]oct-3-en-2-ol(30) derived from attack at
equal rates at both sites of the symmetrical allylic
cation system (27) 1231. That the product-forming cation from the ring-expansion starting with (23) is also
G is demonstrated in the deamination of syn-[7-D]-7-
Table I. Some memory effect selectivities [o].
A. Cations generated from syn-7-norbornenyl substrates
G (%I
[c, dl
Other reagents
NaOAc, 0.068
NaOAc, 0.068 iLiC104, 0. I00
AgC104, 0.123
urea, 0.14
NaOAc. 0.055
NaOAc, 0.055
LiC104, 0.100
AgC104, 0.123 - t
urea, 0.14
AgCI04, 0.123
urea, 0.14
N a N 0 2 , excess
B. Cations generated from anti-7-norbornenyl substrates
NaOAc, 0.068
NaOAc, 0.069 iLiC104, 0.100
NaOAc, 0.068
LiCIOI, 0.250
AgC104, 0.122-k
urea, 0.14
AgC104, 0.122
urea, 0.14
3.6 [gl
4.5 [gI
3.5 [g]
C. Cations generated from 7-norbornylmethyl substrates
T ( “C
Other reagents
[211 J. A. Berson and J. J . Gajewski, J. Amer. chem. SOC. 86,
5020 (1964).
I221 Cf. also R . K . Bly and R. S . Bly, J. org. Chemistry 31, 1577
[23] H. L. Goering and M. F. Sloan, J. Amer. chem. SOC.83,1992
(1961); H. L. Goering, R. W . Greiner, and M. F. Sloan, ibid. 83,
1391 (1961); H. L. Goering and D . L. Towns, ibid. 85,2295 (1963).
[24] a) N. A. Le Be/ and J. E. Huber, J. Amer. chem. SOC.85,
3193 (1963); b) cf. R. R. Fraser and S. O’Farrell, Tetrahedron
Letters 1962, 1143.
NaOAc, 0.06
1 20
NaOAc, 0.06
2.5 [jl
N a N 0 2 , excess
6.6 111
NaNOz, excess
5.7 [iI
PVA [k]
H 2 0 [ml
H2O [ml
2.0, 2.0
[d, il
2.3 [dl
2.0 [il
NaN02, excess
1.6, 1.9
tn, j l
[a] 40 % water, 60 % dioxane (viv). [bl Ratio of products of the G L). [dl Corrected for
series to those of the L. [cl 100 ( G - L)/(G
drift (see text). [el Ratio of products of the L-series to those of the G .
[f] 100 (L - G)/(L T G ) . [gl This value may be as much as 0.5 unit too
low because of difficulty in analysis for a minor L-component. [h] Ratio
of products derived from migration of the R2 vs Rl-labeled bridges.
[i] Duplicate runs. [j] A correction for drift is not yet available for this
result, which is therefore a minimum measure of the true selectivity.
polyvinyl acetate sufficient to cause a 15-fold increase in
[k] PVA
viscosity. [I] 90 % formic acid, 10 % pyridine (vlv). [ml 94 % water,
6 % acetic acid (v/v). [n] Duplicate combustion and falling drop analyses for deuterium. [o] Data for parts A and B of Table are from
refs. [21, 251, and D. Donald, unpublished observations. For part C.
ref. I261 and W.J. Libbey, unpublished observations. [p] Ns =p-NO.C6H4S02. [q] Bs = p-Br-CaHB02.
Angew. Chent. internat. Edit.
1 VoI. 7 (1968) No. 10
norbornenylamhe (23), R = D, X = NH2, which
gives1251 product from G (R = D), in which exactly
half the deuterium is at each bridgehead [(30),R = D].
Similarly, the product distribution from L generated
from (24) is essentially the same 121 251 as that generated from (29)1241, where the major product is the
tricyclic alcohol (31).
Table 1 shows that in all cases, the ratio (G/L) of
products from the two cations is higher from the syn
starting material ( 2 3 ) than from the anti ( 2 4 ) .
The ratio G/L has a range of about 200, from 5:l in
the syn series to 1 :40 in the anti. In a formal conformational representation these results fall into the same
pattern observed in (12) and (13), where chair-boat
isomers, the ring-expanded cations (14) and (15), are
imaginable. The specificities of the follow-up rearrangements would be plausible if (25) and (26) were
twisted in the sense shown, since the p-orbital axis
(dotted line) at C-2 in each then would point toward
the (C-l)-(C-7) o-bond in (25) and toward the xbond in (26). Which cation [(25) or (26)] is the more
selective in the second rearrangement step is difficult
to ascertain, since the quasi-symmetric cation (27) itself has an intrinsic but unknown selectivity for further bond delocalization to G or L. The observed product distribution from, for example, the syn substrate
(23) therefore results not only from the competition
that measures the memory effect [the relative rates of
G and (25) + (27)] but also from the “natural” ratio that determines the distribution of the quasisymmetric cation [the relative rates of (27) + G and
ence for migration of the bridge that was anti to the
leaving group in the starting material and thereby directly measures the memory effect.
Rearrangements of anti-labeled substrates (32) (X -NH2 or arenesulfonate) and of the corresponding synlabeled seriesL26l (Table 1) show that memory is
preserved 1274. In the acetolysis of the p-bromobenzenesulfonate (32), for example, the experimental ratio
(37)/(38) of 1.4 leads, after correction for drift, to a
competition ratio k, /k, = 2. That is, about one-third
of the twisted cations (33) are trapped by bond rearrangement, and the remainder achieve the symmetrical condition 127bI.
7. Crossover via a Truly Symmetrical Cation.
Ring-Expansions of 7-Norbornylmethyl
The two ethylene bridges of a 2-bicyclo[2.Z.2]octyl
cation can be distinguished from each other with an
isotopic label, which permits the examination of
memory effects in the saturated 7-norbornylmethyl
system (32). The unsymmetrical cation, shown here as
a twisted species (33) for explication, can preserve the
distinction between the labeled and unlabeled bridges
if migration of one of them occurs at a rate that is not
very much slower than conversion to the symmetrical
cation (34).
8. The Nature of the “Twisted” Cations
So far we have resorted to a purely phenomenological
description of the species responsible for memory effects. From the results of further studies of the systems
already mentioned 128,291 and those of additional ringexpansions in the nortricyclylmethyl[301, l-methyl-2,3-
[25] J . J . Gaiewski, Ph. D. Dissertation, University of Wisconsin,
[26] a) J. A . Berson and M . S. Poonian, J. Amer. chem. SOC.88,
170 (1966); b) M . S. Poonian, Ph. D. Dissertation, University of
Wisconsin, 1966.
[27a] This is not a consequence of a secondary isotope effect,
since the migratory aptitude of a syn- or anri-C~DzH2bridge is
observed to be virtually the same as that of a syn- or nnti-CZH4
bridge, respectively [261.
[27b] The 33 yo efficiency of preservation of memory is derived
o n the assumption that bond rearrangement (33) + (35) is completely stereospecific. To the extent that some direct trapping in
the opposite stereochemical sense occurs 1/33) + (36)], the calculation overestimates the importance of the symmetrization
path ( 3 3 ) --f (34).
[28] W. J . Libbey, unpublished observations.
[29] D. Donald, unpublished observations.
[30] a) J . A. Berson, R. G. Bergman, G. M. Clarke, and D. Wege,
J . Amer. chern. SOC.YO, 3236 (1968); b) J.A. Berson, G. M . Clarke,
D. Wege, and R. G. Bergman, ibid. 90, 3238 (1968); c) J. A . Berson, D. Wege, G. M . Clarke, and R. G. Bergman, ibid. 90, 3240
The “natural” partition ratio for (34) is unity because
the follow-up rearrangements (34)
(35) and (34)
4 (36) necessarily proceed at the same rate (except
for a small secondary isotope effect). Therefore, after
correction for the drift that partially interconverts their
precursor cations (35) and (36) [16,18,191 the excess
of 6,7-labeled 2-bicyclo[3.2.l]octy1 product (37) over
the 3,4-labeled counterpart (38) represents the prefer.j
Angew. Chem. internut. Edit. / Vol. 7 (1968)
No. 10
and l-methyl-2norbornenyl-syn and anti-7-methyl ‘321 systems [(39),
(40), (4I), and (42)],we have concluded that the mul-
tiple rearrangements are most readily formulated as
step-wise rather than concerted processes. The “twisted” cations therefore are true metastable intermediates,
occupying minima in the potential energy surface of
the reaction. The evidence for this conclusion comes
from experiments on extramolecular capture and from
the influence of methyl substitution on the memory
8.1. Capture of the Once-Rearranged “Twisted”
Species by External Nucleophiles
Scheme 2 shows a path providing for formation of
product by nucleophilic capture of the first-formed,
once-rearranged cation (e.g., R‘ @) in the ring-expansion
route. Evidence for the occurrence of this reaction at a
Bs = p-Br-C&-SO,
symmetrical nonclassical ion (16), some of them are
trapped before the bond delocalization step. The ringexpansion of optically active ( I 2 ) thus leads to only
partially rather than completely racemized endo-bicyclo[3.2.l]octyl product [e.g. ( I ~ u )[151.
] However,
the second rearrangement step apparently is fast
enough to outstrip nucleophilic capture in some cases.
For example, in the syn- and anti-7-norbornenylmethyl systems, the amount of bicyclo[2.2.2]oct-5-en2-yl product is not observably greater when the
“twisted” cationic intermediates (25) and (26) are
generated by ring-expansion 121 251 than when the
system is entered by conventional solvolyses of the
endo- [231 and exo-bicyclo[2.2.2]oct-5-en-2-y1
r z a 1 substrates (28) and (29).
In the analogous saturated system, 7-norbornylmethyl (32), the mixture of 2-bicyclo[2.2.2]- and 2bicyclo[3.2.l]octyl derivatives formed in the ringexpansion contains a proportion of 2-bicyclo[2.2.2]octyl product only slightly greater than that obtained[18-201 in solvolysis by conventional n- or 5routes [26,2*1.
External nucleophilic capture of the cationic intermediate after one rearrangement step is also detectable in the ring-expansion of optically active 3-nortricyclylmethyl derivatives (39) [30,331.
rate competitive with that of the second rearrangement
step (intramolecular trapping) is available from the
chemistry of the chair 2-bicycl0[3.2.l]octyl cation (14).
Although a substantial fraction of these cations elude
capture by external nucleophile and rearrange to the
131 1 J. Foley, unpublished observations.
(321 J. M . McKenna, unpublished observations.
Rearrangement of one of the two nonequivalent ring
members from C-7 to the methylene side chain produces a twisted tricyclo[]0ctyl cation (43)
which can lose optical activity either by conversion to
the symmetrical ion (44) or by bond delocalization,
which converts it into optically inactive non-classical
ion (45), an intermediate o r transition state for the
second rearrangement step. This second step occurs
readily when the cations are generated by conventional
solvolyses of tricyclo[]0~tyl
derivatives (46),
as is shown by the equilibration of a deuterium label
equally between C-4 and C-5 in the product from
[4D]-(46)[30b, 33b3. The tricyclo[]octyl product from optically active tricyclo[]octyl
substrate (46) is virtually completely racemized [3ObI, yet
that from the ring-expansion retains about 24 % of
1331 Previous studies on this and related systems have been carried out in the racemic series by a) K. 5. Wiberg and G . Wenzinger, J. org. Chemistry 30, 2278 (1965); b) R. R. Sauers, J . A .
Beisler, and H. Feilich, ibid. 32, 569 (1967); c) A. K . Colter and
ibid. 30, 2462 (1965).
Angew. Chem. internat. Edit. / Vol. 7 (1968) 1 No. 10
its optical purity in arenesulfonate acetolysis or in
deamination. From the absolute configurations of the
starting nortricyclylmethyl substrate 130a1 and tricycl0[]octyl product 130b1, it is established
that the nucleophilic trapping of the twisted cation
(43) occurs from the same side as the departure of X
(OSO2-CsHcBr(p) or N 2 9 from (39).
8.2. “Twisted” Cations as True Metastable
first is completed, the rearrangement of R’-X to M’@
then is a concerted double migration passing over only
a single transition state and competing with processes
R‘-X + ROO and/or R’-X
M“O. Recent evidence
on the behavior of unsymmetrically substituted systems
strongly favors the first of these possibilities.
In the ring-expansion of the syn-[2,3-D2]-1-methyl-7norbornylmethyl derivative (40), the unsymmetrically
placed bridgehead substituent renders nonequivalent
the bonds a and b available for the first rearrangement
step. Migrations of a or b lead to the corresponding
Scheme 2 shows the “twisted” cation R’@,derived by
ring expansion, as a branching point in the mechanism.
By a second rearrangement step, it can become the
twice-rearranged cation M’O, and by a separate process, it can become symmetrical (or quasi-symmetrical)
ion Roe.
(SO). R
(SI), R
(4O), R = CHs
(32), R = H
= H
(52), R = CH3
(49), R = CH3
Scheme 3.
The question might be raised whether the energy surface
could include a small minimum after the transition state of
the ring-expansion but that this minimum is by-passed by
points descending from the transition state with more than
enough excess vibrational energy to overcome the small second barrier. R’O might then be considered to be a vibrationally excited form ofthe symmetrical cation ROO. It will seldom
if ever be necessary, however, to consider explicitly such
vibrationally excited entities in solution. Collisions with solvent molecules will deactivate the excited R’@ by energy
transfer at a rate that is within an order of magnitude of a
molecular vibration. A reaction of R ’ e competing with collisional deactivation (e.g., the second rearrangement step
R’O + M’O) therefore would have t o have an activation
energy not much greater than RT 1((
k c a l p o l e even at
100 “ C ) . The small second barrier in question would become
no obstacle, and the reaction would have, for all practical
purposes, only one transition state.
9. Unsymmetrically Substituted Systems.
Change in the Competition Ratio
If R’G is an energy minimum, the rearrangement from
R‘-X to M’@ proceeds in two discrete steps, the
second of which competes with symmetrization to ROO.
If the second rearrangement step is initiated before the
Angew. Chem. internnt. Edit.
i Vol. 7 (1968) 1 No. 10
“twisted” cations (47) and (50). The distribution of
starting material between these two branches is not of
great concern in the present context, as long as some
of the reaction goes by each path. The mechanistically
significant branching points occur at the next rearrangement stage. If the “twisted” cations are true
intermediates and their stereochemical behavior is
controlled by Scheme 3, then competition occurs
between intramolecular entrapment by a second rearrangement (kl and k ; ) and symmetrization by some
process (k2 and k i ) .
Although the nature of the symmetrization step remains to be discussed, it seems reasonable to suppose
that these rates (k2 and k;), whether they involve a
conformational change or some kind of extramolecular process (see Section ll),should not be sensitive to
methyl substitution. On the other hand, the rates of the
second migration step ( k l and k;) should be very
The second step (-x) in the path initiated by -b involves conversion of a secondary cation (50) to a rearranged secondary cation (52). The methyl substituent is at a remote position and should exert only a
feeble effect. Therefore, the -x step should proceed at
a rate k ; not greatly different from that in the “twisted” cation (51) lacking the methyl group. In the second
rearrangement step (-x) of the path initiated by -a,
the “twisted” secondary cation (47) is converted into a
rearranged tertiary cation (49). The methyl group is
directly attached to the site of developing positive
charge and provides a large kinetic driving force. One
would predict, therefore, that kllkz in (47) should be
greater than k ; / k ; in (50); that is, the selectivity for
total amounts of products derived from the -a and
-b branches are about equal, but whereas the part
derived from the -b branch shows[341 a selectivity
( k ; / k ; ) that is nearly the same as that observed1261
in the unsubstituted series (32) the selectivity in the
-a branch is substantially higher 1341.
These results are difficult t o reconcile with a concerted double
rearrangement mechanism having a n energy surface like Fig. 2,
Fig. 2. Energy surface f o r the acetolysis of [2,3-D21-syn-1-methyl-7norbornylmethyl p-nitrohenzenesulfonate 140) by a double rearrangement mechanism.
where the t w o rearrangement
telescoped into one. In t h e -a
doubly rearranged cation (49)
reactants would pass over either
steps in each branch a r e
branch, leading t o tertiary
or its [6,7-D&isomer, t h e
single transition state ax or
stereospecific migration of x in the branch of t h e ring
expansion of (40)initiated by -b should be about the
same as that in (32), but this selectivity in the -a
branch should be substantially higher.
These predictions are confirmed in the acetolysis of
[2,3-D~]-syn-l-methyl-7-norbornylmethyl-p-nitrobenzenesulfonate (40), X = OSO2-C6H4 -NOz(p). The
Fig. 3. Energy surface f o r the acetolysis of [2,3-D2]-syn-l-methyl-7hornylmethyl p-nitrobenzenesulfonate (40) hy a stepwise mechanism.
Symm. = symmetrization
1341 a) J . Fofey, unpublished observations; b) Similar investigations are under way in the unsaturated series I-methyl-syn- and
I-methyl-anri-7-norbornenylderivatives (J. M. McKennn, unpublished observations); c) The selectivity in the second stage of
ring expansion of a 1-methyl-2-endo-norbornylmethyl
is considerably higher than in the case of the unsubstituted compound (H. Junge, unpublished).
Angew. Chem. internat. Edit. / Vol. 7 (1968)
No. 10
single transition state ay. In the -b branch, leading t o secondary doubly rearranged cation (52) or its [6,7-D2]-isomer.
reactants would pass over single transition states bx or by.
T h e selectivities in each branch would be determined by t h e
differences Fay- Fax and Fby - Fbx. T h e experimental
facts given would have t o be interpreted as being caused by a
much larger difference between t h e competing transition
states in t h e -a branch than in t h e -b branch (Fay- Fax>
F b y ~-Fbx). T h e difficulty with this interpretation is that it
is not obvious why o n e of t h e four transition states should
be so much higher in energy than t h e other three.
On the other hand, a very natural explanation for the
observed behavior emerges from the step-wise mechanism (Fig. 3). Once the first barrier a or b is passed
in the ring-expansion step, intermediate once-rearranged secondary cations [(47) and (SO)] of very
nearly equal energies result. The relative heights of the
a and b barriers for this first step are not readily predicted, but it would not be surprising if they were
nearly the same and if, consequently, the starting
material were about evenly divided between the -a
and -b branches, as is observed. The major point,
however, concerns the second step. There the further
partitioning of the individual once-rearranged “twisted” cations so as to favor -x vs. symmetrization
specifically in the a branch is clearly to be expected.
The “twisted” cations R ’ 3 are thus most simply
formulated as true intermediates. They are unsymmetrical and can be trapped intramolecularly 01’
intermolecularly before symmetrization. The major
remaining problem concerns the source of their asymmetry.
10. Possible Intramolecular Origins of Memory
In the earliest examples of the memory effect [ringexpansion of the 2-norbornylmethyl derivatives (12)
and (13)], the first-formed cations (14) and (IS) were
thought to be conformational isomers of the chairboat cyclohexane type. Although it is easy to visualize
an energy barrier for interconversion of these cations,
and such a barrier may well contribute to the observed mechanistic insulation from each other of the
2-endo and 2-eno-norbornylmethyl systems, it is clear
that cyclohexane-type conformational isomerism is
not a requirement for the operation of the memory
effect. In the case of 2-bicyclo[2.2.2]octy1 cation, molecular models suggest that any barrier between the twisted form (33) and its enantiomer must be very small.
Further, although one perhaps might not be surprised
if bicyclo[2.2.2]octane preferred to exist in a slightly
twisted conformation so that the vicinal hydrogenhydrogen repulsions of the symmetrical eclipsed form
would be relieved 1351, large departures from the
Angew. Chem. internat. Edit.
1 Vol. 7 (1968) / No. I 0
eclipsed form would not be expected r36f, 36~1. The
weight of physical evidence from X-ray crystallography and infrared - Raman spectroscopy on the
parent hydrocarbon and some of its derivatives is
compatible with the eclipsed or almost eclipsed structure [361. It seems likely that any conformational
barrier separating twisted and eclipsed forms or
enantiomeric twisted forms must be small. Tntroduction of a double bond into the bicyclo[2.2.2]octyl cation system to give bicyclo[2.2.2]octeny1 cation results
in an even stiffer structure. If one, nevertheless, insists upon a conformational rationalization of the observed behavior in these cases, the memory effect then
constitutes a very sensitive probe for small conformational minima.
Nonclassical bonding might be another imaginable
intramolecular source of asymmetry. Thus, in the
ring-expansion of 7-norbornylmethyl derivatives (32),
the “twisted” cation might be assigned the nonclassical structure (531, in which partial bonding of C-4
to C-2 and C-3 makes the two faces of C-2 nonequivalent. This explanation is not very satisfactory, since
opening of the mesomeric bridge should occur with
inversion, whereas neither C-6 nor C-7 is in a favorable position to accomplish this. The reason for the
observed specificity C-6 > C-7 in the second rearrangement step therefore is not immediately evident. A possible explanation comes from consideration of a model
which shows that the (C-l)-(C-7) bond axis lies almost in the nodal plane of the (C-2)-(C-3) x system.
The (C-l)-(C-7) 0 electrons therefore cannot interact with the 5c system. In contrast, the (C-Y)-(C-6)
bond axis assumes an angle of about 60” to the ::
nodal plane. The (C-l)-(C-6) 0 electrons cannot approach from the preferred rear side of the x system, but
if bond migration does take place in spite of this, then
the only bond interaction possible is that between the
(C-l)-(C-6) 0 electrons and the front side of the z
system. If this interpretation is correct, the memory
effect is a much more sensitive test for nonclassical
intermediates than any previously used technique.
[35] a) R. B. Turner, W . R. Meador, and R . E. Winkler, J. Amer.
chem. SOC.79, 4116 (1957); b) In any case, since this type of
repulsion is absent from the (C-2)-(C-3) bridge of the symmetrical
2-bicyclo~2.2.2]octyI cation, it is not obvious that the hydrocarbon is a good model.
[36] a) A . H . Nethercot, j r . and A . Javan, J. chem. Physics 21, 363
(1953); b) J . J. McFarlane and I. G. Ross, J. chem. SOC.(London)
1960, 4169; c) T. Wada, E. Kishida, Y. Tomia, H. Suga, S. Seki,
and 1. Nirta, Bull. chem. Soc. Japan 33, 1317 (1960); d) G. S.
Weiss, A . S. Parkes, E. R. Nixon, and R. E. Hughes, J . chem.
Physics 41, 3759 (1964); e ) M . P. Marrocchi, G. Sbrana, and
G. Zerbi, J. Amer. chem. SOC. 87, 1429 (1965); f) A . F. Cameron,
G. Ferguson, and D. G . Morris, Chem. Commun. 1968, 316, have
found by X-ray analysis a (C-I)-(C-2) vs. (C-3)--(C-4) dihedral
angle of 3 jn the p-bromobenzenesulfonate of bicyclo[2.2.2]octyl-1-methanol; g) The X-ray structure analysis of bicyclo[2.2.21octane-1,4-dicarboxylicacid carried o u t by 0 . Ermer and
.I. D. Dunitz, Chem. Commun. 1968, 567, gave a dihedral angle
of 0
11. Possible Extramolecular Origin of Memory
rounding ions. For a uni-univalent electrolyte in
aqueous solution at 25°C and a concentration of
0.05 M, the mean radius of this ionic atmosphere of a
given ion is about 13 A 1391.
Since the intermediates in these rearrangements are
formed in solution, any temporarily asymmetric arrangement of this extramolecular environment might
persist long enough to evoke a correspondingly asymmetric response, even from a cation of intrinsically
symmetrical constitution. Several potential sources of
such effects might be imagined.
A Wagner-Meerwein rearrangement of the carbonium
ion would displace the central charge by a distance of
about 1.5 A. If this rearrangement were fast relative to
the rate at which the ionic atmosphere could adapt itself to the new location of positive charge, there would
exist a finite time during which a non-equilibrium
state prevailed. Even a symmetrical cation (e.g., 2bicyclo[2.2.2]octyl) when formed by rearrangement
would have a temporarily asymmetric ionic atmosphere and in principle could behave in an asymmetric
Perhaps the most obvious of these is ion-pair return.
In the case of 7-norbornylmethyl arenesulfonate solvolysis, for example, stereospecific ion-pair return from
(32), X == SOZ-C6H4-Br(p)
to [5,6-Dz]-bicyclo[2.2.2]octyl-2-exo-p-bromobenzenesulfonate
(54) followed by anchimerically assisted solvolysis of the latter would produce qualitatively the observed pattern
of the memory effect, that is, preferred migration of
the anti bridge in the second rearrangement step.
In detail, however, this process cannot be responsible for all
of the memory effect, which occurs not only in solvolysis but
also with equal or greater selectivity in deamination e.g. of
(321, X = N2@. The return process in dearnination would involve molecular nitrogen, which if it traps the once rearranged cation at all, should d o so with much lower efficiency than p-bromobenzenesulfonate. Further, even in
solvolysis not all of the rearrangement can pass through the
ion-pair return product (541, since independently synthesized
(54) rearranges with considerably higher selectivity than does
(32) [26bl. There must then be some mechanism for symmetrization on the path between (321 and (541, which in the ionpair context would be ascribed most simply to separation of
the ions into independently solvated species. This separation
process should be extremely sensitive to solvent, as has been
observed with other ion-pair + solvated-ion reactions. For
example, ion-pair return, which is pronounced in the acetic
acid solvolysis of 3-p-anisyl-2-butyl derivatives, is almost completely obliterated in the more proficient ion-solvating solvent, formic acidl37J. If the memory effects observed with
(32) were caused by ion-pair return, they should be drastically
diminished by similar changes of solvent, but they d o not
show this response. Especially striking is the behavior of the
syn-7-norbornenylmethyl series (23) (Table 1). Although
there seems to be a small temperature effect, the selectivity
is nearly invariant over a large range of solvents, leaving
groups, and electrophilic catalysts [21,25,38J.
The same results argue against a variant of the ion-pair return hypothesis in which the counter-ion, instead of capturing the cation covalently, merely remains in its vicinity but
stays predominantly o n one side of it.
Starting from a purely electrostatic point of view, one
might develop a more general hypothesis of the origin
of memory effects. The positive charge in a carbonium
ion interacts with all the other ionic charges in the
solution to produce an equilibrium distribution of sur[37] S . Winstein and G. C. Robinson, J . Amer. chem. SOC. 80,
169 (1958).
[38] D. Donald, unpublished observations.
How fast must the characteristic second rearrangement
step be in order to compete with relaxation of the ionic
atmosphere to the normal, symmetrical condition?
The relaxation time 7 for a model electrolyte, tetraethylammonium acetate, in aqueous solution can be
calculated 1391 to be approximately lO-lO/c sec,
where c is the molar concentration of ions. At 0.05 M,
7 = 2x10-9 sec.
The second rearrangement step thus would have to be
an extremely fast reaction with a rate only a few
thousand times slower than that of a molecular vibration. Since not much is known about the absolute rates
of such rearrangements, one has no strong basis a
priori upon which to evaluate the hypothesis. Fortunately, however, the whole matter is subject to experimental test, since the relaxation time and thereby
the memory effect should decrease with increasing
total ionic concentration. The experimental results [381
with 7-norbornenylmethyl solvolyses show that the
selectivities are unaffected by a three-fold increase in
ionic concentration. Therefore, the memory effect in
this system (and, by implication, in the others) does
not originate in an asymmetric ionic atmosphere.
Although the asymmetric atmosphere may well exist
temporarily around the newly-formed cation, it
either relaxes much more rapidly than the second rearrangement can occur, or else its contribution to the
asymmetric behavior of the cation is negligibly small.
A superficially similar but more readily defensible
argument can be constructed in terms of the intimate
details of the solvation shell of the freshly-formed rearranged ion. Little is known about the relaxation rates
of solvent structures, but it seems likely that the necessary reorganization in hydroxylic media must involve
readjustments of hydrogen-bonded aggregates or networks in the immediate vicinity of the cation. These
processes should be rather insensitive to the macroscopic viscosity, since long-range diffusion is not a
major contributor. The evidence is at least consistent
with this picture. The memory effect (selectivity) in
the 7-norbornylmethyl arenesulfonate solvolysis system is hardly affected by the fifteen-fold change in
macroscopic viscosity resulting from a change of pure
[39] Cf. H . S. Harried and B. B. Owen: Physical Chemistry of
Electrolytic Solutions. Reinhold Publishing Company, Inc., New
York 1953.
A n g e w . Chem. internat. Edit. Vol. 7 (1968) [ NO. I0
acetic acid to an acetic acid-polyvinyl acetate solution r26bI.
A guess at a reasonable upper limit for the solvent reorganization rate derives from the study of the dissociation of c-aminocaprolactam dimers, which occurs at a rate between lo9 and 1010 sec-1 in various
solvents [40,411. Since the reorganization of solvent
clusters under consideration in connection with
memory effects would require several such elementary
steps, the rate would be substantially lower than this
12. Conclusions
Although it is plausible to attribute some of the memory
effects to the occurrence of conformationally isomeric
cations, there are many cases in which the conformational barriers between the “twisted” and symmetrical
forms are so low that non-classical or extramolecular
causes for the observed effects must be considered.
The systems exhibiting memory effects provide uniquely suitable devices for revealing the existence of sub[40] K . Bergman, M. Eigen, and L. De Mneyer, Ber. Bunsenges.
species. The problem of defining the nature of these
intermediates is at the heart of carbonium ion chemistry. We hope that the observations reported here
will provide some stimulus to the more intensive
study of the microscopic structure of solutions, which
may provide the key to an understanding of the extramolecular group of memory effects.
The physical phenomena underlying memory effects
are very probably present whenever carbonium ions
are formed in solution. In the most general terms, one
can conclude that a description of the “structure” of a
carbonium ion in solvolytic medium is incomplete
without a specification of the means by which it is
This work has been supported in part by the National
Institute of Arthritis and Metabolic Diseases (Grant
A M 07505), the Petroleum Research Fund, the Wisconsin Alumni Research Foundation, and the National
Science Foundation, to whom we are grateful. All of the
experimental observations and many of the concepts
have been contributed by the dedicated and enthusiastic
group of collaborators whose names are given in the text
references. To work with them has been both challenge
and reward.
physik. Chern. 67, 819 (1963).
Received: December i t . 1967
[A 661 I € )
G e r m a n version: Angew. Chem. 80. 765 (1968)
[41] M . Eigen, Discuss. Faraday SOC.39, 130 (1965).
Dehydration of Alcohols on Aluminum Oxide
The dehydration of alcohols on y-aluminum oxide, which yields water, olefins, andlor
ethers, was studied with the aid of kinetic methods and I R spectroscopy. The unimolecular
olefin formation probably proceeds via a surface compound, in which an alcohol molecule
is joined by I W O angular H-bonds l o an OH group and an oxygen ion on the surface. The
ether formation (bimolecular reaction), on the other hand, requires OH groups and oxygen and aluminum ions on the surface. The ether is formed f r o m a surface alkoxide group
and molecularly adsorbed alcohol.
1. Introduction
The heterogeneous alumina-catalyzed dehydration of
alcohols has been known since the end of the 18th
century. Nevertheless, it is only recently that efforts
have been made to explain the elementary processes in
heterogeneously catalyzed eliminations, whereas the
elucidation of reaction mechanisms in the liquid phase
has reached a much more advanced stageC11. In the
[*I Priv.-Doz. Dr. H. Knozinger
Physikalisch-Chemisches Institut der Universitat
8 Miinchen 2 , Sophienstr. I t (Germany)
[l] I). V . Banfhorpe: Elimination Reactions. Elsevier, Amsterdam 1963.
Angew. Chem. internat. Edit. / VoI. 7 (1968) 1 Xu. 10
dehydration on aluminum oxide (which has been
mainly studied with ethanol as the alcohol), efforts
have often been made to find analogies with the reaction in the liquid phase. An excellent review of developments up to 1960 has been published by Winfield [21.
Widely differing views are still held concerning the
mechanism of the heterogeneously catalyzed dehydration of alcohols on aluminum oxide. In the case of aliphatic and some other alcohols the situation is complicated, particularly by the simultaneous occurrence
of a bimolecular reaction leading to ether formation.
[ 2 ] M. E. Winfield in P . H . Emmett: Catalysis. Reinhold, New
York 1960, Vol. VII, p.93.
79 1
Без категории
Размер файла
1 231 Кб
effect, ion, rearrangements, memory, multiple, carbonium, stereochemistry
Пожаловаться на содержимое документа