THEO

CHEM
ELSEVIER Journal of MolecularStructure (Theochem) 398-399 (1997) 169-179

Exploration of pericyclic reaction transition structures by quantum

mechanical methods: competing concerted and stepwise mechanisms’
K.N. Houk*, Brett R. Beno, Maja Nendel, Kersey Black, Hi Young Yoo, Sarah Wilsey,
Jeehiun K. Lee

Abstract

Density functional theory and multiconfigurational SCF calculations have been applied to a number of pericyclic reactions,
including cycloadditiona, electrocyclizations and sigmatropic shifts. Emphasis is on the competition between concerted and
stepwise mechanisms, comparisons of computed and experimental activation energies and isotope effects, and the performance
of MP2, CASSCF and DFT calculations. Various diradical processes, such as the vinylcyclopropane rearrangement and
cyclobutane isomerizations, were also studied to test the performance of DFT with diradical processes. 0 1997 Elsevier Science
B.V.

Ke\M.vrds: Cycloadditions; Sigmatropic shifts: Diradical intermediates: B3LYP; DFT calculations; Transition structures

1. Introduction available. We have recently explored the applications

of DFT methods, and much of this presentation
Pericyclic reactions have been a central focus of demonstrates how DFT is useful for the understanding
mechanistic study since the creation of the concept of the competition between concerted and stepwise
by Woodward and Hoffmann thirty years ago [I]. processes involving diradical intermediates.
Theory has made a special contribution to our under-
standing of the mechanisms of these ubiquitous pro-
cesses [2]. Here we summarize the state of the art in 2. Diels-Alder reactions
computations on these important organic reactions
and give examples from a wide variety of pericyclic The Diels-Alder reaction has been studied by most
reactions. computational methods, and there is wide agreement
To draw sound conclusions and make predictions, on the nature of the concerted transition state [X-5].
we must have confidence in the computational This simple reaction constitutes a particularly strin-
methods we use. For that reason, as we investigate the gent test of theory since both concerted and stepwise
chemistry, we also compare computational methods mechanisms are possible. The concerted pathway
with each other and with experimental data when involves a closed-shell species all along the reaction
pathway, while the stepwise involves open-shell
* Corresponding author. species. We and others have used complete-active-
’ Presented at WATOC ‘96, Jerusalem, Israel. 7- I2 July 1996. space multiconfigurational SCF calculations for the

0166.1280/97/$17,00 0 1997 Elsevier Science B.V. All rights reserved

P/I SO 166. I280(96)04970-6
170

1.382

Concerted T. S. Anti T. S. Anti Intermediate

CASSCF/6-31G* and Becke3LYP/6-31G* bond lengths are in parentheses and plain text, respectively.
All bond lengths are in A.

Fig. I. Comparison of CASSCF/6-3 IG* and Becke 3LYP/6-3 1G* Wuctures for the Diels-Alder reaction of butadiene and ethylene

butadiene/ethylene reaction, but to obtain appropriate exchange-correlation functional, the DFT method
activation energies requires the inclusion of dynamic would produce correct spin densities. However, with
electron correlation, using at least MP2 or CI 141. We approximate functionals, spin contamination of
have also studied the reaction with Becke3LYP/6- singlet diradical wavefunctions occurs. The spin
3 1G* density functional calculations. These are com- correction procedure produces a curious cusp at the
pared to the CASSCF/6-31G* geometries in Fig. 1. onset of spin contamination, as has been shown in our
The general comparison is excellent! Furthermore, the study [5] and pointed out in studies of a number of
energetics are quite reasonable as compared to avail- related systems by Chen and Schlegel [7]b. On the
able experimental data [S]. other hand, the singlet-triplet gaps are improved
The unrestricted DFT geometries obtained on the markedly by spin projection, according to results
stepwise pathway have some spin contamination. In from Cramer and Smith [6]b.
UHF, it is common practice to remove this spin con- We have taken the position of reporting energies
tamination by spin projection, and this has been done from both the contaminated and the spin-corrected
for DFT wavefunctions also [6]. However, the proce- wavefunctions, since the experimental energies
dure is controversial, and there is reason to believe appear to be intermediate between the two sets of
that the spin projection should not be applied to spin- computed energies.
contaminated DFT wavefunctions [7]. With an exact An exacting test of computational methods was

reaction to 98.9% completion;

J + y reisolate re~~~~s “I’ ’ o
*
0

143 - 2186 g (1.8 - 26.4 mol)

combined reactants
Fig. 2. High precision ‘H and “C kinetic isotope effects
 K.N. Houk et al.Noumul ofMolucular Structure (Theochew) 398-399 (1997) 169-179 171

recently made available due to Singleton’s measure- the isotope effects were calculated by both methods
ments of high-precision isotope effects for the Diels- [9]. The DFT predictions are remarkably close to
Alder reaction, shown in Fig. 2 181. The Becke3LYP experiment, and the tiny derivations from perfection
and RHF transition structures have been located, and for H-7 and H-9 may be due to a small equilibrium

Becke3LYP/6-31G* 2H Secondary KIEs Becke3LYP/6-31G* 13C KIEs

1.025
Cl l

fi 1.020
HI0 / c4
l

3 HI z 1.015
g 0.950 z /
.-E H9 .ik 1.010
!z l

!?’ 0.925 ::
W I // w 1.005

0 0

E
d
s!
6
o\ z
d
Ca&ated”mE Calculated KIE
“\

RHF/6-31G* 2H Secondary KIEs RHF/6-31G* 13C KIEs

1.oOO I .030
CH3 - Dl 7
8 I .025
2
w 0.975
1.020
‘iii ?I
2
@ 0.950 t 1.015
E I
z -L.
x & 1.010
&j 0.925 rz
1.005

0.900

Fig. 3. Comparison of RBecke3LYP/6_3lG* and RHF/6&31G* kinetic Isotope effects

172 K.N. Houk et ul.Nourr~al of Molecular Structure (Theochenr) .?98-399 (I 997) 169- I79

.. .. . .

lEnthalpies are in kcal/mol. RBecke3LYP/6_31G* used for closed-shell species and

UBecke3LYP/6-31G* used for open-shell species.

Fig. 4. Enthalpy profile (298 K) for the Diels-Alder reaction of cyclopentadiene and ethylene.

percentage of the gauche isomer of isoprene in the here the DFT study of stationary points on the ground-
ground state [9]. A comparison of calculated and state surface. Fig. 4 summarizes these results. Of
experimental kinetic isotope effects is given in Fig. 3. particular note is the predicted 12.8 kcal mol-’
We have begun a collaboration with Ahmed ‘enthalpy of concert’ [ 121. This is larger than pre-
Zewail on the ‘transition state spectroscopy’ for a dicted for the butadiene-ethylene reaction, since the
retro-Diels-Alder reaction. The Zewail group has diene is constrained to a geometry essentially ideal for
applied femtosecond spectroscopic techniques to the concerted transition state.
study the retro-Diels-Alder reactions of norbornene In the retro-Diels-Alder reaction, the stepwise
and norbornadiene [lo]. A femtosecond laser pulse mechanism involves breaking of one bond and
produces excited states of the reactants which rapidly pivoting around the other to form the anti-biradical.
decay, presumably on the ground state surface. With There is no gauche-in biradical, because there is
norbornene, a transient with the mass of norbornene no barrier to closure of the hex-2-en-1,6-diyl
decays with a halflife of 160 fs, while a species with biradical when the termini are gauche [5]. This
the mass of cyclopentadiene appears with a halflife of contrasts with the butane- 1,4-diyl, which does have
30 fs and disappears with a halflife of 220 fs [lo]. a small symmetry-imposed barrier when the
This reaction has also been studied by more con- radical centers are gauche. After bond breaking in
ventional kinetic techniques [ 111, and there are exten- norbornene to form the anti diradical, there is an
sive data available which can be interpreted using additional barrier to cleavage to cyclopentadiene
detailed quantitative information about the potential plus ethylene.
surfaces obtained from calculations. The diradical intermediates have spin-contaminated
The excited state studies are in progress. We report wavefunctions, according to B3LYP calculations.
 K.N. Houk et uI./Journnl of Molecular Structure (Theochrm) 398-399 (1997) 169-179 173

D=66” D=70”

GF

Gl

0
1 1
II+II
D=118”

D=l80”
\‘

!I

GT

TM

+
II II
Fig. 5. (a) Gauche tetramethylene and connected transition structures. (b) Trans.tetramethylene and connected transition structures.
174 K.N. Houk et trl./Journcil of Molecular Structure (Theochew) 398-399 (1997) 169- 179

SDCVZPEt
Relative Energy kcalimol - UB3LYP/ZPE

+3.0 c
GF GT

+l.O -

0.0 -

-1.0 -

-2.0 -

-3.0 -

7 Doubleday, C. J. Am. Chem. Sot. 1993, 115, 11968-I 1983.

Fig. 6. SDCI and DFT stationary pomts of tetramethylene

Spin correction gives somewhat altered energetics, which has been studied thoroughly by CASSCF and
but the same general conclusions may be made multireference configuration methods by Doubleday
about the relative energies of concerted and stepwise [ 131 and by Robb et al. [ 141.
pathways, as well as the nature of the rate-determining Fig. 5(a) and (b) show the structures of the gauche
step in the stepwise process. and anti stationary points of the tetramethylene surface
computed by UBecke3LYP methods. The nearby sta-
tionary points are shown as well. The gauche and anti
3. Tetramethylene biradicals have considerable communication with the
radical centers, as reflected in the long C2-C3 bond
The DFT method with the Becke3LYP functional length (1 S98 i and 1.612 A, respectively). The
appears to be remarkably successful in the balanced Becke 3LYP geometries of the intermediates and
treatment of concerted mechanisms and diradical spe- transition structures for isomerization, closure and clea-
cies. To test the performance of B3LYP for diradical vage compare very favorably with those obtained by
surfaces, we investigated the tetramethylene surface, CASSCF. CISD and UKS methods. The energetics are

AH+= 33.5 kcal/mol AH’= 31.9 kcal/mol AH+= 33.6 kcal/mol AH+= 30.0 kcal/mol
AS*= -13.8 eu AS*= -14.2 eu AS*=-11.7eu

Doering,
W. “on E.; Toscano. V. G. K. A. Dykstra Huntsman. W. D. Hopf, H.
and Beasley, 0. H. Ph.D. Dissertation, and Wrfsters. tf. J. Tetrahedron Left. 1972. 34, 3571
Telrahedron 1971, 27, 5299. Ohio University 1972 J. Am. Chem. Sot. 1967, 69,342

Fig. 7. Experimental activation parameters for the four prototypical hydrocarbon [3,3]-sigmatropic shifts
also similar, as shown by the comparison given in 1,4-diyl was found by CASSCF/6-3 lG* to be compe-
Fig. 6. titive [ 181. Although RHF results did predict the con-
Earlier in this paper we noted that the diradical certed mechanism, MP2 results were quite different
formed from a diene and ethylene closes without a and led to considerable controversy [3].
barrier when the radical centers approach each We have explored all of the prototype [ 3,3]-sigma-
other. By contrast. the gauche biradical closure occurs tropic hydrocarbon rearrangements, as shown in
with a twist of one terminus, which overcomes the Fig. 7. There are four basic reactions, only the first
orbital symmetry forbiddenness of a [7r% + R’S] and last of which involve degenerate rearrangements.
process. This difference is of general significance, The figure also shows known experimental acti-
since it indicates that even diradical pathways will vation parameters for the parent or methylated
have stereochemical preferences. Ongoing work on systems [ 19-221.
the vinylcyclopropane rearrangement shows the The relevant transition structures have been located
importance of such effects in controlling the by RHF, CASSCF and B3LYP techniques. At the
stereoselectivity of this reaction [ 151. CASSCF/6-3 1G* level, only stepwise transition struc-
tures are found. Fig. 8 shows a summary of the transi-
tion structures, and Fig. 9 shows the activation
4. Cope rearrangements and related energies calculated by various methods and compared
[3,3]-sigmatropic shifts with experimental data as well.
The experimental data are remarkable in their
The parent Cope rearrangement has attracted nearly constant activation energies, regardless of the
extraordinary mechanistic attention, both experi- type of multiple bond involved in the reaction. RHF
mental and theoretical. For the reaction of 1,5-hexa- and CASSCF techniques, which neglect dynamic
diene, Borden et al. [ 161 and Davidson et al. [ 171 have correlation involving electrons outside the active
concluded that the reaction involves a concerted space, both overestimate the activation energy and,
mechanism; the stepwise process via a cyclohexane- in general, predict an increase in activation energy

d,j id2

METHOD d dl d2 dl d2 dl d2
RHF/6-31G* 2.046 1.989 1.960 1.965 1.914 1.900 1.900
CASSCF/6-31G* 2.189 1.872 1.670 1.698 1.817 1.646 1.859
Becke3LYP/6-31G* 1.969 1.824 1.862 1.771 1.926 1.763 1.763

Fig. 8. TransItion structure\ of the four hydrocarbon [3,3]-Ggmatropic shifts. Formmg and breaking bond lengths are shown
176

RHF/6-31G’
. . . . . . . . . . . . . . . ..___
52.6

. .._.__

CASSCFIS-3lG’
RHF/3-2111.. . .

. ........ .. .. ... .
CASSCFB-21 G 39.6

. . . ..~........._______.__..__.._.......... ~.35.!____
- “,;:z 34.0

h~!?!F?!F&??G: ,..........._..__._
BLYPIG-31G’

[* This value from M. Dupuis, C. Murray and E. R. Davidson]

Fig. 9. Calcuiated and experimental activation energies for [3.3]-sigmatropic shifts.

when an increasing number of R electrons are The DFT results here, like those mentioned earlier,
involved in bonding changes. are remarkably good for hydrocarbon pericyclic reac-
The problem is corrected by MP2 or by DFT cal- tions [23-251, even with modest basis sets. Whether
culations, which restore a balanced treatment of or not this will be a general finding for hetero systems
dynamic correlation. There is a correlation error of is still uncertain.
6-8 kcal mol-’ for each electron pair involved in
bonding changes or T bond which is distorted from
its equilibrium value in the transition state, whether or 5. [5,5]-sigmatropic shift
not it is one of the bonds directly involved in bonding
reorganization. We have applied several computational techniques

Fig. 10. Known [5,5]-sigmatropic shifts.

 K.N. Houk et ul./Jo~rnccl of Moltwlnr Structure (Thtwhem) 39X-399 (1997) 169-179 17-l

Fig. 10 shows examples of [5,5]-sigmatropic shifts

in the tetrahydroheptalene series. These reactions are
geometrically constrained to follow [5a,5a] stereo-
selectivity, regardless of the details of the mechanism.
We have investigated the [5,5]-sigmatropic shift of
the parent [5s,5s] tetrahydroheptalene [R = H, E =
H]. In addition, we have examined the mechanism
of the parent acyclic compound. Comparison of the
two stereoselectivities allows one to determine which
of these thermally allowed stereochemistries of reac-
tion is favored inherently.
All methods applied to this problem indicate that
the rate-determining transition state of this reaction is
very close to an intermediate formed by formation of
one u bond. This is consistent with the energetic
estimates. Whereas cyclohexan-1,4-diyl, a potential
A intermediate in the Cope rearrangement, is estimated
to be 35 kcal mol-’ above 1Shexadiene, only
[5a,5a] Intermediate 1.5 kcal mol-’ above the experimental activation
AE = 24.3 kcal/mol barrier, the corresponding biradical from 1,3,5,9-
decatetraene is only 18.6 kcal mol-’ above the tetra-
Fig. I I RBecke 3LYP/6-3 lG* intermediate in the [SaSa]-sigma-
ene, not including ring strain in the lo-membered
tropic rearrangement of 5,5a,lO, IOa-tetrahydroheptalene.
intermediate.
to a novel sigmatropic shift involving ten electrons,
the [5,5]-sigmatropic shift. Examples of this reaction
have been reported by the Hafner group in Darmstadt c-0 c-a .
[26], and our theoretical studies are in collaboration
with those of Hafner, Lindner and colleagues [27]. Due to the size of the system, it has not been

[%,5s] Intermediate [5a,5a] Intermediate

AE = 27.1 kcaYmo1 AE = 33.4 kcaYmo1
Fig. 12. RBecke 3LYP/6-3lG* intermediates in the [5,5] sigmatropic rearrangement of (Z,z)-I .3,7,9-decatetraene.
 K.N. Houk rt nl.Noun~crl ofMolecular Structure (Throchem) 398-399 (19Y7~ 16YSl79

6. Conclusion

DFT with the B3LYP functional is adequate for

comparisons of concerted and diradical mechanisms.
The B3LYP functional is remarkably robust for the
study of pericyclic reaction of hydrocarbons. even in
comparisons of concerted with stepwise diradical
pathways. We have shown how detailed surfaces
can be found and understood, and how new insights
into stereoselectivities can be obtained.

Acknowledgements
[3s,3s] Intermediate
AE = 10.4 kcaVmo1 We are grateful to the National Science Foundation
Fig. 13. RBecke 3LYP/6-3lG* intermediate in the [3s,3s] sigma-
for financial support of this research. We thank
tropic rearrangement of 2.5divinyl-I ,5-hexadiene. Professors Klaus Hafner and H.-J. Lindner for colla-
boration and hospitality in Darmstadt. JKL thanks
the National Institutes of Health for a postdoctoral
possible to explore the surface thoroughly, but the fellowship (1 F32GM 17460-01).
best method available, B3LYP, indicates that a
diradical intermediate is formed (Fig. 11).
The acyclic case shows a preference for the [%,%I
References
stereochemistry. Diradical intermediates are predicted
for both processes. Fig. 12 shows both intermediates.
[I] R.B. Woodward and R. Hoffmann. The Conservation of
While we have not yet located the transition structures Orbital Symmetry, Academic Press, New York. 1966.
on either side of the intermediates, they should be very [2] K.N. Hook, Y. Li and J.D. Evanseck, Angew. Chem.. 104
similar in structure to these intermediates. The [5s,5s] (1992) 71 l-739: Int. Ed. En@., 31 (1992) 682-708.
stereochemistry is favored, and calculations on the [3] K.N. Houk, J. Gonzalez and Y. Li. Act. Chem. Rec.. 28 (1995)
8 I-90.
pentadienyl fragments distorted into transition state
[4] Y. Li and K.N. Houk, J. Am. Chem. Sot.. I15 (1993) 747%
geometries indicate that this is the result of the smaller 7485.
distortion required to achieve the 5s geometry. In par- [5] E. Goldstein. B.R. Beno and K.N. Houk. J. Am. Chem. Sot.,
ticular, there is much less distortion of the p system in 1 18 (1996) 6036-6043. and references therein.
the suprafacial mode than in the antarafacial. [6] (a) K.E. Edgecombe and A.D. Becke. Chem. Phys. Lett., 244
(1995) 427-432; (b) C.J. Cramer and B.A. Smith, J. Phys.
There has long been discussion about how radical-
Chem., 100 (I 996) 9664-9670.
stabilizing substituents at the 2 and 5 positions of 1,5- [7] (a) J.A. Pople, P.M.W. Gill and N.C. Handy, Int. J. Chem., 56
hexadiene influence the transition state geometry (1995) 303-305: (b) W. Chen and H.B. Schlegel. J. Chem.
of the Cope rearrangement. The [5,5]-sigmatropic Phys., IO1 (1994) 5957-5968.
shift is related in that the radical centers of the inter- [8] D.A. Singleton and A.A. Thomas. J. Am. Chem. Sot.. I I7
( 1995) 9357.
mediate are stabilized by allylic resonance. For com-
[9] B.R. Beno, K.N. Hook and D.A. Singleton, J. Am. Chem. Sot.,
parison, we located the intermediate found in the I IO ( 1996) 9904-9905.
Cope rearrangement of 2,5-divinyl- 1,5-hexadiene. [IO] B.A. Horn, J.L. Herek and A.H. Zewail. J. Am. Chem. Sot..
This is shown in Fig. 13. The activation energy for I IO (1996) 8755-8756.
formation of this species is lowered by 13 kcal molJ’ [1 I] B. Jursic and 2. Zdravkovaki, J. Chem. Sot., Perkin Trans., 2
(1995) 1223: B.S. Jursic, J. Mol. Struct. (Theochem). 358
relative to the parent, about one-half the expectation
(I 995) 139; (a) W.C. Hemdan, W.B. Cooper, Jr. and M.J.
for the stabilization of two radical centers by allylic Chambers, J. Phys. Chem., 68 (1964) 2016-2018; (b) J.J.
resonance. The mechanism is altered to a stepwise Kiefer, S.S. Kumaran and S. Sundaram, J. Chem. Phys., 99
one. (1993) 353 I-3541 : (c) J.R. Barker and G.D. King, J. Chem.
 K.N. Houk et al.Noumal of MolecularStructure (Themhem) 398-399 (1997) 169-179 179

Phys.. I03 (1995) 4953-4966; (d) R. Walsh and J.M. Wells, [I91 W. van E. Doering. V.G. Toscano and G.H. Beasley Tetra-
J. Chem. Sot.. Perkin Trans., 2 (1976) 52. hedron, 27 (1971) 5299.
[12] W.V.E. Doedng, W.R. Roth, R. Breuckmann, L. F&e, H.-W. [20] K.A. Dykstra, PhD dissertation, Ohio University (1972).
Lennartz, W.-D. Fessner and H. Prinrbach, Chem. Bet-., I21 [2l] W.E. Huntsman and H.J. Wristers, J. Am. Chem. Sot., 89
(1988) 1-9. (1967) 342.
[I31 C. Doubleday, Jr, J. Am. Chem. Sot., I I5 (1993) I I 968- [22] H. Hopf. Tetrahedron Lett., 34 (1972) 3571.
I 1983. [23] 0. Wiest and K.N. Houk, Topics in Current Chemistry, 183
[ 141 F. Bemardi, A. Bottani. P Celani, M. Olivucci. M.A. Robb and ( 1996) l-24.
A. Venturini. Chem. Phys. Len.. 192 (1992) 229-23.5. [24] 0. Wiest, K.A. Black and K.N. Houk. J. Am. Chem. Sot., I I6
[ 151 M. Nendel, 0. Wieat and K.N. Houk, unpublished results. (1994) 10336-10337.
[ 161 D.A. Hrovat, K. Morokuma and W.T. Borden. J. Am. Chem. [25] 0. Wiest. K.A. Black and K.N. Houk, J. Am. Chem. Sot.. I I7
Sot., I I6 (1994) 1072. (199.5) 8.594-8599.
[ 171 P.M. Kozlowskt, M. Dupuis and E.R. Davidson. J. Am. Chem. [26] K. Hafner, H.J. Lindner. W. Luo, K.-P. Meinhardt and T. Zink,
sot., I I7 (1995) 774. Pure Appl. Chem., 65 (1993) 17-25.
[I81 W.T. Borden and E.R. Davidson, Act. Chem. Res.. 29 (1996) [27] J. Fennen. K. Hafner. H.J. Lindner, B. Beno, K. Nakamura and
67. K.N. Houk, in preparation.