5630 J . Phys. Chem. 1990, 94.

5630-5636

TABLE VI: Ewgy Jliffemces R e f e m to C&+ in k d / m o P
molecule basisb SCF MP2W MP2-R 12(v) MP4(v)
C3H7+ (1) s 0 0 0
C3H7+ (3) s 0.2 0.7 0.6
C4+7+ (5) 0.8 1.5 1.6
oGeometries from P. v. R. Schleyer et al.; see ref 16. bSmall basis:
see Table 1. Reference 16.
graphically. On a conventional MP2 or MP4 level structure 2
lies -4 kcal/mol above structure 1, while structures H lie within
1.5 kcal/mol above the chiral structure 1 . In view of the sim-
ilarity of structures 3 and 4 we have computed the higher sym-
metric one (3)-although it has two imaginary frequenciesI6-
together with structures 1 and 5.
The results are collected in Table VI. They are very close to
those of ref 16. The conclusions of Schleyer et a1.I6are confirmed.
4. Concluding Remarks
We were able to show that the MP2-Rl2 method can be applied
to problems of structural chemistry, like that of the relative
stability of various isomers of carbocations. Since the error in
the correlation energy is significantly reduced with respect to
conventional MP2 calculations, a gain in reliability is achieved.
Nevertheless, our results must be regarded as preliminary for
two reasons. (a) MP2 is only a first step in a hierarchy toward
an exact treatment of electron correlation. (The next steps need
not be perturbation theory of higher order but may for example
be of coupled-cluster type.) In order to get new recommended
values, one must combine MP2-RI2 with an extrapolation from
MP2 to an exact treatment. This introduces some uncertainty.
(b) The MP2-Rl2 calculations converge, upon augmentation of
the basis, much faster toward the exact nonrelativistic results than
do conventional calculation. However, for MP2-RI 2 calculations
0 the basis must not be too small. Here we were obliged to use
0.6 smaller basis sets than ought to have been taken but which we
1.4
could not afford due to limited computer resources.
In order to get conclusive answers to problems of the type
studied here, one should be able to implement the R12 concept
into coupled-cluster or CEPA-like methods and to speed up the
computer code. Work along these lines is in progress.
Note Added in Proof. A recent paper by C. Liang, T. P.
Hamilton, and H. F. Schaefer 111 (J. Chem. Phys. 1990,92,3653)
is consistent with our results for C2H3+,especially as far as the
estimate of the basis incompleteness error is concerned. However,
we do not think that the accuracy achievable a t present allows
to make definite conclusions about the effects of the incomplete
treatment of electron correlation in the CCSDT-1 method.
Acknowledgment. The authors thank U. Fleischer for sug-
gesting the application of the MP2-Rl2 method to the problem
of the stability of the carbocations and P. v. R. Schleyer and M.
Biihl for the data on the MP2/6-31G* optimized geometries and
for valuable comments. These results were first presented at the
symposium 40 years of quantum chemistry in Athens, GA,
October 1989, organized by N. C. Handy and H. F. Schaefer 111
in honor of J. A. Pople. The computations were performed on
the Cyber 205 of the computer center of the Ruhr-Universitat
Bochum. Financial support by Fonds der Chemie is gratefully
acknowledged.
Registry NO. 1, 151 35-49-6; C2Hjt, 14604-48-9; C2HSt, 14936-94-8;
C,HT+, 19252-53-0; C2H2, 74-86-2; C2H4, 74-85-1.

Ab Inftio ContormaioMll Analysis of Cyclohexane

David A. Dixon*
E. I. du Pont de Nemours and Company. Central Research and Development Department,
Experimental Station, Wilmington, Delaware 19880-0328

and Andrew Komomicki*

Polyatomics Research Institute, San Antonio Rd., Suite 420, Mountain View, California 94043
(Received: November 9, 1989; In Final Form: March 5, 1990)

The inversion of cyclohexane has been studied by using ab initio molecular orbital theory. Six stationary points were found
and characterized as minima or transition states. With a double-{ basis augmented with a set of d-type polarization functions
on each carbon, the geometries were optimized, and force constants were calculated. An MP-2 correlation correction was
also evaluated with a much larger basis of triple-{plus polarization quality. The lowest energy structure is the chair conformation
of Dw symmetry. The other minimum is found to be the twist-boat structure of D2symmetry, which lies 6.9 kcal/mol higher
in energy. A boat structure of C, symmetry is a transition state, along a pseudorotational coordinate connecting two twist-boat
structures. This transition-state structure lies 7.9 kcal/mol above the chair. The interconversion between chair structures
via the twist boat is accomplished through either of two transition states, one of C2 symmetry, the other of C, symmetry.
Both of these transition states lie about 12 kcal/mol above the chair structure. Knowledge of the zero-point correction and
the resulting entropies of activation yields values of AS and AG that are in good agreement with experimental values. Finally,
the evaluation of the force constants at each of these transition-state structures has allowed us to analyze the torsional motion
of the ring skeleton.

Introduction potential energy hypersurface as either minima or maxima and

Cyclohexane plays a central role in conformational analysis. labeled them by their appropriate point-group symmetry. The
Much of our basic understanding a n be traced back to the work chair of D3d symmetry and the twist-boat of D2 Symmetry were
of Hendrickson, who characterized several structures on the Considered as minima. A boat structure of C ~symmetry,
U ~ ~ e d

Contribution no. 4416. ( I ) Hendrickson, J. 8. J . Am. Chem. Soc. 1961, 83,4537.

0022-3654/90/2094-5630$02.50/0 0 1990 American Chemical Society

Ab Initio Conformational Analysis of Cyclohexane
to connect equivalent twist-boat structures, and a separate, lower
symmetry transition state was seen to connect the twist-boat
manifold with the manifold of chair structures. This model was
reformulated in an elegant description of the normal modes by
Pickett and Strauss? who showed that the transition-state structure
should also be fluxional, Le., pseudorotating with low barriers.
The above two theoretical models empolyed molecular mechanics
to model the energetics of the rearrangement as have a number
of other author^.^ The only molecular orbital investigation is that
of Komornicki and McIver: who employed a semiempirical
method. Using gradient techniques, they showed that the chair
and twist-boat structures are true minima (all positive curvature
directions) while the boat structure was a transition state (one
negative curvature direction) connecting twist-boat structures.
They found two separate transition states of C, and C2symmetry
connecting the chair and twist-boat manifolds. These investigators
also suggested that the C, and C2structures were connected by
a low-energy pseudorotation mode. Any structure lying along this
mode would be of comparable energy and could effectively act
as a transition state connecting the chair to the twist-boat manifold.
The approximations involved in the semiempirical method pro-
duced relative energies that were too small when compared with
the available experimental data due in part to an overestimation
of the C-H bond lengths by -0.10 A. To the best of our
knowledge there has not appeared any a b initio investigation of
the various conformers of cyclohexane. There have however, been
a number of a b initio calculations on five-membered rings con-
taining a t least four carbon atoms. These have focused on the
nature of the pseudorotation process following the ring puckering
coordinates of Cremer and P ~ p l e .These
~ calculations have been
done on cyclopentanes and cyclopentenes,b pyrrolidines?b and
other systems.&
A number of N M R experiments on the interconversion of
cyclohexane have been done in solution. For CsDIIH,the value
of AH is 10.8%xl/mol in CS2 whereas Jonas and ~o-workers~
have reported a value for AH*of 10.0 f 0.3 kcal/mol for C6HI2
in methylcyclohexane. More recently, Ross and True* measured
the activation energy in the gas phase for the chair-to-chair in-
terconversion process using N M R on a dilute sample of C6Hl2
in SF6. They found AH*= 12.1 i 0.5 kcal/mol and AS* = 5.7
f 0.5 eu. This value for AH*differs from the results obtained
in the solution experiments. The reverse reaction has also been
investigated experimentally. In a very elegant experiment, Anet
and co-workers9 trapped the twist-boat structure in an Ar matrix
at 20 K. By warming the matrix, they determined a value of AG*
= 5.27 f 0.05kcal/mol for the twist-boat to chair isomerization.
We are interested in studying a variety of conformational
phenomena using a b initio molecular orbital theory. A number
of our investigations have been motivated by our need to better
understand the energetics of conformational changes in polymeric
subunits.Q Also, as a prelude to the study of the conformations
of sugars and carbohydrates, we have undertaken an investigation
~~
(2)Pickett. H. M.; Strauss, H. L. J . Am. Chem. Soc. 1970,92,7281.
(3) (a) Wilberg, K. B.; Boyd, R. H. J . Am. Chem. Soc. 1972,94,8426.
(b) Allinger, N . L. J . Am. Chem. Soc. 1977,99,8217.(c) Schneider, H.J.;
Gschwendtner, W.; Weipnd, E. F. J. Am. Chem. Soc. 1979,101,7195.(d)
Offmbach. J. L.; Strauss, H. L.; Graveron-Dmmilly,D. J. Chem. Phys. 1978,
69,3441.
(4)Komornicki, A.; Mclver, J. W., Jr. J. Am. Chem. Soc. 1973,95,4512.
Scc also: Stanton. R. E.; McIver, J. W., Jr. J. Am. C k m . Sac. 1975,97,3632,
for a study of the group theoretical aspects of this problem.
(5) Cremer, D.; Pople, J. A. J . Am. Chem. Soc. 1975,97,1354.
(6)(a) Cremer, D.; Pople, J. A. J . Am. Chem. Soc. 1975,97,1358.(b)
Pfafferot, G.; Oberhamma, H.; hggs, J. E.; Caminati, W. J. Am. C k m . Soc.
1985,107,2305.Pfafferot, G.; Oberhammer, H.; Bogge,J. E.; Caminati, W.
J . Am. Chem. Soc. 1985,107. 2309. (c) Serianni. C.; Chipman, D. J . Am.
Chem. Soc. 1987,109,5297.
(7) (a) Anet, F.A. L.; Anet, R. In Dynamic Nuclear Magnetic Resonance
Spectroscopy;Jackman, L. M., Cotton, F. A,, Eds.; Academic Press: New
York, 1975;p 543 and references therein. (b) Anet, F. A. L.; Bourn, A. J.
R. J . Am. Chem. Soc. 1%7,89, 760. (c) Hasha, D. L.; Eguchi, T.; Jonas,
J. J . Am. Chem. Soc. 1982,104,2290.
(8) Ross, B. D.; True,N. S. J . Am. Chem. Soc. 1983,105, 1382,4871.
(9)Squillacote, M.; Sheridan, R. S.;Chapman, 0. L.; Anet, F. A. L. J .
Am. Chem. Soc. 1975, 97,3244.
The Journal of Physical Chemistry, Vol. 94, No. 14, 1990 5631
of the conformations of six-membered rings with focus initially
on cyclohexane because of its central role in conformational
analysis.
Methods
We have calculated the geometries, energies, and force fields
of several structures of cyclohexane using a b initio molecular
orbital theory. These include structures of Du, D2, C, C2,and
C, symmetry. The starting basis set was of double-f quality in
the valence space with contraction coefficients and exponents from
Dunning and Hay.I2 The carbon atom basis set was then aug-
mented by a set of d polarization functions yielding a basis set
of the form (9sSpld/4~)/[3~2pld/2s]denoted as DZP. Additional
energy calculations, which utilized the geometries obtained with
the smaller basis set, were also performed with a more extensive
basis derived from the (10s6p) primitive set of Huzinaga13 and
contracted by Dunning to a triple-f set of the form [5s3pld]
on the carbon atoms. For this set, the polarization functions are
based on a least-squares fit of two primitive gaussians.I5 The
hydrogen basis is also of polarized triple-{ quality derived from
a (5slp) primitive set and contracted to [3slp]. This basis set
is labeled TZP. Starting from the D3,,, D2, and C, symmetry
geometries of Komornicki and McIver: we optimized the geom-
etries with the program H O N D O ~using
~ gradient techniques. The
starting C, and C2symmetry structures were modified to account
for the problems with the C-H bond lengths. Analytic second
derivatives* and MP-2 correlation energy correction^'^ were
calculated with the program G R A D S C F . ~ ~
Although the boat structure of C, symmetry can be optimized
by using energy minimization methods (the negative curvature
direction of a2 symmetry), the C2and C, structures cannot (the
negative curvature direction is of a l or a symmetry). The C2and
C, structures were gradient optimized by using the u optimization
in GRADSCF to find the transition state. Minimization of the
energy, E(x), with respect to the nuclear coordinates x would lead
to the D3,, symmetry structure since there are no symmetry
constraints. We therefore chose to minimize the following a(x)
function:
4 4 = cI g m
where gt is the gradient component of the ith coordinate.21 As
discussed below, the C, symmetry structure is not a transition state
but rather a higher energy stationary point with the two negative
curvature directions. The direction corresponding to the smaller
of the two imaginary frequencies is asymmetric, breaking C,
symmetry. The C, symmetry geometry was distorted along the
a direction yielding a Cl symmetry structure, which was then
gradient optimized by using the u optimization.
(10) Laskowski, 8. C.; Jaffe, R. L.; Komornicki, A. b t . J. Quanrum Chem.
1986, 29, 563. Laskowski, B. C.;Jaffe, R. L.; Komornicki, A. Applied
Quantum Chemistry; Smith, V. H., Jr., Ed.; D. Riedel: Dordrecht, The
Netherlands, 1986;p 347.
( 1 1 ) Dixon, D. A. J . Phys. Chem. 1986, 90, 2038. Dixon, D. A.; van
Catledge. F. A. Inr. J . Supercomput. Appl. 1988, 2,no. 2,62.
(12)Dunning, T. H., Jr.; Hay, P. J. In Methods ofElectronic Structure
Theory; Scheafer 111, H. F., Ed.; Plenum Press: New York, 1977;p 1 .
(13) Huzinaga, S.Comput. Phys. Rep. 1985,2, 279.
(14)Dunning, T. H. J . Chem. Phys. 1971,55, 716.
(15) Stewart, R. F. J . Chem. Phys. lW9,50, 2485.
(16)(a) Dupuis, M.; Rys, J.; King, H. F. J . Chem. Phys. 1976,65,1 1 1 .
(b) King, H. F.; Dupuis, M.; Rys, J. National Resource for Computer
Chemistry Software Catalog, Vol 1, Program QH02 (HONDO), 1980.
(17) (a) Komornicki, A.; Ishida. K.; Morohma, K.; Ditchfield, R.; Con-
rad, M. Chem. Phys. Lett. 1977.45.595. McIver, J. W.;Komornicki, A. Ibid.
1971,10,303.(b) Pulay. P. In Applications ofElectronic Structure Theory;
Schaefer 111, H. F., Ed.;Plenum Press: New York, 1977;p 153.
(18)King, H. F.;Komornicki, A. In Geometrical Deriuatiues of Energy
Surfaces and Molecular Properties; Jergmson, P., Simons, J., as.;
NATO
AS1 series C. Vol. 166,D; Reidel: Dordrecht, 1986,p 207. King, H. F.;
Komornicki, A. J . Chem. Phys. 1986,84,5645.
(19) (a) Meller, M. S.;Plessct, Phys. Reu. 1934,46,618. Pople, J. A.;
Binkley, J. S.;Seeger, R. Int. J . Quantum Chem. Symp. 1976,10, 1 .
(20)GRADSCF, an ab initio gradient program system designed and written
by A. Komornicki at Polyatomics Research Institute.
(21)Mclver, J. W., Jr.; Komomicki, A. J. Am. Chem.Soc. 1972,94,2625.
5632 The Journal of Physical Chemistry, Vol. 94, No. 14, 1990
TABLE I: Conford beqetks ( k d / p d ) for C y c k k x r ~
g" AEoh(SCF) AEO4MP-2)
DZ+P Basis
chair (Dy) ff od
twist boat (4) 6.79 6.95
boat (Cd 7.92 8.44
cz (W 12.21 12.42
c* 12.44 12.92
CI (TS) 12.25 12.49
TZ+P Basis
chair (Du) 0.V 0.d
twist boat (D2) 6.95 6.82
boat (Cd 8.10 8.30
Cz (TS) 12.35 12.30
Ct US) 12.42 12.38
C8 12.62 12.82
a Number of imaginary frequencies; zero-point corrections utilized the DZ+P basis set results. Magnitude of any imaginary frequencies. E-
(SCF)= -234.237 518 au. dE(MP-2) = -235.003660 au. 'E(SCF) = -234.283824 au. fE(MP-2) = -235.191 304 au.
Results and Discussion
The lowest energy structure as expected is the chair form of
Dw symmetry, and the second derivative analysis shows that it
is a true minimum. Energies for the other structures relative to
the chair structure are given in Table I. Molecular graphics views
of the five important structures are shown in Figure 1. Important
geometry parameters for the various structures are given in Table
11. The Cartesian coordinates of the five important structures
are given as supplementary material (see the paragraph at the
end of the paper).
Only the structure of the chair form has been determined
experimentally.22 The results from electron diffraction are also
given in Table 11. The agreement between the electron diffraction
structure and our calculated structure is quite good considering
that in the experimental refinement the hydrogens were assigned
equal bond distances and angles. We find B(CCH,) to be greater
than B(CCH,,) by 1.1" and our value for B(H,CH,) is 3.4"
smaller than the experimental value. We do note that the positions
of the hydrogens are very difficult to obtain from experiment.
The twist-boat structure of Dz symmetry is also a minimum
and lies 6.8 kcal/mol higher than the chair at the SCF level and
7.0 kcal/mol higher at the MP-2 level with the DZP basis. The
geometry of the twist-boat structure is similar to that of the chair
structure. The bond distance r(C C,) increases by 0.01 A while
A
r(C2C2/)decreases by only 0.003 relative to r(CC) in the chair
structure. The ring angles B(C2ClCi) and B(CIC2C2))both in-
crease relative to O(CCC) in the chair structure. The two torsional
angles about the C & ' bonds increase by 6", whereas the other
four C-C torsional angles decrease by 26" when compared to the
chair structure. The parameters involving the hydrogens show
only small changes.
The boat structure of C , symmetry is a transition state. The
magnitude of the frequency in the negative curvature direction
is calculated to be 104i an-'.This boat structure lies 7.9 kcal/mol
higher than the chair at the SCF level and 8.4 kcal/mol higher
at the MP-2 level with the DZP basis. Relative to the Dz sym-
metry twist-boat, the C , symmetry boat is 1,1 kcal/mol higher
at the SCF level and 1.5 kcal/mol higher at the MP-2 level with
the DZP basis. The negative curvature direction connects two
twist-boat structures with the symmetric C, boat. This low value
of the imaginary frequency is consistent with the low 103 cm-'
lowest vibrational mode calculated in the twist-boat structure. The
value of r(C2Ci) increases by 0.021 A and r(CICz)decreases by
0.01 A in the boat structure as compared to the twist-boat structure
and is similar to the value for r(CC) in the chair conformation.
The bond angle O(C,C,Ci) increases by 1.4O and O(C2CIC2/)
decreases by 0.3" in the boat structure as compared to the
twist-boat structure. Both bond angles are still larger than e(CCC)
in the chair structure. The torsional angles about the two C , + '
bonds are zero by symmetry, and torsions about the other four
C-C bonds are 5" less than the value in the chair structure.
(22) Giese, H. J.; Buijs, H. R.; Mijihoff. J. J . Mol. Strucr. 1971, 9, 447.
Dixon and Komornicki
N. freqb AZPE AH(0 K) AH(298 K)
0 0
0 -0.13
1 104 -0.19
1 243 -0.15
2 248,84
1 243 -0.19
0.0 0.0 0.0
-0.13 6.7 6.9
-0.19 8.1 7.9
-0.15 12.2 12.0
-0.19 12.2 12.1
The C2symmetry structure is a transition state connecting the
chair with the twist-boat m a n i f ~ l d . ~The negative curvature
direction is of a l symmetry, and the magnitude of the imaginary
frequency is 2431 cm-I. The negative curvature direction con-
necting the twist boat structures via the C , boat is essentially
orthogonal to the motion connecting the twist boat with the chair.
This transition-state structure lies 12.2 kcal/mol higher than the
chair at the SCF level and 12.4 kcal/mol higher at the MP-2 level
with the DZP basis. The geometry of this transition state has
four atoms, Cl, Cl', C,, C i nearly coplanar with one atom above
(C{) and one atom below (C,) the plane, where the atoms above
and below the plane are bonded to each other. The C-C bond
lengths between the atoms in the plane are elongated by
0.018-0.023 A relative to the reactant, while the remaining three
bonds are O.OO4-0.009 A shorter than r(CC) in the chair structure.
The CCC angles at the carbons in the plane are 5.2-8.6O larger
than the angles in an idealized tetrahedral structure. The bond
angles at the carbons above and below the plane (C, and C,l) are
approximately tetrahedral. The torsional angles about the three
bonds in the plane (Cl-Cl', CI-C2,CI'-Ci) are nearly zero, two
being 7.1 " and one is 13.7". The torsional angle about the C3-C3'
bond is 14O greater and the remaining two torsional angles are
8" less than a torsional angle in the chair structure, while the
parameters for the hydrogens are similar to those in the chair
structure.
The C,symmetry structure is not a transition state but is rather
a higher energy stationary point since it has two negative curvature
directions. The larger magnitude imaginary frequency corresponds
to the totally symmetric mode, which connects the twist-boat
manifold to the chair manifold. The smaller imaginary frequency
direction is asymmetric and breaks C,symmetry. The C, sym-
metry structure is virtually isoenergetic with the C2symmetry
structure at the SCF level and 0.5 kcal/mol higher at the MP-2
level with the DZP basis.
u optimization of a CI symmetry structure derived from the
C,symmetry structure leads to a stationary point that is a tran-
sition state. The surface in this region is very flat, and large
geometry changes lead to only very small changes in energy. A
variety of structures within 0.1 kcal/mol were found that could
serve as transition states. These calculations provide support for
the conclusions of the ~emiempirical~ and molecular mechanics
studies'-' that the flat nature of the potential energy surface in
this region leads to a fluxional transition state for chair to boat
interconversion. The energy of the CI symmetry transition state
is 12.3 kcal/mol above the chair structure at the SCF level and
12.1 kcal/mol above the chair at the MP-2 level with the DZP
basis. The CIsymmetry structure is higher in energy than the
C2symmetry transition state at the SCF level. The correlation
corrections however are small and are in opposite directions for
the CI and C, symmetry structures. The CIsymmetry structure
is thus lower in energy at the MP-2level using the DZP basis.
The C,structure is characterized by an imaginary frequency of
2263 cm-l and a lowest real torsion frequency of 10 cm-'. Due
Ab Initio Conformational Analysis of Cyclohexane The Journal of PhysicaI Chemistry, Vol. 94, No. 14, 1990 5633

E
H
''

A -7 B

,/

4,
D

E
Figure 1. Molecular graphics view of (A) DM (chair), (B)D2(twist boat), (C) C, boat, (D) C2(transition state), and (E)C,(transition state) conformers
of cyclohexane. The dark atoms are the carbons and the light atoms are the hydrogens. The carbon radius is !.5 times the covalent radius and the
hydrogen radius is 2.0 times the covalent radius.
to the nature of this stationary point, as a transition state for the and could equally well be considered as a transition state for the
interconversion and lying on the pseudorotational coordinate, there chair interconversion. The value of the lowest real frequency,
is potentially a multitude of such structures occupying a large corresponding to pseudorotation in the CI structure, is lower than
volume of coordinate space that would have very similar energies the value of 52 cm-l found in the C, structure.
 5634 The Journal of Physical Chemistry, Vol. 94, No. 14, 1990 Dixon and Komornicki
TABLE 11: Optimized Geometry Parameters for the Dy, D b Ct, and CI Structures of Cycb&xa&
D2 (Twist-Boat) DM (Chair)b

.I
112.6
111.8
1.534 (1.528) e(ccc) 111.1 (111.0)
1.089 (1.119) B(CCH,,) 109.2 (108.8)
109.7 1.087 (1.119) B(CCH ) 110.3 (108.8)
109.3 55.8 (55.9) B(H,,Ca,) 106.6 (110.0)
106.0
109.6 C, (Transition State)
109.9 Hi
109.0 I
110.0
106.2

WH2 IG
112.3
1 1 3.2
1.528
1.536
1.556
1.558
1 S46
1.528
1.089
108.8
1 1 1.5
115.8
118.8
117.1
113.3
110.2
109.8 1.086 110.5
109.3 1.087 110.7
106.3 I .088 109.8
108.9 1.086 106.9
108.6 1.086 1 1 1.1
110.1 1.084 108.3
109.8 I .085 109.2
106.0 1.085 110.2
,. I .085 106.4

i 109.8
105.7
107.4
108.8
107.6
108.2
105.2
1.557 118.1 107.4
1.552 114.7 109.2
1.528 109.8 108.6
1.525 107.7 108.4
1.085 108.3 105.4
1.085 108.9 110.7
1.086 107.9 107.5
1.087 105.2 109.4
1.086 109.7 109.5
1.089 108.8 106.2
13.7 109.9
7.1 107.5
47.8 105.9
69.3 109.4
110.7
1 1 1.1
109.3
106.6
Bond distances (r) in angstroms. Bond (e) and torsion ( 7 ) angles in degrees. Experimental values in parentheses from ref 20.

The geometry of the C , symmetry transition state is best de-
scribed as a half-chair twisted away from C, symmetry. Two of
the C-C bond lengths are elongated by 0.020-0.025 A as com-
pared to r(CC) in the chair structure. There are three e(CCC)
bond angles with values 4.7-7.7O larger than e(CCC) in the chair
structure. There is one torsional angle about a C-C bond (C,-C,)
near the value found in the chair and one torsion CI-C2within
12 of the value in the chair structure. Three of the torsions have
quite small values, 50 about the C3-C4bond and = 1 2 O and =20
about the C4-C, and C,-C6 bonds. Again the parameters for the
hydrogens do not show large variations from the values for the
chair structure. The deviations from the chair structure of the
CI and C2symmetry structures are very similar, and their energies
differ by less than 0.1 kcal/mol.
Our calculated values for the CI and C2symmetry transition
states are in agreement with the gas-phase NMR A P values and
not with the solution data. Vibrational corrections are small
relative to the energy changes. To be sure that the use of a larger
basis set would not change the energetics, the relative energies
were calculated with a TZP basis set (192 basis functions as
compared to 114 basis functions for the DZP basis). The relative
energies do not change by more than 0.2 kcal/mol on improvement
Ab Initio Conformational Analysis of Cyclohexane
of the basis set. The twist-boat and boat structures become more
stable by 0.1 3 and 0.14 kcal/mol at the MP-2 level. The energy
of the C2structure also decreases by 0.12 kcal/mol. The energy
of the CI structure increases by 0.12 kcal/mol.
Knowledge of the force field at the transition state allows us
greater insight into the most important motions involved in this
interconversion. Large amplitude motions of six-membered rings
have been analyzed for many years in terms of the three lowest
frequency vibrations with the remainder of the motions either
frozen or constituting a thermal bath. This is basically the sim-
plified model of Picket and Straus~.~ In our case the normal modes
that compond to the three lowest frequency modes are all torsions
of the ring skeleton. An examination of the force constants in
terms of these symmetrized-valence coordinates shows that all
of the diagonal force constants are positive. The negative curvature
directions arise from the large and negative off-diagonal force
constants. As an example we can examine the C, symmetry
structure. In the a' block, two of the torsional force constants
are small, with values of 0.087 and 0.161 mdyn, while the off-
diagonal coupling matrix element is -0.2 15 mdyn. Likewise, in
the a" block we find that one of the torsions has a diagonal force
constant that is nearly zero, while the others are much larger. This
zero-frequency torsion is however most strongly coupled to the
C - C stretching coordinate, with the coupling being negative. It
is this 2 X 2 block of the force constant matrix which ensures that
the C, structure is a maximum and not a transition state. A similar
analysis of the C2structure indicates that the coupling in the a2
symmetry block is not strong enough to produce two negative
curvature directions. The magnitude of these torsional force
constants further indicates that a harmonic approximation may
not be an adequate model for these large amplitude motions. An
extension of these calculations to include anharmonic terms in
the potential energy surface is, however, beyond the scope of this
investigation.
We have used our calculated frequencies to include the zero-
point corrections AZPE to the isomerization energies. This yields
a AE at 0 K (for these isomerizations hE (0 K) = AH(0 K).) The
values for AH(0K) are given in Table I along with the: AZPE
corrections. A subsequent correction AAH'(298K) can be made
to correct the AH*%to 298 K (see Table I). We find that the
energy required to interconvert the chair isomers is 12.1 kcal/mol
via the C I or 12.0 kcal/mol via the C2transition states, in excellent
agreement with the gas-phase NMR result of Ross and True8 of
12.1 f 0.5 kcal/mol. For the reverse process from the twist-boat
structure to the chair, we obtain an activation energy of 5.5
kcal/mol via the CI or 5.3 kcal/mol via the C, transition states
at 298 K, in good agreement with the value of 5.27 f 0.05
kcal/mol obtained by Anet and co-workers,9 who equated AG*
with AH'. The only disagreement between theory and experiment
is with the solution-phase values for AH*for the isomerization
of the chair. This aspect is discussed below. Our results suggest
that the molecular mechanics energies for the various conformers
(e.g., those from MM-2)3bare too low relative to the chair and
that these should be reevaluated. A similar conclusion has been
reached with regard to the rotational barriers in n - b ~ t a n e . The
~~
recently published improvement in the molecular mechanics force
field,%MM3, yields a greater energy separation between the chair
and twist-boat structures when compared with the earlier MM2
values. However, the MM3 energy is still 1 kcal/mol lower that
our results. Furthermore, the MM3 predicted energies of the boat
as well as the transition state for the chair interconversion lie 1-2
kcal/mol lower than our values.
From the structural parameters and vibrational frequencies,
the entropy change, AS' for the isomerization of the chair and
twist boat can be calculated. In this case only rotation and
vibration contribute since the translational terms cancel and the
molecules are in their ground electronic states. The calculation
of the vibrational and rotational contributions to AS' is com-
plicated by the presence of the pseudorotation mode in the
(23) Ragavachari, K. J. Chem. Phys. 1984,81, 1383.
(24) Allinger, N. L.; Yuh, Y. H.; Lii, J.-H. J . Am. Chem. Soc. 1989,111,
8555.
The Journal of Physical Chemistry, Vol. 94, No. 14, 1990 5635
transition state. We have calculated AS*in two different ways.
The first is to calculate AS*for each structure from the computed
structural and vibrational data by using the standard formulas
of statistical mechanics. As a test, we calculated the value of A S
for the chair structure at 298 K to be 71.53 eu, which compares
well with the experimental= values of 71.41 f 0.2 and 71.27 eu.
The only complication to consider in the calculation of AS*is the
use of proper symmetry numbers for the transition states. There
has been considerable discussion in the literature concerning use
of symmetry numbers in absdute rate theory calculations. The
symmetry number for cyclohexane is 6 (using Du symmetry), and
following the work of Pechukas?6 we assign a symmetry factor
of 1 to the CI and C2transition states. The symmetry number
for the D2twist-boat structure is 4, and the symmetry number
for the C, boat is 2. Thus the symmetry factor contribution to
the entropy for isomerization of the chair through the C, or C2
transition states is 3.56 eu, and for the twist-boat through the CI
or C2transition states is 2.75 eu. The value of AS' from the chair
through the CI transition state is then 8.19 eu (using a lowest real
torsional frequency of 10 cm-') and from the chair through the
C,transition state it is 4.98 eu. The actual value of AS*calculated
in this fashion is very dependent on the value of the lowest real
frequency for the transition state. The average of our two cal-
culated values for AS' is 6.58 eu. Alternatively, one can calculate
AS*following the treatment of Pickett and Strauss. From their
work, we can back calculate a value for AS' of 6.22 eu at 300
K for the pseudorotation motion. This value incorporates both
the low pseudorotation vibrational mode and the appropriate
symmetry number changes. Values of AS*for the chair, CI,and
C2structures can then be calculated by using symmetry numbers
of one and neglecting the lowest lying vibrational mode (pseu-
dorotation) in the transition states. The values for AS*for the
CI and Cz structures now only differ by 0.08eu. We calculate
2.52 eu for the CI transition state and 2.60 eu for the C2transition
state. Thus the entropies of activation for the chair to chair
interconversion process are positive by either method. The two
sets of calculated values bracket the value of 5.7 f 0.5 eu obtained
by Ross and True.8 The solution-phase value7bfor AS*of 2.8
f 0.5 eu is in reasonable agreement with the lower range of our
calculated values. Hbffner2' et al. report a value of AS*= 4.6
eu in solution in reasonable agreement with the value of Ross and
Truee8 !t is important to note that the dominant contributions
to AS*are from the symmetry number changes and from the low
frequency of the pseudorotation mode. Without these contribu-
tions AS' would be zero or negative.
Given the above discussion, values for AG* (Table 111) for the
chairchair interconversion can be calculated for comparison with
experimental values. Our calculated values range from 1 1.3 to
10.5 kcal/mol at 300 K depending on the value of AS* used. This
agrees with the high end of the range of experimental data which
goes from True's value of 10.4 f 0.2 kcal/mol to the solution
values of 10.0 and 10.1 kcal/mol. These results indicate that the
value of AG* does not depend strongly on the phase even though
there is a much stronger dependence for the values of AH*and
AS'. Our results are in agreement with the experimental gas-
phase values for AH* and are consistent with the solution- and
gas-phase values for AS'. The results suggest that solvation
decreases AH* even for this nonpolar hydrocarbon. The NMR
experiments of Jonas and c o - w o r k e r ~provided
~~ the first exper-
imental proof of the predictions of stochastic models for reactions
in solution. They showed that the ring inversion in cyclohexane
in a variety of solvents is accelerated on an increase in pressure.
The observed activation volume is dependent on the pressure and
on the solvent viscosity and is negative. These results are consistent
(25) Dorofeeva, 0. V.; Gurvich, L. V.; Jorish, V. S. J . Phys. Chem. Re/.
Dam 1986. 15, 437. Aston, J. G.; S m z . G. J.; Fink, H. L. J. Am. Chem. Soc.
1943,65, 1135. Beckett, C. W.; Pitzer, K. S.; Spitzer, R. J. Am. Chem. Soc.
1947.69.2488.
(26) Pollak, E.; Pechukas, P. J . Am. Chem.Soc. 1978,100,2984. Coulson,
R. D. J . Am. Chem. Soc. 1978,100.2992.
(27) Hbffner, D.; Lesko, S. A.; Binsch, G.Org. Mugn. Reson. 1978, 11,
179.
5636 The Journal of Physical Chemistry, Vol. 94, No. 14, 1990 Dixon and Komornicki
TABLE III: AS' (a)
rod AG' (kd/md) CalcuhtioaS for The calculated values of AS* and AG*are in -good agreement
- with
Cyclohexane Isomerizations the experimental values.
C,(strd) Cz(strd) C,(pseudo) Cz(pseudo) expt The calculations provide additional insights into the behavior
Chair-Chair (300 K) of the various conformers. There are the expected significant steric
AS' 8.19 4.98 2.52 2.60 5.7 f 0.5 interactions due to torsional angles, which are close to zero degrees,
AH' 12.1 12.0 12.1 12.0 12.1 f 0.5 in all of the conformers except for the chair. There are also
AG' 9.6 10.5 11.3 11.3 10.4 f 0.2 distortions in the C C bond lengths and CCC bond angles in the
C,and C2transition states. The energetics of the various con-
Twist BoatChair (100 K) formers differ from the values usually assumed in classical organic
AS* 4.06 3.64 2.70 2.70
AH' 5.5 5.3 5.5 5.3 texts.B Our calculated values of the relative energies of the twist
AG' 5.1 5 .O 5.2 5.0 5.27 f 0.05 boat and boat are roughly 2 kcal/mol higher than the values
usually assigned. Furthermore, the geometry of the boat structure
"Standard: calculated by using standard formulas of statistical me- distorts so that the 'flagpole" hydrogens are 2.47 A apart as
chanics. Pseudo: calculated by using the pseudorotational contribu- opposed to the values of 1.8-1.85 A normally used. This is ac-
tion to hs' from ref 2. See text. complished by bending the CCC bond angles and lengthening of
the CC bonds. The usual models of the energetics of cyclohexane
with the difference in energies that we calculate for the free accord 4 kcal/mol to eclipsing of eight C-H bonds and 2 kcal/mol
molecule and those observed in solution. to the "flagpole" interactions. We would suggest that it is more
The other experimental value that can be compared with these appropriate to attribute 6 kcal/mol for eclipsing interactions (four
calculations involves AG* for the twist boat to chair intercon- CHI groups and eclipsed CC bonds). The remaining two kcal/mol
version. The value of AS' is 2.26 eu for the twist boat to the CI are due to distortions that relieve the "flagpole" interactions.
transition state and is 1.80 eu for the twist boat through the C, The calculations also indicate that a variety of structures are
transition state at 298 K. At 100 K, the value of AS* calculated essentially isoenergetic within the limits of reliability possible in
in the traditional fashion is 4.06 eu through the CI transition state such calculations. Furthermore, a discussion of a transition-state
and 3.64 through the C2 transition state. The values of AG* structure as belonging to either of the above-named symmetries
(loOK) are 5.1 and 5.0 kcal/mol through the C, and C, transition loses meaning, especially if we consider the very small energy
states, respectively, in good agreement with the experimental value differences found here. Although we have estimated the corre-
of 5.27 f 0.05 kcal/mol at 70 K. Values for AS* were not lation energy corrections, these were done using SCF-optimized
calculated below 100 K as the classical approximations for the geometries. Further investigations using a more sophisticated
translational and rotational contributions are no longer valid. We treatment, which would include the optimization and evaluation
can calculate AS' approximately, by subtracting the symmetry of force constants, might indicate that correlation changes the
number contribution to the entropy (3.56 eu) from 5.51 eu, the computed force field sufficiently to alter the description of either
value of AS' due to the pseudorotation and symmetry number the C2 or C, structures as transition states or maxima. Fur-
contribution at 100 K. Adding 2.75 eu for the symmetry number thermore, a more extensive treatment should describe the large
change going from the twist boat to the chair to this gives an amplitude motions in a more rigorous manner, other than the
estimated value of 4.70 eu for the pseudorotation/symmetry simple harmonic treatment done here. However, the agreement
number component of AS'. This yields AS*= 2.70 eu for both with the experimental AG* and AS* values shows the present
the C, and C2transition states, which yields a AG*= 5.2 kcal/mol treatment to be very reliable.
for the isomerization proceeding through the CI transition state
and 5.0 kcal/mol via the C2transition state. These numbers are Acknowledgment. A.K. acknowledges support from NASA
in good agreement with the experimental value. through Grant NCC2-499. We also thank H. Strauss and W.
E. v. Doering for helpful discussions.
Conclusions
Registry No. Cyclohexane, 110-82-7.
Our calculations show that the chair structure and twist-boat
structures are minima on the conformational energy hypersurface Supplementary Material Available: Cartesian coordinates of
with the twist-boat structure 6.9 kcal/mol above the chair. The the five important structures (3 pages). Ordering information
various twist boats are connected by a boat transition state 7.9 is given on any current masthead page.
kcal/mol above the chair. There are two transition states that
are approximately isoenergetic connecting the chair and twist-boat
manifolds. Transition states of C2and CI symmetry are 12.1 (28) Eliel, E. L.; Allinger, N. L.; Angyal, S. J.; Morrison, G. A. Confor-
kcal/mol above the chair and 5.1 kcal/mol above the twist-boat mational Analysis, American Chemical Society: Washington, D.C., 198 1.
EM, E. L. Stereochemistry of Carbon Compounds; McGraw-Hill: New
at 300 K. These results are consistent with a pseudorotating York, 1962. Roberts, J. D.; Caserio, M. C. Basic Principles of Organic
transition state connecting the chair and the twist boat manifolds. Chemistry; W.A. Benjamin: New York, 1965; pp 103-1 11.