Published on Web 04/02/2008

The Mechanism of the Rhodium(I)-Catalyzed [2 + 2 + 1]

Carbocyclization Reaction of Dienes and CO: A
Computational Study
William H. Pitcock Jr., Richard L. Lord, and Mu-Hyun Baik*
Department of Chemistry and School of Informatics, Indiana UniVersity,
Bloomington, Indiana 47405
Received February 3, 2008; E-mail: [email protected]

Abstract: The rhodium(I) catalyzed [2 + 2 + 1] carbocyclization of tethered diene-enes to afford substituted

hexahydropentalenones with high levels of diastereoselectivity was modeled using density functional theory.
Previously, this transformation was observed to be facile, whereas the analogous bis-ene substrate could
not be cyclized under any reasonable conditions. To establish a conceptual understanding of the
carbocyclization mechanism and to identify the functional role of the diene fragment we analyzed the
simulated reaction mechanisms using the two parent systems. We discovered a thus far unrecognized,
but intuitively plausible, role of the CO ligand for controlling the electron density at the metal center, which
affects the feasibility of oxidative addition and reductive elimination steps that are key components of the
mechanism. Our calculations suggest that the diene moiety is unique and required because of its ability to
undergo a η2f η4 reorganization allowing for the thermoneutral expulsion of one CO ligand, which in turn
generates an electron-rich, coordinatively saturated Rh(I) center that can efficiently promote the oxidative
addition with a low barrier. A number of functionalization strategies were considered explicitly to derive a
rational plan for optimizing the catalysis and to expose the roles of the different components of the
reactant-catalyst complex.

Introduction alkene-ynes, Wender postulated that the diene fragment could

be used as a replacement for alkynes.31–34 Indeed, the use of
The use of transition metals to promote otherwise difficult tethered diene-enes allowed for the convenient construction of
organic transformations is a powerful approach to constructing cyclopentanones. Curiously, the presence of the diene fragment
complex polycyclic systems efficiently. In particular, the is critical, as analogous bis-ene substrates failed to react under
exquisite control of stereochemistry displayed by transition metal
catalysts and the ability to install multiple stereocenters (11) Wender, P. A.; Husfeld, C. O.; Langkopf, E.; Love, J. A. J. Am. Chem.
simultaneously are two of the most attractive features of Soc. 1998, 120, 1940–1941.
organometallic catalysis. While the breadth of the [m + n + o] (12) Wender, P. A.; Barzilay, C. M.; Dyckman, A. J. J. Am. Chem. Soc.
carbocyclization variants is wide (e.g., [4 + 2], [5 + 2], [4 + 2001, 123, 179–180.
(13) Evans, P. A.; Robinson, J. E.; Baum, E. W.; Fazal, A. N. J. Am. Chem.
2 + 2], etc.),1–18 the focus of this work is on the three component Soc. 2002, 124, 8782–8783.
rhodium(I)-catalyzed [2 + 2 + 1] carbocyclization. One of the (14) Evans, P. A.; Robinson, J. E.; Baum, E. W.; Fazal, A. N. J. Am. Chem.
apparent requirements for the [2 + 2 + 1] carbocyclization is Soc. 2003, 125, 14648–14648.
(15) Gilbertson, S. R.; DeBoef, B. J. Am. Chem. Soc. 2002, 124, 8784–
a “complex” π-system, either alkynes or allenes.19–30 Building 8785.
off the observation that diene-ynes reacted much faster than (16) Evans, P. A.; Baum, E. W. J. Am. Chem. Soc. 2004, 126, 11150–
11151.
(17) Evans, P. A.; Baum, E. W.; Fazal, A. N.; Pink, M. Chem. Commun.
(1) Wender, P. A.; Jenkins, T. E. J. Am. Chem. Soc. 1989, 111, 6432– 2005, 63–65.
6434. (18) Baik, M. H.; Baum, E. W.; Burland, M. C.; Evans, P. A. J. Am. Chem.
(2) Wender, P. A.; Jenkins, T. E.; Suzuki, S. J. Am. Chem. Soc. 1995, Soc. 2005, 127, 1602–1603.
117, 1843–1844. (19) Evans, P. A.; Robinson, J. E. J. Am. Chem. Soc. 2001, 123, 4609–
(3) Jolly, R. S.; Luedtke, G.; Sheehan, D.; Livinghouse, T. J. Am. Chem. 4610.
Soc. 1990, 112, 4965–4966. (20) Khand, I. U.; Knox, G. R.; Pauson, P. L.; Watts, W. E. J. Chem. Soc.,
(4) Gilbertson, S. R.; Hoge, G. S. Tetrahedron Lett. 1998, 39, 2075–2078. Chem. Commun. 1971, 36.
(5) Gilbertson, S. R.; Hoge, G. S.; Genov, D. G. J. Org. Chem. 1998, 63, (21) Hanson, B. E. Comments Inorg. Chem. 2002, 23, 289–318.
10077–10080. (22) Brummond, K. M.; Kent, J. L. Tetrahedron 2000, 56, 3263–3283.
(6) Wender, P. A.; Takahashi, H.; Witulski, B. J. Am. Chem. Soc. 1995, (23) Koga, Y.; Kobayashi, T.; Narasaka, K. Chem. Lett. 1998, 249–250.
117, 4720–4721. (24) Kobayashi, T.; Koga, Y.; Narasaka, K. J. Organomet. Chem. 2001,
(7) Wender, P. A.; Sperandio, D. J. Org. Chem. 1998, 63, 4164–4165. 624, 73–87.
(8) Wender, P. A.; Glorius, F.; Husfeld, C. O.; Langkopf, E.; Love, J. A. (25) Jeong, N.; Lee, S.; Sung, B. K. Organometallics 1998, 17, 3642–
J. Am. Chem. Soc. 1999, 121, 5348–5349. 3644.
(9) Wender, P. A.; Fuji, M.; Husfeld, C. O.; Love, J. A. Org. Lett. 1999, (26) Jeong, N.; Sung, B. K.; Choi, Y. K. J. Am. Chem. Soc. 2000, 122,
1, 137–139. 6771–6772.
(10) Trost, B. M.; Toste, F. D.; Shen, H. J. Am. Chem. Soc. 2000, 122, (27) Brummond, K. M.; Chen, H. F.; Fisher, K. D.; Kerekes, A. D.;
2379–2380. Rickards, B.; Sill, P. C.; Geib, S. J. Org. Lett. 2002, 4, 1931–1934.

10.1021/ja800856p CCC: $40.75  2008 American Chemical Society J. AM. CHEM. SOC. 2008, 130, 5821–5830 9 5821
ARTICLES Pitcock Jr. et al.

Scheme 1. Rhodium-Catalyzed [2 + 2 + 1] Carbocyclization of

Diene-enes and Bis-enes

similar conditions (Scheme 1). Why the diene fragment is so

important was not understood to date. Functionalized bis-ene
analogues with coordinating heteroatoms or extended π-con-
jugation (4a-f) all failed to react, thus indicating that the role Figure 1. Mechanistic rationale for diene-ene [2 + 2 + 1] with Rh(CO)2Cl
of the second double bond is not trivial. Simply providing an as catalyst.
additional metal binding site or facilitating π-conjugation is
clearly not enough to promote carbocyclization. We applied
density functional theory (DFT)35 to construct models of this
reaction with the aim of identifying the function of the diene
fragment during catalytic turnover and to highlight the deficiency
of the bis-ene system. A systematic series of structural changes
of the reactant-catalyst complex was constructed to highlight
Figure 2. Possible Rh(CO)2Cl chlorine isomers during initial binding.
our conceptual understanding of the mechanism.
Computational Details minima or saddle points of the potential energy surface. Note that
by entropy here we refer specifically to the vibrational/rotional/
All calculations were carried out using the spin-restricted DFT translational entropy of the solute(s); the entropy of the solvent is
formalism as implemented in the Jaguar 6.036 suite of ab initio implicitly included in the dielectric continuum model. Solvation
quantum chemistry programs. Geometry optimizations were per- energies were evaluated by a self-consistent reaction field (SCRF)44,45
formed with the B3LYP37–39 functional and the 6-31G** basis set approach based on accurate numerical solutions of the Poisson-
with rhodium represented using the Los Alamos LACVP basis that –Boltzmann equation.46 In the results reported, solvation calcula-
includes relativistic effective core potentials.40–42 The energies of tions were carried out at the optimized gas-phase geometry
the optimized structures were reevaluated by additional single-point employing the dielectric constant of
) 10.42 (1,2-dichloroethane).
calculations on each optimized geometry using Dunning’s correla- As is the case for all continuum models, the solvation energies are
tion-consistent triple- basis set,43 cc-pVTZ(-f), that includes the subject to empirical parametrization of the atomic radii that are
standard set of polarization functions. For Rh, we use a modified used to generate the solute surface. We employ the standard set of
version of LACVP where the exponents have been decontracted optimized radii in Jaguar for H (1.150 Å), C (1.900 Å), O (1.600
to match the effective core potential with a triple- quality basis. Å), Cl (1.974 Å), Br (2.095 Å), and Rh (1.464 Å). Partial atomic
Vibrational frequency calculations based on analytical second charges reported are based on atomic charges fit to the electrostatic
derivatives at the B3LYP/6-31G** level of theory were carried out potential.47–49
to derive the zero-point energy (ZPE) and entropy corrections at The energy components have been computed following the
room temperature utilizing unscaled frequencies. Vibrational cal- standard protocol. The free energy in solution phase G(sol) was
culations were also used to confirm proper convergence to local calculated as follows:

(28) Brummond, K. M.; Chen, H. F.; Mitasev, B.; Casarez, A. D. Org. G(sol) ) G(gas) + Gsolv (1)
Lett. 2004, 6, 2161–2163.
(29) Bayden, A. S.; Brummond, K. M.; Jordan, K. D. Organometallics G(gas) ) H(gas) + TS(gas) (2)
2006, 25, 5204–5206. H(gas) ) E(SCF) + ZPE (3)
(30) Brummond, K. M.; Gao, D. Org. Lett. 2003, 5, 3491–3494.
(31) Wender, P. A.; Deschamps, N. M.; Gamber, G. G. Angew. Chem., where G(gas) is the free energy in gas phase, Gsolv is the free energy
Int. Ed. 2003, 42, 1853–1857. of solvation as computed using the continuum solvation model,
(32) Wender, P. A.; Deschamps, N. M.; Williams, T. T. Angew. Chem., H(gas) is the enthalpy in the gas phase, T is the temperature (298.15
Int. Ed. 2004, 43, 3076–3079.
(33) Wender, P. A.; Croatt, M. P.; Deschamps, N. M. Angew. Chem., Int.
Ed. 2006, 45, 2459–2462. (44) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.;
(34) Wender, P. A.; Croatt, M. P.; Deschamps, N. M. J. Am. Chem. Soc. Ringnalda, M. N.; Sitkoff, D.; Honig, B. J. Phys. Chem. 1996, 100,
2004, 126, 5948–5949. 11775–11788.
(35) Parr, R. G.; Yang, W. Density Functional Theory of Atoms and (45) Friesner, R. A.; Murphy, R. B.; Beachy, M. D.; Ringnalda, M. N.;
Molecules; Oxford University Press: New York, 1989. Pollard, W. T.; Dunietz, B. D.; Cao, Y. X. J. Phys. Chem. A 1999,
(36) Jaguar 6.0; Schrödinger, Inc., Portland, Oregon, 2003. 103, 1913–1928.
(37) Becke, A. D. Phys. ReV. A 1988, 38, 3098–3100. (46) Edinger, S. R.; Cortis, C.; Shenkin, P. S.; Friesner, R. A. J. Phys.
(38) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652. Chem. B 1997, 101, 1190–1197.
(39) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. ReV. B 1988, 37, 785–789. (47) Chirlian, L. E.; Francl, M. M. J. Comput. Chem. 1987, 8, 894–905.
(40) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270–283. (48) Woods, R. J.; Khalil, M.; Pell, W.; Moffat, S. H.; Smith, V. H.
(41) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310. J. Comput. Chem. 1990, 11, 297–310.
(42) Wadt, W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284–298. (49) Breneman, C. M.; Wiberg, K. B. J. Comput. Chem. 1990, 11, 361–
(43) Dunning, T. H. J. Chem. Phys. 1989, 90, 1007–1023. 373.
5822 J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008
Rh(I)-Catalyzed Carbocyclization of Tethered Diene-Enes ARTICLES

Figure 4. Mechanistic rationale for diene-ene [2 + 2 + 1] with Rh(CO)Cl

as catalyst.

Figure 5. Possible Rh(CO)Cl chlorine isomers during initial binding.

Scheme 3. Possible CO Insertion Regioisomeric Intermediates;

Partial Charge q Is Indicated for 8

Figure 3. Reaction energy profile for diene-ene [2 + 2 + 1] using

Rh(CO)2Cl.

Scheme 2. Possible CO Insertion Regioisomeric Intermediates;

Partial Charge q Is Indicated for 9

when structures were very similar), the QST-optimized transition

states were refined by an unrestricted transition state search.
Results and Discussion
In many catalyst systems promoting organic transformations,
the exact composition of the active catalyst is not known, which
poses a difficult challenge to designing computer models of these
K), S(gas) is the entropy in the gas phase, E(SCF) is the reactions. A variety of possible structures must be explored
self-consistent field energy, i.e., “raw” electronic energy as carefully to identify a plausible catalyst complex that is most
computed from the SCF procedure, and ZPE is the zero-point consistent with both experimental and theoretical considerations.
energy. For our model system, we used the reaction of diene-ene 1 with
To locate transition states, the potential energy surface was first the most common catalyst found for this reaction, [Rh(CO)2Cl]2.
explored approximately, using the linear synchronous transit (LST) Given the presence of CO in excess, we must also consider the
method,50 followed by a quadratic synchronous transit (QST)51 presence of additional CO ligands at any point on the potential
search using the LST transition state as an initial guess. In QST,
energy surface. The precatalyst [Rh(CO)2Cl]2 is commonly
the initial part of the transition-state search is restricted to a circular
curve connecting the reactant, initial transition-state guess, and the assumed to fragment in solution to give structurally identical
product, followed by a search along the Hessian eigenvector that 14-electron d8 fragments, Rh(CO)2Cl (6), which bind the starting
is most similar to the tangent of this curve. In certain cases (i.e., material in two distinct binding modes. The first involves
binding 1 by utilizing only the alkene fragment and the proximal
olefin of the diene to yield an 18-electron species (7, Figure 1).
(50) Halgren, T. A.; Lipscomb, W. N. Chem. Phys. Lett. 1977, 49, 225–
232. After forming 7, the complex undergoes oxidative addition
(51) Peng, C. Y.; Schlegel, H. B. Isr. J. Chem. 1993, 33, 449–454. via 7-TS to form 8, in which the three contiguous stereocenters
J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008 5823
ARTICLES Pitcock Jr. et al.

Figure 6. Computed structures for diene-ene binding, oxidative addition, and metallacyclopentane.

are simultaneously set. This results in a Rh(III), 16-electron

adduct. Since the reaction is carried out under 1 atm of CO
pressure, subsequent binding of an additional CO ligand to form
the 18-electron complex 9 is feasible. This pseudo-octahedral
complex now has the appropriate geometry and proximity to
allow for facile CO insertion traversing 9-TS to form the
rhodacycle 10, which can undergo reductive elimination to form
the desired product 2 and regenerate the catalytic species 6. One
immediate question that arises is that of catalyst configuration.
While one particular orientation is shown (7, Figure 1), it is
possible to have two alternate isomers (Figure 2). Our studies
suggest that the chloride ligand prefers the axial position trans
to the pendant olefin by 3.5 kcal/mol over 7a and 0.60 kcal/
mol over 7b.
The computed structures are distorted trigonal-bipyramidal
and feature a slightly stronger binding to the isolated olefin as
opposed to the proximal olefin of the diene. Another regio-
chemical issue to consider involves the direction of CO insertion.
The CO ligand may either insert into the methine- or the
methylene-rhodium bond to afford 10 and 10a, respectively
(Scheme 2). While this detail has no effect on the overall
outcome of the reaction, it is a priori not clear how these two
possible pathways will affect the reaction mechanism and must
therefore be considered explicitly. Our calculations show that
metallacyclohexanone 10a is thermodynamically favored over
10 by 5.5 kcal/mol in solution-phase free energy. While initial
arguments for this preference could be based on sterics, a more
fundamental reason exists for this energetic advantage. The
computed partial atomic charges47–49 on the methylene and
methine carbons are -0.1 and 0.5, respectively, indicating that
the methylene carbon is more nucleophilic than the methine Figure 7. Reaction energy profile for diene-ene [2 + 2 + 1] using
carbon. This charge polarization difference reflects of course Rh(CO)Cl.
on the stronger electron-withdrawing ability of the sp2 carbons
related systems also indicated that the oxidative addition is
in the olefinic unit assisted by the resonance stabilization of
typically rate determining in Pauson-Khand [2 + 2 + 1]
the additional charge compared to a simple hydrogen fragment.
carbocyclizations.53 The final step of the reaction, the reductive
The complete reaction energy profile for the mechanism elimination to complete the carbocyclization, is associated with
outlined above is illustrated in Figure 3. Oxidative addition is transition state 10a-TS at a relative energy of 17.3 kcal/mol.
rate determining with a prohibitively high barrier of 32.1 kcal/ Although 7 is a plausible starting point for the carbocycliza-
mol, which would translate to a very slow reaction even at tion, it is also possible that the diene fragment prefers to bind
elevated temperatures. The remaining steps of the reaction in an η4 fashion instead of the η2 binding mode shown in Figure
mechanism are reasonable with low barriers connecting each 1. In this case, the 12-electron fragment Rh(CO)Cl serves as
of the intermediates. While a plausible transition state could the catalytically competent species giving rise to a reaction
not be obtained for the migratory insertion of CO despite pathway shown in Figure 4. 12 is unlikely to exist, however,
intensive efforts, we do not expect the barrier to be high enough as CO loss from 6 is unfavorable by 44.8 kcal/mol. Analogous
to give this step any physical meaning. Ziegler and co-workers to the mechanism discussed above, initial binding affords adduct
demonstrated that migratory insertion of CO into Rh-alkyl 13 that undergoes oxidative addition through transition state 13-
bonds is relatively facile in methanol carbonylation, as compared TS to give metallacycle 14. Addition of a CO ligand gives
to Ir where insertion is rate determining.52 Previous studies on intermediate 8, followed by CO insertion giving access to the

(52) Cheong, M.; Schmid, R.; Ziegler, T. Organometallics 2000, 19, 1973– (53) Wang, H.; Sawyer, J. R.; Evans, P. A.; Baik, M. H. Angew. Chem.,
1982. Int. Ed. 2008, 47, 342–345.
5824 J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008
Rh(I)-Catalyzed Carbocyclization of Tethered Diene-Enes ARTICLES

Figure 8. Combined reaction energy profile for diene-ene [2 + 2 + 1].

metallacyclohexanone 15, which can reductively eliminate with a free energy barrier of 21.0 kcal/mol to give intermediate
product 2 regenerating the catalyst 12. As was the case with 14 that is 1.4 kcal/mol lower in energy than the reactant
Rh(CO)2Cl, it is possible for catalyst 12 to be bound to the complex. CO binding to 14 to form 8 should be facile, albeit
diene-ene in one of two orientations (Figure 5). Our calculations slightly uphill. While the CO insertion transition state 8-TSa
indicate that isomer 13 is energetically preferred by ∼2 kcal/ could not be located, it can be estimated in a similar fashion as
mol over 13a. For understanding the role of the diene fragment, for the reaction of Rh(CO)2Cl. Interestingly, the rate-determining
it is crucial to recognize that binding the diene fragment in an step for this mechanistic route is not the oxidative addition,
η4 fashion to the metal center gives rise to a three-legged piano unlike what we found for the “high CO pressure” manifold
stool type of coordination geometry. This structural feature is described above. With a free energy of activation of 42.4 kcal/
maintained throughout the oxidative addition, and even in mol associated with the transition state 15a-TS, the reductive
metallacycle 14 where the metal center is in a slipped σ-allyl elimination has become the rate-determining step. A kinetic
environment, as illustrated in Figure 6. This system also allows barrier of this magnitude cannot be overcome under realistic
for the formation of regioisomeric intermediates (Scheme 3). conditions.
For this pathway, CO insertion into the methylene-rhodium The two slightly different reaction pathways discussed above
bond to form 15a is favored over 15 by 3.1 kcal/mol and the illustrate an interesting mechanistic scenario. In the high CO
partial charges indicate again that the methylene carbon is pressure case, the CO insertion and reductive elimination are
significantly more nucleophilic than the methine carbon. highly accessible with low barriers, but the initial oxidative
The full reaction energy profile for the carbocyclization of addition is difficult to complete. In contrast, the low CO pressure
diene-ene 1 using 12 as the catalytically competent metal case gives kinetically facile oxidative addition, but affords
complex is shown in Figure 7. We label this profile as the “low intermediates that are unable to complete the reaction due to
CO pressure case,” as we expect the low-coordination Rh- high barriers for the reductive elimination step. The two
complex to exhibit increased population under reduced CO pathways considered above contain the common intermediate
pressure conditions, whereas Rh(CO)2Cl discussed above is 8, which can serve as a lynchpin to combine the two mechanistic
expected to be the dominant composition of the catalyst under scenarios into one reaction energy profile, shown in Figure 8.
high CO-pressure conditions. The oxidative addition step Initial complexation leads to 7, followed by loss of a CO
associated with the transition state 13-TS is easily accessible molecule to give the η4-complex 13. Interestingly, this is a
J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008 5825
ARTICLES Pitcock Jr. et al.

Figure 9. Mechanistic rationale for bis-ene [2 + 2 + 1] with Rh(CO)2Cl

as catalyst.

Figure 10. Rh(CO)2Cl positional isomers for bis-ene binding.

Table 1. Energy Components for the Intramolecular Binding of the

Diene vs the Intermolecular Binding of CO in kcal/mol
step ∆H(gas) -T∆S(gas) ∆G(gas) ∆Gsolv ∆G(sol)
7 f 7-TS 34.18 1.14 35.32 -3.22 32.10 Figure 11. Reaction energy profile for bis-ene [2 + 2 + 1] using
7 f 13 12.40 -8.73 3.67 -2.97 0.70 Rh(CO)2Cl.
13 f 13-TS 20.82 -13.15 21.75 -0.79 20.96

thermoneutral process, suggesting that the η4-binding compen-

sates perfectly for the loss of CO. Note that this transformation
should be entropically favored by approximately 10 kcal/mol
at room temperature due to the release of translational entropy
of CO. Since the entropy change for the intramolecular Rh-η2
f Rh-η4 rearrangement is expected to be significantly lower,
the binding enthalpy of the diene does not have to fully
compensate for the loss of Rh-CO bond energy. The computed
energy components are shown in Table 1 and confirm our
intuitive estimates. The 7 f 13 transformation, which encom-
passes the expulsion of CO and rearrangement of the diene
fragment, is enthalpically (∆H) uphill by 12.4 kcal/mol,
reflecting on the fact that the Rh-CO bond is stronger than the
new Rh-π bond formed. However, the release of CO is
entropically favored, and our calculations estimate the energy
gain associated with this process to be 8.7 kcal/mol. Differential Figure 12. Mechanistic rationale for bis-ene [2 + 2 + 1] with Rh(CO)Cl
as catalyst.
solvation energy contributes 3.0 kcal/mol in favor of 13 to make
this transformation overall thermoneutral. These energy com- transition state 8-TSa to form rhodacyclohexanone 15a, where
ponents are important and will become highly relevant for another bifurcation is encountered. The unproductive reductive
understanding the role of the diene, as will be discussed below. elimination barrier via 15a-TS prohibits conversion to the
The thermoneutrality suggests that a rapid pre-equilibrium may desired product along path C, however, 15a could also directly
be established between the two different Rh-complexes in the bind CO to form 10a on path B. Alternatively, if path B is
presence of excess CO. The oxidative addition is best ac- followed from common intermediate 8 then CO binding leads
complished by following path B with an activation energy of to rhodacyclopentane 9 which followed by migratory insertion
21.7 kcal/mol to form the slipped σ-allyl rhodacycle 14 with a also leads to 10a. Both post rate determining pathways to 10a
solution-phase free energy of -0.7 kcal/mol. Binding of another are likely operable and reductive elimination through this
CO molecule leads to the common intermediate 8, where intermediate leads to the product-catalyst adduct 11. Oxidative
pathway bifurcation is encountered again. If path C is followed, addition is rate determining with a barrier of 21.7 kcal/mol in
the system will undergo CO insertion associated with the this combined final mechanism, suggesting that the carbocy-
5826 J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008
Rh(I)-Catalyzed Carbocyclization of Tethered Diene-Enes ARTICLES

Figure 13. Reaction energy profile for bis-ene [2 + 2 + 1] using Rh(CO)Cl.

Table 2. Energy Component for the Binding of Various π-Bases in one CO ligand is less effective in promoting reductive elimina-
the Bis-ene System in kcal/mol
tion than intermediate 10a, which is formed by coordinating
step ∆H(gas) T∆S(gas) ∆G(gas) ∆Gsolv ∆G(sol) an additional CO. Our calculations also suggest a potential
17 f 22 27.94 -10.83 17.10 -3.62 13.48 functional role for the diene moiety. The rate-determining step
22 f 26 -5.61 11.22 5.61 3.58 9.19 is the oxidative addition, as discussed above. The coordination
26 f 26-TS 31.53 2.34 33.88 -3.38 30.50 of the electron-rich π-component after removal of the CO ligand
22 f 27 -2.55 9.23 6.68 4.47 11.15
26 f 27-TS 29.21 2.69 31.91 -0.40 31.51
from the initial reactant complex 7 helps to further increase the
22 f 28 -3.29 13.32 10.02 4.61 14.63 electron density of the metal center, thus, promoting oxidative
26 f 28-TS 29.55 1.08 30.63 -4.16 26.47 addition. To better understand this synergistic interplay between
22 f 29 0.44 10.70 11.14 5.23 16.37 the CO and diene ligand, we examined the simple bis-ene
26 f 29-TS 26.69 3.36 30.05 -5.22 24.83 substrate 4 that cyclizes to afford 5, as summarized in Figure
9.
clization of 1 should be a facile process, in good agreement Analogous to the mechanism discussed above, initial adduct
with experimental observations.34 17 can undergo oxidative addition traversing the transition state
The different reactivity as a function of the number of CO 17-TS to afford the metallacyclopentane 18. CO binding
ligands coordinated to the Rh-center is fairly easy to understand. followed by insertion affords metallacyclohexanone 20, which
For the oxidative addition it is desirable for the metal to be can reductively eliminate to give 4 and regenerate the catalyst
electron rich. In 7 the π-withdrawing nature of the CO ligand 6. Again, we observe an energetic preference for the chloride
reduces the electron density at the rhodium center, making the ligand being in the axial position disposed trans to the olefinic
oxidative addition more difficult. CO loss and the subsequent methine protons by 8.8 kcal/mol over 17a and 3.0 kcal/mol over
η4 binding of the diene lead to a much more electron-rich metal 17b (Figure 10). Unlike the diene-ene system, CO insertion
center, which facilitates the oxidative addition. Analogously, isomers need not be considered in the bis-ene system as the
reductive elimination prefers an electron-poor metal center and reactant 4 is symmetrical.
the π-acidity of the CO ligand can be used again to control the The reaction energy profile shown in Figure 11 suggests that
electron density at the metal center. Intermediate 15a with only the reaction will not occur under realistic conditions. The
J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008 5827
ARTICLES Pitcock Jr. et al.

Figure 14. Binding and activation energies for π-donors in the bis-ene system.

oxidative addition transition state 17-TS is prohibitively high state, leading to the high activation energy. In conclusion, our
in energy, giving a barrier of 36.7 kcal/mol. As was the case calculations provide a plausible explanation for why the
with the diene-ene, the latter part of the catalytic cycle is carbocyclization of the bis-ene substrate is not possible. For
reasonable, with the highest activation energy being that of the the oxidative addition step to become kinetically feasible, one
reductive elimination at 20.2 kcal/mol. This activation energy of the CO ligands has to be replaced by a less electron-
is low because the CO ligands continue to make the metal withdrawing ligand, which was conveniently possible in the
electron deficient, allowing for easy reduction. diene reactant by binding the diene fragment in η4-fashion. With
We have considered that the metal fragment Rh(CO)Cl may this conceptual picture at hand, it is easy to understand why
act as the catalytically competent species (Figure 12). In the reactant functionalizations shown in Scheme 1 failed to give
absence of the diene-fragment that can provide an additional the desired reaction. Most notably, our proposal suggests that
π-ligand fragment to occupy the empty coordination after the conceptually plausible π-conjugation does not play a
removal of the CO ligand, this mechanistic scenario gives rise significant role. Thus, replacing the vinyl moiety of 1 with a
to a four coordinate Rh(I) 16-electron complex 22 as the reactant phenyl group in 4f or any other noncoordinating π-components
complex. The computed reaction energy profile is shown in are not expected to facilitate the desired carbocyclization.
Figure 13. Not surprisingly, the oxidative addition transition Our mechanistic analysis of the [2 + 2 + 1] carbocyclization
state 22-TS is extremely high in energy with a free energy of of bis-ene substrates suggests a few rational design strategies.
43.5 kcal/mol. Thus, this reaction is unlikely to occur under First of all, we may question whether or not the intramolecular
any reasonable reaction conditions. The metal center in the five- π-base binding to afford the 18-electron complex 13 may be
coordinate 18-electron complex 17 is already electron deficient mimicked by an intermolecular π-base. Several π-base ligands
by virtue of the two CO ligands, and removing one of the CO were evaluated. The ligand binding energies and the activation
ligands that function as σ-donors in addition to being π-with- energies for the oxidative addition are summarized in Table 2
drawing ligands results in an even more electron-deficient 16- and visualized in Figure 14. As predicted, the presence of the
electron complex that is unable to promote the oxidative addition additional π-base makes the oxidative addition easier. For
reaction. In a similar arrangement as with the diene-ene example, the energy difference between complex 29, where
Rh(CO)Cl case, the reductive elimination step is also prohibi- isobutene is used as the additional π-base ligand, and the
tively high, with an activation energy of 33.7 kcal/mol. Relative corresponding transition state 29-TS is 24.8 kcal/mol, which is
to 20-TS in the Rh(CO)2Cl bis-ene catalytic cycle, 24-TS is notably lower than the energy difference between 22 and 22-
electron rich, which results in a destabilization of the transition TS that was computed to be 30.0 kcal/mol (Figure 13).
5828 J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008
Rh(I)-Catalyzed Carbocyclization of Tethered Diene-Enes ARTICLES

Scheme 4. Computed Energies of [2 + 2 + 1] Using Rh(CO)3+ as

Catalyst (Energies Are Given in kcal/mol)

However, the binding of the π-ligand is in all cases energetically

uphill, unfortunately. In the case of isobutene, the intermolecular
ligand exchange of CO with the π-base ligand requires a free
energy penalty of 29.9 kcal/mol, resulting in an overall activation
barrier of 54.7 kcal/mol relative to the initial reactant complex
17. The energy components enumerated in Table 2 reveal that
the entropic advantage of releasing CO, observed for the
intramolecular process 7 f 13 (Table 1) is lost. Whereas the
release of CO, 17 f 22, is again preferred entropically by 10.8
kcal/mol, essentially the same amount of energy, 11.2 kcal/mol,
is lost when the free π-base is coordinated in the reaction 22
f 26. Similar trends can be observed for all four π-bases tested
in this work (Table 2). This observation highlights the reason
why tethering the π-base to the bis-ene manifold, as is the case
in the diene-ene substrate, is advantageous. By exchanging the
CO ligand with an intramolecular π-component, the enthalpic
benefit of ligand binding can be leveraged without having to
pay the translational entropy penalty of a conventional inter- Figure 15. Reaction energy profile for bis-ene [2 + 2 + 1] using
Rh(CO)2Br.
molecular ligand exchange. This insight also suggests that there
is nothing “magical” about the diene unit s tethering π-bases
in a fashion that will allow for a sterically relaxed binding should the halogen ligand and subsequent addition of a CO ligand does
in principle facilitate the oxidative addition in general. We are not give rise to the desired lowering of the transition-state
currently exploring this strategy, considering a number of energy, as discussed above, we questioned how sensitive the
different linkers. Another interesting observation illustrated in transition-state energy is to ligand exchange at this position in
Figure 14 is that increasing the electron-donating abilities of general. The simplest test case is of course to exchange the
the π-base ligands has little influence on the activation barrier, chloride ligand by the less electronegative bromide group. The
as all of the transition states located fall in a narrow range of complete catalytic cycle using Rh(CO)2Br as the active fragment
energies. Thus, changing the electronics of the π-base is not a is shown in Figure 15. The barrier for the oxidative addition
promising avenue for reaction design. associated with the transition state 32-TS, 31.3 kcal/mol, is
A second plausible strategy to be considered for controlling significantly lower than in the chloride analogue where the
the reaction is the removal of the chloride ligand. Experimen- barrier was computed to be 36.7 kcal/mol. However, 31.3 kcal/
tally, the addition of silver salts is often observed to be beneficial mol is still too high to be feasible under mild reaction conditions.
in carbocyclization reactions involving metal chlorides. Presum- In accord with the diene-ene and Rh(CO)2Cl mediated bis-ene
ably, the silver salts remove halides from the metal center, giving [2 + 2 + 1] reaction, the reductive elimination (35-TS) is
access to Rh(CO)x fragments. Removing halogen ligands that possible, with an activation energy of 21.9 kcal/mol. Interest-
are σ-electron withdrawing is a plausible strategy for increasing ingly, the analogous iodide complex shows practically identical
the electron density at the metal center to facilitate oxidative energies and structures (see Supporting Information for details).
addition. We modeled this scenario by using Rh(CO)3+ as the For completeness, we have also simulated the reaction energy
catalytically active fragment to give the reactant complex 30 profile using the Rh(CO)Br fragment as the catalytically active
(Scheme 4). For the oxidative addition of 30, we were able to component, and its energy profile is shown in Figure 16. While
locate the transition state 30-TS with a solution-phase free the oxidative addition via 37-TS has been lowered by 4.6 kcal/
energy of 35.8 kcal/mol, which is very comparable to 36.7 kcal/ mol compared to 22-TS, the decreased electronegativity of
mol obtained for 17-TS. Loss of CO to form 31 is still bromine was not enough to bring the energy requirements of
endergonic, albeit 5.1 kcal/mol lower in energy than 22. The this catalytic cycle into a more useful range. As before, a
oxidative addition traversing 31-TS is still prohibitively high, continuing trend in the use of Rh(CO)Br is that the reductive
but at 39.9 kcal/mol, it is 3.7 kcal/mol lower in energy than elimination (39-TS) is also energetically unfavorable with a
22-TS. Thus, we conclude that the removal of the chloride barrier of 34.6 kcal/mol.
ligand does not influence the overall energetics of the reaction
Conclusions
notably.
The final functionalization strategy that we considered relates In conclusion, we have constructed a complete reaction energy
to the nature of the halogen ligand. Whereas the removal of profile for the rhodium(I)-catalyzed [2 + 2 + 1] Pauson-Khand
J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008 5829
ARTICLES Pitcock Jr. et al.

Figure 16. Reaction energy profile for bis-ene [2 + 2 + 1] using Rh(CO)Br.

carbocyclization, delineating the relationship between the explorations of the Pauson-Khand reactions. Our study dem-
functional groups of the reaction and the catalyst composition. onstrates clearly that the increase of scope and functional group
We found that the addition and removal of CO to control the tolerance is likely to require a more complicated Rh-catalyst
electron density at the metal center is crucial to facilitating the that gives a higher degree of structural design.
two most demanding steps of the catalysis, the oxidative addition
to form the metallacycle and the reductive elimination to liberate Acknowledgment. We thank the NSF (0116050 and CHE-
the final product. Diene-ene substrates make excellent carbocy- 0645381) for financial support. M.H.B. is a Cottrell Scholar of
clization candidates due to the presence of the additional π-base the Research Corporation and an Alfred P. Sloan Fellow of the
fragment in the diene moiety, which allows the metal to be Alfred P. Sloan Foundation. W.H.P. and R.L.L. thank the
coordinatively saturated during the critical oxidative addition Department of Education for financial support through GAANN
step, after releasing a CO ligand. Bis-enes, on the other hand, fellowships and R.L.L. thanks the Merck Research Corporation
lack viable π-base fragments that can undergo rearrangement for a graduate fellowship.
to facilitate CO loss. We have explored a few plausible
functionalization strategies, and while we were thus far unable Supporting Information Available: Cartesian coordinates of
to identify a reactant-catalyst combination that may afford a computed structures, energies, vibrational frequencies, and
successful carbocyclization of simple bis-enes, as seen for the additional discussion. This material is available free of charge
diene-ene reactant, we have demonstrated and classified the via the Internet at http://pubs.acs.org.
conceptually plausible strategies in both qualitative and quan-
titative sense. This work forms the foundation for further JA800856P

5830 J. AM. CHEM. SOC. 9 VOL. 130, NO. 17, 2008