Cundari Th.R. -- Computational Organometallic Chemistry-0824704789
.pdf208 |
Taber et al. |
4.COMPUTATIONAL ANALYSIS OF THE RELATIVE STABILITY OF ZIRCONACYCLES: ZINDO
This observation of the thermal reversibility of diene cyclozirconation led us to reinvestigate the earlier systems (10). Indeed, while cyclozirconation of 1,6- heptadiene 14 at room temperature for 1 h, followed by bromination, gave predominantly the trans-fused dibromide 15 (Scheme 4), as reported, cyclozirconation at low temperature ( 78°C to 0°C) followed by bromination gave predominantly the cis dibromide 16. We hypothesized that diene 14 under kinetic conditions gave the cis zirconacycle, which then equilibrated to the more stable trans zirconacycle on reaching room temperature. Qualitatively, the preference for the trans-fused metallacycle 17 is understandable, since the cis-fused metallacycle 18 is folded, with one of the cyclopentadienyl rings extending toward the concave face of the ring system. The trans-fused metallacycle 17 is extended and so is not destabilized by such a steric interaction.
While this qualitative picture was interesting, we needed a method that would give us a quantitative assessment of the relative stability of equilibrating zirconacycles. A survey of different computational methods (ZINDO (11,12), molecular mechanics, and density functional theory) led to the conclusion that ZINDO (Scheme 5) was the most reliable for calculating the relative stabilities of the diastereomeric zirconacycles. Using ZINDO, the trans-5,5-zirconacycle 17 was calculated to be more stable than the cis-5,5-zirconacycle 18 by 2.5 kcal/ mol.
We had observed that cyclozirconation of diene 4 at room temperature (Scheme 6), even overnight, gave after oxygenation predominantly the cis diol 8. Yet, ZINDO calculations (Scheme 5) indicated that the trans zirconacycle 19 should be more stable than the cis zirconacycle 20. Given this evidence, we re-
Scheme 4
A Computationally Designed Diene |
209 |
Scheme 5
peated the cyclozirconation at higher temperature and found that we did receive, after oxygenation, the expected trans diol 21. We hypothesize that the 6/5 zirconacycle (19/20) is more stable than the 5/5 zirconacycle (17/18), so the activation energy for equilibration is higher for the 6/5 zirconacycle.
5.NATURAL PRODUCT SYNTHESIS: ( )-HALICLONADIAMINE
Carbonylation of the equilibrated zirconacycle 22 gave the cyclopentanone 23 (Scheme 7), a valuable synthon for the construction of natural products. This simple, one-step procedure for the preparation of the trans-fused cyclopentanone
Scheme 6
210 |
Taber et al. |
Scheme 7
22 laid the foundation for our recent enantioselective synthesis of ( )-haliclonad- iamine 23 (13).
6.GLOBAL EQUILIBRATION OF DIASTEREOMERIC ZIRCONACYCLES: ( )-ELEMOL
The equilibration of 17 and 18 or of 19 and 20 (Scheme 5) can be accomplished by exchanging the π face of just one of the two alkenes. The next question to address was whether global equilibration, that is, sequential dissociation and readdition of Zr to each face of each alkene, could be achieved under the conditions of cyclozirconation. We therefore investigated the cyclozirconation of diene 24 (Scheme 8), readily prepared from α-terpineol (14). There are four diastereomeric
Scheme 8
A Computationally Designed Diene |
211 |
Scheme 9
zirconacycles 25–28 that could be formed from 24. On the basis of the relative stabilities (ZINDO) of these diastereomers (Scheme 8), we predicted that global equilibration of the zirconacycles followed by oxygenation should give 29. In the event, cyclozirconation of 24 at 80°C for 5 hours, followed by oxygenation, gave predominantly diol 29, as predicted (Scheme 9).
It was equally striking that cyclozirconation under kinetic conditions (60°C, 3 hours), followed by oxygenation, gave only one of the two possible cis diastereomers of the product diol. Diol 30 derives from the more stable of the two possible cis zirconacycles, 27 and 28. This suggests that ZINDO may possibly also prove useful for predicting the kinetic products from such diene cyclozirconations.
7.COMPUTATIONAL DIENE DESIGN: ( )-ANDROST-4-
ENE-3,16-DIONE
Having established that intramolecular diene cyclozirconation can be carried out under conditions of either kinetic or thermodynamic control, and having shown that semiempirical calculations (ZINDO) can be used to predict the relative stabilities of diastereomeric zirconacycles, we next undertook the computationally based design of a diene such that cyclozirconation would be directed toward a desired diastereomer.
Our initial objective was the construction of the steroid skeleton (e.g., 3, Scheme 10) with control of both relative and absolute configuration. We first considered a B → BCD construction, starting with diene 31 (Scheme 11). Unfortunately, computational analysis (ZINDO) predicted that the undesired cis-fused zirconacycle 33 would be more stable than the desired trans-fused zirconacycle 32. The prospects did not improve with the acetonide 34. Again (Scheme 11), computational analysis (ZINDO) predicted that the cis-fused 36 would be more stable than the desired trans-fused 35.
It was clear that the protecting group on the diol had to introduce steric bulk underneath the ring system of the tricyclic zirconacycle, to destabilize the cis diastereomer. After considering several other alternatives, we settled on the
212 |
Taber et al. |
Scheme 10
Scheme 11
A Computationally Designed Diene |
|
|
|
213 |
|||
TABLE 1 Cyclozirconation/Carbonylation of 1 |
|
|
|
||||
|
|
|
|
|
|
|
|
|
T (°C) |
t (h) |
% yield |
2 |
δ 72.3 |
δ 63.2 |
δ 61.6 |
|
|
|
|
|
|
|
|
1 |
80 |
5 |
28 |
49 |
20 |
10 |
21 |
2 |
80 |
12 |
42 |
58 |
10 |
7 |
25 |
3 |
90 |
2 |
48 |
47 |
20 |
12 |
21 |
4 |
100 |
1 |
19 |
60 |
14 |
13 |
13 |
5 |
80 |
10 |
63 |
52 |
21 |
8 |
19 |
6 |
80 |
14 |
26 |
47 |
19 |
10 |
24 |
7 |
80 |
24 |
10 |
56 |
8 |
5 |
31 |
|
|
|
|
|
|
|
|
menthonide 1. This introduced steric interactions such the desired trans-fused 37 was predicted to be more stable than the cis-fused 38. For each of these three dienes (31, 34, and 1), the other two diastereomeric zirconacycles were predicted to be significantly less stable (from 31, the other trans-fused diastereomer (39) was calculated at 9.6 kcal/mol, while the other cis-fused diastereomer (40) was calculated at 9.9 kcal/mol, compared to 33).
Cyclozirconation conditions were varied (Table 1) to optimize the yield of 2. In each case, the crude diastereomeric mixture of zirconacycles was carbonylated, and the yield and the ratios of the mixture of four product ketones (easily discerned by their oxygenated methines, 13C NMR δ 73.1 (2), δ 72.3, δ 63.2, δ 61.6) were recorded. At temperatures in excess of 80°C (entries 1–4), substantial thermal degradation set in. Returning to 80°C (entries 2, 5–7), it was apparent that while the proportion of 2 was still increasing at 12 h, thermal degradation was again competing, lowering the overall yield. Pure 2 was isolated from the mixture of four product ketones by crystallization, and the structure was established by X-ray crystallography. Ketone 2 was carried over several steps to ( )-androst-4-ene-1,16-dione 3.
To assure ourselves of the role of the menthone ketal in the cyclozirconation of 1, we also effected cyclozirconation (80°C, 10 h) of the diol 40 (Scheme 12),
Scheme 12
214 |
Taber et al. |
using an extra two equivalents of n-BuLi to deprotonate the alcohols. Cyclozirconation and carbonylation proceeded smoothly, but as would be expected from the calculations (Scheme 11) the cis-fused product 41 was dominant (65%), and the trans-fused ketone 42 was only 13% of the mixture of product ketones.
8. DIRECTIONS FOR THE FUTURE
These exciting results established the validity of this computational approach. This is, however, just the beginning. Although we could push the proportion of 2 (Scheme 10) to 5:1 and even higher, with longer times and/or higher temperatures for the cyclometallation step, the yields of product dropped off, due to thermal degradation of the metallacycle. In addition to extending the cyclozirconation to more challenging dienes, we are therefore also exploring other metal and ligand combinations to effect intramolecular diene cyclometallation. Our objective is to establish a metal/ligand combination such that full equilibration of the cyclometallation products can be achieved efficiently.
Rothwell (15) recently reported that reduction of (ArO)2TiCl2 43 (ArOH2,6-diphenyl phenol) (Scheme 13) in the presence of 1,7-octadiene 4, with warming only to room temperature, led to the trans-fused titanacycle 45. We prepared 43 from the commercially available 2,6-diphenylphenol and repeated this cyclotitanation, oxygenating the intermediate titanacycles to give diols 8 and
21. It is apparent that the 6/5 titanacycle is equilibrating much more rapidly (4 h, rt) than the 6/5 bis-Cp zirconacycle (3 h, 70°). If ZINDO calculations are valid
with the titanacycles (we have not yet established this), it is also apparent that
Scheme 13
A Computationally Designed Diene |
215 |
Scheme 14
the much more sterically demanding (ArO)2Ti favors the trans ring fusion by a somewhat larger margin than does the Cp2Zr (for Cp2Zr, trans is favored by 2.8 kcal/mol).
We prepared and purified the airand moisture-sensitive complex 43 to carry out these studies. We have also found that it is possible to generate this complex in situ, with the starting ArOH (two equivalents) and two equivalents of BuLi, followed, at rt, by TiCl4, and then at 78° by the diene and two more equivalents of BuLi. The cyclization results are the same as with the preprepared complex 43. We will use this latter approach to quickly screen a variety of other alcohols and diols, including enantiomerically pure diols such as BINOL (1,1′- bi-2-naphthol) and Taddol (α,α,α′,α′-tetraaryl-1,3-dioxolane-4,5-dimethanol), in this cyclization.
Our primary interest with 43 is to develop an alternative procedure for diene cyclometallation such that the intermediate metallacycles will equilibrate at lower temperature and more efficiently than is observed with Cp2ZrCl2. It is also striking that with more substituted dienes, the trans metallacycles derived from 43 are also favored over the cis by a more substantial margin than with Cp2ZrCl2. For diene 1 (Scheme 14), for instance, the trans titanacycle 46 is calculated to be more stable than the cis titanacycle 47 by 3.3 kcal/mol.
REFERENCES
1.DF Taber, W Zhang, CL Campbell, AL Rheingold, CD Incarvito. J Am Chem Soc 122:4813–4814, 2000.
2.(a) For an overview of carbonylative diene and enyne cyclometallation, see LS Hegedus, Transition Metals in the Synthesis of Complex Organic Molecules. 2nd ed. Sausalito, CA: University Science Books, 1999. For other recent references to carbonylative cyclometallation, see: (b) Z Zhao, Y Ding, G Zhao. J Org Chem 63: 9285–9291, 1998. (c) E-I Negishi, J-L Montchamp, L Anastasia, A Elizarov, D Choueiry. Tetrahedron Lett 39:2503–2506, 1998. (d) Y-T Shiu, RJ Madhushaw, W-T Li, Y-C Lin, G-H Lee, S-M Peng, F-L Liao, S-L Wang, R-S Liu. J Am Chem
216 |
Taber et al. |
Soc 121:4066–4077, 1999. (e) M Murakami, K Itami, Y Ito. J Am Chem Soc 121: 4130–4135, 1999. (f) FA Hicks, NM Kablaoui, SA Buchwald. J Am Chem Soc 121: 5881–5898, 1999.
3.For recent examples of steroid total synthesis, see: (a) M Kurosu, LR Marcin, TJ Grinsteiner, Y Kishi. J Am Chem Soc 120:6627, 1998. (b) PA Grieco, SA May, MD Kaufman. Tetrahedron Lett 39:7047–7050, 1998. (c) PA Zoretic, H Fang, A Ribeiro. J Org Chem 63:7213–7217, 1998. (d) C Heinemann, M Demuth. J Am Chem Soc 121:4894–4895, 1999, and references cited therein.
4.WA Nugent, DF Taber. J Am Chem Soc 111:6435–6437, 1989.
5.SL Buchwald, RB Nielsen. Chem Rev 88:1047–1058, 1988.
6.The literature contained several reports of cyclization of 1,7-octadiene by early transition metal complexes to isomeric mixtures of metallaindanes: (a) JX McDermott, ME Wilson, GM Whitesides. J Am Chem Soc 98:6529–6536, 1976. (b) RH Grubbs, A Miyashita. J Chem Soc, Chem Commun 864–865, 1977. (c) SJ McLain, CD Wood, RR Schrock. J Am Chem Soc 101:4558–4570, 1979. (d) KI Gell, J Schwartz. J Chem Soc, Chem Commun 244–246, 1979.
7.E-I Negishi, FE Cederbaum, T Takahashi. Tetrahedron Lett 27:2829–32, 1986.
8.DF Taber, JP Louey, JA Lim. Tetrahedron Lett 34:2243–46, 1993.
9.Negishi independently observed the reversibility of intermolecular diene cyclozirconation. T Takahashi, T Fujimori, S Takashi, M Saburi, Y Uchida, CJ Rousset, E-I Negishi. J Chem Soc, Chem Commun 182–183, 1990.
10.DF Taber, JP Louey, Y Wang, WA Nugent, DA Dixon, RL Harlow. J Am Chem Soc 116:9457–9463, 1994.
11.Both ZINDO and molecular mechanics were used as implemented on the Tektronix CAChe workstation. For leading references to ZINDO, a semiempirical program that has been parameterized for the first two rows of transition metals, see: (a) MC Zerner, GW Loew, RF Kirchner, UT Mueller-Westerhoff. J Am Chem Soc 102: 589–599, 1980. (b) WP Anderson, TR Cundari, RS Drago, MC Zerner. Inorg Chem 29:1–5, 1990.
12.Although ZINDO was originally parameterized to give good spectroscopic results, it had also been used in studies of the energetics and structures of transition metal– based catalytic systems: (a) GL Estiu, MC Zerner. J Phys Chem 97:13720–13729, 1993. (b) GL Estiu, MC Zerner. Int J Quantum Chem 26:587, 1992.
13.DF Taber, Y Wang. J Am Chem Soc 119:22–26, 1997.
14.DF Taber, Y Wang. Tetrahedron Lett 36:6639–42, 1995.
15.MG Thorn, JE Hill, SA Waratuke, ES Johnson, PE Fanwick, IP Rothwell. J Am Chem Soc 119:8630–8641, 1997.
9
Rhodium-Mediated Intramolecular C–H
Insertion: Probing the Geometry of the
Transition State
Douglass F. Taber
University of Delaware, Newark, Delaware
Pascual Lahuerta and Salah-eddine Stiriba
University of Valencia, Valencia, Spain
James P. Louey
Sacred Heart University, Fairfield, Connecticut
Scott C. Malcolm
Harvard University, Cambridge, Massachusetts
Robert P. Meagley
Intel Corporation, Hillsboro, Oregon
Kimberly K. You
BASF Corporation, Wyandotte, Michigan
1. INTRODUCTION
The power of Rh-mediated intramolecular C–H insertion can be seen in the cyclization of the α-diazo ester 1 (Scheme 1). Although four diastereomers could have been formed from this cyclization, only 2, the key intermediate for the synthesis of the dendrobatid alkaloid 251F 3, was in fact observed. This outcome, as explained in detail shortly, had first been predicted computationally. This chapter summarizes our computational approach toward understanding the transition state (‘‘point of commitment’’) for these Rh-mediated cyclizations. As we discuss at the end of this chapter, there is yet much left to be learned.
217