Chemical Kinetics and J. Org

pubs.acs.org/joc

Mechanism and Regioselectivity of the Osmium-Catalyzed
Aminohydroxylation of Olefins
Dominik Munz and Thomas Strassner*
Physikalische Organische Chemie, Technische Universit€t Dresden, Mommsenstrasse 13, 01062 Dresden, a Germany [e-mail] Received November 12, 2009

The mechanism and regioselectivity of the osmium-catalyzed aminohydroxylation of olefins was investigated in detail by density functional theory (B3LYP/6-31G(d)) calculations in the gas phase and with the CPCM-solvent model. A systematic variation of the catalyst system (OsO4 and various nitrogen sources) and the substrate’s electronic situation was conducted. Activation barriers could be correlated to Hammett values and linear Gibbs free energy relations could be determined. Experimental results, which indicated an electronic influence on the regioselectivity, could be confirmed and appear to be predictable. The reaction follows a [3þ2] mechanism. We additionally report results on the experimentally observed competing dihydroxylation reaction and the ligand-induced reaction rate acceleration.

Introduction
Sharpless et al. reported in the 1970s an aza-analogon of the osmium-catalyzed cis-vicinal dihydroxylation (DH)1 of alkenes, the aminohydroxylation (AH).2,3 In 1996 it was rendered asymmetric4 and extended to a large variety of substrates in the following. The AH is synthetically important as it provides straightforward access to the aminoalcohol fragment present in a broad variety of natural products5,6
*To whom correspondence should be addressed. Tel.: þ49 351 46338571.
Fax: þ49 351 46339679.
(1) Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem. Rev.
1994, 94, 2483–2547.
(2) Sharpless, K. B.; Patrick, D. W.; Truesdale, L. K.; Biller, S. A. J. Am.
Chem. Soc. 1975, 97, 2305–2307.
(3) Sharpless, K. B.; Chong, A. O.; Oshima, K. J. Org. Chem. 1976, 41,
177–179.
(4) Li, G.; Chang, H.-T.; Sharpless, K. B. Angew. Chem., Int. Ed. 1996, 35,
451–454.
(5) Bergmeier, S. C. Tetrahedron 2000, 56, 2561–2576.
(6) Li, G.; Sharpless, K. B. Acta Chem. Scand. 1996, 50, 649–651.
(7) Bodkin, J. A.; McLeod, M. D. J. Chem. Soc., Perkin Trans. 1 2002,
2733–2746.

DOI: 10.1021/jo902421n r 2010 American Chemical Society

Published on Web 02/03/2010

and has been extensively reviewed.7-14 The reaction is usually carried out in alcohol/water solvent mixtures as shown in Scheme 1. The chiral ligand is often derived from the Cinchona alkaloids and the catalytically active species is formed in situ from an osmium(VI) salt like K2OsO2(OH)4 and a stoichiometric nitrogen source like Chloramine-M.
Frequently used nitrogen sources are sulfonamide,8 carbamate,1 amide,1 or tert-butyl2,15 compounds. Generally different regio- and stereoisomers can be formed.
By the right choice of ligands it often seems to be possible to tune the enantioselectivity, but the setting of a particular
(8) Kolb, H. C.; Sharpless, K. B. Transition Metals for Organic Synthesis,
2nd ed.; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, Germany,
2004; Vol. 2, pp 309-326.
(9) Muniz, K. Chem. Soc. Rev. 2004, 33, 166–174.
(10) O’Brien, P. Angew. Chem., Int. Ed. 1999, 38, 326–329.
(11) Bolm, C.; Hildebrand, J. P.; Muniz, K. Catalytic Asymmetric
Synthesis, 2nd ed.; Wiley-VCH: New York, 2000; pp 399-428.
(12) Bayer, A. Compr. Asymmetric Catal., Suppl. 2004, 2, 43–71.
(13) Nilov, D.; Reiser, O. Adv. Synth. Catal. 2002, 344, 1169–1173.
(14) Lohray, B. B.; Bhushan, V.; Reddy, G. J.; Reddy, A. S. Indian J.
Chem., Sect. B: Org. Chem. Incl. Med. Chem. 2002, 41B, 161–168.
(15) Rubinstein, H.; Svendsen, J. S. Acta Chem. Scand. 1994, 48, 439–444.

J. Org. Chem. 2010, 75, 1491–1497

1491

JOC Article
SCHEME 1. Aminohydroxylation Reaction with the Oxidans
Chloramine-M (H3CSO2NCl Na)

regioselectivity remains challenging,16 as it is controlled by multiple factors. Steric and electronic contributions of the ligand,17,18 substrate,8,18-21 and hydrophobic effects due to the solvent19 have been experimentally observed and investigated. Janda et al. have proposed a substrate-based methodology18 to explain the factors controlling the regioselectivity.
These are often mutually dependent on each other and experimentally hard to determine, but can be calculated by quantum-chemical calculations.
This work for the first time systematically addresses the electronic influence of substituents R on the O3OsNdR unit as well as of the reacting olefin on the regiochemistry of the
AH. As the ligands in general are too big to be calculated by high-level DFT calculations we used model systems to evaluate the different factors which might influence the regioselectivity of the reaction.
Computational Details
All calculations were performed with Gaussian-03,22 using the density functional/Hartree-Fock hybrid model Becke3LYP
23-26
and the split valence double-ζ (DZ) basis set 6-31G(d)27 for
C, H, O, and S. The Hay-Wadt28 effective core potential (ECP) was used for osmium, as it has been successfully employed in former studies on the respective dihydroxylation29 and diamination30 reaction. No symmetry or internal coordinate constraints were applied during optimizations. All reported ground state structures were verified as being true minima by the absence of negative eigenvalues in the vibrational analysis.
Transition state (ts) structures were located with use of the
Berny algorithm31 and it was verified that the Hessian matrix has only one imaginary eigenvalue. The identities of all transition states were confirmed by animating the negative eigenvector coordinate with MOLDEN32 and intrinsic reaction
(16) Donohoe, T., J.; Chughtai, M. J.; Klauber, D. J.; Griffin, D.;
Campbell, A. D. J. Am. Chem. Soc. 2006, 128, 2514–2515.
(17) Tao, B.; Schlingloff, G.; Sharpless, K. B. Tetrahedron Lett. 1998, 39,
2507–2510.
(18) Han, H.; Cho, C.-W.; Janda, K. D. Chem.;Eur. J. 1999, 5, 1565–
1569.
(19) Reddy, K. L.; Sharpless, K. B. J. Am. Chem. Soc. 1998, 120, 1207–
1217.
(20) Harding, M.; Bodkin, J. A.; Issa, F.; Hutton, C. A.; Willis, A. C.;
McLeod, M. D. Tetrahedron 2009, 65, 831–843.
(21) Morgan, A. J.; Masse, C. E.; Panek, J. S. Org. Lett. 1999, 1, 1949–
1952.
(22) Frisch, M. J. et al. Gaussian03, Rev. E 01; Gaussian Inc.: Wallingford,
CT, 2004.
(23) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200–1211.
(24) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B: Condens. Matter Mater.
Phys. 1988, 37, 785–789.
(25) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(26) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J.
J. Phys. Chem. 1994, 98, 11623–11627.
(27) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56,
2257–2261.
(28) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299–310.
(29) DelMonte, A. J.; Haller, J.; Houk, K. N.; Sharpless, K. B.; Singleton,
D. A.; Strassner, T.; Thomas, A. A. J. Am. Chem. Soc. 1997, 119, 9907–9908.
(30) Deubel, D. V.; Muniz, K. Chem.;Eur. J. 2004, 10, 2475–2486.
(31) Schlegel, H. B. J. Comput. Chem. 1982, 3, 214–218.
(32) Schaftenaar, G.; Noordik, J. H. J. Comput.-Aided Mol. Des. 2000,
14, 123–134.

1492

J. Org. Chem. Vol. 75, No. 5, 2010

Munz and Strassner coordinate (IRC) calculations. Approximate Gibbs free energies
(ΔG) and enthalpies (ΔH) were obtained via thermochemical analysis of frequency calculations. This takes into account zeropoint effects, thermal enthalpy, and entropy corrections. All energies reported are Gibbs free energies or enthalpies at 298 K, using unscaled frequencies. Atomic partial charges were predicted with the NPA model.33 Solvation corrections were applied with use of the CPCM-model34-36 as implemented in
Gaussian-03 and we found that the calculated selectivities exhibited a dependence on solvation parameters. After a thorough screening of different solvation parameters in optimizations, single points of the experimental permittivity ε of a 1:1 tert-butanol:water mixture (ε = 37.8 F/m ; εinf = 2.69 F/m) were performed on optimized structures.37 This solvent mixture is most commonly used in AH reactions.7 A solvent’s mean
˚
˚ molecular radius of 2.16 A, a numerical density of 0.012 A-3 derived by the experimental density,37 and UAKS cavities38,39 were utilized. This approach of determining a solvent’s radius was recently used by Goddard et al.40 and does in our opinion reproduce the experimental conditions best. It is described in detail in the Supporting Information. ΔGCPCM corresponds to the addition of rotational and zero-point corrected vibrational energies in the gas phase at 298 K to the electronic energy in solution (ΔECPCM) as we could not optimize all structures in solvent calculations. The neglect of vibrational and entropic corrections to the gas phase energy seems to be justified as test calculations on selected transition states revealed an error of less than 1 kcal mol-1.

Results and Discussion
To calculate the electronic effects of the nitrogen source and the substrate on the regioselectivity of the AH, it is necessary to first discuss the underlying mechanism ([3þ2] or
[2þ2]) of the AH (periselectivity). We also need to calculate competing reaction pathways, which raises the question of amino- versus dihydroxylation (chemoselectivity). The ligand introduces additional questions: Does the reaction proceed by syn or anti addition? Can we observe a ligandinduced acceleration? Contrary to the DH we also have to address the question of the regioselectivity (R/β) of the AH if the substrate carries different substituents. As electrondonating or -withdrawing substituents at the nitrogen source have been experimentally shown to have an influence on the regiochemistry,8 also different substituents at the nitrogen source have been investigated.
First we would like to discuss the concerted [3þ2] cycloaddition of the olefin to the ligated osmium compound 2 in analogy to the Criegee mechanism41,42 of the DH (Scheme 2).18
A competing [3þ2] mechanism without ligand coordination via intermediates 3 has to be considered, although it will
(33) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899–
926.
(34) Barone, V.; Cossi, M. J. Phys. Chem. A 1998, 102, 1995–2001.
(35) Miertus, S.; Scrocco, E.; Tomasi, J. Chem. Phys. 1981, 55, 117–129.
(36) Klamt, A.; Schueuermann, G. J. Chem. Soc., Perkin Trans. 2 1993,
799–805.
(37) Tabellout, M.; Lanceleur, P.; Emery, J. R.; Hayward, D.; Pethrick,
R. A. J. Chem. Soc., Faraday Trans. 1990, 86, 1493–1501.
(38) Barone, V.; Cossi, M.; Tomasi, J. J. Chem. Phys. 1997, 107, 3210–
3221.
(39) McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639–5648.
(40) Xu, X.; Kua, J.; Periana, R. A.; Goddard, W. A. III Organometallics
2003, 22, 2057–2068.
(41) Criegee, R. Justus Liebigs Ann. Chem. 1936, 522, 75–96.
(42) Torrent, M.; Deng, L.; Duran, M.; Sola, M.; Ziegler, T. Organometallics 1997, 16, 13–19.
(43) Only the most stable regioisomeres are shown in Schemes 2 and 3.

Munz and Strassner
SCHEME 2. [3þ2] Mechanism of the Aminohydroxylation
(AH) and Dihydroxylation (DH)43

almost certainly lead to the experimental observation of a lower overall enantioselectivity, as no chiral ligand is involved in the rate-determining reaction step (Scheme 2).
In every case two different regioisomers may result: the
R-amino 5r and the β-amino products 5β. Those are being hydrolyzed consecutively in situ to yield the aminoalcohol.
Furthermore competing dihydroxylation mechanisms leading to 6 had to be considered.
We chose the substituents R1 on the imido nitrogen atom and R2 on the olefin due to their assigned Hammett substituent parameters.44,45 R1 defines mesyl (mes_), tosyl (tos_), hydrogen (h_), cyano (cn_), or methyl (me_) substituents, R2 different substrates, which are ethylene (a), acrolein (b), propene (c), butadiene (d), and acrylonitrile (e).
To model the electronic effect of the coordination of the chiral nitrogen ligand L, ammonia has been chosen. This simplification (with precedence in the literature)29,46-48 was necessary considering the number of calculated structures and the size of the large cinchona-derived ligands. Test calculations with the ligand NMe3 showed only minor energy differences of less than 0.5 kcal mol-1 compared to those of ammonia.
According to the original proposal the reaction could also follow a formal [2þ2] cycloaddition of the alkene to give the osmaazetidin intermediate 7a, which might rearrange under coordination of the nitrogen ligand L to the osmium azaglycolate 5a (Scheme 3)49 and the osmaoxetane intermediate 8a, which can rearrange either to 5a or the DH product 6a. Moreover a [2þ2] cycloaddition of ethylene to compound 2 might yield 9a (which could rearrange to 5a) as well as 10a (which could rearrange to afford 5a or 6a).
(44) Hammett, L. P. J. Am. Chem. Soc. 1937, 59, 96–103.
(45) Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165–195.
(46) Ess, D. H. J. Org. Chem. 2009, 74, 1498–1508.
(47) Ujaque, G.; Maseras, F.; Lledos, A. Eur. J. Org. Chem. 2003, 833–
839.
(48) Strassner, T.; Drees, M. THEOCHEM 2004, 671, 197–204.
(49) Sharpless, K. B.; Teranishi, A. Y.; Backvall, J. E. J. Am. Chem. Soc.
1977, 99, 3120–3128.
(50) Dapprich, S.; Ujaque, G.; Maseras, F.; Lledos, A.; Musaev, D. G.;
Morokuma, K. J. Am. Chem. Soc. 1996, 118, 11660–11661.
(51) Pidun, U.; Boehme, C.; Frenking, G. Angew. Chem., Int. Ed. 1997,
35, 2817–2820.
(52) Deubel, D. V.; Frenking, G. J. Am. Chem. Soc. 1999, 121, 2021–2031.
(53) Deubel, D. V.; Schlecht, S.; Frenking, G. J. Am. Chem. Soc. 2001,
123, 10085–10094.

JOC Article
SCHEME 3. [2þ2] Mechanism of the Aminohydroxylation
(AH) and Dihydroxylation (DH)

For the DH isotope effect studies and quantum-chemical calculations revealed a significant preference for the [3þ2] mechanism,29,42,50-54 which is also the favored mechanism in the case of the Os(VIII)-catalyzed diamination.30 However, investigations on complexes of the general type LMO3
(where L could also be oxygen) indicated that the activation barriers strongly depend on the ligand L and the transition metal M, which might even lead to a situation where a [2þ2] mechanism is favored.53,55,56 It was also shown that the activation barriers for the [3þ2] cycloaddition of ethylene to LMO3 type complexes can be predicted according to the
Marcus theory.57 We therefore did not expect the reaction to proceed via a [2þ2] mechanism and checked only for one case, the reaction of the nitrile-substituted imidotrioxoosmium complex with ethylene.
Peri- and Chemoselectivity. We calculated the aminohydroxylation transition states for the [3þ2] cycloadditions of ethylene to the cyano-substituted imidotrioxoosmium complexes cn_1 (leading to cn_3a) and cn_2 (with L = NH3, leading to cn_5a) (Scheme 2) and compared them to the [2þ2] reaction pathway. Additions across the OsdN bond lead to the osmaazetidine intermediates cn_7a and cn_9a, whereas addition across the OsdO bond affords the osmaoxetanes cn_8a and cn_10a, which could rearrange to cn_5a and cn_6a
(Scheme 3). Table 1 gives enthalpies and Gibbs free energies of transition states and products with the ammonia ligand in the anti-position relative to the imido group.
The [2þ2] cycloadditions (entries 1-4) are kinetically and thermodynamically disfavored by more than 25 kcal mol-1 compared to the [3þ2] mechanism (entries 5 and 6).
Although we only checked one example, we can expect that also for the substituted olefins the [2þ2] reaction pathway is not favored.
For the cycloaddition step of the [3þ2] addition a ligandinduced reaction-rate acceleration of 2.7 kcal mol-1 was calculated. The coordination of the model ligand NH3 leads
(in agreement with Hammond’s postulate) to shorter O-C
˚
˚
(2.277 vs 2.301 A) and N-C (2.397 vs 2.455 A) distances in the transition state (Chart 1).
To study the chemoselectivity (DH versus AH) we investigated the effect of the five electronically different substituents R1 (Scheme 2), but restricted the variety of alkenes to
(54) Deubel, D. V. Angew. Chem., Int. Ed. 2003, 42, 1974–1977.
(55) Chen, X.; Zhang, X.; Chen, P. Angew. Chem., Int. Ed. 2003, 42, 3798–
3801.
(56) Pietsch, M. A.; Russo, T. V.; Murphy, R. B.; Martin, R. L.; Rappe,
A. K. Organometallics 1998, 17, 2716–2719.
(57) Gisdakis, P.; Roesch, N. J. Am. Chem. Soc. 2001, 123, 697–701.

J. Org. Chem. Vol. 75, No. 5, 2010

1493

JOC Article

Munz and Strassner

TABLE 1. Periselectivity: Activation (ΔHq) and Reaction Enthalpies (ΔH) as well as Gibbs Free Energies (ΔGq and ΔG) of cn_1/cn_2 with Ethylene in
Solvent (in kcal mol-1 at 298.15 K)a entry reaction
ΔH qCPCM (ΔHq)
ΔGqCPCM (ΔGq)
ΔHCPCM (ΔH)
ΔGCPCM (ΔG)
1
2
3
4
5
6 a cn_1 f cn_7a cn_1 f cn_8a cn_2 f cn_9a cn_2 f cn_10a cn_1 f cn_3a cn_2 f cn_5a

37.7 (42.4)
32.3 (32.9)
29.2 (37.5)
27.8 (40.2)
2.0 (0.8)
-0.2 (-0.1)

52.2 (56.8)
45,9 (46,6)
44.7 (53.0)
41.9 (54.3)
15.2 (14.0)
12.5 (12.7)

28.8 (35.5)
-5.0 (-3.8)
28.3 (33.5)
1.2 (4.6)
-53.7 (-49.1)
-62.1 (-55.6)

42.9 (49.6)
9,2 (10.4)
43.8 (51.1)
15.6 (18.9)
-38.7 (-34.1)
-46.6 (-40.2)

Gas phase results are given in parentheses.

CHART 1. [3þ2] Transition States for the Reaction of
OsO3NCN and OsO3(NH3)NCN with Ethylenea

FIGURE 1. Gibbs free reaction energies ΔG of the anti-addition 1 þ NH3 f 2 (in kcal mol-1). Gas phase results are given in parentheses. SCHEME 5.
Selectivity
a

Ammonia Ligand Addition and Effects on Overall

˚
Bond lengths are given in A, angles in deg.

SCHEME 4.

Anti- and Syn-Addition of the Nitrogen Ligand

ethylene. In all DH cases the syn-[3þ2] transition states (e.g., leading to cn_6, Scheme 4) turned out to be favored, whereas for the AH transition states an anti-alignment of the NH3 group with respect to the imido group is lower in energy (e.g., leading to cn_5a, Scheme 4). Additional data can be found in the Supporting Information.
Gibbs free reaction energies for the anti-addition (1 þ NH3 f 2,
Scheme 2) of ammonia to the five different trioxoimidoosmium compounds 1 is summarized in Figure 1. It is exergonic for strongly electron withdrawing substituents (OsO3NCN: -3.5 kcal mol-1) and endergonic in the case of methyl or hydrogen substituents (OsO3NH: 3.3 kcal mol-1; OsO3NMe: 4.1 kcal mol-1). Syn-addition is less favorable; the results are given in the Supporting Information.
As the reaction involves the coordination of the amine ligand and the formation of the product as shown in
Scheme 5, both steps have to be considered in order to predict the overall reaction rate acceleration effects for the different substituents R1 (Table 2). According to the
Curtin-Hammett principle only ΔΔGq is decisive for the overall selectivity of competing reactions.
(58) Transition states mes_5a_ts like mes_5cβ_ts in Figures 2, 3, 5 were obtained due to convergence of forces.

1494

J. Org. Chem. Vol. 75, No. 5, 2010

Table 2 gives the calculated activation barriers leading either to the DH or AH products. In all cases the DH was calculated to be higher in activation energy compared to the
AH. Except in the case of the hydrogen substituent (entry 2) a ligand-induced reaction rate accelerating effect of the AH was predicted. The major difference between the DH and AH transition states is the preferred orientation of the nitrogen ligand which in the case of the DH prefers to be syn while for the AH transition states the anti-position of the ligand is preferred. Regioselectivity. To analyze the effect of the different substituents R1 (Scheme 2) as well as the electronic effects of the alkene substituents R2 (Scheme 2) we calculated all
[3þ2] transition states leading to the 5r- and 5β-regioisomers. All calculated structures, their energies, and coordinates are given in the Supporting Information. The results for the regioselectivity of the [3þ2] addition are summarized in Figure 2. The calculated β-selectivity is given by the difference ΔGqR - ΔGqβ.
The β-addition is predicted to be preferred not only kinetically, but also thermodynamically. The Hammett plots for the nitrile (cn_) and sulfonamide (mes/tos_) substituents are given in Figure 3. The substituents R1
(Scheme 2) have been chosen by their Hammett parameters to reflect the experimentally used substituents. From the

JOC Article

Munz and Strassner
TABLE 2.

a

Activation Energies ΔG6¼ for the [3þ2] Additions of OsO3N-R1 to Ethylene According to Scheme 5 (in kcal mol-1)a 58

Gas phase results are given in parentheses.

FIGURE 2. [3þ2] β-addition is preferred over the R-addition (in kcal mol-1). Gas phase results are given in parentheses.58

FIGURE 3. Hammett plot of the β-addition of the nitrile (cn_), tosyl (tos_), and mesyl (mes_) systems.

calculated results it can be predicted that substituents with higher Hammett constants will lead to higher activation energies. The Hammett plots (ΔΔG is given relative to ethylene) show a large deviation for butadiene, in case of R1 = CN even the inverse selectivity toward the R-product is predicted
(Figure 3). We therefore analyzed the NPA charges of all transition states (given in the Supporting Information) as well as the Kohn-Sham HOMO-LUMO gaps. The amount of net charge transfer from the alkene to the transition metal compound is higher in the case of R-addition than β-addition, whereas the β-addition transition states can be characterized by a larger dipolarity of the double bond of the reacting olefin. The donation of electron density from the

FIGURE 4. Correlation between the activation barriers of the
[3þ2] transition states tos_5ar-tos_5er and tos_5aβ-tos_5eβ to tabulated p-Hammett constants of the alkenes.

alkene to the metal seems to be an important contribution.
A similar observation by charge decomposition analysis had been made in the case of the addition to OsO2(NH2)2 or
OsO4.30,54
The calculated β-selectivity is in agreement with the experimentally observed nitrogen addition to the less substituted carbon atom of the olefin. In the symmetric
J. Org. Chem. Vol. 75, No. 5, 2010

1495

JOC Article
CHART 2.

TABLE 3. transition state cn_5a_ts cn_5br_ts cn_5bβ_ts cn_5cr_ts cn_5cβ_ts cn_5dr_ts cn_5dβ_ts cn_5er_ts cn_5eβ_ts Munz and Strassner

Transition States of the CN Substituted System

Calculated Activation Energies and Selected Bond Lengths, Angles, and Dihedrals of the Transition States Shown in Chart 2
O-Os-N angle
O-C bond length N-C bond length C-C bond length
Os-N-C-C dihedral angle
ΔGq (kcal
˚
˚
˚
mol-1)
(deg)
(A)
(A)
(A)
(deg)
95.8
95.2
94.8
94.9
95.2
95.3
94.5
94.4
94.3

2.301
2.128
2.263
2.253
2.329
2.197
2.339
2.075
2.252

2.456
2.423
2.309
2.492
2.433
2.581
2.345
2.401
2.251

1.357
1.372
1.370
1.365
1.363
1.374
1.373
1.378
1.375

0.00
2.3
1.3
11.6
-11.1
11.8
-14.6
11.9
-5.6

12.5
15.3
14.9
12.8
12,2
12.3
12.9
16.5
16.4

FIGURE 5. Gibbs free energies of activation for [3þ2] β-transition states.

aminohydroxylation3,59 as well as in the asymmetric version, the nitrogen atom usually ends up away from the most electron withdrawing group of the olefin.8,18,60-62
This is also in agreement with the experimental observation, that conjugated systems like cyclohexadiene or E-1phenylpropene often show a preference for the R-addition.3
In the literature this is sometimes also explained by ligand/ solvent interactions.19,62 But as we can correlate activation barriers to tabulated Hammett substituent values of the olefins (e.g., for the tosyl substituent in Figure 4) we propose a relation between electron donating substituents and decreasing preference for β-addition in general. This is nicely
(59) Herranz, E.; Sharpless, K. B. J. Org. Chem. 1978, 43, 2544–2548.
(60) Li, G.; Angert, H. H.; Sharpless, K. B. Angew. Chem., Int. Ed. 1997,
35, 2813–2817.
(61) Rudolph, J.; Sennhenn, P. C.; Vlaar, C. P.; Sharpless, K. B. Angew.
Chem., Int. Ed. 1997, 35, 2810–2813.
(62) Bruncko, M.; Schlingloff, G.; Sharpless, K. B. Angew. Chem., Int.
Ed. 1997, 36, 1483–1486.

1496

J. Org. Chem. Vol. 75, No. 5, 2010

demonstrated by the decreasing distance between the linear regression lines for R- and β-addition for small Hammett σ-values (Figure 4).
All R-transition states show more asynchronicity (Chart 2,
Table 3) resulting in longer distances to the imido nitrogen atom. We observed the same trend for all systems, but for clarity of the presentation decided to only show the transition states for the cyano system.
Figure 5 compares the Gibbs free energies for the transition states leading to the β-products. The difference of the activation energies of all 20 transition states is relatively small (11.2-16.9 kcal mol-1). The lowest activation energy was calculated for an electron withdrawing substituent
(R1 = mes) and butadiene, an electron rich olefine, the highest for an electron donating substituent (R1 = me) and butadiene. In general the combination of electron donating substituent and electron rich substrate is unfavorable. But electron withdrawing groups on the alkene as well as on the

Munz and Strassner

metal also lead to a higher barrier (e.g., R1 = cn; acrylonitrile).
The substituent R1 plays an important role and systems with electron withdrawing groups are predicted to lead to better results than those with electron donating groups (propene,
R1 = cn (12.2 kcal mol-1); R1 = tos (13.4 kcal mol-1%2