Nonracemic and 18 O-labeled 4-tosyloxyprotoadamantane-4-of the tosylate oxygen atoms), elimination (6%), and solvolysis (34%) competitively. 2-Hydroxyadamantane-1-carbonitriles 10c, 11c were prepared. In 60 % dioxane, the exo-OTs isomer 10c afforded ca. 97% of 2-tosyl-carbonitrile ( 15) and 4-hydroxyprotoadamantane-3-carbonitrile (exo/endo = 18/22 เด 2.8) were formed with ee ี 97%. oxyadamantane-1-carbonitrile ( 14) with ee = 98% and with 23% equilibration of the tosylate oxygen atoms. The data According to these results, the chirality (unsymmetrical bridging) of the parent 2-adamantyl cation is not detectably point to a tight ion pair which undergoes nearly complete and stereospecific recombination. The endo-OTs isomer 11c
affected by 1-CN whereas the endo selectivity of the parent 4-protoadamantyl cation (symmetrical bridging) is elimi-underwent internal return to give endo-4-tosyloxyprotoadamantane-3-carbonitrile (20, 60% yield, 51% equilibration nated by 3-CN.
Scheme 1
The influence of electron-withdrawing substituents on the generation of carbocations has been studied extensively [1] [2] . Much less attention has been directed to 1,2-carbon shifts of destabilized carbocations. A fundamental question regards the degree to which ฯ delocalization is affected by destabilizing substituents. In previous approaches to the problem, cyano and/or perfluoroalkyl groups were attached to bicyclo[2.1.1]hex-2-yl [3] , 2-norbornyl [4] , bicyclo-[3.2.0]hept-2-yl [5] , and 2-norpinyl substrates [6] . The present work extends our studies to the 4-protoadamantyl ว 2adamantyl rearrangement.
The 4-protoadamantyl ว 2-adamantyl rearrangement, constituting an important step in the synthesis of adamantoid hydrocarbons [7] , has been investigated experimentally [8] [9] as well as computationally [10] . The stereochemistry of the rearrangement was elucidated by means of deuterium-labeled, nonracemic substrates (Scheme 1) [11] . The exo substrate 1 was found to give 2-adamantyl products 3 with ee ี 97%. The endo substrate 6 afforded 4-protoadamantyl products rather unselectively (exo-5, 8.1% ; endo-5, 3.9% ; 8, 3.6%), along with 3 (76%, ee ี 97%). The data indicate that the chiral, bridged 2-adamantyl cation 2 does not significantly interconvert with an achiral, open species whereas both versions of the 4-protoadamantyl cation, 4 standard procedure was applied to [ 18 O]-9, substantial oxyand 7, intervene. We have now replaced the deuterium labels gen exchange was observed. However, the reaction of [ 18 O]of these intermediates with cyano groups. 9 with trimethylsilyl cyanide and subsequent hydrolysis of the silyl ethers 10b, 11b [13b] proceeded without significant loss of 18 O. In order to resolve the enantiomers of 9, the Preparation of Substrates ketone was treated with (ฯช)(2S, 3S)-dibutyl tartrate to give the spiroacetals 12a and 13a which were separated by re-The addition of hydrogen cyanide to 4-protoadamantanone (9) [12] was reported to give a mixture of the cy-versed-phase chromatography. The peaks of the butyl esters 12a, 13a overlapped less than those of the ethyl esters 12b, anohydrins 10a and 11a [13] . The isomers were now separated by HPLC, and the configuration was assigned by X-13b; the isomer eluting first was obtained with good purity.
Hydrolysis afforded (ฯช)-9 whose ee (95.2 ยฑ 0.2%) was esti-ray analysis of the 3,5-dinitrobenzoate 11d [14] . When the