Bridging Versus Hydride Shift in Gaseous Cations: Hydroxy as a Vicinal Substituent

Primary cations of the form XCH2CH2 (1) are unstable with respect to the bridged isomer cyclo-(CH2)2X (2) or the hydride-shift isomer CH3CHX (3).Two independent hydrogen isotope experiments for X ) OH demonstrate that gaseous HOCH2CH2 (1a) gives less of its bridged isomer, protonated oxirane (2a), than of its hydride-shift isomer, protonated acetaldehyde (3a).Metastable ion decompositions of ionized HOCH2CH2OPh show that the radical cation decomposes via an ion-neutral complex containing 3a and phenoxy radical. Deuterium labeling studies exhibit no detectable interconversion of the two sp3-carbons, implying that complexes containing 2a are not formed to a measurable extent. An independent neutral product study looks at the radioactive oxirane and acetaldehyde produced when tritium on the methyl group of gaseous CH2TCHTOH undergoes radioactive decay. Loss of a betaparticle and a helium atom forms transient, tritiated 1a. Ions from isomerization of this radiolabeled primary cation were deprotonated by Me3N and gave a >97% yield of acetaldehyde. Quantitative assessment of possible routes to acetaldehyde (including rearrangement of excited 2a) implies that 1 undergoes hydride shift at least 5 times faster than bridging by neighboring oxygen. Nonconjugated primary carbocations are notoriously labile and rearrange rapidly to more stable isomers. For â-substituted ethyl cations (1) two rearrangement pathways compete with one another: neighboring group participation via bridging (i) versus hydride shift (ii). These pathways tend to operate in concert with bond heterolysis when solvolysis produces cations with the nominal structure 1. In solution, bridging predominates when X is an amino group1 or a halogen (Cl, Br, or I),2 but it is not easy to form 1 (even nominally) when X ) OR3 (although more highly substituted carbon skeletons are known to exhibit oxygen bridging1). In the gas phase, bridging to form 2b or 2c has been reported when radical cations heterolyze to yield 1 in cases where X ) NH2 or F.5,6 Competition between bridging and hydride transfer is known for neighboring OH in reactions of higher homologues, such as loss of neutral YH from CH3CH(OH)CH(YH)CH3 ions in the gas phase,7 but the simplest system 1 (or its nominal equivalents) is less well documented for X ) OH. Bridged ions (2) containing oxygen have higher heats of formation than their acyclic isomers (3). Gaseous, protonated acetaldehyde (3a) is ∆H3af2a ) 110 kJ mol-1 (26 kcal mol-1) more stable than protonated oxirane (2a).8 This is close to the value for X ) NH2, ∆H3bf2b ≈ 100 kJ mol-1,8,9 where kinetic control favors 2b. For X ) halogen, thermodynamic control should also favor pathway ii unless the bridging atom is bromine or larger.10 Bridging tends to reflect intramolecular SN2-type chemistry, as shown by stereochemical experiments on gaseous 2,3disubstituted butanes.7 Epoxides form rapidly in solution from YCH2CH2X via loss of leaving group Y when X ) O-, because the anionic nucleophile rapidly effects backside displacement. Internal SN2 also takes place when X is an amine. This paper compares rearrangement of “free” gaseous 1 with gas-phase C-Y bond heterolysis in XCH2CH2Y ions. The outcome is not what might have been expected on the basis of higher homologues. The most reliable published ab initio calculations suggest that â-hydroxyethyl cation (1a) does not correspond to a stable geometry.11 Ion 1a should have, at best, a fleeting existence, even when rigorously isolated from other molecules. Two experimental techniques are available for probing gaseous primary cations of this nature. Under one set of experimental conditions, 1a nominally intervenes during C-Y heterolysis of a HOCH2CH2Y precursor, in which eq 1 probably operates in concert with the bond fission. The other experiment examines “free” 1a created on the time scale of a single molecular vibration. In either case we find that the favored rearrangement occurs via pathway ii. X Abstract published in AdVance ACS Abstracts, August 15, 1997. (1) Streitwieser, A., Jr. SolVolytic Displacement Reactions; McGrawHill: New York, 1962. (2) Olah, G. A. Halonium Ions; Wiley-Interscience: New York, 1975. (3) Olah, G. A.; White, A. M.; O’Brien, D. H. In Carbonium Ions, Olah, G. A., Schleyer, P. v. R., Eds.; Wiley-Interscience: New York, 1973; Vol. IV, pp 1697-1781. (4) Van de Sande, C. C.; Ahmad, S. Z.; Borchers, F.; Levsen, K. Org. Mass Spectrom. 1978, 13, 666-670. (5) Ciommer, B.; Schwarz, H. Z. Naturforsch. 1983, 38B, 635-638. (6) Nguyen, V.; Cheng, X.; Morton, T. H. J. Am. Chem. Soc. 1992, 114, 7127-7132. (7) Angelini, G.; Speranza, M. J. Am. Chem. Soc. 1981, 103, 38003806. (8) Lias, S. G.; Liebman, J. F.; Holmes, J. L.; Levin, R. D.; Mallard, W. G. J. Phys. Chem. Ref. Data 1988, 17, Suppl 1. (9) Lossing, F. P.; Lam, Y.-T.; Maccoll, A Can. J. Chem. 1981, 59, 2228-2231. (10) Reynolds, C. H. J. Am. Chem. Soc. 1992, 114, 8676-8682. (11) For X ) OH (1a) SCF calculations using the 6-31G* basis set have been reported, which predict a stable classical â-hydroxyethyl cation (Benassi, R.; Taddei, F. J. Mol. Struct. 1990, 205, 177-190). Higher level computations, however, exhibit a negative force constant (Bock, C. W.; George, P.; Glusker, J. P. J. Org. Chem. 1993, 58, 5816-5825). These latter authors calculate a 1.2 eV energy barrier (MP2/6-31G*//MP2/6-31G*) for isomerization of 2a to 3a. (1) 8342 J. Am. Chem. Soc. 1997, 119, 8342-8349 S0002-7863(96)04475-7 CCC: $14.00 © 1997 American Chemical Society Experimental Section Mass-Resolved Ion Kinetic Energy Spectra (MIKES). Mass spectrometric experiments were performed on VG ZAB-2F and VG ZAB-2SE (B-E) double-focusing instruments with 70 eV electron ionization. Commercial 2-phenoxyethanol (4) was used without further purification, while 4-R,R-d2, 4-â,â-d2, C6D5OCH2CH2OH (4-d5), and PhOCD2COOH were prepared (>98 atom % D) as previously described.6 1,1,2,2-Tetradeuterio-2-phenoxyethanol (4-R,R,â,â-d4 >98 atom % D) was prepared by LiAlD4 reduction of PhOCD2COOH. O-deuterated phenoxyethanols were prepared by exchange with D2O in the ZAB source. Neutral Products from 3H Decay. Ditritiated ethanol 9 was freshly prepared by transesterification of tritiated ethyl docosanoate (specific activity ∼0.05 Ci μmol-1 from catalytic reduction of vinyl docosanoate with tritium gas, performed at the National Tritium Labeling Facility) with docosanol. This tritiated ester had been purified by preparative HPLC a few weeks before conversion to 9. The transesterification was catalyzed by titanium tetradocosoxide and performed in a sealed glass apparatus under vacuum. Vapor of the radiolabeled ethanol product was condensed in a liquid nitrogen-cooled bulb containing frozen NMe3 and the apparatus sealed and allowed to come to room temperature, where its contents were entirely gaseous. Thus, a 0.02 Ci sample of gaseous tritiated ethanol was sealed in a 1 L Pyrex bulb with 0.5 atm of NMe3 and allowed to stand in the dark at room temperature for 11 months (2.9 × 107 s ) 0.075 half-lives). In this interval 5% of the tritium nuclei decay. The gaseous reaction mixture was then condensed at liquid nitrogen temperature, and a large excess of unlabeled oxirane was vacuum transferred into the bulb and mixed with the condensate. A portion was transferred to a sealed tube and reacted with phenol (containing 8 mol % sodium phenoxide) to give PhOCH2CH2OH, which was converted to its N-phenylcarbamate ester, recrystallized 3 times from ethanol and once from CCl4, and radioassayed by liquid scintillation counting. From the specific activity of tritium in this solid derivative we infer the yield of tritiated oxirane to have been e0.02 mCi. The remainder of the sample was diluted with a large excess of unlabeled acetaldehyde in ethanol solution, converted to the semicarbazone, recrystallized three times from ethanol, and radioassayed by liquid scintillation counting. From the specific activity of tritium in that solid derivative we infer the yield of tritiated acetaldehyde to have been 0.9 ( 0.1 mCi. Theoretical Calculations. Ab initio SCF calculations were performed by means of SPARTAN (Wavefunction, Inc.) and GAUSSIAN 94 (Gaussian, Inc.), with geometries optimized using Hartree-Fock (HF) based methods and the 6-31G** basis set. The basis set superposition error (BSSE) was estimated using the counterpoise method, and SCF zero-point energy differences at 6-31G** are scaled by a factor of 0.89 in determining calculated values of ∆H. To minimize spin contamination, single-point density functional calculations were performed on the HF-optimized geometries at B3LYP using