Catalysis of Organic..

206 Fructose Hydrogenationcleave the oxygen to hydrogen bond of the hydroxyl group on the anomeric carbonto produce an adsorbed cyclic alkoxy species and an adsorbed hydrogen atom. Thisadsorbed hydrogen atom will then be well positioned to either return to the adsorbedalkoxy moiety to reproduce the hydroxyl group or to attack the ring oxygen resultingin a shift of electron density to the adsorbed alkoxy group for the development of aπ-bond thereby creating an acyclic adsorbed ketose. In other words, this is a surfaceinduced adsorbed ketal to adsorbed hydroxy ketose equilibrium via a 1,3intramolecular hydrogen atom shift. It has already been proven unambiguously withdeuterium labeling that olefin isomerization can occur via the perpendicularadsorption of the π-bond onto the metal, the abstraction of the lower allylic hydrogenatom (due to its closer proximity to the surface) to give an adsorbed hydrogen atomand a π-allyl species and the attack of that adsorbed hydrogen atom to the bottomside of the opposite end of the adsorbed π-allyl to produce the isomer (23). Thiswork also showed that there was no measurable mixing between the hydrogeninvolved in this isomerization and the nearby chemisorbed hydrogen. Althoughthere are clearly differences between olefin isomerization and the proposedequilibrium of the adsorbed ketal and hydroxy ketose species, there is no mechanisticreason to exclude the occurrence of a 1,3 intramolecular hydrogen atom shift duringthis equilibrium. Even if neighboring chemisorbed hydrogen atoms do participate,this would not change the conclusions of the mechanism proposed here.If the developing π-bond from this equilibrium is able to become perpendicularto the metal surface with very little or no strain and if the metal is able to stabilizethis developing π-character, there will be a shift to a more ketoselike species that canbe hydrogenated at a lower activation energy. If a particular tautomeric form is notable to accommodate this metal-stabilization of the developing π-bond, it will haveless keto character and this will undergo a reaction that resembles more thehydrogenolysis of the ketal at considerably higher activation energy. The reactionwill also have a higher activation energy and the adsorbed intermediate will have lessketo-character if the catalytic metal is not very effective at stabilizing the π-bondcharacter as it is being developed. Thus, the preferred reaction intermediate will beadsorbed by both the ring oxygen and the hydroxyl oxygen on the anomeric carbon(C2), where the bond between the ring oxygen and C2 is less than single and thebond between C2 and its hydroxyl oxygen is more than single. The partiallyadsorbed hydrogen atom from the C2 hydroxyl group will then reside somewherebetween the C2 hydroxyl oxygen and the ring oxygen. A nearby chemisorbedhydrogen atom could then attack the anomeric carbon to give a “half-hydrogenated”adsorbed alkoxy species that can easily react with another chemisorbed hydrogenatom to afford the polyol. Figure 19 displays the possible flat and edge adsorbedintermediates proposed here and since the above mentioned developing π-bond willneed to be perpendicular to the C1-C2-C3 plane, it is obvious that the orientation ofthis π-bond will be parallel to the metal surface for all of the edge adsorbed species.This lack of metal-stabilization for the edge adsorbed intermediates, means that theywill not be the major contributors to the resulting product distribution under normalreaction conditions and this also refutes the preferred adsorbed surface intermediatesproposed by Makkee et al (6).