CCL: hemiketal formation, transition state optimization

 Sent to CCL by: "Dr. Paul Elsinghorst" [paul.elsinghorst^-^gmail.com]
 I'm having trouble optimizing a transition state when going from an
 open ketone/phenol to a cyclic hemiketal.
 I chose the cyclic hemiketal with a proton still attached to the
 alcohol oxygen (positive formal charge) and a negatively charged
 ketone oxygen as the input structure for TS optimization. I
 preoptimized the structure applying a MMFF94 optimization and
 subjected that to ORCA giving:
 ! RKS B3LYP TZVP TightSCF SmallPrint Grid4 NoFinalgrid
 ! PAL8 NoSOSCF SlowConv OptTS NumFreq COSMO(Water)
 %geom Calc_Hess true
 end
 * xyzfile 0 1 TAXI_6_TS1.xyz
 Unfortunately ORCA optimizes the structure towards the open
 ketone/phenol which is obviously not the TS but finally a local
 minimum.
 Does someone know what's going wrong here? A further question is: does
 someone know, whether the proton in hemiketal formation is the one
 > from the alcohol migrating to the ketone? Is there a rule for this or
 does it differ depending on solvent? E.g. in protic solvents it might
 be relayed from solvent and in aprotic solvents it might migrate.
 I appreciate any ideas/help. Especially if I'm totally wrong in my approach.
 Thanks, Paul