Hi Roland

QST input for Gaussian can be made as long as you know the definite structures of reactants and products. Now, in your case, the structures of product is not known because the hydrogen atom leaves the propyl part to some infinite distance. in your case, i will suggest that you play around with the bond length of C-H and try Berny optimization of TS. 

there is another approach through which you can make qst input but that will be just an approximation. you are trying to break a C-H bond, and covalent bonding between C and H vanishes almost completely when the distance between them is more than 2.5 Angstroms. if you place a hydrogen atom at a distance of about 2.5 angstroms from carbon in products, then qst (approximate) can be made like the one shown here in attachement.

thanks