On the validity of the quasi-steady state approximation of bimolecular reactions in solution