after trying to obtain a minimum structure of a complex with several explicit water molecules, the solvent tends to react with the complex's ligand. Is there a way to avoid this during optimization by using some Gaussian09 controls in the input file?