On the breakdown of the Koopmans' theorem for nitrogen