One exception to the trend of increasing ionisation energy across a period in e.g. period 2 is going from N to O. There is no loss of exchange energy in I1 of O, but there is 2K lost in I1 of N, hence more energy is required to remove the valence electron of N.
If Zeff increases from N to O, and n stays constant, then
$$E=-R_H \frac{{Z_{eff}}^2}{n^2} $$ must become more negative, and I = -E, which implies that I increases, when in actual fact it decreases.
Is this equation therefore flawed?