Your argument is slightly wrong, as you have realized from the fact that its extension would require violation of conservation of energy for any substance with k>1 or k<1. Imagine your plate is exceptionally large, to make edge effects negligible, and in free space, with a charge per unit area Q...