Here's what I understand about this argument. Versions of it has been used by Kripke and Chalmers and here by Plantinga to argue that we are not our bodies or brains. The argument goes like this: Imagine yourself without a body. That seems easy enough. Can you imagine your body without its body? It seems not. That would be patently contradictory. Thus, your mind must have at least one property that your body doesn't: namely the modal property of possibly existing without a body. By Leibniz's law, they cannot be identical. That law simply says that everything has the property it has. If A has a property B does not have, they are not identical by that law. Your mind would have the modal property of possibly existing without a body while your body cannot exist without itself.
There are at least two ways to attack this. One may wish to deny Leibniz's law for modal properties; that is, one may wish to assert that it is possible for some things to have some modal properties they don't have. But I don't see this as a plausible avenue. It seems to me that Leibniz's law is true unrestricted, true for all properties, genuine, relational and modal.
So I will concentrate on the other possible objection which I will term the epistemic constraint to modality (ECM). It may seem plausible at first that anything one can imagine being true may be true in some possible world. Your car may be green when it is actually red. We know that your car has the modal property of possibly being green. How do we know that? We just imagine it so and if we can imagine it so, it is possible it is so. That's how we seem to know what is and isn't possible in the broadest sense. I can imagine your car being green when it's red and thus it must have the modal property of possibly green. I cannot imagine your car being green all over and red all over at the same time and thus your car must not have the modal property of being possibly green and red all over at the same time.
But does this kind of epistemic access always cut modality at its joints? Cut it precisely such that the space of epistemic possibility covers the same space as the modal space? Let's imagine we live 500 years ago. We would know what water was. It's that clear, potable, odorless liquid. Can we imagine it not being H2O? Yep, I believe we can. We could imagine it being XYZ for example. But we now know that water = H2O. It could not have been possible that water did not turn out to be H2O because water = H2O is a necessary identity statement. This is a case where our epistemic space over steps the boundaries of possibility. Thus zombies may not be possible though we can (now) imagine them.
Defenders of the claim that epistemic space and modal space are perfectly overlapping can deny that people 500 years ago really can imagine that water = H2O because they are not thinking about water when they imagine it not being H2O. But this doesn't seem like a way to defend that claim because then how would we know we are talking about or believing anything about some putative object when we imagine it being so (having some property) when science may tell us one day that it does not have that property? It doesn't seem that my ability to refer and to think about things are so dependent on contingencies in future scientific discovery.