1. A First Cause possibly exists. (Premise)
There doesn't appear to be any contradiction with the idea of a First Cause, and given the arguments against an infinite regress of non-temporal causes, (1) may even be an understatement.
2. Whatever is possible is either contingent or necessary. (Definition)
3. Whatever is contingent can be actualized. (Premise)
Think of some contingent entity, such as a tree. We know that trees and other contingent entities are capable of being actualized, e.g. transitioned from a state of potentiality only to a state of actuality. (3) seems more reasonable than not.
4. A First Cause cannot be actualized. (Premise)
Assume the opposite. If a First Cause were ever actualized, then it is either: a) actualized by something external to itself, or b) self-actualized. (a) is impossible, since if anything else actualized the First Cause, then the First Cause wouldn't really be first at all. (b) is also impossible, since nothing can actualize itself. In order to actualize itself, an entity must first exist to be actualized, which is contradictory. Because (a) and (b) are necessarily false, it must also be necessarily false that a First Cause can be actualized. Hence, (4) is correct.
5. Therefore, a First Cause is necessary. (From 1 - 4).
In his excellent book, The Cosmological Argument, atheistic philosopher William Rowe doesn't dispute any of the premises of Duns Scotus' argument. Rather, what Rowe attempts to do is demonstrate that if this argument were correct, it would lead to all kinds of absurdities:
Surely it is possible for an everlasting star to exist. The stars that exist are presumably not everlasting--for each star, let us suppose, there was a time before which it did not exist and there will be a time at which it ceases to exist. But this seems to be an empirical fact and not a matter of conceptual or logical necessity. The idea of an everlasting star does seem to be a non-contradictory idea, even if no star is in fact everlasting. Let us grant, then, that
i. it is possible for an everlasting star to exist.
Now clearly we must grant that
ii. it is impossible for an everlasting star to come into existence. (If x comes into existence then by definition x is not everlasting.)
Moreover, since if something is produced by something else then there was a time before which it did not exist, we have
iii. it is impossible for an everlasting star to be produced by something else. 
Rowe concludes that if Scotus' cosmological argument (SCA) is correct, it can also be used to demonstrate that an everlasting star exists. But since we know that everlasting stars do not exist, something must be wrong either in the validity of the argument or in one of its premises.
The problem with Rowe's counter-argument is that it assumes a type of modality not necessarily entailed by Scotus. Even a star that is everlasting is being actualized by the matter that composes it. The analogy, then, of a non-actualized everlasting star seems to be incoherent and disanalogous to the SCA. For, the First Cause that Scotus envisages is simply not actualized at all. In other words, the First Cause doesn't require anything non-essential to itself to sustain its existence. Presumably, the matter of an everlasting star could logically be reconfigured.
Rowe, much to his credit, includes a footnote of this point. What the SCA entails is logical modality, and not metaphysical modality only. Rowe's objection is only successful if the SCA were to only apply metaphysical modality to the First Cause. After relating the possibility that the SCA may be including logical modality, Rowe simply moves on to discuss the Leibnizian Cosmological Argument (LCA), and specifically Samuel Clarke's version of the LCA.
 William L. Rowe, The Cosmological Argument, Fordham University Press, 1998, pp. 52-53.