In a grazing collision, any velocity orthogonal to the radius is the result of any residual orbital velocity the impactor may have had prior to the collision. There will be a radial velocity vector in the direction of the impactee exactly equal to the escape velocity. THAT energy comes from the object falling down the gravity well. Its inescapable. Its Newton.

Now it is certainly possible that an enormous cloud of debris was ejected after the collision which collected in a ring about the primary, and eventually a body coalesced from that. As I mentioned earlier, that's probably how our own moon formed.

Its easy to think of this situation as if they were billiard balls striking glancing blows on the felt. But billiard balls don't gravitationaly attract each other.

And it is dynamically impossible for two bodies to interact in such a way that they wind up in orbit about each other unless there is either a collision or some third body which provides the additional momentum and energy required. Two bodies approaching from infinity will either collide, or swing about each other and go their separate ways.

And don't forget, at least one of the bodies will probably be tidally disrupted by tidal forces prior to the collision. It would be like a melon being hit by a cloud of buckshot.