I’m not understanding what you mean, might be a language/terminology problem. In the picture and in the example file, the bone high lighted can’t really be a control bone. It is part of the ik chain, and can’t control anything. By chain, I mean bones that are connected together and under control of the ik constraint. In your case, you set the ik chain length to 2, so you have 2 bones in that chain, neither of them can be controlled because they are controlled by the ik constraint. Your control bones are the hand and the polearm bones.
I think of a chain of bones as being a series of bones that are connected in some way. Either they have child/parent relationships and are connected (so the chain would end when there is no connected child/parent), or there is some other way that separates the bones (for example, you have 4 bones that are each parented to the next bone and all but the last one has inherit rotation set on them - then the chain would end where the inherit rotation of bones ends). So this is why I’m confused by this:
My question is: Is it possible to use bones for controlling in an ik chain or should I use 3 chains: 1 chain of bones for deforming, 1 chain for controllers, and 1 for IK.
because the controller bones wouldn’t make a chain.
Anyhow, in the example picture, you don’t want the high lighted bone to be rotated that far, so move the polearm bone forward and down and it shouldn’t rotate that high. I really don’t have good luck at limiting the rotation of bones in ik chain, usually it seams that if you limit rotation on one axis, the bone will flip and rotate on the other 2 axis to achieve pretty much the same position.