Which d orbitals can form sigma, pi, delta bonds?

If two $\ce{d_{xy}}$ orbitals approach each other on $x=y, z=0$, would a sigma bond be formed? I would think so.

Can $\ce{d_{z^2}}$ form pi bond with another $\ce{d_{z^2}}$? (as all others can on $x$ or $y$ or $z$ as internuclear axis?) I am not sure but overlapping similar to p could be observed.

I would assume all except $\ce{d_{z^2}-d_{z^2}}$ form delta bond. (both of same type like $\ce{d_{xy}}$ with $\ce{d_{xy}}$)

What are the faults in my logic?

• Atomic orbitals forming molecular orbitals are a way of thinking about chemical bonding. Neither atomic orbitals nor molecular orbitals really exist so they cannot be observed either. Do you have an example? One can then maybe tell you what "kinds of bonds" the d-Orbitals could form there. The concepts you are referring to have been given up by most people a long time ago and are mainly used today as a tool to teach people wrong things. They can, however, be relevant under very specific circumstances, hence an example would be great. – AMT Jan 18 '17 at 16:00
• sigma bond is overlapping on the internuclear axis and pi is perpendicular to it. LCAO, doesn't have mathematical problems to me, I am a novice though. I would just ask what stops from a sigma or pi bond to be formed, if two lobes of same phase overlap, theoretically, some sort of bonding molecular orbital would be formed right and with it a abmo? unless there is +,+ and +,- interference equally, then it would be non bonding. – Mrigank Jan 18 '17 at 16:14