As it is aligned along coordinate axes, the dx2−y2 usually forms σ bonds in preference to π bonds, seeing as how ligands bond along the coordinate axes as the internuclear axes.
An example is in square planar transition metal complexes...
![Adapted from Inorganic Chemistry, Miessler et al., 5th ed., pp. 365, 377]()
...or in octahedral transition metal complexes:
![Adapted from Inorganic Chemistry, Miessler et al., 5th ed., pp. 365, 368]()
One "exception" (among others) is trigonal bipyramidal transition metal complexes, where the dx2−y2 does not quite provide a direct σ interaction. In that case, it only is a partially direct overlap and not an ideal σ bond.
![Inorganic Chemistry, Miessler et al., 5th ed., pg. 386]()
The dx2−y2 orbitals would lie along the coordinate axes defined by ligand "2" and the wedge bond, whereas the ligands would lie on the corners of the dashed triangle.
[See this answer for further detail on the dx2−y2 orbital.](https://socratic.org/questions/what-is-the-significance-of-3dxsqaure-ysquare-in-the-d-subshell)
On the other hand, the dxy, dxz, and dyz tend to form π bonds, and rarely δ bonds.
![Inorganic Chemistry, Miessler et al., 5th ed., pg. 370]()
One note here is that usually the internuclear axis is the z axis, but with some exceptions.
The above diagram for octahedral complexes uses the convention that in polyatomic compounds, the y axis is the internuclear axis in the coordinates of the outer (non-central) atoms.
