The CO trans to a phosphine ligand competes for metal pi donation with a weaker pi acceptor than CO itself, whereas a cis CO has another CO trans to it. The trans-to-phosphine carbonyl can therefore benefit from stronger M–CO pi backbonding; loss of one of the four cis carbonyls is commonly favored. Capture of the incoming phosphine at that vacancy gives the cis bis(phosphine) product. The fourfold statistical availability of cis sites reinforces an electronic preference but does not establish it by itself.

This is the carbonyl cis effect, rather than the square-planar trans effect applied without qualification. A fuller explanation compares the five-coordinate dissociation transition states and their ligand site preferences. Product geometry alone cannot prove which CO first departed, because the intermediate can rearrange before capture. Isotopically labeled CO exchange is a more discriminating experiment; the classic studies explicitly separate this issue from product stereochemistry.^[ref]

A dissociative substitution begins with slow CO loss,followed by capture of the incoming ligand in coordination chemistry. A more electron-releasing $L$ increases metal-to-CO pi backbonding in the six-coordinate reactant, stabilizing it and strengthening its M–CO attachments. Removing CO sacrifices a useful pi acceptor from that electron-rich center. Thus the activation free energy for CO dissociation often rises and the substitution slows. Within an otherwise matched series, $k_d\propto \exp [-\Delta G^{‡}/(RT)]$ captures this sensitivity. The rate depends on the relative energies of reactant and dissociation transition state, so the trend is conditional: large changes in steric hindrance, geometry or mechanism can override the electronic effect.

The nitrite ion is an ambidentate ligand: its Lewis structures leave donor lone pairs on both nitrogen and oxygen. In the N-bound mode, $\mathrm{M}-\mathrm{N}(\mathrm{O}{)}_2$, the nitrogen lone pair makes a sigma bond to the metal, while a suitable unoccupied nitrite molecular orbital can accept metal $d$ electron density by pi backbonding. Thus N-bound nitro is a sigma donor and pi acceptor. The nitrite group remains bent at nitrogen; N coordination does not make it a linear nitrosyl.

In the O-bound mode, $\mathrm{M}-\mathrm{O}-\mathrm{N}=\mathrm{O}$, an oxygen lone pair makes the metal–oxygen sigma bond. A second occupied oxygen orbital can donate sideways into an available metal orbital: O-bound nitrito is a sigma and pi donor, rather than the strong pi acceptor of the N-bound description. oxygen's greater electronegativity lowers its donor orbital energies, but the occupied oxygen orbitals still explain its pi-donor classification. Actual mixing depends on metal oxidation state and geometry.

A further possibility is bidentate coordination, $\mathrm{M}({\kappa }^2\text{-}O,O\text{-}\mathrm{N}{\mathrm{O}}_2)$, with both oxygen atoms attached to one metal. Each attachment has oxygen-to-metal sigma donation; additional pi donation depends on orbital alignment. nitrite can also be a bridging ligand between metals. Neither alternative should be confused with linkage isomerism, which specifically compares attachment through different atoms at the same binding site.