One of the most general reactions exhibited by coordination compounds is that of substitution, or replacement, of one ligand by another. This process is depicted in a generalized manner by the equation ML_{x− 1}Y + Z → ML_{x− 1}Z + Y for a metal complex of coordination number x. The ligands L, Y, and Z may be chemically similar or different. (Charges have been omitted here for simplicity.)

A class of substitution reactions that affords the widest possible comparison of different metal ions is the replacement of water in the coordination spheres of metal-aqua complexes in aqueous solution. The substitution may be by another water molecule (which can be labeled with the isotope oxygen-18 to permit the reaction to be followed) or by a different ligand, such as the chloride ion. Reactions of both types occur as shown below (oxygen-18 is indicated by the symbol ).

Many such reactions are extremely fast, and it has been only since 1950, following the development of appropriate experimental methods (including stopped flow, nuclear magnetic resonance, and relaxation spectrometry), that the kinetics and mechanisms of this class of reactions have been extensively investigated. Rates of substitution of metal-aqua ions have been found to span a wide range, the characteristic times required for substitution ranging from less than 10⁻⁹ second for monopositive ions, such as hydrated potassium ions, to several days for certain more highly charged ions, such as hexaaquachromium(3+) and hexaaquarhodium(3+). The rate of substitution parallels the ease of loss of a water molecule from the coordination sphere of the aqua complex and thus increases with increasing size and with decreasing charge of the metal ion. For transition metal ions, electronic factors also have an important influence on rates of substitution.

There are two limiting mechanisms (or pathways) through which substitution may occur—namely, dissociative and associative mechanisms. In the dissociative mechanism, a ligand is lost from the complex to give an intermediate compound of lower coordination number. This type of reaction path is typical of octahedral complexes, many aqua complexes, and metal carbonyls such as tetracarbonylnickel. An example of a dissociative reaction pathway for an octahedral complex of cobalt is as follows:

The associative mechanism for substitution reactions, on the other hand, involves association of an extra ligand with the complex to give an intermediate of higher coordination number; one of the original ligands is then lost to restore the initial coordination number. Substitution reactions of square planar complexes, such as those of the nickel(2+), palladium(2+), and platinum(2+) ions, usually proceed through associative pathways involving intermediates with coordination number five. An example of a reaction following such a pathway is

A characteristic feature of this class of reactions is the sensitivity of the rate of substitution of a given ligand to the nature of the ligand in the trans position. The trans ligand activates a ligand for replacement as follows, in decreasing order:

CO, CN⁻, C₂H₄> PR₃, H⁻> NO₂⁻, I⁻, SCN⁻> Br⁻, Cl⁻> NH₃, H₂O.

The trans effect may be used for synthetic purposes; thus, the reaction of the tetrachloroplatinate(2−) ion with ammonia yields cis-diamminedichloroplatinum, whereas the reaction of the tetraammineplatinum(2+) ion with the chloride ion gives the trans isomer, trans-diamminedichloroplatinum. The reactions are shown below.

In both reactions, the trans effect causes the introduction of the ligand trans to chloride rather than trans to ammonia.

Lability and inertness

In considering the mechanisms of substitution (exchange) reactions, Canadian-born American chemist Henry Taube distinguished between complexes that are labile (reacting completely in about one minute in 0.1 M solution at room temperature [25 °C, or 77 °F]) and those that are inert (under the same conditions, reacting either too slowly to measure or slowly enough to be followed by conventional techniques). These terms refer to kinetics (reaction rates) and should not be confused with the thermodynamic terms unstable and stable, which refer to equilibrium. For example, as mentioned above, most cyanide complexes are extremely stable (they possess very small dissociation constants); yet, if their rate of exchange with carbon-14-labeled cyanide, as represented in the following equation,

[M(CN)_x]^y−+ x¹⁴CN⁻⇌ [M(¹⁴CN)_x]^y−+ xCN⁻,

is measured, [Ni(CN)₄]²⁻ and [Hg(CN)₄]²⁻ are found to be labile, whereas [Mn(CN)₆]³⁻, [Fe(CN)₆]⁴⁻, [Fe(CN)₆]³⁻, and [Cr(CN)₆]³⁻ are inert. On the other hand, [Co(NH₃)₆]³⁺, a kinetically inert complex, is thermodynamically stable in acidic solution. Inertness may result from the lack of a suitable low-energy pathway for the reaction. In short, stable complexes possess large positive free energies of reaction (ΔG), whereas inert complexes merely possess large positive free energies of activation (ΔG^*).

While the existence of geometric or optical isomers (see above Isomerism) in the solid state or in solution at nonequilibrium concentrations is evidence supporting the inertness of the complex, this does not constitute absolute proof. Conversely, the possibility of intramolecular rearrangement means that failure to isolate geometric isomers or to resolve the racemic mixture into optical isomers is not absolute proof of lability.

Taube has interpreted lability of complexes according to their electronic configuration in terms of VB theory. Labile complexes are either of the outer orbital type (outer d orbitals involved in bonding—e.g., sp³d² as opposed to d²sp³ [inner orbital] for octahedral complexes) or of the inner orbital type with at least one vacant d orbital (available for accommodation of a seventh group during the [associative] substitution reaction).

Isomerization

Coordination compounds that exist in two or more isomeric forms (see above Isomerism) may undergo reactions that convert one isomer to another. Examples are the linkage isomerization and cis-trans isomerization reactions depicted below.

The first of these has been shown to proceed intramolecularly (i.e., without dissociation of the nitrite ligand), whereas the second probably occurs through dissociation of one of the water-molecule ligands.