Crystal Field Stabilization Energy (CFSE)

CFSE remains the standard first-pass explanation for why hydration enthalpies, lattice energies, and ionic radii of first-row transition-metal M²⁺ and M³⁺ ions deviate from smooth trends predicted by charge and size alone — the double-humped curves that appear in essentially every inorganic chemistry textbook trace directly back to the arithmetic developed here. In mineralogy and materials science, CFSE site-preference energies (worked out in Example 2) explain why particular transition-metal ions selectively occupy octahedral versus tetrahedral holes in spinel-structured oxides (used in magnetic recording media, ferrite cores, and battery cathode materials such as LiMn₂O₄-type spinels), and why olivine and pyroxene silicate minerals in the Earth's mantle partition Fe²⁺, Ni²⁺, and Cr³⁺ so unevenly among their available coordination sites. In bioinorganic chemistry, CFSE arguments underlie why heme iron flips between high-spin (five-coordinate, deoxygenated) and low-spin (six-coordinate, oxygenated) states in hemoglobin and myoglobin — a spin-state change with a measurable CFSE difference that couples directly to the protein's quaternary conformational change (the basis of cooperative oxygen binding). Ligand substitution kinetics also trace to CFSE: complexes with maximal octahedral CFSE loss on going to a five-coordinate transition state (like d³ Cr³⁺ or low-spin d⁶ Co³⁺, both near the −1.2Δₒ to −2.4Δₒ range) are kinetically inert, which is precisely why classic substitution-inert coordination complexes used in mechanistic studies (Werner's original cobalt-ammine complexes among them) were chosen from exactly these configurations. Finally, CFSE reasoning foreshadows two closely related phenomena developed elsewhere: the systematic thermodynamic trends across the whole d-block (melting points, ionic radii, hydration enthalpies) collected under CFSE and Thermodynamic Trends, and the geometric distortions that occur when orbital degeneracy remains even after crystal-field splitting, treated under Jahn-Teller Distortion.

Ask the tutor: For octahedral high-spin d⁴ (e.g. Cr²⁺ or Mn³⁺), work out both the CFSE and why this configuration is the textbook trigger for a large Jahn–Teller distortion.

Picture a free, gaseous Mⁿ⁺ ion with its d electrons distributed with perfect spherical symmetry across five degenerate d orbitals. Now bring up six ligands along the ±x, ±y, ±z axes to build an octahedral complex. The ligands' electron density repels the metal d electrons, and because the d orbitals are not spherically symmetric, they don't all feel that repulsion equally. The two orbitals that point directly at the ligands — $d_{z^2}$ and $d_{x^2-y^2}$, together called $e_g$ — are pushed up in energy. The three orbitals that point between the ligand positions — $d_{xy}$, $d_{xz}$, $d_{yz}$, together called $t_{2g}$ — are pushed down. This splitting, of total magnitude $\Delta_o$ (or 10Dq), is the entire content of Octahedral Crystal Field Splitting. Here is the subtlety that CFSE exploits: splitting the orbitals does not, by itself, change the average energy of the d shell. If you spread one electron in each of the five orbitals (a spherical, uniform distribution — the "barycenter"), the total energy is unchanged from the free ion regardless of $\Delta_o$. The splitting only pays off energetically when electrons preferentially occupy the lower set. Since $t_{2g}$ is triply degenerate and $e_g$ doubly degenerate, conservation of the barycenter fixes the two levels at

(the barycenter rule: $3(-0.4\Delta_o) + 2(+0.6\Delta_o) = 0$). Any electron placed in $t_{2g}$ saves $0.4\Delta_o$ relative to the spherical average; any electron placed in $e_g$ costs $0.6\Delta_o$. CFSE is simply the sum of these savings and costs over the actual ground-state configuration, with one more twist: if the field is strong enough to force two electrons into the same orbital before all orbitals singly occupy (low-spin), each forced pairing costs an extra pairing energy $P$. Tetrahedral geometry inverts the whole picture. With four ligands sitting between the Cartesian axes (at alternate corners of a cube), it is now the $t_2$ set ($d_{xy}, d_{xz}, d_{yz}$) that points more nearly toward the ligands and rises in energy, while the $e$ set ($d_{z^2}, d_{x^2-y^2}$) drops. The barycenter rule again fixes the split, this time relative to $\Delta_t$:

Because there are only four ligands instead of six, and because none of them point directly at any d orbital lobe (they approach along diagonals, a geometrically less efficient overlap), $\Delta_t$ is intrinsically much smaller than $\Delta_o$ for the same metal and ligand set — close enough to $\frac{4}{9}\Delta_o$ that pairing almost never wins in tetrahedral complexes, which is why virtually all tetrahedral transition-metal complexes are high-spin.

The other essential ingredient is the pairing energy, $P$. Putting two electrons in the same spatial orbital costs energy from Coulombic repulsion (partly offset by a loss of exchange stabilization if the two electrons had been in different orbitals with parallel spins). For octahedral $d^4$–$d^7$ configurations, there is a genuine choice: keep electrons unpaired and spread across both $t_{2g}$ and $e_g$ (high-spin, cost $P$ avoided but $e_g$ populated) or pair them up within $t_{2g}$ before touching $e_g$ (low-spin, cost extra $P$ but keep all electrons in the low-lying set). The field "chooses" whichever configuration has lower total energy — this is exactly the branch point analyzed in High-Spin vs Low-Spin Complexes, and CFSE is the quantity that makes the comparison numerical rather than qualitative.

The general octahedral CFSE, counting $n_{t_{2g}}$ electrons in $t_{2g}$ and $n_{e_g}$ in $e_g$, with $n_p$ electron pairs forced beyond the free-ion's own minimum pairing, is

where: $n_{t_{2g}}, n_{e_g}$ are electron counts in each set (summing to the total d-electron count $n$); $\Delta_o$ is the octahedral splitting energy; $n_p$ is the number of electron pairs present beyond those already required in the free (spherical-field) ion's high-spin ground state; $P$ is the pairing energy per extra forced pair. Reading it: every electron in $t_{2g}$ subtracts $0.4\Delta_o$ from the ledger (stabilizing), every electron in $e_g$ adds $0.6\Delta_o$ (destabilizing), and every pairing forced specifically by the crystal field (not one that would have occurred anyway in a spherical field) adds a fixed penalty $+P$. CFSE is conventionally reported as a negative number when the configuration is net stabilized, which is the case for essentially all non-$d^0, d^5(\mathrm{HS}), d^{10}$ configurations. The tetrahedral analogue, with $n_e$ electrons in the $e$ set and $n_{t_2}$ in the $t_2$ set,

where: the roles of the coefficients are swapped relative to octahedral (because $e$ is now the low-energy set), and $n_p$ is almost always zero because $\Delta_t$ is rarely large enough to overcome $P$. A convenient practical relationship, used to compare the same metal ion and ligand set in the two geometries, is

where: this arises from the reduced number of ligands (4 vs 6, roughly a factor of $2/3$ in electrostatic field) and a purely geometric factor ($2/3$ again, from the different overlap of orbitals with ligand positions relative to axes), multiplying to $4/9$. It is the reason virtually every tetrahedral d-block complex is high-spin: $\Delta_t$ almost never exceeds typical pairing energies (~15,000–30,000 cm⁻¹ range for $P$ vs $\Delta_t$ typically 3,000–7,000 cm⁻¹).

Example 1 — Foundational: CFSE of octahedral high-spin d⁵ (Mn²⁺) and d⁶ (Fe²⁺), both weak-field. High-spin $d^5$ places one electron in each of the five orbitals: $t_{2g}^3 e_g^2$, with zero forced pairing (every orbital singly occupied, matching the free-ion maximum-multiplicity ground state exactly, so $n_p = 0$).

This is the diagnostic signature of high-spin $d^5$: exactly zero net CFSE, because the configuration is the closest analogue to the spherical barycenter itself (one electron per orbital, matching the barycenter's implicit assumption). This is precisely why Mn²⁺ (high-spin d⁵) sits almost exactly on the smooth Born–Landé-type trend line for hydration and lattice energies across the first transition series, with no anomalous extra stabilization. Now $d^6$ high-spin, using the electron counts $t_{2g}^4 e_g^2$ found in Derivation 2, with $\Delta_o(\mathrm{Fe^{2+}}, \mathrm{H_2O}) \approx 10{,}400\ \mathrm{cm^{-1}}$ and using $n_p = 1$ (one pair, forced beyond the zero-pairing baseline of $d^5$):

Using the electrostatic part alone (the part usually plotted against lattice/hydration data, since $P$ is roughly constant across the series and often omitted from such plots as a common offset):

Converting using $1\ \mathrm{cm^{-1}} = 0.01196\ \mathrm{kJ\,mol^{-1}}$:

So high-spin Fe²⁺ gains roughly 50 kJ mol⁻¹ of extra stabilization that Mn²⁺ does not — exactly the size and sign of anomaly seen in the experimental hydration-enthalpy curve, where Fe²⁺ sits measurably below (more exothermic than) the smooth reference line through Ca²⁺, Mn²⁺, and Zn²⁺. Limiting check: setting $\Delta_o \to 0$ (no field at all, i.e. a genuinely free ion) collapses both results to CFSE $\to 0$, as required — with no splitting, there is no stabilization to speak of, and the barycenter and real configuration coincide.

Example 2 — Applied: comparing octahedral and tetrahedral site preference for Ni²⁺ in a spinel-type lattice. Consider a hypothetical (illustrative) partitioning question: does high-spin d⁸ Ni²⁺ prefer the octahedral or tetrahedral hole in a mixed-oxide spinel host, all else equal? Take $\Delta_o = 8{,}500\ \mathrm{cm^{-1}}$ for Ni²⁺–O²⁻ (a representative oxide value) and use $\Delta_t = \frac{4}{9}\Delta_o$. Octahedral d⁸: fills $t_{2g}^6 e_g^2$ (six electrons pair up fully in $t_{2g}$, two singly occupy $e_g$ — this is forced regardless of field strength for $d^8$, since there is no high-spin/low-spin choice once $n=8$ leaves only two electrons for the two $e_g$ orbitals). Pairing here ($t_{2g}^6$ = three pairs) is identical in both weak and strong field, so no incremental pairing correction is needed for comparing sites (it cancels in fair comparisons, or is included consistently on both sides):

Tetrahedral d⁸: fills $e^4 t_2^4$ (four electrons fill $e$ completely — two pairs — and four fill $t_2$ with one pair and two singly occupied, by Hund's rule within the triple degeneracy):

With $\Delta_t = \frac{4}{9}(8{,}500\ \mathrm{cm^{-1}}) \approx 3{,}778\ \mathrm{cm^{-1}}$:

The octahedral site stabilizes Ni²⁺ by roughly $10{,}200 - 3{,}022 = 7{,}178\ \mathrm{cm^{-1}}$ (about 86 kJ mol⁻¹) more than the tetrahedral site. This large octahedral site-preference energy is exactly why Ni²⁺ (d⁸) essentially never occupies tetrahedral holes in spinel-type oxide lattices, forcing other cations (commonly d⁰, d⁵-HS, or d¹⁰ ions with zero or negligible CFSE in either geometry) to take the tetrahedral sites instead — a real structural consequence with no fitting parameters beyond $\Delta_o$ and the fixed $4/9$ ratio. Limiting check: if $\Delta_t$ were (incorrectly) taken equal to $\Delta_o$ rather than $\frac{4}{9}\Delta_o$, the octahedral preference would shrink to $-1.2\Delta_o - (-0.8\Delta_o) = -0.4\Delta_o$, understating real site-preference energies by more than a factor of two — a check that confirms why the $4/9$ geometric/electrostatic factor, not equality, must be used.

Example 3 — Exam/competition level: full CFSE and spin-state analysis for octahedral d⁷ Co²⁺ with two different ligand sets, including the pairing-energy tie-break. Take Co²⁺ (d⁷) in two complexes: $[\mathrm{Co(H_2O)_6}]^{2+}$ with $\Delta_o = 9{,}300\ \mathrm{cm^{-1}}$, and a strong-field analogue with $\Delta_o = 20{,}000\ \mathrm{cm^{-1}}$ (illustrative, to force a low-spin regime), taking $P(\mathrm{Co^{2+}}) \approx 22{,}500\ \mathrm{cm^{-1}}$ (a representative literature-range value for this ion). High-spin d⁷: configuration $t_{2g}^5 e_g^2$ — five electrons in $t_{2g}$ means one full pair plus three singly occupied ($t_{2g}$ can only hold 3 orbitals × 2 = 6 electrons max; with 5 electrons, Hund's rule gives 3 singly occupied + 1 pair, i.e. one forced pair beyond the maximum-multiplicity reference), and two singly occupy $e_g$. Forced pairs relative to the spherical/free-ion HS baseline: $n_p = 1$.

Low-spin d⁷: configuration $t_{2g}^6 e_g^1$ — $t_{2g}$ completely filled (three pairs), one electron singly in $e_g$. Total pairs = 3, i.e. two pairs beyond the HS baseline's one pair, so incremental pairing relative to HS is $+2P$ (not $3P$ absolute, since we track the difference from HS consistently, matching Derivation 2's method):

Compare at $\Delta_o = 9{,}300\ \mathrm{cm^{-1}}$ (aqua complex):

$E_{\text{HS}} \ll E_{\text{LS}}$: high-spin strongly favored, matching the well-known experimental fact that $[\mathrm{Co(H_2O)_6}]^{2+}$ is high-spin with three unpaired electrons (paramagnetic, $\mu_{\text{eff}} \approx 3.9$–$4.9\ \mu_B$ depending on orbital contribution, consistent with $S = 3/2$). Compare at $\Delta_o = 20{,}000\ \mathrm{cm^{-1}}$ (illustrative strong-field case):

High-spin still wins even at this unusually large $\Delta_o$. Solving the crossover algebraically (setting $E_{\text{HS}} = E_{\text{LS}}$):

With $P \approx 22{,}500\ \mathrm{cm^{-1}}$, crossover requires $\Delta_o \approx 45{,}000\ \mathrm{cm^{-1}}$ — larger than essentially any realistic octahedral ligand field for Co²⁺ (even strong-field ligands like CN⁻ rarely push a d⁷ ion's $\Delta_o$ that high while the ion remains +2). This is exactly why bona fide low-spin octahedral Co²⁺ is exceedingly rare in practice, in contrast to d⁶ Fe²⁺/Co³⁺, which cross over at the much more accessible condition $\Delta_o = P$ (Derivation 2). The factor-of-two difference in crossover conditions ($\Delta_o = P$ for $d^6$ vs $\Delta_o = 2P$ for $d^7$) is a genuine, checkable structural feature of the pairing arithmetic, not a coincidence of the chosen numbers. Limiting check: as $P \to 0$ (hypothetically no pairing penalty at all), the crossover condition $\Delta_o = 2P \to 0$, meaning any nonzero splitting would immediately favor low-spin — correctly recovering the trivial limit where pairing costs nothing and the field always wins by filling the lowest orbitals first.

The chemical application had to wait for a second generation of theorists — notably John Van Vleck in the 1930s and 40s, who reframed Bethe's group-theoretical splitting in terms of transition-metal d-orbital energetics and magnetism, and Leslie Orgel in the 1950s, who popularized the qualitative diagrams (the now-familiar $t_{2g}/e_g$ splitting pictures) and connected them explicitly to observable thermodynamic and spectroscopic anomalies across the transition series. It was Orgel's 1952 analysis of hydration and lattice energies — precisely the double-humped curve motivating this article's opening — that turned CFSE from an abstract orbital-energy bookkeeping exercise into a working predictive tool that inorganic chemists still use directly today. The synthesis of Bethe's electrostatic group theory with Van Vleck's magnetism and Orgel's thermodynamic applications is usually now taught as a unified "crystal field theory," even though it was assembled from three distinct research programs across three decades.