Crystal field theory: predicting color and magnetism

A pale-blue copper sulfate solution, a deep-red iron(III) thiocyanate, a black manganese(IV) oxide, and a colorless zinc chloride all sit on the same shelf. They share a periodic-table row and a common d-block heritage, so why do their colors look nothing alike? Crystal field theory (CFT) is the smallest working model that answers this. It gets Ksp qualitative trends wrong and it gets covalent bonding wrong, but for color, magnetic moment, and geometry stability of first-row transition metal complexes it is the right cheap tool. This guide covers the split, the two big regimes (high spin vs low spin), and the four predictions that fall out of one diagram.

The idea in one paragraph

An isolated transition-metal ion has five degenerate d orbitals. When six ligands approach it octahedrally along the ±x, ±y, ±z axes, the d orbitals that point at the ligands (dz² and dx²-y², collectively labeled e_g) are electrostatically destabilized more than the ones that point between the ligands (dxy, dxz, dyz, collectively t_2g). The result is a splitting: two orbitals up, three orbitals down, with an energy gap Δ_o (Δ for “delta”, subscript o for “octahedral”). Electron configuration in the split levels determines color, magnetism, and thermodynamic stability. That is the entire theory.

$\underbrace{d^{n}\text{ in gas phase}}_{\text{5 degenerate}} \xrightarrow{6\ \text{ligands, octahedral}} \underbrace{t_{2g}^{a} e_g^{b}}_{a + b = n,\ \Delta_o \text{ apart}}$

What sets Δ_o: the spectrochemical series

The size of the split — Δ_o — depends on which ligands and which metal ion make up the complex. Empirically, ligands rank on a spectrochemical series from weakest field (smallest Δ_o) to strongest (largest):

$I^{-} < Br^{-} < Cl^{-} < F^{-} < H_2O < NH_3 < en < NO_2^{-} < CN^{-} \approx CO$

Metal-ion effects: Δ_o grows moving down a group (4d and 5d metals split about 40-80% more than 3d metals) and grows with oxidation state (Fe³⁺ splits more than Fe²⁺ with the same ligand set). Two of Miessler §10 numbers give the scale: for [Fe(H₂O)₆]³⁺ Δ_o ≈ 13,700 cm⁻¹; for [Fe(CN)₆]³⁻ Δ_o ≈ 35,000 cm⁻¹. Same metal, same oxidation state, different ligand, roughly 2.5× different split.

Filling the diagram: high spin vs low spin

For d⁴ through d⁷ configurations, electrons have a choice: pack four electrons into the lower t_2g (paying the pairing energy P for the fourth) or spread them across all five orbitals up to e_g (paying Δ_o but no extra pairing). The chosen configuration is whichever is energetically cheaper.

• Δ_o < P (weak field): electrons prefer to occupy separate orbitals — high spin, maximum unpaired electrons.
• Δ_o > P (strong field): electrons prefer to pair up in t_2g rather than climb to e_g — low spin, minimum unpaired electrons.

For d¹-d³ and d⁸-d¹⁰ there is no choice — the “spin state” is fixed regardless of Δ_o. Only d⁴, d⁵, d⁶, d⁷ have a high/low distinction.

Two consequences you can predict at a glance from the ligand and metal:

• Weak-field ligand (I⁻, Br⁻, Cl⁻, H₂O) → high spin (unless the metal is 4d/5d, which always low-spin).
• Strong-field ligand (CN⁻, CO, NO₂⁻) → low spin.
• 4d and 5d transition metals → almost always low spin (Δ_o is naturally large).

Predicting color

An electron in the t_2g can absorb a photon of energy Δ_o and get promoted to e_g. The absorbed wavelength is $\lambda = hc/\Delta_o$. The transmitted (or reflected) color is the complement of the absorbed wavelength on the color wheel. So a complex absorbing red light looks green; one absorbing yellow looks violet.

For [Ti(H₂O)₆]³⁺ (d¹), Δ_o ≈ 20,300 cm⁻¹, which puts absorption at 493 nm (green-cyan). The complement is red-violet, and the solution is indeed pale purple. This is one of the cleanest tests of CFT because d¹ has only one possible transition — no complications from multiple electron configurations.

Not every d-configuration produces a strong color. d⁰ (Ti⁴⁺, Sc³⁺, Zn²⁺-precursors) and d¹⁰ (Zn²⁺, Cu⁺, Ag⁺) cannot make a d→d transition — either no electron to promote or no vacancy to promote it into. These ions form colorless complexes (barring a charge-transfer transition, which is a different animal and shows up in permanganate, chromate, and other high-oxidation-state complexes). If a solution is deeply colored despite a d⁰ or d¹⁰ metal, charge transfer is running the show.

Predicting magnetism

The spin-only formula for the magnetic moment of a d-electron complex, in Bohr magnetons (μ_B), is:

$\mu_{spin} = \sqrt{n(n+2)}$

where n is the number of unpaired electrons. For low-spin [Fe(CN)₆]³⁻ (d⁵ low-spin, n=1): μ = √3 = 1.73 μ_B. For high-spin [Fe(H₂O)₆]³⁺ (d⁵ high-spin, n=5): μ = √35 = 5.92 μ_B. Both have iron(III), both are octahedral, and their magnetic moments differ by a factor of 3.4 because of a single ligand-substitution decision. Housecroft §21 has the full data set.

Diamagnetic vs paramagnetic falls out of the same count: n = 0 → diamagnetic (repelled by a magnet); n ≥ 1 → paramagnetic (attracted). A rough magnetic-susceptibility measurement is often the fastest way to distinguish high-spin from low-spin in a lab context.

Worked example — predicting color and magnetism of [Fe(CN)6]4−

Jahn-Teller distortion

Octahedral complexes with an odd electron count in the e_g level distort spontaneously — one axis elongates or compresses to remove the electronic degeneracy. This is the Jahn-Teller theorem, and its most common signature is Cu²⁺ (d⁹, one hole in e_g). Copper(II) complexes are almost never a perfect octahedron; they are axially elongated. The four equatorial bonds are typically 1.9-2.0 Å; the two axial are 2.3-2.5 Å. This is why Cu²⁺ makes such a mess of textbook geometry problems — its “octahedral” is a distortion of the octahedron. Shriver §20 has the eigenvalue argument.

Complexes with odd electron count in t_2g (d¹, d², low-spin d⁴, low-spin d⁵) also distort, but the effect is much smaller because t_2g points between ligands and the electron-ligand interaction is weaker.

Tetrahedral and square-planar cases

Not every complex is octahedral. Two secondary geometries have their own splitting patterns:

• Tetrahedral (4 ligands, T_d symmetry): the split inverts — dxy, dxz, dyz (labeled t_2 here) point closer to the tetrahedral ligands and go up; dz² and dx²-y² (labeled e) point away and go down. The gap is smaller, Δ_t ≈ (4/9) Δ_o for the same metal and ligand. Because Δ_t is small, tetrahedral complexes are almost always high-spin.
• Square planar (4 ligands, D_4h): derived from octahedral by removing the two axial ligands. dz² drops dramatically; dx²-y² stays high. Common for d⁸ metals (Pd²⁺, Pt²⁺, Ni²⁺ with strong-field ligands), where all eight electrons pair in the lower four orbitals and the highest (dx²-y²) is empty — very stable, diamagnetic. Cisplatin is the therapeutic example.

Practice problems

Where to go next

• The full molecular-orbital treatment (LFT: ligand-field theory) fixes CFT’s covalency blindspot and correctly predicts π-donor/π-acceptor effects on Δ_o. Miessler §10.3 is the standard bridge.
• The Ksp calculator applies to transition-metal salts too, but be careful with color: many transition-metal hydroxides and sulfides precipitate with dramatic color changes that hint at the geometry of the solid-state complex, not just its solubility.
• The polyatomic ions cheatsheet is worth having open when assigning complex-ion charges — the counting is where mistakes accumulate.

The value of CFT is exactly its cheapness: a five-line splitting diagram plus a spectrochemical series predicts color, magnetism, and geometry for hundreds of complexes. When you need bonding covalency or metal-metal quadruple bonds, upgrade to ligand-field or full MO — but for a first look at any d-block complex, CFT is still the tool of choice.