Jahn-Teller Distortion

A symmetric molecule that can't sit still

Picture a copper(II) ion at the center of six identical water molecules, arranged as a perfect octahedron — three mutually perpendicular axes, six equal Cu–O bonds. It looks like it should be the most stable, most symmetric arrangement possible. It isn't. Within that perfect cage the d-electrons are in an awkward bind, and the molecule responds by spontaneously stretching two of the bonds and squeezing the other four. The cube collapses into a slightly squashed-then-stretched shape. That spontaneous, symmetry-lowering deformation is the Jahn-Teller distortion.

The reason is electronic, not steric. In 1937 Hermann Jahn and Edward Teller proved a sweeping theorem: any non-linear molecule whose electronic ground state is orbitally degenerate is geometrically unstable. If two (or more) orbitals of the same energy are unequally occupied, the molecule can always find a distortion that splits those orbitals, lowers the energy of the filled one, and so lowers the total energy. Nature takes the free lunch. The only molecules exempt are linear ones (where the proof breaks down) and molecules with no orbital degeneracy in the first place.

So the headline is counter-intuitive: maximum symmetry is sometimes a position of maximum energy, a hilltop the molecule rolls off. The distorted, lower-symmetry shape is the real ground state.

The d-orbital splitting, step by step

Start from crystal field theory. In an ideal octahedron the five d-orbitals split into a lower t₂g set (d_xy, d_xz, d_yz — lobes point between the ligands) and an upper e_g set (d_z² and d_x²−y² — lobes point straight at the ligands), separated by the crystal-field splitting Δ_o.

Now count electrons for Cu²⁺, a d⁹ ion. The t₂g set is full (6 electrons). The eg set holds the remaining 3 electrons in 2 orbitals — so one eg orbital is doubly occupied and the other singly occupied. But which one? By symmetry d_z² and d_x²−y² are degenerate, so the electron has no preference. That tie is the degeneracy the theorem forbids.

The molecule breaks the tie by distorting. If it elongates the two bonds along z, the ligands along z move away, so any orbital with density along z is repelled less and drops in energy:

                IDEAL Oh            ELONGATED (z stretched)
                                    ┌──── dx²−y²   ↑  (rises)
   eg  ── dz², dx²−y² ──            │
        (degenerate)        ───────┤
                                    └──── dz²       ↓  (drops, holds 2 e⁻)

                                    ┌──── dxy        ↑  (small rise)
   t₂g ── dxy,dxz,dyz ──    ───────┤
        (degenerate)                └──── dxz, dyz   ↓  (small drop)

Elongating along z lowers d_z² (its big lobe points along the now-distant z ligands) and raises d_x²−y². We put the two electrons in the lowered d_z² and the one electron in the raised d_x²−y². The net change: two electrons go down by δ₁/2 each while one goes up by δ₁/2, a net stabilization of δ₁/2 in the eg set. The t₂g set splits too (by a smaller amount δ₂) but is full, so its splitting is energy-neutral. Result: the molecule is more stable elongated than it was symmetric. That is the entire mechanism.

The vibronic energy balance

The distortion is a competition between an electronic gain that is linear in the distortion coordinate Q and an elastic (bond-strain) penalty that is quadratic. Write the energy of the lower split level as:

E(Q) = ½ k Q²  −  |A| Q

   ½ k Q²   = elastic cost of bending bonds (k = force constant)
   |A| Q    = electronic stabilization, linear in distortion (A = vibronic coupling)

Minimize by setting dE/dQ = 0:

dE/dQ = k Q − |A| = 0     →     Q₀ = |A| / k

E(Q₀) = ½ k (A/k)² − |A|(A/k) = −A² / (2k)    ← the Jahn-Teller stabilization energy E_JT

Two consequences fall straight out of this. First, because the linear term always beats the quadratic one for small Q (the slope at Q = 0 is −|A| ≠ 0), the minimum is never at Q = 0 — the symmetric structure is always unstable, exactly as the theorem promised. Second, E_JT = A²/(2k): a large vibronic coupling or a soft (small-k) bond gives a deep distortion. For Cu(II) aqua and ammine complexes E_JT is typically 10–25 kJ/mol, with Q₀ corresponding to an axial bond lengthening of ~0.2–0.4 Å.

The full picture is a two-dimensional "Mexican-hat" potential in the (Q_θ, Q_ε) plane of the eg distortion mode. Including only the linear term gives a continuous circular trough (the molecule could rotate its long axis freely). Adding the cubic anharmonic warping term carves three equivalent minima into the brim — the three choices of elongation axis (x, y, or z) — separated by three saddle points. The depth of those wells versus thermal energy kT decides whether the distortion is static or dynamic.

Which configurations distort — and how much

The theorem's force depends entirely on where the degeneracy lives. eg orbitals point at the ligands, so an uneven eg occupancy creates a strong distortion. t₂g orbitals point between ligands, so an uneven t₂g occupancy creates only a weak one. Configurations with a fully symmetric occupancy show no first-order effect at all.

dⁿ configuration	Example ion	Degenerate set	Jahn-Teller strength
d¹, d² (high-spin)	Ti³⁺, V³⁺	t₂g	Weak
d³	Cr³⁺	none (t₂g³)	None
high-spin d⁴	Cr²⁺, Mn³⁺	eg (eg¹)	Strong
high-spin d⁵	Mn²⁺, Fe³⁺	none (t₂g³eg²)	None
high-spin d⁶, d⁷	Fe²⁺, Co²⁺	t₂g	Weak
low-spin d⁷	Co²⁺, Ni³⁺	eg (eg¹)	Strong
d⁸	Ni²⁺	none (t₂g⁶eg²)	None
d⁹	Cu²⁺	eg (eg³)	Strong
d¹⁰	Zn²⁺	none (filled)	None

The mnemonic: an odd number of electrons in the eg set is the trigger. high-spin d⁴ (eg¹), low-spin d⁷ (eg¹), and d⁹ (eg³) are the three big distorters. Note d⁸ Ni²⁺ has eg² (one electron in each eg orbital, equally) — symmetric, no effect — which is why Ni²⁺ octahedra stay regular while Cu²⁺ octahedra never do.

Worked example: stabilizing [Cu(H₂O)₆]²⁺

The hexaaquacopper(II) ion is the textbook case. X-ray and EXAFS measurements give an elongated octahedron with four short equatorial bonds and two long axial bonds:

[Cu(H₂O)₆]²⁺   equatorial Cu–O ≈ 1.96 Å   (4 bonds)
               axial      Cu–O ≈ 2.40 Å   (2 bonds)
               elongation Δr ≈ 0.44 Å

Estimate the electronic stabilization. The eg splitting δ₁ induced by the distortion is roughly 8,000–9,000 cm⁻¹ (read off the splitting of the d-d band). Converting to molar energy:

δ₁ ≈ 8,500 cm⁻¹
1 cm⁻¹ = 11.96 J/mol
δ₁ ≈ 8,500 × 11.96 ≈ 101,700 J/mol ≈ 102 kJ/mol  (full eg gap)

Net eg stabilization = δ₁ / 2  (two e⁻ down δ₁/2, one e⁻ up δ₁/2)
                     ≈ 51 kJ/mol  raw electronic gain
minus the elastic strain cost ½kQ² ≈ 30–40 kJ/mol
→ E_JT (net) ≈ 10–20 kJ/mol

So the molecule pays ~35 kJ/mol to bend the bonds and earns ~51 kJ/mol back electronically, netting roughly 15 kJ/mol of stabilization — small in absolute terms (about 6× thermal energy at room temperature) but more than enough to fix the geometry. That modest depth is also why the effect can become dynamic: the wells are shallow enough that thermal hopping between the three axes is fast at 298 K but freezes out on cooling.

Static vs dynamic Jahn-Teller

The same molecule can look distorted or undistorted depending on temperature and timescale, because there are three equivalent elongation axes and the molecule may sample all of them.

	Static Jahn-Teller	Dynamic Jahn-Teller
Geometry observed	One fixed elongated octahedron	Time-averaged regular octahedron
Well depth vs kT	Wells ≫ kT (frozen in)	Wells ≲ kT (hopping fast)
Typical temperature	Low T (≲ 20–100 K)	Room temperature / above
X-ray structure	Three distinct bond lengths	Six apparently equal bonds
EPR spectrum	Anisotropic (g∥ ≠ g⊥)	Isotropic, averaged g
Origin of averaging	—	Pseudorotation among x, y, z axes
Classic example	K₂CuF₄ at low T	Cu(II) in MgO host at 300 K

The distortion never actually vanishes in the dynamic case — the molecule is elongated at every instant. It is only the average over the three axes, sampled slower than the hopping rate, that masquerades as a regular octahedron. Cool it down and the molecule gets trapped in one well, and the static three-bond-length pattern appears in the diffraction data.

Where Jahn-Teller shows up

• The color of copper(II) salts. The split eg/t₂g levels turn the single d-d band of an ideal octahedron into a broad, asymmetric absorption near 600–800 nm. That is the chromophore behind the familiar pale blue of copper(II) sulfate solution and the deep blue of the tetraammine ion.
• Manganite colossal magnetoresistance. In perovskites like LaMnO₃ the Mn³⁺ ions are high-spin d⁴ — strongly Jahn-Teller active. Cooperative ordering of the elongation axes throughout the lattice (orbital ordering) couples directly to the material's magnetic and electronic transport, and is central to the colossal magnetoresistance exploited in read heads.
• Lithium-ion battery cathode fade. When LiMn₂O₄ spinel is over-discharged, Mn averages below +3.5 and Jahn-Teller-active Mn³⁺ accumulates. The cooperative tetragonal distortion strains the lattice, cracking particles and causing the well-known capacity fade. Doping to suppress Mn³⁺ is a major battery-engineering lever.
• The fullerene cation. The C₆₀⁺ ion and many open-shell organic radicals show Jahn-Teller distortions that split otherwise-degenerate molecular orbitals — relevant to organic semiconductors and to the structure of charged fullerides.
• Enzyme active sites. Type-2 copper centers in proteins exploit Jahn-Teller flexibility: the soft axial bonds let the metal switch coordination geometry cheaply during redox cycling, tuning the Cu(II)/Cu(I) potential.

Common misconceptions and pitfalls

• "The theorem predicts the distortion." It predicts that a distortion occurs, not its size, sign, or direction. Whether a complex elongates or compresses, and by how much, comes from higher-order vibronic and anharmonic terms the first-order theorem says nothing about.
• "Jahn-Teller applies to any unsymmetrical molecule." No — it applies only to an orbitally degenerate electronic ground state. A d³ or d⁸ octahedron has a non-degenerate ground state and shows no first-order effect, no matter how you draw it.
• "Linear molecules distort too." The original theorem explicitly excludes linear molecules; their bending modes don't lift the degeneracy to first order. Linear systems instead show the related Renner-Teller effect, a second-order coupling.
• "It's a steric/crowding effect." It's purely electronic. Six identical small ligands with no steric clash still drive the distortion because the degeneracy is in the electrons, not the atoms.
• "A regular Cu(II) octahedron in an X-ray structure disproves Jahn-Teller." That is almost always the dynamic effect averaging the three axes. Variable-temperature EPR or low-T diffraction reveals the underlying distortion.
• "Filled or half-filled shells can still distort." d⁰, d⁵ (high-spin), d⁸, and d¹⁰ have symmetric occupancies and are Jahn-Teller-inactive in their ground state. Only an unequal population of degenerate orbitals triggers the effect.