The Jahn-Teller Effect

What the Jahn-Teller effect actually is

Start with a perfect octahedron: a metal ion at the center, six identical ligands at equal distances. Symmetry demands that certain d-orbitals be exactly equal in energy — the two e_g orbitals (d_z² and d_x²−y²), and the three t_2g orbitals (d_xy, d_xz, d_yz). If the electrons fill those degenerate levels unevenly — say one electron in a two-fold degenerate set — the molecule faces a paradox. There is no symmetric way to place that electron; putting it in d_z² and putting it in d_x²−y² are equivalent, and neither is preferred.

Hermann Jahn and Edward Teller proved in 1937 that such a molecule cannot stay symmetric. Their theorem says: for any non-linear molecule in an orbitally degenerate electronic state, there is always a distortion that lowers the energy. The molecule breaks its own symmetry, splitting the degenerate orbitals into a lower and a higher one, dropping the electrons into the lower one, and pocketing the energy difference. The distortion is not optional — it is forced by the mathematics of degeneracy plus a non-zero coupling between electrons and nuclear motion.

Crucially, the trigger is orbital degeneracy, not spin degeneracy. A high-spin d⁵ ion (Mn²⁺, Fe³⁺) has an orbitally non-degenerate ⁶A_1g ground state — every orbital singly occupied, perfectly symmetric — so it does not distort. The paradox only appears when a degenerate set is partially and unequally filled.

The mechanism: vibronic coupling, step by step

The Jahn-Teller effect is a competition between two energies as the molecule distorts along a symmetry-breaking coordinate Q (for an octahedron, the relevant mode is the e_g stretch that lengthens two trans bonds while shortening the other four).

1. Electronic energy falls linearly. As soon as the octahedron distorts, the degenerate orbitals split. One goes down in energy, the other up. Because the occupancy is uneven, the electrons preferentially sit in the orbital that dropped, so the electronic energy decreases linearly in the distortion: E_elec ≈ −A·Q. A linear term with no minimum at Q = 0 is exactly what "unstable" means.
2. Elastic energy rises quadratically. Bending bonds away from their ideal lengths costs strain energy like a spring: E_elastic ≈ +½k·Q².
3. The sum has a minimum off-center. Adding them, E(Q) = −A·Q + ½k·Q², is a parabola whose vertex sits at a non-zero displacement Q₀ = A/k. The molecule slides into that off-center well. The energy it gains there is the Jahn-Teller stabilization energy E_JT = A²/2k.

This linear-electronic-versus-quadratic-elastic tug of war is the entire physics. The name for the coupling between the electronic state and the nuclear motion is vibronic coupling (vibrational + electronic). It is the same term A that appears in far-reaching places — conical intersections, spectroscopic band shapes, and the theory of superconductivity in doped fullerides.

    Symmetric octahedron          Tetragonally elongated
    (degenerate eg, unstable)     (degeneracy removed, stable)

         L                              L
         |                              :        (long axial bonds)
     L — M — L      ── distort ──→   L — M — L    (short equatorial)
         |                              :
         L                              L

    E(Q) = −A·Q + ½k·Q²   →   minimum at Q₀ = A/k,   E_JT = A²/2k

Why some ions distort hard and others barely at all

The magnitude of A — how steeply the electronic energy falls — depends on which orbitals hold the degeneracy, because that sets how strongly moving a ligand perturbs the orbital.

• e_g degeneracy → strong effect. The e_g orbitals (d_z², d_x²−y²) point directly along the M–L axes. Lengthening two bonds strongly de-stabilizes one orbital and stabilizes the other, so A is large and Q₀ is big: bond-length differences of 0.1–0.3 Å that are unmistakable in an X-ray structure. This is the case for an odd number of e_g electrons: high-spin d⁴ (one e_g electron: Mn³⁺, Cr²⁺) and d⁹ (three e_g electrons = one e_g hole: Cu²⁺).
• t_2g degeneracy → weak effect. The t_2g orbitals point between the ligands, so distorting the octahedron barely changes their overlap. A is small, Q₀ is a few hundredths of an Å, and the distortion is usually smeared out by vibrations. Uneven t_2g fillings occur for d¹, d², low-spin d⁴/d⁵, and high-spin d⁶/d⁷ — all technically Jahn-Teller active, but weakly so.
• No effect at all. Configurations that fill each degenerate set evenly are non-degenerate and stay symmetric: d³ (t_2g³), high-spin d⁵ (t_2g³ e_g²), low-spin d⁶ (t_2g⁶), d⁸ (t_2g⁶ e_g²), and d⁰/d¹⁰. These are the "immune" configurations.

A one-line rule of thumb: look at the higher (e_g) set first. An odd e_g count means a strong, structurally obvious distortion; an even e_g count means at most a weak t_2g distortion.

Why elongation, not compression — and the Mexican hat

The first-order theorem guarantees a distortion but does not say whether the two trans bonds get longer or shorter — both a tetragonal elongation and a tetragonal compression remove the e_g degeneracy and both lower the energy to first order. In practice, almost every isolated Cu(II) complex elongates: two long axial bonds and four short equatorial bonds.

Two effects break the tie in favor of elongation. First, higher-order (second-order vibronic and anharmonic) terms in the potential slightly favor elongation. Second, the electronic bookkeeping is marginally better: elongation drops the doubly occupied d_z² and raises the singly occupied d_x²−y², so two electrons fall while only one rises — a net stabilization. For [Cu(H₂O)₆]²⁺ this gives axial Cu–O ≈ 2.3 Å and equatorial Cu–O ≈ 2.0 Å.

Because x, y, or z can each be the elongated axis, the potential energy surface as a function of the two e_g distortion coordinates is not a single well but a continuous circular trough — the famous "Mexican-hat" (sombrero) potential. Three equivalent minima sit in that trough (plus three saddle points for compression). Whether the molecule sits still in one well (static effect) or hops between them (dynamic effect) is the subject of a later section.

Worked example: assigning [Cu(H₂O)₆]²⁺ and [Cu(NH₃)₆]²⁺

Take the classic teaching case, the hexaaqua and hexaammine copper(II) ions, with the K₂CuF₄ lattice as a cooperative counterpart.

    Cu²⁺ is d⁹  →  t2g⁶ eg³  →  one hole in the eg set  →  STRONG Jahn-Teller.

    [Cu(H₂O)₆]²⁺   axial Cu–O ≈ 2.30 Å   equatorial Cu–O ≈ 1.95–2.00 Å
    [Cu(NH₃)₆]²⁺   axial Cu–N ≈ 2.60 Å   equatorial Cu–N ≈ 2.05 Å  (very elongated)
    K₂CuF₄ lattices show cooperative eg ordering of the long axes

1. Count d electrons. Copper(II): atomic number 29, so Cu²⁺ has 3d⁹.
2. Fill the octahedral levels. t_2g⁶ e_g³ — the e_g set holds three electrons, i.e. one hole, an unequal (degenerate) occupancy.
3. Predict. Odd e_g occupancy ⇒ strong distortion ⇒ tetragonal elongation with two long axial bonds and four short equatorial bonds.
4. Confirm spectroscopically. The single broad, asymmetric d-d band of aqueous Cu²⁺ near 12,000–13,000 cm⁻¹ (which gives the pale blue color) is broad precisely because the Jahn-Teller distortion splits the electronic levels, so the one nominal transition becomes several closely spaced components — a textbook fingerprint. A truly octahedral d⁹ ion would show a sharper band.

Contrast this with Ni(II), d⁸ (t_2g⁶ e_g²): the e_g set is evenly half-filled, non-degenerate, so [Ni(H₂O)₆]²⁺ is a regular octahedron with all six Ni–O bonds ≈ 2.05 Å. Same period, same charge, one electron different — and one distorts hard while the other does not.

Jahn-Teller vs neighboring ideas

	Jahn-Teller effect	Static crystal-field splitting	Pseudo (second-order) Jahn-Teller
Requires degeneracy?	Yes — orbital degeneracy in the ground state	No — always splits d-orbitals by field	No — needs two close non-degenerate states
Origin	First-order vibronic coupling	Electrostatic ligand field	Second-order mixing of ground + excited state
Energy dependence on Q	Linear (−A·Q) then quadratic recovery	Independent of geometry distortion	Quadratic — lowers force constant, may go negative
Classic example	Cu(II) d⁹, Mn(III) d⁴ elongation	Δoct in [Ti(H₂O)₆]³⁺	Pyramidal NH₃, pyramidal PH₃, ferroelectric BaTiO₃
Removes degeneracy?	Yes — that is the whole point	Sets up the splitting pattern	No degeneracy to remove; softens a bond
Linear molecules?	Excluded (theorem) — see Renner-Teller	N/A	Allowed — e.g. the bending of CO₂⁻

Static vs dynamic: what X-rays actually see

Sitting in one of the three equivalent Mexican-hat wells, the molecule has picked a long axis and holds it — a static Jahn-Teller effect. A diffraction experiment then resolves two long and four short bonds. This is the norm for concentrated Cu(II) and Mn(III) solids at ordinary temperatures, where a strong E_JT of ~1,500 cm⁻¹ ≫ kT keeps the distortion frozen.

But if E_JT is smaller, or the temperature higher, thermal energy lets the distortion hop rapidly between the three wells (long axis flipping among x, y, z faster than the experiment's timescale). This dynamic Jahn-Teller effect averages the three orientations, and the ion appears undistorted — an apparently regular octahedron with three equal, elongated bond lengths (all showing the same intermediate value). The tell-tale sign is unusually large, anisotropic thermal-displacement ellipsoids in the crystal structure, and a temperature-dependent EPR spectrum that sharpens into an axial signal on cooling. Cool a dynamic system enough, or impose an asymmetric lattice strain, and it "freezes out" into a static distortion — a genuine phase transition in many manganites and copper fluorides.

Where it matters: from manganites to lithium batteries

• Colossal magnetoresistance in manganites. In perovskites like La₁₋ₓCaₓMnO₃, the balance between Jahn-Teller-active Mn³⁺ (d⁴) and non-distorting Mn⁴⁺ (d³) governs whether the lone e_g electron localizes (trapped as a lattice polaron by the distortion) or delocalizes (hops freely, carrying current and ferromagnetism). A magnetic field can tip this balance, collapsing the resistance by orders of magnitude — colossal magnetoresistance. The Jahn-Teller electron-lattice coupling is the switch.
• Lithium-ion cathodes. LiMn₂O₄ spinel and Mn-rich layered oxides depend on Mn³⁺/Mn⁴⁺ cycling. When too much Mn³⁺ (d⁴) accumulates on discharge, cooperative Jahn-Teller distortion triggers a cubic-to-tetragonal phase change that cracks particles and fades capacity — a central failure mode battery engineers design around. Ni³⁺ (low-spin d⁷ with one e_g electron) is likewise Jahn-Teller active in NMC cathodes.
• Molecular magnetism and spectroscopy. The broad, structured d-d bands of Cu(II) and Ti(III) and Mn(III) complexes carry Jahn-Teller signatures; EPR of Cu(II) resolves the axial g-tensor that the elongation creates.
• The C₆₀ anion and organic radicals. Adding an electron to buckminsterfullerene gives an orbitally degenerate anion that Jahn-Teller distorts; the same coupling, when the fullerides are doped, contributes to their superconductivity. Many aromatic radical cations (e.g. benzene⁺) distort for the identical reason.
• Photochemistry via conical intersections. A vibronic coupling of exactly the Jahn-Teller mathematical form creates conical intersections — funnels where excited states decay to the ground state in femtoseconds. These govern the cis-trans isomerization of retinal in vision and the photostability of DNA bases against UV damage.

Who found it, and when

The theorem was proved in 1937 by Hermann Arthur Jahn and Edward Teller in a short, group-theoretical paper ("Stability of polyatomic molecules in degenerate electronic states"). Legend, repeated by Teller himself, holds that Lev Landau challenged them at a Copenhagen meeting to prove that a degenerate state must be unstable; Jahn and Teller then worked through every degenerate point group case by case and found that non-linear molecules always have an active distortion mode, while linear molecules are the sole exception. Teller is better remembered for the hydrogen bomb, but this early result is a cornerstone of inorganic and solid-state chemistry.

The linear-molecule exception was worked out by Rudolph Renner (with Teller) and is called the Renner-Teller effect: in a linear molecule the bending mode couples to the degenerate electronic state only quadratically, so it softens the bending vibration rather than forcing a static distortion. The extension to nearly-degenerate (rather than exactly degenerate) states — the pseudo-Jahn-Teller effect — was developed largely by Isaac Bersuker from the 1960s onward and explains molecular shapes (why NH₃ is pyramidal, why BaTiO₃ is ferroelectric) that have no ground-state degeneracy at all.

Common pitfalls and subtleties

• Confusing spin and orbital degeneracy. Only orbital (spatial) degeneracy triggers the effect. High-spin d⁵ is spin-degenerate (S = 5/2) but orbitally non-degenerate, so it does not distort.
• Forgetting the linear-molecule exception. The theorem explicitly excludes linear geometries; CO₂ in a degenerate state bends via the Renner-Teller (quadratic) mechanism, not a first-order Jahn-Teller distortion.
• Assuming a "regular" X-ray structure means no effect. An apparently undistorted octahedron with abnormally large thermal ellipsoids is very often a dynamic Jahn-Teller ion averaging its three wells — cool it and the distortion appears.
• Reading too much into the sign. First-order theory does not predict elongation vs compression; you need higher-order terms. Do not claim the theorem "requires" elongation — it merely requires a distortion.
• Expecting a big effect from t_2g ions. Ti³⁺ (d¹) and low-spin Fe³⁺ are formally Jahn-Teller active but the distortion is tiny and often undetectable — do not over-interpret it in a synthesis problem.