Special and High-Symmetry Point Groups

Octahedral symmetry is the working assumption behind the entire d-orbital splitting diagram used to rationalize color, magnetism, and reactivity in transition-metal chemistry — every time a textbook draws the t2g/eg splitting for a first-row transition-metal hexaaqua or hexaammine complex, it is implicitly invoking Oh symmetry, and any deviation (Jahn–Teller distortion, a trans-disubstituted complex) is described as a lowering from that reference point, using Assigning Point Groups by Flowchart logic to name the resulting subgroup (D4h, C4v, etc.). Tetrahedral symmetry governs the geometry at sp³ carbon centers throughout organic chemistry and at many important catalytic and structural metal sites — zinc in carbonic anhydrase and many zinc-finger proteins sits in an approximately tetrahedral N/O/S environment, and the vibrational and electronic spectroscopy of tetrahedral oxoanions (SO₄²⁻, PO₄³⁻, ClO₄⁻) is routinely interpreted using the Td vibrational analysis worked out above (one Raman-only A1 "breathing" mode, degenerate E and T2 modes) to fingerprint these anions in minerals, fertilizers, and biological samples via infrared and Raman spectroscopy. Icosahedral symmetry moved from a mathematical curiosity to a central chemical fact with the 1985 discovery of C₆₀ (buckminsterfullerene) and the subsequent explosion of fullerene and carbon-nanomaterial chemistry; endohedral fullerenes (metal atoms trapped inside the C₆₀ cage), fullerene-based superconductors (alkali-metal-doped C₆₀ salts), and fullerene derivatives used as electron acceptors in organic photovoltaics all inherit their electronic structure, in the undistorted parent cage, from Ih symmetry. Icosahedral symmetry also appears in inorganic cluster chemistry — the closo-borane B₁₂H₁₂²⁻ is a near-perfect Ih icosahedron of borons, a benchmark example in cluster (Wade's rules) chemistry — and, well beyond chemistry proper, in the protein shells of many spherical viruses, whose capsids assemble with icosahedral symmetry because it is the highest-symmetry, most efficient way to enclose a volume with a finite number of identical repeating protein subunits. Linear-molecule symmetry (C∞v/D∞h) remains the daily workhorse of rotational and vibrational spectroscopy: microwave and infrared spectroscopists use the mutual exclusion rule for D∞h molecules (like CO₂, central to greenhouse-gas infrared absorption and atmospheric remote sensing) and the simpler, unrestricted selection rules of C∞v molecules (like CO and HCN, used as interstellar-medium tracers in radio astronomy) as an immediate diagnostic of molecular symmetry from spectral appearance alone.

Ask the tutor: How do I figure out which irreducible representations span the six M–L σ-bonds in an octahedral complex, once I know the point group is Oh?

The ordinary point-group flowchart works by asking sequential yes/no questions: is the molecule linear? does it have a unique principal axis? are there perpendicular C₂ axes? This works beautifully for the vast majority of molecules, which have at most one axis of order 3 or higher. But a small, chemically vital set of shapes have several high-order axes simultaneously, and for these the flowchart's sequential logic becomes awkward — there is no single "principal axis" to start from, because several axes are equally principal. The practical fix is to recognize the shape directly.

Linear molecules (C∞v and D∞h). A linear molecule has a C∞ axis along the internuclear line: rotation by any angle about that axis is a symmetry operation, hence the "∞." There are only two possible point groups for a linear molecule, and the distinguishing question is a single one: does the molecule possess a center of inversion? If the two ends of the molecule are different (HCN, OCS, HF, any heteronuclear diatomic), there is no inversion center and the group is C∞v, containing E, C∞ (all rotations), and an infinite number of σv planes containing the axis — but no σh, no i, no S∞. If the two ends are identical or the molecule is symmetric about its center (N₂, CO₂, C₂H₂, H₂), the group is D∞h, which adds a C₂ axis perpendicular to C∞ (in fact infinitely many, since any line through the center perpendicular to the main axis is a C₂), a σh, an inversion center i, and an S∞ axis. The mnemonic is blunt but reliable: symmetric linear molecule → D∞h; unsymmetric linear molecule → C∞v. No other outcome is possible for a linear shape.

Tetrahedral symmetry (Td). Methane is the canonical example. Place carbon at the centroid of a tetrahedron with the four hydrogens at the vertices. The full symmetry group has order 24 and contains: the identity E; 8 C₃ operations (rotations by 120° and 240° about each of 4 axes, one through each C–H bond, i.e., 4 axes × 2 non-trivial rotations = 8); 3 C₂ operations (rotations by 180° about axes that bisect opposite H–C–H angles — there are 3 such axes, each passing through the midpoints of two opposite edges of the tetrahedron); 6 S₄ operations (two non-trivial S₄ rotations about each of the same 3 axes that carry the C₂'s, since S₄² = C₂ on those axes — 3 axes × 2 = 6); and 6 σd mirror planes (each containing one C₂ axis and two of the four C–H bonds, bisecting the H–C–H angle between the other two). Total: 1 + 8 + 3 + 6 + 6 = 24. Notice there is no inversion center and no σh — tetrahedral symmetry is "achiral but not centrosymmetric," a subtlety worth sitting with: Td molecules can still be superimposable on their mirror image (the σd's provide the achirality) even though there's no single point of inversion symmetry.

Octahedral symmetry (Oh). SF₆, or an idealized six-coordinate complex ML₆ with all six ligands identical, has an octahedron of ligands around the central atom. Octahedral symmetry has order 48 and is best understood as the symmetry of a cube (an octahedron and a cube are dual polyhedra — vertices of one correspond to face-centers of the other — and share the same symmetry group). The operations: E; 8 C₃ (about the 4 body diagonals of the cube, i.e., the 4 three-fold axes through opposite triangular faces of the octahedron — 4 axes × 2 = 8); 6 C₄ and 3 C₂ about the same 3 axes through opposite vertices of the octahedron (C₄ and C₄³ give 6 operations, and C₄² = C₂ gives the additional 3, one per axis); 6 C₂′ about axes through the midpoints of opposite edges (6 such axes, only C2 itself nontrivial, giving 6); i (the inversion center, present because opposite vertices/ligands are related by inversion through the metal); 6 S₄ (about the same axes as C₄, since S₄ and C₄ are distinct classes); 8 S₆ (about the same axes as C₃); 3σh (planes containing 4 ligands each, perpendicular to a C₄ axis); 6σd (diagonal planes). Total 1+8+6+3+6+1+6+8+3+6 = 48. Octahedral symmetry does have both an inversion center and mirror planes of both types — a much "richer" group than Td despite having the same underlying idea (a highly symmetric polyhedron of ligands).

Icosahedral symmetry (Ih). The icosahedron (20 triangular faces, 12 vertices, 30 edges) and its dual, the dodecahedron (12 pentagonal faces, 20 vertices, 30 edges), share point group Ih of order 120. C₆₀ is the truncated icosahedron — imagine slicing off each of the 12 vertices of an icosahedron to create 12 new pentagonal faces alongside the original 20 (now hexagonal) faces, giving the familiar soccer-ball pattern of 12 pentagons and 20 hexagons, 60 vertices (one carbon at each), 90 edges. Ih contains 6 C₅ axes (through opposite pairs of pentagonal faces, or in C₆₀ through opposite pairs of the 12 pentagons), 10 C₃ axes (through opposite pairs of the 20 hexagonal faces), 15 C₂ axes (through opposite pairs of edges), an inversion center, 15 σ planes, and corresponding S₁₀ and S₆ improper axes. The full operation count of 120 is what makes Ih the largest point-symmetry group encountered in ordinary chemistry.

The unifying quantitative fact behind every high-symmetry point group is that the order of the group (total number of symmetry operations, h) is fixed by the polyhedron's own rotational structure, and this order in turn controls the dimensions of irreducible representations you will meet when reading the character table.

where: $h$ is the order of the group (total operation count), the sum runs over conjugacy classes $i$, $n_i$ is the number of operations in class $i$, and $g_i$ is a formal weight of 1 (this is just the statement that $h$ equals the sum of the sizes of all conjugacy classes).

where: $h$ is again the group order, the sum runs over the group's irreducible representations $\Gamma$, and $\ell_\Gamma$ is the dimension (1, 2, or 3) of each irreducible representation. This is the sum-of-squares rule, and it is the single most useful consistency check when you sit down to read or reconstruct a character table for Td, Oh, or Ih.

where: $E$ is the number of edges of the polyhedron and $F$ the number of faces — this reflects the general fact that the rotational subgroup order of a regular polyhedron equals twice its edge count (each edge sits on a symmetry-related boundary between two rotation-equivalent face orientations), and the full point group (including reflections) typically doubles this again when an inversion center or extra mirrors are present.

Reading the equations. The sum-of-squares rule is not a curiosity — it is a hard constraint. For Td (h = 24) the irreducible representations are A₁, A₂ (both 1-dimensional), E (2-dimensional), T₁, T₂ (each 3-dimensional): checking, $1^2+1^2+2^2+3^2+3^2 = 1+1+4+9+9 = 24$. For Oh (h = 48) you get A1g, A2g, Eg, T1g, T2g, A1u, A2u, Eu, T1u, T2u — ten irreps, and $2(1+1+4+9+9) = 48$, i.e., the Oh irreps come in gerade/ungerade pairs because Oh = Td × Ci in a formal sense (Oh contains inversion, so every Td-like irrep doubles). For Ih (h = 120) the irreps are Ag, T1g, T2g, Gg, Hg (dimensions 1,3,3,4,5) plus their ungerade partners, and $2(1+9+9+16+25) = 2(60) = 120$ — the appearance of 4- and 5-dimensional irreducible representations (labeled G and H) is a signature that you have reached icosahedral symmetry, since no lower-symmetry point group needs representations that large.

Example 1 (foundational) — classify OCS and compare with CO₂. OCS (carbonyl sulfide) is linear: O=C=S, with bond lengths d(C–O) ≈ 1.16 Å and d(C–S) ≈ 1.56 Å, deliberately unequal to make the point starkly. Ask the single distinguishing question: is there a center of inversion? Inversion through the central carbon would have to send O to the position currently occupied by S and vice versa, but these are different atoms at different distances (1.16 Å vs. 1.56 Å) — inversion is not a symmetry operation. Therefore OCS is C∞v, with operations E, C∞(all angles), and an infinite set of σv planes containing the molecular axis; there is no σh, no i, no S∞. Contrast CO₂: O=C=O, both C–O distances equal at 1.16 Å by symmetry, so inversion through carbon does map the molecule onto itself (swap the two identical oxygens). CO₂ is D∞h, gaining i, σh, S∞, and an infinite family of C₂ axes perpendicular to the main axis, on top of everything C∞v has. The lesson: it is not "does the molecule look symmetric" but "are the two halves chemically identical" that decides C∞v vs. D∞h — and this directly controls infrared activity, since D∞h's inversion center forces the mutual exclusion rule (no vibration is both IR- and Raman-active), while C∞v has no such restriction.

Example 2 (applied) — verify that SiCl₄, an ideal AX₄ tetrahedron, really has 24 distinct operations, and use this to predict its vibrational spectrum pattern. SiCl₄ has Si at the center with four Cl at tetrahedral vertices, Si–Cl bond length 2.02 Å, angle Cl–Si–Cl = 109.47° exactly (the ideal tetrahedral angle, $\arccos(-1/3)$). Confirm Td applies: all four ligands identical (rules out lower symmetry like C3v or C2v, which would arise if one Cl were replaced by, say, Br), and the angle is the exact tetrahedral value (a distorted AX₄ with unequal angles would only be C2v or lower). With Td confirmed, the 3N − 6 = 3(5) − 6 = 9 vibrational degrees of freedom decompose (by standard reduction using the Td character table) as

Checking dimensions: $1 + 2 + 2(3) = 1+2+6 = 9$, matching 3N−6 exactly, a necessary consistency check. Of these, A1 (the symmetric Si–Cl stretch, all four bonds stretching in phase) is Raman-active only; the two T2 modes are both IR- and Raman-active (T2 always transforms as x,y,z in Td, so it carries dipole-moment character); E is Raman-active only. This pattern — a strong, simple Raman line for A1 and IR-active bands appearing at the T2 frequencies — is the diagnostic fingerprint used experimentally to confirm that a given AX₄ species really is a symmetric tetrahedron rather than a distorted one, since any distortion away from Td would split the T2 and E modes into multiple non-degenerate components.

Example 3 (exam/competition level) — distinguish [Co(NH₃)₆]³⁺-type Oh from a hypothetical trigonally-distorted ML₆ (D3d), using operation counts, and connect to the number of expected ¹H NMR environments. Consider an idealized ML₆ complex where all six M–L distances are equal (2.10 Å, say) and all L–M–L angles between adjacent ligands are exactly 90°: this is Oh, order 48, as derived above. Now distort it by stretching along one C₃ axis (a trigonal elongation, the kind that occurs in some Jahn–Teller-active d⁹ or d⁴ octahedral complexes when the distortion happens to preserve three-fold rather than four-fold character) so that the angles between ligands along that axis and the equatorial ligands change from 90.0° to, say, 92.5°, while keeping the three-fold rotational relationship intact. This distortion destroys the three C₄ axes (a C₄ rotation would no longer map the distorted octahedron onto itself, since bond angles are no longer uniformly 90°) but preserves one C₃ axis (the distortion axis itself) and the three C₂′ axes perpendicular to it that survive as C2 in the lower symmetry, along with the inversion center (i is preserved because opposite ligands are still related by it) and a set of σd planes now numbering 3 rather than 6. The result is D3d, order 12: E, 2C₃, 3C₂, i, 2S₆, 3σd — check: $1+2+3+1+2+3 = 12$, and indeed $12 = 48/4$, consistent with the distortion breaking a symmetry element (the C4 axes and 3 σh/3σd of the extra kind) that had multiplied the group order by exactly 4. Now connect to spectroscopy: in the idealized Oh complex, all six ligands are symmetry-equivalent (related by the full 48-operation group), so a technique sensitive to ligand environment (say, ¹H NMR on NH₃ protons, or IR-active M–L stretches) reports one ligand environment. In the D3d-distorted complex, the three "axial-ish" ligands along the C₃ axis... but wait — in a trigonal (not tetragonal) distortion of an octahedron, the six ligands actually split into two sets of three (not axial/equatorial in the usual 4+2 sense, but two triangular faces related by inversion, twisted relative to each other, forming a trigonal antiprism). Applying the D3d character table's orbit-counting: the C₃ axis and the i both relate the two triangular faces to each other, so the six ligands still fall into a single symmetry-equivalent set of 6 under D3d (since the group's operations connect all six ligands transitively), and only one NMR environment is predicted for the ligand protons even after the distortion — the distinguishing experimental signature of a trigonal twist (versus a tetragonal Jahn–Teller elongation, which does produce two chemically distinct environments, 4 equatorial + 2 axial, because C₄ symmetry about a single unique axis is not transitive on all six positions). This worked example shows how operation-counting directly predicts spectroscopic multiplicity, and why not every "distortion of an octahedron" gives the same qualitative NMR pattern.

The systematic classification of molecular point groups into families like Cn, Dn, and the special cubic/icosahedral groups grew directly out of nineteenth-century crystallography, long before quantum mechanics gave chemists a reason to care about representation theory. The crucial synthesis for chemistry was made by Friedrich Hund and, independently in spectroscopy, by Eugene Wigner and Hans Bethe, who imported the mathematical theory of finite groups (already well developed by Arthur Cayley, Camille Jordan, and Ferdinand Georg Frobenius in the abstract) into physical problems: electronic structure in crystal fields, and vibrational selection rules in molecules.