Point Groups as Mathematical Groups

Modern vibrational spectroscopy — infrared and Raman analysis used routinely in pharmaceutical quality control, polymer characterization, and mineralogy — depends on the fact that a molecule's normal modes transform as irreducible representations of its point group, a classification that only makes sense because the operation set is a genuine group. Catalysis researchers designing chiral ligands for asymmetric hydrogenation (industrially important for synthesizing single-enantiomer pharmaceuticals) use point-group arguments to predict which ligand symmetries ($C_2$, $C_1$, or higher) are compatible with enantioselective binding pockets. Crystallographers solving structures by X-ray diffraction assign space groups built from the same 32 crystallographic point groups first tabulated by Schoenflies and Fedorov, and every space-group table used by modern crystallographic software silently relies on the closure and inverse axioms holding for the constituent operations. In computational chemistry, quantum chemistry packages (Gaussian, ORCA, Q-Chem) automatically detect a molecule's point group at the start of a calculation and use its multiplication table to block-diagonalize the Hamiltonian and Fock matrices by symmetry, dramatically speeding up electronic structure calculations on symmetric molecules like buckminsterfullerene ($I_h$, order 120) or benzene ($D_{6h}$, order 24) — savings that scale directly with how large and non-abelian the point group is. Materials scientists studying phonon modes and photonic band structure in symmetric nanostructures, and NMR spectroscopists predicting equivalent nuclei via symmetry-related permutations, all lean on the same closure-associativity-identity-inverse skeleton established here.

Ask the tutor: Can you walk me through why $C_3\sigma_v \ne \sigma_v C_3$ in $C_{3v}$ using the actual hydrogen-label permutations?

Every symmetry operation of a molecule is a rigid transformation of three-dimensional space — a rotation, reflection, inversion, or improper rotation — chosen so that after applying it, every atom lands on a position previously occupied by an atom of the same kind. The nuclear framework looks unchanged, even though individual atoms may have permuted among equivalent sites. Because these operations are built from rotations and reflections, they preserve all interatomic distances and angles. This is the crucial geometric fact underlying everything else: a rigid motion of a rigid motion is still a rigid motion. Consider ammonia, NH₃, a trigonal pyramid with $C_{3v}$ symmetry. Its operations are $E$ (do nothing), $C_3$ and $C_3^2$ (rotations by 120° and 240° about the principal axis), and three vertical mirror planes $\sigma_v$, $\sigma_v'$, $\sigma_v''$, each containing the axis and one N–H bond. That is six operations total. Suppose you first reflect through $\sigma_v$ and then rotate by $C_3$. Geometrically, "reflect then rotate" is some new rigid motion that fixes the central nitrogen and permutes the three hydrogens among themselves. Because the only rigid motions doing that (consistent with the pyramidal shape) are exactly the six operations listed above, the product $C_3 \sigma_v$ must equal one of them. It cannot be some seventh, unlisted transformation, because there is no seventh symmetry operation left to be — the list of six is already the complete symmetry-preserving set for a triangular arrangement of three identical H atoms around the axis. This is closure in action, and you can watch it happen by literally rotating a paper model. Group theory adds one more layer of structure: it insists that composition be checked as a genuine algebraic operation, with associativity, an identity, and inverses, so that the set together with composition satisfies the same abstract axioms as, say, the integers under addition (only finite, and non-commutative in general). Once verified, an enormous machinery becomes available — the group's character table, irreducible representations, and reduction formulas — because these are theorems about any group satisfying the axioms, not special facts about NH₃.

The multiplication table below is built by physically performing each pair of operations in sequence and reading off which single operation reproduces the result. The convention, matching matrix multiplication, is that in the product $\hat{A}\hat{B}$, operation $\hat{B}$ is performed first, then $\hat{A}$.

Notice the asymmetry across the diagonal: $C_3 \sigma_v = \sigma_v''$ but $\sigma_v C_3 = \sigma_v'$ (read the table with row-first, column-second convention above). This confirms $C_{3v}$ is non-abelian — order of operations matters, exactly as it must for a group describing rotations and reflections in three dimensions.

A set $G = \{\hat{E}, \hat{A}, \hat{B}, \hat{C}, \dots\}$ of symmetry operations, together with the rule "perform one operation, then another," is a group if and only if four conditions hold.

Closure. For every pair of operations in the set, their product is also in the set:

where: $\hat{A}\hat{B}$ means "apply $\hat{B}$ first, then $\hat{A}$," and both operations, and their composite, are symmetry operations of the same molecule. Reading: no sequence of symmetry operations can ever produce a "new" transformation outside the original list — the set is self-contained under composition.

Associativity. For any three operations,

where: grouping of three sequential operations does not affect the final result, only the order left-to-right matters. Reading: this holds automatically because all symmetry operations are represented by linear (matrix) transformations acting on coordinates, and matrix multiplication is associative — you never need to check this by hand for a molecule.

Identity. There exists an element $\hat{E} \in G$ such that

where: $\hat{E}$ is the "do nothing" operation, always a symmetry operation of any molecule. Reading: every point group contains the identity; it is the operation that trivially leaves every atom exactly where it started.

Inverse. For every $\hat{A} \in G$ there exists $\hat{A}^{-1} \in G$ such that

where: $\hat{A}^{-1}$ is the operation that exactly undoes $\hat{A}$. Reading: rotations are undone by rotating the opposite way ($C_n^{-1} = C_n^{n-1}$), reflections and $C_2$ rotations and inversion are each their own inverse ($\sigma^2 = E$, $C_2^2=E$, $i^2=E$), and improper rotations $S_n$ have inverses built from $S_n^{n-1}$ or $S_n^{2n-1}$ depending on parity.

Order of the group. The number of elements in $G$ is called the order, denoted $h$. For $C_{3v}$, $h = 6$; for $C_{2v}$, $h=4$; for $D_{6h}$ (benzene), $h = 24$. A further structural fact, essential for later work with character tables, is that operations split into classes: elements $\hat{B}$ and $\hat{B}'$ are conjugate (in the same class) if there exists some $\hat{A}$ in the group with

where: conjugation by $\hat{A}$ represents "the same kind of operation viewed from a rotated/reflected perspective." In $C_{3v}$, the two mirror-related rotations $C_3$ and $C_3^2$ form one class, and the three vertical mirrors $\sigma_v,\sigma_v',\sigma_v''$ form another, because each type of operation can be transformed into another member of its own class by conjugation with a group element. This is why $C_{3v}$'s character table has only 3 classes ($E$, $2C_3$, $3\sigma_v$) rather than 6 separate columns.

Example 1 — Foundational: verifying the inverse axiom for $S_4$ in point group $S_4$ (e.g., idealized 1,3,5,7-tetrasubstituted cyclooctatetraene skeleton). The group $S_4 = \{E, S_4, C_2, S_4^3\}$ has order $h=4$ (here $C_2 = S_4^2$, since two successive $S_4$ operations — each a 90° rotation combined with a reflection — combine to a pure 180° rotation, because the two reflections cancel). Verify that every element has an inverse within the set. Take $\hat{A} = S_4$ (rotate 90° about $z$, then reflect through $xy$). Its inverse must satisfy $S_4 \hat{A}^{-1} = E$. Try $\hat{A}^{-1} = S_4^3$ (three successive applications of $S_4$, i.e., rotate 270° then reflect three times, net effect: rotate $270°$ and reflect once since reflections are involutions and $\sigma_h^3=\sigma_h$):

Four successive $S_4$ operations means four 90° rotations (total $360°=0°$) and four reflections through $\sigma_h$ (an even number, so they cancel: $\sigma_h^4 = E$). Hence

confirming $S_4^{-1} = S_4^3$, and since $S_4^3$ is already listed as a member of the group, the inverse axiom holds for this element. Checking $C_2$: $C_2 \cdot C_2 = C_2^2 = C_4^4 = E$ (a 360° rotation), so $C_2$ is its own inverse, again found inside the set. Every element's inverse is present — the inverse axiom is satisfied, and no operation is "missing" from the table, which is exactly what closure guarantees in advance.

Example 2 — Applied: using the $C_{3v}$ table to predict a product relevant to NH₃ vibrational analysis without recomputing coordinates. Suppose in an infrared analysis of ammonia you need $\sigma_v' \cdot C_3^2$ — the operation equivalent to first rotating by 240° then reflecting through $\sigma_v'$ — because this composite appears when reducing the basis of N–H bond-stretch displacement vectors into irreducible representations (a step that eventually leads into Matrix Representations of Symmetry Operations). Rather than re-deriving this by direct coordinate geometry, read the multiplication table constructed earlier (row $= \sigma_v'$, column $=C_3^2$): From the Cayley table, row $\sigma_v'$, column $C_3^2$ gives $\sigma_v''$:

This means the composite transformation is physically identical to a single reflection through the third mirror plane $\sigma_v''$ — a fact you can double check by tracking where H₁, H₂, H₃ go under each step: $C_3^2$ sends (H₁,H₂,H₃)→(H₂,H₃,H₁) [rotating by 240°, equivalently −120°, cycles labels backward by one], then $\sigma_v'$ (the mirror through H₂) fixes H₂ and swaps H₁↔H₃, giving net permutation (H₁,H₂,H₃)→(H₃,H₂,H₁)→ wait, composing carefully: apply $C_3^2$ first to get (H₂,H₃,H₁) in the H₁,H₂,H₃ slots, then apply $\sigma_v'$ which fixes slot-2 and swaps slots 1,3, giving (H₁,H₃,H₂) read in original slot order — which is exactly the permutation produced by a single reflection through the plane containing H₃, i.e. $\sigma_v''$, matching the table exactly. Because this product was already guaranteed by closure to equal some member of $\{E,C_3,C_3^2,\sigma_v,\sigma_v',\sigma_v''\}$, the table lookup is trustworthy shorthand for a full geometric recomputation.

Example 3 — Competition-level: proving a general theorem — every group of order $h$ that is prime must be cyclic, and identifying which point groups this excludes. This is a pure-group-theory result (Lagrange's theorem corollary) with direct chemical payoff: it tells you, sight unseen, the multiplication-table structure of any point group whose order is a prime number, such as $C_3$ (order 3, e.g., triphenylphosphine idealized without the phenyl twist) or $C_5$ (order 5).

Claim. If $|G| = p$ for prime $p$, then $G$ is cyclic — generated entirely by repeated application of a single operation — and therefore abelian.

Proof sketch relevant to point groups. By Lagrange's theorem, the order of any subgroup of $G$ must divide $|G|=p$. Since $p$ is prime, its only divisors are 1 and $p$ itself. Take any non-identity element $\hat{A}\in G$ and consider the cyclic subgroup it generates, $\langle \hat{A}\rangle = \{E, \hat{A}, \hat{A}^2, \dots\}$. This subgroup has order dividing $p$; since $\hat{A}\ne E$, its order is not 1, so it must be exactly $p$. But a subgroup of order $p$ inside a group of order $p$ is the whole group. Hence $G = \langle \hat{A}\rangle$: every element of $G$ is some power of $\hat{A}$, and powers of a single element always commute ($\hat{A}^m\hat{A}^n = \hat{A}^{m+n} = \hat{A}^n\hat{A}^m$), so $G$ is abelian.

Chemical consequence. Point group $C_n$ for prime $n$ (e.g. $C_3$, $C_5$, $C_7$) is automatically cyclic and abelian — its entire multiplication table is generated from just $C_n, C_n^2, \dots, C_n^{n-1}, E$ with no independent reflections or rotations to introduce non-commutativity. This immediately tells you, without constructing a single Cayley table entry, that a molecule belonging to pure $C_3$ (chiral, propeller-shaped, no mirror planes — e.g., idealized triarylphosphine with a helical twist) has only one-dimensional irreducible representations that come in complex-conjugate pairs, since abelian groups have $h$ distinct 1-D irreducible representations (a fact that becomes central once you reach Classifying the Point Groups and construct that molecule's character table). By contrast, $C_{3v}$ (order 6, not prime) is not forced to be abelian by this theorem — and indeed the Cayley table above showed explicitly that it is not, since $6=2\times3$ is composite and admits the non-abelian structure built from a normal cyclic subgroup $\{E,C_3,C_3^2\}$ together with the three reflections.