Classifying the Point Groups

Vibrational spectroscopy is the most immediate beneficiary: infrared and Raman selection rules are read directly off the point group via the transformation properties of $x, y, z$ (IR-active) and the quadratic functions $x^2, xy$, etc. (Raman-active) in the character table, and the mutual exclusion rule (no mode is both IR- and Raman-active) holds precisely in centrosymmetric groups like $D_{6h}$ or $D_{\infty h}$ that contain $i$. Vibrational spectroscopists routinely use the point group of a candidate structure (say, distinguishing $D_{3h}$ planar $\mathrm{NO_3^-}$ from a hypothetically pyramidalized $C_{3v}$ form) to predict how many IR bands should appear, then compare against experiment to confirm molecular shape — this is exactly how the planarity of the nitrate ion and the pyramidal shape of ammonia were confirmed independently of diffraction data. Crystal-field and ligand-field theory in inorganic and materials chemistry leans entirely on the octahedral ($O_h$) and tetrahedral ($T_d$) point groups to predict $d$-orbital splitting patterns, and the difference between $O_h$'s $e_g/t_{2g}$ splitting and $T_d$'s inverted $e/t_2$ splitting — a direct consequence of the differing group representations — governs the color, magnetism, and spin state of transition-metal complexes used in catalysts, pigments, and MRI contrast agents. Fullerene and nanomaterial chemistry depends on the $I_h$ icosahedral point group to explain the extraordinary stability and single ${}^{13}$C NMR signal of $\mathrm{C_{60}}$ (all 60 carbons are symmetry-equivalent under the 120 operations of $I_h$), while carbon nanotubes and related cage compounds are classified using $D_{nh}$ or $D_{nd}$ depending on their cap structure. Metal–organic frameworks and coordination polymers are designed around specific point-group-symmetric building blocks (often $D_{3h}$ trigonal or $O_h$ octahedral nodes) precisely because the resulting crystal symmetry — governed by Point Groups as Mathematical Groups combined with translational symmetry — determines pore geometry and gas-adsorption selectivity. Pharmaceutical chirality assessment routinely starts from the point-group chirality criterion: a drug candidate's point group must contain no $S_n$ of any order for the molecule to be optically active, and modern computational chemistry packages automatically assign point groups (via the same logical decision tree developed here) as a preprocessing step before running vibrational-frequency or NMR-shielding calculations, since exploiting symmetry can reduce a quantum chemical calculation's cost by an order equal to the point group's order.

Ask the tutor: How do I turn this catalog into a fast, reliable step-by-step procedure for assigning the point group of an unfamiliar molecule?

Every point group classification starts from the same physical question: pick the axis of highest rotational symmetry, call its order $n$, and then ask what else survives. Does a mirror plane contain that axis ($\sigma_v$)? Is there a mirror perpendicular to it ($\sigma_h$)? Are there $n$ two-fold axes perpendicular to the main axis, turning the single axis into a whole family related by rotation ($D_n$ structure)? Is there an inversion center? Is there a pure improper rotation axis $S_n$ with no accompanying reflection or $C_2$ set? Answering these questions in sequence is exactly how the standard families arise, and it is worth deriving each family's operation list explicitly rather than memorizing it as a table.

The cyclic ladder: $C_n$, $C_{nv}$, $C_{nh}$. Start with a molecule possessing only a single $C_n$ axis and nothing else — no mirrors, no perpendicular $C_2$s. Repeated application of the rotation generates the operations $E, C_n, C_n^2, \dots, C_n^{n-1}$, and no more, since composing any two rotations about the same axis gives another rotation about that axis. This is the group $C_n$, of order $n$, abelian, isomorphic to the cyclic group $\mathbb{Z}_n$. An example is $\mathrm{H_2O_2}$ frozen in a non-planar twisted conformation with only $C_2$ symmetry (order 2), or triphenylphosphine-like propeller structures with $C_3$ symmetry and no mirror planes. Add vertical mirror planes containing the $C_n$ axis and you get $C_{nv}$: the operations are $E$, the $n$ rotations $C_n^k$, and $n$ reflections $\sigma_v$ (or $\sigma_v$ and $\sigma_d$ alternating if $n$ is even), giving order $2n$. Water ($n=2$, one $C_2$ and two $\sigma_v$'s, order 4) and ammonia ($n=3$, order 6) are the textbook cases. Add instead a single mirror plane perpendicular to the $C_n$ axis and you get $C_{nh}$: operations are $E$, $C_n^k$ for $k=1,\dots,n-1$, $\sigma_h$, and the products $S_n^k = C_n^k \sigma_h$. Because $\sigma_h$ commutes with rotations about the axis it lies perpendicular to, $C_{nh}$ is always abelian, of order $2n$. Trans-1,2-dichloroethylene, planar and with a $C_2$ axis perpendicular to the molecular plane which itself serves as $\sigma_h$, is $C_{2h}$.

The dihedral ladder: $D_n$, $D_{nh}$, $D_{nd}$. Now suppose the molecule has, in addition to the principal $C_n$ axis, exactly $n$ two-fold axes perpendicular to it (each related to the next by the $C_n$ rotation). This doubles the order to $2n$ and produces $D_n$: propeller-shaped molecules like a $D_3$-symmetric tris-chelate metal complex with no mirror planes at all (this is the paradigm case of chirality in coordination chemistry). Add a $\sigma_h$ on top of the $D_n$ skeleton and the group becomes $D_{nh}$, order $4n$, gaining $n$ vertical mirrors $\sigma_v$ automatically as products of $C_2'$ and $\sigma_h$, plus $S_n$ operations. Benzene, planar and highly symmetric, is $D_{6h}$ of order 24. Instead of $\sigma_h$, add $n$ dihedral mirrors $\sigma_d$ that bisect the angles between adjacent perpendicular $C_2$ axes, and you get $D_{nd}$, also order $4n$, but now containing an $S_{2n}$ axis instead of $S_n$. Staggered ethane is the classic $D_{3d}$ example (order 12), and allene is $D_{2d}$ (order 8).

Pure improper rotation groups $S_n$. A separate, easily missed family consists of molecules whose only nontrivial symmetry element is a single improper rotation axis $S_n$ (with $n$ necessarily even for a genuinely new group, since odd $S_n$ always implies $\sigma_h$ and $C_n$ separately, collapsing into $C_{nh}$). The group $S_n$ for even $n$ has order $n$ and contains $E, S_n, C_{n/2}, S_n^3, \dots$ — a mix of rotations and improper rotations generated by the single operation $S_n$. A tetraphenyl-substituted allene-like skeleton or certain spiro compounds realize $S_4$ symmetry (order 4): this group is easy to miss because it looks deceptively like it should be $C_4$ or $D_{2d}$ unless you check carefully for the improper axis alone.

The exceptional high-symmetry groups. When multiple $C_n$ axes with $n \geq 3$ coexist, related by the symmetry of a Platonic solid, the cyclic/dihedral bookkeeping is superseded by three special groups: $T_d$ (tetrahedral, order 24, methane), $O_h$ (octahedral, order 48, $\mathrm{SF_6}$), and $I_h$ (icosahedral, order 120, $\mathrm{C_{60}}$). These contain multiple $C_3$, $C_4$ or $C_5$, and $C_2$ axes intersecting at the center, plus mirror planes and (for $O_h$, $I_h$) an inversion center.

Linear molecules. Finally, linear molecules have a $C_\infty$ axis (any rotation angle is a symmetry operation) and an infinite number of $\sigma_v$ planes containing that axis. Without a center of symmetry (heteronuclear diatomics like CO, or asymmetric linear molecules like HCN) the group is $C_{\infty v}$. With a center of symmetry (homonuclear diatomics like $\mathrm{N_2}$, or symmetric linear molecules like $\mathrm{CO_2}$) it is $D_{\infty h}$, which additionally contains an infinite set of perpendicular $C_2$ axes, $\sigma_h$, and $i$.

The order of a finite point group $|G|$ equals the total count of distinct symmetry operations, and this number is fixed once you know the family and the value of $n$:

where:

• $n$ is the order of the principal proper rotation axis,
• $|G|$ counts $E$ plus all rotations, reflections, improper rotations, and inversion present,
• the doubling from $C_{nv}/C_{nh}$ to $D_{nh}/D_{nd}$ reflects the extra factor of 2 from the $n$ perpendicular $C_2$ axes.

Reading it: each step up the ladder (adding $\sigma$'s, then perpendicular $C_2$'s, then both) multiplies the order by 2, because each new type of operation is generated by combining the old ones with one new generator, and a finite group generated by adding one involution to a group of order $m$ has order at most $2m$ — here the bound is always saturated. The improper rotation operation itself is defined as

where:

• $\sigma_h$ is reflection through the plane perpendicular to the $C_n$ axis,
• $C_n$ is rotation by $2\pi/n$ about that axis,
• the two factors commute because $\sigma_h$ acts only on the coordinate along the axis while $C_n$ acts only in the perpendicular plane.

Reading it: $S_n$ is "rotate then reflect," and this single composite operation is what turns a mundane rotation axis into the seed of an entire independent point-group family ($S_4$, $S_6$, $S_8, \dots$) when it exists without $C_n$ and $\sigma_h$ separately being symmetry elements in their own right. The polarity and chirality criteria, essential companions to classification, are:

where:

• polarity requires the dipole vector along the principal axis to be unmoved by every operation, which fails the instant a $\sigma_h$, $i$, or perpendicular $C_2$ would flip it,
• chirality requires the molecule to be non-superimposable on its mirror image, which is guaranteed precisely when no improper operation of any order is present.

Reading it: only $C_n$, $C_{nv}$, and $C_s$ can be polar; only $C_n$ and $D_n$ (the purely-rotational groups, with no $\sigma$ or $i$ at all) can be chiral. This is why chiral tris-chelate complexes are always $D_3$, never $D_{3h}$ or $D_{3d}$ — those higher groups carry mirror planes that would destroy chirality.

Example 1 (foundational): boric acid fragment analog — planar $\mathrm{BF_3}$. $\mathrm{BF_3}$ is planar and trigonal, with boron at the center and three fluorines at $120°$ in the molecular plane. Identify the principal axis: a $C_3$ axis perpendicular to the plane through boron, since rotating by $120°$ permutes the three (identical) fluorines cyclically and reproduces the molecule. Check for perpendicular $C_2$ axes: yes — each B–F bond axis is itself a $C_2$ axis, since rotating $180°$ about a B–F bond swaps the other two fluorines with each other while fixing the one along the axis, reproducing the molecule. There are three such axes, one per B–F bond, consistent with $D_3$ structure so far, giving a provisional order of $2 \times 3 = 6$. Check for $\sigma_h$: the molecular plane itself, perpendicular to the $C_3$ axis, is a mirror plane (every atom lies in it, so it is trivially a symmetry element). This confirms $D_{3h}$, not merely $D_3$. Apply the order formula derived above:

List the operations explicitly to confirm: $E$, $C_3$, $C_3^2$ (2 operations from the $C_3$ axis), $3\,C_2'$ (the three in-plane two-folds), $\sigma_h$ (the molecular plane), $S_3$, $S_3^5$ (improper rotations, note $S_3^2 = C_3$ so only the genuinely new powers count, giving 2 operations), and $3\,\sigma_v$ (vertical planes each containing the $C_3$ axis and one B–F bond). Total:

matching the formula exactly. $\mathrm{BF_3}$ is $D_{3h}$, order 12 — the same group as the planar $\mathrm{NO_3^-}$ and $\mathrm{CO_3^{2-}}$ ions, and structurally the trigonal-planar archetype.

Example 2 (applied): distinguishing $\mathrm{XeF_4}$ ($D_{4h}$) from a hypothetical puckered analog ($C_{4v}$). Square-planar $\mathrm{XeF_4}$ has Xe at the center with four F atoms at the corners of a square, all coplanar. The principal axis is $C_4$ perpendicular to the plane through Xe. Rotating by $90°$ cycles the four fluorines and reproduces the molecule; $C_4^2 = C_2$, $C_4^3$ complete the cyclic rotation subgroup. Perpendicular $C_2$ axes: each F–Xe–F diagonal is a $C_2$ axis (2 of them, through opposite F pairs) and each Xe–F "edge bisector" direction is also a $C_2$ axis (2 more, bisecting adjacent F–F edges) — four perpendicular $C_2$ axes total, matching $n=4$. $\sigma_h$: yes, the molecular plane containing Xe and all four F atoms is a mirror. This immediately promotes the group beyond $D_4$ to $D_{4h}$, with order

Now suppose instead the same $\mathrm{XeF_4}$ skeleton were forced into a shallow umbrella pucker (a hypothetical, non-planar distortion with Xe displaced out of the F4 plane, as sometimes invoked in vibrational analysis of out-of-plane bending modes). The $C_4$ axis survives (rotating $90°$ about the axis through Xe perpendicular to the mean F4 plane still cycles the fluorines). But now:

• $\sigma_h$ is destroyed, because reflecting through the (former) F4 plane would move Xe to the opposite side, not reproduce the structure.
• The perpendicular $C_2$ axes through F–Xe–F diagonals are destroyed too, because rotating $180°$ about an in-plane axis would flip Xe to the other side of the F4 plane, not leave it fixed (Xe is off that axis' plane if puckered asymmetrically) — more carefully, for a symmetric umbrella pucker where Xe sits exactly on the $C_4$ axis, the diagonal $C_2$'s actually survive since they pass through Xe and swap F pairs while Xe stays put on the axis; but the vertical mirror planes containing the $C_4$ axis and a pair of trans F atoms do survive, since Xe's displacement is along the axis itself, invariant under any reflection containing that axis.

Working through this carefully: with Xe on the axis but displaced along it, the operations present are $E$, $C_4$, $C_4^2$, $C_4^3$, and $4\,\sigma_v$ (each containing the $C_4$ axis and passing through either a pair of trans-F atoms or bisecting F–F edges) — but not $\sigma_h$ and not the perpendicular $C_2$'s, because those would require reflecting or rotating Xe to the "wrong side," which the pucker forbids. This gives

The comparison is instructive: identical atomic connectivity and an identical $C_4$ axis, but the presence or absence of just $\sigma_h$ and the perpendicular $C_2$'s cuts the group order in half, from 16 down to 8, and — crucially for spectroscopy — converts a nonpolar molecule ($D_{4h}$, no dipole allowed along the axis since $\sigma_h$ would have to leave any axial dipole invariant, which it cannot unless it's zero) into a polar one ($C_{4v}$, dipole along $C_4$ is fully allowed since nothing forbids it).

Example 3 (competition-level): classifying a $D_{3d}$ vs $D_{3h}$ ambiguity in a bridged biphenyl-type cage, with explicit operation count. Consider a hypothetical triptycene-like cage built from three identical bridging "blade" units connecting two bridgehead atoms, arranged in a paddlewheel with three-fold symmetry, where the two bridgeheads are chemically identical (so the molecule has an "up-down" symmetry too) but the blades are twisted by some torsion angle $\theta$ relative to a fully eclipsed reference, mimicking the ethane-type staggering problem.

Case A: $\theta = 0°$ (eclipsed blades, perfectly aligned top and bottom). The principal axis is $C_3$, along the bridgehead–bridgehead line. Since the blades are eclipsed, there is also a $\sigma_h$ perpendicular to the $C_3$ axis exactly halfway between the two bridgeheads (reflecting top blade onto bottom blade, atom for atom), provided the bridgeheads are identical. There are also three $C_2$ axes perpendicular to the main axis, each passing through the midpoint of one blade, swapping the two bridgeheads. This is the full $D_{3h}$ recipe:

Case B: $\theta = 60°$ (fully staggered blades, top rotated $60°$ relative to bottom — the ethane-like limit). Now $\sigma_h$ is destroyed: reflecting top onto bottom no longer superimposes the twisted blades since they are no longer eclipsed (a top blade now points into the gap between two bottom blades). But a dihedral mirror plane $\sigma_d$ survives instead — the plane that bisects the angle between one bottom blade and the nearest top blade, containing the $C_3$ axis, does map the structure onto itself, because the $60°$ stagger makes top and bottom blades related by exactly this reflection combined with the $C_3$ axis geometry. The perpendicular $C_2$ axes (through blade midpoints, swapping bridgeheads) still survive, since a $180°$ rotation swaps top-for-bottom and each blade maps onto the correspondingly staggered blade at the opposite bridgehead — this is a standard feature of staggered $D_{nd}$ structures. This is $D_{3d}$:

Same order, same value of $n=3$, but a genuinely different group — the character tables differ (in particular $D_{3d}$ contains $S_6$ and $i$, since $S_6^3 = i$ when $n=6$ in the improper axis of $D_{3d}$, whereas $D_{3h}$ contains $S_3$ and no inversion center at all). This is exactly the eclipsed-versus-staggered ethane logic ($D_{3h}$ for eclipsed ethane, $D_{3d}$ for staggered ethane), generalized to a bridged cage, and it demonstrates that "order 12" alone never determines the point group — the specific mirror-plane type ($\sigma_h$ vs $\sigma_d$) and the presence or absence of $i$ must always be checked directly.

Limiting case check: as $\theta \to 0°$ continuously from the staggered geometry, the dihedral mirror smoothly rotates into coincidence with becoming a proper $\sigma_h$ at $\theta = 0°$ exactly, and the group formally jumps from $D_{3d}$ to $D_{3h}$ — a genuine discontinuity in symmetry classification even though the geometry itself varies continuously, which is the standard and correct behavior for point-group assignment: symmetry groups do not interpolate, they jump at special (measure-zero) configurations.

The systematic classification of molecular point groups did not spring from chemistry at all — it descends from nineteenth-century crystallography, where mathematicians and mineralogists worked out the finite groups of rotations and reflections that could leave a crystal lattice invariant. The crucial figure bridging that mathematical crystallography into a language chemists could use for individual molecules and their vibrational spectra was Hans Bethe, whose 1929 paper on crystal-field splitting first applied group-theoretic character tables to the electronic states of ions in symmetric environments, and, even more directly relevant here, the physical chemist Robert S. Mulliken, who in the 1930s established the now-universal labeling convention ($A_1$, $A_2$, $B_1$, $B_2$, $E$, $T$, and so on) for irreducible representations of molecular point groups — a notation so standard that every modern character table still bears his symbols. Mulliken's genius was recognizing that the crystallographic point-group classification, developed for periodic lattices by mathematicians like Auguste Bravais and later formalized by Arthur Schoenflies (whose notation — $C_n$, $D_n$, $T_d$, $O_h$ — is the very notation used throughout this article), could be lifted wholesale and applied to individual, non-periodic molecules. Schoenflies notation, introduced originally for the 32 crystallographic point groups relevant to periodic crystals, turned out to also cleanly describe the finite symmetry groups of free molecules (with the added freedom of $C_5$, $C_7$, icosahedral $I_h$, and other axes forbidden in periodic crystals by the crystallographic restriction theorem, but perfectly allowed for isolated molecules like ferrocene or $\mathrm{C_{60}}$).