Introduction to Point Groups

Point-group assignment is the first, indispensable step in almost every symmetry-based prediction used across modern chemistry. In vibrational spectroscopy, the point group of a molecule (water's $C_{2v}$, benzene's $D_{6h}$, SF₆'s $O_h$) determines exactly which of its normal modes are infrared-active, Raman-active, both, or neither — the mutual exclusion rule for centrosymmetric molecules (any point group containing $i$) is a direct, immediate consequence of a molecule's point-group assignment, and is routinely used to confirm or rule out proposed structures, for instance distinguishing a planar $D_{2h}$ isomer of N₂O₄ from a bent, lower-symmetry alternative purely from which bands appear in IR versus Raman spectra. In ligand-field and molecular-orbital theory, the point group of a coordination complex (octahedral $O_h$, tetrahedral $T_d$, square-planar $D_{4h}$, or a distorted lower-symmetry variant from Jahn–Teller effects) fixes exactly how the metal d-orbitals split and which orbitals are allowed to mix with ligand orbitals of matching symmetry — the entire predictive machinery of crystal-field splitting diagrams traces back to correctly identifying the point group first. Pharmaceutical chemistry and stereochemistry lean on point-group logic constantly: a molecule is chiral if and only if its point group contains no improper operation ($\sigma$, $i$, or $S_n$) at all, a criterion sharper and more reliable than the older, error-prone "look for a mirror plane by eye" heuristic, and essential in classifying drug enantiomers and predicting optical activity. Materials science uses point-group classification to predict which crystals can be piezoelectric, ferroelectric, or optically nonlinear — properties that are strictly forbidden by symmetry in centrosymmetric point groups (a crystal class containing $i$ cannot show a bulk piezoelectric response, for instance), so point-group screening is a first-pass filter in searching for new nonlinear-optical or ferroelectric materials. In spectroscopic instrumentation and computational chemistry software, automated point-group detection from Cartesian coordinates is a built-in preprocessing step before running vibrational analysis, NMR symmetry-equivalence assignment, or symmetry-adapted molecular orbital construction — the entire enterprise still resting on the same foundational definition given here: the complete set of operations fixing one common point. This foundational classification feeds directly into Assigning Point Groups by Flowchart, the systematic procedure for reading off a molecule's point group from its geometry, and into the deeper formal treatment in Symmetry Operations as a Mathematical Group, where the closure property demonstrated above becomes the basis for character tables and irreducible representations — tools that in turn underlie the selection-rule and orbital-splitting applications described above, all of which build on the operations first catalogued in The Complete Set of Symmetry Operations.

:::quiz[In the fac/mer-[Co(NH₃)₃Cl₃] example, what distinguishes the two isomers' point groups?]

• fac is C₁ and mer is Cs, since mer has no symmetry at all
• fac is C₃ᵥ (h=6) and mer is C₂ᵥ (h=4), because fac has a three-fold axis relating all three Cl's while mer does not ✓
• Both isomers belong to the same point group, Oₕ, since both are based on an octahedron
• fac is Oₕ and mer is Td, since both retain full octahedral symmetry

Ask the tutor: How do I use the flowchart to systematically decide a molecule's point group without missing an operation?

Every symmetry operation from the complete catalogue — the identity $E$, rotations $C_n$, reflections $\sigma$, the inversion center $i$, and improper rotations $S_n$ — has one thing in common: applied to a molecule, it produces a configuration that is physically indistinguishable from the original, atom for atom. The defining property that makes these operations belong together as a single mathematical unit is geometric: for a finite molecule, every one of its symmetry operations leaves at least one point in space completely stationary. That point may be an atom (the nitrogen in NH₃, the boron in BF₃) or it may be an empty point in space with no atom sitting there at all (the center of a square-planar PtCl₄²⁻ ion, or the midpoint of the C–C bond in ethane's staggered conformation). Whichever it is, that fixed point is the same for every operation in the set — it is the common origin about which all rotation axes pass, in which all mirror planes intersect, and at which any inversion center sits. This is precisely why the collection is called a point group: "point" because a point is held fixed, "group" because the operations satisfy the four axioms of a mathematical group (closure, associativity, identity, inverses) — a claim developed rigorously elsewhere, but whose most tangible consequence is closure: compose any two symmetry operations of the molecule, in either order, and the result is always itself one of the operations already in the set. Nothing new is ever generated. Consider NH₃ again. Its full operation set is $\{E, C_3, C_3^2, \sigma_v, \sigma_v', \sigma_v''\}$ — six operations total, with a single $C_3$ axis through N and three vertical mirror planes each containing N and one N–H bond. Apply $C_3$ then $\sigma_v$: the composite motion turns out to be identical to applying $\sigma_v''$ alone. No new eleventh operation appears; the set is finite and closed. This closed set of six operations is what chemists mean when they say "ammonia belongs to point group $C_{3v}$." Contrast this with a symmetry operation that does not fix any point: translation. Sliding an infinite crystal lattice by one unit cell along an axis maps the crystal onto itself, but every point in space moves — nothing stays fixed. Translational symmetry, and the screw axes and glide planes that combine translation with rotation or reflection, belong to the theory of space groups, used for describing the symmetry of extended periodic solids. Point groups are reserved for symmetry that fixes a point, which is exactly the situation for isolated, finite molecules (and, locally, for the immediate coordination environment around an atom in a crystal, which is why point-group ideas remain essential even in solid-state chemistry). This restriction to "at least one point fixed" is not a minor technicality — it is the entire reason a finite, tabulable list of point groups exists at all, while the list of space groups is a much larger, separate catalogue (230 in three dimensions) built from combining point symmetry with translational symmetry. The practical payoff of identifying the complete set, rather than just noting "there's a $C_2$ axis" or "there's a mirror plane," is that partial information is not enough to predict molecular behavior. Two molecules can each possess a $C_2$ axis and yet belong to entirely different point groups — one might have only that axis and nothing else ($C_2$), another might have the $C_2$ axis plus two vertical mirrors ($C_{2v}$), another the $C_2$ axis plus a horizontal mirror and an inversion center ($C_{2h}$). These three point groups predict different numbers of infrared-active vibrations, different dipole-moment behavior, and different orbital symmetry labels, even though a lazy glance would tag all three as "has a $C_2$ axis." Assigning the point group means enumerating everything — every axis, every plane, the inversion center if present, every $S_n$ — and recognizing which self-consistent, named collection they constitute.

A point group is formally defined by two joint conditions on a set $G$ of symmetry operations:

where:

• $R_i$ ranges over every symmetry operation the molecule possesses ($E$, all $C_n^k$, all $\sigma$, $i$ if present, all $S_n^k$),
• $h$ is the order of the group — the total count of distinct operations,
• $P$ is the single point in space left invariant by every $R_i$ simultaneously (not just by some of them).

Reading: the order $h$ is not the number of types of symmetry element (axis, plane, etc.) but the total count of individual operations, including all distinct powers of rotations. For $C_{3v}$ (NH₃), $h = 6$: the two elements "$C_3$ axis" and "three $\sigma_v$ planes" generate six operations ($E$, $C_3$, $C_3^2$, $\sigma_v$, $\sigma_v'$, $\sigma_v''$), and $h=6$ is what enters every subsequent group-theoretic calculation (character table dimension checks, reducible representation reduction formulas). The second defining relationship is closure under the group composition law:

where:

• $R_i \circ R_j$ means "first perform $R_j$, then perform $R_i$" (operation composition, read right-to-left as with function composition),
• the result $R_k$ must already be a named member of the same set $G$ — never an operation outside it.

Reading: this is the algebraic heart of "point group." It is what separates a point group (a genuine mathematical group) from a mere unstructured list of symmetry elements you happened to notice. If composing two operations you found ever produced a "new" operation not already accounted for, your list would be incomplete — a sure sign a symmetry element was missed during assignment. A third relevant relationship, used constantly downstream, is the order–subgroup constraint: every point group of order $h$ can only contain subgroups whose order $h'$ divides $h$ exactly,

where:

• $h'$ is the order of any subset of $G$ that is itself a closed group (a "subgroup"),
• $h$ is the order of the full point group.

Reading: this is a powerful consistency check when assigning point groups. If you have tentatively identified $D_{4h}$ (order 16) but also think you see a leftover $C_5$ operation, something is wrong — 5 does not divide 16, so a $C_5$ operation cannot coexist inside a group of order 16. This single arithmetic fact catches many hand-drawing errors before they propagate into a wrong point-group label.

Example 1 — Foundational: enumerating the full point group of trans-N₂F₂ (planar, $C_{2h}$). trans-difluorodiazene, F–N=N–F in the trans, planar configuration, is a clean, low-symmetry test case — not as trivial as water, not as demanding as an octahedral complex. Draw the molecule flat on the page: N=N horizontal, one F up-and-left off the left nitrogen, the other F down-and-right off the right nitrogen (the trans arrangement, so the two F atoms are on opposite sides of the N=N axis). Locate every symmetry element systematically. Rotations. Look for any axis, other than the trivial 360° one, about which rotation reproduces the structure. There is a $C_2$ axis perpendicular to the N=N bond, passing through its midpoint, lying in the molecular plane. Rotating 180° about this axis swaps the two nitrogens with each other and swaps the two fluorines with each other, and because the arrangement is trans (F's on opposite sides), this rotation maps the molecule exactly onto itself. Check for any other rotation axis: no $C_3$, $C_4$, etc. are present — only this one $C_2$. Reflections. The molecule is planar, so the plane containing all four atoms is itself a mirror plane, $\sigma_h$ (called "horizontal" because it is perpendicular to the principal $C_2$ axis, which lies in that very plane — the convention labels $\sigma_h$ relative to the axis direction, not to "up/down" in a drawing). Check for a $\sigma_v$ plane (one that would contain the $C_2$ axis and be perpendicular to the molecular plane): reflecting through such a plane would need to leave each F where it is or swap the F's onto matching positions without violating the trans arrangement, but a vertical mirror here would map "F up-left" to "F up-right" — not matching the actual trans structure, since the real second F sits down-right. So no $\sigma_v$ exists. Inversion. Check the geometric center (midpoint of the N=N bond). Inverting the up-left F through this center sends it to a position down-and-right of the right nitrogen — exactly where the real second F sits (trans geometry is built for this). Inverting each N through the center swaps N with N. So $i$ is present. Improper rotations. An $S_2$ operation is identical to $i$ (already counted). Check $S_1$: that's just $\sigma$ (already counted, as $\sigma_h$). No independent higher $S_n$ appears. Collect the full, closed set:

This four-element set is exactly the point group $C_{2h}$. Verify closure directly: $C_2 \circ \sigma_h$ (rotate 180° about the axis in the plane, then reflect through that very plane) works out to be identical to the single operation $i$ — consistent, since $S_2 \equiv i$ and a $C_2$ axis perpendicular to a $\sigma_h$ plane always generates an $i$ this way. No fifth operation appears; the set closes on itself, confirming $C_{2h}$ with $h=4$.

Limiting-case check: if instead the two F atoms were placed on the same side (the cis isomer), the $\sigma_h$ (molecular plane) and a $C_2$ axis perpendicular to N=N within the plane would survive, but the inversion center would be destroyed (inverting "up-left F" through the center in the cis structure lands on a point where no atom sits), while now two $\sigma_v$ planes appear instead. The point group changes to $C_{2v}$, $h=4$ again but a different set of four operations — illustrating that order alone ($h=4$) does not fix the point group; the actual operations matter.

Example 2 — Applied: distinguishing fac- and mer-[Co(NH₃)₃Cl₃] by point group, relevant to isomer counting in coordination chemistry. Octahedral CoA₃B₃ complexes exist as two geometric isomers, facial (fac) and meridional (mer), and their point groups are exactly what different spectroscopic behavior (numbers of IR-active Co–Cl and Co–N stretches) is traced back to. fac isomer: the three Cl ligands occupy one triangular face of the octahedron (mutually cis to each other), and the three NH₃ occupy the opposite face. Picture the $C_3$ axis running through the centroid of the Cl₃ face, through the metal, to the centroid of the (NH₃)₃ face — rotating 120° about this axis permutes the three Cl's among themselves and the three NH₃'s among themselves, reproducing the structure. Three vertical mirror planes each contain the $C_3$ axis and one Cl–Co–N(H₃) pair (each plane bisecting the other two Cl's and other two NH₃'s pairwise). No $C_2$ axes perpendicular to $C_3$, no $\sigma_h$, no $i$. Full operation set:

mer isomer: the three Cl's occupy one edge-sharing meridian (two Cl's trans to each other, one Cl trans to an NH₃), and correspondingly for the NH₃'s. There is no $C_3$ axis at all here (the three Cl's are not related by any 3-fold rotation, since one is trans to Cl and the other two are trans to N). The only nontrivial rotation is a $C_2$ axis passing through the unique Cl (the one trans to NH₃) and the unique NH₃ (trans to Cl), lying in the plane that also contains the other Cl–Co–Cl and N–Co–N arrangement. There is one mirror plane, containing that $C_2$ axis and the two mutually-trans Cl's-and-NH₃'s arrangement — actually, careful enumeration shows the mer isomer retains a single $\sigma_v$ containing the $C_2$ axis (the plane holding the unique trans Cl–Co–N axis and bisecting the equatorial Cl–Co–N pairs). Full operation set:

Careful recount: $\{E, C_2, \sigma_v\}$ is only three elements, but a $C_2$ and one $\sigma_v$ containing it, together with $E$, indeed closes at three elements only if $C_2 \circ \sigma_v$ reproduces one of $\{E, C_2, \sigma_v\}$ — check: $C_2$ then $\sigma_v$ (with $\sigma_v$ containing the $C_2$ axis) composes to a second, different mirror plane $\sigma_v'$ perpendicular to the first one but still containing $C_2$, unless the geometry forces $\sigma_v' = \sigma_v$. For mer-[CoA₃B₃], only one true mirror survives (the equatorial ligands are not arranged symmetrically enough to generate a second one), so in fact composing $C_2$ with the single $\sigma_v$ must give back an operation in the set — and indeed here $C_2 \circ \sigma_v = \sigma_v$ is impossible unless $C_2 = E$, so the resolution is that the correct, fully closed operation set for mer is

with the second mirror plane $\sigma_v'$ being the plane perpendicular to the first, both containing the same $C_2$ axis (this is the standard $C_{2v}$ closure: two perpendicular mirrors intersecting along a $C_2$ axis always generate each other's product as the $C_2$ itself, and the full four-element set is required for closure — a single mirror plus a $C_2$ can never close on just three elements, matching Lagrange's constraint since 3 does not evenly support a mirror-times-rotation product staying inside a 3-element set).

Result: fac is $C_{3v}$ ($h=6$), mer is $C_{2v}$ ($h=4$) — different point groups, hence different selection rules; fac shows fewer distinct IR-active M–Cl stretching bands than mer, a difference routinely used to assign geometric isomers experimentally from vibrational spectra alone.

Example 3 — Exam/competition level: identify the point group of a staggered, unsymmetrically-substituted ethane derivative, 1,2-dichloro-1,2-difluoroethane in its anti, staggered conformation with (R,R) configuration at both carbons — a case where naive pattern-matching to "just like ethane" fails. Consider ClFHC–CHFCl in the staggered conformation with the two C–Cl bonds anti to each other (dihedral Cl–C–C–Cl = 180°) and with both stereocenters (R). Because both carbons carry four different substituents (H, F, Cl, and the other CHFCl group), this is a chiral molecule, and a first instinct might be "staggered ethane derivatives are like ethane, so surely there's a $C_3$ or at least a $C_2$." Check every candidate axis explicitly. A $C_2$ axis perpendicular to the C–C bond, bisecting it, would need to swap carbon 1 and carbon 2 while mapping each substituent onto a like substituent. Rotating 180° about such an axis sends the Cl on C1 to where the Cl on C2 sits (consistent), and the F on C1 to where the F on C2 sits (consistent) — so far so good — but it also must swap the H on C1 with the H on C2, and swap the local "handedness" arrangement around each carbon. Because the configuration is (R,R) — not (R,S) — the $C_2$ operation, which physically inverts the local spatial arrangement of substituents around the bisecting axis in the same way a proper rotation does (rotations preserve handedness), would need the two stereocenters to be related as mirror images of each other for the operation to work as a true symmetry element combined with reflection; but a pure $C_2$ rotation does not invert handedness, and checking coordinates explicitly (place C1 at $(-0.77,0,0)$ and C2 at $(0.77,0,0)$ Å, with substituents at tetrahedral angles) shows that a genuine 180° rotation about the perpendicular bisector maps the (R) center at C1 onto a position requiring an (R) center at C2 as well only if the dihedral and substituent placement are precisely antisymmetric in the right way — for the (R,R), anti, staggered arrangement specified, direct coordinate rotation confirms the $C_2$ axis perpendicular to C–C, bisecting it and lying along the direction that bisects the anti Cl···Cl arrangement, does map the molecule onto itself. Check for any mirror plane: a molecule with two (R) stereocenters related by a proper rotation (not by reflection) is chiral overall — no $\sigma$, no $i$, no $S_n$ can be present, because any of these would require an (R) center to be reflected into an (S) center somewhere in the molecule, and there is none here to pair with (both are R). Check for a $C_3$ along the C–C axis: a true $C_3$ would require three-fold identical substitution around each carbon, but each carbon carries three different groups (H, F, Cl) rather than three identical ones, so no rotation less than 360° about the C–C axis itself can be a symmetry operation. Full operation set:

This is the point group $C_2$ — order 2, containing only the identity and one proper rotation, no reflections, no inversion, exactly consistent with the molecule being chiral (only proper rotations present; see the general rule that any point group containing $\sigma$, $i$, or $S_n$ describes an achiral object, developed fully elsewhere). The exam-trap here is assuming "ethane-like backbone" automatically implies $C_{3v}$/$D_{3d}$-type richness; specific substitution pattern and stereochemistry collapse the group all the way down to order 2.