Assigning Point Groups by Flowchart

Point group assignment by flowchart is not a museum exercise; it is the first move in nearly every symmetry-based analysis a working chemist or spectroscopist performs. In vibrational spectroscopy, the point group of a molecule determines exactly how many IR-active and Raman-active fundamental modes are expected, via the reducible representation of the $3N$ Cartesian displacements decomposed against the character table of that specific group — get the point group wrong, and every subsequent selection-rule prediction is wrong too. This is routine in identifying metal-carbonyl isomers: $\text{Fe(CO)}_5$'s trigonal-bipyramidal $D_{3h}$ structure predicts two IR-active C–O stretches, a number directly checked against experimental infrared spectra to confirm the geometry rather than, say, a square-pyramidal alternative. In transition-metal and organometallic chemistry, sandwich compounds like ferrocene, staggered ($D_{5d}$) versus eclipsed ($D_{5h}$), are distinguished exactly along the lines of the third worked example, with real energetic and reactivity consequences — the rotational barrier between conformers, small as it is, still shows up in variable-temperature NMR and in the fine details of photoelectron spectra. In main-group and coordination chemistry, correctly placing five-, six-, and seven-coordinate complexes into $D_{3h}$, $D_{4h}$/$O_h$, or $C_{2v}$ frames underlies the construction of ligand-field splitting diagrams, since the pattern of $d$-orbital splitting is read directly from how the metal $d$ orbitals transform under that specific point group's operations. Beyond spectroscopy, point-group logic feeds directly into crystallography and materials science (the 32 crystallographic point groups constrain which space groups a material can adopt, in turn constraining piezoelectric and ferroelectric behavior — a material lacking a center of inversion, checkable by exactly this flowchart's early inversion/mirror questions, is a prerequisite for piezoelectricity). It also underlies computational chemistry practice: quantum chemistry packages exploit a molecule's assigned point group to block-diagonalize the Hamiltonian by symmetry, cutting computational cost for highly symmetric species like $\text{SF}_6$ ($O_h$) or benzene ($D_{6h}$) by orders of magnitude — software routines that automate exactly the flowchart described here as a first preprocessing step before any electronic structure calculation begins.

Ask the tutor: How would the flowchart classify a real, slightly puckered (non-planar) benzene-like ring — does small deviation from idealized $D_{6h}$ actually change the assigned point group?

A molecule's point group is the complete set of symmetry operations — proper rotations $C_n$, improper rotations $S_n$, mirror reflections $\sigma$, and inversion $i$ — that map the molecule onto an indistinguishable copy of itself, all operations sharing at least one fixed point. The flowchart method never tries to enumerate that whole set directly. Instead it asks a sequence of targeted questions, each of which either terminates the search (assigning a group immediately) or narrows the remaining candidates, until only one group is consistent with all the answers. The logic of the ordering is worth understanding, not just memorizing. Linear molecules and the five high-symmetry polyhedral groups ($T_d$, $O_h$, $I_h$, and their unrehearsed rotation-only cousins $T$, $O$, $I$) are structurally unlike everything else: they possess multiple $C_n$ axes with $n \geq 3$ related by the point group itself, so the "one principal axis" logic that governs every other group breaks down. It is therefore efficient to identify these by their unmistakable geometric signatures — linear atoms, or a shape built from a tetrahedron, octahedron, or icosahedron — before applying any axis-counting procedure. Once those special cases are excluded, every remaining molecule has at most one axis of highest order $n$ (there can be more than one axis of a given lower order, but only one — or one set related by that axis — of the maximal order). This is the principal axis, and the entire rest of the classification is organized around it: how many $C_2$ axes sit perpendicular to it, and what mirror planes contain or bisect it.

Two structural facts justify the order of the branches. First, checking linearity and the polyhedral special cases early is cheap and avoids the axis-counting machinery misfiring on molecules where "the" principal axis is not unique. Second, within the generic branch, the perpendicular-$C_2$ test is placed before the mirror-plane test because it is the more discriminating question: it cleanly separates the $C_n$-type groups (cyclic, only one axis "direction") from the $D_n$-type groups (dihedral, an extra set of two-fold axes perpendicular to the main one), and only after that split does asking about $\sigma_h$, $\sigma_v$, or $\sigma_d$ produce an unambiguous subscript.

The flowchart is not a set of arbitrary rules; it is a direct consequence of group-closure requirements. If a molecule has a principal axis $C_n$ and a set of $C_2$ axes perpendicular to it, the group generated by $C_n$ and one $C_2$ necessarily contains $n$ such $C_2$'s (obtained by conjugating one $C_2$ by successive powers of $C_n$), which is why the test is "are there $n$ perpendicular $C_2$'s," never "is there just one."

where:

• $C_n$ is the principal rotation by $2\pi/n$,
• $C_2$ is a two-fold axis perpendicular to $C_n$,
• $C_2^{(k)}$ is the image of that axis after conjugation, i.e. the same $C_2$ axis rotated by $k \times (360^\circ/n)$ about the principal axis.

Reading: because group elements must close under multiplication, a single perpendicular $C_2$ automatically drags along $n-1$ more, spaced evenly around the principal axis. This is why real molecules never show "$n-1$ out of $n$" perpendicular two-fold axes — if you find even one, the full set of $n$ is guaranteed, and you are in a $D_n$-type group. The second governing relationship concerns how $\sigma_h$, $\sigma_v$, and $S_n$ combine, which is what fixes the final subscript once the $C_n$ vs. $D_n$ branch is chosen:

where:

• $S_n$ is an improper rotation: rotate by $2\pi/n$, then reflect through the plane perpendicular to the axis,
• $\sigma_h$ is the "horizontal" mirror plane, perpendicular to the principal axis,
• $C_n$ is the principal proper rotation.

Reading: whenever both $C_n$ and $\sigma_h$ are present, $S_n$ is automatically present too (it is their product) — you never need to search for $S_n$ separately once you've confirmed $\sigma_h$. This is why the flowchart tests for $\sigma_h$ directly rather than hunting for improper rotations from scratch; $S_n$ falls out for free. Finally, the count of vertical planes is pinned down by the dihedral angle between adjacent mirrors:

where:

• $\theta_{\sigma\sigma}$ is the angle between two adjacent vertical mirror planes (whether $\sigma_v$ or $\sigma_d$) that both contain the principal axis,
• $n$ is the order of the principal axis.

Reading: for a $C_n$ or $D_n$ molecule with vertical mirrors, there are exactly $n$ of them (for $C_{nv}$/$D_{nh}$) evenly spaced at $180^\circ/n$, which both confirms you've found all the mirrors and lets you predict where an unfound one should be if the count looks short.

Example 1 (foundational): water, $\text{H}_2\text{O}$. Water is bent, with an $\angle$H–O–H of $104.5°$ (not $109.5°$ or $120°$ — an off-round number confirming this is not accidentally tetrahedral or trigonal geometry masquerading as bent). Step 1, linear? No — three atoms with a bond angle of $104.5°$, well short of $180°$. Step 2, cubic/icosahedral? No — only three atoms, no possibility of multiple equivalent $C_3$/$C_5$ axes. Step 3, principal axis? The $C_2$ axis bisects the H–O–H angle, passing through O. Rotating by $180°$ about this axis swaps the two H atoms and leaves O fixed — the molecule is indistinguishable after the operation. Order confirmed: $n = 2$, this is the highest-order axis present (no $C_3$ or higher is geometrically possible with only two terminal atoms). Step 4, perpendicular $C_2$'s? Look for a second two-fold axis perpendicular to the first. Any axis lying in the molecular plane but perpendicular to the principal $C_2$ would have to map H onto O or O onto H — impossible for chemically distinct atoms. No such axis exists. Branch: $C_n$ family, not $D_n$. Step 5, mirror planes. Two candidates: the plane containing all three atoms (call it $\sigma_v$, since it contains the principal axis), and the plane perpendicular to the H–H direction, also containing the principal axis and bisecting the H–O–H angle (call it $\sigma_v'$). Both contain the $C_2$ axis, so both are vertical mirrors, not $\sigma_h$ (a true $\sigma_h$ would have to be perpendicular to $C_2$, i.e., in this case the plane containing just the two H's and perpendicular to the H–O–H bisector — but that plane does not contain O and is not a symmetry element of the actual bent molecule). Step 6, subscript check via the mirror-count formula. With $n=2$, the formula $\theta_{\sigma\sigma}=180^\circ/n = 90^\circ$ predicts two vertical mirrors at $90°$ to each other — exactly the molecular plane and the perpendicular plane through O, matching what was found by inspection.

Assignment: $C_{2v}$. As a limiting-case check: if the H–O–H angle were forced open to exactly $180°$ (a hypothetical, since real water is bent), the molecule would become linear and reroute entirely into the $D_{\infty h}$ branch at Step 1 — confirming the flowchart's linearity gate is doing real work and that $C_{2v}$ is specific to the bent geometry.

Example 2 (applied): phosphorus pentachloride, $\text{PCl}_5$, trigonal bipyramidal. $\text{PCl}_5$ in the gas phase has three equatorial Cl at $120°$ to each other in a plane through P, and two axial Cl at $90°$ to the equatorial plane, on opposite sides of P (axial–P–axial $= 180°$). Step 1, linear? No — clearly a branched, non-collinear structure. Step 2, cubic/icosahedral? No — five ligands around one center is not a tetrahedron (4), octahedron (6), or icosahedron (12); no set of equivalent $C_3$/$C_5$ axes related by other symmetry operations exists here. Step 3, principal axis? The axial–P–axial line is a $C_3$ axis: rotating $120°$ about it permutes the three equatorial Cl cyclically while leaving both axial Cl fixed (since they lie exactly on the axis). Check higher order: is there a $C_4$ or $C_5$? No — five ligands split 3+2 cannot support a five-fold or four-fold rotation that maps the molecule onto itself, since the equatorial and axial sets are chemically/geometrically distinct (different P–Cl bond lengths, different angles) and cannot be permuted into each other. So $n = 3$ is the maximal order. Step 4, perpendicular $C_2$'s? Consider an axis passing through P and one equatorial Cl, lying in the equatorial plane, perpendicular to the $C_3$ axis. Rotating $180°$ about this axis: the chosen equatorial Cl stays fixed, the other two equatorial Cl's swap with each other (they are $120°$ apart, symmetric about the chosen one), and the two axial Cl's swap with each other (both perpendicular distance from this in-plane axis, related by the $180°$ flip). This is a valid symmetry operation. By the conjugation relation, since $n=3$, there must be $3$ such axes total, one through each equatorial Cl — and indeed there are, by the three-fold symmetry of the equatorial plane. Branch: $D_n$ family, with $n=3$. Step 5, mirror planes. The equatorial plane itself (containing P and all three equatorial Cl, perpendicular to the axial–P–axial $C_3$ axis) is a candidate $\sigma_h$: reflecting through it swaps the two axial Cl (mirror images across the plane) and leaves each equatorial Cl in place (they lie exactly in the plane). This reflection maps the molecule onto itself. Since $\sigma_h$ is perpendicular to $C_n$ and present, this is the deciding test.

Assignment: $D_{3h}$. Cross-check with the $S_n = \sigma_h C_n$ relation: since both $C_3$ and $\sigma_h$ are confirmed present, $S_3$ must also be a symmetry element (rotate $120°$, then reflect through the equatorial plane) — and indeed this composite operation does map $\text{PCl}_5$ onto itself, consistent with $D_{3h}$'s character table, which lists $S_3$ without it needing to be independently rediscovered. Limiting case: if the two axial Cl were somehow chemically different from each other (say one replaced by F, giving an axial-F/axial-Cl asymmetric trigonal bipyramid), the $\sigma_h$ reflection would fail (F cannot map onto Cl), collapsing the symmetry down to $C_{3v}$ — showing the flowchart is sensitive to exactly the substitution that should break the mirror.

Example 3 (exam/competition level): a staggered $M(\eta^5\text{-}C_5H_5)_2$ sandwich (staggered ferrocene-type geometry, $D_{5d}$). Consider a metallocene with two parallel cyclopentadienyl (Cp) rings, each a regular pentagon, related by a $36°$ stagger (i.e., the top ring's vertices sit directly over the bottom ring's edge midpoints, rotated by exactly half the $72°$ vertex spacing — this is the staggered, "anti-prismatic" conformation, as opposed to eclipsed ferrocene which is $D_{5h}$). Step 1, linear? No. Step 2, cubic/icosahedral? No — a sandwich with two 5-membered rings is not one of the five Platonic families; there is no set of multiple equivalent $C_3$ or $C_5$ axes unrelated to a single stacking direction (only the one $C_5$ through the metal, perpendicular to both rings, is present). Step 3, principal axis. The axis through the metal center, perpendicular to both Cp rings, is $C_5$: rotating $72°$ maps each ring onto itself (cyclic permutation of its five carbons) and, because both rings share the same rotation axis and the stagger is uniform, maps the whole sandwich onto itself. No higher-order axis is possible (rings are 5-membered, capping at $C_5$). Step 4, perpendicular $C_2$'s. Consider an axis through the metal, lying in the plane that bisects the M–ring distance, passing through one carbon of the top ring and — because of the $36°$ stagger — through the midpoint of an edge of the bottom ring (not through a carbon). Rotating $180°$ about this in-plane axis: it swaps the top and bottom rings (top ring's carbons map to bottom ring's carbons, since a $180°$ flip exchanges "up" and "down"), and because the stagger angle is exactly half of $72°$, this flip is a genuine symmetry operation — each atom lands exactly on another atom's original position. By the conjugation relation with $n=5$, there are $5$ such perpendicular $C_2$ axes (one for roughly every top-ring carbon position, spaced $72°$ apart around the $C_5$). Branch: $D_n$ family, $n=5$. Step 5, mirror planes — the subtle part. Is there a $\sigma_h$ (horizontal mirror, perpendicular to $C_5$, i.e., the plane exactly midway between the two rings)? Reflecting through this midplane would have to map the top ring onto the bottom ring directly. But the rings are staggered by $36°$: a carbon on the top ring sits over an edge-midpoint of the bottom ring, not over a carbon. A mirror reflection through the midplane preserves each atom's angular position (only flips its height), so it would try to map a top-ring carbon to an angular position where the bottom ring has no atom — this fails. No $\sigma_h$. Now check for $\sigma_d$ (vertical planes, containing the $C_5$ axis, each bisecting the angle between two adjacent perpendicular $C_2$ axes found in Step 4). Such a plane, containing the principal axis and running "between" two of the $C_2$'s, correctly reflects each carbon of the top ring to another carbon of the top ring, and independently each carbon of the bottom ring to another carbon of the bottom ring, consistent with the staggered geometry. Checking the angle: with $n=5$, adjacent mirrors should sit at $\theta_{\sigma\sigma} = 180^\circ/5 = 36^\circ$ — matching the stagger angle exactly, which is the geometric fingerprint of a $\sigma_d$-type (dihedral, bisecting) plane rather than $\sigma_v$.

Assignment: $D_{5d}$. As the limiting-case check: rotate the top ring back by $36°$ so the two rings become perfectly eclipsed (each carbon directly over a carbon). Repeating Step 5 now shows the midplane reflection does work (top ring maps onto bottom ring atom-for-atom), giving $\sigma_h$ instead, and the vertical mirrors reposition to contain the $C_2$'s directly ($\sigma_v$) rather than bisecting them — the group becomes $D_{5h}$, the eclipsed-ferrocene case, exactly the well-known contrast between real ferrocene's two nearly isoenergetic rotamers.

The Schoenflies notation that the flowchart's labels are drawn from — $C_n$, $D_n$, $S_n$, the subscripts $v$, $h$, $d$ — was developed by the German mathematician Arthur Moritz Schoenflies in the 1890s, originally to classify the 230 crystallographic space groups, a problem he solved independently and nearly simultaneously with the Russian crystallographer Evgraf Fedorov. Schoenflies's notation was built for infinite periodic lattices, but it transferred almost unchanged to finite, discrete molecular symmetry once quantum chemists in the 1930s realized that the same closure and conjugacy arguments governing crystal symmetry govern the vibrational and orbital symmetry of individual molecules. The systematic application of group theory to molecular spectroscopy and bonding — the very context in which "assign the point group first" becomes a load-bearing step — was substantially the work of Hans Bethe, Eugene Wigner, and later, in the chemistry-facing form most students encounter, textbook authors like F. Albert Cotton, whose Chemical Applications of Group Theory (1963) codified much of the flowchart logic used today.