Improper Rotations and Sn Axes

Recognizing $S_n$ axes is not an antiquarian classification exercise; it is load-bearing machinery in several live areas of chemistry. In chirality-driven drug design, the presence or absence of any $S_n$ axis is the rigorous test used (often implicitly, via software that checks all rotation-reflection symmetry elements) to determine whether a candidate molecule is chiral and therefore whether its enantiomers must be separated and independently tested for biological activity — thalidomide-type tragedies are precisely failures to appreciate that mirror-image molecules built from the same connectivity can have entirely different pharmacology. In vibrational spectroscopy, whether a normal mode is infrared-active or Raman-active is decided by symmetry selection rules built directly from the full point-group character table, which always includes the $S_n$ operations as distinct symmetry classes with their own characters; $D_{2d}$ allene's $S_4$ axis, for instance, directly controls which C–H stretching and bending modes are IR-forbidden but Raman-allowed, a rule of mutual exclusion that follows from the presence of an inversion-type element in the full operation set. In materials science and nanotechnology, improper rotation axes classify the point-group symmetry of fullerenes, dendrimers, and metal–organic cages; $\mathrm{C_{60}}$'s icosahedral $I_h$ point group contains numerous $S_{10}$ axes, and their presence is directly responsible for the vanishing of certain forbidden vibrational and electronic transitions used to fingerprint fullerene purity by IR and Raman spectroscopy. In organometallic and coordination chemistry, staggered versus eclipsed conformations of sandwich compounds like ferrocene are distinguished exactly by whether a genuine $S_{10}$ (staggered, $D_{5d}$) or $S_5$-with-$\sigma_h$ (eclipsed, $D_{5h}$) axis is present, which in turn governs selection rules in their electronic and vibrational spectra and subtly affects packing and reactivity. Finally, in crystallography and solid-state chemistry, improper rotoinversion axes (the crystallographic cousin of $S_n$, denoted $\bar n$ in Hermann–Mauguin notation) are essential to the complete classification of the 32 crystallographic point groups and 230 space groups that organize all ordered solids, connecting this molecular-scale idea directly to bulk materials science.

Ask the tutor: How do I systematically search a molecule's geometry for a hidden $S_n$ axis when it isn't obvious by inspection, and how does that search feed into assigning the full point group?

Begin with what a proper rotation $C_n$ does: it takes every point $(x,y,z)$ of a molecule and swings it through an angle $2\pi/n$ about a fixed axis, landing every atom on top of an atom of the same kind, leaving the molecule looking unchanged. This is something you can do with your hands — pick up a model, twist it, set it back down. It is called "proper" precisely because it preserves the handedness of three-dimensional space: if you built a right-handed coordinate frame glued to the molecule, after a $C_n$ rotation it is still right-handed. An improper rotation $S_n$ does something a rigid hand can never do. It consists of two steps applied in sequence to every point in the molecule: 1. Rotate by $2\pi/n$ about an axis (call it the $z$-axis) — exactly as in $C_n$. 2. Reflect every point through the plane $\sigma_h$ that is perpendicular to that same axis (i.e., flip the sign of the $z$-coordinate). Crucially, $S_n$ is defined as the product of these two steps performed as a single indivisible symmetry operation. Neither step alone need be a symmetry operation of the molecule — often neither the bare rotation $C_n$ nor the bare reflection $\sigma_h$ maps the molecule onto itself individually, and yet the combination does. This is exactly what happens in staggered ethane: rotating the back $\mathrm{CH_3}$ group by $60°$ relative to the front one is not by itself a symmetry operation (it eclipses hydrogens with nothing), and reflecting through the plane bisecting the C–C bond is not a symmetry operation either — but rotate by $60°$ and then reflect, and every hydrogen lands exactly on a hydrogen. Because reflection reverses handedness (turns a right glove into a left glove), $S_n$ is called an improper operation. A physical object and its $S_n$-transformed image are related the way your right hand is related to your left hand seen in a mirror — you cannot superpose them by rotation alone. This is the deep reason $S_n$ governs chirality: if a molecule possesses any $S_n$ symmetry element, then the molecule's mirror image can be rotated back on top of the original molecule, so the molecule is superposable on its mirror image and is achiral. If no such axis exists for any $n$, the two mirror images are genuinely different objects, and the molecule is chiral. Two special low-order cases deserve names of their own, purely for historical and notational convenience, since chemists met them before the general pattern was recognized:

• $S_1$: rotate by $2\pi/1 = 360°$ (which does nothing) then reflect through the perpendicular plane. The net effect is a pure reflection. So $S_1 = \sigma$.
• $S_2$: rotate by $2\pi/2 = 180°$ then reflect through the perpendicular plane. Rotating $(x,y,z)$ by $180°$ about $z$ gives $(-x,-y,z)$; reflecting through $xy$ then flips the $z$: $(-x,-y,-z)$. That is precisely spatial inversion through the origin. So $S_2 = i$.

This is the central unifying fact of the whole topic: mirror planes and inversion centers are not separate kinds of symmetry element — they are $S_1$ and $S_2$. Every point group symbol you will ever write ($C_{2v}$, $D_{3h}$, $T_d$, $O_h$, …) is built from only two families of geometric operation, $C_n$ and $S_n$, once you accept $\sigma = S_1$ and $i = S_2$.

Notice in the figure that the intermediate point $A'$ (after rotation only, before reflection) need not correspond to any atom of the real molecule — it is a bookkeeping fiction. Only the final point $A''$ must land on a genuine atomic position for $S_n$ to be a true symmetry operation. The second geometric subtlety, essential for later derivations, is what happens when you apply $S_n$ twice. Two rotations by $2\pi/n$ compose to a rotation by $4\pi/n$; two reflections through the same plane cancel completely ($\sigma_h^2 = E$). So:

where:

• $S_n^2$ — the operation $S_n$ applied twice in succession,
• $C_n^2$ — proper rotation by $2 \times (2\pi/n) = 4\pi/n$,
• $\sigma_h^2 = E$ — two reflections through the same plane restore every point.

This single line is the algebraic seed of nearly everything nontrivial about $S_n$ axes, worked out in full below.

The matrix representation makes the geometry completely explicit. Taking the $S_n$ axis along $z$:

where:

• the upper-left $2\times2$ block is the ordinary $C_n$ rotation matrix acting on $(x,y)$,
• the $-1$ in the bottom-right entry is the $\sigma_h$ reflection acting on $z$,
• $\theta = 2\pi/n$ is the rotation angle for one application of $S_n$,
• the two operations commute here because the reflection plane is perpendicular to the rotation axis, so $\sigma_h C_n = C_n \sigma_h$ always, regardless of order of application.

Reading: an $S_n$ operation is fully specified by an axis and an order $n$ — exactly like $C_n$ — but every application also flips the coordinate along that axis, so successive applications alternate between "still on the same side" and "on the other side" whenever $n$ is even, and cycle through both sides an odd number of times when $n$ is odd. The determinant of this matrix is

where: the determinant of any proper rotation matrix is $+1$ and the determinant of any reflection matrix is $-1$; this is the algebraic litmus test that distinguishes proper from improper operations in general, in any point group, for any axis orientation.

Reading: every improper operation has determinant $-1$; every proper rotation has determinant $+1$. This single number is why $S_n$ operations can never be achieved by physically rotating a rigid model — a rigid rotation can only ever produce $\det = +1$ transformations. The generalized power law for repeated $S_n$ operations is:

where:

• $k$ is the number of times $S_n$ is applied in succession,
• $C_n^k$ is a proper rotation by $k\theta = 2\pi k/n$,
• the extra $\sigma_h$ survives only when $k$ is odd, since each pair of reflections cancels.

Reading: an even number of applications of $S_n$ is indistinguishable from a pure rotation; only an odd number of applications retains the "improperness" (the mirror character) of the original operation.

Example 1 (foundational) — locating $S_4$ in allene, $\mathrm{H_2C{=}C{=}CH_2}$. Allene has the two $\mathrm{CH_2}$ planes mutually perpendicular, related by the $sp$-hybridized central carbon's two orthogonal $\pi$ systems. Place the $C{=}C{=}C$ axis along $z$, with the left $\mathrm{CH_2}$ group lying in the $xz$ plane and the right $\mathrm{CH_2}$ group lying in the $yz$ plane (a $90°$ dihedral twist between the two ends — this is the actual geometry, not an idealization). Number the four hydrogens: $H_1, H_2$ on the left group (in the $xz$ plane, symmetric about $z$), $H_3, H_4$ on the right group (in the $yz$ plane, symmetric about $z$). Take representative coordinates with the central carbon at the origin, left carbon at $z = -a$, right carbon at $z=+a$, and C–H bond projections making an angle $\beta$ with the $z$-axis:

with $d, a, c > 0$ specific nonzero numbers set by the actual bond lengths and angles (using asymmetric, nonzero, non-round values on purpose: $d = 0.93$ Å, $a = 0.66$ Å, $c = 0.51$ Å are realistic-order-of-magnitude values, though the argument below only needs the pattern, not the exact numbers). Apply $S_4$ about the central $z$-axis: rotate $90°$ about $z$, then reflect through the $xy$ plane ($z \to -z$). A $90°$ rotation about $z$ sends $(x,y,z) \to (-y,x,z)$ (using the standard counterclockwise convention). Then flip $z$:

Apply this to $H_1 = (+d, 0, -a-c)$:

Apply it to $H_2 = (-d, 0, -a-c)$:

Apply it to $H_3 = (0, d, a+c)$:

Apply it to $H_4 = (0,-d,a+c)$:

Every hydrogen maps onto another hydrogen; the carbon skeleton (all on the $z$-axis) is trivially preserved. So $S_4$ is a genuine symmetry operation of allene, even though allene has no mirror plane containing the $C{=}C{=}C$ axis and no inversion center (check: $i$ would send $H_1=(d,0,-a-c)$ to $(-d,0,a+c)$, which is not one of the four listed hydrogen positions — inversion fails). Allene is achiral (for unsubstituted allene) purely by virtue of this hidden $S_4$ axis, exactly the situation flagged in Step 4 of the second derivation above. Checking the algebraic prediction: since $4$ is even, $S_4^2$ must equal $C_{4/2} = C_2$. Apply the $S_4$ map twice:

which is indeed the standard $C_2$ rotation about $z$ (rotate $180°$: $x\to -x$, $y \to -y$, $z$ unchanged) — confirming $S_4^2 = C_2$, and this $C_2$ is exactly the well-known $C_2$ axis of allene along the cumulated-diene axis. This molecule belongs to point group $D_{2d}$.

Example 2 (applied) — staggered ethane and the $S_6$ axis, with a limiting-case check. Staggered ethane, point group $D_{3d}$, has the C–C bond along $z$. The front $\mathrm{CH_3}$ hydrogens sit at angles $0°, 120°, 240°$ around $z$ at height $+h$; the back hydrogens sit at angles $60°, 180°, 300°$ (staggered, offset by $60°$) at height $-h$, each at radial distance $r$ from the axis. Take one representative front hydrogen:

Apply $S_6$: rotate by $60°$ about $z$, then reflect $z \to -z$. Rotating $(r,0,h)$ by $60°$ gives $\left(r\cos 60°, r \sin 60°, h\right) = \left(\tfrac{r}{2}, \tfrac{r\sqrt3}{2}, h\right)$. Reflecting flips the last coordinate:

This is exactly the back hydrogen at angle $60°$, height $-h$ — a genuine atomic position, since the back set is staggered by $60°$ from the front set by construction. So $S_6$ is confirmed as a true symmetry operation of staggered ethane, and this is the operation that a lone $\sigma_h$ (mid-plane reflection, which alone would send $(r,0,h) \to (r,0,-h)$, landing between two real hydrogens, not on one) and a lone $C_6$ (which does not exist here, since the molecule only repeats every $120°$, i.e., has $C_3$, not $C_6$) each individually fail to provide. Since $6$ is even, group theory demands $S_6^2 = C_{6/2} = C_3$. Check directly: applying $S_6$ twice means rotating by $120°$ total and reflecting twice (which cancels):

which is indeed the next front hydrogen around — the ordinary $C_3$ symmetry of the methyl group, confirming the hidden-axis rule with completely asymmetric, non-special numerical coordinates ($r$, $h$ arbitrary positive numbers, $60°$ a genuinely non-trivial offset angle).

Limiting-case check: what happens to $D_{3d}$ ethane's $S_6$ if the molecule is artificially forced into the eclipsed conformation ($D_{3h}$)? Then the back hydrogens sit at $0°, 120°, 240°$ (no $60°$ offset), and $S_6(H_A)$ computed above, $\left(\tfrac r2, \tfrac{r\sqrt3}2, -h\right)$ (the position at angle $60°$), no longer coincides with any real hydrogen (which now sit only at $0°,120°,240°$ on the back face). So $S_6$ is lost upon eclipsing — exactly matching the textbook fact that eclipsed ethane ($D_{3h}$) has $\sigma_h$ and $S_3$ but not $S_6$, while only the staggered rotamer ($D_{3d}$) has the $S_6$ axis. This confirms the derivation is sensitive to the actual geometry, not just the "spinning cylinder" caricature of ethane.

Example 3 (exam/competition level) — counting the operations generated by an $S_5$ axis, and identifying the point group $D_{5d}$ vs $C_{5v}$ pitfall. Consider a hypothetical eclipsed–staggered bipyramidal fragment with a genuine $S_5$ axis (this occurs, for instance, in appropriately staggered metallocene-type or ferrocene-like $D_{5d}$ frameworks with five-membered rings offset by $36°$). Using the odd-$n$ result from the derivation ($n=5$ is odd, so the cyclic set has order $2n = 10$), list every distinct operation generated by repeated application of $S_5$, tracking rotation angle and parity of reflection count explicitly:

Read off the pattern: at $k=5$ (odd, half of $2n=10$) you land on $\sigma_h$ alone — the rotation angle is a full $360°$ (identity rotation) but an odd number of reflections have accumulated, leaving a bare mirror. Only at $k=10$ does the full identity $E$ reappear. This confirms explicitly, with a genuinely odd, non-trivial $n=5$, the general Step 3 result: an odd-order $S_n$ axis generates a cyclic set of $2n$ distinct operations, including a hidden bare $C_5$ rotation subgroup (the even-$k$ terms: $C_5, C_5^2, C_5^3, C_5^4, E$) and a hidden bare $\sigma_h$ (the $k=5$ term). The exam-trap: a student seeing "$S_5$ axis present" might reflexively write $C_{5v}$ (mistaking the mirror planes of $C_{5v}$, which are vertical $\sigma_v$ planes containing the axis, for the horizontal $\sigma_h$ that $S_5$ actually forces). But the $\sigma_h$ generated at $k=5$ is horizontal (perpendicular to the axis), not vertical — a completely different symmetry element with a completely different effect on axial versus equatorial substituents. A molecule with a true $S_5$ axis and the accompanying $C_5$ and $\sigma_h$ it generates, plus perpendicular $C_2$ axes, is properly classified $D_{5h}$ (if $\sigma_h$ survives as an actual, standalone symmetry element on its own — i.e., $n$ odd) or, in the staggered case with no standalone $\sigma_h$ but a genuine $S_{10}$, $D_{5d}$. Getting the parity argument right is exactly what prevents this misassignment.

The systematic classification of molecular symmetry operations into proper and improper rotations, and their organization into point groups, is usually credited to the broader development of crystallographic and molecular group theory in the late nineteenth and early twentieth centuries, but the chemist most responsible for making $S_n$-based point-group language the working vocabulary of spectroscopists and structural chemists was Hans Bethe, whose 1929 treatment of crystal-field splitting introduced systematic character-table methods to chemistry, soon merged with the mathematical crystallography of Arthur Schoenflies, whose notation ($C_n$, $S_n$, $D_n$, etc. — the "Schoenflies notation") is the one used throughout this article and throughout the chemical literature today.