Stereoisomerism: Geometric (cis-trans, fac-mer) Isomers

Two bottles sit on a shelf, both labeled PtCl₂(NH₃)₂. Both have perfectly correct molecular formulas. Both are stable solids you could weigh out to the milligram. One is a potent anticancer drug. It's used in millions of chemotherapy infusions every year. The other is essentially inactive. Sometimes it's even antagonistic to the first. Nothing differs in what atoms are present. Only how they are arranged in space around the platinum center differs. This is cisplatin versus transplatin. The difference between them is the subject of this article: geometric stereoisomerism in coordination compounds. Stereoisomers share identical connectivity — the same atoms bonded to the same atoms. But they differ in spatial arrangement. This sets them apart from Structural (Constitutional) Isomerism in Complexes, where ligands trade places between coordination sphere and counter-ion. It also differs from linkage isomers, which bond through different donor atoms. Geometric (cis-trans and fac-mer) isomerism is the simplest kind of stereoisomerism. No chirality is required, just a fixed, rigid polyhedron. It must have more than one distinguishable way to place a given set of ligands on its vertices. The moment you have a square-planar or octahedral complex with two or more different ligand types, geometry starts asking combinatorial questions. The answers turn out to have life-and-death pharmacological consequences. They also carry industrial catalytic consequences, and deep structural-diagnostic value. The big picture is this: coordination geometries are rigid frameworks. These include square planar (D₄ₕ ideal symmetry) and octahedral (Oₕ ideal symmetry) geometries. Each has a fixed set of vertex positions. Placing ligands of different identity on these vertices in inequivalent ways generates isomers. These isomers cannot be interconverted without breaking and reforming bonds. A rare intramolecular twist mechanism is the only exception. For square-planar MA₂B₂ and MA₂BC species, this yields cis and trans isomers. For octahedral MA₃B₃ species, this yields fac (facial) and mer (meridional) isomers. Both distinctions are best understood using Octahedral, Tetrahedral, and Square-Planar Geometries as the geometric substrate. Both distinctions matter because cis/trans and fac/mer isomers differ in symmetry, dipole moment, reactivity, and spectroscopic signature. As cisplatin dramatically shows, they also differ in biological activity. This article builds the full geometric and group-theoretical picture of these isomer classes. It shows how to enumerate them rigorously without missing or double-counting isomers. It shows how symmetry (point groups) predicts distinguishing physical properties like polarity and IR/NMR pattern. It offers worked counting arguments for real complexes. It also covers the chemistry (synthesis, interconversion, and trans effect). This chemistry lets chemists deliberately steer a reaction toward one isomer or the other. It closes the loop toward Optical Isomerism and Chirality. Some geometric isomers (especially fac and certain low-symmetry cis arrangements) are themselves chiral. This layers a second axis of stereoisomerism on top of the first.

Key takeaways

Geometric isomers share the same connectivity (same bonds) but differ in the spatial arrangement of ligands on a rigid coordination polyhedron.
Square-planar MA₂B₂: cis (A's adjacent, 90°) vs trans (A's opposite, 180°) — only two isomers possible.
Octahedral MA₃B₃: fac (three A's on one triangular face, all mutually cis) vs mer (three A's along a meridian, two trans to each other) — only two isomers possible.
Point-group symmetry differs between isomers (e.g., trans-MA₂B₂ is D₄ₕ, cis is C₂ᵥ) and directly predicts polarity, dipole moment, and vibrational/NMR spectrum multiplicity.
Cisplatin (cis-[PtCl₂(NH₃)₂]) is a clinically vital anticancer drug; transplatin is clinically inactive — the same atoms, radically different biology.
Key idea: rigid coordination polyhedra have a finite, symmetry-countable number of distinct ligand arrangements, and these arrangements are physically and chemically distinguishable, not just formal bookkeeping.

The core insight

On a rigid polyhedron, the number of geometrically distinct ways to arrange a given multiset of ligands is fixed by the polyhedron's symmetry group; each distinct arrangement is a separate, isolable, physically real compound with its own symmetry, polarity, and reactivity.

Where this lives today

The cisplatin/transplatin contrast remains the textbook example precisely because it is not merely textbook: cisplatin, carboplatin, and oxaliplatin (all cis square-planar Pt(II) complexes) are front-line chemotherapeutics used against testicular, ovarian, bladder, and other cancers, precisely because the cis geometry allows the two adjacent leaving groups to be sequentially displaced by adjacent purine bases on one strand of DNA, producing the characteristic 1,2-intrastrand d(GpG) cross-link that bends the double helix roughly 30-40° and stalls replication and transcription machinery, triggering apoptosis in rapidly dividing cells. Transplatin's chlorides, pointing in opposite directions, cannot form the same intrastrand link and instead tend toward interstrand cross-links or protein adducts with different (and clinically far less useful, sometimes counterproductive) biological consequences — an object lesson in how geometry alone, with identical connectivity, dictates biological fate. Fac/mer isomerism matters throughout homogeneous catalysis: many industrially important octahedral catalysts and precatalysts (rhodium and iridium hydroformylation and hydrogenation catalysts, for example) exist as fac/mer mixtures whose ratio affects catalytic activity and selectivity, since the isomer with a vacant or labile site cis to the substrate-binding site often reacts at a very different rate than the isomer where that same ligand sits trans. Synthetic chemists deliberately exploit the trans effect (the kinetic labilization of a ligand trans to certain strong π-acceptor or good σ-donor ligands such as CO, CN⁻, PR₃, or H⁻) to direct substitution reactions toward a specific cis or trans product — this is literally how cisplatin itself is synthesized industrially and in the lab, using the trans effect of chloride (stronger than ammonia) to selectively install the second chloride cis to the first, starting from [PtCl₄]²⁻ and NH₃. Geometric isomer assignment by vibrational spectroscopy (the IR-active mode-counting argument developed above) and multinuclear NMR (the ³¹P counting argument in Worked Example 2) remain everyday tools in synthetic inorganic and organometallic laboratories, used routinely to characterize new complexes without requiring a full single-crystal X-ray structure for every reaction product — though X-ray diffraction remains the definitive method whenever ambiguity persists, especially for complex asymmetric ligand sets like the [M(bipy)(NH₃)₂Cl₂] system explored above. Finally, this framework is the direct stepping-stone to Optical Isomerism and Chirality: certain geometric isomers (fac arrangements of three different bidentate chelates, or low-symmetry cis/mer arrangements lacking any improper symmetry element) are themselves chiral, meaning a full structural description of a coordination compound often requires specifying both its geometric isomer type (cis/trans, fac/mer) and its absolute configuration (Δ/Λ) — two independent, stackable layers of stereochemistry built on the same rigid polyhedral scaffold.

:::quiz[What is the point group of trans-[PtCl₂(NH₃)₂]?]

C₂ᵥ
C₃ᵥ
D₂ₕ ✓
Td

:::quiz[Why does fac-[MA₃B₃] give a single ³¹P NMR signal for three B = PPh₃ ligands?]

The three P atoms are related by a C₃ rotation axis, making them symmetry-equivalent ✓
Phosphorus nuclei never show chemical shift differences in octahedral complexes
The complex is fluxional at room temperature and averages all signals
Only one PPh₃ ligand is actually bonded to the metal

Ask the tutor: Why does the trans effect of chloride, rather than its thermodynamic trans influence, control which isomer of [PtCl₂(NH₃)₂] actually forms during synthesis?

The physical picture

Start with the square plane. Four coordination sites lie at the corners of a square around the metal, all mutually cis (90° apart) except for the two pairs directly across the square from each other, which are trans (180° apart). For a complex MA₂B₂ (two A ligands, two B ligands, A ≠ B), there is only one way to place the two A's adjacent to each other (cis) and only one way to place them opposite each other (trans) — these two arrangements are not interconvertible without breaking bonds, and they are genuinely different molecules. The cis isomer has the two mirror planes of C₂ᵥ symmetry, a net dipole moment (the two M–A and two M–B dipole components do not cancel), while the trans isomer has full D₂ₕ (often idealized D₄ₕ) symmetry with a horizontal mirror plane and a center of inversion, and by symmetry its dipole moment is exactly zero because the two A–M bonds point in opposite directions and cancel, as do the two B–M bonds. Now consider the more general square-planar case MA₂BC (two identical A, one B, one C — all different from A and each other). Here the two A ligands can be placed either cis or trans to each other; once that choice is fixed, B and C fill the remaining two positions, but since B ≠ C there is nothing left to permute — the structure is fully determined. So MA₂BC also yields exactly two geometric isomers, distinguished only by whether the A–M–A angle is 90° or 180°. Move now to the octahedron. Six coordination sites sit at the vertices of an octahedron, and each vertex has four cis neighbors (90°) and one trans neighbor (180°, through the metal). For an octahedral MA₃B₃ complex, the three A ligands can be arranged in exactly two topologically distinct ways. In the first, all three A's occupy one triangular face of the octahedron — every A is cis to every other A. This is the facial (fac) isomer, and it has C₃ᵥ symmetry (a genuine three-fold rotation axis through the center of the A₃ face and the opposite B₃ face). In the second, the three A's are arranged so that two of them are trans to each other and the third A completes a "belt" around the equator together with the three B's — this is the meridional (mer) isomer, tracing a great circle (a "meridian") around the octahedron, with only Cₛ (one mirror plane) symmetry. These are the only two topologically distinct ways to place three identical ligands among six octahedral vertices when the other three are also identical to one another — a result you can verify by brute-force enumeration of the $\binom{6}{3} = 20$ ways to choose 3 of 6 vertices for the A ligands, which collapse under the octahedral symmetry group Oₕ (order 48) into exactly two orbits.

Figure

Square-planar cis and trans MA₂B₂: symmetry and dipole cancellation

Figure

Octahedral fac and mer MA₃B₃: face vs meridian

The core ideas

The count of geometric isomers for a given ligand set on a given polyhedron is not arbitrary — it is dictated by the symmetry of the coordination polyhedron via Burnside's lemma / orbit counting, though for the simple compositions treated at this level (MA₂B₂, MA₃B₃, MA₂B₂C₂, etc.) direct geometric reasoning suffices and gives the same answer. For a general labeling problem — place $n_A$ ligands of type A, $n_B$ of type B, … on the $N$ vertices of a polyhedron with symmetry group $G$ of order $|G|$ — the number of geometrically distinct isomers is

n_{\text{isomers}} = \frac{1}{|G|}\sum_{g \in G} \chi(g)

where:

$|G|$ is the order of the rotation group of the polyhedron (proper rotations only — 24 for the octahedron, 8 for the square, since reflections are not physical motions but relevant for a separate chirality question),
$\chi(g)$ is the number of vertex-labelings left unchanged (fixed) by the symmetry operation $g$ ,
the sum runs over every rotational symmetry operation of the polyhedron.

Reading: this is Burnside's lemma applied to combinatorics of labeled vertices — it says the number of physically distinct arrangements equals the average number of labelings fixed by each symmetry operation, not the naive $\binom{N}{n_A, n_B, \dots}$ count (which overcounts by conflating rotationally-equivalent structures). A second, more chemically transparent relationship is the polarity/symmetry correspondence: a molecule has zero permanent dipole moment if and only if its point group contains a symmetry element (a $C_n$ axis with $n>1$ combined with perpendicular $C_2$ 's, or an inversion center, or $S_n$ ) that forces the dipole vector to vanish. Formally,

\vec{\mu} = \sum_i q_i \vec{r}_i

where:

$\vec{\mu}$ is the total molecular dipole moment vector,
$q_i$ is the effective partial charge on atom $i$ ,
$\vec{r}_i$ is its position vector from the centroid,
the sum runs over all atoms in the molecule.

Reading: any symmetry operation of the molecule's point group must leave $\vec{\mu}$ invariant (since it leaves the whole molecule invariant); if the group contains an operation that reverses some component of $\vec{\mu}$ (like inversion, which sends $\vec r_i \to -\vec r_i$ ), that component must be exactly zero. This is why trans-MA₂B₂ (D₂ₕ, has $i$ ) is nonpolar while cis-MA₂B₂ (C₂ᵥ, no $i$ ) is polar.

Derivation ACounting geometric isomers of octahedral MA₃B₃ by direct enumeration

Goal: show rigorously that MA₃B₃ octahedral complexes have exactly two geometric isomers (fac and mer), not more, by reducing the raw combinatorial count using the octahedral rotation group.

Step 1 — set up the raw count. Label the six octahedral vertices $1,\dots,6$ with $1$ – $2$ trans (an axial pair, "top/bottom"), and $3,4,5,6$ the four equatorial vertices arranged in a square, with $3$ – $5$ trans and $4$ – $6$ trans. We must choose 3 of these 6 vertices to hold ligand A (the other 3 automatically hold B). The naive number of ways is

\binom{6}{3} = 20 .

Not all 20 labelings are geometrically distinct: the octahedron's rotation group $O$ (order 24) can rotate one labeled structure into another, and such rotated copies are the same physical molecule, just redrawn.

Step 2 — classify the 20 vertex-triples by trans-pair content. Each triple of vertices either (a) contains no trans pair at all (all three mutually cis — this is a triangular face) or (b) contains exactly one trans pair plus one more vertex. It cannot contain two trans pairs, since a trans pair uses up 2 of the 3 chosen vertices, leaving only 1 more vertex, which can be trans to at most one already-chosen vertex given the octahedron only has 3 trans pairs total. Enumerate: the octahedron has 8 triangular faces, hence 8 triples of type (a) — but each face and its "antipodal" (opposite) face are related by the inversion operation, not a proper rotation, so under the pure rotation group these 8 faces fall into orbits; direct check shows all 8 triangular faces are related to each other by proper rotations of the octahedron (rotating the top face to any of the 8 faces), so type (a) is a single orbit of size 8.

Step 3 — the "one trans pair + one more" triples. There are 3 trans pairs in the octahedron (top-bottom, and the two equatorial diagonals), and for each trans pair there are 4 remaining vertices to serve as the third member of the triple, giving $3 \times 4 = 12$ triples of type (b). Check the total: $8 + 12 = 20$ , matching Step 1.

Step 4 — show each class is a single orbit under rotation. The octahedral rotation group acts transitively on the 8 triangular faces (any face can be rotated to any other face — this is geometrically the statement that the octahedron looks the same from any of its 8 faces), so all 8 type-(a) triples are the same isomer: this is fac. Likewise, the rotation group acts transitively on the 12 type-(b) triples (any trans-pair-plus-one arrangement can be rotated into any other, since the group can rotate any trans axis to any other trans axis, and then rotate about that axis to bring the "extra" vertex to any of the 4 remaining equatorial positions): all 12 are the same isomer, mer.

Step 5 — conclude. The 20 raw labelings collapse into exactly $8 + 12 = 20$ accounted for in exactly 2 orbits (sizes 8 and 12), so

n_{\text{isomers}}(\text{MA}_3\text{B}_3, O_h) = 2.

This matches Burnside's lemma applied to the full calculation (summing fixed-labeling counts over all 24 rotations of $O$ and dividing by 24), confirming that fac and mer are the only two geometric isomers — there is no hidden third arrangement.

Go deeper · Derivation B · 2 minSymmetry-based prediction of dipole moment and IR-active stretch count for cis vs trans MA₂B₂

Derivation BSymmetry-based prediction of dipole moment and IR-active stretch count for cis vs trans MA₂B₂

Goal: derive, from point-group symmetry alone (without needing detailed force constants), why cis-[MA₂B₂] shows a nonzero dipole and more IR-active M–L stretching bands than trans-[MA₂B₂], giving a symmetry-based diagnostic that chemists actually use to assign cis/trans structures from vibrational spectra.

Step 1 — assign point groups. Trans-MA₂B�2, with the two A ligands 180° apart and two B ligands 180° apart, has the symmetry elements: $E$ , $C_2$ (through M, perpendicular to plane, and also the two in-plane $C_2$ axes along the A–M–A and B–M–B lines), $\sigma_h$ (the molecular plane), two $\sigma_v$ , and $S_2 = i$ (inversion center at M). This is point group $D_{2h}$ . Cis-MA₂B₂ has the A–M–A angle at 90°, so only a single $C_2$ axis bisecting the A–M–A and B–M–B angles survives, plus two mirror planes containing that axis (one is the molecular plane, one perpendicular to it through M) — point group $C_{2v}$ , with no inversion center.

Step 2 — apply the inversion argument to the dipole. In $D_{2h}$ , inversion $i$ sends the position vector of each A ligand to the position of the other A ligand (since they are related by $i$ through M), and likewise for the two B's. Since the two A ligands are chemically identical, their bond-dipole contributions are equal in magnitude and, after inversion, point exactly oppositely — so they must cancel:

\vec{\mu}_{\text{trans}} = \vec{\mu}_{A_1} + \vec{\mu}_{A_2} + \vec{\mu}_{B_1} + \vec{\mu}_{B_2} = \vec 0

because $\vec\mu_{A_2} = -\vec\mu_{A_1}$ and $\vec\mu_{B_2} = -\vec\mu_{B_1}$ by the inversion symmetry. In $C_{2v}$ (cis), there is no inversion center, so no such cancellation is forced: the M–A and M–B bond dipoles add with a nonzero resultant along the $C_2$ axis,

\vec{\mu}_{\text{cis}} = 2\left(\mu_{MA}\cos45^\circ + \mu_{MB}\cos45^\circ\right)\hat{u} \neq \vec 0

where $\hat u$ bisects the A–M–A angle and $\mu_{MA}, \mu_{MB}$ are the individual M–A and M–B bond dipole magnitudes (positive numbers for polar bonds); this is generically nonzero unless $\mu_{MA}$ and $\mu_{MB}$ happen to cancel exactly by coincidence, which does not occur for chemically distinct A, B.

Step 3 — count IR-active M–L stretching modes by symmetry species. For $D_{2h}$ trans-MA₂B₂, the four M–L stretches decompose (by standard reducible-representation reduction on the four bond vectors) into $A_g + B_{1g} + B_{2u} + B_{3u}$ (schematically — exact labels depend on axis convention). Only the $u$ (ungerade) combinations are IR-active, because IR activity requires a mode to transform like $x,y,$ or $z$ , and in a centrosymmetric group these transform only as $u$ representations (the mutual exclusion rule: no mode can be both IR- and Raman-active in a molecule with inversion symmetry). This leaves 2 IR-active M–L stretches for trans (asymmetric combinations), while the symmetric $A_g$ combination is Raman-active only and IR-silent.

Step 4 — repeat for $C_{2v}$ cis. With no inversion center, mutual exclusion does not apply: all four stretching combinations reduce to symmetry species ( $2A_1 + B_1 + B_2$ in $C_{2v}$ ) that are all IR-active, since $A_1, B_1, B_2$ all transform as translations ( $z$ , $x$ , $y$ respectively) in $C_{2v}$ . This gives up to 4 IR-active M–L stretching bands for cis.

Step 5 — state the diagnostic. Trans-MA₂B₂ (centrosymmetric, $\vec\mu=0$ ) shows a simpler IR spectrum in the M–L stretching region (fewer bands, since some modes are IR-silent by the mutual exclusion rule) and is formally nonpolar, while cis-MA₂B₂ ( $C_{2v}$ , $\vec\mu \neq 0$ ) shows a richer IR spectrum (more allowed bands) and is polar. This symmetry argument — not detailed force-constant calculation — is precisely how vibrational spectroscopy is used to assign cis versus trans geometry to an unknown square-planar complex.

Worked examples

Example 1 (foundational): Enumerate and classify isomers of [Pt(NH₃)₂Cl₂], the cisplatin/transplatin system. [Pt(NH₃)₂Cl₂] is square planar (Pt(II), d⁸, strongly favors $dsp^2$ square-planar coordination as predicted by crystal field stabilization of the $d_{x^2-y^2}$ orbital being left empty). The formula is MA₂B₂ with A = NH₃, B = Cl. By the direct geometric argument (no group theory needed for four vertices), there are exactly two arrangements:

cis-[PtCl₂(NH₃)₂]: Cl–Pt–Cl angle = 90°, NH₃–Pt–NH₃ angle = 90°. Point group $C_{2v}$ . This is cisplatin, the clinical anticancer drug. It has a nonzero dipole moment (calculated dipole ≈ 10 D range experimentally, reflecting the substantial Pt–Cl versus Pt–N bond polarity difference) and, crucially, has two adjacent, mutually cis leaving-group chlorides that can be sequentially displaced by the two exocyclic N7 nitrogens of adjacent guanine bases in DNA, forming a 1,2-intrastrand cross-link that kinks the DNA duplex and blocks replication.
trans-[PtCl₂(NH₃)₂]: Cl–Pt–Cl angle = 180°, NH₃–Pt–NH₃ angle = 180°. Point group $D_{2h}$ (planar, centrosymmetric). This is transplatin. Its two chlorides point in opposite directions and, geometrically, cannot reach two adjacent bases on the same DNA strand to form the same kind of tight intrastrand cross-link; it forms different, less-productive (often interstrand or protein-trapping) adducts and is clinically essentially inactive, at conventional doses even considered to work against the therapeutic goal in some contexts.

Check against the enumeration rule: MA₂B₂ on 4 square-planar vertices gives exactly $n_{\text{isomers}} = 2$ , matching direct inspection — there is no missing third isomer, because swapping the labels of the two identical NH₃ ligands or the two identical Cl ligands returns the same structure (not a new isomer), and the only geometrically distinguishable choice is whether the A's are adjacent (cis) or opposite (trans).

Example 2 (applied): Determine the point group, polarity, and expected number of ³¹P{¹H} NMR environments for fac- and mer-[Co(NO₂)₃(PPh₃)₃] (fictitious but chemically reasonable octahedral MA₃B₃ system with A = NO₂⁻, B = PPh₃, used here as a clean, asymmetric worked case with a spectroscopically simple B-donor). Take the general octahedral MA₃B₃ complex [CoA₃B₃] with A = NO₂⁻ and B = PPh�3. fac isomer: the three PPh₃ (B) ligands occupy one triangular face, all mutually cis to each other, and the three NO₂⁻ (A) ligands occupy the opposite face. This isomer has a genuine $C_3$ axis passing through the centroids of the two triangular faces, plus three vertical mirror planes containing that axis and each unique Co–B bond — point group $C_{3v}$ . Because $C_{3v}$ contains a $C_3$ axis (which is not a $C_2$ , so it doesn't force cancellation the way inversion does) but crucially no inversion center and no perpendicular $C_2$ 's, the molecule retains a net dipole moment along the $C_3$ axis — fac isomers of MA₃B₃ type are always polar. By the $C_3$ symmetry, all three PPh₃ ligands are related to one another by the rotation, hence chemically and magnetically equivalent: fac-[CoA₃B₃] shows a single ³¹P NMR resonance (one peak, since $C_3$ interconverts the three P environments by symmetry — they are homotopic). mer isomer: two PPh₃ ligands are trans to each other (axial), and the third PPh₃ is in the equatorial position trans to one of the NO₂⁻ ligands. The point group is $C_s$ (only one mirror plane survives, containing the unique equatorial B and bisecting the axial B–Co–B axis). This mirror plane relates the two axial B ligands to each other, so those two are equivalent, but the equatorial B is distinct (it lies in the mirror plane, unrelated to the other two by any symmetry operation). Hence mer-[CoA₃B₃] shows two distinct ³¹P NMR resonances, in a 2:1 intensity ratio (two equivalent axial P's, one unique equatorial P) — a spectroscopically clean diagnostic that instantly distinguishes fac (1 peak) from mer (2 peaks, 2:1) without needing crystallography. This NMR counting argument generalizes the reasoning used in Step 3 of the second derivation: point-group symmetry directly determines how many chemically inequivalent nuclei exist by asking which atoms are related to which others by a symmetry operation of the molecule (its point group's atomic orbits).

Example 3 (exam/competition level): For [Ru(bipy)(NH₃)₂Cl₂], predict all distinct geometric (and note any additional) isomers, and justify the total count using symmetry-orbit reasoning. This is a mixed-ligand octahedral complex: one bidentate chelate (bipy, occupying two cis coordination sites because its bite angle geometrically forbids spanning a 180° trans arrangement), two NH₃, and two Cl. Because bipy is constrained to be cis (chelate rings cannot span trans positions on an octahedron — the ligand backbone is far too short to reach both axial-opposite vertices), the remaining four vertices (after fixing the two adjacent bipy sites) form effectively a square-planar-like set of four positions around the "equator" relative to the bipy, into which NH₃, NH₃, Cl, Cl must be placed. Label the octahedron vertices: bipy occupies vertices 1 and 2 (a cis-related edge). The remaining four vertices 3, 4, 5, 6 form a "belt" in which 3 is trans to 5, and 4 is trans to 6 (with 1 and 2 each trans to one of 3,4,5,6 too, but bipy already fixes those). We must place two Cl and two NH₃ on vertices {3,4,5,6}. The relevant possibilities, distinguished by whether the two Cl are mutually cis or mutually trans:

Cl trans to Cl (and consequently NH₃ trans to NH�3): this gives one isomer where both cis-pairs of monodentate ligands are trans-disposed to their own kind. Call this isomer (i).
Cl cis to Cl (NH₃ also forced cis to NH₃, since only 4 sites and 2+2 ligands): but within this "cis" family, there are still two sub-cases depending on whether the Cl that is cis to the other Cl is also cis or trans to a specific nitrogen of bipy — because the two bipy nitrogens (vertices 1, 2) are themselves inequivalent relative to the belt once we fix which vertex is "trans to vertex 3" versus "trans to vertex 4." Careful bookkeeping (place Cl at 3 and 4 adjacent; NH₃ at 5 and 6) versus (Cl at 3 and 6 adjacent on the other side) actually generate structures related by the mirror symmetry of bipy itself (bipy has a local $C_2$ axis swapping its two nitrogens), so these turn out to be the same molecule, not two more isomers.

Careful enumeration (standard result, confirmed by symmetry-orbit counting analogous to the fac/mer derivation but now with the reduced symmetry set by the fixed cis-bipy edge) gives three total geometric isomers for [M(bipy)(NH₃)₂Cl�2]-type complexes: 1. Cl trans to Cl, NH₃ trans to NH₃ (both "trans" relative to each other across the belt). 2. Cl trans to NH₃, Cl trans to NH₃ (the two Cl's cis to each other, straddling one bipy nitrogen each, with each Cl trans to an NH₃) — two versions related by which bipy-N each is nearer, but which collapse or remain distinct depending on whether bipy's own local symmetry maps one version onto the other. 3. A third arrangement in which one Cl is trans to a bipy nitrogen. This example is included specifically to show that once a chelate ring or three-plus different ligand types are involved, isomer counting requires the same orbit-counting rigor as the fac/mer derivation — naive "2 choices, so 2 isomers" reasoning under-counts, and careful attention to which vertices are related by the actual residual symmetry (here, only the local mirror of bipy plus overall molecular symmetry) is essential. This complexity is exactly why total isomer counts for asymmetric polydentate octahedral complexes are typically settled by explicit model-building or computational enumeration rather than by inspection, and why many real syntheses of such complexes are reported alongside X-ray structures to settle which isomer actually formed.

Limiting cases, scope & common traps

Case	What happens	Why
Square-planar MA₄ or MA₃B	Only one isomer possible	No inequivalent vertex partition exists — all A's related by symmetry, or only one B has no "partner" to be cis/trans to
Square-planar MA₂BC (A₂ + 2 different singles)	Exactly 2 isomers (A's cis or trans)	Once A–A relationship fixed, B and C positions are forced (no further choice)
Square-planar MABCD (four different ligands)	3 geometric isomers	Three distinct ways to choose which ligand is trans to A
Octahedral MA₄B₂	2 isomers: cis (B's 90°) and trans (B's 180°)	Same logic as square-planar MA₂B₂, applied to the B pair; A₄ automatically arranges into the remaining positions with no further choice
Octahedral MA₃B₃	Exactly 2 isomers: fac, mer	Proven by direct orbit enumeration (worked derivation above)
Octahedral MA₂B₂C₂	6 geometric isomers	More inequivalent trans-relationships are possible; requires careful orbit counting, not naive combinatorics
Tetrahedral MA₂B₂	Only 1 isomer — no cis/trans distinction	Every pair of vertices in a tetrahedron is symmetry-equivalent (all angles are 109.5°, all pairs are "cis"-like); there is no "trans" position in $T_d$
Chelate rings with short bite angle	Cannot span trans positions	Geometric/steric constraint — reduces the isomer-counting problem to fewer free vertices
High-spin vs low-spin electronic configuration	Does not change isomer count	Geometric isomerism is a structural/topological question, independent of $d$ -electron count (though ligand field strength can bias which isomer is thermodynamically preferred)

Common trapMisconception: "cis and trans are just names for different rotamers, so they interconvert freely"

Cis and trans (or fac and mer) isomers are not conformers or rotamers — you cannot interconvert them by simply rotating single bonds, because the M–L bonds in question are the coordination bonds themselves, fixed by the rigid polyhedral geometry. Interconversion, when it occurs at all, requires either full ligand dissociation/re-association (an associative or dissociative substitution mechanism) or, in special cases, a genuine intramolecular twist mechanism (e.g., Bailar twist or Ray–Dutt twist for octahedral complexes, or a rare pseudorotation pathway). These are real, sometimes slow, sometimes catalyzed chemical processes with activation energies — not free rotations. Confusing geometric isomerism with conformational flexibility is one of the most common errors in early exposure to coordination stereochemistry.

The chemist behind it

Origins · 1 minAlfred Werner and how this was discovered

Alfred Werner

1866-1919 · Coordination chemistry, Zurich, turn of the 20th century

Alfred Werner, working at the University of Zurich in the 1890s, faced a foundational puzzle: classical valence theory (fixed, directional bonds satisfying simple integer valence) could not explain why cobalt(III) ammine-chloride complexes of the same empirical formula existed in multiple distinct colored forms with different chemical reactivities, nor why some conducted electricity differently in solution (differing numbers of free chloride ions, as measured by silver nitrate precipitation and conductometric titration). Werner proposed that metal ions possess a fixed "coordination number" (commonly six) defining a secondary valence, geometrically arranged around the metal — for six-coordination, an octahedral arrangement, not the then-more-obvious hexagonal-planar alternative. Crucially, Werner predicted that MA₄B₂-type octahedral complexes should show exactly two isomers (cis and trans), while a hexagonal-planar arrangement of six ligands would predict three distinguishable isomers (1,2-; 1,3-; 1,4- relationships around a hexagon). He resolved this experimentally, most famously by painstaking isomer counts on cobalt ammine complexes and, later, spectacular resolution of an octahedral complex into optically active enantiomers (containing no carbon at all), definitively ruling out any exclusively carbon-based origin for optical activity and cementing the octahedral model. Werner received the Nobel Prize in Chemistry in 1913, the first inorganic chemist to be so honored, for this structural theory of coordination compounds.

History

- 1893 — Werner proposes the coordination theory: fixed coordination number, geometric (often octahedral) arrangement of ligands around a central metal.

- 1907 — Werner and collaborators isolate and distinguish cis/trans (praseo/violeo) isomers of [CoCl₂(NH₃)₄]⁺ salts, matching predictions of the octahedral model over rival hexagonal-planar models.

- 1911–1914 — Werner resolves optically active, purely inorganic octahedral cobalt complexes (e.g., hexol), demonstrating chirality without carbon and confirming three-dimensional octahedral geometry.

- 1913 — Werner awarded the Nobel Prize in Chemistry for his coordination theory of complexes.

- 1965 — Barnett Rosenberg discovers the antitumor activity of cis-[PtCl₂(NH₃)₂] (cisplatin) during electrolysis experiments on bacterial cell division, launching platinum-based chemotherapy and giving cis/trans isomerism direct clinical stakes.

What exactly distinguishes geometric (cis-trans/fac-mer) isomerism from structural isomerism?

Structural (constitutional) isomers differ in connectivity — which atoms are bonded to which — such as ionization isomers (a ligand swaps places with a counter-ion) or linkage isomers (a ligand binds through a different donor atom). Geometric isomers, by contrast, have identical connectivity: the same set of ligands bonded through the same donor atoms to the same metal, in the same numbers. What differs is purely the three-dimensional spatial arrangement of those ligands on the rigid coordination polyhedron — which vertex holds which ligand. Because connectivity (bonding pattern) is unchanged, geometric isomers typically have very similar chemical formulas-derived properties (same molecular weight, same overall bond count) but can differ sharply in symmetry, polarity, and reactivity, precisely because those properties depend on 3D shape, not just on the bonding graph.

Why can't cis and trans isomers simply interconvert by rotating a bond, the way conformers do?

Bond rotation (as in ethane conformers) works for single bonds where the two halves of the molecule can spin relative to each other around the bond axis without breaking anything. In a coordination complex, the "positions" that distinguish cis from trans are coordination sites on a rigid polyhedron defined by the metal-ligand bonding framework itself — there is no single bond whose rotation would move a ligand from a cis site to a trans site while leaving all bonds intact. Converting cis to trans genuinely requires breaking at least one metal-ligand bond and re-forming it elsewhere (dissociative or associative substitution) or, for octahedral complexes, an intramolecular polytopal rearrangement (twist mechanism) that momentarily distorts the octahedron toward a trigonal-prismatic transition state before relaxing to the new arrangement — a real chemical process with a measurable activation barrier, not a free rotation.

Is fac always more (or less) stable than mer, or does it depend on the ligands?

It depends on the ligands, primarily through a competition between electronic (trans-influence/π-backbonding) effects and steric effects. When the A ligands are strong π-acceptors (like CO or NO₂⁻) that prefer to avoid being mutually trans to each other (to spread out competition for the same set of metal d-orbitals available for back-donation), fac tends to be favored, since fac has no A trans to A at all. When steric bulk dominates and ligands prefer to minimize mutual cis (90°) crowding, mer, which reduces the number of very close ligand-ligand cis contacts among the bulky groups, can be favored instead. Both isomers are frequently observed as kinetic products depending on synthetic route, and many syntheses report thermodynamic isomerization (fac to mer or vice versa) occurring slowly on heating or with a catalyst, showing that the preference is a genuine energetic balance, not a fixed rule.

How does the trans effect help chemists deliberately synthesize the cis (not trans) isomer of platinum complexes like cisplatin?

The trans effect is the empirical (and now mechanistically well-understood, largely associative-mechanism-based) observation that certain ligands, especially strong π-acceptors and soft, polarizable ligands like Cl⁻, Br⁻, CN⁻, CO, and PR₃, kinetically labilize whatever ligand sits trans to them, accelerating its substitution. Starting from [PtCl₄]²⁻ and adding two equivalents of NH₃, the first NH₃ substitutes a chloride; because Cl⁻ has a stronger trans effect than NH₃, the second incoming NH₃ preferentially substitutes the chloride that sits trans to a chloride already in the molecule (since that position is more labilized) rather than the one trans to the newly installed NH₃. The result, worked through systematically, is that the two remaining chlorides end up cis to each other in [PtCl₂(NH₃)₂] — precisely the cisplatin geometry — making the trans effect a rational synthetic tool for steering isomer outcome, not just a spectator observation.

Can a fac or mer isomer also be chiral (optically active)?

Yes, and this is exactly the layering that leads into optical isomerism. The fac isomer of an octahedral complex bearing three different bidentate chelating ligands, or an MA₃B₃ complex with additional asymmetric substitution, can lack any improper symmetry element (mirror plane, inversion center, or improper rotation axis) even though it retains its fac topology, making it chiral with a non-superimposable mirror-image partner (its enantiomer). The mer isomer, having only a single mirror plane ( $C_s$ ) in the simplest MA₃B₃ case, is by contrast achiral in that specific case (the mirror plane itself is a symmetry element that guarantees superimposability on its mirror image) — but more complex mer arrangements with additional distinct ligands can become chiral too. The general rule is always the same: chirality is decided by the absence of any improper symmetry element ( $\sigma$ , $i$ , or $S_n$ ) in the specific isomer's actual point group, checked case by case.

Does geometric isomerism occur in coordination numbers other than 4 and 6, such as 5-coordinate complexes?

Yes. Five-coordinate complexes, typically trigonal bipyramidal (TBP) or square pyramidal (SP), also show geometric isomerism, though the counting is more involved because TBP geometry has two inequivalent vertex types from the start (two axial, three equatorial), so even an MA₄B complex already has two isomers (B axial or B equatorial) without needing two different multiply-occurring ligands. Square pyramidal complexes similarly distinguish apical from basal positions. These five-coordinate cases are conceptually the same enumeration exercise as the square-planar and octahedral cases treated here — rigid polyhedron, inequivalent vertex classes fixed by symmetry — just with a geometry (TBP) that starts out less symmetric (only $D_{3h}$ , not $O_h$ ) so that even simple ligand sets can already show isomerism.

Why does Werner's historical resolution of a purely inorganic octahedral complex into optical isomers matter for the cis/trans-fac/mer story specifically?

Werner needed to rule out a rival hexagonal-planar model for six-coordination, which predicted three (not two) isomers for MA₄B₂-type complexes (ortho/meta/para-like relationships around a hexagon), against the octahedral model's prediction of exactly two (cis/trans). Experimentally finding only two isomers for compounds like [CoCl₂(NH₃)₄]⁺ supported the octahedral geometry directly — this is the same enumeration logic used throughout this article. His later resolution of a chiral, all-inorganic hexol complex into non-superimposable mirror images went a step further, proving the three-dimensional (not planar) character of the octahedron beyond doubt, since only a genuinely three-dimensional, chirality-permitting arrangement could support optical activity without any carbon stereocenter. Both results together cemented the geometric framework — cis/trans, fac/mer, and eventually chirality — as physically real, testable predictions rather than bookkeeping conventions.