Coordination Numbers and Common Geometries

Coordination number and geometry assignment is not an antiquarian classification exercise; it is load-bearing in modern chemistry across many fields. In catalysis, the interconversion between CN 4 (tetrahedral/square planar) and CN 5 intermediates is the literal mechanism of associative ligand substitution in square-planar Pt(II) and Pd(0/II) catalysis — cross-coupling reactions (Suzuki, Negishi, Heck) proceed through five-coordinate trigonal-bipyramidal-like transition states exactly because CN 5 sits at a shallow energy saddle, making the associative pathway kinetically accessible. In bioinorganic chemistry, metalloenzyme active sites are classified first by coordination number: zinc enzymes (carbonic anhydrase, carboxypeptidase) are almost always CN 4 or 5 at Zn²⁺, reflecting d¹⁰ Zn²⁺'s lack of LFSE preference and hence its versatility in accommodating whatever geometry the protein scaffold and substrate demand — a flexibility that is itself catalytically useful, allowing the metal to shuttle between four- and five-coordinate states during turnover. In materials science and lanthanide/actinide separations, high coordination numbers (7–9) govern the design of selective extractants for nuclear fuel reprocessing and rare-earth purification, since subtle differences in ionic radius across the lanthanide series (the lanthanide contraction) shift the preferred coordination number by one unit across the series, and ligands engineered to match a specific CN can achieve remarkable selectivity between otherwise nearly identical +3 ions. In MRI contrast agent design, Gd³⁺ chelates (like Gd-DTPA) are built around CN 8–9 coordination spheres, deliberately leaving one coordination site open for a rapidly exchanging water molecule — the entire clinical function of the contrast agent depends on getting the coordination number (and hence the number of "spare" sites) exactly right. In solid-state and materials chemistry, coordination geometry around metal centers in extended lattices (perovskites, spinels, zeolitic frameworks) directly sets electronic band structure and magnetic ordering, linking this same geometric vocabulary to superconductivity and magnetism research. And in instrumentation, X-ray absorption spectroscopy (EXAFS) is used specifically to determine coordination number and geometry in non-crystalline or biological samples where diffraction is unavailable, making the concepts here directly experimentally measurable in systems from metalloproteins to industrial catalysts in operando.

:::quiz[A hypothetical CN 7 complex [ZrF₇]³⁻ is expected to adopt which geometry, and why?]

• Cubic, because F⁻ is small enough to allow 90° angles throughout
• Trigonal prismatic, because Zr⁴⁺ is d⁰ with strong π-backbonding preferences
• Pentagonal bipyramidal, since small similar-sized F⁻ ligands with no strong steric asymmetry favor this near-degenerate CN 7 minimum ✓
• Square antiprismatic, the universal CN 7 geometry

Ask the tutor: Why does Berry pseudorotation make PF₅'s ¹⁹F NMR spectrum show only one peak, even though the static structure has two chemically distinct fluorine environments?

Imagine building a coordination complex atom by atom: place the metal cation at the origin, then bring up donor atoms (or lone-pair-bearing ligand atoms) one at a time. Each new ligand is repelled electrostatically and sterically by the ones already present — this is the same electron-pair repulsion logic as VSEPR, just applied to bonds that are typically more ionic/dative than covalent. For a fixed number of points confined to a sphere around a central atom, there is a "best" arrangement that maximizes the average angular separation between them; this is the classical Thomson problem in electrostatics, and its solutions are exactly the polyhedra chemists memorize: 2 points → linear (180°), 3 → trigonal planar (120°), 4 → tetrahedral (109.5°), 6 → octahedral (90°), 8 → square antiprism or cube. Five and seven are the interesting outliers: there is no single "best" arrangement, so nature offers close-energy alternatives (trigonal bipyramidal vs. square pyramidal for 5; pentagonal bipyramidal vs. capped octahedral vs. capped trigonal prismatic for 7), and real molecules often sit at a compromise point or interconvert rapidly between them. Onto this purely steric picture, transition metals add an electronic overlay: partially filled d-orbitals are not spherically symmetric, so certain geometries stabilize specific d-electron configurations more than others. A d⁸ ion like Ni²⁺, Pd²⁺, or Pt²⁺ combined with strong-field ligands gains enormous extra stability by adopting square planar (CN 4) rather than tetrahedral geometry, because square planar splits the d-orbitals so as to leave the highest, most destabilized orbital (dx²−y²) completely empty — worth a large ligand-field stabilization energy (LFSE) that tetrahedral geometry cannot offer. This is why [Ni(CN)₄]²⁻ is square planar and diamagnetic while [NiCl₄]²⁻ is tetrahedral and paramagnetic: same ion, same charge, but CN⁻ is strong-field enough to make the square-planar electronic payoff worth the steric cost, whereas Cl⁻ is not. Ligand size and bite angle further modulate everything. Bulky ligands (like triphenylphosphine, PPh₃) physically cannot pack six around a small metal center, capping the coordination number below what electronics alone would predict — this is the origin of Tolman's cone angle concept. Conversely, small, hard, high-charge-density ligands like F⁻ or O²⁻ can push coordination numbers up past 6 for large enough metal ions (lanthanides routinely reach CN 8–9 with small O-donor ligands). Chelating polydentate ligands (from Chelating and Polydentate Ligands) further constrain geometry by fixing bite angles that may not match the ideal polyhedral angle, forcing distortions that are diagnostic in their own right.

The central quantitative tool for CN 4–6 is ligand-field stabilization energy (LFSE), which supplies the electronic bias on top of steric preference. For an octahedral field split into t₂g (lower, 3-fold) and eg (upper, 2-fold) sets separated by Δₒ:

where:

• $n_{t_{2g}}$, $n_{e_g}$ = number of electrons in the t₂g and eg sets
• $\Delta_o$ = octahedral crystal-field splitting energy
• the coefficients −0.4 and +0.6 come from the barycenter rule (weighted average orbital energy must stay at zero, and 3 orbitals below / 2 above must balance: 3(−0.4) + 2(0.6) = 0)

For tetrahedral fields the splitting inverts (e below, t₂ above) and shrinks by roughly a factor of 4/9 because there are fewer ligands and they don't point directly at any orbital lobe:

where:

• $\Delta_t$ = tetrahedral splitting for the same metal/ligand pair
• the 4/9 factor arises from the smaller number of ligands (4 vs 6) and the fact that none of the tetrahedral ligand positions point directly along an orbital lobe (geometric overlap integral scaling)

For square planar fields, four in-plane orbitals shift dramatically and the SPE gain for a d⁸ ion becomes large enough to override tetrahedral's smaller steric strain; the qualitative one-electron ordering is

where:

• ligands lie along ±x and ±y, so $d_{x^2-y^2}$ points straight at all four ligands and is pushed highest
• $d_{z^2}$ has a doughnut in the xy-plane but no z-axis ligands, so it is destabilized less
• a filled set through $d_{z^2}$ with an empty $d_{x^2-y^2}$ (d⁸ low-spin) captures essentially all the stabilization available, at zero magnetic cost (diamagnetic)

Reading: LFSE tells you how much extra stability a given d-count gets from a given geometry; comparing LFSE(octahedral) vs LFSE(tetrahedral) vs LFSE(square planar) for the same electron count is exactly how you predict whether a d⁸ ion "prefers" square planar over tetrahedral, and by how much (in units of Δₒ). For CN 5, no single elegant formula decides TBP vs SPY; instead a continuous distortion coordinate — the Berry pseudorotation coordinate, characterized by the τ₅ geometric index — interpolates smoothly between them:

where:

• $\alpha, \beta$ are the two largest L–M–L angles in the five-coordinate complex ($\beta \geq \alpha$)
• $\tau_5 = 1$ for ideal trigonal bipyramidal (α = 120°, β = 180°)
• $\tau_5 = 0$ for ideal square pyramidal (α = β = 180°... actually both basal trans angles equal, giving β − α = 0)
• most real CN 5 complexes fall between 0.1 and 0.9, i.e., neither pure shape

Example 1 (foundational): assign CN and geometry for [Co(NH₃)₆]³⁺, [Zn(NH₃)₄]²⁺, and [Ag(NH₃)₂]⁺. For [Co(NH₃)₆]³⁺: six neutral NH₃ ligands, each monodentate through N, all directly bonded to Co³⁺. Count donor atoms: 6. Co³⁺ is d⁶; with strong-field NH₃ (moderately strong, and Co³⁺ complexes are almost always low-spin octahedral regardless), the electronic preference and the steric preference for CN 6 coincide exactly on octahedral geometry (Step 4 of the second derivation). Predicted: octahedral, low-spin d⁶, diamagnetic — matches the well known orange-yellow, diamagnetic [Co(NH₃)₆]³⁺. For [Zn(NH₃)₄]²⁺: four monodentate NH₃, CN = 4. Zn²⁺ is d¹⁰ — completely filled d-shell, so LFSE = 0 in any geometry (tetrahedral or square planar), meaning sterics decide alone. Four ligands around a small, symmetric ion with no directional d-orbital demand adopt the sterically optimal tetrahedral arrangement (109.5°, maximizing ligand separation for 4 points on a sphere). Predicted and observed: tetrahedral. For [Ag(NH₃)₂]⁺: two monodentate NH₃, CN = 2. Ag⁺ is d¹⁰, again LFSE = 0, but here the s–d mixing effect specific to d¹⁰ metals (Ag⁺, Au⁺, Hg²⁺, Cu⁺) favors linear, 2-coordinate geometry through hybridization stabilization of a filled dz² with an empty (n+1)s/(n+1)p orbital along the bonding axis, rather than a simple hard-sphere argument. Predicted and observed: linear, 180°.

Example 2 (applied): rationalize why [NiCl₄]²⁻ is tetrahedral and paramagnetic while [Ni(CN)₄]²⁻ is square planar and diamagnetic. Both are Ni²⁺, d⁸, CN = 4. The metal and oxidation state are identical; only the ligand identity changes. Cl⁻ is a weak-field ligand (near the low end of the spectrochemical series), so it cannot generate a large enough splitting to make emptying the high-energy dx²−y² orbital in square planar geometry worthwhile; the complex reverts to the sterically favored tetrahedral arrangement. In tetrahedral d⁸: e⁴t₂⁴ configuration with unpaired electrons in the t₂ set (2 unpaired, by Hund's rule across 3 orbitals holding 4 electrons) — paramagnetic, consistent with the pale, weakly colored, high-spin [NiCl₄]²⁻. CN⁻ is a very strong-field ligand (top of the spectrochemical series, strong π-acceptor via empty π* orbitals). It generates a large enough splitting that the electronic payoff of leaving dx²−y² empty (Step 3 of the first derivation) exceeds the steric cost of compressing four ligands into a plane. Low-spin d⁸ square planar has all eight electrons paired across the four lower orbitals — diamagnetic, matching the intensely colored, diamagnetic [Ni(CN)₄]²⁻. This pair is the single most commonly cited illustration in inorganic courses of "same metal, same charge, different geometry, different magnetism," precisely because it isolates the ligand-field-strength variable so cleanly.

Example 3 (exam/competition level): predict and rationalize the coordination number and geometry of [Y(H₂O)₈]³⁺ vs [Ce(H₂O)₉]³⁺, and estimate the τ-type distortion expected for a hypothetical CN 7 species [ZrF₇]³⁻. Y³⁺ has an ionic radius of about 1.02 Å (CN 8, Shannon radii), and lanthanide/early actinide ions of similar or larger radius (Ce³⁺ ≈ 1.14 Å at CN 9) routinely exceed CN 6 because these ions are d⁰ (no LFSE to speak of for f-electrons in the weak, poorly-shielding 4f/5f shells) and are simply large relative to a small O-donor ligand like water. With no strong electronic geometric preference, the coordination number is set almost purely by radius-ratio-type packing: bigger cation radius allows more ligands to fit around it without steric clash. Y³⁺ (smaller) adopts CN 8, typically as a (distorted) square antiprism or dodecahedron, while Ce³⁺ (larger, plus the lanthanide contraction not yet fully engaged this early in the series) reaches CN 9, typically tricapped trigonal prismatic — the canonical CN 9 shape, seen also in [Nd(H₂O)₉]³⁺. For CN 8, quantify the choice between the two common geometries — square antiprism (D4d) and dodecahedron (D2d) — using the logic that these are the two local energy minima of the CN 8 point-repulsion problem (analogous to the TBP/SPY situation at CN 5, but here the two minima are close enough in energy that both are observed depending on ligand sterics and crystal packing forces); no single universal rule picks one over the other, and interconversion between them is often facile (a fluxional lesson worth remembering, paralleling Berry pseudorotation at CN 5). For the hypothetical [ZrF₇]³⁻ (CN 7, Zr⁴⁺ radius ≈ 0.78 Å at CN 7, small F⁻ ligands): CN 7 has three competing idealized geometries — pentagonal bipyramidal (D5h), capped octahedron (C3v), and capped trigonal prism (C2v) — with very small energy differences (typically under a few kcal/mol by DFT estimates), so real CN 7 species are frequently found close to one of these ideal shapes but easily distorted by crystal packing or specific ligand sterics. Zr⁴⁺ with small, similar F⁻ ligands (no strong steric asymmetry) is expected to sit close to the pentagonal bipyramidal ideal, matching known ZrF₇³⁻ salts, which crystallize as (distorted) pentagonal bipyramids. The take-home lesson: as coordination number climbs past 6, the energy landscape flattens, several polyhedra become nearly degenerate, and small perturbations (crystal packing, minor ligand asymmetry, even temperature) can tip the observed structure from one idealized geometry to a neighboring one.