Square-Planar Crystal Field Splitting

Square-planar Pt(II) and Pd(II) complexes are the backbone of modern homogeneous catalysis: Pd(0)/Pd(II) cycles in Suzuki, Heck, and Negishi cross-coupling reactions pass through square-planar Pd(II) intermediates whose ligand substitution kinetics (associative, via a five-coordinate trigonal-bipyramidal transition state) are a direct consequence of the accessible, empty, high-energy $d_{x^2-y^2}$ orbital pointing exactly at the site of incoming nucleophile attack. Wilkinson-type and related Rh(I)/Ir(I) square-planar d⁸ catalysts (themselves isoelectronic with Pt(II)/Pd(II)) drive industrial hydrogenation and hydroformylation, their oxidative-addition chemistry likewise gated by the same empty high-lying orbital. In medicine, cisplatin — cis-diamminedichloroplatinum(II) — is a square-planar Pt(II) d⁸ complex whose geometry is mechanistically essential: only the cis isomer can chelate adjacent guanine bases on DNA to form the crosslink responsible for its anticancer activity, while the trans isomer (same square-planar splitting, different substitution pattern) is pharmacologically inactive. The rigid, substitution-inert-but-tunable square-planar Pt(II) center, governed by exactly the splitting diagram derived above, is the structural reason cisplatin and its descendants (carboplatin, oxaliplatin) function as targeted DNA-damaging agents rather than indiscriminately reactive species. In materials science, square-planar d⁸ complexes of Pt(II), Pd(II), and Au(III) with extended aromatic or cyanide ligands are prized for their photoluminescence and metal–metal (often called metallophilic, $d_{z^2}$–$d_{z^2}$) interactions in the solid state, exploited in OLED phosphors and molecular electronics; the low-lying, partially populated $d_{z^2}$ orbital derived directly in this splitting diagram is precisely the orbital responsible for these intermolecular Pt···Pt and Au···Au stacking interactions. Bioinorganic chemistry relies on the same picture for vitamin B₁₂'s cobalt corrin (formally octahedral but with a strongly tetragonally distorted, near-square-planar equatorial field) and for synthetic metalloporphyrin mimics of heme and chlorophyll, where the square-planar-like equatorial ligand field sets the metal's spin state and reactivity toward axial ligand binding.

:::quiz[Why is octahedral d⁸ (e.g. [Ni(H2O)6]2+) always paramagnetic regardless of ligand field strength?]

• Because dz² and dx²−y² are never degenerate in Oh
• Because the eg set has only two orbitals, forcing one electron into each by Hund's rule for the last two electrons ✓
• Because water is always a weak-field ligand
• Because Ni2+ cannot achieve a d8 configuration in octahedral fields

Ask the tutor: How would the square-planar splitting diagram and its energies change if the four ligands were strong π-acceptors like CO instead of pure σ-donors?

Begin with an octahedral complex, ligands along $\pm x, \pm y, \pm z$, giving the familiar splitting into $t_{2g}$ (three orbitals: $d_{xy}, d_{xz}, d_{yz}$, lower) and $e_g$ (two orbitals: $d_{z^2}, d_{x^2-y^2}$, higher), separated by $\Delta_o$. Now slide the two ligands on the $z$-axis away from the metal, symmetrically, while keeping the four equatorial ligands fixed. This is a tetragonal elongation. Any orbital with electron density along $z$ feels reduced repulsion from the departing ligands and drops in energy; any orbital with density concentrated in the $xy$-plane feels increased relative repulsion from the four ligands that are now proportionally "closer" (or at least unchanged while their axial competitors have weakened) and rises. Four orbitals split apart from the original two levels. From $e_g$: $d_{z^2}$ (strong axial lobes, big $\pm z$ density) drops steeply; $d_{x^2-y^2}$ (density concentrated in-plane, pointing directly at the four remaining ligands) rises further still, ending up the highest orbital of all. From $t_{2g}$: $d_{xy}$, whose lobes lie in the $xy$-plane between the ligand directions, rises somewhat because equatorial ligand repulsion is now felt more acutely without competing axial destabilization elsewhere in the diagram; $d_{xz}$ and $d_{yz}$, which have equal density split between the $z$-axis and the $xy$-plane, drop because the axial component of their density loses its repulsive partner and remains a degenerate pair (by the surviving $C_4$ symmetry, $xz$ and $yz$ stay equivalent under reflection through the plane). Carried to the limit of complete axial ligand removal (idealized square-planar, $D_{4h}$ symmetry), the four-level pattern is

read low to high energy. Crucially, note that $d_{z^2}$ drops so far that it can fall below $d_{xy}$ — a nontrivial, ligand-dependent detail that some textbooks omit; the ordering of $d_{z^2}$ relative to $d_{xy}$ can invert depending on the specific ligand's $\sigma$- and $\pi$-donor/acceptor character, but the identity of the extremes — $d_{xz}/d_{yz}$ lowest, $d_{x^2-y^2}$ highest by a wide margin — is essentially universal for real square-planar d⁸ complexes.

The second key physical fact is electron count. In a strong-field square-planar d⁸ ion, the eight d electrons fill the lower four orbitals — $d_{xz}, d_{yz}, d_{z^2}, d_{xy}$ — completely, each doubly occupied, and leave $d_{x^2-y^2}$ empty. Because $d_{x^2-y^2}$ sits far above the rest, this is a large HOMO–LUMO gap, all electrons are paired, and the complex is diamagnetic. Compare an octahedral d⁸ ion: it must place two electrons in $e_g$ (three in each $t_{2g}$ orbital, then one in each $e_g$ orbital by Hund's rule for a modest $\Delta_o$, or paired in one $t_{2g}$... actually for d⁸ in $O_h$, six electrons fill $t_{2g}$ completely and two electrons fill $e_g$ singly, giving $t_{2g}^6 e_g^2$, invariably paramagnetic with two unpaired electrons regardless of field strength, since $e_g$ has only two orbitals to distribute two electrons into with parallel spin preferred). Square-planar geometry is the only common geometry in which a d⁸ ion can achieve full electron pairing in exactly four low-lying orbitals with a genuinely large gap to the fifth — this is the physical root of d⁸ ions' famous preference for square-planar coordination.

The correlation between octahedral and square-planar splitting parameters is normally expressed relative to $\Delta_o$, the octahedral splitting parameter, using empirical/semi-quantitative relations first systematized for tetragonally distorted complexes:

where:

• $\Delta_o$ is the parent octahedral splitting for the same metal–ligand pair (a hypothetical or actual six-coordinate reference),
• energies are measured relative to the spherical (barycenter) average, so the weighted mean over all five orbitals is zero,
• these particular coefficients come from angular-overlap/point-charge treatments of the limiting square-planar geometry (axial ligands fully removed) and vary modestly with the specific ligand and metal.

Reading this: the barycenter rule from Octahedral Crystal Field Splitting still applies — the sum of all five orbital energies, weighted by degeneracy, is zero, because crystal field splitting redistributes energy without changing the total (it is not a real bond-energy input, just an electrostatic reshuffling of the $d$-orbital manifold). Check the weighted sum with the coefficients above:

which confirms internal consistency (the barycenter condition holds exactly with these coefficients). The square-planar splitting parameter analogous to $\Delta_o$ or $\Delta_t$ is usually defined as the HOMO–LUMO-type gap between $d_{xy}$ (or, in some conventions, the whole lower set) and $d_{x^2-y^2}$:

where:

• $\Delta_{sp}$ is loosely called $\Delta_1$ in older spectroscopic literature on square-planar d⁸ complexes,
• the relation shows that even though square-planar has fewer ligands than octahedral, the top orbital $d_{x^2-y^2}$ is pushed up almost as far above the mean as $\Delta_o$ itself, because all four ligand point-charges/dipoles are now concentrated in-plane with no axial ligands to "share" the repulsive burden.

The overall crystal field stabilization energy (CFSE) for a configuration in square-planar geometry is computed exactly as in octahedral or tetrahedral cases: sum each electron's orbital energy (in units of $\Delta_o$, using the coefficients above) using the observed spin state, and compare to zero (no CFSE, spherical field).

Example 1 (foundational): construct the CFSE of low-spin square-planar Ni(CN)₄²⁻ and confirm diamagnetism. Ni²⁺ is d⁸ ($[Ar]3d^8$). Cyanide is an exceptionally strong-field ligand (very high in the spectrochemical series), driving Ni²⁺ into a square-planar, low-spin, diamagnetic configuration rather than the tetrahedral or octahedral alternatives typical of weaker-field Ni²⁺ complexes. Fill the four square-planar orbitals from lowest energy up, obeying the Pauli principle, with 8 electrons:

All eight electrons occupy four completely filled, spin-paired orbitals. There are zero unpaired electrons — the complex must be diamagnetic, exactly as observed experimentally for [Ni(CN)₄]²⁻ (measured magnetic moment $\mu \approx 0$, versus $\mu \approx 2.83\ \mu_B$ expected for two unpaired electrons in a hypothetical paramagnetic alternative). Using the correlation coefficients, the CFSE is (from the derivation above) $-2.56\,\Delta_o$, a large stabilization consistent with cyanide's very large $\Delta_o$ for Ni²⁺ (on the order of 30,000–34,000 cm⁻¹ range for related strong-field Ni(II)–CN systems, though the exact experimental $\Delta_o$ for a hypothetical six-coordinate Ni(CN)₆⁴⁻ reference is not directly measurable since that species does not form; the comparison is conceptual/thermodynamic rather than a literal spectroscopic subtraction).

Limiting case check: if cyanide were instead a very weak-field ligand (hypothetically, keeping only the geometry fixed), $\Delta_o \to$ small, and the pairing-energy penalty computed in the derivation would dominate, making square-planar geometry electronically unfavorable relative to tetrahedral — correctly predicting that weak-field Ni²⁺ complexes (e.g., [NiCl₄]²⁻, [Ni(H�2O)₆]²⁺) are not square-planar, but tetrahedral or octahedral respectively, both paramagnetic with two unpaired electrons.

Example 2 (applied): why PtCl₄²⁻ is square-planar and diamagnetic while the isoelectronic-in-charge NiCl₄²⁻ is tetrahedral and paramagnetic. Both ions are d⁸ metal dichlorides with four chloride ligands, yet they adopt opposite geometries: [PtCl₄]²⁻ is square-planar and diamagnetic; [NiCl₄]²⁻ is tetrahedral and paramagnetic (two unpaired electrons, $\mu\approx 2.9$–3.2 $\mu_B$ typically observed for tetrahedral Ni(II)). The geometric divergence, despite an identical ligand and identical formal d-electron count, is a direct test of the CFSE-versus-pairing-energy competition derived above. Two factors both push in the same direction. First, $\Delta_o$ (and by extension $\Delta_{sp}$, roughly $1.05\Delta_o$ by the relation derived above) scales up sharply down a group: for the same ligand set, $\Delta_o(\text{Pt}^{2+}) \gg \Delta_o(\text{Ni}^{2+})$, typically by a factor of roughly 1.5–2× per row descended (3d → 4d → 5d), because the more diffuse 5d orbitals overlap ligand orbitals more strongly and 5d orbitals are less contracted by electron–electron repulsion. Second, the mean pairing energy $P$ decreases down a group (heavier d-block metals have smaller interelectronic repulsion per orbital due to more diffuse orbitals), so the pairing-energy penalty in the derivation's Step 5, $\Delta E_{sp-oh} = -1.36\Delta_o + P$, becomes overwhelmingly negative (square-planar favored) for Pt²⁺: $\Delta_o$ is large and $P$ is comparatively small. For Ni²⁺ (3d), $\Delta_o$ from chloride (a weak/moderate-field ligand low in the spectrochemical series) is small, while $P$ for the compact 3d orbitals is comparatively large, so $-1.36\Delta_o + P > 0$: pairing costs dominate, and tetrahedral geometry — which has an even smaller CFSE magnitude but avoids the extra forced pairing altogether, and also minimizes ligand–ligand steric repulsion for only four large chloride ions — wins out. This is a clean illustration of how the same qualitative formula, applied with correctly scaled parameters across a group, flips the predicted ground-state geometry.

Example 3 (exam/competition level): estimate $\Delta_{sp}$ for [PtCl₄]²⁻ from its d–d absorption spectrum and check internal consistency with the correlation formula. Square-planar Pt(II) complexes typically show three spin-allowed d–d transitions (from filled $d_{xz,yz}$, $d_{z^2}$, $d_{xy}$ up to the empty $d_{x^2-y^2}$ level, in $D_{4h}$ symmetry with appropriate orbital-symmetry labels $^1A_{1g}\to {}^1A_{2g}, {}^1B_{1g}, {}^1E_g$ manifold, though d–d bands in Pt(II) are often obscured by intense charge-transfer absorption). Suppose (for a representative strong-field square-planar Pt(II) chloride-type complex) the highest-energy, most clearly assigned spin-allowed d–d band — corresponding to the $d_{xy}\to d_{x^2-y^2}$ transition, i.e. $\Delta_{sp}$ itself — is observed at $\tilde\nu = 20{,}300\ \text{cm}^{-1}$. Using the correlation-diagram relation derived above, $\Delta_{sp}\approx 1.05\,\Delta_o$, back out the equivalent octahedral splitting parameter that would characterize a hypothetical six-coordinate PtCl₆⁴⁻-type reference with the same metal–ligand bond character:

Compare this to the experimentally known $\Delta_o$ for the genuinely octahedral hexachloroplatinate(IV) ion, [PtCl₆]²⁻ (Pt in the +4 oxidation state, admittedly a different oxidation state and thus not a strictly rigorous comparison, but illustrative of the right order of magnitude): $\Delta_o$ for Pt(IV)–Cl is reported in the range of roughly 20,000–23,000 cm⁻¹ for octahedral hexahalide platinum(IV) complexes. The back-calculated value of 19,300 cm⁻¹ sits close to, if somewhat below, this range — a reasonable outcome given that Pt(II) has one less positive charge than Pt(IV) and would be expected to produce a somewhat smaller $\Delta_o$ for the same ligand, everything else equal (lower oxidation state → weaker ligand field, consistent with periodic trends established for Octahedral Crystal Field Splitting).

Consistency/limiting check: if $\Delta_{sp}$ were instead measured as exactly $1.05\Delta_o$ with $\Delta_o$ taken as a round number, say $\Delta_o = 20{,}000\ \text{cm}^{-1}$, the predicted $\Delta_{sp}$ would be $21{,}000\ \text{cm}^{-1}$ — demonstrating that the correlation coefficient of $1.05$ is a genuinely small perturbation on $\Delta_o$ (square-planar splitting is comparable in magnitude to octahedral splitting for the same ligands, not dramatically larger or smaller), which is why d–d spectra of square-planar and octahedral complexes of the same metal/ligand family often fall in overlapping spectral windows.

Square-planar splitting itself is not attributed to a single discoverer in the way that, say, a named reaction is. It emerged as a natural corollary once Bethe's crystal-field group theory was combined with the specific geometric case of a $D_{4h}$ point group, and once inorganic chemists in the 1950s–60s (notably including Orgel's widely used qualitative treatments of transition-metal stereochemistry) explained the observed magnetism and color of square-planar Ni(II), Pd(II), Pt(II), and Au(III) complexes with exactly this orbital-splitting picture. The experimental anchor — diamagnetic square-planar Ni(CN)₄²⁻ versus paramagnetic tetrahedral NiCl₄²⁻, both d⁸ — became one of the standard textbook demonstrations that geometry, not merely electron count, determines magnetism in the d-block, cementing square-planar crystal field theory's place in the coordination chemistry canon.