Origin of Colour in Coordination Complexes

The direct descendants of this simple $t_{2g}/e_g$ picture are everywhere in modern materials and analytical chemistry. Gemstone color is a textbook crystal-field story: ruby is $\text{Al}_2\text{O}_3$ with a trace of $\text{Cr}^{3+}$ substituting for $\text{Al}^{3+}$ in an octahedral oxide environment, and its characteristic red color (and the basis of the first working laser, the ruby laser) comes directly from spin-allowed and spin-forbidden $d$-$d$ transitions of $\text{Cr}^{3+}$ split by the surrounding oxide ligand field; emerald is beryl with the same $\text{Cr}^{3+}$ ion in a slightly different (weaker-field) oxide environment, shifting $\Delta_o$ and hence the color from red to green. In biochemistry, the color changes of hemoglobin (bright red oxygenated vs. darker venous deoxygenated blood) and the intense green of chlorophyll both trace to d-d and charge-transfer/π-system transitions of metal centers (Fe and Mg respectively) whose ligand environment is tuned by the surrounding protein or porphyrin scaffold — exactly the same $\Delta_o$-tuning logic used here for simple aqua and ammine complexes. Industrial pigments — cobalt blue ($\text{CoAl}_2\text{O}_4$), chrome green, Prussian blue (an iron(II)/iron(III) cyanide charge-transfer pigment) — are engineered by choosing metal, oxidation state, and ligand/lattice environment specifically to place $\Delta_o$ (or a charge-transfer band) at the wavelength needed for the desired hue. Analytically, colorimetric and spectrophotometric assays throughout chemistry, environmental monitoring, and clinical laboratories use exactly this principle in reverse: measure the absorption spectrum of a metal-ligand complex (formed deliberately, e.g., Fe²⁺ with 1,10-phenanthroline giving an intense red complex used for iron determination) and use the known relationship between concentration and absorbance (the Beer–Lambert law) to quantify trace metal ions. Understanding why the complex is colored — that the ligand field must split the d orbitals by a visible-range $\Delta_o$, and that the transition must be at least partially allowed — is what lets analytical chemists choose or design ligands that produce strongly colored, sensitive, selective complexes for a given metal ion.

Ask the tutor: How do the Laporte and spin selection rules determine why some d-d bands are so much weaker than others, even when they're allowed by energy?

Start with a bare transition-metal ion floating in a vacuum, surrounded by nothing. Its five d orbitals ($d_{xy}, d_{xz}, d_{yz}, d_{x^2-y^2}, d_{z^2}$) are degenerate — same energy, same shape category, just different orientations in space. A spherically symmetric electric field (imagine smearing the ligands' negative charge uniformly over a sphere around the metal) raises all five orbitals' energy equally, by electron-electron repulsion with the surrounding charge, but does not split them apart. No splitting means no energy gap between d orbitals, so there is no photon that could drive a d-to-d electronic transition. This is why a free Cu²⁺ ion in the gas phase or a d⁰ ion like Ca²⁺ shows no d-d color signature — one has no split levels to jump between with the right spacing, the other has no d electrons at all. Now let the ligands localize onto six specific points — the vertices of an octahedron along $\pm x, \pm y, \pm z$. The electron density of the ligands (lone pairs, in the classic point-charge/crystal-field picture) now interacts unevenly with the five d orbitals, because the d orbitals themselves point in different directions. The $d_{x^2-y^2}$ and $d_{z^2}$ orbitals (grouped as $e_g$) point directly at the incoming ligands along the axes; the $d_{xy}, d_{xz}, d_{yz}$ orbitals (grouped as $t_{2g}$) point between the axes, into the octahedral faces. An electron in an $e_g$ orbital experiences much stronger repulsion from the ligand lone pairs than one in a $t_{2g}$ orbital, so $e_g$ rises in energy and $t_{2g}$ falls, relative to the (hypothetical) spherical-field average. The gap between them is defined as $\Delta_o$ (subscript $o$ for octahedral), and by the "center of gravity" (barycenter) rule the total energy is conserved: $t_{2g}$ drops by $0.4\,\Delta_o$ and $e_g$ rises by $0.6\,\Delta_o$ relative to the barycenter, so that $3(-0.4\Delta_o) + 2(0.6\Delta_o) = 0$.

The color we perceive is governed by subtraction, not addition. White light contains all visible wavelengths; if the complex absorbs a band centered, say, in the green-yellow region (~500–560 nm) to drive the $t_{2g} \to e_g$ transition, the light transmitted (or reflected) toward the eye is deficient in that band and rich in everything else. The eye and brain interpret that remaining mixture as the complementary color on a standard color wheel — absorbed green-yellow light produces a perceived violet/purple solution, which is exactly the classic case of $[\text{Ti(H}_2\text{O})_6]^{3+}$. Two more geometric cases matter and are worth sketching qualitatively before the equations: tetrahedral fields and the Jahn–Teller-distorted octahedron. In a tetrahedral field, the splitting pattern inverts — three orbitals ($t_2$: $d_{xy}, d_{xz}, d_{yz}$) rise and two ($e$: $d_{z^2}, d_{x^2-y^2}$) fall, because in $T_d$ symmetry none of the d orbitals point directly at a ligand, but the $t_2$ set points closer to the tetrahedral vertices than the $e$ set. The splitting $\Delta_t$ is intrinsically smaller than $\Delta_o$ for the same metal-ligand pair (roughly $\Delta_t \approx \tfrac{4}{9}\Delta_o$), because there are only four ligands instead of six, and none sit exactly on an orbital lobe. Tetrahedral complexes of a given metal/ligand combination therefore tend to absorb at lower energy (longer wavelength) than the corresponding octahedral complex — one visible manifestation is that tetrahedral Co(II) complexes (e.g., $[\text{CoCl}_4]^{2-}$) are often intensely colored blue, distinctly different in hue and intensity from the pink of octahedral $[\text{Co(H}_2\text{O})_6]^{2+}$.

The Jahn–Teller effect adds a further wrinkle relevant to color: certain electron configurations in an octahedral field (notably $d^9$, high-spin $d^4$, and low-spin $d^7$) are unevenly filled in a way that makes a perfectly symmetric octahedron unstable. The complex distorts — typically two trans bonds elongate — which splits $e_g$ and $t_{2g}$ further into sub-levels. This is why $\text{Cu}^{2+}$ (d⁹) complexes often show broad, sometimes asymmetric or split absorption envelopes rather than one clean symmetric peak: the “d-d band” is really an overlapping pair of transitions from the Jahn–Teller-split levels.

The central relationship links the size of the ligand-field splitting to the wavelength (or wavenumber) of the photon absorbed:

where:

• $\Delta_o$ is the octahedral crystal-field splitting energy (matches the $t_{2g}\to e_g$ gap for a $d^1$ or simple case),
• $h$ is Planck's constant ($6.626\times10^{-34}\ \text{J s}$),
• $c$ is the speed of light ($2.998\times10^{8}\ \text{m/s}$),
• $\nu$ is the frequency and $\lambda$ the wavelength of the absorbed photon,
• $\tilde\nu = 1/\lambda$ is the wavenumber, conventionally reported in $\text{cm}^{-1}$, the practical spectroscopic unit.

Reading: the wavelength at the absorption maximum in a UV-vis spectrum is a direct, back-of-envelope measurement of $\Delta_o$ (for the simplest one-electron cases; multi-electron ions need term-symbol / Tanabe-Sugano corrections, which is where the Spectrochemical Series treatment goes further). A larger $\Delta_o$ means absorption shifts to shorter wavelength (bluer light absorbed, since $\tilde\nu \propto \Delta_o$), and the observed color shifts toward the complementary color of blue — typically toward orange/red/yellow appearances. The second core relationship is the barycenter (center-of-gravity) rule, which fixes how the octahedral splitting is partitioned between the two levels:

where:

• the coefficients $3$ and $2$ are the orbital degeneracies of $t_{2g}$ and $e_g$ respectively,
• $-0.4\,\Delta_o$ (often written $-\tfrac{2}{5}\Delta_o$) is the stabilization energy per electron in $t_{2g}$,
• $+0.6\,\Delta_o$ (written $+\tfrac{3}{5}\Delta_o$) is the destabilization energy per electron in $e_g$,
• the sum equals zero because splitting redistributes energy without changing the average (barycenter) orbital energy.

Reading: this identity is what lets you compute crystal field stabilization energies (CFSE) for any $d^n$ configuration once you know how electrons are distributed between $t_{2g}$ and $e_g$, and it is the reason the $t_{2g}$ set is disproportionately stabilized (lower by more than half the total gap) relative to how much $e_g$ is destabilized — a consequence purely of the 3:2 degeneracy ratio.

Example 1 — Foundational: the $d^1$ ion $[\text{Ti(H}_2\text{O})_6]^{3+}$. Ti³⁺ has the configuration $[\text{Ar}]3d^1$. In the octahedral aqua complex, that single electron occupies the lower $t_{2g}$ set in the ground state. The only possible electronic transition is $t_{2g}^1 e_g^0 \to t_{2g}^0 e_g^1$, and because there is exactly one electron, there is no interelectronic repulsion to complicate the picture — this is the textbook-clean case. The measured absorption maximum is $\lambda_{\max} \approx 493\ \text{nm}$ (broad band, actually a bit asymmetric due to Jahn–Teller distortion in the $e_g^1$ excited state, but the peak position is what we use for $\Delta_o$). Converting:

So $\Delta_o \approx 20{,}300\ \text{cm}^{-1} \approx 243\ \text{kJ/mol}$ (as computed in the derivation above). The absorbed light is green-blue (~493 nm sits at the blue-green boundary); the complementary color transmitted is red-violet/purple, and indeed $[\text{Ti(H}_2\text{O})_6]^{3+}$ solutions are a recognizable violet-purple. Limiting case: if we imagine artificially shrinking $\Delta_o$ toward zero (e.g., an extremely weak-field, highly ionic hypothetical ligand set), $\lambda_{\max} \to \infty$ — the transition drifts into the far infrared and the complex would appear colorless (or white/pale) because no visible photon is absorbed, even though the $d$-orbital splitting is nonzero. This confirms that a nonzero $\Delta_o$ is necessary but not sufficient for visible color — the gap specifically has to land in the visible window.

Example 2 — Applied: comparing $[\text{Ni(H}_2\text{O})_6]^{2+}$ (green) and $[\text{Ni(NH}_3)_6]^{2+}$ (blue/violet) to demonstrate the spectrochemical shift quantitatively. Ni²⁺ is $d^8$, octahedral high-spin in both cases (Ni²⁺ octahedral complexes are essentially always high-spin regardless of ligand field strength, since $d^8$ has only one possible high-spin arrangement: $t_{2g}^6 e_g^2$). Three spin-allowed d-d bands are typically observed for $d^8$ octahedral Ni(II), but the diagnostic one for this comparison is the lowest-energy band, roughly assignable (via Tanabe-Sugano correlation) to $\Delta_o$ itself. For $[\text{Ni(H}_2\text{O})_6]^{2+}$, the relevant band sits near $\tilde\nu_1 \approx 8{,}500\ \text{cm}^{-1}$ ($\lambda \approx 1{,}176\ \text{nm}$, actually just outside the visible, in the near-IR) but a second band near $15{,}400\ \text{cm}^{-1}$ ($\lambda \approx 649\ \text{nm}$, red) and a third near $\sim 25{,}300\ \text{cm}^{-1}$ ($\lambda \approx 395\ \text{nm}$, violet) fall in or near the visible, together carving out absorption in the red and violet-blue while leaving green/yellow relatively transmitted — consistent with the classic pale green color of aqueous Ni²⁺. For $[\text{Ni(NH}_3)_6]^{2+}$, ammonia is higher in the spectrochemical series than water, so all three bands shift to higher energy (shorter wavelength): the analogous bands appear near roughly $10{,}750\ \text{cm}^{-1}$, $17{,}500\ \text{cm}^{-1}$, and $28{,}200\ \text{cm}^{-1}$. Converting the middle band:

compared with $649\ \text{nm}$ for the aqua complex — a clear blue-shift of about 78 nm on replacing water by ammonia, exactly as the spectrochemical series predicts ($\text{NH}_3$ ranks higher than $\text{H}_2\text{O}$). This shift moves the strongest absorbed region from red/orange toward yellow-orange, and the transmitted/perceived color of the ammine complex shifts from pale green toward blue-violet, matching what is observed experimentally when concentrated aqueous ammonia is added to a nickel(II) solution. Limiting case check: if instead of ammonia we used a much weaker-field ligand than water (say, a hypothetical very weak halide-like donor), all three bands would red-shift further; if the shift were large enough to push even the second (currently red-absorbing) band into the near-infrared, essentially no visible photon would be absorbed strongly, and the complex would look nearly colorless or very pale despite having a full octahedral $d^8$ splitting — again showing that "has split d orbitals" and "is visibly colored" are related but not automatically identical statements.

Example 3 — Exam/competition level: rationalizing why $[\text{Mn(H}_2\text{O})_6]^{2+}$ is almost colorless (very pale pink) despite Mn²⁺ having five d electrons. Mn²╫ is $d^5$. In a weak field like water, Mn²⁺ is high-spin: $t_{2g}^3 e_g^2$, with all five electrons unpaired (each of the five d orbitals singly occupied). Consider what a "spin-allowed" d-d transition would require: promoting an electron from $t_{2g}$ to $e_g$ while keeping total spin multiplicity unchanged. But every orbital is already singly occupied with parallel spin (the ground term is $^6A_{1g}$, a sextet — all five electrons unpaired, $S = 5/2$). Any transition that moves an electron between orbitals within the $d^5$ manifold necessarily pairs up two electrons somewhere (since there are exactly five orbitals and five electrons, moving one electron forces a doubly-occupied orbital and a spin flip), converting the sextet ground state into a quartet ($S=3/2$) excited state. This means every d-d transition available to high-spin $d^5$ is spin-forbidden — it requires a change in total spin multiplicity, which quantum mechanically requires spin-orbit coupling to occur at all, making these transitions extremely weak (typically molar absorptivities $\varepsilon < 1\ \text{L mol}^{-1}\text{cm}^{-1}$, compared to normal spin-allowed d-d bands with $\varepsilon \sim 1$–$100\ \text{L mol}^{-1}\text{cm}^{-1}$, and to fully-allowed charge-transfer bands with $\varepsilon \sim 1{,}000$–$50{,}000\ \text{L mol}^{-1}\text{cm}^{-1}$). This is precisely why aqueous Mn²⁺ solutions are almost colorless — the very faint pink observed corresponds to $\varepsilon$ values around $0.01$–$0.04\ \text{L mol}^{-1}\text{cm}^{-1}$, orders of magnitude weaker than typical spin-allowed transitions such as those of Ni²⁺ or Co²⁺. This case foreshadows the full Selection Rules for Electronic Transitions treatment (which formalizes the spin-selection rule $\Delta S = 0$ alongside the Laporte parity rule), but the qualitative reasoning — count the electrons, check whether any rearrangement within $t_{2g}/e_g$ can occur without flipping a spin — is fully derivable from the crystal field diagram alone, without invoking group theory machinery. Limiting case check: contrast this with high-spin $d^5$ Fe³⁺, which is isoelectronic with Mn²⁺ and shows the identical spin-forbidden weak-color behavior (pale, near-colorless $[\text{Fe(H}_2\text{O})_6]^{3+}$ in strongly acidic solution — the yellow/brown color usually seen for iron(III) in less acidic solution is instead due to hydrolysis products and charge-transfer bands, not this d-d manifold). This independently confirms that the spin-forbidden argument is a property of the $d^5$ high-spin electron count itself, not an accident specific to manganese.

Crystal field theory as Bethe formulated it was initially a tool for solid-state physicists studying paramagnetic ions in ionic crystals (rare-earth and transition-metal-doped salts), not a mainstream chemical theory. It took roughly two decades, and the parallel efforts of chemists such as Leslie Orgel and John Van Vleck (who helped connect the electrostatic model to magnetic susceptibility measurements), before crystal field theory was widely adopted by coordination chemists to explain the colors, magnetism, and thermodynamic stabilities of transition-metal complexes in the 1950s. The theory was later refined into ligand field theory (adding covalency/molecular-orbital character to the purely ionic picture), but the core $t_{2g}/e_g$ splitting diagram, the barycenter rule, and the $\Delta_o$ notation all trace directly back to Bethe's original 1929 electrostatic-symmetry analysis.