Selection Rules for Electronic Transitions

Selection-rule intensity analysis is a routine diagnostic tool in modern inorganic and bioinorganic chemistry, not a historical curiosity. In coordination chemistry teaching labs and research, the intensity of a $d$-$d$ band is often the fastest experimental clue to geometry: a strongly colored ($\varepsilon > 100$) $\text{Co}^{2+}$ or $\text{Ni}^{2+}$ solution signals tetrahedral or low-symmetry coordination, while a pale one signals octahedral coordination, without needing a crystal structure. This logic underlies rapid solution-phase geometry assignments for newly synthesized complexes before X-ray data are available. In bioinorganic chemistry, metalloprotein active sites are frequently studied by exactly this reasoning: the intensely colored blue copper proteins (plastocyanin, azurin) owe their striking $\varepsilon \sim 3000$–$6000\ \text{M}^{-1}\text{cm}^{-1}$ absorption near 600 nm not to a $d$-$d$ transition (which would be weak) but to a fully allowed cysteine-thiolate-to-Cu(II) LMCT transition — precisely the kind of Laporte- and spin-allowed process foreshadowed here and developed further under Charge-Transfer Spectra: LMCT and MLCT. Distinguishing that intense LMCT band from the much weaker underlying $d$-$d$ bands of the same copper center is a direct, everyday application of the intensity hierarchy built from these two rules. In materials and photophysics, lanthanide luminescence (used in LED phosphors, MRI contrast agent design, and bioimaging labels) depends critically on the fact that $f$-$f$ transitions are Laporte-forbidden and only weakly vibronically/$J$-mixing-allowed, giving lanthanides their famously sharp, long-lived, low-intensity emission lines — a feature deliberately engineered around (via sensitizing "antenna" ligands that absorb strongly and transfer energy) in commercial lanthanide-based phosphors and time-resolved fluorescence immunoassays. In transition-metal photocatalysis and solar energy research, spin-forbidden and Laporte-forbidden $d$-$d$ excited states are often long-lived (because they cannot radiatively decay quickly either), making them useful reservoirs of excitation energy — the same forbiddenness that makes absorption weak also makes the corresponding excited-state lifetime long, a duality exploited in ruthenium and iridium photosensitizer design. Finally, in analytical and forensic chemistry, the extinction coefficients discussed here set the detection limits and required path lengths for colorimetric metal-ion assays, directly translating a symmetry argument into a practical instrumental design constraint.

:::quiz[Which of the following best explains why $[\text{Ti(H}_2\text{O})_6]^{3+}$ shows a weak but clearly visible $d$-$d$ band rather than exactly zero absorbance?]

• The Laporte rule does not apply to $d^1$ ions
• Vibronic coupling from asymmetric vibrations transiently removes the center of symmetry, mixing in allowed character ✓
• Water ligands are always Laporte-allowed donors
• The spin selection rule is violated in this complex

:::quiz[In the transition-dipole factorization $\mu_{if} = [\text{spatial integral}]\times[\text{spin overlap integral}]$, why does $\hat\mu$ not appear in the spin integral?]

• Because $\hat\mu$ is always zero for transition metals
• Because the electric dipole operator is built from position operators only and does not act on spin coordinates ✓
• Because spin is not a real physical property
• Because Laporte's rule already accounts for spin

Ask the tutor: How does Jahn–Teller distortion interact with the Laporte rule to produce fine structure within a single "forbidden" d-d band?

An electronic absorption band's intensity is controlled by the transition dipole moment,

where $\hat{\mu} = -e\sum_i \vec r_i$ is the electric dipole operator (a vector operator, odd under inversion, and spin-independent). The measured molar absorptivity $\varepsilon$ is proportional to $|\vec{\mu}_{if}|^2$ (more precisely to the oscillator strength $f \propto \nu |\vec\mu_{if}|^2$). If the integral is forced to vanish by symmetry, the transition is "forbidden" in the idealized limit; if it is symmetry-allowed, the transition is "allowed" and can carry oscillator strength approaching unity ($\varepsilon \sim 10^4$–$10^5$). Two independent constraints can kill this integral. First, parity: if the molecule has a center of inversion, every wavefunction transforms as either gerade ($g$, symmetric under inversion) or ungerade ($u$, antisymmetric). Since $\hat\mu$ is itself ungerade, the integrand $\psi_f^* \hat\mu \psi_i$ is odd under inversion unless $\psi_i$ and $\psi_f$ have opposite parity. A transition between two states of the same parity ($g\to g$ or $u\to u$) integrates identically to zero over all space — this is the Laporte rule. All $d$ orbitals are $g$ (they transform as gerade combinations under inversion, since $d$ orbitals have $\ell = 2$, even), so any purely $d$-$d$ transition in a centrosymmetric complex is Laporte-forbidden. Second, spin: $\hat\mu$ contains no spin operators at all — it acts only on spatial coordinates. Writing a total wavefunction as $\psi = \phi(\text{space})\,\chi(\text{spin})$, the transition dipole factors as

The spin overlap integral $\int \chi_f^*\chi_i\,d\sigma$ vanishes unless the initial and final spin functions are identical, which (for well-defined total spin states) requires $\Delta S = 0$ and $\Delta M_S = 0$. This is the spin selection rule. A transition from a high-spin ground state to a different-spin excited state — e.g. a spin-flip within a $d^5$ ion — is forbidden by this factor regardless of what the orbital part does. The escape routes matter as much as the rules. Real molecules vibrate, and an asymmetric (ungerade) vibrational distortion instantaneously removes the center of symmetry, mixing a small amount of opposite-parity character into a nominally gerade electronic state — this is vibronic coupling, and it is why octahedral $d$-$d$ bands are weak but not literally zero ($\varepsilon \sim 1$–$100$, not $\varepsilon = 0$). Real molecules also have spin–orbit coupling, $H_{SO} = \zeta(\vec r)\,\vec L\cdot\vec S$, which mixes states of different $S$ (since $\vec L \cdot \vec S$ does not commute with $S^2$ alone) and gives spin-forbidden transitions a small but nonzero intensity, scaling steeply with atomic number.

The oscillator strength of an electronic transition is

where: $f$ is the dimensionless oscillator strength (theoretical maximum $\approx 1$ for a fully allowed transition), $\nu$ is the transition frequency, $m_e$ and $e$ are the electron mass and charge, and $\vec\mu_{if}$ is the transition dipole moment integral defined above. This connects the abstract symmetry integral directly to the experimentally measured, integrated absorption band area (and hence to $\varepsilon_{\max}$).

Reading: $f$ is large only when $|\vec\mu_{if}|^2$ is large; every selection rule discussed here is really a statement about which terms make $\vec\mu_{if}$ zero, nearly zero, or fully nonzero by symmetry, and $f$ (hence $\varepsilon$) inherits that hierarchy directly. The group-theoretic version of the Laporte/orbital selection rule is that the transition is allowed only if the direct product of irreducible representations contains a component that transforms as $x$, $y$, or $z$:

where: $\Gamma_i, \Gamma_f$ are the irreducible representations of the initial and final electronic states, and $\Gamma_{\hat\mu}$ is the representation spanned by $(x,y,z)$ in the point group of the molecule. If the product does not contain the totally symmetric representation $A_1$ (or $A_{1g}$), the integral is forced to be exactly zero by symmetry, and the transition is rigorously forbidden at that order.

Reading: in $O_h$, $(x,y,z)$ transform as $T_{1u}$ (ungerade). Since $g \otimes u \otimes g = u$ never contains $A_{1g}$, any $g \to g$ transition — including all $d$-$d$ transitions — is forced to zero. This is the compact statement of the Laporte rule for centrosymmetric point groups, and it generalizes the parity argument to non-inversion symmetry operations too (mirror planes, rotation axes) for point groups that lack a true center of symmetry, like $T_d$. The spin selection rule is stated most simply as

where: $S$ is the total spin quantum number of the electronic state (multiplicity $= 2S+1$) and $M_S$ its projection. This is exact in the absence of any spin-dependent term in the Hamiltonian; it is only relaxed by spin–orbit coupling, which physically mixes small amounts of different-$S$ character into "pure" spin states.

Example 1 (foundational): confirming the Laporte- and spin-classification of the spin-allowed $d$-$d$ bands of $[\text{Ti(H}_2\text{O})_6]^{3+}$. $\text{Ti}^{3+}$ is $d^1$; in $O_h$ symmetry the single electron occupies $t_{2g}$ in the ground state ($^2T_{2g}$) and is promoted to $e_g$ in the excited state ($^2E_g$) by absorbing a photon near 20,300 cm$^{-1}$ (492 nm), giving the characteristic purple color. Check both rules: Spin rule: both states have $S = \tfrac12$ (doublet, one unpaired electron throughout) — $\Delta S = 0$, so the transition is spin-allowed. Laporte rule: both $t_{2g}$ and $e_g$ are built from $d$ orbitals, which are gerade. The product of state symmetries is

Since $g \otimes u \otimes g = u$ (gerade $\times$ ungerade $\times$ gerade is always ungerade), this product cannot contain $A_{1g}$ (totally symmetric, gerade), so the transition is Laporte-forbidden. Net classification: Laporte-forbidden, spin-allowed. Predicted intensity regime: $\varepsilon \sim 1$–$100\ \text{M}^{-1}\text{cm}^{-1}$. Experimentally, $\varepsilon_{\max} \approx 5\ \text{M}^{-1}\text{cm}^{-1}$ at the band maximum — squarely in the predicted range, and the band additionally shows a shoulder from a Jahn–Teller-split $e_g$ excited state, a refinement beyond the scope of the selection rule itself but consistent with weak, vibronically-allowed intensity.

Example 2 (applied): why tetrahedral $\text{CoCl}_4^{2-}$ is roughly 100× more intensely colored than octahedral $[\text{Co(H}_2\text{O})_6]^{2+}$. Both are high-spin $\text{Co}^{2+}$ ($d^7$, $S = 3/2$ quartet ground states), so both transitions of interest are spin-allowed ($\Delta S = 0$) in each geometry — the spin rule does not distinguish them. The difference is purely a Laporte/parity story. $[\text{Co(H}_2\text{O})_6]^{2+}$ is (to good approximation) $O_h$ symmetric — the aqua ligands sit at the vertices of an octahedron, and the complex possesses a true center of inversion at the metal. Its spin-allowed $^4T_{1g} \to ^4T_{1g}(P)$ transition near 19,400 cm$^{-1}$ is Laporte-forbidden by the same $g\otimes u \otimes g$ argument as Example 1, giving $\varepsilon_{\max} \approx 5$–$10\ \text{M}^{-1}\text{cm}^{-1}$ (pale pink). $\text{CoCl}_4^{2-}$ is $T_d$ symmetric — a tetrahedron has no center of inversion at all. There is no gerade/ungerade classification in $T_d$; metal $d$ orbitals split into $e$ and $t_2$ representations (not $e_g$/$t_{2g}$), and the relevant transition $^4A_2 \to {}^4T_1(P)$ near 15,000–16,000 cm$^{-1}$ has

which does contain $A_1$ (verified by standard $T_d$ character-table reduction), so the transition is not Laporte-forbidden in the strict sense — there is no parity operation to forbid it. The absence of an inversion center means $d$ orbitals can mix with $p$ orbitals on the metal (both $t_2$ in $T_d$), giving the transition genuine, static (not just vibronically-borrowed) allowed character. Experimentally $\varepsilon_{\max}(\text{CoCl}_4^{2-}) \approx 600\ \text{M}^{-1}\text{cm}^{-1}$ near 15,000 cm$^{-1}$ — about 100-fold more intense than the octahedral aqua complex, even though the electron count, spin state, and even the qualitative $d$-orbital promotion are entirely analogous. The sole structural difference — loss of the inversion center on going from $O_h$ to $T_d$ — accounts for the entire two-order-of-magnitude intensity jump. This is exactly why deep royal-blue tetrahedral cobalt complexes (used as a classroom color test for $\text{Co}^{2+}$ coordination geometry) look so much richer than the pale rose color of the aqua ion.

Example 3 (exam/competition level): rank-order and rationalize the intensities of four bands, then predict the effect of iodide coordination on a spin-forbidden band. Four absorption bands are reported for iron(III) and manganese(II) complexes: (a) $[\text{Fe(H}_2\text{O})_6]^{3+}$: very weak, sharp band at 12,600 cm$^{-1}$, $\varepsilon \approx 0.1$ (assigned $^6A_{1g} \to {}^4T_{1g}$). (b) $[\text{Mn(H}_2\text{O})_6]^{2+}$: extremely weak, extremely sharp band at 24,900 cm$^{-1}$, $\varepsilon \approx 0.01$ (assigned $^6A_{1g} \to {}^4E_g, {}^4A_{1g}$). (c) $[\text{FeCl}_4]^-$: LMCT band at 22,700 cm$^{-1}$, $\varepsilon \approx 5000$. (d) $[\text{OsCl}_6]^{2-}$: spin-forbidden band, $\varepsilon \approx 30$. Rank and rationalize. First classify each by both rules. (a) $\text{Fe}^{3+}$ is high-spin $d^5$ ($S = 5/2$, sextet ground state $^6A_{1g}$). The excited state $^4T_{1g}$ has $S = 3/2$ (quartet). $\Delta S = 1 \neq 0$: spin-forbidden. Also $A_{1g} \to T_{1g}$ is $g \to g$ in $O_h$: Laporte-forbidden. Doubly forbidden → extremely weak, consistent with $\varepsilon \approx 0.1$. (b) Same ion type, $\text{Mn}^{2+}$ high-spin $d^5$, ground state $^6A_{1g}$, excited states $^4E_g, {}^4A_{1g}$ also quartets. $\Delta S = 1$: spin-forbidden; $g \to g$: Laporte-forbidden. Doubly forbidden, and additionally these particular excited terms are orbitally non-degenerate in the sense that they arise from a $(t_{2g})^3(e_g)^2$ configuration with the same orbital occupation pattern as the ground term, so they are essentially independent of ligand-field strength (nearly horizontal on a Tanabe–Sugano diagram) and correspondingly razor-sharp and even weaker: $\varepsilon \approx 0.01$, among the weakest transitions in all of inorganic spectroscopy. (c) $[\text{FeCl}_4]^-$: this band is assigned as ligand-to-metal charge transfer (chloride lone pair → iron $d$), which is spin-allowed (no change in overall spin required for a one-electron transfer that preserves total $S$) and, critically, Laporte-allowed in the sense that it is not subject to a $g$-parity restriction at all — the initial state is predominantly ligand-centered ($u$ combinations of Cl lone pairs in $T_d$) and the final state metal-centered, and more importantly $T_d$ has no inversion center to forbid anything. Fully/near-fully allowed → $\varepsilon \approx 5000$, the largest of the four by orders of magnitude. (d) $[\text{OsCl}_6]^{2-}$: also spin-forbidden and Laporte-forbidden by the same formal argument as (a)/(b) (octahedral, $g\to g$, $\Delta S \neq 0$), yet its intensity, $\varepsilon \approx 30$, is roughly 100–1000× stronger than the analogous $3d$ spin-forbidden bands in (a) and (b). The reason is the spin–orbit coupling constant: $\zeta(\text{Os}^{3+/4+}) \sim 3000$–$4000\ \text{cm}^{-1}$ versus $\zeta(\text{Fe}^{3+}) \approx 460\ \text{cm}^{-1}$ and $\zeta(\text{Mn}^{2+}) \approx 300\ \text{cm}^{-1}$. Since the borrowed intensity in Step 5 of the second derivation scales as $(\zeta/\Delta E)^2$ in the observed intensity (moment squared), an order-of-magnitude increase in $\zeta$ translates into roughly two orders of magnitude increase in $\varepsilon$, exactly the pattern seen.

Overall rank: $\varepsilon(c) \gg \varepsilon(d) \gg \varepsilon(a) > \varepsilon(b)$, i.e., $5000 \gg 30 \gg 0.1 > 0.01$ — a span of nearly six orders of magnitude, entirely rationalized by two selection rules and one periodic trend (spin–orbit coupling strength). Predicted effect of iodide coordination on a spin-forbidden band such as (a) or (b): replacing $\text{H}_2\text{O}$ (a weak-field, light-atom donor) with a heavier halide donor increases spin–orbit coupling contributions felt by the metal through covalency and ligand spin–orbit character, and also increases vibronic activity through lower-frequency, larger-amplitude M–I stretches, so spin-forbidden bands are expected to gain modest additional intensity (though the dominant scaling with $Z^4$ still comes primarily from the metal center, not the ligand).

Limiting-case check 1 — vanishing vibrational amplitude. As $T \to 0\,\text{K}$, the thermal population of the ungerade vibrational modes that make vibronic coupling possible decreases (though even at $T=0$ the zero-point vibration persists, so intensity does not literally vanish). Experimentally, cooling octahedral $d$-$d$ spectra from room temperature toward liquid-helium temperature measurably decreases the intensity of vibronically-allowed $d$-$d$ bands (though a residual zero-point contribution remains), a direct experimental confirmation that vibronic coupling — not some fixed intrinsic transition moment — is the source of the observed intensity in Laporte-forbidden bands. This is the correct limiting behavior: it would be a red flag if a supposedly "vibronically borrowed" band's intensity were temperature-independent.

Limiting-case check 2 — heavy-atom limit for spin-forbidden bands. As $Z$ increases within a spin-forbidden series (e.g. Mn$^{2+}$ $\to$ Ru$^{3+}$ $\to$ Os$^{3+}$, all with comparably placed spin-forbidden ligand-field terms), $\zeta$ increases roughly as $Z^4$, and since borrowed intensity scales as $(\zeta/\Delta E)^2$, the observed molar absorptivity of spin-forbidden bands should increase roughly as $Z^8$ within an isostructural, isoelectronic series (modulated by how $\Delta E$, the energy gap to the spin-orbit-mixed state, also varies). This steep trend is exactly why spin-forbidden transitions, essentially undetectable for $3d$ ions, become routinely observable diagnostic bands for $4d/5d$ complexes — the correct qualitative limit, confirmed by comparing Mn(II) ($\varepsilon \sim 0.01$) to Os(III/IV) analogues ($\varepsilon \sim 10$–$50$).