Fluorescence and Phosphorescence: The Quantum Picture

Organic light-emitting diodes (OLEDs) are the most direct industrial descendant of this physics. Early fluorescent OLEDs were capped at 25% internal quantum efficiency because electrical excitation statistically produces triplets three times as often as singlets, and those triplets were simply wasted as heat via nonradiative $T_1 \to S_0$ decay. Phosphorescent OLEDs solved this by using heavy-metal (Ir, Pt) organometallic emitters whose enormous spin-orbit coupling makes $k_p$ fast enough to compete directly with nonradiative loss, harvesting both singlet and triplet excitons for light — now standard in commercial displays. A further refinement, thermally activated delayed fluorescence (TADF) emitters, engineers a very small $S_1$–$T_1$ gap so that thermal energy can reverse intersystem crossing, recycling triplets back into the fast fluorescent channel without needing a heavy metal at all. Photodynamic therapy in oncology exploits the triplet channel directly: a photosensitizer drug is excited, undergoes efficient ISC (often aided by heavy-atom-like or extended-conjugation design) to a long-lived triplet, and then transfers that energy to molecular oxygen, generating cytotoxic singlet oxygen ($^1\Delta_g$) that destroys tumor tissue locally. Fluorescent protein and small-molecule dye photophysics governs the entire toolkit of fluorescence microscopy, FRET-based biosensors, and single-molecule localization microscopy, where quantitative knowledge of $k_f$, $k_{IC}$, and $k_{ISC}$ (including photobleaching pathways that often proceed through the reactive triplet state) determines probe brightness, blinking behavior, and photostability. Glow-in-the-dark pigments (strontium aluminate phosphors, older zinc sulfide formulations) are solid-state analogs where trap states play the role of a long-lived reservoir, conceptually paralleling the molecular triplet but governed by defect-level physics rather than pure spin-orbit coupling. Finally, atmospheric and combustion chemistry rely on the same singlet-triplet framework to understand singlet oxygen generation, triplet-sensitized photochemistry in the upper atmosphere, and phosphorescence-based oxygen sensors used in biomedical and environmental monitoring, since $O_2$ quenching of phosphorescence is sensitive enough to measure dissolved oxygen concentration optically.

Ask the tutor: How would the rate expressions change for a d-d or metal-to-ligand-charge-transfer transition in a transition-metal complex, where spin-orbit coupling is already large in the ground state?

Consider a molecule promoted by a photon from the ground singlet state $S_0$ to some excited singlet $S_n$ ($n \geq 1$), typically by absorption into a higher vibronic level of an excited electronic state. What happens next is a race down a ladder of electronic and vibrational levels, and every rung of that ladder is governed by quantum mechanical selection rules. The first and fastest event is vibrational relaxation: the molecule sheds excess vibrational energy to the solvent bath (collisional or intramolecular vibrational redistribution) on a $10^{-13}$–$10^{-11}$ s timescale, landing in the vibrationless level of $S_n$. If $n > 1$, internal conversion (IC) — a nonradiative, spin-allowed transition between states of the same spin multiplicity but different electronic character — rapidly cascades the population down to $S_1$, again typically in picoseconds. This cascade is the microscopic content of Kasha's rule: because the energy gaps between higher excited states ($S_2 \to S_1$, etc.) are usually small compared to the $S_1 \to S_0$ gap, and because nonradiative rates fall off steeply with increasing energy gap (the "energy-gap law"), internal conversion from $S_2, S_3, \ldots$ down to $S_1$ vastly outpaces any competing process. Radiative and further nonradiative decay essentially always starts from a thermally equilibrated $S_1$ — this is why emission spectra are largely independent of the excitation wavelength. From $S_1$, three channels compete: 1. Fluorescence: $S_1 \to S_0 + h\nu$, spin-allowed ($\Delta S = 0$), rate constant $k_f$. 2. Internal conversion: $S_1 \to S_0$ nonradiatively, spin-allowed, rate constant $k_{IC}$. 3. Intersystem crossing (ISC): $S_1 \to T_1$ nonradiatively, formally spin-forbidden ($\Delta S = 1$), rate constant $k_{ISC}$. The observed fluorescence quantum yield and lifetime are set by the sum of all depopulation channels from $S_1$:

If ISC occurs, the molecule now sits in $T_1$, typically below $S_1$ in energy (Hund's-rule-like exchange stabilization of the triplet). From $T_1$ the only routes back to $S_0$ are phosphorescence ($T_1 \to S_0 + h\nu$, rate $k_p$) and nonradiative triplet decay ($T_1 \to S_0$, rate $k_{nr}^T$, itself partly reliant on a second spin-orbit-mediated ISC event back to the singlet manifold). Both are spin-forbidden, so $T_1$ can live orders of magnitude longer than $S_1$ — hence the "afterglow."

The vertical purple arrow is absorption (Franck-Condon vertical, obeying the Electronic Transitions and the Franck-Condon Principle); the orange squiggles are nonradiative vibrational cascades (IC); the green squiggle marked ISC is the spin-forbidden crossing that only proceeds because $\hat H_{SO}$ mixes $S_1$ and $T_1$ character; the blue solid arrow is fluorescence; the red dashed arrow, drawn dashed specifically to flag its forbidden character, is phosphorescence. The essential quantum ingredient distinguishing fluorescence from phosphorescence is the spin-orbit coupling operator,

This term is absent from the nonrelativistic electrostatic Hamiltonian used to build pure spin states; it originates from the relativistic correction (effectively a magnetic interaction between an electron's own spin magnetic moment and the magnetic field generated, in its rest frame, by its orbital motion around the nuclear charge). Because $\hat H_{SO}$ contains $\hat{\boldsymbol l}\cdot\hat{\boldsymbol s}$, it can connect states differing in both orbital and spin angular momentum in a compensating way — it is the only term in a typical molecular Hamiltonian that can mix a singlet and a triplet built from the same orbital configuration, or from orbitally-different configurations. Its magnitude scales steeply with nuclear charge ($\xi \propto Z_{\text{eff}}^4$ for a hydrogenic estimate), which is the quantum mechanical root of the heavy-atom effect.

The unifying quantitative tool is Fermi's golden rule, which gives a first-order perturbative transition rate between an initial quantum state and a manifold of final states:

where:

• $\hat H'$ is the perturbation responsible for the transition (the electric-dipole/radiation-field coupling for a radiative process; the spin-orbit or vibronic coupling operator for a nonradiative one),
• $\langle f|\hat H'|i\rangle$ is the matrix element connecting the initial and final molecular (electronic $\times$ vibrational $\times$ spin) states,
• $\rho(E_f)$ is the density of accessible final states at the resonance energy (photon modes for emission; vibrational/phonon levels of the lower electronic state for nonradiative decay).

For an allowed electric-dipole radiative transition between electronic states of the same spin, the emission rate constant (spontaneous emission, Einstein $A$ coefficient generalized to a molecule) is

where:

• $\tilde\nu$ is the transition wavenumber (cm$^{-1}$) of the $0$–$0$ emission band,
• $\boldsymbol\mu_{fi} = \langle \psi_f|\hat{\boldsymbol\mu}|\psi_i\rangle$ is the electric transition dipole moment between the electronic-vibrational states,
• the $\tilde\nu^3$ factor is the photon density of states weighting — higher-energy transitions radiate intrinsically faster, all else equal.

For a spin-forbidden transition (ISC or phosphorescence), the bare electric-dipole operator does not connect states of different total spin ($\langle {}^3\Psi|\hat{\boldsymbol\mu}|{}^1\Psi\rangle = 0$ exactly, because $\hat{\boldsymbol\mu}$ is spin-independent and spin states of different $S$ are orthogonal). The transition becomes allowed only because the true stationary states are not pure spin states — spin-orbit coupling mixes a small amount of singlet character into the "triplet":

where the sum runs over singlet states $S_n$ (including $S_0$) that can spin-orbit couple to $T_1$, and the admixture coefficient is a standard nondegenerate perturbation theory mixing coefficient. The phosphorescence transition dipole is then approximately

which is a borrowed transition dipole — small because the numerator (spin-orbit matrix element, typically $1$–$100\ \text{cm}^{-1}$ for light atoms, up to $1000$–$5000\ \text{cm}^{-1}$ for heavy atoms) is much smaller than typical electronic energy gaps in the denominator ($10^4$ cm$^{-1}$).

Example 1 — Foundational: converting a radiative rate constant to an oscillator strength and back. Anthracene fluoresces with $\tau_{S_1} = 5.1\ \text{ns}$ and a fluorescence quantum yield $\Phi_f = 0.36$ (aerated solution values, chosen deliberately non-round). Find $k_f$, $k_{IC}+k_{ISC}$ (lumped nonradiative rate), and the radiative rate expected from a measured extinction coefficient corresponding to a transition centered at $\tilde\nu = 26\,700\ \text{cm}^{-1}$ ($375$ nm) with integrated absorption strength giving $|\boldsymbol\mu_{fi}|^2 = 2.6\ \text{D}^2$ ($1\ \text{D} = 3.336\times10^{-30}\ \text{C}\,\cdot\,\text{m}$). From the definitions,

Now check this against the golden-rule/Einstein expression, converting $|\boldsymbol\mu_{fi}|^2 = 2.6\ \text{D}^2 = 2.6\times(3.336\times10^{-30})^2\ \text{C}^2\text{m}^2 = 2.894\times10^{-59}\ \text{C}^2\text{m}^2$. Using the SI form of the rate law,

with $\tilde\nu = 26\,700\ \text{cm}^{-1} = 2.670\times10^{6}\ \text{m}^{-1}$, so $\tilde\nu^3 = 1.903\times10^{19}\ \text{m}^{-3}$:

Numerator of the prefactor: $16\pi^3 = 496.1$; $496.1\times1.903\times10^{19} = 9.441\times10^{21}$. Denominator: $3\times8.854\times10^{-12}\times6.626\times10^{-34} = 1.760\times10^{-44}$. Prefactor $= 9.441\times10^{21}/1.760\times10^{-44} = 5.364\times10^{65}\ \text{m}^{-3}\text{C}^{-2}\text{m}^{-2}\text{s}^{-1}$ (units work out to s$^{-1}$ per C$^2$m$^2$ once multiplied by $|\mu|^2$). Multiplying by $|\boldsymbol\mu_{fi}|^2$:

This is the intrinsic radiative rate computed from absorption strength via the Strickler-Berg relation (order-of-magnitude estimate here, since a full Strickler-Berg treatment integrates over the whole absorption band rather than using a single $\tilde\nu$ and $\mu$). It lands within a factor of $\sim 4.5$ of the kinetically measured $k_f = 7.06\times10^7\ \text{s}^{-1}$ — the residual difference is expected because the single-frequency point-dipole estimate omits the frequency-cubed weighting integrated properly across the vibronic band envelope; the qualitative lesson stands: both routes independently give $k_f$ on the order of $10^7$–$10^8\ \text{s}^{-1}$, consistent with a fully spin- and symmetry-allowed $\pi\pi^*$ transition.

Limiting-case check: if the transition dipole moment vanished ($|\boldsymbol\mu_{fi}| \to 0$, as for a symmetry-forbidden transition), $k_r \to 0$ and all population would have to decay through $k_{IC}$ or $k_{ISC}$ alone — exactly what happens for centrosymmetric molecules undergoing formally Laporte-forbidden transitions, which show anomalously long $S_1$ lifetimes.

Example 2 — Applied: heavy-atom effect on ISC yield and phosphorescence in naphthalene versus 1-bromonaphthalene. Naphthalene has $\Phi_{ISC} \approx 0.23$ and phosphorescence is barely observable at room temperature (long-lived, easily quenched by oxygen and vibrational deactivation). 1-Bromonaphthalene, with a heavy bromine substituent, has $\Phi_{ISC} \approx 0.72$ and shows strong, easily observed phosphorescence even at moderate temperatures. Rationalize this using the $Z_{\text{eff}}^4$ scaling of spin-orbit coupling and estimate the ISC rate ratio. The internal heavy-atom effect increases $\hat H_{SO}$ locally at the bromine center. Using the hydrogenic-like scaling $\xi \propto Z_{\text{eff}}^4$ as a crude guide (recognizing this is a single-electron approximation, but useful for order-of-magnitude comparison): comparing effective nuclear charge felt by valence electrons near Br ($Z_{\text{eff}} \approx 7.6$ for outer bromine electrons in Slater's rules) versus near C ($Z_{\text{eff}} \approx 3.25$),

Because the ISC rate scales as $k_{ISC} \propto |\langle S_1|\hat H_{SO}|T_1\rangle|^2$, and the matrix element itself scales roughly linearly with the local spin-orbit coupling constant $\xi$ (the heavy atom contributes an admixed atomic-like spin-orbit term into the molecular matrix element), the rate should scale as roughly the square of the $\xi$ ratio if the coupling were purely localized and linear in $\xi$; more realistically (since only a fraction of the transition density sits on Br) the enhancement is attenuated. Empirically the observed ISC quantum yield increases from $0.23$ to $0.72$, i.e., a change in the branching ratio rather than the full $30\times$ naive prediction — because $\Phi_{ISC} = k_{ISC}/(k_f+k_{IC}+k_{ISC})$ saturates as $k_{ISC}$ grows large relative to the other channels. Back-solving,

gives ratios of $0.23/0.77 = 0.299$ for naphthalene and $0.72/0.28 = 2.571$ for 1-bromonaphthalene, an increase of a factor of

in the effective $k_{ISC}/(k_f+k_{IC})$ ratio — smaller than the naive $30\times$ single-atom estimate because the spin-orbit matrix element depends on the overlap of the relevant molecular orbitals with the bromine center, not simply the atomic $\xi$ value, but unambiguously in the direction and order of magnitude the heavy-atom argument predicts.

Limiting-case check: as $Z_{\text{eff}} \to$ that of a very light atom (e.g., replacing Br with F, $Z_{\text{eff}}$ much smaller), $\xi \to$ small, ISC should slow further and $\Phi_{ISC}$ should drop below the parent hydrocarbon value — consistent with the general trend that among halonaphthalenes, ISC efficiency orders as I > Br > Cl > F ≈ H.

Example 3 — Exam/competition level: predicting relative phosphorescence lifetimes from El-Sayed's rule and computing an approximate zero-field splitting-based lifetime bound. Two carbonyl compounds, benzophenone and biphenyl-4-carbaldehyde-like systems (illustrative), are compared. Benzophenone has its lowest triplet as $^3(n,\pi^*)$ character with $T_1$ at $E_T \approx 24\,400\ \text{cm}^{-1}$ (69.8 kcal/mol), while a hypothetical all-hydrocarbon aromatic analog has $T_1$ purely $^3(\pi,\pi^*)$ at a similar energy. Both undergo $T_1 \to S_0$ phosphorescence, but the ground state $S_0$ is $^1(\pi,\pi^*)$-like (fully paired $\pi$ system) in both. Predict which shows the shorter phosphorescence lifetime and estimate $k_p$ orders of magnitude using El-Sayed's rule.

Reasoning. Phosphorescence itself is $T_1 \to S_0$; whether it is efficient depends on the spin-orbit matrix element connecting $T_1$ to nearby singlet configurations that can donate oscillator strength to $S_0$. For benzophenone, $T_1$ is $^3(n,\pi^*)$ while $S_0$ (and the strongly allowed $S_n$ that could donate intensity) are $(\pi,\pi^*)$-type. By El-Sayed's rule, spin-orbit coupling is large precisely when the orbital character changes between the coupled states —

is a symmetry-allowed (large matrix element) spin-orbit channel, because the operator $\hat{\boldsymbol l}$ effectively rotates electron density from the in-plane, non-bonding $n$ orbital into the out-of-plane $\pi$ system, generating substantial orbital angular momentum change. In contrast, an all-$(\pi,\pi^*)$ system undergoing

has no accompanying orbital-type change, so $\langle S_0|\hat H_{SO}|T_1\rangle$ is intrinsically much smaller (typically by one to two orders of magnitude in the matrix element, translating to two to four orders of magnitude in rate, since rate $\propto$ matrix element squared).

Quantitative estimate. Typical experimental $k_p$ values bear this out: aromatic hydrocarbons with $^3(\pi,\pi^*)$ lowest triplets (e.g., naphthalene) show $k_p \sim 10^{-2}$–$10^{0}\ \text{s}^{-1}$ (phosphorescence lifetimes of seconds), whereas aromatic ketones/aldehydes with $^3(n,\pi^*)$ lowest triplets (e.g., benzophenone, benzaldehyde) show $k_p \sim 10^{2}$–$10^{3}\ \text{s}^{-1}$ (lifetimes of milliseconds). Using the borrowed-dipole formula from the derivation above with representative numbers: take $\langle S_n|\hat H_{SO}|T_1\rangle \approx 30\ \text{cm}^{-1}$ for the $(n,\pi^*)$-type coupling (El-Sayed-favored) versus $\approx 1\ \text{cm}^{-1}$ for the $(\pi,\pi^*)$-type coupling (El-Sayed-disfavored), with a common energy denominator $E_{T_1}-E_{S_n} \approx 5000\ \text{cm}^{-1}$ and a common donor transition dipole $|\boldsymbol\mu_{0n}| = 3\ \text{D}$:

Given a base fully-allowed rate $k_r^{\text{allowed}} \sim 2\times10^{8}\ \text{s}^{-1}$ (typical for a strong $\pi\pi^*$ singlet transition, $\tilde\nu \approx 25{,}000$ cm$^{-1}$, $|\mu|=3$ D), the phosphorescence rates come out to

giving phosphorescence lifetimes $\tau_p^{(n,\pi^*)} = 1/k_p \approx 0.14\ \text{ms}$ and $\tau_p^{(\pi,\pi^*)} \approx 0.125\ \text{s}$ — a difference of three orders of magnitude that tracks real experimental trends (ketone/aldehyde phosphorescence lifetimes of hundreds of microseconds to milliseconds, versus aromatic hydrocarbon phosphorescence lifetimes of tenths of a second to several seconds).

Limiting-case check: if the two states $S_n$ and $T_1$ became orbitally identical in every respect (the artificial limit of "no orbital change at all," $\hat{\boldsymbol l}$ matrix element exactly zero), $c_n \to 0$ and $k_p \to 0$ exactly — phosphorescence would vanish and $T_1$ would only decay by nonradiative pathways, exactly the pure-spin, zero-spin-orbit-coupling limit used to justify $\langle S_0|\hat{\boldsymbol\mu}|T_1\rangle = 0$ in Step 1 of the derivation above.

Michael Kasha, working in the mid-twentieth century United States, brought physical rigor to what had been largely empirical photoluminescence observations, formalizing the vibrational-relaxation cascade argument that now bears his name and clarifying the electronic-state language (singlet/triplet, Jablonski-diagram bookkeeping) still used to teach photophysics. His work sits alongside Aleksander Jablonski's original diagrammatic scheme (from which the now-universal "Jablonski diagram" takes its name) and M. A. El-Sayed's mechanistic explanation, via orbital symmetry, of why some spin-forbidden intersystem crossings are dramatically faster than others.