Absorption and Emission Processes

A sodium streetlamp glows a distinctive orange-yellow. A ruby laser fires a pulse of deep red light. Both phenomena trace back to the same three quantum events: absorption, spontaneous emission, and stimulated emission. Every spectrochemical instrument, from a simple colorimeter to a laser-induced fluorescence detector, is built on the bookkeeping of photons moving electrons between quantized Energy Levels and Molecular Transitions states. This topic matters because it is the mechanistic foundation underneath nearly all optical analytical chemistry. Before a chemist can trust that absorbance is proportional to concentration, they need more. The same applies before trusting that a fluorescence signal reports on analyte identity. They need a working model of how atoms and molecules trade energy with the electromagnetic field. This means understanding how fast these events happen. It also means understanding what governs their probabilities. Additionally, it means understanding why some excited states glow. Others quietly dissipate as heat instead. The big picture is a population problem. A collection of molecules distributed over ground and excited states responds to a light field. This response follows three rate processes first unified by Einstein in 1917. These are absorption (population moves up, driven by the field). Also stimulated emission (population moves down, driven by the field). And spontaneous emission (population moves down, driven by nothing but the excited state's own instability). Layered onto these radiative pathways are nonradiative relaxation channels. These include internal conversion, intersystem crossing, and collisional quenching. They compete for the same excited-state population. Together they determine whether a molecule fluoresces, phosphoresces, or simply warms its surroundings. This article builds the rate-equation picture of these three processes from first principles. It connects Einstein's A and B coefficients to measurable quantities like molar absorptivity and radiative lifetime. It also works through realistic numerical examples spanning atomic absorption, laser gain media, and fluorescence quantum yields. It sets up the machinery that The Beer-Lambert Law and Introduction to Molecular Fluorescence both depend on.

Key takeaways

Absorption, stimulated emission, and spontaneous emission are the three Einstein processes linking a two-level system to the radiation field.
Einstein's A coefficient sets the natural (radiative) lifetime; the B coefficients set the field-driven rates and are proportional to each other and to A.
At thermal equilibrium, spontaneous emission dominates unless the photon energy is small or the population inversion is engineered (as in a laser).
Radiative lifetime, natural linewidth, and molar absorptivity are three faces of the same transition dipole moment.
Excited states relax by competing pathways — fluorescence, intersystem crossing, internal conversion, and collisional quenching — and the fluorescence quantum yield is the fraction that emits.
Real spectral lines are broadened by lifetime (natural), Doppler, and pressure effects, which set the ultimate resolution of any optical spectrometer.
Key idea: every optical spectroscopic signal is a rate-equation competition between one radiative pathway you want to measure and several others that steal population from it.

Where this lives today

Atomic absorption spectroscopy (AAS) for trace metal analysis relies directly on the absorption side of the Einstein picture: a hollow-cathode lamp emits sharp spontaneous-emission lines matched to the analyte's own absorption transitions, and Beer–Lambert absorbance measures ground-state atomic population in a flame or graphite furnace. Molecular fluorescence spectrometers exploit the radiative decay pathway of the Jablonski diagram, using fluorescence quantum yield and lifetime to quantify analytes at parts-per-trillion levels — far below what absorbance alone can detect, because fluorescence is measured against a dark background rather than as a small difference between two large light intensities. Time-resolved fluorescence lifetime imaging microscopy (FLIM) in biomedical research directly measures $\tau_{\text{obs}}$ , using its sensitivity to local $k_q[Q]$ (oxygen concentration, pH, ion binding) as a quantitative reporter of cellular microenvironments — an application built entirely on the second derivation's competing-pathways framework. Laser-induced fluorescence (LIF) detection, used in capillary electrophoresis and environmental monitoring, requires balancing pump intensity against the ratio of stimulated to spontaneous rates worked out in Example 1, to avoid saturating the transition. Solid-state lighting (LEDs) and laser gain media (Nd:YAG, Ti:sapphire, diode lasers used in every DVD/Blu-ray reader and telecommunications system) are direct engineering applications of the population-inversion condition from Example 3, chosen specifically because their upper laser levels have long spontaneous lifetimes (metastable states), allowing population to build up before decaying. Even the humble sodium streetlamp and neon sign are living demonstrations of the Einstein A coefficient converting electrical excitation into visible spontaneous emission.

Ask the tutor: How does the Einstein B coefficient framework connect quantitatively to the molar absorptivity used in the Beer-Lambert law?

The physical picture

Consider two nondegenerate energy levels of an atom or molecule, a lower state 1 with energy $E_1$ and an upper state 2 with energy $E_2$ , separated by $\Delta E = E_2 - E_1 = h\nu_{21}$ . A beam of light with spectral energy density $\rho(\nu)$ at frequency $\nu_{21}$ passes through a sample containing $N_1$ molecules per unit volume in state 1 and $N_2$ in state 2. Three things can happen. First, a photon of the right energy can be absorbed, promoting a molecule from state 1 to state 2. This is stimulated absorption: its rate per unit volume is proportional to both $N_1$ and $\rho(\nu_{21})$ . Second, a photon of the same energy can strike a molecule already in state 2 and stimulate it to drop to state 1, emitting a second photon that is coherent with the first — same frequency, phase, direction, and polarization. This stimulated emission is the physical basis of laser gain. Third, a molecule in state 2 can decay to state 1 entirely on its own, with no external field needed, releasing a photon in a random direction with random phase. This spontaneous emission is what makes a fluorescent dye glow under UV light with no coherent output, and it is the process responsible for finite excited-state lifetimes. Einstein showed in 1917 — years before quantum electrodynamics existed to derive it from first principles — that if a collection of two-level atoms is bathed in blackbody radiation and required to reach thermal equilibrium, consistency with the Boltzmann distribution and Planck's radiation law forces fixed relationships among the rate constants for these three processes. Get any two, and the third is determined. This is why the spontaneous emission rate — an apparently "un-drivable" process — can nonetheless be predicted from how strongly a transition absorbs light, a fact exploited constantly in spectrochemical analysis to relate absorption cross sections to fluorescence lifetimes. Once a molecule is excited, its fate is not necessarily to re-emit a photon of the same energy. Vibrational relaxation, internal conversion between electronic states of the same spin multiplicity, intersystem crossing to a different spin multiplicity, and collisional (dynamic) quenching by solvent or dissolved oxygen all compete with radiative decay. The experimentally observed excited-state lifetime and fluorescence quantum yield are the net result of all these parallel first-order (or pseudo-first-order) decay channels running simultaneously.

Figure

Three Einstein processes for a two-level system

Figure

Jablonski diagram: radiative and nonradiative pathways after excitation

The core ideas

The Einstein rate equations for a two-level system exposed to isotropic radiation of spectral energy density $\rho(\nu_{21})$ (units J·s·m⁻³, energy density per unit frequency) are:

\frac{dN_1}{dt}\bigg|_{\text{abs}} = -B_{12}\,\rho(\nu_{21})\,N_1

\frac{dN_2}{dt}\bigg|_{\text{stim}} = -B_{21}\,\rho(\nu_{21})\,N_2, \qquad \frac{dN_2}{dt}\bigg|_{\text{spont}} = -A_{21}\,N_2

where:

$N_1, N_2$ = number density of molecules in the lower and upper states (m⁻³)
$B_{12}$ = Einstein coefficient for stimulated absorption (m³·J⁻¹·s⁻²), driven by the field
$B_{21}$ = Einstein coefficient for stimulated emission (same units as $B_{12}$ )
$A_{21}$ = Einstein coefficient for spontaneous emission (s⁻¹) — a first-order rate constant
$\rho(\nu_{21})$ = spectral energy density of the radiation field at the transition frequency

At thermal equilibrium with a blackbody field, detailed balance and the Boltzmann/Planck distributions force the relations:

g_1 B_{12} = g_2 B_{21}, \qquad A_{21} = \frac{8\pi h \nu_{21}^3}{c^3}\,B_{21}

where:

$g_1, g_2$ = degeneracies of the lower and upper states
$h$ = Planck's constant, $6.62607\times10^{-34}\ \text{J}\,\cdot\,\text{s}$
$c$ = speed of light, $2.99792\times10^{8}\ \text{m}\,\cdot\,\text{s}^{-1}$
$\nu_{21}$ = transition frequency (Hz)

Reading: the second relation says spontaneous emission is not an independent phenomenon — it is stimulated emission "seeded" by vacuum fluctuations of the electromagnetic field, and its rate scales as $\nu^3$ . This is why spontaneous decay is negligible for microwave/radiofrequency transitions (NMR relaxes over seconds) but dominant for UV-visible electronic transitions (fluorescence lifetimes of nanoseconds). The net rate of upward vs. downward transitions under illumination combines all three processes:

\frac{dN_2}{dt} = B_{12}\rho(\nu_{21})N_1 - B_{21}\rho(\nu_{21})N_2 - A_{21}N_2

where each term has the meaning given above. At true equilibrium $dN_2/dt = 0$ , and substituting the Boltzmann ratio $N_2/N_1 = (g_2/g_1)e^{-h\nu_{21}/k_BT}$ into this equation reproduces Planck's law for $\rho(\nu)$ — the calculation Einstein originally used to derive the $A/B$ relationships in the first place. The radiative (natural) lifetime of the excited state, $\tau_{\text{rad}}$ , is set purely by spontaneous emission:

\tau_{\text{rad}} = \frac{1}{A_{21}}

where $A_{21}$ is the spontaneous emission rate constant defined above. This is the lifetime a molecule would show if spontaneous emission were the only decay pathway available. Finally, the Einstein $B$ coefficient connects directly to the experimentally measured integrated absorption cross section — and hence to molar absorptivity $\varepsilon$ , the quantity underlying Beer's law:

B_{12} = \frac{c}{8\pi h \nu_{21}}\cdot\frac{g_2}{g_1}\, A_{21}

where all symbols are as defined above; this is simply the first Einstein relation solved for $B_{12}$ in terms of $A_{21}$ .

Derivation AFrom the Einstein relations to the radiative lifetime and oscillator strength

Goal: show how a purely theoretical quantity, $A_{21}$ , connects to something a chemist can extract from a UV-vis spectrum — the integrated molar absorptivity — via the classical oscillator strength $f$ .

Step 1. Start from the quantum mechanical expression for the spontaneous emission rate of an electric-dipole transition between states 1 and 2, obtained from time-dependent perturbation theory (Fermi's golden rule applied to vacuum-field coupling):

A_{21} = \frac{16\pi^3 \nu_{21}^3}{3\varepsilon_0 h c^3}\,|\boldsymbol{\mu}_{21}|^2

where $\boldsymbol{\mu}_{21}$ is the transition dipole moment (C·m) connecting states 1 and 2, and $\varepsilon_0$ is the vacuum permittivity. This says: transitions with large transition dipoles (strongly allowed, high symmetry-permitted overlap) decay spontaneously much faster than weak, symmetry-forbidden ones.

Step 2. Define the dimensionless oscillator strength $f_{12}$ , which compares the actual transition to that of a classical electron oscillating harmonically at frequency $\nu_{21}$ :

f_{12} = \frac{8\pi^2 m_e \nu_{21}}{3h e^2}\,|\boldsymbol{\mu}_{21}|^2

where $m_e$ is the electron mass and $e$ is the elementary charge. Substituting $|\boldsymbol{\mu}_{21}|^2$ from this into the Step 1 expression for $A_{21}$ eliminates the microscopic dipole moment in favor of the measurable $f_{12}$ :

A_{21} = \frac{2\pi e^2 \nu_{21}^2}{\varepsilon_0 m_e c^3}\, f_{12}

Step 3. Relate $f_{12}$ to the integrated absorption band area, which is what a spectrometer actually records. The Smakula-type relation for a solution-phase transition (SI, with $\varepsilon(\tilde\nu)$ in units of M⁻¹cm⁻¹ and wavenumber $\tilde\nu$ in cm⁻¹) is:

f_{12} = 4.32\times10^{-9}\int \varepsilon(\tilde\nu)\, d\tilde\nu

where the numerical prefactor absorbs the physical constants ( $\varepsilon_0$ , $m_e$ , $e$ , $c$ , and unit conversions from cm⁻¹ and M⁻¹cm⁻¹ to SI). Combining Steps 2 and 3 gives the radiative lifetime purely in terms of the measured absorption spectrum and the transition frequency:

\tau_{\text{rad}} = \frac{1}{A_{21}} = \frac{\varepsilon_0 m_e c^3}{2\pi e^2 \nu_{21}^2\, f_{12}}

Step 4. The takeaway: a strongly absorbing chromophore (large $\varepsilon_{\max}$ , large integrated band area, hence large $f_{12}$ ) has a short natural radiative lifetime. This single fact links a UV-vis absorption spectrum (an equilibrium, ensemble measurement) to a fluorescence lifetime (a kinetic, single-molecule-averaged quantity) — the conceptual bridge from absorption spectroscopy to fluorescence spectroscopy.

Go deeper · Derivation B · 1 minCompeting decay pathways: fluorescence quantum yield and observed lifetime

Derivation BCompeting decay pathways: fluorescence quantum yield and observed lifetime

Goal: derive how the measured excited-state lifetime and fluorescence quantum yield depend on all competing first-order decay channels, not just spontaneous radiative emission.

Step 1. After a short excitation pulse populates $N_2(0)$ molecules in the excited state, population decays via several parallel, independent first-order pathways: radiative decay ( $k_r = A_{21}$ , fluorescence), internal conversion ( $k_{IC}$ ), intersystem crossing ( $k_{ISC}$ ), and collisional quenching by some quencher Q at concentration $[Q]$ with rate constant $k_q$ (pseudo-first-order term $k_q[Q]$ ). The total loss rate is the sum, because these are independent competing first-order processes acting on the same population:

-\frac{dN_2}{dt} = \left(k_r + k_{IC} + k_{ISC} + k_q[Q]\right) N_2 \equiv k_{\text{obs}}\,N_2

Step 2. Integrating this single first-order ODE from $N_2(0)$ gives simple exponential decay, exactly analogous to radioactive decay or first-order chemical kinetics:

N_2(t) = N_2(0)\, e^{-k_{\text{obs}} t}

The observed (measured) fluorescence lifetime is the reciprocal of the total rate constant, not just the radiative one:

\tau_{\text{obs}} = \frac{1}{k_{\text{obs}}} = \frac{1}{k_r + k_{IC} + k_{ISC} + k_q[Q]}

Step 3. The fluorescence quantum yield $\Phi_f$ is defined as the fraction of excited molecules that decay via the radiative (fluorescent) channel specifically — the branching ratio for that one pathway among all the competing first-order rates:

\Phi_f = \frac{k_r}{k_r + k_{IC} + k_{ISC} + k_q[Q]} = k_r\,\tau_{\text{obs}}

Step 4. Two limiting cases fall out immediately. If all nonradiative channels vanish ( $k_{IC}=k_{ISC}=k_q[Q]=0$ ), then $k_{\text{obs}} = k_r$ , $\tau_{\text{obs}} = \tau_{\text{rad}} = 1/A_{21}$ , and $\Phi_f \to 1$ — every excited molecule fluoresces, recovering the purely radiative result of the first derivation. Conversely, if quenching is fast ( $k_q[Q] \gg k_r$ ), $\Phi_f \to k_r/(k_q[Q]) \to 0$ and the observed lifetime collapses toward $1/(k_q[Q])$ — this is exactly the regime exploited in Stern–Volmer fluorescence quenching assays, where $\Phi_{f,0}/\Phi_f = 1 + k_q\tau_{\text{rad}}[Q]$ is used to measure oxygen or quencher concentration.

Worked examples

Example 1 — Foundational: comparing spontaneous and stimulated emission rates for a visible transition. A sodium atom's 3p → 3s transition (the D-line, $\lambda \approx 589.0$ nm) has a measured spontaneous emission (Einstein $A$ ) coefficient of $A_{21} = 6.2\times10^{7}\ \text{s}^{-1}$ . Suppose this atom sits inside a moderately intense light field with spectral energy density $\rho(\nu_{21}) = 1.0\times10^{-15}\ \text{J}\,\cdot\,\text{s}\,\cdot\,\text{m}^{-3}$ at the resonance frequency (roughly the value produced by a few mW of laser power focused to a 100 μm spot, spectrally narrow). The degeneracies are $g_1 = 2$ (³S₁/₂ ground term) and $g_2 = 4$ (²P₃/₂ upper term, using the $D_2$ line specifically). First find the transition frequency:

\nu_{21} = \frac{c}{\lambda} = \frac{2.998\times10^{8}\ \text{m/s}}{589.0\times10^{-9}\ \text{m}} = 5.091\times10^{14}\ \text{Hz}

Now compute $B_{21}$ from the Einstein relation $A_{21} = \dfrac{8\pi h \nu_{21}^3}{c^3} B_{21}$ :

B_{21} = \frac{A_{21}\, c^3}{8\pi h \nu_{21}^3} = \frac{(6.2\times10^{7})(2.998\times10^{8})^3}{8\pi (6.626\times10^{-34})(5.091\times10^{14})^3}

Working the numerator: $(2.998\times10^8)^3 = 2.694\times10^{25}\ \text{m}^3\text{s}^{-3}$ , so the numerator is $6.2\times10^7 \times 2.694\times10^{25} = 1.671\times10^{33}$ . Working the denominator: $(5.091\times10^{14})^3 = 1.319\times10^{44}\ \text{Hz}^3$ , and $8\pi(6.626\times10^{-34}) = 1.665\times10^{-32}$ , giving denominator $= 1.665\times10^{-32}\times1.319\times10^{44} = 2.196\times10^{12}$ .

B_{21} = \frac{1.671\times10^{33}}{2.196\times10^{12}} = 7.61\times10^{20}\ \text{m}^3\text{J}^{-1}\text{s}^{-2}

The stimulated emission rate is $B_{21}\rho(\nu_{21}) = (7.61\times10^{20})(1.0\times10^{-15}) = 7.6\times10^{5}\ \text{s}^{-1}$ . Comparing rates: spontaneous emission ( $6.2\times10^7\ \text{s}^{-1}$ ) exceeds stimulated emission ( $7.6\times10^5\ \text{s}^{-1}$ ) by a factor of about 80 under this field intensity. This is the everyday regime of ordinary absorption/emission spectroscopy: spontaneous decay dominates unless the field is made extremely intense (as inside a laser cavity), confirming why fluorescence — not stimulated emission — is what a bench spectrofluorometer detects.

Limiting check: as $\rho(\nu_{21}) \to 0$ (dark, no field), the stimulated term vanishes entirely and decay is purely spontaneous with lifetime $\tau_{\text{rad}} = 1/A_{21} = 1/(6.2\times10^7\ \text{s}^{-1}) = 16\ \text{ns}$ , consistent with the known ~16 ns lifetime of the sodium D-line — a useful sanity check on the whole calculation.

Example 2 — Applied: fluorescence quantum yield of a dye from lifetime and radiative rate data. Fluorescein in pH 9 buffer has a measured (observed) fluorescence lifetime of $\tau_{\text{obs}} = 4.10\ \text{ns}$ and a radiative rate constant, obtained independently from its integrated absorption spectrum via the Strickler–Berg method (an application of the first derivation above), of $k_r = 2.15\times10^{8}\ \text{s}^{-1}$ . The fluorescence quantum yield is:

\Phi_f = k_r\, \tau_{\text{obs}} = (2.15\times10^{8}\ \text{s}^{-1})(4.10\times10^{-9}\ \text{s}) = 0.8815

Rounding to appropriate significant figures (three, matching the input data): $\Phi_f = 0.881$ , close to fluorescein's textbook value of ≈ 0.90–0.95 depending on pH and ionic strength — the modest difference here is typical of solvent/pH-dependent nonradiative loss variation, and is consistent with reported literature spread. Now suppose a chemist adds potassium iodide, a collisional quencher, at $[\text{KI}] = 0.050\ \text{M}$ , and observes the lifetime drop to $\tau_{\text{obs}}' = 2.85\ \text{ns}$ . The added nonradiative rate is:

k_q[Q] = \frac{1}{\tau_{\text{obs}}'} - \frac{1}{\tau_{\text{obs}}} = \frac{1}{2.85\times10^{-9}} - \frac{1}{4.10\times10^{-9}}

k_q[Q] = 3.509\times10^{8} - 2.439\times10^{8} = 1.070\times10^{8}\ \text{s}^{-1}

Dividing by $[Q] = 0.050\ \text{M}$ gives the bimolecular quenching rate constant:

k_q = \frac{1.070\times10^{8}\ \text{s}^{-1}}{0.050\ \text{M}} = 2.14\times10^{9}\ \text{M}^{-1}\text{s}^{-1}

This value is close to (somewhat below) the diffusion-controlled limit in water (~1×10¹⁰ M⁻¹s⁻¹), consistent with iodide quenching fluorescein via a moderately efficient (not fully diffusion-limited) collisional/electron-transfer mechanism — exactly the kind of number reported in Stern–Volmer quenching studies used to probe solvent accessibility of fluorophores in proteins.

Limiting check: if $k_q[Q] \to 0$ (no quencher present), $\tau_{\text{obs}}' \to \tau_{\text{obs}}$ , recovering the unquenched lifetime, as required.

Example 3 — Exam/competition level: population inversion condition and gain threshold in a laser medium. A three-level laser medium has upper laser level 2 and lower laser level 1 with degeneracies $g_2 = 3$ and $g_1 = 1$ . The spontaneous emission rate is $A_{21} = 1.0\times10^{3}\ \text{s}^{-1}$ (a metastable transition, e.g., analogous to the Cr³⁺ transition in ruby), and the transition wavelength is $\lambda = 694.3\ \text{nm}$ . The medium is pumped so that $N_2 = 2.40\times10^{18}\ \text{cm}^{-3}$ while the lower level retains a thermal population governed by ordinary Boltzmann statistics at $T = 300\ \text{K}$ relative to the ground state, with $N_1 = 3.00\times10^{17}\ \text{cm}^{-3}$ .

(a) Determine whether population inversion (net optical gain) exists. For light amplification via stimulated emission to exceed absorption, the degeneracy-weighted inversion condition is:

\frac{N_2}{g_2} > \frac{N_1}{g_1}

because stimulated emission from level 2 competes against absorption into level 1, and the B coefficients (hence rates per molecule per state) are weighted through the $g_1B_{12}=g_2B_{21}$ relation.

\frac{N_2}{g_2} = \frac{2.40\times10^{18}}{3} = 8.00\times10^{17}\ \text{cm}^{-3}

\frac{N_1}{g_1} = \frac{3.00\times10^{17}}{1} = 3.00\times10^{17}\ \text{cm}^{-3}

Since $8.00\times10^{17} > 3.00\times10^{17}$ , population inversion exists — this medium can amplify light at 694.3 nm, i.e., stimulated emission events outnumber absorption events for photons resonant with this transition.

(b) Estimate the net stimulated emission rate advantage. Using $g_1 B_{12} = g_2 B_{21}$ , the net rate of photon generation (stimulated emission minus absorption) per unit volume, at a given $\rho(\nu_{21})$ , is proportional to:

\left(N_2 - \frac{g_2}{g_1}N_1\right) B_{21} = \left(2.40\times10^{18} - 3\times3.00\times10^{17}\right)B_{21} = \left(2.40\times10^{18} - 9.00\times10^{17}\right)B_{21}

= 1.50\times10^{18}\,B_{21}\ \text{cm}^{-3}

This positive effective population difference, $1.50\times10^{18}\ \text{cm}^{-3}$ , is what must be compared against cavity/output losses to determine whether the laser reaches threshold; a larger degeneracy-weighted inversion produces proportionally larger net gain per photon pass.

Limiting check: if the pump were removed so $N_2 \to 0$ and $N_1$ relaxed toward its full thermal (ground-state-dominated) value, the bracket $\left(N_2 - \frac{g_2}{g_1}N_1\right)$ would become negative, gain would turn to net absorption, and the system reverts to ordinary (non-lasing) absorptive behavior — the expected limiting case for any un-pumped medium at thermal equilibrium.

Limiting cases, scope & common traps

Situation	What dominates	Consequence
Low-frequency transitions (NMR, EPR, $\nu \sim 10^7$ – $10^{10}$ Hz)	Stimulated processes ( $A_{21}\propto\nu^3$ vanishingly small)	Relaxation is slow (seconds); saturation and stimulated transitions dominate signal
UV-visible electronic transitions ( $\nu\sim10^{14}$ – $10^{15}$ Hz)	Spontaneous emission, under ordinary light levels	Fluorescence lifetimes ~ns; laser-level field intensities needed for stimulated emission to compete
Highly allowed transitions (large $\varepsilon_{\max}$ , large $f$ )	Fast radiative decay	Short $\tau_{\text{rad}}$ ; often high $\Phi_f$ unless nonradiative channels are also fast
Heavy-atom-containing chromophores	Enhanced intersystem crossing ( $k_{ISC}$ ) via spin–orbit coupling	Lower $\Phi_f$ , higher phosphorescence yield
Oxygen-saturated aqueous solution	Collisional quenching ( $k_q[\text{O}_2]$ )	Shortened observed lifetimes; must degas for accurate $\Phi_f$ or lifetime measurements
Population inversion ( $N_2/g_2 > N_1/g_1$ )	Net stimulated emission (gain)	Laser action possible; ordinary thermal samples never satisfy this without external pumping

The chemist behind it

The rate-equation framework connecting absorption and emission is due to Albert Einstein, working not as an experimentalist but as a theorist reasoning from thermodynamic consistency. In 1916–1917, a full decade before Schrödinger's wave equation and Dirac's quantum theory of radiation existed, Einstein asked a deceptively simple question: what rate laws for absorption and emission are required so that a gas of atoms bathed in blackbody radiation settles into the Boltzmann distribution while the radiation field itself obeys Planck's law? He found that spontaneous emission — a rate process with no classical analog — was mathematically unavoidable, and in doing so predicted stimulated emission (the mechanism behind the laser) almost forty years before Townes, Schawlow, and Maiman built working devices.

Origins · 1 minAlbert Einstein and how this was discovered

Albert Einstein

1879–1955 · Kinetic theory of radiation, Berlin/Zurich, 1916–1917

Working during the transitional period between the "old" quantum theory of Bohr and Planck and the full quantum mechanics of the 1920s, Einstein derived the A and B coefficients using pure statistical-thermodynamic reasoning, without any quantum electrodynamic machinery for spontaneous emission itself. His 1917 paper "Zur Quantentheorie der Strahlung" ("On the Quantum Theory of Radiation") is the direct theoretical ancestor of every laser, LED, and fluorescence instrument built since.

History

- 1900 — Max Planck introduces energy quantization to explain blackbody radiation.

- 1913 — Niels Bohr proposes quantized atomic energy levels and photon-emission transitions.

- 1917 — Einstein derives the A and B coefficients and predicts stimulated emission.

- 1928 — Rudolf Ladenburg experimentally confirms stimulated emission ("negative dispersion") in neon discharges.

- 1960 — Theodore Maiman builds the first working laser (ruby), realizing Einstein's stimulated-emission prediction as a practical device.

Why does spontaneous emission rate scale with ν³ while stimulated processes do not?

The spontaneous emission rate arises from the coupling of an excited atom to the zero-point (vacuum) fluctuations of the electromagnetic field, and the density of available photon modes per unit frequency interval — the photon "phase space" — increases as $\nu^2$ , with an additional factor of $\nu$ entering through the photon energy itself in the emission rate expression. This combines to give the $\nu^3$ dependence seen in $A_{21} = (8\pi h\nu^3/c^3)B_{21}$ . Stimulated processes, by contrast, are driven by the actual photon field present, whose intensity $\rho(\nu)$ is set externally (by a lamp or laser) rather than by mode density, so no extra $\nu^3$ factor appears. This is precisely why building X-ray or gamma-ray lasers is far harder than building visible lasers: at very high $\nu$ , spontaneous emission is enormously fast and hard to outcompete with stimulated processes.

How is Einstein's B coefficient related to molar absorptivity ε measured on a UV-vis spectrophotometer?

Molar absorptivity is an experimentally convenient macroscopic quantity (units M⁻¹cm⁻¹) describing how strongly a solution attenuates light per unit concentration and path length. The Einstein $B_{12}$ coefficient is the underlying microscopic rate constant for photon absorption per molecule per unit spectral energy density. They are linked through the integrated absorption cross section: integrating $\varepsilon(\tilde\nu)$ over an absorption band gives the oscillator strength $f_{12}$ (via the Smakula-type relation), and $f_{12}$ connects directly to $B_{12}$ through the standard quantum-mechanical expression for the transition dipole moment. In short, $\varepsilon(\tilde\nu)$ is the spectrum a chemist measures, while $B_{12}$ is the rate constant a theorist writes down — the same physical information in two different units.

Does every excited molecule eventually emit a photon?

No. Radiative decay (fluorescence or phosphorescence) is only one of several competing first-order pathways available to an excited state. Nonradiative pathways — internal conversion (rapid relaxation to a lower state of the same spin multiplicity, releasing energy as heat/vibrations), intersystem crossing (to a different spin multiplicity), and collisional quenching by solvent, oxygen, or other quenchers — can dominate, especially in flexible or heavy-atom-containing molecules. The fluorescence quantum yield $\Phi_f = k_r/k_{\text{obs}}$ quantifies exactly what fraction of excited molecules do emit; many important chromophores (including most transition-metal complexes without special ligand design) have $\Phi_f$ close to zero, decaying almost entirely as heat.

What is the practical difference between fluorescence and phosphorescence in terms of these rate processes?

Fluorescence is radiative decay from an excited state back to the ground state without a change in spin multiplicity (e.g., singlet-to-singlet), so it is spin-allowed and typically fast, with rate constants $k_r \sim 10^7$ – $10^9\ \text{s}^{-1}$ and lifetimes in the nanosecond range. Phosphorescence is radiative decay from a state of different spin multiplicity than the ground state (typically triplet-to-singlet), which is formally spin-forbidden and only occurs because spin–orbit coupling mixes some singlet character into the triplet state. This makes phosphorescence rate constants much smaller ( $k_p \sim 10^{-2}$ – $10^3\ \text{s}^{-1}$ ), giving lifetimes from microseconds to seconds — long enough that collisional quenching (especially by dissolved oxygen) almost always outcompetes phosphorescence at room temperature in fluid solution, which is why phosphorescence is typically only observed in rigid, degassed, or frozen media.

How does population inversion relate to ordinary absorption spectroscopy, where inversion never occurs?

In an ordinary absorption experiment at thermal equilibrium, the Boltzmann distribution guarantees $N_1/g_1 > N_2/g_2$ for any transition with $E_2 > E_1$ at any positive temperature, so net absorption always occurs — a beam passing through the sample loses intensity, exactly as required for Beer–Lambert-type measurements to make physical sense. Population inversion, $N_2/g_2 > N_1/g_1$ , is an explicitly nonequilibrium condition that requires external energy input (optical pumping, electrical discharge, chemical reaction) to force population "uphill" against the natural Boltzmann trend. This is precisely why ordinary light sources illuminating ordinary samples never spontaneously amplify light, and why building a laser always requires an active pump mechanism rather than simply intense illumination.

Why do fluorescence measurements generally offer better sensitivity than absorption measurements for trace analysis?

Absorption measures the difference between two large light intensities ( $I_0$ and $I$ ), so at low analyte concentration this difference becomes a small number subtracted from a much larger one, degrading the signal-to-noise ratio as concentration drops (a "small difference of large numbers" problem inherent to transmittance measurements). Fluorescence, by contrast, is measured as a signal appearing against an essentially dark background (with the excitation light spectrally and spatially rejected by filters/monochromators), so the fluorescence intensity itself is directly and roughly linearly proportional to concentration at low concentrations, without any large background to subtract. This is why fluorescence detection routinely reaches picomolar-to-femtomolar sensitivities in techniques like laser-induced fluorescence detection, while absorption-based techniques are typically limited to micromolar-to-millimolar ranges without preconcentration.

What physically broadens an absorption or emission line beyond a single sharp frequency?

Three mechanisms dominate. Natural (lifetime) broadening arises from the time–energy uncertainty principle: a finite excited-state lifetime $\tau$ implies an energy uncertainty $\Delta E \sim \hbar/\tau$ , giving a Lorentzian lineshape with linewidth inversely proportional to $\tau$ — shorter-lived states have broader lines. Doppler broadening arises because atoms/molecules in a gas or flame move with a thermal velocity distribution, so each absorbs/emits at a slightly Doppler-shifted frequency depending on its velocity component along the observation axis, producing a Gaussian profile that grows with $\sqrt{T/M}$ (temperature over mass). Pressure (collisional) broadening arises from collisions interrupting the phase of the emitted/absorbed wave, adding an additional Lorentzian contribution that grows with pressure and collision frequency. In atomic absorption spectroscopy, all three combine (as a Voigt profile) to set the practical linewidth that a hollow-cathode lamp's sharp emission line must match for efficient absorption.