Recombination Lines and Nebular Diagnostics

Point a spectrograph at a cloud of ionized gas — an H II region around newborn O stars, the ejected envelope of a dying star lit as a planetary nebula, or the narrow-line region of an active galaxy — and the spectrum returns as a forest of sharp emission lines on a faint continuum. Almost everything we know about the conditions and chemistry of ionized gas across the Universe is read from the ratios of those lines, which come in two physically distinct families answering two different questions. Recombination lines — the hydrogen Balmer series Hα and Hβ, the helium lines — form when a free electron is captured and cascades down the bound levels; their strength counts recombinations, hence ionizations, hence the ionizing-photon budget of the exciting stars. Collisionally excited "forbidden" lines — green O III λλ4959, 5007, red N II and S II, blue O II λ3727 — form when a thermal electron kicks an ion into a metastable level that then radiates; their emissivities depend steeply on electron temperature and density. Divide two recombination lines for dust reddening, two forbidden lines for the electrons or the temperature, and a forbidden line by a recombination line for an ionic abundance. This article assembles that full diagnostic pipeline from atomic physics: it takes the two-level ion and the recombination cascade down to their governing equations and shows exactly how a raw spectrum inverts into density, temperature, reddening, and chemistry.

Key takeaways

Recombination lines (Hα, Hβ, He) are emitted as a captured electron cascades down bound levels; their emissivity scales as $n_e n_p\,\alpha^{\mathrm{eff}}(T_e)$ , so they count ions and the ionizing-photon rate.
Collisionally excited forbidden lines exist only because their metastable upper levels have tiny decay rates, giving critical densities $n_c=A_{ul}/q_{ul}\sim10^2$ – $10^7\,\mathrm{cm^{-3}}$ inside the nebular range.
Two forbidden lines of different excitation energy (O III λ4363/λ5007) read the electron temperature via a Boltzmann factor; two nearly degenerate levels of different critical density (S II λ6716/λ6731) read the density, sliding from 1.45 to 0.44.
The intrinsic Case B ratio Hα/Hβ ≈ 2.86 is nearly temperature-independent, so the observed Balmer decrement measures dust reddening and dereddens every other line.
Dereddened forbidden-to-Hβ ratios then give ionic abundances; summed over ionization stages they give elemental abundances such as O/H.
Key idea: recombination-line strength counts ions, while collisionally excited line ratios read $T_e$ (levels of different excitation energy) and $n_e$ (levels of different critical density) — together they invert a nebular spectrum into density, temperature, reddening, and abundances.

The core insight

Recombination-line strength counts ions (emissivity $\propto n_e n_p$ ), while collisionally excited forbidden-line ratios read $T_e$ from levels of different excitation energy and $n_e$ from levels of different critical density — together they invert nebular spectra into density, temperature, and abundances.

Where this lives today

Nebular diagnostics are the heart of modern extragalactic chemistry. Integral-field surveys — MaNGA, SAMI, MUSE — take a spectrum at every pixel across thousands of galaxies, and this pipeline runs at each one: fit the lines, deredden with the Balmer decrement, read $n_e$ from S II and $T_e$ from O III λ4363, and map oxygen abundance to trace metallicity gradients and the gas flows that build galaxies. The James Webb Space Telescope has pushed the same thermometer past redshift 7, detecting λ4363 less than a billion years after the Big Bang. The BPT diagram — O III λ5007/Hβ against N II λ6583/Hα, each a forbidden line over a nearby recombination line so reddening cancels — remains the standard tool separating star-forming galaxies from active nuclei, and the Hα calibration of Example 3 underpins the cosmic star-formation history — with photoionization codes CLOUDY and MAPPINGS tying it together.

The physical picture

The recombination family is pure accounting. A hot star pours out photons above 13.6 eV; each ionizes a hydrogen atom, and in steady state each is balanced by a recombination. The captured electron rarely lands in the ground state directly; usually it is caught into an excited level and tumbles down the ladder, emitting one photon per rung. Land on $n=2$ and it is a Balmer photon: Hα for $3\to2$ , Hβ for $4\to2$ . So the Hβ flux counts recombinations, hence ionizations, hence the star's ionizing photons — the nebula's ledger of the stellar photon budget, and the reason Hα luminosity became the workhorse tracer of star formation. The second family is stranger. Forbidden lines are magnetic-dipole and electric-quadrupole transitions among the low-lying levels of ions like O $^{++}$ , N $^{+}$ , and S $^{+}$ , "forbidden" only in that $A_{ul}$ is tiny — $10^{-2}$ to a few s $^{-1}$ instead of $10^{8}$ . A collision excites the ion into a metastable level, and a race begins: either it decays radiatively, emitting the forbidden photon, or another collision returns the energy to the thermal pool with no photon. The density where these are equally likely is the critical density $n_c=A_{ul}/q_{ul}$ : below it radiative decay wins and the line is bright, above it collisions quench it. Because $n_c$ for the optical forbidden lines lands inside the nebular range ( $10^2$ – $10^7\,\mathrm{cm^{-3}}$ ), these lines blaze in nebulae and vanish in the laboratory.

Figure

Two families of nebular lines. Left a captured electron cascades down the hydrogen ladder emitting Hβ (4→2) and Hα (3→2) while Lyman photons stay trapped in Case B, fixing the intrinsic ratio Hα/Hβ = 2.86. Right a metastable ion level empties either radiatively at rate A_ul or by collisions at rate nₑ q_ul, the two balancing at the critical density n_c = A_ul/q_ul ≈ 7×10⁵ cm⁻³ for O III λ5007

A temperature diagnostic reads the tail of the Maxwell–Boltzmann distribution. Take two metastable levels of one ion at very different energies — in O $^{++}$ the $^1S_0$ level sits ~2.8 eV above $^1D_2$ . The fraction of electrons energetic enough to reach the higher level goes as $e^{-\Delta E/k_BT_e}$ , so the auroral line from $^1S_0$ (λ4363) brightens sharply relative to the nebular lines from $^1D_2$ (λλ4959, 5007) as the gas warms, while the density dependence cancels because both belong to one ion. A density diagnostic does the opposite: two levels of nearly equal energy but different decay rates, hence different critical densities, whose populations flip as density crosses between the two critical values — so the ratio slides from a low-density to a high-density limit, nearly independent of temperature.

Figure

The two master diagnostics as functions observers invert. Left the S II 6716/6731 doublet ratio slides from its low-density plateau 1.45 down to the high-density limit 0.44 as nₑ crosses ~2×10³ cm⁻³. Right the O III (4959+5007)/4363 ratio falls steeply from ≈480 to ≈40 as T_e rises from 8000 to 20000 K, reading ≈210 at 10⁴ K — one pair weighs the electrons, the other takes the temperature

The core equations

Every recombination line traces back to one captured electron, so its emissivity is the recombination rate per volume times the photon energy, with the atomic physics in an effective coefficient.

j_{\lambda}=\frac{h\nu}{4\pi}\,n_e\,n_p\,\alpha^{\mathrm{eff}}_{\lambda}(T_e) \qquad (1)

where: $j_\lambda$ — volume emissivity (erg s $^{-1}$ cm $^{-3}$ sr $^{-1}$ ); $n_e,\,n_p$ — electron and proton densities; $\alpha^{\mathrm{eff}}_\lambda(T_e)$ — effective recombination coefficient (cm $^3$ s $^{-1}$ ), a slow $\sim T_e^{-0.9}$ function; $h\nu$ — photon energy. Reading: the emissivity is $\propto n_e n_p$ , so the line is a counter of encounters, only weakly temperature-dependent. Integrated along the sightline (with $n_p\approx n_e$ ), the surface brightness is set by the emission measure — the quantity behind free–free emission and pulsar dispersion too.

EM=\int n_e^2\,\mathrm{d}\ell,\qquad I(\mathrm{H}\beta)\propto EM \qquad (2)

where: $EM$ — emission measure (cm $^{-6}$ pc); $\mathrm{d}\ell$ — path length. The forbidden family lives by the competition between radiative and collisional de-excitation; equating the two defines the critical density, with a universal collisional rate coefficient.

n_c=\frac{A_{ul}}{q_{ul}},\qquad q_{ul}=\frac{8.629\times10^{-6}}{g_u\,T_e^{1/2}}\,\Upsilon_{ul}\;\;\mathrm{cm^3\,s^{-1}} \qquad (3)

where: $A_{ul}$ — decay rate $u\to l$ (s $^{-1}$ ); $q_{ul}$ — collisional de-excitation coefficient; $\Upsilon_{ul}$ — thermally averaged collision strength; $g_u$ — upper-level statistical weight. The emitting-level population follows from statistical equilibrium of a two-level ion — linear in $n_e$ when radiation wins, saturated when collisions do.

\frac{n_u}{n_l}=\frac{n_e\,q_{lu}}{A_{ul}+n_e\,q_{ul}}\;\;\xrightarrow{\;n_e\ll n_c\;}\;\;\frac{n_e\,q_{lu}}{A_{ul}} \qquad (4)

where: $n_u,\,n_l$ — level populations; $q_{lu}=(g_u/g_l)\,q_{ul}\,e^{-\Delta E/k_BT_e}$ by detailed balance. Reading: the low-density emissivity inherits the Boltzmann factor $e^{-\Delta E/k_BT_e}$ — the temperature sensitivity that makes forbidden-line ratios thermometers. For the thermometer itself, the classic O III relation packages the two emitting levels of one ion into a near-explicit function of $T_e$ .

\frac{j(\lambda4959)+j(\lambda5007)}{j(\lambda4363)}=\frac{7.90\,\exp\!\left(3.29\times10^{4}/T_e\right)}{1+4.5\times10^{-4}\,n_e\,T_e^{-1/2}} \qquad (5)

where: the exponent carries $\Delta E/k_B=3.29\times10^4\,\mathrm{K}$ (the $^1S_0$ – $^1D_2$ gap, ≈2.8 eV); the small $n_e T_e^{-1/2}$ term is negligible until $n_e$ nears the λ4363 critical density $\sim10^7\,\mathrm{cm^{-3}}$ . Reading: below that, the ratio is a steep, nearly pure function of $T_e$ — the standard direct-method thermometer. The density gauge uses two nearly degenerate levels of one ion, cancelling temperature and isolating $n_e$ through the different critical densities.

\frac{I(\lambda6716)}{I(\lambda6731)}\;:\;1.45\;\xrightarrow{\;n_e:\,0\to\infty\;}\;0.44 \qquad (6)

where: λ6716 from $^2D_{5/2}$ ( $A\approx2.6\times10^{-4}\,\mathrm{s^{-1}}$ , $n_c\approx1.5\times10^3\,\mathrm{cm^{-3}}$ ), λ6731 from $^2D_{3/2}$ ( $A\approx8.8\times10^{-4}\,\mathrm{s^{-1}}$ , $n_c\approx3.9\times10^3\,\mathrm{cm^{-3}}$ ). With $n_e$ and $T_e$ fixed, a dereddened forbidden-to-Hβ ratio gives an ionic abundance — Hβ counts H $^+$ by recombination, the forbidden line counts X $^{i+}$ by collisions.

\frac{n(\mathrm{X}^{i+})}{n(\mathrm{H}^{+})}=\frac{I(\lambda)}{I(\mathrm{H}\beta)}\,\frac{\alpha^{\mathrm{eff}}_{\mathrm{H}\beta}\,h\nu_{\mathrm{H}\beta}}{\varepsilon_\lambda(n_e,T_e)} \qquad (7)

where: $\varepsilon_\lambda(n_e,T_e)$ — emissivity per ion of the forbidden line; $I(\lambda)/I(\mathrm{H}\beta)$ — dereddened ratio. Reading: summing ionization stages with an ionization correction factor gives the elemental abundance X/H. Before any of that, fluxes must be dereddened — and the recombination family supplies its own dust gauge, the Balmer decrement.

c(\mathrm{H}\beta)=\frac{1}{f(\mathrm{H}\beta)-f(\mathrm{H}\alpha)}\,\log_{10}\!\left[\frac{(I_{\mathrm{H}\alpha}/I_{\mathrm{H}\beta})_{\mathrm{obs}}}{2.86}\right] \qquad (8)

where: $c(\mathrm{H}\beta)$ — logarithmic extinction at Hβ; $f(\lambda)$ — reddening curve, $f(\mathrm{H}\beta)=0$ and $f(\mathrm{H}\alpha)\approx-0.34$ , so the denominator $\approx+0.34$ is positive; 2.86 — intrinsic Case B ratio. Reading: dust dims blue Hβ more than red Hα, pushing the observed ratio above 2.86; the excess yields the extinction applied to every line.

Derivation AEmissivity of a two-level ion and its density and temperature limits

The goal is to derive the emissivity of a collisionally excited forbidden line and extract the two limits that turn line ratios into diagnostics: a low-density regime where the emissivity is linear in $n_e$ and carries a Boltzmann factor (so different-energy levels read temperature), and a high-density regime where the level saturates to LTE and the line is quenched. Model the ion as a ground level $l$ and one metastable level $u$ split by $\Delta E=h\nu_{ul}$ , in a thermal electron gas at density $n_e$ , temperature $T_e$ , radiative pumping negligible.

Step 1. Statistical equilibrium of the upper level equates excitations up to de-excitations down by both channels,

n_l\,n_e\,q_{lu}=n_u\,(A_{ul}+n_e\,q_{ul}).

Why: in steady state what goes up comes down, radiatively ( $A_{ul}$ ) or collisionally ( $n_e q_{ul}$ ).

Step 2. Solve for the upper population with $n_l\approx n_{\mathrm{ion}}$ (almost every ion is in the ground term),

n_u=n_{\mathrm{ion}}\,\frac{n_e\,q_{lu}}{A_{ul}+n_e\,q_{ul}}.

Why: the two denominator terms are the two competing fates of the excited ion, and they contain both diagnostic limits.

Step 3. Each radiative decay emits one photon, so the emissivity is $A_{ul}n_u$ times $h\nu_{ul}/4\pi$ ,

j_{ul}=\frac{h\nu_{ul}}{4\pi}\,A_{ul}\,n_u=\frac{h\nu_{ul}}{4\pi}\,n_{\mathrm{ion}}\,\frac{n_e\,q_{lu}\,A_{ul}}{A_{ul}+n_e\,q_{ul}}.

Why: only the radiative channel makes photons; collisions silently return the energy.

Step 4. Low-density limit $n_e\ll n_c$ with detailed balance,

j_{ul}\;\longrightarrow\;\frac{h\nu_{ul}}{4\pi}\,n_{\mathrm{ion}}\,n_e\,q_{lu}\;\propto\;n_e\,e^{-\Delta E/k_BT_e}.

Why: every excitation becomes a photon, so the line is a collision counter stamped with the temperature-sensitive Boltzmann factor — the basis of the $T_e$ diagnostic.

Step 5. High-density limit $n_e\gg n_c$ , where the level reaches its Boltzmann value,

j_{ul}\;\longrightarrow\;\frac{h\nu_{ul}}{4\pi}\,n_{\mathrm{ion}}\,\frac{g_u}{g_l}\,e^{-\Delta E/k_BT_e}\,A_{ul}.

Why: the line saturates and is density-independent (collisional quenching); a density pair of equal $\Delta E$ but different $A_{ul}$ keeps $n_e$ -sensitivity while the Boltzmann factors cancel. *Verified: reduces to the standard two-level result (Osterbrock & Ferland ch. 3); O III λ5007 gives $n_c=A/q\approx7\times10^5\,\mathrm{cm^{-3}}$ from $A\approx0.028\,\mathrm{s^{-1}}$ , $\Upsilon\approx2.3$ , $g_u=5$ , $T_e=10^4$ K.*

Go deeper · Derivation B · 1 minCase B recombination and the Balmer decrement

Derivation BCase B recombination and the Balmer decrement

The goal is to show why the intrinsic Hα/Hβ ratio is a fixed number ≈2.86 that barely moves with temperature, and to turn that constancy into a dust-reddening measurement. The key is Case B: the gas is optically thick to the Lyman lines, so each is reabsorbed and reprocessed and only cascades terminating on $n\ge2$ escape — fixing the cascade branching and making the Balmer ratios pure atomic constants.

Step 1. Form the ratio of Hα and Hβ emissivities from equation (1); the densities cancel,

\frac{I_{\mathrm{H}\alpha}}{I_{\mathrm{H}\beta}}=\frac{\alpha^{\mathrm{eff}}_{\mathrm{H}\alpha}\,\nu_{\mathrm{H}\alpha}}{\alpha^{\mathrm{eff}}_{\mathrm{H}\beta}\,\nu_{\mathrm{H}\beta}}.

Why: sharing the $n_e n_p$ prefactor, the intrinsic ratio depends only on atomic coefficients and frequencies, not on how much gas there is.

Step 2. Insert Case B coefficients at $10^4$ K ( $\alpha^{\mathrm{eff}}_{\mathrm{H}\alpha}\approx1.17\times10^{-13}$ , $\alpha^{\mathrm{eff}}_{\mathrm{H}\beta}\approx3.03\times10^{-14}\,\mathrm{cm^3\,s^{-1}}$ ) with $\nu_{\mathrm{H}\alpha}/\nu_{\mathrm{H}\beta}=4861/6563$ ,

\left(\frac{I_{\mathrm{H}\alpha}}{I_{\mathrm{H}\beta}}\right)_{\mathrm{int}}\approx3.86\times0.741\approx2.86.

Why: the cascade favors Hα, and the coefficients pin the anchor value 2.86.

Step 3. The coefficients each scale as $\sim T_e^{-0.9}$ but nearly in lockstep, so the ratio drifts only from ≈3.04 at 5000 K to ≈2.75 at 20000 K. Why: a diagnostic that barely moves with the unknown temperature is what a reddening standard needs.

Step 4. Compare observed to intrinsic through the reddening law and invert for the extinction (equation 8),

\left(\frac{I_{\mathrm{H}\alpha}}{I_{\mathrm{H}\beta}}\right)_{\mathrm{obs}}=2.86\times10^{-c(\mathrm{H}\beta)\,[f(\mathrm{H}\alpha)-f(\mathrm{H}\beta)]}.

Why: with $f(\mathrm{H}\alpha)<f(\mathrm{H}\beta)=0$ the exponent is positive, so reddening raises the observed ratio; solving gives $c(\mathrm{H}\beta)$ , which then dereddens every flux before any abundance ratio. *Verified: intrinsic ratio 2.86 at $T_e=10^4$ K, $n_e=10^2\,\mathrm{cm^{-3}}$ (Osterbrock & Ferland Table 4.4); sign convention gives positive extinction, matching standard reddening formulae.*

Limiting cases

Everything hinges on where $n_e$ sits relative to the critical density $n_c=A_{ul}/q_{ul}$ ; the recombination lines add a fixed reference point.

Case	Result	Meaning
Low-density $n_e\ll n_c$	$j_{ul}\propto n_e\,n_{\mathrm{ion}}\,e^{-\Delta E/k_BT_e}$	Every excitation yields a photon; a pure collision counter. Different- $\Delta E$ ratios give $T_e$ .
High-density $n_e\gg n_c$	$j_{ul}\propto n_{\mathrm{ion}}\,\tfrac{g_u}{g_l}e^{-\Delta E/k_BT_e}A_{ul}$	Boltzmann (LTE) populations; the line saturates and is collisionally quenched.
Case B recombination	$I(\mathrm{H}\alpha)/I(\mathrm{H}\beta)\approx2.86$ at $10^4$ K	Optically thick to Lyman lines; the ratio barely varies with $T_e$ , a clean reddening probe.
Density gauge saturated	$n_e\gtrsim n_c(\mathrm{S\,II})\sim4\times10^3\,\mathrm{cm^{-3}}$	S II, O II doublets flatten; switch to higher- $n_c$ pairs (Ar IV, C III] UV, IR lines).
Metal-rich / cool nebula	$T_e\lesssim7000$ K, λ4363 lost	Cooling hides the auroral line; fall back on strong-line calibrations (R $_{23}$ , N2, O3N2).

Common trapTreating every bright optical line by the same rule

The commonest blunder is forgetting that the two families obey opposite rules. Recombination-line ratios (Hα/Hβ, He I/Hβ) are nearly $T_e$ - and $n_e$ -independent and set the reddening and hydrogen scales; collisionally excited ratios are steep in $T_e$ and $n_e$ and set the physical conditions. Three failures follow: applying a density diagnostic above its critical density (S II beyond $\sim4\times10^3\,\mathrm{cm^{-3}}$ ) returns a meaningless flattened ratio; using a temperature diagnostic where the auroral line is partly quenched biases $T_e$ high; and forming an abundance ratio from observed rather than dereddened fluxes underestimates the bluer lines and corrupts every metallicity — the Balmer decrement must be applied first, always.

Worked examples

1 · Dust reddening from the Balmer decrement (fully worked)

A giant H II region has observed $\mathrm{H}\alpha/\mathrm{H}\beta=4.3$ ; Case B predicts 2.86. Find $c(\mathrm{H}\beta)$ , $E(B-V)$ , $A_V$ , and the correction to the λ5007 flux.

Step 1. The excess over 2.86 is dust: $\log_{10}(4.3/2.86)=\log_{10}(1.503)=\mathbf{0.177}$ . Positive confirms reddening.

Step 2. Divide by $f(\mathrm{H}\beta)-f(\mathrm{H}\alpha)\approx0.34$ (equation 8): $c(\mathrm{H}\beta)=0.177/0.34=\mathbf{0.52}$ — Hβ dimmed by $10^{0.52}\approx3.3\times$ .

Step 3. With the $R_V=3.1$ curve, $E(B-V)\approx1.97\times0.177=\mathbf{0.35}$ mag, hence $A_V=3.1\,E(B-V)=\mathbf{1.1}$ mag.

Step 4. λ5007 lies close to Hβ ( $f\approx-0.04$ ), so its correction is only $10^{0.52\times0.04}\approx1.05$ — 5%. The decrement 4.3 gives $c(\mathrm{H}\beta)=0.52$ , $E(B-V)=0.35$ , $A_V=1.1$ mag; nearby lines need small corrections but blue O II λ3727 ( $f\approx0.26$ ) is boosted ~37% — which is why diagnostic pairs are chosen close in wavelength.

2 · The critical density of O III λ5007 (fully worked)

Compute the critical density of the O III $^1D_2$ level at $T_e=10^4$ K.

Step 1. Atomic data: $A_{ul}=A(5007)+A(4959)\approx0.0206+0.0068\approx\mathbf{0.028\;s^{-1}}$ ; $\Upsilon\approx2.3$ ; $g_u=5$ .

Step 2. From equation (3): $q_{ul}=(8.629\times10^{-6}\times2.3)/(5\times100)=\mathbf{4.0\times10^{-8}\;cm^3\,s^{-1}}$ .

Step 3. $n_c=A_{ul}/q_{ul}=0.028/(4.0\times10^{-8})\approx\mathbf{7\times10^{5}\;cm^{-3}}$ . Far above the $10^2$ – $10^4\,\mathrm{cm^{-3}}$ of H II regions, so λ5007 is essentially never quenched and is the brightest optical coolant. The auroral λ4363 ( $^1S_0$ ) has $n_c\sim10^7$ , higher still — which is why the λ4363/λ5007 thermometer stays density-independent across the nebular range.

3 · Ionizing photons and star-formation rate from Hα (fully worked)

A dust-corrected H II region has $L(\mathrm{H}\alpha)=10^{41}\,\mathrm{erg\,s^{-1}}$ . Estimate $Q(\mathrm{H}^0)$ and the star-formation rate.

Step 1. Each ionizing photon is matched by a recombination, a fixed fraction of which yield Hα: $\alpha^{\mathrm{eff}}_{\mathrm{H}\alpha}/\alpha_B\approx1.17\times10^{-13}/2.59\times10^{-13}\approx\mathbf{0.45}$ per recombination.

Step 2. With $h\nu_{\mathrm{H}\alpha}=3.03\times10^{-12}\,\mathrm{erg}$ (1.89 eV): $Q(\mathrm{H}^0)=\dfrac{L(\mathrm{H}\alpha)}{0.45\,h\nu_{\mathrm{H}\alpha}}=\dfrac{10^{41}}{0.45\times3.03\times10^{-12}}\approx\mathbf{7.3\times10^{52}\;s^{-1}}$ .

Step 3. Kennicutt: $\mathrm{SFR}=7.9\times10^{-42}\,L(\mathrm{H}\alpha)\approx\mathbf{0.79\;M_\odot\,yr^{-1}}$ . The nebula needs ~ $7\times10^{52}$ ionizing photons per second and traces ~ $0.8\,M_\odot\,\mathrm{yr^{-1}}$ — but only after the Example 1 dust correction, without which the SFR is underestimated.

Physics of the portrait

For sixty years the brightest nebular lines had no home in the periodic table — the green λ5007 line was even assigned to a hypothetical element, "nebulium." The resolution proved to be not a new element but a new regime: transitions so slow they can radiate only in a near-perfect vacuum.

Origins · 1 minIra Sprague Bowen and how this was discovered

Ira Sprague Bowen

1898–1973 · identification of the nebular forbidden lines

Ira Sprague Bowen, an American physicist trained in Robert Millikan's spectroscopy laboratory at Caltech, cracked a puzzle carried since 1864, when William Huggins found that planetary nebulae shine in bright emission lines dominated by a green pair near 5007 and 4959 Å with no laboratory counterpart. In 1927 Bowen showed there was no nebulium: the lines were forbidden magnetic-dipole and electric-quadrupole transitions among the low-lying levels of doubly-ionized oxygen and singly-ionized oxygen and nitrogen — transitions with rates so tiny they are collisionally quenched at laboratory density but radiate freely in the near-vacuum of a nebula. His identification turned the strongest nebular lines from a mystery into the field's most powerful diagnostics: the λ4363/λ5007 thermometer and every abundance measurement rest on knowing which atomic levels those lines come from. Bowen later designed the Palomar spectrographs and directed the Mount Wilson and Palomar Observatories; his 1927 paper remains a model of laboratory atomic physics unlocking an astrophysical spectrum.

History

1864

— Huggins finds that planetary nebulae shine in bright emission lines, opening the nebulium puzzle

1927

— Bowen identifies the nebular lines as forbidden transitions of ionized oxygen and nitrogen

1938

— Baker and Menzel formulate Case A and Case B hydrogen recombination and the cascade theory

1981

— Baldwin, Phillips and Terlevich introduce the BPT emission-line classification diagram

2006

— Osterbrock and Ferland publish the second edition of the standard nebular reference

Going further

Texts. Osterbrock & Ferland, Astrophysics of Gaseous Nebulae and Active Galactic Nuclei, 2nd ed. — the definitive treatment (cascades ch. 4, diagnostics ch. 5) · Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., §12.1–12.3 · Draine, Physics of the Interstellar and Intergalactic Medium, ch. 14–18 · Dopita & Sutherland, Astrophysics of the Diffuse Universe · Binney & Tremaine, Galactic Dynamics, 2nd ed. · Ryden, Introduction to Cosmology (cosmological recombination as background).

Provenance. The two-level equilibrium result, $n_c=A_{ul}/q_{ul}$ , and the rate prefactor $8.629\times10^{-6}$ follow Osterbrock & Ferland (2006) chs. 3–5 and Draine (2011) chs. 14–18. Case B Hα/Hβ = 2.86 at $T_e=10^4$ K, $n_e=10^2\,\mathrm{cm^{-3}}$ is Osterbrock & Ferland Table 4.4; $\alpha^{\mathrm{eff}}_{\mathrm{H}\alpha}=1.17\times10^{-13}$ , $\alpha^{\mathrm{eff}}_{\mathrm{H}\beta}=3.03\times10^{-14}$ , $\alpha_B=2.59\times10^{-13}\,\mathrm{cm^3\,s^{-1}}$ are standard. Equation (5) is the Osterbrock form with $\Delta E/k_B=3.29\times10^4$ K; S II $n_c(6716)\approx1.5\times10^3$ , $n_c(6731)\approx3.9\times10^3\,\mathrm{cm^{-3}}$ with limits 1.45 and 0.44. Note the sign in equation (8): with $f(\mathrm{H}\beta)=0$ , $f(\mathrm{H}\alpha)\approx-0.34$ , the denominator $>0$ gives positive extinction. A verifier should confirm the two-level density limits, the 2.86 ratio, $n_c(\mathrm{O\,III}\,\lambda5007)\approx7\times10^5\,\mathrm{cm^{-3}}$ , the worked reddening ( $c(\mathrm{H}\beta)=0.52$ ), and the Hα-to-SFR conversion.

Prerequisites. Phases of the Interstellar Medium · Interstellar Gas Heating and Cooling Processes · Interstellar Extinction and Reddening. Leads to. Interstellar Gas Heating and Cooling Processes · Thermal Instability and the Two-Phase Medium · Interstellar Dust Grains.

Ask the tutor: Ask me to walk a real SDSS or JWST emission-line spectrum through the full pipeline — deredden, weigh the electrons, take the temperature, and turn it into an oxygen abundance — one step at a time.

What is the difference between Case A and Case B recombination?

Both describe how a captured electron cascades down the hydrogen levels; they differ in the fate of Lyman-series photons. Case A assumes the nebula is optically thin to all lines, so every photon escapes. Case B assumes it is optically thick to the Lyman lines (realistic for H II regions and planetary nebulae), so a Lyman photon is reabsorbed on the spot and reprocessed until it emerges as a lower-series photon plus Lyα. Case B thus suppresses recombination to $n=1$ and slightly boosts the Balmer lines: Hα/Hβ is 2.86 in Case B versus about 2.74 in Case A at $10^4$ K. Essentially all nebular work uses Case B.

Why are λ4363 and λ5007 called "auroral" and "nebular" lines?

The names come from the levels. In O $^{++}$ the ground configuration has three terms: $^3P$ (ground), $^1D_2$ , and $^1S_0$ highest. Transitions from $^1D_2$ to $^3P$ give the strong green "nebular" lines λλ4959, 5007; the transition from $^1S_0$ down to $^1D_2$ gives the fainter "auroral" line λ4363, named for the analogous transitions that light the terrestrial aurora. Because λ4363 comes from a level ~2.8 eV higher, its strength relative to the nebular lines is exponentially sensitive to temperature — which makes the auroral-to-nebular ratio the standard direct thermometer.

Why measure density before temperature, and temperature before abundance?

Because you unwind the diagnostics in order. The temperature ratio (equation 5) has a small density correction, so you first read $n_e$ from S II or O II, then nail $T_e$ from O III λ4363/λ5007. Abundance (equation 7) then depends on the emissivity $\varepsilon_\lambda(n_e,T_e)$ , which carries a Boltzmann factor — so a wrong $T_e$ propagates exponentially, shifting O/H by ~0.1 dex for a 10% temperature error. The chain is deredden, $n_e$ , $T_e$ , then abundances; reordering corrupts the answer.

What happens when the gas is denser than the critical density?

The line saturates and stops working as written. Below $n_c$ the emissivity grows linearly with density; above $n_c$ collisions de-excite the level before it radiates, the population reaches its Boltzmann value, and the emissivity per ion goes density-independent — the line is collisionally quenched. For the density gauges this means S II and O II ratios flatten above a few $\times10^3\,\mathrm{cm^{-3}}$ and lose their grip on $n_e$ . To probe denser gas you move to higher-critical-density pairs: Ar IV or Fe III in the optical, C III] in the ultraviolet, or infrared fine-structure lines reaching $n_c\sim10^5$ – $10^6\,\mathrm{cm^{-3}}$ .

Why are diagnostic line pairs chosen close together in wavelength?

Because reddening then nearly cancels between the two lines, so the ratio is robust even before a perfect dust correction: attenuation is smooth in wavelength, so two lines tens of angstroms apart are dimmed almost equally. That is why the S II doublet and the O III lines make robust diagnostics, and why the BPT diagram pairs O III λ5007 with nearby Hβ and N II λ6583 with nearby Hα. The Balmer decrement is the deliberate exception: it uses a wide baseline (6563 versus 4861) precisely so that reddening does not cancel — that spread is the signal it measures.

How does helium fit in, and why does it matter cosmologically?

Helium has its own recombination lines (He I λ5876, λ4471; He II λ4686 in harder fields) forming by the same cascade physics, so their strengths relative to Hβ give the helium-to-hydrogen ratio once $T_e$ and $n_e$ are known. This reaches beyond nebulae: measuring the primordial helium abundance in the most metal-poor H II regions — gas almost untouched by stellar processing — tests Big Bang nucleosynthesis and constrains the baryon density and the number of light neutrino species. The same accounting that weighs ionized gas nearby becomes, in the lowest-metallicity nebulae, a probe of the first three minutes of the Universe.