The Metastable Helium 1083 nm Triplet: Tracing Exoplanet Atmospheric Escape

What the 1083 nm triplet is: a forbidden level of orthohelium

Neutral helium has two electrons whose spins can be paired (parahelium, singlet states) or aligned (orthohelium, triplet states). The lowest triplet level, denoted 2³S, sits about 19.8 eV above the ground state. Crucially, it is metastable: a radiative decay from 2³S down to the 1¹S ground state would require both flipping an electron spin and emitting a photon via a forbidden transition, so the state is extraordinarily long-lived — its radiative lifetime is roughly 2.2 hours (~7,900 s), versus nanoseconds for an ordinary excited level.

• The observable feature is the 2³S → 2³P transition at 1083.3 nm in the near-infrared.
• It is a triplet because the upper 2³P level is split by fine structure into three sublevels, giving lines at about 10832.06, 10833.22, and 10833.31 Å — the latter two blend at low resolution.
• Because 2³S is a long-lived reservoir, even a small fraction of helium parked there can produce a deep, measurable absorption line.

This same triplet is familiar in the Sun, where it forms the chromospheric He I 10830 line used in solar physics.

The mechanism: XUV builds the reservoir, mid-UV drains it

The population of the 2³S level is set by a competition of atomic processes, worked out for exoplanets by Oklopčić & Hirata (2018). The dominant channel is photoionization followed by recombination:

• Stellar XUV and extreme-UV photons (energies above 24.6 eV) ionize helium in the planet's upper atmosphere, creating He⁺ and free electrons.
• When He⁺ recombines with an electron, roughly 3/4 of recombinations cascade into triplet states, funneling atoms into the metastable 2³S reservoir — more efficient than direct collisional excitation in most planetary winds.
• The reservoir is drained by (i) further photoionization of the 2³S atoms by near-UV/optical photons (the 2³S ionization threshold is only ~4.8 eV, i.e. wavelengths shortward of ~2600 Å), (ii) electron collisions, and (iii) the slow forbidden radiative decay.

The key scaling insight: metastable helium is abundant when a star emits hard XUV but relatively little mid-UV. This is why K dwarfs — hot coronae, modest near-UV — are the best hosts, while quiet, mid-UV-bright stars can wash the signal out entirely.

Characteristic numbers and a worked example

Consider WASP-107b, the discovery target: a ~0.1 M_Jupiter, Neptune-mass planet with a Jupiter-like radius (a "super-puff") orbiting a K6 dwarf in 5.7 days. Spake et al. (2018) measured an excess absorption of 0.049 ± 0.011% in the helium band. Translating a helium signal into a mass-loss rate uses an energy budget or Parker-wind model of the outflow.

• Inferred mass-loss rate: about 10¹⁰ – 3×10¹¹ g s⁻¹, i.e. roughly 0.1–4% of the planet's mass per gigayear.
• Thermosphere temperatures probed: typically 6,000–13,000 K, hot enough to drive a transonic hydrodynamic wind.
• Outflow velocities inferred from line broadening and blueshifts: up to several to ~10 km/s in the bulk, with high-velocity tails to tens of km/s.

The energy-limited escape estimate scales as Ṁ ∝ ε·F_XUV·R_p³ / (G·M_p), where F_XUV is the stellar high-energy flux, R_p and M_p the planet radius and mass, and ε the heating efficiency — so low-density, low-gravity planets under bright XUV stars lose mass fastest.

How it is observed: high-resolution transmission spectroscopy

The line is measured during a transit: as the planet crosses the star, its extended, helium-bearing exosphere absorbs an extra sliver of light at 1083 nm, deepening in and out of transit. Because 2³S atoms extend well beyond the optical planetary disk — often to several planetary radii and into a comet-like tail — the helium transit can be deeper and longer than the white-light transit.

• Space: the first detection used HST/WFC3's G102 grism at low resolution. JWST (NIRSpec) has since resolved the line and tracked the WASP-107b outflow in detail.
• Ground: high-resolution near-IR spectrographs now dominate — CARMENES (Calar Alto), SPIRou and GIANO (CFHT/TNG), Keck/NIRSPEC, and the Habitable-zone Planet Finder. High resolution resolves the triplet, its Doppler shift, and its shape.

Early ground-based confirmations came rapidly: Nortmann et al. (2018) detected helium escaping WASP-69b with CARMENES, and Allart et al. (2018) mapped the outflow of HD 189733b. Over 50 helium-triplet transit campaigns now exist, and the line has overtaken Lyman-α as the field's workhorse escape tracer.

How it compares to Lyman-α, H-alpha, and metal lines

Atmospheric escape can be probed with several tracers, each with a different niche:

• Lyman-α (121.6 nm): the historic tracer (HD 209458b, 2003), sensitive to neutral hydrogen but observable only from space, and its line core is destroyed by interstellar absorption and airglow, so only the far wings survive.
• H-alpha (656.3 nm): ground-based, but requires enough stellar Lyα to pump hydrogen into n=2; best for ultra-hot planets around active stars.
• Metal lines (Fe, Mg, Na, C, O in the UV/optical): reveal the composition of the escaping gas but usually need space UV or exceptional conditions.

The helium triplet's edge is practical: a clean near-IR window, ground accessibility, and a formation physics tied directly to stellar XUV heating — the very driver of escape. Its main caveat is that a null result is ambiguous: no helium signal can mean little escape, an unfavorable stellar SED (too much mid-UV), a low He/H ratio, or high-altitude clouds — not necessarily a stable atmosphere.

Significance, famous cases, and open questions

The helium triplet has turned photoevaporation — long invoked to explain the "radius valley" and hot-Neptune desert in the exoplanet population — into something directly observable. By watching planets shed atmosphere in real time, it tests models of how sub-Neptunes may strip down to rocky super-Earths.

• WASP-107b: the discovery world; JWST later tracked its extended, escaping helium in unprecedented detail, refining the mass-loss picture.
• WASP-69b: shows a spectacular comet-like helium tail trailing the planet.
• HAT-P-11b, GJ 3470b, HD 189733b: a growing sample spanning warm Neptunes to hot Jupiters.

Open questions include: reconciling helium-derived and Lyα-derived mass-loss rates; the true helium-to-hydrogen abundance in these winds (fractionation may deplete helium); the strong and sometimes surprising dependence on stellar activity and coronal composition; and why some expected detections (e.g. WASP-80b) come up empty. Non-detections around active stars now hint that stellar high-energy spectra, not just planet properties, control the signal.

Atmospheric-escape tracers compared: metastable helium 1083 nm versus hydrogen lines

Property	He I 1083 nm triplet	H I Lyman-α (121.6 nm)	H-alpha (656.3 nm)
Wavelength band	Near-IR	Far-UV	Optical
Observable from ground?	Yes (clean NIR window)	No (absorbed by atmosphere)	Yes
ISM / geocorona contamination	Negligible	Severe (core absorbed)	Low
What populates the level	XUV photoionization + recombination into 2³S	Ground-state H in the wind	Excited n=2 hydrogen (needs strong Lyα pumping)
Depends strongly on	Stellar XUV-to-mid-UV flux ratio, He/H abundance	Stellar Lyα flux, neutral H column	High Lyα flux, active stars
First exoplanet detection	WASP-107b, 2018	HD 209458b, 2003 (Vidal-Madjar)	HD 189733b, 2010s