Exoplanets
Transmission Spectroscopy
Sniffing an exoplanet’s air during transit
Transmission spectroscopy is the technique of reading which colors of starlight an exoplanet’s atmosphere absorbs as the planet crosses its star, turning those missing wavelengths into a chemical fingerprint of the air. During transit, a thin ring of atmosphere — only tens of kilometers thick — is backlit by the stellar disk. Molecules in that ring absorb specific wavelengths, so the planet looks fractionally larger at those colors, deepening the transit by a few hundred parts per million. The resulting curve of transit depth versus wavelength is the transmission spectrum, and its peaks identify water, carbon dioxide, methane, sodium and more.
- Signal size (hot Jupiter)~100–1000 ppm extra depth
- Signal size (rocky/M dwarf)~10–50 ppm
- Atmosphere probedterminator annulus, ~5–10 scale heights
- Key moleculesH₂O, CO₂, CH₄, CO, SO₂, Na, K
- Workhorse instrumentsHST/WFC3, JWST NIRSpec & NIRISS (0.6–12 µm)
- First clear detectionNa in HD 209458 b (Charbonneau et al. 2002)
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
What transmission spectroscopy actually measures
When a planet transits, it blocks a fraction of the starlight equal to the ratio of areas, (Rp/Rstar)². That number is the transit depth. The crucial insight behind transmission spectroscopy is that the planet’s effective radius is not a single value — it depends on wavelength. At colors where the atmosphere is transparent, starlight skims close to the solid (or cloud-top) surface and the planet looks small. At colors where a molecule absorbs strongly, that same starlight is blocked higher up, so the opaque silhouette grows and the transit gets deeper.
Plot transit depth against wavelength and you get the transmission spectrum: a baseline set by the bulk planet radius, with bumps wherever molecules in the atmosphere soak up specific wavelengths. Each bump is absorption by a particular species, and the pattern of bumps fingerprints the molecules present. The geometry that makes this possible is special: we only ever sample the terminator — the thin day–night boundary ring backlit by the star — never the planet’s full disk.
Why the signal is so small — scale height
The extra depth from the atmosphere is set by how thick the absorbing annulus is compared to the star. The natural thickness unit is the atmospheric scale height, H = kBT / (µ g), the altitude over which pressure drops by a factor of e. Hot, low-gravity, low-molecular-weight (hydrogen-rich) atmospheres are puffy and have large H; cool, high-gravity, heavy (CO₂- or N₂-dominated) atmospheres are compact.
A useful estimate for the wavelength-dependent change in transit depth is
Δδ ≈ 2 Rp · NH · H / Rstar²
where NH ≈ 5–10 is how many scale heights a strong band probes. For a hot Jupiter (Rp ≈ 1.4 RJup, T ≈ 1300 K, H₂/He air with µ ≈ 2.3) around a Sun-like star, H ≈ 500–600 km and Δδ lands near a few hundred parts per million — measurable. For an Earth-sized planet with a heavy nitrogen atmosphere around a Sun-like star, H shrinks to ~8 km and the signal collapses below 10 ppm, which is why temperate-rocky atmospheres are studied mostly around small, nearby M dwarfs, where Rstar is small and the ratio is larger.
| Target type | Scale height H | Atmosphere µ | Extra depth Δδ | Feasible with |
|---|---|---|---|---|
| Hot Jupiter (Sun-like star) | ~500 km | ~2.3 (H₂/He) | ~200–500 ppm | HST, JWST, ground |
| Warm Neptune | ~150–300 km | ~2.5–4 | ~50–150 ppm | HST, JWST |
| Sub-Neptune (M dwarf) | ~100 km | ~2–18 (uncertain) | ~30–100 ppm | JWST |
| Rocky, heavy air (M dwarf) | ~8–15 km | ~28–44 (N₂/CO₂) | ~10–30 ppm | JWST (multi-transit) |
| Rocky, heavy air (Sun-like star) | ~8 km | ~29 (N₂/O₂) | < 5 ppm | not yet feasible |
How a spectrum is extracted
The observation is conceptually a division. You record the spectrum of the star continuously — for hours — covering the out-of-transit baseline and the full transit. Out of transit you see only the star. In transit you see the star minus the planet’s opaque disk, plus the wavelength-dependent extra blocking from the atmospheric ring. Dividing the in-transit spectrum by the out-of-transit baseline removes the star’s intrinsic colors and the instrument’s throughput, leaving the transit depth in each wavelength channel.
In practice every spectral channel is treated as its own transit light curve. Fitting the depth of each light curve gives the apparent planet radius at that wavelength; stacking them produces the spectrum. Because the per-channel signal is buried under stellar photon noise and instrument systematics, the technique rewards bright host stars, long baselines, and repeated transits that can be co-added.
- Optical, low resolution (HST STIS, ground): alkali lines (Na 589 nm, K 770 nm), Rayleigh scattering slope toward the blue, broad cloud signatures.
- Near-IR, low resolution (HST WFC3, JWST NIRISS/NIRSpec): water at 1.4 & 1.9 µm, CO₂ at 4.3 µm, CH₄ at 3.3 µm, CO at 4.7 µm, SO₂ at 4 µm.
- High-resolution, ground (R > 25,000; CRIRES+, ESPRESSO): resolves individual lines and uses the planet’s orbital Doppler shift to separate planetary lines from stellar and telluric ones, and to clock terminator winds.
Clouds, hazes, and stellar contamination
Two effects routinely flatten or fake features. Aerosols — condensate clouds or photochemical hazes — act as gray absorbers high in the atmosphere, raising the opaque radius and clipping off the bottom of molecular bands. A high cloud deck can mute a water feature by a factor of several, which is exactly what muddied many early hot-Jupiter spectra. JWST has since pushed deep enough, and to long enough wavelengths, to recover features that clouds had hidden — and even to detect silicate clouds directly (e.g., WASP-17 b).
The second effect is the transit light source effect: the planet only blocks the part of the star directly behind it, but the spectrum is normalized against the whole, possibly spotted, stellar disk. Unocculted spots and plages stamp stellar absorption features into the data that can be mistaken for planetary water or distort retrieved abundances. For active M dwarfs this contamination can rival the planetary signal, so modern analyses fit a stellar heterogeneity model jointly with the atmosphere.
Where it fits among atmosphere methods
| Method | When observed | What it probes | Region sampled |
|---|---|---|---|
| Transmission spectroscopy | During transit (planet in front) | Absorption fingerprints, scale height, winds | Terminator ring, upper atmosphere |
| Secondary-eclipse / emission | Planet behind star | Thermal emission, dayside temperature, inversions | Dayside, deeper layers |
| Phase curve | Full orbit | Day–night contrast, heat transport, hot-spot offset | Whole planet vs. orbital phase |
| Direct-imaging spectroscopy | Planet spatially separated | Full-disk emission/reflection spectrum | Wide-orbit, self-luminous planets |
Transmission spectroscopy is uniquely suited to the upper, low-pressure layers near the terminator and to molecules that are spectrally active in transmitted light. Combined with emission and phase-curve data, it builds a three-dimensional picture of an alien atmosphere from a single unresolved dot of light.
Milestones
- 2002 — first detection. HST sees neutral sodium in the atmosphere of hot Jupiter HD 209458 b, the first molecule/atom detected in an exoplanet atmosphere by transmission.
- 2007–2018 — the water era. HST WFC3 maps the 1.4 µm water band across dozens of hot Jupiters, revealing a continuum from cloudy to clear and abundances often below solar.
- 2022 — JWST opens up. First JWST transmission spectra deliver an unambiguous CO₂ detection (WASP-39 b), then SO₂ from photochemistry, plus CO and water across one spectrum.
- 2023–present — small planets. JWST begins probing sub-Neptunes and rocky worlds around M dwarfs, where heavy atmospheres and stellar activity make every part per million count.
Common misconceptions
- It images the planet. No — the planet is an unresolved point; the spectrum comes from a brightness ratio over time.
- It sees the whole atmosphere. Only the terminator ring, and only the upper layers above any cloud deck.
- A flat spectrum means no atmosphere. It can equally mean high clouds masking the features, or a heavy high-gravity atmosphere with a tiny scale height.
- Bigger planets are always easier. What matters is the planet-to-star area ratio and the scale height; a small planet around a small M dwarf can beat a big planet around a giant star.
- Detected molecules sit on the surface. Transmission probes high-altitude gas; surface chemistry is largely invisible.
Frequently asked questions
What is transmission spectroscopy?
Transmission spectroscopy measures how much starlight an exoplanet's atmosphere absorbs at each wavelength during transit. As the planet crosses its star, a thin ring of atmosphere backlit by the star absorbs specific colors. Wavelengths where molecules absorb show a slightly deeper transit, because the planet's opaque silhouette is effectively larger there. Plotting transit depth against wavelength yields the transmission spectrum — a chemical fingerprint of the air.
How big is the atmospheric signal?
Tiny. The transit itself is ~1-2% for a hot Jupiter and ~0.01% for an Earth around a Sun-like star. The extra absorption from the atmosphere is roughly the ratio of the atmosphere's annular area to the stellar disk, which is about 2·R_p·N·H / R_star². For a hot Jupiter this is ~100-1000 parts per million; for a temperate rocky planet around an M dwarf it is only ~10-50 ppm, which is why such detections demand JWST-class precision.
Which molecules can it detect?
Common targets include water (H2O) with strong bands near 1.4 and 1.9 microns, carbon dioxide (CO2) at 4.3 microns, methane (CH4) near 3.3 microns, carbon monoxide (CO) near 4.7 microns, and sulfur dioxide (SO2). In the optical, neutral sodium (Na, 589 nm) and potassium (K, 770 nm) produce sharp lines. JWST has added CO2, SO2, and even silicate clouds to the inventory of detected exoplanet atmospheric species since 2022.
Why do clouds and hazes ruin the spectrum?
High-altitude aerosols are gray absorbers: they block all wavelengths roughly equally, raising the planet's opaque radius and flattening the molecular features below it. A cloud deck at, say, 1 millibar truncates the deeper, information-rich part of the atmosphere. This is why many early HST spectra of hot Jupiters looked muted — the water bands were partly hidden under hazes — and why retrievals must fit a cloud-top pressure alongside abundances.
How is a transmission spectrum actually measured?
You record the combined star-plus-planet light at high spectral resolution continuously, before, during, and after transit. Dividing the in-transit spectrum by the out-of-transit baseline removes the star's intrinsic colors and leaves the wavelength-dependent transit depth. Each spectral channel becomes its own light curve; fitting all of them gives the planet's apparent radius versus wavelength. The technique only works for transiting planets, so it pairs naturally with the transit method.
What can fool a transmission spectrum?
Stellar contamination is the big one: unocculted starspots and faint plages on the host star imprint stellar features that mimic or distort planetary ones — the so-called transit light source effect. Other pitfalls include instrument systematics, the assumption of a uniform terminator (the day-night boundary may differ in temperature and chemistry), and refraction at low altitudes. Robust results require repeated transits and careful stellar modeling.