How Do We Detect Exoplanet Atmospheres — and What Are Biosignatures?

One of the oldest questions humans have asked is whether we are alone in the universe. For most of history, it was purely philosophical — there was no way to test it. That is changing. With the James Webb Space Telescope (JWST) and its planned successors, astronomers now have instruments capable of measuring the atmospheres of planets orbiting other stars and identifying some of the molecules in them.

This capability — atmospheric characterization of exoplanets — is one of the most active frontiers in astronomy. It does not yet allow us to detect life with confidence. But it is building the observational and theoretical framework to one day do so. Understanding how it works requires a little physics of light, some atmospheric chemistry, and a clear sense of what life does to a planet that geology alone cannot.

Key takeaways

• Most exoplanet atmosphere measurements use transmission spectroscopy: comparing how much starlight a transiting planet blocks at different wavelengths.
• The first exoplanet atmosphere detection was sodium on the hot Jupiter HD 209458 b, reported in 2002 using Hubble.
• In 2022, JWST made the first unambiguous detection of carbon dioxide in an exoplanet atmosphere, on the hot gas giant WASP-39 b.
• A biosignature is a gas or pattern that is hard to explain without life; the strongest cases involve gases in chemical disequilibrium, such as oxygen plus methane.
• JWST found that TRAPPIST-1 b, an Earth-sized planet, likely has little or no atmosphere.
• Claims of dimethyl sulfide (DMS) on K2-18 b have not survived independent reanalysis; as of 2025, several teams found the evidence insufficient.

The challenge: reading atmospheres across light-years

Exoplanets are extraordinarily faint objects sitting next to blindingly bright stars. In visible reflected light, an Earth-like planet around a Sun-like star is roughly ten billion times fainter than its host — like trying to see a firefly beside a searchlight from kilometres away. Separating the two is beyond current technology for rocky planets in all but a handful of favourable cases.

Fortunately, nature offers indirect methods that extract atmospheric information without ever seeing the planet as a separate dot. Almost all of them depend on one idea that dates to the 19th century: every atom and molecule absorbs and emits light only at specific wavelengths, so a spectrum is a chemical fingerprint. The story of how that was discovered is told in how the spectroscope revealed what stars are made of.

A short history of exoplanet atmospheres

Year	Milestone	Instrument
1995	First planet found orbiting a Sun-like star (51 Pegasi b)	Ground-based radial velocity
1999	First observed transit of an exoplanet (HD 209458 b)	Ground-based photometry
2002	First exoplanet atmosphere: sodium in HD 209458 b's transmission spectrum (Charbonneau et al.)	Hubble Space Telescope (STIS)
2005	First direct detection of thermal light from exoplanets via secondary eclipses	Spitzer Space Telescope
2008	Ground-based confirmation of sodium in HD 209458 b (Snellen et al.)	Subaru telescope
2021	JWST launched (December 25)	—
2022	First unambiguous CO₂ detection in an exoplanet atmosphere (WASP-39 b)	JWST NIRSpec
2023	Thermal emission of TRAPPIST-1 b measured; consistent with a bare rock	JWST MIRI
2023	Methane and CO₂ reported on K2-18 b, with a tentative DMS hint	JWST NIRISS + NIRSpec
2025	Joint reanalyses find insufficient evidence for DMS on K2-18 b	JWST archival data

The 2002 sodium detection is a good illustration of how the field works. Charbonneau and colleagues observed four transits of HD 209458 b with Hubble and found the planet blocked slightly more light in the narrow sodium band than in neighbouring wavelengths — by about two parts in ten thousand. That is a tiny number, yet it was enough to show the planet had an atmosphere, and the fact that the signal was weaker than their simple models predicted hinted at clouds or haze, a theme that has dominated the field ever since.

Transmission spectroscopy: reading the atmosphere's fingerprint

The most widely used method is transmission spectroscopy, and it relies on a fortunate geometric alignment.

When a planet transits — passes in front of its host star from our point of view — it blocks a small fraction of the star's light. The fraction blocked, called the transit depth, equals the ratio of the planet's disk area to the star's. But the planet is not a perfectly sharp disk. A thin ring of atmosphere surrounds it, and starlight filters through that ring on its way to us. At wavelengths where a molecule in the atmosphere absorbs strongly, the ring is opaque higher up, so the planet looks slightly bigger and the transit is slightly deeper. At wavelengths where the atmosphere is transparent, the planet looks smaller.

Measure the transit depth at hundreds of wavelengths and you get a transmission spectrum — a plot of apparent planet size versus wavelength. Peaks mark absorption bands: water near 1.4 microns, carbon dioxide at 4.3 microns, methane around 3.3 microns, and so on.

Why some planets are easier than others

The size of the atmospheric signal depends on the scale height, the vertical distance over which atmospheric pressure drops by a factor of about 2.7. Scale height grows with temperature and shrinks with surface gravity and with the average mass of the gas molecules. A hot, puffy gas giant with a hydrogen-rich atmosphere has a huge scale height and produces large features. A cool, dense rocky planet with a nitrogen or carbon dioxide atmosphere produces tiny ones.

The size of the host star matters too. Because the signal is a ratio of areas, the same planet produces a deeper transit and a stronger atmospheric signal around a small star. This is why small, cool M-dwarf stars like TRAPPIST-1 are the prime targets for studying rocky planets today.

Planet type	Example	Typical atmospheric signal in transit	Feasibility with JWST
Hot Jupiter / hot Saturn	WASP-39 b	Hundreds to over a thousand parts per million	Routine; high-precision spectra
Sub-Neptune	K2-18 b	Tens to ~100 ppm	Achievable with several transits
Rocky planet around M dwarf	TRAPPIST-1 planets	~10–100 ppm, if an atmosphere exists	Very demanding; many transits
Earth twin around Sun-like star	(none yet known)	Around 1 ppm or less	Beyond JWST; needs future missions

(Signal ranges are order-of-magnitude illustrations based on scale-height physics, not measurements of specific planets.)

What can go wrong

Transmission spectroscopy requires the planet to transit at all — only a small fraction of planetary systems are aligned edge-on to us. It also requires extraordinarily stable instruments, multiple transits to build up signal, and careful modelling of the star. Starspots and bright faculae on M dwarfs can imprint their own spectral features on the data, mimicking or hiding molecules in the planet's atmosphere. This "transit light source effect" is one of the biggest current headaches in rocky-planet work.

Clouds and hazes are another. A high cloud deck flattens the spectrum because it blocks light at all wavelengths equally. A flat spectrum can therefore mean no atmosphere, a high-altitude cloud layer, or an atmosphere made of heavy molecules with a small scale height — and distinguishing those possibilities usually takes additional observations.

Other methods: thermal emission, phase curves and direct imaging

Secondary eclipse (thermal emission) spectroscopy works when a planet passes behind its star. Just before eclipse, a telescope measures the combined light of star plus planet; during eclipse, only the star. The difference is the light the planet itself emits from its dayside. Measured across wavelengths, it yields the dayside temperature and, when signals are strong enough, molecular features. This method is easiest for hot planets close to their stars, which glow strongly in the infrared, and it is how JWST tested whether TRAPPIST-1 b has an atmosphere.

Phase curves track a planet's brightness through its whole orbit as different hemispheres face us. Because many close-in planets are tidally locked, with one side permanently facing the star, a phase curve reveals how much heat an atmosphere carries from day side to night side. A large day–night contrast suggests a thin or absent atmosphere; a small one suggests winds redistributing heat.

High-resolution ground-based spectroscopy takes a different route. Large telescopes on the ground split light so finely that individual molecular lines become visible, and the planet's orbital motion Doppler-shifts those lines relative to the star's. Cross-correlating the spectrum against templates of known molecules can pull out detections of water, carbon monoxide, iron vapour and more on hot giant planets.

Direct imaging spectroscopy blocks the star's glare with a coronagraph or an external starshade and records the planet's own spectrum. Ground-based instruments such as GPI on the Gemini telescope and SPHERE on ESO's Very Large Telescope have characterized young, hot giant planets far from their stars. Directly imaging an Earth-like planet around a Sun-like star requires a dedicated future space telescope.

What are biosignatures?

A biosignature is any substance, object or pattern whose origin specifically requires a biological agent. In exoplanet atmospheres, the candidates are gases whose presence, abundance or combination is difficult to explain through geology or photochemistry alone. The concept builds on a 1965 suggestion by James Lovelock that life could be detected remotely by looking for an atmosphere in strong chemical disequilibrium. For the wider scientific context, see our overview of astrobiology and the science of life beyond Earth.

Oxygen (O₂) and ozone (O₃)

Molecular oxygen is the canonical biosignature. Earth's atmosphere is about 21% oxygen, produced by photosynthesis — first by cyanobacteria, later by algae and plants. Oxygen began accumulating during the Great Oxidation Event roughly 2.4 billion years ago, although it took hundreds of millions of years more to reach modern levels. Without continuous replenishment, oxygen would be gradually removed by reactions with rocks and volcanic gases. The biochemistry behind its production is described in how photosynthesis works.

The problem is false positives. Ultraviolet light can split water vapour high in an atmosphere; the hydrogen escapes to space and oxygen can be left behind. On planets around very active M dwarfs, which bombard their planets with ultraviolet radiation for billions of years, models suggest this could build up substantial abiotic oxygen. Carbon dioxide photolysis on dry planets is another proposed pathway.

Ozone forms when ultraviolet light breaks O₂ into oxygen atoms that then combine with remaining O₂. On Earth it forms the stratospheric ozone layer, which absorbs ultraviolet light from the Sun, and it also has a strong infrared absorption band near 9.6 microns. Because even a small amount of oxygen produces a detectable ozone layer, ozone is a sensitive proxy for an oxygenated atmosphere — useful because O₂ itself has relatively weak spectral features in the infrared.

Methane (CH₄)

Methane on Earth comes from both biology (methane-producing microbes in wetlands, rice paddies and animal guts) and geology (hydrothermal systems and serpentinization, in which water reacts with iron-rich rock). In today's oxygen-rich atmosphere, methane survives only about a decade before being destroyed by hydroxyl radicals, so maintaining it requires a large, continuous source.

Methane's real power as a biosignature comes in combination with oxygen. The two gases react with each other, so finding both at significant levels implies something is continuously producing each — a thermodynamic disequilibrium that is difficult to explain without life. Mars offers a cautionary example of how hard methane alone is to interpret: the Curiosity rover has measured fluctuating background methane in Gale Crater that orbiters have struggled to confirm, as discussed in the Curiosity methane mystery.

Nitrous oxide (N₂O)

Nitrous oxide is produced mostly by microbes in Earth's nitrogen cycle, with only minor abiotic sources such as lightning. Its infrared bands overlap with those of water and carbon dioxide, making it hard to detect. Some researchers have argued that on planets with different biospheres, N₂O could accumulate to higher levels than on Earth and become a more practical target.

Dimethyl sulfide (DMS)

Dimethyl sulfide, (CH₃)₂S, is produced on Earth overwhelmingly by marine phytoplankton. Because there is no known significant abiotic source on Earth, it was proposed as a biosignature for ocean worlds. It became famous through K2-18 b, discussed below, where the evidence for it has proven fragile.

Biosignature candidates compared

Gas	Main source on Earth	Known or proposed abiotic sources	Key infrared feature(s)	Strength as a biosignature
O₂	Photosynthesis	Water or CO₂ photolysis with hydrogen escape	0.76 µm (A-band), weak in infrared	Strong in context; false positives possible around M dwarfs
O₃	Photochemistry of O₂	Same as O₂	~9.6 µm	Sensitive proxy for O₂
CH₄	Methanogenic microbes	Volcanism, serpentinization, impacts	~3.3 µm, ~7.7 µm	Ambiguous alone; strong with O₂
N₂O	Microbial nitrogen cycle	Lightning, minor photochemistry	~4.5 µm, ~7.8 µm	Strong, but hard to detect
DMS	Marine phytoplankton	Possible photochemical or cometary sources proposed	~3.4 µm, mid-infrared bands	Speculative; low abundances overlap with other molecules

The false positive problem

The central challenge in biosignature science is false positives: non-biological processes or data artefacts that mimic biological signals. This is not theoretical caution — the history of the field is full of examples.

The Venus phosphine episode

In 2020, a team led by Jane Greaves reported detecting phosphine (PH₃) in the clouds of Venus using the James Clerk Maxwell Telescope and the ALMA array, at around 20 parts per billion. On Earth, phosphine in the atmosphere is associated with life and industry, and its presence in Venus's acidic clouds seemed hard to explain. The announcement triggered intense interest in possible life in the Venusian cloud layer.

Independent reanalyses challenged it quickly. A team led by Geronimo Villanueva of NASA Goddard reanalysed the same millimetre-wave data and found no evidence of phosphine, placing upper limits that conflicted with the original claim. They pointed out that the phosphine line at 266.944 GHz lies within about 1.3 km/s of a sulfur dioxide line, closer than the data's spectral resolution could separate. (Villanueva et al., arXiv:2010.14305) The original team revised its estimate downward, to a few parts per billion, and has continued to argue for a signal. As of the mid-2020s, the detection remains disputed rather than confirmed. Our article on why Venus is so hot covers the planet's wider atmosphere.

The episode is a model of how science should work: a bold claim, rapid scrutiny, and an honest acknowledgment that the data cannot settle the question.

The Viking precedent

An earlier cautionary tale comes from Mars. In 1976, the Viking landers' Labeled Release experiment produced results that initially looked like microbial metabolism, but other instruments found no organic molecules, and most scientists concluded the reaction was chemical, probably caused by oxidizing compounds in the soil. That debate is retold in the Mars Viking life detection experiments.

How scientists guard against false positives

For oxygen in particular, the known risks include:

• Dry, carbon-dioxide-rich planets close to active M-dwarf stars, which might accumulate abiotic oxygen through photolysis.
• Planets that lost large amounts of water early on, potentially leaving behind oxygen-rich atmospheres.
• Planets with low amounts of non-condensing gases such as nitrogen, which may allow water to reach the upper atmosphere and be split.

The opposite problem exists too: on a strongly reducing world rich in hydrogen, oxygen produced by life would be consumed rapidly, so its absence does not prove absence of life. Earth itself had life for perhaps a billion years or more before oxygen became abundant.

The solution is context. Rather than looking for a single "smoking gun" gas, researchers aim to characterize the full atmosphere: multiple gases and their ratios, the planet's mass, radius and temperature, the host star's ultraviolet output and age, and whether the overall chemistry is out of equilibrium in a way that requires a continuous, unexplained source.

JWST's results so far

JWST, launched on December 25, 2021, observes in the infrared from about 0.6 to 28 microns, the range in which many key molecules have strong bands. Its large mirror and stable orbit around the Sun–Earth L2 point make it far more sensitive than previous telescopes for this work. Our explainer on how JWST works covers the observatory itself.

WASP-39 b: carbon dioxide, unambiguously

One of JWST's first exoplanet results came from WASP-39 b, a hot gas giant about 700 light-years away. The Early Release Science team measured its transmission spectrum from 3.0 to 5.5 microns and found a prominent carbon dioxide absorption feature at 4.3 microns, detected at 26 sigma — the first unambiguous CO₂ detection in any exoplanet atmosphere. (JWST ERS team, arXiv:2208.11692) Follow-up observations of the same planet detected sulfur dioxide, interpreted as the product of photochemistry driven by the star's light — the first clear evidence of photochemistry on an exoplanet.

WASP-39 b is not a candidate for life. Its significance is technical: it showed that JWST can deliver the precision the field had been planning for decades.

TRAPPIST-1: rocky worlds, sobering results

The TRAPPIST-1 system, about 40 light-years away, hosts seven roughly Earth-sized planets around a small, cool M-dwarf star, several of them in or near the habitable zone. It is JWST's flagship system for rocky-planet atmospheres.

TRAPPIST-1 b. Using five secondary eclipses observed with the MIRI instrument at 15 microns, a team led by Thomas Greene detected the planet's thermal emission at 8.7 sigma. The brightness matched what would be expected if only the dayside reradiated the starlight it receives, with little or no atmosphere carrying heat to the nightside, and no carbon dioxide absorption was detected. (Greene et al., arXiv:2303.14849) The simplest interpretation is a bare or nearly bare rock, although later work noted that some atmospheric scenarios, including hazy ones, cannot be fully excluded.

TRAPPIST-1 c. A similar 15-micron eclipse measurement published later in 2023 ruled out a thick, Venus-like carbon dioxide atmosphere for the second planet, although a thinner atmosphere remains possible.

TRAPPIST-1 e, f and g. The habitable-zone planets are cooler and produce weaker signals, so they require many transits. The first JWST transmission spectra of TRAPPIST-1 e, reported in 2025, neither clearly favoured nor ruled out an atmosphere; they disfavoured hydrogen-rich, Venus-like and Mars-like atmospheres but left nitrogen-rich ones possible. Observations are ongoing, and stellar contamination from the active host star is a major complication.

These results do not mean the TRAPPIST-1 planets are lifeless. They do show that the innermost planets of an active M dwarf may struggle to keep atmospheres, a key question for the habitability of the most common type of star in the galaxy.

K2-18 b: a case study in caution

K2-18 b is a planet about 8.6 times Earth's mass and 2.6 times its radius, orbiting a cool dwarf star in the habitable zone roughly 120 light-years away in the constellation Leo. In September 2023, a team led by Nikku Madhusudhan of the University of Cambridge reported JWST detections of methane and carbon dioxide, and a scarcity of ammonia, from just two transits. They argued this was consistent with a "Hycean" world — a hydrogen-rich atmosphere over a global water ocean. The same announcement mentioned a possible signal of DMS, which ESA's release described as "less robust" and in need of further validation. (ESA, 2023)

In April 2025, the same group reported new mid-infrared observations with JWST's MIRI instrument showing features they interpreted as DMS and/or the related molecule dimethyl disulfide (DMDS), at roughly three sigma. The claim drew wide media attention.

Independent researchers were not persuaded. Other teams reanalysed the near-infrared data and found methane robustly but no statistically significant evidence for DMS. A joint analysis of all of JWST's K2-18 b transmission data from 0.6 to 12 microns, led by Rafael Luque, concluded there was "insufficient evidence" for DMS or DMDS, found that other methyl-bearing molecules such as ethane fit the data about as well, and estimated that about 25 more MIRI transits would be needed to test the claim properly. (Luque et al., arXiv:2505.13407) Separate statistical critiques argued that the earlier significance depended on the choice of models compared.

There is also debate about the planet itself. Alternative models describe K2-18 b as a gas-rich mini-Neptune with no habitable surface, or as a world with a magma ocean beneath a thick envelope, either of which could produce methane and carbon dioxide without an ocean or life. As of 2026, the scientific consensus is that K2-18 b has a carbon-bearing atmosphere, and that DMS has not been detected to the standard any biosignature claim requires.

What would count as a confirmed biosignature?

A credible detection of life beyond Earth would be one of the most significant discoveries in history. The bar for scientific consensus is accordingly high. The community broadly expects:

• Robust detection: high statistical significance, reproduced with independent data reductions and ideally independent instruments.
• Correct identification: the spectral feature must be attributable to the claimed molecule and not to an overlapping one.
• Exclusion of abiotic sources: every known geological, photochemical and stellar mechanism considered and ruled out for that specific planet.
• Planetary context: a planet whose size, temperature and stellar environment make the proposed biology plausible.
• Multiple lines of evidence: ideally several gases in disequilibrium, not a single molecule.

NASA has proposed a graded scale, the Confidence of Life Detection (CoLD) framework, that describes progress from an initial hint to independent confirmation, partly to help communicate claims more carefully to the public.

The next generation of telescopes

JWST is not designed to search for life on Earth twins around Sun-like stars; the signals are too small. The next steps include:

• Extremely large ground-based telescopes, such as ESO's Extremely Large Telescope under construction in Chile, which will pair huge collecting areas with high-resolution spectrographs suited to nearby rocky planets around M dwarfs.
• ESA's Ariel mission, planned to survey the atmospheres of hundreds of mostly warm and hot exoplanets to study planetary chemistry statistically.
• NASA's proposed Habitable Worlds Observatory (HWO), a large space telescope that builds on the earlier LUVOIR and HabEx mission concepts. Its central goal is to directly image roughly Earth-sized planets in the habitable zones of Sun-like stars and search their reflected-light spectra for oxygen, ozone, water and methane.

Machine learning is also increasingly used to find planets and to speed up the analysis of their spectra, as described in how AI is revolutionizing exoplanet detection.

Common misconceptions

"Finding water means finding life." Water vapour is common on gas giants that cannot host life as we know it. It is a requirement for life as we know it, not evidence of it.

"JWST can photograph alien planets' surfaces." For transiting planets, JWST measures total light at different wavelengths; it does not resolve the planet as an image. Even future direct-imaging missions will see Earth-like planets as single points of light.

"A biosignature detection would prove aliens exist." At best, a single gas would be evidence that needs testing against every alternative. Scientists talk about probabilities and confidence levels, not proof.

"No oxygen means no life." Earth was inhabited for a long period before oxygen accumulated, and life on other worlds might use different chemistry.

FAQ

Q: How can a telescope detect the atmosphere of a planet it cannot see? A: When a planet passes in front of its star, some starlight filters through its atmosphere. Molecules absorb specific wavelengths, so the planet appears slightly larger at those wavelengths. Measuring that tiny change across the spectrum reveals which molecules are present.

Q: What was the first exoplanet atmosphere ever detected? A: Sodium in the atmosphere of the hot Jupiter HD 209458 b, reported in 2002 by David Charbonneau and colleagues using the Hubble Space Telescope.

Q: Has JWST found life on another planet? A: No. JWST has detected molecules such as carbon dioxide, water, methane and sulfur dioxide in exoplanet atmospheres. A claimed hint of dimethyl sulfide on K2-18 b was not supported by independent joint reanalyses published in 2025.

Q: Why are M-dwarf stars such popular targets? A: Their small size makes a planet's transit and atmospheric signal proportionally larger, and their habitable-zone planets orbit close in, so transits happen often. The downside is that many M dwarfs are magnetically active, which can strip atmospheres and contaminate the data.

Q: What is the strongest kind of biosignature? A: Most researchers consider chemical disequilibrium the strongest: for example, oxygen and methane together, which destroy each other and must both be continuously replenished.

Q: When could we realistically detect life on an exoplanet? A: Nobody knows. Searching Earth-like planets around Sun-like stars requires a mission like the proposed Habitable Worlds Observatory, which is still in early development and will take many years to build and launch.

Q: Could life exist on planets with no detectable biosignatures? A: Yes. Life might be too sparse, too chemically subtle, or hidden below a surface, as is possible on ice-covered moons such as those discussed in the search for life on Europa and Enceladus.

The bottom line

Reading the atmosphere of a planet light-years away was science fiction a generation ago. Since 2002, it has become a routine measurement for giant planets, and JWST is now pushing it to rocky worlds. Biosignature science has matured alongside: it has learned, through episodes like Venus phosphine and K2-18 b, that a single tentative molecule is not enough, and that context, independent confirmation and humility are essential. We may find nothing — or we may one day find a planet whose air cannot be explained without life. Either result would teach us something profound, and we will only know by looking.

---

Related Reading

• How Do We Actually Find Exoplanets?
• The Fermi Paradox: Where Is Everybody?
• How Big Is the Universe and How Much of It Can We Actually See?
• Dark Matter and Dark Energy Explained

Sources

• Greene, T. P. et al., "Thermal emission from the Earth-sized exoplanet TRAPPIST-1 b using JWST" (Nature, 2023): https://arxiv.org/abs/2303.14849
• JWST Transiting Exoplanet Community ERS Team, "Identification of carbon dioxide in an exoplanet atmosphere" (Nature, 2023): https://arxiv.org/abs/2208.11692
• ESA, "Webb discovers methane, carbon dioxide in atmosphere of K2-18 b" (2023): https://www.esa.int/Science_Exploration/Space_Science/Webb/Webb_discovers_methane_and_carbon_dioxide_in_atmosphere_of_K2-18_b
• Luque, R. et al., "Insufficient evidence for DMS and DMDS in the atmosphere of K2-18 b" (A&A, 2025): https://arxiv.org/abs/2505.13407
• Villanueva, G. L. et al., "No evidence of phosphine in the atmosphere of Venus by independent analyses" (Nature Astronomy, 2021): https://arxiv.org/abs/2010.14305