Exoplanets: How We Find Worlds We Cannot See

Directly photographing an exoplanet is close to impossible in most cases. The planet is perhaps a billion times fainter than its star and separated from it, as seen from Earth, by a fraction of an arcsecond. It is the equivalent of photographing a firefly next to a searchlight from several hundred kilometres away.

So almost every known exoplanet was found indirectly, by watching what it does to its star. Each method has a distinctive blind spot, and understanding those blind spots is the difference between reading the catalogue correctly and badly misreading it.

The radial velocity method

A planet does not orbit its star. Both orbit their common centre of mass, so the star traces a small circle of its own as the planet goes round.

That wobble is detectable in the star's spectrum. When the star moves toward us its light is blueshifted slightly; when it moves away, redshifted. Measuring that periodic shift gives the orbital period and a minimum mass for the planet.

The precision required is remarkable. Jupiter makes the Sun move at about 12 metres per second — walking pace. Earth manages roughly 9 centimetres per second. Modern spectrographs like ESPRESSO at the Very Large Telescope now approach that regime, but only just.

This was the method behind the first confirmed planet around a Sun-like star: 51 Pegasi b, announced by Michel Mayor and Didier Queloz in 1995, which won them a share of the 2019 Nobel Prize.

The blind spot: radial velocity favours massive planets on short orbits, because those produce the largest and fastest-repeating wobbles. It also yields only a minimum mass, because an orbit seen face-on produces no line-of-sight motion at all. A planet could be far heavier than the measured figure and you would not know from this method alone.

The transit method

If a planet's orbit happens to be aligned edge-on to us, the planet passes in front of its star once per orbit and blocks a fraction of its light. Measure that periodic dip and you get the orbital period and the planet's radius.

The dips are small. Jupiter crossing the Sun would block about 1% of the light; Earth about 0.008%. Detecting that requires photometry stable to parts per million, which is why the method needed space telescopes to become productive.

NASA's Kepler mission stared at a single patch of sky containing about 150,000 stars for four years, and it transformed the field — over 2,700 confirmed planets, and crucially the statistics to conclude that planets are the norm rather than the exception. TESS, launched in 2018, surveys the whole sky for transits around brighter, nearer stars that are better suited to follow-up.

Combining a transit radius with a radial-velocity mass gives density, and density distinguishes a rocky world from a gas envelope. Most of what we claim to know about exoplanet composition comes from having both measurements for the same object.

The blind spot: geometric alignment. Only a small fraction of planetary systems happen to be edge-on from our viewpoint. For an Earth-like orbit around a Sun-like star, the odds are under half a percent. The vast majority of planets simply never transit as seen from here, and no amount of observing time will reveal them this way.

Gravitational microlensing

Mass bends light. When one star passes precisely in front of a more distant one, its gravity focuses the background star's light and briefly brightens it. If the foreground star has a planet, the planet adds a short secondary spike to that brightening.

This method reaches planets the others cannot: worlds on wide orbits, low-mass planets, and objects thousands of light-years away toward the galactic centre. It is also the main way we detect rogue planets — worlds bound to no star at all, drifting through the galaxy after being gravitationally ejected from the system that formed them. Population estimates suggest these may be extremely common, plausibly outnumbering the stars.

The blind spot: alignments are chance events that never repeat. You get one observation and no possibility of follow-up, ever. NASA's Nancy Grace Roman Space Telescope is designed in part to run a large microlensing survey and finally put statistics on this population.

Direct imaging

Sometimes it does work. With a coronagraph blocking the starlight, adaptive optics correcting atmospheric blur, and a favourable target — a young, massive, hot planet on a wide orbit around a nearby star — you can photograph the planet itself.

The HR 8799 system, with four imaged planets whose orbital motion has been visibly tracked over years, is the standout case. JWST has since imaged planets and taken spectra directly, which is far more informative than a dot: a spectrum reveals atmospheric composition.

The blind spot: it only works for young, hot, massive planets far from their stars. That is a small and unrepresentative slice of the population.

What the catalogue actually says

With over 5,000 confirmed planets, some patterns are firm and others are artefacts of method.

Planets are ordinary. The statistical conclusion from Kepler is that most stars have planets. In a galaxy of a few hundred billion stars, planets number in the hundreds of billions.

The most common type does not exist here. Super-Earths and mini-Neptunes — between Earth and Neptune in size — are the most abundant class found, and our solar system contains nothing in that range. There is a genuine gap in the distribution around 1.5 to 2 Earth radii, the "radius valley", probably marking where stellar radiation strips a planet's hydrogen envelope.

Hot Jupiters are over-represented. Giant planets orbiting closer than Mercury were the first big surprise, because planet formation theory said giants form far out where ice is available. They must migrate inward. But they also produce the largest signals in both radial velocity and transit surveys, so the catalogue overstates how common they are.

The habitable zone, and its limits

The habitable zone is the orbital range where a planet could sustain liquid surface water given adequate atmospheric pressure. Too close and water boils away; too far and it freezes.

It is a useful first filter and a poor final one. Venus sits at the inner edge of the Sun's habitable zone and is 464 °C under a runaway greenhouse. Mars sits near the outer edge and lost its atmosphere when its magnetic field died. Both are in the zone. Neither is habitable.

Meanwhile Europa and Enceladus, far outside any habitable zone, have subsurface liquid oceans kept warm by tidal flexing rather than sunlight. If we are honest, the solar system's most promising places for life are all outside the habitable zone.

The complication is sharpest around red dwarfs, which host most of the galaxy's planets. Their habitable zones are so close in that planets there are likely tidally locked, with one hemisphere in permanent day. Worse, red dwarfs flare violently, and the TRAPPIST-1 system — seven Earth-sized planets, several in the habitable zone, 40 light-years away — is the test case JWST is currently working through. Early results for the innermost planets suggest thin atmospheres or none at all, which is not encouraging for the class as a whole.

What comes next

The frontier has moved from finding planets to characterising their atmospheres. Transmission spectroscopy is the key technique: during a transit, a sliver of starlight filters through the planet's atmosphere, and molecules there absorb specific wavelengths, leaving fingerprints in the spectrum.

JWST has detected carbon dioxide, methane, sulfur dioxide and water in exoplanet atmospheres this way. The eventual goal is a biosignature — a combination of gases that chemistry alone cannot maintain. Oxygen alongside methane is the classic example, since the two react away rapidly and their persistent coexistence implies continuous replenishment.

Nothing of the kind has been confirmed. Claims in this area have a consistent history of weakening under scrutiny, and abiotic explanations are usually harder to eliminate than the initial announcement suggests. But the instruments capable of settling the question are now being designed, and it is a reasonable expectation that this generation will get a defensible answer.