Planet or eclipsing binary?

A transit survey finds dips in starlight. Only some of them are planets. When a periodic, shallow dip turns up in a light curve, the most likely explanation is often not a planet but a pair of stars eclipsing each other, either the target itself or a fainter star whose light is mixed with it. Separating the two is most of the work of finding planets with the transit method.

Why a dip alone is ambiguous

An eclipsing binary is a pair of stars whose orbit is lined up so that each passes in front of the other. When the eclipse is central and the stars are similar, the dips are deep, often tens of percent, and nobody would mistake them for a planet. But three ordinary situations make a binary look like a transit of a few tenths of a percent to a percent:

• Grazing eclipses. If one star only clips the edge of the other, only a small fraction of the light is blocked.
• A small companion. A low-mass star or brown dwarf can be about the size of Jupiter. Crossing a Sun-like star, it produces a dip of the same depth as a giant planet. Size alone cannot tell them apart; mass can.
• Blends. A faint eclipsing binary that falls in the same pixels as a brighter target star adds its eclipses to the target’s light. A 50% eclipse in a star 100 times fainter than the target appears as a 0.5% dip. Survey cameras have large pixels (TESS pixels are 21 arcseconds across), so a blend can sit behind almost any target.

Clues in the light curve

Shape

A planet is small compared with its star, so it moves fully onto the disc quickly and the dip has a flat or gently curved bottom: a U shape. A grazing or similar-sized companion takes about as long to enter and leave as it spends in eclipse, which gives a V shape. A V-shaped dip is not proof of a binary, since a planet crossing near the edge of its star also gives one, but it is a warning.

Odd and even dips

If the two stars differ, their eclipses differ in depth. When the period is found at the wrong value, half the true one, the alternating primary and secondary eclipses fold together as one signal. Comparing the depths of the odd-numbered and even-numbered dips catches this. A planet’s transits are the same depth every time.

The secondary eclipse

When the companion passes behind its partner, the system dims again by the light of the hidden body. For a star that is often easy to see. For a planet it is tiny, because the planet gives off almost no light compared with its star: tens to hundreds of parts per million even for hot Jupiters at red and infrared wavelengths. A clear secondary eclipse at a depth no planet could produce points to a star.

Light between the dips

Two stars in a close orbit pull each other into slightly elongated shapes, and the brightness rises and falls twice per orbit as we see them side-on and end-on. This ellipsoidal variation, along with small brightness changes from the stars’ motion, reveals companions far more massive than planets.

Does the shape fit the star?

The duration and shape of a transit, together with the period, set the density of the star being transited. If that density disagrees badly with what is known about the target, from its spectrum or from Gaia, the dip may be happening on a different star.

Where the light comes from

For a blend, the light curve can look perfect while the dip comes from the wrong star. The test is to look at where the light dims. Comparing images taken during the dip with images taken outside it, and subtracting them, shows which star faded. If the difference image is centred on a neighbour rather than the target, the signal belongs to the neighbour. Catalogues of nearby stars, chiefly from Gaia, then show whether any neighbour is bright enough to produce the observed depth.

Follow-up observations

• High-resolution imaging. Speckle imaging and adaptive optics resolve companions within an arcsecond or less that survey images cannot separate.
• Spectra. A spectrum shows whether the target is a single star, gives its size and temperature, and can reveal a second set of spectral lines from a companion.
• Radial velocities. A stellar companion makes its partner move at kilometres per second. Jupiter makes the Sun move at about 12.5 metres per second, Earth at about 9 centimetres per second. A modest set of velocity measurements rules out a star; a precise set measures the planet’s mass.
• Transits in other colours. A planet blocks nearly the same fraction of light at every wavelength. A blend of two stars of different colours often does not.

Validation

Some planets are too small, or their stars too faint, for their mass to be measured. They can still be validated statistically: every false-positive scenario is modelled, its probability is computed from the light curve, the imaging and the population of stars in that part of the sky, and the planet is accepted when the false-positive probability is small enough. This is how many of the smaller planets found by Kepler and TESS were confirmed.

That is why careful work uses three words. A signal is a dip that has been detected but not checked. A candidate has passed these tests. A planet has been confirmed by further observation or validated statistically. For the basics of how a transit works, see How the transit method finds planets.