Exoplanets
Photoevaporation Valley: How Stellar X-rays Carve the Radius Gap
Sort every close-in small planet Kepler ever found by size and a startling gap opens up between 1.5 and 2.0 Earth radii (R⊕) — a valley where planets are roughly twice as rare as on either side. Two peaks flank it: a rocky ridge near 1.3 R⊕ and a puffier one near 2.4 R⊕. This is the radius valley (or Fulton gap), and one leading explanation is photoevaporation: for the first ~100 million years of a system's life, the young host star's saturated X-ray and extreme-ultraviolet (XUV) output blowtorches the primordial hydrogen-helium envelopes off close-orbiting planets.
A planet born with a thin, easily stripped atmosphere loses it entirely and lands on the rocky peak; one born with a slightly more massive core keeps its gas and stays a sub-Neptune. Almost nothing survives at the intermediate size, because that configuration is precisely what evaporation destroys — carving the valley into the planet population itself.
- RegimeClose-in planets, orbital period P < 100 days
- Key numberValley at ~1.8 R⊕ (gap spans 1.5–2.0 R⊕)
- Driven bySaturated stellar X-ray/EUV (XUV), Lx/Lbol ~ 10⁻³
- First describedOwen & Wu, Lopez & Fortney (2013); Fulton et al. gap (2017)
- Observed withKepler/K2, CKS spectra + Gaia DR2 radii
- Matters forSplit between rocky super-Earths and gassy sub-Neptunes
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.
A gap in the planet population, hiding in plain sight
Among the thousands of planets Kepler found, the most common are between Earth and Neptune in size — a class with no Solar System analog. In 2017, B.J. Fulton and the California-Kepler Survey (CKS) team measured host-star radii spectroscopically, tightening planet radii to ~5–10% precision, and revealed that this population is not a smooth continuum but bimodal. Occurrence rate peaks near 1.3 R⊕ and 2.4 R⊕ with a factor-of-≥2 deficit between 1.5 and 2.0 R⊕ — the radius valley.
The gap is only seen among close-in planets (orbital periods P < 100 days), exactly where a host star's high-energy radiation is intense. That geography is the tell: the valley is not primordial but sculpted, marking a boundary between planets that kept a hydrogen-helium envelope and those stripped down to bare rock. Whether it is carved chiefly by photoevaporation or a competing process is the central open debate, but its existence is now robust across Kepler, K2, and TESS.
The mechanism: an X-ray blowtorch on a young planet
A rocky core (a few Earth masses) accreting from a protoplanetary disk grabs a hydrogen-helium envelope of a few percent of its mass. Crucially, this thin layer inflates the planet's radius enormously — a 1% envelope can double the observed size. So a planet's radius is set less by its rock than by whether it holds that gas.
Young stars spin fast and are magnetically active, keeping their coronal X-ray/EUV luminosity saturated at Lx/Lbol ≈ 10⁻³–10⁻⁴ for the first ~100 Myr. These XUV photons deposit energy high in the atmosphere, heating it to ~10⁴ K and driving a hydrodynamic (Parker-like) wind that flows off at km s⁻¹ speeds. The escape is roughly energy-limited: Ṁ ≈ η·L_XUV·R_p³ / (4·G·M_p·a²), where efficiency η ~ 0.1 and a is orbital distance. A planet whose envelope binding energy is smaller than the integrated XUV energy received loses everything; a slightly heavier core holds on. The threshold cleaves the population in two.
Characteristic numbers, scales, and the valley's slope
The valley sits near 1.8 R⊕, with peaks at ~1.3 and ~2.4 R⊕. It is confined to P < 100 days (roughly a < 0.4 AU) where XUV flux is high. The relevant photon budget is delivered in the first ~10⁷–10⁸ yr while the star is X-ray saturated; total integrated XUV fluence, not instantaneous flux, sets the outcome.
A sharp prediction is the valley's slope with orbital period. Because more distant planets receive less XUV, the stripping boundary shifts to smaller radii outward, giving d log R_valley / d log P ≈ −0.11 — a shallow negative slope confirmed observationally (Van Eylen et al. 2018 found ≈ −0.09 to −0.11). Energy-limited escape also predicts a dependence on host-star mass and XUV history: the valley center rises with stellar mass. The two flanking populations differ in bulk density — ~5–8 g cm⁻³ rocky worlds below, ~1–3 g cm⁻³ volatile-rich ones above — consistent with envelope retention.
How it is detected: Kepler radii, spectra, and Gaia
The valley is a statistical feature, so it lives or dies on precise planet radii, which come from a transit depth times the stellar radius: R_p = R⋆·√(depth). Early Kepler stellar radii were too uncertain (~25%), smearing the gap away. The CKS program (Petigura, Fulton, Howard 2017) took high-resolution optical spectra of ~1300 Kepler hosts to pin down R⋆, and CKS VII folded in Gaia DR2 parallaxes, sharpening radii to a few percent and exposing the bimodality cleanly.
K2 photometry independently recovered a valley near 1.9 R⊕, and TESS extends it to brighter, nearer stars. CKS X (Petigura et al. 2022) mapped the gap versus stellar mass, metallicity, and age, finding the valley center increases with stellar mass — a photoevaporation-favoring signature. Direct evaporation is also caught in the act: escaping hydrogen is seen in transit as deep absorption in Lyman-α and the metastable helium 10830 Å triplet (e.g., HD 209458 b, GJ 3470 b), confirming that XUV-driven winds are physically real.
Where it operates, and how it differs from its rivals
The valley operates on close-in super-Earths and sub-Neptunes (roughly 1–4 R⊕) around FGKM stars, during the first hundred Myr when high-energy flux is highest. It does not apply to giant planets (too deep a gravity well) or distant planets (too little XUV). The same rocky-vs-gassy split appears around M dwarfs, but shifted, because their XUV histories and habitable-zone distances differ.
The chief rival is core-powered mass loss (Ginzburg et al. 2018; Gupta & Schlichting 2019): here the planet's own cooling luminosity from a hot core, powered by the star's bolometric flux, drives escape over ~Gyr — no XUV blowtorch required. It predicts nearly the same bimodality and slope, so distinguishing them is hard. The cleanest discriminant is stellar mass and age dependence: core-powered models lack an explicit XUV/stellar-mass term, while photoevaporation predicts a non-zero one. A third option is a primordial gap set at formation, needing no mass loss at all.
Significance and open questions
The radius valley is one of exoplanet science's cleanest population-level fingerprints of a physical process, turning millions of years of atmospheric evolution into a measurable notch in a histogram. It reframes how we read planet sizes: a planet's radius is often a record of what it lost, not what it formed with. That matters for habitability — worlds near the valley may be stripped rocky cores or shrunken sub-Neptunes with very different surfaces and volatile budgets.
The central unresolved question is which mechanism dominates. Recent work (e.g., Berger et al. 2023) argues core-powered mass loss may better fit current Kepler statistics, while others favor photoevaporation or a blend of both operating on different planets. Resolving it needs precise stellar ages (asteroseismology), young-cluster planets caught mid-evolution, better XUV spectra of low-mass stars, and mass measurements to break the radius-composition degeneracy. TESS, JWST helium-triplet observations, and PLATO will supply exactly these tests.
| Property | Below valley (super-Earth) | Above valley (sub-Neptune) |
|---|---|---|
| Typical radius | ~1.3 R⊕ | ~2.4 R⊕ |
| Composition | Rocky/iron core, atmosphere stripped | Core + a few % H/He envelope by mass |
| Bulk density | ~5–8 g cm⁻³ (Earth-like) | ~1–3 g cm⁻³ (volatile-rich) |
| Envelope fate | Fully photoevaporated | Envelope retained (deeper gravity well) |
| Photoevaporation prediction | Valley slips to smaller R at long P; scales with stellar mass | Slope d log R / d log P ≈ −0.11 |
| Core-powered rival | Bolometric-driven, no stellar-mass term; same rocky/gassy split | Distinguished by M⋆ dependence + age |
Frequently asked questions
What exactly is the photoevaporation valley?
It is the observed scarcity of close-in exoplanets with radii between about 1.5 and 2.0 Earth radii, with a dip centered near 1.8 R⊕. Discovered clearly by Fulton et al. (2017) in the California-Kepler Survey, it separates rocky super-Earths (peak ~1.3 R⊕) from gas-enveloped sub-Neptunes (peak ~2.4 R⊕). Photoevaporation is one leading explanation for why that intermediate size is depleted.
Why does stripping gas create a gap rather than a smooth range of sizes?
Because a thin hydrogen-helium envelope inflates a planet's radius dramatically — even 1% of the mass in gas can double the size. So planets cluster either with an envelope (large) or without one (small), with little in between. Any planet whose envelope is marginal gets fully stripped and shrinks past the intermediate range, so the intermediate radius is inherently unstable and stays empty.
What powers the evaporation, and when does it happen?
High-energy X-ray and extreme-ultraviolet (XUV) photons from the young host star. For roughly the first 100 million years the star's magnetic activity keeps this emission saturated at Lx/Lbol ≈ 10⁻³–10⁻⁴. These photons heat a planet's upper atmosphere to ~10⁴ K, driving a hydrodynamic wind. Most envelope loss is finished within this early, high-activity window.
How is the valley actually measured?
By building a precise histogram of planet radii for close-in planets. Radius comes from transit depth times the stellar radius, so accurate stellar radii are essential. The California-Kepler Survey used high-resolution spectra, later combined with Gaia DR2 parallaxes, to reach a few-percent precision on radii — sharp enough to reveal the bimodal distribution that earlier, fuzzier data washed out.
How can we tell photoevaporation from core-powered mass loss?
Both predict a similar bimodal distribution and a slope of about d log R / d log P ≈ −0.11, so they are hard to separate. The best discriminants are host-star mass, age, and XUV dependence: photoevaporation predicts the valley location shifts with stellar mass and high-energy history, while core-powered mass loss (bolometrically driven over Gyr) has essentially no explicit stellar-mass term. Precise ages and young-cluster planets help decide.
Do we ever see atmospheric escape directly?
Yes. Escaping hydrogen produces deep absorption during transit in the Lyman-α line, and escaping helium shows up in the metastable 10830 Å (1083 nm) triplet. Planets like HD 209458 b, HD 189733 b, and GJ 3470 b show extended, evaporating atmospheres, confirming that XUV-driven winds are a real, ongoing physical process — even if their integrated effect on the whole population is still debated.