Exoplanets
The Sub-Neptune Desert: Two Rival Engines Behind the Hot-Neptune Gap
Point Kepler at any Sun-like star and you can find a Neptune-sized world in a 30-day orbit, a 100-day orbit, a 300-day orbit — but almost never inside 3 days. In the plane of orbital period versus planet radius, a sharp triangular hole opens up: at periods below ~3 days and radii between roughly 2 and 6 R⊕, planets that should be common are simply missing. This is the Neptunian desert (also called the hot-Neptune gap, evaporation desert, or sub-Jovian desert), one of the most striking sculpted features in the entire exoplanet population.
What makes it a genuinely deep problem is that the desert is bounded on two sides — a lower edge of small planets and an upper edge of larger ones — and the best current explanation invokes two physically unrelated engines: XUV photoevaporation stripping the light planets, and high-eccentricity migration plus tidal disruption forbidding the heavy ones from parking so close.
- RegimePeriods ≲ 3 days, radii ~2–6 R⊕ (masses ~0.02–0.8 M_J)
- Key numberUpper/lower boundaries meet near P ≈ 10–15 days
- Driven byXUV photoevaporation (lower edge) + tidal-disruption-limited high-e migration (upper edge)
- First describedSzabó & Kiss 2011; delimited by Mazeh, Holczer & Faigler 2016
- Observed withKepler, K2, TESS transits + radial-velocity masses
- Matters forAtmospheric loss physics, giant-planet migration, super-Earth origins
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 the desert is, and why it matters
The Neptunian desert is a region of near-total emptiness in the joint distribution of orbital period and planet size (or mass). Its diagnostic shape is a triangle in the period–radius plane whose apex sits near P ≈ 10–15 days and whose base runs along the shortest periods, P ≲ 1 day. Hot Jupiters (R_p ≳ 10 R⊕) survive comfortably down to sub-day orbits; small rocky planets and lava worlds (R_p ≲ 1.8 R⊕) survive too. But the intermediate Neptunes — the single most abundant class of planet at longer periods — vanish.
This matters because it is a sculpting signature, not a detection bias: transits and radial velocities are easiest for close-in Neptunes, so their absence is real and physical. The desert therefore encodes the two processes that most powerfully reshape close-in planets: atmospheric escape and orbital migration. It is a natural laboratory for both, and it directly connects to the origin of the abundant super-Earth population, which may partly be desert Neptunes stripped down to their cores.
The mechanism, step by step: two engines
Lower boundary — photoevaporation. A close-in Neptune sits in a fierce bath of stellar X-ray and extreme-UV (XUV) radiation. These photons are absorbed high in the H/He envelope, heat it to ~10⁴ K, and drive a hydrodynamic ("blow-off") wind. Mass loss scales roughly as Ṁ ≈ η L_XUV R_p³ / (4 G M_p a²), so it soars as the orbit shrinks (∝ a^−2). Inside a critical flux the envelope's gravitational binding energy loses to the integrated XUV energy over ~100 Myr–1 Gyr, and a Neptune is stripped to a bare super-Earth. That removes small-radius planets from short periods.
Upper boundary — tidal disruption. Massive planets get close not by disk migration but by high-eccentricity migration: a companion excites the orbit via planet–planet scattering or Lidov–Kozai cycles, and tides then circularize it. Tidal friction parks the planet near a_F ≈ 2 r_tide, twice the distance at which the planet would be torn apart. Below M_p ~ 0.2 M_J a Neptune's low mass makes r_tide large, so it cannot circularize close enough — carving the upper edge.
Characteristic numbers, scales, and the key relations
The boundaries were quantified by Mazeh, Holczer & Faigler (2016) from Kepler data. In the period–radius plane the desert's edges are power laws in log space: the upper boundary falls as roughly R_p ∝ P^(−1/3), while the lower boundary rises as R_p ∝ P^(2/3); the two meet near P ≈ 10–15 days. In period–mass, the desert spans roughly 0.02–0.8 M_J below ~3 days.
The tidal-disruption radius is r_tide ≈ η (M*/M_p)^(1/3) R_p with η ≈ 2.7 from giant-planet simulations, and high-e migration halts at a_F ≈ 2 r_tide (Owen & Lai 2018). Because r_tide grows for low-mass planets, the minimum reachable period increases as mass drops — exactly the observed upper slope. On the photoevaporation side, the critical XUV fluence needed to unbind an envelope depends on core mass; cores of ~10–15 M⊕ sit right at the survival threshold, so the lower edge tracks the stripping boundary. The divide between the two engines lands near M_p ≈ 0.2 M_J.
How it is observed and measured
The desert emerges only from large, homogeneous transit catalogs. Kepler (2009–2013) provided the first statistically robust census — thousands of planets with precise radii from transit depth (ΔF/F ≈ (R_p/R*)²) — followed by K2 and, crucially, TESS (2018–), whose all-sky survey of bright stars enables radial-velocity mass measurements with HARPS, ESPRESSO, NIRPS, and CARMENES. Radial velocities matter because the desert is sharpest in the period–mass plane, and only spectroscopic masses distinguish a puffy stripped core from a dense Neptune.
Distinguishing the two engines observationally is an active program. Photoevaporation predicts escaping hydrogen, detectable as deep transits in the Lyman-α line (1216 Å) and the metastable helium triplet at 10830 Å (near-IR). High-eccentricity migration predicts residual orbital eccentricity and spin–orbit misalignment, measured via the Rossiter–McLaughlin effect. A handful of "desert-dwelling" survivors — LTT 9779 b, TOI-849 b, and others — are prime targets for exactly these diagnostics.
Where it operates, and how it differs from the radius valley
The desert is a feature of close-in planets around main-sequence FGKM stars, carved over the first ~0.1–1 Gyr while XUV luminosity is high and while dynamical excitation from companions is still active. Its exact boundaries shift with stellar type: M dwarfs, with different XUV histories and lower masses, show a desert offset in insolation, and its shape depends on host mass and metallicity.
It should not be confused with the radius valley (the Fulton gap near 1.8 R⊕), though the two share a cause. The radius valley is a bimodality in planet radius — a scarcity of planets between super-Earths (~1.3 R⊕) and mini-Neptunes (~2.4 R⊕) — that photoevaporation (or core-powered mass loss) also produces, but it exists across a wide period range. The Neptunian desert is a two-dimensional void in period–radius space, requiring the additional dynamical, tidal-disruption engine on its upper edge. The valley strips envelopes; the desert also forbids arrival.
Open questions and significance
The single-engine debate is largely settled — both photoevaporation (Owen & Wu; Owen & Lai) and tidal-disruption-limited high-e migration (Matsakos & Königl 2016) are needed — but the balance between them, and whether core-powered mass loss (driven by the cooling core's internal luminosity rather than stellar XUV) contributes to the lower edge, remains open. The rare desert survivors are the sharpest test: LTT 9779 b, an ultra-hot Neptune at a 0.79-day period, should have been stripped, yet retains a metal-rich atmosphere — perhaps because its high mass or reflective clouds resist escape.
Whether the desert's M-dwarf edge differs fundamentally from the FGK case, how stellar age and rotation set the XUV dose, and whether some Neptunes are actively evaporating right now (as recent eccentric-sub-Neptune candidates suggest) are all live. As a bridge between atmospheric physics and orbital dynamics, the Neptunian desert remains one of the field's most productive natural experiments.
| Property | Lower boundary (small planets) | Upper boundary (large planets) |
|---|---|---|
| Dominant mechanism | XUV photoevaporation of H/He envelope | High-eccentricity migration halted at tidal-disruption limit |
| Planet mass regime | ≲ 0.2 M_J (Neptune cores ~10–15 M⊕) | ≳ 0.2 M_J (sub-Saturns to gas giants) |
| Physics balancing | Envelope binding energy vs. absorbed XUV energy | Orbital circularization radius vs. Roche/tidal radius |
| Boundary slope (period–radius) | R_p increases with P (∝ P^~2/3) | R_p decreases with P (∝ P^~−1/3) |
| Key relation | Ṁ ∝ L_XUV a^−2; envelope stripped below critical flux | a_F ≈ 2 r_tide, r_tide ∝ (M*/M_p)^(1/3) R_p |
| Outcome for planet | Neptune → stripped super-Earth (rocky core) | Cannot circularize inside limit → no close-in Neptunes |
Frequently asked questions
Why is it called a 'desert' and not just a gap?
Because it is a genuinely empty region, not merely underdense, and because it is bounded on multiple sides — like a desert hemmed in by fertile land. In the period–radius plane, hot Jupiters lie above it, small rocky planets below it, and longer-period Neptunes to its right, but the short-period Neptune interior is nearly barren. The term also distinguishes it from a one-dimensional dip like the radius valley.
Why do two different mechanisms make the same feature?
Because the desert has two edges with opposite slopes, and each edge is a different physical limit. The lower (small-planet) boundary is a survival limit set by XUV photoevaporation, which strips light H/He envelopes. The upper (large-planet) boundary is an arrival limit set by tidal disruption during high-eccentricity migration: massive planets can't circularize inside about twice their Roche radius. The two lines meet near P ≈ 10–15 days, closing the triangle.
How does photoevaporation actually strip a planet?
Stellar X-ray and extreme-UV photons are absorbed at the top of the H/He envelope, heating it to roughly 10⁴ K. That energy exceeds the gas's local gravitational binding, launching a transonic hydrodynamic wind — 'blow-off.' Mass loss rises steeply as the orbit shrinks (roughly as a^−2), so close-in low-gravity Neptunes lose their entire gaseous envelope within ~0.1–1 Gyr, leaving a bare rocky or icy core: a super-Earth.
What is high-eccentricity migration and why does it bound the upper edge?
It's a route to close-in orbits in which a companion (via planet–planet scattering or Lidov–Kozai/von Zeipel–Lidov–Kozai cycles) pumps a planet to high eccentricity, and tidal friction then shrinks and circularizes the orbit. Tides park the planet near a_F ≈ 2 r_tide. Since the tidal radius r_tide ∝ (M*/M_p)^(1/3) R_p is larger for lower-mass planets, Neptunes can't circularize close enough to reach the shortest periods — carving the desert's upper boundary.
How is the Neptunian desert different from the radius valley (Fulton gap)?
The radius valley is a bimodality in planet radius near 1.8 R⊕ — a scarcity of planets between super-Earths and mini-Neptunes — spanning a broad range of periods and explained mainly by atmospheric loss. The Neptunian desert is a two-dimensional void in period–radius space at periods below ~3 days, and its upper edge requires an extra dynamical (tidal-disruption) engine that the radius valley does not.
Are there any planets inside the desert, and why do they matter?
Yes — a handful of 'desert survivors' like LTT 9779 b (an ultra-hot Neptune on a 0.79-day orbit) and TOI-849 b (a dense remnant core). They matter because they are natural probes of the sculpting physics: measuring their masses, eccentricities, spin–orbit angles (via Rossiter–McLaughlin), and escaping-gas signatures (Lyman-α and the 10830 Å helium triplet) tests directly which engine dominates and why these worlds escaped it.