Ercolano Group · University Observatory Munich (LMU)

How does a planet‑forming disc die?

Three master's thesis projects on the physics that disperses protoplanetary discs — X‑ray heated winds, gaps carved by planets, and the dust left behind.

FormatComputational astrophysics, M.Sc. thesis
GroupErcolano Group, LMU Munich
FocusDisc winds & planet–disc interaction
Three projects below
About
the group

Protoplanetary discs don't fade quietly — they're heated, blown apart, and carved up by the very star and planets they're feeding. Our group models this end stage of disc evolution: X‑ray driven photoevaporative winds, the gaps that planets and winds open together, and the dust that gets left stranded when the gas is gone.

The three projects below sit at different points of the same story — from the microphysics of what heats and cools the wind, to how that heating shapes a forming cavity, to what happens to solids once the gas starts to disappear. Each is a self-contained M.Sc. project, but they connect, and a student on one is welcome to talk to whoever's working on the others.

RT · Cooling physics

Cooling the wind: a better temperature recipe for X‑ray photoevaporation

Radiative transfer · look‑up tables · hydrodynamics

The big picture

Stellar X‑rays heat the surface of a protoplanetary disc enough to unbind gas and launch a slow wind that eventually disperses the disc. To predict how fast that happens, our hydro models need to know the gas temperature everywhere — but solving the full radiative transfer at every hydro step is far too expensive.

The open problem

Our usual shortcut assumes radiative equilibrium: a Monte Carlo radiative transfer code pre‑computes temperature as a function of local density, distance from the star, X‑ray luminosity and column density, stored as look‑up tables in the ionization parameter ξ. It's fast, and it's what lets us scan a large parameter space. But when the same physics was run with a full on‑the‑fly thermochemical code instead (Sellek et al. 2024), several cooling channels turned out to be missing from the equilibrium tables — molecular cooling isn't included at all. In particular, O–H collisional cooling is already modelled in our radiative transfer code but never fed back into the temperature tables, and it looks like exactly the channel that's cooling the wind region in the full model and suppressing the mass‑loss rate relative to our simple approach.

What you'd do

  • Extend the radiative equilibrium calculation to include O–H collisional cooling
  • Regenerate the ξ–column density look‑up tables with this channel included
  • Re‑run the 2D hydro wind models with the new tables
  • Compare mass‑loss rates against the full thermochemical benchmark of Sellek et al. (2024)
  • Run a disc population synthesis using the improved mass-loss rates.

Why it matters

If this closes the gap, we get a cheap, physically motivated recipe for exploring how mass‑loss rates depend on stellar spectrum and X‑ray luminosity — without needing a full thermochemical run for every point in parameter space.

Monte Carlo RTPythonHydrodynamicsPLUTOBenchmarking
ξ – N_col LOOK-UP TABLE N_col → ξ WIND TEMPERATURE O–H cooling rad. equil. + O–H, target
Left — the ξ vs. column-density table interpolated during a hydro step. Right — schematic: adding O–H cooling should pull the equilibrium temperature curve down toward the full thermochemical result.
GAP CROSS-SECTION wind not pressure- supported, falls back direct irrad. screened here cavity wall diffusion + direct irradiation — both needed for the gap's true temperature
A gap opened by a planet or by photoevaporation: the region above it is under-pressured, so any thermal wind launched there tends to fall back and partially refill the cavity.
3D‑RT · Gap dynamics

Does the wind fall back? Testing gap temperatures with 3D radiative transfer

3D radiative transfer · post-processing · disc dispersal

The big picture

When a planet — or photoevaporation itself — opens a gap in the disc, our previous work showed something counter-intuitive: the region just above the gap is under-pressured, so a thermal wind launched there isn't properly pressure-supported and partly falls back in, refilling the cavity (Weber et al. 2022). The same effect turns out to slow down disc dispersal itself, because the gap struggles to fully open while viscosity, lack of wind, and this fall-back keep feeding it (Weber et al. 2026).

The open problem

The picture rests on a radiative-equilibrium temperature estimate inside the gap, and reviewers have pushed back on exactly that: a forming cavity isn't simply irradiated from the star — direct starlight is partly screened by the cavity wall, and thermal diffusion from the hot wall itself should also matter. Our simple approach doesn't capture this effect properly.

What you'd do

  • Take existing hydro simulations from Weber et al. (2022, 2026) as the base state
  • Post-process them with a 3D radiative transfer code
  • Compare the resulting gap temperature structure to our original radiative-equilibrium assumption
  • Test whether the fall-back / slow-dispersal picture still holds up

Why it matters

This directly answers a standing criticism of two of the group's papers, and either strengthens or usefully revises a picture that changes how long we think discs take to fully clear.

3D radiative transferPost-processingPythonHydro simulation data
Dust · Planetesimals

Dust traps at the disc's end: modelling planetesimal formation during dispersal

Dust dynamics · Lagrangian particles / dust fluid · disc dispersal

The big picture

Disc dispersal turns out to be messier than the textbook picture. Modelled properly in 2D, the gap that eventually clears the inner disc opens much more slowly, and gas keeps crossing from the outer to the inner disc for a long time — accretion onto the star stays high even as the disc appears to be dying (Weber et al. 2026).

The open problem

The slowly-opening cavity still creates a pressure maximum at its edge, and pressure maxima are efficient dust traps. As the gas depletes toward the end of the disc's life, particles with Stokes number ~1 — the ones that drift and get trapped most efficiently — shift to progressively smaller physical sizes. That suggests the trap becomes efficient across a wide range of grain sizes right when it matters most: the disc could look like a transition disc in dust continuum emission while still looking gas-rich and actively accreting in every other tracer.

What you'd do

  • Run a dust module on top of existing dispersal hydro simulations — Lagrangian super-particles, and/or a two-fluid dust code with growth and fragmentation
  • Track how trapping efficiency evolves as the gas depletes and the gap slowly opens
  • Test whether trapped dust can locally grow toward planetesimal-forming conditions
  • Run a radiative transfer code to compare dust continuum emission from the simulations to real transition-disc observations

Why it matters

It offers a physical explanation for discs that look "transitional" in dust but not in gas or accretion — and a possible last site of planetesimal formation before the disc disappears entirely.

Dust dynamicsLagrangian particlesRadiative transferCDust growth/fragmentationPythonPlanet formation theory
PRESSURE BUMP / DUST TRAP gas pressure max as gas depletes, the trapped grain-size range (St ≈ 1) shifts smaller — trap widens
Dust of a range of sizes drifts toward the pressure maximum at the gap edge and concentrates there — a possible site for continued grain growth even as the gas disperses.