Planetary system architectures in the final stages of planet formation

The gist The final stages of planet formation determine the final assembly and composition of planets through ongoing collisions between planetesimals, which produce debris discs that trace system dynamics and volatile delivery.

Scientific context and open questions

When dust- and gas-rich protoplanetary discs dissipate after a few Myr, they leave behind planets and planetesimal belts: rings of asteroidal and cometary bodies, analogous to the Solar System’s Asteroid and Kuiper Belt (e.g. Hughes et al. 2018) Continuing collisions between these planetesimals release second-generation dust and gas, observed as a debris disc This debris makes the belts detectable in scattered light and thermal emission across wavelengths ranging from the infrared to the millimetre We refer to cold outer belts as exo-Kuiper belts (Sec. 2.2, 2.3) and to warm inner belts as exo-Asteroid belts (Sec. 2.1, 2.3) Planetary systems then enter the final phase of planet formation, lasting up to a few 100 Myr (e.g. Hughes et al. 2018), in which collisions, giant impacts, and volatile delivery continue to shape the final assembly of planets, while the planets interact with the surrounding disc (e.g. Lisse et al. 2009) Dust and gas track these processes directly: their composition traces rocky and icy planetesimals, and thus the bulk composition of rocky planets

Terrestrial planet formation: the aftermath of planetary impacts

Terrestrial planet formation in the inner regions of planetary systems continues after the gas-rich protoplanetary disc has dissipated, with collisions between planetary embryos (giant impacts) dominating late-stage interactions and releasing dust and gas into the inner few au between ∼10 and a few 100 Myr (e.g. Morbidelli et al. 2012) Evidence suggests an era of giant impacts in the Solar System, e.g., Earth’s Moon, Mercury’s iron enrichment, and the Martian hemispheric dichotomy (Wyatt & Jackson 2016, and references therein) Despite this being a crucial phase, it remains largely unconstrained in extrasolar systems because the dust lies within a few au of the star (Sub-)mm imaging is ideal for this as the disc-to-star contrast is orders of magnitude higher than at shorter wavelengths, revealing the terrestrial regions free of stellar contamination Only HD 172555, at ∼ 29 pc – one of the closest giant-impact systems – has been resolved in both dust and CO gas emission with ALMA (Roumeliotis et al. 2026) The CO gas, at ∼5 au from the star, would be quickly photodissociated if released by a collisional cascade, so it points to planetary atmospheric stripping during the impact Tens of other ∼ 10 to ∼ 100 Myr-old systems show warm IR dust (100s of K) in the terrestrial region that is too bright to be explained by Asteroid belts alone These extreme debris discs (e.g. Wyatt et al. 2007; Balog et al. 2009; Fujiwara et al. 2012) are thought to result from large collisions between Moon- to Mars-sized bodies and can vary on timescales of months to years (e.g Moor´ et al. 2021; Su et al. 2026) Characterising them requires the sensitivity to survey large samples and the resolution to image gas and dust structures in their terrestrial regions With ALMA2040, many giant impact systems could be resolved and monitored (Fig. 2)

Detecting and resolving debris discs around cool M dwarf stars

M dwarfs make up ∼70% of stars (Bochanski et al. 2010) and are the most favourable hosts for detecting terrestrial planets (Shields et al. 2016), yet their exo-Kuiper belts remain almost unexplored In decades of infrared and optical searches (e.g. Gautier et al. 2007; Lestrade et al. 2009; Avenhaus et al. 2012; Sgro & Song 2021), most candidates proved to be stellar flares, background galaxies or protoplanetary discs Herschel was the first to reach the cold dust around the nearest field M dwarfs, with detection rates of ∼ 2-5% (black in Fig. 3; Lestrade et al. 2012; Kennedy et al. 2014, 2018) ALMA increased this to 11% in the younger β Pic moving group (BPMG; blue in Fig. 3; Cronin-Coltsmann et al. 2023) and JWST has since added detections in the TW Hydrae Association (TWA; gold in Fig. 3; Palatnick et al. 2025; Crotts et al. 2025) Of the ∼13 known M dwarf debris discs, only 6 have been detected in the millimetre, and of these only four are well resolved Yet a statistical analysis of Planck data suggests that ∼ 10% of M dwarfs within 80 pc have debris emission (Nibauer et al. 2020), consistent with Luppe et al. (2020), who argue that these belts are not absent, but simply too faint for current mm facilities Current ALMA is well suited for bright belts around luminous stars (Matra et al. 2025), but reaching M dwarf belts requires a transformative sensitivity increase

Detailed debris disc substructures and planet-disc interactions

Debris discs trace the formation and dynamical history of planetary systems Just as Jupiter’s and Neptune’s current orbits and migration histories sculpted the structures of the Asteroid and Kuiper belt respectively (trapping planetesimals in mean motion resonances, clearing unstable orbital radii, pumping up orbital inclinations and eccentricities; Wisdom 1982; Fernandez & Ip 1984; Malhotra 1995) the detailed spatial distribution of dust in debris discs can reveal the presence, orbits and masses of nearby, sculpting exoplanets (see e.g. Mouillet et al. 1997; Wyatt et al. 1999; Quillen 2006; Pearce et al. 2022) Because many planets predicted from debris disc architectures lie well below the detection limits of current exoplanet-hunting facilities, debris disc substructures can offer unique views of the otherwise inaccessible population of exoplanets in the Galactic neighbourhood (typically cool Saturn- and Neptune-analogues) With a transformative increase in sensitivity at (sub-)mm wavelengths (Sec. 3), ALMA2040 offers three capability enhancements to study debris disc substructures and planet-disc interactions i) Detecting or ruling out features predicted by disc-planet interaction models; e.g. co-orbital populations (Fig. 1), gaps, spiral arms, resonant populations and warps, turning ambiguous detections into robust constraints on unseen outer planet populations ii) An order-of-magnitude increase in the population of discs available for high angular-resolution mm-studies: in 100 h, an ARKS-like survey would reach > 200 exo-Kuiper belts with ALMA2040, compared to <40 belts with current ALMA (Fig.

1. Enhanced Planet Formation Modeling This improved system can now detect or rule out features predicted by disc-planet interaction models; e.g., co-orbital populations (Fig. 1), gaps, spiral arms, resonant populations and warps. This allows for more robust constraints on unseen outer planet populations by moving from ambiguity to constraint.

2. M Dwarf Belt Census and Characterization The system can perform a statistical analysis of Planck data suggests that ∼10% of M dwarfs within 80 pc have debris emission, allowing it to move beyond the brightest ∼25% to survey the faint majority of exo-Kuiper belts, specifically addressing the question: What fraction of M dwarfs, the most common planet hosts, harbour exo-Kuiper belts.

3. Giant Impact System Monitoring The system can conduct high-resolution monitoring for terrestrial planet formation by enabling routine high-resolution imaging to characterise the composition and evolution of debris, allowing for a direct measurement of the width of the disc, which is directly related to the mass of the largest impacting planet.

4. Volatile Delivery Tracing The improved system can perform detailed gas measurements across spectral types and ages by resolving CO, C I and potentially H2O and OH, allowing it to test whether the gas in massive debris discs is secondary or primordial and constrain volatile delivery to terrestrial planets.

5. Population Statistics Across System Architecture The AI system can shift research from individual case studies to population-level statistics across stellar host mass, age, and system architecture, enabling the quantification of how dust structure evolves based on factors like system age and spectral type.

The processes in the last stages of planet formation determine the final assembly and composition of planets. After the protoplanetary discs dissipate, a system of planets and planetesimals -- analogues of the Asteroid and Kuiper Belts -- remains. Ongoing collisions between these bodies produce second-generation dust and gas that form a debris disc. Resolved images of such belts allow us to trace the dynamical history of these systems, including giant impacts and planet-belt interactions, during the crucial era from 10 to a few 100 Myr of terrestrial planet formation. High-resolution mm and optical/near-IR imagers have revolutionised our view of the brightest planetesimal belts, providing first glimpses of diverse structures, many of which show plausible planet imprints, while ALMA has revealed gas in a growing number of systems. Yet these results are strongly biased toward the brightest, mostly young belts around early-type stars, and even for these, crucial details remain undetectable at the required resolution due to their typically low surface brightness. A future (sub-)mm facility with an order of magnitude leap in continuum sensitivity and 10x higher line sensitivity would transform our understanding of late-stage planet formation by enabling the first statistically meaningful surveys of faint outer planetesimal belts across all spectral types, imaging planetary collisions in the inner regions of planetary systems, and gas measurements that trace volatile delivery to terrestrial planets. For the first time, this would reach the faint majority of belts, rather than only the rare bright discs accessible today.