betaFeedback

WorksPhysics and Astronomy

Missing Outer Rings in the exoALMA Sample: Physical Mechanisms Suppressing Dust Ring Formation Despite Strong Pressure Bumps

Abstract

The exoALMA survey has revealed candidate gas pressure bumps at radii beyond the observed millimetre continuum discs in several protoplanetary systems, raising the question of whether these regions could host undetected dust structures. We investigate this for nine of the exoALMA discs by reconstructing gas surface density profiles from midplane pressure gradient measurements derived from 12^{12}CO and 13^{13}CO ALMA Band 7 observations. We evolve the dust using DustPy across a grid of turbulence and fragmentation parameters (α∈{10−4,10−3}α\in \{10^{-4}, 10^{-3}\}; vfrag∈{1,10} m s−1v_\mathrm{frag} \in \{1, 10\}\,\mathrm{m\,s^{-1}}), followed by radiative transfer calculations with RADMC-3D to give predictions for the continuum emission. We find that the dominant outcome across the explored parameter space is efficient outer dust trapping, leading to prominent outer continuum rings that should be detectable at ALMA Band 7, in tension with the observations. The 13^{13}CO-based reconstructions almost universally produce detectable outer rings, whereas the 12^{12}CO-based reconstructions show more variation. We further test delayed formation of the outer pressure bump and enhanced outer-disc turbulence as possible suppression mechanisms of ring formation. Enhanced turbulence weakens the predicted outer continuum emission most effectively, although the efficiency of the mechanism depends on the system explored and on where the turbulence transition is placed. Delayed bump formation generally does not prevent rapid outer dust accumulation.

Introduction

High-resolution millimetre continuum observations with ALMA have shown that protoplanetary discs commonly have substructures in the form of rings, gaps, spirals and azimuthal asymmetries (e.g. [1, 82] (Andrews et al. 2018; Huang et al. 2018; Long et al. 2018; Cieza et al. 2021; Jennings et al. 2022; Guerra-Alvarado et al. 2025; Bosschaart et al. 2026)). Early surveys of the brightest and most spatially extended discs, such as the DSHARP survey (Andrews et al. 2018; Huang et al. 2018) demonstrated that these features are widespread. Similar structures were also identified across discs in Taurus and Ophiuchus star forming regions (Long et al. 2018; Cieza et al. 2021). Subsequent work has moved towards showing the high prevalence of substructures throughout the broader disc population: van der Marel et al. [82] found that ring/gap systems span a wide range of stellar ages and luminosities, while parametric and non-parametric visibility fitting techniques demonstrated that substructures could be recovered at sub-beam resolution, including in systems with fainter and more compact emission (e.g. Long et al. 2019; Jennings et al. 2022). More recently, Bosschaart et al. (2026) found that rings and gaps are a near-universal trait of extended discs (Rdust>30 auR_{\mathrm{dust}} > 30\ \mathrm{au}) in nearby star-forming regions. However, this picture may not extend to the entire disc population; Guerra-Alvarado et al. (2025) found that two-thirds of the Lupus discs are compact (Rdust<30 auR_{\mathrm{dust}} < 30\ \mathrm{au}) and look smooth at ∼4\sim4 au resolution, suggesting that substructures could be more prominent primarily in the larger and more massive systems. The AGE-PRO large programme (Zhang et al. 2025), which targeted a representative sample spanning the disc lifetime in the Ophiuchus (≲1\lesssim1 Myr), Lupus (∼1−3\sim1-3 Myr) and Upper Scorpius region (∼2−6\sim2-6 Myr), provided some further insights. Visibility fitting of the continuum emission of the resolved discs revealed substructures in ∼80%\sim80\% in the Ophiuchus discs, providing support for early formation of substructures (Vioque et al. 2025), in agreement with the results from Hsieh et al. (2025) showing that dust substructures are already common in Class I discs. Substructures were similarly detected in the majority of well-resolved discs across the regions in the AGE-PRO sample; however, many of the compact discs (particularly in Lupus) are only marginally resolved, and some showed tentative visibility variations suggestive of unresolved structure (Vioque et al. 2025). Consistent with this, comparisons with dust evolution modelling suggest that the evolution of spectral index and continuum flux of the AGE-PRO sample are better reproduced by models with dust traps (Kurtovic et al. 2025; Luo et al. 2026). This could imply that substructures are common even in compact discs but are hidden below current resolution limits. The nature and prevalence of substructures in the faint and compact discs are therefore still open questions.

These observations have motivated extensive efforts to link observed dust structures with underlying disc physics. Disc substructures are commonly interpreted as signatures of dust trapping at local gas pressure maxima, where the radial inward drift of solids is halted (e.g. Weidenschilling 1977; Whipple 1972). These pressure maxima can arise through a diverse set of physical mechanisms, including gravitational interactions with planets embedded in the disc (Pinilla et al. 2012; Dong et al. 2015; Dipierro et al. 2015b), self-gravitating spiral arms (Dipierro et al. 2015a), and magnetohydrodynamic effects such as zonal flows and dead-zone edges (e.g. Johansen et al. 2009; Flock et al. 2015) (see Bae et al. (2023) for a comprehensive review of proposed mechanisms). In such regions, particles accumulate and grow, producing the ring-like features seen in millimetre continuum emission. Theoretical and numerical studies have shown that these pressure perturbations can efficiently concentrate dust and create long-lived substructures (e.g. Pinilla et al. 2012; Dullemond et al. 2018; Stammler et al. 2019). Observational support for dust trapping has emerged from ALMA observations, where gas kinematic perturbations and deviations from Keplerian rotation are found to be spatially associated with continuum rings and gaps, consistent with underlying pressure structure (e.g. Pinte et al. 2018; Teague et al. 2018; Rosotti et al. 2020; Izquierdo et al. 2023).

Line-intensity maps of CO isotopologues are only an indirect probe of the midplane pressure structure, since the mapping from the gas column density to pressure is complicated by optical depth and thermal effects. Comparisons of the CO line emission and continuum substructures find that CO gaps tend to coincide with dust gaps but are often shallower and some deep dust gaps lack any clear CO counterpart (Zhang et al. 2021). However the statistical significance of any continuum and line emission substructure correlation has been questioned (Jiang et al. 2022).

A more direct tracer is therefore the gas kinematics: deviations of the rotation curve from Keplerian rotation reflect the radial pressure gradient in the disc midplane, and can reveal pressure perturbations even where the corresponding change in gas column density is weak or ambiguous. This motivates the use of rotation curves to locate candidate pressure substructures.

The exoALMA survey (Teague et al. 2025) has enabled uniform, high-resolution analyses of gas emission and inferred midplane properties for a sample of nearby protoplanetary discs. By combining high spectral and spatial resolution observations of CO isotopologues, exoALMA gives detailed constraints on the gas kinematics and thermodynamic structure of several bright discs, offering a powerful means of probing embedded perturbations such as those induced by forming planets. Analysis of exoALMA rotation curves has revealed substructures in the radial derivative of the midplane gas pressure in several systems (Stadler et al. 2025). Many of these features were shown to coincide with known dust continuum rings, supporting the interpretation that local pressure maxima act as efficient dust traps. However, a notable subset of systems exhibits additional pressure-gradient substructures at radii beyond the observed millimetre continuum disc edge. As illustrated in LkCa 15 in Figure 1, the pressure structures within R90,obsR_{90,\mathrm{obs}} are well matched by the observed continuum rings, while those at large radii lack any corresponding millimetre emission.

Figure 1. Illustration of the missing dust rings in pressure bumps located outside the dust disc, shown here for LkCa 15 as a representative example. Inset: ALMA Band 7 (∼870 μm\sim870\,\mu\mathrm{m}) continuum image. Plot: Mid-plane pressure-gradient profile, dln⁡P/dln⁡R\mathrm{d}\ln P/\mathrm{d}\ln R, taken from exoALMA 12CO{}^{12}\mathrm{CO} observations and combining small- and large-scale measurements. Local variations in the pressure gradient trace candidate pressure perturbations that may act as dust traps. Within R90,obsR_{90,\mathrm{obs}} (black dashed vertical line) the pressure structures coincide with observed continuum rings (grey dashed vertical lines represent the locations of the observed bright rings), while beyond the dust disc, there is no millimetre emission at the location of the gas pressure bump.

These outer gas substructures should, in principle, act as efficient dust traps, yet they do not appear to produce any detectable continuum rings. Understanding why is therefore essential for interpreting the relationship between gas dynamics and dust evolution in protoplanetary discs. This raises a key question: are solids genuinely absent at these radii, or could dust be present but difficult to detect due to grain growth, fragmentation, radial drift, or opacity effects? Some observational evidence for the latter has recently been identified: Das et al. (2026) found extended low surface brightness, “halo”, emission in the outer regions of three systems, including DM Tau from the exoALMA sample. This component accounts for 20–30% of the total submillimetre flux and a significant fraction of the total dust mass, yet is cold and faint enough to be missed entirely in shallower surveys. Whether efficient trapping at a resolved outer gas substructure produces a localised continuum ring, more diffuse halo-like emission, or no detectable signature at all depends on the dust microphysics governing how solids respond to the bump. This is the question we address through forward modelling of dust evolution and synthetic continuum emission.

In this work, we consider two questions. First, if the outer pressure substructures inferred from exoALMA kinematics can act as dust traps, do standard dust evolution models predict outer continuum rings that exoALMA should have detected? Second, where such rings are predicted but not observed, which physical conditions render the trap sufficiently permeable, or the trapped dust sufficiently faint (the “hidden dust” scenario) to escape detection? To answer these questions, we reconstruct Σg(r)\Sigma_{\mathrm{g}}(r) for nine exoALMA discs from the 12CO{}^{12}\mathrm{CO} and 13CO{}^{13}\mathrm{CO} pressure-derivative profiles of Stadler et al. (2025), evolve the dust with DUSTPY across a grid of turbulence and fragmentation parameters (α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\,\mathrm{m\,s^{-1}}), and predict Band 7 continuum emission with RADMC-3D.

This paper is organised as follows. § 2 summarises the modelling pipeline (in Figure 2). § 3 presents the result of the entire sample, two contrasting discs, and two suppression tests. § 4 discusses the implications and limitations, and § 5 summarises our conclusions.

Overview of the modelling pipeline

Figure 2. Overview of the modelling pipeline. For each of the nine discs, we reconstruct two gas surface density profiles from the 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} pressure-derivative profiles of Stadler et al. (2025), using the midplane temperature of Galloway-Sprietsma et al. (2025), the gas mass of Trapman et al. (2025) and the characteristic radius of Longarini et al. (2025). Each profile is evolved with DUSTPY for four (α,vfrag)(\alpha, v_{\mathrm{frag}}) configurations and post-processed with RADMC-3D, giving 72 reference models whose Band 7 emission beyond R90,obsR_{90,\mathrm{obs}} (Curone et al. (2025)) is classified as Ring, Halo or No outer mm. Candidate outer-bump regions (dashed) are identified from the rotation curves for comparison only and do not enter the models. The suppression tests (orange) modify the gas history or the turbulence profile for selected discs.

Methods

Overview

Figure 2 summarises our modelling pipeline. For each of the nine discs, we reconstruct two independent gas surface-density profiles from the 12CO{}^{12}\mathrm{CO} and 13CO{}^{13}\mathrm{CO} pressure-derivative measurements of Stadler et al. (2025) (Section 2.3), evolve the dust in each for four (α,vfrag)(\alpha, v_{\mathrm{frag}}) combinations (§ 2.4), and post-process every model into a synthetic Band 7 image whose emission beyond the observed continuum edge is classified as Ring, Halo or No outer mm (§ 2.5). This gives 72 reference models (9 discs ×\times 2 tracers ×\times 4 configurations).

Independently, we identify candidate outer pressure substructures from the rotation curves (§ 2.2); these are used only for comparison with the predicted emission and do not enter the models. Finally, we test two mechanisms that could suppress outer ring formation; a late-forming outer bump and enhanced outer-disc turbulence, on a subset of discs (§ 2.6). The stellar and disc properties of the sample are listed in Table 2.

Discdd (pc)M⋆M_{\star} (M⊙M_{\odot})TeffT_{\mathrm{eff}} (K)L⋆L_{\star} (L⊙L_{\odot})MgasM_{\mathrm{gas}} (M⊙M_{\odot})RcR_{\mathrm{c}} (au)
AA Tau1350.7943501.100.024156
DM Tau1440.4537150.240.102240
HD 342823091.6195201.000.148370
J16151561.1443500.600.207167
J18421511.0747800.800.009231
J18521471.0347800.600.05787
SY Cha1820.7737920.470.017112
V4046 Sgr721.7343700.500.03599
LkCa 151571.1443651.000.055150

Table 2. Stellar and disc properties of the nine exoALMA systems modelled in this work. MgasM_{\mathrm{gas}} is the chemically estimated gas mass adopted as the normalisation for the reconstructed Σg(r)\Sigma_{\mathrm{g}}(r) profiles (Section 2.3). M⋆M_{\star} and dd are from Teague et al. (2025); TeffT_{\mathrm{eff}}, L⋆L_{\star} and MgasM_{\mathrm{gas}} from Trapman et al. (2025); RcR_{\mathrm{c}} from Longarini et al. (2025).

Identifying candidate regions for outer bumps

To identify candidate outer pressure-substructure regions, we used radial profiles of azimuthal velocity residuals from exoALMA data products (Stadler et al. 2025). The residual δvϕ=vrot−vKep\delta v_{\phi}=v_{\mathrm{rot}}-v_{\mathrm{Kep}} measures how much the observed gas rotation deviates from Keplerian. We express this as a fractional deviation,

Δvϕ=100×δvϕvKep(1)\Delta v_{\phi}=100\times\frac{\delta v_{\phi}}{v_{\mathrm{Kep}}} \tag*{(1)}

with significance

S=ΔvϕσΔvϕ,σΔvϕ=100×σδvϕvKep.(2)S=\frac{\Delta v_{\phi}}{\sigma_{\Delta v_{\phi}}},\qquad\sigma_{\Delta v_{\phi}}=100\times\frac{\sigma_{\delta v_{\phi}}}{v_{\mathrm{Kep}}}. \tag*{(2)}

Here, σδvϕ\sigma_{\delta v_{\phi}} is the statistical uncertainty on the azimuthal velocity residual provided in the Stadler et al. (2025) data products, reflecting the azimuthal scatter of the rotation curve within each radial annulus. We adopt the rotation curves measured from the high surface brightness sensitivity image cubes (0.30″ beam; Loomis et al. 2025), which offer the higher sensitivity to trace δvϕ\delta v_{\phi} in the faint outer disc. We searched only beyond R90R_{90}, the radius enclosing 90% of the Band 7 continuum flux (Curone et al. 2025), since inside R90R_{90} the pressure maxima broadly coincide with continuum rings (Stadler et al. 2025, Figure 1), whereas our question concerns pressure substructures without a continuum counterpart. As a spatial resolution criterion, we used the effective beam size θbeam=θmajθmin\theta_{\mathrm{beam}}=\sqrt{\theta_{\mathrm{maj}}\theta_{\mathrm{min}}} (the geometric mean of the beam axes, converted to au), requiring candidate regions to be wider than one beam.

We selected candidate regions as follows. We identified all connected zones beyond R90R_{90} where the gas is super-Keplerian (Δvϕ>0\Delta v_{\phi}>0). A zone was kept only if (i) it contains at least one point with Δvϕ>1%\Delta v_{\phi}>1\% and S>3S>3, ensuring a physically meaningful and statistically significant velocity excess, and (ii) its total width exceeds one beam, Δr>θbeam\Delta r>\theta_{\mathrm{beam}}, ensuring the feature is spatially resolved. Our use of super-Keplerian rotation as the selection criterion is physically motivated: the sign of Δvϕ\Delta v_{\phi} follows that of dln⁡P/dln⁡r\mathrm{d}\ln P/\mathrm{d}\ln r, so that super-Keplerian rotation (Δvϕ>0\Delta v_{\phi}>0) corresponds to a positive pressure gradient. A genuine pressure bump, where the radial drift of solids reverses, produces a transition from sub- to super-Keplerian rotation and back to sub-Keplerian rotation with increasing radius.

However, we note that sub-Keplerian bumps in the gradient will still perturb the pressure and can locally decelerate the inward drift of solids, producing a “traffic jam” pile-up (e.g. van der Marel & Pinilla [81]), but without reversing the drift.

This classification depends on the adopted stellar mass through vKepv_{\mathrm{Kep}}, so marginal features could shift across the threshold (Appendix § 9). The criterion is used only to identify and report candidate regions; the dust evolution models are driven by the full pressure-gradient profile and are hence unaffected by how a region is labelled.

Applying this criterion to the 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} rotation curves yields markedly different numbers of candidate regions (Figure 3). 12CO^{12}\mathrm{CO} identifies regions in 8 of the 9 systems, while 13CO^{13}\mathrm{CO} identifies regions in only 2 (DM Tau and HD 34282). In the remaining systems, except for J 1852, 13CO^{13}\mathrm{CO} velocity deviations are present beyond R90R_{90} but fall below the S>3S > 3 significance threshold. This could reflect lower signal-to-noise ratio (SNR) in the 13CO^{13}\mathrm{CO} profiles rather than a genuine absence of outer pressure structure. The velocity profiles and selected zones are shown in Appendix § 9.

Gallery of exoALMA continuum images for the nine discs modelled in this work

Figure 3. Gallery of exoALMA continuum images (using the robust parameter of −0.5-0.5) for the nine discs modelled in this work (data from Curone et al. (2025)). Shaded regions indicate candidate outer pressure-substructure zones identified from super-Keplerian velocity deviations in the Stadler et al. (2025) rotation curve profiles, exterior to the observed dust disc; pink and blue regions correspond to 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} selections respectively. The synthesised beam is shown in the lower-left corner of each panel.

Reconstructing Σg(r)\Sigma_{\mathrm{g}}(r) from pressure derivatives

We present our modelling approach in which the gas surface density profile input to DUSTPY is derived directly from the observed midplane pressure-derivative profiles of Stadler et al. (2025):

dln⁡Pgdln⁡r(r),(3)\frac{\mathrm{d}\ln P_{\mathrm{g}}}{\mathrm{d}\ln r}(r), \tag*{(3)}

rather than from a parametric prescription. This allows the dust evolution to be constrained by the actual kinematic structure of each disc.

We reconstruct Σg(r)\Sigma_{\mathrm{g}}(r) in four steps, detailed in Appendix § 8: (i) we blend the small- and large-scale pressure-derivative profiles of Stadler et al. (2025) and extend them beyond the observed radial range with a Lynden-Bell & Pringle (1974) profile; (ii) we integrate to obtain Pg(r)P_{\mathrm{g}}(r) up to a multiplicative constant; (iii) we convert pressure to surface density assuming a Gaussian vertical density profile and the disc-specific midplane temperature of Galloway-Sprietsma et al. (2025); and (iv) we normalise the profile to the total gas mass. The normalisation is necessary because the pressure gradient constrains only the shape of Pg(r)P_{\mathrm{g}}(r), not its amplitude. We adopt the chemically estimated gas masses of Trapman et al. (2025), which are independent of the kinematic analysis used to derive the pressure gradients. These are typically a factor of ∼\sim2–3 lower than the dynamical masses of Longarini et al. (2025); since the predicted ring brightness scales with the gas normalisation, our predictions are conservative in this respect.

We produce two independent Σg(r)\Sigma_{\mathrm{g}}(r) profiles per system using the 13CO^{13}\mathrm{CO} and 12CO^{12}\mathrm{CO} pressure-derivative measurements from Stadler et al. (2025) as separate model inputs. Since 13CO^{13}\mathrm{CO} is less optically thick than 12CO^{12}\mathrm{CO}, it traces deeper into the disc, and should in principle more accurately reflect the thermal and density structure of the midplane where dust trapping occurs (Galloway-Sprietsma et al. 2025).

Dust evolution with dustpy

We evolve dust in the reconstructed gas backgrounds using dustpy (Stammler & Birnstiel (2022)). dustpy solves the Smoluchowski coagulation-fragmentation equation for the dust surface density distribution Σd(r,a)\Sigma_{\mathrm{d}}(r,a), simultaneously modelling radial drift, turbulent diffusion, and collisional grain growth and fragmentation (Birnstiel et al. (2010)). The gas surface density is held fixed throughout: gas evolution is switched off, so the reconstructed pressure bumps are imposed directly through Σg(r)\Sigma_{\mathrm{g}}(r) rather than maintained through a radially varying α\alpha (e.g. Stammler et al. (2023)); feedback from dust onto the gas is neglected.

Dust is initialised with a dust-to-gas mass ratio of 0.01 and an MRN grain size distribution (Mathis et al. (1977)) between amin⁡=0.1 μma_{\min} = 0.1\,\mu\mathrm{m} and amax⁡=1.1 μma_{\max} = 1.1\,\mu\mathrm{m}. The initial distribution is set up with the dustpy option allowDriftingParticles set to True, so the full initial dust reservoir is retained at all radii, including grains that would already be in the drift limit at t=0t = 0.

At the low gas surface densities of the outer disc, even small grains have relatively large Stokes numbers, St≃πaρs2Σg\mathrm{St} \simeq\frac{\pi a\rho_{\mathrm{s}}}{2\Sigma_{\mathrm{g}}} in the Epstein regime, where aa is the grain radius and ρs\rho_{\mathrm{s}} the grain material density, and begin to drift from the start of the simulation; this early depletion of the outer dust reservoir is followed self-consistently in our models.

We evolve all models for 1 Myr. This should be interpreted as a characteristic dust evolution timescale following the formation of the bumps, rather than the actual disc age. The ages in our sample are poorly constrained and span a wide range (∼\sim1–24 Myr) (Huang et al. 2017), and the relevant timescale for dust accumulation in the bump is set by local drift and growth timescales rather than the age of the disc. Therefore, our models should not be intended to represent the full evolutionary history of each disc, but instead to assess whether these observed pressure structures are capable of producing detectable outer continuum emission, if given sufficient time for dust to accumulate.

For each model we record the dust surface density distribution Σd(r,a)\Sigma_{\mathrm{d}}(r,a), from which we derive the local fragmentation and drift barriers (afraga_{\mathrm{frag}}, adrifta_{\mathrm{drift}}) to characterise the grain size distribution (see Figure 7b and Figure 7a). Full model parameters are summarised in Table 1.

ParameterValue
Radial grid (rmin⁡r_{\min})0.5 au (20 au for select systems*)
Radial grid (rmax⁡r_{\max})Set by Stadler et al. (2025) data extent
Radial grid pointsScaled with rmax⁡r_{\max}
Initial dust size distributionMRN power-law (Mathis et al. (1977))
MRN slope3.5
Grain size grid (amin⁡a_{\min}, amax⁡a_{\max})10−510^{-5} cm, 10 cm
Fragmentation velocity (vfragv_{\mathrm{frag}})1, 10 m s−1^{-1}
Turbulence parameter (α\alpha)10−410^{-4}, 10−310^{-3}
CO tracer12CO{}^{12}\mathrm{CO}, 13CO{}^{13}\mathrm{CO} (independent models)
Dust evolution time1 Myr
Gas surface density Σg(r)\Sigma_{\mathrm{g}}(r)Reconstructed from Stadler et al. (2025) pressure derivatives; normalised to chemically estimated gas masses from Trapman et al. (2025)
Gas evolutionOff (Σg\Sigma_{\mathrm{g}} fixed)
allowDriftingParticlesTrue
Midplane temperature Tmid(r)T_{\mathrm{mid}}(r)Power-law fit from Galloway-Sprietsma et al. (2025)
Characteristic radius (RcR_{\mathrm{c}})From Longarini et al. (2025)
Initial dust-to-gas ratio0.01
Dust opacityRicci et al. (2010)

Table 1. Parameters for the dustpy dust evolution models. Varied parameters are in bold. *For systems (DM Tau, HD 34282) the inner boundary is set to 20 au, due to computational issues, though this should not affect the outer disc region of interest.

Dust density distributions for J1615 and LkCa 15

Figure 7. Dust density distribution after 1 Myr using the exoALMA gas surface density reconstructions derived from the 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} observations for J1615 (top panels) and LkCa 15 (bottom panels). White and red dashed curves indicate the drift and fragmentation barriers, respectively. Vertical dashed lines mark the dust substructures identified in the Band 7 continuum observations (Curone et al. 2025), with magenta lines corresponding to rings. The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4},10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\ \mathrm{m\,s^{-1}}.

We explore four combinations of turbulence parameter and fragmentation velocity,

α∈{10−4, 10−3},vfrag∈{1, 10} m s−1,(4)\alpha\in\{10^{-4},\,10^{-3}\}, \qquad v_{\mathrm{frag}} \in\{1,\,10\}\ \mathrm{m\,s^{-1}}, \tag*{(4)}

spanning ranges motivated by observational and laboratory constraints. Dust settling analyses of highly inclined discs suggest α≲10−3\alpha\lesssim10^{-3} (e.g. Villenave et al. 2022; Villenave et al. 2025), consistent with gas evolution modelling of the AGE-PRO sample which favours α∼2\alpha\sim2–4×10−44 \times10^{-4} (Tabone et al. 2025). For fragmentation velocity, silicate grains fragment near vfrag∼1 m s−1v_{\mathrm{frag}} \sim1\ \mathrm{m\,s^{-1}} while icy grains may survive up to ∼10 m s−1\sim10\ \mathrm{m\,s^{-1}}, though recent experiments suggest even icy aggregates can be fragile (e.g. Gundlach & Blum 2015; Musiolik & Wurm 2019). Models with α=10−4\alpha= 10^{-4} and vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}} were found to best reproduce observed dust distributions in the AGE-PRO sample (Luo et al. 2026); consistent conclusions have been reported by Jiang et al. (2024) and Tong et al. (2025) for other discs. The broader parameter grid is included to assess the sensitivity of the outer ring predictions to these choices, and in particular to explore scenarios where efficient grain growth concentrates mass into large, poorly emitting grains that would be undetectable at millimetre wavelengths despite being present in significant quantities.

Synthetic continuum images and detectability

We post-process DUSTPY outputs using RADMC-3D (Dullemond et al. 2012) to generate synthetic Band 7 (∼\sim870 μm) continuum emission maps, using the following approach:

3D density structure

The 1D radial dust surface density distribution Σd(r,a)\Sigma_{\mathrm{d}}(r,a) from DUSTPY is expanded onto a 3D spherical grid assuming a Gaussian vertical dust profile (Pohl et al. 2016),

ρd(R,z,a)=Σd(R,a)2π Hd(R,a)exp⁡(−z22Hd2(R,a)),(5)\rho_{\mathrm{d}}(R,z,a) = \frac{\Sigma_{\mathrm{d}}(R,a)}{\sqrt{2\pi}\,H_{\mathrm{d}}(R,a)}\exp\left(-\frac{z^2}{2H_{\mathrm{d}}^2(R,a)}\right), \tag*{(5)}

where Hd(a)H_{\mathrm{d}}(a) is the dust scale height for each grain size bin, computed following the standard settling prescription (Dubrulle et al. 1995),

Hd(a)=Hgαα+St(a),(6)H_{\mathrm{d}}(a) = H_{\mathrm{g}}\sqrt{\frac{\alpha}{\alpha+\mathrm{St}(a)}}, \tag*{(6)}

with HgH_{\mathrm{g}} the gas scale height and St(a)\mathrm{St}(a) the Stokes number of grains of size aa.

Dust temperature

The dust temperature is computed with a RADMC-3D thermal Monte Carlo calculation, using the stellar effective temperature and luminosity of each system listed in Table 2. This temperature accounts for irradiation and shadowing by the modelled inner disc, and is computed independently of the midplane temperature Tmid(r)T_{\mathrm{mid}}(r) used in the dust evolution. For LkCa 15, we confirmed that adopting the DUSTPY Tmid(r)T_{\mathrm{mid}}(r) instead does not change the classification of the outer emission at any of the wavelengths tested.

Dust opacities

We adopt dust opacities from Ricci et al. (2010), computed using optool (Dominik et al. 2021) for a composition of 21% graphite, 36% amorphous carbon, 28% water ice, and 15% silicate with 40% porosity. Ricci opacities have been shown to reproduce millimetre fluxes and spectral indices of disc populations better than the DSHARP prescription (Stadler et al. 2022; Delussu et al. 2024), motivating our choice.

Synthetic images

Inclined synthetic images are generated at the observed inclination and position angle of each disc and convolved with a 2D Gaussian matching the observational beam. We also generate synthetic images at longer wavelengths (1.3–8.6 mm) for selected systems, to evaluate whether outer dust concentrations might become more detectable at longer wavelengths.

Radial profiles and detectability

Deprojected radial intensity profiles are extracted from the convolved synthetic images by rotating by the position angle, dividing the declination axis by cos⁡(i)\cos(i), and azimuthally averaging in radial bins of one-fifth of the synthesised beam FWHM, i.e. ≈5\approx5 bins per beam.

Detectability is assessed against the observed sensitivity. Since the synthetic images are computed in absolute units at the distance of each source and convolved with the observed beam, we compare them directly with the rms noise of the observed Band 7 images, obs (Curone et al. 2025): an outer emission feature is considered detectable if the synthetic intensity exceeds 3σobs3\sigma_{\mathrm{obs}} beyond R90,obsR_{90,\mathrm{obs}}.

The synthetic disc radius R90,synR_{90,\mathrm{syn}} is then calculated from the masked model image. The synthetic images are noiseless and are not processed through the interferometric uvuv-sampling and imaging of the observations, so this criterion represents an idealised detection limit: emission just above 3σobs3\sigma_{\mathrm{obs}} would thus correspond to a marginal detection in real data, and very extended, low-surface-brightness emission may be partly filtered out by the interferometer.

For LkCa 15, the inner 20 au is masked in the synthetic images prior to profile extraction, as the inner disc continuum emission is poorly constrained (Sierra et al. (2025)).

We classify the resulting outer emission into three categories. Ring signifies a distinct local intensity maximum at r>R90,obs+0.5θbeamr > R_{90,\mathrm{obs}} + 0.5\theta_{\mathrm{beam}}; halo denotes emission detectable above the threshold but with no local maximum, the profile declining monotonically from the observed continuum disc edge, together with R90,syn>R90,obs+θbeamR_{90,\mathrm{syn}} > R_{90,\mathrm{obs}} + \theta_{\mathrm{beam}}; and no outer mm indicates no emission extending meaningfully beyond R90,obsR_{90,\mathrm{obs}}. Here θbeam=θmajθmin\theta_{\mathrm{beam}} = \sqrt{\theta_{\mathrm{maj}}\theta_{\mathrm{min}}} is the geometric mean beam FWHM evaluated per disc. R90,synR_{90,\mathrm{syn}} is computed from all emission above 3σ3\sigma. We adopt the term halo by analogy with the extended low-surface-brightness emission reported by Das et al. (2026), albeit here it refers to a morphological classification of our synthetic images.

Suppression tests: late bump and enhanced turbulence

In order to explore potential physical mechanisms that could suppress the formation of a detectable dust ring despite the presence of an outer gas pressure bump, we perform two additional tests on the case-study systems LkCa 15 and J1615, and additionally test enhanced-turbulence on AA Tau, for comparison. We focus on these discs as they bracket the range of behaviours in the 12CO{}^{12}\mathrm{CO} sample: LkCa 15 shows the strongest configuration dependence, with two of the four parameter combinations already falling below the detection threshold in the outer disc, whereas J1615 produces detectable emission in every configuration. In both, the outer bump lies well inside the radial extent of the kinematic data, so it can easily be removed and reintroduced without extrapolating the reconstructed profile. AA Tau produces detectable outer rings in every configuration and therefore represents the strongest tension with the observations; we do not adopt it as a primary case study because its reconstructed Σg\Sigma_{\mathrm{g}} profile increases with radius, which is atypical of the sample.

Late bump introduction

Whether these outer pressure structures are long-lived, as would be expected for gaps induced by planets, or transient features formed by e.g. recent infall or zonal flows, is unknown. If the bump formed only recently, the outer disc may host little solid material: either insufficient time has passed for the drifting grains to reach and accumulate at the pressure maximum, or most solids have already drifted inward past that location at the stage before the bump formed. Therefore we explore a late-bump scenario to test whether delayed formation of the outer pressure structures can suppress the formation of a detectable continuum ring.

If the bump is induced by a forming planet, the planet must first grow sufficiently massive to open a gap and perturb the gas pressure profile, which occurs on timescales comparable to the local orbital and viscous timescales (Pinilla et al. 2015). Alternatively, late infall of material from the surrounding envelope onto the outer disc could generate a new pressure feature at large radii (Das et al. 2026). Zhao et al. (2025) demonstrated that an infall induced pressure bump can efficiently trap dust and trigger streaming instability, forming a planetesimal belt, but only while the bump persists before being smoothed by viscous diffusion at the local viscous timescale.

To test this, we first evolve the disc for t=1t = 1 Myr with a gas surface density where the outer pressure bump has been removed. The bump region is replaced by a smooth cubic spline in log-log space, connected to the observed profile on either side of the bump and matching the local slope at both boundaries (shaded region in Figure 4). The bridged interval spans 100–410 au for LkCa 15 and 100–220 au for J 1615. In both cases the smoothed (no-bump) profile is renormalised to conserve the total disc gas mass.

Figure 4. Gas surface density profiles used in the late-bump test for LkCa 15 (left) and J 1615 (right), derived from the exoALMA 12CO^{12}\mathrm{CO} pressure-derivative reconstructions (Section 2.3). The reference profile (solid blue line) is the full reconstructed Σg(r)\Sigma_{\mathrm{g}}(r) including the outer pressure bump. The no outer bump profile (yellow solid line) is constructed using a smooth outer bridge. The shaded region hence indicates the bridge interval where the bump is removed. The dashed black lines indicate the edges of the radial range covered by the observed pressure-derivative profiles (labelled in au); inside and beyond these radii the profile is extended analytically (Appendix § 8). In the late-bump test, dustpy is first run for 1 Myr using the no outer bump profile as the gas background; the outer bump is then introduced by switching to the reference profile and the system is evolved for a further 0.5 Myr (see § 2.6).

We then introduce the perturbed outer profile, reconstructed as described in Section 2.3, and evolve for an additional 0.5 Myr. The gas surface density profiles before and after bump introduction are shown in Figure 4. This test follows a similar approach to Pinilla et al. (2015), who modelled the response of a dust disc to a pressure bump introduced after significant prior dust evolution.

Enhanced outer-disc turbulence

Regions of low ionisation in the disc midplane suppress the magnetorotational instability (MRI; Balbus & Hawley 1991), creating so-called dead zones where turbulence is weak and α\alpha is low. Beyond the dead zone, where ionisation from cosmic rays or stellar radiation is sufficient to sustain MRI, turbulence can be substantially higher. The location and extent of dead zones depend on the disc column density, the ionisation sources, and the magnetic field structure (e.g. Dzyurkevich et al. 2013; Flock et al. 2015; Delage et al. 2022).

Enhanced turbulence exterior to the main dust disc could suppress the formation of a detectable outer continuum ring through two related effects: (i) stronger turbulent diffusion counteracts the concentration of solids at the pressure maximum, reducing trapping efficiency; and (ii) higher turbulent velocities increase grain collision speeds, promoting fragmentation and limiting grain growth to smaller sizes with lower millimetre opacity (Birnstiel et al. 2010; Pinilla et al. 2012).

To test this, we implement a step-like turbulence profile,

α(r)={αin,r<Rdzαout,r≥Rdz,(7)\alpha(r) = \begin{cases} \alpha_{\mathrm{in}}, & r < R_{\mathrm{dz}} \\ \alpha_{\mathrm{out}}, & r \ge R_{\mathrm{dz}}, \end{cases} \tag*{(7)}

placing RdzR_{\mathrm{dz}} 5 au interior to the observed continuum disc edge R90,obs−5R_{90,\mathrm{obs}} - 5 au, i.e. 151, 165 and 134 au for LkCa 15, J1615 and AA Tau respectively. We adopt αout∼10−2\alpha_{\mathrm{out}} \sim10^{-2}, consistent with estimates for MRI-active outer disc regions (Flock et al. 2015), while αin\alpha_{\mathrm{in}} takes the reference values from Table 1. We do not apply α=10−2\alpha= 10^{-2} throughout the disc, since such high turbulence would suppress grain growth and equally suppress the observed inner rings. Although this step-function profile is a simplified phenomenological prescription and real ionisation transitions are expected to be smoother, it allows us to test whether the magnitude of turbulence enhancement required to suppress the outer ring is physically plausible. Physically, such a radial increase in turbulence is expected since the gas surface density drops in the outer disc, allowing cosmic rays to penetrate into the midplane and ionise it sufficiently to activate MRI. We note that because the gas surface density is held fixed in our simulations, no pressure bump forms at the α\alpha-transition itself as it would in models where gas is evolved (see e.g. Pinilla et al. 2016).

Results

We present dust evolution and radiative transfer results for nine exoALMA discs with reconstructed outer gas pressure substructures, representing the subset of the full 15 target exoALMA sample for which published pressure derivative data products suitable for forward modelling are available.1 Detailed figures for all nine systems are presented in the Appendix (Figure E4–Figure E18).

Dust density distribution for DM Tau across four dust evolution model configurations

Figure E4. Dust density distribution after 1 Myr for DM Tau, using the exoALMA gas surface density reconstructions derived from the 12CO^{12}\mathrm{CO} (top) and 13CO^{13}\mathrm{CO} (bottom) observations. White and red dashed curves indicate the drift and fragmentation barriers, respectively. Vertical magenta dashed lines mark dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The four columns represent different dust evolution model configurations spanning α∈{10−4, 10−3}\alpha\in\{10^{-4},\,10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\ \mathrm{m\,s^{-1}}.

Figure E18. RADMC synthetic observations of the dustpy 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} models. The fluxes are normalised to the peak. The inset shows the observation in B7 (Curone et al. 2025). The white ellipses indicate dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The white dashed lines indicate the 3σ3\sigma threshold based on the sensitivity of the observations. The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4},10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\ \mathrm{m\,s}^{-1}.

Sample-wide outcome

Figure 5 summarises whether each model produces detectable outer millimetre emission beyond the observed continuum disc edge (numerical values in Appendix D). Figure 6 and Figure E3 show the deprojected radial intensity profiles of the synthetic continuum emission for each disc and dust-evolution configuration, derived from the 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} reconstructions respectively.

Figure 5. Predicted outer Band 7 emission for the 72 reference models, for the 12CO^{12}\mathrm{CO}-based (left) and 13CO^{13}\mathrm{CO}-based (right) gas reconstructions. Rows are discs and columns the four (α,vfrag)(\alpha, v_{\mathrm{frag}}) configurations. Each symbol indicates whether emission above 3σ3\sigma beyond R90,obsR_{90,\mathrm{obs}} forms an outer ring, a halo, or is absent (§ 2.5); the symbols are schematic, not model images. The asterisk marks the J1842 model with no emission above 3σ3\sigma at any radius. Numerical values are given in Table D1 and Table D2.

DiscConfigOuter mm emissionR90,obsR_{90,\mathrm{obs}} (au)R90,synR_{90,\mathrm{syn}} (au)Candidate Rbump,gasR_{\mathrm{bump,gas}} region (au)
AA Tauα=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Ring139431268–460
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Ring545
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Ring552
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Ring546
DM Tauα=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}No outer mm20244200–340
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}No outer mm32
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}No outer mm34
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}No outer mm26
HD 34282α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}No outer mm23983243–382
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}No outer mm58
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}No outer mm48
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}No outer mm50
J1615α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Halo170384170–252
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Halo347
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Halo369
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Ring391
J1842α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Halo8520685–110
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Ring226
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Halo100
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}No outer mm—
J1852α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Ring70275—
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Ring203
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Halo169
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Ring215
LkCa 15α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}No outer mm15666159–254; 336–395
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Ring164
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Ring160
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}No outer mm36
SY Chaα=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}No outer mm19863198–334
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}No outer mm39
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}No outer mm41
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}No outer mm41
V4046 Sgrα=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Halo6127972–212
α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}Ring295
α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Ring169
α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}Ring257

Table D1. Summary of the synthetic Band 7 continuum outcomes for the 12CO^{12}\mathrm{CO}-based dust-evolution models. The outer mm emission column classifies the morphology of any emission detected above 3σ3\sigma beyond the observed continuum disc edge (R90,obsR_{90,\mathrm{obs}}, from Curone et al. (2025)). The final column lists the candidate outer gas pressure-substructure regions inferred from the 12CO^{12}\mathrm{CO} kinematics.

Table D2. Summary of the synthetic Band 7 continuum outcomes for the 13CO^{13}\mathrm{CO}-based dust-evolution models. The outer mm emission column classifies the morphology of any emission detected above 3σ3\sigma beyond the observed continuum disc edge (R90,obsR_{90,\mathrm{obs}}, from Curone et al. (2025)). The final column lists the candidate outer gas pressure-substructure regions inferred from the 13CO^{13}\mathrm{CO} kinematics.

DiscConfigOuter mm emissionR90,obsR_{90,\mathrm{obs}} (au)R90,synR_{90,\mathrm{syn}} (au)Candidate Rbump,gasR_{\mathrm{bump,gas}} region (au)
AA Tauα=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring139304—
AA Tauα=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring286
AA Tauα=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring285
AA Tauα=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring281
DM Tauα=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}No outer mm202186200–448
DM Tauα=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}No outer mm80
DM Tauα=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring296
DM Tauα=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring432
HD 34282α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring239562243–521
HD 34282α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring553
HD 34282α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring552
HD 34282α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring543
J1615α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo170413—
J1615α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring365
J1615α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring340
J1615α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring359
J1842α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring85181—
J1842α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring178
J1842α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring176
J1842α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring178
J1852α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring70180—
J1852α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo137
J1852α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Halo104
J1852α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring145
LkCa 15α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring156453—
LkCa 15α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring547
LkCa 15α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring461
LkCa 15α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring506
SY Chaα=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo198259—
SY Chaα=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring355
SY Chaα=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring333
SY Chaα=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring334
V4046 Sgrα=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo61247—
V4046 Sgrα=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Ring253
V4046 Sgrα=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring148
V4046 Sgrα=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring265

Table D2.

Azimuthally averaged, deprojected radial intensity profiles of synthetic Band 7 continuum emission for the full disc sample

Figure 6. Azimuthally averaged, deprojected radial intensity profiles of the synthetic Band 7 continuum emission for the full disc sample, computed from radmc-3d images convolved to the observed beam of each source. Each panel shows the four dust-evolution model configurations explored in this work, α∈{10−3,10−4}\alpha\in\{10^{-3},10^{-4}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\,\mathrm{m\,s^{-1}}, for the 12CO^{12}\mathrm{CO}-based reconstructions. Intensities are shown in absolute units on a logarithmic scale. The horizontal dotted line marks the observed 3σ3\sigma sensitivity of each disc, so emission above it would be detectable at the exoALMA sensitivity. The dashed vertical line marks R90,obsR_{90,\mathrm{obs}} and the grey solid lines the observed dust rings (Curone et al. 2025). Pink (12CO^{12}\mathrm{CO}) and blue (13CO^{13}\mathrm{CO}) shaded regions indicate the candidate outer gas substructures inferred from the kinematic analysis. The inner 20 au of LkCa 15 is masked following Sierra et al. (2025).

Figure E3. As Figure 6, but for the 13CO^{13}\mathrm{CO}-based reconstructions.

The dominant outcome across the sample is efficient outer dust trapping: 21 of our 36 12CO^{12}\mathrm{CO}-based models and 34 of the 36 13CO^{13}\mathrm{CO}-based models produce emission detectable above 3σ3\sigma beyond the observed continuum disc edge (9 targets ×\times 4 parameter combinations per tracer). Across a wide range of (α,vfrag)(\alpha, v_{\mathrm{frag}}) combinations, the reconstructed gas profiles frequently produce substantial dust pile-ups at large radii, resulting in prominent outer continuum rings in the synthetic Band 7 emission assuming the same angular resolution and sensitivity as the exoALMA observations.

A notable result is that the 13CO^{13}\mathrm{CO}-based reconstructions almost universally produce detectable outer continuum rings across the explored parameter space (Figure 5, right). Even in the few marginal cases, the synthetic continuum disc extends substantially beyond the observed R90,obsR_{90,\mathrm{obs}}, indicating efficient outer dust trapping. Most systems and configurations produce an outer peak emission (ring) and all models yield synthetic continuum emission extending well beyond the observed millimetre disc radius.

In contrast, the 12CO^{12}\mathrm{CO}-based reconstructions exhibit substantially greater diversity (Figure 5, left). While AA Tau, J1615, J1842, J1852, and V4046 Sgr commonly produce detectable outer rings, several systems and parameter combinations do not generate significant outer continuum emission.

Within the 12CO^{12}\mathrm{CO} sample, LkCa 15 shows the strongest configuration dependence, with two of the four configurations producing no detectable outer emission; we examine it in § 3.2.

SY Cha, DM Tau, and HD 34282 do not produce a detectable outer ring across all explored 12CO^{12}\mathrm{CO} configurations. In addition, the model with α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\ s}^{-1} for J 1842 also lacks detectable outer continuum emission; in this model no emission is detected above 3σ3\sigma at any radius (the corresponding panel in Figure E10 appears bright only because each image is normalised to its own peak), because efficient trapping in the outer bump grows grains to sizes with low Band 7 opacity and prevents dust from leaking through to the inner disc (Figure E9). In some of these cases, the dust evolution models nevertheless show narrow outer dust rings with efficient growth (see Appendix Figures E9 and E15), meaning that the absence of observable Band 7 emission does not necessarily imply the absence of dust trapping in the outer disc.

Dust density distribution after 1 Myr for J 1842

Figure E9. Dust density distribution after 1 Myr for J 1842, using the exoALMA gas surface density reconstructions derived from the 12^{12}CO (top) and 13^{13}CO (bottom) observations. White and red dashed curves indicate the drift and fragmentation barriers, respectively. Vertical magenta dashed lines mark dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1, 10\}\,\mathrm{m\,s^{-1}}.

Figure E10. RADMC synthetic observations of the dustpy 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} models. The fluxes are normalised to the peak. The inset shows the observation in B7 (Curone et al. 2025). The white ellipses indicate dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The white dashed lines indicate the 3σ3\sigma threshold based on the sensitivity of the observations. The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1, 10\}\ \mathrm{m\,s^{-1}}.

The radial intensity profiles in Figure 6 and Figure E3 show that many of the predicted outer continuum rings are spatially associated with the candidate outer pressure-substructure regions inferred from the gas kinematics. In several systems, including AA Tau, J1615, and V4046 Sgr, the synthetic outer continuum rings are located within or near the shaded radial regions corresponding to the velocity-selected outer gas substructures.

However, this correspondence is not universal. For example, J1852 produces a prominent outer continuum ring despite lacking a formally significant super-Keplerian velocity region in the observational constraints. This illustrates the point made in § 2.2: the super-Keplerian label is a reporting criterion rather than a prerequisite for trapping, and regions that do not meet this criterion can still host substantial dust concentrations in the models.

In several systems, the locations of the 12CO^{12}\mathrm{CO}-based continuum rings also overlap with outer substructures identified in the 13CO^{13}\mathrm{CO} models. Furthermore, both the radial location and prominence of the synthetic outer rings vary between dust evolution configurations, indicating that the detailed dust response depends sensitively on the adopted (α,vfrag)(\alpha, v_{\mathrm{frag}}) combination.

Two sample discs: LkCa15 and J1615

We now examine the two case-study discs selected in § 2.6 in more detail.

Figure 7 shows the dustpy dust surface density distributions for the four explored (α,vfrag)(\alpha, v_{\mathrm{frag}}) configurations. Across both systems, the dust evolution behaviour depends strongly on the balance between fragmentation, radial drift, and dust trapping. The low-α\alpha, high-vfragv_{\mathrm{frag}} configurations generally produce the most efficient grain growth and strongest outer dust concentrations, while higher turbulence and lower fragmentation thresholds yield smaller grain sizes and more diffuse dust distributions.

For J1615, all four reference configurations for both 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} models produce substantial outer dust concentrations that remain detectable in Band 7 continuum images (see Figure 7a and Figure 8a). The 12CO^{12}\mathrm{CO} models predict an extended halo of continuum emission, whereas the 13CO^{13}\mathrm{CO} models produce bright outer rings. In both cases, the predicted emission lies well beyond the observed millimetre disc radius and remains prominent after beam convolution, making J 1615 a clear example where the reference models substantially over-predict the observable outer continuum emission.

Figure 8. RADMC-3D synthetic observations of the dustpy results using the 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} data of J 1615 (top panels) and LkCa 15 (bottom panels). The fluxes are normalised to the peak. The inset shows the observation in Band 7 (Curone et al. 2025), which is also normalised to peak. The white ellipses indicate dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The white dashed lines indicate the 3σ3\sigma threshold based on the sensitivity of the observations. The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1, 10\}\ \mathrm{m\ s^{-1}}. The inner 20 au of LkCa 15 is masked following Sierra et al. (2025).

For LkCa 15 the outcome is configuration-dependent. The model with α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\ s}^{-1} produces a narrow outer dust concentration in the dustpy evolution model (Figure 7b), but the corresponding synthetic Band 7 emission falls below the adopted detection threshold. The model with α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\ s}^{-1} also does not produce an outer ring due to efficient fragmentation and drift of grains.

However, the R90,synR_{90,\mathrm{syn}} for both configurations is substantially more compact than the observed value. In contrast, the model with α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}} and the model with α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}} produce detectable outer continuum rings near the observed outer substructure location.

This behaviour illustrates that the presence of an outer dust concentration in the dust-evolution models does not necessarily guarantee detectable millimetre continuum emission. In the α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}} case, dust trapping does occur: the pressure bump concentrates solids into a narrow outer ring. This ring remains undetectable at Band 7 with the exoALMA resolution and sensitivity. The corresponding continuum feature is however well resolved, spanning ≈75\approx75 au or roughly seven beams, so beam dilution is not the limiting factor. Rather, efficient grain growth concentrates most of the mass into grains that are inefficient millimetre emitters, leaving a continuum feature whose peak excess above the local profile baseline is around 20 times below the 3σ3\sigma threshold of the observations.

We tested whether the large grains might instead be detectable at longer wavelengths by computing synthetic images at 1.3 mm, 1.6 mm, 3 mm, 6 mm and 8.6 mm (ALMA Bands 6 down to 1). For each band and trial beam size, we measured the peak excess of the outer feature above a locally interpolated radial-profile baseline and calculated the rms required for a 3σ3\sigma detection as one-third of this excess. We compared these required sensitivities with equivalent rms values obtained by scaling the Band 7 rms to preserve the same optically thin dust-mass sensitivity, assuming Fν∝κνBν(Tdust)F_\nu\propto\kappa_\nu B_\nu(T_{\mathrm{dust}}), κν=2.3 (ν/230 GHz)0.4 cm2 g−1\kappa_\nu= 2.3\,(\nu/230\,\mathrm{GHz})^{0.4}\,\mathrm{cm^2\,g^{-1}} and Tdust=20 KT_{\mathrm{dust}} = 20\,\mathrm{K}. The required sensitivities remain at least a factor of ≈20\approx20 more stringent than the corresponding scaled values at every wavelength and beam size tested, so no 3σ3\sigma outer emission is recovered at any of these longer wavelengths either. The surface brightness of the outer feature decreases with increasing wavelength, so observing at longer wavelengths provides no net advantage for detecting this feature.

Suppression tests

Figure 9 summarises the outcome of the two suppression tests described in § 2.6, applied to the 12CO^{12}\mathrm{CO}-based models.

Figure 9. Outcome of the suppression tests for the 12^{12}CO-based models, compared with the reference models. Late bump: outer bump introduced after 1 Myr and evolved for a further 0.5 Myr; enhanced α\alpha: αout=10−2\alpha_{\rm out}=10^{-2} beyond RdzR_{\rm dz} (§ 2.6). Symbols as in Figure 5. Numerical values are given in Table D3.

Late introduction of the outer bump

Since long-lived pressure bumps efficiently accumulate drifting solids, delayed formation of the outer gas substructure may reduce the dusty material available for trapping and outer ring formation (Pinilla et al. 2015; Stadler et al. 2022). We first test whether delaying the formation of the outer bump by 1 Myr (Figure 4) suppresses the outer emission.

For LkCa 15, Figure 11 and Figure 9 show that the late bump prescription does not provide a robust mechanism for suppressing outer continuum emission. For the α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}} configuration, the late-bump introduction suppresses outer ring emission. However, for the other configuration that produces an outer ring in the reference model of LkCa 15 (i.e. α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s^{-1}}), the ring is not suppressed. We also note that for α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\,s^{-1}}, a faint halo-like emission becomes detectable, which was not present in the reference model.

Dust evolution and synthetic Band 7 continuum emission for the late-bump LkCa 15 models

Figure 11. Dust evolution and synthetic Band 7 continuum emission for the late-bump LkCa 15 models. The outer gas pressure substructure is introduced after 1 Myr of dust evolution and the system is subsequently evolved for an additional 0.5 Myr. The panels show the resulting DUSTPY dust surface-density distributions together with the beam-convolved RADMC-3D synthetic continuum images. White dashed contours indicate the observational 3σ3\sigma sensitivity threshold.

Hence we find that detectable outer Band 7 emission can still develop rapidly after the bump is introduced. This is because significant solid material remains in the outer disc during the no-bump phase, either retained at large radii by slow radial drift or already present at the bump location. This material is rapidly swept up into the pressure maximum once the bump is introduced, forming a detectable ring within the additional 0.5 Myr of evolution.

In J 1615 (Figure 10 and Figure 9), the late-bump introduction is even less effective. All four configurations retain detectable outer emission after the additional 0.5 Myr, with much larger R90,synR_{90,\mathrm{syn}} (300–420 au) compared with R90,obsR_{90,\mathrm{obs}} (170 au), and peak outer signal-to-noise ratios of 25–69σ\sigma (Figure E19).

Dust evolution and synthetic Band 7 continuum emission for the late-bump J 1615 models

Figure 10. Dust evolution and synthetic Band 7 continuum emission for the late-bump J 1615 models. The outer gas pressure substructure is introduced after 1 Myr of dust evolution and the system is subsequently evolved for an additional 0.5 Myr. The panels show the resulting DUSTPY dust surface-density distributions together with the beam-convolved RADMC-3D synthetic continuum images. White dashed contours indicate the observational 3σ3\sigma sensitivity threshold.

Figure E19. Azimuthally averaged Band 7 radial intensity profiles for the enhanced-α\alpha and late-bump models. The dashed and dash-dotted vertical lines indicate R90,obsR_{90,\mathrm{obs}} and the minimum search radius for outer emission (R90,obs+0.5θbeamR_{90,\mathrm{obs}} + 0.5\theta_{\mathrm{beam}}), respectively, while the dotted horizontal line shows the 3σ3\sigma detection threshold. The legend reports the model classification and the peak signal-to-noise ratio of the outer emission, showing the detectability and relative strength of the predicted outer dust emission.

Overall, these results indicate that delayed bump formation alone is generally insufficient to suppress outer ring formation: the solids are rapidly concentrated once the pressure maximum appears. Suppression by this mechanism would additionally require the outer disc to be depleted of solids at the time the bump forms; this could happen if most of the material has already drifted inward, or if the outer disc is replenished by gas-rich infall carrying very little dust. If outer substructures form after the inner ones, as expected for inside-out planet formation ([40] (Danti et al. 2025)), the outer bump could appear even later than the 1 Myr assumed in our test. A later bump would encounter a smaller outer solid reservoir, since radial drift continues to deplete the outer disc, whereas an earlier bump would encounter a larger one. The outcome is therefore set by the solid mass outside the bump when it forms; for a sufficiently depleted reservoir, no detectable ring forms, as in our AA Tau test.2

Enhanced outer-disc turbulence

Lastly, we explored whether enhanced turbulence exterior to the main millimetre disc could suppress outer dust trapping and reduce the resulting continuum emission, using the step-like turbulence profile with αout=10−2\alpha_{\rm out}=10^{-2} beyond the transition radius RdzR_{\rm dz} defined in § 2.6.

Figure 12, Figure 13 and Figure 9 show that this prescription substantially weakens the outer continuum rings relative to the reference models. In LkCa 15, all four enhanced-turbulence configurations fail to produce detectable outer Band 7 emission beyond the observed continuum disc edge.

Dust evolution and synthetic Band 7 continuum emission for J 1615 models

Figure 12. DUSTPY dust evolution results and synthetic Band 7 continuum emission for the enhanced-turbulence J 1615 models using the 12CO^{12}\mathrm{CO} data. A step-like turbulence profile with αout=10−2\alpha_{\mathrm{out}} = 10^{-2} beyond RdzR_{\mathrm{dz}} is applied. White dashed contours indicate the 3σ3\sigma observational threshold.

Dust evolution and synthetic Band 7 continuum emission for LkCa 15 models

Figure 13. DUSTPY dust evolution results and synthetic Band 7 continuum emission for the enhanced-turbulence LkCa 15 models using the 12CO^{12}\mathrm{CO} data. None of the four enhanced-turbulence configurations produces detectable outer Band 7 emission beyond the observed disc edge.

For J1615 the suppression is only partial: all four enhanced-turbulence configurations retain detectable outer emission, but its prominence is substantially reduced. The peak signal-to-noise ratio of the outer emission falls from 2525–69σ69\sigma in the late-bump models to 88–25σ25\sigma with enhanced turbulence, and R90,synR_{90,\mathrm{syn}} decreases from 347347–391391 au in the reference models to 274274–311311 au (Figure E19; Table D3). The emission is nonetheless classified as halo for every configuration and still extends well beyond R90,obs=170R_{90,\mathrm{obs}}=170 au, so enhanced turbulence weakens but does not remove the predicted outer emission in this disc.

Table D3. Summary of the special-case dust-evolution models (enhanced outer-disc turbulence and late-bump introduction) for LkCa 15, and J1615, and including the enhanced α\alpha of AA Tau. Outer mm emission classifies the morphology of emission detected above 3σ3\sigma beyond R90,obsR_{90,\mathrm{obs}} as defined in § 2.5. No (No outer mm) indicates no emission extending meaningfully beyond R90,obsR_{90,\mathrm{obs}}.

DiscConfigOuter mm emissionR90,obsR_{90,\mathrm{obs}} (au)R90,synR_{90,\mathrm{syn}} (au)Candidate Rbump,gasR_{\mathrm{bump,gas}} region (au)
AA TauEnhanced-α\alpha, α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo139159268–460
AA TauEnhanced-α\alpha, α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo166
AA TauEnhanced-α\alpha, α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Halo167
AA TauEnhanced-α\alpha, α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Halo170
LkCa 15Enhanced-α\alpha, α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}No*15669159–254; 336–395
LkCa 15Enhanced-α\alpha, α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}No*47
LkCa 15Enhanced-α\alpha, α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}No*44
LkCa 15Enhanced-α\alpha, α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}No*41
LkCa 15Late bump, α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo156160159–254; 336–395
LkCa 15Late bump, α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}No*150
LkCa 15Late bump, α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Ring188
LkCa 15Late bump, α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}No*54
J1615Enhanced-α\alpha, α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo170279170–252
J1615Enhanced-α\alpha, α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo274
J1615Enhanced-α\alpha, α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Halo291
J1615Enhanced-α\alpha, α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Halo311
J1615Late bump, α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo170420170–252
J1615Late bump, α=10−4\alpha= 10^{-4}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\,\mathrm{m\,s^{-1}}Halo363
J1615Late bump, α=10−3\alpha= 10^{-3}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Halo300
J1615Late bump, α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\,\mathrm{m\,s^{-1}}Halo314

Table D3.

We applied the same prescription to AA Tau, where the outer emission is confined much closer to the observed disc radius (R90,syn=159R_{90,\mathrm{syn}}=159–170170 au against R90,obs=139R_{90,\mathrm{obs}}=139 au); the extended outer rings present in the reference models are largely removed, although a detectable halo persists in all four configurations (66–11σ11\sigma).

Among the mechanisms explored in this work, enhanced outer-disc turbulence is the most effective at weakening the predicted outer continuum emission, and in LkCa 15 it removes it entirely. It is not, however, a general solution: for J1615 and AA Tau a detectable halo survives in every configuration, and the improvement over the late-bump is mainly quantitative rather than qualitative.

Discussion

Standard dust evolution models frequently produce outer continuum rings

In general the 13^{13}CO-based reconstructions produce broader or higher-contrast pressure maxima in the outer disc than the corresponding 12^{12}CO ones, leading to more efficient dust trapping. Since 13^{13}CO is less optically thick, it should in principle trace deeper, colder disc layers closer to the midplane than 12^{12}CO (e.g. Pinte et al. 2018; Teague et al. 2018), so systematic differences in the inferred gas structure could plausibly also contribute to the stronger outer trapping behaviour recovered in the 13^{13}CO models. However, it is unclear whether the difference in 13^{13}CO and 12^{12}CO is physical. It could also reflect the different radial coverage and SNR of the 13^{13}CO and 12^{12}CO rotation curves. Differences in optical depth, emitting layers, and the radial extent of the reconstructed gas profiles may thus all influence the resulting dust evolution outcomes.

An important consistency check is provided by the inner disc: many of the pressure bumps within the observed continuum disc co-locate well with dust rings in the Band 7 observations (e.g. Curone et al. 2025), indicating that these pressure maxima are efficient, long-lived dust traps: had they been transient they too should fail to retain solids and lack associated rings. The contrast with the outer disc is therefore informative. Rather than these strong outer pressure bumps failing to trap dust, the absence of outer rings point to a physical difference in the inner and outer disc environments, for example: in turbulence level, ionisation, or the lifetime of the outer substructure relative to the local drift timescale of solids. We explored two such possibilities in this work: enhanced outer-disc turbulence (§ 3.3.2), and tested a scenario where the outer bump formed late (§ 3.3.1).

We note that the inner disc profiles are (partially) extrapolated analytically rather than constrained by kinematic data, so the inner disc models should not be interpreted as predictions for the observed inner substructures. In fact, our models do not produce many of the observed rings for this reason.

Can outer dust remain hidden?

Although most models explored in this work produce detectable outer continuum rings, a limited subset of configurations suggests that dust may remain partially hidden at millimetre wavelengths despite the presence of an outer gas pressure substructure. LkCa 15 provides the clearest example (§ 3.2). Its two configurations without detectable outer emission suppress the ring through different mechanisms. In the (α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\ s}^{-1}) case, high turbulence combined with a low fragmentation threshold keeps grains in the fragmentation-limited regime, preventing efficient growth to mm sizes and allowing small grains to drift inward past the bump rather than accumulating there. We therefore focus on the α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s}^{-1} case as the more physically interesting hidden-dust scenario, noting that the combination of high turbulence and low fragmentation velocity is also not

favoured by recent dust evolution constraints ((Luo et al. 2026; Jiang et al. 2024; Tong et al. 2025)).

In the α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\,s}^{-1} case, the trapped mass is concentrated into a radially narrow ring of efficiently grown grains with low sub-mm opacity. The resulting Band 7 emission is therefore dominated by the co-located, much less massive small-grain population, producing a broad, faint continuum ring rather than a bright ring. Similar behaviour has previously been discussed in the context of evolved grain populations and wavelength-dependent dust-trapping efficiencies (Pinilla et al. 2012; Birnstiel 2024). Since the outer feature is resolved and is not recovered at any longer wavelength (§ 3.2), its faintness is set by opacity rather than angular resolution.

We note that the substructures considered here differ from the unresolved-trap scenario invoked for Elias 2-24 by Das et al. (2026): our candidate regions are by construction spatially resolved, spanning more than one beam (§ 2.2). The LkCa 15 result shows that even a resolved gas substructure can produce a dust ring that evades detection at Band 7, because efficient grain growth leaves the emission too faint.

The tentative Band 3 detection of an outer ring at ∼156\sim156 au in LkCa 15 by Sierra et al. (2025) is consistent with this hidden-dust picture: the ring may be present but too grain-growth-dominated to produce a 3σ3\sigma detection at Band 7. A similar situation may apply to HD 34282: Ma et al. [47] present multi-wavelength observations in which the dust disc extends tentatively beyond the Band 7 continuum edge, at radii where our 13CO{}^{13}\mathrm{CO} models independently predict an outer dust ring across all four parameter configurations (Figure 5). We note, however, that the outermost ring reported by Ma et al. [47] exhibits a prominent azimuthal asymmetry indicative of a dusty vortex. That ring may therefore arise from azimuthal vortex trapping rather than from the axisymmetric pressure bump assumed in our 1D models, so the comparison should be regarded as qualitative.

Overall, hidden dust remains possible in specific systems, gas tracer and parameter combinations, and the emerging tentative detections in LkCa 15 and HD 34282 suggest this scenario may be physically possible in at least some exoALMA discs. However, it does not emerge as a robust generic explanation for the observed absence of outer continuum rings across the full sample.

What could suppress the outer rings?

Among the mechanisms explored in this work, enhanced outer-disc turbulence provides the most promising route to suppressing observable outer continuum rings. Although our adopted turbulence profile is phenomenological, it is a plausible scenario where changes of disc ionisation lead to MRI activation in the outer parts of the disc ((Flock et al. 2015; Delage et al. 2022)). If turbulence increases beyond the main millimetre disc, dust trapping at outer pressure maxima may become substantially less efficient, reducing the likelihood that these structures produce detectable continuum rings. In this picture, the absence of outer dust emission would not necessarily imply the absence of pressure substructures, but rather a change in the local dust-evolution environment. Spatially varying turbulence therefore offers a potential explanation for the apparent disconnect between the outer gas pressure structures inferred from exoALMA kinematics and the observed continuum morphology. This can potentially be tested by measuring disc ionisation using a combination of different molecular lines, such as HCO+^{+} and N2_2H+^{+}, as has been done for the disc around TW Hya, where a low-turbulence “dead zone” was inferred extending from the inner edge of the disc out to 50–65 au ((Cleeves et al. 2015)).

The efficiency of this mechanism depends on where the transition radius RdzR_{\mathrm{dz}} is placed. Our step-like prescription places RdzR_{\mathrm{dz}} just interior to R90,obsR_{90,\mathrm{obs}}, which leaves a detectable halo in J 1615 and AA Tau (Figure E19). Placing the transition further inward, for example at Rc/2R_{\mathrm{c}}/2, would expose a larger fraction of the dust-rich solid reservoir to strong turbulent diffusion and fragmentation, and could plausibly suppress the outer emission more effectively, at the cost of also modifying the inner disc where continuum rings are observed. The required magnitude and location of the turbulence enhancement should therefore be regarded as poorly constrained by the present models; a systematic exploration of RdzR_{\mathrm{dz}} is beyond the scope of this work but is a natural next step.

Disc ionisation models generally predict a dead zone with low viscosity (α<10−2\alpha< 10^{-2}) whose outer boundary lies at radii of tens of au ((Flock et al. 2015; Delage et al. 2022)), generally interior to the outer bump radii in our sample (∼150\sim150–400400 au). The outer disc regions explored here may therefore already reside in an MRI-active regime with high viscosity.

The halo emission reported by Das et al. (2026) in DM Tau is consistent with a recently replenished outer disc, supporting the possibility that some outer pressure structures formed at late times. Das et al. (2026) favour late infall over long-lived dust traps for IM Lup and DM Tau on the basis of the small grain sizes inferred there, reserving the unresolved-trap interpretation for Elias 2-24, where cm-sized grains are present in the halo. They further note that dust drift and growth timescales in these regions are shorter than the disc ages, so smooth outer dust reservoirs should not persist under standard radial drift theory. However, our late-bump experiments show that residual solids are rapidly trapped once a pressure maximum appears (§ 3.3.1), so delayed bump formation alone is unlikely to explain the absence of outer continuum rings unless the outer disc is already depleted of solids when the bump forms.

A complementary possibility is that the outer pressure substructures are not static but dynamic, appearing and disappearing over time ((Stadler et al. 2022)). Rather than permanently trapping solids, transient bumps drive episodic variations in the local drift rate; over successive cycles they could enhance the net inward drift, progressively depleting the outer disc of the material needed to build a detectable ring. This scenario would require the innermost pressure maxima to be comparatively long-lived, consistent with their observed continuum rings, while the outer substructures are shorter-lived and leaky. Such a spatial division in bump lifetime could naturally connect to the spatially varying turbulence discussed above, since the change in ionisation or MRI activity that raises the outer turbulence could also influence whether an outer pressure feature persists.

If the outer bumps are carved by embedded planets, the planets themselves could reduce the trapping efficiency. A massive planet stirs the dust in and around its gap, raising the dust scale height and the effective diffusion near the pressure maximum, which makes the trap leakier and the resulting ring fainter ((Binkert et al. 2023)). This effect is not captured by our 1D models, in which the trap leakiness is set only by α\alpha, and testing it would require 3D planet–disc simulations.

Several of the mechanisms discussed above, namely enhanced turbulence, transient bumps and planet-induced stirring, would reduce the efficiency with which the outer bumps retain solids, in addition to any effect on the grain size, which raises the question of how permeable a trap must be for its ring to disappear. To quantify the permeability of the outer bumps, we computed the trap blocking efficiency BB of Houge et al. (2026) for the late-bump models of LkCa 15 and J 1615 (Appendix C). We find no single permeability threshold below which the outer ring disappears: in LkCa 15 the configuration with a detectable ring has an intermediate B≈0.2B \approx0.2, whereas the most blocking trap (B≈0.75B \approx0.75) remains undetectable because the retained mass resides in large grains, and in J 1615 a halo is detected although the bump is nearly fully permeable, reflecting the large outer dust reservoir. The visibility of the outer emission is therefore set by the grain sizes of the trapped dust and the mass of the outer reservoir, rather than by the trap permeability alone.

Alternatively, part of the solids accumulated in the outer bumps may have already formed into planetesimals. Dust concentrated in pressure maxima can reach midplane dust-to-gas ratios high enough to trigger the streaming instability, and Stammler et al. (2019) proposed ongoing planetesimal formation of this nature may explain the moderate optical depths of the DSHARP rings. Three-dimensional simulations showed that planetesimal formation can occur efficiently in dust-trapping rings even in the presence of turbulence (Xu & Bai 2022), and Zhao et al. (2025) found that infall-induced pressure bumps can likewise form planetesimal belts. Mass that is locked in planetesimals is invisible at (sub-)millimetre wavelengths, so efficient planetesimal formation in the outer bumps would lower the predicted ring brightness. In our models the midplane dust-to-gas ratio exceeds unity beyond R90,obsR_{90,\mathrm{obs}}, in regions where the vertically integrated dust-to-gas ratio is also enhanced above its initial value, in 12 out of the 72 reference models, 10 of which have α=10−4\alpha= 10^{-4} and vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\ s^{-1}}, the configuration with the most efficient trapping and the largest and most settled grains. In several of these, the vertically integrated dust-to-gas ratio at the bump approaches or exceeds unity, a regime in which the back-reaction of the dust on the gas which we neglect would become important and planetesimal formation would be expected well before such concentrations are reached. Since we do not include planetesimal formation, the predicted outer emission is an upper estimate in this respect. However only 9 out of 55 models with detectable outer emission reach the conditions, so planetesimal formation could reduce the predicted emission in the most efficient traps but cannot on its own explain the absence of outer rings across the entire sample.

Limitations and future work

A key source of uncertainty in this work arises from the reconstruction of the gas pressure structure itself. The exoALMA pressure-derivative profiles are inferred from deviations in the observed rotation curves and therefore inherit uncertainties associated with finite spatial resolution, beam smearing, noise, and assumptions about the underlying disc geometry and velocity structure. Because our modelling relies on derivatives of the inferred pressure profile, even relatively small uncertainties in the reconstructed velocity field and mid-plane gas pressure derivatives can translate into significant variations in the resulting gas surface density structure (Andrews et al. 2024; Veronesi et al. 2024).

Converting measurements of dln⁡P/dln⁡rd\ln P/d\ln r into gas surface density profiles, Σg(r)\Sigma_{\mathrm{g}}(r), is also intrinsically non-unique. In our reconstruction framework, the shape of Σg(r)\Sigma_{\mathrm{g}}(r) is determined by the inferred pressure-gradient profile, while the total gas mass is fixed using thermochemical mass estimates (Trapman et al. 2025) and the temperature structure adopted from the exoALMA modelling (Galloway-Sprietsma et al. 2025). Consequently, uncertainties in the pressure gradients, temperature structure, and treatment of regions outside the kinematically constrained domain can affect both the contrast and radial extent of the reconstructed pressure maxima. This is particularly important because dust trapping efficiency depends strongly on the morphology of these pressure profiles.

All models start with a uniform dust-to-gas ratio of 0.01 out to the edge of the grid. If the dust disc were initially more compact than the gas disc, the smaller outer dust reservoir would generally reduce the predicted outer emission.

The dust evolution uses the power-law midplane temperature of Galloway-Sprietsma et al. (2025), whereas the synthetic images use the RADMC-3D dust temperature (§ 2.5). For LkCa 15, adopting the DUSTPY temperature does not change the classification of the outer emission. Since the Band 7 intensity of the cold outer disc scales approximately linearly with the dust temperature, the models whose outer emission lies close to the 3σ3\sigma threshold, such as the LkCa 15 late-bump models (3–5σ\sigma; Figure E19), are the most sensitive to this choice, while the strongly detected cases are not. In addition, because the inner disc profiles are partly extrapolated, the shadowing of the outer disc by the modelled inner disc is only approximate.

A further limitation is that observational uncertainties in the pressure-gradient profiles are not explicitly propagated through the gas reconstruction and DUSTPY modelling pipeline. The current models are generated from a single reconstructed Σg(r)\Sigma_{\mathrm{g}}(r) profile rather than an ensemble spanning the allowed uncertainty range. A more complete treatment would require propagating rotation-curve uncertainties through the gas reconstruction, dust evolution, and radiative transfer calculations. While computationally prohibitive with full DUSTPY simulations, future work could employ semi-analytical dust evolution models such as the two-population framework of Birnstiel et al. (2012) or the three-population TriPoD model (Pfeil et al. 2024) to enable ensemble modelling across the allowed pressure-profile uncertainty range.

Another limitation in the present framework is its 1D axisymmetric geometry. Several discs in the sample exhibit azimuthal asymmetries that our models cannot capture: HD 34282 in particular hosts a bright azimuthal arc consistent with a dust-trapping vortex (Ma et al. [47]), and the exoALMA kinematic data reveal arc- and spiral-like velocity residuals in this system (Wölfer et al. 2025; Fukagawa et al. 2026). For any disc where the outer gas substructure is itself azimuthally asymmetric, the 1D assumption will overestimate the radially averaged trapping efficiency, and predicted ring brightnesses should be interpreted accordingly.

Conclusions

In this work, we used the midplane pressure-derivative profiles published by exoALMA from 12CO{}^{12}\mathrm{CO} and 13CO{}^{13}\mathrm{CO} observations (Stadler et al. 2025), together with chemically estimated gas masses (Trapman et al. 2025), to reconstruct gas surface density profiles for nine protoplanetary discs, and modelled their dust evolution and synthetic continuum emission with dustpy and radmc-3d. We explored how different combinations of α\alpha and vfragv_{\mathrm{frag}} affect the formation and observability of outer dust rings associated with gas pressure substructures beyond the observed millimetre continuum discs, tested two mechanisms that could suppress them, and quantified the permeability of the outer bumps.

Our main conclusions are:

  1. The reconstructed outer pressure substructures should generally be observable. In 21 of the 36 12CO^{12}\mathrm{CO}-based models and 34 of the 36 13CO^{13}\mathrm{CO}-based models, efficient outer dust trapping produces emission above 3σ3\sigma beyond the observed continuum disc edge at the resolution and sensitivity of the exoALMA Large Program, in tension with the observations. Trapping also occurs in regions where the velocity-deviation profiles are not formally super-Keplerian, but where variations in the pressure gradient produce traffic-jam pile-ups.

  1. The 13CO^{13}\mathrm{CO}-based reconstructions produce detectable outer emission almost universally, whereas the 12CO^{12}\mathrm{CO}-based ones show greater diversity. Whether this difference is physical or reflects the different radial coverage and signal-to-noise of the two tracers remains unclear.

  1. Only a limited set of models suppresses the outer continuum emission, and many of these also produce continuum discs more compact than observed. In LkCa 15, two configurations show no detectable outer emission: α=10−3\alpha= 10^{-3}, vfrag=1 m s−1v_{\mathrm{frag}} = 1\ \mathrm{m\ s^{-1}}, where the dust does not grow to (sub-)millimetre sizes, and α=10−4\alpha= 10^{-4}, vfrag=10 m s−1v_{\mathrm{frag}} = 10\ \mathrm{m\ s^{-1}}, where large grains form and are concentrated in a narrow ring that remains too faint to detect. The latter case shows that some degree of “hidden dust” is possible in specific cases, although the outer ring does not become detectable at longer wavelengths in our models. Tentative outer continuum emission has recently been reported in LkCa 15 at Band 3 (Sierra et al. 2025) and in HD 34282 ([47]).

  1. We tested two mechanisms for suppressing outer ring formation: a late-time introduction of the outer pressure bump in LkCa 15 and J 1615, and enhanced outer-disc turbulence (αout∼10−2\alpha_{\mathrm{out}} \sim10^{-2}) in LkCa 15, J 1615 and AA Tau. Enhanced turbulence is the most effective: it removes the outer emission in LkCa 15 in all configurations, but a detectable halo persists in J 1615 and AA Tau. Its efficiency depends on the assumed location of the turbulence transition, which we do not explore systematically. Delayed introduction of the outer bump does not robustly suppress ring formation in either system, as residual solids are rapidly swept up once the bump appears.

  1. Using the trap blocking efficiency B\mathcal{B} of Houge et al. (2026), we find no single permeability threshold below which the outer ring disappears. The visibility of the outer emission is therefore set by the grain size distribution of the trapped dust and by the mass of the outer dust reservoir, rather than by the trap permeability alone.

Spatially, as well as temporally (though not measurable) varying turbulence remains the most promising explanation for the missing outer rings. It could be detected observationally by measuring the ionisation structure of the outer discs, for example with HCO+^{+} and N2_2H+^{+}, and by deeper continuum observations sensitive to faint, extended outer emission.

Acknowledgements

LL would like to thank the Science and Technology Facilities Council (STFC) for funding support through a PhD studentship.

PP acknowledges funding from the UK Research and Innovation (UKRI) under the UK Government’s Horizon Europe funding guarantee from ERC (under grant agreement No 101076489).

We acknowledge Myriam Benisty and Fabiola Antonietta Gres for insightful discussions about this project. We also thank the anonymous referee for their helpful suggestions. This paper makes use of the following ALMA data: 2021.1.01123.L. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ. We acknowledge the use of artificial intelligence (Claude Opus 5.5) for the purpose of improving the text’s readability. We also acknowledge the use of Claude to help draft the python code to plot Figure 5 and 9.

Data Availability

The data underlying this article are available from the corresponding author upon request.

Gas surface density reconstruction

The pressure-gradient from Stadler et al. (2025) are derived from deviations of the observed azimuthal velocity from Keplerian rotation. The SNR is typically lower in the inner disc due to beam smearing being more significant at small angular separations and the fractional difference from Keplerian rotation caused by a given pressure perturbation is smaller in the fast-rotating inner disc. Conversely, the outer disc yields more robust kinematic constraints (Stadler et al. 2025). This is because the radial annulus is resolved by more beams at larger radii, reducing the per-annulus velocity uncertainty and because the pressure scale height increases with radius and therefore more readily resolved in the outer disc.

We reconstruct gas surface density profiles Σg(r)\Sigma_{\mathrm{g}}(r) consistent with the observed pressure-derivative constraints and the chemically estimated gas mass in four steps.

Step 1 – Blending small and large-scale profiles

The exoALMA data products from Stadler et al. (2025) provide two independent measurements of dln⁡Pg/dln⁡rd \ln P_{\mathrm{g}}/d \ln r per system: a high-resolution small-scale profile and a lower-resolution large-scale profile. In the overlap region we combine them as

dln⁡Pgdln⁡r∣blend=(1−w)dln⁡Pgdln⁡r∣small+wdln⁡Pgdln⁡r∣large,(8)\left.\frac{d \ln P_{\mathrm{g}}}{d \ln r}\right|_{\mathrm{blend}} = (1-w)\left.\frac{d \ln P_{\mathrm{g}}}{d \ln r}\right|_{\mathrm{small}} + w\left.\frac{d \ln P_{\mathrm{g}}}{d \ln r}\right|_{\mathrm{large}}, \tag*{(8)}

where the logistic weight is

w(r)=11+exp⁡[−(r−Rsw)/Δr].(9)w(r) = \frac{1}{1+\exp[-(r-R_{\mathrm{sw}})/\Delta r]}. \tag*{(9)}

Here RswR_{\mathrm{sw}} is the midpoint of the overlap region and Δr\Delta r is 20% of its width. Beyond the outer limit of the observed data, we extend the profile analytically using the Lynden-Bell & Pringle (1974) pressure-derivative tail,

dln⁡Pgdln⁡r∣tail=−γ−(2−γ)(rRc)2−γ+q2−32,(10)\left.\frac{d \ln P_{\mathrm{g}}}{d \ln r}\right|_{\mathrm{tail}} = -\gamma-(2-\gamma)\left(\frac{r}{R_{\mathrm{c}}}\right)^{2-\gamma}+\frac{q}{2}-\frac{3}{2}, \tag*{(10)}

with γ=1\gamma=1, temperature power-law index qq from Galloway-Sprietsma et al. (2025), and characteristic radius RcR_{\mathrm{c}} from the CO rotation curve modelling of Longarini et al. (2025). The inner profile is similarly extended inward using a Lynden-Bell & Pringle (1974) profile that matches at the innermost observed radius.

Step 2 – Integration

We obtain the pressure shape by trapezoidal integration,

ln⁡Pg(r)=∫r0rdln⁡Pgdln⁡r′ dln⁡r′,(11)\ln P_{\mathrm{g}}(r)=\int_{r_0}^{r}\frac{d\ln P_{\mathrm{g}}}{d\ln r'}\,d\ln r', \tag*{(11)}

Equation A4 is integrated with Pg(r0)≡1P_{\mathrm{g}}(r_0)\equiv1, where r0r_0 is the innermost radius of the observed (small-scale) pressure-derivative profile. This choice is arbitrary and has no effect on the final profile, since Σg(r)\Sigma_{\mathrm{g}}(r) is subsequently rescaled in Step 4 to match the disc mass estimated from chemical tracers (Trapman et al. 2025).

Step 3 – Conversion to surface density.

Assuming a Gaussian vertical density profile, the mid-plane gas density is

ρg=Σg2πH,(12)\rho_{\mathrm{g}}=\frac{\Sigma_{\mathrm{g}}}{\sqrt{2\pi}H}, \tag*{(12)}

so that the midplane pressure is

Pg=ρgcs2=Σgcs22πH,H=csΩ,(13)P_{\mathrm{g}}=\rho_{\mathrm{g}}c_{\mathrm{s}}^2=\frac{\Sigma_{\mathrm{g}}c_{\mathrm{s}}^2}{\sqrt{2\pi}H},\qquad H=\frac{c_{\mathrm{s}}}{\Omega}, \tag*{(13)}

where the isothermal sound speed is

cs=kBTmidμmp,(14)c_{\mathrm{s}}=\sqrt{\frac{k_{\mathrm{B}}T_{\mathrm{mid}}}{\mu m_{\mathrm{p}}}}, \tag*{(14)}

with mean molecular weight μ=2.37\mu=2.37. The midplane temperature Tmid(r)=T0(r/100 au)qT_{\mathrm{mid}}(r)=T_0(r/100\,\mathrm{au})^q is a power-law fit to the disc-specific temperature profiles derived by Galloway-Sprietsma et al. (2025) from thermochemical modelling of the exoALMA CO isotopologue emission, with T0T_0 and qq determined for each system.

Inverting the pressure equation gives

Σg(r)=2πH(r)Pg(r)cs2(r).(15)\Sigma_{\mathrm{g}}(r)=\frac{\sqrt{2\pi}H(r)P_{\mathrm{g}}(r)}{c_{\mathrm{s}}^2(r)}. \tag*{(15)}

Step 4 – Mass normalisation

The profile is rescaled so that

Mgas=2π∫rmin⁡rmax⁡Σg(r) r dr(16)M_{\mathrm{gas}}=2\pi\int_{r_{\min}}^{r_{\max}}\Sigma_{\mathrm{g}}(r)\,r\,dr \tag*{(16)}

matches the chemically estimated disc mass from Trapman et al. (2025). Uncertainties in MgasM_{\mathrm{gas}} and Tmid(r)T_{\mathrm{mid}}(r) are not propagated through the pipeline; their impact is discussed in § 4.4.

The gas masses from Trapman et al. (2025) are estimated from N2_2H+^+ and C18^{18}O line fluxes combined with thermochemical modelling. We adopt these as our fiducial normalisation as they provide a spatially integrated constraint on the total gas content that is independent of the kinematic modelling used to derive the pressure-gradient profiles. A comparison with the dynamically measured masses from Longarini et al. (2025) shows agreement within a factor of ∼3\sim3, with the chemical masses typically a factor of 2.3−1.0+0.72.3^{+0.7}_{-1.0} lower than the kinematic ones (Trapman et al. 2025). This systematic offset propagates directly into the amplitude of Σg(r)\Sigma_{\mathrm{g}}(r) and therefore into the predicted outer dust ring brightness; since our predicted ring fluxes scale directly with the gas normalisation, models using dynamic masses would predict proportionally brighter rings.

We produce two independent Σg(r)\Sigma_{\mathrm{g}}(r) profiles per system using the 13^{13}CO and 12^{12}CO pressure-derivative measurements from Stadler et al. (2025) as separate model inputs. Since 13^{13}CO is less optically thick than 12^{12}CO, it traces deeper into the disc, and should in principle more accurately reflect the thermal and density structure of the midplane where dust trapping occurs (Galloway-Sprietsma et al. 2025). In practice, we find that the 13^{13}CO based reconstructions systematically produce stronger outer pressure contrasts and more efficient dust trapping than the 12^{12}CO ones (see Sect. 3). We discuss this further in § 4.1.

Candidate region selection

We stress that this classification is sensitive to the assumed stellar mass. Whether a region is labelled super- or sub-Keplerian depends on the absolute Keplerian velocity and hence on M⋆M_{\star}, whereas the sign of the δvϕ\delta v_{\phi} gradient does not. The exoALMA stellar masses are constrained to ≈2\approx2–5% (Longarini et al. 2025), which corresponds to a ≈1\approx1–2.5% uncertainty in vKepv_{\mathrm{Kep}} since vKep∝M⋆1/2v_{\mathrm{Kep}}\propto M_{\star}^{1/2}, so a modest revision of M⋆M_{\star} could shift marginal features across the threshold. We emphasise, however, that this criterion is used only to identify and report candidate outer pressure substructures; it does not enter the dust evolution modelling, which is driven directly by the full mid-plane pressure-gradient profile and therefore captures the drift behaviour, including traffic-jam pile-ups, regardless of how a region is labelled. The stellar-mass sensitivity thus affects only which regions we flag as candidate bumps, not the predicted dust distribution.

Adopting the broad super-Keplerian extent as the region boundary, rather than only the highly significant pixels, means the reported rinr_{\rm in}–routr_{\rm out} range reflects the full zone of super-Keplerian rotation and therefore the full spatial extent over which a pressure bump could plausibly trap drifting solids. The resulting zones should be interpreted as candidate gas-pressure substructure regions, not confirmed dust traps (see Appendix Figs. E1 and E2).

Trap blocking efficiency and pebble fluxes

To quantify the permeability of the outer pressure bumps, we adopt the trap blocking efficiency of Houge et al. (2026),

B(t)=1−Mcrossbump(t)Mcrosscontrol(t)(17)\mathcal{B}(t)=1-\frac{M_{\rm cross}^{\rm bump}(t)}{M_{\rm cross}^{\rm control}(t)} \tag*{(17)}

where McrossM_{\rm cross} is the dust mass that has crossed a reference radius rgapr_{\rm gap} since time t0t_0 in a simulation with (bump) and without (control) the pressure perturbation. B=0\mathcal{B}=0 corresponds to a fully permeable perturbation, B=1\mathcal{B}=1 to complete blocking, and B<0\mathcal{B}<0 to inward transport faster than in the unperturbed disc.

The late-bump experiments (§ 2.6) provided us with the required pairs of simulations. After the shared 1 Myr phase on the no-bump gas profile, each model is continued for 0.5 Myr both with the outer bump introduced and, as a control, without it. Both runs therefore start from an identical dust distribution and differ only in the outer pressure perturbation. We set t0=1t_0=1 Myr, when the bump is introduced, and evaluate B\mathcal{B} at t=1.5t=1.5 Myr. We compute McrossM_{\rm cross} as the decrease of the dust mass outside rgapr_{\rm gap},

Mcross(t)=Md(r>rgap,t0)−Md(r>rgap,t),(18)M_{\rm cross}(t)=M_{\rm d}(r>r_{\rm gap},t_0)-M_{\rm d}(r>r_{\rm gap},t), \tag*{(18)}

which includes both advective and diffusive transport, and, being computed from mass differences, is insensitive to the snapshot sampling, and the pebble flux as Fpeb(r,t)=−dMd(r′>r,t)/dtF_{\rm peb}(r,t)=-\mathrm{d}M_{\rm d}(r'>r,t)/\mathrm{d}t.

The perturbation introduced in the late-bump test consists of a pressure depression interior to the bump, followed by the bump itself. We thus place rgapr_{\rm gap} at the minimum of Pbump/PcontrolP_{\rm bump}/P_{\rm control} interior to the bump, at 85 au for LkCa 15 and 60 au for J 1615, analogous to measuring the flux through the gap of a planet-induced trap. In order to test the sensitivity to this choice, in Table C1 we list B\mathcal{B} for a range of rgapr_{\rm gap}. For the LkCa 15 case, B\mathcal{B} is robust to the choice of rgapr_{\rm gap} for the two configurations with vfrag=10v_{\rm frag}=10 m s−1^{-1}, but varies between ≈0.05\approx0.05 and 0.4 for α=10−4\alpha=10^{-4}, vfrag=1v_{\rm frag}=1 m s−1^{-1}; in J 1615, ∣B∣≲0.15\lvert\mathcal{B}\rvert\lesssim0.15 for all configurations at and inside the depression. Immediately after the bump is introduced, B(t)\mathcal{B}(t) shows short transients (including B<0\mathcal{B}<0) as the steepened pressure gradient in the depression briefly accelerates the inward drift; the values in Table C1 refer to t=1.5t=1.5 Myr.

Table C1. Trap blocking efficiency BB at t=1.5t = 1.5 Myr (Eq. (17)) for the 12CO^{12}\mathrm{CO}-based late-bump models, for several reference radii rgapr_{\mathrm{gap}}. Values at the adopted rgapr_{\mathrm{gap}} (pressure depression interior to the bump) are in bold. MdM_{\mathrm{d}} is the dust mass outside the adopted rgapr_{\mathrm{gap}} when the bump is introduced (1 Myr). The last column gives the late-bump classification of the outer Band 7 emission (Table D3).

LkCa 15
(α,vfrag/m s−1)(\alpha, v_{\mathrm{frag}}/\mathrm{m\,s}^{-1})MdM_{\mathrm{d}} (M⊕M_{\oplus})70 au85 au100 au120 auOutcome
(10−3, 1)(10^{-3},\,1)470.01−0.10−0.050.07Halo
(10−4, 1)(10^{-4},\,1)430.050.060.250.40No outer mm
(10−3, 10)(10^{-3},\,10)260.200.220.230.24Ring
(10−4, 10)(10^{-4},\,10)170.710.750.810.87No outer mm
J 1615
(α,vfrag/m s−1)(\alpha, v_{\mathrm{frag}}/\mathrm{m\,s}^{-1})MdM_{\mathrm{d}} (M⊕M_{\oplus})45 au60 au80 au100 auOutcome
(10−3, 1)(10^{-3},\,1)5510.15−0.04−0.20−0.20Halo
(10−4, 1)(10^{-4},\,1)5390.020.120.380.63Halo
(10−3, 10)(10^{-3},\,10)3060.040.040.040.03Halo
(10−4, 10)(10^{-4},\,10)3730.020.010.020.01Halo

Table C1.

Table C1 compares B\mathcal{B} with our classification of the late-bump models. In LkCa 15, B\mathcal{B} ranges from ≈0\approx0 to ≈0.75\approx0.75 but does not map monotonically onto the predicted continuum morphology: the configuration with a detectable outer ring has B≈0.2\mathcal{B}\approx0.2, whereas the two configurations without detectable outer emission have B≈0.06\mathcal{B}\approx0.06 and ≈0.75\approx0.75. The most blocking trap (α=10−4\alpha=10^{-4}, vfrag=10v_{\rm frag}=10 m s−1^{-1}) remains undetectable because the retained mass resides in large grains with low Band 7 opacity (Section 4.2). In J 1615 the bump is nearly fully permeable (∣B∣≲0.15\lvert\mathcal{B}\rvert\lesssim0.15) in all configurations, yet every model produces a detectable halo, which therefore is indicative of the large outer dust reservoir (≳300 M⊕\gtrsim300\,M_{\oplus} beyond 60 au) rather than efficient trapping. These values are not directly comparable to those of Houge et al. (2026), who quote B\mathcal{B} after 5 Myr for planet-induced traps at 5–40 au. Moreover, we only compute B\mathcal{B} for the late-bump models, since for the reference models we had no bump-free counterpart. Figure C1 shows the pebble flux at rgapr_{\rm gap} and at 1 au over the full 1.5 Myr. At 1 au, the bump and control runs differ by less than a factor of ∼1.5\sim1.5 within the 0.5 Myr after the bump forms, reflecting the long drift time from the outer disc to the inner disc. Therefore, on this timescale the outer bumps do not measurably change the delivery of pebbles to the inner disc probed by JWST.

Figure C1. Pebble flux through rgapr_{\mathrm{gap}} (top; 85 au for LkCa 15, 60 au for J 1615) and through 1 au (bottom) for the 12CO^{12}\mathrm{CO}-based late-bump models. Solid lines show the shared no-bump phase (0–1 Myr) followed by the run with the outer bump introduced at 1 Myr (dotted vertical line); dashed lines show the control run continued without the bump. Negative (outward) fluxes are not shown on the logarithmic axis. The apparent discontinuity at 1 Myr is a sampling effect: the first phase snapshots are separated by up to 0.5 Myr, so the fluxes shown there are averages over long intervals, whereas the flux rises steeply towards 1 Myr. Interval-averaged fluxes are continuous across the restart.

Model outcome tables

Tables D1–D3 list, for every model, the classification of the outer Band 7 emission (Ring, Halo or No outer mm; § 2.5), the synthetic disc radius R90,synR_{90,\rm syn} compared with the observed R90,obsR_{90,\rm obs}, and the candidate outer pressure-substructure regions inferred from the kinematics (§ 2.2). Tables D1 and D2 give the reference models for the 12CO{}^{12}\mathrm{CO}- and 13CO{}^{13}\mathrm{CO}-based reconstructions, summarised in Figure 5, and Table D3 gives the suppression tests, summarised in Figure 9.

Additional Targets and Supporting Figures

We provide additional results for the exoALMA systems not shown in the main text. For each system we include the dust evolution models and the resulting Band 7 RADCMD3D images for the four (α,vfrag)(\alpha,v_{\rm frag}) combinations.

Figure E2. As Figure E1, but derived from the 13CO^{13}\mathrm{CO} rotation curves. Candidate outer pressure-substructure regions satisfying the significance threshold (Δvϕ>1\Delta v_{\phi} > 1 per cent, S>3S > 3) are identified in only 2 of the 9 systems (DM Tau and HD 34282). In some of the remaining systems, velocity deviations are present but fall below the adopted significance threshold, and are therefore not selected as candidate regions by our criterion. Note that the super-/sub-Keplerian classification depends on the assumed stellar mass, regions near the threshold should be regarded as marginal.

Figure E1. Radial profiles of the fractional azimuthal velocity deviation δvϕ/vKep\delta v_{\phi}/v_{\mathrm{Kep}} [%] for the nine exoALMA discs modelled in this work, derived from the 12CO^{12}\mathrm{CO} rotation curves of Stadler et al. (2025). Error bars show the 1σ1\sigma uncertainties on the measured velocity residuals. The dash-dotted vertical line marks R90,obsR_{90,\mathrm{obs}}, the radius enclosing 90 per cent of the Band 7 continuum flux (Curone et al. 2025). The red scale bar in the lower-left corner of each panel indicates the effective beam size θbeam\theta_{\mathrm{beam}} in au. The open circle marks the peak velocity deviation within each candidate region. Grey shaded regions indicate the candidate outer pressure-substructure zones identified as described in § 2.2; candidate regions are identified in 8 of the 9 systems from the 12CO^{12}\mathrm{CO} data. Note that the super-/sub-Keplerian classification depends on the assumed stellar mass, regions near the threshold should be regarded as marginal.

RADMC synthetic observations of the dustpy 12CO and 13CO models

Figure E5. RADMC synthetic observations of the dustpy 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} models. The fluxes are normalised to the peak. The inset shows the observation in B7 (Curone et al. 2025). The white ellipses indicate dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The white dashed lines indicate the 3σ3\sigma threshold based on the sensitivity of the observations. The four columns represent different dust evolution model configurations spanning α∈{10−4, 10−3}\alpha\in\{10^{-4},\,10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\ \mathrm{m\,s^{-1}}.

Figure E6. Dust density distribution after 1 Myr for AA Tau, using the exoALMA gas surface density reconstructions derived from the 12CO^{12}\mathrm{CO} (top) and 13CO^{13}\mathrm{CO} (bottom) observations. White and red dashed curves indicate the drift and fragmentation barriers, respectively. Vertical magenta dashed lines mark dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1, 10\}\ \mathrm{m\ s^{-1}}.

RADMC synthetic observations of the dustpy models

Figure E7. RADMC synthetic observations of the dustpy 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} models. The fluxes are normalised to the peak. The inset shows the observation in B7 (Curone et al. 2025). The white ellipses indicate dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The white dashed lines indicate the 3σ3\sigma threshold based on the sensitivity of the observations. The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1, 10\}\ \mathrm{m\ s^{-1}}.

Figure E8. dustpy dust evolution results and synthetic Band 7 continuum emission for the enhanced-turbulence AA Tau models using the 12^{12}CO data. A step-like turbulence profile with αout=10−2\alpha_{\mathrm{out}} = 10^{-2} beyond RdzR_{\mathrm{dz}} is applied. White dashed contours indicate the 3σ3\sigma observational threshold.

Dust density distribution after 1 Myr for J 1852

Figure E11. Dust density distribution after 1 Myr for J 1852, using the exoALMA gas surface density reconstructions derived from the 12CO^{12}\mathrm{CO} (top) and 13CO^{13}\mathrm{CO} (bottom) observations. White and red dashed curves indicate the drift and fragmentation barriers, respectively. Vertical magenta dashed lines mark dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1, 10\}\ \mathrm{m\,s^{-1}}.

Figure E12. RADMC synthetic observations of the dustpy 12CO{}^{12}\mathrm{CO} and 13CO{}^{13}\mathrm{CO} models. The fluxes are normalised to the peak. The inset shows the observation in B7 (Curone et al. 2025). The white ellipses indicate dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The white dashed lines indicate the 3σ3\sigma threshold based on the sensitivity of the observations. The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\ \mathrm{m\,s^{-1}}.

Dust density distribution after 1 Myr for HD 34282

Figure E13. Dust density distribution after 1 Myr for HD 34282, using the exoALMA gas surface density reconstructions derived from the 12CO{}^{12}\mathrm{CO} (top) and 13CO{}^{13}\mathrm{CO} (bottom) observations. White and red dashed curves indicate the drift and fragmentation barriers, respectively. Vertical magenta dashed lines mark dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\ \mathrm{m\,s^{-1}}.

Figure E14. RADMC synthetic observations of the dustpy 12CO^{12}\mathrm{CO} and 13CO^{13}\mathrm{CO} models. The fluxes are normalised to the peak. The inset shows the observation in B7 (Curone et al. 2025). The white ellipses indicate dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The white dashed lines indicate the 3σ3\sigma threshold based on the sensitivity of the observations. The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1, 10\}\,\mathrm{m\,s^{-1}}.

Dust density distribution after 1 Myr for SY Cha

Figure E15. Dust density distribution after 1 Myr for SY Cha, using the exoALMA gas surface density reconstructions derived from the 12CO^{12}\mathrm{CO} (top) and 13CO^{13}\mathrm{CO} (bottom) observations. White and red dashed curves indicate the drift and fragmentation barriers, respectively. Vertical magenta dashed lines mark dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4}, 10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1, 10\}\,\mathrm{m\,s^{-1}}.

Figure E16. RADMC synthetic observations of the dustpy 12CO{}^{12}\mathrm{CO} and 13CO{}^{13}\mathrm{CO} models. The fluxes are normalised to the peak. The inset shows the observation in B7 (Curone et al. 2025). The white ellipses indicate dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The white dashed lines indicate the 3σ3\sigma threshold based on the sensitivity of the observations. The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4},10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\ \mathrm{m\,s^{-1}}.

Dust density distribution after 1 Myr for V 4046

Figure E17. Dust density distribution after 1 Myr for V 4046, using the exoALMA gas surface density reconstructions derived from the 12CO{}^{12}\mathrm{CO} (top) and 13CO{}^{13}\mathrm{CO} (bottom) observations. White and red dashed curves indicate the drift and fragmentation barriers, respectively. Vertical magenta dashed lines mark dust rings identified in the Band 7 continuum observations (Curone et al. 2025). The four columns represent different dust evolution model configurations spanning α∈{10−4,10−3}\alpha\in\{10^{-4},10^{-3}\} and vfrag∈{1,10} m s−1v_{\mathrm{frag}} \in\{1,10\}\ \mathrm{m\,s^{-1}}.

References

  1. [1]ALMA Partnership et al., 2015, ApJ, 808, L3
  2. [40]Lau T. C. H., Birnstiel T., Drążkowska J., Stammler S. M., 2024, A&A, 688, A22
  3. [47]Ma X., et al., 2026, The Astrophysical Journal, 1001, 124
  4. [81]van der Marel N., Pinilla P., 2025, Dust Evolution in Protoplanetary Disks, Springer Nature Switzerland, Cham, pp 1–41, doi:10.1007/978-3-319-30648-3_201-1, https://doi.org/10.1007/978-3-319-30648-3_201-1
  5. [82]van der Marel N., Dong R., di Francesco J., Williams J. P., Tobin J., 2019, ApJ, 872, 112

Paper details

vlnc:W_n22eujeu
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Claude Opus 5.5

View statement
Looking for related papers…
Contents