Context. The circumstellar envelopes of AGB stars are sites of rich molecular chemistry. The present understanding of C-rich AGB chemistry largely relies on observations of the archetypal carbon star IRC+10 216. Current molecular abundance estimates for carbon stars are based either on single-dish spectra sampling a range of excitation conditions, or on interferometric mapping of a few lines. Aims. We aim to estimate the circumstellar abundances of SiO, SiS, and their most abundant isotopologues (29SiO, 30SiO, 29SiS, 30SiS, and Si34S) for a sample of five carbon stars. This study compares molecular abundances across the sources, tests chemical modelling predictions, and examines whether IRC+10 216 is representative of the broader carbon star population. Methods. We derived molecular abundances using detailed 1D non-local thermodynamic equilibrium (non-LTE) radiative transfer (RT) modelling, constrained by both morphological and excitation information obtained from spatially resolved ALMA maps and single-dish observations. We further compared the derived abundances to chemical modelling results. Results. We obtain good fits to the SiO and SiS line profiles, and derived well-constrained abundance profiles and reliable isotopic ratios for all sources except AFGL 3068. While the SiS peak abundances are very similar across the sample (2.0 × 10−6–4.7 × 10−6), we find that the SiO peak abundances of the rest of the stars are a factor of ~5 larger than that of IRC +10 216. The e-folding radii (Re) are in the range 1.3 × 1016 − 7.0×1016cm for SiO and 6.0 × 1015 − 1.0×1017 cm for SiS. The Re increases with gas density for both SiO and SiS. Our RT models cannot simultaneously fit the low- and high-J SiO lines of IRC+10216. Chemical models reproduce the derived SiO abundance profiles well, while over-predicting the SiS Re values. Conclusions. Our models highlight the necessity of having spatially resolved observations across a broad range of excitation conditions to robustly constrain molecular abundance profiles, while also making evident the limitations inherent in 1D RT modelling using simplified (circum)stellar models. We find that the currently assumed SiS photodissociation rate in chemical models is underestimated.

Context. Mass loss in oxygen-rich asymptotic giant branch (AGB) stars remains a longstanding puzzle, as the dust species detected around these stars appear too transparent to drive winds through the absorption of radiation alone. The current paradigm consists of outflows driven by photon scattering and requires relatively large grains (similar to 0.3 mu m). Whether the necessary number of grains with the required scattering properties exist around AGB stars remains to be determined empirically. Aims. We test whether the dust grains observed around the oxygen-rich AGB star R Doradus can drive its stellar wind by combining, for the first time, polarimetric constraints with elemental abundance limits and force balance calculations. We examine Fe-free silicates (MgSiO3), aluminium oxide (Al2O3), and Fe-bearing silicates (MgFeSiO4) to determine whether any dust species can generate sufficient radiative pressure under physically realistic conditions. Methods. We analysed high-angular-resolution polarimetric observations obtained with SPHERE/ZIMPOL at the Very Large Telescope (VLT) and modelled the circumstellar dust using the radiative transfer code RADMC-3D. Dust optical properties were computed using Optool for both Mie and the distribution of hollow spheres (DHS) scattering theories. By systematically exploring a six-dimensional parameter space, we derived constraints on dust grain sizes, density profiles, and wavelength-dependent stellar radii. For models that successfully fit the observations, we analysed the results taking into consideration recent models for the gas density distribution around R Dor, and applied a multi-criteria zone analysis incorporating gas-depletion constraints and radiation pressure thresholds to assess dust-driven wind viability. Results. We find sub-micron MgSiO3 and Al2O3 grains (up to 0.1 mu m) regardless of scattering theory considered, and a two-layer dust envelope with steep density profiles (r(-3.4) to r(-4.1)). Despite matching observed scattered-light patterns, these grains generate insufficient radiative force under physically realistic gas-to-dust mass ratios, even when assuming complete elemental depletion. Silicates containing Fe could theoretically provide adequate force, but would sublimate in critical acceleration regions and require implausibly high silicon-depletion levels. Conclusions. Our findings for R Doradus show insufficient radiation pressure from scattering on grains, suggesting that dust alone cannot drive the wind in this star and that additional mechanisms may be required.

Mass loss in oxygen-rich asymptotic giant branch (AGB) stars remains poorly understood, as the dust detected around them appears too transparent to drive winds through absorption alone. The current paradigm invokes outflows driven by photon scattering on relatively large grains (∼0.3 μm), but whether such grains exist in sufficient quantities remains uncertain. We test whether the dust around the oxygen-rich AGB star R Doradus can drive its wind by combining polarimetric constraints, elemental abundance limits, and force-balance calculations. We examine Fe-free silicates (MgSiO_3), alumina (Al_2O_3), and Fe-bearing silicates (MgFeSiO_4) to evaluate whether any species can produce enough radiative pressure under realistic conditions. Using high-angular-resolution polarimetric observations from SPHERE/ZIMPOL at the VLT, we model the circumstellar dust with RADMC-3D and explore a broad parameter space in grain size, density structure, and wavelength-dependent stellar radius. For models consistent with the observations, we assess wind feasibility using updated gas-density profiles, elemental depletion constraints, and radiation-pressure thresholds. Although several dust configurations reproduce the observed scattering patterns, none generate sufficient radiative force at realistic gas-to-dust ratios, even under maximal elemental depletion. Our results for R Doradus indicate that photon scattering on dust cannot by itself launch the wind, implying that additional mechanisms must contribute.

The transport of energy through convection is important during many stages of stellar evolution1,2, and is best studied in our Sun3 or giant evolved stars4. Features that are attributed to convection are found on the surface of massive red supergiant stars5-8. Also for lower-mass evolved stars, indications of convection are found9-13, but convective timescales and sizes remain poorly constrained. Models indicate that convective motions are crucial to produce strong winds that return the products of stellar nucleosynthesis into the interstellar medium14. Here we report a series of reconstructed interferometric images of the surface of the evolved giant star R Doradus. The images reveal a stellar disk with prominent small-scale features that provide the structure and motions of convection on the stellar surface. We find that the dominant structure size of the features on the stellar disk is 0.72 ± 0.05 astronomical units. We measure the velocity of the surface motions to vary between -18 and +20 km s-1, which means that the convective timescale is approximately one month. This indicates a possible difference between the convection properties of low-mass and high-mass evolved stars.

AGB stars are major contributors to the chemical enrichment of the ISM through nucleosynthesis and extensive mass loss. Most of our current knowledge of AGB atmospheric and circumstellar chemistry, in particular in a C-rich environment, is based on observations of the carbon star IRC+10216. We aim to obtain a more generalised understanding of the chemistry in C-rich AGB CSEs by studying a sample of three carbon stars, IRAS15194-5115, IRAS15082-4808, and IRAS07454-7112, and test the archetypal status often attributed to IRC+10216. We performed spatially resolved, unbiased spectral surveys in ALMA Band 3. We identify a total of 132 rotational transitions from 49 molecular species. There are two main morphologies of the brightness distributions: centrally-peaked (e.g. HCN) and shell-like (e.g. C$_2$H). We estimated the sizes of the molecular emitting regions using azimuthally-averaged radial profiles of the line brightness distributions, and derived abundance estimates. Of the shell distributions, the cyanopolyynes peak at slightly smaller radii than the hydrocarbons, and CN and HNC show the most extended emission. The emitting regions for each species are the smallest for IRAS07454-7112. We find that, within the uncertainties of the analysis, the three stars present similar abundances for most species, also compared to IRC+10216. We find that SiO is more abundant in our three stars compared to IRC+10216. Our estimated isotopic ratios match well the literature values for the sources. The observed circumstellar chemistry appears very similar across our sample and compared to that of IRC+10216, both in terms of the relative location of the emitting regions and molecular abundances. This implies that, to a first approximation, the chemical models tailored to IRC+10216 are able to reproduce the observed chemistry in C-rich envelopes across roughly an order of magnitude in wind density.

The mass loss experienced on the asymptotic giant branch (AGB) at the end of the lives of low- and intermediate-mass stars is widely accepted to rely on radiation pressure acting on dust grains formed in the extended AGB atmospheres. The interaction of convection, stellar pulsation, and heating and cooling processes cause the density, velocity and temperature distributions in the inner regions of the envelope to be complex, making the dust-formation process difficult to calculate. Hence, characterising the extended atmospheres and inner outflow empirically is paramount to advance our understanding of the dust-formation and wind-driving processes. To this end, we observe the AGB star R Dor using ALMA and modelled the $^{12}$CO $v=0, J=2-1$, $v=1, J=2-1$ and $3-2$ and $^{13}$CO $v=0, J=3-2$ lines using the 3D radiative transfer code LIME up to a distance of $\sim 4$ times the radius of the star at sub-mm wavelengths. We find a complex velocity field with structure down to scales at least equal to the resolution of the observations. The observed maps are well reproduced assuming spherical symmetry for the gas temperature and density distributions. We find the radial profiles of these two quantities to be very steep close to the star and shallower for radii larger than $\sim 1.6$ times the stellar sub-mm radius. This change is consistent with the transition between extended atmosphere and outflow. We constrain the standard deviation of the stochastic velocity distribution in the large-scale outflow to be $\lesssim 0.4$ km/s. We observe two emission blobs in the CO $v=0, J=2-1$ line and find their gas densities and radial velocities to be substantially larger than those of the surrounding gas. Monitoring the evolution of these blobs will lead to a better understanding of the role of these structures in the mass-loss process of R Dor.

Aims. We aim to resolve the spatial and kinematic sub-structures in five detached-shell sources to provide detailed constraints for hydrodynamic models that describe the formation and evolution of the shells. Methods. We use observations of the 12 CO (1-0) emission towards five carbon-AGB stars with ALMA. The data have angular resolutions of 0.3 arcsec to 1arcsec and a velocity resolution of 0.3 km/s . This enables us to quantify spatial and kinematic structures in the shells. Results. The observed emission is separated into two distinct components: a more coherent, bright outer shell and a more filamentary, fainter inner shell. The kinematic information shows that the inner sub-shells move at a higher velocity relative to the outer sub-shells. The observed sub-structures confirm the predictions from hydrodynamical models. However, the models do not predict a double-shell structure, and the CO emission likely only traces the inner and outer edges of the shell, implying a lack of CO in the middle layers of the detached shell. Previous estimates of the masses and temperatures are consistent with originating mainly from the brighter subshell, but the total shell masses are likely lower limits. Conclusions. The observed spatial and kinematical splittings of the shells appear consistent with results from hydrodynamical models, provided the CO emission does not trace the H2 density distribution in the shell but rather traces the edges of the shells. It is therefore not possible to constrain the total shell mass based on the CO observations alone. Complementary observations of, e.g., CI as a dissociation product of CO would be necessary to understand the distribution of CO compared to H2.

Aims. Our goal is to study the long-term mass-loss rate characteristics of asymptotic giant branch (AGB) stars through wind-wind and wind-interstellar medium interaction. Methods. Far-ultraviolet (FUV) images from the GALEX survey are used to investigate extended UV emission associated with AGB stars. Results. FUV emission was found towards eight objects. The emission displays different shapes and sizes; interaction regions were identified, often with infrared counterparts, but no equivalent near-ultraviolet (NUV) emission was found in most cases. Conclusions. The FUV emission is likely attributed to shock-excited molecular hydrogen, considering the lack of NUV emission and the large space velocities of the objects, and makes it possible to trace old structures that are too faint to be observed, for instance, in the infrared.

Aims. In this paper, we aim to constrain the dust mass and grain sizes in the interaction regions between the stellar winds and the interstellar medium (ISM) around asymptotic giant branch (AGB) stars. By describing the dust in these regions, we aim to shed light on the role of evolved low-mass stars in the origin of dust in galaxies. Methods. We use images in the far-infrared (FIR) at 70 and 160 mu m to derive dust temperatures and dust masses in the wind-ISM interaction regions around a sample of carbon-rich and oxygen-rich AGB stars. The dust temperature and mass are determined in two ways: first, directly from the data using the ratio of the measured fluxes and assuming opacities for dust with a constant grain size of 0.1 mu m, and then using three-dimensional dust-radiative transfer models spatially constrained by the observations. Each of the radiative transfer models contains one constant grain size, which is varied between 0.01 and 5.0 mu m. Results. We find that the observed dust mass in the wind-ISM interaction regions is consistent with mass accumulated from the stellar winds. For the carbon-rich sources, adding the spatial constraints in the radiative transfer models results in preferentially larger grain sizes (approximate to 2 mu m). For the oxygen-rich sources, the spatial constraints result in overly high temperatures in the models, making it impossible to fit the observed FIR ratio irrespective of the grain size used, indicating a more complex interplay of grain properties and the stellar radiation field. Conclusions. Our results have implications for how likely it is for the grains to survive the transition into the ISM, and the properties of dust particles that later act as seeds for grain growth in the ISM. However, our results for the oxygen-rich sources show that the derivation of dust properties is not straight forward, requiring more complex modelling.

We aim to constrain the dust mass and grain sizes in the interaction regions between the stellar winds and the ISM around asymptotic giant branch stars. By describing the dust in these regions, we aim to shed light on the role of low mass evolved stars in the origin of dust in galaxies. We use images in the far-infrared at 70 micron and 160 micron to derive dust temperatures and dust masses in the wind-ISM interaction regions around a sample of carbon-rich and oxygen-rich asymptotic giant branch (AGB) stars. The dust temperature and mass are determined in two ways. First directly from the data using the ratio of the measured fluxes and assuming opacities for dust with a constant grain size of 0.1 micron. We then perform 3D dust-radiative transfer models spatially constrained by the observations to consistently calculate the temperature and mass. For the radiative transfer models each model contains one constant grain size, which is varied between 0.01 micron to 5.0 micron. We find that the observed dust mass in the wind-ISM interaction regions is consistent with mass accumulated from the stellar winds. For the carbon-rich sources adding the spatial constraints in the radiative transfer models results in preferentially larger grain sizes (approx. 2 micron). For the oxygen-rich sources the spatial constraints result in too high temperatures in the models, making it impossible to fit the observed far-infrared ratio irrespective of the grain size used, indicating a more complex interplay of grain properties and the stellar radiation field. The results have implications for how likely it is for the grains to survive the transition into the ISM, and the properties of dust particles that later act as seeds for grain growth in the ISM. However, the results for the oxygen-rich sources show that the derivation of dust properties is not straight forward, requiring more complex modelling

Context. We report the detection of two lines at millimetre wavelengths towards the immediate surroundings of the post-giant and most likely post-common-envelope star HD 101584 using high-angular-resolution ALMA observations. The circumstellar environment of this object is rich in different molecular species, but we find no viable identifications in terms of molecular lines. Aims. We aim to determine whether or not these lines can be attributed to the Rydberg transitions – X30 α and X26 α – of neutral atoms of elements heavier than carbon. Methods. A simple model in strict local thermodynamic equilibrium for a warm-gas environment of the moderate-temperature star ( T eff ≈ 8500 K) was constructed to corroborate our findings. A geometrically thin, disc-like geometry seen face-on was chosen and a distance of 1 kpc. Results. The observed flux densities of the lines and the continuum at 232 and 354 GHz can be reproduced using ≈10 −3 M ⊙ of gas at a temperature of ≈2800 K and a hydrogen density of ≈10 12 cm −3 , assuming solar abundances for the elements. The gas lies within a distance of about 5 au from the star (assuming a distance of 1 kpc). The ionisation fraction is low, ≈3 × 10 −5 . The origin of such a region is not clear, but it may be related to a common-envelope-evolution phase. With these conditions, the line emissions are dominated by Rydberg transitions within the stable isotopes of Mg. A turbulent velocity field in the range 5.5–7.5 km s −1 is required to fit the Gaussian line shapes. An upper limit to the average magnetic field in the line-emitting region of 1 G is set using the Zeeman effect in these lines. Conclusions. We speculate that Rydberg transitions of heavy elements may be an interesting probe for the close-in environments of other moderate-temperature objects like AGB stars, red supergiants, yellow hypergiants, and binaries of various types.

Aims. We aim to constrain the sizes of, and investigate deviations from spherical symmetry in, the CO circumstellar envelopes (CSEs) of 16 S-type stars, along with an additional 7 and 4 CSEs of C-type and M-type AGB stars, respectively. Methods. We map the emission from the CO J = 2-1 and 3-2 lines observed with the Atacama Compact Array (ACA) and its total power (TP) antennas, and fit with a Gaussian distribution in the uv- and image planes for ACA-only and TP observations, respectively. The major axis of the fitted Gaussian for the CO(2-1) line data gives a first estimate of the size of the CO-line-emitting CSE. We investigate possible signs of deviation from spherical symmetry by analysing the line profiles and the minor-to-major axis ratio obtained from visibility fitting, and by investigating the deconvolved images. Results. The sizes of the CO-line-emitting CSEs of low-mass-loss-rate (low-MLR) S-type stars fall between the sizes of the CSEs of Cstars, which are larger, and those of M-stars, which are smaller, as expected because of the differences in their respective CO abundances and the dependence of the photodissociation rate on this quantity. The sizes of the low-MLR S-type stars show no dependence on circumstellar density, as measured by the ratio of the MLR to terminal outflow velocity, irrespective of variability type. The density dependence steepens for S-stars with higher MLRs. While the CO(2-1) brightness distribution size of the low-density S-stars is in general smaller than the predicted photodissociation radius (assuming the standard interstellar radiation field), the measured size of a few of the high-density sources is of the same order as the expected photodissociation radius. Furthermore, our results show that the CO CSEs of most of the S-stars in our sample are consistent with a spherically symmetric and smooth outflow. For some of the sources, clear and prominent asymmetric features are observed which are indicative of intrinsic circumstellar anisotropy. Conclusions. As the majority of the S-type CSEs of the stars in our sample are consistent with a spherical geometry, the CO envelope sizes obtained in this paper will be used to constrain detailed radiative transfer modelling to directly determine more accurate MLR estimates for the stars in our sample. For several of our sources that present signs of deviation from spherical symmetry, further high-resolution observations would be necessary to investigate the nature of, and the physical processes behind, these asymmetrical structures. This will provide further insight into the mass-loss process and its related chemistry in S-type AGB stars.

Context. This study is a follow up to the previous analysis of lower-angular resolution data in which the kinematics and structure of the circumstellar envelope (CSE) around the S-type asymptotic giant branch (AGB) star π1 Gruis were investigated. The AGB star has a known companion (at a separation of ~400 AU) that cannot explain the strong deviations from spherical symmetry of the CSE. Recently, hydrodynamic simulations of mass transfer in closer binary systems have successfully reproduced the spiral-shaped CSEs found around a handful of sources. There is growing evidence for an even closer, undetected companion complicating the case of π1 Gruis further. Aims. The improved spatial resolution allows for the investigation of the complex circumstellar morphology and the search for imprints on the CSE of the third component. Methods. We have observed the 12CO J = 3–2 line emission from π1 Gruis using both the compact and extended array of Atacama Large Millimeter/submillimeter Array (ALMA). The interferometric data have furthermore been combined with data from the ALMA total power array. The imaged brightness distribution has been used to constrain a non-local, non-local thermodynamic equilibrium 3D radiative transfer model of the CSE. Results. The high-angular resolution ALMA data have revealed the first example of a source on the AGB where both a faster bipolar outflow and a spiral pattern along the orbital plane can be seen in the gas envelope. The spiral can be traced in the low- to intermediate-velocity (13–25 km s−1) equatorial torus. The largest spiral-arm separation is ≈5.′′5 and consistent with a companion with an orbital period of ≈330 yr and a separation of less than 70 AU. The kinematics of the bipolar outflow is consistent with it being created during a mass-loss eruption where the mass-loss rate from the system increased by at least a factor of five for 10–15 yr. Conclusions. The spiral pattern is the result of an undetected companion. The bipolar outflow is the result of a rather recent mass-loss eruption event.

Context. The asymptotic giant branch (AGB) marks the final evolutionary stage of stars with initial masses between ~0.8 and 8 M ⊙ . During this phase, stars undergo copious mass loss, which contributes significantly to the enrichment of the interstellar medium. The well-accepted mass-loss mechanism requires radiation pressure acting on dust grains that form in the density-enhanced and extended AGB stellar atmospheres. The details of the mass-loss process are not yet well understood, however. For oxygen-rich AGB stars, which are the focus of this study, the dust grains that drive the wind are expected to scatter visible light very efficiently because their sizes are relative large. Aims. We study the distribution of dust in the inner wind of oxygen-rich AGB stars to advance our understanding of the wind-driving process. Methods. We observed light scattered off dust grains that form around three oxygen-rich AGB stars (W Hya, SW Vir, and R Crt) with mass-loss rates between 10 −7 and 10 −6 M ⊙ yr −1 using the extreme-adaptive-optics imager and polarimeter SPHERE/ZIMPOL with three filters centred at 0.65, 0.75 and 0.82 μ m. We compared the observed morphologies and the spectral dependence of the scattered light between the three sources and determined the radial profile, per image octant, of the dust density distribution around the closest target, W Hya. Results. We find the distribution of dust to be asymmetric for the three targets. A biconical morphology is seen for R Crt, with a position angle that is very similar to those inferred from interferometric observations of maser emission and of mid-infrared continuum emission. The cause of the biconical outflow cannot be inferred from the ZIMPOL data, but we speculate that it might be the consequence of a circumstellar disc or of the action of strong magnetic fields. The dust grains polarise light more efficiently at 0.65 μ m for R Crt and SW Vir and at 0.82 μ m for W Hya. This indicates that at the time of the observations, the grains around SW Vir and R Crt had sizes <0.1 μ m, while those around W Hya were larger, with sizes ≳0.1 μ m. The asymmetric distribution of dust around R Crt makes the interpretation more uncertain for this star, however. We find that polarised light is produced already from within the visible photosphere of W Hya, which we reproduce using models with an inner dust shell that is optically thick to scattering. We fit radiative transfer models to the radial profile of the polarised light observed around W Hya and find a steep dust density profile, with steepness varying considerably with direction. We find the wind-acceleration region of W Hya to extend to at least ~7 R ⋆ . This is in agreement with theoretical predictions of wind acceleration up to ~12 R ⋆ , and highlights that ZIMPOL observations probe the crucial region around AGB stars where dust forms and is accelerated.

Context. This is the first publication from the DEATHSTAR project. The overall goal of the project is to reduce the uncertainties of the observational estimates of mass-loss rates from evolved stars on the Asymptotic Giant Branch (AGB). Aim. The aim in this first publication is to constrain the sizes of the 12CO emitting region from the circumstellar envelopes around 42 mostly southern AGB stars, of which 21 are M-type and 21 are C-type, using the Atacama Compact Array (ACA) at the Atacama Large Millimeter/submillimeter Array. The symmetry of the outflows is also investigated. Methods. Line emission from 12CO J = 2→1 and 3→2 from all of the sources were mapped using the ACA. In this initial analysis, the emission distribution was fit to a Gaussian distribution in the uv-plane. A detailed radiative transfer analysis will be presented in a future publication. The major and minor axis of the best-fit Gaussian at the line center velocity of the 12CO J = 2→1 emission gives a first indication of the size of the emitting region. Furthermore, the fitting results, such as the Gaussian major and minor axis, center position, and the goodness of fit across both lines, constrain the symmetry of the emission distribution. For a subsample of sources, the measured emission distribution is compared to predictions from previous best-fit radiative transfer modeling results. Results. We find that the CO envelope sizes are, in general, larger for C-type than for M-type AGB stars, which is as expected if the CO/H2 ratio is larger in C-type stars. Furthermore, the measurements show a relation between the measured (Gaussian) 12CO J = 2→1 size and circumstellar density that, while in broad agreement with photodissociation calculations, reveals large scatter and some systematic differences between the different stellar types. For lower mass-loss-rate irregular and semi-regular variables of both M- and C-type AGB stars, the 12CO J = 2→1 size appears to be independent of the ratio of the mass-loss rate to outflow velocity, which is a measure of circumstellar density. For the higher mass-loss-rate Mira stars, the 12CO J = 2→1 size clearly increases with circumstellar density, with larger sizes for the higher CO-abundance C-type stars. The M-type stars appear to be consistently smaller than predicted from photodissociation theory. The majority of the sources have CO envelope sizes that are consistent with a spherically symmetric, smooth outflow, at least on larger scales. For about a third of the sources, indications of strong asymmetries are detected. This is consistent with what was found in previous interferometric investigations of northern sources. Smaller scale asymmetries are found in a larger fraction of sources. Conclusions. These results for CO envelope radii and shapes can be used to constrain detailed radiative transfer modeling of the same stars so as to determine mass-loss rates that are independent of photodissociation models. For a large fraction of the sources, observations at higher spatial resolution will be necessary to deduce the nature and origin of the complex circumstellar dynamics revealed by our ACA observations.

Context. The observed orbital characteristics of post-asymptotic giant branch and post-red giant branch (post-RGB) binaries are not understood. We suspect that the missing ingredients needed to explain them probably lie in the continuous interaction of the central binary with its circumstellar environment. Aims. We aim at studying the circumbinary material in these complex systems by investigating the connection between the innermost structures and large-scale structures. Methods. We perform high-angular resolution observations of HD 101584 in the near-infrared continuum. HD 101584 has a complex structure as seen at millimeter wavelengths, with a disk-like morphology and a bipolar outflow due to an episode of a strong binary interaction. To account for the complexity of the target, we first perform an image reconstruction and use this result to fit a geometrical model to extract the morphological and thermal features of the environment. Results. The image reveals an unexpected double ring structure. We interpret the inner ring as having been produced by emission from dust located in the plane of the disk, and the outer ring having been produced by emission from dust that is located 1.6 [D/1kpc] au above the disk plane. The inner ring diameter (3.94 [D/1kpc] au) and temperature (T = 1540 ± 10 K) are compatible with the dust sublimation front of the disk. The origin of the out-of-plane ring (with a diameter of 7.39 [D/1kpc] au and a temperature of 1014 ± 10 K) could be episodic ejection or a dust condensation front in the outflow. Conclusions. The observed outer ring is possibly linked with the blue-shifted side of the large-scale outflow seen by the Atacama Large Millimeter/submillimeter Array and may trace its launching location to the central star. Such observations place morphological constraints on the ejection mechanism. Additional observations are needed to constrain the origin of the out-of-plane structure.

This is the first publication of the DEATHSTAR project. The goal of the project is to reduce the uncertainties of observational estimates of mass-loss rates from Asymptotic Giant Branch (AGB) stars. Line emission from 12CO J=2-1 and 3-2 were mapped using the ACA. In this initial analysis, the emission distribution was fit to a Gaussian distribution in the uv-plane. Detailed radiative transfer analysis will be presented in the future. The axes of the best-fit Gaussian at the line center of the 12CO J=2-1 emission gives a first indication of the size of the emitting region. Furthermore, the fitting results, such as the major and minor axis, center position, and the goodness of fit across both lines, constrain the symmetry of the emission distribution. We find that the CO envelope sizes are, in general, larger for C-type than for M-type AGB stars, which is expected if the CO/H2 ratio is larger in C-type stars. Furthermore, a relation between the 12CO J=2-1 size and circumstellar density is shown that, while in broad agreement with photodissociation calculations, reveals large scatter and systematic differences between the stellar types. The majority of the sources have CO envelopes that are consistent with a spherically symmetric, smooth outflow. For about a third of the sources, indications of strong asymmetries are found. This is consistent with previous interferometric investigations of northern sources. Smaller scale asymmetries are found in a larger fraction of sources. These results for CO envelope radii and shapes can be used to constrain detailed radiative transfer modeling of the same stars so as to determine mass-loss rates that are independent of photodissociation models. For a large fraction of the sources, observations at higher spatial resolution will be necessary to further investigate the complex circumstellar dynamics revealed by our ACA observations.

Chemical evolution of the ISM and galaxies depends critically on stellar mass loss. We describe current efforts and future needs and opportunities to characterize AGB outflows: driving mechanisms, outflow rates, underlying fundamental physical and chemical processes such as dust grain formation, and dependency of these on metallicity.