1. Name of program and authors: Mass loss and outflow velocities in Magellanic Cloud AGB stars Martin Groenewegen 2. Science goal: Stars with masses below about 8 solar masses shed almost their entire atmosphere at the end of their life during the Asymptotic Giant Branch phase. During this phase of strong mass loss (up to 10(-4) solar masses/year) a circumstellar envelope (CSE) is maintained containing both dust and molecules. The dust radiates mainly in the infrared, up to the sub-mm. Modelling the Spectral Energy Distribution of an AGB star, and its broad-band dust features in the Near- and Mid-IR range provides information about the DUST-mass loss rate and the composition of the dust. However, one of the input parameters in such modelling is the expansion velocity of the wind. The molecular content of the CSE is traced through its most abundant molecule besides H2, namely CO. Heterodyne observations of CO do not only provide information on the GAS-mass loss rate from the peak intensity (assuming a CO/H2 ratio), but additionally the outflow velocity, and the systemic velocity, useful for kinematic studies. Combining IR/sub-mm with heterodyne observations are necessary to obtain the best constrained values on the outflow velocity, TOTAL mass loss and dust-to-gas ratio of AGB stars. Combining data from the Galaxy, LMC and SMC allows one to study the effect of metallicity on AGB mass loss and outflow velocity. Both are theoretically predicted to be smaller at lower metallicity, and the latter is partially confirmed by observations of H20 masers in LMC sources which indicate a 20% smaller outflow velocity (e.g. van Loon et al. 2001, A&A 368, 950). The proposed observations would allow to study a statistically significant sample of both O- and C-stars in both SMC and LMC. Observations of the type discussed above are standard for our galaxy, with maybe 300 AGB stars being detected in CO at mm wavelengths. Regarding the Magellanic Clouds, continuum Near-, Mid- and Far-IR fluxes will be and have been obtained with ISO, SPITZER, and HERSCHEL, and a few 100 AGB stars with mass loss are known in the Clouds. Now, for the first time with ALMA, heterodyne observations of CO lines of AGB stars in the Magellanic Clouds are within reach. (Note that OH and SiO maser detections exist of very few AGB stars and supergiants in the Clouds, but these data are not easily related to the mass loss [OH], or come from a region where the outflow is still being accelerated [SiO]. In addition, many AGB stars, especially at low-Z, are carbon-rich and hence do not show OH or SiO maser lines !). Expected main-beam temperatures have been calculated using a model for the well studied Galactic carbon star CW Leo (Groenewegen et al., 1998, A&A 338, 491), and putting it at 50 kpc. The predicted temperatures in the CO(1-0), (2-1), (3-2), (6-5) line, in a beam of 1.0*(230 GHz/ freq) arcsec are 0.25, 0.47, 0.49, 0.49 K, respectively. In one hour of integration with a velocity resolution of 0.4 km/s, the rms temperatures are, respectively, 0.05, 0.05. 0.08, 0.34 K, according to the ALMA sensitivity tool. This indicates that the CO(2-1) line is the most sensitive one, in accordance with experience for Galactic sources, and is therefore the line selected for the observations. The beam-size is estimated from the CO photodissociation radius in a source like CW Leo, when located at 50 kpc. The high velocity resolution is preferred for 2 reasons. First, one may realistically expect some sources to have expansion velocities of only a few km/s due to the effect of the lower metallicity in the Clouds. Second, the proposed observations do not try to resolve the CO shell; in fact the beam size is tailored to the CO photodissociation radius. However, it might be possible in the brightest sources to get a hint for possible asymmetries in the CSE by looking for deviations from the expected standard parabolic profile. This has been done for IRC 10 216 (Groenewegen & Ludwig, 1998, A&A 339, 489) but requires high velocity resolution. When such deviations would be detected one might then try to resolve these CSEs. 3. Number of sources Based on previous work using IRAS and ISO, about 50 AGB stars in LMC and SMC have the required high luminosities and mass loss rates (i.e. typically 10(-6) msol/yr and larger) to be detectable in 1h of integration in the CO(2-1) line. 4. Coordinates: 4.1. RA and DEC: LMC (5h40 -69d) and SMC (1h -73d) 4.2. Moving target: no 4.3. Time critical: no 5. Spatial scales: 5.1. Angular resolution: 1 - 1.5 arcsec 5.2. Range of spatial scales/FOV: 1 to 20 arcsec 5.3. Single dish: no 5.4. ACA: no 5.5. Subarrays: no 6. Frequencies: 6.1. Receiver band: Band 6 6.2. Lines and Frequencies 12CO(2-1) at 230 GHz 6.3. Spectral Resolution (km/s): 0.4 km/s 6.4. Bandwidth or spectral coverage: 100 km/s 7. Continuum flux density: 7.1. Typical value: N.A. 7.2. Continuum peak value: N.A. 7.3. Required continuum rms: N.A. 7.4. Dynamic range in image: N.A. 8. Line intensity: 8.1. Typical value: 0.5 K 8.2. Required rms per channel: 0.05 K 8.3. Spectral dynamic range: >5 9. Polarization: no 10. Integration time per setting: 1h per star 11. Total integration time for program: 50 hours ************************************************************ Review by Fredrik Schoeier and reply by Martin Groenewegen 1) There appears to be a significant overlap with project 1.8.2 (Aalto et al.). The Aalto et al. project has the same basic scientific motivation but will also aim at obtaining a slightly larger sample of stars with a much larger spread in mass-loss rates. The importance of studying AGB stars in the Magellanic Clouds is of course to see the how the mass loss depends on metallicity but also that you will get rid of any systematic effects introduced in the analysis, through the distance estimates, and possibly get more reliable correlations of wind-properties with, e.g., stellar parameters such as luminosity. This, is were the Aalto et al. proposal is more ambitious. There is indeed some overlap with project 1.8.2. I will briefly discuss the overlap and differences. Proposal 1.8.2 covers both continuum and line observations. The current DRSP does not consider continuum observations as they are only feasible for very few objects, in fact much fewer than suggested by Aalto et al. (see ``Will the LSA detect continuum or line emission from AGB stars in the LMC ?'', Groenewegen M.A.T., in: ``Science with large millimetre arrays'', Dec. 1995, Garching, Ed. P. Shaver, Springer Verlag, p. 164). The quoted continuum sensitivity in DRSP 1.8.2 of 0.025 mJy in 1 hour at 345 GHz is ONE-sigma only. Therefore objects need to be significantly brighter than this to have their photometry reliably determined in a reasonable time. In my opinion this implies much fewer than 40 potential targets. The proposals overlap in the line observations. Aalto et al. mention 20 objects at high mass-loss rates and 40 at mass-loss rates down to 10(-6) msol/yr. The current proposal mentions 50 stars (at implied mass loss rates of 10(-6) and higher). 2) The reviewer agrees with the predicted line intensities. In the proposal text it is not mentioned that the lines are typically 20-40 km/s broad for these high mass-loss-rate objects. Why is a resolution of 0.1 km/s requested? 1-2 km/s is enough to get a very good handle on also the line profile (This is also the resolution requested by Aalto et al.). 3) It should also be pointed out that the sources will generally not be resolved (however the Aalto et al. project will use 0.05 arcsec resolution for some 20 sources with high mass loss rates) and no spatial information of the emission obtained. That said, ALMA is still the best suited telescope for a project like this given the fact that the compact emission is just too diluted in small single-dish telescopes like, eg., APEX. Points 2 and 3 will be addressed together. One difference between Aalto et al. and the current proposal is the required velocity resolution. The former requires 1 km/s, while I preferred 0.1 km/s. Aalto et al also suggest to work at even lower resolution to push the detection of stars at even slightly lower mass-loss rates. One has to be careful on this issue however. The theory of dust-driven winds predicts that at lower metallicity the terminal velocity should be lower, which is confirmed by observations of H2O maser emission (e.g. van Loon et al. 2001, A&A 368, 950) indicating an about 20% smaller expansion velocity in LMC sources, suggesting even lower velocities in SMC sources. In a Galactic sample of 300 carbon stars [with mass loss rates above 10(-6)] the outflow velocities range from 7 to 40 km/s with a mean of 18.7 km/s (Groenewegen et al. 2002, A&A 390, 511). Therefore one has to consider velocities in the range 4 to 30 km/s and means of 10-15 km/s in SMC and LMC. I believe it is safer to go for a high velocity resolution (of 0.4 km/s or better, AND IN THE DRSP I NOW REQUIRE 0.4 km/s INSTEAD OF 0.1 km.s) and have sufficient redundancy from the beginning, and then bin in velocity a posteriori if required, than to go for a resolution of 1 km/s which we already know will not be sufficient to get an accurate determination of the velocity in some sources. The high velocity resolution is preferred for another reason, which was in my mind when I wrote the DRSP, but which I have now made explicit. The proposed observations do not try to resolve the CO shell; in fact the beam size is tailored to the CO photodissociation radius. However, it might be possible in the brightest sources to get a hint for possible asymmetries in the CSE by looking for deviations from the expected standard parabolic profile. This has been done for IRC 10 216 (Groenewegen & Ludwig, 1998, A&A 339, 489) but requires high velocity resolution. When such deviations would be detected one might then try to resolve these CSEs. A final difference between the two DRSP is the fact that 1.8.2 plans to observe the CO(3-2) line, while I propose CO(2-1). As already mentioned in the DRSP, it is expected that the CO(2-1) line will provide the higher S/N.