Circumstellar HI and CO around the carbon stars V1942 Sgr and V CrB
Y. Libert, E. Gerard, C. Thum, J.M. Winters, L.D. Matthews, T. Le Bertre
aa r X i v : . [ a s t r o - ph . S R ] O c t Astronomy&Astrophysicsmanuscript no. COHIcarbonstars c (cid:13)
ESO 2018November 20, 2018
Circumstellar H i and CO around the carbon starsV1942 Sgr and V CrB Y. Libert , E. G´erard , C. Thum , J.M. Winters , L.D. Matthews , and T. Le Bertre LERMA, UMR 8112, Observatoire de Paris, 61 Av. de l’Observatoire, 75014 Paris, France GEPI, UMR 8111, Observatoire de Paris, 5 Place J. Janssen, 92195 Meudon Cedex, France IRAM, 300 rue de la Piscine, 38406 St. Martin d’H`eres, France MIT Haystack Observatory, O ff Route 40, Westford, MA 01886, USAReceived 11 September 2009 / Accepted 20 October 2009
ABSTRACT
Context.
The majority of stars that leave the main sequence are undergoing extensive mass loss,in particular during the asymptotic giant branch (AGB) phase of evolution. Observations showthat the rate at which this phenomenon develops di ff ers highly from source to source, so that thetime-integrated mass loss as a function of the initial conditions (mass, metallicity, etc.) and of thestage of evolution is presently not well understood. Aims.
We are investigating the mass loss history of AGB stars by observing the molecular andatomic emissions of their circumstellar envelopes.
Methods.
In this work we have selected two stars that are on the thermally pulsing phase of theAGB (TP-AGB) and for which high quality data in the CO rotation lines and in the atomic hy-drogen line at 21 cm could be obtained.
Results.
V1942 Sgr, a carbon star of the Irregular variability type, shows a complex CO lineprofile that may originate from a long-lived wind at a rate of ∼ − M ⊙ yr − , and from a young( < ∼ years) fast outflow at a rate of ∼ − M ⊙ yr − . Intense H i emission indicates a detachedshell with 0.044 M ⊙ of hydrogen. This shell probably results from the slowing-down, by sur-rounding matter, of the same long-lived wind observed in CO that has been active during ∼ years. On the other hand, the carbon Mira V CrB is presently undergoing mass loss at a rate of2 10 − M ⊙ yr − , but was not detected in H i . The wind is mostly molecular, and was active for atmost 3 10 years, with an integrated mass loss of at most 6.5 10 − M ⊙ . Conclusions.
Although both sources are carbon stars on the TP-AGB, they appear to developmass loss under very di ff erent conditions, and a high rate of mass loss may not imply a highintegrated mass loss. Key words.
Stars: AGB and post-AGB – Stars: carbon – (Stars:) circumstellar matter – Stars:individual: V1942 Sgr – Stars: individual: V CrB
1. Introduction
Low- to intermediate-mass stars, at the end of their main-sequence evolution, become first hydro-gen shell-burning red giants (RGB − Red Giant Branch − stars), then hydrogen and helium shell-burning red giants (AGB − Asymptotic Giant Branch − stars). In this second phase they may un- Libert et al.: H i and CO around carbon stars dergo mass loss at a very large rate ( > − M ⊙ yr − ), even so large that it has a decisive e ff ect ontheir late evolution (Olofsson 1999). They are surrounded by expanding envelopes of gas and dustthat have been extensively observed with radio molecular lines and infrared continuum emission.These tracers are used to estimate mass-loss rates. However the estimates are somewhat ambigu-ous because the mass-loss rate of a given source may vary on many di ff erent timescales. The masschange as a function of time due to mass loss is thus di ffi cult to evaluate, and to connect with stel-lar evolution models. Furthermore molecular lines probe an extent of the circumstellar shell (CS)that is limited by photo-dissociation, and therefore furnish information mainly on the inner parts ofCSs, and on the recent mass loss.To try to circumvent these di ffi culties we have started a systematic programme of observationsof red giants in the line of atomic hydrogen at 21 cm. We have published some of our results in sev-eral recent papers, and first reports on sizeable samples have been presented by G´erard & Le Bertre(2006, hereafter GL2006) and Matthews & Reid (2007, hereafter MR2007). A major di ffi culty ofthis programme is the confusion caused by the 21 cm emission from the Interstellar Medium (ISM)that is located on the same line-of-sight as the source of interest. This has a strong impact on theobservations which have to be conducted with a specific approach, and on the data processing thataims at providing spectra corrected for the ISM emission. Perhaps more confounding, as circum-stellar matter is expected to be at some stage injected in the ISM, the confusion by the local ISMmight actually be at least partly of stellar origin, i.e. caused by material ejected at an earlier stageof evolution.In addition to observing systematically the H i line at 21 cm in a large sample of sources withdi ff erent properties, it is also important to choose objects for which the Galactic confusion is lowand / or can be tracked easily, and therefore corrected accurately. The detailed study of such spectrashould serve as a guide to exploit the data that are obtained in more di ffi cult situations.Here we present our results on two carbon stars, V1942 Sgr and V CrB, for which the confusionis not a serious problem, and which have H i properties that di ff er radically. Both are N-type carbonstars (CGCS 4229 and CCCS 2293, respectively) and have already been detected in CO rotationallines (Olofsson et al. 1993a). However the only available CO spectrum of V1942 Sgr had a poorsignal-to-noise ratio, and for our study it appeared essential to also obtain new CO data of highquality.
2. V1942 Sgr
V1942 Sgr is classified as a long-period irregular variable (type Lb). Lebzelter & Obbrugger (2009)have compared the lightcurve properties of Semi-Regular (SR) and Lb variables, and concluded thatLb stars can be seen as an extension of the SRs towards shorter periods and smaller amplitudes.V1942 Sgr is a carbon star on the TP-AGB with a C / O ratio around 1.12 (Olofsson et al. 1993b).Bergeat et al. (2001) estimate its e ff ective temperature, T e ff , at 2960 K. The parallax measured byHipparcos (1.87 ± ⊙ . From the data obtained by IRAS in the mid-infrared there is no direct evidence thatthe star is undergoing mass loss (the low resolution 8-22 µ m spectrum is featureless). However,Olofsson et al. (1993a) discovered emission in the CO(1-0) rotational line centered at V lsr = –31.5km s − , close to the expected radial velocity of V1942 Sgr (V lsr = –32.0 km s − from the General ibert et al.: H i and CO around carbon stars 3 Fig. 1.
Frequency-switched H i
21 cm spectrum obtained with the NRT on the position ofV1942 Sgr. The spectrum enlarged by a factor 20 is also shown as a dashed line. The emissionfrom V1942 Sgr is clearly detected at –33 km s − .Catalogue of Stellar Radial Velocities). The expansion velocity, V exp = − , is surprisinglylarge for an Lb variable. From this spectrum Sch¨oier & Olofsson (2001) derive a mass loss rate of ∼ − M ⊙ yr − (at 535 pc). Extended emission associated with V1942 Sgr was discovered byIRAS (Young et al. 1993a). The 60 µ m data show a resolved shell of external radius 3.2 ′ , i.e. 0.50pc. i observations The H i emission was observed with the Nanc¸ay Radio Telescope (NRT), between March 2007 andJuly 2009, for a total of 85 hours. The NRT beamwidth (FWHM) at 21 cm is 4 ′ in right ascension(RA) and 22 ′ in declination (Dec). An ’on’ source frequency-switched spectrum is presented inFig. 1. The main emission peaks at 50 K around V lsr = − . A narrow emission feature with apeak of ∼ − , is visible on top of the 0.4 K blue wing of the main peaknear 0 km s − .Position-switched spectra were also obtained with on-position taken at the position ofV1942 Sgr and o ff -positions, at ± ′ , ± ′ , ± ′ , ± ′ , ± ′ , ± ′ , ± ′ , ± ′ , and ± ′ . Thecomparison between the spectra obtained for di ff erent values of the throw shows that the interstel-lar emission varies linearly with o ff set in the velocity range from –100 to –20 km s − . It meansthat the H i background emission around V1942 Sgr shows a gradient in RA which is constant foreach velocity. This situation is similar to that encountered for EP Aqr and Y CVn (Le Bertre &G´erard 2004, Figs. 3 and 7). In such a case the source emission can be extracted from the position-switched spectra by subtracting the contribution of the interstellar emission, which is estimated byinterpolation between the two extreme o ff -positions.The intensity of the emission detected from the source in the position-switch spectra is constantwith throw from ± ′ to ± ′ . Therefore the source is mostly confined to the central beam (i.e. ± ′ Libert et al.: H i and CO around carbon stars Fig. 2. H i line profile of V1942 Sgr. The spectrum has been smoothed to a resolution of 0.32 km s − .The dashed line is a fit obtained with the model described in Sect. 4.1.in RA; see GL2006, Sect. 2.1). The spectrum obtained at the star’s position is shown in Fig. 2.It has a shape similar to that obtained on Y CVn (Libert et al. 2007) with a narrow emission linesuperposed on a pedestal extending from –39 to –27 km s − . The narrow emission is centered atV lsr = –32.9 km s − and has a quasi-gaussian profile of width 2.95 km s − (FWHM) and peak in-tensity 168 mJy. The pedestal is also centered at ∼ –33 km s − and has an intensity of ∼ ± / Jy for the NRT at 21 cm). We havealso searched for H i emission at blueshifted velocities down to –48 km s − , and redshifted veloci-ties up to –18 km s − (see the CO spectra in Sect. 2.2). We set an upper limit of 2 mJy for emissionover this range. Nevertheless, we suspect residual features at –46, –26, and –24 km s − , possiblypeaking at ∼ ff -positions are then determined by subtracting the individualposition-switched spectra (on–o ff ) and the contribution of the interstellar emission (assuming thatit varies linearly with RA) from the central spectrum. Furthermore we have obtained data with theon-positions at 11 ′ north and 11 ′ south, and o ff -positions at ± ′ , ± ′ , and ± ′ , and with the on-positions at 22 ′ north and 22 ′ south, and o ff -positions at ± ′ . All these data are used to constructthe flux density map of the source presented in Fig. 3.On this map we see that the intensity at –2 ′ west is almost the same as on the star, and thereforeconclude that the source is slightly o ff set west from the stellar position. Assuming a gaussiandistribution of the intensity, we estimate that the H i source is centered at 0.6 ′ ( ± ′ ) west inRA and at 0 ′ ( ± ′ ) in Dec. The size (FWHM) would then be ∼ ′ in RA and < ′ in Dec. Theintegrated flux in the map is 0.71 Jy × km s − . Assuming that the emission is optically thin and thatatomic hydrogen is at a temperature well above the background ( < ∼ + HI = . − d R S V dV , with M HI in M ⊙ , d in pc, and R S V dV in Jy × km s − , this flux translates to 0.048 M ⊙ of atomic hydrogen at 535 pc. ibert et al.: H i and CO around carbon stars 5 Fig. 3.
Map of the 21 cm H i emission of V1942 Sgr. In each box the label at upper left gives theposition (RA, Dec) with respect to the central star in arcminutes. CO observations of V1942 Sgr were obtained at the IRAM 30-m telescope equipped with EMIR(Eight MIxer Receiver) on June 23, 2009 under average conditions (precipitable water vapor,pwv ∼
10 mm). The two rotational lines, 1-0 and 2-1, were observed in parallel. The four EMIRbands were selected to detect the two orthogonal polarizations at 115.2712 and 230.5380 GHz(T sys ∼
400 K, ∼
800 K, respectively). High spectral resolutions of 20 kHz and 40 kHz respectively(hence 0.05 km s − ) were obtained with the VESPA (Versatile SPectrometer Array) backends. Thetelescope beamwidths are 21 ′′ (at 115 GHz) and 11 ′′ (at 230 GHz), and the observations were madein the wobbler-switching mode using a throw of 60 ′′ in azimuth.The spectra are shown in Fig. 4. These new spectra reveal that the line profiles are compositewith two components centered on the same central velocity, but with di ff erent widths, like those ob-served by Knapp et al. (1998) and Winters et al. (2003) in several late-type giants, mostly oxygen-rich stars of the SR variability type. The emission extends from –48 to –18 km s − , and thereforewe confirm the large expansion velocity ( > ∼
12 km s − ) estimated by Olofsson et al. (1993a).Each line profile was fitted with two parabolae in order to derive representative expansionvelocities (Table 1). We estimate the mass loss rates and photo-dissociation radii associated witheach component using the same approach as in Winters et al. (2003). We adopt a CO / H mass ratioof 1 × − . The di ff erences in the mass loss rates and photo-dissociation radii estimated from thetwo lines are comparable to the uncertainties of the fits. Libert et al.: H i and CO around carbon stars Fig. 4.
CO (2-1, upper panel) and (1-0, lower panel) spectra of V1942 Sgr obtained with the IRAM-30m telescope. The fits used to derive the wind parameters are also shown (see Table 1).
Table 1.
CO line parameters of V1942 Sgr. Formal uncertainties are given in parentheses. V lsr V exp T mb ˙M R CO km s − km s − K M ⊙ yr − cmCO (1-0) –33.75 (0.25) 17.5 (0.5) 0.12 (0.01) 6.1 (0.3) 10 − − − −
3. V CrB
V CrB is a metal-poor ([M / H] = –1.35) carbon star on the TP-AGB with a C / O ratio around 1.10(Abia et al. 2001). It is a Mira of period 358 days. Using the period-luminosity relation for carbonMiras of Whitelock et al. (2006), Guandalini (2009) determines a luminosity of 5600 L ⊙ and adistance of 547 pc. T e ff is estimated at 2090 K (Bergeat et al. 2001). At such a low temperature (i.e.less than 2500 K) molecular hydrogen is expected to be the dominant species in the atmosphereand in the inner envelope of V CrB (Glassgold & Huggins 1983). Indeed photospheric H has beendetected in the near-infrared (2.122 µ m) by Johnson et al. (1983). This Mira is presently undergoingmass loss, since, for instance, it shows clear SiC dust emission at 11.3 µ m (Goebel et al. 1981).The source has also been detected in the CO(1-0) and CO(2-1) rotational lines by Olofsson et al.(1993a), and more recently in CO(3-2) by Knapp et al. (1998). Contrary to V1942 Sgr, only onevelocity component is visible. The central velocity is at V lsr = –99.0 km s − , the expansion velocity,V exp = − , and the mass loss rate, ˙M ∼ − M ⊙ yr − (at 547 pc). With the Plateau-de-Bure IRAM interferometer Neri et al. (1998) find a source of size of 7 ′′ in CO (1-0). On the otherhand IRAS has not detected extended far-infrared emission associated with V CrB (Young et al.1993a, their Table 1). ibert et al.: H i and CO around carbon stars 7 Fig. 5.
Frequency-switch H i
21 cm spectrum obtained with the NRT on the position of V CrB. Thebar indicates the velocity range of the CO emission.V CrB was also observed in H i with the NRT for a total of 44 hours. The frequency-switchedspectrum shows no feature around the expected velocity of –99.0 km s − (Fig. 5), and only a weakinterstellar emission of at most 0.2 K around –99 km s − .We obtained H i data in the position-switch mode of observation with o ff -positions at ± ′ , ± ′ , ± ′ , ± ′ , ± ′ , ± ′ ± ′ and ± ′ . As for V1942 Sgr we find that the interstellar emissionvaries linearly with o ff set, in the velocity range –120 to –80 km s − . We are thus confident that theinterstellar contamination can be corrected for accurately. The source is not detected at the star’sposition and at ± ′ in RA (Fig. 6, in which we have averaged the spectra obtained at + ′ and + ′ , and at – 4 ′ and – 6 ′ , in order to improve the sensitivity). By integrating over the velocity rangedefined by the CO emission, (i.e. –106 to –92 km s − ), an upper limit of 4 mJy × km s − can be seton the intensity of the H i emission at the V CrB’s position in the area defined by the 4 ′ × ′ NRTbeam. It translates to an upper limit of 3 10 − M ⊙ in atomic hydrogen at 547 pc. As the source wasnot found to be extended by IRAS, we do not expect much material outside the NRT beam.
4. Interpretation
The CO line profiles observed in V1942 Sgr have a characteristic composite profile. This kindof profile has been interpreted as evidence for a succession of mass loss events with di ff erentoutflow velocities by Knapp et al. (1998) and Winters et al. (2003). However the interferometricdata obtained on EP Aqr, a source with such profiles, are di ffi cult to explain with this scenario(Winters et al. 2007). Furthermore, in other cases, X Her (Kahane & Jura 1996) and RS Cnc (Libertet al. 2009), there is evidence that the broad components originate in a bipolar flow. Bipolar flowsare believed to develop at the end of the AGB phase when the stars are about to begin their evolutiontowards the planetary nebula phase (e.g. Sahai et al. 2007). Libert et al.: H i and CO around carbon stars Fig. 6. H i spectra obtained on V CrB (middle), and at + ′ east (top) and –5 ′ west (bottom) aftercorrection for interstellar contamination (Sect. 3). The spectra have been smoothed to a resolutionof 0.64 km s − . The bar indicates the velocity range of the CO emission.The H i spectrum obtained on the star’s position shows a pedestal of width 10 km s − that couldbe a counterpart of the narrow CO (1-0 and 2-1) components that have about the same width. Themass in hydrogen corresponding to this pedestal is ∼ − M ⊙ . Assuming 90 % in H and 10 % in He, in number (i.e. a mean molecular weight, µ , of 1.3), and adopting a mass loss rate of 1 10 − M ⊙ yr − from the narrow CO components (see Table 1), the timescale would be 6 10 years, andthe radius 0.31 pc ( ≡ ′ ). The stellar e ff ective temperature (2960 K) is large enough that we expectmost of the hydrogen in atomic form.On the other hand, there is no H i counterpart to the broad CO components at a level of 2 mJy.This seems to indicate that the broad components correspond to a quite recent phenomenon. Indeed,adopting an upper limit of 2 mJy, the flux is < × km s − , and the mass in atomic hydrogen isat most 5 10 − M ⊙ . The timescale is then < ∼ years.The H i spectra obtained on V1942 Sgr are very similar to those obtained by Le Bertre & G´erard(2004) and Libert et al. (2007) on Y CVn, a well documented carbon star with a detached shelldiscovered by IRAS (Young et al. 1993a) and imaged by ISO (Izumiura et al. 1996). On the centralposition we detect a pedestal of width 10 km s − , twice the expansion velocity measured for thenarrow CO components. On all positions, we detect a narrow line of width, FWHM ∼ − .This narrow profile proves that the stellar wind from V1942 Sgr is slowed down at some distancefrom the central star. Young et al. (1993b) have interpreted the detached shells that were revealedby IRAS at 60 µ m as the e ff ects of a slowing-down of stellar outflows by surrounding interstellarmatter. Elaborating on this hypothesis and using the formalism of Lamers & Cassinelli (1999),Libert et al. (2007) developed a model in which the inner radius of a detached shell corresponds ibert et al.: H i and CO around carbon stars 9 Table 2.
Model parameters (d =
535 pc). ˙M (in hydrogen) 0.69 10 − M ⊙ yr − µ
61 10 yearst DS yearsr ′ )r f ′ )r ′ )T ( ≡ T − ), T +
20 K, 746 KT f ( = T ) 81 Kv ( ≡ v − ), v + − , 1.27 km s − v f − v − n − , n + − , 1.8 H cm − n − f , n + f − , 2.15 H cm − n − M r < r (in hydrogen) 4.2 10 − M ⊙ M DT , CS (in hydrogen) 3.7 10 − M ⊙ M DT , EX (in hydrogen) 6.3 10 − M ⊙ The notations are as in Libert et al. (2007). In particular, t DS is the formation time of the detached shell,M DT , CS is the mass of the circumstellar component of the detached shell, and M DT , EX is the external massaccreted in the detached shell. to the location where the stellar wind is abruptly slowed down from V exp to ∼ V exp /
4. The outerradius corresponds to the location where external matter is compressed by the expanding shell(bow shock). They applied this model to Y CVn and obtained excellent fits to the H i line profilesobserved at di ff erent pointings, on and around the star’s position.We are using the same model for V1942 Sgr. For the freely expanding wind (r < r ) we adopt avelocity of 5.0 km s − , i.e. half the width of the pedestal, which corresponds also to the narrow COcomponents. The broad CO components have no obvious counterpart in H i . They probably tracea short lived structure that is restricted to the central part of the circumstellar shell and that has noe ff ect on the large scales probed at 21 cm. The mass loss rate, ˙M = − M ⊙ yr − , is adoptedfrom the CO line fitting (Table 1). The central velocity is taken at –32.9 km s − .We obtain a good fit to the di ff erent H i line profiles (Figs. 2 and 7) with the parameters givenin Table 2. The external radius, r ∼ ′ , that we have adopted is compatible with that derived byYoung et al. (1993a) from IRAS data at 60 µ m, r ext ∼ . ′ , but not the internal radius (r = ′ ,versus r int ∼ . ′ ). In our model, r is constrained by the parameters obtained from the low velocityCO components and by the intensity of the H i pedestal. We assume that the inner shell is too smallcompared to the IRAS beam at 60 µ m (FWHM ∼ ′ ) to have been reliably constrained. V CrB was not detected in H i . As there is no significant Galactic confusion, we are quite confidenton our upper limit of 3 10 − M ⊙ in atomic hydrogen. For a source losing atomic hydrogen witha mass loss rate of 2.1 10 − M ⊙ yr − , it would correspond to a timescale of 2000 years ( µ = i and CO around carbon stars Fig. 7.
Comparison between the H i line profiles obtained on V1942 Sgr and the detached-shellmodel discussed in Sect. 4.1. Top: average of the two spectra at + ′ (east) and –2 ′ (west). Middle:average of the two spectra at + ′ (north) and –11 ′ (south). Bottom: average of the four spectra at( + ′ , + ′ ), (–2 ′ , + ′ ), ( + ′ , –11 ′ ), and (–2 ′ , –11 ′ ). ibert et al.: H i and CO around carbon stars 11 Fig. 8.
Atomic hydrogen column density profile for the V1942 Sgr model. The vertical lines markthe radii r (0.31 pc), r f (0.41 pc), and r (0.47 pc), which define the detached shell.However, the stellar e ff ective temperature is so low (2090 K) that hydrogen should be in molecularform in the atmosphere and outwards (Glassgold & Huggins 1983), until it is photo-dissociatedby the interstellar radiation field. To estimate the distance, R ph , at which this happens, we followthe approach of Morris & Jura (1983). Assuming a mean intensity of the ultraviolet radiation be-tween 912 and 1100 Å of 1.9 10 photons cm − s − sr − , and that 0.11 of all the absorptions leadto a dissociation, we get R ph ∼
410 ˙M / , with R ph in pc and ˙M in M ⊙ yr − . For a mass loss rateof 2.1 10 − M ⊙ yr − , we obtain R ph = ≡ ′ ), which corresponds to a dynamical time of ≈ years (V exp = − ).Therefore the non-detection of V CrB in H i implies that it has not been undergoing mass loss atthe present rate for more than 3.2 10 years. Furthermore, when comparing with V1942 Sgr, whichis at the same distance, we can state that V CrB has not gone through the same phase of massloss during the past 5 10 years, because if it had done so it would have been easily detected likeV1942 Sgr.This reasoning assumes that molecular hydrogen is not self-protected within small-scale struc-tures that could develop in the stellar outflow. However the non-detection by IRAS of an extendedemission around V CrB (Young et al. 1993a) agrees with our conclusion that mass loss has startedonly recently. The V1942 Sgr proper motion measured by Hipparcos is 10.98 mas yr − in RA and –5.10 mas yr − in Dec. When corrected for solar motion towards apex, and for a distance of 535 pc, it translates to6.45 mas in RA and –2.28 mas in Dec. This implies a motion in the plane of the sky at a velocity of17 km s − , and at a position angle, PA = ◦ . Accounting for the radial velocity, V lsr = –33 km s − ,we obtain a 3D space velocity of 37 km s − . The o ff set with respect to the central star that we findin the H i map might therefore be an e ff ect of the motion of V1942 Sgr relative to the surrounding i and CO around carbon stars ISM. Such a deformation in H i has already been noted in several cases: Mira (Matthews et al.2008), RX Lep (Libert et al. 2008), RS Cnc (MR2007 and Libert et al. 2009). GL2006 noted alsothat many H i sources are o ff set with respect to the central stars. A visual inspection of the IRASmap at 60 µ m of V1942 Sgr (Improved Reprocessing of the IRAS Survey : Miville-Deschˆenes &Lagache 2005) reveals that the image is slightly elongated in RA and shifted west by ∼ / ≡ . ′ ), in agreement with our H i map. Finally, it is worth noting that the central velocity in H i is–32.9 ± − , whereas in CO it is –33.5 ± − . The e ff ect is small but consistent withan interaction between the external shell of V1942 Sgr and its local ISM. Shifts in velocity of theH i emission towards the LSR have already been reported in several red giants (GL2006, Matthewset al. 2008).From their study of circumstellar shells resolved by IRAS, Young et al. (1993b) find that, amongnearby AGB stars detected in CO, Miras, in contrast to Semi-Regulars, are in general unresolved.They suggest that the latter have been losing matter for a longer time than the former. In theirH i survey of evolved stars GL2006 obtained results that agree with this suggestion. Although theirsample was small, several Miras with high mass loss rate could not be detected in H i , whereasSRs were often easily detected. The high quality data that we have obtained on V1942 Sgr andV CrB strengthen the case of SRs undergoing mass loss for a longer time than Miras. One normallyassumes that SRs evolve into Miras, and it is puzzling to find no relics of this SR phase aroundseveral Miras. SRs might evolve directly in the post-AGB phase, as suggested also by the presenceof bipolar outflows which has been reported in several cases (Kahane & Jura 1996, Libert et al.2009).Young et al. (1993b) also find that extended sources are observed preferentially around carbonstars and GL2006 obtained a higher rate of detection of carbon stars in H i . However, the caseof V CrB seems to suggest that some stars could reach the carbon-rich stage without undergoingsubstantial mass loss previously. In a systematic investigation of the relations between mass lossand red giant characterstics, Winters et al. (2000) find that the mass loss rate depends criticallyon stellar parameters such as the e ff ective temperature, which controls the dust formation, and theluminosity, which controls the radiation pressure. V CrB may have switched only recently from theB-regime (with a low and, presently, undetected wind) to the A-regime with a wind at a few 10 − M ⊙ yr − .Although both V1942 Sgr and V CrB are carbon stars on the TP-AGB phase, their history ofmass loss during the past 5 10 years seem to di ff er radically. If it is correct that the bipolar shapingis a signpost of the end of the AGB, V1942 Sgr (and also the sources with composite CO line pro-files) might be on the verge of leaving this phase. Both sources have about the same C / O abundanceratio, 1.12 for V1942 Sgr (Olofsson et al. 1993b) and 1.10 for V CrB (Abia et al. 2001), and thesame luminosity, 5200 and 5600 L ⊙ , respectively. Also both sources have a low C / C abundanceratio, 30 for V1942 Sgr (Abia & Isern 1997) and 10 for V CrB (Abia et al. 2001), as compared to ∼
40 for the majority of carbon stars in the AGB phase. The explanation of such low abundanceratios is not known, but could be due to a non-standard mixing process occurring in low-mass starsat the base of the convective stellar envelope (“cool bottom processing”, Nollett et al. 2003). ibert et al.: H i and CO around carbon stars 13
5. Conclusions
The combination of high velocity resolution CO and H i data is a promising tool to investigate thehistory of mass loss by evolved stars. The low level of Galactic H i emission and the absence ofsmall-scale structure in this emission have allowed us to obtain H i data of high quality on V1942Sgr and V CrB with the NRT. We have also obtained high quality CO (1-0) and (2-1) spectra ofV1942 Sgr with the IRAM 30-m telescope.For V1942 Sgr, the CO spectra exhibit composite profiles, that reveal a low velocity wind of ∼ − M ⊙ yr − and a high velocity wind, possibly bipolar, of ∼ − M ⊙ yr − . The comparisonwith the H i spectrum shows that this high velocity wind is recent with an age of at most 10 years.On the other hand, the low velocity wind appears to have filled a cavity of ∼ years. Follow-upobservations with the VLA and ALMA would help to improve this scenario, or possibly lead to anew scheme. The narrowness of the H i line profile in V1942 Sgr brings new evidence that AGBstellar winds are slowed down by their surrounding medium as surmised by Young et al. (1993b).For V CrB, the CO spectra that have been published reveal an outflow with expansion veloc-ity, 6.5 km s − , and mass loss rate, 2.1 10 − M ⊙ yr − . The non-detection in H i of V CrB sets anupper limit of 3.2 10 years for the age of this outflow. In such case of a star with low e ff ectivetemperature, molecular hydrogen data are obviously needed to constrain better the history of massloss. Acknowledgements.
The Nanc¸ay Radio Observatory is the Unit´e scientifique de Nanc¸ay of the Observatoire de Paris,associated as Unit´e de Service et de Recherche (USR) No. B704 to the French Centre National de la Recherche Scientifique(CNRS). The Nanc¸ay Observatory also gratefully acknowledges the financial support of the Conseil R´egional de la R´egionCentre in France. We thank the IRAM Director, P. Cox, for allowing the CO observations of V1942 Sgr to be made onDirector’s time (D01-09). IRAM is supported by INSU / CNRS (France), MPG (Germany), and IGN (Spain). We are gratefulto C. Abia, M. Busso, R. Guandalini, and A. Jorissen for enlightening discussions, and to the referee for careful commentsthat helped us to improve the manuscript. This research has made use of the SIMBAD and VizieR databases, operated atCDS, Strasbourg, France and of the NASA’s Astrophysics Data System.
References
Abia, C., Busso, M., Gallino, R., Dom´ınguez, I., Straniero, O., & Isern, J., 2001, ApJ, 559, 1117Abia, C., & Isern, J., 1997, MNRAS, 289, L11Bergeat, J., Knapik, A., & Rutily, B., 2001, A&A, 369, 178G´erard, E., & Le Bertre, T., 2006, AJ, 132, 2566 (GL2006)Glassgold, A.E., & Huggins, P.J., 1983, MNRAS, 203, 517Goebel, J.H., Bregman, J.D., Witteborn, F.C., Taylor, B.J., & Willner, S.P., 1981, ApJ, 246, 455Guandalini, R., 2009, in preparationIzumiura, H., Hashimoto, O., Kawara, K., Yamamura, I., & Waters, L.B.F.M., 1996, A&A, 315, L221Johnson, H.R., Goebel, J.H., Goorvitch, D., & Ridgway, S.T., 1983, ApJ, 270, L63Kahane, C., & Jura, M., 1996, A&A, 310, 952Knapp, G.R., Young, K., Lee, E., & Jorissen, A., 1998, ApJS, 117, 209Lamers, J.G.L.M., & Cassinelli, J.P., 1999, ”Introduction to Stellar Winds”, Cambridge University PressLe Bertre, T., & G´erard, E., 2004, A&A, 419, 549Lebzelter, T., & Obbrugger, M., 2009, Astron. Nachr., 330, 390Libert, Y., G´erard, E., & Le Bertre, T., 2007, MNRAS, 380, 1161Libert, Y., Le Bertre, T., G´erard, E., & Winters, J.M., 2008, A&A, 491, 789Libert, Y., Winters, J.M., Le Bertre, T., G´erard, E., & Matthews, L.D., 2009, A&A, submitted i and CO around carbon stars Matthews, L.D., Libert, Y., G´erard, E., Le Bertre, T., & Reid, M.J., 2008, ApJ, 684, 603Matthews, L.D., & Reid, M.J., 2007, AJ, 133, 2291 (MR2007)Miville-Deschˆenes, M.-A., & Lagache, G., 2005, ApJS, 157, 302Morris, M., & Jura, M., 1983, ApJ, 264, 546Neri, R., Kahane, C., Lucas, R., Bujarrabal, V., & Loup, C., 1998, A&AS, 130, 1Nollett, K.M., Busso, M., & Wasserburg, G.J, 2003, ApJ, 582, 1036Olofsson, H., 1999, IAU Symposium n ◦◦