A second post-AGB nebula that contains gas in rotation and in expansion: ALMA maps of IW Car
V. Bujarrabal, A. Castro-Carrizo, H. Van Winckel, C. Sanchez Contreras, M. Santander-Garcia
AAstronomy & Astrophysics manuscript no. iwcar c (cid:13)
ESO 2018November 8, 2018 L etter to the E ditor A second post-AGB nebula that contains gas in rotation and inexpansion: ALMA maps of IW Car
V. Bujarrabal , A. Castro-Carrizo , J. Alcolea , H. Van Winckel , C. S´anchez Contreras , and M. Santander-Garc´ıa , Observatorio Astron´omico Nacional (OAN-IGN), Apartado 112, E-28803 Alcal´a de Henares, Spaine-mail: [email protected] Institut de Radioastronomie Millim´etrique, 300 rue de la Piscine, 38406, Saint Martin d’H`eres, France Observatorio Astron´omico Nacional (OAN-IGN), C / Alfonso XII, 3, E-28014 Madrid, Spain Instituut voor Sterrenkunde, K.U.Leuven, Celestijnenlaan 200B, 3001 Leuven, Belgium Centro de Astrobiolog´ıa (CSIC-INTA), Ctra. M-108, km. 4, E-28850 Torrej´on de Ardoz, Madrid, Spain Instituto de Ciencia de Materiales de Madrid (CSIC). Calle Sor Juana In´es de la Cruz 3, E-28049 Cantoblanco, Madrid, Spainaccepted 02 December 2016
ABSTRACT
Aims.
We aim to study the presence of both rotation and expansion in post-AGB nebulae, in particular around IW Car, a binarypost-AGB star that was suspected to be surrounded by a Keplerian disk.
Methods.
We obtained high-quality ALMA observations of CO and CO J = − Results.
Our observations clearly show the presence of gas components in rotation, in an equatorial disk, and expansion, which showsan hourglass-like structure with a symmetry axis perpendicular to the rotation plane and is probably formed of material extractedfrom the disk. Our modeling can reproduce the observations and shows moderate uncertainties. The rotation velocity corresponds toa central stellar mass of approximately 1 M (cid:12) . We also derive the total mass of the molecule-rich nebula, found to be of ∼ − M (cid:12) ; the outflow is approximately eight times less massive than the disk. From the kinematical age of the outflow and the mass valuesderived for both components, we infer a (future) lifetime of the disk of approximately 5000–10000 yr. Key words. stars: AGB and post-AGB – circumstellar matter – radio-lines: stars – planetary nebulae: individual: IW Car
1. Introduction
Keplerian disks around post-AGB stars have been proven to bevery elusive. Gas in rotation has been directly observed in onlytwo nebulae to date, the Red Rectangle and AC Her (Bujarrabalet al. 2013b, 2015), by means of interferometric mm-wave mapsof CO lines. The Red Rectangle and AC Her belong to a classof binary post-AGB stars with low-mass nebulae and with sev-eral independent lines of evidence of disks (e.g., Van Winckel2003; de Ruyter et al. 2006; Gezer et al. 2015; Bujarrabal et al.2013a). They are characterized, in particular, by spectral energydistributions (SEDs) with a NIR excess that indicates hot dustclose to the stellar system. Single-dish observations of CO and CO mm-wave emission in these post-AGB stars systematicallyyielded characteristic line profiles, which are strikingly similarto those of the Red Rectangle and AC Her and to those expectedto be emitted by relatively extended Keplerian disks (Bujarrabalet al. 2013a). A slowly expanding component was also proposedto be present in this class of objects from those CO data. ALMAmaps of CO lines in the Red Rectangle indeed show a bipo-lar low-velocity outflow (Bujarrabal et al. 2013b, 2016), veryprobably formed of gas extracted from the disk and containinga mass approximately ten times smaller. Such a component wasalso confirmed from CO maps of another of these NIR-excesspost-AGBs, 89 Her (Bujarrabal et al. 2007). Rotation was notactually resolved in 89 Her, but a small disk could be confinedto the prominent central condensation. In this source (and prob-ably in others observed in single-dish), the contribution to the total emission of the outflow is dominant and the outflow con-tains a mass at least comparable to that of the compact disk. Onthe other hand, no sign of outflow was found in the maps of ACHer, in which the expanding gas is probably very di ff use.The evolution of these objects is not well known (e.g., DeMarco 2014) and could be very di ff erent from that of high-mass(pre)planetary nebulae. However, both kinds of sources share re-markable properties, such as dominant axial symmetry, whichhas been proposed to be associated with rotating disks (e.g.,S´anchez Contreras et al. 2002; Soker 2001; Balick & Franck2002). Therefore, the study of our objects, the only post-AGBones in which disks are detected, could be relevant to understandthe formation of post-AGB nebulae in general. Moreover, disksare observed in binary post-AGB stars that, surprisingly, showorbits of insu ffi cient size to accommodate an AGB star (Gezeret al. 2015; Van Winckel et al. 2009). In the best-studied neb-ula, the Red Rectangle (Bujarrabal et al. 2013b, 2016), the totalangular momentum of the disk is not negligible and, providedthat all momentum comes from the binary system, would implya significant decrease of the distance between the stars. Betterconstraints on the structure and evolution of the disks are there-fore imperative to studying the orbital evolution of these systemsand their late evolution.We present ALMA maps of IW Car that clearly show bothrotating and expanding gas. A simplified model is comparedwith the observations, allowing us to estimate the main nebularparameters. IW Car is an RV Tau variable (Kiss et al. 2007) thatbelongs to the class of NIR-excess post-AGB stars mentioned a r X i v : . [ a s t r o - ph . S R ] D ec . Bujarrabal et al.: A second post-AGB nebula that contains gas in rotation and in expansion: ALMA maps of IW Car Fig. 1.
ALMA maps per velocity channel of CO J = − ±
5, 15, 45, and 135 mJy / beam (equivalent to ± LSR velocities are indicated in each panel (upper-left corner) and the insertin the last panel shows the beam width.
Fig. 2.
Predictions of our nebula model for CO J = − ∼ cm has been derived from model fitting of the SED,although the nebular shape has not been well studied. IW Car isprobably a double stellar system (with poorly known orbital pa-rameters) and shows a particularly high atmospheric depletion.Following those authors we adopt a distance of 1 kpc for thissource, but we stress that this value is uncertain.
2. Observations
We present maps of IW Car in the CO and CO J = − λ = ∼
110 minutes of acquisitions were obtained onsource. The data were first calibrated with the CASA softwarepackage. The quasars J0522-3627 and J1058 + (cid:48)(cid:48) × (cid:48)(cid:48) − , which was degraded for our final mapsbecause of the poor S / N at high velocities, to a resolution of0.85 km s − for CO J = − − for CO J = −
2. By comparisonwith the APEX single-dish profile (Bujarrabal et al. 2013b), weconclude that a small fraction of ∼
25% of the flux has been fil-tered out in the maps of CO J = −
2. The percentage of lostflux could be higher in the line wings, at ± − from thevelocity center, but remains moderate throughout, < ∼ ∼ ∼ (cid:48)(cid:48)
3. Results
In Fig. 1, we show our ALMA maps per velocity channel ofthe CO J = − ∼ ◦ .
2. Bujarrabal et al.: A second post-AGB nebula that contains gas in rotation and in expansion: ALMA maps of IW Car
Fig. 3.
ALMA maps per velocity channel of CO J = − ±
5, 15, 45, and 135 mJy / beam (equivalent to ± LSR velocitiesare indicated in each panel (note that the velocity resolution is higher than in Fig. 1). The last panel shows the continuum emission,contours: 5, 15, 45, and 135 mJy / beam.The structure is comparable to that found in the similar object 89Her (Sect. 1; but smaller in angular units). The strikingly similardistributions of both structures strongly suggest that these lobesare expanding, as confirmed by the model we describe below.Fig. 3 shows our maps of CO J = −
2. A hint of the lobes can beseen, but their intensity is certainly lower than for CO J = − CO J = − ◦ . The Keplerian pattern is very obvious inboth CO and CO diagrams. We recall that 89 Her shows ex-panding lobes similar to those found here, but no rotation wasdetected in it, although Bujarrabal et al. (2007) argued in favorof the presence of rotation within the central unresolved clump.Among the direct observational facts, we underline the par-ticularly high peak brightness found in the maps, of almost 100K. These high values, very similar to the high brightness foundin the Red Rectangle (Bujarrabal et al. 2013b, 2016), and thecomparable peak intensities found for CO and CO J = − CO) and that the temperatures in IWCar should be high and similar to those deduced for the RedRectangle, > ∼
100 K in central regions. We also point out the pres-ence of blueshifted absorption of the relatively intense emissionof central regions by outer and cooler expanding gas approach-ing us, as shown by the low negative contours that appear at ap-proximately –35 km s − in Figs. 1 and 4; this e ff ect is expectedin expanding nebulae and is in fact also found in our modeling,see further discussion in App. A.3. The S / N ratio and angular resolution of our data are moderateand, moreover, only maps of CO and CO J = − CO J = − J = − CO J = − Fig. 4.
Position-velocity diagrams from our ALMA maps of CO J = − = –15 ◦ .Contours and the rest of the imaging parameters are the same asin the channel maps. The pointed lines show approximate cen-troids in velocity and position.In view of the similar outflows and disks found in IW Carand in 89 Her and the Red Rectangle (Sect. 1), we base ourmodeling on the properties we previously derived for those ob-jects. The width of the disk is not easy to determine from thedata because of the relatively low angular resolution, we adopt ashape similar to that found in the Red Rectangle. In our best-fitmodel, the expansion velocity is radial and its modulus is as-sumed to vary with the distance to the equator and to the axis(see App. A.1). The disk rotation is Keplerian with V (2 10 cm) = − , which corresponds to a central stellar mass of ∼ M (cid:12) ; this mass value is reasonable for a low-mass post-AGBdouble system, though it is smaller than that found in the RedRectangle ( ∼ M (cid:12) ). The velocity value is uncertain by ap-proximately ±
25% (also taking into account the uncertainty inthe inclination, see below). The uncertainty of the deduced cen-tral mass is 1 + . − . M (cid:12) . The axis inclination with respect to theplane of the sky is poorly determined; it cannot be either toolarge or too small in order to explain the easy detection of bothexpansion and rotation. We adopt a value of ∼ ◦ ( ± ◦ ).We assume LTE populations for the involved rotational lev-els. This is a reasonable assumption for low- J CO transitions inthe dense material expected in our sources, n > cm, sincetheir Einstein coe ffi cients are more than ten times smaller thanthe typical collisional rates; see further discussion in Bujarrabalet al. (2013b, 2016). The use of LTE may introduce some uncer-
3. Bujarrabal et al.: A second post-AGB nebula that contains gas in rotation and in expansion: ALMA maps of IW Car
Fig. 5.
As in Fig. 1 but for CO J = − X ( CO) ∼ − , to ease thecomparison with previous results on this object (Bujarrabal et al.2013a). In order to match the relatively low CO / CO inten-sity ratio, we need a low abundance ratio, we deduce X ( CO) ∼ − ; similar low ratios are often found in similar objects. CO J = − CO J = −
2, confirming that the COemission is not very opaque. Our models yield values of τ ( CO J = −
2) smaller than ∼ n and T ) are assumed to depend solely on the distance tothe center and the equator with potential laws. See more detailsin the Appendix; the density distribution is shown in Fig. 6. Thetotal mass derived from our fitting is 4 10 − M (cid:12) , very similarto the value found by Bujarrabal et al. (2013a) from an analy-sis based on very preliminary information on the properties ofthe nebula and only CO single-dish profiles. The mass of theoutflow is approximately 8 times smaller than that of the disk.
4. Summary and conclusions
We present high-quality ALMA observations of CO and CO J = − ∼ ◦ . Thisstructure is very similar to that detected in a similar object, 89Her (Bujarrabal et al. 2007). In the position-velocity diagramsalong the perpendicular direction, for PA = –15 ◦ , Figs. 4 and5, we easily recognize the presence of a disk in Keplerian rota-tion, probably in the equator of the nebula. Similar rotating diskshad been well identified before in only two post-AGB nebulae,the Red Rectangle and AC Her, which also belong to the sameclass of binary stars. The expected presence of both rotating andexpanding gas had been well confirmed before in only the beststudied of these objects, the Red Rectangle, where the outflowwas shown to probably be extracted from the disk (Sect. 1).Our modeling of the observations is simple but is able toreproduce them and confirms our conclusions above (Sect. 3.1;Figs. 2 and 6). We find the Keplerian rotation to be compatiblewith a central stellar mass of ∼ M (cid:12) , reasonable for a post-AGB(double) star of this kind. High temperatures of approximately100 K were typically found in the nebula. The total nebular massis found to be ∼ − M (cid:12) , with the mass of the outflow beingapproximately eight times smaller than that of the disk. Thesemass values are ∼ Fig. 6.
Structure and distributions of the velocity and density (incolor) of our best-fit model for the disk and outflow. Only thevelocity of the outflow is shown. The inclination of the disk withrespect to the line of sight is shown by the red-dashed line.Rectangle, but the disk / outflow mass ratios are very similar inboth nebulae and the general structure and dynamics are compa-rable.In view of the similar properties we find in IW Car and theRed Rectangle, we conclude that the expanding gas in IW Caralso comes from the disk. From the mass values and outflowkinematics in IW Car, we can derive the typical kinematicaltime required to form the outflow ( ∼ / preplanetary lifetimes of these objects, andthe disks could even survive in their (probable) planetary nebulaphase. We conclude that the coexistence of rotating equatorialdisks and gas in expansion extracted from it is probably a sys-tematic property of this class of post-AGB nebulae and that theseare long-living structures. Acknowledgements.
ALMA is a partnership of ESO (representing its memberstates), NSF (USA) and NINS (Japan), together with NRC (Canada) and NSCand ASIAA (Taiwan), in cooperation with the Republic of Chile. The JointALMA Observatory is operated by ESO, AUI / NRAO and NAOJ. We made use ofthe ALMA dataset ADS / JAO.ALMA
References
Balick, B., & Frank, A. 2002, ARA&A, 40, 439Bujarrabal, V., van Winckel, H., Neri, R., et al. 2007, A&A, 468, L45Bujarrabal, V., Alcolea, J., Van Winckel, H., Santander-Garc´ıa, M., & Castro-Carrizo, A. 2013a, A&A, 557, A104Bujarrabal, V., Alcolea, J., Van Winckel, H., et al. 2013b, A&A, 557, L11
4. Bujarrabal et al.: A second post-AGB nebula that contains gas in rotation and in expansion: ALMA maps of IW Car
Bujarrabal, V., Castro-Carrizo, A., Alcolea, J., & Van Winckel, H. 2015, A&A,575, L7Bujarrabal, V., Castro-Carrizo, A., Alcolea, J., et al. 2016, A&A, in pressDe Marco, O. 2014, Asymmetrical Planetary Nebulae VI Conference, 122de Ruyter, S., van Winckel, H., Maas, T., et al. 2006, A&A, 448, 641Gezer, I., Van Winckel, H., Bozkurt, Z., et al. 2015, MNRAS, 453, 133Giridhar, S., Rao, N. K., & Lambert, D. L. 1994, ApJ, 437, 476Kiss, L. L., Derekas, A., Szab´o, G. M., Bedding, T. R., & Szabados, L. 2007,MNRAS, 375, 1338Soker, N. 2001, ApJ, 558, 157S´anchez Contreras, C., Sahai, R., & Gil de Paz, A. 2002, ApJ, 578, 269Van Winckel, H. 2003, ARAA, 41, 391Van Winckel, H., Lloyd Evans, T., Briquet, M., et al. 2009, A&A, 505, 1221
Appendix A: Further discussion on our simplemodeling
A.1. Model description
The determination of the structure and dynamics of the nebulaaround IW Car is di ffi cult because of the lack of informationon this source, in particular on the main properties of the neb-ula. For that reason, we have chosen a simple modeling withfew free parameters. See a representation of out model nebulain Fig. 6 and our model predictions in Fig. 2 and Figs. A.1 andA.2. The model nebula is in some way a compromise betweenthose we used to describe the observations of 89 Her and the RedRectangle (Bujarrabal et al. 2007, 2016).In the rotating equatorial disk, we assume a purely Keplerianrotation velocity in the central biconical region (Fig. 6), closerthan R K = cm, with V rot ( R K = − ). In outer parts,we assume that there is also expansion at 3 km s − , superposed torotation, whose modulus in this case decreases inversely propor-tional to the distance (following the law of angular momentumconservation; the same was found for the outer part of the disk ofthe Red Rectangle, which has been studied in much more detail).The density of the disk is assumed to vary with the distance tothe star following a simple potential law, n ∝ r − . with a valueof 4 10 cm − at half distance and 4 10 cm − at the point wherethe velocity changes to purely Keplerian, R K . The temperaturevaries proportionally to r − . , with a temperature of T =
200 Kat R K .In the hourglass-like outflowing component, we assume ra-dial velocity, with a modulus depending on the distances to theequator h and to the axis p (varying proportionally to h and lin-early with p out – p , where p out is the maximum value of p for agiven value of h ); see Fig. 6. This law is similar to that found forthe Red Rectangle. The temperature is assumed to vary againwith potential laws depending on the distance to the center: T ( r ) ∝ r − . , and a typical T (10 cm) =
85 K. At 2 10 cm, hightemperatures of approximately 200 K are attained. The fitting isimproved if we assume a more complex variation for the out-flow density, depending again on h and p ; we adopted a law pro-portional to h − . and linear with p out – p . Density values rangetypically between ∼ and 10 cm − , see Fig. 6. A.2. Uncertainties in the model parameters
Some parameters describing the structure and dynamics showsignificant uncertainties due to the lack of data on this neb-ula. The disk width is not resolved, so its structure is uncertain.However, its diameter is accurate within ∼ CO measuredintensities (see below).The disk dynamics also show uncertainties. In particular, thepoint at which the rotational velocity law changes is di ffi cult todetermine. The outflow velocity field in the disk is also not easyto describe, but the presence of two regimes is necessary, be-cause otherwise the predicted position-velocity cuts are signifi-cantly di ff erent from the observational data. See an example inFig. A.3 of predictions in which a purely Keplerian field is as-sumed. The outflow velocity modulus is constant and equal to 3km s − , it probably represents an order of magnitude of the de-partures from a Keplerian law we must introduce to avoid theseproblems, more than a well defined velocity field determinedfrom model fitting. Due to the poor information on this parame-
5. Bujarrabal et al.: A second post-AGB nebula that contains gas in rotation and in expansion: ALMA maps of IW Car ter and its meaning, we did not explore possible complex laws,though we recognize that we probably followed an overly simpledescription. However, the rotation velocity is better constrainedbecause we can identify it in the position-velocity diagrams (seeSect. 3.1).The main properties of the hourglass-like component are alsodi ffi cult to study in detail from the existing data. We assumed ra-dial expansion, one of the simplest laws for the velocity. Radialexpansion is a simple definition and the expected case if launch-ing takes place along the force lines of a (locally) radial mag-netic field. A linear variation of the absolute value is in gen-eral expected if most accelerations take place at the beginningof the process; in our case, variation depending on the latitudeand p out – p (as for the density, see A.1) leads to somewhat betterpredictions than dependence on the distance to the center. Wecan exclude some other simple cases, such as constant velocitymodulus, expansion in the axial direction, and expansion paral-lel to the equator, for which the predictions do not match the ob-servations. However, other velocity laws, such as that found forthe outflowing gas in Red Rectangle by Bujarrabal et al. (2016),which is more complex, would lead to results compatible withthe data (although involving a higher number of parameters). Westress that the launching mechanism itself is obviously very un-certain and its possible nature is a weak argument to support anyvelocity field; a theoretical discussion on such processes is be-yond the scope of this letter. The total size, general shape, andoverall velocity of the outflowing gas are however relatively wellconstrained, since they are given by the observed structure atseveral km s − from the central velocity and the total velocityextent of the emission.As mentioned in Sect. 3.1, the assumption of LTE level pop-ulation simplifies a lot our analysis. Nevertheless, we must keepin mind that the rotational temperatures we use in our calcu-lations may only represent J -averaged excitation temperaturesand not true kinetic temperatures. Since high- J levels can obvi-ously be less populated in reality than for LTE, these rotationaltemperatures are probably a lower limit to the actual ones in theoutflowing, relatively di ff use gas.Another consequence of the assumption of thermalized levelpopulations is that it is not possible to distinguish the e ff ectsof the density and the relative CO abundances, X ( CO, CO).Fortunately, X seems relatively well constrained in the best stud-ied objects, particularly X ( CO), which is the basic parameter todetermine the total density and mass because of the lower opac-ity of CO lines. Our previous works always yielded X ( CO)ranging between 10 − and 2 10 − , both for post-AGB disksand for young PNe in general. We will adopt X ( CO) ∼ − to ease the comparison with previous results on this ob-ject (Bujarrabal et al. 2013a). In order to match the relativelylow CO / CO intensity ratio, we need a relatively low abun-dance ratio. We adopt X ( CO) ∼ − ; similar low ratios are of-ten found in similar objects. CO J = − CO J = −
2, confirming that the CO emission is not very opaque.Our model yields typical values of the optical depth τ ( CO J = − ∼ τ ( CO J = − < ∼ CO J = −
2, we find optical depths in the densest parts ofthe disk τ ( CO J = − ∼
1. However, τ ( CO J = − ∼ CO intensity. The value of the total mass is mainly af- fected by the assumed X ( CO) value and depends slightly onother parameters such as the geometry and kinematics.
A.3. Comparison of the predictions with the observationaldata
Fig. 2 shows the observational and synthetic maps per veloc-ity channel and Figs. A.1 and A.2 show the predicted position-velocity diagrams along the rotating disk. The model repro-duces most of the observed features, but some problems arestill present. The major one concerns the faint extended emis-sion, mostly at moderate velocities further than approximately2 km s − from the central velocity. The predicted features arealways somewhat wider. This is very probably due to the signifi-cant loss of flux in the interferometric observation (Sect. 2). Thisrelatively extended and weak emission is expected to be partic-ularly a ff ected by the significant instrumental flux loss, whichmeans that we can expect noisier data and a significant overes-timate of the predictions. Because of the steep logarithmic scalewe use, changes at a level of approximately 1 /
30 of the peakbrightness (our first contour) are very clearly shown in the fig-ures, even if their e ff ect on the total emission is moderate. Thisissue also appears in the synthetic position-velocity diagrams, atintermediate velocities and at ∼ ± (cid:48)(cid:48) ff ect in our modeling, but it cannot disappear withoutsignificantly a ff ecting other predictions, like the total velocityextent, which become incompatible with observations.We also note that the observed emission extends significantlymore at less negative LSR velocities; an asymmetry that we donot try to introduce in our model, which assumes symmetry withrespect to the equator and axis. Obviously, the actual nebula doesnot exactly show such symmetries, which is a common issuewhen modeling post-AGB nebulae.Another e ff ect of the presence of expanding gas is the weakabsorption that appears at approximately –35 km s − . The ab-sorption is noticeable in the di ff erent shapes observed betweenrelatively redshifted and blueshifted features (Figs. 1 and 2)and in the di ff erent blue / red sides of the p-v diagram (Figs.4, A.1), with blueshifted emission systematically weaker. Thisphenomenon depends on subtle geometrical and velocity coin-cidences and is very di ffi cult to model. Our code predicts ab-sorption features similar to those observed but, because of theuncertain treatment of this e ff ect, we will not try to fit it. Fig. A.1.
Position-velocity diagrams predicted from our simplemodel for the CO J = −
6. Bujarrabal et al.: A second post-AGB nebula that contains gas in rotation and in expansion: ALMA maps of IW Car
Fig. A.2.
As for Fig. A.1 but for CO J = −
2. Syntheticposition-velocity diagram along the equatorial direction, to becompared with Fig. 5.
Fig. A.3.
As for Fig. A.1 but assuming that the outer disk showsno expansion velocity. The rest of the model parameters are keptthe same. The predicted position-velocity field is significantlydi ff erent from the observations (Fig. 4).erent from the observations (Fig. 4).