Astronomical Science DOI: doi.org/10.18727/0722-6691/5004

The Nearby Evolved L2 as a Portrait of the Future Solar System

1,2 Pierre Kervella Five billion from now, the Sun will tudes in the visible, L2 Pup has experi- Miguel Montargès3 grow into a red , more than a enced a remarkable, slow photometric Anita M. S. Richards4 hundred times larger than its current size. dimming over the last decades by more Ward Homan5 It will also experience intense mass loss than 2 magnitudes in the visible. Bedding Leen Decin5 in the form of a stellar wind. The end et al. (2002) interpreted this long-term Eric Lagadec6 product of its evolution, seven billion dimming as the consequence of the Stephen T. Ridgway7 years from now, will be a star obscuration of the star by circumstellar Guy Perrin2 — about the size of the Earth and ex­­ dust. Iain McDonald4 tremely dense (density ~ 5 × 106 g cm–3). 8 Keiichi Ohnaka We observed L2 Pup on the night of This metamorphosis will have a dramatic 21 March 2013 with NAOS CONICA impact on the planets of the Solar Sys- (NACO) as part of a survey aimed at 1 Unidad Mixta Internacional Franco-­ tem, including the Earth. While Mercury imaging the circumstellar environments Chilena de Astronomía (CNRS UMI and Venus will be engulfed by the giant of selected nearby evolved (Kervella 3386), Departamento de Astronomía, star and destroyed, the fate of the Earth et al., 2014a). We used 12 narrow-band Universidad de Chile, Santiago, Chile is still uncertain. The brightening of the filters spread in wavelength between 1.04 2 LESIA (UMR 8109), Observatoire de Sun will make the Earth hostile to life in and 4.05 µm. We processed the image Paris, PSL Research University, CNRS, about one billion years. As an aside, cubes using a serendipitous imaging UPMC, Université Paris-Diderot, France assuming that life appeared on Earth approach (also known as “Lucky imag- 3 Institut de Radioastronomie Millimétri­­ 3.7 billion years ago (Ohtomo et al., 2014), ing”). Using 8400 very short exposures que, Saint-Martin d’Hères, France this implies that life on Earth has already (8 milliseconds each) in each filter, we 4 Jodrell Bank Centre for Astrophysics, exhausted ~ 80 % of its development were able to freeze the residual atmos- Dept. of Physics and Astronomy, Uni- time. However, we do not know whether pheric perturbations. After selecting the versity of Manchester, United Kingdom our then-lifeless rock will be destroyed by best 50 % of the series of images, we 5 Institute of Astronomy, Katholieke the burgeoning Sun, or survive in orbit recentred and averaged them to obtain ­Universiteit Leuven, Belgium around the white dwarf. the 12 final, diffraction-limited images. 6 Laboratoire Lagrange (UMR 7293), They were finally deconvolved using a ­Université de Nice-Sophia Antipolis, To address the question of the impact of point spread function (PSF) calibrator

CNRS, Observatoire de la Côte d’Azur, the final phases of on star. The morphology of L2 Pup in these Nice, France planetary systems, hydrodynamical mod- images (Figure 1) was very surprising. 7 National Optical Astronomy Observa­ els have been proposed (see, for exam- From a seemingly double source between tory, Tucson, USA ple, the recent review by Veras, 2016). 1.0 and 1.3 µm, the star became a single 8 Instituto de Astronomía, Universidad But observational constraints on the star- source with an east-west extension at Católica del Norte, Antofagasta, Chile planet interaction models are still rare. 2.1 µm, and exhibited a spectacular spiral Planets of (AGB) loop at 4.0 µm! stars are embedded in complex circum- The impact of the dramatic terminal stellar envelopes and are vastly outshone We proposed (Kervella et al., 2014a) that phases of the lives of Sun-like stars on by their parent star. The observation of L2 Pup is surrounded by an equator-on their orbiting planets is currently uncer- this critical phase of planetary system dust disc. In this framework, the opacity tain. Observations with NAOS CONICA evolution thus presents considerable, and and thermal emission of the dust change and SPHERE/ZIMPOL in 2014–2015 as yet unsolved, technical challenges. As dramatically between 1 and 4 µm, provid- have revealed that the nearby red giant a result, there currently exists only indi- ing a natural explanation for the changing star is surrounded by an rect evidence of the presence of planets aspect of the circumstellar envelope. At almost edge-on disc of dust and gas. orbiting AGB stars (Wiesemeyer et al., 1 µm, the dust efficiently scatters the light We have observed several remarkable 2009). In addition, the masses of AGB from the star, which therefore appears features in L2 Pup: plumes, spirals, and stars are notoriously difficult to estimate masked behind the dust band. The dust a secondary source (L2 Pup B) which from observations, preventing accurate becomes progressively more transparent is embedded in the disc at a projected determinations of their evolutionary as the wavelength increases. At 2.2 µm, separation of 2 au. ALMA observations states. the thermal emission from the hot inner have allowed us to measure a mass of rim of the disc is observable through the

0.659 ± 0.043 M⊙ for the central star. dust. A colour composite image of This indicates that L2 Pup is a close The discovery of the disc of L2 Puppis L2 Pup in the near infrared is presented analogue of the future Sun at an age with NACO in Figure 2. It shows the intrinsically red of 10 Gyr. We also estimate the mass of colour of the equatorial dust band due

L2 Pup B to be 12 ± 16 MJup, implying At a distance of only 64 pc (van Leeuwen, to the stronger scattering at shorter that it is likely a planet or a brown 2007), L2 Pup is the second nearest wavelengths. dwarf. L2 Pup therefore offers us a AGB star after R Doradus. In addition to remarkable preview of the distant future its regular pulsation with a period of Our hypothesis of an edge-on dust disc of our Solar System. 141 days and an amplitude of ~ 2 magni- was initially relatively fragile. But a 3D

20 The Messenger 167 – March 2017 1.04 μm 1.26 μm 2.12 μm 4.05 μm

N E 50 mas 5 au

Figure 1. NACO deconvolved images of L2 Puppis elongated, and the subtraction of the Sparks et al. (2008) and Kervella et al. at a range of wavelengths from 1 to 4 μm. From central star revealed the presence of a (2014b). Kervella et al. (2014a). second source, L2 Pup B, at a separation of 32 milliarcseconds (mas) from the star Apart from the scattered light on the radiative transfer model using the to the west (Figure 3). disc’s upper and lower surfaces, a very

RADMC-3D code (Dullemond, 2012) striking signature in the pL map (Figure 3, ­confirmed that this interpretation of the The polarimetric imaging capability of right panel) comes from the two plumes NACO images is consistent with the ZIMPOL has been particularly important emerging from the disc. Their high observations, reproducing convincingly in revealing the structure of the envelope degree of polarisation (~ 30 %) indicates both the spectral energy distribution and of L2 Pup (Figure 3, right). As the star that they contain dust and that the scat- the morphology of the disc (Kervella et is embedded in a dust-rich environment, tering angle is large at θ ~ 50°. These thin al., 2014a). the scattering of the starlight by the dust plumes, whose transverse diameter is grains induces a linear polarisation of smaller than 1 au, have a length of more the photons. For small dust grains, the than 10 au. The large scattering angle

Stunning features from SPHERE polari- degree of linear polarisation pL is a implies that they emerge from the disc metric imaging smooth function of the scattering angle, close to perpendicular, and propagate in

with a maximum pL value obtained for the northern cone cavity that is otherwise We took advantage of the Science Verifi- ~ 90° scattering. Knowing the degree of essentially devoid of dust. cation of the Spectro-Polarimetric High- polarisation therefore allowed us to esti- contrast REsearch instrument mate the scattering angle θ over the The map of the degree of polarisation (SPHERE) to observe L2 Pup in visible envelope. Knowing θ and the projected also shows well-defined local maxima at light imaging polarimetry with the Zurich position of the dust relative to the star in a radius of 6 au, symmetrically east and IMaging POLarimeter (ZIMPOL) camera the image, then allowed us to retrieve the west of the star. The degree of polarisa- (Kervella et al., 2015). The observation 3D distribution of the scattering material. tion reached at these positions is very was carried out on the night of 7 Decem- This polarimetric tomography technique high, up to pL = 60 % in the R-band, cor- ber 2014. The combination of the high was previously employed, for example, by responding to scattering of the light at brightness of L2 Pup and good seeing resulted in spectacular image quality. The Strehl ratio produced by the SPHERE adaptive optics reached more than 40 % at a wavelength of 646 nm for an atmos- pheric seeing of ~ 0.7 arcseconds. In one hour of telescope time, the observation of L2 Pup and a PSF calibrator star (β Col) confirmed unambiguously our hypothesis of an edge-on circumstellar dust disc (Figure 3, upper left). In addition to reveal- ing the overall geometry of the dust disc 0.1ೀ — as relatively thin, moderately flared Figure 2. Colour com- posite image of L Pup and extending to at least 13 au — the 2 at infrared wavelengths collected images in the V- and R-bands assembled from NACO showed remarkable structures in the Jupiter observations. The orbits Earth Saturn Uranus of four Solar System envelope of L2 Pup (Figure 3, upper right). We detect several spiral arms in the neb- planets are represented in the lower part of the ula, as well as two intriguing thin plumes. figure to give the overall

The central source appeared significantly scale of the L2 Pup disc.

The Messenger 167 – March 2017 21 Astronomical Science Kervella P. et al., Evolved Star L2 Puppis as a Portrait of the Future Solar System

V-band p Plume 2 L Plume 2 Plume 1 Northern cone Loop Plume 1 Max. pL Spiral 1 Segment Max. pL Edge A B 3 au Band AB

Max. pL Max. N N Spiral 2 pL

Southern cone E E 5 au 100 mas 5 au 100 mas

Figure 3. Left: Colour composite intensity image of 0.1 0.20.3 0.40.5 L2 Pup assembled from SPHERE/ZIMPOL V- and R-band images. Middle: nomenclature of the observed structures in the circumstellar environment high brightness of L2 Pup and the high two-thirds the mass of our Sun (0.653 ± of L2 Pup. Right: degree of linear polarisation, pL, of sensitivity of the array. We detected 0.043 M⊙). It is remarkable that the error L2 Pup measured with ZIMPOL.­ The two plumes ­several molecular emission lines in the budget of this measurement is fully domi- sketched in the middle panel are also indicated. selected spectral windows, from 12CO, nated by the uncertainty in the parallax 13 29 CO, H2O, SO2, SiO, SiS and SO. of L2 Pup, as the error from the ALMA We focus our discussion here on the data fit is only ± 1.7 %. ~ 90°. We interpret these maxima as the 29SiO(ν = 0, J = 8–7) molecular line at a scattering of the starlight by the inner rim rest frequency of 342.981 GHz. The con- The coincidence of the radius where the of the dust disc, that is, the minimum tinuum subtracted integrated intensity rotation becomes sub-Keplerian and the radius at which the dust is present with a in this line is presented in Figure 4 (upper radius of the inner rim (maximum of pL, high density. The existence of an inner left panel). Figure 3) of the dust disc observed with rim is due to the very high brightness of ZIMPOL indicates that the rotation of the the central star (2000 times Solar lumi- Our analysis is based on the position- gas becomes sub-Keplerian precisely nosity), which would lead to the sublima- velocity diagram (PVD) formalism. The when it enters the dust disc. This very tion of any dust grains closer to the star PVD is computed by integrating the interesting effect is likely due to the vis- (Homan et al., 2017). ALMA spectro-imaging data cube over cous coupling of the gas and dust within a 0.02 arcsecond wide pseudo-slit the disc. The dust is sensitive to the aligned with the disc plane (east-west strong radiative pressure from the central An analogue of the future Sun, 5 billion direction). The resulting diagram (Figure star, and is thus subject to a reduced years from now 4, lower left) shows the velocity of the effective gravity resulting in a slower molecular gas as a function of the posi- orbital velocity than the gas. As the inter-

L2 Puppis was observed with the tion along the slit. Thanks to the excellent nal cavity within 5 au contains very little ­Atacama Large Millimeter/submillimeter signal-to-noise ratio of the ALMA data, dust, the gas rotates freely there. But Array (ALMA) in Cycle 3 in Band 7 we can measure very accurately the when it reaches the inner rim of the dust (340 GHz, wavelength ~ 0.8 mm) using orbital motion of the gas in the disc disc, it is slowed down by friction with the longest available configuration of the plane, and more precisely the maximum the dust, which explains its sub-Keplerian interferometer with baseline lengths of up velocity of the gas as a function of the rotation. to 16 km (Kervella et al., 2016). This con- separation from the centre of rotation. We figuration provided the highest possible determined that this maximum velocity From the measured mass of L2 Pup, angular resolution currently achievable by profile closely follows Kepler’s law up to a ­stellar evolution models predict that its ALMA, with a beam size of only 15 mas. radius of 5–6 au from the star. Figure 4 age is ~10 Gyr. The mass, radius and It is interesting to remark that this resolu- (lower left and right panels) clearly shows pulsation period are all consistent with tion matches very well that of SPHERE/ the two rotation regimes: Keplerian within a mass identical to that ZIMPOL in the visible (16–20 mas). We the inner 5 au (that is, v ~ R–1/2), and sub- of the Sun. The missing third of a solar –0.85 selected velocity resolutions of 100 to Keplerian beyond 6 au (v ~ R ). A mass has been lost by L2 Pup during its 400 m s–1 for the molecular line spectral smooth transition is observed between 5 evolution. The extraordinary coincidence windows, corresponding to a spectral and 6 au. From the Keplerian motion of of having a Sun-like AGB star in the close resolution of R ~ 1 000 000. This power- the inner 5 au, we could determine with vicinity of the Sun thus provides us with ful combination of very high spatial and very high accuracy (± 6.6 %) the present a privileged preview of the very distant spectral resolution is permitted by the mass of the AGB star, finding that it is future of our own star.

22 The Messenger 167 – March 2017 29 Figure 4. Upper left: SiO(ν = 0, J = 8–7), ν0 = 342.9805 GHz ZIMPOL NR-band (645 nm) 0.3 0.3 ALMA integrated inten- sity image in the 29SiO (ν = 0, J = 8–7) molecu- lar line. The dark spot 0.2 0.2 on the star is caused by gas absorption along ) ) the line of sight. Upper right: ZIMPOL image in 0.1 0.1 R-band at the same seconds seconds scale. Lower left: Posi- rc rc tion-velocity diagram (a (a et et (PVD) constructed from 0.0 0.0 fs fs the east-west pseudo-

of slit displayed in the n of on io

ti upper left panel. Lower –0.1 –0.1 right: PVD in logarithmic velocity and radius Declina Declin at scales to show the dif- ferent power law –0.2 –0.2 regimes of the velocity profile. The domain and positions of various fea- 0.00 0.19 0.38 0210000 00000 tures are indicated –0.3 –0.3 according to the legend. 0.3 0.20.1 0.0–0.1–0.2 –0.3 0.3 0.20.1 0.0–0.1–0.2–0.3 The envelope of the Right ascension offset (arcseconds) Right ascension offset (arcseconds) maximum velocity profile 30 defines the central star mass, and the inflection 40 29SiO(ν = 0, J = 8–7) Background: West points corresponding to the inner edge of the ν0 = 342.9805 GHz 30 dust disc are shown for 20 25 the east and west edges. 20 )

10 ) –1

–1 15 s m s m (k (k

.0 0 ty 10 loci Ve y — 33 locit

Ve –10 5

Photosphere Inner rim –20 Companion (ALMA) M sin i = 0.653 ± 0.011 M ๬ V ≈ Rα with α = –0.85 ± 0.06 East inflection –5 0510 15 20 West inflection –30 0.3 0.20.1 0.0 –0.1 –0.2 –0.3 0.5 1251020 Position offset on sky (arcseconds) Radius (au)

A candidate planet orbiting in the disc of We estimated the mass of the companion convert this offset into a mass of 12 ±

L2 Pup by comparing the position of the centre 16 MJup for this companion, which corre- of rotation of the molecular disc and the sponds to a planet or a low mass brown The ALMA data in the continuum showed position of the continuum emission peak. dwarf. the presence of an excess of emission The centre of rotation of the molecular in the western wing of the disc (Figure 5), disc is very well defined from the ALMA We also detected excess emission in the contributing 1.3 % of the central star’s PVD (Figure 4), and it corresponds to the PVD of the molecular lines (ν = 0, flux. This emission coincides almost centre of gravity of the mass enclosed by J = 3–2) of 12CO and 13CO at the location ­perfectly with the secondary source the disc. The continuum emission peak of the candidate planet, at a velocity of –1 L2 Pup B detected with SPHERE/ZIMPOL provides the position of the AGB star. We 12 km s (Figure 6). We propose that this one before the ALMA observations. determined that the two points are coin- emission comes from the extended

This continuum emission could be pro- cident within 0.55 ± 0.75 mas. As we molecular envelope accreted by L2 Pup B duced by a dusty envelope or by an know the mass of the central star and the from the wind of the AGB star. This

accretion disc surrounding L2 Pup B. current separation of L2 Pup B, we can measured velocity is consistent with the

The Messenger 167 – March 2017 23 Astronomical Science Kervella P. et al., Evolved Star L2 Puppis as a Portrait of the Future Solar System

line-of-sight projection of the velocity Figure 5. Composite view of L Pup in visible vector of a planet in circular Keplerian 2 light (ZIMPOL image in revolution at a radius of 2.4 au from a the R-band, blue col- 0.65 M⊙ central star, that is, with an ours) and ALMA contin- orbital period of 5 years. uum (orange colours). Plume The central star light has been subtracted from This candidate planet is, to our knowl- the ALMA image to bet- edge, the first planet whose emission ter show the companion is detected with ALMA. We also observe object. The size of the a signature at 6 au in the PVD that may star's is represented to scale. be linked to the plume #1 detected in The white circle in the the ZIMPOL images (yellow arrow in Fig- bottom left corner rep- ure 6). resents the resolution of Red Giant ALMA candidate planet the image (beam size).

The persistence of L2 Pup B between the ZIMPOL and ALMA observations is an indication that this is a compact object rather than a coreless aggregate of gas and dust. The strong Keplerian shear and the radiation pressure would very 5 au quickly blow such a clump into the dust 0.1ೀ disc (R > 6 au). Moreover, the presence of a plume originating from the position of L2 Pup B and launched perpendicular companion. Figure 7 shows our hypothe- (Khouri et al., 2016) and W Hya (Ohnaka to the dust disc plane is an indication sised structure of the detected features in et al., 2016), although they are at a com- that accretion is occurring. Such a possi- the L2 Pup system. We propose that the parable stage of their evolution to L2 Pup, bility is plausible, as the Roche lobe of a large dust loop observed at 4 µm in the do not host axially symmetric circum­ planet embedded in the dust- and gas- NACO images is created by dust conden- stellar structures but irregular, mostly rich environment surrounding L2 Pup will sation in the shadow of L2 Pup B. The spherical envelopes. We suggest that the sweep up a considerable volume and highly collimated plume is ejected from a axial geometry of the envelope of L2 Pup therefore accrete a significant quantity of putative accretion disc surrounding the is due to the presence of its companion. material. planet. Its biconical shape is also strongly evoca- tive of the bipolar planetary nebulae, and

The presence of a candidate planet orbit- Compared to other AGB stars observed L2 Pup may hold the key to demonstrat- ing L2 Pup provides a remarkable test at high angular resolution, the morphol- ing the link between binarity (including case to observe the interactions between ogy of the circumstellar disc of L2 Pup sub-stellar companions) and bipolarity of the evolved star’s wind and a low-mass stands out as remarkably singular. R Dor planetary nebulae.

12CO(ν = 0, J = 3–2) 13CO(ν = 0, J = 3–2) Companion (SPHERE) 30 Inner rim 30 –55 1 5 10 Companion (ALMA) 0 5 Photosphere M sin i = 0.653 M 25 ๬ 25

20 20 Figure 6. Emission of L2 Pup B in two CO ) ) molecular isotopologue –1 –1 12 s s lines ( CO to the left 15 15 13 m m and CO to the right) in (k (k position-velocity dia- y y grams with logarithmic ocit ocit l l velocity and radius Ve Ve 10 10 scales. The emission from the east wing of the disc has been sub- tracted. The positions of various key features are indicated in the ­legend, cf. Figure 4. The yellow arrow marks a 5 5 feature tentatively 0.5 11250 0.51 2510 ascribed to the Plume 1 Radius (au) Radius (au) (cf., Figure 3).

24 The Messenger 167 – March 2017 Figure 7. Proposed Lopez et al., 2014; Millour et al., 2016) will ­configuration of the dust enable milliarc­second spectro-imaging of disc of L2 Pup and its candidate planet is the photosphere and structures present shown from a pole-on around the star (plumes, loop, compan- (upper) and equator-on ion). We plan to develop the observations (lower) perspective. of L2 Pup with these new instruments to monitor its pulsation cycle and its close- in molecular envelope (the molsphere, located within 1–3 stellar radii). The inter-

m action of the molsphere with the orbiting ri r planet L2 Pup B may be an important Inne ingredient in the accretion of gas. v v0 B (ALMA, 2015.84) The forthcoming Extremely Large Tele- A scope (ELT) will provide an angular reso- lution of 60 mas at 10 µm, 12 mas at B (ZIMPOL, 2014.93) 2.2 µm and 3 mas in the visible. Observ-

ing L2 Pup with the ELT will be easy: it is very bright (mV ~ 7.5, mK ~ –2), and thus provides an excellent natural guide

on star for adaptive optics wavefront sens- ti ing. With its 0.4 arcsecond extension, the ansi

R R Tr circumstellar disc is easily resolved spa- 0 B au tially, and high-resolution spectroscopy 0 2468 10 12 will provide a Doppler map of the molec- ular envelope. The interactions of the Keplerian Sub-Keplerian Earth candidate planet with the dust disc and the wind of the central star will thus be observable in very fine detail, revealing the accretion of a stellar wind onto a planetary-mass object.

Loop As an older sibling of our Sun, and thanks Plum to the powerful existing and foreseen

e high angular resolution instrumentation

available at the VLT and ALMA, L2 Pup will undoubtedly continue to surprise us in the coming years with key discoveries pertinent to the distant future of the Solar System.

A References B Bedding, T. R. et al. 2002, MNRAS, 337, 79 Dullemond, C. P. 2012, Astrophysics Source Code Library, 1202.015 Eisenhauer, F. et al. 2011, The Messenger, 143, 16 Prospects for future observations determine its mass from its Keplerian Homan, W. et al. 2017, A&A, submitted Kervella, P. et al. 2014a, A&A, 564, A88 velocity amplitude. Kervella, P. et al. 2014b, A&A, 572, A7 The longest ALMA baselines of 16 km Kervella, P. et al. 2015, A&A, 578, A77 combined with the shortest wavelengths Optical interferometry with the Very Large Kervella, P. et al. 2016, A&A, 596, A92 (Bands 9 and 10) will soon give access Telescope Interferometer (VLTI) commis- Khouri, T. et al. 2016, A&A, 591, A70 Lopez, B. et al. 2014, The Messenger, 157, 5 to an angular resolution ~ 6 mas. Assum- sioning instrument VINCI has been used Millour, F. et al. 2016, Proc. SPIE, 9907, 99073 ing a mass of 12 MJup for L2 Pup B, its by Kervella et al. (2014a) to measure the Ohnaka, K., Weigelt, G. & Hofmann, K.-H. 2016, Roche lobe has a diameter of about angular diameter of the central AGB star A&A, 589, A91 0.6 au, or 10 mas. This means that it will of L Pup. The VLTI second-generation Ohtomo, Y. et al. 2014, Nature Geoscience, 7, 25 2 Sparks, W. B. et al. 2008, AJ, 135, 605 be possible to resolve the putative accre- instruments GRAVITY (Eisenhauer et al., van Leeuwen, F. 2007, A&A, 474, 653 tion disc around L2 Pup B (particularly 2011) and the Multi AperTure mid-Infrared Veras, D. 2016, Royal Society Open Science, 3, 150571 in CO molecular emission) and directly SpectroScopic Experiment (MATISSE: Wiesemeyer, H. et al. 2009, A&A, 498, 801

The Messenger 167 – March 2017 25