A&A 489, L5–L8 (2008) Astronomy DOI: 10.1051/0004-6361:200810450 & c ESO 2008 Astrophysics Letter to the Editor Diameter and photospheric structures of Canopus from AMBER/VLTI interferometry, A. Domiciano de Souza1, P. Bendjoya1,F.Vakili1, F. Millour2, and R. G. Petrov1 1 Lab. H. Fizeau, CNRS UMR 6525, Univ. de Nice-Sophia Antipolis, Obs. de la Côte d’Azur, Parc Valrose, 06108 Nice, France e-mail: [email protected] 2 Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany Received 23 June 2008 / Accepted 25 July 2008 ABSTRACT Context. Direct measurements of fundamental parameters and photospheric structures of post-main-sequence intermediate-mass stars are required for a deeper understanding of their evolution. Aims. Based on near-IR long-baseline interferometry we aim to resolve the stellar surface of the F0 supergiant star Canopus, and to precisely measure its angular diameter and related physical parameters. Methods. We used the AMBER/VLTI instrument to record interferometric data on Canopus: visibilities and closure phases in the H and K bands with a spectral resolution of 35. The available baselines (60−110 m) and the high quality of the AMBER/VLTI observations allowed us to measure fringe visibilities as far as in the third visibility lobe. Results. We determined an angular diameter of / = 6.93 ± 0.15 mas by adopting a linearly limb-darkened disk model. From this angular diameter and Hipparcos distance we derived a stellar radius R = 71.4 ± 4.0 R. Depending on bolometric fluxes existing in the literature, the measured / provides two estimates of the effective temperature: Teff = 7284 ± 107 K and Teff = 7582 ± 252 K. Conclusions. In addition to providing the most precise angular diameter obtained to date, the AMBER interferometric data point towards additional photospheric structures on Canopus beyond the limb-darkened model alone. A promising explanation for such surface structures is the presence of convection cells. We checked such a hypothesis using first order star-cell models and concluded that the AMBER observations are compatible with the presence of surface convective structures. This direct detection of convective cells on Canopus from interferometry can provide strong constraints to radiation-hydrodynamics models of photospheres of F-type supergiants. Key words. stars: fundamental parameters – stars: individual: Canopus – supergiants – methods: observational – techniques: high angular resolution – techniques: interferometric 1. Introduction performed a detailed spectroscopic study of Canopus, and other intermediate-mass stars, in order to investigate the influence of The evolved star Canopus (α Carinae, HD 45348) is the sec- = − rotation on the internal mixing of elements. They conclude that ond brightest star (V 0.72) in the night sky, just af- the abundance of many stars of their sample, including Canopus, ter Sirius. Although Canopus is classified as a F0II star in are better explained by models including rotation effects in the the SIMBAD database, several works (e.g. Achmad et al. evolutionary models. 1991; Desikachary & Hearnshaw 1982) present it as a F0 su- To constrain evolutionary models it is crucial to have pre- pergiant (F0Ib), which is more adapted to the measured lu- cise measurements of fundamental stellar parameters, such as minosity of L 13 000−15 000 L (e.g. Jerzykiewicz & the effective temperature Teff, luminosity L,radiusR. If the dis- Molenda-Zakowicz 2000; Smiljanic et al. 2006). tance d and the angular diameter / are known, then a quite direct During their evolution, intermediate-mass stars like Canopus method to estimate R is to apply the simple relation R = 0.5/ d. (M 10 M) leave the red giant branch (RGB) and enter the ff The most direct and precise measurements of / can be obtained blue-giant region, significantly increasing their e ective temper- by modern long baseline interferometers. ature and performing the so-called blue-loop. Even if most stars In this work we present a precise determination of the angu- found in the blue-giant region of the HR diagram are experi- lar diameter (and other derived physical parameters) of Canopus encing their blue-loop phase, there are still puzzling questions from interferometric observations performed with the AMBER concerning this phase of stellar evolution (e.g. Xu & Li 2004). focal instrument (Petrov et al. 2007), installed at ESO-VLTI Another unresolved issue in the evolution of intermediate- (Glindemann et al. 2004) located at Cerro Paranal, Chile. The mass stars concerns some discrepancies between observed and observations and data reduction are described in Sects. 2 and 3. predicted abundances. Additional internal mixing processes are The results and conclusions are outlined in Sects. 4 and 5. invoked to explain the observations. Smiljanic et al. (2006) Based on observations performed at the European Southern 2. Observations Observatory, Chile under ESO Program 079.D-0507. Table 1 is only available in electronic form at The AMBER observations were obtained at low spectral reso- http://www.aanda.org lution Rs = λ/Δλ 35 in the H and K bands (LR-HK mode), Article published by EDP Sciences L6 A. Domiciano de Souza et al.: Diameter and photospheric structures of Canopus from AMBER/VLTI interferometry AMBER-DRS (amdlib v1.21) combined with similar selection procedures developed by us, as they were not yet fully imple- mented in the mainstream software. Calibrated visibilities (V) and closure phases are then ob- tained from selected and averaged visibilities and closure phases of Canopus and the calibrator. The uncertainties associated with the calibrated visibilities provided by AMBER-DRS v2.1 corre- spond in general to ∼<0.02V. These low values take into account, for example, fundamental noises and the uncertainty on the cal- ibrator’s angular diameter (Table 1), but do not include time variations of the atmospheric plus instrumental transfer function of AMBER/VLTI. Thus, for each observing night we estimated this additional visibility uncertainty (σV (λ)) from the several ob- servations of the calibrator, averaging over the three baselines. Depending on the night and on the wavelength λ, the relative uncertainty was estimated as σV (λ)/V(λ) 0.03−0.09 in the H band and σV (λ)/V(λ) 0.02−0.06 in the K band. This addi- tional σV (λ) was quadratically added to the corresponding un- certainties from AMBER-DRS v2.1. Fig. 1. The uv-plane coverage of the AMBER/VLTI observations of We investigated the presence of correlated noise in the Canopus spanning the H (green symbols) and K (black symbols) bands AMBER data (Li Causi et al. 2008) as all data taken before with a spectral resolution of 35. As a reference the dotted circles in- September 2007 were affected by an electromagnetic interfer- dicate the expected positions of the first and second visibility minima. Spatial frequencies range from the first to the third visibility lobes. ence producing spurious fringes on the detector array. These spurious fringes were corrupting the interferometric data qual- ity. The AMDC software was developed by Li Causi et al.2 to using three of the 2 m-class Auxiliary Telescopes (ATs) placed filter out these spurious fringes. We applied the AMDC soft- at the VLTI stations A0, K0, and G1. ware on select observing files to check whether this procedure A relatively complete uv-plane coverage was obtained changes significantly the interferometric observables measured thanks to observations performed over 3 nights (2007 April 6−8) on Canopus. We found that, since the star is very bright, the ff spanning 2 h each night and several projected baselines rang- correlated noise does not a ect the observations in the range of their statistical error bars (relative differences are of the or- ing from 60 m to 110 m (Fig. 1). Each AMBER observing −3 file consists of 1000 frames (one interference and three photo- derof10 ). Therefore, we neglect this step in all our data metric channels) recorded with an integration time of 0.026 s. A processing. complete AMBER observing block is typically composed of 5 Currently, there is no standard procedure dedicated to the or 10 observing files recorded on the target followed by similar spectral calibration for AMBER, which introduces an uncer- observations of the calibrator (HD 79917). These files allow the tainty on the wavelength λ. We could partially correct this ef- measurement of one set of calibrated squared visibilities (one at fect by comparing Earth atmospheric transmission curves with each baseline) and closure phases in the H and K bands. Further the AMBER/VLTI low resolution spectra of Canoups recorded details on the observations are given in the online Table 1. in the J, H,andK bands. From the positions of the strong atmospheric absorption features between the J and H,andH and K bands we could determine that the original λ given by 3. Data reduction AMBER should be corrected to new values λ according to the equation λ = λ + δλ,whereδλ = −0.08 μm. Other possibilities The data was reduced using the standard routines of the AMBER to correct λ exist, in particular we have also tried to determine Data Reduction Software (called amdlib) version 2.11. Tatulli this correction by directly fitting δλ as a free parameter of our et al. (2007) describe the principles of the AMBER-DRS rou- model. However, we found that the procedure using the atmo- tines allowing the conversion of raw-data frames into individual spheric transmission gives more robust results. We also estimate complex visibilities. that the uncertainty on the corrected wavelength σ is 0.03 μm The best individual complex visibilities of an observing λ (1 pixel on the detector). As shown in the next sections, such block are then selected and averaged to obtain uncalibrated vis- a relatively high value (σ 2%λ) has a direct consequence ibilities and closure phases.