A&A 377, L9-L13 (2001)
D. Barrado y Navascués1,2 -
M. R. Zapatero Osorio3 -
V. J. S. Béjar4 -
R. Rebolo4,5 -
E. L. Martín6 -
R. Mundt2 - C. A. L. Bailer-Jones2
1 - Departamento Física Teórica, C-XI-506, Universidad Autónoma de Madrid, 28049 Madrid, Spain
2 - Max-Planck-Institut für Astronomie, Königstuhl 17, 69117 Heidelberg, Germany
3 - Division of Geological & Planetary Sciences, California Institute of Technology, MS 150-21, Pasadena, CA 91125, USA
4 - Instituto de Astrofísica de Canarias, 38205 La Laguna, Tenerife, Spain
5 - Consejo Superior de Investigaciones Científicas, CSIC, Spain
6 - Institute of Astronomy, University of Hawaii at Manoa, 2680 Woodlawn Drive, Honolulu, HI 96822, USA
Received 29 May 2001 / Accepted 15 August 2001
We have obtained low resolution optical spectra of 15 isolated planetary mass objects (IPMOs) in the Orionis cluster, and derived spectral types by comparison with nearby M and L dwarfs. The spectral types are in the range late M - mid L, in agreement with our expectations based on colors and magnitudes for bona fide members. Therefore, most of these objects have masses below the deuterium burning limit. About 2/3 show H in emission at our spectral resolution. From our spectroscopic and photometric data, we infer that three IPMOs in this sample may be binaries with components of similar masses. These results confirm that the substellar mass function of the Orionis cluster, in the form dN/dM, keeps rising in the planetary domain.
Key words: giant planet formation - open clusters and associations: individual: Orionis - stars: brown dwarfs
Very recently, we have discovered 18 very faint, red objects in the Orionis open cluster (5 Myr, 352pc), using optical and infrared photometry (Zapatero Osorio et al. 2000). If they are indeed members of the association, their masses would be below 18 (1047 ). Other low mass objects have been found in the young cluster IC 348 (Najita et al. 2000) and in the Trapezium (Lucas & Roche 2000; Lucas et al. 2001). They are quite intriguing, since those with masses below 13 would be unable to sustain any nuclear reaction at any time (in particular deuterium fusion). Therefore, they would have masses in the planetary domain and some authors have dubbed them free-floating planets, non-fusors, or isolated planetary mass objects (IPMOs). These names rely on the mass of the objects, which can be estimated from observations and comparisons with models, whereas their origin or formation mechanism cannot be known for sure. Note that these mass values are model dependent and should be taken with some caveats. Since the first spectroscopic confirmation of the nature of some IPMOs in Orionis and the Trapezium (Zapatero Osorio et al. 2000; Lucas et al. 2001), they have become very interesting for two different reasons: (i) they extend the sequence of low mass objects beyond the deuterium burning threshold (Saumon et al. 1996; Chabrier et al. 2000); and (ii) their very existence poses a challenge to our understanding of how they have been created during the collapse and fragmentation of molecular clouds, since so far no model has been able to predict the formation of objects in isolation in this low mass range (e.g. Bodenheimer 1998). Other alternatives have been recently suggested, such as the formation in multiple systems as stellar embryos and the ejection from the system before they accrete enough material to become stars or brown dwarfs (Boss 2001; Reipurth 2001; Bate 2001).
Here, we present low resolution optical spectroscopy of isolated planetary mass candidates in the Orionis cluster. In our analysis, we assume a likely cluster age of 5Myr, which relies on several observational facts combined with theory: (i) the 5Myr-isochrone provides the best fit to the location of stellar and substellar Orionis members in color-magnitude diagrams (Béjar et al. 2001), (ii) there is no evidence of lithium depletion in low mass stellar members indicating that the cluster is younger than 8Myr in order to account for total lithium preservation (Zapatero Osorio et al. 2001), (iii) high mass star models including mass loss calculations (Meynet et al. 2001) predict that the central multiple star (Orionis itself, O9.5 V, still burning hydrogen on the main sequence phase) has to be younger that about 7Myr. Therefore, we believe the age of 5Myr is very realistic, being 7 Myr an upper limit for the age of the association.
|Figure 1: VLT/FORS spectra of isolated planetary mass candidate members of the Orionis cluster. Top panel: bright objects, data smoothed using a boxcar of 5 pixels. Lower panel: faint objects, data smoothed using a boxcar of 5 pixels and a Gaussian function with a 12 sigma kernel. Note the logarithmic scale in the y-axis. For comparison, we include two L1.5- and L6-type field objects and SOri47.|
|Open with DEXTER|
Our spectroscopic data were collected with the Very Large Telescope Unit #1 at the Paranal Observatory of the European Southern Observatory during Dec. 23-27, 2000. We used the FORS1 spectrograph and the multi slit capability. FORS1 has a pixel scale in the standard resolution, yielding a field of view of . For our spectroscopic data, we used the 150I grism and the order-blocking filter OG590. With a slit width of 1.4 , our resolution is , as measured in the comparison arcs. Our sample of Orionis IPMOs was selected from Zapatero Osorio et al. (2000). In total, we observed 14 out of the 18 original objects, together with SOri47, discovered previously (Zapatero Osorio et al. 1999). Table 1 lists magnitudes and colors. Exposure times are in the range 2400-16800 s.
The data were reduced using standard procedures within the IRAF environment. Individual exposures were added together. Then, the spectra were extracted using the "apall'' package within IRAF, fitting the sky to remove the emission lines and the background. The wavelength calibration was performed using HeArHgCd comparison arcs taken with the same configuration. Then data were flux calibrated using spectrophotometric standards. Finally, we improved the signal to noise ratio (S/N) by smoothing the spectra with a boxcar of 5 pixels (). The spectra corresponding to faintest objects were also convolved with a Gaussian function of 12 sigma kernels (). Figure 1 displays the spectra of our Orionis targets. The upper panel shows the brightest targets ( mag), with a good signal to noise ratio ( S/N=100-40). while the lower panel displays fainter objects, down to I=23.77 mag, with a much poorer quality. Note that when, due to the generally low S/N at the bottom of the molecular bands, the subtraction of the sky spectrum was not good enough, we have modified artificially the bad spectral range. These areas are replaced with horizontal segments in the modified spectra. In addition to the Orionis targets, we observed several nearby field objects of M and L spectral type, for comparison purposes (to derive spectral types).
|Figure 2: Spectral type against J magnitude. Solid circles represent data from this study, whereas open circles correspond to data from Béjar et al. (2001). The location of the brown dwarf-planetary mass domain borderline in Orionis is at around J=17.9mag at the age of 5Myr. The lines represent several 5Myr isochrones from Baraffe et al. (1998) - thin lines - and Chabrier et al. (2000) - thick lines, which were obtained for different temperature scales (low gravity by Béjar et al. 2001 - dashed lines; high gravity by Basri et al. 2000 - solid lines, and different gravities by Luhman 1999 - dotted lines). See text.|
|Open with DEXTER|
Figure 1 also depicts several relevant spectral features observable in the optical, like lines from alkali elements (K I 7665 and 7699Å, Na I 8183 and 8195Å, Cs I 8521 and 9843Å, and Rb I 7800 and 7948Å, dotted vertical lines), and molecular absorptions of TiO, FeH, CrH (vertical thin dashed lines), and of VO and H2O (vertical thick dashed lines). In particular, the differences can be appreciated, both in the slope of the pseudo-continuum as well as the change in the strength and width of the K I resonance doublet, VO and TiO bands, etc. Detailed studies of characteristics of late M and field L very low mass stars and brown dwarfs can be found in Kirkpatrick et al. (1999) and Martín et al. (1999). Figure 1 indicates that, with this resolution, it is possible to distinguish key spectral features and attempt a spectral classification of the Orionis candidate members. Spectral types have been assigned following the scheme proposed in Martín et al. (1999). We have measured flux ratios between several bands (pseudo-continuum, VO, TiO, etc.) and compared them with M- and L-type, nearby field objects. For the faintest object in the sample, having a low S/N, we derived a spectral type based on the slope of the continuum. Our final spectral types and uncertainties are listed in Table 1. Note that the indices we have used, in particular PC3 (Martín et al. 1996), are based on field, older objects, which have higher surface gravity than members of Orionis. Evolutionary models predict gravities about for very low mass objects at young ages around 5Myr. L-type sources (very cool temperatures) display optical energy distributions characterized by strong atomic lines of Na I and K I (Allard et al. 2001; Pavlenko et al. 2000), which markedly dominate the shape of the far-red wavelengths. These lines are very sensitive to gravity (Allard et al. 2001): low gravity L-type spectra display significantly less intense alkaline lines. Because the blue side of the PC3 index is located very close to the K I resonance doublet, it provides earlier L spectral types for low gravity objects than for high gravity ones. The M-classes are not so much affected (Béjar 2001) by the gravity effect. On the contrary, the near-infrared spectra of these types are governed by water vapor absorptions, which are steeper for low gravities (Allard et al. 2001). Our latest optical spectral types appear shifted by 1-2 subclasses towards warmer L-types compared to the near-infrared classification given in Martín et al. (2001) for the objects in common. The objects in our target list define a neat spectral sequence: the fainter the magnitude and redder the color, the cooler the spectral class. We show the relation between spectral type and J magnitude in Fig. 2. Open circles correspond to Orionis brown dwarfs from Béjar et al. (2001), whereas solid circles stand for the data studied here. Estimated masses for each magnitude appear on the right-hand side of the diagram on the basis of the 5Myr dusty isochrone by Chabrier et al. (2000). Albeit this is the likely age of the cluster (see Sect. 1), we note that the location of the borderline between brown dwarfs and IPMOs ( mag at 5Myr) does not change by a large amount for other possible ages such as 3Myr ( mag), 7Myr ( mag), and the oldest value of 10Myr ( mag). For comparison purposes, Fig. 2 also includes several 5Myr isochrones using the grainless models of Baraffe et al. (1998) and dusty models of Chabrier et al. (2000). These models provide magnitudes in the filters of interest and are represented as thin and thick lines, respectively. Essentially, there is no difference in this diagram between dusty and grainless models for objects warmer than about L1. We have adopted several calibrations in order to derive spectral types from effective temperatures:
|Figure 3: Detail around the H line. No smoothing or convolution with Gaussian functions have been performed.|
|Open with DEXTER|
The equivalent width of the H
line at 6563 Å is
considered as an age indicator in M-dwarfs, and it is usually
associated to stellar activity. In general, the stronger the emission
line for a given spectral type, the younger the object. We have
identified this feature in emission for the first time in such low
mass objects, and measured its equivalent width (EW). The results are
listed in Table 1. For those objects we do not see H
emission we can impose an upper limit to the EW at 10Å and
20Å, depending on their brightness. Nine objects out of 14 (the
spectrum of one of them does not reach this wavelength) have a
significant emission in H,
a strong evidence that they are
young and probable members of the stellar association. A close-up of
the area around H
is depicted in Fig. 3. We note that very few
emissions have been detected in similar spectral type field
objects (Kirkpatrick et al. 1999; Gizis et al. 2000), and the EWs
of the lines are typically below 10Å. The origin of this feature
is not clear for IPMOs. Actually, the emission could be due to the
presence of mass accretion from a gas-dust disk (Muzerolle et al.
2000). If this is true this fact might be indicating that these
objects have formed in isolation by direct collapse and cloud
fragmentation, and they are not runaways from embryonic multiple
systems. A diagram illustrating the H
behaviour as a function
of (I-J) color is presented in Fig. 4. It seems that Hemission is larger for cooler objects. One of our targets, SOri55,
may have experienced a flare-like episode displaying a considerable
variability in the H
EW (Zapatero Osorio et al. 2001).
Moreover, SOri47 also has a variable H
emission, since we
measure a EW of 25Å on the VLT spectrum while Zapatero Osorio et al. (1999) found an upper limit of 6Å.
|Figure 4: H equivalent width against the color (I-J). Detections and upper limits appear as circles and triangles, respectively.|
|Open with DEXTER|
Using VLT/FORS1 low resolution spectrograph, we have derived spectral types for 15 isolated planetary mass object (IPMO) candidates in the Orionis cluster discovered by Zapatero Osorio et al. (2000). All of these objects but one (SOri61) appear to be bona fide members of the cluster, since they follow a well defined sequence in the magnitude versus spectral type and color diagrams. A significant fraction of the IPMOs presents H in emission, and at least two of them show variability in this activity indicator (SOri47 and SOri55). Since their membership in the cluster is confirmed with our optical spectroscopic data, the masses of these planetary objects must be in the range 18-8 adopting a cluster age of 5Myr. The combined information provided by the color-magnitude and the spectral type-magnitude diagrams indicates that SOri47 is a likely photometric binary. Other two objects (SOri56 and SOri60) might also be photometric binaries. The spectra presented here prove that membership is essentially correct in Zapatero Osorio et al. (2000). Therefore, the substellar mass function derived by Béjar et al. (2001), is not biased by spurious members in the planetary mass domain.
We thanks the ESO staff at Paranal Observatory. The referee, G. Chabrier, has helped to improve the original version with his comments and suggestions. Partial financial support was provided by the Spanish DGES project PB98-0531-C02-02 and CICYT grant ESP98-1339-CO2.