A Constraint on brown dwarf formation via ejection: radial variation of the stellar and substellar mass function of the young open cluster IC2391

Abstract

We present the stellar and substellar mass function of the open cluster IC2391, plus its radial dependence, and use this to put constraints on the formation mechanism of brown dwarfs. Our multiband optical and infrared photometric survey with spectroscopic follow-up covers 11 square degrees, making it the largest survey of this cluster to date. We observe a radial variation in the mass function over the range 0.072 to 0.3Msol, but no significant variation in the mass function below the substellar boundary at the three cluster radius intervals analyzed. This lack of radial variation for low masses is what we would expect with the ejection scenario for brown dwarf formation, although considering that IC2391 has an age about three times older than its crossing time, we expect that brown dwarfs with a velocity greater than the escape velocity have already escaped the cluster. Alternatively, the variation in the mass function of the stellar objects could be an indication that they have undergone mass segregation via dynamical evolution. We also observe a significant variation across the cluster in the colour of the (background) field star locus in colour-magnitude diagrams and conclude that this is due to variable background extinction in the Galactic plane. From our preliminary spectroscopic follow-up to confirm brown dwarf status and cluster membership, we find that all candidates are M dwarfs (in either the field or the cluster), demonstrating the efficiency of our photometric selection method in avoiding contaminants (e.g. red giants). About half of our photometric candidates for which we have spectra are spectroscopically-confirmed as cluster members; two are new spectroscopically-confirmed brown dwarf members of IC2391. Comment: 57 pages, 20 figures, accepted for publication in ApJ

Full text

Page 1

arXiv:0909.0842v2 [astro-ph.SR] 22 Oct 2009
A constraint on brown dwarf formation via ejection: radial
variation of the stellar and substellar mass function of the young
open cluster IC 2391
S. Boudreault and C. A. L. Bailer-Jones
Max-Planck-Institut f¨ ur Astronomie, K¨ onigstuhl 17, Heidelberg, GERMANY 69117
ABSTRACT
We present the stellar and substellar mass function of the open cluster
IC 2391, plus its radial dependence, and use this to put constraints on the for-
mation mechanism of brown dwarfs. Our multiband optical and infrared pho-
tometric survey with spectroscopic follow-up covers 11 square degrees, making
it the largest survey of this cluster to date. We observe a radial variation in
the mass function over the range 0.072 to 0.3M⊙, but no significant variation
in the mass function below the substellar boundary at the three cluster radius
intervals analyzed. This lack of radial variation for low masses is what we would
expect with the ejection scenario for brown dwarf formation, although consider-
ing that IC 2391 has an age about three times older than its crossing time, we
expect that brown dwarfs with a velocity greater than the escape velocity have
already escaped the cluster. Alternatively, the variation in the mass function of
the stellar objects could be an indication that they have undergone mass seg-
regation via dynamical evolution. We also observe a significant variation across
the cluster in the colour of the (background) field star locus in colour-magnitude
diagrams and conclude that this is due to variable background extinction in the
Galactic plane. From our preliminary spectroscopic follow-up to confirm brown
dwarf status and cluster membership, we find that all candidates are M dwarfs
(in either the field or the cluster), demonstrating the efficiency of our photo-
metric selection method in avoiding contaminants (e.g. red giants). About half
of our photometric candidates for which we have spectra are spectroscopically-
confirmed as cluster members; two are new spectroscopically-confirmed brown
dwarf members of IC 2391.
Subject headings: stars: low-mass, brown dwarfs, mass function — open clusters:
individual (IC 2391)

Page 2

– 2 –
1.INTRODUCTION
The origin and evolution of brown dwarfs (BD) remains a fundamental open question.
BDs have masses bridging the lowest mass hydrogen-burning stars and giant planets, so any
picture of star and planet formation is incomplete if it cannot account for BDs. Several for-
mation mechanisms have been proposed, including star-like formation from the compression
and fragmentation of a dense molecular cloud, planet-like formation in a circumstellar disk,
and the dynamical interruption of a star-like accretion process.
There are observational signatures which may be used to distinguish between these
scenarios, such as the distribution of binaries, the presence and properties of circumstellar
disks, the (initial) mass function (MF) and kinematics (see Luhman et al. 2007 for a re-
view of observational signatures on the formation of BDs). Work over the past ten years
has seen considerable success in measuring the MF into the BD regime in several clusters,
including σ Orionis (Gons´ alez-Garc´ ıa et al. 2006; Caballero et al. 2007; Lodieu et al. 2009),
the Orion Nebula Cluster (ONC) (Hillenbrand et al. 2000; Slesnick et al. 2004), IC 2391
(Barrado y Navascu´ es et al. 2004; Platais et al. 2007) and the Pleiades (Moraux et al. 2003;
Lodieu et al. 2007). These are only examples among many other analysis of stellar and
substellar populations.
The comparison of the MF in clusters with different properties (e.g. the different density
clusters Taurus and ONC; clusters with different ages, e.g. Chabrier 2003) has led some work-
ers to draw conclusions about the relative efficiency of possible BD formation mechanisms
(e.g. Kroupa & Bouvier 2003). While some observations (Luhman et al. 2007) and theoret-
ical works (Padoan & Nordlund 2004; Hennebelle & Chabrier 2008) conclude in a common
formation mechanism for BDs and stars, some studies has suggested that BDs could form
by massive disc fragmentation (Stamatellos & Whitworth 2008), photoevaporation of the
accretion envelope (Hester et al. 1996), or interruption of the accretion process (Reipurth
2000; Reipurth & Clarke 2001). For instance, Bate & Bonnell (2005) have performed hy-
drodynamical simulations of star formation from fragmentation of molecular clouds. They
concluded that objects which end up as BDs stop accreting before they reach the hydrogen
burning limit because they are ejected from the dense gas soon after their formation by
dynamical interaction in unstable multiple systems.
This ejection scenario in some cases predicts a higher velocity dispersion and spatial
spread of BDs in comparison to stellar objects, which in turn may be observed as a variation
in the MF with radius (Kroupa & Bouvier 2003). On the other hand, other work have shown
that if BDs are formed by ejection, the velocity distribution could be the same for BDs and
stars (Bate 2009). Muench et al. (2003) observed a radial variation in the MF of IC 348
measured over 0.5 to 0.08M⊙, but no variation was detected in the BD regime. In a study

Page 3

– 3 –
of the spatial distribution of substellar objects in IC 348 as well as Trapezium in the Orion
Nebula Cluster, Kumar & Schmeja (2007) observed the stellar objects to be more clustered
than the substellar ones, which they took as evidence in favour of the ejection scenario. By
looking at the spatial distribution of the Taurus stellar and substellar population, Guieu et al.
(2006) observed a gradient in the BD abundance relative to stars, which they conclude as in
favour of the ejection scenario.
In this paper, we present the results of a program to study, in detail, the MF of one of
the nearest and richest open clusters, IC 2391. This has an age of 50Myr measured from
lithium depletion (Barrado y Navascu´ es et al. 2004) or 40Myr from main-sequence fitting
(Platais et al. 2007). The Hipparcos distance is 146.0+4.8
metalicity and extinction are [Fe/H]= −0.03 ± 0.07 and E(B − V ) = 0.01, (Randich et al.
2001). Because of its proximity and youth, this cluster has been subject to several studies
(e.g. Barrado y Navascu´ es et al. 2004, 2001, 1999; Koen & Ishihara 2006; Siegler et al. 2007;
Platais et al. 2007). Barrado y Navascu´ es et al. (2004) and Dodd (2004) measured the MF
down to the substellar limit, but the MF for confirmed member with a completeness limit
at lower masses has not yet been determined.
−4.5pc (Robichon et al. 1999) and the
Since IC 2391 is not as young as IC 348 (age∼2Myr from Muench et al. 2003) and
Trapezium (age∼0.8Myr from Muench et al. 2002), one may expect it has already lost a
significant fraction of its substellar population to evaporation by dynamical evolution. Using
the tidal radius and mass of IC 2391 estimated by Piskunov et al. (2007) (7.4pc and 175M⊙),
we compute the escape velocity as ve=0.4km/s and the crossing as time tcross=17Myr. (We
stress that the crossing time is just an order-of-magnitude quantity, 2R/v, where we have
adopted for R the tidal radius of 7.4pc from Piskunov et al. 2007 and the velocity dispersion
of ∼0.85km/s from Platais et al. 2007). Assuming a minimum value for the number of cluster
members as the objects reported by Dodd (2004) and Barrado y Navascu´ es et al. (2004) (125
and 33 respectively, together a total lower limit of 158 objects), we estimate that the lower
limit of the relaxation time for this cluster is trelax∼105Myr. Furthermore, in a numerical
simulation of open clusters and the population of BDs members, Adams et al. (2002) shows
that there would still be more than 80% of the original BD population in the cluster even
after about 10 crossing times, assuming that BDs and stars have a similar velocity dispersion.
Therefore, IC 2391 is still young enough for a radial study of its very low mass star (VLMS)
and BD populations, considering the fact that mass segregation occurs on a timescale of
order one relaxation time (Bonnell & Davies 1998; although recent work by Allison et al.
2009 suggest that mass segregation can occur on a smaller time than a relaxation time, at
least for more massive stars).
The paper is structured as follows. We will first present the data set, reduction procedure

Page 4

– 4 –
and calibration in §2. We then discuss our candidate selection procedure in §3, present the
survey results in §4 and then discuss the radial variation of the MF in §5. The preliminary
spectroscopic follow-up is presented in §6 followed by our conclusions in §7.
2. OBSERVATIONS, DATA REDUCTIONS AND CALIBRATIONS
2.1. Observations
The survey consists of 35 34×33 arcmin fields extending to 3 degrees from the center
of the cluster and centered on RA=08:40:36 DEC=-53:02:00 (Figure 1).
fields will be referred to as the deep fields while the other 31 other fields will be referred
as the radial fields and outward fields. (We make a distinction since they were observed
with different exposure times and different filters, as will be specified below). The fields
were chosen to extend preferentially along lines of constant Galactic latitude in an attempt
to reduce systematic errors in any established cluster MF gradient which could arise from
contamination by a Galactic disk population gradient. The total coverage of our survey is
10.9 sq. deg. This compares to 2.5 deg2in the survey of Barrado y Navascu´ es et al. (2001).
The central 4
The optical observations were carried out in four runs with the Wide Field Image (WFI)
on the 2.2m telescope at La Silla (Baade et al. 1999) in : 24 January - 9 February 1999, 20
- 24 January 2000, 10 - 23 March 2007 and 15 - 18 May 2007. The WFI is a mosaic camera
comprising 4×2 CCDs each with 2k×4k pixels delivering a total field of view of 34×33 arcmin
at 0.238 arcsec per pixel. The deep fields in our survey were observed in four medium bands
filters, namely 770/19, 815/20, 856/14 and 914/27 (where the filter name notation is central
wavelength on the full width at half maximum, FWHM, in nm) and one broad band filter,
Rc. The radial fields were observed in Rc, 815/20 and 914/27 while the five outward fields
were not observed in Rc.
These filters were chosen to sample the spectra of late M and early L dwarfs to improve
selection over, say, Rc− Ic, and to minimize the Earth-sky background plus any nebular
emission (as the filters are in regions of low emission).
filters are shown in Figure 2. For all radial fields, we have used an exposure time of 15, 10
and 25 minutes for the 815/20, 914/27 and Rcfilters respectively. For each deep field we
obtained an integration time of 65 minutes for each of 770/19 and Rcand from 50 to 155
minutes for 815/20, from 25 to 261 minutes for 856/14, and 50 to 80 minutes for 914/27. We
additionally obtained short exposures for all fields to extend the dynamic range to brighter
objects. The photometry from the short exposures were combined with the photometry of
our long exposures for our analysis. In order to improve the determination of their low mass
The pass band function for all

Page 5

– 5 –
status (via a better determination of spectral type and luminosity), we also observed all
radial fields, including the outward fields, in the J–band using the Cam´ era PAnoramique
Proche-InfraRouge (CPAPIR) on the 1.5m telescope at Cerro Tololo, Chile (runs on 28
February - 3 March 2007 and 10 March 2007). However, we did not get J–band photometry
for the deep fields. CPAPIR consists of one Hawaii II detector of 2k×2k pixels for a field
of view of 35×35 arcmin with a pixel scale of 1.03 arcsec per pixel. All fields were observed
with a total exposure time of 30 minutes. The J filter of CPAPIR is centered at 1 250 nm
with a FWHM of 160 nm. A detailed list of the fields observed with pointing, filter used,
exposure time and 10σ detection limit, is given in Table 1.
Our 10σ detection limit is J=17.7 and 914/27=20.5 for the radial and deep fields re-
spectively (which corresponds to ∼0.03M⊙ for both cases). However, we can’t expect to
detect all objects down to these magnitudes. The completeness spanning from the brightest
objects without saturation down to the 10σ detection limit is estimated by taking the ratio
of the number of objects detected to the predicted number of detections and assuming a
uniform distribution of stars along the line of sight in the fields of our survey. The predicted
number of detections is derived very simply by extrapolating to the detection limit a linear
fit to the histogram of the number of detections as a function of magnitude (Figure 3). The
completeness of the radial part of the survey is 91.8% while for the deep part it is 82.7%.
The spectroscopic observations were carried out with HYDRA, a multi-object, fiber-
fed spectrograph on the 4m telescope at Cerro Tololo on the nights of 6 and 7 January
2007. Only two fields could be observed: the deep fields 15 and 20 (3 and 2 exposures of
45 minutes respectively). We used the red fiber cable with the KPGLF grating (632 lines
mm−1) and a blaze angle of 14.7◦(no blocking filter was used) and . This gives us a coverage
of 6429–8760˚ A centered at 7593˚ A and a spectral resolution of 4.0˚ A.
2.2. Reduction and Astrometry
The standard CCD reduction steps (overscan subtraction, trimming and flat-fielding
for the WFI data and dark subtraction, flat-fielding for CPAPIR data) were performed on
a nightly basis using the ccdred package under IRAF. For WFI data we used the dome
flat both for pixel-to-pixel variation correction and to correct the global illumination, while
for CPAPIR data we used superflat (obtained by combining science image frames for each
nights). For WFI data, we reduced each of the 8 CCDs in the mosaic independently and
in the final step scaled them to a common flux response level. We made an initial sky
subtraction via a low-order fit to the optical data, and for the infrared data by subtracting
a median combination of all (unregistered) images of the science frames. Images were fringe