Mon. Not. R. Astron. Soc. 346, 441–446 (2003)
Star-like activity from a very young ‘isolated planet’
J. S. Greaves,1 W. S. Holland1 and Marc W. Pound2 1 UK
Astronomy Technology Centre, Royal Observatory, Blackford Hill, Edinburgh EH9 3HJ
of Astronomy, University of Maryland, College Park, MD 20742, USA
2 Department
Accepted 2003 August 7. Received 2003 August 4; in original form 2002 December 19
ABSTRACT
The discovery of isolated bodies of planetary mass has challenged the paradigm that planets
form only as companions to stars. To determine whether ‘isolated planets’, brown dwarfs and
stars can have a common origin, we have made deep submillimetre observations of part of
the ρ Oph B star formation region. Spectroscopy of the 9-Jupiter-mass core Oph B-11 has
revealed carbon monoxide line wings such as those of a protostar. Moreover, the estimated
mass of outflowing gas lies on the force versus core-mass relation for protostars and protobrown
dwarfs. This is evidence for a common process that can form any object between planetary and
stellar masses in a molecular cloud. In a submillimetre continuum map, six compact cores in
ρ Oph B were found to have masses presently below the deuterium-burning limit, extending
the core mass function down to 0.01 M with the approximate form dN /dM ∝ M −3/2 . If
these lowest-mass cores are not transient and can collapse under gravity, then isolated planets
should be very common in ρ Oph in the future, as is the case in the Orion star formation region.
In fact, the isolated planetary objects that may form from these cores would outnumber the
massive planets that have been found as companions to stars.
Key words: stars: formation – stars: low-mass, brown dwarfs – planetary systems: formation.
1 INTRODUCTION
Compact objects of substellar mass are generally divided into brown
dwarfs and planets. Brown dwarfs, often referred to as ‘failed stars’,
have masses below the mass of approximately 0.075 M (Baraffe
et al. 1998) required for hydrogen-burning, and although initially
sought in young star clusters (Rebolo et al. 1996) are now being
found in the field (Leggett et al. 2002). Extrasolar planets have also
been recently discovered, using the radial velocity technique (Marcy
& Butler 2000) where the stellar velocity is seen to vary periodically due to the gravitational pull of the orbiting planet. The most
massive planets observed exceed 10 Jupiter masses (0.01 M ), and
so a definition was sought dividing these bodies from the lowestmass brown dwarfs. Although not universally accepted, a convenient definition is that a ‘planet’ has a mass below the deuteriumburning boundary which occurs at 12–13 Jupiter masses. Below this
mass, less than approximately half of the initial deuterium is burned
(Burrows et al. 1997). In contrast to these inactive planets, a brown
dwarf has early-time nucleosynthesis of deuterium, while a star is
massive enough for hydrogen nucleosynthesis on the main sequence.
A further part of the division between planets and brown dwarfs
may be their location in space (Najita, Tiede & Carr 2000). The
paradigm is that planets are formed out of gaseous discs around
E-mail: [email protected] (JSG); [email protected] (WSH); mpound@astro.
umd.edu (MWP)
C
2003 RAS
stars, while brown dwarfs form out of cloud cores in the same manner as stars. However, this distinction has recently become blurred.
Several of the radial velocity detections are of companion bodies
with minimum masses that exceed 12–13 Jupiter masses (Udry et al.
2002); indeed, the prototypical brown dwarf Gl 229 B (Nakajima
et al. 1995) is in orbit around an early-M type star. Thus there is
only partially a ‘brown dwarf desert’. i.e. a lack of bodies of tens of
Jupiter masses as companions. Conversely, some faint red objects
seen free-floating in space have masses below the deuterium-burning
boundary. These have been described as ‘isolated planets’ and have
primarily been detected in young star clusters in Orion (Lucas &
Roche 2000; Zapatero Osorio et al. 2000). The exact masses depend on a number of assumed model parameters, but where the
cluster age must be a few Myr or younger, a number of convincing
candidates remain of as little as 3 Jupiter masses (Zapatero Osorio
et al. 2002). Thus, to retain the conventional definition of the origins
of planets and brown dwarfs, it may be necessary to invoke gravitational capture or expulsion: that is, companion brown dwarfs could
have been captured by a star and isolated planets could have been
removed from their parent discs during a stellar close encounter
(Li 2002).
The aim of our survey was to investigate the origins of lowmass bodies by imaging at the youngest evolutionary stage, as in
the early work of Pound & Blitz (1993, 1995). For ‘free-floating’
objects this is the analogous phase to pre-stellar cores and protostars (Ward-Thompson, André & Kirk 2002), which are found in
molecular, optically dark clouds. The lowest-mass cores may then
442
J. S. Greaves, W. S. Holland and M. W. Pound
Figure 1. 850-µm image of the ρ Oph B1 region; east is to the left and
north to the top. The brightest peak is B1-MM3 at 180 mJy beam−1 and
the contours are spaced by 25 mJy beam−1 . The flux peaks are identified by
symbols: small circles are the 11 low-mass compact cores with flux densities
of 34 to 142 mJy beam−1 ; stars show the two cores associated with young
stars; large circles are five extended dust clumps that are larger than the
beam. The bright core B1-MM3 is marked, as is the Oph B-11 core which
lies at 16h 27m 14.0s , − 24◦ 28 31 in J2000 coordinates. The tick marks on
the RA and Dec. axes are spaced every 30 arcsec with respect to this position.
be identifiable as protoplanets or protobrown dwarfs, assuming that
the core does not accrete further gas from the cloud, and that a substantial fraction of the present-day mass ends up in a final compact
object. The advantage of a core survey is that masses may be deduced directly from the optically thin thermal dust grain emission
(Hildebrand 1983), with very little model dependence, and that (unlike in the infrared) the number of additional line-of-sight objects
is very low. In the submillimetre, distant dusty galaxies comprise
the main background and these are fainter (Scott et al. 2002) than
the flux level needed to detect protoplanets. We present the results
of a submillimetre continuum survey in the ρ Ophiuchi cloud, plus
follow-up spectral line observations of one candidate protoplanet;
finally some conclusions are drawn concerning the future numbers
of isolated planets and brown dwarfs.
2 O B S E RVAT I O N S
The continuum observations were made with the SCUBA camera
(Holland et al. 1999) on the James Clerk Maxwell Telescope in 1997
August and 2000 August, and followed-up with spectroscopic observations using the B3 receiver (Cunningham et al. 1998) between
1999 August and 2000 July. The results of 2 h of SCUBA imaging
are shown in Fig. 1; the field of view is 2.7 arcmin at a wavelength
of 850 µm. The spectra of the CO J = 3–2 transition at 345.796
GHz (862 µm) are shown in Fig. 2.
2.1 Continuum data
The SCUBA data were taken in the standard jiggle-map mode
(Jenness et al. 2002) in which the sky signal was subtracted by rapid
Figure 2. CO J = 3–2 spectra in the vicinity of Oph B-11, offset vertically
for clarity. Top, on-source position showing the outflow-like line wings;
middle, position 15-arcsec west with a faint blue wing; bottom, average of
10 other spectra in a 30-arcsec wide grid. The vertical scale is atmosphericcorrected brightness temperature (T ∗A ) and the horizontal scale is velocity in
the local standard of rest frame. The line peak is > +10 K; negative spikes
are an artefact of position-switching on to cloud emission.
chopping to nearby off-source areas. The ρ Oph region is crowded
and, in fact, the large-scale cloud emission could be as bright as
1 Jy beam−1 (Johnstone et al. 2000). It was therefore impossible to
chop to completely blank sky; the chop throw used was ±120 arcsec at an angle of 120◦ east of north in the RA–Dec. coordinate
frame, away from the main cloud ridge. Short-period sky changes
(Archibald et al. 2002) were removed using between three and five of
the 37 850-µm bolometers, in the blankest region of each of the four
frames taken on three different nights. The final reduction steps were
weighting each bolometer according to its intrinsic noise level, and
regridding all the data into an RA–Dec. coordinate frame. The absolute flux of any very extended emission is not preserved due to the
chopping, and the sky subtraction is only approximate with a small
number of bolometers, but these factors are not relevant to the extraction of compact sources. The image was finally smoothed with a
30-arcsec full-width half-maximum Gaussian (twice the beamwidth
of 15 arcsec) and this was self-subtracted to leave only the point-like
and moderately compact cores. This process will affect flux estimates only if the cores possess diffuse extended envelopes; there is
no evidence of this, for example, for the core Oph B-11 discussed
below, which has beam-like dimensions even in the pre-subtraction
image.
The maximum flux before the self-subtraction was 0.58 Jy
beam−1 at 850 µm. Inspection of scan-maps (Johnstone et al. 2000)
suggests that the main B1 ridge was already reduced in flux by approximately 15 per cent due to chopping on to fainter emission at
the off positions. After the subtraction of the smoothed map, the
maximum flux (white in Fig. 1) is 0.18 Jy per 15-arcsec beam. The
standard deviation of 3-arcsec map pixels is 6.3 mJy and the equivalent noise per diffraction-limited beam is 2.7 mJy; the background
level uncertainty is ±5 mJy beam−1 . The fluxes of the compact cores
are taken from the self-subtracted image and are listed in Table 1; all
the cores were detected reproducibly from night to night. The flux
values will be underestimated only if the chop throw fell directly on
to another compact core, and inspection of the scan-maps finds few
such off-field objects albeit for a noise level of ∼10 mJy beam−1
(Johnstone et al. 2000). Calibration used 850-µm skydips (zenith
C
2003 RAS, MNRAS 346, 441–446
Star-like activity of an isolated planet
443
Table 1. Identification of the submillimetre cores. Quantities listed are previous identifications of millimetre and infrared sources, core
positions in J2000 coordinates, flux per beam at 850 µm, dust-plus-gas mass in Jupiter masses, extended or compact structure (see the
text) and references for the source names. Where there are several source names, the core coordinates given above apply. The two stellar
cores are multiple and/or extended in the near-infrared (Allen et al. 2002). Sources are listed in order of decreasing flux density. Masses
are lower limits where the core is more extended than the beam.
Source
identifications
RA (2000)
Dec. (2000)
S 850
(mJy beam−1 )
M core
(M Jup )
Structure
References
Oph B-13
B1-MM3
SMMJ 16272-2429
16h 27m 12.s 7
−24◦ 29 53
177
>40
Extended
16h 27m 12.s 2
16h 27m 15.s 1
16h 27m 11.s 9
16h 27m 11.s 8
16h 27m 10.s 8
16h 27m 17.s 6
−24◦ 30 01
−24◦ 30 20
−24◦ 29 29
−24◦ 29 03
−24◦ 29 23
−24◦ 28 57
Pound & Blitz (1995)
Motte et al. (1998)
Johnstone et al. (2000)
142
132
125
123
107
107
32
>30
>28
>28
>24
>24
Compact
Extended
Extended
Extended
Extended
Extended, stellar
Oph B-4
IRAS 16242-2422
IRAS 16243-2422
IRS 244
B1B2-MM2
IRS 249
Oph B-11
16h 27m 19.s 2
16h 27m 14.s 1
16h 27m 10.s 5
16h 27m 12.s 1
16h 27m 17.s 8
16h 27m 15.s 2
16h 27m 16.s 1
16h 27m 19.s 0
16h 27m 12.s 9
16h 27m 14.s 0
16h 27m 15.s 1
−24◦ 28 44
−24◦ 29 17
−24◦ 28 28
−24◦ 27 54
−24◦ 28 17
−24◦ 29 24
−24◦ 29 49
−24◦ 29 46
−24◦ 28 07
−24◦ 28 31
−24◦ 28 20
opacities of 0.21–0.34), with Uranus and Mars as flux calibrators,
and is accurate to approximately ±5 per cent at 850 µm (Jenness
et al. 2002). A 450-µm image was obtained simultaneously, but since
the atmospheric transmission at the source elevation was 12 per cent
at best, only the main ridge was detected.
64
63
55
54
54
49
47
45
44
39
34
15
14
12
12
12
11
11
10
10
9
8
Compact, stellar
Compact
Compact
Compact
Compact
Compact
Compact
Compact
Compact
Compact
Compact
Pound & Blitz (1995)
Greene & Young (1992)
Motte et al. (1998)
Greene & Young (1992)
Pound & Blitz (1995)
the lower-quality mixer were baseline-subtracted and co-added with
a relative weight of 0.5. Only a linear baseline was required for the
remaining map points, which were observed when the receiver performance was more stable. The rms noise in the on-source spectrum
is 8 mK TA∗ with the data regridded to 5-MHz spectral channels; the
equivalent value 15 arcsec to the west is 7 mK.
2.2 Spectral line data
The J = 3–2 rotational transition of CO was observed towards the
compact source Oph B-11 (described below). The sky emission was
removed by position-switching to an offset of 1200 arcsec east, and
standard three-load calibration was employed; the spectra are presented on the atmosphere-corrected TA∗ antenna temperature scale
and can be converted to main-beam brightness by dividing by 0.63.
The backend was an autocorrelation spectrometer and the total bandwidth was variously 125 and 250 MHz, hence the noise is lower in
the central 125 MHz of the passband (compared with the ends) in
the co-added spectrum. The total on-source time was 2.2 h, and
further spectra were observed around the core for a further 11.5 h.
These spectra covered a grid with a diameter of 30 arcsec and a
cell-spacing of 7.5 arcsec (roughly half of the FWHM beam-size
of 14 arcsec), although only alternate spaces in the grid were filled.
Fig. 2 shows the on-source spectrum, the spectrum 15 arcsec to the
west, and the co-added result of 10 other map points from the grid.
A broad baseline ripple was present in some of the on-source data
from one of the two mixers, and was subtracted in the region away
from the wing velocities. A comparison spectrum with only a narrow line was observed 90 arcsec to the west of the B-11 core, and
used to establish the shape of the polynomial baseline outside the
region ±30 km s−1 . Direct subtraction of this polynomial yielded a
very similar result to fitting the baseline of the on-source spectrum
directly. Once the region of line wings was established, results from
C
2003 RAS, MNRAS 346, 441–446
3 R E S U LT S
Results are presented in the context of previous surveys of ρ Oph.
These include the wide-field 850-µm image of the main star-forming
regions by Wilson et al. (1999), Johnstone et al. (2000) and the
similar survey at 1250 µm by Motte, Andre & Neri (1998), both
of which found numerous cores down to gas masses of ∼0.1 M .
The present SCUBA survey, although of a limited area, extends the
mass range down to 0.01 M . At shorter wavelengths, ISOCAM
7- and 14-µm observations revealed the young stellar population
(Bontemps et al. 2001), K-band spectra have been used to assign
spectral types down to approximately 0.02 M (Luhman & Rieke
1999), and NICMOS 1.1- and 1.6-µm imaging has revealed further
embedded sources down to brown dwarf masses (Allen et al. 2002).
Together these surveys can be used to construct the mass function for
unevolved cloud cores, deeply embedded protostars and young stars
with dispersing circumstellar envelopes. The stars detected in the
near-infrared are thought to be mostly less than 1 Myr old (Luhman
& Rieke 1999), while the youngest protostellar phases typically last
only a few 104−5 yr (Whitworth & Ward-Thompson 2001).
3.1 Census of low-mass cores
The 850-µm image in Fig. 1 is centred between the B1 and B2
regions of star-forming cores (Motte et al. 1998), with the main B1
444
J. S. Greaves, W. S. Holland and M. W. Pound
ridge visible at the lower right. To verify that known cores have been
detected, a comparison was made with the deep DCO+ survey of
Pound & Blitz (1995). Of the five clumps listed by these authors
which lie in the region mapped with SCUBA, three are recovered
as peaks, one lies within the B1 ridge, and one was not recovered
(coincident with the negative map defect near the centre of Fig. 1).
The lowest in submillimetre flux is Oph B-11, which was suggested
as a candidate protobrown dwarf by Pound & Blitz (1995); the
estimated gas mass (dependent on the assumed DCO+ abundance)
was 21 M Jup . This object is still point-like with our 15-arcsec beam
and the identification is unambiguous with only a 5-arcsec position
shift, well within the resolution limits of the original survey. Having
established that the SCUBA map was recovering known compact
cores, this core was selected for spectroscopic follow-up (described
below).
To identify all the cores in Fig. 1, a single contour was plotted
over the image with a steadily reduced flux level (in increments of
≈10 mJy beam−1 ). Closed loops emerging during this process were
then identified as cores, provided that they surrounded a clear peak
(not a trough) and were of diameter comparable to the FWHM beam
size of 15 arcsec. This simple procedure (see also the discussion in
Johnstone et al. 2000) eliminates a priori assumptions concerning
core shapes. The diameter constraint is applied because even a pointlike source should appear as wide as the beam at the half-maximum
flux contour; smaller contours are likely to be chance groups of
pixels with positive noise values. Only one such object was rejected
(as implausibly small), and at ≈27 mJy beam−1 this is fainter than
the peaks adopted as cores. The peaks identified by this process also
have high confidence because they were reproduced from night to
night in the three observing sessions, but it is cautioned that within
the main ridge the flux contrast is low so some pairs of cores are not
well resolved. However, objects such as B-11 (marked in the figure)
are clearly separated; in fact Pound & Blitz (1995) had already found
this source to be isolated in both position and velocity. The list of
map peaks and their fluxes is given in Table 1.
18 cloud cores were identified in total, of which five are still
somewhat extended even after self-subtracting the map. Since these
could be contaminated by residuals of underlying cloud emission,
they may not represent independent self-gravitating regions and so
are eliminated from further discussion. Two further map peaks are
associated with young stars or multiple systems (Greene & Young
1992; Allen et al. 2002) and so probably represent remnant circumstellar gas. This leaves 11 candidates for unevolved cores or
protoplanets/protobrown dwarfs.
3.2 Core masses
Direct mass estimates can be made from the optically thin dust emission. We assume a dust temperature of 12–20 K in this region (Motte
et al. 1998) and a 1250-µm emissivity of 0.01 cm2 g−1 as used for
protostellar cores (Motte et al. 1998), in which a gas-to-dust mass
ratio of approximately 100 is implicitly included. The emissivity
scales as ν β and comparison of the bright core B1-MM3 common
to the 1250- and 850-µm maps gives a temperature-corrected value
(Wilson et al. 1999) of β = 1.3 ± 0.4 and hence an emissivity of
β 850µm = 0.014–0.019 cm2 g−1 . The flux of Oph B-11 is 39 mJy
beam−1 and the corresponding mass of gas and dust at approximately 160-pc distance1 is 9 ± 4 M Jup , where the error is dominated
1
Knude & Hog (1998) find a distance of approximately 120 pc from reddening data towards parts of ρ Oph; adopting this value would reduce the
core masses by a factor of 0.56.
by the temperature uncertainty. The other 10 compact cores have
masses of 8–32 M Jup and half are nominally also below the deuterium burning limit. None of the cores observed is massive enough
to form a hydrogen-burning star.
The mass estimates are most affected by the dust emissivity, for
which the absolute value depends on grain composition and varies by
factors of ∼2 (Ossenkopf & Henning 1994). Motte et al. (1998) have
argued that the emissivity should be a factor of 2 lower for pre-stellar
cores than cores showing signs of protostellar activity; if this is also
true for cores of substellar masses then some of the values listed in
Table 1 should be doubled. However, it is likely that not all of the
present core mass may be accreted on to a final compact object such
as a planet or brown dwarf. Also, the more isolated cores are unlikely
to grow in total mass, i.e. they are assumed to have decoupled from
the general cloud. If the accretion efficiency is analogous to that
for low-mass stars (Matzner & McKee 2000), ∼50 per cent of the
present-day core mass will end up in the compact object. Thus the
possibly higher mass (if the emissivity is in error) will tend to be
cancelled by the loss in mass due to accretion inefficiency.
Given these uncertainties, the mass estimates of Table 1 are likely
to be close to the final masses of the objects formed by the cores. Six
of these cores are presently below a deuterium-burning boundary
set at 12 M Jup . The most massive compact core cannot form more
than a 0.065-M brown dwarf, even increasing the emissivity by a
factor of 2 and assuming perfect accretion efficiency.
3.3 Evolutionary status of Oph B-11
The Oph B-11 core is spatially unresolved, with an estimated maximum radius in the 850-µm map of approximately 6 arcsec (1000
au). The mass of any gravitationally bound core is reflected in the
internal velocity dispersion of its gas, and a gravitationally bound,
isothermal core of mass 9 ± 4 M Jup and 1000 au radius should exhibit a FWHM linewidth (Pound & Blitz 1993) of v = 0.18 ±
0.04 km s−1 . In the previous molecular line survey (Pound & Blitz
1995), v = 0.21 ± 0.07 km s−1 was measured for the DCO+ J =
3–2 transition, consistent with a bound core. The core cannot contain a more massive object such as an invisible young star or brown
dwarf, because the gravitational linewidth would be measurably
increased. Therefore, the data show that Oph B-11 is a strong candidate for an ‘isolated planet’ at a very young age, prior to emerging
in the infrared. Without higher-precision velocity data, we cannot
establish whether this core has some collapse (infall) motions, or is
gravitationally stable, or indeed marginally bound so that it could
be dispersed by bulk motions within the cloud. We therefore go on
to search for signs of evolutionary activity, to test against the latter
possibilities.
If low-mass isolated planets can form by the same process as
young stars, then common phenomena might be expected. These
would include dispersal of a gaseous envelope, the formation of an
accretion disc, and generation of a jet-like outflow. Discs have, in
fact, been detected at ages of ∼106 yr, for example around the 8–
12 M Jup object GY11 in the ρ Oph stellar association (Testi et al.
2002). Outflows are an indicator of a younger age, only ∼104−5 yr,
and are expected to be very faint for substellar objects by simple
scaling (Wolk & Beck 1990). However, it has been speculated that
even bodies as low in mass as Jupiter could have driven an outflow
at early times (Quillen & Trilling 1998).
We therefore searched for an outflow signature from Oph B-11,
mapping the CO J = 3–2 line over a 30-arcsec region. The spectrum
centred on the core (Fig. 2) has both red- and blueshifted line wings
relative to the bulk velocity of the ambient cloud emission. (Note that
C
2003 RAS, MNRAS 346, 441–446
Star-like activity of an isolated planet
the width of the bright central CO line reflects large-scale motions in
the ambient cloud, e.g. de Geus, Bronfman & Thaddeus 1990, and
is also exaggerated in the figure because only the lowest ∼1 per cent
of the intensity scale is plotted. Hence the linewidth appears much
greater than in the DCO+ line, Pound & Blitz 1995, which traces
only the dense gas in the core.)The red plus blue wing emission is
detected at the 9σ level, with Tmb dv = 1.34 ± 0.15 K km s−1 .
Only one other spectrum showed any evidence of wings, with a
4σ
15 arcsec to the west, and
detection of blueshifted emission
Tmb dv = 0.32 ± 0.08 K km s−1 over three spectral channels.
The average of 10 other grid points (lying northeast to northwest
and southeast to southwest of the core) shows no wings. Accidental
position-switching on to slightly redshifted ambient cloud emission
meant that we are unable to identify a possible red-wing counterpart
15 arcsec east of the core.
Other phenomena such as intermittent turbulence (Falgarone et al.
1994) can produce low-level line wings, but in this case red- and
blueshifted emission would be expected at random map positions.
The pattern of both wings centred on the core, a single wing to
one side, and no wings in the orthogonal direction is exactly that
expected of a bipolar outflow. A further test of whether this could
be the protoplanet equivalent of a molecular outflow is to examine
the energetics involved. For protostellar outflows, there is a good
correlation (Hogerheijde et al. 1998) of outflow force, given by
M flow v 2max /r max , with the mass of the circumstellar envelope traced
by the dust emission.
The outflow force has been estimated for Oph B-11 assuming
that the flow covers an area of approximately a telescope beamwidth
(15 arcsec). It was further assumed that the CO emission is optically
thin, the gas excitation temperature is similar to the dust temperature as found for protostellar flows (Hogerheijde et al. 1998), and the
CO:H2 abundance is 10−4 . The total gas mass in the outflow is then
approximately 2 × 10−5 M and the force is 8 × 10−7 M km s−1
yr−1 . This value is plotted together with protostellar values in Fig. 3,
and lies on the same slope observed at the higher masses. Both the
morphological and energetic data therefore support the identification of a protoplanetary outflow.
445
Figure 3. Outflow force plotted against dust flux at 1-mm wavelength.
The cross symbols are previous data (Hogerheijde et al. 1998) in bins of
width 0.5 on a log-Jansky scale. The Oph B-11 flux has been corrected
from 0.85 to 1 mm and from 160- to the 140-pc distance of the other
sources. The dotted vertical error bar represents an estimated up to one
order of magnitude increase in the outflow force when corrected for the
unknown inclination (Cabrit & Bertout 1990); the other data points were explicitly corrected for inclination. The vertical dashed lines mark the stellar–
brown dwarf and brown dwarf–planet boundaries at approximately 0.075 and
0.012 M , for a dust temperature of 20 K.
cretion phase in the ρ Oph cloud (Luhman & Rieke 1999). Thus
given the expectation that outflow acts to halt bulk gas infall and
thus establishes the final mass of the central object (Velusamy &
Langer 1998), the Oph B-11 source may reach its eventual mass at
approximately the same time as young stars in the cloud.
4 DISCUSSION
4.1 Substellar mass function
A number of cores of mass suitable to form brown dwarfs and isolated planets have been detected in the star-forming regions of ρ
Oph. Without further spectral line data we cannot determine how
many of these are gravitationally bound; it is possible that faint
cores could be transient fluctuations due, for example, to turbulence
in the cloud. The study by Pound & Blitz (1995) in fact found a
few examples of unbound cores, in the higher-mass regime above
0.1 M . Only in the best-studied case, Oph B-11, can we conclude
that the linewidth and thermal-emission mass are consistent with a
gravitationally bound core. In this interesting object, we then found
a spectral profile strongly indicative of outflow activity, although the
core mass is only 9 ± 4 M Jup . The grain emissivity is a source of uncertainty in the mass, but we have made the reasonable assumption
that it is equal to values in protostellar cores with similar outflow
activity. The core mass then lies in the isolated planet regime rather
than that of brown dwarfs; the maximum mass of 13 M Jup will not
support substantial deuterium burning (Burrows et al. 1997).
The final mass of the compact object evolving in the Oph
B-11 core may be determined by the outflow, which has an estimated mass-loss rate of 3 × 10−8 M yr−1 . This implies that the
core mass would be reduced by half in approximately 105 yr, which
is also the approximate time-scale for protostars in the steady ac-
The six cores with masses below 12 M Jup comprise a source density of 1 arcmin−2 . For comparison, the previous wide-area ρ Oph
surveys (Motte et al. 1998; Johnstone et al. 2000) identified approximately 50 cores with masses above the hydrogen-burning limit.
These cores lie within six clouds with a total area ≈150 arcmin2 ,
so the corresponding source density is 0.3 arcmin−2 . There is also
a similar number of stars visible in the infrared, with fig. 13 of
Luhman & Rieke (1999) showing 41 stellar objects for the cloud
core region of AV > 50 versus 46 starless clumps to the same approximate extinction (Motte et al. 1998). Thus the cores of isolated
planet mass are roughly three times as common as cores of stellar
mass, or their successors, the young infrared stars.
A more detailed analysis suggests that the power-law slope of
dN /dM ∝ M −3/2 derived from the shallow submillimetre surveys
(Motte et al. 1998; Johnstone et al. 2000) could continue at least as
steeply into the substellar regime. The ratio dN /dM is estimated
at 1–2 arcmin−2 M−1
at 0.1 M , based on the seven to 13 objects
with masses of 0.08–0.12 M found in the previous surveys. Thus
with a slope of −1.5 this ratio is expected to rise to 20–40 arcmin−2
M−1
around 0.015 M . The value calculated from Fig. 1 is 80 ± 25
arcmin−2 M−1
, from the 11 compact cores. This result is somewhat
higher than the expected value and thus consistent with a core mass
C
2003 RAS, MNRAS 346, 441–446
446
J. S. Greaves, W. S. Holland and M. W. Pound
function rising continuously, or even steepening slightly, into the
substellar regime. This result must be regarded as preliminary, as
the slope would flatten if many of the cores were transient features;
the exact value of the slope is also affected by the small-number
statistics. A survey covering a bigger area is needed to fill in the statistical gap between approximately 0.04 and 0.1 M , which suffers
from incompleteness in the shallow surveys and insufficient map
size in the present study.
Recent theoretical models (Padoan & Nordlund 2002) have predicted a mass function for cores assuming they collapse from gravitationally unstable regions created within a turbulent cloud. Since
few of the lowest-mass regions are dense enough to be gravitationally unstable, few substellar objects are produced and the mass
function turns over at approximately 0.4 M . This turnover point
agrees with that derived from infrared star counts in ρ Oph (Luhman
& Rieke 1999), although this is somewhat dependent on cluster age
and completeness. Our new result, showing that in the earlier evolutionary stages there is no turnover in the source counts, is difficult
to explain in terms of current theory unless the great majority of the
faint cores are only transient objects. Even the one object confirmed
here by spectroscopy has strong implications for the mass function: Padoan & Nordlund (2002) find a thousand times fewer 0.01
than 0.4-M objects in the ‘typical’ cloud in their fig. 1. Luhman
& Rieke (1999) observe approximately 10 young stars around this
mass in the entire cloud, hence there should be 1 objects as low
in mass as Oph B-11. Finally, the existence of a generally declining
substellar function is in doubt. For example, Zapatero Osorio et al.
(2002) suggest it is still rising at planetary masses in the σ Orionis
cluster, and also source counts of brown dwarfs very close to the
Sun (d 8 pc) by Reid et al. (1999) indicate a rising power-law
slope of −1 to −2. This would be consistent with a value of ∼ −1.5
over a wide mass regime in ρ Ophiuchus.
5 CONCLUSIONS
A number of convincing candidates for the earliest evolutionary
stages of isolated planets have been identified in the ρ Ophiuchi
star-forming region. If these are mostly gravitationally bound, the
future mass function in this region should rise smoothly from stars
to brown dwarfs to isolated planets, with the latter being the most
numerous. The best studied low-mass core has a molecular outflow
akin to that of a protostar, suggesting that all of these bodies can,
in fact, form via a common process, starting from self-gravitating
cloud regions. Finally, we note that protoplanet candidates are three
time more numerous than their stellar-mass counterparts in ρ Oph,
whereas only one in five of the known planetary systems around
stars has even one planet above 5 Jupiter masses.2 Thus, high-mass
planetary bodies forming in isolated space could easily outnumber
their counterparts formed in discs around stars.
AC K N OW L E D G M E N T S
The James Clerk Maxwell Telescope is operated by the Joint Astronomy Centre, on behalf of the UK Particle Physics and Astronomy
Research Council, the National Research Council of Canada and the
Netherlands Organization for Pure Research. Part of this work was
undertaken by JSG and WSH while at the Joint Astronomy Centre.
2
See table at http://www.obspm.fr/encycl/encycl.html
JSG is currently supported by the Sir Norman Lockyer Fellowship
of the Royal Astronomical Society. We wish to thank Leo Blitz for
early input to this project, and a referee for insightful comments.
REFERENCES
Allen L.E., Myers P.C., Di Francesco J., Mathieu R., Chen H., Young E.,
2002, ApJ, 566, 993
Archibald E.N. et al., 2002, MNRAS, 336, 1
Baraffe I., Chabrier G., Allard F., Hauschildt P.H., 1998, A&A, 337, 403
Bontemps S. et al., 2001, A&A, 372, 173
Burrows A. et al., 1997, ApJ, 491, 856
Cabrit S., Bertout C., 1990, ApJ, 348, 530
Cunningham C.T. et al., 1998, in Phillips T.G., ed., Proc. SPIE Vol. 3357,
Advanced Technology MMW, Radio, and Terahertz Telescopes. Int. Soc.
Opt. Eng., Bellingham, WA, p. 620
de Geus E.J., Bronfman L., Thaddeus P., 1990, A&A, 231, 137
Falgarone E., Lis D.C., Phillips T.G., Pouquet A., Porter D.H., Woodward
P.R., 1994, ApJ, 436, 728
Greene T.P., Young E.T., 1992, ApJ, 395, 516
Hildebrand R.H., 1983, QJRAS, 24, 267
Hogerheijde M.R., van Dishoeck E.F., Blake G.A., van Langevelde H.J.,
1998, ApJ, 502, 315
Holland W.S. et al., 1999, MNRAS, 303, 659
Jenness T., Stevens J.A., Archibald E.N., Economou F., Jessop N.E., Robson
E.I., 2002, MNRAS, 336, 14
Johnstone D., Wilson C.D., Moriarty-Schieven G., Joncas G., Smith G.,
Gregersen E., Fich M., 2000, ApJ, 545, 327
Knude J., Hog E., 1998, A&A, 338, 897
Leggett S.K. et al., 2002, ApJ, 564, 452
Li Z., 2002, ApJ, 574, L159
Lucas P.W., Roche P.F., 2000, MNRAS, 314, 858
Luhman K.L., Rieke G.H., 1999, ApJ, 525, 440
Marcy G.W., Butler R.P., 2000, PASP, 112, 137
Matzner C.D., McKee C.F., 2000, ApJ, 545, 364
Motte F., Andre P., Neri R., 1998, A&A, 336, 150
Najita J.R., Tiede G.P., Carr J.S., 2000, ApJ, 541, 977
Nakajima T., Oppenheimer B.R., Kulkarni S.R., Golimowski D.A.,
Matthews K., Durrance S.T., 1995, Nat, 378, 463
Ossenkopf V., Henning T., 1994, A&A, 291, 943
Padoan P., Nordlund Å., 2002, ApJ, 576, 870
Pound M.W., Blitz L., 1993, ApJ, 418, 328
Pound M.W., Blitz L., 1995, ApJ, 444, 270
Quillen A.C., Trilling D.E., 1998, ApJ, 508, 707
Rebolo R., Martin E.L., Basri G., Marcy G.W., Zapatero-Osorio M.R., 1996,
ApJ, 469, L53
Reid I.N. et al., 1999, ApJ, 521, 613
Scott S.E. et al., 2002, MNRAS, 331, 817
Testi L., Natta A., Oliva E., D’Antona F., Comeron F., Baffa C., Comoretto
G., Gennari S., 2002, ApJ, 571, L155
Udry S., Mayor M., Naef D., Pepe F., Queloz D., Santos N.C., Burnet M.,
2002, A&A, 390, 267
Velusamy T., Langer W.D., 1998, Nat, 392, 685
Ward-Thompson D., André P., Kirk J.M., 2002, MNRAS, 329, 257
Whitworth A.P., Ward-Thompson D., 2001, ApJ, 547, 317
Wilson C.D. et al., 1999, ApJ, 513, L139
Wolk S.J., Beck S.C., 1990, PASP, 102, 745
Zapatero Osorio M.R., Béjar V.J.S., Martı́n E.L., Rebolo R., Barrado y
Navascués D., Bailer-Jones C.A.L., Mundt R., 2000, Sci, 290, 103
Zapatero Osorio M.R., Béjar V.J.S., Martı́n E.L., Rebolo R., Barrado y
Navascués D., Mundt R., Eislöffel J., Caballero J.A., 2002, ApJ, 578,
536
This paper has been typeset from a TEX/LATEX file prepared by the author.
C
2003 RAS, MNRAS 346, 441–446