A&A 415, 265-272 (2004)
DOI: 10.1051/0004-6361:20034378
V. Hambaryan1 - A. Staude1 - A. D. Schwope1 - R.-D. Scholz1 - S. Kimeswenger2 - R. Neuhäuser 3,4
1 - Astrophysikalisches Institut Potsdam,
An der Sternwarte 16, 14482 Potsdam, Germany
2 -
Institut für Astrophysik der Universität Innsbruck, Technikerstr. 25,
6020 Innsbruck, Austria
3 -
MPI für extraterrestrische Physik, Gissenbachstrasse 1, 85740 Garching, Germany
4 -
Astrophysikalisches Institut, Universität Jena, Schillergässchen 2-3, 07745 Jena, Germany
Received 23 September 2003 / Accepted 7 October 2003
Abstract
We report on the discovery of a very low mass (VLM)
star in the solar neighborhood, originally identified as an
optical counterpart of a flaring X-ray source detected in the
ROSAT All-Sky survey. Optical spectroscopy and infrared photometry
consistently reveal a spectral type of
and a
distance of
pc. The optical counterpart of
1RXS J115928.5-524717 shows a large proper motion of
/year.
1RXS J115928.5-524717 is the fourth object among the VLM stars displaying a huge
X-ray flare, reaching the unprecedent value of
.
Key words: methods: data analysis - X-rays: stars - stars: low-mass, brown dwarfs - stars: distances - stars: flare - stars: individual: 1RXS J115928.5-524717
Imaging sky surveys performed during the last decade at different wavelengths (optical, near-infrared) have uncovered a large population of very low-mass (VLM) stars and brown dwarf candidates in star forming regions, open clusters and in the general galactic field.
According to an estimate presented by Reid et al.
(2003) the current 10 pc sample is only 75%
complete and the level of incompleteness is somewhat higher in the
southern sky and with the latest type objects.
With spectral types of M 7 and later, these objects, sometimes called "ultracool M dwarfs'', are so faint optically that even nearby ones eluded searches based on the optical, near-infrared and high proper motion sky surveys, and, therefore a detection of a new VLM star in the solar vicinity is of vital importance in the determination of fundamental parameters such as the luminosity function, the mass function, the kinematics of those stars and other properties.
The strength of H-emission increases as cooler,
lower-mass stars are considered (Hawley et al. 1996; Gizis
et al. 2000). The emission is up to 100% at spectral
types M 7-M 9. The observed decline in H
-emission for old (>1 Gyr) L-type dwarfs in the field suggests that the activity
borderline is close to objects of substellar masses (Gizis et al. 2000; Basri 2001; Mohanty & Basri
2003).
Analyzing X-ray properties of VLM stars based on ROSAT observations, Fleming et al. (1993) showed that there is no apparent decrease in X-ray flux (hence coronal heating efficency) down to spectral type M 5, as was suggested by Mullan (1984) analyzing Einstein data.
Flares as another property of chromospheric and coronal activity
represent a phenomenon which is important in ultracool dwarfs,
because at these low temperatures (and masses) these objects are
fully convective, and variability is often attributed to
rotational modulation of star spots produced by magnetic activity.
In solar-type stars it is believed to be due to the so-called
dynamo. This mechanism no longer operates in VLM stars and brown dwarfs, but as these objects are fully convective,
a turbulent dynamo could come into operation (see Chabrier &
Baraffe 2000, and references therein). This means
that the change in interior structure is expected to result in a
change of the field sustaining dynamo, and, therefore X-ray
emission properties of late-type stars.
Flares were reported for a number of VLM stars, observed in ultraviolet, X-ray, and radio frequencies, as well as in optical spectra or via photometry. Bursts of X-ray emission interpreted as coronal flares were detected from the long-known M 8V dwarf vB 10 (Fleming et al. 2000), the M 9V dwarf LHS 2065 (Schmitt & Liefke 2002) by ROSAT and the M 9 brown dwarf LP 944-20 (Rutledge et al. 2000) by Chandra observations.
vB 10 showed a far-ultraviolet flare observed by Linsky
et al. (1995) using the GHRS on the Hubble Space Telescope. Radio emission
has been observed from LP 944-20 and some other ultracool dwarfs
(Berger et al. 2001; Berger 2002).
From the intensity of
the continuous emission and flaring at 8.5 GHz detected in these
objects, Berger (2002) concluded that the emission
mechanism is most likely synchrotron radiation and not of thermal
origin. This indicates the presence of magnetic fields and electron
column densities similar to those inferred for earlier-type
flaring M dwarfs.
![]() |
Figure 1:
![]() ![]() ![]() |
Open with DEXTER |
H-flares were observed on several occasions, e.g. in the
M 8.5 APMPM J2354-3316CM 9.5 (Scholz et al.
2003a) and M 9.5 dwarfs
2MASSW J0149090+295613 (Liebert et al.
1999), BRI 0021-214 (Reid et al.
1999) and even in the cooler L5-dwarf
2MASS J01443536-0716142 (Liebert et al.
2003).
Thus, it has become apparent that the level of quiescent
chromospheric and coronal activity as measured by the H emission line and X-ray fluxes relative to the total bolometric
flux generally declines with spectral type later than M 7.
However, no clear change in flare activity diagnostics is found in VLM stars (Fleming et al. 1995; Mokler & Stelzer 2002; Martín & Bouy 2003).
We performed a systematic search for X-ray variability in the ROSAT All-Sky survey data (Hambaryan et al. 2001). Our technique is customized for the detection of flare-like variability. Here we report on the identification of the optical counterpart of the strong flaring X-ray source 1RXS J115928.5-524717 as a new ultracool dwarf star in the solar neighborhood by optical spectroscopy and astrometry.
During the ROSAT All-Sky Survey (RASS) the whole sky was observed in an unbiased fashion. This, therefore, is well suited to study of variability in different types of celestial objects in the X-ray range (0.1-2.4 keV).
Using a Bayesian change point detection method (see Sect. 3.1), we analyzed the RASS Bright Source Catalog (BSC) for variability. The method works on unbinned data, i.e. the photon arrival times of RASS event tables. We found that out of 18811 BSC sources 642 show significant variations (Hambaryan et al. 2001).
In the framework of this project the X-ray source 1RXS J115928.5-524717 was
identified as variable. Our identification
procedure of optical counterparts of RASS sources typically
consists of several steps and first includes the automatic
extraction of DSS images and a visual search for possible
counterparts in an error circle with radius 25
.
In this
particular case, as one sees in Fig. 1, there is no
unique optical counterpart of 1RXS J115928.5-524717 and our dedicated follow-up
optical spectroscopy in such a case starts with the brightest
object in the X-ray error circle. However, while inspecting
finding charts obtained at different epochs, we found that the
brightest object in the DSS II R-band image (epoch 1991.12, the
closest one to the RASS observation, cf. Table 2) has changed its
position significantly. Further investigation showed that it is a
high proper motion star and also bright on an H
image
taken by the UKST (epoch 1998.32).
Other fainter objects in the X-ray error circle cannot be completely removed from further consideration as the counterpart of the 1RXS J115928.5-524717. Nevertheless, we concentrated in our follow-up study on the brightest one.
The RASS was performed in such a way that the sky was scanned
along great circles with the ROSAT X-ray telescope; the scanning
period was equal to the orbital period, i.e. 96 min.
A given source was scanned for at least 2 days and during a single
scan, it remained in the field of view for typically 10-30 s.
The X-ray source 1RXS J115928.5-524717 is included in the RASS Source Catalogue
(1RXS) with a count rate of
cts s-1 and
was observed during January 9-11, 1991 (ROR number 932622).
The arrival times of photons registered in the RASS during
this observation of 1RXS J115928.5-524717 were extracted
using the position and extraction radius available in the BSC (Voges et al. 1999). The Good Time Intervals
were determined in a manner described by
Belloni et al. (1994), in which the whole extraction
circle (
)
was completely within the field of view
of the detector.
Low-resolution spectroscopy and photometry were performed with DFOSC at the 1.54 m Danish telescope at ESO on July 14 and 17, 2002. The nights were not exactly photometric but sufficient for the spectroscopy. We obtained three spectra of 1RXS J115928.5-524717 with the DFOSC grism #5 and grism #15.
Standard calibration exposures were taken for bias subtraction, flat fielding and flux calibration with spectrophotometric standard stars. Wavelength calibration was done with arc-lamp spectra and additionally inspected with the night-sky lines.
We also obtained two medium-resolution spectra of the likely
counterpart of 1RXS J115928.5-524717 on 4 August, 2002 at the 3.5 m-New
Technology Telescope (NTT) of European Southern Observatory (ESO)
at La Silla with the ESO Multi Mode Instrument (EMMI) in the red
medium dispersion mode. Grating #6 was used, providing a FWHM spectral resolution of 4-5 Å (with the 1.0
slit)
and covered a spectral range of
6000-9100 Å.
All reductions were performed with MIDAS. The CCD 44-82 currently
in use with DFOSC is a thinned chip, so there is strong fringing
seen on the red side of 7000 Å, and it is also visible in
Gunn-i images. For the spectra, the fringing could be
satisfactory removed with the flat-field below 8000 Å. For the
i-band images the dome flat-field correction removed the fringes
completely.
![]() |
Figure 2: The X-ray light curve is displayed for the ROSAT All-Sky survey observation of 1RXS J115928.5-524717. Solid and dashed lines represent exponential decay with and without the data point corresponding to OBI #7, with vignetting-corrected peak count rate = 0.93. |
Open with DEXTER |
In order to study the time behavior of 1RXS J115928.5-524717 we performed a timing analysis of the dataset using a Bayesian change point detection approach developed by Scargle (1998,2000). It is very well suited for a statistical examination when the arrival times of individual X-ray photons are registered (see Hambaryan et al. 1999; Schwope et al. 2002).
The essence of the method is that it subdivides a given data set into intervals with a piecewise constant X-ray count rate according to Poissonian statistics. The application of this procedure to the RASS-detected photons finds three change points, i.e. the data set was decomposed into four observational segments with no variation within them and highly significant variations of count rates between them.
Table 1: ROSAT All-Sky survey X-ray observations of 1RXS J115928.5-524717.
Figure 2 and Table 1 show the RASS X-ray lightcurve of 1RXS J115928.5-524717 in the form of average count-rate for each observational interval. In Fig. 2 the PSPC count rate is plotted vs. time in Julian days. The most dominant light curve feature is a giant X-ray flare that occurred between JD 2448266.0 and JD 2448266.3. Owing to the data gaps and sparse statistics it is not possible to determine the flare onset time.
In total (source + background), 52 photons were registered during the
RASS observations. As a background level we used a value given
in the ROSAT BSC (Voges et al. 1999) for this observation,
estimated for the field of view (vignetting corrected 0.00087 counts/sec/arcmin2).
Thus, from these, 52 photons
are attributed
to the flare of 1RXS J115928.5-524717. They were recorded in the two subsequent
observational intervals OBI # 5 and 6 (cf. Table 1). In
OBI #7 just one photon was registered, while during the next one
(OBI # 8) out of 5 registered photons 3 are attributed to the
source. In order to estimate the intrinsic source count rate and
confidence limits for each observational interval we used a
Bayesian approach for low numbers of counts developed by Kraft et al. (1991). Instead of the widely used approach of simply
subtracting the mean number of background counts from the observed
number of counts it correctly takes into account Poissonian
fluctuations in the number of counts.
![]() |
Figure 3:
ESO 1.54-m/DFOSC spectra ( left panel) of 1RXS J115928.5-524717 compared
with those of LP 944- 20, a known M 9.5 brown dwarf (Tinney
1996,1998; Reid et al. 1995), and the M 7.0
dwarf vB8 (Gizis et al. 2000). An arbitrary constant has
been used to separate the spectra. The location of features
typical of late-type stars are labelled, including metal oxide and
hydride absorption bands. A fragment of the 1RXS J115928.5-524717 ESO NTT
EMMI+Grism#6 spectrum ( right panel) with H![]() |
Open with DEXTER |
In the last two columns of Table 1 a mode of the probability distribution function of vignetting-corrected intrinsic source count rates and credible regions of the source count rates are given. In Fig. 2 these Bayesian credible regions (posterior bubble 0.6827) for each estimated value of the source count rate are plotted.
An estimate of the quiescent X-ray count rate is possible after exclusion of the flare OBIs. Excluding OBIs #5 and 6 and application of the above procedure to all remaining photons gives an upper limit to the count rate of 0.015 s-1, excluding OBI #8 as well gives an upper limit of 0.011 s-1. It should be noted that the source was clearly detected during three observational intervals (OBI # 5, 6 and 8).
Assuming an exponential flare decay we determined the flare decay
e-folding time for two cases, either including or excluding OBI #7. This can be justified by assuming that the flaring structure
is hidden behind the star due to rotation. In the first case the
e-folding time is
days, in the second case it
is 0.07 days.
Despite the observed small number of counts (
and
)
we performed a crude spectral analysis, binning
the data with a constant Signal/Noise ratio of 2, i.e. into bins,
which contained between 3 and 7 (
)
and between 0 and 2 (B)
counts, respectively, in the energy range 0.11-2.46 keV. Note
that no photons were registered above 1.25 keV in the
region.
We fit the resulting spectrum with an assumed Raymond-Smith plasma
model with negligible absorption, implemented in XSPEC version 11.1 (Arnaud 1996). Indeed, applying
from Paresce (1984) and using a distance of 11.1 pc (see 3.2.1) we obtain
.
With this fixed interstellar
absorption parameter the best fit according to the C statistics
gives
(90% confidence) and a time-averaged unabsorbed flux of
.
An application to the data of OBI #5 and #6 separately gives
and unabsorbed
flux of
in the energy range 0.1-2.4 keV for OBI #5 and
and
for OBI #6, respectively.
![]() |
Figure 4: Relationships between spectral type and the spectral index for 4 different spectral indices, showing M 7-L 8 dwarfs (small, filled circles). The lines are second-order polynomial, least-squares fits. The open diamond corresponds to the star vB8, the open triangle to the brown dwarf LP 944-20. The spectral type estimates of 1RXS J115928.5-524717 with filled large symbols are depicted using 1.54 m Danish DFOSC (diamond and triangle) and NTT EMMI (circle) spectra. |
Open with DEXTER |
The spectrum of the proper motion object taken by us with DFOSC is shown in Fig. 3. For comparison we show the spectra of the late M dwarfs vB 8 (spectral type M 7) and LP 944-20 (spectral type M 9.5) taken with the same instrumental set up in the same observing run. The likely counterpart of 1RXS J115928.5-524717 shows similar spectral features as the two template stars.
To assign a spectral type to our object we used two
different approaches based on the strengths of spectral features
and the slope of the spectrum. First we compiled a set of spectral
standards using 74 late type dwarfs (from M 0 to L 8) and
then employed spectral classification methods. In the first
method, spectral indices of individual features (in total 16) were
measured and compared with a standard sequence, following Cruz &
Reid (2002) and Hawley et al. (2002). We
computed their spectral indices to determine the
relationships between spectral type and individual spectral
features, suggested by the same authors. In Fig. 4 some of
these relationships are shown.
The second method, described by Henry et al. (2002), tries to match the overall shape of the observed spectrum to template spectra (see also Hawley et al. 2002).
The method works very well for the known late-type M dwarfs LP 944-20 and vB 8 also observed by us. We could reproduce the published spectral subclass with an uncertainty of half a subclass (see Fig. 4).
For each of these relations one gets an independent estimate of
the spectral class. We take the spectral type to be the average of
those estimates rounded to the nearest half spectral type. The
resulting spectral type of 1RXS J115928.5-524717 determined consistently by both
methods is
.
We identified 1RXS J115928.5-524717 with the Two Micron All Sky Survey source
2MASS 11592743-5247188 with magnitudes
,
and
.
Following Gizis et al. (2000) we estimate MKfrom the
color using the relation
If we instead derive the absolute magnitude of the likely
counterpart of 1RXS J115928.5-524717 via the spectral type,
While the large proper motion of the object was already discovered
during the visual inspection of Digitized Sky Survey data at two epochs only, an accurate proper motion determination was done on
the basis of a larger set of measurements with different epochs
available in SuperCOSMOS Sky Survey (SSS) data (Hambly et al. 2001a, 2001b, 2001c). With a
location in the overlap region between two UK Schmidt telescope (UKST) survey fields we were lucky to identify the object in the
SSS catalogues at eight different epochs with three different
passbands (
,
and
). In
addition to the SSS data, two further measurements from the
SuperCOSMOS H-alpha survey (SHS) (Parker &
Phillips 1998) were available (see
Table 2).
The position of the object in the recently completed all sky data
release of the Two Micron All Sky Survey (2MASS, Cutri et al.
2003) and in the second release of DENIS (DEep
Near-Infrared Survey, Epchtein et al. 1997) now available
at http://vizier.u-strasbg.fr/viz-bin/Cat?B/ denis was included in
the proper motion solution. The 2MASS JHK
and DENIS IJK
photometry is also listed in Table 2.
The most recent epoch astrometry was obtained from DFOSC acquisition images in the R and I bands. The astrometric calibration was done using the ESO Skycat tool with 16 faint SSS reference stars located around the target, which did not show significant proper motions and which were not affected by image crowding in the SSS data. Two independent astrometric calibrations for the DFOSC R and I image yielded very similar results. The average result was used as one additional epoch in the proper motion solution.
Table 2: Positions and photometry of 1RXS J115928.5-524717 measured at different epochs.
Since our object (1RXS J115928.5-524717) is moving across
a crowded field of background stars, some of the SSS measurements
are affected by overlapping images, as can also be noted in
the SSS image parameters. There also seems to be a large error (1
)
in the
coordinate of the DENIS data.
Therefore, we excluded three of the SSS positions and the DENIS position given in Table 2 from the final proper motion
solution, yielding a significant improvement in the proper motion
accuracy (see Table 3).
![]() |
= | ![]() |
|
![]() |
= | ![]() |
Table 3: Proper motion solutions.
![]() |
Figure 5:
The X-ray
emission level of late-type dwarfs in the solar neighborhood (see,
for details text). Filled symbols indicate H![]() |
Open with DEXTER |
Our follow-up optical spectroscopy and proper motion study allowed
us to determine the spectral type, M 9 0.5, and the absolute
magnitude
of the most likely optical
counterpart of the flaring X-ray source. It turned out that this
object is a nearby,
pc, high proper motion,
/yr, previously unknown member of
the solar neighborhood.
In order to compare the X-ray properties of 1RXS J115928.5-524717 with other
ultracool stars we estimated
.
From a
spectral fit (Sect. 3.1) we estimated time averaged and
flare peak X-ray luminosities of
and
,
using
unabsorbed fluxes in the passband 0.1-2.4 keV. The luminosity
uncertainties were determined by a combination of uncertainties of
the distance and of the parameters of the spectral fit.
For an estimate of an upper limit of the X-ray luminosity of 1RXS J115928.5-524717 in its quiescent state we converted from upper limit count
rates (0.011 and 0.015) to fluxes by means of an energy conversion
factor
which was computed by us in XSPEC assuming a
Raymond-Smith plasma (model raymond ) with temperature
and absorption column density
(see also Hünsch et al. 1999; Fleming et al. 2003). Using a
distance of
pc we arrived at an upper limit value
of
for the quiescent X-ray emission of 1RXS J115928.5-524717.
Dahn et al. (2002) list bolometric magnitudes for two M 8.5,
five M 9 and three M 9.5 dwarfs in the immediate solar
neighborhood. An average gives
mag.
This leads to the
.
Therefore, we estimate the ratios
at the flare peak and an
upper limit of
in the
quiescent state which can be compared with measurements of other
late-type dwarfs.
In Fig. 5 we display the dependence of X-ray
activity on spectral type for M dwarfs, using data from the
RASS catalogue of nearby stars (Hünsch et al. 1999).
To derive bolometric magnitudes we used the relation
We also include in this diagram the only four known examples of VLM stars with detected X-ray emission in the solar neighborhood. All these stars have quiescent X-ray fluxes which are compatible with zero. They all show pronounced X-ray flares, the newly discovered object 1RXS J115928.5-524717 being the most luminous among them.
The ROSAT surveys of both field and cluster stars show a maximum
value of about
.
Since this ratio may
be regarded as an efficiency measure of the coronal heating
processes and of the activity related phenomena, the maximum value
indicates a still not well understood saturation
phenomenon.
While the final word on the quiescent X-ray emission down to the bottom of the main sequence at spectral type M 8/9 is not known, (see also the discussions in Fleming et al. 1995,2003; Mokler & Stelzer 2002; Martín & Bouy 2003) the flare observations of the very late M dwarfs suggest that their X-ray luminosity is independent of the spectral subclass.
Any nearby ultracool dwarf, like this new one, is a promising
target for direct imaging searches for sub-stellar companions,
because the primary is nearby (i.e. high physical separation) and
faint (i.e. no overshining companions).
Acknowledgements
We thank our colleagues T. Granzer (Astrophysikalisches Institut Potsdam) and R. Schwarz (Universitäts-Sternwarte Göttingen) for their assistance in the reduction of DFOSC photometric data. We have made use of the ROSAT Data Archive of the Max-Planck-Institut für extraterrestrische Physik (MPE) at Garching, Germany, data products from the SuperCOSMOS Sky Surveys at the Wide-Field Astronomy Unit of the Institute for Astronomy, University of Edinburgh and of the ESO Skycat Tool (version 2.5.3.). We have also used data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. This project is supported in part by the German BMBF under DLR grant 50 OX 0201.