A&A 376, L17-L21 (2001)
DOI: 10.1051/0004-6361:20011042
T. Shahbaz1 - R. Fender2 - P. A. Charles3
1 - Instituto de Astrofísica de Canarias 38200 La Laguna,
Tenerife, Spain
2 -
Astronomical Institute "Anton Pannekoek'', University of
Amsterdam and Center for High Energy Astrophysics,
Kruislaan, 403, 1098 SJ, Amsterdam, The Netherlands
3 -
Department of Physics & Astronomy, University of Southampton,
Southampton, SO17 1BJ, UK
Received 6 April 2001 / Accepted 19 July 2001
Abstract
We report on low-resolution spectroscopy of GX339-4 during
its current, extended X-ray
"off'' state in May 2000 (r=20.1) obtained with the VLT Focal Reducer/low dispersion
Spectrograph (FORS1). Although we do not positively detect the secondary star
in GX339-4 we place an upper limit of 30 percent on the contribution of a
"normal'' K-type secondary star spectrum to the observed flux. Using this limit for
the observed magnitude of the secondary star, we find a lower limit for the
distance of GX339-4 to be 5.6 kpc.
Key words: stars: individual: GX339-4 stars X-rays: stars accretion, accretion disks black hole physics
The optical counterpart of GX339-4 was identified by Doxsey et al. (1979) as a
blue star. Subsequent observations showed that it exhibited a wide
range of variability depending on its X-ray state; from V=15.4 to 20.2 (Motch
et al. 1985; Corbet et al. 1987) when it is in the X-ray "low'' and "off''
states, while V=16-18 (Motch et al. 1985) when it is in the X-ray "high''
state. Simultaneous optical and soft X-ray (3-6 keV) observations
showed a remarkable anti-correlation during a
transition from an X-ray "low'' to "high'' state (Motch et al. 1985), the cause
of which was unknown. However, Ilovaisky et al. (1986) showed that there are
times when the optical and X-ray fluxes are correlated.
Callanan et al. (1992) reported a
possible orbital period of 14.8 h from optical photometry.
GX339-4 is of great interest in the class of black hole X-ray binaries, because its black hole candidacy is established by its Cyg X-1-like X-ray variability and multiple states, yet it is the only low-mass X-ray binary member in its class to be "steady'' (apart from the occasional "off'' state it is usually X-ray active) as opposed to transient (i.e. systems which undergo episodic X-ray outbursts and which usually last for several months and then are X-ray quiet for many years; see e.g. Charles 1998). This may be related to the mass of the compact object but, at present, there is no dynamical mass estimate available (which would establish its black-hole nature), since there has been no spectroscopic detection of the mass-losing star. This is very difficult during X-ray "on'' states due to the brightness of the X-ray irradiated disc, but GX339-4 occasionally enters an extended "off'' state when the disc contribution is greatly reduced. In this letter we report on VLT medium resolution spectroscopy of GX339-4 taken during its current "off'' state in order to search for the spectral signature of the mass-losing star.
Object | Date | UT | Exposure | Seeing | Comments |
(2000) | start | time (s) | (
![]() |
||
HD 157423 | 05/11 | 08:57 | ![]() |
0.63 | K1III |
HD 155111 | 05/11 | 08:40 | ![]() |
0.68 | K5III |
HR 5265 | 08/23 | 23:07 | ![]() |
0.00 | K3III |
HR 5178 | 08/22 | 08:57 | ![]() |
0.00 | K5III |
HR 6169 | 08/23 | 08:57 |
![]() |
0.00 | K7III |
V821 Ara | 06/04 | 06:52 | 1200 | 0.74 | GX339-4 |
V821 Ara | 06/04 | 07:16 | 1200 | 0.69 | GX339-4 |
![]() |
Figure 1:
R-band image of GX339-4 taken with the VLT on 4th June 2000.
The exposure time was 30 s and the field of view is ![]() ![]() |
Open with DEXTER |
We obtained r-band Gunn images of GX339-4 prior to the spectroscopic
observations for acquisition purposes. The integration time was 30 s and the
images were corrected for the bias level and flat-fielded in the standard way.
We performed optimal aperture photometry (Naylor 1998) on GX339-4, which is
clearly resolved (see Fig. 1), and several nearby comparison stars. The
seeing during these observations was 0.7 arcsecs. We calibrated the
data using photometric standard stars taken on the same night as part of the
ESO calibration programme. We find
for GX339-4.
Acquisition images (see Sect. 2.1) of the GX339-4 field (see Fig. 1),
clearly revealed the star 1.06
arsecs North-West of GX339-4 in the chart published by
Callanan et al. (1992). However, under the excellent observing conditions,
our new image reveals
that this object can actually be resolved into two stars. These stars were close enough
to pose a potential problem in the spectroscopic data analysis, and
so the slit was rotated
to a position angle of 112.8 degrees East of North, to avoid them contaminating
the GX339-4 spectrum.
![]() |
Figure 2: A cut along the slit showing the spatial profile of GX339-4 and the blend of stars B+C. Only rows 1075 to 1080 which are clear from contamination were used in the extraction of the spectrum. |
Open with DEXTER |
The data reduction and analysis was performed using the Starlink FIGARO package, the PAMELA routines of K.Horne and the MOLLY package of T.R. Marsh. Removal of the individual bias signal was achieved through subtraction of a median bias frame. Small scale pixel-to-pixel sensitivity variations were removed with a flat-field frame prepared from observations of a tungsten lamp. Calibration of the wavelength scale was achieved using 4th order polynomial fits. The stability of the final calibration was verified with the OH sky lines at 6300.3 Å and 6363.8 Å whose position was accurate to within 0.6 Å.
One-dimensional spectra were extracted using the optimal-extraction algorithm
of Horne (1986). However, to minimize contamination from the blend of stars B
and C, we extracted only a few rows of the CCD; as shown in Fig. 2, these
were rows 1075 to 1080. The signal-to-noise ratio of the final extracted
GX339-4 spectrum was 50 in the continuum.
![]() |
Figure 3:
RXTE ASM soft X-ray (2-10 keV) light curve of GX339-4 spanning
the period 1 January 1997-1 September 2000 (1 Crab ![]() |
Open with DEXTER |
Wu et al. (2001) have recently suggested that the H
profiles are
different in the high-soft and low-hard states; the high-soft state being
characterized by a double-peaked profile which arises from the irradiated
accretion disc, whereas in the low-hard state the single-peaked profiles arise
from an outflow. However, it should be noted that if there is a bright spot
in GX339-4 then H
emission from this would fill-in the double-peaked
profile arising from the accretion disc, resulting in a single-peaked
profile. A bright spot is a common feature in most of the X-ray transients
(e.g. A0620-00; Marsh et al. 1994).
Phase-resolved high spectral resolution data will be necessary to investigate
this further.
![]() |
Figure 4: From top to bottom: Variance-weighted average spectrum of GX339-4, the blend of stars B+C; K1III, K3III, K5III and K7III template stars. The spectra have been normalized and shifted vertically for clarity. IS and ATM indicate interstellar and atmospheric features respectively. |
Open with DEXTER |
![]() |
Figure 5:
Left: Close-up of the HeI 5876 Å+ NaI interstellar
absorption doublet, 6678 Å and 7065 Å emission lines. The spectra have been
normalized and offset for clarity.
The HeI 5876 Åline is clearly double-peaked. Right: A close-up of
the H![]() |
Open with DEXTER |
The average density of a star that fills its Roche lobe is determined solely
by its orbital period. Assuming that the orbital period of GX339-4 is 14.83
hr (Callanan et al. 1992) and that the secondary star does fill its Roche
lobe, then the mean density of the star is g cm-3. A K1 main
sequence star would have
g cm-3, implying that the radius of
the secondary must be a factor of 1.6 greater than that of a main
sequence star in order for it to fill its Roche lobe. It is therefore most
likely to be evolved, similar to the secondary star in Cen X-4 (Shahbaz et al. 1993).
The GX339-4 spectrum does not show any signs of obvious absorption features
from the secondary star (see Fig. 4). However, it is still possible to
determine an
limit to the contribution of such a star (Dhillon et al. 2000).
This is done by subtracting a constant times the normalized
template spectrum from the normalized GX339-4 spectrum, until spectral
absorption features from the template star appear in emission in the GX339-4
spectrum. The value of the constant
at this point represents an upper
limit to the fractional contribution of the secondary star. The contribution
also depends on the spectral type of the template used. As the spectral type
of the secondary star in GX339-4 is unknown, we used stars in the range
K1III-K7III and find upper limits to the secondary star
contribution to lie in the range 20-30%. It should be noted that giant stars
have weaker metallic absorption lines (near H
)
compared to sub-giant
stars. This implies that our estimates for the secondary star's contribution
are upper limits.
We can place a lower limit to the distance of GX339-4, by comparing
our observed upper limit for the secondary star's magnitude with that
expected for a Roche-lobe filling secondary star in orbit around a
black hole. Using the observed magnitude of r=20.1 (the Hemission line contribution to the observed flux is negligible and so
all the light is assumed to arise from the secondary star) and our
upper limit for the secondary star's contribution to the observed
light, f< 30%, we find that the secondary star must have a r-band
magnitude fainter than 20.4. Using the ellipsoidal model described in
Shahbaz et al. (1993) we determine a lower limit to the
magnitude of the secondary star and the distance to the
source. Although there were no X-ray measurements during the time of
Callanan's optical observations, the X-ray luminosity of GX339-4 in
its "off'' state, within a few years of their observations was
reported to be
erg s-1 (Ilovaisky et al. 1986).
If we assume that the X-ray luminosity of GX339-4 in its
"off'' state was similar, then the optical light curve of a system
with this X-ray luminosity would be dominated by X-ray heating and
thus would appear single-humped and would be modulated on the orbital
period. Hence, the interpretation of Callanan et al. (1992) that the
modulation they observe is the orbital period would seem correct. If
the orbital period is 14.8 hrs (Callanan et al. 1992) and assuming a
binary mass ratio of 10 (black hole mass/secondary star mass), a black
hole of 10
(median mass observed; Miller et al.
1998), an inclination of 15 degrees (Wu et al. 2001) and a mean
secondary star temperature of 4300K [cf. the secondary star in Cen
X-4 (Chevalier et al. 1989) which has a similar orbital period to
GX339-4] and a colour excess of
EB-V=1.2 (Zdziarski et al. 1998)
we find a lower limit to the distance of 5.6 kpc. This value is
consistent with crude distance estimates determined from the systemic
velocity of GX339-4;
kpc (Zdziarski 1998).
The lack of absorption line features in our GX339-4 spectrum is
puzzling. One possibility, that cannot be ruled out is that the
secondary star has a much earlier spectral type. We can use the
average scale height of black hole X-ray binaries and the Galactic
latitude to estimate the distance to GX339-4. Using a scale height of
500pc (White & van Paradijs 1996) and b=-4.3 degrees, the
distance is
7 kpc. At this distance, and given our observed
magnitude, the secondary star would require a spectral type later than F8
(i.e. <6200 K). Note that the optical spectrum of such a star near
H
would appear featureless, i.e. one would not expect to see
strong absorption lines.
Acknowledgements
TS was supported by a EC Marie Curie Fellowship HP-MF-CT-199900297.