A&A 437, 899-904 (2005)
DOI: 10.1051/0004-6361:20053060
M. G. Watson1 - T. P. Roberts1 - M. Akiyama2 - Y. Ueda3
1 - Department of Physics & Astronomy, University of Leicester,
Leicester, LE1 7RH, UK
2 -
Subaru Telescope, National Astronomical Observatory of Japan, 650
North A'ohoku Place, Hilo, HI 96720, USA
3 -
Institute of Space and Astronautical Science, Sagamihara,
Kanagawa
229-8510, Japan
Received 21 December 2004 / Accepted 6 April 2005
Abstract
We report the serendipitous discovery of several
unresolved X-ray sources lying in the prominent spiral arms of the galaxy
KUG 0214-057 in XMM-Newton observations. The location of
these
X-ray sources strongly suggests that at least three, and possibly four, of
these may be physically related to the galaxy. The luminosity of each
of these sources at the distance of KUG 0214-057 is >
(0.3-10 keV), making each a strong candidate
ultraluminous
X-ray source (ULX). Using the ULXs objects as a metric implies that this
relatively low-mass
galaxy may be experiencing rather intense starburst activity. The
serendipitous discovery of these ULXs objects suggests that such objects
are not a negligible component of the overall extragalactic X-ray source
population.
Key words: X-rays: galaxies - galaxies: starburst
![]() |
Figure 1:
a) and b) ( left and centre panels): the soft band (0.5-2 keV) XMM-Newton X-ray
image centred on KUG 0214-057. Data shown is for the 3 EPIC cameras
combined in both colour and contour representations. The
circles, plotted with nominal 5 arcsec radius, mark the 5 sources
discussed in the text, labelled with source identifiers as in Table 1.
Note that source E is strongly saturated in the image shown in a).
The location of the nucleus of KUG 0214-057 is marked with a
diamond in each panel. The X-ray image has been
smoothed with a Gaussian with ![]() ![]() ![]() |
Open with DEXTER |
Perhaps the most interesting result to emerge from Chandra
observations of starburst galaxies has been the detection
of significant
populations of ULXs, many of which have extreme luminosities
i.e. close to or in excess of
,
in the most
active star forming regions of these galaxies (e.g. NGC 3256, Lira et al. 2002; the Antennae, Fabbiano et al. 2001; the Cartwheel, Gao et al. 2003). Whilst it is possible that some of these ULXs may be
powered by IMBHs formed in young, dense star clusters (as now seems
likely for the extreme ULX in M 82; Portegies Zwart et al. 2004), this
formation route is unlikely for the majority of ULXs in these systems,
which are instead likely to be "ordinary'' high-mass X-ray binaries
powered by one or more of the mechanisms outlined above (King 2004).
Regardless of their origin, it is now becoming apparent that the
presence (and integrated luminosity) of multiple ULXs may be a new
gauge of high rates of star formation activity (Grimm et al. 2003; Persic et al. 2004), and indeed the numbers of galaxies
containing ULXs has been demonstrated to rise proportionally with the
star formation rate out to
(Hornschmeier et al. 2004).
Most of the ULXs known have been discovered in pointed observations of
specific galaxies, i.e. essentially in surveys of external galaxy populations,
although their discovery has often been fortuitous. In this paper we report
the serendipitous discovery of several unresolved X-ray sources lying in the
prominent spiral arms of the galaxy KUG 0214-057 in XMM-Newton
observations of this region of sky, made as part of the Subaru XMM-Newton Deep Survey (SXDS). The location of these X-ray sources strongly
suggests that at least three, and possibly four, of these may be physically related to the galaxy
as opposed to foreground or background objects. The luminosity of each of these
sources, if at the distance of KUG 0214-057, is >
,
placing all three towards the high luminosity end of
the ULX regime.
These results thus correspond to the discovery of ULXs as part of an unbiased
extragalactic survey.
Table 1: Summary properties of the X-ray sources in the region of KUG 0214-057.
Here we concentrate on the "SDS-6'' field, which was observed for a total
exposure time 50 ks. The SDS-6 field (OBSID: 0112370701,
observations made on 2002-08-08/ 2002-08-09) contains a small region
notable in having a relatively high local source density. Our attention was
first drawn to this source complex because of the rather striking pattern: a
single bright source surrounded by a partial ring of fainter objects (Fig. 1a,b).
As is
discussed later, there is no reason to suppose that the bright X-ray source
and fainter objects are physically related, but it is very likely that several
of the faint X-ray sources are physically related to a galaxy they are
spatially coincident with.
Table 1 presents the basic parameters of the five X-ray sources detected by
XMM-Newton in this region of sky taken from the complete X-ray source
catalogue for the SXDS (Ueda et al. 2005, in preparation). The catalogue is
based on an analysis of the summed EPIC images, i.e. the EPIC MOS1, MOS2 and
pn data combined; all the results discussed in this paper relate to the
combined EPIC data.
Source
parameters are those derived from the SAS task emldetect which
determines the source parameters by fitting the instrumental point spread
function (PSF) at candidate source positions as part of the standard source
detection and parameterisation approach. All 5 sources reported in Table 1 are detected at relatively high significance (likelihoods 30 in the combined EPIC data; sources C & D have the lowest likelihoods) and are
confirmed by visual examination of the EPIC pn images and in the combined
EPIC MOS images (the lowest significance sources are only marginally
visible in the individual EPIC MOS images).
The count rates
reported are background-subtracted values corrected for vignetting and
correspond to the count rates integrated over the entire instrument PSF.
Figures 1a,b show the
XMM-Newton X-ray image with the source locations overlaid. By
cross-correlating the full set of X-ray sources detected in the SXDS region
with the Subaru optical object catalogue, the relative astrometry between the
Subaru and XMM-Newton astrometric frames has been established to
<1 arcsec. The positions quoted in Table 1 have been corrected for small
systematic shifts in the XMM-Newton astrometric frame, i.e. essentially
to the reference frame defined for the Subaru imaging which has an absolute
astrometric accuracy better than 0.5 arcsec. The statistical error
(1)
on the X-ray source positions is <0.5 arcsec for the brightest
source (E) and ranges from 1.2 to 1.8 arcsec for the four fainter
sources listed. We adopt a 5 arcsec radius for the X-ray source positional
uncertainty.
![]() |
Figure 2: Optical B-band ( top row) and R-band ( bottom row) finding charts for the four ULX candidate X-ray sources (A, B, C, D from left to right) and the background AGN (E). The finding charts are based on the Subaru SuprimeCam SXDS imaging. X-ray error-circles, plotted with nominal 5 arcsec radius, are shown in green. Note that the two right hand images are displayed with different colour table to enhance the visibility of faint objects. |
Open with DEXTER |
The location of the X-ray sources on the optical sky is shown in
Fig. 1c. The optical image shown is the R-band image obtained as
part
of the Subaru SuprimeCam observations of the SXDS. The full SXDS
optical imaging consists of five colours (BVRi'z') and extends to very
faint limits (e.g.
mag; AB; 5
)
with
excellent seeing (
0.8 arcsec) (Furusawa et al. 2005, in preparation).
As can be seen in Fig. 1c, all five X-ray sources lie in the outer regions of the 15 mag galaxy
KUG 0214-057 (Miyauchi-Isobe & Mahehara 1998; KUG 0214-057 =
MCG -01-06-080/ PGC 008726).
KUG 0214-057 is a barred
spiral galaxy at z=0.018 (corresponding to a distance
Mpc), lying in the galaxy group NOG 151 with NGC 881 and NGC 883 (Giuricin et al. 2000). Three of these sources (A, B and C) are located in the outer spiral arms of the galaxy, whilst the
other two (E, D) are outside the visible optical envelope.
The brightest source, E, has been identified with an AGN at z=0.634 as
part of the ongoing optical follow-up of the SXDS X-ray sources (Akiyama et al. 2005, in preparation); the optical counterpart is the brightest object
within the X-ray error-circle shown in Fig. 2. As KUG 0214-057 is
at z=0.018, the X-ray source E is clearly an unrelated background
object.
![]() |
Figure 3:
X-ray colour-colour plots for the five X-ray sources: the
four ULX candidates (red points and error bars) and the source
identified with an AGN (E; blue square, error bars are ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
The location of the remaining four objects is most plausibly explained by them
being located within the galaxy itself. The evidence is arguably strongest for
sources A, B and C which are embedded in regions of the galaxy
rich in knots of nebulosity which, on the basis of their broad-band optical
colours, are very likely to be star-formation regions (see Fig. 2 and Sect. 3.3). The evidence is weaker for D which lies somewhat outside the
visible optical envelope of the galaxy, although not far from the end of the
apparently disrupted spiral arm to the W. If this is correct, the broad band
X-ray luminosities of these sources range from
to
,
making them ULXs with luminosities
towards the high end of those observed in external galaxies.
These X-ray sources might alternatively be foreground or background objects
unrelated to the galaxy. Based on the well-established high Galactic
latitude X-ray
source
relationship, the a priori chance
probability of finding three or four X-ray sources at these fluxes within
1 arcmin radius of the galaxy is <1% and the probability of finding them
in the smaller region defined by the spiral arms even lower. More than 80% of
faint high latitude X-ray sources are AGN, making them the most likely
background sources. At the flux levels of the four sources under discussion,
the AGN counterparts have mean optical magnitude
mag. Although
counterparts at these magnitudes are normally trivially detected in the Subaru
imaging observations, we estimate that the surface brightness inside the
envelope of KUG 0214-057 is high enough to preclude detection of
counterparts
fainter than
22 mag. Thus only a small fraction of potential background
AGN counterparts would actually be detectable.
A second X-ray diagnostic of the nature of these sources is temporal
variability. This is a particularly important test for relatively distant
sources such as these ULX candidates, where the spatial resolution of XMM-Newton corresponds to physical scales of 2 kpc, hence a
detection of variability (particularly on short timescales) strongly favours
an origin for the X-ray emission in a single accreting source as opposed to an
unresolved complex of many sources. However, the limited nature of the raw
data (100-200 counts per ULX candidate from all three EPIC detectors
combined) precludes all but the simplest of variability tests. Comparative
photometry of the four ULX candidates, after splitting the observation into
three equal-length intervals (
15.9 ks apiece), revealed no significant
variability. A Kolmogorov-Smirnov (K-S) test against the hypothesis of a
constant source flux for the time series data divided into 5 ks intervals
indicated no significant variability in three of the sources, but source B was found to have a
probability of being variable (
significance). Hence at least one of the ULX candidates shows some
temporal evidence of being a single, accretion-powered source.
The X-ray properties of sources A & C and in particular B appear
consistent with known ULXs. Source D has slightly unusual X-ray colours,
perhaps consistent with a buried young supernova remnant, which would make
it a very rare source - only 10 such sources with peak X-ray
luminosities in the ULX regime have ever been observed (Immler & Lewin 2003;
this rarity also argues that the chances of the other ULX candidates
also being such sources is very low indeed). There is of course still a
possibility that some of the candidate ULXs are no more than a
superposition of less-luminous sources clustered within the
5 arcsec diameter XMM-Newton beam (
1.8 kpc at the distance of KUG 0214-057). However, this would require a remarkable density of X-ray
sources - for illustration, the surface luminosity density of the 43 detected X-ray
sources in the central 1-arcmin (
5.5 kpc diameter) region of the
Antennae, calculated from the catalogue of Zezas et al. 2002 expressed as
/area, is
.
In KUG 0214-057 one would require a higher density of
for any of our candidate ULXs to be a
superposition of many fainter sources. Hence the most likely
interpretation must be that we are viewing individual, accretion-powered
ULXs.
![]() |
Figure 4:
FOCAS optical spectrum of brightest knot in nebulosity within
error-circle of source B (see Fig. 2). The bright emission lines are
identified as H![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
KUG 0214-057 was not detected by IRAS and hence there is no mid-IR data to
constrain the star formation rate. Similarly, there are no corresponding
measurements of the flux in H,
UV or radio continuum, hence these
classic methods for deriving a star forming rate are also not available.
However, recent work has shown that the integrated X-ray luminosity of
luminous high mass X-ray binaries in star forming galaxies can also be a
measure of star formation rate (see e.g. Grimm et al. 2003).
We therefore estimate an X-ray star forming rate in the following manner. We
assume a starburst-like X-ray luminosity function slope of
(Kilgard et al. 2002) for sources in KUG 0214-057, and normalise this to the
integrated flux of the four detected sources (
)
over the
range (i.e.
between the minimum detected luminosity of the ULXs, and the apparent break
luminosity for galaxian luminosity functions noted by Grimm et al. 2003). We then integrate below this range (i.e. between
)
to estimate the "missing'' flux from discrete
sources below our detection limit
,
which totals
.
Hence we estimate a total X-ray flux from luminous point-like X-ray
sources of
.
From Fig. 7 of Grimm et al. (2003) this converts to a reasonably high star formation
rate in the range of
yr-1.
To put this star formation rate in perspective, we estimate the mass of
KUG 0214-057 using the stellar M/L correlation coefficients of Bell & De Jong (2001).
In particular, we follow Colbert et al. (2004) and use coefficients
from the formation with bursts model, and we use optical/IR magnitudes from NED of B = 15.26 (Maddox et al. 1990) and K = 13.16 (2MASS measurement).
These convert to give
,
and hence
.
Finally, by normalising the star formation rate to the galaxy mass we find
that KUG 0214-057 has a rate of
4 (in units of
),
which from Table 1 of Grimm et al. (2003) is very similar to the
rate per unit mass in the archetypal local starburst M 82 (
3.6).
However, in contrast to M 82, the locations of both the ULXs and optical knots
demonstrate that the star formation activity appears to be prevalent
throughout the spiral arms of KUG 0214-057, rather than dominant in the
nucleus. This is another indicator that the high star formation rate is due
to an interaction: similar star formation morphology and ULX locations are
seen in more nearby interactions, e.g. in NGC 4485/90 (Roberts et al. 2002)
and M 51 (Terashima & Wilson 2004). However, we note that both the star
formation rates and the integrated and peak luminosities of the X-ray source
populations are far higher in KUG 0214-057. Indeed, Ptak & Colbert (2004)
estimate that only
of galaxies possess one or more ULXs with
;
KUG 0214-057 potentially has
four. This high incidence of ULXs is only bettered by the most intense local
starbursts, usually in galaxy collisions e.g. NGC 3256 (Lira et al. 2002), the
Cartwheel (Gao et al. 2003).
Acknowledgements
This work is based in part on data collected at Subaru Telescope, which is operated by the National Astronomical Observatory of Japan, financially supported in part by a Grant-in-Aid for Scientific Research (No. 15740126) by the Japanese Ministry of Education, Culture, Sports, Science and Technology.