A&A 387, 201-214 (2002)
DOI: 10.1051/0004-6361:20020347
F. Haberl1 - C. Motch2 - F.-J. Zickgraf3
1 - Max-Planck-Institut für extraterrestrische Physik,
Giessenbachstraße, 85748 Garching, Germany
2 -
Observatoire de Strasbourg, 11, rue de l'Universite,
67000 Strasbourg, France
3 -
Hamburger Sternwarte,
Gojenbergsweg 112, 21029 Hamburg, Germany
Received 21 December 2001/ Accepted 5 March 2002
Abstract
We report the identification of the ROSAT all-sky survey source 1RXS J154814.5-452845 as
new intermediate polar and present the results from follow-up optical and
X-ray observations. The source shows pulsations with a period of 693 s
both in the optical and X-ray light curves and the detection of a synodic
frequency strongly suggests that this is the rotation period of the white
dwarf. Although the one day aliasing and the sparse optical data
coverage does not allow to unambiguously identify the orbital period,
the most likely values of 9.37 h and 6.72 h add 1RXS J154814.5-452845 to the intermediate
polars with the longest orbital periods known.
The optical spectrum displays features from the late type secondary and
shows the presence of broad absorption lines at
H
and higher order Balmer lines which may be a signature of the
white dwarf atmosphere, very similar to V 709 Cas (RX J0028.8+5917 Bonnet-Bidaud et al. #!2001AetA...374.1003B!#).
The average X-ray spectra as obtained by the EPIC instruments on board
XMM-Newton show hard emission typical for this class of objects but also
the presence of soft blackbody-like emission similar to that seen from soft intermediate
polars and thought to arise from the white dwarf surface heated by the
hard X-rays. The best fit model comprises thermal emission from multi-temperature
plasma in collisional ionization equilibrium with a
continuous temperature distribution up to a maximum of
60 keV,
an Fe fluorescence line at 6.4 keV and with equivalent
width of 260 eV and a blackbody component with kT of 86 eV. The hard X-ray emission
is absorbed by matter covering 47% of the X-ray source with an equivalent
hydrogen density of
1023 cm-2. The remaining hard emission is absorbed
by a much reduced column density of
cm-2 as is the soft blackbody emission.
Pulse-phase spectroscopy around spin maximum and minimum reveals that the flux
variations are mainly caused by a change in the temperature distribution with
higher intensity (a factor of
3 in the 1 keV emission) seen from the lower
temperature plasma during spin maximum. The absorption in the high column
density matter only decreases marginally during spin maximum. The emission
characteristics are consistent with the accretion curtain scenario and features
in the X-ray pulse profiles indicate that we observe one pole of the white dwarf
and our line of sight is nearly parallel to the curtain at spin minimum while
at maximum we have a more direct view to the cooling post shock accretion flow.
Key words: binaries: close - stars: individual: 1RXS J154814.5-452845 - stars: novae, cataclysmic variables - X-rays: stars
The ROSAT Galactic Plane Survey project (RGPS, Motch et al. 1991) to identify X-ray sources detected at low galactic latitudes
(b
< 20) led to the discovery of new cataclysmic
variables (Haberl & Motch
1995; Motch et al. 1996).
Cataclysmic variables (CVs) are close
binary systems in which the low mass secondary star fills
its Roche lobe and transfers matter onto the white dwarf primary.
Depending on the magnetic field strength of the white dwarf the accretion
flows via an accretion disc or directly couples
onto the magnetic field lines of the white dwarf in the case of a strong
magnetic field. Magnetic cataclysmic variables are usually divided
into polars (AM Her systems) in which the strong magnetic field
forces the two stars into co-rotation and intermediate polars with shorter
white dwarf spin periods of typically a few hundred seconds
(IPs, for reviews of this latter sub-class see
Patterson 1994; Hellier 1995, 1999).
IPs accrete from the accretion disc via an azimuthally extended
accretion curtain which causes high photo-electric absorption.
The X-ray spectra of most intermediate polars show a hard,
absorbed bremsstrahlung continuum plus emission lines from highly ionized
iron. This emission originates from the matter which free-falls towards
the white dwarf creating a strong shock near the surface with temperatures kT up to several tens of keV. The hard X-rays heat the white dwarf surface,
creating a soft blackbody-like component as is generally observed in polars
(Beuermann & Thomas 1993) where the accretion stream is of pencil-like
geometry. Additional heating may be caused by dense blobs in the accretion
flow which directly penetrate the white dwarf atmosphere.
The temperature of the emission is thought to decrease with increasing size of
the accretion region on the white dwarf (King & Lasota 1990).
Therefore, the absence of detectable soft X-ray emission in classical IPs
may be explained by a too low temperature and/or suppression of soft X-rays
due to considerable absorption. The discovery of little absorbed,
soft X-ray emission from a small number of "soft'' intermediate polars in
ROSAT data (Mason et al. 1992; Haberl et al. 1994) together with
relatively high magnetic field strengths (e.g. Potter et al. 1997)
which collimates the accretion stream is consistent with this picture.
Here we present the discovery of a new cataclysmic variable. Optical
observations together with detailed X-ray follow-up using ROSAT and XMM-Newton
clearly identify 1RXS J154814.5-452845 as new intermediate polar.
IP | HR1 | HR2 |
1RXS J154814.5-452845 |
![]() |
![]() |
RX J0028.8+5917 |
![]() |
![]() |
RX J1712.6-2414 |
![]() |
![]() |
RE 0751+141 | -0.83/-0.79 | -0.09/-0.15 |
RX J0558.0+5353 |
![]() |
![]() |
ROSAT detected 1RXS J154814.5-452845 during the all-sky survey (Trümper 1982)
with
cts s-1 in the PSPC detector (Pfeffermann et al. 1987).
The source is listed in the bright source catalogue (Voges et al. 1999)
with hardness ratios (defined as
HR1 = (H-S)/(S+H) and
HR2 = (H2-H1)/(H1+H2) where S, H, H1 and H2denote count rates in the 0.1-0.4 keV, 0.5-2.0 keV, 0.5-0.9 keV
and 0.9-2.0 keV bands, respectively) which indicate a moderately hard
X-ray spectrum in the ROSAT band. Table 1 compares the hardness
ratios of 1RXS J154814.5-452845 to four other ROSAT-discovered IPs extracted from the
RRA PSPC source catalogue from
pointed ROSAT observations (ROSAT Consortium 2000). Classical IPs like RX J0028.8+5917 and
RX J1712.6-2414 show high absorption, for which HR1 is very sensitive to
with values approaching +1. The "soft'' IPs RE 0751+14 and RX J0558.0+5353
show little absorption (negative HR1) but also softer intrinsic spectra
(negative HR2) caused by a strong soft blackbody spectral component which
dominates the spectrum in the ROSAT band. The hardness ratios of 1RXS J154814.5-452845
are between those of the two groups, suggesting less absorption than seen
from classical IPs and/or the existence of a soft spectral component.
The source was included in the RGPS bright source sample for optical identification. We observed
the field of 1RXS J154814.5-452845 for the first time on 1997, June 6 with DFOSC and the LORAL
CCD at the
ESO-Danish 1.54 m telescope. The X-ray source was readily identified with a
blue
object exhibiting Balmer, He I, He II and N III-C III
4640-60 lines in emission. We show in Fig. 1
the mean of two 10 mn long spectra obtained with a 2.5 slit through grism #7 (
3860-6810 Å; 10 Å FWHM resolution). Figure 2 shows the position of the
optical counterpart on a V band image.
![]() |
Figure 1: Mean optical spectrum of the optical counterpart of 1RXS J154814.5-452845 as observed in June 1997. |
Open with DEXTER |
![]() |
Figure 2: V band finding chart for 1RXS J154814.5-452845. The positions of the three comparison stars, A, B and C used for differential photometry are also shown. |
Open with DEXTER |
We organized an optical observation campaign in May 1998 involving both photometric and
spectroscopic monitoring at the ESO-Dutch 0.9 m and ESO-Danish 1.54 m telescopes, respectively.
During three consecutive nights from May 7 to 9 UT we obtained optical V band photometry using
the standard CCD camera equipped with the TK512CB # 33 CCD. In order to ensure the best
compromise between time resolution and observing efficiency the CCD was windowed and the exposure
time was set to 30 s. With this instrumental setting, the mean time between two consecutive
exposures was 53 s. A total of 753 frames were acquired over 11.5 h. Raw images were
corrected for bias and flat-field using standard MIDAS procedures. Instrumental magnitudes were
computed using the SExtractor package (Bertin & Arnouts 1996)
for the optical counterpart and 3
comparison stars, named A, B and C whose positions are indicated in Fig. 2. Weather
conditions were not photometric during the first and the third night. However, the accuracy of
the differential photometry remains good with average errors below the 1% level. From standard
star measurements in June 1997 and May 1998 we derive the following magnitudes for the
comparison stars;
,
and
.
The photometric uncertainty is dominated by the colour transformation. The
mean magnitude of 1RXS J154814.5-452845 recorded in May 1998 was V = 14.61, somewhat fainter than in June 1997,
V = 14.46. Spectroscopic observations took place from May 15 till 17 UT. At the ESO-Danish 1.54 m
telescope, we used the same DFOSC setting as in 1997 but with a smaller entrance slit of 1.5. With grism #7 the wavelength range was
3860-6840 Å with a 5.7 Å FWHM resolution). Because of the smaller slit used in 1998 than in 1997, the spectral
resolution improved but no accurate flux calibration could be established. In particular,
differential refraction distorted the shape of the flux calibrated spectrum. The first spectrum
had an exposure time of 20 mn and the remaining 8 were 15 mn long. Arcs from He and Ne lamps
provided the wavelength scale. As for photometric data, MIDAS procedures were used to correct 2-d frames and extract and calibrate spectra in wavelength and flux.
Radial velocities of Balmer and of He II
4686 emission lines vary relatively slowly within a single
night. The total amplitude is 200 km s-1 for H
.
Balmer and He II velocities seem roughly in
phase although there are large differences in mean velocities, higher Balmer lines being more blue
shifted than H
.
We list in Table 2 the measured emission line velocities. In order
to search for orbital motion we computed periodograms for the various lines using the method
outlined in Lomb (1976)
and Scargle (1982)
which is suitable for unequally spaced data.
Spectroscopic data are consistent with a rather long orbital period, larger than 5 h . Because of
the strong one day aliasing and sparse coverage it is not possible to unambiguously derive the
orbital period. We find that the two most probable periods are
h and
h. We show in Fig. 3 the H
radial velocity curve folded with a
period of
h which gives the best sine fit. At this period, He II
4686 velocities
have a more structured pattern of variation. However, the overall shape of the radial velocity
curve changes with period and clearly more spectroscopic data are needed to determine the true
orbital period and velocity structure.
A rather strong indication in favour of a long orbital period is the presence of several metallic
absorption lines in the red part of the spectrum. We compare in Fig. 4
the red part of
the mean optical spectrum obtained in 1998 with a template K4 III spectrum extracted from the
Jacoby et al. (1984)
library of stellar spectra. The resolution of the library spectra
(4.5 Å) compares well with that of our observations. In order to better highlight the
similarities we diluted the late type spectrum by a factor 5 using a constant continuum and plotted
together the rectified spectra normalized to the average continuum. The Mg "b" bands and the
5206, 5270, 6362, 6497 Å atomic line blends are visible as well as the Ca I
6162 Å line. The Na I doublet is well marked, however, part of its strength could be
due to interstellar absorption as indicated by the presence of a relatively deep
5780 Å band. The large absorption at
6280 Å is probably a mixture of telluric O2 and
diffuse interstellar bands. It is not possible to classify the late type spectrum with great
accuracy. Torres-Dodgen & Weaver (1993)
have identified useful lines for classifying stars in the
near-infrared. The absence of TiO bands at
6080-6390 Å indicates a spectrum earlier
than K7V. The marked
6497 Å blend and Ca I line suggests an early K type.
Finally, the presence of the Fe I, Ti I Cr I blend at
6362 Å may
indicate an evolved type. Unfortunately, the absorption line depths on individual 15 mn long
spectra are not strong enough to measure a reliable radial velocity curve of the late type
companion.
Julian date | H![]() |
H![]() |
H![]() |
He II |
-2 400 000.5 d | ![]() |
|||
50948.28693 | +94.1 | +33.8 | +12.0 | +28.9 |
50948.32557 | +46.6 | -30.2 | -34.1 | -9.8 |
50948.36179 | +2.9 | -61.5 | -69.5 | -23.9 |
50948.41196 | -62.4 | -102.8 | -121.1 | -59.5 |
50949.23678 | -99.6 | -141.2 | -167.1 | -119.4 |
50949.29622 | -91.7 | -120.1 | -132.0 | -101.5 |
50949.34037 | -53.3 | -45.7 | -52.4 | -89.3 |
50949.40147 | +41.7 | +43.8 | +11.3 | +17.1 |
50950.41219 | -56.5 | -69.7 | -60.2 | -28.9 |
A spectral type hotter than K7 and possibly early K is compatible with the long orbital period
derived from our spectroscopy. The mean density of the Roche lobe filling star can be expressed as
g cm-3
(Frank et al. 1992).
The shortest
possible period of 6.72 h implies densities comparable to those of a main sequence K5 star. If the
orbital period is 9.37 h a main sequence star would have a G type. Although this cannot be ruled
out completely from our data, the requirement for stable mass transfer of
(Frank et al. 1992)
would rather suggest a slightly evolved K secondary instead.
This picture is consistent with the spectral type distribution of secondary stars in cataclysmic
variables compiled by Smith & Dhillon (1998)
which shows that no system below 4.3 h period has a secondary earlier than K.
Standard evolutionary sequences for unevolved donor stars undergoing
mass-transfer predict a K2 spectral type if the orbital period of
the CV is 6.72 h (Baraffe & Kolb 2000).
Alternatively, a K type secondary
in a 9.37 h period CV would only be possible if the mass donor star is
slightly evolved (Beuermann et al. 1998).
Therefore,
in the absence of confirmed orbital period and secondary mass estimates,
the evolutionary status of the mass donor star cannot be determined.
![]() |
Figure 3:
Example of radial velocity curves: H![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 4: The red part of the averaged spectrum of 1RXS J154814.5-452845 as observed in 1998 (top) plotted together with a template stellar spectrum of a K4 III star diluted with a constant continuum. Several faint metallic lines visible in the late type spectrum also appear in that of 1RXS J154814.5-452845, revealing the late type secondary (see text). |
Open with DEXTER |
Another outstanding feature of 1RXS J154814.5-452845 is the presence of broad absorption lines at H
and
higher order Balmer lines. Troughs are seen up to H
in the blue part of the spectrum
(see Fig. 5). In that respect, the spectrum of 1RXS J154814.5-452845 is strikingly similar to that of
the ROSAT discovered classical intermediate polar V 709 Cas (RX J0028.8+5917 Bonnet-Bidaud et al. 2001).
These authors argue that the broad absorptions are the signature of the non heated part of the
accreting star and derive
and
T = 23 000 K for the DA white dwarf. Similar conclusions
may be driven for 1RXS J154814.5-452845. One should however take this interpretation with some caution since
similar broad Balmer lines in absorption with superposed narrow emissions are sometimes seen from
accretion discs in UX UMa stars (see e.g. and references therein Warner 1995).
We show in Fig. 6
the V light curve of 1RXS J154814.5-452845 for the three consecutive nights of
observations. Significant photometric activity occurs all along the light curve on time
scales of minutes. No conspicuous large long term variations indicating a possible orbital modulation is seen
although the hollow at the beginning of the second night may represent some minimum. In order to
search for coherent modulations we computed the Lomb-Scargle periodogram of the entire time series.
Figure 7 shows the resulting power spectrum. A clear signal is detected at
cycle/day or
s. This period is consistent with the most probable
X-ray period derived from the long HRI observation (
s, see below).
Although the two closest one day aliases of the photometric period cannot strictly be ruled
out on a statistical basis, none of them appears consistent with any of the possible HRI aliases
(see Fig. 8). The natural aliasing frequencies of ground based and space borne
instruments are obviously different. It is therefore very likely that the optical pulsation
occurs at the same frequency as in X-ray and probably represents the spin period of the accreting
white dwarf unless the accretion is disc-less as supposed in RX J1712.6-2414 for instance
(Buckley et al. 1997).
The absence of large power at harmonic frequencies is consistent with the sinusoidal
shape of the folded light curve shown in Fig. 9. The full amplitude of the folded V light curve is 0.036 mag. well in the range of that observed in other intermediate polars
(see e.g. Patterson 1994; Welsh & Martell 1996).
![]() |
Figure 5: Blue part of the average 1998 spectrum of 1RXS J154814.5-452845. The broad absorption features underneath the emission spectrum could be the signature of the white dwarf photosphere. |
Open with DEXTER |
We investigated further the presence of beat frequencies by applying the Lomb-Scargle periodogram
to data with the main modulation at
P = 692.75 s removed. The resulting periodogram, plotted in
Fig. 10, exhibits an excess power in the interval 120 to 125 cycle/day. The
highest peak is at
d-1 (
s). At this given
frequency, the formal probability of the peak to be real is 1.7%
assuming a white noise local distribution. The value of the peak
frequency agrees very well with that expected for the synodic frequency (
d-1) computed assuming the best orbital period of
h, derived from the
optical spectroscopy, and a white dwarf spin period of
s, derived from the
optical photometry. However, an orbital period of
6.7 h cannot be strictly ruled out by
the observations. At the synodic period the light curve is nearly sinusoidal and has a full
amplitude of 0.023 mag. The detection of a synodic frequency confirms that the main X-ray and
optical period is the true rotation period of the white dwarf.
![]() |
Figure 6: Optical V band light curve of 1RXS J154814.5-452845 obtained in May 1998 with the ESO-Dutch 0.9 m telescope. |
Open with DEXTER |
We proposed two observations of 1RXS J154814.5-452845 in the guest observer program of
ROSAT using the HRI detector (David et al. 1996). During the first short observation
in 1996, Sep. 13 (RH400792, net exposure time of 1.97 ks) the average count rate was
cts s-1. It was dedicated to derive an accurate X-ray position:
,
with
an error of
dominated by systematic boresight uncertainty.
After the optical identification as cataclysmic variable and IP candidate
a second ROSAT HRI observation (RH300624) was performed in 1998, Sep. 11-14, with a net
exposure of 17.89 ks. The flux with an average count rate of
cts s-1
was similar to the first HRI observation and also to the survey detection
(the HRI to PSPC count rate conversion factor is 3.0 for a hard spectrum).
The second HRI observation, performed near the end of the ROSAT mission,
suffered from unstable attitude and no accurate position could be derived.
However, the data can be used for a temporal analysis. The HRI light
curve (0.1-2.4 keV) shows sinusoidal intensity variations by a factor
of about two. A Z2 test (Buccheri et al. 1983) yields formally
a period of
s consistent with the value derived from the
optical observations and
confirming 1RXS J154814.5-452845 as IP, however, the large number of short observation
intervals typically lasting less than 2000 s causes many alias peaks.
Periods of 682.2 s, 627.5 s and 618.4 s are of statistically similar
significance, but are inconsistent with any of the possible alias periods
seen in the optical (Fig. 11).
The folded soft X-ray light curve obtained from the HRI data using the 693.3 s
period is presented in Fig. 12.
![]() |
Figure 7:
Periodogram of the V band time series obtained in May 1998 at the
ESO-Dutch telescope. A clear periodic signal at
![]() ![]() |
Open with DEXTER |
![]() |
Figure 8: Detail of the periodogram of the V band time series obtained in May 1998 at the ESO-Dutch telescope. The best optical photometric frequency is consistent with the most likely HRI determination. |
Open with DEXTER |
![]() |
Figure 9: The V band light curve obtained in 1998 folded with the best period of P = 692.75 s. The mean full amplitude of the modulation is 0.036 mag. |
Open with DEXTER |
![]() |
Figure 10:
Periodogram of the 1998 V band time series demodulated from the main
signal at
![]() |
Open with DEXTER |
![]() |
Figure 11: Z2 test of the ROSAT HRI data. |
Open with DEXTER |
![]() |
Figure 12: ROSAT HRI light curve of 1RXS J154814.5-452845 in the 0.1-2.4 keV energy band folded with a period of 693.3 s. |
Open with DEXTER |
XMM-Newton (Jansen et al. 2001) observed 1RXS J154814.5-452845 as part of the
telescope scientist guaranteed
time during satellite revolution 137. The observation (id 0105460301)
on 2000, Sep. 7 lasted for about 20 ks and included the EPIC instruments.
The Reflection Grating Systems were switched off due to technical
problems and the Optical Monitor was not used due to a bright star in the
field. Here we report on the results from the European Photon Imaging
Cameras based on MOS (EPIC-MOS1 and -MOS2, Turner et al. 2001)
and PN CCD X-ray detectors (EPIC-PN, Strüder et al. 2001) which
are mounted behind the three telescopes (Aschenbach et al. 2000). The background in the EPIC detectors was low
during the whole observation with the exception of a short interval of
400 s when it increased from 1-2 cts s-1 to about 10 cts s-1 per PN CCD.
Source and background events were extracted from circular regions with radius
40 around the source position and a nearby source-free area in
the image.
The light curves in the 0.2-8.0 keV band obtained from PN and combined
MOS data is plotted in Fig. 13. Clearly seen are the periodic
700 s intensity variations on top of a more gradual change on a
time scale of about 7000 s. A timing analysis using a Fast Fourier
Transform (FFT) technique shows a single peak at the spin frequency
with no detectable peaks at harmonic or sideband frequencies originating
from the beat between spin and binary orbital period. This is consistent
with a sinusoidal variation as is seen in the ROSAT HRI data
and a long orbital period in excess of the duration of the
observation of
20 ks as is suggested by the optical data. The FFT power
spectrum is shown in Fig. 14. There is some
signature for a modulation on a longer time scale of 4000-8000 s in the
power spectrum as was already seen in the light curve. Such a
variation might be caused by the binary orbit, but the optical data are more
consistent with a longer orbital period.
A
folding analysis was used to derive the best value of
s
for the spin period. For comparison with Fig. 11 the periodogram is
shown in Fig. 15. The contiguous XMM-Newton observation leads to a
unique identification of the spin period, however, the shorter time base of the
observation causes a larger uncertainty.
![]() |
Figure 13:
X-ray broad-band (0.2-8.0 keV) light curves of 1RXS J154814.5-452845 obtained from the XMM-Newton
EPIC PN (top) and summed MOS1 and MOS2 (bottom) data with a
time binning of 40 s. The average background contribution is
![]() ![]() |
Open with DEXTER |
![]() |
Figure 14: Power spectrum derived from of the XMM-Newton EPIC-PN data (0.2-8.0 keV). The vertical lines indicate the frequencies of first, second and third harmonics which are not detectable in the power spectrum. |
Open with DEXTER |
![]() |
Figure 15:
![]() |
Open with DEXTER |
The EPIC-PN data were folded with a period of 692.3 s in different energy bands (Fig. 16). The 0.2-2.0 keV profile is consistent with that obtained from the HRI data (0.1-2.4 keV). The smaller statistical errors in the PN light curve allow to identify two narrow features. A dip before the pulse maximum (or peak on the increasing flank) and an asymmetric minimum are significant. The intensity dip is more prominent at energies 2.0-8.0 keV while the minimum is more symmetric at high energies. The hardness ratio (ratio of the count rates in the 2.0-8.0 keV to the 0.2-2.0 keV band) shows an increase during intensity minimum. The asymmetry in the soft light curve minimum reflects in a deep dip in the hardness ratio, reaching the low level at intensity maximum. The intensity dip is apparently energy independent and not visible in the hardness ratio. The behaviour of increased hardness during intensity minimum is often observed from IPs and is generally explained in the accretion curtain scenario by increased absorption during pulse minimum caused by the curtain. The sharp and short (<0.2 in phase) decline in hardness ratio near intensity minimum then suggests that the line of sight at this phase is nearly parallel to the curtain, allowing an almost free view down to the white dwarf surface. As the white dwarf rotates the curtain quickly intersects the line of sight causing increased absorption. The hardness ratio dip, the sinusoidal pulse variation with a single peak and the narrow optical emission lines suggest that we see emission from one pole only (if we indeed look down nearly parallel to the curtain the opposite pole in a dipole field should be hidden by the white dwarf at this spin phase). One should mention here that EX Hya also shows sinusoidal pulse variations while we see emission from both poles as indicated by, however, broad optical emission lines (Hellier et al. 1987).
![]() |
Figure 16: Folded light curves of 1RXS J154814.5-452845 derived from XMM EPIC-PN data in two energy bands together with their hardness ratio (2.0-8.0 keV/0.2-2.0 keV). The folded light curves are normalized to the average count rate (3.67 cts s-1 in the 0.2-2.0 keV band and 1.84 cts s-1 in the 2.0-8.0 keV band and are plotted versus an arbitrary phase. |
Open with DEXTER |
To further investigate the emission properties of 1RXS J154814.5-452845 spin-averaged energy spectra from the EPIC detectors were accumulated. From the PN data single- (pattern 0) and double-pixel (pattern 1-4) events were selected into two separate spectra to utilize the better energy resolution for single-pixel events. For the two MOS spectra pattern 0-12 events were used. The four spectra were simultaneously fit with the same spectral model only allowing different normalizations between the two MOS and the PN spectra to account for possible inter-calibration problems in absolute fluxes and the different temporal coverage of the instruments (Fig. 13). Spectral response files provided by the hardware teams were used (epn_ff20_sY9_medium.rmf and epn_ff20_dY9_medium.rmf for PN single and double events released in March 2001; m1_medv9q19t5r4_all_15.rsp and m2_medv9q19t5r4_all_15.rsp for MOS released in Feb. 2001). In the past various spectral emission and absorption components were required to fit X-ray spectra of IPs. These include emission from hot thin plasma and soft blackbody radiation (dominating the spectrum of soft IPs in the ROSAT band). In the following different combinations of these models are compared with the observed EPIC spectra. In all models absorption partially covering the X-ray source is applied and an iron fluorescence line is included.
A spectral model consisting of hot thermal bremsstrahlung
suffering absorption by matter partially covering the X-ray source together
with an iron line complex (a fluorescent line from nearly neutral iron
and two lines from Fe XXV and Fe XXVI) as it was representing X-ray spectra from
classical IPs in the past gives no acceptable fit to the EPIC spectra of 1RXS J154814.5-452845.
We used the MEKAL model (Mewe et al. 1985) for thermal
plasma emission from XSPEC 11.0.1 to account for the
bremsstrahlung component and the two lines from highly ionized Fe
(model A, see Table 3). This resulted in a reduced
of 3.5 with a large excess
flux below 0.8 keV.
Improved fit results were obtained by adding a blackbody component to account
for the low-energy excess (model B), improving the reduced
to 1.30.
In this model the low absorption component of the partial absorber was also applied to the
blackbody emission. Around 1 keV the fit is still not acceptable, again showing excess flux.
This excess indicates iron L emission as is also seen in high resolution
spectra from the IP RE 0751+14 = PQ Gem (Burwitz et al. 2001).
For iron L emission a lower temperature of the emitting matter is required
and we added another MEKAL component (soft thermal
component in contrast to the
hard thermal
component) to the low absorbed part of the
model spectrum (model C). This gives an acceptable fit over the whole
energy band with reduced
of 1.26 for 2113 d.o.f.
high absorption | low absorption | temp. | metal | ![]() |
d.o.f.a | ||||||
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
Fe
![]() |
distr. ![]() |
abund. | |||
[1023 cm-2] | [keV] | [1021 cm-2] | [keV] | [eV] | [keV] | [eV] | |||||
A | 1.53 | 10.0 | 0.1 | 10.0 | - | - | 286 | - | 0.34 | 7444 | 2117 |
B |
![]() |
![]() |
![]() |
13.7 | ![]() |
- | ![]() |
- |
![]() |
2748 | 2115 |
C |
![]() |
![]() |
![]() |
14.3 | ![]() |
![]() |
![]() |
- |
![]() |
2657 | 2113 |
D |
![]() |
![]() |
![]() |
59 | ![]() |
- | ![]() |
1.0d | 0.98 +0.21-0.12 | 2552 | 2115 |
E |
![]() |
60c,d |
![]() |
60 | ![]() |
- | ![]() |
![]() |
0.90 +0.10-0.06 | 2544 | 2115 |
For model A no errors were determined because of the unacceptable fit.
a Degrees of freedom. b Parameter linked to temperature of high absorption MEKAL component. c Maximum temperature ![]() d Fixed parameter. |
The co-existence of soft (1 keV) and hard (14 keV) thermal emission indicates
that we actually observe a wide range of temperatures. A multi-temperature
model was used by
Done & Magdziarz (1998)
to fit ASCA and Ginga spectra from BY Cam.
Such a model is available in the XSPEC package, based on the MEKAL
model with emission measures which follow a power-law in temperature
(the emission measure from temperature T is proportional to (
)
).
To fit this model we replaced the two MEKAL components by the multi-temperature component,
suffering absorption again by matter partially covering the X-ray source.
As first approach we fixed the parameter
at 1.0 (model D in Table 3).
The fit further improved to a reduced
of 1.21 for 2115 d.o.f. (the
decreased
by
100 although the number of free parameters was reduced by 2). Alternatively
allowing
as free fit parameter the maximum temperature is not constrained
and was therefore fixed at 60 keV, the value derived with fixed
.
This model (E in Table 3) finally yields the best fit (reduced
of 1.21)
although statistically not different from model D.
The spectra (PN singles+doubles, MOS1 and MOS2) are shown in Fig. 17
together with the best fit model E.
As derived from the normalizations of the differently absorbed multi-temperature emission
components, the high column density matter covers the plasma by 47% while
53% are absorbed with a much reduced column density.
The slope of the optical spectrum indicates that the column density of
the low absorption component, which corresponds to
,
could entirely be of interstellar origin.
The flux determined from the spectra in the 0.2-10.0 keV band was 2.28
erg cm-2 s-1,
2.34
erg cm-2 s-1 and 2.30
erg cm-2 s-1 for MOS1, MOS2 and PN respectively.
PN single and double spectrum were forced to have the same normalization in the fit and
therefore yield the same flux. The fluxes derived from the three detectors
agree remarkably well, it should be noted however that the used data does not
cover identical time intervals. The spectra are not corrected for point spread
function losses outside the source extraction radius. The used radius of
40 should encompass 93% of the flux (Aschenbach et al. 2000) with
little energy dependence.
The intrinsic 0.2-10.0 keV flux (with absorption set to 0) of the multi-temperature emission
component is 3.3
erg cm-2 s-1 and the bolometric blackbody flux is 1.5
erg cm-2 s-1, yielding a
ratio for blackbody to plasma emission of
46%.
The spectra around the Fe K emission line complex are shown enlarged in
Fig. 18. Iron fluorescence, Fe XXV and Fe XXVI lines
are clearly resolved
in the PN single-events and MOS spectra. The fluorescence line energy derived from
the fits is
keV (models C-E). The equivalent width (Fe
)
is listed in
Table 3. The Fe XXV and Fe XXVI line intensities require an
under-abundance of iron in models A-C. We therefore allowed to vary the metal
abundance in the MEKAL
components, but linked all metal abundances in all MEKAL components to one single
free parameter, which is mainly determined by the iron lines. The best fit values for
the metal abundance varies considerably between models B-C with
0.5 solar and the
multi-temperature models D and E where it is consistent with the solar value
(Table 3). For the equivalent hydrogen densities given in Table 3
solar abundances in the absorbing matter are assumed. At least for the high absorption
column, which is intrinsic to the 1RXS J154814.5-452845 system, reduced metal abundances by a factor of
two would increase the hydrogen column density by the same factor.
There is indication for a stronger Fe XXV line than the model predicts.
Producing more He-like iron emission relative to H-like iron emission
requires a lower temperature than the overall fit (which includes the
bremsstrahlung continuum) predicts. Fits to RXTE and BeppoSAX spectra of
IPs extending to
energies up to nearly 100 keV yield reduced temperatures when a hard emission
component as arises from reflection on the white dwarf surface is
included in the spectral model (de Martino et al. 2001).
![]() |
Figure 17: EPIC MOS (blue and light blue) and PN (single-pixel events green, double-pixel events red) spectra of 1RXS J154814.5-452845 together with the best fit model E (see text). The bottom panel shows the residuals. Somewhat larger deviations near 0.5 keV and 2.2 keV may be caused by calibration uncertainties around instrumental absorption edges. |
Open with DEXTER |
![]() |
Figure 18: Enlarged part of Fig. 17 showing the Fe line complex in the EPIC spectra of 1RXS J154814.5-452845. The colour coding is as in Fig. 17. |
Open with DEXTER |
![]() |
Figure 19:
The spectral model components (model C described in the text) compared for spin
minimum (bottom) and maximum (top). Most of the intensity variations are caused by the
soft thermal emission (light blue).
This emission component shows an Fe-L line complex near 1 keV and is reduced
by a factor of ![]() |
Open with DEXTER |
Spectra were accumulated using events from spin phase 0.45-0.95 (spin maximum) and 0.95-0.45 (spin minimum, see Fig. 16). Subtracting the spin minimum from spin maximum spectrum shows an increasing count rate from 0.2 keV up to a maximum around 0.5 keV and then a decline down to 2 keV. This suggests that mainly the two soft components (blackbody and 1 keV MEKAL in model C) change their intensity during the white dwarf spin revolution. In particular the decrease of the difference spectrum towards low energies indicates that absorption changes play a minor role.
Model/ | high absorption | low absorption | temp. | |||||
Spin phase | ![]() |
norm
![]() |
![]() |
norm
![]() |
norm![]() |
norm
![]() |
Fe
![]() |
distr. ![]() |
[1023 cm-2] | [10-3] | [1021 cm-2] | [10-3] | [10-4] | [10-4] | [eV] | ||
C/Min. |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
- |
C/Max. |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
- |
E/Min. |
![]() |
![]() |
![]() |
![]() |
![]() |
- | 277![]() |
0.95![]() |
E/Max. |
![]() |
![]() |
![]() |
![]() |
![]() |
- | ![]() |
![]() |
A fit to spin minimum and maximum spectra using models C and E shows that
temperatures and element abundances do not change. Therefore, these parameters
were fixed at the values derived from the spin averaged spectra. The
normalizations and column densities derived from the fits using models C and E
are compared for
spin maximum and minimum spectra (for the case of PN) in Table 4.
As suggested by the difference spectrum, the column density of the
high absorption component
shows only a marginal change between spin minimum and maximum. The parameters for
absorption and temperature agree within the errors, although there is an
indication for some
decrease during spin maximum.
Also the
values of the
low absorption components are consistent within errors. From the low absorption
components the soft thermal component (model C) shows the strongest
intensity increase
during spin maximum of at least a factor of 2.0 while the blackbody and the hard
thermal components vary at most by factors 1.7 and 1.3, respectively. This is
illustrated in Fig. 19 where the model components are drawn for
spin minimum and maximum. In model E the intensity increase of the soft
thermal emission is reflected in a decrease of the power-law index
of
the temperature distribution.
The equivalent width of the Fe fluorescence line varies very little between spin minimum and maximum. A small decrease during spin maximum is insignificant given the large errors. If the fluorescence line would mainly originate on the white dwarf surface a correlation of its equivalent width with other emission from the surface may be expected. However, the blackbody emission increases during spin maximum suggesting that a large part of the line intensity is probably produced in the cold pre-shock matter of the accretion curtain which strongly absorbs the hard X-ray emission. In this case little changes in the equivalent width are expected as long as the line emitting volume remains visible to the observer. In contrast to 1RXS J154814.5-452845 de Martino et al. (2001) detect variations in the equivalent width of the Fe fluorescence line from RX J0028.8+5917 during a short interval in spin phase (of duration 0.2) which we may miss due to our coarse phase resolution.
Together with the dip in hardness ratios around spin minimum
the spectral variations seen between spin minimum and maximum suggest that we are
viewing the white dwarf nearly pole-on. If the soft thermal component originates
in the accretion curtain below the shock region, the variation in its intensity
would then be caused mainly by the varying aspect angle under which the
curtain is observed during the spin revolution
and suggests beaming of the radiation preferentially out of the sides of the
accretion curtains. This is consistent with the
decrease in covering fraction of the high column density absorber
from 51% at spin minimum to 46% at maximum as calculated from the norm
values
for model E given in Table 4.
Our extensive optical and X-ray observations establish 1RXS J154814.5-452845 as an intermediate polar
with rather unique properties. Optical spectroscopy and photometry indicate an orbital
period of 6.72 h or 9.37 h, probably together with GK Per (47.9 h), V 1062 Tauri
(9.96 h) and AE Aqr (9.88 h) among the longest known from IPs.
The optical spectrum is strikingly similar to that of V 709 Cas (Bonnet-Bidaud et al. 2001)
with broad absorption lines at H
and higher order Balmer lines.
Bonnet-Bidaud et al. (2001) argue that these are produced in the non heated part of the white
dwarf atmosphere. Therefore, 1RXS J154814.5-452845 may be also a rare case where the three constituents
of the system, hot white dwarf, accretion disc and mass donor star can be detected
altogether at optical wavelength. This peculiarity opens prospects for an accurate
determination of system parameters.
All X-ray observations of 1RXS J154814.5-452845 found the source at a similar X-ray flux level. Using the spectral model parameters derived from the EPIC spectra expected ROSAT PSPC and HRI count rates of 0.400 cts s-1 and 0.146 cts s-1, respectively, were simulated. This shows that the source was brightest during the Sep. 1996 observation, a factor of 1.5 brighter than in Sep. 2000 when it was at the lowest observed intensity level.
To fit the X-ray spectra of 1RXS J154814.5-452845 as obtained by the EPIC instruments (0.15-12 keV)
a complex model is required. Thermal emission from a hot multi-temperature plasma
in collisional ionization equilibrium together with absorption by matter partially
covering the hard X-ray source represents the spectrum above 1 keV.
An approach with a two-temperature model suggests that we see a wide range of
temperatures covering at least 1 to 14 keV. High resolution spectra of magnetic
CVs are consistent with a continuous temperature distribution
(Mauche 2001;
Burwitz et al. 2001) as is expected from the cooling matter in the post-shock accretion
column. The fit of a model with a continuous temperature distribution results in a
maximum plasma temperature of keV, relatively unconstrained by the
limited bandpass of the EPIC instruments. The temperature of the post-shock region
expected from the simple shock model
with the mean
molecular weight
and the hydrogen mass m
,
for a 1
white
dwarf with a radius of 0.008
(for a mass radius relation see e.g. Panei et al. 2000) is
60 keV.
However this may be regarded only as upper limit because spectral models including
other hard spectral components result in decreased shock temperatures
(Beardmore et al. 2000). Fitting Ginga LAC spectra of V1223 Sgr with and without
taking into account reflection of the hard bremsstrahlung from the white dwarf
surface these authors found shock temperatures of 30
+12-4 keV and 57 keV,
respectively. Reducing the shock temperature to 30 keV translates to a white
dwarf mass of
0.7
.
A more reliable shock temperature and
from this a better constrained estimate for the white dwarf mass requires the
measurement of the hard X-ray spectrum at energies above 10 keV.
At energies below 1 keV a blackbody component with a temperature kT of 86 eV
is needed to fit the EPIC spectra. The bolometric blackbody flux is
46% of the intrinsic 0.2-10 keV hot plasma emission. This fraction is
lower than that observed from AM Her systems and soft IPs, where the blackbody
flux exceeds the bremsstrahlung emission. A correlation of the flux ratio
with the magnetic field strength at the main accreting pole suggests increasing
suppression of bremsstrahlung in strong magnetic fields
(Beuermann & Burwitz 1995).
The low flux ratio derived for 1RXS J154814.5-452845 is consistent with a magnetic field strength
below 10 MG as generally anticipated for classical IPs.
The temperature of 86 eV is higher than that observed from AM Her systems and
soft IPs which show typical values of 20-40 eV (e.g. Matt et al. 2000)
and 50-60 eV (Duck et al. 1994; Haberl et al. 1994), respectively.
Models for the accretion process onto magnetic white dwarfs predict that the effective
temperature decreases with increasing size of the area covered by accretion on the
white dwarf (King & Lasota 1990). The lack of soft X-ray emission from IPs
supported this view and suggested temperatures too low for soft X-ray emission to be
detected. The case of 1RXS J154814.5-452845 and in this sense also the soft IPs show an opposite
behaviour: the temperature increases from AM Her systems via soft IPs to 1RXS J154814.5-452845, i.e.
from high magnetic field strength (small accretion area) to low magnetic field strength
(large accretion area). This suggests that either 1RXS J154814.5-452845 is a very peculiar case or that
the temperature characterizing the soft emission is determined mainly by other accretion
parameters like the mass accretion rate or mass and radius of the white dwarf.
Hellier (1999) also discussed that blobby accretion im AM Her systems
may lead to a larger footprint on the white dwarf, while in IPs accretion of more
homogenous matter from the accretion disc probably results in a narrow radial confinement
of the accretion curtain, compensating for the larger range in azimuth.
For an absorbing column density of 1023 cm-2 the expected equivalent width of the
Fe fluorescent line should be of the order of 100 eV (Inoue 1985).
It is however not clear if the spherical geometry of the absorbing matter around
the early type star in a high mass X-ray binary as used by Inoue (1985)
can be transfered to the situation
we have in magnetic CVs with a more cylindrical geometry of the accretion stream.
From a tall cylinder with the line of sight along the cylinder axis high absorption
is expected but the cylinder contains relatively little matter to produce the
fluorescence line. In a flat cylinder the absorption is low but the fluorescence
line stronger. In the accretion curtain scenario of IPs the geometry may be closer
to the latter case and indeed be the main reason for the high equivalent width of the
Fe fluorescence line observed in 1RXS J154814.5-452845 and other IPs. To estimate the Fe line
contribution from reflection by cold matter on the white dwarf surface a more
detailed modeling of the Fe line production in the accreting matter is required
and arguments based on the results of Inoue (1985) sometimes used in the literature should be taken with caution.
The spectral analysis of 1RXS J154814.5-452845 has shown a strong model dependence of the derived metal abundances. While one- or two-temperature plasma emission models result in abundances of 50% solar, the models with continuous temperature distribution yield values compatible with solar elemental abundances. This is likely caused by insufficient spectral resolution in particular at low energies which does not allow to properly disentangle line from continuum emission. To constrain elemental abundances of the accreted matter high resolution spectra obtained from grating spectrometers are required and values derived from broad-band, medium resolution spectra should be treated with caution.
The follow-up observations of 1RXS J154814.5-452845 at optical and X-ray wavelength have revealed a new
intermediate polar with interesting properties. The pulsations with a period of 693 s
most likely reflect the rotation period of the white dwarf and the optical data suggest
a long orbital period of more than 6 hours. Optical spectra are consistent with a late
type companion of spectral type K and the presence of broad absorption
features underneath
the Balmer emission lines could be the signature of the white dwarf photosphere, similar to
the classical intermediate polar V 709 Cas. The improved spectral resolution of the
XMM-Newton EPIC instruments together with the large collecting area of the X-ray mirrors
allowed a detailed study of the broad band 0.1-12 keV spectrum. The spectrum can be
modeled by multi-temperature emission from hot plasma originating in the post-shock
accretion region with a maximum temperature of 60 keV, an iron fluorescence line
with an equivalent width of
260 eV and a 86 eV blackbody component. About 47% of
the hard emission is absorbed by matter partially covering the X-ray source with a
column density of 1023 cm-2 while the remaining part and the blackbody emission are
attenuated by a reduced column density of 1.5
cm-2. Intensity variations of
individual spectral components between spin minimum and maximum suggest that the spin
modulation is mainly caused by viewing effects and only little changes in absorption.
The X-ray spectrum of 1RXS J154814.5-452845 and in particular the detection of soft blackbody emission
make this IP an interesting case between soft IPs and classical hard IPs.
Acknowledgements
The ROSAT and XMM-Newton projects are supported by the German Bundesministerium für Bildung und Forschung / Deutsches Zentrum für Luft- und Raumfahrt (BMBF/DLR), the Max-Planck-Gesellschaft and the Heidenhain-Stiftung.