Issue |
A&A
Volume 512, March-April 2010
|
|
---|---|---|
Article Number | A22 | |
Number of page(s) | 12 | |
Section | Stellar structure and evolution | |
DOI | https://doi.org/10.1051/0004-6361/200811319 | |
Published online | 24 March 2010 |
Cassiopeiae: an X-ray Be star
with personality![[*]](/icons/foot_motif.png)
R. Lopes de Oliveira1 - M. A. Smith2 - C. Motch3
1 - Universidade de São Paulo, Instituto de Astronomia, Geofísica e
Ciências Atmosféricas, R. do Matão 1226, 05508-090 São Paulo, Brazil
2 - Catholic University of America, 3700 San Martin Drive, Baltimore,
MD 21218, USA
3 - Université de Strasbourg, CNRS, UMR 7550, Observatoire
Astronomique, 11 rue de l'Université, 67000 Strasbourg, France
Received 10 November 2008 / Accepted 6 January 2010
Abstract
An exciting unsolved problem in the study of high energy processes of
early type stars concerns the physical mechanism for
producing X-rays near the Be star Cassiopeiae. By now
we know that this source and several ``
Cas analogs''
exhibit an unusual hard thermal X-ray spectrum, compared both to normal
massive stars and the non-thermal emission of known Be/X-ray binaries.
Also, its light curve is variable on almost all conceivable timescales.
In this study we reanalyze a high dispersion spectrum obtained by
Chandra in 2001 and combine it with the
analysis of a new (2004) spectrum and light curve obtained by XMM-Newton.
We find that both spectra
can be fit well with 3-4 optically thin, thermal components
consisting of a hot component having a temperature
-14 keV,
perhaps one with a value of
2.4 keV, and two with well defined
values near 0.6 keV and 0.11 keV.
We argue that these components arise in discrete (almost monothermal)
plasmas. Moreover, they cannot be produced within an integral gas
structure
or by the cooling of a dominant hot process. Consistent with earlier
findings, we also find that the Fe
abundance arising from K-shell ions is significantly subsolar and less
than the Fe abundance from L-shell ions.
We also find novel properties not present in the
earlier Chandra spectrum, including a dramatic
decrease in the local photoelectric absorption of soft X-rays, a
decrease in the
strength of the Fe and possibly of the Si K fluorescence
features, underpredicted lines in
two ions each of Ne and N (suggesting abundances that
are
1.5-
and
solar, respectively),
and broadening of the strong Ne X Ly
and O VIII Ly
lines.
In addition, we note certain traits in the
Cas spectrum
that are different from those of the fairly well studied analog
Cas - in this sense the stars have different
``personalities.''
In particular, for
Cas
the hot X-ray component remains
nearly constant in temperature, and the photoelectric absorption of the
X-ray plasmas can change dramatically. As found by previous
investigators of
Cas,
changes in flux, whether occurring slowly or in rapidly evolving
flares, are only seldomly accompanied by variations in hardness.
Moreover,
the light curve can show a ``periodicity'' that is due to the presence
of
flux minima that recur semiregularly over a few hours, and which can
appear again at different epochs.
Key words: stars: emission-line, Be -
stars: individual: Cas
- X-rays: stars
1 Introduction
The observed properties of Cas
(B0.5 Ve; mV = 2.25)
in ultraviolet to infrared wavelengths have
led to major discoveries related to the Be-phenomenon since its
discovery as the first of what became known as ``Be stars'' (Secchi 1867). However, its X-ray
emission is not typical of this kind of object (White
et al. 1982).
The X-ray emission of
Cas
is dominated by a hot plasma component (
-12 keV), with mildly
high luminosity (
1032-33 erg s-1)
and variable flux (Smith
et al. 2004, and references therein).
Be stars at large are softer X-ray emitters (
keV) with lower
luminosity (
1032 erg s-1)
and little variability (e.g., Berghöfer
et al. 1997). In contrast, most of the well
investigated
Be/X-ray binary systems are Be+neutron star systems and show a high
luminosity (
1033 erg s-1)
and a distinctly nonthermal high
energy distribution. Notably, no large X-ray outburst has been
observed in
Cas.
This is at variance with the behavior witnessed in many classical
Be/X-ray systems.
Cas
is now known to be in a 204-day binary, with an
eccentricity variously determined to be near e = 0
and 0.26 (Harmanec
et al. 2000; Miroshnichenko et al. 2002).
Little is known about the secondary, except
that it is likely to have a mass in the range 0.5-2
.
Recently, X-ray properties similar to those of
Cas
have been observed in a small but growing number of Be stars: the
Cas-like
stars (Motch
et al. 2007; Lopes de Oliveira et al. 2006;
Lopes de
Oliveira 2007).
A conclusive explanation for the X-ray emission of Cas
and its analogs is lacking.
The first suggestion to explain the hard X-ray emission of
Cas
was that it is powered by accretion of matter from the Be wind or disk
onto
an accreting neutron star (White
et al. 1982). More recently, some have advocated the
presence of an accreting white dwarf companion. This suggestion
comes from several X-ray characteristics of cataclysmic variable
systems (e.g., Murakami
et al. 1986; Kubo et al. 1998),
such as their thermal nature and only moderate X-ray luminosity.
However,
the analogy is incomplete upon further scrutiny, and X-rays would have
to come from accretion onto a white dwarf representative of a new class
of Be/X-ray binaries (Lopes de Oliveira
et al. 2007) and have a high
efficiency of mass accretion energy to X-ray flux.
A second proposed scenario suggests that the X-ray emission of
Cas
is a consequence of magnetic interaction between its stellar surface
and a Be (Keplerian decretion) circumstellar disk that entrains a
magnetic field. This idea is based in part on the correlations of the
UV and optical variabilities with the X-ray light curve and also
indirect evidence of magnetic field observed in this star. This
evidence takes the form
of migrating subfeatures running blue-to-red through optical and UV
line profiles (Smith
et al. 1998b; Yang et al. 1988) and
also the discovery of
a gray, robust, 1.21581-day feature in the star's light curve. This
periodicity must be very near or identical to the star's rotation
period (Smith et al. 2006).
The inference of a disk connection is
based on the cyclical aspect and reddish tinge of the optical cycles
that
correlate with the X-ray ones. It is also supported by patterns of
variability in the C IV and Si IV lines that are
occasionally seen in the
X-ray light curve (Cranmer et al.
2000).
One piece of evidence for this interaction comes from the observation
of highly redshifted absorption UV lines,
suggestive of material being ejected from the circumstellar environment
toward the star with enough energies to produce
10 keV X-rays when they impact the star (Smith & Robinson 1999).
Both the magnetic disk and accretion interpretations have
important astrophysical implications.
The former suggests that disk dynamos operate and also that these stars
are possible proto-magnetars. The
accretion interpretation would imply the presence of a neutron star
in an unusual accretion regime, or a white dwarf with novel
properties. The presence of
a white dwarf is still speculative, but such objects are predicted to
be common as secondaries in models of evolution of binaries with B
primaries. The resolution of the mystery of the production of X-rays in
`` Cas
stars'' could lead to a breakthrough in either of these fields.
We have focused our efforts to try to understand the origin of
the X-rays of Cas
and analog Be stars. In this paper we confine ourselves to the X-ray
properties of
Cas
from high and medium resolution spectroscopy and timing studies using
data obtained by XMM-Newton satellite. Chandra
HETG data are also reinvestigated.
2 Previous
X-ray observations of
Cas
Several Rossi X-ray Timing Explorer (RXTE)
observations of the light curve of Cas detailed by
Smith et al. (1998a)
and Robinson & Smith (2000)
(hereafter ``SRC98'' and ``RS00'', respectively) have disclosed that
the light curve undergoes
variations on rapid (flaring), intermediate (several hours) and long
(2-3 month cycles) timescales. We summarize the flaring
results first as follows:
- flares (shots) are ubiquitous, except during brief periods of very low X-ray flux. The reoccurrence of these lulls is often cyclical;
- individual flare profiles are narrow and symmetrical in shape;
- collectively flares show a log-normal distribution in energy;
- the flares show an approximately 1/f distribution down to the photon limit of the instrument (for RXTE about 4 s). Occasional groups of flares (aggregates) last as long as a few minutes, and indicate the end of this red noise dependence.
- spectrophotometry of shots indicates that their flux is usually difficult to distinguish from that of the underlying basal flux. That is, the two components have about the same high temperature of 10-12 keV.

Long-term variations have also been reported with cycle lengths of 50-91 days and amplitudes of a factor of three (Robinson et al. 2002, ``RSH02''). These are so far well correlated with corresponding cycles in optical bands (Johnson B and V). Although the optical amplitudes are 100 times smaller than the X-ray ones they are still much larger than in terms of luminosity output and therefore cannot be the result of reprocessing of X-rays. Rather, a common mechanism appears to mediate both variabilities.
Modulations with several hours timescales were suggested by
previous X-ray satellites.
Frontera et al. (1987)
reported a modulation with timescale of 1.67 h from EXOSAT observation
on 1984 December 7, while Parmar
et al. (1993) argued from a reanalysis of this data
that such an oscillation arise from statistical fluctuations in the red
noise spectrum of the source. Also, Parmar
et al. (1993) have not found periodic oscillation in
other EXOSAT data obtained on 1984 December 25-26.
A
2.3 h
oscillation was suspected by Haberl
(1995) from ROSAT observation on 1993 July 16-17.
Owens et al. (1999)
found weak evidence for a
0.6 h
and 2.6 h in BeppoSAX observation of
Cas
carried out on 1998 July 20-21. None of these suspected detections
could be found by subsequent monitorings
of the star with the RXTE (e.g., Smith et al. 2000).
![]() |
Figure 1:
a) RGS1+RGS2 light curve combining 1st and
2nd orders and the RGS1 background light curve, and EPIC pn light
curves in the b) 0.8-2 keV and
c) 2-10 keV X-ray bands, and the respective
d) hardness ratio. Time bins of 100 s. Upper limits
for error bars at 1 |
Open with DEXTER |
Pre-high resolution spectroscopy of Cas has
demonstrated that its X-ray
emission is dominated by a thermal component with
-12 keV
(Murakami
et al. 1986; Parmar et al. 1993; Kubo et al.
1998; Owens
et al. 1999). These spectra exhibited
Fe XXV and Fe XXVI Ly
K line
strengths consistent with this temperature, but they are consistent
only with a subsolar Fe abundance of 0.2-0.4
.
Prior to this paper, the only high dispersion spectrum of Cas
was obtained
on 2001 August 10
by the Chandra High Energy Transmission Grating
(HTEG) discussed
by Smith et al. (2004)
(hereafter ``S04''). This spectrum
showed a complex structure caused by plasma radiating in at least
3-4 temperatures ranging from about 12 keV to about
0.15 keV. The hot component in turn consisted
of two subcomponents with very different (
1023 cm-2
and
cm-2)
column densities. While the Fe K
lines again indicated abundances of only
other Fe lines arising from the L-shell give abundances which
were consistent with the solar value. The strengths of other lines were
also consistent with solar abundances.
3 Observations and data reduction
Cas
was observed by the XMM-Newton X-ray observatory on
2004 February 5 during about 68 ks in the
revolution 762 (ObsId 0201220101). This observation was
performed with the EPIC pn camera running
in the fast timing mode (timing resolution
of 0.03 ms), connected to the thick
optical filter, and with
the high spectral resolution RGS1 and RGS2 cameras. The optical monitor
was blocked as usual for bright optical sources, and the central CCDs
in the MOS1/2 cameras were switched off in order to avoid overloading
the telemetry.
Data reduction has been made using the Science Analysis System (SAS)
software v8.0.1. All data were reprocessed using the pipeline EPPROC
(for EPIC pn camera) and RGSPROC
(for RGS cameras)
tasks. For the timing analysis, we use the Z2n
Rayleigh (Buccheri et al.
1983) and the Scargle/Midas (Scargle
1982) peridograms, and the Xronos
package.
For spectral fits, the XSPEC
software v11.3.2 was applied.
The EPIC pn data obtained in the fast
timing mode have a doubtful calibration and increased noise especially
at softer energies (<0.5 keV; Guainazzi
2008) and, to be on the safe side, timing analysis was
restricted to broadbands at the 0.8-10 keV energy range
without relevant loss of informations. For spectral studies, since we
have high quality spectra from RGS1 and RGS2 cameras covering the soft
X-rays (6.2-38 Å; or
0.3-2 keV), we use the EPIC pn
data as complementary data covering the hard part of the spectrum
(1.25-4.1 Å; or
3-10 keV).
We noticed that the inclusion of the low energy part of the
EPIC pn data produces systematic residuals
that can be connected to calibration uncertainties.
Because of the high count rate of
Cas, the
inclusion of times with slight background flares during the XMM-Newton
observation has little impact
on its EPIC pn and RGS light curves. However, we
opted for excluding these times in the spectral analysis. The resulting
exposure times in the flare-free good time intervals were
48 ks for the RGS1 and RGS2 cameras and 51.2 ks for
EPIC pn.
We reinvestigated the Chandra data of Cas
obtained on 2001 August 10 with the HETG during about
52 ks (ObsID 1895) with the CIAOv4.0, reported by Smith et al. (2004). Our
purpose was to compare the Chandra and XMM-Newton
spectra of
Cas
from the same analysis techniques and thus to insure that
any differences between the derived parameters are not due to the model
fitting programs. Our reanalysis of the Chandra
spectrum with XSPEC and families of mekal models
gives results consistent with those reported by S04, obtained
from Sherpa and APEC models.
4 Timing studies
The soft (0.8-2 keV) and hard (2-10 keV) light curves of



![]() |
Figure 2: Power spectrum. Top: from EPIC pn events at 0.8-10 keV, rebinned as a geometrical series. Bottom: from 0.8-10 keV EPIC pn data binned to 100 s; the dashed lines represent the confidence levels. |
Open with DEXTER |
These findings are similar to those reported by SRC98, though RS00 found that shot fluxes tended to exhibit a lower (softer) ratio.
Figure 2a exhibits the power spectrum of the EPIC pn events from the combined spectral bands shown in Fig. 1. A key point from this figure is that at frequencies above about 0.003 Hz the slope of the distribution is significantly steeper than -1. This was also noted by SRC98 and RS00, who found slopes of -1.23 and -1.36 in 1996 and 1998, respectively. The difference between those values was already marginally statistically significant, and the current result is certainly even more significantly different from the 1996 result. As noted by RS00, the difference between these slopes and -1 is most likely due to a relative prevalence of strong, longer-lived flares as compared to rapid short ones in these data. RS00 reported 1/f slopes in their 1998 light curves that were intermediate between the 1996 slope of -1 and the present steeper one. The explanation for the knee at about 0.003 Hz is again likely to be caused by the absence of discrete flares with timescales longer than a few minutes.
![]() |
Figure 3: Autocorrelation from 0.8-10 keV light curve binned to 500 s. P1 refers to result of Fig. 2b. |
Open with DEXTER |
The break is due to the dominance of apparently random variations that often occur in the X-ray light curve on timescales of about a half hour or longer. At least some of these appear to be connected with variations in the UV continuum, which Smith et al. (1998a) have associated with the partial occultation of the star by rotationally advected translucent corotating clouds.
Another interesting feature emerging from the autocorrelation
analysis of a light curve obtained by the RXTE satellite (Fig. 3) was the
reappearance of periodic flux lulls.
Following the procedure of Robinson
& Smith (2000)
we have cross-correlated both the fluxes in our XMM-Newton
light curve and also the reciprocals of these fluxes. The latter
exercise was carried out because RS00 had found that
the autocorrelation of the direct fluxes produced no distinguishable
features, whereas the autocorrelation of the reciprocal fluxes produced
dramatic variations with lags of spacing of 7.5 h and
higher multiples. These authors reasoned that this different behavior
could come about if X-ray active centers at various latitudes of the
rotating star would not produce marked features in the autocorrelation
curve. However, features would be observed from the curve generated
from
the reciprocal fluxes if only a few longitudes of the star were not
covered
with X-ray activity. Our results, shown in Fig. 3 for the
direct flux, produced rather similar curves for both
the direct and reciprocal fluxes, indicating that the distribution of
X-ray centers had rearranged themselves dramatically between the two
epochs
of observation. The curve itself shows evenly spaced maxima every 10 156 s
(2.8 h; Fig. 2b). On a
subsequent study of six RXTE light curves of
Cas,
Robinson et al. (2002)
discovered that autocorrelation peaks were most apparent in light
curves associated with flare fluxes. These authors found that the
features are centered at different time lags at different epochs, and
occasionally did not occur at all. At various times these authors found
spacings at 7.5 h, 3.5 h, and 5.8 h. The
2.8 h spacing we now find appear to be a recurrence of the
same cycle length these authors found during a long RXTE observation in
2000 December. As Robinson
et al. (2002) found, the autocorrelation peaks seem
to
be caused by an absence of flux (lulls) at different times. As such,
they seem to hint at the existence of X-ray relaxation cycles in the
Cas
environment. In contrast, Lopes de
Oliveira et al. (2007) found no such lulls in the
light curve of
Cas.
![]() |
Figure 4:
The high resolution fluxed spectrum of |
Open with DEXTER |
5 Spectroscopic properties
5.1 Comparison with previous Chandra HETGS spectrum
We have obtained a XMM-Newton spectrum in order to
investigate time-dependent differences and also to exploit the higher
effective aperture of the system, albeit with lower resolution at
higher energies. The XMM-Newton provides coverage
over longer wavelengths (1-38 Å)
than an Chandra spectrum with the high and medium
energy gratings (
1.5-25 Å).
Our XMM-Newton spectrum shows the same
general properties as the 2001 Chandra spectrum
(Fig. 4).
Both spectra show continua consistent with a multi-component thermal
model that includes a number of Lyman
lines of hydrogen-like ions, helium-like ions, as well as a few
Lyman
lines. Fluorescent lines from lower ion stages of Fe and Si are also
visible. The high energies are dominated by the presence of the
``Fe K complex'' (Fe XXVI Ly
,
Fe XXV, and fluorescent Fe K lines) in the
range 1.7-2 Å. The two spectra of this aggregate look similar,
except that the fluorescence feature is weaker in the XMM-Newton
spectrum. However, our RGS spectrum reveals many more details than the Chandra
one because of the XMM-Newton's larger effective
aperture as well
as an extension to longer wavelengths and
because in 2004 the soft X-rays of
Cas were
attenuated far less than
in 2001. These circumstances allowed us to probe to lower temperatures
and to better judge the number of components required to fit both lines
and continua in the soft X-ray region.
5.2 Selection of multi-component modeling
Our models are the results of fitting both the continua and the line
strengths in a multivariable solution with either the mekal
formulation for optically thin plasmas, or a related model,
vmekal, in which individual elemental abundances can
be determined independently. A Gaussian line was
included to describe the fluorescence feature arising from low-ion
stages of Fe at 1.94 Å (6.4 keV), not incorporated in
the mekal code. In all cases we adopted the phabs
code to describe the photoelectric absorption. When required, we
assumed the Hipparcos distance of 188 pc for Cas
(Perryman 1997).
The high quality of our EPIC pn and RGS
spectra allowed us to determine
in detail the thermal components required to fit the line and continuum
fluxes. This was especially important in allowing us to test whether
the
plasma has a continuous range of temperatures. Initial explorations
showed that contributions from 3 or 4 temperatures
are
necessary. In addition, otherwise mediocre fits to the continua
(especially
the soft continuum) required us to consider a two-absorption column
model. We found first the same hot temperature that many other
investigators have reported. We will refer to this component as kTQ.
This component has a 12 keV temperature and dominates the
total X-ray flux.
The carbon and nitrogen line spectra are emitted by a cool plasma with
a temperature kT1 of 0.1 keV.
The unabsorbed luminosity of this component, integrated over the range
0.2-12 keV, is
6.8-8
1031 erg s-1.
One or two intermediate (warm) temperatures are likewise required to
explain the presence of lines of oxygen, neon, magnesium, silicon,
a few Fe ions with L-shell configurations
(Fe XVII-XXIV), and the Fe XXV line
(Figs. 4
and 5).
Before refining the initial 3 and 4 component models further,
we attempted to determine whether these multiple thermal components
were discrete or could be part of a continuous distribution, as might
be expected in a single, thermally differentiated plasma. To continue
tests using this initial set of simple models, we froze
all but one temperature in our 4-T model and computed models
for the remaining temperature specified over graduatedsteps. We made
movies of these results overplotted with the observations and
determined those values of the scanned temperature which allowed us
to judge the best agreement with the observations. This technique also
allowed us to note whether predicted lines are not observed as well as
to test whether the principal plasma components are essentially
monothermal.
Figure 6
demonstrates
this for kT2 in M2 (4-T),
which is necessarily confined within the
range 0.5 and 1 keV. This plot shows the absence of
Fe XXIII and strong Fe XXIV
lines, as marked on our 10-13 Å segment
of our RGS2 spectrum
(RGS1 shows a gap at this wavelength). These lines would be stronger
visible if the temperature were in the range
0.7-0.9 keV, and yet they are not seen. This fact eliminates
the
possibility that kT2 is a
distribution of temperatures extending to higher temperatures than
0.7 keV. On the lower bound our models predict the presence of
the Fe XVIII 14.2 Å line for
temperatures 0.5 keV,
and this is not visible either. In addition,
the abrupt disappearance of components from the O VII complex
in the models shows that the absence of intermediate temperature plasma
extends from 0.5 keV to about 0.2 keV. Altogether
these diagnostics indicate that the kT2
component in the
Cas
environment is limited to a very narrow range
of temperatures centered near 0.6 keV and that it is distinct
from a lower or higher plasma component. This statement can probably be
extended to the spatial separation of the warm and cool plasmas as
well.
![]() |
Figure 5:
The Fe K |
Open with DEXTER |
Proceeding with this analysis, the weak Mg XII Ly line
hints the existence of a warm kT3
component at
2 keV,
although
some of our models have managed to achieve fits without it. The absence
of yet other Fe L lines indicates that there is
likely no plasma emission in the range of 2.5 to at least
5 keV. Also, the presence of the Fe XXVI Ly
feature, the main line diagnostic in the hard X-ray
spectrum of
Cas,
requires a temperature of at least 8 keV.
The Fe XXV Ly
line
requires a value significantly lower than this limit, and this means
that it cannot be formed in the kTQ plasma
component alone. In Fig. 6 a feature
due to Ne X Lyman
is present. This feature, visible in both the
individual RGS1 and RGS2 spectra,
cannot be easily fit with our thermal equilibrium (mekal)
models.
It is quite possible that the kTQ
component consists of some distribution of temperatures around a mean
value of 12-13 keV. This inference is supported by the
occasional color variations in the high energy light curve
of Cas
(SRC98), suggesting the presence of hot many separated exploding gas
volumes having a small range of temperatures.
The upshot of these considerations the presence and absence of various lines in this spectrum forces the conclusion that the plasma does not exhibit a continuous Differential Emission Measure (DEM). Rather, there are gaps over the ranges of about 2.5-8 keV, 0.7-1 keV, and 0.2-0.5 keV. S04 had suspected that a continuous DEM was inappropriate but could not state it conclusively because of the pronounced photoelectric absorption of the Chandra soft-energy spectrum.
![]() |
Figure 6:
The observed RGS2 spectrum in the 10-13 Å region for
three trial values of kT2
bracketing our
model M2, in each case offset vertically for convenience. This
figure depicts the Ne X Ly |
Open with DEXTER |
Table 1: Spectral fit results using the 3-T and 4-T models.
5.3 Model fitting
The above procedures allowed us to refine our initial 3-T and 4-T
models, utilizing both line and continuum flux information. Further
attempts to search for a continuous Differential Emission Measure
from the continuum alone, such as with a cooling flow
cemekl led to significantly degraded fits even in
the continuum.
The fits converged for keV
and
.
Freezing
to 12 keV produced an unacceptably high
.
We next tried a composite
model cemekl + mekal. This
resulted in temperatures of
keV
and
keV
for each component, respectively, and a
.
This
model failed to describe the continuum at high energies and
underpredicted
the following features: the Fe XXV line in the
Fe K
complex,
the Fe L-shell lines, OV III Ly
,
the fir triplet of O VII,
N VII Ly
,
and Ne X Ly
.
Allowing the cemekal
parameter to vary did not improve the fits.
Table 1
shows the results of our analysis in terms of three basic models.
Columns 2, 3, and 4 list the values determined for
the temperature, column density, the modeled unattenuated flux, the
resulting percentage of the total X-ray flux, and
emission measures for the three and four component models,
and the Fe abundance derived from the Fe XXV and
Fe XXVI Ly ions.
In all cases the high temperature component we designate as
kTQ was
found to lie near 12-14 keV.
Column 2 of Table 1 shows the
solution for a 3-T model,
``(M1)'', that includes a cool and warm components (kT1
and kT2, respectively) as
well as the dominant one, kTQ.
The cool/warm component fluxes were attenuated by a column
while the hot one was attenuated by
.
Our 4-T model (Col. 3 of Table 1; ``M2'')
gives the solution for the addition of a warm component having kT3
and also attenuated by
.
The final column of the table, ``(M3)'', is the 3-T solution again.
However, this time the hot component was attenuated by both
and
columns. In this model, the dominant hot subcomponent (kTQ),
affected by the individual
column, contributes to
76%
of the total flux. The second hot subcomponent (kTQ'),
warm (kT2), and cool (kT1)
components were affected by a common
absorption column. Figures 7
and 8
show the unfolded spectrum for each model,
allowing the reader to judge which of the various observed lines is
formed in which plasma component. In our solutions we found that
cm-2
and
-
cm-2.
By comparison,
the UV and H
determined ISM column density to
Cas from the
literature is a scant 1-
cm-2,
which leaves only a little room for absorption of the soft X-rays
within
the source.
The 4-T fit we found included components kT1,
kT2, kT3,
and
kTQ having
values of about 14 keV, 2.4 keV, 0.6 keV,
and
0.1 keV, respectively. For the 3-T models, the kT3
component was omitted,
but this omission must be compensated for in the
Fe XXVI Ly
strength by
decreasing the temperature from 14 to 12.5 keV.
![]() |
Figure 7: Unfolded spectrum from RGS1 and RGS2 for each model in Table 1. The blue, green, cyan (light blue) and red lines correspond to the kT1, kT2, kT3 and kTQ (and kTQ'), respectively, while the black lines correspond to the composite model. |
Open with DEXTER |
![]() |
Figure 8:
Unfolded spectrum from EPIC pn, centered at the
Fe K |
Open with DEXTER |
It is worth noting that any of our models give us the desired .
The residuals are mostly in the lines, not in the continuum, and the
relatively
high - although even acceptable - values of
(Table 1)
are due in part to an inadequate description of the individual line
profiles.
A key point to take away from the Table 1 is that although 84-90% of the total flux is radiated by the kTQ (or kTQ+kTQ') component(s), almost all the lines except for the Fe K complex is contributed by the cooler kT1-3. The hot kTQ component therefore provides no kinematic information. However, the possibility of multiple absorption columns and the presence of the fluorescence features of Fe and Si do provide potential geometrical descriptions for the emitting volumes.
Although S04's solution for the 2001 spectrum was similar in
that it
required a 4 thermal component fit, there were some small
differences in
detail that are significantly different. One of the clearest
differences
is with the precise temperature of a warm component (their ``
keV cf. the
keV
herein). A small increase in the O VIII/O VII
Lyman
line
ratio for
the XMM-Newton spectrum is indicative of the higher
value we find. If a warm temperature is a quasi-permanent feature of
the X-ray spectrum, it has definitely shifted from 0.4 keV to
0.6 keV between 2001 and 2004. The hot
(
12-14 keV)
and cool (
0.1 keV)
component temperatures
overlap. Although we cannot be sure if
2 keV plasma
existed in 2004, its possible presence is consistent with emission
powered by a similar thermal component in the 2001 spectrum in
their
2001 Chandra observation.
As already noted, one of most important differences from
earlier results is that no high absorption column was needed
to fit this spectrum. S04 found that the Chandra
spectrum
required a fitting with two hot plasma absorption columns. The first
contributed 10-30% of the 12 keV component and was
attenuated by an high absorption column of 1023 cm-2.
The second component, which is the dominant flux
contributor to the observed spectrum at high energies, was attenuated
by a column of
cm-2.
In contrast, most of
the X-ray flux observed by XMM-Newton
is affected by one dominant absorption column equivalent to
-3
1020 cm-2,
as derived from our
3-T and 4-T model with a single absorption column. However, for our
models we were only able to fit the continua of our combined
EPIC pn and RGS spectra with two absorption
columns. Although the absorption of the
column, at
2-
cm-2,
is very low, it has
much influence because it affects most (62-96%) of the total emission
(see Table 1).
The other absorption column has only a minor influence in absorbing the
flux even though it is higher by a factor of 7-10 (
cm-2).
We note that the temperatures derived for each component in the models
with one, as opposed to two, absorption columns are consistent with one
another.
The two column model also improved the descriptions of the lines formed
at low energies, such as
Ne X Ly
,
and O VIII Ly
.
The improvement of the fit with the second column over
a model with a single column
is supported by the F-test statistic, stating with a probability of
only
that the derived decrease in
associated with the second column column is due to chance. The
important conclusion from this is that the absence of a high absorption
column as derived from the XMM-Newton observation
suggests a dramatic rearrangement of cold circumstellar gas in the
vicinity of the X-ray sources between 2001 August and 2004 February.
Contrary to what is arguable for the 2001 epoch, it cannot be said that
in 2004 the circumstellar disk was in front of a significant fraction
of the soft X-ray source(s).
In order to estimate the origin of the long-wavelength lines
in our models,
we constructed a lineless continuum by excluding of the spectral
regions lines
in the 10-17 Å range and fit the resulting pure continuum
spectrum to a
bremsstrahlung model. This model gave a value a slightly hotter
temperature
13.2 keV for our hot component with a
of 1.17. The addition of a second bremsstrahlung component did
not improve the fits (
keV;
of 1.17).
This result suggests that the determination of the low component
temperatures, kT1 and kT2,
is strongly driven
by the lines alone.
Although all of our discussion has taken place in the context
of equilibrium models, we agree with Smith et al. (2004), who
suggested that
the warm and/or cool plasma components could also result from
nonequilibrium
ionization processes, e.g. from the sudden impact of gas parcels
ejected
by flares into a dense stationary medium such as the Be disk. We
suspect this could also be the reason why we observe Lyman transitions,
such as the Ly
transitions of Ne X and O VIII, that are not easily
fit with equilibrium mekal models.
A similar explanation might be also important in understanding the
Fe XXVI Ly
line
at 8.3 keV of HD 110432, reported by Lopes de Oliveira et al. (2007),
and which could not be fit with standard models for the ranges of
temperature considered for this feature. The strong Lyman
features hint at excitation temperatures in
these plasmas that are higher than the ionization temperatures we
found in Table 1.
5.4 Abundances and anomalous line strengths
For the most part the line strengths are consistent
with solar abundances in our 3-T and 4-T models. The
first and most conspicuous exception to this general result is the
finding from Table 1
that the Fe abundance from K-electron
Fe ions is
.
This is a factor of two lower than any other studies in the literature
(the S04 result was
). Because
K-shell ion abundances of 0.1
and 0.24
are significantly different from one another, our result clarifies in a
new way that the Fe abundance derived from K-shell ions
changes with time. Thus, the K-shell result may not reflect a global
elemental abundance.
As also found by S04, the Fe abundance from L-shell ions is
significantly higher than the Fe abundance from K-shell ions. We
pursued the investigation of the Fe abundance from the Fe L-shell lines
with the models M1, M2 and M3. In a new set of models we fixed other
variables and solved for the Fe abundance of the soft temperature
components kT1 and kT2.
Our models suggest that the Fe L-shell abundance could be 0.4
to .
This exercise verifies the distinctness of the Fe abundance arising
from the
L-shell lines on one hand and the K-shell lines on the other.
For completeness note that a similar K- and L-shell Fe anomaly was
reported by Lopes de Oliveira
et al. (2007) for
Cas.
The second apparent departure from solar abundances comes
from the enhanced strengths of the lines of hydrogen and helium-like
ions of neon and nitrogen. The formal abundances we find from our
models
are
and
,
assuming solar abundance for the other metals in the softest plasma and
allowing for line broadening and bulk velocities. However, the allowed
range of Ne abundances depends on the number and nature of the
parameters left free in the fit, the most important being the assumed
Fe L-shell abundance and the line broadening velocity. The hottest
component dominates the NeX line flux with a smaller (
1/3)
contribution from the
thermal component. For instance, using bvapec instead of mekal and
leaving Fe L-shell abundance free yields slightly lower abundances with
and
.
This effect is due to the presence of a FeL-shell forest around the
NeX L
line
and to the fact that the continuum level depends on the description of
the FeL-shell lines. Although the limits for the abundances are not
well determined, there is a clear excess above the solar values for N
and Ne. The NeX line peak velocity is in the range of -110 to
+230 km s-1 and displays a
broadening velocity comparable to that of other bright emission lines.
A similar analysis was applied to the Chandra spectrum in
order to determine whether the Ne enhancement was present or not in
2001 - it is not possible to expand the investigation to N abundance
due to the low SNR of the NVII L line. The
non-velocity broadened mekal model yields
consistent with the value obtained by Smith
et al. (2004). However, since the observed NeX line
apppears significantly wider than modeled, we used the bvapec
model to take into account such a broadening. The larger NeX EW impacts
the determination of the abundance yielding
(or
if the oxygen and iron L-shell abundances are left free, in which case
we obtain
).
The broadened line profile fits the observed Ne line much better, with
peak velocities then in the range of -60 to
260 km s-1 and broadening
velocities of
km s-1.
These values are
consistent with those derived from the RGS spectrum.
We note carefully that whereas the fluxes in these lines scale linearly with abundance, the equivalent widths, as formally defined with respect to the neighboring continuum flux, scale far more slowly because of the contribution to the local bound-free opacities from ions in a N-Ne rich plasma. Thus, we found an increase of only 35% with respect to the equivalent width measured in the Chandra spectrum. Our models with XSPEC validate this mild increase in equivalent width with abundance in detail. Given this reality, we anticipate that the true abundance errors are larger than those XSPEC computes based on photon statistics. Nonetheless, the anomalous excess cannot be discounted. Moreover, the possibility that the X-ray environment could be nitrogen-rich by a factor of 3-4 is unremarkable because enhancements are already a hallmark of massive stars evolving off the main sequence, although the reasons for this are still under discussion (Hunter et al. 2008). However, the apparent neon enhancement is of much greater interest because in a stellar evolution context neon is produced by carbon burning shortly before supernova detonation or in the interiors of some white dwarfs, and dredging to the surface of such elements is possible by various mixing processes.
Suggested alternatives to the straightforward abundance
interpretation of the line strengths are the following:
(i) instrumental artifact or a cosmic ray; (ii) an inappropriate
temperature used in the modeling; and
(iii) the line is strengthened by a microturbulent like broadening.
Each of these possibilities may be dismissed in turn. (i) is unlikely
because separate spectrum extractions from the positive and negative
detector halves show consistent profiles.
Likewise, we may rule out (ii) because a temperature of
0.6 keV is already ideal for Ne X formation, and a
stronger feature cannot be produced
given a standard abundance. Possiblity (iii) would require a large
optical
depth and strengthening of the line. However, even though
the column density is probably high enough for photons to experience
more than one mean free path through their
transits across the medium, they are nonetheless scattered coherently.
Therefore the line widths are unaltered by these histories.
With these possibilities ruled out, we are forced to
conclude that the neon and nitrogen abundances in the X-ray plasmas
of Cas
are high.
In considering the anomalous Fe abundance derived from
Fe-K lines, but normal abundance from Fe-L lines in the Cas
spectrum, S04 suggested that this could arise from an inverse
FIP (first ionization potential) effect
in the
Cas
environment that is similar to that found
in the coronae of the Sun and other magnetically
and X-ray active cool stars, such as AB Dor (Güdel et al. 2003).
Although the details are still unclear, it appears that in a magnetic
plasma low density environment differential ponderomotive forces,
produced
by wave heating and dependent on their first ionization potential, can
prevent certain ions from migrating across a plasma and resulting in an
altered measured chemical abundance (Laming
2004). Whether such a process is actually active
in the environment of
Cas,
let alone whether it extends to
Ne9+ ions, must be considered
speculative. Its attractiveness lies
in its potential to explain the different Fe abundances derived from
K-shell and L-shell ions.
Similarly, now in the framework of the accretion model, it is not
clear, first,
how Ne could be preferentially ejected from the white dwarf
atmosphere or,
second, how Ne could be enhanced without enhancing at the same time
C, O, and Mg abundances.
5.5 The Fe K
complex
In evaluating the equivalent widths (EW) of each emission iron line of
the
Fe K complex, we used the 5-10 keV photons acquired
during low background phases of the XMM-Newton
satellite's orbit. Our approach was to limit consideration of spectra
accumulated only
during low (L) and high (H) flux states observed in Cas
(defined as the lowest and highest one-third fluxes of the total
distribution in our light curve) and also
during some 1000 time windows in which the detector background
was low (see Fig. 1-b).
Finally,
we applied an absorbed bremsstrahlung model in order to describe the
underlying continuum and three Gaussian lines to
account the iron lines. Table 2 and
Fig. 5
show the results. The measured centroid energy of each modeled Gaussian
line is, within the errors, in agreement with the theoretical values
for the fluorescent, helium-, and hydrogen-like components of the
Fe K
complex.
It is possible that there may be a slight inverse sensitivity of the
fluorescence strength (relative to the Fe XXVI line),
though this
inference is of marginal statistical significance.
Table 2:
Parameters of the emission lines of the Fe K
complex from a bremsstrahlung + 3 Gaussian lines model.
In our EPIC pn spectrum the
strength of the fluorescence feature (
mÅ) is weaker than
when Chandra observed it (
mÅ; Smith et al. 2004),
in line with the fact that the attenuation of soft X-rays by cold
matter is less.
This suggests that the fluorescence
emission feature is formed at least partially by
the same medium that aborbs the soft-X ray flux.
Curiously, the analysis in the low and high-state spectra reveals that the FeK fluorescence feature may be marginally stronger in the low flux case. Although such investigation could not be expanded to a quantitative analysis of all spectral range because the relatively limited signal-to-noise in the final spectra for RGS1/2, we suspect that there is weak evidence for a weakening of the SiK feature for the low flux spectrum.
5.6 Helium-like diagnostics of electron densities
Our spectrum covers the regions of the He-like Ne IX,
O VII and N VI complexes - but the N VI
complex falls partially onto gaps in the CCD for RGS1 and RGS2. Each of
these is comprised of a so-called
fir (forbidden/intercombination/resonance) line
triplet (see Fig. 4),
and the ratio of their intensities can indicate
whether the dominant excitation process for producing this triplet
is collisional or photoionization. For example, if collisions are
dominant, the ratio
obtains, whereas if photoionizations dominate the exitation
(e.g. Porquet
& Dubau 2000; Porquet et al. 2001).
From the O VII fir complex in our XMM-Newton
spectrum,
we estimate
for
Cas.
The N VI rif components are poorly
measured because they fall within the gaps of the RGS2 detector.
Nonetheless,
by computing their ratios we find nearly the same value,
.
This suggests that the plasma is in the classical domain of collisional
dominance.
As it happens, the firratio alone cannot
distinguish between quenching of the
forbidden transition by collisions or by photoexcitations by a nearby
strong UV source, such as the Be star.
Table 3: Parameters of the strongest emission lines.
5.7 Velocity broadening and shifts in warm plasma lines
In Table 3
we list the observed centroid wavelength, the equivalent width and the
flux of lines for those lines for which meaningful measurements could
be made. The best observed lines in the soft X-ray spectrum,
Ne X Ly
and O VIII Ly
profiles are noticeably broadened. We evaluate the broadening of the
lines by using the bapec model,
a velocity- and thermally-broadened emission model from
collisionally-ionized
diffuse gas
. The Gaussian
sigma for velocity broadening converges to
400 km s-1for
a model like M1, M2 and M3 in Table 1, replacing
the mekal by the bapec code.
The line broadening was also measured on individual lines, not only on
the whole spectrum. For example, we found a velocity broadening of
km s-1for
the O VIII line.
The RGS velocity is consistent with that reported by S04 (
km s-1).
We applied the same analysis to two RGS spectra of AB Dor
acquired in 2000 and 2006 (obsID 0126130201 and
0160363001 with up-to-date calibration) and found in each case that the
broadening of the O VIII line was less than about
70 km s-1 at the 90%
confidence level, proving that the lines of
Cas are indeed
broadened.
In order to check for velocity shift in the lines, we run the
models in Table 1
leaving the redshift parameter free. We found a overall redshift of 200 km s-1
- and values for the
parameters and
consistent
with those shown in Table 1 - slightly
larger than the typical error on the absolute wavelength scale of
7 mÅ,
or
140 km s-1.
5.8 Do our cool or warm plasma component arise in a radiative wind?
It has been believed for some time that X-ray emission in O and early
B stars arises in sites associated with shocks distributed in
the stars radiative winds (e.g. Leutenegger et al. 2006;
Cohen 2008).
Analyses of HETG spectra of
other stars in the same general region of the H-R Diagram, for
example of Sco
(B0.2 V),
Cru
(B0.5 III), and
Oph
(O9.5 V) by Cohen (2003),
Cohen (2008),
and Waldron (2005),
respectively, paint a complex picture.
The Cohen et al. analyses show that the winds of at least some
early B stars cannot be described by a standard picture of
wind-shocks, and indeed wind structures of at least
Sco
and
Oph
may be influenced by the effects of channeling
by magnetic fields in the regions surrounding the star. The
temperatures of the gas are not yet known to vary but taken as a group
range from about 3 to 25 MK. We note that the X-ray
emission measures associated with winds in these stars have not been
found to be variable so far.
Cas
has such a wind, as evidenced by the profiles of doublets
in UV resonance absorption lines of several ions, notably
C IV, N V, and Si IV. These profiles exhibit
Discrete Absorption features at
1000-1100 km s-1 and edge velocites of
-1800 km s-1 (Doazan 1982;
Smith
et al. 1998b; Kaper et al. 1996; Smith et al.
1998a). Our analysis shows the presence of soft X-ray
components whose temperatures, fluxes, and emission measures
(Table 1)
are consistent with those of normal massive stars reported by Walborn et al. (2009),
for example. It is thus expected that at least part of such emission is
emanating from the radiatively driven wind of
Cas.
However, as can be seen in Fig. 7, the
hottest component dominates also the soft part of the spectrum of
Cas
and the contribution of a pure radiative wind is overshadowed by this
hot component.
Establishing the presence of a luminosity change in the soft components
supporting a non-wind origin is a complicated issue because of the
presence of the Be disk itself (which may shield a part of the wind)
and because the characteristics of the
Cas' spectra in
2001 and 2004 are different from each other.
It is worth noting that the X-ray 0.2-12 keV (unabsorbed)
fluxes from the soft (
keV) component in
2001 and 2004 are consistent with each other (<
erg s-1
from Chandra and 1-
10-11 erg s-1
from XMM-Newton), while the flux from the
keV
component has increased between 2001 and 2004 (from 0.7-2.2 to 2.5-
erg s-1).
In conclusion, we cannot discard an additional contribution to the soft
plasma in addition to that of a radiatively driven wind usually
observed in massive stars.
6 Conclusions
We have reported the second high dispersion analysis of the
X-ray spectrum of the X-ray anomalous B0.5e star Cas
obtained by the
XMM-Newton in 2004, and we find the following
characteristics
at this epoch:
- a)
- the emission is due to an optically thin, thermal medium
comprised of 3 to 4 discrete components but dominated
by a hot component having
-14 keV. The temperature of at least one component (
keV) has definitely shifted since the Chandra HTEG observation in 2001. It is possible that a
keV component exists too. If so it may represent plasma with nearly the same temperature found by S04. Further, the presence of a cool component with
keV is consistent with a value found by S04. This may or may not have the same cause as the wind-shocked fluxes emitted from other early B and Be stars. However, if so, it is a lower temperature than expected. Only part (at most) of the kT2 plasma could be produced in the wind, and if so the large line widths evidenced in this component indicate a larger turbulent broadening than is typical of winds in other early B stars.
- b)
- The hot kTQ component appears stable and any distribution of temperatures of individual sites around this mean value must be small.
- c)
- A subsolar abundance of iron is derived from the
Fe XXV and Fe XXVI Ly
features, in agreement with several other determinations. As S04 also found, this FeK abundance is significantly different of the abundance found from Fe-L ion lines. We have also discovered that [FeK] changes with time and was significantly higher in 2004.
- d)
- The light curve of this star again shows ubiquitous,
rapid flaring. This was also found by Lopes de Oliveira et al. (2007) in
Cas. However, unlike
Cas, color changes occurred only seldomly, and evidently not at all on the few hour timescale noted in
Cas.
- e)
- Our light curve shows a quasi-periodic lull every 2.8 h in 2004, similar to the cyclical lulls of 3.5 h, 7-7.5 h, and 5.8 h noted by RSH02. These appear to occur in most epochs.
- f)
- A thick absorption column affected 25% of the hot component in August 2001, but it had disappeared by February 2004.
- g)
- Apparently variable Fe K and perhaps Si K fluorescent features are present. These emissions correlate with an absorption column (point f) that attenuates soft X-rays.
- h)
- Broadening of warm (kT2)
component lines was noticeable at both epochs but may have increased
marginally from 2001 (
0.4 keV) to 2004 (
0.6 keV).
- i)
- The strengths of lines of two ions each of N and Ne are underpredicted for XSPEC models with solar abundances. Upon consideration of alternative explanations, we have interpreted these as evidence of abundance enhancements. However, we do not understand the cause of these enhancements.
- j)
- Given the identification of the Lyman
feature in a few ions formed in the warm and cool plasmas, there is more than a hint of nonequilibrium processes in the environment of
Cas. This suggests formation in an (at most) intermediate density environment, which is clearly separate from the very high density plasma in which the flares are formed (Smith et al. 1998a).
- k)
- The unabsorbed flux of
Cas at 0.2-12 keV in 2004 from XMM-Newton (
1.7-
erg s-1) is consistent with the value observed in 2001 from Chandra (
2-
erg s-1).



Several changes we have noted are as remarkable as they were
unexpected. Most especially, we have noted changes in the geometry of
the circumstellar environment as reflected in the disappearance of one
of the two columns, and affecting 25% of the hard emission. The strong attenuation
of the 2001 soft X-ray spectrum was clearly evident.
In addition, the strengths of the K fluorescence emission
features
decreased in line with the decrease of the column absorption,
suggesting that the part of the emission is caused by scattering of
hard photons through the attenuating column that was present at the
earlier time.
The reduced attenuation of the soft X-rays partially accounts for the improved ability to study the nature of the warm and cool plasma emissions. As a result, it is finally clear that these components are almost monothermal and therefore are not part of an integral structure with a smoothly varying thermal emission measure, such as an accretion column or a cooling flow plasma. Moreover, the broadening of the lines has increased during the 2001-2004 interval, either because of an increase in a quasi-turbulence or the splitting of a former single region into two with different projected radial velocities.
In contrast to the warm component, a study of the variations in the Fe K-shell lines with temperature shows that the hot plasma need not consist of a uniform kTQ value. Rather, we believe that small variations in the hardness that are occasionally observed in our data and those discussed by SRC98, RS00, and RSH02 can be understood as variations of the instantaneous average temperature resulting from the rapid evolution of small number of flares and nearby basal emission regions.
The unique properties of each of the thermal components, the changing absorption column geometry, and increases in line broadening all provide new hints to the mechanism responsible for the X-ray production. Although on one hand, the disappearance of the strong absorption column and the lack of correlation between the fluorescence features no longer support the argument that the gas in the Be star's circumstellar disk strongly interact with the hard X-rays, the discreteness of the plasma components argues that they are likely to occupy distinct volumes.
Future generations of X-ray telescopes such as the
International X-ray Observatory (IXO) will be important in refining our
understanding of the spatial distribution of emission volumes of Cas
and its analogs. Their observations promise to address such question as
whether the observed line broadening can be tied to the rotational
velocity of the Be star and to resolving distinct sources,
such as corotating active regions.
R.L.O. acknowledges financial support from the Brazilian agency FAPESP (Fundação de Amparo à Pesquisa do Estado de São Paulo) through a Postdoctoral Research Fellow grant (number 2007/04710-1). We gratefully acknowledge the XMM-Newton User Support Group, in particular Nora Loiseau, Jan-Uwe Ness, and Matteo Guainazzi, for their help with problems on SASv8.0.1. We would like to thank the anonymous referee for his/her comments.
References
- Berghöfer, T. W., Schmitt, J. H. M. M., Danner, R., & Cassinelli, J. P. 1997, A&A, 322, 167 [NASA ADS] [Google Scholar]
- Buccheri, R., Bennett, K., Bignami, G. F., et al. 1983, A&A, 128, 245 [NASA ADS] [Google Scholar]
- Cohen, D. H., de Messières, G. E., MacFarlane, J. J., et al. 2003, ApJ, 586, 495 [NASA ADS] [CrossRef] [Google Scholar]
- Cohen, D. H., Kuhn, M. A., Gagne, M., et al. 2008, MNRAS, 386, 1855 [NASA ADS] [CrossRef] [Google Scholar]
- Cranmer, S. R., Smith, M. A., & Robinson, R. D. 2000, ApJ, 537, 433 [NASA ADS] [CrossRef] [Google Scholar]
- Doazan, V. 1982, in B Stars with and without Emission Lines, ed. A. Underhill & V. Doazan, NASA SP-456, 326 [Google Scholar]
- Ehle, M., de la Calle, I., Díaz Trigo, M., et al. 2008, XMM-Newton Users' Handbook, Issue 2.6, 15 July 2008 [Google Scholar]
- Frontera, F., Dal Fiume, D., Robba, N. R., et al. 1987, ApJ, 320, L127 [NASA ADS] [CrossRef] [Google Scholar]
- Guainazzi, M. 2008, XMM-SOC-CAL-TN-0018, Issue 2.7.2 (4 November 2008) [Google Scholar]
- Güdel, M., et al. 2003, ASP Conf. Ser., 277, 221 [Google Scholar]
- Haberl, F. 1995, A&A, 296, 685 [NASA ADS] [Google Scholar]
- Harmanec, P., Hadrava, P., Stefl, S., et al. 2000, A&A, 364, 85 [Google Scholar]
- Howk, J. C., Cassinelli, J. P., Bjorkman, J. E., et al. 2000, ApJ, 534, 348 [NASA ADS] [CrossRef] [Google Scholar]
- Hunter, I., Brott, I., Lennon, D. J., et al. 2008, A&A, 676, L29 [Google Scholar]
- Kaper, L., Henrichs, H., Nichols, J. S., et al. 1996, A&AS, 116, 257 [Google Scholar]
- Kubo, S., Murakami, T., & Corbet, R. H. D. 1998, PASJ, 50, 417 [NASA ADS] [CrossRef] [Google Scholar]
- Laming, J. M. 2004, ApJ, 614, 1063 [NASA ADS] [CrossRef] [Google Scholar]
- Leutenegger, M. A., Paerels, F. B. S., Kahn, S. M., & Cohen, D. H. 2006, ApJ, 650, 1096 [NASA ADS] [CrossRef] [Google Scholar]
- Lopes de Oliveira, R. 2007, Ph.D. Thesis, Universidade de São Paulo, Brazil, and Université Louis Pasteur Strasbourg I, France [Google Scholar]
- Lopes de Oliveira, R., Motch, C., Haberl, F., Negueruela, I., & Janot-Pacheco, E. 2006, A&A, 454, 265 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
- Lopes de Oliveira, R., Motch, C., Smith, M. A., Negueruela, I., Torrejón, J. M. 2007, A&A, 474, 983 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
- Miroshnichenko, A. S., Bjorkman, K. S., & Krugov, V. D. 2002, PASP, 114, 1226 [NASA ADS] [CrossRef] [Google Scholar]
- Motch, C., Lopes de Oliveira, R., Negueruela, I., Haberl, F., & Janot-Pacheco, E. 2007, in Active OB-Stars: Laboratories for Stellar and Circumstellar Physics, Proceedings of the conference held 29 August-2 September, 2005 at Hokkai-Gakuen University, Sapporo, Japan, ed. S. Stefl, S. P. Owocki, & A. T. Okazaki, San Francisco: Astronomical Society of the Pacific, ASP Conf. Ser., 361, 117 [Google Scholar]
- Murakami, T., Inoue, H., & Agrawal, P. C. 1986, ApJ, 310, L31 [NASA ADS] [CrossRef] [Google Scholar]
- Owens, A., Oosterbroek, T., Parmar, A., Schultz, R., Stüwe, J. A., & Haberl, F. 1999, A&A, 348, 170 [NASA ADS] [Google Scholar]
- Parmar, A., Israel, G., Stella, L., & White, N. 1993, A&A, 275, 227 [NASA ADS] [Google Scholar]
- Perryman, M. A. C. 1997, The Hipparcos and Tycho Catalogues, ESA SP-1200 [Google Scholar]
- Porquet, D., & Dubau, J. 2000, A&AS, 143, 495 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
- Porquet, D., Mewe, R., Dubau, J., Raassen, A. J. J., & Kaastra, J. S. 2001, A&A, 376, 1113 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
- Prinja, R. K. 1989, MNRAS, 241, 721 [NASA ADS] [CrossRef] [Google Scholar]
- Robinson, R. D., & Smith, M. A., 2000, ApJ, 540, 474 (RS00) [NASA ADS] [CrossRef] [Google Scholar]
- Robinson, R. D., Smith, M. A., & Henry, G. W. 2002, ApJ, 575, 435 [NASA ADS] [CrossRef] [Google Scholar]
- Scargle, J. D. 1982, ApJ, 263, 835 [NASA ADS] [CrossRef] [Google Scholar]
- Secchi, A. 1867, Astron. Nachr., 68, 63 [Google Scholar]
- Smith, M. A., & Robinson, R. D. 1999, ApJ, 517, 866 [NASA ADS] [CrossRef] [Google Scholar]
- Smith, M. A., Robinson, R. D., & Corbet, R. H. D. 1998a, ApJ, 503, 877 (SRC98) [NASA ADS] [CrossRef] [Google Scholar]
- Smith, M. A., Robinson, R. D., & Hatzes, A. P. 1998b, ApJ, 507, 945 [NASA ADS] [CrossRef] [Google Scholar]
- Smith, M. A., Cohen, D. H., Gu, M. F., et al. 2004, ApJ, 600, 972 (S04) [NASA ADS] [CrossRef] [Google Scholar]
- Smith, M. A., Henry, G. W., & Vishniac, E. 2006, ApJ, 647, 1375 [NASA ADS] [CrossRef] [Google Scholar]
- Walborn, N. R., Nichols, J. S., & Waldron, W. L. 2009, ApJ, 703, 633 [NASA ADS] [CrossRef] [Google Scholar]
- Waldron, W. L. 2005, ASPC, 337, 329 [NASA ADS] [Google Scholar]
- White, N. E., Swank, J. H., Holt, S. S., & Parmar, A. 1982, ApJ, 263, 277 [NASA ADS] [CrossRef] [Google Scholar]
- Yang, S., Ninkov, Z., & Walker, G. A. H. 1988, PASP, 100, 233 [NASA ADS] [CrossRef] [Google Scholar]
Footnotes
- ... personality
- This work is based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA.
- ...
Xronos
- http://heasarc.nasa.gov/docs/xanadu/xronos/xronos.html
- ...
XSPEC
- http://heasarc.nasa.gov/docs/xanadu/xspec/index.html
- ...
continuum
- The cemekl/cevmkl code describes a multi-temperature
plasma based on the mekal code, in which the emission measures
of the plasma scale with their temperature as (
)
;
corresponds to the adiabatic case.
- ... gas
- http://cxc.harvard.edu/atomdb/
All Tables
Table 1: Spectral fit results using the 3-T and 4-T models.
Table 2:
Parameters of the emission lines of the Fe K
complex from a bremsstrahlung + 3 Gaussian lines model.
Table 3: Parameters of the strongest emission lines.
All Figures
![]() |
Figure 1:
a) RGS1+RGS2 light curve combining 1st and
2nd orders and the RGS1 background light curve, and EPIC pn light
curves in the b) 0.8-2 keV and
c) 2-10 keV X-ray bands, and the respective
d) hardness ratio. Time bins of 100 s. Upper limits
for error bars at 1 |
Open with DEXTER | |
In the text |
![]() |
Figure 2: Power spectrum. Top: from EPIC pn events at 0.8-10 keV, rebinned as a geometrical series. Bottom: from 0.8-10 keV EPIC pn data binned to 100 s; the dashed lines represent the confidence levels. |
Open with DEXTER | |
In the text |
![]() |
Figure 3: Autocorrelation from 0.8-10 keV light curve binned to 500 s. P1 refers to result of Fig. 2b. |
Open with DEXTER | |
In the text |
![]() |
Figure 4:
The high resolution fluxed spectrum of |
Open with DEXTER | |
In the text |
![]() |
Figure 5:
The Fe K |
Open with DEXTER | |
In the text |
![]() |
Figure 6:
The observed RGS2 spectrum in the 10-13 Å region for
three trial values of kT2
bracketing our
model M2, in each case offset vertically for convenience. This
figure depicts the Ne X Ly |
Open with DEXTER | |
In the text |
![]() |
Figure 7: Unfolded spectrum from RGS1 and RGS2 for each model in Table 1. The blue, green, cyan (light blue) and red lines correspond to the kT1, kT2, kT3 and kTQ (and kTQ'), respectively, while the black lines correspond to the composite model. |
Open with DEXTER | |
In the text |
![]() |
Figure 8:
Unfolded spectrum from EPIC pn, centered at the
Fe K |
Open with DEXTER | |
In the text |
Copyright ESO 2010
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.