Spectral Analysis of RXTE Observations of A3667

0.1 MB·

by  Yoel Rephaeli

February 2, 2008
Spectral Analysis of RXTE Observations of A3667
Yoel Rephaeli1,2, and Duane Gruber3
4 1Center for Astrophysics and Space Sciences, University of California, San Diego, La Jolla,
0
0 CA92093-0424
2
2School of Physics and Astronomy, Tel Aviv University, Tel Aviv, 69978, Israel
n
a 34789 Panorama Drive, San Diego CA 92116
J

5
1

1
v
2
ABSTRACT
1
3
1 X-ray emission from the cluster of galaxies A3667 was measured by
0
4 the PCA and HEXTE experiments aboard the RXTE satellite during
0
∼
/ the period December 2001 - July 2002. Analysis of the 141 ks RXTE
h
p observation and lower energy ASCA/GIS data, yields only marginal ev-
-
o idence for a secondary power-law emission component in the spectrum.
r
t
s The 90% confidence upper limit on nonthermal emission in the 15-35
a
: keV band is determined to be 2.6 × 10−12 erg-cm−2 s−1. When com-
v
i bined with the measured radio flux and spectral index of the dominant
X
r region of extended radio emission, this upper limit implies a lower limit
a
∼
of 0.4µ G on the mean, volume-averaged, intracluster magnetic field
in A3667.
Subject headings: Galaxies: clusters: general — galaxies: clusters: in-
dividual (A2256) — galaxies: magnetic fields — radiation mechanisms:
non-thermal
1
1. Introduction (Rephaeli, Gruber & Blanco 1999, Fusco-
Femianoetal.1999,Rephaeli&Gruber2002),
The spectral and spatial resolutions of cur-
A2256 (Fusco-Femiano et al.2000, Rephaeli
rent X-ray satellites allow a more realistic de-
& Gruber 2003), A2319 (Gruber & Rephaeli
scription of gas properties and a search for
2002),A119&A754(Fusco-Femianoetal.2003a),
new phenomena in clusters of galaxies. Spec-
andinthemoderatelydistantclusterRXJ0658-
tral diagnostics alone can reveal the presence
5557 (Petrosian 2003). However, the claimed
of a second spectral component that may in-
NT emission in A754 could possibly be from
dicate either a more complex temperature
a radio galaxy (Henriksen, Hudson & Tittley
distribution than a simple isothermal (with
2003).
its associated primary thermal emission), or
Results of two different analyses of a sec-
an appreciable energetic electron population
ond BeppoSAX observation of Coma were
that radiates non-thermally. Clearly deter-
reported very recently: According to Ros-
mining that the gas is non-isothermal is im-
setti & Molendi (2003), the full PDS dataset
portant for a more precise description of IC
∼
(with a total on-source exposure of 166
gas properties, evolution, and for estimation
ks) no longer shows significant evidence for
of the cluster total mass, as well as for our
a NT component, a claim that is disputed
ability to use the gasasa more precise cosmo-
by Fusco-Femiano et al.(2003b). Analysis of
logicalprobe. Thedetectionofanappreciable
the same data by the latter authors yields a
level of non-thermal (NT) emission in clus-
very significant NT compoent at a level that
ters with measured extended regions of radio
is only slightly lower than originally reported
emission (e.g.,, Rephaeli 1977, 1979) is also
by Fusco-Femiano et al.(1999).
very important as an essential second observ-
Since NT electron populations have a wide
ablewhichisneeded inorder todetermine the
energy range, their emission couldpossibly be
strength of magnetic fields, and for estimat-
detected also in the EUV range. It has been
ing densities andenergy content ofrelativistic
claimed that low energy emission observed in
electrons (and protons).
a few clusters by ROSAT and, in particular,
While considerable work is being done to
−
by the EUVE (in the 65 245 eV band), is
investigate the temperature structure of intr-
in excess of what is predicted from thermal
acluster gas, relatively few searches for non-
emission by IC gas, and that the excess emis-
thermal(NT)emissionwerecarriedout. These
sion is NT (e.g.,, Lieu et al.1996, Sarazin &
began with archivel analysis of HEAO-1 data
Lieu 1998, Bowyer et al.1999, 2003). Due to
(Rephaeli,Gruber&Rothschild1987,Rephaeli
the very limited spectral range of the EUVE
& Gruber 1988), and continued with CGRO
measurements this identification is uncertain.
(Rephaeli, Ulmer & Gruber 1994) and ASCA
Analysis of line and continuum XMM mea-
(Henriksen 1999) observations, yielding only
surements of soft emission from a sample
upper limits on spectral power-law compo-
of clusters seems to suggest that the excess
nents. The improved sensitivity and wide
emission is thermal emission from warm gas
spectral band of the RXTE and BeppoSAX
(Kaastra et al.2003).
allowed a more detailed spectral analysis of
The number of clusters observed at high
long exposure measurements that resulted in
(> 30 keV) energies is only a small fraction
significant evidence for NT emission in Coma
2
of the ∼ 40 clusters in which extended radio sults in 141 ks of screened PCA data, spaced
emission has already been measured. It is of irregularly over the observing period. These
considerableinteresttoenlargethesmallsam- were collected in two of the 5 detectors. For
ple in order to begin a more systematic study the HEXTE, which beam-switches observa-
of NT phenomena in clusters. Here we report tions with 32-second dwells between source
the results of an analysis of ∼ 141 ks RXTE and background fields, and has in addition
observation of the ‘merging’ cluster A3667. about 50% detector dead time, the net obser-
vation time was 54.6 ks with each of the two
2. Observations and Spectral Analysis clusters. A systematic error of 0.8% per en-
ergy channel was added in quadrature to the
A3667, a rich southern cluster at z =
statistical error of the PCA data; no system-
0.055,hasabimodalgalaxydistribution,large
atic error was used with HEXTE data. On
velocity dispersion, distorted X-ray morphol-
timescales oftwo weeks orlonger, thelimitto
ogy, and complex extended regions of radio
variability observed with PCA was less than
emission(e.g.,,Rottgeringet al.1997). These
1%. Becauseofthemuchlowersignaltoback-
features are thought to indicate that the clus-
ground, corresponding HEXTE limits to vari-
ter is undergoing strong merging activity. Of
ability are weaker, about 20%.
particular relevance is the very large, elon-
To extend the spectral range lower, analy-
gated and off-center region of radio emission.
sis was conducted jointly with archival ASCA
The spectrum of this dominant source (point
GISobservation(of1995April16)lastingand
sources are estimated to contribute only a
39 ks. GIS2 and GIS3 0.8 – 8.0 keV spectra
few percent of the total emission) can be de-
were accumulated in a field with a diameter
scribed in terms of an overall, single value of
of 14 arcminutes centered on the cluster, a re-
the spectral (energy) index, α ∼ 1.1, in the
gion that included the great majority of the
frequency range ∼ 0.4−2.3 GHz, and a flux
cluster emission. SIS data for the observation
of 5.5 ± 0.5 Jy at 843 MHz (Rottgering et
were found unsuitable for analysis because of
al.1997).
noise. Systematic errors were not added to
Analysis of ASCA observations of A3667
the GIS data.
yielded a mean gas temperature of 7.0 ± 0.6
Preliminary spectral analysis on the direct
∼ ′
keV over a large region with 22 radial ex-
instrument datashowed no features which de-
tent (Markevitch et al.1999). More recently
parted on small scales (i.e. spectral lines or
Chandrameasurements haveledtoamorede-
edges) from the thermal form which clearly
tailed temperature and brightness structure
dominates the bulk of the observed cluster
in the central region showing evidence for
spectrum. We therefore found it appropri-
large temperature gradients, including a re-
ate to combine data in order to make small
∼
gion where the gas temperature is 11 keV,
differences of χ−2 from model fitting more
possibly due heating of the gas merger shocks
distinguishable against the statistical noise of
(Vikhlinin et al.2001).
the highly oversampled data. As a first step
∼
A3667 was observed with RXTE for 167
we combined the data from the two PCA de-
ks during the period December 2001 - July
tectors and also the data of the two HEXTE
2002. The application of data selection cri-
clusters. For all three instruments we then
teria recommended by the RXTE project re-
3
joined the counts in adjacent energy channels
into wider energy bins at a density of two to
three per detector energy resolution element.
Doing so reduced the number of energy bins
in the analysis from over one thousand to 60.
These channels are displayed in the figure. It
should be noted that confidence intervals for
parameter estimates, as shown in the Table,
depend on differences of χ−2, and are only
weakly sensitive to choices of binning.
Both the RXTE and ASCA data are col-
lected essentially from the entire cluster, so
that the joint fits here need not consider pos-
sible spectral differences resulting from gra-
dients in the cluster. To account for pos- Fig. 1.— RXTE and ASCA/GIS spectrum
sible inter-calibration errors between the in- of the A3667 with folded Raymond-Smith
struments, adjustable scaling constants are (kT ≃ 7.3) model. ASCA/GIS data are
employed. These are treated as ‘uninterest- shown in (green) asterisks, (black) crosses
ing’ parameters in the fitting, provided (as is are PCA data, and HEXTE data points are
the case here) that they float to values within marked with (red) circles (with 68% error
5 – 15% of unity, consistent with our experi- bars). The total fitted spectrum is shown
ence inanalyzing other sources withthiscom- with a histogram, and deviations between the
bination of instruments. model and data points (normalized to the
We have considered three spectral models: respective standard error) are shown in the
thermal emission from isothermal gas (based lower panel.
on a Raymond-Smith emission code), two-
temperature thermal, and a thermal plus a
keV thermal component, or to 51.3 (56 dof)
power-law model. The hydrogen column was
withanextrapower-lawofindex2.1,asdeter-
fixed at the Galactic value, NH ≃ 4.7 ×1020 mined from the radio spectrum. If the power
cm−2. With the lower threshold at 0.8 keV,
law index is allowed to vary, its best-fit pho-
the GIS data are insensitive to the value of
ton index assumes the rather steep value of
NH unless it is much larger. Indeed, if this 5.0, but it is almost unconstrained. With
value is allowed to float in the fits a best-
four “interesting” parameters, – temperature,
fit is found which is only slightly lower than
abundance, and normalization of the thermal
our assumed value. Using these models, fits
component, plus power-law normalization –
to the joint data provide only weak evidence the change in χ2 (Lampton et al.1976) gives
of the need for an extra component beyond
90% error limits for the 2-20 keV power-law
isothermal. The χ2 of 56.4(57 degrees of free- flux of (0.0−8.1)×10−11 erg-cm−2 s−1. The
dom [dof]) for an isothermal fit is acceptable.
temperature for the main spectral component
Inclusion of a second component reduces χ2 is in the range ∼ 7.3 − 7.5 keV for all three
modestly to 46.6 (55 dof) with an extra 0.9 cases, with formal (1σ) errors of ∼ 0.1 −0.3
4
keV. Femiano et al.(2001); in the 15-35 keV range
This RXTE observation yields evidence for thefluxlimitis4.2×10−12 ergscm−2s−1,mod-
asecond spectralcomponent which isjustsig- erately higher than our limit in the same en-
nificant at the90%confidence level. With the ergy band.
spectral photon index fixed at 2.1, the value
3. Discussion
predicted from the radio spectrum, the best-
fit 2-20 keV flux is (4.0 ± 1.8) × 10−12 erg-
A3667isthefourth(followingComa,A2319
cm−2 s−1. The 90% confidence upper limit on
& A2256) in a sample of clusters with ex-
the 15-35 keV flux is 2.6 × 10−12 erg-cm−2
tended radio emission whose RXTE observa-
s−1 (with four relevant parameters). The
tions were analyzed by us, and the only one
fraction of the total flux in the best-fit sec-
for which we do not find clear evidence for a
ondary component, whether low-temperature
second spectral component. With no spatial
thermal or power-law, is small, of the order
information, theabsence ofevidence forasec-
of a few percent. We report these values, to-
ondary thermal emission component does not
gether with 90% confidence errors, as “sec-
yield useful information on a large scale tem-
ondary flux fraction” in the Table for three
perature gradient. In a cluster with extended
interesting energy bands. Nevertheless, given
radio emission, the main interest in obtaining
the possible effects of background subtraction
anupperlimit onNTemission stems fromthe
and calibration errors, we feel that we cannot
fact that it sets a lower limit on the mean,
claimdetectionofasecondcomponentat90%
volume-averaged value of the magnetic field
confidence, even thoughthis is a formalresult
in the central region of the cluster. With the
of the analysis.
above value of the flux upper limit, and a
∼
A3667 was observed by BeppoSAX for
spectral index of 2.1 inferred from the radio
113 ks in May 1998 and October 1999. Anal- spectrum, we set a lower limit of ∼ 0.4 µG on
ysis of the measurements (Fusco-Femiano et
the mean value of the magnetic field.
al.2001) showed marginal evidence (formally
In assessing themeaning of this limit it has
significant atthe∼ 2.6σ level) forasecondary
to be realized that radio emission in A3667
power-law component, and this only if the
has a complex morphology, with very differ-
gas temperature was fixed at the value (7
ent field values in different regions. From a
keV) determined from previous ASCA mea-
more basic point of view, it should also be re-
surements. Clearly, the need to assume a
membered that several implicit assumptions
value for the temperature, rather than de-
are usually made in linking the synchrotron
termining it in a simultaneous fit to the pa-
and Compton formulae in order to deter-
rameters of both components (a procedure
mine Brx fromradio andX-raymeasurements
thatisdifficulttoaccomplishwithBeppoSAX
(Rephaeli1979,Goldshmidt&Rephaeli1993).
due to the lack of spectral overlap between
These are too often ignored, especially when
the PDS and the lower energy MECS exper-
values of Brx are contrasted with field val-
iments), introduced a substantial uncertainty
ues deduced from co-added statistical analy-
in the deduced significance of any power-law
sis of Faraday rotation measurements (e.g.,,
emission. Therefore, only an upper limit on
Clarke, Kronberg, and B¨ohringer 2001). The
nonthermal emission was reported by Fusco-
latter method is also prone to substantial in-
5
Table 1: Results of the spectral analysis
Parameter Single Thermal Double Thermal Thermal + Power-law
kT (keV) 7.3±0.2 7.5+0.7 7.3±0.2
1 −0.3
kT (keV) 0.9+3.7
2 −0.8
Secondary flux fraction
2-10 keV 0.003+0.155 0.045+0.048
−0.003 −0.045
0.8-40 keV 0.016+0.153 0.052+0.056
−0.016 −0.052
Abundance (solar) 0.22±0.04 0.22±0.05 0.24±0.05
All quoted errors are at the 90% confidence level.
herent (e.g.,, Newman, Newman & Rephaeli Ensslin T.A., et al.2003, ApJ, in press (astro-
2002) and systematic (Rudnick & Blundell ph/0301552)
2003, but see Ensslin et al.2003) uncertain-
Fusco-Femiano, R., et al.1999, ApJ, 513, L21
ties.
Fusco-Femiano, R. et al.2000, ApJ, 534, L7
Since the number of clusters observed at
Fusco-Femiano, R. et al.2001, ApJ, 552, L97
high X-ray energies is small, the observation
of additional clusters (either with or without Fusco-Femiano, R. et al.2003a, Proceedings
known extended radio emission) is of obvi- of‘MatterandEnergyinClustersofGalax-
ous interest, even if only an upper limit is ies’, ASP, Conf. Ser., eds: S.Bowyer &
obtained on NT emission: While observations C-Y.Hwang, astro-ph/0207241
of more clusters with extended radio emission
Fusco-Femiano,R.etal.2003b,astro-ph/0312625
are very desirable, observations of other clus-
Goldshmidt, O., & Rephaeli, Y., 1993, ApJ,
ters, interesting in their own right, are also
411, 518
useful as a control sample.
Gruber, D.E., & Rephaeli, Y. 2002, ApJ, 565,
We thank the referee for useful comments. 877
This project has been supported by a NASA Henriksen, M., et al.1999, ApJ, 511, 666
grant at UCSD.
Henriksen, M., Hudson, D.S., & Tittley, E.
2003, astro-ph/0306341
REFERENCES
Kaastra, J.S. et al.2003, A&A, 397, 445
Bowyer, S., et al.1999, ApJ, 526, 592
Lampton,M.,Margon,B.,&Bowyer, S.1976,
Bowyer, S., et al.2003, ApJ, submitted
ApJ, 208, 177
Clarke, T.E., Kronberg,P.P.,&B¨ohringer, H. Lieu, R., et al.1996, ApJ, 458, L5
2001, ApJ, 547, L111
Markevitch, M. et al.1999, ApJ, 521, 526
6
Newman, W.I., Newman, A.L., & Rephaeli,
Y. 2002, ApJ, 575, 755
Petrosian, V. 2003, proccedings of IAU XXV
Rephaeli, Y. 1977, ApJ, 212, 608
Rephaeli, Y. 1979, ApJ, 227, 364
Rephaeli, Y., & Goldshmidt, O. 1992, ApJ,
397, 438
Rephaeli, Y., & Gruber, D.E. 1988, ApJ, 333,
133
Rephaeli, Y., & Gruber, D.E. 2002, ApJ, 579,
587
Rephaeli, Y., & Gruber, D.E. 2003, ApJ, 595,
137
Rephaeli, Y., Gruber, D.E., & Blanco, P.R.
1999, ApJ, 511, L21
Rephaeli, Y., Gruber, D.E., & Rothschild,
R.E. 1987, ApJ, 320, 139
Rephaeli, Y., Ulmer, M., & Gruber, D.E.
1994, ApJ, 429, 554
Rossetti, M., & Molendi, S. 2003, A&A, in
press (astro-ph/0312447)
Rottgering, H.J.A., et al.1997, MN, 290, 577
Rudnick, L,&Blundell, K.M.2003,ApJ,588,
143
Sarazin, C.L., & Lieu, R. 1998, ApJ, 494,
L177
Vikhlinin, A. et al.2001, ApJ, 551, 160
This 2-column preprint was prepared with the AAS
LATEX macros v4.0.
7