Transcript ppt

Timing Analysis of the Isolated Neutron Star RXJ0720.4–3125
Silvia Zane, Frank Haberl, Mark Cropper, Vyacheslav Zavlin,
David Lumb, Steve Sembay & Christian Motch.
(MSSL/UCL, University of Strasbourg, MPE Garching, ESTEC, Univ. of Leicester)
A: Introduction
RXJ0720.4–3125 is a nearby, isolated Neutron Star
detected by ROSAT during a Galactic plane survey
(Haberl et al., 1997) and recently re-observed with XMMNewton on 2000 May 13 (Paerels et al. 2001, Cropper et
al., 2001) and 2001 November 21.
B:Timing Analysis
Table 1 shows the different observations used in our
analysis. The major datasets are from the two XMM
observations, and from the 1996 Nov. 3 Rosat pointing.
Table 1
We derive two pairs of values P, dP/dt which cannot be
further discriminated between on statistical grounds
(Table 2). Both fits have dP/dt 3-6 10-14 s/s. This is
the most accurate spin-down measure presented so
far for a dim NSs and, for the first time, it allows a
discrimination between models.
The source shows all the common characteristics of
the other 6 ROSAT Neutron Stars candidates (dim NSs,
see Treves et al., 2000). In particular:
• high X-ray to optical flux ratio Lx/Lopt>1000
• soft X-ray thermal spectrum (Tbb~ 86 eV)
• low column density (NH ~ 6 x1019 cm–2 ).
The source is also pulsating with P ~ 8.4 s.
Until a few years ago two mechanisms were proposed
for dim NSs: accretion from the interstellar medium
onto an old NS or cooling of a younger object. More
recently, based on the similarity of the periods, it has
been suggested a possible evolutionary link between
dim NSs, Anomalous X-ray pulsars (AXPs), and soft
gamma-ray repeaters (SGRs).
Two kind of “unified'’ scenarios have been then
proposed. In the first one, the 3 classes are powered
by dissipation of a decaying, super-strong magnetic
field (B 1014 -1015 G). In this case, dim NSs are the
descendants of SGRs and AXPs and RXJ0720.4-3125
may be the closest old-magnetar. Alternatively, the 3
classes may contain standard NSs (B  1012 G)
endowed by a fossil disk (Alpar, 2001). In this case,
dim NSs in the propellor phase would be the
progenitors of AXPs and SGRs, the latter having
entered an accretion phase.
Our data originate from instrumentation with widely different sensitivities:
typical count rates vary from 0.3 ct/s for Rosat HRI to  6 ct/s for
XMM/PN. However, none of these count rates is sufficiently high for a
normal distribution of counts to be expected, thus standard discrete Fourier
Transforms are not directly applicable. For sparse data and event list data,
we use instead Rayleigh Transforms (i.e. de Jager 1991, Mardia 1972). It
is also crucial for us to define precisely the confidence intervals for the
derived quantities, in particular the period P. We do this by constructing
MLP (maximum likelihood periodogrammes), which make no assumptions
on data distribution, and using the C-statistics (Cash 1979). The
uncertainty in the period and the 2 can be read directly from the y-axis of
the MLP (see e.g. Fig.1).
We begin performing an MLP assuming dP/dt=0 on each of
the longer pointing: R93, R96b, X00a, X00b (see Fig.1).
There is no ambiguity in the period determinations and a
linear least square fit using the 68% formal errors in the
MLP gives P0 = 8.39113 ± 0.00011 s, dP/dt = 0.0 ± 5.5 10-13
s/s (P0 is referenced to the start of the R93 run).
Further information about this puzzling source can
be obtained by the spin history. Magnetars will
spin-down at a rate dP\dt 10-11(B/10 14 G)2/P ss-1 ,
due to magneto-dipolar losses. A measure of dP/dt
for RXJ0720.4-3125 is therefore crucial, as well an
accurate tracking of its spin history.
Here we present a combined timing analysis of
XMM, Chandra and Rosat data, spanning a period
of ~7 years.
Recently, there are some indications that the time-stamping of XMMNewton data may need fine-adjustments at the level of our derived
accuracy. Further investigations by the SOC are in progress and, if
necessary, a re-computation of some of these results may be
required.
Fig.3: The 68 and 90% MLP contours
for the two best-fit solutions of Table
2. Left: solution (1); right: solution (2).
Fig.4: The datasets folded on periods
(1) (left) and (2) (right) of Table 2.
C: Discussion
Recently, Paerels et al. (2001) presented XMM spectra of RXJ
0720.4-3125. The absence of electron or proton cyclotron
resonance in the RGS range excluded magnetic fields of (0.3–
2.0)1011 and (0.5–2.0)1014 G (see Zane at al., 2001).
Based on the same XMM observation, Cropper et al. (2001)
presented the pulse-shape analysis. They derived an upper limit
on the polar cap size, showing that an emitting region larger than
60°-65° can be rejected at a confidence level of 90%. Whatever
the mechanism, the X-ray emitting region is therefore confined to
a relatively small fraction of the star surface. They also found that
the hardness ratio is softest around the flux maximum. The same
has been later discovered by Perna et al. (2001) in some AXPs.
Cropper et al. (2001) suggested two possible explanations: either
radiation beaming (as in their best-fitting model) or the presence
of a spatially variable absorbing matter, co-rotating in the
magnetosphere. The latter may be the case if the star is propelling
matter outward (Alpar, 2001).
Table 2. The first two pairs are the best-fit (P,dP/dt)
values. 2 is the difference between the 2 of a
given solution and that of solution (1). See Fig.3 for
errors.
Fig.1. Left: MLPs for three long datasets, R96d, X00a (PN) and X00b (PN), showing the periodicity at 8.391 sec. These
constrain the selection of the strongest and second-strongest dips in the MLPs for the R98 and Ch00 datasets respectively
(right). The vertical line denotes a period of 8.39113 sec. The 68% and 90% confidence levels are at 2 = 1.0 and 2.71 for
one degree of freedom.
This upper limit in dP/dt=0
permits an unambiguous
determination of the peaks
in the Ch00 and R98 power
spectra. Adding these to the
linear square fit gives P0 =
8.39107 ± 0.00005 s, dP/dt =
2.7 10-13 ± 2.5 10-13 s/s.
The 68, 90 and 99%
confidence level are shown
in Fig.2, as well as the 68
and 90% intervals derived
from
X00a.
With
the
improved (P0, dP/dt) values,
we perform an MLP on the
combined ROSAT 1993 and
1996 datasets. As a result,
the
confidence
contour
break up into small region
(aliases) in the (P0, dP/dt)
plane (see zoom in Fig.2).
With this further restriction,
we finally do the MLP on all
the data.
The refined value of dP/dt reported here is consistent with
the measure of Haberl et al. (1997), but two orders of
magnitude lower. The first implication is that RXJ0720.43125 is hardly to be spinning down due to a propeller
torque. Accretion from a fossil disk implies 2 10-11d2100<
dP/dt < 2 10-9d2100 s/s, where d100 is the sources distance
normalized at 100 pc: the value reported here is well below
this range. On the other hand, the measured spin-down is
considerable and, if interpreted as due to magneto-dipolar
losses, it gives a magnetic field as high 21013 G. The
corresponding spin-down age is P/(2dP/dt)  3106 yrs,
which is higher but, given the numerous uncertainties, not
too far from the cooling age of 5105 yrs.
Table 3. Predicted source age and primordial field for three different mechanisms of Bdecay, simulated as in Colpi et al. (2000). For each decay law, the two solutions correspond
to the two best-fitting pairs of value of Table 2. In all cases, the source is assumed to be born
with P = 1ms.
Fig.2. The 68, 90 and 99% contour for a linear least squares fit of
R 93, R96d, R98, X00a, X00b and Ch00 (continuum elliptical
regions). Parallel lines are the 68 and 90% contour of X00a PN;
tiny elliptical regions are the 68 and 90% contour for the combined
R93 and R96 datasets (see zoom).
It is now fundamental to assess the field evolution and to
understand if the source came through an history of B-decay
or if the magnetic field has been almost constant over its
lifetime. We take for simplicity three different models for the
field decay: Hall cascade and ambipolar diffusion in the
solenoidal or irrotational mode. The laws are taken as in Colpi
et al. (2000). As we see, fast decaying processes as Hall
cascade predict a very low age for the source, which is difficult
to reconcile with its present luminosity and with the relatively
large number of detected close-by objects of this class. It
seems therefore more plausible that the B-field of RXJ0720.43125 only had a relatively small change over the evolution, in
which case the present source age is  106 yrs.