A&A 420, 937-944 (2004)
DOI: 10.1051/0004-6361:20035700
E. van der Swaluw1,2 - T. P. Downes3 - R. Keegan3
1 - Dublin Institute for Advanced Studies, 5 Merrion Square, Dublin 2,
Ireland
2 -
FOM-Institute for Plasma Physics Rijnhuizen, PO Box 1207, 3430 BE Nieuwegein, The Netherlands
3 -
School of Mathematical Sciences, Dublin City University, Glasnevin, Dublin 9, Ireland
Received 17 November 2003 / Accepted 8 March 2004
Abstract
We present a model of a pulsar wind nebula evolving inside its
associated supernova remnant. The model uses a hydrodynamics code to simulate
the evolution of this system when the pulsar has a high velocity.
The simulation distinguishes four different stages of pulsar wind nebula
evolution: the supersonic expansion stage, the reverse shock interaction
stage,
the subsonic expansion stage and ultimately the bow shock stage.
The simulation bears out that, due to the high velocity of the pulsar, the
position of the pulsar is off-centered with respect to its
pulsar wind nebula, after the passage of the reverse shock. Subsequently the pulsar
wind nebula expands subsonically until the event of the bow shock formation, when
the motion of the pulsar becomes supersonic. The bow shock formation event occurs
at roughly half the crossing time, when the pulsar is positioned at 0.677 times
the radius of the supernova remnant blastwave, in complete agreement with analytical
predictions. The crossing time is defined by the age of the supernova remnant when
the pulsar overtakes the blastwave bounding the supernova remnant.
The results of the model are applied to three supernova remnants: N157B, G327.1-1.1 and W44. We argue that the head of the pulsar wind nebula, containing the active pulsar, inside the first two systems is not bounded by a bow shock. However, in the case of W44 we argue for a scenario in which the pulsar wind nebula is bounded by a bow shock, due to the supersonic motion of the pulsar.
Key words: pulsars: general - stars: supernova remnants - shocks waves - hydrodynamics
A young SNR becomes dynamically more interesting in those cases where the
collapse of the progenitor star preceding the supernova explosion yields
a pulsar: a rapidly rotating neutron star. In those cases the dynamics of the
central region of the SNR is dominated by the continuous injection of
energetic particles by a relativistic pulsar wind driven by the spin-down
energy of the pulsar. The pulsar wind is terminated by a strong MHD shock
(Rees & Gunn 1974), and drives a pulsar wind nebula (PWN) in the interior
of the young SNR. The dynamics of the PWN is coupled to the evolution of
the SNR, because the total energy release over the pulsar's lifetime is small
(
erg) compared with the total mechanical energy of the
SNR (
erg).
Several authors (Reynolds & Chevalier 1984; van der Swaluw et al. 2001; Blondin et al. 2001; Bucciantini et al. 2003) have considered the evolution of a centered PWN inside an evolving SNR. In these systems, the initial stage of the PWN is the supersonic expansion stage: the pulsar wind bubble is bounded by a strong PWN shock propagating through the freely expanding ejecta of the SNR. A transition to the reverse shock interaction stage takes place when the reverse shock collides with the PWN shock, and subsequently crushes the pulsar wind bubble (van der Swaluw et al. 2001; Blondin et al. 2001). This reverse shock interaction stage is characterised by an unsteady expansion of the pulsar wind bubble (van der Swaluw et al. 2001), due to the reverberations from the violent collision between the reverse shock and the PWN shock. The expansion of the PWN proceeds subsonically when these reverberations have vanished. The expansion is subsonic because the surroundings of the PWN have been reheated by the passage of the reverse shock: the PWN shock, bounding the swept-up shocked ejecta around the hot pulsar wind bubble, has disappeared.
In this paper we discuss the evolution of PWNe inside SNRs, for pulsars with a constant kick velocity: initially the PWN starts its expansion at the center of the SNR, however the pulsar motion will move the pulsar wind cavity along as it moves through the SNR interior. We present results from hydrodynamical simulations for such a system, which distinguishes all three evolutionary stages mentioned above, i.e. the supersonic expansion stage, the reverse shock interaction stage and the subsonic expansion stage. However at the end of the simulation an additional stage can be distinguished, when the head of the PWN, containing the active pulsar, deforms into a bow shock due to the supersonic motion of the pulsar.
The supersonic expansion stage in the simulation shows a PWN which is off-centered with respect to the twofold shock structure of the SNR, due to the kick velocity of the pulsar. Therefore the timescale on which the reverse shock collides with the complete shock surface bounding the PWN can be a significant fraction of the total lifetime of the PWN when the reverse shock interaction stage starts. We use a semi-analytical approach to estimate the timescale of the collision process, which is shown to scale roughly with the pulsar velocity. The results from the simulation are in almost complete agreement with these semi-analytical calculations. Due to the high velocity of the pulsar, its position inside the PWN is strongly off-centered after the passage of the reverse shock. Ultimately, at the end of the simulation, when the pulsar is approaching the shell of its SNR, the head of the PWN, containing the active pulsar, is deformed into a bow shock at a time and position, in complete agreement with analytical predictions made by van der Swaluw et al. (1998).
We determine the evolutionary stage of the PWNe inside the SNRs N157B, G327.1-1.1, and W44, using our model. For the first two systems, we argue that the position of the pulsar at the head of its PWN is a result of the passage of the reverse shock and the high velocity of the pulsar: the heads of these PWNe are not bounded by a bow shock. Therefore these PWNe are either in the reverse shock interaction stage or the subsonic expansion stage. The PWN inside the SNR W44 however, is shown to be a good candidate for having a bow shock nebula around its pulsar.
The initial stage of PWN evolution is characterised by a hot pulsar wind bubble,
bounded by a strong PWN shock, propagating through the freely expanding ejecta
of the SNR. A transition to the subsonic expansion stage occurs via the
reverse shock interaction stage. This interaction stage starts when the reverse shock collides with
the PWN shock. In the case of a centered pulsar (
), the reverse
shock collides with the PWN shock surface instantaneously, due to the
spherical symmetry of the SNR and the PWN. However, when the pulsar has a kick
velocity, there will be an associated timescale on which the reverse shock collides
with the complete surface of the PWN shock. This is the first stage of the
reverse shock interaction stage. Next, the pulsar wind bubble oscillates back and forward
due to the presence of reverberations from the passage of the reverse shock (van der Swaluw
et al. 2001). The reverse
shock interaction stage ends when these reverberations have vanished and the pulsar
wind bubble proceeds its expansion subsonically. In this section we use a
semi-analytical approach to derive a timescale on which the reverse shock
collides with the complete surface of the PWN shock.
McKee & Truelove (1995) give analytical approximations for the trajectories of the
forward shock and the reverse shock of a SNR in the case of a uniform ambient medium. Their
equations for the trajectories of both shocks are normalised to a timescale
,
which marks the age of the remnant when it has swept up roughly 1.61 the ejected mass
.
Their equations describe the expansion of an isolated SNR in the free
expansion stage and the Sedov-Taylor stage. The trajectory of the forward shock
converges to the Sedov-Taylor solution when the SNR age
.
Internal
(radiative) cooling is not included in their model, therefore the pressure-driven
snowplow stage is not described (see however Blondin et al. 1998).
The reverse shock hits the center of the SNR at approximately 5 times the
transition time
,
putting an upper limit on the age
,
when
the reverse shock has collided with the complete shock surface bounding the PWN:
![]() |
(1) |
An equation for the radius of the PWN shock, when it is
interacting with the freely expanding ejecta of its SNR was given by van der Swaluw et al. (2001), where a constant pulsar wind luminosity
was taken:
![]() |
(3) |
![]() |
Figure 1:
Radii of the forward and the reverse shocks of a SNR with an
explosion energy of
E0=1051 erg and a total ejected mass of
![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 2:
The timescale
![]() |
Open with DEXTER |
We follow van der Swaluw et al. (1998) by calculating the Mach number
of the pulsar as it
propagates through the SNR interior, using the Sedov-Taylor solution
(Sedov 1959). Figure 3 shows the Mach number
as a function
of the age of the remnant t, normalised to the crossing time
.
Alternatively Fig. 4 shows the Mach number
as a function
of the position of the pulsar
,
normalised to the position of the SNR
blastwave
.
![]() |
Figure 3:
The Mach number of the pulsar
![]() ![]() ![]() ![]() |
Open with DEXTER |
The figures show that the PWN head, containing the active pulsar, is deformed into a
bow shock (i.e. when
)
at half the crossing time when the pulsar
is positioned at a radius of 0.677 times the radius of the SNR blastwave.
The Mach number will slowly increase after the formation of the bow shock, due to the
decrease of the sound speed, as the PWN is approaching the shell of the remnant. When
the pulsar breaks through the shell, the Mach number of the pulsar equals
(van der Swaluw et al. 2003).
A lower limit for the pulsar velocity can be derived, such that the bow shock
formation occurs while the SNR is in the Sedov-Taylor stage.
We follow van der Swaluw et al. (2003), who
already give a lower limit for the pulsar velocity in order for the pulsar to
cross the SNR shell in its Sedov-Taylor stage. They use the transition time
calculated by Blondin et al. (1998) as the age of the SNR when the transition
from the Sedov-Taylor stage to the pressure-driven snowplow stage occurs:
![]() |
(5) |
![]() |
(6) |
![]() |
Figure 4:
The Mach number of the pulsar
![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
We use a second order, properly upwinded hydrodynamics code (described in Downes & Ray 1999) to simulate the dynamics of the interaction between a pulsar wind and a supernova remnant when the pulsar has a high velocity. The hydrodynamics equations are intergrated in cylindrical symmetry, and the boundary conditions are taken as gradient zero everywhere except on the r=0 boundary, where they are set to reflecting.
In these simulations, shocks reach the edge of the grid, and hence some inaccuracy will be introduced by the assumption of gradient zero boundary conditions. We can get a feel for the importance of this inaccuracy by considering the shape of the blastwave close to the boundaries (see Figs. 7 and 8). There will be a deformation of the shock near these boundaries if the boundary conditions are causing a significant error in the results. It can be seen that there is little, if any, such deformation. Therefore we conclude that gradient zero boundary conditions are sufficiently accurate for these simulations.
![]() |
Figure 5:
Logarithmic gray-scale plot of the density distribution at an age
![]() |
Open with DEXTER |
![]() |
Figure 6:
Logarithmic gray-scale plot of the density distribution at an age
![]() |
Open with DEXTER |
The simulations are performed in the rest frame of the pulsar. The pulsar
velocity equals
km s-1.
An expanding SNR is created by initialising a sphere of radius 0.25 pc, or
21 grid cells, with a high thermal energy and density such that the total
energy contained in the sphere is
E0=1051 erg, while the ejecta mass is
.
Note that this material also has a velocity of
1000 km s-1 in the negative z direction.
Initially, the ambient density and pressure are uniform with a uniform
velocity of 1000 km s-1 in the negative z direction. The ambient
density is
g/cm-3 (corresponding with
), and the pressure is chosen such that the ambient temperature
is
K. We use a uniform grid on which the resolution has been
taken as
cm.
A realistic pulsar wind is highly relativistic with an adiabatic heat ratio
.
The (shocked) stellar ejecta and ISM material on the other
hand are non-relativistic, with
.
The hydrodynamics code we use
does not include relativistic hydrodynamics, so we make the approximation of
using a value of the adiabatic index
.
Recently Bucciantini et al.
(2003) have performed a simulation, using a relativistic hydrodynamics code, for a
centered pulsar wind nebula inside a supernova remnant. The same simulation was
performed earlier by van der Swaluw et al. (2001) using a non-relativistic
hydrodynamics code (with a value for the adiabatic heat ratio
). The
two simulations show a qualitatively similar and quantitatively almost similar
behaviour. This confirms the validity of the approximation we make in this paper.
The pulsar wind luminosity has been approximated using a constant value, in order to
resolve the pulsar wind termination shock throughout the whole simulation. The results
of the simulation will not change qualitatively, because the total energy injected
by the pulsar during its stay in the SNR interior
.
Therefore this approximation still
ensures that the evolution of the PWN is largely determined by the evolution of the
whole supernova remnant.
![]() |
Figure 7:
Logarithmic gray-scale plot of the density distribution at an age
![]() |
Open with DEXTER |
![]() |
Figure 8:
Logarithmic gray-scale plot of the density distribution at an age
![]() |
Open with DEXTER |
The initial stage of the PWN evolution is the supersonic expansion stage. Figure 5 shows the density profile of the PWN/SNR system during this stage. One can clearly distinguish the four different shocks of the system: the pulsar wind termination shock, the PWN shock, the reverse shock and the forward shock. One clearly observes the off-centered position of the PWN with respect to the twofold shock structure of the SNR, caused by the motion of the pulsar. Furthermore the pulsar wind cavity is roughly spherically symmetric. This results from the high sound speed inside the pulsar wind bubble, which smooths out pressure perturbations rapidly, yielding an almost uniform pressure distribution in the PWN.
Figure 6 shows the density profile of the PWN/SNR system,
when the front of the PWN shock has already collided with the reverse shock. At this
stage the pulsar position starts to get off-centered with respect to the PWN structure.
This is caused by the reverse shock interaction, which raises the presssure inside the
disturbed parts of the bubble, which results in pulsar wind material flowing towards
the back of the PWN. Furthermore the increase in the pressure inside the front part of
the pulsar wind bubble pushes the forward termination shock backwards (towards the pulsar)
as well. On the other hand, the downstream pressure at the backward termination shock
has not been influenced by the reverse shock interaction yet. Therefore the pulsar wind
cavity structure is no longer spherically symmetric, as can be seen in Fig. 6.
The total time on which the reverse shock collides with the complete shock structure bounding
the pulsar wind bubble equals
years in the simulation, after which
the age of the SNR equals
years. These timescales are in almost complete agreement
with the analysis performed in Sect. 2.1 of this paper, which validates the approach made
in that section to derive the collision timescale
.
Figure 7 shows the PWN/SNR system shortly after the passage of the reverse shock. The pulsar is positioned at the head of the PWN, due to the passage of the reverse shock and the high velocity of the pulsar. The PWN structure consists of roughly two parts, a relic PWN and the head of the PWN, as was described in Sect. 2.2 of this paper. After the passage of the reverse shock, the reverse shock interaction stage is continued by the reflected and transmitted shock waves, which propagate through both the pulsar wind bubble and the surrounding ejecta of the SNR. This stage was described in detail by van der Swaluw et al. (2001) for a spherically symmetric PWN/SNR system. We observe a qualitatively similar process, in which the pulsar wind bubble is oscillating. The dynamics are more complicated compared with the simulations of van der Swaluw et al. (2001), due to the asymmetry of the PWN/SNR system caused by the motion of the pulsar. During this stage of the reverse shock interaction, the PWN is oscillating between expansion and compression and the simulations reveal Rayleigh-Taylor and Kelvin-Helmholtz instabilities. These instabilities lead to the mixing of ejecta and pulsar wind material, as was also observed in the simulations performed by Blondin et al. (2001). Therefore at the end of the reverse shock interaction stage, the relic PWN consists of a mixture of pulsar wind material and ejecta. In our simulation the relic PWN will move off the grid, before the bow shock formation occurs around the head of the PWN.
At the end of the simulation, as the pulsar approaches the shell of the SNR, the
head of the PWN, containing the active pulsar, deforms into a bow shock, due to the
supersonic motion of the pulsar. The simulation shows that the bow shock formation
event occurs at roughly
,
when the position
of the pulsar
with respect to the radius of the blastwave
equals
.
This clearly demonstrates the
validity of the analytical approximation made in Sect. 2.2 of this paper.
Figure 8 shows the density profile of the PWN/SNR system after the bow shock formation. The
curvature of the SNR shock is small compared with the bow shock structure, which validates
the assumptions made by van der Swaluw et al. (2003) to model the break-through event.
After the passage of the reverse shock, the PWN consists of a relic PWN and the head of the PWN, containing the active pulsar. Figures 7 and 8 show the density profiles of the PWN/SNR before and after the formation of the bow shock. For both cases the pulsar is located inside the head of the PWN. The synchrotron maps from PWNe will therefore show a similar morphology, which makes it hard to determine the presence of a bow shock based on the observed PWN/SNR morphology. However, we have demonstrated that the following criteria can be used to determine the presence of a bow shock around the head of the PWN:
N157B is a young SNR dominated by plerionic emission from the PWN inside this
remnant. The age of the remnant is approximated to be 5000 yrs (Wang &
Gotthelf 1998) and contains a 16 ms pulsar (Marshall et al. 1998). The velocity
of the pulsar is high (
km s-1), if one assumes that the pulsar
was born at the central region of the bright radio emission (Lazendic et al. 2000).
Wang & Gotthelf (1998) argue for a bow shock interpretation of the PWN in N157B:
the supersonic motion of the pulsar has deformed the PWN around the pulsar into a
bow shock. In this scenario the SNR N157B is a young variant of the SNR CTB80, which
is thought to have a PWN bow shock located close to the shell of the remnant (see e.g.
Strom 1987; Kulkarni et al. 1988; Migliazzo et al. 2002). It is remarkable
though, that the spindown luminosity of the pulsar inside N157B (
erg/s) is so much higher compared with the spindown luminosity from the
pulsar inside CTB80 (
erg/s). If there is a bow shock
around the PWN inside N157B, this implies that an upper limit for the age of the SNR,
when the pulsar crosses the shell, is approximately 10 000 years (using the current
age of 5000 years and the criterion that bow shock creation occurs at half the crossing
time). This is again a remarkable contrast with the current age of CTB80 of 100 000 years,
which is close to the crossing time of this SNR. Furthermore from Fig. 2 of Wang et al.
(2001) it seems as if the position of the pulsar is more or less centered in the SNR. This
is in contrast with what one would expect from the analysis performed in Sect. 2.2 of
this paper, which predicts
.
We propose a scenario for SNR N157B in which the contradictions mentioned above disappear. Following the results from Sects. 2 and 3, we suggest that the central bright parts of the radio and X-ray emission inside N157B correspond with the relic PWN, being blown in the initial stage of the PWN, when it was expanding supersonically. The PWN inside N157B has just collided with the reverse shock; this collision and the high velocity of the pulsar has off-centered the pulsar with respect to the PWN structure as was discussed in the Sects. 2 and 3.3 of this paper. Therefore the head of the PWN is not bounded by a bow shock: the PWN is in the aftermath of the reverse shock interaction stage or in the beginning of the subsonic expansion stage.
Another example of a young SNR, containing a PWN with the position of the pulsar
off-centered with respect to its PWN is G327.1-1.1. No pulsar has been detected
inside this remnant so far. Sun et al. (1999) presented a radio contour
map using MOST data overlaid with X-ray data from ROSAT. The X-ray emission is
centered around a finger of radio emission sticking out of a central radio bright
region, indicating the presence of a pulsar wind. Following Sun et al. (1999) the
SNR can be modelled in X-rays by the
following set of parameters:
E51=0.23, n0=0.10,
km s-1 and
an age of
.
Using Eq. (4) to calculate
we get
a value of
years. The age of the system therefore is less than half
the crossing time, implying the absence of a bow shock. Furthermore the displacement
of the PWN finger (i.e. the head of the PWN), containing the pulsar, does not satisfy
the other criterion for the presence of a bow shock, i.e.
.
This criterion is less restrictive, since one might
introduce an angle between the pulsar velocity and the observer such that the ratio
.
Notice however that this will not influence the age
of the remnant!
Therefore we propose a scenario for G327.1-1.1 in which there is no bow shock around the
head of the PWN: the pulsar has been positioned at the head of the PWN due to the high
velocity pulsar and the passage of the reverse shock. The central bright part of the remnant
corresponds with the relic PWN. The finger of the PWN corresponds with the head of the
PWN, which contains the active pulsar. Because of the larger value of the ratio
and the larger age with respect to the PWN inside SNR N157B we
favour the scenario where the PWN is in its subsonic expansion stage.
The SNR W44 is an older remnant than the previous two SNRs discussed.
Furthermore only a small fraction of the radio emission from this remnant is
characterised by plerionic emission (Frail et al. 1996). Taking the
characteristic age of the pulsar, 20 000 years (Wolszcan et al. 1991), as
the age of this remnant yields an upper limit of
years
for the crossing time of this remnant. This age is much closer to
the age of the SNR CTB80. The displacement of the pulsar position only marginally
violates the condition for bow shock formation, i.e.
.
We therefore conclude that the PWN observed in the SNR W44 corresponds with the
head of the PWN, which has been deformed into a bow shock.
We have considered the case of a PWN interacting with a SNR, for which the associated pulsar is moving at a high velocity through the interior of its SNR. The model we discussed made use of a hydrodynamics code. One could distinguish four different stages in the simulation: the supersonic expansion stage, the reverse shock interaction stage, the subsonic expansion stage and the bow shock stage. Below we summarise the most important results from our model:
1) a roughly spherically symmetric relic PWN;
2) a head, containing the pulsar, directed towards the SNR shell.
Acknowledgements
This work was part-funded by the CosmoGrid project, funded under the Programme for Research in Third Level Institutions (PRTLI) administered by the Irish Higher Education Authority under the National Development Plan and with partial support from the European Regional Development Fund.