Mon. Not. R. Astron. Soc. 356, 359370 (2005) doi:10.1111/j.1365-2966.2004.08437.

x
1E 1207.45209: a low-mass bare strange star?
R. X. Xu

School of Physics, Peking University, Beijing 100871, China

Accepted 2004 September 24. Received 2004 September 24; in original form 2004 August 2
ABSTRACT
Both rotation- and accretion-powered low-mass bare strange stars are studied, with particular
regard to their astrophysical appearance. It is suggested that low-mass bare strange stars,
with weaker ferromagnetic elds than that of normal pulsars, could result from the accretion-
induced collapse of white dwarfs. According to its peculiar timing behaviour, we propose
that the radio-quiet object, 1E 1207.45209, could be a low-mass bare strange star with a
polar surface magnetic eld of 6 10
10
G and a radius of a few kilometres. The low-mass
bare strange star idea is helpful in distinguishing neutron and strange stars, and is testable by
imaging pulsar-like stars with the future Constellation-X telescope.
Key words: dense matter elementary particles stars: neutron pulsars: general pulsars:
individual: 1E 1207.45209.
1 I NTRODUCTI ON
Astrophysics offers an alternative channel for us to explore the fun-
damental laws of nature, and the study of quark stars obeys exactly
this spirit. It is a clear goal for laboratory physicists to nd a quark
gluon plasma (or quark matter) in order to study the problem of the
elemental colour interaction, whereas the detection of astrophysical
quark matter may offer a shortcut. Though one may conventionally
think that pulsars are normal neutron stars (Lattimer & Prakash
2004), it is still an open issue whether pulsar-like stars are neutron
or quark stars (Madsen 1999; Weber 1999; Glendenning 2000), as
no convincing work, either theoretical from rst principles or ob-
servational, has conrmed BaadeZwickys original idea that super-
novae produce neutron stars. Therefore, the question of detecting
astrophysical quark matter is changed to be: how can one identify a
quark star?
One kind of frequently discussed quark stars are those with
strangeness, namely strange stars, which are very likely to exist.
A few ways have been proposed (e.g. cooling behaviour, mass
radius relations, etc.) by which neutron and strange stars could be
distinguished. However, the peculiar nature of a quark surface was
not been noted until 1998 (Usov 2002; Xu 2003c). As the strange
star model (Xu 2003d) may work well for understanding the various
observations of pulsar-like stars, including glitches and free preces-
sions (Zhou et al. 2004), there are at least three natural motivations
for studying such stars with low masses.
(1) The formation of low-mass strange stars is a direct conse-
quence of the presumption that pulsar-like stars are actually quark
stars rather than neutron stars. One cannot rule out this unfortunately
neglected possibility now either from a rst-principle approach or

E-mail: rxxu@bac.pku.edu.cn
astrophysical observations. This paper is an attempt to draw astro-
physicists attention to this relevant investigation.
(2) A low mass may be helpful for identifying strange stars.
Bare strange stars can be very low mass with small radii, while
normal neutron stars cannot. It is well known that the masses
and radii of neutron and strange stars with almost the maximum
mass are similar;
1
none the less, low-mass neutron and strange
stars have remarkably different radii (Alcock, Farhi & Olinto 1986;
Bombaci 1997; Li et al. 1999). As a result of the colour conne-
ment by itself rather than gravitational binding, a bare strange star
could be very small, e.g. the radius with mass M M

is, from
equation (10),
R = 1.04 10
6

B
1/3
60
(M/M

)
1/3
cm, (1)
and R = (4.8, 2.3, 1.0) km for M = (10
1
, 10
2
, 10
3
) M

if
the bag constant

B = 60 MeV fm
3
. However, the radii of neu-
tron stars with 0.5 M

are generally greater than 10 km, and the

minimum mass of a stable neutron star is 10
1
M

, with radius
R 160 km (approximately two orders of magnitude larger than
that of low-mass bare strange stars with similar masses) (Shapiro
& Teukolsky 1983). It is consequently possible that we could
distinguish neutron and strange stars by direct measurements of
the radii
2
of low-mass pulsar-like stars by using X-ray satellites.
Fortunately, low-mass neutron stars have been noted recently
1
This is the reason that it is generally believed that neutron and strange stars
cannot be distinguished by measuring only their masses or radii, and that one
should try to compare the observationally determined massradius relations
with the theoretical ones in order to identify strange stars.
2
Also the very distinguishable massradius (M R) relations of neutron
and quark stars are helpful. Astudy of the LaneEmden equation with n =
2
3
(corresponding to the state of a non-relativistic neutron gas with low mass)
results in M R
3
, whereas, for low-mass quark stars, M R
3
, due to the
colour connement of quark matter.
C _
2004 RAS
360 R. X. Xu
(Carriere, Horowitz & Piekarewicz 2003), the radii of which may
correlate with that of
208
Pb because of the stellar central densities
being near the nuclearmatter saturation density. It is then worth
studying low-mass strange stars in order to obtain crucial evidence
for quark stars.
(3) The conventional method for estimating the polar magnetic
elds of radio pulsars has to be modied if we have no denite
reason to rule out the possibility that low-mass bare strange stars
could exist in the Universe (see Section 3.1 for details).
Additionally, the identication of a low-mass strange star, with
mass 0.1 M

, may also tell us whether the star is bare, as the

radius of a bare strange star is much smaller than that of one with a
crust in the low-mass limit (Xu 2003a; Bombaci, Parenti & Vida na
2004).
The layout of the rest of this paper is as follows. First, a phe-
nomenological view of strange quark matter is introduced in Sec-
tion 2. After a study of the general nature of rotation-powered and
accretion-driven low-mass bare strange stars in Section 3, we fo-
cus our attention on the central compact object 1E 1207.45209
in Section 4, with some investigations of other potential candidates
being given in Section 5. Although the major points of the paper are
to propose candidates for low-mass quark stars according to astro-
physical observations, an effort to probe into the origin of such stars
is attempted in Section 6. Finally, conclusions and discussions are
presented in Section 7.
2 QUARK MATTER PHENOMENOLOGY
Strange quark stars are composed of quark matter with almost equal
numbers of u, d and s quarks. There are actually two different kinds
of quark matter to be investigated in laboratory physics and in as-
trophysics, which appear in two regions in the quantum chromody-
namics (QCD) phase diagram (Fig. 1). Quark matter in laboratory
physics and in the early Universe is temperature-dominated (tem-
perature T 0, with baryon chemical potential
B
0), while that
in quark stars or as cosmic rays is density-dominated (T 0,
B

0). Previously, Monte Carlo simulations of lattice QCD (LQCD)

were only applicable for cases with
B
= 0. Only recently have
attempts been made with
B
,=0 (quark stars or nuggets) in LQCD.
We have then to rely on phenomenological models to speculate on
the properties of density-dominated quark matter.
In different locations on the diagram (Fig. 1), besides the interac-
tion strength between quarks and gluons being weak or strong, the
vacuum would have different features and is thus classied into
two types: the perturbative-QCD (pQCD) vacuum and the non-
T

E
a
r
l
y

U
n
i
v
e
r
s
e
Quark stars?
Quark-Gluon Plasma
Hadrons
2SC
CFL
Clustering:
fluid QM
Clustering:
solid QM
< qq > = 0
< qq > = 0
< qq > = 0
Figure 1. Schematic illustration of the QCD phase diagram.
perturbative-QCD vacuum. The coupling is weak in the former,
but is strong in the latter. Quarkantiquark (and gluon) condensa-
tions occur in a QCD vacuum (i.e. the expected value of qq) ,= 0),
but not in a pQCD vacuum. The chiral symmetry is spontaneously
broken in the case where the vacuum is changed from a pQCD to
a QCD vacuum, and quarks then become massive constituent ones.
LQCD calculations (Kogut et al. 1991) show that the value of qq)
increases when the colour coupling becomes strong (i.e. as the tem-
perature or the baryon density decreases). Therefore, we note that
the quark deconnement and the chiral symmetry restoration may
not take place at the same time.
Considerable theoretical efforts have been made to explore the
QCDphase diagram. When T or
B
are extremely high, there should
be a quarkgluon plasma (QGP) phase because of the asymptotic
freedom, and the vacuum is of the pQCD type. However, in a rel-
atively lower-energy limit, especially in the density-dominated re-
gion, the vacuum is phase-converted to a QCD one, but the quarks
couldbe still deconned. The investigationof the possibilitythat real
quarks may also be condensed (i.e. qq) ,=0) simultaneously when
qq) ,= 0, so-called colour-superconducting (CSC) phases is an ac-
tive area of research (for recent reviews, see, e.g., Ren 2004; Rischke
2004). Actually two CSC phases are currently discussed. One cor-
responds to Cooper pairing among the two avours of quarks (u
and d) only, called the two-avour colour superconductivity (2SC)
phase, in the case where the s quark is too massive to participate.
Another one occurs at higher
B
in which s quarks are relatively
less massive and are thus involved in Cooper pairing, this is called
the colouravour locked (CFL) phase.
However, another possibility cannot be ruled out: qq) = 0
while qq) ,= 0. When T is not high, along the reverse direction
of the
B
axis, the value of qq) increases, and the colour cou-
pling between quarks and gluons becomes stronger and stronger.
The much stronger coupling may favour the formation of n-quark
clusters (where n is the number of quarks in a cluster) in this
case (Xu 2003b). Such quark clusters could very likely be anal-
ogous to clusters moving in nuclei, which are well known
in nuclear physics. Recent experimental evidence for multiquark
(n > 3) hadrons may increase the possibility of quark clustering.
The clusters are localized
3
to become classical (rather than quan-
tum) particles when the thermal de Broglie wavelength of clusters
h/

3mkT < l [3n/(4 f n

b
)]
1/3
(where m is the mass of
the clusters, l is the mean cluster distance, n
b
is the baryon num-
ber density and f is quark avour number), assuming no interaction
between the clusters. Calculations based on this inequality for f =
3 show that cluster localization still exists even at temperatures of
T 1 MeVif n 10
2
. In addition, the interaction in between, which
is neglected in the inequality, would also favour this localization.
In the case of a negligible interaction, quark clusters would be-
come a quantum system at low temperatures. However, the in-
teraction is certainly not as weak as for the vacuum for QCD
( qq) ,= 0).
Now, a competition between condensation and solidication ap-
pears, as in the case of laboratory low-temperature physics. Quark
matter would be solidied as long as the interaction energy be-
tween neighbouring clusters is much larger than that of the ki-
netic thermal energy. This is why only helium, of all the elements,
shows superuid phenomenon though other noble elements have a
3
By local we mean that quark wavefunctions almost do not overlap. In
this sense, localized clusters can still move from place to place when T is
high, but could be solidied at low T.
C _
2004 RAS, MNRAS 356, 359370
1E 1207.45209: a strange star? 361
similar weak interaction strength due to lled electrons shells. The
essential reason for the occurrence of CSC is that there is an attrac-
tive interaction between two quarks at the Fermi surface. However,
as discussed, a much stronger interaction may result in quark clus-
tering and thus a solid state of quark matter. In conclusion, a new
phase with qq) ,= 0 but qq) = 0 is suggested for insertion in the
QCD phase diagram (Fig. 1), which could exist in quark stars.
Astrophysics may teach us about the nature of density-dominated
quark matter in the case of these theoretical uncertainties. There are
at least two astrophysical implications arising from the proposed
solid quark matter state.
(1) Quark stars in a solid state can be used to explain naturally the
observational discrepancy between glitches and the free-precession
of radio pulsars (Link 2003), as a solid quark star is just a rigid-like
body (no damping precession), and glitches would be the result of
star quakes. Actually, it is found (Zhou et al. 2004) that the general
nature of the glitches (i.e. the glitch amplitudes and time intervals)
could be reproduced if the solid quark matter has properties of shear
modulus = 10
3034
erg cm
3
and critical stress
c
= 10
1824
erg cm
3
.
(2) Ferromagnetization may occur in solid quark matter, without
eld decay in the stars. The magnetic eld plays a key role in the
life of a pulsar, but there is still no consensus on its physical origin
although some relevant ideas (e.g. ux conservation during collapse
and dynamo actions) have appeared in the literature. Whereas, an
alternative suggestion is the generation of strong magnetic elds
by spontaneously broken ferromagnetism in quark matter (Tatsumi
2000). One of the advantages of a ferromagnetic origin could be their
unchangeable nature as there is no convincing evidence that elds
decay in isolated pulsar-like stars.
4
However, the magnetic domain
structure may be destroyed by the turbulent motion in a uid quark
star. This worry does not exist if pulsars are solid quark stars (Xu
2003b). Quark clusters with magnetic momentummay exist in solid
quark stars. Solid magnetic quark matter might then magnetize it-
self spontaneously at sufciently low temperatures (below its Curie
critical temperature) by, for example, the ux-conserved eld. Fer-
romagnetic saturation may result in a very strong dipole magnetic
eld. We therefore presume a ferromagnetic origin for pulsar elds
in the following calculations.
3 LOW- MASS BARE STRANGE STARS
3.1 Rotation-powered phase
The energy conservation for an orthogonal star (i.e. the inclination
angle between the magnetic and the rotational axes is =90
o
) with
a magnetic dipole moment , a moment of inertia I and an angular
velocity gives

=
2
3I c
3

3
. (2)
This rule is maintained quantitatively for any , as long as the brak-
ing torques due to magnetodipole radiation and the unipolar genera-
tor are combined (Xu &Qiao 2001). For a star with a polar magnetic
eld B and radius R, the magnetic moment
=
1
2
BR
3
, (3)
4
The apparent eld decay in the P

P diagramcould have arisen fromknown
selection effects, based on the simulations (Wakatsuki et al. 1992).
if the elds are in a pure dipole magnetic conguration or if the star
is a uniformly magnetized sphere. This results in a conventional
magnetic eld derived from P and

P(P = 2/is the spin period)
of
B = 6.4 10
19

P

P G, (4)
if typical values of I = 10
45
g cm
2
and R = 10
6
cm are assumed.
Note that the eld is only half the value in equation (4) if one simply
suggests = BR
3
(Manchester & Taylor 1977).
However, I and R for neutron stars change signicantly for dif-
ferent equations of state, or for different masses even for a certain
equation of state (Lattimer & Prakash 2001). This means that the
typical values may actually be not approximately constant. The
inconsistency becomes more serious if pulsar-like stars are in fact
strange quark stars as such a star could be as small as a few hundred
baryons (strangelets).
Let us compute P and

P for a quark star with a certain mass M
and radius R. First, we approximate the momentum of inertia to be
I . 2MR
2
/5 (i.e. a star with uniform density). This approximation
is allowed for low-mass strange stars (Alcock et al. 1986). In this
case, the magnetic eld derived from P and

P is then (Xu, Xu &
Wu 2001), from equations (2) and (3),
B =

0.6P

Pc
3/2
M
1/2
R
2
/
= 5.7 10
19
M
1/2
1
R
2
6

P

P G, (5)
where M
1
= M/M

and R = R
6
10
6
cm. As the strong magnetic
elds of pulsars are suggested to be of ferromagnetc origin, we then
assume the magnetic momentum to be
= aM

bR

, (6)
where {a, ; b, } is a parametric set. If the magnetized momentum
per unit volume is a constant
v
, one has {a =0, =0; b =4
v
/3,
=3}. In the case where the magnetized momentum per unit mass
is a constant
m
, one then has {a =
m
, = 1; b = 0, = 0}.
From equations (2) and (6), one arrives at
P

P =
20
2
(aM

bR

)
2
3c
3
MR
2
. (7)
This equation shows that a pulsar with certain initial parameters (M
and R) evolves along constant (P

P). Integrating equation (7), one
obtains the pulsar age T,
T =
P
2
P
2
0
2P

P
, (8)
where P
0
is the initial spin period. The age T = T
c
P/(2

P) when
P
0
_ P.
The mass of a pulsar is detectable dynamically if it is in a binary
system, but precise mass estimates are only allowed by the mea-
surement of relativistic orbital effects. It is therefore necessary to
determine the model parameters (
m
and
v
) in equation (7) from
pair neutron star systems (Thorsett & Chakrabarty 1999; Lyne et al.
2004). Four possible cases are investigated and the results of these
calculations are listed in Table 1. The massradius relations for the
calculation are in the regime of strange quark matter described by
a simplied version of the MIT bag model, in which the relation
between the pressure P and the density is given by
P =
1
3
( 4

B), (9)
where the bag constant

B is chosen to be 60 MeV fm
3
(lower
limit) and 110 MeV fm
3
(upper limit), respectively. It is assumed
that the mass and radius of a star do not change signicantly during
quark-clustering and solidication as the star cools.
C _
2004 RAS, MNRAS 356, 359370
362 R. X. Xu
Table 1. Pair neutronstars andthe model parameters derived. Models numbered14are for {a =
m
, = 1; b = 0, = 0;

B = 60MeV fm
3
], {a =
m
, =
1; b = 0, = 0;

B = 110 MeV fm
3
], {a = 0, = 0; b = 4
v
/3, = 3;

B = 60 MeV fm
3
and a = 0, = 0; b = 4
v
/3, = 3;

B = 110 MeVfm
3
,
respectively, where

B is the bag constant. The former two are for constant magnetic momentum
m
(G cm
3
g
1
) per unit mass, while the latter two are for
constant momentum
v
(G) per unit volume.
Pulsars P (ms)

P (s s
1
) M ( M

)
m
(model 1)
m
(model 2)
v
(model 3)
v
(model 4)
J1518
a
40.94 2.73E20 1.56 4.27E7 2.31E8
B1534
a
37.90 2.42E18 1.34 4.00E6 3.10E6 2.10E9 3.52E9
B1913
a
59.03 8.63E18 1.44 9.25E6 6.89E6 4.98E9 8.99E9
B2127
a
30.53 4.99E18 1.35 5.15E6 3.97E6 2.72E9 4.57E9
B2303
a
1066 5.69E16 1.30 3.28E4 2.56E4 1.71E11 2.80E11
J0737A
b
22.70 1.74E18 1.34 2.62E6 2.03E6 1.38E9 2.31E9
J0737B
b
2773 0.88E15 1.25 6.65E4 5.19E4 3.43E11 5.61E11
Note: a denotes Thorsett & Chakrabarty (1999), b denotes Lyne et al. (2004).
The values of
m
and
v
are grouped into two classes in
Table 1. One class has higher
m
or
v
for normal pulsars (B2303
and J0737B only), but the other (of millisecond pulsars) has a lower

m
or
v
. We just average the values of B2303 and J0737Bfor indi-
cations of normal pulsars; model 1,
m
= 4.97 10
4
G cm
3
g
1
;
model 2,
m
=3.88 10
4
G cm
3
g
1
; model 3,
v
=2.57 10
11
G and model 4,
v
=4.21 10
11
G. According to equation (7) and
the massradius relations of strange stars [calculated numerically,
using the state equation (9) and the TolmanOppenheimerVolkoff
(TOV) equation], choosing the model parameters above, we can nd
values of P

P for certain masses, which are represented in Fig. 2.
It is found that the P

P value is limited by either the maximum
mass (models 1 and 3) or by massradius relations (models 2 and
4). The maximum P

P value could be a few 10
15
s.
In model 1 with
m
=5 10
4
Gcm
3
g
1
, a pulsar with a certain
mass (2, 1.5 and 1 M

, respectively) evolves along constant P

P
(solid lines in Fig. 3).
These three lines pass through almost the middle region of nor-
mal pulsars. It is thus suggestive that the distribution of scattered
points of normal pulsars could be the result of the variation of
m
,
rather than that of the pulsar mass. Actually, if the mass is xed to
be 1.5 M

, most of the normal pulsars are between

m
=5 10
3
and 5 10
5
G cm
3
g
1
. However, in model 1 with lower
m
for
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
0.5
1
1.5
2
2.5
x 10
15
Mass (in solar mass)
P

P
d
o
t
Model 1
Model 2
Model 3
Model 4
Figure 2. The P

P value based on equation (7). Model 1,
m
= 4.97
10
4
G cm
3
g
1
; model 2,
m
= 3.88 10
4
G cm
3
g
1
; model 3,
v
=
2.57 10
11
G and model 4,
v
= 4.21 10
11
G.
10
3
10
2
10
1
10
0
10
1
10
22
10
20
10
18
10
16
10
14
10
12
10
10
10
8
Spin period
P
e
r
i
o
d

d
e
r
i
v
a
t
i
v
e
a
c
b
d
e
Figure 3. The P

P diagram of pulsars. Three solid lines from the top are
for constant P

P values of pulsars with masses of 2, 1.5 and 1 M

, re-
spectively, in model 1 (
m
= 5 10
4
G cm
3
g
1
). Dotted lines are
also for constant P

P, but lower
m
(10
6
G cm
3
g
1
). These 10 dotted
lines from the top are for pulsars with mass M

of 10
1
, 10
2
, 10
3
,
10
4
, 10
5
, 10
6
, 10
7
, 10
8
and 10
9
M

, respectively. The lines la-

belled a, b, c, d and e are of constant potential drops . See the text for
the parameters for these ve lines. The pulsar data were downloaded from
http://www.atnf.csiro.au/research/pulsar/psrcat.
millisecond pulsars (e.g. 10
6
G cm
3
g
1
), the line with mass
1 M

of constant P

P does not pass through the centre of millisec-
ond pulsar points. It is unlikely that the variance of
m
is responsible
for this unless the
m
values of those millisecond pulsars listed in
Table 1 are not representative. Therefore, we propose alternatively
that some of the millisecond pulsars could be low-mass bare strange
stars, with mass _M

. Similar conclusions can be drawn fromthe

other models (i.e. models 24).
For low-mass bare strange stars, the value P

P can be obtained
analytically. The mass (M

) of bare strange stars can be well

approximated by (because of equation 9, Alcock et al. 1986),
M =
4
3
R
3
(4

B), (10)
where

B = (60110) MeV fm
3
, i.e. (1.071.96) 10
14
g cm
3
.
Combining equations (7) and (10), one arrives at
P

P =
320
3

2
m
9c
3

BR, (11)
C _
2004 RAS, MNRAS 356, 359370
1E 1207.45209: a strange star? 363
for models 1 and 2, and
P

P =
20
3

2
v
9c
3
R

B
, (12)
for models 3 and 4 in the low-mass limit.
Lines of constant potential drops in the P

P diagram. The po-
tential drop in the open-eld-line region is essential for pulsar mag-
netospheric activity. We adopt only model 1 in the following. From
equations (3), (6) and (10), one has
B =
32
3

B
m
. (13)
This shows that the polar elds of homogeneously magnetized quark
stars, with a certain
m
, of different low masses are approximately
the same. The potential drop between the centre and the edge of a
polar cap is (Ruderman & Sutherland 1975)
=
2
2
c
2
R
3
BP
2
. (14)
In the case of approximately constant
m
, equation (14) can be
conveniently expressed as, from equation (13),
=
64
3
3c
2

B
m
R
3
P
2
. (15)
From equations (11) and (15), one has
P

P
3
=
2.026 10
6
c
7

B
2

5
m
. (16)
However, if the variance of pulsar masses (or radii) is smaller than
that of
m
, it is better to express the potential drop as (fromequations
5 and 14),
= 2

16
5c

BR
5
PP
3
, (17)
where equation (10) is included. The lines of constant are drawn
in Fig. 3, based on both equation (16) (dashed lines labelled a, b
and c, with a slope of
1
3
) and equation (17) (dash-dotted lines la-
belled d and e, with a slope of 3). The parameters for these lines
are as follows: (a) = 10
12
V,
m
= 10
6
G cm
3
g
1
; (b)
= 10
10
V,
m
= 10
6
G cm
3
g
1
; (c) = 10
12
V,
m
= 10
7
G cm
3
g
1
; (d) = 10
12
V, R = 10 km and (e) = 10
11
V,
R = 1 km.
Pair production mechanisms are essential for pulsar radio emis-
sion. A pulsar is said to be dead if the pair production condition
cannot be satised. A general review of the understanding of radio
pulsar deathlines can be found in Zhang (2003). Although a real
deathline depends upon the dynamics of the detailed pair and pho-
ton production, the deathline can also be conventionally taken as a
line of constant potential drop . It is found in Fig. 3 that the slope of
constant is
1
3
(or 3) if the scattering distribution of pulsar points
in the P

P diagram is due to different masses (or polar eld B) but
with constant
m
(or mass or radius). The deathline slope may be
expected to be between
1
3
and 3 if the distributions of mass and
polar eld are combined.
3.2 Accretion-dominated spin-down
The physical process of accretion on to rotating pulsar-like stars with
strong magnetic elds is very complex but is essential to ascertain
the astrophysical appearance of the stars (e.g. the variation of X-ray
ux, the evolutionary tracks, etc.), which is still not well enough
understood (Lipunov 1992). Nevertheless, it is possible and useful
to describe the accretion semiquantitatively.
For an accretion scenario in which the effect of kinematic energy
of accreted matter at innite distance is negligible (such as the case
of supernova fall-back accretion), three typical radii are involved.
The radius of a light cylinder of a spinning star with period P is
r
l
=
cP
2
= 4.8 10
9
P cm. (18)
If all of the accretion material is beyond the cylinder, the star and the
accretion matter could evolve independently. The magnetospheric
radius, dened by equating the kinematic energy density of free-fall
particles with the magnetic one B
2
/(8), is
r
m
=

B
2
R
6

2GM

2/7
, (19)
where

M is the accretion rate. In the low-mass limit of bare strange
stars, considering the massradius relation of equations (10) and
(19) becomes
r
m
=

3
32G

1/7

B
1/7
B
4/7
R
9/7

M
2/7
= 0.064

B
1/7
60
B
4/7
R
9/7

M
2/7
, (20)
where the bag constant

B =

B
60
60 MeV fm
3
. If a star is ho-
mogeneously magnetized per unit mass (i.e. in models 1 and 2),
according to equation (13), one has from equation (20)
r
m
=

32
3

3
27G

1/7

B
3/7

4/7
m
R
9/7

M
2/7
= 4.9 10
7

B
3/7
60

4/7
m
R
9/7

M
2/7
. (21)
As a result of the strong magnetic elds around a spinning star,
matter is forced to corotate, and both gravitational and centrifugal
forces work. At the so-called corotating radius r
c
, these two forces
are balanced,
r
c
=

GM
4
2

1/3
P
2/3
= 1.2 10
3
M
1/3
P
2/3
. (22)
In the low-mass limit, one has from equations (10) and (22)
r
c
=

4G
3

1/3

B
1/3
RP
2/3
= 145

B
1/3
60
RP
2/3
. (23)
In another case, in which the kinematic energy at innity is not
zero [i.e. interstellar medium (ISM) or stellar wind accretion], be-
sides those three radii, an additional one is the accretion radius r
a
,
at which the total energy (kinematic and gravitational ones) is zero,
r
a
= 2GMV
2

, (24)
where V

is the relative velocity of the star with respect to the

surrounding media. The motion of matter only at a radius <r
a
could
be affected by gravity, and the mass capture rate is then

M
c
= r
2
a
V

= 4G
2
M
2
V
3

, (25)
where is the density of the diffusion material.
As a result of centrifugal inhibition, as the radius of matter nearest
to the star could be r
m
, massive accretion on to the stellar surface is
impossible when r
m
>r
c
. This is the so-called supersonic propeller
spin-down phase. A star spins down to the equilibrium period P
eq
,
C _
2004 RAS, MNRAS 356, 359370
364 R. X. Xu
dened by r
m
= r
c
. In the low-mass limit, one has, from equa-
tions (20) and (23),
P
eq
= 0.72G
5/7

B
5/7
B
6/7
R
3/7

M
3/7
, (26)
or assuming a homogenous magnetic momentum per unit mass,
from equations (21) and (23),
P
eq
= 15G
5/7

B
1/7

6/7
m
R
3/7

M
3/7
. (27)
However, accretion with rate

M on to the stellar surface is not possi-
ble, although the centrifugal barrier is not effective when P > P
eq
,
until the star spins down to a so-called break period (Davies, Fabian
& Pringle 1979; Davies & Pringle 1981; Ikhsanov 2003),
P
br
= 60
16/21
30

M
5/7
15
M
4/21
1
s
= 36

B
4/21
60
B
16/21
12

M
5/7
15
R
12/7
6
s
= 0.49

B
12/21
60

16/21
m6

M
5/7
15
R
12/7
6
s, (28)
in the low-mass limit, where equations (3), (10) and (13) have been
included, and the convention Q =10
n
Q
n
has been adopted. Pulsars
with periods between P
eq
and P
br
do still spin-down. This phase
is called subsonic propeller. Only a negligible amount of accre-
tion matter can penetrate into the magnetosphere (on to the stellar
surface) during both the supersonic and subsonic propeller phases
(Ikhsanov 2003), and the expected accretion luminosity is thus very
low.
How can one determine quantitatively the spin-down rate when a
pulsar is in those two propeller phases? No certain answer is known
for the propeller torques, nor even for the accretion conguration
(disc or sphere). None the less, if the spin-upeffect of matter accreted
on to the stellar surface is neglected, the spin-down rate can be
estimated according to the conservation laws of angular momentum
and/or rotational energy (Davies et al. 1979). The escape velocity at
radius r is

2GM/r. Approximating the stellar angular-momentum

loss rate 2I

P/P
2
to that of accretion material near r
m
(based on
angular-momentum conservation), we have

P
J
=

2
M
1/2
I
1

Mr
1/2
m
P
2
=
5

6G
8
3/2

B
1/2
R
7/2

Mr
1/2
m
P
2
, (29)
where equation (10) is introduced. However, in the case of energy
conservation, I

= GM

M/r
m
, one arrives at

P
E
=
G
4
2
MI
1

Mr
1
m
P
3
=
5G
8
2
R
2

Mr
1
m
P
3
.
(30)
4 THE CASE OF 1E 1207. 45209
The radio-quiet central compact object in the supernova remnant
PKS 1209-51/52, 1E 1207.45209, is a unique pulsar-like star
which is worth noting as we have a signicant amount of infor-
mation concerning it: the rotating period P = 0.424 s, the cy-
clotron energy E
cyc
= 0.7 keV (Bignami et al. 2003), the age
T 7 kys estimated from the remnant, with an uncertainty of
a factor of 3 (Roger et al. 2003), the timing properties (Zavlin,
Pavlov & Sanwal 2004) and the thermal X-ray spectrum of long-
time observations (De et al. 2004). The distance to the remnant
is d = 1.33.9 kpc, the X-ray ux in a range of 0.48 keV is
2.3 10
12
erg cm
2
s
1
and the corresponding X-ray luminos-
ity is then L = (0.474.2) 10
33
erg s
1
(Pavlov, Sanwal &
Teter 2003). However, the more we observe, the knottier the prob-
lem astrophysicists have in creating a model to understand its
nature.
Two issues are addressed rst. One concerns its absorption lines.
The lines at 0.7, 1.4 and 2.1 keV (and possibly 2.8 keV) are iden-
tied, which are phase-dependent (Mereghetti et al. 2002). These
imply a cyclotron origin for the features (Bignami et al. 2003; Xu
et al. 2003), although this possibility was considered to be unlikely
when discovered by Chandra (Sanwal et al. 2002). However, there
are still two questions relevant to this issue.
(1) Where does the absorption form (near the stellar surface or in
the magnetosphere)? An e

plasma surrounding a magnetized neu-

tron star, maintained by the cyclotron-resonance process, was sug-
gested to prevent a direct detection of the stellar surface in the X-ray
band, the existence of which seems to explain the age dependence of
the effective radiating area (Ruderman 2003). The cyclotron lines
would thus form at a height where resonant scattering occurs. In
this regime, all neutron stars with high B-elds should present cy-
clotron absorption in their thermal X-ray spectra, but this conicts
with the observations. In addition, the physics of the plasma is still
not well studied, and its density and stability are not certain. An
alternative and intuitive suggestion is that the line-forming region is
near the stellar surface. In this case, we may need a bare quark sur-
face, with an electron layer of density 10
32
cm
3
and a thickness
of a few thousand fermi, in order to explain these absorption dips.
As a result of the degeneracy of electrons, only electrons near the
surface of the fermi-sea can be excited to higher levels, the number
of which is energy-dependent. For instance, the number of electrons
that resonantly scatter photons with energy 1.4 keVcould be about
double those with 0.7 keV. The number of electrons that are re-
sponsible for cyclotron-resonance of photons with higher energy,
is therefore larger, although the absorption cross-section is smaller.
Another factor, which may also favour more electrons absorbing
higher-energy photons, could be the radiative transfer process (e.g.
a certain layer might be optically thick at 0.7 keV, but optically
thin at 2.1 keV), but a detailed consideration of this is necessary
in the future. We conclude that it is safe to assume a surface origin
for the cyclotron-resonant lines.
(2) Is the cyclotron resonance in terms of electrons or pro-
tons? The fundamental electron cyclotron resonant lies at E
e
=
11.6B
12

1 r
s
/R keV, while the proton resonance lies at E
p
=
6.3B
12

1 r
s
/R eV, where r
s
2GM/c
2
is the Schwarzschild ra-
dius. For a star with10
6
cmand1 M

, the factor

1 r
s
/R = 0.84.
In the case of lower-mass strange stars, r
s
/R R
2
is smaller, and
the factor is closer to 1. We thus just approximate the factor to be
1, the eld is then B = 6 10
10
G in terms of electrons and B =
10
14
G for protons. If the lines are of a proton origin, there are still
two scenarios. One is that the multipole elds have a strength of
B
m
10
14
G, but the global dipole eld B
p
3 10
12
G is much
smaller in order to reconcile the spin-down rate expected fromequa-
tion (2). This means that the stellar surface is full of ux loops with
a typical length l
loop
, which can be estimated to be (Thompson &
Duncan 1993)
l
loop

B
p
B
m
R 10
5
R
6
cm. (31)
The maximumrelease of energy due to magnetic reconnection could
be (B
2
m
/8)l
3
loop
10
42
erg. If the dynamical instability takes
place on a short time-scale of 1 s, as observed in soft -ray re-
peaters, bursts with 10
42
erg s
1
might have been detected in 1E
1207.45209; but we do not. Another scenario is that the dipole
C _
2004 RAS, MNRAS 356, 359370
1E 1207.45209: a strange star? 365
eld of 1E 1207.45209 is 10
14
G, but the accreted material on
to the stellar surface contributes a positive angular momentum. The
magnetodipole radiation should spin-down the object at a rate of

P = 2 10
10
s s
1
, based on equation (4), which is much larger
than is observed (10
14
s s
1
). This discrepancy might be circum-
vented if accreted matter contributes a positive momentum. Yet, this
can only be possible when P > P
br
, which results in an accretion
rate, according to equation (28), of

M > 0.9 10
15
g s
1
and an
X-ray luminosity of L
x
> 10
35
erg s
1
, which is much larger than
observed (L 10
33
erg s
1
) for typical neutron star parameters. In
bothpictures, however, the strange thingis whydoes this star that has
a magnetar eld not show magnetar activity (e.g. the much higher
luminosity 10
3435
erg s
1
of persistent X-ray emission, observed
in anomalous X-ray pulsars and soft -ray repeaters)? In addition,
there is still no answer to why the feature strength is similar at 0.7
and 1.4 keV (and even the appearance of a line at 2.1 keV), due
to the high massenergy of protons (10
3
times that of electrons),
as discussed in the case of SGR 1806-20 (Xu et al. 2003). We there-
fore tend to favour an electron-cyclotron origin for the absorption
features.
Another issue is that concerning its detected radius. In principle,
one can obtain the radius of a distant object by detecting its spectrum
(tting the spectrum gives a temperature T if a Planckian spectrum
approximation is good enough) and ux F, through F = T
4
R
2
/d
2
(where R is the radiation radius, not the coordinate radius R
coord
in
the Schwarzschild metric. R = R
coord
/

1 r
s
/R
coord
if the spec-
trum is Planckian), if the distance d is measured by other methods
(e.g. parallax). However, we are not sure whether the thermal spec-
trum of 1E 1207.45209 is really Planckian or not, and we do not
knowthe distance either. None the less, as the spectrumalso depends
on the ISM absorption (the neutral hydrogen density is assumed to
be known), one may t the spectrum using the free parameters T,
d and R. Note that radii determined in this way are highly uncer-
tain. A radius of 1 km was obtained in single-blackbody mod-
els by ROSAT (Mereghetti, Bignami & Caraveo 1996) and ASCA
(Vasisht et al. 1997) observations, whereas a 10-km radius was sug-
gested in an atmosphere model of light elements by Zavlin, Pavlov
& Tr umper (1998). An XMMNewton observation recently yielded
a two-blackbody radii t: R = 0.8 and 4.6 km for hotter and cooler
components, respectively (De et al. 2004). The possibility that 1E
1207.45209 may have a radius much smaller than the 10 kmvalue
of conventional neutron stars is thus not ruled out.
Now we turn to a low-mass bare strange star model for 1E
1207.45209. Besides the absorption features, the most outstand-
ing nature, which makes the pulsar puzzling enough, should be its
timing behaviour: it does not spin-down stably but seems to spin-up
occasionally. Furthermore, two or more probability peak frequen-
cies are identied during each of the ve observations (Zavlin et al.
2004). Pulsar glitches and Doppler shift in a binary system were
proposed for the spin-up (De Luca et al. 2004; Zavlin et al. 2004),
but the multifrequency distributions are still not well understood.
An alternative model proposed here is that the pulsar is a low-mass
bare strange star which is at a critical point of its subsonic propeller
phase, P P
br
.
Steady accretion on to a magnetized and spinning star is possible
onlyif P > P
br
, whenmagnetohydrodynamic (e.g. RayleighTaylor
and Kelvin-Helmholtz) instabilities occur at the magnetospheric
boundary (Arons & Lea 1976; Elsner & Lamb 1984). Before the
onset of the instabilities, the accretion plasma can only penetrate
into the magnetosphere by diffusion, with a rate much smaller than

M (Ikhsanov 2003). Propeller torque may spin-down a star to a pe-

riod of P P
br
. At this period, steady accretion does not occur until
the density just outside the boundary increases to a critical density
of
crit
7
m
when the RayleighTaylor instability occurs (Elsner
&Lamb 1984), where
m


M/(4r
2
m

2GM/r
m
). The star should
then be spun up by a steady accretion torque.
5
However, increas-
ing the spin-frequency may dissallow the necessary condition for
steady accretion, P > P
br
. The star will then return back to a sub-
sonic propeller phase when the decreasing period is low enough,
and it spins down again. We could therefore expect an erratic timing
behaviour when a pulsar is at this critical phase, P P
br
. The object
1E 1207.45209 could be an ideal laboratory for us to study the
detailed physics of such an accretion stage.
For 1E 1207.45209, setting P
br
=0.424 s, B
12
=0.06, one has

m
= 9.110
6

B
1
110
G cm
3
g
1
from equation (13), which is close
to the values for millisecond pulsars (Table 1). The accretion rate
reads, from equation (28), as

M = 10
14

B
4/15
60
R
12/5
5
g s
1
, (32)
and the magnetospheric radius is then, from equation (20),
r
m
= 2.5 10
7

B
1/15
60
R
3/5
5
cm. (33)
As for the instantaneous spin-down rate in the subsonic propeller
phase, we apply equations (29) and (30) for an estimation, according
to the momentum and energy conservation equations, respectively.
One then has,

P
J
= 1.9 10
12

B
4/5
60
R
4/5
5
,

P
E
= 1.3 10
13

B
1/5
60
R
1/5
5
.
(34)
These results imply that the instantaneous period increase could be
one or two orders larger than the averaged one (10
14
s s
1
) if R
5
1. A precise measurement of instantaneous

P
inst
may tell us the
actual radius. In fact, a possible spin-up has been noted in the XMM
Newton observations of August 2002, with an instantaneous period
decrease rate of

P
inst
= (36) 10
14
, whereas a spin-down was
noted in Chandra observations of 2003 June, with

P
inst
210
13
much larger than the averaged one (Zavlin et al. 2004). The conjec-
tured radius is then likely to be 1 km, based on this instantaneous
rate and the spin-down rule of equation (30), and 1E 1207.45209
could be a low-mass bare strange star (M 10
3
M

). Such a high
accretion rate (10
14
g s
1
) could certainly not be from capture of
interstellar mediummatter by the star, but could be due to a fallback
ow, since the age is relatively young (7 kys). The age discrepancy
between the supernova remnant (SNR) age and that
6
of P/(2

P) is
not surprising as the pulsar is not dominantly rotation powered. It
depends on the detailed physics of magnetospheric boundary inter-
actions to solve the age problem for this pulsar.
The accretion X-ray luminosity, for steady accretion with a rate of

M, could be GM

M/R 10
32
erg s
1
, which is comparable with
the observed L 10
33
erg s
1
. This hints that the X-ray radiation of
1E 1207.45209 could be powered by both cooling and accretion.
Note that, from equation (2), the energy-loss rate due to magne-
todipole radiation from such a low-mass star is only 10
24
erg s
1
,
which is much smaller than the X-ray luminosity L. Therefore, such
5
In the case of wind-fed accretion in binary systems, the star will spin-up
(or down) if the accreted material has a positive (negative) angular momen-
tum. Whereas, in the case of fallback disc accretion or ISM-fed accretion, the
momentum of accreted matter should be positive, and the steady-accretion-
induced torque leads the star to spin-up. We are not considering accretion in
a binary system here.
6
See equation (8), for the case of P
0
_0.424 s.
C _
2004 RAS, MNRAS 356, 359370
366 R. X. Xu
pulsar-like stars have negligible magnetospheric activities, and can
be observed only if they are near and young enough.
The models proposed, including this strange-star model and that
of Zavlin et al. (2004), can be tested by future timing observations.
The Doppler shift model should be ruled out if future pulses do not
arrive at the times expected by the model. If the random spin nature
can be conrmed as a general nature in more precise observations,
we may tend to suggest P P
br
for 1E 1207.45209, as glitching
pulsars usually spin stably except during glitches.
5 OTHER CANDI DATES FOR LOW- MASS
STRANGE STARS
A rst candidate pulsar for low-mass strange stars was suggested
in 2001 (Xu et al. 2001): the fastest rotating millisecond pulsar
PSR 193721 (P = 1.558 10
3
s,

P = 1.051 10
19
s s
1
). In
order to explain the polarization behaviour of its radio pulses and the
integrated prole (pulse widths of the main pulse and the interpulse,
and the separation between them), this pulsar is assumed to have a
mass <0.2 M

and a radius <1 km. The polar magnetic eld is

8.2 10
8
G based on equation (4) for conventional neutron stars,
but could be 2.2 10
9

B
1/2
60
G if a radius of R = 1 km and equa-
tions (5) and (10) are applied.
The lowmass may actually favour a high spin frequency during
the birth of a bare strange star. The dened Kepler frequency of such
stars could be approximately constant,

k
=

GM
R
3
= 1.1 10
4

B
1/2
60
s
1
, (35)
with a prefactor of 0.65 at most for M M

and R 10
6
cm
(Glendenning 2000). The initial rotation periods of strange stars
are limited by the gravitational radiation due to r-mode instability
(Madsen 1998). At this early stage, the stars are very hot, with
temperatures of a few 10 MeV, and we may expect a uid state for
the quark matter, without colour superconductivity. The critical
satises the equation
1

gw

sv

bv
= 0, (36)
where the growth time-scale for the instability (the negative sign
indicates that the model is unstable) is estimated to be,

gw
= 3.85 10
81

6
M
1
R
4
, (37)
and
sv
and
bv
are the dissipation time-scales the due to shear and
bulk viscosities, respectively,

sv
= 1.85 10
9

5/3
s
M
5/9
R
11/3
T
5/3
,

bv
= 5.75 10
2
m
4
100

2
R
2
T
2
,
(38)
and
s
is the coupling constant for the strong interaction, T is
the temperature and m
100
is the strange quark mass in units of
100 MeV.
The calculated results, based on equation (36), are shown in
Fig. 4. It is found that low-mass bare strange stars can rotate very
quickly, even faster than the Kepler frequency
7
and one would then
not be surprised that the fastest rotating pulsar could be a low-mass
bare strange star. However, though it needs advanced techniques for
data collection and analysis to detect a submillisecond radio pulsar,
we have not found one yet (Edwards, van Straten & Bailes 2001;
7
Note that the surface matter is not broken at the super-Kepler frequency,
due to the self-binding of quark matter.
10
5
10
6
10
7
10
8
10
9
10
10
10
11
10
3
10
4
10
5
10
Temperature T
A
n
g
u
l
a
r

f
r
e
q
u
e
n
c
y

R = 10km, m100 = 1.5, B = 60

R = 10km, m100 = 3, B = 60
R = 1km, m100 = 1.5, B = 60
R = 1km, m100 = 1.5, B = 110

k
=1100/s
R
=
5
k
m
, m
1
0
0
=
3
, B
=
6
0
6
Figure 4. Temperature dependence of the maximum angular frequency in
rotating bare strange stars, due to the gravitational instability in the r-modes.
The coupling constant
s
=0.1. Other parameters are shown for each types
of lines. The bag constant

B is in units of MeV fm
3
.
Han et al. 2004). This negative result could be due to: (1) the dy-
namical process not resulting in a submillisecond rotator; (2) no
magnetospheric activity existing for very low-mass strange stars
where

P is very small, as the potential drop is not high enough to
trigger pair production (see Section 3.1). In the latter case, a nearby
submillisecond pulsar could be found by X-ray observations as a
spinning hotspot, powered by rotation and/or accretion on the stellar
surface.
RX J1856.5-3754 is another candidate. RX J1856.5-3754 could
be a low-mass quark star, based on the X-ray spectrum, but the main
puzzle is the origin of its optical radiation, the intensity of which is
approximately seven times that extrapolated from the Rayleigh
Jeans law for the X-ray spectrum (Burwitz et al. 2003). If RX
J1856.5-3754is a spinningmagnetizedstar, its magnetosphere could
be surrounded by a spherically quasi-static atmosphere, in which
the plasma temperature is of the order of the free-fall temperature
(Lipunov 1992; Ikhsanov 2003),
T
ff
=
GMm
p
kr
m
, (39)
where m
p
is the proton mass and k is Boltzmanns constant. The
dissipation of stellar rotation energy, as well as the gravitational
energy of accreted matter, may heat the envelope, which could be
responsible for the UVoptical emission. The soft-component-tted
parameters could thus be for this quasistatic envelope, with temper-
ature <33 eV and radius >17 km. Assuming T
ff
<33 eV and r
m
>
17 km, one has a lowlimit
8
for the stellar mass M >4 10
7
M

or
radius R > 0.1 km. While the hard-component-tted stellar radius
is R =4.4 km (Burwitz et al. 2003). One can also infer an accretion
rate of

M 4 10
10
g s
1
, from equation (21) for
m
= 10
6
G
cm
3
g
1
. The X-ray luminosity due to accretion is then > 3 10
26
erg s
1
. This model for the soft component could be tested by more
observations in submillimetre bands, and in optical and UV bands,
as the quasi-static atmosphere could also be effective in radiating
infrared photons if it is dusty.
8
For blackbody radiation, r
m
T
2
is a constant. According to equation (39),
one nds M r
m
T (r
m
T
2
)T
1
.
C _
2004 RAS, MNRAS 356, 359370
1E 1207.45209: a strange star? 367
Strange quark matter with mass _ M

could be ejected by a
massive strange star (M

) during its birth or by the collision of

two strange stars, and such low-mass matter may explain a few as-
trophysical phenomena (Xu &Wu 2003): the planets around pulsars
could be quark matter with mass 10
2328
g, while very low-mass
strange quark matter (called strangelets) with baryon numbers of
10
9
may be ultrahigh-energy cosmic rays beyond the Greisen
ZatsepinKuzmin (GZK) cut-off. The bursts of soft -ray repeaters
could be due to either starquake-induced magnetic reconnection or
a collision between a strange planet and a solar-mass bare strange
star. The chance of a collision would not be low if both objects
formed during the same supernova, or in the same binary system.
Some of the transient unidentied EGRET sources (Wallace et al.
2000) may represent such collision events, the gravitational energy
release of which would be
E
g

GM
a
M
b

R
3
a
R
3
b

1/3
= 2 10
23

B
2
60
R
3
a
R
3
b

R
3
a
R
3
b

1/3
erg, (40)
where a and b denotes two objects composed of strange quark
matter. The released energy could be 10
45
erg if R
a
10
5
cm and
R
b
10
4
cm. This strong (colour) interaction may result in photon
emission by various hadron process (e.g. hadron annihilation), with
energy 100 MeV (the EGRET telescope covers an energy range
from30 MeVto over 20 GeV), and could be another way to produce
strangelets.
Mergingquarkstars, rather thanneutronstars (Eichler et al. 1989),
may result in cosmic -ray bursts (GRBs), which could help to
eliminate the baryon load problem. The released energy is 10
53
erg
during the collision of two quark stars with 10
6
cm. The residual
body would be expected to rotate rapidly, and such a high spin
frequency may result in a beaming pattern of emission. A reball
with low baryon contamination in this colour interaction event may
favour the emission of photons and neutrinos with high energy.
Alternatively, rapidly rotating quark stars as a residue of hypernovae
could also be possible as the central engines. Only non-baryonic
particles (e.g. photons, neutrinos and e

) can be radiated massively

from the quark surfaces, as the stellar temperature is initially very
high (>10 MeV). A reball then forms, and the Usov mechanism
(Usov 1998) of pair production in a strong electric eld near a quark
surface may play an important role at that time. The rotation may
cause very irregular light curves.
6 THE ORI GI N OF LOW- MASS
STRANGE STARS
This is a real problem which is difcult to answer with certainty at
present. Several relevant scenarios and arguments are suggested in
the following section, though, in this paper, we focus on proposing
low-mass strange star candidates and trying to attract the readers
attention to such stars neglected previously.
6.1 The origin of millisecond pulsars
An open debate on this issue took place in 1996 (Bhattacharya et al.
1996). Millisecond pulsars are recycled pulsars in low-mass X-ray
binaries, the magnetic elds of which decay (by, for example, en-
hanced Ohmic dissipation, diamagnetic screening effect, etc.) dur-
ing the accretion process, according to the standard model. After
years of searching for coherent millisecond X-ray pulsations, ve
accretion-driven millisecond pulsars have been discovered since
1998 (Wijnands 2004), which are important for testing the model.
Besides the old problems (e.g. birth rate, millisecond pulsars with
planets, etc.), new puzzling issues are raised in the standard argu-
ments (Rappaport, Fregeau & Spruit 2004).
However, the puzzles may disappear if rapidly spinning low-mass
bare strange stars result directly from accretion-induced collapse
(AIC) of white dwarfs, but could be covered by normal matter
if they had high-accretion phases in their latter evolution (while
core collapses may produce only normal pulsars with mass M

and radius 10
6
cm). Normal neutron stars created by AIC have
been investigated with great effort (Fryer et al. 1999). A detonation
wave, separatingnuclear matter andquarkmatter, shouldforminside
white dwarfs, and propagate outwards, if quark stars are born via
AIC. Though the possibility of forming a massive strange star with
M

by AIC might not be ruled out, the reason that AIC produces
low-mass strange star might be simple: the density and temperature
in an accreting white dwarf may not be so high that the detonation
ame reaches near the stellar surface. A low-mass quark star could
form if the detonation surface were far below the stellar surface.
The newborn low-mass bare strange stars could rotate very
rapidly, and may even be super-Keplerian (Fig. 4), and would spin-
down if the accretion rates were not very high. Unless the mass were
smaller than (0.10.3) M

, the thickness of crust could be negligi-

ble
9
(Xu 2003a), and the maximum X-ray luminosity at the break
period (

M = 2.3 10
16
P
7/5
br3

B
12/15
60

16/15
m6
R
12/5
5
g s
3
from equa-
tion 28) would be approximately,
L
x
=
GM

M
R
2.8 10
34
P
7/5
br3

B
9/5
60

16/15
m6
R
22/5
5
erg s
1
. (41)
A strange star with low mass may explain the low time-averaged
accretion luminosity in bursting millisecond X-ray pulsars as L
x

R
4.4
, if it is assumed that bursting X-ray pulsars are in a critical stage
of P P
br
. It is worth noting that, at this stage, the real accretion
luminosity could be much lower than L
x
presented in equation (41)
because only part of the accretion matter with rate

M could bombard
the stellar surfaces directly if P < P
br
.
It is generally suggested that, during an iron-core-collapse super-
nova, the gravitationally released energy (E
g
10
53
erg) is almost
entirely in the form of neutrinos, 10
2
of which is transformed
into the kinetic energy of the outgoing shock and 10
4
of which
contributes to the photon radiation. However, this idea has not been
as successful in modern supernova simulations (Buras et al. 2003;
Liebendoerfer 2004), as the neutrino luminosity could not be large
enough for a successful explosion even in the models with the in-
clusion of convection below the neutrinosphere (two-dimensional
calculations). We note that the bare quark surfaces may be essential
for successful explosions of both types of iron-core collapse and
AIC. The reason is that, because of the colour binding, the photon
luminosity of a quark surface is not limited by the Eddington limit. It
is possible that a prompt reverse shock could be revived by photons,
rather than neutrinos. A hot quark surface, with temperature
10
T >
10
11
K, of a newborn strange star
11
will radiate photons at a rate of

E
p
> 4R
2
T
4
7 10
50
R
2
5
T
4
11
erg s
1
, (42)
9
The crust thickness could be much smaller than given by previous calcu-
lations for the static and the cold case due to a high penetration rate during
hot bursts.
10
This is estimated using just gravitational energy release. The temperature
of low-mass bare strange stars could also be >10
11
K, as each baryon would
release 10100 MeV during the quark phase transition.
11
The thermal conductivityof quarkmatter is muchlarger thanthat of normal
matter in protoneutron star crusts. The surface temperature of protoneutron
stars cannot be so high, otherwise a signicant amount of stellar matter
would be expelled as wind (the neutron star mass might then be very small).
C _
2004 RAS, MNRAS 356, 359370
368 R. X. Xu
while the Thomson-scattering-inducedEddingtonluminosityis only
L
Edd
=
64
2
cGm
p
3
T

BR
3
10
35

B
60
R
3
5
erg s
1
. (43)
This means that the photon emissivity may play an important role in
both types of supernova explosions (i.e. for the birth of solar-mass
as well as low-mass bare strange stars).
Strange stars born in this way are certainly bare, as any normal
matter cannot survive fromthe strongphotonbursts. Ahighfall-back
accretion may not be possible due to massive ejecta, rapid rotation
and strong magnetic elds, and such stars could stay bare as long as
the accretionrates are not veryhigh, as accretedmatter withlowrates
could penetrate the Coulomb barrier (Xu 2002). As a white dwarf
collapses to a state with nuclear or supranuclear densities, strange
quark matter seeds may help to trigger the transition from normal
matter to quark matter. However, the seeds may not be necessary,
as the transition could occur automatically at that high density.
In the core-collapse case, the total photon energy could be much
larger than the energy (10
2
E
g
10
51
erg) with which the outer
envelope should be expelled, as the time-scale for a protostrange
star with T 10
11
is usually more than 1 s. In the AIC case,
the binding energy of a progenitor white dwarf with mass M

and radius 10
9
cm is E
bin
3 10
50
erg, and a minimum mass
3 10
3
M

of bare strange stars would release an amount of

every E
bin
if each baryon contributes approximately 50 MeV af-
ter conversion from hadron matter to strange quark matter. In this
sense, such explosions to produce low-mass strange stars are pow-
ered by the phase-transition energy, rather than the gravitational
energy. Certainly, the lower limit can be smaller if the progenitor
white dwarf is less massive. It is then not surprising that AIC may
produce low-mass strange stars as long as a strange quark phase
conversion
12
can occur in the centre of a white dwarf with much
higher temperature and density. The mass of the residual bare quark
stars may depend on the details of combustion of nuclear matter
into strange quark matter, especially on the last detonation surface
where the phase transition can no longer occur. Certainly, this sur-
face would be determined by various aspects of the microphysics
(e.g. how much energy per baryon is released during the phase con-
version). In the case where the kick energy is approximately the
same, only solar-mass millisecond pulsars can survive in binaries
as low-mass pulsars may be ejected by the kick.
Recently, it was suggested by Popov (2004) that low-mass com-
pact objects can only form by fragmentation of rapidly rotating
protoneutron stars, and that such objects should have large kick ve-
locities. It is worth noting that AIC-produced low-mass compact
stars might not have large kick velocities, which may serve as a pos-
sible test of the models of formation mechanisms. On one hand, if
the kick energy E
kick
MV
2
(V is the kick velocity) is the dominant
part of the gravitational energy GM
2
/R, then one nds V

M.
On the other hand, if E
kick
is mainly part of the phase-transition
energy M, then low-mass stars could have a similar V to that
of strange stars with M

. Fragmentation might hardly occur for

quark stars due to the colour connement.
Additionally, AIC-created pulsar-like stars may help to explain
various astrophysical phenomena, the recent work including the
kick velocities of millisecond pulsars (Tauris & Bailes 1996), the
r-process nucleosynthesis of heavy (baryon number A>130) nuclei
(Fryer et al. 1999; Qian & Wasserburg 2003), the ultrahigh-energy
12
The critical conditionmaydependonvarious parameters, e.g. the chemical
composition, the accretion rate, the thermal history, the stellar rotation, etc.
protons accelerated in the pulsar magnetospheres (de Gouveia Dal
Pino & Lazarian 2001), as well as the numerical calculations of
millisecond pulsar formation in binary systems (Tauris et al. 2000).
None the less, besides the high-mass stars, it is interesting to deter-
mine whether an LI-star (a low- and intermediate-mass star, with
mass 2 M/M

8) can die via a violent event (e.g. a super-

nova) and possibly produce a low-mass strange star after an unstable
nuclear explosion, by detailed calculations and/or observations.
6.2 The origin of pulsar magnetic elds
The different values of
m
(or
v
) of normal pulsars and millisecond
pulsars (Table 1) may result fromtheir dissimilar physical processes
at birth (iron-core collapse or AIC). The magnetic momentum per
unit baryon,
b
, could be dependent on the number, n, of quarks
in each cluster. It is suggested that
b
of pulsars created by core
collapse is bigger than that by AIC according to Table 1. A nucleon
has a magnetic momentum of about the nuclear magneton,
N
=
5 10
24
erg G
1
, and the corresponding value
m

N
/m
p

3 G cm
3
g
1
, which is much larger than that observed in pulsars.
This hints that the quark clusters in solid quark stars may have a
magnetic momentum per baryon 10
(46)
orders weaker than that
of nucleons.
7 CONCLUSI ONS AND DI SCUSSI ONS
General properties of both rotation- and accretion-powered low-
mass bare strange stars have been presented. It has been sug-
gested that normal pulsars with M

masses are produced af-

ter core-collapse supernova explosions, whereas millisecond pul-
sars with (0.11) M

(and even lower) masses could be the

remains of accretion-induced collapses of massive white dwarfs.
These different channels for pulsar formation may result in two
types of ferromagnetic elds: weaker for AIC (
m
10
6
G cm
3
g
1
) and stronger for core collapse (
m
10
4
G cm
3
g
1
). We note that the low-mass quark stars involved in this pa-
per also have small radii, and hence may be distinguished from
low-mass quark stars with large radii (Alaverdyan, Harutyunyan &
Vartanyan 2001).
Some potential astrophysical appearances relevant to low-mass
bare strange stars are also addressed. We suggest that the radio-
quiet central compact object, 1E 1207.45209, is a low-mass bare
strange star with a polar surface magnetic eld 6 10
10
G and
likely to be a few kilometres in radius, and it is now at a critical
point in the subsonic propeller phase, P P
br
, in order to un-
derstand its timing behaviour. A newborn low-mass strange star
could rotate very rapidly, even with a super-Kepler frequency. The
radius of the dim thermal object, RX J1856.5-3754, is R > 0.1
km if its soft UVoptical component radiates from a spherically
quasi-static atmosphere around it. It is proposed that some of the
transient unidentied EGRET sources may result from the col-
lisions of two low-mass strange stars. It is worth noting, in our
sense, that the so-called massive compact halo objects, discovered
through gravitational microlensing (Alcock et al. 1993), could also
be low-mass quarkstars formedfromevolvedstars, rather thanquark
nuggets born during the QCD phase transition of the early Universe
(Banerjee et al. 2003).
The mass of strange quark matter could be as low as a few hun-
dred baryons (strangelets). Strangelets can be evaporated through
the bare surfaces of new-born quark stars, or produced during the
collision of two (low-mass) quark stars. Strangelets with 10
89
baryons could be detected as ultrahigh-energy cosmic rays (Xu &
C _
2004 RAS, MNRAS 356, 359370
1E 1207.45209: a strange star? 369
Wu 2003). In this sense the discrepancy between the observational
uxes (Anchordoqui et al. 2004) of AGASAand of HiRes/Flys Eye
might be explained, as solid strangelets at initially low temperature
should be heated enough to ionize the atmosphere and would thus
result in low uorescence radiation in the latter detector.
Can we conrm the small radius of a low-mass bare strange star
by a direct observation of future advanced space telescopes? This
work might be done by the next generation Constellation X-ray
telescope (to be launched in 20092010), which covers an energy
band of 0.25100 keV. The radii, R, of neutron stars are generally
greater than 10 km (R of 0.1-M

mass neutron stars is 160 km).

If pulsars are neutron stars, their surfaces should be imaged by
the Constellation-X with much higher space resolution, as long as
the separation between the four satellites is greater than d/R
3
8
d
100 pc
/R
6
km. (Note that the wavelength of an X-ray photon
with 10 keV is 10
8
cm, the distance to a neutron star is d =
d
100 pc
100 pc). However, if these objects are bare strange stars
with low masses, Constellation-X may not be able to resolve their
surfaces.
We also suggest that more electron cyclotron lines could be
detected by future telescopes. The eld strength allowing an ab-
sorption detectable in Constellation-X is accordingly from 2
10
10
to 9 10
12
G, while that for UVISS (its spectroscopy is
in two ranges: 125320 and 90115 nm) is from 3 10
10
to
10
11
G. The detection of lines in radio pulsars (especially in
low-eld millisecond ones) is interesting and important. We may
distinguish neutron or strange stars through the constraint of the
massradius relations, by knowing the cyclotron-determined B and
the timing result

P

P (e.g. see equation 5 for low-mass bare
strange stars), in the case that the lines form just above the stellar
surfaces.
The properties of radio emission from millisecond pulsars are
remarkably similar to those of normal pulsars (Manchester 1992),
although the inferred polar elds range approximately four orders,
based on equation (4). However, their elds range only 2 orders
when the ingredient of mass changing is included, according to
Table 1 and equation (13), if pulsars are actually strange stars. Why
has no millisecond pulsar with characteristic age T
c
< 10
8
yr been
found (or why are most millisecond pulsars so old)? The answer
could be that the initial periods, P
0
, of millisecond pulsars spread
over a wide range of 1 ms (or even smaller) to 50 ms, so that P
is not much larger than P
0
[it is thus not reasonable to estimate age
using T
c
= P/(2

P)]. Such a distribution of P
0
may be relevant to
their birth processes from AIC.
Can a core-collapse supernova also produce a low-mass bare
strange star? This possibility could not be ruled out in principle.
Likely astrophysical hints could be that the thermal X-ray emission
and rotation power of such a star should be lower than expected pre-
viously. Additionally, the cooling history of a low-mass strange star
should be signicantly different fromthat of solar-mass ones. Obser-
vationally, two isolated low-eld weak radio pulsars could be low-
mass normal pulsars (Lorimer et al. 2004): PSR J06092130 (P =
55.7 ms,

P = 3.1 10
19
) and PSR J22351506 (P = 57.9 ms,

P = 1.7 10
19
). The polar elds inferred from equation (5) of
low-mass bare strange stars could be higher than that from equation
(4) for rotation-powered pulsars, as one has
R
6
= 2.9 10
15

B
60
B
2
12
P

P, (44)
from equations (5) and (10). We can therefore obtain radii of
rotation-powered pulsars in the case that electron cyclotron absorp-
tion fromtheir surfaces is detected by advanced X-ray spectrometry.
If these two pulsars have polar elds B = 5 10
10
G, the radii of
PSR J06092130 and J22351506 are thus 2 and 1 km, respec-
tively. The very weak radio luminosity might be due to a very small
rotation energy I
2
/2 R
5
and a small potential drop in equa-
tion (14). In this sense, many low-mass bare strange stars may not
be detectable in the radio band.
It is proposed to search for low-mass bare strange stars, especially
with masses of (10
1
10
3
) M

, by re-processing the timing data

of radio pulsars. The systemic long-term variation of the timing
residual of the millisecond pulsar (B193721) might uncover a
companion star with mass 10
2
M

(Gong 2003). The companion

masses of pulsar/white-dwarf binaries are estimated to be a few
0.1 M

(Thorsett &Chakrabarty1999). Are all of these companions

real white dwarfs (or are some of themjust low-mass strange stars)?
Onlysome of the companions (a fewtens of percent) of pulsar/white-
dwarf systems have been detected optically. A further study on this
issue is then surely necessary.
Finally, it is essential to probe strange quark stars through vari-
ous observations of the environments of millisecond pulsars (plan-
ets, accretion discs), in order to distinguish these two scenarios for
the nature of the millisecond pulsars: to be (A) recycled or (B) of
supernova origin. In case (A), planets and residual accretion discs
could be around such pulsars, but possible mid- or far-infrared emis-
sion has still not been detected (Lazio & Fischer 2004; L ohmer,
Wolszczan & Wielebinski 2004). Another point against case (A)
is that it is observed that planets orbiting main-sequence stars can
only form around stars with high metallicities, but the planet cap-
tured by PSR B1620-26 is in a low-metallicity environment (the
globular cluster M4). In addition, the formation of pulsar planets
is still a matter of debate (Miller & Hamilton 2001). In case (B),
however, relevant observations could be well understood, as a pul-
sar (with possibly low mass) and its planet(s) may be born together
during a supernova (see the discussion at the end of Section 4 and
Xu & Wu 2003), and no infrared emission can be detected if no
signicant supernova fall-back disc exists. If the rst scenario is
right, infrared radiation from both the discs and the planets
13
could
be detectable by the Spitzer Space Telescope and by the present
SCUBA-1 or future -2 detectors of JCMT 15-m ground telescope.
However, if the latter is true, only negative results can be con-
cluded. Surely these are exciting and interesting subjects for when
these advanced telescopes are operational. Besides the radio-loud
pulsars, it is also valuable to detect submillimetre emission from
radio-quiet pulsar-like compact objects discovered in high-energy
X-ray bands, in order to nd hints of quark stars. In the model
presented, compact centre objects (CCOs, e.g. 1E 1207.45209)
and dim thermal neutron stars (DTNs, e.g. RX J1856.53754)
might have submillimetre radiation from the cold material around
the stars; but no submillimetre emission is possible if no accretion
occurs there. It is then very interesting to test and constrain the
models through observations of CCOs and DTNs at submillimetre
wavelengths.
In conclusion, if pulsar-like stars are strange quark stars, part of
them should consequently be of low mass unless one can convince
oneself that no astrophysical process results in the formation of low-
mass quark stars. As they are also X-ray emitters, we may expect
that some of them could have been detected by Chandra or XMM
Newton (e.g. the Chandra 1 Ms X-ray survey).
13
Comet-like planets with density of 1 g cm
3
are much larger and more
grainedthanstrange planets with10
14
gcm
3
, andconsequentlycontribute
more infrared emission.
C _
2004 RAS, MNRAS 356, 359370
370 R. X. Xu
ACKNOWLEDGMENTS
I would like to acknowledge discussions with Dr X. D. Li (for the
propeller phase) and Dr Bing Zhang (for pulsar deathlines), and to
thank various members of the pulsar group of Peking university for
stimulating discussions. This work is supported by National Nature
Sciences Foundation of China (10273001) and the Special Funds
for Major State Basic Research Projects of China (G2000077602).
REFERENCES
Alaverdyan G. B., Harutyunyan A. R., Vartanyan Yu. L., 2001, Astrophysics,
44, 265 (astro-ph/0409538)
Alcock C., Farhi E., Olinto A., 1986, ApJ, 310, 261
Alcock C. et al., 1993, Nat, 365, 621
Anchordoqui L., Dova M. T., Mariazzi A., McCauley T., Paul T., Reucroft
S., Swain J., 2004, preprint, hep-ph/0407020
Arons J., Lea S. M., 1976, ApJ, 207, 914
Banerjee S., Bhattacharyya A., Ghosh S. K., Raha S., Sinha B., Toki H.,
2003, MNRAS, 340, 284
Bhattacharya D. et al., 1996, in Johnston S., Walker M., Bailes M., eds,
ASPC Ser. Vol. 105, Pulsars: Problems and Progress. Astron Soc. Pac.,
San Francisco, p. 545
Bignami G. F., Caraveo A., Luca A. D., Mereghetti S., 2003, Nat, 423,
725
Bombaci I., 1997, Phys. Rev. C, 55, 1587
Bombaci I., Parenti I., Vida na I., 2004, ApJ, 614, 314 (astro-ph/0402404)
Buras R., Rampp M., Janka H.-Th., Kifonidis K., 2003, Phys. Rev. Lett., 90,
241101
Burwitz V., Haberl F., Neuh auser R., Predehl P., Tr umper J., Zavlin V. E.,
2003, A&A, 399, 1109
Carriere J., Horowitz C. J., Piekarewicz J., 2003, ApJ, 593, 463
Davies R. E., Pringle J. E., 1981, MNRAS, 196, 209
Davies R. E., Fabian A. C., Pringle J. E., 1979, MNRAS, 186, 782
de Gouveia Dal Pino E. M., Lazarian A., 2001, ApJ, 560, 358
De Luca A., Mereghetti S., CaraveoP. A., Moroni M., Mignani R. P., Bignami
G. F., 2004, A&A, 418, 625
Edwards R. T., van Straten W., Bailes M., 2001, ApJ, 560, 365
Eichler D., Livio M., Piran T., Schramm D. N., 1989, Nat, 340, 126
Elsner R. F., Lamb F. K., 1984, ApJ, 278, 326
Fryer C., Benz W., Herant M., Colgate S. A., 1999, ApJ, 516, 892
Glendenning N. K., 2000, Compact Stars. Springer-Verlag, Berlin
Gong B. P., 2003, preprint, astro-ph/0309603
Han J. L., Manchester R. N., Lyne A. G., Qiao G. J., 2004, in Camilo F.,
Gaensler B. M., eds, Proc. IAU Symp. 218, Young Neutron Stars and
Their Environments. Astron, Soc. Pat., San Francisco, p. 135
Ikhsanov N. R., 2003, A&A, 399, 1147
Kogut J. B., Sinclair D. K., Wang K. C., 1991, Phys. Lett., B263, 101
Lattimer J. M., Prakash M., 2001, ApJ, 550, 426
Lattimer J. M., Prakash M., 2004, Sci, 304, 536
Lazio J., Fischer J., 2004, AJ, 128, 842
Li X. D. et al., 1999, Phys. Rev. Lett., 83, 3776
Liebendoerfer M., 2004, in Proc. 12th Workshop on Nuclear Astrophysics,
preprint,astro-ph/0405029
Link B., 2003, Phys. Rev. Lett., 91, 101101
Lipunov V. M., 1992, Astrophyics of Neutron Stars. Springer-Verlag, Berlin
L ohmer O., Wolszczan A., Wielebinski R., 2004, A&A, 425, 763
Lorimer D. R. et al., 2004, MNRAS, 347, L21
Lyne A. G. et al., 2004, Sci, 303, 1153
Madsen J., 1998, Phys. Rev. Lett., 81, 3311
Madsen J., 1999, in Hadrons in Dense Matter and Hadrosynthesis, Proc. 11th
Chris Engelbrecht Summer School held in Cape Town. Springer-Verlag,
Berlin, p. 162
Manchester R. N., 1992, in Hankins T. H., Rankin J. M., Gil. J. A., eds,
Magnetospheric Structure and Emission Mechanics of Radio Pulsars,
Proc. of IAU Colloq. 128, Pedagogical Univ. Press, p. 204
Manchester R. N., Taylor J. H., 1977, Pulsars. Freeman, San Francisco
Mereghetti S., Bignami G. F., Caraveo P. A., 1996, ApJ, 464, 842
Mereghetti S., De Luca A., Caraveo P. A., Becker W., Mignani R., Bignami
G. F., 2002, ApJ, 581, 1280
Miller M. C., Hamilton D. P., 2001, ApJ, 550, 863
Pavlov G. G., Sanwal D., Teter M. A., 2004, in Camilo F., Gaensler B. M.,
eds, Young Neutron Stars and Their Environments, Proc. IAU Symp.
218, p. 239 (astro-ph/0311526)
Popov S. B., 2004, preprint, astro-ph/0403710
Qian Y.-Z., Wasserburg G. J., 2003, ApJ, 588, 1099
Rappaport S. A., Fregeau J. M., Spruit H., 2004, ApJ, 606, 436
Ren H. C., 2004, Lecture notes at CCAST Workshop on Progresses in Color
Superconductivity, Beijing, Dec. 811, preprint, hep-ph/0404074
Rischke D. H., 2004, Prog. Part. Nucl. Phys., 52, 197 (nucl-th/0305030)
Roger R. S., Milne D. K., Kesteven M. J., Wellington K. J., Haynes R. F.,
1988, ApJ, 332, 940
Ruderman M. A., 2003, in 4th AGILE Science Workshop on X-ray
and Gamma-ray Astrophysics of Galactic Sources, preprint, astro-
ph/0310777
Ruderman M. A., Sutherland P. G., 1975, ApJ, 196, 51
Sanwal D., Pavlov G. G., Zavlin V. E., Teter M. A., 2002, ApJ, 574, L61
Shapiro S. L., Teukolsky S. A., 1983, Black Holes, White Dwarfs, and
Neutron Stars: the Physics of Compact Objects. Wiley, New York
Tatsumi T., 2000, Phys. Lett. B, 489, 280
Tauris T. M., Bailes M., 1996, A&A, 315, 432
Tauris T. M., van den Heuvel E. P. J., Savonije G. J., 2000, ApJ, 530, L93
Thompson C., Duncan R. C., 1993, ApJ, 408, 194
Thorsett S. E., Chakrabarty D., 1999, ApJ, 512, 288
Usov V. V., 1998, Phys. Rev. Lett., 80, 230
Usov V. V., 2002, eConf C010815, 36 (astro-ph/0111442)
Vasisht G., Kulkarni S. R., Anderson S. B., Hamilton T. T., Kawai N., 1997,
ApJ, 476, L43
Wakatsuki S., Hikita A., Sato N., Itoh N., 1992, ApJ, 392, 628
Wallace P. M., Grifs N. J., Bertsch D. L., Hartman R. C., Thompson D. J.,
Kniffen D. A., Bloom S. D., 2000, ApJ, 540, 184
Weber F., 1999, J. Phys. G: Nucl. Part. Phys., 25, R195
Wijnands R., 2004, in van den Heuvel E. P. J., t Zand J. J. M., Wijers R. A.
M. J., eds, Proc. Symp. The Restless High-Energy Universe, preprint,
astro-ph/0309347
Xu R. X., 2002, ApJ, 570, L65
Xu R. X., 2003a, Chin. J. Astron. Astrophys., 3, 33
Xu R. X., 2003b, ApJ, 596, L59
Xu R. X., 2003c, in Li X. D., Wang Z. R., Trimble V., eds, IAU Symp. 214,
High Energy Processes, Phenomena in Astrophysics. Astron. Soc. Pac.,
San Francisco, p. 191
Xu R. X., 2003d, in Cheng K. S., Leung K. C., Li T. P., eds, Stellar As-
trophysics a Tribute to Helmut A. Abt. Kluwer, Dordrechet, p. 73
(astro-ph/0211563)
Xu R. X., Qiao G. J., 2001, ApJ, 561, L85
Xu R. X., Wu F., 2003, Chin. Phys. Lett., 20, 806
Xu R. X., Wang H. G., Qiao G. J., 2003, Chin. Phys. Lett., 20, 314
Xu R. X., Xu X. B., Wu X.J., 2001, Chin. Phys. Lett., 18, 837
Zavlin V. E., Pavlov G. G., Tr umper J., 1998, A&A, 331, 821
Zavlin V. E., Pavlov G. G., Sanwal D., 2004, ApJ, 606, 444 (astro-
ph/0312096)
Zhang B., 2003, Acta Astronomica Sinica (Supplement), 44, 215 (astro-
ph/0209160)
Zhou A. Z., Xu R. X., Wu X. J., Wang N., 2004, Astropart. Phys., 22, 73
This paper has been typeset from a T
E
X/L
A
T
E
X le prepared by the author.
C _
2004 RAS, MNRAS 356, 359370