SXP 323 - an unusual X-ray binary system in the Small Magellanic Cloud

Abstract

Spectroscopic observations taken with the VLT/UVES telescope/instrument are presented of the unusual Small Magellanic Cloud (SMC) X-ray binary system SXP 323 = AX J0051-733. This system shows a clear modulation at 0.71d in long term optical photometry which has been proposed as the binary period of this system. The high resolution optical spectra, taken at a range of phases during the 0.71d cycle, rule out this possibility. Instead it is suggested that this long-term effect is due to Non Radial Pulsations (NRP) in the Be star companion to SXP 323. In addition, the spectra show clear evidence for major changes in the (V/R) ratio of the double peaks of the Balmer lines indicative of asymmetries in the circumstellar disk. The complex structure of the interstellar lines are also discussed in the context of the SMC structure. Comment: Accepted in MNRAS

Full text

Page 1

arXiv:astro-ph/0506693v1 28 Jun 2005
Mon. Not. R. Astron. Soc. 000, 1–6 (2002) Printed 5 February 2008(MN LATEX style file v2.2)
SXP 323 - an unusual X-ray binary system in the Small
Magellanic Cloud⋆
M. J. Coe1, I. Negueruela2& V.A. McBride1
1School of Physics and Astronomy, Southampton University, SO17 1BJ, UK
2Departamento de Fisica, Ingeniera de Sistemas y Teora de la Seal, Escuela Politcnica Superior,
University of Alicante, Ap.99, 03080 Alicante, Spain.
June 2005
ABSTRACT
Spectroscopic observations taken with the VLT/UVES telescope/instrument are
presented of the unusual Small Magellanic Cloud (SMC) X-ray binary system SXP 323
= AX J0051-733. This system shows a clear modulation at 0.71d in long term optical
photometry which has been proposed as the binary period of this system. The high
resolution optical spectra, taken at a range of phases during the 0.71d cycle, rule out
this possibility. Instead it is suggested that this long-term effect is due to Non Radial
Pulsations (NRP) in the Be star companion to SXP 323. In addition, the spectra show
clear evidence for major changes in the (V/R) ratio of the double peaks of the Balmer
lines indicative of asymmetries in the circumstellar disk. The complex structure of the
interstellar lines are also discussed in the context of the SMC structure.
Key words: stars:neutron - X-rays:binaries - Magellanic Clouds
1 INTRODUCTION AND BACKGROUND
The Be/X-ray systems represent the largest sub-class of
massive X-ray binaries. A survey of the literature reveals
that of the 115 identified massive X-ray binary pulsar sys-
tems (identified here means exhibiting a coherent X-ray
pulse period), most of the systems fall within this Be coun-
terpart class of binary. The orbit of the Be star and the
compact object, presumably a neutron star, is generally wide
and eccentric. X-ray outbursts are normally associated with
the passage of the neutron star close to the circumstellar
disk (Okazaki & Negueruela, 2001).
X-ray satellite observations have revealed that the Small
Magellanic Cloud (SMC) contains an unexpectedly large
number of High Mass X-ray Binaries (HMXB). At the time
of writing, 47 known or probable sources of this type have
been identified in the SMC and they continue to be discov-
ered at a rate of about 2-3 per year, although only a small
fraction of these are active at any one time because of their
transient nature. Unusually (compared to the Milky Way
and the LMC) all the X-ray binaries so far discovered in
the SMC are HMXBs, and equally strangely, only one of
the objects is a supergiant system, all the rest are Be/X-ray
binaries. A recent review of these systems may be found in
Coe et al. (2005).
⋆Based on observations collected at the European Southern Ob-
servatory, Paranal, Chile (ESO 71.D-0093)
The source that is the subject of this paper, SXP 323
= AX J0051-733, is one of these SMC systems. It was re-
ported as a 323s pulsar by Yokogawa & Koyama (1998) and
Imanishi et al (1999). Subsequently Cook (1998), using MA-
CHO optical photometry, identified a 0.7d optically variable
object within the ASCA X-ray error circle. The system was
discussed in the context of it being a HMXB by Coe &
Orosz (2000) who presented some early OGLE data on the
object identified by Cook (1998) and modelled the system
parameters. Coe & Orosz identified several problems with
understanding this system, primarily that if it was a binary
then its true period would be 1.4d and it would be an ex-
tremely compact system. In addition, the combination of the
pulse period and such a binary period violates the Corbet
relationship for such systems (Corbet, 1986). A discussion
of the object as a possible triple system was presented in
Coe et al. (2002).
In order to try and understand this unusual system bet-
ter, high resolution optical spectra were obtained at a variety
of phases throughout the 1.4d period.
2OBSERVATIONAL DETAILS
Spectroscopy of the optical counterpart to SXP 323was ob-
tained with the Ultraviolet and Visual Echelle Spectrograph
(UVES) mounted on the 8.2-m VLT/UT2 (Kueyen). UVES
was used in the standard Diochroic #1 mode. The blue arm
was equipped with a 2048 × 4096 15-m pixel thinned EEV,

Page 2

2
M.J. Coe et al.
Obs. date Obs. time (UT)TJDPhase
9 June 2003
12 June 2003
13 June 2003
23 June 2003
4 July 2003
6 July 2003
11 July 2003
28 July 2003
31 July 2003
09:12
08:48
09:35
08:13
06:21
05:43
05:12
08:45
03:56
2799.88
2802.87
2803.90
2813.84
2824.76
2826.74
2831.72
2848.86
2851.66
0.00
0.22
0.68
0.71
0.14
0.93
0.97
0.17
0.13
Table 1. Table of VLT observations. The phase given in the
last column is that with respect to the precise OGLE period of
0.7081d.
providing coverage of the λλ 3821–4520˚ A range. The red
arm was equipped with a mosaic formed by a similar EEV
CCD (covering λλ 4727–5804˚ A) and a MIT/LL CCID-20
chip (covering λλ 5818–6835˚ A). The spectral resolution of
this configuration is 40,000.
Nine 2000-s exposures, each with a SNR of ∼35/pixel,
were obtained in service mode at random times. The 9 sets
are spread over the possible 1.4d period and give a good
coverage of different phases. The dates and phases of the
observations are presented in Table 1.
3 SPECTRAL CLASS
In spite of the higher resolution and S/N ratio, the data can-
not provide a better classification than the spectrum pre-
sented in Coe et al. (2002) as the range λλ 4500−4700˚ A,
rich in metallic lines is absent. A co-added blue spectrum,
with rather high S/N, shows no evidence of SiIV 4089˚ Aor
SiII 4128˚ A. Though the low metallicity of the SMC may
result in weaker lines, this is only compatible with a main-
sequence classification. The spectral type is constrained to
lie in the B0.5V–B1V range. In view of the complete ab-
sence of SiIV, we slightly favour the later classification over
the B0.5V preferred in Coe et al. (2002).
4SEARCH FOR BINARY MOTION
One of the main goals of the VLT observations was to search
for possible binary motion at either 0.71d or twice that value.
As can be seen from Table 1, a wide range of phases were
covered in the possible 0.7d or 1.4d cycle. To investigate pos-
sible radial velocity motion in any of the line features, the
spectra were subjected to the autocorrelation analysis func-
tion RVSAO within IRAF. The first spectra was assumed to
be at a phase 0.0 and subsequent spectra were then cross-
correlated against this one. This exercise was carried out for
both the blue data (3750 - 4500˚ A) and the middle spectral
range (4800 - 5700˚ A). The results from the latter are shown
in Figure 1 which set an upper limit of ∼30km/s for any ra-
dial velocities. The corresponding blue band upper limit was
slightly larger at ∼50km/s. There was no matching pattern
between the two sets of data implying a null result for any
binary motion.
Figure 1. Limits on any radial velocity motion as a function of
the 0.7d phase.
Figure 2. Average Balmer line profiles for the first four of the
series. The discontinuity to the left of Hβ is due to a slight mis-
match in adjacent echelle orders.
5BALMER LINE PROFILES
The VLT spectra cover all the Balmer lines from Hα to the
limit of the series. The first four Balmer lines are presented
in Figure 2. From this figure it is possible to see at least two
things: there is a clear progression in the strength of the
emission as we move down the series. While Hα and Hβ are
fully in emission, the emission components in upper Balmer
lines are weaker. This is perfectly normal in a Be star. Sec-
ondly, the emission component is clearly double peaked with
V and R components.
It is of great interest to use the emission features to
investigate possible V/R variations that might be indicative
of global one-armed oscillations, and hence asymmetries in
the circumstellar disk. To this end, the V/R ratio for the
Hβ line was measured throughout the series of observations
and the results are presented in Figure 3. It is clear from
this figure that a clear general trend exists in the ratio, the
values changing from ∼0.6 to ∼0.9 over 52 days. The other
Balmer lines showed a similar trend.

Page 3

SXP 323
3
Figure 3. Variations in the V/R ratio as a function of time for
the H beta line.
Line Feature 1
km/s
Feature 2
km/s
Feature 3
km/s
CaII H
CaII K
Na D
Na D
0
0
5
128
130
132
137
204
198
200
20010
Average4 132201
Table 2. Table of measured interstellar line features. The error
on each measurement is approximately ±5km/s.
6INTERSTELLAR LINE FEATURES
The high quality, high resolution spectra permit a detailed
investigation of the interstellar line profiles in the direction
of SXP 323. Specifically two sets of IS lines were investi-
gated: the CaII H & K lines (3968.5˚ A and 3933.7˚ A respec-
tively), and the Na doublet (5890.0˚ A and 5895.9˚ A). Fig-
ure 4 presents the region around the CaII H line and shows
the complex structure often associated with IS lines towards
the SMC (see, for example, the extensive set of examples in
Wayte, 1990). The profiles of the CaII H & K are almost
identical to each other, and confirm the presence of at least
three significant absorption features. These are indicated in
Figure 4 and listed in Table 2.
The region around the Na doublet is presented in Fig-
ure 5 and the values also given in Table 2. From the Na group
one can immediately see similarities and differences with the
CaII line profiles. The main similarity is the presence of at
least three major absorption features at essentially the same
velocities to those in the CaII lines. The other two features
seen in the CaII lines are weaker and cannot be ruled out as
also being present in the Na D lines.
The main difference, is the striking reversal in the rela-
tive strengths of the the line sets between the CaII and Na
features. In the CaII lines the strongest feature is the one
at the highest velocity (204km/s) while the weakest is the
one at the lowest velocity (0km/s). The complete opposite
is true for the Na lines.
Figure 4. Interstellar line features associated with the CaII H
line
Figure 5. Interstellar line features associated with the Na D lines.
7 POSSIBLE NRPS
If the possibility of binary motion is excluded as an explana-
tion for the 0.71d modulation, then other explanations must
be found. One possible alternative would be Non-Radial Pul-
sations (NRP) from the Be star. Such effects are well docu-
mented in B-type stars - see, for example, Uytterhoeven et
al, 2004. In addition, Fabrycky (2005) has suggested a NRP
explanation for two periods of 0.28d and 0.65d seen in the
OGLE data for SXP 702 (= XMMU J005517.9-723853).
The presence of strong emission components in most of
the lines renders the search for profile variability extremely
difficult. The Hei 4026˚ A was selected as showing particu-
larly little emission contamination and its profile was stud-
ied closely to search for any evidence of features moving in
phase with the 0.71d period. The result of this is shown in
Figure 6 in which the top line is the average profile for the
He4026˚ A line. This profile was then subtracted from each
of the individual nine spectra and the spectra were then
stacked in order of their occurence in the 0.71d phase - see
Table 1. It is not apparent from Figure 6 that there are
features left that move any semblance of phase coherence
through the spectra. Though the data are of a sufficiently
high resolution, the signal to noise limits the detection of
any such features. However, a moving feature such as that

Page 4

4
M.J. Coe et al.
Figure 6. The top line shows the average profile of the 4026˚ A he-
lium line. The other lines are the individual spectra after the av-
erage profile has been subtracted. They are arranged in order of
their occurence in the 0.71d phase, starting from the bottom at
phase 0.0. Also visible is the 4009˚ A helium line.
shown in Figure 1 of Uytterhoeven et al (2004) should have
been apparent.
One other signature of NRP behaviour would be a tem-
perature effect during the pulse cycle. To investigate this,
the MACHO data on this object were used (Coe & Orosz,
2000) and a colour index determined by simply subtracting
the red and blue bands. Figure 7 shows the result of folding
the colour data at the period of 0.71d and compares the re-
sult to the blue band data folded in the same manner. Also
shown in the top panel of this figure is the best fit sine wave
to the colour data. This sine wave was fitted with no con-
straints on its amplitude or phase. The amplitude of this sine
wave is 0.0006 magnitudes and the phase is such that the
star is slightly bluer by this amount when at its brightest.
8 DISCUSSION
8.1Binary motion
The upper limit results presented for the amplitude of any
radial velocity behaviour in SXP 323 (30 - 50km/s) may
be compared to the expected RV values for the 1.4d orbit
suggested by Coe & Orosz (2000). Assuming no eccentricity,
the expected velocities for a Keplerian orbit of a 15 Mo
star and a 1.4 Mo companion are of the order 480km/s for
Figure 7. Top panel : MACHO colour versus the phase of the
0.71d modulation; the line through the histogram is that of the
best-fit sine wave. Lower panel : the MACHO blue data folded at
the same 0.71d period and with the same phase as the top panel.
such an orbit. Implicit in this calculation is the assumption
that the system is not being viewed pole-on. To check this
assumption, the stellar vsini was determined using the He
4026˚ A line and the conversion formula presented in Equation
4 from Steele, Negueruela & Clark (1999). This method gives
vsini = 371±5 km/s for SXP 323, a higher value then any
those authors obtained for a long list of Be stars in their
work. It is therefore exceedingly unlikely that this star is
being viewed pole-on. Furthermore the clearly split Balmer
lines indicate that the circumstellar disk is also far from
being viewed pole-on. It is therefore possible to rule out
binary motion with a period of 0.71d as the explanation for
the sinusoidal modulation of the photometric light curve.
Laycock et al (2005) suggest that, based on X-ray ac-
tivity cycles, the correct binary period for this system is
108±18d. This period is much closer to the expected bi-
nary period of ∼180d based upon a rigorous interpretation
of the Corbet diagram (Corbet, 1986). Of course, the ac-
tual Corbet diagram exhibits quite a spread of values and it
is quite feasible that the correct binary period is the X-ray
one. Such a period, assuming a circular orbit, implies speeds
of ∼120km/s, which again exceeds the observational upper
limits presented here. A period of ∼180d and a circular orbit
would give radial velocity amplitudes of ∼100km/s.
However, most Be/X-ray binaries present very eccentric
orbits and the presence of regular outbursts in a wide orbit
(such as a 108-d or 180-d period would imply) is generally
associated with high eccentricity (cf. Okazaki & Negueru-

Page 5

SXP 323
5
ela 2001). In such a situation, we should have observed the
source close to periastron in order to see significant veloc-
ity changes, but the data presented here cover only 51 d,
which is less than half the X-ray period. Therefore the ab-
sence of obvious radial velocity variations within the 30 km/s
limit reported here does not rule out such a wide orbit. It
makes, though, exceedingly unlikely that the true binary
period could be as short as 0.71 d.
8.2V/R variability
From the measured V/R variability presented in Figure 3
it is possible to estimate the time for a complete cycle of
V/R changes to be at least ∼ 0.3yr. Such timescale does not
difer significantly from those seen in other Be/X-ray binaries
(e.g., Negueruela et al. 1998). SXP 323 is thought to be a
B1 star (Coe et al 2002) and, though there is not a strong
variation in V/R timescales with spectral class (Mennickent
& Vogt, 1991), such a value would be consistent with other
measurements of stars of a similar type. The presence of
obvious V/R variability in the profiles of emission lines also
argues strongly against a close binary. The existence of V/R
variability in Be stars is explained by the theory of global
one armed oscillations in circumstellar disks (Okazaki 2000).
The presence of global one-armed oscillations requires
the existence of a large disk around the Be star, extending to
several R∗ (Okazaki 2000). Such disk would not have space
to form in a close binary with a period of 0.7 - 1.4d.
8.3 NRP
Previous authors have shown that Be stars exhibiting cyclic
photometric variations show a weak colour variation with
phase. Spear, Mills & Snedden (1981) studied the Be star
28 Cygni (spectral type B3Ve) and determined a (U-B) vari-
ation of ∼0.01 mags in phase with the 0.7d cycle. As in the
observations presented here, the star got slightly bluer when
it got brighter, but the magnitude of the change in 28 Cyg
seems somewhat larger than that seen in SXP 323 (red-
blue ∼0.001 mags. However, Spear, Mills & Snedden (1981)
concluded that they were not seeing NRP behaviour but a
phenomenom related to the rotation of the photosphere.
The case for NRP behaviour is, unfortunately, not sup-
ported by evidence of line profile changes in these VLT data.
However, this effect may well be masked by the changes in
the circumstellar disk as indicated by the evolution of the
V/R ratio in Hβ (Figure 3). A data set taken over a much
shorter period of time (∼few days) could be much more suc-
cessful in exploring this possibility.
8.4 Interstellar lines
The multiple structure seen in the interstellar lines is indica-
tive of the complex spatial structure of the SMC.
Wayte (1990) obtained CaII absorption spectra of 17
stars in the SMC and presented their profiles. It is clear from
studying their results that there is a huge range of variations
in these profiles across the SMC reflecting its complex struc-
ture. The nearest object to SXP 323 in the data of Wayte is
Sk 35 which reveals features at a heliocentric velocity of ∼50,
∼115, ∼160 and ∼190km/s. Though some of these are close
to our observed features, there are significant differences as
well (for example, our data show nothing at 115km/s). This,
though, may not be too surprising given the large spread of
velocity features seen by Wayte across the SMC.
Danforth et al (2002) used FUSE data on 37 objects to
identify two major features: one at 125km/s and another
at 156km/s. Both of those features are probably present
in our data, but at slightly different velocites (132km/s
and ∼167km/s). The same authors also report a feature at
∼180km/s often seen in the Bar - where SXP 323 lies. This
velocity lies between two of the main components reported
here at 167km/s and 201km/s.
Most authors agree that there are major separate struc-
tures throughout the SMC and our data reflect this level of
detail. Studies of many of the other 40-50 SXP sources will
add vital information to this complex question.
9CONCLUSIONS
The VLT observations reported here conclusively rule out
the possibility of any binary motion as an explanation for
the 0.71d period seen in the optical counterpart to SXP
323. This fact combined with the slight colour change seen
in the photometric data suggests that instead this is a phe-
nomenum associated with the Be star - either NRP or the
rotational modulation of photospheric structures. However,
the probable stability, and lack of any phase change in the
entire MACHO and OGLE data sets of many years (Coe
et al 2002) strongly supports the the NRP case. Future ob-
servations over a much shorter time base should be able to
confirm this directly.
In addition, the spectra show clear evidence for major
changes in the (V/R) ratio of the double peaks of the Balmer
lines indicative of asymmetries in the circumstellar disk. Fi-
nally, the complex structure of the interstellar lines present
clear evidence for multiple structures within the SMC.
10 ACKNOWLEDGEMENTS
Most of the results presented here are based on observations
collected in Service Observing mode at the European South-
ern Observatory, Chile under programme 71.D-0093(A). IN
is a researcher of the programme Ram´ on y Cajal, funded
by the Spanish Ministerio de Educaci´ on y Ciencia and the
University of Alicante, with partial support from the Gener-
alitat Valenciana and the European Regional Development
Fund (ERDF/FEDER). This research is partially supported
by the Spanish MEC under grant AYA2002-00814.
REFERENCES
Coe M.J. & Orosz J.A. 2000 MNRAS 311, 169.
Coe M.J., Haigh N.J., Laycock S.G.T., Negueruela I. &
Kaiser C.R. 2002 MNRAS 332, 473.
Coe M.J., Edge W.R.T.E., Galache J.L., McBride V.A.,
2005, MNRAS 356, 502.
Cook. K 1998 IAUC 6860.
Corbet R.H.D. 1986 MNRAS 220, 1047.
Danforth C. W., Howk J. C., Fullerton A. W., Blair W. P.,
Sembach K. R. 2002 ApJS 139, 81.