A&A 418, 103-116 (2004)
DOI: 10.1051/0004-6361:20034488
M. Messineo 1 - H. J. Habing 1 - K. M. Menten 2 - A. Omont 3 - L. O. Sjouwerman 4
1 - Leiden Observatory, PO Box 9513, 2300 RA Leiden, The Netherlands
2 -
Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
3 - Institut d'Astrophysique de Paris, CNRS & Université
Paris 6, 98bis Bd Arago, 75014 Paris, France
4 -
National Radio Astronomy Observatory, PO Box 0, Socorro NM 87801, USA
Received 9 October 2003 / Accepted 19 December 2003
Abstract
We present a compilation and study of DENIS, 2MASS,
ISOGAL, MSX and IRAS 1-25 m photometry for a sample of 441 late-type stars in the inner Galaxy, which we previously searched for
86 GHz SiO maser emission (Messineo et al. 2002). The comparison of the
DENIS and 2MASS J and K
magnitudes shows that most of the SiO targets are indeed variable stars. The MSX colours and the IRAS
[12]-[25] colour of our SiO targets are consistent with those of
Mira type stars with dust silicate feature at 9.7
m feature in
emission, indicating only a moderate mass-loss rate.
Key words: stars: AGB and post-AGB - infrared: stars - stars: variables: general - stars: circumstellar matter - masers - Galaxy: stellar content
To understand the Galactic structure and kinematics it is important to combine the kinematic information and the stellar properties, e.g. luminosities, which can provide a distance estimate. Good photometry on infrared point sources toward the inner Galaxy is now available from large surveys such as DENIS (Epchtein et al. 1994), 2MASS (Cutri et al. 2003), ISOGAL (Schuller et al. 2003; Omont et al. 2003) and MSX (Egan et al. 1999; Price et al. 2001). Since the high extinction toward the inner Galaxy precludes studies at optical wavelengths, these infrared data permit a unique view of its stellar population. The combination of near- and mid-infrared photometry enable us to examine the nature of the stars, i.e. to derive their luminosities, mass-loss rates, and to discriminate againts foreground stars.
To improve the line-of-sight velocity statistics, we conducted 86 GHz
SiO maser line observations of 441 late-type stars in the inner Galaxy (
,
|b| <
1) with the IRAM 30-m telescope (Messineo et al. 2002, hereafter Paper I). This paper is part
of a series devoted to characterise the properties, i.e. mass-loss
rates and luminosities, of the 441 sources previously targeted to
search for 86 GHz SiO maser emission (Paper I).
Here (Paper II) we present the available near- and mid-infrared
photometry of the targeted sources ("SiO targets'' hereafter). In
another paper (Messineo et al. 2004a, hereafter Paper III)
we deal with extinction
correction and finally in the last paper (Messineo et al. 2004b, hereafter
Paper IV) we
compute and analyse the luminosities of the SiO targets.
![]() |
Figure 1: Location of the 441 SiO targets in Galactic coordinates. The 379 MSX counterparts are shown as open circles, the 267 ISOGAL counterparts as crosses. Overlap of MSX and ISOGAL sources resemble filled symbols. Four points fall outside the figure. |
Open with DEXTER |
Table 1: Number of counterparts of our SiO targets.
The spatial distribution of the 441 targets is shown in Fig. 1. These SiO targets are divided into two subsamples: 253 sources were selected from the ISOGAL catalogue, ("the ISOGAL sample''), and 188 sources from the MSX catalogue, ("the MSX sample''). The ISOGAL and MSX samples are examined to test whether both are drawn from the same "parent population''. Brightness variability is studied by a comparison of the DENIS and 2MASS photometry.
The structure of the paper is as follows: in Sect. 2 we identify our SiO targets in various infrared catalogues and collect their magnitudes finding for many stars up to fourteen different measurements. In Sect. 3 we compare the statistical differences between our ISOGAL and MSX samples. In Sect. 4 we summarise additional information found with SIMBAD, e.g. variability and other types of masers, and we derive the probability of association between the radio maser and the infrared counterpart. The brightness variability of the stars is discussed in Sect. 5. In Sects. 6 and 7 we analyse the mid-infrared colours of the stars and compare them with those of OH/IR stars. The main conclusions are summarised in Sect. 8.
The individual source numbers (e.g. #99) are taken from Table 2 (86 GHz SiO maser detections) and Table 3 (non-detections) in
Paper I unless otherwise indicated. The SiO maser
emission in this paper generally refers to the 86 GHz (
)
SiO maser only and not to the 43 GHz (J=1-0) SiO masers. Velocities in this paper refer to line-of-sight velocities
with respect to the Local Standard of Rest.
We cross identified all SiO targets, whether taken from the ISOGAL or from the MSX database, with all infrared catalogues available to us, the DENIS, 2MASS, ISOGAL, MSX and IRAS survey catalogues. The results are summarised in Table 1. In the following, we briefly recall the criteria used for the selection of the ISOGAL and the MSX samples and describe the modality of the cross-correlations. More details on the selection criteria can be found in Paper I.
Sources were selected from a preliminary version of the ISOGAL
catalogue by their extinction-corrected 15 m magnitude, [15]0,
and their (K
0-[15]0) and (
[7]0-[15]0) colours approximately
corrected for extinction (Sect. 7.3.2 Paper I). The
brightest 15
m sources, [15]0<1.0, and those with
([7]
0-[15]0) < 0.7 and with (K
0-[15]0) < 1.95 were excluded
since they are likely to be foreground stars or non-AGB stars or AGB
stars with very small mass-loss. Further, sources with
[15]0 > 3.4were excluded since they are likely to show SiO maser emission fainter
than our detection limit of 0.2 Jy. Sources with (
[7]0-[15]0) >
2.3 were excluded since they are likely to be compact HII regions or other young stellar objects or planetary nebulae. Those
with (K
0-[15]0) > 4.85 were excluded because they are likely to
be OH/IR stars with a high mass-loss rate or young stellar objects.
Moreover, known OH/IR stars were discarded as the kinematic data are
already known.
The Midcourse Space Experiment (MSX) is a survey at five mid-IR bands
ranging from 4.3 m [B1 band] to 21.4
m [E band], with a
sensitivity of 0.1 Jy in A band (8.28
m) and a spatial
resolution of 18.3
(Price et al. 2001). The survey covers the
Galactic plane to
latitude. Version 1.2 of the
MSX-PSC (Egan et al. 1999) lists more than 300 000 point sources with an
rms astrometric accuracy of
2
.
The MSX catalogue gives
the source flux density, F, in Jy. Magnitudes are obtained adopting
the following zero-points: 58.49 Jy in A (8.26
m) band, 26.51 Jy in C (12.12
m) band, 18.29 Jy in D (14.65
m) band and
8.80 Jy in E (21.41
m) band (Egan et al. 1999).
For the MSX source selection we used flux densities in the A and Dbands which have wavelength ranges roughly similar to the ISOGAL 7 and
15 m bands. We selected those non-confused, good-quality sources
in A and D band (flag > 3), which show variability in the Aband. We avoided the reddest stars,
FD/FA > 2.3 (A-D > 2.2 mag). We also avoided the bluest and most luminous stars with
FD/FA
< 0.6 (
A-D < 0.75 mag) and FD > 6 Jy (D < 1.2 mag) since they
are likely to be foreground stars or supergiants
(Schultheis et al. 2003; Schuller 2002). Furthermore, following the
classification of Kwok et al. (1997) of IRAS sources with low-resolution
spectra, we used the C to E band ratio to discard very red
(
FE/FC > 1.4,
C-E > 1.55 mag) sources, which are likely to be
young stellar objects or OH/IR stars with thick envelopes. Known OH/IR stars were excluded.
![]() |
Figure 2:
Associations between our SiO ISOGAL
targets and the MSX catalogues. The distribution of the angular
separations (
![]() |
Open with DEXTER |
Out of the 188 SiO targets selected from the MSX catalogue, only 14 are located in an ISOGAL field, and we found ID-PSC identifications
for all of those within 5
from the MSX positions. For the
remaining 174 sources we searched the DENIS PSC.
As the ISOCAM 15 m and MSX D filters are similar, we compared
the ISOCAM 15
m magnitudes, [15], and the MSX D band
magnitudes, D, of the 154 sources detected at 15
m in both
surveys and found good agreement (Fig. 3); the
average difference D-[15] is -0.04 magnitude and the standard
deviation is 0.3 mag, resulting from the combination (
0.15 mag)
of the photometric errors of both catalogues and from the possible
intrinsic source variability.
![]() |
Figure 3:
Difference between the MSX band Dmagnitude, D, and the ISOCAM 15 ![]() |
Open with DEXTER |
![]() |
Figure 4:
Associations of our SiO MSX targets
in the DENIS catalogue. Top panel: distribution of K![]() ![]() ![]() ![]() |
Open with DEXTER |
The distribution of angular separations of the K
band counterparts
identified is shown in Fig. 4. The mean and
median separations are 3.0
and 2.7
with a standard
deviation of 1.8
,
respectively, which is consistent with the
expected scatter due to positional uncertainties for the MSX and DENIS
catalogues.
To find the distribution of chance associations we again
searched the nearest neighbours after shifting the coordinates of the
SiO targets by 30
.
The resulting distribution of separations
and magnitudes is also shown in Fig. 4. The real
associations are on average much brighter and closer than the chance
associations. To compute the expected number of incorrect
identifications we divided the "chance'' distribution in magnitude
bins,
,
and in each bin computed the normalised
cumulative chance distribution of the separations, f(m,r), i.e., the
fraction of all chance associations with those magnitudes and within a
radius r. We then computed the sum
over all
"real'' associations, each being characterised by m and r. Thus,
f(m,r) gives the probability for an association to be spurious, and
the sum over all sources yields the total expected number of spurious
identifications. When we consider only the brightest identifications
for the 154 sources with a K
< 9 mag counterpart, only two
spurious identifications are expected. For the 34 possible
identifications with K
> 9 mag, we would expect three to be
spurious. In total, we expect that about five of our identifications
may be incorrect.
Since for the fainter possible counterparts the chance of a false
identification is higher, we looked for brighter possible
identifications somewhat further away, which due to the lower surface
density of brighter sources might have a higher probability to be the
actual counterpart. In a few cases (#231, #243, #367, #405, #406,
and #424) we found brighter sources somewhat further away but with a
lower value of f(m,r) than the closest identification. This
suggested that these brighter sources were more likely to be the
correct counterparts, which we therefore retained. For these MSX
targets with dubious near-infrared association, we additionally
examined other MSX sources in their surrounding and checked for
possible astrometric shifts between the DENIS and MSX coordinates which
could uniquely identify the correct near-infrared counterpart.
However, because of the low MSX source density (1 source per
), only few associations could be confirmed
in this way. Seven sources could only be associated with a K
> 11 mag counterpart (#162, #273, #363, #403, #417, #423, #443) within
15
(beam size of the IRAM telescope).
![]() |
Figure 5:
2MASS K![]() ![]() |
Open with DEXTER |
To avoid misidentification, because of the high source density of
the 2MASS survey and because the DENIS counterparts are mostly
brighter than 11 mag in K,
we limited the 2MASS K
magnitude to
K
< 11. However, seven sources in our sample, #162, #273,
#363, #403, #413, #421 and #423, could be associated only with
2MASS sources fainter than K
= 11 (those sources are also faint
in DENIS) within 15
,
which is the beam size of the IRAM
telescope. Positional associations were confirmed via overplotting
the 2MASS counterpart image with both the SiO targets and the DENIS
sources. Finding charts were obtained for all the stars with 2MASS
images, an example of which is given in Fig. 5. We
found 439 2MASS counterparts and missed only two. In fact, after image
inspections, the potential 2MASS counterparts for two sources, #224
and #298, were eliminated as their positions on the 2MASS images were
marked by artifacts.
There is a non-Gaussian tail at large separations. We have
individually checked all the sources with separation larger than 3
and note that they all have K
< 6. We attribute the
large separation to saturation of the DENIS detector. Saturated
pixels are an obstacle to the correct determination of the source
centroid and this affects the astrometry of saturated
stars. Furthermore, most of these bright sources do not have any I associations and therefore the J/K
astrometry is kept. The mean,
median and standard deviation of the separations between the ISOGAL SiO targets and the 2MASS
associations with K
>6.5 mag are 0.4
,
0.3
and
0.4
;
while for 2MASS associations with K
<6.5 mag they are
1.9
,
2.1
and 1.5
.
![]() |
Figure 6: Distribution of the angular separations between the SiO targets (positions as in Paper I) and the 2MASS associations. The continuum line shows the distribution of the ISOGAL SiO targets and thecorresponding y-axis is on the left side. The dashed line shows the distribution of the MSX SiO targets and the relative y-axis is onthe right side. |
Open with DEXTER |
![]() |
Figure 7:
Upper panel: separation of the
possible IRAS counterparts, expressed in IRAS sigma units, against
angular separation. Crosses indicate IRAS sources with possible MSX
counterparts closer than those associated with the SiO targets; these
are likely to be unrelated to the SiO targets. The dotted horizontal
line is the upper limit that is selected. Lower panel: the
continuum line shows the distribution of the separations (in sigma
units) of the possible IRAS counterparts. The dashed line shows the
distribution of the chance associations obtained shifting the source
coordinates by 250
![]() ![]() |
Open with DEXTER |
We selected only the 165 IRAS associations within 3.5
error
ellipse of the IRAS PSC, to reduce the chances of spurious
associations to
2% (3 sources). A comparison of 12
m
fluxes of the prospective counterparts with the ISOGAL and MSX fluxes
(Fig. 8) shows a good agreement, confirming that
the IRAS identifications are proper. About 35% of our SiO targets
have counterparts in the IRAS PSC: 65% of the MSX sample and 17% of
the ISOGAL sample. Of those, 96% are detected at 12
m, 87% at
25
m, 6% at 60
m and 4% at 100
m, and 56% are reported
in the IRAS catalogue as variables (flag > 80).
![]() |
Figure 8:
Difference between the MSX magnitude in
the C band, and the IRAS 12 ![]() ![]() ![]() ![]() |
Open with DEXTER |
Within a radius of 5
to our SiO targets we searched for
possible visual counterparts in the Tycho 2 and USNO-A2.0 catalogues.
We found possible associations for 85 SiO targets, of which 27 are
validated by corresponding I band counterparts. Their R magnitudes
range from 10.9 to 17.9 and B magnitudes from 13.4 to 20.6.
Considering that the pulsation amplitude of Mira stars increases at
shorter wavelenghts and can be up to 8 mag in the visual
(Smak 1964), all those visual stars are possible counterparts of
our SiO targets. Most of them are located at latitude
,
but the extinction value inferred by their colours
are much smaller than the median of their surrounding stars
(Paper III). Therefore they are likely to be foreground
stars. Two SiO targets, #7 with
kms-1) and #139 with
kms-1), have extinction value
inferred by their colours
= 5.5 and 3 mag, respectively,
consistent with the median extinction of their surrounding stars
(Paper III). Furthermore, the velocity of #139 is
inconsistent with being a foreground star. Therefore we conclude that
they are likely located in the Galactic bulge, in regions of low
interstellar extinction. In fact, #139 is located in the optical
window W0.2-2.1 at
(Stanek 1998; Dutra et al. 2002), and #7 has an extinction value typical of
bulge ISOGAL fields with
(Ojha et al. 2003).
Table 2:
Infrared counterparts of the SiO
targets.
The identification numbers are the same as in
Table 2 and Table 3 of Paper I.
![]() |
Figure 9: Mid-infrared colour-magnitude diagrams. Left-hand panel: all data from ISOGAL. Right-hand panel: all data from MSX. Filled circles indicate targets from the ISOGAL sample and crosses indicate targets from the MSX sample. The two samples have similar mid-infrared colours, due to selection criteria. The numbers of MSX and ISOGAL targets between the two panels vary as explained in Sect. 2.3. |
Open with DEXTER |
A large variety of names exists to indicate oxygen-rich AGB stars
characterized by different pulsation properties and/or mass-loss rate:
semi-regular (SR) stars and Mira stars (H
in emission, visual
pulsation amplitude larger than 2.5 mag), large amplitude variables
(LAV), long period variable (LPV) stars (when their periods are longer
than 100 days), and OH/IR stars (with 1612 MHz OH maser emission). In
the IRAS colour-colour diagram the oxygen-rich AGB stars are
distributed on a well-defined sequence of increasing shell opacity and
stellar mass-loss rate (e.g. Habing 1996), which goes from SRs and
Miras with the bluest late-type colours and the 9.7
m silicate
feature in emission, to the coldest OH/IR stars with the reddest
colours and the 9.7
m silicate feature in absorption. The
sequence of increasing shell opacity corresponds also to an increasing
K
- [15] or K
- [12] colour
(e.g. Olivier et al. 2001; Whitelock et al. 1994; Ojha et al. 2003, Paper I).
SiO maser emission is generated in the envelopes of mass-losing AGB
stars, close to their stellar photospheres and it occurs more
frequently towards oxygen-rich Mira stars than towards other AGB stars
(including SR and OH/IR stars) (Bujarrabal 1994; Nyman et al. 1993).
Therefore, for our 86 GHz SiO maser survey we selected the brightest
sources at 15 m with colours of Mira-like stars
(Paper I). The ISOGAL and MSX samples were selected to
have similar 7 and 15
m colours, as shown in Fig. 9. For the ISOGAL sample a range of intrinsic
(K
- [15])-colour was selected (see Sect. 2.1), but for
the MSX sample no near-infrared counterparts were available at the
time of the observations, and thus no K
magnitudes. As an
alternative to the (K
- [15]) criterion, for the MSX sample we
imposed an upper limit to the ratio of the fluxes in the E(21
m) and C (12
m) bands (
C-E < 1.55 mag). Both criteria
were defined in order to avoid non-variable objects and thick envelope
objects, but these criteria are not equivalent. The emission in the K
band is dominated by the stellar emission attenuated by the
circumstellar absorption, while circumstellar dust emission
contributes strongly to the mid-infrared radiation.
AGB stars with thin envelopes have
1.5 mag, while sources
with
1.5 mag are AGB stars with thick envelopes, post-AGB
stars and young stars (Lumsden et al. 2002; Sevenster 2002). Our sample of
SiO targets includes very few sources with C-E> 1.5.
The K - [15] and K
- [12] are good indicators of mass-loss rate
for AGB star with shells at few hundred Kelvin. All Miras have redder
K
- [12] than non-Mira stars from 1.8 mag up to 6 or even 14 mag
(e.g. Olivier et al. 2001; Whitelock et al. 1994). Thick-envelope OH/IR stars
have typically K
- [15] >4 (Ortiz et al. 2002). The ISOGAL SiO target
sample is characterised by a smaller range of (K
- [15]) or
(K
- D) values than the MSX sample, as shown in Fig. 10, suggesting that the MSX sample includes a tail of
sources with optically thick envelopes. However, this will be
verified after correction for extinction
(Papers IV, III).
Figure 10 shows a strong correlation, between K
and
(K
- [15]) or (K
- D), which is due to the way we selected our
original sample. In fact, due to the general correlation of SiO maser
emission and infrared luminosity (Bujarrabal et al. 1987), we
selected only sources with
[15] < 3.4 in order to be able to detect
the expected SiO line. The [15] and D magnitudes range from
3.4 to
1.0. This narrow range generates the correlation
seen in Fig. 10.
Figure 10 also shows that the two samples, ISOGAL and MSX, overlap largely, although there are some minor systematic differences. There is a vertical shift between the ISOGAL and MSX sequence, with the MSX sources brighter for a given colour. This effect is due to the different sensitivity between the MSX D band and ISOGAL surveys. MSX targets have on average a brighter D (or [15]) magnitude than the ISOGAL targets (see Fig. 9). This fact translates in differences in distance between the two samples and suggest that the MSX stars are closer on average; the two samples are also distributed differently in longitude (see Fig. 1).
On the basis of this comparison in the following we combine the
results obtained with the ISOGAL sample with those obtained with the
MSX sample, taking into account that there are some minor differences
between the two samples.
![]() |
Figure 10:
2MASS K![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
Sources #7, #286 and #303 are included in the catalogue of late-type stars in the inner Galactic region by Raharto et al. (1984) as spectral types M 6, M 7 and M 6.5, respectively.
#153 is a well known Mira star, TLE 53, with a period of 480 d,
located in Baade's window (e.g. Glass et al. 1995).
Our sample also includes 15 LPVs found by Glass et al. (2001) within
0.3
from the Galactic center and 19 candidate variable stars
from the list of Schultheis et al. (2000) (listed in Tables 2
and 3 of Paper I).
A few sources are given in the literature as possible red supergiants or extremely luminous AGB stars: #25, #32, #92 and #295 correspond to sources #6, #8, #31 and #5 of Nagata et al. (1993), respectively; #356 is classified as bulge M supergiant by Raharto (1991) and Stephenson (1992) also lists it among distant luminous early type stars.
#127 (IRAS 17500-2512), #178 (IRAS 18040-2028), #188 (IRAS
18060-1857), #252 (IRAS 18285-1033), #265 (IRAS 18367-0507)
and #434 (IRAS 18415-0355) are listed by (Kwok et al. 1997) among
sources detected with the IRAS Low Resolution Spectrometer, in the
range 8-23 m and with a resolution,
,
20-40. Four spectra are noisy or incomplete, while the spectra
of #178 (IRAS 18040-2028), #434 (IRAS 18415-0355) are classified
as featureless; they are probably evolved stars with negligible
amounts of circumstellar dust. For those two objects, at longitudes 9.7 and 28.6
respectively, we also compute a moderate mass-loss
rate of
yr-1(Paper III).
#164, IRAS 17590-2412, is classified as a Li K giant star by
de La Reza et al. (1997). There is a significant difference between the SiO
heliocentric velocity
kms-1 and the optical
heliocentric velocity
kms-1
(Torres 1999; de la Reza, priv. communication)
Since the SiO maser velocity is
usually coincident with the stellar velocity within few kms-1 (e.g. Habing 1996), we suggest that the SiO emitter is not
associated with the G8II star, IRAS 17590-2412/PDS 482, which
however is the only mid-infrared source within the IRAM beam and the
association between the ISOGAL and DENIS source is of excellent
quality (flag = 5).
#189, IRAS 18059-2554, is given by Lynch & Rossano (1990) as a possible
member of the globular cluster NGC 6553. The stellar line-of-sight
velocity, obtained through the SiO maser line, is 161.2 kms-1. Because
the cluster mean line-of-sight velocity is 7 kms-1 with
kms-1(Coelho et al. 2001), we conclude that IRAS 18059-2554 is not a
member of the cluster.
Twenty-eight of our 86 GHz SiO targets were previously observed for 43 GHz SiO maser emission. These are discussed in Sect. 4.5 of
Paper I. For the strongest 86 GHz maser sources
within 2.2
of the Galactic Centre from Paper I
we recently used the Very Large Array (VLA) to observe the two 43 GHz
SiO maser lines (v=0 and v=1) simultaneously (Sjouwerman et al. 2004).
We excluded from our SiO maser survey the OH/IR stars detected by Sevenster et al. (2001,1997a,b), Sjouwerman et al. (1998) and Lindqvist et al. (1992). However, due to intrinsic source variability and the limited sensitivity of the Sevenster et al. surveys, we still included 4 OH/IR stars, as found with a SIMBAD search: #181, #226 and #257, all with detected SiO maser emission, coincide in position and velocity with OH9.84+0.01, OH17.43-0.08 and OH25.05+0.28, respectively (Blommaert et al. 1994); #409 (IRAS 18142-1600), not detected in our SiO survey, corresponds to OH #280 (OH14.805+0.150) listed by Te Lintel Hekkert et al. (1989).
![]() |
Figure 11:
Upper panel: difference between
2MASS and DENIS J magnitudes versus the DENIS J magnitude.
Filled squares represent the ISOGAL SiO targets. For comparison small dots
show ID-PSC and 2MASS associations obtained in several ISOGAL fields,
which have a distribution of their J variations consistent with a
Gaussian distribution centred at zero. No correction for offset in
the photometric zeropoint was applied. Open circles represent the MSX
targets, which have indications of variability at mid-infrared wavelengths.
The error bars shown within the box are ![]() ![]() ![]() ![]() |
Open with DEXTER |
The J and K
filters used by DENIS and 2MASS are similar and
therefore the measurements obtained during the course of the DENIS and
2MASS surveys are directly comparable. For non variable sources, the
differences between the J magnitude of DENIS and 2MASS and the K
magnitude of DENIS and 2MASS are smaller than 0.15 mag (Schultheis & Glass 2001; Delmotte et al. 2002, and
present work).
The DENIS observations were performed between 1996 and 2000, while the
relevant 2MASS observations were performed between 1998 and 2000; AGB
variables have periods from 50 to 1000 days, therefore the interval of
time between the observations makes it possible to derive variability
information.
![]() |
Figure 12: The difference of the 2MASS and DENIS J magnitude versus the time between the 2MASS and DENIS observations. |
Open with DEXTER |
![]() |
Figure 13:
Difference between the 2MASS and DENIS J magnitudes versus the difference of the 2MASS and DENIS K![]() |
Open with DEXTER |
For each of the 61 ISOGAL fields containing our SiO targets we
retrieved the corresponding 2MASS sub-catalogue, and cross-correlated
the ID-PSC and the 2MASS point source positions. For our ISOGAL
sample the difference between the 2MASS and DENIS J and K magnitudes is shown as a function of the corresponding DENIS magnitude
in Fig. 11. Figure 12 shows the
difference between the 2MASS and DENIS J magnitudes plotted against
the time between the 2MASS and DENIS observations. The distribution
of the magnitude variations of the ISOGAL targets is different
compared to that of random field objects. The Kolmogorov-Smirnov test
gives a zero probability for the ISOGAL targets and field stars to be
extracted from the same population. For 55% of our ISOGAL stars, the
difference between both the 2MASS and the DENIS J and the 2MASS and
the DENIS K
magnitudes is larger than 3 times the dispersion
measured in the corresponding field. Therefore, our sample contains
mostly variable stars.
Due to the simultaneity of the J and K
measurements in both the
DENIS and 2MASS surveys, a correlation is expected between the
variation in the J magnitude (
)
and in the K
magnitude
(
K
). As shown in Fig. 13 such a correlation
exists. A linear least squares fit yields
For comparison, in Fig. 14, we also show the relation between the pulsation amplitudes in the J and K SAAO bands for two different samples of oxygen-rich stars in the solar neighbourhood (Olivier et al. 2001) and the South Galactic Cap (SGC) (Whitelock et al. 1994). The two samples have a different period distribution; most of the stars from the solar neighbourhood sample have periods between 500 and 700 d, while most of the SGC stars have periods between 150 and 450 d. Overplotting our best-fit, we see that it aligns well with the distribution of the two samples of LPVs.
A monitoring program of the near-infrared magnitudes of our SiO maser sample will provide pulsation periods and estimates of the source distances through the period-luminosity relation.
There is no correlation between the variability indication and the
detection of SiO maser emission.
![]() |
Figure 14: Pulsation amplitude in J band versus amplitude in K band for two samples of oxygen-rich variable stars: one in the solar neighbourhood from Olivier et al. (2001) (starred points) and another in the South Galactic Cap from Whitelock et al. (1994) (filled circles). Superimposed is our best fit from Fig. 13. |
Open with DEXTER |
![]() |
Figure 15:
Distribution of the IRAS [12]-[25]
colours,
![]() ![]() |
Open with DEXTER |
We cannot determine the distribution of our target sources in the
two-colour IRAS diagram of van der Veen & Habing (1988) since for most of our
sources we have only upper flux limits at 60 m. Figure 15 shows the distribution of the IRAS [12]-[25]
colours and recalls the regions of the IRAS two-colour diagram that
separate different classes of evolved stars with circumstellar
envelopes, from the bluer Mira to the thick envelope OH/IR stars. The
[12]-[25] colours of our selected sources range from -1.0 to 0.4, peaking at -0.2, corresponding mostly to regions IIIa and
VIb. Region IIIa represents sources with moderate dust emission,
being populated mostly by oxygen-rich stars with silicate emission
(Kwok et al. 1997; van der Veen & Habing 1988). Region VIb, whith 60
m excess,
contains a mixed population of early type stars with line emission and
planetary nebula; however this region is scarcely polulated.
Considering the distribution of the IRAS good quality sources, only 5% of them are located in region VIb. We conclude that our targets
are mostly Mira stars with moderate mass-loss rate, in agreement with
the selection criteria. For comparison, the distribution of the
[12]-[25] colours of the OH/IR stars of Sevenster (2002) is also
shown in Fig. 15. OH/IR stars are distributed over a
larger and redder colour range, although partially overlapping with
the colours of the SiO sample; they can have significantly (
1 mag) redder colours than the SiO targets. This can not be accounted
for by interstellar reddening, which is only 0.1-0.2 mag for
mag. The stars in our sample brighter than K
< 6.5mag are bluer than those with K
> 6.5 mag. They are likely to be
mostly foreground stars with thinner shell, but still IRAS detectable
due to their proximity.
#177 (IRAS 18039-2052) is the only source with [12]-
[25] > 1.0,
which resembles those colours of an embedded young stellar object.
However, methanol and water maser emission was unsuccessfully searched
for toward this source
(Molinari et al. 1996; Palla et al. 1991; MacLeod et al. 1998). Furthermore, the detection of
SiO maser emission confirms that the source is a late-type star, since
SiO maser emission is extremely rare in star-forming regions, with
only three (extremely luminous) sources detected to date
(e.g. Snyder & Buhl 1974; Ukita et al. 1987; Engels & Heske 1989). Though both the 12 m
and 25
m IRAS flux densities have good quality, their association
is unreliable. In fact, the corresponding MSX source has measurements
in the A, C and D bands consistent with the 12
m IRAS
detection, but is not detected at 21
m.
Sevenster (2002) analysed the mid-infrared properties of her OH/IR sample using IRAS and MSX data. Studying a possible correspondence between regions in the two-colour IRAS diagram and the MSX A-Cversus D-E plane she suggested that the MSX diagram can distinguish between the AGB and the post-AGB phases. The transition from a blue (<1.8) to red (>1.8) A-C colour may correspond to a transition off the AGB to proto-planetary nebulae: the star had its last thermal pulse, and ceased to be variable. The transition from a blue (<1.5) to a red (>1.5) D-E colour indicates a later evolutionary transition, when mass-loss starts to drop down of several order of magnitudes and there is the onset of the fast wind. Most of our SiO targets with a clear MSX counterpart show A-C<1.8and D-E<1.5, as expected for AGB stars, and comparable to the bulk of Sevenster's OH/IR sample.
According to Sevenster' criteria, only two objects, #76 and #99,
which are both SiO maser emitters, are likely to be post-AGB stars.
The odd colour of #99 is due to an extremely high flux density of 39 Jy measured in the C band, while the flux density in both the Aand D bands is only 1.7 Jy. However, the inspection of the
MSX C band image does not confirm the presence of a such bright
object and we conclude that the C photometry (despite its good flag)
of #99 is unreliable. Source #76, LPV 12-352 (Glass et al. 2001), is
located on a region of extended emission which is associated with a
star forming region (Schuller 2002); this could have affected the
mid-infrared colours of #76. In conclusion: the anomalous position
of the sources #99 and #76 is probably due to the assignment of
incorrect magnitudes.
The A-C versus D-E colour diagram is useful to locate post-AGB
stars, which have redder colours due to their colder envelopes.
However, this diagram can hardly distinguish between different
thickness of the envelopes of AGB stars. Using a sample of IRAS
sources with IRAS low resolution spectra (Kwok et al. 1997),
Lumsden et al. (2002) showed that a different combination of MSX filters
can distinguish between circumstellar envelopes with silicate feature
at 9.7 m in emission and in absorption. Miras and OH/IR stars
with the silicate feature in emission are located below the black body
line in the C-D versus A-E diagram, while OH/IR stars with
silicate feature in absorption lie above this line. In optically
thick envelopes, self-absorption causes a decrease of the flux in the C and A bands, leading to an increase in C-D. Figure 16 shows that our SiO targets are distributed like
objects with the silicate feature in emission. We note that the MSX
two colour plot is not corrected for reddening. However, the
reddening correction makes the sources bluer in the C-D colour
independently of the adopted extinction law. Furthermore, in the
C-D versus A-E diagram the distribution of the ISOGAL and MSX
samples are similar. For comparison, in Fig. 17 the
MSX colour-colour plots of the Sevenster's OH/IR stars is shown, which
are distributed over a wider and redder range of colours.
![]() |
Figure 16:
MSX colour-colour plot for our SiO maser
targets, similar to Fig. 5 of Lumsden et al. (2002). Stars represent AGB stars, open circles early post-AGB stars, following the classification
of Sevenster (2002). Black body spectra follow the continuous
line. A reddening vector for ![]() ![]() |
Open with DEXTER |
![]() |
Figure 17: MSX colour-colour plot as in Fig. 5 of Lumsden et al. (2002) for the OH/IR sample of Sevenster (2002). Symbols are as in Fig. 16, plus crosses which represent late post-AGB stars, following the classification of Sevenster (2002). |
Open with DEXTER |
As described in Paper I, we initially selected the SiO
targets from the ISOGAL and MSX catalogues on the basis of their near-
and mid-infrared colours, and their 15 m magnitudes. We tried to
select objects with colours typical of pulsating AGB stars with thin
envelopes (Mira-like stars), while avoiding OH/IR stars and other
sources with thick circumstellar envelopes.
Our analysis of the targeted stars' multi-band photometry showed that
these selection criteria were quite reliable. A comparison between
the DENIS and 2MASS data shows that most of them are variable stars,
and moreover the correlation between the J and K band brightness
variations is similar to that found in local dust-enshrouded Mira
variable stars (Olivier et al. 2001).
The IRAS [12]-[25] colours of the SiO targets confirms that they
populate mostly the region IIIa of the van der Veen and Habing
classical two-colour IRAS diagram, which is a region of stars with
moderate mass-loss rates and with silicate feature at 9.7 m in
emission. The distribution of the [12]-[25] colours of the SiO
targets overlaps with the distribution of the colours of OH/IR stars,
which however are distributed over a larger and redder range of
colours (mostly in region IIIa and IIIb). Following the work of
Sevenster (2002) and Lumsden et al. (2002), those properties can be
translated and seen in the MSX C-D vs. A-E diagram. The SiO
targets have a narrower C-D colour range and are located below the
black-body line, differently from the thick-envelope OH/IR stars.
The two subsamples, the MSX-selected objects and the ISOGAL selected objects have very similar infrared properties but differ slightly in the average apparent magnitude, the MSX sample being on average a little brighter. This difference is, however, smaller than the spread in magnitudes of each subsample. In forthcoming papers we usually will combine the two subsamples into one.
Acknowledgements
We are grateful to G. Simon for providing the DENIS data and to M. Sevenster for her constructive criticism. We thank F. Bertoldi, M. Johnston-Hollitt and F. Schuller for their careful reading and commenting of an earlier version of the manuscript.
We acknowledge using the cross-correlation package CataPack developed by P. Montegriffo at the Bologna Observatory.
The DENIS project is supported, in France by the Institut National des Sciences de l'Univers, the Education Ministry and the Centre National de la Recherche Scientifique, in Germany by the State of Baden-Würtemberg, in Spain by the DGICYT, in Italy by the Consiglio Nazionale delle Ricerche, in Austria by the Fonds zur Förderung der wissenschaftlichen Forschung and the Bundesministerium für Wissenschaft und Forschung. The IRAS data base server of the Space Research Organisation of The Netherlands (SRON). This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. This research made use of data products from the Midcourse Space Experiment, the processing of which was funded by the Ballistic Missile Defence Organization with additional support from the NASA office of Space Science. This research has made use of the SIMBAD data base, operated at CDS, Strasbourg, France. The work of MM is funded by The Netherlands Research School for Astronomy (NOVA) through a netwerk 2, Ph.D. stipend.