A&A 441, 1099-1109 (2005)
DOI: 10.1051/0004-6361:20042485
R. Tylenda1,3 - N. Soker2 - R. Szczerba1
1 - Department for Astrophysics, N. Copernicus Astronomical Center,
Rabianska 8, 87-100 Torun, Poland
2 -
Department of Physics, Technion-Israel Institute of
Technology, 32000 Haifa, Israel
3 -
Centre for Astronomy, N. Copernicus University, 87-100 Torun, Poland
Received 6 December 2004 / Accepted 9 June 2005
Abstract
We summarize and analyze the available observational
data on the progenitor and the environment of V838 Mon.
From the available photometric data for the progenitor of
V838 Mon we exclude the possibility that the object before eruption
was an evolved red giant star (AGB or RGB star).
We find that
most likely it was a main sequence or pre-main sequence star of
.
From the light echo structure and evolution we conclude that the reflecting dust
is of interstellar nature rather than blown by V838 Mon in the past.
We discuss the IRAS and CO data for interstellar medium observed near
the position of V838 Mon. Several interstellar molecular regions have
radial velocities similar to that of V838 Mon, so dust seen in the light echo
might be related to one of them.
Key words: stars: early-type - stars: binaries: close - stars: circumstellar matter - stars: individual: V838 Mon - ISM: reflection nebulae - ISM: structure
V838 Mon is a star caught in eruption at the beginning of
January 2002 (Brown 2002).
The eruption, as observed in optical wavelengths,
lasted about three months, and was composed
of two or three major peaks.
After developing an A-F supergiant spectrum
at the optical maximum at the beginning of February,
the object showed a general tendency to evolve to lower effective
temperatures. In April 2002 it almost disappeared from the optical but remained
very bright in infrared becoming one of the coolest M-type supergiants yet
observed. Detailed descriptions of the spectral and
photometric evolution of V838 Mon can be found in a number of papers
including Munari et al. (2002b), Kimeswenger et al. (2002),
Kolev et al. (2002),
Osiwaa et al. (2003), Wisniewski et al. (2003), Crause
et al. (2003) and Kipper et al. (2004).
The nature of the V838 Mon eruption is enigmatic. As discussed by Soker & Tylenda (2003), thermonuclear models (classical nova, He-shell flash) seem to be unable to explain this type of eruption. Therefore other mechanisms such as a stellar merger model (Soker & Tylenda 2003) or a giant swallowing planets scenario (Retter & Marom 2003) have been proposed.
The global fading of V838 Mon in optical after outburst has enabled us to discover a faint hot continuum in short wavelengths (Desidera & Munari 2002; Wagner & Starrfield 2002) later classified as coming from a normal B3 V star (Munari et al. 2002a). This strongly suggests that V838 Mon is a binary system which can be an important fact for identifying the outburst mechanism.
V838 Mon has received significant publicity due to its light echo, which was discovered shortly after the main eruption in February 2002 (Henden et al. 2002), and was seen in images by the HST (Bond et al. 2003). The light echo was used, e.g. in Bond et al. (2003, 2004), to claim that the echoing matter was ejected by V838 Mon in previous eruptions. This conclusion was disputed by Tylenda (2004), who examined the evolution of the light echo and concluded that the dust illuminated by the light echo was of interstellar origin rather than produced by mass loss from V838 Mon in the past.
van Loon et al. (2004) argue that there are multiple shells around V838 Mon, which were ejected by V838 Mon in previous eruptions. Hence they reason that prior to eruption V838 Mon was an asymptotic giant branch (AGB) star. These authors have also analyzed the light echo with more recent observations than in Tylenda (2004), and argue that the echoing dust was ejected by V838 Mon in past eruptions.
In the present paper we collect and discuss the data available on the progenitor of V838 Mon. This includes the archival photometric measurements done in the optical and infrared before 2002, results of analysis of the evolution of the light echo after the eruption, as well as available data on regions of interstellar matter (ISM) near the position of V838 Mon. In the case of erupting stars, conclusions drawn from the progenitor usually are very important for constraining the mechanism of the eruption. An analysis of the observational data for V838 Mon during and after its eruption is done in another paper (Tylenda 2005).
Table 1 lists the photometric results for V838 Mon prior to its outburst.
Columns 1 and 2 give the names and the effective wavelengths of the
photometric bands. The magnitudes and the error estimates in Col. 3 are
from the references given in the last column.
Optical magnitudes have been taken from Kimeswenger et al.
(2002) and Goranskij et al. (2004).
Munari et al. (2002b, 2005)
have also estimated magnitudes of the V838 Mon progenitor. However, the results
given in these two references differ by 1 mag. We do not take them into account
as it is not clear
what caused such large differences (Munari et al. 2005 do not comment
on this). However, if one takes mean values from
Munari et al. (2002b, 2005) they do not significantly differ from
those quoted in Table 1.
The object has also been observed in infrared surveys.
magnitudes
can be found in the 2MASS data while the DENIS experiment measured
the
bands.
Table 1: Photometry of the V838 Mon progenitor.
Note that different measurements have been based on observations taken at different epochs. However, the fairly constant B magnitude obtained in Goranskij et al. (2004) between 1928-1994 shows that the progenitor of V838 Mon was not significantly variable.
As can be seen from Table 1, for four photometric bands we have two
independent measurements. In the case of the B and J magnitudes the agreement
is good. The values in the I and K bands are discrepant by
0.2 mag. As it is difficult to judge which result is more reliable,
for futher analysis we have adopted mean values in the bands for which two
measurements have been available.
An analysis of the progenitor has to take into account the B-type companion
discovered by Munari et al. (2002a). It accounts for about half the brightness
of the progenitor. It seems most reasonable to assume that V838 Mon
and its B-type companion form a binary system.
The main argument obviously comes from the observed positions.
From the instrumental crosses of stars seen on the HST images
taken in September-December 2002
(http://hubblesite.org/newscenter/archive/2003/10/,
see also Bond et al. 2003) one can
deduce that the central object in the B images (dominated by the B-type
companion) very well coincides with the central object in the I image
(dominated by V838 Mon itself) and that both stars cannot be separated by
more than
.
In the HST field (
)
there are
10 field stars of similar brightness as V838 Mon
before outburst and the B-type companion.
In this case the probability that due to a random coincidence one of
these stars is separated by
from
V838 Mon is
.
Next, as discussed below, the binary hypothesis leads to
a consistent interpretation of the observational data of the progenitor.
Observational determination of the distance and reddening
to V838 Mon itself and its B-type companion,
summarized and discussed in Tylenda (2005), give consistent results,
in the sense that
there is no significant difference in the results for both objects.
Therefore in most of our discussion we assume that V838 Mon and its B-type
companion are at the same distance and suffer from the same interstellar
extinction. In some cases, however, we relax this assumption and discuss
the consequences of that.
![]() |
Figure 1:
Spectrophotometry of the progenitor of V838 Mon.
Full symbols - observed magnitudes from Table 1.
Part a) ( left panel): ![]() ![]() |
Open with DEXTER |
As mentioned above, from spectroscopy Munari et al. (2002a) have identified the hot companion of V838 Mon as a typical B3 V star. Indeed their photometric results obtained in September and October 2002, i.e. V =16.05, B-V = 0.68 and U-B = -0.06 (Munari et al. 2005), can be well reconciled with the standard B3 V colours (see Schmidt-Kaler 1982) provided that the object is reddened with EB-V = 0.90 (see open symbols and crosses in Fig. 1a).
The brightness of the B-type companion can also be deduced from the photometric
results of Crause et al. (2005). Between September 2002 and January 2003
(AJD 529-668 in Table 2 of Crause et al.) V838 Mon was
practically constant in
while steadily brightening in R and I. This
behaviour of the object was also noted by
Munari et al. (2002a, 2005) and indicates, in
accord with their spectroscopic results, that the
magnitudes were dominated
by the B-type companion during this time period. From the most reliable results of
Crause et al., i.e. those for dates not marked with an asterisk in their Table 2, obtained
between AJD 529-668 one derives (mean value
standard deviation)
,
,
and
.
Thus the object
is by
0.2 mag. fainter in V than in Munari et al., while the B-V value
if interpreted with the B3 V standard gives
EB-V = 0.72. In this case however
the object seems to be not blue enough in U-B. A better agreement with
the above results derived from Crause et al. is obtained for a B4 V standard
and
EB-V = 0.71 (see open symbols and crosses
in Fig. 1b).
From the beginning of October 2002 V838 Mon has also been
measured by Goranskij et al. (2004).
From their results obtained in October-December 2002 one
derives
and
(no U measurements have been
done during this time period). Thus the V magnitude is in between the values of
Munari et al. and Crause et al., while the B-V value is closer to that of
Munari et al. and implies
.
A large scatter in the B measurements of Goranskij et al.
should however be noted.
Later on in this section we consider two cases depending on whether the photometric data for the B-type companion are adopted from Munari et al. (2005) or from Crause et al. (2005). The differences in the magnitudes between these two references are extreme (the data from Goranskij et al. 2004 are in between them) so these two cases allow us to see how the results of our analysis depend on uncertainties in the photometry of the B-type companion.
Figure 1 presents our interpretation of the available photometric data done assuming that both V838 Mon and the B-type companion have the same reddening. In the discussion we also assume that both components are at the same distance, namely that they form a binary system. In both parts of the figure full symbols display the observed magnitudes from Table 1. The best fits, shown with full curves, have been made using the least square method and the intrinsic photometric colours for the main sequence stars taken from Schmidt-Kaler (1982), Johnson (1966), Koornneef (1983) and Bessell & Brett (1988) (for more details on the fitting procedure see Tylenda 2005).
In part (a) open symbols show the
photometry of
the B-type companion taken from Munari et al. (2005) fitted with
a standard B3 V star shown with crosses.
The full curve presents the best fit to the full points obtained with
a standard B1.5 V star
added to the B3 V companion. Both spectral components have been reddened with
EB-V = 0.9. The ratio of the luminosity of the B1.5 component to that of B3
is 1.9. This ratio is somewhat too low for the B1.5 V and B3 V stars
but given the uncertainties in the observational data we can conclude
from Fig. 1a as follows. The progenitor of V838 Mon was a binary
system consisting of two early B main sequence stars. V838 Mon itself was
probably somewhat brighter, hotter and more massive than its companion.
The system is young, i.e.
yr (main sequence lifetime
of a 9
star, typical for B2 V). Note that it is excluded that
V838 Mon was an evolved B1.5 star as then it would have been significantly more
luminous than the B3 main sequence companion.
Figure 1b adopts the parameters of the B-type companion derived
from the photometry of Crause et al. (2005), i.e. a B4 V star reddened
with
EB-V = 0.71. In this case in order to reproduce the observational
data for the progenitor an A0.5 V standard star has been added to the B4 V companion.
The luminosity ratio of the A component to the B one is 0.43. This is much larger
than the ratio for the main sequence of the same types which, according to
Schmidt-Kaler (1982), is 0.02
.
Also the possibility that the A component was an evolved star, e.g.
a giant evolving towards the red giant branch (RGB), the AGB, or a post-AGB
star, can be ruled out.
In this case it would be expected to have been initially (while being on
the main sequence) more massive and thus, at present, significantly more luminous
than the B4 main sequence companion. Therefore the only possibility within
the binary hypothesis is that the A0.5 star is in the pre-main-sequence phase.
The system would thus be very young.
Judging from the luminosity of the A-type component,
if
1300
is assumed for the B4 V component,
its mass would be
and the age of the system
would be of
yr (Iben 1965).
In summary, although the uncertainties in the observational data for the
progenitor and for the B-type companion do not allow us to unambiguously
identify the nature of V838 Mon the above discussion allows us to put rather
narrow constrains if the most probable hypothesis of binarity is adopted.
In this case V838 Mon is a system
consisting of two intermediate mass stars.
V838 Mon itself certainly was not an evolved star, e.g. RGB, AGB, post-AGB.
It is either slightly more massive than its B-type companion, i.e. 8-10 ,
and was on the main sequence
prior to eruption, or is somewhat less massive,
,
being in the
pre-main-sequence phase. The system is young, with the age estimated between
and
yr.
The abundances in V838 Mon obtained by Kipper et al. (2004) are
reminiscent of those in the so-called HAEBE stars (e.g. Acke & Waelkens 2004)
which are believed to be more massive analogues of the T Tauri stars.
Therefore it is likely that the V838 Mon system is
still partly embedded
in the interstellar complex from which it has been formed. Indeed,
as discussed in Sect. 3.3, near the position of V838 Mon there are
several star-forming regions with radial velocities close to that of V838 Mon
and its B-type companion.
This also fits well
the conclusion of Tylenda (2004) and Sect. 3.1 that the circumstellar
dust producing the light echo of V838 Mon is most probably of interstellar
origin.
Munari et al. (2005) have made an analysis of
the photometric data for the V838 Mon progenitor similar to ours.
Their conclusion is qualitatively similar to ours in the sense that
the progenitor was an early-type star.
However, contrary to our main-sequence or pre-main-sequence hypothesis,
Munari et al. conclude
that the V838 Mon outburst was that of an evolved star of initial mass
of
,
at present in a region occupied by Wolf-Rayet stars in the
HR diagram and having
K.
However, in a case like this
we should see a bright HII region surrounding V838 Mon. The observed light
echo (discussed in Sect. 3.1) shows that there is a lot of
diffuse matter extending from
pc up to at least
pc.
The 50 000 K star of Munari et al. (2005), assuming
EB-V = 0.9 and
a distance of 8 kpc, would have a luminosity of
.
Using model results of Stasinska (1990), for abundances depleted
by a factor of 2 relative to standard values (V838 Mon lies at the outskirts of the
Galactic disc), we can estimate that a star like
this would be able to ionize the surrounding matter up to
pc
if its density is
H atoms cm-3 (
scales as
). The emission line spectrum would
be dominated by [OIII] and Balmer lines
([OIII]
Å/H
), while
the H
luminosity would be
.
For an observer (at 8 kpc
and
EB-V = 0.9) it would look like a nebula with a diameter of
and an H
flux of
erg s-1 cm-2. The resultant H
surface
brightness of
erg s-1 cm-2 arcsec-2is typical for many extended planetary nebulae, e.g. those in the Abell
(1966) catalogue (observed H
fluxes and nebular diameters
can be found in Acker et al. 1992).
Thus the nebula would be rather easy to discern observationally,
especially that in H
and [OIII]
Å it would
times brighter than in H
.
Yet no emission-line nebula have been discovered around the position
of V838 Mon (Orio et al. 2002; Munari et al. 2002b).
Thus the idea of Munari et al. (2005) that V838 Mon prior outburst
could have been as hot as 50 000 K is not consistent with the observations.
From the observed lack of any significant emission nebula around V838 Mon
we can conclude that before the outburst the star was cooler than
30 000 K,
i.e. of a spectral type not earlier than B0. Munari
et al. (2005) also consider that the progenitor
could have had
K (although they argue that this
is not likely). This solution is practically the same as our case
of a B1.5 star in Fig. 1a
which we interpret as an early B-type main sequence star.
As discussed above it is evident that V838 Mon was not a typical red giant
nor an AGB star prior to eruption
if V838 Mon and its B-type companion form a binary system.
The only way to reconcile the RGB or AGB hypothesis is to assume that
the B-type companion has nothing to do with V838 Mon and that the coincidence of
the two objects in the sky is purely accidental. Then one may assume that V838 Mon
is less reddened than the B-type companion and a cooler star can be fitted to
the observations. Let us consider that the B-type companion has the parameters
derived from the observations of Crause et al. (2005), i.e. B4 V reddened
with
EB-V = 0.71, as then the fits give later spectral types for V838 Mon
than if the results of Munari et al. (2002a) were adopted.
Let us also use the standard supergiant spectra (intrinsic colours taken from the
same references as the main sequence ones) to model the contribution
from V838 Mon. This is more relevant with the RGB/AGB hypothesis and also results in
later spectral types from the fits than the main sequence spectra.
EB-V = 0.5 seems to be a lower limit for the extinction towards V838 Mon
(see discussion of different observational determination in Tylenda 2005).
Assuming this value, the best fit to the observations is obtained for the spectral
type F1 (effective temperature 7500 K). If, in spite of observational
determination, the extinction is pushed to its limit, i.e.
EB-V = 0.0 is
assumed, the fit gives G7 (effective temperature
4700 K). Thus there is
no way to reconcile an M-type star with the observational data.
This conclusion is obvious if one realises
that the B-R colour for an unreddened M-type star is
2.7 while that of
the V838 Mon progenitor was
1.3.
From the above results we can firmly conclude that V838 Mon was not an
AGB star. If one still does not want to leave the AGB hypothesis and argues that it
best explains the existence of the circumstellar matter seen in the echo and infrared
images, then the only way to reconcile it with the photometric data is to say that
V838 Mon had quite recently left the AGB and prior to eruption was in the post-AGB phase.
However in this case its luminosity would be close to
and its
distance would have to be
55 kpc in the case of the F1 type and
EB-V = 0.5 or
even
90 kpc if one prefers G7 and
EB-V = 0.0. Thus a typical spectral types
of the post-AGB stars (F-G) would require rather unacceptable conditions, i.e.
a very low reddening and
a very large distance putting V838 Mon in the extreme outskirts of the Galaxy.
For the observationally acceptable range of the extinction,
i.e. EB-V between 0.7 and 1.0 (Tylenda 2005), V838 Mon would have been
of the B9-B0 type and, assuming the typical post-AGB luminosity as above, its distance
would be between 7 and 30 kpc. Thus the only possibility within the AGB
- post-AGB hypothesis, which is not excluded by
the photometric data, is that V838 Mon was a B-type post-AGB star.
However, as discussed above, the binarity with the B-type main
sequence component is, in this case, excluded in spite of similar estimated distance
ranges (7.5-12.5 kpc for the B-type companion, see Tylenda 2005), close
values of interstellar extinction and the same positions in the sky of both objects.
If one adds that the duration of the phase when the post-AGB star can be classified
as B-type is typically 103 years it is clear that this solution is extremely
improbable.
Finally let us discuss the giant hypothesis. As discussed above it is certain that
V838 Mon was not a typical RGB star, i.e. of K-M spectral type, prior to eruption.
The latest acceptable spectral type, obtained assuming the lower limit of
EB-V = 0.5, is F0-A5. At the lower limit of the distance of 5 kpc
(Tylenda 2005) the star would have a luminosity of 40-50 which is more or less consistent with the standard giant luminosities for
these spectral types (Schmidt-Kaler 1982). For the more probable reddening,
i.e.
EB-V = 0.7 - 1.0, we have to move to the B types and correspondingly larger
luminosities and distances. Thus the hypothesis that V838 Mon was a giant before
its eruption is not excluded provided that it was an early (A-B) type giant.
The binarity with the B-type companion is excluded in this case (as discussed above).
The star would be quite massive,
,
not far evolved
from the main sequence and thus be in a fast evolutionary phase (time scale
years). The object would be quite rare in the stellar
population although not as rare as the B-type post-AGB case considered above.
The discussed case would however have difficulties in explaining the origin
of the circumstellar matter seen in the light echo. The wind from an early type
giant would not be enough. On the other hand, the giant, being significantly
older (most probably at least as old as 108 years) than the B-type binary system
considered above, would have little chance to still reside in a dense interstellar
cloud.
The phenomenon of light echo observed in V838 Mon during and after the outburst suggests that there is much dusty matter in the vicinity of the object. Since the light echo works as a sort of scanner, an analysis of the light echo images in different epochs should provide detailed information on the dust distribution near the object which would be important for constraining the nature of the object. Unfortunately, in spite of numerous images obtained at different observatories (including HST) no elaborate study of the light echo has been done as yet. So far the most detailed, but still very simple, analysis has been done by Tylenda (2004) on five echo images obtained with HST between 30 April and 17 December 2002. His conclusion is that the dust distribution does not show any signs of spherical symmetry and that dust is likely to be of interstellar origin rather than due to past mass loss from V838 Mon.
We have extended the analysis of Tylenda (2004)
using two recent echo
images obtained on 21 Oct. 2003 at the USNO
(http://www.ghg.net/akelly/v838lar3.jpg) and on 8 Feb. 2004 with the HST
(http://hubblesite.org/newscenter/newsdesk/archive/releases/2004/10/). On these images we have measured the outer rim of the light echo.
Then a least square fit of a circle to
the measurements has been done in the same way as in
Tylenda (2004). The results of the fits are given in the last two lines
of Table 2. The first five lines in the table repeat the results
from Table 1 of Tylenda (2004) as the uncertainties there have been
slightly overestimated. The first column of Table 2
shows the time of observations,
,
given in days since
1 January 2002.
The radius of the echo,
,
and its uncertainty
are given in the second column. The next
two columns show the (x,y) position of the centre of the fitted circle relative
to the central star. Note that x points to west while y is to north.
The last column gives the angular distance of the echo centre
from the central star.
All the results are in arcsec.
Following Tylenda (2004) we adopt in our analysis that the zero age
of the echo is t0 = 34 days (since 1 Jan. 2002).
Table 2:
Results of fitting a circle to the outer edge of the light echo
of V838 Mon. Time of observations,
,
is in days
since 1 January 2002. Results are in arcsec.
![]() |
Figure 2:
Left panel:
the best fit of a plane model to the observed evolution
of the light echo radius. Symbols with error bars -
the values and uncertainties of ![]() |
Open with DEXTER |
Filled symbols in the left panel of Fig. 2 display
the evolution of the echo radius,
,
with time. The full curve shows the best fit to the data
of a plane slab model,
i.e. of Eq. (17) in Tylenda (2004), obtained
for
kpc and
pc, where d is the distance
between the light source and the observer while z0 is the distance of
the dust slab from the source.
However, similarly to
Tylenda (2004), the
minimum of the fit is
quite shallow and extended along
.
The right panel of
Fig. 2 shows the 95% confidence region of the fit.
From this figure one can conclude that
the distance to V838 Mon is
kpc. Recently, Crause et al. (2005)
have analysed the light echo evolution from their observations done
at the SAAO. Their results obtained for the sheet model are well within the
hatched region in the right panel of Fig. 2.
![]() |
Figure 3:
Migration of the echo centre from the central star.
Left panel: (x, y) positions of the echo centre
from Cols. 3 and 4 of Table 2. The size of
the symbols is proportional to the echo radius given in
Col. 2 of Table 2. Note that
the x and y axes point to west and north, respectively.
The central star is at (x=0, y=0).
Right panel: evolution of the distance of the
echo centre from the central star,
![]() |
Open with DEXTER |
As can be seen from Table 2, the centre of the light echo
has been migrating from the central object. This migration, displayed
in Fig. 3, has kept the same pattern for 2 years.
The following two conclusions can be drawn from
Fig. 3.
First, since the appearance of the light echo
its center has been moving away from the
central object in roughly the same (north-east) direction,
as can be seen from the left panel
of Fig. 3. Second, the distance of
the echo centre from the central star has been increasing linearly with time,
as shown in the right panel of Fig. 3.
As discussed in Tylenda (2004), the fact that the light echo has an outer edge means that the dusty medium producing the echo has a boundary in front of the central star. However, the fact that the echo edge is not centered on the star and that the distance of the echo centre from the star does increase with time shows that this dust boundary is not spherically symmetric with respect to the central object.
Following the theory of the light echo, as e.g. summarized in Tylenda (2004),
the only reasonable interpretation of the observed evolution of the
outer edge of the light echo is that the dust boundary in
front of V838 Mon is more or less in the form of a plane inclined to the line
of sight.
A linear fit to the observed evolution of the distance of the echo centre
from the central object, i.e. last column in Table 2 or
symbols in the right panel of Fig. 3,
gives the relation
In April 2002, when V838 Mon faded in the optical, the light echo started developing
an asymmetric hole in the centre. As analyzed in Tylenda (2004)
this clearly shows that there is
a dust-free region around V838 Mon and that this empty region is strongly asymmetric.
The inner edge of the dusty region in the southern directions
is at 0.10-0.15 pc from the central object whereas in the opposite directions
it is at least 10 times further away.
It can be noted that the IRAS PSC source 07015-0346 coincides
with the position of V838 Mon within
a position elipse of
(see also Kimeswenger et al. 2002). The
measured fluxes at 100
m and 60
m are 4.6 Jy and 1.4 Jy, respectively,
while at 25
m and 12
m the catalogue gives only an upper limit of
0.25 Jy. When fitted with a simple dust emission model, i.e. emissivity
proportional to
,
the IRAS fluxes give a dust temperature,
,
of
30 K. For a central source of
(two B3 V stars, see
Sect. 2) this value of dust temperature is reached at
0.2 pc
(see e.g. Eq. (7.56) in Olofsson 2004).
Thus the IRAS fluxes can be consistently interpreted as due to inner parts
of the dusty region inferred from the light echo analysis. In particular,
they give evidence that there is no significant amount of dust at distances
pc, thus confirming the existence of the central dust-free region.
The strongly asymmetric central dust-free region would be very difficult
to understand if the echoing dust were produced by a past mass loss from V838 Mon.
The hole would imply that mass loss stopped a certain time ago, e.g. 104 yr
for the 0.10-0.15 pc inner dust rim if a wind velocity of
is assumed.
However the hole asymmetry
would imply that in the opposite direction either mass loss stopped 10 times earlier
or the wind velocity was 10 times higher. Neither of these two possibilities seems
to be likely.
Instead, as discussed in Tylenda (2004), the asymmetric hole is easy to understand if the echoing dust is of interstellar origin. Then it is natural to suppose that V838 Mon is moving against the ISM. If possessing a fast wind it would create a hole largely asymmetric along the direction of the movement. Indeed, the structure of the inner edge of dust in Fig. 5 of Tylenda (2004) well resembles stellar wind bow shocks investigated in e.g. Van Buren & McCray (1988) and Wilkin (1996). The nearest rim in the southern directions, being at 0.10-0.15 pc from the central star, would correspond to a region where the stellar wind collides head-on with the ambient medium.
Let us assume that a star losing mass at a rate,
,
and a velocity,
,
is moving in the ISM
of number density, n0, with a relative velocity of
.
A swept-up shell is created in the form of a bow shock and in the up-stream direction
this takes place where the wind ram pressure is comparable
to that of the ambient medium (see e.g.
Van Buren & McCray 1988, Wilkin 1996), i.e.
A B3 main sequence star has a luminosity of
(Schmidt-Kaler 1982).
Thus, according to the relation of Howard & Prinja (1989), the expected
mass loss rate would be
.
If V838 Mon is a young binary system, as discussed in Sect. 2, its velocity
relative to the ISM should be rather low. Assuming
between
and the above
mass loss rate, Eq. (4) yields
cm-3if
r0 = 0.10-0.15 pc. This result is uncertain but it indicates that V838 Mon
is imbedded in a relatively dense ISM. In principle, the
density could be estimated from the echo brightness.
Unfortunately there is no such estimate available, although
the presence of the bright echo
suggests that the density of the ambient medium must be significant.
A fast stellar wind colliding with a circumstellar medium
should produce X-rays. V838 Mon was observed with Chandra a year after the outburst
by Orio et al. (2003). The object was not detected and the upper limit
to the X-ray luminosity is
(for a distance of 8 kpc).
The kinetic power of a wind with parameters as above
(
expanding at
)
is
,
thus it is below the X-ray limit, although only by a factor of 2.
However, the X-ray luminosity from a colliding wind is usually much below the
wind kinetic power (e.g. Soker & Kastner 2003). A hot bubble
is usually formed and, if the radiative cooling time is long, the shocked gas cools
adiabatically by expansion of the bubble. In our case the cooling time
is estimated to be above 108 years, i.e. much longer that the life time
of a B3 main sequence star. In the case of a bow-like inner edge
of the circimstellar matter, as discussed above, a hot bubble need not be formed.
Instead the shocked gas may flow along the edge and escape through the open side,
thus expanding, cooling adiabatically and radiating very little X-rays.
van Loon et al. (2004) have also analyzed
the expansion of the light echo.
Their analysis has been based on their measurements of the echo diameter on images
from different sources. However, as also noted in Crause et al. (2005),
their diameters are
systematically smaller by factor of 2.5-2.7 than any other measurements available
in the literature (Munari et al. 2002b; Tylenda 2004;
Crause et al. 2005, see also numerous individual measurements
in IAU Circ. in 2002), including our present results in Table 2.
It is curious that these authors do not note this discrepancy and do not
comment on it.
In any case it can be concluded that the whole analysis of the light echo
made by van Loon et al. is questionable as it has been based on wrong data.
In particular, their rather speculative interpretation of the outer edge of
the echo in October 2003-February 2004 as being produced by scattering under
right angles does not hold. With the correct values of the echo diameter in these
dates this interpretation would imply a distance of 2 kpc (and not 5.5 kpc
as written in van Loon et al.). which is much too low compared with any other
distance estimates from the light echo evolution (Bond et al. 2003;
Tylenda 2004; Crause et al. 2005).
van Loon et al. (2004, hereafter vLERS) from their analysis of
the IRAS and MSX images and
the CO maps claim that V838 Mon is surrounded by three shells. The innermost,
seen from the MSX, is highly irregular and has dimensions of
.
The second one would be the elliptical one referred from the IRAS with dimensions of
.
The largest one, having a diameter of
,
is suggested
from the CO maps. On this basis vLERS conclude that V838 Mon is a low mass
AGB star experiencing thermal pulses. As we have shown in Sect. 2,
the photometric data on the V838 Mon progenitor exclude the AGB hypothesis.
However, even if the latter is not taken into account,
we find severe problems with the results of vLERS and
their interpretation.
First questions arise when analyzing the innermost structure seen
in the band A (8.3 m) of MSX. Unlike the two
outer shells (IRAS and CO) showing an elliptical or circular symmetry
this structure is very distorted and does not show any kind of symmetry
with respect to V838 Mon.
It does not resemble objects involving AGB, like
planetary nebulae, envelopes of AGB and post-AGB stars.
Usually, as in planetary nebulae for example, the ISM
affects external regions so one would expect to see more distortion in
the IRAS and CO shells than in the MSX structure.
The dimensions of the IRAS and CO shells seem to be too large to be
compatible with the hypothesis that they have been produced by an AGB mass
loss. Adopting a distance to V838 Mon of 8 kpc
(Tylenda 2004) the radius of the IRAS shell is 20 pc while
that of the CO shell is
70 pc.
(Munari et al. 2005 argue for 10 kpc as the
most probable distance, which would make radii even
larger thus strengthening our conclusions below.)
The largest observed AGB dust shells
have radii of 2-3 pc (Speck et al. 2000) while the CO shells
are usually well below 1 pc (Olofsson 2004).
We can thus conclude that the IRAS and CO shells, if real, are not typical
AGB shells.
Below we show that
it is unlikely that AGB shells could survive and be observed as fairly
symmetric structures at distances of 20-70 pc from the central star.
As a mass-losing star moves relative to the ISM, the shell's segment
in the up-stream direction (the side facing the ISM) is slowed down,
until it is stopped at a time
,
when the leading edge is at a
distance
from the star.
Next the up-stream segment of the shell is pushed by
the ISM toward the central star.
At the same time the shell's segment in the down-stream direction
is expanding at a constant, undisturbed rate.
Soker et al. (1991) have derived simple analytical expressions for these
parameters which can be used here.
Let a shell of mass, Ms, expand with a velocity
,
and let
be the relative velocity of the mass-losing star and
the ISM.
Also let
be the mass density of the ISM, and n0the total number density of the ISM. For a distance of
150 pc from the galactic plane, we can scale the ISM
density with
,
which
corresponds to
.
Following Soker et al. (1991) we define the radius of a sphere which
contains an ISM mass equal to the shell mass
For an AGB star
.
The typical star-ISM
velocity at
150 pc from the galactic plane
is
.
For
and assuming
,
we get from Eqs. (5), (6) and (8)
pc and
pc.
Even for an extreme case of
,
(which is a generous upper limit for
an AGB shell, especially if it were ejected in a thermal pulse from a low
mass star, as suggested in vLERS)
and
(say, expansion velocity of
and
), we find
pc
and
pc.
Given the observed diameter of the CO shell of
140 pc we can conclude
that an AGB shell would have been seriously disturbed by the ISM before reaching
these dimensions, its up-stream part would have to be significantly brighter
(because of significant accretion of the matter from the ISM)
than the opposite part and the central star would very likely be now observed
outside the up-stream rim. All this is not observed.
Table 3:
Interstellar regions within
from the position of
V838 Mon.
The above estimates are supported by observations of planetary nebulae.
From Eq. (5) we see that when a typical planetary
nebula shell of
reaches a radius of
3 pc,
it is expected to be highly distorted by the ISM.
Indeed, examining the list of planetary nebulae interacting with the ISM
compiled by Tweedy & Kwitter (1996, their Table 3), we find that all
the planetary nebulae in this list have radii <5 pc.
Most of the large planetary nebulae are highly distorted; not is only the central
star not at the center, but the shells are neither
circular nor elliptical.
Some planetary nebulae, like NGC 6826
NGC 2899 and A 58 (surrounding the final helium shell flash star V605 Aql),
have very large, diameters
10-40 pc, IRAS structures around them
(Weinberger & Aryal 2004; Clayton & De Marco 1997).
They are all severely distorted. Clayton & De Marco (1997) argue that
structures of this size are
swept-up ISM dust, rather than AGB mass-loss shells.
The above analysis rises a question: are the shells claimed in vLERS indeed real and related to V838 Mon?
It is difficult
to discuss the nature of the emission seen in the MSX image as it has been
recorded only in the A band (8.3 m) image. No emission is seen in the
bands C (12.1
m), D (14.6
m) and E (21.4
m). This is
perhaps due to the highest sensitivity of band A.
The elliptical structure around V838 Mon in the IRAS image shown in Fig. 1 of vLERS at first sight looks convincing. However, in the whole field of this figure it is easy to fit several ellipses of similar sizes and similar orientations as the one drawn by vLERS. One possible explanation is that the image pattern seen in Fig. 1 of vLERS might be spurious, i.e. of instrumental and/or image processing origin. Another likely interpretation is that this is a general pattern of the interstellar diffuse emission in this region and thus it has nothing to do with V838 Mon. The discussed region is a part of an extended infrared emission related to several molecular clouds and HII regions near the direction to V838 Mon (see Sect. 3.3). Whether or not a part of this emission is physically related to V838 Mon is an important question but it cannot be decided just from the image.
Figure 3 of vLERS has been derived from a compilation of
surveys in the CO 1-0 line made by Dame et al. (2001). However, if one
takes the composite map from Fig. 2 of Dame et al. and expands near the
position of V838 Mon the resultant image is not the same as that in vLERS.
In particular, there is no emission to the left and upper-left of the position
of V838 Mon [i.e. (V838 Mon) and
(V838 Mon)], so no bubble-like
structure is seen. Apparently Dame et al. (2001) considered this emission
as statistically insignificant. Indeed, the data for this region come from
a low resolution (0
)
and low sensitivity survey of Dame et al.
(1987). Thus the emissions seen in Fig. 3 of vLERS and coming
from this last survey is uncertain and might be spurious.
This is supported by the fact that
two large, faint patches seen to the left and upper-left of the V838 Mon bubble
in Fig. 3 of vLERS (not seen in Fig. 2 of Dame et al. 2001)
cannot be identified with any known CO cloud
while two known CO regions, namely regions (9) and (13) in
Table 3 (see Sect. 3.3), are not seen in the image of vLERS.
The presence of the light echo proves that there is dusty matter around
V838 Mon. As argued in Tylenda (2004) and in Sect. 3.1 of the
present paper this matter is likely to be of interstellar character. This notion
is supported by the conclusion of Sect. 2 that V838 Mon is likely to be
a young binary system, as well as
by the fact that having the Galactic coordinates,
,
the object is located near the Galactic plane.
Therefore it is important to investigate observational data on the ISM in the
vicinity of V838 Mon. In this section we discuss the available data from the IRAS
and CO surveys.
![]() |
Figure 4:
The 100 ![]() |
Open with DEXTER |
Figure 4 shows the IRAS image at 100 m centered at the position
of V838 Mon. At this wavelength practically all emission seen comes from
dust (interstellar or circumstellar). As can be seen from the figure
V838 Mon is located in a faint diffuse emission probably related to bright
regions near the Galactic plane.
The numbers in the figure show the positions
of molecular and HII regions listed
in Table 3, which are within
of the position
of V838 Mon. The table
gives the galactic coordinates of the regions, their usual names, values of T*A taken from Wouterloot & Brand (1989),
resulting from
CO line observations, and types of the regions.
The references to the data are given in the last column of the table.
As can be seen from Table 3 the regions marked in Fig. 4
can be divided into two groups from the point of view of their positions, brightness
and
.
The brightest regions near and slightly
below the Galactic plane (regions 5, 6, 8, 12) have
in the range
of 20-30 km s-1.
The heliocentric radial velocity of V838 Mon is not well known as it
has been estimated from
outburst spectra. The results are within 55-65 km s-1(Kolev et al. 2002; Kipper at al.
2004; M. Miko
ajewski - private communication).
That of the B-type companion is
64 km s-1 (T. Tomov - private
communication). Using the
results of Dehnen & Binney (1998) this can be transformed to
km s-1. Thus the above regions
are most probably at much smaller distances (2-3 kpc if interpreted with
the Galactic rotation curve of Brand & Blitz 1993) than V838 Mon and they
are simply seen in front of the object.
However the regions lying above the Galactic plane (regions 1, 2, 3, 4, 7, 9, 13)
have
between 47-57 km s-1. When interpreted with the rotation curve
of Brand & Blitz (1993) their distances are in the range of 6-8 kpc. Thus
these regions are located much closer to V838 Mon and a physical relation between
one of them and the matter seen in the light echo is quite possible.
Region (10), whose apparent position is closest to that of V838 Mon,
has been listed in Magakian (2003) as a reflection nebula related
to a 9 magnitude B9 star HD 53135 (LS -03 15) estimated to be at
a distance of 2 kpc (Vogt 1976, Kaltcheva & Hilditch
2000). Thus this region is probably located well in front of V838 Mon.
Interstellar Na I lines in the spectrum of V838 Mon during eruption showed
two components at heliocentric velocities of 37 and
64 km s-1(Zwitter & Munari 2002; Kolev et al. 2002; Kipper et al.
2004). When transformed to
the figures become
26
and
53 km s-1. Thus both line components can be interpreted as due to ISM
related to the above two groups of the ISM regions.
Detailed observations of the vicinity of V838 Mon in molecular lines might be
important for discussing the nature of V838 Mon.
The goal of this paper is to use available observational data on the progenitor and environment of V838 Mon to better constraint the nature of its eruption. Below we summarize and discuss our main findings and conclusions.
(1) The nature of the progenitor.
In Sect. 2 we have analyzed the photometric data available
for the progenitor. Most likely the progenitor was
a young binary system consisting of two
intermediate mass (5-10 )
stars. V838 Mon itself was either
a main sequence star of similar mass as its B-type companion or a slightly
less massive pre-main-sequence star. The system is very wide as the B-type
companion observed today does not seem to be affected by the eruption.
From the maximum photospheric radius of V838 Mon during eruption
(Tylenda 2005) we can estimate that the separation of the components
is
so the orbital period is >12 years.
The B-type companion was probably not involved in the eruption of
V838 Mon, at least directly. The hypothesis of a young binary system
is also supported by the position of the object near the Galactic plane
and the conclusion of Sect. 3 that V838 Mon is probably embedded in the ISM.
A less likely hypothesis is that the presently observed B-type companion does not form a binary system with V838 Mon. In this case, other than a B-type main sequence star, the progenitor could have been an A-B spectral type giant evolving from the main sequence or a B-type post-AGB star. These two possibilities however involve a short (giant in the Hertzsprung gap) or very short (post-AGB) evolutionary phase. As it is an old object in these cases it would not be expected to reside inside or close to dense ISM regions.
We can safely excluded the possibility that before eruption V838 Mon was of spectral type K-M, so it could not have been a typical RGB or AGB star.
(2) The light echo and the Galactic environment of V838 Mon.
In several studies the light echo was used to argue that
the light-reflecting dust was
expelled by V838 Mon in previous eruptions (e.g. Bond et al. 2003,
vLERS).
As argued by Tylenda (2004), and discussed here in Sect. 3.1,
the data strongly suggests that dust is of ISM origin.
The dust structure derived from the echo analysis does not show any
hint of spherical symmetry. On the contrary,
the outer boundary of the echoing dust in front of the object can be
approximated by a plane at a distance
of 3.5 pc from V838 Mon and inclined at
an angle of
to the line of sight. The strongly asymmetric
dust-free region in the near vicinity of V838 Mon, inferred from
the central hole in the echo, is interpreted as produced by the V838 Mon
progenitor (and possibly its B-type companion) moving relative to
the local ISM and sweeping out the medium by its fast wind.
As discussed in Sect. 3.3, there are several interstellar
molecular regions seen in the IRAS image and CO surveys probably located
near V838 Mon. The local
ISM seen in the light echo is therefore likely to be related to one or
some of them.
Acknowledgements
This has partly been supported from a grant No. 2 P03D 002 25 financed by the Polish State Committee for Scientific Research, as well as by the Israel Science Foundation We thank T. M. Dame for providing his CO data for a field centered on V838 Mon.
Note added in proof. Thomas Dame (private communication) has recently confirmed our discussion in the last paragraph of Sect. 3.2. The emission that van Loon et al. (2004) plotted as the upper-left section of their CO loop comes from the survey of Dame et al. (1987) and is way down in the noise of this survey.