A&A 395, 161-167 (2002)
D. P. K. Banerjee - N. M. Ashok
Physical Research Laboratory, Navrangpura, Ahmedabad 380 009, India
Received 10 June 2002 / Accepted 27 August 2002
Near - IR, multi - epoch, spectroscopic and photometric observations of the enigmatic, eruptive variable V838 Mon in the JHK bands are reported. One of the unusual features of the spectra is the detection of several strong neutral TiI lines in emission in the K band. From the strength of these lines, the mass of the ejected envelope is estimated to be in the range 10-7 to 10 . The spectra also show the strong presence of the first and second overtone 12CO bands seen in the K and H bands. The CO bands show a complex evolution. Deep water bands at 1.4 m and 1.9 m are also seen later in the object's evolution. Blackbody fits to the JHK photometric data show that V838 Mon has evolved to temperatures between 2400-2600 K by 130 days after outburst. The spectra at this stage have the general characteristics of a very cool M giant.
Key words: stars: individual: V838 Mon - stars: supergiants - infrared: stars - stars: novae - techniques: spectroscopic
The eruptive variable V838 Mon was first reported to be in outburst on 6 January 2002 by Brown (2002). A first maximum () in its lightcurve was reached around 11 January. Subsequently there was a slow steady decline followed by a second strong outburst on 2 February which changed the brightness by 4.3 mag to a peak value of V= 6.7. The overall evolution of the light curve has thus been complex. The initial spectra of the object in the optical have shown several emission lines, in general having P Cygni profiles, of BaII, LiI and those of several s process elements (Munari et al. 2002a). However, the most striking development was the detection of an expanding light echo around the object seen most clearly in U band images (Munari et al. 2002a). Being a rare phenomenon, this heightened interest in an already puzzling object, resulting in several IAU circulars on the subject and also warranting special Director's Discretionary Time allocation on the HST (Bond et al. 2002). The light echo seems to be caused by scattered light from dust shells which existed even before the outburst. This is supported by the fact that the progenitor of V838 Mon was detected by IRAS in the 60 and 100 m bands indicating the presence of low temperature dust. In this work, we present results from JHK observations of V838 Mon which have been made at five, fairly evenly-spaced epochs and which should help in following the temporal evolution and understanding the nature of V838 Mon - questions of considerable importance and interest at present.
Near-IR JHK spectra at a resolution of 1000 were obtained at the
Mt. Abu 1.2 m telescope using a Near Infrared Imager/Spectrometer
HgCdTe NICMOS3 array. We present here the spectroscopic
observations of five days viz. 2 February, 25 March, 9 April,
2 May and 14 May 2002. It may be noted that the 2 February spectra were
acquired between 2.729 - 2.762 UT and coincide with the time when the
second outburst of V838 Mon was just underway as indicated by VSNET
reports. VSNET reports, centered around the second outburst, give photometric
values of V = 10.67, 10.708, 8.193 and 8.02 at epochs of February
1.487, 1.858, 2.799 and 2.913 UT from which an idea can be
obtained when the outburst began.
In each of the J,H and K bands a set of at least two spectra were taken
and sometimes as many as ten. In each set the star was offset to two
different position of the slit (slit width = 2 arcsec). The
signal to noise ratio of the spectra, as determined using IRAF, is moderate
and ranges between 30-50 in J, 30-60 in H and 30-80 in the K band.
The exposure times for the spectra presented here are as follows
(given in order of J,H and K): 2 Feb. - 120, 90, 90s; 25 March
- 15, 10, 15s; 9 April - 15, 10, 15s; 2 May - 120, 45, 45s and 14 May
- 120, 60, 60s. Spectral calibration was done using the OH sky lines
that register with the spectra. The comparison stars that were used for
ratioing the spectra, were either HR 2714 or HR 3314 in all cases. The spectra
of the comparison stars were taken at similar air-mass as that of V838 Mon
and the ratioing process thereby removes the telluric lines.
|Figure 1: The J band spectra of V838 Mon are shown at different epochs. The spectra have been offset from each other for clarity. The comparison spectra of HR 4517 is from Wallace et al. (2000).|
|Open with DEXTER|
|Figure 2: The H band spectra of V838 Mon are shown at different epochs. The spectra have been offset from each other for clarity. The comparison spectra of HR 4517 is from Meyer et al. (1998) while that of WW Ser is from Lancon & Roca-Volmerange (1992). The marked features are discussed in the text.|
|Open with DEXTER|
|Figure 3: The K band spectra of V838 Mon are shown at different epochs. The spectra have been offset from each other for clarity. The comparison spectra (see text) of HR 4517 is from Wallace & Hinkle (1997) while that of WW Ser is from Lancon & Roca-Volmerange (1992). The 12CO bands and also the TiI lines around 2.2 m can be seen.|
|Open with DEXTER|
The JHK spectra are shown in Figs. 1-3. We first discuss the spectral features that are observed and later discuss the shape of the continuum in the JHK bands. We have compared our spectra, with the aim of classification and line identification, with the spectral classification catalogs of JHK spectra compiled by Meyer et al. (1998), Wallace & Hinkle (1997), Lancon & Roca-Volmerange (1992), Wallace et al. (2000) and Forster Schreiber (2000). The spectra from some of these last-mentioned references are available in electronic form from the ADC & CDS data centers and also from the site http://www.noao.edu/archives.html. To guide the reader in interpretation/comparison of our observed data, we have added two reference spectra to Figs. 1-3 - one at the bottom and the other at the top. The selected comparison spectra of HR 4517 (spectral type M1 III) and WW Ser (M8 III) are close to the spectral type of V838 Mon as on 2 February and 14 May. However it was not possible to add a comparison J band spectra for 14 May because V838 Mon had cooled to spectral class M8 or 9 by then and similar comparison spectra were not available. It must be pointed out that the spectra of HR 4517 in Figs. 1-3 have been smoothed by a seven point moving average to degrade them somewhat from a higher resolution (R = 3000) to a comparable resolution as our observations. Further the slope of the J and H band spectra of HR 4517 have been tilted to match those of our spectra because in the original data the spectra had been normalized to be flat. The slope correction does not alter the spectral features at all and the moving-average smoothing (instead of a convolution with the instrument function) should adequately serve the primary purpose of enabling a comparison between spectral features. The spectra for WW Ser, which are from Lancon & Roca-Volmerange (1992) are at a resolution of 500.
In the J band, the only feature which appeared weakly but persistently in emission is Paschen beta at 1.2818 m. We give the equivalent widths for this line in case they can be used in conjunction with other hydrogen line data (obtained in other studies) to study some properties of V838 Mon (for e.g. interstellar extinction towards it). The measured equivalent widths are 2.7, 5.0, 5.9, 8.3 and 6.9 Å on 2 February, 25 March, 9 April, 2 May and 14 May respectively. The errors in measurement typically lie between 10 to 20 percent of the measured values. The H band spectra show a multitude of spectral features which are rather characteristic of the spectrum of cool M type giant stars. Some of the principal spectral features listed by Meyer et al. (1998) which characterize the H band spectra of stars are also seen here. Among these are the second overtone 12CO ( ) bands whose positions are marked in Fig. 2. Although weak, the v = 3-0 (1.558 m), 4-1 (1.578 m), 5-2 (1.598 m) and 6-3 (1.619 m) bands are consistenly seen. The positions of the higher vibrational transitions are also marked in the diagram but they are not clearly present in the data. Three other features (marked 1, 2, and 3 in Fig. 2) are also persistent. Feature 1 at 1.5711 m could be the HI Brackett 15 line while feature 2 at 1.5892 m is attributable to SiI (Forster Schreiber 2000). Feature 3 is seen in emission at all phases and although it coincides well with the position of MgI 1.7113 m it is expected to be in absorption had it been due to MgI (Meyer et al. 1998). The spectra in the H band, in general, resemble those of later M type giant stars, but not completely. However it must be pointed out that the spectra of V838 Mon is a superposition of photospheric absorption lines and also emission lines from the ejecta and their coaddition can cause deviations from standard stellar spectra.
The K band spectra have two striking features. The first of these are
several emission lines seen around 2.2
m which were first visible
on 9 April, peaked in strength on 2 May and again weakened by 14 May. A magnified section around these lines is shown in Fig. 4. The
preliminary indication that they could be due to TiI was their presence
in absorption, at similar wavelengths, in the high resolution spectra
of Arcturus (Hinkle et al. 1995). We believe
these lines are due to neutral titanium (Banerjee & Ashok 2002) after
having matched the observed wavelengths of these lines with the
laboratory spectrum of TiI as given by Forsberg (1991).
It is quite remarkable that all the
significantly strong lines listed by Forsberg (1991) in the K band appear
in our spectrum. The laboratory wavelengths of these lines are marked
in Fig. 4. As can be seen there is a good match. The intensities of
the lines also match fairly well the intensities given by Forsberg (1991).
We therefore feel the identification is secure. The detection of TiI
emission lines in the K band is certainly rare in eruptive variables
or even in
astronomical objects - in fact it's unclear whether these lines have
ever been seen before and their presence adds to the mystery of V838 Mon.
Being in emission, they cannot be of photospheric
origin (not in a cool object) and must therefore arise from the circumstellar
ejecta. Except for the 2.2627
m line which arises from a transition
between the a3G4-z3F4 levels of the
configuration, all the other lines arise from transitions between different
levels (not listed here) of the
configuration. The excitation energies for the lower levels
of all the TiI lines shown in Fig. 4 range between 1.734 to 1.749 eV
except for the 2.2639
m for which it is
1.879 eV (the ionization potential of TiI is 6.83 eV).
|Figure 4: A magnified section of the K band spectra showing the unusual presence of several TiI lines in V838 Mon.|
|Open with DEXTER|
From the observed strength of the TiI lines we estimate the amount of
TiI in the ejecta and from this calculate the mass of the shell. The
observed luminosity in an individual TiI line is first calculated which
will be given by:
By equating Eqs. (2) to (1) the number of TiI atoms in the shell
can be found.The two unknown quantities of Eq. (2) viz.
are found as follows.
The value of A21 for any particular line is given by
The number of TiI atoms in the shell, as computed from Eqs. (1) to (4), is listed in Table 1 along with some of the other relevant parameters needed in the calculations. The W values in Table 1 are for the 2 May 2002 observations when the TiI lines were strongest and the signal-to-noise ratio in the continuum was about 50. To reduce errors in the measured equivalent widths, an average W was measured from at least two (at times four) different spectra. However, as can be seen from Fig. 4, the 2.2217, 39, 80 and 2.2317 lines are blended (blend 1) and so also the 2.2627 & 2.2639 lines (blend 2) and so the individual equivalent widths cannot be measured. To calculate for these lines is still possible by the following method. The combined equivalent width of blend 1 and 2 was measured separately. The line luminosity corresponding to this combined equivalent width for a blend (as calculated from Eq. (1)) will be equal to the summation of the right hand side of Eq. (2) for all the blended lines in the group. Thus an average for the blended lines can be determined. The equivalent widths for the blend 1 and 2 lines in Table 1 is therefore the combined equivalent width. In the calculations we have adopted a value of d = 790 pc (Munari et al. 2002a), T = 3600 K and U = 23 (for T = 3600 K). The temperature of the shell is not known, but is possibly low, in the range 3000 to 5000 K. Such a possibility is indicated because, as mentioned earlier, the TiI lines are all low excitation lines. Secondly the presence of CO which is seen in emission in the K band (discussed subsequently) is generally associated with this temperature range. From Table 1 it is seen that , as derived from the different lines, is reasonably consistent with a variation of 5 among the individual values. We use a mean value of for calculating an important parameter viz. the mass of the shell. Assuming that the and abundances in the shell are adequately represented by cosmic abundances (Allen 1976), the mass of the ejecta can be determined. This is found to have a value . Similar calculations, as described above, have been computed for different temperatures yielding a shell mass ranging between to for a variation in the temperature between T = 2600 to 5000 K respectively. Within the uncertainties of the parameters used, a reasonable constraint for the shell mass would be in the range between 10-7 to . The derived value of compares reasonably well with lower values of envelope masses determined for novae. In novae an average value of can be taken though deviations by more than an order of magnitude (on both lower and higher side) are often seen (Williams 1994). It may be noted that derived here may be underestimated because we have assumed that the TiI emission originates entirely from the matter ejected in the present eruption and not from any pre-existing material. Further, even if the TiI emission arises only from the ejected material, it could be arising only from a fraction of it. Thus the shell mass may be underestimated and this may be borne in mind while trying to classify V838 Mon into known categories of eruptive variables - an aspect which is discussed later in Sect. 4.
|Obs. date (UT)||J||H||K|
|3.614 May 2002||5.15 0.14||4.16 0.15||3.63 0.14|
|14.604 May 2002||5.43 0.04||4.43 0.07||3.72 0.03|
The E(B-V) value for
V838 Mon is uncertain and lies between 0.25 to 0.8 (Munari et al. 2002a).
Following Munari et al. (2002a), we have adopted a midpoint value of 0.5.
We have used Koornneef's (1983) relations viz.
AV= 3.1 E(B-V),
AJ= 0.265 AV,
AH= 0.155 AV and
AK= 0.090 AV to correct for interstellar
extinction. Absolute flux calibration was done by adopting zero magnitude
fluxes from Koornneef (1983). The broad band JHK fluxes from V838 Mon
are shown in Fig. 5
as filled circles. These broad band fluxes were fitted by black body curves.
Different black body curves generated for temperature increments of
100 K were tried - the best fit was decided by least squares
minimization of the deviation. Black body fits of 2600 and 2400 K are seen
to fit the data reasonably well for the 3 and 14 May data respectively.
A temperature of 2600 K for 3 May is identical with the findings
of Munari et al. (2002b)
at around the same time. The overall temporal evolution of V838 Mon since
outburst has been towards cooler temperatures. This is confirmed by a large
body of photometric data as given by Munari et al. (2002a,b) and
also from reports
in several IAU circulars. We have superposed the JHK spectra of 2 and
14 May in Fig. 5 so as to juxtapose them side by side. This enables to
highlight the strong dips seen in the spectra of 2 and 14 May between
the J & H and the H & K bands. However in doing so, we assume
that errors caused by applying photometric data of 3 May to the spectra of
2 May are marginal because of the small time-difference involved.
The strong absorption bands between the near-IR bands is attributed to water
vapor in V838 Mon which if present in the atmospheres of cool stars is known
to cause deep and broad absorption features at 1.4
m (Lancon & Rocca-Volmerange 1992; Terndrup et al. 1991).
These absorption features are seen
to be most prominent in giant stars even cooler than M 5 (Lancon &
Roca-Volmerange 1992). However, an additional
factor that could enhance the H band hump is the H
which has an opacity minimum at 1.6
m. It must be mentioned that
a slightly elevated H band continuum near 1.6
m was also seen
during the 26 January observations of Lynch et al. (2002).
|Figure 5: The computed fluxes from V838 Mon on 3 and 14 May 2002 as derived from broad-band JHK photometry are shown here by the filled circles. The flux for the 14 May data can be read off directly from the graph whereas the 2 May data has been off-set by 2.5 flux units for clarity. Blackbody fits to these fluxes, at temperatures of 2600 and 2400 K, are shown by the smooth bold lines. The JHK spectra, represented by the irregular wavy lines, are also superposed in the figure. See the text for additional details.|
|Open with DEXTER|
V838 Mon does not also fit easily into the scenario of a born again AGB star like FG Sge, V605 Aql and Sakurai's object. Most of the discrepancies that arise in such a scheme are given by Munari et al. (2002a) including the fact that the rise to maximum is too fast in comparison with known cases. To this we add that not only the rise time, but also the decline time of V838 Mon is much faster than the known cases. For the rise and decline timescales, the reader may refer to Duerbeck et al. (2000) for Sakurai's object, Harrison (1996) for V605 Aql and Fig. 1 of Blocker & Schonberner (1997) for FG Sge.
Bond et al. (2002) have proposed that the closest resemblance of V838 Mon is to a red variable star (M 31 RV) that erupted in M 31 (Rich et al. 1989; Mould et al. 1990). This view is shared by Munari et al. (2002a) who also cite the outburst of V4332 Sgr (Martini et al. 1999) as being similar to the present case and propose that V838 Mon, M 31 RV and V4332 Sgr belong to a new genre of astronomical objects. Unfortunately there is no near-IR spectroscopic data for either M 31 RV or V4332 Sgr with which we can compare our spectra to check for similarities. However, there is a point of dissimilarity between M 31 RV and V838 Mon that arises from our data. As mentioned earlier, in the case of V838 Mon an envelope mass of 10-7 to 10 was found - a value comparable to novae shell masses. In the the case of M 31 RV the mass of the ejected shell is estimated to be much higher i.e. in the range of 0.1 to . These figures are arrived at by assuming equipartition between the kinetic energy of the shell and the total radiative energy i.e. by equating the total energy of 1046 ergs radiated in the first 100 days (Mould et al. 1990) to the kinetic energy of the shell computed for expansion velocities in the range 100-500 km s-1. Iben & Tutukov (1992) have modeled M 31 RV in the scenario of a very cold white dwarf accreting matter at a very slow rate from its binary companion and show that under such conditions thermonuclear runaway will not take place until the accreted mass is much larger than in models which represent the novae phenomenon. As a result larger envelope masses and outburst luminosities are predicted - as is found in the case of M 31 RV. How V838 Mon fits into such a model needs to be addressed and worked out. Other similarities and differences between V838 Mon, M 31 RV and V4332 Sgr have been brought out by Munari et al. (2002a) and Martini et al. (1999). Based on such comparisons, there is a fair possibility that all three objects belong to a new class of astronomical objects.
The research work at Physical Research Laboratory is funded by the Department of Space, Government of India. We thank Varricatt W.P. of Joint Astronomy Centre, Hawaii for help in photometric reductions. We thank the anonymous referee whose constructive comments helped improve the paper. This work has made use of data available from the ADC and CDS data centers and also from data available at the following websites viz. http://www.noao.edu/archives.html, http://kurucz.harvard.edu/linelists.html and