4 Discussion

There are not many objects in the Milky Way displaying so strong emission lines of H I and Fe II as MWC 314 does. Nearly all of them discovered so far have been recently collected in a catalog of Thé et al. (1994). Their physical parameters and evolutionary state may vary in a very large range from extremely massive relatively young supergiants (MWC 349) to low-mass proto-planetary nebulae (OY Gem). The nature of some of these stars, like MWC 84, MWC 930 (Miroshnichenko 1995), is still uncertain. However, the main characteristics of the stars and their environments are uncertain even for well-observed objects. One of the main problems causing incomplete knowledge of these objects is the lack of high-resolution spectroscopy and long-term photometric observations. High-resolution spectroscopy is important because the strong emissions of the circumstellar envelopes existing around such stars severely veil photospheric spectra. On the other hand, these observations help to constrain the shape of the gaseous component of the envelopes and to estimate distances toward the objects using radial velocities (e.g., Humphreys et al. 1989).

4.1 Stellar wind

In general, the emission line profiles observed in the spectra of MWC 314 are double-peaked without blue-shifted P Cyg-type absorptions. Such a structure is sharply different that of some other peculiar supergiants and LBVs like P Cyg, HD 316285, AG Car, while similar profiles are observed in such B[e] stars as MWC 349, MWC 84, OY Gem. This implies that the shape of the circumstellar envelope, where the emissions originate, deviates from spherical and that it is not viewed edge-on. Miroshnichenko (1996) modeled the MWC 314 Balmer line profiles in spherical geometry and converged theoretical profiles to match the observational resolution, which was about 2 Å for his data. However, the results of his original calculations, which were done with a resolution of nearly 0.5 Å, always showed the presence of P Cyg-type absorption components. Now with a 5 times better resolution in the H$\alpha$region we still do not see any absorption component in the line profile. They are weak in the helium lines originating close to the stellar surface where density is significantly larger than in the region of the Balmer line formation. The same lines in the spectrum of P Cyg, which do not show evidence of a non-spherical wind (Najarro et al. 1997), have very strong absorption components. This strongly favors a non-spherical wind from the star. In a forthcoming paper we will model profiles of the hydrogen and helium lines using a disk-like matter distribution around MWC 314.

The weakness of the forbidden iron lines indicate that the stellar wind of MWC 314 does not contain extended low density regions. The widths of most emission lines are 130 $\pm$ 5 km s^-1 FWHM and 160 $\pm$ 5 km s^-1 at the continuum level (FWZI). Only H$\alpha$and H${\beta}$have FWZI of $\sim 1700$ and 900 km s^-1 respectively which can be due to electron scattering effects. This is not surprising in such a dense wind with the physical parameters as quoted in the introduction. The strength of the Ca I IR triplet, which shows a larger intensity relatively to the continuum (4.3-5.8) among similar stars (Lopes et al. 1992; Andrillat et al. 1996, 1997; Jaschek et al. 1996a,b), is probably another evidence of the wind high density.

4.2 Effective temperature

Our new results confirmed the suggestion of Miroshnichenko (1996) that MWC 314 is an early B-type high-luminosity star. We detected its photospheric lines for the first time, however such well-known luminosity indicators as Si IV and O II lines have not been found in the spectrum. This is mainly due to the presence of a large number of emissions located at the positions of many photospheric lines, abundance anomalies (see below), and a limited spectral range of the obtained spectra. On the other hand, Fe III emission lines of multiplets 115 and 117 are usually observed in the spectra (Fig. 4) of early B-type hypergiants with $M_V \leq 8\hbox{$.\!\!^{\rm m}$}0$ (Wolf & Stahl 1985). These lines have radial velocities similar to those of Fe II and [Fe II], which are formed in the stellar wind. Nevertheless, the Fe III lines are narrower than the other iron lines and always single-peaked, indicating a different formation zone, which might be the polar regions of the non-spherical stellar wind (see a model for B[e] supergiants of the Magellanic Clouds by Zickgraf et al. 1986).

The weak signes of He II lines in the spectrum (their presence needs further confirmation at a higher resolution and signal-to-noise ratio) constrains an upper level of $T_{\rm eff}$at nearly 27000 K (Schmutz et al. 1991). The presence of other high excitation lines, like [Fe III] and [S III], also point out to an early-B spectral type.

The photospheric lines can be seen at a resolution of less than 1 Å. Our list of photospheric absorptions identified in the spectrum of MWC 314 does not contain oxygen lines. The same situation has been previously reported for such LBVs as AG Car and HR Car (Hutsémekers & van Drom 1991) and is suggested to be due to a N/O overabundance in the atmosphere. Other similar stars (e.g., P Cyg and MWC 300) show O II photospheric lines along with those of N II. In any case, such anomalies imply that MWC 314 is a highly evolved star. The presence of the Ne I lines gives this suggestion further support.

In order to estimate $T_{\rm eff}$ of MWC 314 we compared characteristics of some of its photospheric lines showing dependence on stellar temperature with those of other supergiants. An atlas by Lennon et al. (1992) was used in the blue region ($\lambda\lambda$ 4200 - 4900 Å), while spectra with echelle-spectrometers of the 1-meter SAO telescope (Musaev 1993, 1996) were examined in the red region ($\lambda\lambda$ 5400 - 6500 Å). Both data sets were obtained with a resolution close to those of our data, i.e. $\sim$ 0.3-0.8 Å. Nearly 20 stars of spectral types O9 - B8 was used from the atlas and 7 B2 - B8 stars from the SAO archive.

In the blue region we found only one line, N II (M. 5) at 4630.5 Å, for the $T_{\rm eff}$ estimate. Its strength increases with $T_{\rm eff}$ decrease for the spectral types O9 - B2, then decreases towards B4, and remains weak and almost constant at lower temperatures. This gives a spectral type for MWC 314 of B0 or B3.5. Two lines of S II (M. 6) at 5453.8 and 5473.6 Å, two lines of N II (M. 3) at 5679.6 and 5686.2 Å, four lines of Ne I (Ms. 1 and 5) at 6143.1, 6163.6, 6334.4, 6402.3 Å and two lines of Al III (M. 2) at 5696.5 and 5722.7 Å were used in the red region. The N II and Al III lines weaken sharply from B2 to B8, while the other lines used strengthen gradually from B2 to B5 remaining stable with further temperature decrease. Spectra of two hotter stars, HD 190603 (B1.5) and HD 38771 (B0.4), obtained at SAO were also examined. They contain the very weak S II lines and only traces of the Ne I lines and were not included in Table 5. The Ne I lines give a spectral type of B2.5 for NWC 314, N II and S II - B3.5, while Al III - B4-5. An example of such a comparison is presented in Fig. 8.

  

Figure 8: Comparison of the photospheric lines in the spectra of MWC 314 (dashed line) and HD 183143 (B7 Ia, solid line). The spectrum of HD 183143 was taken at the 1-meter SAO telescope and shifted to the photospheric radial velocity of MWC 314. Units of the wavelengths and intensities are the same as in Fig. 1

Thus, the photospheric lines characteristics suggest that MWC 314 has a spectral type of nearly B3 or B0, fixing the lower limit of $T_{\rm eff}$at approximately 15000 K. This value seems to be rather low and inconsistent with appearance of the strong He I emission lines, which are not observed in B3 stars (e.g., Zickgraf et al. 1986). Furthermore, Walborn (1980) noted that a Si III (M. 4) line at 5739.8 Å, which is seen in the MWC 314 spectrum, appears only in early-B stars (B0.5 and B1.5). Additionally, Miroshnichenko (1996) showed that $T_{\rm eff} <$20000 K is less consistent with the SED of MWC 314 in the UV region than 25000 and 30000 K.

On the other hand, B0-B1 supergiants display a strong blend of C III lines at 4647-4650 Å, which is not evident in MWC 314 because of veiling by a blend of the Mn II and Fe II emissions. Furthermore, the absence of the C III lines can also be explained by a carbon depletion. Thus, the underabundance of oxygen and carbon combined to the overabundance of nitrogen could be an evidence of CNO processing. In this case, a helium enrichment of the photosphere is expected. The Ne I lines in MWC 314 are stronger than in any other supergiant taken for comparison (Table 5). The B0-B1 supergiants do not show these lines. This feature of MWC 314 might be due to its larger mass ($\sim 80\ M_\odot$,Miroshnichenko 1996) and, hence, faster evolution. Thus, B0, corresponding to $T_{\rm eff}$$\sim$ 25000 K, fits the whole picture of the MWC 314 spectrum better.

Table 5. Photospheric line characteristics in MWC 314 and other supergiants

The peak intensities are given in units of the continuum while the equivalent widths in Angströms. For each of the comparison stars their spectral types and references to them are given. The references are as follows: 1 - Kopylov (1958); 2 - Bartaya et al. (1994); 3 - Chentsov & Luud (1989). Equivalent widths for HD 58350 were taken from Underhill & Fahey (1973).

 

Table 6: Radial velocities

• [$^{\rm a}$] all detected lines besides those at $\lambda$5876 and 7065 Å,
• [$^{\rm b}$] only the lines at $\lambda$5876 and 7065 Å,
• [$^{\rm c}$] Fe II, Ti II, Cr II,
• [$^{\rm d}$] all forbidden lines except [Ca II],
• Radial velocities are given in km s^-1, numbers of the lines used to determine the mean velocity are given in the last column.

4.3 Radial velocities, distance, luminosity

The spectral resolution of our data allows to study radial velocities of different lines. Only the lines with clearly resolved peaks uncontaminated by nearby features having comparable intensities have been selected for radial velocity measurements. It is seen from Table 6 that there is a significant difference in velocities of different types of lines. The heliocentric velocities of the photospheric lines are grouped around +81 km s^-1, those of the lines of ionized Fe, Ti, and Ca as well as of the Balmer lines around +41 km s^-1, while the lines of He I and Si II have velocities close to +20 km s^-1. The paschen lines, those of Ca II, the strongest Fe II (mainly of Mult. 42) lines, He I 5876 and 7065 Å lines, and emission components of Na I D-lines have radial velocities close to +35 km s^-1. The velocities of the diffuse interstellar bands are close to zero. The correction for the solar motion in the Galaxy, which should be applied to these values to reduce them to the local standard of rest, is +14 km s^-1. These data can be used to estimate the systemic velocity and, finally, distance toward the object employing differential rotation of the Galaxy. To do this the systemic velocity should be separated from other motions. The photospheric lines represent radial velocity of the star itself which might have its own peculiar motion. The lines of He and Si are formed in the very inner parts of the wind and their velocities can be affected by strong acceleration of the outflowing matter close to the star. The lines of other metals originate at larger distances from the star and most probably have radial velocities close to that of the system. The hydrogen lines forming in a very extended region of the wind have mean velocities comparable to those of the ionized metals. Taking into account the correction to the local standard of rest we can estimate the systemic velocity as +55 $\pm$2 km s^-1. To calculate relation between distance (D) from the Sun and radial velocity in the object's direction we used a second order expansion of the standard equation for circular rotation with R₀ = 8.5 kpc, A₁ = 16.1 km s^-1kpc^-1, and A₂ = -0.7 km s^-1kpc^-1 (Dubath et al. 1988). The systemic velocity corresponds to $D = 3.0~ \pm$ 0.2 kpc, which is in excellent agreement with the previous estimate of Miroshnichenko (1996).

We should note that this method can be applied only to the objects having wind velocities much larger than those measured for the star itself using photospheric lines. If an object has additional components of its motion (e.g., orbital motion in a binary system) the problem becomes much more complicated. This method has already given reasonable results for such LBV candidates as AG Car (Humphreys et al. 1989), HR Car (Hutsémekers & van Drom 1991), and He3-519 (Smith et al. 1994). However, in other cases one should be careful estimating distance using radial velocities and the galactic rotation curve. For example, for MWC 300 ($v_{\rm rad} = 26~ \pm$ 3 km s^-1, Winkler & Wolf 1989) it gives D = 2.1 kpc, which is inconsistent with hypergiant classification for the star. In our case the estimate is confirmed by a strong reddening, complex structure of the Na I D_1,2 lines, and the relation between A_V and D in the MWC 314 direction (Miroshnichenko 1996).

With our refined values of the distance and effective temperature we can now re-estimate stellar parameters of MWC 314. Since the spectrum contains many emission lines we have to correct the visual flux for their contribution in order to derive absolute visual magnitude of the star. This correction is not expected to be large and we calculate it to realize how strong it could affect the result. Taking into account 172 emission lines detected in a spectral region corrsponding to the Johnson V-band (Kornilov et al. 1991) we found that their net effect is only $0\hbox{$.\!\!^{\rm m}$}05$ added to the continuum level. Using the mean level of the object's visual brightness $V = 9\hbox{$.\!\!^{\rm m}$}9$ and the value of interstellar extinction $A_V = 5\hbox{$.\!\!^{\rm m}$}6$(Miroshnichenko 1996) in combination with D = 3 kpc derived here one can calculate $M_V = - 8\hbox{$.\!\!^{\rm m}$}1$. The bolometric correction corresponding to $T_{\rm eff}$ = 25000 K, $-2\hbox{$.\!\!^{\rm m}$}5$ (Gubotchkin & Miroshnichenko 1991), gives an absolute bolometric magnitude $M_{\rm bol} = - 10\hbox{$.\!\!^{\rm m}$}6$ or log $L_{\rm bol} / L_\odot$ = 6.1, and, consequently, stellar radius $R_* = 60\ R_\odot$.

The accuracy of the luminosity and radius estimates are set by the uncertainties of A_V, $T_{\rm eff}$, and D, which are approximately 10% each. The latter give an uncertainty of $0\hbox{$.\!\!^{\rm m}$}25$ in M_V, about 30% in $L_{\rm bol}$, and 50% in R_*. A lower limit for R_* is set by the $A_V \sim D$ relation in the vicinity of MWC 314 (Miroshnichenko 1996), which shows a constant increase of A_V up to 2 kpc from the Sun. At this minimum distance the star would have a radius of 50 $R_\odot$.The upper level of $R_*, 90\ R_\odot$, gives log $L_{\rm bol} / L_\odot$ = 6.5 and, therefore, D = 4.3 kpc. An independent estimate of D can be obtained by means of radio continuum observations. The radio flux at the lower D limit would be 5 times larger than that at the upper one, while the expected value at 3 kpc is about 4 mJy according to the model calculations of Miroshnichenko (1996).