4 The kinematic structure of the nebula

Our high-resolution echelle spectra allow us to analyze the kinematic structure of the S 119 nebula in great detail. At a FWHM resolution of 8 km s^-1 our observation fully resolve the global structure of the Doppler ellipse of the expansion of the S 119 nebula.

4.1 The overall expansion

We find a radial velocity of the center of expansion at $156 \pm 2$ km s^-1. This is in agreement with earlier findings that the radial velocity of the S 119 system is well below that of the main part of the LMC. The echellograms show background H$_\alpha$ emission along the entire slit at all positions, which presumably results from an H  II region within the LMC. The radial velocity of this H  II region of $\sim$264 km s^-1 is consistent with it being located in the LMC. With respect to this H  II region, the S 119 nebula moves with about 100 km s^-1 more along the line of sight.

All spectra passing over the central part of the spherical main body of the nebula show an expansion ellipse as expected for a spherical expansion. We derive a maximum expansion velocity of $25.5 \pm 2$ km s^-1 at the central position in both slits that cross the central star (see pv-diagrams for Slit Center and for Slit PA  =  213 at the position 0 $^{\prime \prime }$ in Figs. 3, 4). As we move away from the geometric center of the nebula, the slits (Slit 2.5N and Slit 2.5S) prove the decrease of the expansion, that reduces to 15.5 km s^-1 and 21.4 km s^-1 respectively.

We measured sizes of the Doppler ellipses in our echelle data and get diameters of 6 $.\!\!^{\prime\prime}$0 (Slit 2.5N), 9 $.\!\!^{\prime\prime}$1 (Slit Center), and 6 $.\!\!^{\prime\prime}$8 (Slit 2.5S), respectively. These sizes were determined from the spectra and not from the binned pv-diagrams. Given the considerably different spatial resolutions these values agree with those derived from the HST images (5 $.\!\!^{\prime\prime}$7, 8 $.\!\!^{\prime\prime}$6, and 6 $.\!\!^{\prime\prime}$2, respectively).

For a spherical shell of radius R a spherically symmetric expansion with velocity $v_{\rm exp}$ would lead to an observed radial velocity relative to the systemic velocity

$$v_{\rm rad,rel}^{\rm spher} (x,y) = v_{\rm exp} \left( \frac{R^2 - x^2 - y^2}{R^2} \right)^{1/2}$$ (1)

at a location $\left\{x,y\right\}$ in a Cartesian coordinate system centered on the star. In the present case, we would assume $R = 4\hbox{$.\!\!^{\prime\prime}$ }5$ and $V_{\rm exp} = 25.5~$km s^-1 from our data. If we orient the coordinate system such that the y-axis is along the slit direction with y = 0 being the star's projected position onto the slit and x being the offset of the slit with respect to the star, we get for the maximum radial velocities $v_{\rm rad, rel, max} = v_{\rm rad, rel}^{\rm spher} (x=0,y) = 25.5~\hbox{km~s$^{-1}$ } \sqrt{1 - \left( y / 4\hbox{$.\!\!^{\prime\prime}$ }5 \right)^2 }$. In Fig. 5 we give a comparison between the absolute value of the radial velocities (corrected for the systemic velocity) at the star's position projected onto the slit, and the model values. If it were a purely spherical expansion, the measured values should fall onto the dashed model curve). We find a clear deviation from a purely spherical expansion, for instance in the north-south asymmetry in Slits 2.5N and 2.5S.

Figure 5: A position-velocity diagram to compare the measurements with a model of a spherically expanding shell (dashed curve). The errors of the measurement are $\pm ( 0.5 {-} 1 )~$ km s-1, and thus of about the size of the dot symbols or smaller.

Both the red-shifted and the blue-shifted components of the expansion ellipse have a FWHM of 0.3 - 0.5 Å (13 - 23 km s^-1), compared to an instrumental FWHM of 8 km s^-1. The echellograms in Figs. 3 and 4, in particularly the small inserts of the expansion ellipse reveal a clumpy sub-structure. Most prominent in Slit Center, the blue-shifted wing of the ellipse shows brightness variations (at a roughly constant FWHM). These variations are most likely due to the brighter knots identified in the HST image, which were intercepted by the slits. However, due to the large difference in resolution between our spectra and the HST image, no clear identification of the individual knots is possible.

4.2 The outflow

Figure 6: Overlay of the HST image of S 119 and the spectrum Slit PA  =  213 around the [N  II]$\lambda$6583 Å line showing the location of the outflow.

Figure 7: Ratio map of the [N  II] $\lambda$6583 Å and H$_\alpha$ lines of spectrum Slit PA  =  213. In this gray scale plot of the two-dimensional spectrum we clipped the background so that it is at a near noise-free zero value and slightly smoothed the data with a $3\times 3$ median to improve the visibility of low surface brightness features. Note the high [N  II] $\lambda$6583Å/H$_\alpha$ ratio of the ring and the outflow region compared to the background H  II region.

In addition to the spherical expansion we find in all spectra clear indications for a much higher velocity expansion structure (for an example, see Fig. 6). It extends spatially from $\sim$ $+4\hbox{$^{\prime\prime}$ }$ to $\sim$ $+7\hbox{$^{\prime\prime}$ }$ and in velocity space from the end of the expansion ellipse to more than 280 km s^-1 redshifted. Only in Slit 2.5N this emission feature appears to be detached from the expansion ellipse, and merges with the background H$_\alpha$ emission. To ensure that this high velocity component is part of the S 119 nebula and not a structure in the background emission, we determined its [N  II]$\lambda$6583 Å/H$_\alpha$ ratio. If the feature contains CNO processed material we expect a significantly larger value for an LBV nebula than for an uncontaminated H  II region. Nebulae around LBVs do have a higher content of nitrogen due to CNO processed material that is mixed into outer regions of the star, which will be ejected and forms the nebula. The higher nitrogen abundance of the nebula can be traced by the [N  II]$\lambda$6583 Å/H$_\alpha$ ratio. Typical examples of this ratio for LBVNs are 0.4 - 0.9 for HR Car (Hutsemékers & van Drom 1991; Weis et al. 1997), 0.7 for AG Car (Thackeray 1977; Smith et al. 1997), or even as high as 3 - 7 in $\eta$ Car (Davidson et al. 1982; Meaburn et al. 1987, 1996; Weis et al. 1999). For the spherical nebula around S 119 we derive a ratio of $0.6 \pm 0.1$, for the high velocity component we still find $0.5 \pm 0.1$ (Fig. 7). Both ratios are well within the range found in LBVNs, and are significantly larger than the ratio of the H  II region in the background of which we observe [N  II]$\lambda$6583 Å/H $_\alpha = 0.04$. While not far from the limit of our spatial resolution, the combination of spatial and velocity resolution makes it unlikely that our [N  II]$\lambda$6583 Å/H$_\alpha$ determination is contaminated. The [N  II]$\lambda$6583 Å/H$_\alpha$ values derived for the nebula agree well with earlier measurements of Nota et al. (1994).

The largest velocity of the high-velocity feature was detected in the Slit Center position with 283.1 km s^-1, almost 130 km s^-1 faster than the center of expansion of the central nebula. The velocity of this component increases approximately linearly with distance from the star's projected position. The echellograms do not show any traces of this feature within the Doppler ellipse of the spherical nebula. All these properties strongly point towards this feature being a component of the nebula around S 119 and not a projection effect.

In Fig. 8 we combine the plots of all pv-diagrams available to us. A dashed line marks the radial velocity background H  II region. Both, the spherical expansion of the nebula's main body as well as the high-velocity component are clearly visible. Due to its larger overall offset from the star, in Slit 5S (diamond shape symbols) the outflow seems to set on at a smaller projected offset from the star ( $+ 2\hbox{$^{\prime\prime}$ }$ as compared to $+4\hbox{$^{\prime\prime}$ }$ for the other slits). This is caused by the effect that the slit intercepts the outflow region earlier.

In Slit 2.5S (square shape symbols) at a projected position of $- 5\hbox{$^{\prime\prime}$ }$ another feature appears which moves slightly faster by $\sim$ $20~{\rm km~s}^{-1}$. This feature most likely can be assigned to a part of the filament SE-1 identified in Sect. 3.

Figure 8: Combination of all pv-diagrams with a PA  =  125$^\circ$ of Figs. 3 and 4. The dashed line indicates the radial velocity of the background LMC H  II region.

4.3 X-ray emission?

A 21 ks integration with the ROSAT HRI shows no X-ray emission of the nebula. At a distance of 50 kpc and for a Raymond-Smith plasma this corresponds to an upper flux limit of $\sim$ $1\times 10^{33}~{\rm erg~s}^{-1}$, almost independent of assumed temperatures between 0.1 and 1 keV (see Technical Appendix to the ROSAT Call For Proposals, ROSAT Appendix F). Thus we can only demonstrate that this limit is consistent with the conditions in the S 119 nebula:

• For an energy conserving bubble, for instance, we find an expected X-ray flux of $\sim$ $2\times 10^{32}~{\rm erg~ s}^{-1}$(Weaver et al. 1977; Chu et al. 1995; with a mechanical wind luminosity of $8.1\times 10^{40}$ erg s^-1, an electron density of 700 cm^-3, an electron temperature of 7000 K, and an expansion age of $2\times 10^4$ yr).
• Moreover, even for the highest outflow velocities ($\sim$10² km s^-1), may lead to post-shock temperatures at a possible interface with the ambient medium of $T_{\rm ps} \approx 1.3\times 10^5~{\rm K}$ and thus give rise to UV radiation rather than X rays detectable by ROSAT.