Molecular Hydrogen Emission in the Wolf-Rayet Nebula NGC2359

Nicole St-Louis, René Doyon, François Chagnon and Daniel Nadeau

Département de Physique, Université de Montréal and Observatoire du Mont Mégantic,
C.P. 6128, Succ. Centre Ville, Montreal, H3C 3J7, Canada




Abstract

We report on the first direct detection of molecular hydrogen (H2) emission in the interstellar medium in the vicinity of a Wolf-Rayet star. The spatial distribution of the excited molecular gas associated with NGC2359 is filamentary and lies mainly on the border of the ionized gas, as traced by optical emission lines such as Halpha or [OIII]5007. The typical H2 brightness in the filaments is 5 x 10-5 ergs s-1 cm-2 s-1str-1 and the total H2 luminosity detected is ~ 4 LSun. The detected line flux in the 1--0 S(1) transition of H2 at 2.122 microns could equally be explained by shock excitation or by fluorescence from the strong ultraviolet flux of the Wolf-Rayet star. The morphological distribution of the H2 filaments is not inconsistent with either mode of excitation. Although the ubiquity of this phenomenon needs to be confirmed, the relatively high level of 1--0 S(1) H2 emission detected in this WR nebula indicates that hot stars could potentially contribute a significant fraction of the total H2 emission of young starburst galaxies.





1. Introduction

Wolf-Rayet (WR) stars, because of their advanced evolutionary state and dense stellar winds (Mass loss rate = 0.8 -- 8 x 10-5 MSun yr-1 ; Abbott & Conti 1987 and vinfinity = 1000 -- 3000 km s-1 ; Prinja, Barlow & Howarth 1990) play an important role in physically shaping the interstellar medium and in its chemical and kinematical evolution. Although they are few in number, there is no doubt that their impact is significant as they input into the interstellar gas approximately 50% of the total wind energy provided by all stellar types taken together, at least in the solar vicinity (Abbott & Conti 1987). In addition, they supply a copious amount of gas enriched by either hydrogen or helium burning products, which has a definite impact on the abundance of elements such as 4He, 12C, 17O and 22Ne (Meader 1992). Furthermore, as a WR star represents the stage in massive-star evolution that immediately precedes the supernova explosion , the study of the composition, density distribution and kinematics of the interstellar gas in the vicinity of this type of star will yield a better understanding of the initial conditions in the interstellar medium prior to a supernova explosion.

Perhaps the most spectacular examples of the interaction between the winds of these hot stars and the interstellar medium are the WR nebulae consisting of ionized gas observed in lines such as Halpha and [OIII]5007 . They have been classified in three different types according to their formation mechanism: diffuse , radiatively excited H II regions, ejecta-type nebulae and wind-blown bubbles (Chu 1981). In addition , neutral hydrogen voids and shells thought to be associated with the ionized gas have been discovered around several WR stars in our Galaxy (e.g. Niemela & Cappa de Nicolau 1991 and references therein; Dubner et al. 1992; Arnal 1992; Arnal & Cappa 1996; Cappa et al. 1996).

NGC2359, one of the first three WR nebulae to be identified by Johnson & Hogg (1965), is the prototype wind-blown bubble. Its distance has been estimated by many authors (see Goudis et al. 1994 for a summary) and ranges from 3.5 kpc to 6.9 kpc. For the remainder of this paper, we will adopt a distance of 5 kpc. Abundance studies by several authors (Peimbert et al. 1978; Talent & Dufour 1979; Esteban et al. 1990) have demonstrated that the nebula shows very little chemical enrichment from processed stellar material. NGC2359 appears to consists of two distinct parts. The first is a U-shaped diffuse HII region that is thought to have been created by the O-star ancestor of the WR star (Dufour 1989). Within this region, which is seen most strikingly in [NII]6584 (Schneps & Wright 1980), lies the filamentary bubble blown by the wind of the WN4 star HD 56925 (=WR7 in the catalog of van der Hucht et al. 1981). This slightly egg-shaped nebula, which can readily be observed in Halpha or [OIII], is colliding with the diffuse HII region on its eastern side. Beyond this diffuse HII region, the optical nebula is partially obscured by a molecular cloud detected in CO (Schneps et al. 1981). The part of the nebula that is obscured is revealed by high-resolution 20 cm VLA observations by the same authors. Three distinct molecular clouds have been identified: the first borders the southern part of the U-shaped nebula and is observed to have the same velocity as the ambient interstellar medium around the nebula (VLSR=55 km s-1, see below). The second bounds the bubble and the U-shaped nebula on their eastern side and is moving at VLSR=37 km s-1. This cloud is therefore thought to be compressed and accelerated by the wind-blown bubble. Finally, the third cloud which is located in the south-east has a velocity of VLSR=67 km s-1 and is probably either foreground or background.

The kinematics of this nebula have been extensively studied and were found to be rather complex. Schneps & Wright (1980) (see also Schneps et al. 1981) were the first to recognize that the emission from the Halpha or [OIII] lines have three distinct origins: diffuse emission from the large HII region and, from the bubble, both thick filaments which are stationary and a thin expanding membrane which is overrunning them. The systemic velocity of the ambient gas is VLSR ~ 55 km s-1 as measured by several authors (Georgelin & Georgelin 1970, v=+56 plus or minus 5 km s-1; Lozinskaya 1973, v=+55 plus or minus 25 km s-1; Pismis, Recellas-Cruz & Hasse 1977, v=+53 plus or minus 8 km s-1; Treffers & Chu 1982, v=+52 plus or minus 3 km s-1; Goudis et al. 1983, v=+54 plus or minus 5 km s-1; Goudis et al. 1994, v=+52 plus or minus 3 km s-1). The bubble expansion velocity was estimated to 18, 30, 15--30 and 26 km s-1 respectively by Treffers & Chu (1982), Goudis et al. (1983), Chu (1988) and Goudis et al. (1994).

The value of the total mass of this wind-blown bubble is still very controversial. Schneps et al. (1981) estimated the mass of ionized gas to be ~ 16 MSun (for a distance of 5 kpc) but Van Buren (1986) based on dynamical arguments suggests that it might have a neutral component and that the total mass of the bubble is, in fact, much higher than revealed by just the ionized gas. In an attempt to detect a possible molecular component to the bubble in NGC2359, we have obtained a series of narrow-band H2 images covering most of the optical nebula. The observations are described in Section 2 and our findings are discussed in the context of the present models for this wind-blown bubble in Section 3. In Section 4, we summarize our results.



2. Observations and data reduction

We observed NGC 2359 in 1993 September/October with the Canada-France-Hawaii telescope (CFHT) on Mauna Kea, Hawaii, using the MONtreal Infrared CAmera which is equipped with a HgCdTe NICMOS detector from Rockwell International (MONICA; Nadeau et al. 1994). With the setting used for the observations, the pixel scale of the detector was 0.25 '' and therefore the 256 x 256 pixels of the camera corresponded to a field of view of approximately 1' x 1'. A preliminary scan of the entire 5' x 7' optical nebula was made using a narrow band (Delta lambda / lambda = 1 %; lambdac=2.122 microns) H2 filter centered on the 1--0 S(1) transition. Relatively strong emission was discovered in the southern part of the diffuse U-shaped H II region. Further observations were therefore concentrated in this area. In total, five 60-second scans were made with a total area covered of 4.5' x 3.4'. Approximately 100 images were combined to form the final mosaic which is presented as part of a three-color image in the inset of Figure 1. We have also obtained K-band and Brgamma mosaics of the nebula but no significant emission was detected.

Before creating the final H2 mosaic, each individual image was reduced in the following way. First a median sky was formed from the image being treated and three images on each side. After subtracting the sky from the image, it was divided by the flat field and corrected for the geometrical distortion. We also corrected for the variable sky transmission by using stars in the overlap regions of adjacent images. It was not necessary to subtract the continuum from the line images since the non-detection of any significant flux in the K-band or Brgamma images indicates that the contribution from continuum emission is negligible. Finally, once the mosaic was complete, we used UKIRT Faint Standards number 8 and 9 (Casali & Hawarden 1992) to carry out the flux calibration.



3. Results

In the montage of Figure 1, we show in the top left a global view of the optical nebula taken with an [OIII]5007 filter. The white box in the southern part of the U-shaped nebula indicates the region where the infrared observations were concentrated. These are presented in the inset, showing a three-color image of the southern part of NGC2359. The H2, Halpha and [OIII]5007 emission is coded in red, green and blue respectively. All three intensity scales are logarithmic. The optical images used to create Figure 1 were kindly provided by Miller & Chu (1993). This figure shows that the excited H2 component of NGC 2359 consists of long filaments which generally do not coincide with strong Halpha or [OIII] emission but rather are located on the immediate border of the ionized gas or in regions of very low ionized flux. H2 emission has also been detected in the northern part of the nebula but we have not taken sufficiently long exposures for significant quantitative results to be obtained. Therefore for the remainder of this paper, we will concentrate on the images of the southern part of the nebula.





Figure 1 : [OIII]5007 image of NGC2359 from Miller & Chu (1993). The total field of view is approximately 13' x 13'. The inset is a three color image of the southern part of the WR nebula NGC2359. The red corresponds to a mosaic of H2 emission, the blue to [OIII]5007 and the green to Halpha. The field of view of this image is 4.5'x 3.4'.



The brightest filament follows exactly the southern border of the U-shaped nebula, but there is also H2 emission within it. In particular, there are three filaments oriented in the northeast-southwest direction that fall in regions of very low optical Halpha or [OIII] emission. The northernmost filament is amazingly strait; it is not clear what can produce such a structure. It has no special orientation with respect to the WR star. At the northern tip of these three filaments lies a very faint curl-shaped structure. Note that this together with the three filaments were also detected in the light of [NII]6584 by Schneps & Wright (1980). In fact, the entire H2 flux distribution we have detected is extremely similar to the distribution of [NII]6584 in this region. Those authors have interpreted the [NII] emission as indicating the location of the ionization front.

Also intriguing is the X-shape structure observed in H2 and located just below the center of the three-color image. The lower section of the X is part of a longer filament that borders the optical emission. A clue to the origin of the upper part of the X is obtained by examining Figure 2 in which we show the same three-color image but with the color contrast adjusted in such a way that only the very brightest Halpha and [OIII] emission is visible. This figure strongly supports the suggestion made by Schneps & Wright (1981) that the bright arc-shaped structure just below the main bubble is actually part of another bubble which is colliding with the main WR nebula. The southern part of this second bubble is revealed in bright Halpha (green) emission superposed on H2 emission which forms the upper part of the X mentioned above (yellow arc). This southern part of the bubble was not recognized in previously published optical images as it was always confused with lower level emission from the U-shape nebula. In Figure 2 it can be seen that the upper arc-shaped structure below the main WR bubble is in fact composed of a mixture of Halpha and [OIII] emission (seen in turquoise) except in the very inner part which is seen only in [OIII] (blue).





Figure 2: Same as in the inset of Figure 1 but with the color contrast adjusted in such a way that only the brightest levels of optical emission are visible.



Although it is difficult to distinguish in the reproduction of Figure 1, another interesting characteristic of the H2 emission is that it does not completely fill the region of low optical emission but rather is concentrated on its northwestern border, towards the direction of the WR star. The same statement can be made for the curl-shaped structure at the northeastern tip of the filaments. As an example, Figure 3 shows a cut perpendicular to the northernmost H2 filament. We show the average intensity of 75 pixels along this filament of Halpha in the top panel and H2 in the bottom panel on a distance of 90 pixels perpendicular to the filament. It can be seen that the excited molecular gas is located in a much more narrow region that the Halpha trough and that it is not centered within it. Instead, it seems to be located on its north-westerly border.





Figure 3: Perpendicular cut across the northernmost H2 filament. In the top panel the mean Halpha flux of 75 pixels along the filament is shown as a function of position and in the bottom panel the same is shown for the H2 flux. The intensity units are arbitrary.



The typical surface brightness in the filaments is 5 x 10-5 ergs s-1 cm-2 str-1 with maximum values reaching 40 x 10-5 ergs s-1 cm-2 str-1. This is comparable to what has been detected in planetary nebulae and supernova remnants (Graham et al. 1987, 1993). In order to estimate the total 1-0 S(1) H2 luminosity in our field, we first removed the stars from our mosaic. This was done by fitting each star with a Gaussian and replacing it by a smooth background. Then we plotted a histogram of the number of pixels with a given intensity in the range -100 to +100 using 200 bins (1 ADU/s for each bin). This histogram consists of noise together with the H2 emission from the nebula. The latter is, of course, always positive. We assumed the noise to be symmetrical and used the negative part of the histogram to perform a spline fit to the curve. We then created a positive mirror curve to the fitted negative spline. Finally, we subtracted the fitted curve and its mirror counterpart to the original histogram curve. The remainder is the emission from the H2 gas. For a distance of 5 kpc, we estimate the integrated 1-0 S(1)H2 luminosity for this region to be ~ 4 LSun.



4. Discussion

4.1 The H2 excitation mechanism -- shocks or fluorescence?

The Halpha/[NII] and Halpha/[SII] line ratios clearly show that the ionized gas at various positions in the nebula, including positions close to where the H2 has been detected, is photoionized rather that being shock-excited (cf. Goudis et al. 1994). It is therefore natural to ask whether fluorescence by UV photons from the WR star could excite the H2 gas.

Black & van Dishoeck (1987) have calculated detailed models of fluorescent excitation of H2 in interstellar clouds. We will use their Figure 5 in which they plot the total H2 emission (Itot) as a function of the ultraviolet scaling factor of the radiation field (IUV) for clouds of various densities to determine if the necessary conditions are satisfied for NGC2359.

From our observed flux in the 1--0 S(1) transition, we estimate the total flux emitted in the H2 lines to be Itot=3 x 10-3 ergs s-1$^{-1}$ cm-2 str-1. This was determined by adopting model 14 of Black & van Dishoeck (1987) in which the 1--0 S(1) line is found to contribute 1.6% of the total H2 line emission.

The UV scaling factor of the stellar radiation field corresponds to the ratio of the number of non-ionizing UV photons reaching the cloud to the number of UV photons in the background radiation. It can be written as
IUV=310(NUV/1048 s-1)(d/pc-2)

where NUV is the number of non-ionizing photons capable of exciting H2 and d is the distance between the star and the emitting gas. We also adopted a mean background of UV photons of 2.7 x 1011 photons s-1 m-2 as measured by Draine (1978) for the solar neighborhood.

From the model atmospheres of Kurucz (1979), Puxley et al. (1990) estimated the number of H ionizing photons and the number of corresponding non-ionizing photons capable of exciting H2 (NUV; between 912--1108 angstroms) for stars of various temperatures. For the purpose of our study, it will be sufficient to assume that the radiation field of our WR star can be approximated by the models of Kurucz (1979). Esteban et al. (1993) have calculated the number of ionizing photons emitted by the WR star HD 56925 (5--13 x 1048 s-1). From Puxley et al. (1990) we find that this is similar to a star of ~ 40000 K and that the corresponding number of non-ionizing photons is roughly 8 x 1048 sU-1. With the adopted distance of the star of 5 kpc we estimate the distance between the star and the H2 gas in the brightest filament (d) to be approximately 5 pc. From these values, we obtain a UV scaling factor of ~100.

These values do not in fact correspond to any valid cloud model presented in Figure 5 of Black & van Dishoeck (1987). However, this model assumes that we are viewing the interstellar cloud ``face-on''. For optically thin emission , if we observe it with an angle, we must multiply the calculated H2 flux by a geometrical enhancement factor because of the longer optical path. With a modest value of ~ 5, we can easily explain the observed flux with an interstellar cloud having a density of approximately 1000 cm-3, which is close to the value for the southern cloud estimated by Schneps et al. (1981) from their CO observations. Note that this corresponds only to an average density; the density in the filaments could be much higher in which case a smaller geometrical enhancement factor would be required. Therefore it is clear that UV fluorescence is a valid candidate for the excitation mechanism of the observed molecular gas.

One can also ask if the observed spatial distribution of the H2 emission is compatible with the excitation mechanism of the H2 gas being UV fluorescence. Black & van Dishoeck (1987) demonstrated that fluorescent H2 emission increases with density and decreases with the square of the distance from the exciting star. This can possibly explain why we do not detect any H2 radiation in the molecular cloud located immediately to the east of the bubble. Indeed, Schneps et al. (1981) find that the mean density of the molecular cloud to the east is three times lower than the one in the south of the nebula which is consistent, at least qualitatively, with our non-detection of H2 emission.

In this scenario, we suggest that the H2 filaments we are seeing within the U-shaped nebula are areas of higher density. The higher extinction would explain why the ionized gas is not visible in these regions (or why it is not ionized) and self-shielding by the high density gas would also insure that the H2 molecules are not dissociated by the intense ultraviolet radiation field of the WR star. Self-shielding could also explain why the excited H2 gas we observe is only located on the north-western border of the filament. But if fluorescence is indeed the excitation mechanism of the molecular gas, one can then wonder what has produced such a filamentary density distribution in the first place. It has been suggested that the filaments in the wind-blown bubble were a consequence of the interaction of the bubble with clumps in the surrounding interstellar medium. Perhaps these have a similar origin and are a consequence of the work of the ancestor O-star wind on the surrounding interstellar medium.

However, the filamentary structure is also reminiscent of what is observed for supernova remnants (i.e. Graham et al. 1991) in which shocks are identified as the excitation mechanism of the molecular gas. Following Graham et al. (1991), the total 1--0 S(1) H2 flux can be written as

I1-0 S(1)= 9.0 x 10-6 (nH2/100 cm-3)(vs/25 kms-1)(cos(theta))-1 ergs s-1cm-2 str-1.


Using a typical density of 103 cm-3, a slow shock velocities of ~ 15 km/s and a small geometrical enhancement factor of ~ 2.5, we estimate that a partly dissociating shock is indeed capable of reproducing the observed 1--0 S(1) H2 flux. However, shocks should also produce considerable levels of [FeII]1.644-micron emission which we have not detected in an image consisting of five 60-second CVF observations of the brightest filament located in the pink box in the inset of Figure 1. This is a surprising result as the ratio of [FeII]/H2 1-0 S(1) is usually much greater that unity (i.e. Graham et al. 1987) for shocks, while in our case, we estimate it to be smaller that 0.14 (3 sigma). One possible explanation for the lack of [FeII] emission is the presence of the strong UV flux from the WR star which could prevent the recombination of Fe behind the shocks, in which case we should expect significant emission from [FeIII]. Just as the fluorescent H2 emission, the H2 emission from shocks also scales with the density and therefore, once again, this can explain why we do not detect anything on the eastern side of the bubble. In this model, the regions of low optical emission would also be zones of higher density in which the H2 has survived the intense UV flux through self-shielding. The relatively slow-moving shocks would be working their way through the dense filaments in the northwest-southeast direction and in doing so excite the H2 molecules. The fact that the excited gas is only located on the northwest border of the Halpha trough would indicate that the shock has not moved beyond this point yet. The shock would have to be non-radiative as line ratios show that the gas observed in the optical is photoionized.

There is also the possibility of a combined scenario in which the H2 filamentary structure was shaped by the passage of a shock front in the past but is presently being excited by the ultraviolet photons from the WR stars. Once again, the regions we have detected in H2 flux would then trace the zones of higher density. Finally, the only satisfactory way in this case to determine which H2 excitation mechanism dominates is to gather spectroscopic information and measure H2 line ratios. In addition to clearly identifying the dominant mode of excitation these line ratios will also allow us to determine which type of shock is present.

Schneps & Wright (1980) suggest that there might be a second ionizing star in this region located in the southern part of the U-shaped nebula. They propose that this star is blowing a bubble which is colliding with the WR bubble, creating the bright curved structure seen at the top left of the three-color image of Figure 1. Figure 2 strongly supports this suggestion and H2 gas is seen to border the southern part of the secondary bubble (yellow arc). The additional star is located very close to the H2 gas we have detected and therefore may contribute significantly to the excitation of the molecular gas.



4.2 The Origin of the H2

One of the main questions that remain to be answered about this nebula concerns its history and formation mechanism. The mass of ionized gas has been estimated from the 20 cm radio flux by Schneps et al. (1981) to be ~ 16 MSun (for a distance of 5 kpc). It is therefore likely that the bubble is mainly composed of swept-up gas as confirmed by abundance studies. Based on this mass, Treffers & Chu (1982) estimate that the kinetic energy in the shell is much less than 1% of the kinetic energy in the wind of the WR star integrated over the lifetime of the shell, which is more than a factor of 20 smaller than predicted by wind-blown bubble theory (Weaver et al. 1977). Treffers & Chu (1982) conclude that the bubble is in a momentum conservation phase and that energy losses have occurred. Schneps et al. (1981) propose that the non-moving dense filaments of the bubble may have been the result of interactions of the expanding tenuous bubble with small inhomogeneities in the external interstellar medium. In support of this hypothesis, Goudis et al. (1983) find that the [OIII] lines are much broader in the filaments than in the expanding bubble, which is consistent with a post-shock phenomenon interpretation. Goudis et al. (1994) then suggest that the radiative losses occur at the boundaries between clumps and shocked wind flows (Hartquist & Dyson 1993).

The alternative interpretation is that a large fraction of the bubble mass consists of neutral material and that the kinetic energy obtained from the ionized gas is underestimated by a large factor. Van Buren (1986) presents a dynamical estimate of the kinetic efficiency parameter and of the radial momentum conservation parameter that takes into account the neutral material and finds values much closer to those predicted by the theory for energy-conserving bubbles. He therefore suggests that a shell of neutral HI or H2 gas might be found associated with the ionized gas. This is possible if the ionization front is trapped inside the wind-blown bubble. By comparing the number of Lyman continuum photons from the WR star with the number of recombinations in the shell, Van Buren (1986) estimates that the ionization front could easily be trapped for NGC2359.

In support of this suggestion, Marston (1991) has presented IRAS observations of this nebula. From pointed 50 and 100 micron fluxes he detects dust heated at 33 K, mainly in the region to the south of the bubble. He estimates the mass of the dust to be 1.35 MSun. For a normal gas to dust ratio of ~ 100, this indicates that the mass of the shell is much higher than the 16 MSun measured from ionized gas. Using 135 MSun for the total mass of the shell, Marston (1991) finds that the bubble's kinetic efficiency parameter and the radial momentum conversion parameter are consistent with the energy conserving case. The small difference can be accounted for by energy losses in the clumpy interstellar medium.

The observations we have made were aimed at attempting to detect the molecular component of the bubble predicted by Van Buren (1986). The brightest filament seems to follow the general shape of the U-shaped nebula in the southern part and probably indicates the position of the border of the neighboring molecular cloud. The distribution of the very weak emission we have detected in the northern part also seems to follow this pattern although higher signal-to-noise data are required to confirm this. High spatial resolution observations in CO would clearly establish the limits of the molecular cloud. The filaments located within the U-shaped optical nebula most likely trace areas of higher density in which the H2 has not been dissociated. The strong ultraviolet radiation from the WR star photoionizes the matter inside the U-shaped nebula and excites the H2 molecules on its border or in the denser parts within it by fluorescence. Alternatively, slow shocks could be moving away from the star exciting the H2 gas on their way.

The location of this WR nebula right next to a dense molecular cloud has clearly had a profound influence on its present appearance. Nebulae around massive stars are generally thought to evolve according to the so-called three-wind model. First the star begins its life as an O star and blows a hole in the surrounding medium which can reach 50 pc in size. These are frequently associated with the HI voids and shells detected in the radio (e.g. Niemela & Cappa de Nicolau 1991 and references therein; Dubner et al. 1992; Arnal 1992; Arnal & Cappa 1996; Cappa et al. 1996). Then there is a phase of slower but much denser wind when the star undergoes a red supergiant (RSG) or luminous blue variable (LBV) phase, depending on its initial mass. Finally, when the star reaches its WR phase the wind becomes once again fast but less dense and sweeps up the material ejected from the previous evolutionary phase. In some cases, this wind-blown bubble can overtake the RSG or LBV ejecta and burst through.

In the case of NGC 2359, the boundary conditions vary in different locations around the star and therefore the evolution of the surrounding nebula should be slightly different. The star is located immediately to the west of the border of a dense molecular cloud detected by Schneps et al. (1981). On its western side the bubble can easily expand but on its eastern side it runs into the molecular cloud. Therefore, in its main-sequence phase, the star probably blew and ionized a hole which had a much larger extent on its western side than on its eastern or southern side because of the large density contrast. Dufour et al. (1989) suggest that this gave rise to a blister-type nebula which is the suggested origin of the U-shaped HII region. It would be interesting to try to detect very faint Halpha emission on the western side of the U-shaped nebula. Later, depending on its initial mass, the star underwent a RSG or LBV phase with a very high mass-loss rate but a very low terminal velocity. Assuming a typical expansion velocity of ~ 10--25 km/s, the wind would reach a distance of ~ 1--2.5 pc in a lifetime of ~ 105 years. Depending on the extent of the cavity on the eastern side, this material could be distributed roughly symmetrically around the star. Finally, the star entered its WR phase and blew the relatively spherical filamentary 3 pc bubble visible today. Inhomogeneities in the circumstellar material generated the non-moving, dense filaments which can be seen together with the thin membrane of the bubble. The yet unsolved question of the mass of the WR bubble is critical as the model described above would only be acceptable if the mass of the WR nebula is found to be relatively modest.

In the scenario described above the molecular gas we have detected would be part of the ambient interstellar medium swept-up by the star when it was on the main sequence (O star). The H2 could be either fluorescent or shock excited. We have not, however, detected a molecular component directly associated with the WR wind-blown bubble. This does not exclude the existence of a neutral component to it as it is possible that the hydrogen molecules have been dissociated by a shock or by the strong ultraviolet flux from the star. In this context, 21 cm HI observations would be complementary to the observations presented here and are extremely important to establish the real mass of the WR bubble.



4.2 WR nebulae as an extragalactic H2 source

It is well known that many galaxies are strong 1--0 S(1) H2 emitters. It is generally thought that the sources of this emission are diverse ranging from young stellar objects such as Orion to supernova remnants and planetary or reflection nebulae. Can H2 emission from hot star nebulae (O or WR) such as the one presented here be a significant contributor to the observed extragalactic flux? Fischer et al. (1987) and Puxley et al. (1988, 1990) have shown that UV fluorescence from hot stars is indeed a viable mechanism to explain the H2 emission in some AGNs and starburst galaxies. However, emission from an individual O or WR star had not previously been detected. Assuming the level of H2 emission detected from NGC 2359 is typical, is it sufficiently high to constitute a significant fraction of the measured extragalactic sources?

As a test, we will examine the case of the prototype WR galaxy He 2--10. This galaxy is a blue compact dwarf in which features resembling WR bands were first detected by Allen, Wright & Goss (1976). Later, Conti (1991) defined the WR galaxy class and published the first catalog with 37 members. In order to be included in this class, emission line galaxies only need to satisfy a single criterion: display broad HeII4686 emission. This indicates that these galaxies harbor a large number of WR stars.

Vacca & Conti (1992) present long-slit (1.5'' x 101 '') optical spectra of 10 WR galaxies including He 2--10. From the observed luminosity in the CIV5808 and HeII4686 lines, they estimate the number of WC and WN stars in the nucleus of the galaxy to be ~ 400 and 293 respectively. Using a method developed by Vacca (1991) based on the number of ionizing photons required to produce the observed recombination spectrum, they find that the number of main-sequence O stars is ~ 4400. The total number of hot stars in the nucleus of He 2--10 is therefore ~ 5100.

Lumsden, Puxley & Doherty (1994) present infrared spectroscopy of the nucleus of He 2--10. They find a total 1--0 S(1) H2 emission of 6 plus or minus 0.4 x 10-18 W m-2 in a 3'' x 6'' slit. For NGC2359, we find a total 1--0 S(1) H2 luminosity of ~ 4 LSun which corresponds, for a distance of 5 kpc, to a flux of 16 x 10-22 W m-2. Adopting this as a typical value, the estimated 5100 hot stars could contribute as much as 8 x 10-18 W m-2 or a value of the same order of magnitude as the total detected 1--0 S(1) H2 flux from the nucleus of He 2--10.

Because of their short evolutionary timescales, it is very likely that massive stars are still located relatively close to their parent molecular cloud. Therefore, unless the hydrogen molecules are mostly dissociated, the stars have good chances of exciting the surrounding H2 molecules either by UV fluorescence or by shocks from wind-blown bubbles. The WR stars themselves are few in number compared to O stars and therefore only contribute about 10% of the total massive star input. Clearly, a more systematic study of galactic hot stars is required to determine how widespread H2 emission is in the vicinity of WR and OB stars.



5. Conclusions

In this paper, we have presented IR line-emission observations of the WR nebula NGC 2359. We report on the first direct detection of emission from H2 gas in the vicinity of a WR nebula and propose that the observed filamentary distribution of molecular gas traces regions of higher density either on the border of the neighboring molecular cloud or in denser areas of non-dissociated H2 molecules within the region mainly composed of ionized gas. The excitation mechanism can be either ultraviolet fluorescence or shocks and the spatial distribution of the H2 is not inconsistent with either mechanism. The present data can unfortunately not determine the relative importance of the two possible excitation mechanisms because both can explain the observed level of H2 flux. Spectroscopic observations of several H2 transitions with the aim of calculating specific line ratios are required to achieve this.

In view of the relatively high level of H2 emission detected in NGC2359, hot star nebulae have the potential of contributing a significant fraction of the total H2 emission observed in emission line galaxies, particularly in those displaying recent starburst activity. It is however essential to establish the ubiquity of this phenomenon among galactic O and WR stars in order to be able to judge its importance on extragalactic scales.



Acknowledgments

We are grateful to Grant Miller and You-Hua Chu for making their optical images available to us and to Luc Turbide for help with putting together Figures 1 and 2. We wish to thank the Natural Sciences and Engineering Research Council (NSERC) of Canada and the Fonds pour la Formation de Chercheurs et l'Aide à la Recherche (FCAR) of Québec for financial support.



References

Abbott, D.C., & Conti, P.S. 1987, Annual Reviews of Astronomy and Astrophysics, 25, 113.

Allen, D.A., Wright, A.E., & Goss, W.M. 1976, Monthly Notices of the Royal Astronomical Society, 177, 91.

Arnal, E.M. 1992, Astronomy & Astrophysics, 254, 305.

Arnal, E.M., & Cappa, C.E. 1996, Monthly Notices of the Royal Astronomical Society, 279, 788.

Black, J.H., & van Dishoeck, E.F. 1987, Astrophysical Journal, 322, 412.

Cappa, C.E., Dubner, G.M., Rogers, C., & St-Louis, N. 1996, Astronomical Journal, 112, 1104.

Casali, M., & Hawarden, T. 1992, JCMT-UKIRT Newsletter, No. 4, 33.

Chu, Y.H. 1981, Astrophysical Journal, 249, 195.

Chu, Y.H. 1988, Publication of the Astronomical Society of the Pacific, 100, 986.

Conti, P.S. 1991, Astrophysical Journal, 377, 115.

Draine, B.T. 1978, Astrophysical Journals, 36, 595.

Dubner, G.M., Giacani, E., Cappa de Nicolau, C.E., & Reysono, E. 1992, Astronomy & Astrophysics, 96, 505.

Dufour, R.J. 1989, Rev. Mexicana, Astron. Astrof., 18, 87.

Esteban, C., Vilchez, J.M., Manchado, A. & Edmunds, M.G. 1990, Astronomy & Astrophysics, 227, 515.

Esteban, C., Smith, L.J., Vilchez, J.M., & Clegg, R.E.S. 1993, Astronomy & Astrophysics, 272, 299.

Fischer, J., Geballe, T.R., Smith, H.A., Simon, M., & Storey, J.W.V. 1987, Astrophysical Journal, 320, 667.

Georgelin, Y.P., Georgelin, Y.M. 1970, Astronomy & Astrophysics, 6, 349.

Goudis, C., Hippelein, H., & Munch, G. 1983, Astronomy & Astrophysics, 117, 127.

Goudis, C.D., Christopoulou, P.-E., Meaburn, J., & Dyson, J.E. 1994, Astronomy & Astrophysics, 285, 631.

Graham, J.R., Wright, G.S., & Longmore, A.J. 1987, Astrophysical Journal, 313, 847.

Graham, J.R., Wright, G.S., Hester, J.J., & Longmore, A.J. 1991, Astronomical Journal, 101, 175.

Graham, J.R., Serabyn, E., Herbst, T.M., Matthews, K., Neugebauer, G., Soifer, B.T., Wilson, T.D., & Beckwith, S. 1993, Astronomical Journal, 105, 250.

Hartquist, T.W., & Dyson, J.E. 1993, Q. J. R. astr. Soc., 34, 57.

van der Hucht, K.A., Conti, P.S., Lundstrom, I., & Stenholm, I. 1981, Space Science Reviews, 28, 227.

Johnson, H.M., & Hogg, D.E. 1965, Astrophysical Journal, 142, 1033.

Kurucz, R.L. 1979, Astrophysical Journals, 40, 1.

Lozinskaya, T.A. 1973, Soviet Astronomy, 17, 317.

Lumsden, S.L., Puxley, P.J., & Doherty, R.M. 1994, Monthly Notices of the Royal Astronomical Society, 268, 821.

Maeder, A. 1992, Astronomy & Astrophysics, 264, 105.

Marston, A.P. 1991, Astrophysical Journal, 366, 181.

Miller, G.J., & Chu, Y.H. 1993, Astrophysical Journals, 85, 137.

Nadeau, D., Murphy, D.C., Doyon, R., & Rowlands, N. 1994, Publication of the Astronomical Society of the Pacific, 106, 909.

Niemela, V.S., Cappa de Nicolau, C.E. 1991, Astronomical Journal, 101, 572.

Peimbert, M., Torres-Peimbert, S., Rayo, J.F. 1978, Astrophysical Journal, 220, 516.

Pismis, P., Recillas-Cruz, E., & Hasse, I. 1977, Rev. Mex. Astron. Astrophys., 2, 209.

Prinja, R.K., Barlow, M.J., & Howarth, I.D. 1990, Astrophysical Journal, 361, 607.

Puxley, P.J., Hawarden, T.G., & Mountain, C.M. 1988, Monthly Notices of the Royal Astronomical Society, 234, 29P.

Puxley, P.J., Hawarden, T.G., & Mountain, C.M. 1990, Astrophysical Journal, 364, 77.

Schneps, M.H., & Wright, N. 1980, Sky and Telescope, 59, 195.

Schneps, M.H., Haschick, A.D., Wright, E.L. & Barrett, A.H. 1981, Astrophysical Journal, 243, 184.

Talent, D.L., & Dufour, R.J. 1979, Astrophysical Journal, 233, 888.

Treffers, R.R., & Chu, Y.H. 1982, Astrophysical Journal, 254, 569.

Vacca, W.D. 1991, Ph.D. thesis, Univ. Colorado, Boulder.

Vacca, W.D., & Conti, P.S. 1992, Astrophysical Journal, 401, 543.

Van Buren, D. 1986, Astrophysical Journal, 306, 538.

Weaver, R., McCray, R., Castor, J., Shapiro, P., & Moore, R. 1977, Astrophysical Journal, 218, 377.


--------


Retour à la page maîtresse du groupe d'astrophysique de l'Université de Montréal