A&A 374, 682-690 (2001)
S. Ehlerová 1 - J. Palous 1 - W. K. Huchtmeier 2
1 - Astronomical Institute, Academy of Sciences of the Czech Republic, Bocní II 1401, 141 31 Prague 4,
2 - MPIfR, Auf dem Hügel 69, 53121 Bonn, Germany
Received 27 February 2001/ Accepted 18 May 2001
We describe H I observations of a field in the Milky Way centered on , made by the Effelsberg radiotelescope. The field contains one previously identified H I supershell, GS061+00+51 (Heiles 1979); apart from it we find several new structures. We also study the H I distribution in the vicinity of four H II regions, S86, S87, S88 and S89. We confirm the existence of the shell GS061+00+51, and we find that it has two smaller neighbours, spherical shells with . We identify at least one more regular shell at ; and one blown-out shell at . In two cases we are able to connect H II regions with features in the H I distribution (S86 and S87), in two other cases no connection is found. Apart from quite regular H I shells we see a number of non-coherent objects, which are probably a result of the turbulence in the interstellar medium.
Key words: ISM: bubbles - ISM: supernova remnants - ISM: H II regions - radio lines: ISM
Turbulence creates in the interstellar medium (ISM) many structures, typically dense sheets, clumps and low-density holes. The majority of these structures are transient. Many of them have an irregular, patchy appearance; however, some may look like ordinary, regular objects.
Another type of structures found in the ISM are H I shells and holes. We agree with Walter & Brinks (1999) that there is a difference between turbulent structures and H I shells, at least in the sense that most turbulent structures show very little consistency if any in the position-velocity (or position-position) space, while H I shells do. This, of course, does not mean that the turbulent medium does not influence the shape and evolution of H I shells. It does, and as a first guess we may estimate that shapes of shells in a turbulent medium will be more irregular than in a smooth medium.
We observed a field in the galactic plane which contains the supershell GS061+00+51 (Heiles 1979). Our observations have four times higher resolution than the survey of Weaver & Williams (1973) used for the previous identification. Apart from the shell GS061+00+51 and its surroundings we study the rest of the datacube and try to identify new shells and shell-like structures.
In the observed field four optical H II regions are known (S86, S87, S88 and S89; Sharpless 1959), at least one of them (S86) is connected to an OB association (Vul OB1). The angular dimensions of the mentioned H II regions are greater than or comparable to the resolution of our observations, and therefore we should be able to see their imprint in the H I distribution.
In 1997 (March-June) we observed a field in the Milky Way centered on , with the 100 m radiotelescope in Effelsberg at the frequency 1.4 GHz (21 cm) of the neutral hydrogen line. The frequency switching mode was used. The bandwidth 1.56 MHz was split into 512 channels with widths of 3 kHz, or 0.64 . The primary beamwidth of the Effelsberg radiotelescope at 21 cm is 9.4 arcmin, observations were made with a spacing of 4 arcmin (the pixel size). Each spectrum was integrated for 15 s. The data were calibrated using the standard S7 procedure (Kalberla et al. 1982) and a linear baseline was subtracted. Observations were made in 6 runs, each dataset was calibrated separately. Data were not corrected for stray radiation, because we observed a small field and were mostly interested in the differential effect of the observed emission against the background. Radio frequency interference may be present in the observational data.
To check our observations we compared them with data from the Leiden-Dwingeloo H I survey (Hartmann & Burton 1997), which had a resolution of . To match the Dwingeloo beam we averaged spectra from 81 pixels, i.e. (36 arcmin)2. A comparison between our data and the Dwingeloo survey is shown in Fig. 1, where good agreement between the two datasets is visible.
|Figure 1: A comparison between the Leiden-Dwingeloo survey (thick line) and our observations (thin line). Our spectra are artificially offset by 10 K.|
|Open with DEXTER|
To be ranked among H I shells, a structure must fulfill several criteria:
Due to the size of the field, many structures are only partly visible, and those we do not describe here (of the known H I shells GS064-0.1-97 (Heiles 1979) is seen in channel maps as a partial arc). Another previously known structure, the shell GS061+00+51, lies fully in the observed field.
|Figure 2: graph of the shell GS59.9-1.0+38. The left panel shows the spectrum through the center of the shell (solid line) and the average spectrum of the surrounding area (dotted line). The right panel shows their difference , where both walls (receding and approaching) belonging to the structure and the hole (corresponding to the swept-up region) are nicely recognizable.|
|Open with DEXTER|
The spectrum of an H I shell (in an ideal case) contains two peaks corresponding to the intersection of the line of sight with dense walls and a depression corresponding to the hole. These features are superimposed on the spectrum emitted by the surrounding ISM. To disentangle the two contributions, we subtract the background emission from the spectrum through the structure; the shell features should then appear. The column densities of the gas swept up into the wall can be - under some assumptions on the shape and dimensions of the shell - transformed into the mass of the shell and the initial volume density of the ISM.
However, to determine the background spectrum is difficult, first because of the unknowns in the velocity field and gas distribution which shape the spectrum, secondly because of the turbulent character of the ISM. The simplest way to define the background is to take the average of the emission from a region around the studied line of sight (in the case of studying the spectrum through the H I shell the region should contain the whole structure). This approach has its drawbacks, but at least it smears out the small scale inhomogeneities. When applied to artificial datacubes, we find, that often this method leads to a slight underestimation of the real values (lower column densities of the swept-up gas and lower masses). An important "spoiler'' is the non-zero velocity dispersion of the gas, both in the shell and in the ISM. How important this effect is, depends on the ratio between the velocity dispersion and the expansion velocity. The line widths of the walls correspond to the real velocity dispersion in the gas swept in the shell. The line width of the hole is not so easy to classify and so we abstain from any deductions.
It is also possible to estimate masses purely from the dimensions of the shell and an assumed (or estimated or fitted) density n0 of the ISM. This approach, due to the variability of n0 on many scales, does not lead to better or more reliable results.
To calculate kinematic distances of shells we use the rotation curve of Wouterloot et al. (1990).
The total energy
required to create the H I shell
is estimated using the Chevalier (1974) formula
|( )||( )||()|
|Figure 3: The shell GS59.9-1.0+38 in a velocity channel ( ) and the lv cut ( ). The grey scale goes from white (the lowest temperature) to black (the highest temperature). The range of temperatures in the left panel is (22.9 K, 61.5 K), contour values are 26.8 (1.9) 57.7 K; in the right panel the range is (0.5 K, 88.8 K), contour values are 9.3 (4.4) 80.0 K.|
|Open with DEXTER|
Quantities derived from the graph (see Fig. 2) are summarized in Table 1. FWHM gives the width of the line (if the line profile is Gaussian, the dispersion FWHM); is the column density of H I swept up into the wall (or missing in the hole); is the derived mass of H I swept up in walls (or missing in the hole), assuming the radius of the shell to be 35 pc.
The velocity dispersion of the gas swept into the shell is quite small (1.5-2.0 ) which is in agreement with the expected high cooling rate in dense walls.
The masses derived from walls and a hole are not the same,
but this is not very surprising, given the method and uncertainties
in deriving the background (see the section "
As a reasonable estimate we adopt the value of the total mass
where 0.7 is the solar abundance of
The shell GS59.9-1.0+38 is probably young, from the analytical solution (Sedov 1959) we estimate its expansion age as 1.5 Myr.
|( )||( )||()|
|Figure 4: The shell GS59.7-0.4+44 in a velocity channel ( ) and the bv cut ( ). The grey scale goes from white (the lowest temperature) to black (the highest temperature). The range of temperatures in the left panel is (4.5 K, 51.6 K), contour values are 9.2 (2.4) 46.8 K; in the right panel the range is (0.0 K, 91.3 K), contour values are 7.7 (4.6) 82.0 K.|
|Open with DEXTER|
GS59.7-0.4+44 (see Fig. 4) is another small spherical structure, in fact it is nearly a twin of GS59.9-1.0+38. Like GS59.9-1.0+38, GS59.7-0.4+44 lies close to the tangential point. Its radius is 24 arcmin (30 pc), its expansion velocity is 14 (for the explanation of the seemingly lower expansion velocity in the bv diagram see the previous section). Table 2 summarizes observed properties of the shell.
A reasonable mass estimate is
The age of the shell is small, only about 1 Myr.
|Figure 5: The shell GS061+00+51 in a velocity channel ( ) and the lv cut ( ). Three crosses in the left panel show centers of H I shells GS061+00+51, GS59.9-1.0+38 and GS59.7-0.4+44. The grey scale goes from white (the lowest temperature) to black (the highest temperature). The range of temperatures in the left panel is (7.0 K, 66.9 K), contour values are 13.0 (3.0) 61.0 K; in the right panel the range is (0.0 K, 102.6 K), contour values are 8.7 (5.2) 92.2 K.|
|Open with DEXTER|
This is one of Heiles' expanding shells (Heiles 1979), the only complete shell in our field which was known before. Its first detection and description can be found in Katgert (1969). A receding part of the shell is not visible. The approaching hemisphere is seen clearly, and is quite interesting. It is not a classical elliptical shell, but, especially at lower velocities, it resembles a crescent (see Fig. 5). We can think of two possibilities to explain this shape:
The properties of the shell are:
The dimensions of the structure as given by Heiles (1979) are slightly higher than our values, which is caused by 1) the fact, that the resolution of the Effelsberg radiotelescope is higher than that of the survey in which Heiles identified the shells: viz. the H I survey of Weaver & Williams (1973) with a spatial resolution of and a velocity resolution of 2 ; and 2) uncertainties in defining the precise boundaries of the shell - while there is no doubt about the existence and position of the structure, it is not completely clear, if all adjoining depressions belong to it.
Obviously (see Fig. 5), shells GS061+00+51, GS59.9-1.0+38 and GS59.7-0.4+44 are neighbours. GS061+00+51 is older and bigger than the other two, but not old enough to trigger secondary star formation in the walls, which could result in the creation of new small shells on the rim of the old structure. We may be witnessing propagating star formation in one cloud (or a cloud complex) which started at higher galactic longitudes and propagates toward the lower longitudes. The difference in ages of GS061+00+51, GS59.9-1.0+38 and GS59.7-0.4+44 is about 3-4 Myr, which suggests that the speed of the shock front compressing the gas and triggering the star formation is around 40 (this is a lower limit since we do not take into account the differences in radial distances). In a few million years the three bubbles should merge.
|( )||( )||()|
|Figure 6: The shell GS62.1+0.2-18 in a velocity channel ( ) and the bv cut ( ). The grey scale goes from white (the lowest temperature) to black (the highest temperature). The range of temperatures in the left panel is (7.3 K, 79.1 K), contour values are 14.5 (3.6) 72.0 K; in the right panel the range is (3.7 K, 92.9 K), contour values are 12.6 (4.5) 84.0 K.|
|Open with DEXTER|
This is a comparatively spherical shell in the outer Galaxy (see Fig. 6); it lies at a distance of 9.6 kpc. Its radius is 40 arcmin, or 110 pc (in the l direction it is 120 pc, 100 pc in the b direction). The expansion velocity is 13 . Only one wall is seen reliably. Table 3 summarizes properties of the shell.
One or more probably several supernovae were needed to create the shell GS62.1+0.2-18; its age is 5 Myr.
|Figure 7: The shell GS60.1-1.1-54 in a velocity channel ( ) and the bv cut ( ). The grey scale goes from white (the lowest temperature) to black (the highest temperature). The range of temperatures in the left panel is (6.3 K, 59.4 K), contour values are 11.6 (2.7) 54.0 K; in the right panel the range is (2.4 K, 56.9 K), contour values are 7.8 (2.7) 51.5 K.|
|Open with DEXTER|
The shell GS60.0-1.1-54 is a highly non-spherical structure (Fig. 7). It consists of a roughly spherical hole centered on ( ), connected with a cone which opens to the halo, closed by an arc. The shell lies in the outer Galaxy, at a distance of 13.7 kpc. Its dimensions are about (500 pc) in the b-direction, the maximum diameter in the l-direction is (400 pc). Though it is quite extended in the b-direction, it is not an object in the Koo et al. (1992) catalog of galactic worm candidates.
The H I shell GS60.0-1.1-54 is an irregular structure, however, it is probably not unique in the Milky Way. Its shape and dimensions are similar to the Aquila supershell (Maciejewski et al. 1996). For a possible scenario how to create such a structure compare the rightmost panel of Fig. 3 in Korpi et al. (1999) showing results of MHD simulations. The structure shown resembles the observations quite well, both in shape and dimensions.
The shell GS60.0-1.1-54 does not show the approaching hemisphere, i.e. it is open at one side (or the wall is negligible). The receding hemisphere is visible: the small "hole'' around changes diameter as expected from the expanding structure with an expansion velocity of 9 . The spectrum through also shows the expansion (17 ). The different expansion velocities are quite consistent with the idea that the fastest deceleration of the shell takes place in the densest part of the Galactic disk. The blown-out part at high latitudes changes shape and dimensions with velocity, though not in a very regular way. The best estimate of the expansion velocity is 9 . Table 4 gives the column densities in different positions inside the shell.
The shell GS60.0-1.1-54 is very irregular and therefore we have not estimated its energy, as this is very unreliable.
|Figure 8: H II regions in the observed field. The pixel map is the H I column density between with the S86 region (left panel); H I in with regions (from left to right) S87, S88 and S89 (right panel). Positions of H II regions are indicated by big crosses and circles corresponding to dimensions of regions (S89 is very small, with the diameter only barely exceeding the size of the pixel, it lies near to the edge of the map at the upper cross-section of the lines). The grey scale goes from white (the lowest column density) to black (the highest density). The range of densities in the left panel is ( , ), contour values are ( ) ; in the right panel the range is ( , ), contour values are ( ) .|
|Open with DEXTER|
There are four optical H II regions in the observed field; S86,
S87, S88 and S89 (Sharpless 1959). We examine the distribution
of the neutral hydrogen in their vicinity. Table 5 gives
the properties of the four regions taken from Blitz & Fich (1982).
|name||l||b||( )||d (arcmin)|
This H II region is associated with the Vul OB1 association (NGC 6823). At the position of S86 there is a clear hole in the H I distribution, visible between . The H II region lies inside the hole, its dimensions are comparable to dimensions of the hole (see Fig. 8). The hole is stationary.
The coincidence of the H I hole and the H II region is consistent with the idea, that most gas in the vicinity of the OB association is ionized and therefore not observed in 21 emission. The hole does not expand, which may mean that no SN has exploded so far in the cluster (which is consistent with age estimates of NGC 6823: 2-7 Myr; Massey et al. 1995).
S87 is a source observed in optical, infrared, radio recombination lines (RRL) and molecular line emission (Barsony 1989; Onello et al. 1991). It has a compact core surrounded by an extended structure oriented south-east (i.e. perpendicular to the galactic plane). It interacts with a molecular cloud.
The H II region S87 lies inside the hole in the H I distribution, visible between . Again, this hole is stationary.
S88 is also observed in RRLs, molecular line emission, infrared and optical (Wood & Churchwell 1989; Onello et al. 1991). The region has an ultracompact core with a complex, multi-peaked structure.
S88 probably lies at the boundary between a dense sheet of gas and a more rarefied medium. At the position and the radial velocity of the region there is a small hole visible in a few velocity channels around , but definitely not as pronounced as in the case of S86 or S87. This hole is a part of the bigger empty region (see Fig. 8).
S89 lies in a dense region (see Fig. 8). It is not situated inside any hole, at least not in the predicted velocity range, but it lies just on the edge of a small hole, visible between . The physical association of these two structures, an H II region and an H I hole, is unclear, but cannot be excluded.
In two out of four cases (S86, S87) we find a clear trace of the Strömgren sphere in the H I distribution, i.e. a stationary hole. In one case (S88) the connection H II region - H I hole is not very obvious - there are depressions at the position of the region, but nothing really convincing. Maybe simply the gas distribution in the vicinity of S88 is so chaotic, that the nice Strömgren sphere does not exist. The region S89 does not lie inside a hole, but on the edge of one.
The chance coincidence of unrelated H II regions and H I holes cannot be excluded, because of the distance ambiguity, but at least for S86 and S87 the probability of this coincidence is small, as not only the positions and radial velocities, but also the dimensions of H II regions and H I holes agree.
The area where all these H II regions lie, i.e. , is a very turbulent region, full of structures on many scales (in Ehlerová 2000, it was described as a strange kind of a complex, multicomponent H I shell GS60.1-0.3+15). This is partly the reason why none of the H I holes mentioned was identified as an independent H I shell.
The field contains a rich variety of structures. Due to its limited size, selection effects play heavily against any statistical or general considerations and we can only describe individual structures.
Summing up, it seems that there are two types of "shell-like'' structures found in the H I distribution. The first, formed by consistent structures that are coherent in the position-velocity space, is less abundant than the second type, which contains non-coherent objects. We believe that these second type structures are created mainly due to the turbulence in the ISM. We identify the first group of objects with structures known as H I shells, as they fulfill the usual criteria put on shells. This is good news concerning the existence of H I shells. The bad news is the fact that there is no well-defined boundary between the two types of structures.
Authors gratefully acknowledge financial support by the Grant Agency of the Academy of Sciences of the Czech Republic under the grant No. A3003705/1997 and support by the grant project of the Academy of Sciences of the Czech Republic No. K1048102. SE would like to thank MPIfR for the hospitality during her stay in Bonn.