In order to test the second scenario we used XMM-Newton and observed the X-ray emission of three edge-on galaxies known to possess extended diffuse ionized gas (DIG) and radio continuum halos and search for associated star formation induced hot X-ray halos. This, together with previously obtained multi-wavelength observations would allow us to check the infall vs. the outflow model.
In addition, the large collecting area and the enhanced sensitivity of XMM-Newton would reliably constrain essential parameters of the ISM, such as the temperature and the filling factor of the hot ionized medium.
So the main issue is to figure out: "What causes the formation of gaseous X-ray halos in non-starburst, but actively star forming spiral galaxies?"
It is commonly believed that supernova (SN) explosions and/or stellar winds from young and hot stars are the main trigger mechanisms. Theoretical models predict an intense transport of gas and momentum between the disk plane and the halo (so called disk-halo interaction). Collective SNe explosions within star forming regions can produce overpressured flows of hot X-ray emitting gas that are driven off the plane far into the halo. This process gives rise to phenomena known as "galactic fountains" (de Avillez 2000) or "chimneys" (Norman & Ikeuchi 1989). The results of the violent energy input can be seen in different wavelength regions, such as in the radio or in the optical. Superbubbles and filaments as well as extended worm structures have been observed in several galaxies, including the Milky Way (e.g., Heiles 1992, Koo et al. 1992, Reynolds et al. 2001, Tüllmann et al. 2003).
Although the conceptual basis seems to be well understood, direct observational evidence whether these processes may also trigger the creation of a X-ray halo is still scarce. Very recent cosmological models, however, claim a completely different mechanism, namely accretion from the IGM with subsequent infall of hot gas onto the disk plane (Benson et al. 2000, Toft et al. 2002). These models can account for extended star formation histories of isolated spiral galaxies and also give an answer to the so called ``G-dwarf problem'' (e.g., Rocha-Pinto & Maciel 1996).
The most apparent disadvantage of both scenarios is their lack of observational evidence, especially in the X-ray regime. Since our group has been working extensively on star formation induced outflows in edge-on galaxies, we want to test this model approach in detail and establish the missing link between optical, radio and X-ray halos of normal star forming galaxies.
Independent evidence for an interstellar disk-halo connection comes from the presence of a thick layer (1.5 - 10kpc) of ionized hydrogen, called "diffuse ionized gas" (DIG). This gas phase is known to be blown out into the halo by correlated SNe and was first discovered in the edge-on galaxy NGC891 over a decade ago (Rand et al. 1990, Dettmar 1990, for a review see Dettmar 1992).
During recent years significant progress could be achieved in this field, e.g., by carrying out a comprehensive Halpha survey of 74 edge-on spirals, by estimating the ejected DIG mass, and deriving an empirical set of parameters which indicates the presence of prominent DIG halos (Rossa & Dettmar 2000, Rossa & Dettmar 2003). As can be seen from Fig. 6.1, it is the star formation rate (SFR) which determines whether starburst but also normal star forming galaxies possess DIG halos or not. As the SFR is related to the SN rate this would directly support the outflow model.
Fig. 6.1: Diagnostic DIG diagram for our survey galaxies with IRAS detections, plotted together with the starburst sample of Lehnert & Heckman (1995). The horizontal dashed line marks the threshold for IRAS warm galaxies (log(S_60/S_100)> -0.4). There is a clear threshold above which starburst and non-starburst galaxies possess pervasive DIG halos (Figure courtesy J. Rossa).
Moreover, it could be shown that DIG (HII), radio continuum emission (HI), and magnetic field vectors are well aligned in the halos of some star forming spirals (e.g., Dahlem et al. 1997, Tüllmann et al. 2000). In an ongoing analysis this correlation could be verified for at least 13 of the survey galaxies for which radio data is available Rossa & Dettmar 2003).
Finally, recent studies claim not only a correlation between DIG and HI but also with X-rays (e.g., Wang et al. 2001, Read & Ponman 2001). Interestingly, each gas component seems to trace the others if just a certain star formation intensity is exceeded. For starburst galaxies, this correlation could be established making extensive use of the Chandra satellite (Strickland et al. 2004). However, for normal star forming spirals it is still lacking.
In order to trace this correlation down to lower X-ray luminosities (similar to Fig. 6.1) and thus allow to distinguish between SN or infall induced X-ray halos, new sensitive observations of extended X-ray gas in halos of non-starburst (but still actively star forming) spiral galaxies are required. Although ROSAT contributed significantly to the field with studies of NGC891 (Bregman & Pildis 1994), NGC4631 (Wang et al. 1995), and NGC4666 (Dahlem et al. 1997), clear evidence is still scarce.
It was argued that the presence of all these components of the ISM (DIG, HI, and X-rays) in halos is related to the SFR of the underlying disk (Dettmar 1992, Dahlem et al. 1994, Rand 1996, Rossa & Dettmar 2003). However, it is not clear why this empirical coincidence of different gas phases occurs and holds. Is this a consequence of different cooling times and densities? From two examples it becomes obvious that the detailed physics of the involved processes is only poorly understood:
(1) The global energetics of the X-ray halo phenomenon is highly dependent on the temperature of the gas. This temperature, however, is not at all well constrained from the ROSAT mission (e.g., (Wang et al. 1995) derive a two temperature model for NGC4631) and one may even expect a non-equilibrium situation (Breitschwerdt & Schmutzler 1994) which would lead to much more complex spectra than current "standard" models allow for.
(2) due to the limited spatial resolution and sensitivity of previous X-ray observations it is unclear whether the X-ray halos are fed by individual sources (such as giant OB associations) or whether a wide spread hot medium is driving a wind on more global scales.
Fig. 6.2: X-ray data of NGC891, NGC3044, NGC3221, NGC3628, NGC3877, NGC4631, NGC4634, NGC4666, and NGC5775 obtained with XMM-Newton. Red colors address the energy range from 0.2-1.0keV, green denotes energies between 1.0 and 4.5keV, while bluish colors represent the hard energy band from 4.5-12keV. The yellow ellipse indicates the extent of the Halpha-disk. Except NGC3044 and NGC3877, all these galaxies possess a prominent hot ionized gas halo as evidenced by the extended emission in the soft energy band.
All scientific results are discussed in two papers by Tüllmann et al. (2006a,b).
X-ray images of NGC3628, NGC4631, NGC4666, NGC4634, and NGC5775 are available as high-resolution versions at
the XMM-Newton Image Gallery.
Please mind also our joined ESA and A&A press release from December 14, 2005.