INTRODUCTION
DIG MORPHOLOGY
DIG AND OTHER GASEOUS HALOS
EMISSION LINE RATIOS AS DIAGNOSTICS OF IONIZATION AND
HEATING
REFERENCES
Published in Publications Astronomical Society of Australia, vol. 15, no. 1, p. 106-110
Abstract. We review observations of diffuse ionized gas (DIG) in edge-on spiral galaxies. Deep imaging has revealed a variety of morphologies for "extraplanar" (above the HII region layer) or "halo" DIG: some show widespread prominent layers of truly diffuse gas and filamentary structures, others show just one or a few patches of extraplanar emission, while still others show no detectable extraplanar DIG at all. The key galactic property which appears to govern the prominence of the DIG layer is the star formation activity. Smaller scale connections between DIG and local star formation activity can be seen in some of the galaxies.
DIG halos are also related to radio continuum, X-ray and HI halos. Those galaxies with the most prominent DIG layers also have the most prominent radio halos. In NGC 891, there are spatial correlations between extraplanar DIG, radio emission, X-ray emission and HI, and these in turn are related to the underlying disk star formation activity.
Progress on identifying the sources of ionization and heating of the gas has come from spectroscopy, imaging in different emission lines, and Fabry-Perot observations. By comparing with theoretical predictions, much evidence is found for photo-ionization by disk stars as the primary ionization and heating mechanism, but recent observations are beginning to point towards the need for additional sources of ionization and/or heating.
Keywords: galaxies: ISM - galaxies: halos - galaxies: spiral - galaxies: structure - radiative transfer
INTRODUCTION
DIG MORPHOLOGY
DIG AND OTHER GASEOUS HALOS
EMISSION LINE RATIOS AS DIAGNOSTICS OF IONIZATION AND
HEATING
REFERENCES
With the advent of the AAO-UKST Galactic Plane
H
survey and
the several other Galactic surveys discussed in this volume, an
opportunity arises for progress in understanding the origin,
structure, and ionization of DIG through the comparison of these
surveys with observations of external edge-on galaxies. These
galaxies offer the advantage of allowing the entire DIG layer to be
quickly imaged (albeit at relatively low linear resolution), while
avoiding the difficulty inherent in Milky Way observations of
translating angular heights of emission to physical heights. Studies
of the diffuse ionized layer of the Milky Way will allow, for example,
the vertical structure of the layer and the dependence of line ratios
on height from the Galactic plane and on the underlying disk
environment to be compared with results from external edge-ons. The
importance of DIG as an ISM component and the leverage it provides on
understanding the ISM as a whole is discussed elsewhere in this volume
by Reynolds, Ferrara, Elmegreen and others. The goal of this review
is to summarize our understanding of DIG in edge-on galaxies at the
time when these surveys are getting underway.
NGC 891, having an optical appearance similar to the Milky Way, and being nearby (D = 9.5 Mpc) and almost perfectly edge on (Swaters 1994), was the first target in the search for extraplanar DIG layers (Rand, Kulkarni, & Hester 1990; Dettmar 1990). There, a bright, widespread DIG halo was found, showing truly diffuse emission as well as several quasi-vertical filaments of height 2-3 kpc. The halo could be traced up to z = 3.5 kpc, but recent deep spectroscopy (Rand 1997; see below) shows that it extends to at least z = 5.5 kpc. Furthermore, the spectra confirm the earlier result that there are two vertical components to the emission, with electron density scale heights (assuming that the gas temperature and filling factor are constant with z) of 1 kpc and 5-6 kpc. The emission is concentrated to within about R = 8 kpc of the center of the galaxy, with the brightest region at about R = 5 - 8 kpc on the north side and a rather sharp cutoff beyond this region. Disk star formation, as traced by [CII] 158µm emission (Madden et al. 1993), shows a radial profile with very similar asymmetries (Rand 1994).
Almost without exception, the
H filaments have their footprint
in bright HII regions in the disk
(Rand 1994).
Some are quite narrow and may be walls of chimneys
(Norman & Ikeuchi 1989),
others are broader and may be "ionization cones" : regions where
relatively exposed O and B stars have been able to ionize large
volumes of diffuse gas above and below the midplane
(Miller & Cox 1993,
Dove & Shull 1994).
In the case of the former, the connection
with HII regions suggests that star formation persists in these ~ 1 kpc
regions of disk for at least as long as the ~ 107 yr
dynamical time
(Norman & Ikeuchi 1989)
to produce the chimneys.
This connection between filaments and HII regions is also seen in
NGC 2188
(Domgörgen, Dahlem,
& Dettmar 1996)
and NGC 55
(Ferguson, Wyse, &
Gallagher 1996;
Wyse, this volume).
Since these observations, over a dozen edge-ons have been observed
with similar sensitivity, with the result that bright, extended DIG
layers as in NGC 891 are the exception rather than the rule
(Table 1).
Other galaxies with bright layers include NGC 4631 (although only
above the central few kpc;
Rand, Kulkarni & Hester
1992)
NGC 5775
(Dettmar 1992),
and NGC 3079
(Veilleux, Cecil, &
Bland-Hawthorn 1995).
However, most galaxies seem to show only one or a few isolated
patches or filaments of emission above the HII region layer, while
some show no extraplanar DIG at all
(Rand 1996).
The prominence of
the extraplanar DIG emission is related to the star formation activity
in the disk. Insofar as far infrared luminosity (LFIR)
is a
tracer of star formation activity (debate continues on what fraction
of the emission comes from dust heated by the ISRF; however, one
cannot use H emission as a
star formation tracer because of
the severe extinction problem), then the galaxies with high surface
densities of star formation (LFIR /
D252) generally have the
brightest and most prominent extraplanar DIG layers
(Rand 1996).
This general connection is strengthened by the afore-mentioned correlation
of halo DIG and disk [CII] emission profiles in NGC 891 and the
filament-HII region associations, and it is probably more accurate to
say that active star-forming regions tend to have associated
extraplanar DIG.
|
Current directions in this area include attempts at better characterization of the background emission to allow heavier smoothing, thus making possible searches for emission at much fainter levels. Examples are the "shift and stare" technique (Donahue, Aldering, & Stocke 1995) and "charge shuffling" (Bland-Hawthorn, this volume). In this way, one can hope to determine the extent of halos at low column densities and their possible relevance as QSO absorption line systems.
The galaxies with the brightest extraplanar DIG layers also tend to
have bright and extended radio continuum halos
(Rand 1996).
Examples are NGC 891 (e.g.
Dahlem, Dettmar, & Hummel
1994),
NGC 4631 (e.g.
Golla & Hummel 1994),
and NGC 5775
(Dettmar 1992;
Duric, Irwin, & Bloemen
1996).
At the other end of the scale, NGC 4244, with no
indications of extraplanar DIG
(Walterbos & Braun
1996),
was not detected at all in radio continuum
(Hummel, Sancisi, & Ekers
1984).
Dettmar (1992)
reports an association between filaments of radio and
H
emission in the halo of NGC 5775. One might expect an
H-radio
correlation if magnetic fields associated with the thermal gas serve
to confine the cosmic-ray electrons as they try to escape the disk
(N. Duric, pers. comm.). Observational connections between radio halo
and star formation distributions in edge-ons has also been discussed by
Dahlem, Lisenfeld, &
Golla (1995).
Most of the galaxies in Table 1
have not been observed with sufficient sensitivity and resolution to
search for a radio halo.
Diffuse X-ray halo emission is also difficult to detect and has been
searched for in only a few of these edge-ons with ROSAT.
NGC 891
(Bregman & Pildis 1994;
Bregman & Houck 1997)
and NGC 4631
(Wang et al. 1995)
are notable for having relatively bright X-ray halos
occupying about the same radial range as the
H halos.
However, this trend is not perfect: patches of diffuse halo X-ray
emission have been found in NGC 4565
(Vogler, Pietsch, &
Kahabka 1996),
which shows no extraplanar DIG.
NGC 891 has a very thick disk of 21-cm emission, and
a careful study
of the origin of this high-z gas has been carried out by
Swaters (1994).
The kinematic information allows inner-disk high-z gas,
which may be accelerated by supernova activity, to be distinguished
from outer-disk gas, which may be due to warps and flares. The result
is that the most prominent inner-disk high-z emission is coincident
with the brightest region of halo DIG and radio emission, sitting
above the brightest [CII] 158 µm emission in the disk (all at
about R = 5 - 8 kpc on the north side). This HI is detected up to 2.5
kpc above the disk. There does not seem to be a good correlation
between individual HI and
H filaments, however,
although the
lower (20") resolution of the HI may hamper the comparison. Other
bright star forming edge-ons (NGC 3079, NGC 5775) show HI arcs, shells
and filaments extending several kpc from the plane
(Irwin & Seaquist 1990;
Irwin 1994).
Connections between these and
H features
are not yet clear.
The emergent scenario, then, which needs to be tested further, is that star formation activity, when sufficiently intense, leads to the elevation of gas from the thin, star-forming layer, through chimneys and superbubbles, leading to halo DIG, HI and X-ray emission. Meanwhile, diffusion and possibly localized regions of convection of cosmic rays produced in supernovae (e.g. Duric, Irwin, & Bloemen 1996) produce cosmic ray halos, traceable through radio continuum emission.
In the Reynolds layer, photo-ionization by massive stars in the disk
is considered the most likely primary ionization mechanism (Reynolds,
this volume). The available energy easily satisfies that required to
maintain the layer, and photo-ionization models [e.g.
Mathis 1985;
Sokolowski 1994
(partially reproduced in
Bland-Hawthorn, Freeman,
& Quinn 1997)]
have been the most successful in explaining the behavior
of line ratios (enhanced [SII]
6716, 6731 /
H,
[NII]6583 /
H, very weak [OI]
6300 /
H, [OIII]
5007 /
H relative to HII
regions are due to dilution of the stellar radiation as it propagates
away from the Galactic plane). However, there are observations that
these models cannot explain - one of which is the weakness of HeI
5876 /
H
(Reynolds & Tufte 1995;
Reynolds, this volume), a ratio with a fairly straightforward
interpretation in terms of the hardness of the ionizing radiation [see also
Heiles et al. (1996)
for an even more extreme result from radio recombination
lines]. The He I emission is much weaker than expected from models
given the high [NII] /
H and [SII] /
H values [see
Domgörgen & Mathis
(1994)].
The forbidden lines are
highly temperature sensitive and their interpretation is complicated
by issues of abundances, gas heating, and depletion of important
coolants such as Fe, Si, and Ca
(Sokolowski 1994).
But regardless of the forbidden lines, the low HeI /
H is problematic in itself
and may indicate that we do not understand stellar atmospheres in the
extreme UV well enough (see e.g.
Cassinelli et al. 1995),
and/or which spectral types of massive stars
are most important for the ionization
(Reynolds & Tufte 1995;
Heiles et al. 1996).
It is also not clear to what degree other sources of ionization contribute: shocks (which may play a role in irregulars; Martin 1996), turbulent mixing layers (Slavin, Shull, & Begelman 1993), cooling fountain gas (Shapiro & Benjamin 1993) and decaying neutrinos (e.g. Sciama 1995).
In NGC 891, the run of [NII] /
H vs. z at R =
5 kpc on the north side has been measured by
Rand (1997).
This shows a smooth increase with z, from 0.4 in the midplane to
1.4 at z = 2 kpc (east
side of midplane) and z = 4 kpc (west side). Previous, less sensitive
spectroscopy by
Dettmar & Schulz (1992)
and Keppel et al. (1991)
also showed an increase with z for [NII] /
H as well as
[SII] / H. The same trend in
[SII] / H was also seen
in narrow-band images
(Rand, Kulkarni, & Hester
1990).
The spatial
behavior is as expected in photo-ionization models (the ratios rise
with distance from the disk ionizing sources as the radiation field is
diluted) but values as high as 1.4 are difficult to explain
(Sokolowski 1994).
Note that in M31 (Walterbos, this volume) and the
Milky Way, [NII] /
H is roughly constant at
only 0.3-0.5, and well modeled by
Domgörgen & Mathis
(1994).
It is not
clear whether the lower value is due to a less dilute radiation field,
or whether features of some models of Sokolowski designed to raise the
value, such as a very high upper IMF cutoff, do not apply in these cases.
HeI is detected in NGC 891, but, as in the Reynolds layer,
the weakness of HeI /
H implies a much softer
spectrum than indicated by [NII] /
H, and also softer than
expected for a population of ionizing stars
(Rand 1996).
A measurement of the gas
temperature would help in understanding the forbidden line strengths,
but only upper limits exist from the non-detection of the
[NII]5755 line: 13,000 K
(east side) and 10,000 K (west side).
Veilleux et al. (1995)
have formed an [NII] /
H map of
NGC 3079, where again there is a general increase
with distance from HII regions, reaching similar values as in NGC 891.
Golla, Dettmar, &
Domgörgen (1996)
present long-slit
spectra of NGC 4631, and again find an increase of [NII] /
H
and [SII]6716 /
H height from the
plane. However,
the rate of increase of the two ratios with z is about the same
(both running from a minimum of about 0.13 in the plane to a
maximum of about 0.5 at z = 1 kpc in their slit position
"C"), whereas their models predict a stronger trend in
[SII] / H than in [NII] /
H (because of a more
dramatic change in the predominant ionization state of S between HII
regions and diffuse gas). The solution of this discrepancy is not
clear, but again it points towards a departure from existing models.
Finally,
Ferguson, Wyse, &
Gallagher (1996)
present H+[NII], [SII] and [OII]
3727 images of
NGC 55,
and find that both [SII] /
H+[NII] and
[OII] / H+[NII] increase
with height from the plane (like N
and S, O becomes increasingly singly ionized with greater distance
from the HII regions). The latter ratio shows a very strong contrast,
from about 0.1 in the midplane to as high as 2 in the halo. Again,
such high values are hard to explain.
Clearly, more work needs to be done to constrain the sources of ionization and heating. One approach is to model a combination of photo-ionized gas with shock-ionized gas using various shock speeds, and turbulent mixing layers of various temperatures, as has been done for the much brighter diffuse emission of irregulars by Martin (1996). While such composite models may be able to explain the observations, one still desires a physical scenario which joins these several processes together. For instance, in a model where the diffuse halos are fed by superbubble and chimney activity, what sort of shocks are predicted to run through the halo gas as a result, what sort of mixing interfaces should exist, and how might line ratios change with environment?
As for extra sources of non-ionization heating, possibilities include
photoelectric heating from dust grains
(Reynolds & Cox 1992)
and the dissipation of turbulence
(Minter & Balser 1997).
The latter authors
include a heating rate from turbulence derived from scintillation
observations in a photo-ionization model of the DIG and can thereby
explain simultaneously [SII] /
H, [NII] /
H, and
[HeI] / H at low-z in
the Reynolds layer if the composite
stellar temperature is low enough.
Acknowledgements
I would like to thank the organizers for a stimulating and well-run workshop, and the referee, R. Walterbos, for helpful comments.
