arXiv:0801.1069v1 [astro-ph] 7 Jan 2008
Vertically Extended Neutral Gas in the Massive Edge-on Spiral
Richard J. Rand
Department of Physics and Astronomy, University of New Mexico, 800 Yale Blvd, NE,
Albuquerque, NM 87131
Robert. A. Benjamin
Department of Physics, University of Wisconsin at Whitewater, 800 West Main Street,
Whitewater, WI 53190
We present Very Large Array 21-cm observations of the massive edge-on spiral
galaxy NGC 5746. This galaxy has recently been reported to have a luminous
X-ray halo, which has been taken as evidence of residual hot gas as predicted
in galaxy formation scenarios. Such models also predict that some of this gas
should undergo thermal instabilities, leading to a population of warm clouds
falling onto the disk. If so, then one might expect to find a vertically extended
neutral layer. We detect a substantial high-latitude component, but conclude
that almost all of its mass of 1.2 − 1.6 × 109M⊙most likely resides in a warp.
Four features far from the plane containing about 108M⊙are found at velocities
distinct from this warp. These clouds may be associated with the expected infall,
although an origin in a disk-halo flow cannot be ruled out, except for one feature
which is counter-rotating. The warp itself may be a result of infall according to
recent models. But clearly this galaxy lacks a massive, lagging neutral halo as
found in NGC 891. The disk HI is concentrated into two rings of radii 1.5 and
3 arcminutes. Radial inflow is found in the disk, probably due to the bar in this
galaxy. A nearby member of this galaxy group, NGC 5740, is also detected. It
shows a prominent one-sided extension which may be the result of ram pressure
– 2 –
Subject headings: galaxies: ISM — galaxies: spiral — galaxies: kinematics and
dynamics — galaxies: formation — galaxies: evolution — galaxies: individual
(NGC 5746, NGC 5740)
Gaseous thick disks or halos in spiral galaxies hold promise for answering key ques-
tions about galaxy formation and evolution. These vertically extended layers are multi-
phase, with detailed studies of the nearest edge-on galaxies revealing halos of neutral hydro-
gen (e.g. Irwin 1994; Swaters et al. 1997), diffuse ionized gas (DIG; e.g. Rand et al. 1990;
Dettmar 1990; Rossa & Dettmar 2003b), hot X-ray emitting gas (e.g. Bregman & Pildis
1994; Strickland et al. 2004; T¨ ullmann et al. 2006b), radio continuum emission (e.g. Irwin et al.
1999; Dahlem et al. 2001) and dust (e.g. Howk & Savage 1999; Alton et al. 2000; Irwin & Madden
Especially for the extended DIG, X-ray, dust and radio continuum components, it has
been well established that their prominence is correlated with the star formation activ-
ity in the underlying disk (e.g. Rand 1996; Rossa & Dettmar 2003a; Howk & Savage 1999;
T¨ ullmann et al. 2006a; Dahlem et al. 2006). Large shell-like and vertically oriented filamen-
tary structures are also seen in many of the more actively star-forming edge-ons, in, e.g., DIG
(Rand et al. 1990) and HI (e.g. Lee et al. 2001). All of this evidence has led to a picture of the
origin of these layers in a star-formation-driven disk-halo cycle, which has been theoretically
modeled as a general galactic fountain flow (Shapiro & Field 1976; Bregman 1980) with later
models incorporating the fact that mass and energy input into the halo may efficiently occur
through localized structures such as supershells and chimneys (Norman & Ikeuchi 1989).
As for the Milky Way, the WHAM survey (Haffner et al. 2003) has characterized in
detail the so-called Reynolds layer of vertically-extended DIG (Reynolds et al. 1973), reveal-
ing further instances of possible superbubbles (Reynolds et al. 2001). Very relevant to this
paper are the High Velocity Clouds (HVCs) and Intermediate Velocity Clouds (IVCs). For
the latter, a few are known to be 0.3–4 kpc above the midplane, with metallicities close to
solar, and it is quite possible that these clouds originate in a disk-halo flow (Wakker 2004,
and references therein). The situation for the HVCs is different, however. Not considering
the contribution from the Magellanic Stream (Putman et al. 2003), some well studied com-
plexes are found to be many kpc from the plane (e.g. 8–10 kpc for part of complex A, >4
kpc for complex C; Wakker 2004 and references therein). Although information is scarce,
– 3 –
their metallicities are also well below solar (e.g. Collins et al. 2007), suggesting that such
clouds are not part of a disk-halo flow but may be extragalactic clouds that are mixing with
metal-enriched gas in the halo as they fall onto the disk.
Such a population of infalling warm clouds is a prediction of recent models of galaxy
formation, in which galaxies are still growing by this mechanism. The idea that galaxies form
by gas cooling out of a shock-heated hot halo was presented by White & Rees (1978). This
hot gas should be thermally unstable and given to fragmentation (Field 1965; Fall & Rees
1985; Murray & Lin 1990). Maller & Bullock (2004) explored the consequences of such an
inflow of fragmenting material. The remaining hot gas has such a low density that it can
support itself against infall for a long time. This may explain the ”over-cooling” problem
in galaxy formation models that leads to overly massive galaxies and the finding that most
of the baryons in the universe never ended up in galaxies (Benson et al. 2003 and references
therein). Instabilities in cooling inflows have also been studied by Kaufmann et al. (2006)
and Sommer-Larsen (2006). The cloud populations and distributions in these high-resolution
models vary, with clouds having an uncertain but probably high ionization fraction.
An alternative explanation to the ”over-cooling” problem in the most massive galaxies
is feedback. Reheating of the halo gas by core-collapse supernovae or a (possibly recurrent)
AGN prevents the gas from cooling and flowing in (Binney 2004). Heating by Type Ia
supernovae is considered by Wang (2005). It is also argued that, except in the most massive
halos, infalling gas never heats to the virial temperature but flows in cold (Binney 1977;
Birnboim & Dekel 2003; Binney 2004; Kereˇ s et al. 2005; Dekel & Birnboim 2006).
The rotation of gaseous halos may also provide clues as to their origin and the physical
processes occurring within them. In recent years, the manner in which rotation speeds change
as a function of height has begun to be characterized in the DIG (Heald et al. 2006a,b, 2007;
Oosterloo et al. 2007). Heald and coworkers have measured the gradient in rotation speed
with height (dVrot/dz) in three edge-ons which form a decreasing sequence of star forming
activity and DIG scale-height: NGC 5775 (–8 km s−1kpc−1), NGC 891 (–17 km s−1kpc−1),
and NGC 4302 (–30 km s−1kpc−1). The gradients, when expressed in terms of km s−1per
DIG scale-height, show much less range: –15 to –25 km s−1(scale height)−1. The authors
show that simple ballistic models of disk-halo flow (Collins et al. 2002) greatly underpredict
the magnitude of these gradients and also predict the wrong trend with scale-height. Two
possible resolutions are: 1) the physics of disk-halo flows are not well described by ballistic
models and that hydrodynamical effects such as pressure (Barnab` e et al. 2006) or magnetic
or viscous forces (Benjamin 2002) dominate the rotation, or 2) part or all of the gas does
not originate in a flow from the high-angular-momentum disk, but rather from infalling low-
angular-momentum primordial gas. For the DIG, the latter explanation must also account
– 4 –
for morphological and other connections with ongoing star formation discussed above.
For NGC 891, dVrot/dz has also been measured for the HI and is –15 km s−1kpc−1
(Oosterloo et al. 2007), in good agreement with the DIG value . The authors conclude from
this gradient that at least some of the 1.2 × 109M⊙HI halo must be accreted, low angular
momentum material (see also Fraternali & Binney 2006).
Furthering our understanding of both galactic infall and disk-halo cycling, as well as
possible interactions between the two, would be made easier if one could isolate galaxies
where one or the other origin is expected to obtain. High star-forming galaxies like NGC
5775 presumably have gaseous halos dominated by disk-halo flows. The recent discovery of a
bright ( Lx= 7.3±3.9×1039erg s−1) X-ray halo around the massive, nearby (29.4 Mpc is the
commonly adopted distance), low star forming edge-on Sb galaxy NGC 5746 (Pedersen et al.
2006; Rasmussen et al. 2006) presents a challenge to disk-halo models as the X-ray luminosity
clearly exceeds that expected (T¨ ullmann et al. 2006a; Strickland et al. 2004) for a galaxy
with no detected DIG halo (Rand 1996; in an image with an Emission Measure noise level of
3.7 pc cm−6) and little star formation. Rand (1996) roughly characterized the star formation
rate per unit disk area of many edge-ons using the far infrared luminosity measured by the
Infrared Astronomical Satellite (IRAS) divided by the optical disk area: LFIR/D2
calculated from the NASA/IPAC Extragalactic Database1). NGC 5746 is one of the lower of
the 16 edge-ons thus characterized, at 4 × 1039erg s−1kpc−2. Rather, the X-ray luminosity
puts NGC 5746 on the expected steep relationship Lx∝ V7
models (Toft et al. 2002), and, in such an interpretation, it is only because of the very high
rotation speed (measured in this paper to be about 310 km s−1) that this residual hot halo
can be detected at all. The metallicity of the hot gas is found to be low, at about 0.04 solar
(Rasmussen et al. 2006), but is uncertain and could be biased toward low values (see Buote
2000). But overall, this galaxy may well be an attractive test case where disk-halo cycling
rotfor hot halos in galaxy formation
There are some caveats to this interpretation, however. First, NGC 5746 is in a group
of 26 cataloged members (Giuricin et al. 2000), and there may well have been interactions
in the past leading to gas in the group environment. Second, although there is little ongoing
star formation, it may be that Type Ia supernovae or an AGN create a hot wind or provide
energy to keep a pre-existing halo heated, as discussed above. However, Rasmussen et al.
(2006) argue that the likely Type Ia SN rate in a galaxy like NGC 5746 would be far too low to
explain the X-ray luminosity. They also argue that there is little evidence for at least current
1The NASA/IPAC Extragalactic Database (NED) is operated by the Jet Propulsion Laboratory, Cali-
fornia Institute of Technology, under contract with the National Aeronautics and Space Administration.
– 5 –
nuclear activity in this galaxy, and that a soft, thermal X-ray halo would not be expected
for an AGN outflow anyway. We do note, however, that NGC 5746 is classified as a LINER
by Carrillo et al. (1999), and there is a compact X-ray source (Gonz´ alez-Mart´ ın et al. 2006)
at the center, indicating at least low level activity at present, although no compact radio
source has been detected (Nagar et al. 2005). If the metallicity of the hot gas is indeed as
low as 0.04, these alternative sources are not very likely.
If this X-ray halo is a residual of galaxy formation and thus an indication of missing
baryons, then an obvious question is whether the predicted thermal instabilities are occur-
ring, leading to an infalling warm component. If it exists, the lack of a DIG halo may simply
mean that there are few disk sources capable of significant ionization. But there may be a
neutral halo. Either a detection or an upper limit will constrain galaxy formation models
and the “missing baryon” question, and shed light on the origin of the Milky Way’s HVCs,
more so if individual clouds can be seen. A detection of a significant neutral halo would also
be challenging to explain in terms of a disk-halo cycle origin.
We have therefore observed NGC 5746 in 21-cm emission with the Very Large Array
(VLA), as described in §2. We analyze high-latitude emission in §3, and end with a brief
discussion in §4 in terms of the theories of the halo gas discussed above.
NGC 5746 was observed in the C array of the VLA on 2007 January 7, 8, 12 and 13.
Phase calibration was achieved through observations of VLA calibrator 1445+099 about
every 30 minutes. Observations of 3C286 and 3C48 were used for flux and bandpass calibra-
tion. Sixty three spectral channels were employed, centered at 1725 km s−1, with channel
width 20.85 km s−1, while online Hanning smoothing yielding a velocity resolution of twice
this width. A total of about 28.3 hours were spent observing NGC 5746. Eight of the 27
antennae were unavailable for most of the final track, while two were absent on the other
three dates. Smaller amounts of data were also lost due to high winds, interference, and
equipment failures. Data were inspected for high amplitudes and any suspect data excised.
Some small baseline corrections for a few antennae were made, based on later observations
in the same configuration. The calibration of all four tracks is of very high quality.
High amplitude visibilities were clipped before mapping. Dirty maps of each track were
made to check for additional problems before concatenating the four uv datasets. Continuum
was subtracted in the uv-plane using line-free channels at either end of the spectrometer with
the AIPS tasks UVSUB and UVLSF. This was very successful. The AIPS task IMAGR was
– 6 –
used to produce clean maps using clean boxes that covered the emission from NGC 5746 as
well as NGC 5740, which is also detected, albeit somewhat outside the 31.9’ FWHM primary
beam. The pointing center is R.A. 14h44m56.4s, Decl. 1◦57’ 16” (J2000). The primary
data cube discussed here was made with no uv-tapering and has a resolution of 15.5”x14.8”
(2.2 x 2.1 kpc) at a P.A. of −3.6◦. A cube with 61.3”x58.1” resolution (8.7 x 8.3 kpc), at
a P.A. of 45.5◦, created via uv-tapering is also discussed here. All maps have 1024x1024
pixels of 3” size, and were made with uniform uv-weighting with the IMAGR ’ROBUST’
parameter equal to 0. The noise in a single channel of the full-resolution cube is 0.23 mJy
(beam)−1. No primary beam correction was made for maps of NGC 5746. Primary beam
attenuation reaches 4% at the ends of the major axis. Maps of NGC 5740 are corrected for
primary beam attenuation; the response at the center of this galaxy has dropped to 0.36, and
thus structure may not be accurately mapped. No other galaxies in the NGC 5746 group
are within the mapped region. Observed velocities are on the heliocentric scale. The noise
corresponds to an HI column density of 4.6×1019cm−2averaged over one resolution element
for optically thin gas. In a zeroth-moment map made using all data and all channels, a 5σ
detection of a point source corresponds to 2.3 × 106M⊙, using the conversion from (e.g.)
Zwaan et al. (1997). All masses are of total atomic gas and include a correction of 1.36 for
Finally, the GIPSY task BLOT was used to blank emission-free regions in each channel,
reducing the noise in moment maps. Moment maps were made with various strategies to
eliminate as much noise as possible while retaining as much emission as possible that appears
to be real in the channel maps and position-velocity (pv) diagrams (given its signal-to-noise
ratio and continuity over multiple velocity channels). The zeroth-moment map presented
here for the full-resolution cube includes all emission in the blotted cube above the 1σ level
in each pixel in two consecutive channels. For the 60” resolution cube, such a strategy
introduced too many features in the zeroth-moment map that did not appear real in the
data cube, and thus a 2.5σ cutoff was used. A concern with the velocity resolution of 42 km
s−1is that many real spectrally narrow features may be rejected by this strategy.
Channel maps of the full-resolution cube are shown in Figure 1. Zeroth- and first-
moment maps made from the blotted full-resolution cube are shown in Figure 2. The zeroth-
moment map is overlaid on an Digitized Sky Survey red image in Figure 3. Also labeled
in Figure 3 are the four extraplanar features discussed in §3.5. The emission from NGC
5746 is dominated by what appears to be a highly inclined disk at a position angle (PA) of
– 7 –
350◦(measured CCW from N to the receding side of the major axis) with a large central
depression of about 2’ diameter. In this sense it is somewhat reminiscent of the Infrared Space
Observatory ISOCAM 12µm image from Bendo et al. (2002), which has the appearance of
an inclined ring of approximate diameter 2.3’ (19.7 kpc). It is possible the HI hole is larger
but limited resolution has made it appear smaller.
– 8 –
Fig. 1.— Channel maps from the full-resolution cube for NGC 5746. Contour levels are 3,
6, 12, 24, 36 and 48 times the 1 σ noise in column density units of 4.6 × 1019cm−2. Each
map is labeled with its heliocentric velocity.
– 9 –
Fig. 2.— Left: Zeroth-moment map of NGC 5746 from the full-resolution cube. Contour
levels are 1, 2, 4, 8, 17 and 32 times 1020cm−2. The beam is shown in the lower right corner.
Right: First-moment map from the full-resolution cube. Contours run from 1500 to 2020 km
s−1in 40 km s−1increments from south to north.
– 10 –
Fig. 3.— Zeroth-moment map of NGC 5746 from the full-resolution cube overlaid on a
Digitized Sky Survey red image. Contour levels are as in Figure 2 (left). The approximate
locations of the centers of the four features discussed in §3.5 are also indicated.
– 11 –
At fainter levels, a component extended along the minor axis is clearly detected. It
is resolved into individual clouds and filaments to some degree. This is in contrast to the
HI map of NGC 891 (see Figure 1 of Oosterloo et al. 2007) where the halo is smoother in
appearance despite better linear resolution (1.4 kpc vs. 2.1 kpc). Emission is detected up to
about 70” (10 kpc) from the midplane. The northern end of the disk also suggests a warp.
The total flux in the map in Figure 2 is 33.7 Jy km s−1, yielding a total atomic mass of
9.4×109M⊙. For comparison, the total observed flux found by Springob et al. (2005) using
the Arecibo telescope is 38.8 Jy km s−1.
Figure 4 shows the zeroth- and first-moment maps from the 60”-resolution cube. The
total mass in this map is 8.8 × 109M⊙, slightly less than in the full-resolution map. We
attribute this to the somewhat higher cutoff that was necessary to eliminate noisy features
in the moment map at this resolution. However, it also indicates that there is very little
diffuse, low surface brightness emission that is missed in the full resolution cube. The ends
of the disk in this map also suggest a warp, but with the opposite sense of bending than is
evident at lower radii in the full resolution map.
– 12 –
Fig. 4.— Left: Zeroth-moment map of NGC 5746 from the 1’-resolution cube. Contour
levels are 1, 2, 4, 8, 16, 32, 64 and 92 times 1.3×1019cm−2. The beam is shown in the lower
right corner. Right: First-moment map from the 1’-resolution cube. Contours run from 1460
to 2020 km s−1in 40 km s−1increments from south to north.
– 13 –
3.1.Modeling the Atomic Mass Distribution and Rotation
Our main goal here is to understand the origin of the faint, vertically extended com-
ponent seen in Figure 2 through full modeling of the data cubes. Although we are able to
quantitatively constrain many of the model parameters, we do not carry out a full search
of parameter space, not least because of asymmetries in the data, but rather explore and
constrain a few well motivated types of models.
We begin by constraining the radial density profile at the midplane. As a first estimate,
we employ the GIPSY task RADIAL, which fits radial profiles of the column density inte-
grated vertically through the disk to averaged major axis emission profiles for highly inclined
galaxies. The observed major axis emission profile is shown in Figure 5 as the solid line.
Since in the below modeling we will use axisymmetric density distributions, we also show
a profile where the north and south sides have been averaged to produce an axisymmetric
profile (dashed line). The fitted axisymmetric profile is shown as the dotted line. The radial
column density profile derived by RADIAL is shown in Figure 6 as the filled circles. The disk
is dominated by two rings of radius 1.5’ and 3’ and width about 1’, while the profile confirms
the aforementioned central depression of radius 1’. RADIAL reproduces the observed pro-
file in Figure 5 very well, except for a moderate overestimate of the emission in the central
depression. This overestimate cannot be rectified by creating a more prominent central hole
but is instead a result of projected emission from gas at larger radii; it presumably indicates
an asymmetry in the galaxy. The other symbols in Figure 6 show the deviations from the
basic profile featured in the models described below.
– 14 –
Fig. 5.— The solid line shows the observed major axis emission profile averaged over the
minor-axis extent of the emission in NGC 5746. The dashed line shows the average of the
north and south sides of the observed profile. The dotted line shows the fitted profile from
the GIPSY task RADIAL.
– 15 –
Fig. 6.— Filled circles show the axisymmetric radial profile of the column densities integrated
vertically through the disk derived by RADIAL for NGC 5746. This profile is used for the
unwarped component of Models A and B. Open circles show the modification necessary to
model the extended emission along the minor axis as a warp in Models A and B. Triangles
show the slightly lower profile for the disk component in the lagging halo Models C, D, and
E. The ’x’ and the open squares show the column density of the ringlike lagging halo Models
D and E, respectively. See text for further details of the models.
– 16 –
We use the major axis pv diagram (Figure 7) and the envelope tracing method to
estimate the rotation curve. The method is described in Sofue & Rubin (2001). We use
η = 0.3 (see Sofue & Rubin 2001 for the definition of this parameter). We assume the
inclination is close enough to 90◦so that no correction for projection is necessary.
there is little modeled column density in the central 1’ radius, the rotation curve is very
poorly constrained there. In fact, long-slit spectra and a peanut bulge indicate that NGC
5746 is clearly barred, with the bar elongated more across the line of sight than along it
(Kuijken & Merrifield 1995; Bureau & Freeman 1999). The bar manifests itself kinematically
in a steeply rising component within major axis offsets of ±0.25′in pv diagrams of emission
lines (indicative of x2orbits perpendicular to the bar), which is clearly not seen in HI. For
our rotation curve we have simply extended the value at R = 1.15′inwards. Our adopted
rotation curve is shown in Figure 8. The dynamical mass within R = 5′(42.8 kpc), given by
M(R) = RV2(R)/G, is 7.9 × 1011M⊙.
– 17 –
Fig. 7.— At the top, middle and bottom are shown major-axis position-velocity diagrams
for NGC 5746 from the full-resolution cube, Model B and Model D respectively (see text for
model details). Contour levels are as in Figure 1.
– 18 –
Fig. 8.— The filled circles show the rotation curve for NGC 5746 derived from envelope
fitting. The open circles show the assumed extension at low radii.
– 19 –
We next need to constrain the inclination, i. For nearly edge-on galaxies, tilted-ring
programs such as ROTCUR in GIPSY cannot do this accurately, especially if a warp along
the line of sight or a lagging halo is suspected and contributes significantly to the velocity
field. However, i can be estimated by assuming axisymmetry in the ringlike appearance of the
emission at the center of the zeroth-moment map. Hence, we use our adopted rotation curve
and radial column density profile as inputs to the GIPSY program GALMOD (a program
which allows construction of model galaxies with specified radial column density profiles,
rotation curves, and various forms for the vertical density distribution) in order to generate
a zeroth-moment map and match it to the data. The radial bins in all our models have a
width of 15” (2.1 kpc). The models produced by GALMOD are convolved to the resolution
of the data. From the measured separation of the two sides of the ring along the minor axis
near the galaxy center in Figure 2, we find that i must be about 86◦. This value is confirmed
by the modeling described below. ROTCUR was run to constrain other parameters, and
indicates a PA of 350±0.3◦, a systemic velocity of 1733±7 km s−1, and a kinematic center
about 6” or 850 pc west of the pointing center, at R.A. 14h44m56.0s, Decl. 1◦57’ 16”
(J2000). The gas is initially assumed to be in a single exponential layer with scale height 3”
or about 400 pc. The true scale height of the bulk of the emission is not well constrained
because of the resolution and could be substantially lower than 3”.
We next show that the vertical structure cannot be fit by a single component at a given
inclination, or by two exponential components. Figure 9 shows a minor-axis emission profile,
averaged over 3’ along the major axis (solid line). There are clearly tails at both ends of
the profile. The dashed line shows our modeled component for an exponential layer with 3”
scale height. It fits the bulk of the emission well but not the tails. The dotted line shows an
exponential model with a scale height of 18”. It fits the tails well but is broader than the
bulk of the emission and washes out the splitting at low latitudes (which is due to the ringlike
distribution). No combination of two exponentials were found to produce a reasonable fit.
This result does not change if a gaussian or sech2form is used.
– 20 –
Fig. 9.— The solid line in both panels shows the minor axis emission profile averaged over
the inner 3’ (25.6 kpc) of the major axis, in column density units. The dashed and dotted
lines in the top panel show profiles for models as described in the text with exponential scale
heights of 3” and 18”, respectively. The dashed line in the bottom panel shows the profile
for Model A, as described in the text. The profiles for the other models described are almost
identical to that of Model A.
– 21 –
Instead we consider two possibilities to explain the tails. The first is a warp along the
line of sight, and the second is a halo with a box-like (employed for ease of modeling although
it is unphysical) profile in z. The model parameters are summarized in Table 1.
The morphology of the tails turns out to be very much coupled with their kinematics,
so we now consider these together by introducing pv diagrams parallel to the minor axis
at various positions along the major axis which provide good leverage on the parameters of
interest. These are shown in Figure 10, along with several models generated by GALMOD
and variants on that program. The characteristic appearance of the disk component in these
diagrams is one of narrow angular extent at velocities furthest from Vsyswhich broadens
and (where the ringlike structure is evident) splits as velocities move towards Vsys. The
faint high-latitude component has a characteristic appearance which will greatly constrain
its morphology and kinematics. Although it varies significantly among cuts, it generally
manifests itself as an extended, spectrally very narrow component at velocities closer to Vsys
than the disk component, with a definite gradient in Vhelwith b that is more evident in panels
further from the galaxy center. Some east-west asymmetries are evident in this component.
– 22 –
Fig. 10.— Position-velocity diagrams parallel to the minor axis at various offsets along
the major axis from the full-resolution data cube and various models described in the text.
Contour levels are as in Figure 1. The offsets along the major axis are shown in the panels
for the data.
– 23 –
We first consider a warp along the line of sight. In GALMOD, this is achieved by slowly
decreasing the inclination of outer rings. However, in order to reproduce the minor axis
emission profile, the column density in the warped rings had to be increased significantly
from the major-axis-based starting point, as shown in Figure 6. The warp begins at a radius
of 4’ (34.2 kpc), where i = 84◦, dropping to 79◦by 5’ (42.8 kpc) radius (Figure 11). The
maximum displacement of the midplane of the outermost warped ring from the unwarped
disk is about 5 kpc. A slightly larger warping is required for the SE quadrant (top three
panels of Figure 10 at negative minor axis offsets) but is not modeled here. The resulting
model, Model A, appears to match the observed minor-axis pv diagrams very well.
– 24 –
Fig. 11.— Filled circles show the run of inclination (left scale) for Models A and B. Open
triangles show the run of PA (right scale) for Model B. Open circles show the inclination for
Models C, D and E.
– 25 –
Model B is an attempt to incorporate the bending across the line of sight evident in the
zeroth-moment maps. Retaining the run of inclination in Model A, the PA in this model
decreases from 350◦to 346◦over the radius range 3’ to 4’, and then increases back to 5◦by
5’ radius (Figure 11). The asymmetry introduced into the minor-axis pv diagrams (Figure
10) provides a slightly better match to the data at high latitudes, although still probably
underestimating the observed asymmetry. The zeroth-moment map of this model (Figure
12) also exhibits a slight brightness asymmetry with respect to the minor axis at major
axis distances less than 4’ from the center and at heights 30–60” from the plane that is
also apparent on the east side in Figure 2. The model appears to somewhat overestimate
the bending to larger PAs evident in Figure 2 (and, of course, there is too much emission
at the ends of the major axis because of the excess column density in the modeled warp).
However, a model version smoothed to 60” (Figure 12, center), is a reasonable match to the
bending in Figure 4. We note that warps are often asymmetric (Garc´ ıa-Ruiz et al. 2002),
so one should not expect a warp fitted to kinematic information parallel to the minor axis
to match perfectly the morphology along the major axis. A second caveat to this exercise
is that some of the bending along the major axis may be due to outer spiral structure seen
not quite edge-on. In fact, the initial bending to lower PAs in the model is only present to
match the morphology in the moment maps. The subsequent increase in PA is responsible
for the asymmetries in the pv diagrams. As mentioned at the beginning of this section, for
these kinds of reasons, we have not carried out a full exploration of parameter space but
simply present these models as reasonable fits to the data.
– 26 –
Fig. 12.— Left: Zeroth-moment map of Model B, convolved to 15” resolution. Contour
levels are as in Figure 2 (left). Center: Zeroth-moment map of Model B, convolved to 60”
resolution. Contour levels are as in Figure 3. Right: Zeroth-moment map of Model D,
convolved to 15” resolution. Contour levels are as in Figure 2 (left).
– 27 –
We next consider the box-like halo, extending to z = 50” (7.1 kpc). Such a component,
when convolved with the beam, loses its sharp edges and, added in the right proportion to
the disk component, reasonably reproduces the minor-axis emission profile in Figure 9, being
almost indistinguishable from the warp model profile. The disk component column density
is normalized downwards by a factor of 1.24 in these models.
For the halo models, the gradient in Vhelwith b at high latitudes in Figure 10 implies a
rotational velocity that decreases with z, i.e. dVrot/dz < 0. The spectral narrowness implies
that such a halo cannot occupy a large range in radius, otherwise projection of the rotational
velocity would cause broadening of the Vhelprofiles in disagreement with the data.
To illustrate these constraints, we first consider a model where the halo has the same
radial density profile shape as the disk, and a vertical gradient in rotation velocity of –3.5
km s−1(arcsec)−1(–25 km s−1kpc−1). We refer to this as Model C in Figure 10. Although
the amount of high-latitude emission is well matched to the data, there is little such emission
seen at the first solid contour level of this model. This is because the velocity profiles there
are broader with lower peak intensities, extending to velocities much too far from Vsysto be
consistent with the data, and we demonstrate this by including a low-level dashed contour.
Clearly, this model is a poor match to the halo emission.
The narrowness of the high-latitude velocity profiles puts a strong constraint on the
radial range of the halo component, and leads us to consider annular distributions, beginning
with one in which the entire halo is contained in one 15” ring. Model D in Figure 10 features
a ring of central radius R0= 180”, with dVrot/dz = −3.5 km s−1arcsec−1, and a necessarily
high column density of 9×1020cm−2to match the minor axis emission profile. Comparison
of modeled and data pv diagrams for other values of R0constrain this parameter to be in
the range 160–200” (22.8–28.5 kpc). Even for such a narrow ring, the high-latitude velocity
profiles tend to be broader than in the data. For values of dVrot/dz outside the range –3
to –4 km s−1arcsec−1the gradient in the halo component in the pv diagrams is not well
matched. However, a somewhat higher gradient is required for the SE quadrant, and models
indicate a value of dVrot/dz = −5 km s−1arcsec−1. All halo models employ i = 86.25◦to
match the minor-axis emission profile.
The narrowness of the ring also introduces a strong edge-brightening in the halo in
zeroth-moment maps (Figure 12) due to the projection of the ring which is clearly not seen
in the data. This halo would therefore need to be much more asymmetric than the warp
of Models A and B. The normalization of the modeled halo also now depends on the radial
range used to form the minor axis emission profiles. We have chosen to normalize using the
same radial range as above, realizing that the edge-brightening will lead to excess mass in
the modeled halo.
– 28 –
An annulus wider than 1’ (8.6 kpc), although featuring a more plausible column density,
causes the high-latitude velocity profiles to broaden, if only slightly. Model E, where the
annulus is 1.25’ wide and its column density is 1.8 × 1020cm−2, shows this effect.
The major-axis pv diagram for the data and Models B and D are compared in Figure
7. The general shape of the flat part of the pv diagram is reproduced quite well, given
that we do not attempt to model the aforementioned asymmetry in the emission around the
major axis. The general slope of the rising part of the diagram is underestimated. In the
models, the slope is dictated by the fact that most of the gas is in the outer (R = 3′) ring
in our axisymmetric radial profile. Asymmetries, in the sense of gas being more centrally
concentrated in the vicinity of the inner ring at major axis offsets within 1’ of the center,
could be the cause of this discrepancy.
The velocity dispersion is not well constrained because of the somewhat low velocity
resolution. A value of 10 km s−1was used in the models for both disk and halo.
We roughly estimate the fraction of mass in the extended component in two different
ways. First, we simply assume that in the zeroth-moment map of Figure 2, all emission
above a certain height can be attributed to the extended component. Using two values of
this height – 9” and 11” (1.3 and 1.6 kpc) –, its mass is 1.2 − 1.6 × 109M⊙. The second
estimate comes from the models. In the warp model, the excess column density in the
warped region accounts for 1.8×109M⊙. For the halo models, we calculate the mass in the
halo component and find 3.1 × 109M⊙. These are both overestimates, especially the latter,
because the extra column density along the line of sight required by the axisymmetric models
overestimates the emission along the major axis. The estimate from the data themselves is
likely to be superior.
Although both models are inaccurate because of their symmetry, the problem is much
worse for the halo model. This model is also much more contrived in that it requires a very
large mass in a very small radial range. The narrowness of the high-latitude spectra is also
better reproduced by the warp model. We therefore conclude that the extended component
is a warp rather than a halo. Further fine tuning of the warp model may produce a slightly
better match to the data, but the asymmetries preclude a significant improvement, and hence
we do not pursue this further here.
Figure 13 (top panel) shows a pv diagram along the minor axis of the galaxy. A shift
in mean velocity between the positive and negative sides is apparent in the disk component,
– 29 –
while a very small shift also exists in the warp component (the double-peaked nature of the
disk component is due to the inner ring of 1.5’ radius as discussed above). Such a shift is
a potential sign of radial inflow in a not quite edge-on galaxy. All diagrams for major axis
offsets within one beam width of this one show this signature. For the warp, such a shift is
also seen, albeit at a slightly smaller amplitude, in the corresponding pv diagram of Model B
(bottom panel of Figure 13), and we cannot rule out that it is an effect of the warp geometry.
For the disk, since there is strong kinematic evidence of a bar (see §1), the signature would
suggest radial inflow in the inner ring-like gas distribution surrounding the bar. To measure
the inflow amplitude, we have created two spectra from the minor axis pv diagram, averaged
over the range 3–18” in the minor axis direction on the east and west sides. A Gaussian is
fit to each spectrum to find the mean velocity, and half of the velocity difference between
the east and west sides is taken as the inflow velocity. The result is Vinflow= 9.7 ± 0.4 km
– 30 –
Fig. 13.— Minor-axis position-velocity diagram for the full-resolution cube (top) and Model
B (bottom). Contour levels are 1.5, 3, 6 and 12 times the 1 σ noise in column density units
of 4.6 × 1019cm−2.
– 31 –
3.3.A Possible Supershell
A structure in the SW quadrant above the disk has the form of a closed shell – better
seen in Figure 14 – although it is always possible that projection of unrelated features has
given it this appearance. The total mass is about 108M⊙, and the extent parallel to the
major axis is about 3.4 kpc. Position-velocity diagrams parallel to the minor axis for the
northern wall, center and southern wall of the shell are shown in Figure 15. Above about
10” (1.4 kpc) from the plane, the walls show emission at the velocities of the warp as well
as a broad component extending to about the terminal velocity in the northern wall, and
some evidence for line splitting in the southern wall. The northern wall is detected over
nine velocity channels, or about 190 km s−1. The line splitting in the southern wall is five
channels or about 100 km s−1. Given that the velocities in the warp component are due to
the warp geometry, it would be incorrect to conclude that this line broadening/splitting is
due to expansion. Rather, it may just indicate that the feature is extended along the line
of sight. There is no indication of a corresponding ionized feature in the Hα image of Rand
– 32 –
Fig. 14.— Close-up of the zeroth-moment map of NGC 5746 showing the shell-like structure
described in the text.
– 33 –
Fig. 15.— Position-velocity diagrams parallel to the minor axis from the full-resolution cube
for the northern wall, center and southern wall of the possible supershell. Contour levels are
2, 4, 8, 16 and 24 times the 1 σ noise in column density units of 4.6×1019cm−2as in Figure
– 34 –
3.4.A Clumpy Warp?
Figure 2 shows that the high-latitude emission is partially resolved into individual
clumps. Almost thirty such features have been identified by visual inspection and are listed
in Table 2. This selection is not meant to be complete or statistically well defined, but is
simply meant to give a first-order idea of the masses of well detected features. These range
from about 4×106M⊙to 108M⊙, and their masses sum to 7×108M⊙, or about half of the
total mass of the high-latitude emission. The most massive, numbered 18, is the shell-like
structure discussed above. Most are unresolved but a few have a filamentary appearance
and lengths of 3–6 kpc. The line widths are difficult to estimate for many features because
of the velocity resolution of 42 km s−1(FWHM) and the low signal-to-noise ratio of some of
the detections. For example, the significant emission for three of the best detected clouds,
numbered 5, 6 and 8 in Table 2, is confined to two channels and therefore the lines are
unresolved. Others, such as 11 and 28, show emission over 6 − 7 channels, or 125 − 145 km
s−1, and it is not clear if these represent actual internal motions or reflect the extent of the
features along the line of sight, for example.
The 60”-resolution zeroth-moment map (Figure 4) reveals two faint vertical extensions
from the plane in the SE quadrant, reaching heights of 15 and 20 kpc. Summing their
emission above where they appear to merge with the disk, we obtain masses of 7.9×106M⊙
and 6.8 × 106M⊙, for the northern and southern feature, respectively. A faint extension is
also seen off the NW end of the disk. Its mass is about 6.2×106M⊙. The first-moment map
shows that the SE features have about the same velocity as the part of the warp that they
connect to with little change in velocity along their vertical extent.
3.5.High Latitude Features Not in the Warp
We have carefully combed the channel maps and pv diagrams to look for high-latitude
emission which does not have the velocities of the warp component but which may not be
evident in Figure 2. We have found four such features. Position-velocity diagrams for the
first three, at the center of their extent along the major axis, are shown in Figure 16. The
features’ locations are also marked in Figure 3, and their properties are summarized in Table
3. One is an unresolved feature in the NE quadrant of the halo. The second and third are
diffuse features with appreciable extents along the major axis of 0.7’ and 0.6’ (6.0 and 5.1
kpc), respectively. The second feature is also counterrotating and its emission spans more
than 300 km s−1. The fourth is the low-Vhelside of the possible shell (Cloud 18 in Table 2).
– 35 –
Fig. 16.— Position-velocity diagrams parallel to the minor axis from the full-resolution cube
showing emission (indicated by arrows) from three high-latitude features that do not have
the velocities of the warp. From top to bottom, these are Clouds 1, 2, and 3 in Table 3.
Contour levels are 1.5, 3, 6, 12, 24, 36 and 48 times the 1 σ noise in column density units of
4.6 × 1019cm−2.
– 36 –
As this barred (Erwin 2005) Sb galaxy is beyond the half-power radius of the primary
beam, the detailed structure may not be accurate and we limit ourselves to general properties
only. We assume its distance to be the same as that of NGC 5746. Channel maps of the
full-resolution cube are shown in Figure 17. Zeroth- and first-moment maps of NGC 5740
are presented in Figure 18, using all data above 2σ in two consecutive channels. The zeroth-
moment map is overlaid on a Digitized Sky Survey red image in Figure 19. There is a disk
of about 30 kpc diameter and a large extension to the NNW. The disk has a mass of about
5.3×109M⊙and the extension 9.4×108M⊙. For comparison, Schneider et al. (1986) quote
a mass in atomic hydrogen of 7.0 × 109M⊙, scaled to our distance.
– 37 –
Fig. 17.— Channel maps from the full-resolution cube for NGC 5740, corrected for primary
beam attenuation. Contour levels are 3, 6, 12, 24, 36 and 48 times 4.6 × 1019cm−2. Each
map is labeled with its heliocentric velocity.
– 38 –
Fig. 18.— Left: Zeroth-moment map of NGC 5740 from the full-resolution, primary-beam-
corrected cube. Contour levels are 2, 4, 8, 12 and 16 times 1020cm−2. Right: First-moment
map of NGC 5740. Contours run from 1440 to 1740 (the closed contour on the NE side at
Decl. 1◦42’) km s−1in 20 km s−1increments from SW to NE.
– 39 –
Fig. 19.— Zeroth-moment map of NGC 5740 from the full-resolution cube overlaid on a
Digitized Sky Survey red image. Contour levels are as in Figure 18 (left).
– 40 –
A ROTCUR analysis of the disk indicates a PA of 339±0.5◦, an inclination of 58±2◦,
Vsys= 1586 ± 6 km s−1, a dynamical center of R.A. 14h44m24.4s, Decl. 1◦40’ 52”, with
approximate uncertainty 3” in R.A. and 8” in Decl. The first-moment map suggests regular
rotation continues into the NNW extension. Keeping Vsysand the dynamical center fixed,
and fitting the receding half of the galaxy only, the extension exhibits little change in PA
but a continued fall in Vrotand an increase in inclination to 61◦at 2’ and 70◦at 4’. The
derived rotation curve is shown in Figure 20. The dynamical mass within R = 2′(17.1 kpc)
is 1.2 × 1011M⊙.
Morphologically, the NNW extension is similar to extensions in several Virgo Cluster
galaxies, most of which are best explained by ram pressure stripping (Chung et al. 2007).
There is no obvious evidence of compression of the gas on the opposite side of the disk,
however. The condition for ram pressure stripping, as given by Vollmer et al. (2001), is
the velocity of the galaxy through the medium, for gas at surface density ΣISMat a radius
R from the galactic center. At the outskirts of the main disk the typical surface density
is about 2 M⊙pc−2, while Vrot= 175 km s−1. The condition for stripping is then (n/10−3
cm−3)(v/100 km s−1)2> 14. Both required numbers are not constrained. We find that the
one-dimensional velocity dispersion for the catalogued galaxies of the NGC 5746 group is 120
km s−1. The IGM density is unknown. X-ray bright groups are found to be HI-deficient by
Sengupta & Balasubramanyam (2006), who calculate a range of IGM densities from 5×10−4
to 2×10−3cm−3, but it is unknown how extensive X-ray emission is in the NGC 5746 group.
Hence, it is unclear whether the stripping condition can be met. Tidal stripping seems less
likely given the morphology and that the closest group galaxy on the sky, NGC 5746, is 155
kpc away in projection.
rot/R, where ρIGM is the intragroup medium (IGM) density, and vgalis
– 41 –
Fig. 20.— Filled circles show the rotation curve for NGC 5740 derived from a tilted ring
analysis. Open circles show the extension of the rotation curve into the NNW extended
– 42 –
Almost all of the high-latitude gas we have detected can be associated with the warp.
At the level of sensitivity of our data, only the features listed in Table 3, summing to about
108M⊙, may have originated in a disk-halo flow or infall. The heights off the plane of their
centers range from 2 to 8 kpc. One of the more massive ones is counter-rotating and is
unlikely to have originated in the disk. A few×107M⊙worth of counter-rotating clouds
were also discovered in the neutral halo of NGC 891 by Oosterloo et al. (2007).
Are these few clouds consistent with predictions from the halo thermal instability sim-
ulations mentioned in §1? In the simulations of Sommer-Larsen (2006), most of the warm
clouds are within 50 kpc from the center of the galaxy and total 108M⊙, with most having
masses a few times 105to a few times 106M⊙for a galaxy like the Milky Way. By compari-
son, for an M33-like galaxy, Kaufmann et al. (2006) expect clouds of radii 0.1−0.6 kpc and
masses from 105to a few times 106M⊙, totaling ∼ 107−108M⊙depending on the simulation
parameters, confined to within about 10–20 kpc above the disk. If the few clouds in Table 3
are due to infall, we note that their total mass is comparable to these model predictions, with
the caveat that the fraction of neutral gas is uncertain in the simulations. The small number
of detected clouds is in fact comparable to that in the “very high resolution” simulation of
Sommer-Larsen (2006), where the mass resolution is comparable to that of our data.
For galaxies like the Milky Way, Maller & Bullock (2004) predict 2×1010M⊙of mostly
ionized clouds of mass 5 × 106M⊙and size ∼ 1 kpc, extending to 150 kpc from the center
of the galaxy. At our sensitivity, it is difficult to rule out with confidence such a widespread
population of clouds from the HI data alone.
It would be interesting to know how much mass in warm clouds forms in simulations
of more massive systems like NGC 5746, and where it is located. The cooling rate, at least,
of the hot gas in this galaxy can be estimated from the observations, but is very uncertain.
The rate inferred by Rasmussen et al. (2006), who exclude the projected disk area where
there may be contamination by other sources of hot gas, is only 0.2 M⊙(yr)−1. The rate is
modest because the estimated halo temperature is rather high, at 0.56 keV. However, most
of the halo cooling is likely to be occurring in the excluded area, and their models of massive
disks suggest that the cooling rate could be 5–10 times higher as a result. In summary, it
is unclear whether the observed clouds are consistent with theoretical expectations for halo
cooling in a massive disk galaxy like NGC 5746.
What fraction of these halo clouds do we expect to be fully or mostly ionized? This
depends on the incident ionizing flux, the mass, and most importantly, the bounding pressure
on the cloud. If we assume that the ionized gas in a cloud is in thermal pressure equilibrium
– 43 –
with an outside pressure, the density of a cloud is given by n−2 = 4.55T−1
n−2 = nH/10−2cm−3is the normalized hydrogen density of the cloud, T4 = T/(104K)
is the normalized temperature, and [p3/k] = [p/k]/(103cm−3K) is the normalized thermal
pressure. (We use here ρ/nH = 2.4 × 10−24for solar metallicity gas, ntot = 2.2nH, and
ne = 1.2nH for fully ionized plasma). For a constant density spherical cloud, the cloud
radius is given by R = (532 pc)M1/3
We assume that the outer skin of this cloud will be ionized by a (plane-parallel) flux
of H ionizing photons, φ5= φ/(105photons cm−2s−1). The metagalactic flux is estimated
to be φ5< 0.45 (Sternberg et al. 2002, and references therein), while high velocity clouds
in the Galactic halo experience φ5= 2 (Tufte et al. 2002). The thickness of this skin, ℓ, is
determined by the balance of ionizations and recombinations, φ = αBn2
recombination coefficient is αB∼= 3 × 10−13cm3s−1. (Note that for optically thick clouds,
φ = αB EM, so that the surface averaged emission measure is a simple measure of the
ionizing flux.) In this case, the skin depth is given by ℓ = (35.7 pc)φ5T2
eℓ where the Case B
With these approximations, the ionized volume and mass of the cloud is given by fion=
1 − [1 − (ℓ/R)]3or fion= 3(ℓ/R) − 3(ℓ/R)2+ ... for small (ℓ/R). The condition for a cloud
to be fully ionized is
< (197 cm−3K) φ3/5
For very low ionization fraction (fion< 5%), this criterion is well approximated by
= (380 cm−3K)f−3/5
so that an ionization fraction of less than 1% requires [p/k] > (6022 cm−3K) φ3/5
The sensitivity of the ionization fraction of gas clouds to the outside pressure has been
studied by Ferrara & Field (1994), Wolfire et al. (1995), and Maloney & Putman (2003)
with comparable results. The normalization constant relating the pressure to the ionization
fraction and cloud mass will depend on the assumed geometry. The presence of central
condensation due to dark matter halos can also lower somewhat the pressure necessary to
form neutral clouds (Sternberg et al. 2002).
These considerations show that not only are X-ray halos associated with galaxy forma-
tion more likely to form condensations, but that the higher pressures of such halos result in
a much larger neutral fraction for the condensation. Using the density and temperature esti-
mate for the X-ray halo of NGC 5746 (Rasmussen et al. 2006), the halo pressure is given by
– 44 –
[p/k] = 6200 cm−3K. Such a high bounding pressure would guarantee that clouds embedded
in this hot gas will be primarily neutral. It is also interesting to note that the bounding
pressure for the cloud can be either the thermal gaseous halo pressure or the ram pressure
associated with a cloud’s motion through the halo. The one cloud showing counter rotation
presumably has a much higher ram pressure than most of the rest of neutral gas detected
and could be expected to have a higher neutral fraction.
At larger radius than that traced by the X-ray gas, it remains possible that the pressure
will drop to such a point that a large population of fully ionized clouds might be present,
but uncertainties in the X-ray data and analysis prevent such pressure variations from being
assessed.Although the predicted emission measure from the front and back face of an
individual cloud is a relatively low EM= 2 × (0.11 pc cm−6) φ5, a population of unresolved
smaller clouds in the beam might boost the Hα emission to detectable levels. However,
processes such as thermal conduction and evaporation should set a lower limit on the mass
of cloud that can survive (Maller & Bullock 2004). We have reexamined the Hα image of
Rand (1996), smoothed to 1.5” resolution – yielding a noise level of 2.2 pc cm−6– and find
no evidence for such bright clouds in the halo.
To attempt to constrain the origin of the neutral gas halo in this and other galaxies,
we can compare the neutral halo mass and mass fraction with estimates for other galaxies
and look for correlations with other parameters (Table 4). In this comparison one should
keep in mind that some of the galaxies are edge-on, one (NGC 6946) very much face-on,
and some at intermediate inclinations (NGC 4559, NGC 2403 and NGC 253), and thus the
characterization of halo gas necessarily varies from galaxy to galaxy. In the galaxies not
viewed edge-on, the extraplanar gas is identified by its anomalous velocities compared to
the bulk of the emission. We also note that the sensitivity and linear resolution of these
observations vary significantly.
NGC 5746 ranks near the bottom of the table in terms of halo mass and mass fraction
and also has one of the lowest levels of star formation activity, as estimated from LFIR/D2
lending support for an origin in a weak disk-halo flow. However, there is no discernable trend
among the galaxies in this direction. NGC 6946 and NGC 253 stand out as having little HI in
their halos given their star forming activity. The halo HI in the latter is particularly puzzling
as it is found only on one side of the disk, away from the center of the galaxy (Boomsma et al.
2005). It is possible that the neutral gas has been swept radially outward by the central
starburst outflow or that much of the halo gas is ionized. In NGC 6946, Boomsma (2007)
note that the mass of halo gas more than doubles in a map at 64” resolution vs. one at 22”
resolution. The rotation speeds of the galaxies span a factor of three, and thus the ability of
supernova power to raise gas off the plane should vary significantly from galaxy to galaxy.
– 45 –
In particular, the high fractions of halo HI in NGC 4559, NGC 2403 and UGC 7321 could
be in part due to their low mass. Finally, there does not seem to be any correlation between
halo mass or mass fraction and rotation speed alone.
However, at this point it is not unreasonable to conclude for NGC 5746 that the three
non-counterrotating clouds in Table 3 may just as likely be due to a relatively inactive disk-
halo cycle as to infall. One possible further clue to their origin would be infrared emission,
and indeed imaging observations of this and other edge-ons with the Spitzer Space Telescope
have been proposed. A detection of dust would argue against an origin in low-metallicity
A larger sample of edge-ons needs to be observed with comparable sensitivity and angu-
lar resolution to understand how the mass and kinematics (rotational lags in particular) of
neutral halo gas relates to other galactic properties such as star formation activity, assuming
that such gas can always be kinematically distinguished from warps and flares. A better
measure of such activity than the somewhat crude tracer employed here would be the 24µm
surface density as measured by the Spitzer Space Telescope (Calzetti et al. 2005), hence 24µm
maps of a large sample of edge-ons would be of great benefit in such comparisons. For NGC
5746, deeper observations with higher spectral resolution are clearly called for in order to
carry out a more complete census of halo gas that can be distinguished from the warp, so
that its spatial distribution, kinematics, and cloud masses can be compared to the various
models discussed here.
It should also be remembered that NGC 5746 is in a group environment, and some of
the high-latitude gas may owe its origin to previous encounters, although the apparently low
metallicity of the hot gas argues against encounters with other large galaxies.
We reemphasize that our best warp model (Model B) is not necessarily a unique fit to
the data, especially because of asymmetries, but does demonstrate that a warp is a superior
explanation to a lagging halo. It is not surprising that a warp is present in NGC 5746, as they
are common in disk galaxies (e.g Garc´ ıa-Ruiz et al. 2002). As far as the warp parameters
are concerned, we note that it is generally the case in such galaxies (Briggs 1990) that the
line of nodes increases in the direction of galaxy rotation for radii beyond the Holmberg
radius, which is 4.5’ for NGC 5746 (Holmberg 1958). Our warp model implies an increase
beginning just inside this radius. More recently, van der Kruit (2007) finds evidence from
edge-ons that warps begin just beyond the radius where a truncation is evident in the stellar
disk, but finds no indication of such a truncation in NGC 5746.
Finally, there is the possibility that the warp itself is caused by infall. The origin
of warps is debated, but the possibility receiving most attention recently is that they
– 46 –
are due to infalling gas with an angular momentum vector tilted with respect to the in-
ner disk (Jiang & Binney 1999; Shen & Sellwood 2006).
Shen & Sellwood (2006), in which particles are injected into an outer torus inclined at a
fixed angle with respect to the main disk, the warp can persist for several Gyr, making it
difficult to infer the accretion history from the mere presence of a warp. That is, the large
gas mass in the warp may reflect accretion over a substantial period of time and not nec-
essarily demand a high accretion rate, as would more likely be the case such a large mass
were contained in a lagging halo. On the other hand, if the centrally concentrated X-ray
halo emission reflects the hot gas density, then one would expect any warm clouds recently
formed by cooling out of the hot phase to be more centrally concentrated.
In the N-body simulations of
We have observed the massive edge-on spiral NGC 5746 with the VLA, in an attempt to
find a vertically extended component of atomic gas as predicted by recent galaxy formation
models. A high-latitude component has been discovered, but almost all of its mass, summing
to 1.2 − 1.6 × 109M⊙– or about 15% of the total HI mass – is more readily explained as
a warp than as a halo. The warp must be asymmetric in its column density distribution
and geometry, but this is not unusual. If the high-latitude component is a lagging halo, it
must feature a large mass in a rather narrow radial range, be centered at a large radius,
and have a much larger degree of asymmetry than the warp. Even then, such a model has
difficulty reproducing the narrowness of the high-latitude spectra as well as the warp model.
We therefore conclude that a warp is the more likely explanation. The warp itself may be a
result of infall according to recent models. It partially resolves into clouds of mass 4 × 106
M⊙to 108M⊙, accounting for about half the mass in the warp.
We have found four high-latitude features at velocities distinct from the warp, totaling
about 108M⊙. These could be accreting onto the disk, or they may originate in a disk-
halo flow, although there is no other sign of such a flow in this galaxy with little star
forming activity. One cloud is counter-rotating and must have an external origin. The
cloud properties are roughly comparable to those expected in the galaxy formation models
of Kaufmann et al. (2006) and Sommer-Larsen (2006) where warm clouds form by thermal
instabilities in the residual hot halos. However, these models have not been tailored to
galaxies as massive as NGC 5746, and it would be interesting to compare our results with
such a model. Given the calculated pressure in the hot halo, we expect any clouds to be
primarily neutral. Alternatively, we cannot rule out a weak disk-halo flow as the origin for
these clouds, except for the counter-rotating one. How neutral gaseous halos relate to other
– 47 –
galaxian properties needs to be tested with more observations.
The disk shows the signature of radial inflow at the level of about 10 km s−1. This
is most likely due to the bar, which is known to have a strong kinematical signature from
optical long-slit spectra.
The group member NGC 5740 is also detected, albeit outside the half-power points of
the primary beam. The most important result here is a broad extension of gas to the NE,
which is more likely to be due to ram pressure than tidal stripping, although whether the
condition for such stripping is met is unknown.
We thank J. Mulchaey for discussions regarding X-rays in groups, and L. Sparke for
discussion about warps. We also thank an anonymous referee for many useful comments.
The VLA is operated by the National Radio Astronomy Observatory, which is a facility
of the National Science Foundation operated under cooperative agreement by Associated
Universities, Inc. We thank the VLA staff for making these observations possible. This
research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated
by the Jet Propulsion Laboratory, California Institute of Technology, under contract with
the National Aeronautics and Space Administration.
Alton, P. B., Rand, R. J., Xilouris, E. M., Bevan, S., Ferguson, A. M., Davies, J. I., &
Bianchi, S. 2000, A&AS, 145, 83
Barbieri, C. V., Fraternali, F., Oosterloo, T., Bertin, G., Boomsma, R., & Sancisi, R. 2005,
A&A, 439, 947
Barnab` e, M., Ciotti, L., Fraternali, F., & Sancisi, R. 2006, A&A, 446, 61
Bendo, G. J., Joseph, R. D., Wells, M., Gallais, P., Haas, M., Heras, A. M., Klaas, U.,
Laureijs, R. J., Leech, K., Lemke, D., Metcalfe, L., Rowan-Robinson, M., Schulz, B.,
& Telesco, C. 2002, AJ, 123, 3067
Benjamin, R. A. 2002, in Astronomical Society of the Pacific Conference Series, Vol. 276,
Seeing Through the Dust: The Detection of HI and the Exploration of the ISM in
Galaxies, ed. A. R. Taylor, T. L. Landecker, & A. G. Willis, 201–+
– 48 –
Benson, A. J., Bower, R. G., Frenk, C. S., Lacey, C. G., Baugh, C. M., & Cole, S. 2003,
ApJ, 599, 38
Binney, J. 1977, ApJ, 215, 483
—. 2004, MNRAS, 347, 1093
Birnboim, Y., & Dekel, A. 2003, MNRAS, 345, 349
Boomsma, R. 2007, PhD thesis, Kapteyn Astronomical Institute, University of Groningen
Boomsma, R., Oosterloo, T. A., Fraternali, F., van der Hulst, J. M., & Sancisi, R. 2005,
A&A, 431, 65
Braun, R., & Walterbos, R. A. M. 1992, ApJ, 386, 120
Bregman, J. N. 1980, ApJ, 236, 577
Bregman, J. N., & Pildis, R. A. 1994, ApJ, 420, 570
Briggs, F. H. 1990, ApJ, 352, 15
Brinks, E., & Burton, W. B. 1984, A&A, 141, 195
Buote, D. A. 2000, MNRAS, 311, 176
Bureau, M., & Freeman, K. C. 1999, AJ, 118, 126
Calzetti, D., Kennicutt, Jr., R. C., Bianchi, L., Thilker, D. A., Dale, D. A., Engelbracht,
C. W., Leitherer, C., Meyer, M. J., Sosey, M. L., Mutchler, M., Regan, M. W.,
Thornley, M. D., Armus, L., Bendo, G. J., Boissier, S., Boselli, A., Draine, B. T.,
Gordon, K. D., Helou, G., Hollenbach, D. J., Kewley, L., Madore, B. F., Martin,
D. C., Murphy, E. J., Rieke, G. H., Rieke, M. J., Roussel, H., Sheth, K., Smith, J. D.,
Walter, F., White, B. A., Yi, S., Scoville, N. Z., Polletta, M., & Lindler, D. 2005,
ApJ, 633, 871
Carrillo, R., Masegosa, J., Dultzin-Hacyan, D., & Ordo˜ nez, R. 1999, Revista Mexicana de
Astronomia y Astrofisica, 35, 187
Chung, A., van Gorkom, J. H., Kenney, J. D. P., & Vollmer, B. 2007, ApJ, 659, L115
Collins, J. A., Benjamin, R. A., & Rand, R. J. 2002, ApJ, 578, 98
Collins, J. A., Shull, J. M., & Giroux, M. L. 2007, ApJ, 657, 271
– 49 –
Dahlem, M., Lazendic, J. S., Haynes, R. F., Ehle, M., & Lisenfeld, U. 2001, A&A, 374, 42
Dahlem, M., Lisenfeld, U., & Rossa, J. 2006, A&A, 457, 121
Dekel, A., & Birnboim, Y. 2006, MNRAS, 368, 2
Dettmar, R.-J. 1990, A&A, 232, L15
Erwin, P. 2005, MNRAS, 364, 283
Fall, S. M., & Rees, M. J. 1985, ApJ, 298, 18
Ferrara, A., & Field, G. B. 1994, ApJ, 423, 665
Field, G. B. 1965, ApJ, 142, 531
Fraternali, F., & Binney, J. J. 2006, MNRAS, 366, 449
Fraternali, F., van Moorsel, G., Sancisi, R., & Oosterloo, T. 2002, AJ, 123, 3124
Garc´ ıa-Ruiz, I., Sancisi, R., & Kuijken, K. 2002, A&A, 394, 769
Giuricin, G., Marinoni, C., Ceriani, L., & Pisani, A. 2000, ApJ, 543, 178
Gonz´ alez-Mart´ ın, O., Masegosa, J., M´ arquez, I., Guerrero, M. A., & Dultzin-Hacyan, D.
2006, A&A, 460, 45
Haffner, L. M., Reynolds, R. J., Tufte, S. L., Madsen, G. J., Jaehnig, K. P., & Percival,
J. W. 2003, ApJS, 149, 405
Heald, G. H., Rand, R. J., Benjamin, R. A., & Bershady, M. A. 2006a, ApJ, 647, 1018
—. 2007, ApJ, 663, 933
Heald, G. H., Rand, R. J., Benjamin, R. A., Collins, J. A., & Bland-Hawthorn, J. 2006b,
ApJ, 636, 181
Holmberg, E. 1958, Meddelanden fran Lunds Astronomiska Observatorium Serie II, 136, 1
Howk, J. C., & Savage, B. D. 1999, AJ, 117, 2077
Irwin, J. A. 1994, ApJ, 429, 618
Irwin, J. A., English, J., & Sorathia, B. 1999, AJ, 117, 2102
Irwin, J. A., & Madden, S. C. 2006, A&A, 445, 123
– 50 – Download full-text
Jiang, I.-G., & Binney, J. 1999, MNRAS, 303, L7
Kaufmann, T., Mayer, L., Wadsley, J., Stadel, J., & Moore, B. 2006, MNRAS, 370, 1612
Kereˇ s, D., Katz, N., Weinberg, D. H., & Dav´ e, R. 2005, MNRAS, 363, 2
Kuijken, K., & Merrifield, M. R. 1995, ApJ, 443, L13
Lee, S.-W., & Irwin, J. A. 1997, ApJ, 490, 247
Lee, S.-W., Irwin, J. A., Dettmar, R.-J., Cunningham, C. T., Golla, G., & Wang, Q. D.
2001, A&A, 377, 759
Maller, A. H., & Bullock, J. S. 2004, MNRAS, 355, 694
Maloney, P. R., & Putman, M. E. 2003, ApJ, 589, 270
Matthews, L. D., & Wood, K. 2003, ApJ, 593, 721
Murray, S. D., & Lin, D. N. C. 1990, ApJ, 363, 50
Nagar, N. M., Falcke, H., & Wilson, A. S. 2005, A&A, 435, 521
Norman, C. A., & Ikeuchi, S. 1989, ApJ, 345, 372
Oosterloo, T., Fraternali, F., & Sancisi, R. 2007, AJ, 134, 1019
Pedersen, K., Rasmussen, J., Sommer-Larsen, J., Toft, S., Benson, A. J., & Bower, R. G.
2006, New Astronomy, 11, 465
Putman, M. E., Staveley-Smith, L., Freeman, K. C., Gibson, B. K., & Barnes, D. G. 2003,
ApJ, 586, 170
Rand, R. J. 1996, ApJ, 462, 712
Rand, R. J., Kulkarni, S. R., & Hester, J. J. 1990, ApJ, 352, L1
Rasmussen, J., Sommer-Larsen, J., Pedersen, K., Toft, S., Benson, A., Bower, R. G., &
Olsen, L. F. 2006, ArXiv Astrophysics e-prints
Reynolds, R. J., Roesler, F. L., & Scherb, F. 1973, ApJ, 179, 651
Reynolds, R. J., Sterling, N. C., & Haffner, L. M. 2001, ApJ, 558, L101
Rossa, J., & Dettmar, R.-J. 2003a, A&A, 406, 493