A kinematic study of the neutral and ionised gas in the irregular dwarf galaxies IC 4662 and NGC 5408
ABSTRACT Abridged. The feedback between massive stars and the interstellar medium is one of the most important processes in the evolution of dwarf galaxies. This interaction results in numerous neutral and ionised gas structures that have been found both in the disc and in the halo of these galaxies. However, their origin and fate are still poorly understood. We here present new HI and optical data of two nearby irregular dwarf galaxies: IC 4662 and NGC 5408. The HI line data were obtained with the ATCA and are part of the Local Volume HI Survey. They are complemented by optical images and spectroscopic data obtained with the ESO NTT and the ESO 3.6m telescope. Our main aim is to study the kinematics of the neutral and ionised gas components in order to search for outflowing gas structures and to make predictions about their fate. We find the HI gas envelopes of both galaxies to extend well beyond the optical discs. The optical disc is embedded into the central HI maximum in both galaxies. However, higher resolution HI maps show that the HI intensity peaks are typically offset from the prominent HII regions. While NGC 5408 shows a fairly regular HI velocity field, which allows us to derive a rotation curve, IC 4662 reveals a rather twisted HI velocity field, possibly caused by a recent merger event. We detect outflows with velocities between 20 and 60 km/s in our Halpha spectra of both galaxies, sometimes with HI counterparts of similar velocity. We suggest the existence of expanding superbubbles, especially in NGC 5408. This is also supported by the detection of FWHMs as high as 70 km/s in Halpha. In case of NGC 5408, we compare our results with the escape velocity of the galaxy, which shows that the measured expansion velocities are in all cases too low to allow the gas to escape from the gravitational potential of NGC 5408. This result is consistent with studies of other dwarf galaxies. Comment: 23 pages, 16 figures, accepted for publication by MNRAS
Mon. Not. R. Astron. Soc. 000, 1–21 (2002)Printed 28 April 2010(MN LATEX style file v2.2)
A kinematic study of the neutral and ionised gas in the
irregular dwarf galaxies IC4662 and NGC5408?
Janine van Eymeren1,2,3†, B¨ arbel S. Koribalski3‡,´Angel R. L´ opez-S´ anchez3,
Ralf-J¨ urgen Dettmar2, Dominik J. Bomans2
1Jodrell Bank Centre for Astrophysics, School of Physics & Astronomy, The University of Manchester, Alan Turing Building,
Oxford Road, Manchester, M13 9PL, UK
2Astronomisches Institut der Ruhr-Universit¨ at Bochum, Universit¨ atsstraße 150, 44780 Bochum, Germany
3Australia Telescope National Facility, CSIRO Astronomy and Space Science, P.O. Box 76, Epping, NSW 1710, Australia
Accepted 2010 April 26. Received 2010 April 22; in original form 2010 February 10
The feedback between massive stars and the interstellar medium is one of the most
important processes in the evolution of dwarf galaxies. This interaction results in nu-
merous neutral and ionised gas structures that have been found both in the disc and
in the halo of these galaxies. However, their origin and fate are still poorly understood.
We here present new Hi and optical data of two Magellanic irregular dwarf galaxies
in the Local Volume: IC4662 and NGC5408. The Hi line data were obtained with the
Australia Telescope Compact Array and are part of the “Local Volume Hi Survey”.
They are complemented by optical images and spectroscopic data obtained with the
ESO New Technology Telescope and the ESO 3.6m telescope. Our main aim is to
study the kinematics of the neutral and ionised gas components in order to search
for outflowing gas structures and to make predictions about their fate. Therefore, we
perform a Gaussian decomposition of the Hi and Hα line profiles.
We find the Hi gas envelopes of IC4662 and NGC5408 to extend well beyond the op-
tical discs, with Hi to optical diameter ratios above four. The optical disc is embedded
into the central Hi maximum in both galaxies. However, higher resolution Hi maps
show that the Hi intensity peaks are typically offset from the prominent Hii regions.
While NGC5408 shows a fairly regular Hi velocity field, which allows us to derive a
rotation curve, IC4662 reveals a rather twisted Hi velocity field, possibly caused by a
recent merger event. We detect outflows with velocities between 20 and 60kms−1in
our Hα spectra of both galaxies, sometimes with Hi counterparts of similar velocity.
We suggest the existence of expanding superbubbles, especially in NGC5408. This is
also supported by the detection of FWHMs as high as 70kms−1in Hα, which cannot
be explained by thermal broadening alone. In case of NGC5408, we compare our re-
sults with the escape velocity of the galaxy, which shows that the measured expansion
velocities are in all cases too low to allow the gas to escape from the gravitational
potential of NGC5408. This result is consistent with studies of other dwarf galaxies.
kinematics and dynamics – galaxies: structure
galaxies: individual (IC4662, NGC5408) – galaxies: ISM – galaxies:
?The radio observations were obtained with the Australia Tele-
scope which is funded by the Commonwealth of Australia for
operations as a National Facility managed by CSIRO. All optical
observations were collected at the European Southern Observa-
tory, Chile, Proposal-Nos.: 047.01-003, 51.1-0067, 69.D-0143(A),
† E-mail: Janine.Eymeren@rub.de
The interplay between massive stars and the interstellar
medium (ISM) has a large effect on the formation and the
evolution of galaxies, especially of dwarf galaxies. Photoioni-
sation is the most likely interaction process. However, stellar
winds and supernovae (SNe) explosions of massive stars also
contribute significantly to the energy input into the ISM.
The radiative and mechanical feedback heat the gas and
arXiv:1004.4757v1 [astro-ph.CO] 27 Apr 2010
2van Eymeren et al.
drive it outwards, sweeping up the ambient gas into a thin
shell. A superbubble filled with hot gas evolves and begins
to expand into the ISM. Due to Rayleigh-Taylor instabili-
ties, the outer shell can rupture and the hot gas can vent
out through so-called chimneys into the halo (Norman &
Numerous ionised gas structures such as supergiant
shells or filaments close to the galactic disc of dwarf galaxies,
but also at kpc-distances from any place of current star for-
mation, have been detected on deep Hα images (e.g., Hunter
et al. 1993; Bomans et al. 1997; Hunter & Gallagher 1997;
L´ opez-S´ anchez & Esteban 2008). Spectroscopic observations
of the Hα line revealed that most of the ionised gas struc-
tures expand from the disc into the halo of their host galaxies
(e.g., Marlowe et al. 1995; Martin 1998; Bomans 2001; van
Eymeren et al. 2007). This leads to the question of what the
fate of the gas is. One scenario is that the gas cools down
in the halo and eventually falls back onto the galactic disc
(outflow, galactic fountain scenario, Shapiro & Field 1976).
However, it might also be possible that the gas escapes from
the gravitational potential by becoming a freely flowing wind
(galactic wind). The detection of large amounts of hot gas
in the intergalactic medium (IGM) and the generally low
metal content of dwarf galaxies support this scenario.
blow-out in dwarf galaxies (Mac Low & Ferrara 1999) pre-
dict that at least a part of the hot gas has enough kinetic
energy to leave the gravitational potential of its host galaxy
and to enrich the IGM. The relatively low escape velocities
of dwarf galaxies should facilitate the removal of substan-
tial amounts of gas (Larson 1974). According to Ferrara &
Tolstoy (2000) galactic winds occur in dwarf galaxies with
gas masses up to 109M?. However, the fate of the gas does
not only depend on the mass of the host galaxy and there-
fore its gravitational potential, but it also strongly depends
on the morphology of the ISM distribution. For a spherical
distribution, the ISM seems to be more resistant to ejection
(Silich & Tenorio-Tagle 2001).
A recent detailed kinematic study of the neutral and
ionised gas in the two irregular dwarf galaxies NGC2366
and NGC4861 revealed several outflows in each galaxy (van
Eymeren et al. 2009a,b). The measured expansion velocities
were of the order of 20 to 50kms−1, which is, in compari-
son with the escape velocities of the host galaxies, too low
to allow the gas to leave the gravitational potential. This
result confirms earlier observations: no convincing case of a
galactic wind in local dwarf galaxies has been reported so
far (Bomans 2005). Note that galaxies like M82 are not typ-
ical dwarf galaxies as they show strong starbursts, are more
luminous and have a higher mass.
Altogether, this shows that despite all the detailed ob-
servational studies and simulations, we do not fully under-
stand what is going on in these galaxies. Galactic winds
are still thought to be necessary ingredients for their forma-
tion and evolution, but direct evidence seems to be difficult
to get. In order to improve our knowledge about the pro-
cesses happening in dwarf galaxies, we performed a multi-
wavelength study of altogether four irregular dwarf galax-
ies (van Eymeren 2008). The results for NGC2366 and
NGC4861 have already been published (van Eymeren et al.
2009a,b, see also above). We here concentrate on the two
remaining galaxies IC4662 and NGC5408. We obtained Hi
line data as well as optical images and spectroscopic data.
Some basic properties of both galaxies are given in Table 1.
IC4662 (HIPASS J1747–64) is classified as a barred
irregular galaxy of Magellanic type (IBm). Its distance of
DTRGB=2.44Mpc was obtained by Karachentsev et al.
(2006) and hence makes it the nearest known representa-
tive of blue compact dwarfs (Karachentsev et al. 2006). It
seems to be a rather isolated galaxy, belonging to no known
groups. de Vaucouleurs (1975) describes IC4662 as a fore-
ground galaxy in the direction of the NGC6300 group.
NGC5408 (HIPASS J1403–41) is classified as an
IB(s)m galaxy. It was first studied by Bohuski et al. (1972)
who found that its nucleus consists of several bright Hii
regions and appears to be undergoing a violent burst of
star formation. As the galaxy reveals an ultra-luminous
X-ray source, NGC5408X-1, very close to the main Hii
regions, it was a popular object to study over the last
decade (e.g., Soria et al. 2004, 2006; Strohmayer et al.
2007; Lang et al. 2007; Kaaret & Corbel 2009). This
X-ray source has recently been argued to harbour an
intermediate-mass black hole (Strohmayer & Mushotzky
2009). The distance of NGC5408 of DTRGB=4.81Mpc
was obtained by Karachentsev et al. (2002). Its position
on the sky puts NGC5408 in the Centaurus A group. The
closest known neighbour appears to be ESO325-G?001 at
a projected distance of 208?.
This paper is organised as follows: in Sect. 2 we give
an overview over the observations and the data reduction.
In Sect. 3 we describe and compare the morphology of the
neutral and ionised gas. Section 4 contains the kinematic
analysis of both gas components including a search for ex-
panding gas. The results are discussed in Sect. 5, which is
followed by a summary in Sect. 6.
2OBSERVATIONS AND DATA REDUCTION
Deep R-band and Hα images are available for both galaxies.
Using the ESO Multi-Mode Instrument (EMMI) attached to
the ESO New Technology Telescope (NTT), we obtained a
600s exposure of IC4662 in R-band and a 1800s exposure in
Hα. A 600s R-band image of NGC5408 as well as a 1200s
Hα image, both observed with the ESO Faint Object Spec-
trograph and Camera (EFOSC) attached to the ESO 3.6m
telescope, have been taken from the ESO archive. The data
reduction was performed using the software package IRAF
(Tody 1993), and included standard procedures of overscan-
and bias-subtraction as well as a flatfield correction. Addi-
tionally, we removed cosmic rays by running the IRAF ver-
sion of L. A. Cosmic (van Dokkum 2001). In order to get the
pure Hα line emission, we first scaled the flux of the stars
in both the continuum and the Hα images and subsequently
subtracted the continuum image from the Hα image. The
seeing was 1.??1 and 0.??8 during the observations of IC4662
and NGC5408 respectively. We used adaptive filters based
on the H-transform algorithm (Richter et al. 1991) to stress
weaker structures in the Hα images and to differentiate them
A kinematic study of the gas in IC4662 and NGC54083
Table 1. The basic properties of IC4662 and NGC5408.
opt. diameter [?]
opt. diameter [kpc]
position angle [◦]
336 ± 29
3.0 × 1.6
2.1 × 1.1
12.33 ± 0.09
−14.91 ± 0.1
0.14 ± 0.01
537 ± 33
2.6 × 1.6
3.6 × 2.2
12.59 ± 0.09
−16.12 ± 0.09
0.44 ± 0.03
302 ± 3
130.0 ± 12.0
1.83 ± 0.17
506 ± 3
61.5 ± 6.7
3.36 ± 0.37
Note: The blue luminosity is calculated using a solar B-band magnitude of 5.48 mag. – References: (1) de Vaucouleurs et al. (1991), (2)
Karachentsev et al. (2006), (3) Karachentsev et al. (2002), (4) Lauberts & Valentijn (1989) [ESO Uppsala], (5) Schlegel et al. (1998),
(6) Koribalski et al. (2004) [HIPASS BGC].
Table 2. Imaging – some observational parameters.
Exp. Time [s]
Spatial res. [??]
608 (R), 596 (Hα)
642 (R), 692 (Hα)
from the noise. All images are displayed in Fig. 1. Some ob-
servational details are given in Table 2.
2.1.2 Medium-resolution long-slit spectroscopy
We used archival medium-resolution long-slit spectra of
NGC5408, which have a spectral resolution of about
60kms−1, as measured from the night sky lines. As the
target of these observations was the ultra-luminous X-ray
source NGC5408X-1, all spectra were obtained at almost
the same position. Therefore, we only analyse one slit posi-
tion as a representative. The seeing was 0.??9.
We performed further medium-resolution long-slit spec-
troscopy not only of NGC5408, but also of IC4662 from May
1st to May 3rd 2006. As we aimed for very deep spectra and
were mainly interested in the Hα line, only the red part of
EMMI was read out. We used grating #7 centred on the Hα
line with a wavelength coverage of 1300˚ A and a dispersion
of 0.41˚ A/pix. With a slit width of 1??we achieved a spectral
resolution of 112kms−1, again as measured from the night
sky lines. The pixel size is 0.??332 and the seeing was about
1??. Spectra of a thorium comparison lamp were taken for
the wavelength calibration.
We obtained three spectra each at different positions
across the two galaxies. The slit positions were chosen to
intersect prominent shell and filamentary structures as visi-
ble on the Hα images. Some details of the observations are
given in Table 3.
The data reduction of all spectra (including the archival
one) was performed by us using IRAF and included
overscan- and bias-subtraction, flatfield correction and a
wavelength calibration. A geometric distortion correction
was not necessary because the deviations were smaller than
one pixel. Again, cosmic rays were removed by running L. A.
Cosmic. We also performed a background subtraction to re-
move the contamination by the night sky lines.
The spectra and their analysis are presented in Sect. 4.4.
They were binned by three pixels in spatial direction, which
corresponds to about 1??in order to match the seeing (see
above). At positions of very weak emission, ten pixels were
summed up to improve the signal to noise ratio (S/N).
4van Eymeren et al.
Figure 1. Optical imaging of the Magellanic irregular dwarf galaxies IC4662 (top) and NGC5408 (bottom). Left panel: R-band image;
right panel: continuum-subtracted Hα image. The images of IC4662 were obtained with the ESO NTT, the images of NGC5408 with
the ESO 3.6m telescope. In order to locate the positions of the main star clusters and the main Hii regions respectively, the high intensity
areas in both images are highlighted in white.
Table 3. Long-slit spectroscopy – some observational details.
ObjectTelescope / InstrumentDateSlit No.Exp. Time
NGC 5408ESO NTT / EMMI01.05.06
IC 4662ESO NTT / EMMI02.05.06
Note: Slit1b of NGC5408 denotes the archival spectrum. The position angle PA is defined to increase counter-clockwise with north
A kinematic study of the gas in IC4662 and NGC54085
2.2ATCA Hi synthesis observations
Hi line observations of IC4662 and NGC5408 were obtained
with the Australia Telescope Compact Array (ATCA), and
are part of the “Local Volume Hi Survey” (LVHIS1; Korib-
alski 2008; Koribalski et al. 2010).
The first frequency band was centred on 1418 MHz with
a bandwidth of 8 MHz, divided into 512 channels. This gives
a channel width of 3.3kms−1and a velocity resolution of
4kms−1. The ATCA primary beam is 33.?6 at 1418 MHz,
which is sufficient to fully map both galaxies and their sur-
In order to ensure excellent uv-coverage and sensitiv-
ity to large-scale structures, IC4662 was observed for a full
synthesis (12h) in each the EW367, 750A, and 1.5D config-
urations. The lower resolution data of NGC5408 (375 and
750D array) were taken from the archive and complemented
by high-resolution data within LVHIS (1.5A array).
The data reduction was carried out with the MIRIAD
software package (Sault et al. 1995) using standard proce-
dures, including flux- and phase calibrations. Using a first
order fit to the line-free channels in the Hi data set, the
20cm radio continuum was separated from the Hi emission.
We created two different sets of data cubes, a low-resolution
cube to detect the extended emission and a high-resolution
cube to resolve the inner structure and to make the Hi data
comparable to the optical data. For the low-resolution cubes
of both galaxies, we excluded the longest baselines, which
are all baselines to the distant antenna 6 (CA06). The low-
resolution data of IC4662 were made using ’natural’ weight-
ing of the uv-data in the velocity range covered by the Hi
emission in steps of 4kms−1. The low-resolution data of
NGC5408 were made using ’robust’ weighting. The high-
resolution data of both galaxies were created by including
all baselines to CA06 and using the same weighting as for
the low-resolution cubes. The resulting beam sizes are given
in Table 4.
1Jybeam−1corresponds to an Hi column density of
2.55 × 1020atoms cm−2(IC4662, low-res.), 2.64 × 1021
atoms cm−2(IC4662, high-res.), 2.89 × 1020atoms cm−2
(NGC5408, high-res.). The moment maps (integrated in-
tensity map, intensity-weighted mean velocity field, and the
velocity dispersion) were created from the Hi data cubes
by first isolating the regions of significant emission in ev-
ery channel and afterwards clipping everything below a 2.5σ
threshold. This final step of the data reduction process and
the subsequent analysis of the Hi data were performed with
The Groningen Image Processing System (GIPSY, van der
Hulst et al. 1992), complemented by some IRAF tasks.
2.77 × 1021
1LVHIS is a large project that aims to provide detailed Hi inten-
sity maps, velocity fields and 20cm radio continuum observations
for a complete sample of nearby, gas-rich galaxies belonging to the
Local Volume (LV), a sphere of radius 10Mpc centred on the Lo-
cal Group. The ATCA observations include all LV galaxies which
were detected in HIPASS and reside south of δ ≈-30◦. Further de-
tails can be found at http://www.atnf.csiro.au/research/LVHIS/.
Table 4. ATCA Hi data – imaging properties.
Hi synthesised beam
Hi synthesised beam
In order to get a first impression of how the stars and the
gas are distributed, we compared the optical and Hi mor-
phologies of IC4662 and NGC5408.
The R-band image of IC4662 (see Fig. 1, upper left panel)
reveals a slightly elongated, box-like shape with a position
angle of 104◦(Table 1). The Hα image (Fig. 1, upper right
panel) shows a large Hii region complex with an Hα mini-
mum in the south-eastern part of the stellar light distribu-
tion, and extended diffuse, filamentary gas structures out-
side the stellar disc to the north-east. The central two star
clusters are offset by about 350pc to the north-east from the
optical centre (see Table 1), which probably explains the ex-
tended Hα emission in this area. They coincide with the two
brightest Hii regions visible on the Hα image. Additionally,
a star cluster with associated Hα emission that appears to
be detached from the main body of the galaxy can be seen. It
is located 1.?5 or 1.1kpc to the south-east of the centre. Our
deep Hα image shows an Hii region at the same position
that is connected to the main complex by small, compact
Hii regions and diffuse filamentary gas structures. We will
come back to this feature in Sect. 5.5.
The channel maps of the low-resolution Hi data (see
Fig. 2) show that the emission is distributed over a velocity
range from roughly 200 to 400kms−1, i.e., over 200kms−1.
The neutral gas seems to consist of two systems, an inner
system with a position angle of about 135◦and an outer part
with a position angle of roughly 45◦, i.e., perpendicular to
the inner system. The integrated Hi intensity map (upper
left panel of Fig. 3) confirms this feature: it shows an inner
high column density system that is perpendicular to the
outer low column density system. As displayed on the lower
right panel of Fig. 3, the inner system coincides with the
optical extent of IC4662 including the southern Hii region.
We can see an additional Hi maximum to the north-west
of the galaxy, which has no optical counterpart. The Hi
outer diameter is given in Table 5. Note that the neutral
gas is extended by a factor of about six in comparison to
the ionised gas, which is among the largest ratios of Hi over
Hαever detected. We measured the flux density to be FHI =
123Jykms−1, from which we derived a total Hi mass of
MHI = 1.7×108M? (see Table 5). This is within the errors
of the HIPASS value of FHI = (130.0 ± 12.0)Jykms−1(see
Table 1), which means that our interferometric observations
6van Eymeren et al.
Table 5. ATCA Hi properties.
Hi diameter [?]
Hi diameter [kpc]
Hi / opt. ratio
< σ > [kms−1]
10.0 × 6.3
14.0 × 8.8
15.0 × 11.5
10.6 × 8.2
Note: As the velocity field of IC4662 is very distorted, we did
not perform a tilted-ring analysis and therefore could not
calculate any of the kinematic parameters. The systemic
velocity was estimated from the global Hi profile of IC4662.
The lower value of < σ > refers to the outer system of IC4662,
the higher value to the inner, perpendicular system.
recover most of the emission detected on single-dish data of
We also show the Hi intensity distribution at higher
spatial resolution (21??×20??) for a better comparison with
the optical data. In the upper left panel of Fig. 4, the high-
resolution Hi intensity contours are overlaid in white onto
the continuum-subtracted Hα image. Additionally, the in-
nermost contours of the low-resolution Hi intensity distribu-
tion are displayed in black. As the high-resolution contours
show, the central elongated maximum is split into several
smaller, point-like maxima. Interestingly, all these maxima
are clearly offset from the most prominent Hii regions.
Like IC4662, NGC5408 reveals a slightly elongated stellar
disc with a position angle of 62◦, which leads to a box-like
appearance in the optical (see Fig. 1, lower left panel). The
main star-forming knots are located along a chain running
from the centre to the main star formation area in the south-
west. In Hα, the galaxy is dominated by a bright complex of
Hii regions in the south-west (Fig. 1, lower right panel) that
coincides with the brightest star clusters. In the eastern part,
the ionised gas becomes more patchy with a few holes in the
distribution and several filaments emanating from the disc.
Indeed, it seems that the eastern part of the galaxy ends in a
superbubble as suggested by the shell-like structures visible
in the north- and south-east and the big hole in between.
Diffuse gas can also be seen across the southern part of the
Figure 5 shows the low-resolution Hi channel maps of
NGC5408. We detected emission over 130kms−1, from 430
to 560kms−1. In the inner parts the emission is elongated
along the east-west axis (position angle of 90◦), whereas in
the outer parts it is more extended to the north-west and
south-east (position angle of 135◦). The integrated Hi in-
tensity distribution looks fairly symmetric and reveals two
point-like maxima in the centre (see Fig. 5 and upper left
panel of Fig. 6). A comparison with the ionised gas distri-
bution shows that the western Hi intensity maximum co-
incides with the centre of the optical emission in NGC5408
(see Fig. 6, lower right panel). On larger scales, the distribu-
tion of the neutral gas seems to be warped in the outer parts,
which explains the two orientations seen in the channel maps
(Fig. 5). In comparison to the ionised gas, the neutral gas
component is four times more extended, which is again an
unusually high ratio (for the Hi diameter see Table 5). We
measured the flux density to be FHI = 63Jykms−1, from
which we derived a total Hi mass of MHI = 3.4 × 108M?
(Table 5). This is again within the errors of the HIPASS
value of FHI = (61.5 ± 6.7)Jykms−1.
Again, we overlaid the high-resolution Hi intensity con-
tours (20??×20??) in white over the continuum-subtracted
Hα image (Fig. 4, upper right panel). The distribution shows
two prominent maxima which coincide with the point-like
maxima detected on the low-resolution Hi map (black con-
tours). The western extended maximum coincides with the
main Hii region complex in NGC5408, although it is not
fully centred on the brightest Hii region. It is connected
via a bridge with the eastern maximum which is clearly off-
set from the southern shell-like structure mentioned above.
Additionally, the high-resolution Hi intensity distribution
reveals a third, weaker maximum in the north, close to, but
offset from the northern shell-like structure. In agreement
with the distribution of the ionised gas, the Hi column den-
sity between the shell-like structures is very low. This con-
firms our findings in the optical data, i.e., the existence of a
superbubble in the eastern part of NGC5408.
4 KINEMATIC ANALYSIS
The comparison of the optical and Hi morphology in IC4662
and NGC5408 showed that on low-resolution scales, the op-
tical extent coincides very well with the central Hi maxi-
mum. However, the high-resolution Hi maps reveal that the
peaks in the neutral gas distribution are typically offset from
the Hii regions. On larger scales, the optical appearance of
both galaxies is defined by an alignment which is perpen-
dicular to the extended Hi distribution (IC4662) or differs
by at least 45◦(NGC5408).
In this section, we analyse the kinematics of the two
galaxies in order to study the properties of both their neu-
tral and ionised gas components. All given velocities are he-
liocentric velocities measured along the line of sight.
4.1The distorted Hi kinematics of IC4662
The Hi velocity field of IC4662 (Fig. 3, upper right panel;
see also Fig. 4, middle row, left panel) is very disturbed.
The overall velocity gradient runs from the north-east with
velocities of 220kms−1to the south-west with velocities of
380kms−1. The direction of the rotation of the neutral gas
seems to change from the western part of the galaxy to the
eastern part, causing a sudden change of the position angle
A kinematic study of the gas in IC4662 and NGC54087
Figure 2. Low-resolution Hi channel maps of IC4662 (contours) superposed on our R-band image. For display purposes we show the
maps with a velocity resolution of 12kms−1. The original channel spacing is 4kms−1.The noise is about 0.9mJybeam−1. Contours are
drawn at −2.5 (−3σ), 2.5 (3σ), 5, 10, 20, 40, 80, and 160mJybeam−1. The synthesised beam of 70??×62??(natural weighting, excluding
CA06) is displayed in the lower left corner of the first channel map. The optical centre of the galaxy is marked by a white cross in the
same channel map. The corresponding heliocentric velocities are indicated in the upper right corner of each channel. 2?correspond to
of almost 90◦. This change is possibly connected with the
two perpendicular systems as discussed in Sect. 3.1. The
Hi velocity dispersion map (lower left panel of Fig. 3, over-
laid are the Hi intensity contours; see also Fig. 4, lower
left panel) shows a dispersion of 20kms−1in the inner and
11kms−1in the outer parts, except for two maxima with al-
most 40kms−1east of the extended Hi intensity maximum.
The large-scale twist in the velocity field makes it im-
possible to derive a rotation curve for IC4662. We extracted
spectra from the Hi data cube to get a better understand-
ing of the kinematics of the neutral gas. They are shown
in Fig. 7: each box represents the sum of all spectra within
70??×70??(corresponding to one beam size). Stars denote
spectra with a flux scale which is four times larger than
that of the other spectra. It is obvious that across the whole
distorted area, the Hi line is often split into two, some-
times clearly separated components. The difference of about
70kms−1is always the same, although the absolute veloci-
ties vary according to the rotation of IC4662.
As already mentioned in Sect. 2.2, we created the veloc-
ity field using the intensity-weighted mean (iwm). However,
as soon as the velocity distribution of the Hi emission is not
8van Eymeren et al.
Figure 3. Hi moment maps of IC4662, combining three arrays, but excluding baselines to CA06, which leads to a synthesised beam
of 70??× 62??(natural weighting). Top left: the Hi intensity distribution. Contours are drawn at 0.06 (3σ), 0.16, 0.4, 0.8, 1.6, 3.2, and
6.4Jybeam−1kms−1where 1Jybeam−1corresponds to a column density of 2.55 × 1020atomscm−2. Top right: the Hi intensity-
weighted mean velocity field. Contour levels range from 220 to 380kms−1in steps of 10kms−1. The systemic velocity of 302kms−1,
estimated from the global Hi profile, is marked in bold. Bottom left: the Hi velocity dispersion. Overlaid are the same Hi intensity
contours as mentioned above. Bottom right: continuum-subtracted Hα image. The spatial resolution was smoothed to fit the resolution
of the Hi data. Overlaid in black are again the Hi intensity contours. Overlaid in grey are the Hi velocity dispersion contours at
symmetric, the velocity values are biased to the longest tail
of the velocity profiles (de Blok et al. 2008). Therefore, one
might argue that the iwm velocity field does not represent
the true velocities. A recently more popular approach is to
fit the line profiles with Gauss-Hermite polynomials, which
allows to define the peak velocities more accurately. This be-
comes important when components overlap. In the IC4662
spectra, the two components are clearly separated with one
of them dominating the Hi profile. Hence, the difference be-
tween the iwm and the Hermite velocity field should be neg-
ligible. Nevertheless, we created the Hermite velocity field
by fitting Gauss-Hermite h3 polynomials to all line profiles
using the GIPSY task xgaufit (Fig. A1, middle panel). The
right panel of Fig. A1 shows the residuals after subtracting
the Hermite velocity field from the iwm. As we expected, the
residuals are mostly close to zero, except for an area in the
north-east. A look at the spectra (Fig. 7) reveals that the
S/N in this area is quite low, which can often not be han-
dled properly by the GIPSY moments task which we used to
create the iwm velocity field. As the large-scale distortion is
still very pronounced in the Hermite velocity field, we think
that it is not a consequence of using the ’wrong’ method
to create the velocity field, but that it is a real feature in
4.2The Hi velocity field of NGC5408
NGC5408 shows a symmetric Hi velocity field (Fig. 6, up-
per right panel; see also Fig. 4, middle row, right panel)
with a velocity gradient running from the south-east with
velocities of 450kms−1to the north-west with velocities of
550kms−1. The gradient is smooth, even at both ends of the
disc where the Hi distribution looks warped. It even seems
that it better follows the large scale structure than the in-
ner east-west alignment. A local, quite prominent distortion
of the velocity field can be seen south of the Hi intensity
maxima. The Hi intensity distribution, however, appears to
be smooth and undisturbed (see upper left panel of Fig. 6).
The Hi velocity dispersion (lower left panel of Fig. 6, over-
laid are the Hi intensity contours; see also Fig. 4, lower right
panel) shows two peaks at 25kms−1, which are located on
either side of the spur in the velocity field.
Again, we extracted the spectra from the Hi data cube
in order to analyse the profiles (see Fig. 8). Each box equals
a beam size of 60??×60??. Stars denote spectra with a flux
A kinematic study of the gas in IC4662 and NGC54089
Figure 4. High-resolution Hi moment maps of IC4662 (left) and NGC5408 (right). Upper row: a comparison of the Hα and Hi line
emission in the inner discs of both galaxies. The continuum-subtracted Hα emission is shown as grey-scale, while the black and white
contours indicate the Hi emission at low and high angular resolution respectively. The high-resolution contours of IC4662 are drawn
at 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4, and 1.6Jybeam−1kms−1where one beam corresponds to 21??× 20??; the ones of NGC5408 at 1.2,
1.6, 2.0, 2.4, 2.8, 3.2, 3.6, and 4Jybeam−1kms−1where one beam corresponds to 20??× 20??. Middle row: Hi velocity fields. Contour
levels are given in steps of 10kms−1and range from 260 to 340kms−1(IC4662) and from 480 to 530kms−1(NGC5408). The systemic
velocities (302kms−1and 502kms−1respectively) are marked in bold. Lower row: Hi velocity dispersion maps. The high-resolution
Hi intensity contours are overlaid as in the top panels.
10 van Eymeren et al.
Figure 5. Low-resolution Hi channel maps of NGC5408 (contours) superposed on the R-band image. For display purposes we show the
maps with a velocity resolution of 12kms−1. The original channel spacing is 4kms−1. The noise is about 1mJybeam−1. Contours are
drawn at -3 (−3σ), 3 (3σ), 5, 10, 20, 40, 80, and 160mJybeam−1. Otherwise the same as in Fig. 2. The synthesised beam is 71??×54??
(robust weighting, excluding CA06). 2?correspond to 2.8kpc.
scale which is four times larger than that of the other spec-
tra. In general, the Hi profiles look quite simple. The lines
are broad with FWHM of 35 to 50kms−1, but often also
symmetric with no second component clearly standing out.
The high velocity dispersion in the south coincides with an
area of very low S/N, which is probably again due to the in-
ability of the moments task to properly handle noisy spectra.
However, at the eastern edge we detected two components
which are separated by about 40kms−1.
We tested the quality of the iwm velocity field by sub-
tracting the Hermite velocity field: the residuals are very
small except for the area of high velocity dispersion. As in
the case of IC4662, this can be explained by the low S/N
and the line-splitting. For a direct comparison we refer the
reader to van Eymeren et al. (2009) where the Hi kinemat-
ics of several dwarf galaxies (among them NGC5408) have
been analysed by using the Hermite velocity fields.
4.3The Hi rotation curve of NGC5408
In order to make predictions about the fate of the gas, we
need to know some of the kinematic parameters like the in-
clination and rotation velocity of NGC5408. Therefore, we
derived a rotation curve by fitting a tilted-ring model to the
A kinematic study of the gas in IC4662 and NGC540811
Figure 6. Hi moment maps of NGC5408, combining three arrays, but excluding all baselines to CA06, which leads to a synthesised
beam of 71??× 54??(robust weighting). Top left: the Hi intensity distribution. Contours are drawn at 0.2 (3σ), 0.4, 0.8, 1.2, 2.0, 3.2,
and 4.0Jybeam−1kms−1where 1Jybeam−1correspond to a column density of 2.89 × 1020atomscm−2. Top right: the Hi intensity-
weighted mean velocity field. Contour levels range from 450 to 550kms−1in steps of 10kms−1. The systemic velocity of 502kms−1,
as calculated from fitting tilted-rings to the velocity field, is marked in bold. Bottom left: the Hi velocity dispersion, overlaid are the
same Hi intensity contours as mentioned above. Bottom right: continuum-subtracted Hα image. The spatial resolution was smoothed
to fit the resolution of the Hi data. Overlaid in black are again the Hi intensity contours. Overlaid in grey are the Hi velocity dispersion
contours at 20kms−1.
observed velocity field (GIPSY task rotcur, Begeman 1989).
Initial estimates for the relevant parameters (systemic veloc-
ity, the coordinates of the dynamic centre, the inclination,
and the position angle) were obtained by interactively fit-
ting ellipses to the Hi intensity distribution with the GIPSY
task ellfit. The width of the rings was chosen to be half the
spatial resolution, i.e., 31??. In order to get the most precise
values, three different approaches were made by always com-
bining receding and approaching side. The upper left panel
of Fig. 9 shows the resulting curves: first, we kept the ini-
tial estimates fixed in order to derive the rotation velocities
(green (light grey) triangles). The best-fitting values for all
parameters were derived by allowing only one parameter to
vary with radius while keeping the remaining ones fixed. As
no significant variation of any of the parameters over radius
was noticed, we derived an average value (given in Table 5).
The black triangles represent the resulting rotation veloci-
ties. The error bars indicate the values of the receding (top)
and approaching side (bottom) when treated separately. In
order to reproduce the result of the iterative approach, the
so-derived parameters were all left free (red (dark grey) tri-
angles). As the filling factor of the rings drops significantly
beyond a radius of 320??, we did not fit the outer parts. That
means we also miss most of the distortion at the southern
edge of the galaxy.
NGC5408 has a slowly rising rotation curve coming to
a plateau at a radius of 250??. Receding and approaching side
show a very similar kinematic behaviour up to a radius of
320??, confirming the impression of an evenly rotating galaxy.
The initial estimates already defined the Hi kinematics quite
well as the green and black triangles are in very good agree-
ment. The ’left-free’ approach is also in good agreement with
the other two approaches, except for some scatter between
0??and 100??. This is still far away from the disturbed region,
which can therefore not explain the scatter.
The iterative approach results in an inclination angle
of 62◦, a position angle of 302◦, and a systemic velocity
of 502kms−1(see also Table 5). The systemic velocity is
12van Eymeren et al.
Figure 7. Hi velocity profiles of IC4662. We show the low-resolution Hi velocity field (see Fig. 3, upper right panel) with the Hi profiles
overlaid. Each box represents the sum of all spectra within roughly one beam size (70??×70??). Stars denote spectra with a flux scale
which is four times larger than that of the other spectra.
slightly lower than the HIPASS velocity of 506kms−1(Ko-
ribalski et al. 2004). In comparison to the parameters derived
from optical data, the inclination of the neutral gas is higher
by 10◦. The value of the position angle of the neutral gas is
dominated by the extended emission. As the optical disc is
differently aligned (see Sect. 3.2), the Hi position angle of
302◦differs significantly from the optical position angle of
In order to prove the reliability of the calculated values,
a model velocity field was created by using the best-fitting
parameters (Fig. 9, lower left panel) and afterwards sub-
tracted from the original velocity field (upper right panel).
The residual map is shown on the lower right panel of Fig. 9
with the 3σ contour from the low-resolution Hi intensity
map and the systemic velocity contour overplotted. It is ob-
vious that the original velocity field is very well described by
the derived kinematic parameters. The residuals are gener-
ally at ±5kms−1, except for one region in the south, which
shows residuals of about 15kms−1. This region adjoins the
already mentioned distortion in the velocity field and the
high velocity dispersion. We examine this issue in more de-
tail in Sect. 4.4.2.
4.4A detailed kinematic analysis of the Hi and
Hα line data
We now want to study the Hi and Hα line profiles of both
galaxies in more detail and to compare the peak velocities
of the neutral and ionised gas. Therefore, we performed a
Gaussian decomposition by interactively fitting the Hi and
Hα line emission (IRAF task splot). Only detections above
a 3σ threshold were considered.
Figures 7 and 8 have already shown that the Hi profiles
of both galaxies are sometimes split into at least two com-
ponents. For a better comparison with the Hα images, we
used the high-resolution Hi data cubes (see Sect. 2.2) to per-
form the Gaussian decomposition. As the extracted spectra
are often quite noisy, especially in the outer parts of both
galaxies, only the central parts are fitted. Figures 10 and 11
show the resulting maps: a map with the blue-shifted gas
on the left, a map with the main component (component of
highest intensity) in the middle and a map with red-shifted
gas on the right. The velocities are averaged over 20??×20??
in case of IC4662 and 18??×18??in case of NGC5408, which
roughly corresponds to one beam size. For better visuali-
sation, we overlaid the 3σ Hi intensity contour from the
low-resolution data and the 3σ Hα intensity contour from
the Hα images (Fig. 1).
We then performed a Gaussian decomposition of the Hα
line profiles as measured from the long-slit spectra. We mea-
A kinematic study of the gas in IC4662 and NGC540813
Figure 8. Hi velocity profiles of NGC5408. We show the low-resolution Hi velocity field (see Fig. 6, upper right panel) with the Hi
profiles overlaid. Each box represents the sum of all spectra within roughly one beam size (60??×60??). Stars denote spectra with a flux
scale which is four times larger than that of the other spectra.
sured the peak velocities in order to create position-velocity
(pv) diagrams (Figs. 12, lower row and 13, middle left panel
and lower row). The Hα velocities are indicated by + sym-
bols. The corresponding Hi velocities (solid lines) were ex-
tracted from the peak velocity fields of the main component
(Figs. 10 and 11, middle panels). Both the Hi and the Hα
velocities are not corrected for inclination. Figures 12 and
13 also show the slit positions, plotted over the Hα image
of each galaxy. Position0 in the pv diagrams is marked by a
small circle; the arrows indicate increasing distance from 0
in a positive sense. Examples illustrating the quality of the
spectra and the analysis are given for both data sets in the
upper right panel of Fig. 12 and in the upper and middle
right panels of Fig. 13.
At first glance, the Hα emission line is quite broad along
all slits. Observations of the ionised gas in other dwarf galax-
ies revealed outflows with expansion velocities of about 20 to
50kms−1(e.g., Schwartz & Martin 2004; van Eymeren et al.
2009a,b). For the data used in our analysis, this implies that
it will be difficult to separate blended lines because of the
relatively low spectral resolution (60kms−1for the archival
spectrum of NGC5408, 112kms−1for all other spectra).
Therefore, we fitted the profiles with only one Gaussian un-
less a fit with two Gaussians gave reasonable values, i.e.,
FWHMobs > FWHMinstr. In many cases, a single Gaus-
sian fit led to large residuals, which indicates that the Hα
line is a superposition of at least two components (see, e.g.,
Fig. 12, upper right panel).
In Fig. 14 we present the FWHMs, corrected for
instrumental broadening FWHM2
instr, of all fitted lines vs. the integrated intensity.
Different shades of grey indicate FWHMs measured in spec-
tra from different slit positions. The black stars denote the
data of the archival spectrum of NGC5408.
In the following subsections, we describe the kinematics
of the neutral and ionised gas, point out peculiarities, and
compare the kinematic behaviour of both gas components.
4.4.1 IC 4662
The high-resolution Hi velocity maps of IC4662 show that
the neutral gas often consists of a single component that
follows the overall rotation (Fig. 10). Red-shifted gas can
be seen in the filamentary north-eastern part of the opti-
cal disc of IC4662, which is offset from the main compo-
nent by about 55kms−1. We did not find any more outflow-
ing neutral gas associated with the optical disc. However,
we detected blue-shifted gas north-west and south-east of
the optical disc with expansion velocities of about 40 to
50kms−1with respect to the main component, whose mor-
phology gives the impression of some kind of bipolar outflow.
We then used the Hi velocity values of the main com-
14van Eymeren et al.
Figure 9. The Hi rotation curve of NGC5408 created by fitting a tilted-ring model to the observed velocity field. Top left: different
approaches for deriving the rotation curve. The black triangles represent the iterative approach, the error bars indicate receding and
approaching sides. The green (light grey) curve was derived by taking the initial estimates and keeping them fixed, the red (dark grey)
curve by taking the best-fitting parameters and leaving them free. Top right: the observed Hi velocity field (see Fig. 6). Bottom left:
the model velocity field created by taking the best-fitting parameters. Bottom right: the residual map after subtracting the model from
the original velocity field. The 3σ contour from the low-resolution Hi intensity map and the systemic velocity are overplotted in black.
Figure 10. Gaussian decomposition of the Hi line profiles of IC4662, which were extracted from the high-resolution cube. We show the
blue-shifted (left panel), main (middle panel) and red-shifted (right panel) components of the Hi velocities. The 3σ Hi intensity contour
from the low-resolution cube and the 3σ Hα intensity contour from the Hα image (Fig. 1) are overlaid in black for better visualisation.
We averaged the velocities over 20??×20??, which roughly corresponds to one beam size.
A kinematic study of the gas in IC4662 and NGC540815
Figure 11. Gaussian decomposition of the Hi line profiles of NGC5408. The same as in Fig. 10. We averaged the velocities over
18??×18??, which roughly corresponds to one beam size.
Figure 12. Gaussian decomposition of the Hα line profiles of IC4662. Upper left panel: the continuum-subtracted Hα image. White
arrows indicate the slit positions, white circles mark position 0. Upper right panel: an example spectrum (slit3 at r = 110??) which
shows the flux in arbitrary units vs. wavelength. The black solid line indicates the original spectrum, the light grey line marks the Gaussian
fit. The residuals after subtracting the Gaussian from the original profile are shown in dark grey. Note that the spectral resolution is about
2.5˚ A, which corresponds to 112kms−1. Lower panels: the pv diagrams of the Hα emission. The + symbols represent the Hα velocities
obtained from the spectra, the solid line the Hi velocities obtained from the high-resolution velocity map of the main component (see
Fig. 10, middle panel).
ponent as a reference value for the velocities of the ionised
gas. We obtained optical spectra of IC4662 along three slit
positions (Fig. 12, lower row). Slit 1 intersects the southern
Hii region, the main Hii region complex, and the diffuse
gas in between. South of the main Hii region complex, i.e.,
from −160??to −80??, the ionised gas roughly follows the
Hi velocities. In most cases, however, the subtraction of the
fitted Gaussian from the observed profile reveals two resid-
uals which are symmetrically distributed. The FWHM is
measured to be 70kms−1, which suggests a blue- and a red-
shifted component with a velocity offset of 35kms−1with
respect to the Hi line. Between −80??and −15??, the ionised
gas appears to be slightly red-shifted by 20 to 30kms−1with
respect to the neutral gas.