Transcript An analysis of the HI component of the Magellanic Bridge

An analysis of the HI component of
the Magellanic Bridge
Erik Muller:
University of Wollongong
Australia Telescope National Facility
Supervisors:
Bill Zealey (UOW)
Lister Staveley-Smith (ATNF)
Overview: HI observations of the
Magellanic Bridge
• HI Data collection and reduction
• Magellanic Bridge HI expanding Shell census
– Selection Criteria
– Shell formation mechanisms
• Statistical tools applied to Bridge HI dataset
– Spectral Correlation Function
– Power Spectrum
• Preliminary results for search of CO in emission.
• A quick look at some new HI/Hα features in the
Bridge.
The Magellanic System:
• Detected in HI (spin-flip transition of Neutral Hydrogen) by Kerr, Hindman
& Robinson, 1954 with a 36ft Potts hill (Sydney,Australia) reflector.
• Their nearness (SMC ~60kpc, LMC ~50kpc) makes them an excellent
laboratory in which to observe physical processes with high spatial resolution
• Magellanic system comprises five elements:
–
–
–
–
–
Large Magellanic Cloud (LMC) (Kim 1998)
Small Magellanic Cloud (SMC) (Staveley-Smith 1998, Stanimirovic 1999)
The Magellanic Stream
(Putman, Gibson, Stanimirovic etc. 1998)
The Leading Arm
(Putman 2000)
The Magellanic Bridge
(Mathewson & Cleary 1984)
• Bridge spans the ~14kpc from western edge of LMC to eastern edge of SMC
– Formed through tidal interaction of SMC with LMC (Simulations predict 150200 Myr old - eg. Gardiner & Noguchi 1996)
– Populated by young O-B (>7 Myr), as well as older, stars. (eg. Irwin et al, 1995)
The Magellanic System in HI:
Peak Pixel map, Linear trans. func. Tmax=0.3 MJy/beam
LMC, R~50kpc
Parkes Multibeam data obtained by Putman, M.E.
To the Magellanic Stream
The Magellanic Bridge l~14kpc
To the Leading Arm
SMC, R~60kpc
HI Data collection & Reduction:
• 144 pointings with ATCA (375m configuration)
– ~16 minutes/pointing
• Scanning with Parkes multibeam (inner seven beams)
– Scanning rate: 1o/min
• ATCA Data reduced with MIRIAD
– conventional procedures for data flagging and calibration
• Parkes data reduced on line with ‘LIVEDATA’
– Bandpass calibrations, velocity corrections
• Cube Merging:
– Parkes and ATCA data merged post-convolution using IMMERGE
(Stanimirovic, PhD, 1999)
• Resulting cube:
–
–
–
–
~7ox7o region, Vel range~100-350 km/s (Heliocentric)
RMS ~ 15.2 mJy/Beam (eq 1.7x1018 cm2 for each channel)
98” spatial resolution
~2x108 M (SMC ~4x108 M)
Right Ascension-Declination
RMS=15.2 mJy/beam
(1.7x1018 atm cm-2)
Velocity-Declination
?
38 km/s [VGSR]
8 km/s [VGSR]
Right Ascension-Velocity
Peak pixel maps of ATCA/Parkes HI
datacube
Total observed HI Mass=200x106 M
Mass of centre region=72x106M
(2 x 4.7)kpc cylinder ρ=0.2 atm cm-3
(2 x 4.7 x 5)kpc slab ρ=0.06 atm cm-3
Bridge HI Shell census:
• Search and census of HI shell features reveal 163 expanding
shells within a ~1.9x108 M of HI.
• Strict shell selection criteria must be satisfied
• Bridge shells have kinematic ages similar to those found in the
SMC, and parameters similar to other galaxies
• Distribution of shell centres and positions of OB associations is
generally very poor.
• Roughly, High HI column density and spatial distribution of OB
associations correlate well. Not quite so well at a more detailed
level.
Formation mechanisms of HI expanding
Shells:
• Stellar wind and SNe driven shells (Weaver et al, 1977):
– Hot, energetic stars ionise local gas, and blow open an
expanding sphere of hot gas.
– Study by Rhode et al (1999) on Holmburg II galaxy find that
the distribution and brightness of HOII clusters do not
support SNe as expansion mechanism.
• HVC collisions (Tenorio-Tagle 1987, 1988; Ehlerova & Palous, 1996)
– Capable of producing low energy, spherical expanding
structures for impacts by low Ek clouds. Rc ~10pc
– Difficult/impossible to differentiate from stellar wind
formation mechanism.
– Distribution of impact sites should be uniform.
Formation mechanisms of HI expanding
Shells (cont):
• Gamma Ray Bursts (Efremov, Elmegreen & Hodge, 1998;
Loeb & Perna,
1998)
– Release relatively large amounts of energy (10% of
progenitor mass) ~1053 erg
• Shells formed from GRB are more energetic for lower
radii and more quickly expanding shells.
• GBR frequency in a our galaxy ~0.1 Myr –1 (Portegies
Zwart, & Spreeuw, 1996). Given the relatively low age
and stellar population of the Magellanic we expect this
frequency to be significantly reduced.
• Ram pressure (Bureau et al, 2001)
– ISM Inhomogeneities external to, and impacting on Bridge
gas will expand any ‘dimples’ and isolated low-density
regions into much larger voids and holes.
Shell selection Criteria
•
Adapted from Puche et. al. (1992)
i.
A (rough) ring shape in all three projections (RA-Dec, RA-Vel, DecVel), must be present across the velocity range occupied by the shell
ii.
Shape is observable across at least three velocity channels (~5km/s)
iii. Ring Must be rim-brightened relative to ambient column density of
channel
Criteria target rim brightened, expanding spherical structures (not
cylindrical or blown out volumes)
To reduce subjectivity, criteria must be strictly satisfied!
•
We assume a stellar wind model (Weaver, 1977)
HI, OB associations and Shells appear to correlate well.
HI Peak Pixel map. Size and location of 163 Magellanic Bridge HI expanding shells.
Crosses locate OB associations (Bica et al. 1995) Ret
Comparison of Magellanic Bridge shells to
SMC population:
5 Rs 
Ts=2V 
 exp
Rs 5 Ts -3 no 
Ls=1.5x10 100pc 106yr cm-3L


 

5
Magellanic Bridge SMC
Mean
Stdev
Mean
Stdev
Kinematic Age (Myr)
6.2
3.4
5.7
2.8
Shell Radius (pc)
58.6
33.2
91.9
65.5
Expansion Velocity (km/s)
6.5
3.8
10.3
6.3
Energy (log [ergs])
48.1
51.8 (n=1 cm3)
• Bridge shells, compared to the SMC population are (on average)
• Marginally older, 60% smaller + expand 60% more slowly
• Much less energetic:
– Mean energy of Bridge: 48.1 (log erg)
– Mean energy for SMC: 51.8 (log erg).
MB and SMC HI shell population
Dynamic
age
Radius
Expansion
Velocity
Decreasing shell
Decreasing
radius with
expansion velocity
increasing RA
with increasing RA
Discontinuity
at
Discontinuity
• Generally
continuousat
MB/SMC
MB/SMC
age distribution
transition
transition(from
(from
• Slightstrict
dominance
by
critera)
strict critera)
older shells at higher RA
Comparison of power law parameters of
expanding HI structures from other surveys
αx = 1-γx
Holmberg
II
SMC
Magellanic
Bridge
(Puche et al.
1992)
(StaveleySmith et al.
1997)
Number of Shells
51
509
163
Expansion Velocity αv
2.9±0.6
2.8±0.4
2.6±0.6
Shell Radius αr
2.0±0.2
2.2±0.3
3.6±0.4
• αv is in agreement with other systems
• αr is much steeper for the Magellanic Bridge population
– Due to a strict selection criteria that manifests as an overall deficiency of
small radii shells, and ultimately as an older shell population.
Distribution of OB, shells and HI
Black: Left and mean integrated HI in 1.5’x1.5’ region, centred on 198 OB
positions.
White: Right and entire integrated intensity map, binned into 1.5’2 areas.
•OB positions generally correspond
regions of higher HI
•50% of OB associations correlate
with a Mean HI column
density>1.2x1021 atm cm-2
•Be wary of selection effects
HI around OB associations
•HI ramps almost
linearly to centre of
OB positions
•Excess of HI <80pc
of association centre,
in disagreement with
Grondin & Demers,
1993.
•OB associations
don’t appear to sweep
out a significant hole
in the local HI.
Diamonds: Mean HI averaged in concentric annuli around OB catalogued positions.
Triangles: Mean HI averaged in concentric annuli, offset 90pc (10 pixels) south of OB centres
Error bars mark one standard error of the mean, vertical line marks resolution of Parkes observations
by Matthewson, Cleary & Murray (1974)
Distribution of OB associations and
HI shells
• As shown earlier: Visually, OB associations, HI and shell
centres appear to correlate reasonably well.
• A more quantitative study shows that:
– ~50% of shells have one or more OB association within 8’
(140pc)
– ~18% of shells have one or more OB associations within 3.5’
(60pc) (mean shell radius)
• A study of the variation of the HI column density around OB
associations shows in fact and excess within ~80pc of each
association (!)
These call into question the commonly used wind/SNe model
Do alternative expansion theories help?
Formation mechanisms of HI expanding
Shells:
• Stellar wind and SNe driven shells (Weaver et al, 1977 and Chevalier, 1974)
– Hot, energetic stars ionise local gas, and blow open an
expanding sphere of hot gas.
• The most recent burst of star formation 10-25Myr ago (Demers &
Battinelli 1998) , C/W mean shell kinematic age ~6Myr
• ‘Constant energy input rate is generally invalid’ (Shull, & Saken 1995)
• Input from WR and stellar wind at 3~10Myr for coeval and non-coeval
associations, mis-estimation of age by up to 40% - lower limit of
starburst date by Demers & Battinelli
• Bridge Associations & Clusters are very poorly populated, typically N ~
8 (N increases towards SMC), some Associations & Clusters ‘may be of
type later than O-B’ (priv comm. Bica 2002)
• Very poor correlation with Bridge Shells (see also Rhode et al. 1999)
• A local HI excess(!) appears co-incident with OB association positions.
Formation mechanisms of HI expanding
Shells:
• Gamma Ray Bursts
(Efremov, Elmegreen & Hodge, 1998; Loeb & Perna, 1998)
– Release relatively large amounts of energy (10% of progenitor
mass) ~1053 erg
• Unlikely to be a significantly frequent occurrence to explain observed
shell population for relatively young and poorly-populated stellar
component of the Bridge.
• GRB Expansion velocities are ~10-2 of velocity for shells of sizes found
in the Bridge.
Formation mechanisms of HI expanding
Shells:
• HVC collisions (Tenorio-Tagle 1987, 1988, Ehlerova & Palous, 1996)
– Capable of producing low energy, spherical expanding
structures for impacts by low Ek clouds. Rc ~10pc
• surface distribution of shells shows a tendency higher HI col. dens.
– HVC collisions in tenuous gas may create very asymmetrical and
fragmented shells – will not be included in the survey!
– An further survey of incomplete shell-like features is necessary
before much can be said about the effectiveness of this mechanism.
• Ram pressure drag (Bureau et al, 2001)
– Mechanism proposed that enlarges ‘dimples’ into deeper holes
through interaction of Gas with nearby inhomogeneities
• Hole produced this way would not be a complete shell
• Holes form this way appear on ‘skin’ of HI mass, MB shells generally
deeply embedded.
Statistical tools:
• Statistical tools provide a means to
– compare populations of similar objects between different
systems
– Understand and model general trends and behaviours.
– Distinguish between sub-populations
• Spectral correlation function (SCF): Measures spectral
similarity as a function of radial separation
• Power spectrum analysis (PS): Measures power as a
function of scale, and as a function of velocity range.
• Both SCF and PS have been used to infer information
about the third spatial dimension.
Spectral Tools 1:
• Spectral Correlation function: (Rosolowsky et al. 1999)
– Compares two spectra separated by Δr, and makes an
estimate of their ‘similarity’ (see later)
– A 2D map of mean SCF shows rate of change (or degree of
correlation) of SCF with Δr and θ
– Has been used to confirm a characteristic length for the scale
height of the LMC, by measuring the radius of de-correlation
(Padoan et al. 2001)
– In this case, SCF shows that MB spectra has a longer decorrelation length in the east-west direction. (Tidal
stretching)
Spectral Tools 2:
• Spatial power spectrum
– Used to show the range of spatial scales present in source
– Highlights any process favouring a particular scale. (Eg.
Elmegreen, Kim, Staveley-Smith, 2001)
– Using velocity averaging, is can be used to show the relative
contributions of density and velocity dominated fluctuations.
(Lazarian & Pogosyan, 2001)
– P-spect shows no characteristic scale in the MB.
Spectral Correlation function
How it works:
So (r ,  
So (r ) 
S    
So (r, r )  1 
2


T
(
r
,
v
)

T
(
r


r
,
v
)
r
 T (r, v)   T (r  r , v)
2
r
S0, N (r )  1 
1
Q(r) 
N
r
r
1
Q( r )
2
T
(
r
,
v
)
dv
r
W
2
r
Δr Δr Δr
SCF output maps:
T maps
55 pixels
37 pixels
(=2/3 NT)
SCF maps
Fits in E-W and N-S directions (central 5 rows/columns)
ΣT=1.0x106 K.km/s
ΣT=7.5x105 K.km/s
ΣT=8.4x105 K.km/s
ΣT=9.4x105 K.km/s
ΣT=1.0x106 K.km/s
ΣT=1.1x106 K.km/s
ΣT=1.1x106 K.km/s
ΣT=1.1x106 K.km/s
•Positive and negative departures from log-log fit after a varying length
(~250-380pc ~ 14’-22’ at R=60kpc)
•-ve departures seem to exist only for sub images where signal is lower
and less well distributed throughout region of interest.
Spatial Power spectrum
• Measures the rate of change of power with spatial scale
– Works on Fourier inverted image data (edges are rounded by
convolution with a small (~30 pixels) gaussian)
– Channels with significant signal selected (60 channels)
– Filtered to reduce leakage from low spatial frequencies
(image convolved with 3x3 unsharp mask, then divided back
into FFT data)
– Un-observed UV data is masked out.
– Power-law fit to remaining dataset (γ) (use IDL poly_fit).
– A range of velocity increments are examined to determine
the relative contributions of density (thin regime) and
velocity (thick regime) fluctuations.
Spatial Power spectrum cont.
ATCA + Parkes data
(+Gaussian rounding)
FFT (im2+r2)
Spatial Power spectrum cont.
Power law fit for
Brightness2 [K2]
γ – velocity binsize
Transition from thin to thick regime
(velocity to density dominated regime)
General result:
• All Power spectra, for all velocity bins are featureless and well
fit with by a single power law:
• No processes present that lead to a dominant scale (c/w LMC)
• More ‘3 dimensional’ than the LMC (Similar to SMC). i.e. no
characteristic thickness.
• Power spectra steepen for increasing velocity bin size
(ΔV~<20km/s)
• Transition from ‘thin’ velocity dominated (spectral ΔV ~< integrated
ΔV thickness) to thick, density dominated regime.
• γ changes from ~-2.90 - ~-3.35, consistent with Kolmogorov
Turbulence. (Lazarian & Pogosyan, 2000)
• Source of turbulence?
– Processes that do not show a scale preference:
• Stirring & instabilites from tidal force of LMC and SMC?
• Energy deposition into ISM from stellar population?
PS from other systems:
• LMC (Elmegreen, Kim & Staveley-Smith, 2001)
• much steeper; γ ~<2.7 (Entire velocity range, two linear fits)
• LMC spectra turns over at r~100pc
– attributed to line-of-sight thickness of LMC.
• SMC (Stanimirovic, 1999; Stanimirovic & Lazarian, 2001)
• SMC and MB cover same range of γ:
– γSMC~ 3.4 at ΔV ~100km/s
– γMB~ 3.3 at ΔV ~100km/s
• linear (featureless) over entire range of Δv
• does not appear to approach a characteristic Δv
• Galaxy (Dickey et al. 2001)
• Analysed for smaller range of Δv (0-20 km/s)
• Inner Galaxy γ ~ -2.5 - -4, consistent with Kolmogorov turbulence.
• All systems show steepening of γ with ΔV.
SMC and Galaxy γ with ΔV
SMC γ with ΔV. (Stanimirovic & Lazarian,
2001)
Galaxy γ with ΔV. (Dickey et al 2001)
(N.B. Inverted γ scale, linear ΔV scale)
Star-formation sites in the
Magellanic Bridge
• The Bridge has a young stellar population
– O-B star ages range >7 Myr
• SMC has v. low Metallicity (Rubio et al 1993)
– ZMB < ZSMC
• Bridge formation began ~150-200 Myr ago
• Stars are forming in the Bridge
– Where are they forming? - Any molecular clouds?
• Some recently discovered H2 in the Eastern Magellanic Bridge
(Lehner, astro-ph 0206250, 2002)
• Any other sites of star formation (Lehner et al 2001, Smoker et al 2000)?
Candidate 12CO(1-0) emission sites
•Local 60m Maximum = dust = surface chemistry.
•Local HI Maximum = absorbing layer and raw material
HI Int. Int., 60m contours:
HI Integrated intensity
(0.4-3.4 +1.9MJy/str)
60m
Pointing Offsets by 1 BW (~45”)
HI and CO spectra
suggest the CO emission
region is imbedded
within an HI cloud:
Right: CO Spectra at pointing 1
(solid line and Left axis) overlaid
on HI spectra (dotted line and
right axis).
•
12CO(1-0)
detected in emission in the Magellanic Bridge
– Star-formation through molecular cloud collapse
– Star-formation is a current, active process in the Bridge, even close to
SMC.
– Stellar population develop in situ, and are not transported from remote
location.
• Further Evidence of Star-formation in a low-metallicity, tidal
structure
Hα structures in the Magellanic Bridge.
– One shell has been previously detected in Hα emission that
corresponds to an HI shell (Meaburn 1985)
• Age calculations from spectroscopic studies disagree with HI shell
age estimates by ~10%. (Graham, Meaburn & Bryce (2001)
• Contains a UV source (FAUST 392)
– Other Hα + HI shells and Hα filaments have since been
found
• All existing Hα structures in the Bridge are nearby to a FAUST
object. (within 2-3”)
• Pending further investigations
– HI + Hα shells are not significantly different
from mean Magellanic Bridge HI shell in terms
of radii, expansion velocity, energy etc.
Hα structures
HI:
Hα
Summary:
• General appearance:
– ATCA and Parkes have uncovered chaotic and intricate structure of HI
comprising the Magellanic Bridge.
– Loops, filaments and clumps observable to smallest scales of 98” (~29pc)
– Much of the Bridge is bifurcated into two velocity sheets, converging at
~2hr 30min
– Large loop R~1kpc off the northeastern edge of SMC.
• Shell survey:
– 163 shells found within the Magellanic Bridge
– Kinematic age is consistent with that of , shells of the SMC although
Magellanic Bridge shells are considerably smaller and less energetic.
– Power law distribution of expansion velocity is consistent with HoII and
SMC.
– Strict selection criteria is insensitive to incomplete and fragmented shells
Summary:
• Shells, stars and HI :
– Good correlation of HI with OB assocations and Clusters, and also with
HI shells (NB. Selection criteria), Poor correlation of OB associations
and clusters with expanding shell centres
– HI distribution about OB associations and Clusters shows a mean excess
at short radii (<80pc), and a decreasing slope with increasing radii
• Shell formation:
– Shell Energies and spatial distribution do not agree with theories of
formation by stellar wind by OB associations and Clusters or by SNe
– Frequency of GBRs is too low to be generally applied to Magellanic
Bridge shells.
– HVCs are capable of producing the observed structures, however, the
surface distribution shows preferential distribution (selection effects!),
and many shells are found deeply embedded throughout the HI Bridge.
– Alternatives ??
Summary:
•
12CO(1-0)
detected in emission for the first time in the
Magellanic Bridge
– Indication of star formation from molecular cloud collapse
(c/w shock triggered star formation)
– Star formation within tidal structure and a very low
metallicity environment
• This presents a unique and opportunity to study star-formation under
these conditions at high spatial resolution
• Hα filaments throughout the Bridge:
– A few are shown to be associated with an expanding HI
structure.
– All have FAUST objects nearby, the likely source of
ionisation.
Summary:
• SCF
– In general, decorrelation of spectra separated by Δr occurs at ~200-400pc
• Estimated thickness of MB is ~5kpc, based on distance measurements for
two OB associations separated by ~7’ (Demers & Battinelli, 1998)
• Results of SCF are difficult to interpret in the same way for LMC, PS
analysis may help.
– SCF behaves strangely for datacubes containing low S/N
– The line of minimum rate of change of SCF is points almost, but not
quite, E-W, towards the SMC and LMC.
• Power Spectra
– There is no suggestion of a departure from a power law fit to MB spatial
power spectra, despite a decorrelation at ~200-400pc found using SCF.
(c/w Padoan et al, 2001)
– PS shows transition from γ =~-2.9 to γ =-3.35, through thin to thick
regime, consistent with Kolmogorov turbulence.