High mass star cluster formation

The compact mm sources of W51 and Sgr B2

and what this suggests for analysis of ALMA-IMF

Adam Ginsburg

Source Extraction

Two methods were explored in Ginsburg+ 2017 section 3.1.1:

Dendrogram extraction:

Automated, reproducible
Problematic for extended structures we're not interested in (HII regions)
Difficult to use if noise varies (e.g., if bright sources produce noisy sub-fields)

Hand extraction

Mildly tedious, unreproducible (e.g., on simulations)
Reliable: sources are real if you say they are
Difficult to estimate completeness

Hand-selected cores

Dendrogram-selected cores

Hand-selected cores

Dendrogram-selected cores

Source Photometry (Section C.2)

Contour-based photometry is not great for core or protostar analysis:

Can't compare contour-based to hand-extracted
Can't (shouldn't) use contours to extract from other wavelengths (e.g., free-free correction)
Contours require 'truncation correction' because they only go to the specified noise threshold, but signal goes all the way down

So, we use circular apertures.

Multiple apertures yield a radial profile
"Concentration" is a measure of source structure

Free-free contamination measurement

Temperature Inference

Interpolation of HiGal data does poorly; hot (~100-200K sources dominate)

Fred will tell us about PPMAP later?

Used line \(T_B\) to get lower-limit temperature:

Assuming dust & gas are mixed, \(T_{dust}\geq T_{B,max}\)
For each source, extract spectrum across full band
Fits to each line for each source, bad ones rejected

Effect of adopting temperature peak

Which species is dominant?

Spatial and Spectral distribution

Sources are extended: What constitutes a 'core'?

We fit power-laws to the flux and mass distributions

Fit each aperture excluding free-free fluxes
Powerlaw fit for apertures 0p2: 1.631315423503212+/-0.08083165707886612     xmin: 0.0064744025027009995    n: 61
Powerlaw fit for apertures 0p4: 1.5658370824846624+/-0.07128877156346468     xmin: 0.011868132040355701    n: 63
Powerlaw fit for apertures 0p6: 1.5529861072430189+/-0.07080261581763395     xmin: 0.01943433602461    n: 61
Powerlaw fit for apertures 0p8: 1.9104050128602124+/-0.1785451125325736     xmin: 0.24321248022004    n: 26
Powerlaw fit for apertures 1p0: 1.4623503245632266+/-0.06070957184993224     xmin: 0.026001245374699996    n: 58
Powerlaw fit for apertures 1p5: 1.420579507665507+/-0.054754788083823426     xmin: 0.03052176796733    n: 59

Fit each aperture excluding free-free MASSES, THEN starless-only also
Powerlaw mass fit for apertures 0p2: 1.912047687552428+/-0.11975772845046809     xmin: 5.20167867846    n: 58
Powerlaw mass fit for starless apertures 0p2: 2.6088696890146474+/-0.41540836745144233     xmin: 4.47311821623    n: 15
Powerlaw mass fit for apertures 0p4: 1.815021606866669+/-0.10435282362110364     xmin: 10.2104888538    n: 61
Powerlaw mass fit for starless apertures 0p4: 2.3926434280095217+/-0.359578986925739     xmin: 8.45440665685    n: 15
Powerlaw mass fit for apertures 0p6: 1.7553097307339227+/-0.09670750130612725     xmin: 15.0538010421    n: 61
Powerlaw mass fit for starless apertures 0p6: 2.8898077334723236+/-0.5050723618235171     xmin: 16.4954024498    n: 14
Powerlaw mass fit for apertures 0p8: 1.785955620178369+/-0.11115090974796064     xmin: 29.1094782263    n: 50
Powerlaw mass fit for starless apertures 0p8: 5.749488311948332+/-2.1240357447714393     xmin: 49.827658381    n: 5
Powerlaw mass fit for apertures 1p0: 1.8731582232383897+/-0.13163355495886897     xmin: 51.7280690487    n: 44
Powerlaw mass fit for starless apertures 1p0: 5.273575112776541+/-1.7446797339599336     xmin: 55.5714845757    n: 6
Powerlaw mass fit for apertures 1p5: 2.7917077019490932+/-0.4479269254872733     xmin: 420.574512148    n: 16
Powerlaw mass fit for starless apertures 1p5: 3.2306486231067537+/-0.7053930308535842     xmin: 47.5659895928    n: 10

Flux slope is shallower for bigger apertures
"Starless" much steeper than "starry"
Mass slopes steeper than flux slopes

Completeness Testing

In the paper, I did not discuss completeness testing at all. However, I have well-tested ideas of how to do this.

This code, and several related scripts, take Aquila and Perseus data, rescale them to 1mm at d=5 kpc, push them through the ALMA simulator with the exact UV coverage of the observations, then extracts a dendrogram catalog from them.

When I attempted this, the CASA process was buggy and unsatisfactory. Details are recorded in some blog posts:

I don't remember the conclusions, but I think they were that we wouldn't see anything (or at least, not enough to characterize completeness) from Aquila/Perseus at the distance of W51, which isn't terribly surprising.

Completeness testing usefully

The framework I've set up will work (demo on next few slides)

To make this useful, we need more reasonable simulations to compare against, e.g., hydro sims with PPP-identified cores, or synthetic IMFs⇾CMFs⇾images