Slide 0 - Instructions:
  • To move slides, use the arrow keys or swipe on your mobile device
  • To see the speaker notes, press "s"
  • To go to full screen, press "f"
  • To print as PDF, go to this URL: ?print-pdf, then print.
  • To get a PDF with speaker notes, add ?print-pdf&showNotes=true to the URL.

High-Mass Star Formation:
Feedback and Fragmentation

original title: thermal and chemical feedback during the early stage of high-mass star formation
$\tau_{travel} = 36 h$, $L_{travel} = 18,462$ km, N=4
  • Postdocs: Miriam Garcia Santa Maria (2024-2025), Allison Towner (2020-2023)
  • PhD: Desmond Jeff [PhD 2025], Theo Richardson [PhD 2025; postdoc], Alyssa Bulatek, Nazar Budaiev, Savannah Gramze, Taehwa Yoo
  • Postbac: Derod Deal, Aden Dawson
  • Supported by NSF 2008101, 2206511, CAREER 2142300, STSCI 1905, 2221, 3523, 5365, 6151, Astropy
Collaborators: ACES team, ALMA-IMF team, DIHCA team. John Bally, Ashley Barnes, Cara Battersby, Roberto Galván-Madrid, Jonathan Henshaw, J. M. Diederik Kruijssen, Steven Longmore, Xing Lu, Fanyi Meng, Elisabeth A.C. Mills, Juergen Ott, Justin Otter, Álvaro Sánchez-Monge, Peter Schilke, Daniel Walker, Erik Rosolowsky, Eric Koch, Ciriaco Goddi, Brett McGuire, Dick Plambeck, Melvyn Wright, Johan van der Walt, Carlos Roman-Zuniga, Henrik Beuther, Kei Tanaka, Yichen Zhang, Timea Csengeri, Fabien Louvet, Nichol Cunningham, Frederique Motte, Patricio Sanhueza, Thomas Nony, Yohan Pouteau, Melisse Bonfand, Fernando Olguin, Sylvain Bontemps, and many others
Slides available at https://keflavich.github.io/talks/sfde2025.html
or from my webpage →talks

Massive star clusters are detected in other galaxies, resolved in ours

NGC 253 protoclusters (Leroy+2018)
NGC 4945 protoclusters (Emig+2020)

Proto-clusters were common in the early universe

Two ALMA large programs probe these conditions:
ALMA-IMF:
  • 15 High-mass protoclusters probing young (few stars) to old (blowing out gas)
  • Two ALMA bands looking at cold gas & dust
  • Data published, analysis ongoing
ACES:
  • The whole Galactic Center, with low-SF and high-SF clouds
  • One ALMA band, biggest mosaic made.
  • Data all delivered, mostly reduced

Most stars form in dense regions

Lada & Lada 2003:
5-10% in bound clusters
in our Galaxy
JWST life cycle of clusters: W51e → W51 IRS2 → Wd2 → Wd1

Massive star clusters are not isolated

IRS 2
Main

Massive star clusters are not isolated

Massive star clusters are not isolated

forming cluster [ALMA] alongside the giant HII region

Cores fragment. Massive stars cook their neighbors in hot cores.

Hot Cores

Hot cores are chemically rich sites of high-mass star formation.
Bounded at ~100 K

Hot Cores

Hot cores are chemically rich sites of high-mass star formation.
Accretion is ongoing
all MYSOs make hot cores

Hot cores:

My story of high-mass star formation:
  • They start as normal-ish looking seeds:
    there is no massive prestellar core.
    [ALMA-IMF, ALMA-ATOMS, ALMAGAL all confirm this]
  • They accrete for some time, during which their core must be replenished
  • They have a disk most of this time, but it is not a "classic" T Tauri disk, it's small and constantly changing directions: accretion is chaotic
  • Massive YSOs spend most of their accretion mass (maybe time) luminous, cooking their surroundings: all MYSOs make hot cores
Theo Richardson's models turn this story into predictions

Theo Richardson
Including source count and flux predictions spanning whole clusters
Richardson+ in prep
See Josh Peltonen's talk next...
My story of high-mass star formation:
  • They start as normal-ish looking seeds:
    there is no massive prestellar core.
    [ALMA-IMF, ALMA-ATOMS, ALMAGAL all confirm this]
  • They accrete for some time, during which their core must be replenished
  • They have a disk most of this time, but it is not a "classic" T Tauri disk, it's small and constantly changing directions: accretion is chaotic
  • Massive YSOs spend most of their accretion mass (maybe time) luminous, cooking their surroundings: all MYSOs make hot cores
No massive starless cores in ALMAGAL or ALMA-IMF
Coletta+ 2025 ALMAGAL
(stuff above $\sim10 \mathrm{~M}_\odot$ is protostellar)
My story of high-mass star formation:
  • They start as normal-ish looking seeds: there is no massive prestellar core.
    [ALMA-IMF, ALMA-ATOMS, ALMAGAL all confirm this]
  • They accrete for some time, during which their core must be replenished
  • They have a disk most of this time, but it is not a "classic" T Tauri disk, it's small and constantly changing directions: accretion is chaotic
  • Massive YSOs spend most of their accretion mass (maybe time) luminous, cooking their surroundings: all MYSOs make hot cores

Massive, hot cores exist

MYSOs do have cores, those cores must grow in parallel
~200 $\mathrm{~M}_\odot$ hot core
My story of high-mass star formation:
  • They start as normal-ish looking seeds: there is no massive prestellar core.
    [ALMA-IMF, ALMA-ATOMS, ALMAGAL all confirm this]
  • They accrete for some time, during which their core must be replenished
  • They have a disk most of this time, but it is not a "classic" T Tauri disk, it's small and constantly changing directions: accretion is chaotic
  • Massive YSOs spend most of their accretion mass (maybe time) luminous, cooking their surroundings: all MYSOs make hot cores
Chaotic accretion in another ~200 M$_\odot$ hot core

The outflow (& disk) around W51 North changed direction by ~50 deg in < 100 years.

0.25-0.5 M accreted

We don't know the frequency of these events; they are likely bright transients, but most often heavily dust-obscured
DIHCA: Digging into the Interior of Hot Cores with ALMA
Olguin+ 2025/in prep: kinematic mass measurements of 31 objects with $M>8 M_\odot$ (see his poster)

The 'disks' are messy.
My story of high-mass star formation:
  • They start as normal-ish looking seeds: there is no massive prestellar core.
    [ALMA-IMF, ALMA-ATOMS, ALMAGAL all confirm this]
  • They accrete for some time, during which their core must be replenished
  • They have a disk most of this time, but it is not a "classic" T Tauri disk, it's small and constantly changing directions: accretion is chaotic
  • Massive YSOs spend most of their accretion mass (maybe time) luminous, cooking their surroundings: all MYSOs make hot cores
  • Massive YSOs spend most of their accretion mass (maybe time) luminous, cooking their surroundings: all MYSOs make hot cores

Many stars in clusters form from cooked gas.

Cooking may suppress further fragmentation.
Our program: Cores → YSOs
(no YSO, 1 YSO, >1 YSO) = (25%, 30%, 45%) W51-E, (53%, 39%, 8%) IRS2

Taehwa Yoo+: fragmentation
toward W51
(no YSO, 1 YSO, >1 YSO) = (25%, 30%, 45%) W51-E, (53%, 39%, 8%) IRS2

Taehwa Yoo+: fragmentation
toward W51
(no YSO, 1 YSO, >1 YSO) = 17%, 37%, 45% Sgr B2 ("cores" are bigger)

Budaiev+ 2024: fragmentation in Sgr B2
20% of YSOs in W51 lie within the $>100$ K contour
Trend: More massive cores fragment more.


Yoo+ resub.: W51
Budaiev+ 2024: Sgr B2
Suggestive Trend: More massive cores make more massive single stars


Yoo+ resub.: W51
Budaiev+ 2024: Sgr B2

The number of detectable fragments goes up, but the mass gets more concentrated into the most massive object.
Fragmentation summary:
  • N(YSO) > N(core) [maybe obvious, but now empirical]
  • More massive cores contain more fragments
  • ...but more massive fragments dominate their cores more
  • 20% of YSOs reside in hot cores: exposed to hot core chemistry even if they didn't form there

See also Qiuyi Luo's poster on multiplicity: similar data, but different analysis
My story of high-mass star formation:
  • They start as normal-ish looking seeds:
    there is no massive prestellar core.
    [ALMA-IMF, ALMA-ATOMS, ALMAGAL all confirm this]
  • They accrete for some time, during which their core must be replenished
  • They have a disk most of this time, but it is not a "classic" T Tauri disk, it's small and constantly changing directions: accretion is chaotic
  • Massive YSOs spend most of their accretion mass (maybe time) luminous, cooking their surroundings: all MYSOs make hot cores
Results so far have been (mostly) published and well-digested.
Now for some fresh JWST results, starting with pictures
Sgr B2 with JWST on MEERKAT: Ionized gas shows layers of the cloud
MIRI reveals deeply embedded gas
MIRI 25 micron shows the outflow:
first IR light from within Sgr B2 N
There is an overall asymmetry in the star formation seen both in JWST (HII regions) and ALMA (embedded YSOs)
What ALMA sees, JWST doesn't: 0/700 point source matches, though still investigating extended sources (HII regions, outflow cavities)
480 405 187
182 162 140
360 335 210
480 410 405
Interesting feature highlight reel
IRS2E is saturated at all wavelengths $\lambda >2 \mu\mathrm{m}$
The extinction-producing layers seen at short wavelengths are not the ALMA star-forming gas:
this hints at edge-on filamentary (or planar) structures

The Galactic Center Dust ridge...
...has a foreground cloud in front
We measure ice via stellar absorption
  • YSO model grid is testing accretion models
  • Sgr B2 is forming & fragmenting YSOs - asymmetrically
  • W51 fragments are dominated by most massive
  • JWST measures ice abundance across the galaxy
  • We can measure metallicity with ice
past here is probably stuff to exclude

W51 e2e: Massive YSO

CS v=0 J=1-0 and v=0 J=2-1 masers may trace the disk?

M = 24-10+12M
if the masers trace a disk

CS maser conditions

van der Walt+ 2020
  • Top: CS J=1-0, Bottom: CS J=2-1
  • Red: Consistent w/W51e2e observations
  • Masers do not coexist; require different specific CS column
    (N2-1=1015.6, N1-01016.1 cm-2)
  • Require high abundance (XCS > 10-5)
  • Hot (300-500 K), moderate-density (n~105 cm-3): Disk surface? Or outflow cavity wall?
How does the CMF become the IMF?

Some cores fragment, some disappear


Taehwa Yoo+: fragmentation
toward W51

What's inside the cores?

What's inside the cores?

What's inside the cores?

Top-heavy IMFs occur in clusters...
NGC 3603 & Westerlund 1

SSCs are common in starburst nuclei and drive galactic outflows

NGC 253 protoclusters (Leroy+2018)
NGC 4945 protoclusters (Emig+2020)

Star formation drives the evolution of the universe

Most stars in most galaxies formed long ago

Galaxies were smaller & denser back then

STSCI; Pappovich, Ferguson, Faber, Labbe

ALMA-IMF:
The CMF is shallow (top-heavy) in HMSFRs

Pouteau+ 2022 W43-MM2/3 CMF
Motte+ 2018 W43-MM1 CMF: α≳-1

ALMA-IMF:
The CMF is shallow, and steepens with time?

Louvet+, subm
ALMA-IMF Line Data: CH3CN, CH3CCH
Temperature measurements with per-pixel rotation diagrams
Jeff+ 2024 (CH3OH in CMZ), Wyrowski+ in prep (CH3CN in ALMA-IMF)
Hot cores in ALMA-IMF: From rare objects to a population
Cores with line forests
TD>50 K
TG ≳100K

Hot core overview:

  • 9 HCs in W43-MM1 (Brouillet+ 2022)
  • ~60-70 HCs in ALMA-IMF sample from CH3OCHO (Bonfand+ 2024; left)
  • ~10% of continuum cores are within hot cores
  • CH3CN temperature maps (Wyrowski+ in prep)

ALMA-IMF Key Results summary

  • Rich, science-ready data (Ginsburg+ 2022, Cunningham+ 2023)
  • CMF is top-heavy in HMSFRs (Pouteau+ 2022, Louvet+ 2024)
  • Core fragmentation is not 1-to-1 [WIP] (Budaiev+ (CMZ), Yoo+ (W51), Louvet+ (W43), ...)
  • 10% of $M\gtrsim1\mathrm{~M}_\odot$ cores are hot cores (Brouillet+ 2022, Bonfand+ 2024, Wyrowski+)
  • Outflow feedback builds over time, sets initial conditions for many cores (Nony+ 2022, Towner+ 2023)

More Brinary disks

Ginsburg+ 2023: New sample. Miriam Garcia Santa-Maria is following up

Briny chemistry:

Salts, SiS, SiO, and PN are seen together [but limited spectral coverage]

Clusters are sites of interactions & collisions


The BN/I/x interaction is the poster case of accretion ended by dynamical interaction.
...even though Orion is only a medium-mass proto-open-cluster

Classic model: a "core mass function" maps to the IMF

Many alternatives, no consensus on which is best

However, a core is not a core.
Cores change states

...but the naïve version doesn't work

ALMA-IMF "Core" Mass CDF

The CMF in protoclusters is shallower than the IMF
Louvet+ 2024
Context: Recent CMF measurements
Many! Differ in: resolution, algorithm, selection.
Nearby surveys are only-starless; more distant are starless + protostellar

Publication Distance Ncores Mmin Mmax Resoln [au] Slope "stage" Figure
Louvet+ 2024 2-6 330 1.64 200 2000 0.97 HM
Zhang+ 2024 0.4 927 0.3 20 8000 1.4 LM
Armante+ 2024 2.4 80 0.03 13.2 2000 1.44 HII
Cheng+2024 4.5 183 0.5 10 3000 1.15 HM
Pouteau+ 2023 6 205 0.8 70 2500 0.93 HM
Li+ 2023 0.7 570 4 200 15000 1.35 LM
Suárez+ 2021 7.1 40 2.0 59 1000 1.11 HII
Könyves+ 2020 0.4 292 0.2 20 8000 1.33 LM
...
How does the CMF become the IMF?
Many cores are protostellar - so there's a mass correction
The path to better mass measurement:
YSO modeling → luminosity functions
Richardson, Ginsburg, Indebetouw, Robitaille 2024 (2401.12810)
NSF 2008101:
"How are stellar masses set?"
Theo Richardson
The path to better mass measurement:
YSO modeling → luminosity functions
NSF 2008101: "How are stellar masses set?"
Theo Richardson

Hot cores in the Galactic center: Distributed MYSOs

Desmond Jeff+ 2024

Ten hot cores in Sgr B2 DS
outside the massive clusters
TG ~ 200-500 K
M ~ 200 - 2900 M
(proto-O-stars / clusters)
~5% of cores are hot cores
Sgr B2 DS: More massive cores than the Disk
Desmond Jeff+ 2024

Several different scenarios: mix of mechanisms

How is star formation in high-mass clusters different?

Cartoon of high- and low-mass star formation

Main difference: massive stars affect their surroundings

Classic HII region feedback:
O-stars clear out their environment

Accreting massive young stars affect their environment

Accreting massive young stars affect their environment

Accreting massive young stars affect their environment

The characteristic fragmentation scale


The Jeans Mass MJ is the mass where gravity and thermal pressure are balanced.

MJ ∝ T3/2 ρ−1/2

The characteristic fragmentation scale is larger

Jeans Mass MJ ∝ T3/2 ρ−1/2

Feedback affects dense gas

ALMA + VLA + GBT together give multiple temperature probes on multiple scales.
High-mass protoclusters are filled with gas warmed by feedback.
Ginsburg+ 2017, Machado+ in prep

The cartoon in the context of HMSF

These high mass cores suppress low-mass star formation (LMSF) in their vicinity. They reduce or prevent LMSF in the cores of stellar clusters.

More extreme: 'cooperative accretion'

With enough high-mass stars forming concurrently, massive stars may prevent fragmentation entirely.
If they still have enough gravity to bind the gas, the remaining gas is forced onto the most massive gravitational sinks.

Is star formation different in other environments?
In the Galactic Center?

Is star formation different in other environments?
In the Galactic Center?

Is star formation different in other environments?
In the Galactic Center?

Is star formation different in other environments?
In the Galactic Center?

Is star formation different in other environments?
In the Galactic Center?

The CMZ

$\sim10^8$ M$_\odot$ of gas in $\sim200$ pc, 10% of Galactic star formation

The Central Molecular Zone of the Galaxy represents one extreme of star forming conditions in the Galaxy

The proto-Super Star Cluster Sgr B2 dominates SF in the CMZ

Two $>10^4$ M$_\odot$ clusters within $<3$ pc; $10^6$ M$_\odot$ of gas in $\sim10$ pc

In denser (parts of) galaxies, more stars form in clusters

Γ is the fraction of stars forming in bound clusters
Galaxy averages

The "Bound Cluster Fraction" is predicted higher in the CMZ

Γ is the fraction of stars forming in bound clusters
Galaxy averages
CMZ prediction

The "Bound Cluster Fraction" is higher in CMZs

Γ is the fraction of stars forming in bound clusters
Galaxy averages
CMZ prediction
Sgr B2 data

The "Bound Cluster Fraction" was higher in the past

Zoom in to Sgr B2

The proto-Super Star Cluster Sgr B2 dominates SF in the CMZ

Two $>10^4$ M$_\odot$ clusters within $<3$ pc; $10^6$ M$_\odot$ of gas in $\sim10$ pc

The proto-Super Star Cluster Sgr B2 dominates SF in the CMZ

Two $>10^4$ M$_\odot$ clusters within $<3$ pc; $10^6$ M$_\odot$ of gas in $\sim10$ pc

Comparison of Sgr B2 (one cloud, 200 pc2) with
ALMA-IMF (15 SFRs, 53 pc2)

3mm catalog → Simplistic mass inferences

At this sensitivity, all are M>8 YSOs
Mtot = 8 M
0 M N(M) dM

8 M N(M) dM
= 96 M

The "Bound Cluster Fraction" is higher in CMZs

Γ is the fraction of stars forming in bound clusters
Galaxy averages
CMZ prediction
Sgr B2 data

The CMZ: ACES

The first complete survey of the CMZ with 2.4" resolution (previous best was ~15")
between ~2 microns and 10 cm.
Results forthcoming! Continuum, line data papers, catalogs, kinematic analyses, filament identifications all in prep....
... but the first result is unrelated (?) to star formation:

Brinary disks

The SrcI disk has gas-phase salt (NaCl, KCl) and water (H2O).
So it's brine.
(blame Adam Leroy for this term)
IRAS16547A/B (Tanaka+ 2020) have (unresolved) salt water disks

Brine lines measure dynamical mass


NaCl v=1 J=18-17 Stack of v=[0,1] Ju=[18,17]
SrcI

15 M 30 M 40 M


What governs the star formation rate?

Turbulent ISM models

Turbulent ISM models

Turbulent ISM models

ALMA enables protostar counting in
distant, massive clouds

Sgr B2: the most massive & star-forming cloud in the Galaxy
Compare YSO counts in Sgr B2 and the CMC

Is there a threshold?

Is there a threshold?

A threshold separates Sgr B2 from The Brick

Walker+ 2021

3mm Luminosity Function

What are the sources?

At this sensitivity, all are M>8 YSOs

Mass measurements: Optically thin, isothermal dust



TD estimated with PPMAP fits to Herschel data
(~6" resolution)

Simple models assuming TD ~ f(M) don't change CMF much

We can do better with YSO models and TG measurements

From YSO counts to the IMF?

How do we measure the CMF if the cores all have YSOs in them?
Mapping accretion histories
(left: IS, right: TC)
onto the Robitaille 2017 model grid

Key addition:
Envelope mass!
SPICY-ALMA-IMF:
Richardson-enhanced Robitaille+ 2017 model grid fits including ALMA data

UG team:
Sydney Petz

Brice Tingle

Morgan Himes

Our Galaxy's center, the CMZ, has denser gas than the Galactic average

Cold Dust
Hot, ionized gas
Hot dust/PAHs
Bonus slides past here

Final segment:

Astrochemistry:
Hot Cores
Ices
Line Surveys
F187N F182M F150W
F210M F187N F182M
F212N F210M F187W
F300M F212N F210M
F360M F300M F212N
F405N F360M F300M
F410M F405N F360M
F466N F410M F300M
F466N F410M F405N
F480M F410M F360M
F480M F466N F410M
F200W F182M F115W
F212N F200W F182M
F356W F212N F200W
F410M F356W F212N
F444W F410M F356W
F466N F444W F410M
The Brick isn't forming many stars
The Brick is icy
The Brick is icy
Building new tools: Spectral Line Survey of The Brick
Alyssa Bulatek: Spectral Line Survey of The Brick
  • Dasar line at 107 GHz
  • CH3OH only in absorption
  • Occurs at density $n<10^6 \mathrm{cm}^{-3}$
(me), Desmond Jeff, Savannah Gramze, Theo Richardson, Nazar Budaiev, Alyssa Bulatek, Taehwa Yoo, Miriam Garcia Santa Maria

Higher CMF + top-heavy cluster IMF = top-heavy CMZ IMF

High-mass clusters (denser regions) have top-heavier IMFs

1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]
1mm Dust
[ALMA]
870 μm Dust
[APEX/ATLASGAL]

ALMA-IMF data highlights

Gas flows in N2H+ filaments (Álvarez-Gutiérrez+ 2024)
Collapse is slower on larger scales, but fast enough to matter.

ALMA-IMF: Continuum Data → core catalogs

Pouteau+ 2022 W43-MM2/3

ALMA-IMF: CMF measurement & YSO counting

(these papers are all now submitted or accepted, but this is a screenshot that includes everyone's photos...)
So how do high-mass stars form?
Is HMSF different from LMSF?

Yes! They have more neighbors.