SQUALO Survey: Clump Accretion & Fragmentation

• The survey establishes that high-mass clumps with infall signatures fragment early, linking parsec-scale inflow with compact cores.
• It employs ALMA Band 6 and Hyper source-extraction techniques alongside simulations to quantify fragment masses, spacing, and evolution.
• Findings support a clump-fed accretion model where magnetic fields play a crucial role in moderating fragmentation and core growth.

Searching arXiv for SQUALO survey and related Rosetta Stone papers. SQUALO, an acronym for Star formation in QUiescent And Luminous Objects, is an ALMA Band 6 ($1.3\,\mathrm{mm}$) and Band 3 ($3\,\mathrm{mm}$) survey of 13 high-mass clumps selected at different evolutionary stages but with a common signature of parsec-scale infall. Its central purpose is to connect clump-scale accretion and sub-parsec fragmentation, and thereby test whether massive stars are assembled primarily from isolated, pre-existing cores or by continuing accretion from a larger clump reservoir. The first survey paper presents the $1.3\,\mathrm{mm}$ continuum analysis and argues for a clump-fed accretion mechanism in high-mass star-forming objects (Traficante et al., 2023). A later synthetic-observation study, built explicitly around SQUALO, used radiative MHD simulations post-processed through the exact ALMA observing strategy and source-extraction pipeline of the survey to interpret what the observed fragmentation statistics imply about the underlying physics, particularly magnetic regulation and unresolved multiplicity (Nucara et al., 15 Jul 2025).

1. Scientific rationale and conceptual framework

SQUALO was designed to test the relation between the clump scale and the fragment/core scale in massive star formation. In the survey definition, the clump scale is traced by Hi-GAL and single-dish line data at approximately $0.1$–$0.3\,\mathrm{pc}$, whereas the fragment/core scale is traced by ALMA at approximately $0.01$–$0.05\,\mathrm{pc}$ (Traficante et al., 2023). The survey therefore targets a specific unresolved issue: whether the formation of high-mass stars is core-fed, in the sense of massive, quasi-static pre-stellar cores whose mass reservoir is largely fixed before collapse, or clump-fed, in the sense of hierarchical, continuous accretion from the parent clump and filamentary environment.

The observational strategy is intentionally selective. Rather than sampling all massive clumps, SQUALO focuses on clumps that already show evidence of infall at parsec scales. In that sense, the survey is not a general census of high-mass clumps; it is a controlled experiment on a set of accreting systems. This framing matters for interpretation, because the survey asks how an already accreting clump fragments, whether fragment properties track parent-clump properties, whether fragmentation is compatible with thermal Jeans expectations, and whether there exist massive clumps that remain weakly fragmented despite ongoing evolution (Traficante et al., 2023).

The first SQUALO results favor a dynamic, multi-scale accretion picture. The youngest clumps are already fragmented, fragment masses and surface densities correlate with clump properties, and fragment separations decrease with evolution. The survey paper interprets this pattern as a sequence in which non-thermal motions are important initially, gravitational collapse becomes increasingly dominant, and fragments continue to accrete from the parent clump rather than behaving as isolated, pre-assembled units (Traficante et al., 2023).

2. Survey construction, sample definition, and observing strategy

The 13 SQUALO clumps were assembled from the Hi-GAL compact source catalog cross-matched with MALT90. Traficante et al. identified 213 clumps with characterized dust SEDs and gas kinematics, among which 21 showed blue-asymmetric HCO$^+$ $(1\!-\!0)$ profiles indicative of infall. From these, SQUALO selected 10 clumps with $M_{\rm cl} \ge 170\,M_\odot$, $3\,\mathrm{mm}$0, $3\,\mathrm{mm}$1, relative isolation from other bright clumps, and clear blue-skewed HCO$3\,\mathrm{mm}$2 $3\,\mathrm{mm}$3 or HNC $3\,\mathrm{mm}$4 profiles. Three additional massive $3\,\mathrm{mm}$5-quiet clumps were then added to cover the earliest stages (Traficante et al., 2023).

The final sample spans $3\,\mathrm{mm}$6. By the survey’s evolutionary convention, $3\,\mathrm{mm}$7 corresponds to “quiescent” or $3\,\mathrm{mm}$8-quiet clumps, $3\,\mathrm{mm}$9 to early protostellar objects, and $1.3\,\mathrm{mm}$0 to more evolved IR-bright or H II region phases. The clumps have masses $1.3\,\mathrm{mm}$1–$1.3\,\mathrm{mm}$2, radii $1.3\,\mathrm{mm}$3–$1.3\,\mathrm{mm}$4, surface densities $1.3\,\mathrm{mm}$5–$1.3\,\mathrm{mm}$6, dust temperatures $1.3\,\mathrm{mm}$7–$1.3\,\mathrm{mm}$8, virial parameters $1.3\,\mathrm{mm}$9–1.22, and distances $0.1$0–$0.1$1. All 13 show blue-skewed HCO$0.1$2 $0.1$3 or HNC $0.1$4 profiles consistent with clump-scale infall and inferred accretion rates of $0.1$5 (Traficante et al., 2023).

The ALMA observations were obtained in project 2018.1.00443.S. Band 6 provides the continuum data analyzed in the first paper, while Band 3 line observations are reserved for later work. The continuum setup combines the $0.1$6 and $0.1$7 arrays, with synthesized beams of approximately $0.1$8–$0.1$9; at the source distances, this corresponds to linear resolutions of approximately $0.3\,\mathrm{pc}$0–$0.3\,\mathrm{pc}$1, or about $0.3\,\mathrm{pc}$2–$0.3\,\mathrm{pc}$3 (Traficante et al., 2023). In Rosetta Stone III, the SQUALO observing strategy was mimicked in detail using the reference clump HIGALBM24.0116+0.4897 (“HG24”), with ALMA Band 6 centered at $0.3\,\mathrm{pc}$4, a synthetic beam of $0.3\,\mathrm{pc}$5, a cell size of $0.3\,\mathrm{pc}$6, Briggs weighting with robust $0.3\,\mathrm{pc}$7, and tclean imaging in mosaic mode with a multiscale deconvolver (Nucara et al., 15 Jul 2025).

This observing design has an immediate methodological consequence. The interferometric data recover compact fragments within the clumps but filter out a significant fraction of the very extended envelope. The survey therefore does not attempt to measure the full clump mass budget from ALMA alone; instead, Hi-GAL characterizes the parent clumps while ALMA isolates the compact substructure (Traficante et al., 2023).

3. Fragment identification and derivation of physical properties

SQUALO explored two extraction methods, Astrodendro and Hyper, and adopted Hyper for the main analysis. Hyper was favored because it is designed for compact sources on strongly varying backgrounds, includes deblending, and treats sources as 2D Gaussian components in a way consistent with Hi-GAL source definitions (Traficante et al., 2023). The procedure was to estimate the rms of each primary-beam-corrected continuum map using sigma-clipping, identify compact peaks with peak flux $0.3\,\mathrm{pc}$8, fit 2D Gaussians, and retain sources with aspect ratio $0.3\,\mathrm{pc}$9 and FWHMs not larger than $0.01$0 the synthesized beam. This yielded 60 initial fragments, of which 5 were rejected as spurious, producing a final sample of 55 fragments (Traficante et al., 2023).

For each fragment, the radius $0.01$1 is defined from the geometric mean of the Gaussian major and minor FWHM axes. The resulting radii are $0.01$2, indicating that the detected objects are resolved or moderately resolved structures that may still host sub-fragmentation below the survey resolution (Traficante et al., 2023). Rosetta Stone III used the same Hyper-based source-finding and photometry pipeline on synthetic SQUALO maps, with detection on primary-beam-uncorrected images, photometry on primary-beam-corrected images, a $0.01$3 threshold, second-order polynomial background modeling, and compact elliptical source selection with semi-major and semi-minor axes between 1 and 2 beam FWHM and aspect ratio $0.01$4 (Nucara et al., 15 Jul 2025).

Fragment masses are derived from integrated $0.01$5 fluxes using the standard optically thin dust-emission relation

$0.01$6

with $0.01$7 from Preibisch et al. (1993), already including a gas-to-dust ratio of 100 (Traficante et al., 2023). SQUALO assigns a single temperature to all fragments in a given clump according to $0.01$8: $0.01$9 for $0.05\,\mathrm{pc}$0, $0.05\,\mathrm{pc}$1 for $0.05\,\mathrm{pc}$2, and $0.05\,\mathrm{pc}$3 for $0.05\,\mathrm{pc}$4. Surface density is then

$0.05\,\mathrm{pc}$5

The fragment masses span $0.05\,\mathrm{pc}$6, and the completeness mass across the sample is $0.05\,\mathrm{pc}$7–$0.05\,\mathrm{pc}$8 (Traficante et al., 2023).

The survey also evaluates clump-scale Jeans quantities using the clump density and either the thermal or non-thermal velocity dispersion. The clump virial parameter is written as

$0.05\,\mathrm{pc}$9

The dimensionless ratios $^+$0 compare the minimum fragment separation to the Jeans length, and $^+$1 compares the mass of the most massive fragment to the Jeans mass (Traficante et al., 2023). These ratios are central to the survey’s interpretation of fragmentation as initially non-thermal and increasingly gravity-dominated.

4. Empirical findings from the $^+$2 continuum survey

The basic fragmentation statistics are straightforward. Twelve of the 13 clumps are fragmented, and the number of fragments per clump ranges from 1 to 9. The most fragmented source is HIGALBM327.3918+0.1996, with 9 fragments. Only one clump, HIGALBM343.7560−0.1629, shows a single fragment (Traficante et al., 2023).

The youngest objects are not monolithic. All three $^+$3-quiet clumps are already fragmented: HIGALBM24.0116+0.4897 has 2 fragments with masses of approximately $^+$4–$^+$5, HIGALBM28.1957−0.0724 has 4 fragments with masses of approximately $^+$6–$^+$7, and HIGALBM31.9462+0.0759 has 4 fragments with masses of approximately $^+$8–$^+$9 (Traficante et al., 2023). This is one of the survey’s principal arguments against the expectation that massive clumps routinely host single, massive pre-stellar cores at the observed scales.

The exceptional case, HIGALBM343.7560−0.1629, is physically notable rather than merely observationally unresolved. It lies at $(1\!-\!0)$0 with $(1\!-\!0)$1 and contains a single fragment with $(1\!-\!0)$2, $(1\!-\!0)$3, and $(1\!-\!0)$4. At SQUALO’s linear resolution of approximately $(1\!-\!0)$5, the fragment remains unresolved into subcomponents. Because other clumps at similar $(1\!-\!0)$6 and larger distances are clearly fragmented, the survey interprets this source as requiring an additional support mechanism, plausibly strong magnetic fields (Traficante et al., 2023).

The Jeans analysis places the fragmentation in a gravo-turbulent regime rather than a purely thermal one. For the thermal Jeans comparison, $(1\!-\!0)$7 lies in $(1\!-\!0)$8–7.04 for most clumps, so fragment separations exceed the thermal Jeans length. For the non-thermal Jeans comparison, $(1\!-\!0)$9 lies in 0.28–1.47; most clumps have values below unity, indicating that non-thermal motions alone do not explain the observed structure. The most massive fragments exceed the thermal Jeans mass by large factors, while $M_{\rm cl} \ge 170\,M_\odot$0 is mostly 0.01–0.93, except for two clumps where it is about 1.7 (Traficante et al., 2023). The survey interprets this as evidence that non-thermal motions are important but are themselves coupled to gravitational contraction rather than acting as a static support term.

Evolution enters most clearly through fragment spacing rather than fragment count. The correlation between $M_{\rm cl} \ge 170\,M_\odot$1 and number of fragments is mild, with Pearson $M_{\rm cl} \ge 170\,M_\odot$2, but the minimum distance between fragments decreases with evolutionary stage. In addition, the number of fragments is strongly anti-correlated with the thermal Jeans ratio, with $M_{\rm cl} \ge 170\,M_\odot$3, so more highly fragmented clumps are closer to thermal Jeans spacing (Traficante et al., 2023). This suggests a sequence in which initial fragmentation occurs at larger, non-thermally regulated separations, after which collapse drives fragments into denser, more thermally supported configurations.

The survey reports several cross-scale correlations. Total fragment mass $M_{\rm cl} \ge 170\,M_\odot$4 correlates with clump mass $M_{\rm cl} \ge 170\,M_\odot$5 with $M_{\rm cl} \ge 170\,M_\odot$6 and $M_{\rm cl} \ge 170\,M_\odot$7. The densest fragment surface density $M_{\rm cl} \ge 170\,M_\odot$8 correlates with clump surface density $M_{\rm cl} \ge 170\,M_\odot$9 with $3\,\mathrm{mm}$00 and $3\,\mathrm{mm}$01. Total fragment mass shows a moderate positive correlation with clump accretion rate, $3\,\mathrm{mm}$02, $3\,\mathrm{mm}$03. The clump virial parameter is anti-correlated with total fragment mass, with $3\,\mathrm{mm}$04 and $3\,\mathrm{mm}$05, while $3\,\mathrm{mm}$06 versus $3\,\mathrm{mm}$07 follows the familiar $3\,\mathrm{mm}$08 relation with $3\,\mathrm{mm}$09 (Traficante et al., 2023). The survey interprets these trends as evidence that fragment growth is dynamically linked to the parent clump’s mass reservoir, density, and boundedness.

5. Synthetic reinterpretation through Rosetta Stone III

Rosetta Stone III extends SQUALO by constructing an end-to-end simulation-to-observation pipeline tuned to the survey’s actual continuum strategy (Nucara et al., 15 Jul 2025). It uses 24 radiative MHD simulations of collapsing high-mass clumps, performs full dust radiative transfer at $3\,\mathrm{mm}$10, post-processes the resulting maps in CASA to mimic the exact ALMA setup of SQUALO, and applies the same Hyper source-extraction and photometry pipeline used on the real data. The simulations are initialized as uniform spheres with $3\,\mathrm{mm}$11 or $3\,\mathrm{mm}$12, radius $3\,\mathrm{mm}$13, temperature $3\,\mathrm{mm}$14, Mach numbers $3\,\mathrm{mm}$15 and 10, and mass-to-flux ratios $3\,\mathrm{mm}$16. Two random seeds for the turbulent velocity field give 24 distinct simulations (Nucara et al., 15 Jul 2025).

The simulations are evolved until sink formation efficiency reaches approximately 0.15 for $3\,\mathrm{mm}$17 models and approximately 0.30 for $3\,\mathrm{mm}$18 models, corresponding to roughly $3\,\mathrm{mm}$19 after the first sink appears. Each snapshot is projected along three orthogonal lines of sight, yielding 244 3D cubes and 732 2D sky maps at $3\,\mathrm{mm}$20 (Nucara et al., 15 Jul 2025). Rosetta Stone III then uses the SFE–$3\,\mathrm{mm}$21 calibration

$3\,\mathrm{mm}$22

to place synthetic and real SQUALO clumps on the same fragmentation-versus-evolution diagrams (Nucara et al., 15 Jul 2025).

Several technical results of the synthetic pipeline are directly relevant to the interpretation of SQUALO maps. When the exact observing setup is reproduced, the synthetic ALMA images recover approximately $3\,\mathrm{mm}$23–95% of the total flux, with earlier stages losing more flux because their emission is more extended. A generic simalma pipeline with idealized tracks would under-recover large-scale flux by up to approximately $3\,\mathrm{mm}$24 in early stages, broaden the beam by approximately $3\,\mathrm{mm}$25, and both create spurious fragments and miss real ones. By contrast, the customized SQUALO-tuned pipeline yields synthetic rms values of $3\,\mathrm{mm}$26–$3\,\mathrm{mm}$27, with mean $3\,\mathrm{mm}$28, similar to the real survey dynamic range (Nucara et al., 15 Jul 2025).

At SQUALO-like resolution of approximately $3\,\mathrm{mm}$29, the synthetic clumps show 2 to 14 fragments per field, with an average of about 7. This range overlaps the survey’s observed 1 to 9 fragments per clump, but with an important caveat: the simulations do not produce unfragmented clumps at this resolution (Nucara et al., 15 Jul 2025). The dominant physical parameter controlling multiplicity is magnetic field strength. For $3\,\mathrm{mm}$30, weak-field quasi-hydrodynamic models with $3\,\mathrm{mm}$31 preferentially show more fragments than strongly magnetized models with $3\,\mathrm{mm}$32, while the difference between $3\,\mathrm{mm}$33 and 10 is much weaker. This synthetic result is used to argue that the low-multiplicity end of the SQUALO sample is more naturally associated with strong magnetic regulation than with moderate changes in turbulence (Nucara et al., 15 Jul 2025).

Rosetta Stone III also clarifies what a SQUALO “fragment” means physically. Across 393 Seed-2 synthetic maps, the analysis identifies 2749 fragments and 16209 sinks. About $3\,\mathrm{mm}$34 of fragments have at least one sink counterpart, but only approximately $3\,\mathrm{mm}$35 correspond to a single sink, while approximately $3\,\mathrm{mm}$36 contain multiple sinks. About $3\,\mathrm{mm}$37 of fragments have no sink at that snapshot. Conversely, approximately $3\,\mathrm{mm}$38 of all sinks fall within at least one Hyper fragment, leaving approximately $3\,\mathrm{mm}$39 without a continuum-fragment counterpart (Nucara et al., 15 Jul 2025). This shows that a $3\,\mathrm{mm}$40 SQUALO fragment is neither a one-to-one proxy for an individual protostar nor necessarily already protostellar.

The same synthetic framework supports the clump-fed interpretation. Defining sink formation efficiency as

$3\,\mathrm{mm}$41

and fragment formation efficiency as

$3\,\mathrm{mm}$42

Rosetta Stone III finds $3\,\mathrm{mm}$43 at all times. In early phases, $3\,\mathrm{mm}$44 while $3\,\mathrm{mm}$45, and the two converge as sinks accrete from fragments (Nucara et al., 15 Jul 2025). Synthetic fragments and sinks both gain mass throughout the evolution, favoring the same hierarchical picture advanced in the survey paper: clumps feed fragments, and fragments feed sinks.

6. Limitations, interpretive boundaries, and legacy

SQUALO’s principal limitation is sample size: 13 clumps and 55 fragments. The survey nonetheless gains leverage by restricting the sample to clumps with established infall signatures, so its trends are physically focused rather than statistically broad (Traficante et al., 2023). Another limitation is that the first paper is continuum-only. Fragment masses and surface densities depend on assumed dust temperatures, and the dominant systematic uncertainty is therefore thermal rather than photometric. The survey tested plausible temperature ranges and found that individual masses can shift by factors of approximately 4–5 in extreme cases, but Monte Carlo experiments showed that the principal correlation trends are robust (Traficante et al., 2023).

Rosetta Stone III sharpened these temperature and resolution caveats. It showed that using a single clump-scale temperature works reasonably well at moderate $3\,\mathrm{mm}$46, but that early fragments can be colder than assumed and highly evolved fragments hotter than assumed. At high $3\,\mathrm{mm}$47, true fragment temperatures can exceed $3\,\mathrm{mm}$48, even $3\,\mathrm{mm}$49, while SQUALO still assumes $3\,\mathrm{mm}$50; this tends to over-estimate fragment masses in the most evolved clumps (Nucara et al., 15 Jul 2025). It also showed that interferometric filtering preferentially suppresses extended early-stage emission, so low-contrast young fragments are easier to miss, and that the $3\,\mathrm{mm}$51 resolution merges close pairs and substructure into single “fragments” (Nucara et al., 15 Jul 2025).

On the simulation side, Rosetta Stone III explicitly acknowledges that its clumps are initially uniform spheres, not filament-embedded hubs; that only accretion luminosity is included, without outflows or H II regions; that the magnetic field geometry is initially uniform; and that the parameter space does not extend below $3\,\mathrm{mm}$52 or above $3\,\mathrm{mm}$53 (Nucara et al., 15 Jul 2025). These limitations matter most for the SQUALO clumps with the weakest observed fragmentation, because those systems lie below the synthetic multiplicity envelope and plausibly require either stronger magnetization or additional physics.

Within those boundaries, SQUALO occupies a distinct niche among high-mass ALMA surveys. It is complementary to ATLASGAL+ALMA, CORE, ALMA-IMF, ASHES, and related $3\,\mathrm{mm}$54-quiet surveys, but its defining feature is the combination of a broad $3\,\mathrm{mm}$55 span with explicit pre-selection on clump-scale infall (Traficante et al., 2023). The empirical survey and its Rosetta Stone reinterpretation converge on the same general picture: massive clumps selected for infall are already fragmented at early stages, fragment properties remain coupled to clump properties, fragment spacing decreases with evolution, and magnetic fields are likely crucial in the minority of systems where fragmentation is strongly suppressed (Traficante et al., 2023, Nucara et al., 15 Jul 2025).

A plausible implication is that SQUALO should be read not as a census of compact cores, but as a survey of fragmentation within accreting, dynamically evolving clumps. In that reading, its major contribution is the direct linkage between parsec-scale inflow and the compact structures seen by ALMA, together with a calibrated synthetic framework showing how interferometric filtering, unresolved multiplicity, temperature assumptions, and magnetic regulation shape the observable fragmentation pattern (Traficante et al., 2023, Nucara et al., 15 Jul 2025).