Science Highlights

Understanding how stars form and evolve is one of the most fascinating challenges in astronomy. A key piece of this puzzle is the stellar multiplicity—the frequency, separations, and mass ratios of stars that form in pairs or higher-order groups. Massive stars, which make up only 0.01% of all formed stars,  play a pivotal role in shaping galaxies and the universe. Understanding their life cycle—90% of which unfolds alongside at least one companion—is therefore of paramount importance. Yet, they are particularly difficult to study since they are heavily bloated in their dusty envelopes and are born in distant, dense, and young star clusters. These observational challenges leave significant gaps in our understanding of how their multiplicity evolves over time. 

To help fill this gap, our team conducted high-resolution observations to explore the properties of stars that have just formed or are still in the process of forming, within four massive, near-primordial clusters. We identify potential binary or multiple star systems, study how their properties—such as separation and mass ratios—vary within and between clusters. In this framework, multiplicity serves as a tool to trace the initial conditions that drive the formation of massive stars in multiple systems.

To study the primordial multiplicity properties, the clusters must have similar chronological ages with slight variations and be young enough to ensure that significant dynamical interactions have not yet occurred. This increases the likelihood of observing systems in a near-primordial multiplicity state. The selected clusters—Hourglass Nebula, RCW 108, DBS 113, and DBS 121—are observable in the near-infrared and submillimeter wavelengths due to their significant dust emission. These clusters are located at a distance of 1.5 ± 0.3 kpc on average and can be classified by age.  Based on this cluster classification, we trace the evolution of multiplicity and companion fraction between 1.0 Myr and 2.8 Myr. 

Our preliminary results are derived from NACO K-band data, with Chinmaya, a PhD student in our team, leading this aspect of the project.  The figure shows one of the fields of DBS121 for which we show the identified sources, hence potential companions per bin of separation. We find that multiplicity properties evolve with cluster age, as the other clusters show a gradually increasing trend in the number of stars, thereby indicating a possible increase in companion fraction with cluster age. To extend this study, we recently included VVV CL100 (7.5 Myr) to determine whether this trend persists in older clusters. 

Emma, a postdoc in our team, successfully led an ALMA proposal to enhance this analysis. Data acquisition is now underway, promising exciting new discoveries in the evolution of  multiplicity properties in massive star-forming clusters. Stay tuned!

We are currently building the CCAT Heterodyne Array Instrument (CHAI) to be operated at the FYST-telescope. CHAI is a 64-pixel high-resolution spectrometer for two frequency bands around 460 GHz (650 µm) and 800 GHz (370 µm). For optimum instrument stability, CHAI uses balanced SIS mixers, which receive their Local Oscillator (LO) signal through an input port separate from the measurement signal input.

In order to distribute the LO power to the individual mixers, we developed a waveguide power splitter with an extremely even splitting ratio over the required bandwidth. Since the LO source is distributed to up to 32 mixers, we are employing a 5-level cascade of this divide-by-two splitter. Excessive asymmetry in the splitting ratio of a single unit would result in unacceptable power differences between the outputs of the cascade. Each of our splitting stages uses a simple Y-junction, upgraded by a drop-in chip, which suppresses crosstalk between the two outputs.

In order to illustrate the suitability of the concept for a multi-level cascade of binary splitters, we built the first section of CHAI’s LO distribution network: a 3-level cascade, which results in 8 equal output ports to serve one row of 8 mixer blocks in CHAI’s focal plane. The photograph shows a device manufactured as E-plane split block at our precision workshop. Waveguide dimensions are 0.460 mm x 0.230 mm for the low-frequency band of CHAI (420 – 510 GHz). Visible at each junction are the chips (600 µm x 60 µm) for the isolation between the output ports, which were manufactured in our microfabrication lab.

The graph shows the output power distribution measured at the 8 output ports of the 3-level cascade over the nominal bandwidth. No systematic asymmetry can be seen in the signal. The maximum power scatter – including significant measurement noise – is below 40% of the nominal output power – well within our tolerance range.

Title figure: Photograph of the two split-block halves of the 8-way power divider, showing the waveguides with the isolator chips. Output ports are numbered.

To date, the growth mechanism of supermassive black holes (SMBHs) is a scientific mystery. If we consider the accretion rate of the SMBH in our Milky Way, Sagittarius A* (Sgr A*), and the age of the universe, a discrepancy of several magnitudes in its mass opens up. One proposed idea to overcome the mismatch of accretion rate and age of the universe is merging events between intermediate-mass black holes (IMBHs) that ultimately form SMBHs. However, only around 10 IMBHs in our entire universe have been confirmed by observations, which poses a significant challenge to the theory of merging black holes. In Peißker et al. (2023c) and Peißker et al. (2024b), we have analyzed the densest stellar cluster, IRS 13, in our Milky Way, only about 0.3 lightyears away from Sgr A*. This massive embedded cluster shows two distinct generations of stars, implying independent and plausible triggered, star formation events. Until recently, it was unclear how and why this cluster so close to an SMBH seems to preserve its shape. As we show in both related publications, the cluster comprises three distinctive components. One of the components is associated with the dense core of the cluster, whereas the other one shows signs of evaporation. We argue that IRS 13 is the remanent of a more massive cluster that plunges into the gravitational well of our central SMBH. One key aspect of this analysis is the young age of the two stellar populations. From the young age and plausible star formation channels, we derive an unusually short cluster migration timescale. One explanation for the rapid infall of the cluster could be the presence of an IMBH inside the cluster. In Peißker et al. (2024b), we confirm that the presence of an IMBH is highly likely. Due to the evaporation nature of the cluster, it is expected that we identified one of the first pre-merger setups between an SMBH and IMBH to date.

Stars form within molecular clouds, dense regions of cold gas primarily composed of molecular hydrogen. These clouds provide the necessary conditions for the formation of stars, including low temperatures and high densities, which allow gravitational forces to overcome thermal pressure and initiate the collapse of gas. To fully understand the process of star formation and therefore the evolution of galaxies, it is crucial to study the properties of molecular clouds—such as their mass, density, distribution, and relation to the galactic environment.

The Physics at High Angular resolution in Nearby GalaxieS (PHANGS) collaboration aims to create a comprehensive view of star formation and the lifecycle of gas and dust in nearby galaxies, using state-of-the-art facilities. In particular, by leveraging the James Webb Space Telescope (JWST)’s infrared capabilities, the PHANGS-JWST program has provided astonishing views of 19 galaxies in wavelengths that are typically obscured by dust in the optical range, reaching unprecedented resolutions and sensitivities.

We used observations of the emission from Polycyclic Aromatic Hydrocarbons (PAHs, e.g. complex organic molecules that emit strongly in the mid-infrared and are associated with photodissociation regions) from PHANGS-JWST to generate molecular gas maps of the 19 galaxies.

The application of the Spectral Clustering for Molecular Emission Segmentation (SCIMES) on the JWST data allowed the identification of more than 50,000 highly-resolved molecular clouds. SCIMES is a machine learning-based code that utilises graph theory concepts to segment out molecular clouds from the more diffuse medium by preserving their intrinsic morphology and internal structure.

Our preliminary results suggest that the molecular cloud mass spectra—specifically their steepness and truncation mass—are strongly influenced by the surrounding dynamical environment. This indicates that certain physical conditions may be more favourable to the formation of high-mass clouds than others.

Figure caption: Upper row: Left image shows the galaxy’s intensity map; right image displays dust structures identified by SCIMES with a greyscale intensity background and a colour bar indicating 7.7 μm intensity. Bottom row: Cumulative mass distributions of molecular clouds in different environments. Dotted black lines represent simple power-law fits; solid black curves are truncated power-law fits. Fit parameters—γ (spectral index), M0 (maximum mass), N0 (count in the distribution). The grey region represents the Poisson errors on the counts.

Line Intensity Mapping (LIM) is an emerging technique in radio-astronomy that scans vast fractions of the sky with a large beam and detects the integrated emission of all sources along the line of sight without resolving individual objects. This approach enables probing the high-redshift Universe including the contribution from intrinsically faint sources that traditional surveys miss due to their flux-limit thresholds. These peculiarities make LIM an ideal tool to probe the nature of dark matter (DM).

Most particle-physics candidates for DM fall into the class of thermal relics (i.e. particles that were once in thermal equilibrium with the rest of the Universe). In this case, the velocity dispersion of the particles at early times turns out to be inversely proportional to their mass. This implies that less massive particles can freely stream out of shallow potential wells and, de facto, inhibit the formation of low-mass structures. Therefore, cold DM (CDM, with negligible velocity dispersion) and warm DM (WDM, with a free-streaming length of the order of 0.1 Mpc) give rise to a different mass spectrum of DM halos within which galaxy formation takes place.

Using the halo-model approach, we make forecasts for the constraints that LIM of the 150 μm fine-structure transition of [Cii] can set on the mass of the DM particles. Ionised carbon is a promising tracer that should be present also in low-mass halos, contrary to neutral hydrogen that cannot be shielded from the UV background after cosmic reionisation. We compress the data into the isotropic power spectrum and use Bayesian inference marginalising over the uncertain faint-end slope of the [Cii] luminosity function (LF). 

Our results are shown in the figure as a function of the survey area and for two different measurements of the bright-end of the [Cii] LF (optimistic/pessimistic). Assuming a CDM scenario, we find that LIM can rule out WDM particle masses up to 2–3 keV, which makes this technique competitive with other probes, such as the Ly-α forest. Our study demonstrates that, taking into account the current limits on the LF, the [Cii] power spectrum is dominated by sources hosted in relatively massive halos and this diminishes its constraining power on the WDM mass. 

Paper: Marcuzzo et al. (2024) 

Modelling the molecular gas content of galaxies is a highly non-linear, multi-scale problem in astrophysics. On one hand, it is necessary to simulate galaxies in realistic environments as they are affected by outflows and gas accretion from the cosmic web. On the other hand, molecular-cloud chemistry is regulated by conditions on sub-parsec scales.

To overcome this challenge, we have developed a new sub-grid model, HYACINTH – HYdrogen And Carbon chemistry in the INTerstellar medium in Hydro simulations – that can be embedded into cosmological simulations of galaxy formation to calculate the non-equilibrium abundances of molecular hydrogen and its carbon-based tracers, namely, CO, C, and C⁺ on the fly. Our model captures the effects of the ‘microscopic’ (i.e., unresolved) density structure on the ‘macroscopic’ (i.e., resolved) chemical abundances in cosmological simulations using a variable probability distribution function of sub-grid densities within each resolution element. 

The chemical abundances from HYACINTH are in good agreement with observations of nearby and high-redshift galaxies. We are now running a suite of cosmological galaxy formation simulations with HYACINTH that will allow us to address fundamental questions regarding the contribution of different galaxies to the global H₂ budget and the reliability of different molecular gas tracers across ISM conditions and galaxy environments at high redshifts (z ≳ 3).

The paper describing the model has been accepted for publication in A&A. 
DOI: 10.1051/0004-6361/202449640

Figure caption: Column density maps of different species in a pre-simulated galaxy (from Tomassetti et al. 2015) post-processed with HYACINTH. The first two panels show the distribution of total and molecular hydrogen from the simulation, while the other panels show the species obtained in post-processing. CO is concentrated in regions with the highest N(H₂), while C and C⁺ are more widespread; C⁺ even extends out to regions lacking a significant amount of H₂and closely mirrors the total gas distribution.

Many places in space are too far away to learn about them by sending spacecraft there. So they cannot be examined directly but instead, we can learn about them by analyzing their emitted light.
Due to quantum mechanics, each molecule has a set of characteristic transition lines that uniquely identify it. When these transition lines are found in the emitted spectrum, we can be sure that the respective molecule appears in the observed object.
However, to identify molecules in space, we first have to understand their characteristic patterns in the laboratory. We do so by measuring the rotational spectrum of the molecules in our experiment and then fitting quantum mechanical models to them. These models can then be used by astronomers to identify the molecules in space and also to infer the physical conditions of the corresponding regions in space. For example, the temperature can be deduced from intensity relations, the pressure from the lineshape, and the molecule’s abundance can be inferred from its intensity.
Here, we measured the pure rotational spectrum of phosphabutyne (C2H5CP) for the first time and analyzed its vibrational ground state as well as its three singly 13C-substituted isotopologues. This will allow astronomers to search for phosphabutyne in space and determine the prevailing conditions of the corresponding regions. The figure shows a section of the measured broadband spectrum on top, highlighting the pattern repeating with the total angular momentum quantum number J, while the zoom-in on the bottom highlights the very good agreement between the calculated and measured spectrum, especially when applying a small shift of 126 MHz.
This work was performed in collaboration with Jean-Claude Guillemin (University Rennes) and Michael E. Harding (KIT).

What is the structure and chemical composition of gas that may feed future star formation? Before interstellar gas turns dense enough to form new stars it is not fully molecular yet but in some so far unknown transitional state. A special case of such gas clouds are given by high-latitude clouds representing material that may fall onto the plane of the Milky Way.
In a recently accepted paper by Nicola Schneider and collaborators we reported results from SOFIA/upGREAT observations of a number of diffuse and high latitude clouds in the Milky Way, in particular the Draco, Spider, Polaris and Musca clouds. In only one of them, the Draco cloud, we detected emission of ionized carbon in spite of it being neither the densest cloud nor the one irradiated the strongest by known sources.
When trying to model the emission of all observations in terms of an astrochemical model of a photon-dominated region it turns out, that the model is not able to simultaneously explain the strength of the continuum emission from dust and the ionized carbon line. The ionized carbon detection and also the non-detections suggest a very low impinging UV field well below what is actually observed taking all the known stars in the environment. For the Draco cloud we can explain the difference by the additional energy from a shock that is produced when the cloud is hitting the interstellar gas of the Milky Way and some shielding of the cloud by interstellar dust in atomic gas. However, for the other clouds, we do not have a consistent explanation yet.
The paper combined the results from a fruitful collaboration between the CRC 1601 subprojects B2 and A6.

Figure caption: Schematic illustration of the illumination of a high-latitude cloud by UV radiation from stars in the plane of the Milky Way. The orange cylinder along the path between star and cloud indicates the dust column which attenuates the UV-field.

What roles do different stellar feedback processes play in governing star formation? From ionizing and non-ionizing radiation to stellar winds and supernovae, these forces interact with the surrounding stellar nurseries. However, understanding the precise significance of each process in shaping star formation remains an ongoing challenge.

We present new magneto-hydrodynamic (MHD) simulations conducted within the SILCC framework, exploring the multiphase interstellar medium (ISM) within a patch of a stratified galactic disk. Our study incorporates a self-consistent modeling of non-ionizing far-ultraviolet (FUV) radiation emitted by stellar clusters, aiming to understand its impact on star formation and the chemical composition of the nearby ISM. We observe locally intense interstellar radiation fields (ISRF) with values up to G₀ ≈ 10⁴ (in Habing units), contrasting with the canonical solar neighborhood value of G₀ = 1.7.

Our findings suggest that while FUV radiation influences star formation, its role in regulating the star formation rate (SFR) appears less significant compared to other stellar feedback mechanisms such as ionizing UV radiation, stellar winds, and supernovae. Additionally, our chemical analysis reveals enhancements in both the warm-ionized medium (WIM) and the cold-neutral medium (CNM) beyond the vicinity of stellar clusters, indicating a complex interplay influenced by the self-consistent and highly variable FUV radiation field, fostering the presence of a diffuse molecular hydrogen gas phase.

Further details will be available in Rathjen et al., currently in preparation.

Figure: Overview of the simulated ISM. Shown are the edge-on views of the total gas (Σ_gas, 1st panel), molecular hydrogen (Σ_H2, 4th panel), and ionized hydrogen (Σ_H⁺, 5th panel) column densities, as well as mass-weighted gas (T_gas, 2nd panel) and dust (T_dust, 3rd panel) temperatures and ionizing photon energy density (e_γ, 6th panel), the effective G₀ field (G_eff, 7th panel), and cosmic ray energy density (e_CR, 8th panel) in projection. The star-forming galactic ISM is concentrated around the midplane. White circles in the 1st panel indicate active star clusters.

To learn more about the formation and evolution of massive stars it is important to confront simulations and observations.
It is useful to interpret the observational data and to extract the cores’ physical parameters,. We can address e.g. the question how massive cores fragment and form (massive) stars, or how long the young, massive stellar objects are embedded in their parental core.

Doing so, we simulate a collapsing core scenario of a subvirial, 1000 M☉ core with an initial radius of 1 pc and a linear magnetic field of 100 μG, which is a birthplace of massive stars.

For the post-processing we use RADMC-3D, which is an open source radiative transfer code that is based on the Monte-Carlo method. Here, we present synthetic observations of the dust emission (top left panel). The advantage of our simulations is that we calculate the dust temperature self-consistently, hence taking into account radiative heating by all young stars as well as shock heating. Thus, RADMC-3D is directly working on the simulated dust temperature. These results are post-processed with CASA, where different, respectively a combination of, possible ALMA channels and predictable water vapour (pwv) in the atmosphere can be simulated.
We show the results for synthetic observations in different ALMA channels (labeled with AMLA TM1, TM2 and ACA; bottom panels), as well as their combination for two predictable water vapour settings (top middle and top right panel). In synthetic observations most of the structures in the less dense environment are not visible anymore; however, the emission of the main star forming regions remain.

This work was performed in collaboration with Dr. Álvaro Sánchez-Monge