RESEARCH ACTIVITY

My research activity revolves around the InterStellar Medium (ISM). The ISM is the complex intertwining of gas phases that fills the volume of a galaxy, between its stars. As an illustration, Fig. 1 shows an image of the Orion nebula. We can clearly see stars mixed with swirls of dusty gas. Stars are born out of the gravitational collapse of interstellar clouds. At the end of their lifetime, stars return a significant fraction of their matter to the ISM, whether exploding as supernovae or gently expelling their envelope as planetary nebulae. This is why we talk about an interstellar lifecycle, as stars deplete and replenish the interstellar matter reservoir. Each successive generation of stars also enriches the ISM in heavy elements (C, N, O, Fe, etc.), changing its thermodynamic and radiative properties. This cycle is however not endless. Young galaxies (irregular galaxies) are gas-rich, whereas old galaxies (elliptical galaxies) are almost totally devoid of interstellar gas, because most of it is locked in old stars.

The interstellar lifecycle is thus what drives the evolution of a galaxy. Yet, the process of star formation and its impact in return on the ISM is still poorly known. This is a major open question in modern astrophysics. This is important because our understanding of galaxy evolution, from the dark ages to the contemporary epoch, determines our understanding of the evolution of the Universe. My research work is aimed at tackling some particular aspects of this cycle.

I focus my research on cosmic nanoparticles, also vulgarly called dust. These particles are small, solid-state compounds made of the most abundant heavy elements in the ISM (C, O, Si, Mg, Fe, etc.) arranged predominantly in silicate and carbonaceous compounds. The size of these grains ranges in the ISM from a few Å (a few atoms) to a few 100 nm. Cosmic nanoparticles account for less than 1% of the mass of the interstellar matter, but they have a crucial role in several physical processes.

1. They extinct the UV-to-near-InfraRed (IR; λ≅100 nm to λ≅1 μm) light emitted by stars. The absorbed power is re-emitted thermally in the infrared/submillimeter (λ≅3 μm to λ≅800 μm). For instance, the interstellar emission in Fig. 1 comes mainly from the grains. The gas emission is much dimmer. For that reason, star forming regions are totally opaque at UV and visible wavelengths. The stellar radiation can not escape their interstellar dusty cocoon. These regions have to be studied via their dust emission, in the IR.
2. Dust grains play an important role in the heating and cooling of the neutral gas. Grain surfaces are also catalysts to chemical reactions, forming molecules, including the most abundant one in the Universe, H₂.
3. Because of their ubiquity, and the high power they radiate, cosmic nanoparticles are used as proxies to studying numerous phenomena, including star formation.

The physical properties of these cosmic nanoparticles are not well constrained. We do not precisely know their exact composition, size distribution and shape. Moreover, there is clear evidence that these physical properties evolve as a function of the environment, under the effects of stellar radiation, gas density, shocks and metallicity. The dust lifecycle, describing how grains are formed, processed and destroyed in the ISM is schematically represented in Fig. 2. An important part of my work consists in constraining the timescales of the individual processes that are represented.

The dust lifecycle can be studied from different viewpoints. It however requires an empirical approach. Most of my research work focuses on nearby galaxies for the following reasons.

1. They are close enough to be studied with a lot of details and accuracy. In the closest ones, the Magellanic clouds, we can resolve individual clouds with modern infrared observatories. More distant galaxies do not provide this luxury, but they bring a more statistical view on ISM properties.
2. Nearby galaxies show a wide diversity of physical conditions, as represented in Fig. 3. In particular, they have a large range of star formation activity, some being quiescent, other undergoing a starburst. They also span two orders of magnitude in metallicity, the mass fraction of heavy elements. Metallicity is a chemical age tracer. Low-metallicity galaxies, which happen to be dwarf galaxies, are the best local analogues to distant primordial systems.

By modeling nearby galaxies, one can better understand dust evolution, if one considers each object as a snapshot of galaxy evolution at a given stage.

---

Most of the physical information is contained in spectra. A spectrum is the power emitted per unit wavelength (or photon energy). The spectrum of a typical galaxy is shown in Fig. 4. We see that there is a continuum emission, varying smoothly as a function of wavelength, originating from different processes (stellar emission, dust emission, etc.). There are also several lines, with sharp spikes of energy in a very narrow wavelength band. These lines originate in the resonance of electrons in atoms or molecules. The shape and level of the continuum, as well as the intensity, width and central wavelengths of gas lines, contains a wealth of information about the physical conditions in the observed region. Historically, it is spectroscopy that transformed astronomy into astrophysics, as it allowed scientists to quantify the temperature, density, physical state, abundances of atoms from very far away.

When the spectrum is only coarsely sampled, we rather talk about Spectral Energy Distribution (SED), but it is the same thing. The SED of the galaxy represented in Fig. 4 covers a large part of the electromagnetic spectrum, from the UV to the centimeter radio wavelengths. Photons with wavelengths shorter than λ=0.0912 μm are able to ionize hydrogen. They are emitted by the most massive stars, which also happen to be the hottest ones. Stellar emission dominates the UV-to-near-IR range. In the IR, thermal emission from dust grains dominates. These grains are on average at 20 kelvins, 20 degrees above the absolute 0. Gas lines are emitted over the whole electromagnetic range and the gas continuum dominates mainly at radio wavelengths.

A lot of the information about the grain properties can be inferred from their IR emission. This is a large part of my activity. I develop SED models to interpret IR observations of galaxies. Comparing models to observations is however not a straightforward matter. There is an immense literature on the subject.

Inferring physical quantities from observations, that are fatally tainted by uncertainties, requires a statistical framework. There are two philosphies: the frequentist and Bayesian approaches. For a complete discussion about these two methods, I refer the reader to Chapter V of my HDR. In brief, the advantage of the Bayesian method, which I favor, allows one to derive the probability distribution of physical parameters, accounting for all the sources of observational uncertainty. I wrote a computer code to achieve this goal as accurately as possible, in the case of IR SED fitting: HerBIE (HiERarchical Bayesian Inference for dust Emission; Galliano, 2018).

I also work with other methods. I have followed the recent interest in machine-learning techniques. I worked with several students applying them to ISM questions. In particular:

• Optimizing dust SED models with machine-learning methods (Tedros & Galliano, in prep.);
• Using Natural Language Processing to automatically explore the literature looking for molecules that could explain DIBs (van den Broek d'Obrenan et al. 2023, JIMIS, vol. 11).

This last study was quite original, as we showed that some molecules found in the eye pigments of several sea animals contained molecular groupings of interstellar interest (chromophores) that could account for several DIBs.

---

Interstellar dust grain properties:

computation and implementation in SwING: PDF.

Photometric filters:

color corrections and calibration uncertainties: PDF.

Robust estimators:

of the moments of a distribution: PDF.

Displaying uncertainties:

in correlation plots: PDF.