Figure 1: This image shows atomic hydrogen emission from the interstellar medium of the Milky Way. We measure emission as a function of Doppler-shifted velocity...


From The Stanford Physics Department Annual Newsletter 2023/24

Main content start

The Stuff Between The Stars

By Susan Clark

When you think of a galaxy like our Milky Way, you might think of a massive knot of dark matter, or of the stars that can be seen splashed across the sky on a dark night. But galaxies are also swirling cauldrons of gas and dust: the material that makes up the interstellar medium, or what you might colloquially call the “stuff between the stars”. 

Figure 1: This image shows atomic hydrogen emission from the interstellar medium of the Milky Way. We measure emission as a function of Doppler-shifted velocity, and this three-color image is made by mapping the emission in three adjacent velocity bins to the red, green, and blue components of the image. This figure was adapted from Clark et al. 2019.

The interstellar medium is turbulent and permeated by magnetic fields. It is the material out of which new stars form, and into which some old stars explode. Understanding the flow of matter and energy in the interstellar medium is part of the modern astrophysical quest to understand star formation, the evolution of galaxies, and ultimately our cosmic origins.

Gas in the interstellar medium spans an enormous range of physical states, from hot, diffuse plasma to the cold and dense molecular clouds that are the birthplaces of new stars. We seek to understand the distribution of this material, and the physics that governs flows of gas and transitions between interstellar phases. These are formidable challenges that we undertake from our single vantage point: a smallish, rocky planet some eight kiloparsecs from the center of the Milky Way. The Cosmic Magnetism and Interstellar Physics Group at Stanford is leading the way, working to elucidate the physics of the diffuse and magnetic universe using theory, simulations, and vast quantities of data.

From our little corner of the cosmos, we decipher our gaseous galactic home by collecting light from across the electromagnetic spectrum. One particularly useful window that we have on our universe is line emission from atomic hydrogen. Atomic hydrogen is abundant in galaxies and has a hyperfine transition that corresponds to a photon with a wavelength of 21 centimeters. We observe this “21-cm line” with radio telescopes, measuring both emission and absorption as a function of frequency, or equivalently, Doppler-shifted velocity. Figure 1 shows a region of this 21-cm emission from the nearby interstellar medium. 

Figure 2: The architecture of the scattering transform as applied to images of atomic hydrogen emission. This figure from Lei & Clark 2023.

Recently, our group has been pursuing new probes of the physical properties of this gas. The neutral medium traced by atomic hydrogen emission is thermally bistable, meaning that there are two stable thermal phases with very different densities and temperatures that coexist at typical interstellar pressures. We call these the “cold neutral medium” and the “warm neutral medium”. The cold neutral medium has typical temperatures ~50-100 K, with densities of tens of atoms per cubic centimeter near the Sun. The warm neutral medium is comparatively tenuous, with less than an atom per cubic centimeter and typical temperatures ~6000 - 1 0,000 K. 

Directly constraining the thermodynamic properties of this gas requires both emission and absorption measurements – and absorption measurements require a bright background radio source, like a quasar, against which we can detect interstellar absorption. However, our group has found a promising new avenue: we have shown that substantial physical information is encoded in the spatial distribution of the gas. In other words, we can use clever computational image analysis tools to probe interstellar medium physics directly from hyperspectral data like that illustrated in Figure 1.

Recently, graduate student Minjie Lei led work that quantified the spatial distribution of 21-cm emission with the scattering transform, a statistical technique that originated in the mathematics literature and draws inspiration from the structure of convolutional neural nets, but requires no training (Figure 2). By applying this transform to images of atomic hydrogen emission and comparing to absorption-line measurements, we discovered a link between the morphology of the interstellar gas and its thermodynamic state. This points to exciting new directions for studying the phases of the interstellar medium from emission data alone.

Our group has also discovered connections between gas morphology and the interstellar magnetic field structure. This is an exciting time for our group and for the field, as we find creative new approaches for the analysis of vast troves of astrophysical data. We are designing new algorithms that allow us to quantify image-space morphology in gas and dust emission– and finally probing the rich physics that is etched into the beautiful complexity of the interstellar medium.


Assistant Professor Susan Clark is an astrophysicist, with primary research interests in cosmic magnetic fields, magnetohydrodynamic processes, and the interstellar medium.