Title: The JWST EXCELS survey: Too much, too young, too fast? Ultra-massive quiescent galaxies at 3 < z < 5

Authors: A. C. Carnall, F. Cullen, R. J. McLure, D. J. McLeod, R. Begley, C. T. Donnan, J. S. Dunlop, A. E. Shapley, K. Rowlands, O. Almaini, K. Z. Arellano-Córdova, L. Barrufet, A. Cimatti, R. S. Ellis, N. A. Grogin, M. L. Hamadouche, G. D. Illingworth, A. M. Koekemoer, H.-H. Leung, C. C. Lovell, P. G. Pérez-González, P. Santini, T. M. Stanton, V. Wild

First Author’s Institution: Institute for Astronomy, University of Edinburgh, Royal Observatory, Edinburgh

Status: Published on Arxiv [open access]

If you’ve been browsing popular science articles in the last year or so, you may have seen a variation on the following headline: “JWST disproves Big Bang!” or: “New Galaxies force astronomers to completely rewrite cosmology.” What are these articles talking about? Well, they’re right that something super interesting is going on. JWST, since its launch in 2021, has used its cameras that are specifically designed for looking deep into the universe’s past to smash the record for oldest galaxy over and over again. This is amazing, but you quickly run into a problem if you are a cosmologist; are these galaxies so old that they disagree with how old we expect the universe to be?

A nice metaphor is this: what would you do if you believed the universe to be 50 years old, but then met someone who looked 80? Well, you’d have basically two options. Either the universe is older than you thought, or your understanding of how to guess people’s ages is slightly wrong. This is the issue we are facing with these ultra distant galaxies: do they violate our leading theory of cosmology or did they just fall behind on their skincare routine? Solving this mystery is the goal of today’s authors. In this Astrobite, we explore if a careful measurement of some of these ancient galaxies can determine if we need to rewrite cosmology!

What’s my age again?

What do we mean when we say a galaxy “looks old” anyway? Well, galaxies aren’t alive [citation needed] but they do have a sort of lifespan. A galaxy needs gas to form stars, specifically cold gas. Hot gas has too much thermal energy to collapse down to stars. Once a galaxy runs out of cold gas, we say it has been quenched, or is quiescent. There are many ways a galaxy can be quenched, such as being strangled by its neighbors or when the accretion disk of its central supermassive black hole is so bright that it pushes out or heats all of the gas in its host galaxy. 

These quiescent galaxies are where all the trouble starts. JWST has been able to spot galaxies not long after the Big Bang which are already quite massive and seem to be already quenched.  One might even say these galaxies live fast and die young if one was trying to, say, shoehorn a slightly dated hip hop reference into a pop science article.  A 2024 study used JWST to spot ZF-UDS-7329 (catchy name, yeah?), a massive quiescent galaxy that seems to have formed most of its mass by a redshift of 11, a paltry 400 million years after the big bang. The study argues that such a massive quiescent galaxy should not have been able to form so quickly under our best model of cosmology, ΛCDM.

To see if we actually have a problem here, today’s authors decide to take a closer look at this ancient galaxy and three other potentially problematic cases. To do so, spectroscopic data is key. In observational astronomy, there are essentially two types of measurements one can make: spectroscopic or photometric. Spectroscopic measurements measure the brightness of an object as a function of frequency, resulting in a spectrum that yields a ton of information about the object, see Figure 2. The problem is it’s very slow, and when time is literally money (telescope builders and operators have to make a living somehow) it may not be the best approach. Photometric measurements measure the intensity of light in large wavelength bands, drastically reducing the time it takes to make the measurement but one can infer much less about the object in question. The authors are interested in learning a lot about very few objects, so clearly a spectroscopic approach is better. Plus, the authors are able to dedicate more telescope time to these few galaxies, allowing for a higher spectroscopic resolution than previously attained, by a factor of about 10!

Analyzing The Spectra

Wait wait wait,  we’ve completely dodged the question. We asked how a galaxy could look old, and we learned about quenching, but we didn’t actually say how we could tell if a galaxy was quenched? This is where the spectroscopic data comes in. Just like how you can determine a person’s age in real life, there are many key giveaways to a galaxy’s age, and the more accurate look you can get, the better you can determine this. One relatively easy-to-spot age marker is color. Old stars tend to be dimmer and redder, so as a galaxy stops forming stars and its current stars age, we say it becomes “red and dead.” However, a galaxy may have stars of all different ages and hence of different colors, making its age harder to pin down. This is where the star formation history (SFH) of a galaxy becomes important. For example, imagine a very old galaxy that suddenly has a huge uptick in its star formation, creating a bunch of bright blue stars. Even though the galaxy is very old, it might seem young because there’s a bunch of baby stars bouncing around. We’d say that this galaxy’s SFH has a starburst (unfortunately they’re not a sponsor) so to get an accurate age measurement we would need to account for it. Therefore, to get a good picture of the age, we need to find what star formation history best fits the spectra we observe. There are many ways of figuring out what the best SFH for a given galaxy is, but today’s authors focus on using a Bayesian statistics approach. Using a code specifically designed for this purpose, the authors are able to fit the SFH as well as other key parameters such as metallicity, stellar mass, or interstellar dust content of the galaxy.

SFHs and Halo Masses

So what do they find? Pushing the spectra through the fitting code yields the star formation history for each of the four galaxies no problem, see Figure 3. A slightly trickier question is: are these star formation histories in tension with our ΛCDM cosmology?

The issue is that ΛCDM does not make a direct prediction on the size of galaxies in our universe at a given time, it only makes a prediction on the size of dark matter haloes as a function of time, called the halo-mass function. Using the predicted size of dark matter haloes to infer the size that galaxies should be is quite difficult, as this depends on a lot of complicated, random-seeming physics factors. To handle this, astronomers might typically infer the allowed stellar mass by multiplying the allowed halo mass by the stellar fraction (fraction of the mass of the galaxy in the form of stars) and then the baryon fraction (fraction of the universe’s mass in the form of matter particles); i.e., if we know an upper allowed limit on halo mass, we can estimate the limit on galaxy stellar mass by multiplying the halo mass by the average ratio of dark matter to stars in the universe. This method has some problems as we are sweeping a lot of physics under the rug. The authors therefore choose to add one layer of sophistication, performing this process directly on the SFH itself. Therefore, they can see if the predicted SFH is ever in conflict with the halo mass function at different points in time. This can be a very important additional step to take, as the halo mass function can vary relatively quickly at different redshifts. Finally, the authors choose to add one more wrinkle to the issue, as that stellar fraction can be a tricky number to pin down. The authors therefore hedge their bets and choose two cases, a modern estimate of the distribution of this value from a recent paper, versus a case with very efficient star formation, i.e. the stellar fraction is always one. They then compare the allowed stellar mass by ΛCDM to the SFH of these 4 galaxies. An example for one of the galaxies is shown in Figure 4. 

Interestingly, three out of the four galaxies tested have SFHs that are producing galaxies too massive to be accommodated by ΛCDM using the first model. However, using the second “stellar fraction is always 1” model, they are relatively allied, only ever disagreeing with ΛCDM to about 1 sigma. In other words, it is decently likely the disagreement is a statistical fluke. This means that, while they would require an unusually high percentage of their mass to be in the form of stars, they are not cosmology-breaking. 

The authors come to the conclusion that it is more likely that these super ancient galaxies age and form stars slightly differently than our current understanding, rather than ΛCDM being incorrect. However, this is the exact opposite of what an earlier study found on the same galaxy, and the SFH for the galaxy that both studies found look roughly the same; how could they get different results? The authors hypothesize that the difference comes down to the earlier paper’s assumption that the dark matter halo would remain constant in size, while today’s paper argues that it is more likely that the dark matter halo evolves in step with the galaxy. This means that the halo mass in today’s paper was slightly lower and hence more likely to exist in the early universe, resulting in it no longer being cosmology breaking. 

 So ΛCDM is saved! Or, at least, this “impossibly early galaxy” problem is not the smoking gun we thought it was, though ΛCDM does have plenty of other problems that mean it cannot be the final, ultimate cosmological model. The authors are also quick to point out that in order for ΛCDM to still be valid, they had to push our galaxy formation models to the absolute edge of what is allowed. Therefore, it is extremely important we understand why galaxy formation was so different in the early universe to make sure these conclusions are valid. In the meantime, ΛCDM remains triumphant as the leading model of cosmology, even if cracks are beginning to show. 

Astrobite edited by Brandon Prie

Featured image credit: Kat Nurminsky