Search everything:

Formation and evolution of galaxies

The study of galaxy formation and evolution aims to address the basic question of how the Milky Way and other types of galaxy we see around us today came to be. This not only involves understanding the properties of nearby galaxies in great detail, but also looking into the very distant Universe, since the long light travel times from distant galaxies allow us to examine the properties of galaxies at different times in the past. This way we can build up an empirical picture of how they have evolved over cosmic time and compare this to numerical models.

Some of the key questions we are trying to answer are:

  • what is the cosmic history of star formation and black hole growth and how does this depend on fundamental properties such as mass?
  • what is the nature of star formation in the early Universe?
  • how do galaxies acquire and process their gas?
  • how does environment affect the evolution of galaxies?
  • how does feedback (from stars and black hole growth) affect galaxy evolution?

Galaxy formation and evolution research at CAR covers a range of areas that touch upon all of these themes, and our strong team of staff, fellows post-docs and students are engaged in a number of different projects.

The earliest galaxies

(Curtis-Lake)

The search for and characterisation of the earliest galaxies in the Universe is one of the four main science goals of the Webb Space Telescope.  Webb can see further back in time than Hubble, but also, probing into the infrared, we can learn more about galaxies Hubble had identified by observing more of the electromagnetic spectrum.  For the last decade or so, we had candidate galaxies inhabiting the Universe at high redshifts ( out to z~11.9) but spectroscopic follow-up was incredibly challenging due to i) their intrinsic faintness and ii) the absorption of the brightest UV line (Lyman-alpha, an emission line of Hydrogen from the first excited state to the ground state) by intervening Neutral Hydrogen in the early Universe.  These candidates were identified from the rest-frame Ultra-violet which gives information about the present star-forming state.  Some information about the rest-frame optical could be gleaned from the Spitzer space telescope to understand something about the previous stellar mass growth, but this was impeded greatly by the confusion between stellar continuum and nebular emission line contributions to Spitzer imaging.

With the Webb space telescope, we can identify new candidates, notably further away than what Hubble could see.  Then, thanks to NIRSpec we can follow up candidates with sensitive near-infrared spectroscopy, targeting hundreds of galaxies at once.  Thanks to the JADES survey (NIRCam imaging and NIRSpec follow-up spectroscopy) we have set the redshift frontier for highest redshift spectroscopically confirmed galaxies twice over (JADES-GS-z12-0, JADES-GS-z13-0, Curtis-Lake et al. 2023, NatAstron and JADES-GS-z14-0, JADES-GS-z14-1, Carniani et al. 2024, Nature accepted).  The first spectra showed indications of soft Lyman-alpha drops indicative of neutral hydrogen, inferred at the time to be present in the IGM, later confirmed for JADES-GS-z12-0 to be due to a large reservoir of neutral gas within the galaxy (d’Eugenio et al. 23)  thanks to the incredibly deep NIRSpec observations taken in parallel to the NIRCam imaging of the JADES Origins Field (JOF, Eisenstein et al. 23).

Deep NIRCam imaging and NIRSpec spectroscopy from the JADES Survey are revealing intriguing chemical abundances (super-solar C/O in JADES-GS-z12-0, high N/O at low metallicity, e.g. Bunker et al. 23Curti et al. 24), and a flatter low-metallicity slope to the mass-metallicity relation at low stellar masses (Curti et al. 23).  The survey is now complete and Curtis-Lake follow up on these observations with SED-fitting, exploiting the photoionisation models implemented in the BEAGLE SED fitting tool (Chevallard & Charlot 2016 iap.fr/beagle, Curtis-Lake is current lead developer) to characterise key statistical relations of galaxy populations from the full sample (the main sequence of star forming galaxies and the mass-metallicity relation), as well as using BEAGLE-AGN (Vidal-Garcia, Plat, Curtis-Lake et al. 23) to probe obscured AGN properties.

The evolution of dwarf galaxies

(Kaviraj)

Dwarf galaxies dominate the galaxy number density, making them critical for our understanding of galaxy evolution. However, while much is known about the evolution of massive galaxies out to high redshift, the dwarf regime remains largely unexplored. This is largely due to the fact that past large surveys like the SDSS are too shallow to detect typical dwarfs outside the very nearby Universe (z<0.02). The dwarfs that do appear in these surveys outside our local neighbourhood have anomalously high star formation rates, which boost their luminosity, making them detectable in shallow surveys. However, they also make them biased towards blue systems which predominantly have late-type morphology.


New and forthcoming surveys like the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP) and the Legacy Survey of Space and Time (LSST) are poised to revolutionise our understanding of galaxy evolution by providing complete and unbiased samples of hundreds of thousands of dwarfs at low and intermediate redshift. At Hertfordshire we are using the HSC-SSP to explore key questions about dwarf galaxy evolution, in preparation for our future efforts using LSST. Some important questions are: what processes regulate star formation in dwarf galaxies? what is the morphological mix of dwarfs? is AGN feedback important in the dwarf regime? how does dwarf evolution differ from what is known in massive galaxies?

The chemical evolution of galaxies

(Kobayashi, Yates)

The mix of chemical elements (particularly those heavier than helium) in the gas and stars of galaxies can significantly impact how they evolve and appear when observed. This is because, as stars and supernovae enrich galaxies with freshly synthesised heavy elements, cooling rates in the interstellar medium  (ISM) are boosted, the efficiency of star formation is modified, and the subsequent stellar evolution inside stars is altered. It is therefore paramount for us to have an accurate picture of the composition and distribution of chemical elements in galaxies, so that we can understand their overall evolution.

Research conducted by researchers at CAR is designed to do exactly this, by measuring the abundance of key elements in galaxies using the latest observational instrumentation. For example, by leveraging data from the MaNGA and MUSE integral field units, the true abundance and distribution of oxygen (the most abundant heavy element in the Universe) in the ISM of nearby spiral galaxies has been determined using highly accurate measurement techniques (Yates et al. 2020, 2021a, 2024). In collaboration with external colleagues, these data have been supplemented by measurements from JWST of galaxies in the more distant (i.e. older) Universe which host super-bright gamma-ray bursts (Schady, Yates, et al. 2024). The combination of these nearby and distant measurements has allowed us to constrain the chemical evolution of galaxies over many billions of years, revealing a less rapid evolution than was previously assumed (Yates et al. 2021b, 2024).

Future work on these topics will look at the chemistry of galaxies at even earlier times, fully harnessing the potential of JWST, as well as the implications of this chemical evolution on the conditions for habitability across cosmic time.

Map of the spatial distribution of oxygen across the ISM in NGC4900
Map of the spatial distribution of oxygen across the ISM in nearby spiral galaxy NGC 4900. The IFU spectra were taken from the MUSE-MAD sample (Erroz-Ferrer et al. 2019) and oxygen abundances were re-measured using the method of Easeman et al. (2024). Adapted from Yates et al. (2024).
Stacked radial profiles of oxygen abundance in galaxies
Stacked radial profiles of the oxygen abundance in the ISM of nearby spiral galaxies, separated into three stellar-mass bins. Square symbols denote re-measurements from the MUSE-MAD sample of Erroz-Ferrer et al. (2019), and circle symbols denote measurements from the MaNGA sample of Yates et al. (2020). These data are compared to the stacked radial profiles from the L-Galaxies 2023 galaxy evolution simulation in blue (Yates et al. 2024, see [hyperlink to the "Theory & Modelling" webpage] for more details). Adapted from Yates et al. (2024).

Submillimetre surveys

(Coppin, Geach, Stevens, Smith, Hardcastle)

Observations in the submillimetre part of the electromagnetic spectrum are sensitive to the cold interstellar dust in galaxies, and informs us about the rate of star formation and total dense gas content in galaxies. At CAR we are heavily involved in the SCUBA-2 Large eXtragalactic Survey (S2LXS; and previously the Cosmology Legacy Survey (S2CLS), Herschel-ATLAS and HerMES projects) as well as several ALMA follow-up campaigns of these luminous submm sources at higher resolution.

The S2CLS was the largest of the JCMT Legacy Surveys and is the largest and most sensitive survey of its kind ever conducted, producing the first samples of thousands of extragalactic sources selected in the 450um and 850um submillimetre wavebands, an order-of-magnitude improvement in the sample sizes of previous surveys at these wavelengths (Geach et al. 2013, Coppin et al. 2015).  The next generation survey, S2LXS, is now the largest contiguous area of extragalactic sky (centred on the XMM-LSS field) mapped by JCMT at 850um to date. The wide area of the S2LXS XMM-LSS survey allows us to probe the ultra-bright ( S_850um>15mJy ), yet rare submillimetre population (Garrett et al. 2023). These dedicated survey programmes have revolutionized our understanding of submillimetre galaxies, and indeed galaxy formation in general, with enormous and lasting legacy value, as well as providing a springboard for future exploitation of data from the Atacama Millimeter Array (ALMA; Garrett et al. in prep), LOFAR, James Webb Space Telescope (JWST) and the Square Kilometer Array (SKA).

Herschel-ATLAS is the largest open time extragalactic survey that was conducted with the Herschel Space Observatory, and has detected around 250,000 galaxies based on their dust content alone, from the local Universe out to the highest redshifts. To date, the survey has produced in excess of 50 refereed publications, many led from CAR, on a diverse range of topics from studying the properties of dust in galaxies (e.g. Dunne et al. 2010, Smith et al. 2013), their star formation histories, stellar masses and multi-wavelength properties (e.g. Smith et al. 2012, 2014, Jarvis et al. 2010, Rowlands et al. 2014, Davis et al. 2015), studying the relationship between star formation and AGN (e.g. Hardcastle et al. 2010, 2013, Bonfield et al. 2011, Kalfountzou et al. 2014, Gurkan et al. 2015), linking galaxy properties to their merger history (e.g. Kaviraj et al. 2013) and environments (e.g. Coppin et al. 2011, Burton et al. 2013), and finding large samples of lensed galaxies (Negrello et al. 2010).

Panchromatic SEDs derived using star-forming galaxies in the H-ATLAS
Panchromatic SEDs from Smith et al. (2012), derived using star-forming galaxies in the H-ATLAS; binned according to specific star formation rate, and offset vertically, these empirical templates are cooler than other available models, and include uncertainties as a function of wavelength for the first time.

Cosmic Web

(S. de Souza, Devereux)

The cosmic web, an awe-inspiring latticework structure, stands as one of the most significant physical patterns in the Universe. Formed through the hierarchical growth of large-scale structures, its existence has been validated by extensive surveys and cosmological simulations. This intricate network comprises galaxy groups, clusters, filaments, sheets, and vast cosmic voids. The filamentary components, mapped by surveys like the Sloan Digital Sky Survey (SDSS) and the Cosmic Evolution Survey (COSMOS), highlight its extensive reach and complexity. The distribution of baryonic matter within the cosmic web correlates with dark matter, influencing galaxy properties such as stellar masses and luminosity. Observations of filaments reveal phenomena like quiescent galaxies and hot gas, and studies suggest that interactions within filaments may enhance star formation rates, adding to our understanding of cosmic evolution.

In our latest work, we introduce a novel cosmic web finder adaptive to spherical and light-cone geometries, building on the classical subspace constrained mean shift (SCMS) algorithm. This new method, the Spherical and CONic Cosmic wEb finder (SCONCE), estimates the matter or galaxy density function on the celestial sphere or 3D light-cone, recovering filaments as density ridges. Our approach enhances the accuracy of cosmic web structure identification, particularly in regions with higher declination, providing a more detailed and precise cosmic map. This research marks a significant step forward in mapping the Universe's grand structure, offering deeper insights into the cosmic web's role in shaping galaxies.

Advanced Galaxy Classification with Machine Learning

(Spindler)

Machine learning techniques have been used in astronomy to classify the structures of galaxies for nearly a decade, and recent advances in both labelling techniques and algorithm design have led to very large galaxy morphology catalogues using recent imaging surveys. However, it is also possible to use machine learning to identify the more complex internal structures of galaxies, and not just the global morphology labels. Using volunteer classifications from Galaxy Zoo 3D, a neural network called Zoobot3D has been trained that can classify galaxy structures, such as bars and spiral arms, down to the pixel level in extragalactic imaging. These segmentation maps can then be used to study and analyse a variety of properties and processes that affect the structure and evolution of late-type galaxies.

Zoobot3D uses an advanced neural network architecture, called a U-NET, which extracts the locations and extents of spiral arms and bars from images of galaxies. When compared with analytical segmentation models and the original volunteer maps used to the train the model, expert astronomers almost unanimously preferred the maps created by Zoobot3D in a blind test. An example of this comparison is shown in the first figure below. The maps can also be used to measure physical properties of a galaxy, such as the extent of the bar shown in the second figure below, which are in good agreement with previous citizen science measurements of this property. In the future Zoobot3D will be used to create segmentation maps from the millions of late-type galaxies in the DESI survey, and new labelling technologies will allow for larger, more accurate training data to be gathered from next-generation astronomy surveys like LSST and Euclid.

Comparison of galaxy classification structures
Comparisons of different segmentation techniques for identifying interior galaxy structures. Left: DESI images of five galaxies with various morphologies. Left-centre: Results from Masters at al. (2021) created by combining masks drawn by citizen scientists in the Galaxy Zoo: 3D project; the red areas in these images show the location of the bar, and the blue the spiral arms, the intensity of the colour represents the number of volunteers who labelled a given area. Right-centre: Results of the sparcfire algorithm (Davis et al. 2014), a numerical model that identifies bars and spiral arms analytically; the algorithm does not distinguish between bars and spiral arms, and only provides a binary label for each pixel. Right: Results of Zoobot3D, with the same colour scheme as the volunteer masks; the intensity of the pixels shows the certainty of the model that a given label is correct.
Images showing the extent of the bar
A demonstration of how the results of Zoobot3D can be used to measure physical properties of a galaxy. The first column shows the DESI image of a spiral galaxy, the second shows the Zoobot3D predictions for the spiral arm and bar masks, and the third shows just the bar with the extremes marked with yellow points. By measuring the distance between these points in pixels, we can convert this to a physical size based on the distance to the galaxy and the resolution of the image.