Abstract
One of the stunning surprises from JWST’s first year and a half of observations of the extragalactic sky is that not only are we detecting large numbers of galaxies in the early universe but that many of these galaxies are resolved and show great structural detail. For the first time, we can quantify the morphological structure of galaxies well into the epoch of reionzation and constrain how these galaxies grew over cosmic time. Here we present an analysis of the morphological and size evolution of galaxies in the early universe, from z=3-9, using deep multi-band NIRCam imaging from the public survey fields CEERS and COSMOS-Web. Our morphological measurements include quantitative measures, such as surface brightness profile fitting and non-parametric fits, visual morphologies, and machine learning approaches. We find that galaxies at this early epoch have a wide diversity of morphologies, including compact unresolved sources, disky structures, irregular features and multiple components, and a significant fraction that appear to be involved in a merger. Future work with larger surveys like COSMOS-Web will enable statistical analyses and the training of machine learning algorithms to classify galaxy morphologies and identify galaxy mergers.
1. Introduction
One of the most fundamental properties of a galaxy is its physical structure. We have known for about 100 years that galaxies in our nearby universe typically have two different structures: 1) disk galaxies with spiral arms and sometimes a bar through their center, 2) elliptical galaxies consisting of stars with more randomized orbits. Galaxies that do not fall into these two types are considered to be irregular. However, galaxies in the early universe are significantly different as these structures were in the process of forming. Galaxies were overall smaller, clumpier, and messier. However, it is still an open question of when the very basics of today’s structures were put into place.
Over the age of the universe, while galaxies were evolving structurally, many other properties were also evolving in tandem. Their stellar masses grew as their gas reservoirs were turned into stars. Their chemical compositions changed as subsequent generations of stars formed heavier elements. Their central supermassive black holes grew, occasionally entering an active growth face and appearing as an active galactic nucleus (AGN). The overall level of star formation and AGN activity in the universe changed, ramping up to a peak from the early universe to a period known as cosmic noon (1 < z < 3) and then coming down from that peak to the present day level of star formation. Some galaxies grew passively on their own while others were involved in major or minor mergers that altered their properties. The structural evolution of galaxies over this time period provide clues to the physical processes at work driving all of these changes along their evolutionary pathways.
JWST has opened a new window in our understanding of galaxy structure. Prior to JWST, our observations of structural detail with Hubble were limited to z ∼ 3, where Hubble could probe the rest-frame optical emission from galaxies. JWST’s coverage of the near to mid-infrared wavelengths makes it ideal for investigating the rest-frame optical structure of galaxies, where the bulk of stars emit their light, out to very high redshifts. Additionally, the incredible sensitivity and increased resolution allow us to see features that are smaller and fainter than ever before. JWST’s first observations with surveys such as CEERS (Bagley et al. 2023) have revealed the detailed morphologies in the early universe for the first time, while large surveys such as COSMOS-Web (Casey & Kartaltepe et al. 2023) enable detailed statistical analyses of hundreds of thousands of galaxies.
2. Morphological Measurements
Our initial study of galaxy morphologies in the early universe was conducted using images from the CEERS survey’s first observations taken in June 2022. We selected a sample of galaxies with previous measurements from HST to evaluate the improvement / change in morphologies as measured by JWST. These galaxies span the redshift range z = 3 − 9. We measured these morphologies visually, adapting the visual classification scheme of Kartaltepe et al. (2015), as well as quantitatively, using Galfit[1] (Peng et al. 2002; Peng et al. 2010), GalfitM,[2] and the Python package Statmorph[3] (Rodriguez-Gomez et al. 2019). Figure 1 shows the CEERS NIRCam mosaic with a handful of example postage stamp cutouts highlighting the range of morphologies observed with JWST at these redshifts.
3. Morphological Evolution
Our sample contains the full range of morphological types across all redshift and stellar masses. Over the entire redshift range, only 16 and 18 galaxies are classified as Point Source/Unresolved or Unclassifiable, respectively. Figure 2 shows the fraction of the total number of galaxies that each morphological class makes up as a function of redshift. Overall, 56% of the galaxies at z > 3 have a visually identifiable disk component, dropping from ∼ 60% at z = 3 − 4, to ∼ 45% at z ∼ 5, to ∼ 30% at z > 6. 38% of the galaxies at z > 3 have a visually identifiable spheroidal component, decreasing from 42% to 26% between z = 3 and 4.5, then varying between ∼30-40% beyond z = 4.5. This is largely driven by the similar decrease and then increase in the Spheroid Only group. Part of this apparent trend at higher redshifts may be due to small number statistics and part may be due to a number of selection effects. For example, there is a possibility that we miss fainter extended features in some of these systems at high redshift. It is also possible that a larger fraction of galaxies at higher redshift are small enough to be at the resolution limit of NIRCam, given the expected size evolution of galaxies, and are therefore more round and compact in appearance.
43% of the galaxies at z > 3 have irregular features and this fraction remains roughly constant across the full redshift range due in part to the fraction of Disk+Irregular galaxies being roughly constant at 20% and then decreasing while the fraction of Irregular Only galaxies is at roughly 10-15% and then increases to 20% by z = 4.5. Note that the total fractions of objects that are All Disks, All Spheroids, or All Irregular do not add up to one due to the overlapping objects in each of these classes.
Finally, we note that the fraction of point sources and unclassifiable objects remains at 0-2% across most of the redshift range. At z > 6, 13% of galaxies are unresolved and 8% are unclassifiable, corresponding to 5 and 3 individual galaxies, respectively, in this redshift bin. We remind the reader that the above percentages correspond to galaxies that were bright enough to be detectable with HST CANDELS imaging and may not be representative of the overall galaxy sample detectable by JWST at these redshifts.
Figure 3 shows the morphological fractions as a function of redshift based on CANDELS HST imaging and using the visual classifications of Kartaltepe et al. (2015) for all z > 3 galaxies in all five CANDELS fields (1375 galaxies in total). Based on the HST imaging alone, a smaller fraction of galaxies at z = 3.0 − 4.5 have disks (∼40%) and a larger fraction are pure spheroids (∼20% at z = 3.0 − 5.0). The fraction of galaxies that are only irregular is small and drops with redshift, from ∼5% at z > 6. The fraction of galaxies that are unclassifiable rises sharply, ∼5% at z = 3.5 to∼35% at z=5.5 to∼80% at z>7. Likewise,∼30% are unresolved at z=5−7.
The large difference seen between the HST and JWST morphologies at these redshifts is expected and is due to the difference in depth and wavelength coverage. 488 galaxies were flagged by at least one classifier as having a different morphology in the JWST images compared to the HST images. A significant number of galaxies with disks were previously identified as spheroids because of their compact central morphologies, with low surface brightness disk features that only became visible with deeper imaging (see, for example, Conselice et al. 2011; Mortlock et al. 2013; Kartaltepe et al. 2015). This suggests that some fraction of the spheroidal galaxies observed with JWST, particularly those that are faint and/or at higher redshifts, possibly have unobserved disks as well. It is not likely that these disks would previously have been identified as irregular, except for some at the low redshift end, as these irregular features are also too faint to be easily identified at these redshifts with HST. At the low redshift end (z = 3 − 3.5), some disks may have been classified as irregular if the HST data only picked up the brighter star forming clumps rather than the underlying disk structure.
Figure 4 compares the distribution of the Sérsic indices and sizes of galaxies from the Santa Cruz Semi-analytic model (SAM, Yung et al. 2022a), TNG50 (Costantin et al. 2022), and TNG100 (Rose et al. 2023) to the distribution measured from CEERS galaxies. The distributions from the SAM have very similar peaks with a narrower distribution, which holds for all three redshift bins. The Sérsic index for both TNG50 and TNG100 peak at lower values than the CEERS galaxies and have narrower distributions at all redshifts. At z = 3 − 4, TNG50 galaxies have larger sizes than TNG100 galaxies and even larger than both the SAM galaxies and the observed CEERS galaxies. At z > 4 the distributions match more closely. At all redshifts, the simulations do not contain the smaller (lower Re) more compact (larger n) galaxies that we observe with JWST CEERS imaging.
Figure 4 also compares the measured axis ratio and asymmetry value for the TNG50 and TNG100 galaxies to the distribution from CEERS. In all three redshift bins, the axis ratios of the TNG50 and TNG100 galaxies match each other well, but peak at higher b/a (∼0.6) and fall off more sharply at lower values than the observed CEERS galaxies. At z = 3 − 4, the asymmetry distributions for TNG50, TNG100, and CEERS are well-matched, but the TNG50 and TNG100 distributions shift toward lower (more negative) values at higher redshift. Negative asymmetry values are unphysical and typically result from low S/N sources, where the source is very close to the background level that is being subtracted when making the asymmetry measurement.
Overall, the agreement between our measurements for the z > 3 JWST CEERS galaxies and the various simulations is encouraging. The differences seen (for example, the difference in axis ratio and the lack of small compact galaxies in the simulations) are worthy of a more in-depth look in order to determine if there are selection effects impacting the results or if there is an actual physical difference between galaxies in these simulations and those in the real observed universe.
4. Merger Identification
Galaxy mergers are particularly difficult to identify, especially at high redshift where the low surface brightness merger signatures are lost. JWST’s increased sensitivity enables the identification of major and minor mergers that would previously have been missed. New machine learning techniques can also be applied to large datasets and tested to quantify how well they work to identify galaxy mergers. We have identified mergers visually, and categorized them into Groups 1, 2, and 3 based on how many classifiers identified them as a merger, irregular, or marked various merger signatures (such as tidal tails and double nuclei). The Group 1 sample is the most confident, though has smaller numbers.
We then tested two different machine learning techniques: random forests and a convolutional neural network DeepMerge (Ćiprijanović et al. 2020). We first used the simulated CEERS imaging of Rose et al. (2023) to train and then test the performance using both the simulated images (where the merger history is known) and the visual classifications from CEERS. Overall, we find that each of the two methods hits a ceiling of ∼ 70% at identifying both mergers and non-mergers correctly. The performance for identifying mergers can be improved, but only at the expense of incorrectly classifying non-mergers.
Figure 5 shows the Gini coefficient G vs. M20 for the observed CEERS dataset, color-coded by visual classification. G quantifies the relative distribution of the galaxy’s flux, whereas M20 is the second-order moment of the brightest 20% of the galaxy’s flux. G vs. M20 does not appear to effectively separate visually classified mergers and non-mergers, which is to be expected since G – M20 is not sensitive to all stages of a merger. The second and third panels show the merger classifications predicted by the random forest and the DeepMerge network, respectively, color-coded by the merger probability. Probabilities higher than 0.5 mean the object was classified as a merger while probabilities lower than 0.5 mean the object was classified as a non-merger. The second panel shows a potential trend where objects with very low probabilities from the random forest are far below the merger discriminating line while objects with higher probabilities are closer to or above the line. The third panel shows that most galaxies were classified as mergers with relatively high probabilities, and there appears to be no clear trend in relation to the merger discriminating line.
Figure 6 shows the simulated CEERS and observed CEERS merger rates. In each panel, the black line is the theoretical Illustris merger rate derived from Rodriguez-Gomez et al. (2015) assuming a merger timescale of 0.5 Gyr, for comparison. The second and third panels show the simulated CEERS merger rate as determined by the random forests and by the DeepMerge network, respectively, and are in good agreement with each other. Like the simulated merger rates, the observed merger rates in the first and second panel are underestimated compared to the theoretical Illustris merger rate. The DeepMerge observed merger rate appears higher than the simulated CEERS merger rates, the theoretical Illustris merger rate, and the observed merger rates in the first panel due to the fact that DeepMerge overestimated the number of mergers. The random forest observed merger rate is in better agreement with the observed merger rates in the first panel. In addition, the observed merger rates all appear to decrease or remain constant at z > 4 while the simulated mergers rates increase at z > 4.
5. Next Steps
One and a half years after JWST’s first observations, we now have a better understanding of galaxy structure in the early universe and how that structure has evolved with time. We know that early galaxies were diverse and structures such as disks and spheroids were already in place very early on in our Universe’s history. However, many questions remain that can only be answered with large area surveys like COSMOS-Web that provide statistics for many thousands of galaxies, or very deep surveys such as NGDEEP that will be able to detect the faintest structures.
References
Bagley, M.B., Finkelstein, S.L., Koekemoer, A.M., et al. 2023, ApJL, 946, L12, doi: 10.3847/2041-8213/acbb08.
Cameron, E. 2011, PASA, 28, doi: 10.1071/AS10046.
Casey & Kartaltepe, Casey, C.M., Kartaltepe, J.S., Drakos, N.E., et al. 2023, ApJ, 954, 31, doi: 10.3847/1538-4357/acc2bc.
Ćiprijanović, A., Snyder, G.F., Nord, B., & Peek, J.E.G. 2020, Astronomy and Computing, 32, 100390, doi: 10.1016/j.ascom.2020.100390.
Conselice, C.J., Bluck, A.F.L., Ravindranath, S., et al. 2011, MNRAS, 417, 2770, doi: 10.1111/j.1365-2966.2011.19442.x.
Costantin, L., Pérez-González, P.G., Vega-Ferrero, J., et al. 2022, arXiv e-prints, arXiv:2208.00007. https://arxiv.org/abs/2208.00007
Kartaltepe, J.S., Mozena, M., Kocevski, D., et al. 2015, ApJS, 221, 11, doi: 10.1088/0067-0049/221/1/11.
Kartaltepe, J.S., Rose, C., Vanderhoof, B.N., et al. 2023, ApJL, 946, L15, doi: 10.3847/2041-8213/acad01.
Mortlock, A., Conselice, C.J., Hartley, W.G., et al. 2013, MNRAS, 433, 1185, doi: 10.1093/mnras/stt793.
Peng, C.Y., Ho, L.C., Impey, C.D., & Rix, H.-W. 2002, AJ, 124, 266, doi: 10.1086/340952.
Peng, C.Y., Ho, L.C., Impey, C.D., & Rix, H.-W. 2010, AJ, 139, 2097, doi: 10.1088/0004-6256/139/6/2097.
Rodriguez-Gomez, V., Genel, S., Vogelsberger, M., et al. 2015, MNRAS, 449, 49, doi: 10.1093/mnras/stv264.
Rodriguez-Gomez, V., Snyder, G.F., Lotz, J.M., et al. 2019, MNRAS, 483, 4140, doi: 10.1093/mnras/sty3345.
Rose, C., Kartaltepe, J.S., Snyder, G.F., et al. 2022, arXiv e-prints, arXiv:2208.11164. https://arxiv.org/abs/2208.11164
—. 2023, ApJ, 942, 54, doi: 10.3847/1538-4357/ac9f10
Somerville, R.S., Popping, G., & Trager, S.C. 2015, MNRAS, 453, 4338, doi: 10.1093/mnras/stv1877.
Somerville, R.S., Olsen, C., Yung, L.Y.A., et al. 2021, MNRAS, 502, 4858, doi: 10.1093/mnras/stab231.
Yung, L.Y.A., Somerville, R.S., Finkelstein, S.L., Popping, G., & Davé, R. 2019, MNRAS, 483, 2983, doi: 10.1093/mnras/sty3241.
Yung, L.Y.A., Somerville, R.S., Ferguson, H.C., et al. 2022a, arXiv e-prints, arXiv:2206.13521. https://arxiv.org/abs/2206.13521
—. 2022b, MNRAS, 515, 5416, doi: 10.1093/mnras/stac2139.
[1] https://users.obs.carnegiescience.edu/peng/work/galfit/galfit.html
[2] https://www.nottingham.ac.uk/astronomy/megamorph/
[3] https://statmorph.readthedocs.io/en/latest/