Not Just a Dot: the complex UV morphology and underlying properties of Little Red Dots

TL;DR

Not Just a Dot investigates the nature of Little Red Dots (LRDs) at z~4–8 by combining ultra-deep JWST/NIRCam SW imaging, HST data, and NIRSpec spectroscopy to quantify rest-frame UV morphologies, perform comprehensive SED fitting with and without AGN components, and apply emission-line diagnostics. The work finds a mixed population: about 30% of LRDs show extended, irregular UV morphologies consistent with mergers or clumpy star formation, while the majority remain compact yet still exhibit disturbed UV structures; AGN contributions are present in several fits, with high dust attenuation and BH masses estimated for broad-Hα sources. Bolometric luminosities and BH masses derived under different assumptions are broadly consistent with other recent LRD studies, though some extreme sources challenge standard baryon conversion limits. The combination of UV morphology, SED modeling, and spectroscopic diagnostics demonstrates that LRDs are diverse and likely powered by a combination of star formation, AGN activity, and dynamical interactions, underscoring the value of morphology as a diagnostic tool for high-redshift galaxy evolution.

Abstract

We analyze 99 photometrically selected Little Red Dots (LRDs) at $z \approx 4-8$ in the GOODS fields, leveraging ultra-deep JADES NIRCam short-wavelength (SW) data. We examine the morphology of 30 LRDs, while the remaining 69 appear predominantly compact, with sizes $\leq 400$ pc and no extended components even in stacked SW images. However, their unresolved nature may partly reflect current depth limitations, which could prevent the detection of faint diffuse components. Among the 30 morphologically analyzed LRDs, 50% show multiple associated components, while the rest exhibit highly asymmetric structures, despite appearing as single sources. This diversity in rest-frame UV morphologies may point to interactions or strong internal feedback. We find median stellar masses of $\log_{10}(M_{\star}/M_{\odot}) = 9.07_{-0.08}^{+0.11}$ for pure stellar models with $A_{V} \approx 1.16^{+0.11}{-0.21}$ mag, and $\log{10}(M_{\star}/M_{\odot}) = 9.67^{+0.17}{-0.27}$ for models including AGNs with $A{V} \approx 2.74^{+0.55}_{-0.71}$ mag, in line with recent studies suggesting higher masses and dust content for AGN-fitted LRDs. NIRSpec spectra are available for 15 sources, six of which are also in the morphological sample. Broad H$α$ is detected in 40% (FWHM = 1200-2900 km/s), and one source shows broad H$β$ emission. Emission line ratios indicate a composite nature, consistent with both AGN and stellar processes. Altogether, these results suggest that LRDs are a mixed population, and their rest-frame UV morphology reflects this complexity. Morphological studies of larger samples could provide a new way to understand what drives their properties and evolution.

Figures (10)

• Figure 1: The photometrically selected LRD sample in the GOODS-N and GOODS-S fields at $z\approx4-8$, alongside other recent literature samples: labbe_population_2023akins_cosmos-web_2024barro_extremely_2024kokorev_census_2024perez-gonzalez_what_2024. Our GOODS-S sample overlaps by 70% with perez-gonzalez_what_2024 and by 75% with kokorev_census_2024. The differences arise because perez-gonzalez_what_2024 include sources above $z\approx 8$, which we did not consider in this work, and kokorev_census_2024 included sources for which we did not have coverage in both NIRCam/F115W and NIRCam/F200W, preventing us from fully applying our color criteria (see Section \ref{['section2']}).
• Figure 2: We display two examples from our sample of 99 photometrically selected LRDs in the GOODS-N and GOODS-S fields. For each source, we present two sets of $3\text{$^{\prime\prime}$}\times3\text{$^{\prime\prime}$}$ RGB postage stamps: one using the SW bands (F090W, F115W, and F200W) and another with the classic RGB colors (F090W, F277W, and F444W). A visual comparison between these two sets reveals that, for these two sources—which also have high SNR in the NIRCam SW bands—the morphology appears more complex at the shorter wavelengths compared to the classic compact morphology typically associated with LRDs at longer wavelengths (as highlighted in the classic RGB, in red), suggesting that they are not just a dot.
• Figure 3: Example statmorph output images (middle and rightmost images in each image triplet) along with the associated SW RGB image of galaxy emission (leftmost image), showing the resulting MID and $A_S$ values that comprise our quantitative LRD morphology study. The middle image of each source image triplet shows the segmentation map used specifically to calculate the I statistic, where each distinct intensity maximum is highlighted with a different color. The sources with a single (blue) region represent single sources of emission, however 87% of them show non-zero M statistic values, indicating that the spatial footprints of multiple distinct source regions were detected (see Section 3 and the Appendix for a discussion of the M and I statistics). In the galaxy segmentation image used to calculate shape asymmetry, contained within the $r_{max}$ aperture (cyan line), $A_S$ values greater 0.2 mark a strong asymmetry/disturbance. It can be seen from the representative LRD examples shown here that they present highly asymmetric and complex morphologies, including systems with multiple sources, irregularly shaped single sources with multiple distinct regions of emission, and bright point-like sources embedded in fainter extended emission.
• Figure 4: An adaptation of Figure 5 in freeman_new_2013, where the MID statistics for measuring galaxy morphologies are introduced, but with the classic asymmetry parameter, A, replaced by the shape asymmetry parameter, $A_S$; see Section 3.1. The red symbols are akin to the merger candidates in the referenced work, as they represent the LRDs in our sample exhibiting multiple distinct emission and spatial components (indicated by non-zero I and M values, respectively). Close pairs of sources with similar brightnesses and sizes, such as major-merger candidates, would show I and M values tending towards a value of 1, while lower values of these two statistics would suggest a minor-merger candidate, or a single source with a relatively small and faint 'companion' UV clump of emission. The blue symbols represent those LRDs identified with only a single (I = 0), asymmetric ($A_S > 0.2$) source of emission, but with most ($87\%$) showing multiple non-contiguous pixel regions by their non-zero M statistic values (similar to the “non-regular” non-merger candidates in freeman_new_2013). In short, all of the LRDs examined in the morphological analysis display irregular and extended features. The black lines represent the fit to the multi-source/multi-component LRDs (red points), demonstrating the expected positive linear relationship between these statistics. Based on the random forest regression and classification analysis of the combined MID and A statistics measured for 1639 galaxies in HST/WFC3 H- and J-band images in freeman_new_2013, both the visually labeled (red) multi-component systems (classified as purely mergers in their study) and the (blue) non-merging irregular galaxies are detected with $\approx78\%$ accuracy (i.e., the percentage of correctly classified non-regular or merging galaxies).
• Figure 5: Left panel:$M_{\star}$ as a function of redshift for our sample of 99 photometrically selected LRDs in the GOODS fields. In this plot, $M_{\star}$ comes from the bagpipes runs with AGN. No significant differences arise when plotting, instead, $M_{\star}$ adopting stellar models only. Gray points represent galaxies from JADES DR2/DR3. The large sample of LRDs from akins_cosmos-web_2024 is presented for comparison. The sources with spectra are denoted by yellow stars. Grey points represent the JADES sources in both GOODS-S and GOODS-N. The allowed $M_{\star}$ as a function of redshift for two different star formation efficiencies values are also shown for the JADES area and the full sky. Right panel:$L_{Bol}$ as a function of redshift. We computed $L_{Bol}$ from the intrinsic model SED (i.e., before any dust attenuation) by using the monochromatic luminosity at 5100 Å and a bolometric correction of 9 (richards_spectral_2006). For comparison, we plot the LRD sample from akins_cosmos-web_2024 along with some other recent literature, divided into two groups: confirmed BL AGNs (larson_ceers_2023harikane_jwstnirspec_2023maiolino_jades_2023ubler_ga-nifs_2023bogdan_evidence_2024maiolino_small_2024parlanti_ga-nifs_2024ubler_ga-nifs_2024) and red AGNs (kokorev_uncover_2023furtak_high_2024greene_uncover_2024matthee_little_2024). Assuming an Eddington ratio = 1, we show what $L_{Bol}$ would correspond to log$_{10}(M_{BH}/M_{\odot}) = 6-8$ (horizontal dashed lines).