The Y Dwarf Population with HST: unlocking the secrets of our coolest neighbours -- III. Near-Infrared Photometry

TL;DR

This study delivers a homogeneous, high-precision near-infrared photometric dataset for 21 Y dwarfs using HST/WFC3 across F105W, F125W, and F160W, enabling tight colour–magnitude diagrams when combined with new parallaxes. By applying refined PSF-fitting photometry and careful zero-point calibration, the authors trace well-defined Y-dwarf sequences and explore multiple colour–magnitude and colour–colour spaces, including HST–Spitzer and HST–JWST combinations. Comparisons to atmospheric model grids (ATMO, Lacy & Burrows, and Sonora Elf Owl) reveal that no single model reproduces all observed colours; low-metallicity grids often provide the best global fits, but significant degeneracies among metallicity, C/O ratio, vertical mixing, and cloud physics persist. The results underscore the necessity of uniform, precise datasets to constrain the cooling sequence and guide improvements in atmospheric models for the coldest substellar objects, with JWST offering a path to extending these constraints to the coldest regime. Overall, the work establishes empirical baselines that challenge current models and inform the interpretation of similarly cold exoplanet atmospheres.

Abstract

Y dwarfs represent the coldest class of brown dwarfs, with effective temperatures below 500K, and provide unique analogues to cold giant exoplanets. We present a large compilation of uniform near-infrared photometry from the Hubble Space Telescope for 21 Y dwarfs across multiple WFC3/IR filters, including the F105W, F125W and F160W bands. We employed refined PSF-fitting and calibration procedures to reach photometric uncertainties at the 0.02-0.05 mag level for most targets. Combined with precise parallax measurements, our data reveal well-defined Y-dwarf sequences in near-infrared colour-magnitude diagrams, observed with a markedly improved tightness. Known photometric trends emerge with minimal scatter, including the continuous redward progression in F125W-F160W with decreasing temperature, and the blueward trend in F105W-F125W with possible hints of a reversal around 350K. Comparisons to the ATMO, Sonora Elf Owl, and Lacy & Burrows atmospheric models highlight systematic discrepancies, in particular F105W-F125W and F105W-F160W colours predicted to be too red. Low-metallicity grids provide the best fits to the global Y-dwarf population, although closer inspection across wavelengths shows that these preferences likely reflect compensating effects in missing or incomplete physics rather than true population-level abundances. While some atmospheric diversity is expected among Y dwarfs, their tight observational sequences and systematic offsets from model predictions reveal that key physical and chemical processes remain inadequately captured in current grids. Our results underscore the importance of high-precision, internally consistent datasets in robustly tracing the Y-dwarf cooling sequence and providing the empirical constraints needed to advance theoretical models at the lowest temperatures.

Figures (9)

• Figure 1: HST CMDs for our 21 Y dwarfs, showing all combinations of WFC3/F105W, F125W and F160W colours against F125W absolute magnitude. The target with a black edge is the known binary WISE 0336$-$0143 Calissendorff2023, unresolved in HST observations. The effective temperatures in the colour bar are reported in Table \ref{['t:photometry']}. The grey points in the top panel show synthetic photometry for L and T dwarfs from the SpeX Prism Spectral Library SpeX2014. The bottom panel shows zoomed-in views of the regions within the grey boxes in the top panels.
• Figure 2: Comparison of the improved tightness in NIR CMDs for Y dwarfs from our HST observations (black) relative to the compilation of YJH MKO photometry from Leggett2021 (grey). Filled symbols mark objects present in both samples, and only a few open symbols indicate objects missing from one of the datasets: WISE 2354$+$0240 from our sample which lacks $Y$-band photometry (left and middle panels), and 4 Y dwarfs at the bright end of the ground-based sample that are not part of our HST sample. While plotted on the same scales for direct comparisons of the local scatters and size of uncertainties, it is important to note that axes correspond to different filter systems, and magnitude or colour values should not be directly compared between datasets.
• Figure 3: HST--Spitzer CMDs for our 21 Y dwarfs, showing WFC3/F125W absolute magnitudes against F125W--[4.5] and [3.6]--[4-5] colours (first two panels), and IRAC/[4.5] absolute magnitudes against F125W--[4.5] colours in the right-most panel. Spitzer data come from Kirkpatrick2021. The colour bar and symbols are the same as in Figure \ref{['f:CMD_NIR-wide']}, with bottom panels showing zoom-in views of the shaded regions in the top panels.
• Figure 4: HST vs. JWST colours for the 16 of our 21 Y dwarfs with uniform NIRCam photometry from the data in Albert2025. The first three panels display all combinations of colour indices in our main WFC3 wide bands against NIRCam F150W--F480M colours, and the rightmost panel shows the close 1:1 relationship between the JWST colours and F160W--[4.5] from HST and Spitzer covering equivalent spectral regions.
• Figure 5: Comparison of the measured photometry for our 21 Y dwarfs (grey circles) to atmospheric models from the ATMO family Phillips2020Leggett2021. The different line colours correspond to grids of varying model parameters: three states of chemical (dis)equilibrium for ATMO 2020 (top), and metallicities of $-$[0.5], [0.0] and $+$[0.3] dex for ATMO++ (bottom), with dashed and solid lines corresponding to surface gravities $\log(g)$ of 4.0 and 5.0, respectively. Squares, diamonds and crosses indicate the points along each model curve where the effective temperature T$_\mathrm{eff}$ is 500 K, 400 K and 300 K, respectively. In the right-most panels, the $\overline{\chi}^2_\nu$ values represent the global reduced chi-squares of the sub-models in the corresponding colours to the full population (see Section \ref{['population-fits']}.)