Euclid: Discovery of bright $z\simeq7$ Lyman-break galaxies in UltraVISTA and Euclid COSMOS

TL;DR

The paper leverages UltraVISTA DR6 and Euclid PV imaging, combined with deep optical (HSC) and Spitzer data, to construct two $z\simeq7$ LBG samples and measure the rest-frame UV luminosity function down to and below the knee. Through two SED-fitting paths (with and without Euclid photometry), it achieves a cleaner, more complete LF, finds a knee near $M_{\rm UV}^*\approx -21.14$, and shows that a double-power-law form better captures the bright end when bright measurements are included. The inclusion of Euclid data reduces contamination from ultra-cool dwarfs and reveals fainter galaxies, enabling a joint comparison with JWST results that hints at a gradual evolution of the bright-end slope. The study also forecasts strategies for Euclid Deep Fields to remove dwarfs morphologically and identifies a strong LAE candidate, highlighting the synergy of ground-based, Spitzer, HSC, and Euclid data for exploring the Epoch of Reionisation. Overall, it provides a clean, wide-area view of the $z\simeq7$ LF, informing models of early galaxy formation and guiding future Euclid-driven surveys.

Abstract

We present a search for $z\simeq7$ Lyman-break galaxies using the $1.72 \, \rm{deg}^2$ near-infrared UltraVISTA survey in the COSMOS field, reaching $5\,σ$ depths in $Y$ of 26.2. We incorporate deep optical and Spitzer imaging for a full spectral energy distribution (SED) fitting analysis. We find 289 candidate galaxies at $6.5\leq z \leq 7.5$ covering $-22.6 \leq M_{\rm UV} \leq -20.2$, faint enough to overlap with Hubble Space Telescope studies. We conduct a separate selection by including complementary Euclid performance verification imaging (reaching $5\,σ$ depths of $26.3$), yielding 140 galaxies in $0.65 \, \rm{deg}^2$, with 38 sources unique to this sample. We compute the rest-frame UV luminosity function (UV LF) from our samples, extending below the knee ($M^*=-21.14^{+0.28}_{-0.25}$). We find that the shape of the UV LF is consistent with both a Schechter function and double-power law (DPL) at the magnitudes probed by this sample, with a DPL preferred at $M_{\rm UV}<-22.5$ when bright-end results are included. The UltraVISTA+Euclid sample provides a clean measurement of the LF due to the overlapping near-infrared filters identifying molecular absorption features in the SEDs of ultra-cool dwarf interlopers, and additional faint galaxies are recovered. A comparison with JWST LFs at $z>7$ suggests a gentle evolution in the bright-end slope, although this is limited by a lack of robust bright-end measurements at $z>9$. We forecast that in the Euclid Deep Fields, the removal of contaminant ultra-cool dwarfs as point sources will be possible at $J_{\rm E} < 24.5$. Finally, we present a high-equivalent-width Lyman-$α$ emitter candidate identified by combining HSC, VISTA, and Euclid broadband photometry, highlighting the synergistic power these instruments will have in the Euclid Auxiliary Fields for identifying extreme sources in the Epoch of Reionisation.

Figures (21)

• Figure 1: Limiting magnitudes, or modal depths ($5\,\sigma$) of the photometric filters used in this work within the COSMOS field. The line widths represent the FWHM of the filter transmission curves, and the depths are reported in Table \ref{['tab:Depths']}. The HSC and VISTA filters are shown in black and orange, respectively, and the four filters are labelled. We also show example SEDs of a UCD and a LBG with the solid and dashed lines, respectively. The model photometry for the UCD and the LBG are shown by the stars and circles respectively, and are colour-coded by their filters. Note that the NIR filters cover the gaps between the VISTA filters. These wavelengths are inaccessible from the ground due to the atmospheric absorption.
• Figure 2: The COSMOS UltraVISTA ultravista footprint with the COSMOS PV footprint overlaid. Optical imaging from HSC-SSP DR3 Aihara22 covers the full UltraVISTA area. The overlapping area between and UltraVISTA covers $0.65 \deg^2$.
• Figure 3: SED fitting of a candidate LBG, EUCL J100041.40$+$020157.5 (hereafter LBG 10004$+$02015), at $z=6.90$. Additional candidates are presented in Appendix \ref{['sec: candidate seds and stamps']}. Left: the SED fitting without data, as part of the U-only selection (see Sect. \ref{['sec: candidate selection']}). Right: the SED fitting including data, as part of the U+E selection. The HSC, VISTA and photometry are shown by the points, diamonds and squares, respectively. The photometry is coloured following Fig. \ref{['fig:filters']}. Non-detections are replaced with $2\,\sigma$ upper limits. We also show the filter widths for VISTA and at the top of the plot. The Spitzer/IRAC model photometry and upper limit in the $3.6\,\micron$ filter is also shown. The blue curve shows the best high-redshift solution, and the grey open circles are its expected model photometry. The orange curve shows the best dusty low-redshift solution, and the red curve shows the best UCD solution. The legend in the top right shows the redshift and $\chi^2$ of the galaxy solutions, and the $\chi^2$ and spectral type of the UCD solution. The inset panel shows the redshift probability distribution for this source. The inclusion of data reinforces the exclusion the UCD solution. Also note that the inclusion of data prefers a bluer slope.
• Figure 4: The $z\simeq7$ LBG sample from this work plotted in photometric redshift $z_{\rm{phot}}$ and absolute rest-frame UV magnitude $M_{\rm{UV}}$ space. Top: the U-only sample (blue). Bottom: the U+E sample (red). The red squares with a black outline indicate galaxies which are not recovered in the U-only sample, and are thus unique to the U+E sample. In both panels, we also show the candidates from Varadaraj23 in the XMM-LSS and ECDF-S fields, and the HST-selected candidates from Bouwens21. The dashed grey lines show the 30%, 40%, and 50% completeness limits, as derived from the injection-recovery simulation (see Sect. \ref{['sec:completeness']}). We also show the marginalised distributions in $z_{\rm{phot}}$ and $M_{\rm{UV}}$ as normalised histograms, and also overplot the distribution of the other sample as the thinner, fainter line for comparison. The mean uncertainties are shown on the bottom left.
• Figure 5: A comparison of the UV LF at $z\simeq7$ measured with the two samples presented in Fig. \ref{['fig:z-Muv']} and Sect. \ref{['sec: candidate galaxies']}. The smaller blue circles show the LF points calculated from the U-only sample. The larger red circles show the LF points calculated from the U+E sample. We use the same binning in both cases, and the LF values are presented in Table \ref{['tab:LF values']}. The brightest bin contains no galaxies from the U+E sample, so we show a $1\,\sigma$ upper limit whose value is noted in Table \ref{['tab:LF values']}. We show results from McLure13, Finkelstein15, Bowler17, Bouwens21, harikane22, Varadaraj23, harikane24, and Franco25. Also shown are the best-fit Schechter and DPL fits found by Bouwens21 and harikane24 respectively.