A simple numerical experiment on the dust temperature bias for Lyman break galaxies at $z\gtrsim 5$

TL;DR

This study uses numerical experiments to investigate potential observational biases in measuring dust temperatures of high-redshift Lyman break galaxies, finding that observed high dust temperatures are likely genuine.

Contribution

It demonstrates that ALMA observations do not bias against high dust temperatures in $z extgreater5$ LBGs and proposes future observations with a 30-m Antarctic telescope to better measure dust temperatures.

Findings

ALMA tends to miss high-$T_d$ objects at total dust luminosity.

High $T_d$ in $z extgreater5$ LBGs is likely real.

Future 450 μm observations can improve dust temperature measurements.

Abstract

Some studies suggest that the dust temperatures ($T_\mathrm{d}$) in high-redshift ($z\gtrsim 5$) Lyman break galaxies (LBGs) are high. However, possible observational bias in $T_\mathrm{d}$ is yet to be understood. Thus, we perform a simple test using random realizations of LBGs with various stellar masses, dust temperatures, and dust-to-stellar mass ratios, and examine how the sample detected by ALMA is biased in terms of $T_\mathrm{d}$. We show that ALMA tends to miss high-$T_\mathrm{d}$ objects even at total dust luminosity. LBGs are, however, basically selected by the stellar UV luminosity. The dust-temperature bias in a UV-selected sample is complicated because of the competing effects between high $T_\mathrm{d}$ and low dust abundance. For ALMA Band 6, there is no tendency of high-$T_\mathrm{d}$ LBGs being more easily detected in our experiment. Thus, we suggest that the observed trend of high $T_\mathrm{d}$ in $z\gtrsim 5$ LBGs is real. We also propose that the 450 $\mu$m band is useful in further clarifying the dust temperatures. To overcome the current shallowness of 450 $\mu$m observations, we examine a future Antarctic 30-m class telescope with a suitable atmospheric condition for wavelengths $\lesssim 450~\mu$m, where the detection is not confusion-limited. We find that, with this telescope, an $L_\mathrm{IR}$-selected sample with $\log(L_\mathrm{IR}/\mathrm{L}_{\odot})>11$ is constructed for $z\gtrsim 5$, and detection in the intermediate-$M_\star$ (stellar mass) range [$9<\log (M_\star /M_{\odot})<9.5$] is much improved, especially at high $T_\mathrm{d}$.