First Infrared-based implications for the dust attenuation and star formation of typical Ly$α$ emitters
By stacking publicly available deep Spitzer/MIPS 24 $μ$m and Herschel/PACS images for 213 $z \simeq 2.18$ Ly$α$ Emitters (LAEs) in GOODS-South, we obtain a strong upper limit to the IR luminosity of typical LAEs and discuss their attenuation curve for the first time. The $3σ$ upper limit $L_{\rm TIR}^{3σ}= 1.1 \times 10^{10} L_\odot$, determined from the MIPS data providing the lowest limit, gives $IRX \equiv L_{\rm TIR}/L_{\rm UV} \leq 2.2$. Here we assume that the local calibration between the 8 $μ$m emission and the dust SED shape and metallicity applies at high redshifts and that our LAEs have low metallicities as suggested by previous studies. The inferred escape fractions of Ly$α$, $16$--$37%$, and UV continuum, $\ge 44%$, are higher than the cosmic averages at the same epoch. The SMC attenuation curve is consistent with the IRX and the UV slope $β= -1.4^{+0.2}_{-0.2}$ of our stacked LAE, while the Meurer's relation (Calzetti curve) predicts a 3.8 times higher IRX; we also discuss the validity of PACS-based $L_{\rm TIR}^{3σ}$ allowing the Meurer's relation. SED fitting using the Calzetti curve also gives a $\sim 10$ times higher SFR than from the $L_{\rm TIR}^{3σ}$ and $L_{\rm UV}$. With $M_{\star}=6.3^{+0.8}_{-2.0} \times10^8 \mathrm{M_{\odot}}$, our LAEs lie on a lower-mass extrapolation of the star formation main sequence at $z \sim 2$, suggesting that the majority of $z \sim 2$ LAEs are mildly star forming with relatively old ages of $\sim 200$ Myr. The faint $L_{\rm TIR}^{3σ}$ implies that LAEs contribute little to the faint ($\gtrsim 100 μ$Jy) submm number counts by ALMA.