We study the infrared ( IR ) properties of high–redshift galaxies using deep Spitzer 24 , 70 , and 160 µm data .
Our primary interest is to improve the constraints on the total IR luminosities , L _ { \mathrm { IR } } , of these galaxies .
We combine the Spitzer data in the southern Extended Chandra Deep Field with a K _ { s } -band–selected galaxy sample and photometric redshifts from the Multiwavelength Survey by Yale-Chile .
We used a stacking analysis to measure the average 70 and 160 µm flux densities of 1.5 < z < 2.5 galaxies as a function of 24 µm flux density , X-ray activity , and rest–frame near-IR color .
Galaxies with 1.5 < z < 2.5 and S _ { 24 } = 53 - 250 \mu Jy have L _ { \mathrm { IR } } derived from their average 24-160 µm flux densities within factors of 2–3 of those inferred from the 24 µm flux densities only .
However , L _ { \mathrm { IR } } derived from the average 24–160 µm flux densities for galaxies with S _ { 24 } > 250 \mu Jy and 1.5 < z < 2.5 are lower than those inferred using only the 24 µm flux density by factors of 2–10 .
Galaxies with S _ { 24 } > 250 \mu Jy have S _ { 70 } / S _ { 24 } flux ratios comparable to sources with X-ray detections or red rest–frame IR colors , suggesting that warm dust possibly heated by AGN may contribute to the high 24 µm emission .
Based on the average 24–160 µm flux densities , nearly all 24 µm–selected galaxies at 1.5 < \hbox { $z _ { \mathrm { ph } } $ } < 2.5 have \hbox { $L _ { \mathrm { IR } } $ } < 6 \times 10 ^ { 12 } L _ { \odot } , which if attributed to star formation corresponds to \Psi < 1000 M _ { \odot } yr ^ { -1 } .
This suggests that high redshift galaxies may have similar star formation efficiencies and feedback processes as local analogs .
Objects with \hbox { $L _ { \mathrm { IR } } $ } > 6 \times 10 ^ { 12 } L _ { \odot } are quite rare , with a surface density \sim 30 \pm 10 deg ^ { -2 } , corresponding to \sim 2 \pm 1 \times 10 ^ { -6 } Mpc ^ { -3 } over 1.5 < z < 2.5 .