We analyze the multi-wavelength photometric and spectroscopic data of 12 ultraluminous infrared galaxies (ULIRGs) at z ~ 1 and compare them with models of stars and dust in order to study the extinction law and star formation in young infrared (IR) g
alaxies. Five extinction curves, namely, the Milky Way (MW), the pseudo MW which is MW-like without the 2175 Angstrom feature, the Calzetti, and two SN dust curves, are applied to the data, by combining with various dust distributions, namely, the uniform dust screen, the clumpy dust screen, the internal dust geometry, and the composite geometry with a combination of dust screen and internal dust. Employing a minimum chi square method, we find that the foreground dust screen geometry, especially combined with the 8 - 40 M_sun SN extinction curve, provides a good approximation to the real dust geometry, whereas internal dust is only significant in 2 galaxies. The SN extinction curves, which are flatter than the others, reproduce the data of 8(67%) galaxies better. Dust masses are estimated to be in excess of ~ 10^8 M_sun. Inferred ages of the galaxies are very young, 8 of which range from 10 to 650 Myr. The SN-origin dust is the most plausible to account for the vast amount of dust masses and the flat slope of the observed extinction law. The inferred dust mass per SN ranges from 0.01 to 0.4 M_sun/SN.
We apply the supernova(SN) extinction curves to reproduce the observed properties of SST J1604+4304 which is a young infrared (IR) galaxy at z = 1. The SN extinction curves used in this work were obtained from models of unmixed ejecta of type II supe
rnovae(SNe II) for the Salpeter initial mass function (IMF) with a mass range from 8 to 30 M_sun or 8 to 40 M_sun. The effect of dust distributions on the attenuation of starlight is investigated by performing the chi-square fitting method against various dust distributions. These are the commonly used uniform dust screen, the clumpy dust screen, and the internal dust geometry. We add to these geometries three scattering properties, namely, no-scattering, isotropic scattering, and forward-only scattering. Judging from the chi-square values, we find that the uniform screen models with any scattering property provide good approximations to the real dust geometry. Internal dust is inefficient to attenuate starlight and thus cannot be the dominant source of the extinction. We show that the SN extinction curves reproduce the data of SST J1604+4304 comparable to or better than the Calzetti extinction curve. The Milky Way extinction curve is not in satisfactory agreement with the data unless several dusty clumps are in the line of sight. This trend may be explained by the abundance of SN-origin dust in these galaxies; SN dust is the most abundant in the young IR galaxy at z = 1, abundant in local starbursts, and less abundant in the Galaxy. If dust in SST J1604+4304 is dominated by SN dust, the dust production rate is about 0.1 M_sun per SN.
We present the detailed optical to far-infrared observations of SST J1604+4304, an ULIRG at z = 1.135. Analyzing the stellar absorption lines, namely, the CaII H & K and Balmer H lines in the optical spectrum, we derive the upper limits of an age for
the stellar population. Given this constraint, the minimum {chi}^2 method is used to fit the stellar population models to the observed SED from 0.44 to 5.8um. We find the following properties. The stellar population has an age 40 - 200 Myr with a metallicity 2.5 Z_{sun}. The starlight is reddened by E(B-V) = 0.8. The reddening is caused by the foreground dust screen, indicating that dust is depleted in the starburst site and the starburst site is surrounded by a dust shell. The infrared (8-1000um) luminosity is L_{ir} = 1.78 +/- 0.63 * 10^{12} L_{sun}. This is two times greater than that expected from the observed starlight, suggesting either that 1/2 of the starburst site is completely obscured at UV-optical wavelengths, or that 1/2 of L_{ir} comes from AGN emission. The inferred dust mass is 2.0 +/- 1.0 * 10^8 M_{sun}. This is sufficient to form a shell surrounding the galaxy with an optical depth E(B-V) = 0.8. From our best stellar population model - an instantaneous starburst with an age 40 Myr, we infer the rate of 19 supernovae(SNe) per year. Simply analytical models imply that 2.5 Z_{sun} in stars was reached when the gas mass reduced to 30% of the galaxy mass. The gas metallcity is 4.8 Z_{sun} at this point. The gas-to-dust mass ratio is then 120 +/- 73. The inferred dust production rate is 0.24 +/- 0.12 M_{sun} per SN. If 1/2 of L_{ir} comes from AGN emission, the rate is 0.48 +/- 0.24 M_{sun} per SN. We discuss the evolutionary link of SST J1604+4304 to other galaxy populations in terms of the stellar masses and the galactic winds.
