Title: Hα Equivalent Widths from the 3D-HST survey: evolution with redshift and dependence on stellar mass
Authors: Mattia Fumagalli et al.
First Author’s Institution: Leiden Observatory, Leiden University
In her Russell Prize Lecture at the Anchorage AAS meeting, Sandy Faber discussed the history of star formation in a lambda-cold dark matter (LCDM) Universe. LCDM predicts the hierarchical clustering of dark matter halos exceedingly well, but our understanding of the other key component of galaxy evolution (the infall and collapse of ordinary matter) is still incomplete, since it involves complicated “baryonic processes” because gravity is not the only force that dictates the interaction of ordinary matter. One of the most fundamental quantities arising from these processes is the cosmic star formation rate density, which depends on the star formation rate (SFR) in galaxies and the galaxy luminosity function (i.e., how much star formation is occurring in galaxies and how many galaxies there are). Unfortunately, studies of SFRs in galaxies are plagued by observational uncertainties, especially regarding the evolution of SFR with redshift. In a paper recently submitted to the Astrophysical Journal Letters (ApJL), Mattia Fumagalli and his collaborators report on the history of star formation from 0.8 < z < 1.5 using measurements of the Hα line, making it possible for the first time to directly compare the SFRs of galaxies the local Universe with those at intermediate redshift.
A galaxy’s star formation rate is not an observable quantity and must be inferred from other observations. The most common star formation indicators are a galaxy’s (1) ultraviolet continuum, (2) infrared luminosity, and (3) the strength of recombination emission lines (such Hα and [OII] being emitted in HII regions around young stars). These indicators work because they are essentially tracing the amount of energy produced by young stars, either directly (in the case of the ultraviolet continuum) or “reprocessed” by dust when it is absorbed and re-emitted thermally in the infrared. Likewise, the Hα luminosity is generated when ionized hydrogen in HII regions recombines to an excited state and the electron cascades down to lower energies. Each of the methods has its own shortcomings, and building a picture of how star formation changes with cosmic time is especially challenging, since all of these methods are typically not available for a given galaxy, due to the redshifting of its spectrum. For example, the Hα line at 6563 Å is redshifted into the infrared for z > 0.5, but ground-based infrared spectrometers have to contend with a much higher background and are generally less sensitive than their optical counterparts, limiting astronomers’ ability to efficiently detect Hα in intermediate- and high-redshift galaxies. It is possible to compare SFR estimates from different indicators, but uncertainties in the calibration of these schema is a definite hurdle.
Fumagalli et al. avoid the problem of comparing SFRs from different methods by using data from a new survey with the Hubble Space Telescope, 3D-HST. 3D-HST is a near-infrared spectroscopic survey covering ~600 square arcminutes, situated in the same fields as other extragalactic surveys like GOODS and COSMOS. The total survey area is a little less than the apparent size of the moon and contains roughly 10,000 galaxies with z > 1. The spectra themselves are low-resolution slitless spectra obtained using a grism and the WFC3 camera on Hubble, with wavelength coverage from ~11,000 Å to ~16,500 Å (most of the J and H bands, for infrared astronomy aficionados). This means Hα falls in the spectra of galaxies with 0.7 < z < 1.5.
The galaxies included in 3D-HST sample have measured redshifts from a combination of their spectra and photometry, and their masses are estimated using spectral energy distribution fitting, which attempts to match the observed photometry with stellar population synthesis models. The Hα fluxes are determined by fitting a Gaussian line profile to the flux above the continuum in the region of the spectrum where Hα is expected to fall. Measurements for other samples at lower and higher redshift are taken from the Sloan Digital Sky Survey (SDSS), the VIMOS Very Large Telescope Deep Survey (VVDS), and a paper by Dawn Erb et al. in 2006.
A more interesting quantity than just Hα flux is the Hα equivalent width, or EW(Hα), which is shown as a function of stellar mass in Figure 1. Since EW(Hα) is just the ratio of Hα luminosity (which comes from reprocessed ultraviolet light from young stars) to the continuum luminosity near the Hα line (which comes from all the old stars), it is effectively a measure of the amount of current star formation compared to the amount of previous star formation. Therefore, unlike Hα flux, which probes the total SFR in a galaxy, EW(Hα) is a proxy for the star formation rate per unit mass (i.e., the specific star formation rate, sSFR).
The data show that at every redshift, the galaxies with the highest stellar mass tend to have the lowest EW(Hα) (or, equivalently, sSFR), which is consistent with previous observations. The authors quantify the trend by averaging EW(Hα) in 0.5 dex-wide mass bins (i.e., 10.0 < log(M*/M☉) < 10.5) for only the star forming galaxies (those having EW(Hα) with signal-to-noise greater than 3) and also for the entire sample (where EW(Hα) is set to zero for galaxies with a EW(Hα) signal-to-noise less than 3). Overall, the Hα equivalent width is 5 times higher for the low-mass galaxies compared to the high-mass ones. They also consider the evolution of EW(Hα) with redshift. For a given mass bin, EW(Hα) increases with redshift like (1+z)1.8, which is independent of mass.
Converting measurements of Hα to an exact sSFR requires an understanding of how much the measurement of the Hα line has been affected by extinction from dust. The amount of extinction can be measured if more than one line in the Balmer series (of which Hα is the brightest) is observed, as this allows astronomers to compare the observed ratio between the lines with what it would be without any extinction (i.e., the Balmer decrement). Without this, the authors naively assume zero extinction and use the Kennicutt-Schmidt law to convert their Hα fluxes to SFRs. They find that the low-mass galaxies have sSFRs 15-20 times greater than the high-mass galaxies and that sSFR also increases with increasing redshift, but does so more quickly than EW(Hα), with sSFR ∝ (1+z)3.2 (Figure 2). The relation is steeper because the mass-to-light ratio also evolves with redshift.
The paper by Fumagalli et al. is an exciting new step in understanding the star formation history of the Universe, but leaves plenty of room for new work. In particular, the team looks forward to performing the same analysis on the full 3D-HST sample, which will include grism spectroscopy from ACS as well as WFC3 and allow them to measure Balmer decrements and determinate accurate sSFRs.