Final Report Summary - LENSGAL (Investigating the properties of the early universe with lensed galaxies)
Dwarf galaxies at high redshift are believed to be the building blocks from which modern day galaxies are assembled. Observations from large 8 – 10 m class telescopes have propelled our understanding of galaxy evolution in the early universe. However, our current knowledge is based on dedicated surveys, which focus only on high-mass galaxies because these are simply the most luminous ones, but the luminous and massive galaxies only contribute by 1% in numbers to the entire galaxy population in the distant universe. Recent technological advances with large telescopes combined with sensitive instruments, has allowed us to observe the fainter counterparts of the distant galaxy population in greater detail in order to understand their role in the evolving universe. This project explores gravitational lenses, which amplify the light from background galaxies and brings them within reach of detailed spectroscopic observations using large telescopes. The two primary objectives of the project are: 1) To construct models of the mass distribution of the foreground cluster lenses and determine magnification factors by measuring lensed galaxy redshifts. 2) To characterize the physical properties of lensed galaxies.
Work performed:
Part 1a: Lens models of Cluster lenses.
The primary data set for the project is VLT/X-shooter spectra obtained for arc and arclets detected in Hubble Space (HST) images of massive clusters. A large fraction of HST time has recently been allocated to legacy surveys of lensing clusters. Two such major surveys are called CLASH and Frontier Fields Survey. In order to infer the intrinsic properties of the background source galaxies, it is essential to determine the mass distribution in the foreground lenses and infer the magnification factors through lens models. The information needed for lens models are lens and source redshifts as well as projected distances between lenses and background sources. We obtained spectroscopic data for 12 arcs towards 10 massive clusters, including CLASH clusters (Christensen et al. 2012a, MNRAS 427, 1953). The locations of the multiple imaged sources were determined from HST data, and we used Lenstool (Jullo at al. 2007) to derive the lens models.
Part 1b: Lens models of group lenses and galaxy-galaxy lenses.
Building on our expertise with research projects on lensed galaxies (Pettini, Christensen et al. 2010; Christensen et al. 2010; Grillo & Christensen 2011) selected from the Sloan Digital Sky survey, in particular from the survey called Cassowary (http://www.ast.cam.ac.uk/ioa/research/cassowary ), we obtained additional X-shooter data for several systems in the period 2012-2013. Primarily, we aimed at lenses that were particularly massive, both in their stellar mass and dark matter components. Due to bad weather, the data from 2012 were only sufficient for analysis of the lens properties, not the detailed physical properties of the background sources.
For the CASSOWARY survey, lens model accuracies are determined by the spatial resolution of the images. In order to improve the quality of the data compared to the SDSS images, we have obtained optical and near IR images from the Nordic Optical Telescope in 2013. The survey is on-going as we aim to obtain data for a complete sample of 69 targets.
Part 2: Physical properties of lensed galaxies
The data for the arcs in the cluster lensing project (part 1a) was used to measure redshifts between 1
With additional data obtained for part 1b, we have been able to measure detailed physics of some of the Cassowary sources.
Another way to find strong gravitational lenses is provided by the Herschell satellite, which discovered some extremely luminous sub-mm bright sources. Their luminosities are so bright that they are likely to be lensed background galaxies. However, some sources do not have measured redshifts. We have obtained data for 13 sources in 2013, again using X-shooter. The data has been reduced, and partly analysed.
Main results:
We have demonstrated that intrinsically low-mass galaxies with morphologies dominated by compact regions have bright emission lines that are detectable in even short integration times. UV bright galaxies appear to have unusual sets of UV emission lines that have not been detected previously at high-z. High-z galaxies typically have larger star-formation rates (SFRs) and extreme UV flux as well as higher ionization parameters (Christensen et al. 2012a). These high ionization parameters imply that UV lines are relatively brighter than seen in the local universe.
We developed a code to combine spectra and HST photometric data points as input to a photometric redshift code (HyperZ), which allows us to determine stellar masses for the galaxies. Using lensing we are able to investigate masses over a much larger dynamical range (7 < log M* < 11) compared to field studies (9.5< logM*<11). This enables us to explore the intrinsic low-mass end of the luminosity function. Their intrinsic SFRs are on the average also two orders of magnitude smaller than in field galaxies. The strong emission lines detected for the sources also allow us to derive the oxygen metallicities of the galaxies using conventional strong emission line diagnostics. It is known that the three properties: mass, metallicity and SFRs follow a scaling relation in the local universe and out to z~2 (Mannucci et al. 2010). With the lensed galaxies, we demonstrate the scaling relations FMR at z>2 can be extended towards lower stellar masses, but that the scatter increases at higher redshifts (Christensen et al. 2012a).
One of the key problems of measuring metallicities in high-z galaxies is that the diagnostics are calibrated for local galaxies, and that various diagnostics may give metallicities that vary by up to a factor of ~6 (Kewley & Ellison 2008). One of the reasons might be the higher ionization parameters and SFRs in the high-z galaxies. By detecting temperature sensitive emission lines, we compare direct- and indirect methods for metallicity calculations and demonstrate that the R23 diagnostics (Pilyugin et al. 2005) is valid for low-mass galaxies at z>2. With the 4 new measurements, we have more than doubled the number of high-z galaxies where this was possible (Christensen et al. 2012b, MNRAS 427, 1973; Christensen et al. 2013; James et al. 2014).
For the CASSOWARY early type lenses we determine accurate total masses both from lens models and from measurements of stellar velocity dispersions in the lens galaxies. By modelling the lens galaxies spectral energy distribution to derive stellar masses we demonstrate that group lenses contain a larger dark matter fraction relative to lower mass galaxies, but smaller than in galaxy clusters (Grillo et al. 2013).
We search for Lya emission from the z=7.6 candidate lensed galaxy, but it is not detected to a very deep level (equivalent width < 4 A). Nevertheless, we demonstrate that the continuum emission is detected and a clear break in the spectra suggests a redshift of 7.5+/-0.2 (Watson, Christensen et al. 2014, Nature submitted).
Expected final results:
The use of R23 as a valid diagnostics to determine metallicities at z>2, was recently questioned by Steidel et al. (2014), who argue that the O3N2 diagnostics is a better choice (although based only on 3 data points). Closer examinations of their results combined with ours, we are able to demonstrate that R23 is best for low-metallicity galaxies while O3N2 is more appropriate for more massive and metal-rich galaxies. Clearly, the debate will continue for the next couple of years. Further data from UV-bright lensed galaxies is likely to provide the best solution, because we are able to analyse a large dynamical range in stellar masses.
With the data we have assembled in part 1b, we aim to explore the nature an origin of extremely massive lenses at intermediate redshifts. Their velocity dispersions (400-500km/s) as well as compact morphologies, for some of the galaxies, suggest that at least some of the rare and massive compact galaxies discovered at z>2 survive several Gyr (Grillo et al. in prep.).
An analysis of the kinematics of lensed sources show that the velocity dispersion of strong emission lines is clearly correlated with the galaxy stellar masses. This stellar-mass Tully-Fischer relation has been investigated at z<1 (Kassin et al. 2007), but with gravitational lenses we extend the analysis towards lower mass- and higher redshift galaxies (Christensen et al. 2014 in prep). With such a relation, we can infer galaxy masses based on the detection of single emission lines at any redshifts.
For the highest redshift source, we have combined the X-shooter spectra plus HST images with ALMA 226 GHz observations. With a 5-sigma ALMA detection, this is the first time a high redshift ‘normal’ galaxy has been detected. The paper has been submitted to Nature (Watson, Christensen et al. 2014, submitted).