Unresolved stellar populations in high redshift galaxies

Stellar populations in high-redshift galaxies cannot be resolved in single stars, this means that we need of observed integrated quantities, like magnitudes and spec- tra, to investigate their stellar content. Stellar populations synthesis models are fundamental and powerful tools to model the integrated properties of unresolved systems. Key information concerning galaxy formation and evolution can be pro- vided by using physical parameters, like stellar mass, age, star formation rate, metal and dust content. This is the reason why evolutionary stellar population synthesis models are already widely used in codes performing fits of the spectral energy distributions (SEDs) of massive sample of high redshift galaxies observ- able in large photometric and spectroscopic surveys. However, the accuracy of the inferred parameters depends on the reliability of the adopted models. Major uncertainties come from the stellar ingredients and from assumptions adopted in stellar evolutionary models. Therefore, it is crucial to calibrate these models on systems for which we have accurate and independent information about physical parameters, like age and metal content. In this context, Galactic globular clusters (GGCs) play a key role: their stellar population is characterized by uniform age and metallicity and also, it can be resolved in each fundamental stellar evolutionary phases. Thus, the main aim of this work is to empirically calibrate stellar population synthesis models against a sample of ∼ 70 GGCs for which we have unprecedented accurate multi-band photometry from far-UV to mid-IR. Firstly, we analysed a fundamental stellar evolutionary phase, namely the hor- izontal branch (HB) phase, characterized by stars burning helium in their core and hydrogen in a surrounding shell. We took advantage of our sample of GGCs combining the homogeneous and accurate optical ground-based V and I photome- try with HST/F606W (∼ V) and HST/F814W (∼ I) space-based photometry. We introduced a new HB index, τ HB , defined as the ratio between the areas subtended by the cumulative number distribution in magnitude and in colour of all HB stars. We found a linear correlation between τ HB and the absolute ages of the clusters in our sample. We also found a quadratic anti-correlation with [Fe/H] which be- comes linear after the removal of age effect in τ HB values. Moreover, we were able to identify a sub-sample of eight "second parameter" clusters, peculiar according 3to their τ HB values and showing bluer HB morphology when compared to typical clusters of similar metallicity. After analysing the GGCs, we investigated the HB morphology in a sample of 19 nearby Local Group dwarf spheroidal (dSph) galaxies: eight of them are Milky Way satellites, nine are Andromeda satellites, while two of them are isolated galaxies. For the Milky Way dwarfs, we have data in ground-based U BVRI data, while for the rest of the sample we have photometry in HST/F475W (∼ B) or HST/F606W (∼ V) and HST/F814W (∼ I). We found that nearby dSphs cover a limited range in HB morphologies, indeed they range from HBR 0 ∼ 1 (Andromeda III) HBR 0 ∼ 2.5 (Ursa Minor) and from τ HB ∼ 1 (Leo II) to τ HB ∼ 5 (Ursa Minor). This indicates that HB morphologies are systematically redder than in GGCs. Moreover, they display a modest variation both in metallicity and age, thus suggesting a relevant environmental effect when moving from GGCs (M = 10 4 − 10 6 M ) to dSphs (M = 10 7 − 10 9 M ). To complete our investigation concerning the resolved stellar populations, we exploited our GGC multi-band photometry to calibrate stellar population synthesis models. Indeed, we employed integrated observed/synthetic data in GALEX/FUV, GALEX/NUV bands and HST/F275W. In the optical range we used homo- geneous and accurate U BVRI ground-based and HST/F606W and HST/F814W space-based photometry. Moreover, we used the first Pan-STARRS data release in grizy filters. In the near-IR wavelengths, we employed the JHK s photometry from 2MASS, while in the mid-IR we calculated synthetic integrated magnitudes in four IRAC filters on board of SPITZER. We performed an empirical calibration of Bruzual & Charlot (2003) stellar population synthesis model. In particular, we compared the inferred cluster ages from SED fitting to the ones obtained through photometric techniques found in the literature. We found a difference of ∆Age between 3Gyr and 10Gyr for most of the clusters in the sample. Moreover, ∆Age seems to have a mild correlation with metallicity, in the sense that it seems to decrease as metallicity increases. Finally, we compared the rest-frame colours pre- dicted by Bruzual & Charlot (2003) model to the ones obtained by other stellar population synthesis models. We found that, in the optical-NIR wavelength range, they agree quite well with observations. On the contrary, in the FUV band, they systematically under-estimate colours in the metal-rich regime ([Fe/H]>-1), while they over-estimate NIR colours in the metal-poor regime ([Fe/H]<-1). After having analysed the resolved simple (GGCs) and composite (dSph galax- ies) stellar populations, we dedicated our efforts to investigate the unresolved stel- lar populations in high redshift galaxies through the galaxy stellar mass function (GSMF), which is a fundamental statistical tool to investigate the stellar mass assembly in galaxies from the re-ionization era to present time. We compute the GSMF through the non-parametric 1/V max method in 0.2 < z < 4.0. We combined the five CANDELS fields (GOODS-South, GOODS-North, EGS, COSMOS, UDS), the Frontier Fields Parallels (Abell 2744, MACSJ0416, MACSJ0717, MACSJ1149) 4and the Ultravista/COSMOS field for a total area of 1.65 deg 2 . This sample al- lows us to explore the GSMF from low masses (∼ 10 8 M ) to a very high-mass end (∼ 10 12 M ), thanks to both deep photometry and large area. Our total GSMF is in good agreement with the one estimated in previous works by different authors. As shown by previous studies, we found a progressive steepening of the faintest region of the function, while its "knee" shows a slight decrease with redshift. Finally, we analysed the GSMF of star-forming and quiescent populations. We found that, while the GSMF of the star-forming population has a shape similar to total one, thanks to our deep photometry, we are able to observe a bi-modality in the quies- cent GSMF. This bi-modality seems to be connected, according to empirical and theoretical evidence available in the literature, with different physical mechanisms able to cease the star formation. Finally, in the context of a more extended project in which we plan to re- calibrate SPSMs on high redshift galaxies, we decided to firstly investigate the impact of the SPSM calibration on the GSMF of quiescent galaxies. Indeed, we would expect to find an age difference in stellar populations of quiescent galax- ies similar to the one found for GGCs. In fact, these stellar systems are domi- nated by old (t>10 Gyr) stellar populations, therefore, an error on inferred ages could also affect the quiescent GSMF at low redshifts. To overcome this prob- lem, we identified a new prescription in our SED fitting code to avoid too young, non-physical age estimates for quiescent galaxies. The new recipe provided accu- rate both re-calibrated stellar masses and GSMF. The age estimates, as expected, mildly (< 0.1dex) affects the GSMF at low redshift (z < 1.5) and only at the lowest mass end (M < 10 8 M )