ABSTRACT
We present our results on the structure and activity of massive galaxies at z = 1–3 using one of the largest (166 with M⋆ ⩾ 5 × 1010 M☉) and most diverse samples of massive galaxies derived from the GOODS-NICMOS survey: (1) Sérsic fits to deep NIC3 F160W images indicate that the rest-frame optical structures of massive galaxies are very different at z = 2–3 compared to z ∼ 0. Approximately 40% of massive galaxies are ultracompact (re ⩽ 2 kpc), compared to less than 1% at z ∼ 0. Furthermore, most (∼65%) systems at z = 2–3 have a low Sérsic index n ⩽ 2, compared to ∼13% at z ∼ 0. We present evidence that the n ⩽ 2 systems at z = 2–3 likely contain prominent disks, unlike most massive z ∼ 0 systems. (2) There is a correlation between structure and star formation rates (SFRs). The majority (∼85%) of non-active galactic nucleus (AGN) massive galaxies at z = 2–3, with SFR high enough to yield a 5σ (30 μJy) 24 μm Spitzer detection, have low n ⩽ 2. Such n ⩽ 2 systems host the highest SFR. (3) The frequency of AGNs is ∼40% at z = 2–3. Most (∼65%) AGN hosts have disky (n ⩽ 2) morphologies. Ultracompact galaxies appear quiescent in terms of both AGN activity and star formation. (4) Large stellar surface densities imply massive galaxies at z = 2–3 formed via rapid, highly dissipative events at z > 2. The large fraction of n ⩽ 2 disky systems suggests cold mode accretion complements gas-rich major mergers at z > 2. In order for massive galaxies at z = 2–3 to evolve into present-day massive E/S0s, they need to significantly increase (n, re). Dry minor and major mergers may play an important role in this process.
Export citation and abstract BibTeX RIS
Footnotes
- 10
In Section 4, we use a Chabrier IMF for SFR estimates. Using a Chabrier IMF rather than a Salpeter IMF in estimating the stellar mass would lower the values by a factor of 0.25 dex or less.
- 11
While Figure 6 of Conselice et al. (2011) shows that ∼15%–20% of bright (20 < HAB < 23) galaxies with spectroscopic redshifts are catastrophic outliers in photometric redshift with δz/(1 + z) > 0.5, it should be noted that there are no catastrophic outliers with such large δz/(1 + z) > 0.5 among the 44 galaxies with spectroscopic redshifts in our sample of massive (M⋆ ⩾ 5 × 1010 M☉) galaxies at z = 1–3. The outliers with δz/(1 + z) > 0.5 in the GNS survey have stellar masses below the cutoff value of our sample or/and lie outside its redshift range.
- 12
For the more extended galaxies multiple component (e.g., bulge and disk) decomposition was attempted with limited success and this is discussed in Section 7.1.
- 13
The Hubble types are based on the bulge-to-total light ratio (B/T), which we measured with bulge–disk and bulge–disk–bar decomposition of z ∼ 0 g-band images.
- 14
HyLIRGs are defined to have LIR ⩾ 1013 L☉.
- 15
The mid-IR selection criteria of Lacy et al. (2004) and Stern et al. (2005) were investigated but considered unreliable. Contamination from high-redshift star-forming galaxies drastically reduces their accuracy (e.g., Donley et al. 2008). Applying these methods at z = 1–3 would add more false-positives than true AGNs.
- 16
The MGC catalog assigns the "E/S0" Hubble type and unfortunately does not allow us to identify Es separately.
- 17
The range in PSF FWHM comes from differing positions in the NIC3 field and the PAM values used to create the synthetic PSFs.