The Bright and Dark Sides of High-Redshift starburst galaxies from {\it Herschel} and {\it Subaru} observations

TL;DR

This study analyzes high-redshift starburst galaxies, revealing that most star formation occurs in heavily obscured regions, and suggests these galaxies are analogous to local ULIRGs, likely resulting from major mergers.

Contribution

It provides detailed spectroscopic analysis of z~1.6 starburst galaxies, highlighting their obscured star formation and metal-rich nature, and links them to local ULIRGs and merger activity.

Findings

Most star formation occurs in optically-thick, obscured regions.

Starburst galaxies are metal-rich outliers from the typical metallicity-SFR relation.

High electron densities suggest intense star-forming environments.

Abstract

We present rest-frame optical spectra from the FMOS-COSMOS survey of twelve $z \sim 1.6$ \textit{Herschel} starburst galaxies, with Star Formation Rate (SFR) elevated by $\times$8, on average, above the star-forming Main Sequence (MS). Comparing the H$\alpha$ to IR luminosity ratio and the Balmer Decrement we find that the optically-thin regions of the sources contain on average only $\sim 10$ percent of the total SFR whereas $\sim90$ percent comes from an extremely obscured component which is revealed only by far-IR observations and is optically-thick even in H$\alpha$. We measure the [NII]$_{6583}$/H$\alpha$ ratio, suggesting that the less obscured regions have a metal content similar to that of the MS population at the same stellar masses and redshifts. However, our objects appear to be metal-rich outliers from the metallicity-SFR anticorrelation observed at fixed stellar mass for the MS population. The [SII]$_{6732}$/[SII]$_{6717}$ ratio from the average spectrum indicates an electron density $n_{\rm e} \sim 1,100\ \mathrm{cm}^{-3}$, larger than what estimated for MS galaxies but only at the 1.5$\sigma$ level. Our results provide supporting evidence that high-$z$ MS outliers are the analogous of local ULIRGs, and are consistent with a major merger origin for the starburst event.

Tables

Table 1. Table 1 Physical properties of the z ∼ 1.6 similar-to 𝑧 1.6 z\sim 1.6 SB sample.

PACS-ID	RA	Dec	zspec	log(M${}_{\star }{}^{})$	SFRFIR    $\pm 1\sigma$	F(H$\alpha$) $\pm 1\sigma$	F([NII]6583) $\pm 1\sigma$
	hours	deg		M⊙	M${}_{\odot }{}^{}y{r}^{-1}$	${10}^{-16}erg{s}^{-1}c{m}^{-2}$	${10}^{-16}erg{s}^{-1}c{m}^{-2}$
300	09:58:24.32	2:15:15.10	1.6706	10.4	332 $\pm$ 8	1.88 $\pm$ 0.04	0.38 $\pm$ 0.03
299	09:59:41.31	2:14:42.80	1.6467	10.1	326 $\pm$ 14	1.24 $\pm$ 0.32	0.43 $\pm$ 0.26
455	09:59:43.88	2:38:08.20	1.6696	10.4	275 $\pm$ 11	1.45 $\pm$ 0.12	0.39 $\pm$ 0.06
491	10:00:05.19	2:42:04.70	1.6366	10.4	209 $\pm$ 21	1.15 $\pm$ 0.08	0.35 $\pm$ 0.06
830	10:00:08.73	2:19:02.50	1.4610	10.7	336 $\pm$ 34	1.53 $\pm$ 0.09	0.41 $\pm$ 0.05
472	10:00:08.95	2:40:10.50	1.5988	10.3	661 $\pm$ 66	0.78 $\pm$ 0.04	0.24 $\pm$ 0.04
135	10:00:15.72	1:49:48.00	1.6508	10.3	188 $\pm$ 6	0.46 $\pm$ 0.19	0.15 $\pm$ 0.16
175	10:00:34.62	1:55:25.40	1.6664	10.4	164 $\pm$ 9	1.32 $\pm$ 0.07	0.30 $\pm$ 0.03
682	10:01:23.96	1:52:28.60	1.4681	10.6	418 $\pm$ 8	1.21 $\pm$ 0.19	0.58 $\pm$ 0.15
197	10:01:34.46	1:58:47.70	1.6005	10.7	394 $\pm$ 11	0.82 $\pm$ 0.07	0.32 $\pm$ 0.08
787	10:02:27.95	2:10:04.70	1.5234	10.6	811 $\pm$ 81	1.42 $\pm$ 0.05	0.75 $\pm$ 0.03
251	10:02:39.64	2:08:47.10	1.5847	10.0	165 $\pm$ 16	2.71 $\pm$ 0.07	0.59 $\pm$ 0.04

Equations

$$A_{\mathrm{H}\alpha ,\mathrm{BD}}=\frac{2.5}{k_{\mathrm{H}\beta }-k_{\mathrm{H}\alpha }}\log [\frac{H\alpha /H\beta }{2.86}]\times k_{\mathrm{H}\alpha }$$

$$A_{H\alpha ,IRX}=2.5\log \times (1+\frac{{\mathrm{SFR}}_{\mathrm{FIR}}}{{\mathrm{SFR}}_{\mathrm{H}\alpha ,\mathrm{obs}}}).$$

Full text

1. Introduction

The majority of Star Forming (SF) galaxies at all redshifts form stars in a quasi-steady state along the Main Sequence (MS), a correlation between their stellar mass (M*⋆) and the Star Formation Rate (SFR, Noeske et al., 2007; Elbaz et al., 2007; Renzini & Peng, 2015). MS galaxies also form a tight sequence in the M⋆-metallicity (Z) plane (the Mass-Metallicity Relation MZR, see e.g., Tremonti et al., 2004; Maiolino et al., 2008), with metallicity increasing with M⋆. The scatter in the MZR is reduced when considering the anti-correlation of Z with the SFR at fixed M⋆* (e.g. Ellison et al., 2008). Mannucci et al. (2010) found such relation to be invariant up to $z\sim 2.5$ and called it the Fundamental Metallicity Relation (FMR). The interpretation of this relation is that the SFR is enhanced by upward fluctuations in the gas inflow rate from the cosmic web, while such inflow dilutes the metal content of the system (e.g., Lilly et al., 2013). A fourth key parameter of SF galaxies is their dust content, which correlates with their M*⋆*, SFR and Z so that more massive objects, as well as most SF or metal-rich galaxies tend to host larger dust reservoirs, thus suffering higher extinction (Garn & Best, 2010; Pannella et al., 2014; Tan et al., 2014).

Besides the MS galaxy population, a population of outliers has been observed at all redshifts, supporting extreme SFRs for their M*⋆*. While such MS-outliers contribute only $\sim 10\%$ to the cosmic Star Formation History (SFH) (Rodighiero et al., 2011), they may represent a key phase in galaxy evolution, having been considered among the progenitors of passively evolving ellipticals (e.g. Cimatti et al., 2008). Local outliers appear to be mainly Ultra-Luminous InfraRed Galaxies (ULIRGs) undergoing a major merger that funnels gas into the nucleus, enhances the Star Formation Efficiency (SFE) and triggers a StarBurst (SB, Sanders & Mirabel, 1996). These systems have strong dust extinction (Monreal-Ibero et al., 2010) and complex kinematic and gas properties (Rich et al., 2015), with strong nuclear outflows and shock dominated regions (Westmoquette et al., 2009). The nature of the high-$z$ counterparts to these SB galaxies is however still debated, as it is not clear yet to which extent their high SFR is uniquely due to a higher SFE, as if experiencing a different mode of SF (e.g. Daddi et al., 2010; Tacconi et al., 2013, 2017; Sargent et al., 2014; Silverman et al., 2015a), or to a higher gas content, or a combination thereof (Genzel et al., 2015; Scoville et al., 2016). Studies of the metal content of high-$z$ SBs may shed light on their nature as the gas-phase metallicity reflects their recent SF activity and is a crucial input when estimating their gas content via either the CO luminosity or the dust continuum emission (e.g. Genzel et al., 2015; Tacconi et al., 2017).

In this work we present an analysis of the Inter Stellar Medium (ISM) properties of 12 Herschel-selected SB galaxies at $1.4\leq z_{\rm spec}\leq 1.7$ with H$\alpha$ detection via near-IR spectroscopy from the FMOS-COSMOS survey (Silverman et al., 2015b). These galaxies have SFRs from $4\ \rm{to\ over}\ 10\times{\rm SFR}_{\rm MS}$ and are analyzed in comparison to the MS population, taking advantages of our complementary studies on MS galaxies at the same redshift (Zahid et al., 2014; Kashino et al., 2013, 2017). Throughout this Letter we adopt a Chabrier (2003) IMF, standard cosmology ($H_{0}=70\rm kms^{-1}Mpc^{-1},\Omega_{\rm m}=0.3,\Omega_{\Lambda}=0.7$), AB magnitudes and a Calzetti et al. (2000) extinction law.

2. Data set and sample selection

SB galaxies are identified for lying above the MS, hence a careful definition of the MS at the redshift of interest is required. We defined the MS equation (${{\rm log(SFR}/{[M}_{\odot}\mathrm{/yr])}}=0.906\times\mathrm{log}(M_{\star}{/[M}_{\odot}\mathrm{])}-7.798$, blue solid line in Fig. 1) from a sample of $\sim 3000$ star-forming Bzk galaxies (Daddi et al., 2004) at $1.4\leq z_{\rm phot}\leq 1.7$ in the COSMOS field, for which we computed M*⋆* and SFR by fitting their Spectral Energy Distribution (SED), using broad-band photometry from the Laigle et al. (2016) catalog and the hyperzmass code (Bolzonella et al., 2000) with Bruzual & Charlot (2003) stellar populations synthesis models and constant SFH.

Our SBs are drawn from a Herschel sample (see Rodighiero et al. 2011 for details about the PACS photometry) at $1.4\leqslant z_{\rm phot}\leqslant 1.7$ having near-IR spectroscopy from the FMOS-COSMOS survey. FMOS observations with the H-long and J-long gratings allows us to detect H$\alpha$, [NII]6549,6583 and H$\beta$,[OIII]4959,5007 emission lines, respectively. Kashino et al. (2013, 2017) and Silverman et al. (2015b) contain further details about the spectroscopic observations and the data analysis. We also refer the reader to our companion papers for the description of emission line fitting and flux measurements, as well as the stacking technique adopted to construct average spectra as shown in Fig. 2.

SB sources analyzed in this work are selected as outliers from the MS defined above, with $\rm SFR\geq 4\times\rm SFR_{MS}$ (see Fig. 1), following Rodighiero et al. (2011). Our selection may overlap with samples of Sub-mm selected Galaxies (SMGs). However, SMGs galaxies represent a mixed population (Rodighiero et al., 2011; Roseboom et al., 2013) that include both massive MS galaxies as well as objects qualified as of SBs according to our criterion. Instead, our criterion selects a pure sample of SB galaxies by construction.

For each Herschel-FMOS source, we compute $M_{\star}$ using homogeneous photometry and methodology as applied for MS galaxies, while fixing the redshift to the FMOS $z_{\rm spec}$. SFRs are measured from the bolometric IR luminosity ($L_{\rm FIR}$), using the calibration of Kennicutt (1998), rescaled to a Chabrier IMF by dividing by a factor of 1.7. The $L_{\rm FIR}$ are computed by integrating the far-IR SED over the range $\lambda\in[8-1000]\mu$m, with the SEDs being derived by fitting the Magdis et al. (2012) templates to the Herschel photometry (PACS 100 and 160 $\mu$m, SPIRE 250, 350 and 500 $\mu$m).

AGN candidates among the SBs are identified and discarded using the BPT diagram and the Kewley et al. (2013) dividing line at $z\sim 1.6$ (when H$\alpha$, H$\beta$, [OIII]5007 and [NII]6583 are detected), the [NII]${}_{6583}/\rm H\alpha$ ratio (if only [NII]6583 and H$\alpha$ are available), X-ray detections and inspection of the mid-IR SED (Marcella Brusa, private communication). With these criteria we identify 8 AGN, resulting in an AGN fraction among the off-MS sample of $\sim\ 40$%, substantially higher than the $\sim\ 8$% identified on the MS by Kashino et al. (2017). The average spectrum of these AGN candidates differs from the one of the “purely SF” population, showing broader H$\alpha$ and [NII]6549,6583 emissions and an additional broad component on the H$\alpha$ emission.

The sample analyzed here includes 12 H$\alpha$-detected SBs (red filled circles in Figure 1, upper panel of Fig. 2). Table 2 summarizes the main physical properties of these sources. Among them, 8 objects have a J-long follow-up (blue circles in Fig. 1, lower panel of Fig. 2). Seven of these objects have molecular gas measurements through ALMA CO 2-1 observations, indicating an average gas fraction $f_{\rm gas}\sim 50\%$, which implies very high SFE and short depletion timescale (Silverman et al., 2015a, $\sim 40$ Myrs).