SDSS-IV MaNGA-resolved Star Formation and Molecular Gas Properties of Green Valley Galaxies: A First Look with ALMA and MaNGA

MaNGA is an on-going integral field unit (IFU) survey on the SDSS 2.5 m telescope (Gunn et al. 2006), as part of the SDSS-IV survey (SDSS Collaboration et al. 2016; Blanton et al. 2017). MaNGA makes use of a modification of the BOSS spectrographs (Smee et al. 2013) to bundle fibers into hexagons (Drory et al. 2015). Each spectra has a wavelength coverage of 3500–10,000 Å, and instrumental resolution ∼60 km s⁻¹. After dithering, MaNGA data have an effective spatial resolution of 25 (FWHM; Law et al. 2015), and data cubes are gridded with 05 spaxels.

We make use of the Pipe3D pipeline (Sánchez et al. 2016a) to model the stellar continuum with 156 templates with 39 ages and 4 stellar populations that were extracted from a combination of the synthetic stellar spectra from the GRANADA library (Martins et al. 2005) and the MILES project (Sánchez-Blázquez et al. 2006; Vazdekis et al. 2010; Falcón-Barroso et al. 2011). Details of the fitting procedures are described in Sánchez et al. (2016b). In short, a spatial binning is first performed in order to reach an S/N of 50 accross the entire field of view (FoV) for each datacube. A stellar population fit of the coadded spectra within each spatial bin is then computed. The stellar population model for spaxels with continuum S/N > 3 is then estimated by rescaling the best fitted model within each spatial bin to the continuum flux intensity in the corresponding spaxel, following Cid Fernandes et al. (2013) and Sánchez et al. (2016a). The stellar mass surface density (${{\rm{\Sigma }}}_{* }$) is then obtained using the stellar mass derived for each spaxel and then normalized to the physical area of one spaxel. We derive the emission line fluxes following the same procedure described in Belfiore et al. (2016). Briefly speaking, the fittings are performed on continuum subtracted spectra using sets of Gaussians (one per line) with a common velocity. The dust attenuation is corrected by using the Balmer decrement, adopting the Calzetti (2001) attenuation curve with Rv = 4.05 and a theoretical value for the Balmer line ratio (Hα/Hβ = 2.86) taken from Osterbrock & Ferland (2006), assuming case B recombination. SFR is then estimated based on this extinction corrected Hα flux using the conversion given by Kennicutt (1998a) with the Salpeter IMF. Similarly, we convert the spaxel-based SFR into the SFR surface density (${{\rm{\Sigma }}}_{\mathrm{SFR}}$) by normalizing it to the spaxel area. At at fixed extinction curve and IMF, the uncertainty in the SFR estimate is proportional to that of the Hα flux and is less than 33% given that we only limit to our analysis to spaxels with S/N (Hα) > 3.

We show in the left panel of Figure 1, the locations of the sSFR, defined as the SFR divided by the stellar mass (${M}_{* }$), versus ${M}_{* }$ for 2730 MaNGA galaxies (black dots), from an internal release (labeled MPL5), very closely equivalent to Data release 13 (SDSS Collaboration et al. 2016), by integrating the Pipe3D results from individual MaNGA spaxels. The three green valley targets (MaNGA 1-596678, 1-114956, and 1-596598) for the ALMA follow-up, highlighted by the color-coded stars, were drawn from the first 118 galaxies observed by MaNGA at the time when the ALMA proposal was prepared. They are randomly selected to be massive galaxies that lie below the star-forming main-sequence relation with different separations from the main sequence, ΔsSFR, defined as the offset in log(sSFR) relative to the main-sequence value (i.e., log(sSFR)–log(sSFR_MS)). Previous studies have revealed significant differences in the slope and normalization of the main sequence (e.g., see Speagle et al. 2014). The selection of the star-forming population, the method determining the star formation rate and stellar mass, as well as the IMF, have a strong effect in determining the properties of the main sequence. In light of this complexity, we compute our own value of sSFR_MS based on the Pipe3D results to be self-consistent. The sSFR of the main sequence is determined to be ∼${10}^{-10.18}$ yr⁻¹, as shown in the blue solid line, by taking the median sSFR of galaxies with log(sSFR/yr⁻¹) > −10.6. Our derived sSFR_MS is close to the $z\sim 0$ value (sSFR ∼${10}^{-10.09}$ yr⁻¹) derived using the empirical sSFR versus redshift relation given in Equation (13) of Elbaz et al. (2011). We also require the targets to be accessible by ALMA and we do not impose the constraint on the predicted CO abundance when selecting the targets. We number them 1 to 3 (hereafter Galaxy 1, Galaxy 2, and Galaxy 3) according to their ΔsSFR (see Table 1). Although Galaxy 1 lies close to the lower edge of the star-forming main sequence on the global sSFR—${M}_{* }$ plane, all three galaxies are referred to as "green valley galaxies" loosely in this work.

Zoom In Zoom Out Reset image size

Table 1.  Properties of the Three MaNGA Galaxies

ID	MaNGA ID	R.A.	Decl.	Redshift	$\mathrm{log}$(M${}_{\star }$/${M}_{\odot }$)	$\mathrm{log}$ $\left(\tfrac{\mathrm{SFR}}{{M}_{\odot }\,{\mathrm{yr}}^{-1}}\right)$	ΔsSFR	$\mathrm{log}$(M${}_{{\rm{H}}2}$/${M}_{\odot }$)	$\mathrm{log}$(M${}_{{\rm{H}}{\rm{I}}}$/${M}_{\odot }$)	$\mathrm{log}$ $\left(\tfrac{\mathrm{SFE}}{{\mathrm{yr}}^{-1}}\right)$ b
				(${D}_{r}$ a)
1	1-596678	332.89284	11.79593	0.02695	10.88	0.46	−0.24	9.47c	10.21	−9.0
				(114.7 Mpc)
2	1-114956	332.79873	11.80073	0.02702	10.36	−0.3	−0.48	8.98d	9.87	−9.2
				(115.0 Mpc)
3	1-596598	331.12290	12.44263	0.02659	10.98	0.075	−0.73	⋯	9.7e	⋯
				(113.2 Mpc)

Notes.

^aComoving radial distance. ^bThe global SFE estimated based in the single-dish CO measurements. ^cData taken with JCMT by Ting Xiao et al. ^dData taken from Saintonge et al. (2012). ^eGalaxy 1 is in the edge of the GBT beam (5.5' away) of Galaxy 3 and both galaxies are at very similar redshifts, so the H i can be attenuated flux from the H i linked to that in the edges of the beam. If there is no H i detected linked to Galaxy 3, the upper limit (assuming a width of 400 km s⁻¹) would be $\mathrm{log}$(M${}_{{\rm{H}}{\rm{I}}}$/${M}_{\odot }$) = 9.22 instead.

Download table as:  ASCIITypeset image

These three objects were recently observed as part of the H i-MaNGA program at the Robert C. Byrd Green Bank Telescope (GBT), which is obtaining H i 21 cm observations of a large sample of MaNGA galaxies (AGBT17A_012, PI: K. Masters). Galaxies 1 and 2 have H i gas fractions comparable to that of the normal H i galaxies, while Galaxy 3 is below the ALFALFA scaling relation (see Figure 2(c) of Huang et al. 2012). In addition, Galaxy 1 and Galaxy 2 were also observed in CO (2–1) with JCMT (PI: Ting Xiao) and CO (1–0) with IRAM (Saintonge et al. 2012), respectively, from which the total H₂ mass can be derived. The beam size of JCMT is 22'' and is 325 for IRAM. The general properties of the three green valley galaxies are summarized in Table 1. We assume the CO(2–1) to CO(1–0) ratio to be 0.7 when calculating the total H₂ mass of Galaxy 1.

Zoom In Zoom Out Reset image size

Molecular gas observations in ¹²CO(1–0) were carried out with ALMA in Cycle 3 on 2016 January using the Band 3 receiver (project code: 2015.1.01225.S; PI: Lihwai Lin). The baseline ranges from 15 to 310 meters. The largest structure that we expect to be sensitive to is about 36'' (∼20 kpc). Thus, the missing flux should be negligible. Uranus was observed as the flux calibrator for Galaxy 2, and Neptune was used for Galaxy 1 and Galaxy 3. The phase and bandpass of the observations of Galaxy 1 and Galaxy 2 were calibrated with J2232+1143 and J2222+1213 , respectively, and J2200+1030 and J2148+0657 for Galaxy 3. The on-source time is ∼1 hr for each galaxy.

Our spectral setup includes one line targeting ¹²CO (1–0). The window has a bandwidth of 0.937 GHz (2500 km s⁻¹), with a channel width of 3906.250 kHz (10.1 km s⁻¹). The data were processed by pipeline (version r35932 and r36660) in the Common Astronomy Software Applications package (CASA, version 4.5.1 r35996 and 4.5.3 r36115).

The task CLEAN was employed for deconvolution with a robust = 0.5 weighting (Briggs). We adopted a user-specified image center, pixel size, and restoring beamsize to match the image grid and the spatial resolution of the MaNGA images during the CLEAN process. The user-specified image center is ∼01 away from the original center in the ALMA observations. We adopt a geometric mean beamsize of the user-specified beam, 25 × 25 (∼1.4 × 1.4 kpc), similar to that of the native beamsize reported by the CLEAN (26 × 22). We have confirmed that all results remain unchanged if we instead use the original image center and restoring beamsize. Sensitivities of the three observations are almost identical. The final cubes have channel widths of 10.1 km s⁻¹ and rms noise (${\sigma }_{\mathrm{rms}}$) of ∼0.5 mJy beam⁻¹. Integrated intensity maps were created from the cubes with a clip in noise of 1.5-${\sigma }_{\mathrm{rms}}$. Varying the clipping threshold from 2σ to 1.3σ results in a change of the CO flux from −15% to +10% with respect to the case using 1.5σ. Since the ALMA observations have larger fields of view than MaNGA, the edge of ALMA maps were cut off to match the image size of MaNGA. The H₂ mass surface density (${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$) is computed from the CO surface density by adopting a conversion factor (${\alpha }_{\mathrm{CO}}$) of 4.3 ${M}_{\odot }$ (K km s⁻¹ pc²)⁻¹ (e.g., Bolatto et al. 2013).

We can compare the total CO flux and/or H₂ mass obtained by integrating the ALMA results with those based on single-dish observations for two of our targets. We find that the total ALMA CO(1–0) flux for Galaxy 1 is in good agreement with that derived from the JCMT CO(2–1) observation if adopting a conventional CO(2–1)/CO(1–0) ratio = 0.7. For Galaxy 2, which is part of the COLD GASS sample, its ALMA-integrated H₂ mass is a factor of 1.9 lower than the value listed in the COLD GASS catalog when applying the same ${\alpha }_{\mathrm{CO}}$. We suspect that this discrepancy may be related to the method of aperture correction used in the COLD GASS estimation for this object.

Recently, it has been found that kiloparsec-scale ${{\rm{\Sigma }}}_{\mathrm{SFR}}$ traces well with the underlying ${{\rm{\Sigma }}}_{* }$ for star-forming galaxies (Sánchez et al. 2013; Cano-Díaz et al. 2016; Abdurróuf & Akiyama 2017; Hsieh et al. 2017). This relation may be responsible for the observed tight correlation between the global SFR and ${M}_{* }$. In the right panel of Figure 1, we show the kiloparsec-scale ${{\rm{\Sigma }}}_{\mathrm{SFR}}$ versus ${{\rm{\Sigma }}}_{* }$ relation for the three green valley galaxies. The data points from the bulge and disk regions are shown in solid circles and contours, respectively. The dotted line represents the best-fit of the resolved main-sequence relation obtained for the MaNGA star-forming population (Hsieh et al. 2017): log $\left(\tfrac{\mathrm{SFR}}{{M}_{\odot }\,{\mathrm{yr}}^{-1}}\right)$ = −10.33 + log $\left(\tfrac{{M}_{* }}{{M}_{\odot }}\right)$. For Galaxy 1, the disk almost lies on the resolved main sequence, while the bulge is only slightly below the line. On the other hand, the bulge of Galaxy 2 shows significant departure from the resolved main sequence. We note that the central Hα emission of Galaxy 3 is dominated by broad emission lines associated with an AGN, and therefore the ${{\rm{\Sigma }}}_{\mathrm{SFR}}$ from Hα is an upper limit in the central part of this galaxy. In spite of this, it is clear that both the bulge and disk regions of Galaxy 3 are systematically below the resolved main-sequence relation. There is also a trend that the disk sSFR declines from Galaxy 1 to Galaxy 3, following a similar behavior of the global sSFR. Assuming galaxies evolve with declining sSFR, our results would indicate that the sSFR in the bulge departs from the resolved main sequence first, followed by the disk as the global sSFR decreases.²⁰

Figure 2 shows the optical image, ${{\rm{\Sigma }}}_{\mathrm{SFR}}$, ${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$, star formation efficiency (SFE; defined as SFR/${M}_{{{\rm{H}}}_{2}}$), and the gas fraction (${f}_{\mathrm{gas}}$; defined as ${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$/(${{\rm{\Sigma }}}_{* }$+${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$)) for the three green valley galaxies. We can see that the spatial distributions of these quantities are diverse among the three galaxies. For example, both ${{\rm{\Sigma }}}_{\mathrm{SFR}}$ and ${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$ peak in the central part of Galaxy 1, whereas ${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$ is more evenly distributed in Galaxy 2 and greater in the outskirts in Galaxy 3. Interestingly, all three galaxies show increasing ${f}_{\mathrm{gas}}$ with radius. We note that MaNGA achieves nearly uniform sensitivity across its IFU bundles, which is sufficient to detect the continuum at high S/N in the outskirts of these objects. The observed increases in gas fraction is, therefore, not driven by low S/N in the outskirts of galaxies. To quantify these differences, we next compare the relations among various quantities. The upper panels of Figure 3 show the ${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$ versus ${{\rm{\Sigma }}}_{* }$ relation. For Galaxy 1, ${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$ scales with ${{\rm{\Sigma }}}_{* }$ for both bulge and disk regions. On the other hand, for Galaxies 2 and 3, ${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$ is quite uniform across the bulges, despite the fact that the ${{\rm{\Sigma }}}_{{{\rm{H}}}_{2}}$ correlates with ${{\rm{\Sigma }}}_{* }$ in disks. The lower panel shows the median gas fraction as a function of ${{\rm{\Sigma }}}_{* }$ in bulges (circles) and disks (stars). The gas fraction in the bulges varies significantly among the three galaxies by 1.6 dex, being lower toward the Galaxy 3. On the other hand, the gas fractions in disks are comparable in the three cases, although slightly lower in Galaxy 3. Except for Galaxy 1, ${f}_{\mathrm{gas}}$ in the bulges is significantly lower than in the disks for the other two galaxies.

Zoom In Zoom Out Reset image size

Next, we explore the relation between the SFR surface density and the gas surface density, the so-called "Kennicutt-Schmidt" relation (Kennicutt 1998a), shown in the upper panels of Figure 4. Only spaxels with S/N (CO) > 2 are displayed. The SFE versus ${{\rm{\Sigma }}}_{* }$ is shown in the lower panel.²¹ Except for Galaxy 3 for which the central Hα emission is contaminated by the broad-line region associated with the AGN, the SFEs of the other two bulges are at a similar level and are moderately lower compared to their corresponding disks. Similarly, the disk regions show a wider spread in the resolved SFE. Even though the SFEs are similar in some regions among the three galaxies, their median values systematically decline from Galaxies 1 to 3 by a factor of 3. Since two of our targets also have single-dish CO observations, we can compare the resolved SFE with the global SFE measurements as listed in Table 1. The global SFE is in good agreement with the resolved SFE for Galaxy 1. On the other hand, the global SFE is close to the lower end of the resolved SFE distribution for Galaxy 2. We note that the later is caused by the factor of 1.9 excess in the total CO (1–0) flux estimated by the COLD GASS single-dish measurement compared to the integrated ALMA flux (see Section 2.2).

Zoom In Zoom Out Reset image size

To address the relative importance in controlling the sSFR between gas fraction and SFE, we plot ${f}_{\mathrm{gas}}$ and SFE against sSFR and compute the Spearman's correlation coefficient ρ as shown in Figure 5. The data points associated with the bulge regions in Galaxy 3 are excluded in this analysis given that we cannot measure the SFR directly due to the AGN contamination. For bulges, the relation between ${f}_{\mathrm{gas}}$ an sSFR is stronger than that between SFE and sSFR as indicated by the ρ values, suggesting that the sSFR of bulges is mainly controlled by ${f}_{\mathrm{gas}}$. On the other hand, it is observed that both local ${f}_{\mathrm{gas}}$ and SFE correlate with local sSFR in disks, and the local relations are common among the three disks. For comparison, we also plot the global SFE versus sSFR relations of the COLD GASS sample for galaxies with secure CO detections (Saintonge et al. 2011) in Figure 5 after correcting for the differences in the adopted IMF and ${\alpha }_{\mathrm{CO}}$. Our data points in the disk regions are systematically below the best-fit lines of the COLD GASS sample. This discrepancy may come from the fact that these two samples are averaged over different physical scales. The spatially resolved observations tend to sample CO bright regions, resulting in lower SFE than the global averages. An alternative explanation is that green valley galaxies may form a different correlation from the main sequence. Observations covering a wider range of galaxy populations are needed to conclude whether the observed SFE versus sSFR relation is universal.

Zoom In Zoom Out Reset image size