HIGHEST REDSHIFT IMAGE OF NEUTRAL HYDROGEN IN EMISSION: A CHILES DETECTION OF A STARBURSTING GALAXY AT z = 0.376

We are observing one pointing in the COSMOS field with the B configuration of the VLA, which corresponds to a spatial resolution of ∼25 kpc at z = 0.376. The first 178 hr (Phase I) were observed in 2013 (E. Momjian et al 2016, in preparation).

For this preliminary study we imaged the interval 0.36 < z < 0.39 (1025–1045 MHz), which includes a wall with over 250 galaxies with spectroscopically known redshifts, and 60 of these have predicted M_{H i} > 10¹⁰M_⊙. The cube was made with a robustness parameter of 0.8 in CASA, it has a spatial scale of 2048 × 2048 of 2'' pixels, and a total of 320 channels of 62.5 kHz (18 km s⁻¹ at z = 0.376). We first Hanning smoothed the data to increase the signal-to-noise ratio (S/N) and then iteratively subtracted the continuum in the image plane, leading to a typical noise per channel of 75 μJy beam⁻¹. We searched for H i by eye around a subset of the 60 predicted gas-rich galaxies. We are still in the process of searching for H i around other galaxies directly and via stacking.

We detect H i in the galaxy J100054, which has a predicted M_{H i} = 2.8 × 10¹⁰ M_⊙. The H i channel maps (Figure 1(a)) show that the emission is mostly at the 2–3σ level (with a 4σ peak at the position of the galaxy). Although the emission is weak, we trust the detection because the signal is at the location and velocity of the galaxy and it appears in several consecutive independent channels. The emission is seen in seven panels, making it 875 kHz wide, which translates to a velocity width of 246 ± 36 km s⁻¹ after correcting for the channel width. We calculate z_{H i} = 0.3764 ± 0.0002, assuming that the center of the emission is that shown in Figure 1(a), panel 5. We have high confidence in the emission seen in panels 2–5, as it is at the 3σ level with a smooth velocity gradient and there are no negative contours of high significance. The emission in panels 6–8 is weaker and its velocity does not follow a clear pattern. We generate the total H i distribution map (Figure 1(b)) by adding the emission seen in the seven channels after smoothing and applying a 1.25σ cutoff. We also make a moment map excluding the emission seen in panels 6–8, and the morphology is almost identical. We calculate M_{H i} for both maps, M_{H i} = (2.9 ± 1.0) × 10¹⁰ M_⊙ for the one presented here and M_{H i} = 2 × 10¹⁰ M_⊙ from the map only including panels 2–5, showing that the H i in panels 6–8 does not contribute much. As seen in Figure 1(b), the H i is asymmetric and very extended, encompassing some potentially nearby companions. Because the low-level emission in the southern extension is rather sensitive to the noise, the morphology is uncertain. We also include two integrated H i spectra in Figure 1(c), one that integrates over the optical disk of the galaxy and another one that is centered on the H i emission. We calculate a S/N = 7 from the H i spectrum by adding up the signal in the range 1031.5–1032.5 MHz, and dividing by the rms calculated in the range 1033–1038 MHz (taking into account the number of channels). Last, Figure 1(d) shows a position–velocity diagram that further shows the significance of the detection.

Zoom In Zoom Out Reset image size

The last panel of Figure 1(a) (bottom right) shows the continuum emission from the line-free channels. In addition, we have continuum data from the commensal survey CHILES Continuum Polarization (CHILES Con Pol).²² The source has a total flux density of 240 ± 10 μJy at 1.45 GHz, which is consistent with our measurement and the COSMOS-VLA data (Schinnerer et al. 2007). We measure a spectral index of −0.8 ± 0.15, which is the typical value for star-forming galaxies (e.g., Magnelli et al. 2015).

As a follow-up to our H i detection, we observed J100054 using the Redshift Search Receiver (RSR) at the LMT (Erickson et al. 2007) on 2015 April 8 and May 21 for 1 hr each under excellent weather conditions (T_sys ≡ 100 K). The spectrum covering a frequency range 73–111 GHz is obtained with a spectral resolution of 31.25 MHz (∼110 km s⁻¹ at z = 0.376) to search for possible CO(1–0) emission. The effective beam size is 25'' at 84 GHz and telescope pointing is checked on the nearby QSO J0909+013 before each observing session. When averaged together with the 1/σ² weight, the final spectrum has an rms noise of σ = 1.2 mJy.

A single bright line is detected with a S/N = 9.7 centered at 83.7816 GHz (Figure 2) and is interpreted as that of the CO(1–0) transition at z = 0.3759 ± 0.0002. This line is fully resolved, as shown by the zoom-in spectrum shown in the inset, and the derived full width at half maximum (FWHM) is 413 ± 62 km s⁻¹ after correcting for the instrumental resolution. The narrow CO line peak (FWHM is 317 ± 56 km s⁻¹), centered on the optical redshift z = 0.376, sits atop a broad base, which accounts for the observed broadline width. The measured CO line integral of 3.12 ± 0.32 Jy km s⁻¹ translates to L_CO = (2.3 ± 0.2) × 10¹⁰ K km s⁻¹ pc². The H₂ mass depends on the adopted conversion factor, which ranges from α = 0.8 M_⊙/(K km s⁻¹ pc²) for interacting galaxies (Downes & Solomon 1998) to the Galactic conversion factor of α = 4.3 M_⊙/(K km s⁻¹ pc²) (Bolatto et al. 2013). This results in a range of possible values of ${M}_{{{\rm{H}}}_{2}}$ = (1.8–9.9) × 10¹⁰ M_⊙.

Zoom In Zoom Out Reset image size

As J100054 is located within the COSMOS field, extensive multiwavelength data are available, and it is detected in ultraviolet, optical, IR, and by the VLA-COSMOS continuum survey. A deep HST I-band image of J100054 (Figure 3(a)) shows that it is a massive regular barred spiral with clear spiral arms and a prominent bulge. In addition it shows a number of fainter galaxies, suggesting that this may be a small group. The information for the possible companions is scarce, as the four are very faint with r > 23 and only have measured photometric redshifts. Only one of the companions has a photometric redshift similar to J100054, but spectroscopic data are necessary to confirm this and determine whether the other ones are indeed background objects.

Zoom In Zoom Out Reset image size

We obtained an additional long-slit spectrum with the SOAR 4.1 m telescope and the Goodman High Throughput Spectrograph on 2015 April 17 to confirm the redshift of J100054. A slit of 168 width was oriented north–south on the galaxy and the 400 l mm⁻¹ grating yielded spectral coverage 3500–7500 Å at a resolution of R = 495. The 1 hr exposure confirms a galaxy at z = 0.3758 ± 0.0001 with strong [O ii] 3727 Å emission and Balmer absorption, indicating a post-starburst stellar population. The spectrum shows no signs of AGN activity, as seen in Figure 3(b).

The optical and IR counterparts are found exploiting the likelihood ratio technique (Sutherland & Saunders 1992) with the optical photometric data (Capak et al. 2007), Spitzer Space Telescope MIPS 24 μm (Le Floc'h et al. 2009), and the Herschel Space Observatory (Oliver et al. 2010; Lutz et al. 2011). The best-fit SED is obtained through the Le Phare (Arnouts et al. 1999; Ilbert et al. 2006) with the M82 galaxy template for the optical-mid IR regime and the Chary–Elbaz galaxy template for the far-IR regime at the fixed redshift of z = 0.37. The best-fit SED yields the total L_IR(8–1000 μm) = (5.7 ± 0.4) × 10¹¹ L_⊙ and the L_FIR(40–120 μm) = (3.0 ± 0.2) × 10¹¹ L_⊙, confirming that this object is a LIRG.

We compute ${M}_{\ast }=(8.7\pm 0.1)\times {10}^{10}\;{M}_{\odot }$ using the Spitzer IRAC data (Eskew et al. 2012). Following Murphy et al. (2011), we calculate the SFR using the FUV, IR, and radio data SFR_FUV = (0.7 ± 0.1) M_⊙ yr⁻¹, SFR_IR = (85 ± 6) M_⊙ yr⁻¹, and SFR_{1.4 GHz} = (72 ± 3) M_⊙ yr⁻¹ (errors come from the photometric uncertainties). The SFR_FUV is negligible, indicating that most of the star formation is dust-obscured. We adopt SFR_IR for the rest of the paper.