The Extended [C ii] under Construction? Observation of the Brightest High-z Lensed Star-forming Galaxy at z = 6.2

The target of the ALMA observations, MACS0308-zD1, was identified in the Reionization Lensing Cluster Survey (RELICS; Coe et al. 2019) in the background of the lensing cluster MACS0308+26. The galaxy was first identified as one of the triply imaged galaxies in the field at z > 6 (Acebron et al. 2018), and our target is the brightest among the three images (R. A. = 03: 08: 53.407, decl. = + 26: 44: 58.93, which originally given the ID of MACS0308-c4.3 in Acebron et al. 2018). MACS0308-zD1 was found as the apparently UV-brightest z > 6 galaxy having m_AB = 23.0 mag, and an estimated photometric redshift of z_ph ∼ 6.2–6.3 (Salmon et al. 2020). Welch et al. (2023) further analyzed MACS0308-zD1 by forward modeling of the galaxy in the source plane and found an updated robust photometric redshift of z_ph = 6.21, and that MACS0308-zD1 consists of a bright clump: MACS0308-zD1.1, and a faint diffuse component. The bright clump was found to have a radius of r = 27 ± 5 pc in its source plane, and to have an extremely high-SFR surface density of Σ_SFR = 900 ± 300 M_⊙ yr⁻¹ kpc⁻². This Σ_SFR is the highest value ever found at z ∼ 6 star-forming galaxies (Kennicutt & Evans 2012), suggesting that the clump could be an environment similar to the starburst galaxies. The diffuse component of MACS0318-zD1 has a radius of r = 900 ± 200 pc, and has Σ_SFR ∼ 0.22 ± 0.08 M_⊙ yr⁻¹ kpc⁻². Detailed source plane analysis in Welch et al. (2023), thus, showed that MACS0318-zD1 consists of higly star-forming clump and a diffuse extended structure.

MACS0308-zD1 is found to be gravitationally magnified by a factor of ∼20 (Acebron et al. 2018; Salmon et al. 2020; Welch et al. 2023). However, such a large magnification factor is usually uncertain. Zitrin et al. (2015) found a typical uncertainty of ∼60% for a magnification factor of ∼20 depending lens models. Further, magnification changes across the stretched structure of the galaxy. While the bright clump has a magnification of 22, the farthest tail has a magnification of 15 (Welch et al. 2023). In this paper, we use a fiducial magnification factor of μ = 20 for our analysis, and apply a conservative systemic uncertainty of 60% for the magnification factor where applicable.

MACS0308-zD1 has been observed in the ALMA program #2021.1.00143.S (PI: Fudamoto). The observation scanned to search the [C ii] 158 μm emission line at a sky frequency range between 245 and 273 GHz that covers entire photometric redshift probability of the target (Acebron et al. 2018; Salmon et al. 2020; Welch et al. 2023). To efficiently scan the frequency range, the observations used four separate spectral tunings (T1 to T4; see Figure 1). The correlator was set up in dual polarization and frequency domain mode, covering 1.875 GHz bandwidth having a native spectral resolution of ∼9 km s⁻¹.

Zoom In Zoom Out Reset image size

All tunings were observed during multiple observing runs using the ALMA cycle-8 C43-3 configuration. The on-source time was ∼0.5, ∼1.6, ∼1.0, and ∼1.0 hr, respectively. The precipitable water vapor (PWV) during the observations was 0.545, 1.450, 0.544, 0.636 mm for tunings T1, T2, T3, and T4, respectively. The relatively large variations of the on-source time were mainly due to variations in the weather conditions.

For the data calibration, we asked East Asian ALMA Regional Center for the data calibration support ⁴⁰ to deliver the calibrated data that are produced using the pipeline code created during the second quality assurance procedure (i.e., ScriptForPI.py). The observed data were analyzed using The Common Astronomy Software Application (CASA) version 6.2.1 (CASA Team et al. 2022).

To search the [C ii] 158 μm emission line, we created a data cube combining all four tunings using CASA task tclean using NATURAL weighting scheme, a spectral binning of 30 km s⁻¹, a pixel scale of 01, and an image size of 30'' × 30'' (i.e., a large enough sky area that contains the primary beam attenuation <50%). With CASA task tclean, we performed the synthesized beam deconvolution using a fixed stopping threshold of 3 times the pixel-by-pixel RMS for all channels and the maximum iteration of 500 times. The RMSs were separately calculated and used for tclean processes of each T1, T2, T3, and T4 tuning data to take into account the variations of the average RMS of each tunings. The conservative deconvolution threshold does not affect the final data products as there are no bright sources in this velocity binning that could contaminate the resulting image products. The produced data cube has a synthesized beam full width at half maximum (FWHM) of 131 × 073, 094 × 068, 111 × 078, and 103 × 073 for tunings 1, 2, 3, and 4, respectively. The achieved average pixel-by-pixel RMSs are 0.28, 0.30, 0.21, 0.23 mJy beam⁻¹ with 30 km s⁻¹ bins, for T1, T2, T3, and T4, respectively. The variations in the sensitivity by a factor of up to 1.4 are due to the variations of the PWV under which each tuning was observed and also due to the weak atmospheric absorption features (see Figure 1).

Using the location of the rest-frame UV emission as a prior of the spatial position of [C ii], we found the [C ii] 158 μm emission line from MACS0308-zD1 at a frequency of 263.6995 ± 0.004 GHz (Figure 1).

The λ_rest ∼ 158 μm continuum image of MACS0308-zD1 was created using CASA task tclean with the multifrequency synthesis mode. During the continuum imaging, we conservatively excluded the whole spectral window that contains the entire [C ii] 158 μm emission line. Since we only excluded one spectral window out of the 16 total spectral windows from our four spectral setups, the exclusion does not affect the resulting sensitivity of the continuum map. The synthesized beam deconvolution was performed using NATURAL weighting scheme, using a stopping threshold of 27 μJy beam⁻¹ (i.e., 3 × the pixel-by-pixel RMS of the dirty image) and by setting the maximum number of iteration to 500 times. Again, the deconvolution thresholds do not affect the result, as there are no bright sources in the continuum image. As a result, we obtained a dust continuum image of MACS0308-zD1 that has a final pixel-by-pixel RMS of 8.9 μJy beam⁻¹ (Figure 2).

Zoom In Zoom Out Reset image size

In the continuum image, there is only a ∼2σ positive signal close to but offset from the rest-frame UV position of the MACS0308-zD1 (Figure 2). Several previous studies found that typically a significance of >3.5σ is required to avoid contamination of noise signals (e.g., Béthermin et al. 2020; Inami et al. 2022). Therefore, we conclude that the dust continuum of MACS0308-zD1 is not detected neither from the bright clump MACS0308-zD1.1 or diffuse component.

As the dust continuum is not detected, we estimated the 3σ upper limit of the rest-frame 158 μm continuum of whole MACS0308-zD1 as <27 μJy, adopting 3 × the pixel-by-pixel RMS. Assuming a dust temperature of ∼45 K and β = 1.8 (Béthermin et al. 2020; Sommovigo et al. 2022) for a modified blackbody emission, we estimated a 3σ upper limit of the observed infrared (IR) luminosity of <1.9 × 10¹⁰ L_⊙ in the image plane. This means L_IR < 9.5 × 10⁸ L_⊙ in the source plane assuming μ = 20.

We extracted the [C ii] emission line and created the moment-0 map in an iterative way. First, we created a moment-0 map integrating over an arbitrary frequency range that clearly include the [C ii] emission line. Second, we extracted the [C ii] spectrum using the moment-0 maps as a mask. Namely, we used pixels that had >2σ signals in the moment-0 map, and extracted integrated spectra that were again used to create moment-0 map using the channels having continuous positive signals. The procedures are repeated until the channels used to create moment-0 map converged. A few iterations were enough to converge, with the final results shown in Figure 3.

Zoom In Zoom Out Reset image size

In this way, we obtained the moment-0 image and the spectrum of the [C ii] emission line integrated over the whole galaxy component. As a result, we found that [C ii] emission line of MACS0308-zD1 is detected from the bright clump MACS0308-zD1.1. We did not find any clear [C ii] emission line from the diffuse component. Using the integrated moment-0 map, we measured the total [C ii] flux density using CASA task imfit; 2D Gaussian fitting. The measured quantities are summarized in Table 1.

Table 1. Summary of the ALMA Measurements

[C ii] Component	Redshift	[C ii] FWHM	f[CII]×μa	L[CII]	${M}_{{{\rm{H}}}_{2}}$
		(km s−1)	(Jy km s−1)	(L⊙)	(M⊙)
Total	6.2078 ± 0.0002	⋯	0.62 ± 0.18	3.0 ± 1.9 × 107	9.1 ± 5.5 × 108
Narrow	6.2078 ± 0.0002	110 ± 20	0.37 ± 0.10	1.7 ± 1.1 × 107	5.1 ± 3.1 × 108
Broad	6.2058 ± 0.0005	230 ± 50	0.30 ± 0.09	1.4 ± 0.9 × 107	4.1 ± 2.5 × 108

Note. We used the single-component 2D Gaussian fits to measure integrated fluxes.

^a The measured values (except for the [C ii] flux densities) are corrected for gravitational lensing magnification of μ = 20 with 60% of systematic uncertainties quadratically added.

Download table as:  ASCIITypeset image

The [C ii] emission line of MACS0308-zD1.1 deviates from a single Gaussian profile, and shows a blueshifted excess (Figure 3). We separated these components and estimated the properties of each of the [C ii] lines. As a first step, we performed a least square fitting using a two-component Gaussian function to the integrated 1D spectrum of the [C ii] emission. This resulted in a reasonable fit with only small residuals that were consistent with random noise. We found that the best fitting results arose from the combination of a narrow and a broader Gaussian (left panel of Figure 3).

We then created moment-0 maps for both the narrow and the broader [C ii] emission by integrating frequency ranges of 263.62–263.72 GHz and 263.73–263.92 GHz, respectively. These frequency ranges were selected to separately image moment-0 maps of the two different components, and at the same time, to avoid contamination of different components.

Based on the results of the two-component Gaussian fitting, we estimated the fractions of contamination for each maps. Using the above frequency range to integrate and the results of the two-component Gaussian fitting, we estimated that ∼7% of the narrow [C ii] component contaminates the broad image, while ∼8% of the broader [C ii] component contaminates the narrow [C ii] moment-0 maps. Thus, the contamination from other components has only a minor impact on the moment-0 maps and measured fluxes for each component.

3.3.1. The Narrow [C ii] Emission Component

The moment-0 image of the narrow [C ii] component shows a 6σ detection. The emission is cospatial with the rest-frame UV emission of the bright clump: MACS0308-zD1.1 seen by the HST F160W filter (top panel of Figure 4). The significant detection from the cospatial location implies that the narrow component is associated with the UV-bright star formation activity of the MACS0308-zD1.1 and not associated with the underlying diffuse component. We extracted the narrow component of the [C ii] emission using the moment-0 map (top left panel of Figure 4). We measured a velocity FWHM of the narrow component by fitting 1D Gaussian and found the FWHM of 110 ± 20 km s⁻¹. We note that the line width lies at the lower bound of the FWHM of [C ii] emission lines thus far observed from high-redshift galaxies (e.g., Béthermin et al. 2020).

Zoom In Zoom Out Reset image size

3.3.2. The Broader [C ii] Emission Component

We derive a 3σ upper limit to the observed IR luminosity of <1.9 × 10¹⁰ L_⊙ assuming a dust temperature of 45 K. The 3σ upper limit IR luminosity in the source plane is ≲9 × 10⁸ L_⊙, applying a magnification factor of μ = 20. The strong upper limit on the IR luminosity implies an extremely low dust-obscured SFR of SFR_IR < 0.1 M_⊙ yr⁻¹ using the conversion of ${\mathrm{SFR}}_{\mathrm{IR}}=1.2\times {10}^{-10}\,{L}_{\mathrm{IR}}\,{M}_{\odot }\,{\mathrm{yr}}^{-1}\,{L}_{\odot }^{-1}$ (Madau & Dickinson 2014; Inami et al. 2022). Given the extreme brightness of MACS0308-zD1 in the rest-frame UV (SFR_UV = 3 ± 1 M_⊙ yr⁻¹ for zD1.1 and 7 ± 2 M_⊙ yr⁻¹ for diffuse component; Welch et al. 2023), we derive a very low dust-obscured fraction of the star formation activity (f_obs = SFR_IR/SFR_total) of only ≲1% to 3%, meaning that the star formation activity of MACS0308-zD1 is almost completely dust-free.

The major uncertainty in estimating the IR luminosity from a single ALMA observation is the unknown dust temperature; increasing the assumed dust temperature can produce larger IR luminosities (e.g., Faisst et al. 2020; Bakx et al. 2021; Fudamoto et al. 2023; Algera et al. 2023). We examined this effect and its impact on the dust-obscured SFR. For a high dust temperature case of 60 K, the demagnified 3σ upper limit IR luminosity is 5.5 × 10⁹ L_⊙, which yields a 3σ upper limit SFR of SFR_IR < 0.3 M_⊙ yr⁻¹. Even for an extreme dust temperature of 80 K, we find 8.0 × 10⁹ L_⊙ and 0.8 M_⊙ yr⁻¹ for demagnified 3σ upper limit of IR luminosity and dust-obscured SFR, respectively. Although such extremely high dust temperatures are not yet confirmed in this redshift range (e.g., Algera et al. 2023), even with these extreme assumptions, we still derived low obscured fractions of star formation activity of ≲10% (T_d = 60 K) to <26% (T_d = 80 K). Thus, MACS0308-zD1 has relatively little dust-obscured activity, even after accounting for rather extreme uncertainties in the dust temperature.

This low f_obs contrasts strongly with the relatively high obscured fractions (∼50%) seen for massive M_* > 10^9.5 M_⊙ galaxies at z ≳ 5 (Fudamoto et al. 2020; Schouws et al. 2022; Algera et al. 2023). Nonetheless, it is consistent with the very blue UV color of MACS0308-zD1 (UV slope of β ∼ −2.2; Welch et al. 2023). The nearly dust-free environment is indicative of the metal-poor and young environment of the star formation activity. Recent studies found large [O iii]/[C ii] luminosity ratios from observations of such metal-poor and UV-bright galaxies (i.e., L_[OIII]/L_[CII] ≳ 3; Inoue et al. 2014; Hashimoto et al. 2019; Carniani et al. 2020). These studies argue that such high large [O iii]/[C ii] luminosity ratios of high-redshift galaxies might indicate low photodissociation region covering fractions, hard ionizing fields, and/or low metallicities (e.g., Harikane et al. 2020; Sugahara et al. 2022; Witstok et al. 2022). [O iii] 88 μm emission line observations would be able to reveal such a high [O iii]/[C ii] can also be seen in the extremely star-forming condition of MACS0308-zD1.1. With the extremely high Σ_SFR, observations of MACS0308-zD1.1 may provide a sample to extend studies of star-forming conditions in high-redshift galaxies. Follow-up observations with ALMA will be important to further characterize conditions of the ISM of the star-forming region of the MACS0308-zD1.

4.2.1. SFR–[C ii] Correlation

It is highly debated whether the correlation between SFRs and L_[CII] evolves as a function of redshift. Previous studies have shown that galaxies with low SFRs or metallicities show extremely faint [C ii] emission, denoted "[C ii]-deficit" (e.g., Matthee et al. 2019; Harikane et al. 2020), that are also predicted by theoretical studies (e.g., Vallini et al. 2015; Liang et al. 2023). However, Schaerer et al. (2020) demonstrated that, based on large samples of galaxies obtained in the ALMA large program ALPINE (Béthermin et al. 2020; Faisst et al. 2020; Le Fèvre et al. 2020), there is no significant change of the observed L_[CII]–SFR relation as a function of redshift, and pointed out that the previously obtained "[C ii]-deficit" is largely affected by the uncertain estimations of the dust-obscured SFRs of high-redshift galaxies.

With the measured [C ii] luminosity of MACS0308-zD1.1 and estimated SFR (3 ± 1 M_⊙ yr⁻¹), MACS0308-zD1.1 appears consistent with the SFR–L_[CII] relation derived in the local galaxies (e.g., De Looze et al. 2014) and for high-redshift galaxies where no-evolution of the relation is found (Schaerer et al. 2020), but deviates from the "[C ii]-deficit" relation suggested in some of the previous studies (Harikane et al. 2020; Liang et al. 2023). This is also true for the total galaxy MACS0308-zD1 (i.e., the bright clump plus the diffuse component; indicated by the white star in Figure 5). Here, we assume that the SFR_IR is negligible, given the faint IR luminosity upper limits and low SFR_IRs found in Section 4.1. Thus, our observation supports a nonevolving picture for the SFR–L_[CII] correlation in the star-forming clump-scale structure inside distant galaxies (Figure 5). This result suggests that the SFR–L_[CII] holds even for low-SFR and potentially metal-poor galaxies at high redshift.

Zoom In Zoom Out Reset image size

4.2.2. Molecular Gas Mass

We found multiple velocity components in the [C ii] emission line: a narrow FWHM ∼110 km s⁻¹ [C ii] that is spatially coincident with the UV-bright star-forming region, and a broader FWHM ∼230 km s⁻¹ [C ii] that is spatially extended and potentially offset (Table 1). The offset both in centroid velocity and dispersion in the same region is difficult to explain with simple rotation (Figure 6).

Zoom In Zoom Out Reset image size

As the narrow emission is spatially coincident with the UV emission of the galaxy, we associate it with the star formation activity of the UV-bright clump, MACS0308-zD1.1. Assuming a gas mass of M_H2 /M_⊙ = 5.1 × 10⁸, and a radius of r = 27 pc, the line-of-sight velocity dispersion of a gas cloud will be $\sqrt{{\rm{G}}M/3r}=167\,\mathrm{km}\,{{\rm{s}}}^{-1}$, assuming virial equilibrium, where G is the gravitational constant. This value is consistent with the measured line width of the narrow component, suggesting that the ISM is supported by internal turbulent motion of the gas cloud.

On the other hand, the origin of the broad [C ii] emission is less clear. The broad line width and the velocity offset suggest that the broad [C ii] emission line could arise from a physically distinct region from the narrow [C ii] line. Additionally, the spatial extent of the broad [C ii] component shows that the broad [C ii] is spatially offset from the UV-bright star-forming region and the [C ii] emission associated with the UV-bright clump. Although the estimated virial velocity of the cloud is uncertain without a confident size measurement, the fact that the linewidth of the broad component appears super-virial suggests that this component of the gas cloud might not be bound. Combined with the extremely high density of star formation activity (Σ_SFR ∼ 900 M_⊙ yr⁻¹ kpc⁻²), circumstantial evidence suggests that the broad [C ii] might arise from gas outflowing from the intense star-forming region, which is shown to exist in a larger >10 kpc scale (Fujimoto et al. 2020; Ginolfi et al. 2020; Pizzati et al. 2020, 2023).

An alternative interpretation of the broad [C ii] component would be gas clumps generated as a result of the fragmentation of cold gas streams (i.e., inflow). Previous studies predicted such fragmentation of the cold gas streams in the inner halo (∼0.2 R_v), where R_v is the virial radius (e.g., Pallottini et al. 2017; Mandelker et al. 2018). After the fragmentation of cold gas streams, the masses of each clump are predicted to be order of ∼10⁵–10⁷ M_⊙. The estimated gas mass from our observation (M_H2 ∼ 5 × 10⁸ M_⊙) suggests that several clumps exist along the line of sight or the case that we are observing down the barrel of the stream, which increase observed gas mass and velocity width of the line.

More quantitative examination requires several follow-up observations, such as higher-resolution and higher-sensitivity ALMA [C ii] follow-up and JWST's rest-frame optical emission to reveal the stellar distribution down to much deeper limits. Also, to examine the extremely dense environment of the star-forming ISM, ALMA observations using higher resolution or observations of a neutral dense gas indicator such as [O i] will be needed.