Unusually High Gas-to-Dust Ratios Observed in High-Redshift Quiescent Galaxies

TL;DR

Spilker et al. address whether dust continuum reliably traces molecular gas in high-redshift quiescent galaxies by combining ALMA 870 $\mu$m observations with CO(2--1) data for five $z\sim1$ systems, a broader literature comparison, and SIMBA simulations that model dust production and destruction. They find a pronounced deficit in dust emission relative to CO, implying $\delta_{\rm{GDR}}$ values from about $3\times10^2$ up to $>10^3$, consistent with SIMBA’s prediction that dust is heavily suppressed in quiescent ISM. This result challenges the use of dust continuum as a gas proxy in such galaxies and suggests CO-based methods are required for reliable gas masses, even when dust would be observationally more efficient. The work emphasizes substantial scatter and potential mass-dependent effects in $\delta_{\rm{GDR}}$ and advocates for more CO follow-up and refined modeling of dust evolution to enable robust gas-mass inferences for quiescent galaxies at high redshift.

Abstract

Tracking the cold molecular gas contents of galaxies is critical to understand the interplay between star formation and galaxy growth across cosmic time. Observations of the long-wavelength dust continuum, a proxy for the cold gas, are widely used in the high-redshift community because of their ease and efficiency. These measurements rely on the assumption of a molecular gas-to-dust mass ratio, typically taken to be GDR ~ 100 in massive, metal-rich systems. We present Atacama Large Millimeter/submillimeter Array (ALMA) observations of the 870um dust continuum in a sample of five massive quiescent galaxies at z ~ 1 with existing detections of CO(2-1). We find surprisingly weak dust emission, falling a factor of >~0.4-0.8 dex below the typical correlation between CO and continuum luminosity. We interpret this dust deficiency as evidence for unusually high GDR in these galaxies, which we calculate to range from 300 to at least 1200. Our results and other observations from the literature are generally compatible with predictions from the SIMBA cosmological simulation that dust is preferentially destroyed in quiescent galaxies. Ultimately, we conclude that the dust continuum is a highly unreliable tracer of the molecular gas in high-redshift quiescent galaxies. As a consequence we may know much less about the cold gas contents of this population than previously thought.

Figures (5)

• Figure 1: Tapered ALMA 345 GHz cutout images of our quiescent galaxy sample. Contours are drawn at $\pm$2, 3$\sigma$, where $\sigma$ varies from 20.6--22.6$\mu\rm{Jy}$/beam. Only C130284 and C132776 are detected at $>3\sigma$. Dashed white circles indicate the 2" photometric apertures.
• Figure 2: Our sample of $z\sim1$ quiescent galaxies (diamonds) is empirically faint in dust continuum emission, regardless of any calibration or conversion factors. Compared to star-forming galaxies at low (circles) or high redshifts (squares), our sample galaxies have rest-frame 850 $\mu\rm{m}$ continuum luminosities weaker by 0.4 to more than 0.8 dex (2$\sigma$ limit) given the observed CO luminosity. The dashed line and grey shaded regions show the best-fit relation and inner $\pm$1,2$\sigma$ scatter from scoville23.
• Figure 3: SIMBA exhibits a tight relationship between sSFR and $f_{\rm{H}_2}$ but often a sharp decrease in $f_{\rm{dust}}$ in quenched galaxies whitaker21b. These trends are in broad agreement with observations, including our own. Our measurements are shown as large diamonds, color-coded according to $\delta_{\rm{GDR}}$, and the stacked non-detection as a large square. Individual quiescent galaxies from the literature are shown as circles (for detections) or left-facing triangles (2$\sigma$ upper limits). Literature stacking experiments are shown similarly as open symbols. The galaxy ALMA.14 hayashi18 is the only literature source with a measured $\delta_{\rm{GDR}}$. Galaxies from SIMBA are shown as small dots, color-coded according to $\delta_{\rm{GDR}}$. The dotted lines in the left-hand panel are lines of constant gas depletion time $t_{\rm{dep}}\xspace = M_{\rm{H}_2}\xspace$/SFR; the lines in the right-hand panel show the same if one also assumes $\delta_{\rm{GDR}}\xspace=100$. Note that the $\delta_{\rm{GDR}}$ color scaling uses 2$\sigma$ limits where needed; $\delta_{\rm{GDR}}$ is more directly visualized in Figure \ref{['fig:gdrssfr']}.
• Figure 4: We find values of $\delta_{\rm{GDR}}$ much higher than the typically-assumed value of 100 in our sample of $z\sim1$ quiescent galaxies -- even the most 'dust-rich' object has $\delta_{\rm{GDR}}\xspace \approx 300$, and we find a 2$\sigma$ limit $\delta_{\rm{GDR}}\xspace \gtrsim 1200$ from our stack of undetected galaxies. Our observations, and others from the literature, are generally consistent with predictions from SIMBA. We highlight ALMA.14 hayashi18 as the only other object with a measured $\delta_{\rm{GDR}}$, and MRG-2129 whitaker21amorishita22 with a 2$\sigma$upper limit $\delta_{\rm{GDR}}\xspace \lesssim 50$. Symbols otherwise are as in Fig. \ref{['fig:fhtfdust']}; all limits are 2$\sigma$.
• Figure 5: In contrast to the mild decline of $f_{\rm{H}_2}$ with increasing $M_{\rm{star}}$ at all sSFRs, SIMBA predicts no clear trend between $f_{\rm{dust}}$ and $M_{\rm{star}}$ for quiescent galaxies. Observed galaxies scatter both above and below the expected $f_{\rm{H}_2}$, while the most massive quiescent galaxies have lower $f_{\rm{dust}}$ than predicted. Circles and errorbars / shaded regions correspond to the median and 1$\sigma$ interval of SIMBA simulated galaxies, while diamonds, squares, and small circles are observed galaxies as in previous figures. All upper limits (downward triangles) are 2$\sigma$.