Molecules with ALMA at Planet-forming Scales (MAPS) VI:
Distribution of the small organics HCN, C₂H, and H₂CO

The velocity-integrated emission, or “zeroth moment”, maps of the HCN, C₂H, and H₂CO $3-2$ and $1-0$ lines are shown in Fig. 1, the dust continuum emission at 260 GHz (with $\sim$0$\farcs$1 angular resolution) is shown in the upper row for comparison. A description of the continuum emission images can be found in Sierra & MAPS team (2021). For HCN $J=3-2$ and $J=1-0$, all hyperfine lines were included to generate the zeroth-moment maps, while for C₂H only the two brightest (and blended) lines are included in the $N=3-2$ line, and the first and brightest component in the $N=1-0$ line. For more details on the zeroth-moment map generation process, see Law & MAPS team (2021a).

The $3-2$ lines have higher fluxes than the lower-energy $1-0$ lines for all disks. HCN $J=3-2$ is strongly detected in all disks, and presents a variety of substructures, including central holes, single, and multiple rings. HCN $J=1-0$ presents similar spatial distributions to the $J=3-2$ line in AS 209, HD 163296, and MWC 480. Several rings are seen toward HD 163296, including a faint ring in the outer disk (centered at $\sim 400$ au), far beyond the dust continuum edge. Toward IM Lup, the HCN $J=1-0$ line emission is fainter and more extended than in the other disks, resulting in a lower-S/N zeroth-moment map. Toward GM Aur, the HCN $J=1-0$ line is mostly detected in the central regions of the disk, although an outer diffuse ring is seen in the radial profile (see also Law & MAPS team, 2021a).

The C₂H $N=3-2$ and $N=1-0$ lines have spatial distributions and radial extents similar to HCN in AS 209, HD 163296, and MWC 480. Toward IM Lup and GM Aur, the $N=3-2$ line is fainter than in the other disks, resulting in lower-S/N zeroth-moment maps, and the $N=1-0$ line is mostly detected in the central regions of the disks, similar to HCN.

Fewer substructures are observed for H₂CO compared to HCN and C₂H, although a few ring-like structures are seen, in particular in HD 163296. The H₂CO line emission is more extended than HCN and C₂H, and presents a central hole of varying depths in all the disks except GM Aur and MWC 480, where only a tentative dip is observed. GM Aur presents exceptionally bright emission compared to the other disks.

Normally, for molecules other than ¹²CO and ¹³CO, it is not possible to derive the height of the emission directly from the spatially resolved image cubes (e.g., Rosenfeld et al., 2013; Pinte et al., 2018a). This is because other lines are typically fainter and arise in less elevated disk layers. However, thanks to the higher inclination and relative proximity of the HD 163296 disk, and the high angular resolution and sensitivity of the MAPS data, we can distinguish emission arising in a flat disk from emission arising in the elevated layers above the disk midplane. Indeed, the HCN and C₂H $3-2$ lines observed toward HD 163296 show a clear asymmetry along the disk minor axis that cannot be explained with emission from a flat disk. Part of the asymmetry, also seen as projected ellipses that look shifted to the south with respect to the disk center, is due to the fact that we are observing the emission from the near cone that is facing us. This effect is clearly seen in the zeroth-moment map, and corresponds to the heights estimated in Law & MAPS team (2021b), i.e., $z/r\lesssim 0.1$. See Law & MAPS team (2021b) for a detailed description of the emission surfaces for HCN and C₂H in the MAPS disk sample.

To better characterize the spatial distribution of the emission we used the radial_profile function of the Python package GoFish (Teague, 2019) to produce deprojected radial intensity profiles from the zeroth-moment maps, considering the disk inclination and position angle (see Table 1). Although it is possible to take into account the vertical height of the emission to derive the radial intensity profiles (Law & MAPS team, 2021a), here we assume a flat disk but produce the deprojected profiles by averaging the emission within a $30^{\circ}$ wedge along the disk major axis. The C₂H lines were not detected at sufficiently high S/N toward IM Lup and GM Aur, so we averaged the emission azimuthally to produce the deprojected radial profiles. For additional details on the deprojected radial profiles, see Law & MAPS team (2021a). The resulting profiles for the $3-2$ (solid) and $1-0$ (dashed) lines are shown in Fig. 2. For comparison, the profiles of the 260 GHz and 90 GHz dust continuum emission are shown in the first row.

In general, the HCN and C₂H emission is more compact than that of H₂CO — their radial extents are smaller by a factor $\sim 2$. Moreover, their radial extents are similar to that of the dust continuum emission, although faint HCN and C₂H emission is observed up to $\sim 400$ au in HD 163296. IM Lup is an exception, as it presents similar radial extents in all molecular lines. One possibility to explain the lack of HCN and C₂H emission in the outermost disk regions where H₂CO is observed is that the H₂ gas density is not high enough to excite the lines. Indeed, for gas temperatures between 10 and 50 K, the HCN $3-2$ line has a critical density of $(6-10)\times 10^{6}~{}\rm{cm}^{-3}$, while the H₂CO $3-2$ line has a critical density that is 10 times lower (Shirley, 2015). However, models of these disks find that gas densities are $>10^{7}~{}\rm{cm}^{-3}$ in the outer disk midplanes, enough to excite these lines (e.g., Zhang & MAPS team, 2021). This suggests that the abundance of HCN and C₂H decreases in the outermost disk regions where H₂CO is still abundant. We note that no additional emission arises in the outermost disk regions in the lower-resolution (0$\farcs$3) images that have better S/N.

The radial profiles are consistent with profiles previously observed at lower angular resolution (0$\farcs$5) for HCN and C₂H in IM Lup, AS 209, HD 163296, and MWC 480 (Bergner et al., 2019), and for H₂CO in all sources (Pegues et al., 2020). However, our higher angular resolution observations reveal additional radial substructures, in particular for HCN and C₂H. We focus on the $3-2$ lines which show more structure than the $1-0$ lines, due to the higher signal-to-noise ratios and angular resolution (see Law & MAPS team, 2021a for additional substructure analysis).

To derive the column densities, we assume that the emission is in local thermal equilibrium (LTE) and that the emission fills the beam. The LTE assumption is reasonable, considering the high densities present in protoplanetary disks. Cataldi & MAPS team (2021) performed a non-LTE fit to the HCN lines and found similar column densities, suggesting our LTE assumption is a good approximation, at least for HCN. The line intensity is computed as

where $B_{\nu}$ is the Planck function, $\rm{T}_{\rm{ex}}$ is the excitation temperature, $\rm{T}_{\rm{bg}}$ is the background temperature, and $\tau_{\nu}$ is the line optical depth. The background temperature $\rm{T}_{\rm{bg}}$ is fixed to the maximum of the continuum brightness temperature or the cosmic microwave background (CMB) temperature (2.73 K).

At the line center, the optical depth is given by

where $A_{ul}$ is the Einstein coefficient, $\nu$ is the frequency of the line, $\sigma$ is the Gaussian line width and N_u is the upper-level column density,

Here, N_tot corresponds to the total column density of the molecule, $g_{u}$ is the upper-level degeneracy, $E_{u}$ is the upper-level energy, and $Q(\rm{T}_{\rm{ex}})$ is the partition function. The spectroscopic constants and partition functions were taken from the CDMS and JPL catalogs and are listed in Table 2.

The free parameters in the fit are the total column density N_tot, the FWHM $\Delta V=2\sqrt{2\ln 2}\sigma$, and the excitation temperature in the case of HCN and C₂H, as their lines present hyperfine structure. For H₂CO, the excitation temperature is fixed because only a single line is available. We also allow for a small offset from the source velocity $v_{lsr}$, constrained between $-$0.25 and 0.25$~{}\rm{km~{}s}^{-1}$. We explore the parameter space using the emcee package (Foreman-Mackey et al., 2013). The best fit is taken as the 50th percentile of the samples in the marginalized distributions, and the associated errors are computed from the 16th and 84th percentiles.

We follow a method similar to the one implemented in Bergner et al. (2019). The HCN and C₂H lines present hyperfine structure that is spectrally resolved. We fit the observed spectra assuming that all the hyperfine lines have the same excitation temperature. Each hyperfine line is assumed to be a Gaussian, such that the total line opacity is given by

where $\tau_{i,0}$ is the line center opacity of the ith component, $v_{i}$ is the relative velocity of each component with respect to the brightest line, and $v_{lsr}$ is the source velocity. The relative intensities $r_{i}$ as well as the relative velocities $v_{i}$ are listed in Table 2.

The width of spectral lines is dominated by thermal broadening, which depends on the gas temperature:

where $m$ is the molecule’s mass. Under the assumption of LTE, $\rm{T}_{\rm{gas}}=\rm{T}_{\rm{ex}}$. However, we find that the observed line widths for HCN and C₂H are broader than their thermal widths would imply. This is most likely produced by beam smearing, where line emission from different radii in the disk that have different velocities are combined in a single beam. In addition to beam smearing, it is possible that emission from both the back and front sides of the flared disk, which have slightly different projected velocities, are being combined in the observed spectra. The inner disk is particularly affected by these issues, which makes it difficult to fit the lines without artificially decreasing the line opacity or increasing the excitation temperature. A way to alleviate this problem is to first compute the line intensity assuming a purely thermal line width, then compute the total flux, and finally redistribute this flux in the line but now for the observed and broader line width. With this method, the fit to the lines is better, and the inferred line opacity—and hence the column density—is more realistic. The same solution was applied to fit the CN and DCN lines in the MAPS disks (Bergner & MAPS team, 2021; Cataldi & MAPS team, 2021).

The fit to the observed spectra was done in two steps. First, we fit the $1-0$ and $3-2$ lines simultaneously to obtain a better constraint on the excitation temperature. The tapered (0$\farcs$3) versions of the $3-2$ line cubes were used for this step, to match the resolution of the $1-0$ line cubes. We included priors on the excitation temperature informed by the gas temperature traced by the optically thick ¹²CO 2–1 line. Law & MAPS team (2021b) found that the HCN and C₂H emission arises from relatively flat surfaces $z/r\lesssim 0.1$ in AS 209, HD 163296, and MWC 480, while the CO isotopologues arise in more elevated disk layers. We thus restrict the rotational temperature between 10 K and the gas temperature inferred from the peak brightness temperature of the optically thick ¹²CO 2–1 line, $\rm{T}_{\rm{gas}}$(¹²CO) (Law & MAPS team, 2021b), for both HCN and C₂H. The results for the simultaneous fit to the $1-0$ and $3-2$ lines are shown in Appendix B. The resulting excitation temperature and column density profiles are shown in Fig. B.1, and the gray curves show the $\rm{T}_{\rm{gas}}$(¹²CO) profiles for each disk. HCN presents rotational temperatures between 15 and 40 K, while C₂H presents higher values of $15-60$ K. The column densities are on the order of $10^{13}-10^{15}~{}\rm{cm}^{-2}$ for both molecules. Despite the lower angular resolution of 0$\farcs$3 of the data used in these analyses, two column density rings are resolved in both HCN and C₂H for HD 163296, and a single ring or central component for the other sources.

In the second step, and to obtain a better-resolved column density profile, we fit the high angular resolution $3-2$ lines only. We used the excitation temperature found in the previous step (T_ex,0) as an initial parameter, and constrain $\rm{T}_{\rm{ex},0}-10$ K $<\rm{T}_{\rm{ex}}<\rm{T}_{\rm{ex},0}+10$ K for both HCN and C₂H. Figure 3 shows a selection of observed spectra and fit for each species and disk. Note that the radial extent is different for each disk. The lines are fit well in most of the disk, but poorly fit in the inner disk, where the lines are substantially broader, resulting in the individual hyperfine components being completely blended. For this reason, we used the ¹²CO or ¹³CO temperature as the excitation temperature in regions where the fit does not work properly. For C₂H, the $\rm{T}_{\rm{gas}}$(¹²CO) was used for all sources. For HCN, the $\rm{T}_{\rm{gas}}$(¹²CO) was used for IM Lup, GM Aur, and AS 209. For the two disks around Herbig Ae stars, however, the $\rm{T}_{\rm{gas}}$(¹³CO) was used instead, since the $\rm{T}_{\rm{gas}}$(¹²CO) produced a large discontinuity in the column density profile.

An independent retrieval of the HCN column density and excitation temperature is presented in Bergner & MAPS team (2021) and Cataldi & MAPS team (2021), using a method similar to the one presented here but with different constraints on the excitation temperature. A comparison between the HCN column densities presented in these three papers is shown in Appendix C. The resulting column density profiles and excitation temperatures are consistent with the ones presented here (see Fig. C.1).

For H₂CO, we fit a single Gaussian to the $J=3-2$ line in each radial bin. In order to constrain the H₂CO excitation temperature, a combination of lines with different upper level energies is needed. Since we only observed one line, the excitation temperature is fixed and the column density was computed for a range of excitation temperatures between 20 and 50 K so that we are not a priori assuming the vertical location of the H₂CO emitting layer. The range of temperatures is based on previous estimates. For example, Pegues et al. (2020) recently showed that the bulk H₂CO emission in four Class II disks arises in $11-40$ K gas, using multi-line H₂CO observations with similar energy ranges ($E_{u}=20-70$ K). In particular, for MWC 480 they found a low excitation temperature of $\sim 21$ K. Similarly, toward HD 163296 the disk-averaged H₂CO excitation temperature has been estimated to be $\sim 24$ K (Carney et al., 2017; Guzmán et al., 2018). Finally, toward the older ($\sim 8$ Myr; Sokal et al., 2018) TW Hya disk, Terwisscha van Scheltinga et al. (2021) recently found that H₂CO emits mostly from a $30-40$ K layer, corresponding to an elevated disk layer of $z/r\geq 0.25$.

We found that the column density is not very sensitive to $\rm{T}_{\rm{ex}}$; the difference in column density is only a factor $\sim 2$ for this range of temperatures. The observed spectra with the best-fit overlaid are shown in the bottom panels in Fig. 3. As was found for HCN and C₂H, the lines are fit well in most of the disk, but poorly fit in the inner $\sim 30$ au, especially in HD 163296 and MWC 480. This is probably due to contamination from the back side of the disk.

The resulting column density profiles for HCN, C₂H, and H₂CO are shown in Fig. 4. Two profiles are shown for H₂CO, corresponding to $\rm{T}_{\rm{ex}}$ of 20 and 50 K. The gray shaded area marks the inner regions where $\rm{T}_{\rm{gas}}$(¹²CO) or $\rm{T}_{\rm{gas}}$(¹³CO) was used as the excitation temperature in the fit. The horizontal bars show the spatial resolution of the profiles for each molecule and disk.

The HCN and C₂H column densities range from $10^{12}$ to $10^{15}~{}\rm{cm}^{-2}$, although the HCN column density may reach $\gtrsim 10^{16}~{}\rm{cm}^{-2}$ in the inner $30$ au of GM Aur, HD 163296, and MWC 480. Overall, HCN and C₂H present similar column density profiles, except for that HCN has a central component in GM Aur, HD 163296, and MWC 480, while the inner disk seems to be depleted of C₂H in all sources but toward GM Aur. GM Aur presents a central dust cavity (see Fig. 2), which could result in enhanced C₂H abundance in the inner disk. However, there is evidence that the total H₂ gas is also depleted inside the dust cavity (Dutrey et al., 2008; Huang et al., 2020; Bosman & MAPS team, 2021b). Another difference between HCN and C₂H is in the outer $>150$ au disk. The HCN column density seems to reach a plateau of a few times $10^{12}~{}\rm{cm}^{-2}$, which is not present for C₂H.

The H₂CO column density profiles are consistent with the ones derived by Pegues et al. (2020) for all sources using different H₂CO transitions and at lower angular resolution. H₂CO is present at higher column density at larger disk radii compared to HCN and C₂H. Although the spatial resolution for the H₂CO column density profiles is lower than for the smaller organics, a few distinct rings can be seen in the profiles. GM Aur and AS 209 present less structure, while at least two rings or local maxima are seen for HD 163296 and MWC 480. The H₂CO column density varies between the sources, ranging between $10^{12}$ and $10^{14}~{}\rm{cm}^{-2}$ in the inner 200 au. The largest column densities are seen toward GM Aur and IM Lup, which together with AS 209 are the coldest disks in the sample if we consider their stellar mass and the large disk masses that have been estimated (Cleeves et al., 2016; McClure et al., 2016).

Figure 5 shows the excitation temperature profiles inferred for HCN and C₂H from fitting the higher angular resolution $3-2$ lines only, informed by the $\rm{T}_{\rm{ex}}$ obtained from the simultaneous fit to the $1-0$ and $3-2$ lines. The $\rm{T}_{\rm{gas}}$(¹²CO) and $\rm{T}_{\rm{gas}}$(¹³CO) profiles are shown in gray. At least for 3/5 sources, the C₂H excitation temperatures are higher than those derived for HCN. HCN temperatures range from 20 to 30 K, while C₂H temperatures range from 20 to 60 K. The C₂H excitation temperatures in the outer disk regions ($>100$ au) of AS 209, HD 163296, and MWC 480 are consistent with the temperatures derived by Bergner et al. (2019) at lower angular resolution. Excitation temperatures for HCN do not vary substantially within the disk sample. However, the inferred $\rm{T}_{\rm{ex}}$ values for C₂H are higher for HD 163296 and MWC 480 than for the other three disks. This is consistent with disks around Herbig Ae stars being warmer than disks around lower-mass T Tauri stars, and suggest that C₂H arises in warmer elevated disk layers compared to HCN.

The inferred excitation temperatures are low ($20-30$ K) for both HCN and C₂H in IM Lup and GM Aur, and do not vary substantially across the disks. Toward AS 209, the C₂H excitation temperature profile shows a ring-like distribution similar to the column density ring seen in Fig. 4, reaching a peak $\rm{T}_{\rm{ex}}\sim 40$ K in one of the dust gaps. HCN, on the other hand, shows a decreasing $\rm{T}_{\rm{ex}}$ profile with radius. Toward HD 163296, the C₂H and HCN excitation temperatures show two rings, similar to the distribution of the column densities, although the rings are more pronounced for C₂H than for HCN. Toward MWC 480, the $\rm{T}_{\rm{ex}}$ of HCN and C₂H decrease with radius in a smoother profile compared to HD 163296. Finally, in AS 209 and HD 163296, the $\rm{T}_{\rm{ex}}$ peaks of C₂H spatially coincide with the column density peaks. We do not find the same correlation for the other sources or for HCN.

Chemical models predict HCN and C₂H emission arises in elevated disk layers where UV photons drive the gas to temperatures of $40-50$ K and dominate the chemistry (Bergin et al., 2016; Cleeves et al., 2018). However, a recent analysis of observations toward a sample of 16 disks at lower ($>$0$\farcs$5) angular resolution resulted in low excitation temperatures of $5-10$ K and $10-40$ K for HCN and C₂H, respectively (Bergner et al., 2019). Bergner et al. (2019) suggested that the origin of the lower-than-expected excitation temperatures could be due to subthermal emission or beam dilution due to unresolved chemical substructure. The higher angular resolution MAPS data reveals these substructures across a sample of disks for the first time. In addition, the $1-0$ line observations now allow us to better constrain the molecular excitation temperature. While we do find HCN excitation temperatures that are higher than inferred by Bergner et al. (2019) toward these disks, they remain relatively low ($<30$ K). For C₂H, on the other hand, we find higher excitation temperatures that are more consistent with the gas temperature at the expected height of the emission, in particular for the warmer disks HD 163296 and MWC 480, which are also disks that present the most substructure in their emission.

The inferred excitation temperatures for HD 163296 are consistent with the brightness temperature distributions of the HCN and C₂H $3-2$ lines found by Law & MAPS team (2021b). These brightness temperatures provide a lower limit to the gas temperature because these lines are partly optically thin. These results, combined with the emission heights estimated by Law & MAPS team (2021b) of $z/r\lesssim 0.1$, suggest that the HCN emission arises in vertically deeper and colder layers than C₂H.

For IM Lup and GM Aur, the $\rm{T}_{\rm{ex}}$ of C₂H and HCN are very similar, which suggests that their emission arises from similar disk layers. However, given our uncertainties, we cannot rule out a vertical stratification. Indeed, detailed chemical models of IM Lup suggest that HCN arises in deeper layers compared to C₂H. (Cleeves et al., 2018). Finally, it is worth mentioning that the C₂H excitation temperature is lower for the colder disks around T Tauri stars, which show fewer substructures compared to the disks around Herbig Ae stars. This could be due to intrinsic lower gas temperatures or to small-scale substructure not resolved in our observations.

CO is widely used to trace the gas in disks because H₂ does not have strong transitions in the low temperature range that characterizes disks. Thus, we compare the column density profiles of the small organics with that of CO. Figure 6 shows the normalized column density profiles of HCN, C₂H, and H₂CO, as well as the CO column densities derived from the C¹⁸O 2–1 line by Zhang & MAPS team (2021). In general, the HCN and CO column densities tend to be more centrally peaked than that of C₂H, suggestive of an active warm cyanide chemistry in the inner disk. In the outer disk, HCN and C₂H show similar radial distributions up to $\sim 150$ au, where the C₂H column density starts to drop faster than HCN and CO. There is notably good spatial correlation between the profiles toward HD 163296 between $50<r<150$ au.

In general, the HCN and CO column densities correlate well, except in AS 209, where significant differences are observed. In particular, an anticorrelation is observed between CO and both HCN and C₂H. Alarcón & MAPS team (2021) used chemical models to investigate the gas structure in AS 209, and showed that this anticorrelation can be explained by CO chemical processing, which produces a decrease in the CO abundance and an elevated C/O ratio ($>2$) in the gas that boosts the formation of C₂H. Bosman & MAPS team (2021a) also found that elevated C/O ratios ($\sim 2$) are needed to explain the observed C₂H column densities for AS 209, MWC 480, and HD 163296. C₂H is therefore very sensitive to the gas-phase C/O abundance ratio in the disk atmosphere (Bergin et al., 2016; Cleeves et al., 2018; Miotello et al., 2019).

If the formation of HCN and C₂H is the result of a UV-dominated chemistry, we might expect to see enhanced column densities inside the dust gaps. The depletion of millimeter dust at these locations should allow the deeper penetration of UV photons, and thus efficient formation of C₂H and HCN. However, the penetration of UV photons will also depend on the geometry of the source (for example, the flaring of the disk). It is nevertheless worth comparing the structures observed in mm dust and molecular emission. The position of the dust continuum gaps, as measured by Law & MAPS team (2021a ; but see references therein), are marked by the gray dashed vertical lines in Fig. 6. We only include the gaps that are resolved at $\sim 0\farcs 1$ resolution. However, several other gaps are seen in higher angular resolution observations of the dust continuum emission (Huang et al., 2018, 2020). The peak of the C₂H column density seen toward IM Lup, AS 209, MWC 480, and the first peak seen toward HD 163296, coincides with the location of the innermost dust gaps, although the C₂H structures are generally broad compared to the dust gaps. The second column density peak seen at $\sim 110$ au toward HD 163296 does not spatially coincide with any of the dust gaps, revealing that there is not a universal connection between dust depletion and C₂H or HCN formation. The weak correlation between dust structures and HCN and C₂H chemical substructures could be due to the HCN and C₂H emission arising in disk elevated layers, which are not always impacted by the dust millimeter substructure that trace the midplane. Moreover, recent studies have shown that the H₂ density seems to be much less altered than the dust inside the dust gaps, and chemical effects play a more important role in setting molecular abundances (e.g., Teague et al., 2018b; Alarcón & MAPS team, 2021; Zhang & MAPS team, 2021).

If photochemistry drives the formation of HCN and C₂H, we could also expect to see enhanced emission in the outer disk where the gas density starts to drop allowing UV photons to penetrate the disk. The flaring of the outer disk could also help to capture more UV photons from the star. Although the HCN and C₂H emission is compact compared to H₂CO, some faint emission is indeed detected at large radii, far beyond the dust millimeter edge in IM Lup, GM Aur, and HD 163296. As an example, Fig. 7 shows radial structures seen in the tapered images for HD 163296, the disk that presents the most substructure in molecular line emission. A power-law stretch was used in the y-axis, to enhance the low-S/N structure seen in the outer disk. A faint ring that is seen in both HCN and C₂H near 400 au is also seen in CN (Bergner & MAPS team, 2021). The ring is not clearly seen in H₂CO in the radial profile, but a ring structure is visible at large radii in the zeroth-moment map, in particular along the disk minor axis (see Fig. 1). No millimeter dust grains are present at these larger radii, but small dust grains are present, as suggested by scattered light images (e.g., Monnier et al., 2017; Avenhaus et al., 2018; Rich et al., 2020). A small dust grain population is also present in the disk elevated layers where HCN and C₂H emission seems to arise. This small grain population is likely regulating the HCN and C₂H chemistry. An increased amount of UV photons will also contribute to enhanced HCN destruction in the outer disk. Indeed, Bergner & MAPS team (2021) find an increasing CN/HCN profile with radius, indicative of increased HCN photodissociation relative to CN in the lower-density outer disk (see also Guzmán et al., 2015). All this suggests that an active photochemistry is at play in the outer disk beyond the millimeter dust emission around HD 163296. Even so, as mentioned in the previous section, the C/O ratio in the gas may still be more important than UV exposure for the formation of HCN and C₂H in the outer disk.

Several efforts have been made in recent years to find planets forming in protoplanetary disks. The presence of several planets has been proposed to explain the gaps seen in the dust continuum emission and the deviation from Keplerian rotation in CO observed in a few disks, including HD 163296 (Teague et al., 2018a; Pinte et al., 2018b, 2020; Teague & MAPS team, 2021) and MWC 480 (Teague & MAPS team, 2021). While dust depletion in the midplane can be inferred from the millimeter dust continuum observations, it has been harder to infer the gaps are also depleted in gas, since all millimeter lines, including CO isotopologues, may reflect chemistry variations rather than true gas density variations. Indeed, chemistry has a nonlinear response to gas density. See Rab et al. (2020) for detailed thermochemical modeling of HD 163296 in particular. Here, we compare the locations of the planets proposed for HD 163296 and MWC 480 with the structures observed in HCN, C₂H, and H₂CO.

In Fig. 7, we compare the emission of HCN, C₂H, and H₂CO with the locations of the putative planets in HD 163296 (dashed vertical lines). Besides the two bright rings seen in the inner 150 au, two fainter rings are resolved in the outer disk for both HCN and C₂H (see also Law & MAPS team, 2021a). The planet predicted at 83 au by Teague et al. (2018a) and Isella et al. (2016) is located inside a dust millimeter gap, which coincides with the gap seen in HCN and C₂H. This gap is also recovered in the column density profiles (Fig. 4). The correlation between the planet location and a decrease in the C₂H column density was previously hinted by Bergner et al. (2019). Although the gaps seen for HCN and C₂H are much shallower than what is observed for the millimeter dust, these observations suggest the gap is also depleted of gas.

Toward MWC 480, Teague & MAPS team (2021) proposed that a planet near 245 au could explain the observed CO deviations from Keplerian rotation. However, HCN and C₂H are not detected at these large radii. At smaller radii ($<200$ au), no perturbations could be confirmed in the CO gas kinematics. However, a gap is seen in HCN at around 70 au, which roughly corresponds to the location of a millimeter dust gap. The gap is not seen in C₂H. Hydrodynamical simulations suggest that a $\sim 2.3$ M_J planet located at an orbital radius of $\sim 78$ au could produce the observed dust gap (Liu et al., 2019). If there is a corresponding gas gap, then this suggests that HCN may be more sensitive to gas density perturbations than C₂H, and could potentially be used as a diagnostic to identify massive planets in protoplanetary disks.

H₂CO could be a better tracer of the colder gas in the outer disk than HCN and C₂H, which seem to trace warmer elevated layers in the disks. The use of H₂CO as a tracer of cold gas depends on its dominant formation pathway and on the efficiency of desorption mechanisms. H₂CO can form from both gas phase and grain surface chemistry. The dominant reaction in the gas-phase at low ($\lesssim 100$ K) temperatures is the neutral-neutral reaction

The grain surface chemistry pathway involves the successive hydrogenation of CO ices (Hidaka et al., 2004; Watanabe et al., 2004; Fuchs et al., 2009)

This pathway can only be efficient in regions where CO has started to freeze out onto dust grains, i.e., beyond the CO snowline and below and around the CO snow surface, depending on whether the CO hydrogenation is faster than the CO desorption. The light blue vertical lines in the bottom panels of Fig. 6 mark the estimated location of the CO snowline in each disk as measured by Zhang & MAPS team (2021). CO snowlines are located at smaller radii in the colder disks around T Tauri stars, compared to the warmer HD 163296 and MWC 480 disks, where the CO snowline is well-resolved by our observations. The first H₂CO ring seen toward HD 163296 and MWC 480 spatially coincides with the estimated location of the CO snowline, suggesting that CO hydrogenation is an important (if not dominant) formation pathway of H₂CO in disks.

In the warm inner disk, H₂CO is expected to form mainly in the gas phase, since CO cannot freeze out onto dust grains. The inner regions of the MAPS disks all show central H₂CO depressions, although they are less pronounced for GM Aur and AS 209. This does not mean that H₂CO is depleted in the inner disk; it could purely be a result of the observed line being more sensitive to cold gas. Indeed, higher-energy lines have shown centrally peaked emission profiles suggesting a warmer H₂CO component is present in the inner disk (Loomis et al., 2015; Öberg et al., 2017).

In the outer disk, both gas-phase and CO hydrogenation could contribute to the overall H₂CO emission. In Section 3.2.3 we showed that at least ${\sim}$50% of disks may show some spatial links between H₂CO substructures and the outer edge of the continuum disk. We also observe spatial links between the H₂CO column density profiles and the outer edge of the continuum disk, which are marked by the gray vertical lines in Fig. 6. In particular, the second ring or bump component seen in the H₂CO column density profiles of HD 163296 and MWC 480 spatially coincides with the millimeter dust edge. A similar coincidence is seen toward IM Lup and GM Aur, where a shoulder-type structure is seen near the continuum edge. This suggests that the release of ices at large distances becomes more efficient, either through thermal desorption due to thermal inversion (Cleeves, 2016), or by nonthermal desorption due to the penetration of UV photons (e.g., Podio et al., 2020; van’t Hoff et al., 2020). Another possibility is that other ice fragments are released into the gas phase in the outer disk, which then boost the gas-phase chemistry enhancing the abundance of H₂CO.

The H₂CO traced by our observations could also form predominantly in the gas phase through reaction (6), with little contribution from the CO hydrogenation pathway (7). This could happen if nonthermal desorption of H₂CO ices is not efficient and H₂CO remains frozen onto dust grains. In this scenario, the emission would arise in more elevated and warmer disk layers. Although chemical models have shown that pure gas-phase chemistry cannot account for the observed H₂CO abundances in the outer disk around DM Tau (Loomis et al., 2015), observations of edge-on Class I disks have shown that H₂CO emission arises in the warmer surface layers, as revealed by X-shaped patterns in their zeroth-moment maps (Podio et al., 2020; van’t Hoff et al., 2020). These observations, however, targeted the H₂CO $3_{12}-2_{11}$ line which has a relatively large upper-level energy of $\sim 33$ K, and is therefore more sensitive to the warmer elevated disk layers. Lower-energy lines could trace colder gas present at intermediate disk layers. In order to confirm the origin of the H₂CO emission, multi-line observations are needed (e.g., Pegues et al., 2020). The results toward Class II disks found by Pegues et al. (2020) suggest a possible trend between the location of the H₂CO emitting layer and the age of the disk, or perhaps a difference between disks around T Tauri and Herbig stars. Future multi-line observations toward a larger sample of disks with a consistent set of lines may help to elucidate whether this trend is real.

We can estimate the amount of HCN, C₂H, and H₂CO present in the inner disk regions. The gas masses within $50$ au are listed in Table 3. Because of the uncertainties of the fit in the inner disk, we extrapolated the column density profiles inward, assuming a constant column density starting from the radius where we obtain a good fit (see Fig. 4). This means that the HCN masses estimated for GM Aur, HD 163296, and MWC 480, which tend to have centrally peaked N profiles, could be larger than what we have estimated here. The total amount of HCN, C₂H, and H₂CO is, however, much larger because a large fraction of the organic reservoir is expected to reside in icy mantles. Indeed, as discussed before, the emission of these small organics arises in disk layers with temperatures of $20-60$ K, which are lower than the freeze-out temperature of these molecules. Assuming a conservative ice-to-gas ratio of 1000 (these ratios could be larger, as discussed in Öberg & MAPS team, 2021), we find a large amount ($10^{23}-10^{26}$ g) of gas phase plus ice mantle HCN, C₂H, and H₂CO in the inner disks. Interestingly, the relative amounts of these organics varies between the disks. For example, H₂CO is $\sim 5$ times more abundant than HCN and C₂H in IM Lup, while H₂CO is less abundant than HCN and C₂H in the rest of the disks. HCN is more abundant (by a factor $\sim 10-50$) than both C₂H and H₂CO in GM Aur, HD 163296, and MWC 480, and AS 209 has a slightly larger C₂H mass than HCN.

We can also compare the amount of organics with respect to water ice in these disks. The total water mass inside $50$ au was estimated from the models presented in Zhang & MAPS team (2021), and are listed in Table 1 of Öberg & MAPS team (2021). The abundance of total gas and ice HCN, C₂H, and H₂CO relative to water are listed in the right columns of Table 3. The fractions range from 0.5% to 1.1% and 0.02% to 1.8% for HCN and C₂H, respectively, not taking into account IM Lup, which has much lower organic fractions of $<0.001\%$. The fractions for H₂CO are lower, ranging from 0.001% to 0.08%. Typical abundances with respect to water in cometary ices range from 0.1% to 0.6% for HCN and from $0.1-1\%$ for H₂CO (Mumma & Charnley, 2011), which are consistent with our organic fraction estimates.

The MAPS disks are therefore rich in organic species, suggesting that future comets formed in these disks could efficiently deliver water and other key organics to rocky planets forming in the inner disk.