PROTOPLANETARY DISKS IN THE ORION OMC1 REGION IMAGED WITH ALMA

We mapped a region around the OMC1 BN/KL outflow in the ONC (e.g., Snell et al. 1984; Bally et al. 2015a). The primary goal of the program (as proposed) is to study the outflow in molecular lines, and this will be the topic of a companion paper. Here we focus on the compact circumstellar disks within the imaged region. The map is comprised of 108 mosaicked pointings in the northwest region of the OMC1 outflow and 39 pointings in the southeast region. The fields were observed at 230 GHz frequency (Band 6), corresponding to a wavelength of 1.3 mm. Observations were taken between 2014 July 19 and 2015 April 05.

The angular resolution of the observations was approximately 1''. At the distance to Orion, ∼400 pc (e.g., Kraus et al. 2007; Menten et al. 2007; Sandstrom et al. 2007), the linear resolution is approximately 400 au. Thus, any objects with radii smaller than ∼200 au will be unresolved by these observations. ALMA Compact Array (ACA) observations were also obtained as part of this program in order to map the extended outflow well, but we do not use these data here since our focus is on the compact disks.

We observed continuum emission using four 1.875 GHz wide bands centered at 216, 228, 231, and 233 GHz. Substantial parts of the bandpass contained spectral line emission and were flagged for the continuum data reduction. Averaging continuum data over all spectral windows, the effective central frequency is 225 GHz.

One 2 GHz window was devoted to CO(2-1) at 231 GHz, and two other windows targeted C¹⁸O and SiO at 218 GHz, and SO at 216 GHz. All spectral windows had a frequency resolution of about 1 MHz, providing a velocity resolution of 1.3 km s⁻¹.

We adopted the reduced ALMA data products. The reduction included heavy flagging to remove strong spectral lines from the continuum, as well as the standard flux, bandpass, and gain calibrations. The sources used for calibration included J0607-0834, J0541-0541, J0423-013, J0725-0054, Uranus, and Ganymede. The uv data were inverted using a Briggs robust weighting parameter of 0.5, resulting in a synthesized beam with dimensions ∼14 × 08. The rms varies across the field because of strong extended emission from OMC1. Typical noise values in clean regions of the continuum image are 0.5 mJy/beam. The noise is 5 mJy km s⁻¹ in a typical spectral line channel, although in channels with strong emission from the OMC1 outflow the rms can rise to ∼20 mJy km s⁻¹.

We searched for sub-millimeter emission toward the positions of known near-IR sources, because detection of sub-millimeter continuum flux would indicate disks (as opposed to embedded spherical sources; e.g., Beckwith et al. 1990). We therefore make use of previous near-IR observations that identified cluster members (Hillenbrand & Carpenter 2000). We supplement this catalog with new data, because previous observations did not cover the complete region we mapped with ALMA. We observed the OMC1 region in Orion with GeMS at the Gemini South telescope between 2012 December 30 and 2013 February 28 (see Bally et al. 2015a). We observed in the K_s filter, as well as the narrow Fe[ii] and H₂ filters, producing images with angular resolutions of ∼006. The K_s image is shown in Figure 1.

Zoom In Zoom Out Reset image size

We used the IDL implementation of the find algorithm to identify stellar sources in the K_s-band image. We cross-referenced these with the Hillenbrand & Carpenter (2000) catalog, and kept only previously unidentified objects. We also visually inspected all sources, and discarded several spurious detections in the bright central regions of the OMC1 outflow. After this process, 40 new detections remain. We performed aperture photometry for these sources, and used previous measurements of Hillenbrand & Carpenter (2000) for stars in our field to calibrate the flux scale. Source positions and magnitudes are listed in Table 1.

Without proper motions, the cluster membership of these newly identified sources is uncertain. However, all objects are also detected in narrowband filters tuned to Fe[ii] in the H-band and H₂ in the K-band. Comparison of the color–magnitude relations of new detections and previously cataloged sources in Hillenbrand & Carpenter (2000) suggests that they trace the same stellar population. This suggests that contamination by background or foreground objects, while possible, is likely no worse than in the sample from Hillenbrand & Carpenter (2000).

The detection rate among near-IR-selected sources that lie within the ALMA mosaic is ∼8%, lower than the detection rate of ∼20% for HST-identified proplyds (all but two of which are also included in the near-IR selected sample). This discrepancy may reflect a selection bias caused by strong free–free emission in proplyds that lead to easier detection at millimeter wavelengths (despite a low intrinsic dust disk mass). Among our sample, seven proplyds have strong contributions from free–free emission to the observed millimeter-wavelength flux. In contrast, only two near-IR-selected targets (that are not also identified as proplyds) are strongly affected by free–free emission. Assuming these objects would not have been detected if not for their strong free–free emission, the discrepancy in detection rates becomes smaller.

However, some difference remains, perhaps because the near-IR sample is more likely to include some diskless stars, or foreground or background objects. Given the selection of proplyds based on their morphologies and observed emission from ionized gas, their status as cluster members is indeed easier to determine.

The distribution of disk masses is plotted as a histogram in Figure 5. The vast majority of detections have dust disk masses ≲0.03 M_J (or total gas+dust masses ≲3 M_J assuming a gas-to-dust ratio of 100). The noise in the maps means that less massive disks would probably have been undetected. Given that most cluster members were not detected in our observations, we infer that the peak of the disk dust mass distribution likely lies at masses smaller than 0.03 M_J.

Zoom In Zoom Out Reset image size

There are a number of objects with inferred dust+gas disk masses ≳0.01 M_⊙. From Table 2, we see that 12 sources have disk masses above this level, representing ∼25% of all detections, and ∼2% of all cluster members surveyed.

The intense photoionizing radiation of θ¹ Ori C and other Trapezium stars can lead to photoevaporation of disks in the ONC (e.g., Johnstone et al. 1998; Scally & Clarke 2001). One might therefore expect to see some dependence of disk mass on cluster radius, since disks in the cluster outskirts would be less affected by photoionizing radiation.

Intriguingly, the inner disk fraction may decrease at larger cluster radii (Hillenbrand et al. 1998), contrary to this expectation. Recent work shows no evidence for a decrease in inner disk fraction near to O stars in the ONC or other regions (Richert et al. 2015). However, inner disks are tighly bound to their central stars and may be less affected by photoevaporation than outer disks.

At millimeter/sub-millimeter wavelengths, where observations are sensitive to large, massive disks, previous observations suggested a truncated disk mass distribution within 0.03 pc of θ¹ Ori C, whereas a broader distribution exists at larger cluster radii (Mann et al. 2014). This dichotomy was interpreted as evidence that stars within the EUV-dominated region of the Trapezium stars experience significant disk evaporation, while disks can survive intact in the FUV-dominated region at larger cluster radii.

Kinematics suggest that the stellar population is unrelaxed, and younger than the crossing time for the entire cluster (e.g., Tobin et al. 2009). However, the crossing time within a fraction of a parsec is substantially shorter than the age of the region. It is therefore unclear whether stars that currently lie within 0.03 pc of θ¹ Ori C have been there for a substantial fraction of the lifetime of the cluster. On the other hand, it may be that the ionization front from θ¹ Ori C has turned on recently, uncovering proplyds as it advances through the cloud.

Our observations support the conclusion of Mann et al. (2014). As seen in Figure 6, disk masses within about 0.03 pc range from 0 to <10 M_J, while the range extends to >30 M_J at larger distances. This result is not statistically significant based on our data set. The Fisher exact test suggests that there is a 9% probability that a single disk mass distribution applies within 0.03 pc and at larger radial distances.

Zoom In Zoom Out Reset image size

If we include the ALMA data of Mann et al. (2014), which are sensitive to the same range of disk masses probed in this work, we can improve the significance of this result. Using both data sets, the Fisher exact test gives only a 0.7% probability of a single distribution, suggesting a difference between the disk population within and beyond ∼0.03 pc of θ¹Ori C significant at ∼3σ. Even if we exclude the two very massive disks in Mann et al. (2014) at ≳1 pc distance, the probability of a single distribution is still only ∼1%.

Previous investigations of near-IR excess emission showed the inner disk fraction for stars in Orion to be largely independent of stellar age and mass, although there are indications of a paucity of disks around very massive stars (Hillenbrand et al. 1998; Lada et al. 2000). However, this correlation has not been studied in the ONC for massive disks as traced by millimeter/sub-millimeter emission.

Observations of Taurus suggest a significant correlation between disk and stellar mass (Andrews et al. 2013). While we see no such correlation in the ONC, at present the statistical significance of this result is limited both by the small number of ALMA detections and the small number of spectroscopically determined stellar masses for extincted cluster members. To date, stellar masses have been determined spectroscopically for less than half of the sources in our ALMA-detected sample (Hillenbrand 1997; Luhman et al. 2000; Slesnick et al. 2004; Hillenbrand et al. 2013; Ingraham et al. 2014). We plot disk mass versus stellar mass for this sub-sample in Figure 7. In the near future, deeper ALMA observations will yield a larger sample of disk masses, and near-IR spectroscopy will provide stellar masses of embedded cluster members in the ONC.

Zoom In Zoom Out Reset image size

While we determined dust masses above, and converted them to total gas+dust masses using an assumed gas-to-dust ratio, one can also use spectral line data to attempt to measure gas masses directly (e.g., Miotello et al. 2014; Williams & Best 2014). Models presented in Williams & Best (2014) suggest that a disk with total mass of 0.01 M_⊙ should exhibit a ¹²CO(2-1) flux of ∼0.1–5 Jy km s⁻¹. While this spans a wide range (indeed, Miotello et al. 2014; Williams & Best 2014, suggest using line ratios for more accurate gas mass estimates), even the lower limit is higher than the detection threshold in our line maps, ∼12 mJy km s⁻¹. Thus, we would expect that most disks in our sample detected in 1.3 mm continuum emission with ALMA should be detected in CO.

The models in Williams & Best (2014) included a range of disk sizes, with radii as small as 30 au. Since ¹²CO(2-1) is generally optically thick, still smaller disks—at fixed mass—might produce lower flux. However, inferred sizes of proplyds are typically larger than the lower limit in Williams & Best (2014). The mean disk radius for a sample of 134 proplyds observed with HST is 35 au (Vicente & Alves 2005). Many proplyds, including some within 0.03 pc of θ¹Ori C, have radii greater than 100 au (Vicente & Alves 2005; Eisner et al. 2008). Thus it seems unlikely that the lack of CO detections can be explained by unusually compact disks.

The non-detection of any disk in the CO data suggests a relative lack of gas in these disks. The gas-to-dust ratio is likely much smaller than 100. This is consistent with previous results: Williams & Best (2014) inferred values of the gas-to-dust ratio of ≲20 in Taurus. In the ONC, where strong ionization from θ¹C and the other Trapezium stars can cause strong photoevaporative winds from circumstellar disks, we expect still lower gas-to-dust ratios (e.g., Throop & Bally 2005).

Detecting gas from disks in the ONC likely requires better sensitivity than the observations presented here. Assuming a gas-to-dust ratio of 10, the distribution of M_dust values shown in Figure 5 suggests that we would need sufficient sensitivity to detect ≲1 M_J of gas. Using the lower limit from Williams & Best (2014), this could represent CO(2-1) flux ≲10 mJy km s⁻¹.

One caveat is that the strong molecular emission from the Orion cloud could obscure gas emission from disks lying in the background. While most proplyds appear to reside in front of the cloud, it is possible that some gas-rich disks in the background could be missed.

There are strong detections of a few large proplyds at relatively large cluster radii (e.g., Williams et al. 2014; Bally et al. 2015b). Such large, gas-rich disks would be more likely to photoevaporate if they resided close to the Trapezium stars. Thus, the gas content of many, if not most, disks in the ONC, may be substantially lower than these examples.

The disk mass distribution shown in Figure 5 indicates that only a small percentage of ONC cluster members have masses comparable to or larger than the MMSN. Adopting 10 M_J as the minimum value of the MMSN (Weidenschilling 1977), we find 12 disks above this level. Compared to the ∼600 near-IR and optically identified cluster members surveyed, this represents a detection rate of only 2%.

Such a low detection rate may be at odds with the known existence of giant planets around other stars (e.g., Wright et al. 2011). Around FGK stars, the detection rate of planets more massive than Jupiter is ∼10%, and extrapolation to larger orbital radii suggests that up to 20% of FGK stars in the Galaxy may host giant planets (Cumming et al. 2008). The detection rate around M dwarfs is ∼5–10 times lower⁵ , ∼2% (Cumming et al. 2008; Johnson et al. 2010). The distribution of stellar masses in the ONC peaks at 0.2 M_⊙ (Hillenbrand 1997; Slesnick et al. 2004), implying most cluster members are M dwarfs.

A giant planet detection rate near 2% may therefore be be an appropriate benchmark in this case. Since ∼2% of ONC cluster members in our survey have disk masses similar to the MMSN, the planet formation potential in this young region seems close to the final outcomes seen around main-sequence M dwarfs in the field.

However, the disk masses plotted in Figure 5 assume a gas-to-dust ratio of 100. The non-detection of any sources in CO emission (as well as previous results for young stars in Taurus; Williams & Best 2014) suggests that this ratio may be substantially lower. In that case, the fraction of disks with total masses greater than 10 M_J could approach zero.

According to the approach used to derive the MMSN, this would imply that none of the systems in the ONC can form Jupiter-mass planets. However, this conclusion can be avoided if we reexamine a key assumption in the MMSN model. Specifically, that model assumes that all of the heavy elements in Jupiter and other planets had to condense out from solar-composition gas. If there is sufficient dust to build planetary cores (perhaps from an earlier condensation event), then a much smaller amount of gas is needed to build the planetary atmospheres. For example, a 30 M_⊕ solid core, massive enough to accrete a large gaseous atmosphere (e.g., Pollack et al. 1996), requires only ∼1 M_J of solar-composition gas to become a Jupiter-like planet.

It is the mass of disk solids, which determine the formation timescale of a crossover-mass core, that provides a tighter constraint on planet-formation potential. A more accurate definition for the minimum mass of dust and gas needed to form a planetary system like ours may be >0.1 M_J of dust and >1 M_J of gas. This argument suggests that 2% is the correct number for gauging giant planet formation potential, since a gas+dust mass of 10 M_J in Figure 5 corresponds to 0.1 M_J of dust.

Accounting for uncertain dust opacity (i.e., mass hidden in already-grown pebbles or planetesimals), the solid content of disks in the ONC may be still higher than inferred above. Lower dust opacities, which account for this hidden mass of solids, are physically plausible as long as they do not push the high end of the inferred disk mass distribution into a gravitationally unstable regime. Assuming most stars in our sample have masses ≲1 M_⊙, we would surmise that disks could not have total masses larger than ∼0.1 M_⊙ (100 M_J) because disks more massive would become gravitationally unstable.

Figure 5 suggests that adjusting the distribution up substantially, to account for uncertain opacity, is not physically plausible if the gas-to-dust ratio is 100, since it would produce some disks with more than 100 M_J of material. However, if the gas-to-dust ratio is only 10, then such an adjustment is allowed on physical grounds. Such an adjustment might also produce many systems with sufficient solids to form giant planet cores within gas disk lifetimes.