F

Report of the Panel on the Interstellar Medium and Star and Planet Formation

INTRODUCTION

Stars and planets form from gas and dust initially present in the interstellar medium (ISM). The ISM is complex, highly structured, and dynamic. It consists of gas at temperatures ranging from ~10⁷ K or more down to 10 K or lower, with densities ranging over many orders of magnitude, threaded by magnetic fields. Stellar energy input in the form of radiation, winds, and explosions shape the ISM, heating, ionizing, and dissociating atomic and molecular gas, driving the cycling of material through the different phases, dispersing and accumulating dense gas, and even expelling gas into the circumgalactic medium (CGM). Stellar winds and explosions enrich the ISM in heavy elements and dust essential to the formation of planets and life, at the same time as infall from the CGM adds gas with generally lower heavy element abundances. In this complex, turbulent, and dynamic environment, clouds of dense molecular gas are produced that are the sites of star formation. Within these molecular clouds, dense cores form that eventually gravitationally collapse and often fragment further to form stars with a wide range of masses. At least some of the angular momentum of the parent cores is retained during gravitational collapse, resulting in the formation of rotating circumstellar disks where planets form. The scope of this report of the Panel on the Interstellar Medium and Star and Planet Formation spans these widely disparate phases and structures over huge ranges of scale that are nevertheless inextricably linked.

Major advances in our understanding of the ISM and star and planet formation have been made over the past decade, encompassing the star-forming activity in nearby galaxies, the structure and properties of gas and dust in the local ISM, the fragmentation and collapse of dense gas to form stars and circumstellar disks, and the properties of those disks that seed the formation of planetary systems. To provide context for the recommendations in the coming decade, here the panel briefly outlines some of the significant progress in these fields.

Studies of the galactic ISM have provided much more detailed characterizations of structures in both the dense and the diffuse medium over the past decade. Far-infrared imaging and spectroscopy from Herschel emphasized that filamentary structures are ubiquitous in dense molecular clouds (Figure F.1, left), while observations from the Planck spacecraft and the Galactic Arecibo L-Band Feed Array HI Survey (GALFA)

21 cm survey at the Arecibo Observatory demonstrated that filamentary structure is also prevalent in the diffuse medium. The polarized sub-mm emission measured by Planck (Figure F.1, middle) showed that dust grains are efficiently aligned by magnetic fields, which exhibit coherent structure over large scales and provide important tests for models of ISM dynamics.

Significant progress was also made in characterizing the spatial distribution of gas and dust in the local ISM. Velocity-resolved observations of [CII] 158 μm emission with both the Herschel spacecraft and the Stratospheric Observatory for Infrared Astronomy (SOFIA) have helped quantify the significant fraction of molecular gas that is not traced by CO emission (“CO-dark” gas). Large-scale multiband stellar surveys from the Sloan Digital Sky Survey (SDSS) and the Panoramic Survey Telescope and Rapid Response System (Pan-STARRS), coupled with state-of-the-art statistical methods, have helped develop novel 3D models for the spatial distribution of dust and regional variations in the reddening curve. In the near term, distances and stellar characterizations from future Gaia data releases will enhance the power of these methods.

On larger scales, the distribution of molecular gas in nearby galaxies has been mapped in exquisite detail using the Atacama Large Millimeter/submillimeter Array (ALMA) (Figure F.1, right), showing how the properties of molecular clouds depend on the local environment. In particular, comparisons of ALMA CO maps to tracers of massive star formation on matched scales have revealed systematic variations in the star formation efficiency of molecular clouds, as well as the breakdown of star formation laws on cloud scales where the life cycles of star-forming clouds set by feedback or dynamics dominate. Observations with ALMA, Herschel, and optical integral field units (IFUs) have detailed the launching of outflows driven by feedback from massive star formation, revealing the cycling of ISM material into the CGM and the intergalactic medium (IGM) of galaxies.

Shifting to local studies of Milky Way molecular clouds, there have been substantial developments in our understanding of the many relevant physical scales and processes associated with star formation. Surveys of local star-forming regions with Spitzer provided the first complete censuses of their low-mass protostars and analyzed their spatial distributions relative to the cloud structures. Herschel observations provided the far-infrared data needed to determine protostellar luminosities and identify the youngest objects for further study (Figure F.2, left). Resolved submillimeter to centimeter to cm continuum and spectral line imaging with ALMA and the Karl G. Jansky Very Large Array (VLA) detected multiple protostars interacting with their disks while still embedded in their natal envelopes (Figure F.2, middle). Measurements of protostellar disk rotation and the kinematics of infalling envelopes from ALMA spectral line studies at high spatial resolution enabled measurements of the central masses, critical data for testing theories of protostar formation.

Parallaxes from Gaia and very long baseline interferometry (VLBI) have substantially improved estimates of young star luminosities and thereby ages, an essential aspect of characterizing the star formation histories of molecular clouds as well as for establishing protoplanetary disk lifetimes. Additional measurements of precise kinematics of young stellar populations from Gaia have increased the number of known nearby clusters and moving groups, facilitating more robust investigations of their initial mass functions, chemical homogeneity, and multiplicity statistics as a function of age. Photometric monitoring programs, at modest cadence from the ground (e.g., the Palomar Transient Factory [PTF]) and high cadence from space (e.g., the Microvariability and Oscillations of Stars telescope, Kepler), have revealed diverse protostar and pre-main sequence variability that signals ubiquitous, complex, and variable accretion phenomena.

Complementing this observational progress, computational modeling of star formation has advanced considerably over the past decade. Three-dimensional, adaptive resolution simulations including radiative transfer, magnetohydrodynamics (MHD), and other physics (accretion, outflows) are now routine (if still expensive). Radiative transfer post-processing has also enabled more direct comparisons between the models and data. In many instances, these improvements have highlighted shortcomings in current models of important physical processes, necessitating ongoing study.

On still smaller scales, new insights have emerged from high-resolution imaging of circumstellar disks (especially with ALMA), including constraints on the physical conditions, dynamics, and chemical abundances in the environments where planets form and grow, preliminary looks at demographic relationships between disk and host parameters, and the signatures of small-scale substructures in the spatial distributions of disk material. These latter features, in the forms of gaps, rings, spirals, and arcs on ~few astronomical unit (AU) scales, appear to be pervasive in early high-resolution samples, with major implications for models of planet formation and planet-disk dynamical interactions (Figure F.2, right). Many additional insights into planet formation have become available, with a particular highlight being the comprehensive survey of short-period exoplanets (primarily from Kepler) and their initial characterization. Unique information has also been gleaned on the nature of Jupiter’s core from the Juno mission, while New Horizons has provided the first exploration of a possibly primordial body in the Kuiper Belt.

On the theoretical side, novel ideas have emerged to better explain the transport of material and formation of solid bodies in disks during the planet formation epoch, including wind-driven accretion and accelerated planetary core growth via pebble accretion. Computational advances now allow simulations of these and other key processes from first principles, which can be compared against astrophysical data or in situ measurements. Laboratory experiments, including microgravity studies of dust particle collisions, chemical measurements of ice mantle formation and sublimation, and analyses of meteorites, provide critical inputs for both theoretical models and the interpretation of astronomical data.

Building on these many achievements over the past decade, this appendix discusses the opportunities for making further progress in characterizing and understanding the state of the ISM in the Milky Way and nearby galaxies, star formation, and planet formation, greatly assisted and informed by the contributions of more than 150 science white papers from the broader community.

Below, four key science questions and one discovery area identified by this panel are discussed and are listed for convenience near the end of this appendix in Table F.1. A summary of the capabilities, facilities, and future needs to answer these questions appears in Table F.2.

F-Q1. HOW DO STAR-FORMING STRUCTURES ARISE FROM, AND INTERACT WITH, THE DIFFUSE INTERSTELLAR MEDIUM?

While star formation occurs in molecular clouds, there remains considerable uncertainty on the mechanisms that control the formation and evolution of those clouds in the Milky Way and other galaxies. Interstellar gas must cool and condense over many orders of magnitude to reach the densities and temperatures necessary for star formation. The transitions in density, temperature, and chemical state are driven by turbulence, stellar feedback, nonequilibrium chemistry, and dust evolution. Disentangling these processes and characterizing their dependence on galactic environment in the Milky Way and beyond are critical challenges for the next decade.

Over the coming decade, observations will be capable of delineating the 3D structure of the gas, dust, and magnetic field, providing a new and dramatically detailed view of the dynamic ISM: “ground truth” to test theoretical models. Observations of other galaxies have less spatial detail, but will allow us to probe the star-forming ISM under different conditions, and over a range of metallicities. Milky Way observations will provide complementary data, including 21 cm emission, CO emission, dust extinction, interstellar absorption lines, starlight polarization, and polarized submillimeter emission from dust. These huge data sets will be a challenging opportunity for big data methodologies—for example, automated analysis pipelines run “locally” at the site of data acquisition that can distill multiterabyte datacubes into manageable data products, obviating the need for massive transfers and immediate human intervention.

F-Q1a. What Sets the Density, Temperature, and Magnetic Structure of the Diffuse ISM, Enabling the Formation of Molecular Clouds?

To understand the origin of star-forming clouds, we first must understand the state of diffuse gas in the Milky Way and other galaxies. The neutral gas can be characterized using emission and absorption in the 21 cm line to derive the neutral hydrogen (HI) spin temperature, revealing the cold neutral medium (CNM) and warm neutral medium (WNM) mass fractions, and the temperature in each phase. To constrain how the CNM/WNM fraction depends on galactic environment, HI absorption measurements in nearby galaxies are required. Imaging cold neutral gas structures in emission is also critical to uncover the cold gas dynamics, organization, and connection to star formation.

Our view of the Milky Way’s ISM has been hampered by seeing it in projection. As a result, we have little information about volume density; observed quantities are line-of-sight averages; constraints on kinematics are highly incomplete; and studies of the 3D magnetic field are compromised. We are on the verge of a revolution, enabled by big data methods to exploit the stellar distances provided by Gaia, to construct a backbone for a 3D view of the ISM. Photometric and spectroscopic surveys plus Gaia distances already have been used to create spatial maps of dust extinction. High-resolution optical spectroscopy of absorption lines (NaI, KI, CaII, CH, CN, C₂) toward stars of known distances would allow the gas to also be dissected in 3D; 21 cm emission components can then be associated with optical absorption lines at the same velocity. Large surveys of stellar polarization (e.g., Polar-Areas Stellar-Imaging in Polarisation High-Accuracy Experiment [PASIPHAE]) and filamentary HI features can outline the spatial structure of the magnetic field. Early results from these approaches are spectacular. For example, the 3D structure of the Orion A molecular cloud has been found to differ greatly from its projected structure. The first generation of 3D dust maps also show spatial variations of the extinction curve across the MW, revealing regional evolution of dust properties.

F-Q1b. How Do Molecular Clouds Form from, and Interact with, Their Environment?

Studying the formation of molecular clouds requires observations of regions where dramatic changes in temperature, density, and chemistry are occurring. In particular, it is critical to track the “CO-dark” gas, where CO is underabundant. Ultraviolet (UV) observations from space offer a uniquely powerful window into this process, by directly observing the absorption lines from HI, H₂, CO, CII, and many other molecules, atoms, and ions. With high spectral resolution, the dynamics and physical state of the gas can be characterized. Building on the existing framework of stellar distances from Gaia, UV spectroscopy toward stars with known distances can provide a 3D view of Milky Way molecular cloud formation.

While UV spectroscopy can provide detailed information, the reach of such observations is limited to conditions where individual stars can be resolved (e.g., the Local Group with current facilities). The fine structure transitions of various atoms and ions provide means to trace gas in a variety of phases both in the Milky Way and nearby galaxies. A benefit of the far-IR lines, such as [CII] 158 μm and [OI] 63 μm (Figure F.3) is that they are relatively easy to excite and thus easily detectable even out to high redshifts with ALMA. A challenge is that they represent the contributions from gas in many different physical states. Enabling the full diagnostic potential of these lines requires high-velocity resolution far-infrared (IR) observations coupled with matched resolution diagnostics of the ionized, atomic, and molecular gas phases (e.g., Hα, HI, and CO) to separate the emission arising from various phases. First, associating components of [CII] with HI can probe the thermal pressure in the diffuse ISM, a critical ingredient that mediates the phase transition between CNM and WNM. Second, the line intensities provide information on the heating of the gas, controlled by the photoelectric effect from small dust grains. Last, any components not associated with Hα, HI, or CO may arise from gas that is in transition between phases, most importantly the “CO-dark” H₂, of critical importance for measuring total molecular gas content especially at low metallicity.

Dust is critical to all stages of the transition between diffuse to molecular gas, through its roles in thermal balance, chemistry, and shielding. Our current understanding shows that the size distribution, composition, and overall abundance of dust relative to gas (the dust-to-gas ratio) change within the ISM. As these properties vary, the efficacy of the dust in its ISM roles is modified. Over the next decade, we need to develop a comprehensive picture of how dust evolves in the ISM, both in the Milky Way and in extragalactic environments. This will require observational constraints as well as theoretical models and simulations that track the dust life cycle. High-resolution UV spectroscopy can provide measurements of the depletion of heavy elements in the Milky Way and nearby galaxies; low-resolution UV spectroscopy

through mid-IR spectroscopy can trace the extinction curve; and mid-IR spectroscopy with JWST will better characterize the silicate absorption bands and other features. X rays can provide detailed information on dust mineralogy and dust-to-gas ratio, via measurements of absorption edges and scattering halos. Last, laboratory studies of candidate materials at all wavelengths are vital.

F-Q1c. How Does Injection of Energy, Momentum, and Metals from Stars (“Stellar Feedback”) Drive the Circulation of Matter Between Phases of the ISM and CGM?

The ISM is largely driven by energy from stars. Over the next decade, it is critical to perform increasingly realistic MHD simulations that include transitions between ionized, atomic, and molecular gas phases, dust evolution, cosmic rays, and the inflow and outflow to the CGM (an example is shown in Figure F.1). Simulations on many different scales are needed, from galaxy scales (cell sizes of tens of pc) to molecular cloud scale (where the cell size may be below 0.1 pc). Including stellar feedback in the form of injection of energy and momentum as well as metal enrichment by nucleosynthetic products is required.

Enabling Capabilities

To characterize the neutral gas in nearby galaxies, HI 21 cm absorption and emission observations on cloud scales ~1” (~100 pc for D <= 20 Mpc) with improved sensitivity are needed (this requires ~10× the

collecting area of the VLA). To take full advantage of the stellar parallaxes from Gaia (or elsewhere) to map the 3D structure of the Milky Way’s ISM, high-resolution (R >60,000) optical spectroscopy, with sensitivity to obtain spectra for all V <15 O, B, and A stars of known distances, will enable revolutionary studies of the temperatures, densities, and kinematics of the diffuse gas in the Milky Way. Pilot studies of feasibility for using a larger sample of F, G, and K stars for absorption line studies are crucial to densely sample structures on cloud scales. High-resolution (R >10⁵) spectroscopy (to resolve line profiles of absorption from cold gas) in the UV (~1,000–3,000 Å) toward stars with significant extinction is needed to probe the formation of molecular gas. This requires effective apertures ~3–5× present Hubble Space Telescope (HST) Cosmic Origins Spectrograph (COS) capabilities. Similarly, expanded starlight polarimetry surveys to map the ISM magnetic field toward stars with known distances are needed to map out the structure of the Milky Way’s magnetic field.

Wide-field (up to ~10 deg² for Milky Way regions), high-sensitivity mapping of velocity-resolved (~0.1 km/s) far-IR lines ([CII] 158 μm, [OI] 63 μm, and others) in the Milky Way and nearby galaxies is needed to study the “CO-dark” gas, along with matched resolution HI, CO, and Hα observations from existing facilities. Observational (X-ray to millimeter), theoretical, and laboratory studies are important for better characterization of the properties and evolution of dust, and its role in ISM thermal and ionization balance. Last, making full use of the proposed expanded set of observations will require galaxy simulations including realistic feedback, multiphase gas, radiative transfer, CGM/ISM inflow and outflow, cosmic ray acceleration and transport, and a live dust model.

F-Q2. WHAT REGULATES THE STRUCTURE AND MOTIONS WITHIN MOLECULAR CLOUDS?

Molecular clouds are structurally complex, with substructure arising from MHD turbulence, chemistry, and self-gravity. The processes that drive the turbulence remain unclear. Stellar feedback, in the form of flows external to the cloud driven by expanding HII regions and supernova blastwaves, is thought to play an important role. Within star-forming clouds, energy and momentum are also injected by protostellar outflows, and cosmic rays play key roles in heating the gas. Last, the role of gravity in driving supersonic motions cannot be neglected.

F-Q2a. What Processes Are Responsible for the Observed Velocity Fields in Molecular Clouds?

Large-scale motions within molecular clouds are observed to be highly supersonic. The kinematics (in projection) are currently studied using cold gas tracers such as CO 1-0 or 2-1. Maps of higher-excitation far-infrared lines—such as [CII] 158 μm, [OI] 63 μm, rotational transitions of OH, CH⁺ and other hydride ions, and high-J rotational transitions of CO—would be invaluable to trace gas heating by turbulent dissipation. With high spectral resolution, such maps will reveal the kinematics of the warm gas. To clarify the role of the magnetic field in turbulence and cloud structure, we need maps of polarized dust emission to reveal the geometry, and Zeeman effect measurements of field strength using species such as CN and OH. Also needed are MHD simulations, on various length scales, that include realistic gas physics, including chemistry and line emission.

F-Q2b. What Is the Origin and Prevalence of High-Density Structures in Molecular Clouds and What Role Do They Play in Star Formation?

The densest gas, in which stars form, generally comprises only a few percent of the total cloud mass, leading to low global star formation efficiencies. An observational census of the dense gas as a function of interstellar environment, and understanding how dense structures form and evolve, has important implica-

tions for understanding galactic-scale star formation. Moreover, dense structures set the initial conditions for subsequent collapse to stars and disks.

In nearby molecular clouds, the densest gas often appears filamentary. The pervasiveness of filaments in dust continuum images is one of the key results from Herschel. While filaments likely dominate the mass budget of the dense molecular gas where stars form, the understanding of their formation, fragmentation, as well as the degree to which they contain sub-structure remain controversial. Furthermore, it is not clear whether filaments are a widespread and critical step in star formation across galaxies of different properties. The key way to study filament formation, growth, and dispersal is via dense-gas kinematics using molecular species like ammonia, N₂H⁺, deuterated molecules, and CO isotopologues (Figure F.4). High-resolution observations can untangle and measure the gas flows within molecular clouds that assemble filaments, and search for infall motions and velocity oscillations along filaments that lead to core formation. While ALMA and the Green Bank Telescope are making important strides in this area, the progress is slow owing to the limited mapping speed for sensitive multitracer observations (e.g., lines tracing lower-density gas, high-density gas, and shocks). In tandem, for a large sample of star-forming clouds, we need high spatial resolution (<0.1 pc) mapping of the magnetic field in the filaments and cores to constrain field geometry and the extent to which magnetic fields provide support against gravitational collapse (Figure F.4). While ALMA and SOFIA are making important strides in this direction, larger samples probing diverse interstellar environments are essential.

In external galaxies, detailed studies of dense gas have been limited to only the closest galaxies, owing to the faintness of the relevant lines. While we cannot resolve individual dense structures, systematic measurements of the physical state of the cold and dense ISM across the full range of galactic conditions and environments found in the local universe will uncover processes that regulate the fraction of the star-formation feedstock—dense

molecular gas. Density, excitation, and chemistry play central roles in many theories of star formation; linking these quantities to galaxy structure and galaxy evolution across redshift requires multitransition, multispecies (HCN, HCO⁺, CO isotopologues, different excitation lines) mapping of a diverse set of galaxies.

F-Q2c. What Generates the Observed Chemical Complexity of Molecular Gas?

To fully interpret observations of molecular gas, we need to invest in testable chemical theories that can explain observed chemical abundances. Current astrochemical models fail to fully explain the complexities of observed molecular abundance ratios over a range of densities. In addition, observational efforts are needed to place rigorous constraints on models of chemical evolution at each stage of the star and planet formation process. In particular, understanding the formation pathways and excitation of large astronomical molecules, all the way to (pre)biotic molecules such as glycine and glyceraldehyde (the simplest sugar), and connecting molecule formation with processes happening in icy mantles of dust grains is critical. Laboratory studies are vital for measuring and quantifying critical pathways within astrochemical reaction networks.

Enabling Capabilities

Far-IR/submillimeter line imaging, ideally with high spectral resolution (R >5e4), is needed to distinguish heating by turbulent dissipation, stellar radiation, and/or cosmic rays. To address the role of magnetic fields in cloud structure and dynamics, high-resolution (<0.1 pc, ~10” [arcsec] to resolve filaments in Milky Way clouds) maps of polarized dust emission are needed, along with maps of circularly polarized emission from CN with <0.1 pc (~10”) resolution and sensitivity to Stokes V <0.5 mK at 0.2 km/s resolution. MHD simulations with chemistry, covering a range of scales from molecular cloud environments to dense cores, are essential to interpret observations by providing synthetic spectra and magnetic field maps for statistical comparisons.

High spatial (<0.1 pc) and velocity (<0.1 km/s) resolution radio and millimeter lines tracing gas at a variety of densities (radio: N₂H⁺, NH₃, deuterated molecules, millimeter: N₂H⁺, CO isotopologues) for Milky Way clouds are needed to study filaments. Far-IR and submillimeter polarization maps are important to trace magnetic field structure on the scale of molecular clouds and their substructures (<0.1 pc).

Cloud-scale (~100 pc, <1” at D = 20 Mpc) radio and millimeter spectroscopy of large samples of nearby galaxies to detect HCN, HCO+, CO isotopologues, and different excitation lines (>10× fainter than CO) are crucial to measuring the physical state of dense gas, and then to relate that to environment and the star formation efficiency.

Laboratory measurements and modeling of ices, gas phase chemistry, dust surface reactions, and the spectra of complex molecules are essential to understanding dense gas phases. Deep (RMS noise of 100 μJy/beam) centimeter/millimeter line surveys with spectral resolution to overcome line confusion (<0.1 km/s) and detect complex molecules with the angular resolution ~1” are required to isolate local environmental conditions.

F-Q3. HOW DOES GAS FLOW FROM PARSEC SCALES DOWN TO PROTOSTARS AND THEIR DISKS?

The fundamental challenge of star formation is to understand how processes spanning an enormous range of scales, from parsec-scale turbulent flows to sub-AU disk accretion, combine to produce the apparently universal stellar initial mass function (IMF) and planet-forming disks. Existing or nearly completed facilities, if complemented by some crucial future investments, will enable substantial progress in our understanding of critical fragmentation and accretion processes in the coming decade.

F-Q3a. How Do Dense Molecular Cloud Cores Collapse to Form Protostars and Their Disks?

The initial phase of star formation is controlled by the collapse and fragmentation of dense molecular gas onto a disk, which then transports mass inward and angular momentum outward. This process is dictated by the density structures of dense molecular gas “cores,” which are yet to be well resolved observationally. Moreover, the lifetimes of these cores are unknown: Do they represent a fixed mass reservoir from which bound stellar systems accrete, or do they evolve over the collapse time scale? Observations at submillimeter to centimeter wavelengths at the highest resolution available (~500 AU) for a small sample of cores reveal that they diverge from simple assumptions of spherical symmetry and solid body rotation (Figure F.5).

Addressing these fundamental questions demands more complex modeling, and that will require measurements of the internal kinematics and density fluctuations on scales below 1,000 AU for a much larger sample of cores. Insights on how magnetic fields mediate the collapse process can also be obtained from measurements of field morphologies—for example, from polarized submillimeter to centimeter dust emission and Zeeman splitting from key molecular line tracers. With such refined observational constraints, we can assess core lifetimes, their susceptibility to fragmentation and binary star formation, and ultimately the link between core masses and the stellar IMF.

F-Q3b. How Do Protostars Accrete from Envelopes and Disks and What Does This Imply for Protoplanetary Disk Transport and Structure?

The large-scale mass transfer intrinsic to core collapse continues down to smaller scales, with material transported through the disk and onto the central star. A better understanding of these disk transport processes (see also section F-Q4, below) is an important complement to the detailed studies of the core

structures outlined above. One compelling avenue for new insights on these processes comes from time-domain measurements, tracking broadband and spectroscopic variability on a range of time scales.

Figure F.5 illustrates the wide range of variability that should be characterized. High cadence (~seconds to minutes) targeted photometric surveys will identify low-amplitude, rapid variability in the inner disk and on the protostellar surface (flares) up to time scales of hours to days owing to a host of processes, from quasi-regular accretion via disk-magnetosphere interaction to (still poorly understood) extinction events. The latter may be produced by MHD turbulence or even signal the presence of perturbing bodies (e.g., planets). In contrast, deep, large-area surveys (e.g., the Vera Rubin Observatory) will provide complementary data on rarer, high-amplitude variability, as with outbursts from the FU Orionis disk, which dramatically affect both the masses and structures of stars and their disks. Because these outbursts are infrequent and long-lived (decades), samples of >10⁵ young stars are necessary to constrain their properties and occurrence rates (i.e., a census of regions stretching beyond ~2 kpc). Furthermore, Spitzer and Kepler monitoring campaigns showed that low-amplitude, medium-duration (~month) variability is common; there is a vast parameter space of unexplored time-domain behavior poised for discovery. On any time scale, a comprehensive characterization of time-variable accretion activity requires both photometric and spectroscopic studies of large samples, spanning a range of masses and evolutionary states. Near-simultaneous spectroscopic measurements at optical to near-IR wavelengths (ideally to ~5 μm) are needed to determine accretion rates and accurate stellar parameters (mass, radius, effective temperature).

F-Q3c. Is the Stellar Initial Mass Function Universal?

The form of the stellar IMF provides a fundamental test of star formation theories and is crucial to the interpretation of the composite spectra of distant galaxies. Whether the IMF is universal or depends upon the star-forming environment is not a settled question. To characterize mass functions in extremely dense, massive regions requires the study of clusters—in which most stars form—and associations nearer the galactic center and in nearby galaxies. The Magellanic Clouds offer an especially compelling opportunity to probe the effects of metallicity, which has been shown in local samples to have an impact on stellar multiplicity. To reach these distant environments requires better spatial resolution to deal with stellar crowding, and improved sensitivity to spectroscopically probe the peak of the stellar mass function, ideally to masses ≤0.2 solar masses, or K magnitudes ~23 at 50 kpc, with resolution <0.004 arcsec.

Enabling Capabilities

Measurements of the spatial distributions of physical conditions (densities, temperatures) and kinematics (rotation, collapse, and turbulent motions) in star-forming cores require sensitive, high-resolution mapping capabilities of key molecular gas and dust tracers accessible from submillimeter to centimeter wavelengths. Rotational transitions of CO isotopologues, ammonia, simple molecules (CN, HCN), and chemically important ions (N₂H⁺) are crucial diagnostics, along with very deep dust continuum data at a range of wavelengths and with full polarization capabilities. Observations at the longest wavelengths (~few centimeters) are especially important for probing high-mass star formation in environments with very high dust extinction, as well as better penetration of high optical depths in the densest parts (generally smallest-scales) of infalling low-mass cores. Those observations need sub-arcsecond spatial resolution (<1,000 AU for massive star-forming cores at ~kpc distances) and velocity resolutions of ~50–100 m/s in the spectral lines (to constrain protostar masses, subsonic turbulent line widths, and small collapse motions) at ~10× improved sensitivity compared to current facilities (e.g., the VLA) to assemble measurements for a sufficiently large sample in diverse environments.

Characterizing young variable objects requires high signal-to-noise (≳50) optical and near-IR spectroscopy with R ≳20,000 on ≳8 m telescopes, to identify faint protostellar absorption features (diminished by veiling continuum emission), disentangle photospheric emission from hot accretion continua and lines, measure nonthermal line profiles, and uncover distant, heavily extinct, star formation. To optimally synchronize these capabilities with time-domain surveys will also require a sophisticated alert system to trigger and prioritize follow-up. The Arizona–NOIRLab Temporal Analysis and Response to Events System (ANTARES) software “instrument” is an excellent example of how the community might achieve such goals.¹ Open-source platforms enable broad swaths of the community to develop unique alert brokers for a range of systems. Collaboration with computer scientists will also be key to training brokers to find new young star phenomena that were too rare or faint for previous discovery.

The requisite deep photometry in a narrow-field, moderate-resolution (R ~4,000) spectroscopy, and precise astrometry to distinguish cluster membership, can be done only with adaptive optics (AO)-equipped Extremely Large Telescopes (ELTs) with enough sensitivity to at least reach the peak of the IMF at ~0.2 M_☉ in the Magellanic Clouds (K ≳23). The James Webb Space Telescope (JWST) and the Nancy Grace Roman Space Telescope (formerly the Wide Field Infrared Survey Telescope [WFIRST]) can provide wider field IR imaging to study outer cluster regions, crucial to address mass segregation. Wide-field optical/NIR imaging and spectroscopy of many nearby low-mass star-forming regions might in sum provide a test of the low-density IMF.

F-Q4. IS PLANET FORMATION FAST OR SLOW?

A robust determination of the epoch of planet formation is a challenge for the coming decade. Are planets made slowly, over a time scale spanning their natal disk lifetimes? Or do they form quickly, at an early phase that overlaps with the epoch of star and disk formation? The former is the classical theoretical prediction and suggests that disk properties should be interpreted as the initial conditions for planet formation. The latter implies instead that those properties may already be modified by dynamical interactions with young planetary systems. Building on recent advances, a direct route to answering this question focuses on the origins and demographics of substructures in circumstellar disks. That work needs to be supplemented with efforts to advance our capabilities to quantify the physical and chemical conditions of the disk material. Progress in these areas will facilitate an improved understanding of angular momentum transport in the disk and accretion onto the star, and particularly how those processes are related to the turbulence that impacts many aspects of planetary growth.

F-Q4a. What Are the Origins and Demographics of Disk Substructures?

High-resolution observations at near-IR and submillimeter wavelengths have revealed ubiquitous, small (~few astronomical units) perturbations to disk structures in the forms of rings, gaps, spirals, arcs, and other asymmetries. As an illustration, Figure F.6 shows a gallery of disks imaged at high resolution in the millimetercontinuum. These disk substructures are likely to be the hallmarks of localized particle concentrations at gas pressure maxima. Pressure maxima in disks can be produced by various (magneto)hydrodynamic instabilities, and in this interpretation, the observed substructures are fluid mechanical features that may catalyze the planet formation process. Alternatively, pressure maxima and substructures could be the signposts of dynamical interactions with an already-formed generation of planets. That hypothesis would falsify the classical planet formation models that predict very slow time scales in the outer regions of disks, and

___________________

¹ A. Saha, Z. Wang, T. Matheson, G. Narayan, R. Snodgrass, J. Kececioglu, C. Scheidegger et al., 2016, “ANTARES: Progress Towards Building a ‘Broker’ of Time-Domain Alerts,” Proceedings of SPIE 2016: Observatory Operations: Strategies, Processes, and Systems VI (A.B. Peck, R.L. Seaman, and C.R. Benn, eds.), SPIE Proceedings Volume 9910, Bellingham, WA.

point instead to more efficient mechanisms to assemble giant planets with (at least initially) long orbital periods (perhaps employing rapid planetesimal formation and pebble accretion).

The physical mechanisms at play in both possible interpretations are broadly understood, but substantial computational work is required to make quantitative predictions that capture the interaction of magnetohydrodynamic, thermal, and chemical effects in disks. The key observational clues on the origins of disk substructures are expected to come from detailed characterizations of their morphologies (e.g., gap shapes), kinematics (non-Keplerian deviations), and physical conditions (density contrasts, pressure gradients, particle size distributions, etc.). Enhanced spatial resolution, particularly at more transparent continuum wavelengths (i.e., the centimeter radio bands) and in molecular spectral line emission, are essential, both to resolve the larger substructures and to probe the currently inaccessible features in the inner few astronomical units of nearby disks. Ultimately, the information gleaned from these detailed characterizations will benefit substantially from demographic explorations of how substructure properties depend on external factors like the host mass, age, and environment.

F-Q4b. What Is the Range of Physical Environments Available for Planet Formation?

The metamorphosis of disk material into planetary systems involves a complex set of physical processes. Models for those processes are always limited by imperfect knowledge of the planet formation environment—the spatial variations of temperatures, densities (for gas and solids), chemical composition, particle sizes (and other microphysical properties), gas dynamics (turbulence), and magnetic fields. Over

the past decade, considerable progress has been made on vetting the approaches available to constrain those environmental properties. In the next decade, we are poised to transition from that exploratory phase into a quantitative era that enables progress toward more predictive models of disk behavior.

The physical conditions in disks can be determined from a range of data sets and methodologies. The most promising options for determining the temperature, density, chemical, and dynamical structure of the gas rely on spatially and spectrally resolved observations of molecular emission lines with a variety of optical depths and excitation states through the submillimeter bands and into the far-infrared (also, an important link to ices will be available in the mid-infrared with JWST). To anchor estimates of column densities onto an absolute scale, reference to measurements of a more direct gas mass tracer, particularly the isotopologue HD, would be beneficial. For the solids, the spatial variation in the particle densities and size distributions can be constrained with spatially resolved maps of the spectral shape and linear polarization of the submillimeter through centimeter continuum emission. Magnetic field strengths and geometries may be accessible through the Zeeman effect in various spectral features throughout the submillimeter bands.

F-Q4c. How Do Turbulence and Winds Influence the Evolution of Structure in Disks?

The paradigm that turbulence enables the angular momentum transport that drives accretion in disks has been challenged over the past decade. Angular momentum loss in a magnetohydrodynamic disk wind has emerged as an important alternative, which may be more compatible with (still limited) observations of minimal (nonthermal) spectral line broadening and the strong confinement of dust continuum features (both radially, in substructures, and vertically, in characteristic heights) in disks. Aside from better characterizing the different potential transport process, continued efforts to measure turbulence are highly desirable because of the diverse roles it plays in planet formation—from limiting the collisional growth of solids, to affecting the formation of planetesimals, and beyond to regulating the structures of gaps and the migration of protoplanets in gas disks.

Enabling Capabilities

The science issues of this question are best addressed with high-resolution measurements of the continuum and spectral line emission accessible to submillimeter to centimeter interferometers. Current facilities reach 20–70 mas resolution (3–10 AU for nearby disks), with limited surface brightness sensitivity. Efforts to resolve larger substructures and discover new features that are smaller or located within ~10 AU of their host stars are essential, with a targeted factor of ~5 improvement in resolution (to ~5 mas, or sub-AU scales). Substantial sensitivity improvements will be necessary to probe the (~percent level) polarization and spectral line emission on those scales, as well as especially faint but chemically important emission lines at coarser resolution. An appropriate metric is an order of magnitude decrease in the noise level for a fixed integration compared to current facilities (e.g., VLA, ALMA). High-resolution access to the centimeter/radio bands are crucial: the lower continuum optical depths there offer unique access to the properties of disk solids, particularly in the inner disk (terrestrial planet region), and measurements of free-free emission and H recombination lines that could be the key to quantifying disk winds. Those interferometric capabilities would strongly benefit from complementary observations of spectral lines throughout the far-infrared (particularly from water vapor and HD), as well as improvements in high-resolution images of disk structures in optical/infrared scattered light from dust grains entrained in the gas. The latter can be achieved with similar targets for improvements in resolution (~5× better) and sensitivity (~10× deeper) as indicated for the submillimeter to centimeter interferometers.

DISCOVERY AREA: DETECTING AND CHARACTERIZING FORMING PLANETS

Opportunities to characterize young planets and their circumplanetary material while they are still embedded in their natal circumstellar disks will lead to major advances in our understanding of the planetary accretion process, the formation of early atmospheres and satellites (moons), and the origins and evolution of both the solar system and the exoplanet population. Observations of dynamically cleared gaps in circumstellar disks may point to a population of young giant planets with masses ~ 10 M_⊕to 10 M_Jup and orbital separations of ~10–100 AU. At least one example of a system with actively accreting, young, giant planets has been found in a cleared disk gap, as shown in Figure F.7. Current observations with 8–10 m telescopes are sensitive to only the most massive planets, but extrapolations to ~30 m class facilities and reasonable improvements in instrumental capabilities can help find and characterize many more examples. In the longer term, the goal is to detect forming planets across a broad range of masses and orbital distances, transforming our understanding of planet formation and early planetary system evolution.

F-DA1. How Do Planets and Their Satellites Grow?

The transport of material from the circumstellar disk to a planet is mediated by a circumplanetary disk (CPD). By analogy to the circumstellar disk/host star system, we expect a variety of accretion diagnostics from the CPD/planet system, including Balmer line emission from magnetospheric funnel flows and excess UV/blue continuum from accretion shocks. These features have been observed in the few available CPD candidates associated with massive (>few M_Jup) planets (e.g., Figure F.7). Such measurements constrain the protoplanet accretion rate, informing models that describe how giant planet envelopes grow from flows across the gaps that planets sculpt in their parent disks. The physical conditions, spatial structures, and dynamics of CPDs are controlled by the combined gravitational potential of the protoplanet and host star, the local heating, and the mechanics of mass transfer from the circumstellar disk reservoir. Direct imaging measurements of the CPD spectral energy distribution (SED) would provide crucial insights on the thermal structure, and therefore constraints on the protoplanet luminosity and mass. Measurements of the CPD at submillimeter to centimeter wavelengths are sensitive to the CPD mass (in solids) and potentially its size, which scales with the planet’s Hill radius (and thereby mass). Taken together, these measurements offer foundational boundary conditions for models of satellite formation. Ideally, they would be complemented with measurements of the CPD gas from molecular emission lines, as well as millimeter- to centimeter-based estimates of the widths, depths, and kinematic perturbations of the associated gaps in the circumstellar disks (all sensitive to the protoplanet mass), to provide multiple diagnostics of CPDs and their planet hosts.

F-DA2. What Are the Atmospheres of Long-Period Giant Planets Like at Their Formation Epoch?

Direct imaging detections of young planets at wide orbital separations present compelling opportunities for spectroscopic characterization (e.g., see Figure F.7). The ability to measure absorption features from relatively extended planetary atmospheres would enable the first compositional analyses of planets at their formation epoch. These can be compared with measurements of their highly irradiated counterparts at small orbital separations (from transit spectroscopy) or their more evolved analogs around main-sequence stars (found via direct imaging) to better understand the chemical evolution and diversity of giant planet atmospheres. Moreover, estimates of atmosphere properties would help extrapolate a spectral sequence from young low-mass stars and brown dwarfs to offer some first-order benchmarks for key planetary properties (particularly masses). This would enable a rough empirical calibration of the link between the properties

of planets and the disk gaps they sculpt, permitting a rubric to extrapolate toward the planet properties needed to explain more subtle disk features.

F-DA3. How Do the Orbital Architectures of Planetary Systems Evolve?

When an ensemble of protoplanets (and their CPDs) have been imaged, we can infer the giant planet mass function and its variation with orbital separation at the formation epoch. Then, by empirically characterizing how these planets perturb their disks as a function of their mass, we could extrapolate to indirectly infer constraints on the mass function into the super-Earth regime by measuring more subtle disk substructures. These young planetary architectures could then be compared with their more evolved counterparts around main-sequence hosts found from microlensing surveys and direct imaging campaigns. That comparison would provide novel constraints on the distribution of planetary migration histories. For example, if more long-period giant planets are identified at early times, it may indicate that the radial velocity and transit populations of exoplanets could have migrated from much more extended initial architectures.

Enabling Capabilities

The goals associated with this discovery area can be achieved with improved capabilities for deep direct imaging campaigns in the optical/IR, particularly when including modest resolution spectroscopy. Success will require deeper contrast limits (to search for long-period Saturn analogs around a range of

host star masses requires near-IR contrast limits ~15× better than current ground-based 8–10 m telescope capabilities), improved adaptive optics systems for higher Strehl ratios to broaden the search for planets orbiting fainter hosts (e.g., R ~13–14 or better), smaller inner working angles to probe orbital separations of ~5–10 AU (~30–60 mas), and better resolution (~10 mas) to optimize contrast by resolving the associated disk gap and potentially the CPD. These goals are achievable from the ground with “extremely large” aperture telescopes (i.e., ~30 m diameter mirrors). There would be considerable value added with complementary instrumentation in the thermal infrared, ~3–5 μm (L and M bands), where planet and CPD contrasts are significantly improved. An instrument that can access the 10 and 20 μm windows (N and Q bands) would optimize the CPD contrasts for both the host star and surrounding disk. Moreover, assembling a more complete SED, from optical to mid-IR, will help disentangle bulk atmosphere properties of protoplanets and the thermal structure of CPDs. Spectroscopic capabilities coupled to the adaptive optics system are also essential, both to probe atmosphere features and for accretion diagnostics. Integral field units or analogous technology (including interferometric capabilities) with resolving power of at least a few hundred (and preferably up to a few thousand) are necessary. Links to the substructures these protoplanets sculpt in their natal disks will require the improved observational capabilities for submillimeter to centimeter interferometry data discussed in detail in the previous section.

THEORY AND MODELING

In tandem with the improved observational capabilities discussed earlier, progress will be further enabled by theoretical and modeling advances, which are enabling capabilities for all of the science questions. In common with other fields, computational progress in ISM, star formation, and planet formation problems requires ongoing improvements in the size and efficiency of large, multiscale numerical simulations that include MHD and self-gravity. The largest simulations—for example, of the launching of galactic winds owing to star-formation feedback into the ISM (F-Q1)—are at the limit of what is possible given existing supercomputers. Radiative transfer is required to create synthetic observations that can be compared against data, and in some problems as a dynamical actor (F-Q1, F-Q2, F-Q3, F-Q4, F-DA). Funding for the development of faster, more accurate, and more scalable algorithms in these areas is key, as it can improve the fidelity of simulations more rapidly than is possible from iterative advances in machine size (F-Q1, F-Q2, F-Q3, F-Q4). Also common with other fields is the likelihood that, over the next decade, a greatly increased fraction of computer resources will be devoted to machine learning (F-Q1, F-Q2, F-Q3). Machine learning has multiple potential applications, including the identification of rare time-domain events in large data sets and the development of fast ways to compare simulations quantitatively against data.

The ISM, star formation, and planet formation also involve key physical processes that are less commonly encountered elsewhere, including non-ideal MHD and multiple-fluid effects, complex time-dependent chemistry and dust evolution, and planetary growth processes (F-Q1, F-Q2, F-Q3, F-Q4). Many of these physical processes have important effects on fluid motions, temperature, ionization, chemistry, and dust properties that are not yet adequately understood. Hundreds of diffuse interstellar bands are observed, but only a single carrier (C₆₀⁺) has been identified, accounting for only three of the observed features (F-Q1). Uncertainties in collisional excitation cross sections limit the accuracy of temperature and abundance determinations from emission lines. Laboratory and atomic and molecular astrophysics needs to be supported to exploit and interpret astronomical data (F-Q1, F-Q2, F-Q3).

TABLE F.1 Science Questions and Discovery Area: Interstellar Medium and Star and Planet Formation

TABLE F.2 Summary of Capabilities, Facilities, and Future Needs to Address the Science Questions and Discovery Area

NOTE: XMM: X-ray Multi-Mirror Mission; MMT: formerly Multiple Mirror Telescope; LBT: Large Binocular Telescope.