The week I started graduate school, the first science projects were announced for the new Atacama Large Millimeter/submillimeter Array (ALMA) telescope in Chile. This groundbreaking facility uses dozens of radio antennas working in concert to create images as detailed as those made by a single telescope 16 kilometers wide. With this extreme resolution, ALMA can see deeper and farther in millimeter- and submillimeter-wavelength light than any previous telescope. I leaped at the opportunity to join one of its first projects—a study of a disk of dust and rubble around a nearby star called AU Mic. The subject of our observations was something scientists had never seen in such detail before ALMA was built. Dust and rubble might not sound that exciting, but they are the raw materials planets are made of, and this observatory was giving us a chance to see the process in action.

It took another year for the data to be delivered. Modern astronomy is often done at a distance: rather than spending long nights at the remote mountain observatory, all we had to do was submit a computer script that told the telescope what to observe and when. Then we waited patiently (or, more often, impatiently) for our observations to be scheduled and completed. I can still remember the anticipation, the butterfly feeling in my stomach as I waited for the data download and, when it was finally ready, the awe when the image appeared on my computer screen—a long, thin blob of light with three bright spots: one in the center and two on either side at the edges.

What we were glimpsing was a solar system growing up. The central spot was actually the star, which we now know is flaring, sending bursts of high-energy particles out into space. The other two bright spots marked the edges of a disk of debris circling the central star, akin to the Kuiper Belt that orbits our sun. We think this band is the rubble left over after planets formed around AU Mic, a young M dwarf star about 32 light-years away. Other scientists have recently discovered two planets in the system: one about the mass of Jupiter and the other about the mass of Saturn, both orbiting fairly close to their star. Now we have an unprecedented opportunity to see how the material in the disk evolved and interacted with the newly formed planets.

Since that early image, the capabilities of ALMA have continued to expand, and the array now has new dishes, higher resolution and more wavelength coverage. Meanwhile the study of circumstellar disks and planet formation has exploded. ALMA has taken several hundred planetary baby pictures, helping us to build a new view of how such systems form and revealing troves of planets we never could have detected otherwise.

Baby Planets

Stars form out of vast regions of gas and dust called molecular clouds. The typical density of empty space is only one atom per cubic centimeter, but the thickest areas of molecular clouds can reach densities 10,000 to one million times this norm. When these spots, or “cores,” become dense enough, they start to collapse under their own gravity to make stars. At the same time, the initial rotation of the collapsing core and the conservation of angular momentum naturally form a disk surrounding the newly born star. Astronomers call these collections of dust and gas “circumstellar” (meaning “around stars”) disks.

When stars are still very young (only a few million years old), their circumstellar disks are relatively huge, often with about 1 to 10 percent of the mass of the central star in a typical system. For a star like the sun, that amounts to a disk with roughly 100 times the mass of Jupiter. These young, massive Frisbees are “protoplanetary” because we think this is where planets are actively forming. Rock, metal and ice condense out of the disk to form planetary seeds. As seeds start to collide and stick together, they grow larger and larger until they have enough gravity to start attracting more material through a process known as accretion. The baby protoplanets orbit within the disk and continue accumulating material, carving out gaps in the disk in a game of planetary Pac-Man. Nearly all stars that are younger than a few million years are surrounded by disks that most likely harbor a zoo of new planetary systems.

The protoplanetary disk phase lasts for several million years. After that point, most of the gas and dust from the initial circumstellar disk has cleared. How this clearing happens and over what timescales are areas of active research, but we think that a lot of the dust and gas in the original disk either migrates inward and falls onto the central star or is blown out by strong stellar winds. After approximately 10 million years, all that is left is a mature star surrounded by a new planetary system and a disk of remnant asteroids and comets. The total mass of this leftover material is low—likely less than 10 percent of the mass of Earth. Although there may still be enough mass in these “debris disks” to form small terrestrial planets or Pluto-like bodies, you can think of them as the fossil record of earlier planet formation. Their structure is sculpted through gravitational interactions with the newly formed planets, and their composition gives us clues as to what material was originally built into those planets.

HIGH IN CHILE'S Atacama Desert, the ALMA observatory uses dozens of antennas in tandem to capture images of distant planetary systems. Credit: CLEM and Adri Bacri-Normier ( and ESO

Astronomers first discovered debris disks when the Infrared Astronomical Satellite (IRAS) was launched in 1983. It was the first satellite to survey the entire sky at infrared wavelengths (12 to 100 microns; a human hair is roughly 75 microns across). You can think of infrared radiation as heat. When IRAS scanned the infrared sky, astronomers discovered that many stars looked brighter than expected. Why? The answer proposed was dust. If these stars were surrounded by disks of dust, the grains would get heated by the star and then radiate thermal emission in the infrared range. From this inference, a new area of research was born. In fact, the first four debris disks discovered by IRAS—Vega, Beta Pictoris, Epsilon Eridani and Fomalhaut—are still studied and puzzled over today.

By using infrared telescopes to search for such bright spots, astronomers have confirmed that at least 20 to 25 percent of stars are surrounded by debris disks. Given our picture of how planetary systems form, one might logically conclude that all stars should be surrounded by remnant material—after all, statistics from the Kepler mission tell us that every star in the galaxy has at least one orbiting exoplanet. In fact, debris disks are probably more common than we know. Even our solar system has its own system of multiple debris disks—the asteroid belt and the Kuiper Belt. Yet the solar system is actually dust-poor compared with the systems around other stars we have been imaging. In fact, the deepest infrared surveys to date have been able to identify only disks with dust masses roughly an order of magnitude higher than what we see in our solar system. Does that make our cosmic home an oddball? We are not sure yet. We have been studying the most massive, most extreme disks, but there are probably many more low-mass disks to be found that will help us put our own planetary system into context.

Although astronomers began to infer the presence of dusty disks from early infrared observations in the 1980s, they did not know what they looked like. Before improvements in telescope technology were made in the 1990s and 2000s, only a single star system—Beta Pictoris—had been resolved. Notably, the Hubble Space Telescope employed coronagraphic imaging, a technique astronomers use to block the light from the central star in order to see dimmer surrounding objects, to image light scattering off small dust grains in circumstellar disks. Although many of these early images were indistinct, they gave the first indication that circumstellar disks actually have extended, complicated structures. In the case of the debris disk around Beta Pictoris, the first Hubble images showed a warp in the inner regions of the disk that astronomers thought might indicate an unseen planet. Direct imaging later confirmed this baby world.

A New Telescope

The wavelength of light that we see reflected from dust roughly corresponds to the size of the dust grains—optical and near-infrared light comes from small dust grains tens of microns in size, whereas far-infrared and millimeter-wavelength imaging is sensitive to larger grains similar in size to sand. We think that these larger grains are better tracers of the underlying structure of circumstellar disks. Within a disk, there is a continuous cascade of collisions. Large comets and asteroids crash into one another and get ground down into smaller and smaller dust grains. The most massive objects in the disk are called planetesimals, and their locations are shaped by interactions with other planets in the system. If we can locate the planetesimals, that information can be used to infer the presence of unseen planets, even if we can never observe those large bodies directly.

The tiniest dust grains are easily moved around by interactions with interstellar gas or are simply blown out by winds and radiation from the star itself. But because the larger sandlike grains are less affected by such forces, they offer us the best opportunity to uncover the underlying disk structure and unseen planets through their gravitational influence.

Credit: Nigel Hawtin

Therefore, we want to look at long wavelengths to study disk structure and to search for signatures of unseen planets. It seems straightforward—but of course, there is a catch. The resolution of a telescope is equal to the observing wavelength divided by the diameter of the telescope. Thus, as you increase the wavelength from the optical to the millimeter range, you have to dramatically increase the size of the telescope to achieve the same resolving power. Hubble has a diameter of 2.4 meters, which gives a resolution of 0.13 arc second for observations at a one-micron wavelength. If you wanted to achieve the same resolution at a wavelength of one millimeter, you would need to increase the telescope’s diameter by a factor of 1,000 to more than two kilometers! We cannot build a telescope that large, so we have to use a technique called interferometry. Essentially, instead of a single two-kilometer-diameter telescope, an interferometer spreads multiple smaller telescopes out over two kilometers and combines their signals to achieve equally high resolution.

ALMA, which took its first images in 2011, is still the world’s most powerful interferometer. Located at an elevation of roughly five kilometers in Chile’s Atacama Desert, ALMA has 66 antennas that can be relocated to span baselines (the distance between any two antennas) of 150 meters to 16 kilometers. If you are familiar with the Washington, D.C., area, picture the White House Ellipse: in its most compact configuration, ALMA would fit entirely within it. In its most extended configuration, it would span the entire Capital Beltway. With such advancements in both sensitivity and resolution, we can now image fainter objects in greater detail than ever before. It is not an overexaggeration to say that ALMA has revolutionized our understanding of circumstellar disks.

In one of its first blockbuster disk images, taken in 2014, ALMA imaged HL Tau, a young system probably less than 100,000 years old. The photograph revealed that what had been assumed to be a continuous disk was carved into multiple rings and gaps. Given the young age of the system, if these gaps are actually sculpted by baby planets, planet formation must start earlier than originally thought. In another notable discovery, in 2018, the DSHARP (Disk Substructures at High Angular Resolution Project) survey looked at 20 protoplanetary disks with high resolution and found that every one of them had rings and gaps, and some even showed spiral structure. Apparently such features are not unique to HL Tau but are instead ubiquitous to young circumstellar disks.

Planet Detection

In addition to teaching us about the process of planetary formation, studying disks is also a good way to detect exoplanets we would otherwise be unable to find.

Telescope missions such as Kepler and TESS (the Transiting Exoplanet Survey Satellite) and many ground-based surveys have so far detected thousands of exoplanets. Yet most of these planets are more massive or are closer to their host star than the planets in our solar system are. These types of planets are not necessarily more common, though; they are simply easier for us to find. The two top methods of detecting exoplanets are the transit technique, which looks for periodic dimming of stars when planets orbit in front of them, and the radial velocity method, which traces planets through observation of the slight change in velocity they cause in their host stars because of their gravitational pull. Both methods favor large planets with short orbits because multiple orbits must be observed to confirm a detection, which means that astronomers using these methods might be missing a lot of planets. Neptune, for instance, has an orbital period of roughly 165 years: if you were studying our solar system from a different star, you would be waiting a very long time before you saw it transit the sun even once. The few planets we do know about that are at Neptune-like distances from their host star have been detected via direct imaging, which uses coronagraphy—blocking the light from the host star—to image the planet itself. This approach has its own observational biases, however, favoring young systems where the planets still retain significant heat from their formation.

To put the architecture of the solar system in context, we must be able to detect giant planets at large distances from their host stars in old systems. Now, with ALMA, this can be done by using the resolved structure of circumstellar disks, providing a powerful complement to other methods of exoplanet detection.

We can find Neptune-like planets, for instance, by studying features of disks sculpted by planets orbiting within them, such as warps, clumps and other asymmetries. In our own solar system, the classical Kuiper Belt is quite narrow because of the gravitational influence of Neptune. We think that during the early evolution of the solar system, Neptune initially formed closer to the sun and then migrated outward, sweeping up much of the remnant material in its wake to create the Kuiper Belt seen today. If we observe similar structures in extrasolar debris disks, we can use them to infer the presence of unseen Neptune analogues.

We can also learn more about planets we already know of by studying the disks they inhabit. The HR 8799 system has four directly imaged giant planets orbiting between analogues of our own asteroid belt and Kuiper Belt. With millimeter interferometry, we can resolve the structure of the system’s outer Kuiper Belt analogue and determine the location of its inner edge. If we assume that the outermost planet in the system is responsible for carving out the disk, we can use the location of the inner edge to constrain the possible mass of the planet as roughly six Jupiter masses. That may not seem like a significant feat, but it is far more precise than our previous best estimate of the planet’s mass, which relied on theoretical models of how planets cool and dim over time. Using the disk’s structure, we can provide an important independent check on those models.

ALMA observations of younger protoplanetary disks show a wealth of detailed structure; rings and gaps seem to be present in nearly every system. If all those gaps are carved by planets, we can assume there is a large population of unseen ice-giant planets present. Tying structure in young systems directly to planets is challenging, however, because other effects complicate modeling efforts. Older, more evolved systems are easier to interpret, but so far very few of these debris disks exhibit multiring structure. Recently we discovered a new gap in the HD 15115 debris disk located beyond where Pluto orbits in our system. Dynamical modeling suggests that this gap represents an ice-giant planet with a mass slightly less than that of Saturn. I suspect that as we obtain deep, high-resolution images of more of these evolved systems, more planet-induced features will come to light.

Furthermore, beyond the structure of circumstellar disks, we can also study their composition. Because these disks are the reservoirs and fossil records of planet formation, their composition is intimately tied to the composition of planets in these systems and to their formation history. Numerous common molecules emit light at millimeter wavelengths because of the bending and stretching of their molecular bonds. Scientists have detected dozens of organic molecules (including carbon monoxide, formaldehyde, methanol and ammonia, among many others) in the large gas reservoirs present in protoplanetary disks.

Our research has also uncovered a new mystery: Traditionally debris disks were assumed to be gas-poor because their initial gas reservoirs should be cleared within a few million years. ALMA has revealed that a number of debris disks contain carbon dioxide gas, but we interpret this as the result of comets colliding in the disk and releasing trapped ice in the form of gas as they are ground into small dust grains. A few systems challenge this picture, though, because they contain such a large amount of gas that it would take an unrealistically high rate of cometary collisions to produce it. This discovery prompts a question: Is it possible for primordial gas to remain in these disks for tens of millions of years? As of yet, we do not have an answer.

DEBRIS DISKS, imaged by ALMA, represent a later stage of evolution than protoplanetary disks, after a star and its planets have formed. Their bright bands of rubble are akin to our solar system's Kuiper Belt. Credit: ALMA (ESO, NAOJ and NRAO)

A Multiwavelength Future

It has been exciting for me to grow up as a scientist while the field of planet-formation research has grown up around me. I began working on my Ph.D. as ALMA first opened its eyes on the sky, and I am beginning my first faculty position as we move into an exciting new future of multiwavelength astronomy. ALMA has revolutionized our understanding of circumstellar disks, revealing complexities in structure and chemical composition that could have only been guessed at a few decades ago. But ALMA cannot answer all the questions we want to explore. All the debris disks I have discussed in this article are analogues of the Kuiper Belt, cold rings of dust in the outer regions of their solar systems. So far astronomers have struggled to image an analogue of the asteroid belt—we can still detect such features only through their excess infrared light, as we did in the early days with IRAS.

To image the inner regions of extrasolar systems, we need shorter wavelengths that are sensitive to hotter dust. The James Webb Space Telescope (JWST) is due to launch in 2021, and we expect it to take the first picture of one of these asteroid belt analogues. Beyond that, JWST will operate at wavelengths that directly trace emission from silicates (minerals such as olivine and pyroxene, which are also found on Earth) and that constrain the mineral composition of disk grains.

Looking even further into the future, the next generation of “Extremely Large Telescopes” is being constructed now, and these instruments will see their first light in the mid- to late 2020s. These telescopes will have diameters greater than 24 meters, more than five times larger than any current ground-based telescopes, and they may be able to directly image some of the planets we can only infer now from ALMA disk observations.

The Decadal Survey on Astronomy and Astrophysics—a field-wide effort to decide on priorities for future funding—is underway now. Under consideration are four NASA flagship missions that could make huge advances in planetary science in the 2030s and beyond. The Origins Space Telescope, a cryogenically cooled infrared observatory, could trace how water from star-forming regions ends up in circumstellar disks, provide statistics on low-mass disk populations, and much more. Other candidates such as the Large Ultraviolet/Optical/Infrared Surveyor and the Habitable Exoplanet Observatory are direct-imaging missions that could detect and characterize many exoplanets, some of which could be Earth-like.

Regardless of which of these missions is ultimately selected, the one thing I know for sure is that our understanding of the solar system and of its formation and its place in the universe of exoplanet systems is changing every day. The butterfly feeling in your stomach while you wait to see what each new observation looks like—it never goes away.