The First Photograph of Another Earth

Astronomers are working today on the camera that will see habitable planets in other solar systems.

Subaru Telescope
The Subaru Telescope in Hawaii is an early tester of advanced new detectors for astronomy.

Subscribe to Air & Space Magazine Now

This story is a selection from the August issue of Air & Space magazine

Astronomers have devised ingenious methods for finding objects that were, as late as 2010, thought to be rare: Earth-size planets orbiting other stars. So far, more than 4,000 exoplanets of all sizes have been discovered, leading scientists to estimate that, in the Milky Way alone, there are 100 million habitable planets. And yet, despite the recent discoveries, one of the greatest goals of the scientific community remains elusive: an image of a habitable, Earth-like planet orbiting in another solar system. Capturing an image is the only way to confidently identify a small rocky planet among the thousands discovered and to study its atmosphere, assess its orbit, and determine whether it might be capable of supporting life. The conundrum for astronomers is that—unlike stars, superheated gases, and other bright objects—Earth-like planets are too dim to be seen by telescopes.

Now the first image of such a world finally might be within reach. Last summer, a team of astronomers from the University of California Santa Barbara installed a powerful new kind of camera on the eight-meter Subaru telescope at Mauna Kea, Hawaii—the so-called Astronomy Precinct, where a dozen observatories with 13 telescopes have set up shop.

This camera relies on a sensor called a microwave kinetic induction detector, or MKID. The sensor makes the MKID Exoplanet Camera (MEC) more than a camera; it’s a photon counter, capable of providing the arrival time, location, and energy of every single photon that hits the detector.

MKIDs are much faster and more sensitive than the charge-coupled devices (CCDs) widely used for digital imaging. Looking for a dim exoplanet through the fluctuating distortions of Earth’s atmosphere requires levels of speed and sensitivity beyond what CCDs have provided. That’s why some astronomers are convinced that these new sensors could become a transformative technology, gradually replacing many of the other instruments in today’s telescopes.

* * *

Ben Mazin, the UCSB researcher who led the development of the new exoplanet camera, points out that MKIDs aren’t exactly new. The technology was conceived in 1999 at a coffee shop by his graduate school adviser at Caltech, Jonas Zmuidzinas (pronounced zehmoid-zenahs), and Rick LeDuc, at NASA’s Jet Propulsion Laboratory. Four years later, they published a journal paper describing the detection of X-rays with MKIDs.

“Starting in about the 1980s, it was obvious that there were two directions that detectors needed to go to gain sensitivity,” Mazin says. First, detectors would have to get colder. The circuitry in CCDs produces heat, and the heat can create false images, or noise. Second, they would have to react to far less energy than that which is required to activate CCDs.

CCDs are semiconductors; they work like light switches. They’re normally in the “off” position, but if you add enough energy—by bombarding them with photons, for example—they turn to the “on” position, allowing an electric current to run through them. The new MKIDs rely on superconductors, which turn on when struck by a very small amount of energy. Because the amount of energy required to turn them on is much smaller, they can detect much dimmer light sources.

That’s why the search for exoplanets is an excellent testing ground for MKIDs. Exoplanets are hard to detect. The earliest confirmed discovery wasn’t until 1992, and almost all of the exoplanets discovered since then have been found by instruments using indirect methods. Astronomers don’t typically “see” exoplanets. Rather, they infer their presence through methods such as measuring the effect of an exoplanet’s gravity on its sun, or how much the light of its sun dims when the exoplanet passes in front of it (see “Our New Planet Hunter").

Direct imaging, though, is dependent on the light generated by or reflected by the object being photographed. The CCD cameras most telescopes use consist of an array of tiny sensors embedded on a silicon chip, with each sensor representing a single pixel. The sensor creates and stores a small electric charge proportional to the amount of light (energy) that hits it. Later, a computer measures the charge of each sensor and uses that information to compile an image. The stronger the charge, the brighter the pixel.

A handful of exoplanets have been detected through direct imaging, using a powerful telescope and a highly sensitive camera to actually take a picture of the planet. In fact, by combining images taken over several years at the Keck Observatory, astronomers were able to create a jerky, time-lapse video of four planets orbiting a star known as HR 8799, 129 light-years from Earth. But these planets are two to three times larger than the gas giants in Earth’s solar system.

So far, astronomers have been able to take pictures only of bright gaseous planets many times larger than Jupiter, and nearly hot enough to be stars themselves. Cool, rocky Earth-like planets, by contrast, don’t shine; they emit only the light they reflect from their star. In addition, the gas giants that were photographed by the Keck Observatory were all much farther from their star than Jupiter is from our sun. Small, potentially habitable planets will be orbiting much closer to a star, in the so-called Goldilocks Zone, roughly defined as just warm enough to support liquid water crucial for life. The closer a planet is to its star—especially a cool planet, like Earth—the more difficult it will be to distinguish its dim, reflected light from the booming glare of its sun.

This task is far beyond the capacity of today’s telescopes or the current generation of cameras. That’s where the MKIDs come in.

Mazin lists the key metrics that astronomers use to characterize a camera. The first is “quantum efficiency,” which is simply the likelihood that a photon hitting the array will be detected. The quantum efficiency of a modern CCD is more than 90 percent, so it is able to detect nine out of 10 of the arriving photons. For MKIDs, quantum efficiency is above 70 percent, but more complex anti-reflection coatings could push that figure higher.

The next measure of a camera is the level of noise, or false signals, generated by the detectors. In addition to the “thermal noise” introduced by the heat CCDs produce, so-called “read noise” can pollute images. Read noise is caused by the amplifiers that are used to read out the amount of charge that’s accumulated in each pixel of a CCD. “The nice thing about the MKIDs is that, when they’re performing correctly, they don’t have either of these noise sources,” Mazin says. “MKIDs are essentially noise-free.”

Another property of detectors is spectral resolution—how accurately they can determine the wavelength of the incoming light. CCDs have no spectral resolution. In order to determine the energy of the photons, scientists channel some of the light to another instrument. This leads to light loss—a real problem when you’re observing dim objects like exoplanets.

One of the benefits of MKIDs, Mazin says, is that they have inherent spectral resolution. “They can actually tell you the wavelength of every arriving photon.” That’s crucial to the search for habitable exoplanets, since spectroscopic analysis is what enables astronomers to determine the composition of a planet’s atmosphere.

* * *

Not everyone is as enthusiastic about MKIDs as Mazin. Bruce Macintosh, the principal investigator for the Gemini Planet Imager—an extreme adaptive optics system at the Gemini Observatory in Chile—is skeptical that MKIDs would replace other instruments, like spectrographs. “They don’t have very high spectral resolution,” he says.

And then there’s the difficulty in operating an MKID camera. All superconductors require very low temperatures. The MKID camera will operate below one degree Kelvin. Maintaining these low temperatures is complicated and expensive. Macintosh says there are existing instruments that accomplish pieces of what MKIDs do but don’t “involve mucking around with the liquid helium you need for superconducting.”

Markus Kasper, the European Southern Observatory project manager of the Spectro-Polarimetric High-contrast Exoplanet Research instrument—an extreme adaptive optics system at the Very Large Telescope in Chile—acknowledges some of the advantages of the new technology, but he also sees distinct limits.

Kasper points out that there are already certain kinds of CCDs that can count photons and that some CCDs, when used in combination with multiple filters, can measure the energy of photons.

Mazin acknowledges that MKIDs have some inherent limitations. But the MKID camera is a test model, starting small to help instrumentalists learn how it can best be used. In the meantime, Mazin notes, the technology is well suited to the camera’s current mission: offering an unprecedented look at dim objects.

While MKIDs do not have a high enough spectral resolution for studying bright objects, Mazin expects this capability will improve and says that, for now, “if you care about the faintest galaxies or this faint planet around a nearby star, then that low resolution is good enough.”

Similarly, although the MKID camera has only a 20,000-pixel array—dwarfed by the megapixel CCD cameras deployed on today’s big telescopes, let alone the gigapixel cameras on the next generation of observatories—Mazin expects MKID arrays to grow in size. But for now he says that “when you’re looking at a planet around a nearby star, all the action is happening right near the star. There’s no point in having a very big field of view, because what you’re interested in is happening in a very small spatial area.”

And, while CCDs can measure the wavelength of photons using sequential filters, Mazin says, in addition to losing some photons, “you pay a penalty in observation time, equivalent to the number of filters you use.”

The one thing you can’t beat about MKIDs is their speed. With most scientific CCDs, depending on their design, it can take anywhere from 30 seconds to several minutes to read out data. By contrast, data can be downloaded from an MKID array within half a millisecond.

Speed is critical in addressing one of the central problems of ground-based telescopes: atmospheric turbulence. Light is distorted as it passes through the atmosphere, and modern telescopes compensate by means of a technology called adaptive optics. A basic adaptive optics system uses a sophisticated camera to measure distortions and a mirror that can change its shape to remove them. The Keck Observatory, for instance, uses a deformable mirror that makes its telescopes 10 times sharper.

Today there is a new generation of devices—extreme adaptive optics—without which the search for Earth-like exoplanets would be impossible. In fact, it was a sophisticated cluster of extreme adaptive optics instruments on the Subaru telescope that brought the MKID Exoplanet Camera to Mauna Kea.

* * *

Subaru is the ideal place to perfect the use of MKIDs because, unique among the world’s big telescopes, it was designed as an open system to incorporate and test new technologies. Olivier Guyon, the MacArthur “Genius Award” winner who runs the Subaru telescope’s extreme adaptive optics program, says the MKID camera’s speed is a key trait that makes it well suited for obtaining images of dim exoplanets while correcting for atmospheric distortions.

“The atmosphere is being pushed by the wind, so those distortions are constantly changing,” Guyon says. “We have to go very fast.”

In other words, Subaru’s optics suite has one millisecond to tell the deformable mirror how to change shape to make the appropriate correction. “We don’t get a lot of light for measurements because we’re essentially shortening our measurement time,” says Guyon.

At the heart of the Subaru telescope’s extreme adaptive optics system is an instrument package called a pyramid wavefront sensor. But the sensor is equipped to react only to visible light, even as astronomers rely on near-infrared wavelengths to obtain images of dim objects. Since the MKIDs in the camera are fast, they can supplement the wavefront sensor, cleaning up the near-infrared image in real time.

Mazin’s plan is to use the MKID Exoplanet Camera (MEC) on Subaru to obtain images of exoplanets orbiting nearby stars that have been previously discovered through indirect methods. “We also tend to look at young stars, because the planets will still be hot from the heat of their formation,” he says. Jupiter, for instance, has cooled down to 200 degrees Kelvin some four billion years after our solar system formed. But, Mazin says, there could be younger gas giant planets elsewhere, between 700 and 1,400 degrees Kelvin, which would “glow like a light bulb in the near infrared.”

Right now, he adds, the best adaptive optics systems can image a planet about a million times fainter than its star. But an Earth-like exoplanet will be even dimmer. “What we hope to do with the MEC is to break through that contrast floor so we can start to see planets that are 10 million or 100 million times fainter,” Mazin says. “That opens up a huge discovery space of planets.”

Notice that he doesn’t mention small habitable exoplanets. That’s because no one really believes that Subaru’s extreme adaptive optics suite will ever see an Earth-size, rocky planet in reflected light. That’s beyond the power of even the 10-meter Keck telescope. But it may just be possible for another observatory currently under development. When the Thirty Meter Telescope (TMT) achieves first light sometime in the late 2020s it will produce images more than 12 times sharper than those from the Hubble Space Telescope. With that kind of power, astronomers will be able to study the supermassive black hole at the center of the Milky Way, the formation of distant galaxies, and—with the help of sensitive detectors such as MKIDs—Earth-like exoplanets.

Christophe Dumas, the head of operations at TMT, knows that this next generation of telescopes will need a new, more powerful quiver of instruments. That’s why Subaru’s open-system strategy is so important, he says.

With prototype instruments, like MEC, you need to do all sorts of experiments and bench-testing at the smaller, 8- to 10-meter-class telescopes “so that you understand what the challenges are and how to overcome them,” he says. “You have to do that to make your instrument better so you can be very efficient when you move it to a 30-meter-class telescope.”

Astronomer Doug Simons, the executive director of the Canada-France-Hawaii Telescope, Subaru’s neighbor on Mauna Kea, foresees a time when MKIDs could replace many astronomical instruments, including CCDs and spectrographs.

“With a very small package, they’re starting to push into a realm where they’re able to measure all the properties of every single photon,” he says. “That has nothing to do with the telescope. It’s all solid-state physics and modern-age electronics. I’m an instrumentalist by trade.... In a sense, we may have ‘solved’ instrumentation. And I don’t think that’s an overstatement, because there’s only a finite amount of information in each photon. So, if you’ve measured everything about that photon, I don’t know what more you do with it.”

Meanwhile, Mazin’s team continues to demonstrate that there are always opportunities for new applications. With an MKID, Mazin says, each photon gets time-tagged to about two millionths of a second so that it’s possible to know the precise moment it hits the array as measured against Coordinated Universal Time. (UTC is kept by atomic clocks in combination with measurements of the Earth’s rotation.) “The time tags open up a whole new domain of techniques,” Mazin says. His team recently submitted a paper on using photon arrival times to differentiate between photons that come from a star and those that come from its orbiting exoplanets—a method that can improve the MEC’s contrast even further.

For now, Mazin says, his lab has a monopoly on MKID research, because other scientists have been slow in embracing the new technology. That’s changing though, he says, pointing out that there are “now four groups in Europe who are making real, significant efforts in MKIDs, and I think eventually they’ll start fielding instruments.” Looking ahead, Mazin says, “MKIDs will have an increasingly large role to play in astronomic instrumentation in the next decade.” That’s when the true test will come: when humanity might finally glimpse another Earth circling a distant star.

Get the latest stories in your inbox every weekday.