On a clear night, away from city lights, you might see a beautiful structure arched across the sky: our home galaxy, the Milky Way. Since ancient times, humans have marveled at the dark dust clouds silhouetted against the milky background. Just four centuries ago Galileo pointed his telescope at the heavens and found that the milk was the blended light of countless stars.

The architecture of the Milky Way has just been revised again. Using an entirely new type of telescope, we and our collaborators have discovered colossal structures that tower over the galactic center and extend tens of thousands of light-years into space. These luminous lobes have gone unnoticed for so long because they glow brightest in gamma rays, which cannot pass through our atmosphere.

We do not yet know what is creating these Fermi bubbles, as we have called them. But they appear to be driven by mysterious processes happening deep within the center of our Milky Way galaxy—a chaotic region where a supermassive black hole churns whirlpools of hot gas, while violent supernovae bloom like daffodils out of the rich soil of stellar nurseries.

Like many unexpected discoveries, we found the Fermi bubbles nearly by accident. But now that we have found them, we have begun to meticulously map out their features. Our giant bubbles of the Milky Way promise to reveal deep secrets about the structure and history of our galaxy.

The Surprise Discovery
The first hint that something was amiss in the inner galaxy came not from gamma rays but from microwaves. The year was 2003, and I (Finkbeiner) was trying to better understand the origin of the universe using data from the Wilkinson Microwave Anisotropy Probe (WMAP), at the time the latest, greatest cosmology satellite. I was a postdoctoral fellow at Princeton University, studying how nearby interstellar dust obscured the signal from WMAP's intended target—microwaves from the dim afterglow of the big bang. The dust is interesting in its own right, but to a cosmologist it is like smudges on a windshield, something to be modeled and subtracted from the data.

The dust was not the only thing I had to remove. Since astronomers are forced to observe the cosmos from inside the Milky Way, I also had to subtract the microwave signals created by energetic particles such as electrons that fly through the galaxy. In 2003 astronomers already had a fairly sophisticated understanding of these signals, but something did not fit. I could model most of the galactic emission, but when I tried to subtract it in the inner part of the galaxy, there was always something left over. I named this leftover signal the “microwave haze.”

This mysterious signal coming from the center of the galaxy had no known explanation, but astronomers quickly came up with ideas. The most exciting possibility was that the haze was evidence of hidden dark matter. No one knows what dark matter is, only that it interacts with ordinary matter through gravity. Scientists expect that gravity will pull dark matter toward the center of the galaxy. In the dense cloud of dark matter in the Milky Way's core, dark matter particles will occasionally collide, perhaps creating an electron and a positron in the process.

Even if we cannot see dark matter, we should be able to see these particles. As they twist and turn through the galactic center's tangled knot of magnetic fields, they should emit what is called synchrotron radiation—the luminous exhaust of charged particles forced to make a turn.

The microwave haze we were seeing could have been an artifact of synchrotron radiation generated by dark matter. But how were we to tell for sure? The very same electrons that produce synchrotron microwaves would have an additional consequence: they should also be colliding with existing photons and accelerating those photons up to extremely high energies in a process called inverse Compton scattering.

Soon a consensus developed: if the microwave haze was being caused by high-energy electrons—perhaps as a consequence of dark matter annihilation—then we should also be able to find high-energy gamma rays. We turned to the Fermi Gamma-ray Space Telescope, which was designed to study gamma rays in space.

Data from the Fermi satellite was released to the public on August 24, 2009. By this time I had become a professor, and my then postdoctoral fellow Gregory Dobler and I rushed to make our first gamma-ray maps of the galaxy. After a few long days and nights, we found a hazy excess of gamma rays in the inner galaxy that appeared to match the microwave haze. We and our collaborators quickly submitted a paper arguing that the signals were related. We asserted that they are both probably caused by a high-energy population of electrons in the center of the galaxy, but we did not speculate about the source of the electrons.

The next shoe took a bit longer to drop. In October 2009 I was in my office remaking some figures in our first paper with newly released Fermi data. I had noticed that the original gamma-ray data showed faint edges—clear borders where the signal dropped off precipitously. In astronomy, sharp features usually come from transient events. For example, a supernova may send out a shock wave that appears as a distinct edge in our telescopes. In time, sharp features tend to smooth out and fade away.

If dark matter were causing the gamma-ray signal, then the drop-off should have been smooth—fading gently as you looked farther away from the galactic center—because dark matter annihilation would have been going on for billions of years. Any sharp edges would have dissipated long ago.

In the first batch of Fermi data, the edges had looked so ratty that we just chalked them up to noise in the signal and ignored them. Now they were appearing in the new data again, and I started to wonder. I showed them to my then graduate students Meng Su and Tracy Slatyer, who agreed that they were real. Then Su really jumped in and started to work—I think almost continuously without sleep—on deriving the exact shape of the edges. Within a matter of days we totally changed our opinion about what was in the data. Dark matter was out. Bubbles were in. In May 2010 Su, Slatyer and I submitted a paper to the Astrophysical Journal describing the structures and naming them “Fermi bubbles” in honor of the Fermi telescope.

Bubble Makers
Even though nobody had expected to find bubbles made of high-energy cosmic rays jutting tens of thousands of light-years above the Milky Way, perhaps it should not have been that shocking.

Many other galaxies have bubbles, too. We can see them in x-rays and radio waves, and if we had better gamma-ray telescopes we would probably find them shining in gamma rays as well.

We understand the processes that create the bubbles in many of these other galaxies. In some cases, the bubbles trace their origin to a gigantic black hole—often with the mass of billions of suns—that anchors the galaxy's center. As material from the galaxy falls toward the black hole, it begins to spin like the water draining out of a bathtub. This whirlpool of hot gas and dust creates intense magnetic fields that power jets of radiation and cosmic-ray particles that may inflate the bubbles.

We know that the Milky Way galaxy also has a supermassive black hole at its center, but we have never observed a strong jet of intense radiation streaming out of its core. (If a jet exists, it is not pointed our way, and thank goodness for that.) So we do not have direct evidence that this process is inflating the Fermi bubbles.

On the other hand, a large gas cloud—the Magellanic stream—sits high above the galactic center. If a jet of radiation were pointed there, it would temporarily strip electrons free from atoms in the cloud. As the electrons and ions came back together, they would produce so-called recombination radiation.

Astronomers have found exactly this. Perhaps there was an intense episode of accretion onto the Milky Way's central black hole about a million years ago—accretion that generated high-energy jets and ultraviolet radiation that knocked the electrons around in the Magellanic stream. This event could have also created the Fermi bubbles.

Alternatively, galaxies such as the nearby M42 galaxy have bubbles that are by-products of intense star formation near their centers.* In a stellar nursery, stars form in many different sizes. The more massive a star is, the faster it burns its nuclear fuel. When the fuel runs low, the star's core collapses and releases an enormous amount of energy that rips off the outer layers of the star in a supernova explosion, leaving a neutron star or black hole behind. These supernovae create a wind of particles that can inflate bubbles around a galactic center.

We know that the center of the Milky Way has also been a region of intense star formation. Several thousand stars around the central black hole are only about six million years old—mere toddlers in cosmic time. Yet if extremely massive stars also formed in this same stellar nursery, six million years would be long enough for them to have already exploded as supernovae. These supernovae would have driven a wind of hot gas out from the galactic center—a wind that might have been powerful enough to inflate the bubbles.

Next Steps
As we see, the story of the Fermi bubbles is wound tightly with the history and evolution of the Milky Way. The bubbles can also teach us about the physics of how black holes pull in nearby matter and how high-energy cosmic rays interact with interstellar gas. Although structures like the Fermi bubbles exist in other galaxies, the bubbles allow us to study these systems up close.

To this end, we are trying to observe the bubbles using the entire electromagnetic spectrum. One of the most amazing things about the bubbles is that they are large and luminous in gamma rays but nearly invisible in other frequencies. We hope that new data from the Planck spacecraft, which has mapped microwave radiation across the entire sky, will provide important clues. We are also attempting to map the bubbles in x-rays, although we are limited by current technology. The bubbles are giant structures that tower over the galaxy, but nearly all x-ray satellites have a narrow field of view. The challenge is akin to mapping a mountain range while peering through a paper towel tube.

It took three centuries after Galileo's discovery that the Milky Way is made of stars for astronomers to understand that our galaxy is just one of many billions of galaxies spread throughout the cosmos. With any luck, we will come to understand the true significance of the Fermi bubbles in less time than that.

*Erratum (9/3/14): Due to an editing error in this sentence, M42 is referred to as a nearby galaxy; it is a nebula. The authors intended to use for their example M82, which is a galaxy.