Measuring Polarization	Modeling Geometries	Resolving Spectra

"New" stars, novae, have been known since antiquity, but it was not until the 1930s that the German astronomer Walter Baade and the Swiss astronomer Fritz Zwicky, both working at Mount Wilson Observatory in Southern California, coined the name supernova for the brightest of these temporary wonders.

So bright are supernovae, in fact, that they immediately seemed to offer a new way to measure vast cosmic distances. Objects whose brightness is not intrinsically similar do not make good standard candles, however, and it soon became apparent that, as a group, supernovae are not much alike.

Early on, astronomers classified two main types. The spectra of Type I supernovae lack hydrogen lines, while the spectra of Type II have these lines. But within types there is a great deal of variation, and after theorists learned more about the precursors and remnants of supernova explosions, the presence or absence of hydrogen lines seemed less than fundamental.

"Core collapse" supernovae begin as bloated, massive stars, which may be rich in hydrogen; after exploding they leave behind a neutron star or sometimes a black hole. Type II and Types Ib and Ic supernovae are all core-collapse supernovae and all follow this general scenario. The reason Types Ib and Ic lack hydrogen in their spectra is because the hydrogen blows away before the final explosion.

Type Ia supernovae explode by a very different mechanism. They begin as tiny white dwarf stars with no hydrogen at all. Typically made of carbon and oxygen, they pack about 60 percent of the sun's mass into an object roughly the size of Earth. If they gain enough additional mass -- most likely by accreting it from an orbiting companion star, which can be of almost any composition -- they explode as titanic thermonuclear bombs, leaving nothing behind.

Because the critical mass that triggers the thermonuclear blast is always the same, about 1.4 times the mass of our sun, Type Ia's -- unlike all other kinds of supernovae -- are not only extraordinarily bright but remarkably similar in their brightness. Here at last were the long-sought astronomical standard candles.

A new way to study supernovae

Until a few years ago, most of the information about how supernovae explode came from observing the way their spectra change as they increase to maximum brightness and then fade away, a progression known as a light curve.

But in the mid-1990s a young astronomer from China, working at the loneliest major observatory in North America -- the McDonald Observatory in the Davis Mountains of West Texas -- pioneered a new way to study supernovae, a method called spectropolarimetry.

Lifan Wang

Lifan Wang began his academic career as an undergraduate major in electrical engineering in China's University of Science and Technology. There he was persuaded to take up astronomy by astrophysicist Lizhi Fang, a man known worldwide for his outspoken advocacy of democracy. In June, 1989, demonstrations for democracy in China ended at Tiananmen Square. Professor Fang sought asylum in the U.S. embassy and later emigrated to the West.

Meanwhile his former student, Lifan Wang, was pursuing graduate studies at the European Southern Observatory, headquartered in Garching, Germany. It was at ESO that young Wang scored his first astronomical coup. He was a member of the group that in December of 1989 used the ESO's New Technology Telescope in the Atacama Desert of Chile to discover the stunning rings around supernova 1987A, a core-collapse supernova in the Large Magellanic Cloud, practically next door to our own Milky Way.

A few years later, in 1996, Wang secured a Hubble Fellowship from NASA. As a postdoctoral fellow at the University of Texas at Austin, he set out to settle a question about supernovae that could not be readily answered from studying their light curves and spectra: what shape are their explosions?

The vast majority of stars, including supernovae, are too far away to reveal their shape directly; to a telescope, a supernova is a point-source of light. But the way the light is polarized encodes a subtler indication of shape.

"I got interested in polarimetry starting at the University of Texas," says Wang. "We studied polarimetry with wide-band filters and a simple instrument that took us six months to calibrate. The telescope we used had a small aperture -- sometimes we were up three or four nights in a row, integrating the data -- but it did the job because it was very stable. And in the beginning, we had no competition."

Polarization is the orientation of the electric part of an electromagnetic wave. Light commonly becomes polarized by scattering; for example, light scattered from a horizontal surface is polarized horizontally and can be blocked or absorbed by vertically polarizing material, like Polaroid sun glasses.

Unlike reflected light, light from a supernova is randomly polarized by scattering from electrons. Since vertical polarization in one component of a beam of light is canceled if another component is polarized at right angles to it, a spherical star or explosion shows no net polarization at all -- because every randomly polarized component of the beam is canceled by another.

But if the explosion is not spherical -- if, for example, it looks oval from the point of view of the observer -- then the vertically polarized light in the long dimension will exceed the horizontally polarized light in the short dimension; the excess vertical polarization is not canceled. Thus an aspherical explosion shows net polarization.

Two ways to explode

By the time Wang had collected polarimetric data on some 15 different supernovae, he saw that there were two populations with markedly different degrees of polarization.

"Core-collapse supernovae were generally polarized, with about one percent net polarization," Wang says. "With Type Ia supernovae, on the other hand, there was no convincing evidence of any polarization -- at best, less than 0.3 percent polarization. This was unexpected and quite striking."

Unexpected, because a core-collapse supernova starts from a single massive star, which one might reasonably assume to be spherical. A Type Ia supernova, on the other hand, if the widely accepted model is correct, involves a white dwarf star, an orbiting companion, and an accretion disk -- an inherently asymmetric configuration.

	The first studies of supernova spectropolarimetry were conducted with the 2.1-meter Otto Struve Telescope at the University of Texas's McDonald Observatory.

Some unusual Type Ia supernovae found at the time did show asphericity, and there were hints that others might as well. Wang and Andy Howell -- then a graduate student at Texas, later with Berkeley Lab's Physics Division, and now continuing his supernova work at the University of Toronto -- found good evidence for asphericity in the Type Ia supernova 1999by, but being subluminous, that is, peculiarly dim, it was far from a normal specimen.

Another Type Ia, SN 1997bp, showed up to one percent polarization in individual spectral lines, but its spectrum as a whole indicated no polarization. The telescope Wang and his colleagues were using was not powerful enough to resolve their questions.

By contrast, the systematic polarization of Type II and other core-collapse supernovae was unequivocal. The University of Texas astronomers soon had enough polarimetric data to theorize about the way core-collapse supernovae explode. Their proposal: as the giant's core collapses into a neutron star or black hole, two powerful jets are formed that break through the star's overlying envelope to form a highly asymmetric explosion.

This persuasive model of core-collapse supernovae with energetic jets accounts for other odd phenomena sometimes observed in Type II or Type Ib and Ic supernovae that other models can't explain, including high-velocity ejected material and gamma-ray bursts.

Ironically, this coherent model for asymmetric core-collapse supernovae made the mystery of unpolarized normal Type Ia supernova even more acute. Wang doubted that astronomical events of such violence could truly be spherically symmetric.

Breakthrough

Late in 2000 Lifan Wang joined Berkeley Lab's Physics Division, a new base for continuing the supernova research he had begun years before at the European Southern Observatory. In September, 2001, Wang and his longtime ESO collaborator Dietrich Baade were using a spectropolarimetry instrument mounted on one of the four extremely stable 8.2-meter telescopes of the ESO's aptly named Very Large Telescope in Chile. They received word that the noted amateur astronomer Berto Monard in South Africa had reported a supernova in galaxy NGC 1448, about 55 million light years away in the southern constellation Horologium, the Clock.

SN 2001el was studied with the FORS1 spectropolarimeter mounted on ESO's VLT-Melipal telescope.

Wang and his colleagues trained their telescope on the supernova before it reached maximum brightness, and they continued observing it until the explosion faded away. Their data analysis was announced in the spring of 2003: they had found the first net polarization -- thus the first asymmetry -- in a normal Type Ia supernova.

Wang and his colleagues were able to show that at peak brightness the exploding star was slightly flattened, with one axis shorter by about 10 percent. By a week later, however, the visible explosion was virtually spherical.

Paradoxically, this discovery may strengthen the standing of Type Ia supernovae as dependable standard candles, by providing a partial explanation for the uncertainty that persists in measuring their brightness. "If all Type Ia supernovae are like this, it would account for a lot of the dispersion in brightness measurements," says Wang. "They may be even more uniform than we thought."

Moreover, although the asymmetry of SN 2001el persisted beyond maximum brightness, spherical symmetry began to dominate about a week after maximum. This suggests that information from the fading part of Type Ia supernovae light curves can be used to reduce uncertainties in the relation between their distance and brightness.

"It's not because the supernova is changing shape, but because we are seeing different layers of it," says Wang. As outer layers expand, they grow diffuse and eventually transparent, allowing the inner layers to become visible. "When it explodes, the outer part is aspherical, but as we see lower down, the dense inner core is spherical."

These clues to the way Type Ia supernovae burn as they explode promise to reveal much about the stellar systems that led to the explosion in the first place.

The Shapes of Exploding Stars, part 2

Additional information

• More about using polarimetry to study the shape of SN 2001el
• "Spectropolarimetry of SN 2001el in NGC 1448: asphericity of a normal Type Ia supernova," by Lifan Wang, Dietrich Baade, Peter Hoeflich, Alexei Khokhlov, J. Craig Wheeler, Daniel Kasen, Peter E. Nugent, Saul Perlmutter, Claes Fransson, and Peter Lundqvist, appeared in the Astrophysical Journal, 2003, vol 591, p 1110.