The oldest, farthest type Ia supernova was a lucky catch
Working together, NERSC and the Hubble Space Telescope set an astronomical record
Supernova 1997ff was found the first time by a deliberate search of the Hubble Deep Field-then found again three weeks later, this time by happy accident.
Berkeley Lab astrophysicist Peter Nugent, working with Adam Riess of the Space
Telescope Science Institute, used an IBM SP supercomputer at NERSC to analyze data
from an exploding star that had been caught once on purpose and twice by accident by
NASA's Hubble Space Telescope.
Nugent's analysis confirmed that SN 1997ff, with a redshift of 1.7, is the oldest and most
distant Type Ia supernova ever seen. Riess and Nugent presented their findings at a
press conference held at NASA headquarters in Washington on April 2, where they
discussed the importance of their discovery for cosmology.
"This supernova is consistent with the cosmological model of an accelerating universe, a
universe mostly filled with dark energy," Nugent says. "It argues against the notion that
observations of distant Type Ia supernovae may be systematically distorted by
intervening gray dust or the chemical evolution of the universe."
Moreover, says Nugent, "the supernova is so ancient that it allows us to glimpse an era
when matter in the universe was still relatively dense and expansion was still slowing
under the influence of gravity. More recently the dark energy has begun to predominate
and expansion has started to speed up."
The farthest candle
Two international groups of astronomers
and physicists-the Supernova Cosmology
Project, headquartered at Berkeley Lab
and led by Saul Perlmutter of the Physics
Department, and the High-Z Supernova
Search Team, led by Brian P. Schmidt at
the Australian National
University-discovered the accelerating
expansion of the universe by using Type Ia
supernovae as "standard candles" to
measure cosmological parameters.
Type Ia spectra and light curves (their
rising and falling brightness over time) are
all nearly alike, and they are bright enough
to be seen at very great distances. With a
redshift (or z) of about 1.7, says Nugent, "supernova 1997ff is some 11.3 billion years
old, much older-and much fainter-than the previous record of z equal 1.2, which
corresponds to an age of about 9.8 billion years old."
He adds that a supernova at redshift 1.7 "is too far away to have been visible from the
surface of the Earth. Only a space-based telescope could have found it."
Faint as it is, SN 1997ff is actually brighter than its extreme redshift would suggest. "This
drives a stake through the heart of alternatives to the accelerating universe," said
University of Chicago cosmologist Michael Turner at the NASA press conference.
Because the theory of the accelerating universe is based on observations that distant
supernovae are dimmer and thus farther away than their redshifts might otherwise
suggest, some critics had suggested that maybe Type Ia supernovae in the early
universe were just dimmer to begin with, or that some form of "gray dust," like a neutral
density filter on a camera, filters their light and only makes them seem dim.
But SN 1997ff dates from a time before acceleration, so early that the expansion of the
universe was still slowing under the influence of gravity. Instead of being dimmer than
expected, it is brighter, an effect that rules out both gray dust filters and the inherent
dimness of the most ancient supernovae.
The one that almost got away
SN 1997ff was first found, on purpose, by Ron Gilliland of the Space Telescope Science
Institute and Mark Phillips of the Carnegie Institute of Washington, during the last week
of December, 1997. Gilliland and Phillips turned the Hubble Space Telescope on the same
patch of sky recorded in the renowned Hubble Deep Field of typical galaxies, looking for
bright spots which, after spurious or doubtful signals had been rigorously eliminated,
might prove to be supernovae. They found two good candidates.
Gilliland and Phillips asked Nugent to help them determine what these discoveries implied
for the rate at which high-redshift supernovae might occur in the universe as a whole.
The three published a report in 1999 suggesting that one of their two candidates, SN
1997ff, was probably a Type Ia with a redshift greater than z = 1.32. Because it had
been observed in only one range of frequencies, however, the uncertainties were too
great to use the supernova for cosmological estimates.
At high redshifts, much of an astronomical
object's characteristic spectrum is shifted into
the infrared. Without additional infrared
observations, no useful cosmological
information could be derived from SN 1997ff,
nor could its type be positively identified. It
seemed unlikely that anyone had made such
Enter serendipity. Only 25 days after the initial
observation, Rodger Thompson of the
University of Arizona had begun doing tests of
a small portion of the Hubble Deep Field with
NICMOS, an instrument aboard the space
telescope that makes images in the near
infrared. Although Thompson had not been
looking for supernovae, many of his images
accidentally included SN 1997ff and its host
"Twenty-five days later may seem like a long
time, but highly redshifted objects are moving away from us so fast that time dilation is
large," Nugent remarks. "At a redshift of 1.7, three and a half weeks in our frame of
reference is only about nine days of elapsed time for the supernova itself."
Six months later another set of infrared images of the same region, made by Mark
Dickinson of the Space Telescope Science Institute, caught the now greatly faded
supernova and its host galaxy once again. Once more, luck had provided a missing piece
of the puzzle: by digitally subtracting the new image of the host galaxy from images
made when the supernova was bright, Nugent proposed to Dickinson, much of the
remaining uncertainty about the supernova and its host could be eliminated.
He had no takers until July of last year, however, when Adam Riess, independently
intrigued by the accumulating data, queried Nugent about doing cosmology on an
unnamed supernova at a redshift "around 1.65." There was only one such supernova;
soon Riess and Nugent were collaborating.
"Adam had the monumental task of reducing the observed NICMOS infrared data," said
Nugent, "while I concentrated on comparing the reduced data to known supernovae and
various sets of cosmological parameters." By painstakingly eliminating other possibilities,
they determined that SN 1997ff was almost certainly a Type Ia supernova at a redshift
of 1.7, first seen eight days after it exploded.
"Now we could do the cosmology," Nugent says.
If the luck holds, what comes next?
SN 1997ff supports the model of a universe consisting of about one third matter and
ordinary energy and about two thirds "dark energy," which acts to overcome gravity.
Most important, says Nugent, SN 1997ff proves that while the most distant supernova
currently cannot be seen from ground telescopes, they can be observed from space-and
they can provide vital information about the most basic cosmological questions, including,
perhaps, the nature of the dark energy itself.
"The results from SN 1997ff are one of the best arguments for the SNAP satellite,"
Nugent says. SNAP- for SuperNova/ Acceleration Probe- is a multi-institution,
multi-agency proposal led by Saul Perlmutter and Michael Levi of the Lab's Physics
Division. It would fly a 2-meter telescope and employ a CCD camera far larger and more
sensitive than any previous astronomical imager, especially in the near infrared.
The discovery of SN 1997ff underscores the role of luck in science - although as Michael
Turner paraphrased Louis Pasteur, "Serendipity favors the prepared mind"-it also
illustrates the value of cooperation among agencies such as NASA and the Department of
Not to mention among rival groups: Peter Nugent is a member of the Supernova
Cosmology Project, and Adam Riess is a member of the High-Z Supernova Search Team.
To see the recorded press conference, or for more information about SN 1997ff, visit