Section 4: Mapping the Expansion with Exploding Stars
The galaxies we see with telescopes are massive objects, and the presence of mass in the universe should slow its expansion. Starting in the 1950s, astronomers saw their task in studying cosmology as measuring the current rate of cosmic expansion (the Hubble constant), and the rate at which gravity was slowing down that expansion (the deceleration.) With that information, they would be able to measure the age of the universe and predict its future. The deceleration would show up as a small but real departure from Hubble's Lbaw when the Hubble diagram is extended to very large distances, in the order of a few billion light-years. That's roughly 1,000 times farther away than the galaxies Hubble studied in the 1920s. The inverse square law tells us that galaxies that are 1,000 times farther away are (1/1,000)2 times as bright. That's a million times dimmer, and it took several decades of technical advances in telescopes, detectors, and astronomical methods to compensate for this giant quantitative difference. Today's telescopes are 16 times bigger, our light detectors are 100 times as efficient, but we need to use much brighter stars than the Cepheids to look back far enough to see the effects of a changing rate of cosmic expansion.
Figure 7: The spectrum of a Type Ia supernova, shown here, distinguishes it from other supernova types.
Source: Recreated from High-Z Supernova Team data, courtesy of Robert Kirshner. More info
Fortunately, nature has provided a brighter alternative. Some stars die a violent death in a blast of thermonuclear energy as a supernovae (SN) explosion. For a few weeks, a single star shines as brightly as 4 billion stars like the sun. These thermonuclear supernovae, known as Type Ia supernovae, can be recognized from their spectra and, with some effort, used as standard candles for measuring cosmic distances.
The supernova challenge
Using Type Ia supernovae to measure cosmic acceleration or deceleration is no easy task. First, we have to find the supernovae; then we have to determine their distances and redshifts; and, finally, we have to find enough of them to make our measurement meaningful.
Although SN Ia are bright enough to see over distances of several billion light-years to detect the effects of cosmic acceleration or deceleration, they have one serious drawback: They are rare. Type Ia supernovae explosions take place about once per century in a typical galaxy. We cannot simply select the galaxies whose distances we wish to know. Instead, we have to inspect many galaxies to find the supernovae whose light has just arrived in the past week. Those are the ones we get to measure.
Figure 8: Light curve shape standardization.
Source: © Robert Kirshner. More info
How many galaxies do we need to search? Given that SN Ia appear about once per century per galaxy, and a year has about 50 weeks, we must search 5,000 galaxies to have a good chance of finding one within a week of its maximum light. This is a good job for a "computer"—but not the human kind.
Further, SN Ias don't possess the direct relation between brightness and vibration period exhibited by Cepheid variables. Fortunately, though, they have something similar. The light from an extra-bright SN Ia increases and decreases more slowly over time than that from a dimmer version. Careful study of the light curve can reveal which supernovae are extra bright and which are not so bright. Analyzing the light curve reduces errors in the distance to a single SN Ia to about 10 percent. This makes SN Ia plausible candidates for measuring the effect of cosmic acceleration or deceleration with a modest-sized sample, provided we look at SN Ia that are at a large enough distance that the effect of expansion, speeding up or slowing down, makes a 10 percent difference in the distance.
Finally, mathematical analysis shows that the uncertainty in the average value of something we're trying to measure becomes smaller as the square root of the number of times we repeat the measurement. In this case, a well-measured shape for the light curve of a single Type Ia supernova gave an uncertainty in a single distance of 10 percent. The effect due to cosmic deceleration was also expected to be about 10 percent for supernovae at a distance of a few billion light-years. So, an astronomer who wanted to make the uncertainty in the mean significantly smaller than the signal that would show evidence for cosmic deceleration would need at least nine objects to push the error down to about 3 percent (10 percent/) and about 100 to push the uncertainty in the mean down to 1 percent (10 percent/). Somewhere in that range, where the ratio of the expected signal to the uncertainty in the measurement is a factor of 3 to 10, astronomers might begin to believe that they have really measured something.
Figure 9: This Pan-STARRS telescope will find thousands of supernovae in the next decade.
Source: © Pan-STARRS, University of Hawaii. More info
If we need to search 5,000 galaxies to find one supernova, we'll need independent searches of 50,000 galaxies to find 10 and 500,000 galaxies to find 100. This became practical in the 1990s, when large digital cameras with tens of megapixels were mounted on 4-meter telescopes. In the coming decades, as the detector arrays grow, the hunt will reel in dozens of supernovae every night. Of course, this is still a small fraction of all the supernovae; we estimate that 30 explode somewhere in the universe every second. We have a long way to go before we will see them all.
The Astonishing Discovery of Cosmic Acceleration
Figure 10: Adding high-redshift supernovae to the Hubble diagram revealed the effects of dark energy.
Source: © High-Z Supernova Search Team. More info
Starting in the mid-1990s, the pieces were in place for a direct approach to measuring cosmic deceleration; and by the end of the decade, two international teams were ready to publish their results. In September 1998, the High-Z Supernova Team published measurements of 16 distant and 34 nearby supernovae. To their surprise, the data pointed firmly toward acceleration. Instead of averaging a little brighter than expected for their redshift, this sample showed that the distant supernovae were about 25 percent dimmer than expected. This meant that the expansion of the universe was speeding up. Nine months later, the Supernova Cosmology Project reported on a larger sample of 42 distant supernovae. It agreed with the High-Z Team's results. It looked as if the universe was not decelerating, but accelerating. What could cause this? One possibility was that old bugaboo of cosmological theory, the cosmological constant.
Scientists are naturally suspicious of measurements and interpretations in new areas. However, subsequent work at large and small telescopes has augmented the samples of low redshift and high redshift supernovae into the hundreds. There's no escaping it now: The evidence from SN Ia shows that the universe is speeding up.