Figure 10.7. The Hubble diagram for high redshift supernovae The small departure from the dotted line in the upper panel is the evidence that we five in an accelerating universe. In the lower panel, the 45° dope. Which is just the inverse square laiw. has been removed. The points certainly lie above the downward curving line of lon£ dashes, which is the prediction for i2„ - I with no cosmological constant. Most of the points also lie above the dashed horizontal line which is the prediction for 11„ = 0.3, with no cosmological constant The Oneway to get up to the solid line (which is formally the best fit to the data) is to include the effects of acceleration. Points from both the high-z team and the supernova cosmology project are shown here_ The high-z team points are fewer, but have equal weight because of smaller uncertainties
Figure 10.7. The Hubble diagram for high redshift supernovae The small departure from the dotted line in the upper panel is the evidence that we five in an accelerating universe. In the lower panel, the 45° dope. Which is just the inverse square laiw. has been removed. The points certainly lie above the downward curving line of lon£ dashes, which is the prediction for i2„ - I with no cosmological constant. Most of the points also lie above the dashed horizontal line which is the prediction for 11„ = 0.3, with no cosmological constant The Oneway to get up to the solid line (which is formally the best fit to the data) is to include the effects of acceleration. Points from both the high-z team and the supernova cosmology project are shown here_ The high-z team points are fewer, but have equal weight because of smaller uncertainties scatter were just as helpful in telling something about cosmology as their 42.
And the something was, you needed A to match the data. Since there is an invisible contest between £2m, which slows cosmic expansion, and flA, which speeds expansion up, the supernova results provide information about the difference between the attractive effects of matter and the accelerating effects of dark energy. The supernova results showed that acceleration is winning now, stretching out the distance light has to travel from a supernova at redshift 0.5 to our telescopes. The supernova results measure Qm - £2A, and they showed that this quantity must be smaller than zero. You cannot do that without A, or something very much like it. It's a little like stepping on a scale and finding your weight is below zero—something beyond the usual gravitational attraction must be going on! So far, the supernova data are the only evidence that the universe is accelerating, and the only measurement that showrs the effects of A directly. As Sir Frank Dyson said of the gravitational bending of light, "I was myself a skeptic and expected a different result." Me, too.
The cosmological constant might have been Einstein's biggest blunder and part of Eddington's journey into the theoretical wilderness, but the evidence from supernovae shows that we need it, or something very much like it, to understand the world we live in. This is no longer a matter of esthetics or introspection or stubble from Occam's razor. We need to learn to live with A.
Of course, Brian Schmidt's horror made us take extra steps to be certain that the small extra dimming of distant supernovae was not due to some other effect. If somebody was going to find a flaw in this work, we thought it would be best if we did it ourselves. So we tried hard to see if we could showr our own result was wrong, or misguided, or if we had missed some important source of error that was not described by the statistics of the data points.
We knew it wasn't Malmquist bias. Malmquist bias selects the brightest objects near the limit of a survey. But we weren't seeing supernovae that wrere extra bright, we wrere seeing objects that were extra dim. But there is more than one way to go wrong.
We know that when we iook to redshift 0.5, we're looking hack ahout one-third of the way to the Big Bang, about 5 billion years. So the stars will all be 5 billion years younger. Does age make a difference to supernova properties?
We know that the universe has grown richer in heavy elements, partly through the action of all the supernovae that have blown up in the past 5 billion years. Does chemistry make a difference to supernova properties near and far?
And we know for sure that many astronomical investigations have come to a bad end by misunderstanding dust. Couldn't boring old dust, not acceleration, make the distant supernovae appear dim?
These are serious questions to which the answers are still incomplete. Our job now is to examine these possibilities to see if they have misled us into the temptation of ascribing to cosmology an effect that truly belongs to evolving stellar populations or changing chemical composition or dirt.
As for the ages of stars, we know that galaxies today have stellar citizens with distinct demographics. Elliptical galaxies have very little current star formation, so all the stars are old, like the population of an Arizona retirement community. In contrast, spiral and irregular galaxies often have very active star formation—this is more like Ann Arbor, a town full of boisterous young people as well as a quiet older population. Those galaxies have young stars, including massive stars that blow up as SN II in much less than 5 billion years. They also have a quiet population of old stars that putter around while the young stars live fast, die young, and leave a beautiful neutron star corpse. So different types of nearby galaxies provide places to study the effects of a young or old population of stars.
Interestingly, type la supernovae have been found in all types of galaxies. It is worth looking to see if the SN la in spirals, where there is recent star formation, differ from the SN la in ellipticals where there is not. That would provide a clue to whether looking back in time makes a difference in the brightness of the supernovae. From the Calan/Tololo data plus the CfA data, we have now built up a set of over 50 well-ol>served supernovae in nearby galaxies that lets us examine this question. Every month we observe more. We are going to find out.
At first glance, the news is bad. On average, the SN la found in elliptical galaxies are dimmer than the SN la found in nearby spirals. However, when you use the light-curve shape to correct the luminosity, as we do for both the nearby and the distant sample, Supernovae in ellipticals are indistinguishable from the Supernovae in spirals. This suggests that there may be a real difference in the stars that become Supernovae in spiral galaxies, and presumably in the distant younger galaxies we observe to measure A, but the correction methods we have developed are adequate to deal with this difference. By measuring the shape of the light curve, we iron out the age differences in the Supernovae from 5 billion years in the past.
Does chemistry affect the brightness of Supernovae, somehow making the Supernovae in distant galaxies dimmer? There are several ways to approach this problem. Theory is one path. Peter Höflich, Craig Wheeler, and Friedel Thielemann wrote a paper in 1998 to look into the theoretical possibilities.11 One prediction of supernova theory is that increasing the chemical abundances, as happens in galaxies over time, doesn't affect the spectrum or the overall light emission very much, except in the ultraviolet, where increased abundances are predicted to make SN la dimmer. This is the opposite of the effect we see, where the distant (and presumably slightly anemic) galaxies are dimmer than the nearby objects, which are the ones formed from enriched gas.
The chemical evolution from 5 billion years ago to today is not very extreme. In our galaxy, the chemical abundances 5 billion years ago at the site where the sun formed were precisely the solar abundances we see in the solar system today. Most chemical change in our galaxy and in other galaxies, took place in early violent episodes of star formation. Gas in our galaxy today, 5 billion years after the sun formed, is not much richer in heavy elements than the gas that formed the solar system.
Even so, it would be prudent to took for these effects. With the predictions of theory as a roadmap to action, we are now building up a sample of ultraviolet observations of nearby Supernovae, since that's the part of the spectrum where chemistry matters most. We will see if galaxies with different chemical abundances produce SN la with different ultraviolet light curves and colors, as predicted. This work isn't finished, but so far, there do not seem to be large differences.
We have also compared the spectra of our high z supernovae with the spectra of SN la observed nearby. The mighty Keck is amazingly good at obtaining spectra of the distant objects, using its immense collecting area to gather in the photons from distant SN la, and then sorting them out by wavelength. Alison Coil and Alex Filippenko led our group's effort to study the spectra of high-2: supernovae and to compare them with supernovae in the local neighborhood. The spectra of distant SN la are, within our ability to measure, just the same as the spectra of the nearby SN la going back to SN 1972E and SN 1937C."
The spectrum formed in the expanding atmosphere of an exploded white dwarf depends in a very complex way on the chemistry, velocities, and temperatures throughout the wrecked star. It is hard to imagine that exploding white dwarfs near and far are significantly different in light output, but have somehow conspired to make the spectra the same. Just because we can't imagine something doesn't mean it can't be true, but spectrum measurements test whether distant supernovae are distinctly different from nearby ones. If so, the cosmological interpretation of the supernova results would be suspect. This is a test SN la could have failed, but as far as we can see, they did not.
Dust is trickier. We know how to detect the presence of dust like the dust in our galaxy from the reddening it produces. Adam Riess worked that out in his Ph.D. thesis, and the Tololo crew did something equivalent. We made all our measurements of high-redshift supernovae in two colors specifically to overcome that weakness in the earliest SCP data. But clever theorists can invent dust that is unlike the dust in the Milky Way, and perhaps pixie dust like that really exists. A Harvard astronomy graduate student with a slight contrarian bent, Anthony Aguirre, worked this out. As a beginning student, Anthony had examined the possibility that the microwave background wasn't really from a hot Big Bang, hut
Figure 10.8. Spectra of supernovae. The supernovae observed at high redshift, SN I999ff and SN I999fv are, as rear as we can tell, identical with those seen nearby at similar ages. The spectra have been shifted back to the wavelengths you would observe if you were in the same galaxy as each of the supernovae. Courtesy of Alison Coif, Alex Filippenko, and the High-z supernova team
Rest Wavelength (A)
Figure 10.8. Spectra of supernovae. The supernovae observed at high redshift, SN I999ff and SN I999fv are, as rear as we can tell, identical with those seen nearby at similar ages. The spectra have been shifted back to the wavelengths you would observe if you were in the same galaxy as each of the supernovae. Courtesy of Alison Coif, Alex Filippenko, and the High-z supernova team might be thermal emission from solid particles. Again challenging orthodoxy, Anthony asked whether there could be pixie dust in the universe that dims the light of distant supernovae, but does not leave the fingerprint of reddening. To explain the high redshift supernova results this way, you need dust that dims distant supernovae by about 25 percent and that eludes our color measurements, is this possible?
Anthony knew that the effect that leads to reddening has its origin in the sizes of interstellar dust panicles. When the particle size is comparable to the wavelength of the light wraves, you get reddening. Interstellar dust is not like the big balls of dog hair and sloughed-off human skin that accumulate behind your couch. Interstellar dust is a very fine submicroscopic haze made of carbon and silicon that would be invisible to an ordinary microscope. It is a kind of soot and sand that even the finest white glove test would not reveal. Although that's the kind of dust we know about, Anthony suggested there could be another kind of dust that we don't know about, which is revealed only in the supernova data. These would be big dust grains, so large that they affect all colors almost equally. Big dust grains could lead to absorption without much reddening.
Scientific imagination has to obey reason and cannot violate observed limits. Anthony had to think how to make the dust smoothly distributed throughout the huge regions of space between the galaxies. Otherwise, the light from a supernova would sometimes encounter pixie dust and sometimes not. This would lead to increased scatter in the high-z supernovae, beyond what we observed. So he invented a story, which was not too crazy, for a way to form the dust, make sure it had big grains but no small ones, and expel it from galaxies. He had to be careful not to use more of the element carbon than stars could produce. Anthony found that there was a possibility that this specially constructed dust might exist, and showed that it could conceivably account for the results we were observing in the high-redshift supernovae. While there was no other evidence for this pixie dust, it was not impossible.
If this were a debate, you could ask, in a rhetorical flourish, pounding the podium: "I ask you my friends and fellow coun trymen, which is more likely, a form of intergalactic dust that has hitherto eluded detection or creating out of whole cloth a mysterious new component of the universe—so-called dark energy that purportedly dominates cosmic expansion? Must wre repeat Einstein's notorious stumble? Can we not learn from the past? Must we stride confidently into the abyss of error?"
Luckily, science does not consist of public exhortation. I encouraged Anthony in his work. Of course, if he was right, our measurements were about my least favorite subject, dust, rather than detecting a dramatic newr component of the universe that had been hidden from observers for 80 years. But since we're trying to get at the truth» we should test every link in the chain of inference. Though debate can be amusing, to test Anthony's idea we needed a more sensitive measurement for the existence of his pixie dust.
First, Anthony improved his earlier prediction of exactly what this dust would do. On closer inspection, it was not perfectly gray, dimming all wavelengths equally, but slightly pink, absorbing blue light a little more than red light. To look for this, we have been observing distant supernovae over a larger span of wavelengths, where these subtle effects should show up. In practice, this means getting observations from the blue end of the emitted spectrum out to the infrared wavelengths that lie beyond the range of human vision. So far, we have one well-observed case, SN 1999Q, and there we see no sign of pink dust. In the year 2000, we observed a number of supernovae as carefully as we could over a wide wavelength range, to see if we can put a stake through the heart of gray dust. We put the data in the capable hands of my graduate student, Saurabh Jha. We shall see what he comes up with.
So we have some evidence that the age of the stars, their chemistry, and pixie dust are not the cause of the effect we see: the distant supernovae haven't been shown to be faint for one of these reasons. But we have a stronger way to distinguish a genuine cosmological effect from a misleading systematic effect in the supernovae. If we imagine, for a moment, that the apparent faintness of distant supernovae with redshift z = 0.5 is due to cosmology with a dominant dark energy causing acceleration, we can predict what will happen as wre look even further into the past_
The brightness we measure for a supernova depends on the outcome of a battle between the accelerating effect of £2A and the decelerating effect of £2™. What we see is that dark energy has been winning this tug-of-war in the last 5 billion years or so while the light from supernovae at z = 0.5 has been on its way to us. But what about even more distant supernovae?
if we look into the past of the expanding universe, each chunk of it would have been smaller in the past. Imagine a region, say 500 million light-years on a side, that has stretched out from a smaller volume in the past. The amount of matter in such a big chunk of the universe has not changed appreciably over that time—that's too big a region to have been affected much by the individual motions of galaxies or the growth of structure. So if you look back to redshift of one, each of the sides of a cube that expands along with the universe was smaller by a factor of two back then. If you have the same amount of mass in a smaller volume, that means the density was up by a factor of 2\ a factor of eight, so we would be looking back to a time when the universe was eight times denser.
When you look into the past, you see a denser universe. As you look further back, gets more important compared to £2A. Einstein guarantees that if the total £2 is 1, it stays 1, but the balance between £2m, tugging to slow things down, and £2A, urging the universe to accelerate, ought to shift as you look deeper into the past. The distant past would be dominated by dark matter, not by dark energy.
If you think of the universe starting out from the Big Bang, the first several billion years would be sluggish years of slowing down due to gravity from dark matter, but then, as matter thins out and its density declines, the balance would shift. Dark energy, negligible at first, is destined to dominate cosmic evolution. Slow and steady wins the race. There will come a time, which depends on the precise values of £2™ and £2A, where deceleration due to dark matter loses its grip and acceleration begins.
What makes this story interesting is that wre can already see back to the time when the brakes came off and the gas pedal was mashed to the floor. For reasonable values of £2m and £2* that are consistent with the supernova data we had in 1998, such as Qm =
0.3 and £2A = 0.7, the coasting point that marks the transition from a decelerating universe to an accclcrating one is at a redshift of about 0.7. This is distant, but not more distant than observations we've already made. As wc look further into the past, wc should see less acceleration from dark energy and an increased influence of dark matter slowing things down. And the redshift where this happens is not out of reach.
All the other effects, such as the age of the stellar population, the chemical composition of stars, and absorption by gray dust would increase with redshift. After all, if the ages of the stars matter, looking to bigger redshift means you arc looking at younger stars. If composition matters, as you look farther back, you should see even lower abundances. And if pink pixie dust pervades the universe, you will traverse more of it looking back to higher redshift.
This makes a definite prediction—if the dimming wc see is due to cosmology, then if we look far enough back, back past the coasting point, the extra dimming will grow smaller, and then shift sign to produce extra brightening. On the either hand, if other effects of age, composition, or dust are the cause of distant supernovae appearing dim, then you'd expect more distant supernovae to be extra dim. So we have a way to tell if we've been fcxiled or whether the cffcct we see really is due to the history of cosmic expansion. Look at higher redshift.
To observe the signature of dark energy, wc need to find and measure supernovae out beyond a redshift of one. If they turn out to be extra dim, we're wrong and something that is not cosmology is making the supernovae dim. If they are brighter than you'd otherwise expect, this means we're on the right track. To find out, we needed to shift our attention from z = 0.5 to z = 1 and beyond.
By 1998, we were out on a limb, and we had the means to saw it off ourselves. Better us than somebody else. Worse than the risk of failure was the creeping sense of investment. While everybody on the high-zteam was uneasy at first about a claim that A was real, as our own evidence built up, and the data from the other team gave the same result, we began to get comfortable with A and then getting it bight
Was this article helpful?