Upon listening to researchers from Lawrence Berkeley National Laboratory (Berkeley, CA) describe the Supernova Cosmology project, one is reminded of historical tales about Fifteenth, Sixteenth, and Seventeenth Century explorers and astronomers sailing ships into the unknown to see if the world was flat or round, and peering through primitive telescopes in defiance of the Inquisition to see whether the Earth or the Sun was at the center of the universe.
The Twenty-first Century version of this search for fundamental understanding of nature has extended in scope to much larger questions such as whether the universe is flat or round, and whether some great but unseen universal force is causing it to expand at an increasingly rapid rate. Expansion of the universe aside, however, the scope of our universal questions has certainly expanded far beyond the human limitations of even the most intrepid explorers and observers. Enter optoelectronics.
The Berkeley researchers are developing a CCD detector with the same pixelation as a human eye that also will have both the radiation hardness to survive space missions and a visual range extending from blue-ultraviolet to the near-infrared with a quantum efficiency about two orders of magnitude greater than human vision. Supernova cosmology research may not call for traveling beyond a terrestrial horizon or contemplating the relative motion of celestial bodies, but it does call for meticulously and rapidly collecting light from 10 billion-year-old exploding stars.
The detectors
Conventional astronomical CCDs are designed to be sensitive to the dim blue light emanating from many distant stars, which can be accomplished within depletion thicknesses on the order of one human hair, according to Michael Levi, co-principal investigator for the proposed supernova cosmology satellite. Light coming from 10 billion-year-old supernova across a universe whose rate of expansion is accelerating, however, gets red-shifted in wavelength by several orders of magnitude into the near-infrared, requiring much thicker detectors to achieve reasonable quantum efficiencies. So another member of the Berkeley team, Steve Holland, an electrical engineer in the physics division microsystems laboratory, has designed and begun to construct such devices based upon an existing photodiode that successfully handles challenging imaging needs and environments in particle physics (see Fig. 1).
The key element in the Berkeley CCD design, for which one patent has already been awarded to Holland, is a 300-µm-thick depletion region that provides a quantum efficiency on the order of 90% for collecting near-infrared radiation on a back-illuminated CCD. And while conventional 20-µm-thick devices only actually transfer charge across the top five or 10 microns, Holland's innovative design manages to transfer charge throughout the full 300-µm depth by applying a reverse-bias voltage.
The concept actually works because the thick CCDs are made from very pure and high resistivity (on the order of 10,000 ohm-cm) silicon material, while typical CCDs are fabricated in 20-ohm-cm material. In terms of purity, the phosphorus concentration in the silicon for the thick CCDs is on the order of 1011 per cc, whereas typical CCD material is on the order of 1014.
"For a given voltage, you have a much larger depletion depth in a high resistivity silicon wafer than you would have in a low resistivity wafer," Holland said. To maintain material purity during processing, he has customized gettering techniques from the semiconductor industry. "You are sensitive to states in energy band that are usually generated by metallic impurities, especially in a depletion region," he said. "And in a 300-µm-thick depletion region, it's very important to keep the dark current low."
Maintaining blue sensitivity in such thick devices required another innovation, which Holland described as creating a backside window. "The blue light gets strongly absorbed (within a tenth of a micron)," he said. "So we had to developed a thin backside contact (about 200 Å) that was optically transparent."
Mission specific requirements
In addition to providing high levels of quantum efficiency across a very broad wavelength range, the detectors for the supernova project also must observe a broad window of sky to catch a sufficient sample of supernova for the statistical needs of the project. A billion pixels, similar to the human eye according to Levi, will be required for this task.
Development of 2K x 2K working prototypes have been achieved so far using a donated large-area lithography tool from a closed down Intel plant in Oregon, Holland said. Those prototypes allowed the researchers to demonstrate that the concept could actually scale. And they are currently testing a 2K x 4K version.
"We went up a factor of 100 in area and now we've gone up another factor of 2," Holland said. "These are very large devices so you can't have any killer defects, pretty much in a whole wafer. It's really a yield issue."
The researchers also hope eventually to include the detectors in an orbiting supernova acceleration probe (SNAP) satellite (see Fig. 2). And preliminary tests indicate higher-than-normal radiation resistance, which Holland attributes to the p-channel CCD. "We use an n-type substrate, so we collect holes in our CCD, while all of the other CCDs that are n-channel collect electrons," he said. "Among CCD space people, it has been speculated for a long time, that p-channel CCDs would have better radiation resistance. And that seems to be what we're seeing as well."
For all of their advantages, the thick CCDs also have their drawbacks. "Especially for ground-based astronomy where you have very long exposures, we actually see more background tracks from background radiation than you would in a thin CCD," Holland said. "We are aware of that issue and are investigating solutions."
Grounded applications
While the Berkeley researchers focus on larger arrays and the proposed SNAP mission, similar technology is finding application by others in different areas. X-ray astronomy efforts at Lincoln Laboratory (Lexington, MA) and Marconi (London, England) also have developed deep-depletion CCDs on high-resistivity material at thicknesses on the order of 40 to 80 µm that do not require the near-infrared quantum efficiency of the SNAP project, Holland said. Berkeley patent rights for the photodiode predecessor to the thick CCD also have been licensed and found commercial use in nuclear medicine imaging applications.
Saul Perlmutter, the other co-principal investigator for the SNAP satellite, listed several potential application areas for the detector technology, beyond the proposed space probe. "There is obviously a lot of other science besides supernova that you can do on a telescope," he said. He mentioned process control in the manufacture of computer chips and biological imaging of cellular processes as areas where high quantum efficiency in the near infrared might make an important difference.
Perlmutter's primary interest of course is in probing the "fundamental nature of the energy that pervades all of empty space" and that seems to be causing the expansion of the universe to accelerate (see "Lighting a path to the past with standard candles").
"Now when you talk to the particle physics theorists, they say that they see this mystery as one of the most important questions, perhaps the most important question, that they know of today in fundamental physics," he said. "So the fact that this kind of chip gives us a tool to address these very mysterious and somewhat philosophical questions is very exciting."
Describing his own perspective as "parochial in the terms of technology development," Perlmutter also argued that even though this emerging detector technology may eventually prove useful in the semiconductor industry, the inspiration for it actually came from a passionate pursuit of fundamental science as opposed to a desire for more efficient "chip defect hunts."
"It's the kind of thing that you're more likely to do if you're out for a much more fundamental question that's driving people," he said.
Lighting a path to the past with standard candles
In 1998, astronomy researchers at Lawrence Berkeley National Laboratory and elsewhere observed that the universe, instead of contracting as everyone expected it to under the influence of gravity, was in fact expanding at an accelerating rate, according to Michael Levi, co-principal investigator for the proposed supernova astronomy probe (SNAP) satellite. "Things do not accelerate in nature unless something drives that acceleration," he said. And the source of the observed universal acceleration has yet to be determined.
"The current data that we have are fairly scarce based on using supernova as a cosmic yardstick to measure the expansion rate," he said. The data have been collected by observing about 100 relatively nearby type 1A supernovae. Astronomers call type 1A supernovae "standard candles," because the supernovae give off the same amount of light every time one blows up, Levi said. "So if you know how bright something is, then by measuring its intensity with an electronic sensor like a CCD, you can know how far away it is."
The light emitted from these very hot standard candles is basically blue when viewed in its rest frame. "The characteristic spectral lines get red-shifted as the photons come traveling towards us over the eons [from 10 billion-year-old supernovae], and they get red shifted by the amount that the universe expanded while they were in transit," Levi said. "They basically stretch with the universe, so we can plot as a function of distance how much the universe stretched. It's a very simple experiment."
So the Berkeley researchers have developed a thick CCD for absorbing the red light. And prototypes have already been placed on telescopes. "We have one actually in operation now at Kitt Peak," Levi said. One is being planned to go into the Keck Telescope on Mauna Kea. There is one on Mount Hamilton, Lick Observatory. Those are some examples. We're collecting data on them and starting to grow a user community."
Ultimately, the researchers hope to fly the detectors on a proposed supernova acceleration probe (SNAP) satellite, which would, if approved, significantly swell the data pool by detecting about 2000 supernovae a year over the course of a three-year mission. The projected launch date is in 2007 or 2008.