In a modest, two-story office building in the heart of Silicon Valley, a series of experiments that could change forever how scientists hunt for new materials is taking place. In one lab, a robotic arm sealed within a tabletop-sized vacuum chamber is intent on synthesizing electronic compounds. The robot selects a ceramic wafer from what looks like a small stack of compact discs and draws the wafer to a central chamber a foot away. A beam of electrons blasts the disc, blowing ceramic vapor against tiny squares on a shiny silicon wafer. Shutters inside the vacuum chamber click open and closed to control precisely how much of the vapor hits each square. The robot puts the first ceramic disc away and selects another. The process is repeated until the silvery wafer is coated with dark squares, each a potential new high-temperature superconductor.
Down the hall, another diminutive robotic arm whisks back and forth across a benchtop. The arm’s needle-shaped tip squirts a few drops into dozens of wells positioned in a plastic tray the size of a paperback book. Each well holds a different mix of chemicals and, before long, each will contain a type of plastic never made before. One of those novel polymers could become a choice material for high-strength structures, electrical insulation, or biological implants.
Welcome to the headquarters of a startup called Symyx-and perhaps to the future of materials prospecting. In this new strategy, borrowed from chemistry and biotechnology, automated machines rapidly synthesize and sift through anywhere from dozens to tens of thousands of novel materials in hopes of hitting pay dirt. It is a big change from how materials scientists have traditionally worked, following precise recipes-and occasional spurts of inspiration-to mix chemicals in test tubes tediously cooking up new materials one at a time.
Taking a Clue from Nature
Though weinberg and his colleagues at Symyx are the first to try to commercially apply combinatorial techniques to materials research, they didn’t invent the process. In fact, they were beaten out by a few billion years by a very creative innovator: Evolution. Cells have the ability to create a wide variety of molecules based on a limited number of building blocks and then select the ones that function best. In this familiar evolutionary process, cells create an enormous variety of DNA and protein molecules by arranging common building blocks in a different order. Natural selection does the rest.
Beginning in the early 1980s, researchers began imitating nature’s example. They started creating collections of peptides-short proteins that can bind to cell receptors and thereby regulate cell function. Just how well this regulation takes place depends on how tightly a peptide binds to a receptor, which itself depends on getting just the right sequence of peptide building blocks, amino acids. Researchers invented several methods that made it possible to arrange amino acids in different combinations and track the products they made. They found that they could easily create thousands of peptides in nothing flat. By testing these compounds for activity in cells, researchers could quickly home in on the most chemically active peptide and work out its structure.
These early successes didn’t win many converts among those who design new therapeutic drugs for a living. “There was enormous resistance from medicinal chemists in the beginning,” says Joseph Hogan, founder and chief scientific officer of ArQule-a Medford, Mass.-based combinatorial startup. “They felt it was completely inelegant and ugly” compared with the traditional approach of rationally designing and then painstakingly synthesizing compounds.
The approach also faced practical limitations. Because enzymes in the stomach break down peptides, most researchers considered them poor drugs. But the idea was in the air, and before long, new research teams showed that the basic strategy could go beyond peptides and turn out small organic compounds similar to those that make up most drugs.
By the beginning of the 1990s the craze for high-speed chemistry was sweeping through the pharmaceutical industry. Startups sprang to life to commercialize combinatorial know-how. Flush with hundreds of millions of dollars from investors, these companies set about creating libraries of potential drugs with as many compounds as big pharmaceutical companies had hoarded on their stockroom shelves during the past 100 years. Not to be left out, Big Pharma companies, such as Glaxo Wellcome and Merck, leaped into the fray starting their own combinatorial research efforts and striking deals with combinatorial chemistry startups. “In the mid-1980s, traditionalists were laughing at the idea of the combinatorial synthesis of drugs,” says Weinberg. “But they’re not laughing now.”
Back to the Future
Schultz is betting that for materials science, the present is like the late 1980s all over again. In 1995, Schultz-a Berkeley chemist who holds a joint position at the Lawrence Berkeley National Laboratory (LBNL)-teamed up with LBNL physicist Xiao Dong Xiang and others to create a combinatorial library of materials rather than drug candiates. The group first made arrays of 128 different compounds, each a potential high-temperature superconductor, and each a tiny speck just 200 millionths of a meter across. The Berkeley team and others went on to create libraries of phosphors, data storage materials, polymers, catalysts, and even electronic devices.
For all these diverse materials, the basic strategy is the same: Make a lot of compounds at once, then scan them simultaneously to see which works best. To make the superconductor array, for instance, the Berkeley team sprayed seven different inorganic oxides one at a time through a mask. By using a series of different masks to control the deposition of each oxide, the researchers created a checkerboard of compounds in which each 200-micron square on the board contained a different combination of elements. The entire chip was then processed and screened for activity.
But making such arrays turns out to be the easy part; it’s much harder to pick winners. “It doesn’t make a lot of difference if you can make 100,000 compounds at once if you still have to test them one by one,” says Brandeis University chemist Gregory Petsko, who is also a scientific adviser to ArQule. Rapid screening methods are widely available in drug discovery research to detect desired biological activity. But equivalent screens for measuring most physical properties, such as flexibility and electrical conductivity, simply don’t exist yet.
“How do you measure the strength of a nanogram of material?” asks Luke Schneider, who heads the combinatorial effort at SRI International, a consulting and research firm in Menlo Park, Calif. “Nobody has developed that technology yet.” Further, combinatorial approaches require measurements of thousands of compounds at once. “There’s a whole new technology that has to be built,” says Schneider.
Several groups are trying to develop convenient methods for rapidly testing the properties of huge batches of different materials. Symyx found its new blue phosphor earlier this year by simply shining ultraviolet light on an array of candidate phosphors to see which glowed the brightest. Other high-speed screens are in the works. Last year, Xiang and his LBNL colleagues invented a new high-speed scanning microscope that they use to screen arrays for electronic properties. Richard Wilson and his colleagues at the University of Houston have been experimenting with an infrared sensor for tracking the activity of arrays of catalysts by looking at the heat given off during reactions.
Although the hunt is on for new screens, most of the success in developing combinatorial materials has come in designing libraries of interesting new compounds. Recently, the Berkeley team staked out more new territory by reporting the first combinatorial array of electronic devices. In this case, the researchers made simple devices called ferroelectric capacitors, used to store information as packets of electrical charge on DRAM (dynamic random access memory) computer chips. Computer companies hope to shrink DRAM chips to even smaller dimensions. But the materials currently used to confine the electrical charge fail when they’re layered too thin, causing current to seep out like water from a leaky bucket.
To find new “buckets” that don’t leak as much, Xiang and co-workers built an array of several thousand capacitors, each with a charge-confining layer made of a slightly different ceramic alloy. The group found that a particular combination of barium, strontium, and titanium, spiked with a touch of tungsten, was the best yet at stopping the leak. The new material is not likely to find its way into devices immediately because it still must prove itself on other grounds, such as fitting in with current chip-making practices. But it offers a promising new lead.
Though capacitors and phosphors are tempting targets for these revolutionary combinatorial methods, the big payoff could prove to be catalysts. Catalysts are key to a myriad of commercial processes, ranging from plastic manufacturing to the production of high-volume chemicals to emission-control devices in cars. Come up with a catalyst to make a better-or cheaper-commodity plastic, and you stand to win big. “You can warp markets with those things,” says Hogan.
Despite the economic incentives, researchers have a tough time designing catalysts. Catalysis is a notoriously complex process, and catalysts are finicky creatures; each works best under its own set of conditions, such as temperature, pressure, and concentrations of reactants. Figuring out how these variables affect the catalyst is extraordinarily difficult. As a result, polymer chemistry has long been part science and part art, with chemists relying heavily on intuition-and sheer luck-to find new catalysts. “Nobody knows how to design the ideal catalyst from scratch,” says Petsko.
The complexity of the materials makes discovering new catalysts a prime testing ground for combinatorial chemists. In 1996, researchers led by Amir Hoyveda and Marc Snapper at Boston College turned in one of the first reports on creating libraries of different catalysts. And now just about everybody else, including Symyx, ArQule, SRI, and DuPont are trying to do the same thing.
Despite the progress, combina-torial chemistry still must prove itself in materials research. And while combinatorial methods went from scientific oddity to rising star in the drug business within several years, success in the materials industry could be tougher to achieve.
Fast screening, it turns out, is not the only headache. Researchers must also come up with faster methods for determining the exact molecular structure of each compound. That is particularly difficult for crystalline materials such as high-temperature superconductors, says Xiang. Even if scientists know the exact chemical composition of a square in the array, the material can adopt a variety of structures, in the same way that precipitation can come down as rain, hail, or snow.
And beyond such research hurdles loom even more daunting commercialization challenges. “Finding a good material is not enough,” says Xiang. Researchers must figure out how to scale-up production from nanograms to tons. Even if a substance can be produced in relatively large quantities, bulk materials often behave very differently than do thin films. A compound that acts as a high-temperature superconductor when it is a thin film can behave completely differently as a bulk powder. “There are a lot of just plain doubters who question [whether] all this can be done,” says Bob Ezzell, a chemist at Dow Chemical.
Many don’t want to take the risk. “Most research managers with budget responsibilities don’t want to take a gamble” on an unproved technology, says Gerald Koermer, a chemist at Engelhard. “Their tendency is to hold back.”
But combinatorial proponents are not daunted. The technology, says SRI’s Schneider, escalates the research arms race, allowing its users to come up with new products faster and cheaper than competitors. And in a business where winners and losers are often determined in patent court, combinatorial chemistry could allow companies to sew up rights to new technologies before other companies even get wind of an emerging field, says Schneider. In an initial patent “it’s very difficult to cover everything you would like to cover,” Schneider explains. By speeding up the discovery process, he says, “combinatorial chemistry allows you to cover more of the world.”
The process also works in reverse. “It also makes it easier for your competitor to get around your technology,” by allowing them to quickly explore hundreds or thousands of alternative compounds to one already on the market, says Schneider. As a result, Schneider believes that in the near future, chemicals and materials companies will be more or less forced to use combinatorial efforts to prevent competitors from pirating their core businesses.
Just when that will happen is anybody’s guess. And it will take a radical change in thinking. “Research really hasn’t changed much since Madame Curie,” says Schneider. Combinatorial chemistry, he adds, “represents a major change in the research mindset. Making that change is hard to get people to do.” For researchers to be convinced that combinatorial chemistry is the wave of the future for materials science and not just a passing swell, “it’s really going to take a hit,” says Schneider. But if and when someone gets that first big hit, he says, “everyone will follow and say, God, I can’t believe we haven’t been doing this all along.”’
This startup wants to copy you into an embryo for organ harvesting
With plans to create realistic synthetic embryos, grown in jars, Renewal Bio is on a journey to the horizon of science and ethics.
VR is as good as psychedelics at helping people reach transcendence
On key metrics, a VR experience elicited a response indistinguishable from subjects who took medium doses of LSD or magic mushrooms.
This nanoparticle could be the key to a universal covid vaccine
Ending the covid pandemic might well require a vaccine that protects against any new strains. Researchers may have found a strategy that will work.
This artist is dominating AI-generated art. And he’s not happy about it.
Greg Rutkowski is a more popular prompt than Picasso.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.