Curtarolo Materials Laboratory
     Curtarolo Materials Laboratory
     home      people      publications      courses      aflow      aflow-online      "binary alloy library"               

"Key" Genetic Search of Novel Electrical Crystals
Pratt Duke News [website,article]

Duke University Office of News & Communications

FOR RELEASE: May, 14th 2012
CONTACT: Richard Merritt
(919) 660-8414

URHAM, N.C. - After creating a master "ingredient list" describing the properties of more than 2,000 compounds, Duke University engineers have now developed the a key to turning these ingredients into the next generation of electrical components.

The electrical compounds are known as topological insulators (TI), man-made crystals that are able to conduct electrical current on their surfaces, while acting as insulators throughout the interior of the crystal. Discovering TIs has become of great interest to scientists, but because of the lack of a rational blueprint for creating them, researchers have had to rely on trial-and-error approaches, which to date have had limited success.

Because of their unique properties, TIs can be created so that they can not only conduct electricity more efficiently, but can be fashioned to be much smaller that conventional wires or devices, making them ideal candidates to become quantum electronics devices, the Duke researchers said.

The "key" developed by the Duke investigators is a mathematical formulation that unlocks the data stored in the database and provides specific recipes for searching TIs with the desired properties.

"While extremely helpful and important, a database is intrinsically a sterile repository of information, without a soul and without life. We need to find the materials `genes'," said Stefano Curtarolo, professor of mechanical engineering and materials sciences and physics at Duke's Pratt School of Engineering and director of the Duke's Center for Materials Genomics. "We have developed what we call the `topological descriptor,' that when applied to the database can provide the directions for producing crystals with desired properties."

Finally the team discovered a new class of systems that could have been hardly anticipated without such a genetic approach.

The results of the Duke research was reported online in the journal Nature Materials. The work was supported by the Office of Navy Research and the National Science Foundation.

In November, Curtarolo and colleagues reported the establishment of a materials genome repository ( which allows scientists to stop using trail-and-error methods for combining different elements to create the most efficient alloys for a promising method of producing electricity. The project developed by the Duke engineers covers many thousands of compounds, and provides detailed recipes for creating most efficient combinations for a particular purpose, much like hardware stores mix different colors of paint to achieve the desired result. The project is the keystone of the newly formed Duke's Center for Materials Genomics.

The new descriptor developed by the Duke team basically can determine status of any specific combination of element under investigation. On one end of the spectrum, Curtarolo explained, is "fragile."

"We can rule those combinations out because what good is a new type of crystal if it would be too difficult to grow, or if grown, would not likely survive," Curtarolo said.

A second group of combinations would be a middle group termed feasible, but what excites Curtarolo the most are those combinations found to be robust. These crystals are those that are stable and able to be easily and efficiently produced. Also, and just importantly, these crystals can be grown in different directions, which gives them the advantage of tailored electrical properties by simple growth processes.

While TIs generally are currently in the experimental stage, Curtarolo believes that with this new tool, scientists should have a powerful framework for engineering of wide variety of them.

First author of the paper was Kesong Yang, post-doctoral fellow in Curtarolo's laboratory. Other members of the team were Duke's Shidong Wang, Wahyu Setyawan, Pacific Northwest Laboratory and Marco Buongiorno Nardelli, University of North Texas and the Oak Ridge National Laboratory.

# # #

New Elemental Cookbook Guides Efficient Thermoelectric Combinations
Pratt Duke News [website,article]

Duke University Office of News & Communications

FOR RELEASE: November 29, 2011
CONTACT: Richard Merritt
(919) 660-8414

New Elemental Cookbook Guides Efficient Thermoelectric Combinations

Note to editors: Stefano Curtarolo can be reached at (919) 660-5506 or Photographs of the researchers and calcite images are available.

DURHAM, N.C. A materials genome repository developed by Duke University engineers will allow scientists to stop using trail-and-error methods for combining different elements to create the most efficient alloys for a promising method of producing electricity.

These thermoelectric materials produce electricity by taking advantage of temperature differences and are currently being used in such applications as deep space satellites to campsite coolers. In the past, scientists have not had a rational basis for combining different elements to produce these energy-producing materials.

The project developed by the Duke engineers covers thousands of compounds, and provides detailed "recipes" for creating most efficient combinations for a particular purpose, much like hardware stores mix different colors to achieve the desired result. The database is free and open to all (

Stefano Curtarolo
"We have calculated the thermoelectric properties of more than 2,500 compounds and have calculated all their energy potentials in order to come up with the best candidates for combining them in the most efficient ways," said Stefano Curtarolo, associate professor of mechanical engineering and materials sciences and physics at Duke's Pratt School of Engineering. "Scientists will now have a more rational basis when they decide which elements to combine for their thermoelectric devices."

The results of the Duke team's work were published on line in the journal Physics Review X.

A thermoelectric device takes advantage of temperature differences on opposite sides of a material - the greater the temperature difference, the greater energy potential. Different material combinations may be a more efficient method of turning these temperature differences into power, according to Shidong Wang, a post-doctoral fellow in Curtarolo's lab and first author of the paper.

Thermoelectric materials can be created by combining powdered forms of different elements under high temperatures - a process known as sintering. Not only does the new program provide the recipes, but it does so for the extremely small versions of the particular elements, known as nanoparticles. Because of their miniscule size and higher surface areas, nanoparticles have properties unlike their bulk counterparts.

"Having this repository could change the way we produce thermoelectric materials," Wang said. "With the current trial-and-error method, we may not be obtaining the most efficient combinations of materials. Now we have a theoretical background, or set of rules, for many of the combinations we now have. The approach can be used to tackle many other clean energy related problems."

Thermoelectric devices are currently used, for example, to provide power for deep-space satellites. The side of the device facing the sun absorbs heat, while the underside of the device remains extremely cold. The satellite uses this temperature difference to produce electricity to power the craft.

Shidong Wang
The Duke researchers believe that the use of thermoelectric devices - which the new database should help fuel - could prove especially effective in cooling microdevices, such as laptop computers.

Wang and Curtarolo made use of data collected by the consortium, a cloud-distributed repository for materials genomics. It currently comprises electronic structures, magnetic and thermodynamic characterization of inorganic compounds. The project, started by Duke scientists, is sponsored by the Office of Naval Research, the National Science Foundation and the U.S. Department of Homeland Security.

Duke's Wahyu Setyawan, as well as Zhao Wang and Natalio Mingo of France's Atomic Energy and Alternative Energies Commission, were also part of the research team.

# # #

New Approach Gives On-the-Spot Information on Sample's Origins
Pratt Duke News [website,article]

Duke University Office of News & Communications

FOR RELEASE: November 17, 2010
CONTACT: Richard Merritt
(919) 660-8414

New Approach Gives On-the-Spot Information on Sample's Origins

Note to editors: Stefano Curtarolo can be reached at (919) 660-5506 or Kristin Poduska can be reached at 709.864.8890 or Photographs of the researchers and calcite images are available.

DURHAM, N.C. - Mother Nature makes her crystals right, ancient humans not always so much.

The ability to tell the difference can be important to archeologists in the field who discover an unknown sample and need to know whether the material formed as a result of natural processes or was created by humans. This can be a crucial bit of information in determining the ancient activities that took place at a site, yet archeologists often wait for months for the results of laboratory tests.

Now, however, an international team of physicists, archeologists and materials scientists has developed a process that can tell in a matter minutes the origin of samples thousands of years old. The new device is easily portable, and works by "lifting off" the spectral fingerprint of a material by use of infrared light.

The first material tested was the mineral calcite, commonly found in rocks like limestone, which forms over millions years in sediments. These rocks can also contain the mineralized shells of sea creatures.

In the Nov. 17th issue of the journal Advanced Materials, Stefano Curtarolo, associate professor of mechanical engineering and materials sciences and physics at Duke University, and Kristin Poduska, associate professor of physics at Memorial University in Newfoundland, and their colleagues at the Wezimann Institute of Science in Israel describe the new approach, which has already been successfully tested in archeological sites in Israel.

"The key to determining a sample's origin lay in figuring out how well the crystal structure is organized," Curtarolo said. "Naturally occurring calcite crystals are tightly organized, while a material created by humans from calcite is usually far less organized."

However, interpreting the information obtained using traditional methods is not only time-consuming, but tricky, since such factors as particle size and the alignment of the atoms within the crystals can send out conflicting information.

"For this reason, getting useful and reliable information about the sample usually requires careful and time-intensive sample preparation with highly specialized equipment," Poduska said.

The current scientific team used infrared spectroscopy to take advantage of the fact that different molecular units absorb light differently, yielding distinct spectral peaks, or molecular fingerprints. They put a sample through a series of grindings - sometimes as many as a dozen - while taking detailed infrared spectroscopy readings after each one. By analyzing the absorption peaks at different points in the grinding process, as the particles got smaller and smaller, they could tease out the effects of size and organization.

For example, Curtarolo said, an archeologist finds a sample and knows that it is calcite, but what cannot be determined at the site is whether it is a naturally occurring mineral, or a plaster, ash, or other building material made of calcite. Plaster is made by heating limestone, grinding it up and mixing it with water.

"We've shown in the field that our method can quickly detect subtle differences in the organization of a crystal by decoupling the two factors that influence the spectral peaks," Poduska said. "Our method is particularly powerful because the direct measurement of particle size is not needed, and it can be used with any crystal that can be excited by infrared light."

Last summer, a team of archeologists and scientists from the Weizmann Institute successfully tested the new approach at an ancient site in central Israel at Tel Safit, close to where King David is thought to have slain Goliath.

"Whenever they found something white, they would call us over to do tests," Poduska said. "We were able to confirm whether the sample was rock or plaster, which helps us decide how to proceed at the excavation site."

Other members of the team, from the Weizmann Institute of Science, were structural biologists Steve Weiner and Lia Addadi, nuclear physicist Elisabetta Boaretto, PhD student Lior Regev, and theoretical physicist Leeor Kronik.

The researchers were supported by the Weizmann Institute, the Lize Meitner Minerva Center for Computational Quantum Chemistry, and the National Science Foundation.

# # #

Smaller is better in the viscous zone
Pratt Duke News [website,article]

Duke University Office of News & Communications

FOR RELEASE: 4 p.m. ET, October 21, 2010
Contact: Richard Merritt
Duke University

Smaller is better in the viscous zone

DURHAM, N.C. -- Being the right size and existing in the limbo between a solid and a liquid state appear to be the secrets to improving the efficiency of chemical catalysts that can create better nanoparticles or more efficient energy sources.

When matter is in this transitional state, a catalyst can achieve its utmost potential with the right combination of catalyst particle size and temperature, according to a pair of Duke University researchers. A catalyst is an agent or chemical that facilitates a chemical reaction. It is estimated that more than 90 percent of chemical processes used by industry involve catalysts at some point.

This finding could have broad implications in almost every catalyst-based reaction, according to an engineer and a chemist at Duke who reported their findings on line in the American Chemical Society's journal ACS-NANO. The team found that the surface-to-volume ratio of the catalyst particle – its size -- is more important than generally appreciated.

"We found that the smaller size of a catalyst will lead to a faster reaction than if the bulk, or larger, version of the same catalyst is used," said Stefano Curtarolo, associate professor in the Department of Mechanical Engineering and Materials Sciences.

"This is in addition to the usual excess of surface in the nanoparticles," said Curtarolo, who came up with the theoretical basis of the findings three years ago and saw them confirmed by a series of intricate experiments conducted by Jie Liu, Duke professor of chemistry.

"This opens up a whole new area of study, since the thermo-kinetic state of the catalyst has not before been considered an important factor," Curtarolo said. "It is on the face of it paradoxical. It's like saying if a car uses less gas (a smaller particle), it will go faster and further."

Their series of experiments were conducted using carbon nanotubes, and the scientists believe that same principles they described in the paper apply to all catalyst-driven processes.

Liu proved Curtarolo's hypothesis by developing a novel method for measuring not only the lengths of growing carbon nanotubes, but also their diameters. Nanotubes are microscopic "mesh-like" tubular structures that are used in hundreds of products, such as textiles, solar cells, transistors, pollution filters and body armor.

"Normally, nanotubes grow from a flat surface in an unorganized manner and look like a plate of spaghetti, so it is impossible to measure any individual tube," Liu said. "We were able to grow them in individual parallel strands, which permitted us to measure the rate of growth as well as the length of growth."

By growing these nanotubes using different catalyst particle sizes and at different temperatures, Liu was able to determine the "sweet spot" at which the nanotubes grew the fastest and longest. As it turned out, this happened when the particle was in its viscous state, and that smaller was better than larger, exactly as predicted before.

These measurements provided the experimental underpinning of Curtarolo's hypothesis that given a particular temperature, smaller nanoparticles are more effective and efficient per unit area than larger catalysts of the same type when they reside in that dimension between solid and liquid.

"Typically, in this field the experimental results come first, and the explanation comes later," Liu said. "In this case, which is unusual, we took the hypothesis and were able to develop a method to prove it correct in the laboratory."


The research was supported by the Office of Naval Research, the National Science Foundation, the Department of Energy and the National Council of Science and Technology (CONACYT), Mexico. Duke's Thomas McNicholas and Jay Simmons, as well as Felipe Cervantes-Sodi, Gabor Csanyi and Andrea Ferrari, University of Cambridge, U.K., were also members of the team.

Duke scientist wins presidential award from department of Defense
Pratt Duke News [website,article]

Duke University Office of News & Communications

FOR RELEASE: 4 p.m. ET, Friday, December 19, 2008
CONTACT: Deborah Hill
(919) 660-8403

*Duke scientist wins presidential award from department of Defense*
Note to editors: Stefano Curtarolo can be reached at (919) 660-5506 or An image of him is available.

DURHAM, N.C. - In recognition of his discovery and characterization of novel combinations of elements, Duke engineer and physicist Stefano Curtarolo, Ph.D., has received a Presidential Early Career Award for Scientists and Engineers (PECASE).

The award, the highest honor given to scientists by the federal government, also carries $1 million in research support over five years. Many federal agencies participate in the PECASE program - Curtarolo was recommended by the Department of Defense's Office of Naval Research (ONR), which had granted him a Young Investigator Award in 2007 (ONR-YiP).

Curtarolo received the award Dec. 19 during a ceremony at the White House.

Curtarolo, who joined the Duke faculty in 2003, is an associate professor in the Department of Mechanical Engineering and Materials Sciences and in the Department of Physics.

As a theoretician, Curtarolo uses supercomputers to search for and test novel combinations of elements for specific purposes, or to better understand existing materials. He searches for such materials as novel titanium alloys for marine structures, new superconductors, ceramic materials for nuclear detection, and metallic nanoparticles for growing nanotubes and fuel cell alloys for energy conversion.

Specifically, he searches for these combinations at the nano scale, where the macroscopic laws of physics don't always apply. Because he works in such small systems, sometimes involving a few hundred atoms at a time, supercomputers are employed to calculate all the possible combinations of materials and their properties. For example, he used the supercomputer at the Texas Advanced Center at the National Science Foundation - sponsored University of Texas - Austin to discover thermodynamic instabilities in iron nano-catalysts governing the minimum size of nanotubes that can be grown.

"This grant will allow me and my team to extend our research in current areas as well as into exciting new avenues," Curtarolo said. "We expect to expand our investigations into such areas as titanium alloys, nano-catalysts for energy production and conversion, new materials for detecting radiation and strategies for reducing friction between quasi-crystalline surfaces."

Curtarolo also is involved in improving access to research for students with disabilities and those from under-represented minorities.

In addition to the ONR awards, Curtarolo received a CAREER award from the National Science Foundation in 2007.

For the PECASE award, the Office of Naval Research cited Curtarolo's "research on physics and thermodynamics of superconducting materials, topological transitions of quasi-crystalline thin films, size-induced instabilities in nano-catalysts; and for mentoring minority graduate students."

After receiving his undergraduate training at the University of Padova, Italy, Curtarolo earned an M.S. degree in condensed matter physics from Pennsylvania State University and a Ph.D. in materials science from the Massachusetts Institute of Technology in 2003.


New 'Metal Sandwich' May Break A Superconductor Record, Theory Suggests
DUKE News [website, article]
Pratt Duke News [website,article]

Duke University Office of News & Communications

FOR IMMEDIATE RELEASE: Monday, May 8, 2006

Durham, N.C. -- After an exhaustive data search for new compounds, researchers at Duke University's Pratt School of Engineering have discovered a theoretical "metal sandwich" that is expected to be a good superconductor. Superconductive materials have no resistance to the flow of electric current.

The new lithium monoboride (LiB) compound is a "binary alloy" consisting of two layers of boron -- the "bread" of the atomic sandwich -- with lithium metal "filling" in between, the researchers said. Once the material is synthesized, it should be superconductive at a higher temperature than other superconductors in its class, according to their results. The researchers reported their findings in the May 5 online edition of the journal Physical Review B, Rapid Communications.

"To the best of our knowledge, this alloy structure had not been considered before," said Stefano Curtarolo, professor of mechanical engineering and materials sciences at Duke's Pratt School. "We have been able to identify synthesis conditions under which the LiB compound should form. And we believe that if the material can be synthesized, it should superconduct at a higher temperature, perhaps more than 10 percent greater, than any other binary alloy superconductor."

"The significance of the work is not only the discovery of lithium monoboride itself, but also that this opens the door to finding derivatives that could aid in the search for additional novel superconductors," added Aleksey Kolmogorov, lead author of the study and a postdoctoral fellow at the Pratt School. He said that once a new superconductive material is identified, scientists typically can manipulate the substance -- twisting it or doping it with other elements to create related structures that might have even more appealing properties. Superconductors have the potential to produce more efficient electronics and electric generators, according to the researchers. The materials also have unique magnetic capabilities that may enable their use in transportation applications, such as "levitated" trains that glide over their tracks with virtually no friction. However, today's superconductors perform only when cooled to extremely low temperatures near absolute zero, which is -459.67 degrees Fahrenheit, or 0 degrees Kelvin. This requirement makes their use prohibitively expensive, the researchers said.

The first superconductive material was identified in 1911 when a Dutch scientist cooled mercury to 4 degrees Kelvin, the temperature of liquid helium. Since then, scientists have discovered superconductivity in various materials, including other pure elements, complex ceramics, and binary alloys.

Since 1986, ceramics have held the overall record for highest superconducting temperature -- currently 138 degrees Kelvin. Among pure elements, lithium, when contained under pressure, holds the record at 20 degrees Kelvin.

Recently, scientists scored an unexpected breakthrough with the discovery of superconductivity in the simple binary alloy magnesium diboride (MgB2), Curtarolo said. This compound holds the current temperature record for its class at 39 degrees Kelvin, and it has attracted much attention because it can be produced relatively easily from two abundant elements.

"The physics of the superconductivity in MgB2 is now well understood," Kolmogorov said. "However, MgB2 has been shown to be such a unique superconductor -- finely tuned by nature -- that attempts to improve it or use it as a model for finding even better superconducting materials have so far been fruitless."

Curtarolo and Kolmogorov decided it was time to try something else. Using a theoretical data-mining method developed by Curtarolo, the pair scoured a database of experimental and hypothetical compounds, looking for other possible configurations of binary alloys and tweaking their compositions.

In the process, the team stumbled onto "a path to a new metal sandwich structure consisting of stacks of metal and boron layers," Curtarolo said.

Additional calculations identified the binary alloy lithium monoboride as a promising candidate that might be both structurally stable and superconductive at temperatures that exceed those of the current binary alloy record-holder.

"It's a very thin line, because as you try to increase the temperature at which a material becomes superconducting, the material tends to lose its stability," Kolmogorov said. "But we think lithium monoboride should be stable and superconduct at temperatures greater than 39 degrees Kelvin."

"It was like spotting a $100 bill on the street," Curtarolo said of the finding. "It seemed impossible that this could be real and that no one had seen it before."

The researchers are now conducting more precise theoretical calculations of LiB's "critical temperature" -- that is, the temperature at which it becomes superconductive -- with computational support from the San Diego Supercomputer Center at the University of California, San Diego.

The material will have to be synthesized before experimental tests can confirm any of the theoretical results, the researchers said. They added that this won't be an easy process, as manufacturing lithium monoboride will require extremely high temperatures and pressures.

For more information, contact: Kendall Morgan, Pratt School of Engineering | 919-660-8414 | Relaxation path from FCC-V2 to MS1

Prediction of new crystal structure phases in metal borides:
a lithium monoboride analog to MgB2
[pdf, PRB]

DUKE-PSU Researchers Aim for UltraLow-Friction Machine Parts with Computer Model of Quasicrystal Metal
DUKE News [website, article]
Pratt Duke News [website,article]

Duke University Office of News & Communications

FOR IMMEDIATE RELEASE: Thursday, Sept. 15, 2005

CONTACT: James Todd
(919) 681-8061

Note to editors: Stefano Curtarolo can be reached at (919) 660-5506 or

DURHAM, N.C. -- Duke University materials scientists have developed a computer model of how a "quasicrystal" metallic alloy interacts with a gas at various temperatures and pressures. Their advance could contribute to wider applications of quasicrystals for extremely low-friction machine parts, such as ball bearings and sliding parts.

Quasicrystals, like normal crystals, consist of atoms that combine to form structures -- triangles, rectangles, pentagons, etc. -- that repeat in a pattern. However, unlike normal periodic crystals, in quasicrystals the pattern does not repeat at regular intervals. So, while the atomic patterns of two crystalline materials rubbing together can line up and grind against one another, causing friction, quasicrystalline materials do not, and thus produce little friction.

Quasicrystalline metalic alloys are already used in a handful of commercial products, including as a coating for some non-stick frying pans because they combine the scratch- and temperature- resistant properties of a polymer such as Teflon with the heat conduction property of metals.

However, a major technical obstacle remains to using quasicrystal materials to minimize friction between surfaces sliding against one another, the scientists said. Microscopic surface contaminants, such as atmospheric gases, can come between the surfaces and interfere with the materials' high lubricity. The gases form a thin layer of molecules over the alloy surface-- typically in a crystalline pattern -- which masks the desirable surface properties of the underlying quasicrystal, they said.

The researchers' computer model of the effect of adsorbed gas on the quasicrystal alloy of aluminum, nickel and cobalt will be published in an upcoming issue of the journal Physical Review Letters. Their research was funded by the National Science Foundation.

"We are interested in quasicrystals because they are scratch-resistant and they have very little friction," said Stefano Curtarolo, lead author of the paper and a professor of materials science in Duke's Pratt School of Engineering. "So they are promising for sliding interfaces in machines and applications where the potential for scratching might be involved."

Metals were believed to have only periodic crystalline structures until 1984, when materials scientists reported discovery of the first metallic alloy with a quasicrystalline structure. Since then, scientists, including Curtarolo, have sought to explore the properties and applications of quasicrystals.

The challenge Curtarolo, Duke graduate student Wahyu Setyawan and their colleagues at Penn State University address in their paper is how to preserve the low-surface-friction property of a quasicrystal in the presence of a gas.

In previous experiments, Curtarolo's Penn State colleagues Nicola Ferralis, Renee D. Diehl, Raluca Trasca and Milton W. Cole had found that when xenon gas is exposed to their quasicrystal alloy, a single layer of xenon first forms in a quasicrystal pattern on top of the alloy, but by the time two or more layers formed, the xenon atoms develop a crystalline structure.

They chose to experiment with xenon, which does not react chemically with most metals, so they could consider the physical interaction of the gas and the metallic alloy, without complicating chemical interactions. In the experiments, the number of layers formed by the xenon atoms varies with the experimental temperature and pressure.

"If you have very little xenon gas, it's going to follow the aperiodic symmetry of the quasicrystal; if you have a lot, it's going to follow the periodic structure of xenon," Curtarolo said. "This change from quasicrystal to periodic crystal -- that's what we want to know about."

Cutarolo and his colleagues modeled in their computer simulation this transition from a single layer of xenon with quasicrystalline properties to multiple layers with crystalline properties. The simulation is consistent with experimental data.

The simulation is available online at . In the simulation, the image on the left is of the average position of the xenon atom, the image on the right is of the electron diffraction pattern used to determine the position of the atoms and the graph on the bottom gives the density of the xenon gas.

"This model tells us how we might be able to control the transition and preserve the low-friction property of quasicrystals," Curtarolo said. "It's a step towards understanding how quasicrystals interact with gases in the atmosphere and how we could eventually use them in real machines."

5- to 6-fold transition

Duke University: Stefano Curtarolo (left) and Wahyu Setyawan. (photo Jim Wallace)

Penn State (left to right): Milton W. Cole, Raluca Trasca, Nicola Ferralis and Renee Diehl.

MIT team mines for new materials with a computer
MIT News [website, article]

For Immediate Release
MONDAY, NOV. 17, 2003
Contact: Elizabeth A. Thomson, MIT News Office
Phone: 617-258-5402

CAMBRIDGE, Mass.--A computational technique used to predict everything from books that a given customer might like to the function of an unknown protein is now being applied by MIT engineers and colleagues to the search for new materials.

The team's ultimate goal: a public online database that could aid the design of materials for almost any application, from nanostructure computer components to ultralight, high-strength alloys for airplanes.

The technique, known as data mining, uses statistics and correlations to search for patterns within a large data set. Those patterns can then be used to predict an unknown. ", for example, tracks a customer's past purchases, then uses data mining to suggest, based on those purchases, additional books the customer might like," said Dane Morgan, a research associate in the Department of Materials Science and Engineering (DMSE).

The technique also has applications in science. Applied to a protein database containing "essentially all the known data on protein structure," Morgan said, data mining "assists researchers in exploring the structure, properties and functions of other proteins."

Now Morgan and colleagues have shown that data mining can also make the search for new materials easier. They describe their work in a recent paper in Physical Review Letters.

Authors are Stefano Curtarolo (Ph.D. 2003); Morgan; Gerbrand Ceder, the R.P. Simmons Professor of Computational Materials Science; Kristin Persson, an MIT postdoctoral associate when the work was conducted; and John Rodgers of Toth Information Systems in Canada. Curtarolo, now an assistant professor at Duke University, wrote his thesis on the work and is continuing to develop it in collaboration with his MIT colleagues.

Throughout history, scientists have created new materials with novel characteristics by experimentation, essentially melting together existing materials, then painstakingly characterizing the structure of the resulting product. "The behavior of any material flows from its structure," Morgan noted.

With state-of-the-art computational techniques, or ab initio methods, engineers can now do "virtual" screenings of potential materials. A computer predicts what structure and properties a given mixture might have, based on fundamental equations of quantum mechanics. Ceder's Lab for Computational Materials Science specializes in ab initio calculations.

Even these virtual screenings, however, can be time-consuming and costly because "there are still so many possible structures for any given material that it's impractical for the computer to explore them all," Morgan said.

The new MIT technique "establishes patterns among the many thousands of different possible structures" for a given mixture of materials, he said. "These patterns can then be used to greatly reduce the number of structures the computer has to explore."

To date, the MIT team has tested the technique on a relatively small homegrown database. Recently, however, they received funding from the National Science Foundation to produce a public online database "that will allow the whole computational materials community to contribute calculated data," Morgan said.

The team is excited that the materials database will allow the "recycling" of data from past ab initio computer calculations and laboratory experiments. "Until now, researchers have made no formal use of their older calculations, simply starting again with each new material, thereby throwing away a huge amount of information," Curtarolo said.

"Just as recycling old cans allows one to avoid waste, the ability to recycle old calculated data will avoid wasted and useless calculations in the future. In addition, old calculations for already investigated systems might be used to predict properties of new systems.

"We believe this database and associated data-mining tools will become a standard tool for scientists studying new materials systems," Curtarolo concluded.

The Lab for Computational Materials Science is funded by the Department of Energy and the MIT Center for Materials Science and Engineering through the MRSEC program of the National Science Foundation. In addition, Hewlett-Packard recently donated a million-dollar supercomputer to the lab.


     Energy Materials Laboratory - Curtarolo Group - Mechanical Engineering and Materials Science - Duke University, 144 Hudson Hall, Box 90300, Durham NC 27708