C&EN Classifieds
Home | This Week's Contents  |  C&EN ClassifiedsSearch C&EN Online

Related Stories
[C&EN, June 4, 2001]

[C&EN, Jan. 22, 2001]

Combinatorial Synthesis Hits The Spot
[C&EN, July 3, 2000]

Combinatorial Chemistry - Redefining The Scientific Method
[C&EN, May 15, 2000]

Materials A La Combi
[C&EN, May 15, 2000]

Related People
E-mail this article to a friend
Print this article
E-mail the editor
 Table of Contents
 C&EN Classifieds
 News of the Week
 Cover Story
 Editor's Page
 Government & Policy
  Government & Policy
 ACS News
 Digital Briefs
 ACS Comments
 Career & Employment
 Special Reports
 What's That Stuff?
 Pharmaceutical Century

 Hot Articles
 Safety  Letters

 Back Issues

 How to Subscribe
 Subscription Changes
 About C&EN
 Copyright Permission
 E-mail webmaster
August 27, 2001
Volume 79, Number 35
CENEAR 79 35 pp. 59-63
ISSN 0009-2347
[Previous Story] [Next Story]

Combinatorial synthesis is leading to new alloys, catalysts, and magnetic semiconductors


HIGH-TECH FILMS This molecular beam epitaxy system at Pacific Northwest National Laboratory was used to grow high-quality films of a room-temperature magnetic semiconductor discovered by a Japanese team using the combinatorial approach.

Ainissa G. Ramírez often jokes that if graduate students in materials science were to find out about the combinatorial approach to materials discovery, they would get out of graduate school much sooner. That's because this synthetic approach allows researchers to prepare hundreds or thousands of different chemical compositions using the same small set of starting materials--all in one experiment. And by using newer high-speed screening techniques, it's possible to examine the properties of those compositions much more quickly and efficiently than in the past. Thus, researchers, like Ramírez, who are in search of promising new materials, can now explore whole ranges of different compositions in less time than it previously took to prepare and study just one sample.

"In my laboratory, we tend to look at hundreds of samples, so why not have all those hundreds of samples on one wafer?" says Ramírez, a materials scientist at Agere Systems, formerly the Microelectronics Group of Lucent Technologies in Murray Hill, N.J.

The combinatorial approach was embraced about a decade ago by the pharmaceutical industry as a faster, more economical way to discover and optimize drug candidates. And since the mid-1990s, this approach has increasingly been applied to the search for new materials with desirable electronic, magnetic, luminescent, catalytic, mechanical, or other properties.

The adoption of the combinatorial approach has altered the usual pattern of research in materials science, Ramírez points out. Traditionally, "you make a sample, analyze its composition, and then test it for the properties you desire." But now that you can prepare a wafer containing a large number of different (and unknown) compositions, "you have to switch gears: You first test the wafer for the properties you want. Then, if you find a range of desirable properties, you go back and determine what compositions are associated with those properties."

Some combinatorial experiments involve wafers that contain a specific number of discrete spots, each one having a unique composition. The wafers Ramírez uses in her studies, on the other hand, are coated with a thin-film continuum of compositions--specifically, of metal alloys.

RAMIREZ IS SEARCHING for an alloy that could be used as the contact material in a microrelay, a miniaturized mechanical switch for future microelectromechanical devices. A microrelay is somewhat like a telegraph clicker: It has two contacting surfaces--one on a base and one on a cantilever arm--through which an electrical current passes. The switching occurs very rapidly: Microrelays will need to perform through a billion or more switching cycles without sticking (one possible failure mechanism). So Ramírez is seeking a hard, conductive, and wear-resistant alloy for the microrelay's contacting surfaces.

For several reasons, gold is a good starting point for these investigations. But gold is too soft by itself, so it has to be hardened by addition of another metal. Alloying, though, degrades electrical conductivity, so one has to find the right balance between hardness and conductivity.

To efficiently explore a range of gold alloys, Ramírez prepares so-called continuous compositional spreads of gold with a second metal such as cobalt, lead, or antimony. These binary spreads can be prepared by simultaneously depositing gold and the second metal onto a wafer from two sputter guns located on opposite sides of the wafer. One gun deposits gold; the other, cobalt, for example. Each gun generates a thickness gradient of material on the wafer, with the most material being deposited nearest the gun, and the deposition rate falling off exponentially with distance from the gun. Different amounts of the two metals are deposited at each point on the wafer, producing a smoothly varying, continuous spread of alloy compositions. The process is exciting for materials scientists, she explains, because "you can generate a phase diagram of a binary system on a wafer." In a similar way, a ternary thin-film alloy spread can be prepared very quickly using a three-gun arrangement.

Ramírez' work on these alloy spreads is still at an early stage, and she hasn't yet published her results. But she has discussed some aspects of this research at meetings like the Knowledge Foundation-sponsored conference on combinatorial materials discovery that was held last January in San Diego.

"THE MOST INTERESTING results we've had so far involve the effects of annealing on the electrical properties of these spreads," she tells C&EN. Although adding a second metal to gold degrades the conductivity, some of this lost conductivity can be recovered by annealing the spread at around 300 ºC. "So that is encouraging," she says.

She also has been able to assess the deformation properties of the films by moving a nanoindenter--a nanoscale diamond tip--across the wafer and pushing it into the surface at various locations to measure the stress and strain. Not surprisingly, the cobalt-rich alloys were found to be harder than the gold-rich alloys.

Ramírez says she has identified some promising candidate materials for use in microrelays. At this time, however, she's not at liberty to divulge any details beyond the fact that the materials are based on noble alloys.

Combinatorial synthesis of thin films also has been used in the search for other types of solid-state functional materials, including superconductors, dielectrics, phosphors, and heterogeneous catalysts. Unfortunately, the thin-film approach doesn't lend itself to screening structural materials; for these, the bulk properties are key. But now a new combinatorial method, developed by materials scientist Ji-Cheng Zhao at General Electric corporate research and development in Schenectady, N.Y., can be used to generate compositional variations in bulk samples.

HEAT HELPS A gold-antimony compositional spread demonstrates how a heat treatment can improve the electrical conductivity of Au-Sb alloys. The film spread contains the most Au-rich compositions on the left, with Sb levels increasing toward the right. The colors indicate the conductivity at various spots on the film, with black indicating the most conductivity and red indicating the least. A comparison of the two patterns indicates that most of the compositions display a higher conductivity after the heat treatment.
Researchers in search of promising new materials can now explore whole ranges of different compositions in less time than it previously took to prepare and study just one sample.

MATERIALS SCIENTISTS have long known that if two (or more) metal or ceramic slabs of different composition are pressed together at high temperature, atoms from each slab will diffuse into adjoining slabs, forming new phases. This approach has long been used to determine phase diagrams and diffusion coefficients, Zhao notes, but it hasn't been used to survey the properties of the new phases formed. Zhao has now coupled this traditional synthetic technique with newer microscale techniques for measuring compositions and properties, creating what he calls "a powerful combinatorial approach" for structural (load-bearing) materials.

In recent issues of Advanced Engineering Materials [3, 143 (2001)] and the Journal of Materials Research [16, 1565 (2001)], Zhao describes this approach for studying a complex system involving four components: molybdenum; nickel; iron; and Inconel 706 (IN706), a commercial superalloy consisting of (by weight) about 42% nickel, 38% iron, 16% chromium, and smaller amounts of several other elements. The goal of the experiment was to study how adding various amounts of molybdenum, nickel, and iron to the alloy would affect its properties.

To promote the interdiffusion of these four materials, Zhao assembled machined components made of these materials into a metallic "birthday cake." The entire outside of the cake--the "frosting"--was pure nickel. Buried inside was a smaller cake, consisting of four quarter slices--one each of the four starting metals. In this arrangement, every metal slice was in contact with all the other metal slices. Zhao ensured good contact by welding the assembly together and subjecting it to high temperature and high pressure. Then the entire cake--with frosting--was baked at 1,100 ºC for more than two months to give the constituents plenty of time to intermix by diffusion.

After heating, the cake was cut in half horizontally, and the cut surfaces were polished flat. Zhao was then able to chemically analyze the various phases that had formed (using electron probe microanalysis) and determine their mechanical properties. From these data, for example, he mapped out the Ni-Mo-Fe phase diagram and found that it agrees very well with that measured earlier from equilibrated alloys. And by performing nanoindentation tests, he found that iron has little effect on the hardness and elastic modulus (stiffness) of a major phase known as the g-phase, whereas molybdenum causes significant hardening but has little effect on stiffness.

Metallurgists have long known that molybdenum is a better hardener than iron, Zhao remarks. But actual data on how the hardness varies with composition have been difficult to come by because the traditional approach of studying one alloy at a time is expensive and time-consuming. The combinatorial approach is much more efficient, he says.

COMBINATORIAL METALLURGY General Electric's Zhao shows off two "diffusion multiples"--assemblies of different metallic components that are welded together and heated for prolonged periods to promote the formation of intermetallic compounds through interdiffusion. At right is a cross-sectional view of one such diffusion multiple consisting of Inconel 706, nickel, molybdenum, and iron components.

Zhao also has mapped out IN706-Mo-Fe phases and evaluated their properties. This kind of multicomponent phase diagram can be used directly for designing new alloys and for improving the computational design of materials, he says.

Zhao and his coworkers have studied many other alloy systems using this combinatorial approach and have uncovered promising new phases that may lead to new commercial alloys. He firmly believes that the diffusion method is a quicker and less expensive way to search for useful alloy compositions.

Combinatorial chemistry is also speeding up the search for new metal-based catalysts. In the Netherlands, for example, chemists at Delft University of Technology and Avantium Technologies, Amsterdam, have demonstrated the use of high-speed synthesis and screening techniques to find new titanium-silsesquioxane catalysts for the epoxidation of alkenes [Angew. Chem. Int. Ed., 40, 740 (2001)].

Silsesquioxanes are polyhedral frameworks made of Si–O–Si linkages. Imagine, for example, a cube where each edge is a linear Si–O–Si moiety and the silicon atom at each corner has an organic (R) group attached. Now imagine the related open structure in which one of the corner silicons is missing and each of its nearest-neighbor silicons carries a hydroxyl group. This not-quite-cubical trisilanol can react with titanium(IV) isopropoxide [Ti(OiPr)4] to yield a closed-cube silsesquioxane that has a titanium instead of a silicon at one corner. This titanium-silsesquioxane structure is known to be an active epoxidation catalyst. But its main interest to chemists is as a model compound: Because its titanium center is attached to siloxy groups, this soluble compound can be used as a convenient stand-in for insoluble oxidation catalysts in which titanium centers are grafted to, and supported by, silica.

SCIENTISTS HAVE established that in the most active heterogeneous titanium-silica catalysts, the titanium centers are all four-coordinate. But they haven't determined whether, for optimum performance, the titanium center must have four siloxy groups attached to it, or whether three siloxy groups and one hydroxy would work as well.

One way to investigate this question would be to make trisilanols having different R groups on the silicon, react them with Ti(OiPr)4 to produce different titanium silsesquioxanes, and then determine how well these titanium-containing cubes catalyze alkene epoxidation. The rub is that the preparation of trisilanols normally is time-consuming and requires a number of purification steps. But the Delft researchers recognized that a combinatorial approach could be used to synthesize a library of trisilanols faster and at lower cost than the traditional route.

Trisilanols and other silsesquioxanes can be prepared by hydrolyzing a silane such as RSiCl3 to RSi(OH)3, which then condenses to form various RSiO1.5 frameworks (the chemical definition of a silsesquioxane).

The Delft researchers, led by professor Thomas Maschmeyer of the Laboratory of Applied Organic Chemistry & Catalysis, used a robotic synthesis workstation designed by Avantium Technologies to perform many related syntheses in parallel, trying different solvents and trichlorosilanes in various combinations. The crude silsesquioxane products were treated with Ti(OiPr)4, and the resulting titanium silsesquioxanes were evaluated for their catalytic activity in the epoxidation of 1-octene with an organic hydroperoxide.

The greatest epoxidation activity was found for the titanium silsesquioxane where R = cyclopentyl. This is a known epoxidation catalyst, so this experiment did not lead to any new catalysts with superior performance. Nevertheless, the researchers note, the study shows that silsesquioxane precursors to the soluble titanium catalysts can be synthesized in a much faster and cheaper way, without the need to purify the precursors before catalyst preparation.

This approach has increasingly been applied to the search for new materials with desirable electronic, magnetic, luminescent, catalytic, mechanical, or other properties.
THE WORK ALSO uncovered some interesting catalytic trends that "would have been difficult to identify without a combinatorial approach," in the words of the research team, which includes Paolo P. Pescarmona at Delft, Ian E. Maxwell at Avantium Technologies, and Jan Kees van der Waal, who is affiliated with both organizations. They found, for example, that the catalysts with the highest catalytic activities were derived from silsesquioxane structures that were synthesized in acetonitrile. The next most effective solvent was found to be acetone, which is the solvent commonly mentioned in literature reports of the synthesis of the open trisilanol.

Can a combinatorial search for titanium-based catalysts lead instead to a new material that shows promise in a totally different area of technology? The answer is yes, based on the experience of a team of Japanese researchers.

More than two years ago, Hideomi Koinuma, a professor of materials chemistry at the Frontier Collaborative Research Center and Ceramic Materials & Structures Laboratory at Tokyo Institute of Technology, Yokohama, and his coworkers began looking for new types of diluted magnetic semiconductors. These are semiconductors--normally nonmagnetic--that are doped with small amounts of a magnetic element to make them ferromagnetic (the state in which all the magnetic moments line up spontaneously so that the material behaves like a magnet).

All the known diluted magnetic semiconductors were based on nonoxides such as gallium arsenide or zinc selenide. Koinuma was interested in looking for such materials among semiconducting oxides such as zinc oxide. Indeed, theoretical calculations suggest that doping ZnO with 3d transition metals (those from scandium to zinc in the periodic table) could lead to a ferromagnetic material. To explore this possibility, Koinuma and coworkers used a laser deposition technique to prepare combinatorial thin-film libraries of ZnO doped with 3d transition metals. "The result was negative," Koinuma tells C&EN. "In none of the films could we confirm ferromagnetism."

As part of another project, Koinuma's group later used the same laser method to make thin-film libraries of titanium dioxide doped with 3d transition metals. Since this semiconducting oxide has catalytic properties, the Japanese scientists were interested in looking for doped TiO2 films that would be able to catalyze the decomposition of environmentally harmful chemicals such as dioxin. It occurred to Koinuma that the TiO2 libraries also should be checked for the possibility that a ferromagnetic material had been created.

The Yokohama researchers used a scanning SQUID (superconducting quantum interference device) microscope to take magnetic images of the TiO2 libraries and were surprised to find that oxide samples doped with up to 8% cobalt were ferromagnetic. "We observed no sign of such ferromagnetic behavior" in combinatorial TiO2 films containing other transition metals, the researchers reported last February in Science [291, 854 (2001)].

Magnetic semiconductors have been known for decades, but all the previously reported ones lose their magnetic properties at temperatures well below room temperature. The remarkable thing about cobalt-doped TiO2, Koinuma and coworkers found, is that it is ferromagnetic at room temperature and above.

SUCH A MATERIAL could be a boon for quantum computing and the emerging area of spintronics. Spintronic devices exploit an electron's spin to carry information, rather than its charge. Electron spins would therefore provide the basis for advanced computers in which magnetic data storage and electronic processing functions are integrated on the same chips, or for quantum computers, which would rely on coherent spin states to transmit and store information.

Koinuma's room-temperature magnetic semiconductor has another big selling point: It is colorless and transparent. So the material offers an unusual combination of potentially useful properties. It might, for example, allow transistor and magnetic memory functions to be integrated into a thin-film display. One such possibility is electronic paper.

Considering the potential, it's not too surprising that Koinuma's Science paper sparked a rush at other labs to make the same material. Scott A. Chambers, a materials chemist in the William R. Wiley Environmental Molecular Sciences Laboratory at Pacific Northwest National Laboratory (PNNL) was one of those who was intrigued by the Japanese results. He felt that the magnetic properties of the material could be improved if it were made in a more carefully controlled fashion.

Koinuma and coworkers grew films of cobalt-doped TiO2 (in the anatase form) in a vacuum by blasting TiO2 and Ti0.5Co0.5O2 targets with a laser--a process that can generate several different atoms and molecular fragments. Condensation of these energetic fragments on a substrate typically leads to a film containing many defects, Chambers explains. He and his coworkers, on the other hand, grew anatase films from single atomic beams of cobalt, titanium, and oxygen that were separately generated and controlled. This allowed them to make higher quality films that display considerably better magnetic properties than the Japanese material, Chambers says. For example, the magnetic strength of the PNNL material (as measured by the magnetic moment per cobalt atom) is about four times greater than what Koinuma and coworkers reported.

In addition, the PNNL researchers, in collaboration with scientists at IBM Almaden Research Center in San Jose, Calif., have determined that the cobalt-doped TiO2 remains ferromagnetic up to at least 250 ºC. And they say they have ruled out the possibility that the ferromagnetic behavior is due to inclusions of cobalt metal in the anatase TiO2 lattice. Chambers presented these results earlier this month at the Spintronics 2001 conference in Washington, D.C., and also has submitted them for publication.

The film-growth technique used at PNNL--oxygen-plasma-assisted molecular beam epitaxy--allows high-quality films of the magnetic semiconductor to be grown. But that technique is not amenable to rapid combinatorial searches because it can grow only one film composition at a time, according to Chambers. So the Japanese work was "a very important step," he says, because the laser deposition method they used is "a good way to screen a lot of materials."

The serendipitous discovery that TiO2 can be doped to form a room-temperature magnet illustrates the "big advantage" of combinatorial synthesis and high-speed screening, Koinuma says. As these methods become more widely used in materials research, serendipitous discoveries are likely to occur more often.

[Previous Story] [Next Story]


Chemical & Engineering News
Copyright © 2001 American Chemical Society

C&EN Classifieds
Home | Table of Contents | News of the Week | Cover Story
Business | Government & Policy | Science/Technology
Chemical & Engineering News
Copyright © 2001 American Chemical Society - All Right Reserved
1155 16th Street NW • Washington DC 20036 • (202) 872-4600 • (800) 227-5558

CASChemPortChemCenterPubs Page