LOW-RESOLUTION (ABOVE) AND HIGH-RESOLUTION IMAGES (BELOW) OF CLEAN PLATINUM EXPOSED TO A MIX OF CARBON MONOXIDE AND OXYGEN.  THE IMAGES SHOW AN ORDERED STRUCTURE OF CARBON-MONOXIDE MOLECULES FORMING A MOIRÉ PATTERN DUE TO THE MISMATCH WITH THE PLATINUM LATTICE, AFTER THE SAMPLE WAS HEATED TO 456 K.

Traditionally, STM is done in high vacuum; some chemical reactions, however, including important catalytic reactions, occur only under pressure. There is no dependable way to extrapolate from high-vacuum STM experiments to high pressure research on catalysis.

"A while ago the idea of uniting these two communities -- the high-vacuum STM community and the people who investigate catalysis under high pressure and temperature -- just jumped out at me," says Miquel Salmeron of Berkeley Lab’s Materials Sciences Division, who in 1993 built the first scanning tunneling microscope capable of working with samples under heat and pressure.

"Some kinds of atomic-scale studies can only be done in vacuum, because many of the techniques used in them -- low-energy electron diffraction or electron microscopy, for example -- are based on electrons traveling out of the sample, and gas molecules diffuse these electrons," Salmeron explains. "With STM, however, although high vacuum is useful to keep the sample and the microscope’s probe tip from reacting with air or other gases, there’s no intrinsic reason it has to be done in vacuum."

Recently Salmeron and his colleagues, including Gabor Samorjai, John Jensen, and Keith Rider, have used the new technique to discover unsuspected features of catalytic reactions on the surface of platinum at the atomic scale.

A catalyst is a material or chemical that speeds up reactions without itself undergoing permanent change. In the catalytic converters of automobiles, platinum is used to promote the conversion of exhaust gases -- carbon monoxide and unburned hydrocarbons -- to carbon dioxide and water. What happens when carbon monoxide adheres to an exposed platinum surface?

Salmeron and his colleagues showed that the answer critically depends on pressure and temperature. Despite what some researchers had argued, the molecular structure of carbon monoxide adsorbed on platinum at ambient temperature and high pressure is fundamentally different from the structures seen in vacuum at an extremely cold temperature, even when similar amounts cover the surface.

"In a cold vacuum," says Salmeron, "carbon monoxide molecules form ordered structures that are different and may not represent equilibrium structures but rather metastable ones, frozen in place. Under heat and pressure, however, the carbon monoxide covers the surface uniformly. This layer is not only in equilibrium with the gas; due to the much higher temperature, it also has had time to find its true spatial equilibrium arrangement."

A close-packed hexagonal lattice of carbon monoxide overlies the platinum substrate atoms, in the ratio of three carbon monoxide molecules to approximately four platinum atoms, forming a striking hexagonal moiré pattern. "Moiré patterns result when two regular arrays overlap and give rise to a third periodicity," Salmeron explains.

It was Salmeron’s inspiration to mount a scanning tunneling microscope inside an air-tight chamber immediately adjacent to a traditional high-vacuum chamber. The platinum crystal is first prepared in the high-vacuum chamber; samples can also be characterized here using standard techniques such as low-energy electron diffraction and Auger spectroscopy. The cleaned specimen is drawn out of the vacuum chamber into the second chamber by a long rod and positioned under the microscope tip; here it can be imaged in vacuum first, if desired.

A CLEAN SINGLE-CRYSTAL METAL SAMPLE IS PREPARED IN THE HIGH-VACUUM CHAMBER ON THE RIGHT, THEN TRANSFERRED TO THE CHAMBER CONTAINING THE SCANNING TUNNELING MICROSCOPE, WHICH CAN BE PRESSURIZED AND HEATED.

Then, after a gate valve is closed between the chambers, one or more gases can be introduced. In addition to carbon monoxide with platinum, Salmeron and his colleagues have studied a mixture of carbon monoxide and molecular oxygen. Although pressures used to date have been one atmosphere or less, pressures of hundreds of atmospheres are possible.

Samples are heated -- up to 500 degrees Kelvin, if desired -- using an ordinary movie-projector lamp whose hot filament, sealed inside a glass bulb, cannot react with the gas.

After working with platinum, carbon monoxide, and oxygen, Salmeron and his colleagues have investigated the action of the gases with another important catalyst, rhodium. Under heat and pressure, patterns never seen before appear, quite different from the hexagonal moiré patterns evident on platinum. Studies like these may lead to a fundamental understanding of catalytic processes and an ability to control and improve them.

"There is so much chemistry to be investigated at the atomic level," says Salmeron, "that much of what we are doing is simply establishing a technique for working under realistic conditions. We hope researchers will get enthusiastic about using it on many other problems."

"High pressure adsorbate structures studied by scanning tunneling microscopy: CO on Pt(111) in equilibrium with the gas phase," by Jensen et al, appeared in the 9 February 1998 Physical Review Letters. Work by Salmeron and his colleagues on rhodium is in press.